«

J...

2.).) .1 § 5

:ta

.m

 

 

 

..
an}; .9.
.

1... mmlmmvo:

lit
“a“. r!

an.

.15!- 5:,4?
.‘Litunwr .-
«5:153.- if.
Eff:

_ 2
41.3.: a

. .35. L.
as...
2' - '1. I I t
3: I42.» x at... antenflxzf .
’1‘. u- ‘l ‘33 h0‘l.’li3}’.¢h K «
Q. I - o-V T
tux... o). .. .
III-5 2.
‘4 : [311:1
- ....

n a.
lie.
. bk: (.1.

31‘ 2|.

.3! 1...“ .. s!
.‘c. .\ (-1.? In?
1......1931... ..

3.. 91....

r a ('11.! .3 Run!)
In?» t . .II

.I {.v $14 3v...

. 5 In: 3:}.

.1. .Ivuv. "ntliliwruzg.

.1 Influx. . » ..

 

l HEOIO

fl
1'»

Z (:0

 

LIBRARY

Michigan State
Unlverslty

 

 

 

This is to certify that the

dissertation entitled

Life History Implications of Ribosomal
RNA Gene Copy Number in Escherichia goii

presented by

Bradley Scott Stevenson

has been accepted towards fulfillment
of the requirements for

Ph. D. degree in Microbiology

 

 

Vex @ at

Major professor

Date 4/1912000

MSU is an Affirmattw Action/Equal Opportunity Institution 0-12771

PLACE iN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

MAYIOZODI
0209 09.

101204
JUL 01 2004‘

 

'v

 

 

 

 

 

 

 

 

 

 

 

 

 

 

moo mums-p.14

 

LIFE HISTORY IMPLICATIONS OF RIBOSOMAL RNA GENE COPY NUMBER
IN ESCHERICHIA COLI

BY

Bradley Scott Stevenson

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Microbiology and Center for Microbial Ecology

2000

ABSTRACT

LIFE HISTORY IMPLICATIONS OF RIBOSOMAL RNA GENE COPY NUMBER
IN ESCHERICHIA COLI

By

Bradley Scott Stevenson

Unlike the majority of prokaryotic genes, the ribosomal RNA (rRNA) genes
can be found in multiple copies on the chromosome. The number of rRNA gene
copies varies widely between different prokaryotic species, but among isolates of
the same species the number of rRNA gene copies appears to be conserved.
The research in this dissertation investigates the potential selective forces behind
the occurrence of multiple rRNA gene copies in prokaryotes, and the proposed
link between rFiNA gene copy number and the life history of a species. The
hypotheses tested state that multiple rRNA operons are an advantage in
environments characterized by fluctuations in nutrient availability, and that fewer
rRNA operons are an advantage under stable conditions where growth is limited
by nutrient availability. The relationship between rRNA gene copy number and
life history was investigated in strains of Escherichia coli in which additional rRNA
genes were added on plasmids or deleted from the chromosome. Additional
plasmid-borne rRNA gene copies caused decreased growth rates and higher
concentrations of RNA in slow growing batch cultures, suggesting multiple rRNA

gene copies are a metabolic expense at slower growth rates. The competitive

ability of each rFlNA gene deletion strain (ArmA, ArmB, and ArmAB) was
measured relative to a wild type control strain in batch and chemostat
competition experiments representing fluctuating and stable environments,
respectively. Individual growth parameters (maximal growth rate, lag time, Ks,
death rate, and yield) for each strain were measured and used in a dynamic
computer model programmed to simulate the competition experiments. The
model represented the dynamics of two populations competing for the same
limiting nutrient, and accurately predicted the outcome of competition
experiments under batch or chemostat culture conditions. As predicted by the
model, the control strain was shown to have a higher competitive ability in batch
culture competition experiments with the rRNA gene deletion strains, due to
slightly faster growth rates and shorter lag times. These results support the
hypothesis that multiple rFlNA operons are an advantage when nutrient
availability fluctuates. The chemostat competition experiments however, did not
support the hypothesis that fewer rFlNA genes are an advantage in stable,
limiting nutrient conditions, because the rRNA deletion strains had lower
competitive ability than the control strain. Interestingly, the decrease in relative
fitness measured in chemostats depended upon which rFlNA operon was deleted
from the chromosome (either rrnA or rmB), suggesting that the seven rRNA
operons in E. coli may have differential properties that are poorly understood.
Further research to confirm this observation and study the mechanism for

differentiation between the rrnA and rmB operons is currently underway.

DEDICATION

To my parents Will and Sue, and my wife Katie, for your endless love, guidance,
and support.

ACKNOWLEDGMENTS

I am thankful for the support of my parents, Will and Sue, my wife, Katie,
and all of my family, throughout my life. The guidance of my mentor, Thomas
Schmidt, was essential to my development as a scientist and is a resource that
will continue to shape my career. I thank the rest of my graduate committee,
Helmut Bertrand, John Breznak, Richard Lenski, and Michael Thomashow, for
their guidance and insight. Thanks to all of the past and present members of the
Schmidt lab; Bonnie Bratina, Dan Buckley, Les Dethlefsen, Joel Hashimoto, Joel
Klappenbach, and Paul Lepp, as well as my fellow graduate students, Keri
Byzek, Joe Graber, Deb Hogan, and Mark Johnson, for their assistance,
suggestions, lively discussion, and humor. Thanks to Kate Delaney and Rachael

Collet for their logistical support of various experiments.

TABLE OF CONTENTS

LIST OF TABLES .............................................................................................. Vlll
LIST OF FIGURES ............................................................................................... X
CHAPTER 1 ......................................................................................................... 1
1.1 Central observations ................................................................................... 1
1.2 The rRNA genes ......................................................................................... 4
1.3 Non-adaptive explanations for multiple rRNA operons and their
conservation ..................................................................................................... 6
1.4 Multiple rRNA operons as an adaptation .................................................... 8
1.5 Role of rRNA operon copy number as a life history trait ............................. 9
CHAPTER 2 ....................................................................................................... 12
2.1 Introduction ............................................................................................... 12
2.2 Materials and Methods .............................................................................. 13
2.3 Results ...................................................................................................... 16
2.4 Conclusions .............................................................................................. 22
CHAPTER 3 ....................................................................................................... 23
3.1 Introduction ............................................................................................... 23
3.2 Materials and Methods .............................................................................. 27
Bacterial strains, plasmids, and phage ........................................................ 27
Bacterial growth conditions, media .............................................................. 30
General cloning procedures ........................................................................ 30
PCR and XL PCR amplification ................................................................... 31
Visualization of rRNA operons using southern hybridization ....................... 32
Deletion of the rmB operon ......................................................................... 33
Deletion of the rrnA operon ......................................................................... 38
3.3 Results ...................................................................................................... 41
3.4 Discussion ................................................................................................ 47
Deletion of rRNA operons ........................................................................... 47
Conclusion .................................................................................................. 50
3.5 Lessons learned during construction of rRNA deletion strains .................. 50
Generalized transduction between strains of E. coli .................................... 50
Gene conversion ......................................................................................... 51
Marker gene orientation .............................................................................. 54
Temperature sensitive plasmid ................................................................... 54
CHAPTER 4 ....................................................................................................... 55
4.1 Introduction ............................................................................................... 55
4.2 Materials and Methods .............................................................................. 58
Bacterial strains ........................................................................................... 58
Media and culture conditions ....................................................................... 58
Measurement of specific growth parameters ............................................... 62
Measurement of other physiological traits ................................................... 65
Batch culture competition experiments ........................................................ 66
Chemostat culture competition experiments ............................................... 67
Calculation of chemostat-specific growth parameters ................................. 69
4.3 Results ...................................................................................................... 69

vi

Specific growth parameters ......................................................................... 69

Measured physiological traits ...................................................................... 73
Batch culture competition experiments ........................................................ 75
Chemostat competition experiments ........................................................... 77
Calculated chemostat-specific growth parameters ...................................... 81
4.4 Discussion ................................................................................................ 81
Effect of rRNA copy number on relative fitness in fluctuating nutrient
conditions .................................................................................................... 83
Effect of rRNA operon copy number on relative fitness in stable nutrient
conditions .................................................................................................... 86
Possible impact of tRNA imbalance ............................................................ 89
Proposed explanation of cumulative results ................................................ 90
Potential limitations of comparing rRNA deletion strains to control strains.. 92
Conclusion .................................................................................................. 93
CHAPTER 5 ....................................................................................................... 95
5.1 Introduction ............................................................................................... 95
5.2 Methods .................................................................................................... 97
Identification of variables ............................................................................. 97
Parameterization ....................................................................................... 1 00
Modeling method ....................................................................................... 100
Sensitivity analyses ................................................................................... 101
Verification of model performance ............................................................. 102
Validation of model performance ............................................................... 103
5.3 Results .................................................................................................... 103
Sensitivity analyses ................................................................................... 103
Model Verification ...................................................................................... 108
Model Validation ........................................................................................ 1 16
5.4 Discussion .............................................................................................. 121
CHAPTER 6 ..................................................................................................... 127
APPENDIX A .................................................................................................... 136
APPENDIX B .................................................................................................... 138
APPENDIX C .................................................................................................... 140
REFERENCES ................................................................................................. 141

vii

LIST OF TABLES

Table 2.1 Specific growth rates (it) of the host strain and transformants

carrying pE2200 or pE2211, grown in different media

Table 2.2 Concentration of RNA (fg/fl) of the host strain and

transformants carrying pE2200 or pEZ211, grown in different media

Table 2.3 Specific growth rate (it) for pE2211ASaI transformants in

different media

Table 3.1 Bacterial strains

Table 3.2 Plasmids and Phage

Table 4.1 Growth parameters measured for each strain‘al

Table 4.2 Chemostat-specific growth parameters

Table 5.1 Parameter values and results for a simulation of previously

published batch culture competition experiments (76)

viii

Table 5.2 Parameter values for simulations of previously published

chemostat culture competition experiments (37)

Table 5.3 Comparison of reported and calculated J parameters for

chemostat model verification

Table 5.4 Calculated J parameters for chemostat model

validation at different dilution rates (D)

LIST OF FIGURES

Figure 1.1 Phylogenetic relationship and rRNA operon copy number of
environmental isolates and representative prokaryotic species. Major

phylogenetic groupings are indicated on the right. Figure taken from (46).

Figure 1.2 Typical organization of prokaryotic rRNA operon.

Figure 2.1 (A) Map of pEZ211 showing the 7.0 kb rrnB operon of E. coli
cloned into pBR322 between the BamHI and SgrAI restriction sites. (8)

Linearized representation of rrnB insert in pEZ211, removal of rm promoters P1

and P2 in pE2200, and internal deletion in pEZ211ASaI.

Figure 2.2 Cellular RNA content (lg/cell) as a function of growth rate, p.
(h") for the evolved host strain (A), and transformants with pE2200 (O) or
pEZ211 (0) compared with previously published values for E. coli B (A; (11)).

Error bars represent sample standard deviations.

Figure 3.1 Location (in min) of the seven rRNA operons (rm) and the
origin of replication (on‘C) on the E. coli K-12 chromosome (22). Arrows indicate

direction of rRNA operon transcription.

Figure 3.2 The (A) general organization (19), and (B) tRNA genes of the

seven E. coli rRNA operons (61, 62).

Figure 3.3 Steps used to delete the rmB operon from the REL 4548 and

BS 4548 genome, creating the BS 200 and BS 201 (ArmB) strains.

Figure 3.4 Regions of the E. coli chromosome containing the rrnA and
rmB operons, showing the amplified flanking sequences (UpA, DnA, UpB, and
DnB; white boxes), structural rRNA genes (168, 238, and SS; dark grey boxes),
tRNA genes (tRNA“°, tRNAA'813, and tRNAG'”2; dark grey boxes), and adjacent

genes (trkH, hemG, mobB, murl, murB; light grey boxes).

Figure 3.5 Oligonucleotide primers used throughout rRNA operon deletion
procedure. The non-target specific tails at 5’ end of each primer are shown in
boxes, the recognition sequences of restriction endonuclease are labeled and

shown in shaded area of box, and the arrows indicate the restriction sites.

Figure 3.6 Southern hybridization of genomic DNA of each strain digested
with Pqu and hybridized with a 168 DIG-labelled probe. Hybridized bands
corresponding to each of the seven rRNA operons in the control strain are
labeled on the left. Sizes (kb) of each band of the A HindlII DNA marker are

labeled on the right.

xi

Figure 3.7 XL PCR amplification of the wild type rrnB operon (lane 5, 6.1
kb), rmB-specific marker allele (lane 6, 5.4 kb), plasmid-borne UszDnB deletion
fragment (lane 7, 1.8 kb), and chromosomal rrnB deletions (lanes 9-16, 1.8 kb).
Fragments sizes of a 1 kb DNA ladder in lanes 1 and 19 are marked to the left of

lane 1.

Figure 3.8 XL PCR amplification of wild type rrnA (lane 2) and rmB (lane
11) operons, intermediate rrnA-specific marker allele (lane 3), rrnA deletion in BS
101 (lane 4), rrnB deletion in B8 201 (lane 12), and both rrnA and rmB deletions
from BS 301 strain isolates (lanes 5-9 and 13-17 respectively). Fragment sizes

of a 1 kb DNA ladder (lanes 1, 10, and 18) are marked to the left of lane 1.

Figure 3.9 Southern hybridization of a repaired chromosomal rmB
operon. Hybridized DNA fragments that correspond with the wild type rmB
operon (1), the marker rmB allele (2), and the hybrid rmB operon (3) are labeled

with arrows.

Figure 4.1 (A) Diagram of chemostat culture system. (B) Sideview of

individual chemostat culture vessel.

Figure 4.2 The death rate is shown as the change in population density

(CFU/ml) of each strain in batch monocultures from 12 to 24 hours.

xii

Figure 4.3 (A) Cell yield (106 cells/pg glucose), and (B) cell volume (fl) for
the control (black), ArrnB (dark grey), ArmA (light grey), and ArmAB (white)
strains are shown for chemostat monocultures at slow (0.11 h") and fast (0.56
h") dilution rates. Error bars represent the standard error of the measurements

of two independent cultures.

Figure 4.4 Relative fitness values for the control (black), ArmB (dark
grey), ArmA (light grey), and ArmAB (white) strains in batch culture competition
experiments with the control strain. The value for the control represents the
relative fitness of the Ara+ versus the Ara' control strain. Error bars represent

standard errors of the measurements.

Figure 4.5 Relative fitness of each rRNA deletion strain in chemostat
competition experiments with slow dilution rates (0.11 h") is represented as the
population ratio (vs. control) versus the number of population doublings. The
data for the control represents the fitness of the Ara+ strain relative to the Ara'
strain. Error bars represent the standard error of four independent cultures for

each strain.

Figure 4.6 Empirical (solid lines with symbols) and simulated (dotted lines)

relative fitness of ArmB in fast (black) and slow (dark grey) chemostat

competition experiments are represented as the population ratio (ArmB vs.

xiii

control) versus the number of population doublings. Error bars indicate the

standard error of four independent cultures.

Figure 4.7 Empirical (solid lines with symbols) and simulation (dotted
lines) relative fitness of ArrnA in fast (black) and slow (dark grey) chemostat
competition experiments are represented as the population ratio (ArmA vs.

control) versus the number of population doublings. Error bars represent the

standard error of four independent cultures.

Figure 4.8 Empirical relative fitness values for the control (black), ArrnB
(dark grey), ArmA (light grey) and ArmAB (white) strains in batch culture

competition experiments, and relative fitness values based on simulations using

all measured parameters or each parameter individually.

Figure 5.1 Map showing the relationship between stocks ([1), converters (O),
and flows (073:5) of the model programmed in the object-oriented language
STELLA.

Figure 5.2 Batch culture sensitivity analyses. The population ratio
(Population 1/Population 2) after 24 h simulations (labeled as Sensitivity) was
plotted against the ratio of the tested parameters (uMAx1/uMAx2, lagZ/lag1, and

KSZ/Ks1).

xiv

Figure 5.3 Sensitivity analyses of chemostat simulations. The sensitivity
(change in population ratio (Pop1/Pop2) per doubling) of model parameters
under fast (0.81 h") and slow (0.11 h") chemostat dilution rates is plotted against

the ratio of the parameters (ltMAxl/ILMAXZ) and (K32/Kst).

Figure 5.4 Batch culture verification data. The dynamics of the ancestor
(left axis, solid line) and derived (left axis, dashed line) populations, and the
concentration of glucose (right axis, dotted line), using the estimated parameters

in (76) are shown.

Figure 5.5 Population density dynamics for chemostat verification

simulations using parameters in Table 5.2 for experiment A, B, and C.

Figure 5.6 Validation of model performance under batch culture
conditions. Empirical versus simulation relative fitness values are shown for the

control (black), ArmB (dark grey), ArrnA (light grey), and ArmAB (white) strains.

Figure 5.7 Population ratio dynamics for slow chemostat validation
simulations of direct competition experiments with (A) ArrnB, (B) ArmA, and (C)

ArmAB strains vs. the control strain using parameters in Chapter 4.

XV

Figure 5.8 Chemostat validation at fast dilution rates. Empirical and
simulation data for (A) ArrnB/control and (B) ArmA/control population ratio versus

doublings during fast chemostat dilution rates (0.81 h")

xvi

CHAPTER 1
INTRODUCTION
1.1 Central observations
The rRNA genes can be found in multiple copies on a prokaryotic
chromosome, whereas the majority of genes are present as single copies. The
numbers of rRNA genes can vary widely between different prokaryotic species
(Figure 1.1) (27, 69). Many species contain only one or a few copies of the rRNA
genes, but some have up to 14 as in Clostridium beijen’nckii (77). While the
number of rRNA genes can vary between different prokaryotic species, among
isolates of the same species the number of genes seems to be conserved (13).
The research in this dissertation attempts to address why some prokaryotes
contain multiple copies of the rRNA genes, and why the numbers of copies vary

between different prokaryotic species.

Figure 1.1 Phylogenetic relationship and rRNA operon copy
number of environmental isolates and representative prokaryotic species.
Major phylogenetic groupings are indicated on the right. Figure taken

from (46).

Escherichia coli
:: Haemophilus influenzae
r— Vibrio chloerae
L— symbiont of K alfnedi

HF3-SZ1027
I E Pseudomonas sfutzeri
Pseudomonas putida

HSS-SZ4542
KBS-LCZ
HS7-S24561
Variovorax sp.
KBS-LC8
KBS-LC1

HF2-821012
_E KBS-EC4
HF5-821 125

I Burkholden'a cepacia
‘I— Neisseria gononhoeae
HSZ-SZ4520
Bradyrhizobium japonicum
HS6-SZ4556
HS1-823321
HSS-SZ4538
Agrobacterium tumefaciens
Sinorhizobium meliloti
HS4-324541
Caulobacter crescentus
Sphingomonas sp. RB2256
Rhodobacter sphaeroides
Rickettsia prowazekii
Helicobacter pylon”
—: Camp ylobacter jejuni
Bacteroides fragilis
Bacteroides uniformis
KBS-ECZ

KBS-LC6
l E KBS-LC4
KBS-LCQ

r— Synechococcus PCC6103
1— Anabaena cylindrica

—‘E: Treponema pallidum
Bone/la burgdon‘eri

KBS-EC3
Arthrobacter globiformis
HF 1 -821 003

KBS-ECS
Mycobacterium Ieprae

: Mycobacterium smegmatis
Streptomyces venezuelae
Streptomyces ambofaciens
HF6-822243
Mycoplasma pneumoniae
Streptococcus pneumoniae
Listeria monocytogenes
KBS-ECl

HF 4-821 032

Bacillus cereus

Bacillus subtilis

Clostridium beijerinckii 13
—: Clostridium acetobutylicum 11 -

C Aquifex pyrophilus
Thennus therrnophilus

 

 

 

 

 

Proteobacteria

 

 

 

 

m

 

 

i|_.ll__.ll_——__ILJl

SPR CYN ClF/B

 

High G+C
Gram Positive

Swssmub-lwmflNd-b-thI-lNNNN-l-Ih-bmwn-tu-Ln—pr”aaaaggmamugnanagggdmgq
LowG+C

 

 

 

 

 

N O
l_I
TRM

Figure 1.1

1.2 The rRNA genes

DNA sequences within the rRNA genes are highly conserved among all
organisms. The rRNA genes (in prokaryotes, 168, 238, and SS rDNA) are
expressed from strong promoters that are tightly regulated to coordinate the
amount of rRNA (and therefore ribosomes) within the cell. Expression of the
rRNA genes is controlled by the physiological condition of the cell. Many other
genes involved in protein synthesis are controlled in connection with the
expression of the rRNA genes. For instance, the binding of rRNA with ribosomal
proteins (r-proteins) relieves the r-protein promoters from feedback inhibition.
The rRNA genes are typically organized into operons that have one or more
strong promoters, followed by the 168 rRNA gene, an internally transcribed
spacer (ITS) region that may contain one or more tRNA genes, the 23S and SS
genes, and occasionally one or more distally-located tRNA genes followed by
termination sequences (Figure 1.2). There are exceptions to this consensus,
especially among some of the Archaea, in which the rRNA operons lack ITS
tRNAs, or where one or all of the rRNA genes are distributed around the

chromosome, behind their own promoters (as reviewed in 28).

Multiple Internally Transcribed Distal
Promoters Spacer Spacer

P1 P2 |.__| H
rrs tRNA(s) n1 m

l l l

16S rRNA 23S rRNA 58 rRNA

Figure 1.2 Typical organization of prokaryotic rRNA operon.

1.3 Non-adaptive explanations for multiple rRNA operons and their
conservation

There are several possible non-adaptive explanations for the presence of
multiple rRNA operons on a prokaryotic chromosome and the variation in rRNA
operon copy number among different prokaryotes. Some of the explanations that
have been explored suggest that rRNA operon copy number in prokaryotes may
be related to genome size, the result of genetic drift, and/or largely determined by
evolutionary history. Duplications of regions of the genome that include the
rRNA genes could lead to larger genome sizes as well as an increase in number
of rRNA operons. The rRNA genes themselves represent obvious sites where
unequal recombination events could lead to the duplications mentioned above.
The number of rRNA operons however, is not a function of genome size. Two
recent studies have concluded that there is not a significant correlation between
genome size and rRNA gene copy number (27, 46). Two examples that illustrate
the poor correlation between rRNA gene copy number and genome size are
Bradyrhizobium japonicum, which has a very large genome (~8.7 Mb) and one
copy of the rRNA genes, and Haemophilus influenzae, which has a small
genome (~1.8 Mb) but six copies of the rRNA genes (26, 50).

A lack of conservation in the sequence, organization, and number of rRNA
operons might be expected if the operons were subject to considerable genetic
drift. As discussed above however, large portions of the sequence and the
overall organization of rRNA operons are generally conserved among all

prokaryotic species. When many isolates of the same prokaryotic species have

been surveyed, the number of prokaryotic rRNA operons is highly conserved
(9.9. 13). The number of rRNA operons might differ between certain serovars of
Vibrio cholerae (43) or strains in Salmonella typhimurium (4), but in both of these
cases the difference was by a single operon, constituting a small deviation in
overall rRNA copy number. It is unlikely that genetic drift is the sole factor that
determines the rRNA operon copy number of prokaryotic species.

Any comparative analysis must take into consideration the independence
of each observation. It is therefore important to consider the impact of phylogeny
on rRNA operon copy number. For example, a correlation between a high
number of operons and a shared physiological trait might be used to argue a
causative relationship. If all of the species that shared this physiological trait and
the high rRNA operon copy number were also closely related, an equally
compelling argument could be made that this correlation was due to their
common ancestry. Also, if the rRNA operon copy number of a prokaryotic
species were largely determined by evolutionary history, closely related species
would be expected to have the same number of rRNA operons. To the extent
that bacterial diversity has been studied however, phylogeny does not seem to
play an obvious role in determining the number of rRNA operons in prokaryotes
(Figure 1.1, (69)). There is no obvious link between phylogeny and rRNA operon

copy number beyond the species level.

1.4 Multiple rRNA operons as an adaptation

The occurrence of multiple rRNA operons in prokaryotes could be the
result of an adaptation to certain environmental conditions. Multiple rRNA
operons could allow faster rates of rRNA synthesis, faster growth, and the ability
to quickly respond to more favorable growth conditions. The rRNAs are the end
product of the rRNA operons; and unlike other genes the rRNA gene products
cannot be amplified through translation. Multiple rRNA operons on a
chromosome can amplify rRNA synthesizing capacity by offering more sites for
rRNA synthesis within the cell to meet the large demand for ribosomes under fast
growth. The synthesis of rRNA constitutes over 50% of the transcriptional
activity of a fast-growing Escherichia coli cell, and one rRNA operon would not be
sufficient to supply all of the ribosomes needed at the fastest growth rates (11).
Faster growth rates were therefore thought to be the main reason for the
occurrence of multiple rRNA operons in E. coli.

Multiple rRNA operons do allow for higher rates of rRNA synthesis and
growth, but recent evidence suggests that maximal growth rates may not be the
main determinant of rRNA copy number (11, 16). Up to four of the seven rRNA
operons in E. coli can be inactivated without significantly affecting the rate of total
rRNA synthesis (16), suggesting that E. coli cells contain more rRNA operons
than are needed at their maximal observed growth rates. This observation has
been used to suggest that although multiple rRNA operons can support faster
rates of rRNA synthesis, they increase the cell’s ability to quickly respond to

fluctuating nutrient conditions (17).

1.5 Role of rRNA operon copy number as a life history trait

The number of rRNA operons on the prokaryotic chromosome is well
conserved among isolates of the same species (9.9. 13), but rRNA operon copy
number is not necessarily conserved between even closely related species
(Figure 1.1, and (69)). The conservation of rRNA operon copy number within a
species but not between different species is a common life history pattern. The
life history pattern of a species is a concept developed by macroecologists to
describe the lifetime pattern of growth, differentiation, storage and, most
importantly, reproduction. An observed life history is the result of long-term
evolution, as well as the more immediate forces of the environment in which a
species is and has been living (9). The number of rRNA operons could be a trait
that has a major impact on the life history pattern of a species. The main goal of
the work presented here was to investigate the mechanistic connection between
the life history of an organism and its rRNA operon copy number.

The first hypothesis tested in this dissertation research stated that multiple
rRNA operons are an advantage to an organism in environmental conditions
defined by fluctuations in nutrient availability. Cells with more rRNA operons
have more sites from which to synthesize rRNA after an influx of nutrients.
Multiple rRNA operons can therefore result in faster growth rates, as well as
provide the cell with a larger capacity for de novo ribosome synthesis following a

shift to favorable growth conditions, such as an influx of nutrients.

The inactivation of rRNA operons in strains of E. coli may not significantly
affect growth rate, but the time required for the cells to respond to nutrient
fluctuations increases considerably (17). Additionally, E. coli cells growing at
very slow rates accumulate excess ribosomes, which could offer the cell an
advantage upon a shift to more favorable growth conditions (47). In Chapter 3,
one or two rRNA operons were deleted in strains of E. coli. These deletions
were unique from all previous manipulations of chromosomal rRNA operons (3,
5, 18, 23) because they did not introduce any additional marker genes and the
tandem rRNA operon promoters were removed, causing them to be
transcriptionally silent. These strains with 5 to 7 rRNA operons were used in
Chapter 4 to study the impact of rRNA operon copy number on cell physiology
and the strains’ ability to compete in environments with frequent fluctuations in
limited nutrient availability.

The second specific hypothesis tested addressed the potential trade-off
for multiple rRNA operons. It stated that cells with fewer rRNA operons are at an
advantage under conditions in which growth is controlled by stable, limiting
nutrient concentrations. This hypothesis was based upon evidence that E. coli
cells with multiple rRNA operons have excess ribosomes at very slow growth
rates (47). These excess ribosomes are most likely the result of constitutive
expression of the rRNA operons (68) and therefore; cells with fewer rRNA
operons would contain fewer excess ribosomes at very slow growth rates. The
excess ribosomes might be an advantage upon an influx of nutrients, but their

synthesis also represents a large metabolic expense to cells under stable

10

conditions in which they will not be utilized (47). Cells with fewer rRNA operons
that contain fewer excess ribosomes would not experience the same metabolic
expense of diverting limited biosynthetic resources away from cell division. In
Chapter 2, plasmid-borne rRNA operons were used to study the effect of
additional rRNA operons on E. coli cells under various growth rates. Also, direct
competition experiments between cells differing in rRNA operon copy number
were run in chemostats with slow dilution rates in Chapter 4. These experiments
were used to test for any competitive advantage that may exist in the rRNA
deletion (Arm) strains (Chapter 3), compared to the wild type control strain.

The empirical measurements of individual growth parameters and
cumulative relative fitness (Chapter 4) of each strain developed in Chapter 3
were used in the computer model described in Chapter 5. This model was
programmed to simulate the direct competition of two populations for a shared
limiting nutrient in either batch or chemostat culture conditions. After
programming and verifying the performance of the model, the individual growth
parameters of each strain of E. coli were used to simulate the direct competition
experiments. The simulation results were used as a validation of the model’s
ability to accurately simulate the dynamic changes in population density and
nutrient concentration in direct competition experiments under both batch and

chemostat culture conditions.

11

CHAPTER 2‘1

GROWTH RATE-DEPENDENT ACCUMULATION OF RNA FROM PLASMID-

BORNE RIBOSOMAL RNA OPERONS IN ESCHERICHIA COLI

2.1 Introduction

Most prokaryotic genes are present as a single copy on the chromosome.
Exceptions to this generality include the ribosomal RNA (rRNA) genes, which are
frequently organized into an operon. The number of rRNA operons can range
from one to as many as thirteen copies per chromosome (69). This variation and
the absence of any obvious correlation between rRNA operon copy number and
organismal phylogeny (69) led us to investigate the potential adaptive
significance of rRNA operon copy number.

It has been assumed that the number of rRNA operons is directly
proportional to maximal growth rates (42). Although an organism with one rRNA
operon may not be able to achieve the maximum growth rates obtainable by an
organism with a higher multiplicity of rRNA operons, deletion of one or two rRNA
operons in E. coli has a marginal effect on maximal growth rate (16, 23). The
inactivation of single or multiple rRNA operons in E. coli does, however, influence
the time required to shift to faster growth rates upon encountering more favorable
growth conditions (17). The enhanced capacity for a rapid response to favorable
growth conditions suggests that multiple rRNA operons may be an evolutionary

adaptation and advantage to organisms that experience fluctuating growth

 

a Stevenson and Schmidt 1998. J. Bacteriol. 180 (7): 1970-72
1 2

conditions (19, 69). A potential tradeoff for this enhanced capacity is the
synthesis of superfluous rRNA and ribosomes under constant, slow growth
conditions. While transcription of the rRNA operons is proportional to growth rate
in moderate to fast growing cells, the same relationship does not exist at slow
growth rates (19, 32, 69). Under slow growth conditions in a chemostat, the

presence of “excess” ribosomes has been observed in populations of E. coli (47).

2.2 Materials and Methods

The potential disadvantages of multiple rRNA operons in E. coli were
investigated in this study by increasing rRNA gene dosage using plasmid-borne
rRNA operons. E. coli strain B/r which had been propagated for 10,000
generations in glucose minimal medium (56) was selected for this study to
minimize changes in growth rate due to adaptation to laboratory growth
conditions. Cells were transformed with plasmids derived from pKK3535, which
contains the rrnB operon from E. coli (12). Plasmid pE2211 contains the
functional rmB operon (Figure 2.1A) and was made by digesting pKK3535 with
the restriction endonucleases SgrAI and MunI (Boehringer Mannheim,
Indianapolis, IN) to remove the upstream A promoter PL57 (44). Sequence
determination of this region revealed a larger fragment of 1,756 bp was removed,
apparently due to “star” or non-specific restriction activity of SgrAI. This and
subsequent digestions left overhanging ends which were polished with T4 DNA
polymerase (New England Biolabs, lnc., Beverly, MA) and re-ligated using T4

DNA ligase (Boehringer Mannheim). Restriction, polymerase, and ligation

13

reactions were performed according to manufacturers’ protocols. The rRNA
promoters P1 and P2 were deleted in a second plasmid, pE2200 by digesting

pKK3535 with the restriction endonucleases SgrAI and BgIII (Figure 2.1 B).

14

 

 

 

B
Plasmid
pEZ21 1
pE2200
8’ S?
D D
pE221 1 ORF 1W Mom: 2—
ASal

Figure 2.1 (A) Map of pEZZ11 showing the 7.0 kb rmB operon of E. coli
cloned into pBR322 between the BamHI and SgrAI restriction sites. (B)
Linearized representation of rrnB insert in pEZ211, removal of rrn promoters P1
and P2 in pEZ200, and internal deletion in pE2211ASal.

15

Growth rates for each transformant were determined by measuring the
change in optical density (O.D.6oo) as a function of time. Growth rate
experiments were initiated by inoculating 5 ml of Luria broth (LB) (67) from
culture stocks preserved at -80°C. Following overnight incubation at 37°C,
triplicate conditioning cultures were inoculated (1:100) in 5 ml test medium. The
test media consisted of modified Davis minimal medium (DM) (56) or LB. DM
and LB were both amended with 200 ug/ml ampicillin when used to grow
transformants. DM was further amended with either 1 mg/ml sodium acetate
(Ac), sodium acetate plus 40 pg/ml each of 10 essential L-amino acids (Sigma
Chemical Co., St. Louis, MO) (Ac-AA), 1 mg/ml glucose (Glc), or glucose plus
amino acids (Glc-AA). LB was amended with 1 mg/ml glucose (LB-Glc).
Aliquots (0.5 ml) of the conditioning cultures in mid-exponential growth phase
were used to inoculate triplicate 50 ml cultures of specific test medium. All liquid
cultures were aerated via vigorous shaking (2200 rpm) at 37°C. To determine if
differences in growth rates were correlated with the accumulation of RNA,
concentrations of RNA were determined by a scaled-down orcinol assay (20) for
at least three replicate cultures of the host strain and transformants carrying
either pE2200 or pE2211 grown in each test medium. Since RNA concentrations
are proportionate to growth rate (11), comparisons of RNA concentrations were

made after normalizing the data to specific growth rate.

2.3 Results

Effects attributed to carriage of the pEZ plasmid backbone and promoter-
less rRNA operon were assessed by comparing the growth of pE2200
transformants with the host strain. There was no consistent difference between

the growth rates of transformants carrying the control plasmid pEZ200 and the

16

host strain (SE = 96%; Table 2.1). Similarly, the presence of pE2200 had little
effect on the concentration of RNA relative to the host strain on any medium (Y =

95%; Table 2.2).

Table 2.1 Specific growth rates (it) of the host strain and transformants carrying

pEZ200 or pE2211, grown in different media

 

Specific growth rate (u)a

Relative growth rateb

pEZ200/ pEZZ11/

 

 

Medium host strain pE2200 pE2211 host pEZ200
strain
Acetate 0.41 (0.01) 0.35 (0.02) 0.23 (0.01) 85 66
Acetate-AA 0.66 (0.01) 0.67 (0.04) 0.45 (0.00) 102 67
Glucose 1.09 (0.01) 0.99 (0.05) 0.66 (0.04) 91 67
Glucose-AA 1.19 (0.02) 1 .15 (0.05) 0.98 (0.07) 97 85
LB-Glucose 1.60 (0.01) 1.71 (0.18) 1.45 (0.31) 107 85

 

values are the sample

standard deviations are in parentheses.
b Relative growth rate is the ratio of specific growth rates depicted as % (i.e.

multiplied by 100); slow growth values are grouped by dark shading, fast growth

values are grouped by light shading.

17

mean of at least three replicate cultures, sample

Table 2.2 Concentration of RNA (fg/fl) of the host strain and transformants

carrying pEZ200 or pE2211, grown in different media.

 

Relative RNA conc.
Concentration of RNA (fg/fl)“ (normalized to p)b
pEZ200/
host pE2211/
Medium host strain pEZ200 pEZ211 strain pE2200

 

Acetate 26.8 (2.1) 25.0 (1.7) 33.3 (3.9) 109 ‘ 202
Acetate-AA 40.2 (4.9) 40.1 (3.4) 54.1 (8.7) 98 l 201

Glucose 56.2 (3.6) 44.8 (2.5) 74.0 (9.8) 88 1 248°
Glucose-AA 79.5 (8.1) 72.1 (5.8) 81.4 (5.8) 94 132
LB-Glucose 122.5 (2.8) 115.7 (5.6) 122.4 (3.4) 88 125

 

8‘ values represent the sample mean of at least three replicate cultures,

sample standard deviations are in parentheses; cell volumes (fl) of 5%
formalin-fixed cells were determined using a Coulter Counter Channelizer
(Coulter Electronics, Inc., Miami).

I” ratio of normalized (divided by u) RNA concentrations depicted as %;
shading is as described in Table 2.1.

c normalized RNA concentration of transformants carrying pE2211 was
218% that of the host strain, suggesting that the determination of RNA
concentration for cells containing pE2200 and growing on Glc, was
anomolously low.

There was however, a dramatic effect on both growth rate and cellular
RNA concentration when the rRNA promoters were present on the plasmid-borne
rRNA operon (pE2211). At slower growth rates (Ac, Ac-AA, and Glc), cells with
pE2211 grew an average of 33% slower than those with the control plasmid,
pEZ200 (Table 2.1). Along with slower growth rates, extra rRNA operons
resulted in normalized RNA concentrations, which were at least 100% higher
than the control (Table 2.2). Transformants carrying pE2211 not only had higher

concentrations of RNA per unit volume (fg/fl), but also more RNA per cell (Figure

2.2).

18

300

 

250 -

200 1

150-

RNA (fg/cell)

100'

50-

 

 

 

 

0 0.5 1.0 1.5 2.0
Specific Growth Rate, p(h'1)
Figure 2.2 Cellular RNA content (fg/cell) as a function of growth rate, (I
(h'1) for the evolved host strain (A), and transformants with pEZ200 (O) or

pEZ211 (0) compared with previously published values for E. coli B (A; (11)).
Error bars represent sample standard deviations.

19

The accumulation of RNA in slowly growing E. coli cells may be due to
constitutive expression of the rRNA operons from the P2 rRNA promoter (9, 18).
Since rRNA synthesis is linked to the expression of ribosomal proteins and other
genes, the overproduction of rRNA would have the effect of diverting limiting
resources to the production of unnecessary ribosomes, resulting in a decreased
growth rate (32, 34, 41). The additional RNA in transformants carrying pE2211
does not appear to be due to an increase in plasmid copy number because in
comparable experiments with a slightly different host strain and rRNA operon-
containing plasmid constructs, Baracchini and Bremer (8) reported a 41%
decrease in growth rates and only slight differences in the plasmid copy number
at various growth rates.

The effect of extra rRNA operons on both growth rate and cellular RNA
concentration was diminished when cells were grown in media which led to faster
growth rates (Glc-AA and LB-Glc). Transformants carrying pEZ211 grew 15%
slower than those with pE2200 (Table 2.1), and had an average of 29% more
RNA per unit volume when normalized to specific growth rate (Table 2.2). The
smaller difference between RNA concentrations of pE2200 and pE2211
transformants suggests a more effective means of compensation for the
additional rRNA operons at higher growth rates. The mechanism of
compensation is most likely transcriptional regulation of the operons (41).

Nomura and colleagues (41) made direct measurements of rRNA
synthesis rates with similar plasmid constructs at faster growth rates and

reported that cells were able to down regulate rRNA synthesis from all rRNA

20

operons within the cell as a result of feedback inhibition. The growth rate of
these populations was reduced by 10%. This reduction was attributed to an
imbalance in tRNA pools resulting from the increased expression of tRNAs
located on the plasmid-borne operon. The rRNA synthesis rate in these cells
was 19% higher than the control when normalized to growth rate (41), suggesting
that modest RNA overproduction may have also contributed to the reduced
growth rate.

A third plasmid (pE2211ASal) was used to investigate the possibility that
the observed changes in growth rate were due to the sequestration of RNA
polymerase by the numerous promoters of the rRNA operons. The non-
functional rRNA operon of pE2211ASaI was developed by digesting pE2211 with
Salt which resulted in the deletion of a majority of the structural RNA genes
(Figure 2.18). There was no consistent deviation in the doubling times of cells
expressing non-functional rRNA (pE2211ASaI transformants) relative to the
control (Table 2.3) suggesting that the presence of the promoters alone is not
sufficient to explain the changes in growth rate. Cells carrying pE2211ASaI were
elongated, frequently occurred in chains, and tended to form aggregates at
slower growth rates. Although the altered cell morphology had no obvious affect
on growth rate estimation, cell enumeration was ineffective via particle counting
or direct microscopy. As a result, the concentration of RNA could not be

determined for cells with pEZ211ASaI.

21

Table 2.3 Specific growth rate (it) for pE221 1ASal

transformants in different media.

 

Specific growth Relative growth

 

 

 

rate ((11)‘El rate
pEZ21 1ASaV
Medium pEZ21 1ASal pE2200
Acetate 0.38 (0.04) 108
Acetate-AA 0.60 (0.05) 88
Glucose 0.82 (0.03) 83
Glucose-AA 1 .14 (0.18) 99
LB-Glucose 1 .62 (0.15) 95

 

8'” see Table 2.1, footnotes a and b

2.4 Conclusions

The presence of extra rRNA operons in a population of E. coli resulted in
overproduction of RNA and decreased growth rates. These effects were
exaggerated at slower growth rates, suggesting that regulation of rRNA synthesis
was overwhelmed at slow growth rates. The metabolic expense associated with
ribosome overproduction when nutrient availability is low may be compensated
for by the capacity to rapidly shift up growth rate in response to an influx of
nutrients. The potential advantage of a rapid shift in growth rate may seldom be
realized in stable, low nutrient environments where the diversion of limited
resources towards the production of excess ribosomes would clearly be
disadvantageous. Understanding the advantages and disadvantages of multiple
rRNA operons in different environmental conditions should provide insight into

the selective pressures that influence rRNA copy number.

22

CHAPTER 3

REMOVAL OF RIBOSOMAL RNA OPERONS FROM THE ESCHERICHIA COLI
CHROMOSOME

3.1 Introduction

The Escherichia coli chromosome has seven ribosomal RNA (rRNA)
operons (45). The rRNA operons are located on the half of the chromosome
closest to the origin of replication, which can further increase the number of rRNA
operon copies in rapidly growing cells containing multiple replication forks (Figure
3.1, (7, 22)). Transcription of the rRNA operons is in the same direction as DNA
replication, reducing the conflict between these two processes on the same
region of the chromosome (11). All of the rRNA operons in E. coli (rmA, B, C, D,
E, G, and H) contain upstream activation sequences, strong tandem promoters,
structural RNA genes (168, 238, and 58), and some tRNA genes (Figure 3.2A,

reviewed in (32)).

23

 

Figure 3.1 Location (in min) of the seven rRNA operons (rm) and the
origin of replication (oriC) on the E. coli K-12 chromosome (22). Arrows indicate
the direction of rRNA operon transcription.

24

A

 

Multiple lntemally Transcribed Distal
Promoters Spacer Spacer
P1 P2 |__( H
rrs tRNA(s) m m‘
168 rRNA 23S rRNA SS rRNA
B rm Operon ITS tRNA(s) Distal tRNA(s)
B, E, G Gluz -
C Gluz Asp1, Trp
A "81, Ala1 B -
D "81, Ala13 Thr1
H "81, Ala1B Asp1

 

Figure 3.2 The (A) general organization (19), and (B) tRNA genes of the
seven E. coli rRNA operons (61,62).

25

Sequence composition and organization of the seven rRNA operons is
very similar yet, some differences do exist (7). Each rRNA operon has either the
tFtNAG'“2 gene (rmB, C, E, and G), or the tRNA"°1 and tRNAA'am genes (rrnA, D,
and H) in the internally transcribed spacer region (ITS) between the 168 and 233
genes, and a few also contain distally located tRNA genes (Figure 3.2B). The
similarity between the rRNA operons on the same chromosome creates potential
sites for homologous recombination, which could explain the maintenance of the
highly conserved regions of each operon via gene conversion, and could also
play a major role in genomic rearrangements (reviewed in (40)). The subtle
sequence differences among the rRNA operons could also lead to differential
expression of the operons or functional differences of their products (18).

Previous studies of rRNA operons and their function have been based on
chromosomally located rRNA operons inactivated by either the insertion of
marker genes (16-18), or deletion of part or all of an rRNA operon (3, 6, 23).
These constructions are unsuitable for directly determining the specific effects of
rRNA operon deletion because they result in the expression of additional marker
genes, synthesis of truncated transcripts from intact rRNA promoters, or the
deletion of adjacent genes. In contrast, the rRNA operon deletions reported here
were designed to enable the direct determination of the ecological and
physiological effects of rRNA operon copy number in E. coli. These deletions
removed the majority of two E. coli rRNA operons (rrnA and rmB) including their

promoters, resulting in transcriptionally “silent” deletions.

26

3.2 Materials and Methods
Bacterial strains. plasmids, and phage

All bacterial strains and all plasmids and phage used or created in this
study are listed in Tables 3.1 and 3.2. These tables list the names, relevant
genotype or phenotypes, and the source for each entry. The rRNA operons were
deleted from two isogenic strains of E. coli (REL 4548, and BS 4548) that come
from the same line, which has evolved for 10,000 generations in glucose minimal
medium batch cultures (56). The two strains differ only by a single point
mutation, which is neutral and renders BS 4548 capable of utilizing the sugar
arabinose (Ara*), while the parental strain REL 4548 cannot (Ara'). Both strains

represent a well-studied line of E. coli with a stable genetic background.

27

Table 3.1 Bacterial strains

 

Strains Relevant genotype or Source
phenotype

TOP10 Lac-, recA, endA1 lnvitrogen, chemically competent cells

REL 4548 Ara' R.E. Lenski (56)

BS 4548 Ara+ spontaneous Ara‘ mutant of REL 4548

D308 recD F.W. Dahlquist (66)

DBS 222 rmB, Sues, Kanr D308 linear transformant of pR8222,
UpB:sacB-neoanB fragment

DAS 113 rrnA‘, Sucs, Kanr D308 linear transformant of pRA113,
UpA:sacB-neoanA fragment

BS 211 Ara”, rmB, SUCS, KanIr P1 transduction of BS 4548 with 088
222 Iysate

BS 210 Ara', rmB, SUCS, Kanr P1 transduction of REL 4548 with 088
222 Iysate

BS 111 Ara‘“, rrnA', Sucs, Kanr P1 transduction of BS 4548 with DAS
113 Iysate

BS 110 Ara', rrnA', Sues, Kan’ P1 transduction of REL 4548 with DAS
113 Iysate

BS 31 1 Arafl rrnA', Arr/)8, SUCS, P1 transduction Of 33 111 With DAS 113

Kan' Iysate
BS 310 Ara', rrnA', ArrnB, Sucs, P1 transduction of BS 110 with DAS 113
Kant Iysate

BS 201 (ArrnB) Ara+, ArrnB allelic exchange of UpB:sacB-neo:DnB in
BS 211 with UszDnB from pKTS203

BS 200 (ArmB) Ara‘, ArmB allelic exchange of UpB:sacB-neo:DnB in
BS 210 with UszDnB from pKT8203

BS 101 (ArmA) Ara*, ArmA allelic exchange of UpA:sacB-neo:DnA in
BS 111 with UpA:DnA from pKTS102

BS 100 (ArrnA) Ara', ArrnA allelic exchange of UpA:sacB-neo:DnA in

BS 301 (ArmAB) Ara“, ArrnA, ArmB
BS 300 (ArmAB) Ara‘, ArmA, ArmB

28

BS 110 with UpA:DnA from pKTS102
allelic exchange of UpA:sacB-neoanA in
BS 311 with UpA:DnA from pKT $102
allelic exchange of UpA:sacB-neoanA in
REBS 310 with UpA:DnA from

pKTS102

 

Table 3.2 Plasmids and Phage

 

Plasmids Relevant genotype or Source
phenotype
pUC19 Amp', lacZ’ Pharmacia
pBR322 Amp’, Tet', Pharmacia
pCR 2.1-TOPO Amp', Kanr lnvitrogen
pKK3535 Amp', rrnB operon J. Brosius
pC1 Amp', rrnA operon C.L. Squires (18)
pl8279 Amp’, sacB-neo cassette I.C. Blomfield (10)
pMAK705 Cam', rep‘s, lacZ C.M. Hamilton (36)
pUPB1 Amp', UpB upstream-rmB(UpB) fragment (Ndel
and BamHI, 817bp) cloned into
pUC19
pUBD4 Amp', UpB:DnB downstream-rmB(DnB) fragment
(BamHI and Xbal, 991 bp) cloned
into pUPB1
pR3222 Amp', UpB:sacB-neo sacB-neo cassette (BamI-ll, 3,825 bp)
cassetteanB from plB279 cloned into pUBD4
pKTSZOS Cam', rep”, UpB:DnB UpB:DnB fragment from pCR-UBD
(Hindlll and Xbal) cloned into
pMAK705
pCR-UBD Amp', Kan', UpB:DnB Ligation of pCR 2.1-TOPO with
UpB:DnB fragment (from pUBD4
plasmid DNA)
pCR-UpA Amp', Kan', UpA Ligation of pCR 2.1-TOPO with
upstream-rmA (UpA) fragment
(from REL 4548 genomic DNA)
pCR-DnA Amp', Kan', DnA Ligation of pCR 2.1-TOPO with
downstream-rrnA (DnA) fragment
(from pC1 plasmid DNA)
pUPA2 Amp', UpA UpA fragment (Ndel and Kpnl,
1667bp) from pCR-UpA cloned into
pUC19
pUADI 1 Amp', UpA:DnA DnA fragment (Kpnl and Xbal, 941
bp) from pCR-DnA cloned into
pUPA2
pRA1 13 Amp’, Kan’, sucroses, saclBtnedO 0:82;“? (Bézmlrl. 3.825 tfipl
. - c one in o n res ric ion site 0
Eggjgéggio pUAD11 via blunt-ended ligation
pCR-UAD Amp', Kan', UpA:DnA Ligation of pCR 2.1-TOPO with
UpA:DnA fragment (from pUAD11)
pKTS102 Cam', rep‘s, UpA:DnA UpA:DnA (Xbal, 2,609 bp) fragment
from pCR-UAD cloned into
pMAK705
Phgqg P1 vir Vir' L. Snyder (72)

 

29

Bacterial growth conditions. medfi

Liquid broth cultures were grown in 16 x 100 mm culture tubes, shaking at
225 rpm and 37°C unless otherwise specified. The liquid medium used was
Luria-Bertani (LB) broth (67), sometimes containing one or more of the following
antibiotics [Amp, ampicillin (100 mg/ml); Kan, kanamycin (50 mg/ml); or Cam,
chloramphenicol (30 mg/ml)]. Solid agar medium most commonly used was LB
with 1.5% (w/v) Bacto agar (Difco Laboratories; Detroit, MI) added. Antibiotics
were present in the concentrations listed above when specified. The other solid
agar medium used (LBSUC) was LB agar lacking NaCl but containing 6% (w/v)
sucrose. The soft agar medium used in the generalized transduction procedure

below was LB containing 0.75% (w/v) Bacto agar.

General cloning procedures

Plasmid and genomic DNA were digested with restriction endonucleases
according to manufacturer protocols (NEB; New England Biolabs lnc.; Beverly
MA). DNA fragments with either blunt ends or cohesive ends were ligated using
the Prime Efficiency blunt-ended DNA ligation kit (Eppendorf-5 Prime lnc.;
Boulder, CO). Plasmid and genomic DNA were harvested from cells using the
Wizard SN Plasmid DNA Purification Kit (Promega Corp.; Madison, WI), and the
Tissue DNA Purification Kit (Qiagen lnc.; Valencia, CA), respectively. When
necessary, electrophoresis in agarose gels (0.6 — 1.0% agarose in 0.5x TBE)
was used to isolate DNA fragments after restriction digestion. Once isolated,
DNA fragments of the correct size were cut out of the agarose gel, and DNA was

recovered using the Qiaex ll DNA recovery kit (Qiagen Inc.).

30

DNA was transformed into either One Shot TOP10 chemically competent
E. coli cells (lnvitrogen; Calsbad, CA) according to manufacturer’s protocols, or
cells of other E. coli strains made chemically competent using a protocol found in
the Transformer Site-Directed Mutagenesis Kit product protocol (PT1130-1;
Clonetech Laboratories; Palo Alto, CA).

Generalized transduction was carried out with a virulent mutant of phage
P1 (P1vir). The protocol used varied from most common ones (59, 64) in that the
lysate was prepared on LB agar plates with a LB soft agar overlay (Appendix A)

to obtain a higher titer (usually 109 to 1010 PFU/ml).

PCR and XL PCR amplification

DNA fragments under 4 kb were PCR amplified using Biolase DNA
polymerase (ISC; lntermountain Scientific Corporation; Kaysville, UT) in the
following reaction conditions [1x reaction buffer (ISC), 2 mM MgCl2, 2.5 mM each
dNTP, ~200 ng template DNA, 7.5 uM each primer, 25 or 100 pl final reaction
volume]. A Perkin Elmer GeneAmp PCR System 9600 thermal cycler (PE
Biosystems; Norwalk, CT) was used for these reactions with the following
temperature regime: 95°C for 3 min, 30 cycles of (94°C for 30 sec, 60°C for 30
sec, and 72°C for 45 sec), 72°C for 10 min, and hold at 4°C indefinitely. When
potentially larger DNA fragments (>4 kb) were to be PCR amplified, the
GeneAmp XL PCR kit (PE Biosystems) was used, which utilizes the proofreading
Tfh DNA polymerase. Reactions were set up according to the manufacturer’s

protocols except that total reaction volumes of 25 ul were sometimes used, and

31

colony material was occasionally used as the template. The temperature regime

used was the same as above except that each cycle was kept at 72°C for 10 min.

Visualization of rRN_A operons using southern hybridization

 

Southern hybridization was used to visualize rRNA operons of each
strain of E. coli (Appendix B). Genomic DNA was digested with Pvull, separated
according to size via electrophoresis in agarose gels (0.6% agarose in 0.5x TBE
buffer), and transferred to Magnacharge nylon transfer membranes (Micron
Separations Inc.; Westborough, MA) using the VacuGene XL Vacuum Blotting
System (Amersham Pharmacia Biotech; Piscataway, NJ). After cross-linking the
DNA to the membrane with the Stratlinker UV Crosslinker (Stratagene; La Jolla,
CA), a digoxigenen-dUTP labeled probe (DIG High Prime labeling and detection
kit; Boehringer Mannheim Corporation (BMB); Indianapolis, IN) specific for a
conserved region (positions 8-356) of the E. coli16$ rDNA gene, was hybridized
to the immobilized fragments of genomic DNA containing an rRNA operon. After
hybridization, the membrane was washed and hybridized probe was detected
according to the manufacturer’s protocol (BMB). Each rRNA operon was
visualized by the hybridized probe and identified by its unique fragment size.
Genomic DNA from cells that contained modified rRNA operons or rRNA operons
replaced by marker alleles or deletions, either showed altered fragment sizes or

no corresponding fragment.

32

Deletion of the rmB operon

The first step in the procedure used for deleting the rmB operon (outlined
in Figure 3.3) from the genomes of REL 4548 (Ara') and BS 4548 (Ara") was to
construct a marker allele for the rmB operon. Two fragments of sequence (UpB
and DnB, Figure 3.4) that flanked the rmB operon were amplified from the REL
4548 genome using the UpB and DnB primers in Figure 3.5. The UpB fragment
contained sequence only found at one site on the chromosome. The
downstream (DnB) fragment included the 3’ end of the rmB operon, most of
which was conserved among the other rRNA operons. The UpB fragment was
cloned into pUC19 (pUPB1, Table 3.1), and then the DnB fragment was cloned
into pUPB1 adjacent to UpB, to form pUBD4 (Figure 3.3). This plasmid was
used in creating the marker allele and served as the source of the deletion allele
(UpB and DnB) used in the final step of the rmB deletion, discussed below.

A marker cassette containing both selectable and counter-selectable
markers (sacB:neo) from pl8279 (10), was cloned in between UpB and DnB on
pUBD4 creating the rmB-specific marker allele UpB:sacB:neo:DnB on the
plasmid pRB222 (Figure 3.3). The selectable marker gene (neo) encodes for
resistance to neomycin and kanamycin, which was used as the selective agent.
The counter-selectable marker gene (sacB) encodes for the exoenzyme
levansucrase in Bacillus subtilis (30). Expression of this gene in E. coli and other
Gram-negative prokaryotes is lethal when the cell is grown in the presence of

sucrose (29).

33

Intermediate strains of REL 4548 and BS 4548 were then constructed (BS
210 and BS 211, respectively; Figure 3.3), in which the marker allele replaced
the rrnB operon. First, a linear fragment of DNA containing the marker allele was
used to transform strain D308, a recD mutant strain of E. coli (66). The resulting
transformant (088 222) was kanamycin resistant (Kan’) and sucrose sensitive
(Sues) due to the marker allele that had replaced the wild type rrnB operon.
Generalized transduction (59, 64) with phage P1 vir was then used to move the
marker allele into both REL 4548 and BS 4548, creating the intermediate strains
of each neutral marker type (Ara' and Ara“, respectively).

The plasmid pKT S203 was constructed by cloning the UpB:DnB deletion
allele from pUBD4 into the temperature sensitive plasmid pMAK705 (36) (Figure
3.3). The intermediate strains were transformed with pKTS203 at the permissive
temperature (30°C). Transformant colonies were grown in LBKanCam at the non-
permissive temperature (44°C), selecting for cells with the plasmid integrated
into the chromosome at the marker allele via recombination between the
homologous UpB or DnB sequences. Transferring these cultures to fresh LB
without antibiotics at 44°C selected for cells in which a second recombination
event occurred, liberating the plasmid from the chromosome. If the second
recombination event occurred between the region of homology opposite from
where integration occurred (UpB or DnB), the deletion allele was exchanged with
the marker allele, and the resulting free plasmid contained the marker allele
(shown in Figure 3.3, (as discussed in 10, 36)). Sucr cells were isolated on LBSUC

agar; these cells had exchanged the marker allele with the deletion allele on the

34

plasmid, subsequently lost the plasmid, and therefore contained no markers
(ArmB in Figure 3.3). Screening of these isolates for the Sucr Kans Cams
phenotype, XL PCR amplification of the deletion region, and southern
hybridization of the remaining rRNA operons confirmed the deletion of the rmB

operon.

35

Figure 3.3 Steps used to delete the rmB operon from the REL 4548

and BS 4548 genome, creating the BS 200 and BS 201 (ArrnB) strains .

36

Isolate and

purify
sacB:neo

genome

 

5298376

0

s \a

P1 transduction

””648” MW") fl

 

II?)
o
0303);, 04/4

 

    
 

BS 210 or BS 21
D308 (intermediate strain)

genome 088 222 genome
genome

 

   
       

Transformation,
plasmid integration,
allelic exchange

 

 

   

OQQ 000
2 pKT8203 BS 200 or BS 201
7 (AnnB) genome
Cam repTS
Cam rep"-S
Figure 3.3

37

Deletion of the rrnA operon

The rrnA operon was deleted from the genome of REL 4548, BS 4548,
and both ArrnB strains in the same manner in which the rrnB operon was deleted
above. Regions upstream (UpA) and downstream (DnA) of the rrnA operon
(Figure 3.4) were amplified using rrnA-specific primers (Figure 3.5), and cloned
adjacent to one another on pUC19 to form pUAD11. A marker allele
(UpA:sachneoanA) was constructed as above, on pRA113, and its linear DNA
fragment was used to transform D308 E. coli cells. These transformants (DAS
113) served as the donor strain to transduce the marker allele into REL 4548, BS
4548, and both ArrnB strains, forming intermediate strains. The UpA:DnA
deletion allele from pCR-UAD (Table 3.1) was cloned into the pMAK705 plasmid
backbone, forming pKTS102. This plasmid was used to transform the
intermediate strains at the permissive temperature (30°C). After growth in LB
with, and then without antibiotics at 44°C, cells that had exchanged the marker
allele for the deletion allele and lost the resulting plasmid were isolated on LBS,uc
agar. These cells were screened for the Sucr Kans Cams phenotype, the rrnA
and rmB regions were amplified, the remaining rRNA operons were visualized
through southern hybridization to confirm the rRNA operon deletions. The strains

that contained the rrnA deletion were designated ArrnA and the strain with both

the rrnA and rmB deletions was designated ArrnAB.

38

 

 

 

 

Figure 3.4 Regions of E. coli chromosome containing the MA and rmB
operons, showing the amplified flanking regions (UpA, DnA, UpB, and DnB;
white boxes), structural rRNA genes (168, 238, and 58;dark grey boxes), tRNA
genes (tRNAI'e, tRNAA'a, and tRNAG u; dark grey boxes), and adjacent genes
(trkH, hemG, mobB, murl, murB; light grey boxes) are shown.

39

5' 3.

UpB forward 'CCG AAT TAC ATA TGiA CCG TGC TGG TGT TTG AC

UpB reverse :GAG GGA chl GCA ACA TTC AAC CAA ATC A

DnB forward ECGA GGA TCCI CCA TCG CTC AAC GGA TAA AA

DnB reverse iCTA CTC TAG ACC TGA TGC AAA AAC GAG GCT AGT TTA

----------------

UpAforward :GCG AAT TAC ATA TGC CCT CAC GCC ATC CTC TTT TAT

(Kpnl
UpA reverse :A’Aé'c'TA'cdc TGA CCG CGC ATT TTT TAT TCT

ijnI
DnAforward iAAé'ETA'béIC CAT CGC TCA ACG GAT AAA A

DnAreverse iCTA CTC TAG AiCG TCG CAT CCG GCA TTT TTT T

Figure 3.5 Oligonucleotide primers used throughout rRNA operon
deletion procedure. The non-target specific tails at the 5' end of each primer are
shown in boxes, the recognition sequences of restriction endonucleases are
labeled and shown in the shaded area of each box, and the arrows indicate
restriction sites.

40

3.3 Results

The deletion of the rmB and rrnA operons outlined above, relied on the
use of homologous sequence which flanked the regions being deleted (Figure
3.4). In the case of rrnB, the region upstream of the deletion (UpB) included the
3’ end of the gene murl and all sequence directly upstream of the rrnB promoters
P1 and P2. The region of homologous sequence upstream of rrnA (UpA)
included most of the trkH gene from the 3’ end, all of the hemG gene, and ended
directly upstream of the rrnA P1 and P2 promoters. The DNA sequence in these
upstream regions is found nowhere else on the chromosome. The downstream
regions used for rrnB (DnB) and rrnA (DnA) included the 3’ end of each 238
gene, the entire 58 gene, and ended just upstream of the murB gene in the case
of rmB, and upstream of the mobB gene in the rrnA deletion. The majority of the
sequence in the downstream regions includes sequences from the 23S and 58
rRNA genes, which are conserved among all seven rRNA operons.

The majority of the rrnB and rrnA operons, including the promoter
sequences were deleted. Upon deletion, the sequence remaining at the rmB site
included only the UpB and DnB regions adjacent to each other. The intervening
sequence that was removed included the rrnB promoters P1 and P2, the entire
168 rRNA gene, the ITS region including the tRNAG'“2 gene, and the 5’ end of the
238 gene. The deletion of rmB from the chromosomes of both REL 4548 and
BS 4548 (Ara‘ and Ara", respectively) resulted in the strains designated as B8

200 and BS 201 (ArrnB for short). Southern hybridization showed that the

genomic DNA from the ArrnB strain lacks the rmB-specific fragment seen in the

41

control strain REL 4548 (Figure 3.6). The rmB operon deletion was further
verified by amplifying the rmB region of the chromosome using genomic DNA
from ArrnB, and the primers UpB forward and DnB reverse (Figure 3.5) in an XL
PCR reaction. The size of the resulting fragment (1.9 kb, lanes 8-16, Figure 3.7)
corresponded to the size of the deletion allele amplified from pUBD4 (Iane7), and
was easily distinguishable from the wild type rrnB operon (lane 5, 6.1 kb) or the
marker allele (lane 6, 5.4 kb). By removing the rrnB promoters, no transcription of
the remaining rrnB sequence should occur.

The deletion of rrnA from the chromosomes of REL 4548, BS 4548, and
the ArmB strains resulted the strains designated as BS 100 and BS 101 (ArmA
for short) and BS 300 and 88301 (ArrnAB for short). The sequence that was
deleted from rrnA included the rrnA promoters P1 and P2, the entire 168 gene,
the ITS region with the tRNA“°1 and tRNA/”813 genes, and most of the 238 gene
at the 5’ end. Genomic DNA from ArrnA and ArmAB lacks the rrnA-specific
fragment in the southern hybridization (Figure 3.6). As with the ArmB strains
above, the deletion of rrnA was verified in ArrnA and ArmAB by amplifying the
rrnA region of the chromosome with the primers UpA forward and DnA reverse
(Figure 3.5). The rrnA regions amplified from ArmA (2.6 kb, lane 4, Figure 3.8)
and ArmAB (2.6 kb, lanes 5-9) genomic DNA are easily distinguished from the
wild type rrnA operon (7.4 kb, lane 2) and the rrnA-specific marker allele (6.4 kb,
lane 3). Also, the rrnB regions amplified from ArrnAB as above (1.9 kb, lanes 13-

17) are easily distinguished as deletions compared to the wild type rrnB operon

42

(6.1 kb, lane 11). As with the deletion of rrnB, the rrnA deletions left no

promoters or additional marker genes, and did not affect adjacent genes.

43

Control
ArrnB
ArrnA
ArrnAB
Marker

' — 23.1

E \.'““fl- -9.4
3’“ out

H-Uflfl-

«4 - 2.3

Figure 3.6 Southern hybridiztion of genomic DNA of each strain digested
with Pqu and hybridized with a 168 DIG-labelled probe. Hybridized bands
corresponding to each of the seven rRNA operons in the control strain are
labeled on the left. Sizes (kb) of each band of the it HindIII DNA marker are
labeled on the right.

44

12 3 4 5 6 7 8 910111213141516171819

 

Figure 3.7 XL PCR amplification of the wild type nnB operon (lane 5, 6.1
kb), rmB-specific marker allele (lane6, 5.4 kb), plasmid-borne UpB:DnB deletion
fragment (lane 7, 1.8 kb), and chromosomal rmB deletions (lanes 9-16, 1.8 kb).
Fragments sizes of a 1 kb DNA ladder in lanes 1 and 19 are marked to the left of
lane 1.

45

12 34 5 6 7 8 9101112131415161718

10.0
8.0

6.0
5.0

4.0

3.0

2.5

2.0

 

Figure 3.8 XL PCR amplification of wild type nnA (lane 2) and rmB (lane
11) operons, intermediate rmA-specific marker allele (lane 3), rmA deletion in B8
101 (lane 4), rrnB deletion from BS 201 (lane 12), and both rrnA and rmB
deletions from BS 301 strain isolates (lanes 5-9 and 13-17 respectively).
Fragment sizes of a 1 kb DNA ladder (lanes 1, 10, and 18) are marked to the left
of lane 1.

46

3.4 Discussion
Deletion of rRN_Loperons

This study outlines the genetic techniques needed for the deletion of rRNA
operons. The difficulty in manipulating genes that are present in multiple copies
on the same chromosome lies in the cell’s ability to repair these genes through
gene conversion. It was essential that regions of sequence were used, which
unlike the rRNA operons were present in a single copy on the chromosome to
ensure that recombination occurred with the target operon and not with any of
the other rRNA operons.

Previous studies deleting rRNA operons have used inserted marker genes
(3, 18), deleted only part of the rRNA operon leaving the rRNA promoters intact
(3, 6, 18), or altered adjacent genes in the deletion process (23). The cells from
these constructions express additional marker phenotypes, produce truncated
rRNA transcripts from the still-active rRNA promoters, or have adjacent genes
that are also altered, all which may have affects not specifically attributable to the
deletion of rRNA operons. The rRNA operon deletions in this study do not
involve the addition of marker genes, lack active promoters, and adjacent genes
are not affected, making the resulting strains useful for directly determining the
specific fitness or physiological effects of deleting rRNA operons from the
chromosome.

Care was taken in choosing which rRNA operons were deleted to
minimize the perturbation to the tRNA gene pool. rrnA and rmB each represent

one of the two groups of E. coli rRNA operons with respect to the tRNA genes

47

located in their ITS regions. The ITS regions of rrnA, D, and H operons have the
tRNA"°1 and tRNAAIam genes, whereas the ITS regions of the rmB, C, E, and G
operons have the tRNAGqu gene. Unlike the rrnC, D, and H operons on the E.
coli chromosome, neither rrnA nor rmB has tRNA genes located distal to the
operon. The rrnA and rrnB operons are also located close to each other on the
chromosome, which should reduce the difference in gene dosage when the cell
is dividing faster than DNA replication and has multiple copies of the
chromosome within the cell (11).

The procedure used to delete the rmB and rrnA operons was complex but
precise, with strong selection provided for each step. An essential step in the
deletion of each rRNA operon was the construction of an intermediate strain in
which a marker allele has replaced the wild type rRNA operon that is to be
deleted. To get the marker allele to replace the target rRNA operon on the
chromosome, 8 recD strain of E. coli (D308) was transformed with linear DNA
(66). Linear DNA transformants were selected for the Sucs and Kan'
phenotypes. These cells had the wild type rRNA operon replaced by the marker
allele through recombination with the homologous sequence flanking the marker
genes. The upstream homologous sequence allowed for rRNA-specific
recombination and prevented the cell from repairing the mutation with one of the
other rRNA operons (discussed below). The region of sequence used upstream
of the rRNA operon promoters was not part of the operon and therefore was
unique to that location on the chromosome. Generalized transduction with phage

P1 vir was used to move the marker allele from the chromosome of transformed

48

recD cells into a strain of E. call that has evolved in glucose minimal medium for
10,000 generations (56), thereby producing intermediate strains. The potential
limitation caused by the P1 transduction of DNA from the recD cells (derived from
E. coli strain K-12, (66)) into the evolved strains of E. coli (derived from E. coli
strain B, (51, 56)) is discussed below.

The intermediate strains were transformed with a temperature sensitive
plasmid containing the deletion allele, which consisted of the rRNA operon-
specific upstream and downstream regions. Growing these transformants at
non-permissive temperatures with, and than without antibiotics promoted an
exchange of the deletion allele with the marker allele between the plasmid and
the chromosome. The counter-selectable marker sacB simplified the isolation of
cells that had undergone an allelic exchange and lost the resulting plasmid,
therefore containing a deleted rRNA operon and no markers. It was possible
however, to find cells that apparently had a mutation in the sacB gene and were
able to grow on LBS,uc agar. Further screening for KarlS allowed for the
identification of cells that had indeed undergone the allelic exchange among
those that still had the marker allele (Kan') but could somehow grow on LBs,uc
agar. The use of markers that provide strong selection along with the
temperature sensitive plasmid was essential in facilitating the allelic exchange

between the marker allele and the deletion fragment.

49

Conclusion

Further studies of the strains with ArmB, ArmA, and ArmAB have shown
that these deletions are stable for many generations. These strains have
differences in several growth parameters as will be discussed in another chapter,
but overall they are robust and grow well under all tested conditions. This alone
is evidence that neither rmB, rrnA, nor rmB and A together are essential for the

growth of E. coli.

3.5 Lessons learned during construction of rRNA deletion strains
Generalized transduction between strains of E. coli

Generalized transduction was used to move the marker alleles from the
constructed recD strains to the intermediate strains. The recD strains were
derived from an E. coli K-12 strain (66), whereas the REL 4548 and BS 4548
strains and their derivatives are from an E. coli 8 strain (51, 56). Furthermore,
the REL 4548 and BS 4548 strains have been evolved for 10,000 generations
under specific conditions (56). These two different strains of E. coli are relatively
closely related within one subgroup of E. coli strains based on a study of
sequence divergence of 20 enzyme-encoding loci (38). There are some known
regions of sequence differences between their genomes as well as in their
restriction modification systems (71 ). When phage P1 is used in transduction,
between 91 and 100 kb of donor DNA typically enters the recipient cell (64).
Only one known region of intraspecific sequence divergence (the phn operon) is

located anywhere near 100 kb upstream or downstream of the rrnA and rmB

50

operons. Based on the location of the known regions of high intraspecific
sequence divergence, the region of the chromosome that contains the rrnC, A, B
and E operons appears to be well conserved among different strains of E. coli
(58).

The fact that genomic DNA from E. coli K-12 was used to transduce this
evolved line of E. coli B raises the question as to what other donor DNA could
have been incorporated into the recipient 8 strain of E. coli along with the rRNA
operon marker alleles. The effect of incorporating this additional donor DNA on
the ecological and physiological properties measured in these rRNA operon
deletion strains must be considered. In order fully to address this question, a
control strain must be “reconstructed” from the same intermediate strains that
were used to delete the rRNA operons. Transforming these intermediate strains
with a temperature sensitive plasmid that contains the full rRNA operon and
isolating those cells that have replaced the marker allele with the full operon
could accomplish this. Only these reconstructed controls can be considered to
be othenrvise isogenic to the rRNA operon deletion strains, and are currently

under construction.

Gene conversion

Gene conversion apparently took place in an earlier attempt to delete
rrnB, where a marker allele was constructed by inserting the sacB and neo
marker genes into the 168 rDNA gene of a cloned rmB operon. The resulting

construction had the marker genes flanked by 168 rDNA sequence (704 bp

51

upstream and 837 bp downstream) that were conserved among all the other
rRNA operons. A temperature sensitive plasmid carrying the marker allele was
used to promote an exchange of the marker allele with the wild type rmB operon.
Once the marker allele was present on the chromosome, any of the other six
rRNA operons were available to repair this mutation through a non-reciprocal
recombination event. The occurrence of this event was only caught by virtue of
which rRNA operon was used to repair the mutation. If rmG, E, or Cwere used,
the repaired operon did not appear any different from the wild type rrnB operon
when visualized by southern hybridization (Figure 3.8). If however, the repair of
the marker rrnB allele was facilitated by recombination with rrnA, D, or H, the
repaired operon resulted in a band that was a different size and did not
correspond with the size of either the wild type rmB operon or the rmB marker
allele.

The difference in size of these repaired operons appeared because they
were hybrids with regards to their ITS region. An extra Pvull restriction site in the
ITS regions of rrnA, D, and H resulted in a fragment of unique size. The hybrid
fragments were smaller than they would be if they had been repaired back to
rmB-like sequence through recombination with rmG, E, or C, but larger than the
marker allele (Figure 3.9). This self-repair phenomenon illustrates the
importance of flanking homologous sequence in permitting gene conversion

among the rRNA operons.

52

 

Figure 3.9 Southern hybridization of a repaired chromosomal rmB operon.
Hybridized DNA fragments that correspond with the wild type rmB operon (1), the
marker rmB allele (2), and the hybrid rmB operon (3) are labeled with arrows.

53

 

Marker gene orientation

Another lesson that was learned in earlier attempts to delete the rmB
operon is that the orientation of a marker gene has a great impact on the marker
phenotype when it is downstream of the rRNA promoters. When marker genes
were inserted into the 168 gene of a cloned rmB operon (discussed above),
markers were only weakly expressed if their orientation was opposite of rRNA
transcription. In constructions where the markers were in the same orientation of
the rRNA operon, or when the rRNA promoters P1 and P2 were not present,

expression of the marker phenotypes was robust (data not shown).

Temperature sensitive plasmid

The temperature sensitive replication of the plasmid used for allelic
exchange between the marker allele and the rRNA deletion fragment enabled the
selection of cells that had integrated the plasmid into the chromosome, but it was
still possible to isolate cells that contained free plasmid. The temperature
sensitivity of the plasmid pMAK705 and its derivatives was therefore not absolute
when grown at non-permissive temperatures. This plasmid was still a good
vehicle for allelic exchange when the proper selectable and counter-selectable

markers were used, as noted by Blomfield et al (10).

54

CHAPTER 4

EFFECT OF DELETING RIBOSOMAL RNA OPERONS FROM THE

ESCHERICHIA COLI CHROMOSOME

4.1 Introduction

The research in this chapter investigates the potential connection between
the number of rRNA operons on the Escherichia coli chromosome and its life
history. As discussed in Chapter 1, the number of rRNA operons on the
chromosome of a prokaryotic species may be an adaptation to the environmental
conditions in which that species has evolved, and a component of the life history
of that species.

The first hypothesis that was tested states that multiple rRNA operons are
an advantage in environments defined by fluctuations in nutrient availability.
Cells with a higher number of rRNA operons could have a higher capacity for de
novo ribosome synthesis upon an influx of nutrients, or excess protein
synthesizing capacity prior to an influx of nutrients. The rate-limiting step in the
production of ribosomes is the synthesis of rRNA, which is controlled at the level
of transcription. Furthermore, the rRNAs are the end product of the rRNA genes.
Multiple rRNA operons are an obvious way for a cell to increase rRNA
synthesizing capacity.

The second hypothesis that was tested represents the potential tradeoff
for having multiple rRNA operons, stating that fewer rRNA operons are an

advantage in environments defined by the stable and limiting availability of

55

nutrients. E. coli cells growing very slowly (u s 0.2 h") contain excess ribosomes
believed to be the result of constitutive expression of the seven rRNA operons
(47). If the number of excess ribosomes were proportional to the number of
rRNA operons, than cells with fewer rRNA operons would be expected to have
fewer excess ribosomes. The expression of excess rRNA would represent a
metabolic expense to cells in stable, slow growth conditions. If the expression of
excess rRNA resulted in excess ribosomes, the metabolic expense would be
compounded by the cost of synthesizing the rest of the protein synthesizing
machinery, which is regulated with the transcription of rRNA. The diversion of
limited resources towards the production of excess ribosomes would be a
disadvantage to a cell if that cell did not encounter the influx in nutrients needed
to utilize the higher protein synthesizing capacity.

Previous work attempting to study the relationship between rRNA operon
copy number and cell physiology was based upon strains of E. coli with altered
numbers of rRNA operons. The rRNA operon copy number was altered in these
strains through the addition of plasmid-borne operons ((33, 74), Chapter 2) or by
the inactivation (18) or deletion (3, 5, 23) of operons on the chromosome. The
studies using plasmid-borne rRNA operons were limited by unknown or varying
plasmid copy number and could only add rRNA operons to the genome of the
cells. The previous constructions, in which rRNA operons on the chromosome
were either inactivated or deleted, were also unsuitable for the experiments
outlined in this study. It was not possible to discern the effect of rRNA operon

manipulations in these previous constructions from the effects of deleting

56

adjacent genes (23), the expression of truncated rRNA transcripts from intact
rRNA promoters (3, 5), or the additional expression of marker genes (18). New
strains in which the rRNA operons rrnA and rmB were each or both deleted were
constructed with these requirements in mind (as discussed in Chapter 3) and
used in the experiments described below. The deletions included the rRNA
promoters and a large portion of the operon, but not adjacent genes on the
chromosome. These rRNA operon deletion strains were designed to be more
suitable than the previous rRNA operon deletions in determining the direct effect
of rRNA operon deletion on physiological traits and competitive ability of E. coli.
To test the hypotheses discussed above, strains of E. coli containing
either 6 or 5 rRNA operons (ArmA, ArmB, or ArmAB) were used in mixed culture
competition experiments with a wild type control (7 rRNA operons). These
culture conditions resulted in direct competition for a shared nutrient with
fluctuating or constant availability, and gave a direct measurement of the
competitive ability of each strain relative to the control. To determine which
growth parameters contributed to any measurable difference in competitive
ability, the parameters lag time, maximal growth rate, Ks, and death rate were
individually measured for each strain in batch monocultures. Additionally, cell
yield and cell volume were measured to better understand what physiological

traits might be impacted by the deletion of rRNA operons.

57

4.2 Materials and Methods
Bacterial strains

The bacterial strains used in this study are thoroughly described in
Chapter 3. All strains come from the same line of E. coli that has evolved for
10,000 generations under batch culture conditions in glucose minimal medium
(56). These strains are isogenic except for any rRNA deletion and a neutral point
mutation rendering reciprocal strains able to utilize the sugar arabinose (Ara+),
where as the parental line cannot (Ara'). The Ara' and Ara" strains form red and
white colonies, respectively, when grown on tetrazolium arabinose indicator agar
(TA), enabling the identification and enumeration of colonies from each
population in mixed cultures during direct competition experiments (57).

Strains with the rrnA, rmB, and both rrnA and B operons deleted are
designated as ArmA, ArrnB, and ArmAB, respectively. Deleting the same rRNA
operons from strains of each marker type (Ara+ and Ara‘) allowed for reciprocal
competition experiments. The promoters and all but the end of the 23S gene and
the 58 gene were deleted from the targeted rRNA operon on the chromosome

(described in Chapter 3).

Media and culture conditions

Luria Bertanl (LB) broth (67) was used when growing cells from glycerol
stocks stored at -80°C. These cultures were grown in 3 ml volumes, sterile 16 x
100 mm culture tubes, and incubated at 37°C, shaking at 225 rpm overnight.

The liquid medium used for all growth and competition experiments was Davis

58

Minimal broth (DM) (15) supplemented with 2.0 mg/L thiamine hydrochloride.
Glucose was added (0.1, 25, or 1000 mg/L) as the limited nutrient. The specific
concentrations of glucose used are indicated with each experiment below (9.9.
DM+1000 refers to 1000 mg/L glucose). The solid agar medium used was either
LB agar (LB broth with 1.5% [wt/vol] agar added) or tetrazolium arabinose (TA)
indicator agar (57).

Batch cultures were grown in 10 or 50 ml volumes in 50 or 250 ml
Erlenmeyer flasks, respectively, incubated at 37°C, and shaking at 225 rpm for
complete aeration. For all experiments, cultures were first grown in LB from
—80°C glycerol stocks, and then used to inoculate 8 DM conditioning culture with
a specific concentration of glucose and incubated for exactly 24 hours, shaking at
37°C. The inoculum for the conditioning culture was approximately 1:100 of the
population density achieved at the specific glucose concentration. This
conditioning culture was then used to inoculate (1 :100) fresh DM medium for the
experiment performed.

Chemostat cultures were inoculated with conditioning batch cultures and
grown in 65 ml volumes in 8 parallel 3 x 20 cm glass vessels that were enclosed
with a #8 rubber stopper and submersed in a 37°C water bath (Figure 4.1). The
chemostat vessels received medium and air through a thin glass tube that ran
through the top of the vessel and down to 1 cm above the bottom. The sterile
medium (DM+25) was delivered from up to 4 parallel reservoirs (10 liters each) to
the inflow tube by a variable speed peristaltic pump. Air was also pulled in

through the inflow tube and a 0.45 pm HEPA filter, providing ample aeration and

59

mixing. A second glass tube that was connected to a vacuum, entered through
the top of the vessel and extended approximately halfway to the bottom of the
vessel. This outflow tube removed culture medium and cells, maintained the
volume of the culture, and provided the negative pressure that pulled in the air for
aeration and mixing. A sampling port was located on the side of the vessel and
above the level of the culture. This port was fitted with a rubber septum and was
used to aseptically inoculate and sample the chemostat vessel. Sterile, 5 ml
syringes fitted with 5/8 inch, 23 gauge needles were used to inoculate the
vessels, and sterile 1 ml syringes fitted with sterile, 9 inch, 23 gauge, stainless

steel, non-coring needles were used to sample the cultures.

60

A

Culture Medium '
Vacuum Outflow Inflow Peristaltic Medium

Reservoirs

Pump

 
 

\ Pump

 

 

Culture Vessels

\l

. N’O 0
Air Inflow —————> Y Sterile Medium

(0.3 um HEPA Filter)

/\’-:itfi
5;:7'

 
 

   

 

 

 

 

Outflow Tube .
(used medium z;
and vacuum)
Inoculation and /
Sampling Port
5'1-
0‘ Inflow Tube
3:; (Medium and Air)

 

 

Figure 4.1 (A) Diagram of chemostat culture system. (B) Sideview of
individual chemostat culture vessel.

61

Measurement of specific growth parameters
Lag time

The amount of time (h) required for a culture to enter exponential growth
after a 1:100 inoculation with stationary phase cells was estimated and referred
to as lag time. The cultures were treated identically to those in batch culture
competition experiments (described below), and all inoculations and samplings of
cultures were staggered by 15 min intervals to insure precise timing. Each
conditioning culture was used to inoculate 10 ml of fresh DM+25 medium.
Immediately after inoculation and mixing (time, t = 0 hours), a sample of each
culture (approx. 3.5 x 105 CFU/ml) was diluted and an aliquot was spread onto 4
replicate LB agar plates. The cultures were incubated for exactly 4 hours (t = 4),
and another sample of each culture was taken, serially diluted, and an aliquot
was spread onto 4 replicate LB agar plates. The time point of 4 h represented a
convenient time to sample each culture during mid-exponential growth. The agar
plates were incubated overnight and the CFU/ml of culture was determined.

The duration of lag time for each culture was estimated similarly as
described in (55). Exponential growth was extrapolated from the population
density at t = 4, back in time until the extrapolated CFU/ml intersected the initial
CFU/ml at t = 0 h that was actually measured. The point of intersection
represents the elapsed time (lag) after inoculation, before the culture reached
exponential growth, assuming the initiation of exponential growth was
instantaneous. The lag time for each strain (control, ArmA, ArmB, and ArrnAB)

was represented as the mean of 4 cultures (2 Ara“ and 2 Ara' cultures).

62

Maximal growth rate

The maximal growth rate (pMAx) of each strain in DM+1000 medium, and
under the batch culture conditions described above, was estimated by calculating
the slope of the linear regression of natural log-transformed population density
values (O.D.600) versus time during exponential growth. The concentration of
1000 mg/L glucose in DM+1000 was used to achieve cultures with sufficient
optical densities for accurate measurement with a Lambda 3 UV-Vis
spectrophotometer (PE Applied Biosystems). Cultures with larger volumes (50
ml) than the competition experiments were also used to enable the many
samplings needed for spectrophotometric measurement and minimize the impact

upon the entire culture volume.

Ks

The parameter Ks represents the concentration of a growth-limiting
substrate at which a culture grows at half the maximal specific growth rate. This
parameter is a component of the Monod equation, which represents the
relationship between specific growth rate (it) and substrate concentration (8) as

follows;

M = S ' MMAx/(S + Ks)-

The IQ for each strain was calculated using the estimated specific growth
rate (it) of each strain growing on DM+0.1; a concentration of glucose (0.1 mg/L)

very near Ks for the control strain (76). The glucose concentration (8), the

63

specific growth rate (u), and uMAx (determined as described above) were used in
the above equation rearranged to solve for Ks as follows;

Ks = 3 [(HMAX/H) " 1]-
Ks estimates were calculated for three replicate cultures of each strain.

As described above, LB cultures were inoculated from —80°C glycerol
stocks for each strain and incubated overnight at 37°C, shaking at 225 rpm. The
final population density obtainable on DM+0.1 was much lower than the final
population density of the LB overnight cultures. The LB overnight cultures were
therefore serially diluted 1:104 before they were used to inoculate the 50 ml of
DM+0.1 medium used for the conditioning cultures. Aliquots of the conditioning
culture were transferred to fresh DM+0.1 medium. Dilutions of the freshly
inoculated test cultures were spread onto triplicate LB agar plates for
enumeration at time points equaling 0, 2, and then every hour up to 12 hours
after inoculation. The mean CFU/ml values for each culture were Ln-transformed
and plotted versus time. The specific growth rate of each culture was estimated
as the mean slope of these linear regressions, and used with glucose

concentration and pMAx to calculate Ks as described above.

Death rate

The death rate of each strain during stationary phase was estimated as
the rate of decrease in population density between 12 and 24 hours. This
measurement was modeled after those in (76). Triplicate 10 ml DM+25 medium

flasks were inoculated (1 :100) from conditioning cultures as described above,

64

and dilutions of each culture were spread onto triplicate LB agar plates every
hour after inoculation between 12 and 24 hours. The mean CFU/ml for each
culture at each timepoint was Ln-transformed and plotted versus time. The
regression was assumed to be linear, and the inverse slope was reported as
death rate (Le. a positive death rate represented a decrease in population
density). Monitoring of the change in CFU/ml of each strain was continued up to

197 hours after inoculation.

Mefiurrement of ot_her physiological traits
Cell yield

The cell yield that is reported was measured as the number of cells
produced per pg of glucose in chemostat monocultures (equilibrium population
density), and assumes that the utilization of glucose in each culture is equal.
This was measured for each strain by determining the equilibrium population
density (cells/ml) in two independent chemostat monocultures at slow (0.11 h“)
and fast (0.56 h") dilution rates, and dividing by the concentration of glucose in
the medium (25 ug/ml). Population density was determined for 4 aliquots of each
culture by direct enumeration of the fixed cells using a Coulter Counter particle
analyzer (Beckman Coulter, lnc., Fullerton, CA), and subtracting “background”

particles measured in a sample of sterile medium.

65

Cell volume
The mean cell volume (fl) of each strain was also measured using the
same samples used to determine cell yield, and the Coulter Counter particle

analyzer (Beckman Coulter Electronics).

Batch culture competition experiments

The relative fitness of each strain was measured in mixed batch culture
competition experiments with the control strain (as described in 54). Each
competing strain with opposite marker type (Ara" and Are) was grown from
stocks and conditioned for 24 hours in DM+25 (described above). The
conditioning cultures of each competitor were used to inoculate four replicate
culture flasks containing 10 ml DM+25 with each population at 1:200, for an initial
population ratio of 1:1. Each mixed culture was then sampled at time (t) = 0 h
and 24 h, where an aliquot of each culture was diluted when necessary and
spread onto TA agar to enumerate each population. Also after 24 h, a 100 ill
aliquot of each culture was transferred to fresh DM+25 medium. Again after 24
h, these competition cultures were sampled to enumerate each population and
used to inoculate fresh medium. Each competition culture was transferred to
fresh DM+25 at least 4 times. During each period of competition, the cultures
underwent lag phase, exponential growth, and at least 10 hours of stationary
phase before an aliquot was used to inoculate fresh DM+25 to repeat the growth

phase cycle (about 6.6 doublings).

66

The relative fitness of each strain was calculated as the number of
population doublings relative to the control. The initial densities of the Ara’ and
Ara’ competitors, can be represented as N1(0) and N2(0), respectively; where
N1(1) and N2(1) refer to their corresponding densities after 24 hours. The
average rate of increase, or Malthusian parameter (m), for each competitor
during each 24-hour interval (1 d) was estimated as;

m1 = In [N1(1)/N1 (0)1/(1 d).
where m, has units of d", and the mean of these estimates was used for each
strain. The fitness of one strain relative to another, W1; 2, was estimated as the
ratio of the number of doublings of the two competitors, which is identical to the
ratio of their Malthusian parameters (52);
W1; 2 = [Th/"12.
All statistical inferences about relative fitness values were based on at least 4

independent replicate estimates of relative fitness for the same strain.

Chemostat culture competition experiments

Relative fitness was also measured for each strain based upon population
dynamics in mixed chemostat competition cultures with the control strain. At
equilibrium, the ratio of population densities will not change if the competing
populations have the same competitive ability. The ratio of population densities
however, will change over time in favor of any population with a higher

competitive ability than the other competing population. The difference in

67

competitive ability of one population relative to the other (relative fitness) is
represented by the rate at which the ratio of population densities changes.

The entire chemostat system (8 parallel culture vessels, medium (DM+25)
reservoirs, and all tubing) was assembled, clamped, and autoclaved as a closed
system. After autoclaving, the culture vessels were placed in a water bath, the
tubing was attached to the peristaltic pump, and the vacuum pump and waste
trap were connected to the outflow tubes (Figure 4.1). The peristaltic and
vacuum pumps were then turned on to fill the culture vessels. Once filled to the
level of the outflow tube and at a constant temperature of 37°C, each culture
vessel was inoculated (1:200) with conditioning cultures of each competing
population of opposite marker (Ara+ and Ara'). This inoculation was performed
by first mixing the populations in a 5 ml syringe and injecting the inoculum into
the culture vessel through the sampling port.

Upon inoculation, the peristaltic pump was stopped for 6 — 12 hours until
the cultures had reached early stationary phase in the culture vessels. At that
point, the peristaltic pump was turned on at the desired rate and the culture was
sampled (t = 0). A dilution of these samples was spread onto triplicate TA agar
plates to enumerate each competing population. The chemostat cultures were
assumed to have reached equilibrium when the total population density did not
change significantly for at least 3 volume changes. The competition cultures
were sampled periodically (based upon flow rate) to monitor changes in the ratio

of population densities.

68

Calculation of chemostat-specific growth parameters

Using the growth parameters K5 and ilMAx. the J parameter of each strain
was calculated for the slow (0.11 h“) and fast (0.81 h") dilution rates used in the
mixed culture chemostat competition experiments. Bacterial populations present
in chemostats at equilibrium grow at the same rate as the dilution rate (D) and
reduce the concentration of nutrients to the lowest amount needed to grow at that
rate. According to chemostat theory, this nutrient concentration (J) can be
determined using the equation:

J = Ks (D/I).

where r= (lLMAx — D) > 0, the intrinsic rate of increase of a particular species (37).
For a population to exist in a chemostat at equilibrium its maximal specific growth
rate must be faster than the dilution rate (D < WM) and the input concentration of
the limiting nutrient (So) must be greater than J (So > J). Also according to
theory, the population with the lowest J parameter will competitively exclude all
other species with higher J values. The J values for each strain at both dilution
rates were used to predict the outcome of the mixed chemostat culture

competition experiments.

4.3 Results

Specific growth parameters
The mean values of lag time, maximal growth rate, and Ks for each strain
are given in Table 4.1. A set of planned comparisons was used to calculate

which differences between the mean values for each strain were statistically

69

significant (73) (same comparisons as in Appendix C). The effect on lag time
and maximal growth rate was directly correlated with the number of rRNA
operons in each strain. Compared to the control, the mean lag time estimates of
both ArrnA and ArrnB were about 4% (2.4 min) longer than the control, whereas
the lag time of the ArrnAB strains was about 20% (12 min) longer. Wild type E.
coli cells are therefore about 1/3 of a doubling ahead of cells with only 5 rRNA
operons inoculated at the same time, in the same medium, and under batch
culture conditions. Maximal growth rates of the rRNA deletion strains were
slightly slower than the control, and also correlated with the number of deleted
rRNA operons. Opposed to lag times and maximal growth rates, the differences
in Ks values for each strain were not correlated with the number of deleted rRNA
operons. The ArrnB strain showed a slightly lower Ks than the control, but the Ks
of both the ArmA and ArmAB strains increased significantly (29.7 and 26.2%,

respectively) relative to the control.

70

Table 4.1 Growth parameters measured for each straina

 

 

Strain Lag (h)b um 01")" Ks (rig/ml)d
Control 1.43 :t 0.039 1.03 :1: 0.011 0.118 :1: 0.0035

ArmB 1.48 :t: 0.091 1.00 :t 0.012 0.111 a: 0.0025

ArmA 1.48 a: 0.076 1.00 :I: 0.016 0.153 :1: 0.0047
ArrnAB 1.71 a: 0.070 0.99 :I: 0.014 0.149 :I: 0.0116

 

a values are means of at least three measurements :1: standard errors of the

measurements
° planned comparisons show that only ArmAB mean is significantly different from

the control (F123 = 7.37, P < 0.05) and both ArmA and ArmB (F128 = 6.54, P <
0.05)

° planned comparisons show that ArrnB and ArmA means are significantly
different from the control (F144 = 3.09, P < 0.10), and ArmAB mean is significantly
different from the control (F124 = 4.84, P < 0.05) but not ArrnA and ArrnB.
° planned comparisons show that ArrnA and ArmAB means are not different
from each other but are significantly different from ArmB (F13 = 20.12, P < 0.05)
and the control (F13 = 11.50, P < 0.05), which are also not different from each
other

There was no appreciable death rate measured for any of the strains
between 12 and 24 hours after inoculation (Figure 4.2). The population densities
of these cultures did not change significantly [paired ttest (P3 > 0.05)] until after
48 hours, when the number of viable cells dropped rapidly (data not shown).
There was also no discernable difference between the strains in their long-term

survival under these conditions.

71

 

6x107 1

 

 

 

 

 

+Control
+ArmB
—o—ArmA
—A—An'nAB
_ 4x10 . K
E \ % \
2x107 1 r l 1 l l l
12 14 16 18 20 22 24

Time (h)

Figure 4.2 The death rate is shown as the change in population density
(CFU/ml) of each strain in batch monocultures from 12 to 24 hours.

72

[tapas—med physiological traits
Cell yield

The mean cell yield of each strain is shown in Figure 4.3A. These data
seems to coincide with the K5 data in that strains with a deleted rrnA operon are
affected to a larger extent than strains with the rrnB deletion. In slow dilution rate
chemostat cultures (0.11 h") the cell yield of ArrnB is similar to the control strain,
whereas the cell yield of ArrnA and ArrnAB are 39% and 42% lower (F13 =
910.25, P < 0.05), respectively. In faster dilution rate chemostat cultures (0.56 h'
1), no significant difference was measured between the cell yield of all strains

based on planned comparisons.

Cell volume

The mean cell volume of each strain during slow (0.11 h“) and fast (0.56
h") dilution rate chemostat conditions is shown in Figure 4.3B. These data
suggests an inverse relationship to cell yield in slow chemostat conditions
because the mean cell volume of both ArrnA and ArrnAB cells is larger than the
control and ArmB. Based on planned comparisons, the only significant
difference between mean cell volumes was between ArrnAB and the control

under fast chemostat conditions (F13 = 10.83, P < 0.05).

73

 

Yield (105 Cells/pg glucose)
0
's

.0
N

 

 

 

 

.0
o

 

 

 

 

_o
on
i
l
l
l

.0
a:
l

 

 

Cell volume (fl)
9
A

.0
N

    

 

 

Slow IIIII Fast

 

 

 

 

 

0.0

 

Figure 4.3 (A) Cell yield (106 cells/pg glucose), and (B) cell volume (fl) for
the control (black), ArmB (dark grey), ArmA (light grey), and ArrnAB (white)
strains are shown for chemostat monocultures at slow (0.11 h'1) and fast (0.56
h'I) dilution rates. Error bars represent the standard error of the measurements
of two independent cultures.

74

Batch culture competition emeriments

The batch culture competition experiments gave very accurate and
precise measurements of competitive ability under fluctuating nutrient conditions.
The competitive ability of each strain is defined by how quickly it can initiate
exponential growth upon inoculation into liquid culture, the maximal growth rate,
the relationship between the concentration of substrate and specific growth rate
(Ks), and the death rate once the population reaches stationary phase. The
measurement of competitive ability was expressed as relative fitness and refers
to the increase in population density relative to a competing population in a direct
competition experiment (53, 54). The mean relative fitness of each strain was
directly related to the number of rRNA operons in that strain, in that the strains
with fewer rRNA operons were had lower a relative fitness (Figure 4.4). The
results of planned comparisons between the mean relative fitness of each strain
show that only ArrnA and ArmB have relative fitness means that are not
significantly different (F124 = 0.05, P > 0.10; see Appendix C, (73)). The value
given for the control strain represents the difference in competitive ability
between strains with different markers (Ara+ and Ara'). The values given for
each of the deletion strains is the mean of relative fitness values for strains of

both marker types.

75

 

1.05

1.00 -

0.95

0.90 -

Relative fitness

0.85 -

 

 

 

 

 

0.80 -

Figure 4.4 Relative fitness values for the control (black), ArmB (dark grey),
AnnA (light grey), and ArrnAB (white) strains in batch culture competition
experiments with the control strain. The value for the control represents the
relative fitness of the Ara+ versus the Ara' control strain. Error bars represent
standard errors of the measurements.

76

Chemostat competition experiments

Mixed culture chemostat competition experiments between each strain
and the control at two different dilution rates (0.11 and 0.81 h") allowed for the
direct measurement of relative fitness. Relative fitness under equilibrium
chemostat conditions is represented by the change in the ratio of population
densities per doubling. The population ratio will change in favor of the population
with a higher relative fitness under chemostat conditions, and the slope of this
regression indicates the magnitude of the fitness difference.

The relative fitness of each strain under slow chemostat conditions is
shown as the ratio (vs. the control strain) of the population densities versus the
number of doublings in Figure 4.5. The data shown for the control strain
represents the relative fitness for the Ara+ control strain in competition
experiments with the Ara' control strain. The stability of the population ratio in
the control strain competition experiments illustrates the neutrality of the marker
phenotype. The competitive ability of ArmB is also very much the same as that
shown for the control strain. The ArmA and ArmAB strains, however, have a
much lower competitive ability in the slow chemostat relative to the control strain.
The relative fitness of ArrnB (Figure 4.6) and ArmA (Figure 4.7) was also
measured in chemostats with faster dilution rates (0.81 h"). The relative fitness
of ArrnB at faster dilution rates was not discernable from that in chemostats with
slower dilution rates. ArrnA still had a negative relative fitness at fast dilution

rates, but it had a higher relative fitness than at slow dilution rates.

77

1.4

 

 

 

 

 

 

 

1.2
.9 -~ \
q—l
E --
c 0.8—
.9
E
3 _
Q 0.5
O
o.
0.4—
+Control
+ArmB
0.2— —o—ArmA
— —A—ArmAB
0.0 ....,....,...,,
0 5 10 15

Doublings

Figure 4.5 Relative fitness of each rRNA deletion strain in chemostat
competition experiments with slow dilution rates (0.11 h'1) is represented as the
population ratio (vs. control) versus the number of population doublings. The
data for the control represents the fitness of the Ara+ strain relative to the Ara‘
strain. Error bars represent the standard error of four independent cultures for
each strain.

78

1.41

 

 

Population ratio

0.4 —

0.2 1

 

 

000 I l I I I T I I I I T T I T I T I T I l I I I I I

Doublings

Figure 4.6 Empirical (solid lines with symbols) and simulated (dotted lines)
relative fitness of ArrnB in fast (black) and slow (dark grey) chemostat competition
experiments are represented as the population ratio (ArrnB vs. control) versus the
number of population doublings. Error bars indicate the standard error of four

independent cultures.

79

 

Population ratio

 

 

 

 

Doublings

Figure 4.7 Empirical (solid lines with symbols) and and simulation (dotted
lines) relative fitness of WA in fast (black) and slow (light grey) chemostat
competition experiments are represented as the population ratio (ArrnA vs.
control) versus the number of population doublings. Error bars represent the
standard error of four independent cultures.

80

Calcrfiited memostat-specific growth parameters

The J parameters calculated for each strain at each dilution rate correlate
with the outcome of each chemostat competition experiment described above
except for ArrnB at slow dilution rates. According to the calculated J parameter
ArmB would have a higher competitive ability compared to the control strain at
slow dilution rates, which did not occur. It is important to point out however, that
the calculations of J parameters used the means for Ks and pMAx reported in
Table 4.1. Not all of the differences between these means were statistically
significant. Therefore, only the differences in J parameters of strains with a
deleted rmA operon (ArrnA and ArrnAB) with those with a functional rrnA operon
(control and ArmB) should be considered significant, as with the Ks values for

each strain (Table 4.1).

Table 4.2 Chemostat-specific growth parameters

 

 

 

Strains J (jig/ml)al
D°=0.11 D=0.81
Control 0.0141 0.434
ArrnB 0.0137 0.473
ArrnA 0.0189 0.652
ArrnAB 0.0185 0.871

 

a J = Ks-(D/pMAx — D) as described in (37).
b D = dilution rate (h")

4.4 Discussion
This study used strains of E. coli to test the assertion that the rRNA

operon copy number of a prokaryotic species represents an adaptation to

81

environmental conditions, with respect to the pattern of nutrient availability. One
of the basic tenets of bacterial physiology is that macromolecular content (e.g.
RNA concentration) changes with growth rate. The synthesis of rRNA, which is
regulated at the transcriptional level, is the rate-limiting step in ribosome
synthesis and therefore can limit growth. Multiple rRNA operons are an effective
way to increase the cell’s capacity to synthesize rRNA by effectively increasing
rRNA gene dosage.

It has been postulated that a culture of E. coli responds quickly to nutrient
influxes with rapid growth as a result of the seven rRNA operons on the E. coli
chromosome (1, 47). Previous work studying the connection between the
number of functional rRNA operons on the chromosome and cell physiology (5,
17, 23) have used strains in which the effect on rRNA operon copy number could
not be discerned from secondary effects of the rRNA operon manipulations. The
research reported here was based on strains of E. coli containing rRNA operon
deletions, which were designed to eliminate the shortcomings of previously used
rRNA operon deletion strains by removing the majority of the targeted rRNA
operons (including the promoters) but leave all other genes intact. These strains
were therefore used to investigate the connection between rRNA operon copy
number, individual growth parameters, and cumulative competitive ability under

different nutrient availability regimes.

82

gffect of rRNA copv number on relative fitness in fluctuatinthutrient conditions

The results of the batch culture competition experiments support the
hypothesis that multiple rRNA operons are an advantage under culture
conditions that are defined by fluctuating nutrient availability. This was evident
when strains with either one or two deleted rRNA operons were competing with
the control strain for limited nutrients in batch cultures, where nutrient availability
fluctuates with each transfer to fresh medium. There was a direct correlation
between the number of rRNA operons in the cells and relative fitness under
batch culture conditions in glucose minimal medium (Figure 4.4).

The overall relative fitness in batch culture of each competing population
was comprised of a collection of individually measured growth parameters (Table
4.1). The impact of each growth parameter on relative fitness was tested using
the computer model described in Chapter 5 by simulating competition
experiments in which only one, or all of the measured growth parameters differed
between the two competing populations (Figure 4.8). Although the differences
between individual growth parameter means were not always statistically
significant, the cumulative differences between all growth parameters of each
strain significantly effected overall relative fitness (Figure 4.4, Appendix C) and
were accurately depicted in the model simulation results. The simulation data
shows that the decrease in maximum growth rate and increase in lag time
contribute similarly to the decreased relative fitness of strains with 6 rRNA

operons (ArrnA and ArrnB in Figure 4.8). When a second rRNA operon was

deleted, the increase in lag time of the resulting strain (ArrnAB) had a larger

83

impact on relative fitness than maximum growth rate. The amount of time
needed for a stationary phase culture to reach exponential growth upon
inoculation into fresh medium (lag time) increased about 4% relative to the
control strain in ArrnB or ArmA, whereas the lag time of ArmAB was about 20%
longer than the control strain. This amount of time (0.28 h) is appreciable when
one considers that during that time, the control strain is growing exponentially

and beginning to remove the limited amount of nutrients.

84

 

1.02

Relative Fitness

 

 

 

 

 

 

 

 

 

Empirical All Lag “MAX KS

Figure 4.8 Empirical relative fitness values for the control (black), AnnB
(dark grey), ArmA (light grey) and ArmAB (white) strains in batch culture
competition experiments, and relative fitness values based on simulations using
all measured parameters or each parameter individually.

85

Populations of E. coli have evolved under environmental conditions that
are best described as feast and famine, referring to the disparity between
fluctuations in nutrient availability inside animal intestines (63). Measurements of
lag time depend heavily upon previous culture conditions and can therefore be
somewhat arbitrary. The measurement of lag time in this study mimics the exact
conditions faced by cells in batch culture competition experiments, as well as the
conditions in which the control strain had evolved for 10,000 generations (56).
Additionally, lag time measured in these experiments may be more informative
than previous measurements of “shift-up” times (17). In the strains used to
measure “shift-up” times in the previous study, the effect of rRNA operon
inactivation cannot be discerned from the effect of expressing the additional
antibiotic resistance markers. Also, the “shift-up” times reported in the previous
study reflect the amount of time needed to shift from one exponential growth rate
to a higher growth rate. The measured lag times in the present study however,
not only reflect the time needed to make additional protein-synthesizing
machinery (i.e. ribosomes) for growth under the more favorable growth condition,

but to do this starting from a “non-growing” physiological state.

fleet of rRNA operon c0pv number on relative fitness in stable nutrient
conditions

The second hypothesis states that cells with fewer rRNA operons are at
an advantage under stable, nutrient limiting growth conditions. Although the

amount of ribosomes is closely regulated with growth rate in E. coli cells growing

86

at moderate to fast rates (11), E. coli cells maintain excess protein synthesizing
capacity under very slow growth conditions ()i s 0.2 h") (47). It was suggested
that these “excess” ribosomes might offer the cell an advantage upon an influx of
nutrients. In the absence of a fluctuation in nutrient availability however, the
synthesis of excess ribosomes under slow growth rates might be a metabolic
expense that cells with fewer rRNA operons would not experience. Cells with
fewer rRNA operons would therefore be at an advantage under slow, stable,
limiting nutrient conditions where there were no sudden influxes of nutrients.
According to the hypothesis stated above, a negative correlation between relative
fitness and rRNA operon copy number was expected in chemostat competition
experiments. This hypothesis however, was not supported by the experimental
data, and there was no correlation with rRNA operon copy number (Figures 4.5-
4.7).

The rRNA deletion strain ArrnB had a slightly lower mean Ks value (Table
4.1). When the mean growth parameter values were used (regardless of
significance) to simulate chemostat competition experiments (grey dotted line,
Figure 4.6), or determine J parameters (Table 4.2) at slow dilution rates, ArmB
was predicted to have a higher competitive ability than the control strain. There
was no measurable difference however, in the relative fitness of ArmB in
chemostat competition experiments with slow dilution rates (0.11 h"). The
population ratio of ArrnB versus the control did not change under equilibrium
chemostat conditions for nearly 15 doublings of the total population. The

disparity between the simulated and empirical competition outcomes indicates

87

the limit of precision in measuring K5 and the sensitivity of relative fitness to
seemingly small differences in Ks at slow dilution rates. The model simulations
and J parameter determinations more accurately depicted the outcome of
chemostat competition experiments with fast dilution rates between ArmB and
the control strain (black dotted line, Figure 4.6), where the impact of K3 on
relative fitness is smaller.

Concomitant with model simulations and J parameter calculations, the
ArmA and ArmAB strains were considerably less fit than the control strain under
slow chemostat conditions (Figure 4.5, Table 4.2). Their population ratios were
less than 1:10 versus the control in about 10 doublings of the total population.
The low relative fitness of both strains coincided with their increased Ks values,
which were close to 30% higher than either the control strain or ArmB.
Chemostat competition experiments at fast dilution rates showed that although
ArmA still had a higher J parameter and therefore was less fit than the control
strain, the relative fitness of the ArrnA strain was much improved relative to that
in slow chemostat competition experiments (Figure 4.7). Based on chemostat
theory and previously reported data (70), chemostats with faster dilution rates
have higher residual concentrations of the limiting nutrient. The relative fitness of
a strain is less sensitive to differences in Ks under these conditions, which

explains the data above.

88

Possible impact of tRN_A imbalance

The differential effect of deleting the rrnA or rmB operon may be linked to
some of their known differences. It is possible that the sequence divergence
between the rrnA and rmB operons (as with all of the operons) might lead to
differential expression, or result in ribosomes with different properties.
Alternatively, the differential phenotypes could be related to the different tRNA
genes in the internally transcribed spacer (ITS) regions of rmA and rmB. The
rrnA, D, and H operons have the Ile and Alalg tRNA genes in their ITS regions,
whereas the rmB, C, G, and E operons have only the GIU2 tRNA gene. These
tRNA genes are only represented in the rRNA operon ITS regions, whereas other
tRNA genes are present in multiple copies elsewhere on the chromosome. It
was originally thought that deleting the rmA and rmB operons would minimize the
disruption of the tRNA gene pool because they represent one of the two groups
of rRNA operons in E. coli, with respect to their ITS tRNA genes, and neither
operon has any distal tRNA genes. It was hopedothat the effect of deleting the
tRNA genes present in the rRNA operons would be minimal as long as several
intact copies were present in the remaining rRNA operons.

The deletion of a tRNA gene present in multiple copies may have a larger
detrimental effect on competitive ability than anticipated, especially at slower
growth rates. The tRNA genes that are present as multiple copies on the
chromosome (including those in the rRNA operon ITS region) represent those

tRNAs that are relatively more abundant in the pool of tRNAs in the cell, and

89

whose codons are used more frequently as growth rate increases (21 ). These
properties suggest that those tRNA genes present in multiple copies, and their
abundance relative to other tRNA genes, could be important to the efficient
growth of a cell.

It is not possible to discern the effect of deleting rRNA genes from deleting
tRNA genes present on the rRNA operons. None of the experimental results
reported above can be excluded unequivocally from being partially, if not entirely,
a consequence of deleting the tRNA genes along with the rrnA and rmB operons.
Furthermore, the differential effect of deleting the rrnA versus the rmB operon
might simply be the result of deleting the two tRNA genes in rrnA instead of one

tRNA gene in the rmB ITS region.

Pr0pp§ed explanation of cumplative resu_lt§

With the considerations stated above, a possible explanation is presented
that could tie the results of batch culture and chemostat cultures together with the
result of deleting rRNA and tRNA genes. This explanation is intended to
illustrate the possibilities of extending this current line of research. The inability
to discern the effect of deleting rRNA genes from deleting tRNA genes on an
rRNA operon illustrates the need to investigate the impact of both rRNA and
tRNA gene copy number on cell physiology and competitive ability under
fluctuating and stable nutrient conditions.

The effect of deleting rRNA genes from the chromosome of a cell could

impact the cell’s ability to respond to fluctuations in nutrient availability by

90

increasing the amount of time needed for de novo ribosome synthesis. This
impact is supported by the increase in lag time and decrease in relative fitness of
rRNA operon deletion strains in batch culture. During steady state growth
however, the cell might compensate for the deleted rRNA genes by increasing
the amount of synthesis from the remaining copies. The deletion of tRNA genes
however, causes an imbalance in the tRNA gene pool that is not corrected by
increased rRNA synthesis from the remaining rRNA operons; the resulting
proportion of tRNAs is still unbalanced with respect to the deleted tRNA gene.
Compared to deleting the rmB operon, deleting the rrnA operon might result in a
greater impact on the tRNA gene pool because there are two tRNA genes in the
rrnA ITS region (tRNA"° and tRNAA'am) and only two other rRNA operons (rrnD
and H) from which to make these tRNAs. The impact of an imbalance of tRNAs
could result in decreased translational efficiency, as more time is required for the
affected tRNAs to come into contact with the ribosome. One possible response
to a decrease in translational efficiency could be the increase in production of
ribosomes. A cell that needs more ribosomes for the same translational capacity
would therefore require more metabolic energy per doubling event. The
requirement of more metabolic energy would have its largest impact at the lowest
concentration of nutrients.

There is an interesting correlation between the cell yield and cell volume
of the ArrnA and ArmAB strains and their relative fitness in chemostats with slow

dilution rates, which supports the explanation above. The cells of the ArrnA and

ArrnAB strains are larger and have a lower cell yield than those of the control and

91

ArmB strains (Figure 4.3). At faster dilution rates (0.56 h") however, the
difference in either trait between all strains is greatly reduced. This suggests that
cells with a deleted rrnA operon are somehow physiologically different at slow
dilution rates, resulting in larger cells and decreased cell yields. Although cell
yield is not a component of competitive ability in unstructured environments,
there is likely some relationship between the cell yields, cell volumes, Ks and
ultimately, relative fitness of the strains with a deleted rrnA operon.

The difference in Ks values, cell yield, and cell volume between ArmA and
ArmB coincides with relative fitness differences observed in chemostat
competition experiments. It does not however, explain how the deletion of an
rRNA operon affects the cell or why there was a differential effect between the
deletion of rmA and rmB. Work is currently ongoing to confirm these
observations and address these questions. The rmA operon will be replaced in
ArrnAB to test for the “rescue” of the ArrnA phenotype, and then smaller
fragments of the rrnA operon will be deleted in an attempt to identify the specific

region of the operon that is responsible for the phenotype.

Potential limitations of comparing rRNA deletion strains to control strains

As discussed in Chapter 3, part of the procedure used in deleting the
rRNA operons from the E. coli chromosome included the generalized
transduction of genomic DNA from a recD mutant strain of E. coli K-12 into the E.
coli B evolved strains used in this study. It is likely that other regions of the donor

strain’s genomic DNA, including the rRNA operon, also recombined into the

92

genome of the recipient strain. What is not known is if any of these potential
regions of donor DNA is responsible for the effects observed in the rRNA operon
deletion strains.

The E. coli K-12 and B strains are relatively closely related, falling within
the same subgroup of E. coli strains based on 20 enzyme-encoding loci. Only
one of these loci (phn operon) is even close to within 100 kb of the rmA or rmB
operons and therefore possibly co-transduced using phage P1. Nonetheless, it
will be important for control strains to be “reconstructed” from the intermediate
strains by replacing the marker alleles with the entire rRNA operon instead of the
deleted operon as in the deletion procedure (Chapter 3). Any effect caused by
collateral transduction of genes other than the rRNA operon marker alleles will
remain in these reconstructed control strains, which will have the rRNA operon
replaced at the site of the marker allele. The control strains used in this study
have not been transduced with any of the recD donor strain DNA and therefore
do not represent controls that could be nearly as isogenic as the reconstructed

control strains.

Conclufsion

The work reported here supports the hypothesis that multiple rRNA
operons are an advantage to an organism under fluctuating nutrient conditions.
Further experimentation is needed however, in order to separate the effect of
rRNA gene copy number from the tRNA genes that were also deleted as well as

any effect caused by the transduction of E. coli K-12 DNA into the E. coli B strain

93

used in this study. No trade-off for the maintenance of multiple rRNA operons
was measured in this study. Cells with fewer rRNA operons were not at an
advantage in slow, stable, nutrient conditions. Additionally, the deletion of the
rrnA operon caused a marked decrease in relative fitness under chemostat
conditions that was attributable to an increased Ks parameter. Either the rrnA
operon or its tRNA genes were somehow essential for comparable relative
fitness under nutrient limiting growth. This attribute was previously unknown for
any of the rRNA operons and raises many new questions about the differential

properties of the multiple rRNA operons and their tRNA genes in E. coli.

94

CHAPTER 5

DEVELOPMENT OF A COMPUTER MODEL TO SIMULATE THE DYNAMICS

OF TWO POPULATIONS COMPETING FOR A LIMITING NUTRIENT

5.1 Introduction

In natural ecosystems, a multitude of physicochemical influences as well
as the activities of other microorganisms within the same environment influence
microbial populations. A common goal of microbial ecologists is to understand
how microbial populations interact with each other and their environment, and
how these interactions affect the abundance and distribution of these
populations. A fundamental interaction that drives many of these factors is the
availability of nutrients for growth and proliferation of a population. When two or
more populations within a microbial community vie for one or more of the same
resources, the ultimate result can be inhibition of growth and/or survival of some
of these competing populations. This competition for limiting resources is
probably the most common type of interspecies interaction.

Mathematical models are powerful tools that can be used in investigating
the impact of nutrient availability on population dynamics. Over fifty years ago,
Monod derived equations that describe the relationship between the
concentration of a limiting nutrient and the specific growth rate of a microbial
population (60). Relevant parameters of microbial growth include; lag time,

maximum specific growth rate, Monod half-velocity coefficient (Ks), death rate,

95

and yield. Many other equations have been developed to define substrate-
specific growth rates more accurately, however most fit the available data just as
well as the Monod equation, or contain biologically indefinable components (48).
The close fit, extensive testing, and mathematical simplicity of the Monod
equation justifies its use under many circumstances.

In the study reported here, the Monod equation was incorporated into a
computer-based model that was programmed to simulate the dynamic
relationships between microbial populations and the availability of the limiting
nutrients. An object-oriented software package was used to program the model,
which makes it easy to expand and use to simulate the direct competition
between populations in which the limiting nutrients are supplied in a batch,
continuous, or pulsed manner. All of the individual growth parameters mentioned
above are incorporated into the model and can be set to match those of any
competing population, as well as the beginning densities of those populations.

This chapter describes the computer model mentioned above, its
behavior, verification, and validation. After programming the model, its behavior
was studied to insure that it correctly simulated the relationships between nutrient
concentration and population growth described in the Monod equation. The
results from previously reported simulations and empirical competition
experiments (37, 76) were compared with simulations using the model reported
here in order to verify the performance of the model. Comparing simulations of
competition experiments described in Chapter 4 with the empirical data further

validated the performance of the model. This model does not incorporate any

96

new definitions of microbial growth, but it does provide an easy way to portray
how individual growth parameters affect the competitive ability of a bacterial
population under fluctuating and stable nutrient conditions. Additionally, the
model represents an easy to use tool to further study the interactions between

well-defined microbial populations competing for the same limiting nutrients.

5.2 Methods
Identification of variables

The model (Figure 5.1) is a three-compartment system composed of the
limiting nutrient (glucose (9 ~L")), and two competing populations (Population 1
and Population 2, (cells -L")). The concentration of limiting nutrient in the culture
vessel is the central state variable of the system. Dynamics of this variable are
determined by the flows; ‘flow in’ from the reservoir, ‘flow out’ from overflow, and
from the ‘utilization’ of each population. The physical “flow in’ and ‘flow out’ of the
culture vessel are determined by the reservoir concentration multiplied by the
dilution rate D (h“), which represents the number of volumes that are cycled per
hour. The growth (cells-(L-h)-1) divided by the yield (cells-g-1) of each
population determines the removal of glucose from the system due to utilization

(9'(L'h)-1) by each population.

97

 

GLUCOSE PARAMETERS

 

 

 

 

 

 

 

 

 

Iriiialpuse Interval Dirlionralle
lirxon
tow Ghmse
a; Flowout
Constant . .
flow Ulizallon1 Ulialon2
Reoervo'roonc.
POPULATION 1 POPULATION 2
Popualion1 6 W0

Dealh1 '9“?
eYeUT

Y

 

ar

:.Spgt31 WWI .. Mm1 .. 3139132 mmz " M11892
Lagl'rne1 MaxguMhlatBZ Lagt'rneZ de‘

 

 

 

 

Figure5.1 Mapshovlingflterelafionshlpbetweensmdts(0).oonverlers(l:l),and
flows(°‘5‘)ofthenndelpmgammedinflleobjed-aientedlanguageSTELLA

 

An additional flow into the glucose stock was programmed into the model
to facilitate simulations of fluctuations in the amount of glucose entering the
system. This flow ‘flux flow’ represents a second reservoir (‘flux reservoir’) of
limiting nutrient that can be added to the total flow into the glucose stock. The
‘flux flow’ is controlled by the function (PULSE), which pulses an input of a
specified size. In this case the pulse input is the ‘dilution rate’ multiplied by the
‘flux reservoir’ concentration of limiting nutrient (g-L“). In using the PULSE
function, the user has the option of setting the time at which the PULSE will first
fire (‘initial pulse’) as well as the interval between subsequent PULSEs
(‘interval’).

The flows, ‘growth’ and ‘death’ determine the dynamics of each
population. The rate of growth of each population follows the Monod equation
(60) as follows,

ll = uMAx°[S]/Kc.+[S]. (1)
where u is the specific growth rate, pMAx is the maximum growth rate, [S] is the
substrate concentration, and K3 is the Monod half-velocity coefficient ([8] where
p. is equal to uMAx/Z). Equation (1) defines the relationship between the specific
growth rate of a population relative to the concentration of a limiting nutrient. in
addition to the above parameters, lag time acts to delay the growth of a
population at the beginning of a simulation, for the duration of the ‘lag time’ (h),
and is programmed using logic statements. The ‘Iag growth’ converter returns a

value of zero growth for the specified duration of ‘Iag growth’.

99

The flow ‘death’ represents the overall death rate of the population.
During batch culture simulations the ‘dilution rate’ is set = 0, and therefore the
flow ‘death’ is equal to the parameter ‘death rate’. Under continuous culture
conditions the ‘dilution rate’ is set > O, and ‘death’ is equal to the dilution rate
multiplied by the population density and represents the washout of cells from the

chemostat.

Parameterization

The parameterization of the model for each simulation was performed
using the “Control Panel” interface. Here, the values for initial population size,
Ks, maximum growth rate, lag time, and death rate can be set for each
population. Parameters defining the reservoir and initial vessel concentrations of
glucose, dilution rate, and the flux flow can also be set in the control panel.
Other controls present in the control panel interface and other locations
throughout the model include navigation and simulation controls. These controls
allow the user to navigate quickly from one location to another, specify the step

times and duration of simulations, and run, stop, or pause a simulation.

Modeling method

This deterministic model was programmed with STELLA® Research 5.1.1
software. The intuitiveness of this object-oriented modeling language allows the
user to easily understand and program the relationships between the system

components, and make additions or amendments to the model. Data generated

100

during simulations was exported to spreadsheets in Microsoft Excel ®, and
Microcal Origin ® for further statistical and graphical analysis. The data
exchange was made possible via the dynamic data-exporting feature of the
modeling software. Another benefit of the STELLA software package is cross
compatibility, allowing any model to be used on several computing platforms.

All simulations were run using the Euler’s method of derivation with a step
time of 0.00625 h (delta time, DT = 0.00625). This “optimum” DT was
determined by sequentially decreasing the step time from 1.0 h by 1/2 until the
simulation resulted in no quantitative change in population densities and no
additional dampening of oscillations occurred. The “1/2 test” was used as
suggested by the STELLA technical documentation because of the binary
arithmetic the computer uses for calculation (39). At the DT = 0.00625, any
further 1/2 decreases did not affect the outcome of the simulations, but chemostat
simulations could not be run longer than 200 hours before the limit of maximum
step times was reached. There is an internal limit of ~32,767 step times set
within the software. If this limit would be exceeded, a warning message will

appear suggesting a larger DT or shorter simulation.

Sensitiviy analyses

The model was tested for sensitivity to changes in the parameters;
pMAx, lag time, and Ks. Sensitivity was determined during batch and chemostat
culture conditions by simulating competitions between populations that were

identical except for the parameter being tested. Under batch culture conditions,

101

the sensitivity of the system to changes in the ratio of parameter values (e.g.
pMAxlluMAXZ) was determined as the effect on the ratio of population densities
(Pop 1/Pop 2) after 24 hours. Under chemostat conditions, the sensitivity of the
system was measured as the change in the population ratio per doubling at fast

(0.81 h") and slow (0.11 h") dilution rates.

Verification of model performance

The behavior of the model was verified using data that was previously
modeled with Monod kinetics (76). When a batch culture competition was
simulated, both experimental and theoretical data (Table 5.1) (76) were
compared to simulation results (Table 5.1, Figure 5.4). Quantitative comparison
was performed using empirically and mathematically derived values of relative
fitness (W) (54, also described in Chapter 4).

In the case of continuous culture simulations, the model was
parameterized with data for three experiments from Hansen and Hubbell (37)
(Table 5.2). Their empirical and simulation data were compared to simulation
results (Figure 5.6). In these simulations, the relative change in population
densities indicates relative fitness and represents the rate at which one
competitor excludes another. Quantitative comparison of simulation results was
also made on the basis of the J parameter for each population (37, also
described in Chapter 4). The J parameter represents the subsistence

concentration of the limiting resource for a population at equilibrium. The

102

population with the lowest J value will competitively exclude all others with higher

J values.

Validation of model performance

Simulations of competition experiments described in Chapter 4 were run
using independently measured growth parameters for each strain (Table 4.1),
and the specific population densities and nutrient parameters of the direct
competition experiments that were simulated. The simulation results were
compared to relative fitness values calculated for batch culture and the
population ratio dynamics for chemostat competition experiments in order to

validate the model performance.

5.3 Results
Sensitivitv analvses

Under batch culture conditions, the outcome of the simulation was most
sensitive to MAX (Figure 5.2). As the MW of Population 1 was increased relative
to Population 2 (a higher umax ratio), the population ratio (Pop 1/Pop 2) increased
proportionately in favor of the population with the higher uMAx. The model was
slightly less sensitive to changes in lag time, and rather insensitive to changes in
Ks. As predicted by the Monod equation, increases in either lag time or lg had a
negative effect on relative fitness. For the purpose of comparing absolute model
sensitivity, the effect on population ratio (Pop1/Pop2) was plotted against

increases in lag time and Ks of Population 2 relative to Population 1. The

103

sensitivity of these parameters was as expected and indicates proper model

behavior.

104

2.0— /
1.8— /

g
\

Sensutrvrty
F .

4 lag o
/ ./
1.2 — 0/

/I /'/./

- . Ks
././ “W
1.0 /¢ —— —— ——

1.0 1.2 1.4 1.6 1.8 2.0
Ratio of tested parameter

 

 

 

 

Figure 5.2. Batch culture sensitivity analyses. The population ratio
(Population 1/Population 2) after 24 h simulations (labeled as Sensitivity) was
plotted against the ratio of the tested parameters (“MAX““MAXZ' lagZ/lag1, and
KSZIKS1).

105

The model was sensitive to changes in WAX under chemostat culture
conditions in a similar manner as in batch culture conditions (Figure 5.3). Unlike
batch culture conditions however, the model was also sensitive to changes in Ks.
The model was more sensitive to Ks at slow dilution rates (0.11 h") than at fast
dilution rates (0.81 h"). As the K3 of a population decreased (K32/Ks1 increases
in Figure 5.3), the population ratio increases in favor of the population with the
smaller Ks value. Under chemostat conditions, the sensitivity of the model to
differences in lag time was not relevant because chemostats represent

equilibrium conditions in which the cells are in steady state growth.

106

 

0.3 1
—D'— “MAX faSt
+ ”MAX SIOW
1 —Ar— KS fast
+ KS slow

 

 

 

 

.0
N
1

Sensitivity

0.1 —

 

 

1.0 1.2 1.4 1.6 1.8 2.0

Ratio of tested parameter
Figure 5.3 Sensitivity analyses of chemostat simulations. The sensitivity
(change in population ratio (Pop1IPop2) per doubling) of model parameters

under fast (0.81 h") and slow (0.11 h") chemostat dilution rates is plotted
against the ratio of the parameters (pMAx1/uMAx2) and (KSZIKS1).

107

Model Verification

Overall model behavior during batch culture simulations was correct in
that the growth of each population was delayed for the amount of lag time given,
followed by a period of exponential growth at the specified maximal growth rates
(Figure 5.4). Following with Monod’s equation, growth was limited by nutrient
availability. The model behavior was further verified by comparison of simulation
and empirical data from Vasi et al. (76) to the simulation results (Table 5.1).
When relative fitness values were calculated with the simulation data, the model
was able to account for 90% of the empirical relative fitness reported for the
derived strain (Table 5.1). As was the case for Vasi et al (76), this only
represents 60% of the actual gain in relative fitness of the derived strain. They
went on to show that this discrepancy was best explained by an under-estimation
of maximum growth rate based on the increase in biovolume versus cell
numbers. When biovolume was accounted for in the adjusted maximal growth
rate, the simulation results were not different than experimental data. With the
same considerations, this would show a satisfactory verification of the model

behavior and performance under batch culture conditions.

108

Table 5.1. Parameter values and results for a simulation

of previously published batch culture competition
experiments (76)

 

 

 

Organism
Fitness components parental derived
Lag time (h) 1.5264 1.2470
um, (11") 0.7726 0.8887
Ks (10'5 g-liter") 7.27 8.80
Other parameters of
simulation
Glucose (g-L") 0.25 0.25
Initial population 5.00 5.00
(108 cells-L")
Yield (1012 cells-g") 2.6352 1 .8386
Relative fitness (W)
empirical (1 ) 1 .3486
simulation (1 ) 1.21
simulation a (1) 1.21

 

a all other values are from (76).

109

 

 

 

 

 

 

11 = , , , , , 0.025
8
57‘ —0.020 2
.1 0
£2 2
'5
‘33 -0.015 a
£5 -1
:9, 5
t ~0.010 2
c7) - a
5 5 ‘5
. _ O
o . 0.005 8
, (D
2 I"
8 , ‘ , , , , 0.000 3
0 4 8 12 16 20 24

TIME (Hours)

Figure 5.4 Batch culture verification data. The dynamics of the ancestor
(left axis, solid line) and derived (left axis, dashed line) populations, and the
concentration of glucose (right axis, dotted line), using the estimated parameters
in (76) are shown. ’

110

The model was also run as a chemostat using data from Hansen and
Hubbell (37) (Table 5.2). Each of the competition experiments reported in their
study was simulated, and the results were compared to their simulation and
empirical data. This represents a qualitative verification of the model under these
conditions since no exact data for population densities were given in the
published study. The population dynamics in model simulations (Figure 5.5)
corresponded to those reported by Hansen and Hubbell, and the J values

calculated for each population were identical to reported values (Table 5.3).

111

 

.ooammmc

 

 

3.6 6-9 x md N6 No. x v MO 2
. . . om m B: 1
mm @0me moo 6-9me md .5958 m m.& 0 1
. . .ooqumc
mac 6.? xm F No mow xw mad
. . . moo 9.6:
mm 6-038 moo 6-9me md mopxw mO m
Ed vow x fim Qm «or x m mmmO<d
Qm vow x F 5.0 6-9 x o.m ma Now x m mO <
A} «.0 5 3.9 cs 3.9 $92.8 20: 3.2.82 8585
o om 1 av. 22> cozmsaoa .mEc. .mzofimm E68233
8.2.20 mcozmiaod

fig mucoEtoaxo 533388 93.30 588650 8:253 2.630365 .6 9.25366 Lo. mo:_m> LoHoESmd .N.m 63m...

Figure 5.5 Population density dynamics for chemostat verification

simulations using parameters in Table 5.2 for experiment A, B, and C.

113

Log cells/L

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

— -8
----- PA0283
2 I I I I
0 20 40 60
6 ..
g --------- C-8 nal'spec5
§ ——-C-8 nal*"specr
8’
_l 2 _
1 a
0 I I I I I
0 20 40 60 80 100
6 _
5 _ ...............................................................................
-J.
g
8 --------- C-8 nallspecs
8’
.1 2 4 C-8 nalsspecr
1 _
o I I ' r ' I ' I ' l
0 20 40 60 80 100 120

Time (Hours)

Figure 5.5

114

Table 5.3 Comparison of reported and calculated J

parameters for chemostat model verification

 

 

 

J parameterAg-L"?
Strain Reportedb Calculated
Ac C-8 2.40 x 10'7 2.40 x 10‘7
PAO283 2.19 x 10'5 2.18 x 10'5
B C-8 nal’specs 1.98 x 10'7 1.98 x 10'7
C-8 nalsspec' 1.35 x 10” 1.36 x 10'7
C C-8 nal'specs 1.98 x 10'7 1.98 x 10'7
C-8 nalngec' 2.01 x 10'7d 2.01 x 10‘7

 

a J parameter was calculated as follows; J = Ks-(D/r),

where r = ”W - D.

values taken from (37).

° letters correspond to panel labels for verification

simulations represented in Figure 5.5
d this value represents J calculated from reported Ks, D,

and pMAx values. A J value of 1.99 x 10'7 g-L'1 was

reported (37).

115

 

Model Validation

Model validation is based on the ability to simulate the modeled system
using independent empirical data. Using all parameters measured for each
population (Table 5.4), the model was able to simulate the relative fitness of each
strain very closely (Figure 5.6), thus validating its performance under batch
culture conditions. When run as a chemostat, the model also closely simulated
empirical competition results (Figures 5.7 and 5.8), except in the case of ArrnB at
slow dilution rates (0.11 h"). The calculated J parameters and simulation data
predicted that ArmB would be more fit than the control strain at slow dilution
rates. This discrepancy and its possible implications are discussed below.
These simulations were very dependent upon the proper parameterization of
initial population ratios. The exact initial population of each empirical competition

was used for the simulations instead of an assumed 1:1 ratio.

Table 5.4 Calculated J parameters for chemostat model

validation at different dilution rates (D)

 

J parameter (g-L")al

 

 

 

Strain D = 0.11 h“ 0.81 h"
Control 1.41 x 10'2 4.39 x 10'1

ArrnB 1.38 x 10'2 4.79 x 10‘1

ArrnA 1.90 x 10‘2 6.58 x 10'1
ArrnAB 1.88 x 10'2 6.89 x 10'1

 

a J parameter was calculated as follows; J = Ks-(D/r), where
r = “MAX - D-

116

 

1.02

1.00 ‘

0.98 ‘

0.96 ‘

0.94 ‘

Relative fitness

0.92 7

0.90 ‘

 

 

 

 

 

 

Empirical Simulation

Figure 5.6 Validation of model performance under batch culture
conditions. Empirical versus simulation relative fitness values for the control
(black), An'nB (dark grey), ArmA (light grey), and ArmAB (white) strains.

117

Figure 5.7 Population ratio dynamics for slow chemostat validation
simulations of direct competition experiments with (A) ArmB, (B) ArmA,
and (C) ArrnAB strains vs. the control strain using parameters in Chapter

4.

118

$4

mil

(3:18:61

Population ratio Population ratio

Population ratio

 

1.4

 

 

1.2 -———— A — .. _--.-

 

 

 

 

 

 

 

0.8 ~ _*- -—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.6 —-— -- -- ~—
0.4 1 - , LL, , ~ W —— » ______ ——-._ .—
....... ArmB
0.2 ,w, —— —— _ . .
Simulation
0.0 1 I T l l r T
0 2 4 6 8 10 12 14 16
1.4
B ....... W
1.2 L" " . . 1,,
Simulation

 

 

 

 

 

 

1.01: 7:71;" - - —

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.8 7‘

0.6 7 ‘

0.4 —

0.2 '

0.0 I I l I l
0 2 4 6 8 10 12

1.4

121 C,,_.__ ---— "-A- AnnAB
‘" simulation

 

 

 

 

 

 

 

 

Doublings

Figure 5.7

119

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A -- . .
1,2 «- _ I - __ Simulation __.~
-f-.’. 1.0 -»
E .1
E 0.8 -
<0
3 0.6
8 _
o. 0.4
0.2 ~—
0.0 . , , 1 . .
0 10 20 30 40 50 60
1.4
12 B simulationfi
o . ~ — ---I---AnnA
E 1.0‘ _ ,
.5 0.8
(U _-
3 0.6
Q '- .—
8 0.4-
0.2 _. 41......-.-.-T
0.0 T I r I fir
0 10 20 30 40 50 60
Doublings

Figure 5.8 Chemostat validation at fast dilution rates. Empirical and
simulation data for (A) ArmB/Control and (B) ArrnA/Control population ratio
versus doublings during fast chemostat dilution rates (0.81 h'1)

120

 

 

5.4 Discussion

The computer model reported here was designed to simulate the
dynamics of two populations that are competing for a shared limiting nutrient.
This model incorporated the dynamic changes in population growth and nutrient
availability based upon the Monod equation, which describes the relationship
between limiting nutrient concentration and specific growth rates of bacterial
populations (60). The Monod equation was used because it is mathematically
simple and accurately describes the relationship between nutrient concentration
and growth rates under most conditions (14, 48). An object-oriented
programming language and its supporting software package (STELLA, High
Performance Systems, Inc.) were used to program this three-component model
and all of the underlying relationships. An advantage to using this approach was
the ease with which the model could be changed, expanded, and used.

A major problem in modeling is the balance of sufficient realism with
complexity. If a model does not contain at least the most important mechanisms,
it does not describe reality; whereas if a model is too complex, it is almost
impossible to test experimentally (24, 65). No matter how complex a model may
be, it is still a simplification of the more complex system being modeled. This
simplification allows the most important interactions within a system to be
understood, even without knowing all of the components or interactions within a
system. Interactions that are incorrectly represented in the design of a model are

often identified by its performance. Assumptions are made in the design of any

121

 

 

model, and those that are inappropriate also affect model performance. Two
possible limitations of the model presented here stem from the use of the Monod
equation to define the growth of a population and the accurate measurement of
the growth parameters for each competing population. Specific assumptions of
this computer model are that the environment is homogenous and there are no
direct interactions between populations (e.g. inhibition, predation, synergism).
There was also no allowance in the model for the internal storage of nutrients or
the requirement of nutrients for the maintenance of viability between nutrient
pulses. The above assumptions and omissions do not substantially affect the
accuracy of the simulation data discussed below.

The model correctly predicted the population dynamics in both batch and
chemostat culture conditions, which was evident in the nature and degree that
model outcomes were sensitive to changes between the various growth
parameters of each population. In batch culture conditions, there was a linear
relationship between maximum specific growth rate and gains in relative fitness
of a population, but smaller gains in relative fitness resulted from similar
decreases in lag time (Figure 5.2). Unlike maximum growth rate and lag time,
only very large changes in Ks (orders of magnitude) could result in comparable
relative fitness gains. This was predictable since populations growing in batch
cultures experience very low nutrient concentrations (where K; has the largest
effect on growth rate) for only a small period of time before entering stationary
phase. When chemostat cultures were simulated, the sensitivity of relative

fitness to changes in growth parameters was different in several ways.

122

 

 

 

Populations that are maintained in chemostat cultures are in steady state growth,
and therefore lag times are no longer relevant. The relative fitness of a
population was still very dependent upon maximal growth rates, regardless of
dilution rates (assuming D s ILMAx) (Figure 5.3). Kg however, played a much
larger role in determining relative fitness in chemostats than in batch culture
conditions. Relative fitness in chemostat conditions was more sensitive to
changes in Ks because populations growing in chemostats consistently
experience very low concentrations of nutrients. The concentration of limiting
nutrient in a chemostat vessel decreases with the dilution rate (70). At slower
dilution rates therefore, relative fitness is more sensitive to differences in Ks.

The accurate simulation of competition experiments using growth
parameter data from parental and evolved strains of E. coli (Figure 5.4 and Table
5.1) verified the performance of the model under batch culture conditions. The
discrepancy between the simulation and empirical relative fitness values from
Vasi et al. (76) was due to an over estimation of yield. When this was accounted

for, there was no longer a difference between simulation and empirical values

 

(76). The performance of the model under batch culture conditions was further
validated by the accurate simulation of competition experiments between several
strains of E. coli described in Chapter 4 (Table 4.1). The simulations of batch
culture competition experiments that were based upon data from Chapter 4
accurately measured the relative fitness of each strain compared to empirical

values (Figure 5.6).

123

 

The performance of the model was also tested under chemostat
conditions by its accurate simulation of competition experiments using data from
Hansen and Hubbell (37). The foremost predictive parameter for the outcome of
competition under continuous culture conditions is based upon chemostat
kinetics and the Monod equation. This parameter, J (37), represents the
concentration of limiting nutrient on which a population can subsist in continuous
culture at a given dilution rate. According to chemostat theory, the population
with the lowest J value will out-compete all others with higher J values (37, 75).
Two populations with equal J values are predicted to coexist, even if they have
different values for MW and Ks (37). The model successfully predicted all of the
above-mentioned outcomes (Table 5.2, Figure 5.5). The model performance
under chemostat conditions was validated by its ability to accurately predict the
outcome of all the chemostat competition experiments described in Chapter 4,
with the possible exception of ArrnB versus the control strain at a slow dilution
rate (Table 5.3, Figure 5.7 and 5.8).

Unlike the ArrnA and ArmAB strains, the measured Ks of ArrnB was
conspicuously lower than the control strain. Comparison of the model
simulations with the empirical competition experiments suggests that the K3 is
higher for ArmB or lower for the control than measured. The compliance of batch
culture and fast chemostat simulations to empirical results suggests that
measurements of maximum growth rate and lag time are accurate for ArrnB.
Based upon its measured Ks value, ArmB is predicted to outcompete the control

strain at slow dilution rates (Table 5.3, Figure 5.7A), where Ks has a larger effect

124

 

 

on relative fitness. This predicted outcome does not occur in the empirical
experiments, suggesting that the accuracy of the Ks measurements was
exceeded under the slow dilution rate chemostat competition experiments
between ArrnB and the control strain. The Ks measurements are relatively
accurate considering that reported Ks values for E. coli can differ by orders of
magnitude depending on the strain, the history of the culture, and the laboratory
in which the measurements were made (70). The simulation results of the slow
dilution rate competition experiments between ArrnB and the control illustrate
how the model can be used to identify inaccurate parameter values.

This model contributes to the study of microbial ecology in several ways.
It offers tremendous power in predicting the outcome of direct competitions
between populations for a limiting nutrient through the use of growth
characteristics measured in pure culture, thus eliminating the need for mixed
cultures. The comparison of empirical to simulation results can also be used as
an indication of other parameters that may be affecting the competition
(production of inhibitory compounds, cross feeding, etc.). The limitations of the
model are also linked to its attributes. The accuracy of growth parameters
determined in pure culture can have an impact on the predicted outcomes,
especially in the case of maximal growth rate and Ks (70). These variables are
determined during steady state growth in batch cultures and K5 for instance, may
actually change during chemostat or non-equilibrium conditions (35, 49). Further
study of the assumptions behind these measured growth parameters and their

effect on modeled systems would increase the accuracy and dependability of

125

 

 

 

predictions made by the model, and could provide useful insight into the forces

that drive population dynamics during resource competition.

126

 

 

 

CHAPTER 6

SUMMARY AND CONCLUSIONS

The research in this dissertation investigated the role of rRNA operon
copy number as a life history trait and its connection with competitive ability in
prokaryotes. Several central observations about rRNA operon copy number in
prokaryotes suggest this trait should be considered in terms of the life history of a
species. Multiple copies of the rRNA operons can be found on the chromosome
of certain prokaryotes, whereas almost all other genes are present as single
copies. The number of rRNA operons is highly conserved among isolates of the
same species (as in 13) with a few known exceptions (4, 43). Although
conserved among isolates of the same species, rRNA operon copy number is not
strictly conserved between different species, regardless of phylogeny (69). The
strict conservation of rRNA operon copy numbers within a species but not
between different species is a common pattern for life history.

The hypotheses tested in this dissertation were based on the presumption
that the number of rRNA operons in prokaryotes is largely an adaptation to
environmental conditions; genome size, genetic drift, and phylogenetic
constraints are not major deterministic factors (see Chapter 1). The first
hypothesis states that multiple rRNA operons are an advantage in environments
that are defined by fluctuations in nutrient availability. The advantage of multiple
rRNA operons is thought to be due to higher rates of rRNA synthesis that permit

quicker responses to favorable growth conditions and faster growth rates. The

127

second hypothesis states that fewer rRNA operons are an advantage in
environments with stable, limiting nutrient availability, and is based on the
metabolic expense of over-expressing multiple rRNA operons at slow growth
rates.

To test these hypotheses, the effect of rRNA operon copy number on cell
physiology and competitive ability was measured in othen~ise isogenic strains of
E. coli. The number of rRNA operons was manipulated in these strains by either
the introduction of rRNA operon-containing plasmids (Chapter 2, (74)), or the
deletion of one or two rRNA operons from the chromosome (Chapter 3). In
previous studies, strains of E. coli containing manipulations of the rRNA operons
were used to test the effect of rRNA operon copy number on various aspects of
cell physiology. The strains constructed in these previous studies were
unsuitable for the experiments in this dissertation because of the methods used
to manipulate rRNA operon copy numbers, and therefore, some of the results of
the previous studies are also called into question. Some limitations of using
plasmid-borne rRNA operons, as in (41) and Chapter 2 (74), are that the plasmid
copy number is either unknown or can vary, the plasmid can be lost without
proper selection, and the present rRNA operon copy number can only be
increased, not decreased. Other alterations of E. coli rRNA operons either
contained additional antibiotic resistance genes that were expressed from the
promoters of the inactivated rRNA operon (18), produced truncated rRNA
transcripts from intact promoters of partially deleted rRNA operons (3, 5), or

affected adjacent genes on the chromosome (23). The experiments in this

128

dissertation required that the effect of altering rRNA operon copy numbers be

discernable from the secondary effects of manipulating the rRNA operons on the

 

chromosome. It was therefore necessary to construct new strains with deletions
of the chromosome that included the rRNA promoters and a large portion of the
rRNA operon, but not other genes (discussed in Chapter 3). The effect of
altering the rRNA operon copy number on cell physiology and competitive ability
was directly measured in the resulting strains because the known promoters of
effected rRNA operons were deleted and other genes on the chromosome were L,
not disrupted.

It was postulated that the seven rRNA operons of E. coli are linked to
the cell’s ability to respond quickly to fluctuations in nutrient availability (1, 47).
Direct evidence for this connection was the increased amount of time a strain of
E. coli required to respond to more favorable growth conditions after the
inactivation of rRNA operons (17). The inactivation of up to four of the seven
rRNA operons did not have “a significant impact” upon maximal growth rates
(17), which led to the conclusion that the benefit of E. coils multiple rRNA
operons was not higher maximal growth rates, but quicker responses to
fluctuations in nutrient availability. Unfortunately, the strains used in this study
leave some doubt as to whether the increased amount of time it took for a strain
to respond to favorable growth conditions was due to having fewer intact rRNA
operons or expressing the additional antibiotic resistance genes used to

inactivate the rRNA operons.

129

The results from Chapter 4 support the connection between rRNA operon
copy number and E. coli’s ability to respond quickly to fluctuating nutrient
availability (Chapter 4, Table 4.1). There is also a correlation between rRNA
operon copy number and relative fitness in batch culture competition experiments
(Chapter 4, Figure 4.4), which further supports the hypothesis that multiple rRNA
operons were an advantage when nutrient availability fluctuates. When
compared to ArrnAB, the higher relative fitness of the control strain (7 rRNA
operons) in batch culture conditions was largely due to its ability to more quickly '. .
respond to an influx of nutrients (i.e. shorter lag time). Relative fitness, as
measured in batch culture competition experiments, can be largely affected by
seemingly minor differences in lag time and maximal growth rate between two
competing populations. The relative impact of each growth parameter is
illustrated in the batch culture sensitivity analyses of the computer model
(Chapter 5, Figure 5.2), and their effect on empirical relative fitness
measurements (Chapter 4, Figure 4.8).

Increasing the rRNA operon copy number of E. coli by the addition of
plasmid-borne rRNA operons had a negative effect on growth rate, accompanied
by an over expression of RNA, in media that supported slow doubling times
(Chapter 2, 74). These data suggested that multiple rRNA operons are a
metabolic burden at slow growth rates. A decrease in this metabolic burden
however, was not measured with strains containing fewer rRNA operons in
chemostat competition experiments (Chapter 4, Figures 4.5-4.7). Under the

chemostat conditions tested, strains of E. coli with fewer rRNA operons were not

130

at an advantage in environments with stable nutrient conditions. These results
suggest that, under stable slow growth conditions, multiple rRNA operons in E.
coli are not a metabolic expense that can be relieved by deleting two of its seven
rRNA operons. These cells can either control the level of constitutive expression
equally well from 5 and 7 rRNA operons, or the decrease in metabolic expense
could not compensate for some detrimental effect of deleting the rRNA operons.

The number of rRNA operons on the chromosome of a prokaryotic E4.
species may be a component of its life history. An observed life history is the
lifetime pattern of growth, differentiation, storage and, most importantly,
reproduction. The availability of nutrients is arguably one of the major selective
forces imposed upon populations of prokaryotes. The effect of rRNA copy
number on E. coli in batch culture competition experiments suggests that multiple
rRNA operons contribute to E. coli’s fitness in environments characterized by
fluctuating nutrient availability. The hypotheses tested in this dissertation would
extend these findings to all prokaryotic species and suggest that the number of
rRNA operons on the chromosome is an indication of the life history of a
prokaryotic species. These hypotheses are also supported by the correlation
between rRNA operon copy number and the rate of colony appearance of soil
isolates (46). After spreading a soil sample onto dilute nutrient agar, isolates that
form visible colonies in the first few days have about 5 rRNA operons, whereas
isolates that form colonies up to two weeks later have 1 to 2 rRNA operons on

their chromosomes.

131

Additional circumstantial evidence that support the hypotheses of this
dissertation can be found in the abundant examples in which the number of rRNA
operons on the chromosome of certain prokaryotes coincide with their presumed
life history. Prokaryotic species such as the bioluminescent symbiont
Kryptophanaron alfredi from the Caribbean flashlight fish (78), the parasitic
mycoplasmas (2) and mycobacteria (31 ), and the marine oligotrophic
ultramicrobacterium Sphingomonas sp. strain RB2256 (25) have one to two
rRNA operons. These prokaryotes exist in stable and/or poor nutrient
environments, tend to have relatively slow growth rates, and respond slowly, if at
all, to fluctuations in nutrient availability. Prokaryotic species with a relatively
high number of rRNA operons on their chromosome include E. coli with 7, Vibrio
spp. with 8 to 10, and Bacillus subtillus with 10 rRNA operons. Compared to the
prokaryotes with one or two rRNA operons listed above, these high rRNA operon
copy number prokaryotes tend to respond more quickly to fluctuations in nutrient
availability with relatively fast growth rates.

It is necessary to consider other possible explanations for the
experimental results in this dissertation. One step of the procedure used to
delete the rRNA operons in this study could have led to the incorporation of
donor DNA other than the intended allele during generalized transduction with
phage P1 (discussed in Chapter 3). The deletion of rRNA operons in this study
needs to be complemented by replacing the deleted region of the chromosome
with the full rRNA operon. The resulting strains would therefore be

“reconstructed” controls that would allow any effect of co-transformed DNA

132

regions to be directly measured and accounted for. Also as discussed in Chapter

4, it is not possible to distinguish the effect of the deletion of rRNA genes from

 

the deletion of tRNA genes on the rrnA and rmB operons. Interestingly, there
was a much larger effect when the rrnA operon was deleted from the
chromosome of E. coli than when the rmB operon was deleted, which was most

obvious under slow dilution rate chemostat conditions (Chapter 4, Figure 4.5).

.‘q
E

Previously, the effect of deleting tRNA genes along with the rest of the rRNA

operons was not apparent, but it is possible that the secondary effects of rRNA

1:"
r

operon deletions could have obscured the impact of deleting the tRNA genes in
these studies. Additionally, the effects of rRNA operon deletions have not been
studied in chemostats at dilution rates as slow as in this dissertation (0.11 h").
Deleting the rrnA operon in the ArmA and ArmAB strains resulted in a
much larger decrease in relative fitness than measured for the ArmB strain in
chemostat competition experiments with slow dilution rates (Chapter 4, Figure
4.5). The large decrease in relative fitness was due to an increase in Ks for both

rrnA deletion strains. Also at slow dilution rates, ArmA and ArmAB strains had

lower yields and larger cell volumes than either the ArrnB or control strains
(Chapter 4, Figure 4.3). A possible explanation for these results is presented in
Chapter 4, which suggests that the tRNA genes present in the internally
transcribed spacer (ITS) regions of the rrnA and rmB operons may be causing
the measured effects.

The rrnA ITS region contains the two tRNA genes tRNAA'au; and tRNA"°1

whereas, the rrnB ITS region contains tRNAG'"2 gene. The tRNA genes in each

133

rRNA operon ITS region represent three of the most abundant and most
frequently used tRNAs in the cell. There is a direct correlation between the
number of copies of each tRNA gene and the concentration of their respective
tRNAs. Additionally, the molar ratio of tRNA per ribosome increases as growth
rate decreases (21). All of these factors may explain why a disruption in the
tRNA gene pool could cause large effects under slow chemostat conditions, as
with the rrnA operon. The differential effect of deleting rrnA or rrnB could be
related to the two tRNA genes in the rrnA ITS region, which are present in only
three copies on the chromosome, versus the single tRNA gene in the rmB ITS
region that is present in four copies.

It is also possible that either the expression or the products of the rrnA

operon are somehow more essential under the slow growth, low nutrient

conditions of the slow dilution rate chemostats. Specialization of rRNA operons

might occur when multiple rRNA operons are maintained on the same

chromosome. Any detrimental effect of mutations in one copy of an essential

gene might be lessened by the expression of remaining unaffected copies on the

same chromosome. The specialization of rRNA operons in E. coli could also
explain the data in Chapter 4, which showed a substantial negative effect on

relative fitness when rrnA was deleted, but not rmB. Previous studies suggest

there is some differential expression of E. coli’s seven rRNA operons (up to 1.4-

fold), but not specifically between the rrnA and rmB operons (18). The

expression of individual E. coli rRNA operons however, has not been

134

 

investigated under the slow chemostat conditions used in the research reported
here.

The work in this dissertation highlights the differential effect between
complete deletions of the rrnA and rmB operons at slow growth rates.
Previously, the deletion of tRNA genes along with the rest of the rRNA operon
was considered to play only a minor role as long as intact copies of these genes

either remained elsewhere on the chromosome (17) or were provided on a

r... min—min"

plasmid (5). The results of this dissertation illustrate the importance of discerning
the effect of deleting tRNA genes from deleting the rRNA genes on an rRNA
operon. Work continues in an effort to identify which portions of the rrnA operon
(including the tRNA genes) can be linked to the phenotype observed in the ArmA

and ArmAB strains.

135

APPENDIX A

P1 vir LYSATE PREPARATION

The protocol for the preparation of P1 virlysates was adapted from L.
Snyder (personal communication), and uses LB agar plates with a LB soft agar
overlay. A culture of donor cells that was in late log or early stationary growth
phase in LB was amended to 1 mM CaClz from a 1 M stock. A 100 pl aliquot of
the amended culture and 50 ml of phage P1 vir at a titer of approximately 109-101O
PFU/ml was added to 5 ml of soft agar also amended to 1 mM CaClz. The soft
agar was mixed and then poured onto moist LBKan agar. These overlayed plates
were incubated overnight at 37°C. Three discernable, isolated plaques were
picked with sterile toothpicks, which were then placed in 0.2 ml sterile saline
(amended to 1 mM CaClg) in a sterile 14 ml disposable polypropylene culture
tube. The phage were allowed to diffuse away from the toothpicks into the saline
for 30 min to several hours. Next, 50 pl of donor bacteria (as above) was added
and the phage were allowed to adsorb to the bacteria for 3 min. After the phage
had adsorbed to the donor bacteria, 3 ml of soft agar (50°C, 1 mM CaClz) was
added, mixed, poured onto an LBKam agar plate, and incubated overnight at 37°C.
When confluent plaques are visible, 4 ml of sterile saline was layered on the agar
surface and the phage are allowed to diffuse into the saline at 4°C overnight.

The saline is then decanted into a sterile polypropylene centrifuge tube and

centrifuged for 10 min at 5,000 x g and 4°C to pellet any bacterial cells and

136

 

debris. The supernatant is decanted into a fresh centrifuge tube and 100 pl of
chlorophorm is added to lyse any remaining bacteria. The phage titer of the
lysate was determined (usually 109 to 1010 PFU/ml) and transduction was

performed as described in Chapter 3.

137

APPENDIX B

VISUALIZATION OF rRNA OPERONS USING SOUTHERN HYBRIDIZATION

Approximately 1 pg of genomic DNA was digested with 10-20 units of the
restriction endonuclease Pqu in a total volume of 50 pl at 37°C for >2 hours to
allow for complete digestion. There is one Pvull restriction site in the 23S gene
of all seven rRNA operons, and one in the ITS region of rrnA, D, and H. A
second restriction site lies upstream of the rRNA operons, giving each 168-
containing restriction fragment a different, distinctive size. Each of the 168-
containing fragments represents an rRNA operon and can be resolved by
electrophoresis on an agarose gel (0.6% agarose in 0.5x TBE buffer). By using
the E. coli genome sequence to find each unique upstream restriction site, the
expected fragment size of each rRNA operon was determined and used for
positive identification of each operon.

The separated DNA fragments were transferred to a Magnacharge nylon
transfer membrane (Micron Separations Inc.; Westborough, MA) using the
VacuGene XL Vacuum Blotting System (Amersham Pharmacia Biotech;
Piscataway, NJ). The vacuum blotting was performed with a vacuum of 50 mbar,
using the protocol provided with the vacuum blotting system. The DNA was then
UV cross-linked to the membrane using the Stratlinker UV Crosslinker, model
1800 (Stratagene; La Jolla, CA) and the autocrosslink function (120 mjoules/cm2

from a 254 nm light source).

138

 

l--m-—w
A. ‘ .
U .
‘ .

 

Hybridization of the probe to the target DNA, post-hybridization washing of
the membrane, and detection of the probe was all performed according to the
protocols provided with the DIG High Prime labeling and detection kit (Boehringer
Mannheim Corporation (BMB); Indianapolis, IN). The probe used to hybridize the
168-containing fragments was synthesized by PCR-amplification of E. coli16$
DNA with the universal Eubacterial primers 8 Forward (5’-
AGAGTTTGATCCTGGCTCAG-3’) and 519 Reverse (5’-
GTA‘I‘I'ACCGCGGCTGCTGG-3’), resulting in a DNA fragment that is
homologous with a conserved region (positions 8-356) of the E. coli168 rDNA
gene. This probe was labeled with digoxigenen-dUTP following the protocol
provided with the DIG High Prime labeling and detection kit. After 12-16 hours of
hybridization, post-hybridization washes, and the addition of the detection
(CSPD) reagent, the membrane was wrapped in Saran Wrap and exposed to
Lumi-Film Chemiluminescent Detection Film (BMB) for 30-60 min. The individual

rRNA operons, if present, were visualized after the film was developed.

139

 

APPENDIX C

STATISTICAL ANALYSES OF RELATIVE FITNESS VALUES IN BATCH

CULTURE COMPETITION EXPERIMENTS

The differences between mean relative fitness values of each strain were
analyzed for significance using a single factor ANOVA and planned comparisons

among means (73). The ANOVA showed a significant difference among the

 

mean relative fitness values (F333 = 11.18, P = 5.35x10’5). The planned
comparisons showed which differences were significant, and were based upon
predetermined hypotheses about the effect of rRNA operon deletions on relative
fitness in batch culture competition experiments. Table 0.1 shows the

comparisons, the F statistic, and whether the difference can be considered

 

 

 

significant.
Table C.1
Comparison F124 Significance“
Control vs. ArrnA and ArrnB 6.06 ***
Control vs. ArrnAB 32.46 11*
ArrnA and ArmB vs. ArrnAB 16.95 ***
ArrnA vs. ArrnB 0.05 NS
ArmA vs. ArrnAB 13.54 m
ArmB vs. ArrnAB 1 1.92 11*

 

a *** indicates P < 0.025, NS indicates not significant or P > 0.1

140

Escherichia coli grown in phosphate-limited chemostats. J Mol Biol.
86(1 ): 1 -9.

Amlkam, D., G. Glaser, and S. Razin. 1984. Mycoplasmas (Mollicutes)

REFERENCES
Alton, T. H., and A. L. Koch. 1974. Unused protein synthetic capacity of
have a low number of rRNA genes. J Bacteriol. 158(1):376-8.
1

"5

Ammons, D., J. Rampersad, and G. E. Fox. 1999. SS rRNA gene
deletions cause an unexpectedly high fitness loss in Escherichia coli.
Nucleic Acids Res. 27(2):637-42.

Elna—w

Anderson, P., and J. Roth. 1981. Spontaneous tandem genetic
duplications in Salmonella typhimurium arise by unequal recombination
between rRNA (rm) cistrons. Proc Natl Acad Sci U S A. 78(5):3113-7.

Asal, T., C. Condon, J. Voulgaris, D. Zaporojets, B. Shen, M. AI-Omar,
C. Squires, and C. L. Squires. 1999. Construction and initial
characterization of Escherichia coli strains with few or no intact
chromosomal rRNA operons. J Bacteriol. 181 (12):3803-9.

Asal, T., D. Zaporojets, C. Squires, and C. L. Squires. 1999. An
Escherichia coli strain with all chromosomal rRNA operons inactivated:
complete exchange of rRNA genes between bacteria [see comments].
Proc Natl Acad Sci U S A. 96(5):1971-6.

Bachellier, S., E. Gilson, M. Hofnung, and C. W. Hill. 1996. Repeated
sequences, p. 2015-7. In F. C. Neidhardt, R. C. III, J. L. lngraham, E. C.
C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter,
and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium:
Cellular and Molecular Biology, vol. 2. ASM Press, Washington, DC.

Baracchini, E., and H. Bremer. 1991. Control of rRNA synthesis in
Escherichia coli at increased rm gene dosage. Role of guanosine
tetraphosphate and ribosome feedback. J Biol Chem. 266(18):11753—60.

Begon, M., J. L. Harper, and C. R. Townsend. 1996. Ecology. 3rd ed.
Blackwell Sciences International, Oxford, UK.

141

 

 

10.

11.

12.

13.

14.

15.

16.

17.

18.

Blomfield, I. C., V. Vaughn, R. F. Rest, and B. I. Eisenstein. 1991.
Allelic exchange in Escherichia coli using the Bacillus subtilis sacB gene
and a temperature-sensitive pSC101 replicon. Mol Microbiol. 5(6):1447-
57.

Bremer, H., and P. P. Dennis. 1996. Modulation of chemical composition,
p. 1561-1562. In F. C. Neidhardt, R. C. III, J. L. lngraham, E. C. C. Lin, K.
B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H.
E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: Cellular
and Molecular Biology, 2 ed, vol. 2. ASM Press, Wahsington, DC.

Brosius, J., A. Ullrich, M. A. Raker, A. Gray, T. J. Dull, R. R. Gutell,
and H. F. Noller. 1981. Construction and fine mapping of recombinant
plasmids containing the rrnB ribosomal RNA operon in Escherichia coli.
Plasmid. 6:112-118.

Bruce, J. L., R. J. Hubner, E. M. Cole, C. I. McDowell, and J. A.
Webster. 1995. Sets of EcoRI fragments containing ribosomal RNA
sequences are conserved among different strains of Listeria
monocytogenes [published erratum appears in Proc Natl Acad Sci U S A
1995 Oct 24;92(22):10441]. Proc Natl Acad Sci U S A. 92(11):5229-33.

Button, D. K. 1985. Kinetics of nutrient-limited transport and microbial
growth. Microbiol Rev. 49(3):270-97.

Carlton, B. C., and B. J. Brown. 1981. Gene Mutation, p. 222-242. In P.
Gerhardt (ed.), Manual of methods for general bacteriology. American
Society for Mirobiology, Washington, DC.

Condon, C., S. French, C. Squires, and C. L. Squires. 1993. Depletion
of functional ribosomal RNA operons in Escherichia coli causes increased
expression of the remaining intact copies. Embo J. 12(11):4305-15.

Condon, C., D. Liveris, C. Squires, I. Schwartz, and C. L. Squires.
1995. rRNA operon multiplicity in Escherichia coli and the physiological
implications of rm inactivation. J Bacteriol. 177(14):4152-6.

Condon, C., J. Philips, 2. Y. Fu, C. Squires, and C. L. Squires. 1992.
Comparison of the expression of the seven ribosomal RNA operons in
Escherichia coli. Embo J. 11(11):4175-85.

142

[2‘
I“

 

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

Condon, C., C. Squires, and C. L. Squires. 1995. Control of rRNA
transcription in Escherichia coli. Microbiol Rev. 59(4):623-45.

Daniels, L., R. S. Hanson, and J. A. Phillips. 1994. Chemical Analysis,
p. 512-554. In P. Gerhardt (ed.), Methods for General and Molecular
Bacteriology. American Society for Microbiology. Wahsington, D. C.

Dong, H., L. Nilsson, and C. G. Kurland. 1996. Co-variation of tRNA
abundance and codon usage in Escherichia coli at different growth rates.
J Mol Biol. 260(5):649-63.

Ellwood, M., and M. Nomura. 1982. Chromosomal locations of the genes
for rRNA in Escherichia coli K-12. J Bacteriol. 149(2):458-68.

Ellwood, M., and M. Nomura. 1980. Deletion of a ribosomal ribonucleic
acid operon in Escherichia coli. J Bacteriol. 143(2):1077-80.

Essener, A. A., J. A. Roels, and N. W. F. Kossen. 1983. Theory and
application of unstructured growth models: Kinetic and energetic aspects.
Biotechnol. Bioeng. 25:2803-2841.

Fegatella, F., J. le, S. Kjelleberg, and R. Cavlcchloli. 1998.
Implications of rRNA operon copy number and ribosome content in the
marine oligotrophic ultramicrobacterium Sphingomonas sp. strain RB2256.
Appl Environ Microbiol. 64(11):4433-8.

Fleischmann, R. D., M. D. Adams, 0. White, R. A. Clayton, E. F.
Kirkness, A. R. Kerlavage, C. J. Bult, J. F. Tomb, B. A. Daugherty, J.
M. Merrick, and et al. 1995. Whole-genome random sequencing and
assembly of Haemophilus influenzae Rd [see comments]. Science.
269(5223):496-51 2.

Fogel, G. B., C. R. Collins, J. Li, and C. F. Brunk. 1999. Prokaryotic
genome size and SSU rDNA copy number: Estimation of microbial relative
abundance from a mixed population. Microb Ecol. 38(2):93-113.

Garrett, R. A., J. Dalgaard, N. Larsen, J. Kjems, and A. S. Mankin.
1991. Archaeal rRNA operons. Trends Biochem Sci. 16(1):22-6.

143

29.

30.

31.

32.

33.

34.

35.

36.

37.

Gay, P., D. Le Coq, M. Steinmetz, T. Berkelman, and C. I. Kado. 1985.
Positive selection procedure for entrapment of insertion sequence
elements in gram-negative bacteria. J Bacteriol. 164(2):918-21.

Gay, P., D. Le Coq, M. Steinmetz, E. Ferrari, and J. A. Hoch. 1983.
Cloning structural gene sacB, which codes for exoenzyme levansucrase of
Bacillus subtilis: expression of the gene in Escherichia coli. J Bacteriol.
153(3): 1 424-31 .

Gonzalez-y—Merchand, J. A., M. J. Garcia, S. Gonzalez-Rico, M. J.
Colston, and R. A. Cox. 1997. Strategies used by pathogenic and
nonpathogenic mycobacteria to synthesize rRNA. J Bacteriol.
179(22):6949-58.

Gourse, R. L., T. Gaal, M. S. Bartlett, J. A. Appleman, and W. Ross.
1996. rRNA transcription and growth rate-dependent regulation of
ribosome synthesis in Escherichia coli. Annu Rev Microbiol. 50:645-77.

Gourse, R. L., and M. Nomura. 1984. Level of rRNA, not tRNA, synthesis
controls transcription of rRNA and tRNA operons in Escherichia coli. J
Bacteriol. 160(3): 1 022-6.

Gourse, R. L., Y. Takebe, R. A. Sharrock, and M. Nomura. 1985.
Feedback regulation of rRNA and tRNA synthesis and accumulation of
free ribosomes after conditional expression of rRNA genes. Proc Natl
Acad Sci U S A. 82(4):1069-73.

Grover, J. P. 1991. Resource competition in a variable environment:
Phytoplankton growing according to the variable-internal-stores model.
Am. Nat. 138(4):811-835.

Hamilton, C. M., M. Aldea, B. K. Washburn, P. Babitzke, and S. R.
Kushner. 1989. New method for generating deletions and gene
replacements in Escherichia coli. J Bacteriol. 171 (9):4617-22.

Hansen, 8. R., and S. P. Hubbell. 1980. Single-nutrient microbial
competition: qualitative agreement between experimental and theoretically
forecast outcomes. Science. 207(4438):1491-3.

144

 

38.

39.

40.

41.

42.

43.

44.

45.

46.

Herzer, P. J., S. Inouye, M. Inouye, and T. S. Whittam. 1990.
Phylogenetic distribution of branched RNA-linked multicopy single-
stranded DNA among natural isolates of Escherichia coli. J Bacteriol.
172(11):6175-81.

High Performance Systems, I. 1997. Stella Technical Documentation.
High Performance Systems, Inc., Hanover, NH.

Hill, C. W., S. Harvey, and J. A. Gray. 1990. Recombination between
rRNA genes in Escherichia coli and Salmonella typhimurium, p. 335-340.
In K. Drlica and M. Riley (ed.), The Bacterial Chromosome. American
Society for Microbiology, Washington, DC.

Jinks-Robertson, S., R. L. Gourse, and M. Nomura. 1983. Expression
of rRNA and tRNA genes in Escherichia coli. evidence for feedback
regulation by products of rRNA operons. Cell. 33(3):865-76.

Jinks-Robertson, S., and M. Nomura. 1987. Ribosomes and tRNA, p.
1358-1385. In F. C. Neidhardt, J. L. lngraham, K. B. Low, B. Magasanik,
M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella
typhimurium: Cellular and Molecular Biology, 2 ed, vol. 2. American
Society for Microbiology Press, Washington DC.

Khetawat, G., R. K. Bhadra, S. Nandi, and J. Das. 1999. Resurgent
Vibrio cholerae O139: rearrangement of cholera toxin genetic elements
and amplification of rm operon. Infect Immun. 67(1):148-54.

Kingston, R. E., R. R. Gutell, A. R. Taylor, and M. J. Chamberlin. 1981.
Transcriptional mapping of plasmid pKK3535: quantitation of the effect of
guanosine tetraphsophate on binding to the rmB promoters and a A
promoter with sequence homologies to the Cll binding region. J Mol Biol.

1 46:433-449.

Kiss, A., B. Sain, and P. Venetianer. 1977. The number of rRNA genes
in Escherichia coli. FEBS Lett. 79(1):77-9.

Klappenbach, J. A., J. M. Dunbar, and T. M. Schmidt. 2000. Ribosomal
RNA operon copy number reflects ecological strategies of bacteria. Appl
Environ Microbiol. 66(4).

145

 

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

Koch, A. L. 1971. The adaptive responses of Escherichia coli to a feast
and famine existence. Adv Microb Physiol. 6:147-217.

Koch, A. L. 1982. Multistep kinetics: choice of models for the growth of
bacteria. J Theor Biol. 98(3):401-17.

Kovarova-Kovar, K., and T. Egli. 1998. Growth kinetics of suspended
microbial cells: from single-substrate- controlled growth to mixed-substrate
kinetics. Microbiol Mol Biol Rev. 62(3):646-66.

Kundig, C., H. Hennecke, and M. Gottfert. 1993. Correlated physical
and genetic map of the Bradyrhizobium japonicum 110 genome. J
Bacteriol. 175(3):613-22.

Lederberg, S. 1966. Genetics of host-controlled restriction and
modification of dioxyribonulceic acid in Escherichia coli. J Bacteriol.
91 (3):1029-1036.

Lenski, R. E. 1988. Experimental studies of pleiotropy and epistasis in
Escherichia call. I. Variation in competitive fitness among mutants
resistant to virus T4. Evolution. 42:433-440.

Lenski, R. E., J. A. Mongold, P. D. Sniegowski, M. Travisano, F. Vasi,
P. J. Gerrish, and T. M. Schmidt. 1998. Evolution of competitive fitness
in experimental populations of E. coli: what makes one genotype a better
competitor than another? Antonie Van Leeuwenhoek. 73(1):35—47.

Lenski, R. E., M. R. Rose, S. C. Simpson, and S. C. Tadler. 1991. Long-
term experimental evolution in Escherichia call. I. Adaptation and
divergence during 2,000 generations. Am Nat. 138(6):1315-1341.

Lenski, R. E., V. Souza, L. P. Duong, Q. G. Phan, T. N. M. Nguyen, and
K. P. Bertrand. 1994. Epistatic effects of promoter and repressor
functions of the Tn10 tetracycline-resistance operon on the fitness of
Escherichia coli. Mol Ecol. 3:127-135.

Lenski, R. E., and M. Travisano. 1994. Dynamics of adaptation and
diversification: a 10,000-generation experiment with bacterial populations.
Proc Natl Acad Sci U S A. 91 (15):6808-14.

146

 

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

Levin, B. R., F. M. Stewart, and L. Chao. 1977. Resource-limited growth,
competition, and predation: a model and experimental studies with
bacteria and bacteriophage. Am Nat. 111:3-24.

Milkman, R. 1996. Recombinational exchange among clonal populations,
p. 2663-2684. In F. C. Neidhardt, R. C. III, J. L. lngraham, E. C. C. Lin, K.
B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H.
E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: Cellular
and Molecular Biology, 2 ed, vol. 2. ASM Press, Wahsington, DC.

Miller, J. H. 1972. Experiments in Molecular Genetics, p. 201-205. Cold
Spring Harbor Laboratory, Cold Spring Harbor, NY.

Monod, J. 1949. The growth of bacterial cultures. Ann Rev Microbiol.
3:371-394.

Morgan, E. A., T. Ikemura, L. Lindahl, A. M. Fallon, and M. Nomura.
1978. Some rRNA operons in E. coli have tRNA genes at their distal ends.
Cell. 13(2):335-44.

Morgan, E. A., T. Ikemura, and M. Nomura. 1977. Identification of
spacer tRNA genes in individual ribosomal RNA transcription units of
Escherichia coli. Proc Natl Acad Sci U S A. 74(7):2710-4.

Poulsen, L. K., T. R. Licht, C. Rang, K. A. Krogfelt, and S. Molin. 1995.
Physiological state of Escherichia coli BJ4 growing in the large intestines
of streptomycin-treated mice. J Bacteriol. 177(20):5840-5.

Provence, D. L., and R. C. III. 1994. Gene transfer in Gram-negative
bacteria, p. 332-334. In P. Gerhardt, R. G. E. Murray, W. A. Wood, and N.
R. Kreig (ed.), Methods for General and Molecular Bacteriology. ASM
Press, Washington, DC.

Roels, J. A., and N. W. F. Kossen. 1978. On the modeling of microbial
metabolism. Prog lndust Microbiol. 14:95-203.

Russell, C. B., D. S. Thaler, and F. W. Dahlquist. 1989. Chromosomal
transformation of Escherichia coli recD strains with linearized plasmids. J
Bacteriol. 171 (5):2609-13.

147

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

Sambrook, J. 1989. Appendix A: Bacterial media, antibiotics, and
bacterial strains, p. A.1-A.6. In E. F. Fritsch and T. Maniatis (ed.),
Molecular Cloning: A Laboratory Manual, 2 ed, vol. 3. Cold Spring Harbor
Laboratory Press, Cold Spring Harbor, NY.

Sarmientos, P., J. E. Sylvester, S. Contente, and M. Cashel. 1983.
Differential stringent control of the tandem E. coli ribosomal RNA
promoters from the rrnA operon expressed in vivo in multicopy plasmids.
Cell. 32(4): 1 337-46.

Schmidt, T. M. 1997. Multiplicity of Ribosomal RNA Operons in
Prokaryotic Genomes, p. 221-229. In F. J. deBruijn, J. R. Lupski, and G.
M. Weinstock (ed.), Bacterial Genomes: Physical Structure and Analysis.
International Thomson Publishing, New York.

Senn, H., U. Lendenmann, M. Snozzl, G. Hamer, and T. Egli. 1994. The
growth of Escherichia coli in glucose-limited chemostat cultures: a re-
examination of the kinetics. Biochim Biophys Acta. 1201 (3):424-36.

Sharp, P. M., J. E. Kelleher, A. S. Daniel, G. M. Cowan, and N. E.
Murray. 1992. Roles of selection and recombination in the evolution of
type I restriction-modification systems in enterobacteria. Proc Natl Acad
Sci U S A. 89(20):9836-40.

Snyder, L. 1998. Phage P1 strain and transduction protocol.

Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principles and practice
of statistics in biological research, 3 ed. W. H. Freeman and Company,
New York.

Stevenson, B. S., and T. M. Schmidt. 1998. Growth rate-dependent
accumulation of RNA from plasmid-borne rRNA operons in Escherichia
coli. J Bacteriol. 180(7):1970-2.

Tilman, D. 1977. Resource competition between planktonic algae: an
experimental and theoretical approach. Ecology. 58:338-348.

Vasi, F., M. Travisano, and R. E. Lenski. 1994. Long-term experimental
evolution in Escherichia coli. ll. Changes in life-history traits during
adaptation to a seasonal environment. Am Nat. 144(3):432-456.

148

77.

78.

Wilkinson, S. R., D. I. Young, J. G. Morris, and M. Young. 1995.
Molecular genetics and the initiation of solventogenesis in Clostridium
beijerinckii (formerly Clostridium acetobufylicum) NCIMB 8052. FEMS
Microbiol Rev. 17(3):275-85.

Wolfe, C. J., and M. G. Haygood. 1993. Bioluminescent symbionts of the
Caribbean flashlight fish (Kryptophanaron alfredi) have a single rRNA
operon. Mol Mar Biol Biotechnol. 2(4):189-97.

149