Serum‐free suspension cultured human cells can produce a high‐level of recombinant human erythropoietin

Currently, the majority of commercial recombinant therapeutic proteins are produced in nonhuman expression systems. Human cells are capable of producing recombinant proteins with post‐translational modifications that are more similar to native proteins which reduce immunogenicity. This article describes the potential of the human cell lines Sk‐Hep‐1, HKB‐11, and Huh‐7, cultured in serum‐free suspension conditions, to produce a complex recombinant glycoprotein, human erythropoietin (EPO). Recombinant cell lines were generated by lentiviral transduction and sorted by flow cytometry. All the recombinant cells presented high‐specific cell growth rates and a high level of rhEPO production. Maximum rhEPO concentrations achieved by Sk‐Hep‐1, HKB‐11, and Huh‐7 cells were 112.5 μg/mL, 112.7 μg/mL, and 571 μg/mL, respectively. The levels of rhEPO production by Huh‐7 cells were higher than the levels commonly reported in the literature for serum‐free suspension cultures. The rhEPO produced demonstrated biological activity measured through in vitro differentiation of CD34+ cells. In view of this, we demonstrate that Sk‐Hep‐1, HKB‐11, and Huh‐7 cell lines have good characteristics to be used as host cells for the production of complex recombinant glycoproteins, with special emphasis on Huh‐7 due to enhanced EPO production.

humans. 1,2 Chinese hamster ovary (CHO) cells are the most commonly used mammalian cell-based expression system, 3 being responsible for 60%-70% of all recombinant products produced. This can be explained by the fact that CHO cells are capable to produce recombinant proteins at high productivity with titers up to grams per liter of protein, to synthesize human-like glycosylation patterns, are well adapted to serum-free suspension cultures, are refractory to a number of human viruses, and have established regulatory track record. 4,5 Other cell lines, such as baby hamster kidney cells (BHK21), murine myeloma cells (NS0 and Sp2/0), human embryonic kidney cells 293 (HEK293), and human fibrosarcoma cell (HT-1080) are used for therapeutic protein production as well. 6,7 Nonhuman mammalian cells have proven reliable for recombinant protein production, synthesizing proteins that are similar to those naturally occurring in humans. However, there are some differences in glycosylation patterns among mammalian species, which have had a significant impact on the quality of the recombinant protein produced. 4,8 CHO cells, for instance, may not adequately perform glutamic acid -carboxylation, post-translational modification required for activity of some therapeutic recombinant proteins such as coagulation factors VII and IX. 3 The main structural differences between recombinant proteins produced in nonhuman vs human mammalian cells include the presence of the nonhuman sialic acid N-glycolylneuraminic acid (Neu5Gc) and the absence of (2,6) linkages of sialic acid to galactose. 3,9 The presence of these N-glycans can contribute to immunogenicity and neutralization of the corresponding recombinant therapeutic proteins since humans have demonstrated the ability to produce circulating antibodies against them. [10][11][12][13][14] Neu5Gc uptake either via consumption of animal-sourced food or of contaminated biopharmaceuticals may lead to the onset of chronic inflammation and is suspected to increase the risk of cancer and heart attack. 15 Using human cells for recombinant protein production is a viable way to overcome issues related to nonhuman cell-based expression systems. Human cells can generate more natural, human-like recombinant proteins with higher quality characteristics including lower immunogenicity, greater biological activity, and increased half-life. 16 There are human cell-based expression systems that have demonstrated promising outcomes for recombinant protein production 6 : AGE1.HN (Probigen AG; a human cell line developed from primary neuronal precursor cells), CAP cells (CEVEC's Amniocyte Production; an immortalized cell line based on primary human amniocytes), HEK293, PER.C6 (Crucell; a human embryonic retina cell line), and HT1080 (human fibrosarcoma cell. [17][18][19][20][21] However, variations in glycan structure persist even among human cells, for example between differentiated normal human cells and tumor cells or when comparing erythropoietin (EPO) produced recombinantly to EPO obtained from natural sources. [22][23][24][25] The human cell lines used in this study, Sk-Hep-1, HKB-11, and Huh-7, were already successfully used for the production of recombinant coagulation factors VII, VIII, and IX [26][27][28][29][30] and adapted to serum-free suspension cultivation. 31 In this research, we selected human EPO as a model protein to further evaluate the potential of these cells to produce complex recombinant glycoproteins.

Lentiviral production and titration
For lentiviral production, HEK293T cell line was cultured in 75-cm 2 T-flasks containing 10 mL of DMEM culture medium supplemented with 10% fetal bovine serum and transfected using a mixture of DNA and the cationic polymer polyethyleneimine. For transfection, the three plasmids were used in the following proportions: 20 μg of vector p1054, 13 μg of vector pCMVΔR8.91, and 7 μg of vector pMD2 VSVG. After 48 hours of transfection, the supernatant was collected, centrifuged at 450×g for 5 minutes at 4 • C, filtered (0.45 μm filter), and stored at −80 • C. A gene transfer assay was performed in HEK293 to titer viral particles produced in HEK293T. Briefly, HEK 293 cells were plated at 3 × 10 5 cells/well in six-well plates and aliquots of 10, 20, 50, 100, and 500 μL of the lentivirus-containing supernatant were added to the cells. Cells were maintained in culture for 4 days and then trypsinized for flow cytometry analysis (GFP measure) for viral titer quantification.

Generation of stable rhEPO producing-cell lines
Aliquots containing 1 MOI (Multiplicity of infection) of the produced lentiviral particles were added to the cultures of SK-Hep-1, HKB11, and Huh-7 cell lines. At the time of transduction, 1 mL of the cell suspensions were seeded at a concentration of 3 × 10 5 cells/well in a 6-well plate. After 72 hours, the cells were taken for analysis of green fluorescent protein (GFP) expression as well as cell viability by flow cytometry. All assays were performed in duplicate. To obtain a homogenous recombinant cell population, recombinant cells were sorted by flow cytometry using the BD FACSAria cytometer. The selected cells were recultured, and after the third passage, they were reanalyzed for the percentage of postsorting GFP positive cells. The cells were then expanded and frozen to create the recombinant cell banks.

Selection of the best media for rhEPO production
With the objective to select an appropriate culture medium for the production of rhEPO, two culture media were evaluated for each cell: the media where cells presented better growth rates and viability during adaptation process 31 and the Freestyle 293 Expression media (Thermo Fischer Scientific). For this purpose, 1 × 10 6 cells/mL of each cell line was inoculated into 20 mL of each media in 125 mL Erlenmeyer flask and cultured under the aforementioned culture conditions. The production of rhEPO was monitored for 72 hours. Every 24 hours, 1 mL aliquots were removed, and after centrifugation at 1200 rpm the supernatants were frozen at −20 • C for further analysis. Cell densities and viability were measured by the dye exclusion method using Trypan blue (0.4%) in a hemocytometer. . For absolute quantification, a 1054-EPO standard curve was constructed using a 10-fold serial dilution ranging from 10 1 to 10 6 plasmid copies per well. With the knowledge of the molecular weight of the plasmid and the insert, the corresponding plasmid copy number was determined. 34 The standard curve was generated by linear regression of the Ct values plotted against the logarithm of their template copy numbers. The absolute gene copy number in samples was quantified by analyzing 100 ng of genomic DNA of each cell line. The absolute number of copies was divided by the corresponding number of genome molecules related with 100 ng of genomic DNA of each cell line. Each qPCR run was performed in triplicate.

Relative mRNA expression
The transduced cell lines were analyzed for the level of rhEPO gene expression by RT-qPCR (Real Time-Quantitative PCR). For this, the total RNA was extracted from 1 × The volumes of all reactions were achieved using pure nuclease-free water. Data analysis was performed according to the relative expression of the target gene to its control, through equation 2 −ΔΔCt . 35 Quantitative PCR was carried out in 7500 real-time PCR System (Applied Biosystems), 40 cycles were performed consisting of 50 • C for 2 minutes, 95 • C for 10 minutes, 60 • C for 1 minute. Normalization was performed using the geometric mean of the Ct (Cycle threshold) of the reference genes. The reference genes were the endogenous 2 microglobulin and -actin genes, and each qPCR run was performed in triplicate.

Characterization of growth, metabolism, and rhEPO production by recombinant cells
After determining the best media for each cell, a kinetic characterization was performed regarding cell growth, metabolism, and rhEPO production. For this, 5 × 10 5 cells/mL were inoculated into a 125-mL Erlenmeyer flask containing 20 mL of each media. Cultures were maintained for 312 hours. Every 24 hours, 2 mL aliquots were collected, and after centrifugation at 1200 rpm the supernatants were frozen at −20 • C for the subsequent analysis. The culture volume was adjusted to 20 mL by adding fresh media for each 2 mL aliquot taken. Glucose, lactate, glutamine, and glutamate concentration in the culture supernatants was carried out using the YSI 2700 biochemical analyzer (Yellow Springs Instruments). Ammonia concentration was determined using an ion-selective electrode (Orion 720A + , Thermo Scientific, South Logan, Utah) according to the manufacturer's instructions.
The maximum specific growth rate ( max ) was estimated from the slope of the best-fitting line of the logarithm of viable cell density vs time during the exponential growth phase. Specific EPO productivity (qEPO), specific consumption, or production rates of glucose (qGlc), lactate (qLac), and ammonium (qNH 4 ) were calculated by plotting the products or substrate concentrations against the values of the integral of viable cells (IVC) during exponential growth phase, except for qEPO in which IVC was estimated from 48 hours to 216 hours, a time-period related to the rhEPO production phase. A positive signal for specific rate indicates production and negative consumption. The IVC is the area under the viable cell growth curve and was estimated by the software Origin 8.5. The glucose-lactate yield coefficient was calculated as shown in the equation below: where, Δlac is the variation of lactate concentration in the beginning and end of the exponential phase and Δlac is the variation of glucose concentration in the same interval.

In vitro terminal erythroid differentiation assay
The biological activity of the rhEPO produced by the recombinant cells was evaluated by the ability to induce erythroid differentiation. For this, peripheral blood mononuclear cells (PBMC) were obtained from the leukoreduction system chambers of apheresis platelet donors, with informed consent, according to the institutional ethics committee (CAAE 98950818700005440). CD34 + cells were isolated from PBMC by magnetic-positive selection-cell sorting (Miltenyi Biotec).
The culture method to enrich the erythroblast population was adapted from Freyssinier et al 36 and Hu et al. 37 This differentiation protocol was set into two phases. In phase 1, CD34 + cells were cultured for 7 days in Iscove's Modified Dulbecco's Medium (ThermoFisher) supplemented with 15% (v/v) of BIT9500 serum substitute (Stem Cell Technologies), 100 U/mL Penicillin-Streptomycin (ThermoFisher), 2 mM GlutaMax (ThermoFisher), 100 ng/mL Stem Cell Factor, 100 ng/mL interleukin IL-6, and 10 ng/mL interleukin IL-3 (Miltenyi Biotec). In phase 2, cells were cultured for 12 days in the previously described media with the replacement of IL-6 for 2 U/mL EPO either in-house recombinantly expressed in human cell lines SK-Hep-1, HKB-11, and Huh7 or commercially available CHO-expressed EPO (R&D Systems). The NC was cultured in EPO-deprived phase 2 media for the same period. Cells were cultivated at 37 • C, 5% CO 2 , and 85% humidity.
Cultured cells were immunophenotyped at days 0, 4, 8, and 12 of phase 2. Briefly, 1 × 10 5 cultured cells were washed and suspended in 2%FBS-PBS and then labeled with FITC-conjugated anti-CD235a, also known as Glycophorin A (GPA) (BD Biosciences). Labeled cells were then analyzed in a FACSAria I cytometer (BD Biosciences). Flow cytometry data analysis was conducted in FCS Express 6.06.0014 (DeNovo Software).
Cell morphology was analyzed by May-Grünwald-Giemsa (MGG) staining of cytospin samples. At days 0, 4, 8, and 12 of phase 2, 1 × 10 5 cells were washed and suspended in 200 μL of 1× PBS and then centrifuged in microscopy glass slides at 100×g for 5 minutes. After drying, slides were stained in May-Grunwald (Sigma MG500) for 5 minutes, washed in PBS for 1 minute, stained in Giemsa solution (Sigma GS500) for 15 minutes, and rewashed in PBS for 1 minute. The cells were imaged using an inverted microscope (Carl Zeis Axiovert A1).

Statistical analysis
All statistical analyses were conducted in GraphPad Prism version 6.05 for Windows (GraphPad Software).

RESULTS AND DISCUSSION
The host cell line chosen to produce the glycoprotein has a strong influence on the glycosylation since different host systems may express different glycosylation enzymes and transporters that influence the specificity and heterogeneity in glycosylation patterns. 4 Consequently, the selection of the appropriate host system is extremely crucial for the production of therapeutic glycoproteins. 1 To study the potential of human cell lines for industrial production of recombinant glycoproteins, we have generated recombinant human cell lines under suspension serum-free conditions by a lentiviral transduction approach. We have chosen EPO as our model for this study based on its molecular complexity, mainly due to its glycan profile. EPO is a glycoprotein that contains approximately 40% carbohydrate in its structure. 38 Carbohydrate structures are critical for many biological properties of recombinant proteins, these include pharmacokinetics, secretion, stability, receptor recognition and antigenicity, protein conformation, and biological activity. 39

Recombinant rhEPO producing-cells were successfully generated
Lentiviral-based gene transfer enables the generation of recombinant cell lines with higher and stable levels of transgene expression when compared to classical DNA delivery methods. 40 A lentivirus-derived vector carrying the coding region of a rhEPO gene, whose expression is linked to the GFP through an internal ribosome entry site, was used for the generation of stable recombinant Sk-Hep-1, HKB-11, and Huh-7 cells. We transduced the cells at a MOI of 1. Figure 1 shows the percentage of GFP positive cells (A) as well as the cell viability (B) after transduction. HKB-11 cells presented a higher percentage of positive GFP cells, 63%, followed by Sk-Hep-1 and Huh-7 with 50% and 34%, respectively. Regarding viability, only Huh-7 presented a decline, showing that this cell line can be more sensitive to the transduction processes. To enrich the population of rhEPO producing cells, GFP-positive cells were sorted by flow cytometry reaching levels >90% for all cell lines ( Figure 1C).

rhEPO production and levels of mRNA expression are not related to the gene copy number
After the generation of the pool of recombinant cells, we evaluated different culture media to select an appropriate for the production of rhEPO. We compared the culture medium previously selected for each cell line during the adaptation process (CDM4CHO for Sk-Hep-1 and HKB-11 and CD293 AGT cells for Huh-7 cells) 31 with Freestyle 293 medium, a culture medium widely used for recombinant protein production. To evaluate which media had the highest rhEPO productivity, each culture was monitored for 72 hours. As shown in Figure 2, Sk-Hep-1 (A) and HKB-11 (B) cell lines achieved a maximum rhEPO concentration of 38.35 μg/mL and 36.65 μg/mL when cultured in CDM4CHO medium, and 8.3 μg/mL and 16.55 μg/mL when cultured in Freestyle 293, respectively, confirming CDM4CHO as the best medium for cell growth as well as for rhEPO production. For the Huh-7 cells (C), the best rhEPO production was achieved with the Freestyle 293 medium when compared to CD293 AGT, reaching a maximum concentration of 130 μg/mL and 94.96 μg/mL, respectively, demonstrating that the medium elected for cell growth was not the one that presented the best volumetric protein production. rhEPO specific production of each cell line was also evaluated. As shown in Figure 3, Huh-7 cells had higher specific production than Sk-Hep-1 and HKB-11 cell lines.
We also compared the gene expression of the cells to the nontransduced controls and the cell line that presented the highest relative expression was the Huh-7, with the rhEPO gene 122 294-fold higher than its nontransduced control, followed by the Sk-Hep-1 line with expression 2352-fold and the HKB-11 1097-fold higher than its nontransduced control (Figure 4). At this point, these results corroborate with the higher rhEPO production by Huh-7.
Detrimental effects on karyotypic stability are related to the high gene copy number integrated into the genome of mammalian cells, especially after long periods of cultivation. High gene copy number might impair cell growth, cell survival and proliferation of surviving cells, outbalancing any beneficial effect of enhanced specific productivity. 41,42 On the other hand, high levels of integrated copies into the genome can improve the productivity of the cell. 43 In this research, we used an MOI of 1 to favor a low number of integrations per cell, targeting the stability of the recombinant cell populations. As shown in Table 1, HKB-11 cells had the higher number of inserted copies, with a total of 10.1 copies/cell followed by the Sk-Hep-1 with 8.3 copies/cell, whereas Huh-7 cells had a significantly lower value, with 1.97 copies/cell. Huh-7 cells had a lower gene copy number compared to Sk-Hep-1 and HKB-11, suggesting that The analysis was carried out by qPCR using 100 ng of genomic DNA extracted from SK-Hep-1, HKB11, and Huh-7 cell lines.

TA B L E 1 Number 1054-EPO lentiviral vector copies per cell
the high level of expression achieved by this cell line was not related to the number of copies integrated into the cell genome, but to the transcriptional activity of the integration site. Putting these results together it is possible to note that there was no correlation between the number of copies integrated into the genome of the cells and the rhEPO production level.

Growth and metabolic characterization
As shown in Figure 5   end of the culture. Cell viability remained above 90% throughout the culture. The maximum specific growth rate during the exponential growth phase was 0.0237 hour −1 ( Table 2). HKB-11 cells ( Figure 5B) showed strong cell growth up to 120 hours, reaching a maximum cell density of 5.74 × 10 6 cells/mL, with a maximum specific growth rate of 0.0218 hour −1 ( Table 2). These cells had no stationary phase as with the Sk-Hep-1 cell, with a decrease in cell density and viability in the final days of culture. At the end of 312 hours, both cell density and viability fell by approximately 50% in comparison to their respective maximum points. Huh-7 cells ( Figure 5C) showed a cell growth pattern similar to HKB-11 cells for up to 96 hours of culture; reaching a maximum cell density of 4.36 × 10 6 cells/mL, after that the cell density began dropping, resulting in a final cell concentration of 1.7 × 10 6 cells with a viability of 30% and a maximum specific growth rate of 0.0312 hour −1 ( Table 2). The cell density values achieved by all the cell lines are compatible to values obtained for the culture of the HKB-11 and SK-Hep-1 cells 31 and other human cells such as HEK-293 and PER.C6 under suspension serum-free conditions. [44][45][46][47][48] Understanding the metabolism of cell lines is an essential step in improving the performance of mammalian cell culture processes. Knowing the metabolism of the lineages allows for the characterization of substrates and metabolites that should be monitored and controlled in the culture medium to increase cellular growth and productivity. For Sk-Hep-1 F I G U R E 5 Growth kinetic and metabolism of recombinant Sk-Hep-1, HKB-11, and Huh-7 cell lines cultured under serum-free suspension conditions. Growth, glucose consumption, lactate production, glutamine consumption, glutamate production, and ammonia production by the Sk-Hep-1 (A), HKB-11 (B), and Huh-7 (C) cells (n = 2) cells it is possible to observe a decrease in glucose concentration for up to 120 hours of cultivation. Lactate, which is a product of glucose metabolism, showed an increase in the concentration at the start of cultivation, reaching 35 mM at 96 hours and remaining in the range of 18-32 mM until the end of cultivation. There was no total consumption of glutamine during cultivation, it reaches a minimum concentration of 1.25 mM in 240 hours, and remains in this level until the end of the culture. Regarding ammonia, which is a byproduct of glutamine and glutamate metabolism, Sk-Hep-1 cells had a rapid increase in ammonia concentration, reaching a maximum concentration of 2.5 mM at the end of the culture.
For the HKB-11 cell line it was not possible to observe a drop in glucose concentration in the first 2 days, but from then on it began to fall, stabilizing after 192 hours and maintaining a concentration of approximately 30 mM. Lactate increased significantly in the first 48 hours, subsequently reaching its maximum concentration in 264 hours (46 mM). The levels of glutamine consumption and glutamate production in the HKB-11 cells were lower than in Sk-Hep-1 cells. The concentration of ammonia increases more slowly than the Sk-Hep-1 cells, which can be correlated with the lower production of glutamate and consumption of glutamine. The maximum concentration of ammonia reached was 2 mM in 288 hours.
During the Huh-7 cultivation the glucose concentration began to fall after 24 hours, reaching its minimum concentration in 312 hours. The cells were more sensitive with the glucose drop and the increase of the concentration of the metabolites produced in comparison with the other cell lines tested, showing more significant and pronounced decreases in the number of total viable cells. The Huh-7 line reached a lactate concentration of 40 mM in 96 hours and a maximum ammonium concentration of 3 mM in 168 hours. All metabolites increased more markedly than in other lines and at higher levels. It is possible to infer from this that the Huh-7 cell had a considerably higher metabolic flux than those of the previous lines.
One of the main consequences of the metabolic behavior of in vitro cultured animal cells is the exacerbated formation of the lactate and ammonium ions. When these products exceed specific critical concentrations, they inhibit cell growth and may decrease the production of recombinant protein. 49,50 Lactate can cause a significant reduction in the pH of the medium, which may inhibit cell growth, but this effect can be avoided with a control system that keeps the pH within an optimal range. However, the primary mechanism by which lactate appears to affect cultures negatively is increasing osmolarity of culture media.
In all cultures, glucose exhaustion was not observed. All cell lines showed a rapid accumulation of lactate in the first 72 hours, which may be related to rapid growth. In general, concentrations less than 20 mM do not affect the cell, whereas 20-40 mM disrupt productivity, and above this value cell growth is inhibited. 51,52 As can be seen in Figure 5B and C, HKB-11 and Huh-7 cells reach lactate inhibitory concentration for mammalian cells in 264 and 96 hours respectively, which may be related to the decrease of viable cells and with cell growth limitation for Huh-7 cells (96 hours); Sk-Hep-1 cells did not reach the lactate concentration value considered inhibitory. Changes in extracellular pH and osmolarity as a result of increased lactate production are thought to be responsible for cell growth inhibition. 53 Also significant is the influence of this metabolite on the quality of the recombinant product, especially on glycosylation. Lactate has been reported to inhibit cell growth and cell metabolism and decrease productivity in CHO, BHK cells, and hybridomas, in part due to changes in osmolarity of culture environment. 50,54,55 The ammonium ion has a more significant impact on the culture of mammalian cells in vitro than lactate because concentrations of between 2 and 5 mM of ammonium ion can inhibit growth and impair productivity. 51 Furthermore, a high concentration of ammonium has been shown to impair the glycosylation pattern of rhEPO by decreasing the terminal sialylation, the content of the O-linked glycan, and the tetra-antennary N-glycan. 56 Sk-Hep-1 cells reached the ammonium concentration considered to be inhibitory in approximately 96 hours of culture when cells enter in the stationary growth phase. At this point lactate is also near inhibitory concentration. From then on the culture entered a stationary phase where it remains until the end of the culture, without presenting decay of the total viable cells. The HKB-11 cells showed a decrease in the number of viable cells shortly after reaching their maximum concentration in 120 hours, and the ammonium and lactate levels in this period are not considered as inhibitory. 51 The Huh-7 cell line reached both ammonium and lactate inhibitory concentration in the first 96 hours, which correlates with cell growth limitation and decay of viable cells. The intensity of Huh-7 metabolism could be a burden associated with higher mRNA expression of rhEPO as well as higher protein yield as shown in the previous results. The higher level of rhEPO expression could have led to the accumulation of unfolded or misfolded protein in the endoplasmic reticulum (ER) lumen of the Huh-7 cells inducing ER stress that can be detrimental and affect cell viability. 57 Regarding glucose specific consumption (qGlc) the values obtained for each cell were very similar to those found in another human cell line, HEK293, with a range from 1.75 to 7.92 PMol/cell/day. For qLac, HKB-11 showed a low value (1.128 PMol/cell/day) compared to those found in literature for HEK293 (3.05 to 16 PMol/cell/day). [58][59][60][61][62][63][64] For glucose-lactate yield coefficient (Ylac/Glc), HKB-11 also presented a low value (0.444 mmol/mmol) compared to other human cell lines (0.5-2.3 mol/mol), 59,61-64 which can be explained by a lactate consumption in exponential phase, a common phenomenon in HEK293, [60][61][62] one of the cell employed to establish HKB-11 cell line. 65

Huh-7 cells achieved higher levels of rhEPO production than Sk-Hep-1 and HKB-11 cells
rhEPO production was monitored throughout the 216-hours culture period using the ELISA assay. The volumetric rhEPO production of Sk-Hep-1, HKB-11, and Huh-7 achieved a maximum concentration of 112.5 μg/mL, 112.7 μg/mL, and 571 μg/mL (Table 2, Figure 6), respectively, and the production of all cells was sustained during the whole process, achieving the higher concentration at the end of the culture ( Figure 6). Corroborating with the previous results, Huh-7 cells produced higher rhEPO compared to HKB-11 and Sk-Hep-1 cells, which achieved virtually the same volumetric production. This yield is comparable to some results obtained using well-established production platforms for recombinant glycoproteins, like CHO, BHK, and HEK293. In 2002, Yoon and colleagues, analyzing the effects of temperature on the F I G U R E 6 Kinetic of rhEPO production of Sk-Hep-1, HKB-11, and Huh-7 cells cultured under serum-free suspension conditions. SK-Hep and HKB-11 cells were cultured in CDM4-CHO medium and Huh-7 cells in Freestyle 293 medium (n = 2) production of rhEPO by CHO cells achieved a maximum concentration of 109.2 μg/mL of rhEPO in cultures maintained at 33 • C and 42.8 μg/mL at 37 • C. 66 The qEPO values obtained by CHO cells (2.16 pg/cell/day at 37 • C and 8.4 pg/cell/day at 33 • C) were also comparable to the ones obtained in our present work, except for Huh-7 where the value was considerable higher. In a similar study using CHO cells, Ahn and colleagues reached a maximum concentration below 20 μg/mL in cultures maintained at 37 • C and approximately 140 μg/mL in cultures maintained at 32 • C. The highest qEPO obtained was 9.2 ± 0.9 pg/cell/day at 30 • C culture, 67 which is lower than the 22.272 ± 5.568 pg/cell/day by the Huh-7 cells used in this study.
After 13 days of culture, Gaillet and colleagues achieved a maximum rhEPO concentration of 206 μg/mL using stable CHO cells generated by lentiviral transduction and cultured in serum-free conditions. In that study, an MOI of 200 was used for the transduction of the cells with an average of 19 gene copies integrated per cell. 43 Also using lentiviral vectors for generation of stable CHO cells, Baranyi and colleagues 68 generated clones producing on average 50-250 μg/mL of rhEPO. In a recent work Chin and colleagues generated HEK293 cells producing EPO, adapted then to serum-free suspension conditions and obtained cells pools expressing 3680 to 21 730 U/mL (24.5-144.9 μg/mL) in batch mode. To further optimize rhEPO production the authors cultured the cells in a bioreactor operated under feed-batch conditions, where the cells were able to produce 92 700 U/mL (696 μg/mL) with qEPO of 18.1 pg/cell/day, values lower than obtained in this work for Huh-7 cells cultured in batch in Erlenmeyer flasks.
Despite the high expression of rhEPO by Huh-7 cells, its culture showed an inhibitory concentration of both ammonium and lactate, which can inhibit growth and decrease cell viability and may interfere with the glycosylation of this protein. However, there are some strategies to decrease the concentration of these metabolites during cultivation, such as substitution of glutamine by some ammoniagenic substrates such as glutamate, pyruvate, alpha-ketoglutarate, or a few dipeptides, fed-batch for glutamine replacement or limitation of the initial glutamine concentration 69 ; and in lactate case, a fed-batch strategy using galactose/glutamate. 51,70 The maximum rhEPO concentration obtained in the cultures of Sk-Hep-1, HKB-11, and Huh-7 cells are comparable and even superior in the case of Huh-7 cells to those found under similar conditions of serum-free or nonserum-free suspension cultures. Sk-Hep-1, HKB-11, and Huh-7 cells have also proven to be able to produce high amounts of other complex recombinant glycoproteins. In 2017, a study using the Sk-Hep and HKB-11 cells showed that these cells were able to secrete 2.6-5.0-fold more recombinant human coagulation factor VII (rhFVII) (585.4-1176.6 ng/mL) than BHK-21 cells (222.6 ng/mL) under static serum-supplemented culture. 27 Bomfim and colleagues showed that the amount of recombinant human coagulation factor IX (rhFIX) secreted by SK-Hep-1 cells was 1.6-fold higher than by 293T cells and the biological activity of rhFIX secreted by SK-Hep-1 cells was 4-fold higher (186 mUI/10 6 cells) than by 293T (47 mUI/10 6 cells). 26

Produced rhEPO was able to sustain differentiation of CD34+ cells into erythroid cells in vitro
To evaluate the biological activity of the produced rhEPOs, we performed an erythroid terminal differentiation assay with peripheral blood-derived hematopoietic progenitor cells. 37 To achieve this, we measured the expression of GPA, a major Representative images of the morphology of cytospinned MGG stained cells from day 0 and day 12 from each group (63× magnification). The negative control represents CD34 + cells cultured in EPO-deprived phase 2 media. *Two-way ANOVA and Turkey's multiple comparison correction test computed with 95% Confidence Interval erythroid surface marker 71 during a 2-phase culture system. On Day 0 of Phase 2 of each independent experiment, samples presented similar levels of GPA positive cells ( Figure 7B). Except for the NC (Phase 2 culture deprived of rhEPO), all samples displayed an important increase of GPA positive cells throughout the experiment, with the peak on Day 4. Of note, cells treated with rhEPOs showed a comparable performance to the cells treated with the commercial EPO regarding the percentage of GPA positive cells. Since erythropoiesis is an EPO-dependent process, it is expected that samples deprived of EPO should not terminally differentiate. 72 This fact was supported by the morphologic characteristics of the samples among the different groups ( Figure 7A). NC cells preserved a morphology similar to the early erythroid progenitors, maintaining a Day 0-like aspect. In contrast, cells cultured either with the commercial or rhEPOs presented a higher number of terminally differentiated erythroid cells and even enucleated reticulocyte-like cells. These results showed that the rhEPO produced by Sk-Hep-1, HKB-11, and Huh-7 cells under serum-free conditions was functional and biologically active, with no significant difference when compared to a commercially available EPO produced in CHO cells.

CONCLUSION
In the present study, we show that the Sk-Hep-1, HKB-11, and Huh-7 cell lines were easily genetically modified to produce a recombinant protein in serum-free suspension conditions. Besides, the cells presented good cell growth characteristics and were able to produce high levels of a recombinant complex glycoprotein, EPO. HKB-11 cells stood out for their ability to grow at high cell densities in batch culture. The Huh-7 cell line showed superior results when compared to Sk-Hep-1 and HKB-11 cells regarding the level of EPO produced, which makes it a promising alternative for the production of recombinant proteins. This cell line however produced the higher amount of ammonia. The effect of this by-product on recombinant protein quality attributes must be evaluated. Further studies are also needed in order to develop specific serum-free culture media for these cells as well as to establish optimized culture conditions to enable an enhanced cell growth and recombinant protein production profile. The scalability and compatibility of these cells with a bioreactor are also essential parameters that need to be evaluated. The glycosylation profile of the recombinant proteins produced by the cells is also a relevant feature to be analyzed, since this feature can demonstrate the quality and heterogeneity of the protein produced, as well as influence the selection of the later production strategies.