Keywords
Ex vivo Expansion; Mesenchymal Stromal Cells; Extracellular Vesicles; stirred tank reactor; Feeding regime; continuous perfusion; Fed-batch

INTRODUCTION

In the last decades, human mesenchymal stromal cells (MSC) have emerged as a main player in the field of Regenerative Medicine. Up to 448 clinical trials with human MSC were running within phases I and II including 48 for the treatment of SARS-CoV-2 Induced Acute Respiratory Failure (www.clinicaltrials.gov, terms searched: “mesenchymal stromal cells”, “mesenchymal stem cells” and “mesenchymal stem/stromal cells”, December 2022). Human MSC have been defined as plastic adherent, fibroblast-like cells, expressing cell surface markers such as CD73, CD90, CD105, CD29, CD44, CD49a-f, while being negative for CD14, CD19, CD34, CD45 and HLA-DR expression and can differentiate into multiple mesoderm-type cell lineages. These multipotent cells are present in several tissues including bone marrow (BM), adipose tissue (AT), Wharton’s jelly (WJ), dental pulp, synovial fluid, placenta, amniotic fluid and others. Due to their ability to home to injury sites and promote a regenerative microenvironment, combined with their multilineage differentiation potential, MSC have been widely employed in clinical trials for blood-related diseases (Le Blanc et al., 2008; Prasad et al., 2011), autoimmune diseases (Duijvestein et al., 2010; Garcia-Olmo et al., 2005; Mazzini et al., 2010) and tissue engineering (Horwitz et al., 2002; Horwitz et al., 1999), with encouraging results. More recently, these beneficial effects of human MSC have been attributed mostly to their paracrine activity, exerted through MSC-secreted modulating factors, rather than their engraftment into injured tissues (Kraitchman et al., 2005; Toma et al., 2009). Indeed, emerging evidence associates these paracrine effects to MSC-derived extracellular vesicles (EVs), which play a central role in cell-to-cell interaction and communication in a pleiotropic way (Balbi et al., 2017; Camussi et al., 2013; Zhang et al., 2014). The EVs released by the donor cells can be uptaken by nearby cells or distant cells, subsequently modulating recipient cells (Phinney & Pittenger, 2017). The universally used expression “EVs” comprises all vesicle subtypes (exosomes, microvesicles and apoptotic bodies) and is a highly heterogeneous pool concerning size range (30-5000 nm), origin, content (proteins, lipids, genetic material and organelles such as mitochondria), biochemical and biophysical features, and biological functions. Compared to cell therapies, the use of EVs as cell-free therapeutic products presents several potential advantages namely: (i) EVs are relatively safer, as they are completely non-replicative and not mutagenic (Elsharkasy et al., 2020); (ii) EVs have also a low risk of inducing microvasculature obstruction upon administration due to their smaller size; (iii) EVs have long circulating half-life and the ability to cross the blood brain barrier (BBB) (Cerri et al., 2015; Moon et al., 2019); (iv) EVs have a simpler composition than parental cells, although still complex and with a bioactive cargo; (v) EVs can used as delivery systems with increased efficacy and homing capacity (de Almeida Fuzeta et al., 2022; Katsuda et al., 2013; Pascucci et al., 2014); (vi) EVs as non-living biological products are more resistant to manipulation than living cells; (vii) the possibility of using reduced doses in vivo to achieve a therapeutic response, as EVs can evade phagocytes (Baglio et al., 2015); and (viii) EVs can be potentially stored with no need of potentially toxic cryoprotectants at -20°C for six months without loss of their biochemical activity (Alvarez-Erviti et al., 2011; Sun et al., 2010; Webber & Clayton, 2013). Overall, the regulatory aspects for producing EV-based products for therapeutic strategies is expected to be less complicated than for any therapy based on in vitro expanded cells. Concerning their therapeutic use, EVs can be used as biomedicines, taking advantage of their natural intrinsic medicinal effects, which can be further improved by manipulating the parental cell towards the production of more specialized and efficient EVs (Fan et al., 2020; Haraszti et al., 2018). Alternatively, EVs can be used as drug delivery systems (DDS), by loading various types of therapeutics, including genetic material and drugs (Malhotra et al., 2019).
From a manufacturing perspective, EV production comprises, sequentially: culture of the parental cell line (i.e. MSC); their collection or harvest from the conditioned medium (i.e. culture supernatant); and purification. Until now, MSC-EV production have been performed using serum-containing media, mostly of animal origin, in planar systems under static conditions, which are limited in terms of cell productivity; their non-homogeneous nature results in concentration gradients (e.g. pH, metabolites); are difficult to monitor; and ultimately unsafe as animal-origin material represent a risk of microbial/prion contamination (Sotiropoulou et al., 2006). Moreover, since the concentration of MSC-EV in the conditioned medium is typically low, scale-out to hundreds of T-flasks or multiplate flask systems is required, involving an extensive handling for feeding/harvesting procedures, to achieve a significant final mass production. Additionally, several differences were observed, in terms of the MSC-derived EVs cargo, between EVs isolated from culture supernatants of MSC expanded under different culture conditions, stressing the importance of controlling all culture process parameters to obtain a consistent EV content (Hyland et al., 2020; Kay et al., 2021). In this context, it turns necessary to move forward to fully controlled reactors (i.e. stirred tank reactors (STR)) to establish a reproducible and scalable process to produce MSC-EV. In addition, the use of defined serum-/xenogeneic-free (S/X-free) culture medium formulations could result in substantial improvements for MSC-EV production in terms of reproducibility, stability and quality, while ensuring the approval of regulatory agencies. Besides pH, dissolved oxygen and temperature, another important bioprocess parameter that is crucial to monitor and control during STR cultures is the nutrient/metabolite concentration profiles in culture. For that reason, it is important to design efficient feeding schemes able to maximize cell densities in a cost-effective way. Different strategies can be considered for human cell cultures: batch, where no culture medium is added or withdrawn during culture; fed-batch, where fresh culture medium is added discretely/continuously to the STR but no culture supernatant is withdrawn; and continuous perfusion operation with cell retention, where there is an automated continuous replenishment/removal of fresh/exhausted culture medium. Fed-batch is the feeding operation mode of choice in the pharmaceutical sector due to the high cell densities and consequently high product titers (e.g. antibodies) attained, as well as its simplicity, among other features. On the other hand, perfusion cultures allow higher cell productivities and a steady state operation, as well as better cell physiology control that is crucial when cells are the target product. As main disadvantages, perfusion cultures are generally more costly and sometimes difficult to implement, especially due to the cell retention device (e.g. continuous centrifuges, tangential flow membrane filters and spin-filters). In this work, a scalable S/X-free process to produce MSC and MSC-EVs in a fully controlled STR was successfully established. Two different feeding operation modes were compared in what concerns MSC yields and identity and EVs production: fed-batch (FB) and FB combined with continuous perfusion (CP). The use of FB followed by CP is expected to combine the advantages of the two processes, as no important autocrine factors are withdrawn from the STR during the first days, when the cell number is still low (FB process), and CP process, when the cell number increases and the need for culture medium exchange in deemed needed.
The manufacturing platform established herein is expected to pave a new way for the development of MSC-based therapies, eliminating time- and labour-consuming procedures aiming at the scalable production of well-defined MSC populations as well as their secreted EVs to boost their medical uses.

MATERIALS AND METHODS

2.1 Human cells

The human MSC derived from bone marrow [MSC(M)] used in this study are part of the cell bank available at the Stem Cell Engineering Research Group (SCERG), iBB-Institute for Bioengineering and Biosciences at Instituto Superior Técnico (IST). MSC were previously isolated/expanded according to protocols previously established at iBB-IST (Dos Santos et al., 2010). Bone marrow aspirates were obtained from IPO Lisboa from healthy donors after written informed consent according to the Directive 2004/23/EC of the European Parliament and of the Council of 31 March 2004 on setting standards of quality and safety for the donation, procurement, testing, processing, preservation, storage and distribution of human tissues and cells (Portuguese Law 22/2007, June 29), with the approval of the Ethics Committee of the respective clinical institution.

2.2 MSC(M) expansion under static conditions

Prior to bioreactor inoculation, cryopreserved MSC(M) were thawed and expanded at a cell density of 3000 cells/cm2 with StemProTM MSC SFM XenoFree supplemented with 1% (v/v) GlutaMAXTM and 1% (v/v) Antibiotic-Antimycotic (10,000 units/mL of penicillin, 10,000 µg/mL of streptomycin and 25 µg/mL of Amphotericin B) (StemPro culture medium) on pre-coated T-flasks with CELLstartTM Substrate [diluted 1:100 in phosphate buffered saline (PBS)]. At 70-80% cell confluence, MSC were detached from the flasks by adding TrypLETM Select Enzyme (diluted 1:10 in PBS) for 7 min at 37ºC. Cell number and viability were determined using the Trypan Blue exclusion method. All reagents were acquired from GibcoTM, Thermo Fisher Scientific, USA.

2.3 MSC(M) expansion under stirred conditions

Non-porous SoloHill® plastic microcarriers (Sartorius, Germany) were prepared according to manufacturer’s instructions. Briefly, 20 g of plastic microcarriers were autoclaved at 121ºC for 20 minutes, washed once with PBS and coated with CELLstartTM Substrate (diluted 1:100 in PBS) for 2 h at 37ºC, using a 250 mL Bellco spinner flask (Bellco Glass, Inc., USA), equipped with 90º paddles and a magnetic stir bar, with an agitation of 40 rpm. After the coating process, microcarriers were washed once with StemPro culture medium. 50x106 MSC(M), previously expanded under static conditions with StemPro culture medium for 2 passages, were mixed with the pre-coated plastic microcarriers and the microcarrier-cell suspension (100 mL) was transferred to a 2 L STR (F0 BABY Bioreactor; Bionet, Spain), equipped with a three-blade pitched impeller, and dissolved oxygen (DO), pH, and temperature probes, already containing 500 mL of StemPro culture medium. For process monitoring and control, the F0 BABY was coupled to ROSITA software (Bionet). Process parameters were set to pH 7.2, temperature at 37ºC and DO of 20% of air saturation (Dos Santos et al., 2014; Fernandes-Platzgummer et al., 2016). pH control was performed by adding CO2 through overlay (50 ccm), DO concentration was maintained through overlay (100 ccm) with a mixture of gases (air, CO2, N2) and temperature was controlled via thermal jacket. The cell seeding on the microcarriers (first 24 h) was performed under continuous stirring at 40 rpm and afterwards the agitation was increased to 60 rpm. STR cultures were operated under FB and FB/CP mode. For FB cultures, after 3 days with 600 mL of volume, fresh culture medium was added at days 3, 6 and 7 (final volume of 1750 mL). For the FB/CP cultures, from day 3 to day 5.5, 700 mL of fresh culture medium was added at a constant rate of 0.2 mL/min and after reaching a volume of 1300 mL, the STR started to operate under continuous perfusion at the same flow rate (0.22 day-1 dilution rate) until day 11. Cell-containing microcarriers were maintained inside the reactor, by using a Spin Filter mounted in the agitation shaft. To maintain glucose concentration between 2-6 mM, pulses of concentrated glucose (100 g/L, Sigma) were added to the STR daily (days 3-8) or every 2 days (days 4-12) for FB and FB/CP cultures, respectively. At days 9 and 13, the conditioned medium (MSC-CM) from the STR cultures operated under FB and FB/CP, respectively, was separated from the cell-containing microcarriers, streamed through a 0.22 µm filter and stored at -80ºC for EVs processing and characterization.

2.4 Monitoring of cell culture in the STR

Cell Counts and Viability
STR culture sampling was performed using a previously established protocol (Fernandes-Platzgummer et al., 2014). Briefly, every day, duplicate samples of evenly mixed culture were collected from the STR and transferred to 2 mL tubes. After the microcarriers settle, the supernatant was collected for glucose/lactate analysis and the cell-containing beads were washed with PBS and incubated with TrypLE solution for 7 min in the thermomixer (Eppendorf AG, Germany) at 37º C and 750 rpm. Afterwards, cells were separated from the microcarriers with a 100 µm cell strainer and the cell number and viability were assessed by using the Trypan Blue exclusion method.
Growth Rate and Doubling Time Calculation
The growth kinetics of MSC(M) cultured under stirred conditions on microcarriers was also characterized. For the exponential phase (considering the death rate constant negligible), the maximum specific growth rate, \(\mu\)max (day-1), was calculated as dXv/dt=\(\mu\)max×Xv, where Xv is the viable cell number for a given time (t). After calculating \(\mu\)max, the doubling time, td (day), was calculated as td\(\ =\ \)ln(2)/\(\mu\)max.
Glucose/Lactate Analysis
Glucose and lactate concentrations in the samples retrieved form the STR cultures were determined by membrane-bound immobilized enzyme quantification in a YSI 2500 Biochemistry Analyzer (YSI, USA)
DAPI staining (Nuclear integrity)
Every 2 days, samples of cell-containing microcarriers were stained with DAPI (1 mg/mL) (4’,6-diamidino-2-phenylindole dihydrochloride, Sigma) for 5 min at room temperature (RT) and protected from light, to observe cell adhesion and distribution on the microcarriers.
Immunophenotypic analysis
MSC(M) were analysed by flow cytometry before and after expansion under stirred conditions, using a panel of mouse anti-human monoclonal antibodies for the expression of CD90 (PE-conjugated), CD73 (phycoerythrin (PE)-conjugated), CD14 (PE-conjugated), CD19 [PE-conjugated], CD34 [fluorescein isothiocyanate (FITC)-conjugated], CD45 (FITC-conjugated) and HLA-DR (PE-conjugated) (all from Biolegend, USA); anti-CD105 (PE-conjugated, InvitrogenTM, Thermo Fisher Scientific, USA) was also used. Briefly, cells were incubated with each antibody for 20 min in the dark and at RT, washed with PBS and then fixed using a solution of 1% (v/v) PFA in PBS. Samples were analysed in a FACScalibur (Becton Dickinson, USA) flow cytometer and CellQuestTMsoftware (Becton Dickinson) was used for acquisition. A minimum of 10 000 events were collected for each sample. Analysis was performed using the FlowJo software (Tree Star, USA).

2.5 MSC-EVs separation by precipitation

MSC-EVs were isolated from the MSC-CM (thawed on ice) using the Total Exosome Isolation Kit (InvitrogenTM) according to manufacturer’s instructions (de Almeida Fuzeta et al., 2020). Briefly, 50 ml of MSC-CM were incubated with half the volume of Total Exosome Isolation reagent overnight at 4ºC. Afterwards, samples were centrifuged for 1h at 10,000×g and 4ºC, the supernatant was removed, and the MSC-EV enriched pellet was resuspended in 1 ml of cold RNase-Free PBS (InvitrogenTM) diluted in UltraPureTM DNase/RNase-free Distilled Water (InvitrogenTM). MSC-EV were aliquoted and stored at -80ºC in Eppendorf Protein Lowbind Tubes (Eppendorf AG) until further use.

2.6 MSC-EV characterization

Nanoparticle Tracking Analysis (NTA)
NTA was performed using NanoSight LM14C instrument equipped with a 405 nm laser (Malvern Panalytical, United Kingdom) and the software NanoSight 3.1 (Malvern Panalytical, United Kingdom). Silica 100 nm microspheres (Polysciences, Inc.) were routinely analysed to check instrument performance (Gardiner et al., 2013) and NTA acquisition and post-acquisition settings were optimized and kept constants for all samples. Particle concentration and size distribution on MSC-EVs diluted samples (1:50) were determined according to protocols optimized at iBB described elsewhere (de Almeida Fuzeta et al., 2020).
Protein Quantification
Protein concentration on MSC-EVs samples was measured using Micro BCATM Protein Assay Kit (Thermo ScientificTM) following manufacturer’s guidelines. Briefly, bovine serum albumin (BSA) standards and samples were diluted 1:2 with an equal volume of reagent solution and incubated for 2 h at 37ºC. Absorbance was measured at 562 nm using the plate reader (BioTek Synergy NEO, software Gen5 version 2.09). Three replicates were quantified for each sample. A linear fit was applied to the BSA standards, and the resulting equation was used to determine each sample concentration from its absorbance measurement.
Purity Assessment
The purity of the MSC-EV samples was given by the particle to protein ratio (PPR), which is the ratio between the particle concentration determined by NTA and the protein concentration measured by MicroBCA protein quantification.
Western Blot
The detection of Calnexin, Synthenin, CD63, CD81 and GAPDH through Western Blot was performed according to the protocol optimized at SCERG and described elsewhere (de Almeida Fuzeta et al., 2020). Primary antibodies included anti-Calnexin (1:1000, BD, USA), anti-Synthenin (1:1000, Abcam, UK), anti-CD63 (1:1000, Genetex, USA), anti-CD81 (1:500, Abcam, UK) and anti-GAPDH (1:1000, Santa Cruz Biotechnology, USA). Secondary antibodies included Goat anti-Mouse IgG (H + L) Cross-Adsorbed Secondary Antibody, HRP (1:5000, InvitrogenTM) and Goat anti-Rabbit IgG HRP-conjugated (1:1000, R&D Systems, UK). Image acquisition was performed on iBrightTM CL1500 Imaging System (InvitrogenTM).
Transmission Electron Microscopy (TEM)
TEM Imaging was performed following negative staining protocol. Briefly, 100 Mesh copper grids were coated with formvar/carbon and glow discharged right before use. MSC-EV samples were mixed (1:1) with formaldehyde 4% in 0.1 M PBS (final concentration 2% formaldehyde in 0.05 M PBS) and incubated for 5 min at RT. Then samples added to the grids and were incubated for 5 min to promote adhesion of EVs to the grids. Next, washing in 10 drops of distilled water was performed and samples were stained in 2 drops of uranyl acetate 2% by incubation for 5 min at RT in the dark followed by sample imaging. A TecnaiTM G2 Spirit BioTWIN Transmission Electron Microscope (TECNAI Tecnogroup, Italy) operating at 120k was used, and data was collected with an Olympus-SIS Veleta CCD Camera (OLYMPUS, Japan).

2.7 Statistical Analysis

Results are presented as mean ± standard error of the mean (SEM) of the values obtained from different MSC donors (i.e., biological replicates). When appropriate, comparisons between experimental results were determined by the non-parametric Mann–Whitney U test and statistically significance was assessed with a p-value less than 0.05.

RESULTS

3.1 Influence of the feeding strategy on MSC(M) proliferation

Based on previous work from our group (Dos Santos et al., 2014), a S/X-free microcarrier-based culture system was successfully established for MSC(M) proliferation using a 2 L-scale fully controlled STR operated under FB and FB/CP modes. 50×106 MSC(M) from 3 donors were expanded on CellStart-coated plastic microcarriers with S/X-free StemPro culture medium, which varied over culture time as shown in Figure 1A. During the adhesion stage (first 24 h of cultivation), a minimum agitation speed of 40 rpm was used, which simultaneously allowed to keep the microcarriers in suspension, maximizing the cell-microcarrier contact necessary for cell adhesion, and minimized cell death due to agitation (Hewitt et al., 2011). Subsequently, the agitation speed was increased to 60 rpm to avoid/minimize the formation of aggregates. After 24 hours, cell adhesion efficiencies of (34±5.8)% and (40±5.9)% corresponding to (1.9±0.19)×107 and (2.1±0.23)×107 cells were attained for FB and FB/CP, respectively. Throughout the STR experiments, for both feeding operation modes, cell viabilities were always above 90%, and the fraction of live/dead cells found in suspension was (5.8±1.3)%, after day 2. The growth curves depicted in Figure1B display an exponential phase leading to maximal cell numbers of (3.0±0.12)×108 and (5.3±0.32)×108 corresponding to cell densities of (2.0±0.51)×105 and (4.1±0.90)×105cells/mL (Figure1B) and fold increase (FI) values in total cell number of 16±2.1 and 24±5.5 (Figure1C), obtained at days 8 and 12 of cultivation for FB and FB/CP cultures, respectively. No major differences were observed in terms of cell number until day 7 for the two feeding operation modes tested. After this day, cell growth ceased in the STR operating under FB mode. Specific growth rates (µmax) and doubling times (td) were also calculated for the two STR feeding operation modes according to “Materials and Methods” and were 0.6±0.1 day-1 and 1.3±0.2 days; and 0.5±0.1 day-1 and 1.5±0.2 days, respectively, for FB and FB/CP cultures.