Industrial Biotech: What Does it Really Mean to “Begin with The End In Mind?”

By H. Brett Schreyer, PhD, Lee Enterprises Consulting, Inc.

Special to The Digest

In the industrial fermentation industry, many experts preach “start with the end in mind” with regards to bioprocess development.  What does that exactly mean?  And when do you start thinking about the end?  The short answer is the sooner you start planning for your final process, the better.  As a bio-product progresses through strain screening, bench-scale process development, piloting, and to large scale production, changes to the process become progressively more expensive.  You don’t want to be making significant changes to a plant you just built when production is about to begin.  Such a change could cost on the order of millions to tens of millions of dollars in change orders, equipment, installation, and lost production time.  Let’s dig a little deeper into what it means to “start with the end in mind.”

Start with Techno-Economic Analysis

The most important example of beginning with the end in mind is the work required even before the first stain is made.  This is the techno-economic analysis (TEA) that builds up the business model and justifies the business case to develop a fermentation-based product.  TEA is critical in estimating key performance metrics (e.g. fermentation titer and yield, downstream recovery) necessary for the project to be economically successful and to determine sensitivity to inputs, such as feedstock costs.  The model should be continuously updated as new information comes in.  The net result of TEA drives go/no go decisions and prioritizes research goals that need to be achieved before the process can be scaled up.  There is much information already written on the value of TEA and how to perform it, so it won’t be discussed here.  For more information on TEA, the following articles are good starting points:

1. Chris Burk, “Techno-Economic Modeling for New Technology Development,” CEP Magazine, January, 2018 (https://www.aiche.org/resources/publications/cep/2018/january/techno-economic-modeling-new-technology-development)
2. Daniel A. Lane, “Getting the Most out of Technoeconomic Analyses,” Biofuels Digest, Oct 8, 2018 (https://www.biofuelsdigest.com/bdigest/2018/10/08/getting-the-most-out-of-technoeconomic-analyses/)

What will be discussed in this article is what beginning with the end in mind means at the R&D level.  This will be discussed through three examples:  strain selection, media development, and integration of upstream and downstream processes.

When Selecting a Strain, Choose Wisely

Early stage entrepreneurs in industrial biotech may not even be thinking about the final process when selecting a host strain to engineer as the immediate priority may just be getting proof of concept to secure funding, or the initial stages of strain development may have been completed by an academic institution not focused on commercialization.  So if the host strain has already been locked in, then you’ll need to be ready to deal with the typical traits of that strain at scale because the host strain, such as commonly used Escherichia coli or Saccharomyces cerevisiae, can influence process robustness, repeatability, strain stability, performance metrics (production rate, titer, yield), and operating costs at large scale.

If one truly starts with the end in mind, then one needs to think about the targeted attributes of the large-scale fermentation process when selecting an organism for production.  The rule of thumb is that yeasts (e.g. S. cerevisiae) are more hardy and robust at large scale, but can be more difficult and time consuming to engineer as metabolic pathways are more compartmentalized in eukaryotes and genetic tools and methods may be limited.  Bacteria, on the other hand, are relatively easier to genetically or metabolically engineer and so the iterative process of strain optimization can be faster, but prokaryotes like E. coli tend to be sensitive to fermentation characteristics at large scale (such as non-homogeneity or higher partial pressure of carbon dioxide, or pCO₂) and they are susceptible to phage infection.  Sensitivity to large scale operation can lead to batch variability and not achieving results seen at small scale, while phage infection can lead to lost batches, down time, and lost revenue.  There are many industrial biotech companies that started with E. coli or other common bacteria, but later switched to yeast for some of these reasons.  Such a switch can add months to years to a commercialization timeline.  There are of course other considerations that go into strain selection, such as product toxicity to the host strain, intellectual property, if the product is secreted, or whether a GMO (genetically modified organism) can be used, but the point is that all of these factors need to be considered at the beginning when selecting a strain.

Here is an example.  Let’s say a company wants to develop a bio-based chemical for a growing market and they know there is consumer interest in a renewable, green option to the petrochemical derived product.  The company has performed economic analysis and are confident they can produce it at an equivalent cost and possibly cost advantaged compared to the petrochemical process.  From the economic analysis, in order to stay cost advantaged, operating costs need to be minimized.  To do so, they sketch out a fermentation process that operates at lower temperature to lessen heat removal and that is not highly aerobic to keep cooling requirements low and to minimize aeration and agitation energy consumption.  Further, to decrease fermenter down time and shorten turn-around time, i.e. to make sure the fermenters are up and producing as much as possible, the company places requirements on the strain to be robust so as to maintain batch consistency, has low risk to phage infection, and has a relatively consistent productivity so semi-continuous fermentation is an option (such as draw and fill).  All of these considerations point to the selection of a yeast as the production strain.  The downside is that, depending on the product, it may take longer compared to a bacteria to engineer the yeast with the desired rate, yield, and titer, and the productivity of a yeast strain may not be as high versus bacteria.  The consequence of lower productivity means in order to achieve the same annual production, more capital investment is needed in the size or number of fermenters.  A nice example of the influence of host selection is given in the paper Lau et al. Biotechnology for Biofuels 2010, 3:11 (http://www.biotechnologyforbiofuels.com/content/3/1/11).

Conversely, if time to get to large scale production is critical (due to competitors, market conditions, financial runway, etc.), and the cost margins are not as tight, i.e. TEA indicates the minimum rate of return can still be achieved with relatively higher operating costs (possibly due to a process requiring high aeration and cooling, for example), then selecting a bacteria (like E. coli) may make sense.  Strain modification and optimization will likely be quicker than with a yeast, so achieving key milestones (to secure additional funding, for instance) and getting to scale will probably not take as long.  Once the decision is made to select E. coli as the production strain, then it behooves the biologists and fermentation engineers to develop strategies to deal with the strain’s misgivings; otherwise it may not achieve production targets at scale.  Take phage infection for example.  As part of production strain development, resistance to common phages should be engineered into the strain.  Further, process engineers need to develop a plan on how to respond to a phage infection in the production plant as it will happen.  If these strain and process development activities are incorporated into the schedule from the beginning, then impact on the timeline will be minimal.

Regardless of the strain, testing of sensitivities to variables that may cause issues at large scale needs to be integrated into fermentation process optimization from the beginning of development.  For instance, an E. coli based process may be sensitive to pCO₂, which means titer, rate, and yield will not reproduce at scale where CO₂ concentrations of several percent are common.  If early in development the process is found to be sensitive to pCO₂, then there is time to respond before scale-up.  One option is to engineer the strain to be less sensitive to CO₂, yet this may disrupt the metabolic balance of the production pathway and thus extend the time to develop a scale-ready strain.  Another option is to develop a fermentation process that will minimize pCO₂, such as high aeration to improve CO₂ stripping.  The important point is to select a strain that is more likely to give you the desired fermentation production characteristics at scale and to start testing strains early to scale sensitive parameters to avoid surprises on start-up.  As it is easy to focus on hitting near term goals, it is just as important not to neglect the long-term targets of the production strain to ensure successful and reproducible scale-up.

Media Preparation at Large-Scale can Present Challenges

What about media formulation?  If you start with the end in mind, what does that look like?  Let’s start at the end, large-scale media preparation, and work backwards towards early stage media development.

With large scale fermentation, especially above about 20 m³ fermenters, media is not sterilized in situ (in the vessel) but with high temperature, short time (HTST) heat exchangers.  At higher temperatures, above the laboratory standard of sterilizing at 121°C, the sterilization time can be shortened to minutes or even seconds and still achieve the same log-kill ratio.  This allows for continuous sterilization of large volumes of media in a short amount of time and for rapid turn-around of production fermenters.  Process engineers and scientists need to pay attention to the media recipe to avoid precipitation of salts (a common example is struvite, NH₄MgPO₄) and identify components that can’t be heat sterilized (e.g. vitamins, complex nutrient sources, etc.).  Ideally, other than sugar the recipe has no components susceptible to precipitation or that are damaged by heat.  If a recipe can’t be formulated to avoid precipitation, then the ingredient that causes precipitation has to be filter sterilized or separated from other components during heat sterilization.  Components that are damaged by heat have to be filter sterilized, and any filter sterilized ingredients then have to be aseptically batched into the fermenter.  It is important to note that filter sterilization does not remove phage, increasing risk to a lost batch and lost revenue.  The net effect is that media prep is more complicated which lengthens the turn-around-time (time to prepare the fermenter for the next batch) and increases risk to mistakes and to batch to batch variation.

Another rule of thumb at large scale is to add the minimum amounts of media components necessary to achieve fermentation metrics.  In other words, don’t add media salts in excess that have to then be removed in downstream.  This means most salts are at or near zero concentration at time of harvest.  Over the long run this saves material costs going into the process, but also reduces purification costs.  Again, having a good TEA model will determine the magnitude of such savings.

As the strains in the lab are coming closer to meeting the targeted fermentation metrics, the media recipe should also be becoming closer to what will be used at large scale.  At lab scale this looks like a scaled-down media prep process that does not necessarily have to be run with every bench-scale fermentation, but should be run regularly, such as when a change to the strain or the process is made.  The scaled-down process minimizes the component concentrations and sterilizes as many of the components together as possible in order to replicate large scale sterilization as best as possible.  Ideally only the sugar is sterilized separately.  This also may mean the media may have to be sterilized in a concentrated form as it is unlikely at scale that the media mix tank will be the same volume as the starting volume in the production fermenter, but more likely 1/5^th to 1/10^th the volume.  As we move further back in strain and process development, efforts will include removing any complex media components (e.g. yeast extract) from growth and production medium formulations, removing necessity of antibiotics, and reducing or removing extraneous media components, such as through optimizing key nutrients and trace elements.

To summarize, early in fermentation process development, the fermentation team needs to outline media development through the course of the project.  At key milestones of the project, the team needs to determine when to remove any complex media components, when antibiotics need to be removed from the formulation (if present), and when to prepare and test a scale-ready recipe.  Forethought needs to be given on how to make these changes without influencing strain screening and process optimization.  The sooner these items are resolved at lab scale, the less time and money will be spent solving these issues at pilot scale.  Waiting until production scale to tackle such things makes production runs very expensive experiments and potentially jeopardizes the project.

It’s All One Process

The final example of beginning with the end in mind is integration of upstream and downstream processes.  Although upstream and downstream process development are usually separate groups in a company, they should not be siloed.  From the beginning of development, it is important to envision the entire process of converting raw materials to finished product as a single process, and that fermentation and purification (along with all supportive processes such as logistics, utilities, waste treatment, packaging, etc.) are intricately connected.  For example, as previously mentioned in media development, not reducing the media recipe to the bare minimum means downstream will have to specify filtration or chromatography operations to remove the excess media components, adding to both capital and operating expenses. 

Not long after initial fermentation proof of concept experiments are shown to be successful, downstream process development should begin in earnest and downstream process (DSP) team members should regularly communicate not only with fermentation, but also strain engineering.  As fermentation process optimization can significantly alter the impurity profile and influence downstream design and operation, likewise changes to the strain can also change the impurity profile.  In fact, it is not uncommon for the DSP team to notice the sudden presence of an unknown impurity.  The new impurity is likely concentrated in downstream processing and changes performance of a unit operation and/or product purity.  Through collaborative problem solving, the DSP, fermentation, strain engineering, and analytical teams may discover the impurity is the result of a significant change to the production strain and went unnoticed through strain screening and fermentation.  Just as important, the DSP team may find that a single impurity drives significant capital and equipment costs in purification, such as an extra unit operation or oversized equipment.  This is where DSP, fermentation, and strain engineering groups collaborate to reduce the concentration of the problematic impurity.  This could be solved in fermentation by harvesting sooner, for example, or through strain engineering by optimizing enzymes in a pathway.  Even if fermentation titer and yield take a slight hit, the reduction in downstream capital costs and operating costs may more than make up for it in the overall economics.  This brings us back to the first point made in this article and why having a techno-economic model is so important, i.e. to quickly determine the influence of process changes on the economics of producing the product and building a plant, and to use this information to drive process decisions.

Envision the Final Process

“Beginning with the end in mind” is a methodology when developing a process for large scale production.  At each step of the process, scientists and engineers developing the strain, the fermentation process, and the purification process need to continuously be asking themselves, “What does this look like at production scale?”  By planning ahead and anticipating problems, a company that “begins with the end in mind” will have a more successful, less expensive, and quicker scale-up to production.

About the Author

Dr. Schreyer is an expert with Lee Enterprises Consulting, Inc.   He has a Ph.D. in Chemical Engineering from the University of Connecticut and over 15 years’ experience in microbial fermentation process development, tech transfer, media formulation, scale-up/scale-down, process and techno-economic modeling, and project management.  He has served as Senior Fermentation Engineer for Process Development and Tech Transfer and Scale Up for several large fermentation companies and has taken projects from the lab, thru pilot and full commercialization.  Brett is excellent at directing projects, budget preparation, technical lead in writing service agreements, on-site management of tech transfer and start-up at fermentation and downstream facilities, and in guiding clients in screening strains, developing media, and optimizing and characterizing fermentation processes at bench scale in preparation for scale-up.