111.! ,

.vl
o. \
JP
_ I.
I‘m
In.
’( _
J, .5 A
. n

I‘ll"

lo III-1|.
i
4 0|;

"\
I

It.

. s:
.((,.)\H.
r [I I

‘- III|

ll'lll‘.‘

dale],

J
t

!

IF.-

’1

' V;
I
o I "‘I '
‘ u, -
.. .‘,
a
...
. ‘
‘I | ‘ E
. .
.5 v ' I ‘:
' ‘. ,u ”.v
‘ i '

1 . t
0.. I1
A.. 5‘ 0.

5'. I ..

. L. , .

bl
5
.

t v-

n: .r

. .

-.

7}

V41
f

H
i

 

H?

J!

II :
,,‘. .

i

.....
I F?)

"J
{

2H9

if.

§
F:
1"”

-f
'7' | an?

t1

IN"
4*, 1*ng

f
' 73ml

 

J

.35).} I.
J!
‘

'i

m
H

  

t’l
!
\li

II

‘I
f u - | ‘
'D
l t H;
‘. ‘zhr ‘ 2
M

 

1 i
”L;0

"if!

   

"1

bu“; '

    

 

' 3' ' n ‘I
, H .l
'37 Hi, h‘

 

.
'1!le
L“‘l’ H

 

0..u.’6
0' v

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

m“ I. 293 00563 0656

3|.IBRARY
Mkihi'gcm State
University

 

 

 

This is to certify that the

dissertation entitled

ON ESTIMATION OF SOME DENSITY FUNCTIONALS
UNDER REGRESSION AND ONE SAMPLE MODELS

presented by

Pathma Sarath Thewarapperuma

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in StatIStICS

W

Hira L. Kou] ‘*“">
Major professor

 

Date W

MSUI'J an Affirmative Action/Equal Opportunity Institution 0-12771

 

MSU ’

LIBRARIES
.—:—.

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

ON ESTIMATION OF SOME DENSITY FUNCTIONALS
UNDER REGRESSION AND ONE SAMPLE MODELS

BY

Pathma Sarath Thewarapperuma

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Statistics and Probability

1987

ABSTRACT

In the analysis of the general linear model
Y1 = xi'a + e1 , the scale parameter If2(x)dx appears
as a factor in the asymptotic variance of the rank
estimator g of B . A histogram type estimator was
constructed in Koul, Siever and McKean (1987) for this
parameter which does not depend on the symmetry of the
error distribution. The window width h of the kernel
function is random and is such that vfihn + constant in
probability as n ->9. This is different from the usual
assumption that lfihn-r(3 , hn 4- 0 as n 4~~ . Here it
is shown that this estimator is asymptotically normal.

An estimator is also constructed under one sample
model for the functional ' [f2(x)dH(F(x)),which appears
as a factor in the asymptotic variance of certain minimum
distance estimators. The function R or its derivative h
need not be bounded. This estimator is shown to be consist

and asymptotically normal under mild assumptions on density

function f and the function H .

To my wife Premitha

and my daughter Anushka

iii

ACKNOWLEDGEMENTS

I wish to express my sincere thanks to Professor
Hira L. Koul for his guidance, advice, encouragement, and
specially for his patience during the preparation of
this dissertation.

I would also like to thank Professors Dennis
Gilliland, Habib Salehi and Joseph Gardiner for serving
on my committee.

Finally, special thanks go to Ms. Carol J. Case for

her excellent typing of the manuscript.

iv

INTRODUCTION AND SUMMARY

Consider the linear model

= t
Y1 xi-B +81 1_<_i_<_n,

where e1, 1 :_i.: n are independent, identically
distributed with cumulative distribution function F and

th

density 1', .xi' is the pi row of aknown nxp design matrix

X and 5 is a parameter vector.

The scalar parameter

Y = 7(f) = Ifd¢(F)
appears as a factor in the asymptotic variance of the rank
estimate of fi . Koul, Sievers and McKean (KSM, 1987)
constructed a new estimator for y, which does not require
the symmetry of the error distribution.‘

The window width hn of the kernel that is used
is 0(n-i) and therefore nibn does not tend to zero as
n tends to w . This means that the methods that are
available cannot be applied to obtain the asymptotic
normality of this estimator.

In Chapter I, it is shown that this estimator is
asymptotically normal in the special case ¢(u) c u .

During the course of the proof, we obtain a
result similar to that obtained by Scheweder (1975) for a
special kernel under a different set of assumptions using
Hajek's projection technique.

Next, consider the one sample model. Suppose that

1, ..., Xn are random variables with the common

distribution function F and density f .

X

The density functional
T s T(f,F) = ff2(x)dH(F(x))
appears as a factor in the asymptotic variance of the.
minimum distance estimators of various models (Boos 1981,
Koul and DeWet 1981). The function .H is defined on (0,1).
The function H or its first derivative h is not assumed
to be bounded. However, some assumptions are made on the

behavior of H (and h) near 0 and 1. An example of the

type of functions considered is given by

H(x) = log ng , h(x) = §f%:§7 .

In Chapter 2., we construct an estimator for T .
Estimators have been constructed for similar functionals
by various authors (Ahmad & Lin 1976, Scheweder 1975,
Cheng & Serfling 1983). In all these it is assumed that
either H or one of its first two derivatives is bounded
and therefore these methods cannot be applied for
constructing estimators for T .

The estimator defined in Chapter 2 is shown to be
consistent and asymptotically normal under mild conditions

on density f and the function H .

vi

Chapter
0

1

TABLE OF CONTENTS

INTRODUCTION AND SUMMARY.

ASYMPTOTIC NORMALITY OF AN ESTIMATOR OF
THE SCALE PARAMETER IN LINEAR REGRESSION.

1.1 Introduction and Preliminaries.

1.2 Asymptotic Normality,
Theorems and Proofs.

AN ESTIMATOR FOR THE SCALE PARAMETER
APPEARING IN THE MINIMUM DISTANCE
ESTIMATION OF THE LOCATION PARAMETER.

2.1 Introduction and Preliminaries.

2.2 Consistency of the estimator.

2.3 Asymptotic normality of the estimator.

BIBLIOGRAPHY

vii

Page
(v)

31
34
36

42

CHAPTER I

1.1. Introduction and preliminaries.

 

Consider the linear model

= ' ' =
Yi xi 8 + e1 1 1, 2, ... n

where el, e2, ... e are independent identically

n
distributed (i.i.d.) random variables with cumulative
distribution function F and density f, xi' is as is
th

usually the case, the 1 row of a known nxp design matrix K

and‘ B is a pxl parameter vector.

The rank estimates of B are the values of 8 which

minimizes Jaeckel's (1972) dispersion function given by

545

D (B) =

¢ a(R(Yi — Xi'B ))(Yi - Xi'B-)y

i=1

where R(Zi) denotes the rank of Zi among Zl’ Z2, ... Zn’
and the rank scores are generated by

a(i) =¢KF$I) for a non-decreasing function ¢

defined on (0,1).

For a complete rank analysis of B one requires an estimate
of the scale parameter

v = Ifd¢<F>
For example this appears in the asymptotic variance of the
R-estimates of B . (See Koul (1984, monograph) or
Hettmansperger (1984) for a complete discussion in the case

of Wilcoxon scores.)

In Koul, Sievers and McKean (KSM, 1987) an estimate of 'y

was propsed based on the following arguments. If g and C
are independent random variables with distribution functions
F and ¢(F),respective1y,then the distribution function of k_;|
is H(y)===]iT(y + X) - F(-y + x)}d¢(F(X)). y > 0 .

AHere and throughout the following if limits of integration
are not specified then they extend from -w to +00

When F has density f, density of H at 0, from the
definition 0f 7,18 27. This suggests that an estimate of y

may beaohtained by estimating the slepe of H(y) at y = 0.

Note that, by a probability"transformation we can write
H(y) =(!1{F(y + F_1(t)) - F(-y + F'1(t))}d¢(t)

where F—1(t) = inf {xz F(x) 1 t} .

Let E‘ be an estimate of 8 . Then the residuals Zi ,
are defined by

= ... ' ° ::
Zi Yi xi 8 , 1 1, 2, ... n

Let Fn be the empirical distribution function of the Zi's.

An estimate of H(y) can thus be defined as

l _ _
I {Fn(y + Fn 1m) - Fn(-y + Fn 1(t))}d¢(t).

Hn(y) o

y > 0

—1

where Fn (t) = inf {xz Fn(x) 1 t}

n .
Then H ( ) = .1. X E { J. _ 1:}.
, 1= 3=

1
(12(1) ‘ Z<i>' i 3)

If tn is chosen near zero, an estimate of y is

 

Y = Hn<tn)
n 2tn
1 E l I
Let f (x) = ' 1( x - Z. <‘t )
n 2ntn j=1 J n

It is then possible to write
Yn = Ifn<x)d¢(Fn<x))

In KSM the choice of tn was determined as follows.
th

Let tn 0 be the a quantile of the empirical
,
distribution function Hn' Define estimate §n a of Y by
“g = H (II-$1”. ) / Zn-ét ' O< a < 1
n,a n n,d n,d’

When c e C = {¢:(0,1) + (0,1) : o is non-decreasing,

right continuous, bounded and ¢(O) = O, ¢(1) = 1.},

A

then the consistency of Yn a uniformly over all t e C

was shown in KSM under the following assumptions.

(A1) (§f§)-1 for all n 1 p and
max xi'(§f§)-lxi + O as n + m
1 i,i i n
(A2) F has a uniformly continuous, bounded density f,
f > O a.e. Lebesgue.
-1 x -
(A3) ILé (B - 8)” = Op(1) where A = (£15) % and

||-|| is the Euclidean norm.

We will here obtain the? asymptotic normality of this

estimator for the special case ¢(u) = u ._ In this case,
H" (y) =91- . 22- 1 ' "
n 2 - <-|,Z~--Z Loy)
n ij 1 J

' 1
= — - 1 . , —1 ,<

:3

Thus we want to show that
. -é _ —§ ‘ 2
/K{Hn(n tn,a)/2n tn,a — If (x)dx}

is asymptotically normal for O .< a '< 1
We modify Hn(:y)_ process in the following manner.

Since [[A'1(B-B)H =z. 010(1) ,

(-xi-xjrAA’ltfi-s) (xi-xjmt where ”in ~ : a < ..

II
(I.

and bij (.xi xj)A.

Let Kn(t,y) -

l
DIH
HM
uM
H
(K
(D
I
P
I
U‘
H‘
t...
d
A
:3
to
‘<
v

un(t.Y)

II
E!
IN

:3
A
ri-
‘<
v
m
:3
0.

let Xn(t,y) Kn(t,Y) - unct)y)

We will therefore consider
Kn(t,y) - 2niy ff2(x)dx
and obtain its asympmotic Irormrtics as a process in t and y.
In Theoren1.l weewill show that

(1) sup sup [Xn(t,y) — Xn(0,y)[ -> o in probability as n+°°.
t v

G

('1

Here and throughout the following sup sup means that
t y
the suprema are taken over all t and y such that

IItII'j: a and 0 < y.f b < m .

This means that the asymptotic properties of
Xn(t,y) can be obtained by considering Xn(0,y) . We
obtain these properties in Theorems 2 and 3.

Since un(t,y) = EKn(t.Y)

....1.

_ l .. 2
— E H E §1(Iei - ej - bIj tI f_ n y)

= Ii; . Z I [F(bi'j t'+ n'%y+X)-F(bi'j t-ne_%y+X)3dF(X)
1 J
we have
(t -1ZZ[Gb't+-% Gb't-é]
un .y) - H i j ( ij n Y) ‘ ( ij ‘“ V)

where G(y) = [F(y+x)dF(x) is the distribution function

of e1 e2

In order to consider Xn(t,y) in place of
Kn(t,y) — Z/Hy [f2(x)dx we will first show
in Lemma 1 that

sup lun(t,y) - 2yn% f f2(x)dxl + O as n + w
tiy

<

Suppose that (Al) holds and that the second

0*”.

meal:
derivative" f" of ‘f is bounded. Then,
sup lun(t,y) — 2yné I f2(x)dx| + O as
t.y
; i 2
Proof. lun(t,y) — 2n y [f (x)dx|

|% Z Z I{F(bij't+fiiy+x) - F(bij t-fiiy+X) - 2n'iyf(x)}dr(x)l

i J

Eli-i

g Z I{F(x)+(bij t+n-éy)f(x)+%(bij“t+n-éin
J

'(X)-+

l(b_'t+n-§y)3f”(.x+el) — F(x) - (bfiFt—nnéy)f(x) -

6 u

%(bij't-n‘iy)2f'(x) '.%(bn"t'n’iy)3f"(x+92) ’

2n-éyf (x)}dF(x)I

where 61 and 92 are chosen appropriately.

I§§§f21féybift f(,x) +l écbij: t+n§y)3f' (x+el)

- écbij't—n'iy>3f"(x+ez)}dF(x)l
const - % X Z (Ib1 ftI+ nély|)3 Z
_ ( b..%
3 —3/2 1 J 13

Note that the second term tends to zero as n tends to m .

Next it is shown that % Z I IbijtJB -+ o as n + m ,
i J

This proves the lemma.

Now
1 3 '1
a Z ; Iowan = h- z z |(x1 xj)' At]3
1 J 1 J
_ 1' 2 ._ I
- B’ 2 Z |(C1 - c )tl |(c — c )tl C.—xiA
1 J
1 3
:3- ;g HG, - an2 Hci - cjna, Htllja
. 3
2a 2 2
g n §§<Hciu +Ilcjnv><ncin+ncjn>
3
8 2
:-—§-—max “Gill 2 H61”
1 i
2 1‘1
but 2 ”Ci” = Z xi'azixi = traceZAxi' xiA= p and
1 1 ' i

is therefore bounded

and max “(Gil + 0 as n + w, by assumption.
1

1 3
Thus BEE Ibij'tl +0 as n+oo.

Let us also note that

2 . i i
z IICiH _<_ due,” 21):. = mp“.
i i 1

1.2 Asymptotic normality of Hn (Theorems and proofs)

Theorem 1. Under (A1), (A2) and the assumptions in Lemma 1

 

P
sup sup IXn(t,y) — Xn(0,y)| + O as n + w .

t Y
Proof: sup sup IXn(t,y) - Xn(0,y)| is first written as
t Y

a sup(max) over a discrete set in the following manner.
See Mugantseva (1978) for details of the idea used in the

following subdivision.

Divide the interval [0.6] into Nn subintervals of

equal length by the points y i = 1, 2, ... N

i ' n

Since t = A-1(§ - §§ 6 RP we define points sk in

RP so that the ball of radius a in Rp is divided
-k
into smaller balls of radii 2 n.

We will specify the values of kn and Nn later.

Then

sup sup lxn<t,y> - xn<0.y)l

t y

:max SUP sup IX (.t.y) - X (s .y)|
k llt-skll 16,, y n n k

n

+ max sup an(sk,Y) - Xn(0.Y)I
R Y

Let us first consider the second term. Here we remark
that the full force of the above subdivision is not
needed to show that this term goes to zero as n + w
This can be obtained using a finite partition. But this

division is required to obtain bounds for the first term

in the right: hand (side of theeahove inequality.

Now

max max IXn(sk,y) - Xn(0.Y)I

k y
'1
= max max I— X Z {1 _ , -% -1 _ . _ -%
k y n i j (ei ejfbij sk+n y) (ei ejlpij Sk n y)
-1 -§ + 1 —;
(ei-ejjn y) (ei-ej i -n y)

_G(bijisk+n—%y)+G(bijis —n-%Y)+G(n-%Y)sth—%Y)}I

k

(0

By Lemma .1, we have

sup |G(bij t+n'%y) - Gcbijt-n_%y) — G(n-%y) + GC-n'iy)|
tiy

' -% I -5. £3 2
5 sup |c;(bij t+n y) - G(bij t-n y) — 2n y [i (x)dx|
t.y

+ sup |G(n'iy) - G(-n‘iy) — 2niy fi2(x)dx|
t.y

It is therefore sufficient to consider

1 .
max sup I— Z Z {1 _ . , -§ - 1 _ , _ -§
k y n i j (ei ej:m.. sk+n y) (e. e.:b.. s. n y)

+ 1 _i
(ei—ej:-n “y)}|

Let y be such that yr i y j yr+1 for some r

Then the (i,j)th summand of above

-%

< 1
—- (e.-e«b..'s +n y ) - 1 _ , —§
1~ 13 k r+1 (ei-ejjmij sk-n yr+1)
— —§ + 1 -%
(ei—ejjn yr) (ei-ejjrn yr)
(ei‘ejibij Ski“ yr+1) (ei'ejioij Sk ‘ “ yr+1)

" 1(ei—e._<_’n—é

+ 1 -1 }
J yr+1) (ei-eji—n 2yr+1)

+'{1 g 1

(Gime‘j-f-n yr+1)

_ -§
(61 eji'*‘ yr+1)

1 -. + -
(ei-ejjn éyr) 1(ei-efign iyr)}

10

Using the same procedure a similar lower bound can be

obtained for the above summand. Thus
max mgx lKn(sk,y)-Kn(0.Y)I

mix max IKntsk,yr,1) — Knc0.yr+1)|

(2)

IA

+ max max lKn(O’yr+1) — Kn(0’yr)|
k r
+ Similar terms arising from the lower bound.
We will now show that each term in the right hand side of (2)
tends to,0 in probability as n tends to w .

Given e > O , consider

P(m:x max lKn(skyr+1) - Kn(0’yr+1)|
r
1 .
:_—2 g £‘Var(Kn(sk,yr+l) -'Kn(0.yr+1)),
e

Var(K n(sk’ yr+1) - Kn(0’yr+l))

till—i

2 E g Var(Knij Sk'yr+1) ’ Knij(0’yr+1))

/\

(A

v
+

'1
- I
n J

#.#g COV(Knij(-Sit'yr+1) ‘ Knij(0’yr+l)’

2 i
Knmj(sk,yr+1) " Knmj(0’yr+1))

’1 . ‘ . . = ...x
where Yn13(t'y)%Ie -.3- a J tl 5 n 4y).

Consider the two terms in (3) separately.

n _ r .
Var Knij(°kpyr+1) Lnlj(0iyr+1))

IA

. - - ‘ 2
(lei'ej‘bij skii“ yr+1) (lei ejli“ yr+1'

A.

-1 -
I{F(bm +n ayr+1+x) — F(bij'sk—n 2yr+1+x)

_: -1
+ F(n y 1+x) — F(-n 2y

+ "% I
r+ _‘ r+1 x) — 2[F(n y +b s

r+1 ij kAoix)

_ v(—n-éy + +bij'skY0+X)]}f(X)dX

1U

11

But
F(b 's +n-éy +x) — 2F(b 's AO+n-%y +x)
ij k r+1 ' ij k ° r+1
-i,“ ,
= ‘ I _ " I ‘ I _%
{bij sk (bij skAO)}f(bij skA0+n yr+1+x)

I , i "'2 I
(bij skAo)f(biJ. skA0+n yr+1+x) + 0(1)

Ibij'sk|f(bij's Ao+n'i +x) + 0(1)

k yr+1

A similar expression may be obtained for the remaining
three terms.

But we observed that

1221' 1
- ' ' b- S | < - X I H0 -C H IIS H
n i j 13' k —-n i j 1 J k
i né (constant)
Thus.
l X Z Var(K (s ) - K (O p < ( t) "i
(4) n2 . j nij- k’yr+1 nij ’yr+1 ’—~ cons n -

Covariance term.

 

1
3‘2 2 X Z COV(‘Knij(Sk’yr+l) " Knij(0’yr+l)’
iriim - '

I! _
unmj(sk.yr+1) Knjm(0.yr+1))

1

‘2 i i i Cov(K ..(s.,0,y ) - K .(s ,O,y .) (say)
h i#j#m.- n13 k r+1 nmJ k} r+1

12

égfgfg {E Kn i w(sk,0 yr+1)Km j(sk'0' yr+1)

— EKnij(sk,O,yr+1)EKnmj(sk,O,yr+1)}

We will obtain a bound for the first term.
From what follows it is easy to see that a similar bound

may be obtained for the second term.

Consider,

12 2 Z X E {KnijCSk’yr+1)‘-’ j.,(0 yr+1)}'
ifjfm

n

‘ {Knmj (.Sk! yr+1) "' Knmj(0 , yr+1)}

_ 1 _ . ' _% _%
_ g2 i$§f§ I {F(bij Sk+n yr+l+x) —IF(bijskan yr+1+x)

. _1 _
(5) - F(n §yr+1+x) + F(-n J‘iyrflthxH'

tot-

. -% _
o ' " _ _
{F(bmj sk+n yr+1+x) F(bmj'sk n y +x)

r+1

-F(n-iyr¥1+x)£F(-n-&yr+1+x)}f(x)dx

Note that
F(b 's +n-%y' -4- -F(b 's —n-%y +x)
» ij k r+1 X) - ij k; r+1
p “g? "ii
- *(9 ,yr+1+x) + F("n yr+1+x)
_ -iy -i -1 2

-5 -g g -1 2 ,
' 2n yr+1f(-n yr+1+x) -—4n yr+1f'(x+t2)

_ a *1 '
__ 2n yr+1hij t Skf' (”1,1,3 ) + O,(n )., 51110 9 III and yr+1

are bounded. 0 does not depend on k and r

13

Then (5) can beawritten as

7 '3
‘ ' -1 2 . '7
izai E Z I4n yr+lh1j‘skpmjtskf'tx+w3)f'<x+w4)f(x>dx+ocn >
1mm .3
= M1 t'ls Z 2 Z bij'skbmjtskv+ 0(n 2) where; M1 is a constant.
17‘3'7‘111
1 '1 '.
But I;3 Z Z X b13 skbmJ'skI - ;3IZ(§(xi'xj)'Ask)(g(xm-VJ)UhfiM
i#j#m J
1 2 'M2 2 p
(6). ‘ :3- ;(x‘j'Ask) 3: K 3;]ch = H where M2 is a constant.

From (3), (4), (E), and (6) it follows that
kn
. ~ f . . 5 ’2
Z {Var(,Kn(_sk,yr+1) - Kn(,0,yr+l)_) 5113 2 Nn n ,
k r
where M3 is a constant.

k .
Hence if kn and Nn are such that 2 n'Nn“ ni'd‘ for some
fixed 6 > 0 then

k .
2 n-Nn-n-i + 0 as n + w~.

This completes the proof that the first term of (2) tends

to zero in Irobability as n tends to w .
Next we want to show that

mix [hn(0,yr+1) - Kn(0,yr)|+ O in Irobability as n + m.

It suffices to show that

% é )

_ .- 1 -

- 1 - }

+ O as n + w .

14.

Now ‘Var (Z z'{1

1. j (leiCejl—fn yr+l)

= g g var(1(n-§yr<leai_efi 1'0 yr+1))

+ Z Z Z Cov(1

-é _ -- ,
ifjfm (n yr<|ei ej'fn VH1)

—§ -I -;

1 )_
(n yr (lem-e‘] in yr+1)

Since the computations for the first term is
easier we will only obtain the relevant bounds for the
second term.

Again we need to consider only the term

E 1

DP

-%

_ 1 -1
(n yr<|ei-e.|:;n ifyr+1) (n 2

J yr<lem-e.|:r1

J yr+1)

since the same bound works for the product of the

expected values.

-1 '
yr+1) (n Yr: IgniéjIan

'2

1H

"0

Il§(n'éyr+1+X) - FC-n'éyr+l+x) --F(n 'yr+X)

+ F(—n‘5yr+x)] 2f(x)dx

f(-n-%y

f{2n-&yr+1 r+1+x) - 2n-%yrf(-n-§yr+x) + 0(%)}2f(x)dx

NP

. _% _1 -1. . _
I{2n <yr+1-yr)f<-n 2yr+1+x) + 2n 2yr{fc-n yr+1+x)

- f(—n-%yr+x)} + O(%)}2f(x)dx

{{Zn—%(yr+1-yr)f(-n"%yr+1+X) +

- _}
+ 2n 1yr(yr+1-yr)f'(x+¢3(yr+1-yr)n &) + 0(%)}2f(x)dx.

0 5 ¢ f 1

3

—i
_ 2 ~ -
= 4n 1(yr+1-yr)2 If (-n yr+1+X)£(X)dx + 0(n 3/2)

1
Hence £2 E E % E{Knij(o’yr+1) - Knij(0’yr)}'

2
#1f

‘{Knmj(0’yr+l)-Knmj(O’yr)}

>1 2 '1
< const — Z X Z 2(y -y ) = N - —
— n3 r r+1 r n N 2

+ O as n + m if N + w

This is true if Nn : n 1 for some small 51 > 0

Since the second term may be treated in a similar manner,

these two results now imply that

max max IK (sk,Y) - K (0,y)| + O probability as n + w .
k n n

We will now consider

max sup sup IX (t,y) - X (S..Y)|
k Ilt—skllsén y n “ “

This is where the division of '{t :IItH i a} into small

balls is used. "Recall that sk are the points such that

A

IIt-skll - 6n

an(t.y) - chsk.y)|

'{1 — 1

3H4

=l g

L-IoM

- -1
(ei-ej_<bi .' t+n §y) ('ei—ej-fbij" t _—n 2y)

J
—é

1(ei-ejjbij'sk+n-%y) + 1(ei—ejitrtj'sk-n y)
- G(b..'t+n-%y) + G(b..'t-n—%y) + G(b..'s +n-%y)

13 13 13 k _1
-G(bijsk—n 3y)H.

16

By using lemma 1, it follows that supremum over both
t and y of the nonrandom terms in the above expression

tends to zero as n tends to w .

We also note that

Ibij't-bij'skl f- Ibij I Ht"5k|| If. 013 5n:
where Ibijl = Cij’
' I
Hence bij sk - Cijén i bij t i bij'sk + Cijén .

Let y be such that y Then

r i y : yr+1

1 2 if
n i j (.ei-ejf-Dij t+n y) (ei-ejf-hij t-n Y)

‘1 'é‘+1 t -%'})l
(e.-e. y) (ei-ejjtrtj sk-n y)

I
1 Jibij Sk+n

is bounded from the above by

l ' -1
n z 2 Z { 1(ei-ej: hijlsk+cij5n+n 2yr+l)

-%

_ 1 '
(efejibij Sk‘cijén'“ VH1)

- 1 I "'5 + 1 —%J-
‘ei'ejf-bij Sk+n yr+1) (ei'ej-fbij' Sk-n VH1)
+ 1 __ t “% " 1 I ‘21):
(e1 ejjbij sk+n yr+1) (,ei—ejibi‘j sk n yr+1)
"‘ 1 I ‘i‘ + 1 I "g?
(ei-ejilaij sk+n yr) (ei—ejibij sk-n yr) }

An expression of this form may be obtained as a lower bound
for this term. Since the computations are similar we will
not show this lower bound explicitly in the following

calculations.

VI.

‘A.

PM

17

* 1
Var(-— {-2 1 _ . 42
n 1 jE (e: ejibij Sk+C1j5n+n Yr+1)
- 1 ' . -5
(ei—ejibij'Sk-Cijén-n yr+l)

- 9
(e1 ejilnij sk+n yr+1)

+ 1 , —§ ’]
(ei—ej_<__bij sk—n yr+1).

DIH

2 E gvar(Knij(Sk'Cij 6n+yr+1) " Knij(sk’yr+l))

+

a “A

g COV(Knij(sk’Cij6n+yr+l) - Knij(sk’yr+1) ’

2 §%§¢

Knmj(sk’cmj6n+yr+1) - Knmj(sk’yr+1))

We consider only the term'
(7) EE‘nij(Sk'Cij 6n+yr+1) " Knij(sk’yr+1)]'

All the other terms in the above expression may be treated

in the same manner.

(7) = I F(b 's + C 6 + n-% +x)
1: ij k 13' n yr+1

- C..6 —n-% +x)

 

_ I
F(bij 5k 13 n yr+1
_ t '5 ".12 .- ,.
F(bij sk + n yr+1+x) + F(bij'sk—n yr+1+x) .
I -§ _ v I _ __ "%
|'_:F(bm‘j sk + ij6n+n yr+1+x) F(omj sk ij6n n yr+1+x)

_ u ӎ %
F(bmj Sk+n yr+1+x) + F(bmj' sk — n yr+1+x):lf(x)dx
= I “é ' "%
-; —%
ij 6n E(bmjsk+n yr+1+x) + f(bmj' sk-n yr+l+x)' f(x)dx

+ higher order terms.

18

Since f is bounded, the leading term in this expression

 

is
C C 6 2 ° constant
ij mj n
But, —2 Z X 2 c1 cm n2
n 1 j m J J
6 2
'n
n J 1
‘6 2
‘ n i 2 % A 2 w
= Z n ( |c.—c.| )~(n~§|c -C.|,)~
n2 j E 1 J m m J
2
- 5 n
"' n a
T HICl-CJI
J 1
Thus
2
—2 Z Z Z Ciijj n
n 1 J m
2
26
< n 2 2
_'—H_ Z Z (llCiH + Ilcjll )
J 1
= (constant) ° fig
This shows that
1 2 kn 2
X g :2 Z X Z Cijcmjdn i (constant) ° 2 . Nn . 5n
ifli‘m
2 Nn . 6n + O as n + m

-k k

(Recall that 2 n = 6n and 2 n'Nn"n

Hence

max 8111? SUD g:- 0 Z ELK ..(S ’C 6 +y )

k “t-Sk“ <5 r? n i 1 his k 13 n r+1
—— n b

- Knij(sk’yr+l)}
+ O in probability as n + w .

The second term is

1

DIH

—%
' (l‘ h!
(e.--e._<__bi'j sk+n yr)

-1 -
' a
' Sk+n yr+1) 1 3

2X1 <
i j { (ei-ej___‘biJ

"’ 1 I "é + 1 I _ "%
(ei-ej—Sbij sk-n yr+1) (ei'ejf-bij Sk n y )}

= Kn(5k,ur+1)-Kn(0.yr+1) + Kn(0.yr+1)-Kn(0.yr)
+ Kn(0’yr)-Kn(sk’yr)
The proofs of the first part can now be used to get the

required conditions for this term.

This completes the proof of Theorem 1.

'> In Theorem_2,.we obtain the asymptotic normality of
Xn(0,y) . We will prove this for a general function ¢
since this result is of independent interest. It is assumed
that the function ¢ is twice differentiable everywhere

and that both ¢' and ¢" are (uniformly) bounded.

W‘firu

... x

20

Scheweder (1975, 1981) has obtained this result
for a general kernel under different assumptions using
afiVoneMises type.expansionr In his {roof he assumes
that /th + 0 but in ours /th + constant. Of course,

our result is valid only for the special kernel considered.

'Theorem'Z: Let ¢z [0,1] + [0,1] bela nondecreasing twice
differentiable function such that is first two derivatives

¢' .and ¢" are uniformly bounded and continuous. Then

' J. _ '.J_-_1_ '_ t .
£32 {¢(n) Mn “lfllegiregfliy/VH) for vy 1s

asymptotically enauivalent in probability to
1 , ‘
nilcb (F(.€j))1(.|ei-ej|_<_y//n)

+ 2n'5y I, ¢"(.u)f<.F-‘1(-u)>i{1

-u}du
[0,1] (F (- 93 )1“)

and hence

we) - ¢ci=l)}1 ' a - ¢'(F(,x)>f<.x)dF(x)
H n “ ”eurecmlf-n Y) I

42> N(,0,02) where

2 _ . X -
a = 4y20f1¢"(x)f(r 1(.x)>{2(.1—x)of u¢"(u)f(F 1(a))du

+ 4Ofx'¢'(u)f(F—1(_u)_)du - 4x0f1 ¢"(,u)f(F-l(__u))du}dx

1
+ 16y20I1 4>' (mm-1(1)}? ax - 16y2<ol ¢'<.x)r(r’1<x))dx)2.

21

Before giving the proof of this Theorem we state the

following important corollary.

'Corollary'iz By taking ¢(u) = u in Theorem 2 it follows

 

that for Vy
2
Xn(.O:Y) z? N(-0100)
2 2
where (b = 16y2{ff3(x)dx — ([f2(x)dx)2} = yzcJ (say).

‘Proof’Of Theorem'z:

Since

.,._1 .1 '..
Elsup ; 2 mg- )- “971—1, — B- e» (fi—)Jl(le _e(j)|_<_n-%y)|

¢ Y 1 J (i)
:3 811M Z {|¢'(§—)-¢(3§l)4-§-) -7 -,1;¢"(,l;—1)I_ +’ -,1;|¢'<%)-¢'<~lgl)l.}
my 1 j

1 —é
"'e<,i)‘e(j)I—‘“ Y)

= 2M I{F(x+n-%b) - F(x-n-éb)}dFCx) +»0 , because F is

continuous and by DCT, it suffices to consider

aIH

:21;¢1(n)1(|e(i)—e(.j)|_fn-éy)- 2ni’y I ¢'(F<x>)f(,x)dF(x).

Let wn

and

22

-1
1(.It-SI :in “y)

I wn(t,s)an(s) .

n&(.2§l)—1

(t.s):

fn(t)%

Following Scheweder (1975) let us consider the difference

1
n

where'

and

i g ¢'(anu‘fl ' ¢'(F<e

. )) 1 -1
(J) (|e(i)-e(j)| j n 3y)

n I I{¢'<Fn<t)) - ¢>'(F<t)>}1(|t_sl_f n—éy)an(t)an(s)

zn'éy I l{¢'(Fn(t)) — ¢'(F(t))}wn(t,S)an(t)an(S)

 

I ¢'(Fn(t))-¢'(F(t))

% v ‘
Fn(t) _ F(t) fn(t) n‘{Fn(t)-.(t)}drn(t)

2y
2y I ¢h"(F(t))-¢"(F(t))fn(t)Wn(t)an(t)
+ 2y I¢"(F(t)){fn<t)-f<t)}wn(t)an<t)

+ 2n'iy f¢”(F(t))f(t)Wn(t)de(t)

+ 2n‘5y f¢"(F(t))f(t)Wn(t)dF(t)
Bn1 + Bn2 + Bn3 + Bn4 (say).

-‘¢n"(F(t)) = {¢'(Fn(t)) - ¢'(F(t))}/{Fn(t) -F(t)}

_ é
Wn(t) — n {Fn(t) — F(t)}.

From KSM(1987) it follows that

sup |fn(u) - f(u)l + 0 in probability
u

and therefore fn is uniformly bounded in probability.

Under the assumption that ¢" is uniformly continuous

23

sup I¢n"(u):-¢"(u)I-+ 0 in probability.
u

Wn converges weakly to a Brownian bridge and therefore is
uniformly bounded.in probability.

These imply that Bnl’ an + O in probability as n + w.
Mooreflgfigxas in Scheweder 1975) implies

Bn3 + 0 provided ¢”(u)f(F—1(u)) is of bounded
variation.

From this analysis, it follows that

Z 2 ¢:(%) 1(Ie _e |< n-iy) - 2ynéf¢<r<x>)f<x>dF<x)
i j (i) (j) - ~ '

is asymptotically equivalent to

DIH

<n—éy) +

E E ¢ (F(ej)) 1(lei _ j|_

+ Zn-éy IE I ¢"(u)f(-F-1(u)){1(F(ek)< u) - u}du
ELI} ‘

- Znéy f ¢'(F(x))f(X)dFCX)

Let us note that

DIH

Z Z ¢'(F(ej)) 1 —%
i J

(lei-ej Iii] Y) '

= f? X Z f{¢'(F(.X))F(.n_% y+x)—F(_n‘§y+x)} . f(x)dx

It is now easy to see that

BIH

e.-—e.|__§ n-E

¢'(F(e.))_ 1
E E J (l 1 J

Y)

- 2n%y I¢'(F<x))f<x)dF(x>l

24

+ O as n + w under the assumption that ¢' is bounded.
= ' . . .. ,1

—-f ¢'(F(x)){Fcn’%y+x) -F(-n-éy+x)}f(x)dx,

_* .1
T1 ’ H X 2 Tij '

*3
II

‘% _ n -1 _
2 2n ygé, ¢ (,u)f(F (“)){1(F(;ej)-_<u) u}du.

and T = T1 + T2 .
Define T the projection of T as in Lemma (4.1, Hajek)
(1968).

Conditional expectations of T with respect to ek‘s

can be obtained as follows.

1. E(Tij|ek) = o if k¥7 i. or k #.j

2. = '{I¢'(’F(u))l(,lek-ulfn-%y)f(‘u)'du
- expected value} if j # i = k .
3- = {f¢'<F<ek))1(|ek_ul_<n-ey)f(u)du
— expected value} if j = k e i .
Hence T = lX{{f<1>'(F(.u))+<1>'(F(6’k))}1‘(l(gym—(rial),)f(u)du
- expected value}
+ 2n-iy if ..¢"('u)f(F-1(u)){1(F((1)511) - uldu .

ELIJ.

I.

II

Now‘l§[¢'(F(u)) 1(Iek’ulj n-%y)f(u)du
= E I _i _, ¢'(F(u))f<u)du
eken y,ek+n y]

Efi'mcekm’i’yn -— ¢'(F(.ek-n'%y)).]
-.2
= E 2n‘*y¢'(Ftek))F(ek) + §~§<¢F)"(ek+...>
1
H

But Z(¢F)"(ek+...) - expected value + O a.s.
k

by the SLLN.

Similarly Z{f¢"(F(ek)) 1 4g - expected value}
k

( ek-ulirl y)

is asymptotically

= 2n'5y £{¢'<F(ek))f(ek) - E¢'<F<ek)>f<ek)}.

It is, therefore, sufficient to obtain the
asymptotic normality of
T1 = 2n‘*y E 2{¢'(F(ek))f<ek) - E¢'(F(ek))f<ek)}

+ f¢"(u)f(VF-1(u)) 1(.F(e -- u du

k)._fu)

But Scheweder (1975) has shown that

ll‘é=:> N(O,02) where 02 is given in theorem 2.

A

This shows the asymltotic normality of T . T will
have the same asymptotic distribution if ‘Var(T—f) + 0

as n + w . This is shown next.

26

lei’e; I: {fl-$3?)

Var(T - 'i‘)= Var(—- Z 2 ¢' (F(ej )) 1(
.J

J

- XI{¢‘CF(ej)).+ ¢'(F(u))}1(|e f(u)du)

J l ..—
. — "*~
U. n- 2y )

Let. L1(ei.ej) = {¢'(F(ei))+¢'(F(ej))}1(lei_ejl;3n-%y)
L2(ei) = f{¢'(F(eil)+¢'(F(u))} 1(lei_u[:,,- ey)f(u)du
L3<e1,ej) = chei) + L2<ej)

and let w(ei,ej) = L1(ei,ej) + L3(ei’ej)

Then ‘Var(T - T) var(2n Z Z Wei ))

i eJ

‘1
:§<} j 2n g 1 J

We will now show that

. '1
Var Un := Var (E Z Z ¢(me and
i<j

Var (2: Z w(ei, ej)) tendSto O as n tends to w .

These two results then imply that

Var (T - T) tends to O as n- tends to w .

Since |¢'l is bounded, it is easy to see that

Var (§% 2 ¢(ei,ei)) tends to 0 as n tends to w .
1 .

.2U 1
Also ——2 = fi— . . - _ . .C
n-l (2) ; <§ WCel,eJ) IS a U statistl

27

_ , .1
and thus ——JL—— 'Var U = a" 2(n-2); + C

(n—1)2 n (2) 1 2
where C1 = ‘Var ¢1(e1) ; w1(x) = EW(x,e2)
and C2 = ‘Var ¢(e1,e2)
Now, 91(x) = Ed(x,e2), does not depend on x
and hence C1 = 0

'(n-1)2 “1
Thus Var U = ——————-. — C e C
n n 2 2
4 (-2 )

C2 = Var W(e1,e2)

Var'{¢'<F<e1)) + ¢'(.F(.92))}1(.|el-e2|:nm%y)

- f{¢'F(e1))+¢'(F<_u>.)} 1(lel_u|1n—%y)f(u)du
- I{¢'(_F(62)) + ¢'(F(u>)}1(|92_u|:n-%y)f(u)du

Again, the boundedness of the function I¢'I = ¢',

(since ¢' is non-negative)imp1ies that

92 = Var W(e1,e2)
+ O as n + w .
This proves that 'Var (T - T) +-O as n + w .

Combining these results, we now have that

(-1) " Quill-i“ Y)

1

v 'l _ '3-1
329th) ¢(.-—fi-)}1(lre

- 2y/H f ¢'(F(x))f(x)dF(x)

28

L" 2 2 . . .
':::aN(0,a ) where o is given 1n theorem 2.

In order to grove the.weak convergence of the

process to Gaussian process of the form (yZ), Q i y i b,

where Z is a Gaussian random variable, we need to prove

‘

the following:“We will now consider the case ¢(u) = u
Lemmajci.
lim sup Var (Xn(0,Y) — Xn(0,x)) i constant~(X-y)2,05x,y:b.
n+oo

Proof of Lemma 2:

Recall that Xn(0,y) = Kn(0,y) - ZKn(O,y)

1
h K 0 = — "
w ere n( .Y) n E g 1(|ei - ejl j n éY)

Hence it suffices to prove that

Var (Kn(0,Y) - Kn(0,x)) i constant (x-y)2

n-l 1

Since Kn(0,y) - Kn(0.x) - —§-(g) g g £1¢;i_ej|: n‘éy)

4 1

 

moaHA

in =(

Z'il < -a -%
j (Iei - ejl _ n y) (Iei — ejl i n x)}

) Z
i

<

and Wu = constant.

We have
' _ 2
Var (Kn(0,y) - Kn(0,x)) - (n—l) Var‘Vn
'1

_ 2 . — .
- (“-1) (g) 2(n—2)C3 + C4 9

c3 = Var E(1

and c4 = Var 1

For large n
Var (Kn(0,y) - Kn(0,x)) = 4(n—2)53 + C4
As in theorem 2, it is easy to prove that

C4+O asn-rao.

-5 -.1
and C Var F(e2+n y) - F(eZ—n 2Y)
- F(e +n.%V) + F(e -n_éx)
2 18 > 2

Var [?n’é(y-x)f(e2) + 2n—1y2f!(e2+e'n—éy)

+ 2n—1y2f'(e2+e"n—éy)]

where 91 and 62 are chosen according to
Taylor's theorem.
Thus,
Var (Kn(0,y) — Kn(,o,x)_) = lGVarBy-x)f(e2)] for
sufficiently large n which implies that
lim sup Var (Xn(0,y) - Xn(0,x)) : constant (y-x)2
n + on

This proves the Lemma 2.

30

From Corollary 1, Lemma 2, and Corollary 15-5
from Billingsley we deduce

Proposition.: Under the assumption stated in Lemma 1 and

 

Theorem 1, with ¢(u) = u
xn(;0,~) =—_—_-_,._x¢o), on D[0,b]_

2
where X(y) = yZ, Z is a N(O’01) random variable.

From Theorem 1, Corollary 1 to Theorem 2, Lemma 3

and Lemma 6 (KSM, 1987) it now follows that

/H{HnCn-étn,a/2n-étn,a - ff2(x)dx}

:22: N(O,of) where of = 16{ff3(x)dx - (ffz(x)dxf%

CHAPTER II

2.1 Introduction and preliminaries,

 

Let X be a random variable with distribution function F
and density function f . The density functional
f2(X)

I dF<x>
F(x)(1-F(X))

 

appears as a factor in the asymptotic variance of minimum
distance type estimators in various models (See Anderson -
Darling (1952), Boos (1981), Koul and DeWet (1981)).
(No limits of integration will be shown whenever the
integration extends from -m to +w.)

In this chapter, we proposean estimator for
this type of functionals and obtain its consistency
and asymptotic normality.

Note that the density functional of interest is of the form

T(f) f f2(x)dH(F(x))

I f2(X)h(F(X))dF(X)

where H is a nondecreasing measurable function definedcxlbgfl

h is the a.e. derivative of H and need not be bounded.
We will assume that the function H satisfies

the following properties.

(1) I dH(x) ~ C log n for large n ,
[n-a, 1-n-a

a > 0 and C is a constant which does not depend on n.

(2) O < h(x) < nza for x glé’fl, 1-n-:].

For example,

H(x) log 1-x , h(x) x(1-x) satisfy these

properties.

Estimators for this type of density functionals have not
appeared (to the authors' knowledge) anywhere in the
statistical literature.

The density functionals of the form
T1(f) = f¢<x>i<F<x>>i2<x>dx

were considered by many authors. For example, Ahmad and
Lin (1976) proposed a consistent estimator for T1(i)
under the assumption that the function w has a bounded
first derivative. Cheng and Serfling (1983) also
considered estimation of T1(f); They have shown the
consistency as well as the asymptotic normality of their
estimator with the assumption that the function w has
a bounded second derivative. Scheweder (1975, 1981) has
a similar result but under more restrictive assumptions.
None of these methods can.be applied to T(f) since the
derivatives of H(-) are not bounded.

Note that when ¢

1 the functional T1(f) reduces to

T1*<f) IfZCX)w(F(X))dX

ff(x)¢(F(x))dF(x) and

is similar to T(f) . In fact, our method can be used to

33

*
obtain an estimator for T1 (F) when w (and its

indefinite integral) satisfy (1) and (2).
2.1 'Proposed Estimator oflTCT).
NoW' T E T(f) = I f2(x)h(F(x))dF(x)
=0f1f2(F-1(t))h(t)dt

=011r2(F‘1<.t) )dH<t )

Let X1, ..., Xn be a sample of size n from F ,

and let Fn be the empirical distribution function of
1’ COO, n

Let fn be the kernel estimator of f obtained

from a kernel function K .

 

.1 n
That is, F (u) = — Z 1
n n i=1 (Xi : 11)
£ch = 1— ! Kkflw‘ < )
n ' a a n y
n n
= T: i K<x~xi)
n i=1 an
Th T — ' - 2 ’1 ' l/a
en, n — Tn(fn,Fn) — f fn (Fn (x))dH(x), n > 2

LWJ

is the proposed estimator for T(f), (3 < a <:% and fixed.

34

In the remainder of this note the integral f’°'

[p-a,1—n‘§]

will be denoted by f... ,

Thus, Tn = ‘f'fnz (Fn'lcx) 1 dH(x)

2.2 Consistency of the estimator.

Let the kernel K be such that

(3) K, is a symmetric probability density on R and is
of bounded variation u, flu|K(u)du < w .

(4) K(1) , the derivative cf K is a continuous
function of bounded variation pl,

(5) Window width an is such that

an + 0 and n&(log n)"é an + w as + w .

Theorem 1: Suppose that f has finite Fisher Information

 

and that the derivative f' of f is bounded. Suppose

that K satisfies (3) and (4) above. If né(log n)-1an-+<n

as n + m then Tn is (weakly) consistent for T .
To prove this theorem we need the follbwing lemma

from Schuster (1969).

Lemma(2.4 ,Schuster): If f and its first (r+1)

 

derivatives are bounded and if '{en} is a sequence of
positive numbers such that an = 0(en) then there exist

positive constants C1 and C2 such that

P(sup Ifn('r)(_x) -- f(r)(x)| > an) : Clexp‘{—C2nen2an2r+2}

X

35

Proof of Theorem:

 

Since T(f) is finite, given 5 > 0 select n

sufficiently large so that

T = I r2(r’1(;t))dH(.t) <e

21 .. _
[n '0‘ , 1-n ‘0‘]0
where [n'“, l—n-a]C is the comgflement of [n-a, l-n-a]

in [0,1] and 0<a<&.

Let T = T - T'
a2 a1

=.j'f2<F'1(jt))dH(:ti and

__ - 2 -1
let TM if f (Fn (t))dH(t).
Then ITn-TI _< ITn-Tnal + [Tnu-Tazl + To”.
Next it is shown that both IT -T, | and IT -T I
no: a2 n no:

tend to 0 in probability as n tends to w .

ITna—Ta2|._5 j' |f2(,Fn-1(t)) - f2(F"1(t))|dH(,t)

.I..IDZ(FFn-1(t)) - oz(.t)| dH(t) where p = fF-l

_ ' -1 -
_ .1 my“ (.t)-t)20(.t+6n)o'(t+6n)IdH(..t)
%1= enCt) ‘chosen according to M.V.T.
._<_ ZIIFFn-l - I:|_|_n.2°‘ [11f'(_t)|dt
o

, since [r'l is bounded

r __ 20
SFZK4HFFB'1- IHn
and h(Qx)<n20 in [n-a,1-n-°‘].

Using standard techniques (for e.g. seszoul, 1984)

36

it follows that

1

n’l’Il'F-Fn" - In ope)

Thus, IT .-Tuzl + O in probability as n + w provided<1<i.

no

IT'Tnal if 'lfzn (Fn‘lctn - f2(Fn"1(-t).)I_dH(.t).
‘- -1 . -1 . -1 _
:52! Ifn(Fn (x)} -f(Fn (.tmchn (t))dH(t)
+,I'<:rn (.Fn"1 (-ti). -. f(Fn"1(;t)_))2du(t)

1 2|[f||’_ an - f||_c (log n) + Cllfn- f“? (.109; n)

setting r = 0, en = a(log n)f1, where e > O, in

Lemma(2.4, Schuster) we see that Hfh - fH(log n) + O
2

+°° as 11"”.

in probability provided h(log nf2 n

This completes the proof of Theorem 1.

2.3 Asymptotic.normality.0fizTn'.

 

Theorem 2: In addition to above assumptions on K and

 

F suppose that

(6) ofllpi(t-s) - o'<t>ln(t>dH(t) + o as s + o
(7) 0f1f4(F'—1(t)){h(jt)}2dt. <03 < ... and

(8) f(x)h(F(_x)). <M. < no for some constant M .
Moreover, if ni(log nféan + w as n + u then

0% f Ifn2(.Fn'1(.t)) - f2(F-1(_t))|dll(t) converges in

distribution to a normal random variable with mean 0 and

. 2 .
variance 0 where

37
02 = 4011f4CF'1LtliIthllzdt + .4f{FI1)f' (F-1(.t)dH(t)}2f(.U)du
. (Lu

—8 n. '11 r1(r‘1(.t1.).dH<.,turacumm‘cu))du
F(n)

.4 ' {oflrzm “lat ).ldH.(;t ). 40111:“? <.F“1(-tl)d11(t)}2

“Proof: n5..].[fn2(Fn-1(.t)) — f2(F-1(t);dH(t)

 

(9) = n*..]f2[r,,(F,,'1<.t)) - f(.F"'1(.t)):f(.F‘1(.t))dnm

 

+ n*.:l'[fn(Fn‘1(.t)). - f<.F'1(t)):_2dII(.t) .
Consider the.second term
nU' En(.Fh.-1(.t)) - f(r’lcuflzdngt)
.f Zni fa En(Fn'1(t)) - f(Fn"1(t))}2dn(t)
+ 2n§A.ff{f(Fn-.-1(t)) — f(F_1(t))l Zena)
.5 Zné llfn- fll?‘ C (log n) + 2n*.I"{o(FFn'1(t)) - p (tnzdmt)

- 2
2n5 Ilfn- in2 0 (log n) + ZnfillFFn 1 _ I” .

IA

‘ .. ' ‘- 2
.Ho (we, s.n)| dH(t) 0<el<1

IA

1
2n5 0 Ilfn- f||'2(log n) + ZnéIIFFn—l - Illznzao'f Ip"(t)|2dt .
Use of Schuster's lemma with r = 0 shows that

nilfln - ffl2(log n) + O in probability

provided n§(log n)-1an:3 + a as n + w .

Since [1 Ip'(t)th is finite and since a < 1/4,
0
Znilan'1-]J[2 n2“ [llp'(t)|2dt + O in prob. as n + m.
o .

Thesetwo results show that the second term in (9) tends

to 0 in prob. as n tends to w .

38

Consider the first term in (9).

2n*.["{rn(rn'1<.t)) - “F’lmnuF‘lctnde

2n% I" Ifn(Fn‘1(.tn -- fn(F‘1<t>)}f(.F‘1(t))dH<t)
+2115 I‘Ifne‘lctn -- f(F‘lmmcF'lct))dH<t)

= In1 + In2 (saY)

2n*.I°f<F‘1<t)){onan'lctn - pnmndnm.

H
II

n1

where on = an-l

2n*.f f(F-ICt))(FFn-'1(t) -- t) -

I ’1 -
on (,t+62(FFn (t) - t>)dH<t).
O < 62< 1 .

Next we want to show that %; appearing in the above

expression can be approximated by d ,

First note that,

|2n%.f.f(F_1(t))(FFh-1(12)."-1}).

' {pn' (t+ez<FFn‘1(t>-t) - p'(t+62(FFn-1(t)-t))}d1{(t)l

1 Znélfi‘Fn-l- 1“ "(an o')o|| C (log n)

= 2.1% c “nu-1 - In. Ilrn' - runes n: .

39

Setting r 1 and En = (log n)e , where e > O ,

in Schuster's lemma we see that and- ffllclog n) + O .
. .. k —l

in prob. as n + w prov1ded n (log n) ‘an + w as n + w .
Also [2“; f f(F-4(t)(FFn'1(t)-t)

'{p' (-t+echFn‘1(t)- t) — q'(t)}dH(.t)|

-1

_< 2n§ IIIFFn

1 _
- III! pcth' (“92(an 1m - t)
0 ,
- o'<.t)ldn(.t)‘

Assumption (6) implies that this term tends to zero
as n + w .
These reductions now imply that we need to obtain the

normality of

., -N1
'10:) - 1:}f (F (U) dH(t)

2n; I.f(_F-1(t)) {FFn _1
' f(F (13))

 

+ 2ng.f.f(F-1(t))u{fn(.F_1(t)) - f(F"1(t))}dH(-t) .
s2n§ fawn-la) - t}f' (F—1(t))dll(t)

+2né of.f(_F-1(t)){fn(F-1(,t)) - f(F-1(_t))}dH(t) .

Note that

“F "1(t) — t = 5-{F F-J'FF “1(t) - FF ’1(t)'} + F F ‘1(t) — t

‘ n ' n n. ' ' n ' n n ' '
. -1 ~i

Since Illi‘nFn - III = Op(a) and

ni‘mnF’lFFn'ltt) - FFn'lun = n*(FnF'1(t> - t)

40

it suffices to consider
2n§;I f(F-ICI)){fn(F_1(I)) -f(F'1<t))}dHct)

-2né.,{.f'(F-1Ct)){FnF_'1(t) — menu) .

F 7.1 (.t, ).—x .

= 2n% - 1)dmt)

 

{ I f(F (t)) '- K(
1 ' an an

II M:

all-*1

i

OI°f'(F‘1(t))1(Xigfir'1(t))du<t)

_ j f2(F‘1(t))dH(t) + ['f'(F‘1(t))th(t)}

 

.1 n
= {fi' 121(Yin -‘ An) (_SaY) 0
Then Yin are i.i.d. for each n .
E Y A = I f.f(F-1(t))'$'Y(Fyl(t)-u)dH(t)f(u)du
in - n . ' “ an" an '

- .I f2(F-1(,t))dH(t)

+

I,I.f'(F-1(t))1(,U_fF'—1(.t))dH(.t)f(u)du'

+

.I'f'(F’1ct))th(t)

+ 0 as n + w—.

 

V—l .
CF :t)—u
n

2 1 ~ —1 ' 2
E Yln = flan! f(F (.t))K )dH(,t)} f(u)du

+ I'{ff'(F‘1(_.t))1(_u_5r"1(tndngtnzrgumu

. h -1 .37
(10) - 2 I % .f°f(F-1(tl)K(F gt)’“
n n

 

)dHCt)°

,ffif'(F-1(8)1(M5F'1(s))dHLs)}f(u)du .

 

41

From (8) it follows that

 

- o~ _. ..:“-,]_n - .
I; ,1 HF 1mm? gt) “)dmul :M ..
n n
. '1 . __i ' ngi,£5_fi . _ _2.
Since an I f(F (t))K(-——Ef—-)dH(t) e f (u)h(F(u))
. . n

by D.C.T. it is then clear that the first term 1n<fl9yconverges
to ] fs(x){h(F(X))}zdx .
Second and third terms clearly converge to
I{ {1 f'(F-1(t))dH(t)}zf(u)du and
F(U)

- 2 I{F{1)f'(p‘1(t))du(t)}£3(u)h(F(u))du respectively.
F,u

Since EY converges to

In

1 2 -1 ~ 1 -1
of f (F (t))dH(t) -0[ tf‘(F (t])dH(t)

. . . . . . 2
the limiting variance 15 given by 0

Remark: Assumption (9) is reasonable and can be verified

for the distribution for which the integral
I f2(;<)h(1=(x))d1=(x)

is finite. Normal, double exponential, logistic and Cauchy

distributions are some examples.

42

BIBLIOGRAPHY

Ahmad, I. A. and Lin, P. E. (1976): Nonparametric esti-
mation of a density functional. Florida State

University report, M369.

Anderson, T. W. and Darling, D. A. (1952): Asymptotic
theory of certain goodness of fit criteria based on

stochastic processes, Ann. Math. Stat. ;; 193-212.

Boos, Dennis D. (1981): Minimum distance estimator for
location and goodness of fit.
J, Amer. Stat. Assoc. Z§_ 663-670.

Cheng, K. F. and Serfling, R. J. (1981): On estimation of
a class of efficacy-related parameters.
"Scand. Actuarial g, 1981: 83-92.

Hajek, J. (1968): Asymptotic normality of simple linear
rank statistics under alternatives.
Ann. Math. Stat. 22 325-346.

Hettmensperger, T. P. (1984): Statistical inference based
on ranks. John—Wiley and Sons.

Jaeckel, L. A. (1972): Estimating regression coefficients
by minimizing the dispersion of the residuals.
Ann. Math. Stat. 51_ 1449-1458.

Koul, H. L. (1984): A monograph on weighted empiricals
and linear regression models. Dept. of Statistics
and Probability, Michigan State University.

Koul, H. L. and DeWet, T. (1983): Minimum distance estma-
tion in linear regression model.
Ann. 9; Stat..;; 921-932.

Koul, H. L., Sievers, G. L., and McKean, J.
(1987, to appear): An estimator of the scale para-
meter for the rank analysis of linear models under
general score functions.

43

Mugantseva, L. A. (1977): Testing normality in one dimen-
sional and multidimensional linear regressions.
Theory Prob. Appl. 33 591-602.

Scheweder, T. (1975): Window estimation of the asymptotic
variance of rank estimates of location.
Scand. g; Stat. g_ 113-126.

(1981): Correction note.
Scand. g; Stat._§_ 55-55.

Schuster, E. (1969): Estimation of probability density
function and its derivatives.

Ann. Mata, §L§§4 19 1187-1195.

A DIENCEPHALIC CONTRIBUTION TO SYMPATHETIC NERVE

DISCHARGE IN THE ANESTHETIZED CAT

BY

KURT JAMES VARNER

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree

DOCTOR OF PHILOSOPHY

Department of Pharmacology and Toxicology

1987

ABSTRACT

A DIENCEPHALIC CONTRIBUTION TO SYMPATHETIC NERVE
DISCHARGE IN THE ANESTHETIZED CAT

BY

Kurt James Varner

The purpose of this thesis was to study the role of diencephalic structures in
the control of sympathetic nerve discharge (SND) in the anesthetized cat. This
laboratory (Huang et al., Am. J. Physiol. 252: R645-R652, 1987) previously
showed that the forebrain is responsible for a significant component of inferior
cardiac SND in cats anesthetized with alpha-chloralose. My research assessed the
contribution of various diencephalic regions to the forebrain-dependent compo-
nent of SND in this preparation. The reductions in inferior cardiac SND and blood
pressure produced by midbrain transection in nonlesioned control cats were
compared with those in cats in which diencephalic lesions were made with radio-
frequency current. Lesions of the lateral hypothalamus, posterior medial hypo-
thalamus or medial thalamus, but not of the anterior medial hypothalamus,
attenuated the effects of midbrain transection. These results suggest that the
medial hypothalamus and medial thalamus contribute to SND in the anesthetized
cat.

Tungsten microelectrodes were used to record the spontaneous discharges of
individual neurons in the medial thalamus and hypothalamus to determine whether
these regions contained the cell bodies of neurons that influence basal SND.

Spike-triggered averaging of SND revealed the existence of two types of

Kurt James Varner

hypothalamic and medial thalamic neurons with sympathetic nerve-related acti-
vity in baroreceptor-innervated cats. The activity of type 1 neurons was
synchronized to an aperiodic spike-like component in SND whereas that of type 2.
neurons was synchronized to a 2-to 6-Hz rhythmic component. Microstimulation
at type 1 and type 2 unit recording sites increased SND. These results are
consistent with the possibility that diencephalic neurons contribute to the
rhythmic and aperiodic components of SND in the anesthetized cat.
Spike-triggered averaging of frontal-parietal cortical activity (EEG) re-
vealed that all medial thalamic and most hypothalamic neurons with activity
related to SND also had EEG~related activity. Thus I considered whether the
frontal-parietal cortex contributed to the forebrain—dependent component of SND.
Electrical stimulation of the cortex evoked excitatory sympathetic nerve re-
sponses and activated some type 1 and type 2 diencephalic neurons. However,
frontal lobotomy failed to attenuate the decreases in SND and blood pressure
produced by midbrain transection suggesting that these cortical regions do not

contribute to basal SND in the anesthetized cat.

To my grandmother, Margret Varner whose spirit of scholarship is embodied in
this work and to Laurie whose support and understanding were invaluable.

ii

ACKNOWLEDGEMENTS

I would like to offer my sincere thanks to my advisers Dr. Gerard Gebber
and Dr. Susan Barman, whose guidance, advice, patience and friendship have
allowed me to grow as both a scientist and a person.

I would also like to thank the members of my thesis committee, Dr. Glenn
Hatton, Dr. Gregory Fink, Dr. Irena Grofova and Dr. Robert McCall whose advice
and critiques have been very helpful in the preparation of this dissertation. I
would also like to thank Dr. Zhong-Sun Huang, for his advice and assistance in
performing the lesion and decerebration studies.

Finally, I would like to thank Diane Hummel for her secretarial assistance,

and William Cook and Ronald Winters for their technical help.

iii

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

LIST OF ABBREVIATIONS

INTRODUCTION

1.

Involvement of the cerebral cortex and limbic system in autonomic
cardiovascular function

A. Early evidence of autonomic regulation by the cortex

B. Cardiovascular responses elicited from the infralimbic cortex

Cardiovascular responses evoked from the thalamus

A. Cardiovascular responses elicited by electrical stimulation of
the thalamus

B. Cardiovascular responses evoked by chemical stimulation of
the thalamus

Hypothalamic regulation of the cardiovascular system
A. Cardiovascular responses elicited by electrical stimulation of
the hypothalamus
B. Chemical stimulation of the hypothalamus
C. The defense arousal system: possible source of sympathetic
tone
D. Models of hypertension involving diencephalic systems
1. Retain-dependent renal hypertension
2. Nonrenin-dependent renal hypertension
3. Neurogenic hypertension
4. Mineralocorticoid-salt hypertension
5. Spontaneous hypertension

Generation of rhythmic activity by the diencephalon and brain stem
A. The generation of basal SND
1. Early views on the origin of basal SND
2. Evidence that the brain stem is intrinsically capable of
rhythm generation
3. Proposed mechanisms of SND generation in the brain
stem
4. Possible brain stem regions involved in the generation of
SND

iv

Page
viii
ix

xii

~10)“

10
11

11
14

19
23
23

28
29
30
31
31
31
32
34

36

TABLE OF CONTENTS (continued)

B. Thalamic mechanisms for the generation of electrocortical
rhythms
1. Early models of thalamic generation of EEG rhythms
2. Mechanisms for the generation of alpha spindle activity
3. The thalamic reticular nucleus as a generator of tha-
lamic spindle sequences
4. Origin of the cortical delta and theta rhythms

METHODS AND PROCEDURES

A.

General procedures

1. Baroreceptor denervation

2. Nerve recordings

3. Electrical stimulation of the diencephalon and cortex
4. Histology

Data analysis

1. Quantification of SND and blood pressure

2. Crosscorrelation and autocorrelation analysis
3. Post-R wave interval analysis

4. Post-stimulus analysis '

5. Power spectral density analysis

6. Peri-spike-trigger averaging

7. Interspike interval analysis

8. Statistical analysis

Project A: Midbrain transection and radio-frequency lesions
1. Decerebration procedures

2. Radio-frequency diencephalic lesions

3. Data analysis

4. Histology

Project B: Multiunit diencephalic activity

1. Multiunit diencephalic recordings

2. Electrical and chemical stimulation of the diencephalon
3. Data analysis

Project C: Characterization of diencephalic neurons with sympa-
thetic nerve-related activity

1. Diencephalic unit recordings

2. Electrical stimulation

3. Efffect of baroreceptor reflex activation

4. Data analysis

Page

38
38
40

43
47

50

50
50
51
52
52

53
53

55
55
55

56

56
56
57
57
57

58
58
58
59

59
59

60
60

 

TABLE OF CONTENTS (continued)

F. Project D: Sympathetic responses elicited by cortical stimulation
1. Electrical stimulation of the cortex
2. Diencephalic unit recordings
3. Frontal lobotomy and decerebration

G. Drugs used
RESULTS

1. Identification of diencephalic regions contributing to SND
A. Midbrain transection in nonlesioned cats
B. Anterior medial hypothalamic lesions
C. Large medial hypothalamic lesions
D. Lateral hypothalamic lesions
E. Medial thalamic lesions

II. Synchronization of SND to diencephalic multiunit activity
A. Multiunit unit recordings of diencephalic activity
B. Patterns of relationship between SND and diencephalic activity
1. SND related only to medial thalamic activity
2. SND related to medial and lateral thalamic activity
3. SND related to hypothalamic activity
4. SND unrelated to thalamic activity

C. Electrical and chemical stimulation of the diencephalon
1. Sympathetic nerve responses elicited by electrical
stimulation
2. Intrathalamic injection of picrotoxin

III. Identification of diencephalic neurons with sympathetic nerve-
related activity
A. Identification of diencephalic neurons with sympathetic nerve
related activity
1. Diencephalic unit discharge patterns
2. Synchronization of diencephalic unit activity to inferior
cardiac SND
3. Comparison of unit firing times

B. Detailed analysis of diencephalic neurons with sympathetic
nerve-related activity
1. Detailed analysis of type 1 neurons
2. Detailed analysis of type 2 neurons
3. Electrical stimulation
4. Baroreceptor reflex activation

vi

Page

60
61
61
61

62
63
63

73
78
78

87
87
92
92
95
100
100

105

105
112

118

118
119

119
125

130
130
130
142
147

TABLE OF CONTENTS (continued)

Page
IV. ~Sympathetic nerve responses elicited by cortical stimulation 150
A. Sympathetic nerve responses elicited by cortical stimulation 150

B. Synaptic activation of type 1 and type 2 neurons by cortical
stimulation 15?

C. Effects of frontal lobotomy on the forebrain-dependent com-
ponent of SND 161
DISCUSSION 165
SUMMARY 181

BIBLIOGRAPHY 186

vii

 

Table

LIST OF TABLES

Page
Mean onset latencies of inferior cardiac sympathetic nerve exci-
tation produced by diencephalic stimulation with lO-ms trains of
three pulses of supramaximal intensity (1 mA) 107

Mean onset latencies of inferior cardiac sympathetic nerve re-
sponses produced by cortical stimulation with lO-ms trains of
three pusles or l-ms single shocks of supramaximal intensity (1
mA) 156

viii

£292:

10

11

LIST OF FIGURES

Representative histological sections showing lesions of lateral
hypothalamus (A), posterior hypothalamus (B) and medial tha-
lamus (C)

Comparison of effects of midbrain transection on inferior
cardiac sympathetic nerve discharge (SND) and mean blood
pressure (BP) in nonlesioned control cats with those in cats
with anterior medial hypothalamic or large medial hypothala-
mic lesions

Effect of midbrain transection on inferior cardiac (ICN) and
renal (RN) sympathetic nerve discharge (SND), and mean blood
pressure (BP) in nonlesioned baroreceptordenervated cat anes-
thetized with chloralose

Effects of midbrain transection at stereotaxic plane A3 (A)
and ganglionic blockade (B) on blood pressure (BP), inferior
cardiac (ICN) and renal (RN) sympathetic nerve discharge in a
nonlesioned baroreceptordenervated cat anesthetized with
chloralose

Representative lesion in anterior medial hypothalamus
Representative large lesion in medial hypothalamus
Comparison of effects of midbrain transection on inferior
cardiac sympathetic nerve discharge (SND) and mean blood
pressure (BP) in nonlesioned control cats with those in cats
with lateral hypothalamic lesions

Representative lateral hypothalamic lesion

Comparison of effects of midbrain transection on inferior
cardiac sympathetic nerve discharge (SND) and mean blood
pressure (BP) in nonlesioned contol cats with those in cats with
medial thalamic lesions

Representative medial thalamic lesion

Thalamic activity and sympathetic nerve discharge (SND) in a
baroreceptordenervated cat

ix

Page

65

67

69

71
74

76

79
81

83

85

88

LIST or FIGURES (continued)

£222:

12

13

14

15

16

17

18

19

20

21

22

23

24

Power spectra of arterial pulse wave (AP), sympathetic nerve
discharge (SND) and thalamic activity in two experiments

Selective relationship between medial thalamic activity and
sympathetic nerve discharge (SND)

Relationship between medial thalamic activity and sympathe-
tic nerve discharge (SND) in a decorticate and baroreceptor-
denervated cat

Distribution of diencephalic recording sites in experiments in
which sympathetic nerve discharge (SND) was related to
activity in medial but not in lateral thalamic nuclei

Nonselective relationship between thalamic activity and sym-
pathetic nerve discharge (SND)

Crosscorrelograms showing relationships of sympathetic nerve
discharge (SND) and activity recorded from thalamus and
hypothalamus

Sympathetic nerve responses elicited by diencepahlic stimula-
tion (lO-ms trains of three l-mA pulses applied once every 2 s)
in a CNS-intact (A) and decorticate (B) cat

Depth-threshold curves for sympathetic nerve responses eli-
cited by medial thalamic, hypothalamic and midbrain stimula-
tion with lO-ms trains of three pulses

Changes in sympathetic nerve discharge (SND) produced by
injection of picrotoxin into the medial thalamus

Changes in sympathetic nerve discharge (SND) and frontal-
parietal cortical activity (EEG) produced by the intravenous
administration of a convulsive dose of picrotoxin

Representative recordings of action potentials of diencepahlic
neurons with sympathetic nerve-related activity

Peri-spike-triggered averages of inferior cardiac sympathetic
nerve discharge (SND) for a type 1 hypothalamic and a type 2
hypothalamic neuron in the same cat

Anatomical distribution of diencephalic neurons with sympa-
thetic nerve—related activity

Page

90

93

96

98

101

103

108

110

113

116

120

123

126

LIST OF FIGURES (continued)

Figs
25

26
27

28
29

30

31

32

33

34

35

36

37

Firing times of type 1 (A,B) and type 2 (C, D) hypothalamic
and medial thalamic neurons relative to inferior cardiac
sympathetic nerve discharge (SND)

Type 1 medial thalamic neuron
Type 2 medial thalamic neuron with cardiac-related activity
Type 2 hypothalamic neuron with cardiac-related activity

Type 2 medial thalamic neuron without cardiac-related acti-
vity

Activity of a type 2 medial thalamic neuron synchronized to a
noncardiac-related component of SND

Comparison of changes in inferior cardiac sympathetic nerve
discharge (SND) accompanying diencephalic unit spike oc-
currence and an electrical stimulus applied at site of unit
recording

Peri-spike-triggered and peri-stimulus averages of inferior
cardiac sympathetic nerve discharge (SND)

Effect of baroreceptor reflex activation produced by aortic
obstruction on firing rate of two type 2 hypothalamic neurons

(LII)

Increases and decreases in inferior cardiac sympathetic nerve
discharge (SND) elicited by stimulation (IO-ms trains of three
pulses applied once every two 3) of the motor cortex

Sympathetic nerve responses elicited by stimulation (IO-ms
trians of three l-mA pulses applied once every 2 s) of
prefrontal, frontal (A) and infralimbic (B) cortex

Synaptic activation of a type 1 hypothalamic neuron by single
shock stimulation (1 mA; 1-ms duration delivered once every 2
s) of the motor cortex

Effect of bilateral frontal lobotomy (panels A and B) and
midbrain transection (panels C and D) on inferior cardiac
sympathetic nerve discharge (SND) and mean blood pressure
(BP) in seven baroreceptor-denervated cats anesthetized with
chloralose

xi

Page
128
131

133

135

138

140

143

145

148

151

154

159

162

Diencephalic structures:

aHd

Am
AV
CA
CC
Cd
CI
Ch
CL
CM
En
Fil
Fx
Ha
HbL
HL
Hp
va
H1,H2
1AM
LD
LP
MD
MFB
Mm
NCM
Nva
NR
Pc
Ped
Pf
PVH
R
RE
Rb
RPO
TO
VA
VI.
VM
VPL
VPM
21

LIST OF ABBREVIATIONS

area hypothalamica A.S.
dorsallis ASC
N. anterior medialis C.G.
N. anterior ventralis Co.S.
commissura anterior Cr.S.
corpus callosum E.S.G.
N. caudatus F.G.
capsula interna IL
chiasma optica MI

N. centralis lateralis M.G.
N. centrum medianum 0.3.
N. entopeduncularis O.G.
N. filiformis SI
fornix SII
hypothalamus anterior S.G.
N. habenularis lateralis S.S.G.

hypothalamus lateralis
hypothalamus posterior
hypothalamus ventromedialis
Forel's fields

N. interanteriormedialis

N. lateralis dorsalis

N. lateralis posterior

N. medialis dorsalis

medial forebrain bundle
corpus mammillare

N. centralis medialis

N. hypothalami ventromedialis
N. ruber

N. paracentralis

pedimculus cerebralis

N. parafascularis

N. paraventricularis hypothalami
N. reticularis thalami

N. reuniens

N. rhomboidens

Regio preOptica

tractus opticus

N. ventralis anterior

N. ventralis lateralis

N. ventralis medialis

N. ventralis posterolateralis
N.ventralis posteromedialis
zona incerta

xii

Cortical abbreviations

ansate sulcus
association cortex
coronal gyrus

coronal sulcus

cruciate sulcus
ectosylvian gyrus
fornicatus gyrus
infralimbic cortex
primary motor cortex
marginal gyrus
olfactory bulb

orbital gyrus

primary sensory cortex
secondary sensory cortex
sigmoid gyrus
suprasylvian sulcus

LIST OF ABBREVIATIONS (continued):

Other abbreviations:

EPSP excitatory post synaptic potential
IPSP inhibitory post synaptic potential
IML intermediolateral nucleus

ISI interspike interval

LTF medullary lateral tegmental field
VLM ventrolateral medulla

SHR spontaneuosly hypertensive rat
SND sympathetic nerve discharge

EEG electroencephalogram

xiii

INTRODUCTION

In the anesthetized animal the sympathetic nervous system maintains a level of
basal activity. The first studies to characterize sympathetic activity were
performed by Adrian a_t_ _a_l. (1932) and Bronk a; a_l. (1936) who demonstrated that
in most cases the activity in sympathetic nerve bundles is synchronized into bursts
which are locked in a 1:1 relationship to the cardiac cycle. In addition, SND
contains a respiratory-related periodicity since the amplitude of these cardiac-
related bursts waxes and wanes in relation to the respiratory cycle.

That supraspinal networks generate basal sympathetic tone to the heart and
vasculature was recognized when Bernard (1863) produced a sustained decrease in
blood pressure by transecting the cervical spinal cord. Subsequent studies by
Owsjannikow (1871) and Dittmar (1873) attempted to localize the brain regions
responsible for generating sympathetic tone by monitoring changes in blood
pressure in response to serial transection of the neuraxis in rabbits. In these
experiments, blood pressure was not affected until the level of transection was at
a point 1-2 mm behind the caudal border of the inferior colliculi. On this basis it
was concluded that, at least in the anesthetized animal, the forebrain is not
involved in maintaining the level of resting blood pressure. Transections made
more caudally produced further decreases in blood pressure. When the level of
transection was 4-5 mm above the calamus scriptorius, blood pressure fell to a
level similar to that produced by transection of the cervical spinal cord. Thus it
appeared that the region of the neuraxis responsible for the generation of basal

sympathetic tone lies between the caudal one third of the pans and the rostral two

thirds of the medulla. Alexander (1946) confirmed this by monitoring the effects
of brain stem transection on the level of tonic activity in the cervical and inferior
cardiac sympathetic nerves of cats.

Although it is widely accepted that sympathetic nerve discharge (SND) is
generated in the brain stem of the anesthetized animal, there are several (often
ignored) observations which support the view that the forebrain contributes to
SND in this preparation. First, SND can be locked to a rhythmic 2- to 6-Hz
component of frontal-parietal cortical activity (EEG) in chloralose- (Barman and
Gebber, 1980) or diallylbarbituratedurethane- (Gebber and Barman, 1981) anesthe-
tized cats and chloralose-anesthetized dogs (Camerer a; a}, 1977). Furthermore,
the spontaneous discharges of individual medullary neurons in the cat (Gebber and
Barman, 1981) are temporally related to rhythmic 2- to 6-Hz components in both
SND and cortical activity. Langhorst at a_l. (1975) identified medullary neurons
whose activity is related to a 2- to 6-Hz rhythm in cortical activity and to the
arterial pulse. Second, several laboratories have reported that midcollicular
decerebration in anesthetized animals produces a 30-40 mmHg decrease in blood
pressure (Wang and Chai, 1962; Reis and Cuenod, 1964; Peiss, 1965; Huang at_ 94.,
1987) and a 381796 reduction in postganglionic inferior cardiac SND (Huang g a”
1987). Taken together these data raise the possibility that the forebrain
participates to a significant extent in setting the level of SND in the anesthetized
animal. Moreover, Huang 9.! a1. (1987) demonstrated that the effects of
decerebration on blood pressure and SND are prevented by removal of the medial
diencephalon prior to transection. Therefore, the purpose of this thesis was to
examine the role of the diencephalon in the control of basal SND in the

anesthetized cat.

To place this study in the prOper perspective, I will first review classic and
current literature related to the involvement of forebrain structures in sympathe-
tic ftmction with special emphasis on their role in cardiovascular regulation.
Moreover, since a distinguishing feature of basal SND is its rhythmicity, I will also
review the proposed mechanisms by which SND is thought to be generated in the
brain stem, as well as mechanisms by which the forebrain is thought to generate
rhythmic activity patterns. These reviews are not intended to be global in scope.
Rather their intent is to orient the reader in terms of the rationale and

conclusions of this study.

L Involvement of the Cerebral Cortex and Limbic System in Autonomic
Cardiovascular Function

A. Early evidence of autonomic regulation by the cortex

An extensive review of the early literature regarding cortical regula-
tion of autonomic function has been provided by Hoff a; a_l. (1963). In 1869
Hughlings Jackson first proposed that the ”cortex controls not only voluntary
movements of the body, but the movements of the arteries and viscera as well."
Jackson based his conclusions not on experimental evidence, but on the clinical
observation that somatic and visceral responses accompany epileptic seizures.
Early experimental studies of cortical involvement in cardiovascular function used
electrical stimulation to elicit changes in heart rate and/or blood pressure. In the
mid 1870's, Schiff, Danilewsky and Bochfeontain working independently, reported
that stimulation of the suprasylvian sulcus and the anterior and posterior sigmoid
gyri elicits primarily increases in blood pressure. Changes in heart rate (increases
and decreases) often accompanied the blood pressure responses; however, no
consistent relationship between the direction of the blood pressure response and
the heart rate changes was observed. In 1866, Bechterew and Mislawski reported

that stimulation of the globus pallidus and internal capsule elicits increases in

blood pressure which persist after the elimination of descending motor fibers by
lesioning the motor cortex. These results suggest that motor and autonomic
responses are mediated over separate pathways.

Hoff and Green (1936) used electrical stimulation to map cortical
pressor and depressor sites in ether-anesthetized cats and monkeys. In cats,
pressor responses are consistently elicited from the posterior and anterior sigmoid
gyrus and gyrus proreus (motor and premotor areas). In most cases, a 5 to 10%
increase in heart rate accompanied pressor responses; however, decreases in heart
rate are observed in some cats in which the stellate ganglion and adrenal glands
were removed suggesting that these effects are vagally mediated. Depressor
responses are consistently elicited from the lateral median, medial suprasylvian,
ectosylvian and sylvian gyri. In the monkey, pressor responses are produced by
stimulating areas adjacent to the superior paracentral sulcus (motor and premotor
areas). Depressor responses are occasionally produced by stimulating the margi-
nal gyrus and the anterior tip of the superior paracentral sulcus. No vasoreactive
sites were fmmd in the parietal or occipital cortex. In cats and monkeys pressor
and depressor responses are not the result of current spread to subcortical
structures, since they are abolished by either undercutting or applying local
anesthetic to the cortex. In addition, pressor and depressor responses are
obtained from sites separated by 2—4 mm. Vasomotor responses could often be
elicited in the absence of skeletal muscle movement.

In a subsequent paper, Green and Hoff (1937) compared the vasomotor
responses and changes in blood flow evoked by cortical stimulation in cats and
monkeys anesthetized with ether or diallylbarbiturate. In contrast to the pressor
responses that are elicited by stimulating the motor and premotor cortex in cats
and monkeys anesthetized with ether, stimulation of these same cortical areas in

Dial-anesthetized animals elicits depressor responses or no change in pressure.

Pressor responses elicited from the motor and premotor areas are associated with
an increase m blood flow to the limbs and a decrease in kidney blood flow. This
was true for both cats and monkeys regardless of the anesthetic used. Depressor
responses were often accompanied by a decrease in limb blood flow and an
increase in renal blood flow. Blood flow changes are considered to be active
rather that passive for several reasons. 1) Dilation of the limbs occurs in the
absence of a pressor response. 2) Changes in blood flow persist after removal of
the adrenal glands. 3) The changes in limb blood flow are abolished by
denervation of the limb. 4) Changes in blood flow are observed in the absence of
muscle movement. It was concluded that a mechanism exists whereby the motor
and premotor cortical areas influence the distribution of blood to various body
regions to favor circulation in active muscles.

The possibility that the variability of cardiovascular responses was due
to experimental variables such as type and depth of anesthesia was examined by
Delgado (1960) who studied cortically-evoked cardiovascular responses in con-
scious unrestrained cats, monkeys and humans. In the cat, stimulation of the
metor cortex including the hidden motor cortex usually increases blood pressure,
although depressor responses are sometimes observed. Pressor and depressor sites
are randomly distributed in the motor cortex. Stimulation produces variable
effects on heart rate. Cardiovascular responses occur in the absence of cortical
seizures and seizure activity is seen without accompanying cardiovascular effects,
indicating that the cardiovascular effects are not due to generalized seizure
activity.

Eliasson at_ _a_l. (1952) elicited atrOpine-sensitive vasodilator responses
from the cruciate sulcus in dogs. Associated with the vasodilator response is
either an increase or a decrease in blood pressure. By using electrical stimulation

Eliasson 2! a1. traced a pathway from the cortex to a region in the hypothalamus

located near the midline 1-2 mm above the Optic chiasma. Vasodilator responses
elicited from this hypothalamic area are also atropine-sensitive and are observed
in hind limbs denervated except for their sympathetic innervation. These results
are in agreement with those of Hsu a; a_l. (1942) who also reported that active
vasodilation is elicited by ' stimulation of the motor cortex in chloralose-anesthe-
tized dogs.

Clark 3 9.1- (1968) examined the blood flow changes elicited by
stimulation of the motor cortex in chloralose-anesthetized monkeys and baboons.
They located sites in area 4 from which discrete movement and an increase in
blood flow are elicited in the contralateral limb. Stimulation produced small (5-
10 mmHg) or no change in blood pressure. Blood flow responses are not altered
after preventing muscle movement by administering decamethonium bromide or
surgically transecting the spinal cord at the third or fourth lumbar segment.
Similar responses are also observed in skinned limbs, suggesting that the changes
in blood flow involve changes in muscle perfusion. Stimulus-induced increases in
blood flow are not affected by atropine, suggesting that the changes in flow result
not from the activation of a sympathetic cholinergic vasodilator system, but
rather from the inhibition of sympathetic vasoconstrictor tone. Alternatively,
these effects may reflect the activation of a noncholinergic vasodilator system.
From these results Clarke 3 a_l. suggested that somatomotor and vasomotor
representation of the limbs in the cortex are topographically co-localized.

Hilton a_t_ a_l. (1979), working m althesin (alphaxalone-alphadalone)-
anesthetized cats contradicted the conclusions of Clarke et al. Although muscle
contraction and vasodilation in the contralateral hind limb accompanied stimula-
tion of the motor cortex in the cat, they noted parallel reductions in muscle
contraction and vasodilation after lesioning the contralateral pyramidal tract or

administering gallamine. During recovery from gallamine contractile and blood

flow responses increase in parallel. Vasodilation and muscle contraction (as
verified by electromyographic recordings) are prevented by lumbar cordectomy.
The lack of vasodilation after lumbar cordectomy is not due to disruption of the
sympathetic out flow since an atropine-sensitive vasodilation of the hind limb
muscles is elicited by stimulation of the hypothalamic defense region. They
concluded that the vasodilation elicited by stimulation of the motor cortex is due
to a post-contraction hyperemia, rather than the activation of a cortical
sympathetic vasodilator system.
B. Cardiovascular responses elicited from the infralimbic cortex

The infralimbic cortex (including the anterior cingulate gyrus and
subcallosal cortex) is thought to control autonomic and behavioral variables such
as respiration, gastric motility, blood pressure, pupillary size, attention and
searching. (cf. Kaada, 1960; Lofving, 1961).

Smith (1945) was the first to demonstrate that cardiovascular re-
sponses can be elicited by stimulating the anterior portions of the cingulate gyrus
in monkeys anesthetized with ether. Three types of cardiovascular responses are
observed: 1) a marked increase in blood pressure (20-30 mmHg) without a change
in heart rate, 2) a decrease in arterial pressure and heart rate, 3) a fall in blood
pressure and skipping of cardiac beats. The bradycardia was presumed to be
vagally mediated since it is accentuated by administering the anticholinesterase
eserine. Similar cardiovascular responses are observed in conscious cats and
monkeys (Anand and Dua, 1956).

Lofving (1961), in a very detailed study in the cat, identified pressor
and depressor sites in the medial portions of the rostral cingulate gyrus. The
strongest responses occur upon stimulation of the area around the genu of the
corpus callosum. While some overlap is noted, vasoactive sites are topographical-

ly arranged with depressor sites in the dorsal and pressor sites in the ventral

portions of this region. The pressor responses are associated with a pronounced
decrease in blood flow to skeletal muscles and a moderate decrease in flow to the
intestines. No change in renal blood flow is observed. These responses are
mediated via the activation of vasoconstrictor fibers since they are abolished by
sectioning the white rami (i.e., preganglionic sympathetic fibers). Stimulation of
depressor sites in atropinized, curarized-cats elicits increases in blood flow to
skeletal muscles (most pmnomced), skin and intestines with no change in renal
perfusion. Stimulation of depressor sites in animals having intact vagi reduces
heart rate by up to 40%. However, after sectioning the vagi, heart rate is still
reduced by up to 10% suggesting that the bradycardic response is in part due to
inhibition of sympathetic tone. Electrolytic lesions of the anterior hypothalamic
depressor area abolish the depressor responses suggesting that the subcallosal

depressor area mediates its effects via the anterior hypothalamic depressor area.

II. Cardiovascular Responses Evoked from the Thalamus

A. Cardiovascular responses elicited by electrical stimulation of the
thalamus

Early investigations of thalamic influences on cardiovascular function
monitored blood pressure responses elicited by electrical stimulation (cf. Kabat _e_t_
a_l., 1935). hi 1911 Sachs used the Horsley-Clark stereotaxic instrument and found
that electrical stimulation of the n. retmiens and n. ventralis increases blood
pressure, whereas stimulation of the anterior thalamic nuclei occasionally elicits
depressor responses. In contrast, Kabat a; a_l. (1935) reported that electrical
stimulation of the thalamus in pentobarbital-anesthetized cats fails to change
blood pressure. The only exception is an area along the course of the periventri-
cular fiber tract medial to the habenulopeduncular tract from which pressor and

depressor responses are occasionally elicited.

Stimulation of the medial and intralaminar thalamic nuclei increases
blood pressure and decreases renal blood flow in chloralose-anesthetized dogs
(Takeuchi at al., 1960; 1962) and nembutal-anesthetized rabbits (Jurf and Blake,
1972). The increase in renal vascular resistance was both neural and humoral in
origin as determined in cross perfusion studies (removing humoral influences) and
via direct measurement of adrenal catecholamine release. Pressor responses are
accompanied by a reflex vagal bradycardia. Depressor responses are obtained
from n. ventralis pars medilais and n. lateralis posterior and are most often
accompanied by decreases in renal blood flow with no change in heart rate.
Stimulation of sites in ventral and lateral thalamic nuclei usually produce no
change in either blood pressure or renal blood flow.

Angyan (1978) examined the cardiovascular effects associated with
stimulation of rewarding sites the mediodorsal and anteromedial thalamic nuclei
in chronically instrumented cats. Stimulation at these sites was associated with
increases in blood pressure, respiratory rate, heart rate and a decrease in pulse
pressure. Contrary to studies mentioned above, no depressor sites were located in
the thalamus. Interestingly, stimulation of nonrewarding thalamic sites failed to
evoke cardiovascular responses.

Most recently, Buchanan and Powell (1987) examined the cardiovascu-
lar responses elicited by electrically stimulating the anteromedial and medio-
dorsal thalamic nuclei in conscious rabbits. Stimulation of the anteromedial
nucleus evoked an increase in blood pressure and a biphasic cardiac response (a
brief acceleration followed by a larger deceleration). Phentolamine abolished
both the bradycardia and pressor responses suggesting that the bradycardia results
secondarily from baroreceptor reflex activation. Administration of atropine
abolished the bradycardia suggesting that it is vagally mediated. Stimulation of

the mediodorsal nucleus elicited pressor responses and bradycardia. Phentolamine

10

administration prevents the pressor response but does not affect the bradycardia.
Propranolol attenuates the decrease in heart rate and in combination with
atropine abolishes it, suggesting that the bradycardia is in part vagally mediated
and in part due to sympathoinhibition. Like Angyan (1978), these authors failed to
locate depressor points in the medial thalamic nuclei. No cardiovascular
responses were produced by stimulating the lateral thalamic nuclei.

B. Cardiovascular responses evoked by chemical stimulation of the thala-
mus

An inherent problem in the use of electrical stimulation is its
nonselective activation of cell bodies and axons of passage. For this reason many
investigators have used chemical substances to activate selectively cell bodies
and/or synaptic networks.

Rockhold a; a_l. (1985) reported that heart rate and blood pressure
increase after the injection of the excitatory neurotoxins kainic acid and N-
methyl-D-aspartic acid into the n. reuniens of the thalamus in urethane anesthe-
tized rats. On this basis, it was concluded that the n. reuniens contains cell
bodies having excitatory efferent projections to areas regulating cardiovascular
functions. Due to the proximity of the n. reuniens to the ventricular system and
hypothalamus the possibility must be considered that these cardiovascular effects
reflect the activation of other diencephalic structures. In this regard, these
authors have reported similar cardiovascular responses following the injection of
these substances into the hypothalamus (Rockhold g al., 1987).

A similar study was performed in conscious rabbits by Powell and
Buchanan (1986). Small volumes of glutamate, Gamma aminobutyric acid (GABA)
and carbachol were injected into the n. medialis dorsalis via indwelling cannulae.
Glutamate and carbachol, but not GABA, produced dose related (nmol to mmol
range) increases in heart rate and blood pressure. They noted similar responses

during electrical stimulation of the same region (Buchanan and Powell, in press).

11

Injections into the anteroventral thalamus or reticular nucleus were without
effect. Injection of equal concentrations of glutamate and carbachol into the
lateral ventricles decreases heart rate and blood pressure, suggesting that the
effects of the intrathalamic injection of these substances are not the result of the

spread of the drugs to the ventricular system.

III. Hypothalamic Regulation of the Cardiovascular System

Traditionally the hypothalamus is thought to be involved in the integration
of a wide variety of autonomic and behavioral responses. Several investigators
have even raised the possibility that this area is tonically involved in the control
of the cardiovascular system.

A. Cardiovascular responses elicited by electrical stimulation of the
hypothalamus

Karplus and Kreidl in a series of papers from 1907-1927 first demon-
strated that a variety of autonomic responses are elicited by electrical stimula-
tion of the hypothalamus (Kabat 3 a_1., 193 5). These investigators showed that
electrical stimulation of the ventral surface of the hypothalamus produces
increases in blood pressure, bladder contraction, rate and depth of respiration and
alterations hi gastrointestinal motility. Also observed are increases in the
secretion of sweat, tears and saliva. Cardiovascular responses are not altered by
administering curare, by adrenalectomy or by hypophysectomy; however, section-
ing the renal nerves does attenuate the pressor response.

In the mid 1930's Ranson, Magoun and coworkers (Ranson SE al., 1935;
Kabat _e_t al_., 1935) used the Horsley-Clark stereotaxic instrument to systemati-
cally explore the hypothalamus for vasoactive sites in curarized nembutal-
anesthetized cats. Pressor responses resulted from stimulation of the lateral
hypothalamus, medial forebrain bundle, anterior commissures, periventricular

fibers, posterior hypothalamic nucleus, H1, H2 Forel's fields or the perifornical

12.

region. Kabat and coworkers were unable to elicit pressor responses by
stimulating regions rostral or dorsal to the hypothalamus and therefore concluded
that the hypothalamic pressor responses do not involve the activation of pressor
fibers entering the hypothalamus from the cerebral cortex, corpus striatum or
thalamus. A few depressor sites were found in more caudal and lateral
hypothalamic regions. Stimulation of the region surrounding the anterior commis-
sure decreases the rate and depth of respiration, increases bladder contraction
and often decreases blood pressure (10-20 mmHg). Evidence that the anterior
hypothalamus contains a parasympathetic system was provided by Korteweg at al_.
(1959) who showed that vagotomy prevents the decreases in heart rate and blood
pressure produced by stimulating the anterior hypothalamus.

Wang and Ranson (1941) reported that decreases in heart rate (6-19%)
and depressor responses (18-30 mmHg) are elicited by stimulating the preoptic
area in chloralose-anesthetized cats. Vagotomy does not alter the depressor
response and only partially attenuates the bradycardia. Both vagal and sympatho-
inhibitory mechanisms were postulated to account for these responses.

Morison and Rioch (1937) elicited pressor responses from the hypotha-
lamus in chronic decorticate cats (lacking corticofugal fibers). Pressor responses
are elicited by stimulating the anterior hypothalamus, rostral periventricular
system, caudal septum and the olfactory tubercule. The strongest pressor
responses are elicited from the posterior hypothalamus. In contrast, depressor
responses are elicited from the rostral pole of the amygdala, rostrobasal septum
and rostral wall of the third ventricle. On the basis of evidence from ablation
experiments, it was concluded that a sympathoinhibitory mechanism is located in
the anterior hypothalamus. In contrast to others (Ranson a; g” 1935; Kortweg at_
a_l., 1959), Morison and Rioch did not find evidence of a parasympathetic system in

the anterior hypothalamus.

13

The presence of a sympathoinhibitory system in the anterior hypotha-
lamus of the anesthetized cat was confirmed by Folkow a; _a_l. (1959). Stimulation
of a circumscribed area located just caudal and ventral to the anterior commis-
sure decreases blood pressure (105 mmHg) and heart rate (20%) and increases
blood flow to skeletal muscle, skin and visceral beds. The depressor response is
not blocked by atropine, but is prevented by adrenergic antagonists. After
regional sympathectomy, stimulation of this area elicits parallel decreases in
blood pressure and blood flow to skin, muscle and the viscera. The bradycardia
likely reflects the inhibition of cardio-accelerator fibers in these vagotomized
animals. The authors concluded that stimulation of this anterior hypothalamic
site may produce a generalized inhibition of sympathetic tone by activating a
relay center for cortical sympathoinhibitory systems.

Hilton and Spyer (1 97 1) reported that lesioning the anterior hypothala-
mus in cats attenuates the cardiovascular responses produced by baroreceptor
afferent stimulation, leading them to suggest that the anterior hypothalamus is
involved in integrating the cardiovascular response to baroreceptor reflex activa-
tion. Furthermore, stimulation of this hypothalamic region located ventral and
caudal to the anterior commissure and extending in the dorsal hypothalamus
dorsal to the the fornix, elicits a pattern of cardiovascular effects resembling
that produced by baroreceptor reflex activation. These effects consisted of
decreases in blood pressure (30-50 mmHg), heart rate (25%) and vascular
resistance. Respiratory rate was also decreased. The depressor response is due to
sympathoinhibition, since it is potentiated by carotid artery occlusion and
abolished by administering guanethedine. The bradycardic response is prevented

by sectioning the vagi.

14

Reis and Nathan (1973) showed that bilateral electrolytic lesion in the
anterior hypothalamus of the rat increases arterial pressure. Within five to seven
hours the rats died from pulmonary edema. The hypertension is associated with a
50% decrease in cardiac output and is abolished by administering phentolamine.
The lesion-induced hypertension apparently results from a sympathetically-
mediated increase in total peripheral resistance.

Manning and Peiss (1960) examined the cardiac effects produced by
stimulating pressor areas in the lateral and posterior hypothalamus in vagoto-
mized cats. Three basic types of short latency (1-4 s) cardiac and vascular
responses are observed: 1) an essentially pure sympathetic cardioaccelerator
response without an increase in blood pressure, 2) a vasoconstrictor responses with
equal increases in systolic and diastolic pressure, 3) an augmentor response
represented primarily as an increase in systolic pressure due to an increase in
myocardial contractile force. Augmentor responses are observed without an
accompanying increase in heart rate. These three types of cardiac responses are
observed alone or more commonly in combination. These responses are presumed
to be sympathetically mediated owing to their short latency and abolition by
stellate ganglionectomy. Cardiac augmentor and accelerator responses are
generally concentrated into two zones, one in the periventricular areas and nuclei
bounding the ventricular walls and the other located more lateral, extending into
the area of the medial forebrain bundle, the H1 and H2 Forel's fields and
extending into the subthalamic nuclei.

B. Chemical stimulation of the hypothalamus

Masserman (1937) compared the cardiovascular effects elicited by
electrical and chemical stimulation of the anterior hypothalamus in ether-
anesthetized cats. In contrast to studies discussed above, a 10-60 mmI-Ig rise in

blood pressure and reductions in heart and respiratory rates were produced by

15

electrically stimulating the anterior hypothalamus. The intrahypothalamic injec-
tion of sodium amytal produces a 25-60 mmHg decrease in blood pressure and
increases in heart rate, pulse pressure and respiratory rate. In contrast, the
injection of coramine elicits a slight increase in blood pressure and respiratory
rate.

More recently, Redgate and Gellhorn in a series of papers (Gellhorn
and Redgate, 1955; Redgate and Gellhorn, 1956a; 1956b;) examined the cardio-
vascular effects occurring after the intrahypothalamic injection of barbiturates or
local anesthetics in chloralose-anesthetized cats. The injection of pentothal,
nembutal or procaine into the posterior hypothalamus produces a 10-80 mmHg fall
in blood pressure. The fall in blood pressure is usually accompanied by a decrease
in heart rate. In addition, the action of hypotensive drugs such as acetylcholine or
histamine are potentiated following the injection of barbiturates into the pos-
terior hypothalamus. On this basis they concluded that the posterior hypothala-
mus exerts a strong tonic excitatory sympathetic drive. Injection of pentothal or
nembutal into the anterior hypothalamus between the optic chiasma and the
anterior commissure increases blood pressure and heart rate. These responses are
smaller in magnitude than those produced by posterior hypothalamic injections.
The injection of barbiturates into the anterior hypothalamus reduces the reflexly-
induced bradycardia occurring after the intravenous injection of norepinephrine.
In addition, the magnitude of the pressor response to a given dose of norepine-
phrine is greater following the injection of barbiturate into the anterior hypotha-
lamus. They suggested that the anterior hypothalamus exerts a tonic parasympa-
thetic influence. The results and conclusions of Redgate and Gellhorn must be
viewed with a degree of caution in view of the fact that they used injection

volumes of up to 40 ul/dose in these experiments.

16

hi 1964, Gellhorn demonstrated that electrolytic lesions of the pos-
terior hypothalamus produce decreases in blood pressure and heart rate similar to
those occurring after the intrahypothalamic injection of barbiturates into these
same regions. These posterior hypothalamic lesions destroyed n. mammillaris, n.
ventrolateralis, n. lateralis and n. posterior. In contrast, the injection of metrazol
or strychine into the posterior hypothalamus increases blood pressure and heart
rate. Lesion of the anterior hypothalamus produces even greater increases in
blood pressure and heart rate than does the injection of barbiturates into the
anterior hypothalamus. The anterior lesions destroyed n. preopticus, :1. supra-
opticus, n. superchiasmaticus, and n. hypothalamus anterior.

Lee 5 al. (1972) studied the cardiovascular effects elicited by the
intrahypothalamic injection of the GABA antagonist picrotoxin in the cat. The
injection of picrotoxin (30-40 ug) into either the lateral or posterior hypothalamus
increases blood pressure (64 mmHg) and heart rate (30 beats/min) and produces
cardiac arrhythmias similar to those seen during electrical stimulation of the
same sites. Unlike intravenous administration, the intrahypothalamic injection of
picrotoxin does not produce a reflex bradycardia even though blood pressure often
rises above 250 mmHg. These effects are similar to those elicited by the
injection of picrotoxin into medullary pressor sites and are thought to reflect a
hypothalamic-mediated inhibition of the vagal reflex. The ability of picrotoxin
injections to increase blood pressure indicates that sympathoexcitatory systems in
the .lateral and posterior hypothalamus are under tonic GABAergic inhibitory
influences. Unfortimately, no attempt was made to determine whether these
sympathoexcitatory systems are tonically active by injecting a GABA agonist.
The injection of picrotoxin into depressor sites (as determined by electrical
stimulation) in the anterior hypothalamus produces only slight increases in blood

pressure (14 mmHg) and heart rate.

l7

Dimicco and colleagues (Williford _e_t g, 1980; Schmidt and Dimicco,
1984) identified a sympathoexcitatory system in the periventricular hypothalamus
of the cat which is under GABAergic control. Bicuculline (8-32 ug in 100 nl)
injected into the left lateral cerebral ventricle was confined to the lateral and
third ventricles by cannulating the aqueduct of Sylvius. Within one minute of
bicuculline injection, blood pressure, heart rate and hind limb vascular resistance
increase. The vagally-induced bradycardia normally accompanying the pressor
response is prevented. The intraventricular injection of muscimol (10 ug in 100
nl) during the peak bicuculline-induced cardiovascular response abruptly reverses
these effects. Following muscimol, heart rate and blood pressure are at slightly
less than control levels. These findings support the suggestion of Redgate and
Gellhorn that the hypothalamus exerts a tonic sympathoexcitatory drive. How-
ever, this is not directly tested by administering muscimol alone. These authors
also failed to consider the possibility that these effects reflect an action of
bicuculline in the thalamus or other diencephalic structures.

A pattern of cardiovascular effects similar to that reported by
Dimicco and colleagues was seen after injecting bicuculline or the GABA
synthesis inhibitors isoniazid and 3-mercaptopropionic acid into the posterior
hypothalamus of the rat (Finch and Hicks, 1977). The increases in arterial
pressure, respiratory rate and tachycardia produced by the GABA antagonist and
synthesis inhibitors are reversed by intrahypothalamic muscimol. The injection of
muscimol alone fails to change baseline blood pressure or heart rate. It is
concluded that unlike in the cat, the GABAergic mechanism in the anesthetized
rat is sufficiently active to suppress tonic activity in the sympathoexcitatory
hypothalamic system activated by GABA antagonists.

Histamine (0.01-10 ug/ul) injected into the posterior or anterior

hypothalamus in urethane anesthetized rats also increases blood pressure and

18

heart rate (Finch and Hicks, 1977). These effects are thought to be mediated by
Hi receptors since they are attenuated by the H1 antagonist mepyramine. Similar
volumes of histamine injected into the ventromedial hypothalamus, lateral hypo-
thalamus, preoptic area or medial thalamus fails to elicit changes in heart rate or
blood pressure suggesting that the cardiovascular effects elicited by histamine
involve the activation of systems in the posterior and anterior hypothalamus.

Rockhold _e_t_ al. (1987) showed that the injection of kainic acid and N-
methyl-D-aspartic acid (30-1000 ng dissolved in 100 nl saline) into the area of the
paraventricular nucleus (PVH) in rats increases heart rate and blood pressure and
causes mydriasis, exOpthalmos and increases in respiratory rate. Injection of
these agents into the lateral hypothalamus and lateral cerebral ventricle produced
cardiovascular effects of similar magnitude when 1000 ng but not 30 ng was
administered. Pressor and tachycardic responses are reversed by phentolamine
and propranolol, respectively, suggesting sympathetic mediation of the cardio-
vascular responses. In contrast, injections of L-glutamate into this region
produces only minimal cardiovascular effects. No attempt was made to deter-
mine whether these regions tonically influence sympathetic tone.

8m and Guyenet (1986) elicited increases in arterial pressure and a
reflex bradycardia by injecting L-glutamate into the lateral hypothalamus of the
rat. In addition, the injection of glutamate increases the discharge rate of
medullospinal sympathoexcitatory neurons in the rostral ventrolateral medulla.
This hypothalamic projection to the ventrolateral medulla is thought to be
glutamatergic since the activation of the medullospinal neurons and pressor
response to hypothalamic stimulation were blocked by injecting the glutamate

antagonist, kynurenate into the ventrolateral medulla.

19

C. The defense arousal system: Possible source of sympathetic tone

Fear and rage reactions are associated with stereotypic behavioral and
visceral responses which prepare the organism for flight or attack. Cannon in the
1920's suggested that sympathetic activation during these reactions mimics the
effects of epinephrine (Hilton, 1982). That is, sympathetic activation increases
skeletal muscle _blood flow, while reducing flow to visceral structures, thus driving
blood out of the viscera and into skeletal muscles to prepare for the exertion of
attack or escape.

Hess and collaborators in the 1940's (cf. Hilton, 1982) used electrical
stimulation to map the regions from which the patterns of behavior characteristic
of alerting and ultimately attack or flight (increased stimulus intensity) are
elicited. Behavioral responses which completely mimic those elicited by natural
stimuli are evoked in conscious cats from the tuberal region of the hypothalamus
just medial, ventral and lateral to the fornix. Hess, collectively termed these
responses the defense reaction.

Abrahams gt fl. (1960) used electrical stimulation to map those
regions of the hypothalamus and midbrain from which the cardiovascular compo-
nents of the defense reaction are elicited in anesthetized cats. Stimulation of
reactive sites in the hypothalamus elicit a stereotypic pattern of cardiovascular
responses including increases in skeletal muscle blood flow, blood pressure and
heart rate. Also observed are vasoconstriction in the skin and intestine, pupillary
dilation, retraction of the nictating membrane, piloerection along the back and
tail and an increase in respiratory rate. Small doses of atropine abolish the
increase in skeletal muscle blood flow. The responsive zone consists of a narrow
strip running bilaterally along the length of the hypothalamus. In the tuberal
region the reactive area extends laterally giving rise to a highly reactive site just

dorsal to the cerebral peduncle. Stimulation identified three mesencephalic

20

regions giving similar patterns of responses: 1) in the tegmentum ventral to the
superior colliculus, Z) in the dorsolateral part of the central gray and 3) dorsal to
the cerebral peduncle. The pattern of responses elicited from the central gray
most resembles that occurring during hypothalamic stimulation. The regions that
produce active muscle vasodilation are identical to those identified by Hess and
others as producing the behavioral defense reaction in unanesthetized animals (see
Abrahams gt al., 1960). This fact was confirmed by Abrahams et al. (1960) who
elicited alerting and defense behaviors in conscious cats by stimulating sites
which elicit muscle vasodilation in anesthetized preparations.

hi a subsequent study, Coote g 11, (1973) showed that the region
producing defense-like cardiovascular responses, including active muscle vasodila-
tion extends caudally into the medulla. This region consists of a narrow strip 2.5
mm on either side of the midline, starting ventral to the inferior colliculus and
ending close to the floor of the fourth ventricle. This region is anatomically
distinct from the bilateral hypothalamic cholinergic vasodilator pathway coursing
near the ventral surface of the medulla. Stimulation of the medullary defense
area in conscious animals produces defense-like behaviors; however, these re-
sponses are not as vigorously expressed as those occurring during stimulation of
the hypothalamic or midbrain defense areas. The skeletal muscle dilation elicited
by stimulation of the caudal defense area is only partially reversed by atropine,
suggesting that the vasodilation is due to the activation of both a cholinergic
vasodilator system and the inhibition of sympathetic vasoconstrictor tone.

Hilton _e_t al. (1983) used electrical stimulation to trace an efferent
pathway from the hypothalamic and midbrain defense areas through the ventral
medulla. This pathway courses as a narrow strip ventral to the superior olive and
nucleus of the trapezoid body to the level of the rostral inferior olive where it lies

ventral to the facial nucleus. Guetzenstein e_t al. (1978) and Hilton 2 a_l_. (1983)

21

demonstrated that bilateral application of glycine to the ventral surface of the
medulla overlying the caudal end of the efferent pathway decreases blood
pressure (20-60%), heart rate and cardiac output and stops respiration. Also total
peripheral resistance decreases primarily due to an increase in splanchnic
conductance. Moreover, after applying glycine, stimulation of the hypothalamic
or mesencephalic defense areas fails to elicit a defense response. From these
results Hilton gt a_l. (1983) concluded that glycine blocks a synaptic relay in the
efferent pathway from the defense regions to preganglionic sympathetic neurons.
Furthermore, they suggested that the neurons in the ventral medulla receiving
inputs from the defense area provide a tonic excitatory drive to preganglionic
sympathetic neurons and that the tonic drive originates in the hypothalamic and
mesencephalic defense regions. Thus these authors concluded that the defense
regions are responsible for the maintaining basal sympathetic tone. On the basis
of the data presented, this suggestion is tenuous at best. The authors present no
evidence that the defense regions are indeed tonically active. In addition, no
attempt was made to determine whether other brain regions provide input to the
glycine-sensitive area. For example, Barman and Gebber (1987) showed that the
ventral lateral medulla (which may correspond to the glycine sensitive area)
receives sympathoexcitatory input from the lateral tegmental field of the dorsal
medulla. In addition, it is well known that sympathetic tone is maintained in the
decerebrate animal.

Lovick gt_ gl_. (1984) identified neurons in the glycine sensitive area of
the ventral medulla which are antidromically activated by stimulating the
ventrolateral funiculus of the cat thoracolumbar spinal cord. Approximately half
of these neurons are excited by stimulating the hypothalamic or midbrain defense
regions. In addition, other neurons in the ventral medulla respond to stimulation

of either or both defense regions; however, whether these neurons have spinally

22

projecting axons is not known. They did not address the question of whether the
efferent pathway from the defense regions to these medullary neurons is tonically
active.

Hilton and Redfern (1986) compared the responses produced by chemi-
cal and electrical stimulation of sites in the classical defense regions in order to
locate the cell bodies of neurons responsible for these responses. In conscious and
anesthetized rats, sites from which the defense reaction can be elicited by
electrical stimulation correspond to those identified by Abrahams gt gl, (1960) in
the cat. In contrast, chemical stimulation via the microinjection of the
excitatory amino acid, D,L-homocysteic acid identified four discrete sites within
the electrically-defined defense region from which the defense reaction is
elicited. These four sites were the dorsomedial periaqueductal gray matter, an
area overlying the lateral optic chiasma, the medial tuberal hypothalamus and a
region medial to the lateral lemniscus in the pans. Of these regions the full
pattern of visceral, respiratory and behavioral responses is only elicited by
chemically stimulating the periaqueductal gray. In the remaining three regions,
chemical stimulation elicits decreases rather than increases blood pressure. In
addition, chemical stimulation of the tuberal hypothalamus fails to elicit renal
vasoconstriction. Thus it is concluded that the components of the defense
reaction are generated in many areas of the brain and integrated in the
periaqueductal gray. The rather wide distribution of areas from which the
defense response is elicited by electrical stimulation is thought to reflect the
activation of fibers of passage.

Indirect evidence that the defense region is tonically active and may
generate a component of SND is provided by studies examining the cardiovascular
changes accompanying transitions between awake and sleep states. Mancia _e_t at.

(1970) showed that as cats pass from wakefulness to synchronized and then into

v!

3

23

desynchronized sleep, they display a characteristic pattern of cardiovascular
responses. During synchronized sleep there are slight reductions in heart rate and
blood pressure without much change in total peripheral resistance. During
desynchronized sleep there are greater reductions in cardiac output, total
peripheral resistance and blood pressure. In addition, there is a distinct pattern of
vascular responses, including increased conductance in the skin, mesenteric and
renal vascular beds and a decrease in skeletal muscle conductance. The changes
in vascular resistance are apparently centrally mediated, since, during periods of
desynchronized-like sleep resulting from the administration of physostigmine, the
level of activity on cardiac, renal, splanchnic and lumbar sympathetic nerves
decreases, while activity in vasoconstrictor nerves to skeletal muscles increases
(cf. Coote, 1982). Similar neural responses are reported to occur in other species
during desynchronized sleep (cf. Coote, 1982). Interestingly, the pattern of
cardiovascular responses observed during the transition from the awake state to
desynchronized sleep are the opposite of those resulting from the activation of
the defense reaction. Conversely, the pattern of cardiovascular responses during
awakening from desynchronized sleep mimics that during the defense reaction.
Thus the defense areas appear to be tonically active, with the level of activity in
the system being governed by the behavioral state of the animal.
D. Models of hypertension involving diencephalic systems
Further evidence that the diencephalon can influence cardiovascular
function comes from studies which demonstrate the involvement of diencephalic
structures in several models of experimental hypertension.
1. Renin-dependent renal hypertension
Renal hypertension is produced experimentally by renal artery
stenosis, perinephritis, renal compression or aortic ligation in animals having one

or two kidneys. The initial rise in blood pressure and increase in peripheral

24

resistance in these hypertensive models is associated with increases in plasma
renin and circulating angiotensin II. This early phase of hypertension is known as
renin-dependent renal hypertension. As reviewed by Brody gt g. (1984) angioten-
sin II has both peripheral and central sympathomimetic actions. The pressor
response to intravenous administration of angiotensin II is prevented by cooling or
lesioning the area postrema in dogs, rabbits and cats. In the rat, the region
mediating the central pressor response to intracerebroventricular (icv) angiotensin
II is not the area postrema, but the anteroventral region of the third ventricle
(AVBV). This region incorporating the organum vasculosum of the lamina
terminalis is accessible to both blood-borne and cerebrospinal fluid angiotensin II.
The central action of angiotensin was demonstrated by Fink e_t a_l. (1980), who
compared the pressor responses produced by the intracarotid and intra-aortic
administration of angiotensin II. Since the pressor response to intracarotid
angiotensin is larger, their difference is used as an index of angiotensin's central
pressor effect. The injection of the angiotensin antagonist saralasin into the third
ventricle significantly reduces the central pressor effect of angiotensin 11.
Furthermore, the central pressor effect is also abolished by electrolytically
lesioning the AV3V region, suggesting that this region is responsible for mediating
the pressor effect of circulating angiotensin II.

In a similar study, Lappe and Brody (1984) compared the pressor
and regional vascular responses to intracarotid and abdominal aortic infusions of
angiotensin II in conscious rats. Both the intracarotid and intra-aortic infusion of
angiotensin II produce pressor responses and increases in resistance in the renal,
mesenteric and hindquarter vascular beds. However, these changes are uniformly
greater after intracarotid, than after intra-aortic administration. Ganglionic
blockade with hexamethonium abolishes both the central pressor and mesenteric

Vascular responses; however, renal vasoconstriction is only slightly reduced, while

25

no difference in the hindquarter vasoconstriction is observed. Similarly, the
injection of saralasin into the lateral cerebral ventricle eliminates the pressor
response and significantly attenuates the vasoconstriction in the renal and
mesenteric vasculature. These results indicate that angiotensin acts centrally to
induce vasoconstriction of the hindquarters, renal and mesenteric vasculature.

Haywood gt gt. (1 983) showed that the sustained renin-dependent
hypertension produced by unilateral clipping of a renal artery is prevented by
electrolytically lesioning the AV3V region. In both AV3V lesioned and nonlesioned
rats, clipping of one renal artery produces an initial increase in blood pressure;
however, the level of hypertension is lower in lesioned animals. After five days,
blood pressure in the lesioned animals begins to decrease reaching levels similar
to those in sham-operated, sham-clipped controls within two weeks. In addition,
the hypertensive rats (sham-lesioned, clipped artery) show a tendency to increase
water intake and urine volume, presumably due to the dipsogenic effect of
increasing angiotensin II levels. The AV3V lesioned animals do not show an
increase in water turnover.

The ability of centrally administered angiotensin II to elicit
increases in sympathetic nerve activity was directly demonstrated by Tobey g gt.
(1983). These authors showed that the intra-aortic injection of angiotensin 11
increases splenic and renal sympathetic nerve activity and blood pressure. The
increase in activity is greater on the splenic than on the renal nerve. In contrast,
the icv injection of angiotensin II (20 ug) increases splenic but not renal
sympathetic nerve activity, indicating that central angiotensin II can differential-
ly affect sympathetic nerve activity. The ability of angiotensin II to increase
sympathetic nerve activity is confirmed by Keim and Sigy (1971) who reported
that the intravertebral infusion of angiotensin II increases the level of activity on

preganglionic cervical sympathetic nerves in unanesthetized cats.

26

Hartle gt g_l. (1982) mapped the pathway from the AV3V region
believed to mediate the pressor response to icv administered angiotensin II. By
examining the central angiotensin pressor response after a series of electrolytic
lesions, it was determined that the pathway mediating the pressor response
originates in the AV3V region from where it ascends along the lamina terminalis
to the level of the anterior commissure. The pathway then descends medially
through the anterior hypothalamus. A series of horizontal knife cuts showed the
pathway descends anterior to the PVH and then passes ventral to this nucleus.
The authors concluded that lesions of the AV3V regions which prevent renin-
dependent hypertension most likely interrupt part of this pathway.

2. Nonrenin-dependent renal hypertension

A chronic level of hypertension is produced by removing one
kidney and then either constricting the renal artery (one kidney Goldblatt) or
wrapping the contralateral kidney (one kidney Grollman). Most authors feel that
this form of hypertension is not renin-dependent (Brody g gl_. 1984). For
example, Bumpus gt gt. (1973) showed that the angiotensin II antagonists [He 8]-
and [Sar 1, lie 8] reverse the initial rise in blood pressure resulting from the
ligation of the renal artery in both one- and two-kidney models of hypertension.
However, these antagonists do not lower blood pressure in chronically hyperten-
sive one-kidney Goldblatt animals.

Buggy g g. (1977) reported that lesions of the AV3V region
prevent the hypertension and increase in drinking normally occurring following the
one-kidney Grollman procedure. Lesion of the AV3V region alone produced a
temporary adipsia in one half of the rats, and a transient (1 week) hypertension.
The failure of lesioned animals to develop hypertension is not the result of
decreases in water intake since restricting the water intake of sham-lesioned

animals with renal wrapping does not prevent them from reaching hypertensive

27

levels similar to those of sham-lesioned animals which were allowed to increase
their water intake.

Buggy _e_t 11. (1978) demonstrated that lesion of the AV3V region
six weeks after the initiation of one-kidney Grollman hypertension returns blood
pressure to normotensive levels. Lesion of the AV3V region two weeks after the
development of two-kidney Goldblatt hypertension reduces, but does not com-
pletely return blood pressure to normotensive levels. These results indicate that
the AV3V region is also involved in the nonrenin-dependent maintenance phase of
renal hypertension.

Katholi gt a_l. (1981) studied the contribution of renal afferent
nerve activity to the maintenance phase of hypertension in the one-kidney
Grollman rat. Denervation of the remaining kidney by stripping the adventitia
and painting the renal artery with phenol significantly decreases the chronically
elevated blood pressure in these animals. Denervated animals experience no
change in sodium excretion, urine volume, water or sodium intake, or renal
function compared to sham-operated controls. Since renal denervation failed to
alter water or electrolyte balance, it is suggested that renal-denervation disinhi-
bits a central sympathoinhibitory mechanism.

Hartle and Brody (1982) identified two separate vasoconstrictor
pathways emanating from the AV3V region in the rat. These pathways were
identified by determining which of a series of lesions and knife cuts would block
the pressor response to icv angiotensin II, hypertonic saline or carbachol. The
first, pathway is the periventricular pathway described above (see Hartle gt 31.,
1982). This pathway mediates the pressor responses to icv angiotensin II,
hypertonic saline and carbachol. Furthermore, lesion of this pathway prevents the
development of renin-dependent but not renin-independent forms of hypertension,

and does not block the hemodynamic effects produced by electrically stimulating

28

the AV3V region. Instead, the hemodynamic effects evoked by stimulating the
AV3V are conducted along a laterally directed pathway which eventually joins the
medial forebrain bundle. Although not directly tested, the authors speculate that
the lateral pathway is involved in nonrenin-dependent forms of hypertension which
are prevented by AV3V lesion.

3. Neurogenic hypertension

Hypertension which results from removal of baroreceptor inhibi-
tory input to the central nervous system is called neurogenic hypertension. This
form of hypertension is marked by persistent fluctuations (lability) in blood
pressure. It remains controversial as to whether blood pressure remains elevated
over long periods of time in these animals (see Brody gt 9, 1984).

Bing e_t a_l. (1945) examined the development of neurogenic
hypertension in unanesthetized dogs occurring after sectioning the carotid sinus
and aortic depressor nerves. Denervation produced tachycardia, sustained in-
creases in blood pressure, cardiac output, renal vascular resistance and blood flow
through the forelimb. The authors concluded that these cardiovascular changes
are consistent with an increase in sympathetic tone, however, this was not
demonstrated directly.

Doba and Reis (1973) produced neurogenic hypertension in rats
by electrolytically lesioning the middle third of the n. tractus solitarii (NTS), an
area in which the integration of baroreceptor afferent activity is thought to
occur. The acute effects of NTS lesion include abolition of the baroreceptor
reflexes, an increase in blood pressure without a change in heart rate and a
decrease in respiratory rate. These effects are only present after discontinuation
of the halothane anesthesia. Also observed was a 255% increase in total
peripheral resistance and a 62% reduction in cardiac output. These effects are

attenuated by administering the alpha adrenergic antagonist phentolamine and are

29

thought to reflect an increase in sympathetic outflow occurring after removing
baroreceptor mediated inhibition. The hypertension is prevented or reversed by
midcollicular decerebration indicating that the forebrain is essential for the
development and maintenance of neurogenic hypertension. All rats developing
neurogenic hypertension died within 5 hours as a result of pulmonary edema.

Zhang and Ciriello (1985a,b) reported increases in blood pressure
and heart rate after sectioning the aortic depressor nerves in rats. Both
electrolytic and more selective kainic acid lesions of the PVH before or after
aortic baroreceptor denervation prevents or reverses, respectively, the hyperten-
sion and tachycardia occurring after denervation. Lesion of the PVH had no
effect on heart rate or blood pressure in rats with intact buffer nerves. Kainic
acid lesions destroyed the parvocellular but not the magnocellular neurons in the
PVH.

The possibility that the increases in arterial pressure and heart
rate after sinoaortic denervation are the result of humoral factors was explored
by Alexander and Morris (1982). These investigators measured arginine vasopres-
sin, osmolality and sodium levels in rats up to 7 days after sinoaortic denervation.
While no changes in osmolality or sodium are noted, vasopressin levels are
elevated 15 min and 7 days after denervation, suggesting that an increase in
vasopressin contributes to the neurogenic hypertension. This conclusion was
challenged by Brody gt _a__l. (1984), who reported that administering vasopressin
antagonists fails to prevent the development of neurogenic hypertension occurring
after sinoaortic denervation.

4. Mineralocorticoid-salt hypertension

A combination of dietary salt and repeated administration of

mineralocorticoids such as deoxycorticosterone acetate (DOCA) produces a non-

renin-dependent hypertension. Several causal factors such as elevated vasopressin

30

and sympathetic hyperactivity have been implicated in this model of hypertension.
Berecek gt g_l. (1982) reported that lesion of the AV3V region prevents the
development of DOCA-salt hypertension. The systemic administration of vaso-
pressin to lesioned animals elicits only small increases in blood pressure indicating
that the protective effects of AV3V lesion apparently involve more than the
prevention of vasopressin release.

Takeda and Bunag (1980) compared the level of sympathetic
nerve activity and the magnitude of pressor responses in DOCA-salt hypertensive
and normotensive rats. Under urethane anesthesia, blood pressure and the
frequency of sympathetic nerve firing are significantly higher in DOCA-salt
hypertensive rats than in normotensive rats. Electrical stimulation of the
posterior hypothalamus in both groups increases sympathetic nerve activity and
blood pressure; however, the magnitude of the increases is significantly greater in
the hypertensive rats. Ganglionic blockade by pentolinium reduces blood pressure
and sympathetic nerve activity to similar levels in both groups. Thus it was
concluded that DOCA-salt hypertension is associated with sympathetic hyperacti-
vity.

5. Spontaneous hypertension

The spontaneously hypertensive rat (SHR) was isolated from an
inbred strain of Wistar rats. These rats are characterized by a pronounced
hypertension which develops with age. The hypertension in SHR is thought to
involve a central mechanism.

Judy g a_l. (1976) demonstrated that hypertension in SHR's is
associated with an increase in activity on the splanchnic, cervical, renal and
splenic sympathetic nerves in both anesthetized and conscious SHR‘s. Sympathe-
tic nerve activity and blood pressure increase in parallel with increasing age in

the SHR. This conclusion is disputed by Touw gt 11. (1980) who, by comparing the

31

effects of hexamethonium sympathectomy in conscious adult and juvenile SHR
and normotensive rats, concluded that neurogenic vasoconstrictor tone is not
elevated in SHR's and is thus not directly responsible for the higher blood
pressure.

Although AV3V lesions fail to prevent the development or
maintenance of hypertension in SHR's (Gordon gt gl_., 1982; Buggy gt g, 1978)
several other forebrain mechanisms and structures are thought to be involved in
the development or maintenance of hypertension in the SHR. These include
hyperactivity of sympathetic systems in the posterior hypothalamus (Bunag and
Eferakeya, 1976), exaggerated cardiovascular responsiveness to environmental
stress resembling the alerting-defense reaction (Hallback and Folkow, 1974:

Galeno gt a” 1984) and the amygdala (Galena gt gt” 1982).

IV. Generation of Rhythmic Activity by the Diencephalon and Brain Stem
A. The generation of basal SND
1. Early views on the origin of basal SND

As first demonstrated by Adrian et al. (1932) basal sympathetic
nerve discharge contains cardiac- and respiratory-related rhythms. Historically
there have been two main theories concerning the manner in which the rhythmic
patterns in SND are generated by networks of supraspinal neurons.

The classical theory of how patterns of sympathetic activity are
generated held that sympathetic tone was randomly generated within a diffusely
organized brain stem reticular network. The presence of rhythms in the outflow
of this system was thought to arise as a consequence of periodic activity in
afferents to the generating network. For example the cardiac-related rhythm in
SND was believed by Adrian gt gl_. (1932) and others (Green and Heffron, 1967;

Cohen and Gootman, 1970) to arise from the pulse synchronous activity in

32

baroreceptor afferent nerves. That is, increases in baroreceptor nerve discharge
during systole cause a delayed inhibition of SND, whereas, removal of inhibition
during diastole reflexly increases SND. Similarly, the slower respiratory-related
rhythm in SND was thought to arise from the synchronous activity in vagal
pulmonary lung inflation afferents.

This view of sympathetic rhythm generation is no longer held. It
is now well established that central sympathetic networks in the brain stem are
inherently capable of generating complex output patterns in the absence of
periodic afferent input (cf. Gebber, 1982; 1984).

2. Evidence that the brain stem is intrinsically capable of rhythm
generation

Evidence that the brain stem is inherently capable of rhythm
generation was first presented by Green and Heffron (1967) and Cohen and
Gootman (1969). These authors found that SND in baroreceptor-innervated cats
can contain a noncardiac-related 10 Hz rhythm.

Recently, Barman and Gebber and collaborators have advanced
the theory that the 2- to 6-Hz rhythm in SND which is normally cardiac-locked is
generated in the brain stem and then entrained to the cardiac cycle by the
baroreceptor reflexes. Several lines of evidence support this view. First,
although no longer cardiac-locked, the 2- to 6-Hz rhythm in SND persists after
complete baroreceptor denervation (Taylor and Gebber, 1975; Barman and Gebber,
1980). Therefore, the occurrence of rhythmic bursts in SND (recorded as slow
waves using a band pass of l to 1000 Hz) at frequencies between 2 and 6 Hz is not
the result of pulse synchronous baroreceptor nerve activity. The 2- to 6H2
rhythm in SND is apparently generated in the brain stem since it persists after
midcollicular decerebration (Barman and Gebber, 1980), but not after acute
transection of the spinal cord at the first cervical segment (McCall and Gebber;

1975). Second, slowing of the heart rate by cardiac pacing or vagal stimulation

33

produces dramatic shifts in phase between sympathetic nerve slow waves and the
arterial pulse (Gebber, 1976, Morrison and Gebber, 1982). If the cardiac-related
rhythm in SND is the result of pulse synchronous baroreceptor discharge then the
phase relation between SND and the arterial pulse should be independent of the
heart rate. Third, when the heart rate is slowed to less than 2.5 beats/s, the
declining phase of the sympathetic nerve slow wave begins before the onset of
pulse synchronous baroreceptor afferent nerve activity, thus the decline of the
sympathetic slow wave cycle can not be attributed to baroreceptor-mediated
inhibition of SND. The termination of the slow wave must involve a central
mechanism. Gebber (1976) proposed that an inhibitory phasing mechanism is
responsible for the baroreceptor reflex entrainment of centrally generated
sympathetic slow waves to the cardiac cycle. In this model, the rising phase of
the sympathetic nerve slow wave begins only when baroreceptor inhibition falls to
some minimum level.

The respiratory-related component of SND is observed following
vagotomy indicating that it, at least in part, is of central origin (Cohen and
Gootman, 1970; Preiss e_t 11., 1975; Barman and Gebber, 1976; Bachoo and Polosa,
1987). However, the origin of the respiratory-related rhythm in SND is contro-
versial. Several authors (Cohen and Gootman, 1970; Preiss _e_t_ _a_l. 197 5; Bachoo
and Polosa, 1987) have proposed that this slow rhythm in SND is extrinsically
imposed on sympathetic networks by elements of the brain stem respiratory
oscillator. In contrast, Koepchen and Thurau (1959) and Barman and Gebber
(1976), proposed that the slow component in SND is generated by a sympathetic
oscillator which is normally, but not always, coupled to the respiratory oscillator.
Three observations led to this conclusion: 1) changes in respiratory rate were
accompanied by dramatic shifts in phase relations between sympathetic and

phrenic nerve discharge, 2) slow oscillations of sympathetic and phrenic nerve

34

discharge were not always locked in a 1:1 relation, 3) the sympathetic nerve slow
wave activity persists when respiratory rhythmicity ceases during periods of
hyperventilation.

3. Fraposed mechanisms of SND generation in the brain stem

In the baroreceptor-denervated cat, the rhythm in SND is
characterized by irregularly occurring 2- to 6-Hz slow waves (i.e. variable inter-
slow wave intervals (Taylor and Gebber, 197 5; Barman and Gebber, 1980). Barman
g a_l. (1984) tested the possibility that the irregular 2- to 6-Hz activity pattern of
SND in baroreceptor-denervated cats arises from multiple brain stem generating
networks. By using crosscorrelation and power density spectral analysis, they
foxmd that the frequency components of activity simultaneously recorded from
different postganglionic sympathetic nerves can be markedly different. More-
over, by using peri-spike-triggered averaging (a method of crosscorrelation) they
formd that the activity of some ventrolateral and raphe medullary neurons is more
strongly correlated to the activity on one sympathetic nerve than another.
Evidence that the sympathetic generating circuits are coupled is provided by the
fact that although more strongly correlated to activity in one nerve, the
medullary units with sympathetic nerve-related activity are correlated to activity
in both nerves. Furthermore, considerable overlap is observed in the frequency
spectra of SND recorded from two sympathetic nerves.

Langhorst and collaborators (1975; 1980: 1984) argue against the
notion that there are discrete pools of sympathetic reticular neurons capable of
selective sympathetic control. Rather, they believe that rhythmic activity in a
variety of effector systems including SND is generated by a ”common brain stem
system”. Effector systems under the simultaneous control of this common brain
stem system include the cardiovascular, respiratory, and general activating

systems plus afferent sensory inputs. The output onto the effector system is

35

determined by the integration of convergent afferent inputs from somatic and
vegetative receptors. Individual reticular neurons receiving the convergent inputs
then influence more than one effector system. Support for this hypothesis comes
from correlation and spectral analysis which showed that on a long time scale the
rhythmic discharge patterns of all reticular neurons display a common rhythmic
component having a period of 4 to 10 seconds. While often not the predominant
rhythm, a similar slow 0.1- to 0.25-Hz rhythm is usually observed in the activity
patterns of the effector systems. In addition, power spectral analysis revealed
that there are common rhythmic components in the discharge patterns of some
reticular neurons and the activity patterns in several effector systems. The
authors provided this as evidence that individual reticular neurons influence more
than one effector system.

There are several problems with the common brain stem model.
First, the authors fail to prove that the common slow rhythm in reticular activity
is even generated in the brain stem reticular formation. Second, the discharges of
most reticular neurons are not related to the activity patterns in several effector
systems. This is true in their work and in the work of others (Barman and Gebber,
1981; 1982; 1984). Barman and Gebber demonstrated that most reticular neurons
do not have sympathetic nerve-related activity and those that do often do not
have cardiac-activity and/or EEG-related activity. Third, Langhorst g gt. failed
to address the possibility that the discharges of reticular neurons are related to
activity in more than one effector system because there is coupling between
functionally distinct reticular systems controlling specific effector systems. In
this case the temporal relationship between the discharges of these neurons and
the activity in effector systems may not reflect a role of these neurons in

controlling all effector systems.

36

4. Possible brain stem regions involved in the generation of SND

In recent years electrophysiological and pharmacological investi-
gations have revealed three areas of the medulla which may be involved in the
control of basal SND in the anesthetized animal. The first of these regions is the
rostral ventrolateral medulla (VLM) located caudal to the facial nucleus. Ross gt
gt. (1984) and McAllen _e_t_ gt. (1982) showed that both electrical stimulation and
the application of excitatory amino acids in the VLM elicits dramatic increases in
arterial blood pressure and heart rate. Furthermore, bilateral lesions of this area
lower arterial blood pressure to levels resembling those occurring after acute
transection of the cervical spinal cord (Dampney and Moon, 1980). By using spike-
triggered averaging, Barman and Gebber (1983: 1985) showed that the spontaneous
discharges of some VLM neurons are temporally related to SND in baroreceptor-
denervated and -intact cats. The axons of many of these neurons project to the
intermediolateral nucleus (IML) of the thoracolumbar spinal cord (Barman and
Gebber, 198 5). These neurons are thought to be sympathoexcitatory since their
firing rate decreases during baroreceptor reflex activation.

The second area is the lateral tegmental field (LTF) of the
dorsolateral medulla, including the reticularis parvocellularis and reticularis
ventralis nuclei. Like the VLM, electrical stimulation of the LTF elicits increases
in blood pressure and SND (Dampney and Moon, 1980; Gebber and Barman, 1985).
The injection of excitatory amino acids into the rostral portions of n. reticularis
parvocellularis increases blood pressure in the rabbit (Goodchild and Dampney,
1985). Lesion of this region also produces a profound decrease in blood pressure
(Kumada gt gl_., 1979). The spontaneous discharges of some neurons in the LTF
are temporally related to SND (Barman and Gebber, 1981; Gebber and Barman,
1985). Some LTF neurons with sympathetic nerve-related activity respond to

baroreceptor reflex activation with a decrease in firing rate whereas others show

37

an increase. The axons of LTF neurons with sympathetic nerve-related activity
do not project to the thoracic spinal cord. However, as demonstrated by
antidromic activation sympathoexcitatory LTF neurons project to the region of
the VLM containing sympathoexcitatory neurons which project to the IML
(Barman and Gebber, 1987).

The third area believed to be involved in the control of SND is
the medullary raphe nuclei (raphe magnus, pallidus and obscurus). These nuclei
are contained in the medial medullary depressor region of Wang and Ranson
(1939). Electrical stimulation of this area elicits either decreases or increases in
blood pressure and SND (Adair fl a_l., 1977; Morrison and Gebber, 1982; McCall
and Humphrey, 1985). Lesion of the medial medulla including the medullary raphe
nuclei increases SND (Barman and Gebber, 1978). By using spike-triggered
averaging Morrison and Gebber (1982; 1984; 1985) showed that some neurons in
the medullary raphe nuclei have spontaneous activity temporally related to SND.
The majority (75%) of these neurons are sympathoinhibitory with the rest being
sympathoexcitatory. As with the VLM sympathetic neurons, the axons of some
raphe neurons with activity related to SND also project to the thoracic spinal
cord, however, only sympathoinhibitory raphe neurons terminate in the IML.

Gebber and Barman (1985) compared the firing times of neurons
in the LTF, VLM and raphe nuclei relative to the peak of the first slow wave to
the right of time zero in the spike-triggered averages of SND. The rationale
behind this comparison is that the neurons which are closer to the origin of SND
should fire earlier in the sympathetic cycle. That is, the interval between unit
spike occurrence and the peak of SND in the spike-triggered averages should be
longer for these neurons. Gebber and Barman found that sympathoexcitatory
neurons in the LTF fire on the average, 35 and 43 ms before their counterparts in

the VLM and raphe, respectively. Interestingly, the modal onset latency of

38

synaptic activation of sympathoexcitatory VLM neurons to electrical stimulation
of the LTF (2613 ms) is very similar to the difference in the mean firing times
between LTF and VLM sympathoexcitatory neurons (Barman and Gebber, 1987). A
comparison of sympathoinhibitory neurons showed that LTF neurons fire before
their counterparts in the raphe. A direct connection between these two regions
has yet to be demonstrated. Based on these findings it is concluded that the
sympathetic neurons in the LTF are closer to the origin of SND. Whether the LTF
neurons comprise a sympathetic oscillator which outputs through the VLM and
raphe remains to be determined. Alternatively the LTF may receive input from
some as of yet undefined generator of SND located in the brain stem or the
forebrain.
B. Thalamic mechanisms for the generation of electrocortical rhythms
Under a variety of experimental and behavioral conditions the electri-
cal activity in the cerebral cortex is rhythmic in nature (EEG, electroencephalo-
gram). It is widely accepted that much of this cortical activity originates in the
thalamus (cf. Dempsey and Morison, 1942b). A number of mechanisms by which
these cortical rhythms are generated have been proposed.
1. Early models of thalamic generation of EEG rhythms

Dusser de Barenne and McCollock (1938a, 1938b) used the local
application of strychnine to the cortex and thalamus to demonstrate the func-
tional reciprocity between the thalamic relay nuclei and that portion of the
cerebral cortex to which they project. Undercutting of the projection cortex
abolishes or greatly alters the normal pattern of EEG activity. On this basis, it is
proposed that the reverberating activity in thalamocortical-thalamic circuits is
the source of normal EEG activity.

An alternative model of cortical rhythm generation was proposed

by Morison and Dempsey in the early 1940's (Dempsey and Morison, 1942a,b;

39

Morison and Dempsey, 1942). These authors showed that stimulation of the
thalamus elicits two types of cortical responses. Stimulation of sensory relay
thalamic nuclei elicits a ”primary" or augmenting response in their projection
cortices. These responses with onset latencies of less than 4 ms are characterized
as a positive/negative wave form capable of following stimulus frequencies up to
120 Hz. Stimulation of the nonspecific medial thalamic nuclei at frequencies of
2- to 15-Hz evokes the recruiting response in wide regions of the cortex. The
recruiting response has a long onset latency (20-35 ms) and is characterized by a
monophasic surface negative waveform which waxes and wanes in amplitude with
successive stimuli. The recruiting responses closely resemble, in both contour and
frequency (8-12 Hz), the spontaneous bursts of cortical spindle activity (alpha
spindles) seen in barbiturate-anesthetized or sleeping animals. The recruiting
response and spontaneous cortical spindles are recorded in cortical areas isolated
except for their thalamic connections indicating that the generalized nature of
the recruiting response reflects the activity of a widely distributed system and is
not the result of the intracortical spread of activity from discrete cortical
regions. Based on the similarities, it is concluded that the spontaneous spindle
burst and recruiting responses are identical. Dempsey and Morison concluded that
the spontaneous cortical alpha rhythm is generated in the medial thalamus and
transmitted to wide spread cortical areas via a diffuse thalamic projection system
separate from that relaying sensory information to the the cortex. This
mechanism is proposed in spite of the fact that at this time cortical projections
from many of the medial thalamic nuclei have not been demonstrated.

In a series of papers in the early 1950's, Jasper and colleagues
modified the original hypothesis of Morison and Dempsey (cf. Andersen and
Anderson, 1968). These authors behaved that the midline and intralaminar

thalamic nuclei in conjunction with n. reticularis thalami and n. ventralis anterior

40

constitute a rostral extension of the generalized reticular activating system
proposed by Magoun and Ranson in 1949. They proposed that the pacemaker
activity of the medial thalamic nuclei is relayed to the cortex via n. reticularis
and n. ventrais anterior where this activity 'acts upon the cortex, controlling the
form and rhythm of the background of cortical activity upon which afferent
impulses must act, and regulating the generalized excitatory states of the cortex
as a whole" (Jasper, 1949). This theory had to be discounted when it became clear
that the n. reticularis does not project to the cortex (Scheibel and Scheibel, 1966;
Jones, 1985).
2. Mechanisms for the generation of alpha spindle activity

Purpura and coworkers (Purpura and Cohen, 1962; Purpura and
Shofer, 1963; Purpura gt gt., 1965) prOposed that the nonspecific medial thalamic
nuclei are the major source of synchronizing input to the thalamus. These authors
examined the synaptic events occurring in thalamic neurons during the recruiting
response elicited by stimulation of the nonspecific medial thalamic nuclei.
Intracellular recordings from thalamic neurons during the recruiting response
revealed that the synchronization of neuronal firing is produced by a recurring
sequence of a short-latency (4-15 ms) excitatory postsynaptic potential (EPSP),
followed by a long-latency (15-40 ms), prolonged (80-200 ms) inhibitory postsynap-
tic potential (IPSP). The prolonged IPSP is thought to be the major feature of the
synchronization process since neither spontaneous nor evoked neuronal discharges
of the thalamocortical neuron occur during the IPSP. Rather, the discharges of
thalamocortical neurons are restricted to the short periods between IPSPs. This
EPSP-IPSP sequence is observed in over 70% of the neurons recorded in widely
separated regions of the n. ventralis anterior, n. ventralis medialis, n. ventralis
lateralis and from the intralaminar and midline thalamic nuclei. Thalamic focal

potentials recorded during the recruiting response reveal long-latency (15-40 ms),

41

long-duration (40-80 ms) positive waves presumably reflecting the summation of
EPSP-IPSP sequences throughout the thalamus. These positive thalamic waves
are temporally related to the negative waves in the cortical recruiting response.
Similar EPSP-IPSP sequences are recorded in cells in the ventrobasal complex
during spontaneous alpha spindle activity in the sensory cortex (Maekawa and
Purpura, 1967). High frequency (20 Hz) stimulation of the medial thalamic nuclei
which blocks the recruiting response in the cortex also prevents the development
of prolonged IPSPs in thalamic cells (Purpura and Shofer, 1963). Instead, an
increase in neuronal discharge associated with a steady depolarization or no
change in membrane potential is observed.

Like Purpura and coworkers, Andersen and Eccles (1962) ob-
served that the rhythmic bursting discharge of thalamic neurons is associated with
a recurring sequence of EPSPs and IPSPs. Andersen and Eccles also demonstrated
that rhythmic burst discharges in thalamic relay cells occurs during stimulation of
peripheral afferents or the antidromic activation of thalamocortical neurons,
suggesting that the generation of rhythmic activity is an inherent property of all
thalamic neurons. From these data Andersen and Eccles proposed the facultative
pacemaker theory for the generation of rhythmic burst discharges in thalamic
neurons. This model was later expanded upon by Andersen and Andersson in 1968.
Briefly, this model proposes that the thalamocortical cells excited by afferent
inputs emit excitatory recurrent axon collaterals onto local interneurons. The
interneurons in turn inhibit the thalamocortical neuron. Furthermore, it is
proposed that during the declining phase of the interneuron-induced inhibition, the
thalamocortical neuron becomes hyperexcitable (post anodal exaltation) and
discharges again, thus initiating a repeating cycle of excitation and inhibition.
The post anodal exaltation is thought to be an intrinsic property of the membrane,

thus endowing the neuron with pacemaker prOperties. The possibility that

42

excitatory interneurons help to trigger subsequent neuronal discharges and recruit
other relay neurons is also proposed.

Andersen and Sears (1964) demonstrated that spindle bursts and
8- to 12-Hz focal waves can be recorded in midline and relay thalamic nuclei in
acute and chronically decorticate cats. Spindle activity is also present in relay
nuclei after section of all connections with the nonspecific medial thalamic
nuclei. On this basis, it is concluded that all regions of the thalamus are
inherently capable of generating rhythmic spindle activity via facultative pace-
maker circuits. In contrast to Purpura and coworkers, the medial thalamic nuclei
are not believed to be the major source of spindling activity in the thalamus,
although these regions may exert control over these rhythms. Andersson and
Manson (1971) later demonstrated that rhythmic activity occurring simultaneously
from different thalamic regions in unanesthetized decorticate cats can be quite
different. These data further suggest that the rhythmic activity in the cortex and
thalamus is not generated by or in one specific thalamic region.

Although attractive, the faculatative pacemaker theory is
seriously flawed. For example Purpura and colleagues pointed out that the long
latency and duration of the IPSP in the thalamocortical cells is hard to reconcile
on the basis of a simple feedback inhibitory synapse. These authors also pointed
out that brief periods of stimulation (7 Hz) of the midline thalamic nuclei elicits
EPSP-IPSP sequences in thalamic neurons and spindle waves in the cortex which
persist (in the absence of neuronal spikes) following termination of the stimulus
(Purpura e_t gh, 1966). According to the facultative pacemaker model, the
thalamocortical neuron must discharge in order to observe the IPSPs in thalamic
neurons. Perhaps the strongest evidence against this model comes from anatomi-
cal studies of thalamic nuclei using the Golgi method. These studies failed to

support the facultative pacemaker theory since only 15-20% of thalamocortical

43

neurons have recurrent axon collaterals (Scheibel g a_l., 1973). In Scheibel's words
the small number of collaterals was deemed to be ”an insignificant substrate for
the powerful recursive effects ascribed to them.“

3. The thalamic reticular nucleus as a generator of thalamic spindle
sequences

The n. reticularis thalami is a sheet-like structure encompassing
the rostral and lateral borders of the thalamus. The structure and connections of
this nucleus have been well characterized (see Scheibel and Scheibel, 1966;
Steriade _e_t a_l., 1984; Jones, 1985). The axons of most thalamocortical neurons
emit collaterals in the n. reticularis in a topographical fashion. The axons of
these thalamic reticular neurons project back to those thalamic regions from
which the collateral input arises. Corticothalamic neurons also collateralize in
the reticular nucleus; however, the reticular nucleus does not have cortical
projections.

The early development of the hypothesis that the n. reticularis is
involved in the generation of the 8- to 12-Hz spindle rhythms is reviewed by
Steriade and Hobson (1976). On the basis of mostly anatomical data, the Scheibels
proposed that the rhythmic bursting discharge of the thalamic relay cells is due to
direct feedback inhibition from n. reticularis. Physiological evidence in support
of this hypothesis was provided by Schlag and Waszak who showed that a
reciprocal relationship exists between the burst discharges of n. reticularis and
thalamocortical neurons. During episodes of EEG spindling or high amplitude
cortical slow waves, the discharge rates of neurons in n. reticularis increase while
those of thalamocortical neurons decrease. Moreover, the burst discharges of the
reticularis neurons often spanned the spindle sequence and are thus thought to be
responsible for the spindle-related hyperpolarization (IPSP) of the thalamocortical
neurons. During EEG desynchronization elicited by stimulating the mesencephalic

reticular formation, the discharge rates of neurons in n. reticularis decrease while

44

those of thalamocortical neurons increase. Additional support for this proposal
comes from the work of Houser gt gt. (1980) who demonstrated that many of the
projections from n. reticularis to thalamic relay nuclei are GABAergic.

Further evidence that the n. reticularis is involved hi generating
the alpha spindles is provided by the observation that neurons in the anterior
thalamic nuclei (anterior ventral, anterior medial and anterior dorsal) which do
not receive afferents from n. reticularis do not display spontaneous or evoked
spindle-er discharges (Mulle e_t gt., 1985). This is the case even though these
neuron have intrinsic membrane properties similar to those of thalamic neurons
displaying spindling activity. Furthermore, EEG recorded from the parasplenial
projection cortex of these nuclei also fails to display spindle activity.

Although it is generally accepted that the n. reticularis is
involved in generating the alpha rhythm, Steriade and colleagues (Steriade and
Deschenes, 1984; Steriade _e_t_ a_l., 1985; 1986; 1987) have questioned whether the
regulation of activity in thalamocortical neurons is due solely to a direct feedback
inhibition from n. reticularis. That a more complex mechanism is involved is
suggested by Steriade gt a_l. (1986) who recorded the activity of neurons in
thalamic relay nuclei and n. reticularis during various sleep and awake states in
cats. As in earlier studies, during EEG synchronization n. reticularis neurons fire
in long bursts (increasing then decreasing interspike intervals) which often span
the spindle sequence of the thalamocortical cells. In contrast, during arousal and
awake states (EEG desynchronized) n. reticularis and thalamocortical neurons
both cease spindling and significantly increase their firing rates in parallel. Such
parallelism is not consistent with a direct inhibitory action of reticularis neurons
on thalamocortical neurons.

To account for the state-dependent relationships between the

activity of thalamocortical and n. reticularis neurons, Steriade and colleagues

4S

(Steriade e_t a_l., 1985; 1986; 1987) proposed a model whereby n. reticularis
controls the activity in thalamocortical circuits. Moreover, these authors propose
that the n. reticularis functions as a pacemaker to generate alpha spindling. In
this model, neurons in n. reticularis have direct GABAergic projections to
thalamocortical neurons and to GABAergic interneurons located near the thala-
mocortical neurons. These inhibitory interneurons in turn contact the thalamo-
cortical neurons. The presence of GABAergic interneurons was reported by
Montero and Singer (1985). A central feature of this model is the notion that the
projections from n. reticularis exert differential inhibitory effects on the
thalamocortical and GABAergic interneurons, with the more powerful inhibitory
effects being exerted on the interneurons. Thus state-dependent changes in the
amount of GABA released by the n. reticularis neurons changes the activity
pattern in thalamocortical circuits by changing the ratio of inhibition of the
interneurons and thalamocortical neurons. For example during EEG-desynchroni-
zation, the'n. reticularis cells cease bursting and fire tonically as single spikes at
an increased rate. The GABA released by the n. reticularis neurons inhibits the I
more sensitive interneurons. As a consequence, the thalamocortical neurons
become more excitable due to the disinhibition and fire at a higher frequency.
This proposal is supported by the observation that reticularis-disconnected thala-
mocortical neurons have a lower average firing rate than do their counterparts in
reticularis-intact cats during EEG-desynchronization. Conversely, during EEG-
synchronized states, the n. reticularis neurons begin to discharge in prolonged
bursts. During these bursts the intraburst frequencies are higher than the fastest
tonic discharge rates occurring during desynchronized states and as a conse-
quence, more GABA is released. The increased in GABA release via direct
reticularis inputs is now sufficient to hyperpolarize the thalamocortical neurons.

The hyperpolarization of the thalamocortical neuron triggers the deinactivation of

46

a calcium (Ca'H’) conductance, eventually leading to the spindle burst discharge of
the neuron. At the same time an overabimdance of GABA released by the
bursting n. reticularis neuron saturates the more GABA-sensitive interneurons,
completely inhibiting their firing.

Evidence in support of this model is presented by Steriade et al.
(1985) who examined the discharges of thalamocortical neurons after removing
inputs from the n. reticularis by using either knife cuts or kainic acid lesions.
Removing inputs from the n. reticularis abolished spindle-related rhythms (se-
quences of 7-12 Hz waves recurring at 0.1-0.2 Hz) in the thalamus and ipsilateral
EEG. In addition, n. reticularis-disconnected thalamocortical neurons no longer
display spindle-related sequences of prolonged IPSPs. Instead short duration IPSPs
are recorded. Thalamocortical cells exhibit high frequency spike bursts whose
parameters (number of spikes, intraburst frequency, duration of consecutive
intervals) are the same as those in relay and intralaminar cells having intact
reticularis connections. However, the characteristic pattern of spindle-related
sequences of spike bursts and irregularly occurring single-spike discharges during
non-spindling periods is replaced by solitary bursts occurring with remarkable
regularity at l- to 2-Hz. This regular 1- to 2-Hz burst pattern is thought to be
mediated by the GABAergic interneurons which are disinhibited by removing
inputs from n. reticularis. The increased activity in the interneurons produces a
series of short duration IPSPs in the thalamocortical neurons. The rapidly
occurring IPSPs summate and hyperpolarize the thalamocortical neuron de-
inactivating a low threshold, long duration Ca++ conductance whose activation
leads to the burst response. This conclusion is based on the finding that a
subconvulsive dose of bicuculline administered intravenously changes the dis-
charge pattern of reticularis-disconnected thalamocortical neurons from rhythmic

bursting to irregularly occurring single spikes.

47

Evidence that the n. reticularis functions as a pacemaker in the
generation of spindle rhythmicity during EEG-synchronization is provided by
Steriade g a_l. (1987). In this study extracellular recordings of unit discharges and
focal potentials were made in the rostral pole of n. reticularis after surgically
disconnecting it from all other thalamic nuclei. As previously discussed discon-
nection of n. reticularis abolishes spindle activity in the ipsilateral cortex and
thalamic nuclei (Steriade gt _a_l., 1986). However, neurons in the disconnected n.
reticularis discharge in the biphasic (acceleration-deceleration) pattern of spike
bursts characteristic of neurons in the intact n. reticularis during EEG synchroni-
zation. The spike bursts were of long duration (50 ms-1.5 s) and recurred with a
frequency of 0.1- to 0.3-Hz. Simultaneous recordings of focal potentials through
the recording microelectrode reveal spindle sequences (grouped 7-16 Hz slow
waves recurring at 0.1-0.3 Hz) temporally related to the spike bursts of the n.
reticularis neurons. The authors suggest that the synchronization of the burst
discharge of the reticularis neurons involves dendrodendritic interactions between
the GABAergic reticularis neurons. The presence of dendrodendritic connections
between reticularis neurons has been demonstrated with the electron microscope
(Deschenes e_t. gt., 1985). In addition, the intravenous administration of bicuculline
greatly reduces spike bursts in the deafferentated n. reticularis.

4. Origin of the cortical delta and theta rhythms

Although it is fairly well established that the cortical alpha-
rhythm (8-12 Hz) is generated in the thalamus there is some controversy as to
whether the slower delta and theta rhythms in the EEG are generated in the
thalamus.

Villablanca (1974) examined sleep and behavioral patterns in
chronic athalamic and decorticate cats. During the first 10 days after removing

the thalamus, irregular high voltage slow waves occurring at frequencies of 2- to

48

4-Hz dominate the EEG. Initially, the synchronized EEG pattern is unchanged.
during behavioral arousal, however, after 15 to 25 days, periods of desynchro-
nized-like EEG activity accompany arousal. At no time after thalamic ablation
are 8- to 12-Hz alpha spindles observed in the EEG, nor can they be induced by
administering pentothal. In acute (3-5 days postoperative) decorticate cats,
rhythmic 8- to 12-Hz waves which wax and wane in amplitude are recorded in the
electrothalamogram (EThG). After six postoperative days, the spindle-like
rhythms diminish. During waking irregular, low voltage waves occurring at
frequencies of over 10 Hz are recorded in the EThG. During nonREM sleep the
irregular waves in the EThG increase in voltage and became slower (2- to 6-Hz).
Thus it was suggested that the thalamus is capable of generating at least some of
the slow wave activity in the EEG.

Steriade gt _a_l. (1987) confirmed Villablanca's (1974) observation
that slow wave activity in the EEG persists after removing thalamic influences.
Knife cuts of the corona radiata, which severed corticothalamic connections,
abolish cortical spindle bursts; however, synchronous 0.5- to 4-Hz slow waves
persist in the EEG.

Andersson and Manson (1971) identified three patterns of syn—
chronized thalamic slow wave activity in acute unanesthetized decorticate and
decerebrate cats. The first pattern of activity recorded primarily from the
ventrobasal complex, is characterized by 10- to l4-Hz spindles. The second
pattern is characterized by long lasting irregular slow waves occurring at a
frequency of 4- to 8-Hz. This pattern of activity is recorded primarily in the
dorsolateral thalamus. The third pattern is most often recorded from the medial
thalamic nuclei and consists of 8- to 12-Hz spindle bursts separated by periods of

slower frequency (4- to 6-Hz) slow waves. These findings indicate that the

49

thalamus is capable of generating spindles as well as synchronized slow wave
activity.

Jahnsen and Llinas (1984), by using a guinea-pig thalamic slice
preparation, demonstrated that most thalamic neurons are endowed with intrinsic
membrane properties which allow them to oscillate at frequencies of 9- to lO-Hz
or 5- to 6-Hz. Thus each thalamic neuron is capable of operating as a single cell
oscillator. The central feature of the proposed mechanism for the 5- to 6-Hz
rhythmic discharge is the prolonged afterhyperpolarization following the spike
burst. Increasing the level of hyperpolarization by synaptically generated IPSPs
or by current injection de-inactivates a transient potassium (K+) current and a
Cafi-dependent K+ conductance, thus increasing the duration of the afterhyper-
polarization. This in turn de-inactivates a low threshold Ca“ conductance
triggering a rebound low threshold spike burst which begins the cycle again. The
time course (170 ms) of these events is responsible for the 5- to 6-Hz frequency

oscillations.

METHODS AND PROCEDURES

A. General Procedures
Cats of either sex, weighing between 2.0 and 4.5 kg were used in these
studies. The animals were initially anesthetized with ketamine hydrochloride (10
mg/kg, im) followed by intravenous alpha-chloralose (40 mg/kg, initial dose). The
animals were paralyzed with gallamine triethiodide (4 mg/kg, iv) and artificially
ventilated. The rate (12-20 strokes/min) and stroke volume (50-75 ml/stroke) of
the respirator (Harvard model 607) were adjusted to mimic natural respiration. A
bilateral pneumothoracotomy reduced respirator pump-related movements. Sup-
plemental doses of chloralose and neuromuscular blocking agent were admini-
stered as required. Rectal temperature was maintained at 3711°C by using a heat
lamp. Blood pressure was monitored from either the femoral or brachial artery
with a Statham transducer (Model P23AC) and displayed on a Grass polygraph
(model 7D). Standard techniques were used to record lead II of the electrocardio-
gram (ECG). Drugs were administered through a cannula inserted in the femoral
vein. In approximately 60% of the experiments, a solution of dextran (6% in
saline) and norepinephrine (20 ug/ml) was infused through a femoral vein cannula
to maintain mean arterial blood pressure at or above 100 mmHg. The cats were
placed in a David Kopf Instruments stereotaxic apparatus and spinal investigation
imit.
1. Baroreceptor denervation
In some experiments the baroreceptor nerves (carotid sinus, aortic

depressor and cervical vagus nerves) were sectioned bilaterally. These nerves

50

51

were isolated from a ventral approach after reflecting the trachea and esophagus.
The carotid sinus nerve was identified on both sides at its junction with the
hypoglossal nerve. The cervical, vagus and aortic depressor nerves were isolated
at a midcervical level. Denervation was deemed complete if 1) a sudden rise in
blood pressure failed to inhibit reflexly inferior cardiac SND and 2) sympathetic
slow wave activity was no longer locked in a 1:1 fashion to the cardiac cycle.

2. Nerve recordings

The left postganglionic inferior cardiac sympathetic nerve was iso-
lated retropleurally at its exit from the stellate ganglion after removing the head
of the first rib. The nerve was tied distal to the ganglion with saline-soaked silk
thread. A loop was then tied in the thread and the nerve cut distal to the tie.
Nerve activity was recorded monOphasically after placing the silk loop on the
indifferent pole and the proximal portion of the nerve on the central pole of a
bipolar platinum electrode. To prevent drying, the nerve and ganglion were
covered with an emulsion of silicone release agent (Dow Corning 7 Compound) and
paraffin oil (Fischer Scientific Co.).

The left postganglionic renal sympathetic nerve was isolated retro-
peritoneally through an incision in the left flank. One of the branches was tied
with saline-soaked silk thread and cut between the tie and the nerve's jimction
with the renal artery. A loop was then tied in the thread. Monophasic recordings
were made as just described. The abdominal skin flaps were secured to a frame in
order to form a pool which was filled with warm paraffin oil to cover the nerve.

Nerve activity was amplified by using a capacity-coupled preamplifier
at a band pass of 1 to 1,000 Hz. This allowed the synchronous discharges of the
sympathetic nerves to be viewed as slow waves (i.e. envelopes of spikes) on the

polygraph and oscilloscope (see RESULTS; Figure 11).

52

Electrical cortical activity (electroencephalogram, EEG) was recorded
monophasically between a gold plated electrode (Grass Instruments, Model E4G)
affixed to the frontal bone and a indifferent electrode (an alligator clip) attached
to reflected scalp muscle. Cortical activity was preamplified using a band pass
of 1 to 1,000 Hz and displayed on a polygraph and oscilloscope.

3. Electrical stimulation of the diencephalon and cortex

Square wave pulses (lo-ms trains of three pulses or single shocks) were
delivered to selected sites in the diencephalon and frontal-parietal cortex by using
a Grass S88 stimulator connected to a Grass PSIU-6 constant current unit.
Stimuli were passed through either concentric bipolar electrodes, 0.25 mm
exposed tip (Rhodes Medical Instruments, Model NE—100) or etched monopolar
tungsten microelectrodes (10-30 kilo ohm tip impedance). In the latter case, the
anode (an alligator clip) was attached to reflected scalp muscle. Stimulus current
was measured by monitoring the voltage drop across a 100 ohm resistor in series
with the anode. Potentials evoked on sympathetic nerves by stimulation of the
diencephalon and cortex were computer averaged (Nicolet, Model 1070; see
section 3.4), displayed on an oscilloscope and photographed.

4. Histology

The diencephalon and in some cases the brain stem were removed and
placed in 10% buffered formalin for a period of at least 7 days. In decerebration
studies the completeness and level of midbrain transection was verified by gross
inspection of the brain stem and microscopic evaluation of frontal sections of 30-
um thickness stained with cresyl violet. Transection at stereotaxic plane A3 was
at a level approximately midway between the rostral and caudal borders of the
superior colliculus. Transection at stereotaxic plane A0 was near the midcollicu-
lar level. Only those experiments in which the transections were complete and

the planes of transection were within +0.5 mm of the A3 and APO planes of Jasper

53

and Ajmone-Marsan (1954) are included in the RESULTS. To identify unit (or
multiunit) recording sites and stimulation sties, frontal sections (30-um thickness)
of the diencephalon were cut on a cryostat microtome and stained with cresyl
violet. Diencephalic lesions were reconstructed by projecting serially arranged
frontal sections onto the stereotaxic planes of Jasper and Ajmone-Marsan (1954).
Sites of stimulation and unit recording in the diencephalon were also recon-
structed from stained frontal sections. Since I recorded the depth of all
stimulation and recording sites for each electrode penetration, these sites were
reconstructed using the end of the tracks as reference points. Shrinkage during

fixing was taken into account.

B. Data Analysis

Data were analyzed either on- or off-line. On-line analysis used a Nicolet
MED-80 computer. Off-line analysis from magnetic tape was performed by using
either a Nicolet MED-80 computer or an RC Electronics Computerscope System.
For this purpose arterial pressure, neural activity and pulses derived from the R
wave of the ECG or stimuli applied to the neuraxis were stored on magnetic tape.
The data from tape were analyzed after passing SND, EEG and diencephalic
multiimit activity through an A.P. Circuit Corp. variable filter with the low pass
cut-off frequency set at 50 Hz. The following methods of analysis were used.

1. Quantification of SND and blood pressure

In Projects A and D, after preamplification the sympathetic nerve

signal was led to a Grass 7P10 cumulative integrator. Raw and integrated SND
were displayed on a polygraph (see RESULTS; Figure 4). Changes in integrated
SND were quantified by comparing mean epoch length (integrated records).
Corrections were made for the integration of noise. The noise level was measured

following ganglionic blockade with hexamethonium chloride (5 mg/kg, iv). Mean

54

blood pressure was calculated from pulsatile records of blood pressure recorded on
the polygraph. The mean blood pressure (BP) was calculated using the formula:
BP = Pd + 1/3 (Ps - Pd)
where. Pd is the diastolic pressure and Ps is the systolic pressure.
2. Crosscorrelation and autocorrelation analysis

Autocorrelation was used to extract the rhythmic components in
inferior cardiac SND, EEG, and thalamic and hypothalamic multiunit activity.
Forty seconds of data were analyzed. Crosscorrelation analysis was used to
determine whether SND, multimit diencephalic activity and the EEG contained
common components that were temporally correlated. Normalized crosscorrela-

tion fimctions [ny(T)] N were calculated by using the formula:

 

ny(Tl
R T N =
[ xy( )1 {Rxx(0)Ryy(O)} {} = square root

where ny(T) is the nonnormalized crosscorrelation function in watts (voltage
squared); x and y are the input signals; T is the lag in the crosscorrelation
function; Rxx(0) and Ryy(0) are values (in watts) at zero lag in the nonnormalized
autocorrelograms. The resolution was 20 ms and 10 ms for the crosscorrelation
and autocorrelation analysis.
3. Post-R wave interval analysis

Trigger pulses coincident with the R wave of the ECG were used to
construct normalized averages (minimum of 500 trials) of the arterial pulse wave,
inferior cardiac SND, EEG and diencephalic multiunit activity. Post-R wave
interval histograms of unit discharge were also constructed. These analyses
revealed whether SND, EEG, multiimit diencephalic activity and diencephalic unit

discharges were temporally related to the cardiac cycle.

55

4. Post-stimulus analysis
Trigger pulses coincident with stimuli delivered through metal elec-
trodes in the diencephalon or cortex were used to construct normalized averages
of evoked responses in the inferior cardiac nerve and histograms of unit responses.
From these averages the modal onset latency of synaptic activation or inhibition
of the neuron and onset latency and time to peak of the evoked sympathetic nerve
response were calculated.
5. Power spectral density analysis
This analysis, performed by fast Fourier transform (Bendat and Pier-
sol, 1971), was used to examine the frequency components and relative power (in
watts) of each frequency band in SND, EEG and diencephalic activity. Power is
plotted against frequency. Aliasing of high frequency components into the 0- to
lS-Hz frequency bands was prevented by filtering the signals (15 Hz low pass cut-
off) prior to analysis. Spectra were constructed by averaging six 20-s data blocks.
The resolution of the analysis was 0.1 Hz.
6. Peri-spike-trigger averaging
Normalized averages (minimum of 500 trials) of SND and EEG activity
were acquired before and after the spike trigger (a naturally occurring unit spike).
For this purpose, the imit discharge was passed through a window discriminator.
A standardized pulse coincident with unit spike occurrence was used to trigger
computer sweeps. The portion of the average to the left of time zero was
computed from data that preceded the spike trigger, while the second half (right 0
lag) represents data that followed the spike. Thus it was possible to determine
whether the activity of the neuron was temporally related to changes in SND or
EEG which preceded or followed its discharge. A random pulse train with about
the same frequency as the neuronal spike train was used to trigger a "dummy"

average of SND and EEG. Neurons were considered to have sympathetic nerve-

56

related and/or EEG-related activity if the amplitude of the first peak to the right
of time zero in the spike-triggered averages of SND and EEG exceeded by at least
a factor of three that of the largest deflection in the "dummy" averages.
7. Interspike interval analysis

This routine constructs a histogram of the distribution of the intervals
between consecutive spontaneously occurrhig neuronal spikes. A minimum of 500
spikes are used. This method was used to calculate the mean frequency of
discharge of the individual diencephch units as well as to gain information on the
periodic components in the spike train.

8. Statistical analysis

Data are expressed as mean 1 S.E. The Student's t-test was used to
compare changes in mean blood pressure between lesioned and nonlesioned groups.
Changes in mean SND between lesioned and nonlesioned groups were compared
using the Students's t-test after transforming the data using arcsin square root
transformation. A random block design analysis of variance (ANOVA) was used to
analyze the effect of time on mean blood pressure after midbrain transection.
The least significant difference test (LSD) was used for individual comparisons.
Regression analysis was used to compare the intervals between the spontaneous
unit spike to peak SND and stimulus to peak SND intervals in the spike-triggered
averages of diencephalic neurons with activity related to SND. In all cases, a 95%

confidence level was used as a criteria for significance.

C. Project A: Experiments Involving Midbrain Transection and Radio-fre-
quency Lesions

l. Decerebration procedures
Following removal of the appropriate portions of the parietal bone and
dura mater, serial transections at stereotaxic planes A3 and APO were made with

spatulas constructed so that their diameters and contours corresponded with those

57

of the left and right halves of the midbrain. The spatulas were held in a
stereotaxic electrode manipulator. Left and right hemisections were performed
sequentially at each stereotaxic plane. The two-stage transection was completed
within 90 s without excessive bleeding or subsequent swelling of the brain stem.
The experiment was terminated on those rare occasions when transection caused
excessive bleeding or swelling of the brain.
2. Radio-frequency diencephalic lesions
Diencephalic lesions were made bilaterally with a radio-frequency
generator (Radionics Inc., Model RFG-4). An unmodulated sine wave (500 Hz) was
passed through an electrode (0.7 mm tip diam: 1.5 mm uninsulated tip length)
stereotaxically placed into the diencephalon. Current was adjusted to maintain
the temperature of the tissue contacting the thermal sensing electrode tip at 75°C
for 1 min. A lesion of 1-2 mm in diameter was produced at each site of current
application.
3. Data analysis
Changes in SND and blood pressure were quantified as described in
section 3.1. The level of activity during the 5-min period before midbrain
transection was compared with that during the first 2-min period, the following 3-
min period and then during subsequent 5-min periods after midbrain transection.
Either mean blood pressure or change in mean blood pressure was plotted at these
specified intervals following midbrain transection.
4. Histology
The completeness and level of transection were verified and lesions

reconstructed as described in section A.4.

58

D. Project B: Multiunit Diencephalic Activity

CNS-intact or decorticate cats with intact or sectioned baroreceptor nerves
were used in this study. Portions of the parietal bone and underlying dura were
removed bilaterally providing access to the diencephalon. The exposed neural
tissue was covered with paraffin oil to prevent drying. In those instances in which
decortication was performed, the skull was removed from the coronal suture to
the lambdoidal suture in a rostral caudal direction and bilaterally as close to the
squamosal suture as possible. Decortication was performed in two steps. First, a
bilateral frontal lobotomy was performed 1 mm caudal to the ansate sulcus.
Second, the cortex, hippocampus, corpus callosum and fornix surrounding the
diencephalon and basal ganglia were removed by suction under visual observation
with 6X magnification. The completeness of decortication was evaluated by gross
inspection and subsequent examination of histological sections. In two cases there
was damage to the anterior portions of the caudate and in all cases a small
portion of the pyriform cortex remained connected to the diencephalon.

1. Multiunit diencephalic recordings

Multiunit activity in the thalamus and hypothalamus was recorded

(capacity-coupled preamplifier band pass, 1 to 1,000 Hz) by using concentric
bipolar stainless steel electrodes (Rhodes Medical Instruments, Model NE-100)
with exposed leads (0.5 mm) separated by 0.5 mm. The electrodes were placed
stereotaxically into the diencephalon according to the coordinates of Jasper and
Amjone-Marsan (1954). The multiunit thalamic and hypothalamic activity was
displayed on the polygraph and oscilloscope.

2. Electrical and chemical stimulation of the diencephalon

The diencephalon was electrically stimulated and sympathetic poten-

tials averaged as described in section B.4. In a series of experiments, chemical

stimulation was accomplished by microinjecting the GABA antagonist picrotoxin

59

(5 1.1g in 0.5 ul saline) bilaterally over a 2 minute period into selected thalamic
sites by using a 26 gauge needle attached to a 10 111 Hamilton syringe. The needle
was placed stereotaxically according to the coordinates of Jasper and Amjone-
Marsan (1954).
3. Data analysis

The following methods of computer analysis were used: 1) crosscorre-
lation and autocorrelation analysis. 2) Post-R wave interval averaging. 3) Power
spectral density analysis. 4) Post-stimulus averaging. The methodology and

application of these analyses are described in section B.

E. Project C: Characterization of Diencephalic Neurons with Sympathetic
Nerve-related Activity
1. Diencephalic unit recordings
Metal microelectrodes (Frederick Haer, 2-um tip diam, 2-3 mega ohm
tip impedance, 0.5 mm exposed tip length) were used to record extracellularly the
action potentials of thalamic and hypothalamic neurons from the left side of the
diencephalon in baroreceptor-innervated cats. The indifferent electrode was a
gold-plated disc affixed to the frontal bone. Capacity coupled preamplification
with a band pass of 0.3-3 KHz was used. The position of the recording electrode
was controlled by using a David Kopf Instruments stepping hydraulic microdrive.
Stereotaxic placement of the electrodes was made according to the coordinates of
Jasper and Ajmone-Marsan (1954). The unitary nature of the recordings was
judged on the basis of the constancy of action potential shape, amplitude and
duration. For imits which discharged as high frequency bursts, the comparison
was made for the first spike in successive bursts. The majority (60%) of unit
recordings were from the soma-dendritic region of the neurons. This was

indicated by the presence of an inflection on the rising phase of the action

60

potential (see RESULTS; Figure 2232, C2). These inflections likely reflect
separation of the action potentials of the axon hillock and the soma dendritic
region (Coombs gt gt., 1957).
2. Electrical stimulation
A Grass 888 stimulator and a PSIU-6 stimulus isolation unit were used
to deliver square wave pulses (90-800 uA, 1.0 ms duration) to selected recording
sites in the thalamus and hypothalamus. Stimuli were passed through the
recording microelectrode. The strength of the stimulus was measured as
described in section A.3.
3. Effect of baroreceptor reflex activation
Baroreceptor reflex activation was produced by either the bolus
injection of norepinephrine bitartrate (1-2 jig/kg, iv) or by inflating the balloon-
tipped end of a catheter (Fogerty embolectomy catheter, Model 12 A-100-4F)
placed into the abdominal aorta via the femoral artery. The firing rate of
selected thalamic and hypothalamic neurons was assessed during baroreceptor-
induced reflex inhibition of SND. These procedures do not affect unit activity or
SND in baroreceptor-denervated cats (Barman and Gebber, 1987).
4. Data analysis
The following methods of computer analysis were used: 1) Peri-spike-
triggered averaging. 2) Interspike interval analysis. 3) Post-R wave interval
analysis. 4) Post-stimulus averaging and histogram construction. 5) Autocorrela-
tion analysis. The methodology and applications of these analyses are discussed in

section B.

F. Project D: Sympathetic Responses Elicited by Cortical Stimulation
Experiments were performed on baroreceptor-intact and -denervated cats.

Baroreceptor denervation was performed as described in section A.1. In four

61

animals the meninges on the left cerebral hemisphere were deafferentated by
sectioning the left trigeminal nerve at a point between the medulla and the
sensory ganglion. The trigeminal nerve was exposed by removing portions of the
occipital and basisphenoid bones and the tympanic bulla. EEG activity was
recorded using either a plate electrode (as described in section A.3) or a wick
electrode. Wick electrodes consisted of a small (2x6 mm) piece of cotton wrapped
with silver wire. The electrode was then soaked in saline and placed directly on
the cortical surface. The indifferent lead was an alligator clip placed on crushed
scalp muscle.
1. Electrical stimulation of the cortex
The prefrontal and sensorimotor cortex was exposed bilaterally by
removing the appropriate portions of the frontal and parietal bones and dura
mater as described in section C.1. The exposed cortex was covered with warm
paraffin oil. Electrical stimuli were applied to the cortex using procedures
described in section A.3.
2. Diencephalic tmit recordings
In some experiments the action potentials of medial thalamic neurons
with sympathetic nerve-related activity were recorded. The responses of these
neurons to cortical stimulation were recorded on magnetic tape for later off-line
construction of post-stimulus histograms (see section 3.4).
3. Data analysis
Computer-aided analyses and changes in blood pressure and inferior
cardiac SND before and after lobotomy and A3 transection were quantified as

described in sections 3 and C.3.

62

G. Drugs Used
The following drugs were used in this study. alpha-chloralose (Sigma),
gallamine triethiodide (Lederle), hexamethonium chloride (Mann Research Labora-

tories), atenolol (Sigma) and picrotoxin (Sigma).

RESULTS

L Identification of diencephalic regions contributing to SND

Huang 3 a_l. (1987) reported that decerebration in the anesthetized cat
produces transient (< 30 min) decreases in inferior cardiac postganglionic SND
(reduced to 62:7% of control) and blood pressure (33:4 mmHg). Three observa-
tions led Huang gt gt. to conclude that the transient effects of midbrain
transection (i.e., decerebration) reflect the loss of a forebrain-dependent compo-
nent of SND in the anesthetized cat rather than a nonspecific phenomenon such as
generalized trauma or the mechanical stimulation of a descending sympathoinhibi-
tory system. First, following recovery from the effects of decerebration, a
second transection more caudally in the midbrain (APO) fails to affect SND and
blood pressure. Second, crosscorrelation analysis revealed that a component of
SND is synchronized to frontal-parietal cortical activity only in those experiments
in which subsequent midbrain transection reduces SND > 30%. Third, the effects
of the initial transection are prevented by prior removal of the hypothalamus and
those medial thalamic nuclei located 0 to 4 mm lateral to the midline.

The diencephalic lesions made by Huang et a1. were extensive. Thus, the
purpose of my first set of experiments was to localize those regions of the
hypothalamus and/or medial thalamus responsible for the forebrain-dependent
component of SND. For this purpose, the effects of midbrain transection on SND
and blood pressure in nonlesioned control cats were compared with those in cats in

which selective diencephalic lesions were made with radio-frequency current.

63

64

Figure 1 shows examples of histological sections used to reconstruct the dience-
phalic lesions made in these experiments. The lesions in panels A, 3, and C were
in the lateral hypothalamic region, the posterior hypothalamic region and the
medial thalamus, respectively.

A. Midbrain transection in nonlesioned cats

The effects of midbrain transection on inferior cardiac SND and mean
blood pressure in 23 nonlesioned baroreceptor-denervated cats are summarized in
Figure 2A and B, respectively. SND was reduced to 65:5% of control and mean
blood pressure was reduced 35:3 mmHg in the first 2 min after midbrain
transection at A3. These changes were statistically significant. .There was a
tendency for recovery toward control levels, however, in contrast to the results of
Huang gt g. (1987) SND and blood pressure were still significantly reduced at the
end of the 30 min observation period.

In six of the 23 nonlesioned cats, renal as well as inferior cardiac SND
were recorded. As shown in Figure 3, A3 midbrain transection produced parallel
changes in renal and inferior cardiac SND. Figure 4 shows the polygraph records
of raw and integrated SND and blood pressure from one of the six experiments in
this series. The reductions in SND and blood pressure produced by A3 transection
are shown in panel A. Subsequently, hexamethonium (5 mg/kg, iv) was used to
produce ganglionic blockade (panel B) .

B. Anterior medial hypothalamic lesions

Lesions placed in the anterior medial hypothalamus failed to attenuate
significantly the decreases in inferior cardiac SND and blood pressure produced by
midbrain transection (Figure 2). A3 transection was performed one hour after
completion of the lesions. Mean blood pressure (111:4 mmHg) just prior to
midbrain transection in these six experiments was not different from that (109:4

mmHg) in the 23 nonlesioned control cats.

65

Figure 1. Representative histological sections showing lesions of lateral
hypothalamus (A), posterior hypothalamus (3) and medial thalamus (C). Each
section is from a different cat. Calibration is 2 mm.

66

 

Figure l

67

Figure 2. Comparison of effects of midbrain transection on inferior cardiac
sympathetic nerve discharge (SND) and mean blood pressure (3P) hi nonlesioned
control cats (n=23) with those in cats with anterior medial hypothalamic (n=6) or
large medial hypothalamic (n=7) lesions. All cats were baroreceptor-denervated
and anesthetized with alphachloralose. Changes in SND (% of control) in panel A
were quantified by comparing mean epoch length (integrated records) during 5-
min period before transection with that during the first 2-min period, the
following 3-min period, and then during subsequent 5-min periods after midbrain
transection. Change in BP (mmHg) is shown. *Denotes a statistically significant

difference from corresponding data point on control curve. Values are means _t
SE.

68

J>

 

 

 

     
 

‘2—
2100 j " T
O \ s
U
H(3'
5Q 80 ..x Lg. Med. Hypo. Lesion
as :5 gm. Me‘d. Hypo. Lesion
onrro
g, 60
I-—1—r-—I—r——T——I
o 10 , . 20 30
B Time, mm
C) O . .
I ‘ a
E i'
5‘ ‘20 - /’ " -
:25
<
'40
W
o 10 , 20 30
Time, mm

Figure 2

69

Figure 3. Effect of midbrain transection on inferior cardiac (ICN) and renal (RN)
sympathetic nerve discharge (SND), and mean blood pressure (BP) in 6 nonlesioned
baroreceptor-denervated cats anesthetized with chloralose. SND was quantified
as described for Figure 2. BP (mmHg) is plotted against time after transection

(panel 3). Values are means 1 SE. *Denotes a statistically significant difference
from pretransection level.

 

70

 

Time, min

 

Time, min

Figure 3

 

71

A8333 8n— n m 3 was: 08E. .2m 8m >1 m: was 20H 8m >3 mp mm nomusunzso 8038a, .sofloomcsb £2338 8t.»
.3 a 1; .mxBE 3 03.830 8.58288.»on no non—vow: 3.88 m 5 >893. .nomuoommaoa Emma as”: $2 x88 < 88a
3 2.893. .585 0.8 5330s 020: 385.885 no mesons» 633935 was Bum 688.830 :33 nouflofimofiw «so
vouszonovuuoaooouousn “89832.8: m 3 09:33:. assoc Omaha—38.3 A25 :28.“ was 20: oranges .8233 .35
9380.3 803 .8 AM: 033803 umnozmndm was As m4 ens—a 8880.88 8 nofloomfiwuu £9338 mo 303mm .v mama

a. 0.5me

l: _. “see: a
7 all scenes 1.

 

 

2

It.

 

73

The extent of the anterior medial hypothalamic lesion shown in Figure
5 is typical for this series of experiments. This lesion destroyed most of the
paraventricular hypothalamic nucleus (PVH) and n. filiformis (Fil), and a small
portion of the dorsal hypothalamic area (aHd) between stereotaxic planes A11 and
A13. There was minimal involvement of the posterior hypothalamic (Hp) and
anterior hypothalamic (Ha) nuclei. A small portion of the fornix (Fx) on the right
side was destroyed in this particular experiment.

C. Large medial hypothalamic lesions

More extensive lesions of the medial hypothalamus attenuated the
decreases in inferior cardiac SND and blood pressure produced by midbrain
transection. Thus large medial hypothalamic lesions attenuated the forebrain-
dependent component of SND. The reductions in SND and blood pressure were
significantly smaller in these seven experiments than in nonlesioned animals
during the first 20 min and 10 min, respectively, after midbrain transection
(Figure 2). The absence of significant differences beyond 20 min reflects the
rapid, although partial recovery of SND and blood pressure towards control levels
following midbrain transection in nonlesioned cats. Mean blood pressure (107:5
mmHg) just prior to midbrain transection in this series of experiments was not
different from that in nonlesioned cats. Thus, it appears that between 30 and 60
min were required for complete recovery from the loss of the forebrain-dependent
component of SND produced by either midbrain transection or diencephalic
lesions.

The extent of the large medial hypothalamic lesion shown in Figure 6
is typical for this series of experiments. In addition to PVH and Fil in the anterior
medial hypothalamus, large portions of aHd, posterior hypothalamic (Hp) and the

ventromedial hypothalamic (Nva) nuclei were destroyed at more posterior

74

.88 N 8 838.2580 8880888889 $8.58?
.2 rug «£8808 mm8uun0> .Z .35 “$8.883 838.38; .2 .15 “82088 828.589, .2 .<> $503“? 8308.: .09
“8088989? .Z :8 $802808 .2 .mm “82830308 .2 rm 388852;: 88308889889 .2 .m>m “mm—85089882.
833393 .z .882 $855 52888 “some .882 8888 assume .2 .82 “83883288235 .2 .25 a:
n8mn08o580> 888888893 .83“ «880809 88888509»: in $28833 88888893 in 33838 88888889»:
.8m “558 in “888;: .z ..Fm “3.830.880.8280 .2 .nm $28833 828.5800 .2 JD “88.888 83880 40
“838880 .2 .8 “8:80:80 889.80 .00 “mm—8.380> .8238 .2 $4 m8m~8mn08 82088 .2 .52 $3888 838885093
80.8 .638 Age: 8882:08o82 88 8&3. an 808308 88 0.8 3o: 283830.538 88 80:83 08880.85 6008.3
838p no £083 8 8.888 8 .833 no 808m .8888509: 8808 .8238 8 .832 0>E8808080m .m 0.8me

75

m 0.89..“

 

76

.88 N 8 80383280 23.808
8808 AN “88880820322 888380> .2 .2; “82830800 0280860.“ .80& “88858008888 02 .828808 828.5800
.2 .202 “0.858888 89.30 .82 28.8030." 888.838 .2 .and «828988 888838 .2 .94 “8803 £03m .NE .22 no
888308 05 888 m 0.332 8 a8 0.8 8038303.: 688885083 88808 8 88002 0982 088880802 .c 0.892

77

c 0.39%

 

78

levels. Large medial hypothalamic lesions did not extend into the thalamus nor
did they involve the medial forebrain bimdle (M PB).
D. Lateral hypothalamic lesions

Lesions placed in the lateral hypothalamic region between stereotaxic
planes A9 and All attenuated the decreases in inferior cardiac SND and blood
pressure produced by midbrain transection. As shown in Figure 7, the reduction in
SND was significantly less in these 12 experiments than in nonlesioned cats during
the first 15 min after midbrain transection. The decreases in blood pressure were
significantly different during the first 10 min after midbrain transection. Mean
blood pressure (111:6 mmHg) just prior to midbrain transection in cats with
lateral hypothalamic lesions was not different from that of nonlesioned cats.

The extent of the lateral hypothalamic lesion in Figure 8 is typical for
this series of experiments. The lesion included portions of the lateral hypothala-
mic nucleus (HL), H1, H2, 21 and the MFB. The H1, H2 and 21 are considered as
part of the hypothalamus rather than the subthalamus in this thesis.

E. Medial thalamic lesions

Lesions of the medial thalamus (0 to 3.5 mm lateral to the midline)
between stereotaxic planes A7 and All also attenuated the decreases in inferior
cardiac SND and blood pressure produced by midbrain transection. As shown in
Figure 9, the reduction in SND was significantly less in these seven experiments
than in nonlesioned animals during the first 15 min after midbrain transection.
The decreases in blood pressure were significantly different only at 2 min after
midbrain transection. Mean blood pressure (114:8 mmHg) just prior to midbrain
transection in cats with medial thalamic lesions was not different from that in
nonlesioned cats.

A typical medial thalamic lesion is shown in Figure 10. The lesion

included most of the medial dorsal nucleus (MD), n. rhomboidens (Rh) and the

79

Figure 7. Comparison of effects of midbrain transection on inferior cardiac
sympathetic nerve discharge (SND) and mean blood pressure (BP) in nonlesioned
control cats (n=23) with those in cats with lateral hypothalamic lesions (n=lZ).
Same format as in Figure 2. *Denotes a statistically significant difference from
corresponding data point on control curve. Values are means 1‘. SE.

80

>

—o o

h C

E 100 '

o C

u - '
3 8° ' '

O

‘ . -0 Lot. Hypo. Lesion

£23 _ ‘0 Control

(I)

60
W
20 30

0 10
B Time, min
o: O
I " '
E a
Ex-zo
'0.
m
<
-40
m
Time! min

Figure 7

81

N 3 8030.32.00 .o can m gag 8 as 0.8 aomuom>0unn<

.88
.8282 0280205093 29.08— 038832302 .w 08mg

82

a 290E

 

83

Figure 9. Comparison of effects of midbrain transection on inferior cardiac
sympathetic nerve discharge (SND) and mean blood pressure (BP) in nonlesioned
control cats (n=23) with those in cats with medial thalamic lesions (n=7). Same
format as in Figure 2. *Denotes a statistically significant difference from
corresponding data point on control curve. Values are means 1 SE.

84

J>

      

E

3100] ' '

O

U

”5 80

58‘ .0 Med. Thai. Lesion

Q -a Control

5 60
m
0 10 . 20 30

Time, min

 

Time, min

Figure 9

85

.88 N 8 838.5280 .88—8088.88 .2 ..E «80888 .2 .22 «$88082 $8380.88 .2 .432 “8882008 88.3800 .2 .20
30 803208 05 823 o 88 m 885$ 8 8 0.8 08028303.: .8300“ 028885 82008 0>m8u80m0a0m .3 08mg

86

 

Figure 10

87

lateral habenula (HbL). In addition, portions of the following intralaminar nuclei
were destroyed: centralis lateralis (CL), centralis medialis (NCM), and paracen-
tralis (Pc). Although extensive, medial thalamic lesions did not extend into the

hypothalamus or the thalamic sensory relay and association nuclei.

II. Synchronization of SND t0 diencephalic multiunit activity
A. Multiunit recordings of diencephalic activity

Diencephalic activity was recorded between stereotaxic planes A7 and
All in 17 baroreceptor-denervated. and six baroreceptor-innervated cats. The
region from 0 to 3.5 mm lateral to the midline is referred to as the medial
thalamus. Recordings in this region were made from sites in the following
intralaminar, medial and midline thalamic nuclei: CL, MD, Pc, centrum medianum
(CM), reuniens (RE) and ventralis medialis (VM). Recordings in the lateral
thalamus were made from the following association and sensory relay nuclei:
lateralis dorsalis (LD), lateralis posterior (LP), ventralis posterolateralis (VPL) and
ventralis posteromedialis (VPM); and from nuclei ventralis anterior (VA) and
ventralis lateralis (VL) which project to motor regions of the neocortex (Jones,
1985). Recordings were made from the following hypothalamic nuclei: aHd, HL,
Hp, 111,112. and ZL

Representative recordings of diencephalic activity and inferior cardiac
SND in a baroreceptor-denervated cat are shown in Figure 11. As is the case for
cortical activity in chloralose-anesthetized animals (Camerer g a_l., 1977; Barman
and Gebber, 1980), diencephalic slow waves occurred primarily at frequencies
between 2. and 6 Hz. The frequency of sympathetic nerve slow wave occurrence
was also in the Z- to 6-Hz range. These points are illustrated by the oscillogra-
phic traces in Figure 11A and the power spectra in Figure 12A. As reported by

others (Andersen and Andersson, 1968; Andersson and Manson, 1971), the patterns

88

.88 N 8 888.8880 28808802
688883 88082 88 888808 8 3808.8 .3 800.888 0.080. M88802 m888u8o0 808888008088 89.0.83 803008 2880.3
“0 $8888 82> 888 Q2 80H >8 m 888 22m. .802 >1 3 8 80388880 80380> .88 2: 8 803888—80 88808802
$8888 13> 888 $8888 Q2 .QZm .0>83 0888 83038 .8 m0m880>8 0.83 2.808 «2 $8888 an; 888
D2 8m >3 cm 88 22m 88 >1 2: 8 80388280 80380> 80883 8808002280 888 85 8 >888m08 08808 88888
88883 88.8 22m «0 0008.3 8 880300308 88382 8808808808 .323 888088080808 828.580., 88 52V 888.808
882008 80888 8 0088 80.8 80880008 88888 088885 8.8088 888 88808 .QZm 0888.80 8030.88 4888888
5 08a 083 .3288 .28 0.88008 8003 8030.2 03 83 80.8 08 80088.2. 3. 6008.830 883 80830838088
80 80820808188000.8088 8 8 5va 0988088 0208 03088888 888 58888 0888889 .: 08mg

89

: 08mg

 

02

90

Figure 12. Power spectra of arterial pulse wave (AP), sympathetic nerve
discharge (SND) and thalamic activity in two experiments. Records in A are from
experiment illustrated in Figure 11. Medial and lateral thalamic activity in A
were recorded simultaneously from sites in the nuclei medialis dorsalis (MD) and
ventralis posterolateralis (VPL), respectively. Medial and lateral thalamic
activity were recorded simultaneously from nuclei paracentralis (Pc) and ventralis
posteromedialis (VPM), respectively, in experiment B. Each spectrum is the
average of eight lS-s data blocks. Frequency resolution is 0.06 Hz. Vertical
range of power (voltage squared) is the same in all traces.

91

 

 

 

 

 

 

 

 

 

 

/\ R [AP
A; SND
_A MD
F l l l
0 Frequency, Hz ‘5
B a AP
_4,/¢““Ak,__ Sh“)
A PC
A W
Ff l l _1
O 15

Frequency, Hz

Figure 12

92

of activity simultaneously recorded from medial and lateral thalamic and from
hypothalamic sites could be quite different. In the example shown in Figure 11A,
the predominant frequency of slow wave occurrence in the Z- to 6-Hz range was
higher in MD than in VPL (also see power spectra in Figure 12A). Neither SND
nor thalamic activity contained a cardiac-related component in baroreceptor-
denervated cats. This is indicated by the essentially flat post-R wave (ECG)
averages of SND and thalamic activity in Figure 113. Although not shown SND
contained a cardiac-related component in baroreceptor-innervated cats (Barman
and Gebber, 1980; Gebber, 1980), however, thalamic activity did not.
B. Patterns of relationship between SND and diencephalic activity
1. SND related only to medial thalamic activity

Crosscorrelation analysis was used to determine whether SND
and thalamic activity were temporally related. Inferior cardiac SND was
temporally related to medial but not to lateral thalamic activity in five
baroreceptor-denervated and two baroreceptor-innervated cats. Two of the
baroreceptor-denervated cats were decorticate. Figure 13 shows an example of
this pattern of relationship in a baroreceptor-denervated cat with an intact
neuraxis. The CM -> SND crosscorrelogram (Figure 133) shows inferior cardiac
SND that preceded (left of zero lag) and followed (right of zero lag) medial
thalamic activity. Note that the crosscorrelogram contains a sharp peak near
zero lag. This peak indicates that a component of SND and medial thalamic (CM)
activity were temporally related. The crosscorrelation function (rho value) at the
peak was 0.2. A value of 1.0 signifies a perfect relationship. A rhythm with a
period of approximately 240 ms appears in both the autocorrelogram of CM
activity (Figure 13A) and the CM -> SND crosscorrelogram. Thus, the component
common to SND and medial thalamic activity was in the 2- to 6-Hz range. In

contrast, the VA -> SND crosscorrelogram (Figure 133) was flat, indicating that

93

Figure 13. Selective realtionship between medial thalamic activity and sympathe-
tic nerve discharge (SND). A: autocorrelograms of SND and of activity recorded
from nuclei centrum medianum (CM) and ventralis anterior (VA) of medial and
lateral thalamus, respectively. B: crosscorrelograms. CM -> SND crosscorrelo-
gram shows SND relative to CM activity at zero lag. VA -> SND crosscorrelogram
shows SND relative to VA activity at zero lag. CM -> VA crosscorrelogram shows
relationship between activity at both thalamic sites. Analysis time for correlo-
grams was 40 s. Bin width was 10 ms.

94

 

 

 

 

 

 

 

 

 

 

 

 

 

10 0.20-
SND CM-—>SND
0 0..
r I F l l
1.0 '10 O +1.0
Interval, s Interval, s
10 0.20-
CM VA-DSND
O O-VWW
r
o ‘ Ll) -1'0 0 «71.0
Interval, s Interval, s
10 (130- ll
VA CM ->VA
O 0-
l l l l
‘10 0 +1.0
Interval, s In terval, s

Figure 13

95

SND was unrelated to activity recorded from this lateral thalamic site. The CM
-> VA crosscorrelogram in Figure 133 shows a sharp peak near zero lag indicating
that medial and lateral thalamic activity contained a common component. This
component apparently was not responsible for the relationship between medial
thalamic activity and SND.

As was the case in CNS-intact cats, 2- to 6-Hz slow waves
appeared in SND and thalamic activity of decorticate cats. Figure 14A shows
representative recordings of SND and medial (VM) and lateral (VPL) thalamic
activity in a decorticate cat. The crosscorrelograms in Figure 143, C demon-
strate that SND was temporally related to VM but not to VPL activity.

Recordings were made from 53 medial thalamic and 40 lateral
thalamic sites in the seven experiments in which SND was temporally related only
to medial thalamic activity. The distribution of thalamic recording sites in these
experiments is shown in Figure 15. Twenty-one of the medial thalamic sites had
activity related to SND. The mean interval between zero lag and the first peak to
the right of zero lag in the medial thalamic -> SND crosscorrelograms for these 21
sites (.) was 73:10 ms. SND was related to thalamic activity recorded in CL (2
of 7 cases), CM (4 of 8 cases), MD (8 of 21 cases), Pc (4 of 8 cases) and VM (3 of 6
cases). No relationship (0) was observed between SND and RE activity (3 cases).

2. SND related to medial and lateral thalamic activity

SND was related to both medial and lateral thalamic activity in
seven baroreceptor-denervated and three baroreceptor-innervated cats. Sites
with sympathetic nerve-related activity were distributed, without apparent
concentration, throughout the medial and lateral thalamus. The mean interval
between zero lag and the first peak to the right of zero lag in the medial thalamic
-> SND crosscorrelograms was 77:10 ms. The corresponding value for lateral

thalamic -> SND crosscorrelograms was 61:11 ms. In the representative example

96

68 3 no? 533 Sn .m 3. we? mfisumodou
£30295 now 683 $9322 .amumo~0uuoommouo DZw A: An; .0 .Snumofiouuoommouo OZm At :5 um in; was 55 you
>1 on one QZw new >1 am am noflmhfino 303.35 .bozuoomuou A153 mzmuouflohoumon «2239’ can 343 «23608
335:2, Magus: 05 3 mafia Baum covuouou figfiom 38323 3.8%: was 3:58 .QZm 6.36.80 .8233 $8633?
5 one: 083 Amman” .mmv 0.3329 @003 Eaton on no“ 89: 0.8 @6095 3. .30 counZoaovLou—noomuoudn can
3.33.3006 a 5 52$ 0923036 030: gaging—a? use .5333 883.05 3:58 nooBuon fifiaoflflom .3 05mg

97

E mama

 

 

 

 

 

 

 

 

a. 026 a. 9:2:
6... _ o .5 3.. Q? _ o _ QT
— _ _ b L
02m AI ...; 02w AI 2>
r . r
«.0 N8
U m
...;
_
<<>
02w
_ . _ _ _ 1
8.
an

 

nx—

98

 

.2 es. e .m «mama
am an 0.8 £8303 Banana no 3053603“: .88 — mm .8325de .QZm o» 6330.35 53306 525 mafia unannouaog
Bone @3930 :25 $33.28 ”8330980320 8:.» 639330806 an .QZm o» 6330.. hutfiom 5MB 83m 0zmgnooa0m6
Bonn «30.30 60ao~0 .335: own—Sufi 3.862 5 «on :5 038352;: 65 38.29: 63608 3 32,306 3 6322 an?
5va 0933036 ozon 0205.252: A023 3 munofimuomuo 5 mafia 956.803 0245900536 «0 nomuanmpummn .3 03mg

99

 

 

 

100

shown in Figure 16, SND was as strongly related to lateral thalamic (LP) activity
as to medial thalamic (CL) activity (see peak rho values in crosscorrelograms;
panel B). The predominant frequencies of medial and lateral thalamic slow wave
occurrence were essentially the same in these experiments (see autocorrelograms
of CL and LP activity in Figure 16A). Twenty-nine of 45 medial thalamic sites
and 27 of 43 lateral thalamic sites had activity related to SND in these
experiments.
3. SND related to hypothalamic activity

The relationship between SND and hypothalamic activity was
examined in 19 cats. SND was related to hypothalamic activity as well as to
thalamic activity in 13 cats. SND was related to medial thalamic but not to
lateral thalamic activity in five of these experiments. An example of the
relationship between SND and hypothalamic activity fromone of five experiments
in which SND was temporally related to medial but not to lateral thalamic
activity is illustrated in Figure 17. The mean interval between zero lag and the
first peak to the right of zero lag in the hypothalamic -> SND crosscorrelograms
was 59:13 ms (19 of 31 sites had activity correlated to SND. The distribution of
these sites is shown in Figure 15. Sites with sympathetic nerve-related activity
(C) were located in both the lateral and posterior hypothalamic nuclei. In the
remaining eight cats, hypothalamic and medial- and lateral thalamic activity was
related to SND. In an additional two cats SND was related to hypothalamic but
not to thalamic activity.

4. SND tmrelated to diencephalic activity

SND was unrelated to hypothalamic and medial- and lateral
thalamic activity in one decorticate and three CNS-intact cats with severed
baroreceptor nerves. Activity was recorded from a minimum of 20 diencephalic

sites in each of these experiments.

101

Figure 16. Nonselective relationship between thalamic activity and sympathetic
nerve discharge (SND). A: autocorrelograms of SND and of activity recorded
from nuclei centralis lateralis (CL) and lateralis posterior (LP) of medial and
lateral thalamus, respectively. B: crosscorrelograms. CL -> SND crosscorrelo-
gram shows SND relative to CL activity at zero lag. LP -> SND crosscorrelogram
shows SND relative to LP activity at zero lag. CL -> LP crosscorrelogram shows
relationship between activity at both thalamic sites. Analysis time for correlo-
grams was 40 s. Bin width was 10 ms.

102

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

lO
11
(
o-J o-
l I If I
O lntervol,s 1.0 1.0 lnte?vol,s 10
1.0
017
CL LP-vSND
O O
' fl
[0 '10 0 ’10
Interval, ’ In terval, s
1.0
0.25
LP CL-DLP
O O
I l h
0 Interval; 10 -10 Inleevah r10

Figure 16

103

Figure 17. Crosscorrelograms showing relationships of sympathetic nerve dis-
charge (SND) and activity recorded from thalamus and hypothalamus. Diencepha-
lic activity in this baroreceptor-denervated cat was recorded from sites in N.
medialis dorsalis (MD) of medial thalamus, the lateral hypothalamus (HL) and N.
lateralis posterior (LP) of lateral thalamus. Analysis time for crosscorrelograms
was 40 s. Bin width was 10 ms.

104

 

 

 

 

 

 

 

 

 

 

a16-
A MD->SND
o—m I
r l
-'l.0 0 +10
lntervo|,s
0.18- B
| HL-bSND
[— l
'10 0 +1.0
Intervals
0'06 C LP->$ND
O
9 ('3 +110
-10 lntervo|,s '

Figure 17

105

C. Electrical and chemical stimulation of the diencephalon
An important issue raised by the previous series of experiments is
whether the temporal relationships between diencephalic and sympathetic nerve
activity revealed by crosscorrelation analysis are indicative of a functional
connection between the diencephalon and the sympathetic nerves. Alternatively,
brain stem networks controlling SND and diencephalic circuits controlling other
functions may not be directly connected, but may receive input from a common 2-
to 6-Hz generator. Assuming that a functional connection exists, a second
question which arises is whether the diencephalon contains the cell bodies of
neurons capable of influencing SND. These questions were approached in two
stages. First, electrical stimulation was used to test for a functional connection
between the diencephalon and the sympathetic nerves. Second, chemical stimula-
tion was used to determine whether the thalamus contains the cell bodies of
neurons capable of influencing SND. In this study, the hypothalamus was not
chemically stimulated. In contrast to the thalamus, numerous studies using
chemical stimulation have shown that the posterior hypothalamus (Gellhorn and
Redgate, 1955; Redgate and Gellhorn, 1956a; 1956b; Lee gt g, 1972; Finch and
Hicks, 1977) lateral hypothalamus (Lee g a_l., 1972; Sun and Guyenet, 1986) and
the PVH (Rockhold gt a_l., 1987) contain the cell bodies of neurons capable of
‘ influencing blood pressure and heart rate.
1. Sympathetic nerve responses elicited by electrical stimulation
Increases in SND were elicited by lO-ms trains of three pulses
applied once every 2 s to sites in the diencephalon of 17 cats, three of which were
decorticate. Stimuli were applied to CL, CM, MD, Pc and W of the medial
thalamus, VA, VL and VPL of the lateral thalamus and to sites in the lateral and
posterior hypothalamus. Increases in SND could be elicited by stimulation of sites

in each of these nuclei. The onset latencies of the sympathetic nerve responses

106

elicited from different medial and lateral thalamic nuclei were statistically
indistinguishable. However, the mean onset latencies of the responses produced
by medial or lateral thalamic stimulation were significantly longer than that for
the responses elicited by hypothalamic stimulation (Table 1). This was the case
even when stimuli of supramaximal intensity (1 mA) were used. Typical examples
of averaged sympathetic nerve responses (32 trials) elicited by thalamic and
hypothalamic stimulation in a CNS-intact and in a decorticate cat are shown in
Figure 18.

In five cats the diencephalon was explored to locate those sites
from which sympathetic nerve responses could be elicited with the lowest
stimulus current (<30 uA in some cases). The concentric bipolar stimulating
electrode was moved vertically in 1 mm steps. Threshold current for eliciting a
sympathetic nerve response and the depth of the electrode were recorded for each
site of stimulation. Eleven depth-threshold curves were constructed from these
data. Representative examples are shown in Figure 19. Electrode tracks passing
through the posterior portion of the medial thalamus and underlying midbrain
yielded the type of depth-threshold curve shown in Figure 19A. Note that the site
requiring the least stimulus current (60 uA) for eliciting an increase in SND was
located in CM of the medial thalamus. Thus these responses were most likely not
due to current spread to more ventral structures. Electrode tracks passing more
anteriorly through the medial thalamus and underlying hypothalamus yielded
depth-threshold curves that were broader in contour. In the case shown in Figure
19B, the electrode track extended through MD and Pc of the medial thalamus and
the HL. Note that sympathetic nerve response could be elicited with stimulus
currents below 100 uA applied to widely separated sites in the medial thalamus

and lateral hypothalamus. Sympathetic nerve responses also were routinely

107

TABLE 1. Mean onset latencies of inferior cardiac
sympathetic nerve excitation produced by diencephalic
stimulation with lO-ms trains of three pulses of supra-
maximal intensity (1 mA).

A. Cats with intact neuraxis (n=l4)

 

 

 

No. of Sites ' Mean Onset i S.E.
MTh 78 69 j; l ms‘I
LTh 30 67 1 2 ms"
HTh 27 60 1 2 ms

 

B. Decorticate cats (n=3)

 

 

 

No. of Sites Mean Onset _+_ S.E.
MTh 7 64 i 3 ms“
LTh 3 64 i 4 ms“
HTh 5 52 1 2 ms

 

MTh, LTh and HTh are medial thalamus, lateral thalamus
and hypothalamus, respectively. * indicates that mean
onset latencies of responses elicited by MTh and LTh
stimulation were significantly (P<0.05) longer than that
of the responses evoked by HTh stimulation.

108

Figure 18. Sympathetic nerve responses elicited by diencephalic stimulation (10-
ms trains of three l-mA pulses applied once every 2 s) in a CNS-intact (A) and a
decorticate (B) cat. Diencephalic nuclei are labeled as in Figures 5, 6 and 10.
Calibration is 1 mm. Post-stimulus averages (32 trials) of inferior cardiac
sympathetic nerve responses. Responses are numbered to correspond with the site
of stimulation. Bin width was 1 ms. Vertical calibration is 100 uV and horizontal
calibration is 100 ms.

 

Decorticate

 

 

B

A10

 

Figure 18

110

Figure 19. Depth-threshold curves for sympathetic nerve responses elicited by
medial thalamic, hypothalamic and midbrain stimulation with lO-ms trains of
three pulses. Stereotaxic horizontal plane (H) is plotted against threshold current
(11A) for eliciting a response. Nuclei are labelled as in Figures 5, 6 and 10. Data
in A and B are from different experiments.

111

‘5

 

 

- --+4
+2 -/
VCM a o
/ NR u ......— .2
y gllflllllllzgo
Threshold Current, pA

 

 

 

 

 

I I I I I I I I I I I jO
44
Threshold Current, pA

 

 

Figure 19

112

elicited when the stimulus current applied to lateral thalamic sites was less than
100 uA.
Z. Intrathalamic injection of picrotoxin

Picrotoxin, a CNS stimulant with GABA antagonistic properties
is believed to exert its action on synaptic networks rather than on fibers of
passage (Johnston, 1976). SND was monitored before and after this drug was
injected into the medial thalamus of five baroreceptor-denervated cats. One or
two series of bilateral injections (5 ug picrotoxin in 0.5 111) were made between
stereotaxic planes A9 and A1 1, 0.5 to 1.5 mm lateral to the midline at H+1 or
H+Z. Fast green FCF dye injected at these sites at the end of each experiment
spread to portions of CL, Pc, MD, NCM and RE of the medial thalamus. No dye
was observed in the ventricular system, lateral thalamus (> 3 mm lateral to the
midline) or the hypothalamus.

Figure 20 illustrates the changes in SND observed in 4 of 5 cats
approximately 10 min after picrotoxin was injected into the medial thalamus. The
intrathalamic injection of picrotoxin led to an increase in the amplitude of the 2-
to 6-Hz sympathetic nerve slow waves and a decrease in the variability of
interslow wave intervals (Figure 2011A). The latter effect indicates that SND
became more rhythmic. The reduction in the variability of interslow wave
intervals after picrotoxin administration is reflected by the sharp peak between 1
and 2.5 Hz in the power spectra of SND in Figure ZOIIB. The changes in SND
produced by the injection of picrotoxin into the medial thalamus were accompa-
nied by a rise in blood pressure (20 and 40 mmHg) in 2 of the 4 experiments. The
failure of intrathalamic picrotoxin to raise blood pressure in 2 of 4 animals
suggests that although picrotoxin altered the rhythm in SND in these animals, it
did not sufficiently increase total sympathetic nerve activity to produce a pressor

response. No quantitation of total SND was made in these experiments.

113

Figure 20. Changes in sympathetic nerve discharge (SND) produced by injection
of picrotoxin into the medial thalamus. I: control. II: 10 min after bilateral
injection of picrotoxin (total dose, 10 ug) into the medial thalamus at stereotaxic
plane A10. Traces in IA and HA are from top to bottom blood pressure (BP,
mmHg), SND and frontal-parietal cortical activity (EEG). Vertical calibration is
100 uV for SND and EEG. Horizontal calibration is 500 ms. Traces in IB and IIB
power spectra of SND. Traces in IC and 11C are power spectra of EEG. Total
power in EB was 387% of that in IB. Total power in IIC was 128% of that in IC.
Each spectrum is the average of 20 6—s data blocks. Frequency resolution is 0.1
Hz.

 

114

I Control IIPicrotoxin
I75 A
AP

”LVN/WW Kim
MAM/WWW” °VVVVVVVW

::u./M\l/1V(wwl1 l/W GWW

 

 

 

 

 

 

 

 

 

 

 

]. 5‘4” L1
SBND A“ A SBNDIH -\
. \I \
Wfllf/ \“V‘ ‘3 . // \a
0.. _/ \f‘r‘flw 0‘ _/ WM-
[— I l h I I
0 5 10
E Frequesncy, Hz ‘0 C Frequency. Hz
l- L5-
EEG {Ix/l” V) EEG
WM WM {If ’\.
A "W“ ” W WW
o-J o- f
r l #I r I I
5 IO 0 5 )0
Frequency, Hz Frequency, Hz

Figure 20

115

Recovery of SND and blood pressure to near control levels occurred between 1
and 1 1/2 hr after picrotoxin injection (not shown). After recovery, the same
pattern of results could be obtained following a second series of picrotoxin
injections into the medial thalamus (not shown). In one cat intrathalamic
picrotoxin (110 ug total dose) failed to alter SND.

As seen in the raw records (Figure 201A, IIA) and power spectra
(Figure 201C, IIC) of EEG activity, the intrathalamic injection of picrotoxin
increased the amplitude and frequency of the EEG (4 of 5 cats). However, these
effects were not as dramatic as the effects on the SND as indicated by the
smaller increase in power and the broad peak in the power spectrum of the EEG
(Figure ZOIIC). In two cats larger doses of picrotoxin (60 and 90 1.1g total dose)
changed the synchronous 1- to Z-Hz rhythm in SND into a less synchronous pattern
dominated by periods of spike-like activity. The spikes in SND were locked to
spikes in the EEG in a 1:1 fashion. The degree to which these larger injections
volumes spread was not assessed in these experiments. The periods of synchro-
nized sympathetic and cortical spikes resembled those seen after intravenous
administration of convulsive doses of picrotoxin (0.5-1.0 mg/kg). The polygraphic
records of EEG and SND in Figure 21 are representative of the patterns of
activity observed during picrotoxindinduced (1 mg/kg, iv) seizures in eight cats.
Note the wide variety of activity patterns and the strong 1:1 synchronization of
EEG and SND (panels B, C and D). The changes in blood pressure (increases and
decreases) of 10-35 mmHg shown in the example in Figure 21 are typical of those
accompanying picrotoxin-induced seizure activity in all eight cats.

In contrast to the changes produced by injection of picrotoxin
into the medial thalamus, multiple injections of normal saline (0.5 111) into the
same regions failed to affect SND (3 experiments). SND also was unaffected

when bilateral injections of picrotoxin (IO-20 11g total dose) were made into VL,

116

Figure 21. Changes in sympathetic nerve discharge (SND) and frontalparietal
cortical activity (EEG) produced by the intravenous (iv) administration of a
convulsive dose of picrotoxin. A: control. B, C and D, are 3, 5 and 7 min,
respectively, after iv injection of 1 mg/kg picrotoxin. Traces in A, B, C and D are
from top to bottom blood pressure (BP, mmHg), SND and EEG. Vertical
calibration for SND is 100 11V and EEG is 80 11V. Time base is l s/division.

 

 

 

118

VPL and LP of the lateral thalamus (4 experiments) or when the drug was injected

intravenously in a dose of 250 ug/kg (2. experiments).

III. Identification of diencephalic neurons with sympathetic nerve-related acti-
vity

The experiments with chemical stimulation in this and other studies suggest
that the medial thalamus and hypothalamus contain the cell bodies of neurons
capable of influencing SND. However, the question of whether these cells are
tonically active thus accounting for the temporal relationship between SND and
multiunit diencephalic activity required further investigation. To test this I
recorded from individual neurons in the medial thalamus and hypothalamus.

Whereas the experiments with radio-frequency lesions were performed on
baroreceptor-denervated cats, single neurons were studied in cats with intact
baroreceptor nerves. This allowed me to characterize baroreceptor reflex
influences on diencephch neurons with sympathetic nerve-related activity.
Although the decreases in SND and blood pressure produced by decerebration are
most pronounced in nonlesioned baroreceptor-denervated cats, qualitatively simi-
lar effects are observed in baroreceptor-innervated cats (Huang 31 al., 1987).

A. Identification of diencephalic neurons with sympathetic nerve-related
activity

In 30 cats, the hypothalamus was explored from 8.5 to 14 mm rostral
to the interaural line and between 0.5 and 3.5 mm lateral to the midline for
neurons whose spontaneous discharges were temporally related to postganglionic
inferior cardiac SND. In another 23 cats, the medial thalamus was searched from
7.0 to 11.0 mm rostral to the interaural line and from 0 to 3.0 mm lateral to the

midline for neurons with sympathetic nerve-related activity.

119

1. Diencephalic tmit discharge patterns

The unit discharge patterns illustrated in Figure 22 are typical of
those of diencephalic neurons with sympathetic nerve-related activity. Most
medial thalamic neurons fired in bursts of three or more spikes. Interspike
intervals (18!) within a burst usually were < 10 ms. This was reflected by a large
number of counts in the first bin of the ISI histograms (Figures 26D, 27D, 29D and
30D). A second sharp peak corresponding to a frequency between 2 and 6 Hz
appeared in the ISI histograms for thalamic neurons with regularly spaced bursts
(Figures 27D, 29D and 30D). This peak was absent when the bursts were
irregularly spaced (Figure 26D). Raw records of the regularly spaced bursts of a
medial thalamic neuron are shown in Figure 22A1, whereas the irregularly spaced
bursts of another thalamic neuron appear in Figure 2231. Hypothalamic neurons
rarely discharged in a burst. These neurons were most apt to generate single
spikes or doublets (ISIs < 50 ms; Figure 22C1) that were usually irregularly spaced
(Figure 28D). Representative recordings of inferior cardiac SND and frontal-
parietal cortical activity are shown in Figure 22D.

2. Synchronization of diencephalic unit activity to inferior cardiac
SND

Peri-spike-triggered averaging was used to identify diencephalic
neurons with naturally occurring (i.e., spontaneous) activity temporally related to
inferior cardiac SND. Sixty of 310 hypothalamic neurons (including those in the
21 and H1, H2 Forel's Fields) and 47 of 196 medial thalamic neurons tested had
activity related to SND. Neurons with sympathetic nerve-related activity were
intermingled with those whose discharges were unrelated to SND. Recordings
were made from the vicinity of the cell body in most cases. This was indicated by
the appearance of an inflection on the rising phase of the spikes of 60% of the

neurons with sympathetic nerve-related activity. These inflections are believed

120

Figure 22. Representative recordings of action potentials of diencephalic neurons
with sympathetic nerve-related activity. A, B: two thalamic neurons located in
nucleus medialis dorsalis. C: hypothalamic neuron located in paraventricular
nucleus. Horizontal calibration is 200 ms for A1, BI and Cl, 1 ms for A2, C2, and
2 ms for 32. Vertical calibration is 200 uV for A and 100 Mi for B and C. Note
inflection on rising phase of action potential when spike amplitude decremented
during the burst in B2. Five successive spikes of hypothalamic neuron were
superimposed in C2. Note separation of action potential into two components.
This occurred for the second spike of a doublet. Arrows in A3-C3 mark locations
of neurons. Calibration for histological sections is 2 mm. D: representative
recordings of arterial pressure (AP; mmHg), inferior cardiac sympathetic nerve
discharge (SND), and frontal-parietal cortical activity (EEG). Horizontal calibra-
tion is 1 3. Vertical calibration is 200 11V for SND and 100 uV for EEG.

 

121

 

D
13°]memmwwuwmm0

SND

VMMVWWWNWMWWWWWW

Figure 22

122

to reflect separation of the action potentials of the initial segment and soma-
dendritic regions (Coombs g a_l., 1957). Examples are shown in Figure 2232 and
C2. The inflections were particularly evident when spike amplitude decremented
during a burst or doublet.

Peri-spike-triggered averaging revealed two patterns of relation-
ship between diencephalic unit activity and SND, both of which often occurred in
the same experiment. These patterns are illustrated in Figure 23. The first
pattern was characterized by synchronization of unit activity to an aperiodic
spike-like event in SND. The aperiodic component could be detected in raw
records of inferior cardiac nerve activity (Figure 23A). Neurons with activity
synchronized to the aperiodic component are referred to as type 1 units. A
typical example of the spike-triggered average (based on 729 spikes) of SND for
one of these neurons is shown in Figure 2331. The trace shows the average of
inferior cardiac SND that preceded (left of zero lag) and followed (right of zero
lag) the action potential of the hypothalamic neuron. Unit discharge was followed
by a sharp spike-like increase (upward negative deflection) in SND that was
succeeded by a positive wave and then by a second smaller negative wave. The
portion of the average to the left of zero lag was flat. Thus, type 1 unit
discharges were unrelated to preceding activity in the inferior cardiac nerve.
Diencephalic neurons were considered to have activity synchronized to SND when
the amplitude of the peak closest to zero lag in the spike-triggered average was
at least three times as great as the largest deflection in the "dummy" average of
SND (Figure 2332). "Dummy" averages of SND were constructed by using a
random pulse train whose mean frequency approximated that of the unit spike
train. Fortyfour type 1 neurons were located in the hypothalamus and 34 type 1

neurons were found in the medial thalamus.

123

Figure 23. Peri-spike-triggered averages of inferior cardiac sympathetic nerve
discharge (SND) for a type 1 hypothalamic and a type 2 hypothalamic neuron in
the same cat. A: oscillographic record of SND showing 2- to 6-Hz slow waves and
aperiodic spike-like events (marked by dots). Horizontal calibration is 0.5 s.
Vertical calibration is 200 11V. 31: normalized peri-spike-triggered average (729
trials) of SND for type 1 neuron. Unit spike occurrence is at zero lag. 32:
"dummy“ pulse-triggered average (729 trials) of SND. C1: peri-spike-triggered
(500 trials) of SND for type 2 neuron. C2: "dummy" pulse-triggered average (500
trials). Vertical calibration is 20 (N. Bin width for averages is 4 ms.

124

 

 

 

 

SND
o

Unit-DSND Unit->SND
1 MW 1 W |

Pulse->SND Pulse->SND

r 1 F I
-l.0 0 +1.0 ‘10 O +1.0

Interval, : Interval, s

Figure 23

125

The second pattern of relationship is illustrated in Figure 23C1.
This pattern was characterized by synchronization of unit activity to a 2- to 6-Hz
rhythmic component in inferior cardiac SND. Neurons of this kind are referred to
as type 2 units. Their spike-triggered averages of SND contained a 2- to 6-Hz
rhythmic component both to the left and right of zero lag. This rhythm was
clearly discernible in raw records of SND (Figure 23A). Type 2 neurons were
located in the hypothalamus (n=16) and in the medial thalamus (n=13).

The distributions of type 1 (left side of coronal sections) and
type 2 (right side) neurons in the hypothalamus and medial thalamus are shown in
Figure 24. The two neuronal types generally were intermingled. The hypothala-
mic regions containing these neurons included: Fil, H1, H2, HL, Hp, PVH and the
21. In addition, three type 1 neurons were located in the preoptic region (RPO) of
the hypothalamus. Medial thalamic nuclei containing type 1 and type 2 neurons
included: CL, CM, MD, Pc, RE and VM. CL, CM and PC are part of the
intralaminar complex.

3. Comparison of unit firing times

The firing times of type 1 and type 2 neurons relative to the
spike-er event and 2- to 6-Hz rhythmic components in inferior cardiac SND,
respectively, were assessed from spike-triggered averages. I measured the
interval between unit spike occurrence and the first peak to the right of zero lag
in the average of SND. The distributions of the intervals for type 1 and type 2
neurons are shown in Figure 25. The mean intervals for corresponding unit types
in the hypothalamus and medial thalamus were not significantly different.
Further analysis (not shown) failed to reveal differences in unit firing times on the

basis of location in a particular nucleus.

126

Figure 24. Anatomical distribution of diencephalic neurons with sympathetic
nerve-related activity. Type 1 (Q) neurons are shown on left of coronal sections:
type 2 neurons (A) are shown on right. Calibration is 1 mm. Stereotaxic planes
and abbreviations are the same as those in Figures 5, 6 and 10 with the addition
of: CA, commissura anterior; Ch, chiasma opticum; RPO, Regio preoptica.

127

 

 

 

Figure 24

128

Figure 25. Firing times of type 1 (A,B) and type 2 (C,D) hypothalamic and medial
thalamic neurons relative to inferior cardiac sympathetic nerve discharge (SND).
Number of neurons is plotted against interval between unit spike occurrence and
the first peak to the right of zero lag in the spike-triggered average of SND.
Values are means 1 SE.

129

A Hypo. Type I C Hypo. Type IT.

 

 

 

 

 

 

 

 

 

 

 

 

20—.
_ 9924"“ 85:14m:
§
5 3-
: J‘ V"
2—
o
O E—
Z -l- a
Z
- “6
Frl‘ d... J ..
Z
'1 Fl]
0 I I I I I 0 I I I I I
0 Intervalms 250 0 Intervalms 250

B Thol. Type I D Thol. Type I[

 

 

 

 

 

 

 

 

 

 

 

 

l0 -
_ rL 10636ms "4:13:11:
2
‘0' 3
. _ o
29 i?
Z - ~ -
3.
- 2
° I I FT I ° I I I l I
O Interval, ms 250 0 lnlerval, ms 250

Figure 25

130
3. Detailed analysis of diencephalic neurons with sympathetic nerve-
related activity
1. Detailed analysis of type 1 neurons
Post-R wave analysis was performed off-line for 12 type 1
hypothalamic and 11 type 1 medial thalamic neurons. The relationships between
unit activity, SND, cortical activity and the cardiac cycle were similar for these
23 neurons. A typical example is shown in Figure 26. The discharges of this
medial thalamic neuron were synchronized to spike-like events in SND (Figure
26A) and cortical activity (Figure 263). While the sharp increase in SND followed
the action potential of the neuron (i.e., zero lag in the average), the onset of the
spike-like event in cortical activity preceded the unit action potential. This was
the case for 20 of the 23 type 1 neurons. The discharges of the other three type 1
neurons preceded the onsets of the spike-like events in cortical activity and SND.
As evidenced by the flat post-R wave histograms of :mit discharges, type 1
neurons did not have cardiac-related activity (Figure 26C). Type 1 neurons had
irregular firing patterns as indicated by the absence of sharp peaks beyond the
first bin in their 181 histograms (Figure 26D). The mean firing rates of the 12 type
1 hypothalamic and 11 type 1 medial thalamic neurons were l.8;l_-0.8 and 3.0103
spikes/s, respectively.
2. Detailed analysis of type 2 neurons
Post-R wave analysis was performed for seven type 2 hypothala-
mic and seven type 2 medial thalamic neurons. The mean firing rates of these
hypothalamic and medial thalamic neurons were 2.9:? and 4.9115 spikes/s,
respectively. Five of the hypothalamic and three of the medial thalamic neurons
had cardiac-related activity. Figures 27C and 28C show the post-R wave
histograms (1118 and 1630 R wave triggers, respectively) of the discharges of two
of these neurons. As might be expected, the rhythm (3.6:0.2 Hz) in the spike-

triggered averages of SND for the eight type 2 neurons with cardiac-related

131

Figure 26. Type 1 medial thalamic neuron. A: normalized peri-spike-triggered
(trace 1) and "dummy" pulse-triggered (trace 2) averages (565 trials) of inferior
cardiac sympathetic nerve discharge (SND). Unit spike occurrence and "dummy'l
pulse were at zero lag. Bin width is 4 ms. 3: same for frontal-parietal cortical
activity (EEG). Vertical calibration is 10 1.1V for SND and EEG. C: normalized
post-R wave averages (2595 trials) of arterial pulse (AP), SND and EEG, and
histogram of thalamic unit spike occurrences. Bin width is 1 ms for averages and
10 ms for histogram. Vertical calibration is 20 11V for SND and EEG. D:
interspike interval histogram based on 2700 spikes. Bin width is 32 ms.

132

A '1
Unit->SND

1%

 

 

 

Pulse->SND
\a
I I
-1.0 +1.0
Interval, s

(A
Z
O

 

 

 

r I
Post - R Wave Int, ms 510

B

Unit->EEG

   
 

 

Pulse->556

 
 

 

 

 

 

 

 

2
W I I
-10 0 +1.0
lnterval,s
l849
26 Unit
‘3 ll
8
U
0 I F I
0 lnterval,s 2.0

Figure 26

133

Figure 27. Type 2 medial thalamic neuron with cardiac-related activity. Same
format and abbreviations as in Figure 23 except that "dummy" pulse-triggered
averages are not shown. A, B: peri-spike-triggered averages based on 3011 trials.
Bin width is 4 ms. Vertical calibration is 10 LIV. C: based on 1118 R wave
triggers. Bin width is 1.1 ms for averages and 11 ms for histogram. Vertical
calibration is 20 11V. D: interspike interval histogram based on 2732 spikes. Bin
width is 16 ms.

134

 

 

 

 

 

 

 

Unit->SND Unit->EEG
I fl I
'10 0 +1.0 -10 +1.0
Interval,s Interva|,s
AP
l 3978
SND 1 Unit
EEGW
133
Unit
0 0
U U
r 0 “Thu HI” '1
~90 I I fit I fl
0 Post - R Wave Int, ms 560 0 Intervals 2.0

Figure 27

135

Figure 28. Type 2 hypothalamic neuron with cardiac-related activity. Same
format and abbreviations as in Figure 23 except that "dummy" pulse-triggered
averages are not shown. A, B: peri-spike-triggered averages based on 3070 R
wave triggers. Vertical calibration is 10 1.1V. C: based on 1630 R wave triggers.
Bin width is 1.4 ms for averages and 14 ms for histogram. Vertical calibration is
20 1.1V. D: interspike interval histogram based on 3149 spikes. Bin width is 32 ms.

136

 

 

 

 

A UnIt-I>SN D Unit-(>555
r fl
“1.0 0 +1.0 -1.0 0 +1.0
1 nterva I, s Interval, s

 

 

 

 

 

SND
377- Unit
EEG
108‘ Unit :2
S

22 0

§ U

U

50 I I 0 I

0 Post - R Wave Int., ms 720 0 Intervals 20

Figure 28

137

activity was the same as that (3.610.2 Hz) appearing in the post-R wave averages
of SND (see examples in Figures 27A, C and 28A, C). The frequency of the
rhythm in the spike-triggered average of SND was taken as the reciprocal of the
interval between the two peaks closest to zero lag x 1000. Cortical activity did
not contain a cardiac-related rhythm (Figures 26C-3OC). Nevertheless, the
discharges of 6 of the 8 type 2 neurons with cardiac-related activity were
synchronized to a rhythm in cortical activity whose frequency (3.910.2 Hz) was
close to that of the cardiac-related sympathetic nerve rhythm. The spike-
triggered average (based on 3011 spikes) of cortical activity for one of these
neurons is shown in Figure 273. The peak at 240-256 ms in the ISI histogram (2732
spikes) for this medial thalamic neuron corresponded to a frequency higher than
that of the rhythms in the spike-triggered averages of SND and cortical activity.
This was not surprising since first order 151 analysis reflects the mean interval
between the last spike in a burst and the first spike in the next burst rather than
that between the first spikes in successive bursts. The discharges of two type 2
hypothalamic neurons with cardiac-related activity were not synchronized to
cortical activity (Figure 283).

The remaining two type 2 hypothalamic and four type 2 thalamic
neurons did not have cardiac-related activity even though SND contained a
prominent cardiac-related rhythm. Examples are shown in Figures 29 and 30. The
spike-triggered averages of SND for these neurons contained a rhythm (3.410.3
. Hz) that was the same as that (3.410.3 Hz) appearing in their spike-triggered
averages of cortical activity. The cardiac-related sympathetic nerve rhythm was
3.110.4 Hz in these experiments. Figure 30 shows an example in which there was
a marked difference in the frequencies of the rhythm in the spiketriggered

average (2.8 Hz) and the cardiac period (2.2 Hz). This neuron was located in a cat

138

Figure 29. Type 2 medial thalamic neuron without cardiac-related activity. Same
format and abbreviations as in Figure 23 except that "dummy“ pulse-triggered
averages are not shown. A, B: peri-spike-triggered averages based on 732 trials.
Bin width is 4 ms. Vertical calibration is 10 uV. C: based on 2360 R wave
triggers. Bin width is 1.1 ms for averages and 11 ms for histogram. Vertical
calibration is 20 LIV. D: interspike interval histograms based on 2732 spikes. Bin
width is 32 ms.

139

A B
Unit->SND W W
1
I If

W I
'10 0 +1.0 '10 0 +1.0

Interval s Interval s

 

 

 

 

 

 

27A
”:1 Unit g
E 3
8 U
U
69 I I fl 0 I I .
0 Post - R Wave Int, ms 565 O Intervals 20

Figure 29

140

Figure 30. Activity of a type 2 medial thalamic neuron synchronized to a
noncardiac-related component in sympathetic nerve discharge (SND). Cardiac
period was lengthened by administration of atenolol (0.5 mg/kg, iv). Same format
and abbreviations as in Figure 23 except that "dummy” pulse-triggered averages
are not shown. A, B: peri-spike-triggered averages based on 1100 trials. Bin
width is 4 ms. Vertical calibration is 12 11V for SND and 10 uV for EEG. C:
based on 618 R wave triggers. Bin width is 1.7 ms for averages and 17.5 ms for
histogram. Vertical calibration is 18 LIV. D: interspike interval histograms based
on 1112 spikes. Bin width is 32 ms.

 

141

 

 

 

A Unit->SND B Unit->556
1 I
Pulse->SN D Pulse->556
r ‘I h I
“1.0 0 +1.0 ‘10 0 +1.0
Interval, s Interval, s

N D

 

 

 

 

 

 

 

 

Unit
SND
I 52-1-

31.. Unit g

8
.‘2 U
g II
0
U
2 I I 0 I

0 Post - R Wave Int., ms 875 0 Intervals 20

Figure 30

142

in which the cardiac period had been slowed by administering the beta-adrenergic
antagonist atenolol (0.5 mg/kg, iv). I
3. Electrical stimulation

A square wave pulse (0.5 ms duration) was applied once every 2.5
s at 32 type 1 and nine type 2 diencephalic unit recording sites. Sixty-four trials
were averaged. In each case, stimulation increased SND. The minimum current
needed to increase SND ranged from 18-140 LIA. Using two to three times
threshold current, I measured the interval between the stimulus and the peak
increase in inferior cardiac nerve activity from the post-stimulus average of SND.
This interval was compared with that between unit spike occurrence and the first
peak to the right of zero lag. in the peri-spike-triggered average of SND. The data
points for type 1 unit recording sites in the hypothalamus and medial thalamus are
shown in Figure 31A. The diagonal line is the identity line. By definition, the
slope (i.e., regression coefficient) of the identity line is one and the y-intercept is
zero. The slope (l.1110.25) and y-intercept (-27130 ms) of the regression line
were not significantly different from those of the identity line. The mean
interval between Imit spike occurrence and the first peak to the right of zero lag
in the averages of SND for the 32 type 1 neurons was 10416 ms. The interval
between the stimulus applied to type 1 unit recording sites and the peak increase
in SND was 11813 ms. Figure 32 shows an example of close correlation of the
intervals in the spike-triggered and stimulus-triggered averages of SND. Note
that the contour of the sympathetic nerve potential synchronized to the dis-
charges of this type 1 hypothalamic neuron (panel A) was similar to that of the
sympathetic nerve response evoked by electrical stimulation at the site of unit
recording (panel B).

The slope (2.8511.08) and y-intercept (-2661127 ms) of the

regression line for the data from type 2 unit recording sites were not significantly

143

Figure 31. Comparison of changes in inferior cardiac sympathetic nerve discharge
(SND) accompanying diencephalic unit spike occurrence and an electrical stimulus
applied at site of Imit recording. The interval between unit spike occurrence and
the first peak to the right of zero lag in the peri-spike-triggered average of SND
is plotted against that between the stimulus and the peak increase in SND (as
measured from peri-stimulus average). A: data points for type 1 unit recording
sites. 3: data points for type 2 unit recording sites. hypothalamic site, thalamic
site. The diagonal line in each plot is the identity line. See text for slopes and y-
intercepts of regression lines.

ischvze
sfimflfi
ace and
of 513
SND Is
ecorilli
thalamic

‘14P4

180-

E o

d 0

Z -I

II)
‘5.

I .
.3 0
.§-_

.- 0

'25
D

O

 

O

  
    

 

 

O

Stim. -> pl: SND. ms

Type II:

130°

Unit spike -> pII SND, m:

 

O

I
180

   
  

 

 

0 Stim. -b pk SND, ms

Figure 31

180

145

Figure 32. Peri-spike-triggered and peri-stimulus averages of inferior cardiac
sympathetic nerve discharge (SND). A: spike-triggered average (519 trials) of
SND for a type 1 hypothalamic neuron. Unit spike occurrence is at zero lag. B:
peri-stimulus average (64 trials). Stimulus at zero lag (50 11A, 0.5 ms) was applied
once every 2.5 s at site of unit recording. Bin width is 1 ms. Vertical calibration
is 10 uVinAand40 uVinB.

146

 

 

 

 

 

 

 

A Unit->SND
“IO 0 +10
Interval, s

Stim.—>SND
M)
r I
-I.0 0 +1.0
Interval, s

Figure 32

147

different from those of the identity line (Figure 313). Nevertheless, 7 of the 9
data points fell far from the identity line. Thus, it was not surprising that the
mean interval (68117 ms) between type 2 unit spike occurrence and the peak of
the accompanying 2- to 6-Hz sympathetic nerve slow wave differed considerably
from that (11715 ms) between the electrical stimulus applied to the unit recording
site and the peak increase in SND.
4. Baroreceptor reflex activation

I monitored the effect of baroreceptor reflex activation on the
firing rate of 26 diencephch neurons with sympathetic nerve-related activity in
18 cats with intact carotid sinus, aortic depressor and vagus nerves. Arterial
pressure proximal to the point of aortic obstruction was raised to a level needed
to inhibit maximally SND (Figure 33). Baroreceptor reflex activation was
considered to affect unit firing rate when a change (increase or decrease) of > 20%
was observed during the reflex inhibition of SND. Six of 13 type 1 hypothalamic
neurons were affected by baroreceptor reflex activation even though they did not
have cardiac-related activity. Three of these neurons were inhibited (1.510.4 to
0.410.4 spikes/s) while the other three cells were excited (5.214.1 to 8.114.9
spikes/s). Two of four type 2 hypothalamic neurons were inhibited (2310.3 to
0.510.2 spikes/s) during baroreceptor reflex activation. Both of these neurons had
cardiac-related activity. The baroreceptor reflex-induced response of one of
these cells is shown in Figure 331. Note that the undershoot of brachial arterial
pressure following the release of aortic obstruction was accompanied by a rebound
increase in unit discharge rate. This response is consistent with a reduced level of
baroreceptor nerve activity during the fall in arterial pressure. The type 2
hypothalamic neurons unaffected by baroreceptor reflex activation did not have
cardiac-related activity. The failure of baroreceptor reflex activation to affect

the firing rate of one of these neurons is shown in Figure 3311.

148

Figure 33. Effect of baroreceptor reflex activation produced by aortic
obstruction on firing rate of two type 2 hypothalamic neurons (1, II). IA, IIA:
traces from tOp to bottom are brachial arterial pressure (AP: mmHg),
standardized pulses derived from unit action potentials, inferior cardiac
sympathetic nerve discharge (SND) and time base (1 s/division). Vertical
calibration for SND is 100 uV. I3, I13: histograms of unit firing rate based on 4
episodes of baroreceptor reflex activation for each neuron.

149

 

200 mm

AP

50 \F““::‘:. M
UNIT LIALIUL ”11 J H 1 II II I 'l

 

 

I SN 0 (AllWJI‘Ififlt

L A L A L A L A L

 

 

 

 

 

 

 

0 Time,s
250
APMWW “ 1“-
100
Unit """Hfli Hi i iiiiiiii giaiini- iiiii
l I l I I‘M/l
l SND Mill; 'IMIFWIl I [Jypfli/
‘Z'iB
>~ W
1g-
5;- , ..
0 I I I I I
O Time,s ‘ 25

150

Baroreceptor reflex activation failed to affect the firing rate of
medial thalamic neurons with sympathetic nerve-related activity. Six type 1 and
three type 2 neurons were studied. None of these cells had cardiac-related

activity.

IV. Sympathetic nerve responses elicited by cortical stimulation
The locking of type 1 diencephalic unit activity to a preceding spike-like
event in frontal-parietal cortical activity raised the possibility that this region of
the cortex provides input to sympathetic neurons in the diencephalon. To
determine whether these cortical regions had sympathetic representation I
examined the responses of the inferior cardiac nerve and diencephalic neurons
with sympathetic nerve-related activity to electrical stimulation of the frontal-
parietal cortex.
A. Sympathetic nerve responses elicited by cortical stimulation
This series of experiments was performed in 24 cats, 10 of which were
baroreceptor-denervated. In these animals, sites of stimulation in the frontal-
parietal cortex were along electrode penetrations through the gray matter located
both on the convexities of the gyri and hidden in the depths of the sulci. On
occasion stimulation also included portions of the underlying white matter. In
most cases, cortical stimulation (IO-ms trains of three l-mA pulses) elicited
increases in SND. Occasionally, increases in SND were also elicited by single
shocks of l-ms duration. The post-stimulus average of SND in Figure 34A1 is
typical of an excitatory sympathetic nerve response. In all cases, the increase
(upward negative deflection) in SND was followed by a downward positive
deflection. Such positive wave forms are indicative of a decrease in background
SND time-locked to the stimulus. In some cases, the negative and positive wave

forms could be differentiated by lowering the stimulus current (Figure 34A). In

151

Figure 34. Increases and decreases in inferior cardiac sympathetic nerve
discharge (SND) elicited by stimulation (lO-ms trains of three pulses applied once
every two s) of the motor cortex. A1-3: post-stimulus averages (64 trials) of SND
elicited using stimulus currents of 1 mA, 200 IIA and 50 11A, respectively. 31,32:
Post-stimulus averages (128 trials) of increase and decrease, respectively, in SND
elicited by supramaximal (l-mA) stimulation. Sites of stimulation in the motor
cortex in 31 and 32 were separated by 2 mm. Bin width was 1 ms for all
averages. Vertical calibration is 100 uV for A and 200 uV for 3. Horizontal
calibration is 125 ms for A and 500 ms for B.

In 2%
WWW

153

the example shown in Figure 34A, the ratio of the amplitudes of the negative to
the positive wave forms decreased as the stimulus current was lowered, until at
50 uA (A3) only the positive wave form remained. These results suggest that the
decrease in SND following an excitatory response is, at least in part, due to active
inhibition. In 13 instances (in seven cats), stimulation using supramaximal (1 mA)
currents elicited inhibitory sympathetic nerve responses (Figure 3432). In these
seven cats, stimulation of other cortical sites elicited increases in SND. Figure
3431, 32 compares an excitatory and an inhibitory sympathetic nerve response
elicited by stimulation (1 mA) of the motor cortex in the same cat. The
mechanism responsible for producing the sympathetic nerve responses was not
studied.

As shown in Figure 35, increases in SND could be elicited from widely
separated regions of the frontal~parietal and infralimbic cortex. These regions
included the primary motor (MI, sites 1 and 2), primary sensory (SI, sites 3, 4, 5
and 6), secondary sensory (811, site 7) and infralimbic (IL, sites 1-4 in panel B)
cortex. Although not shown, increases in SND were also elicited from the orbital
gyrus (O.G.) and the parietal association cortex (ASC). The ASC cortex included
the anterior marginal (M.G.) and suprasylvian (S.S.G.) gyri. Sites from which
inhibitory responses were elicited with supramaximal stimulus currents (1 mA)
were dispersed throughout the MI, SI, ASC and O.G. Inhibitory responses were not
elicited from the ST! or IL cortex. Sympathetic nerve responses could be evoked
in the left inferior cardiac nerve by stimulation of either the left or right cerebral
hemisphere (not shown).

The mean onset latencies of the sympathetic nerve responses evoked
by cortical stimulation were calculated for each of the cortical regions listed
above. These values are listed in Table 2. The mean onset latencies of responses

elicited from the different cortical regions were not significantly different from

154

Figure 35. Sympathetic nerve responses elicited by stimulation (10 ms trains of
three l-mA pulses applied once every 2 s) of prefrontal, frontal (A) and
infralimbic (B) cortex. A,B: Dorsal and sagital (approximately 0.5 mm from
midline) views, respectively, of the cat brain and post-stimulus averages (64
trials) of inferior cardiac sympathetic nerve responses. Responses are numbered
to correspond with the site of stimulation. Sites of stimulation in A were 1 to 3
mm deep in gray matter. Bin width was 1 ms in A and 3. Vertical calibration is
100 1.1V for A and 50 uV for 3. Horizontal calibration is 500 ms in A and B.
Cortical structures are labeled as defined by Crouch (1969) and include: A.S.,
ansate sulcus; CC, corpus callosum; C.G., coronal gyrus; Co.S., coronal sulcus;
Cr.S., cruciate sulcus, E.S.G., ectosylvian gyrus; F.G., fornicatus gyrus; M.G.,
marginal gyrus; 0.3., olfactory bulb; O.G., orbital gyrus; S.G. sigmoid gyrus and
S.S.G., suprasylvian gyrus.

155

 

Ink ,
his Kim

 

 

Figure 3 5

156

TABLE 2. Mean onset latencies of inferior cardiac
sympathetic nerve responses produced by cortical stimu-
lation with lO-ms trains of three pulses or l-ms single
shocks of supramaximal intensity (l-mA).

A. Cortically evoked increases in SND

 

 

No. of Sites Mean Onset 1 S.E.
ASC 5 66 1 2 ms
IL 30 72 1 1 ms
O.G. 7 66 1 7 ms
MI 52 64 1 2 ms
51 15 65 1 2 ms
51! 3 60 1 1 ms

 

B. Cortically evoked decreases in SND

 

No. of Sites Mean Onset 1 S.E.

 

combined 13 104 1 2 ms

 

Cortical regions are abbreviated and defined as follows:
ASC, association cortex area 5 (anterior marginal and
suprasylvian gyri); IL, infralimbic area (anterior fornica-
tus gyrus and subcallosal cortex); O.G., orbital gyms; MI,
primary motor area (anterior sigmoid gyrus including the
hidden cortex on the banks of the cruciate sulcus); SI,
primary sensory area (posterior sigmoid ‘and coronal gyri);
51!, secondary sensory area (anterior ectosylvian gyrus).

157

each other. In determining the onset latency for a given cortical site, measure-
ments were made for the potential elicited with l-mA pulses that had the
greatest amplitude along each electrode penetration.

In view of the suggestion by Wall and Pribram (1950) that sympathetic
responses elicited by cortical stimulation result from the activation of meningeal
afferents, I compared the sympathetic nerve responses elicited by stimulation of
the frontal-parietal cortex in cats having intact or sectioned meningeal afferents.
In three cats, two of which were baroreceptor-denervated, the left trigeminal
nerve was sectioned. Both increases (n=20) and decreases (n=4) in SND were
elicited by stimulation of the left O.G., M1, 81 and ASC cortex in these animals.
No attempt was made to elicit responses from the 82 or II. cortex in these
animals. The shape and mean onset latencies of the excitatory and inhibitory
responses evoked from the deafferentated frontal-parietal cortex were similar to
those of the corresponding responses elicited from these same regions in cats
having intact meningeal afferents. For this reason the responses elicited in
meningeally deafferentated and intact cats were pooled for statistical analysis of
onset latencies in Table 2.

B. Synaptic activation of type 1 and type 2 neurons by cortical stimula-
tion

The possibility that cortically-evoked increases in SND are mediated
via a relay in the medial diencephalon was tested by determining whether medial
diencephalic neurons with sympathetic nerverelated activity could be activated
synaptically by cortical stimulation.

Stimulation (10-ms trains of three 1-mA pulses) of the sensorimotor
cortex in seven cats synaptically activated 4 of 4 type 1 hypothalamic and 2 of 4
type 1 medial thalamic neurons. The other 2 type 1 medial thalamic neurons
tested were inhibited by cortical stimulation for 165 and 210 ms. All type 1

neurons tested were related to a spike-er event in the EEG preceding their

158

discharge. In response to cortical stimulation hypothalamic neurons discharged as
a single or a pair of spikes with variable onset latencies. Thalamic neurons
discharged as a burst of 3 to 6 spikes also with variable onset latencies. The
modal onset latencies of synaptic activation were calculated from the post-
stimulus histograms of the units (Figure 363). The modal onset latencies of
synaptic activation ranged from 22 to 26 ms for the type 1 hypothalamic neurons,
while those of the type 1 thalamic neurons were 90 and 105 ms. For those type 1
neurons activated by cortical stimulation a comparison was made to determine
whether the modal onset latency of synaptic activation, plus the time to peak in
the peri-spike-triggered average of SND for the unit was similar to the interval
between the cortical stimulus and the peak of the increase in SND in the post-
stimulus average. If the cortically evoked sympathetic nerve responses involved
the diencephch neurons, these values should be similar. For three of the type 1
hypothalamic neurons the modal onset latency, phis the time to peak in the spike-
triggered average was similar (within 10 ms) to the time to the peak of the
cortically evoked increase in SND. The example in Figure 36 is representative of
the results obtained for these three hypothalamic units. In this example the
modal onset latency (26 ms) plus the time to peak SND in the peri-spike-triggered
average (116 ms) was similar to the time to the peak of the increase in SND (144
ms) evoked by cortical stimulation. The comparison of times to peak and onset
latencies for type 1 medial thalamic neurons showed that the latency of the
cortically-induced activation of the thalamic neurons was more than 50 ms less
than that of the accompanying increase in SND.

Stimulation of the sensorimotor cortex in three cats synaptically
activated 2 of 2 type 2 medial thalamic neurons with modal onset latencies of 45
and 105 ms. Cortical stimulation inhibited the two type 2 hypothalamic neurons

tested for 172 and 72 ms. The discharge patterns of the type 2 neurons in

159

Figure 36. Synaptic activation of a type 1 hypothalamic neuron by stimulation
(IO-ms trains of three 1-mA pulses delivered once every 2 .5 s) of the motor
cortex. A: normalized peri-spike-triggered average (873 trials) of inferior
cardiac sympathetic nerve discharge (SND). Unit spike occurrence is at zero lag.
Bin width is 4 ms. Vertical calibration is 30 11V for SND. B: post-stimulus
histogram (513 trails) of synaptic activation of the unit. Modal onset latency of
synaptic activation of the unit is 27 ms. Bin width is 2 ms. C: normalized post-
stimulus average (513 trials) of inferior cardiac SND elicited by cortical
stimulation. Vertical calibration is 80 1.1V.

160

A B

‘37 Stim.»Unit

 

 

 

 

Unit->SND I g
W U
0
Stim.—>SND
2 I
I I I I
“1.0 0 +10 0 250
Interval, s Interval, ms

Figure 36

161

response to cortical stimulation were similar to those described for the type 1
neurons. All of the type two neurons were related to a rhythmic component in the
EEG. For the type 2 thalamic neurons the sum of the modal onset latency and the
time to peak in the spike-triggered average of SND was at least 40 ms less than
the interval between the cortical stimulus and the peak increase in SND.

C. Effects of frontal lobotomy on the forebrain-dependent component of
SND

Although diencephalic neurons with sympathetic nerve-related activity
can be activated synaptically by stimulation of the frontal-parietal cortex, the
question still remained as to whether these cortical regions contribute to basal
SND in the anesthetized cat. This possibility was tested by first examining the
effects of bilateral frontal lobotomy on blood pressure and inferior cardiac SND.
Subsequently, an A3 transection was performed and the changes in SND and blood
pressure compared with those produced by A3 transection in a group of control
cats.

The results obtained in seven baroreceptor-denervated cats anesthe-
tized with chloralose are summarized in Figure 37. Frontal lobotomy was
performed bilaterally 1-2 mm behind the apex of the ansate sulcus. In five cats,
the plane of transection passed immediately anterior to the genu of the corpus
callosum. In the remaining three cats the transection removed 1 to 2 mm of the
anterior corpus callosum. Frontal lobotomy failed to significantly reduce SND or
mean blood pressure (Figure 37A, 3). One hour after lobotomy, midbrain
transection at A3 reduced SND initially (i.e., 2 min reading) to 72% of control and
mean blood pressure by 44 mmHg (Figure 37C, D). These changes were
statistically significant. Mean blood pressure before A3 transection (11618
mmHg) was not significantly different than that (12317 mmHg) before lobotomy.
The level and time course of the reductions in SND and blood pressure produced

by A3 transection in the lobotomized animals were not significantly different

 

162

.mm H 0508 0.3 m030> .mnau p038o~ono~
N. 05 E 035 A33 300 _ouusou mm 3 mm was QZm no .3300283 30.538 mo 300:0 039—80 Q was 0 £053
.5593 3 Amman: mm 3 09830 .N 0.39% now pan—€080 an @0335"? «03 OZm 033.830 5MB 6030585
300 v0ua>u0n0viuoa000uohen b 5 EB 0.5303 woos .308 can 5va 0903002. 030: 0305098.? 0530.30 .8203:
no a can 0 30:03 .530035 $0.538 can a 0.8 < 0.052: hBoeono~ 338m “0.03:9 no «03E Km 03mg

163

S 23E

 

 

 

 

 

 

 

 

 

 

 

£2 .25... cm... .25...
o-n — O—N — O—— - w o-n - 0% b a“. — w
I 1
V I V
”or u. ISM“.
w
Ioaim Ion. m
m m w w M I I m
c Io
£6 .082. m £8 .052. m
....U . om . o._ . ... o... . a.“ . o._ . w
m m
loo .0 moo .0
I x x
I8 w. 18 m:
..l 3 II 3
87». i ..8_m
m m. .. m.
20:03:02.. m< Hm .4 52030.. H <

164

from those in 23 control animals (Figure 37C, D). These data suggest that the
frontal-parietal cortex did not contribute to the forebrain-dependent component

of SND in the anesthetized cat.

DISCUSSION

The purpose of this thesis was to determine whether the diencephalon is
responsible for a component of basal SND in the anesthetized cat. Several lines
of evidence from this study support this hypothesis. First, radio-frequency lesions
of the hypothalamus or the medial thalamus significantly attenuated the reduc-
tions in SND and blood pressure produced by midbrain transection.

The reductions in SND and blood pressure produced by midbrain transection
were attenuated similarly by either lateral hypothalamic, large medial hypothala-
mic or medial thalamic lesions. Whether these lesions were made at different
points in the same pathway is not known. It is also possible that the forebrain-
dependent component of SND was manifested only when two or more independent
systems were simultaneously active. It is doubtful that attenuation of the effect
of midbrain transection resulted from a nonspecific effect of the diencephalic
lesions. Regarding this point the effectiveness of a lesion was related to the
region rather than the amount of tissue destroyed. For example, small lateral
hypothalamic lesions including a major output pathway to the brain stem [MFB
(Nauta and Haymaker, 1969)] significantly attenuated the effects of midbrain
transection on SND and blood pressure whereas larger lesions in the anterior
medial hypothalamus did not.

Anterior portions of the medial hypothalamus participate in the expression
of several forms of experimental hypertension (see INTRODUCTION, IILD). For
instance, destruction of either the AV3V region (Brody g 11., 1984) or the PVH

(Zhang and Ciriello, 1985a, 1985b) prevents or reverses the hypertension

165

166

produced by sinoaortic or aortic baroreceptor denervation in unanesthetized rats.
In the present study, anterior medial hypothalamic lesions including the PVH
failed to affect significantly the decreases in SND and blood pressure produced by
midbrain transection. Medial hypothalamic lesions attenuated the effects of
midbrain transection only when large portions of aHd, Hp and Nva were
destroyed. Whereas species differences as well as the use of an anesthetic might
explain why anterior hypothalamic lesions failed to attenuate the effects of
midbrain transection in my experiments, it should also be recognized that the
basic phenomenon studied was different from that in unanesthetized rats. I
examined the consequences of diencephalic lesions on the acute reductions in SND
and blood pressure produced by midbrain transection. These acute changes were
short lived (< 1 hour). Moreover, although most pronounced in baroreceptor-
denervated cats, the effects of midbrain transection are qualitatively the same in
cats with intact baroreceptor nerves (Huang et al., 1987). Thus the present study
examined the effects of diencephalic lesions on a forebrain system that contri-
butes to SND in both baroreceptor-innervated and -denervated cats, and whose
1m is rapidly compensated for presumably by heightened activity in neural
circuits located in the brain stem or spinal cord. The mechanism (s) responsible
for the compensatory increase in neural activity remain to be determined. In
contrast, the studies (Brody g! 3., 1984; Zhang and Ciriello, 1985a; 1985b) in
unanesthetized rats focused on the effects of hypothalamic lesions on the chronic
elevation of blood pressure observed in hypertensive models.

Diencephalic lesions were more effective in attenuating the reduction in
SND produced by midbrain transection than the reduction in blood pressure. This
is indicated by the fact that the changes in SND produced by midbrain transection

in lesioned and nonlesioned cats deviated significantly from each other for a

167

longer period of time than did the changes in blood pressure. This was most
evident for the series of experiments with medial thalamic lesions.

Currently, it is unclear why diencephalic lesions were more effective in
attenuating the effects of midbrain transection on SND than on blood pressure.
Inferior cardiac and renal SND were similarly affected by midbrain transection in
nonlesioned cats (see Figures 3 and 4). Thus, the changes in inferior cardiac SND
in these preparations were representative of those occurring in at least one
sympathetic nerve controlling the vasculature. Whether such is also the case in
cats with diencephch lesions was not tested. Consequently, the possibility
remains that the diencephalic regions lesioned exerted greater control over
cardiac sympathetic nerves than vascular sympathetic nerves.

Humoral substances may have played a role in determining the degree to
which blood pressure was reduced by midbrain transection. If so, differential
attenuation of the effects of midbrain transection on SND and blood pressure may
indicate that the lesions destroyed diencephalic regions that were more involved
in controlling SND than the release of these substances. In this regard, although
some of the hypothalamic lesions destroyed the PVH, none of the lesions in this
study destroyed both the PVH and the supraoptic nuclei. Both of these nuclei are
known to be involved in the bumpral control of the circulation through the release
of vasopresssin (Blessing _e_t_ a_l., 1982; Johnson, 1985; Page, 1987).

The second line of evidence supporting the view that the medial diencepha-
lon contributes to SND in the anesthetized cat was obtained in experiments in
which multiunit activity was recorded in the hypothalamus and thalamus. SND
was coupled to ongoing diencephalic activity in both CNS-intact and decorticate
cats. In some experiments, SND was temporally related to medial thalamic and
hypothalamic multiunit activity but not to lateral thalamic activity. In other

cases, SND was coupled to lateral thalamic as well as to medial thalamic and

168

hypothalamic activity. The potential significance of this observation will be
discussed subsequently.

Radio-frequency lesions undoubtedly destroyed diencephalic neurons and
axons of passage. Moreover, multiunit diencephalic activity synchronized to SND
may have been recorded from axons of passage rather than synaptic networks.
Thus, other approaches were required to determine whether the hypothalamus
and/or medial thalamus contain the cell bodies of neurons that contribute to SND
in the anesthetized cat. Evidence that the medial thalamus contains synaptic
networks capable of influencing SND was provided by the results of chemical
stimulation studies. While other investigators (Powell and Buchannan, 1985:
Rockhold gt; a_l., 1985) reported that the injection of CNS stimulants into the
medial thalamus increases blood pressure and heart rate, the present study was
the first to demonstrate directly changes in SND upon chemical stimulation of the
medial thalamus. The intrathalamic injection of picrotoxin increased the ampli-
tude and decreased the variability of the interslow wave intervals (i.e. synchro-
nized) in inferior cardiac SND (see Figure 20). The changes in SND were
accompanied by an increase in blood pressure in 40% of the experiments. This
CNS stimulant with GABA antagonistic properties was chosen in view of the large
GABAergic innervation of the medial thalamus (Houser gt a_l., 1980; Steriade _e_t_
fl” 1985). The microinjection of equal volumes of saline into the medial thalamus
or of picrotoxin into the lateral thalamus failed to elicit sympathetic responses.
Although intrathalamic injections of fast green dye did not spread to the
hypothalamus or the ventricular system, I cannot be completely certain that the
extent of dye spread was representative of that of picrotoxin. With this
qualification, it appears that the medial thalamus contains synaptic networks
capable of generating sympathetic rhythms. This also seems to be the case for

the hypothalamus. Regarding this point, numerous investigators (Gellhorn and

169

Redgate, 1955; Redgate and Gellhorn, 1956a,b; Lee g_t_ 11., 1972; Finch and
Hicks, 1977; Williford 51 a_l., 1980; Schmidt and Dimicco, 1984; Dimicco and
Abshire, 1987; Rockhold e_t_ a_l., 1987) have reported that blood pressure and heart
rate changes are elicited by the injection of neurotoxins and excitatory amino
acids into the hypothalamus of rats and cats.

While the results of chemical stimulation indicate that synaptic networks in
the medial diencephalon can influence SND, these experiments did not answer the
question of whether these networks are the same as those responsible for the
tonic forebrain-dependent component of SND in the anesthetized cat. Thus
experiments were initiated to locate individual diencephalic neurons with natural-
ly occurring (i.e., spontaneous) discharges temporally related to SND. Barman and
Gebber (1982) identified hypothalamic neurons with sympathetic nerve-related
activity. The current study confirmed this finding and, in addition, demonstrated
for the first time that such neurons are also contained in the medial thalamus.
Two types of neurons with sympathetic nerve-related activity were located in
both the hypothalamus and medial thalamus. Type 1 unit activity was synchro-
nized to an aperiodic spike-like event in SND while type 2 neuronal activity was
synchronized to a 2- to 6-Hz rhythm in SND (see Figure 23). Neurons with these
patterns of relationship to SND are also located in the cat medulla (Barman and
Gebber, 1981; 1985; 1987; Gebber and Barman, 1981; Morrison and Gebber, 1985).

The major issue raised by these results is whether the temporal relationships
between diencephalic Imit activity and SND reflect the existence of functional
connections between these neurons and sympathetic nerves. That is, were type 1
and/or type 2 diencephalic neurons contained in networks contributing to SND in
the anesthetized cat? Two lines of evidence support this possibility. First, as
demonstrated by the lesion studies, ablation of either hypothalamic or medial

thalamic regions containing type 1 and type 2 neurons attenuated the reduction in

170

SND produced by decerebration. The distribution of neurons with sympathetic
nerve-related activity in widely separated diencephalic nuclei (see Figure 24)
presumably explains why large lesions were required to attenuate the effect of
decerebration on SND. Small diencephalic lesions significantly attenuated the
reduction in SND produced by decerebration only when they destroyed the MFB, a
major output pathway to the brain stem (Nauta and Haymaker, 1969; Conrad and
Pfaff, 1976).

The second line of evidence supporting the view that at least some of the
diencephch neurons studied contributed to SND was obtained with electrical
stimulation. SND was increased by low current stimuli applied through the
microelectrode used to record from hypothalamic and medial thalamic neurons
with sympathetic nerve-related activity. The interval between the stimulus and
the peak increase in SND often was close to that between unit spike occurrence
and the peak of the accompanying change in SND. This was most evident from
type 1 unit recording sites. These data suggest that at least some diencephalic
neurons with sympathetic nerve-related activity were involved in mediating the
increases in SND produced by electrical stimulation of the diencephalic recording
sites.

An important issue in these studies concerns the sources of the sympathetic
nerve-related activity of diencephalic neurons. Is such activity inherent to
diencephalic circuits or is it imposed on these neurons by inputs from other
sources? The second alternative likely holds for type 1 diencephalic neurons. The
discharges of these cells were also synchronized to a spike-like event in frontal-
parietal cortical activity. The onset of the spike-like event in cortical activity in
most cases preceded the discharge of the type 1 diencephalic neuron. This
observation suggests that the spike-like event in SND was not generated in the

diencephalon. Nevertheless, type 1 diencephalic neurons may have been in the

171

pathway transmitting this activity pattern to sympathetic nerves. As already
discussed, this possibility is supported by the results obtained with microstimula-
tion at type 1 diencephalic unit recording sites. There are at least two ways to
explain why spike-like cortical activity preceded type 1 diencephalic unit activity.
First, the frontal-parietal cortex may have provided an important driving input to
type 1 diencephalic neurons. This possibility seems remote, however, since
frontal lobotomy failed to attenuate the effects of midbrain transection on SND.
This observation indicates that the sympathetic nerve-related activity of these
neurons did not arise in the frontal cortex. However, these observations do not
rule out the possibility that the frontal-parietal cortex contributes to SND under
other conditions. For example, Skinner and Reed (1981) demonstrated that
cryogenic blockade of a frontocortico-hypothalamo-brain stem pathway prevents
ventricular fibrillation in the ischemic heart of emotionally stressed conscious
pigs.

The second possible explanation for the relationship of type 1 Imits to a
spike-like cortical event preceding their discharge is that the cortex and the
diencephalon share input from a common source. In this case, conduction time
from the source to the cortex would have been shorter than from the source to
the diencephalon.

Choosing between these possibilities is complicated by the fact that the
origin of the spike-like activity in the EEG of chloralose-anesthetized animals is
unknown. Under chloralose anesthesia, the EEG is characterized by a pattern of
2- to 6-Hz slow waves and spikes (Winters and Spooner, 1966). A similar EEG
pattern is observed during natural or experimentally-induced global epileptic
seizures (Gloor and Testa, 1974). At this time, neither the origin nor the
mechanism responsible for this epileptic pattern of cortical slow waves and spikes

is known. Studies have implicated the cortex (Fischer and Prince, 1977; Gloor g

172

11., 1977), the medial and intralaminar thalamic nuclei (Quesney 31 g” 1977;
Kostopoulos g; g, 1981) and the reticular formation (Velasco 3 3.1., 1975;
Faingold, 1980) as being involved in the genesis of this EEG pattern.

The origin of the sympathetic nerve-related discharges of type 2 diencepha-
lic neurons is also imknown. SND contains 2- to 6-Hz rhythmic components in
decerebrate, baroreceptor-denervated cats (Barman and Gebber, 1980). This
activity pattern is abolished by acute high spinal cord transection (McCall and
Gebber, 1975). These observations led to the proposal that brain stem circuits are
inherently capable of generating a 2- to 6-Hz component in SND (Barman mid
Gebber, 1980; Gebber, 1980). Forebrain circuits can also generate 2- to 6-Hz
rhythms since this rhythm appears in thalamic (Andersson and Manson, 1971;
Villablanca, 1974) and cortical (Barman and Gebber, 1980; Steriade 51 31., 1987)
activity in decerebrate cats and in guinea-pig thalamic slice preparations, Q zilrg
(Jahnsen and Llinas, 1984). Thus there are at least three ways to explain
synchronization of type 2 diencephalic unit activity to 2- to 6-Hz SND. First, the
diencephalic neurons may comprise a generator of 2- to 6-Hz SND subsidiary to
that located in the brain stem. Second, type 2 diencephalic neurons may receive
input from a 2- to 6-Hz generator of SND located in the brain stem. In this case,
the temporal relationship between type 2 diencephalic unit activity and SND
might not reflect a sympathetic nerve controlling function of these neurons.
Regarding this possibility there are numerous pathways from the VLM, medullary
LTF and raphe nuclei to the diencephalon. These medullary regions contain
neurons with activity locked to the 2- to 6-Hz activity pattern in SND. Neurons
in the VLM have been shown both anatomically and electrophysiologically to
project to the RPO, Ha, PVH, HL and supra0ptic nuclei in the hypothalamus
(Loewy g! 21., 1981; Kaba 31 a_l., 1986). Although a direct projection from the

medullary LTF to the hypothalamus has not been reported, the medullary LTF

173

may influence the hypothalamus via a relay in the VLM (Barman and Gebber,
1987). Alternatively, inputs from the medullary LTF may reach the hypothalamus
via a polysynaptic route through the reticular core (Scheibel and Scheibel, 1958).
A similar pathway may exist whereby LTF influences are transmitted through the
mesencephalic reticular formation and then to the medial thalamus (Scheibel and
Scheibel, 1958). Neither the VLM nor the medullary LTF project directly to the
medial thalamus; however, their influences may reach the thalamus via a relay in
the HL, 21 or aHd (Velayos, 1982; Irle, 1984; Ono and Niimi, 1985). Direct
projections from raphe magnus and obscurus to most nuclei in the medial, lateral
and dorsal hypothalamus in rats and cats have been identified (Bobillier 31 a_l.,
1976; Takagi _e_t; $1., 1930; Bowker, 1986). In addition, Bobillier _e_t_ 31. (1976)
reported direct projections from raphe magnus to CM, CL, Pf and P0, but not to
MD in the cat.

A third possibility which might account for the synchronization of type 2
diencephalic neurons to 2- to 6-Hz SND is that these neurons receive input from a
2- to 6-Hz generator located elsewhere in the forebrain such as in the medial
septum. Wilson _e_ti a1. (1976) demonstrated that the septum of the anesthetized
cat contains pacemaker neurons which discharge in rhythmic bursts at frequencies
of 2- to 6-Hz. These neurons are presumed to generate the hippocampal theta
rhythm. The septum projects to the lateral and medial hypothalamus (Szenta-
gothai, 1968; Palkovits and Zaborszky, 1979) and medial thalamus (Gullery, 1952;
Irle fl a_l., 1984) of the cat.

The relationships between the discharges of type 2 units and the inferior
cardiac nerve suggest that the 2- to 6-Hz rhythmic components in SND arise from
more than one central source. Regarding this point, the peri-spike-triggered
averages of SND for type 2 neurons contained a 2- to 6-Hz rhythm independent of

whether the unit had cardiac-related activity. Although close in frequency, one

174

can presume that the rhythm in the spike-triggered average of SND for type 2
neurons without cardiac-related activity is distinct from the cardiac-related
rhythm. Indeed, the period of the 2- to 6-Hz rhythm in the peri-spike-triggered
average of SND for a thalamic imit without cardiac-related activity was clearly
different from that of the cardiac rhythm when the heart rate was slowed by
atenolol (see Figure 30). Thus, I propose that SND contains both cardiac- and
noncardiac-related 2- to 6-Hz rhythmic components. This hypothesis is indirectly
supported by other observations. The activity of most type 2 neurons was
synchronized to a noncardiac-related 2- to 6-Hz rhythm in cortical activity. The
frequency of the cortical rhythm was the same as that of the rhythm in the spike-
triggered averages of SND for type 2 neurons without cardiac-related activity.
These data may explain why crosscorrelation analysis in some cats reveals a
common 2- to 6-Hz rhythm in SND and cortical activity (Camerer gt_ a_l., 1977;
Barman and Gebber, 1980; Gebber and Barman, 1981; Huang e_t a_l., 1987).
Descending connections from the hypothalamus to autonomic regions of the
brain stem and spinal cord have been studied extensively. The VLM receives
projections from aHd, HL and Hp (Lovick, 1985). In addition, the axons of neurons
in the PVH, HL and aHd pass through the VLM on their way to the IML of the
spinal cord in rats and cats (Kuypers and Maisky, 1975; Saper e1 a_l., 1976; Tucker
and Saper, 1985). Whether the axons of these hypothalamic neurons emit
collaterals in the VLM has not been determined. Hilton and coworkers (1983) by
using electrical stimulation traced a functional pathway from the midbrain
defense regions to the VLM. They speculated that the pathway controls
sympathetic tone even in the anesthetized animal. Hilton and Smith (1984) and
Lovick et al. (1984) showed that medullospinal VLM neurons are synaptically

activated by stimulation of the midbrain and/or hypothalamic defense regions.

175

The raphe nuclei (magnus, pallidus and obscurus) receive direct afferent projec-
tions from HL, HP and H1, H2 in the cat (Abolos and Basbaum, 1981) and from
HL, aHd, PVH, Hp in the rat (Conrad and Pfaff, 1976; Hosoya, 1985; ter Horst and
Luiten, 1986).

In contrast to the hypothalamus, relatively little attention has been paid to
descending connections emanating from medial and intralaminar thalamic nuclei.
There are at least four routes from the medial thalamus that might be involved in
the control of SND. First, a projection from the medial thalamus to the lateral
hypothalamus has been anatomically (Seigel _e_t_ 33.1., 1973) and functionally mapped
by using the 2-deoxyglucose technique (Brutus gt a_l., 1984). Such a pathway might
explain why type 1 and type 2 hypothalamic neurons fired at about the same time
as their counterparts in the medial thalamus (see Figure 25). Alternatively, these
data may indicate that hypothalamic and medial thalamic neurons with sympathe-
tic nerve-related activity received input from a common source. It is also
possible that the thalamic neurons were influenced by their hypothalamic counter-
parts. Indeed, projections from hypothalamic regions containing type 1 and type 2
neurons to the medial thalamus have been demonstrated (Velayos, 1982; Irle, 1984;
Ono and Niimi, 1985).

Two other routes over which medial thalamic neurons might influence SND
include projections to the midbrain. One route includes a monosynaptic connec-
tion from the intralaminar thalamic nuclei to the midbrain tegmentum. This
pathway was defined in rats, mice and cats by using the Golgi technique (Scheibel
and Scheibel, 1967). The second route includes a disynaptic pathway from the
intralaminar and medial thalamic nuclei to the periaqueductal gray (PAG) of the
midbrain tegmentum and the superior colliculus via the n. reticularis thalami
(Scheibel and Scheibel, 1966; Grofova g 31., 1978; Parent and Steriade, 1984).

Regarding these potential pathways for the control of SND, the microinjection of

176

excitatory amino acids into the midbrain tegmentum or PAG increases blood
pressure (Hilton, 1982; Tan 91 _a_l., 1983; McDougall gt_ $1., 1985). Thus these
midbrain regions are believed to contain the cell bodies of neurons that control
SND. Hypothalamic influences on SND may also involve a relay in the PAG since
Ha, aHd, PVH, HL, 21 are known to project to this region (Saper e_t_ 11., 1976; 1978;
ter Horst and Luiten, 1986).

A fourth route over which medial thalamic neurons might influence SND
includes a loop through the prefrontal and infralimbic cortex. CL, MD and Pc
contain neurons projecting to these cortical regions (Scheibel and Scheibel, 1967;
Jones, 1985). Moreover, changes in blood pressure and heart rate can be elicited
by electrical stimulation of these cortical regions (Hoff and Green, 1936; Green
and Hoff, 1937; Hsu e_t_ g, 1942; Kaada, 1951; Eliasson it. a_l., 1952; Hoff g 31.,
1959; 1963; Delagado, 1960; Lofving, 1961; Clarke gt_ _a_1., 1968). It should be
noted, however, that the integrity of thalamocortical projections is not mandatory
for eliciting increases in SND by electrical stimulation of the medial thalamus.
Regarding this point, sympathetic nerve responses elicited by low current stimuli
applied to sites in the medial thalamus were similar in intact and decorticate cats
(see Figure 18).

Although some hypothalamic neurons respond to natural or electrical
activation of baroreceptor afferents (Hilton and Syper, 1971; Calaresu g 11.,
1975) or have cardiac-related activity (Barman and Gebber, 1982), this is the first
study to describe the effects of baroreceptor reflex activation on diencephalic
neurons with sympathetic nerve-related activity. These results further reflect
the heterogeneity of such neurons. Perhaps the most interesting finding was that
some of the hypothalamic neurons and all of the medial thalamic neurons with
sympathetic nerve-related activity did not respond to baroreceptor reflex activa-

tion. This was the case even when the reflex inhibition of inferior cardiac SND

177

was complete (see Figure 33I3, IE3). Some of these neurons may have exerted
sympathoexcitatory actions. If so, their influences on SND would have been
intercepted at a lower level of the neuraxis.

Although it is generally accepted that cardiovascular responses can be
elicited by stimulation of the prefrontal, sensorimotor and infralimbic cortex in
an variety of species (Hoff and Green, 1936; Green and Hoff, 1937; Hsu 31 a_l.,
1942; Kaada, 1951; Eliasson _e_t _a_l., 1952; Hoff e_t a_l., 1959; 1963; Delgado, 1960;
Lofving, 1961; Clarke g 51, 1968), this is the first study to demonstrate directly
that sympathetic nerve responses can be elicited by electrical stimulation of
these cortical regions. Evidence is presented suggesting that both sympathoexci-
tatory and -inhibitory systems are represented in these cortical regions.

In view of the preceding discussion of diencephalic involvement in the
control of SND in the anesthetized cat, a major issue is whether sympathetic
nerve responses elicited by electrical stimulation of the frontal-parietal cortex
involve a synaptic relay in the diencephalon. That is, are diencephalic neurons
with sympathetic nerve-related activity part of a corticosympathetic pathway? If
the diencephalic neurons were contained within pathways mediating cortical
influences to the sympathetic nerves, one would expect that these neurons could
be synaptically activated by stimulation of those cortical areas from which
sympathetic nerve responses were elicited. Furthermore, the sum of the modal
onset latency of synaptic activation of the diencephalic neuron and the interval
between its spontaneous discharge and the peak increase in SND (as measured
from its peri-spike-triggered average) should be close to the interval between the
cortical stimulus and the peak of the increase in SND in the post—stimulus
average. This was the case for 3 of 4 type 1 hypothalamic neurons activated by
stimulation of the cortex, suggesting that these neurons were part of a cortico-

sympathetic pathway. However, such was not the case for the type 1 nor the type

178

2 medial thalamic neurons activated by cortical stimulation, since the intervals
were not similar. Thus, the results are consistent with the possibility that
hypothalamic but not thalamic neurons with sympathetic nerve-related activity
are contained in a pathway responsible for the sympathetic nerve responses
elicited by stimulation of the frontal-parietal cortex. The functional significance
of those type 2 hypothalamic and type 1 medial thalamic neurons inhibited by
stimulation of the cortex remains to be determined.

While the results of my study suggest that cortically-evoked sympathetic
nerve responses are, at least in part, mediated through the hypothalamus,
anatomical studies have failed to demonstrate a direct connection from the
frontal cortex to the hypothalamus in the cat (Szentagothai, 1968; Palkovitz and
Zaborszky, 1969). However, a direct frontocortico-hypothalamic connection has
recently been demonstrated in the rat (Beakstead, 1979; Van der Kooy 2t. a_l.,
1984). In the cat, influences from the frontal cortex may reach the hypothalamus
via a multisynaptic pathway through the septum or amygdala (cf. Brodal, 1981;
Kuypers, 1981).

The possibility that the sympathetic nerve responses elicited by cortical
stimulation involved extra-diencephalic as well as diencephalic pathways must
also be considered. Indeed this is believed to be the case in the monkey (Wall and
Davis, 1951). There are at least two extra-diencephalic pathways by which
cortical sympathetic responses may reach sympathetic nerves. First, these
responses may have involved the activation of a direct cortical projection to the
medullary tegmental field or the PAG (Brodal, 1981; Kuypers, 1981). Second,
some of the responses may have been mediated via a monosynaptic pathway from
the medial frontal cortex to the IML. Such a pathway was recently identified in

the rat by Hurley-Guis g; _a_1., (1986).

179

The possibility that the cortically-evoked sympathetic responses did not
involve the activation of cortical neurons must also be considered. These
responses may have resulted from the antidromic activation of thalamic or
hypothalamic neurons connected via collaterals to caudally projecting sympathe-
tic neurons. In this case, the synaptic activation of diencephalic neurons with
sympathetic nerve-related activity may have been due to collateral input to these
neurons from the antidromically activated thalamocortical or hypothalamocorti-
cal neurons. Both the thalamus (Jones, 1985) and the hypothalamus (Conrad and
Pfaff, 1976) contain neurons with cortically projecting axons.

There are no studies implicating the relay and association nuclei of the
lateral thalamus in the control of SND and blood pressure. Thus, the finding that
multiimit lateral thalamic activity was temporally related to SND in some
experiments was unexpected. There are at least three possible explanations to
account for the synchronization of SND to lateral thalamic activity. First, the
sympathetic nerve-related activity in the lateral thalamus may have resulted
from the hypersynchronization of activity in different thalamic nuclei rather than
simply a role of the lateral thalamus in the control of SND. Regarding this
possibility, the predominant frequency of slow wave activity at medial and lateral
thalamic sites was essentially the same in these experiments. Such occurrences
may reflect the level of cateleptic anesthesia produced by chloralose, since this
agent is known to be capable of hypersynchronizing activity in different thalamic
nuclei and cortical regions (Winters and Spooner, 1966). Direct pathways from the
medial to the lateral thalamus which might account for the synchronization of
activity in these two regions have been demonstrated (Purpura e1 a_l., 1966;
Purpura, 1970).

A second explanation of sympathetic nerve-related activity in the lateral

thalamus is that this activity was recorded from the axons of medial thalamic

180

neurons passing through the lateral thalamus. Evidence against this possibility
comes from the finding that in some experiments activity recorded from medial,
but not lateral thalamic nuclei was related to SND. hi these experiments the
lateral thalamus was extensively searched, suggesting that the lack of sympathe-
tic nerve-related activity in the lateral thalamus was not due simply to the failure
to locate the axons of medial thalamic neurons with activity related to SND.

A third possibility is that lateral thalamic neurons are involved in control-
ling SND. This possibility is supported by the fact that electrical stimulation of
the lateral thalamus increased SND. Whether these responses involved the
activation of lateral thalamic neurons or fibers of passage is not known. The fact
that picrotoxin injected into the lateral thalamus failed to affect SND does not
distinguish between these alternatives. Lateral thalamic "sympathetic” neurons
may be picrotoxin-insensitive. If such neurons exist, there are at least two routes
by which they may influence SND. First, their influences may be relayed through
the cortex which is a major target of lateral thalamic neurons (Scheibel and
Scheibel, 1967; Jones, 1985). Second, they may control SND via the n. reticularis
thalami or medial and intralaminar thalamic nuclei. These nuclei project to the
midbrain and hypothalamus (Scheibel and Scheibel, 1966; 1967 ; Seigel _e_t_ _a_1_., 1973;
Grofova 21 a_l., 1978; Brutus g_t_ 11., 1984; Parent and Steriade, 1984) and receive
input from the lateral thalamic nuclei (Jones, 1985). None of these potential roles

of the lateral thalamic nuclei in the control of SND were tested in this thesis.

SUMMARY

My thesis involved studies on the role of the diencephalon in the control of
sympathetic nerve discharge (SND) in the anesthetized cat. The thesis is
comprised of four main projects. The findings and conclusions of each of these

projects are summarized below.

A. Localization of diencephalic regions contributing to SND in the barorecep-
tor-denervated cat anesthetized with chloralose.

The results of this project are:

1. Midbrain transection at stereotaxic plane A3 produced significant
decreases in blood pressure (3816 mmHg) and SND in the inferior
cardiac (3414%) and renal (3 513%) sympathetic nerves.

2. Radio-frequency lesions of the lateral hypothalamus (including the
MFB), posterior medial hypothalamus or medial thalamus significantly
attenuated the decreases in SND and blood pressure produced by
midbrain transection.

3. Lesions of the anterior medial hypothalamus (including the PVH) failed
to attenuate the effects of midbrain transection.

The conclusions of this project are:

1. As previously suggested by Huang 91 2.1. (1987), the forebrain makes a

significant contribution to SND in this preparation.

181

182

2. The medial hypothalamus and medial thalamus are, at least in part,
responsible for the forebrain-dependent component of basal SND in

this preparation.

Synchronization of SND to diencephch multiunit activity and sympathetic
nerve responses to electrical and chemical stimulation of the diencephalon.
The results of this project are:

1. Multimit activity recorded from the thalamus and hypothalamus was
temporally related to a 2- to 6-Hz component of SND in the majority
of baroreceptor-denervated and —intact cats anesthetized with chlora-
lose.

2. In some CNS-intact and decorticate cats, activity recorded from the
medial thalamus and hypothalamus, but not from the lateral thalamus
was related to SND. In other cats, lateral as well as medial thalamic
and hypothalamic activity was synchronized to SND.

3. Electrical stimulation of the medial and lateral thalamus and hypo-
thalamus elicited increases in inferior cardiac SND in decorticate and
CNS-intact cats.

4. The injection of the GABA antagonist picrotoxin into the medial, but
not the lateral thalamus increased the amplitude of the 2- to 6-Hz
sympathetic slow wave activity and decreased the variability of the
interslow wave intervals.

The conclusions of this project are:

1. A component of basal SND in the 2- to 6-Hz range can be coupled to '
thalamic and hypothalamic activity in both CNS-intact and decorti-

cate cats anesthetized with chloralose.

2.

183

A functional connection exists between the diencephalon and the
sympathetic nerves.
The medial thalamus contains the cell bodies of neurons capable of

influencing SND.

Identification of diencephalic neurons with sympathetic nerve-related acti-

vity.

The results of this project are:

1.

3.

4.

Peri-spike-triggered averaging showed that some diencephalic neurons
(type 1 neurons) were related to a spike-like event in SND following
their discharge but were not related to SND preceding their discharge.
All type 1 neurons were also related to a cortical event which most
often preceded the discharge of the neuron.

Peri-spike-triggered averaging showed that some diencephalic neurons
(type 2 neurons) were related to a 2- to 6-Hz rhythm in SND. All type
2 medial thalamic and most type 2 hypothalamic neurons were also
related to a 2- to 6-Hz rhythm in frontal-parietal cortical activity.
Stimulation through the recording microelectrode at type 1 and type 2
recording sites elicited increases in inferior cardiac SND. In some
instances, the interval between the stimulus and the peak increase in
SND was close to that between unit spike occurrence and the peak of
the accompanying increase in SND.

Post-R wave averaging in combination with spike-triggered averaging
revealed that the discharges of some type 2 diencephalic neurons were
synchronized to a cardiac-related component in SND while those of
other type 2 neurons were locked to a noncardiac-related 2- to 6-Hz

component.

5.

184

Baroreceptor reflex activation altered (increased or decreased) the
firing rate of some type 1 and type 2 hypothalamic neurons. Reflex
activation did not affect the firing of medial thalamic neurons with

sympathetic nerve-related activity.

The conclusions of this project are:

1.

z.

3.

4.

The medial thalamus and hypothalamus in the chloralose-anesthetized
cat contains two types of neurons with sympathetic-nerve related
activity.

At least some diencephch neurons with sympathetic nerve-related
activity are contained in pathways mediating the increases in SND
produced by electrical stimulation of the diencephalon.

Some hypothalamic but none of the medial thalamic neurons with
sympathetic nerve-related activity receive inputs from baroreceptor
afferents.

There is more than one central source of basal 2- to 6-Hz SND in the

chloralose-anesthetized cat.

Sympathetic nerve responses elicited by cortical stimulation.

The results of this project are:

1.

z.

3.

Both increases and decreases in inferior cardiac SND can be elicited
by stimulation of the frontal-parietal and infralimbic cortex.
Stimulation of the frontal (sensorimotor) cortex synaptically activated
medial thalamic and hypothalamic type 1 and medial thalamic type 2
neurons.

For some of the hypothalamic type 1 neurons, the sum of the modal
onset latency of synaptic activation plus the time to peak in the spike-

triggered average of SND was similar to the interval between the

4.

185

cortical stimulus and the peak of the evoked increase in inferior
cardiac SND.

Bilateral frontal lobotomy failed to attenuate the decreases in SND
and blood pressure produced by midbrain transection at stereotaxic
plane A3 in baroreceptor-denervated cats anesthetized with chlora-

lose.

The conclusions of this project are:

l.

2.

3.

The prefrontal, frontal and infralimbic cortex have sympathetic repre-
sentation. Both sympathoexcitatory and -inhibitory systems are
represented.

Some type 1 hypothalamic neurons may be contained in pathways
mediating sympathetic responses to cortical stimulation.

The prefrontal and frontal cortex is not involved in providing the
forebrain-dependent component of SND under the conditions of my

experiments.

Bibliography

BIBLIOGRAPHY

Abolis, LA. and Basbaum, A.I.: Afferent connections of the rostral medulla of the
cat: A neural substrate for midbrain-medullary interactions in the modula-
tion of pain. J. Comp. Neurol. 201: 285-297, 1981.

Abrahams, V.C., Hilton, S.M. and Zbrozyna, A.: Active muscle vasodilatation
produced by stimulation of the brain stem: its significance in the defense
reaction. J. Physiol. (London) 154: 491-513, 1960.

Adair, J.R., Hamilton, B.L., Scappaticci, K.A., Helke, C.J. and Gillis, R.A.:
Cardiovascular responses to electrical stimulation of the medullary raphe
area of the cat. Brain Res. 128: 141-145, 1977.

Adrian, E.D., Bronk, D.W. and Phillips, G.: Discharges in mammalian sympathetic
nerves. J. Physiol. (London) 74: 115-133, 1932.

Alexander, N. and Morris, M.: Elevated plasma vasopressin in the sinoaortic
denervated (SAD) rat. Fed. Proc. 41: 1094, 1982.

Alexander, R.S.: Tonic and reflex fimctions of medullary sympathetic cardio-
vascular centers. J. Neurophysiol. 9: 205-217, 1946.

Anand, B.K. and Dua, S.: Circulatory and respiratory changes induced by
electrical stimulation of limbic system (visceral brain). J. Neur0physiol.
19: 391-400, 1956.

Andersen, P. and Eccles, J .C.: Inhibitory phasing of neuronal discharge. Nature
196: 645-647, 1962.

Andersen, P., Eccles, J.C. and Sears, T.A.: The ventrobasal complex of the
thalamus: Types of cells, their responses and their functional organization.
J. Physiol. (London) 174: 370-399, 1964.

Andersen, P. and Andersson, S.A.: E Physiological basis of the alpha rhythm, _e_d.
by A. Towe, Appleton-Century-Crafts, New York, 1968.

Andersson, S.A. and Manson, J.R.: Rhythmic activity in the thalamus of the
decorticate cat. Electroenceph. clin. Neurophysiol. 31: 2134, 1971.

Angyan, L.: Cardiovascular effects of septal, thalamic, hypothalamic and
midbrain self-stimulation. Physiol. Behav. 20: 217-226, 1978.

Bachoo, M. and Polosa, C.: Properties of the inspiration-related activity of

sympathetic preganglionic neurons of the cervical trunk in the cat. J.
Physiol. 385: 545-564, 1987.

186

187

Barman, S.M. and Gebber, G.L.: Basis for synchronization of sympathetic and
phrenic nerve discharges. Am. J. Physiol. 231(5): 1601-1607, 1976.

Barman, S.M. and Gebber, G.L.: Tonic sympathoinhibition in the baroreceptor-
denervated cat. Proc. Soc. Exp. Biol. Med. 157: 648655, 1978.

Barman, S.M. and Gebber, G.L.: Sympathetic nerve rhythm of brain stem origin.
Am. J. Physiol. 239: R42-R47, 1980.

Barman, S.M. and Gebber, G.L.: Brain stem neuronal types with activity patterns
related to sympathetic nerve discharge. Am. J. Physiol. 240 (Regulatory
Integrative Comp. Physiol. 9): R335-R347, 1981.

Barman, S.M. and Gebber, G.L.: Hypothalamic neurons with activity patterns
related to sympathetic nerve discharge. Am. J. Physiol. 242 (Regulatory
Integrative Comp. Physiol. 11): R34-R43, 1982.

Barman, S.M. and Gebber, G.L.: Sequence of activation of ventrolateral and
dorsal medullary sympathetic neurons. Am. J. Physiol. 245 (Regulatory
Integrative Comp. Physiol. 14): R438-R447, 1983.

Barman, S.M. and Gebber, G.L.: Axonal projection patterns of ventrolateral
medullospinal sympathoexcitatory neurons. J. Neurophysiol. 53(6): 1551-
1566, 1985.

Barman, S.M. and Gebber, G.L.: Lateral tegmental field neurons of cat medulla: a
source of basal activity of ventrolateral medullospinal sympathoexcitatory
neurons. J. Neur0physiol. 57(5): 1410-1424, 1987.

Barman, S.M., Gebber, G.L. and Calaresu, F.R.: Differential control of sympathe-
tic nerve discharge by the brain stem. Am. J. Physiol. 247 (Regulatory
Integrative Comp. Physiol. 16): R513-R519, 1984.

Beckstead, R.M.: An autoradiographic examination of corticocortical and sub-
cortical projections of the mediodorsal-projection (prefrontal) cortex in rat.
J. Comp. Neurol. 184: 43-62, 1979.

Bendat, J.S. and Piersol, A.G.: Random Data: Analysis and Measurement
Procedures, Wiley, New York, 1971.

Berecek, K.H., Barron, K.W., Webb, R.L. and Brody, M.J.: Vasopressin-central
nervous system interactions in the development of DOCA hypertension.
Hypertension 4(suppl. II): II-131-II-137, 1982.

Bernard, C: Lecons sur la Physiologie et la Pathologie du systeme Nerveux, vol.
1, Bailliere, Paris, 1863.

Bing, R.J., Thomas, C.3. and Waples, E.C.: The circulation in experimental
neurogenic hypertension. J. Clin. Invest. 24: 513522, 1945.

Blessing, W.W., Sved, A.P. and Reis, D.J.: Destruction of noradrenergic neurons
in rabbit brain stem elevates plasma vasopressin, causing hypertension.
Science 217: 661-663, 1982.

188

Bobillier, P., Seguin, S., Petitjean, F., Salvert, D., Touret, M. and Jouvet, M.: The
raphe nuclei of the cat brain stem: A topographical atlas of their efferent
projections as revealed by autoradiography. Brain Res. 113: 449-486, 1976.

Bowker, R.M.: The relationship between descending serotonin projections and
ascending projections in the nucleus raphe magnus: a double labeling study.
Neurosci. Lett. 70: 348-353, 1986.

Brodal, A.: Neurological Anatomy in Relation to Clinical Medicine, Oxford Univ.
Press, New York, 1981.

Brody, M.J., Faber, J.E., Mangiapane, M.L. and Porter, J.P.: The central nervous
system and prevention of hypertension. E Handbook of Hypertension 4
Experimental and Genetic Models of Hypertension, 93. by W. de Jong, pp.
474-494, Elsevier, New York, 1984.

Bronk, D.W., Ferguson, L.K., Margaria, R. and Solandt, D.Y.: The activity of the
cardiac sympathetic centers. Am.J. Physiol. 117: 237-249, 1936.

Brutus, MmWatson, Jr., R. E., Shakh, M.B., Siegel, H.E., Weiner, S. and Siegel, A.:
A [ C] 2-deoxyg1ucose analysis of the functional neuronal pathways of the
limbic forebrain in the rat. IV. A pathway from the prefrontal cortical-
medial thalamic system to the hypothalamus. Brain Res. 310: 279-293,
1984.

Buchanan, S.C. and Powell, D.A.: Cardiovascular adjustments elicited by elec-
trical stimulation of dorsomedial and anteriomedial nuclei of the thalamus.
Physiol. Psych. ('m press), 1987.

Buggy, J., Fink, G.D., Johnson, A.K. and Brody, M.J.: Prevention of the
development of renal hypertension by anteroventral third ventricular tissue
lesion. Circ. Res. 40(suppl. 1): 110-117, 1977.

Buggy, J., Fink, G.D., Haywood, J.R., Johnson, A.K. and Brody, M.J.: Interruption
of the maintenance phase of established hypertension by abation of the
anteroventral third ventricle (AV3V) in rats. Clin. Exp. Hyperten. 1(3): 337-
353, 1978.

Bumpus, F.M., Sen, 8., Smeby, R.R, Sweet, C., Ferrario, C.M. and Khosla, M.C.:
Use of angiotensin II antagonists in exper hypertension. Circ. Res. 32-
33(suppl. 1): 150-158, 1973.

Bunag, R.D. and Eferakeya, A.E.: Immediate hypotensive after-effects of
posterior hypothalamic lesions in awake rats with spontaneous, renal or
DOCA hypertension. Cardiovasc. Res. 10: 663-671, 1976.

Calaresu, F.R., Faiers, A.A. and Mogenson, G.J.: Central neural regulation of
heart and blood vessels in mammals. Prog. Neurobiol. 5: 1-35, 1975.

Camerer, H., Stroh-Werz, M., Kreinke, B. and Langhorst, P.: Postganglionic
sympathetic activity with correlation to heart rhythm and central cortical
rhythms. Pfluegers Arch. 370: 221-225, 1977.

189

Clarke, N.P., Smith, O.A. and Shearn, W.D.: Topographical representation of
vascular smooth muscle of limbs in primate motor cortex. Am. J. Physiol.
214(1): 122-129, 196 8.

Cohen, MJ. and Gootman, P.M.: Spontaneous and evoked oscillations in respira-
tory and sympathetic discharge. Brain Res. 16: 265-268, 1969.

Cohen, M.I. and Gootman, P.M.: Periodicities in efferent discharge of splanchnic
nerve of the cat. Am. J. Physiol. 218: 1092-1101, 1970.

Conrad, L.C.A., and Pfaff, D.W.: Efferents from medial basal forebrain and
hypothalamus in the rat. II. An autoradiographic study of the anterior
hypothalamus. J. Comp. Neurol. 169: 221-262, 1976.

Coombs, J.S., Curtis, D.R. and Eccles.: The interpretation of spike potentials of
motoneurons. J. Physiol. (London) 139: 198-231, 1957.

Coote, J .H., Hilton, S.M. and Zbrozyna, A.W.: The ponto-medullary area
integrating the defense reaction in the cat and its influence on muscle blood
flow. J. Physiol. (London) 229: 257-274, 1973.

Coote, J.H.: Repiratory and circulatory control during sleep. J. Exp. Biol. 100:
223-244, 19820

Crouch, J .E.: Text-Atlas of Cat Anatomy. Lea and Ferbiger, Philadelphia, 1969.

Dampney, R.A.L. and Moon, E.A.: Role of the ventrolateral medulla in vasomotor
response to cerebral ischemia. Am. J. Physiol. 239 (Heart Circ. Physiol. 8):
H349-H353, 1980.

Delgado, J.M.: Circulatory effects of cortical stimulation. Physiol. Rev.
40(suppl. 4): 146-171, 1960.

Dempsey, E.W. and Morison, R.S.: The production of rhythmically recurrent
cortical potentials after localized thalamic stimulation. Am. J. Physiol.
135: 293-300, 1942a.

Dempsey, E.W. and Morison, R.S.: The interaction of certain spontaneous and
induced cortical potentials. Am. J. Physiol. 135: 301-308, 1942b.

Deschenes, M., Madariaga-Domich, A. and Steriade, M.: Dendrodendritic synapses
in the cat reticularis thalami nucleus: A structural basis for thalamic
spindle synchronization. Brain Res. 334: 165-168, 1985.

Dimicco, J.A. and Abshire, V.M.: Evidence for GABAergic inhibition of a
hypothalamic sympathoexcitatory mechanism in anesthetized rats. Brain
Res. 402: 1-10, 1987.

Dittmar, C.: Uber die Laga des songenannten Gefasscentrums der medulla
oblongata. Ber. Vehr. Saechs. Wiss. Leipzig Math. Phys. K1. 25: 449-479,
1873.

Doba, N. and Reis, D.J.: Acute fulminating neurogenic hypertension produced by
brain stem lesions in the rat. Circ. Res. 32: 584-593, 1973.

190

Dusser de Barenne, J .G. and McColloch, W.S.: The direct functional interrelation
of sensory cortex and optic thalamus. J. Neurophysiol. 1: 176-186, 1938a.

Dusser de Barenne, J.G. and McColloch, W.S.: Sensorimotor cortex, nucleus
caudatus and thalamus opticus. J. Neurophysiol. 1: 364377, 1938b.

Eliasson, S., Lindgren, P. and Uvnas, 3.: Representation in the hypothalamus and
the motor cortex in the dog of the sympathetic vasodilator outflow to the
skeletal muscles. Acta. Physiol. Scand. 27: 18-37, 1952.

Faingold, C.L.: Enhancement of mesencephalic reticular neruonal responses to
sensory stimuli with pentylenetetrazol. Neuropharmacol. 19: 53-62, 1980.

Finch, L. and Hicks, P.E.: Involvement of hypothalamic histamine-receptors in
the central cardiovascular actions of histamine. Neuropharmacol. 16: 211-
218, 1977.

Fink, G.D., Haywood, J.A., Bryan, W.J., Packwood, W. and Brody, M.J.: Central
site for pressor action of blood-borne angiotensin in rat. Am. J. Physiol.
239 (Regulatory Integrative Comp. 8): R358-R361, 1980.

Fischer, R.S. and Prince, D.A.: Spike-wave rhythms in cat cortex induced by
parenteral penicillin. I. Electroencephalographic features. Electroenceph.

Folkow, 3., Johansson, 3. and Oberg, 3.: A hypothalamic structure with a marked
inhibitory effect on tonic sympathetic activity. Acta Physiol. Scand. 462-
270, 1959.

Galeno, T.M., Van Hoesen, G.W., Maixner, W., Johnson, A.K. and Brody, M.J.:
Contribution of the amygdala to the development of spontaneous hyperten-
sion. Brain Res. 246: 1-6, 1982.

Galeno, T.M., Van Hoesen, G.W. and Brody, M.J.: Central amygdaloid nucleus
lesion attenuates exaggerated hemodynamic responses to noise stress in
spontaneously hypertensive rat. Brain Res. 291: 249-259, 1984.

Gebber, G.L.: Basis for phase relations between baroreceptor and sympathetic
nervous discharge. Am. J. Physiol. 230(2): 263-270, 1976.

Gebber, G.L. and Barman, S.M.: Basis for 2-6 cycle/s rhythm in sympathetic
nerve discharge. Am. J. Physiol. (Regulatory Integrative Comp. Physiol. 8):
R48-R56’ 19800

Gebber, G.L. and Barman, S.M.: Sympathetic-related activity of brain stem
neurons in baroreceptor-denervated cats. Am. J. Physiol. 240 (Regulatory
Integrative Comp. Physiol. 9): R348-R3 55, 1981.

Gebber, G. L.: Brain stem mechanisms and substrates involved in generation of
sympathetic nerve discarge. In Exciation and Neural Control of the Heart,
ed. by M. N. Levy and M. Vassalle, pp. 203-226, Am. Physiol. 800., Williams
and Wilkins, Baltimore, 1982.

191

Gebber, G.L.: Brain stem systems involved in cardiovascular regulation. _Ir_1

Nervous Control of Cardiovascular Function, ed. by W.C. Randall, pp. 374-
350, Oxford Univ. Press, New York, 1984.

Gebber, G.L. and Barman, S.M.: Lateral tegmental field neurons of cat medulla:
A potential source of basal sympathetic nerve discharge. J. Neurophysiol.
54(6): 1498-1512, 1985.

Gellhorn, E. and Redgate, E.S.: Hypotensive drugs [Acetylcholine, mecholyl,
histamine] as indicators of the hypothalamic excitability of the intact
organism. Int. Pharmacodyn. 102: 162-178, 1955.

Gellhorn, E.: The significance of the state of the central autonomic nervous
system for quantitative and qualitative aspects of some cardiovascular
reactions. Am. Heart J. 67(1): 106-120, 1964.

Gloor, P. and Testa, G.: Generalized penicillin epilepsy in the cat: effects of
intracarotid and intravertebral pentylenetetrazol and amobarbital injec-
tions. Electroenceph. clin. Neurophysiol. 36: 499-515, 1974.

Gloor, P., Quesney, L.F. and Zumstein, H.: Pathophysiology of generalized
penicillin epilepsy in the cat: the role of cortical and subcortical structures.
II. topical application of penicillin to the cerebral cortex and to subcortical
structures. Electroenceph. clin. Neur0physiol. 43: 19-94, 1977.

Goodchild, A.K. and Dampney, R.A.L.: A vasopressor cell group in the rostral
dorsomedial medulla of the rabbit. Brain Res. 360: 24-32, 1985.

Gordon, F.J., Haywood, J.R., Brody, M.J. and Johnson, A.K.: Effect of lesions of
the anteroventral third ventricle (AV3V) on the development of hypertension
in spontaneously hypertensive rats. Hypertension 4: 387-393, 1982.

Green, J.H. and Heffron, P.F.: Observations on the origin and genesis of a rapid
sympathetic rhythm. Arch. Int. Pharmacodyn. 169: 403-411, 1967.

Green, H.D. and Hoff, E.C.: Effects of fardic stimulation of the cerebral cortex
on limb and renal volumes in the cat and monkey. Am. J. Physiol. 118: 641-
658, 1937.

Grofova, I., Ottersen, O.P. and Rinvik, E.: Mesencephalic and diencephalic
afferents to the superior colliculus and periaqueductal gray substance
demonstrated by retrograde axonal transport of horseradish peroxidase in
the cat. Brain Res. 146: 205-220, 1978.

Guertzenstein, P.G., Hilton, S.M., Marshall, J.M. and Timms, R.J.: Experiments
on the origin of vasomotor tone. J. Physiol. (London) 275: 78P-79P, 1978.

Gullery, R.W.: Afferent fibers to the dorso-medial thalamic nucleus in the cat.
J. Anat. 93: 403-419, 1959.

Hallback, M. and Folkow, 3.: Cardiovascular responses to acute mental "stress" in
spontaneously hypertensive rats. Acta. Physiol. Scand. 90: 684-698, 1974.

‘ ——_:‘_‘T§

 

192

Hartle, D.R. and Brody, M.J.: Hypothalamic vasomotor pathways mediating the
development of hypertension in the rat. Hypertension 4(suppl. III): IlI-68-III-
71, 1982.

Hartle, D.R., Lind, R.W., Johnson, A.K. and Brody, M.J.: Localization of the
anterior hypothalamic angiotensin II pressor system. Hypertension 4(suppl.-
165, 1982.

Haywood, J.R., Fink, G.D., Buggy, J., Boutelle, S., Johnson, A.K. and Brody, M.J.:
Prevention of two-kidney, one clip renal hypertension in rat by ablation of
AV3V tissue. Am. J. Physiol. 245 (Heart Circ. Physiol. 14): H683-H689,
1983.

Hilton, S.M. and Spyer, R.M.: Participation of the anterior hypothalamus in the
baroreceptor reflex. J. Physiol (London) 218: 271-293, 1971.

Hilton, S.M., Spyer, K.M. and Timms, R.J.: The origin of the hind limb vasodila-
tion evoked by stimulation of the tex in the cat. J. Physiol. (London) 287:
545-557, 1979.

Hilton, S.M.: The defense-arousal system and its relevance for circulatory and
respiratory control. J. Exp. Biol. 100: 159-174, 1982.

Hilton, S.M., Marshall, J.M. and Timms, R.J.: Ventral medullary relay neurons in
the pathway from the defense areas of the cat and their effect on blood
pressure. J. Physiol. (London), 345: 149-16.

Hilton, S.M. and Smith, P.R.: Ventral medullary neurons excited from the
hypothalamic and mid-brain defense areas. J. Autonom. Nerv. Sys. 11: 35-
42’ 1984.

Hilton, S.M. and Redfern, W.S.: A search for brain stem cell groups integrating
the defense reaction in the rat. J. Physiol. (London) 378: 213-228, 1986.

Hoff, E.C. and Green, H.D.: Cardiovascular reactions induced by electrical
stimulation of the cerebral cortex. Am. J. Physiol. 117: 411-422, 1936.

Hoff, E.C., Kell, Jr., J.F., Bourne, J.M. and Velo, A.: Pressor responses to
cortical stimulation in unanesthetized cats and dogs. Fed. Proc. 18: 69-76,
1959.

Hoff, E.C., Kell, J.F. and Carroll, M.N.: Effects of cortical stimulation and
lesions on cardiovascular function. Physiol. Rev. 43: 68-114, 1963.

Hosya, Y.: Hypothalamic projections to the ventral med oblongata in the rat,
with special reference to the nucleus raphe pallidus: A study using
autoradiographic and HRP. Brain Res. 344: 338-3 50, 1985.

Houser, C.R., Vaughn, J.E., Barber, R.P. and Roberts, E.: GABA neurons are the
major cell type of the nucleus reticularis thalams. Brain Res. 200: 341-354,
1980.

193

Hsu, S.-H., Hwang, K. and Chu, H.-N.: A study of the cardiovascular changes
induced by stimulation of the motor cortex in dogs. Am. J. Physiol. 137:
468-472, 1942.

Huang, Z.S., Gebber, G.L., Barman, S.M. and Varner, R.J.: Forebrain contribution
to sympathetic nerve discharge in anesthetized cats. Am J. Physiol. 252
(Regulatory Integrative Comp. Physiol. 21): R645-R652, 1987.

Hurley-Gius, K.M., Cechetto, D.F. and Saper, C.3.: Spinal connections of the
infralimbic autonomic cortex. Soc. Neurosci. Abstr. Part 1, 145.4: 538,
1986.

Irle, E., Markowitsch, H.J. and Streicher, M.: Cortical and subcortical, including
sensory-related, afferents to the thalamic mediodorsal nucleus of the cat.

Jahnsen, H. and Llinas, R.: Ionic basis for the electroresponsiveness and
oscillatory properties of guinea-pig thalamic neurons in vitro. J. Physiol.
(London) 349: 227-247, 1984.

Jasper, H.: Diffuse projection systems: the integrative action of the thalamic
reticular system. Electroenceph. clin. Neurophysiol. 1: 405-420, 1949.

Jasper, H.H. and Ajmone-Marsan, C.: A Stereotaxic Atlas of the Diencephalon of
the Cat. Natl. Res. Council Canada, Ottawa, 1954.

Johnson, A.K.: Role of periventricular tissue surrounding the anteroventral thaird
ventricle (AV3V). la The Regulation of Body Fluid Homeostasis, 3:1. by R.W.
Scbrier, pp. 319-331, Raven Press, New York, 1985.

Johnston, G. A. R.: Physiologic pharmacology of GABA and its antagonists in the
vertebrate nervous system. In GABA in Nervous System Function, e_c1. by E.
Roberts, T. N. Chase and D. 3. Tower, pp. 395411, Raven Press, New York,
1976.

Jones, 3.6.: The Thalamus. Plenum, New York, 1985.

Judy, W.V., Watanabe, A.M., Henry, D.P., Beach, H.R., Murphy, W.R. and Hochel,
G.M.: Sympathetic nerve activity: role in regulation of blood pressure in the
spontaneously hypertensive rat. Circ. Res. 38 (suppl. 2): 11-21-11-34, 197 6.

Jurf, A.N. and Blake, W.D.: Renal response to electrical stimulation in the
septum and diencephalon of rabbits. Circ. Res. 30: 322-331, 1972.

Kaada, B.R.: Somato-motor, autonomic and electrocorticographic responses to
electrical stimulation of rhinencephalic and other structures in primates,
cat and dog. Acta Physiol. Scand. 24(suppl. 83): 285 p., 1951.

Kaada, B.R.: Cingulate, posterior orbital, anterior insular and temporal pole
cortex. In Handbook of Physiology, Sec 1.2, Part 2, pp. 1345-1372, Am
Physiol. 800., Williams and Wilkins, Baltimore, 1960.

194

Kaba, H., Saito, H., Otsuka, K. and Seto, K.: Ventrolateral medullary neurons
projecting to the medial pre0ptic/anterior hypothalamic area through the
medial forebrain bundle: an electrophysiological study in rats. Exp. Brain
Res. 63: 369-374, 1986.

Kabat, H., Magoun, W.H. and Ranson, S.W.: Electrical stimulation of points in the
forebrain and midbrain. Arch. Neurol. and Psychiat. 34(5): 931-955, 1935.

Katholi, R.E., Winternitz, S.R. and Oparil, 8.: Role of the renal nerves in the
pathogenesis of one-kidney renal hypertension in the rat. Hypertension 3:
404-409, 1981.

Keim, K.L. and Sigg, E.B.: Activation of central sympathetic neurons by
angiotensin II. Life Sci. 10(1): 565-574, 1971.

Koepchen, H.P. and Thurau, K.: Uber die Entstebungsbedingungen der atemsyn-
chronen Schwankimgen des Vagustonous (Respiratorische Arrhythmie).
Pfluegers Arch. 269: 10-18, 1959.

Kortweg, G.C.J., Boeles, J.Th.F. and Ten Cate, J.: Influences of stimulation of
some subcortical areas on electorcardiogram. J. Neurophysiol. 20: 100-107,
1957.

Kostopoulos, G., Gloor, P., Pellegrias, A. and Siatitsas, L: A study of the
transition from spindles to spike and wave discharge in feline generalized
penicillin epilepsy: EEG features. Exp. Neurol. 73: 43-54, 1981.

Kumada, M., Dampney, R.A.L. and Reis, D.J.: Profound hypotension and abolition
of the vasomotor component of the cerebral ischemic response produced by
restricted lesions of medulla oblongata in rabbit. Circ. Res. 44: 63-70, 1979.

Kuypers, H.G.J.M. and Miasky, V.A.: Retrograde axonal transport of horseradish
peroxidase from spinal cord to brain stem cell groups in the cat. Neurosci.
Lett. 1: 9-124, 1975.

Kuypers, H.G.J.M.: Anatomy of the descending pathways. E Handbook of Physio-
logy, Sec. 1, Part 1., pp. 597-666, Am. Physiol. Soc., 1981.

Langhorst, P., Stroh-Wertz, M., Dittmar, K. and Camerer, H.L: Faculatative
coupling of reticular neuronal activity with peripheral cardiovascular and
central cortical rhythms. Brain Res. 87: 407-418, 1975.

Langhorst, P., Schultz, 3., Lambertz, M., Schultz, G. and Camerer, H.: Dynamic
characteristics of the 'Unspecific" brain stem system. In Central Interac-
tion Between Respiratory and Cardiovascular Control Systems, ed. by H. P.
Koepchen, S. M. Hilton and A Trzebski, pp. 30-41, Springer-Verlag, Berlin,
1980.

Langhorst, P., Schultz, G. and Lambertz, M.: Oscillating neuronal network of the
"common brain stem” system. In Mechanisms of Blood Pressure Waves, 29.-
by K. Miyakawa, pp. 257-275, Japan Sci. Soc. Press, Tokyo/Springer-Verlag,
Berlin, 1984.

195

Lappe, R.W. and Brady, M.J.: Mechanisms of the central pressor action of
angiotensin II in conscious rats. Am. J. Physiol. 246 (Regulatory Intergra-
tive Comp. Physiol. 15): R56-R62, 1984.

Lee, T.M., Yang, K.L., Kou, J.S. and Chai, C.Y.: Importance of sympathetic
mechanisms on cardiac arrhythmias induced by picrotoxin. Expt. Neurol. 36:
389-398, 1972.

Loewy, A.D., Wallach, J.H. and McKellar, S.: Efferent connections of the ventral
medulla oblongata in the rat. Brain Res. Rev. 3: 63-80, 1981.

Lofving, 3.: Cardiovascular adjustments induced from the rostral cingulate gyrus.
Acta. Physiol. Scand. 53(suppl. 184): 1-83, 1961.

Lovick, T.A., Smith, P.R. and Hilton, S.M.: Spinally projecting neurones near the
ventral surface of the medulla in the cat. J. Autonom. Nerv. Sys. 11: 27-
33, 1984.

Lovick, T.A.: Projections from the diencephalon and mesencephalon to the
nucleus paragigantocellularis lateralis in the cat. Neurosci. 14(3): 853-861,
1985.

Maekawa, K. and Purpura, D.P.: Intracellular study of lemniscal and non-specific
synaptic interactions in thalalmic ventrobasal neurons. Brain Res. 4: 308-
323, 1967.

Mancia, G., Baccelli, G., Adams, D.B. and Zanchetti, A.: Vasomotor regulation
during sleep in the cat. Am. J. Physiol. 220: 1086-1093, 1970.

Manning, Jr., J.W. and Peiss, C.N.: Cardiovascular responses to electrical
stimulation in the diencephalon. Am. J. Physiol. 198(2): 366-370, 1960.

Masserman, J.H.: Effects of sodium amytal and other drugs on the reactivity of
the hypothalamus of the cat. Arch. Neurol. Psychiat. 387: 617-628, 1937.

McAllen, R.M., Neil, J .J. and Loewy, A.D.: Effects of kaini applied to the ventral
surface of the medulla oblongata on vasomotor tone, the baroreceptor reflex
and hypothalamic autonomic responses. Brain Res. 238: 65-76, 1982.

McCall, R.B. and Gebber, G.L.: Brain stem and spinal synchronization of
sympathetic nervous discharge. Brain Res. 89: 139-143, 1975.

McCall, R.B. and Humphrey, S.J.: Evidence for GABA mediation of sympathetic
inhibition evoked from midline medullary depressor sites. Brain Res. 339:
356-360, 1985.

McDougal, A., Dampney, R. and Bandler, R.: Cardiovascular components of the
defense reaction evoked by excitation of neuronal cell bodies in the
midbrain periaqueductal gray of the cat. Neurosci. Lett. 60: 69-75, 1985.

Montero, V.M. and Singer, W.: Ultrastructural identification of soma and neural
processes immunoreactive to antibodies against glutamic acid decarboxylase
(GAD) in the dorsal lateral geniculate nucleus of the cat. Exp. Brain Res.
59: 151-165, 1985.

196

Morison, R.S. and Rioch, D.McK.: The influence of the forebrain on an autonomic
reflex. Am. J. Physiol. 120: 257-276, 1937.

Morison, R. and Dempsey, E.W.: A study of thalamo-cortical relations. Am. J.
Physiol. 135: 281-292, 1942.

Morrison, S.F. and Gebber, G.L.: Classification of raphe neurons with cardiac-
related activity. Am. J. Physiol. 243 (Regulatory Integrative Comp. Physiol.
12): R49-R59, 1982.

Morrison, S.F. and Gebber, G.L.: Raphe neurons with sympathetic-related
activity: Baroreceptor responses and spinal connections. Am. J. Physiol.
246 (Regulatorve Comp. Physiol. 15): R338-R34, 1984.

Morrison, S.F. and Gebber, G.L.: Axonal branching patterns and axonal trajec-
tories of raphespinal sympathoinhibitory neurons. J. Neurophysiol. 53(3):
759-772, 1985.

Mulle, C., Steriade, M. and Deschenes, M.: Absence of spindle oscillations in the
cat anterior thalamic nuclei. Brain Res. 334: 169-171, 1985.

Nauta, W.J.H. and Haymaker, W.: Hypothalamic nuclei and fiber connections. E
The Hypothalamus, g1. by W. Haymaker, E. Anderson and W.J.H. Natua, pp.

Ono, K. and Niimi, K.: Direct projections of the hypothalamic nuclei to the
thalamic mediodorsal nucleus in the cat. Neurosci. Lett. 57: 283-287, 1985.

Owsjannikow, P.: Die tonischen und reflectorischen Centren der Gefassnerven.
Ber. Vehr. Saechs. Wiss. Leipzig Math. Phys. Kl 23: 13 5-147, 1971.

Page, LH.: Hypertension Mechanisms. Grune and Stratton, Orlando, FL, 1987.

Palkovits, M. and Zaborszky, L.: Neural connections of the hypothalamus. In
Anatomy of the Hypothalamus, 951. by P. J. Moragne and J. Pankesepp, pp.
379-510, Marcel Decker Inc., New York, 1979.

Parent, A. and Steriade, M.: Midbrain tegmental projections of nucleus reticularis
thalami of cat and monkey: A retrograde transport and antidromic invasion
study. J. Comp. Neurol. 229: 548-558, 1984.

Peiss, C.N.: Concepts of cardiovascular regulation: past, present and future. In
Nervous Control of the Heart, e_d. by W. C. Randall, pp. 154-197, Williams
and Wilkins, Baltimore, 1965.

Powell, D.A. and Buchanan, S.: Autonomic changes elicited by chemical stimula-
tion of the mediodorsal nucleus of the thalamus. Pharmacol., Biochem,
Behav. 25(2): 423-430, 1986.

Preiss, G., Kirchner, F. and Polosa, C.: Patterns of sympathetic preganglionic
neuron firing by the central respiratory drive. Brain Res. 87: 363-374, 197 5.

Purpura, D.P. and Cohen, 3.: Intracellular recording from thalamic neurons
during recruiting responses. J. Neurophysiol. 25: 621-63 5, 1962.

197

Purpura, D.P. and Shofer, R.J.: Intracellular recording from thalamic neurons
during reticulo-cortical activation. J. Neurophysiol. 26: 496-505, 1963.

Purpura, D. P., Scarff, T. and McMurtry, J. G.: Intracellular study of intranuclear
inhibition in ventrolateral thalamic neurons. J. Neurophysiol. 28: 487-496,
1965.

Purpura, D. P., Frigyesi, T. L., McMurtry, J. G. and Scarff, T.: Synaptic mecha-
nisms in thalamic regulation of cerebello-cortical projection activity. In
The Thalamus, ed. by D. P. Purpura and M. D. Yahr, pp. 153- -170, New York,
1966.

PIn-pura, D.P.: Operations and processes in thalamic and synaptically related
neural subsystems. E The Neurosciences, _e_g. by F.O. Schmitt, pp. 453-470,
Rockefellar Univ. Press, New York, 1970.

Quesney, L.F., Gloor, P., Kratzenberg, E. and Zumstein, H.: Pathophysiology of
generalized penicillin epilepsy in the cat: the role of cortical and subcortical
structures. 1. Systemic application of penicillin. Electroenceph. clin.
Neurophysiol. 42: 640-655, 1977.

Ranson, S.W., Kabat, H. and Magoun, H.W.: Autonomic responses to electrical
stimulation of hypothalamus, preoptic region and septum. Arch. Neurol.
Psychiat. 33: 467-474, 1935.

Redgate, E.S. and Gellhorn, E.: The tonic effects of the posterior hypothalamus
on blood pressure and pulse rate as disclosed by the action of intrahypo-
thalamically injected drugs. Arch. Int. Pharmacodyn. 105: 193-198, 1956a.

Redgate, E.S. and Gellhorn, E.: The alteration of anterior hypothalamic excit-
ability through intrahypothalamic injections of drugs and its significance for
the measurement of parasympathetic hypothalamic excitability in the intact
organism. Arch. Int. Pharmacodyn. 105: 199-208, 1965b.

Reis, D.J. and Cuenod, M.: Tonic influences of rostral brain structures on
pressure regulating mechanisms in the cat. Science Wash. D.C. 145: 64-65,
1964.

Reis, D.J. and Nathan, M.A.: Hypertension and pulmonary edema in rats after
hypothalamic lesions. Circulation 44(suppl. 4): 44, 1973.

Rockhold, R.W., Jin, C.-3., Huang, C. and Farley, J.M.: Excitatory neurotoxin
stimulated cardiovascular responses following diencephalic microinjecitons.
Soc. Neurosci. Abstr. Part 2, 242.1: 825, 1985.

Rockhold, R.W., Jin, C.-B., Huang, C. and Farley, J.M.: Acute tachycardia and
pressor effects following injections of kainic acid into the antero-dorsal
medial hypothalamus. Neuropharm. 26(6): 567-573, 1987.

Ross, C.A., Ruggerio, D.A., Park, D.H., Joh, T.H., Sved, A.F., Fernandez-Parda,
J., Saaverda, F.M. and D.J. Reis.: Tonic vasomotor control by the rostral
ventrolateral medulla: effect of electrical or chemical stimulation of the
area containing C1 adrenaline neurons on arterial pressure, heart rate, and
plasma catecholamines and vasopressin. J. Neurosci. 4: 474-494, 1984.

19s

Saper, C.3., Loewy, A.D., Swanson, L.W. and Cowan, W.M.: Direct hypothalamo-
autonomic connections. Brain Res. 117: 305-312, 1976.

Saper, C.3., Swanson, L.W. and Cowan, W.M.: The efferent connections of the
anterior hypothalamic area of the rat, cat and monkey. J. Comp. Neurol.
182: 575-600, 1978.

Scheibel, M. E. and Scheibel, A. 3.: Structural substrates for integrative patterns
in the brain stem reticular care. In Reticular Formation of the Brain, ed. by
H. H. Jasper and Associates, pp. 31-55, Little, Brown, Boston, 1958.

Scheibel, M.E. and Scheibel, A.3.: The organization of the nucleus reticularis
thalami: a Golgi study. Brain Res. 1: 43-62, 1966.

Scheibel, M.E. and Scheibel, A.3.: Structural organization of nonspecific thalamic
nuclei and their projection toward the cortex. Brain Res. 6: 60-94, 1967.

Scheibel, M.E., Davies, T.L. and Scheibel, A.3.: On thalamic substrates of
cortical synchrony. Neural. 23(3): 300-304, 1973.

Schmidt, 3. and Dimicco, J.A.: Blockade of GABA receptors in periventricular
forebrain of anesthetized cats: effects on heart rate, arterial pressure and
hind limb vascular resistance. Brain Res. 301: 111-119, 1984.

Seigel, A., Edinger, H. and Troiano, R.: A pathway from the mediodorsal nucleus
of thalamus to the hypothalamus in the cat. Expt. Neurol. 38: 201-217,
1973.

Skinner, J.E. and Reed, J.C.: Blockade of a frontocortical-brain stem pathway
prevents ventricular fibrillation of ischemic heart. Am. J. Physiol. 240
(Heart Circ. Physiol. 9): H156-H163, 1981.

Smith, W.K.: The functional significance of the rostral cingular cortex as
revealed by its responses to electrical excitation. J. Neurophysiol. 8: 241-
255, 1945.

Steriade, M. and Hobson, J.A.: Neuronal activity during the sleep-wake cycle.
Prog. Neurobiol. 6: 155-376, 1976.

Steriade, M. and Deschenes, M.: The thalamus as a neuronal oscillator. Brain
Res. Rev. 8: 1-63, 1984.

Steriade, M., Parent, A. and Hada, J.: Thalamic projections of nucleus reticularis
thalami of cat: a study using retrograde transport of horseradish peroxidase
and flourescent tracers. J. Comp. Neurol. 229: 531-547, 1984.

Steriade, M., Deschenes, M., Domich, L. and Mulle, C.: Abolition of spindle
oscillations in thalamic neurons disconnected from nucleus reticularis tha-
lami. J. Neurophysiol. 54(6): 1473-1497, 1985.

Steriade, M., Domich, L. and Oakson, G.: Reticularis thalamic neurons revisited:
activity changes during shifts in states of vigilance. J. Neurosci. 6(1): 68-
81, 1986.

199

Steriade, M., Domich, L, Oakson, G. and Deschenes, M.: The deafferentated
reticular thalamic nucleus generates spindle rhythmicity. J. Neurophysiol.
57(1); 260-273, 1987.

Sun, M.-K. and Guyenet, P.G.: Hypothalamic glutamatergic input to medullary
sympathoexcitatory neurons in rats. Am. J. Physiol. 251 (Regulatory
Intergrative Comp. Physiol. 20): R798-R810, 1986.

Szentagothai, J., Flerko, 3., Mess, B. and Halasz, 3.: Hypothalamic Control of
the Anterior Pituitary, Akademiai Kiado, Budapest, 1968.

Takagi, H., Shiosaka, S., Tohyama, M., Senba, E. and Sakanaka, M.: Ascending
components of the medial forebrain bimdle from the lower brain stem in the

rat, with special reference of raphe and catecholamine cell groups. A study
by the HRP method. Brain Res. 139: 315-337, 1980.

Takeda, K. and Bunag, R.D.: Augmented sympathetic nerve activity and pressor
responsiveness in DOCA hypertensive rats. Hypertension 2: 97-101, 1980.

Takeuchi, J ., Yagi, S., Nakayama, S., Ikeda, S., Uchida, E., Inoue, G., Shintani, F.
and Ueda, .: Experimental studies on the nervous control of the renal
circulation-effect of electrical stimulation of the diencephalon on the renal
circulation. Jap. Heart J. 1(3): 288-299, 1960.

Takeuchi, J ., Yagi, S., Takeda, T., Uchida, E., Inoue, G. and Ueda, H.: Experi-
mental studies on the nervous control of the renal circulation-changes in
renal circulation induced by electrical stimulation of the diencephalon and
effect of several drugs on the circulatory response. Jap. Heart J. 3(1): 57-
72, 1962.

Tan, E., Goodchild, A.K. and Dampney, R.A.L.: Intense vasoconstriction and
bradycardia evoked by stimulation of neurones within the midbrain ventral
tegmentum of the rabbit. Clin. Exp. Pharmacol. Physiol. 10: 305-309, 1983.

Taylor, D.G. and Gebber, G.L.: Baroreceptor mechanisms controlling sympathetic
nervous rhythms of central origin. Am. J. Physiol. 228: 1002-1013, 1975.

ter Horst, G.J. and Luiten, P.G.M.: The projections of the dorsomedial hypotha-
lamic nucleus in the rat. Brain Res. Bull. 16: 231-248, 1986.

Tobey, J.C., Fry, H.K., Mizejewski, C.S., Fink, G.D. and Weaver, L.C.: Differen-
tial sympathetic responses iniated by angiotensin and sodium chloride. Am.
J. Physiol. 245 (Regulatory Integrative Comp. Physiol. 14): R60-R68, 1983.

Touw, K.B., Haywood, J.R., Shaffer, R.A. and Brady, M.J.: Contribution of the
sympathetic nervous system to vascular resistance in conscious young and
adult spontaneously hypertensive rats. Hypertension 2: 408-418, 1980.

Tucker, D.C. and Saper, C.3.: Specificity of spinal projections from hypothalamic
and brain stem areas which innervate sympathetic preganglionic neurons.
Brain Res. 360: 159-164, 1985.

200

Van der Kooy, D., Koda, L.Y., McGinty, J.F., Gerfen, C.R. and Bloom P.E.: The
organization of projections from the cortex, amygdala and hypothalamus to
the nucleus of the solitary tract in rat. J. Comp. Neurol. 224: 1-24, 1984.

Velayos, J .L.: Topographical organization of the prosencephalic afferents to the
mediodorsal thalamic nucleus. Neurosci. Lett. Suppl. 10: S502, 1982

Velasco, F., Velasco, M., Estrada-Villanueva, F. and Machado, J.P.: Specific and
nonspecific multiple unit activities during the onset of pentylenetetrazol
seizures. 1. Intact animals. Epilepsia 16: 207-214, 1975.

Villablanca, J.: Role of the thalamus in sleep control: Sleep-wakefulness studies
in chronic diencephalic and athalamic cats. I_n_ Basic Sleep Mechanisms, 251.
by O. Petre-Quadens and J.D. Schlag, pp. 51-81, Academic Press, New Yok,
1974.

Wall, P.D. and Pribram, K.H.: Trigeminal neuroanatomy and blood pressure
responses from stimulation of lateral cerebral cortex of macaca mulatta. J.
Neurophysiol. 13:409-412, 1950.

Wang, S.C. and Ranson, S.W.: Autonomic responses to electrical stimulation of
the lower brain stem. J. Comp. Neurol. 71: 437455, 1939.

Wang, S.C. and Ranson, S.W.: The role of the hypothalamus and preoptic region in
the regulation of heart rate. Am. J. Physiol. 132: 5-8, 1941.

Wang, S.C. and Chai, C.Y.: Central control of sympathetic cardioacceleration in
medulla oblongata of the cat. Am. J. Physiol. 202: 31-34, 1962.

Williford, D.J., Dimicco, J.A. and Gillis, R.A.: Evidence for the presence of a
tonically active forebrain GABA system influencing central sympathetic
outflow in the cat. Neuropharmacol. 9: 245-250, 1980.

Wilson, C.L., Matter, B.C. and Lindsley, D.B.: Influences of hypothalamic
stimulation upon septal and hippocampal electrical activity in the cat.
Brain Res. 107: 55-68, 1976.

Winters, W.D. and Spooner, C.E.: A neurophysiological comparison of alpha-
chloralose with gamma-hydroxybutyrate in cats. Electroenceph. clin.
Neurophysiol. 20: 83-90, 1966.

Zhang, T'.-X. and Ciriello, J .: Effect of paraventricular nucleus lesions on arterial

pressure and heart rate after aortic baroreceptor denervation in rat. Brain
Res. 341: 101-109, 1985a.

Zhang, T.-X. and Cireillo, J .: Kainic acid lesions of paraventricular nucleus
neurons reverse the elevated arterial pressure after aortic baroreceptor
denervation in the rat. Brain Res. 358: 334-338, 1985b.