1. Introduction
Streptococcus pneumoniae (pneumococcus) is a Gram-positive bacterium that causes human diseases as otitis, sinusitis, pneumonia, meningitis and sepsis. Current pneumococcal vaccines confer protection by generating antibodies to capsular polysaccharides of prevalent serotypes. In order to be effective in children, these polysaccharides have to be covalently linked to carrier proteins, producing conjugate vaccines (Pollard et al., 2009). Despite the availability of these vaccines, the burden of pneumococcal disease remains high due to two main reasons: i) the high price of conjugate vaccines limiting accessibility to the majority of the world population; ii) serotype replacement observed in all countries with universal pneumococcal vaccination programs (Weinberger et al., 2011), as a consequence of limited number of polysaccharides included in vaccine formulations in face of more than 95 known serotypes (Geno et al., 2015).
A promising alternative vaccine is the pneumococcal whole cell vaccine (PWCV), which is composed of a non-encapsulated strain that was genetically modified to delete the autolysin gene and substitute the wild-type hemolytic toxin pneumolysin with a detoxified derivative (Malley et al., 2001 and Lu et al., 2010a). Due to the absence of any polysaccharide capsule, in theory this vaccine could provide broad coverage against all serotypes. Moreover, PWCV induces protection against pneumococcal nasopharyngeal colonization and invasive disease in mice (Lu et al., 2010b). For its production, a culture medium free of animal compounds has been developed (Liberman et al., 2008), as well as the production and inactivation processes following current good manufacturing practices (cGMP) requirements. The cost of PWCV production was estimated to be low, since the process established is relatively simple compared to other pneumococcal vaccines (Gonçalves et al., 2014).
Despite the success in producing the PWCV in cGMP conditions, the high number of doses necessary to immunize the population imposes further developments. In this context, the intensification of the production process of PWCV is worthy to pursue. According to Babi et al. (2016), process intensification can be defined as a process that achieves high efficiency of process equipment, reduction of cost and high yields. Many studies evaluate process intensification for different valuable products, as proteins (Berenjian et al., 2014) and viral vaccines (Tapia et al., 2016); here we evaluate process intensification of PWCV.
Process intensification for increasing pneumococcal cell density is particularly challenging due to the lactic acid produced during fermentation, which inhibits cell growth (Xu et al., 2006). The pneumococci, as other lactic acid bacteria (LAB), are strictly fermentative and nutritionally fastidious bacteria, hence carbohydrates normally are their energy source and the main end-product is lactate, which is also responsible for growth inhibition (Carvalho et al., 2013). Depending on the redox balance, pneumococcus can also metabolize carbohydrates to mixed-acid fermentation, producing acetate and formate (Yesilkaya et al., 2009). In addition to the end-product inhibition, the batch process suffers also from rather low productivity due to the long auxiliary time (Lu et al., 2016).
In the industry, continuous cultures with membrane cell-recycle systems have proven to be efficient for commercial production of lactic acid (Wee and Ryu, 2009, Min-tian et al., 2005, Kwon et al., 2001, Tejayadi and Cheryan, 1995) and biomass of Lactobacillus casei(Aguirre-Ezkauriatza et al., 2010). In contrast to batch fermentation, this operation mode allows the removal of the lactic acid produced, eliminating its inhibitory effects and increasing cell growth, while also improving productivity of lactic acid and biomass by decreasing the auxiliary time. Moreover, the use of bioreactor connected to the tangential flow microfiltration membrane meets two principles of process intensification: integration of operations and integration of functions (Lutze, 2010). For these reasons, we developed a process that uses the same principle of the continuous culture with cell-recycling, i.e., uses a microfiltration membrane to remove the inhibitors and to return the cells to the bioreactor. Since our goal was to increase cell biomass, we did not perform the continuous process during the steady-state, but integrated the up and downstream processing by employing the same microfiltration system for cell-recycling and cell separation when the highest biomass was reached. Thus, this integrated process was called here perfusion-batch, as it differs from the conventional continuous culture with cell-recycling.
Despite the advances of lactic acid production in continuous culture with cell recycling and the advantages of this system to increase cell density, to the best of our knowledge neither the strategy of continuous culture with cell-recycling nor the so-called perfusion-batch has not been applied to date for S. pneumoniae culture. The work presented here aimed at comparing the former processes developed for PWCV production in batch (Gonçalves et al., 2014) and fed-batch (Liberman et al., 2011) with the production in the perfusion-batch using hollow fiber membranes for cell-recycling and cell separation. We also compared microfiltration using hollow-fibers and centrifugation to perform cell separation procedures in order to address intensification on downstream process for batch and fed-batch as well. To evaluate whether the different fermentation and separation procedures would affect vaccine quality, we compared the efficacy of vaccines produced via these methods in mouse models.
2. Material and methods
2.1 Microorganism
The S. pneumoniae RM200 (Rx1 PdT ΔlytA ) strain is derivative of the spontaneous non-encapsulated Rx1, which was genetically engineered for autolysin deficiency (ΔlytA ) to improve cell density and had pneumolysin substituted by a non-toxic pneumolysoid derivative, PdT, with three amino acid substitutions (W433F, D385N, and C428G) to reduce any potential toxicity (Lu et al., 2010a).
2.2 Culture media, buffers and solutions
The composition of all culture media was based on the complex animal component-free culture medium described by Liberman et al. (2008) and is shown on Table 1. BHI-blood agar plates, containing 5% (v/v) sheep blood, 15 g/L agar and 37 g/L of brain heart infusion (Difco, BD), were employed for bacterial enumeration, controlling culture purity and assessing hollow fiber membrane integrity. Automatic pH control of cultivation in the reactor was obtained by addition of 5 M NaOH. The cell washing and harvesting buffer was lactated Ringer’s solution with glucose composed of 5 g/L NaCl, 0.3 g/L KCl, 0.2 g/L CaCl2.2H2O, 3 g/L sodium lactate and 2 g/L glucose. All reagents were of analytical grade.
2.3 Cultivation strategies
All processes were carried out with 10 L medium, except for fed-batch fermentation (initial volume of 8 L) in the bioreactor BioFlo410 (New Brunswick Scientific Company Inc., Edison, NJ, USA) with automatic pH control, temperature and stirring speed. The inoculum was prepared from a cryopreserved working cell bank, 750 µL inoculated in 500 mL of the same medium used in the batch phase in glass bottle and incubated statically at 36.5 ºC until the cell growth reached an optical density (OD) at 600 nm of approximately 2.0. This culture was then introduced into the bioreactor in order to obtain an initial OD around 0.1. The bioreactor cultivation was performed at 36 ºC (± 1 ºC) and the pH was automatically controlled at 7.0. The stirring speed was controlled at 150 rpm. Nitrogen was sparged throughout the fermentation at a flow rate of 0.1 vvm. Polypropylene glycol (Fluent Cane 114, Brenntag, Germany) 30% (v/v) was used as an antifoam agent. Fed-batch and perfusion-batch processes were operated in batch mode until 3-3.5 h and the feeding or perfusion operation started when the OD reached approximately 4.0. Previously, the feeding medium was sparged with N2 in order to decrease the oxygen concentration. The flow-rate was 0.5 L/h for fed-batch process. The initial medium volume in fed-batch was 8.5 L, including the inoculum volume, and the mean final volume was 9.05 L after 3 h of feeding or 6 h of cultivation, when the highest OD was reached in this process. In the perfusion-batch , the working volume of the reactor was maintained at a constant value (10 L) by supplying the feeding medium with the same flow-rate as the permeate of the membrane, i.e., 7.3 L/h, which represents a D = 0.63 h-1, until the highest OD was achieved, then the cells were harvested and washed as described below (item 2.5). A diagram for all three processes integrated with the cell separation system is shown in Figure 1.
2.4 Cell recycling
A polysulfone hollow fiber membrane with 0.1 μm of pore size and 0.92 m² of total filtration area (CFP-1-E-35A, GE Healthcare, Little Chalfont, Buckinghamshire, UK) was employed to remove metabolites as organic acids and return all cells to the bioreactor in the perfusion-batch process, without bleeding. The hollow fiber removed also medium nutrients, which were replaced by the feeding medium 1 from 3 h to 5 h and feeding medium 2 from 5 to 9 h (Table 1). Diaphragm pressure gauges were installed at the hollow fiber inlet and outlets and a peristaltic pump was used for continuous recirculation of fermentation broth through the hollow fiber system (Easy-load tubing pump, EMD Millipore, Merck KGaA, Darmstadt, Germany). The inlet pressure was kept below 10 psi during the operation. Additionally, two peristaltic pumps were used to control permeate flow-rate and feeding flow-rate (504U and 323U, respectively, Watson-Marlow Fluid Technology, Cornwall, UK) at the same flow rates in order to keep the reactor volume constant (Figure 1).
2.5 Cell separation and inactivation
Cell mass was harvested after OD 6.0 was reached, the same as for the GMP production (Gonçalves et al., 2014), or after reaching the highest OD in each operation mode. Two methodologies were applied in order to compare different cell separation methods: tangential microfiltration or centrifugation. The same system described for cell recycling in the perfusion-batch was used for tangential microfiltration (Figure 1). Cells were concentrated to OD 20-50, and then washed 6 times with the same volume of washing buffer. After washing, cell suspensions were adjusted to approximately OD 30. To evaluate the centrifugation method, samples of 33 mL were harvested by centrifugation at 1930 g for 20 min at 18 °C (RC5C, Sorvall, Du Pont Company, Newtown, CT), and then cell pellets were vortex-homogenized with 33 mL of washing buffer. This step was performed 6 times, and cell suspensions were adjusted to approximately OD 30.
After cell washing, cell inactivation was performed as previously standardized (Gonçalves et al., 2014) with 1:4000 (v/v) β-propiolactone (BPL, Sigma-Aldrich) for 30 h at 4° C mixing at 150 rpm (TE-140, Tecnal, Piracicaba, SP, Brazil). Then, the residual BPL was hydrolyzed by incubation at 37 ºC for 2 h mixing at 180 rpm (Series 25 Incubator Shaker, New Brunswick Scientific).
2.6 Analytical methods
Culture samples were taken during the process approximately every 30 min. Cell density was monitored by measuring the OD using a Spectrophotometer (U-1800, Hitachi High-Technologies, Tokyo, Japan). Dry cell weight was evaluated using previously weighed conical tubes, in which 30-45 mL culture samples were inactivated with 2-5% (v/v) formaldehyde (Synth, Diadema, SP, Brazil), depending on the cell density, for up to 18 h at room temperature, and then washed once with PBS by centrifuging at 3200 g for 20 min (5810R, Eppendorf). The cell pellets were dried at 60 °C for at least 2 days until constant weight. Purity was verified by Gram staining and plating samples on BHI-blood agar. Plates were incubated for 3 days at 36 °C into anaerobic jars to check colony morphology, purity and viability. For the analysis of residual sugar and metabolites, 2 mL samples were centrifuged at 17,530 g for 5 min (Mikro 120, Andreas Hettich GmbH & Co., Tuttlingen, Germany), and supernatants were stored at -20 ºC. Glucose, lactate and acetate were analyzed by high performance liquid chromatography (HPLC, SCL-10AVP, Shimadzu Corporation, Kyoto, Japan) using an Aminex HPX 87H column (300 x 7.8 mm, BioRad Laboratories Inc., Hercules, CA, USA), and 5 mM H2SO4 as mobile phase at 0.6 mL/min and 60 °C. The refraction
index detection (RID) was used for glucose analysis and UV detection at 210 nm for organic acids. Chromatograms were analyzed by Class VP software, version 6.14 SP2 (Shimadzu). Finally, the integrity of the hollow fiber membrane was evaluated by plating the filtrate on BHI-blood agar during the process.
2.7 Vaccine quality evaluation
After inactivation, final products were evaluated as described by Gonçalves et al., 2014. Briefly, aspect, pH, OD, bacterial and fungal sterility, endotoxin level were determined. Total and soluble protein contents were measured by Kjeldahl and Lowry, respectively. Also, bacterial identity was evaluated before inactivation.
The protein production profile was analyzed by Western blot to check if there was any difference between fermentation strategies. Briefly, 20 µL of vaccines (50 µg of each sample) were loaded onto precast 4 to 12% Bis-Tris gels (NuPAGE, Invitrogen, Thermo Fisher Scientific, Carlsbad, CA, USA) and separated by electrophoresis. The proteins were transferred onto a nitrocellulose membrane (Biorad) and probed using different antibodies against specific pneumococcal proteins: SP0785, SP2070, SP2145, SP1572 (known as pneumococcal protective protein A - PppA), Pneumolysoid (PdT) and Pneumococcal surface protein A (PspA) or anti-Pneumococcal Whole Cell Vaccine (PWCV) polyclonal serum. Bands were visualized with the Super Signal West Pico Chemiluminescent Substrate Kit and exposed in CL-X Posure Film (both from Thermo Fisher Scientific, Waltham, MA, USA).
Immunogenicity and potency of the vaccines were also evaluated. Groups of female mice (C57BL/6J from Jackson Laboratories, Bar Harbor, Maine, USA) received one or two (at two-week interval) subcutaneous doses of 100 µg of vaccine preparations adsorbed onto 200 µg of aluminum hydroxide (Alum - Al(OH)3; Brenntag North America, Reading, PA, USA). Mice were anesthetized with isoflorane and bled after 12 days of immunization in order to evaluate antibody and IL-17A production according to Campos et al. (2017). One week after bleeding, animals were anesthetized and received a lethal dose of 106 CFU of serotype 3 S. pneumoniae strain WU2 intranasally and illness monitored for 7-8 days. Any ill-appearing animal (defined prior to the initiation of any of the studies) was immediately and humanely euthanized. All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) of Boston Children’s Hospital.
2.8 Kinetics analysis
The mass balance and kinetic parameters were calculated according to the following equations:
\(D=F_{\text{feed}}/V_{\text{sys}}\) (1)
\(S_{\text{total}}={S_{0}+S}_{\text{feed}}-{S_{\text{res}}-S}_{\text{samp}}-S_{\text{perm}}\)(2)
\(\text{acet}_{\text{total}}=\ {\text{acet}_{0}+acet}_{\text{res}}+\text{acet}_{\text{samp}}+\text{acet}_{\text{perm}}\)(3)
\(\text{lac}_{\text{total}}=\ {\text{lac}_{0}+lac}_{\text{res}}+\text{lac}_{\text{samp}}+\text{lac}_{\text{perm}}\)(4)
\(Y_{\text{acet}}=\text{acet}_{\text{total}}/S_{\text{total}}\) (5)
\(P_{x}=\frac{X_{\max}-X_{0}}{t-t_{0}\text{\ x\ }V_{\text{consumed}}}\)(6)
\(P_{\text{lac}}=\frac{\text{lac}_{\max}-\text{lac}_{0}}{t-t_{0}}\)(for batch and fed-batch) (7)
\(P_{\text{lac}}=\text{lac}_{\max}\times D\) (for perfusion-batch) (8)
Where D is the dilution rate (h-1);Ffeed is the feed flow rate (L/h);Vsys is the culture volume (L) of the system, including bioreactor, hollow fiber and tubing;Stotal is total amount of glucose consumed (g);acettotal and lactotal are the total amount of acetate (g) and lactate produced (g), respectively;Yacet/S is the acetate yield on glucose (g acetate produced/g glucose consumed); Px is the biomass volumetric productivity (g dry cell weight/L.h) andPlac is the lactate volumetric productivity (g lactate/L.h); X is the biomass (g dry cell weight); t is the time of cultivation (h); Vconsumed is the total medium used in the process. The index 0 indicates the initial condition, res is the residual amount inside the bioreactor,samp is the amount removed in sampling, perm is the amount present in hollow fiber permeate, max is the maximum value reached.
Linear regression fit was applied to calculate the angular coefficient, which corresponds to the yield coefficients on glucose: YX/S is the biomass yield (g dry cell weight produced/ g glucose consumed), and Y lac/S is the lactate yield (g lactate produced/g glucose consumed). The linear regression was also performed to calculate the lactate yield on biomass (Ylac/X , g lactate /g dry cell weight). The maximum specific growth rate (µmax , h-1) was calculated by the angular coefficient of linear regression fit of Ln(DO) vs. time in the first 3 h of cultivation (batch phase of all three processes).
The total protein of the whole cell vaccine obtained in each fermentation was calculated by multiplying the volume of concentrated bulk product by the total protein concentration measured by Kjeldahl. Then, to estimate the number of doses, the total protein amount was divided by 0.3 mg, which represents an estimate of the human dose of this vaccine (ClinicalTrials.gov, 2014).
2.9 Statistical analysis
Each culture was performed at least in triplicate. The mean of values and standard deviation is presented in the figures. All parameters were analyzed by one-way ANOVA and the means were compared by Tukey’s Multiple Comparison Test. Statistical differences between IgG antibody titers and IL-17A production were evaluated by the Mann-Whitney Utest. Animal survival after challenge was analyzed by Kaplan-Meier method and the log-Rank test to compare the curves. For all analyses, P<0.05 was considered to represent statistical significance.
3. Results and discussion