Note – this is a beta release.  (Release early, release often.)  Still TODO – expand the computational bits, add references, tighten up the prose.  Feedback is appreciated!

Life is rarely unicellular.  In nature, bacteria and fungi live in structured communities such a biofilms that allow them to efficiently utilize nutrients and resist environmental stressors[cite — who?].  However, while the field of synthetic biology has developed tools and processes for engineering complex, dynamic behavior in single cells and monocultures, creating useful multicellular behavior is still on the forefront of bioengineering.  [cite? why?] Engineered multicellular systems may have broad uses in biotechnology and biomanufacturing, for example improving the efficiency of bioproduction of fuels or recombinant protein therapeuticsm[make more concrete].  More fundamentally, creating model multicellular systems may allow us to test hypotheses about their natural counterparts, with implications that span from materials science to human health and medicine [make more concrete]. [Add goal sentence — and a tractable problem w/ undergrads]

Multicellular behavior arises from two basic cellular mechanisms: specialization and communication.  Cells specialize in order to spatially or temporally segregate different functionalities.  For example, cells on the outside of a biofilm may produce metabolically expensive peptidoglycans to provide protection from the environment and attach the colony to an underlying substrate, while the internal cells may adjust their metabolism to “feed” the cells on the outside [cite? better example? who was it at the CILSE kickoff?]  These genetic programs are coordinated by intercellular communication, which can drive the emergence of specialization from a mongenic population. Together, these two behaviors make the population more productive and resilient, increasing the efficiency with which its genetic material propogates.

The baker’s yeast Saccharomyces cerevisiae is uniquely suited for engineering novel multicellular behavior.  It is robust, easy and inexpensive to culture[cite], and it has a remarkable (and expanding) suite of tools for genetic manipulation.  Eighty-five percent of its genes have a putative annotated function[cite], which makes it possible to build synthetic gene networks that interact with native cellular machinery.  Finally, S. cerevisiae has a simple but comprehensive suite of pre-existing multicellular behaviors which may be amenable to co-opting. Cells in a population communicate via mating pheremones, ammonium species and localized changes in pH[cite], which drive programs of gene expression that include a basic palette of morphological modules: FLO genes that promote adhesion [and what else?].  In natural isolates (as well as domesticated laboratory derivatives), these modules give rise to natural multicellular behavior including flocs, mats, and biofilms.  Even stand-alone colonies on solid agar media evince structure and specialization, with the inner cells “feeding” metabolites to the outer cells which undergo meiosis and form spores.

These proto-metazoan behaviors can be leveraged and augmented using a synthetic biology approach.  Synthetic biology uses engineering principles such as modular assembly and computer-aided to design to build new biological systems with pre-specified behaviors.  A number of recent advances have made S. cerevisiae a particularly attractive “chassis” organism to host synthetic gene networks.   First, a number of recently developed methods[moclo, gateway/gibson, ytk] make it easy and inexpensive to assemble together DNA “parts” into large multi-transcription unit networks, then reliably integrate them into the S. cerevisiae genome.  These modular assembly methods have dramatically decreased the cost and effort required to build and test multi-TU synthetic gene networks in S. cerevisiae, which will be necessary for precise, coordinated control over new “genetic programs” required to build novel multicellular behavior.  Second, concerted efforts by CIDAR and others have produced a new generation of biodesign tools that can compose the measured behavior of various genetic “parts” and predict the behavior of the composed systems.  Though still nascent, these design tools promise to leverage “big data”-style machine learning to explore design space more rapidly and completely than is possible experimentally.  Recent results suggest that these tools lead to synthetic gene networks that are more robust than is possible by hand.

By using computer-aided design tools and modular gene network assembly, a synthetic biology approach can leverage both existing S. cerevisiae behaviors and systems from other organisms to build novel multicellular behaviors in yeast.  One promising avenue is to use a “quorum sensing” approach to control density-dependent behavior in axenic cultures, which has immediate biotechnological applications.  For example, natural yeast flocculation is used to separate yeast at the end of a fermentation, but the success of flocculation depends sensitively on the quantities of unfermented sugar, fermentation products (such as ethanol) and oxygen.  The molecular players involved in flocculation (such as the adhesion protein FLO11) are well-understood, and a synthetic gene network that activated them using additional endogenous sensors could make floccing occur more robustly and under a broader range of fermentation conditions.  Alternately, a synthetic quorum sensing system could be used with a batch culture for multi-step biosynthesis of compounds where the final product is toxic to the cell.  At low densities, the cells could produce a non-toxic intermediate, then switch genetic programs at a high cell density to produce the final toxic product.  A synthetic yeast quorum sensing system was developed in the Weiss lab by co-opting the cytokine isopentyladenine from Arabidopsis, but it may require optimization to use in a biotechnological setting.  Alternately, other intercellular signaling platforms (such as the N-acyl homoserine lactone from Vibrio species or peptide-based intercellular signaling from higher eukaryotes) may provide better characteristics (such as a wider on/off ratio.)

These approaches may also be used to study spatial behavior in developing yeast structures as a model of metazoan cellular organization.  As discussed above, S. cerevisiae naturally forms a number of multicellular structures, including flocs, mats, and biofilms.  These structures arise from the systematic execution of a genetic program that drives the establishment of different fates from a monogenic, homogeneous population, and the ability to activate different parts of this program synthetically may lead to insights into both the emergence of these structures in yeast, as well as similar programs of differentiation and development in metazoa.  Synthetic gene networks that interact with natural ones allow us to perturb them in a controlled manner to probe the robustness of the developmental outcome.  Networks that implement a subset of these morphogenic “modules” synthetically (or activate the natural ones via synthetic signals) allow us to probe which ones (and under what dynamic program) are sufficient.  Finally, such a synthetic system may lead to insights into the organization, encoding and transmission of information in developing biological systems.  Because changes in gene expression happen on the same timescale as cell division, teasing apart the dynamics of the development of these structures may only be possible in systems where they can be activated, modulated and perturbed synthetically.

A number of features make this research program particularly well-suited for an undergraduate research environment.  First, it is hypothesis based.  Many synthetic biology projects, inspired by more mature fields of engineering, are focused on meeting an a priori specification: the goal of the project is to build a synthetic gene network that meets that specification.  Unfortunately, the field is still far from mature, and it is frequently unclear at the outset of a project how best to fulfill the specification; thus, many synthetic biology projects devolve into open-ended technology development efforts.  Hypothesis-driven projects, on the other hand, have a defined end-point that is reached when data is gathered that supports or disproves the hypothesis. Based on either biological insight or (more and more commonly) modeling and simulation tools, a student can design a synthetic gene network, then use modern modular cloning techniques to rapidly build the gene circuit and use it to test the hypothesis.

Second, the intersection of biology and engineering holds authentic research opportunities for students of many different skills, backgrounds and interests.  For biologists, this research program presents opportunities to study fundamental biological questions in a relevant, powerful model organism.  For students of computer science, it presents an opportunity to develop expertise in modelling and data analysis.  For mechanical engineers, the opportunity to develop novel tools for culturing and analyzing cells.  Synthetic biology and bioengineering exist at the intersection of all of these fields, and the opportunity to work with a diverse team to address interesting problems carries with it a whole host of soft skills such as project planning, communication, goal-setting, and conflict resolution, the likes of which will be useful as a professional even in fields completely divorced from synthetic biology.

Finally, maintaining a longitudinal research program with a rotating cast of undergraduates presents challenges relating to continuity: the availability of materials, reagents and data generated by past students must be readily available to current researchers to enable building off of eachothers’ work. The key is to prevent the formation of silos (of information or of physical materials.)  A single system for recording experiments and data, and for connecting them to physical locations of materials and reagents, is key for new members of the laboratory to efficiently begin building off of others’ work.  Fortunately, efforts of out the Densmore and Klavins laboratories are producing software to enable precisely this centralization, and I fully intend to build on their efforts to build an open, transparent, and efficient laboratory environment.  Many academic laboratories both inside and outside of synthetic biology face similar problems, and solving it in a limited environment such as the laboratory I propose to establish could have broad implications for enabling similar openness and transparency in other labs.

Cells manipulate matter, energy and information with precision and finesse; writ large, the goal of synthetic biology is to harness these capabilities for engineered systems.  Engineering reliable multicellular systems has implications for biotechnology and medicine, as does the insight to be gained by building model multicellular systems in order to better understand natural ones.  By leveraging the decades of study that have made S. cerevisiae a model for eukaryotic cell biology, as well as more recent developments that have made it an attractive host for synthetic biology, this research program promises to address important questions and enable authentic research experiences for undergraduate scientists.