Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Building a biofoundry

Building a biofoundry A biofoundry provides automation and analytics infrastructure to support the engineering of biological systems. It allows scientists to perform synthetic biology and aligned experimentation on a high-throughput scale, massively increasing the solution space that can be examined for any given problem or question. However, establishing a biofoundry is a challenging undertaking, with numerous technical and operational considerations that must be addressed. Using collated learnings, here we outline several considerations that should be addressed prior to and during establishment. These include drivers for establishment, institutional models, funding and revenue models, personnel, hardware and software, data manage- ment, interoperability, client engagement and biosecurity issues. The high cost of establishment and operation means that developing a long-term business model for biofoundry sustainability in the context of funding frameworks, actual and po- tential client base, and costing structure is critical. Moreover, since biofoundries are leading a conceptual shift in experi- mental design for bioengineering, sustained outreach and engagement with the research community are needed to grow the client base. Recognition of the significant, long-term financial investment required and an understanding of the com- plexities of operationalization is critical for a sustainable biofoundry venture. To ensure state-of-the-art technology is inte- grated into planning, extensive engagement with existing facilities and community groups, such as the Global Biofoundries Alliance, is recommended. Key words: biofoundry; high-throughput; synthetic biology is considered by most to be essential to developing and growing 1. Introduction synthetic biology capacity. The high-throughput capability Currently, over 40 countries have national strategies relating to afforded by access to a biofoundry can satisfy this goal. the ‘bioeconomy’ (the economic potential of bioscience) and/or A biofoundry is an integrated molecular biology facility that synthetic biology, including the USA (1, 2), the UK (3) and the includes robotic liquid-handling equipment, high-throughput Australian Council of Learned Academies (ACOLA) report (4). In analytical equipment, and the software, personnel and data many of these strategies, the growth of synthetic biology capa- management systems required to run the equipment and bilities is identified as critical to scientific and economic com- broader biofoundry capabilities. Biofoundries marry synthetic petitiveness. A comprehensive, national infrastructure platform biology with automation engineering to create new high- Submitted: 13 July 2020; Received (in revised form): 26 October 2020. ; Accepted: 12 November 2020 V Crown copyright 2020. This Open Access article contains public sector information licensed under the Open Government Licence v2.0 (http://www.nationalarchives.gov.uk/doc/ open-government-licence/version/2/) 1 2| Synthetic Biology, 2021, Vol. 6, No. 1 throughput biological solutions that help to build and collaborators and/or clients, internal and external to one’s de- strengthen a Design-Build-Test-Learn (DBTL) approach to bio- partment or institution, is an important step to determining the logical engineering (Figure 1). viability of a facility. Market saturation, too many biofoundries Biofoundries are gaining popularity around the world, with for a given research community or insufficient client availability academic and commercial facilities being established across in smaller research communities, could result in facility failure North America, Europe and Asia-Pacific regions. A Global and loss of the significant establishment investment. One way Biofoundries Alliance (GBA) (5) for noncommercial biofoundries to avoid such an outcome would be a coordinated funding was launched in 2019, with 16 founding members (6, 7); and has scheme by national granting agencies, preferably as part of their already grown to 27 members in 2020. Core aims of the GBA in- national bioeconomy strategy. clude developing best practices across facilities, sharing of in- formation and resources and enhancing visibility and support 2.2 Scale of investment for these facilities. Biofoundries require significant time and human capital to set Establishing a biofoundry is a significant investment and up and operate. The quantum of funding required to establish requires more than simply setting up a well-equipped physical and support a biofoundry is contingent on the anticipated scale space. The emphasis on high-throughput methods requires concomitant attention to software, protocols and the integra- and reach of the facility, but necessarily extends beyond the tion of physical and digital infrastructures to efficiently prepare purchase of robotic, high-throughput equipment to include con- and track samples. In this respect, biofoundries are at the fore- sumables, software and support for skilled personnel to set up front of a paradigm shift in biological engineering toward a and run the facility. When seeking investment, provisions need more automated, design-focused venture. to be made for each of these elements. Long-term support from In this review, we offer an overview of key technical, organi- one’s home institution, including an understanding of the na- zational and operational issues relating to the setup and run- ture and scale of the venture, is critical. ning of a biofoundry. Our focus is on academic rather than commercial biofoundries, although there is clearly overlap be- 2.3 Nature of experiments tween the two, and at least some of our recommendations Through the use of automation to perform high numbers of re- might apply to commercial foundries. We distinguish between petitive, standardized tasks, biofoundries can dramatically in- academic and commercial biofoundries primarily with respect crease the throughput and design space for biological to their locations, with academic foundries being located in aca- engineering (6–8). Modularization of workflows allows for a mix demic institutions or government laboratories, and commercial and match approach and introduces flexibility in services of- facilities operating outside of an academic context. There are fered in the DBTL engineering biology cycle (Figure 1). However, broad differences in the target client base, funding and profit some types of experiments are more amenable to high- models, and sustainability/growth objectives of academic and commercial biofoundries. Academic biofoundries typically fo- throughput, automated workflows than others. Research pro- cus on supporting the research community and translational grams that require continual small (or large) adjustments to ex- activities, while commercial biofoundries have more of a focus perimental workflows should consider whether a full on commercial clients and investment return. The recent crea- biofoundry is the most appropriate solution to their needs, or tion of a GBA to specifically promote and support noncommer- whether bringing key, individual pieces of high-throughput cial biofoundries suggests there are common challenges for the equipment, such as liquid handlers, into their workflows might long-term development and sustainability of these facilities. satisfy their requirements. Focusing on bringing in key, individ- Our goal in this review is to share what we have learned with re- ual pieces of equipment can also be a strategy for lower- spect to establishing and running academic biofoundries. resourced settings looking to grow their biological engineering capabilities. 2. Rationale for Establishing a Biofoundry 3. Institutional and Funding Models Biofoundries deliver capabilities that allow for an accelerated approach to synthetic biology research and application. They Biofoundries have often been started with a block of strategic also facilitate development of economically important bioengi- funding from a public sector entity for the purchase of key ro- neered products and organisms. This underpins the strategic botic equipment, often with insufficient reference to long-term and economic drivers for acquiring this capability, and there is sustainability. The initial block funding must be matched with consequently growing interest and activity worldwide in their other sustained sources of funding, because the costs associ- establishment. However, a biofoundry requires significant—and ated with running a biofoundry are high: they require special- ongoing—investment, and so it is important to establish a clear ized staff, expensive equipment and large volumes of justification and a solid business case before proceeding. As consumables, as well as longer-term equipment maintenance part of business case development, the following key issues and upgrade costs. These costs have been singled out as a chal- should be considered: lenge for ensuring sustainability (7). Approaches vary for sourcing funding, and hinge to some ex- 2.1 Anticipated throughput (market demand) tent on national and institutional funding structures and priori- ties. For biofoundries based in individual research groups or The equipment found in most biofoundries enables a significant academic institutions, ongoing funding may be part of key in- scale-up in throughput. The throughput capacity of biofoun- frastructure allocations. Activities may be heavily subsidizd, or dries is rarely fully utilized; understanding the need or desire (market demand) is critical in order to ensure long-term, effi- grant-funded, to ensure access by institute members. cient use of the capacity of a biofoundry. This capacity typically Biofoundries with a national outlook may be funded by combi- exceeds the needs of a single laboratory, so identifying nations of public/government funding and grant funding, with M.B. Holowko et al. | 3 Figure 1. The Design-Build-Learn-Test cycle. Biofoundries typically focus on the BUILD and TEST aspects of the cycle, but this is not a firm rule; a facility might, for ex- ample, focus more on LEARN capabilities by developing data analysis and machine learning methods. the key driver to support national capability across academia current and prospective user base and an iterative development and industry. strategy is required. As the market scale and client needs often For biofoundries without these ongoing support mecha- require parallel development to the physical infrastructure, it is nisms, the key to medium- and long-term sustainability is the critical to start small and grow organically with a biofoundry creation of a core client base and an appropriate service model facility so that the facility meets the need without overcapitalization. (see Section 4 Development Strategies and Client Engagement). Appraisal of the client base should extend beyond the host in- stitute, to the broader national and international research com- 4.1 Services available munities. These strategies require significant client/community The focus of a biofoundry can be narrow or expansive. In prac- engagement. They can also lead to identification of additional tice, different biofoundries tend to specialize in different facets funding sources and alternative funding models. of synthetic biology, focusing, for example, on a specific organ- ism or application area, or on different types of services or foun- dational technologies within the biodesign space (9). The 4. Development Strategies and Client specific niche of a biofoundry should be determined by client Engagement need and available equipment. Central to the design of a biofoundry is establishing the mode(s) Much of the core work of current foundries focuses on ‘build’ of engagement and interaction with clients and collaborators, and ‘test’ capabilities, specifically high-throughput construction be they academic and/or commercial users. In developing ser- of DNA componentry, transformation of cell lines and basic vice models, it is critical to bear in mind that effective engineer- product validation. Facility clients often want to do their own ing involves a clear understanding of user needs/wishes (i.e. analysis (‘test’) work, or want minimal in-house testing done by user-centered design). This said, the cutting-edge nature of bio- the biofoundry. The modularity of synthetic biology workflows foundries means there may still be limited understanding can allow clients to pick and choose when and where to enter within the broader research community of what a biofoundry and leave a service pipeline, and to build custom pipelines from could do for them (see Section 11). Many researchers are un- a set of interchangeable modules (Figure 2). aware that the technology allowing for automated execution of An example of a basic high-throughput workflow could in- key genetic engineering laboratory protocols is already available clude DNA synthesis, construct assembly, bacterial/yeast trans- and can be routinely employed. Ongoing discussion with the formation, colony picking, classification and DNA/RNA 4| Synthetic Biology, 2021, Vol. 6, No. 1 For facilities operating in a cost-recovery model, client en- gagement necessarily includes discussions around the cost per sample/plate/unit. As automation-oriented consumables can be more costly per unit, it can be useful to provide comparative costing (both time and labor) for the same protocols to be con- ducted without automation. Wherever partial costs are absorbed by core biofoundry funding, the client should be made aware of full project cost relative to charged cost. Furthermore, it is imperative to engage clients in discussions about forward planning for biofoundry access, e.g. in grant applications. The necessity of planning for associated costs should not be under- estimated by the service provider or the client. It should be regarded as similar to planning access to other core facilities or services with such as animal houses or microscopy suites. 4.4 Early collaborators In the development of core biofoundry protocols, it can be use- ful to establish a small number of early collaborations that en- able the development of key workflows and standard operating procedures. This period can be used to determine approximate failure rates in processes, to understand the cost and time com- Figure 2. Modular nature of biofoundry services. Services offered by a given bio- foundry can be easily grouped into functional modules and clients may be mitments of different levels of troubleshooting, and to develop allowed to mix and match modules and/or services to fit their needs better. quality control measures. This information can be crucial for managing future client relationships and allowing for realistic recovery. Biofoundries with limited equipment may choose to projections on workflow time frames and associated costs. partner with, or outsource to, other facilities to complement or extend workflows. This could include sequencing, cell sorting, 5. Site Considerations proteomic or metabolomic studies and data analysis. Some of 5.1 Centralized versus distributed model the most-requested testing capabilities are flow cytometry, microplate reading assays, fluorescence-activated cell sorting Biofoundries can be physically centralized or distributed. Most and high-throughput micro-bioreactors. If a given facility has a facilities are currently centralized, that is, all staff and machines specific focus or application area (e.g. strain construction for are in one location/building/room. Most managers see this as an chemical production; large construct building for plant transfor- advantage with respect to managing communication between mation), then the available equipment and workflows devel- researchers, simplifying workflows. On the other hand, there oped should support those applications. As with other aspects are examples of successful distributed biofoundries, sometimes of biofoundry development, the analytical capabilities attached with significant geographical distribution. Advantages of a dis- to a given facility should align with client needs. tributed model are that it can leverage established capability at distinct sites and service a broader cross-section of the research 4.2 Service models community. However, governance issues can arise with distrib- Biofoundry service models are similar to other scientific service uted models if multiple partner organizations have differing governance approaches or processes. Furthermore, having sites delivery platforms and core facilities, in that the model depends on the funding and the client base. Models in use to date offer in different geographical locations may mean having to comply with multiple sets of regulations. In either model, often the bio- differing degrees of access to the facility, ranging from equipment-only access and training programs to full in-house foundry facilities are colocated with or housed within a research institute, or with a local industry. This embedding can facilitate service. Funding models range from full cost recovery (with or without profit) to partial cost recovery to fully subsidized access. access by users/clients and can promote collaborative informa- tion exchange. A further benefit is that it can also lend the bio- For facilities with both academic and industry clients (10), it is common to see a tiered-cost model combining several different foundry the credibility of the host organization. This is cost-recovery levels. Lower recovery rates are typically geared particularly advantageous in the earliest days of operation and toward academic users and higher recovery from industrial/ establishing a client base. commercial clients, using profits to subsidize the academic clients. 5.2 Physical requirements A second site consideration for biofoundries is the physical 4.3 User engagement attributes of the space housing the facility. While a standard Client engagement and relationship management are central to molecular biology lab can be turned into a biofoundry, depend- any service model. Early discussion points in project engage- ing on resource availability we recommend choosing premises ment include what are client ‘must-haves’, ‘like-to-haves’ and with following features: (i) containment infrastructure neces- what is not necessary with respect to project outcomes. At a sary for the microorganism Risk Group (11, 12) along with insti- higher level, current client needs, and projected future client tutional and regulatory authority approval to work with needs should be drivers for longer-term planning and growth. genetically modified organisms (GMOs); (ii) at least 20 benches This includes plans for new equipment acquisition, protocol de- worth of space (for immediate development); (iii) climate con- velopment and staff recruitment. trol/air conditioning; (iv) vacuum and compressed air lines; (v) M.B. Holowko et al. | 5 access to three-phase power supply; (vi) sufficient Ethernet retaining qualified staff can be challenging, not just from a fi- ports for each instrument to be networked; (vii) enough space nancial perspective but also in terms of supporting career aspi- for at least 1.8 m worth of biosafety cabinet (BSC) class 2 space; rations. Building and running a biofoundry requires and (viii) as open a layout as possible. considerable scientific and technical talent, but career paths for staff in academic biofoundries are not yet well established. Working across multiple projects in a service role can make it 6. Personnel Considerations challenging for a young scientist/engineer to carve out the kind Although it might not seem intuitive for a facility focused on of specific identity or publication output typically associated automation, personnel are critical to the success of a biofoun- with a career in academia. (See Hilgartner (15) for a discussion dry. Indeed, it could be argued that staff are a more important of similar challenges faced during the Human Genome Project). investment than automated equipment in order to build know- This said, there continue to be opportunities to contribute to how and ensure continuity in operation over time. There has fundamental knowledge generation around the setup and been a tendency to overlook this in biofoundry establishment, operation of biofoundries as well as for addressing important requiring post-hoc securing of large and long-term funding. scientific questions, as evidenced by several papers in this col- lection (OUP Synthetic Biology; Biofoundry special issue). Recommended steps for biofoundry managers to pursue include 6.1 Key roles developing workforce retention plans to ensure continued ser- On average, a small biofoundry will likely require a team of 5–10 vice delivery, working with individuals to establish career devel- people, with larger facilities and more distributed foundries in- opment plans, striking an appropriate balance between their volving many more. Typical roles in a biofoundry include a service and knowledge generation activities, and promoting op- high-level manager to oversee biofoundry operations and assist portunities to contribute to research articles. with business development, automation specialists with experi- ence in lab robotics, a software engineer with expertise in Laboratory Information Management Systems (LIMS) and data 7. Hardware Considerations management, data scientists for managing data and integrating Although it might not seem immediately obvious, the central ‘learn’ capabilities, technicians to perform experiments, poten- piece of equipment in a biofoundry is arguably the microplate. tially a system integrator to integrate hardware and software Rather than microcentrifuge tubes, which are the core vessels (given that ‘standardized’ biofoundry solutions do not currently in a classical molecular biology lab, standardized SBS footprint exist) and a scientific director (often appointed as a part-time microplates (Standards ANSI/SLAS 1-2004 through ANSI/SLAS 4- position). The exact distribution among managers, researchers/ 2004), with 96, 384 or 1536 wells per plate, are the primary vehi- developers and technicians will vary across facilities, but with cle for samples in a biofoundry. Thus, machines and software the current challenges facing biofoundry setup it is not unusual under consideration for a biofoundry must be microplate com- to see a focus on employing researchers/developers. This distri- patible. This orientation around the microplate is both a techni- bution may change as the focus of a facility shifts from develop- cal consideration and a major shift in experimental design ment to day-to-day operations. The development strategy of a thinking (see Section 11). given biofoundry (see above) will likely necessitate long-term retention of automation, software and data specialists along- 7.1 Liquid handlers side technicians. Ongoing training and staff development path- ways are also factors to consider in the strategy. The first major hardware investment for a prospective biofoun- dry is likely to be in liquid-handling robots. Most procedures performed with manual pipettes can be adapted to automated 6.2 Knowhow and tacit knowledge pipetting or acoustic liquid handlers. There are numerous in- While the prospect of automated biofoundries might conjure up strument options at a wide range of price points; generally a future requiring little human input, any such future is a long speaking, robot size and reliability are a function of price (with way off at best. The embodied knowhow that staff bring to run- diminishing returns). There is a growing push to diversify the ning a biofoundry is far from trivial. This knowhow is manifest development and availability of Open hardware and accompa- in several ways, from understanding how the particular goals nying low-cost automation platforms, which may offer opportu- and biological idiosyncrasies of a project might influence work- nities to develop lower-cost biofoundries (9, 16–19). flow design, to being able to disentangle biological and mechan- The types of protocols being planned for the facility should ical sources of failure when troubleshooting, to orchestrating all factor into purchasing decisions for liquid-handling robots. For the work needed behind-the-scenes to manage what at a dis- instance, if a wide range of protocols are required, then a han- tance might seem like ‘seamless’ workflows (13). In his doctoral dler with a fair degree of adaptability will be required. Certainly, study of automation in a UK biofoundry, Chris Mellingwood this may be the case in the early stages of biofoundry develop- highlighted the ‘amphibious’ skills of foundry operators, need- ment where a single pipetting handler may be used to execute ing to have deep understandings of both ‘wet’ biological and most processes. Depending on the instrument, the handler ‘dry’ robot behaviors (14). With the new and complex configura- adaptability may be conferred by deck accessories or inter- tions of physical, digital and biological elements being devel- changeable parts. Or adaptability may be conferred in the range oped in biofoundries, the ability to straddle ‘wet’ and ‘dry’ of motion and pipetting head selectivity. For instance, a handler domains is a rare and critical form of expertise. may have interchangeable parts on the pipetting head for selec- tion of pipette tips by number of tips or by tip size (pipetting vol- 6.3 Job security and career prospects ume). Alternatively, a handler may have a single pipetting head Ensuring job security for biofoundry staff is critical to the that can vary the number and size of tips selected. As part of smooth operation of a biofoundry, not least because of the sig- this, consider the throughput expected on the instrument. Will nificant knowhow they build up over time (see above). However, it be necessary to use a 96-barrel pipetting head or will lower (8 6| Synthetic Biology, 2021, Vol. 6, No. 1 Figure 3. Mapping of a manual workflow onto an automated one. This figure depicts a typical mapping exercise where a DNA assembly and subsequent bacterial trans- formation is being considered for automation. In the first step, the reagents used in tubes are mapped to a microplate layout. Next, instead of manual pipetting auto- mated dispensing is implemented. For the small volumes of a PCR reaction or DNA assembly, an acoustic liquid handler would be good choice. Next, a thermal cycler with 96-well or 384-well plate format is required. For workflows that include DNA amplification by PCR, a quality control step to assess amplified fragments is included. In low-throughput, manual methods this would be done using standard gel electrophoresis. In an automated setup, this can be done in an automated DNA analyzer in 96-well format. Next, the heat shock reaction is be setup and executed using pipettes in manual workflow, whereas this can be executed by an automated liquid han- dler. Finally, after transformation and overnight growth colonies are picked—using your tool of choice out of a petri dish or using an automated colony picker in the au- tomated alternative. or 12-barrel) or higher (348-barrel) pipetting capacity be opti- Repeating this process mapping exercise for each of the mal? These considerations should be tempered with research planned core workflows can help to prioritize equipment pur- into the types and expense of consumables required for the in- chases. It is also important to consider what basic analytical strument and commercial kits to be used in processes. equipment might be required for initial assessment of the gen- erated outputs. Some biofoundries make lists available of the specific equipment they use (10, 20–22). 7.2 Additional equipment All biofoundry equipment should be procured with potential Treating liquid handlers as the central element of a given work- future automation integration in mind. It is also worth remem- flow, attention can then turn to additional high-throughput bering that equipment expenses are not restricted to the pur- equipment needed to realize specific elements of the DBTL cy- chase price of a machine. Consumables can be costly, and in a cle. For example, a common biofoundry protocol might be the high-throughput biofoundry, they will be used in greater vol- construction of genetically modified Escherichia coli in 96-well umes than in a nonautomated laboratory. Maintenance con- plates. Each step of this process can be mapped and associated tracts for machines can also run as high as 15% of the original with specific pieces of automated hardware. The input parts are purchase cost per annum. Some foundries choose not to pay typically either synthetic DNA with minimal pre-processing re- maintenance costs on individual equipment, and instead in- quired or DNA amplified by PCR from a source template using a clude potential maintenance costs in annual financial planning. microplate-compatible thermal cycler. A liquid handler with cherry-picking capabilities can then be used to combine specific DNA parts with a plasmid backbone, followed by a round in the 8. Software Considerations thermal cycler to ligate the fragments. Organism transforma- Software is critical to running a biofoundry. Key software com- tion and plating can then be done using a liquid handler with ponents include (i) biodesign automation software, for different appropriate heating and shaking capabilities. For colony pick- ing, there are automated colony picking robots (Figure 3). elements of the DBTL cycle, (ii) an LIMS, for sample and M.B. Holowko et al. | 7 Education Client Engagement Social Considerations Regulatory Compliance Outreach Operational Strategy Market Analysis Marketing Strategy Business Case Biofoundry Funding Infrastructure Maintenance Site Considerations Hardware Procurement Infrastructure Resources Standard Operating Procedures Software licencing/development Human Resources Mentors & Advocates Staff Training Staff Recruitment Figure 4. The biofoundry funnel. This diagram outlines how different elements discussed in this guide interact with each other. They have been listed in the two cen- tral columns, reflecting early- (left) and mid- to late-stage (right) activities, noting that there is often need to parallelize and prioritize different parts of the funnel to match the developing situation. workflow tracking, (iii) DNA screening software for biosecurity above). And developing reliable protocols can require funda- considerations and (iv) a web portal for client interaction. mental re-design of experiments to account for reaction volume kinetics and to match equipment capacity (26, 27). Ideally, these four systems should be interlinked. Some biofoundries are opting to design their own custom- It is also important to consider the level of automation desir- able and/or feasible within a given biofoundry. Developing a ized software, requiring a dedicated, in-house development fully automated engineering platform can be expensive, time- team. There is an ever-growing list of software tools to assist consuming to operationalize, and relatively inflexible when it with different elements of the biodesign automation process comes to offering clients access to modular, customizable serv- (23). There is also widespread sharing within the community, ices. A productive approach can be to start small in terms of aiming at ever-important standardization. However, there is both equipment and automation—slowly growing a ‘machine currently no off-the-shelf software that satisfies all require- park’ piece-by-piece as needs arise and moving. In tandem with ments for a typical biofoundry. Furthermore, numerous gaps this, an automation strategy can evolve from discrete islands of have been noted in the integration of existing computer-aided automation, for processes with clear and repeated workflows, design tools (24). Considerable effort is typically required to toward more integrated systems (28). operationalize any tool (or set of tools) in a facility, especially given that much of the freely available software is ‘buggy’ and not reliably maintained. There are also broader questions being 10. Data Access Considerations raised about the suitability of key metaphors currently under- A fully operational biofoundry will produce large amounts of pinning the development of DNA design tools, in particularly data. Robust mechanisms for managing, storing and accessing challenging the idea of DNA as linear text (25). Alongside the re- data thus become critical. Simply understanding how to use finement of existing tools, some foundational reconceptualiza- and format data produced by a foundry can be a daunting task, tion of DNA design tools may be needed to achieve ambitions of particularly if wanting to ensure they conform to FAIR princi- improved rational design. Before committing to a specific soft- ples (findability, accessibility, interoperability and reusability) ware approach in a new biofoundry, we recommend consulting (29). Developing databases, data management tools and gover- with existing biofoundries to determine current state-of-the-art nance structures require bioinformatics and data science exper- and review operational experiences, as these are changing tise. With the projected volumes and quality of data being rapidly. produced, artificial intelligence algorithms are increasingly be- ing promoted to help mine data and improve the ‘Learn’ capa- 9. Automation and Integration Strategy bilities of biofoundry design-build-test cycles (30). Initial funding for and excitement around biofoundries most of- 10.1 Data inputs ten relates to the purchase of high-throughput equipment. However, such equipment can be challenging to install and in- The data required to complete builds will generally be provided tegrate into biofoundry operation and requires staff with con- to the biofoundry by the client or will be available on publicly siderable experience (9). In the first instance, machines must be accessible genetic sequence databases. If physical DNA samples set up for reliable use, which includes tasks like figuring out (and in some jurisdictions, digital sequence data) are provided which brand(s) of consumables to use. Any given machine must by the client, then the onus will likely fall on them (not the bio- be integrated with potentially multiple pieces of software (see foundry) to ensure that the samples were obtained in 8| Synthetic Biology, 2021, Vol. 6, No. 1 accordance with relevant laws, recognizing that countries have biofoundry better tailor its services and workflows to prospec- sovereign authority over their genetic resources and this re- tive users. quirement can be put into service agreements or contracts Beyond individual biofoundries, the GBA, through a working signed with the clients (Convention on Biological Diversity group, has an outreach program focusing on industry engage- (CBD) (31)). These include any import/export rules and domestic ment, policy, safety and security, and the public (5). Two of the genetic resource access and benefit-sharing policies that have aims of the program are to provide impact greater than which been implemented under the UN’s CBD and its supplementary individual biofoundries can achieve, and to increase the aware- ness about the role and importance of biofoundries. Nagoya Protocol (32). Multiple UN forums are currently discussing capturing ‘digi- Furthermore, sharing best practices across biofoundries tal sequence information’ (including genetic sequence data) in through forums like the GBA could prove advantageous, as a the same access and benefit-sharing regime that regulates ac- means of recruiting a broader client base. These efforts are two- cess to physical biological samples (33). This could have a major way: as well as promoting understanding among potential cli- impact on the operation of biofoundries and synthetic biology ents regarding the possibilities enabled by biofoundries, work is more generally (34). Some countries are already regulating ac- undertaken to understand resistance and hesitation to using cess to genetic sequences from biological samples originating in biofoundries among the broader research community. their countries (e.g. India, Malaysia and Kenya), the use of which may attract benefit-sharing obligations (35). These data- 11.2 Training input considerations will be particularly important for any cli- Several academic biofoundries began with the assumption that ents wishing to commercialize downstream products. The cli- once their facilities were up and running, there would be wide- ent should be able to account for the provenance of all genetic spread demand for their services across (and beyond) their resources and associated genetic sequence data used in the home institution. However, in academia, the design and alloca- R&D process. tion of projects do not always map neatly onto the enlarged de- sign space made possible through biofoundries. For example, 10.2 Data outputs the design space for a protein engineering PhD is frequently Ownership of data produced by a biofoundry will largely depend cast as ‘one student, one protein’ when biofoundry capabilities on its service model and legal obligations. If a client engages the can allow for larger and more combinatorial efforts. Including biofoundry collaboratively, then intellectual property considera- training in the design of research experiments and projects tions will likely be negotiated on a case-by-case basis (along should thus be seen as a core service offered by a biofoundry. with permissions, warranties, assignments, licenses and in- This can be achieved in several ways, for example: working demnities) and outlined in an MoU and/or service agreement with academic researchers to design biofoundry experiments between the client and the biofoundry prior to engagement. that are suited to PhD or master’s projects, and designing stu- Data output considerations are also important for academic dent and postdoctoral residency or internship programs to pro- clients not intending to commercialize their research, especially vide basic biofoundry training. These strategies will provide a given synthetic biology community’s tendency toward open base for generational change in approaches to bioengineering. data. They will need to consider whether the data produced by the biofoundry will be published on public access genetic se- 12. Biosafety and Biosecurity Considerations quence databases, noting that there may be restrictions on 12.1 Biosafety some genetic sequence data from genetic resources originating in certain countries (35). Publication of genetic sequences may Biosafety is about preventing ‘accidental interactions between be a prerequisite to publication in academic journals and can dangerous biological agents and other organisms or the envi- also be a requirement of some research funding agencies. ronment’ (37) and can refer to biosafety practices within the lab and/or risk mitigation for modified organisms used outside the 11. Cultural and Training Considerations lab (34). However, it is recognized that risk mitigation is not in of itself sufficient, as risk identification and response to unpre- 11.1 Outreach dicted consequences of genetic mutation (38) is necessary. Biofoundries are at the forefront of a conceptual shift in bioengi- The biosafety risks and associated containment considera- neering design. In developing robust, flexible, high-throughput tions for biofoundries will largely depend on the host organisms capabilities, the very nature of what can be achieved in a single with which the biofoundry intends to work, but may also de- ‘experiment’ is rapidly expanding. While this might be clear to pend on the DNA expression products (e.g. constructs that in- the operators of biofoundries and their current client base, such crease or confer pathogenicity or virulence) (11). The most a conceptual shift has not yet taken root across the life science common host organisms (e.g. lab strains of E. coli, Saccharomyces disciplines that deal with the engineering (genetic, metabolic, cerevisiae and Bacillus subtilis) are generally considered safe to protein and others) community at large (36). To combat this, handle at Biosafety Level 1 (BSL-1). Different jurisdictions will biofoundries need to engage more proactively, not just in mar- have different standards for working with GMOs and different keting their services, but also in the use of educational tools requirements for biosafety accreditation. In Australia, for exam- such as workshops, conferences, lectures and webinars. These ple, biocontainment facilities are accredited by the Office of the can all assist potential clients in determining how a biofoundry Gene Technology Regulator (OGTR), which requires an may facilitate their research. At a more granular level, direct en- Institutional Biosafety Committee (IBC) to oversee the activities gagement with potential clients to discuss their research goals of the facility. and concerns is beneficial. Automation necessarily involves Internationally, the Cartagena Protocol on Biosafety (2000) to some compromises (26) and understanding what compromises the UN’s CBD (31) regulates the safe transfer, handling and use potential clients would be reluctant to make may help a given of living modified organisms across national borders. Even M.B. Holowko et al. | 9 BOX 1 National regulations relating to responsibility for imported biological material—including organisms, cell cultures, and nucleic acids—may affect biofoundry operations. This is particularly relevant where a country does not have a domestic DNA synthesis capability. For example, in Australia, the Federal Government Department of Agriculture biological import permit conditions mandate that the primary importer is responsible for imported materials (https:// www.agriculture.gov.au/import). It is unclear just how far downstream this responsibility endures and what such a responsibility would mean for biofoundries, which operate as an intermediary or ‘middle-person’ (making and perhaps testing, but not deploying, engineered constructs/organisms). This regulatory posi- tion might make a biofoundry responsible for the activities of clients using organisms and genetic constructs derived from imported biological materials. In the face of ambiguous policies, individual biofoundries may have to negotiate their responsibilities with respect to national policies. countries that are not party to the various protocols of the CBD orders or dsDNA under 200 bp and it is unclear to whom suspi- may have implemented related provisions in their national leg- cious order should be reported) (48), client screening is becom- islation. For example, Australia is party to the CBD but not the ing a more valuable biosecurity measure than sequence Cartagena Protocol, however, Australia’s Gene Technology Act screening (34). 2000 (Cth) is considered sufficient to meet the requirements for There is no clear responsibility for biofoundries to assume national implementation of the Cartagena Protocol (39). similar screening activities of potential clients or projects; how- Biofoundries should conduct induction and refresher training ever, there have been calls for every link in the synthetic biology on the international and national rules for safe handling and R&D chain to be involved in biosecurity screening (44). Even transfer of modified organisms, and appropriate labeling and without hard rules or voluntary guidelines for screening poten- documentation when sending or receiving biological resources tial clients, collaborators or projects, biofoundries will need to and GMOs. exercise discretion in determining what services they are pre- It is unlikely that a biofoundry will be directly involved in pared to provide and to whom (e.g. are they willing to service the environmental release of a genetically modified end- DIY Bio practitioners?) to mitigate potential liabilities, especially product. However, should the genetic constructs be incorpo- when working with pathogens or pathogen-derived sequences. rated in a living or nonliving product destined for environmen- Biofoundries may determine their own broad set of standards tal release, the biofoundry may wish to consider the design and from the outset or choose to make such decisions on a case-by- inclusion of intrinsic biosafety measures, such as DNA signa- case basis. Given the speed of technological advancement and tures, barcodes or watermarks coded into the constructs (40). the inability of law and policy to keep up, there are many legal ambiguities in this space which require further characterization 12.2 Biosecurity and research (see e.g. Box 1). Biofoundries should engage with their local authorities to discuss how best to address such In the context of biofoundries, biosecurity refers to the meas- ambiguities. ures taken to reduce the risk of materials and information being used for nefarious purposes and the dual-use research of con- cern (DURC). Basic biosecurity measures should include physi- 13. Conclusions cal limits on who can access the biofoundry, vetting potential Biofoundries are quickly becoming a central element of the syn- employees, compliance with national biosecurity regulations, thetic biology landscape. With their focus on automation, they and conducting DURC awareness training. are at the leading edge of a larger reconceptualization and reor- Biosecurity measures have typically focused on controlling ganization of bioengineering work that stands to make possible access to physical samples of human, plant and animal patho- new kinds of knowledge and engineering practice (49). gens (e.g. through import and export controls) (40). This However, building a biofoundry is a complicated process, re- includes multilateral export control regimes focusing on dual- quiring specialized equipment and expert, dedicated personnel. use items of interest. The Australia Group (41), for instance, is The running costs can be an order of magnitude higher than a an informal association, of more than 40 countries and the EU, standard laboratory, but biofoundry automation has the poten- with the objective of harmonizing export controls on animal (in- tial in principle to increase the speed, reproducibility and reli- cluding human) and plant pathogens and toxins, genetic ele- ability of acquired results by several orders of magnitude. An ments and GMOs, fulfilling obligations under the Biological and established facility can become the heart of a local or even na- Toxin Weapons Convention (42) and Chemical Weapons tional synthetic biology program, delivering fast and reliable Convention (43). data for many projects and research teams and forming the ba- The focus of international and domestic controls has now sis of new economic development. moved beyond the international movement of physical materi- In this article, we have outlined numerous considerations als to screening orders for synthetic DNA. Biofoundries are key and challenges associated with building and operating a bio- sites for making DNA constructs and are major consumers of foundry (Figure 4). As the number of biofoundries continues to synthetic DNA (44). Most commercial DNA synthesis providers grow, the importance of initiatives like the GBA increases. With follow voluntary guidelines (45–47) for screening orders of their focus on sharing best practices, developing standards and double-stranded DNA (37). Despite the increased costs and po- business models, and working through legal, security and cul- tential delay to service delivery times, industry groups chose to tural concerns, such forums can help speed the development adopt screening guidelines early to address fears surrounding the misuse of synthetic DNA and a general lack of government and maturation of the synthetic biology ecosystem. When plan- oversight (48). Given the shortcomings of synthetic DNA order ning a new biofoundry, we strongly recommend engaging with screening (e.g. it is voluntary, may not include oligonucleotide the GBA community and with individual, established 10 | Synthetic Biology, 2021, Vol. 6, No. 1 biofoundries to gain an understanding of current learnings and World Economic Forum. https://www.weforum.org/agenda/ best practice. Furthermore, engagement with national road- 2019/10/biofoundries-the-new-factories-for-genetic-prod maps and broader technological roadmaps (such as the ucts/ (3 December 2020, date last accessed). 9. Jessop-Fabre,M.M. and Sonnenschein,N. (2019) Improving re- Engineering Biology Research Consortium next-generation bio- economy roadmap) (2) should assist with planning and develop- producibility in synthetic biology. Front. Bioeng. Biotechnol., 7, ing a sustainable business model that is appropriately integrated within a broader national science and innovation 10. Chambers,S., Kitney,R. and Freemont,P. (2016) The Foundry: the DNA synthesis and construction Foundry at Imperial strategy. College. Biochem. Soc. Trans., 44, 687–688. 11. World Health Organization. (2004) Laboratory Biosafety Manual, Authors’ Contributions 3rd edn. Geneva. https://apps.who.int/iris/handle/10665/ 42981, (1 July 2020, date last accessed). All authors contributed equally to the development, writing and 12. Standards Australia. (2010) AS/NZS 2243.3:2010: Safety in review of this article. Laboratories Microbiological Safety and Containment. SAI Global under Licence from Standards Australia Limited and Acknowledgments Standards New Zealand (1 July 2020, date last accessed). We would like to thank Jose Carrasco Lopez, Rosalind le 13. Hammang,A. and Frow,E. (2020) Mapping Synthetic Biology Feuvre, Liz Fletcher, Paul Freemont, Nathan Hillson, Vincent Workflows: An Experimental Workshop. Arizona State Martin, David McClymont, Filippo Menolascina, Nicola University. https://blogs.sps.ed.ac.uk/engineering-life/files/ Patron and Hille Tekotte and other members of the Global 2020/02/Hammang-Frow-2019-Mapping-Synthetic-Biology- Biofoundry Alliance for their invaluable insights into inner Workflows-circulated-report.pdf (1 July 2020, date last accessed). workings of their respective biofoundries. 14. Mellingwood,C. (2019) Amphibious researchers: working with laboratory automation in synthetic biology. Doctoral. The Funding University of Edinburgh. 15. Hilgartner,S. (2013) Constituting large-scale biology: building M.B.H. and M.R. acknowledge support from CSIRO’s a regime of governance in the early years of the Human Synthetic Biology Future Science Platform. E.K.F. acknowl- Genome Project. BioSocieties, 8, 397–416. edges research support from the European Research Council 16. Maia Chagas,A. (2018) Haves and have nots must find a better [EC 616510 ENLIFE]. way: the case for open scientific hardware. PLoS Biol., 16, Conflict of interest statement. None declared. e3000014. 17. Gibney,E. (2016) ‘Open-hardware’ pioneers push for low-cost lab kit. Nature, 531, 147–148. References 18. May,M. (2019) A DIY approach to automating your lab. Nature, 1. National Academies of Sciences, Engineering, and Medicine. 569, 587–588. (2020) Safeguarding the Bioeconomy. The National Academies 19. Storch,M., Haines,M.C. and Baldwin,G.S. (2020) DNA-BOT: a Press, Washington, DC. low-cost, automated DNA assembly platform for synthetic bi- 2. Engineering Biology Research Consortium (EBRC). (2019) ology. Synth. Biol., 5, ysaa010. Engineering Biology: A Research Roadmap for the Next-Generation 20. Johnson,J.R., D’Amore,R., Thain,S.C., Craig,T., McCue,H.V., Bioeconomy. https://roadmap.ebrc.org (1 July 2020, date last Hertz-Fowler,C., Hall,N. and Hall,A.J. (2016) GeneMill: a 21st accessed). doi:10.25498/E4159B. century platform for innovation. Biochem. Soc. Trans., 44, 3. Department for Business, Energy & Industrial Strategy. 681–683. Growing the Bioeconomy: A National Bioeconomy Strategy to 2030. 21. Kabisch-Lab. CompuGene Robotic Platform. http://kabisch-lab. UK Government. https://www.gov.uk/government/publica de/robotics.html (2 July 2020, date last accessed). tions/bioeconomy-strategy-2018-to-2030 (1 July 2020, date 22. Biofoundry,L. What the London Biofoundry Can Do for You. last accessed). https://www.londonbiofoundry.org/capabilities (2 July 2020, 4. Gray,P., Meek,S., Griffiths,P., Trapani,J., Small,I., Vickers,C. date last accessed). and Waldby,C. (2018) Synthetic Biology in Australia: An Outlook 23. Appleton,E., Madsen,C., Roehner,N. and Densmore,D. (2017) to 2030. Report for the Australian Council of Learned Design automation in synthetic biology. Cold Spring Harb. Academies (ACOLA). www.acola.org.au (1 July 2020, date last Perspect. Biol., 9, a023978. accessed). 24. Casini,A., Chang,F.Y., Eluere,R., King,A.M., Young,E.M., 5. Global Biofoundries Alliance (GBA). Global Biofoundries Dudley,Q.M., Karim,A., Pratt,K., Bristol,C., Forget,A. et al. Alliance. https://biofoundries.org/ (1 July 2020, date last (2018) A pressure test to make 10 molecules in 90 days: exter- accessed). nal evaluation of methods to engineer biology. J. Am. Chem. 6. Hillson,N., Caddick,M., Cai,Y., Carrasco,J.A., Chang,M.W., Soc., 140, 4302–4316. Curach,N.C., Bell,D.J., Feuvre,R.L., Friedman,D.C., Fu,X. et al. 25. Szymanski,E. and Scher,E. (2019) Models for DNA design (2019) Author correction: building a global alliance of biofoun- tools: the trouble with metaphors is that they don’t go away. dries. Nat. Commun., 10, 3132. ACS Synth. Biol., 8, 2635–2641. 7. Hillson,N., Caddick,M., Cai,Y., Carrasco,J.A., Chang,M.W., 26. McClymont,D.W. and Freemont,P.S. (2017) With all due re- Curach,N.C., Bell,D.J., Le Feuvre,R., Friedman,D.C., Fu,X. et al. spect to Maholo, lab automation isn’t anthropomorphic. Nat. (2019) Building a global alliance of biofoundries. Nat. Biotechnol., 35, 312–314. Commun., 10, 2040. 27. Meckin,R. (2019) Changing infrastructural practices: routine 8. Freemont,P., Curach,N., Friedman,D.C. and Lee,S.Y. (2019) and reproducibility in automated interdisciplinary biosci- These ‘Biofoundries’ Use DNA to Make Natural Products We Need. ence. Sci. Technol. Human Values, 45, 1220–1241. M.B. Holowko et al. | 11 on Escherichia coli K12. Proc. Natl. Acad. Sci. U S A, 95, 28. Alexanian,T. (2019) The Case for Modular Lab Automation. Zymergen Technology Team, Medium, 2020. https://medium. 10449–10452. com/@ZymergenTechBlog/the-case-for-modular-lab-automa 39. Keiper,F. and Atanassova,A. (2020) Regulation of synthetic bi- tion-c34f214e1276 (6 October 2020, date last accessed). ology: developments under the convention on biological di- 29. Wilkinson,M.D., Dumontier,M., Aalbersberg,I.J., Appleton,G., versity and its protocols. Front. Bioeng. Biotechnol., 8, 310. Axton,M., Baak,A., Blomberg,N., Boiten,J.-W., da Silva 40. Gronvall,G.K. (2019) Synthetic biology: biosecurity and bio- Santos,L.B., Bourne,P.E. et al. (2016) The FAIR guiding princi- safety implications. In: S.K.,Singh and J.H.,Kuhn (eds). Defense ples for scientific data management and stewardship. Sci. against Biological Attacks: Volume I. Springer International Data, 3, 160018. Publishing, Cham, pp. 225–232. 30. Dixon,T.A., Curach,N. and Pretorius,I.S. (2020) 41. The Australia Group. Department of Foreign Affairs, Australia, Bio-informational futures: the convergence of artificial intel- https://www.dfat.gov.au/publications/minisite/theaustralia ligence and synthetic biology. EMBO Rep., 21, 1–5. groupnet/site/en/index.html (1 July 2020, date last accessed). 31. Convention on Biological Diversity (CBD). (1992) Convention 42. Biological Weapons Convention (BWC). (1975) Convention on on Biological Diversity. Rio de Janeiro, 5 June 1992 Nations, U., the Prohibition of the Development, Production and Stockpiling of Ch_XXVII_8. United Nations Treaty Series 1760. https://trea Bacteriological (Biological) and Toxin Weapons and on Their ties.un.org/pages/ViewDetails.aspx? src¼TREATY&mtdsg_ Destruction. United Nations Office for Disarmament Affairs, no¼XXVII-8&chapter¼27 (1 July 2020, date last accessed). London, Moscow, Washington. http://disarmament.un.org/ 32. Nagoya,P. (2010) Nagoya Protocol on Access to Genetic Resources treaties/t/bwc (3 December 2020, date last accessed). and the Fair and Equitable Sharing of Benefits Arising from their 43. CWC. (1997) The Convention on the Prohibition of the Development, Utilization to the Convention on Biological Diversity Nations, U., Production, Stockpiling and Use of Chemical Weapons and on their UNEP/CBD/COP/DEC/X/1 of 29 October 2010, Destruction Chemical Weapons Convention. United Nations, C.N.115.2011.TREATIES-7 of 18 March 2011, Nagoya, Japan. Paris. https://www.opcw.org/chemical-weapons-convention https://treaties.un.org/Pages/ViewDetails.aspx? src¼ (3 December 2020, date last accessed). TREATY&mtdsg_no¼XXVII-8-b&chapter¼27&clang¼_en (1 44. Carter,S. and DiEuliis,D. (2019) Mapping the synthetic biology July 2020, date last accessed). industry: implications for biosecurity. Health Secur., 17, 33. Laird,S., Wynberg,R., Rourke,M., Humphries,F., Muller,M.R. 403–406. and Lawson,C. (2020) Rethink the expansion of access and 45. International Gene Synthesis Consortium (IGSC). Harmonized benefit sharing. Science, 367, 1200–1202. Screening Protocol v2.0. Gene Sequence & Customer Screening to 34. Gronvall,G.K. and Carr,P. (2016) Synthetic Biology: Safety, Promote Biosecurity. https://genesynthesisconsortium.org/ (1 Security, and Promise. CreateSpace Independent Publishing July 2020, date last accessed). Platform, Scotts Valley, California, US. 46. International Association Synthetic Biology (IASB). The IASB 35. Bagley,M., Karger,E., Ruiz Muller,M., Perron-Welch,F. and Siva Code of Conduct for Best Practices in Gene Synthesis. http://op. Thambisetty,S. (2020) Fact-finding Study on How Domestic bna.com.s3.amazonaws.com/hl.nsf/r%3FOpen%3Djaqo-7xqpnr Measures Address Benefit-sharing Arising from Commercial and (1 July 2020, date last accessed). Non-commercial Use of Digital Sequence Information on Genetic 47. Department of Health and Human Services (US HHS). (2010) Resources and Address the Use of Digital Sequence Information on Screening Framework Guidance for Providers of Synthetic Genetic Resources for Research and Development. Annex. Double-Stranded DNA, Public Health Service Act, 42 U.S.C. 241, Convention on Biological Diversity, Ad Hoc Technical Expert Section 301; HSPD-10., 75 FR 62820 pp. 62820–62832. https:// Group on Digital Sequence Information on Genetic www.federalregister.gov/documents/2010/10/13/2010-25728/ Resources,. United Nations, Montreal, Canada, CBD/DSI/ screening- framework-guidance-for-providers-of-synthetic- AHTEG/2020/1/5. https://www.cbd.int/doc/c/428d/017b/1b0c6 double-stranded-dna (1 July 2020, date last accessed). 0b47af50c81a1a34d52/dsi-ahteg-2020-01-05-en.pdf (3 48. Bugl,H., Danner,J.P., Molinari,R.J., Mulligan,J.T., Park,H.O., December 2020, date last accessed). Reichert,B., Roth,D.A., Wagner,R., Budowle,B., Scripp,R.M. et 36. Check Hayden,E. (2014) The automated lab. Nature, 516, al. (2007) DNA synthesis and biological security. Nat. 131–132. Biotechnol., 25, 627–629. 37. Go ´ mez-Tatay,L. and Herna ´ ndez-Andreu,J.M. (2019) Biosafety 49. Keating,P., Limoges,C. and Cambrosio,A. (1999) The auto- and biosecurity in Synthetic Biology: a review. Crit. Rev. mated laboratory. In: M.,Fortun and E.,Mendelsohn (eds), The Environ. Sci. Technol., 49, 1587–1621. 38. Ralph,E.T., Guest,J.R. and Green,J. (1998) Altering the anaero- Practices of Human Genetics. Springer Netherlands, Dordrecht, bic transcription factor FNR confers a hemolytic phenotype pp. 125–142. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Synthetic Biology Oxford University Press

Building a biofoundry

Download PDF
11 pages

Loading next page...
 
/lp/oxford-university-press/building-a-biofoundry-F0XIbf95lV

References (64)

Publisher
Oxford University Press
Copyright
© Crown copyright 2020.
eISSN
2397-7000
DOI
10.1093/synbio/ysaa026
Publisher site
See Article on Publisher Site

Abstract

A biofoundry provides automation and analytics infrastructure to support the engineering of biological systems. It allows scientists to perform synthetic biology and aligned experimentation on a high-throughput scale, massively increasing the solution space that can be examined for any given problem or question. However, establishing a biofoundry is a challenging undertaking, with numerous technical and operational considerations that must be addressed. Using collated learnings, here we outline several considerations that should be addressed prior to and during establishment. These include drivers for establishment, institutional models, funding and revenue models, personnel, hardware and software, data manage- ment, interoperability, client engagement and biosecurity issues. The high cost of establishment and operation means that developing a long-term business model for biofoundry sustainability in the context of funding frameworks, actual and po- tential client base, and costing structure is critical. Moreover, since biofoundries are leading a conceptual shift in experi- mental design for bioengineering, sustained outreach and engagement with the research community are needed to grow the client base. Recognition of the significant, long-term financial investment required and an understanding of the com- plexities of operationalization is critical for a sustainable biofoundry venture. To ensure state-of-the-art technology is inte- grated into planning, extensive engagement with existing facilities and community groups, such as the Global Biofoundries Alliance, is recommended. Key words: biofoundry; high-throughput; synthetic biology is considered by most to be essential to developing and growing 1. Introduction synthetic biology capacity. The high-throughput capability Currently, over 40 countries have national strategies relating to afforded by access to a biofoundry can satisfy this goal. the ‘bioeconomy’ (the economic potential of bioscience) and/or A biofoundry is an integrated molecular biology facility that synthetic biology, including the USA (1, 2), the UK (3) and the includes robotic liquid-handling equipment, high-throughput Australian Council of Learned Academies (ACOLA) report (4). In analytical equipment, and the software, personnel and data many of these strategies, the growth of synthetic biology capa- management systems required to run the equipment and bilities is identified as critical to scientific and economic com- broader biofoundry capabilities. Biofoundries marry synthetic petitiveness. A comprehensive, national infrastructure platform biology with automation engineering to create new high- Submitted: 13 July 2020; Received (in revised form): 26 October 2020. ; Accepted: 12 November 2020 V Crown copyright 2020. This Open Access article contains public sector information licensed under the Open Government Licence v2.0 (http://www.nationalarchives.gov.uk/doc/ open-government-licence/version/2/) 1 2| Synthetic Biology, 2021, Vol. 6, No. 1 throughput biological solutions that help to build and collaborators and/or clients, internal and external to one’s de- strengthen a Design-Build-Test-Learn (DBTL) approach to bio- partment or institution, is an important step to determining the logical engineering (Figure 1). viability of a facility. Market saturation, too many biofoundries Biofoundries are gaining popularity around the world, with for a given research community or insufficient client availability academic and commercial facilities being established across in smaller research communities, could result in facility failure North America, Europe and Asia-Pacific regions. A Global and loss of the significant establishment investment. One way Biofoundries Alliance (GBA) (5) for noncommercial biofoundries to avoid such an outcome would be a coordinated funding was launched in 2019, with 16 founding members (6, 7); and has scheme by national granting agencies, preferably as part of their already grown to 27 members in 2020. Core aims of the GBA in- national bioeconomy strategy. clude developing best practices across facilities, sharing of in- formation and resources and enhancing visibility and support 2.2 Scale of investment for these facilities. Biofoundries require significant time and human capital to set Establishing a biofoundry is a significant investment and up and operate. The quantum of funding required to establish requires more than simply setting up a well-equipped physical and support a biofoundry is contingent on the anticipated scale space. The emphasis on high-throughput methods requires concomitant attention to software, protocols and the integra- and reach of the facility, but necessarily extends beyond the tion of physical and digital infrastructures to efficiently prepare purchase of robotic, high-throughput equipment to include con- and track samples. In this respect, biofoundries are at the fore- sumables, software and support for skilled personnel to set up front of a paradigm shift in biological engineering toward a and run the facility. When seeking investment, provisions need more automated, design-focused venture. to be made for each of these elements. Long-term support from In this review, we offer an overview of key technical, organi- one’s home institution, including an understanding of the na- zational and operational issues relating to the setup and run- ture and scale of the venture, is critical. ning of a biofoundry. Our focus is on academic rather than commercial biofoundries, although there is clearly overlap be- 2.3 Nature of experiments tween the two, and at least some of our recommendations Through the use of automation to perform high numbers of re- might apply to commercial foundries. We distinguish between petitive, standardized tasks, biofoundries can dramatically in- academic and commercial biofoundries primarily with respect crease the throughput and design space for biological to their locations, with academic foundries being located in aca- engineering (6–8). Modularization of workflows allows for a mix demic institutions or government laboratories, and commercial and match approach and introduces flexibility in services of- facilities operating outside of an academic context. There are fered in the DBTL engineering biology cycle (Figure 1). However, broad differences in the target client base, funding and profit some types of experiments are more amenable to high- models, and sustainability/growth objectives of academic and commercial biofoundries. Academic biofoundries typically fo- throughput, automated workflows than others. Research pro- cus on supporting the research community and translational grams that require continual small (or large) adjustments to ex- activities, while commercial biofoundries have more of a focus perimental workflows should consider whether a full on commercial clients and investment return. The recent crea- biofoundry is the most appropriate solution to their needs, or tion of a GBA to specifically promote and support noncommer- whether bringing key, individual pieces of high-throughput cial biofoundries suggests there are common challenges for the equipment, such as liquid handlers, into their workflows might long-term development and sustainability of these facilities. satisfy their requirements. Focusing on bringing in key, individ- Our goal in this review is to share what we have learned with re- ual pieces of equipment can also be a strategy for lower- spect to establishing and running academic biofoundries. resourced settings looking to grow their biological engineering capabilities. 2. Rationale for Establishing a Biofoundry 3. Institutional and Funding Models Biofoundries deliver capabilities that allow for an accelerated approach to synthetic biology research and application. They Biofoundries have often been started with a block of strategic also facilitate development of economically important bioengi- funding from a public sector entity for the purchase of key ro- neered products and organisms. This underpins the strategic botic equipment, often with insufficient reference to long-term and economic drivers for acquiring this capability, and there is sustainability. The initial block funding must be matched with consequently growing interest and activity worldwide in their other sustained sources of funding, because the costs associ- establishment. However, a biofoundry requires significant—and ated with running a biofoundry are high: they require special- ongoing—investment, and so it is important to establish a clear ized staff, expensive equipment and large volumes of justification and a solid business case before proceeding. As consumables, as well as longer-term equipment maintenance part of business case development, the following key issues and upgrade costs. These costs have been singled out as a chal- should be considered: lenge for ensuring sustainability (7). Approaches vary for sourcing funding, and hinge to some ex- 2.1 Anticipated throughput (market demand) tent on national and institutional funding structures and priori- ties. For biofoundries based in individual research groups or The equipment found in most biofoundries enables a significant academic institutions, ongoing funding may be part of key in- scale-up in throughput. The throughput capacity of biofoun- frastructure allocations. Activities may be heavily subsidizd, or dries is rarely fully utilized; understanding the need or desire (market demand) is critical in order to ensure long-term, effi- grant-funded, to ensure access by institute members. cient use of the capacity of a biofoundry. This capacity typically Biofoundries with a national outlook may be funded by combi- exceeds the needs of a single laboratory, so identifying nations of public/government funding and grant funding, with M.B. Holowko et al. | 3 Figure 1. The Design-Build-Learn-Test cycle. Biofoundries typically focus on the BUILD and TEST aspects of the cycle, but this is not a firm rule; a facility might, for ex- ample, focus more on LEARN capabilities by developing data analysis and machine learning methods. the key driver to support national capability across academia current and prospective user base and an iterative development and industry. strategy is required. As the market scale and client needs often For biofoundries without these ongoing support mecha- require parallel development to the physical infrastructure, it is nisms, the key to medium- and long-term sustainability is the critical to start small and grow organically with a biofoundry creation of a core client base and an appropriate service model facility so that the facility meets the need without overcapitalization. (see Section 4 Development Strategies and Client Engagement). Appraisal of the client base should extend beyond the host in- stitute, to the broader national and international research com- 4.1 Services available munities. These strategies require significant client/community The focus of a biofoundry can be narrow or expansive. In prac- engagement. They can also lead to identification of additional tice, different biofoundries tend to specialize in different facets funding sources and alternative funding models. of synthetic biology, focusing, for example, on a specific organ- ism or application area, or on different types of services or foun- dational technologies within the biodesign space (9). The 4. Development Strategies and Client specific niche of a biofoundry should be determined by client Engagement need and available equipment. Central to the design of a biofoundry is establishing the mode(s) Much of the core work of current foundries focuses on ‘build’ of engagement and interaction with clients and collaborators, and ‘test’ capabilities, specifically high-throughput construction be they academic and/or commercial users. In developing ser- of DNA componentry, transformation of cell lines and basic vice models, it is critical to bear in mind that effective engineer- product validation. Facility clients often want to do their own ing involves a clear understanding of user needs/wishes (i.e. analysis (‘test’) work, or want minimal in-house testing done by user-centered design). This said, the cutting-edge nature of bio- the biofoundry. The modularity of synthetic biology workflows foundries means there may still be limited understanding can allow clients to pick and choose when and where to enter within the broader research community of what a biofoundry and leave a service pipeline, and to build custom pipelines from could do for them (see Section 11). Many researchers are un- a set of interchangeable modules (Figure 2). aware that the technology allowing for automated execution of An example of a basic high-throughput workflow could in- key genetic engineering laboratory protocols is already available clude DNA synthesis, construct assembly, bacterial/yeast trans- and can be routinely employed. Ongoing discussion with the formation, colony picking, classification and DNA/RNA 4| Synthetic Biology, 2021, Vol. 6, No. 1 For facilities operating in a cost-recovery model, client en- gagement necessarily includes discussions around the cost per sample/plate/unit. As automation-oriented consumables can be more costly per unit, it can be useful to provide comparative costing (both time and labor) for the same protocols to be con- ducted without automation. Wherever partial costs are absorbed by core biofoundry funding, the client should be made aware of full project cost relative to charged cost. Furthermore, it is imperative to engage clients in discussions about forward planning for biofoundry access, e.g. in grant applications. The necessity of planning for associated costs should not be under- estimated by the service provider or the client. It should be regarded as similar to planning access to other core facilities or services with such as animal houses or microscopy suites. 4.4 Early collaborators In the development of core biofoundry protocols, it can be use- ful to establish a small number of early collaborations that en- able the development of key workflows and standard operating procedures. This period can be used to determine approximate failure rates in processes, to understand the cost and time com- Figure 2. Modular nature of biofoundry services. Services offered by a given bio- foundry can be easily grouped into functional modules and clients may be mitments of different levels of troubleshooting, and to develop allowed to mix and match modules and/or services to fit their needs better. quality control measures. This information can be crucial for managing future client relationships and allowing for realistic recovery. Biofoundries with limited equipment may choose to projections on workflow time frames and associated costs. partner with, or outsource to, other facilities to complement or extend workflows. This could include sequencing, cell sorting, 5. Site Considerations proteomic or metabolomic studies and data analysis. Some of 5.1 Centralized versus distributed model the most-requested testing capabilities are flow cytometry, microplate reading assays, fluorescence-activated cell sorting Biofoundries can be physically centralized or distributed. Most and high-throughput micro-bioreactors. If a given facility has a facilities are currently centralized, that is, all staff and machines specific focus or application area (e.g. strain construction for are in one location/building/room. Most managers see this as an chemical production; large construct building for plant transfor- advantage with respect to managing communication between mation), then the available equipment and workflows devel- researchers, simplifying workflows. On the other hand, there oped should support those applications. As with other aspects are examples of successful distributed biofoundries, sometimes of biofoundry development, the analytical capabilities attached with significant geographical distribution. Advantages of a dis- to a given facility should align with client needs. tributed model are that it can leverage established capability at distinct sites and service a broader cross-section of the research 4.2 Service models community. However, governance issues can arise with distrib- Biofoundry service models are similar to other scientific service uted models if multiple partner organizations have differing governance approaches or processes. Furthermore, having sites delivery platforms and core facilities, in that the model depends on the funding and the client base. Models in use to date offer in different geographical locations may mean having to comply with multiple sets of regulations. In either model, often the bio- differing degrees of access to the facility, ranging from equipment-only access and training programs to full in-house foundry facilities are colocated with or housed within a research institute, or with a local industry. This embedding can facilitate service. Funding models range from full cost recovery (with or without profit) to partial cost recovery to fully subsidized access. access by users/clients and can promote collaborative informa- tion exchange. A further benefit is that it can also lend the bio- For facilities with both academic and industry clients (10), it is common to see a tiered-cost model combining several different foundry the credibility of the host organization. This is cost-recovery levels. Lower recovery rates are typically geared particularly advantageous in the earliest days of operation and toward academic users and higher recovery from industrial/ establishing a client base. commercial clients, using profits to subsidize the academic clients. 5.2 Physical requirements A second site consideration for biofoundries is the physical 4.3 User engagement attributes of the space housing the facility. While a standard Client engagement and relationship management are central to molecular biology lab can be turned into a biofoundry, depend- any service model. Early discussion points in project engage- ing on resource availability we recommend choosing premises ment include what are client ‘must-haves’, ‘like-to-haves’ and with following features: (i) containment infrastructure neces- what is not necessary with respect to project outcomes. At a sary for the microorganism Risk Group (11, 12) along with insti- higher level, current client needs, and projected future client tutional and regulatory authority approval to work with needs should be drivers for longer-term planning and growth. genetically modified organisms (GMOs); (ii) at least 20 benches This includes plans for new equipment acquisition, protocol de- worth of space (for immediate development); (iii) climate con- velopment and staff recruitment. trol/air conditioning; (iv) vacuum and compressed air lines; (v) M.B. Holowko et al. | 5 access to three-phase power supply; (vi) sufficient Ethernet retaining qualified staff can be challenging, not just from a fi- ports for each instrument to be networked; (vii) enough space nancial perspective but also in terms of supporting career aspi- for at least 1.8 m worth of biosafety cabinet (BSC) class 2 space; rations. Building and running a biofoundry requires and (viii) as open a layout as possible. considerable scientific and technical talent, but career paths for staff in academic biofoundries are not yet well established. Working across multiple projects in a service role can make it 6. Personnel Considerations challenging for a young scientist/engineer to carve out the kind Although it might not seem intuitive for a facility focused on of specific identity or publication output typically associated automation, personnel are critical to the success of a biofoun- with a career in academia. (See Hilgartner (15) for a discussion dry. Indeed, it could be argued that staff are a more important of similar challenges faced during the Human Genome Project). investment than automated equipment in order to build know- This said, there continue to be opportunities to contribute to how and ensure continuity in operation over time. There has fundamental knowledge generation around the setup and been a tendency to overlook this in biofoundry establishment, operation of biofoundries as well as for addressing important requiring post-hoc securing of large and long-term funding. scientific questions, as evidenced by several papers in this col- lection (OUP Synthetic Biology; Biofoundry special issue). Recommended steps for biofoundry managers to pursue include 6.1 Key roles developing workforce retention plans to ensure continued ser- On average, a small biofoundry will likely require a team of 5–10 vice delivery, working with individuals to establish career devel- people, with larger facilities and more distributed foundries in- opment plans, striking an appropriate balance between their volving many more. Typical roles in a biofoundry include a service and knowledge generation activities, and promoting op- high-level manager to oversee biofoundry operations and assist portunities to contribute to research articles. with business development, automation specialists with experi- ence in lab robotics, a software engineer with expertise in Laboratory Information Management Systems (LIMS) and data 7. Hardware Considerations management, data scientists for managing data and integrating Although it might not seem immediately obvious, the central ‘learn’ capabilities, technicians to perform experiments, poten- piece of equipment in a biofoundry is arguably the microplate. tially a system integrator to integrate hardware and software Rather than microcentrifuge tubes, which are the core vessels (given that ‘standardized’ biofoundry solutions do not currently in a classical molecular biology lab, standardized SBS footprint exist) and a scientific director (often appointed as a part-time microplates (Standards ANSI/SLAS 1-2004 through ANSI/SLAS 4- position). The exact distribution among managers, researchers/ 2004), with 96, 384 or 1536 wells per plate, are the primary vehi- developers and technicians will vary across facilities, but with cle for samples in a biofoundry. Thus, machines and software the current challenges facing biofoundry setup it is not unusual under consideration for a biofoundry must be microplate com- to see a focus on employing researchers/developers. This distri- patible. This orientation around the microplate is both a techni- bution may change as the focus of a facility shifts from develop- cal consideration and a major shift in experimental design ment to day-to-day operations. The development strategy of a thinking (see Section 11). given biofoundry (see above) will likely necessitate long-term retention of automation, software and data specialists along- 7.1 Liquid handlers side technicians. Ongoing training and staff development path- ways are also factors to consider in the strategy. The first major hardware investment for a prospective biofoun- dry is likely to be in liquid-handling robots. Most procedures performed with manual pipettes can be adapted to automated 6.2 Knowhow and tacit knowledge pipetting or acoustic liquid handlers. There are numerous in- While the prospect of automated biofoundries might conjure up strument options at a wide range of price points; generally a future requiring little human input, any such future is a long speaking, robot size and reliability are a function of price (with way off at best. The embodied knowhow that staff bring to run- diminishing returns). There is a growing push to diversify the ning a biofoundry is far from trivial. This knowhow is manifest development and availability of Open hardware and accompa- in several ways, from understanding how the particular goals nying low-cost automation platforms, which may offer opportu- and biological idiosyncrasies of a project might influence work- nities to develop lower-cost biofoundries (9, 16–19). flow design, to being able to disentangle biological and mechan- The types of protocols being planned for the facility should ical sources of failure when troubleshooting, to orchestrating all factor into purchasing decisions for liquid-handling robots. For the work needed behind-the-scenes to manage what at a dis- instance, if a wide range of protocols are required, then a han- tance might seem like ‘seamless’ workflows (13). In his doctoral dler with a fair degree of adaptability will be required. Certainly, study of automation in a UK biofoundry, Chris Mellingwood this may be the case in the early stages of biofoundry develop- highlighted the ‘amphibious’ skills of foundry operators, need- ment where a single pipetting handler may be used to execute ing to have deep understandings of both ‘wet’ biological and most processes. Depending on the instrument, the handler ‘dry’ robot behaviors (14). With the new and complex configura- adaptability may be conferred by deck accessories or inter- tions of physical, digital and biological elements being devel- changeable parts. Or adaptability may be conferred in the range oped in biofoundries, the ability to straddle ‘wet’ and ‘dry’ of motion and pipetting head selectivity. For instance, a handler domains is a rare and critical form of expertise. may have interchangeable parts on the pipetting head for selec- tion of pipette tips by number of tips or by tip size (pipetting vol- 6.3 Job security and career prospects ume). Alternatively, a handler may have a single pipetting head Ensuring job security for biofoundry staff is critical to the that can vary the number and size of tips selected. As part of smooth operation of a biofoundry, not least because of the sig- this, consider the throughput expected on the instrument. Will nificant knowhow they build up over time (see above). However, it be necessary to use a 96-barrel pipetting head or will lower (8 6| Synthetic Biology, 2021, Vol. 6, No. 1 Figure 3. Mapping of a manual workflow onto an automated one. This figure depicts a typical mapping exercise where a DNA assembly and subsequent bacterial trans- formation is being considered for automation. In the first step, the reagents used in tubes are mapped to a microplate layout. Next, instead of manual pipetting auto- mated dispensing is implemented. For the small volumes of a PCR reaction or DNA assembly, an acoustic liquid handler would be good choice. Next, a thermal cycler with 96-well or 384-well plate format is required. For workflows that include DNA amplification by PCR, a quality control step to assess amplified fragments is included. In low-throughput, manual methods this would be done using standard gel electrophoresis. In an automated setup, this can be done in an automated DNA analyzer in 96-well format. Next, the heat shock reaction is be setup and executed using pipettes in manual workflow, whereas this can be executed by an automated liquid han- dler. Finally, after transformation and overnight growth colonies are picked—using your tool of choice out of a petri dish or using an automated colony picker in the au- tomated alternative. or 12-barrel) or higher (348-barrel) pipetting capacity be opti- Repeating this process mapping exercise for each of the mal? These considerations should be tempered with research planned core workflows can help to prioritize equipment pur- into the types and expense of consumables required for the in- chases. It is also important to consider what basic analytical strument and commercial kits to be used in processes. equipment might be required for initial assessment of the gen- erated outputs. Some biofoundries make lists available of the specific equipment they use (10, 20–22). 7.2 Additional equipment All biofoundry equipment should be procured with potential Treating liquid handlers as the central element of a given work- future automation integration in mind. It is also worth remem- flow, attention can then turn to additional high-throughput bering that equipment expenses are not restricted to the pur- equipment needed to realize specific elements of the DBTL cy- chase price of a machine. Consumables can be costly, and in a cle. For example, a common biofoundry protocol might be the high-throughput biofoundry, they will be used in greater vol- construction of genetically modified Escherichia coli in 96-well umes than in a nonautomated laboratory. Maintenance con- plates. Each step of this process can be mapped and associated tracts for machines can also run as high as 15% of the original with specific pieces of automated hardware. The input parts are purchase cost per annum. Some foundries choose not to pay typically either synthetic DNA with minimal pre-processing re- maintenance costs on individual equipment, and instead in- quired or DNA amplified by PCR from a source template using a clude potential maintenance costs in annual financial planning. microplate-compatible thermal cycler. A liquid handler with cherry-picking capabilities can then be used to combine specific DNA parts with a plasmid backbone, followed by a round in the 8. Software Considerations thermal cycler to ligate the fragments. Organism transforma- Software is critical to running a biofoundry. Key software com- tion and plating can then be done using a liquid handler with ponents include (i) biodesign automation software, for different appropriate heating and shaking capabilities. For colony pick- ing, there are automated colony picking robots (Figure 3). elements of the DBTL cycle, (ii) an LIMS, for sample and M.B. Holowko et al. | 7 Education Client Engagement Social Considerations Regulatory Compliance Outreach Operational Strategy Market Analysis Marketing Strategy Business Case Biofoundry Funding Infrastructure Maintenance Site Considerations Hardware Procurement Infrastructure Resources Standard Operating Procedures Software licencing/development Human Resources Mentors & Advocates Staff Training Staff Recruitment Figure 4. The biofoundry funnel. This diagram outlines how different elements discussed in this guide interact with each other. They have been listed in the two cen- tral columns, reflecting early- (left) and mid- to late-stage (right) activities, noting that there is often need to parallelize and prioritize different parts of the funnel to match the developing situation. workflow tracking, (iii) DNA screening software for biosecurity above). And developing reliable protocols can require funda- considerations and (iv) a web portal for client interaction. mental re-design of experiments to account for reaction volume kinetics and to match equipment capacity (26, 27). Ideally, these four systems should be interlinked. Some biofoundries are opting to design their own custom- It is also important to consider the level of automation desir- able and/or feasible within a given biofoundry. Developing a ized software, requiring a dedicated, in-house development fully automated engineering platform can be expensive, time- team. There is an ever-growing list of software tools to assist consuming to operationalize, and relatively inflexible when it with different elements of the biodesign automation process comes to offering clients access to modular, customizable serv- (23). There is also widespread sharing within the community, ices. A productive approach can be to start small in terms of aiming at ever-important standardization. However, there is both equipment and automation—slowly growing a ‘machine currently no off-the-shelf software that satisfies all require- park’ piece-by-piece as needs arise and moving. In tandem with ments for a typical biofoundry. Furthermore, numerous gaps this, an automation strategy can evolve from discrete islands of have been noted in the integration of existing computer-aided automation, for processes with clear and repeated workflows, design tools (24). Considerable effort is typically required to toward more integrated systems (28). operationalize any tool (or set of tools) in a facility, especially given that much of the freely available software is ‘buggy’ and not reliably maintained. There are also broader questions being 10. Data Access Considerations raised about the suitability of key metaphors currently under- A fully operational biofoundry will produce large amounts of pinning the development of DNA design tools, in particularly data. Robust mechanisms for managing, storing and accessing challenging the idea of DNA as linear text (25). Alongside the re- data thus become critical. Simply understanding how to use finement of existing tools, some foundational reconceptualiza- and format data produced by a foundry can be a daunting task, tion of DNA design tools may be needed to achieve ambitions of particularly if wanting to ensure they conform to FAIR princi- improved rational design. Before committing to a specific soft- ples (findability, accessibility, interoperability and reusability) ware approach in a new biofoundry, we recommend consulting (29). Developing databases, data management tools and gover- with existing biofoundries to determine current state-of-the-art nance structures require bioinformatics and data science exper- and review operational experiences, as these are changing tise. With the projected volumes and quality of data being rapidly. produced, artificial intelligence algorithms are increasingly be- ing promoted to help mine data and improve the ‘Learn’ capa- 9. Automation and Integration Strategy bilities of biofoundry design-build-test cycles (30). Initial funding for and excitement around biofoundries most of- 10.1 Data inputs ten relates to the purchase of high-throughput equipment. However, such equipment can be challenging to install and in- The data required to complete builds will generally be provided tegrate into biofoundry operation and requires staff with con- to the biofoundry by the client or will be available on publicly siderable experience (9). In the first instance, machines must be accessible genetic sequence databases. If physical DNA samples set up for reliable use, which includes tasks like figuring out (and in some jurisdictions, digital sequence data) are provided which brand(s) of consumables to use. Any given machine must by the client, then the onus will likely fall on them (not the bio- be integrated with potentially multiple pieces of software (see foundry) to ensure that the samples were obtained in 8| Synthetic Biology, 2021, Vol. 6, No. 1 accordance with relevant laws, recognizing that countries have biofoundry better tailor its services and workflows to prospec- sovereign authority over their genetic resources and this re- tive users. quirement can be put into service agreements or contracts Beyond individual biofoundries, the GBA, through a working signed with the clients (Convention on Biological Diversity group, has an outreach program focusing on industry engage- (CBD) (31)). These include any import/export rules and domestic ment, policy, safety and security, and the public (5). Two of the genetic resource access and benefit-sharing policies that have aims of the program are to provide impact greater than which been implemented under the UN’s CBD and its supplementary individual biofoundries can achieve, and to increase the aware- ness about the role and importance of biofoundries. Nagoya Protocol (32). Multiple UN forums are currently discussing capturing ‘digi- Furthermore, sharing best practices across biofoundries tal sequence information’ (including genetic sequence data) in through forums like the GBA could prove advantageous, as a the same access and benefit-sharing regime that regulates ac- means of recruiting a broader client base. These efforts are two- cess to physical biological samples (33). This could have a major way: as well as promoting understanding among potential cli- impact on the operation of biofoundries and synthetic biology ents regarding the possibilities enabled by biofoundries, work is more generally (34). Some countries are already regulating ac- undertaken to understand resistance and hesitation to using cess to genetic sequences from biological samples originating in biofoundries among the broader research community. their countries (e.g. India, Malaysia and Kenya), the use of which may attract benefit-sharing obligations (35). These data- 11.2 Training input considerations will be particularly important for any cli- Several academic biofoundries began with the assumption that ents wishing to commercialize downstream products. The cli- once their facilities were up and running, there would be wide- ent should be able to account for the provenance of all genetic spread demand for their services across (and beyond) their resources and associated genetic sequence data used in the home institution. However, in academia, the design and alloca- R&D process. tion of projects do not always map neatly onto the enlarged de- sign space made possible through biofoundries. For example, 10.2 Data outputs the design space for a protein engineering PhD is frequently Ownership of data produced by a biofoundry will largely depend cast as ‘one student, one protein’ when biofoundry capabilities on its service model and legal obligations. If a client engages the can allow for larger and more combinatorial efforts. Including biofoundry collaboratively, then intellectual property considera- training in the design of research experiments and projects tions will likely be negotiated on a case-by-case basis (along should thus be seen as a core service offered by a biofoundry. with permissions, warranties, assignments, licenses and in- This can be achieved in several ways, for example: working demnities) and outlined in an MoU and/or service agreement with academic researchers to design biofoundry experiments between the client and the biofoundry prior to engagement. that are suited to PhD or master’s projects, and designing stu- Data output considerations are also important for academic dent and postdoctoral residency or internship programs to pro- clients not intending to commercialize their research, especially vide basic biofoundry training. These strategies will provide a given synthetic biology community’s tendency toward open base for generational change in approaches to bioengineering. data. They will need to consider whether the data produced by the biofoundry will be published on public access genetic se- 12. Biosafety and Biosecurity Considerations quence databases, noting that there may be restrictions on 12.1 Biosafety some genetic sequence data from genetic resources originating in certain countries (35). Publication of genetic sequences may Biosafety is about preventing ‘accidental interactions between be a prerequisite to publication in academic journals and can dangerous biological agents and other organisms or the envi- also be a requirement of some research funding agencies. ronment’ (37) and can refer to biosafety practices within the lab and/or risk mitigation for modified organisms used outside the 11. Cultural and Training Considerations lab (34). However, it is recognized that risk mitigation is not in of itself sufficient, as risk identification and response to unpre- 11.1 Outreach dicted consequences of genetic mutation (38) is necessary. Biofoundries are at the forefront of a conceptual shift in bioengi- The biosafety risks and associated containment considera- neering design. In developing robust, flexible, high-throughput tions for biofoundries will largely depend on the host organisms capabilities, the very nature of what can be achieved in a single with which the biofoundry intends to work, but may also de- ‘experiment’ is rapidly expanding. While this might be clear to pend on the DNA expression products (e.g. constructs that in- the operators of biofoundries and their current client base, such crease or confer pathogenicity or virulence) (11). The most a conceptual shift has not yet taken root across the life science common host organisms (e.g. lab strains of E. coli, Saccharomyces disciplines that deal with the engineering (genetic, metabolic, cerevisiae and Bacillus subtilis) are generally considered safe to protein and others) community at large (36). To combat this, handle at Biosafety Level 1 (BSL-1). Different jurisdictions will biofoundries need to engage more proactively, not just in mar- have different standards for working with GMOs and different keting their services, but also in the use of educational tools requirements for biosafety accreditation. In Australia, for exam- such as workshops, conferences, lectures and webinars. These ple, biocontainment facilities are accredited by the Office of the can all assist potential clients in determining how a biofoundry Gene Technology Regulator (OGTR), which requires an may facilitate their research. At a more granular level, direct en- Institutional Biosafety Committee (IBC) to oversee the activities gagement with potential clients to discuss their research goals of the facility. and concerns is beneficial. Automation necessarily involves Internationally, the Cartagena Protocol on Biosafety (2000) to some compromises (26) and understanding what compromises the UN’s CBD (31) regulates the safe transfer, handling and use potential clients would be reluctant to make may help a given of living modified organisms across national borders. Even M.B. Holowko et al. | 9 BOX 1 National regulations relating to responsibility for imported biological material—including organisms, cell cultures, and nucleic acids—may affect biofoundry operations. This is particularly relevant where a country does not have a domestic DNA synthesis capability. For example, in Australia, the Federal Government Department of Agriculture biological import permit conditions mandate that the primary importer is responsible for imported materials (https:// www.agriculture.gov.au/import). It is unclear just how far downstream this responsibility endures and what such a responsibility would mean for biofoundries, which operate as an intermediary or ‘middle-person’ (making and perhaps testing, but not deploying, engineered constructs/organisms). This regulatory posi- tion might make a biofoundry responsible for the activities of clients using organisms and genetic constructs derived from imported biological materials. In the face of ambiguous policies, individual biofoundries may have to negotiate their responsibilities with respect to national policies. countries that are not party to the various protocols of the CBD orders or dsDNA under 200 bp and it is unclear to whom suspi- may have implemented related provisions in their national leg- cious order should be reported) (48), client screening is becom- islation. For example, Australia is party to the CBD but not the ing a more valuable biosecurity measure than sequence Cartagena Protocol, however, Australia’s Gene Technology Act screening (34). 2000 (Cth) is considered sufficient to meet the requirements for There is no clear responsibility for biofoundries to assume national implementation of the Cartagena Protocol (39). similar screening activities of potential clients or projects; how- Biofoundries should conduct induction and refresher training ever, there have been calls for every link in the synthetic biology on the international and national rules for safe handling and R&D chain to be involved in biosecurity screening (44). Even transfer of modified organisms, and appropriate labeling and without hard rules or voluntary guidelines for screening poten- documentation when sending or receiving biological resources tial clients, collaborators or projects, biofoundries will need to and GMOs. exercise discretion in determining what services they are pre- It is unlikely that a biofoundry will be directly involved in pared to provide and to whom (e.g. are they willing to service the environmental release of a genetically modified end- DIY Bio practitioners?) to mitigate potential liabilities, especially product. However, should the genetic constructs be incorpo- when working with pathogens or pathogen-derived sequences. rated in a living or nonliving product destined for environmen- Biofoundries may determine their own broad set of standards tal release, the biofoundry may wish to consider the design and from the outset or choose to make such decisions on a case-by- inclusion of intrinsic biosafety measures, such as DNA signa- case basis. Given the speed of technological advancement and tures, barcodes or watermarks coded into the constructs (40). the inability of law and policy to keep up, there are many legal ambiguities in this space which require further characterization 12.2 Biosecurity and research (see e.g. Box 1). Biofoundries should engage with their local authorities to discuss how best to address such In the context of biofoundries, biosecurity refers to the meas- ambiguities. ures taken to reduce the risk of materials and information being used for nefarious purposes and the dual-use research of con- cern (DURC). Basic biosecurity measures should include physi- 13. Conclusions cal limits on who can access the biofoundry, vetting potential Biofoundries are quickly becoming a central element of the syn- employees, compliance with national biosecurity regulations, thetic biology landscape. With their focus on automation, they and conducting DURC awareness training. are at the leading edge of a larger reconceptualization and reor- Biosecurity measures have typically focused on controlling ganization of bioengineering work that stands to make possible access to physical samples of human, plant and animal patho- new kinds of knowledge and engineering practice (49). gens (e.g. through import and export controls) (40). This However, building a biofoundry is a complicated process, re- includes multilateral export control regimes focusing on dual- quiring specialized equipment and expert, dedicated personnel. use items of interest. The Australia Group (41), for instance, is The running costs can be an order of magnitude higher than a an informal association, of more than 40 countries and the EU, standard laboratory, but biofoundry automation has the poten- with the objective of harmonizing export controls on animal (in- tial in principle to increase the speed, reproducibility and reli- cluding human) and plant pathogens and toxins, genetic ele- ability of acquired results by several orders of magnitude. An ments and GMOs, fulfilling obligations under the Biological and established facility can become the heart of a local or even na- Toxin Weapons Convention (42) and Chemical Weapons tional synthetic biology program, delivering fast and reliable Convention (43). data for many projects and research teams and forming the ba- The focus of international and domestic controls has now sis of new economic development. moved beyond the international movement of physical materi- In this article, we have outlined numerous considerations als to screening orders for synthetic DNA. Biofoundries are key and challenges associated with building and operating a bio- sites for making DNA constructs and are major consumers of foundry (Figure 4). As the number of biofoundries continues to synthetic DNA (44). Most commercial DNA synthesis providers grow, the importance of initiatives like the GBA increases. With follow voluntary guidelines (45–47) for screening orders of their focus on sharing best practices, developing standards and double-stranded DNA (37). Despite the increased costs and po- business models, and working through legal, security and cul- tential delay to service delivery times, industry groups chose to tural concerns, such forums can help speed the development adopt screening guidelines early to address fears surrounding the misuse of synthetic DNA and a general lack of government and maturation of the synthetic biology ecosystem. When plan- oversight (48). Given the shortcomings of synthetic DNA order ning a new biofoundry, we strongly recommend engaging with screening (e.g. it is voluntary, may not include oligonucleotide the GBA community and with individual, established 10 | Synthetic Biology, 2021, Vol. 6, No. 1 biofoundries to gain an understanding of current learnings and World Economic Forum. https://www.weforum.org/agenda/ best practice. Furthermore, engagement with national road- 2019/10/biofoundries-the-new-factories-for-genetic-prod maps and broader technological roadmaps (such as the ucts/ (3 December 2020, date last accessed). 9. Jessop-Fabre,M.M. and Sonnenschein,N. (2019) Improving re- Engineering Biology Research Consortium next-generation bio- economy roadmap) (2) should assist with planning and develop- producibility in synthetic biology. Front. Bioeng. Biotechnol., 7, ing a sustainable business model that is appropriately integrated within a broader national science and innovation 10. Chambers,S., Kitney,R. and Freemont,P. (2016) The Foundry: the DNA synthesis and construction Foundry at Imperial strategy. College. Biochem. Soc. Trans., 44, 687–688. 11. World Health Organization. (2004) Laboratory Biosafety Manual, Authors’ Contributions 3rd edn. Geneva. https://apps.who.int/iris/handle/10665/ 42981, (1 July 2020, date last accessed). All authors contributed equally to the development, writing and 12. Standards Australia. (2010) AS/NZS 2243.3:2010: Safety in review of this article. Laboratories Microbiological Safety and Containment. SAI Global under Licence from Standards Australia Limited and Acknowledgments Standards New Zealand (1 July 2020, date last accessed). We would like to thank Jose Carrasco Lopez, Rosalind le 13. Hammang,A. and Frow,E. (2020) Mapping Synthetic Biology Feuvre, Liz Fletcher, Paul Freemont, Nathan Hillson, Vincent Workflows: An Experimental Workshop. Arizona State Martin, David McClymont, Filippo Menolascina, Nicola University. https://blogs.sps.ed.ac.uk/engineering-life/files/ Patron and Hille Tekotte and other members of the Global 2020/02/Hammang-Frow-2019-Mapping-Synthetic-Biology- Biofoundry Alliance for their invaluable insights into inner Workflows-circulated-report.pdf (1 July 2020, date last accessed). workings of their respective biofoundries. 14. Mellingwood,C. (2019) Amphibious researchers: working with laboratory automation in synthetic biology. Doctoral. The Funding University of Edinburgh. 15. Hilgartner,S. (2013) Constituting large-scale biology: building M.B.H. and M.R. acknowledge support from CSIRO’s a regime of governance in the early years of the Human Synthetic Biology Future Science Platform. E.K.F. acknowl- Genome Project. BioSocieties, 8, 397–416. edges research support from the European Research Council 16. Maia Chagas,A. (2018) Haves and have nots must find a better [EC 616510 ENLIFE]. way: the case for open scientific hardware. PLoS Biol., 16, Conflict of interest statement. None declared. e3000014. 17. Gibney,E. (2016) ‘Open-hardware’ pioneers push for low-cost lab kit. Nature, 531, 147–148. References 18. May,M. (2019) A DIY approach to automating your lab. Nature, 1. National Academies of Sciences, Engineering, and Medicine. 569, 587–588. (2020) Safeguarding the Bioeconomy. The National Academies 19. Storch,M., Haines,M.C. and Baldwin,G.S. (2020) DNA-BOT: a Press, Washington, DC. low-cost, automated DNA assembly platform for synthetic bi- 2. Engineering Biology Research Consortium (EBRC). (2019) ology. Synth. Biol., 5, ysaa010. Engineering Biology: A Research Roadmap for the Next-Generation 20. Johnson,J.R., D’Amore,R., Thain,S.C., Craig,T., McCue,H.V., Bioeconomy. https://roadmap.ebrc.org (1 July 2020, date last Hertz-Fowler,C., Hall,N. and Hall,A.J. (2016) GeneMill: a 21st accessed). doi:10.25498/E4159B. century platform for innovation. Biochem. Soc. Trans., 44, 3. Department for Business, Energy & Industrial Strategy. 681–683. Growing the Bioeconomy: A National Bioeconomy Strategy to 2030. 21. Kabisch-Lab. CompuGene Robotic Platform. http://kabisch-lab. UK Government. https://www.gov.uk/government/publica de/robotics.html (2 July 2020, date last accessed). tions/bioeconomy-strategy-2018-to-2030 (1 July 2020, date 22. Biofoundry,L. What the London Biofoundry Can Do for You. last accessed). https://www.londonbiofoundry.org/capabilities (2 July 2020, 4. Gray,P., Meek,S., Griffiths,P., Trapani,J., Small,I., Vickers,C. date last accessed). and Waldby,C. (2018) Synthetic Biology in Australia: An Outlook 23. Appleton,E., Madsen,C., Roehner,N. and Densmore,D. (2017) to 2030. Report for the Australian Council of Learned Design automation in synthetic biology. Cold Spring Harb. Academies (ACOLA). www.acola.org.au (1 July 2020, date last Perspect. Biol., 9, a023978. accessed). 24. Casini,A., Chang,F.Y., Eluere,R., King,A.M., Young,E.M., 5. Global Biofoundries Alliance (GBA). Global Biofoundries Dudley,Q.M., Karim,A., Pratt,K., Bristol,C., Forget,A. et al. Alliance. https://biofoundries.org/ (1 July 2020, date last (2018) A pressure test to make 10 molecules in 90 days: exter- accessed). nal evaluation of methods to engineer biology. J. Am. Chem. 6. Hillson,N., Caddick,M., Cai,Y., Carrasco,J.A., Chang,M.W., Soc., 140, 4302–4316. Curach,N.C., Bell,D.J., Feuvre,R.L., Friedman,D.C., Fu,X. et al. 25. Szymanski,E. and Scher,E. (2019) Models for DNA design (2019) Author correction: building a global alliance of biofoun- tools: the trouble with metaphors is that they don’t go away. dries. Nat. Commun., 10, 3132. ACS Synth. Biol., 8, 2635–2641. 7. Hillson,N., Caddick,M., Cai,Y., Carrasco,J.A., Chang,M.W., 26. McClymont,D.W. and Freemont,P.S. (2017) With all due re- Curach,N.C., Bell,D.J., Le Feuvre,R., Friedman,D.C., Fu,X. et al. spect to Maholo, lab automation isn’t anthropomorphic. Nat. (2019) Building a global alliance of biofoundries. Nat. Biotechnol., 35, 312–314. Commun., 10, 2040. 27. Meckin,R. (2019) Changing infrastructural practices: routine 8. Freemont,P., Curach,N., Friedman,D.C. and Lee,S.Y. (2019) and reproducibility in automated interdisciplinary biosci- These ‘Biofoundries’ Use DNA to Make Natural Products We Need. ence. Sci. Technol. Human Values, 45, 1220–1241. M.B. Holowko et al. | 11 on Escherichia coli K12. Proc. Natl. Acad. Sci. U S A, 95, 28. Alexanian,T. (2019) The Case for Modular Lab Automation. Zymergen Technology Team, Medium, 2020. https://medium. 10449–10452. com/@ZymergenTechBlog/the-case-for-modular-lab-automa 39. Keiper,F. and Atanassova,A. (2020) Regulation of synthetic bi- tion-c34f214e1276 (6 October 2020, date last accessed). ology: developments under the convention on biological di- 29. Wilkinson,M.D., Dumontier,M., Aalbersberg,I.J., Appleton,G., versity and its protocols. Front. Bioeng. Biotechnol., 8, 310. Axton,M., Baak,A., Blomberg,N., Boiten,J.-W., da Silva 40. Gronvall,G.K. (2019) Synthetic biology: biosecurity and bio- Santos,L.B., Bourne,P.E. et al. (2016) The FAIR guiding princi- safety implications. In: S.K.,Singh and J.H.,Kuhn (eds). Defense ples for scientific data management and stewardship. Sci. against Biological Attacks: Volume I. Springer International Data, 3, 160018. Publishing, Cham, pp. 225–232. 30. Dixon,T.A., Curach,N. and Pretorius,I.S. (2020) 41. The Australia Group. Department of Foreign Affairs, Australia, Bio-informational futures: the convergence of artificial intel- https://www.dfat.gov.au/publications/minisite/theaustralia ligence and synthetic biology. EMBO Rep., 21, 1–5. groupnet/site/en/index.html (1 July 2020, date last accessed). 31. Convention on Biological Diversity (CBD). (1992) Convention 42. Biological Weapons Convention (BWC). (1975) Convention on on Biological Diversity. Rio de Janeiro, 5 June 1992 Nations, U., the Prohibition of the Development, Production and Stockpiling of Ch_XXVII_8. United Nations Treaty Series 1760. https://trea Bacteriological (Biological) and Toxin Weapons and on Their ties.un.org/pages/ViewDetails.aspx? src¼TREATY&mtdsg_ Destruction. United Nations Office for Disarmament Affairs, no¼XXVII-8&chapter¼27 (1 July 2020, date last accessed). London, Moscow, Washington. http://disarmament.un.org/ 32. Nagoya,P. (2010) Nagoya Protocol on Access to Genetic Resources treaties/t/bwc (3 December 2020, date last accessed). and the Fair and Equitable Sharing of Benefits Arising from their 43. CWC. (1997) The Convention on the Prohibition of the Development, Utilization to the Convention on Biological Diversity Nations, U., Production, Stockpiling and Use of Chemical Weapons and on their UNEP/CBD/COP/DEC/X/1 of 29 October 2010, Destruction Chemical Weapons Convention. United Nations, C.N.115.2011.TREATIES-7 of 18 March 2011, Nagoya, Japan. Paris. https://www.opcw.org/chemical-weapons-convention https://treaties.un.org/Pages/ViewDetails.aspx? src¼ (3 December 2020, date last accessed). TREATY&mtdsg_no¼XXVII-8-b&chapter¼27&clang¼_en (1 44. Carter,S. and DiEuliis,D. (2019) Mapping the synthetic biology July 2020, date last accessed). industry: implications for biosecurity. Health Secur., 17, 33. Laird,S., Wynberg,R., Rourke,M., Humphries,F., Muller,M.R. 403–406. and Lawson,C. (2020) Rethink the expansion of access and 45. International Gene Synthesis Consortium (IGSC). Harmonized benefit sharing. Science, 367, 1200–1202. Screening Protocol v2.0. Gene Sequence & Customer Screening to 34. Gronvall,G.K. and Carr,P. (2016) Synthetic Biology: Safety, Promote Biosecurity. https://genesynthesisconsortium.org/ (1 Security, and Promise. CreateSpace Independent Publishing July 2020, date last accessed). Platform, Scotts Valley, California, US. 46. International Association Synthetic Biology (IASB). The IASB 35. Bagley,M., Karger,E., Ruiz Muller,M., Perron-Welch,F. and Siva Code of Conduct for Best Practices in Gene Synthesis. http://op. Thambisetty,S. (2020) Fact-finding Study on How Domestic bna.com.s3.amazonaws.com/hl.nsf/r%3FOpen%3Djaqo-7xqpnr Measures Address Benefit-sharing Arising from Commercial and (1 July 2020, date last accessed). Non-commercial Use of Digital Sequence Information on Genetic 47. Department of Health and Human Services (US HHS). (2010) Resources and Address the Use of Digital Sequence Information on Screening Framework Guidance for Providers of Synthetic Genetic Resources for Research and Development. Annex. Double-Stranded DNA, Public Health Service Act, 42 U.S.C. 241, Convention on Biological Diversity, Ad Hoc Technical Expert Section 301; HSPD-10., 75 FR 62820 pp. 62820–62832. https:// Group on Digital Sequence Information on Genetic www.federalregister.gov/documents/2010/10/13/2010-25728/ Resources,. United Nations, Montreal, Canada, CBD/DSI/ screening- framework-guidance-for-providers-of-synthetic- AHTEG/2020/1/5. https://www.cbd.int/doc/c/428d/017b/1b0c6 double-stranded-dna (1 July 2020, date last accessed). 0b47af50c81a1a34d52/dsi-ahteg-2020-01-05-en.pdf (3 48. Bugl,H., Danner,J.P., Molinari,R.J., Mulligan,J.T., Park,H.O., December 2020, date last accessed). Reichert,B., Roth,D.A., Wagner,R., Budowle,B., Scripp,R.M. et 36. Check Hayden,E. (2014) The automated lab. Nature, 516, al. (2007) DNA synthesis and biological security. Nat. 131–132. Biotechnol., 25, 627–629. 37. Go ´ mez-Tatay,L. and Herna ´ ndez-Andreu,J.M. (2019) Biosafety 49. Keating,P., Limoges,C. and Cambrosio,A. (1999) The auto- and biosecurity in Synthetic Biology: a review. Crit. Rev. mated laboratory. In: M.,Fortun and E.,Mendelsohn (eds), The Environ. Sci. Technol., 49, 1587–1621. 38. Ralph,E.T., Guest,J.R. and Green,J. (1998) Altering the anaero- Practices of Human Genetics. Springer Netherlands, Dordrecht, bic transcription factor FNR confers a hemolytic phenotype pp. 125–142.

Journal

Synthetic Biology – Oxford University Press

Published: Dec 16, 2020

Keywords: biofoundry; high-throughput; synthetic biology

There are no references for this article.