Rational Engineering Principles in Synthetic Biology: A Framework for Quantitative Analysis and an Initial Assessment
- First Online:
- Cite this article as:
- Giese, B., Koenigstein, S., Wigger, H. et al. Biol Theory (2013) 8: 324. doi:10.1007/s13752-013-0130-2
- 189 Views
The term “synthetic biology” is a popular label of an emerging biotechnological field with strong claims to robustness, modularity, and controlled construction, finally enabling the creation of new organisms. Although the research community is heterogeneous, it advocates a common denominator that seems to define this field: the principles of rational engineering. However, it still remains unclear to what extent rational engineering—rather than “tinkering” or the usage of random based or non-rational processes—actually constitutes the basis for the techniques of synthetic biology. In this article, we present the results of a quantitative bibliometric analysis of the realized extent of rational engineering in synthetic biology. In our analysis, we examine three issues: (1) We evaluate whether work at three levels of synthetic biology (parts, devices, and systems) is consistent with the principles of rational engineering. (2) We estimate the extent of rational engineering in synthetic biology laboratory practice by an evaluation of publications in synthetic biology. (3) We examine the methodological specialization in rational engineering of authors in synthetic biology. Our analysis demonstrates that rational engineering is prevalent in about half of the articles related to synthetic biology. Interestingly, in recent years the relative number of respective publications has decreased. Despite its prominent role among the claims of synthetic biology, rational engineering has not yet entirely replaced biotechnological methods based on “tinkering” and non-rational principles.
KeywordsSynthetic biologyBibliometric analysisDesignEvolutionRational engineeringTinkering
Since the beginning of the 21st century, engineering principles seem to have become increasingly influential in attempts to manipulate living organisms. In a biological context, “engineering” is no longer restricted to classic biotechnological tasks such as upscaling or fermenter turbation. Engineering has entered the cellular level and is now—quite contrary to former efforts that were focused on culturing conditions—trying to adapt the cells themselves to human needs.
This change can be considered as qualitative—rather than quantitative, just referring to scope and precision. It seems to belong to a process of “instrumentaliz[ing] animate nature” by biotechnology and consequently “turn[ing] organisms into manufactures,” as Pottage and Sherman (2007, p. 545) characterized it. In the case of synthetic biology the common rhetoric is “absolute control,” achieved by elimination of any unadapted function or characteristic of biotechnologically used microorganisms, agricultural crops, or livestock that is not of any use for man. Economic metaphors are prevalent throughout the discourse on synthetic biology. All functions that do not create added value are considered as excrescent and disturbing. Thus, synthetic biology is often defined as “the science of reassembling catalogued and standardized biological components in a systematic and rational manner to create and engineer functional biological designer devices, systems and organisms with predictable, useful and novel functions” (Weber and Fussenegger 2012, p. 21). Improved “efficiency” is considered to be realized by implementing only the target functions and removing all unintended interfering dependencies within cells and with their environment. Mechanisms supporting these functions should become more amenable to planning. Thus, the attempt to strengthen the quantitative and predictive character of biology is based on technological, economic, and reliability considerations (Lazebnik 2002; Cambray et al. 2011).
The guiding idea is to improve efficiency by practices adopted from the engineering point of view. Engineering, whose Latin origin “ingenerare” means to implant, generate, or produce (Mitcham and Schatzberg 2009), changed within the industrial revolution from an artisan profession to a business of scientific practitioners (Layton 1971). This transformation is characterized by a separation of production and design of technical artifacts (Kroes 2009). A number of definitions for engineering refer to intentional design as a preceding step in which individual constraints have to be fulfilled in an iterative process (Banse and Grunwald 2009; Kroes 2009).
Both classic and recent definitions of engineering also stress that “directing the sources of power in nature” is a major claim (Mitcham and Schatzberg 2009, p. 41). Accordingly, the principal characteristic of synthetic biology that was taken from engineering might be the separation of design and fabrication. This is achieved by an iterative process starting at design and construction, continuing via in silico and in vivo evaluation, and ending at production (Brent 2004; Schyfter 2011). Cambray et al. (2011, p. 627) gave a definition for a design process meeting the demands of rationality: “[…] desirable […] is a rational and transparent design process wherein systems are built from understandable components whose interconnected, composite behavior is predictable.” Therefore, rational engineering in the sense of synthetic biology can be defined as a predefined synthesis using completely characterized components to avoid any uncertainties, including unpredictabilities. In other words: rational engineering requires the ability to predict and to control biological processes—whereby biological processes are mainly regarded as deterministic. Modularity, orthogonality, and a robust design are supposed to be required for rational engineering in biology (Heinemann and Panke 2006).
Following Endy, synthetic biology seems to be on the way to a rational engineering approach. However, at the moment it is faced with some uncertainties that give rise to the major challenges of synthetic biology: managing biological complexity, inefficient and unreliable construction, unpredictable changes of system performance, as well as evolutionary changes. Therefore three basic engineering principles should provide the fundament for coping with these challenges and enable rational construction and reliable function: standardization, decoupling, and abstraction (Endy 2005). The basic principle of standardization is established in the development of basic biological parts and standardized conditions for their operation and combination. Decoupling is not only achieved by the separation of design and fabrication, it is also represented by the subdivided layout of systems into different independent “devices.” Finally, abstraction could establish a hierarchy of levels with different complexity as represented by a differentiation in DNA, parts, devices, and systems (Endy 2005). A distinction between parts and devices on the one hand, and systems on the other hand, is the time-dependent and dynamic character of the latter, constituting their relation to systems biology.
However, it is still contested whether rational engineering principles are applicable and adequate for shaping living nature according to human needs (Wimmer, quoted in Breithaupt 2006). The core of the controversy are contrary views of scientists on the pros and cons of variations in biological systems. A rather “rational” group advocates control by suppressing variation, and a rather “evolutionary” group regards “noise” or “randomness,” the expressions of instability, as an elementary part of biological function and development: “Noise is not merely a quirk of biological systems, but a core part of how they function and evolve. … [T]he question of how cells and organisms use and control random variation in their own components to grow, develop and evolve goes right to the heart of many fundamental biological problems” (Eldar and Elowitz 2010). Bujara and Panke (2010) consider techniques based on evolutionary principles (or random variations) as an intermediate step that is necessary due to the still fragmentary knowledge of biological processes, and they assume that it will be overcome on the way to fully rational engineering design. Major limitations of evolution-based methods are seen in the lack of knowledge of the underlying structure for the achieved improvements, including the high number of constructs that have to be evaluated in random or directed evolution approaches. The “evolutionary” group of scientists—instead of attempting to suppress variations—established efficient methods of handling the stochastic character of biological mechanisms, based on evolutionary processes (Dymond et al. 2011; Wang and Church 2011). These methods depend on a selection of accidental or random mutation events. To some extent this might be regarded as a contradiction to the deterministic character of an idealized type of rational construction. Nevertheless, some authors believe that they will complement rational engineering methods in the future (Dougherty and Arnold 2009; Michalodimitrakis and Isalan 2009).
In the following we will show to what extent rational engineering goals have been realized and implemented in synthetic biology. We begin by conducting a systematic analysis of the subfields of synthetic biology and evaluate their aims and specific requirements for rational engineering. Subsequently, a quantitative bibliometric analysis of synthetic biology is carried out, based on search terms that were derived from the aims and requirements identified in the previous step of our analysis. Finally, by analyzing the specialization of the most prominent authors we examine the methodological polarization of the scientific community of synthetic biology.
A Bibliometric Strategy Exploring the Methodological Aspects of Synthetic Biology
In order to determine the relevance and predominance of rational engineering principles in present-day synthetic biology, we make use of (quantitative) bibliometrical methods that allow for a statistical overview that cannot be provided by qualitative analyses.
Queries were executed in Web of Science within “Science Citation Index Expanded.” Unless noted otherwise no time limit was set for the past. The lemmatization function (i.e., alternative spelling for terms without quotation marks is accepted) was activated. Title, abstract, author-keywords, as well as “KeyWords Plus” (cited article titles) were included in the queries.
General Characterizing Terms
It becomes apparent that engineering-related terms like “module” (17 %), “orthogonal” (4 %), “rational” (8 %) and “robust” (11 %) do not occur frequently with “synthetic biology.” Since these terms constitute some of the core-principles of engineering, this result suggests that these basic elements are seldom explicitly referred to. However, the frequent use of “engineer” (42 %) illustrates that its relevance is as accepted as the main principle of synthetic biology: the preceding step of “design” (47 %). At a ratio of 19 %, the term “evolution” is more frequently used within publications related to synthetic biology in comparison to the engineering-related adjectives mentioned above. One third of these publications refer to “directed evolution” (data not shown) which could be a “… a powerful complement to ‘rational’ engineering approaches.” (Dougherty and Arnold 2009, p. 486). The quite frequent co-occurrence of the terms “network” (35 %) or “system” (57 %) with “synthetic biology” indicates the close relation of synthetic biology with systems biology.
A Strategy for the Investigation of Methodological Characteristics
Overview of synthetic biology subdisciplines, divided into the parts-devices-systems scheme, with respective aims for construction related to rational engineering principles
Level of abstraction (based on Endy 2005)
Modular standard parts (BioBricks), synthetic genes
Extending the natural functional spectrum, expression of new functional sequences or proteins
Welch et al. (2011)
Tian et al. (2009)
Sismour and Benner (2005)
Protein design, in vitro evolution
Computational design, extended functionality in biological regulation on the genetic (protein–nucleic acid) and biochemical level (protein–protein)
Kortemme and Baker (2004)
Behrens et al. (2011)
Reliable and robust programming extending the natural functional spectrum, optimization, regulation, detection
Win et al. (2009)
Isaacs et al. (2006)
Genetic circuit construction (including RNA-devices)
Construction of either autonomous or integrated circuits for novel or optimized functions
Greber and Fussenegger (2007)
Nandagopal and Elowitz (2011)
Michalodimitrakis and Isalan (2009)
Carothers et al. (2011)
Reengineering cells—improving yield and productivity
Carothers et al. (2009)
Nielsen and Keasling (2011)
Lynch and Gill (2011)
Systems and chassis
Inter- and intracellular systems
Coordinated combination of different devices (functions)
Bujara and Panke (2010)
Purnick and Weiss (2009)
Weber et al. (2007)
Provision of a “chassis”
Jewett and Forster (2010)
Moya et al. (2009)
Construction of a “chassis”—organization of continuous self-replication
Noireaux et al. (2011)
Microreactors, nano reactors
Construction of an alternative “chassis” without continuous self-replication
Richmond et al. (2011)
Shchukin and Sukhorukov (2004)
Further analysis of the impact of rational engineering including a detailed evaluation for the three abstraction levels of synthetic biology has to include an appropriate assembly of keywords for bibliometric screening. Keywords for rational engineering methods were collected according to their relation to the above-mentioned basic engineering principles of standardization, decoupling, and abstraction in a separated process of rational design and fabrication. Only methods that are regarded to have a high potential of supporting a rational engineering vision, i.e., a reliable, verifiable rational design (Cambray et al. 2011; Ellis et al. 2011) were included in the “rational engineering” classification. In contrast, although (evolutionary) “trial-and-error” or “tinkering” methods may be part of an engineer’s everyday work, they yield no valuable information for advancement of the field, unless followed by systematic examination (Jeremy Knowles cited in Benner et al. 2011).
Search terms for bibliometric analysis
(“artificial protein” OR “artificial RNA”)
((RNA OR enzyme OR protein) design)
(“in vitro evolution” OR “directed evolution”)
(“rational protein design”)
(“de novo” “enzyme design”)
(“computational protein design”)
(“computational design” protein* function)
(“in silico” (“protein design” OR “RNA design”))
((rational design OR computational design) molecul* biol*)
(“computational design” (RNA OR riboswitch OR ribozyme))
(“rational design” (RNA OR riboswitch OR ribozyme))
(“noncanonical amino acid*” (engineer* OR construct* OR design))
(“unnatural amino acid*” (engineer* OR construct* OR design))
(modeling (RNA OR protein))
(biol* “standard* parts”)
(“DNA assembly” genome)
((“whole genome” OR “whole-genome”) AND (synthesis OR assembly))
((cell OR metaboli*) AND “pathway design”)
(“gene regulatory network*”)
(synthetic “regulatory network*”)
(“synthetic gene network*”)
(gene* AND synthetic AND “circuit design”)
(“genetic circuit*” (artificial OR synthetic))
(engineering “gene cluster*”)
(“metabolic pathway*” “rational design”)
(“biosynthetic pathway*” “rational design”)
(“minimal gene set” (design OR engineer* OR construct* OR synthetic))
(chemical-synthesis AND genome AND (engineer* OR construction))
(“synthetic genome” OR “artificial genome”)
((vesicle OR vesicul*) bioreactor*)
(CAD or “computer aided design”)
(CAE or “computer aided engineering”)
(CAM or “computer aided manufacturing”)
(SBML or “markup language”)
(a) “Parts” are molecules that perform a basic biological function (Endy 2005). The category also contains proteins with unnatural aminoacids (Hoesl and Budisa 2011), as well as genes and ribonucleic acids. All these approaches try to expand the functional spectrum and to deliver more options for regulation. At the same time, activities are focused to achieve a highly controlled, rational, and—especially for proteins—automated production of molecules.
At the “parts” level, keywords pointing to the use of rational design (“computational,” “intentional,” or “de novo design”) were categorized as “rational engineering” because these practices enable advancements in the understanding of the underlying molecular mechanisms, and therefore support abstraction, decoupling, and standardization. The same was assumed for approaches using (“in silico”) modeling in parts design. The inclusion of functional RNA molecules (“riboswitches,” “ribozymes”) also reveals a detailed mechanistic understanding, since RNA design is based on a quite good understanding of structure–function relationships (Isaacs et al. 2006).
In contrast, the directed evolution approaches broadly used in present protein design were excluded from the “rational engineering” methods because their development is highly dependent on the system used and less easily transferable to other environments (Arkin and Fletcher 2006).
(b) A combination of parts arranged to perform a human-defined technical function constitutes the “device” level in the abstraction hierarchy of components. Publications of the “device” level were categorized on the basis of approaches to create the information-carrying DNA molecules, since this is currently the limiting technology (Ellis et al. 2011). Modular design with controlled interfaces between modules (“biobrick,” “standard biological part”) is possibly the most-cited rational engineering-inspired method in synthetic biology (Endy 2005; Shetty et al. 2008). The design of devices based on a dynamic, systems biology view of cells (“genetic circuits,” “regulatory networks,” “signal networks”) also represents an important prerequisite to achieve decoupling and standardization (Purnick and Weiss 2009). Therefore, these approaches were included in the “rational engineering” methods. Nevertheless, approaches not regarded as “rational engineering” include devices synthesized in the pattern of complete devices from other organisms. These rely on the solutions realized in an organism by evolutionary contingency that have higher potential for interference and context-dependence (Arkin and Fletcher 2006). Recent methods based on high-throughput recombination and genome-scale synthesis enable more powerful optimization, but also offer little understanding of the genotype-phenotype relationship and are semi-rational at best (Cambray et al. 2011). As non-rational engineering approaches in the “device” category are not defined by characteristic techniques, but rather by what is not done (modularization), no positive search terms for “tinkering” approaches were used.
(c) Finally we considered the term “systems,” which are either formed by combinations of more elementary modules or as a chassis established by top-down or bottom-up approaches. Chassis are minimal cells with basic genetic and biomolecular machinery (Jewett and Forster 2010) or simple vesicular structures. For protocells, these basic functions besides cellular self-replication still represent a highly ambitious goal (Solé et al. 2007). Micro and nano reactors are excluded from these requirements and serve only as delivery vehicles or stationary shells for reactive systems.
At the “systems” level, search strings for the “rational engineering” category include two aspects: (1) design platforms and (2) provision of an independent chassis. An ideal chassis would be a universally usable basis for insertion of devices, or at least usable for a broad range of applications. Bottom-up approaches of genome and compartment synthesis provide the best control over chemical cell composition and an understanding of interactions (Noireaux et al. 2011; Richmond et al. 2011), therefore promising the best chances for abstraction and standardization. Otherwise, minimal cells and genomes derived from simple natural cells in a top-down approach have higher risk of molecular leftovers from the original cell, which can interfere with part function (Andrianantoandro et al. 2006) and were classified as “tinkering.”
A reliable design platform organizing a repeatable and interchangeable workflow with bioinformatical support (“design platform,” “CAM or computer aided manufacturing,” “integrated design”) is necessary for dividing and streamlining the design process, increasing productivity and facilitating standardization (Marchisio and Stelling 2009; Cambray et al. 2011).
Further approaches which aim at creating alternative nucleotides (genetic code expansion), while arguably part of synthetic biology, were not included in our analysis because these approaches reflect rather an alternative approach which may be called “artificial biology,” and remain at an early stage of development not yet integrated in the design process of biological systems (Benner et al. 2011).
The Extent of Rational Engineering Principles
For an estimation of the relevance and prevalence of rational engineering in synthetic biology, the above described terms are used in a query to determine the number of articles where rational engineering principles seemingly dominate the methodology. To minimize the influence of theoretical discussions, review articles were excluded from the search and queries are limited to “articles.” The hits returned by the search were checked by evaluating titles, abstracts, and, in ambiguous cases, complete texts of the articles for the compliance of their approach and methods to the concepts of rational engineering.
In the category ‘‘parts’’, the majority of rational engineering results reported molecules specifically constructed for use in network interactions or aimed at functions on the whole-cell level, pointing to a systems biology-inspired approach in construction, while the “tinkering” results included mostly non-modular and contingent manipulations. In the category “devices”, rational engineering search terms yielded DNA-based oscillators and other functional modules, frequently created with the help of computer-aided design. On the systems level, many of the rational engineering articles referred to bottom-up design enabled by modeling of circuits and networks, while the “tinkering” articles reported mostly smaller experimental manipulations of naturally existing networks and genomes.
The results contained minor overlaps and errors, i.e., a small percentage of hits were returned in more than one or a wrong category (see the supplemental table in Online Resource 1). The robustness of the analysis could thus in the future be improved by setting a threshold for the “relevance” value returned for each result by Web of Science, but in this first establishment of the search framework we chose to include all hits to provide a broad basis for discussion.
For publications before 2009 we obtained a distribution similar to that for the complete query (up to and including 2012) (Fig. 2b), but the results for articles before 2009 yield lower total numbers for parts, devices, and systems—accounting for nearly one-fifth of the hits that were obtained in the current query.
The results demonstrate that rational engineering approaches have permeated all subfields of synthetic biology and a significant part of the articles focus on rational engineering principles. Nevertheless, a rational engineering approach is at present far from dominating the overall amount of work in synthetic biology, as one could be led to believe by some reviews that attempt a unification of the field (Endy 2005; Andrianantoandro et al. 2006; Arkin and Fletcher 2006; Heinemann and Panke 2006).
Furthermore, we analyzed the rational engineering-related articles in synthetic biology in relative numbers by comparing a period from 2009 to 2012 with results for articles that were published before 2009 (Fig. 2c). The results of the merged query for the whole field (by a combination of all search terms) as well as for the different abstraction levels, show a stagnation for the influence of rational engineering principles. Except for “parts” with a stable number of articles, the relative numbers for “devices” and “systems” decreased—despite growing publication numbers for all levels (Fig. 2a, b). The overall frequency of terms related to rational engineering decreased from two-thirds of all research articles in the field of synthetic biology to less than half of the articles. This change is due to a decrease of the respective methods in the abstraction levels “devices” and “systems.”
The observed decrease in publications related to a rational engineering approach may be explained by a number of obstacles that appeared during the attempts to create increasingly complex biological systems. In many areas, typical rational engineering methods turned out to be rather difficult to implement. For instance, the limited predictability of standardized components caused by context-dependencies is one of the most considerable obstacles in genetic design automation and, thus, in rational engineering (Lux et al. 2012). Besides the interfering interactions with the cellular environment, functional connections of synthetic modules may require extensive tuning of input/output-characteristics as well (Kittleson et al. 2012). Moreover, genetic mutations (Arkin and Fletcher 2006) as well as noise (Becskei et al. 2001) have the potential to cause unintended effects in genetic circuits. Finally, evolution causes a loss of functionality in whole populations (Arkin and Fletcher 2006).
The decrease in numbers of publications related to rational engineering might also partially reflect the technical progress in evolutionary methods (Dougherty and Arnold 2009; Marlière et al. 2011; Wang and Church 2011) and in DNA-synthesis, enabling economic coding of complex functions (Gibson et al. 2008) independent of standardized sequences. Simultaneously, “biobricks” have suffered in their popularity due to the big effort that is required to set up a reliable database providing proper descriptions (Kwok 2010).
Our investigations are based on the common designation of methods and objects in the field and, therefore, enable a broad quantitative analysis. However, limitations will probably arise from the proper use of terms by the authors. The keyword-based approach is prone to articles where phrases are only used to describe the authors’ future agendas rather than depicting their present-day practices. Nevertheless, the broad overview provided by the bibliometric analysis should compensate for this disadvantage—at least to some extent.
Our approach provides a framework for the quantitative investigation of methodological preferences in synthetic biology. The analysis presented here reveals the influence of rational engineering in present-day synthetic biology to be relevant for a considerable number of authors and publications. Nevertheless, rational engineering approaches do not account for the majority of publications.
An investigation of co-occurrence of rational engineering-related terms revealed a very limited use of typical object-related qualities of rational engineering. Instead, the term “evolution” is used quite frequently. This may point to the fact that attempts to establish rational engineering often have to be complemented by evolution-based techniques to achieve the expected results (Dougherty and Arnold 2009; Lynch and Gill 2011).
Applying a keyword-based analysis of the extent of rational engineering principles, we showed that they are a subject for slightly half of the articles published up to now. An analysis of authors in the field of synthetic biology with a focus on rational engineering revealed that a majority of them are concerned with both—methods of evolutionary and rational character—without remarkable changes in their absolute publication activity. Only a limited group of scientists is concentrating on rational engineering.
After a decade of synthetic biology development, the influence of rational engineering principles can be interpreted from two sides: a quota of roughly 50 % represents substantial progress compared with the evolution-based methods and manipulation of single genes used in traditional molecular biotechnology. However, it is still far from what many advocates in synthetic biology want to accomplish and what they consider as a unifying approach of the field. One reason certainly is that rational engineering principles are far more difficult to implement in a biological context than in a classical technical environment (Kittleson et al. 2012; Perkel 2013). Apart from obstacles in implementing and configuring systems—in particular with a high degree of complexity (Purnick and Weiss 2009)—limitations could arise as well from the lack of evolutionary stability in engineered biological systems (Arkin and Fletcher 2006). Whether abstraction can still be promoted as a promising strategy in managing complexity will be an important question for future investigations in this context. It remains open whether the suggested rational engineering methods are essential for the development of synthetic biology as an engineering discipline that meets the goals claimed by its advocates. The framework of systematization and evaluation of synthetic biology proposed here in this can serve as a starting point for the continuous assessment of the success in meeting these goals.
According to Kroes (2009), problems with “classic” rational engineering principles are not restricted to the constraints of biological systems. Emergent phenomena in complex technical systems already question the applicability of traditional rational design paradigms. Therefore, the claim of extensive control over the designed system might be over ambitious with regard to biological objects. That result might instead call for alternative design paradigms. To give up this control would be a potential revolution in systems engineering, but it would confront us with a whole new set of problems and challenges in issues of (biological) safety.
We are indebted to the following colleagues: Christian Pade for fruitful discussions, valuable comments, and critically reading the manuscript; and Robin van der Auwera, Jan-Ole Werner, and Sven Rohrdanz for their help in queries and graphic visualization of data. This work has been funded by the German Ministry for Science and Research (BMBF) within the study “Technology Assessment of Synthetic Biology” under code 16I1611. The authors declare no competing interests.