Mammalian cell culture is and will remain the principle vehicle for protein production in upstream manufacturing, according to Florian Wurm, Ph.D., professor of biochemistry at the University of Lausanne and founder of ExcellGene. Dr. Wurm reported that 70% of the $20-billion to $30-billion annual bioproducts inventory is generated with mammalian cells, and the gap separating them from other approaches (yeast, bacteria, baculovirus, transgenic plants, and animals) is only growing wider. “We anticipate that mammalian cells will continue to dominate bioproduction for the next 10 years as new technologies carry the process forward.”
Dr. Wurm spoke at a two-day symposium sponsored by Sartorius in which representatives from the academic and private sectors discussed their approaches to producing more and cleaner proteins, confronting both the upstream and downstream ends of the equation. Better promoters, enhancers, and other genetic elements have contributed to increased upstream production of proteins.
However, Dr. Wurm stated that the role of these elements on the DNA level has been overemphasized, and that the principle source of increased productivity comes from better cell growth and other process-related improvements that were obtained through media modifications, productivity enhancers, and media feeds. Indeed, this strategy has resulted in a 25-fold increase in cell density in cultures from 1986–2007.
“The conventional rationale for obtaining high producers is based on examination of transfected cell lines by low-throughput screening to identify the top producing clones,” Dr. Wurm continued. Subsequently, low-throughput process development strategies were pursued. Biotech firms assumed that highly delicate cells would not survive more robust technologies for growth and thus implemented highly expensive, fully controllable bioreactors. Time-consuming protocols were applied that could take up to 18 months.
ExcellGene has the ability to rapidly scale-up production. It has developed a disposable design, simple bioreactor system as an alternative to a classical stirred bioreactor that does not require oxygen probes and pH electrodes. These small Tubespin™ bioreactors are miniature editions of the much larger 100-L bioreactors. Hundreds or thousands of these reactors can be run in parallel, and vast amounts of data can be collected, allowing optimization over a wide range of culture conditions in a matter of weeks.
“This protocol can deliver better productivity and growth of CHO cells in a simpler and more efficient format than classical bioreactors, while exceeding the latter in performance,” Dr. Wurm asserted.
The heart of ExcellGene’s technology is an orbitally shaken bioreactor. The cylindrical bioreactors, ranging in volume from 1–100 L, are mounted on a circular moving platform. Orbital shaking allows for efficient mixing of suspension cultures, while oxygenating cells through the headspace. The vessel contains a disposable, sterile bag with appropriate connection tubes for seeding, feeding, gas supply, and harvesting of the culture. Compared to classical stirred tank bioreactors, orbitally shaken bioreactors are much simpler and less costly to operate, reported Dr. Wurm.
Published data from his lab established that oxygen can be transferred to cells more efficiently with ExcellGene’s technology than in other systems, an outcome reflected in the growth and productivity from these reactors. ExcellGene disposable reactors at the 100-L scale are equivalent and frequently better than those of fully instrumented, nondisposable, glass, or stainless steel stirred bioreactors, he said.
“Biology plays a dominant role in cell performance,” Dr. Wurm added, believing that the karyotypic makeup of the transfected cell lines needs to be carefully monitored during expansion. The insertion of foreign DNA into these cells is an unpredictable phenomenon. CHO cells have a highly variable and unstable karyotype, given that they are permanently transformed cells that have been cultured in vitro for decades.
Using FISH, Dr. Wurm was able to show the regions that carry the human DNA within the hamster karyotype. These fluorescent spotlights on the chromosomes represent the genetic region that codes for the protein of interest. By periodically monitoring the karyotype of the cells during the scale-up process, it is possible to guarantee the presence of the target marker, assuring that the cell will in fact produce the desired protein.
“There are virtually no diploid cell lines used in bioprocessing today,” Dr. Wurm continued, “and it’s a good thing we didn’t follow recommendations to use them that were current 20 years ago, as it would have set back the industry tremendously.”
Dr. Wurm observed that even using antiquated technologies, “we have seen extremely rapid progress in volumetric yields from batch processes.”
He said the industry is on the cusp of a massive run up in production driven by the demands of the marketplace. This will be realized through development strategies that closely match large-scale systems with resulting easier, shorter run times and much higher cell densities. Fewer reactors will be required as the cells are better adapted and can experience split ratios as high as 1–100, which would have been impossible 20 years ago.
There is good news and bad news on the upstream processing front, Dr. Wurm said. “On the one hand, we are not going to need 200,000-L systems, as process volumes will decline in the next decade. Volumetric yields will rise dramatically, easily exceeding the 10-g/L range. Transient gene expression will dramatically increase the number of proteins processed and recovered.
“The bad news is that the upstream-downstream gap is widening at an alarming rate. New technologies will have to be found and some old technologies will have to be reinvented. Simpler is more foolproof.”
“Current downstream process designs don’t match the upstream process improvements,” observed Alahari Arunakumari, Ph.D., senior director of process development at Medarex. “To meet these demands, we have formulated a comprehensive strategy for integrated process development, rather than attacking the problem piecemeal.”
Cranking Up the Downstream
Medarex has achieved 5-g/L yields of antibodies, but this transition to higher titers demands substantial modification of purification processes. Dr. Arunakumari discussed the evolution of her company’s downstream antibody processing platform, which includes replacement of Protein A, or abandonment of Protein A from the process scheme, and truncation of the number of column chromatography steps from three to two and finally to a single step.
The Protein A affinity capture step is substituted by equally efficient and higher binding cation exchange resins. This non-affinity purification process for human mAbs consists of cation capture and disposable anion membrane polishing chromatography steps. These modifications are absolutely essential, given the increase in antibody yield per liter of culture fluid. Dr. Arunakumari explained how Medarex has successfully scaled-up ion-exchange purification technologies to produce clinical material for multiproduct campaigns.
The modifications in the purification protocol include replacement of low-binding with high-binding resins and replacing the anion exchange column chromatography with a membrane chromatography serving mainly for viral clearance.
The impact of this nonaffinity two-step purification scheme on facility output in terms of batch processing time and cost is profound, while at the same time product purity, contaminant levels, and viral clearance capability were unaffected.
“Membrane adsorbers allow improved purification of high-titre cell culture fermentations and, as such, will have a key role in the evolution of downstream processes,” says Lee Allen, Ph.D., senior scientist, DSP process engineering group of Lonza. Dr. Allen discussed how he and his colleagues have addressed the upstream overproduction dilemma.
The company is faced with two separate production issues, the first tied to small-scale and the other to large-scale production facilities. The small-scale facility in Slough, U.K., supplies material for Phase I/II studies in small batches with rapid turn around time. The large-scale plants generate multiple batches for late-stage clinical trials using Protein A and additional chromatography steps, virus removal and ultrafiltration.
Dr. Allen notes that as titer increases from 1–10 g/L the contribution of chromatography costs to the overall budget increases from 54–84% of the total. Thus, there is a powerful incentive to explore alternatives to traditional purification procedures. Membrane absorbers offer a number of advantages over chromatography beads in that they do not require packing, cleaning, and validation of stainless steel equipment. Because of their increased loading capacity in flow-through mode, there is less buffer consumption and reduced membrane area requirement. Dr. Allen covered the potential avenues by which membrane adsorbers could be used to tackle the downstream bottleneck.
The Lonza team evaluated the Sartobind Protein A membrane adsorbers as a replacement for the packed bed Protein A steps. However, because of substantial protein breakthrough, the use of the Sartobind A 75 product was determined to be unsuitable for replacing current packed bed technology.
When using membrane adsorbers for impurity removal in flow-through mode, a number of groups have found a substantial reduction in cost of goods and reduced personnel and equipment costs with excellent viral clearance.
These observations were reflected in Dr. Allen’s experience, who presented data on the implementation of membrane adsorbers to reduce host cell impurities in cell-culture supernatants and suggested that further optimization of the process will include varying conductivity and pH. “In combination with other purification methods it may have a role to play in future downstream processes that are more cost and time effective,” Dr. Allen concluded.
Economy and Process Robustness
“Do it right the first time and save costs,” said Franz Nothelfer, associate director, purification development, at Boehringer Ingelheim. His group has also faced the issue of large gains in protein production and the inevitable bottleneck that arises at the downstream interface.
According to Nothelfer, the initial process for optimizing the clinical supply of a protein therapeutic needs to have the potential for a commercial process, which requires close attention to cost factors.
Among the most critical factors are time constraints, so the company follows a tight timeline, moving through the early phase of small-scale production of the protein in a three-month period. This requires that the investigator think from a commercial point of view from the very beginning, so that the scale-up into large-volume production can proceed smoothly.
Nothelfer described a specific antibody project, discussing the selection of candidate clones, the initial characterization, the optimization of the protein A purification process, and the use of nanofiltration for virus removal.
In later purification protocols, wash and elution conditions were improved with a 40% saving in buffer usage and 20% reduction in process time. However, as the costs are driven down in other areas, the expense of affinity agents such as protein A becomes more and more dominant and a rate-limiting factor when one attempts to lower overall costs.
These factors have led Nothelfer and his colleagues to consider alternatives to chromatography, specifically the Sartobind Q membrane filter system. In the processing of an immunoglobulin product, DNA was cleared below the detection limit and there was a 20-fold reduction of process time, with diminished product loss.
“There is no generic one-size-fits-all solution for downstream processing,” said Nothelfer, “but the platform technologies must be fast, easy, and robust. In the future, there will be increasing focus on low-cost alternatives to chromatography.”
Needed: Protein A for $10 per kg
The bioprocessing industry has worked its way into a corner that seems to vindicate the adage, “be careful what you wish for, because you might just get it.” As yields of protein per liter of cells spiral through the roof, there appears to be no way that the purification end of bioprocessing can keep up in a cost-effective manner. Indeed the per kilogram cost of protein A now runs into the millions of dollars.
However, there are a number of commercial and academic investigators who are looking at affinity alternatives such as engineering the peptides in the protein A molecule that bind to the immunoglobulin molecule. It may be that these options can be developed and commercialized at a price that is more acceptable than the current tab for protein A and related affinity proteins.
This leaves a number of unanswered questions such as how to overcome the expense of buffers and other solutions whose cost, while trivial at the scale of the lab bench, can mushroom out of control when swimming pools full of buffer are required. This represents the next challenge for bioprocessors.