Systems biology: From the cell to the brain

SITABHRA SINHA1,∗, T JESAN2 and NIVEDITA CHATTERJEE3

1The Institute of Mathematical Sciences, CIT Campus, Taramani, Chennai 600 113, India.2Health Physics Division, Bhabha Atomic Research Centre, Kalpakkam 603 201, India.

3Vision Research Foundation, Sankara Nethralaya, 18 College Road, Chennai 600 006, India.∗e-mail: sitabhra@imsc.res.in

With the completion of human genome mapping, the focus of scientists seeking to explain thebiological complexity of living systems is shifting from analyzing the individual components (suchas a particular gene or biochemical reaction) to understanding the set of interactions amongst thelarge number of components that results in the different functions of the organism. To this end,the area of systems biology attempts to achieve a ‘systems-level’ description of biology by focusingon the network of interactions instead of the characteristics of its isolated parts. In this article, webriefly describe some of the emerging themes of research in ‘network’ biology, looking at dynamicalprocesses occurring at the two different length scales of within the cell and between cells, viz., theinter-cellular signaling network and the nervous system. We show that focusing on the systems-level aspects of these problems allows one to observe surprising and illuminating common themesamongst them.

1. Introduction

The complete mapping of human and othergenomes has indicated that the remarkable com-plexity of living organisms is expressed by lessthan 30,000 protein-coding genes [1]. Thus, theobserved complexity arises not so much from therelatively few components (in this case, genes), asfrom the large set of mutual interactions that theyare capable of generating. In a similar fashion, the302 neurons of the nematode C. elegans enablesit to survive in the wild, much more success-fully than complicated, state-of-the-art robots. It isnot the number of neurons that is the crucialfactor here, but rather their interactions and theresulting repertoire of dynamical responses thatunderlie the survival success of living organismsin a hostile (and often unpredictable) environ-ment. The focus of research in biology is there-fore gradually shifting towards understanding howinteractions between components, be they genes,proteins, cells or organisms, add a qualitatively

new layer of complexity to the biological world.This is the domain of systems biology which aimsto understand organisms as an integrated whole ofinteracting genetic, protein and biochemical reac-tion networks, rather than focusing on the indivi-dual components in isolation [2–4]. While the termitself is of recent coinage [2], the field has hadseveral antecedents, most notably, cybernetics, aspioneered by Norbert Wiener [5] and W RossAshby (who indeed can be considered to be oneof the founding figures of systems neuroscience [6]along with Warren McCulloch [7]) and the gen-eral systems theory of Ludwig von Bertalanffy,which have inspired other fields in addition tobiology.

The recent surge of interest in systems thinkingin biology has been fuelled by the fortunatecoincidence in the advent of high throughputexperimental techniques (such as DNA and pro-tein microarrays) allowing multiplex assays, alongwith the almost simultaneous development ofaffordable high-performance computing which has

Keywords. Systems biology; intracellular signalling; kinase cascades; Caenorhabditis elegans; nematode neuronal network.

199

200 SITABHRA SINHA et al

Figure 1. Dynamical processes mediated through complexnetworks occurring at different length scales in the biologicalworld, from the protein contact networks at the molecularlevel to the network of trophic relations spanning several tensor hundreds of kilometers. In between these two extremes, wesee intracellular networks (e.g., that involved in intracellu-lar signaling), intercellular networks (e.g., neuronal networksin the brain) and networks among individuals comprising apopulation (e.g., the social contact network through whichvarious epidemic diseases spread).

made possible automated analysis of huge volumesof experimental data and the simulation of verylarge complex systems. Another possible stimu-lant has been the parallel growth of the theoryof complex networks (comprising many nodes thatare connected by links arranged according to somenon-trivial topology) from 1998 onwards, whichhas provided a rigorous theoretical framework foranalysis of large-scale networks, ranging from thegene interaction network to the Internet [8,9].Indeed, reconstructing and analyzing biologicalnetworks, be they of genes, proteins or cells, isat the heart of systems biology. The role of such‘network biology’ is to elucidate the processes bywhich complex behavior can arise in a system com-prising mutually interacting components. Whilesuch emergent behavior at the systems level is notunique to biology [10], to explain properties ofliving systems, such as their robustness to environ-mental perturbations and evolutionary adaptabi-lity, as the outcome of the topological structure ofthe networks and the resulting dynamics, is a chal-lenge of a different order. As networks appear atall scales in biology, from the intracellular to theecological, one of the central questions is whetherthe same general principles of network functioncan apply to very different spatial and temporalscales in biology [11] (figure 1). In this article, welook at a few examples of how using a networkapproach to study systems at different scales canreveal surprising insights.

2. Intracellular networks

Inside the biological cell, a variety of networks con-trol the multitude of dynamical processes respon-sible for its proper function [12]. These include

genetic regulatory networks [13,14] by which genesregulate the expression of other genes through acti-vation or inhibition (i.e., by expressing proteinsthat act as promoters or suppressors of other genes)as seen, for example, in the pattern formation stepsthat occur during development from a fertilizedcell into an embryo [15], protein-protein interac-tion networks involving the physical associationof protein molecules that are vital for many bio-logical processes such as signal transduction [16],and, metabolic networks of bio-chemical reactions[17–19], that are responsible for breaking downorganic compounds to extract energy as well asthose which use energy to construct vital compo-nents of the cell such as amino acids and nucleicacids. While the glycolytic pathway that convertsglucose into pyruvate, the first significant portionof the metabolic network to be reconstructed, tookmany years to be elucidated, there are now experi-mental techniques such as the yeast two-hybridscreening method that test for physical interactionsbetween many pairs of proteins at a time, allowingrapid reconstruction of such networks.

One of the most intriguing cellular networksis the protein-protein interaction network that isresponsible for intracellular signaling, the mech-anism by which a cell responds to variousstimuli through an ordered sequence of biochemicalreactions. These reactions regulate processes vitalto the development and survival of the organism(e.g., differentiation, cell division, apoptosis, etc.)by transmitting information from receptors locatedat the cell surface (that receive external signals) tospecific intracellular targets in a series of enzyme-substrate reaction steps [20].

This dynamics is governed by a large num-ber of enzyme molecules, which act as the nodesof the signaling network. Kinases belong to themost commonly observed class of enzymes involvedin cell signaling and they activate target mole-cules (usually proteins, e.g., other types of kinasemolecules) by transferring phosphate groups fromenergy donor molecules (e.g., ATP) to the tar-get, the process being termed phosphorylation. Thesubsequent deactivation of the target molecules bydephosphorylation is mediated by correspondingphosphatases (figure 2).

In the post-genomic era, the focus of researchershas shifted from studying such molecules in iso-lation, to understanding how the set of interac-tions over the complex bio-molecular network canexplain the entire repertoire of cell behavior [21].An essential tool towards such an understand-ing is modeling the dynamics of reactions occur-ring in smaller subnetworks (modules) that act asdistinctly identifiable functional units performingspecific tasks [22]. The detailed knowledge of howindividual modules control signaling dynamics can

SYSTEMS BIOLOGY: FROM THE CELL TO THE BRAIN 201

Figure 2. Dynamics of kinase activation (phosphoryla-tion) and de-activation (dephosphorylation). The substratekinase (S) is activated by its corresponding enzyme (E),which is the kinase located immediately upstream in thecascade. The enzyme-substrate complex (ES) formation is areversible reaction with forward and reverse rates of kf andkr, respectively, The product (P) of the enzyme-substratereaction is the phosphorylated kinase S*, which is generatedby an irreversible reaction step from ES with the rate kp. Thedeactivation reaction from S* to S is mediated by the phos-phatase E′. The enzyme-substrate reaction dynamics can berepresented by the rate equations for the variation of con-centrations of S, ES and P as shown at the top of the figure.

help us in eventually building a coherent picture forthe functioning of the entire cellular network. Here,we focus on such a module, the three componentmitogen-activated protein kinase (MAPK) cascade(figure 3), which is seen in all eukaryotic cells andis involved in many functions, including cell cyclecontrol, stress response, etc., [23].

The single chain cascade, where an input (orsignal) to the MAPK kinase kinase (MAPKKK)is transmitted via MAPK kinase (MAPKK) tothe MAPK whose output is the activation oftranscription factors or other kinases, is well under-stood. However, in many situations, e.g., in thenetwork involved in processing stimulus receivedby the B-cell antigen receptor (a key player inhuman immune response), a branched variation ofthe basic MAPK module is observed. Here, a com-mon MAPKKK sends signals to two different typesof MAPKK and hence, to different MAPK mole-cules (figure 4, inset). We have analyzed a modelof this branched structure to investigate how theelements in the two branches affect each other,although there is no direct interaction betweenthem. We find that a novel mechanism of indi-rect regulation arises solely from enzyme reactiondynamics on the branched structure [24].

The time-evolution of the concentrations fordifferent molecules in the branched MAPK net-work is described by a set of coupled ordinarydifferential equations (ODEs), with each enzyme-substrate (ES) reaction having a reversible ES

Figure 3. Schematic view of the MAP-kinase signal-ing pathway comprising MAPKKK, MAPKK and MAPK.The initial activation of the cascade is due to the signalS, which initiates phosphorylation of MAPKKK, and thecorresponding response is measured in terms of the degree ofresultant activation in MAPK. The phosphorylated productsare denoted by ∗, while P’ase indicates the correspondingphosphatases. Note the dual phosphorylation of MAPKK &MAPK, unlike the single phosphorylation of MAPKKK.

Figure 4. The relative increase in the response of branchB as a function of signal strength, when the phosphorylationof MAPK in branch A is inhibited by an ATP blocker. Theincrease is measured as the ratio of steady-state concentra-tions for MAPKK ∗∗

B and MAPK ∗∗B on inhibiting branch A

to their concentrations under normal conditions. The insetshows a schematic diagram of the branched MAPK reactionnetwork.

complex formation step and an irreversible step ofproduct formation (figure 2). The 32 coupled ODEsin our model are explicitly integrated numericallywithout invoking the quasi-steady-state hypothesisunderlying Michaelis–Menten kinetics. The modelparameters, viz., the different reaction rates andinitial concentrations of the substrate molecules,are adapted from the Huang–Ferrell model [25].

202 SITABHRA SINHA et al

MAPKKK is activated by a signal S whose con-centration is varied in the physiologically plausiblerange of 10−12 to 10−8 moles.

The system is first observed under normalconditions and then perturbed by preventing theactivation of MAPKA. As a result of block-ing the phosphorylation of MAPK in branch A(through preventing the formation of MAPK ∗

A

product from the MAPKA-MAPKK ∗∗A complex,

which can be implemented in an experimentalsituation by using an ATP blocker), the concen-tration of free MAPKKA is significantly reduced.This is because, although MAPKA is not beingactivated, free MAPKA and its correspondingkinase MAPKK ∗∗

A are being steadily depletedfrom the system through ES complex formation.In the absence of product formation, the releaseof MAPKK ∗∗

A can only occur at the relativelyslow rate of the ES-complex unbinding process.As increasing numbers of MAPKK∗∗ molecules aretaken up into complexes with MAPK, there isless available free MAPKK for MAPKKK∗ to bindwith. As a result, there is more MAPKKK∗ avail-able for binding to MAPKKB. This results in sig-nificant amplification of the activity in B branch,with all its downstream reactions being upregu-lated, including an increase in the dual phosphory-lation of both MAPKKB and MAPKB (figure 4).This indirect regulation of activity between the twobranches implies backward propagation of infor-mation along the signaling pathway from MAPKA

to MAPKKK, in contrast to the normal forwarddirection of information flow from MAPKKK toMAPK.

Our results show that, in intracellular signal-ing networks there may be indirect regulation ofactivity between molecules in branched structures.Such long-range effects assume importance in lightof the current paradigm in reconstructing intra-cellular networks where, the observation of up- ordown-regulation of activity for a molecule as aresult of perturbing another molecule is thoughtto be indicative of the existence of a direct inter-action between them. As signaling networks areoften inferred on the basis of such observations, ourresults provide an indication of other effects thatcan explain such dynamical correlations betweenthe activities of different molecules.

3. Systems neuroscience

Going up the scale from cellular to multi-cellularsystems, we come across the systems biology ques-tions related to intercellular communication andhow such systems respond to events in the externalenvironment. Possibly, the most intriguing ques-tions in this domain have to do with the brain

and the nervous system, the area of systems neu-roscience. This field explores the neural basis ofcognition, as well as of motivational, sensory, andmotor processes. Systems neuroscience fills the gapbetween molecular & cellular approaches to thestudy of brain and the behavioral analysis of high-level mental functions. This definition is, however,nebulous as an enormous number of questions canbe considered to be under the umbrella of systemsneuroscience. For example, such questions mayconcern issues of how to identify the neuronal cor-relates of consciousness and how multiple sensorystimuli define perceptions of reality, to much moretangible issues of the projection field of neurons.In an effort to classify the large variety of prob-lems, Sejnowski and van Hemmen in a recent book[26] have divided them into questions about (i) theevolution of the brain, (ii) the organization of thecortex, (iii) computational ability of the brain, and(iv) the organization of cognitive systems. How-ever, these classes cannot always neatly pigeonholeall systems-related questions about the brain, e.g.,how do the computations taking place in the brainget translated into actions and decisions at the levelof the individual organism.

Given the complexity, inaccessibility, and het-erogeneity of the brain, systems neuroscience usestools which are interdisciplinary in nature. Whilecommon behavioral experiments (e.g., involvingthe water-maze test for memory) and neuropsy-chological tests (e.g., showing sensory cues forstudying synaesthesia) still form the basis formany investigations, technological improvementsare allowing freedom in what and how these ques-tions can be addressed. For example, new molec-ular and biophysical tools for the observation andmanipulation of neural circuits allow one to imagesubjects as they are performing specific tasks. Thefield uses a variety of imaging techniques (fMRI,sMRI, DTI, and EEG) as well as behavioral, psy-chological and computational methods. The aim isto use all these tools to have a better understandingof the integrated functioning of large-scale distri-buted brain networks and how disruptions in brainfunction & connectivity impact behavior. Thus,systems neuroscience addresses questions on bothnormal and abnormal functioning of the nervoussystem, and also has the potential to significantlyinfluence developments in the artificial intelligencecommunity.

Exploring cognitive ability and organization ofthe nervous system is obviously easier to studyin a system with limited number of neurons.Not only does it allow the advantage of work-ing with a numerically simpler system, but italso indicates what is the least number of neu-rons required for performing simple behavioralfunctions. The nervous system of the nematode

SYSTEMS BIOLOGY: FROM THE CELL TO THE BRAIN 203

Figure 5. Biological information processing systems canadopt one of two possible structural arrangements: (A) aseries of parallel pathways with relatively little communica-tion (cross-talk) between them, and (B) a densely connectednetwork. The squares indicate the input and output nodes ofthe system, while the circles indicate the intermediate nodesinvolved in signal processing.

C. elegans which has only 302 neurons allows fortractability, unique identification of neurons andavailability of a detailed physical connectivity mapderived from ultrastructural analysis with elec-tron microscope [27]. Computational tools can beused on such data for identifying patterns in thewiring of the C. elegans nervous system [28] aswell as mapping the functional connectivity in spe-cific behavioral circuits and delineating causal rela-tionships between neural activity and behavior.This allows us to ask questions such as why doesnature prefer to use networks rather than a par-allel set of dedicated stimulus-response pathwaysto process information (figure 5). This is an exam-ple of how using the systems methodology allowsone to see common themes in biological networksacross length scales, as simply by changing theidentity of the nodes one can ask the same ques-tion for the nervous system (nodes = neurons, links= excitatory and inhibitory synapses or gap junc-tions, stimulus-response path = reflex arc) and forthe intracellular signaling network considered ear-lier (nodes = kinases, links = phosphorylation anddephosphorylation reactions, stimulus-responsepath = reaction pathway).

The core neural circuits behind some simplebehaviors in the nematode have already beenexperimentally determined by selective and sys-tematic ablation [29]. On the basis of these func-tional circuit neurons, we can aim to identify allneurons involved in a circuit for a particular beha-vior, the inputs that drive activity in each neuron,how input signals received in the sensory neuronsare transformed as they travel through the circuitand finally how the pattern of neuronal depolari-zations and hyperpolarizations over the entire net-work translates into behavior.

The functional circuits are seen to function intandem rather than as isolated modules. To bespecific, if one considers the eight functional cir-cuits for (i) mechanosensation (touch sensitivity),(ii) chemosensation, (iii) thermotaxis, (iv) egglaying, (v) defecation, and, three types of locomo-tion, viz., when (vi) satiated (feeding), (vii) hungry(exploration) and (vii) during escape behavior(tap withdrawal), there are certain neurons whichbelong to more than one circuit. The interneu-rons commonly receive inputs from many modali-ties and are often multifunctional. However, thesensory neurons and sometimes even motorneu-rons may be dedicated for a particular function,as in the egg-laying circuit (occurring only in thehermaphrodite animal) which possess specializedmotor neurons [27,30].

In a recent work [31], we have performed anintegrated analysis of these functional circuits todiscover emergent patterns in the connectivityprofile of the C. elegans neural network. Usingnetwork core decomposition tools we observe cor-relations between a neuron having a critical func-tional role and it occupying a central position interms of network structure. This is evidence for astructural basis of the roles played by neurons inthe functional circuits. In order to identify the fac-tors which determine the connectivity constraintsin a neuronal network, we have looked at the modu-larity of the network with reference to position,function and neuron type. It appears that modulesdefined in terms of network connectivity do notnecessarily correspond to the ganglia defined interms of spatial location of the cell bodies. Thus,wiring economy and developmental constraints donot completely decide the connection structure ofthe network. However circuits sharing a large pro-portion of core neurons do show similarity in termsof their modular spectra which may be interpretedas a requirement to get input from these over-lapping circuits in any perceptual decision-makingprocess.

Our work also showed that while the network hasa preferred direction of information flow, it is nota simple unidirectional propagation from sensoryto motor neurons. The large number of recurrentconnections observed among interneurons sug-gests a hierarchical structure with a densely con-nected core comprising mostly interneurons andless densely connected periphery populated by sen-sory and motor neurons. The betweenness central-ity of the neurons increases with their core ordermembership, indicating that most of the short-est paths between pairs of neurons pass throughneurons belonging to the innermost core. Thiswould allow most of the information propagationto take place via a small select set which are sub-ject to a high level of feedback activity. Such a

204 SITABHRA SINHA et al

hierarchical network (having a densely connectedcore and an overall sparse structure) also preventsindiscriminate global activation of the nervoussystem while at the same time permitting highdensity of connections to allow for high commu-nication efficiency, so that information can prop-agate rapidly from sensory stimulation to motorreaction.

Based on anatomical, physiological and beha-vioral data, the simulation of the dynamics ofneural circuits has already been attempted in afew organisms. In C. elegans, a dynamic networksimulation of the nematode tap withdrawal circuitwas worked out by Niebur and Erdos [32] whileFerree and Lockery [33] demonstrated that simplecomputational rules can predict certain compo-nents of chemotactic behavior. In spite of the abil-ity to process a large variety of sensory inputs,the ultimate outputs of the C. elegans nervous sys-tem are simple. The organism performs a limitedset of motor programs on integration of sensoryinformation and there should be no indiscriminateactivation of either sensory or interneurons of unre-lated circuits. The challenge in front of systemsneuroscience is to explicate the strategies used bynetworks to achieve this goal. The results of thisinvestigation will be useful not only in the con-text of the nervous system but also for intracellularinformation processing networks such as those ofkinases.

References

Papers of particular interest have been highlightedas

∗∗ : of outstanding interest∗ : of special interest

[1] International Human Genome Sequencing Consortium2004 Finishing the euchromatic sequence of the humangenome; Nature 431 931–945.

[2] Kitano H 2000 Perspectives on systems biology; NewGeneration Computing 18 199–216.

This paper is possibly the first to use the term systemsbiology.

[3] Kitano H 2002 Computational systems biology; Nature420 206–210.

[4] Kitano H 2002 Systems biology: A brief overview; Sci-ence 295 1662–1664.

[5] Wiener N 1948 Cybernetics or control and communica-tion in the animal and the machine (Cambridge: MITPress).

[6] ∗Ashby W R 1952 Design for a brain (London:Chapman & Hall).

Long before the rise of modern complex systems the-ory, this book explored the possibility of explainingbasic brain mechanisms by using the tools of nonlineardynamics.

[7] McCulloch W S 1965 Embodiments of mind(Cambridge: MIT Press).

A collection of papers; includes the 1943 paper withWH Pitts that started the field of artificial neuralnetworks.

[8] ∗∗Strogatz S H 2001 Exploring complex networks;Nature 410 268–276.

Somewhat dated by now but still an excellent summaryof the developments in the field of complex networktheory in the late 1990s.

[9] Newman M E J, Barabasi A L and Watts D J 2006The structure and dynamics of networks (Princeton:Princeton Univ. Press).

[10] ∗Anderson P W 1972 More is different; Science 177393–396.

Looks into the question of why biology (and othersciences) is not just ‘applied quantum mechanics’although quantum rules do underlie the naturalworld.

[11] Sinha S 2009 From network structure to dynamics andback again: Relating dynamical stability and connec-tion topology in biological complex systems; in: Dyna-mics on and of complex networks (eds) Ganguly N,Deutsch A and Mukherjee A. (Boston: Birkhauser)[also at arxiv:0804.0977].

[12] Barabasi A L and Oltvai Z N 2004 Network biol-ogy: Understanding the cell’s functional organization;Nature Rev. Genetics 5 101–113.

[13] ∗∗Kauffman S A 1969 Metabolic stability and epige-nesis in randomly constructed genetic nets; J. Theor.Biol. 22 437–467.

A truly pioneering paper that looks at the genetic con-trol of cellular differentiation by studying random net-works of elements controlled by Boolean logic.

[14] Samal A and Jain S 2008 The regulatory network ofE. coli metabolism as a Boolean dynamical systemexhibits both homeostasis and flexibility of response;BMC Systems Biol. 2 21.

[15] ∗von Dassow G, Meir E, Munro E M and Odell G M2000 The segment polarity network is a robust devel-opmental module; Nature 406 188–192.

Suggests that the segment polarity genes in Drosophiliaform a robust subnetwork within the genetic regulatorynetwork.

[16] Tyson J J, Chen K C and Novak B 2001 Network dyna-mics and cell physiology; Nature Rev. Mol. Cell Biol. 2908–916.

[17] Jeong H, Tombor B, Albert R, Oltvai Z N and BarabasiA L 2000 The large-scale organization of metabolic net-works; Nature 407 651–654.

[18] ∗Wagner A and Fell D A 2001 The small world insidelarge metabolic networks; Proc. Roy. Soc. Lond. B 2681803–1810.

Suggests that the metabolic network of E. coli is a‘small-world’ network, which contains evidence aboutthe evolutionary history of metabolism.

[19] ∗Guimera R and Amaral L A N 2005 Functional car-tography of complex metabolic networks; Nature 433895–900.

Finds functional modules in metabolic networks oftwelve organisms and classifies the nodes into univer-sal roles according to their patterns of intra- and inter-module connections.

SYSTEMS BIOLOGY: FROM THE CELL TO THE BRAIN 205

[20] ∗∗Tyson J J, Chen K C and Novak B 2003 Sniffers,buzzers, toggles and blinkers: dynamics of regulatoryand signaling pathways in the cell; Curr. Opin. CellBiol. 15 221–231.

A masterly overview of the modeling approaches forsignaling interactions in the cell.

[21] ∗Weng G, Bhalla U S and Iyengar R 1999 Complexityin biological signaling systems; Science 284 92–96.

Discusses the origin of complex behavior in signalingnetworks.

[22] Alon U 2007 An introduction to systems biology: Designprinciples of biological circuits. (Boca Raton: CRCPress).

[23] Garrington T P and Johnson G L 1999 Organizationand regulation of mitogen-activated protein kinase sig-naling pathways; Curr. Opin. Cell Biol. 11 211–218.

[24] Jesan T, Sarma U, Saha B and Sinha S (in prepara-tion).

[25] Huang C Y and Ferrell J E 1996 Ultrasensitivity in themitogen-activated protein kinase cascade; Proc. Natl.Acad. Sci. USA 93 10078–10083.

[26] Leo van Hemmen J and Sejnowski T J (eds) 2006 23problems in systems neuroscience (New York: OxfordUniv. Press).

[27] ∗White J G, Southgate E, Thomson J N and Brenner S1986 The structure of the nervous system of the nema-tode C. elegans; Phil. Trans. Roy. Soc. Lond. B 3141–340.

The seminal paper on C. elegans somatic nervoussystem describing the complete network connectivitydiagram.

[28] Reigl M, Alon U and Chklovskii D B 2004 Search forcomputational modules in the C. elegans brain; BMCBiology 2 25.

[29] ∗∗Hobert O 2003 Behavioral plasticity in C. ele-gans: Paradigms, circuits, genes; J. Neurobiol. 54203–223.

Gives a brief but comprehensive coverage of the coreneurons in the most important identified functional cir-cuits in the worm; also outlines approaches to analyzethe role of genes in governing behavior.

[30] Desai C, Garriga G, McIntire S L and Horvitz H R1988 A genetic pathway for the development of theCaenorhabditis elegans HSN motor neurons; Nature336 638–646.

[31] Chatterjee N and Sinha S 2007 Understanding themind of a worm: Hierarchical network structure under-lying nervous system function in C. elegans; Prog.Brain Res. 168 145–153.

[32] Niebur E and Erdos P 1993 Theory of locomotion ofnematodes: control of the somatic motor neurons byinterneurons; Math. Biosci. 118 51–82.

[33] Ferree T C and Lockery S R 1999 Computational rulesfor chemotaxis in the nematode C. elegans; J. Compu-tational Neurosci. 6 263–277.