After drug-candidate selection, the design of the drug’s manufacturing-scale purification starts out at lab scale, in which both upstream and downstream processes come under scrutiny. “With the upstream process, you’re trying to optimize the growth of the cells and the production of the protein,” explains Jonathan Royce, global product manager for antibody affinity media in GE Healthcare’s Life Sciences section. He adds, “This extends beyond increasing the concentration of the protein to improving its quality, for instance trying to generate a cell clone that doesn’t produce a lot of aggregated or fragmented protein.”
Here, scaled-down modeling tools such as benchtop bioreactors can help the scale-up process and make optimization less costly. These growth chambers replicate the conditions of their larger-scale counterparts while enabling scientists to monitor multiple cell lines and experiments at the same time. GE’s ReadyToProcess WAVE™25 bioreactor , and Xcellerex™ XDR-10 systems are such examples. Other systems include Eppendorf’s scalable and BioBLU® 3f Single-Use Fermentor and Sartorius Stedum’s widely used ambr® 15 and ambr® 250 bioreactors.
Screening chromatography media
There are many considerations researchers should keep in mind for downstream process workflow. For this, Royce mentions the use of multiwell plates in the early lab-scale stages to screen potential media. These multiwell plates come prefilled with chromatography media covering many different chromatography techniques, including affinity, ion exchange (IEX), hydrophobic interaction chromatography (HIC) and multimodal resins. They support much of the early process design work. “For instance, they allow you to measure uptake curves and adsorption isotherms and [to] perform washing studies,” specifies Royce, adding, “A multiwell plate can allow you to differentiate between 12 or more different candidate resins with just a few rounds of experiments.” Furthermore, he points out this can be done with very little protein input and in a very time-efficient way. GE’s multiwell offering takes shape in the form of its 96-well PreDictor™ plates, which are available in several formats. Bio-Rad’s Foresight™ filter plates also are designed specifically for the resin-screening process and come prepacked with a range of Bio-Rad’s process chromatography resins, offering process scientists convenience and reliability for high-throughput experimentation. Other chromatography screening plates include Sartorius’ Sartobind® 96-well plates, available for either IEX or HIC media and Pall Corporation’s family of AcroPrep™ ScreenExpert plates, among others.
Automated screening solutions
If necessary, further resin-screening work can be performed with miniaturized chromatography columns, which can also be the starting point in the design process. “There is a danger of over-interpretation with resin-screening plates,” cautions Mark Snyder, manager of the Applications R&D Group in Bio-Rad’s Process Chromatography Division. He adds, “They don’t exactly replicate what’s going on in the chromatography column, and a couple of wells could possibly be outliers.” Many chromatography media developers, like Bio-Rad, offer prepacked microcolumns, which better represent larger-scale processes and are designed to be compatible with robotic liquid-handling workstations, such as the Freedom EVO® series from Tecan. These so-called RoboColumn® units are supplied eight columns to a row, which contain a few hundred microliters of each resin. “You make a scale jump of roughly 100-fold when you go from a plate setup to a RoboColumn,” Royce explains. “This allows you to perform more advanced method development as you’re beginning to work with a packed column instead of an aliquot of resin that’s sitting at the bottom of a plate.”
RoboColumns can help users optimize buffers, look at crude input vs. product yield and even begin to assess the clearance of impurities.
Another advantage of this setup is it still enables high-throughput screening (HTS) of chromatography media, as many experiments can be run in parallel on the robotic workstation. Both GE and Bio-Rad offer RoboColumn formats as part of their PreDictor and Foresight ranges, respectively. Several other biotech companies also provide RoboColumns, including Merck Millipore, Pall Corporation, Repligen and Agilent Technologies via its partner company JSB.
Snyder has further advice for this early stage; “When you’re moving from lab scale to process-development scale, it’s important to test several resins from different independent manufacturers and also to test several lots to see if they achieve consistent separation.”
Moving on up
Royce shares, “once you’ve generated some data on the RoboColumns, then it’s typically time to start scaling up to a full-sized column.” Here, process designers work on the milliliter scale, which is roughly another 100-fold increase in volume. This intermediary step is necessary to verify the results of the quality-by-design studies performed earlier in plates or on RoboColumns. “This is now a direct scale-down of what the process will look like during production,” Royce explains, “You’re now working with a similar platform you will be working with at pilot and production scale.” For instance, this may be an ÄKTA-based system, which has a pump and injection valves. “During screening with a plate or a RoboColumn, you operate somewhat in batches; with a lab-scale system like the ÄKTA platform, you can really start to simulate what’s going to happen in production.”
The next step is the pilot stage, “which is the first time a process designer will scale up the process into larger-scale bioreactors upstream,” says Royce. “This is when they will start to work with packed columns that dimensionally represent what they’re going to do in production, i.e., with a column of the same bed height.”
To make this a relatively seamless process—if you’ve opted to use pre-packed columns—it’s important to make sure your chosen resin is available in many formats that also span the entire range of scale.
“We make our resins available with as broad an offering of scales as possible, up to the prepacked 20-liter column we make in our cleanroom facility in Uppsala,” Royce says.
If customers opt to pack their own columns, companies like Bio-Rad have specialists on hand to help process design scientists from the outset. “Packing is important,” emphasizes Snyder. “Our scientists will spend time on-site with the customer and help them pack their columns for the first time.” They will also help troubleshoot issues or improve aspects of the workflow. “We have personnel who’ve been in the biopharma industry for a long time and have a lot of expertise in our resins; they are available worldwide to show researchers how to use them optimally,” Snyder says, adding, “It’s important to have a good working relationship with the resin manufacturer.”
GE has similar support services available in the form of a global team of chromatography specialists. These experts are locally based in all the regions GE serves, and they are available early in the process to help customers with their process design. GE also has several Fast Trak Centers around the globe where customers can receive hands-on help to set up a chromatography process with just a small sample of their target protein. These Fast Trak Centers also provide more detailed training in the form of residential courses throughout the year.
By the time a process is ready to transition to production scale, all the major steps and kinks will have already been worked out in the stages described above.
Important operating windows are created at the lab scale. These include the ranges of pH and salt-concentration tolerance, which are then verified in the pilot-scale runs.
It is important to stress that these parameters need to be set early in the process, and scientists really need to push these ranges to their limits. “A problem that sometimes occurs is when designers make buffer ranges too tight, or that require a degree of accuracy that is not possible on the production floor,” says Snyder. For example, pH ranges that are easily achievable at lab scale are much more difficult to achieve at production scale, where perhaps thousands of liters of buffer are in use. Snyder advises, “To really have a smooth transition to the manufacturing scale, the key is to set realistic parameters. Decide which ranges are critical and which aren’t.”
Struggling to settle on operating parameters? Several pieces of chromatography-modeling software are available to help you calculate them to a high degree of accuracy using statistical and theoretical plate modeling. During early process development, GE’s free Assist software supports the PreDictor workflow from setup of experimental design to data evaluation. It guides the user through the experimental workflow, assists in setting up experiments, keeps track of all data produced and supports evaluation of data. During later stages of development, UNICORN system control software (included with all ÄKTA systems) provides built-in knowledge for planning and controlling runs as well as analyzing results from chromatography, bioreactor and filtration systems. Other software offerings in this space are GoSilico’s Chrome X, AspenTech’s Aspen Chromatography™ platform and DryLab, which was developed by the Molnár-Institute for Applied Chromatography.
Most, if not all of these software options enable the user to keep detailed GMP method documentation, which is another important factor in process transfer. “You really need to keep good records,” says Snyder. “To make the transition to production scale as quick and efficient as possible, you need to keep a thoroughly documented process history.”
The limits to optimizing protein-purification workflow are really left to the imagination of researchers. Tool providers continue to provide advances to the technology that offer better selection specificity, improved isolation yields and greater sample throughput. Together, these improvements are helping scientists to be more efficient in their protein-purification workflows, and that will translate into both time and cost savings and ultimately better characterization of target proteins.