Tissues were fixed at RT, in Bouin’s solution (HT10132; Sigma-Aldrich^®) and paraformaldéhyde (PFA, P6148; Sigma-Aldrich^®) at 4% for 2 hr (explants) and overnight (testis). Then, tissues were dehydrated in a graded series of ethanol in the Citadel 2000 tissue processor (12612613, Thermo Fisher Scientific Inc.) and embedded in paraffin. Sections (3 µm thick) were cut using a microtome (JungRM 2035; Leica Microsystems^© GmbH, Wetzlar, Germany). Three serial tissue sections were mounted on each Polysine^® slide (J2800AMNZ; Thermo Fisher Scientific Inc.) who were coded for future blinded analysis. For each repetition of each experiment, two to three slides separated by a minimum interval of 45 µm were examined and quantify to obtain a more accurate and global assessment of the tissue. Evaluations were assessed on 30 cross-sectioned tubules per slide if possible.

All the antibodies used in this study for immunofluorescence techniques are presented in Table S1 and washes were performed with phosphate buffer saline containing Tween (PBST). First, to assess the entry into meiosis initiation of intratubular cells, we carried out immunofluorescences with the stimulated by retinoic acid gene 8 (STRA8) protein. To unmask antigenic sites, tissue sections fixed in PFA4% were incubated in citrate buffer (T0050, Diapath) at 96°C for 40 min and cooled for 20 min at RT. To avoid nonspecific staining, sections were blocked with a solution of 5% bovine serum albumin (BSA) plus 5% horse serum (HS) for 30 min. Slides were then incubated for 90 min at RT with primary antibody. Then, sections were incubated with secondary antibody coupled to anti-rabbit biotin-conjugated. Finally, sections were incubated with an Alexa Fluor^® 594-conjugated streptavidin for 30 min at RT. Sections were rinsed, dehydrated with ethanol and mounted in Vectashield (H-1000, Vector laboratories, Burlingame, CA, USA) with Hoechst 33342 (B2261; Thermo Fisher Scientific Inc). Negative controls were performed with mouse or rabbit immunoglobulin G (IgG).

For testicular explants and corresponding controls fixed with Bouin’s solution, slides were stained with Hemalun eosin saffron (HES) to have an appreciation of testicular morphology and to appreciate the distinction of the different stages of the germ cells present within the seminiferous tubules over time. We assessed the average surface area of the seminiferous tubules, the number of cells per seminiferous tubule, the intratubular cell density, and the progression of spermatogenesis throughout key time points of the first spermatogenic wave. Images were observed under a light microscope (DM4000B^®; Leica Microsystem^© GmbH) equipped with Leica Application Suite^® software (LAS^®; Leica Microsystem^© GmbH).

Total RNA was extracted from testicular samples using RNeasy Micro kit (74004; Qiagen, Courtabœuf, France) according to the manufacturer’s instructions and stored at -20°C until use. For in vitro cultured testicular explants, the central necrotic area (inherent of long-term culture system) was carefully removed before RNA extraction. To avoid contamination with genomic DNA, extracted RNA was incubated with two units of TURBO DNA-free™ kit (AM1907; Life Technologies™, Carlsbad, California, USA) for 45 min at 37°C. The amount and purity of the RNA samples were measured with a NanoDrop™ spectrophotometer (ND2000; NanoDrop™ Technologies, Wilmington, DE, USA). Only samples with A260:A230 ratio values ≥1.80 were chosen for further analysis. The RNA integrity was also analyzed by capillary gel electrophoresis on RNA 6000 Nano kit chips (5067-1511; Agilent Technologies, Courtabœuf, France) with the Agilent Bioanalyzer 2100 system (G2939BA; Agilent Technologies). Only those samples with an RNA Integrity Number (RIN) value ≥8 were chosen for further analysis. Total RNA extracted material were stocked in liquid nitrogen (LN₂).

RNA-Seq libraries were generated from 800 ng of total RNA using TruSeq Stranded mRNA LT Sample Preparation Kit (20020595; Illumina^®, San Diego, California, USA), according to manufacturer’s instructions. Briefly, following purification with poly(T)-oligo attached magnetic beads, the mRNA was fragmented using divalent cations at 94°C for 2 min. The cleaved RNA fragments were copied into first strand cDNA using reverse transcriptase and random primers. Strand specificity was achieved by replacing dTTP with dUTP during second strand cDNA synthesis using DNA Polymerase I and RNase H. Following addition of a single ‘A’ base and subsequent ligation of the adapter on double stranded cDNA fragments, the products were purified and enriched with PCR (30 sec at 98°C; [10 sec at 98°C, 30 sec at 60°C, 30 sec at 72°C] ×12 cycles; 5 min at 72°C) to create the cDNA library. Surplus PCR primers were further removed by purification using AMPure XP beads (A63881; Beckman-Coulter, Villepinte, France) and the final cDNA libraries were checked for quality and quantified using capillary electrophoresis.

The prepared RNA samples were quantified on a Qubit™ Fluorometer (Q33226; Thermo Fisher Scientific Inc.) and adjusted to 800 ng. Mus musculus organism was chosen with mm10 as genome assembly. The library preparation used was SQ00/SIL-09-PE (for Stranded mRNA-Seq/standard quantity) with read length of SQ00/HS-1×50_40 (with 1×50 bases and an adapter size of 120 bp) and saved in .fastq data files. In total, 36 samples were run on an Illumina^® HiSeq 4000 Sequencing technology (SY-401-4001; Illumina^®) for single-read 50 bp in Poly(A)-enriched mode. Image analysis and base calling were carried out using RTA v.2.7.7 and bcl2fastq v.2.17.1.14. The mRNA-Seq data are deposited at Sequence Read Archive (NCBI) ¹ .

Reads from each individual sample were aligned to the mm10 release of the mouse genome with STAR (version 2.5.2a) (25), using previously published approaches (26–30). Briefly, the STAR program was first run for each fastq file, using the RefSeq transcript annotation (GTF format) of the mouse genome (release mm10). Exon junction outputs were then added to a splice junction set. STAR was next run a second time, complemented by the splice junction dataset, to produce a final alignment file (BAM format) for each sample.

RefSeq transcripts were quantified with StringTie (version 1.3.3), with default settings applied (31). Transcript abundances were next normalized with Ballgown (available in the StringTie suite), expressed as reads per kilobase of exon model per million (RPKM).

The statistical filtration of the genes showing a differential expression across experimental samples was performed using the Annotation, Mapping, Expression and Network (AMEN) suite of tools (32). We first selected 12,565 genes “detectable” or “expressed” genes, defined as those for which abundance levels exceeded 0.5 RPKM in at least one experimental condition (median value of sample duplicates). Next, we compared in vitro vs. in vivo (i.e., D4 vs. 10.5 dpp; D16 vs. 22.5 dpp; and D30 vs. 36.5 dpp) and fresh vs. frozen (6.5 dpp CSF vs. 6.5 dpp and D30 CSF vs. D30) testicular tissue samples and selected genes yielding at least one-fold change greater than or equal to 2.0 (median values of sample duplicates). A linear models for microarray data (LIMMA) statistical test was used to identify genes with significant abundance variations across samples (F-value adjusted with the FDR method: P<0.05) (33). The differentially expressed genes (DEGs) were clustered into eight (C1-C8) and five (F1-F8*) expression patterns with the unsupervised HCPC algorithm for in vitro vs. in vivo and fresh vs. frozen, respectively (34).

The g:GOSt function was used to known functional information sources and detects statistically significantly enriched terms using g:SCS threshold fixed at 0.05. In order to consider only the elements that emerge from the batch for in vitro vs. in vivo, DEGs selected for this analysis have a p_adj<6×10^-8. The g:GOSt database versions used in the present study are Ensembl 104 and Ensembl Genomes 51 ² . In addition to GO Terms with the use of biological process (BP), cellular component (CC), and molecular function (MF) ³ , the g:GOSt includes pathways from Reactome (REAC) ⁴ , and WikiPathways (WP) ⁵ up-to-date data sources.

Reads were aligned against the reference genome (GRCm38) using STAR (2.7.3a; 25) with default parameters. Gene expression was then calculated with HTseq (0.12.4; 35) with –mode intersection-nonempty.

The raw count data from all samples were transformed with the varianceStabilizingTransformation (vst) function from DESeq2. We then calculated the Euclidean distance between samples on the transformed counts and the samples were then clustered using the R function hclust. We also performed a PCA on the vst transformed counts using the 500 most variable genes. The biological samples who did not meet the quality checks were therefore eliminated from this bioinformatic analysis.

Differential analysis was then performed with R, using DESeq2 ⁶ . Each condition was compared to another, and raw p-values were adjusted by the method of Benjamini and Hochberg to control a false discovery rate (FDR) (36). DEGs selected for Volcano Plots analysis have a |Log₂ fold change| greater or equal to 2.0 and a |Log₁₀ p_adj| greater or equal to 3.0.

For KEGGs pathways evaluation, we considered genes with both a log2 fold-change of at least 1.5 and adjusted p-values under 0.001 as differentially expressed here. The g:GOSt function ⁷ from R package g:Profiler was used to detect statistically significantly enriched terms.

An initial analysis was performed to identify patterns with corresponding DEGs. Then, an analysis of pattern’s Gene Ontology (GO) Terms was carried out. Finally, an investigation of known testicular cell genes was settled to characterize biological samples and determine the potential impact of (i) in vitro culture and (ii) freezing procedure. Considering that in vitro culture does not allow achieving a spermatic yield similar to the physiological conditions observed in vivo, we wanted to examine the biological and molecular changes that differ throughout the first wave of spermatogenesis. To this end, we compared global gene expression between cultured testicular explants and physiological corresponding testicular tissue by RNA-Seq. A clustering analysis on the resulting set of genes allowed us to highlight eight expression patterns (termed C1-8) for the impact of in vitro culture on the first spermatogenic wave ( Figures 2A, D ). A total of about 8,456 DEGs were shown between cultured testicular explants and in vivo controls (especially for genes related to advanced stages of spermatogenesis) ( Figure 2C ). C1-C2, C3-C7, and C8 can be regrouped due to their broad expression profiles: cluster C1-C2 corresponds to 4,860 genes transiently expressed in the testis during its maturation and predominantly down-regulated in cultured tissues; cluster C3-C7 corresponds to 3,189 genes poorly expressed during the maturation process but up-regulated in cultured tissues, and cluster C8 corresponding to 407 genes highly expressed only at the end of in vitro culture.

In order to investigate the molecular pathways involved in the alterations of the maturation process throughout in vitro culture, we performed a functional analysis based on a generic GO Terms enrichment on the expression patterns ( Figure S1A ). The patterns C1 (251 GO Terms) and C2 (107 GO Terms) were mainly related in general mechanisms such as cytoplasm, intracellular cell part and organelle, nucleus, metabolic biological processes, and protein binding. Interestingly, the pattern C3 (39 GO Terms) was linked to meiotic cell cycle, chromosome segregation, and nuclear division. A strong up-regulation of genes associated with this pattern can be observed at 22.5 dpp ( Figures 2A, D ), a timing during which pachytene spermatocytes I perform their strict pairing of homologous chromosomes and when recombination form allows crossing-over. The pattern C4 (26 GO Terms) was associated with cilium, sperm motility, microtubule activity, and CatSper complex. These Terms are strongly up-regulated at 22.5 and 36.5 dpp, representing the initiation of gene expression necessary for spermiogenesis (a late stage of spermatogenesis that remains highly impaired in culture). As expected, the pattern C5 (10 GO Terms) corresponding to the majority cell population of elongated flagellated spermatids at 36.5 dpp was related to sperm part, spermatogenesis, acrosome vesicle, and reproduction. The pattern C7 (4 GO Terms) was linked to cytoplasmic part and ion binding. The pattern C8 (102 GO Terms) was associated to response to stress, inflammatory response, immune system process, and cytokine production. Mostly up-regulated for D16 and D30, these transcriptional data suggest an important inflammatory response of the testicular explants during in vitro culture.

Five patterns were highlighted for the impact of freezing procedure ( Figures 2B, E ). A total of only 25 specific genes were considered as differentially expressed between cryopreserved and fresh tissues without the impact of in vitro culture ( Figure 2C ): Cluster F1 (Rpl26, Rpl6, Mrpl42), Cluster F2 (Entpd6, Pdpk1, Farsb, Ndufa3, Zdhhc21), Cluster F3* (Pomc, 4932414J04Rik, Cldn12, Svip), Cluster F4* (Gm10767, Pde7a, Trmt2b, Btbd9, Zfp426, Kcnip1), Cluster F5* (Rps27rt, Ifit1, Ifit3, Myef2, Fxyd1, Ndufc1, Arhgap19). This functional enrichment analysis revealed a high number of overlapping genes (20 DEGs; 44,5%) between freezing procedure and in vitro culture impact. These observations suggest a minor impact of the freezing protocol used to cryopreserve prepubertal testicular tissues.

For the impact of freezing procedure, the patterns F1 (6 GO Terms), F3* (1 GO Terms), and F5* (3 GO Terms) were associated to ribosome, regulation of appetite, and response to interferon/defense, respectively ( Figure S1B ). The interpretation of these data is relatively complex due to the low number of GO Terms per pattern. However, although freezing has very limited impact on testicular tissue, we observe a weak response of frozen and cultured testicular tissue through a delicate persistent response to external aggression.

The presence of the onset of meiotic prophase retinoic acid protein STRA8 into seminiferous tubules follows the same dynamics in vitro as in vivo. However, a lower proportion of STRA8-expressing seminiferous tubules is observed in culture at D4 (17.2 ± 5.28) and D16 (19.5 ± 2.21) compared to 10.5 (33.6 ± 2.06; P<0.001) and 22.5 dpp (19.5 ± 4.95; P<0.01) ( Figures 3B, C ). Freezing procedure does not prevent entry into meiosis of spermatogonia before or after in vitro culture (P≥0.05) ( Figure 3D ).

Regarding the kinetics of the first wave of spermatogenesis, there is a delay at D4 in the timing of entry of germ cells into meiosis with a lower percentage of seminiferous tubules containing Leptotene/Zygotene spermatocyte I cells (59.5 ± 4.24%) in comparison to 10.5 dpp (81.7 ± 4.54%; P<0.0001) ( Figures 4A and C_1,2 ). A lower number of seminiferous tubules were observed with haploid cells in vitro at D16 (34.4 ± 7.43% for round spermatids) and D30 (36.5 ± 5.71% for round spermatids and 24.7 ± 4.69% for elongated spermatids) than in vivo at 22.5 (8.10 ± 0.74% for round spermatids; P<0.0001) and 36.5 dpp (100 ± 0.00% for round spermatids and elongated spermatids; P<0.0001), respectively ( Figures 4A and C_1,2 ). In addition, we observed an increase in intratubular cell density at D4 for the tissue cultured in vitro (11.4 ± 0.14 cells/1000 µm²) compared to in vivo at 10.5 dpp (8.10 ± 0.74 cells/1000 µm²; P<0.0001) ( Figures 4A and C_1,2 ). This observation is due to the fact that testicular explants have a lower seminiferous tubule expansion as well as a lower entry into meiosis, resulting in a higher cell number per µm². This observation is counterbalanced with a strong reduction in cell density at D30 for tissue grown in vitro (5.29 ± 0.20 cells/1000 µm²) compared to in vivo at 36.5 dpp (7.53 ± 0.44 cells/1000 µm²; P<0.01) ( Figures 4_C1-2 ). These observations could be explained by a lower proliferation of germ cells in culture while suggesting a well-known mechanism of disappearance by autophagy of maturing cells that have difficulty passing the pachytene checkpoint.

Regarding the impact of freezing procedure, a lower number of seminiferous tubules were observed with haploid cells at D30 CSF (25.0 ± 4.89% for round spermatids) than at D30 (36.5 ± 5.71% for round spermatids; P<0.05) ( Figures 4A, B and C₃ ). Regarding the cell density obtained after culture of frozen or fresh tissue, no difference could be observed (P≥0.05) ( Figures 4A, B and C₃ ). These observations suggest that testicular tissue freezing does not have a major impact on the first wave of spermatogenesis initiation and development.

Although the proportions of germ cells transcripts were globally comparable between the in vitro cultures and the age-matched controls in vivo, we observed a small decrease in spermatocyte-related transcripts observed at D16 (1.77) compared to 22.5 dpp (4.00) ( Figures 4D_1-2 ). In addition, the levels of protamines (Prm1, Prm2) and spermatid nuclear transition protein 1 (Tnp1) expressions drop at the end of the in vitro cultured condition (29.7) ( Figures 4D₂ ) and even more from frozen tissue (11.4) ( Figures 4D₃ ) compared to in vivo age-matched controls (368) ( Figures 4D₁ ). This observation highlights the low yield of spermatogenesis obtained under non-physiological conditions.

In our experimental setup, the RNA of whole testis or pool of testicular explants were used for sequencing. Because the study of the testis is not limited to germ cells, we studied transcripts derived from somatic testicular cell. In addition to germ cells transcripts (Ddx4, Prm1, Prm2) and in order to understand the contribution of somatic cells in the gene expression differences, we analyzed the expression of Sertoli (Vim, Gata4, Ar, Amh, Sox9), Leydig (Cyp17a1, Cyp11a1, Hsd3b1, Ccn5), and peritubular and blood vessel transcripts (Bgn, Dcn, Acta2) cells genes.

For in vivo, the expression of genes related to Sertoli cells ( Figure 5A₁ ) and peritubular and blood vessel transcripts ( Figure 5C₁ ) decreased during the first wave of spermatogenesis, possibly reflecting the increase in germ cell population, especially at 36.5 dpp with a large presence of spermatids ( Figure 5D₁ ). The analysis of the expression of Leydig cells specific genes show no global difference across the first wave, although we observed an increase of the expression of Cytochrome P450 gene Cyp17a1 at 36.5 dpp ( Figure 5B₁ ). For in vitro culture, equivalent profiles were observed with a lower cell-dilution effect ( Figures 5A₂, B₂ , and C₂ ) due to a lower spermatic yield. However, Cyp17a1 expression did not increase at D4 and throughout the culture while the Leydig cell marker Hsd3b1 is still present at levels comparable to those in vivo, suggesting a defective maturation of the Leydig cells in vitro. Besides, despite the fact that we pooled 12 testicular tissue explants per biological replicate for in vitro conditions, we observed a rather important variability in the expression of genes related to highly differentiated germ cells (Prm1, Prm2) ( Figure 5D₂ ). Concerning the comparison between fresh and frozen tissue, the expression levels of the studied genes are not different from a somatic cell point of view ( Figures 5A₃, B₃ , and C₃ ). Regarding germ cells, a lower yield of spermatogenesis is observed ( Figure D₃ ).

Between 6.5 dpp and 10.5 dpp, very few differences in gene expression (only 6 DEGs) can be highlighted with a reasonable p-value and a consistent gene ratio ( Figure S2A₁ ); afterwards, almost all genes are upregulated during the first wave of spermatogenesis ( Figures S2A_2,3 and C_2,3 ), reporting the successful establishment of the first wave of spermatogenesis. Between 10.5 dpp and 22.5 dpp, we notice a majority of GO Terms focused on the flagellum/cilium, reproduction, and spermatozoon in formation ( Figure S2B₂ ), which is consistent with the onset of spermatogenesis. Between 22.5 dpp and 36.5 dpp, we can observe again GO Terms in connection with the flagellum but especially with nucleus organization and sperm chromatin condensation ( Figure S2B₃ ), corresponding to the establishment of spermiogenesis. As expected, between 6.5 dpp and 36.5 dpp, we observe the overexpression of genes related to the establishment of the first wave of spermatogenesis ( Figures S2C_1,3 and D_1,3 ). In addition, we observe the establishment of the blood testis barrier (BTB) via the apparition of Kyoto Encyclopedia of Genes and Genomes (KEGGs) pathways related to tight junctions, gap junctions, and adherens junctions ( Figure S3A₃ ).

Between 6.5 dpp (corresponding to D0) and D4, we can observe a response of the tissue to the cutting at the beginning of the culture (response to stimulus, wound healing) ( Figure S4B₁ ). In addition, we observe a change in the expression of genes related to GO Terms associated with vascularization (hemoglobin and blood pressure), which is consistent with the degeneration of blood capillaries after a few days of in vitro culture. Despite oxygen supply during organotypic gas-liquid interphase culture system, the absence of vascularization leads to a loss of physiological oxygenation (response to hypoxia and decreased oxygen level). Moreover, the direct exposition to oxygen with a constant artificial oxygenation seems to induce oxidative stress to testicular explants exposed to a more harmful environment (action on CH-OH group of donors and oxidoreductase activity). Finally, a dysregulation of steroid hormones (e.g., testosterone, steroids, and angiogenesis) is present at the beginning of the culture, which raises questions about the status of Sertoli and Leydig cells maturity and/or functionality throughout the in vitro culture. In addition, after only 4 days of in vitro culture we can observe the establishment of KEGG pathways related to steroidogenesis, hypoxy (NF-κB), and inflammation response (cytokines, chemokine, HIF-1) ( Figure S3B₁ ).

Between D4 and D16, we mainly observe the presence of DEGs linked to the establishment of spermatids and spermatozoa (GO Terms related to CatSper complex, sperm, acrosomal complex, axonemal complex, cilium, flagellum, and spermatogenesis) ( Figure S4B₂ ). Between D0 and D16, we observe the overexpression of genes related to the establishment of the first wave of spermatogenesis ( Figures S4C_1,2 and D_1,2 ) and the maintenance of KEGG pathways related to inflammation response between D4 and D16 (complement, natural killer cells mediated cytotoxicity, RIG-I-like receptor, IL-17) ( Figure S3B₂ ).

Even if almost all the genes are up-regulated between D16 and D30, the number of DEGs (only 58 genes) remains very small ( Figure S4A₃ ). We observe a change in the expression of genes related to GO Terms of sperm formation (sperm individualization, gamete generation, multicellular reproductive process, sperm motility) as well as DNA modifications (chromatin condensation, DNA packaging and conformation change) and immune response (innate immune response) ( Figure S4B₃ ). In addition, only eighteen genes were differencially expressed between testicular explants cultured up to D30 compared to D16 with a |Log₁₀ p_adj| greater to 3.0 ( Figure S4C₃ ), reinforcing the idea that an abnormal kinetic throughout the second half of the first spermatogenesis and/or blockage is present at the end of spermatogenesis in vitro with the presence of inflammatory process or immune cell recruitment. Surprisingly, only one gene is down-regulated: Rps3a3 [log₂-fold change=-7.13; p_adj=9.24E-5] ( Figure S4D₃ ). Again, we can observe the maintenance of KEGG pathways related to inflammation response (IL-17, cytokine, chemokine) ( Figure S3B₃ ), symbolizing an increase in inflammation mechanisms throughout the culture.

At D4 ( Figure 7A₁ ), we can observe mainly DEGs related to GO Terms of oxidative stress (antioxidant activity), vascularization (hemoglobin, heme, blood pressure, oxygen), and response to 4 days of culture (complement, cytokine, response to stimulus) ( Figure 7B₁ ). Those results are consistent with the degeneration of blood capillaries and the relative harmfull generated on the testicular explant at the beginning of the culture in comparison to age-matched control. Among the top-ten DEGs, we notice an over-expression of Serpina3n, Complement component 3 ( C3 ), C7, and Il33 as well as an under-expression of Cyp17a1 and F13a1 ( Figures 7D₁ and D₃ ). In addition, we can observe the presence of KEGG pathways related to inflammation response (complement, cytokine), steroidogenesis and cell adhesion/senescence ( Figure S3C₁ ).

At D16 ( Figure 7A₂ ), we observe a dysregulation of genes linked to oxidative stress (oxygen, antioxidant activity), vascularization (hemoglobin, haptoglobin, blood pressure, oxygen transport and binding), response to hypoxia, and growth factors (insulin-like growth factor), response to stimulus/stress, wound healing, and cell adhesion (cell adhesion, biological adhesion) ( Figure 7B₂ ). Among the top-ten DEGs, we notice an over-expression of Igfbp3, Vldlr, Gpx3, Mmp12, Vegfa as well as an under-expression of Cyp17a1 ( Figures 7D₂ and D₃ ). In addition, we can observe the maintenance of KEGG pathways related to inflammation response (complement) and hypoxia (HIF-1) ( Figure S3C₂ ).

At D30, a large proportion of genes are down-regulated when comparing to age-matched corresponding 36.5 dpp ( Figure 7A3 ). DEGs related to GO Terms of response to cell adhesion (cell adhesion, biological adhesion), spermatozoa formation (cilium, sperm flagellum, tubulin, gamete generation), probably present due to the very low proportion of elongated spermatids in vitro compared to those observed physiologically at 36.5 dpp. In addition, GO Terms related to stress, wound healing, and tissue development (system, process, cell, etc.) were observed. Among the top-ten DEGs, we notice an over-expression of Igfbp3, Cfi, Wnt4 as well as an under-expression of Cyp17a1 ( Figures 7C₃ and D₃ ). Again, we can observe the maintenance of KEGG pathways related to inflammation response (complement, TNF signaling) ( Figure S3C₃ ), symbolizing a strong presence of inflammatory reaction from the beginning of the culture and up to 30 days.