trfifl.‘ .V... . .« . . . vv ‘.

 

       

     

        

 

.§.. .3:

 

 

 

 

 

 

       

 

 

 

 

 

     

 
 

 

 

 

 
 

 

 
 
 

 
 

 

 

 
 
 
 
 
 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

dissertation entitled
THE ECOLOGY 0F SPIROCHETES IN METHANOGENIC

BIOREACTORS

presented by

Sherry Lynn Dollhopf

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Microbiology

 

 

5, 3001
I I

 

Dat

MS U is an Affirmative 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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDuo.p65—p. 15

 

THE ECOLOGY OF SPIROCHETES IN METHANOGENIC BIOREACTORS
By

SHERRY LYNN DOLLHOPF

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Doctor of Philos0phy

Microbiology and Molecular Genetics

2001

ABSTRACT
THE ECOLOGY OF SPIROCHETES IN METHANOGENIC BIOREACTORS
By

SHERRY LYNN DOLLHOPF

Methane is a potent greenhouse gas, and the release of methane from ruminant
animals, termites, waste treatment and anaerobic sediments plays a significant role in
global warming models. Spirochetes occur in many methanogenic environments
including termite hindguts, rumen contents, dental plaque, human gastrointestinal and
reproductive tracts, and anoxic sediments. Many spirochete species of medical and
environmental importance are difficult to cultivate and study in vitro. To investigate the
ecology of spirochetes in methanogenic environments, the unusual enrichment of
spirochete species in laboratory scale glucose-fed methanogenic bioreactors was studied
using 16S rRNA and rDNA based molecular techniques, pure culture, and coculture
experiments. The bacteria and archaea communities associated with the spirochetes were
also described.

The enrichment of Spirochaeta and T reponema related species was observed in
multiple bioreactors Operated at low diliution rates (0.06 to 0.17 days") inoculated with
freshwater river sediment, sewage sludge, and rumen contents. Spirochaeta sp. str. R8
was isolated from bioreactor fluid and was shown to compose 25 to 45 % of the
bioreactor community by fluorescent in situ hybridization. Strain R8 fermented glucose
to acetate, ethanol, lactate, hydrogen, and carbon dioxide. These products were formed

and utilized in bioreactor fluid, indicating that the spirochetes played an important role in

the carbon flow within the bioreactors. At dilution rates of 0.25 days'1 and above, the
Spirochaeta sp. str. R8 population decreased, while a population of Clostridium ramosum
related organisms increased. Competition experiments between Spirochaeta sp. str. R8
and Clostridium ramosum str. S9 in continuous culture demonstrated the strain R8 was a
superior competitor for glucose at low dilution rates, provided that a sufficient supply of
coenzyme A precursors were supplied in the medium. Without added coenzyme A
precursors, strain S9 and R8 coexisted in approximately equal numbers at all dilution
rates tested. Coculture of Methanobacterium bryantii or Methanosarcina mazeii with
Spirochaeta sp. str. R8 relieved the coenzyme A requirement and increased cells yields of
strain R8 significantly. Competition of strains R8 and S9 in the presence of
Methanobacterium bryantii allowed strain R8 to outcompete S9 in the absence of extra
coenzyme A. This dissertation demonstrates the importance of cooperative interactions in

the function and structure of microbial communities.

Dedication

To Mike, for being there during each crisis
I have encountered and created in our life together.
Life would not be the same without you.

iv

 

ACKNOWLEDGMENTS

My thanks to everyone, friends and family, that has helped me through each
flaming hoop and unexpected result of my thesis. I am especially thankful to my lab-
mates and classmates (Merry, Denise, Frank, Carol, Ben, Dan, Joe, and others) who have
been there for serious discussions, good times, and everything in-between. Many thanks
also to the members of my committee for being accessible and generous with their
excellent guidance and advice. Lastly, I would like to thank Dr. Tiedje for giving me the
opportunity to be a part of his lab and the Center for Microbial Ecology. Without his
vision and support I would not have had the opportunity to attend numerous conferences
and summer classes, to interact with multi-disciplinary groups, and to pursue my own
research interests. Many graduate students do not have the same opportunities that I have

had, and I am grateful for it. Above all else, I am grateful for his tremendous faith in my

abilities, that in itself has meant the most.

MST

LIST

Clix

 

 

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. ix
LIST OF FIGURES ............................................................................................................ x
CHAPTER 1 — COMPETITION AND COOPERATION IN MICROBIAL COMMUNTTIES ........... 1
Introduction .............................................................................................................. 1
Types of microbial species interactions ..................................................................... 2
Negative interactions and community structure .................................................... 4
Positive interactions and community structure ..................................................... 8
Investigating microbial species interactions ............................................................ 12
Enrichment culture ........................................................................................... 12
Coculture .......................................................................................................... l3
Microsc0py ....................................................................................................... 15
Molecular methods ............................................................................................ 15
Implications ............................................................................................................ 1 9
Thesis overview ..................................................................................................... 20
Experimental design ............................................................................................... 23
References .............................................................................................................. 24
CHAPTER 2 - PHYLOGENCY OF BIOREACI’ OR ORGANISMS .......................................... 30
Introduction ............................................................................................................ 30
Materials & Methods .............................................................................................. 32
Reactor design and operation ............................................................................ 32
Extraction and purification of community DNA ................................................ 35
168 rDNA clone libraries .................................................................................. 36
Bacterial isolates ............................................................................................... 37
Sequencing and sequence analysis .................................................................... 38
16S rRNA oligonucleotide probing ................................................................... 39
Results ................................................................................................................... 42
Bacteria ............................................................................................................ 42
Archaea ............................................................................................................ 49
Clone libraries ............................................................................................. 49

vi

 

16S rRNA oligonucleotide probes ............................................................... 49

Discussion .............................................................................................................. 5 1
References .............................................................................................................. 55

CHAPTER 3 - THE IMPACT OF FERMENTATIVE ORGANISMS ON CARBON FLOW IN

METHANOGENIC SYSTEMS UNDER CONSTANT Low SUBSTRATE CONDITIONS ........... 58
Introduction ............................................................................................................ 58
Materials & Methods .............................................................................................. 60

Bioreactor operation ......................................................................................... 60
Sample preparation and microscopy .................................................................. 60
DNA extraction and T-RF LP ............................................................................ 61
16S rDNA analysis ........................................................................................... 62
Determination of intermediary metabolites ....................................................... 63
Spirochete isolate R8 ........................................................................................ 63
Fermentation products ...................................................................................... 64
Maximum substrate utilization rates .................................................................. 64
Nucleotide sequence accession numbers ........................................................... 65
Results ................................................................................................................... 65
Community structure ........................................................................................ 6S
Intermediary metabolites .................................................................................. 71
Maximum substrate utilization rates .................................................................. 72
Spirochete isolation and fermentation products ................................................. 75
Discussion .............................................................................................................. 75
References .............................................................................................................. 80

CHAPTER 4 - INTERPRETATION OF 168 RDNA T-RFLP DATA: APPLICATION OF SELF

ORGANIZING MAP NEURAL NETWORKS AND PRINCIPLE COMPONENT ANALYSIS ....... 84
Introduction ............................................................................................................. 84
Materials & Methods ............................................................................................... 86

Reactor operation and experimental design ........................................................ 86
Sampling ........................................................................................................... 88
T-RFLP profiles ................................................................................................ 88
Clone libraries and sequencing .......................................................................... 89
Community similarity ........................................................................................ 90
PCA ................................................................................................................... 91
SOM .................................................................................................................. 91

vii

 

Results .................................................................................................................... 92

Community similarity ........................................................................................ 92
PCA ................................................................................................................... 94
SOMs ................................................................................................................ 98
Terminal restriction fragment identification ..................................................... 102
Discussion ............................................................................................................. 108
References ............................................................................................................. 1 13

CHAPTER 5 — INTERSPECIES INTERACTIONS AFFECTING A SPIROCHAETA SPECIES IN A

CARBON-LIMITED METHANOGENIC ENVIRONMENT ................................................... 117
Introduction ........................................................................................................... 1 17
Materials & Methods ............................................................................................. 118

Bioreactor operation ........................................................................................ 1 18
T-RFLP analysis .............................................................................................. 119
FISH ................................................................................................................ 119
Isolation and culture techniques ....................................................................... 122
16S rRNA gene sequencing ............................................................................. 123
Continuous culture ........................................................................................... 123
Results .................................................................................................................. 124
T -RFLP ........................................................................................................... 124
FISH ................................................................................................................ 127
Strain R8 and S9 characteristics ....................................................................... 130
Competition in continuous culture ................................................................... 132
Methanogenic coculture ................................................................................... 135
Discussion ............................................................................................................. 138
References ............................................................................................................. 143
CHAPTER 6 - CONCLUSIONS ....................................................................................... 146

viii

 

 

LIST OF TABLES

Table 1.1. Interactions among microbial species. ............................................................. 3
Table 1.2. Basis of cooperative interactions. .................................................................... 9

Table 1.3. The purpose and potential application of molecular microbial ecology methods
for the study of interspecies interactions. ....................................................................... 18

Table 2.1. Phyla detected by 16S rRNA gene molecular methods in various reactor
communities .................................................................................................................. 43

Table 3.1. Similarity of bacterial HaeIII, HhaI, and Mspl T-RF LP electropherograms as
measured by Horn’s community similarity index (Brower 1997) and Pearson product-
moment correlation coefficient (Pearson 1926) .............................................................. 68

Table 3.2. Production of intermediary metabolites in bioreactor HS and LS as determined
by addition of [II-14C] glucose to the bioreactors. ........................................................... 73

Table 3.3. Fermentation of glucose by isolate R8 and maximum substrate utilization rates
in bioreactor R1 ............................................................................................................. 74

Table 4.1. Morisita community similarity indices for the R1, 82, and M3 bioreactor

communities. ................................................................................................................. 93
Table 4.2. T-RFLP fragments with high factor loading. ................................................. 97
Table 5.1. Growth yields in glucose minimal medium. ................................................ 131

Table 5.2 Growth yields of Spirochaeta strain R8 in coculture with Methanobacterium
bryantii. ...................................................................................................................... 136

ix

 

 

LIST OF FIGURES

Figure 1.1. u — 3 relationship of a copiotroph (A), an organism with high and low affinity
uptake systems (B), and an oligotrOph (C). ...................................................................... 6

Figure 1.2. Flow chart of molecular microbial ecology methods for community analysis
and monitoring of specific phylogenetic groups of microorganisms. RT-PCR = Reverse

transcription polymerase chain reaction. ........................................................................ 16
Figure 1.3. Bioreactor design ......................................................................................... 21

Figure 1.4. Functional groups of microorganisms in glucose-fed methanogenic

bioreactors. ................................................................................................................... 22

Figure 2.1. Reactor operation and sampling. Dotted horizontal arrows indicate
inoculation and solid horizontal arrows indicate reactor operation. Black and gray circles
indicate sampling times for 16S rRNA gene clone libraries and bacterial isolation,

respectively. Vertical arrows indicate an intentional organic shock load. (Facing) ......... 33

Figure 2.2. Maximum likelihood tree of cloned bacterial 16S rRNA gene sequences from
bioreactors and other methanogenic environments. The backbone of this tree was cast
using 696 unambiguously aligned nucleotides between E. coli positions 95 and 1295.
Partial sequences were then appended to the tree using the maximum likelihood
algorithm with 261 unambiguously aligned nucleotides between E. coli positions 154 and
474. The origin of the cloned genes is as follows: RFS == Termite gut (Lilburn); WCHBl
= Aquifer (Dojka); E0 = Reactor P1; HS = Reactors HS 5-8; LS = Reactors LS 1-4; 82 =

Reactor 82; str. = Bacterial isolate from indicated reactor. (Facing) ............................... 40

Figure 2.3. Maximum likelihood tree of cloned spirochete-related l6S rRNA genes from
bioreactors and other methanogenic environments. This tree was calculated in ARB using
312 unambiguousy aligned nucleotides between E. coli positions 154 and 478. The origin
of the cloned genes is indicated as follows: RFS == Termite gut (Lilburn); WCHBI ==

Aqui

Pi

.
wk

met.

Aquifer (Dojka); E0 = Reactor P1; HS = Reactors HS 5-8; LS= Reactors LS 1-4; 82 =
Reactor 82; str. = Bacterial isolate from the indicated reactor. ....................................... 45

Figure 2.4. Neighbor-joining tree of Archaea l6S rDNA clones created from a Jukes-
Cantor distance matrix. The matrix was calculated from analysis of 328 unambiguously
aligned nucleotides of the 16S rRNA gene. Bootstrap values from 100 parsimony

replicates are shown to the left of each node. ................................................................. 48

Figure 2.5. Relative abundance of methanogen-related 16S rRNA in reactors LS 1 and 2
and HS 6 and 7 as determined by 16S rRNA membrane hybridization to oligonucleotide
probes. The abundance of each group is expressed as a percentage of the total amount of

methanogen-related 16S rRNA detected by all of the methanogen-specific probes used. 50

Figure 3.1. T-RFLP electropherograms of the HaeIII digested SSU rDNA amplified from
isolate R8 and bacterial communities LS, HS, and R1. Peaks corresponding to the
restriction fragment length of the most frequently cloned SSU rRNA gene sequences are
labeled. ......................................................................................................................... 66

Figure 3.2. Phylogenetic relationships of clones and isolates in the families
Spirochaetales (A) or Streptococcales (B). Unambiguously aligned nucleotides of the
SSU rRNA gene from E. coli position 111 to 465 (A) and 21 to 1477 (B) were used to
calculate Kimura 2-parameter distance matrices (Kimura 1980). The phylogenetic trees
were created with the neighbor-joining method (Saitou, Nei 1989). Neighbor-joining and
parsimony bootstrap values (50 % or higher) are listed above and below each node,
respectively. Scale bar = 10 % difference in nucleotide sequence positions. Genbank
accession numbers are as follows: Streptococcus bovis, M58835; Streptococcus
macedom‘cus, 294012; Streptococcus mutans, 870358; Streptococcus equi, $70324;
Bacillus subtilis str 168, D26185; Lactobacillus acidophilus subspp. johnsonir’, M99704;
Lactococcus lactis subspp. Lactis, M58837; Streptococcus salivarius, M58839;
Streptococcus parauberis, X89967 ; Brachyspira hyodysenteriae, M57741; T reponema
btyantii, M57737; Spirochaeta zuelzerae, M88725; Spirochaeta sp. str. Mastotermes,
X79548; Spirochaeta caldaria, M71240; Spirochaeta stenostrepta, M88724; Spirochaeta

xi

Pig.
3} a

lfll

aurantia str. J-I, M57740; Spirochaeta litoralis, M88723; Spirochaeta halophila,
M88722; Spirochaeta Sp. str. PLIZMY, RDP short II) str.PL12MY; Spirochaeta sp. str.
Antarctica, M87055; Leptospira inadai, 221634; EO-XIV, AF 149879; EO-XI,
AF 149878; EO-V, 149881; EO-I, AF 149883 (Fernandez et a1. 1999); LSII, AF218216;
and HSI, AF218221 (Fernandez et a1. 2000). (Facing) ................................................... 69

Figure 4.1. Timeline for reactors R1, 82, and M3. Time is measured in number of volume
changes (x-axis not to scale). Sampling times are indicated by numbered circles, which
show the number of volume changes that occurred since the target organic loading rate
was achieved in reactors $2 and M3. Control reactor R1 operated for over 2 years prior to
this study. Reactors 82 and M3 were inoculated from river sediment and municipal
sewage sludge, respectively. Flow rate and HRT are shown above and below the timeline

respectively. .................................................................................................................. 87

Figure 4.2. PCA ordination of Bacteria T—RFLP peak height data for reactors M3 (A and
B) and 82 (C and D) during phase A (A and C) and Phase B (B and D). The amount of
variability accounted for by each factor is on the axes in each panel. Sampling times are

next to each point. ......................................................................................................... 95

Figure 4.3. SOM clustering of T-RFLP fragments in reactor M3 (A) and 82 (B). The line
graphs shown for each fragment represent the change in normalized peak height over
time. The graphs are arranged according to the position of their winning node within the
square SOM output. SOM groups I, II, III and IV are indicated. The letter A before the
fragment length indicates fi'agments from Archaea-specific T-RFLP. Fragment lengths
indicated in color are in the same SOM group in both reactors. (C) A timeline shows

when groups I through IV were present in each reactor. (Facing) ................................... 99

Figure 4.4. Maximum likelihood tree of cloned Bacteria 16S rRNA genes from reactors
M3 and $2 and other methanogenic environments. Clones retrieved in this study are in
bold. The SOM group and terminal restriction fragment length (Group x : y bp) is shown
to the left of each clone. The backbone of this tree was cast in ARB using 969
unambiguously aligned nucleotides between E. coli positions 95 and 1295. Partial

xii

seq-ll

sequences were appended to the tree in ARB using the maximum likelihood algorithm
with 261 unambiguously aligned nucleotides between E. coli positions 154 and 474. The
origin of cloned genes from other methanogenic environments is indicated as follows:
E0, LS, HS = glucose-fed methanogenic reactors (11,12); HA = anaerobic vinasses

reactor (15); RFS = termite hindgut (21); WCHBl = hydrocarbon contaminated aquifer
(9). (Facing) ................................................................................................................ 103

Figure 4.5. Succession of methanogenic phylotypes in reactor S2 (A) Neighbor-joining
tree of cloned Archaea l6S rDNA genes. The number next to each node indicates
bootstrap values from 100 parsimony bootstraps. Numbers to the right of each bracket
indicate the length of the 16S rDNA terminal restriction fiagment for each phylogenetic
group. (B) Archaea community T-RFLP profiles of reactors 82 and M3 after different
HRTs in phase A. ........................................................................................................ 107

Figure 5.1. 16S rRNA gene terminal restriction fragments generated with restriction
enzyme HaeIH and specificity of oligonucleotide probes used for in situ hybridization.
Except for the two organisms noted, all sequences in the database had more that two

mismatches with the oligonucleotide probes designed in this study. ............................ 121

Figure 5.2a. 16S rDNA T-RFLP electropherograms of bioreactor community M3 at

different dilution rates. ................................................................................................ 125

Figure 5.2b. Change in normalized peak height of terminal restriction fragments unique
for Spirochaeta species (closed symbols) and Clostridium ramosum (open symbols)
generated by restriction of amplified bacterial 16S rRNA genes with enzyme HaeIII from
bioreactor communities 82 ('I,E]) and M3 (0,0). The dilution rate applied is shown
along the top axis. ....................................................................................................... 126

Figure 5.3. Micrographs of bioreactor fluid after hybridization with fluorescently labeled
oligonucleotide probes. DAPI (left) and FISH (right) micrographs are shown for identical
fields. Bioreactor community M3 after ten volume changes at D = 0.17 days'1 hybridized
with probe SpiroR8-l77 (A and B). Bioreactor community SZ after eight volume changes

xiii

at D = 0.13 days'1 (C and D) and after seven volume changes at D = 0.50 days'1 (E and F)
hybridized with probe ClostS9-182. ............................................................................ 128

Figure 5.4. Percent of DAPI stained cells that hybridized with probes SpiroR8-l77
(closed symbols) and ClostS9-l82 (open symbols) in the $2 (I) and M3 (0) bioreactor

communities. The dilution rate applied is shown along the top axis. ............................ 129

Figure 5.5a. Population size of Spirochaeta sp. str. R8 (0,.) and Clostridium ramosum
str. S9 (0,0) during competition in continuous mixed culture with 500 ug/mL of
pantothenate supplied. The dashed line indicates the washout rate of a non-growing

population. The dilution rate applied is shown along the top axis ................................. 133

Figure 5.5b. Population size of Spirochaeta sp. str. R8 (6,.) and Clostridium ramosum
str. S9 (0,0) during competition in continuous mixed culture. Data is shown from two
separate competition experiments performed under identical conditions. The dilution rate

applied is shown along the top axis .............................................................................. 134

Figure 5.6. Population size of Spirochaeta sp. str. R8 (O) and Clostrr'dium ramosum str.
89(0) during competition in continuous mixed culture. The arrow indicates the addition
of 1.8 X 108 Methanobactertum bryantii cells to the chemostat. The dashed line indicates
the washout rate of a non-growing population. The dilution rate applied is shown along

the top axis. ................................................................................................................. 137

xiv

IN]

lntrc

6
l"
A ill

a r
liar

lo:

marl

ml
Err.

lli

sea?

0 l
is:

“rm
Mil)

CHAPTER 1

INTERSPECIES INTERACTIONS IN ANAEROBIC MICROBIAL COMMUNITIES

Introduction

Microbial communities affect many aspects of the world in which we live.
Microbial community function impacts human health, agriculture, waste treatment,
nutrient cycling, and global warming, to name a few. Aside from our innate desire to
learn about the world in which we live, our goal in studying microbial communities is to
control and direct microbial functions to preserve and improve the quality of human life.
To control and direct microbial activities, an understanding of the fundamental biological
mechanisms underlying their activities is necessary. The harnessing of the ability of
microbes to produce a variety of commercial products is an example of the successful
application of this reductionist approach.

The difficult task of elucidating the biological processes and interactions within
an entire microbial community may be necessary to understand many ecological and
health-related problems. Along with the typical environmental controls and biological
limitations of a single species that must be understood, multiple species, diverse
metabolisms, and interspecies interactions must be taken into account. In complex
microbial communities with multiple trophic levels and species numbers in the
thousands, this task becomes daunting. The advent of high-throughput molecular
technologies vastly improves our ability to describe and compare complex microbial

communities, which should aid us in gaining an understanding of microbial communities.

The
ammals shc
Ilche pant
play a majo
ecology has
cur unders

natural em'

Types of II

A p
tier micr
unival an
metabolism
completely
llf entire]
Hem, lme:
Elfin little
39365513 j

.\'e
llable 11
ill an m
Slides p1

zlereblal .

The same basic principles that govern the community structure of plants and
animals should apply to microbes as well. Thus we can expect competition for resources,
niche partitioning, trophic interactions, mutualism and other interspecies interactions to
play a major role in the structuring of microbial communities. Much research in microbial
ecology has shown that basic ecological principles apply to microbial communities, but
our understanding of how these principles actually work in complex communities in

natural environments is still in its infancy.

Types of microbial species interactions

A prokaryotic cell within a microbial community interacts with a large number of
other microorganisms simultaneously. The importance of these interactions for the
survival and activity of the cell depends upon the proximity of the other cells and their
metabolism. The nature of the interactions may be positive, negative, or neutral. A
completely neutral relationship between two species in the same environment is rare if
the environment they share allows diffusion of soluble and gaseous molecules between
them. Interspecies interactions range from relatively weak non-specific interactions that
exert little influence on the grth of either species, to strong interactions that are
necessary for the survival of one or both Species in a particular environment.

Negative species interactions include competition, parasitism, and predation
(Table 1.1). Parasitism and predation are common between prokaryotes and eukaryotes,
but are not as common between prokaryotic species. Only a few known prokaryote
species prey upon other prokaryotes. The importance of predatory cells in natural

microbial communities is unclear, but they are ubiquitous in natural environments.

coma-Fh-imvfl— Erhqthl

Am»? mun-5 h.— .ruaJ.J-nv.....:rd._ ..:ed.-.u jaw—:VAZ-uv
EccmuUa-y-nvdP-m TurianIJLuLth—p-m unimh-ediumrph nib—m,-uv.r..iduv Avg uvnvZ-I. Ink-Ltdu Funk-reunhurVNV .~ .~ hv~n~=fl~r

 

00830 3000 80 53> 0323.
30: 000 80303000 025 3023 8
823338 30305 ziwfi .30 095 360% < 3038an
080? 000800 2:00 0:0 0:30:
383 030333 0 008mg 80303000
033 £023 8 8030308 .30 099 300% < 82300383
8033800 :0>_w
23 000:: 03.0% 23 .30 0:0 300— 30
00.3 8030830 000 .303 03 038023

mm :23 >03 0 8 30888 8030—0000 03H Eva—0302
8033800 53m 05 00080
E08330 30: 83 .303 03 038023 3838059
mm :23 >03 0 E 308038 80303000 033. 8038000000305
00300.30
30: mm 305 00303000 005080 :33
80308038 80.3 330003 00303000 0:0 8m=0m=08800

003808 0330830380: 0 8 805080

880800 80 33:38 0030—0000 0:0 833803
.0808 03300380
0 8 05080 880800 80303000 000 8030305
000833 803333:
0 $300008 03 833 30050 A8m=0me08<v
@3058 3 8050 3000 3528 80303003 003300800 008080.303:
@8338: 000 80.580» 803300800 008805

8033 30.30 2000 33:38 8032009! 803300800 0>3§03xm

 

803830D 80,—.

 

.Amv 3:0 5 00000030.. 80.3 303695
80308038 80008038 830.88.. 003800 03 000: 08.03 008800 ._ ._ 033.

Viruses that prey on heterotrophic and phototrophic bacterial populations are also
common in aquatic environments and affect community structure and population
dynamics (2,55).

Positive, or cooperative, interactions between prokaryotic species include
commensal, syntrOphic, and symbiotic relationships (Table 1.1). Both symbiotic and
commensal interactions are common between prokaryotes and eukaryotes. Examples
include the symbiotic fixation of nitrogen in legumes and the commensal microflora of
mammalian gastrointestinal tracts. Among prokaryotes, cooperative relationships appear
to be more important in anaerobic than in aerobic environments. The oxidation of
compounds without highly oxidizing electron acceptors, such as oxygen or nitrate,
requires a large enzymatic machinery and detailed metabolic conversions to gain energy.
Consequently, complete anaerobic oxidation occurs in steps that are catalyzed by
different organisms. The anaerobic food chain that results from the step-wise oxidation of
organic compounds necessitates a greater level of interaction between microbial species
(7,17,45). Highly structured environments, such as microbial mats and biofilms, are also
characterized by a high level of interspecies interactions (postive and negative) due to the
close proximity of cells and microscale chemical gradients that are important for the
metabolism of the organisms within the community (38,53).

Negative interactions and community structure. Competition is an interaction that
has a negative impact on the grth of the interacting species. Two types of competition
between organisms are possible, exploitative competition and interference competition.
Exploitative competition refers to the physical/biochemical race for uptake of substrate,

while interference competition (a.k.a amensalism) refers to the direct infliction of harm

05'
l 4,5

on competing species. The negative impact of competition can be mathematically
represented by a coefficient that decreases the growth rate, or increases the death rate, of
a species depending upon the concentration of other competitors. The species that is
impacted the least by the presence of other competitors (i.e. has the smallest coefficient)
ultimately wins the competition, excluding the other species (24). Interference
competition is common in microbial communities, especially in, soil environments. The
production of a wide variety of antibiotics by fungi and bacteria, and the evolution of
genes to combat these antibiotics, attests the importance of interference competition in
some microbial communities.

Many microbes have similar nutritional and energetic demands that they must
meet from a limited pool of organic and inorganic compounds in natural environments;
thus, exploitative or substrate competition may be the most common type of negative
microbial species interaction. The outcome of resource competition is solely dependent
upon the concentration of the substrate in the environment and the shape of the Monod
growth curve (5 versus p.) of the competing organisms (32,41) (Figure 1.1). The Monod
model of substrate dependent specific growth rate [u = pmax s/(s + K,)] is the most
common mathematical equation used for microbial growth under substrate limiting
conditions because of its simplicity and accuracy in fitting most cases reasonably well
(4). A large number of experimental studies of single substrate competition in chemostat
cultures support this model (13,15,16,25,26,41).

According to the Monod model, bacteria can use two different strategies to
increase their competitiveness for a single substrate (Figure 1.1). One strategy is typified

by oligotrophs, organisms adapted to stable, low-nutrient environments that have high

 

 

 

 

 

“MWWBP
)Two utake
B ---------------------------------- fling-1.321.: ---'1
llmaxl / 4, -"" "" "T systems
/ -
”C l ___________________ var: ________ (C) Oligotmph
max .
/.-"
:3.
/{°'
P'---- -- ----3
----- vr
/: :
./ e
4,»; a
at : :
KSC KSB Kg.A [3]

Figure 1.1 p. versus s relationship of a copiotroph (A), an organism with high and low

affinity uptake systems (B), and an oligotrOph (C).

" l
03' 9h

Ilrl LIIi

re
.3033?

ll, l ll

substrate affinity and relatively low maximum specific growth rates. The other strategy is
typified by copiotrophs, organisms adapted to fluctuating, high-nutrient environments
that have lower substrate affinity and higher maximum specific growth rates. These
strategies are not necessarily mutually exclusive. Many microorganisms have both high
and low afiinity uptake systems that are induced under the appropriate conditions (Figure
1.1); however, obligate oligotrOphs and cOpiotrophs are also known that are specialized
for these two different lifestyles (14).

In anaerobic environments single substrate competition is a major force in
structuring the microbial community. For example, competition for hydrogen in
anaerobic environments is fierce, and organisms that can’t compete in environments with
a hydrogen partial pressure below their thermodynamic limits are only present at very
low levels. Hydrogen-consuming methanogens are generally not active in environments
where nitrate-, sulfate-, or iron-reducers are active because these organisms have a higher
affinity for hydrogen than methanogens. This is not only due to the thermodynamics of
different electron-accepting processes, but also to the lower affinity of methanogens for
hydrogen (29).

Despite intense resource competition, many conditions and adaptations allow for
the co-existence of species. The presence of multiple substrates and the need for more
than one nutrient (e. g. carbon, oxygen, nitrogen, phosphate, etc.) in natural environments
are obvious mechanisms for co-existence of similar species. When growth is limited by
more than one substrate, such as carbon and nitrogen simultaneously, co-exiStence is
possible. Multiple substrate limitation is a large subject and is reviewed elsewhere

(9,10,15,16,31). In addition, the sources of carbon and energy in natural environments are

direrse
such 2
tempo:

4.|'; .’\
U: ,0.

an“;
V

lfll‘,
mgiill:
and

tell ‘0

diverse, potentially creating different niches for microbes to occupy. Other conditions, .
such as predation, environmental perturbations, non-homogeneous micro-habitats,
temporal variability, and interference competition, can disrupt competitive exclusion
during competition for a single substrate, allowing many species to coexist (6,19,23,48).

For instance, studies of glucose-fed methanogenic reactors have found a surprising

amount of redundancy among organisms capable of fermenting glucose. In these

completely mixed reactors competitive exclusion was presumably excluded by the use of
carbon and energy sources other than glucose and the alternating feed-starve cycles

applied to the some of the reactors (11,12,56). This redundancy had important

implications for functional recovery of the reactors after application of an organic shock

load (11,21).

Positive interactions and community structure. Odum (1983) (35) noted that
newly introduced species tend to interact negatively, while species that have evolved
together over long periods of time tend to develop positive interactions. Considering that
microbes are the oldest forms of life on earth, positive interactions among microbes may
well be the most common interspecies interaction occurring in microbial communities.
Quite a few positive species interactions between microbes are understood, and the
transfer of molecules between the cooperating species has been defined (Table 1.2).
Transfer of molecules between cooperating species may be unidirectional or
bidirectional. Cooperative interactions may depend upon transfer of molecules between
two different metabolic types of organisms, or may require specific communication
between co-evolved species. An example of a non-specific unidirectional cooperation is

the consumption of the waste products of one bacterium by another. Even this type of

II t 1.1

UanflCtCSau any—303.32 DO>_C>:_ matzmcpwww..fiv CC.:U~:D::

 

.%°~Av%--nwv-rv-uhh F-ooaanvhu emf-PUmuPJ-uhnvunum ZOmUrUnhmflLrUuuhm nv>m~nerUhnfivaVnJ gnu-av?" twee \fl-mndw‘mndrivah'f. MW ~ mV~n.-w-~.

00308038 3830500sz a

 

0300035

000000.005
000300— 00308803

0300—08 0000003
80300000 0300003
8030.3 .3380

Game + mam T «mm + 0.0m
«:31 N00 AI CAIN + £0

om + 0 0.50.0 T £0: + em + N00
em + N00 T 0 25000 + cm
0200 030083 802005 0303
0000.0

.3. + 0.0 T 22 + 0..
mm + 03800.... T 030 .200 530m

030 @200 08088
80.608 0303

80300000 33830008
8030.3 0.3000

30.. 38800 80303000003080

TV 80380 8.38 00000 80058.? 08830

m 0200.5 TV < 0000.5 w0300w3 3.000808”—

cAommooamv

00030305 TV 000300.00. 830000.800 3.003.502

000000003035 .1 00w000£0§ 0030038 000.308 05000084

08038 mum iv 0000008 ma: 850.80.. Bram

00000< 3 000000.}. 00300308 00m>x0

083000 mm Al 000308 «E 000803 00w003>m
m 00. 03.005. AI < 05. 03.00:. 8000 000m
980003

00030305 T 0003003... 830003-800 3003382

118m

Attogtoods

M0'I

 

0080,3803 8300—02

00202: 08.00080 00308035

 

0030.08.08 0003003 000308008 0030000008. 33800000 0800 .30 50300.5 .N._ 0300.

war
in 1h:
then:

deper

orgar

mire

WEI

COT?

simple interaction creates growth conditions that are different than those in pure culture,
where most of our knowledge of metabolism is gained.

The most well studied cooperative interaction between prokaryotic species in
anaerobic environments is the syntrophic interaction of two metabolically distinct species
in the degradation of a substrate that neither species could degrade alone because of
thermodynamic constraints. The term syntrophic specifically implies that both partners
depend on each other for some metabolic activity that can not be replaced by the addition
of a cofactor or nutrient (46). Syntrophic interactions between hydrogen-oxidizing
organisms, such as methanogens, sulfate-reducers, or homoacetogens, and a variety of
hydrogen-producing fermentative bacteria are necessary for the degradation of fatty
acids, ethanol, glycolic acids, and aromatic compounds under anaerobic conditions (46)
(Table 1.2). Synergistic interspecies hydrogen transfer also benefits sugar-fermenting
organisms by allowing them to produce more hydrogen and oxidized fermentation
products, increasing the energy yield per molecule. Other interspecies interactions that
increase mineralization rates in anaerobic environments are poorly understood. For
example, undefined interactions between saccharolytic spirochetes and cellulolytic
bacteria, such as Spirochaeta caldaria and Clostridium thermocellum, enhance the rate of
cellulose degradation as much as two-fold over the rate of degradation by the cellulolytic
species alone (40,49).

Many anaerobes have nutritional requirements for certain vitamins or amino acids
that may also be the basis of cooperative interactions among microbial species.
Microorganisms found in food products and animal or plant associated environments

commonly require multiple amino acids and vitamins that are found in adequate supply in

10

illiOi

COEl.’

4 .
like,
lithe:

than

um:
Imp.

snif:

their environment. However, a survey of microorganisms revealed that organisms
isolated from environments that would not necessarily have an abundance of vitamins
and amino acids, such as sediments and other aquatic environments, are often
auxotrophic for one to several vitamins such as biotin, thiamine, or cobalamin, suggesting
that these vitamins may be supplied by other community members (this study). In
contrast to aquatic environments, prototrophs were more common in soil environments.
Independent of the environment from which they were isolated, autotrophic and
lithotrophic microorganisms were the most likely metabolic groups to synthesize all
vitamins and amino acids de novo. The excretion of large amounts of various organic
products by autotrophs, especially photoautotrophs, is the basis of many cooperative
interactions. The benefit of these interactions for the autotroph is not obvious, but include
the stimulation of N2 fixation, an increase in local C02 concentrations, and the structural
maintenance of mat communities (17). The only large phylogenetic groups to
homogeneously express prototrophy were the y—proteobacteria (except the genera
Legionella and Haemophilus, and few other cases of single vitamin requirements) and the
Planctomycetes.

Interactions between guilds within anaerobic communities, aside from the
unidirectional transfer of products though the anaerobic food chain, may be very
important for stable ecosystem fimction. In a previous study of methanogenic bioreactors,
shifts in the composition of the methanogenic guild corresponded to dramatic
compositional changes in the fermentative bacterial community. These compositional
changes did not appear to affect ecosystem function (12). It is not clear why a change in

the species composition of hydrogenotrophic and acetotrophic methanogens should affect

11

la 5

alias.

bet)“

iJlC’.

ands

em

lObl

mm

com:

O
—‘

( '(9
93

line

Beg:

tom

fermentative species, or vice versa, as long as both hydrogen and acetate are removed
effectively. This concerted change in the two guilds suggests that direct interactions
between the fermentative and methanogenic guilds may exist that are important for the
fiJnction of both guilds.

Microbial interactions such as those defined above that change degradation rates
and substrate uptake profoundly affect our models of anaerobic mineralization in natural
environments. In addition, species and strain specific cooperative interactions are likely
to be one reason for the difficulty in culturing many microorganisms from natural
environments. Because of the importance of interspecies interactions in microbial
communities, a concerted effort must be made to study the physiology of assemblages of

organisms.

Investigating microbial species interactions

Enrichment culture. The traditional enrichment culture technique developed by
Beijerinck and Winogradsky at the beginning of the twentieth century is one of the
foundations of microbial ecology. Enrichment cultures are especially useful for the study
of interspecies interactions if an enrichment culture is viewed as a way to study a
simplified community that is carrying out a specific biochemical process, rather than as
simply a means to obtain a pure culture. Either continuous or batch enrichment culture
methods can be utilized to simulate the natural environmental conditions more closely.
The enrichment of ammonia-oxidizing bacteria is a good example of how enrichment
cultures may be used in the study of substrate competition. Enrichment cultures for

ammonia-oxidizing bacteria yield different organisms depending upon the concentration

12

of an

3mm

illlllll

inter;

tom:

illElll

The

net.

of ammonia used in the enrichment, indicating that there are at least two types of
ammonia-oxidizing bacteria in nature that are specialized for different concentrations of
ammonia.

Enrichment cultures are even more appr0priate for the study of cooperative
interactions. After simplifying a natural microbial community to those microbes
connected to the process of interest, the members of the enrichment and their roles may
be more easily discerned than in the natural environment. The use of molecular
techniques (see below) to identify the members of an enrichment culture and sub-
culturing of individual members may allow the elucidation of the roles of the dominant
members and further understanding of the process as it occurs in natural environments.
The discovery that a type II methanotroph could oxidize atmospheric concentrations of
methane when cultivated in the presence of Variovorax species was made in such a way
(8).

Coculture. Deliberate coculture of two organisms already in pure culture has also
lead to many important discoveries about interspecies interactions. The coculture of two
organisms in continuous and batch systems is a common method to study competition for
single or multiple substrates, as mentioned above. Cases of coexistence of two species
when competitive exclusion was expected are perhaps the most interesting outcome of
such competition experiments. The coculture of two sulfide-oxidizing bacteria,
Ihiobacillus, a colorless sulfide-oxidizing bacterium, and Thiocapsa, a photosynthetic
purple sulfur bacterium, under oxygen-limiting conditions resulted in the coexistence of
the two species rather than the predicted exclusion of Thiocapsa. The Thiobacillus

Species produced incompletely oxidized Slllfill' compounds under oxygen-limited

13

CORGI

elem

Sill:
W).

i .
Jim

conditions. The Thiocapsa species used the incompletely oxidized sulfur compounds as
electron donors rather than sulfide. Thus, the two species were no longer in competition
for sulfide, but actually cooperated under habitat-simulating conditions (52).

The discovery of syntrophic growth through interspecies hydrogen transfer in the
culture “Methanobacillus omelianskii” has lead to the isolation of a number of syntrophic
organisms. A greater understanding of anaerobic degradation of even and odd-chained
fatty acids, branched chain fatty acids, amino acids, ethanol, and aromatic compounds has
resulted (46). Coculture of sulfide-oxidizing and sulfide-producing microbes has also
lead to a greater understanding of the sulfur cycle in natural systems. The use of coculture
techniques may significantly increase the culturability of environmentally important
microorganisms as well. In a study of rice paddy soils, it was reported that different
species of lactate and ethanol utilizers were isolated depending upon whether or not
Methanospirillum hungatei was used in the enrichment medium. The organisms isolated
did not require the presence of a methanogen in order to oxidize lactate or ethanol, yet the
presence of the methanogen allowed the isolation of the numerically dominant soil
organisms, while without the methanogen, organisms that were not numerically dominant
in the environment were isolated instead (43). The idea of using coculture techniques in
enrichment cultures, even when the process being studied is not thought to require any
syntrophic interactions, may prove very helpful in the isolation of environmentally
important microbes. Studying cocultures of organisms that are known to cohabitate the
same environments may lead to the discovery of new interactions, even if there is no a

priori reason that a coculture would be required for the organisms or processes of

interest.

14

emirl
bent:
Orga:
form
hem

lillEE

tlll‘ll
Elli-I.
Beta
seq;

in:

Cm

Microscopy. The spatial distribution of microorganisms within their natural
environment can be very important for their activity. The efficiency of metabolite transfer
between two cells is inversely proportional to the distance between the two (47).
Organisms that are known to interact in the syntrophic degradation of compounds often
form specific structural arrangements that allow for the efficient transfer of molecules
between the two populations in the environment (20,33). The use of microscopy and
image analysis may therefore be very helpfiil in the study of interspecies interactions.

The use of 168 rRNA-targeted oligonucleotides in fluorescent in situ
hybridization (FISH) allows the visualization of specific microbial populations in
environmental samples (1). In conjunction with confocal laser scanning microscopy, the
arrangement of cells in thin sections of biofilms and aggregrates can also be seen.
Because oligonucleotides can be designed for organisms that are only represented by
sequence information, valuable information about uncultured organisms’ morphology,
position, and associations with other cells in their natural environment or in an
enrichment culture can be described (Figure 1.2). The simultaneous application of FISH,
radiolabeled substrates, and autoradiography can give additional information about which
populations are utilizing different substrates (18,28,36). Information from these
techniques may aid in design of enrichment strategies, including cocultures, for
uncultivated microorganisms. In addition, studying the positioning and associations
formed by cultivated microorganisms could indicate an interspecies interaction that we
have overlooked that may be important for the in situ activity of the microorganisms.

Molecular methods. Our ability to describe the members of a microbial

community has drastically improved in the last three decades due to the development of

15

      
   
   
 

Environmental Sarnple —b Fixed Cells ,
In srtu PCR or

Extracted Nucleic Acids
RNA —> DNA

Real Time PCR

Probing
RT-PCR PCR
C . F' . t Fluorescent Specific PCR
ommumtyér Ingerprm Nucleic Acid Probes Primers
Cloned rRNA genes
Sequencing &

Database Search

Figure 1.2 Flow chart of molecular microbial ecology methods for community analysis
and monitoring of specific phylogenetic groups of microorganisms. RT-PCR = Reverse
transcription polymerase chain reaction. FISH = Fluorescent in situ hybridization.

16

he pol
method
from er
in the l
uncuhi
maths:
or hm.
lisuali.
the stu
tnapsh
enxhro
microl
the se
smith
Ofgtni
heliqj
comp;

Q
VII-‘1“. u
u «E.

b

it!"
ulnci'

the polymerase chain reaction (PCR) and 168 rRNA and DNA based phylogenetic
methods (39,54) (Figure 1.2). Amplification and cloning of 16S rRNA genes directly
from environmental samples has opened our eyes to the vast diversity of microorganisms
in the natural environment, and has allowed the development of methods to learn about
uncultivated microorganisms. Many different applications of 16S rDNA and rRNA
methods allow the detection, and even quantification, of specific strains, species, genera,
or broad phylogenetic groups (Table 1.3). Molecular methods that allow the in situ
visualization of specific organisms, such as FISH, are among the most usefirl methods for
the study of interspecies intereactions. Other methods aim to provide a fingerprint, or
snapshot, of microbial community composition, allowing the rapid comparison of many
environmental samples. The use of molecular techniques to monitor the composition of
microbial communities in enrichment cultures has greatly improved our understanding of
the selective bias of the conditions applied (44). In addition, molecular analysis of
enrichment cultures has resulted in the discovery of new interspecies interactions,
organisms, and physiologies (8,22,30,50)

Another way to elucidate important interspecies interactions is to take a more
holistic approach and study whole communities in their natural environment. Rapid
comparison of microbial communities with DGGE, T -RFLP, and other community
fingerprinting methods have allowed robust comparisons of microbial communities in
different layers of microbial mats, soils, lake strata, freshwater and marine sediments, as
well as changes in community structure over time (3,34,37,42,51,56). Quite a few
microorganisms whose 16S rDNA gene sequences were retrieved from the same

microenvironment repeatedly (i.e. the same depth layer of a microbial mat or stratified

17

£056029Q< 03¢.anan- 335.0904

,3FurVadanuhnvaPh- mamonvbhthqur-m Lune \fluV-aut rival-u brunt Zuvaf—aofc \flxkv—avuo udumh-PUEHJm-hh Luw--Qrd~av--iunv :1».- u-wnJ.-QL-w -~.-.::J-Avnh ‘00-: rvnflnvahuanu asst? ~...~ hv~n~9fl~i

 

0033300 003 000 0800030000030
30 0003000000 00330 .30 003030000

08030030000200
.30 0003000000 30030030 .30 003030000
0003—00
000020300 00 00300000000 03 0003030000
030000 .30 000309000 000 30300002

0000300 000800300 0
00330000000 .30 0003000000 000 30303002

000000 0000000300
00 0030003000 5 0000000000 00000000
00w000 .30 000309000 000 30300002

000500 0000000300 00
0030080000 .30 000309000 000 w0300002

000200 0000000300 00
0030000000 .30 0003000000 000 30300002

0000003
.30 003003 000 0000003000000
.30 0030033002 000 00300330000

000030300 30 00000 03000w0030
.0 00300330000 000 0030030005

3000000300030 000000000 03000
.30 009000 0w0_ 0 .30 00300330000

000000000
03000 0330000 .30 00300330000

003000303 000080000 03003

303E0w03 00008800 0003

003003030»: 030 03 0000000003
000 0300020000000 :00 £305

000000

00300030»: 33 03 0000000005

0003000300 0003 «.703 m0:

Mum 083.303

£2.40
033000 00000 093000 000 >303

35.5 030000 0003330000200
0000. 0000003 003030000 00300.0

 

0000300 000000300 00 A3008 030000003000
003000000000 30 000309000 000 m0300002 003E0w03 3300000000 03003 0w 00030003 30300000
000008 00.30 00.3 0000030 303 000 0000.00 00030090 030000 000 000000000 300000000
003020000330 .30 0mm000 23 003 03003 0003 $53 mm: 300 .30 E05005 000 3008.0 0003 553 m2
00300300< 0000003 000302

 

0003000005 00300000033 .30 030 0.3 .83 000508 32000 30000030 00300—000 .30 003003000 0300000 000 000903 00.0 .mg 0303.

18

12k?)

differ

lmpl

hem
hem
CUHI

prok

HER

lake) were eventually shown to have relatively strong interspecies interactions (37,38).
Competition between species could also presumably be deduced if the size of two

different microbial populations were inversely related over time.

Implications

Interspecies interactions have repeatedly been demonstrated to be extremely
important for the in situ activity of microorganisms in many environments. In most
natural habitats microbes are in close proximity to cells of other species. If bacteria in
pelagic and benthic environments were homogeneously distributed, the average distance
between cells would be between 122 and 12 um. In reality, bacteria are not
homogeneously distributed, but form aggregrates, microcolonies, and biofllms that
contain multiple species, thus the distance between cells in their natural habitat is
probably much less than 12 um on average (38). The study of microorganisms in pure
culture has undoubtedly taught us much about general and environmental microbiology,
but it has also caused many misconceptions of the true kinetic and metabolic capabilities
of microorganisms in their natural habitats. These misconceptions significantly impact
our models of biogeochemical transformations. The advances made in the study of
interspecies interactions imply that the interactions that occur may be quite difi‘erent from
the paradigm of syntrophy formed from examples in anaerobic environments. Exploring
new ideas about interspecies interaction may substantially increase the number of

environmentally important microorganisms that can be cultivated.

19

Thesis ove:
The
anaerobic
understand
mammalian
communiti
seaings an
concentram
nmcunm
infill pl:
mmmuniti
microbiala
Syntrophic
bioreactor
different In
Pm
”leaned;-
Mani Spa
Cellulose.‘
IEl'llllle g1
dioxide, 2
hernnacen

releaned e

Thesis overview

The tight metabolic coupling and interspecies interactions necessary for the
anaerobic degradation of organic matter are fascinating and important for our
understanding of global carbon and nutrient cycling, wastewater treatment, and
mammalian gastrointestinal tract communities. However, natural anaerobic microbial
communities are very complex because of the variety of substrates available in natural
settings and the fluctuation of physical parameters such as temperature, pH, and nutrient
concentrations. Manipulating natural environments and communities is also very
diflicult. Anaerobic bioreactors fed continuously with glucose at a loading rate of 0.5 g-I'
1°day'l provided a manageable environment to examine methanogenic microbial
communities (Figure 1.3). No electron acceptors were added to the chemostats, thus the
microbial community should contain three main guilds of organisms: glucose fermenters,
syntrophic fatty acid and ethanol fermenters, and methanogens (Figure 1.4) The
bioreactors used in this study were inoculated from various sources and operated for
different lengths of time (Figure 2.1).

Preliminary work on the microbial communities of these simple bioreactors
repeatedly documented enrichment of spiral-shaped cells that resembled spirochetes.
Many species of spirochetes are responsible for disease, while others flourish in
cellulose-degrading environments such as salt marsh sediments, the rumen and the
termite gut (5). Most spirochetes ferment sugars to acetate, ethanol, hydrogen, carbon
dioxide, and lactate; however, recent termite-gut isolates were shown to be H2/C02
homoacetogens (27). Spirochete species can be difficult to cultivate, therefore the

repeated enrichment of the spirochetes in the bioreactors provided a unique opportunity

20

00003.33 . .

  

0000303
000

 

   

303
30300003

03300 0000000033 .2 000303

 

 

 

 

 

 

 

 

 

0303 33 H 0000 00300003

0300 e. Z n 000,0 003000000

703.0 no u 200 00003

00mm

~03

000005

.008 2

Flu

 

 

 

 

 

 

 

VJ

0030.03
000.502

 

(lk

 

 

21

0000000005 2003000508 003000023 0_ 08000030000030 .30 000003 00030000 0.3 00035

0000302

00000003 A... 0: 000 0U

    
   

........ (Sinai
........ ficgfim «$0
3003
300030on 0003000502 In
00000.03
.0 mSQObHEAW .........

0000000000 000020 in

. 00000—0 ai..\

22

to Sta
The l

these

him

to study their biology and ecology in an environment that could be easily manipulated.
The following hypotheses were evaluated concerning the ecophysiology of spirochetes in

these methanogenic bioreactors:

Hypothesis 1: Spirochetes are members of the glucose fermenting guild in methanogenic
bioreactors.

Hypothesis 2: Spirochetes dominate the glucose fermenting guild in methanogenic
reactors when the in situ glucose concentration is low because they are adapted
for growth at low substrate concentrations.

Hypothesis 3: Removal of H2/formate by Hz-utilizing methanogen species increases the
growth yield of spirochetes on glucose.

Hypothesis 4: Cooperative interactions other than interspecies hydrogen transfer exist
between methanogens and spirochetes that are important to the ecology of both

microorganisms.

Experimental design

Research presented in chapter 2 explores the microbial diversity and structure of
the bioreactor communities. Chapter 3 addresses the ecological role of spirochetes in the
bioreactor communities (Hypothesis 1) Spirochete strains were isolated from bioreactor
fluid and characterized. 14C-labeled glucose was fed to bioreactor communities to
determine the flow of carbon through the ecosystem. Molecular methods based on 16S

rDNA were used to confirm that the spirochete strains isolated were the dominant

23

organ

glam

alel

U05

mas

organisms in the bioreactors. These studies revealed that spirochetes were fermenting
glucose to hydrogen, acetate, and ethanol in the bioreactor community.

The question of why the spirochetes dominated the glucose-fermenting guild in
the bioreactor communities required whole community, pure culture, and coculture
studies. Chapter 4 presents the effects of dilution rate on the entire bioreactor community,
including changes in spirochete and other glucose-fermenting populations. Chapter 5
describes the results of pure and mixed culture studies, including chemostat competition
of two predominate strains isolated from bioreactor fluid, Spirochaeta sp. strain R8 and
Clostridium ramosum strain S9. The results indicate that spirochetes are able to
outcompete most other glucose-fermenting organisms provided that the spirochetes have
a sufficient source of cofactors or vitamins required for growth.

The final questions about interactions between spirochetes and methanogens are
also presented in chapter 5. Pure culture, coculture, and triculture experiments were used
to demonstrate that the interaction between spirochetes and methanogens involved the
cross-feeding of pantothenate or a similar molecule. The cross-feeding of pantothenate
was also determined to be an important factor in determining the competitive advantage

of the spirochete strain.

References

l. Amann RI, Ludwing W, Schleifer K (1995) Phylogenetic identification and in situ
detection of individual microbial cells without cultivation. Microbiol Rev 59:143-
169

2. Bergh O, Borsheim KY, Bratbak G, Heldal M (1989) High abundance of viruses found
in aquatic environments. Nature 340:467-468

3. Buckley DH, Schmidt TM (2001) The structure of microbial communities in soil and
the lasting impacts of cultivation. Microb Ecol DOI: 10.1007/5002480000108

24

413

5C

10

4. Button DK (1985) Kinetics of nutrient limited transport and microbial growth.
Microbiol Rev 49:270-297

5. Canale-Parola E (1977) Physiology and evolution of spirochetes. Bacteriol Rev
41:181-204

6. De Freitas M], F redrickson AG (197 8) Inhibition as a factor in the maintenance of the
diversity of microbial ecosystems. J Gen Microbiol 106:307-320

7. Dolfing J (2001) The microbial logic behind the prevalence of incomplete oxidation of
organic compounds by acetogenic bacteria in methanogenic environments.
Microb Ecol DOI: 10.1007/s002480000076

8. Dunfield PF, Liesack W, Henckel T, Knowles R, Conrad R (1999) High-affinity
methane oxidation by a soil enrichment culture containing a type II methanotroph.
Appl Environ Microbiol 65: 1009-1014

9. Dykhuizen DE, Davies M (1980) An experimental model: bacterial specialists and
generalists competing in chemostats. Ecology 61: 1213-1227

10. Egli T (1991) On multiple-nutrient-limited growth of microoganisms, with special
reference to dual limitation by carbon and nitrogen substrates. Antonie van
Leeuwenhoek 60:225-234

11. Fernandez A, Hashsham S, Dollhopf SL, Raskin L, Glagoleva O, Dazzo FB, Hickey
RF, Criddle C, Tiedje JM (2000) Flexible community structure correlates with
stable community firnction in methanogenic bioreactor communities perturbed by
glucose. Appl Environ Microbiol 66:4058-4067

12. Fernandez A, Huang S, Seston S, Xing J, Hickey RF, Criddle C, Tiedje JM (1999)
How stable is stable? Function vs community stability. Appl Environ Microbiol
65:3697-3704

13. Fredrickson AG (1977) Behaviour of mixed cultures of microorganisms. Ann Rev
Microbiol 3 1 :63-87

14. Fry JC (1990) Oligotrophs. In: Edwards CA (ed) Microbiology of Extreme
Environments. Open Univeristy Press, Milton Keynes, pp 93—1 16

15. Gottschal JC ( 1985) Some reflections on microbial competitiveness among
heterotrophic bacteria. Antonie van Leeuwenhoek 51 :473-494

16. Gottschal JC (1993) Growth kinetics and competition - some contemporary
comments. Antonie van Leeuwenhoek 63 2299-3 13

17. Gottschal JC, Harder W, Prins RA (1992) Principles of enrichment, isolation,
cultivation, and preservation of bacteria. In: Balows A, Truper HG, Dworkin M,
Schleifer K (eds) The Prokaryotes, A Handbook on the Biology of Bacteria:

25

 

30}

21.}

M
g.‘

"1

241

16. l

Ecophysiology, Isolation, Identification, Applications, vol 1. Springer-Verlag,
New York, NY, pp 149-196

18. Gray ND, Howarth R, Pickup RW, Jones JG, Head IM (2000) Use of combined
microautoradiography and fluorescence in situ hybridization to determine carbon
metabolism in mixed natural communities of uncultured bacteria from the genus
Achromatium. Appl Environ Microbiol 66:4518-4522

19. Hansen SR, Hubbel SP (1980) Single-nutrient microbial competition: Qualitative
agreement between experimental and theoretical forecast outcomes. Science
207:1491-1493

20. Harmsen HIM, Kengen I-IMP, Akkermans ADL, Stams AJM, de Vos WM (1996)
Dectection and localization of syntrophic propionate-oxidizing bacteria in
granular sludge by in situ hybridization using 16s rRNA-based oligonucleotide
probes. Appl Environ Microbiol 62: 1656-1663

21. Hashsham S, Fernandez A, Dollhopf SL, Dazzo FB, Hickey RF, Tiedje JM, Criddle
C (2000) Parallel processing of substrate correlates with greater functional
stability in methanogenic bioreactor communities perturbed by glucose. Appl
Environ Microbiol 66:4050-4057

22. Holoman TRP, Elberson MA, Cutter LA, May HD, Sowers KR (1998)
Characterization of a defined 2,3,5,6-tetrachlorobiphenyl-ortho-dechlorinating
microbial community by comparative sequence analysis of genes coding for 16S
rRNA. Appl Environ Microbiol 64:3359-3367

23. Huston MA (1994) Biological Diversity: The coexistence of species on changing
landscapes. Cambridge University Press, New York, NY

24. Keddy PA (1989) Competition. Chapman and Hall, London; New York

25. Kuenen JG, Gottschal JC (1982) Competition among chemolithotrophs and
methylotrophs and their interactions with heterotrophic bacteria. In: Bull AT,
Slater JH (eds) Microbial Interactions and Communities, vol 1. Academic Press,
London, pp 153-187

26. Kuenen JG, Harder W (1982) Microbial competition in continuous culture. In: Burns
RG, Slater JH (eds) Experimental Microbial Ecology. Blackwell Scientific
Publications, Oxford, pp 342-3 67

27. Leadbetter JR, Schmidt TM, Graber IR, Breznak JA (1999) Acetogenesis from H2
plus C02 by spirochetes from termite guts. Science 283 1686-689

28. Lee N, Nielsen PH, Andreasen KH, Juretschko S, Nielsen JL, Schleifer K-H, Wagner
M (1999) Combination of fluorescent in situ hybridization and

26

ill

b)
C)
l/

32)

‘JJ
(I)
1‘

34;)

D)
co

microautoradiography - a new tool for structure-function analyses in microbial
ecology. Appl Environ Microbiol 65:1289—1297

29. Lovley DR, Dwyer DF, Klug MJ (1982) Kinetic analysis of competition between
sulfate reducers and methanogens for hydrogen in sediments. Appl Environ
Microbiol 43:1373-1379

30. Maltseva O, Oriel P (1997) Monitoring of an alkaline 2,4,6-trichlorophenol-degrading
enrichment culture by DNA fingerprinting methods and isolation of the
responsible organism, haloalkaliphilic Nocardioides sp. strain M6. Appl Environ
Microbiol 63 :4145-4149

31. Mateles RI, Chian SK, Silver R (1967) Continuous culture on mixed substrates.
Mixed cultures - mixed substrates. In: Powell EO, Evans C, Strange RE, Tempest
DW (eds) Microbial physiology and continuous culture. H.M.S.O., London, pp
233-239

32. Monod J (1942) Recherches sur la Croissance des Cultures Bacteriennes. Hermann &
Cie., Paris

33. Nielsen AT, Tolker-Nielsen T, Barken KB, Molin S (2000) Role of commensal
relationships on the spatial structure of a surface-attached microbial consortium.
Environ Microbiol 2:59-68

34. Nubel U, Garcia-Pichel F, Clavero E, Muyzer G (2000) Matching molecular diversity
and ecophysiology of benthic cyanobacteria and diatoms in communities along a
salinity gradient. Environ Microbiol 22217-26

35. Odum HT (1983) Systems Ecology: An Introduction. Wiley, New York

36. Ouvemey CC, Fuhrman JA (1999) Combined microautoradiography-16S rRNA
probe technique for determination of radioisotope uptake by specific microbial
cell types in situ. Appl Environ Microbiol 65:1746-1752

37. Overmann J, Coolen MJL, Tuschak C (1999) Specific detection of different
phylogenetic groups of chemocline bacteria based on PCR and denaturing
gradient gel electrophoresis of 16S rRNA gene fragments. Arch Microbiol
172283-94

38. Overmann J, van Gemerden H (2000) Microbial interactions involving sulfitr
bacteria: implications for the ecology and evolution of bacterial communities.
FEMS Microbiol Rev 24:591-599

39. Pace NR (1997) A molecular view of microbial diversity and the biosphere. Science
276:734-740

27

40F

41F

43R

44$

45?

51L

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

Pohlschroeder M, Leschine SB, Canale-Parola E (1994) Spirochaeta caldaria sp.
nov., a thermophilic bacterium that enhances cellulose degradation by
Clostridium thermocellum. ArchMicrobiol 161 : 17-24

Powel EO (1958) Criteria for the growth of contaminants and mutants in continuous
culture. J Gen Microbiol 182259-268

Ramsing NB, Ferris MJ, Ward DM (2000) Highly ordered vertical structure of
Synechococcus populations within the one-millimeter-thick photic zone of a hot
spring cyanobacterial mat. Appl Environ Microbiol 66: 103 8-1049

Rosencrantz D, Rainey FA, Janssen PH (1999) Culturable p0pulations of Sporomusa
spp. and Desulfovibrio spp. in the anoxic bulk soil of flooded rice microocosms.
Appl Environ Microbiol 65 :3526-3533

Santegoeds CM, Nold SC, Ward DM (1996) Denaturing gradient gel electrophoresis
used to monitor the enrichment culture of aerobic chemoorganotrophic bacteria
from a hot spring cyanobacterial mat. Appl Environ Microbiol 62

Schink B (1992) Syntrophism among prokaryotes. In: Balows A, Truper HG,
Dworkin M, Harder W, Schleifer K (eds) The Prokaryotes, A Handbook on the
Biology of Bacteria: Ec0physiology, Isolation, Identification, Applications, vol 2.
Springer-Verlag, New York, NY, pp 276-299

Schink B (1997) Energetics of syntrophic cooperation in methanogenic degradation.
Microbiol Mol Biol Rev 612262-280

Schink B, Thauer RK (1988) Energetics of syntrophic methane formation and the
influence of aggregation. In: Lettinga G (ed) Granular anaerobic sludge:
microbiology and technology. Pudoc, Wageningen, The Netherlands, pp 5-17

Smith KA, Arah JRM (1985) Anaerobic micro-environments in soil and the
occurrence of anaerobic bacteria. In: Jensen V, Kjoller A, Sorensen LH (eds)
Microbial Communities in Soil. Proceedings of 33rd FEMS-symposium.
Kopenhagen. Elsevier Science Publishing Co., London, pp 247-261

Stanton TB, Canale-Parola E (1980) T reponema bryantii sp. nov., a rumen spirochete
that interacts with cellulolytic bacteria. Arch Microbiol 127:145-156

Teske A, Sigalevich P, Cohen Y, Muyzer G (1996) Molecular identification of
bacteria from a coculture by denaturing gradient gel electrophoresis of 16S
ribosomal DNA fragments as a toll for isolation in pure cultures. Appl Environ
Microbiol 62:4210-4215

Urakawa H, Yoshida T, Nishimura M, Ohwada K (2000) Characterization of depth-
related population variation in microbial communities of a coastal marine
sediment using 168 rDNA—based approaches and quinone profiling. Environ
Microbiol 2:542-54

28

52. van den Ende FP, Meier J, van Gemerden H (1996) Coexistence of aerobic
chemotrophic and anaerobic phototrophic sulfur bacteria under oxygen limitation.
FEMS Microbiol Ecol 19: 141-151

53. Wimpenny J, Manz W, Szewzyk U (2000) Heterogeneity in biofilms. FEMS
Microbiol Rev 24 :661 -671

54. Woese CR (1987) Bacterial evolution. Microbiol Rev 51:221-271

55. Wommack KE, Ravel J, Hill RT, Colwell RR (1999) Hybridization analysis of
Chesapeake Bay virioplankton. Appl Environ Microbiol 65:241-250

56. Zumstein E, Moletta R, Godon J-J (2000) Examination of two years of community
dynamics in an anerobic bioreactor using fluorescence polymerase chain reaction
(PCR) single-strand conformation polymorphism analysis. Environ Microbiol
2:69-78

29

Petr

Hie}

Int:

1116‘.

COT.

CHAPTER 2

PHYLOGENY OF METHANOGENIC REACTOR ORGANISMS
Portions of the work presented in this chapter were published previously in the following
articles: Fernandez, A., Huang, 8., Hashsham, S., Seston, S., Xing, J., Hickey, R.,
Criddle, C., and Tiedje, J. 1999. How stable is stable? Function versus community
composition. Appl. Environ. Microbiol. 65:3697-3704; Hashsham, SA, A. S. Fernandez,
S.L. Dollhopf, F.B. Dazzo, R.F. Hickey, J.M. Tiedje, and CS. Criddle. 2000. Parallel
processing of substrate corresponds with greater functional stability in methanogenic
reactor communities perturbed by glucose. Appl. Environ. Microbiol. 66:4050-4057; and
Fernandez, A., 8A. Hashsham, S.L. Dollhopf, L. Raskin, O. Glagoleva, F.B. Dazzo, RF.
Hickey, C.S. Criddle, and J .M. Tiedje. 2000. Flexible community structure correlates
with stable community function in methanogenic reactor communities perturbed by

glucose. Appl. Environ. Microbiol. 66:4058-4067.

Introduction

There are many diverse environments, natural and man-made, that produce
methane. In all cases our knowledge of the environmental factors and organisms
controlling methanogenesis is incomplete. We perhaps know the most about the
physiology and ecology of organisms involved in the mesophilic, freshwater degradation
of sugars to methane. Methanogenic communities metabolizing simple sugars should
contain three main functional microbial groups, or guilds. The sugar-utilizing organisms
represent one guild. The majority of organisms utilizing sugars in a methanogenic
environment should be fermentative due to the lack of suitable electron acceptors, such as

02, N03“, and SO4'2. A vast array of bacteria from many different phylogenetic groups

30

prod

enx'i:

utiii
H; E

1
0:1.

eh;

1“:

girl

‘5?

can ferment simple sugars. The hydrogen, carbon dioxide, organic acids, and alcohols
produced by this guild are then utilized by the other two guilds found in methanogenic
environments (23).

Acetate, one of the most common fermentative products, is used directly by
acetoclastic members of the second guild - the methanogens. Two genera of acetoclastic
methanogens are known, Methanosarcina and Methanosaeta. Methanosarcina spp.
utilize H2 and CO2, acetate, and methylated amines with generation times of 9 hours on
H2 and CO2 and 17 hours on acetate. In contrast, all described Methanosaeta spp. utilize
only acetate, for which they have a high affinity, and have doubling times greater than 24
hours (13). Other methanogens typically found in methanogenic reactors, such as
members of the families Methanobacteriaceae and Methanomicrobiaceae, utilize formate
and/or H2 and CO2 that is produced during the oxidation of glucose, propionate, butyrate,
ethanol, and other fatty acids (21,23).

The syntrophic guild, the third fimctional group, depends upon H2 utilizing
methanogens to remove H2 and/or formate from the environment. If H2 or formate reach
concentrations above 1 Pa and 10 M, respectively, the oxidation of volatile fatty acids
carried out by syntrophic organisms is no longer thermodynamically favorable (23). Two
main phylogenetic groups of syntrophic bacteria have been described. The first group,
which includes the genera Syntrophobacter, Pelobacter, and various sulfate reducing
strains, is found within the delta subclass of the proteobacteria (10,11). The bacteria in
this phylogenetic group oxidize ethanol, propionate, and other odd numbered fatty acids
in syntrophic associations with hydrogen scavenging bacteria. The other phylogenetic

group of syntrophic organisms, which includes the genera Syntrophomonas,

31

clan

‘DOL
llbu

11a!
. \n.

llal

 

Syntrophospora, and Syntrophus, are found in the low percent G+C subdivision of the
gram-positive bacteria (1,29). These bacteria oxidize butyrate and some aromatic
compounds in syntrophy with hydrogen scavengers (19,20).

In this study our goal was to explore a simple methanogenic community by
cloning and identifying the most common bacteria and archaea in 14 continuously stirred
methanogenic reactors fed glucose as the sole carbon and energy source. This analysis
revealed sequences related to both cultivated and uncultivated prokaryotic species and
corresponded well to other studies of methanogenic environments. In addition, we
compared the species composition of reactors seeded from different sources, operated
under slightly different conditions, or perturbed in some way and found that some species

seemed to be indicative of certain reactor conditions.

Materials & Methods

Reactor design and operation. Sequences were retrieved fi'om similar
continuously stirred, mesothermic (35°C), methanogenic reactors that were part of three
separate studies (Figure 2.1). All reactors were fed via a buffered glucose solution (8 g

l + 6 NaHC03 g-liter’l) and a 40X concentrated nutrient solution as

glucose-liter'
previously described (28,12). In the first study a 1.5 liter reactor, designated reactor P1,
was inoculated from a 16 L continuously fed reactor and supplied glucose in a 2-day
cycle: 16 g-liter'1-day'1 for one day and 0 g-liter'l-day'l the following day, resulting in an
average loading rate of 8 g-liter'l-day'l (Figure 2.1A). A constant dilution rate (0.1 day")
was maintained and a steady state was achieved after 400 days.

In the second experiment, four reactors [designated the high spirochete (HS) set]

were inoculated with fluid from a 16 L methanogenic reactor that had operated for 200

32

B
lhn
Sit
in
Figure 2.1. Reactor operation and sampling. Dotted horizontal arrows indicate 5},
inoculation and solid horizontal arrows indicate reactor operation. Black and gray circles
indicate sampling times for 16S rRNA gene clone libraries and bacterial isolation,

, Ru
respectively. Vertical arrows indicate an intentional organic shock load. I-IRT = hydraulic Fl
retention time.

C
R
F
R
323'
1
Re

33

A

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

977 1437 1505
Municipal ) M1 days da 3 da 8
Sludge 16 L 200 days 900 1036 1325 1470
days days days days
Continously- Pulse- Fed
Fed Reactor Reactor
B 250
days
Mme,» M a
Sludge 16 L 200 days . M1 cont.
C , I (cont) - HS 8 >900 days
°“‘“‘°“‘ 3' Variable Volume 80 as 104
Fed Reactor 250 ml Reactors days days days
1 feel—o—o—H
. e ‘W‘ : #
Municipal M2 31:53 ................ —H
Sludge """" ’ 16 L 60 days §L83 M2 cont.
. Continously- ' LS 4—.L".—> 370 days
:llcles Fed Reactor Variable Volume 32 40
250 ml Reactors days days
till.
. Rumen ....... ) LS. —.—’
13.11 F 101d 210
Variable Volume days
250 ml Reactor
C
i 16 HRT
Rumen R1 (days)
. ....... . L A
FIUId ) 16 L 330
- _ (I
{25333;} 48 16 8 6 4 zdays I'IRT
River ..................... ) 3’3 l I L | #81348).
Sediment 320 I 72 1 60 1 40 I 18
Continously- days days days days days
Fed Reactor 407
diys HRT
. . 16 8 6 4 2 (days)
RLS 1 ..................... ) M3 1 I l : :
eactor 16 L 320 72 60 40 18
Continously- days days days days days
Fed Reactor

days

file;

1633

days with glucose as the sole carbon and energy source (Figure 2.1B). The other four
reactors [designated the low spirochete (LS) set] were inoculated with fluid from a
different methanogenic reactor that had been supplied glucose for 60 days. The HS and
LS sets were operated at steady-state conditions for 80 and 32 days, respectively, during
which time the concentration of glucose in the reactors was below the detection limit (20
uM) of the HPLC methods used (12). Both sets were then perturbed with a shock load of
glucose by instantaneously increasing the concentration of glucose in the reactor liquid to
approximately 38 mM using a concentrated sterile stock solution that also contained a

50:50 mixture of NaHCO3 and KHCO3 in order to provide buffer for the additional

organic acids expected from glucose.

In the third experiment, two reactors were inoculated and compared to a well-
established control reactor (Figure 2.1C). Reactor R1 (Rumen 1) was inoculated from
rumen fluid and operated under the above conditions for 400 days prior to the inoculation
of the two new reactors. The two new reactors were inoculated from sediment of the Red
Cedar River in Lansing, MI and a 250 ml variable volume reactor (LS 1) from the
previous study and designated reactor S2 (Sediment 2) and reactor M3 (Municipal 3),
respectively. Reactor chemical oxygen demand (COD), volative fatty acids (VFAs), and
pH were measured periodically in all reactors to monitor operating efficiency (12,28).

Extraction and purification of community DNA. Community DNA from
reactors P1, HS 5-8, and LS 1-4 was extracted from 10 ml samples using a French press
(15,000 lb/inz) for lysis and a phenol-chloroform DNA extraction method as previously
described (5). Community DNA from reactors M1, R1, 82, and M3 was extracted from 2

ml samples with a bead-beater method (17).

35

tom;
clone
Basil
pl'lm:

1131112
5

W)!

‘U

Rena:

l6S rDNA clone libraries. Bacteria and archaea clone libraries from reactors P1,
HS 5-8, and LS 1-4 were constructed by Ana Fernandez as described in Fernandez et. a1.
1999 (5). Briefly, the bacteria clone libraries were prepared from amplification products
produced with a forward primer which corresponds to nucleotide positions 19 to 38 of the
Escherichia coli rRN A and a reverse primer which corresponds to the complement of
positions 1521 to 1541 (27). The archaea clone libraries were prepared from
amplification products produced with a forward primer which corresponds to nucleotide
positions 50 to 69 of the E. coli rRNA (14) and a reverse primer which corresponds to the
complement of positions 958 (5’-[C/T]CCGGCGTTGA[A/C]TCCAATT-3’) (2). All
clone libraries from reactors LS’, R1, 82, and M3 were constructed by S. DollhOpf.
Bacteria clone libraries from reactor 82 were constructed using the bacteria-specific
primers 8F and 1392R. An archaea-specific clone library from reactor M2 was contructed
using archaea-specific primers 21F and 958R (2).

All amplicons were cloned with the TA cloning kit (Invitrogen, Carlsbad, Calif.)
according to the manufacturer’s instructions. Direct amplification of plasmid DNA from
whole cells of white colonies was accomplished with the following conditions: initial
denaturation (92 C for 2 min) followed by 35 cycles of denaturation (94°C for 30 s),
annealing (67°C for 1 min), and extension (72°C for 3 min) and a single final extension
(72°C for 7 min). Primers complementary to the TA cloning vector 5’-
GCCGCCAGTGTGCTGGAATT-3’ and 5’-TAGATGCATGCTCGAGCGGC-3’ were
used (30). The fragment size of these amplicons was checked by electrophoresis in 1 %
agarose. PCR products of the right size (1,500 bp for bacteria and 900 bp for archaea)

were digested simultaneously for 12 h with two restriction enzymes (HaeIII and HhaI;

36

(lit l
bull:
lesul

SClll'

llfar.

Ram

Gibco BRL, Gaithersburg, MD) according to the manufacturer’s specifications. The
restriction fragments from each clone were separated by electrophoresis on a 3.5 %
(wt/vol) MetaPhor agarose gel (FMC, Indianapolis, IN) in fresh 1X Tris-borate-EDTA
buffer at 4 V/cm in a cold room and observed after staining with ethidium bromide. The
resulting restriction banding patterns were normalized and compared with GelCompar
software (version 3.1; Applied Maths, Kortrijk, Belgium). Clones representing the most
abundant banding patterns for each sample were chosen for sequencing, except in reactor
S2 where clones that potentially matched certain terminal restriction fragments were
chosen for sequencing.

Bacterial isolates. Spirochaeta sp. str. M1-8, M2-8, R8 and M3-10 were isolated
in mineral medium (24) plus 10 % (v/v) clarified rumen fluid, 5 mM glucose, 0.7 %
agarose, and 50 pg ml‘l rifampin adjusted to a pH of 7.2 (15). A dilution series of fluid
from reactors M1, M2, R1, and M3 was done in 10 ml agar shake tubes with a headspace
of N2:C02 (80:20) and incubated at 37°C. Single colonies from the highest dilution tube
in which growth occurred were picked with a sterile Pasteur pipette and serially diluted in
10 ml agar shake tubes. This was repeated three times to ensure that a pure culture was
obtained. Liquid cultures were inoculated with a single colony picked from the highest
dilution tube containing colonies. Clostridium sp. str. S9 was isolated in the same manner
except tryptic soy broth plus 0.2 % (w/v) glucose, 1 mM dithiothreitol, and 50 ug ml”1
rifampin was used instead of mineral medium. Isolate 094 was isolated by A. Fernandez
from reactor P1 (6). Spirochaeta sp. str. R8 has been deposited in the DSMZ under

accession number 13955. Genomic DNA from all isolates was extracted and purified

37

wing

SDCCII

ampli
R‘Stl‘l
colun
the l
preri.
gene
tors:
Dual
{Sim
slid/1
lmm
All?
Den

Rim

using a Qiagen kit. The 16S rRNA gene of each isolate was amplified with bacteria-
specific primers and cloned as described above.

Sequencing and sequence analysis. The 16S rDNA gene fragments were
amplified for sequencing directly from whole cells containing the correct plasmid and
insert as described above. The amplicons were then purified with Microcon-IOO spin
columns (Millipore Corp.) Sequences were obtained via the dye terminator method by
the Michigan State University DNA sequencing facility with modified versions of
previously described sequencing primers targeting conserved regions of the 16S rRNA
gene (27). Alignment of sequences, mask construction, chimera check, and dendrogram
construction were performed with ARB (26) and software provided by the Ribosomal
Database Project (18). Dendrograms were constructed with PAUP' 4.05 software
(Sinauer Associates, Inc. Publishers, Sunderland MA). Genbank accession numbers are:
Methanobacterium formicicum, Z2943 6; Thermotoga maritima, M21774; Planctomyces
Iimnophilus, S3 9795; Verrucomicrobium sp., X99390; Aminobacterium mobile,
AF073521; Aminobacterium colombiens, AF 069287; Anaerobaculum thermoterrenum,
U50711; Candidate division OPll, AF027030; Escherichia coli, D1506l;
Desulfuromonas acetoxidans, M26634; Syntrophobacter fiimaroxidans, X82874;
Syntrophobacter sp. , X94911; C lostridium ramosum, X73440; Eubacterium barkeri,
M23927; Eubacterium hadrum, ARB_7A6488B3; Streptococcus bovis, M58835;
Streptococcus macedonicus, Z94012; Propionibacterium acnes, M61903; Spirochaeta
zuelzerae, M88725; Spirochaeta sp. str. Mastotermes, X79548; Spirochaeta caldaria,
M71240; Spirochaeta stenostrepta, M88724; Spirochaeta aurantia str. J -1, M5 7740;

Spirochaeta halophila, M88722; Spirochaeta sp. str. Antarctic, M87055; Spirochaeta sp.

38

str l
18;?”

Alli

react
Ctllf
imml
chill
The

$621

str. TM3, X97096; Spirochaeta sp. str. Masto., X79548; Treponema ZAS-l, AF093251;
Leptospira santarosai, U12672; EO-IV, AF 149889; LS*-B1, AF 157108; HS-B26,
AF 157107; S2-A through S, AF338764-66 and AF339852-59; WCHBl-63, AF050568;
WCHBl-44, AF050565; WCHBl-31, AF050569; WCI-IBl-40, AF050549; WCHB1-30,
AF050551; WCHBl-91, AF050550; RFS 75, AF068421; RFS 60, AFO68418; RFS 12,
AF068335; RFS 84, AF068428; HA49, U81766; HA73, U81735; EO-XIV, AF 149879;
EO-XI, AF 149878; EO-X, AF 149886; EO-IX, AF 149880; EO-VIII, AF 149882; EO-VH,
AF 149885; EO-VI, AF 149889; EO-V, AF 149881; EO-IH, AF 149888; EO-II, AF 149884;
EO-I, AF149883; LSI, AF219215; LSII, AF218216; LSIII, AF218217; LSIV,
AF218218; LSV, AF218219; HSI-III, AF218220-22; 05.95.32, ; AA-AD; AF339848-51;
AO-I through IV, AF 1498890-93.

16S rRNA oligonucleotide probing. Approximately 10 mL samples from
reactors LS 1, LS 2, HS 6, and H87 were taken immediately before the glucose pulse and
centrifuged for 10 min at 8,000 X g. The supernatant was discarded and the pellets were
immediately stored at —80°C until the nucleic acids could be extracted. Phenol-
chloroform-isoamyl alcohol (100:24:1 [vol/vol/vol]) extraction was used to extract RNA.
The concentration and quality of the extracted rRNA was measured
spectrophotometrically at 260 nm and by electrophoresis through a 10% polyacrylamide
gel. Quantitative rRN A membrane hybridizations were performed as previously
described by Raskin et al. (21). The following reference organisms and probes were used:
Meihanosarcina sp. strain WH2 (S-G-Msar-0821-a-A-24), Methanobacterium bryantii

M.o.H.G. (DSM 862) (S-F-Mbac-0310-a-A-22), Methanosaeta concilli FE (DSM 3013)

39

Figure 2.2. Maximum likelihood tree of cloned bacterial 16S rRNA gene sequences from
bioreactors and other methanogenic environments. The backbone of this tree was cast
using 696 unambiguously aligned nucleotides between E. coli positions 95 and 1295.
Partial sequences were then appended to the tree using the maximum likelihood
algorithm with 261 unambiguously aligned nucleotides between E. coli positions 154 and
474. Scale bar = 10 % difference nucleotide sequence. The origin of the cloned genes is

as follows: HA = anaerobic vinasses reactor (7); E0 = Reactor P1; HS = Reactors HS 5-

8; LS = Reactors LS 1-4; 82 = Reactor 82.

4O

 

es ft:

.25 iii

 

Candidate Divison OP 11
Aminobacterium mobilis
Aminobacterium colombiense
S2-P
S2-K
HA 73
Anaerobaculum thermoterrenum
EO-VIII
EO-V
Spirochaeta sp. str. R8 _fl_

HS I
EO-IX
SZ-S

Spirochaeta caldaria
SZ-H
LS I
EO-I
R Spirochaeta sp. str. Antarctic
r—WCHB 1-63

SZ-A
H L:WCHB 1-44
WCHB 1-3 1

LSIV

1‘ C Iostridium ramosum
Streptococcus bows

LS HI
LS H
LS * -B 1
Streptococcus macedonicus
l_‘ J Eubacterium barkeri
: _' LS V

— HS 111
____ds_tr. P1-O94
Eubacterium hadrum

r—--—'Propionibacterium acnes
. _ IEO-VII
EO-X Planctom c I' h!
HBO-III y es 1mn0p Ius
_ . EO-IV
S2-B
EO-VI
_ SZ-L
T—HA 49
'S2-E
S yntrophobacter sp.
'-—Syntrophobacterfiimaroxidans
Desulfilromonas acetoxidans

 

 

 

 

 

 

 

 

 

SZ-C

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

__{ Escherichia coli
Verrucomicrobium sp
j— 05.95.32

m_[ 182—]

Thermotoga maritima

 

 

 

Methanobacteriumformicicum

41

Result

leattel
rate g
most t
163 r]
phyla

within

*t'lele
Emit
tiller:
Sequel

Splmt

 

(S-G-Msae-O381-a-A—22), and Methanogenium cariaci JRl (DSM 1497) (S-O-Mmic-

1200-a-A-21) (22,25).

Results

Bacteria. Over 800 bacterial 16S rRNA genes were screened from 28 different
reactor samples during the three separate experiments performed. These cloned genes
were grouped according to their restriction enzyme banding patterns and the two to four
most frequent gene types from each sample were chosen for sequencing. The bacterial
16S rRNA genes from the reactors were diverse, with sequences falling in six major
phyla and ten different orders (Figure 2.2). The distribution of cloned 16S rRNA genes
within bacterial phyla was 44.4 % Spirochaetales ( 16), 13.8 % of uncertain affiliation (5),
8.3 % Eubacterium (3), 8.3 % Streptococcus (3), 5.5 % Mycoplasmatales (2), 5.5 %
Propiom'bacterium (2), 5.5 % Thermotogales (2), 5.5 % Anaerobaculum (2), 2.8 °/o
Syntrophobacter (l) and 2.8 % in environmental clone group WCHBl-33 (1). Although
there was significant diversity among the 16S rRNA genes cloned from the reactors,
certain groups and sequences were found repeatedly in different reactors, despite
differences in inoculum source, mode of operation, and amplification primers. Bacterial
sequences retrieved from more than one reactor included sequences related to the genera
Spirochaeta, C Iostridium, Eubacterium, and Streptococcus (Table 2.1).

The largest number of sequences fell into the order Spirochaetales and at least
one Spirochaeta-related sequence or isolate was obtained from 22 of 28 reactor samples
amplified with bacteria-specific primers. The Spirochaeta-related group of sequences

contained three subgroups, designated groups 1, 2, and 3 (Figure 2.3). Group 1 spirochete

42

 

mow. Che. when 3...: what mmm. Was: ems. whEeMNm wry-Se. med
—A— —A— —Am —An —Am —Ah ~Ah

IlvmdquuP-LEAVU Lrvvodwfivhfldmflu W-JCmb-W) nu. InvavrfiunvF-L -uuw—uduer—Adnnu scum <ZV—k rub: >.n~ norJrlvunvmv munu.-T.m-.witnadaehrlvmhxz ~ NH rU~A-~.hl

hnqufieUx

 

.monoa 028202530 9 cogfiwtnz: 23582: <7Eu m3 3 BEE—Sofie

.oo— LQOU EgEEAQOU Eon—mm l Q00 Eon—ES: :3 85.52% 33 55680 588. :8ch,.

Eztfioofiaoaoak n 36950.36wa .Eatmxogozcfie: n Eatefiog $396533: 0 Soon:
.meNEQERanoSm: n 8333353 .mmEEeBgoEBE M 832883 6:888:38: M 3:25.: 58356th n .233
630: 815.650 32:: 855: 28—6 meow <72. m2 m0 @6538 3 888% «Egg :1?

 

 

 

 

 

385 28%
A88 Bo: :35 A88 >6: :wé cognamv Swampsmv
$.an Xenav .xemoA e\emoA escov fimaA e\eoav .xemmA
833925:
Eateaog: Eatefiog 555:5 833335: amEteBoE
65.23: 3.88%: .333: coEEuoaoc 8 Z 35.88% 82 e633: 358.28% “oz egaeaom:
pagans:
shun 5 EU Eztfiogzm
Guam; ED .9:qu Emu Séaucaoxmgw 553.585 Shaw. 5 BO .oaiwu 30
£355.85 £535.86 aficmoeofixefi 28883.35 $886335 38863951 8588343 9.3m. _ 90
:8 Se are Be as. we a8 2 N are 8 £8 mm are 3 3% ow
22 mm mm .mq l v; m.— wl m4 w-w m: w-m mm
Ages—EV 930533 Ago—ES; Aomsauv
es 8 v .x. 8 v .x. 3 v ex. 3 A e. 3 A ex. 8 A .x. 8 A .3535
Eatmeuea: Enhance: Eztmfies 65.23: 6233.4:
3.85%: sateen: sateen: Surfing: 53.2893: Essence: EEEQQS wannabe.
.BEEQEER 53.2893er mmeoBEaefi
32.58: 2% = 96 32.88: 9&2 .5 seem a 9o
.BQQEEQEK topaz—bang Eatfioeeam .33: 90 6.2%. 5 BO 655% = 30 e.oa.§m. E 90 «£38m
82 E: was 2: use 22 mg 98. flee Re as 25 Beam
E E E E E E E

 

.emomzcsfifioo 58.883 33:3 E 30508 330208 8me <75? mm: >9 888% mEmEmwuoeoS AN 2an

43

v6;
CUM
4' G
in _

311$:

23?
Ci

sequences were the most common sequences found and were most closely related to
Spirochaeta caldaria, which belongs to a group of free-living spirochetes, including S.
stenostrepta and S. zuelzerae, that are phylogenetically more similar to host-associated
Treponema species that they are to other free-living Spirochaeta species. Many cloned
sequences from termite gut contents are also most closely related to S. caldaria. Three
isolates, Spirochaeta sp. str. Ml-8, M2-8, and R8, obtained from reactors M1, M2, and
R1, respectively, had identical 16S rRNA genes and belonged to group 1 (Figure 2.3).
Sequences identical to these reactor isolates were cloned directly from 11 of 14 reactors.

Spirochaeta group 2 sequences were 4 % divergent, on average, from group 1.
Group 2 sequences were also quite closely related to cloned 168 rRNA sequences
obtained from aquifer material (Figure 2.3). Sequences from group 2 were retrieved
repeatedly from reactor P1, however, they were less common than group 1 sequences
overall (Table 2.1). One isolate, strain M3-10, from reactor M3 also fell within group 2
(Figure 2.3). Group 3 spirochete sequences were 15 % divergent from the first two
spirochete groups and clustered with cell wall-less organisms morphologically
resembling Mycoplasma species, despite their phylogenetic affiliation with the
spirochetes (7). Group 3 spirochete sequences were obtained from 5 separate samples
from three different reactors (Table 2.1).

The second largest group of 16S rRNA sequences was 20 % divergent from any
cultivated organism. Sequences belonging to this clade were retrieved from reactors P1
and 82 (Table 2.1). The four sequences were 5 % divergent from each other and 7 %
divergent from HA 49, a cloned sequence from an anaerobic vinasses (winery waste)

reactor (8). These sequences could not be affiliated with any bacterial phyla with

44

Pig

in

Rt

Spirochaeta stenostrepla

l

 

RFS 75

 

 

 

0.10

 

LTreponema sp. ZAS-l

rSpirochaeta caldaria

Spirochaeta sp. str. R8

str. M1-8/ s-st

str.M2-8 HSI Groupl
EO-IX HS 11

EO-XIV sz-s

/EO-XI

Mastotermes gut clone
Spirochaeta zuelzerae

RFS 6O

 

 

 

RFS 12
r '— str. M3-10

EO-V Group 2
EO-VIH

'—WCHB1-40
WCHB 1-30

 

 

 

 

 

 

 

Li —WCHB 1-9

——RFS 84

 

 

 

 

 

 

 

 

 

 

 

 

 

l-Eo-I

 

Leptospira santarosai

Spirochaeta aurantia
Spirochaeta halophila

Spirochaeta sp. TM3
Spirochae ta sp. str. Antarctic

Group 3

Figure 2.3 Maximum likelihood tree of cloned spirochete-related 16S rRNA genes

from bioreactors and other methanogenic environments. This tree was calculated in

ARB using 312 unambiguousy aligned nucleotides between E. coli positions 154 and

478. Scale bar = 10 % difference in nucleotide sequence. The origin of the cloned

genes is indicated as follows: RFS = Termite gut (15); WCHBI = Aquifer (3); E0 =

Reactor P1; HS = Reactors HS 5-8; LS= Reactors LS 1-4; 82 = Reactor SZ.

45

conli

m l
ramt

endo

31ml

Were

confidence. Parsimony-based bootstrapping did not support the grouping of these
sequences with the Planctomyces (data not shown). Godon et a1. placed clone HA 49
with the Planctomyces, but they noted that this placement was questionable (8). In the
Ribosomal Database Project (18), clone HA 49 is grouped with Desulfuromonas spp.
within the delta-proteobacteria; however, clone HA 49 and the four related sequences
from our reactors never clustered with Desulfilromonas acetoxidans or other
proteobacteria in any analysis we performed.

Sequences from reactors LS 1-4 (Clone LS IV), S2 (Clone R), and Clostridium sp.
str. R8 were 100 % identical to each other and to the 168 rRNA sequence of Clostridium
ramosum (Figure 2.2). C. ramosum, C. spiroforme, and C. cocleatum form a group of
endospore-forming organisms that are phylogenetically located within the
Mycoplasmatales order, despite their genus name Clostridium. Unlike most clostridia,
these organisms do not produce butyrate as an end product of glucose fermentation. The
sequences were found in reactors LS 1-4 on day 40 after perturbation with glucose and in
reactor 82 on day 407 when flow rates were high (Table 2.1).

Three sequences related to Eubacterium species were found in reactors P1, LS 1-
4, and HS 5-8. Clone LS V, which was retrieved from the LS 1-4 reactors, was 96 %
similar to Eubacterium barkeri, while the other sequences from reactors P1 and HS 5-8
were 92 % similar to Eubacterium hadrum. Although E. barkeri and E. hadrum, have the
same genus name and share many phenotypic traits, such as the production of butyric
acid, their 16S rRNA gene sequences are in two different families within the order

Clostridiales, thus the distance between these two sequences is quite large (Figure 2.2).

46

found
volum:
use 01
30 the .
cridizi
from 2

'

,ruman

sequer
less 5
(05 95
This 1
Shown
segue:

B {h

Sequences closely related to Streptococcus macedonicus and S. caprinus were
found in reactors LS. and LS 1—4. These sequences were confined to the 250 ml variable
volume reactors. Two coexisting clones, EO-VII and EO-X, were also found in only one
type of reactor (Pl). EO-VII and —X had a sequence divergence of 0.5 % and were related
to the Propiom’bacterium genus. Only one cloned sequence related to a known fatty-acid
oxidizing, syntrophic bacterium was found. This sequence, SZ-E, was 0.5 % divergent
from an unnamed Syntrophobacter isolate and 4 % divergent from Syntrophobacter
fitmaroxidans.

The remaining sequences were quite divergent from their closest related
sequence, and some of them can only be tentatively assigned to bacterial phyla. All of
these sequences were found in reactors P1 and S2. Two sequences, one from reactor P1
(05.95.32) and one from reactor 82 (S2-J), cluster with Thermotogales-related sequences.
This grouping is weakly supported by parsimony bootstrapping analysis (data not
shown). Clones P and K from reactor 82 were 90.2 % similar to Clone HA 73, another
sequence obtained by Godon et. al. (8), that clusters with a deep-branching clade known
as the Anaerobaculum thermoterrenum group that contains many sequences and
organisms from anaerobic environments. Clone SZ-A from reactor 82 was placed within
the environmental clone group WCHBl-33, which contains no cultivated organisms, with
fairly high confidence. Lastly, clone SZ-C from reactor 82 had less than 83 % sequence
identity with any sequence in GenBank and could not be confidently affiliated with any
known phyla. Chimera analysis did not indicate a high probability of a chimeric sequence

for clone SZ-C or any other sequence in this study.

47

Methanosarcina barkeri

 

 

 

 

 

 

 

 

 

 

 

 

 

llé Methanosarcina sicilae
AL 98 74 Clone AA
56 Clone AB
100 q Methanosarcina mazei
95 AO-II
Methylcoccoides methyIuIens
Methanosaeta concilii
Methanobrevibacter ruminantium
71 JMethanbacterium formicicum
l AO-I
Methanobacterium byrantii
5§ Clone AC
95 AO-III
—— Clone AD
Methanobacterium defluvii

 

 

 

 

Methanopyrus kandleri

Figure 2.4 Neighbor-joining tree of Archaea 16S rDNA clones created from a Jukes-
Cantor distance matrix. The matrix was calculated from analysis of 328
unambiguously aligned nucleotides of the 16S rRNA gene. Bootstrap values from 100
parsimony replicates are shown to the left of each node. Scale bar = 5 % difference in

nucleotide sequence.

48

sires!

were

Archaea. (i) Clone libraries. Approximately 280 archaeal 16S rRNA genes were
screened from 8 different reactor samples. Only the two to four most frequent gene types
were sequenced from each sample, resulting in the sequencing of seven unique sequences
that cluster within the kingdom Euryarchaeota. The sequence of AO-I, the most frequent
16S rRNA gene in reactor P] from day 900 through 1325, was 100 % similar to the 16S
rRNA gene sequence of Methanobacterium formicicum (Figure 2.4). On days 1437,
1470, and 1505 AO-II, which exhibited a 99.7 % similarity to Methanosarcina mazei,
and AO-III, which was 92.0 % similar to Methanobacterium bryantii, were represented
fairly equally in the clone library. A small number of clones matched the banding pattern
for AO-I on day 143 7, but this pattern was not detected on days 1470 or 1505.

The archaea clone library from reactor M3 yielded four different 16S rDNA
sequences afier screening 30 clones. Clones AA and AB were 98.5 % and 99 % similar to
Methanosarcina sicilae, respectively (Figure 2.4). Two other clones, AC and AD,
composed 17 % of the archaea clone library and grouped with Methanobacterium
species. They were 92.2 % similar to each other and 90.4 % and 87 % similar to
Methanobacterium bryantii. Clone AC and AO-III were only 1.7 % divergent.

(ii) 168 rRNA oligonucleotide probes. Two reactors from HS 5-8 and two
reactors from LS 1-4 were chosen for rRNA membrane hybridization analysis: reactors 6
and 7 from the HS set and reactors 1 and 2 from the LS set. Replicate communities were
similar within each set of reactors, but the HS and LS sets differed significantly (Figure
2.5). In the HS set, the Methanosarcina specific probe S-G-Msar-0821-a-A-24 accounted
for 63% i 1% (Reactor 6) and 61% i 1% (Reactor 7) of the rRNA detected by all of the

methanogen-specific probes used. The Methanosaeta specific probe S-G-Msae-O381-a-

49

% total specific probes

 

 

 

 

2 6 7

r
1

I Methanomicrobiales Methanosaeta
I Methanobacten‘ales I Methanosarcina

Figure 2.5 Relative abundance of methanogen-related l6S rRNA in reactors LS 1 and
2 and HS 6 and 7 as determined by 16S rRNA membrane hybridization to
oligonucleotide probes. The abundance of each group is expressed as a percentage of
the total amount of methanogen-related 16S rRNA detected by all of the methanogen-

specific probes used.

50

Discu

A-22 accounted for 19% i 4% (Reactor 6) and 11% i 5% (Reactor 7). In contrast, in the

LS set the Methanosarcina specific probe detected only 1% i 0.3% of the rRNA detected
by all of the methanogen-specific probes, while the Methanosaeta specific probe detected
up to 38% (Figure 2.5). Additional differences were found among COg-reducing
methanogenic genera. Species belonging to the families Methanomicrobiaceae,
Methanocorpusculaceae, and Methanoplanaceae detected using the probe S-O-Mmic-
12000-a-A-21 accounted for over 60% of the methanogen-specific probes in the LS set
and less than 20% in the HS set. Lastly, the specific probe for the Methanobacleriales
Order detected between 6% and 12% of the methanogen rRNA in the HS set and between
0% and 1% of the methanogen rRNA in the LS set. There may have been other
methanogens present that were not detected by the methanogen-specific probes used in
this study; however, previous studies have found that these probes detect the majority of

methanogens present in mesophilic sewage sludge digesters (9,22).

Discussion

Many of the sequences retrieved from the 14 methanogenic reactors sampled in
this study were closely related to phyla that are expected in a methanogenic environment,
e. g., Spirochaeta, Eubacterium, C lostridium, Streptococcus, Propionibacterium,
Syntrophobacter, Methanobacterium, Methanosarcina, and Methanomicrobiales. The
function of these organisms in methanogenic environments is well known, i.e. the
fermentation of sugars, the oxidation of fatty-acids, and the formation of methane. An
exhaustive 16S rDNA clone analysis of a fluidized-bed anaerobic vinasses reactor found

similar genera (8). Eubacterium, Clostridium, Propionibacterium, Methanobacterium,

51

and .1

51.1614

craq

IHUSL

and Methanosarcina were, found in similar proportions in both studies. In addition, both
studies found more diversity within the bacteria than within the archaea.

In contrast to the Godon study, we found a much greater percentage of spirochete-
and streptococci-related sequences, which may be related to the different substrate and
biofilm-based ecosystem in the fluidized-bed vinasses reactor. The spirochete-related
sequences found in the vinasses reactor were 8 % divergent from Spirochaeta sp. str. R8.
The rest of the spirochete-related sequences from our reactors were not directly compared
to those in the vinasses reactor because the portion of the 16S rRNA gene sequenced in
the two studies for this group did not overlap (8). Many cloned spirochete sequences
from other methanogenic environments, such as termite hind guts and aquifer material,
are closely related to S. caldaria and S. zuelzerae (16). However, none of the termite gut
or aquifer sequences were more than 95 % similar to any of the sequences we retrieved.

A significant number of sequences retrieved from the reactors were related to
unusual phyla or were difficult to affiliate with any clade represented in GenBank or the
RDP. Two of these unusual groups, the unaffiliated sequences (S2-L, EO-III, EO-IV,
EO-VI) and the Anaerobaculum-related sequences (Clone K and Clone P), were most
similar to 16S rRNA genes cloned from the anaerobic vinasses reactor mentioned above
(Figure 2.2). Finding these two unusual groups in both types of reactors suggests that the
function of the corresponding organisms is not related to the substrate or fluidized-bed
nature of the vinasses reactor, but is a function also present in a completely mixed,
glucose-fed, anaerobic reactor. The few cultivated organisms within the Anaerobaculum
clade degrade amino-acids, suggesting this as a possible metabolism for the organisms in

this group that were detected in both reactors, although the cultivated organisms are only

52

distantly related to the sequences found here. The function of the organisms
corresponding to the unaffiliated sequences can only be speculated. Their high frequency
in the glucose-fed reactors would indicate that they ferment glucose, since this is the
largest source of energy in the reactors. If they do utilize glucose, which is probably one
of the most common sugars used in media, it is interesting that we have not cultivated
more organisms with similar 16S rRNA sequences.

Some correlations between reactor efficiency and the dominant bacteria
sequences present in each reactor were evident (Table 2.1). Eubacterium and Clostridium
sequences were especially common when reactor efficiency was perturbed (i.e., less than
95 %). Streptococcus sequences appeared only in the variable volume 250 ml reactors,
indicating that perhaps they were better adapted to stresses (such as occasional oxygen
introduction) that may have been more pronounced in the smaller reactors. Spirochete
sequences were generally the most common sequence type recovered when reactor
efficiency was high. Other phyla, such as the Ihermotogales-related and unaffiliated
sequences, did not seem to be associated with a particular type of reactor or operation
efficiency, and appeared only sporadically in the bioreactor communities. However, the
unaffiliated sequences were associated with the unexplained decrease in efficiency of
reactor P1, which eventually caused the fouling of this reactor (Table 2.1).

Methanobacterium was by far the most common archaea genus cloned from the
reactors. The samples in which few to no Methanobacterium sequences were detected
were taken within the first 50 days of reactor operation (Table 2.1). Based on the number
of sequences retrieved, Methanosarcina species appeared to be the dominant acetoclastic

species in the reactors. Methanosaeta sequences were only detected in samples taken

53

10:1

lac:

firs»;

within the first 50 days of reactor operation. The observation that different methanogen
species were present after the first 50 days of operation suggests that the environment
within the reactors was highly selective for particular methanogenic species.

The composition of the methanogenic community also had an impact on the
operating efficiency of the reactors, and reactors with relatively more Methanosarcina
species were functionally more stable during the glucose perturbation applied to reactor
LS 1-4 and HS 5-8 (4,12). In addition, a change in the particular Methanobacterium
species and in the proportion of Methanosarcina species present in the archaea
community of reactor P1 between days 1324 and 1437 corresponded with an abrupt
change in the bacteria community and a steady decline in reactor operating efficiency.
The cause of the change in both communities is unknown, but the simultaneous change in
both communities suggests an interaction between fermentative and methanogenic
bacteria that goes beyond typical interactions with obligately syntrophic organisms.

The results presented here show that even the highly selective environment of
mesothermic glucose-fed methanogenic reactors can maintain a high level of diversity,
including organisms that have no cultivated relatives, over long periods of time. The
sequences detected in these simplified reactors corresponded closely with those found in
comparable, but more complex, environments. The microbial community structure of
each reactor varied to some extent without having a dramatic effect on reactor function,
however the detection of certain organisms, such as Eubacterium and/or Clostridium
species correlated with sub-standard reactor function. Other organisms, such as
Spirochaeta, Methanobacterium, and Methanosarcina species dominated this

environment when operating conditions were stable. This descriptive study provides

54

He

fife

3' in

9H2

interesting observations that could be the basis of fiarther hypotheses, and should lead to

insights into the interactions of microbial populations in methanogenic environments.

References

l. Auburger G, Winter J (1995) Isolation and physiological characterization of
Syntrophus buswellz'i strain GA fi'om a syntrOphic benzoate-degrading, strictly
anaerobic coculture. Appl Microbiol Biotech 44:241-248

2. DeLong EF (1992) Archaea in coastal marine environments. Proc Natl Acad Sci USA
89:5685-5689

3. Dojka MA, Hugenholtz P, Haack SK, Pace NR (1998) Microbial diversity in a
hydrocarbon- and chlorinated-solvent-contaminated aquifer undergoing intrinsic
bioremediation. Appl Environ Microbiol 64:3869-3 877

4. Fernandez A, Hashsham S, Dollhopf SL, Raskin L, Glagoleva O, Dazzo FB, Hickey
RF, Criddle C, Tiedje JM (2000) Flexible community structure correlates with
stable community firnction in methanogenic bioreactor communities perturbed by
glucose. Appl Environ Microbiol 66:4058-4067

5. Fernandez A, Huang S, Seston S, Xing J, Hickey RF, Criddle C, Tiedje JM (1999)
How stable is stable? Function vs community stability. Appl Environ Microbiol
65 :3697-3 704

6. Fernandez A, Tiedje JM (2000) Characterization of a new microorganism able to
overgrow dominant fermentative bacteria in methanogenic reactors. American
Society of Microbiology General Meeting, Los Angeles

7. Franzman PD, Dobson SJ (1992) Cell wall-less, free-living spirochetes in Antarctica.
FEMS Microbiol Lett 97 :289-292

8. Godon I, Zumstein E, Dabert P, Habouzit F, Moletta R (1997) Molecular microbial
diversity of an anaerobic digestor as determined by small-subunit rDNA sequence
analysis. Appl Environ Microbiol 63 :2802-2813

9. Hansen SR, Hubbel SP (1980) Single-nutrient microbial competition: Qualitative
agreement between experimental and theoretical forecast outcomes. Science
207:1491-1493

10. Harmsen HIM, Kengen HMP, Akkermans ADL, Stams AJM (1995) Phylogenetic
analysis of two syntrophic propionate-oxidizing bacteria in enrichment cultures.
SYSt Appl Microbiol 18:67-73

55

11. Harmsen HJM, Wullings B, Akkermans ADL, Ludwing W, Starns AJM (1993)
Phylogenetic analysis of Syntrophobacter wolinii reveals a relationship with
sulfate-reducing bacteria. Arch Microbiol 160:238-240

12. Hashsham S, Fernandez A, Dollhopf SL, Dazzo FB, Hickey RF, Tiedje JM, Criddle
C (2000) Parallel processing of substrate correlates with greater functional
stability in methanogenic bioreactor communities perturbed by glucose. Appl
Environ Microbial 66:4050-4057

13. Koenig H, Stetter KO (1989) Archaeobacteria. In: Staley JT (ed) Bergey's Manual of
Systematic Bacteriology, vol 2. Williams & Wilkins, Baltimore, MD, pp 2171-
2216

14. Leadbetter JR, Breznak J A (1996) Physiological ecology of Methanobrevibacter
cuticularis sp. nov. and Methanobrevibacter curvatus sp. nov., isolated from the
hindgut of the termite Reticulitermesflavipes. Appl Environ Microbial 62:3620-
3631

15. Leschine SB, Canale-Parala E (1986) Rifampin-resistant RNA polymerase in
spirochetes. FEMS Microbial Lett 35: 199-204

16. Lilburn T, Schmidt TM, Breznak JA (1999) Phylogenetic diversity of termite gut
spirochaetes. Environ Microbial 1:331-345

17. Laefi‘ler FE, Ritalahti KM, Tiedje JM (1997) Dechlorination of chloroethenes is
inhibited by 2-bramaethansulfanate in the absence of methanogens. Appl Environ
Microbiol 63 :4982-4985

18. Maidak BL, Cale JR, Lilburn TG, Parker CTJ, Saxman PIL Stredwick JM, Garrity
GM, Li B, Olsen GJ, Pramanik S, Schmidt TM, Tiedje JM (2000) The RDP
(Ribosomal Database Project) continues. Nucleic Acids Res 28: 173-174

19. McInemey MJ (1992) The genus Syntrophomonas and other syntrophic anaerobes.
In: Balows A, Truper HG, Dworkin M, Harder W, Schleifer K (eds) The
Prokaryotes, A Handbook on the Biology of Bacteria: Ecaphysialagy, Isolation,
Identification, Applications, vol 2. Springer-Verlag, New York, NY, pp 2048-

2057

20. Mclnerney MJ, Bryant MP, Hespell RB, Costerton JW (1981) Syntrophomonas
wolfei gen. nov. sp. nov., an anaerobic, syntrophic, fatty acid-oxidizing bacterium.
Appl Environ Microbial 41 : 1029- 1039

21. Raskin L, Paulsen LK, Naguera DR, Rittmann BE, Stahl DA (1994) Quantification of
methanogenic groups in anaerobic biological reactors by oligonucleotide probe
hybridization. Appl Environ Microbial 60: 1241- 1248

56

22. Raskin L, Stramley JM, Rittmann BE, Stahl DA ( 1994) Group-specific 16S rRNA
hybridization probes to describe natural communities of methanogens. Appl
Environ Microbial 60: 1232-1240

23. Schink B (1997) Energetics of syntrophic cooperation in methanogenic degradation.
Microbiol Mol Biol Rev 61:262-280

24. Shelton DR, Tiedje JM (1984) Isolation and partial characterization of bacteria in an
anaerobic consortium that mineralizes 3-chlorabenzaic acid. Appl Environ
Microbiol 48:840-848

25. Stahl DA, Flesher B, Mansfield HR Montgomery L (1988) Use of phylogenetically
based hybridization probes for studies of ruminal microbial ecology. Appl
Environ Microbial 54: 1079-1084

26. Strunk O, Ludwig W (1997) ARB: Software for phylogenetic analysis. Technical
University of Munich, Munich, Germany

27. Weisburg WG, Barns SM, Pelletier DA, Lane DJ (1991) 16S ribosomal DNA
amplification for phylogenetic analysis. J Bacterial 173:697-703

28. Xing J, Criddle C, Hickey R (1997) Long-term adaptive shifts in anaerobic
community structure in reponse to a sustained cyclic substrate perturbation.
Microb Ecol 33:50-58

29. Zhaa H, Yang D, Woese CR, Bryant MP ( l 990) Assignment of Clostridium bryantii
to Syntrophospora bryantii gen.nav., comb. nav. an the basis of a 16S rRNA

sequence analysis of its cratanate-grawn pure culture. Int J Syst Bacteriol 40:40-
44

30. Zhau J, Davey ME, Figueras JB, Rivkina D, Gilichinsky D, Tiedje JM (1997)

Phylogenetic diversity of a bacterial community determined from Siberian tundra
sail DNA. Microbiology 143:3913-3 919

57

Da

8;”?

CHAPTER 3

THE IMPACT OF FERMENTATIVE ORGANISMS ON CARBON FLOW IN
METHANOGENIC SYSTEMS UNDER CONSTANT Law SUBSTRATE
CONDITIONS

These results will be published in the article: Dollhopf, S.L., Hashsham, S.A.,
Dazzo, F.B., Hickey, R.F., Criddle, CS, and J.M. Tiedje. 2001. The impact of
fermentative organisms on carbon flow in methanogenic systems under constant low

substrate conditions. Appl. Microbial. Biotech. In Press.

Introduction

The conversion of organic waste into methane, i. e. biomethanatian, is a process
that converts organic waste into useable energy, simultaneously reducing pollution and
providing a source of renewable energy. Methane is also a potent greenhouse gas, and the
release of methane from ruminant animals, termites, and anaerobic sediments plays a
significant role in global warming (30). Although many of the bacteria and archaea that are
involved in the anaerobic food chain are well known, the effect of different fermentative
organisms on carbon flow through methanogenic environments has not been well studied.
In sewage sludge and anaerobic sediments, 60 to 80 % of methane was produced from
acetate, with the balance coming from Hz/C02 (5,17,35). Prapionate, a precursor to
acetate, accounted for 15 to 24 % of the methane formed during sewage sludge and

glucose digestion, while butyrate contribution was negligible (8,18). In a study of

58

mesothermic cattle waste digestion, 13 and 8 % of the methane formed originated from
propionate and butyrate, respectively (25). Alcohols have only been shown to be an
important intermediate in the degradation of lactose (7). None of these previous studies
examined the fermentative organisms present or attempted to correlate their physiology
with the intermediary metabolites formed.

Molecular techniques now provide an opportunity to link the microbial community
structure to the functional attributes of biomethanation (1). Other studies have
concentrated on the connection between community structure and function during
perturbations and changes in operating conditions (1,10,14,28,33). The effect of
community structure on system firnction when substrate conditions are continually law has
not been investigated in methanogenic systems. The purpose of this study was to address
the fundamental question: Does community structure affect carbon flow in methanogenic
systems when substrate concentrations are kept continually low? In particular we were
interested in the influence of fermentative bacteria] members on carbon flow, because their
role seems to be overlooked in methanogenic systems. Microbial communities were
characterized using SSU rDNA terminal restriction fragment length polymorphism (T-
RFLP) analysis, SSU rDNA clone libraries, and phase contrast microscopy. Addition of
[U-”C]glucose to bioreactors and maximum substrate utilization rates were used to
identify important intermediates in glucose digestion. We found that acetate and ethanol
were formed in bioreactors containing high number of spirochetes, while acetate and
prOpianate were formed in bioreactors containing high numbers of streptococci. This
suggests that microbial community structure can affect the path of carbon flow under

constant low substrate conditions in methanogenic bioreactors.

59

Materials & Methods

Bioreactor operation. These studies were conducted in two variable volume
methanogenic bioreactors and two continuously stirred tank reactors (CSTRs) that were
operated at 34°C with an effective mean cell residence time (MCRT) of 16 days as
described previously (14). Only the variable volume methanogenic bioreactors were used
for l4C-carbon flow studies. Separate substrate and nutrient solutions were fed
continuously using syringe pumps at a rate of 12.5 and 5 ml/day (0.52 and 0.21 nil/hr),
respectively (14). The volume of the variable volume bioreactors increased from 250 ml to
362 ml over 8 days. Every 8th day 112 ml of suspended culture was removed. This resulted
in a glucose loading rate that varied between 0.4 to 0.28 g L'1 d'l. Reactor fluid from an
18 L CSTR that was inoculated with anaerobic digester sludge from the municipal
wastewater treatment facility at Jackson, MI and had been maintained in the lab for over 3
years was used to inoculate the variable volume bioreactor HS (high-spirochete).
Bioreactor HS was Operated for 250 days before this study commenced. Variable volume
bioreactor LS (low-spirochete) and CSTR R1 were inoculated with rumen contents at the
same time and were operated for 210 and 330 days, respectively, before they were used in
this study.

Sample preparation and microscopy. Samples (10 ml) were taken the day before
addition of [U -”C]glucose for morphological analysis. Freshly collected samples were
dispersed by multiple, rapid passages through a 25-gauge needle, and diluted to slight
turbidity to achieve an ideal spatial density of separated cells (ca. 100 cells per

micrograph) for morphotype recognition. Slides, photomicrographs, and digital images

60

were prepared as previously described (10). Classification of over 2,000 cells per sample
into Operational morphological units (OMU’s) was performed using the CMEIAS© plug-
ins (9,23) operating in UTHSCSA ImageTool Ver. 1.27 software (39). Averages and 95
% confidence intervals of OMU frequency were calculated by assuming that every
phatamicrograph represented an independent sample of the total community. Calculations
of community similarity were performed using EcoStat Ecological Analysis software
(Trinity Sofiware, Plymouth, NH) based on relative abundance of each OMU.

DNA extraction and T-RFLP. Samples (1 ml) for community analysis were taken
the day before addition of [U-“C]glucose and immediately stared at —80°C. DNA was
extracted using a beadbeaterTM method (24) modified by reducing the beadbeating
intervals from 1 min to 30 s to minimize DNA shearing. Bacterial 16S rDNA was
amplified using the primers 8F-Hex (AGAGTTTGATCCTGGCTCAG), which is specific
for the domain bacteria, and 1392R (ACGGGCGGTGTGTRC), a universal reverse primer
(1). The primer 8F-Hex was synthesized and labeled at the 5’ end with the
phosphoramidite dye 5-hexachlorofluarescein by Operon, Inc. (Alameda, CA). The
amplification reaction consisted of 25 ng DNA, 0.5 uM of each primer, 0.2 mM of each
deoxynucleoside triphosphate, 1.5 mM MgC12, 0.2 mg ml1 of BSA, 2.5 ul of 10X Taq
buffer and 1 unit of Taq DNA polymerase (Boehringer Mannheim, Indianapolis, IN) in a
25 pl final volume. The amplification was performed using the following conditions: initial
denaturation (95°C for 3 min) followed by 25 cycles of denaturation (94°C for 30 s),
annealing (55°C for 30 s), and extension (72°C for 30 s) and a final extension of 72°C for
7 min. Three separate amplifications of each sample were performed, pooled, and purified

using Microcon-100 spin columns (Millipore Corp, Bedfard, MA) and digested overnight

61

with either HaeIII, HhaI, or MspI. Aliquots (2 pl) containing approximately 30 ng of the
digested DNA mixture and 2 final of GENESCAN-2500 TAMRA internal lane standard
(Perkin Elmer Applied Biosystems, F aster City, CA) were then loaded onto a Perkin
Elmer Applied Biosystems 373A automated sequencer by the MSU DNA Sequencing
Facility. The resulting chromatograrns were processed with GeneScan 2.1 (Perkin Elmer
Applied Biosystems) to determine the size of each fragment. Community similarity was
calculated using the GelCompar sofiware package (version 4.2; Applied Maths, Kortrijk,
Belgium). Following normalization (no background subtraction), Pearson product-
moment correlation coefficients were used to calculate similarity matrices for each
restriction enzyme digest (31). Alternatively, peaks were manually aligned and used to
compute community similarity indices in EcoStat Ecological Analysis software (Trinity
Software, Plymouth, NH). Individual peaks and peak height were substituted for species
and species abundance, respectively. Identification of organisms corresponding to different
terminal restriction fragments was accomplished by computer simulated restriction digests
of the ribosomal database project (RDP) database, inspection of clone sequences, and
actual amplification and digestion of DNA from 16S rDNA clones and isolates (27).

16S rDNA analysis. Amplification of bacterial SSU ribosomal DNA (rDNA),
cloning, screening, and sequencing were performed as described previously (1 1). Thirty-
six clones were analyzed for each sample and the most frequent clone type was selected
for sequencing. Nearly full length sequences were obtained in both directions for all
sequences. The 16S rDNA of spirochete isolate R8 was amplified as described above for
community rDNA. Alignment of sequences, mask construction, and chimera check were

performed using software provided by the Ribosomal Database Project (26) and ARB

62

(37). Dendrograms were constructed using ARB and PAUP“ 4.05 (Sinauer Associates,
Inc. Publishers, Sunderland, MA).

Determination of intermediary metabolites. Radiolabeled glucose was fed to
the HS and LS bioreactors by replacing the normal substrate feed with an 8 g L'1 solution
of [U-“C]glucose with a specific activity of 0.015 mCi mmol'1 and 7 g L'1 N8HCO3. The
radiolabeled solution was fed continuously for 70 h to both bioreactors. Every 12 h, 1 ml
samples were taken and immediately centrifuged at 10,000 x g for 10 min, acidified with
0.1 M H2804 (100 pl), and held at —20°C until analysis. The samples were analyzed for
volatile fatty acids (VFA’s) using a high pressure liquid chromatograph (Binary LC Pump
250, Perkin Elmer, Norwalk, Conn.) equipped with a 300-mm HPX-87H ion exclusion
column (Bio-Rad Laboratories) connected to a Lambda-Max Model 480 LC
spectrophotometer (Millipore, Milford, Mass.) set at 210 nm and a B detector (Insus
Systems Inc., Tampa, Fla). The mobile phase was 0.013 N H2804 at 0.6 mL min'1 with the
column at 20°C. VFA's were identified by retention time using authentic radioactive and
non-radioactive standards. A standard curve of area versus DPM was calculated using [U-
14C]acetate of a known specific activity and also by collecting fiactions of other
radioactive compounds and counting them on a Tri-Carb 1500 liquid scintillation analyzer
(Packard Instrument Company, Downers Grove, Illinios). A ratio of 1.55 DPM for every
area unit was established. The detection limit for glucose and the individual VFA’s with
this system was approximately 20 M. Soluble compound concentrations were converted
to electron equivalents (e') according to the oxidation state of each compound.

Spirochete isolate R8. The spirochete isolation medium consisted of a mineral

medium (34) plus 10 % (v/v) clarified rumen fluid, 5 mM glucose, 0.7 % agarose, and 50

63

pg ml'l rifampin adjusted to a pH of 7.2 (21). A dilution series of R1 bioreactor fluid was
done in 10 ml agar shakes with a headspace of N22C02 (80:20) and incubated at 37°C.
Single colonies from the highest dilution tube in which grth occurred were picked with
a sterile Pasteur pipette and serially diluted in 10 ml agar shakes. This was repeated three
times to ensure that a pure culture was obtained. Liquid cultures were inoculated with a
single colony picked from the highest dilution tube containing colonies. Rumen fluid was
replaced by a cofactor solution (100x, pH 7.2) containing (per 100 ml): pyridaxal HCl, 10
mg; pyridaxal phosphate, 10 mg; calcium folinic acid, 2 mg; B-NAD, 2 mg; coenzyme A,
2 mg; FAD, 2 mg; nicotinamide, 1 mg; folic acid, 0.1 mg; riboflavin, 0.02 mg; thiamine
pyrophasphate, 100 mg (19). Isolate R8 has been deposited in the DSMZ under accession
number DSM 13955.

Fermentation products. Isolate R8 was grown in spirochete isolation medium for
identification of fermentation products. Eight days after a 5 % (v/v) inoculation with an 8-
day old culture fluid, visible turbidity developed. Samples were taken 6 and 8 days after
inoculation and analyzed by HPLC for VFA’s. Ethanol was measured using an enzyme
assay (ethanol test kit 332-A, Sigma Chemicals, St. Louis, MO). Glucose was measured
using hexokinase test kit 115 from Sigma Chemicals. Hydrogen was measured using a
Hewlett Packard 6890 GC equipped with a mercury reduction gas detector.

Maximum substrate utilization rates. Substrate utilization rate assays were
performed in 160 ml serum bottles containing 25 ml of R1 fluid, 10 mM acetate,
propionate, butyrate, lactate, or ethanol and 10 mM phosphate buffer (pH 7.2). Reaction
rates were determined by measuring the concentrations of acetate, propionate, butyrate,

lactate, and ethanol overtime by the methods indicated above. Methane was measured on

64

a Varian GC (model 3700) equipped with a Porapak Q column and a flame ionization
detector.

Nucleotide sequence accession numbers. The SSU rRNA gene sequences
determined in this study have been deposited in the GenBank database under accession

numbers AF157107 and AF157108.

Results

Community structure. Bioreactors HS and LS were operated under identical
conditions at an applied organic loading rate of 0.3 -- 0.4 g glucose-L"°day'l and were
efficiently processing substrate with total VFA concentrations in the effluent less than 4
mM and chemical oxygen demand removal efficiencies of approximately 90 % (data not
shown). Glucose concentratiOns were below the detection limit (20 pM) in both reactors
throughout the duration of the experiment. Morphological analysis of bioreactors HS and
LS revealed that they contained very different dominant microbial populations. Bioreactor
HS contained two codominant morphotypes, single cocci and spiral-shaped cells, which
comprised 33 i 3 % and 37 i 3 % of the community, respectively. LS was dominated only
by cocci, which comprised 75 i 2 % of the community and had only 0.5 % spiral-shaped
cells. The cocci present in bioreactor LS were phase light and often found in clusters in
contrast to the phase dark singly occurring cocci in bioreactor HS.

Visual inspection of the T-RFLP patterns from a HaeIII restriction digest of
amplified SSU rDNA from the two communities also showed that the HS and LS bacterial
communities were quite different (Figure 3.1). We compared the T-RFLP patterns from all

restriction digests by two methods. The Pearson product moment correlation coefficient

65

50 100 150

200 280 300 350 400

450

 

Bioreactor LS

 

 

 

LS-Bl 266b
/ ( p)

210

 

 

 

 

 

HS-B26 185b
/ ( p)

212

 

 

 

 

 

400- Bioreactor HS
300-
200-
1w-
0
‘00- Bioreactor R1
300-
200-
100

.jWL

 

 

 

 

Isolate R8

 

.WW

 

 

 

Figure 3.1. T-RF LP electropherograms of the HaeIII digested SSU rDNA amplified
from isolate R8 and bacterial communities LS, HS, and R1. Peaks corresponding to the
restriction fragment length of the most frequently cloned SSU rRNA gene sequences

are labeled.

66

calculates similarity based on every point in the chromatogram produced by the laser dye
reader, and does not require the identification and alignment of individual fragments (31).
Correlation coefficients for HS and LS electropherograms from restriction enzymes
HaeHI, HhaI, and MspI digests gave values below 0.50 for all digests (Table 3.1).
Alignment of T-RFLP patterns and identification of each fragment allowed the
computation of Horn’s community similarity index (2) based an abundance of each
terminal restriction fragment. Comparison of LS and HS electropherograms resulted in
Horn similarity indices below 0.2 for all digests (Table 3.1). These results agree with the
correlation coefficients and indicate that the HS and LS bacterial communities were indeed
very different.

The most frequent SSU rDNA clone type from bioreactor HS, represented by
clone HS-BZ6, accounted for 53 % of the total clones analyzed and was most closely
related to Spirochaeta caldaria, with a sequence similarity of 99.5 % (Figure 32A). In
bioreactor LS the most frequent clone type, represented by clone LS-Bl, accounted for 50
% of the clones analyzed and was 98.3 % similar in sequence to Streptococcus bovis
(Figure 32B). Digestion of clones obtained from HS and LS with HaeIII revealed a 185
base pair fragment that corresponded to clone HS-B26 and spirochete isolate R8, and a
266 base pair fragment that corresponded to clone LS-Bl (Figure 3.1). HhaI and MspI
digestions confirmed that the dominant peaks in the HS and LS communities also
corresponded to cloned sequences HS-BZ6 and LS-Bl, respectively (data not shown).

The larger bioreactor R1 also contained a large proportion of spiral shaped cells.
Because its larger volume provided us with reactor fluid to perform activity assays, the

degree of community similarity between HS and R1 was ascertained using T-RFLP

67

Table 3.1. Similarity of bacterial HaeIII, HhaI, and MspI T-RFLP
electropherograms as measured by Ham’s community similarity index and

Pearson product-moment correlation coefficient.

 

 

 

LS R1
HaeIII / HhaI /MspI HaeIII /HhaI / MspI
HS : ll-Iorn 0.14/0.09 / 0.09 0.76 / 0.62 / 0.86
Pearson 0.02 / 0.47 / 0.01 0.85 /0.98 / 0.96
LS : Horn 0.29 / 0.02 / 0.03
Pearson 0.02 / 0.50 / 0.0

 

68

Figure 3.2. Phylogenetic relationships of clones and isolates in the families
Spirochaetales (A) or Streptococcales (B). Unambiguously aligned nucleotides of the
SSU rRNA gene from E. coli position 111 to 465 (A) and 21 to 1477 (B) were used to
calculate Kimura 2-parameter distance matrices. The phylogenetic trees were created
with the neighbor-joining method. Neighbor-joining and parsimony bootstrap values (50
% or higher) are listed above and below each node, respectively. Scale bar = 10 %
difference in nucleotide sequence positions. .Genbank accession numbers are as follows:
Streptococcus bovis, M58835; Streptococcus macedonicus, Z94012; Streptococcus
mutans, $70358; Streptococcus equi, $70324; Bacillus subtilis str 168, D26185;
Lactobacillus acidophilus subspp. johnsonii, M99704; Lactococcus lactis subspp. lactis,
M58837; Streptococcus salivarius, M58839; Streptococcus parauberis, X89967;
Brachyspira hyodysenteriae, M57741; T reponema bryantii, M57737; Spirochaeta
zuelzerae, M88725; Spirochaeta sp. str. Mastotermes, X79548; Spirochaeta caldaria,
M71240; Spirochaeta stenostrepta, M88724; Spirochaeta aurantia str. J-l, M57740;
Spirochaeta litoralis, M88723; Spirochaeta halophila, M88722; Spirochaeta sp. str.
PLIZMY, RDP short ID str.PL12MY; Spirochaeta sp. str. Antarctica, M87055;
Leptospira inadai, 221634; EO-XIV, AF 149879; EO-XI, AF 149878; EO-V, 149881;

EO-I, AF149883; LSII, AF218216; and HSI, AF218221.

69

(2.79:4
“va

b .....

 

 

 

98

Serpulina hydoysenteriae

 

 

are
-¥>\l

 

90

 

fl

 

 

 

 

T reponema bryanti
Spirochaeta stenostrepta

 

 

 

 

Spirochaeta caldaria
Is a l a te R8
H S - B2 6

fl EO-XIV
85 69

7; H81
EO-V

 

 

 

Spirochaeta zuelzerae
Spirochaeta sp. str. Mastotermes

 

Spirochaeta aurantia str. J-I

 

Spirochaeta litoralis

 

 

 

 

 

Spirochaeta halophila
Spirochaeta aflicana

'— Spirochaeta sp. str. PLIZMY

Spirochaeta sp. str. antarctic

 

EO-I

 

100

0.1

Leptospira inadai

 

 

 

100

 

100

Lactococcus lactis

 

 

100

 

JD—i
H

Streptococcus paraubis

F——Streptococcus salivarus
Streptococcus bovis

 

 

99 98 Streptococcus macedom'cus

9S LS-Bl
LSII

 

Streptococcus mutans
Streptococcus equi

 

 

Bacillus subtilis

 

Lactobacillus acidophilus

0.1

 

70

analysis. The dominant terminal restriction fragment in bioreactors HS and R1 was a 185
bp fragment corresponding to the clone HS-B26 and isolate R8 (Figure 3.1). The two
communities also shared other less abundant terminal restriction fragments, as reflected in
the high community similarity values and correlation coefficients between HS and R1
(Table 3.1). Three separate samples taken fi'om bioreactor R1 on the same day gave Horn
similarity indices between 0.90 and 0.99 with a HaeIII digest and two samples taken from
bioreactor R1 two weeks apart yielded a Horn similarity index of 0.83 (data not shown).

Tentative identification of other community members was accomplished through in
silico digestion of sequences in the RDP database. The 210 base pair fragment seen in the
HaeIII digest of the LS community corresponds to the predicted fragment size for
Selenomonas ruminantium, an obligate anaerobe known to rapidly ferment lactate to
propionate (3) (Figure 3.1). A 210 base pair fragment was not found in either the HS or
R1 communities. A fragment of size 212 was found in communities HS and R1. This
fragment was always distinguishable from the 210 base pair fragment in repeated PCR
amplifications and T-RFLP gels of the three communities. Evidence for the presence of
Desulfovibrio species in bioreactors HS and R1 was found in both the HhaI and MspI
digests with terminal fragment lengths of 97 and 509, respectively (data not shown). A
terminal fragment of length 97 after digestion with HhaI is common to many species in the
genus Desulfovibrio. A 509 base pair fragment resulting from digestion with MspI also
corresponds to the species Desulfovibrio profimdus. Neither of these fragments was found
in LS.

Intermediary metabolites. Approximately 13 % of the 0.34 pmol e'-ml'1°hr'l

entering the reactor accumulated in the pool of extracellular intermediary metabolites in

71

bioreactor HS (Table 3.2). Acetate and ethanol contributed fairly equally to these
intermediary metabolites. Although ethanol contributed more electron equivalents than
acetate, slightly more acetate was produced on a molar basis. Labeled propionate was not
detected in the HS bioreactor over the course of this experiment.

In bioreactor LS approximately 10 % of the electron equivalents entering the
system were found in soluble intermediates, similar to bioreactor HS (Table 3.2). The
percentage of electron equivalents found in the soluble products indicates that both
bioreactors were operating efficiently, and agrees with the operation efficiency of
approximately 90 % calculated from chemical oxygen demand measurements of the
bioreactor fluid (data not shown). In contrast to bioreactor HS, acetate was the more
important fermentation intermediate, contributing twice as many electrons to the pool of
intermediary metabolites than propionate. No ethanol was detected in bioreactor LS.

Maximum substrate utilization rates. Utilization rates for acetate, propionate,
ethanol, lactate, and butyrate were determined for bioreactor R1 in order to confirm that
the intermediates detected with [U-“C]glucose were utilized in a community that
contained a high number of spirochetes (Table 3.3). Acetate and ethanol degradation
began immediately, and acetate was formed concomitant with ethanol degradation (data
not shown). The utilization rate for propionate was not significantly different than zero
and butyrate utilization was very low. Lactate was not utilized until 10 h after inoculation
with bioreactor fluid. The headspace of all bottles contained greater than 50 % methane
after the substrate was utilized, indicating that the assay conditions were favorable for

methanogenesis.

72

Table 3.2. Production of intermediary metabolites in bioreactor HS and LS as

determined by addition of [U-MC] glucose to the bioreactors.

 

 

 

 

Production Rates“ Total electron equivalentsb
Intermediate (pmol e'-ml' -hr' ) (%)
HS LS HS LS
Acetate 0.191002 0.23i0.02 5.7 6.8
Ethanol 0.25i0.04 None Detected 7.4 -
Propionate None Detected 0.11:1:002 - 3.3

 

a Production rates and 95 % confidence intervals calculated from linear regression
analysis of dpm accumulation overtime.

” Percent total electron equivalents entering the reactor through glucose.

73

Table 3.3. Fermentation of glucose by isolate R8 and maximum substrate

utilization rates in bioreactor R1.

 

Isolate R8
Fermentation Products

(pmol-100 pmol glucose")

Bioreactor R1
Substrate Utilization Rates "

(pmol‘ml'l'hr'l)

 

Acetate 69
Ethanol 101
Prapionate 0
Lactate 20.8
Butyrate 0
Hydrogen 144

0.411015
0.52i0. 14
0.02i0.02

10 hour lag / 0.26i0.04
0083:005

Not Determined

 

a Utilization rates and 95 % confidence intervals were calculated by linear

regression analysis of substrate disappearance curves from duplicate assays.

74

Spirochete isolation and fermentation products. Serial dilution of R1 bioreactor
fluid in agar shake tubes yielded more than 50 uniform colonies in the 108 dilution tube
and no colonies in the 1010 dilution tube (there was no 109 dilution), suggesting that there
were over 5 X 109 spirochetes ml". All colonies were spherical, had a "cottonball-like"
appearance, and were 2-3 mm in diameter, which is characteristic of spirochete colonies.
Several colonies were picked and serially diluted. Microscopic inspection revealed that all
of the isolates were long, thin spiral shaped cells typically between 20 and 50 pm long and
0.4 pm thick, with an average wavelength of 1.2 pm and average amplitude of 0.3 pm.
This morphology was similar to that of the spirochetes seen in both the HS and R1
bioreactor communities. One isolate, R8, was chosen and serially diluted from isolated
colonies two more times. Microscopic observation did not reveal any contaminating
bacteria. The 16S rRNA gene of isolate R8 was 99.5 % similar to S. caldaria and 100 %
similar to clone HS-B26 (Figure 3.2A). The most common T-RFLP fragment in the R1
and HS bacterial communities also corresponded to the fragment size of isolate R8 (Figure
3.1). Acetate, ethanol, and Hz were the major fermentation products of isolate R8, with

lactate being produced in smaller amounts under these culture conditions (Table 3.3).

Discussion

Both microscopy and T-RFLP indicated that Spirochaeta or T reponema species
comprised a surprisingly high percent of the HS community. Spirochaeta species are
found in many anaerobic environments, but they are generally not numerically dominant
(13). Some Treponema species are pathogenic, but most known species are part of the

normal oral, intestinal, and genital flora of mammals. T reponema species are rarely

75

numerically prevalent outside of their eukaryotic hosts (29). F or example, a molecular
study of a firll-scale methanogenic vinasses (winery-waste) bioreactor reported that only 4
% of cloned 16S rRNA genes were spirochete-related. None of these sequences were over
92 % similar to the 16S rRNA sequences in our bioreactors (12). The termite hindgut is
the only environment in which large proportions of Treponema species have been
observed (22). The dominance of spirochetes in the absence of any eukaryotic host
provides a unique opportunity to study their physiology and ecology, and may provide
clues that will aid in the cultivation of the fastidious T reponema species implicated in oral
and venereal diseases.

The agreement between the metabolism of spirochete isolate R8 and the
intermediary metabolites in communities containing high numbers of spirochetes suggests
that spirochetes controlled the carbon flow in these communities at low glucose
concentrations (less than 20 pM glucose). Spirochete isolate R8 fermented glucose to
acetate, ethanol, and small amounts of lactate in pure culture. Spirochaeta caldaria and
other Spirochaeta species typically ferment glucose to acetate, ethanol, H2 and trace
amounts of lactate through the Embden-Meyerhof pathway (4). Both acetate and ethanol
appeared after the addition of [U-”C]glucose to bioreactor HS and were rapidly degraded
during the maximum substrate utilization assay performed with R1 fluid, confirming that
both substrates were produced and utilized in Spirochaetaceae-dominated methanogenic
communities. In addition, T-RFLP results from two restriction digests suggested that
Desulfovibrio species, which have been found to degrade ethanol to acetate in the
presence of Hz-consuming methanogens (6), were present in both HS and R1

communities. Although isolate R8 also produced small amounts of lactate, the long lag

76

time before the utilization of lactate in the substrate utilization assays suggests that lactate
was not an important intermediate in the HS community.

Free-living spirochete species are thought to be specialized competitors for
glucose and other compounds typically produced during the breakdown of cellulose and
similar compounds (13). Numerous mechanisms of enhanced survival and growth at low
nutrient conditions have been documented in free-living spirochetes, including enhanced
chemotaxis under nutrient limitation, dissimilation of amino acids and endogenous RNA
under starvation conditions, and polyglucose storage (4,13). The repeated enrichment of
spirochetes from different inoculum sources in glucose-fed reactors with stable, low
organic loading rates suggests that spirochetes are indeed adapted to grth at low
substrate concentrations. The convergence of the bacterial communities in bioreactors HS
and R1 is an example of such an enrichment. Whether this is the result of enrichment of
spirochetes from the inoculum or from invasion/cross-contarnination is difficult to
conclude without analyzing samples of the original inoculum.

Bioreactor LS contained a very different and perhaps more typical bioreactor
community and carbon flow than bioreactors HS and R1. T-RFLP and microscopy data
suggest that a Streptococcus species was the most common organism in the LS
community (Figure 1 and 2B). Streptococci are commonly found in methanogenic
digesters. Studies of bacterial populations in anaerobic piggery-waste and sewage sludge
digesters found that approximately 30 - 70 % of the bacteria isolated on a habitat-
simulating medium were Streptococcus species (15,38). The accumulation of radioactive
acetate and smaller amounts of prOpionate from radiolabeled glucose agrees with previous

studies in which acetate and propionate generally account for most of the methane

77

production in glucose digestion (8,18). The presence of streptococci, which typically
produce lactate as well as acetate, suggests that lactate would be observed as a
radiolabeled intermediate from [U-“C]glucose in LS; however, lactate is known to be
rapidly fermented to propionate, while propionate oxidation is a much slower and energy-
poor reaction. The fermentation of lactate to propionate may have been rapid enough to
prevent the detection of lactate as an intermediate in this study. Members of genera such
as Selenomonas, Propionibacterium, Anaerovibrio, Pectinatus, and Desulfobulbus are
known to ferment lactate to propionate and acetate, with propionate being the major
product (3,20,32,36). A terminal restriction fi'agment corresponding to a Selenomonas
species was found in the LS community, suggesting that this type of organism may be
responsible for the fermentation of lactate in the LS community.

In contrast to spirochetes, streptococci are usually thought to be adapted to high
substrate concentrations and fast grth rates. In a shock loading experiment with similar
glucose-fed bioreactors, excess glucose was added to four bioreactors communities
containing large proportions of streptococci. The streptococci remained prevalent during
the shock load, and were apparently able to grow rapidly enough to avoid being overcome
by other species. In Spirochaetaceae-dominated bioreactors the Spirochetes were
overgrown by other species after the addition of excess glucose (10). Many organisms are
known to have multiple enzyme systems that are employed under different substrate
concentrations, and some streptococci have been reported to be facultative oligotrophs, i.
e. able to grow under both high and low substrate conditions (16).

If streptococci are capable of growing well under low substrate conditions, why

then did the LS community and Spirochaetaceae-dominated R1 community, which were

78

inoculated simultaneously from the same source, have two very different communities?
The time between inoculation and analysis of the LS community was shorter (210 days)
than for the R1 community (330 days), which could explain some of the difference in the
two communities. The difference in the operation of bioreactors LS and R1 may also have
inhibited a spirochete-rich community from developing in the smaller, variable volume LS
bioreactor; however, both types of bioreactors have maintained spirochete-rich
communities (10). A 185 bp fragment that corresponds to the dominant spirochete in the
HS and R1 communities is present in the LS community T-RFLP pattern, although the
height of the peak suggests that it is a minor component of the community. Given enough
time under continuous low glucose conditions it is possible that this spirochete would have
outcompeted the streptococci for glucose and taken over the LS community as it did in
the R1 community.

This study illustrates that different fermentative organisms establish different
routes of carbon flow in methanogenic communities when substrate concentrations are
low. The importance of fermentative organisms in the overall pathway of anaerobic
degradation has been somewhat overlooked in community analysis of methanogenic
environments, while the organisms that catalyze the energetically more difficult steps of
fatty acid oxidation and methane formation have been emphasized. Although the model
system used here is much simpler than most bioconversion systems and natural
environments, the general observation that different fermentative organisms can impact the
pathway of organic carbon degradation is nonetheless applicable over a wide variety of
environments. Including the physiologic properties of the fermentative organisms in

methanogenic communities, such as varying substrate affinity, starvation survival and acid

79

tolerance, will improve our ability to model anaerobic degradation and methane

production in natural environments.

References

l. Amann RI, Ludwing W, Schleifer K (1995) Phylogenetic identification and in situ
detection of individual microbial cells without cultivation. Microbiol Rev 59:143-
169

2. Brower JE, Zar JH, van Ende CN (1990) Field and Laboratory Methods for General
Ecology. Wm.C.Brown, Dubuque, IA

3. Bryant MP (1984) Genus IX. Selenomonas. In: Krieg NR, Holt JG (eds) Bergey's
Manual of systematic bacteriology, vol I. The Williams & Wilkins Co., Baltimore,
pp 650-653

4. Canale-Parola E (1977) Physiology and evolution of spirochetes. Bacterial Rev 41 :181-
204

5. Cappenberg TE, Prins RA (1974) Interrelations between sulfate-reducing and methane-
producing bacteria in the bottom deposits of a fresh-water lake. III. Experiments
with l4C-labeled substrates. Antonie van Leeuwenhoek 40:457-469

6. Chartrain M, Zeikus JG (1986) Microbial ecophysiology of whey biomethanation:
characterization of bacterial trophic populations and prevalent species in
continuous culture. Appl Environ Microbiol 51:188-196

7. Chartrain M, Zeikus JG (1986) Microbial ecophysiology of whey biomethanation:
intermediary metabolism of lactose degradation in continuous culture. Appl
Environ Microbiol 51 : 1 80- 187

8. Cohen A, Van Deursen A, Van Andel J G, Breure AM (1982) Degradation patterns and
intermediates in the anaerobic digestion of glucose: Experiments with l4C-labeled
substrates. Antonie van Leeuwenhoek 48337-3 52

9. Dazzo FB, Petersen M (1989) Applications of computer-assisted image analysis for
microscopical studies of the Rhizobium-legume symbiosis. Symbiosis 7:193-210

10. Fernandez A, Hashsham S, Dollhopf SL, Raskin L, Glagoleva O, Dazzo FB, Hickey
RF, Criddle C, Tiedje JM (2000) Flexible community structure correlates with
stable community function in methanogenic bioreactor communities perturbed by
glucose. Appl Environ Microbiol 66:4058-4067

1]. Fernandez A, Huang S, Seston S, Xing J, Hickey RF, Criddle C, Tiedje JM (1999)

80

How stable is stable? Function vs community stability. Appl Environ Microbiol
65:3697-3704

12. Godon J, Zumstein E, Dabert P, Habouzit F, Moletta R (1997) Molecular microbial
diversity of an anaerobic digestor as determined by small-subunit rDNA sequence
analysis. Appl Environ Microbiol 63 :2802-2813

13. Harwood CS, Canale Parola E (1984) Ecology of spirochetes. Annu Rev Microbiol
3 8: 161 -1 92

14. Hashsham S, Fernandez A, Dollhopf SL, Dazzo FB, Hickey RF, Tiedje JM, Criddle C
(2000) Parallel processing of substrate correlates with greater functional stability in
methanogenic bioreactor communities perturbed by glucose. Appl Environ
Microbiol 66:4050-4057

15. Hobson PN, Shaw BG (1974) The bacterial population of piggery-waste anaerobic
digesters. Wat Res 8:507-516

16. Ishida Y, Imai I, Miyagaki T, Kadota H (1982) Grth and uptake kinetics of a
facultatively oligotrophic bacterium at low nutrient concentrations. Microb Ecol
8:23-32

17. Jerris J S, McCarty PL (1965) The biochemistry of methane fermentation using 14C-
tracers. J Wat Pollut Control Fed 37: 178-192

18. Kaspar HF, Wuhrrnann K (1978) Kinetic parameters and relative turnovers of some
important catabolic reactions in digesting sludge. Appl Environ Microbiol 36: 1-7

19. Leadbetter JR, Schmidt TM, Graber JR, Breznak JA (1999) Acetogenesis from H2
plus C02 by spirochetes from termite guts. Science 283:686-689

20. Lee SY (1984) Genus XI. Pectinanus. In: Krieg NR, Holt JG (eds) Bergey's manual of
systematic bacteriology, vol I. The Williams & Wilkins Co., Baltimore, pp 657-658

21. Leschine SB, Canale-Parola E (1986) Rifampin-resistant RNA polymerase in
spirochetes. FEMS Microbiol Lett 35:199-204

22. Lilburn T, Schmidt TM, Breznak JA (1999) Phylogenetic diversity of termite gut
spirochaetes. Environ Microbiol 1:331-345

23. Liu J, Dazzo FB, Glagoleva 0, Yu B, Jain A (2001) CMEIAS: A computer-aided
system for the image analysis of bacterial morphotypes in microbial communities.
Microb Ecol 41

24. Loffler FE, Ritalahti KM, Tiedje JM (1997) Dechlorination of chloroethenes is
inhibited by 2-bromoethansulfonate in the absence of methanogens. Appl Environ
Microbiol 63 :4982-4985

81

25. Mackie R1, Bryant MP (1981) Metabolic activity of fatty acid-oxidizing bacteria and
the contribution of acetate, propionate, butyrate, and C02 to methanogenesis in
cattle waste at 40 and 60°C. Appl Environ Microbiol 41 : 1363-1373

26. Maidak BL, Cole JR, Parker CT, Garrity GM, Larsen N, Li B, Lilburn TG,
McCaughey MJ, Olsen GJ, Overbeek R, Pramanik S, Schmidt TM, Tiedje JM,
Woese CR (1999) A new version of the RDP [Ribosomal Database Project].
Nucleic Acid Res 27: 171-173

27 . Marsh TL, Saxman P, Cole J, Tiedje JM (2000) Terminal restriction fragment length
polymorphism analysis program, a web-based research tool for microbial
community analysis. Appl Environ Microbiol 66:3616-3 620

28. McCartney DM, Oleszkiewicz JA (1993) Competition between methanogens and

sulfate reducers: Effect of COD:sulfate ratio and acclimation. Wat Environ Res
65:65 5-664

29. Miller JN, Smibert RM, Norris SJ (1992) The genus T reponema. In: Balows A,
Truper HG, Dworkin M, Schleifer K (eds) The Prokaryotes, A Handbook on the
Biology of Bacteria: Ecophysiology, Isolation, Identification, Applications, vol 2.
Springer-Verlag, New York, NY, pp 3537-3555

30. Oremland RS (1988) Methanogenic bacteria. In: Zehnder A (ed) Biology of Anaerobic
Microorganisms. Wiley, New York, NW

31. Pearson K (1926) On the coefficient of racial likeness. Biometrika 18:105-117

32. Prins RA (1984) Genus X. Anaerovibrio. In: Krieg NR, Holt JG (eds) Bergey's
manual of systematic bacteriology. The Williams and Wilkins Co., Baltimore, pp
653-655

33. Raskin L, Rittmann BE, Stahl DA (1996) Competition and coexistence of sulfate
reducing and methanogenic populations in anaerobic biofilms. Appl Environ
Microbiol 6223847-3 857

34. Shelton DR, Tiedje JM (1984) Isolation and partial characterization of bacteria in an
anaerobic consortium that mineralizes 3-chlorobenzoic acid. Appl Environ
Microbiol 48:840-848

35. Smith PH, Mah RA (1966) Kinetics of acetate metabolism during sludge digestion.
Appl Microbiol 14:368-371

36. Stams A-JM, Kremer DR, Nicolay K, Weenk GH, Hansen TA (1984) Pathway of
propionate formation in Desulfobulbus propionicus. Arch Microbiol 139: 167-173

37. Strunk O, Ludwig W (1997) ARB: Software for phylogenetic analysis. Technical
University of Munich, Munich, Germany

82

38. Ueki A, Miyagawa E, Minato H, Azuma R, Suto T (1978) Enumeration and isolation
of anaerobic bacteria in sewage digestor fluids. J Gen Appl Microbiol 24:317-332

39. Wilcox CD, Dive SB, McDavid WD, Greer DB (1997) UTHSCSA ImageTool,
University of Texas Health Science Center at San Antonio

83

CHAPTER 4
INTERPRETING 16S RDNA T-RFLP DATA: APPLICATION OF SELF-
ORGANIZING MAPS AND PRINCIPAL COMPONENT ANALYSIS T0 DESCRIBE

COMMUNITY DYNAMICS AND CONVERGENCE

These results will be published in the article: Dollhopf, S.L., Hashsham, SA, and
J .M. Tiedje. 2001. Interpreting 16S rDNA T-RFLP data: Application of self-organizing
maps and principal component analysis to describe community dynamics and

convergence. Microb. Ecol. In Press.

Introduction

The use of molecular methods in microbial ecology has greatly expanded our
ability to describe and monitor microbial communities; however, the collection of large
amounts of data about a particular community does not necessarily lead to an
understanding of the community, especially when noise and bias cloud data analysis. The
problem of interpreting these data hinders our ability to learn what general principles
control microbial diversity, community composition, interspecies interactions, and
microbial processes. The management of microbial communities for practical
applications, such as bioremediation and waste treatment, is also greatly impeded by our
inability to predict community dynamics and function under variable environmental
conditions. To better understand microbial communities, methods need to be developed
that can integrate large amounts of environmental and molecular data as well as

recognize patterns in data that may be relatively noisy.

84

Typically, microbial ecologists have relied on conventional statistical methods,
such as correlation analysis and principal component analysis (PCA), for comparison of
complex communities. These methods do not detect non-linear relationships, are
confounded by experimental variability, and require normal or log-normal distributed
data. Artificial neural networks (ANN 5) may provide an alternative to conventional
statistical methods because they detect non-linear relationships, allow the visualization of
complex data, and remain robust despite experimental variation. In particular, Kohonen
self-organizing maps (SOMs) have been used extensively in biological research for
pattern recognition. Phylogenetic reconstruction, classification of proteins, and genomic
analysis are a few applications of SOMs in molecular biology (18,31,33). Ecological
applications for SOMs are also being explored for classification and modeling of
populations and ecosystems (13,20). Only one study in microbial community analysis has
previously used an ANN for community analysis and classification (25).

This study compares the ability of principal components analysis and self-
organizing map neural networks to analyze 16S T-RFLP profiles of microbial
communities in continuously-stirred, glucose-fed methanogenic bioreactors. The goal
was not only to identify which samples were similar, but also to decipher community
dynamics and describe specific phylotypes, i.e. phylogenetically similar organisms, that
behaved similarly in different reactors. We observed that the composition of both reactor
communities converged over the course of the experiment. In addition, we were able to
identify phylotypes that occurred in both reactors in different phases of reactor operation.

SOM analysis proved most useful in identifying all phylotypes that behaved similarly in

85

both communities, while PCA ordination clearly demonstrated community stability and

dynamics during different periods of reaction operation.

Materials & Methods

Reactor operation and experimental design. Three 16-liter glucose-fed
continuously stirred tank reactors were operated anaerobically at 34°C with a loading rate
of 0.5 g glucose-l'l-day'l and a hydraulic retention time (HRT) of 16 days by addition of
1.0 l-day'1 of buffered glucose solution (8.0 g glucose-1'l + 7.0 g NaHCOyl'l) plus 10
ml-day'1 of concentrated nutrient solution (34). Reactor R1 (Rumen 1) was inoculated
from rumen fluid and operated under the above conditions for over 330 days prior to this
study (Figure 4.1). The other two reactors were inoculated from sediment of the Red
Cedar River in Lansing, MI and from a 250 ml variable volume reactor seeded with
sewage sludge from a previous study (17). These reactors were designated reactor 82
(Sediment 2) and reactor M3 (Municipal 3), respectively. Reactor chemical oxygen
demand (COD), volative fatty acids (VFAs), and pH were monitored biweekly.

The operation of the reactors was divided into two experimental phases, phase A
and phase B (Figure 4.1). Phase A began when all reactors had achieved stable operation
(COD removal above 90 %) at the target organic loading rate of 0.5 g glucose-liter'l'day“
and lasted for 14 volume changes (224 days). All Operating conditions were constant and
percent COD removal remained above 90 % during phase A in all three reactors. During
phase B, the effect of HRT on the microbial community was investigated by decreasing
the HRT incrementally. The IRT of reactors 82 and M3 was manipulated by increasing

the flow rate of the buffered glucose and nutrient solutions, thus increasing the organic

86

Phase A I Phase B
l

 

 

 

 

1.0 1.0 Flow Rate
R1 3 (Uday),
HRT
l
1.0 2.0 3.3 4.0 8.0 Flow Rate
“ (Uday)
0 0 0 0 @119 CD ‘10 ‘9 '39 G) ‘9 (D (~19 ~
16 days 8 days 6 days 4 days 2 days HRT
i
1.0 2.0 . 3.3 . 4.0 8.0 Flow Rate
” ‘ (I/day)
M3 9 9 0 0 ‘96) ‘1) ‘E ‘2’ ‘9 ‘9 ‘D 69 #3 (It
16 days 8 days 6 days 4 days 2 days HRT

Figure 4_1_ Timeline for reactors R1, 82, and M3. Time is measured in number of
volume changes (x-axis not to scale) Sampling times are indicated by numbered circles
which show the number of volume changes that occurred since the target organic loading
rate was achieved in reactors 82 and M3. Control reactor R1 operated for over 2 years
prior to this study. Reactors $2 and M3 were inoculated from river sediment and
municipal sewage sludge, respectively. F low rate and HRT are shown above and below
the timeline respectively.

87

loading rate as well. Each new flow rate was maintained for at least nine volume changes
before a subsequent increase. The HRT of the control reactor was constant throughout
phase B. Percent COD removal remained above 85 % during phase B in all three reactors
with the exception of one sample taken a few days after a short-terrn pump failure after
37 volume changes (data not shown).

Sampling. After reactors 82 and M3 had achieved a stable operating efficiency at
the target organic loading rate, samples (2 mL) from each reactor were taken after 3, 5, 7,
9, 11, 12, and 14 volume changes (Figure 4.1). During phase B, samples (2 mL) from
each reactor were taken immediately before each change of the HRT. Occasionally
samples were taken at other times during phase B to check community stability (Figure
4.1). Samples were stored at -80°C until DNA extraction and T-RFLP analysis.

T—RFLP profiles. DNA was extracted from 2 mL of reactor fluid using a bead-
beater/phenol-chloroform method (22). Bacteria or archaea community 16S rDNA was
amplified using the primer pair 8F-Hex and 1392R (2) or 21F-FAM and 951R (8),
respectively, digested with restriction enzyme HaeIII, loaded on an automated sequencer
(model 373A, Perkin Elmer Applied Biosystems), and processed with GeneScan (v2.1;
Perkin Elmer Applied Biosystems) as previously described (10). Key samples were also
digested with MspI and HhaI to aid in the identification of matching 16S rRNA genes
from clone libraries.

Variability of the relative intensity of each fragment was examined by analyzing
three replicate samples from one reactor in parallel. The same 13 fragments were found
in each replicate and the relative intensity, as ascertained by peak height measured in the

electropherogram, had a standard deviation of 6 %. When the same replicates were

88

amplified and run on three separate T-RFLP gels, the standard deviation increased to 16
%. Only one sample and T-RFLP profile was analyzed for all other time points.

Clone libraries and sequencing. Bacterial clone libraries were created from
reactor 82 samples taken after 3 and 56 volume changes. An archaeal clone library was
created from a reactor M3 sample taken after 3 volume changes. Community DNA
extraction and amplification were performed as described above, except that non-labeled
PCR primers were used for amplification of community 16S rDNA. After amplification
of the community DNA with the proper primer set, the amplicons were cloned into E.
coli with the TA cloning kit (Invitrogen, Carlsbad, CA). Screening and sequencing were
performed as previously described (12). Briefly, the cloned 16S rDNA gene fragments
were amplified from whole cells using plasmid specific primers, checked for the proper
size, and digested with HaeIII to screen for 168 genes that had terminal restriction
fragments matching those found in community T-RFLP analyses. T -RFLP analysis was
also performed on each clone to confirm the size of their terminal restriction fragment for
the enzymes HaeIII, MspI, and HhaI.

Only clones matching predominant terminal restriction fragments found in both
reactors were sequenced and analyzed. The amplified gene fragments were then purified
with Microcon-IOO spin columns (Millipore Corp), and sequenced via the dye terminator
method by the Michigan State University DNA sequencing facility with a universal
primers targeting conserved regions of the 168 rRNA molecule. Approximately 1400
nucleotides were sequenced for clones L and I because they were divergent from
cultivated phyla. Only partial sequences were obtained for the rest of the clones.

Sequences were evaluated with the Check Chimera program of the Ribosomal Database

89

Project (RDP) and judged not to be chimeric
(http://wwwcme.msu.edu/RDP/html/jgalyseshtml). Alignment of sequences, mask
construction, and dendrogram construction were performed with ARB (27). The .
sequences determined in this study have been deposited in the GenBank database under
accession no. AF338765 to AF338766 and AF339848 to AF339859.

Community similarity. The Morisita index of community similarity (1M), which
is based on Simpson’s dominance index, was calculated in Ecostat software (Trinity

Software, Plymouth, NH) using the formula:

_ 22 nunz,

I _.
M (11+12)N1N2

 

where n, is the number of individuals of species i, N is the total number of individuals
sampled, and l is Simpson’s dominance index for each community. Simpson’s dominance

index (I) was calculated using the formula:

£02.02. — 1))
_ N(N - 1)

 

where sis the total number of species in the community (5,32). This index ranges from 0-
1, with 0 indicating that no species are shared between the two communities and 1
indicating complete identity. Because the index takes species abundance into account,
communities that contain the same species but have different species abundance will have
an index value less than 1. These equations were adapted to T-RFLP data by considering
each terminal restriction fragment a separate species and peak height a measure of

species abundance.

90

PCA. The data sets were tested for normality with MINTTAB (v. 13.3]; Minitab
Inc., State College, PA) using the Anderson-Darling, Ryan-Joiner and Kolmogorov-
Smirnov methods. PCA ordinations were calculated in StatView (v. 5.0.1; SAS Inc,
Cary, NC) using correlation matrices produced from T -RFLP profiles that were
normalized such that the sum of the peak heights of identified fragments equaled 10,000
fluorescent units. The complete data set consisted of 15 T-RFLP profiles (one for each
sampling date) with the peak height of 66 unique fragments indicated for each profile.
The data set was separated into individual analyses for each experimental phase and
primer set to reduce the number of factors required to account for 85 % of the variation
observed. All terminal restriction fragments with a peak height greater than 100 units
were included in the analyses.

SOM. Self organizing maps organize large multidimensional data sets by
mapping the input data onto a lower-dimensional array (19). A SOM has two layers: an
input layer, which has the same number of nodes as there are input variables (e. g.
species) and an output layer, which is a 2-D array of nodes with associated weights. Each
input node is connected to every output node. During training, the output nodes, which
initially have randomized weights, compete for the best match to the input. The output
node that best matches the input becomes the winner. The winning node and its
neighboring nodes are then updated so that it is more sensitive to the input with which it
was presented. This competition occurs repeatedly over a set number of iterations,
resulting in clusters of winning nodes that correspond to clusters within the input data
(19). To verify the clustering pattern observed in the output, the training is redone a

number of times, randomizing the initial weights each time. Because the initial weights

91

are randomized, the area of the output in which specific inputs cluster changes from one
round of training to the next, but the grouping of the inputs should remain essentially the
same if the SOM is accurately mapping the input on to the output space (7,25).

All neural computing was performed in NeuroSolutions software (v3.022,
Consultants Level; NeuroDimensions, Gainesville, FL). The complete input data set
included 15 T-RFLP profiles from each reactor containing 46 terminal restriction
fragments from both the bacteria and archaea 16S rDNA primer sets. Terminal restriction
fragments that were only observed once were not included. The T-RFLP profiles from
each reactor were analyzed independently. Data were normalized so the sum of the peak
heights of identified fragments equaled 10,000 fluorescent units, and then transformed
within NeuroSolutions so that all values were between 1 and -—1. The network parameters
were selected based on trial runs and suggested values fiom NeuroDimensions. The input
layer consisted of 46 nodes and the output layer consisted of a square Kohonen map of 15
X 15 nodes (225 output nodes total). The distance between the input and output nodes
was calculated using the Euclidian distance measure. The Kohonen neighborhood size
was initially set to 10 and then decreased linearly by 0.05 over 200 iterations. The initial
learning rate of 0.01 also decreased linearly by 5 X 10" over 180 iterations. Five separate
training runs of 500 iterations each were performed for the M3 and S2 data sets to

confirm the clustering patterns within the SOM output layer.

Results

Community similarity. The 82 and M3 communities were very different at the

beginning of phase A as measured by the Morisita community similarity (Table 4.1).

92

Table 4.1. Morisita community similarity indices for the R1,

82, and M3 bioreactor communities.

 

Sample“ Communities compared

 

$2 to M3 S2 to R1 M3 to R1

3 0.05 0.02 0.30
0.05 0.14 0.32

11 0.11 0.01 0.15
14 0.16 0.49 0.62
20 0.72 0.79 0.78
29 0.72 0.61 0.63
39 0.39 0.66 0.43
49 0.84 0.72 0.63
56 0.71 0.33 0.51

 

“ Sample numbers described in Figure 4.1.

93

During phase A, similarity indices between $2 and M3 varied, but by the end of phase A,
comparison of the two communities yielded a similarity value of 0.72. Community
similarity values between 82 and M3 remained above 0.70 for the rest of the experiment,
except for sample 39 which was taken after a short-terrn pump failure in reactor M3. Both
S2 and M3 communities were also similar to the R1 control community by the end of
phase A. Only the application of a two day HRT caused the 82 and M3 communities to
depart from the control (Table 4.1).

PCA. Distribution analysis of the data with the Anderson-Darling, Ryan-Joiner
and Kolmogorov-Smimov methods all indicated that the data varied significantly from a
normal distribution. The data was skewed to the right, with more species having a
relatively low abundance and few having a very high abundance. However, it has been
shown that PCA can provide meaningfirl ordination of non-normal data (28,29). PCA
ordination of the entire set of T-RFLP profiles for each reactor was very difficult to
interpret because more than three factors were required to account for 85 % of the
variability within the data. To allow a more detailed analysis of the bacteria and archaea
communities, the T-RFLP profiles were separated into data sets for each reactor,
experimental phase, and primer set, resulting in four different PCA ordination plots for
each reactor.

The microbial community in control reactor R] changed very little over the
course of both experiments, and PCA of both the bacteria and archaea data sets fiom

phase A and phase B periods extracted only one principal component factor (data not

shown).

94

 

 

 

 

 

 

A B
A 0.5 1'» 0.5
3 s o g
2 a
:7 a as 9-5 7° . z; m
‘- 9 I...
° II c
E 14 f2 13
c- a:
.051. -0.5
-1» -1
Factor 1 (70 °/o) F actor 11 (74 %)
1 r T
C i 5 0° D
o 7 0 49
3 56 %
0.5 «1- 0-5 i 52
A ’9‘
.\° .\
n N
64.1 .05 05 125-1 -0.5 0.5 l
a; r r A i N .L r ‘- —i
‘5 4 ‘5
E 01 32 E 2‘ 39
7' 43,5, 9 ll “‘ ,0, i 20 9:9
l l ’
4 _ -l at
Factor 11 (46 %) Factor 1 (75 %)

 

Figure 4.2. PCA ordination of bacteria T-RFLP peak height data for
reactors M3 (A and B) and S2 (C and D) during phase A (A and C)
and Phase B (B and D). The amount of variability accounted for by
each factor is indicated on each axis. Sampling times are next to each

point.

95

(i) Phase A. The M3 bacteria community changed during the first 7 volume
changes of phase A, but was more stable after nine volume changes, as indicated by the
tight clustering of samples 9, 11, 12, and 14 in a graph of Factor 1 versus Factor 2 (Fig
4.2A). The amount of explained variability in the data represented by the first factor
increased with time. Restriction fragment 225 accounted for most of the variability in
factor 1 (Figure 42A and Table 4.2). This fragment appeared five volume changes after
start-up and increased steadily for six volume changes, after which its abundance
stabilized. The variability represented by factor 2 decreased over time. Four fragments
that were abundant during the beginning of phase A had high loading values for factor 2
(Table 4.2).

The $2 bacteria community also changed between 7 and 9 volume changes, and
remained relatively stable thereafter (Figure 4.2C). The variability represented by factor 1
increased over time while that represented by factor 2 decreased. Four fragments
explained most of the variability with factor 1, with fragment 185 having the highest
loading (Table 4.2). Six terminal restriction fragments that were abundant early and then
decreased were associated with factor 2 in the S2 community. Of these six fragments, two
were also associated with factor 2 in the M3 community. A third factor was retained that
accounted for 13.6 % of the variability in the 82 community during phase A. Five
terminal restriction fragments that were abundant only in the sample taken seven volume
changes after start-up had high loadings for this third factor (Table 2).

(ii) Phase B. During phase B, the M3 bacteria community was quite stable until
the HRT was decreased to two days, as shown by the tight clustering of samples before

the two day HRT (Figure 4.2B). The one outlying sample before the two day HRT was

96

Table 4.2. T-RFLP fragments with high factor loading.

 

 

 

 

 

Reactor Phase A Phase B

Factor 1 Factor 2 Factor 3 Factor 1 Factor 2
M3 225 78, 161, NA“ 115, 135, 75, 125,
203, 248, 185, 225 253, 258
S2 185, 250, 69, 78, 61, 67, 75, 61, 67, 78, 253,
253, 286 203, 234, 125, 318 75, 125, 258, 309,

267,280 185

“Not applicable

97

likely the result of a temporary pump failure directly before the sampling date. The 82
bacteria community was also stable for the majority of phase B, with only four and two
day HRTs affecting the bacteria community (Figure 4.2D). In both communities the value
of factor I changed very little during phase B. Four fragments in M3 and five fragments
in 82 had high loadings for factor 1. The only fragment with a high loading for factor 1 in
common between both communities was 185 base pairs in length. The amount of
variability explained by factor 2 increased at higher flow rates. Four fragments in each
community had high loadings for factor 2 with fragments 253 and 258 in common
between the two communities (Table 4.2).

(iii) Archaea. The M3 archaea community changed very little during phase A and
phase B. PCA extracted only one factor for each data set. Only during phase A in reactor
82 did significant changes occur in the archaea community. Two factors were extracted
and the amount of variability explained by both changed over the 14 volume changes of
phase A (data not shown). Values for factor 1 (61 %) decreased over time and the
terminal restriction fragments A184, A189, and A215 had high loadings for this factor.
Values for factor 2 (24 %) increased overtime. Only fragment A217 had a high loading
for this factor. Changes in HRT had little effect on the 82 archaea community, and only
one factor was extracted from the 82 HRT archaea data set.

SOMs. Separate analyses of the M3 and 82 T-RFLP data resulted in four
discernible groups of terminal restriction fragments in both outputs (Figure 4.3). The
SOM grouped the fragments according to when they were most abundant in each reactor.
As one moves counter-clockwise from the top left around the square SOM output layer,

the time in which the fragments are abundant in the reactor progresses. Thus, fragments

98

Figure 4.3. SOM clustering of T-RFLP fragments in reactor M3 (A) and S2 (B). The line
graphs shown for each fragment represent the change in normalized peak height over
time. The graphs are arranged according to the position of their winning node within the
square SOM output. SOM groups I, 11, HI and IV are indicated. The letter A before the
fragment length indicates fragments from Archaea-specific T-RFLP. Fragment lengths
indicated in color are in the same SOM group in both reactors. (C) A timeline shows

when groups I through IV were present in each reactor. This figure is presented in color.

99

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CE = _3:

 

 

 

IQDD.

 

 

 

MIMN mmm wk. mom owN
=lz lat! Q E: L: _
mm 863% H9890 m

reins

11 lrt‘é

1e ill“

1 til—r

  

 

 

 

 

a: a3. a: a: E S : o A m m .
w “ n J_ n n n w _- 4_ [_ L S
>2 I E = = H
a: a: a: a3 3 E : a A m m
_ E _ _ _ r u r L _ C m:
n . l m _ _ J _ _ a [
Z I E a L o
5.380 H.555 5980 fig
figF—Emww—E @WWJ Q19 .9 mm 82 SN EN ”R 9.: 2m mom NM: _
E 2: m: on 3 o: E E 9.” mum 8m 3m
BBB a. E EEE BEES
_ ms 3 S
2m 2m 2: T4... ‘gdL m: a: :< 2? EN 8m 8m 3m
33% 22.... ESE a DEE:
WEN... SN 48 Illllll a?
as .EQE a. E
E .- ED E a: me as as
2m _\8m in NR Em. 2?. B E! EEIQ
3m 3m a2 a new £2 a; 82 $2

 

100

in the top left were most abundant at the beginning of phase A, while those in the top
right dominated late in phase B. Fragments located between groups were intermediate
between those groups. Fragments located in the center of the SOM either appeared
sporadically or were present at constant levels throughout both experiments.

The four groups found in each reactor were very similar; however, they were not
completely synonymous (Figure 4.3C). Group I was the same in both reactors and
contained terminal restriction fragments that were abundant very early in phase A and
then reduced drastically or disappeared afterwards. Group H fragments were abundant
during the end of phase A in both reactors; however, group II fragments in reactor 82
persisted longer than those in reactor M3, and had a bimodal abundance pattern (Figure
4.3B).

Group III fragments persisted the longest in both reactors, and were most
abundant from the end of phase A to the beginning of the four-day HRT, thus they
persisted in the reactors for over 300 days. Most of the group III fragments did not
disappear after the four-day HRT started, but decreased in abundance while the group IV
fragments increased. Group IV contained fragments that increased after the HRT of the
reactors was decreased to four days or less. In reactor 82, the pattern was slightly
different than in reactor M3. Group IV fragments in reactor 82 increased transiently 12
volume changes after start-up and then increased again after the HRT was decreased to
four days. The transient increase in group IV fragments in reactor 82 corresponds with

the brief decrease in the fragments of group H in the same reactor (Figure 4.3).

101

Those fragments located in the center of the output were present at constant levels
throughout both experimental phases. In reactor M3, fragments that appeared at fairly
constant levels included the archaea derived fragments A221, A215, and A217.

Terminal restriction fragment identification. (i) Bacteria. Six clones were
sequenced that matched terminal restriction fragments found in both communities. Four
of these six clones matched fragments that were among the eight terminal restriction
fi'agments (out of 48 bacterial fragments identified) that had a maximum normalized peak
height greater than 25 % of the total peak height in a given profile. Of these four
fragments, two of them (fragments 185 and 253) accounted for over 50 % of the total
peak height of a profile in at least one sample. The other two clones sequenced matched
fragments that had a maximum normalized peak height of over 10 % of the total peak
height in a given profile.

From the phase A bacteria clone library from reactor 82 (sample 3), four 163
rDNA clones were retrieved that matched predominant T-RFLP fragments (Figure 4.4).
Clones L and J matched terminal restriction fragments 78 and 203, respectively, and were
classified to SOM group I in both S2 and M3 reactors. T-RFLP analysis of S2 and M3
communities with three restriction enzymes confirmed the presence of matching
fragments for these clones in all three digests, suggesting that the shared T-RFLP
fragments indicate the presence of similar organisms in reactors 82 and M3. Clone L was
95.8 % similar to a sequence found previously in a glucose digesting reactor, and was
also closely related to sequences cloned from an anaerobic vinasses waste digester

(12,15). However, Clone L had only 77 % sequence similarity to any cultivated organism

102

Figure 4.4. Maximum likelihood tree of cloned Bacteria 16S rRNA genes from reactors
M3 and 82 and other methanogenic environments. Clones retrieved in this study are in
bold. The SOM group and terminal restriction fragment length (Group x : y bp) is shown
to the left of each clone. The backbone of this tree was cast in ARB using 969
unambiguously aligned nucleotides between E. coli positions 95 and 1295. Partial
sequences were appended to the tree in ARB using the maximum likelihood algorithm
with 261 unambiguously aligned nucleotides between E. coli positions 154 and 474. The
origin of cloned genes from other methanogenic environments is indicated as follows:
EO, LS, HS = glucose-fed methanogenic reactors (11,12); HA = anaerobic vinasses

163910? (15); RFS = termite hindgut (21); WCHBl = hydrocarbon contaminated aquifer
(9)-

103

>3 5..
.6033

 

Candidate Divison OP 11

__L-_:4minobacterium mobilis
Aminobacterium colombiense
Clone P Group IV : 248 bp
Clone K Group I : 248 bp
HA 73
Anaerobaculum thermoterrenum

[BO-VIII
EO-V
Spirochaeta sp. str. R8 0.10

HS I

EO-IX

Clone S Grou III : 185 bp
Spirochaeta ca ria

Clone H Group II & IV : 67 bp
l|:LS I
EO-I
l Spirochaeta sp. str. Antarctic
—-—-WCI-IBl-63

—r:fl°HB"44
WCHB 1-3 1

____j Clone R Grou IV : 253 b
"i C lostridium ramgsum p

Streptococcus bovis
LS III
LS II

Streptococcus macedonicus
g {Eubacterium barkeri

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CloneL Grou 1:78b
7 'L—HA 49 p p

_[' Syntrophobacter sp.
___l Flyntmphobacterfirmaroxidans
Desu firrom

onas acetoxidans

 

 

 

 

 

 

 

 

_[— Escherichia coli
Verrucomicrobium sp

l Clone J Group I : 203 bp
' Thermotoga maritima

 

 

 

 

Methanobacteriumformicicum

104

represented in the RDP database. Clone J grouped with llrermotoga-related sequences,
but was 20 % divergent from any isolated organism (Figure 4.4).

Clone H matched the 67 bp terminal restriction fragment found in group II in
reactor 52 and group IV in reactor M3. This sequence was most similar to spirochete-
related sequences cloned from glucose digesting reactors on two separate occasions
(11,12). These sequences group with unusual spirochetes that are reported to lack cell
walls and the typical spiral morphology (14). Clone K matched the 248 base pair
restriction fragment and was 96.2 % similar to a cloned sequence from an anaerobic
vinasses waste digester (15). This sequence clustered with the Anaerobaculum
thermoterrenum group that is separate from all other bacteria divisions (23) and includes
many sequences cloned from anaerobic habitats, such as Aminobacterium mobilis, an
organism isolated from anaerobic sludge that degrades amino acids (3), and Synergistes
jonessi, a rumen organism that degrades amino acids and pyridinediols (1).

In the phase B bacteria library from reactor 82 (sample 56), three clones were
found that matched predominant terminal restriction fragments found in both reactors.
Clone S matched the 185 base pair terminal restriction fragment classified to SOM group
III in both reactors (Figure 4.4). Restriction digests of $2 and M3 communities during
this period with HhaI and MspI revealed predominant fi'agments that also matched with
those of clone S. Clone S was 99.5 % similar to S. caldaria, and other sequences closely
related to this group have been observed in seven different glucose-fed methanogenic
reactors in four different studies (lo-12,34). Clone R matched the 253 base pair terminal
restriction fragment classified to SOM group IV in both reactors. Again, additional

restriction digests confirmed the presence of matching fragments in both communities.

105

Clone R was 100 % similar to Clostridium ramosum and to a cloned sequence from a
perturbed glucose anaerobic digester (11). The third clone, Clone P, was 99 % similar to
Clone K from the phase A bacteria library and matched terminal restriction fiagment 248.

(ii) Archaea. The archaea clone library from reactor M3 yielded four different
16S rDNA sequences after screening of 30 clones. Clones AA and AB were 98.5 % and
99 % similar to Methanosarcina sicilae and had a terminal restriction fragment of 220 bp
(Figure 4.5). In silico digestion of the RDP database with the TAP T-RFLP program
(http://wwwcme.msu.edu/RDP/html/apalyses.html) (24) revealed that a 220 bp fragment
was indicative of sequences related to Methanosarcina barkeri and Methanosarcina
siciliae, while a 217 bp fragment was indicative of sequences related to Methanosarcina
mazei. No 16S rDNA clones were obtained that had a 217 bp terminal restriction
fragment, but a clone found previously in a similar bioreactor matches this restriction
fragment (12). The in silico digest also indicated that the only archaea sequence that
yields a 215 bp fragment after digestion with HaeIII was Methanosaeta concilii (Figure
4.5). Digestion with Hhal confirmed the presence of matching fragments in both
communities for the different Methanosarcina and Methanosaeta species. Two other
clones, AC and AD, composed l7 % of the archaea clone library and grouped with
Methanobacterium species. Clones AC and AD had a HaeIII terminal restriction
fragment of only 22 base pairs, which is indistinguishable from the primer fragments in
T-RFLP analysis. Fragments matching this clone were found in HhaI digests. No other

clone types were detected.

106

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A 7 Methanosarcina barkeri j
98 Methanosarcina siciliae 220
—fl5—— 7 Clone AA
5 Clone AB A
100 7 Methanosarcina mazei T 217
95 AO-I I
Methylcoccoides methylatens 3215
Methanosaeta concilii 3215
Methanobrevibacter ruminantium ‘7
Methanobacterium formicicum
Methanobacterium bryantii
95 AO-III
L— Clone AD
Methanobacterium defluvii
Methanopyrus kandleri
B
50 1.00 L 0 200 250 3.00 350
45001
3090: iii/gm“ 215 b
15901
0; - 4
313881 82 Archaea 217 b
1 <—
7 P
1 590 i VCs
01 A_ A
75_Oi $2 Archaea
500 1
25:0: 12 VCS «220 bp
0i , R
31388 SZ Archaea
”90% 14 VCs
0- A _ - L -
313383 M3 Archaea
_ . 4—220 b
1590, 7 VCs P
or , 4g

 

 

 

 

 

107

Figure 4.5. Succession of methanogenic thIOWPCS in “33010? 32- (A) Neighbor -joining
tree of cloned Archaea 16S rDNA genes. The number next to each node indicates
bootstrap values from 100 parsimony bootstraps. Numbers to the right of each bracket
indicate the length of the 16S rDNA terminal restriction fragment for each phylogenetic
group. (B) Archaea community T-RFLP profiles of reactors $2 and M3 after different
volume changes (V Cs) in phase A.

Discussion

Principal components analysis is a very useful tool in ecological studies. PCA is a
method that is applicable to single treatments and descriptive studies, thus allowing
statistical analysis of pseudo- or un-replicated ecosystems, a problem faced in many
ecosystem studies (6). However, PCA is based on the analysis of variance in normally
distributed variables, thus requiring that all variables have a normal distribution. In
addition, it is assumed that the variables are related in a linear fashion. These assumptions
may not be met by the data. One or both of these problems may be corrected by a
transformation of the raw data (i.e. a log transformation); however adherence to these
assumptions is difficult to test, especially in time-series experiments, and they are often
violated by biological data (16,26). There is evidence that PCA is robust even if the data
is not normally distributed; however, to make strong conclusions from a statistical
analysis, the assumptions of the analysis method should be met (28,29).

In contrast, Kohonen self-organizing maps do not make any assumptions about
the distribution of the input variables or their relationships to one another, making SOM
ordination an attractive alternative to traditional ordination methods for biological
systems. In addition, SOMs may be capable of analyzing data sets with more variability
than principal components or correspondence analysis because the dimensionality of the
output space is determined by the investigator. This allows the exploration of different
output dimensions (size and number of axes) to compromise between finding the best fit
and the best visual representation of the data in a smaller n-dimensional space. For
example, to create PCA ordination plots that were interpretable the data set for each

reactor in this study had to be divided into separate analyses for phase A and phase B

108

because more than three principal factors, and thus more than three dimensions, were
required to describe the variability in the data. This was not necessary for the analysis of
the same T-RFLP profiles using the Kohonen self-organizing map algorithm.
Unfortunately, it is harder to ascertain the importance of each axis and the meaning of
distance and directionality in an SOM output, although a correlation between distance
within a SOM output, geometric similarity of the input, and PCA have been demonstrated
(4,13). In addition, the only way to assess the quality of a SOM output is to perform
multiple independent analyses and compare the results. Because of these pitfalls, SOM
analysis is most amendable to low-level data exploration of particularly large and
variable data sets, while PCA and other traditional ordination techniques are more usefiJl
for smaller data sets and rigorous testing of hypotheses.

Despite the differences between the two methods, they yielded complementary
results that demonstrate important aspects of community structure dynamics. The tight
clustering of samples taken between 9 and 14 volume changes in the PCA ordination
plots demonstrated that approximately 9 volume changes were required for the bioreactor
bacteria communities to stabilize. Although the ordination of the phase A S2 and M3
community T-RFLP profiles resulted in strikingly similar patterns, it should be noted that
the factors extracted from the S2 community are different than those extracted from the
M3 community. Thus, the similarity of the ordination patterns for the two communities
does not indicate complete convergence of the M3 and 82 communities. However, both
PCA and SOM analysis identified fragments 78 and 203 as important during the early
start-up phase in both reactors (Table 4.2 and Figure 4.3). Reactor operation was very

stable (above 90 % COD removal) in both the 82 and M3 reactors despite the dynamic

109

nature of the community structure during this period. This finding is in agreement with a
previous study of methanogenic bioreactors, which found that a dynamic community
could sustain a functionally stable system (12).

The clustering of samples seen in the PCA ordination plots of phase B samples
clearly shows which HRTs affected the bacteria community structure. Both communities
were affected by the 2 day HRT, while only the 82 community was significantly affected
by the 4 day HRT (Figure 2B and D). The fragments indicated as important during the 4
and 2 day HRTs by both PCA and SOM analysis overlapped, and all of the fragments
with high loadings for the second factor were found in SOM group IV.

Thirteen of the 66 terminal restriction fragments analyzed (both bacterial and
archaeal) were associated with the same SOM group in both reactors. The PCA indicated
that five fragments that accounted for a significant amount of variability between samples
behaved similarly in both reactors. Both of these results indicate that the two bacterial
communities shared numerically abundant community members during each stage of
reactor operation. The SOM identified more fragments that behaved similarly in both
reactors because all fragments were represented in the output, regardless of their
contribution to the overall variability. Fragments 78, 81, 203, and A192 were found in
group I in both reactors. Three of these fragments had relatively large peak heights,
suggesting that they were not just minor components of the community. In group 11], four
fragments were common between the two reactors. Of these, fiagment 185 was the most
dominant member of the community as judged by its relative peak height. In group IV,
four fragments were in common between the two reactors, with 253 and 258 being the

most abundant.

110

The similarity of community change and response in reactors M3 and S2, as
illustrated by the appearance of the same terminal restriction fragments at the same time,
presumably indicating similar phylotypes and organisms, is intriguing. Although PCR
bias likely renders community sampling incomplete, it should be somewhat
representative and, when kept constant, allows the comparison of communities under
different conditions, as is done here. It appears that a type of successional change
occurred in both reactors as comparable phylotypes appeared transiently at the same time
prior to the establishment of similar stable communities. This observation indicates that
two relatively different communities converged through analogous stages of community
development and responded similarly to changes in environmental conditions. These
observations indicate that microbial community change is not a stochastic process, but
may occur in successional stages, even when the starting communities are very different.
In plant communities, the repeatability and similarity of successional change in many
different communities has long been noted. However, in microbial communities
extensive data showing that succession is repeatable and similar in different communities
is limited. This study indicates that theories such as the resource-ratio hypothesis may be
applicable to microbial community succession as well (30).

Not only are there similarities between the reactors in this study, but the bacterial
sequences retrieved from these reactors are also closely related to those found in other
anaerobic bioreactors (Figure 4.4). Sequences corresponding to two of the most abundant
fragments, 185 and 253, have been found repeatedly in glucose-fed anaerobic reactors
(10, 11, 33). Fragments 78 and 203 corresponded to phylotypes related to sequences

previously found in methanogenic environments that are divergent from cultured

111

organisms (Figure 4.4). The number of sequences repeatedly observed in simple
environments that are not related to cultivated organisms emphasizes our lack of cultured
representatives of microbial diversity even in common meSOphilic environments.

Another interesting convergence occurred in the archaea communities of reactors
S2 and M3. The archaea community in the S2 reactor changed steadily during phase A,
while the M3 archaea community remained relatively unchanged. This was probably a
result of the different inoculum source for the two reactors. The change that occurred in
the S2 archaea community was easily tracked by T-RFLP and led to convergence of the
two archaea communities (Figure 4.5). There was a progression of acetoclastic
methanogen species in the 82 community after start-up that is demonstrated by the
decrease of fragment A215 (Methanosaeta concilii) and the concomitant increase of
fragments A217 and A220 (Methanosarcina species). The dominance of Methanosarcina
over Methanosaeta is unexpected because Methanosaeta species have a higher affinity
for acetate than Methanosarcina species. The observations made possible by T-RFLP can
now be used to form new hypotheses about interactions between these two genera.

In conclusion, describing these types of community shifts over time in many
complex microbial communities is difficult with labor intensive techniques, such as 163
rDNA cloning and fluorescent in situ hybridization, thus the first step of rapid community
fingerprinting (by T-RFLP or similar techniques such as denaturing gradient gel
electrophoresis) and robust data analysis is invaluable to microbial ecology. In addition,
similarities between complex communities are not always apparent by manual inspection
of T-RFLP profiles or similar data, especially when the communities are diverse and/or

dynamic. PCA provided insight into overall community dynamics and continues to be of

112

great importance in analysis of ecological data, but the analysis was complicated by the
large number of terminal restriction fiagments and the amount of change observed
between samples. This problem would be compounded in natural microbial communities
that are subject to more variability than lab-scale, single-substrate, stirred bioreactors.
SOM analysis efficiently reduced the dimensionality of the data, providing a meaningful

ordination of all the input variables, and should be of great help in the analysis of natural

microbial communities in the firture.

Acknowledgements. This research was supported by NSF Grant DEB 9120006 to the
Center for Microbial Ecology. SLD was partially supported by the Michigan State
University Distinguished Fellowship Program. We are also grateful to TL. Marsh, A.P.

Topchy and B. Punch for their helpful discussions concerning this research.

References

1. Allison MJ, Mayberry WK McSweeney CS, Stahl DA (1992) Synergistes jonesii,
gen. nov., sp. nov.: A rumen bacterium that degrades toxic pyridinediols. Syst
Appl Microbiol 15:522-529

2. Amann RI, Ludwing W, Schleifer K (1995) Phylogenetic identification and in situ
detection of individual microbial cells without cultivation. Microbiol Rev 59:143-
169

3. Baena S, Fardeau ML, Labat M, Ollivier B, Garcia JL, Patel BK (2000)
Aminobacterium mobile sp. nov., a new anaerobic amino-acid-degrading
bacterium. Int J Syst Evol Microbial 50:259-264

4. Blayo F, Demartines P (1991) Data analysis: How to compare Kohonen neural
networks to other techniques? In: Prieto A (ed) Artificial Neural Networks.

Springer-Verlag, Germany, pp 469-475

5. Brower JE, Zar JH, van Ende CN (1990) Field and Laboratory Methods for General
Ecology. Wm.C.Brown, Dubuque, IA

113

6. Carpenter SR (1989) Replication and treatment strength in whole-lake experiments.
Ecology 702453-463

7. Chan T-S, Park YS, Moon KH, Cha EY (1996) Patternizing communities by using an
artificial neural network. Ecological Modeling 90:69-78

8. DeLong EF (1992) Archaea in coastal marine environments. Proc Natl Acad Sci USA
89:5685-5689

9. Dojka MA, Hugenholtz P, Haack SK, Pace NR (1998) Microbial diversity in a
hydrocarbon- and chlorinated-solvent-contaminated aquifer undergoing intrinsic
bioremediation. Appl Environ Microbiol 64:3869-3 877

10. Dollhopf SL, Hashsham SA, Dazzo FB, Hickey RF, Criddle CS, Tiedje JM (2001)
The impact of fermentative organisms on carbon flow in methanogenic systems
under constant low substrate conditions. Appl Microbial Biotech In Press

11. Fernandez A, Hashsham S, Dollhopf SL, Raskin L, Glagoleva O, Dazzo FB, Hickey
RF, Criddle C, Tiedje JM (2000) Flexible community structure correlates with

stable community function in methanogenic bioreactor communities perturbed by
glucose. Appl Environ Microbiol 66:4058-4067

12. Fernandez A, Huang S, Seston S, Xing J, Hickey RF, Criddle C, Tiedje JM (1999)
How stable is stable? Function vs community stability. Appl Environ Microbiol
65:3697-3704

13. Foody GM (1999) Applications of the self-organising feature map neural network in
community data analysis. Ecological Modelling 120297-107

14. Franzman PD, Dobson SJ (1992) Cell wall-less, free-living spirochetes in Antarctica.
FEMS Microb Let 972289-292

15. Godon J, Zumstein E, Dabert P, Habouzit F, Moletta R (1997) Molecular microbial
diversity of an anaerobic digestor as determined by small-subunit rDNA sequence
analysis. Appl Environ Microbiol 63:2802-2813

l6. Greig-Smith (1980) The development of numerical classification and ordination.
Vegetatio 42: 1-9

17. Hashsham S, Fernandez A, Dollhopf SL, Dazzo FB, Hickey RF, Tiedje JM, Criddle
C (2000) Parallel processing of substrate correlates with greater functional
stability in methanogenic bioreactor communities perturbed by glucose. Appl
Environ Microbiol 66:4050-4057

18. Hirst ID, Stenberg MJE (1992) Prediction of structural and functional features of
protein and nucleic acid sequences by artificial neural networks. Biochemistry
32:7211-7218

114

19. Kohonen T (1998) The self-organizing map. Neurocomputing 2121-6

20. Lek S, Guegan JF (1999) Artificial neural networks as a tool in ecological modelling,
an introduction. Ecological Modelling 120:65-73

21. Lilburn T, Schmidt TM, Breznak JA (1999) Phylogenetic diversity of termite gut
spirochaetes. Environ Microbiol 1:331-345

22. Loeffler FE, Ritalahti KM, Tiedje JM (1997) Dechlorination of chloroethenes is
inhibited by 2-bromoethansulfonate in the absence of methanogens. Appl Environ
Microbiol 63 24982-4985

23. Maidak BL, Cole JR Lilburn TG, Parker CTJ, Saxman PK Stredwick JM, Garrity
GM, Li B, Olsen GJ, Pramanik S, Schmidt TM, Tiedje JM (2000) The RDP
(Ribosomal Database Project) continues. Nucleic Acids Res 28:173-174

24. Marsh TL, Saxman P, Cole J, Tiedje JM (2000) Terminal restriction fragment length
polymorphism analysis program, a web-based research tool for microbial
community analysis. Appl Environ Microbiol 66:3616-3620

25. Noble PA, Almeida JS, Lovell CR (2000) Application of neural computing methods
for interpreting phospholipid fatty acid profiles of natural microbial communities.
Appl Environ Microbiol 662694-699

26. Potvin C, Raff DA (1993) Distribution-free and robust statistical methods: Viable
alternatives to parametric statistics. Ecology 74:1617—1628

27. Strunk O, Ludwig W (1997) ARB: Software for Phylogenetic Analysis. Technical
University of Munich, Munich, Germany

28. Tabachnick BG, F idell LS (1983) Using Multivariate Statistics. Harper & Row, New
York

29. Thorndike RM (1978) Correlational Procedures for Research. Gardner Press, Inc,
New York

30. Tilman D (1985) The resouce-ratio hypothesis of plant succession. American
Naturalist 125:827-852

31. Toronen P, Kolehmainen M, Wang G, Castren E (1999) Analysis of gene expression
data using self-organizing maps. FEBS Letters 451:142-146

32. Towner H (1998) Ecostat for Windows. Trinity Software, Plymouth, NH

33. Wu C (1993) Classification neural networks for rapid sequence annotation and
automated database organization. Comput Chem 17:219-227

115

34. Xing J, Criddle C, Hickey R (1997) Long-term adaptive shifts in anaerobic
community structure in reponse to a sustained cyclic substrate perturbation.
Microb Ecol 33:50-58

116

CHAPTER 5
INTERSPECIES INTERACTIONS AFFECTING A SPIROCHAETA SPECIES 1N

METHANOGENIC CHEMOSTATs

Introduction

Interspecies interactions are an integral part of microbial community structure and
function in all ecosystems. It is very rare to observe a pure culture of a microorganism in
a natural setting. Even symbiotic and parasitic organisms are interacting with their hosts
and are therefore interacting with another species. Although the study of microorganisms
in pure culture is a convenient and powerful technique that has vastly increased our
knowledge of genetics, biochemistry, and microbial diversity, it is very artificial. Batch
culture methods are also unnatural systems for studying many microorganisms.
Continuous culture systems, such as bioreactors and chemostats, provide a better
simulation of the nutrient limitation that prevails in most environments (9,14,26). The
chemostat provides a means of enriching for organisms that have a low saturation
constant (K) and maximum specific growth rate (pm), which are difficult to select for in
typical batch enrichment cultures, but may be more prevalent in the environment. In
addition, the growth rate of organisms in a chemostat are easily manipulated, making it
possible to study the growth and interactions of species at doubling times closer to those
found in natural environments.

Spiral-shaped microorganisms have intrigued microbiologists since they were first
described over one hundred years ago because of their distinctive shape and fascinating

motility (1). Many spiral-shaped organisms, including the members of the order

117

Spirochaetales (referred to as spirochetes), are slow-growing microorganisms
hypothesized to be adapted to growing at low substrate concentrations. The spirochetes
are a phylogenetic group that is united by common morphological traits, but is very
diverse physiologically, and includes aerobic, anaerobic (facultative and obligate),
fermentative, and homoacetogenic microorganisms (1,15). Spirochetes can be free-living,
host-associated, or pathogenic and are known to cause a variety of human and animal
diseases such as syphilis, Lyme disease, and digital dermatitis. Because of the slow-
growth rates exhibited by many spirochetes they should be amendable to enrichment and
study in continuous culture systems, yet only a few studies have examined the growth of
spirochetes in continuous culture (18,19,25).

In this study we investigated the competitive ability and interspecies interactions
of an obligately anaerobic fermentative spirochete, Spirochaeta sp. str. R8, in glucose-fed
continuously stirred methanogenic bioreactors operated with long generation times
(weeks to days) and low substrate concentrations. We present evidence that fermentative
spirochetes are excellent competitors for glucose, but that their population can be limited
by the supply of required vitamins, such as pantothenate, making their abundance

dependent upon the excretion of these vitamins by other community members.

Materials & Methods

Bioreactor operation. Two 16-liter glucose-fed continuously stirred tank
bioreactors were operated anaerobically at 34°C with a loading rate of 0.5 g glucose/liter-
day and a dilution rate of 0.06 days'1 for 20 volume changes before this study

commenced (5). The dilution rate was achieved by continuous delivery of a buffered

118

glucose solution (8.0 g/L glucose + 7 .0 g/L NaHCO3) and a separate concentrated
nutrient solution using separate parastaltic and syringe pumps, respectively (27).
Bioreactors 82 (Sediment 2) and M3 (Municipal 3) were inoculated with sediment from
the Red Cedar River and fluid from a 250 mL variable volume reactor initially seeded
with municipal sewage sludge, respectively. The dilution rate of the bioreactors was
manipulated by increasing the flow rate through the bioreactors without modifying the
glucose or nutrient concentrations of the influent. Each dilution rate was applied for at
least eight volume changes to allow the community to adapt and stabilize. A third
bioreactor, bioreactor R1 (Rumen 1), served as a reference and was operated at identical
conditions at a dilution rate of 0.06 days'1 throughout the entire experiment. This reactor
was inoculated with rumen fluid and operated for over 650 days (approximately 40
volume changes) prior to this study. Chemical oxygen demand (COD), volative fatty
acids (VFAs), and pH were monitored biweekly as previously described (10).

T-RFLP analysis. Community DNA was extracted from 1 mL samples using a
beadbeaterTM and phenol-chloroform extraction method (4,16). Amplification of bacterial
16S rRNA genes from community DNA with the bacterial specific forward primer 8F-
HEX and the universal reverse primer 1392R was performed as previously described (4).
Amplification products were digested with HaeIII and run on a Perkin Elmer Applied
Biosystems 373A automated sequencer. The resulting chromatograrns were normalized to
10,000 fluorescence units for comparisons of fragment abundance in different samples.

FISH. Slides and samples were prepared as described previously (23). Briefly,
samples (0.5 mL) were fixed in 3 % paraforrnaldehyde and stored in 50 % ethanol at -

20°C. Fixed cell suspensions (1-5 pl) were spread onto 5 mm wells of black-coated 10

119

well slides, dehydrated in 50, 80, and 98 % alcohol and air dried. Hybridization and
washing steps were carried out in 50 ml screw top tubes in a water bath of the desired
temperature. The hybridization buffer consisted of 0.5 M NaCl, 0.02 % SDS, and 0.1 M
TrisHCl (pH 8.0). The wash buffer was identical except for an increase in SDS to 0.05%.
Hydridization stringency was varied by changing the hybridization and washing
temperatures. After washing, all slides were counterstained with 0.1 pg/ml DAPI for 10
min at 4°C. A Zeiss Axioskop model 50 microscope (Zeiss Inc., Thomwood, NY)
equipped with a mercury lamp and a SPOT digital camera (Diagnostic Instruments Inc.,
Sterling Heights, MI) was used to capture all epifluorescent images. Zeiss filter sets
488015, 488026, and 488002 were used for CY3, CY5, and DAPI imaging, respectively.
Two new oligonucleotide probes, SpiroR8-177 and ClostS9-182, were designed
to target 16S rRNA gene sequences closely related to Spirochaeta sp. str. R8 and
Clostridium ramosum str. S9 using the PROBE-DESIGN and PROBE-MATCH tools
included in the ARB software package (24) (Figure 5.1). Probes SpiroR8-177 (5’-
CACAGCCCGGTAACCACATT-3’) and ClostS9-182 (5’-TCACCATGCAGT
GTCCGTACC-3’) were synthesized and labeled with the CY3 or CYS fluorochrome,
respectively, by Molecular Probes (Portland, OR). Unlabeled helper probes that
hybridized to either side of SpiroR8-177 were designed to increase accessibility of the
rRNA in this region (8). The 5’ (5’-GGTATT ACCCACTATTTCTAGUG-3’) and 3’ (5’-
CATAGCTCCTTTCTTTACCAGG—3’) helper probes were added to the hybridization
buffer at the same time as the labeled probe with a final concentration of 1 ng/ml. The

conditions for the new probes were Optimized using positive and negative controls. The

120

scam £5 E 856% woooa oozoofiacogfi on“ 53> mocoLmEmE N 8.5 99: no: ommnmamu o5 E 305309.
__m .88: 952390 oz: 9: .2 flooxm .coszEEE £6 E L2 now: $006 oozoofiscogo Co 36:63.0.
ocm Ecol oE>~co cozoEwQ 53> ooamcocom 95:59., cozoEmE .mc_E._9 ocom «.sz mm: fin 930E

 

an§E§ EutmuomnozmsoE
mEEroEE mooAEoEEmE

 

 

            

mEEmE musoEcmE illi_illJ
MEL. an mmN EDmQES 53.5.5ng
- 2.9
NS. mmamoa m.mw
EEmmBoo E33586 i

58:55 P mESEEQm E35585
EEQEES 32.68203
ESE: 35538.63
E3883 EEDEmoB i

moEooAE mEmmEooAE

€23me mEmmfiEfim w“
EEQE mEmmEoobe

EzotbzooEms E28385

mEEELosam 823820 Ml
E39883 mEmquot Hi
EEmbo mEocoqmt

>-Om
__ - m
:2 do Somcoocfim BOICIIUI

    

Em.
9.0 “E9 3 m? E Em .8 Somcogqm
wtémogam >_x-Om

_ m:
S638 SmmEoEEm.
mfigmocmm SomEoEEm

 

 

 

motmEsomE 53:83
EEO: Segue: m
SEQQEE mcomcootqm
mEmSm SomzooEqm

 

121

optimum hybridization and washing temperatures were determined to be 37°C for
SpiroR8-177 and 43°C for ClostS9-182.

Isolation and culture techniques. Spirocheata sp. strain R8 (DSM 13955) was
isolated from R1 bioreactor fluid as previously described (4). Strain S9 was isolated by
serial dilution of S3 bioreactor fluid in anaerobically prepared tryptic soy broth plus 0.2
% glucose and 50 ug/mL rifampin with a headspace of N22C02 (80:20). The highest
positive dilution tube was then serially diluted in the same medium plus 0.7 % agar.
Routine cultivation was performed at 37°C in mineral medium (22) plus 5 % (v/v) rumen
fluid, 5 mM glucose, IOmM MOPS buffer, 0.4 g/L sodium sulfide nonahydrate, 0.6 g/L
cysteine, 0.13 g/L bacto peptone and 0.13 g/L yeast extract adjusted to a pH of 7.2 with
headspace of N2:C02 (80:20). A IOOX cofactor solution (pH 7.2) containing (per 100
ml): pyridoxal HCl, 10 mg; pyridoxal phosphate, 10 mg; calcium folinic acid, 2 mg; B-
NAD, 2 mg; coenzyme A, 2 mg; FAD, 2 mg; nicotinamide, 1 mg; folic acid, 0.1 mg;
riboflavin, 0.02 mg; thiamine pyrophosphate, 100 mg, was added to enhance growth (15).
For determination of growth requirements, yeast extract, bacto peptone, and rumen fluid
were omitted from the media. Cell yields were determined afier logarithmic growth had
ceased by DAPI staining and filtration of paraformaldehyde-fixed cell suspensions.

All methanogen cultures were obtained from the Oregon Collection of
Methanogens in Portland, Oregon. Methanobacterium bryantii M.o.H. (OCM 110) and
Methanosarcina mazeii S-6 (OCM 26) were grown in the medium described above,
minus glucose and over-pressurized with H2. Cocultures of strain R8 and
Methanobacterium bryantii or Methanosarcina mazeii were transferred in the glucose

medium described above without added yeast extract or cofactors.

122

16S rRNA gene sequencing. PCR amplification of the 16S rRNA gene from
strain S9 was performed on whole cells after one freeze-thaw cycle. The bacterial primer
8F and the universal primer 1392R were used for amplification as previously described
(4). The resulting amplicons were cloned into pCR2.1 per the manufacturers instructions
(Invitrogen, San Diego, CA), screened for homogeneity by restriction analysis, and
approximately 1400 bp were sequenced via the automated dye dideoxy terminator
method by the MSU sequencing facility. Sequence similarity was determined using the
GenBank BLAST function (wwwncbigov).

Continuous culture. Competition experiments were carried out at 37°C in an
anaerobic 500 mL chemostat. A mineral medium (22) plus 5 % (v/v) rumen fluid, 0.13
g/L yeast extract, 0.13 g/L bacto-peptone, 5 mM glucose, 10 mM MOPS buffer, 0.4 g/L
sodium sulfide nonahydrate, and 0.6 g/L cysteine adjusted to a pH of 7.2 with a
headspace of N2:C02 (80:20) was used in all competition experiments. A vitamin
solution was also added to give the following final concentrations (per liter): biotin, 2.0
pg; folic acid, 2.0 ug; pyrodoxin, 10.0 pg; thiamine HCl, 5.0 ug; riboflavin, 5.0 pg;
nicotinic acid, 5.0 ug; cobalamin, 0.1 ug; paraminobenzoate, 5.0 pg; lipoic acid, 5.0 ug;
b-nicotinamide adenine dinucleotide, 2.0 ug; folinic acid, 2.0 ug; flavin adenine
dinucleotide, 2.0 ug; pyridoxal S-phosphate,10.0 ug; thiamine pyrophosphate, 30.0 mg.
Calcium D-pantothenate (500 ug/ml) was added separately in specific experiments. An
Ismatec [PC multichannel peristaltic pump controlled the flow rate of the sterile medium
into the culture vessel. The pH was monitored daily and maintained at 7.1:t0.1.

Organisms were pre-grown in batch culture and anoxicly inoculated into the

chemostat. Each competition experiment was initiated by inoculating the chemostat

123

vessel with strain S9 and allowing the optical density to reach equilibrium before
inoculating with strain R8. At equilibrium, the medium supported approximately 109
chL (based on serial dilution in agar shakes). Samples (1 mL) were withdrawn every
12 or 24 hours and immediately diluted to 107 cells/ml and mounted on an agarose-coated
slide. The number of rod- and spiral—shaped cells in 25 fields of view were manually

counted, resulting in the enumeration of over 500 cells for each sample.

Results

T-RFLP. The effect of different dilution rates on bioreactor communities 82 and
M3 was assessed with 16S rDNA based-T-RFLP analysis. Dilution rates of 0.25 and 0.50
days'1 had the most noticeable effect on the community structure in both reactors, while
little to no change occurred in the R1 reference community (5). In particular, two specific
fragments of 185 bp and 253 bp appeared to be the most affected (Figure 5.2a). These
fragments matched the predicted terminal restriction fragment (T-RF) lengths for specific
groups of Spirochaeta- (185 bp) and Clostridium- (253 bp) related 16S rDNA sequences
previously cloned from similar bioreactors (4,6,7,10) (Figure 5.1). The 185 bp fragment
also matched the predicted T-RF for Spirochaeta sp. str. R8, which was previously
isolated from bioreactor R1. This data suggested that in both the 82 and M3 communities
Spirochaeta-related organisms decreased due to the increase in dilution rate, while

Clostridium-related organisms increased (Figure 5.2b).

124

   

 

 

 

 

 

 

 

 

 

 

so 1 1 zoo 2
40011. D = 0.13 days‘1 €253 bp
Ml l i
I
Q: — l l
40°94. D = 0.17 days'1 ' l
l I
zoooTL :
a MA : - - , r
i 7
4000.) D = 0.50 days'1 1
OJ A AL - : A .

 

 

 

 

 

Figure 5.2a. T-RFLP electropherograms of M3 communities at different dilution rates.
The electropherograms were generated by restriction of amplified bacterial 16S rRNA
genes with the restriction enzyme HaeIII. The 185 bp and 253 bp fragments were
indicative of Spirochaeta and Clostridium species, respectively.

125

.38 a3 2: mac? v80: 3 datum 2:: 3.: mam—2..
com—mam 2E 5:26 05 3 "6:83:00 memo—Em 2E. .AO.OV 92 EB AD.IV mm moficsfifig 888.55 Bot acct 085:0
5:5 macaw 4‘ch mg 3583 BEE—5 mo c2853.. 3 835.com 3058.? :38 55655 Eafismoc 98 22.8%
@828 36on o‘egooxfiw com 033:: 358%..“ 55:63 23:58 mo Emma; anon nod—EEO: E omfifio .56 €sz

 

mowumnu 0838/
cc mm ow mv ow mm on mm 2..
.l lF . _ _ b L - lLI O
n .—
s l fl 8m
f. 89 W
82 W
m.“
\ - 88 mt.
\ 1
.)u
,_ . (s - 88
I

 

Ania—av A— lv . .. cm: . mud . . bmd mud 08m

126

FISH. To confirm results obtained with the PCR-based T-RFLP method,
oligonucleotide probes were designed for specific groups of Spirochaeta- and
Clostridium-related sequences (Figure 5.1). Probe ClostS9-182 was specific for C.
ramosum when hybridized at 43°C and gave strong signals with little background after
hybridization with the bioreactor samples. Probe SpiroR8-l77 was specific when
hybridized at 37°C, but initially gave very weak signals that were difficult to detect. The
use of helper probes boosted the signal noticeably (8).

Between 45 and 55 % of DAPI stained cells hybridized with SpiroR8-l77 in all
samples of the reference reactor R1 (data not shown). Probe SpiroR8-177 hybridized with
22 to 44 % of DAPI stained cells in the 82 and M3 bioreactor samples taken when the
dilution rate was below 0.25 days'1 (Figures 5.3 and 5.4). After ten volume changes at a
dilution rate of 0.25 days'l, the number of cells detected with SpiroR8-177 decreased
slightly in both bioreactor communities, and dropped significantly after four and seven
volume changes at a dilution rate of 0.50 days].

Cells detected with probe ClostS9-182 in samples taken when the dilution rate
was less than 0.25 days'1 gave relatively weak signals (Figure 5.3c and d). In some cells
the DAPI stain showed condensation of the genomic DNA, indicating that at dilution
rates less than 0.25 days'1 some members of the population were in a quiescent or
dormant state, although they were not completely washed out of the bioreactor. Less than
1 % of DAPI stained cells hybridized with ClostS9-182 in all samples of reference

reactor R1. Afier ten volume changes at a dilution rate of 0.25 days'1 in reactors M2 and
S3, the number of cells detected with ClostS9-182 increased from approximately 5 % to

between 15 and 25 % in both bioreactor communities (Figure 5 .4). After four volume

127

 

Figure 5.3. Micrographs of bioreactor fluid afler hybridization with fluorescently
labeled oligonucleotide probes. DAPI (left) and FISH (right) micrographs are shown
for identical fields. Bioreactor community M3 after 10 volume changes at D = 0.17
days" hybridized with probe SpiroR8—l77 (A and B). Bioreactor community S2 afier
8 volume changes at D = 0.13 days'1 (C and D) and afler 7 volume changes at D =
0.50 days‘1 (E and F) hybridized with probe ClostS9-182. Images are presented in
color.

Ashe—av G r 1...:

ems

 

M4~-°

A» W.

.008 no“ 05 mac? 008: 00 .0303
«05 macaw 00:30 0000 023:0 05 9 3:800:00 msfimnm 2E. 00383888 08000003 3 m2 050 A Iv mm 05 E A238,?
:83 $3986 05 238% 883 5.0083 88:. £3 85an 35 2.8 85% :20 0c 2850 .4.“ 050$

m0mSEU 0E:_o>

cm ; ow :. en fin on
F L _ [V 0 0 Tr O

 

 

T—

on

tuoorod

.I‘

 

£03 a l. on... .

 

129

changes at D of 0.50 days“, little change in the percentage of ClostS9-l82 positive cells
was detected, but after seven volume changes they increased to 35 % of DAPI stained
cells in both the 82 and M3 bioreactor communities.

Strain R8 and S9 characteristics. Spirochaeta sp. str. R8 (DSM 13955) was
previously isolated from the control bioreactor R1. The 16S rRNA gene sequence was
only 0.5 % divergent from S. caldaria. Stain R8 ferments glucose to acetate, ethanol,
lactate and hydrogen, a common fermentative pattern among free-living anaerobic
spirochetes (4). No growth occurred in minimal media without pantothenate, and growth
yield increased as the supply of pantothenate was increased. If coenzyme A (CoA) was
added instead of pantothenate, greater growth yields resulted (Table 5.1). Yeast extract
also stimulated grth more than predicted from the assayed amount of pantothenate
present per gram of extract. Thiamine pyrophosphate and a mixture of volatile fatty acids
were also stimulatory for growth in mineral medium lacking rumen fluid and decreased
the number of spheroid bodies present in 4 day old cultures. Rumen fluid (5 % v/v)
provided more robust grth than the volatile fatty acids and thiamine pyrophosphate
combined (data not shown). In minimal glucose medium plus 5 % rumen fluid and 0.13
g/L of yeast extract strain R8 grew at a maximum specific grth rate of 0.14 hr'1
(doubling time of 5 hours).

Strain 89 was isolated from a 109 dilution of fluid from bioreactor 82. The 168
rRNA gene sequence of this strain was 100 % identical to that of Clostridium ramosum.
Strain S9 required B vitamins or yeast extract for grth and had a maximum specific

growth rate of 1.4 hr'l.

130

Table 5.1. Growth yields in glucose minimal medium.

 

 

Additions Yield (cells/ml)“
None No Grth
100 ug/ml Ca pantothenate 0.2 X 107
5% Rumen Fluid, 0.13 g/l YE 0.4 x 107
200 11ng Ca pantothenate, 7.4 X 107
5% Rumen Fluid, 0.13 g/l YB
50 ug/ml Coenzyme A, 9.0 X107

5% Rumen Fluid, 0.13 g/l YE

 

“Original inoculum size of 3.0 X 105 cells/ml

131

Competition in continuous culture. Strain R8 increased rapidly after inoculation
into a stable continuous culture of S9 cells, regardless of whether or not excess
pantothenate was supplied in the medium (Figure 5.5). The washout of S9 cells (and thus
the increase in the relative frequency of R8 cells) was more rapid when excess
pantothenate was supplied. Under these conditions the washout rate of S9 was only
slightly slower than that expected of a non-growing population, indicating that as a
substantial population of R8 established, glucose concentrations were so low that S9
could no longer maintain growth (Figure 5.5a). R8 outcompeted S9 at all dilution rates
employed when excess pantothenate was supplied, even those dilution rates at which R8
was not competitive within the bioreactor community, eventually leading to a population
of 96 % R8 cells and 4 % S9 cells after only four volume changes. The overall increase in
dilution rate over the course of the experiment from 0.17 to 1.0 days'1 corresponds to a
reduction in doubling time of nearly four days (from 4.2 days at D = 0.17 days'1 to 0.7
days at D = 1.0 days'l).

With only yeast extract and rumen fluid supplying a small amount of
pantothenate, strain R8 increased to approximately 40 % of the total cells in the
chemostat, after which a stable coexistence of approximately 40 % R8 and 60 % S9 cells
was established (Figure 5.5b). This ratio of R8 to S9 cells was maintained even when the

dilution rate was increased from 0.17 to 0.25 days‘1 and from 0.25 to 0.50 days".

132

.comoflsaoq wEBEméo: a mo 38 30:33 05 330:2: on: 353 2:. 8:03:58 32:52 nova: 855mb.“
mEofitoaxo 5:309:00 8883 95 Bow .565 mm Sun .3238 08:058ch mo AER: 08 53> 8330 v88: 303588
E season—boo mat—6 AO.OV mm .5 55855 533385 was A 0 .$ mm .5 dm uhogoohmm mo 3% cots—anon .mm.m unamE

 

 

 

momcfio ofiao>
v Wm m “N N m; _ We 0
p r _ _ _ [r .. b P r ~.O
3
U.
_
%
o W
.. 0
. .m
w r 2
o
m ..
o M o 5.”... ..
o o o .0...
...... ...................................... .. situ... 1.3 ”a 35......in .o. o. 00 M
is; a Iv 3 cm... 3... :5 .

 

 

.mcoEcho 30552 .895 con—Stun macofitoaxo 553388 8838 95 89m 8505 mm «SQ .8330 @838 303583
E 8:309:00 mats AO.OV am am EamoEE 533335 98 A. .$ xx .5 .am Seguohnw. mo 3% coca—anon find 059m

$920 8575

 

Wm m Wm N mg

_ _ _ — .l _

134

adhoqdlow %

 

 

 

0A:

ER

FEUI

T83]

WEI

Methanogenic coculture. Coculture of Spirochaeta sp. str. R8 with
Methanobacterium bryantii lead to significant increases in cell yield, without increasing
the amount of acetate (i.e. ATP) produced per mole of glucose utilized (Table 5.2).
Addition of excess hydrogen to the coculture increased the amount of methane produced
and the yield of R8 cells. Addition of hydrogen to a pure culture of strain R8 did not
increase cell yield. Stable cocultures of strain R8 and Methanosarcina mazei or
Methanobacterium bryantii without added pantothenate were transferred over 10 times
and repeatedly gave similar cell yields as pure cultures of R8 with added pantothenate
(data not shown). The addition of M bryantii to a competition experiment between
strains S9 and R8 in medium without excess pantothenate allowed strain R8 to attain a
much higher percent of the total cells than when the two strain were competed alone in
the same medium (Figure 5.6). A stable coexistence of 92 % strain R8 and 8 % strain S9
was maintained for two volume changes after the initial increase in the R8 population in
response to the addition of M. bryantii. One percent of the total cells in this triculture
were F420 autofluorescent, indicating that M. bryantii composed approximately 1 % of the

total population (~108 cells/mL).

135

 

uofificfioc 87F

35:30 03233 mo omSQ/m 2: 03 833/ .8823 2088 ~ Home

.8523 03253 mo Star/on €3.86 8

35:3 88:93 mo omega 2: 8m 33.3w ._E\m=oo “9 X on .«0 our. 8289: :2:er

.2550

-923 AR 2 .o use .8938 68% AR 2 .o .Ezc :25: o\.. m min «:58 BEES 882m E “customs?

Acumozqsc 05

 

DZ 02 Q2 52 x no 5 8:2
83233 95
92 oz bz 52 x v: ”m as _ + snug §§8€§£§
awed AS .03 2 u .93 f: x 33
2 ._ 98 who 52 x om 5:ch 55883333:
60.93 :33 333 sf: x 8.0.3
84 Mao ”to Le x to 252
9208:: @2995 Ame—9E5
nuoosuoa unoosuoa Locoscoa BEEN—8 35 32:22.
7255mm owwuomim 3800< E .>

 

95:on SESBEQSSSMVEN 53» 83308 :33 E mm 58% 3238th mo m2?» £305 .N.m 033.

136

 

do:£:qoq wEBEméo: a mo 88 30:33 8: 3388: 8.: 3:3:
8E. 888883 8: 8 £3 Stub: 538833583: «2 X m: mo 5:63 8: 883%.: Bots 85 .8330 :88: 302588
E 8::388 mats: AOV om 8m £3855 £53885 88 ADV mm 8% dm Smguohmw :0 08m :83?qu .c.m 82mm“:

mowafiogo>
0 Wm m. We v Wm m Wm N n: _ mac o

. p p i, _ . L if _ L . P ~.o

0,! r”

adfitoquow %

 

 

137

Disc

limi

P0P

bio

fer

 

Discussion

The observation of a high proportion of fermentative spirochetes in a glucose-
limited bioreactor community suggests that the spirochetes were outcompeting other
populations for glucose. The abundance of spirochetes in numerous glucose-fed
bioreactors in our studies and others demonstrates that this is not an isolated observation
(3,5-7). In the mineralization of one mole of glucose to methane and carbon dioxide,
enough free energy is released to form approximately 5.5 moles of ATP (21). Glucose-
fermenting organisms perform the most energetically favorable reactions in this
conversion and can capture from 37 to 74 % of the energy released, depending upon the
products formed and the fermentation pathways employed. Most Spirochaeta species
ferment glucose to acetate, ethanol, and lactate via the Embden-Meyerhof pathway.
Spirochaeta species also contain the low potential electron carrier ferredoxin and may be
able to vary the amount of acetate produced depending upon hydrogen concentration in
order to produce more ATP, potentially producing up to 4 moles of ATP per glucose
(1,21).

Strain R8 produced approximately 0.5 moles acetate per mole of glucose utilized
(Table 5.1), which provides an extra 0.5 moles of ATP per glucose in addition to the two
ATP formed during glycolysis. Fermentation patterns of other acetate and ethanol
producing spirochetes demonstrate variable acetate to glucose ratios, resulting in a range
of 2.3 to 3.0 moles of ATP produced per mole glucose (1). In a bioreactor dominated by
spirochetes approximately 1.3 moles of acetate were produced for every 1.0 mole of
ethanol produced (4). Based on this ratio, up to three moles of ATP could have been

produced per mole of glucose utilized in situ. These data indicate that the spirochetes in

138

 

the bioreactor community were probably producing between 2.5 and 3 moles of ATP
glucose, which would result in 45 to 54 % of the total energy available from the
conversion of glucose to carbon dioxide and methane.

Based on these calculations, Spirochaeta sp. str. R8 should have constituted
approximately 50 % of the community if it was the dominant glucose-utilizing
population. The highest frequency of strain R8 cells detected by FISH in a bioreactor
community was 45 % of the total DAPI stained cells. In bioreactor S2, strain R8 only
accounted for 25 °/o of the community, indicating that perhaps other populations were
utilizing glucose in addition to strain R8. An additional consideration is that when
biomass is calculated from cell volume, the contribution of spirochetes can decrease
significantly because of their slender morphology (6). Previous studies found that the
specific products produced by strain R8 (acetate and ethanol) were rapidly utilized by
bioreactor fluid containing a high proportion of spirochetes (4). The data presented here
suggest that strain R8 was a dominant, although likely not the sole, population utilizing
glucose in these bioreactor communities.

The decrease in the abundance of spirochete R8 in bioreactor communities 82 and
M3 when the dilution rate was increased to 0.25 and 0.50 days'1 indicated that conditions
were more favorable for strain R8 at low dilution rates (Figure 5.4). The concomitant
increase in the Clostridium ramosum population suggested that the two strains were in
direct competition for glucose. At the higher dilution rates, glucose concentrations
increased to a point where Clostridium ramosum was the better competitor. However,
when the two strains competed directly in continuous culture, spirochete R8 remained

dominant over Clostridium S9 at dilution rates of 0.25, 0.50, and 1.0 days'1 (Figure 5.5a).

139

Only when pantothenate was limiting the growth of spirochete R8 was Clostridium S9
able to maintain a large population, and the two strains coexisted in relatively equal
numbers (Figure 5.5b). This suggests that the reason for the decrease in the abundance of
spirochetes at high dilution rates was not directly related to glucose competition, but was
the result of some other inhibition of spirochete growth, such as pantothenate limitation.

Spirochaeta sp. str. R8. required CoA for optimum growth. Although
pantothenate could replace this requirement, grth yields were significantly lower even
when higher concentrations of pantothenate were added. Similar results were obtained in
nutritional studies of S. litoralis (11), however most other organisms requiring
pantothenate do not respond to externally supplied CoA. The cell membrane may be
impermeable to CoA, therefore a specific transport mechanism or external degradation of
CoA may be required for uptake. An insufficient intracellular supply of CoA severely
limits protein and phospholipid production in E. coli, and the cessation of amino acid and
protein synthesis eventually causes growth stasis (13). In the heterofermentative lactic
acid bacterium Oenococcus oeni, CoA deficiency causes a shift in glucose fermentation
products away from products that are formed by enzymes utilizing CoA as a cofactor. In
Spirochaeta sp. str. R8, this type of shifi would decrease the amount of acetate and
ethanol produced because the phosphoroclastic reactions involved in their formation
require CoA. This would likely reduce the amount of ATP formed per mole of glucose,
resulting in much less efficient growth.

In these glucose-fed bioreactors, the only source of vitamins and cofactors would
be the excretion of vitamins by community members or the scavenging of vitamins from

dead microbial cells. To support a large population of cells (approximately 5.0 X 109

140

cells/ml (4)) a significant continuous supply of CoA precursors must have been available
to strain R8, therefore the scavenging of vitamins from dead cells would not have been
sufficient. Escherichia coli secretes up to 3 ug/ml of pantothenate during logarithmic
growth (12). The secretion of pantothenate in E. coli is a mechanism for regulating levels
of CoA inside the cell. However, other fermentative organisms screened for pantothenate
excretion excrete two to three orders of magnitude less pantothenate than E. coli (20). No
measurable amount of pantothenate was excreted by Clostridium ramosum str. S9 when
culture fluid was bioassayed with Lactobacillus plantarum (data not shown).

A potential source of CoA precursors for spirochete R8 in the bioreactor
community would be excretion by microorganisms in other trophic groups. Two
methanogens (Methanobacterium btjyantii M.o.H. and Methanosarcina mazeii) screened
in this study relieved the pantothenate requirement of R8 when grown in coculture. The
increase in strain R8 cell yield in coculture was not due to an increase in acetate or ATP
production from glucose (Table 5.2). Very low numbers of methanogens enhanced the
growth of R8, fitrther supporting the view that the methanogens were supplying a vitamin
or cofactor that enhanced the growth of spirochete R8. Methanogens likely supply
pantothenate or another similar molecule that allows strain R8 to synthesize CoA.
However, it is possible that the methanogens supplied amino acids and/or fatty acids
whose synthesis requires CoA, allowing strain R8 to grow well with small amounts of
CoA. Alternatively, the presence of methanogens may have altered culture conditions or
induced expression of genes such that strain R8 could synthesizes CoA de novo.

This study demonstrates that Spirochaeta sp. str. R8 is well adapted to compete

for low concentrations of glucose within a tightly linked anaerobic food web were

141

sufficient vitamins and cofactors are supplied. In general, spirochetes have a high surface
area to volume ratio, which may confer a competitive advantage for substrate and nutrient
uptake (17). They are also able to detect and chemotax towards very low sugar
concentrations (25). However, spirochetes are limited by their requirement for cofactors
and vitamins such as CoA, folate, and thiamine. In fact, spirochetes are usually found in
carbon-limited environments that are not considered oligotrophic, such as the termite
hind-gut, rumen fluid, hydrothermal and marine mats, and sediments (2). All of these
environments are either host-associated and/or involve a tightly linked anaerobic food
web. The newly-discovered homoacetogenic spirochetes also require exogenous vitamins
and cofactors and are limited to the termite gut (15).

Our results demonstrate that vitamin requirements can impact the competitive
fitness of an organism, and may affect the total abundance of a population in a
community. The localization of spirochetes to host-associated and non-oligotrophic
anaerobic environments supports this hypothesis. The supply of vitamins may come from
the host, organisms in other trophic levels, or the breakdown of detritus in these
environments. Spirochetes appear to specialize in metabolic interactions with a host or
other co~occurring organisms for the supply of vital cofactors that are central to the
spirochetes’ energy-producing metabolism. The advantage of these interactions for a
spirochete may be the lowering of maintenance energy, thus allowing for a lower growth
rate and better scavenging of growth substrates. The organisms furnishing the cofactors
required by spirochetes may in turn benefit from the sequestering of all ferrnentable
substrates entering the system, preventing any potential carbon sources from leaving the

community unprocessed.

142

Free-living fermentative spirochetes may be the common ancestors of all extant
spirochete species. Portions of the Embden-Meyerhof energy-yielding pathway exist in
all spirochete species that have been assayed for them, even in obligately aerobic strains
(1). Studying the primitive free-living spirochete species may therefore increase our
understanding of the fastidious disease-causing Treponema and Borellia species. In
addition, understanding the positive and negative interactions occurring within microbial
communities that prevent competitive exclusion may unlock answers to our questions

about the maintenance of astounding microbial diversity in seemingly simple,

unstructured environments.

References

l. Canale-Parola E (1977) Physiology and evolution of spirochetes. Bacteriol Rev
41:181-204

2. Canale-Parola E (1992) Free-living sacchrolytic spirochetes: The genus Spirochaeta.
In: Balows A, Truper HG, Dworkin M, Schleifer K (eds) The Prokaryotes, A

Handbook on the Biology of Bacteria: Ecophysiology, Isolation, Identification,
Applications, vol 2. Springer-Verlag, New York, NY, pp 3524-3536

3. Delbes C, Moletta R, Godon J (2001) Bacterial and archaeal 16S rDNA and 168
rRNA dynamics during an acetate crisis in an anaerobic digestor ecosystem.

FEMS Microbiol Ecol 35: l9-26

4. Dollhopf SL, Hashsham SA, Dazzo FB, Hickey RF, Criddle CS, Tiedje JM (2001)
The impact of fermentative organisms on carbon flow in methanogenic systems
under constant low substrate conditions. Appl Microbiol Biotech In Press

5. Dollhopf SL, Hashsham SA, Tiedje JM (2001) Interpreting 16S rDNA T—RFLP data:
Application of self-organizing maps and principle component analysis to describe
community dynamics and convergence. Microb Ecol In Press

6. Fernandez A, Hashsham S, Dollhopf SL, Raskin L, Glagoleva O, Dazzo FB, Hickey
RF, Criddle C, Tiedje JM (2000) Flexible community structure correlates with
stable community function in methanogenic bioreactor communities perturbed by

glucose. Appl Environ Microbiol 66:4058-4067

143

7. Fernandez A, Huang S, Seston S, Xing J, Hickey RF, Criddle C, Tiedje JM (1999)
How stable is stable? Function vs community stability. Appl Environ Microbiol
65:3697-3704

8. Fuchs BM, Glockner F0, Wulf J, Amann R (2000) Unlabeled helper oligonucleotides
increase the in situ accessibility to 16$ rRNA of fluorescently labeled
oligonucleotide probes. Appl Environ Microbiol 6622603 -3607

9. Harder W, Dijkhuizen L (1983) Physiological response to nutrient limitation. Ann
Rev Microbiol 37:1-23

10. Hashsham S, Fernandez A, Dollhopf SL, Dazzo FB, Hickey RF, Tiedje JM, Criddle
C (2000) Parallel processing of substrate correlates with greater functional
stability in methanogenic bioreactor communities perturbed by glucose. Appl
Environ Microbiol 66:4050-4057

11. Hespell RB, Canale-Parola E (1998) Glucose and pyruvate metabolism of
Spirochaeta litoralis, an anaerobic marine spirochete. J Bact 116:931-937

12. Jackowski 8, Rock CO (1981) Regulation of coenzyme A biosynthesis. J Bacteriol
148:926-932

13. Jackowski S, Rock CO (1986) Consequences of reduced intracellular coenzyme A
content in Escherichia coli. J Bacteriol 166:866-871

14. Jannasch HW (1967) Enrichments of aquatic bacteria in continuous culture. Arch
Microbiol 59:165-173

15. Leadbetter JR, Schmidt TM, Graber JR, Breznak JA (1999) Acetogenesis from H2
plus C02 by spirochetes from termite guts. Science 283 2686-689

16. Loeffler FE, Ritalahti KM, Tiedje JM (1997) Dechlorination of chloroethenes is
inhibited by 2-bromoethansulfonate in the absence of methanogens. Appl Environ
Microbiol 63 :4982-4985

17. Matin A, Veldkamp H (1978) Physiological basis of the selective advantage of a
Spirillum sp. in a carbon-limited environment. J Gen Microbiol 105: 187-197

18. Mikx FH (1997) Environmental effects on the growth and proteolysis of Treponema
denticola ATCC 33 520. Oral Microbiol Immunol 122249-253

19. Mikx FH, Jacobs F, Satumalay C (1992) Cell-bound peptidase activities of
Treponema denricola ATCC 33520 in continuous culture. J Gen Microbiol
138218374842

20. Sahm H, Eggeling L (1999) D-P antothenate synthesis in Corynebacterium

glutamicum and use of panBC and genes encoding L-valine synthesis for D-
pantothenate overproduction. Appl Environ Microbiol 65 : 1973-1979

144

21. Schink B (1997) Energetics of syntrophic cooperation in methanogenic degradation.
Microbiol Mol Biol Rev 61 :262-280

22. Shelton DR Tiedje JM (1984) Isolation and partial characterization of bacteria in an

anaerobic consortium that mineralizes 3-chlorobenzoic acid. Appl Environ
Microbiol 48:840-848

23. Snaidr J, Amann R Huber I, Ludwig W, Schleifer KH (1997) Phylogenetic analysis

and in situ identification of bacteria in activated sludge. Appl Environ Microbiol
63:2884-2896

24. Strunk O, Ludwig W (1997) ARB: Software for phylogenetic analysis. Technical
University of Munich, Munich, Germany

25. Terracciano J S, Canale Parola E (1984) Enhancement of chemotaxis in Spirochaeta
aurantia grown under conditions of nutrient limitation. J Bacteriol 159: 173-178

26. Veldkamp H (1977) Ecological studies with the chemostat. Adv Microb Ecol 1:59-94

27. Xing J, Criddle C, Hickey R (1997) Long-term adaptive shifts in anaerobic

community structure in reponse to a sustained cyclic substrate perturbation.
Microb Ecol 33 : 50-58

145

CHAPTER 6

SUMMARY

The activity of glucose-utilizing organisms is important in aquatic environments.
Total monosaccharide concentrations range from 2.0 uM to 0.02 M, with glucose
consistently accounting for 40 to 50 % of total monosaccharides (1). Dissolved sugars
constitute up to 80 °/o of the biomass of aquatic primary producers and 15 to 90 % of
photo-assimilated carbon is released by prokaryotic or eukaryotic algae during growth. In
mammalian gastrointestinal tracts, ruminant mammals, and wood-digesting insects, sugar
fermentation is also important to the nutrition of the animal. Recently the microorganisms
that carry out fatty acid oxidation and terminal electron accepting processes have
received the most attention in microbial ecology studies. The impact of fermentative
microorganisms on community structure and function has received relatively little
attention in the current revolution of molecular microbial ecology. This thesis has
emphasized the impact of competition and cooperation on the composition of the

fermentative guild in anaerobic microbial communities. In summary:

0 In methanogenic environments species richness, species evenness, and
community change is greater in bacterial communities than in archaeal
communities. In addition, microbial communities in a mass action environment
can be dynamic while ecosystem function is stable.

0 Even the highly selective environment of a mesothermic glucose-fed

methanogenic reactor can maintain a large number of fimctional redundant

146

species over long periods of time. Organisms with no cultivated relatives were
also repeatedly detected in this presumably well-studied environment.

The presence of different fermentative organisms can establish different routes of
carbon flow in methanogenic communities. These differences in community
structure and carbon flow influence the ability of a system to recover fiom
perturbation events.

Although microbial communities are dynamic, community change is not a
stochastic process, and may occur in successional stages, even when the starting
communities are very different. In microbial communities, extensive data
showing that succession is repeatable and similar in different communities is
limited.

Certain Spirochaeta species are successful in glucose-fed methanogenic
bioreactors operated at low dilution rates. Spirochete species in general are
amenable to enrichment and study in continuous culture systems.

Spirocheata sp. str. R8 has a low half saturation constant and maximum specific
grth rate, making it a superior competitor at low glucose concentrations. In
addition, strain R8 benefits from cooperative interactions with co-occurring
species that supply cofactors and vitamins, lowering the maintanence energy

required for grth and allowing growth at lower substrate concentrations.

147

o The competitive ability of a particular species may be affected by a cofactor or
vitamin requirement, and thus the abundance of one species may be linked to
other species in the community that fulfill their nutritional requirements. For
example, methanogens may be important in supplying vitamins to fermentative

organisms in anaerobic environments.

References
1. Mopper K, Dawson R, Liebezeit G, Ittekkot V (1980) The monosaccharide spectra

of natural waters. Marine Chemistry 10:55-66

148

1111 ‘111‘111‘111‘ 1‘ 111 1111 111

2930

     

 

‘ .