Theoretical aspects of Systems Biology

Abstract

The natural world consists of hierarchical levels of complexity that range from subatomic particles and molecules to ecosystems and beyond. This implies that, in order to explain the features and behavior of a whole system, a theory might be required that would operate at the corresponding hierarchical level, i.e. where self-organization processes take place. In the past, biological research has focused on questions that could be answered by a reductionist program of genetics. The organism (and its development) was considered an epiphenomenona of its genes. However, a profound rethinking of the biological paradigm is now underway and it is likely that such a process will lead to a conceptual revolution emerging from the ashes of reductionism. This revolution implies the search for general principles on which a cogent theory of biology might rely. Because much of the logic of living systems is located at higher levels, it is imperative to focus on them. Indeed, both evolution and physiology work on these levels. Thus, by no means Systems Biology could be considered a ‘simple’ ‘gradual’ extension of Molecular Biology.

Introduction

According to the paradigm inherited from Galileo and Newton, later philosophically theorized by Descartes, every phenomenon we observe can be ‘reduced’ to a collection of particles whose movement is governed by linear dynamics rules that drive the overall system toward a deterministic, predictable ‘fate’. This approach was proven to be mistaken, even for apparently ‘simple’ situations characterized by linear dynamics, like the ‘three body problem’, sharply addressed by Henri Poincaré (Barrow-Green, 1997). Reductionism hardly allows us to understand the world's complexity, as was recognized by modern physics at the beginning of the last century (Laughlin, 2005): complex systems exhibit properties and behavior that cannot be understood from laws governing the microscopic parts given that such systems cannot be easily ‘reduced’ or explained by simple deterministic rules (Anderson, 1972).

To date, however, the positivist theoretical framework has survived in Biology under the spoils of “genetic determinism”, which consider genes alone able to drive and determine the development as well as the characteristics of an organism. This is paradoxical when keeping in mind that “it seems odd […] that just when physics is moving away from mechanism, biology and psychology are moving closer to it. If this trend continues […] scientists will be regarding living and intelligent beings as mechanical, while they suppose that inanimate matter is too complex and subtle to fit into the limited categories of mechanism” (Bohm, 1969). In other words, Molecular Biology tries to explain the mysteries of the living being by exclusively considering it as a consequence of a linear translation of the ‘DNA code’. As originally formulated (Crick, 1970), the ‘central dogma’ posits that ‘information’ flows unidirectionally from DNA to proteins, and not the other way around. However, environmental factors do change the genome, by both genetic as well as epigenetic mechanisms (Goldenfeld and Woese, 2007), and many types of molecules participate in ‘information’ transfer from one molecule to another (Barnes and Dupré, 2008). Genomic functions are inherently interactive (isolated DNA is virtually inert) (Shapiro, 2009), and biological processes flow along complex circuits, involving RNA, proteins and context-dependent factors (extracellular matrix, stroma, chemical gradients, biophysical forces) within which vital processes occur (Keller, 2000). Indeed, no simple, one-to-one correspondence between genes and phenotypes can be made (Noble, 2008a, b). Therefore, “the collapse of the doctrine of one gene for one protein, and one direction of causal flow from basic codes to elaborate totally marks the failure of reductionism for the complex system that we call biology” (Gould, 2001).

The concept of “gene” inherited by molecular biology has therefore been broadly revised (Moss, 2006; Pichot, 1999), taking into consideration that gene function is in fact “distributed” along a connection of corporate bodies that interact among them according to a not-linear dynamics (Siegelmann, 1998). Eventually, gene functional expression has lost a lot of its deterministic character after the demonstration of the fundamental stochasticity of gene expression at the single cell level (Elowitz et al., 2002).

The discovery of an irreducible level of stochasticity in single cell gene expression coupled by the substantial invariance of transcriptome profile at the tissue level emphasizes a fundamental question: how to reconcile the existence of stochastic phenomena at the microscopic level with the orderly process finalized observed at the macroscopic level. This situation is somewhat analogous to the behavior of gases, resolved by the classical thermodynamics for equilibrium systems, and further, by the non-equilibrium thermodynamics for dissipative processes (Nicolis and Prigogine, 1989). The theoretical framework provided by non-equilibrium theory contradicts the paradigm proposed by Schrödinger (1944) which was enthusiastically adopted by molecular biology. According to such an approach, “order originated from order”, through the decoding of the information flux from DNA into proteins and, thereby, into tri-dimensional structures: each level of organization was produced by ‘specific’ interactions at the lower level. Thus, cell differentiation and organism development are traditionally described in deterministic terms of program and design, echoing a conventional clockwork perception of the cell at another scale. Accordingly, this conceptualization manifests itself in an all-pervasive vocabulary of “locks”, “keys”, “machineries”, “power”, “signals”, that populate the past and current biological and medical literature. These exchanges consider valid all the familiar implications, consequences and interrelations between concepts used as metaphors. Thereby, methodologies as well as intellectual approaches are coherently shaped according to the aforementioned framework. Little doubt is left about adherence to such a mechanistic view significantly handicaps our ability to adequately comprehend and model biological phenomena (Kurakin, 2005).

Those statements and the widely used concept of “genetic program” are currently challenged by an alternative view for which the order “emerges” at the macroscopic level (cell, tissues) as a consequence of the microscopic stochastic behavior (Kauffman, 1995).

According to the classical deterministic, “instructive” model, cells differentiate and activate functional programs depending on “specific” signals. Every signal is thought to correspond to a “command” of the genetic “program”. According to this deterministic model, all cells answer to the stimulus in the same way. Variability is not contemplated other than for correlated variance (externally imposed variability) or in the form of instrumental variability due to the uncertainty of the measures. On the contrary, the “selective” model posits that variability occurs on a larger scale and cells differentiate as a result of stochastic genetic events (Laforge et al., 2005).

The stochasticity of gene expression, originally proposed in 1983 (Kupiec, 1983), is today supported by a body of experimental data. Stochasticity is an inherent property of the non-linear dynamics of gene expression, which, in turn, can lead to bi-stable states in gene network activity (Becksei and Serrano, 2000): as such, it underlies the behavior of isogenic macromolecules (Xie and Lu, 1999), cells (Hume, 2000; Blake et al., 2003) and organisms (Herndon et al., 2002). Moreover, proteins are less specific than previously thought, and they can interact with different molecular components: in other words, protein interactions are also intrinsically stochastic and are not ‘directed’ by their ‘genetic information’ (Kupiec, 2010). This implies that, notwithstanding that differentiation is a highly precise and reproducible phenomenon, a deterministic mechanism supporting it is not really needed. Indeed, biophysical as well as biochemical interactions between cells and the surrounding microenvironment (stroma, extracellular matrix) converge in sorting and subsequently stabilizing the cellular phenotype, henceforth addressing its differentiation fate (Till, 1981; Balazsi et al., 2011) according to a Darwinian (selective) model of cell differentiation (Kupiec, 1997). Thus, the genome should not be considered a deterministic execution program (Coen, 1976), but rather a ‘database’ from which the dynamics of intra- and intercellular biophysical networks actively choose the desired inputs according the current needs of the system (Atlan and Koppel, 1990). Those features challenge expectations and assumptions of linear causality and reductionism that characterize the current molecular paradigm (Moss, 2006; Kurakin, 2005).

Consequently, scientific research was legitimate to give up models based on linear dynamics that are being substituted by approaches based on far-from-equilibrium systems and upon non-linear mathematical approaches (Kellenberger, 2004; Longo et al., 2012a, b).

A system characterized by non-linear dynamics is confined within a discrete number of configurations (stable states), represented by attractors in a phase-space landscape. Non-linear dynamics lead to symmetry breaking, hence allowing the system to choose among different fates, i.e. stable states or eventually chaotic regimens. Symmetry breaking confers irreversibility to the system, positioning it within the “arrow of the time”, previously “omitted” in classical physics: that is to say the system has now a history and its further evolution shall depend from choices undertaken at the bifurcation points. Moreover, such ‘complex’ systems may display the property of self-organization, characterized by the ‘spontaneous’ emergence of properties and ordered structures in time and space that confer to the system novelty and adaptation to a changing environment. These features are ‘uncommon’ for classical physical objects, characterized by stable symmetries and invariance, whereas in biological systems theoretical symmetries change and they become specified along (and by) their history (Longo et al., 2012a, b). Newtonian physics as well as molecular biology are clearly unfit to address these problems. However, around the middle of last century, researchers of different disciplines provided theories, concepts and methods in order to cope with complexity. Their contributions are coalescing into a new approach: Systems Biology.