Developing new media or enhancing existing media formulations in order to move from static planar vessels into dynamic suspension bioreactors or for improved bioprocess design are well documented in the literature, and knowledge regarding the most critical components in hPSC media has been greatly expanding throughout the last two decades [24]. This has led to a myriad of different media formulations for hPSCs [25–30]. Currently, there exists over 15 commercial media for hPSC culture [24]. However, most of these media formulations were developed with studies focused on optimizing media components and concentrations within static culture platforms and laboratory scale volumes [24, 26, 29]. The current study presents a novel comparison between five common commercial hPSC media (StemFlex, mTeSR1, CTS E8, PluriStem, and NutriStem) in a scalable suspension bioreactor, in which hPSCs grow as aggregates, through short-term and long-term culture. Although some of the commercial media formulations have been tested in various dynamic systems, specifically mTeSR1 [22, 31–36], most others have no published results in suspension bioreactor platforms. Therefore, if a process has been developed using static culture vessels only at a small scale with a specified commercial medium, it is risky to assume that this medium will be translational to scalable suspension culture systems. Moreover, some studies tested PSC culture in suspension bioreactors only by inoculating cells from 2D planar seed train into the bioreactors and analyzing their growth without further passaging between bioreactors. Without data that demonstrates the performance of PSC media in suspension bioreactors, particularly through serial passages, efficiently translating protocols to clinical and commercial manufacturing systems and scales will likely be delayed.
This study demonstrated that, when using the VW bioreactor platform and bioprocess protocols outlined, hPSC expansion in those five commercial media tested herein could be successfully transitioned into a dynamic culture environment for a single bioreactor culture expansion period. The ability to transition all tested media successfully into dynamic culture using specified PIVs highlights the robust nature of the bioprocessing protocols used for this stage of the bioprocess workflow. This success was supported by normal aggregate formation, growth, and morphology along with relative consistencies in cell growth rates and maximum fold expansions achieved for all tested media. The latter is especially impressive as successful expansion has been defined in this study to be meeting or exceeding growth rates from top published reports in current literature [37–39]. Earlier studies in literature testing the translation of PSC expansion from static planar culture vessels into dynamic bioreactor systems report cell growth rates and fold expansions that are significantly lower [40–46] than recent publications [37–39]. It is important to note that in the current study there are some significant differences in the fold increase and average aggregate size between media tested through one bioreactor passage, especially by day 6 of culture. These differences are most likely due to the proprietary components and differences in their respective concentrations specific to the media. For example, NutriStem manufacturer notes there is a low level of basic fibroblast growth factor in their medium. This low concentration combined with the minimal feeding regime used in this 3D bioprocess protocol could contribute to the smaller aggregate diameters and lower fold-expansion results with NutriStem when compared to the StemFlex medium condition. It is, however, difficult to ascertain any conclusions without knowing the actual formulations of each different medium.
Although initial results from this study suggest that all five media tested could be transitioned from 2D to cultivate hiPSCs in 3D bioreactors, the results from the in-vessel dissociation of aggregates expanded in bioreactors and the serial bioreactor-to-bioreactor passage experiments highlighted the risks of drawing upon such conclusions prematurely or extending these conclusions to downstream protocols in the workflow. While some bioprocess workflows may require only one stage of PSC aggregate expansion in the bioreactor before starting differentiation, others will require multiple bioreactor-to-bioreactor passages to generate the required cell numbers for clinical or commercial manufacturing purposes, particularly for allogeneic therapies. For these bioprocess workflows it is critical to evaluate bioprocess harvesting/passaging protocols and expansion in 3D culture for prolonged durations. Most published literature that includes studies with PSC serial passaging in 3D culture systems were designed at a small-scale, and methods used to dissociate aggregate samples from the bioreactor for subsequent passages are not scalable [47, 48]. As such, we tested our previously optimized in-vessel aggregate dissociation protocol to harvest the entire bioreactor volume for serial passaging [5].
When this in-vessel dissociation protocol was employed on the aggregates that had been expanded for 6 days in the various media tested, significant differences were noted between the different media. When dissociating aggregates cultured in StemFlex or mTeSR1, harvest recoveries were ~ 100% with no noticeable cellular debris. In comparison, when dissociating aggregates cultured in CTS E8, PluriStem, or NutriStem, there were significantly lower harvesting efficiencies and a decrease in the cell quality. Most notably, the dissociation process resulted in cell lysis as indicated by the visible DNA strands and cellular debris in the supernatant. While the viabilities of the cells post in-vessel dissociation of the aggregates expanded in bioreactors using CTS E8, PluriStem and NutriStem were lower than observed during the pre-harvest cell sample count, these differences alone do not account for the significant reduction in cell number obtained from the harvest. It is presumed that these lysed strands account for a large fraction of cells that could not be counted post in-vessel dissociation. Since the lysed DNA is known to be a sticky cell substance, it is likely that these strands caused some of the aggregates and single cells to adhere together during the dissociation procedure. These agglomerated cell-aggregate masses were excluded from the recovery of the dissociated single cells from the bioreactor harvest, and these cell losses would account for the large decrease in total cell number post in-vessel dissociation. These results indicate that the choice of media not only impacts cell culture process and quality output variables such as growth rate, aggregate morphology, and aggregate size, but it can have a significant impact on cell harvesting/passaging process parameters and results. Such process steps, including in-vessel aggregate dissociation, are often not considered when designing and testing upstream cell culture processes. The current study highlights the importance of evaluating all steps of the bioprocess (expansion, dissociation/harvesting, and passaging) to truly encapsulate the impact of different input variables.
The different outcomes of aggregate dissociation into viable single cells were most likely due to media components and their impact on cell quality during the first bioreactor expansion phase. For example, shear protectants, such as Pluronic, are commonly added to cell culture media to prevent cell damage or death caused by fluid shear and aeration in agitated suspension bioreactors [49, 50]. As such, the absence of these components can profoundly impact cell quality when cells are exposed to the hydrodynamic environments in bioreactors [44, 51–53]. In our previous studies we have shown that there is an optimal range of hydrodynamic values, and thus agitation rates, for the expansion of high quality hiPSC aggregates in the VW bioreactors. However, to facilitate the dissociation of aggregates into single cells during the in-vessel dissociation process, higher shear forces are required than what we recommended for the culture expansion phase [20]. For the optimized in-vessel aggregate dissociation protocol, a reduced working volume of 20 mL and an agitation rate of 80 rpm was effectively used in the PBS-0.1 Mini bioreactors. This falls outside the suitable hydrodynamic range for hiPSC expansion culture. Under this in-vessel dissociation protocol, the cells are subjected to turbulent fluid conditions where eddies generated by hydrodynamic forces are smaller than aggregate diameters, thus causing aggregates to be sheared apart into single cells. Although these higher forces are required to effectively dissociate the aggregates, it may leave the cells more susceptible to damage. As such, having the cells exposed to shear protectants throughout the culture period, priming them for high shear dissociation, may be necessary to preserve cell quality and viability. Therefore, it may be possible to increase the success of a complete bioprocess in the VW bioreactors when using CTS E8, PluriStem, and NutriStem if they are supplemented with such protectants. If these media are considered for the culture of PSCs in suspension bioreactors, this hypothesis should be verified by testing such shear protectants at various concentrations.
In addition to the immediate observations noted from the in-vessel aggregate dissociation studies, the long-term impacts of the media tested on growth and quality outputs were observed during the subsequent serial passages. It should be noted that there appears to be an extended lag phase present for all media conditions during the second bioreactor passage. Comparing day 1 and day 2 average cell counts indicates they are lower during the second bioreactor culture compared to the first bioreactor culture for all media conditions. This apparent lag could be a result of cell exhaustion at the end of the first bioreactor culture. If the cells were nearing the limits of a factor like dissolved oxygen concentration, pH buffer capacity or nutrient availability, a portion may start to exit the exponential growth phase undergoing proliferation at a reduced rate and then enter a stationary phase. These cells that exit the exponential growth phase are likely to display an increased lag phase in the following passage before re-entering the exponential growth phase. Another observation that was made was the cells cultured in either StemFlex or mTeSR1 maintained consistent growth through three bioreactor passages. Conversely, cells cultured in CTS E8, PluriStem, or NutriStem showed a significantly decreased growth and changes to aggregate morphology, specifically during the third bioreactor expansion culture where there was negligible growth over 6 days. This aligns with the findings from the aggregate dissociation stage of the first bioreactor expansion culture where the aggregates expanded in StemFlex and mTeSR1 were almost completely dissociated into viable single cells with ~ 100% harvest efficiency whereas the cultures with CTS E8, PluriStem, and NutriStem resulted in macroscopically visible cellular debris and reduced viable cell recovery during the passage. The challenges revealed during the in-vessel dissociation-mediated harvest process impacted results in the subsequent passages. Although cells cultured in CTS E8, PluriStem, and NutriStem were able to recover and show some growth in the second passage, by the third passage the cultures experienced extended lag phases with little to no cell recovery in growth. This was mostly likely an indication of compounded cell quality reduction through the serially passaged culture in bioreactors. These amplified consequences again highlight the importance of extended bioprocess testing timelines to identify protocols which will produce clinically or commercially relevant numbers of high quality and functional PSCs.
While key process attributes including growth kinetics, aggregate morphology, and viability are critical in assessing bioprocess robustness, assays that test genetic stability and pluripotency are essential to assessing the maintenance of PSC cell quality and function. It has been noted in literature that the use of enzymatic passaging can increase the risk of genetic instability [28, 54, 55]. Therefore, to validate the success of the overall bioprocess, cell quality should be further analyzed. Of note, the mTeSR1-based culture tested with the 4YA cell line and bioprocess PIVs from our past publications was taken as a positive control when establishing genetic stability and pluripotency maintenance in this study. In our previous publications we have performed in depth genetic (G-banding karyotyping), phenotypic (surface and nuclear marker cell-image staining, polymerase chain reaction, and flow cytometry) and functional (tri-lineage directed differentiation and teratoma formation) testing following serial passages in the VW bioreactor [5, 21]. These studies demonstrated the robust nature of the bioprocess to generate high quality PSCs, allowing the focus of this study to be on evaluating other critical input and output variable relationships throughout a long-term bioprocess.
Throughout the short- and long-term media comparison studies presented here, it was found that, in addition to the mTeSR1 process control condition, the condition cultured with StemFlex medium was successful in maintaining key process attributes (i.e., cell viability/growth kinetics and aggregate size/morphology) throughout the bioprocess. As such, we continued to passage the hiPSCs cultured in StemFlex medium through a total of 10 bioreactor serial passages (60 days) to demonstrate extended bioprocess robustness. During this prolonged testing, not only did the cells maintain growth characteristics and aggregate morphology throughout each passage, but they were also found to maintain genetic stability and pluripotent function. Cell samples from the StemFlex bioprocess condition were taken at the end of the third and tenth bioreactor serial passage for G-banding and teratoma formation assays. At both time points, genetic stability and pluripotent function in-vivo were confirmed. This again highlighted the robust nature of the developed protocol to ensure to produce quality-assured cells. It is recognized in literature that more subtle changes in genetic stability and pluripotency may not be distinguishable using these traditional quality assays [55]. As such, higher resolution quality assays should be employed before finalizing the bioprocess used in clinical or commercial manufacturing settings to ensure safety and efficacy.
iPSC line-to-line variability is another prominent challenge in bioprocess development [44]. As such, in the final part of this study we repeated the long-term serial passage testing with the successful media candidates (StemFlex and mTeSR1) using two additional hiPSC lines from different commercial vendors. The consistent patterns and predictable results of cell growth and aggregate morphology demonstrated in these final experiments further highlight the robust nature of the bioprocess protocols established. Differences in the growth rates throughout the three serial passages using each cell line were shown to be minimal. All conditions demonstrated exponential cumulative multiplication patterns through the three serial passages and normal aggregate growth and morphology. These results build further confidence in the bioprocessing protocols established and provide evidence that these protocols can be translated to other hiPSC lines with minimal optimization required.