Bioindustrial manufacturing readiness levels (BioMRLs) as a shared framework for measuring and communicating the maturity of bioproduct manufacturing processes

Abstract Readiness level (RL) frameworks such as technology readiness levels and manufacturing readiness levels describe the status of a technology/manufacturing process on its journey from initial conception to commercial deployment. More importantly, they provide a roadmap to guide technology development and scale-up from a ‘‘totality of system’’ approach. Commercialization risks associated with too narrowly focused R&D efforts are mitigated. RLs are defined abstractly so that they can apply to diverse industries and technology sectors. However, differences between technology sectors make necessary the definition of sector specific RL frameworks. Here, we describe bioindustrial manufacturing readiness levels (BioMRLs), a classification system specific to bioindustrial manufacturing. BioMRLs will give program managers, investors, scientists, and engineers a shared vocabulary for prioritizing goals and assessing risks in the development and commercialization of a bioindustrial manufacturing process.

Bioindustrial manufacturing processes offer many advantages over strictly chemical processes and are well-positioned to address important global challenges in the coming decades ( Chui et al., 2020 ) . First, biology can be engineered to produce complex materials and substances that have properties unmatched by other manufacturing sectors. This is illustrated by the recent global COVID-19 pandemic. Bioproducts including testing supplies, treatments, and components for vaccines played an important role in the pandemic response. Materials such as polylactic acid are sufficiently robust to allow reallocation toward N95 mask production ( Anon, 2020 ) . On the horizon are more diverse biopolymers, bioplastics, and other bioproducts that have physical or chemical properties absent from the repertoire of products on the market today.
A second advantage of bioindustrial manufacturing is in its ability to reshore supply chains. The feedstocks that support bioindustrial manufacturing are diverse and include agricultural by-products, non-food crop agriculture, industrial waste streams, and gas feedstocks such as CO 2 and methane ( Clomburg et al., 2017 ) . It is conceivable to engineer multiple bioproduction strains that convert distinct feedstocks into the same bioproduct. In this way, fluctuations in feedstock availability or price will have less of an impact on manufacturing output.
Lastly, bioindustrial manufacturing can help mitigate global environmental issues. Pivoting from a petrochemical-based economy to a bio-based economy promises to sequester carbon dioxide from the atmosphere ( Scown & Keasling, 2022 ) , reduce energy requirements for the production and transport of goods, and possibly drive a paradigm change toward smaller, distributed manufacturing facilities ( Clomburg et al., 2017 ) . Innovations in bioindustrial manufacturing will help sustain global ecosystems in the face of population growth by lowering carbon footprints, decreasing pollution, and improving renewable energy utilization efficiency. Even for products that are traditionally sourced from living organisms, like rubber or palm oil, bioindustrial manufacturing offers routes of production that are more sustainable and environmentally friendly.

Bioindustr ial Manufactur ing is Different than Other Manufacturing Sectors
All manufacturing sectors face unique challenges during scaleup. There is a false narrative surrounding bioindustrial manufacturing that the ability of engineered cells to self-replicate eliminates these challenges. Scale-up of a bioindustrial manufacturing process from bench-scale to industrial-scale faces a variety of risks and unknowns, for example, the performance of a fermentation process replicated at production-relevant volumes may exhibit different behavior than predicted from much smaller volumes, depending in part on scale-down experience of the developer ( Delvigne et al., 2017 ;Marques et al., 2010 ;Micheletti et al., 2006 ;Wang et al., 2020 ) . Milliliter-scale fermentation vessels may not accurately reflect strain performance in kiloliter-scale vessels. Many unique challenges exist in downstream processing ( DSP ) . Molecular intermediates and products from fermentations must typically be purified from complex aqueous media. End products of metabolic pathways can share physicochemical properties with shunt metabolites and pathway intermediates that must be purified away using sometimes tedious and expensive separation processes. However, process-guided approaches to strain engineering, coupled with application of downstream processing modeling, can mitigate such issues. For protein-based products purification, workflows often need to maintain the three-dimensional structure throughout DSP/purification so as not to lose activity ( Hearn, 2017 ) . Lastly, because the producing organisms are selfreplicating, maintaining genetic stability is critical to maintaining production quality. The use of scale-down fermentation models can be used to validate genetic stability before going to scale, thus minimizing the potential impact on production scale-up.

New Tools Are Accelerating Bioindustrial Manufacturing Workflows and Lowering R&D Costs for Early Stage Technology Development
The bioeconomy is experiencing rapid growth and has a global impact in excess of $1 trillion ( Chui et al., 2020 ) . Consistent improvement and reduction of costs for DNA synthesis and DNA sequencing are making possible new approaches in early R&D workflows that favor the massively parallel design, construction, and evaluation of hundreds to thousands of putative production strains ( Esvelt et al., 2011 ;Smanski et al., 2014Smanski et al., , 2016Wang et al., 2009 ;Warner et al., 2010 ) . As capabilities for rational genetic design improve, particularly the integration of machine learning capabilities, these libraries will be increasingly focused on highperforming variants ( Lawson et al., 2021 ;Nielsen et al., 2016 ) . New tools and approaches for genome engineering, data analysis, and learning are benefiting the bioindustrial manufacturing community. CRISPR-Cas9 enables rapid and cost-effective generation of genome modifications in virtually any organism that can be transformed with foreign DNA ( Doudna & Charpentier, 2014 ) . dCas9-based tools featuring a deactivated Cas9 enzyme provide control over gene expression and epigenetic state ( Casas-Mollano et al., 2020 ) . Base-and prime-editing tools allow for precise genome manipulation with a lower frequency of undesirable DNA repair modifications ( Anzalone et al., 2020 ) . Large-scale sequencing projects continue to provide new enzyme variants that can be exploited for metabolic engineering. Increased implementation of machine learning and artificial intelligence in biological studies is solving previously intractable problems ( Jumper et al., 2021 ) . Rapid tool advances are providing small and large companies that engineer organisms for bioindustrial manufacturing with greater precision and speed. While these innovations can accelerate ideas to the proof-of-concept stage, continued focus and innovation are needed through the pilot-, intermediate-, and industrial-scale. These innovations might be technical in nature ( e.g., new fermentor designs, new membrane/resin formulations for product recovery ) , or they might solve problems to mitigate business risk ( e.g., supply chain management and quality assurance, multiproduct, or modular manufacturing plants ) . Manufacturing Innovation Institutes ( MIIs ) were established by the U.S. Government to provide the focus and funding needed to advance promising proof-of-concept technologies toward commercial readiness. BioMADE is an MII launched in early 2021 to advance bioindustrial manufacturing of non-medical products, and NIIMBL is an MII launched in 2017 with a focus on manufacturing biomedical products.

Bioindustr ial Manufactur ing Poses Unique Advantages and Faces Distinct Challenges
Bioindustrial manufacturing poses several distinct advantages over conventional methods ( Fig. 1 ) . Diverse metabolic capabilities of engineered host strains lead to multiple feedstocks that can support bioindustrial manufacturing. The fraction of chemical space available to bioindustrial production is large and nonoverlapping with molecules easily produced via synthetic chemistry ( Wetzel et al., 2007 ) . Enantiomerically pure products, often comprising many stereocenters, constrained ring systems, and a range of heteroatom incorporation are commonplace ( Clomburg et al., 2017 ;Erickson & Winters, 2012 ) . Scale-up production processes can be run under relatively mild conditions-typically at moderate temperature, low pressure, at biological pH levels, and with fewer toxic catalysts-resulting in safer, more environmentally benign processes. Perhaps the most unique advantage is the fact that engineered strains are able to grow and reproduce, which changes the economics and thought processes around R&D and industrial scale-up compared to other industries. Fig. 1. Bioindustrial manufacturing utilizes natural or engineered cells, or cell-free systems, to convert feedstocks into manufactured goods, including materials, fuels, fibers, cosmetics, food ingredients, and more. System prototype demonstration in an operational environment TRL8 Actual system completed and qualified through test and demonstration TRL9 Actual system proven through successful mission operations

The Readiness Levels Framework Provides a Shared Vocabulary for Confronting Risk Associated with Technology Development and Manufacturing Scale-up
The technology readiness level ( TRL ) system of classification was developed by NASA in the 1980s to better communicate the maturity status of a new technology ( Table 1 ) . Since then, additional field-specific readiness level frameworks have been implemented ( Buchner et al., 2019 ;Carmack et al., 2017 ;Martínez-Plumed et al., 2021 ) . In the early 2000s, leaders within the Department of Defense spearheaded the development of parallel manufacturing readiness level ( MRL ) classifications to cover elements of the commercialization process missing from the original TRLs. MRLs incorporate a consideration of the production costs and schedules, supply chain robustness, and pressures from competing markets, many of which change as a production process matures from labscale through production-relevant and production-representative environments. There is not a formal link between TRL and MRL scales, but it is not advised to mature the manufacturing readiness further than the technology readiness. Determining the readiness level for a process requires more than a quick look. The chosen level needs to be substantiated by evidence, and this evidence should be limited to demonstrated milestones ( not ongoing or future planned activities ) . As such, it is less useful to describe a process as being ''at'' or ''working on'' MRLx-instead one can describe the technology or manufacturing process as having ''at-tained'', ''demonstrated'', or ''achieved'' an MRL. At the DoD and NASA, technology readiness assessments ( TRAs ) are performed by dedicated teams through a well-defined and rigorous process. For example, a 147-page TRA Guide was produced in 2016 by the U.S. Government Accountability Office ( Government Accountability Office-GAO, 2016 ) .

Translating DoD MRLs to BioMRLs
The path to commercialization for a bioindustrial manufacturing process is sufficiently distinct from other manufacturing sectors to justify the translation of a field-specific framework for BioMRLs. We describe the key criteria for each BioMRL in Table 2 , and below provide more context to the nuanced differences used to distinguish between levels. In drafting this document, we frequently encountered vocabulary that is not common in the bioindustrial manufacturing literature. In these cases, we had to choose between ( i ) replacing vocabulary in the DoD MRLs with terms more familiar to our community or ( ii ) socializing the terms used in the DoD MRLs literature into our community. We did some of each and have included a glossary as Appendix B that should be referenced for key terms. The 10 BioMRLs can be broadly broken up into early, mid-, and late-stage maturity levels ( BioMRL1-3, BioMRL4-7, and BioMRL8-10, respectively ) . Early stage R&D efforts needed to produce the initial proof-of-concept system are encompassed by BioMRL1-3. These MRLs can be satisfied with laboratory-environment experiments and include system-level design ( BioMRL1 ) , biomanufacturing and behavioral characterization of the required component Prior to physical research and development efforts, a study of manufacturing capacity is performed. Criteria include identification and investigation of global trends in the industrial base, manufacturing science, material availability, supply chain, and metrology.

BioMRL2: Manufacturing concepts identified
Key manufacturing concepts have been identified, including broad-based studies that address analysis of material and process approaches, material effects and availability, potential supply chains, needed workforce skillsets, potential future investments, etc. Manufacturing scale and quality requirements for potential markets are identified and analyzed. An understanding of manufacturing feasibility and risk is emerging.

BioMRL3: Manufacturing subsystems or components
Components of the biomanufacturing process have been proven in a laboratory environment. This includes genetic engineering efforts needed to create strains capable of producing the desired products in titers that support the transition to pilot-scale production ( typically in excess of 1 g/L ) . Methods for the purification and analysis of the product of interest are also required but can rely on lab-scale equipment that is not suitable for larger-scale DSP.

BioMRL4 : Independent validation and verification of proof-of-concept
The proof-of-concept system has been demonstrated in a strain suitable for commercial-scale manufacturing and has been independently reproduced/validated/verified. Additionally, an initial assessment of the manufacturability is complete, including preliminary techno-economic analysis ( TEA ) and life-cycle analysis ( LCA ) . This assessment should include plans for the scale-up production ( SUP ) and downstream processing ( DSP ) needed to produce sufficient quantities to allow testing and evaluation by downstream stakeholders. These plans incorporate production-relevant environments. Product quality risks and mitigation plans are documented.

BioMRL5 : Demonstration of prototype unit operations in a production relevant environment
Identification of enabling/critical unit operations is complete. Prototype materials, tooling, and test equipment, as well as personnel skills, have been demonstrated empirically for unit operations in a production-relevant environment. Scale-up production and downstream processing have been performed at suitable scales to deliver sufficient quantities of end-product to downstream stakeholders for testing and evaluation. The TEA has been further refined to assess projected manufacturing costs. A risk management plan to mitigate technical and economic risks is integrated with the manufacturing strategy.
BioMRL6 : Demonstration of a prototype system or subsystem in a production relevant environment Manufacturing processes have been selected for the end-to-end manufacturing pipeline, even if engineering and/or design variables still need to be optimized. Prototype manufacturing processes and technologies, materials, tooling, and test equipment, as well as personnel skills, have been demonstrated on systems and/or subsystems in a production-relevant environment. The TEA is refined based on system performance and is expanded to include inventory control, production scheduling, plant maintenance, and production quality attributes ( PQAs ) . Long-lead and key supply chain elements have been identified, and supply chain risk mitigation strategies exist.
BioMRL7 : Demonstration of systems or subsystems in a production representative environment Detailed system design is complete. Manufacturing processes and procedures have been demonstrated in a production representative environment. Sufficient quantities of product have been made to test packaging and distribution systems. Unit cost reduction strategies, such as statistical process controls ( SPCs ) , are underway in a production representative environment. Quality assurance of supply chains is in place, and procurement schedules for long-lead elements are established. The manufacturing process is sufficient to support low-level commercial manufacturing.

BioMRL8: Manufacturing line demonstrated, ready for low-rate initial production ( LRIP )
This maturity level is associated with manufacturing readiness for entry into LRIP. The detailed system design is complete and sufficiently stable to enter LRIP. All materials, manpower, tooling, test equipment, and facilities are proven on the manufacturing line and are available to meet the planned low-rate production schedule. STE/SIE has been validated in accordance with plans. Manufacturing and quality processes and procedures have been proven and are ready for LRIP. Known technical and business risks pose no significant challenges for LRIP. The cost model and yield and rate analyses have been updated with manufacturing line results. Supplier qualification testing and first article inspections have been completed. The industrial base has been assessed and shows that industrial capability is established to support LRIP.
BioMRL9: Low rate production demonstrated; Capability in place to begin full rate production ( FRP ) Manufacturing has successfully achieved LRIP and is ready to enter FRP. All systems engineering/design requirements have been met such that there are minimal system changes. Major system design features are stable and have been proven in operational tests and evaluations. Materials, parts, manpower, tooling, test equipment, and facilities are available to meet planned rate production schedules. STE/SIE validation is maintained and re-validated as necessary. Manufacturing process capability is at an appropriate quality level to meet customer tolerances. LRIP cost targets have been met. The cost model has been updated for FRP and reflects the impact of continuous improvement.

BioMRL10: FRP demonstrated and lean production practices in place
Engineering/design changes are few and generally limited to continuous improvement changes or obsolescence issues. System, components, and items are in FRP and meet all engineering, performance, quality, and reliability requirements. Manufacturing process capability is at the appropriate quality level. All materials, tooling, inspection and test equipment, facilities, and manpower are in place and have met FRP requirements. Process infrastructure and analytical equipment validation are maintained and re-validated as necessary. Rate production unit costs meet goals, and funding is sufficient for production at the required rates. Continuous process improvements based on risks identified during FRP are ongoing.
parts ( BioMRL2 ) , or initial demonstration of a complete biomanufacturing system in laboratory environments ( BioMRL3 ) . There is a nearly one-to-one correlation between TRLs and MRLs across readiness levels, and at these early stage levels, it is arguable more important to use the TRL designation. MRLs become more important as a manufacturing process increases in scale. Mid-stage manufacturing maturity levels are described by BioMRL4-7. For a process to achieve BioMRL4, the proof-of-concept manufacturing process should have been verified and validated, and a plan for manufacturing at scale should be articulated. Verification and validation should be done in a manner that ensures that it is sufficiently reliable ( and the communication materials describing the process are sufficiently detailed ) to the point that it could be confidently transferred to a new production environment with little impact on performance metrics. BioMRL5 processes have seen process unit operations demonstrated in productionrelevant environments ( Appendix B ) . This is extended in BioMRL6 to demonstrate the ability to carry out all unit operations of a manufacturing system in a production-relevant environment. Because production-relevant environments do not necessarily have all of these unit operations integrated into a single production line, it is not necessary to demonstrate complex scheduling to maximize equipment use for multi-batch production runs at this stage. BioMRL7 requires a shift from production-relevant environments to production-representative environments ( Appendix B ) . At this stage of readiness, all unit operations should be integrated seamlessly, and the aforementioned complex scheduling can be demonstrated.
Late-stage manufacturing maturity levels include BioMRL8-10. In each of these stages, manufacturing is performed in the actual environment that will be used for industrial production. BioMRL8 is achieved when the final manufacturing line has been demonstrated and is ready for low-rate initial production ( LRIP ) . BioMRL9 signifies that LRIP has been demonstrated and that all systems are ready to shift to full rate production ( FRP ) . Finally, BioMRL10 is the most advanced stage of manufacturing readiness. At this level, FRP has been demonstrated, and operational changes to the manufacturing line are limited to continuous improvement changes.

Bioindustr ial Manufactur ing Readiness Assessments ( BioMRAs )
Assessing the maturity of a manufacturing process is done with the help of a detailed criteria matrix that describes the milestones that should be met in 25 business and technology risk categories. These risks can come from a wide range of sources, including material supply chains, manufacturing process designs, capital infrastructure, and access to a well-trained workforce. BioMRAs provide technology developers, regulators, and acquisition managers a shared process for identifying and mitigating risks in advancing manufacturing processes. Thorough and rigorous BioMRAs help to manage and communicate expectations between all stakeholders involved in bringing a product to market.
Proper BioMRAs require a non-trivial dedication of time and resources. BioMRAs should be performed by a team of subject matter experts who can identify both technical and business risks that stand in front of successful product commercialization. The BioM-RLs are not aspirational states but are a set of milestones that can be justified with evidence. Well-executed BioMRAs will seek out this evidence and present it in a detailed report.
The BioMRA criteria matrix developed by this working group is included as Appendix A to this paper. BioMRAs should not be performed with a ''pass/fail'' mentality, and final assessments are expected to be nuanced. For example, the overall summary of a BioMRA could place a manufacturing process at BioMRL5, even if some of the criteria for quality management or facilities have not yet reached BioMRL5 status ( presumably most of the other criteria would be at BioMRL5 or BioMRL6 ) . Identifying the criteria that are lagging behind the general BioMRL highlights vulnerabilities to future process maturation. These vulnerabilities should be addressed before advancing the other criteria to higher maturity levels.
The BioMRA criteria matrix translates the formal criteria for MRAs available via the DoD's website ( dodmrl.com ) into language and examples that are accessible to the community of bioindustrial manufacturing developers, regulators, and funders. Like the BioMRLs, it does not seek to redefine target MRA milestones but merely adds nuance and context to facilitate adoption by bioindustrial manufacturing organizations. We briefly describe the primary criteria categories below that constitute rows in the MRA criteria matrix.

Assessment Categories
These categories make up the rows of the BioMRA criteria matrix. Each category should mature continuously as a manufacturing process advances to higher maturity levels. Categories are presented in the same order as in the DoD's general MRA criteria matrix, and do not represent a listing of priority or importance.

A. Technology and industrial base capabilities:
Requires an analysis of the capability of the National Technology and Industrial Base ( NTIB ) to support the design, development, production, operation, uninterrupted maintenance support, and eventual disposal processes. The NTIB was established by Congress specifically to support national security interests, and currently supports research and development, production, maintenance, and related work in the USA, Canada, the UK, and Australia. This is an important category to consider when the DoD is a possible customer, but if the BioMRLs are used for non-materiel manufacturing processes, this category can be replaced with an assessment of the general industrial base.
B. Design: Relates to the high-level design of manufacturing processes from sourcing materials, assembling unit operations in a SUP and DSP workflow, and identifying manufacturing intensification opportunities. Considerations in this category focus on system-level design and control of key characteristics of the process.

C. Cost and funding:
Requires an analysis of the funding needed to achieve target MRLs and milestones. This category also identifies technical and business risks that might prevent reaching manufacturing cost targets.

D. Materials:
Requires an analysis of the materials used in the manufacturing process, including supply chains for raw materials and the longevity/maintenance risks of materials used for manufacturing equipment and infrastructure.

E. Bioproduction strain:
Requires an analysis of the strain or strains used in the biomanufacturing process, including suitability for industrial biomanufacturing processes, production titers and growth considerations, genetic stability, and rules/regulations/constraints concerning strain safety, shipping, and IP rights.

F. Process capability and control:
Relates to the risks associated with whether manufacturing processes can reach performance levels required by the manufacturing design.

G. Quality management:
Requires an analysis of risks and management efforts to control quality of the final manufactured product and of the incoming supply chains and to ensure that quality is maintained as a process matures to greater manufacturing scales.
H. Manufacturing personnel: Relates to assessments of the workforce to ensure that the required skills, availability, and personnel numbers are accessible to support the manufacturing effort.
I. Facilities: Requires an analysis of the capabilities and capacity of key manufacturing facilities involved in a manufacturing process, including prime manufacturers, subcontractors, suppliers/vendors, and maintenance/repair providers.

J. Manufacturing management:
Requires an analysis of the orchestration and organization of all elements needed for a successful manufacturing process, from design to field/plant integration.
Many of the above categories are further broken down into subcategories in the BioMRA criteria matrix ( Appendix A ) .

Appropriately Scoping the Assessment
When biomanufacturing processes require intersecting technology development, performing an appropriately scoped TRA/MRA can become challenging. For example, imagine a project focused on developing a new analytical sensor for succinic acid that will allow for needed feedback process control for the biomanufacturing of succinic acid. There are two unique MRL maturation paths that could be considered in this case: ( 1 ) how is milestone completion on this project advancing the MRL of succinic acid biomanufacturing ( in which the new sensor is an important process control element and infrastructure element ) or ( 2 ) how is milestone completion on this project advancing the MRL of the new analytical sensor ( which is an important consideration if you need to manufacture/sell this sensor to customers in the biomanufacturing ecosystem ) .
Either MRL maturation path could be important, so it is imperative to be clear at the outset of the MRA to define which path is being assessed. The two paths will not necessarily be at the same MRL at a given time. It is possible to have a fairly advanced process for succinic acid production ( MRL 6/7 ) that does not use the sensor at all. Developing a proof-of-concept sensor and testing it in production-relevant environments ( MRL4/5 for the sensor ) could de-risk process control elements of succinic acid production to merit a step change in its MRL status.
We recommend that users of the BioMRA criteria matrix focus on the business model that is of primary interest to the development team. In the example above, if a team's business model is to develop, manufacture, and sell novel analytical sensors to biomanufacturing companies, then it would be better to focus a rigorous MRA on the sensor itself. Alternatively, if a team's business model is to manufacture succinic acid, and the sensor development is just an important stepping stone to enable reliable manufacturing at larger scales, then succinic acid production should be the focus of the rigorous MRA. In either case, clear communication about what manufacturing process is being assessed is important to avoid confusion.

Interplay Between Scales and Environments
There is language within the criteria matrix below that describes experiments performed in different environments ( laboratory, production-relevant, production-representative, etc. ) . There is an implicit correlation between these environments and the scale at which manufacturing is performed. However, each bioprod-uct is different, and the specific scale ( e.g., in terms of liters of production medium ) that is relevant for these stages in commercial readiness depends on the bioproduct and the scale-up/scaledown expertise of the developer.
Laboratory environments are inherently small-scale and include scaled-down models for production environments. They are able to predict some but not all behaviors of larger scale production runs. Laboratory-scale environments include shake flasks, small-scale ( < 10 L ) bioreactors, and other scale-down minifermentation systems ( e.g., Amber 250 ) . They are commonly used in early stage research and development efforts, including an initial proof-of-concept ( TRL/MRL 3 ) . Laboratory-scale experiments, when coupled with appropriate metrology, can be invaluable for predictably scaling to production-relevant and productionrepresentative environments. Many types of process manipulations that are feasible in laboratory environments are not feasible in production environments.
Production-relevant environments ( BioMRL5/6 ) include pilotscale unit operations that have process control capabilities similar to final production environments. There is no requirement to seamlessly link unit operations in production-relevant environments, and it is common that not all unit operations will be performed in the same facility at this scale. In other words, experiments that accurately test unit operations in isolation are expected at this scale.
Production-representative environments ( BioMRL7 ) must test pilot-to intermediate-scale production runs in a facility that can seamlessly link all of the unit operations for production and DSP that will be required in the eventual commercial-scale production facility. Production-representative environments enable efficiency optimization tied to unit operation scheduling for multiple successive production runs.

Emerging Bioindustrial Platforms
The most mature bioindustrial manufacturing platforms currently are natural or engineered microbial strain fermentations. The language used throughout this paper and the BioMRAs assumes such a platform. However, there are alternative platforms that are maturing rapidly and may feature prominently in the future bioeconomy. For example, engineered plants could be leveraged to achieve an economy of scale-up production not possible with bioreactor-based fermentations ( Mizik & Gyarmati, 2021 ) . Algal, cyanobacterial, or plant-cell culture could be utilized in innovative photosynthetic bioreactor designs to make bioproducts using CO 2 as a feedstock and sunlight as an energy source ( Deprá et al., 2019 ) . Cell-free systems for biomanufacturing have been demonstrated at the proof-of-concept scale and could mature to compete with live-cell-based fermentations for certain products ( Silverman et al., 2020 ) . While the specific language in a BioMRA matrix would have to be tailored for some of these emerging platforms, the general considerations should hold true. For all current and future bioindustrial manufacturing platforms, mitigating manufacturing risk is best done by addressing each of the assessment categories listed above at increasingly relevant manufacturing environments and scales.

Noteworthy Differences Between BioMRLs and Traditional MRLs
There were several modifications made to the standard DoD MRA matrix that we draw attention to here. These changes reflect recognition of the ways in which bioindustrial manufacturing differs from other manufacturing sectors. In most cases, the changes show up as new rows that were added to the BioMRA criteria matrix ( e.g., bioproduction strain, measurement system maturity ) , but in some cases, there is merely a pivot of emphasis within an existing row ( e.g., supply chain managment ) .
Bioproduction strain . The engineered strain that is grown in industrial fermentors to make a bioproduct is critically important in bioindustrial manufacturing. Milestones that pertain to the maturity of the engineered strain did not fit well into the existing criteria rows of the DoD MRA matrix. We created a new category for the bioproduction strain and further subdivided it into two subcategories: ( i ) chassis organism and strain characteristics and ( ii ) strain-environment interface.
Progression from BioMRL1 to BioMRL8, and optionally to BioMRL10 in the chassis organism and strain characteristics row begins by broadly considering potential host organisms that could be engineered to support bioproduction. These are prioritized in an analysis of alternatives ( AoA ) by comparing the technical risk and business risk of adopting each host into the manufacturing process. Advanced design-build-test-learn ( DBTL ) cycles of strain improvement focused on the strain attributes that matter for industrial scale-up are found in the low-to mid-BioMRLs. Mid-to late-BioMRL efforts focus on mitigating strain instability as well as obtaining final regulatory approval for the strain/process.
The second subcategory, strain-environment interface, has many of the same objective functions ( strain growth rate and stability, bioproduct yield, titer, productivity, etc. ) , but focuses on understanding and de-risking the ways that bioproduction environment ( e.g., fermentation medium ) influences these characteristics.
Measurement system maturity . The ability of an experiment to reliably identify changes/improvements in key bioprocess metrics ( titer, yield, productivity, etc. ) is dependent on the number of replicates recorded and the measurement variation. Measurement systems that provide better precision are able to identify smaller changes in key metrics with lower numbers of replicates. Because of this, the efficiency with which an R&D team can capture knowledge throughout the BioMRL escalation ladder is impacted by the quality of measurement systems. This is especially important for mid-and late-BioMRL experiments for two reasons. First, the desired effect size that signals process improvement becomes smaller as a process approaches maximum theoretical yield. Early in strain engineering efforts, when measured titers are only a small fraction of theoretical yield, multiple-fold improvements are possible. Low-precision ( i.e., immature ) measurement systems can be tolerated at this stage, since they are still capable of detecting conditions that lead to several-fold process improvements. However, processes that are already at 50% of theoretical yield cannot give a severalfold improvement with respect to yield alone. Instead, moderate improvements of 2-5% in yield may be the goal for R&D teams, as these can translate to millions of dollars in revenue at target manufacturing scales. High-precision measurement systems that minimize both biological and measurement variance are necessary to detect moderate improvements in bioproduction.
The second reason is that experiments become increasingly more expensive at higher BioMRLs. Advanced measurement systems that allow teams to draw statistically justified conclusions from smaller numbers of replicates allow more rapid and efficient process maturation.
For these reasons, time, effort, and funds dedicated to mitigating risks associated with measurement systems are prioritized throughout the biomanufacturing maturation timeline, and measurement system maturity is included as a new subcategory in the assessment criteria matrix. In some cases, the bioproduct material requirements for performing measurement system maturation studies during early BioMRLs are at odds with the low productivity of manufacturing processes at this stage. Improving access to and affordability of pilot-scale infrastructure to produce the first kilogram of material for testing will benefit the entire field.
Supply chain management . Supply chain management exists as a subcategory in the traditional MRA criteria matrix, but we seek to increase its emphasis in the BioMRA criteria matrix. Feedstock materials that constitute a critical supply chain for bioindustrial manufacturing are subject to many sources of batch variation. For domestic bioindustrial manufacturing, feedstocks may need to be procured from different hemispheres due to seasonality of agriculture. Weather/climate variation from year to year might change important aspects of feedstock quality. For these reasons, establishing a quality management system for incoming feedstocks and other important supply chain materials is arguably more important for bioindustrial manufacturing than for other manufacturing sectors.
Pilot-line terminology . In most cases, we have retained key terms from the DoD MRL Deskbook and translated them for the bioindustrial manufacturing community ( e.g., productionrelevant environment, production-representative environment, etc. ) ( Appendix B ) . However, because the use of the term ''pilotline'' in the DoD MRL literature was commonly misinterpreted by members of our working group, we have elected to remove it from our descriptions of BioMRLs. In the DoD literature, a pilotline environment is the final manufacturing infrastructure that will be used for LRIP and FRP ( i.e., it shows up in MRL8 ) . In contrast, the bioindustrial manufacturing community more commonly associates the terms ''piloting'' or ''pilot-scale'' with smaller-scale experiments that exist between lab-scale and intermediate scale ( i.e., common for BioMRL4-5 processes ) . When we use the term ''pilot'' in this document, we intend to communicate the latter interpretation. We replace the usage of pilot-line in DoD literature with production-line. A production-line environment is established immediately prior to LRIP and FRP.

Perils of R&D in the Absence of a TRL/MRL Framework
The objective functions used to guide DBTL cycles during R&D do not necessarily correspond with commercialization needs. For example, iterative DBTL cycles to improve yields for a product in a chemically defined medium will be wasteful if they do not correspond to higher yields in media suitable for commercial-scale production. In such cases, regular testing of intermediate strains in pilot-scale settings would be important to ensure that R&D of lab-scale strains does not proceed past the point of diminishing returns. Similarly, the final production titer is a convenient metric for lab-scale DBTL efforts, but the value-limiting metric at commercial scale might be the purity of that molecule relative to structurally similar metabolites ( or even the concentration of one specific impurity ) . In such cases, there could be more value in engineering strains to decrease the production of impurities versus pushing further titer increases.
An important consideration that should be addressed early in a TRL/MRL-focused optimization is the specific host organism. Strains for which rapid and easy genetic manipulation tools are available might be ideal for early ( TRL1-3 ) R&D but may not be well suited for the final environment of deployment, whether that is an industrial-scale bioreactor, a human gut, or a farmer's soil. Engineered multi-gene systems are known to vary in their Fig. 2. BioMRLs seek to identify and mitigate risks in scaling manufacturing processes from proof-of-concept to commercial scales. They provide a roadmap for responsible maturation of manufacturing processes.
performance between lab and industrial strains ( Egbert & Klavins, 2012 ;Moser et al., 2012 ) , making this transfer important early in the transition to mid-scale BioMRLs. Changes to genetic design that impact scale-up should also be made in BioMRL4. For example, the impact of plasmid-based versus chromosome-integrated systems, and the use of repetitive genetic elements that decrease genomic stability need to be addressed before advancing too far in the commercialization process ( Chen et al., 2013 ;Englaender et al., 2017 ) . Moving engineered systems into the final host organism early can prevent wasted DBTL effort in a strain that is not commercially viable.
To avoid having new technologies stall in the TRL/MRL4-7 phase, early R&D needs to be done in consideration of the limits and constraints imposed later in the scale-up process ( Fig. 2 ) . This may limit the short-term progress of TRL3-4 technology improvements, which in the absence of oversight could rapidly ''improve'' quickly down routes that in the long-term are dead ends. It is important to take the long view for applied strains and processes in parallel to or soon after demonstrating the first proof-of-concept. This will help de-risk technologies and save money over the entire course of a commercialization effort. Essentially, work with the end in mind.

Conclusion
Bioindustrial manufacturing has special advantages and unique challenges compared to other manufacturing sectors. It can utilize diverse feedstocks, incorporating domestic supply chains to reshore manufacturing processes that have gone overseas. It can make compounds not accessible to bulk chemistry or petrochemical extraction. Producing organisms are self-replicating and can scale to large volumes when process control and measurement systems are sufficiently mature. Challenges may include more complicated DSP/purification requirements and maintaining strain stability and production efficiency. The commercialization process demands that organizations directly confront potential risks to prioritize R&D efforts. The BioMRLs framework provides a shared language for stakeholders to clearly assess the maturity toward or in commercial-scale production. Such a shared framework will help organizations to collaborate in carrying promising technologies across the ''valley of death'' from inception to commercialization. The impact of bioindustrial manufacturing on many sectors of our economy and society will continue to grow. Companies that have successfully shifted from the mindset of ''scaling to learn'' toward one of ''learning to scale'' have brought dozens of products to market in the past two decades ( Hill et al., 2020 ) , and it is exciting to see what the next two decades will bring. Wang, G., Haringa, C., Noorman, H., Chu, J., & Zhuang, Y. ( 2020 ) . Developing a computational framework to advance bioprocess scaleup. Trends in Biotechnology , 38 ( 8 )