An early‐senescence state in aged mesenchymal stromal cells contributes to hematopoietic stem and progenitor cell clonogenic impairment through the activation of a pro‐inflammatory program

Abstract Hematopoietic stem and progenitor cells (HSPC) reside in the bone marrow (BM) niche and serve as a reservoir for mature blood cells throughout life. Aging in the BM is characterized by low‐grade chronic inflammation that could contribute to the reduced functionality of aged HSPC. Mesenchymal stromal cells (MSC) in the BM support HSPC self‐renewal. However, changes in MSC function with age and the crosstalk between MSC and HSPC remain understudied. Here, we conducted an extensive characterization of senescence features in BM‐derived MSC from young and aged healthy donors. Aged MSC displayed an enlarged senescent‐like morphology, a delayed clonogenic potential and reduced proliferation ability when compared to younger counterparts. Of note, the observed proliferation delay was associated with increased levels of SA‐β‐galactosidase (SA‐β‐Gal) and lipofuscin in aged MSC at early passages and a modest but consistent accumulation of physical DNA damage and DNA damage response (DDR) activation. Consistent with the establishment of a senescence‐like state in aged MSC, we detected an increase in pro‐inflammatory senescence‐associated secretory phenotype (SASP) factors, both at the transcript and protein levels. Conversely, the immunomodulatory properties of aged MSC were significantly reduced. Importantly, exposure of young HSPC to factors secreted by aged MSC induced pro‐inflammatory genes in HSPC and impaired HSPC clonogenic potential in a SASP‐dependent manner. Altogether, our results reveal that BM‐derived MSC from aged healthy donors display features of senescence and that, during aging, MSC‐associated secretomes contribute to activate an inflammatory transcriptional program in HSPC that may ultimately impair their functionality.


| INTRODUCTION
Hematopoietic stem and progenitor cells (HSPC) can self-renew and differentiate into all blood components thus serving as a reservoir for mature blood cells throughout life. However, as we age, HSPC functionality is impaired with cells displaying a reduced capacity to maintain tissue homeostasis (Geiger, de Haan, & Florian, 2013).
Indeed, although the number of phenotypically defined HSPC increases with aging, aged HSPC display impaired regenerative capacity when transplanted into host recipients and a differentiation skewing toward the myeloid hematopoietic lineage (Geiger et al., 2013).
Hematopoietic stem and progenitor cells reside in the BM niche, and their function is supported by a variety of both hematopoietic and nonhematopoietic cell types, such as osteoblasts, adipocytes, endothelial, and mesenchymal stromal cells (MSC) (Morrison & Scadden, 2014). Among these, BM-derived MSC are multipotent cells of mesodermal origin capable of adhering to culture dishes, proliferate in vitro, and differentiate into different lineages, including osteoblasts, adipocytes, and chondrocytes. Several studies highlighted the key role of MSC in regulating HSPC fate and promoting engraftment of the rare and more primitive hematopoietic stem cells (HSC) (Kfoury & Scadden, 2015). Even if the molecular mechanisms that regulate HSPC dysfunction in the elderly are thought to be primarily cell-intrinsic, recent insights support the contribution of external cues from the niche to HSPC dysfunction during aging (Adams et al., 2007;Geiger et al., 2013;Mendez-Ferrer et al., 2010). Indeed, changes in the cellular composition of the HSC niche during aging contribute to hematologic decline and involve decreased bone formation, enhanced adipogenesis, increased BM inflammation, and altered HSPC-MSC crosstalk (Mendelson & Frenette, 2014). Consistent with this, some studies, mainly conducted in mouse settings, have reported that MSC decrease in number with aging and display a more pronounced differentiation toward the adipogenic lineage at the expenses of bone formation (Liu, Xia, & Li, 2015). Senescent cells accumulate during aging (Campisi, 2005;Farr et al., 2017) and contribute to tissue dysfunction and impaired tissue regeneration. Senescent cells display an enlarged morphology coupled with a proliferation arrest mediated by the Rb/p16 and p53/p21 pathways. Senescence is also characterized by increased SA-β-Gal activity, persistent DDR activation, and telomeric attrition. Moreover, senescent cells exhibit transcriptional activation of a senescent-associated secretory inflammatory phenotype collectively known as SASP (Coppe et al., 2008). The robust secretion of SASP chemokines/cytokines triggers an inflammatory response that could reinforce senescence in a cell-autonomous fashion and be transferred to surrounding cells through paracrine mechanisms, to amplify the senescence response. To date, the activation of a senescence program in human aged MSC and the interplay between aged MSC and HSPC remain to be elucidated. In this study, we successfully established human BM-derived MSC from young and elderly healthy donors. We investigated the effects of chronological age on MSC properties and found that MSC derived from aged healthy subjects show senescence-like features comprising an enlarged morphology, reduced proliferation capacity, delayed cell cycle progression, and increased levels of SA-β-Gal and lipofuscin. Importantly, we found that aged MSC activate a SASP-like program that contributes in a non cell autonomous manner to impair young HSPC clonogenicity by mediating an inflammatory state in HSPC.

| Aged MSC display reduced clonogenic capacity compared to younger counterparts
To identify molecular determinants of MSC aging, we derived MSC from bone marrow (BM) of healthy subjects belonging to three different age groups: pediatric (<18 years) (n = 4), young adults (18-35 years) (n = 6), and aged subjects (≥70) (n = 12) (See Experimental Procedures section for details). Pediatric and young adult subjects were BM donors for hematopoietic stem cell transplantation (HSCT), while BM samples from aged donors were collected from healthy subjects undergoing primary hip replacement. In the efforts to identify features of healthy chronological aging in MSC and to avoid any confounding effect due to underlying age-related pathologies, we excluded from our analysis subjects with compromised immune functions and previous history of cancer or cancer treatments, even if most elderly subjects in our cohort were diagnosed with hypertension. Of note, none of the elderly subjects under study displayed symptoms of overt frailty upon clinical assessment.
We isolated mononuclear cells (MNC) from BM by density gradient centrifugation; after depletion of CD34 + cells purified from MNC by immunomagnetic selection, we plated the remaining CD34 − fraction for MSC ex vivo expansion in culture medium supplemented with 5% platelet lysate, state-of-the-art protocol for derivation of human MSC for clinical use in transplantation settings and regenerative medicine (Avanzini et al., 2009;Ingo et al., 2016). First, we verified that established MSC from young and aged BM conform to the International Society for Cell & Gene Therapy (ISCT) minimal definition criteria that include plastic adherence and spindle-shape morphology, immunophenotype, and multilineage differentiation potential (Dominici et al., 2006). We first analyzed MSC samples from the three age groups by flow cytometry and evaluated the expression of established MSC surface markers. We observed that MSC from pediatric, young adults, and aged subjects expressed canonical MSC markers including CD90, CD105, and CD73 and were negative for the expression of hematopoietic/endothelial markers such as CD45, CD34, CD14, CD31, and HLA-DR (representative examples shown in Figure 1a). We next evaluated the clonogenic potential of MSC by measuring colonies formed unit-fibroblasts (CFU-F), a property of the more primitive enriched MSC subsets (Sacchetti et al., 2007;Tormin et al., 2010). At the time of MSC derivation, we counted the number of clones generated at 7 and 14 days normalized for the number of MNC seeded. Interestingly, despite some individual donor variability, we did not detect any significant difference in the clonogenic capacity of MSC derived from pediatric and young adult donors (hereafter used as a unified group and defined as "young"). Instead, MSC from aged individuals displayed a significant reduction of CFU-F ability at day 7 compared to younger counterparts that was partially rescued at day 14 (Figure 1b).
In order to identify intrinsic alterations in MSC phenotype during advanced aging and their impact on hematopoietic stem cell functionality, we randomly selected up to 8 aged MSC samples and consistently analyzed them in comparison with up to 6 MSC samples from the "young" group. To minimize the effect of culture-induced alterations on MSC biology and functionality and better resemble MSC properties in the BM, we focused all our analyses on ex vivo expanded MSC at early passages in culture (between passage P2 and P4).
Given that the CFU-F ability was impaired in aged MSC, we evaluated the expression of two different surface markers that are used to identify primitive MSC, CD146 (melanoma cell adhesion molecule -MCAM) (Sacchetti et al., 2007), and CD271 (nerve growth factor receptor-NGFR) (Mabuchi et al., 2013). Flow cytometric analysis revealed a significant decrease in the % of CD146 + cells and reduced intensity of CD146 expression in the analyzed aged MSC when compared to young MSC (Figure 1c-e), whereas the frequency and intensity of CD271 + MSC were not altered during aging .
We next evaluated the differentiation potential of young and aged MSC. Early passage MSC were induced to differentiate into osteogenic and adipogenic lineages by culturing the cells for three weeks (day 21) in the presence of media enriched in dexamethasone, L-ascorbic acid, insulin, methylxanthine, indomethacin, and β-glycerol phosphate for adipocyte differentiation, and dexamethasone, L-ascorbic acid, and β-glycerol phosphate for development of the osteogenic lineage. When we measured the levels of expression of adipogenic genes (PPARγ, FABP4, LPL) at steady state, we did not observe any significant difference among young and aged MSC (Supporting information Figure S1a-c), except for a decrease in PPARγ levels in aged MSC (Supporting information Figure S1a). At day 21 of differentiation, aged MSC displayed a similar ability to differentiate toward adipocytes compared to younger counterparts as revealed by induction of PPARγ levels, a key player involved in the initial phase of adipocyte differentiation, and FABP4 and LPL genes, involved in the later phases of adipocytes commitment (Supporting information Fig-ure S1d-f). Consistent with this, young and aged MSC displayed similar accumulation of lipid droplets upon differentiation induction, as shown by Oil Red O staining (Supporting information Figure S1g). We next evaluated the levels of genes involved in osteogenic differentiation (RUNX2, SPARC, COL1A2) and observed comparable mRNA levels in young and aged MSC at steady state (Supporting information Figure S1h-j). Upon 21 days of differentiation, we observed significant induction of osteogenic genes in young MSC, while the levels of RUNX2, SPARC, and COL1A2 did not reach significance in differentiated aged MSC. Moreover, we reported a decrease in the induction of RUNX2 and COL1A2 in aged MSC compared to young MSC upon differentiation (Supporting information Figure S1k Altogether, these data indicate that MSC were successfully derived from aged BM and displayed an impaired clonogenic potential associated with a significant reduction of mesenchymal primitive cell surface markers and with a mild impairment in the differentiation ability toward the osteogenic lineage.

| An early-senescence state characterizes aged human BM-derived MSC
Having observed a reduced clonogenic capacity of aged MSC compared to younger counterparts, we asked whether aged MSC display reduced proliferation capacity and induction of cell senescence markers even at early passages in culture. First, we noticed that the morphology of aged MSC was more flat and enlarged compared to spindle-shaped young cells, resembling senescent fibroblasts (Figure 2a). Since fully senescent cells are characterized by a stable cell cycle arrest, we measured the proliferative ability of young and aged MSC. We carried out bromodeoxyuridine (BrdU) incorporation assays to evaluate the percentage of MSC in the active S-phase of the cell cycle and detected a significantly lower proliferation rate of aged MSC compared to young MSC at early passages in culture (Figure 2b,c), although 50% of aged MSC were still able to enter in Sphase under these conditions. To better quantify cell cycle progression in single human MSC from young and aged donors, we generated a lentiviral bicistronic reporter vector encoding fluorescent ubiquitination-based cell cycle indicator probes (Fucci system). The lentiviral vector expresses mVenus-hGeminin(1/110) fused to mCherry-hCdt1(30/120) by the T2A peptide using an EF1α promoter that generates optimal levels of gene expression in primary cells (Pineda et al., 2016). These fluorescent reporters allow us to discriminate between three cell cycle phases as G1 by red fluorescence, G1/S by yellow fluorescence, and S/G2/M by green fluorescence.
After a one-hit transduction protocol, we monitored cell cycle transit up to 6 days in live imaging from young and aged MSC. Lentiviral transduction did not alter cell morphology or induced overt cell GNANI ET AL.  When we analyzed the expression levels of cyclin-dependent kinase (CDK) inhibitors associated with senescence in early passages MSC, we observed a significant increase in CDKN1A mRNA levels in aged MSC compared to younger MSC ( Figure 2g) and only a trend toward differential expression of CDKN2A between the two MSC groups ( Figure 2h). These data indicate that aged MSC were not fully arrested in the cell cycle at early passages but rather displayed a slowdown in their proliferation capacity.
Consistent with the induction of a senescence-like state in early passages aged MSC, we observed a significant increase in the percentage of cells with a weak but detectable accumulation of SA-β-Gal cytosolic signal (Figure 2i,j). Similarly, when we evaluated lipofuscin accumulation by the biotin-linked Sudan Black B (SBB) analogue staining (Evangelou et al., 2017), we found that aged MSC at early passages in culture display detectable lipofuscin granules compared to younger counterparts (Supporting information Figure S2a), further supporting the establishment of a "early-senescence" phenotype in BM-derived MSC from aged subjects.

| Human BM-derived aged MSC display modest accumulation of DNA damage, ROS, and activation of DDR
To further delve into the molecular mechanisms of the observed early-senescence state of aged MSC, we evaluated the accumulation of physical DNA damage and the activation of the DDR pathway.
We first performed alkaline comet assay to detect accumulation of both single-and double-strand breaks in young and aged MSC.
Quantitative analysis of a panel of MSC from young and aged donors indicated a significant increase in DNA damage accumulation in aged MSC (Figure 3a,b).
We next evaluated whether aged MSC had increased amount of oxidative stress. We measured the levels of reactive oxygen species (ROS) and reported a significant increase in ROS levels in aged MSC compared to younger MSC (Supporting information Figure S2b). The accumulation of cellular ROS was in line with the decreased expression levels of FOXO4 in aged MSC (Supporting information Fig-ure S2c), a factor that has been previously shown to protect from oxidative stress during cellular homeostasis (Klotz et al., 2015). We also tested whether aged MSC display signs of oxidative DNA damage by measuring oxidation on guanosine residues by immunostaining with an antibody against 8-oxo-dG. Apparently, no increased oxidative DNA damage was detected in aged samples compared to young MSC (Supporting information Figure S2d). We next evaluated

| Aged MSC display a pro-inflammatory SASPlike program
Mesenchymal stromal cells exert potent anti-inflammatory and immune-suppressive functions on cells from the adaptive and innate F I G U R E 1 Biological characterization of young and aged MSC. (a) Representative flow cytometric plots of immunophenotypic characterization of MSC from pediatric, young adults, and aged subjects. Canonical MSC markers: CD90, CD105, and CD73. Hematopoietic markers: CD45, CD34, and CD14. MHC class II marker: HLA-DR. Endothelial marker: CD31. (b) Colony-formation assay (CFU-F) at day 7 and day 14 after the initial seeding. Data are expressed as CFU/10 6 MNC plated and shown as scatter dot plot; lines indicate median values (pediatric, n = 4; young adults, n = 6; aged, n = 12). p-value was determined by Mann-Whitney test; ****p < 0.0001; ***p < 0.001. Overall, these findings indicate that aged MSC display reduced immunomodulatory properties and increased levels of pro-inflammatory molecules and suggest that aged stromal cells may contribute to alterations of the BM niche in a non cell-autonomous fashion.

| Aged MSC secretomes activate inflammatory genes in young HSPC
We next investigated possible changes in the crosstalk between MSC and HSPC during aging. We first evaluated the expression levels of key factors involved in mediating supportive role of MSC in maintaining HSPC homeostasis in BM microenvironment, including CXCL12, VCAM, VEGF, and ANGPT1. None of these factors was significantly altered or diminished in aged MSC compared to those derived from young subjects (Supporting information Figure S4j-m).
Given the above-mentioned activation of a pro-inflammatory transcriptional program in aged MSC (see Figure 4), we tested whether MSC secretomes could affect HSPC function. We collected conditioned medium (CM) from young and aged MSC and cultured umbilical cord blood (CB)-derived CD34 + HSPC admixing hematopoietic stem cell media with MSC-derived CM. We analyzed the effects of aged MSC secretome on the functionality of CD34 + cells by evaluating HSPC immune-phenotype, clonogenic output in methylcellulose assays and gene expression changes ( Figure 5a). As control we F I G U R E 2 Aged MSC are characterized by an early senescent state. (a) Representative brightfield images of BM-derived MSC isolated from young and aged donors at early passages in culture. (b) Representative confocal images of BrdU incorporation assay in young and aged MSC. DAPI indicates nuclei, scale bar = 20µm (young, n = 4; aged, n = 4). Quantification of BrdU-positive cells is shown in (c). p-value was determined by Mann-Whitney test; *p < 0.05. (d-e) Representative cell cycle kinetics of young (d) and aged (e) MSC as determined by mean fluorescent intensity from Fucci2A-transduced cells. (f) Quantification of cell cycle kinetics of MSC from one young and two aged MSC donors measured in hours. 10 cells per each donor were analyzed in live imaging up to 6 days. Each squared dot represents a complete phase of cell cycle. Data from two aged MSC samples were represented as a unified group. The median duration of cell cycle phases, with the minimum and maximum length in brackets (hours), is reported in the table. p-value was determined by Mann-Whitney test; **p < 0.01; *p < 0.05; ns p > 0.05. (g-h) Relative mRNA expression of CDKN1A (g) and CDKN2A (h) as measured by quantitative Real-Time PCR. Gene expression data are represented as 2 −▵CT relative to GUSB housekeeping. Each squared dot represents an individual MSC donor (young, n = 6; aged, n = 8) (red = young; blue = aged); lines indicate median 2 −▵CT values. p-value was determined by Mann-Whitney test; *p < 0.05. (i) Representative pictures and quantification (j) of SA-β-Gal-positive MSC isolated from young and aged donors at early passages. Human senescent fibroblasts (BJ) induced into senescence by irradiation are shown as positive control for SA-β-Gal staining in (i). DAPI was used to stain nuclei. Scale bar = 50 µm (young, n = 4; aged, n = 8). p-value was determined by Mann-Whitney test; **p < 0.01 Albeit not significant, we also reported a trend toward an increase for IL1α and IL6 mRNA levels in cells grown with CM derived from aged MSC (Supporting information Figure S5d,e).
We next tested the hypothesis that aged MSC from subjects with lower burden of chronic inflammation would have limited impact on HSPC functionality. We took advantage of MSC derived Overall, these results indicate that aged MSC release SASP factors that propagate inflammatory signals to neighboring HSPC in a paracrine fashion and may in turn contribute to the reduced clonogenic capacity of recipient HSPC.

| SASP inhibition in aged MSC rescues clonogenicity in young HSPC
In order to assess the direct effects of SASP inhibition on aged MSC and identify the molecular events that regulate the propagation of inflammation from MSC to HSPC, we employed ex vivo SASP inhibitor treatments on aged MSC. We first performed a 6 days long treatment with corticosterone, a glucocorticoid previously reported to dampen activation of SASP in senescent fibroblasts (Laberge et al., 2012). We measured the transcript levels of a panel of pro-inflammatory SASP molecules and found a dose-dependent F I G U R E 3 Aged MSC accumulate DNA damage and display modest DDR activation. (a) Representative fluorescence pictures of DNA damage in young and aged MSC as detected by comet assay; scale bar = 50 µm. (b) Quantification of alkaline comet assay carried out in young and aged MSC at early passages in culture (young, n = 5; aged, n = 5); up to 75 nuclei per sample were analyzed; histograms represent mean olive tail moment value ±SEM of young and aged MSC; p-value was determined by Mann-Whitney test; *p < 0.05. (c) Representative confocal pictures and (d) immunofluorescence quantification for 53BP1 foci-positive cells in young and aged MSC at early passages in culture. Human BJ fibroblasts analyzed 2 hr post irradiation (20 Gy) were shown as positive control for 53BP1 nuclear staining in (c). Nuclei were counterstained with DAPI; scale bar = 20 µm (young, n = 5; aged, n = 5); p-value was determined by Mann-Whitney test; **p < 0.01. (e) Representative z-stack confocal pictures of telomeric signal in young and aged MSC. Each red dot represents a telomere identified by a PNA probe against telomeric sequences. Scale bar = 20 µm. (f) Mean intensity of telomeric signal quantification was calculated with cell profiler (young, n = 5; aged, n = 5). Histograms represent mean values ±SEM of MSC samples analyzed (red = young; blue = aged). At least 20 nuclei were analyzed per sample with identical laser parameters. DAPI was used to stain nuclei. Scale bar = 20 µm.   HSPC for 96 hr before assessing their clonogenic potential in methylcellulose assays (Figure 6b). Whereas HSPC exposed to CM from aged MSC displayed a reduced clonogenic potential compared to control or HSPC exposed to CM from young MSC, HSPC exposed to CM from corticosterone-treated aged MSC gave rise to a significantly higher number of colonies (Figure 6c). We also showed that the inflammatory program of aged MSC could be transferred in a paracrine fashion to HSPC, and likely contribute to  (Lujambio, 2016).

Because NFkB acts as a master regulator of the SASP transcrip
The multipotent potential of MSC and their differentiation ability have been extensively dissected in mouse and human settings. In F I G U R E 6 SASP inhibitors rescue the clonogenic impairment of young HSPC exposed to CM from aged MSC. (a) Relative mRNA levels of IL1α, IL1β, MCP1, IL6, and IL8 revealed by quantitative real-time PCR in late passages aged MSC treated with vehicle CTRL (Ethanol), 0.5 µM, or 2.5 µM corticosterone for 6 days. GUSB was used as housekeeping gene; histograms represent fold change + SD relative to CTRL.  aging. Of note, in our cohort, in order to avoid any confounding effect due to underlying age-related diseases, we excluded from the analysis patients with significant co-morbidities and inflammatory conditions, including cancer and immunological defects.
Our findings indicate for the first time that early senescent BMderived MSC activate a robust pro-inflammatory transcriptional program and display reduced immunomodulatory capacity toward cells of the adaptive immune system. As a consequence, aged MSC contribute to create and sustain an inflammatory milieu in the BM niche.
These data indicate that, in concert with a cell-autonomous dysfunction of aged MSC, extrinsic factors may as well alter MSC crosstalks with other BM cellular components. TGF-β pathway has only recently emerged as a key mediator of the senescence associated with aged MSC derived from osteoarthritic bones (Kawamura et al., 2018).
When measuring the expression levels of TGF-β in our aged MSC, we did not observe any significant increase compared to younger MSC. Instead, the levels of well-established SASP factors, including IL1, IL6, IL8, and MCP1, were higher in aged MSC and actively secreted. In particular, MCP1 was among the most highly expressed SASP factors in aged MSC both at the transcript and protein levels. Consistent with these findings, MCP1 was reported to be epigenetically repressed in umbilical cord blood-derived MSC and activated only when cells reach premature senescence due to prolonged culture conditions (Jin et al., 2016).
Here, we provide evidence that factors secreted by aged MSC drive an inflammatory transcriptional program in young HSPC and contribute to their clonogenic impairment. Over the past decade, a growing body of evidence revealed that inflammatory stimuli alter HSPC fate and functionality by affecting HSPC proliferation/quiescence status, differentiation potential, or HSPC-niche interactions. In particular, it has been reported that chronic inflammation drives HSPC myeloid skewing and leads to HSPC exhaustion during aging (Essers et al., 2009;Haas et al., 2015;Pietras et al., 2016).  Table S1. BJ normal human fibroblasts were purchased from ATCC and grown in DMEM supplemented with 10% FBS (EuroClone), 1% penicillin/streptomycin (Pen/Strep, EuroClone), and 2 mM L-glutamine (L-Glu, EuroClone).

Senescence in BJ cells was induced by irradiation (20 Gy), and cells
were analyzed for senescence markers two weeks post-treatment.
Irradiated BJ cells analyzed 2 hr post-treatment were used a positive control for 53BP1 foci in Figure 3c.

| BrdU incorporation assay
For BrdU incorporation analysis, early passages MSC derived from young and aged donors were seeded at a density of 3 × 10 4 /well GNANI ET AL.

| Alkaline COMET assay
To detect DNA damage in young and aged MSC, we employed alkaline comet assay. Specifically, after trypsinization, 2 × 10 3 MSC were mixed with molten Comet LMAgarose (Trevigen, MD) at a ratio of

| Fluorescence in situ hybridization
To analyze telomeres in early passages young and aged MSC, we

| Purification of CD34 + cells from cord blood
Human CD34

| HSPC colony-forming assays
For the colony-forming unit (CFU) assays, HSPC that were cultured with MSC-derived CM were reseeded onto a 35-mm culture dish (Corning) at a density of 8 × 10 2 cells per dish in triplicate in methylcellulose medium (MethoCult, Stemcell). After 14 days, hematopoietic cell-derived colonies were counted under a light microscope.
The colony types were identified and defined as myeloid (white), erythroid (red), or mixed (gray).

| Statistical analysis
All data are presented as median values or mean ± SD or ±SEM, as indicated. Mann-Whitney test was used for comparisons between two experimental groups. Data were analyzed upon consulting with biostatisticians at CUSSB (University Center for Statistics in Biomedical Sciences) within the San Raffaele Hospital, Milan. Graphs were generated by Prism software v8 (GraphPad Software Inc.). p values <0.05 were considered significant (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).

ACKNOWLEDGMENTS
We thank all members of Di Micco's laboratory for discussion, the