Upstream Processing

BPI Contributor

September 1, 2013

19 Min Read

The “Cell Culture and Upstream Processing” track began many years ago as a conference of its own. In October 2004, IBC Life Sciences brought it together in Boston, MA, with three other events (“Recovery and Purification,” “Production and Economics,” and “Scaling Up from Bench to Clinic”) to create the first BPI Conference.

Just like BPI magazine’s editors, the event producers have depended on industry advisors from the beginning. Our editorial advisory board (EAB) members give generously of their time and expertise to help us keep the magazine content relevant, trustworthy, and forward-looking. Similarly, each events’ group of scientific advisors assist the producers in bringing together the best presenters and covering the most timely topics each year.

Some of our EAB members have been with us since the beginning. And some experts have come back to help IBC Life Sciences for two or more yearly events. For the “Cell Culture and Upstream Processing” track, those have included Xuejun “Sherry” Gu (Eli Lilly), Paul Wu (Bayer), Janani Swamy (Genzyme), Ron Taticek (Genentech), Timothy S. Charlebois (Pfizer), Laurel Donahue-Hjelle (Life Technologies), Kevin Kayser (SAFC), Dennis M. Kraichely (Johnson & Johnson), John Mott (Pfizer), Charles Sardonini (Genzyme), and Thomas Seewoester (Amgen). People such as them — as well as loyal readers and attendees, authors and presenters, advertisers and exhibitors — have been key to our success. We couldn’t do what we do without you!

What’s Different: Ten years ago, cell-line engineering was making great strides toward solving the most pressing issue of the day. A looming crunch in bioprocess capacity was brought to light when Immunex’s Enbrel product was approved for market without adequate supply for an unexpectedly robust market. That and other monoclonal antibody (MAb) drugs needed massive infrastructure to manufacture them with expression titers that were often measured in milligrams per liter of cell culture supernatant. Many companies were looking at media optimization to aid in those cell-line engineering efforts. And thanks to news involving transmissible spongiform encephalopathies (TSEs), most were transitioning from animal-sourced raw materials toward chemically defined synthetic ones.

In 2008’s conference guide, Janani Swamy wrote, “Key challenges for the industry are developing comparability approaches and validation paradigms for fast-track implementation, cross-site scale-up and manufacturing, postapproval changes, and the improvement of legacy processes.” That accurately portrays the nature of most discussions we heard at BPI Conference upstream tracks over the first half of the decade.

What Remains the Same: Many of the issues raised above won’t be unfamiliar to attendees at modern BPI events. Optimization of culture media continues. Scale-up is a perennial issue. And single-use technology — which was only starting to make its case in the early years of the BPI Conference — has become a familiar component of production suite design. It has particular utility in smaller-scale processes such as seed train scale-up and manufacturing of products for preclinical studies and clinical trials. Expression titers are now measured routinely in grams per liter, having surpassed milestone after milestone over the past decade. In fact, they led to a new kind of “capacity crunch” in downstream processing — which will be addressed in the next chapter.

Where Do We Go from Here? The 2013 BPI Conference and Exhibition celebrates a decade of such events back on its home turf: in Boston, MA. The upstream track’s tagline this year is “leveraging novel technologies and process innovation to take cell culture productivity significantly higher, faster, and farther while improving understanding, transparency, and flexibility.” Clearly, the process improvement work is far from over. But companies are approaching it more smartly than ever before, with ever more information and technology at hand. We’re looking at cellular metabolism and process control through data feedback. Quality by design (QbD) and process analytical technology (PAT) are changing the way companies approach cell culture process development. Risk management helps them predict and prevent trouble before it arises.

As always, the BPI Conference is all about the people involved in it and the work that they do. IBC’s Jennifer Pereira spoke with several presenters this past summer about their topics as well as experiences with the event over the years. Here, in Q&A format, is what they had to say.

Michael Butler (University of Manitoba, Canada)

Michael Butler, PhD (a professor in the department of microbiology at the University of Manitoba) will be joining us for the “Cell Metabolism and Physiology: Impact on Production” session on Tuesday morning, 17 September 2013. His presentation is titled, “Apoptosis of CHO Cells Monitored by a Novel Dielectrophoretic (DEP) Cytometer,” and it features unpublished data.

What aspects of cell metabolism are most important to consider in biopharmaceutical production? In the use of mammalian cells over the past couple decades, it’s been important to understand cell metabolism for maximizing yields: both of cells and of product. Up until 10–15 years ago, the focus of a lot of work was energy metabolism. Many cells in batch culture tend to have an inefficient metabolism so that their uptake of glucose and glutamine in particular are high. Metabolic by-products such as ammonia and lactate accumulating in media. When that was realized and studied, the value of fed-batch culture became clear. What you try to do is maintain energy substrates at fairly low concentration, which allows for more efficient cell metabolism and limits the amount of ammonia build-up.

In a fed-batch culture, you can get cells to a fairly high density and (most important) keep them at that density for a longer time. During that period (it might be a stationary phase), production is good. It’s efficient and will produce a lot of glycoproteins that we might want. That’s why we’ve advanced from being able to produce maybe 100 mg/L of a product to now 5–10 g/L of product. The major reason is that understanding of energy metabolism and limiting the accumulation of unwanted by-products.

How does DEP cytometry differs from the more familiar flow cytometry? At the University of Manitoba we’ve built this prototype of a novel dielectrophoretic (DEP) cytometer. What it involves is analysis of individual cells channeled through a microfluidic channel. Through that channel the cells (individually in a low-conductivity medium) are passed over three electrodes. Numbers one and three are detection electrodes. The electrode sensor operates with a fairly high-frequency alternating current that doesn’t affect the cells’ polarizability.

Electrode number two is an actuating electrode. We tried it at different signal frequencies: 0.1–6 MHz. At that range, cells can be polarized. That affects their trajectory through the microfluidic channel, so we can define positive and negative dielectrophoretic signals. It depends on vertical deflection: If cells go one way, they are positive; if they go in another direction, they are negative.

If you imagine a sphere passing this microfluidic channel without an actuator, then the signals you get from detectors one and three would be identical. We define that as a force index of zero. However, if a charged particle passes through, it is affected by the actuation potential of the second electrode, causing deflection either upward or downward and thus either a positive or negative force index, which we can measure by the relative area between the detector signals.

Cell Culture and Upstream Processing Sessions

Tuesday, 17 September 2013

8211;11:45 AM  Cell Metabolism and Physiology: Impact on Production (sponsored by Life Technologies)

12:00–1:00 PM  Concurrent Technology Workshops

1:30–3:15 PM  Improving Speed and Quality of Cell Line Development

Wednesday, 18 September 2013

8:00–9:45 AM  Consistency Through Control of Raw Materials and Media Optimization (sponsored by BD Bioscience)

10:30 AM–12:00 PM  Methods and Strategies to Maximize Throughput of Higher Titer Processes and Culture Intensification

12:00–12:30 PM  Concurrent Technology Workshops

1:45–3:30 PM  Implementing Flexible Manufacturing in Upstream Processing

Thursday, 19 September 2013

8:00–10:15 AM  Innovation at the Interface of Upstream and Downstream Processing

11:00 AM–12:00 PM  Beyond Antibodies: Production of New Modalities

12:00–12:30 PM  Concurrent Technology Workshops

1:40–3:15 PM  Control of Process and Product Quality

3:45–5:15 PM  Models for Perfusion and Fed-Batch Processes

What we’ve been doing is to take a cell sample (say 200–300 cells) from a bioreactor and put it into a low-conductivity medium. The cells pass individually through this microfluidic channel, and we analyze the resulting signals. We detect the different signal patterns between viable and nonviable cells, but we can also do more than that. Toward the end of a culture (as cells move from stationary to a decline phase and go from a viable to nonviable state), we can detect subpopulations — at least seven and sometimes more distinguished by different force indices. The force index we can measure for a 100% viable cell population is highly positive, which starts decreasing and goes into negative values as cells become nonviable. To a certain extent, we can correlate those subpopulations with apoptosis measurements, which are measured traditionally using fluorescent stains. We’ve compared readings from the DEP cytometer with those from a fluorescent-based flow cytometer. And we can correlate those subpopulations to a certain extent.

DEP cytometry is essentially a “markless” flow cytometer, meaning that we don’t have to stain cells fluorescently to be able to detect a signal. Essentially, cell polarizability is the critical feature. So we think this type of flow cytometer has a lot of potential. At the moment, it is a prototype in our lab. We’re seeking commercial partners either to develop this further or look into further applications.

What are some other potential applications? We have looked at the potential to detect the points in a bioreactor at which cells are moving from a viable to a nonviable state. We’re looking at the early transition based on the change in electrical properties, comparing single cells using probes that we can put into a bioreactor. Those are commercially available from Fogale and from Aber Instruments.

The DEP cytometer could be used for looking at variant cells within a pool population. We should be able to detect those with different electrical properties. And that could be used in diagnostics, for example. If an aberrant or diseased cell has different electrical properties, then this type of cytometer should be able to detect that.

What has your experience been with BPI Conferences over the years? I’ve been to a lot of IBC conferences, including the BPI Conference. Many of the conferences have been very focused on upstream processing or glycosylation. The BPI Conference is much larger. I think the first one I went to was in 2004, and I had a very good experience there. I enjoy meeting with industrial colleagues and seeing what areas are of interest in developing bioprocesses in the future.

Yongping Crawford (Genentech)

Yongping Crawford, PhD, is a scientist in early stage cell culture at Genentech, Inc. (part of the Roche Group). She will be joining us for the “Improving Speed and Quality of Cell Line Development” session on Tuesday afternoon, 17 September 2013. Her talk is “Fast Identification of Reliable Hosts for Targeted Cell Line Development from a Limited-Genome Screening Using Combined φC31 Integrase and CRE-Lox Technologies.”

What are the advantages and limitations of nonviral vectors? Viral vectors expose cells to a virus, and there might be some undefined risk associated with that. It may become a regulatory concern. With nonviral vectors, you would not have to worry about that risk.

Some viruses are very good at inserting genes into a host genome right at the “hot spot.” Using those to insert the landing pad or platform for future target integration (TI), you would have a better chance to have that inserted into the highly transcriptional active sites. Nonviral vectors do not have as strong a bias of some viruses. So the chance of landing in the transcriptional active side site is smaller. You have to do more screening to compensate for that.

In our study, we used φC31 integrase system to insert the landing or platform. It has been demonstrated that φC31 integrase exhibits some transcriptional active site bias. We performed only a limited genome screening and established two hosts that support high transgene expression. We then used CRE-Lox technology to reliably cassette-exchange our antibody construct to the active site. The hosts we established supported production of diverse antibodies with similar titers. Up to 1 g/L titer can be achieved when we cassette-exchange in one copy of the antibody gene. We can double the titer when we cassette-exchange in two copies of antibody gene.

How does the diversity of antibodies affect product quality, titer, and other development results? We tested diverse IgG1 and IgG4 antibodies using the TI hosts we established. We tested those that are known to be “good expressors” and one antibody known to be a difficult-to-express molecule. Our TI host supported production of diverse antibodies at similar titers regardless of subclass or previous cell line development performance. If this holds true, it will allow better planning and resource allocation. Using the same TI host may reduce host-dependent variations in product quality. But you still have the molecule-specific product qualities.

We tested five diverse antibodies. We have preexisting cell lines for them developed using traditional methods. The product qualities we got from TI cell-line development were not too dissimilar from those developed using traditional methods.

What are some benefits of your new approach? Because all cells selected after cassette exchange are isogenic, there is no need for extensive ELISA screening for high-titer clones. In our study, we simply scaled up 12 randomly chosen clones and found one that produced the same amount of antibodies as the isogenic pool. The timeline and labor requirements can be significantly reduced when using TI cell line development. So throughput can be improved substantially.

Because TI cell lines have only one copy (or very limited copies) of the antibody genes or transgenes inserted into the genome, it is not likely to be subject to clone instability from repeat induced rearrangement or silencing. For quality control, it will be easy to monitor sequence variants. Unlike cell lines with multiple copies of the transgene, sequence variants in a TI cell line would be easily detected at any stage. All these benefits offer companies a competitive advantage. But of course, the titer needs to be sufficient.

What has your experience been with BPI Conferences over the years? It was the first conference I attended when I joined this field. My managers encouraged me to attend because it would provide me with an overview on bioprocessing and expose me to new technologies in the field. It was a great experience, which is why I dec
ided to present our work at BPI this year.

Rajesh Krishnan (Gilead Sciences)

Rajesh Krishnan, PhD (associate director of cell line and upstream process development at Gilead Sciences) will be joining us for the “Single-Use Innovations in Critical and High-Value Applications” session on Thursday morning, 19 September 2013. His case study is titled, “Investigation and Reduction of Performance Variability in Single-use Cell Culture Bioreactors.”

Can you describe some challenges companies face when implementing single-use technology and bioreactors in particular? A bias still may exist for stainless steel over single-use materials for bioreactors. Many companies have previously invested heavily in facilities — in particular, stainless-steel production bioreactors. Switching would be economically challenging because of that preinvestment. Perhaps the prevailing thought is that large-scale manufacturing should be performed using stainless steel bioreactors.

A lot of that distrust comes from the potential risks of leachables and extractables and how those compounds could affect cells and final products. The concern is magnified because of limited sources for generating resins to make single-use materials and a lack of transparency (from the customer perpsective) into how resins are made and quality controls for those processes. What specific types of tests ensure customers that they won’t get compounds either pulled out from the single-use materials or bound to the bags?

Many single-use companies started off primarily making mixing and storage bags. For bioreactors, you want technical expertise to go into the design. Now you’re seeing more of that expertise with single-use manufacturers, but it was initially a big challenge.

What are the suppliers’ responsibilities in customer support? I think the big responsibility is open communication concerning extractables or leachables testing — or other quality control assessments, including stability studies with the materials and experimental conditions used for testing. Although numerous tests can be performed on single-use materials, those tests may not be representative if they are not done at the proper pH — or alternatively a wide range of pH, temperatures, or chemical exposures.

Vaccine Development and Production Sessions

Wednesday, 18 September 2013

8:15 AM–12:00 PM  High-Concentration Vaccine Delivery, Formulation, and Novel Adjuvants

1:45–3:30 PM  Vaccine Manufacturing and Production

Open collaboration between customers and vendors (in designing testing strategies and studies) would allow all of us collectively to tackle many potential issues up front. Customers should be equally collaborative and open with vendors in sharing data regarding product-quality issues and growth or process variability observed with specific single-use systems.

What are some sources of variability in performance? The biggest source of variability — besides potential issues with resins — is the accuracy of process control. How well are you controlling pH, dissolved oxygen, and mixing in these different bag configurations? Process control can be affected by many parameters, including bag design and the type of probes used. That is where I’ve seen the biggest variability when I’ve assessed single-use bioreactors. Product-quality variability with materials made in different single-use systems has not been really that significant in my experience. Those systems also differ in ease of set-up and use (e.g., probe insertion into bags, placement of bags into holders, and so on). That does not really lead to process variability, but it can affect facility fit of the different bioreactors.

What has your experience been with BPI Conferences over the years? I haven’t been to the BPI Conference in several years. I was targeting smaller conferences around particular focus areas, be it preclinical development or scale-up/scale-down. I did get the opportunity to speak at a BPI Conference in China a few years back, which provided me with an excellent look into the future of CMO and CRO activities in Asia. It would be great to keep bringing that perspective from outside the United States into these US meetings. I had not seen detailed discussions of the type of work and services that could be offered out of Asia until I went there.

Barney Zoro (TAP Biosystems)

Barney Zoro (a product manager at TAP Biosystems) will be presenting a technology workshop at lunch on Thursday, 19 September 2013. His presentation is titled, “ambr™: An Advanced Tool for Automated Optimization of Cell Culture for Biotherapeutics.”

Can you describe some challenges that pharmaceutical companies face in optimizing cell culture processes? At the highest level, key challenges remain the same: Do more with less, decrease your timelines, increase the productivity of your process, improve the quality attributes for your protein. Product quality and time-to-market are very critical things to focus on in the biosimilars area, as well, and we’ve heard that quite a bit.

At the laboratory level, that translates into model scalability, consistency and throughput of those scale-down models, and a focus on shake-flask and bench-top bioreactors. Those systems are pretty manually intensive. And that manual factor can affect consistency and error rates in experiments. It also limits the throughput, which sort of stands in the way of shortening project timelines.

The biggest issue with shake flasks is scalability. Shake-flask systems don’t have proper pH or dissolved oxygen (DO) controls. And the shake flask now has been shown in literature to be a relatively poor model of performance in larger reactors for many cases. The implications of that are serious. If you pick the wrong clones with a shake-flask model, you may find yourself having to put greater effort and more time into process development and optimization studies later on. In the worst-case scenario, there could be a lot of rework if your clones fail to scale up — or even major consequences if a delay means that somebody else gets their biosimilar into the market before you. That could be quite significant.

How can automation technologies help address those issues? Automation provides a high consistency in results, low error rates, and high experimental throughput. Decision-making can be improved through increased resolution that we get from the consistency in automation. Also, it frees up a scientist’s time to do more or deeper data analysis, which can give better decision-making results.

Experimental throughput can be maximized by combining automation with single-use technology. You don’t have to do the cleaning, and the robots can do things very quickly. So what might otherwise be quite lengthy sequential experiments or programs (as you typically find with bench-top bioreactors) can be condensed down into much faster parallel programs.

It’s most exciting when automation delivers a new capability — not just taking a manual process and making it faster and more efficient, but really doing something new that you just could not do before. For example, industry-wide adoption of the ambr microbioreactor has been driven largely by the fact that with really good automated pH and DO controls, you have a much better scale-down bioreactor model than the shake flask.

How does the ambr instrument fit into quality by design (QbD)? Ideally, process understanding is a key part of the QbD approach. Although daily bioreactor studies are essential in that approach, they have been pretty difficult or almost impossible to implement in most companies. That’s really down to capacity limitations with bench-top reactors.

Right now, ambr systems are working away in laboratories across the world, opening up bottlenecks around bioreactor processes. The technology has already enabled a broad
range of bioreactor design of experiment (DoE) studies that just couldn’t be done before. We’ve seen base media and feed media development, feeding strategies, and process optimization studies looking at product titer and also biosimilar product quality attributes. People are performing a number of different DoE experiment designs, mixture designs, factorial designs, and other things. DoE can become a routine matter now. Published literature is available on the scaling ambr DoE optimization results to larger reactors, as well. So I think this is a vital technology for enabling bioreactor DoE studies.

What has been your experience at past BPI Conferences? I’ve attended each of the regional BPI Conferences over the past couple of years and given a couple of presentations here and there. I always come away thinking, It was a really great experience; I’m really glad I went. I think the scientists attending the BPI’s really broad technology forum are very interested and engaged with the latest developments, both in supply technologies and the new biology and process improvements at biopharmaceutical companies.

About the Author

Author Details
Listen Online! These interviews have been edited from transcripts for space and style. You can access the original conversations at

You May Also Like