1. Introduction
Adenovirus (Ad)-based vectors are highly efficient gene transfer vectors widely used in vaccine development [
1,
2] and a variety of gene therapy applications [
3]. The Ad-based vaccine platform represents an attractive strategy as it induces robust humoral and cell-mediated immune responses, with proven safety and vaccine efficacy, and can meet the global demand in a pandemic situation. The development of at least four Ad vector-based COVID-19 vaccines, their excellent protection profiles and the administration of over one billion doses so far have fully elucidated the potential of this vaccine delivery system.
When adenovirus is used for vaccine applications, the price per dose will be an important determining factor of its economic viability. Achieving a cost-effective and scalable manufacturing process will be critical for success [
4], particularly when the diseases targeted by Ad vector-based vaccines have their highest prevalence in resource-poor settings. Therefore, the biomanufacturing process must yield high Ad titers. Currently, the Ad production process used at either the R&D setting or commercial manufacturing is still a typical batch culture process with an optimal cell density between 1 to 2 × 10
6 cells/mL at infection. There is a potential to improve the process productivity by infecting cultures at higher densities; however, many have attempted to improve the volumetric productivity using this strategy, but have not achieved much success [
5,
6,
7]. Adenovirus production is limited by reduced cell-specific productivity when high cell density cultures are infected. The breakpoint related to specific production drop with increasing cell densities at infection depends on the cell culture media. This limitation has been referred to as the “cell density effect” [
8].
The cell density effect has been generally associated with nutrient limitations and/or accumulation of inhibitory metabolites, although the exact nature of these limitations remains largely unknown [
9]. A complete medium replacement at the time of infection has been commonly used as a strategy to reduce nutrient limitations and/or the accumulation of inhibitory metabolites in batch operations. Virus production was significantly higher when cultures were resuspended in the fresh medium compared to those without medium replacement at similar infected cell densities [
10]. Higher volumetric productivity was achieved and maintained at infected cell densities up to 3 × 10
6 cells/mL in shake flask culture [
7].
Perfusion culture offers a continuous supply of fresh nutrients, maintaining required levels of essential nutrients, while removing inhibitory metabolites from the culture. The perfusion process allows the culture to reach a high cell density and provide the nutrients required during the virus production phase to maintain cell-specific virus productivity or minimize the “cell density effect” in cultures infected at higher cell densities. With the development of alternating tangential flow (ATF) technology, perfusion culture is becoming a popular process for the production of viruses and viral vectors [
11]. Cell-specific virus yields could be maintained or the volumetric productivity was proportionally improved with the cell densities up to 7 × 10
6 cells/mL at infection [
12,
13,
14].
Both batch culture with a medium replacement at infection and perfusion culture require additional equipment, and the use and handling of substantial quantities of additional medium, especially at manufacturing scales. Some perceive it as being a complex operation that is more difficult to implement than a fed-batch process. In addition, media consumption is substantially higher, and the operational cost of perfusion culture is much higher.
Fed-batch culture has become the research focus in the pharmaceutical industry to improve process volumetric productivity and reduce production costs. In the fed-batch process, the initial cellular growth is supported by the basal medium. Concentrated feed is added in fed-batch culture to replenish nutrients, sustain the cell growth to high density and improve productivity [
15]. Fed-batch processes are widely used to produce monoclonal antibodies but have only occasionally been employed in the process development for virus production [
16,
17]. Some previous works [
7,
18] employed a fed-batch strategy to alleviate the detrimental effects of lactate and ammonia accumulation on virus production, and slightly improved volumetric virus production. The virus productivity was still lower when compared to the medium replacement strategy. Nutrients provided by the relatively simple feeds might not meet the complexity of nutritional requirements in both the cell growth and viral production process phases [
19]. Recently, some improvement was achieved in fed-batch culture for the production of a chimpanzee adenovirus-vectored SARS-CoV-2 vaccine [
5].
Based on our recent findings that titers of other types of viruses can be increased using commercially available or in-house developed media [
20,
21], we seek to increase Ad5 titers by re-examining the fed-batch culture process. Through testing and customizing media, feed and regimen, we have succeeded in infecting HEK 293 fed-batch cultures at cell densities up to 5 × 10
6 cells/mL while maintaining the cell-specific productivity, resulting in 6-fold improvement in the volumetric titers up to 3 L bench scale bioreactor.
3. Results
Evaluation of culture media supporting HEK 293SF cell growth and Ad5-GFP production in shake flask batch culture: Figure 2A depicts the cell growth profile of the HEK 293SF batch cultures cultivated in four commercial serum-free media and one in-house developed medium (HEK SFM). Ex-Cell 293 medium supported cell growth close to 6 × 10
6 cells/mL, while the maximum viable cell densities supported by the other media were lower; up to 4 × 10
6 cells/mL in HEK SFM and SFM4Transfx-293; and ≤3 × 10
6 cells/mL in SFM4HEK293 and Pro293s-CDM. The cells’ doubling time in the latter was longer at 40 h.
Figure 2B shows the volumetric Ad5-GFP titers in the batch cultures cultivated with the 5 selected and in-house (HEK SFM) cell culture media, and all were infected at a cell density of 1 ± 0.1 × 10
6 cells/mL. The cells produced the highest Ad5-GFP titer at 7.5 × 10
9 vp/mL in SFM4HEK293, slightly higher than titers obtained in HEK SFM at 6.2 × 10
9 vp/mL. The productivity was between 4.1 to 4.5 × 10
9 vp/mL in the batch cultures cultivated in Ex-Cell 293 and SFM4Transfx-293. Lastly, Pro293s-CDM was a poor medium in supporting the production of Ad5-GFP.
By taking into consideration the capacity of the cell culture media to support cell growth and virus production, and the availability of commercial and in-house developed feeds, we selected SFM4Transfx-293 and HEK SFM as basal media for the development of a high cell density fed-batch culture process to improve the volumetric productivity of Ad5-GFP.
Cell growth and Ad5-GFP production in HEK 293SF shake flask fed-batch culture using commercial medium and feed: Figure 3A depicts very strong growth of HEK 293SF cells cultivated in SFM4Transfx-293 medium and fed with CB5. The total cell density (TCD) reached 18.8 × 10
6 cells/mL, and viable cell density (VCD) was more than 15 × 10
6 cells/mL after the culture was fed with a total 15% culture volume of CB5, tripling the maximum cell density obtained in the batch culture. However, the volumetric productivity in the fed-batch culture infected at 2.9 and 4.0 × 10
6 cells/mL reduced drastically instead of increasing (
Figure 3B), indicating that the SFM4Transfx-293—CB5 fed-batch culture process was very promising for reaching a high cell density but did not result in increased Ad5-GFP volumetric titers. This result prompted us to develop in-house feeds to improve the productivity of Ad5-GFP in the fed-batch culture.
Development of in-house feeds to improve the cell growth and Ad5-GFP production in shake flask fed-batch culture process: The HEK SFM medium supported the cell growth to 4.3 × 10
6 cells/mL in batch culture. About a dozen individual nutrients and other supplements such as amino acids, growth factor and trace metal were added alone or in combination to the batch culture when cell cultures reached 3 × 10
6 cells/mL to further increase cell densities. Experimental data (not shown) only indicated a trivial improvement in the maximum cell density. However, the maximum viable cell density, as shown in
Figure 4A, increased to 6.2 × 10
6 cells/mL and 8.0 × 10
6 cells/mL when the culture was fed with 1 g/L of yeast extract (Fed-batch_YE) and 1 g/L of Sheff-Vax ACF (Fed-batch_ACF), respectively. Much higher viable cell density up to 15 × 10
6 cells/mL was obtained when the culture was fed with a series of in-house developed feeds (Feeds 1 to 4; Fed-batch_F1–4) according to the feeding schedule described in
Table 1. Surprisingly, the profile of lactate accumulation in the shake flask batch and fed-batch_F1–4 culture was similar even though the maximum viable cell density in the fed-batch culture was almost four times the density achieved in the batch culture
Figure 4B. However, the ammonia concentration increased dramatically at day 8 when the cell density reached maximum. The osmolality of the fed-batch_F1–4 culture also increased from 304 to 319 mOsm/kg during the last two days of culture (day 8 to 10).
The volumetric productivity was very promising in the shake flask fed-batch culture using HEK SFM media and in-house developed feeds, and infected at different cell densities. The titer increased almost proportionally to the cell density at infection, reaching 3.0 × 10
10 vp/mL in the culture infected at 5 × 10
6 cells/mL, and was 6 times the titer obtained from the batch culture infected at 1.2 × 10
6 cells/mL (
Figure 4C). This promising result prompted us to scale the fed-batch culture process to the bench scale bioreactor to test its scalability.
Challenges in the scale-up of fed-batch culture to 3 L bioreactor: The fed-batch conditions developed for the shake flask culture using HEK SFM medium and in-house feeds were first scaled up to a 3 L bioreactor to test the growth of HEK 293SF cells. In parallel, a control culture taken out from the bioreactor after inoculation was maintained in a shake flask and fed under the same conditions used for the 3 L bioreactor.
Figure 5A reveals that the maximum viable cell density was only 9 × 10
6 cells/mL in the bioreactor culture, much lower than the 16 × 10
6 cells/mL obtained in the control shake flask culture.
The glucose consumption rate in the bioreactor culture was much faster than the control shake flask culture, almost resulting in a glucose depletion on day 4 (
Figure 5B). As a result of the high glucose consumption rate, the concentration of lactate accumulated in the bioreactor culture over the time course was much higher, exemplified by the respective maximum concentration of 38- and 19.6-mM lactate detected in the bioreactor and shake flask culture. Samples taken from the bioreactor and control flask culture over the time course were analyzed for residual concentration of amino acids. However, analytical results revealed no significant difference in the consumption rate of amino acids between the bioreactor and flask cultures. No amino acids were depleted during the time course of culture.
The production of Ad5-GFP was also scaled up to a 3 L bioreactor under the fed-batch conditions. One external shake flask control with culture taken out from the bioreactor after inoculation and one internal control shake flask culture taken out from the bioreactor after the viral infection were conducted parallel to the bioreactor culture.
Figure 5C depicts that the virus productivity was 30.4 × 10
9 vp/mL in the external control culture infected at a cell density of 5.1 × 10
6 cells/mL, while the titer was only 1.8 and 0.5 × 10
9 vp/mL, respectively, in the bioreactor and internal control cultures infected at a cell density of 4.1 × 10
6 cells/mL. The maximum concentration of lactate and ammonia in the bioreactor culture was 20.6 and 2.1 mM, respectively. This result clearly suggests that the virus productivity in the bioreactor and internal cultures was affected by either nutrient limitation, metabolite inhibition or cell status, or a combination of these factors, during cell growth and virus production.
Cell-specific virus productivity of HEK 293SF culture grown to different cell densities over the time course of fed-batch culture process: A set of experiments designed according to
Figure 1 was carried out to better understand the factor(s) causing the drastic decline in the virus production in the fed-batch bioreactor culture. The productivity of fed-batch cultures grown in 3 L bioreactor and 2 L shake flask to different cell densities was investigated.
Figure 6A depicts a decline of cell-specific virus productivity from about 6000 vp/cell in the culture taken from the bioreactor and infected at 1.1 × 10
6 cells/mL to productivity of <100 vp/cell in the culture from the bioreactor and infected at 3.3 or 6.5 × 10
6 cells/mL. This represented a more than 98% decrease in cell-specific productivity and was similar to the result obtained in the bioreactor and internal control cultures described in the previous paragraph (
Figure 5C). There was a significant correlation between the cell-specific productivity (CSP) and the total cell density (TCD) of culture taken from the 3 L bioreactor before the viral infection (CSP = −1021 × TCD + 5641; R
2 = 0.65). The cell-specific productivity also declined in the cultures from the 2 L shake flask and infected at higher cell densities. The percent reduction in specific productivity increased with the increasing cell density at infection: 65%, 74% and 94% for the cultures infected at 3.1 × 10
6, 4.2 × 10
6 and 7.3 × 10
6 cells/mL, respectively, in comparison to the cell-specific productivity obtained in a culture infected at 1.2 × 10
6 cells/mL. There was also a strong correlation between the CSP and TCD (CSP = −892 × TCD + 6032; R
2 = 0.81).
Data in
Figure 6B show that when the cultures harvested at different cell densities from the 3 L bioreactor and 2 L shake flask were centrifuged and cell pellets were resuspended in fresh HEK SFM medium at 1 × 10
6 cells/mL before the viral infection, there was a trend of only mild decline in the cell-specific virus productivity with the increasing cell density in the harvested culture. The cell-specific productivity was in the range of 6500 ± 1050 vp/cell when excluding the highest and lowest productivity (9800 and 3400 vp/cell), indicating that the cell-specific productivity was not dramatically affected by the cell density of culture. This trend was reflected by a weak correlation between the CSP and TCD in the data from the 3 L bioreactor culture (CSP = −549 × TCD + 8152; R
2 = 0.37) or from the 2 L culture (CSP = −439 × TCD + 8786; R
2 = 0.32). This result might suggest that the status (quality) of cells was not the main factor causing the drastic decline in the virus production when the culture was infected at high cell densities without medium exchange.
The maximum concentration of lactate and ammonia was 39 and 2.8 mM, respectively, in the fed-batch bioreactor culture when the cell density was at 6.5 × 10
6 cells/mL and was last withdrawn for the virus infection. The increased concentration of these metabolites might contribute to the decline of virus productivity, but seems less likely to be the main contributor to >95% reduction in the virus productivity based on our previous study [
19]. All of these data suggested that the declined virus productivity in the bioreactor culture infected at higher cell density might be related to nutrient limitation due to a difference in nutrient consumption rates between the shake flask and bioreactor cultures. Supplementing hydrolysates, such as Sheff-Vax ACF (or ACF in short), provides broad-spectrum nutrients and might alleviate nutrient limitation in the bioreactor culture infected at higher cell density. Therefore, we set out to optimize the nutrient supplies in the culture media and feed to maintain cell-specific productivity.
Maintaining cell-specific virus productivity in fed-batch bioreactor culture through the improvement of culture media and feed: HEK SFM medium was fortified with 1 g/L ACF before being used for the bioreactor inoculation. Glucose was included in the first feed to provide an additional 2 g/L glucose (final concentration) in the bioreactor culture to avoid glucose depletion.
Figure 7A depicts the growth profile of HEK 293SF cells in 3 L fed-batch bioreactor culture under improved nutritional conditions. The total cell density at infection was 4.8 × 10
6 cells/mL with a viability of 98%, and reached 6.2 × 10
6 cells/mL with a viability of 73% at 48 hpi. The glucose was not depleted in the culture during the time course. The maximal molar lactate and ammonia concentrations reached 57 and 2.8 mM, respectively, and the osmolality was 347 mOsm/kg at 48 hpi (
Figure 7B).
Volumetric titer of samples taken from the bioreactor at 43 and 48 hpi was 26.3 and 31.3 × 10
9 vp/mL, respectively, corresponding to a respective cell-specific virus productivity of 5479 and 6521 vp/cell. Volumetric productivity of an external control shake flask culture conducted in parallel and infected at 5.2 × 10
6 cells/mL was 33.0 × 10
9 vp/mL (
Figure 7C). The cell-specific productivity from this fed-batch bioreactor or shake flask culture was similar to the results obtained in the shake flask or bioreactor cultures infected at 1 × 10
6 cells/mL without feeding or media replacement before the viral infection (
Figure 6A), demonstrating that the cell-specific productivity could be maintained in the fed-batch bioreactor culture when the nutrient supplements in the culture were balanced. This result also demonstrated the feasibility of improving volumetric virus production through infecting cell culture at higher density in fed-batch culture.
4. Discussion
With recent advancements in the development of culture media, some commercial media can support the growth of HEK 293SF cells up to around 5 × 10
6 cells/mL or higher in batch culture. However, the optimal cell density at infection for maximum volumetric virus production is still less than 2 × 10
6 cells/mL, which was evidenced again in this study. The data in
Figure 2 revealed that there is no correlation between the virus productivity and achievable maximum cell density (R
2 = 0.03) among the 5 cell culture media tested in this study. The observed decrease in the specific productivity at higher cell densities was further exemplified by the 16 × 10
6 cells/mL obtained in the fed-batch culture using commercial SFM4transfx-293 medium and CB5 feed (
Figure 3). In this process, when the culture was infected at 3 or 4 × 10
6 cells/mL, the volumetric productivity dropped by almost one log, down to 6 × 10
8 vp/mL from 4.5 × 10
9 vp/mL obtained in the culture infected at 1 × 10
6 cells/mL. These results clearly show that a culture medium that supports a high density of cell growth does not necessarily warrant higher productivity. This indicates that, although appropriate cultivation strategies can be employed to increase cell densities in culture, it is often difficult to maintain cell-specific and volumetric productivities. This challenge is more likely due to the biphasic processes of cell culture-derived viral and vector production, in which an initial cell growth phase is followed by a virus replication phase initiated by virus infection [
23]. The cell growth phase and virus production phase might have different nutritional requirements.
The development of cell culture processes to increase adenovirus production has been often approached by designing feeding strategies that improve nutrient supply and reduce the accumulation of inhibitory metabolite(s) [
7,
24]. Since most commercial culture media are based on proprietary formulations, very few data have been published on nutritional requirements during cell growth as well as virus production. The lack of basic information on the composition and the complexity of commercial cell culture media are obstacles to developing nutrient-specific strategies and customized media that can support both high cell density and cell-specific productivity. We have exploited our experience in developing fit-for-purpose cell culture media and employed a rational approach through the design of experiments to develop and customize an in-house serum-free medium (HEK-SFM) and feeds that can support a cell density of 5 × 10
6 cells/mL in batch culture and up to 16 × 10
6 viable cells/mL in fed-batch culture (
Figure 4A). More importantly, through the optimization of in-house media and feed, the volumetric virus production dramatically improved to 3.0 × 10
10 vp/mL, and cell-specific virus productivity was maintained in the fed-batch cultures infected at higher cell densities (2.5 and 5 × 10
6 cells/mL).
Scaling up the fed-batch culture in a 3 L bench scale bioreactor to achieve high cell density and Ad5 productivity was challenging. The maximum viable cell density in the fed-batch bioreactor culture did not exceed 9 × 10
6 cells/mL, and the volumetric productivity dropped to 1.8 × 10
9 vp/mL when the fed-batch bioreactor culture was infected at 4.1 × 10
6 cells/mL, suggesting unfavorable conditions in the bioreactor for the cell growth and virus production. The unfavorable conditions could be attributed to increased concentration of accumulated metabolites (such as lactate and ammonia) and/or depleted nutrients (such as glucose) due to increased consumption rate of nutrients caused by shear stress [
25,
26] under the bioreactor or even large shake flask conditions. Interestingly, the cell-specific productivity of the cultures (
Figure 6B) grown to different cell densities and withdrawn from the 3 L bioreactor revealed that the status of cells in supporting the virus production was not dramatically and negatively impacted when the cells were resuspended to fresh media at a density of 1 × 10
6 cells/mL and infected.
Nutrient limitation has been frequently cited as the main cause for decreased platform performance at high cell density. Amino acids, glucose and easily-analyzed components in spent media were initially examined. However, aside from glucose, no significant nutrient depletion was measured. Due to the complexity of medium formulation and the limitation of analytical techniques, it is very challenging to quantify all the potential limiting components in spent media. Many trials, where individual or a group of components were supplemented to the cultures grown to 5 × 106 cells/mL and withdrawn from the 3 L bioreactor before the viral infection, resulted in no or insignificant improvements. A significant increase in the virus production was observed and the cell-specific productivity was maintained only when the culture medium (HEK-SFM) was supplemented with 1 g/L of ACF prior to the cell inoculation. This result might suggest that nutrient consumption is complex in the cell culture process, especially in the biphasic cell growth and virus production process. Identification of potential limiting nutrients is challenging, and the increased consumption of key nutrient(s) in the bioreactor culture could not be effectively compensated by supplementing fed-batch cultures with an individual component in order to improve the virus production. A hydrolysate such as ACF provides a much broader range of nutrients and is a fast approach to offset the depleted key nutrient(s). The result from this study indicated that the decline in the virus production in the 3 L bioreactor fed-batch culture infected at higher cell density was indeed at least partially due to nutrient limitation and could be minimized through nutrient supplementation. To our knowledge, the work reported in this study is the first demonstration of high titer Ad5 production in fed-batch bioreactor culture infected at a cell density of up to 5 × 106 cells/mL (highest so far) while maintaining cell-specific virus productivity.