April 1, 2015 (Vol. 35, No. 7)

Microbioreactors Afford Unique Views of High-Throughput Upstream Development

The conundrum of bioprocessing is that most development decisions concern matters for which the least information exists. What is even worse, this conundrum is never more puzzling than when developers confront the most important decisions.

“With small-scale shaken systems, you typically look at just one variable, say titer or cell growth, and eliminate a large number of possibilities based on that one number,” says Loe Hubbard, global product manager at Pall. Microplates and shake flasks are uncontrolled with poor mixing and mass transfer, and they do not duplicate process conditions. It is, therefore, easy to discard cell lines that might thrive in a bioreactor, or promote shear-sensitive cells that will not.

By providing control, feedback, and consistency, microbioreactors (MBRs) provide more meaningful data—even when they are used to monitor just one parameter.

Small-scale shaken systems will never disappear from bioprocess development, Hubbard says, because cell-line development will always outpace the capacity of strain evaluation teams. She suggests grouping strains according to similar behavior and transferring a few from each group into a controlled system to determine how they perform.

“If people did that routinely,” asserts Hubbard, they would be surprised to find that strains that don’t perform well uncontrolled do quite nicely in controlled systems.” Controlled systems such as MBRs have the additional benefit of greater comparability for cell growth or titer measurements.

Shake flask cultures are almost impossible for standardizing maximum oxygen transfer. “It becomes a function of which shake flask you’re using,” Hubbard states. “You become oxygen limited, so you’re actually measuring the oxygen transfer of your vessel, not the potential of the culture.”

Flasks differ widely in hydrophobicity; glass flasks provide higher kLa, gas composition and evaporation change with the type of closure; and old, worn surfaces interact differently with cells than new, surface-coated flasks. “But people mix and match flasks all the time and take their results as gospel,” Hubbard observes. “You can’t ignore culture variables just because you can’t control them. If you don’t at least consider the variables you normally associate with bioreactors, you can be working for no reason.”

Another common error with shakers is poor inoculum control—essentially varying the number of cells added to each flask, which will ultimately affect both cell density and titers.

To Hubbard, MBRs “don’t replace anything” but instead allow the appropriate use of microplates, shake flasks, and benchtop bioreactors within the range of tasks they do best.

As a rule of thumb, developers use shaken microplates for thousands of clones, shake flasks for twenty to several hundred strains, and microbioreactors down to a half dozen cell lines, below which small benchtop bioreactors become the vessel of choice.

“Microbioreactors,” Hubbard concludes, “are most suited for final screens and early process development.”

Purchasing and Operational Issues

Potential purchasers should consider MBRs with automated pipetting, which according to Dr. Barney Zoro, ambr®15 product manager at Sartorius Stedim Biotech, “allows you to do things you couldn’t before, like adding reagents at four in the morning.” Hands-free sampling or addition of feed or inducers (for microbial cultures) are critical for true scaledown systems. “If you start making compromises, you’ll get away from getting performance that matches larger-scale bioreactors,” Zoro says.

Before the emergence of MBRs, most bioprocess development occurred in shake flasks. “But those lack control for pH and oxygen,” Zoro adds. “You have few options in terms of additions, and there’s a lot of variability due to manual processing.” For example, gas composition changes, agitation stops, and flasks cool down when they’re removed from the incubator.

Like Pall’s Hubbard, Zoro believes that the proper role of MBRs is to help narrow down the number of clones to a handful from around fifty. “Fifty is the magic number because that was traditionally the number that a tech could manage with shake flasks,” Zoro remarks. “It also depends on what kinds of analytics you set up.”

A benchtop bioreactor then confirms performance at MBR scale. Developers should be prepared to verify performance of at least two to five clones, however, because scaling up from 15 mL to 5 L is not always predictable, and running back to screening results to select another candidate wastes time.

ambr bioreactors provide scalable predictions from microscale to manufacturing, according to Sartorius Stedim Biotech.

Minor Trade-Offs

The most serious constraint when running high-throughput upstream optimization studies is the simultaneous need for scaledown and real-time analysis (for example, process analytic technology). “It’s hard to have both,” says Ann D’Ambruoso, product manager, small-scale technologies, Applikon Biotechnology. Sensors for very small-scale bioreactors, for example MBRs, lack the capabilities of those employed in larger vessels. Moreover any process volume drawn for analysis from a milliliter-scale reaction well represents a significant proportion of the total volume.

But whatever their shortcomings, MBRs provide much finer control of dissolved oxygen, pH, and temperature than shake flasks or deep-well microplates. “There is still much to be gained by studying just those factors in a design-of-experiment or screening scaledown experiment, even if you cannot control as closely as at 5 or 50 L,” D’Ambruoso adds.

The MBR’s huge advantage is statistical power—the ability to run replicates that would not be possible at 5 L. Applikon’s micro-Matrix product, for example, provides two dozen 5 mL reactors. “There are certainly plusses and minuses operating at 5 mL scale, but the ease of use and flexibility for today’s microbioreactors makes such experiments extremely valuable in decision-making. You can introduce design-of-experiment variables, and control accuracy is still quite good despite the small volumes,” D’Ambruoso explains.

Some variability exists around MBR inputs and outputs—the best development stage to begin utilizing MBRs, and the most likely post-MBR step or scale.

D’Ambruoso rejects a one-size-fits-all strategy: “True, an MBR is great for screening media and other process conditions. But the information you get out is much more valuable than with other low-volume technologies because cultures grown in MBRs approach the cell densities of a much larger bioreactor.

“You won’t get the exact same process because of the nature of stirring and mixing differences, but cell density is important. It helps process developers get much closer to an informative decision.”

Several Applikon customers, for example, have asked about MBR applications to replace some plate-based screenings. Detecting highly productive clones is significantly easier in a 5 mL MBR than from a 50 μL microplate well.

Sometimes decisions come down to budget and lab space. “Microbioreactors are more expensive than microplates,” D’Ambruoso remarks. “It’s a function of what fits into their facility and what they can afford.”