The donors are from the Italian Volunteers Blood Association (AVIS) and they regularly donate to the Blood Bank of the ASST Spedali Civili of Brescia, according to Italian national legislation (Centro Nazionale Sangue [CNS], 2020a,b). Donations are voluntary and free. Since the plasma from polytransfused donors and multiparous women may contain antibodies against human leukocyte antigens (HLA) and human neutrophil antigens (HNA), we have decided to exclude their donations for hPL production. For organizational reasons, we have limited our selection to non-transfused males between the ages of 18 and 60, negative in the screening for communicable diseases by serology (HIV 1/2, HBV, HCV, and Treponema pallidum) and NAT tests (HIV, HBV, HCV, and WNV) (American Association of Blood Banks [AABB], 2020; Centro Nazionale Sangue [CNS], 2020a,b; European Directorate for the Quality of Medicines & HealthCare of the Council of Europe [EDQM], 2020a; European Union [EU], 2020). It was shown that hMSC do not express or upregulate ABO blood group antigens and that hPL does not exert any detrimental effect on proliferation, regardless of blood group (Moll et al., 2014). However, in order to avoid potential immune activation due to antigens or isoagglutinins and to limit hemolysis during production, we have decided to use group 0 buffy coat and group AB plasma. We used the intermediate by-products of whole blood donations processed within 24 h of collection, which were not necessary for clinical needs.

By taking advantage of the Blood Bank of the ASST Spedali Civili of Brescia, we had the opportunity to apply to the production of hPL instruments and technologies (Supplementary Figure 1) quality and safety requirements routinely used by blood banks in the processing of blood products. By automatic whole blood separation (Fresenius Kabi), fresh frozen plasma (FFP), buffy coat (BC), and red blood cell (RBC) units were derived (Figure 1A). FFP was processed to obtain cryoprecipitate-poor plasma (CPP) used in platelet pool (PLT pool) production. Group AB FFP was filtered for leukocyte depletion, rapidly frozen (−80°C, <45 min, PlasmaFrost ITeM, Angelantoni Life Science), and then allowed to thaw overnight in a monitored blood bank refrigerator (+4°C). The morning after, FFP was centrifuged on high speed (3,000 g, 15 min, +4°C, Heraeus Cryofuge 6000i) to precipitate and remove the cryoglobulin fraction of plasma (Figure 1B). To produce every single PLT pool (Figure 1C), at maximum 24 h time-lapse from collection, six group 0 BC and one unit of group AB CPP were assembled (Teruflex BP-KIT with Imugard III-S PL, Terumo) and gently centrifuged (340 g, 6 min, +22°C, Heraeus Cryofuge 6000i) to obtain platelet-rich plasma (PRP) in-line filtered for leukocyte depletion in a satellite bag. Afterward, PRP was diluted to a platelet concentration of 1.0 × 10⁶/μl with CPP (pool of four donations) and 25–50 Gy irradiated, according to the PLT pool protocol, to reduce the viability of any residual leukocytes. Residual leukocytes were flow cytometry counted (Kit Leucocount, FACSCanto II, BD).

Three hPL batches (A–B–C18) were manufactured by assembling three different PLT pools for each batch (Figure 1D). Assembled pools were rapidly frozen (−80°C, <45 min, PlasmaFrost ITeM, Angelantoni Life Science), stored at −80°C for 24 h in a monitored blood bank freezer, and then thawed at +37°C for 25 min (DH8 Plasma Thawing System, Helmer) to cause platelets’ lysis. The hPL was split in half into two bags. After centrifugation (5,000 g, 25 min, +22°C) for the removal of platelet bodies or debris, half of the product hPL (hPL1c) was aliquoted into 50 ml tubes (BD Falcon conical tubes) and stored at −80 ± 10°C until use. In order to improve the releasing process of growth factors, half of the hPL product further underwent three freeze–thaw cycles (hPL4c), and after the removal of debris by centrifugation, it was aliquoted and stored at −80°C as the one-cycle product. The testing of hPL sterility for fungi and aerobic and anaerobic bacteria resulted negative. The described process allowed to produce batches of hPL with a volume greater than 1,000 ml.

Three hPL batches at four freeze–thaw cycles (hPL4c A–B–C18) were analyzed for their fibrinogen content (ACL TOP 700 LAS, Instrumentation Laboratory Werfen, Milan, Italy), biochemical compositions (Roche Cobas 8000, Risch-Rotkreuz, Switzerland), protein content by electrophoresis (SEBIA Italia s.r.l., Bagno a Ripoli, Firenze, Italy), and hemoglobin (HemoCue, Ängelholm, Sweden) in comparison with a mix of three lots of commercial FBS (Sigma-Aldrich, Darmstadt, Germany).

Proteomic profiles of all the batches of hPL both at one and four freeze/thaw cycles (hPL1c and hPL4c, respectively) and the related CPP, belonging to three different group of donors (groups A, B, and C; n = 18 for each group), were evaluated by nano LC-ESI MS/MS.

Before the mass spectrometric analysis, the nine different samples were albumin- and IgG-depleted using the PROTIA Depletion kit (Sigma-Aldrich) and trypsinized (Re et al., 2019). Briefly, 50 μl of sample derivative diluted in 50 μl of equilibration buffer was loaded onto the immunodepletion column, already equilibrated, and incubated for two consecutive times. The twice-depleted specimen was then collected by centrifugation and combined with a final washing step of 125 μl of equilibration buffer. All the eluted fractions were pooled and the protein concentration was determined using bicinchoninic acid assay on microplates (Pierce-Thermo Fisher Scientific). The enzymatic digestion protocol was based on the filter-aided sample preparation (FASP) strategy. In particular, a volume corresponding to 300 μg of proteins for each sample was diluted with ammonium bicarbonate 50 mM and denaturated by dithiothreitol (20 mM DTT; 95°C for 5 min). After disulfide bond reduction, samples were transferred into the ultrafiltration units (Amicon Ultra-0.5 ml 30 kDa, Millipore). FASP digestion was performed (Chinello et al., 2019). Protein digestion was performed overnight at 37°C adding trypsin. Digestion was stopped by acidification by adding formic acid till 0.1% v/v. Tryptic peptide mixtures were concentrated by a vacuum centrifuge and their concentrations were determined using a NanoDrop^TM spectrophotometer (Thermo Fisher Scientific).

A volume of sample corresponding to about 1 μg of peptide mixtures was injected into the nanoRSLC system for UHPLC separation (Thermo Fisher Scientific) and analyzed by an online coupled tandem mass spectrometer Impact HDTM UHR-QqToF (Bruker Daltonics) (Bosisio et al., 2020). Three technical replicates for each sample were run in order to take into account and reduce the variability associated with the label-free relative quantification approach. Data elaboration including calibration was performed through DataAnalysis^TM v.4.1 Sp4 (Bruker Daltonics, Germany).

Protein identification and quantification were carried out using a proteomic platform based on PEAKS Studio X+ ver. 10.5 (Bioinformatics Solutions Inc., Waterloo, ON, United States). Three LC-MS/MS runs were loaded for each sample. An in-house-constructed UniProt’s reference database of Homo sapiens (accessed June 2018, 557,713 sequences; 200,130,199 residues) was combined with a decoy database and implemented in the software. Only confidently identified peptide features [false discovery rate (FDR) ≤1%] were used for the estimation of the related signal intensity. The normalization of the areas under the curve of the extracted ion chromatograms was performed using TIC (total ion current). Proteins with less than two unique peptides were discarded from the analysis. Only peptides having a quality factor ≥5 were considered for the relative quantification. The non-parametric test integrated in PEAKSQ was used to evaluate the significance of the comparisons. A minimum significance of 13 (p value ≤ 0.05) and a minimum fold change of 1.5 calculated selecting the three most abundant unique peptides were set. Three different levels of stringency in the peptide filtering of the data were established based on the specific distribution of the ratios for the dataset: (i) low stringency criteria (LSC): Ave-Area ≥5E4, peptide ID count: 0, confidently detected sample: 0; (ii) middle stringency criteria (MSC): Ave-Area ≥1E5, peptide ID count: 0, confidently detected sample: 0; and (iii) high stringency criteria (HSC): Ave-Area ≥5E5, peptide ID count: 1, confidently detected sample: 1.

Bone marrow (BM) cells are obtained from bag washouts of scheduled BM donation, after informed consent from the donors. Nucleated cells are recovered by two washings of the filters’ and bags’ residues with 150 ml PBS (Sigma-Aldrich). Mononuclear cells are recovered by gradient separation (Lympholyte-H, Biosera) (400 g, 30 min, +22°C). Viable cells are counted [Trucount tubes, 7-amino-actinomycin D (7-AAD) FACS Canto II, BD] and seeded (3.5 × 10⁵ cells/cm²) in plastic culture flasks (SPL Life Sciences) in Iscove’s medium (Sigma-Aldrich) containing penicillin 100 U/ml–streptomycin 100 μg/ml (Sigma-Aldrich), 2 mM L-glutamine (Sigma-Aldrich), 2.5 μg/ml Fungizone (Sigma-Aldrich), and 10% FBS. Non-adherent cells are removed after 48–72 h (+37°C, 5% CO₂) and the fresh medium was changed every 72 h thereafter. At confluent growth, adherent cells are detached by trypsin 0.25%–EDTA 0.02% (Sigma-Aldrich) treatment and replated at 4 × 10³ cells/cm² in the same medium conditions. At each passage, hMSC (5 × 10⁵ cells/100 μl) antigenic profile is determined (FACSCanto II, BD) by using a four-color combination (CD45-Pe-Cy7, CD73-PE, CD90-APC, CD105-PerCPCy5.5) and single monoclonal antibodies (CD34-PE, CD14-FITC, and CD19-PE). The different tests, all along the entire experimental design, have been repeated with three hMSC batches (hMSC1–2–3).

To test the most effective concentration of hPL4c in hMSC expansion, three different hMSC batches (hMSC1–2–3 at the fourth passage) have been expanded in a medium containing scalar concentrations of hPL4c (A–B–C18) from 0.6 to 20%. Doubling times (DTs) at 4, 7, and 10 days of culture are compared at different hPL concentrations.

Since in the literature hPL is mainly used at 5% in culture medium, we looked for the minimum concentration of heparin to be added to the medium based on this percentage. The same hMSC batches (hMSC1–2–3 at the fourth passage) have been cultured in a medium containing 5% hPL4c (A–B–C18) in the presence of standard 2 IU/ml sodium heparin in comparison with decreasing amounts of sodium heparin (1.2, 0.6, 0.3 IU/ml) and a heparin-free condition. DTs at 4, 7, and 10 days of culture are compared in the different medium conditions.

The biological effects of hPL at one and four freeze/thaw cycles (hPL1c and hPL4c, respectively) have been tested in comparison to a mix (FBS mix) of three commercially available batches of FBS (Sigma-Aldrich), as well as to hPL A17, as previously characterized and tested (Tran et al., 2019). Briefly, hMSC at the fourth passage (p4) are seeded at 4 × 10³ cells/cm² in plastic culture flasks in Iscove’s medium containing penicillin 100 U/ml–streptomycin 100 μg/ml, 2 mM L-glutamine, and 2.5 μg/ml Fungizone (Sigma-Aldrich) in the presence of either 10% FBS mix, 5% hPL1c, 5% hPL4c, or 5% hPL A17. Three different batches (A–B–C18) of hPL1c and hPL4c have been tested. Sodium heparin (0.6 IU/ml) (Veracer) has been added to the medium containing hPL to prevent gellification. The results have been generated in triplicate, performed at different time frames, and repeated with three different hMSC batches (hMSC1–2–3). Half of the hMSC were sacrificed at 4 days and the rest at 7 days of culture. The cells were trypsinized, washed, and resuspended in PBS solution (5% FBS). Cell count and viability were performed in Trucount tubes (BD Biosciences) by the addition of 7-AAD (BD Biosciences) and analyzed by flow cytometry (FACSCanto II, BD Bioscience). hMSC expansion at 4 and 7 days has been calculated as DT expressed in hours¹.

Descriptive statistics and graphical displays, including histograms and boxplots, were obtained prior to all inferential analyses to evaluate the appropriateness of distributional assumptions and for visual assessment. Throughout the manuscript, all measurements are reported as mean ± standard deviation (SD). The comparisons between measurements on two culture conditions were assessed by the Welch two-sample t test, whereas comparisons among multiple (>2) culture conditions were assessed using one-way ANOVA. To adjust for multiplicity in the ANOVA, a Tukey’s HSD procedure was employed for all pairwise comparisons. All tests were two-sided with a significance level set at 5% (0.05). Whenever applicable, the Benjamini–Hochberg FDR procedure was employed for p value adjustment. All the analyses were performed using the freely available language and environment for statistical computing and graphics, R, version 4.0.2².

The biochemical compositions of the three hPL4c batches that were partially fibrinogen-depleted during the production phase turned out to be overlapping, but as expected, they were different in comparison with the FBS mix (Supplementary Table 2). In more details, fibrinogen ranged from 0.173 to 0.191 g/dl in hPL, with a 31% depletion range, while it was undetectable in the FBS mix. Mechanical methods based on the intentional formation of hydrogels in the hPL medium lead to reducing the initial concentration of fibrinogen by over a thousand folds (Laner-Plamberger et al., 2015). As expected, the removal of the cryoglobulin fraction of the plasma demonstrated lower efficiency. However, with a simple method and without adding additives, we obtained fibrinogen concentrations slightly lower than the values published for hPL with similar characteristics (Shih and Burnouf, 2015; Burnouf et al., 2016). The protein content in hPL, ranging from 5.61 to 6.13 g/dl, was found to be comparable to that of other preparations (Shih and Burnouf, 2015; Burnouf et al., 2016). By chemical analysis, glucose resulted to be significantly higher in hPL in comparison with the FBS mix (average hPL 307.6 vs. 70.1 mg/dl). This result is probably attributable to the additives present in the bags used for whole blood donations, while folate resulted to be lower in hPL compared with the FBS mix (3.6 vs. 6.5 ng/ml). Finally, Ca⁺⁺, Na⁺, K⁺, Cl^–, P, Mg, and vitamin B₁₂ have similar concentrations.

An LC-MS/MS label-free strategy was applied to compare “proteomically” the different preparations of hPL (hPL1c and hPL4c) together with the related CPP counterparts and to measure proteomic batch-to-batch variability. Therefore, the same three batches derived from 25 donors (A–B–C18) used for testing the effect of hPL on hMSC expansion were relatively quantified and characterized by an MS-based proteomic approach. Data were filtered with three increasing levels of stringency – low, middle, and high – as detailed before.

Independently from the set parameters, the hPL4c has always shown lower batch-to-batch variability than hPL1c in terms of the number of proteins varied between batches vs. the number of quantifiable proteins (Figure 2A). Applying high stringency criteria, batch-to-batch variability was 17, 21, and 14% for PPP, hPL1c, and hPL4c, respectively (Figure 2B), suggesting that the improvement in cell culture expansion is combined with an increase in the stability of the preparation for hPL4c. Among the differentially abundant proteins observed across the three batches of hPL4c, only three proteins varied across the high, middle, or low data filtering conditions (ALBU_HUMAN, HBA_HUMAN, and KNG1_HUMAN). As a comparison, seven proteins were found to vary for hPL1c (GELS_HUMAN, HBA_HUMAN, KNG1_HUMAN, CO8A_HUMAN, RET4_HUMAN, APOC2_HUMAN, and CO3_HUMAN) and two for PPP (ALBU_HUMAN, ANT3_HUMAN) (Figure 3). Proteomic analysis of the two different human lysate samples hPL1c and hPL4c allowed to identify a total of 383 different protein IDs showing at least one significant and unique peptide (FDR < 1%); 331 of them, corresponding to 86% of the total, were common between the two diverse protocols of preparations. Moreover, eight proteins were present in hPL1c and 44 were present in hPL4c (Supplementary Table 3). Among the proteins that were only detectable in hPL4c, three were recognized as growth factors related to the regulation of insulin-like growth factor (IGF) transport and uptake (SERPIND1, PDIA6, PRKCSH).

We have produced hMSC in large batches from three BM donations (hMSC1–2–3), in order to use each single one along the entire evaluation process. The second-passage hMSC have been stored in liquid nitrogen at the concentration of 0.5 × 10⁶ cells/vial until subsequent use. Up to the third passage, hMSC were cultivated in 10% FBS medium in order to ensure equal culture conditions for subsequent experiments in the different culture media (Supplementary Figure 2A). Before freezing, each hMSC batch was characterized for antigenic profile and differentiation potential, according to minimal accepted criteria (Dominici et al., 2006). The hMSC antigenic profile was proved to be positive (>99%) for CD105, CD73, and CD90 and negative for CD45, CD34, CD14, and CD19 (Supplementary Figure 2B). Osteogenic, adipogenic, and chondrogenic differentiation potentials of hMSC have been determined by using specific differentiating media (Supplementary Figure 2C).

In order to identify the most effective hPL4c concentration for hMSC expansion, we have progressively doubled the concentration of hPL4c (A–B–C18) from 0.6 to 20%, keeping the sodium heparin concentration fixed at 0.6 IU/ml (Supplementary Table 4). In the expansion cultures containing 0.6% hPL4c, cellular proliferation did not proceed for some cultures. The culture medium with 5% hPL4c proved to be the best condition for 10 days of hMSC expansion (average DT 69.37 ± 1.96 h), compared with 1.25% hPL4c (average DT 112.77 ± 6.37 h), 2.5% hPL4c (average DT 86.57 ± 3.45 h), and 10% hPL4c (average DT 71.30 ± 0.75 h). With 10 and 20% hPL4c, the culture medium gelled, and with 20% hPL4c, it was not possible to recover the cells at the time of counting. When using 10% hPL4c with 2 IU/ml heparin, the medium did not gel (average DT 86.30 ± 2.30 h) and also when using 20% hPL4c with 2 IU/ml heparin (average DT 84.17 ± 2.83 h).

Considering that the use of hPL for hMSC expansion requires sodium heparin to be added to the final medium in order to prevent gelling and that, in our hPL4c production process, we reduced the cryoprecipitate fraction of plasma, we have investigated if it were possible to avoid or at least reduce the amount of sodium heparin to be added to the final medium, containing 5% hPL4c. For the three hPL4c batches tested, 0.6 IU/ml was shown to be the lowest concentration of sodium heparin needed to prevent gelling of the culture medium. At 10 days of culture, the average DT was 67.84 ± 5.76 h with 0.6 IU/ml sodium heparin, 80.00 ± 8.54 h with 2 IU/ml sodium heparin, and 72.48 ± 7.66 h with sodium heparin-free medium (0.6 vs. 2 IU/ml, p = 0.0161). The heparin-free and 0.3 IU/ml heparin culture medium gelled, but it was still possible to recover the cells by trypsin treatment (Supplementary Table 4).

Regardless of the hPL and hMSC batches being used, at 4 days of culture, hPL4c showed an increased proliferative effect (mean DT 36.06 ± 4.03 h) compared with hPL1c (mean DT 39.94 ± 1.13 h, p < 0.001), hPL A17 (mean DT 45.71 ± 4.48 h, p < 0.001), and the FBS mix (mean DT 66.73 ± 6.93 h, p < 0.001) (Supplementary Figure 3A). Even considering the results of each single batch at 4 days of culture, the higher proliferation rate is confirmed when using hPL4c vs. hPL1c (A18 p < 0.001, B18 p = 0.01, C18 p < 0.001) (Supplementary Figure 3B). Even at 7 days of culture, hPL4c (mean DT 40.61 ± 1.11 h) confirms an expansion effect on hMSC higher than the hPL1c (mean DT 49.41 ± 2.62 h, p < 0.001), hPL A17 (mean DT 52.66 ± 3.90 h, p < 0.001), and the FBS mix (mean DT 119.70 ± 32.36 h, p < 0.001). The proliferative effect of hPL1c is comparable to that of hPL A17, our previous production with one freeze/thaw cycle (p = 0.004), but it is higher than that of the FBS mix (p < 0.001, Figure 4A). The greater proliferative thrust of hPL4c, compared with hPL1c, is also confirmed by considering the single hPL batches (p < 0.001, Figure 4B). No significant differences in DT were measured between the hPL batches at one and four cycles, respectively. However, a higher coefficient of variation between batches can be noted in hPL1c compared with hPL4c (5.39 vs. 2.74 CV%). In addition, hMSC cultured with a mix of the three hPL1c (mean DT 50.36 ± 2.15 h) had a lower proliferative effect than the single hPL1c B18 batch (p = 0.045); however, this last result needs to be verified with a larger number of measurements. On the other hand, the hPL4c mix (mean DT 41.03 ± 0.83 h) showed no positive or negative effect compared with the single batches (Figure 4B). We also tested the proliferative effect of each single hPL4c at 7 days of culture on three different hMSC batches. While hPL4c A18 has shown no differences in DT for the three hMSC batches (Figure 5A), the hPL4c B18 (Figure 5B) and C18 (Figure 5C) have shown higher proliferative effect on hMSC3 compared with the other hMSC batches (B18 hMSC1 vs. hMSC3, p = 0.014, and hMSC2 vs. hMSC3, p = 0.029; C18 hMSC1 vs. hMSC3, p = 0.018, and hMSC2 vs. hMSC3, p = 0.026).

Overall, the hMSC3 showed reduced DT with all hPL4c batches (mean DT 39.97 ± 1.12 h, median 39.55, CV% 2.81). However, no significant differences were observed between the hPL4c batches when comparing the DT within each hMSC (Figure 5), all resulting <50 h.

Scientific organizations and national entities have been committed to the standardization of the processes and they have released criteria for the harmonization of raw materials such as hPL (Strunk et al., 2018; Biebak et al., 2019; Henschler et al., 2019; European Directorate for the Quality of Medicines & HealthCare of the Council of Europe [EDQM], 2020b; Schallmoser et al., 2020). Although the blood bank standards and procedures (American Association of Blood Banks [AABB], 2020; European Directorate for the Quality of Medicines & HealthCare of the Council of Europe [EDQM], 2020a; European Union [EU], 2020) are the key reference in blood management, we have defined specific requirements for the production of hPL (Table 1). According to the European guidelines (European Directorate for the Quality of Medicines & HealthCare of the Council of Europe [EDQM], 2020a), we have limited hPL residual leukocyte content by irradiation and filtration, far below the accepted value of 1 × 10⁶/unit. Our production process involves batches with 18 BC units to reduce the variability between hPL batches without neglecting the risk of infection. Platelets, diluted in CPP (1 × 10⁶/μl), were subjected to lysis to obtain more than 1,000 ml of hPL. Although not always validated (Strunk et al., 2018; Biebak et al., 2019; Henschler et al., 2019), we have used the freeze/thaw method as the lysis process. The cycle number efficiency was evaluated by comparing proteomic analysis and the proliferative effect of hPL1c vs. hPL4c. Based on the different hPL variables and requirements, we established the release standards and the minimum tolerable batch-to-batch variability (Table 2). Lastly, we have also considered hPL4c stability after 6, 12, and 24 months of storage at −80°C (Supplementary Figure 4). We have verified that the proliferative efficacy of hPL4c batches (A–B–C18), tested on hMSC2, remains stable after 24 months from production (mean DT 40.80 ± 1.71 h).