Systems Biomedicine Concepts and Perspectives 1st Edition Edison T. Liu PDF Download
Systems Biomedicine Concepts and Perspectives 1st Edition Edison T. Liu PDF Download
https://ebookfinal.com/download/systems-biomedicine-concepts-and-
perspectives-1st-edition-edison-t-liu/
https://ebookfinal.com/download/exploring-chemical-concepts-through-
theory-and-computation-1st-edition-shubin-liu/
https://ebookfinal.com/download/face-geometry-and-appearance-modeling-
concepts-and-applications-1st-edition-zicheng-liu/
https://ebookfinal.com/download/making-sense-of-intersex-changing-
ethical-perspectives-in-biomedicine-1st-edition-ellen-k-feder/
https://ebookfinal.com/download/microemulsions-background-new-
concepts-applications-perspectives-1st-edition-cosima-stubenrauch/
https://ebookfinal.com/download/endothelial-biomedicine-1st-edition-
william-c-aird/
Systems Biomedicine Concepts and Perspectives 1st
Edition Edison T. Liu Digital Instant Download
Author(s): Edison T. Liu, Douglas A. Lauffenburger
ISBN(s): 9780123725509, 012372550X
Edition: 1
File Details: PDF, 13.31 MB
Year: 2009
Language: english
Academic Press is an imprint of Elsevier
525 B Street, Suite 1900, San Diego, CA 92101-4495, USA
30 Corporate Drive, Suite 400, Burlington, MA 01803, USA
32 Jamestown Road, London NW1 7BY, UK
No part of this publication may be reproduced, stored in a retrieval system or transmitted in any
form or by any means electronic, mechanical, photocopying, recording or otherwise without the
prior written permission of the publisher.
Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in
Oxford, UK: phone (44) (0) 1865 843830; fax (44) (0) 1865 853333; email: [email protected].
Alternatively visit the Science and Technology Books website at www.elsevierdirect.com/rights for
further information
Notice
No responsibility is assumed by the publisher for any injury and/or damage to persons or property
as a matter of products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions or ideas contained in the material herein. Because of rapid advances in the
medical sciences, in particular, independent verification of diagnoses and drug dosages should be
made
ISBN: 978-0-12-372550-9
10 11 12 13 14 15 10 9 8 7 6 5 4 3 2 1
Overview
Douglas Lauffenburger
VII
VIII Overview
Mechanistic complexity
Sequence Expression Localization Dynamics Populations
Biological complexity
Cells Tissues Organisms Populations
Dynamics
Systems complexity
Single gene Genomic complement Cells
Sequence
em ic
ol e
e
t
pl m
en
m On
ul
m no
ec
co Ge
systems complexity is schematically illustrated technologies and experimental approaches that
in the figure (originally developed by Peter is changing the conduct of biological research.
Sorger for the MIT Computational & Systems The availability of whole genome sequences
Biology Initiative). provides the complete catalog of genetic knowl-
In this book, then, systems biomedicine can be edge of an entire organism. Multiplex sensors
described as an emerging approach to biomedi- such as expression arrays and multi-channel
cal science that seeks to integratively infer, anno- flow cytometry, and high throughput screening
tate, and quantify multi-variate complexity of the maneuvers generate precise and comprehen-
molecular and cellular processes of living systems, sive data. Contemporary and developing com-
with ultimate aim of constructing formal algorith- putational capabilities are sufficiently powerful
mic models for prediction of process outcomes to envision capability for computing model-
from component input. Systems approaches are based inferences and/or predictions even as
characterized by several key attributes: the magnitude of systems (in terms of number
of components and their interactions) and asso-
1. A pursuit of quantitative and precise data;
ciated data-sets continue increase. The differ-
2. The comprehensiveness and completeness of
ence between systems and reductionist biology
the datasets used;
is in the objectivity with which we can analyze
3. A focus on interconnectivity and networks of
complex data and the resolution afforded by the
the component parts;
completeness of the datasets.
4. A willingness to define, measure, and
Furthermore, systems biology does not
manipulate biological complexity;
remain constant from year to year. Obsolescence
5. An interest to computationally (and
occurs in a matter of months. For this reason,
therefore quantitatively) predict outcomes.
this book has been written and assembled as
Certainly it can be said that all of biological a series of linked essays that convey strategies
research historically could be characterized by and processes. The arguments are bolstered by
these descriptors. Any scientific endeavor seeks commissioned chapters on specific topics dis-
to measure and systematize observations (quan- cussed in depth. These should be considered
tification) and, in finding underlying order as examples to clarify points and to stress con-
(model), would allow scientists to predict out- cepts rather than as an encyclopedia of past
come. However, there is an ongoing evolution of knowledge. There will be more departures from
Overview IX
an expected book on systems biology. Our dis- narrative should be heard, starting with Chapter
cussion will extend from model systems to 1 (by Liu) which offers a conceptual introduc-
human biology and pharmacology. We will tion to systems biomedicine.
focus on applications of systems approaches The first section of the book lays experi-
to medical problems and thus the title Systems mental groundwork. It begins with summa-
Biomedicine. Some would demand that true ries of experimental technologies in genomics
systems approaches require precise mathemati- (Chapter 2, by Liu) and proteomics (Chapter 3,
cal models; however, in this book, because of by Hanash), to set a foundation for the observa-
the experimental complexity in human sys- tions and measurements which motivate, popu-
tems, we wish to broaden the inclusion criteria late, and test associated computational models.
for systems biology to qualitative systems and Chapters 4 (by Lauffenburger and Liu along
hypothesis generators. with associated colleagues) and 5 (by Lim)
Our attempt to describe systems medicine is describe molecular networks regulating cell
our final experiment. Often, the most rational functional responses to environmental inputs,
experimental strategy is to identify the simplest, which form a basis for a wide variety of envi-
most definable model system to study and then sioned models. These are followed by presenta-
to construct a computational model around tions of two particular manifestations of these
data output from such systems; ergo, the use of networks – cell/matrix adhesion networks
phage, microbial systems, and yeast. However, (Chapter 6, by Geiger and colleagues) and net-
such systems strategies can now be applied to works regulating stem cell behavior (Chapter 7,
more complex mammalian systems and even by Ng and colleagues).
to study human disease. The experimental sys- The second section of the book focuses on
tems approaches to studying a human problem mathematical and computational methods for
will, by necessity, be different and potentially modeling of these kinds of molecular networks
less complete than attacking a question using and consequent cell behaviors. Chapter 8 (by
prokaryotes simply because the possible solu- Subramaniam and Maurya) starts by outlining
tion space is orders of magnitude greater. fundamental challenges for network modeling,
Nevertheless, productive strategies have been followed by three chapters describing differ-
tried and the outcomes have proven useful even ent modeling approaches. Chapter 9 (by Janes,
in drug development. Woolf, and Peirce) focuses on “high level”
We are attempting to organize this book in approaches, which emphasize relational and
a manner reflecting important distinguishing logical operations of molecular and cellular
characteristics of systems strategies in experi- processes, whereas Chapters 10 (by Doyle and
mental biology and medicine: comprehensive Petzold and associates) and 11 (by Loew and
(even though not exhaustive) and quantitative associates) focus on “low level” approaches in
measurement, using quantitative data to con- which details of physico-chemical mechanism
struct a model of the system, and defining com- are incorporated. This section is rounded out by
plexity as an experimental dependent variable. Chapter 12 (by Sauro and Bergmann) discussing
Finally, we explore the applications of these modeling software.
principles to biomedical problems. Finally, the third section offers some early
Rather than an assembly of independent attempts at application of systems biology per-
entries or chapters, we have composed this book spectives to particular biomedical science areas
as a narrative. Whereas we cannot project how and pharmaceutical industry challenges. With
this book will ultimately benefit our readers, respect to physiological areas, Chapter 13 (by
we suggest that it is best read in sequence as a Hunter and Cooling) directs systems modeling
Overview
Foundations of systems
biology
C H A P T E R
1
Foundations for Systems Biomedicine:
an Introduction
Edison T. Liu
Genome Institute of Singapore, Singapore
o u t l i n e
earlier quantitative traditions. As with any new in complex organisms, and the later generations
field, there are many opinions as to what systems of William Harveys had rendered pulmonary
biology is. For the purposes of this book, systems and cardiac physiology into equations. To a
biology can be described as a discipline that seeks large extent, this approach has been remarkably
to quantify and annotate complexity in biological successful, and has brought us many of the
systems in order to construct algorithmic models medical advances in cardiopulmonary medicine
with which to predict outcomes from compo- and surgery. The cardiac diagnostics from angi-
nent input. Systems biomedicine is an extension ography, to echocardiography, to telemetry in
of these strategies into the study of biomedical which patient physiologic output is monitored
problems. We believe that this demarcation is rel- and automated alerts generated, represent a cul-
evant, given the challenges of the complexity of mination of such research in cardiac function. In
the human organism and the human impact of a sense, physiology was the systems science in
the results of these investigations. medicine. However, these organ-level models
This definition of systems biomedicine high- do not parse with molecular realities, because
lights the difference between quantitative data their unit of measure is in average blood flow,
acquisition and systems biology. The scale of for example, and not in the flow dynamics of
data acquisition in biology today is unparal- the red corpuscle. Therefore, in the past, quan-
leled in history. Analog and descriptive data titative physiologic models could not be unified
such as cellular images are now digitalized and with cellular models and, by scale, to molecular
converted to discrete data points. Genomic- and models. Moreover, the need that assumptions
proteomic-scale information is registered in the be greatly simplified in order to arrive at com-
gigabyte scale per experiment. This reality also putationally tractable models also limited the
demands formal mathematical and algorith- relevance of many physiologic models.
mic conversion of experimental data in biology Now, however, medicine is becoming amena-
in order for them to be simply understood by ble to complexity analysis. The understanding of
the investigator. The interposition of comput- the cell and molecular biology of human disease
ers and their algorithms as an essential part of has dramatically advanced in the past 25 years.
biological research immediately places, at least Whereas the pathophysiology of most human
a rudimentary, mathematical formalism around diseases was previously limited to the analysis
all experiments performed in this fashion. of organ failure, most diseases now have a cel-
Although measuring outcomes is standard lular and molecular explanation. It is precisely
in day-to-day biological experiments, these ear- this reduction to common units of measure—to
lier quantitative approaches do not scale. While the cell and the molecules within the cell—that
detailed biochemical kinetics can be calculated allows systems analyses to be applied across the
for a single biochemical reaction, most com- entire human condition. Therefore, the pump
monly, we have tended to resort to descriptive dynamics of the heart after myocardial infarc-
generalizations when we ascend to physiological tion can be resolved at the same level as pancre-
scales. With current technologies that can acquire atic beta-cell function in diabetes mellitus. There
precise, comprehensive and quantitative data, is convergence.
biological complexity can now be quantitatively The current systems biology now includes two
analyzed. The challenge, however, is to iden- important new characteristics that distinguish
tify the optimal mathematical approaches most it from historical physiology and mathematical
suited for this scale and complexity of analysis. biology. First, there is a focus on complexity; sec-
Physiologists and pharmacologists have ondly, the fundamental unit of study resides in
always sought to quantify inputs and outputs the DNA (and, by association, protein) sequence.
That the unit of measure can be the nucleotide Geneticists have already defined phenotypic
now provides the lingua franca that permits the interactions between genes or alleles as epistasis
direct translation of experimental results from (Phillips, 2008). In many cases, new properties
biochemistry to cell biology, to physiology and emerge: two white flowers that when crossed give
to population genetics. Moreover, the ultra-high- a purple flower, or two genes that when individu-
throughput and multiplex genomic technologies ally mutated give no phenotype, but show a lethal
allow for the digitalization of experimental data outcome when both are mutated. The mathemati-
of such precision and comprehensiveness that the cal representation of epistasis can be:
true complexity of a biological system can actually
be measured and dissected. In all aspects—bio- (1.1)
logical and mathematical—the greatest advance
has been the availability of computational capa- where W is the observed phenotype, x and y
bilities that can match the systems complexity. are the individual effects of each allele at loci x
This reliance on these genomic and computa- and y, is the deviation that is due to epistasis.
tional technologies and datasets that can be trans- Systems biology, however, examines the sum of
muted across species has broadened significantly all epistatic relationships and hopes to uncover
the applicability of systems approaches to very the hierarchy. This, indeed, has been the direc-
complex systems such as human medicine. tion of this line of genetic research. Tong et al.
Other thinkers have expounded on the new (2004) crossed mutations in 132 “query” genes
possibilities in integrating mathematics with into a set of 4700 viable yeast gene deletion
biology. In an excellent essay, Joel E. Cohen mutants to develop a genetic interaction map
(2004) noted that “mathematics is not only biolo- containing more than 4000 functional gene
gy’s next microscope, but in fact is better”. He interactions. Classical genetics converges on
observed that, in biology, enormous complexity systems biology.
of up to 100 million species is built on just a few Kitano (2007) noted the importance of con-
basic elements of carbon, nitrogen, hydrogen trol theory in describing biological systems, and
and oxygen. By contrast, the entire periodic table described the primacy of “robustness” in the
generates only several thousand kinds of miner- design of biological systems. He differentiated
als in the earth’s crust. Thus the entire basis of robustness from homeostasis, in that homeosta-
biology is a complexity that produces ensemble sis seeks to return the system to the original state,
or emergent properties of much greater func- whereas robustness will accommodate migra-
tion than the component parts. Cohen argued tion to another state to achieve survivability. One
that mathematics can also benefit from attacking characteristic of evolvable systems described by
biological problems as it did in working through the Highly Optimized Theory (HOT) states that
problems in physics. Calculus was developed such systems are robust against common per-
in part to help solve the problems of celestial turbations, but are fragile against unusual ones
motion and of optics. Similarly, the multilayered (Carlson and Doyle, 2000). A common example
complexity, interlocking control loops, distrib- is the World Wide Web, which, despite being
uted switch mechanisms and the differential use robust because of its high interconnectivity, has
of the same components over developmental been brought down by specific attacks at hubs
time challenges mathematical and computational of activity. Thus systems robustness is a mat-
solutions. It is likely that new mathematics will ter of “trade-offs.” Mathematical descriptions of
be required to deal with these ensemble proper- robustness have been attempted.
ties and with the heterogeneity of the biological Kitano (2007) provides a representation of
input that feeds into the organismic output. robustness in the following equation, but also
acknowledges that new mathematics may be the construction of a complete functional biolog-
necessary to accommodate these systems con- ical network map for Halobacterium salinarum,
cepts in biology: an Archaea species that thrives in conditions of
high salinity. The final network map describes
“Robustness (R) of the system (s) with regard to
the regulatory functional relationships among
function (a) against a set of perturbations (P):
80% of its genes. The predictive power of
this model was evident in its ability to predict
(1.2)
the transcriptional responses to challenge with
novel environmental conditions or disruption of
The function is the probability for perturbation transcription factors. The predictive capability
‘p’ to take place. P is the entire perturbation space, of this genome-wide, whole-organism predic-
and D (p) is an evaluation function under perturba-
tion (p).” tive model was significant. In order to achieve
this, Bonneau and colleagues accomplished the
following in order to achieve their goal:
Experimental strategies in
1. The 2.6 Mb Halobacterium salinarum genome
systems biology was sequenced and functions were assigned
to each gene using protein sequence and
Systems approaches are characterized by sev- structural similarities (know all the components).
eral key attributes: 2. Cells were perturbed by varying
concentrations of environmental factors
1. The measurement of quantitative and
and / or gene knockouts (perturbation
comprehensive data of an experimental
analysis).
system.
3. The transcriptional changes of all genes
2. Assessment of the relationships between the
using microarrays were determined after
component parts.
each perturbation (genomic readout for
3. Perturbation of the system to detect response
perturbation analysis).
dynamics.
4. Diverse data (mRNA levels, evolutionary
4. Intersection of orthogonal data to arrive at
conservation in protein structure, metabolic
higher-order logic. (Orthogonal data are
pathways, and cis-regulatory motifs) were
defined as datasets derived from different
integrated to identify subsets of genes that
systems, perhaps addressing the same
are co-regulated in certain environment (data
question in which the intersection of the two
integration).
datasets can further resolve a problem: for
5. A dynamic network model was constructed
example, the set of genes with binding sites
for the of influence environmental and
of a transcription factor and the set of genes
transcription factor changes on the expression
that are expressed with overexpression of
of co-regulated genes (model building).
the same transcription factor [see Chapter 4]).
6. The resulting network was explored
5. Derivation of a model of the system that can
using software visualization tools within
be mathematical or qualitative.
an integrator that enables software
6. Correct prediction of output based on the
interoperability and database integration.
model.
This allowed for manual exploration and
The most complete analyses that engage all generation of hypotheses used to plan
these attributes have been made in lower organ- additional iterations of the systems analysis
isms. Bonneau and colleagues (2007) reported (model testing).
Similar strategies have been applied to the 2001). Together with the advent of expression
eukaryotic model system, yeast, with less pre- arrays in 1996 (Shalon, 1996) and their stable
dictive success (Luscombe, 2004; Tong, 2004; Yu, use by 2000, these technologies launched the
2008). Nevertheless, the strategy still requires next phase of growth for systems approaches
the integration of heterogeneous datasets, such to complex organisms like mammals. Network
as transcription factor binding sites, transcrip- analyses have been conducted primarily where
tional profiles and protein–protein interactions the system is cell-based, such as immunol-
(Fig. 1.1). ogy (Kitano, 2006) or cancer (Segal et al., 2005),
or where the tissue is homogeneous such as
the heart (Olson, 2006) or liver (Schadt, 2008).
Interestingly, computer scientists have looked
Systems biomedicine to the natural immune system to develop analo-
gous artificial immune systems for computer
Systems biomedicine is the analysis of medi- system security (Forrest and Beauchemin, 2007).
cal problems using systems approaches; there- There is much to be learned from biological sys-
fore pertinence to the human condition is a tems that have had the benefit of more than a
prerequisite. Given the complexity of mamma- billion years of evolutionary history.
lian systems, are we ready to study the ensemble The experimental systems approaches to
properties of the human model, and are we suf- studying a human problem will, by necessity,
ficiently clever to use these approaches to under- be different and potentially less complete than
stand and to treat human disease? Before 2001, those appropriate for attacking a question using
perhaps, it would have been difficult to answer prokaryotes. Such reconstruction of a regulatory
affirmatively. If access to the complete human network has been difficult in higher organisms,
genome is a prerequisite for a systems analysis, owing to the dramatically increased complex-
only after the sequencing of the human genome ity of the contributing subsystems. Thus the
could this goal be conceived (Lander et al., possible solution space is orders of magnitude
Expression
Protein
profiles after
interaction map
perturbation
Literature
Complete annotation of
genome protein and gene
sequence interactions
Cis-regulatory Metabolic
Network model
elements pathways
Evolutionary Genotyping
evidence associations
Figure 1.1 Input Information that can be Used to Construct a Network Model.
greater than that for lower organisms. Gene following 6 stimulation conditions assessing 6
numbers increase in higher eukaryotes, but this signaling molecules. These states could sepa-
is not the confounding factor: splice variants, rate acute myeloid leukemia cells into signaling
transcription factors binding at great distances classes that corresponded to cytogenetic and
from the transcriptional start sites, gene dupli- clinical parameters (Irish, 2004).
cation, post-transcriptional regulation by micro- It has been said that biology asks six kinds
RNAs and other non-coding RNA species, and of question (Cohen, 2004): How is it built? How
complex post-translational modifications that does it work? How did it begin? What is it for?
change binding affinities all radically augment The remaining two questions are more in the
the complexity of the components. domain of medicine: What goes wrong? How
Despite these challenges, network models of is it fixed? So, systems biomedicine focuses, not
subsystems have been described, for example only on human biology, but also human dis-
for the class of receptor tyrosine kinases (RTKs) ease. Efforts to examine perturbations in gene
(Amit et al., 2007a, 2007b; Katz, 2007). In these and protein networks for clues to disease etiol-
analyses, signaling hubs for the RTKs, such as ogy have been pursued and will be described in
RAF and the phosphoinositide 3 kinase PI3K– subsequent chapters in this book. Most efforts
AKT nodes, are noted to be frequent points of are in the bench-to-bedside direction, but one
attack by oncogenic viruses, in addition to being approach that starts commonly from the patient
sites of de-novo mutations in primary cancers. and is validated at the bench is in human and
Such hubs, independently identified by both population genetics of disease genes.
viruses and cancer mutations, also are effective Human variations in the form of single nucle-
targets for anticancer therapeutics. otide polymorphisms (SNPs) are used to identify
Exploiting kinase networks, Sachs et al. genetic loci statistically associated with disease
(2005) pursued an interesting alternative strat- when compared with control populations. When
egy. Using multicolor / multiparameter flow assessed on a genome-wide basis, this has been
cytometry in which up to 11 different features a powerful, unbiased means of uncovering dis-
can be determined when labeled with different ease-associated genes. When expression arrays
fluorophores, they quantitatively assessed the are coupled with genetic markers, expression
combinatorial presence of specific phosphor- quantitative trait loci (eQTL) can be assigned.
proteins indicative of activated kinases. Because In eQTL analysis, each transcript on the array is
flow cytometry assesses the biochemical state of considered to be a quantitative phenotype and
individual cells, a large number of observations is correlated with the SNP configuration at each
can be accumulated that would otherwise be an locus in the genome (Cheung et al., 2005; Sieberts
average of the population. In this manner, Sachs and Schadt, 2007). cis-eQTL represent those SNPs
and colleagues were able to construct a Bayesian adjacent to the measured gene of which the con-
network from these data. Bayesian network figuration is correlated with transcript levels,
models disclose the dependent effect of each whereas trans-eQTLs are those associated with
biomolecule on the others, and therefore can SNPs that are distant from the transcribed gene.
infer causal relationships. Examining signaling eQTLs in humans have been used as proof of
in T cells, they could construct a network map the genetic basis of gene expression in humans
that faithfully portrayed known and experimen- (Cheung et al., 2008; Spielman et al., 2007). When
tally validated kinase–substrate relationships. viewed on a genome-wide basis, a transcriptional
In a similar fashion, they mapped the signaling network of regulatory “influence” can be dis-
profiles of acute myeloid leukemia cells after cerned by statistical association between individ-
cytokine challenge and found 36 node states, ual SNPs and expression of genes anywhere in
the genome. Schadt and his colleagues at Rosetta REV-ERB accumulates to threshold levels,
Pharmaceuticals have shown that combining it represses BMAL1/CLOCK. This secondary
genotypic data and expression data can increase regulatory loop is not essential for the establish-
precision of the discovery for disease-associated ment of the circadian cycle, but it appears to be
genes (Drake et al., 2005; Zhu et al., 2007). involved in stabilizing the regulatory frame-
work. The oscillator function can be explained
by a time delay in PER/CRY feedback inhibi-
Circadian cycles as a tion of BMAL/CLOCK establishing a composite
relevant model for systems negative network motif with asymmetric tim-
ing. This oscillator is also affected by enzyme-
biomedicine families such as casein kinase 1 (CSNK1 and
CSNK1) that regulate the degradation of criti-
An excellent example of a systems model cal components like the PER protein (Fig. 1.2).
that has medical importance is that of oscil- (Takahashi et al., 2008).
lators as regulators of the circadian rhythm. Peripheral tissues also exhibit autonomous
Oscillators are machines that cycle functions circadian rhythms but are subservient to and
over time and are characterized by an auto- are entrained by the SCN. The SCN coordinates
matic periodicity (see Chapter 4), and the best the peripheral clocks through humoral and neu-
examples of biological oscillators are found in ral signals that are not well understood, and
studies of circadian rhythm. The guiding motif by indirect means such as body temperature,
for all living creatures is the ability to replicate, wakefulness and food intake. Thus the entire
which imparts a cycling of functions. Over evo- circadian system is a hierarchy of subnetworks
lutionary time, there appeared to be an adaptive that extend from the molecular and biochemical
advantage to entrain such physiologic processes level to the physiological level.
to an external clock defined by the day–night Components of the circadian clock are deeply
cycle. In order to do this, most organisms have involved in human physiology and disease. The
found biochemical mechanisms to maintain this most obvious association is with sleep disor-
cycling, and mechanisms to sense the environ- ders. Familial advanced sleep-phase syndrome
ment in order to modulate this periodicity. (FASPS) is an autosomal dominant circadian
The master circadian regulator in mammals rhythm disorder characterized by an abnormal
is in the suprachiasmatic nucleus (SCN) in the phasing of the circadian cycle relative to the
brain. The molecular mechanism underpinning desired sleep–wake schedule. Here sleep onset
this oscillator has been elucidated. The basic and awakening times are 3–4 hours ahead of the
helix–loop–helix containing transcription factor desired times. Through linkage analysis, indi-
CLOCK interacts with BMAL1 to activate tran- viduals with the syndrome were found to har-
scription of the Per and Cry genes. The Period bor a missense mutation, S662G, in the human
(PER) and Cryptochrome (CRY) protein prod- PER2 gene. This S662G mutation disrupts a
ucts heterodimerize and undergo negative feed- phosphorylation site within a casein kinase 1
back to inhibit their own transcription, and that (CSNK1)-binding domain of PER2, resulting
of BMAL1. The PER–CRY repressor complex in the increased turnover of nuclear PER2. As
is degraded during the night, and Clock-Bmal1 evidence that FASPS has heterogeneous genetic
are de-repressed and can then induce transcrip- origins, a mutation in a casein kinase isoform,
tion. There is a secondary feedback loop that CSNK1, was also found in FASPS.
involves the induction of a nuclear hormone Such sleep disorders are rare; however,
receptor, REV-ERB, by BMAL1/CLOCK. When there is the cumulative evidence that molecular
Protein PER–CRY
CSNK1ε/δ
degradation
FBXL3
PER CRY
RORα BMAL–CLOCK
Clock-
BMAL1 CLOCK
controlled
genes
REV-ERBα
REV-ERBα
Figure 1.2 Schematic Representation of the Control Network for Circadian Cycling.
The core organization of the regulatory module is comprised of two interlocking negative-feedback network motifs with
asymmetric timing. BMAL1, brain and muscle Arnt (aryl hydrocarbon receptor nuclear translocator)-like protein-1; CLOCK,
Circadian Locomotor Output Cycles Kaput gene; CSNK1/, casein kinase 1 and ; CRY, Cryptochrome; FBXL3, an E3
ubiquitin ligase; PER, Period; REV-ERB: a retinoid-acid-related orphan nuclear hormone receptor; ROR, a retinoid-acid-
related orphan nuclear hormone receptor.
components of the circadian oscillator may be body weight and sensitivity to insulin shock.
involved in many common disorders. Gene pro- Fibroblasts from these Bmal1 knockout mice
filing experiments demonstrated that up to 10% also cannot undergo adipocyte differentiation
of the testable transcriptome shows circadian (Shimba et al., 2005). Clinically, a link between
periodicity, and that the attributes of these clock- circadian cycles and metabolism has been
regulated genes are highly enriched for metabolic observed. Epidemiologic studies in shift work-
functions. Recall that the nuclear hormone recep- ers have shown an increase in body mass index,
tors, ROR and REV-ERB, are integral parts of and in the rates of incidence of metabolic disor-
the oscillator loop. Extending this analysis fur- ders and cardiovascular events (Ellingsen et al.,
ther, Yang and colleagues (2006) examined the 2007). It is also well understood that the specific
detailed gene expression the 49 nuclear receptors sensitivity to exogenous insulin exhibited by dia-
in mice, and found that 28 display tissue-specific betic patients changes over the time of day. Thus
circadian rhythms. Given the function of nuclear the circadian clock mechanisms are inextricably
receptors in metabolic regulation, their circadian linked to metabolic functions, and may repre-
control provides one explanation for the diurnal sent an adaptive evolutionary response to maxi-
behavior of glucose and lipid metabolism. mizing energy utilization that is dependent on a
Studies in animal models also continue to consistent environmental change—the planetary
uncover associations between clock genes and reality of the day / night cycle (Green et al., 2008).
metabolic phenotypes: homozygous Clock- An intriguing side observation that now has
mutant mice are hyperphagic, obese and exhibit significant ramifications for cancer therapeutics
a metabolic syndrome with hyperlipidemia, fatty is that liver detoxifying genes also show signifi-
liver, high circulating glucose concentrations and cant circadian oscillations and have been shown
low circulating insulin concentrations (Turek to be regulated by clock mechanisms. Doses of
et al., 2005). Bmal1/ knockout mice not only the chemotherapeutic agent, cyclophosphamide,
have abnormal sleep patterns, but also show low given at different times of the circadian cycle can
result in differences in mortality rates—from 20% experimental access. Clearly, the human genome
to 100% (Gorbacheva et al., 2005). Exploring this and proteome is more complex than those of
phenomenon further, the investigators found that yeast and bacteria, and human genetic studies
Clock and Bmal1 knockout mice are sensitive to are more complex than those in mice. Moreover,
the toxic effects of cyclophosphamide, but Cry1 the complexity of a multicellular and multi-
and Cry2 double-knockout mice are resistant. organ system has yet to be configured into the
This resistance was not caused by pharmacoki- equation. To date, the comparative extent of that
netic differences, but appeared to be correlated complexity remains not quite known; therefore,
with cellular insensitivity of B lymphocytes to how much more data and how much more com-
the lymphotoxic effects of this drug. These exper- puting will be necessary to achieve the same
iments validate the clinical observations that tim- coverage as that described for Halobacterium
ing of chemotherapeutic administration has an salinarum is unclear, but will undoubtedly be
effect on drug toxicity and drug effectiveness more than the ratio of the size of our genome to
(reviewed by Takahashi et al., 2008). that of this microbe. However, the approaches
The growing body of knowledge of the mecha- and the opportunities are the same.
nisms around circadian clocks and their impact Of course, in the final analysis, systems bio-
on health has provided opportunities for the medicine, by directly benefiting human health,
development of drugs targeting these molecules. will be a significant endeavor. So any increment
Many of the clock-associated genes are amenable in improvement in prediction will help medicine
to the action of drugs or represent biochemical and benefit society. The challenges, however,
classes amenable to small-molecule modulation: are logistical, computational and organizational.
the melatonin receptors are G-protein-coupled Logistical because first, for obvious ethical rea-
receptors; GSK3 is a kinase that modifies PER, sons, experimentation in human systems is
and REV-ERB casein kinase 1 is another class slower and more ponderous; secondly, human
of kinases; REV-ERB and ROR are nuclear hor- variation will make initial estimates less general-
mone receptors (the ligand for REV-ERB has izable; and thirdly, the further division into organ
been identified as heme). All these targets have systems linked by circulation and endocrine fac-
candidate small-molecule modifiers. This has led tors will increase the number of studies needed
companies to explore the use of cell-based screens in order to complete the human organism. The
to identify molecules that would disrupt or alter computational challenges have been alluded to,
the circadian clock. Cell systems with luciferase and are most critical: massive amounts of data
reporter genes controlled by clock-dependent reg- requiring integration and iterative analysis of
ulatory elements can be used to screen libraries of high computational complexity. The new tech-
small molecules. The readout would be disruption nologies in sequencing, genotyping, proteomics
of the periodicity (reviewed by Liu et al., 2007). and imaging are generating a hyper-exponen-
Thus, starting from a simple oscillator, explana- tial growth in data acquisition that is quickly
tions of human physiology and identification of outstripping the capabilities of most biological
targets for therapy can be explored. laboratories and departments. The physical sci-
ences have pioneered the use of supercomputers
with the capability of handling this challenge.
Conclusion However, the porting of the all biological, genetic
and genomic algorithms to these new plat-
How is systems biomedicine different from forms and their continued development will be
other forms of systems studies? In my opin- a prodigious task. Lastly, the simple fact is that
ion, the differences are only ones of scale and our data standards do not routinely allow for
cross-platform comparisons. Manual curation is platforms is too often bypassed as lacking sci-
still required for most high-level systems integra- entific content. By discounting participation in
tion. There is a need for integration of heteroge- collaborative projects and focusing exclusively
neous data (e.g. protein–protein interaction, RNA on individual effort, University promotion proc-
expression information, biochemical pathways, esses historically encourage faculty insularity.
genomic data and literature-based connections) Graduate student training, restrained by clas-
and for visualization tools that will enable the sical departmental boundaries and focused on
presentation of large-scale data that are interpret- individual faculty projects, is not responsive to
able to bench biologists. the educational requirements for success in inte-
Finally, the organizational challenges, although grative and systems biology. Systems biology is
man- made and therefore surmountable by man, deeply cross-disciplinary.
are also daunting (Liu et al., 2005). These organi- Daunting as these challenges are, the stakes
zational challenges are rooted in the sometimes are high. I believe that systems approaches in
contradictory requirements of systems biology biology will become as common as molecular
research and the operational intentions of our technologies are in current biological investi-
academic and funding institutions. gations. Molecular biology, which was a new
In systems research, scientists with very dif- creature in the 1970s and early 1980s and which
ferent skills (biology, mathematics, engineering, spawned biotechnology companies and insti-
medicine) must be working closely together tutes and departments with “molecular biol-
and have proximity with one another in what ogy” in their title, is now commonplace and
might almost be scientific collectives (Liu, integrated into the fabric of biological teachings.
2009). Traditionally, bioinformatics resided in Current medical investigations are all molecu-
a computer science or biostatistics department, lar medicine. The same will be true of systems
biology in a biochemistry department and a approaches.
genomics center that was functionally disso- Systems Biomedicine, indeed, is here to stay.
ciated from the previous two. However, the
scale of this interaction requires coordinated
resources from the funding agencies, much References
akin to the supercomputing program of the US Amit, I., Citri, A., Shay, T., et al., 2007a. A module of nega-
National Science Foundation. This unfortunate tive feedback regulators defines growth factor signaling.
disconnect would benefit from some concep- Nat. Genet. 39, 503–512.
Amit, I., Wides, R., Yarden, Y., 2007b. Evolvable signaling
tual realignment. Regarding data presentation, networks of receptor tyrosine kinases: relevance of rob-
there is a need to provide more natural inter- ustness to malignancy and to cancer therapy. Mol. Syst.
faces between humans and computers to serv- Biol. 3, 151.
ice the non-expert user. There will be a demand Bonneau, R., Facciotti, M.T., Reiss, D.J., et al., 2007. A predic-
tive model for transcriptional control of physiology in a
for simplified interfaces specifically designed
free living cell. Cell 131, 1354–1365.
for biologists. This does not detract from the Carlson, J.M., Doyle, J., 2000. Highly optimized tolerance:
important need to train the next generation of robustness and design in complex systems. Phys. Rev.
biologists who are mathematically and compu- Lett. 84, 2529–2532.
tationally literate, and the next generation of Cheung, V.G., Spielman, R.S., Ewens, K.G., Weber, T.M.,
Morley, M., Burdick, J.T., 2005. Mapping determinants
mathematicians, computer scientists and engi-
of human gene expression by regional and genome-
neers who are steeped in the nuances of biology. wide association. Nature 437, 1365–1369.
Grants management is often at odds with col- Cheung, V.G., Bruzel, A., Burdick, J.T., Morley, M., Devlin,
lective efforts. Funding for critical technology J.L., Spielman, R.S., 2008. Monozygotic twins reveal
germline contribution to allelic expression differences. Olson, E.N., 2006. Gene regulatory networks in the evo-
Am. J. Hum. Genet. 82, 1357–1360. lution and development of the heart. Science 313,
Cohen, J.E., 2004. Mathematics is biology’s next microscope, 1922–1927.
only better; biology is mathematics’ next physics, only Phillips, P.C., 2008. Epistasis—the essential role of gene inter-
better. PLoS Biol. 2, e439. actions in the structure and evolution of genetic systems.
Drake, T.A., Schadt, E.E., Davis, R.C., Lusis, A.J., 2005. Nat. Rev. Genet. 9, 855–867.
Integrating genetic and gene expression data to study the Sachs, K., Perez, O., Pe’er, D., Lauffenburger, D.A., Nolan,
metabolic syndrome and diabetes in mice. Am. J. Ther. G.P., 2005. Causal protein-signaling networks derived
12, 503–511. from multiparameter single-cell data. Science 308,
Ellingsen, T., Bener, A., Gehani, A.A., 2007. Study of shift 523–529.
work and risk of coronary events. J. R. Soc. Health 127, Schadt, E.E., Molony, C., Chudin, E., et al., 2008. Mapping
265–267. the genetic architecture of gene expression in human
Forrest, S., Beauchemin, C., 2007. Computer immunology. liver. PLoS Biol. 6, e107.
Immunolog. Rev. 216, 176–197. Segal, E., Friedman, N., Kaminski, N., Regev, A., Koller, D.,
Green, C.B., Takahashi, J.S., Bass, J., 2008. The meter of 2005. From signatures to models: understanding cancer
metabolism. Cell 134, 728–742. using microarrays. Nat. Genet. 37 (Suppl), S38–S45.
Gorbacheva, V.Y., Kondratov, R.V., Zhang, R., et al., 2005. Shalon, D., Smith, S.J., Brown, P.O., 1996. A DNA microar-
Circadian sensitivity to the chemotherapeutic agent ray system for analyzing complex DNA samples using
cyclophosphamide depends on the functional status two-color fluorescent probe hybridization. Genome Res.
of the CLOCK/BMAL1 transactivation complex. Proc. 6, 639–645.
Natl. Acad. Sci. USA 102, 3407–3412. Shimba, S., Ishii, N., Ohta, Y., et al., 2005. Brain and muscle
Irish, J.M., Hovland, R., Krutzik, P.O., et al., 2004. Single cell Arnt-like protein-1 (BMAL1), a component of the molec-
profiling of potentiated phospho-protein networks in ular clock, regulates adipogenesis. Proc. Natl. Acad. Sci.
cancer cells. Cell 118, 217–228. USA 102, 12071–12076.
Katz, M., Amit, I., Citri, A., et al., 2007. A reciprocal tensin- Sieberts, S.K., Schadt, E.E., 2007. Moving toward a system
3-cten switch mediates EGF-driven mammary cell genetics view of disease. Mamm. Genome 18, 389–401.
migration. Nat. Cell Biol. 9, 961–969. Spielman, R.S., Bastone, L.A., Burdick, J.T., Morley, M.,
Kitano, H., 2006. The B-cell interactome. Available from http:// Ewens, W.J., Cheung, V.G., 2007. Common genetic vari-
amdec-bioinfo.cu-genome.org/html/BCellInteractome. ants account for differences in gene expression among
html#Publication ethnic groups. Nat. Genet. 39, 226–231.
Kitano, H., 2007. Towards a theory of biological robustness. Takahashi, J.S., Hong, H.K., Ko, C.H., McDearmon, E.L.,
Mol. Syst. Biol. 3, 137. 2008. The genetics of mammalian circadian order and
Kitano, H., Oda, K., 2006. Robustness trade-offs and host- disorder: implications for physiology and disease. Nat.
microbial symbiosis in the immune system. Mol. Syst. Rev. Genet. 9, 764–775.
Biol. 2, 2006–2022. Tong, A.H., Lesage, G., Bader, G.D., et al., 2004. Global map-
Lander, E.S., Linton, L.M., Birren, B., et al., 2001. For the ping of the yeast genetic interaction network. Science
international human genome sequencing consortium. 303, 808–813.
Initial sequencing and analysis of the human genome. Turek, F.W., Joshu, C., Kohsaka, A., et al., 2005. Obesity and
Nature 409, 860–921. metabolic syndrome in circadian Clock mutant mice.
Liu, E.T., 2005. Systems biology, integrative biology, predic- Science 308, 1043–1045.
tive biology. Cell 121, 505–506. Yang, X., Downes, M., Yu, R.T., et al., 2006. Nuclear receptor
Liu, E.T., 2009. Integrative biology—a strategy for systems expression links the circadian clock to metabolism. Cell
biomedicine. Nat. Rev. Genet. 10, 64–68. 126, 801–810.
Liu, A.C., Lewis, W.G., Kay, S.A., 2007. Mammalian circa- Yu, H., Braun, P., Yildirim, M.A., Lemmens, I., et al., 2008.
dian signaling networks and therapeutic targets. Nat. High-quality binary protein interaction map of the yeast
Chem. Biol. 3, 630–639. interactome network. Science 322, 104–110.
Luscombe, N.M., Babu, M.M., Yu, H., Snyder, M., Zhu, J., Wiener, M.C., Zhang, C., et al., 2007. Increasing the
Teichmann, S.A., Gerstein, M., 2004. Genomic analysis power to detect causal associations by combining geno-
of regulatory network dynamics reveals large topologi- typic and expression data in segregating populations.
cal changes. Nature 431, 308–312. PLoS Comput. Biol. 3, e69.
2
Genomic Technologies for Systems
Biology
Edison T. Liu1, Sanket Goel1, Kartiki Desai1 and Mathijs
Voorhoeve2
1
Genome Institute of Singapore, Singapore
2
Duke-National University of Singapore Graduate Medical School, Singapore
o u t l i n e
Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.
ebookfinal.com