Lost in Translation
Very Early Process Development for Biopharmaceuticals
Today, scientists are searching for proteins to be used as biopharmaceuticals in the most remote corners of the plant and animal kingdoms. Drug candidates are cloned, trimmed and modified and heterologously expressed in a handful of relatively well-known host organisms. If things go extremely well for a new API (active pharmaceutical ingredient), it is expressed and milligrams can be purified, proof-of-principle can be shown, toxicity seems to be no issue and the need for a more useful amount for further characterization, usually along with first considerations of conducting a clinical trial, arises.
In about a decade of upscaling and process development for a plethora of different API candidates at the Fraunhofer IME's department for Integrated Production Platforms, we have noticed that at this point, many promising candidates and sanguine dreams (and business plans) die silently. This article tries to highlight a few of putative showstoppers in early process development for biopharmaceuticals and to point at the difficulties that will most likely occur as early as possible.
Quick Success Versus the Search for a Sustainable Expression System
The first pitfalls in process development occur at the very beginning. Unfortunately, the earlier a decision is made that is not far-sighted enough, the more damage will be done in later process stages. The work of years will be reset to the starting point because something in the process simply turns out to be not feasible. Often, in an long and sometimes challenging process of generating an expression clone, decisions are made that are aimed at short-term objectives but cannot be adhered to in later phases. In other words, a quick success is - consciously or not - preferred to longterm sustainability. For example, commercially available bacterial expression kits that are optimized for high-expression levels are used in low-cell-density cultivations on complex media. The kits often feature modular solutions such as support for rare codons by additional plasmids, deletions and additions in the host strain that increase folding efficiency or stability, strong promotors that rely on specific genetic elements and environments and sometimes protein tags for purification or even detection of the foreign protein.
These kits are designed to make life easier, but many elements or modules can cause severe problems when it comes to their usability in large-scale applications or regulatory compliance.
The decision for an expression cassette or the use of a certain expression kit that promises quick success is not wrong in itself. The problem is that early achievements seduce to built on them and continue the easy way without asking how practicable it will be later. As we will see, things that make life easier by rendering careful design of the gene of interest, the expression cassette and the expression system unnecessary in the beginning might cause major discomfort later on.
The Challenge of Scaling-Up
One of the misunderstandings we've been faced frequently is the assumption that a bioreactor is basically a large shake-flask featuring a built-in autoclave. Processes that lead to a "fat band" in a shake-flask or a cell cultivation system do not always maintain their specific productivities when translated to a bioreactor. The term "specific productivity" describes the level of expression or the accumulation of a product per biomass or per cell. Expression kits are often tailored to low-cell density cultivations in very rich media. Under these conditions, optimal supply of nutrients or energy and accumulation of inhibitory metabolites is usually not a problem. In medium with high cell density fermentations, this may become limiting and decrease the specific productivity. Also, complex media compounds such as peptones or additives such as protease inhibitors can be unacceptable from the regulatory or economical point of view. As a result, actual productivities may be lower than the calculated numbers. This can often be fixed by intensive process development, but it needs resources and time to get there.
Scaling-up a laboratory purification process to pilot scale usually includes major changes in the separation steps. Laboratory filtration steps sometimes rely on syringe filters and chromatography in gravity-driven columns. Also, "perform all steps on ice" is not an unusual direction in laboratory protocols, but neither very specific nor easy to in larger scales. Chromatography media suited for reasonable processing times and flow rates can differ fundamentally in terms of resolution and capacity from those used in gravity-flow approaches. Filtration processes such as ultra-filtration or diafiltration, when carried out in tangential flow devices can introduce thermal load to a process that may impair product integrity and cooling down some 50 liters of an intermediate in a carboy takes surprisingly long compared to 50 ml in a falcon tube.
Assuming that a process aimed at the production of a clinical-grade API has been scaled up successfully, analytic results generated by the QC department are likely to hold some surprises. The purity of early-development protein preparations is usually overestimated - sometimes simply due to the detection limits of the method used. All of a sudden, with adequate overloading, bands of unknown identity pop above and below the familiar band. Also, the resolution of up-to date chromatography controllers will more often than not raise questions regarding the homogeneity of the main elution peak by unveiling shoulders or a distinct asymmetry that inevitably call for further investigation and explanation.
Assuming once again that all technical obstacles can be adequately addressed and circumnavigated, there are legal issues that are often ignored until there is is a product, a market or some profit foreseeable. There's always something more urgent or more interesting to take care of than analyzing which bits of somebody else's intellectual property you have been using somewhere along the road. At this point, circumvention of a technology that has worked well for a protein is an extremely painful decision to be made. But it's either that or negotiation of license fees. An easily obtained research license may turn out to be a costly mortgage on the future if not addressed in time.
Time is Cash
Time may be linear or not, but in biopharmaceutical process development it definitely has more dimensions than "time-to-market". It may take half an hour to spin down 500 ml of cultivation broth, but in a pharmaceutical process, you'll still end up with 2 days for the same process step if you account for buffer preparation, run time and cleaning-in-place. The same applies to other process steps in a very similar fashion. Cost of goods also make a remarkably non-linear impression on most researchers. A raw material or consumable order somewhere under 500 Euros or dollars passes through easily without raising any eyebrows and a scale-up factor of 100 is easily asked for. In fact we have been asked several times for a quote on the scale-up of laboratory protocols where the cost of buffers alone exceeded the customer's budget expectations for the entire project.
But these are relatively simple technical issues. More important are problems that cost time, no matter if there is enough money available. We've all heard the calculations of how many dollars of earnings are lost in a month's delay of the launch date of a blockbuster. Such calculations may be rigged a little to impress the audience, but in fact, the further a product is down the pipeline, the more expensive and time-consuming it becomes to clean up unsolved problems.
There are many points to consider in the early process development of biopharmaceuticals and even this very limited and superficial flashlight on the most easily made mistakes in the translation of a product from research to development may be disillusioning. The take-home-message is to involve know-how on the technical, regulatory and legal implications of biopharmaceuticals as early as possible. On the other side, the availability and accessibility of existing information and know-how as well as sophisticated technologies to efficiently produce and thoroughly analyze the biopharmaceuticals was never as abundant as today. Maybe it's time to go back to one of the fundamental virtues of scientists - think twice before you pipet.
Fraunhofer IME - Department for Integrated Production Platforms
Fraunhofer IME is one of 5 Institutes of the Fraunhofer Life science alliance. Building on 20 years of experience of recombinant protein production with an emphasis on antibody-derived products and plant biotechnology, about 150 scientists are working on contract research and development projects focused on Pharmaceutical Product Development, Plant Biotechnology and Applied Ecology at the IME and the closely-attached institute for Molecular Biotechnology at the Aachen University. Dr. Stephan Hellwig is co-heading the Department of Integrated Production Platforms at the IME. The department does contract research in protein production process development and holds a manufacturing authorization for the GMP-compliant production of API for clinical phase one trials. Dr. Hellwig is Head of Manufacturing in the GMP facility.