|
||||||||||||||||
|
||||||||||||||||
|
||||||||||||||||
|
| ||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 3
2
Workshop Proceedings
MICROBIAL SYSTEMS: CASE STUDIES
Because the workshop was formatted as open discussions, the planning
committee of the workshop selected two contrasting case studies to main-
tain focus: on the Geobacter community, which has well-recognized bio-
technologic applications, and on termite hindgut communities. On the basis
of the two case studies, workshop participants outlined broad issues,
including modeling, infrastructure needed, and education and training, even
though some of the points made could be applied to a broader context.
Single-Organism System: Geabacter
To begin the workshop, Derek Lovicy, (professor and head of the
Department of Microbiology at the University of Massachusetts, Amherst)
presented the first case study, Geobacter. The family Geobacteraceae
includes over 40 known species of microorganisms, many of which have a
considerable impact on the environment. According to Lovicy, geobacters
are known for their ability to remediate contaminated water and sediments
and their ability to harvest electric energy from their surroundings. Other
microorganisms play a role in the breakdown of pollutants but Lovicy ex-
plained that Geobacter is special in that it dominates many microbial com-
munities in subsurface environments in which oxygen is absent but iron
oxides are abundant. Molecular studies, which avoid culture bias, have
demonstrated that geobacters account for about half the microbial commu-
nity in environments in which metal reduction is important. Not only are
3
OCR for page 4
4
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
geobacters more prevalent in these environments but they have been found
to be the primary microbial species carrying out such processes of interest
as iron reduction, uranium reduction, and the degradation of petroleum
contaminants. The ability to bring geobacters into the laboratory, where
they can be mass-cultured for biochemical and physiologic studies and for
genetic sequencing, makes them highly valuable for investigating the inter-
. , . . . . . .
action ot a micron ~a" community wits ~ its environment.
Bioremediation
Geobacters have different mechanisms for Fe(III) reduction from other
phylogenetically distinct Fe(III) reducers. Lovicy described the biochemistry
behind their method of iron reduction from Fe(III) to Fe(II) and described
how their unique form of respiration enables the application of geobacters
for bioremediation of groundwater. According to Lovicy, microorganisms
that use insoluble Fe(III) oxide as an electron acceptor can have an impor-
tant function in the carbon and nutrient cycles of aquatic sediments and in
the bioremediation of groundwater. These microorganisms are effective in
breaking down organic compounds, chlorinated solvents, and a variety of
metals. In addition to geobacters' ability to reduce iron because of their
unique respiration, these microorganisms can reduce other metals, includ-
ing gold and uranium. In fact, the species may make up as much as 85% of
the microbial communities in the subsurface during the most active phase
of an in situ uranium bioremediation project.
Harvesting Electric Energy
In addition to their ability to reduce metals at contaminated sites.
geobacters are capable of obtaining electricity from organic waste material.
More specifically, geobacters harvest energy, transferring electrons not only
within the cells but beyond, and the electrons can be harvested to produce
electric current.
Geobacters can inhabit mud, generating electricity while they degrade
pollutants. Furthermore, the microorganisms can be housed in an experi-
mental fuel cell and power devices that are buried in the ocean floor and in
remote locations. Geobacters can serve as biopowered antenna grids or
play a role in novel biosensing devices. They can convert such waste organic
matter as sewage sludge, domestic wastes, and industrial organic wastes to
electricity; and they can convert renewable biomass to electricity instead of
ethanol. They can harvest energy from hot environments with thermo-
philes, organisms that thrive under warm conditions (2 50°C). The study of
geobacters, accordingly, will benefit DOE in remediation, in the develop-
ment of cleaner forms of energy, and in biomass conversion to energy.
OCR for page 5
VDORKSHOP PROCEEDINGS
Genomics
An application of genomics has unraveled some of the key mechanisms
that allow geobacters to compete and thrive in subsurface environments.
Lovicy and his colleagues are pursuing parallel research on previously
cultured geobacters, novel culturing strategies to isolate environmen-
tally relevant geobacters, and environmental genomic DNA of "as-yet-
uncultured" geobacters to build conceptual models for optimizing uranium
bioremediation and electrical energy harvesting.
Sequencing the genome of previously cultured geobacters led to a number
of surprising discoveries. For example, he found a novel mechanism that
allows geobacters to access Fe(III) oxides. Earlier studies had demonstrated
that Geobacter species do not produce electron-shuttling compounds that
might alleviate the need for the microorganism to contact Fe(III) oxide
directly in order to reduce them. That raised the question of how geobacters
could find the Fe(III) oxides that serve as their primary electron acceptor in
subsurface environments. Although geobacters were believed to be non-
motile, inspection of the preliminary genome sequence revealed genes for
flagella slender outgrowth of cells used for swimming and other forms of
locomotion and that suggested that the microorganisms might be motile
after all, at least under some conditions. Laboratory studies allowed Lovicy
to confirm that geobacters are indeed motile, but only when they are grow-
ing with insoluble Fe(III) or Mn(IV) oxide as an electron acceptor. The
expression of the genes for flagella are highly regulated and coupled with
novel chemotaxis to access Fe(III).
The analysis of previously cultured geobacters revealed that the Geobacter
genome comprises a high degree of gene duplication in cytochrome genes
(i.e., certain cytochrome genes have a high percentage of identical se-
quences). Studies of differential gene expression led to the identification of
genes specifically required for Fe(III) oxide reduction.
Other surprises revealed in the genome sequencing include the discovery
of genes for new bioremediation capabilities, such as TNT degradation and
mercury reduction, and novel central metabolism genes that suggest that
geobacters are well adapted for acetate metabolism. Lovicy said that data
from the Geobacter species already sequenced, combined with systematic
manipulative experiments that allow analysis of differential gene expres-
sion, associated physiologic and biochemical studies, and modeling had
proved extremely valuable to his research team. "It has jumped us years
ahead of where we would have been in our research."
The genome sequences of four geobacters are available, and up to 10
Geobacter pure-culture genomes will be available in another year. In-depth
studies of gene expression with whole-genome DNA microarrays and
detailed proteomic studies are under way. An in silica mode! for the central
OCR for page 6
6
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
metabolism of Geobacter has been developed. Preliminary environmental
genomic studies suggest that uncultured Geobacter species in subsurface
environments have, at least in some instances, similar gene arrangements to
Geobacter species available in pure culture. Furthermore, billions of base
pairs of genomic DNA extracted from environments in which geobacters
predominate will be sequenced over the next 3 years.
In the conclusion of his presentation, Lovicy said that application of
pure culture and environmental genomic techniques to studies on geobacters
has enhanced our understanding of its role in environmental processes. He
raised the following issues regarding the application of systems biology in
environmentally relevant microorganisms.
· Although techniques for high-throughput analysis of mRNA and
proteins are available, high-throughput methods for quantitatively evaluat-
ing the metabolome of microorganisms are just as important but are not
readily available.
· Ways to mode! the influence of minor components of a microbial
community without a comprehensive analysis of each organism should be
developed.
· Mlcronlai ecologist snouicr be trained so that they have enough
understanding of basic microbial physiology and environmental science to
design and interpret meaningful high-throughput studies on environmental
microbial processes.
~ ¢- 1 · 1 1 · 1 1 1 1
Multiorganism Systems: Termite Hindgut Communities
Tared Leadbetter (assistant professor of environmental microbiology at
the California Institute of Technology) presented the second case study, on
termite hindgut communities. As many as 250 species of microorganisms-
many of which are strictly anaerobic inhabit the hindgut of a termite,
including bacteria, archaea, and protozoa. The termite shelters the micro-
organisms and feeds them with food that the insect itself cannot digest. The
microorganisms digest the food and metabolize the nutrients into a biologi-
cally available meal for the insect. This exquisite architecture of symbiotic
processes lies at the core of Leadbetter's work. Specifically, his research
focuses on the intricate dynamics of the hindgut community of the California,
or Pacific, dampwood termite, Zootermopsis angusticollis, and on the
mutualistic symbiosis formed between the termite and the microorganisms
. . . . .
wing in its ~ zinc gut.
Z. angusticollis is nothing but a "grinding machine," according to
Leadbetter. The termite feeds on wood, but it cannot digest cellulose, so it
grinds up the chips and passes them to the hindgut, where the flagellated
OCR for page 7
VDORKSHOP PROCEEDINGS
7
protozoa and other microorganisms digest the carbohydrates, converting
the sugars into the acetate that the insect uses for energy.
Termites are of interest to the scientific community for a number of
reasons, including their role as sources of greenhouse gases, such as carbon
dioxide, CO2 and methane, CH4. Those gases are released into the atmo-
sphere by the hindgut microbiota. Because of their high biomass densities,
termites, including their hindgut microbiota, contribute an estimated 2% of
global CO2 emission and about 4% of global CH4 emission. Many termites
are also the sites of intense nitrogen (N2) fixation into biologically useful
forms, an activity that stimulates the insect's proliferation on and mineral-
ization of a nitrogen-poor diet, thereby introducing net protein into local
food webs. Those activities are mediated by the hindgut microorganisms;
any intense perturbation of the hindgut community results in the termite's
loss of the ability to digest wood, to emit CH4, and to fix N2. The more
than 2,000 species of termites each serve as host to a different microbiota.
Because the microbial communities are different, each species of termite has
an array of biochemical processes in its hindgut distinct from other species.
Microorganisms in the Termite Hindgut
"Molecular inventories suggest that there are easily 100-250 microbial
species," Leadbetter said. A typical hindgut has a volume of 1-5 AL, but it
can contain 40,000 to 50,000 large protozoa and other microorganisms. In
one particular gut track isolated from a worker insect, more than seven
species of anaerobic protozoa were identified, as were methanogenic
archaea, and many aerobic and anaerobic bacterial phyla.
Leadbetter described several recent discoveries about termites, includ-
ing steep gradients of pH, oxygen, and hydrogen throughout the gut tract.
Hydrogen is produced in the anoxic core of the gut tract. As it diffuses out
of the gut, its concentrations plummet. In other words, the hydrogen
gradient counters the oxygen gradient, and this leaves the peripheral region
of the gut with very different physical and chemical environments from the
central portion. Because of the physical difference, the distribution of micro-
organisms in different regions of the gut tract is neither random nor
homogeneous. In fact, the termite hindgut can be considered the smallest
bioreactor on Earth. After discussing the peculiarities of the hindgut envi-
ronment, Leadbetter introduced a number of microbial communities that
live in the hindgut and presented an array of biochemical processes that are
carried out by them. He then described aspects of one of the processes,
called acetogenesis the process of generating organic carbon in the form
of acetate from inorganic carbon catalyzed by bacteria which is the focus
of his research at the California Institute of Technology.
OCR for page 8
8
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
Acetogenesis Research
Leadbetter pointed out that the termite gut fluid is typically dominated
by spiral-shaped organisms, which phylogenetically are spirochetes. There
are eight to 10 morphologically distinct types in one termite, but gene
inventories have suggested that there may be as many as 25 or even
100 types of spirochetes in the gut of one termite. Until recently, biologists
did not know the role of these organisms. Leadbetter was able to isolate
some of them, grow them, and test them in the laboratory. He found that
some of the spirochetes are bonafide homoacetogens, anaerobic bacteria
that reduce CO2 to acetate during respiration, also known as H2-CO2
acetogens. About 100 homoacetogens have been characterized, and most
are clostridia and their relatives. The addition of spirochetes to that list is
of interest, not only because of their role in the termite hindgut but because
they represent genetic diversity. Leadbetter wants to know how the spiro-
chetal pathway of acetogenesis compares with the "clostridial" Wood-
Ljungdah! pathway. Diversity among acetogens suggests the presence of
multiple niches even in this 1- to 5-,uL environment. If that is the case, the
diverse groups of acetogens warrant further investigation.
Molting and Reinoculation
At first glance, the orchestration of all the gut activities would seem to
be the most fascinating aspect of termite biology. But the termite also sheds
its hindgut contents when it molts. "The termite is born sterile and acquires
the microorganisms from its parents and siblings," Leadbetter explained.
"Every time it molts, it has to be reinoculated with microorganisms. There-
fore, the whole ecosystem is sensitive to extinction. If the host fails, it's all
over. Or if the microbial community fails the host, it's really all over."
New generations of microorganisms have been passed from termite to
termite in this manner for more than 12 million years. Leadbetter suspects
that the insect would have shed species that are not needed over such a span
of years, so he speculates that most of the 100-250 species of microorganisms
play some role in the hindgut. So far, Leadbetter has been focusing on
robust systems, but he would like to investigate some questions further.
For example, What are the first microorganisms that are reinoculated in the
hindgut after the molt? What role do they play?
Future Research Endeavors
In summarizing his research, Leadbetter pointed out that despite the
widely recognized role of termites as global sources of greenhouse gases,
such as CO2 and CH4, little is known about the function of most of the
OCR for page 9
WORKSHOP PROCEEDINGS
9
hindgut microorganisms, and much less about the insect- ant! atmosphere-
clerived physicochemical controls on lignocellulose mineralization. The
termite hindgut system is ideal for complexity analysis, which focuses on
one metabolic process carried out by various contributors (such as cellulose
digestion by protozoa or methane production by archaea) without consicI-
ering the detailed ancillary physiology of the contributors. Such analysis
could be conducted at the genetic level if specific genes that encode impor-
tant enzymes or transport proteins could be identified. The ecosystem can
also be perturbed and then the effects compared. Termites are amenable to
a range of physiological and ecological studies. The quandary is that many
techniques disrupt the architecture of the system, which is known to be
important. Better experimental designs are needed to disentangle the physi-
cal, chemical, biological, genetic, and temporal architecture of the termite
hindgut.
Leadbetter concluded his presentation with the following question and
remarks: "Even if the complete metagenome of the hindgut community
were available and sorted into all the files it should be in, what information
would that reconstruction provide? I'm excited, and I would love to have
those data. I'm just not sure how far they will go."
EXPLORING MICROBL\L SYSTEMS
The biological processes of microorganisms have been explored at mul-
tiple levels of observation, from gene regulation to the interactions of patho-
gens with their hosts. Recently, scientists have been interested in adopting
so-called! systems approaches to understand and harness complex biological
processes. Although systems analyses have most often focused on the op-
eration of a single cell or on the individual biological processes in an
organism, microorganisms do not typically exist as individuals or even as
pure cultures. Rather, they occur in consortia of related and unrelated
organisms in natural communities or in engineered fermenters. Thus, addi-
tional challenges in systems microbiology research include the need to define
a system (be it a single-cell system or a consortium) and its subsystems and
the need to integrate multiple scales. Including research on microbial
communities in system-based studies could provide a broader perspective
on controls of biological processes and how they operate in and among
. .
mlcroorgamsms.
Definition of a System
Before discussing how best to explore microbial systems, participants
in the workshop considered some characteristics of a mode! biological sys-
tem. Participants decided that such a system must have a fixed number of
OCR for page 10
10
inputs and outputs. It should be a single isolated entity with distinct bound-
aries. A half-cell, for example, is not a good system. A whole cell, however,
is often a reasonably good system. A single-organism community (such as
Geobacter) and a multiorganism system (such as in the termite hindgut) are
other examples of good systems. A biological system, in fact, can be on any
scale. A tidal salt marsh is an example of a whole intact ecosystem with
multiple variables with respect to the flow of water in and out of the sys-
tem. A system can be on the molecular scale. For example, a network of
genes that function by inducing and suppressing each other is a system.
Connecting the coarse-grain description of a single-organism or multi-
organism system with the fine-grain molecular system is a challenge. "There
is a tool, however, to help to connect the dots," said John Doyle (professor
of electrical engineering and control and dynamic systems at the California
Institute of Technology and workshop chair). "That tool is modeling and
modeling enables the additional tools of simulation and analysis."
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
. . . . . . ..
Daniel Drell (DOE) cautioned that modeling a microorganism or a micro-
bial system cannot be done all at once; such systems have to be modeled in
pieces. Those thoughts were echoed by several participants later during the
workshop. Michael Savageau (University of California, Davis) suggested
that one strategy for systems microbiology is to understand the subsystems
and try to put them together. As an example, he said that understanding all
processes at the RNA level would not provide information on how a cell
works, but knowing the interactions between mRNA, protein, metabolites,
receptors, and sensors in a subsystem and connecting the subsystems would
allow an understanding of how the whole cell system functions. Lovley
agreed that measuring every variable possible for example, every protein
or every protein-protein interaction may not be the best approach to study
a system, because of the large amount of data generated. That led to a
discussion on the modularity of microbial systems, that is the extent to
which a system is divided into modules connected by interfaces.
Even though living things are diverse, there is a striking unity among
them, particularly at the microbial level. All forms of life share the genetic
code, many biochemical processes and similarities in cell structure. In fact,
J. Doyle said that microorganisms appear to be well-designed systems with
a high degree of modularity that facilitates domain swapping and horizon-
tal transfer.
J. Doyle predicted that scientists one day will be able to model biologi-
cal systems by taking advantage of the modularity of microorganisms. Gene
swapping appears to be a source of confusion, he said, but the ability of
systems to swap genes suggests a degree of modularity. Systems with a high
degree of modularity are easier to model than those without it. Microbial
ecosystems are very attractive for that reason. As soon as biologists under-
stand more about the principles, the processes and how they function
OCR for page 11
VDORKSHOP PROCEEDINGS
11
together can be sorted out quickly. Although biologists are already familiar
with some of the principles, they know them implicitly, incompletely, and
more at the small-component level than at the network level. Engineering
often moves ahead by integrating modules as much as by creating new ones.
Savageau pointed out that if biologists make the wrong assumptions
about how the subsystems are partitioned, it may be difficult to link all the
subsystems to obtain a picture of the whole system. "There are many simi-
larities between biology and engineering, but there are some differences. In
engineering, you design the systems so that the interfaces are clean. In
biology, you have to discover them." In a microbial consortium, the system
can be partitioned at the community level, the cellular level, and the sub-
cellular level. The proper partitioning and the integration of different levels
of subsystems, particularly given the role of evolution in creating diversity
within those subsystems, pose another challenge to applying systems-based
approach to microbial ecosystems.
Geabacter and Termite Hindgut as Mode! Systems
The workshop planners chose Geobacter and termite hindgut organ-
isms as extreme cases of single-organism and multiorganism systems for
comparison and to maintain the focus of discussion. The workshop partici-
pants considered whether those two systems illustrate all the challenges
faced by microbiologists who apply the systems approach to study other
systems. Participants concluded that the two mode! ecosystems represent
only heterotrophic systems photosynthesis does not play a role in either.
Both are discrete systems that can be studied in isolation, so they do not
illustrate the difficulties of studying such complex systems as soil microbial
systems that cannot be studied in isolation and do not have controlled
inputs. Nevertheless, a systems-level approach to understanding, simu-
lating, and designing cells based on the two communities may contribute
substantially to biologists' genetic engineering of microorganisms and plants
for a variety of industrial and agricultural applications.
In comparing the two systems in a systems-biology context, Gary Sayler
(University of Tennessee, Knoxville) noted a large knowledge gap regarding
their cellular communication and cell-to-cell signaling. Sayler explained
that all kinds of communication go on in the environment. "It occurs
between and within species. It occurs between plants and microorganisms.
And now there is evidence that human hosts communicate with micro-
organisms. Undoubtedly, those processes are coming under fundamental
exploration at all levels, including mathematical simulations. They are
probably operating in microbial systems, too, but we don't exploit those in
ferreting out where they are." Sayler said that it would not surprise him if
Leadbetter came back next year and said that the termite was releasing a
OCR for page 12
2
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
pheromone that was communicating with the methanogens in holding down
population. "There is a lot of exciting work going on in systems biology
that requires the integration of biology, mathematics, and engineering,"
Sayler said.
Leadbetter shared a story about communication between a bacterium
living in close association with a phototroph, an organism that harvests
energy from light. One of the fascinating aspects of this consortium,
Leadbetter said, is that the phototroph contains a chromophore that is
attracted to light of a specific wavelength. The bacterial cell has a flagellum
and can actually swim toward light, but it does not have the chromophore,
so it cannot detect that light. "There has to be some kind of communica-
tion, not just at the transcriptional level, but actual, behavioral, real-time
communication between the phototroph that receives the gradients of light
and the bacterium that swims toward that light. These are two organisms
with very rapid communication related, in this case, to gradients of light."
T. Doyle thought that Leadbetter suggested an interesting example of a
deductive approach that draws inferences about what might be happening
by ruling out what cannot be happening. "You might know, for example,
what bandwidth the control system has to have, and from that you can
infer how fast the signal has to cross the system. Thus, you might know
that the signal probably cannot diffuse through the membrane, so it prob-
ably has to have some coupled receptors between the two organisms or
some other mechanism, such as data channels, to go between two mem-
branes. lust by knowing the speed of response, therefore, you could know
several things about what the mechanism could not be. This cellular com-
munication could not involve transcription, for example, and you could
rule out other things as well."
MODELING THE TOOL FOR UNDERSTANDING SYSTEMS
J. Doyle began the discussion about modeling by stating that "modeling
can play a huge role in helping us to understand biological systems. What is
needed is software infrastructure and new theory for systems biology."
The first point Doyle made is that simulation technology is essential to
connect multiple scales. He added, however, that existing simulation tech-
nology is inadequate because biological dynamics are stiff and stochastic
with lots of random variables and multiple scales. Processes in biological
systems occur in different time scales, and fast dynamics coexist with slow
dynamics. The statement holds true for modeling cells or collections of
cells. Further complicating the dynamics is the fact that biological systems
are extremely stochastic with different counts of random molecules in dif-
ferent time scales. Modelers are still struggling to develop a more unified
view of stiffness that encompasses the different timescales and stochasticity.
OCR for page 13
WORKSHOP PROCEEDINGS
13
Connecting scales in biology is different from connecting scales in physics
and chemistry; only some components of a biological system matter enor-
mously and others don't. Some components in a biological system could be
varied without disrupting the system because of its robustness. To "boot-
strap" (that is, to draw statistical inferences based on a data-based simulation)
an understanding of a system, modeling some of the detailed components
that do not greatly disrupt the system can be deferred until later.
Doyle's second point is that as critical as simulation technology is, simu-
lation alone is not an adequate tool. Suppose that all the genomic data of
an ecosystem are available. In principle, the characteristics of the available
metabolic and cellular processes are known and a mode! could be con-
structed to describe the ecosystem. However, many characteristics would
still be unknown. If new data become available, an exponentially large
number of simulations need to be performed within the parameters to
determine whether the proposed mode! is consistent with all the available
data. Simulation alone is not a scalable way to analyze a model, and the
alternative to simulations requires mathematics a novel approach even
. . . . .
wits fin t" be engineering community.
Finally, Doyle said fundamental laws that govern the organization of
biological networks still need to be discovered. For instance, the conserva-
tion of energy is a fundamental principle that applies in all biological sys-
tems, but other principles are not as well known. If such general principles
were known, it would be recognized that there are only a few ways to do
some things. Once the laws governing how networks can be organized are
known, the key components that should be included in a mode! can be
more easily identified. Therefore, biologists and modelers need to identify
some of the hard constraints in biological networks. Without knowledge of
the governing principles, scientists may never truly understand how a sys-
tem functions even if all its components are known. Most of the governing
principles are related to efficient energy flow, particularly resistance to fluc-
tuations, Doyle explained. "If you don't exploit these laws, the complexity
will be bewildering and overwhelming."
Doyle continued: "Microbial ecosystems are a fantastic domain in
which to study the basic questions in multicellularity that are important at
the core of biology, let alone the obvious immediate applied benefits that
everybody is well aware of in terms of just basic research. Modeling micro-
bial systems is going to be a very good challenge. If the modeling is not well
done, however, or if we don't develop the right new tools, we will never be
able to figure out how these systems work."
Eberhard Voit (Medical University of South Carolina) agreed with
Doyle that taking static snapshots of a system and comparing snapshots are
insufficient for studying regulation of systems. Modeling should be based
on dynamic measurements and detailed study of responses to defined inputs.
OCR for page 16
16
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
hypothesized that there must be an organism performing a particular
reaction in that environment, so he created a culture environment that
emphasized this reaction in which the organism survived and he isolated
those organisms. Doyle thought that was a good test for identifying hard
constraints.
A simple mode! might also provide valuable insight into spatial aspects
of an ecosystem. For example, in the Geobacter case study, models with
few parameters might provide insight into the effect of acetate depletion in
one location. Again, Doyle pointed out an instance in which Lovicy wanted
to explain why sulfur reducers were able to overrun the geobacters and so
he constructed a simple model. Lovicy reasoned that once the geobacters'
preferred electron acceptor is depleted, the next preferred one is sulfate; the
sulfur reducers flourish because of the abundant amount of acetate that is
available for their consumption. Doyle pointed out that although Lovicy
constructed a mode! with few interactions, it provided valuable insights.
Doyle elaborated that it is easy to include just a few more interactions, but
the combination of multiple interactions becomes bewildering. "Fairly
simple models that don't have any genes, and don't have many details, can
help you sort out what might be the next thing to do. You won't always get
the right answer, but it suggests how you might do the next experiment.
That's how you bootstrap yourself into deeper understanding."
To end this part of the discussion, Lovicy referred to his research with
geobacters. He repeated the story about how the use of models predicted
phenotypes that he could never have sorted out only by looking at the
genome. In part, that was because the Geobacter mutant grew better on
iron than the wild type. "That never would have entered my mind."
Are Detailed Data Needed for Modeling?
Timothy Donohue (University of Wisconsin, Madison) was troubled by
T. Doyle's vision that biologists did not have to worry about having all the
data and details. His concern was echoed by several participants. Using
Lovicy's work on Geobacter as an example, Donohue said that he would
like his datasets to be robust enough for a mode! to be able to predict
whether a particular gene is present. Although sequence analysis could
have predicted the presence of a particular gene in Geobacter, Lovicy puri-
fied the samples and demonstrated the presence and location of the gene.
Doyle, however, clarified that data are always incomplete. Some biolo-
gists believe that all the information are in the gene-coding sequence, but
there are large gaps in knowledge of cell dynamics. In dynamic situations,
regulation is much more complex and cannot be predicted from genomics
alone. "The good news is that you can defer knowing all those details. You
can get somewhere and make predictions. You can move ahead without
OCR for page 17
WORKSHOP PROCEEDINGS
17
having to know everything. You can have incomplete data, incomplete
models, and incomplete knowledge; but you use what you have to predict
the next discovery."
Lovicy does not worry about every component of central metabolism in
Geobacter, which he can mode! fairly well, but respiration is not under-
stood and cannot be sorted out by looking at genomic sequences. Until he
has that information, he cannot predict other responses in Geobacter. At
some point, the function of a particular subset of genes needs to be eluci-
dated. He asserted that "every genomic project should study the physiology
of an organism."
Leadbetter agreed with Lovicy's view that the structure and architec-
ture of some ecosystems, such as the tidal salt marsh, can be elucidated on
the basis of a few physical characteristics. For example, if the effects of
oxygen, pH, sulfite, and light are taken into account, the biological struc-
ture of sediment, in which there might be 1,000 species, can be understood.
The stratification of the salt marsh can be clearly modeled on the basis of
knowledge of those four components. A day and night cycle and maybe
seasonality can be included to improve understanding of the underlying
architecture of the community. But every system is different. "I would love
to have that type of information on the termite, but I don't know what the
four components are. If I did, it would change everything. With the termite,
we need to find out what the critical components are before predictions can
be made."
T. Doyle urged the group to keep their eyes on the big picture and to
apply high-value-added details only to models so as to reach their goals.
"People who mode! human diabetes processes are helping in drug design,
but they have no idea what most of the details are. The idea that you have
to know how every gene works with every protein is unrealistic. The point
is to figure out what we can do now that enables us to move ahead without
having to know everything."
Jizhong Zhou (Oak Ridge National Laboratory) supported Doyle's
premise of the need to stay focused on the big picture. Microbial systems
are very much a "black box," and not all the details are necessary for con-
structing models, he said. Hastings spoke of the advances in population
models made possible over the last 15 years by quantitative statistical
approaches for matching them to data. Scientists are gathering and struc-
turing information on what used to be black boxes.
Doyle suggested the importance of focusing on key areas. "For example,
if we were going to look at metabolite concentrations in some detail, some
would give us an enormous amount of information, and some would give
us little. You could spend the next decade spending huge amounts of money
in measuring concentrations of metabolites in glycolysis in Geobacter, for
example, and you would learn almost nothing. But some things are show-
OCR for page 18
18
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
stoppers. For example, understanding of some aspects of respiration is
necessary for knowing how organisms function. So you have to measure in
a targeted way. If you are looking at genes and proteins, you don't want to
do everything. You want high-value-added data."
Hastings suggested that ecologists had also struggled with the issue of
whether detailed data are needed for modeling. "About 60 years ago,
ecologists split into two camps and they haven't talked to each other much
since. One camp looked only at energy flow and exchange of materials.
The other camp looked only at the dynamics of species numbers. We keep
hitting the same issues: can you just look at input and output without con-
sidering which species are and are not present? Or do you have to look
only at the species that are present? Those are fundamental questions.
The group considered the issue from a variety of perspectives and, with
T. Doyle, concluded that it is clear that taking dynamic, in situ measure-
ments is difficult. Modeling can play a big role in combining sequencing,
which provides a static picture and lots of data with a few key dynamic
measurements. That type of work can be expensive, but it is worth doing
because dynamic measurements and responses to inputs are essential for
learning about regulation, according to Voit. There is no doubt that snap-
shots are valuable and necessary, but eventually dynamic experiments and
analyses have to be performed if we are to understand the structure and
function of a whole system.
To conclude the discussion on modeling, Doyle said, "Again, the point
is, you want to have enough information to be able to say, 'I don't know
what all the dynamics are, but I know that the basic dynamics in the cell of
greatest interest in this system occur here.' So you know where to look
without knowing the entire answer. That's the sort of bootstrapping that
we hope will continue to work."
TOOLS FOR SYSTEMS-MICROBIOLOGY STUDIES
After discussing modeling of the two case studies, the group turned its
discussion to the tools, databases, and personnel that are needed to facilitate
the modeling of microbial systems. A systems approach to research in
microorganisms and microbial communities requires novel experimental
tools for detecting and measuring biochemical processes such as gene
expression, metabolic activity, and protein production across the entire
system, whether the system is defined as a genome, a cell, or a cell culture.
To make sense of information generated from measurements of system-
wide activity, assays must be repeated continually, each time monitoring
the system under a different set of conditions.
The statistical analysis of the resulting data is likely to be compu-
tationally intensive, and mathematical and computer models are needed to
OCR for page 19
WORKSHOP PROCEEDINGS
19
capture relationships in a way that enables the visualization and quantifica-
tion of the underlying events that give rise to system-wide activity. Analysis
of the data will be facilitated by knowledge about pathways, molecular
relationships, and operational mechanisms knowledge that can be
acquired through traditional hypothesis-driven studies. Combining data
from multiple research approaches will enhance our understanding of bio-
logical systems.
Experimental Tools
Workshop participants were asked to think about the technologies and
tools that could facilitate systems-microbiology studies. What technologies
are needed for exploring microbial systems? What are the corresponding in
viva and in situ tools that would provide insights into the dynamics? What
technologies are going to be available for dynamic measurements?
Donohue believes that many creative scientists could develop tools for
database management, dynamic measurements, and so on, if they had
appropriate funding. Sayler suggested that an environmental genome for
prokaryotes would be a useful tool. "My students don't even want to make
microarrays", Sayler said, adding that he would like to purchase such tools
as a phylogenetic microarray so that his students could be actively engaged
in the experimental work, such as perturbation experiments to study responses,
instead of spending time in developing a microarray. A genotyping chip
with diagnostic sequences could help to determine which microorganisms
are present in a community, their activity and dynamics. An appropriate
chip-based analysis can be very informative despite its low sensitivity. Sayler
suggested that more commercial biotechnology companies are needed to
produce tools that facilitate and accelerate biological research. One such
company is Biolog, in Hayward, California, which makes the Phenotype
MicroArrays™ that can quantitatively measure thousands of cellular pheno-
types at once. The company also makes systems that can rapidly identify
over 1,900 species of aerobic and anaerobic bacteria, yeasts, and fungi
thttp://www.biolog.com/main.htmI] .
As an example of one funding mechanism that enables investigators to
buy microarray chips, Leadbetter credited the Cystic Fibrosis Foundation
thttp://www.cfri.org/home.htm] for developing an especially creative strat-
egy for generating research about cystic fibrosis. "The Foundation makes
metric chips available for about $150 if you write to them with a short
proposal and say, 'I'm going to study this particular aspect of Pseudomonas
aeruginosa. It may have some connection, some way, or somehow, to cystic
fibrosis.' You can then buy 10 chips for $1,500, and they even give you the
software free. This policy has stimulated research in the field: it changes the
dynamics of researchers who might be afraid to fail with a $2,000 chip."
OCR for page 20
20
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
Lyle Whyte (McGill University) asked what high-throughput tech-
niques, aside from phylogenetic microarrays, funding agencies like DOE
can support. Zhou said that typical microarrays are easy to make but insuf-
ficient for some kinds of investigations. For the termite hindgut, for
example, which may have more than 100 species of microorganisms in a
single insect, new high-density microarrays are necessary. However, even if
there were microarrays for 100 microorganisms from the hindgut of one
termite, there are more than 2,000 species of termites to consider.
Zhou suggested proteomics as another field in which DOE could help
to advance development so that the proteins that carry out a function can
be identified and how they function together can be understood.
Tames Frederickson (Pacific Northwest National Laboratory) empha-
sized the importance of basic biological research. "I think it's safe to say
that we don't know the entire function of networks, even in E. coli, which
has been studied to death. We still don't know the function of 30-40°/O of
genes. Moreover, the known genes may serve functions that we are not
aware of. We haven't even talked about proteins; identifying the proteins
that carry out the work of the cell is important, and we need to understand
how proteins interact and work together to carry out functions at the cellu-
lar level. Many fundamental issues need to be understood before we can
move on to other things."
Frederickson noted that a lot of the technologies being developed-
such as proteomics and transcriptome analysis are based on using a large
biomass of cells, around 109 cells. But there is large variation among
individual organisms. Furthermore, many assays are based on in vitro
techniques that need to be confirmed in viva. There is much room for
developing approaches and technologies that are centered on confirming
such assays.
Many new techniques have been developed to allow single-cel1 observa-
tion, according to Patrick Dennis (National Science Foundation). The
classic example is the observation of bacterial motility, which was observed
recently by taking pictures with a camera. "That's a type of situation where
you can extract a tremendous amount of information by looking at single
cells, rather than populations." Other novel technologies will contribute to
an increased understanding in this field.
Databases and Ontologies
There are no analytic tools for taking data on gene expression and
comparing the experimental results to decipher the biological meaning. For
example, Sayler conducted a study, in collaboration with researchers in
chemistry and medicine, on the influence of natural anticancer compounds
on the human gene array. In the first round of experiments, they narrowed
OCR for page 21
WORKSHOP PROCEEDINGS
21
down the potential anticancer compounds to 2,000 cytokines and
chemokines. "The data that came out of the first series of experiments were
good. The problem was that the biologists had to work virtually on a gene
to gene basis to find out whether a gene response had any relevance to the
cancer process." Tools are needed to analyze such large databases. T. Doyle
commented that although software would aid biologists in comparing
massive amounts of data and deciphering their meaning, it is generally
expensive to develop, and so far no one wants to pay for it.
Donohue emphasized the importance of access to reliable, updated data-
bases and the ability to compare data. He found that databases are over-
whelmed with 2-year old annotation and cannot be updated fast enough.
Different data on a given organism may be found in several databases,
depending on who submitted the data. Moreover, the same organism may
be labeled differently because of outdated information. Often, different
software is built for each microorganism, and this makes cross-comparisons
difficult. Doyle questioned whether the fundamental issue is that the correct
abstractions for organizing biological information need to be developed or
that programmers are still needed to piece together well-developed abstrac-
tions. Donohue believes that it is probably both. If databases are linked
together seamiessly, researchers should be able to access all the records for
a given organism in different databases.
Another too! is an ontology of functional modules that allows researchers
to organize vast amounts of scientific data electronically. (An ontology is a
description of the concepts and relationships pertinent to an entity or com-
munity of entities.) The objective of an ontology is to enable communica-
tion between informatic systems in a way that is independent of individual
system technologies, information, architecture, and application domains.
A structured language that can be put into the database or the ontology
allows a computer to look for associations between data. An ontology
forces scientists to think systematically about intermediate-level architec-
tural issues and about the different levels of systems. Doyle suggested that
not only does a current ontology need to be developed to help today's
researchers in studying systems microbiology, but the framework should
allow ontologies to evolve systematically and robustly. The hardest thing
for software engineers to develop is a system that is both usable now and
also flexible enough to adapt to technologies of the future, Doyle told the
group.
There has been some effort by software engineers to connect databases
and to create a distributed system whereby researchers can view their own
annotation in a central searchable system. Any new database would be
registered to the central system so that researchers can query the central
system and obtain information from different databases. Donohue is not
convinced that the databases are well connected, not to mention the short-
OCR for page 22
22
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
comings of some databases, such as GenBank, the National Institutes of
Health genetic-sequence database that is an annotated collection of all
publicly available DNA sequences. Donohue has found many errors in
GenBank, which he says must be updated. "As an end user, when I go
looking for something, I know it should be set up differently." Not only is
GenBank often outdated, but the databases are not linked properly and are
not "talking to" one another, he said. And how genes are identified is
confusing. "Bacteriologists don't know what's out there. They need a Web
site like the yeast Web page where from a gene you can get chip data,
proteomic data, hybrid data, and functional data and go right to the cita-
tion that's what everyone wants."
Doyle believes that basic research on linking databases still needs to be
done. He suggested the Systems Biology Markup Language (SBML) as a
success story in the context of modeling. SBML is a computer-readable
format for representing models of biochemical reaction networks. The
development of SBML benefited from the fact that few groups were work-
ing on a dynamic systems level and on control-regulation levels, so the
groups could easily agree on standards. Most of the major simulation
systems enable models to be shared and published in forms that other
researchers can use even in different software environments. The key to
SBML's success is having good software engineers who concentrate on
making the system work. Doyle suggested that DOE could conceivably
facilitate development in this field.
Doyle posited the hydraulic system of the Boeing 777 as a perfect
example of systems that had problems initially and great success later. The
hydraulic system of the Boeing 777 has two separate models, he explained:
one describes the layout of the hydraulic lines, and the other the dynamics
of the hydraulic system as a component of the control system. The first
mode! is important because the people who wire up the lines are different
from the people who decide where the structures go. In the past, when
engineers built an airplane, the specifications called for running a hydraulic
line directly through a big piece of metal, which, of course, cannot be done.
They had to decide whether they should move the hydraulic line or drill a
new hole. Before they could drill a new hole, however, they had to consider
whether it would compromise the structure of the plane. Boeing has since
invested about $1 billion in software so that it can test the system virtually
and eliminate those sorts of errors. The software system was expensive, but
Boeing made the money back in saved manufacturing and other costs. The
model of the dynamics of the hydraulic system had to be developed inde-
pendently of the model of the layout. A goal of research is to have more
integrated models, and this is a big issue in biology.
Doyle again pointed to software that is generally extremely expensive
and difficult to build. No one wants to pay the bill, despite the likelihood
OCR for page 23
VDORKSHOP PROCEEDINGS
23
of enormous long-term savings. In the case of systems microbiology, the
fundamental issue is that the correct abstractions for organizing informa-
tion need to be developed. In some cases, the abstractions exist but await
skilled programmers to put them together properly.
LOOKING AHEAD COLLABORATION AND EDUCATION
With the promises of microbiology well within reach, the group
explored what might be done to facilitate exploration in molecular eco-
systems. How might DOE and other funding agencies fuel research and
accelerate the process? What are the challenges of integrating disciplines?
How do we cross disciplinary boundaries in the short term? What can we
do in the long term to educate and train participants to join efforts in the
quest for scientific discovery?
Collaboration
Among the more obvious ways to encourage collaborative projects is to
create more funding opportunities to drive research. Fredrickson suggested
that DOE and other agencies build such goals into their programs. Funding
for each project needs to be increased to accommodate the size and scope of
interdisciplinary research and to convince investigators to collaborate and
form teams that spark innovative research. Fredrickson remarked that
opportunities need to exist before researchers can come together.
Finding funds is not so easy, according to Dennis. The National Science
Foundation (NSF), for example, has a limited budget to support all types of
research; historically, single-investigator grants seem to be the most cost-
effective and yield the most return on the research dollar. "We've just
heard how important basic science is for understanding lac operon, for
example. We can take money away from single investigators and dump it
into large collaborative interdisciplinary projects, but it's not going to have
the same impact."
Dennis went on to say that NSF has a new Frontiers in Integrated Bio-
logical Research program that provides $1 million a year for 5 years. Many
researchers believe that such funding is insufficient to meet the goals at
hand and that pulling together a large team for a short period like 5 years is
not realistic. Furthermore, such collaboration is risky, particularly for
younger researchers who will be judged on whether they got grants from
NSF and what has come out of their laboratories. Whether they partici-
pated in a group project and their achievements in collaboration with others
are considered less important.
Another question centered on what kind of research NSF is looking for.
The agency claims to support creative, imaginative science, but T. Doyle
OCR for page 24
24
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
argued that in general innovative proposals are almost guaranteed to get a
"poor" rating from some reviewers and to be doomed to fail. Dennis con-
firmed that NSF is looking for creative, imaginative science, which he
defined as the kind of small-scale science that might have a big impact down
the line studies on restriction enzymes, for example. Dennis defended his
agency's practice of limiting the lifetime of research grants by explaining
that science evolves. Resources may need to be reallocated to address
different issues at different times.
. ... . . .
SClentlilC community IS Saying."
"NSF must pay attention to what the
Lovicy said that at first collaborating with modelers on Geobacter was
difficult for him; later he accepted that he did not have to know everything
about an organism. "Once I got beyond that, it became a good experience
because the people who do the modeling are engineers, and they ask ques-
tions we never had thought to ask, because they have a different perspec-
tive. I think it's gone incredibly smoothly even though, in our case, we are
on opposite coasts." Hastings was quick to point out that the engineers
doing the Geobacter modeling had to learn enough biology to be able to
communicate with Lovicy; this led the workshop group to discuss biolo-
gists' and engineers' problems in learning enough about each other's field to
be able to communicate.
Undergraduate and Graduate Programs
There was a consensus that an interdisciplinary approach to science
begins at the undergraduate and graduate levels. How to bring it about was
less clear. One suggested strategy involved transforming graduate-level mi-
crobiology courses into one integrative microbiology course to make room
for quantitative courses, but this strategy was unpopular with the biologists
in the group. Lovicy objected to universities' reducing their offering of
microbiology courses. "What if you like microbial genetics?" he asked.
Leadbetter agreed with Lovicy and elaborated on how some of his students
are forced to devote as much as 80% of their time and effort to passing
quantitative courses at the expense of the other courses they are taking.
The students end up not getting much out of any of the courses. "I'm much
more excited about getting people who like what they do and know how to
do different things. Then we can try to build connections between them.
And because they like working together, they will probably communicate."
The workshop participants supported the notion that curriculum con-
stitutes a serious problem. But what courses should be eliminated to make
room for new ones? Can a subject be compressed? How? How can people
in different disciplines be trained with enough overlap to be able to commu-
nicate with each other? How much mathematics does a biologist need?
OCR for page 25
VDORKSHOP PROCEEDINGS
25
Quantitative courses enable biologists to assimilate and trust informa-
tion that they get from computer scientists, said Donohue, "but we need to
talk with our friends in the computer field to know the courses that are
relevant." Donohue explained that the quantitative courses that his biology
students are considering either have many requirements that his students
cannot meet or have no relevance to their work in biology. The group agreed
that courses in statistics, calculus, and computer science would help biolo-
gists in their research and studies. Those courses should be rigorous in
presenting problems and examples in biology, chemistry, physics, and eco-
nomics that science students can relate to.
Dennis suggested that biology needs to be taught in a mathematical
framework, which has rarely been done in the past. In contrast, engineer-
ing and many other disciplines have been teaching mathematics in their
own contexts. Biologists may have been teaching in cubicles chemistry,
biology, physics, mathematics, computer science, and engineering and a
more integrated curriculum needs to be developed. Core courses intended
for students outside a discipline should be designed carefully with a specific
focus, according to Savageau. As it is now, when nonbiologists first see a
biochemistry text, they are overwhelmed by it and quickly lose interest.
However, it is possible to teach the principles of biochemistry without all
the details that specialists think are interesting.
Hastings proposed that scientists change how they think about doing
science. He suggested that scientists in different disciplines consider using a
common language in their collaborative work. Otherwise, they will be
talking past one another. He described a course that he teaches to graduate
students, some of whom are majoring in applied mathematics and in popu-
lation biology. His students in different disciplines communicate with each
other, he said. "If you can get the students to sit down together outside
class and without faculty members around that's the real key. It will
break down the barriers. We need to design projects with questions of
interest to you as a microbiologist and questions of interest to you as an
engineer. That's what develops collaborative skills and builds expertise in
other backgrounds over time. We don't do a good job of teaching people to
collaborate."
Future scientists may be encouraged to collaborate through carefully
designed programs at the undergraduate and graduate levels, but post-
doctoral researchers may be encouraged by the funding of projects that
force them to work across disciplines with biologists, mathematicians, and
computer scientists. Experienced scientists can look for new short courses
specially designed to share knowledge across disciplines. A number of par-
ticipants told the group about various short courses that lasted from a
couple of hours to a couple of weeks. Frank Doyle (University of California,
OCR for page 26
26
PROMISE AND CHALLENGES IN SYSTEMS MICROBIOLOGY
Santa Barbara) stated that the time he spent on such special short courses
had been worthwhile.
Learning enough to be able to speak the language, however, is insuffi-
cient. Working toward something that draws everyone together is the higher
ideal. A number of workshop participants suggested that collaboration
would be encouraged by identifying concrete challenges in subjects of com-
mon interest at the intersection of various disciplines. It is important not to
focus systems microbiology only on microbiology, modeling, and computa-
tional biology; that would exclude knowledge essential to comprehending
the complexities of these systems. Biochemistry, organic and inorganic
chemistry, physics, economics, and other disciplines all play roles.
CONCLUDING REMARKS
In the final discussion, Donohue cautioned that the value of basic bio-
logical research and benchwork such as genome sequencing, which Lovicy
had described earlier as the too! that led to the discovery that geobacters
have genes for flagella should not be neglected. The goal should be to
train a group of well-rounded students and postdoctoral researchers. "Ten
years from now, you don't want to have a group of students and post-
doctoral scientists who know how to look at chip data and make models
but don't know how to purify an enzyme."
T. Doyle, workshop chair, brought the meeting to a close when he said
that he had learned about two important systems over the course of the
workshop and, as an engineer, was becoming as interested in microorgan-
isms as the biologists in the room were. As mentioned earlier, the shared
interest across disciplines is the key to driving the exploration of microbial
ecosystems. But Doyle said that the biology community has yet to recog-
nize the need for specialists to develop theory and software infrastructure in
order to advance systems biology. "One of the things the biology commu-
nity and the funding agencies have to recognize is that if you want to have
technology that works, you have to spend the money to build infrastructure.
Not everyone needs to know how to do that, but somebody does. That is
where systems biology has to go. You must have the infrastructure to
address problems. Microorganisms are the foundation of the biosphere, but
they have been neglected."
Representative terms from entire chapter:
governing principles
