Development of an animal-derived component-free medium for Spodoptera frugiperda (Sf9) cells using response surface methodology

To develop an animal-derived component-free medium for Spodoptera frugiperda (Sf9) growth and green fluorescent protein (GFP) expression. OSF9-ADCFM contained optimum concentrations of CDLC, YE and ST at 0.5% (v/v), 11.0 g/L, and 3.0 g/L, respectively. A mean viable cell concentration of 1.71 ± 0.14 × 105 cells/mL was obtained from 5 passages (P1-P5). The use of both peptones after 10 kDa ultrafiltration had a significant effect on Sf9 cell growth. Grace’s insect medium with 10% FBS gave higher un-infected cell number than SF-900II and OSF9-ADCFM for 4.29 and 5.38 times, respectively. The average cell number of un-infected cells and GFP-fluorescent cells of SF-900II were higher than OSF9-ADCFM 1.25 and 7 times, respectively. In-house OSF9-ADCFM could support growth and GFP expression in Sf9 less than commercial SF-900II. However, it could lower the production cost at least 50% comparing to commercial SF-900II. The development of in- house OSF9-ADCFM would be continued to increase both cell numbers and protein expression in the next step.


Introduction
Insect cells (IC) in conjunction with the baculovirus expression vector system (BEVS), or IC-BEVS, have been widely used to produce recombinant proteins, vaccines, and vectors for gene therapy (Drugmand et al. 2012;Ikonomou et al. 2003;van Oers et al. 2015). Spodoptera frugiperda (Sf9) is a normal cell line capable of growing in serum-free medium without any added growth factors. Sf9 cells can secrete growth factors (IGF-I) to regulate their own cell growth mechanisms (Doverskog et al. 1997). Typically, the basic formulations of the four most commonly used insect cell culture media are Grace's, Hink's TNM-FH, TC-100, and IPL-41. All four media are generally supplemented with 10% serum (Schlaeger 1996). In animal cell culture, sera, especially those derived from fetal bovines or calves (FBS or FCS), have been supplied at 2-20% (v/v) in a basal medium to stimulate cell growth and other cellular activities . Serum contains growth factors, albumin, transferrin, hormones, protease inhibitors, lipids, minerals, and other essential nutrients that facilitate the attachment and spreading of cultured cells (Hewlett 1991;Gstraunthaler 2003;Freshney 2010).
However, there are several disadvantages of using sera in cell culture media, including physiological variability, shelf life and consistency, quality control, specificity, availability, downstream processing, contamination, cost, growth inhibitors, and standardization (Freshney 2010). The collection of serum causes unneeded pain to unborn calves (van der Valk et al. 2004). In addition to the risk of contamination during the serum production process, particularly from transmissible spongiform encephalopathies (TSEs) or bovine spongiform encephalopathies (BSE), the use of animal-derived materials is not recommended for serum-free medium development (Gstraunthaler 2003).
Serum-free media are currently media that are designed to grow a specific cell type or to be used only in the absence of serum. It enables cell culture to be carried out under specific conditions with no confounding variables. Bacto peptone (Keay 1977;Sakoda and Fukusho 1998), peptone and casein hydrolysate (Saha and Sen 1989), tryptose peptone (Zhang et al. 1994a), casein peptone (Dyring et al. 1994), Primatone RL (Schlaeger 1996;Gu et al. 1997;Nyberg et al. 1999), yeastolate or yeast extract (Sung et al. 2004) have been used in combination with a basal medium to formulate animal-derived free medium. Shen et al. (2007) summarized that the yeastolate fraction could promote growth activities of Sf9 ranging from 32 to 80% compared to the control group. Soy peptone, a plant-derived protein hydrolysate, has been widely used for serum-free media development of both mammalian and insect cells (Ikonomou et al. 2003;Kwon et al. 2005;Agathos 2007;Chan and Reid 2016). Apart from yeastolate and soy peptone, chemically defined lipid concentrate is currently commercially available and designed to reduce or replace fetal bovine serum in cell culture media for a wide range of applications, including CHO, hybridoma, insect cell culture growth and maintenance, monoclonal antibody production by hybridomas, and viral expression in insect cells (Hink 1991;Goosen 1991;Schlaeger 1996;Parampalli et al. 2007;Batista et al. 2008).
Although there are many commercially available serum-free media (SFM), animal-derived component-free media (ADCFM), and chemically defined media (CDM), low-cost serum-free media or in-house serum-free media are still challenging alternatives to reduce the cost of insect cell cultures along with baculovirus cultures. Herein, three independent variables, namely, Bacto™ Soytone (ST), Bacto™ yeast extract (YE), and chemically defined lipid concentrate (CDLC), are chosen for optimization as serum substitutes using a systematic design experiment, central composite design (CCD). The growth and maintenance of Sf9 cells in the in-house animal-derived component-free medium were investigated both in static and suspension cultures. The effects of the combination of the microfiltration and ultrafiltration of both ST and YE on Sf9 cell growth were studied. In addition, green fluorescent protein (GFP) expression was also examined in the developed medium.

Cell line and cell culture
The insect cell line Spodoptera frugiperda (Sf9) was purchased from the American Type Culture Collection (ATCC CRL-1711, Manassas, VA, USA). The cells were grown in a 5 mL serum-containing (SC) medium that consisted of Grace's insect medium, 10% (v/v) FBS, 3.3 g/L lactalbumin, and 3.3 g/L yeast extract in a 25 cm 2 tissue culture flask at 28 °C for 4 days. After that, the cells were dislodged from the flasks using a cell scraper (SPL Life Science, Gyeonggi-do, Korea). For cell counting, 10 μL of cell culture sample was diluted by adding 10 μL of a 0.4% trypan blue dye solution and then loaded into a counting chamber. Total cells, live or viable cell concentrations, and dead cell concentrations were measured by using a Countess™ II FL automated cell counter (Thermo Fisher Scientific, USA). For subcultivation, a viable cell concentration of 4 × 10 4 cells/cm 2 obtained from a previous culture was inoculated into a new 25 cm 2 tissue culture flask containing 5 mL of SC medium and then incubated under the same conditions as mentioned above. Cells were passaged every 4 days to a cell density of 1.2 × 10 5 -1.6 × 10 5 cells/cm 2 and maintained in SC medium for 5 passages before use.
The basal medium used in all experiments was modified from Grace's insect medium by adding other supplements. The basal medium had the following composition: Grace's insect medium, 45.7 g/L; sodium hydrogen carbonate, 0.35 g/L; d-glucose, 9 g/L; vitamin B12, 0.00015 g/L; methyl cellulose, 0.1% (w/v); MEM vitamin solution (100x), 0.5% (v/v); and l-glutamine, 0.876 g/L. Pluronic F68 (0.1% w/v) was added to the basal medium for suspension culture. The basal medium was filtered sterile using a 0.22 μm membrane filter before use. For the experiments, the three independent variables to be studied (CDLC, YE, ST) were separately added to the basal medium according to the specified concentration in the design of experiments.

Experimental design
The software Minitab R16 (Minitab Inc., USA) was used for designing the experiments and the statistical analysis of the data. Response surface methodology (RSM) was used to determine the influence of the supplements (CDLC, YE, ST) on the response (viable cell yields). A circumscribed central composite design (CCD) for three independent variables at five levels each was constructed as shown in Table 1. The independent variable coding regions were − α (− 1.682, lowest level), − 1, 0 (middle level), + 1, and + α (+ 1.682, highest level). The experiments were simultaneously performed in triplicate for 4 days. The treatment combinations of the three independent variables on the responses (Y) are presented in Table 2.

Cell culture
The starter cells obtained were scraped and washed three times in phosphate-buffered saline (Ca 2+ , Mg 2+ -free PBS, pH 7.4) by centrifugation at 110×g for 10 min. After that, the cell pellets were resuspended in PBS to obtain a cell concentration of 10 6 -10 7 cells/mL. The initial cell density of 4 × 10 4 cells/cm 2 was placed in each 6-well plate containing 3 mL of a tested medium, and then the plates were incubated at 28 °C for 4 days. The experiment for each tested medium was carried out in triplicate for 4 days. A small volume of suspended Sf9 cells was placed into a 1.5 ml microtube, and then the concentration of viable cells was counted by using a Coun-tess™ II FL automated cell counter as mentioned above.

Statistical analysis
The responses (Y) were first plotted according to the Anderson-Darling statistic in Minitab software (Minitab Inc., USA) for statistical analysis. The probability plot of the responses (as residues) on a specified day of the cultures is constructed. If the p-value of the Anderson-Darling statistic was ≥ 0.05, the data were normally distributed. Next, regression analysis was performed using the analyze response surface design function, and terms including linear, linear + squares, linear + interactions, and full quadratic were selected. The value of the coefficient of determination (R 2 ) was used to evaluate the sufficiency of the model. The selected model depended on the selected terms that always gave a p-value ≤ 0.05. The reduced model based on the condition t-test in the function was evaluated using the R 2 value. A second-order or higher-order polynomial regression model was fitted as described in Eq.
(1) using the data presented in Table 2. (1) where Y is the viable cell yield, b 0 is the intercept value, X is the Factors (X1 = CDLC, X2 = YE, X3 = ST), b z is the Regression coefficient (z = i, ii or ij, where i < j), ε is the Residual error, k = 1, 2, 3… The p-value of the lack-of-fit for the selected model should be ≥ 0.05. A pair sample Student's t-test was used to determine the significance of differences among the groups in the experiments.

Verification study
The optimum concentrations of the three independents in the basal medium (named "OSF9-ADCFM") were validated. Cell inoculums were cultivated in SC medium. The initial viable cell density of 4 × 10 4 cells/cm 2 was used to inoculate a 25 cm 2 tissue culture flask, and then a volume of the culture medium was added up to 5 mL. The cultured flasks were carried out in triplicate. The cells in the cultured flasks were incubated at 28 °C for 3-5 h until cells were attached to the flasks. After cell attachment, the spent medium was carefully removed and gently replaced with 5 mL of OSF9-ADCFM. The cultured flasks were continuously incubated at 28 °C for 4 days. Subsequently, the cells were gently harvested using a cell scraper. The concentration of viable cells was measured as mentioned above. The cells obtained from this first optimized culture were named passage 0 (P0). For subculture in the next passage, the P0 cells were centrifuged at 110×g for 5 min, and then the cell pellets were resuspended in OSF9-ADCFM at a specified volume. After that, the initial viable cell density used for the experiments was equal to that used above, and 5 mL OSF9-ADCFM was used instead of SC media. The efficiency of OSF9-ADCFM in maintaining and supporting the growth of cultured cells was recorded as the viable cell yield, and the cultured cells were continuously subcultured for 5 passages.
Study of the optimum cell density in static culture The optimum cell density was evaluated using various initial cell densities placed in a 25 cm 2 tissue culture flask containing 5 mL of OSF9-ADCFM: 2 × 10 5 , 3 × 10 5 , 4 × 10 5 , and 5 × 10 5 cells/mL (equal to 4 × 10 4 , 6 × 10 4 , 8 × 10 4 and 1 × 10 5 cells/cm 2 , respectively). Each initial cell density was carried out in triplicate. All flasks were incubated at 28 °C for 4 days. The viable cells were dislodged from the flasks using a cell scraper and then counted as previously mentioned.
The combined effect of microfiltration and ultrafiltration of YE and ST on Sf9 cell growth Stock solutions of YE and ST were made by dissolving either YE or ST powder in distilled water. For the experiments, YE and ST were filtered through a 0.22 μm membrane filter (named "Fraction #1ST" and "Fraction #1YE"); then, Fraction #1 was ultrafiltered through a membrane cassette (Pall Corporation, USA) with a molecular weight cut-off of 10 kDa (named "Fraction #2ST" and "Fraction #2YE"). At the lower cut-off, Fraction #2 was ultrafiltered through a membrane cassette with a molecular weight cut-off of 3 kDa (named "Fraction #3ST" and "Fraction #3YE"). The filtered products obtained from Fraction #1, Fraction #2, and Fraction #3 were kept at 2-8 °C before use. For the experiments, the combinations of fractions obtained from both ST and YE (RUN1-9) are shown in Table 3. The final concentrations of YE and ST were 11 g/L and 3 g/L, respectively, while the final concentration of CDLC was 0.5% (v/v). Regarding the final concentrations of both YE and ST in terms of protein content, the total protein concentrations of YE and ST were measured using a NanoDrop (ND-1000) spectrophotometer (Thermo Scientific, USA). The starter cells were grown in OSF9-ADCFM in 175 cm 2 tissue culture flasks, and then the cells were scraped and washed three times in PBS. Cells with a density of 4 × 10 4 cells/cm 2 were placed in 25 cm 2 tissue culture flasks containing 5 mL of each fraction combination and carried out in triplicate. All flasks were incubated at 28 °C for 4 days. Subsequently, the cells in all flasks were scraped and mixed well before a small volume of cell sample from each tested medium was taken to count the viable cell concentration using the method mentioned above. The osmolality of the medium was measured using a Cryoscopic Osmometer (OSMOMAT® 030, Gonotec Gesellschaft für Meẞ-und Regeltechnik mbH, Germany).

Adaptation of Sf9 cells to grow in suspension
Suspension cultures of Sf9 cell growth in OSF9-ADCFM were investigated. An initial viable cell concentration of 4 × 10 4 cells/cm 2 was grown in a 175 cm 2 tissue culture flask containing 30 mL OSF9-ADCFM medium at 28 °C for 4 days. Then, a viable cell density of 3 × 10 5 cells/mL was inoculated in 250 mL spinner flasks containing 50 mL of medium. The spinner flasks were stirred at 50 rpm and 28 °C. Cell culture samples were taken daily to count the viable cells according to the method described above. The subcultivation of cells was carried out when the concentration of viable cells reached 1-2 × 10 6 cells/ mL by taking the viable cell density of 3 × 10 5 cells/ mL and placing in a new spinner flask containing 50 mL of fresh medium. The cells were continuously cultured for 5-7 days. The morphology of cultured Sf9 cells was photographed. The evaluation of cell growth in the spinner flasks was carried out separately for two batches.
Transient transfection of Sf9 cells with plasmid pEGFP-N1 of recombinant green fluorescent protein (GFP) Sf9 cells were separately cultured in each medium (Grace's insect medium plus 10% FBS, SF-900II and OSF9-ADCFM) with an initial cell density of 5 × 10 5 − 1 × 10 6 cells/mL in T75 flasks at 28 °C for 3-4 days. Cells were dislodged using a cell scraper and then counted using a Countess TM II FL automated cell counter based on the trypan blue dye method. Before transfection, cells were subcultured continuously in each medium for 3-5 passages. One hundred microliters of cell suspensions obtained from each medium culture were seeded into a 96-well plate with 2 × 10 4 cells/well. The volume of Cellfectin™ II reagent at 0.2, 1, and 2 µL were separately added into a 1.5 mL microtube containing 10 µL of Grace's insect medium (unsupplemented), and then all microtubes were gently mixed. One hundred-fifty nanograms of plasmid pEGFP-N1 were separately added into a 1.5 mL microtube containing 10 µL of Grace's insect medium (unsupplemented), and then all microtubes were gently mixed. After that, the diluted plasmid pEGFP-N1 was added into the diluted Cellfectin™ II reagent tube, gently mixed, and then incubated at room temperature for 15-30 min. Twenty microliters of the plasmid pEGFP-N1-Cellfectin™ II reagent complexes were added into each well, and then the plate was incubated at 28 ℃ for 72 h The expression of gene encoding a green fluorescent protein (GFP) was monitored daily under a fluorescence microscope. At 72 h post transfection, the cells obtained from each medium were collected and counted using a Countess TM II FL automated cell counter. The residual amount of glucose in each medium was measured using a YSI2900 Biochemistry Analyzer.

Regression analysis of the selected model
The analysis of variance (ANOVA) for the data of viable cell yield on day 4 of culturing indicated that Fraction #2ST + Fraction #2YE  (Table 5). The coefficient of determination or R 2 value for the full quadratic model was 94.70%, while R 2 (adjusted) was also high at a value of 87.88%, indicating that the correlation of all variables and the response could be described by the model. Moreover, it was also found that the p-value of lack-of-fit was relatively high at a value of 0.072 (Table 4), which indicates that the model was adequate for estimating the response. To construct the model, all coefficients, and symbols of all variables in the forms of linear, squares, and interactions were arranged in a series as presented below. The full quadratic model in the form of the mathematical equation is shown in Eq. (2).
The optimum points of three independent variables derived from the Response Optimizer function of Minitab R16 software (Minitab Inc., USA) were 0.5% (v/v) CDLC (X1), 11 g/L YE (X2), and 3 g/L ST (X3). Generally, the cell concentration estimated according to the empirical mathematical model was 1.98 × 10 5 cells/mL. Three-dimensional (3D) response surface plots were created by Design Expert 13.0 software to examine the optimum concentration for all possible combinations (Fig. 1). The interaction between CDLC + YE in this range showed a strong effect on Sf9 growth (p-value ≤ 0.05). Increasing CDLC concentrations resulted in a decline in viable cells, while YE promoted better growth (p-value ≤ 0.05). The interaction between CDLC + ST and ST + YE in the tested range indicated that they slightly affected Sf9 growth (p-value ≥ 0.05). ST and YE have been used extensively as medium substitutes as sources of vitamins, amino acids, peptides, and carbohydrates in serumfree medium to improve cell density and productivity (Posung et al. 2021). Spearman et al. revealed that soy hydrolysate stimulated the growth of CHO cells, while yeast hydrolysate caused a reduction in their growth (Spearman et al. 2014). On the other hand, the yeastolate fraction could promote the growth activities of Sf9 ranging from 32 to 80% compared to the control group. (Shen et al. 2007). Andreassen et al. concluded that YE alone could restore cell growth and replace FBS during bovine muscle cell cultivation in serum-free media (Andreassen et al. 2020). However, which components of yeast extract are responsible for the growth-promoting effects could not be fully described (Mosser et al. 2015). A low dose (1 g/L) of ST could stimulate the growth of keratinocyte cells up to 201%, whereas higher concentrations resulted in growth inhibition (Lee et al. 2008). Lee et al. also suggested that adherent cell cultures may need lower protein sources than suspension cultures, such as CHO cells (Lee et al. 2008). In addition, amino acid utilization in cells results in ammonia byproduct accumulation. Therefore, high amino acids can cause cell growth inhibition by high ammonia content in the medium (Zhang et al. 1994b). In this circumstance, YE could support better Sf9 growth than ST. This might be because the protein compositions of The analyzed model revealed that increasing ST from 3 g/L to 11 g/L led to lower cell yield (data not shown). This inhibition is probably due to some components in ST. Lipids and related components serve biological functions: cell membrane components, nutrient storage and transport, and signal transduction (Yao and Asayama 2017). Formulation of serum-free medium is necessary to include lipids at a concentration range of 10-100 µg/L, whereas insect cells need much higher concentrations at 1,000 µg/L (Shen et al. 2004). Cholesterol in CDLC is an important constituent in serum-free media for Sf9 because they cannot synthesize by themselves (Mitsuhashi 1989). The analyzed model revealed that CDLC at 0.5% (v/v) gave the highest viable cell yield (data not shown), while increasing the concentration in the formulated medium caused a decline in the cell yield. This was probably due to the higher hydrophobicity of CDLC, which directly contacts cells without protection from FBS and could reduce Sf9 cell growth (Batista et al. 2008).

Verification study
The verification study was performed using basal medium supplemented with three independent variables at the optimum concentrations. The medium containing 0.5% (v/v) CDLC, 11 g/L YE and 3 g/L ST was named OSF9-ADCFM. The viable cell concentration of Sf9 cells grown in OSF9-ADCFM during continuous subcultivation for 5 passages (P1-P5) is presented in Fig. 2. In the first passage P0, Sf9 inoculum was first transferred from serum-containing medium in which residual serum proteins in inoculum might stimulate cell growth after replacement with OSF9-ADCFM. A maximum viable cell concentration of 1.91 ± 0.15-1.93 ± 0.15 × 10 5 cells/ mL (P2-P4) was obtained with a relative cell number greater than 107% compared to P0, and thereafter, the viable cell concentration rapidly decreased (1.20 ± 0.10 × 10 5 cells/mL) with a relative cell number of 67%. However, even the number of viable cells could not increase at the last passage P5, but the percent of cell viability was still high enough (97% ± 2%) to be used for long-term subcultivation. Overall, an average viable cell concentration of 1.71 ± 0.14 × 10 5 cells/mL and an average percent cell viability of 96.8% ± 1.0% were obtained from 5 passages (P1-P5).
The obtained average viable cell yields under optimal conditions were in the accepted range of 1.68 × 10 5 − 2.27 × 10 5 cells/mL, as calculated from the empirical mathematical model, indicating the suitability of this model (Posung et al. 2021). The osmolality of OSF9-ADCFM was 511 ± 3 mOsm/kg (n = 3), which was much greater than that of commercial serum-free media at 345-359 mOsm/kg (Doverskog et al. 2000). However, the reported values of medium osmolality for insect cells could vary from 250 to 500 mOsm/kg (Zhang et al. 1994a). For the cultivation of Trichoplusia ni BTI-TN-5B1-4 (High-Five™), it was found that the increase in osmotic pressure from 350 to 500 mOsm/kg caused no significant differences in the cell proliferation rate during the exponential growth phase (Olejnik et al. 2003). Batista et al. (2005) found that Grace's medium supplemented with 2.7 g/L glucose, 8 g/L YE, 0.1% (w/v) Pluronic F-68, 1% (w/v) milk whey ultrafiltrate (MWU), and 3% (v/v) FBS had osmolality up to 438 mOsm/kg, but it could stimulate Sf9 cell growth by approximately fivefold (4.7 × 10 6 cells/mL) compared to Grace's medium containing 10% (v/v) FBS (9.5 × 10 5 cells/mL). In addition, Batista et al. (2008) also found that supplemented IPL-41 containing 6 g/L yeastolate ultrafiltrate, 10 g/L glucose, 3.5 g/L glutamine, 0.5 g/L fructose, 2 g/L lactose, 0.6 g/L tyrosine, 1.48 g/L methionine, and 1% (v/v) lipid emulsion had osmolality up to 460 mOsm/kg, but it could still support growth of Drosophila melanogaster S2 (S2AcGPV) up to the maximal cell concentration of 19 × 10 6 cells/mL. We noticed that a modified Grace's insect medium used as such a basal medium had an osmolality of 413 ± 2 mOsm/kg (n = 3), and it was possible that the addition of YE and ST at 11 g/L and 3 g/L could significantly increase the osmolality of Fig. 1 Viable cell yield in 3D response surfaces (left) and contour plots (right): A effects of CDLC and YE concentration, B effects of CDLC and ST concentration, C effects of YE and ST concentration. All 3D response surfaces and contour plots were created using Design-Expert® v13 software (Stat-Ease, Inc. MN, USA) ◂ OSF9-ADCFM. We assumed that osmolality did not affect Sf9 growth; if so, they would not grow in even the first passage. Lower cells yield in serum-free medium is common because of inadequate nutrients compared to medium containing serum.
Optimum cell density in the static culture The optimum initial cell density in static culture was tested in the range of 2-5 × 10 5 cells/mL in OSF9-ADCFM. The use of a low cell concentration (2 × 10 5 cells/mL) could not achieve a high cell concentration at the end of 5 days of culture (Fig. 3A). However, when the initial cell concentration was higher than 2 × 10 5 cells/mL, the final cell concentration was noticeably higher. An initial cell concentration of 3-5 × 10 5 cells/mL gave almost the same level of final cell concentration. Nevertheless, an initial cell concentration of 3 × 10 5 cells/mL could give the maximum multiplication ratio of 6.69 ± 0.15, while those of the initial cell concentrations of 2 × 10 5 , 4 × 10 5 , and 5 × 10 5 cells/mL were 1.34 ± 0.07, 5.32 ± 0.17, and 4.05 ± 0.09, respectively (Fig. 3B). Parizi et al. concluded that the optimum initial cell density of BHK1 cells in suspension culture was 3 × 10 5 cells/ mL with the highest multiple ratios of 3.7 (Parizi et al. 2017). High cell density engendered competition for obtaining limited nutrients and influenced the pH of the medium. Meanwhile, some signals produced from cells could stimulate the growth of others (Parizi et al. 2017). Spens suggested that increasing the amount of serum or specific growth factors in the medium could decrease the critical initial cell density (cID) (Spens 2006). Moreover, after inoculation of initial cells below cID, viable cell density declined before they started to proliferate (Spens 2006). The optimum initial cell density would be different between the cell lines and medium used. Hence, the optimum initial cell concentration for Sf9 growth in OSF9-ADCFM was 3 × 10 5 cells/mL.

Effect of microfiltration and ultrafiltration components of YE and ST
The effect of the protein fraction on Sf9 growth was determined by microfiltration and ultrafiltration of YE and ST, and then each fraction was combined, as shown in Table 3. The fraction #1 were crude yeast extract and crude soytone, fraction#2 had the protein molecular weight less than 10 kDa and fraction#3 had the protein molecular weight less than 3 kDa. The results showed that the viable cell concentrations obtained from the combinations of microfiltration and ultrafiltration of YE and ST at Run 6, Run 8, and Run 9 were 6.27 ± 0.07 × 10 5 , 5.98 ± 0.13 × 10 5 , and 7.17 ± 0.08 × 10 5 cells/mL, respectively, which were relatively higher than those of other combinations, as shown in Fig. 4. It was noticed that only some combinations of microfiltration and ultrafiltration of both YE and ST could stimulate growth of Sf9 cells, namely, Run 6: Fraction #3ST + Fraction #2YE, Run 8: Fraction #2ST + Fraction #3YE, and Run 9: Fraction #2ST + Fraction #2YE. We observed that the combination of high molecular weight YE and ST (Run 1) could not support good growth of Sf9 compared to others. Moreover, Run 5, which contained #3ST and #3YE, showed the lowest viable cell concentration. Fraction #3 contained all free amino Fig. 2 The verification study of OSF9-ADCFM on Sf9 viable cell yield obtained from continuous sub-cultivation for 5 passages (P1-P5). The viable cells were counted on day 4 of culturing and expressed as a mean value obtained from a triplicated experiment acids, which have molecular weights in the range of 75-204 daltons, and some oligopeptides (Mustățea et al. 2019). However, some carbohydrates may be removed from this fraction. Therefore, there might not be enough nutrients for Sf9 to grow in this combination. Run 1, 2, and 4, which showed similar viable cell concentrations, all contained fraction #1 and lacked fraction #2. The combination in Run 3, and 7, both contained fraction #1 and fraction #2, exhibited better growth than Run 1, 2, and 4. Run 6, 8 and 9 all contained fraction #2. We believed that the high molecular weight of YE and ST in fraction #1 may contain some inhibition factor for Sf9, whereas the 10 kDa in fraction #2 of both YE and ST might be suitable for its growth. However, the scientific reason for such a synergistic effect still requires further explanation. The paired samples Student's t-test indicated that the final cell concentration obtained from Run 9 was significantly different from the other two. In fact, ST and YE supplemented in Run 9 were the same composition as in OSF9-ADCFM. Mendonça et al. (2007) found that low molecular weight (LMW) YE (less than 30 kDa) had a positive effect on the growth of Sf9 cells better than high molecular weight (HMW). Shen et al. (2007) used sequential ethanol precipitation of yeastolate ultrafiltrate (YUF1) combined with other YUFs to culture Sf9 cells, and they showed a synergistic effect to enhance Sf9 growth. Additionally, Chou (2013) noticed that the low molecular weight of yeastolate at less than 10-14.2 kDa could fully support cell growth of IPLB-Sf-21AE (Sf21). Our results revealed that ultrafiltration (10 kDa cut-off) of YE and ST could enhance Sf9 growth, which corresponds to reports from a previous The final cell concentration is expressed as the mean ± standard deviation (error bar) value obtained from a triplicate experiment. Paired samples Student's t test was carried out, and the p-values were as follows: * p-value = 0.010, ** p-value = 0.011, *** p-value = 0.015 study. However, there is not much information on the synergistic effect of ST on insect cells. Nevertheless, integration of an ultrafiltration step in media preparation can eliminate hydrolysate lot variability and ensure reproducible good growth of insect cell lines (Schlaegae 1996;Ikonomou et al. 2003).

Adaptation of Sf9 cells to grow in suspension
The adaptation of Sf9 cells to grow in suspension was carried out in 250 mL spinner flasks containing OSF9-ADCFM with an initial cell concentration of 3 × 10 5 cells/mL. The growth of Sf9 cells was investigated separately in two batches, as shown in Fig. 5. The maximum viable cell concentrations obtained from batches no. 1 and no. 2 were 1.79 × 10 6 and 2.17 × 10 6 cells/mL, respectively. The maximum multiplication ratios of batches no. 1 and no. 2 were 4.96 on day 5 and 6.23 on day 4, respectively. Hensler et al. (1994) believed that the lag phase is population density-associated. They suggested that seeding densities should be higher than 6 × 10 5 cells/mL to overcome the long culture lag of Sf9 growth in serum-free medium (Hensler et al. 1994 for protein expression in mammalian cells, was used. Cell number of un-infected Sf9 cells in Grace's insect medium with 10% FBS, SF-900II and OSF9-ADCFM were 1.21 × 10 6 , 2.82 × 10 5 and 2.25 × 10 5 cells/ml, respectively. The normal medium could give higher cell numbers than SF-900II and OSF9-ADCFM for 4.29 and 5.38 times, respectively. Growth of Sf9 in OSF9-ADCFM was lower than SF-900II only 1.25 times whereas the price of OSF9-ADCFM is cheaper than SF-900II almost 50%. The size of un-infected Sf9 cells in SF-900II and OSF9-ADCFM were 15.53 and 14.02 µm, respectively, which was smaller than that of serum-containing medium of about 17.55 µm (Fig. 7). Transient transfection of Sf9 cells with plasmid pEGFP-N1 of recombinant green fluorescent protein (GFP) expressed a very low cell number of positive GFP-fluorescent even in medium supplemented with serum. The highest number of GFP-fluorescent cells in Grace's insect medium plus 10% FBS was only 34 cells/well while the total cell numbers was 4.57 × 10 5 cells/ml. (4.57 × 10 4 cells/well). It seems that some components in serum might gave good support on Sf9 transfection because SF-900II and OSF9-ADCFM gave very low GFP-fluorescent cells, with an average number of GFP-fluorescent cells less than 13 cells/well in SF-900II and less than 2 cells/well in OSF9-ADCFM (Fig. 8). Higher volume of Cell-fectin™ II reagent was toxic to Sf9 cells. In Grace's insect medium with 10% FBS, cell number was lower than un-infected cells 2.62-3.77 times. Dead cells in SF-900II and OSF9-ADCFM infected with 2 µL Cellfectin™ II reagent reached 85.35% and 94.30%, respectively. The relative low number of GFP-fluorescent cells in these 3 kinds of medium might be because the CMV-IE promoter was not suitable for gene expression in insect cells while polyhedrin promoter (Ppolh) gave better efficiency (Li et al. 2014).

Conclusion
Animal-derived component-free medium development by CCD indicated that CDLC exhibited a strong negative effect, YE had a slightly positive effect and ST had a slightly negative effect on with an initial viable cell concentration of 3 × 10 5 cells/mL, cultured at 28 °C and with a 50 rpm stirring rate. The viable cell concentration of two batch cultures was expressed as the mean value of a triplicated experiment Fig. 6 The morphology of Sf9 cells growing in OSF9-ADCFM in 250 mL spinner flasks containing 50 mL of Run9 medium on day 2 of culture. Black circles indicate cell aggregation. Scale bar is 50 µm viable cell yield. CDLC, YE, and the combination of CDLC and YE were significantly effective on viable cell yield. The optimum points of CDLC, YE and ST of OSF9-ADCFM containing 0.5% (v/v) CDLC, 11 g/L YE and 3 g/L ST. The verification study of Sf9 in OSF9-ADCFM continuously in 5 passages (P1-P5) had an average viable cell concentration of 1.71 ± 0.14 × 10 5 cells/mL, which was in the range of 1.68 × 10 5 -2.27 × 10 5 cells/mL, as calculated from the empirical mathematical model, indicating the suitability of this model. The optimum initial cell concentration was 3 × 10 5 cells/mL. Both peptones after 10 kDa ultrafiltration significantly affected Sf9 growth. Two batches of suspension culture in OSF9-ADCFM of Sf9 cells exhibited that it could adapt to grow in suspension culture with a doubling time and specific growth rate of 28 h and 0.025 h −1 in batch no. 1 and 26 h and 0.027 h −1 in batch no. 2. The maximum viable cell concentrations obtained from batches no. 1 and no. 2 were 1.79 × 10 6 and 2.17 × 10 6 cells/mL, respectively. The results indicated that in-house OSF9-ADCFM could be used for Sf9 culture in both static and suspension culture. Un-infected Sf9 cells in Grace's insect medium plus 10% FBS gave higher cell number than SF-900II and OSF9-ADCFM 4.29 and 5.38 times, respectively. GFP-fluorescent cells in medium supplement with serum reached the highest cell number of 34 cells/well which was less than 1% of total cell number. In OSF9-ADCFM, un-infected cells number and GFP-fluorescent cells were lower than SF-900II 1.25 and 7 times, respectively. Even though the efficiency of the developed OSF9-ADCFM could not reach the same level as the commercial SF-900 II. However, OSF9-ADCFM can also be used in cell culture to reduce the cost of animal cell cultivation. The development would be continued to increase both cell numbers and protein expression in the next step.