Biological and functional characterization of bone marrow-derived mesenchymal stromal cells from patients affected by primary immunodeficiency

Mesenchymal stromal cells (MSCs) represent a key component of bone marrow (BM) microenvironment and display immune-regulatory properties. We performed a detailed analysis of biological/functional properties of BM-MSCs derived from 33 pediatric patients affected by primary immune-deficiencies (PID-MSCs): 7 Chronic Granulomatous Disease (CGD), 15 Wiskott-Aldrich Syndrome (WAS), 11 Severe Combined Immunodeficiency (SCID). Results were compared with MSCs from 15 age-matched pediatric healthy-donors (HD-MSCs). Clonogenic and proliferative capacity, differentiation ability, immunophenotype, immunomodulatory properties were analyzed. WB and RT-qPCR for CYBB, WAS and ADA genes were performed. All PID-MSCs displayed clonogenic and proliferative capacity, morphology and immunophenotype comparable with HD-MSCs. PID-MSCs maintained the inhibitory effect on T- and B-lymphocyte proliferation, except for decreased inhibitory ability of SCID-MSCs at MSC:PBMC ratio 1:10. While HD- and CGD-MSCs were able to inhibit monocyte maturation into immature dendritic cells, in SCID- and WAS-MSCs this ability was reduced. After Toll-like Receptor priming, PID-MSCs displayed in vitro an altered gene expression profile of pro- and anti-inflammatory soluble factors. PID-MSCs displayed lower PPARγ levels and WAS- and SCID-MSCs higher levels of key osteogenic markers, as compared with HD-MSCs. Our results indicate that PID-MSCs may be defective in some functional abilities; whether these defects contribute to disease pathophysiology deserves further investigation.

(RAG-1 deficiency) samples. In agreement with previous publications 29 , in the presence of HD-MSCs the levels of anti-inflammatory cytokines (IL-6, IL-10 and TGF-β) increased; a similar behavior was observed when PID-MSCs were added to the co-cultures. In particular, the amount of IL-6 detected was significantly increased in PHA stimulated PBMCs co-cultured in the presence of HD-MSCs (31 ng/ml, SEM ± 19; P < 0.05) and WAS-MSCs (30.1 ng/ml, SEM ± 4.5; P < 0.001), as compared with PHA stimulated PBMCs alone (4.5 ng/ml, SEM ± 7); by contrast, in the presence of SCID-MSCs and CGD-MSC only a trend for superior levels of this cytokines was observed (P = NS as compared with the condition without MSCs). An increase in IL-10 levels, although not statistically significant, was observed in the presence of both HD-and PID-MSCs as compared with the condition without MSCs (P = NS), the only exception being CGD-MSCs where the levels of IL-10 were lower as compared with the condition without MSCs (P = NS). Moreover, levels of TGF-β were higher in the presence of each PID-MSCs, although the P values were not statistically different as compared with HD-MSCs. The levels of pro-inflammatory cytokines (IL-2 and IFN-γ) decreased in MSC:PBMC co-cultures from both HDand PID-MSCs, as compared with PHA-stimulated PBMCs. In case of WAS-MSC co-cultures, the amount of IFN-γ was significant lower (597 pg/ml, SEM ± 370; P < 0.01), as compared with PHA stimulated PBMCs alone (6479 pg/ml, SEM ± 4920; see Fig. 2B); statistics for IL-2 and IFN-γ in the other disease groups could not be performed due the limited number of samples available for analysis.
In order to assess the immune-regulatory effect of PID-MSCs on B lymphocytes, we measured B-cell proliferation and plasma-cell generation after 7-day MSC:PBMC co-culture stimulated with CpG in an allogeneic setting. Both HD-and PID-MSCs were able to significantly inhibit B-cell proliferation and plasma-cell generation, as compared with PBMCs + CpG alone (P < 0.05 for CGD-and WAS-MSCs and P < 0.01 for HD-, ADA-and SCID-MSCs) (see Fig. 3). Moreover, no statistically significant differences were found in the ability to influence B-cell functionality between PID-MSCs and HD-MSCs (P = NS for the 4 disease groups). Inhibitory effect of PID-MSCs on monocyte maturation. To investigate whether PID-MSCs maintain the ability to modulate also innate immune responses, we focused on MSC effect on differentiation and maturation of monocyte-derived dendritic cells (DCs). Both HD-and PID-MSCs were cultured in transwell inserts with peripheral blood-derived CD14 + monocytes in the presence of IL-4 and GM-CSF, cytokines known to be able to promote the differentiation into immature CD1a + DCs (iDCs). We observed that both HD-and CGD-MSCs were able to efficiently inhibit iDCs generation; in particular, the mean percentage of CD14 + cells after co-culture with MSCs at MSC:Monocyte ratio 1:10 was 74% (SEM ± 18; P < 0.05 as compared with stimulated monocytes in the absence of MSCs) for HD-MSCs and 92% (SEM ± 6.7; P < 0.01 as compared with stimulated monocytes in the absence of MSCs) for CGD-MSCs. ADA-and SCID-MSCs displayed a significantly reduced ability to inhibit iDCs differentiation as compared with HD-MSCs with a mean percentage of CD14 + cells after co-culture of 22% and SCID-MSCs (+SCID-MSC). Each bar represents the percentage of residual proliferation of 10 5 PBMCs, in the presence of two different MSC:PBMC ratios (MSC:PBMC ratio of 1:2 and 1:10), calculated by measuring 3H-thymidine incorporation after 72 hours co-culture. We referred to PBMC proliferation alone (in the absence of MSCs) as 100% and this percentage was used to normalize PBMC proliferation in the presence of MSCs. Mean ± SEM of multiple experiments performed at least twice on 9 HD-MSCs, 7 CGD-MSCs, 12 WAS-MSCs, 6 ADA-MSCs and 5 SCID-MSCs (each point being in triplicate) is reported. P values lower than 0.05 were considered to be statistically significant (*p < 0.05; ***p < 0.001). (B) Levels of anti-inflammatory and proinflammatory cytokines in supernatants of co-cultures of 3 HD-MSCs, 3 WAS-MSCs, 2 CGD-MSCs and 2 SCID-MSCs (RAG-1 deficiency) with PBMCs, after 72-hour incubation with PHA. Results are expressed as pg/ml of the mean ± SEM. P values lower than 0.05 were considered to be statistically significant (*p < 0.05).
Furthermore, for a more meaningful characterization of MSC effect on the differentiation of monocytes into DCs, we looked also at the expression of HLA-DR and CD86 after 7-day co-culture. We found that, in the absence of MSCs and in presence of GM-CSF and IL-4, CD1a + cells expressed some levels of both HLA-DR and CD86, as reported for classic iDCs 30,31 . In line with data published da Gao et al. 30 , after co-culture with both HD-and PID-MSCs, monocytes displayed a lower, although not statistically different, expression of CD86 at all MSC:Monocyte ratios, as compared with monocytes stimulated alone (P = NS; see Fig. 4B). The levels of HLA-DR tended to increase in the presence of MSCs, both from HDs and PIDs, although the difference was not statistically significant as compared with monocytes stimulated alone (P = NS for all MSCs sample) (Fig. 4C). In this case, no differences were found between HD-and CGD-MSCs and SCID-and WAS-MSCs.

Effect of priming with TLR3 and TLR4 agonists on immune-regulatory genes expression in PID-MSCs.
Previous studies have demonstrated that MSCs express several TLRs and that stimulation with specific TLR agonists is able to polarize MSCs into a pro-inflammatory phenotype (MSC1) or an immunosuppressive one (MSC2) 32,33 . In order to better understand MSC ability to be activated by the surrounding microenvironment in diseases such as PIDs in which severe infections and inflammation are frequently present, we analyzed the effects of TLR signaling on PID-MSCs, as compared with HD-MSCs. To this aim, we performed gene expression analysis of typical MSC immune-regulatory molecules after PID-MSC stimulation with LPS, poly I:C or the combination of LPS + poly I:C. Our results indicate that PID-MSCs display in vitro an altered gene expression profile of pro-and anti-inflammatory soluble factors after TLR stimulation (see Fig. 5A,B for anti-and pro-inflammatory soluble factors, respectively). In particular, as far as CGD-MSCs are concerned, gene expression analysis showed that after TLR4-priming, CGD-MSCs displayed significantly higher levels of CXCL10 (2.1-fold increase, SEM ± 0.5; P < 0.01), as compared with HD-MSCs. When the mixture of LPS and poly I:C was employed to challenge CGD-MSCs, both Rantes and CXCL10 increased.
When WAS-MSCs were stimulated through TLR-3 or with the mixture LPS + poly I:C, we observed a significant decrease in IDO levels as compared with HD-MSCs (0.16-fold increase, SEM ± 0.07; P < 0.001) and a significant increase in HGF levels (17.1-fold increase, SEM ± 8; P < 0.05 as compared with HD-MSCs). Meanwhile, TLR-4 stimulation and the mixture LPS + poly I:C induced a significant decrease in the expression of pro-inflammatory cytokine genes (P < 0.01 for Rantes, P < 0.001 for CXCL9, P = NS for CXCL10) in WAS-MSCs. With regard to ADA-and SCID-MSCs, when TLR-3 was primed and the mixture LPS + poly I:C was employed, HGF expression significantly increased as compared with HD-MSCs (1.5-fold increase, SEM ± 0.3, P < 0.05 for ADA-MSCs and 6.8-fold increase, SEM ± 4, P < 0.01 for SCID-MSCs with poly I:C stimulation; 3.3-fold increase, SEM ± 1.5, P < 0.01 for ADA-MSCs and 3.3-fold increase, SEM ± 0.9, P < 0.01 for SCID-MSCs with the mixture LPS + poly I:C); by contrast, TLR-4 stimulation and the mixture LPS + poly I:C were associated with significantly reduced expression of pro-inflammatory molecules for both ADA-and SCID-MSCs, as compared with HD-MSCs (P < 0.001 for Rantes, CXCL9 and CXCL10). . P values lower than 0.05 were considered to be statistically significant (*p < 0.05; **p < 0.01; ***p < 0.001).

PID-MSCs ability to differentiate into osteoblasts and adipocytes.
In order to assess the differentiation capacity of PID-MSCs in comparison with HD-MSCs, cells were induced to differentiate in osteoblasts and adipocytes and analyzed by histological staining. As shown in Fig. 6, all PID-MSCs stained positive with Oil Red O, which reveales the formation of lipid droplets (Fig. 6A); moreover, they showed calcium depositions positive for Alizarin Red (see Fig. 6B) and responded to alkaline phosphatase (ALP) reaction with Fast Blue (not shown).

Discussion
With the aim of extensively characterizing their in vitro biological and functional properties, we isolated and expanded ex-vivo BM-derived MSCs from patients affected by PIDs. For a more meaningful interpretation of data, we compared PID-MSCs results with those of their HD-derived counterparts.
Our data demonstrate that MSCs can be consistently generated and expanded ex-vivo from PID patients. PID-MSCs are morphologically similar to HD-MSCs and express the combination of surface markers commonly employed for identifying BM-derived HD-MSCs 26 . When the clonogenic and proliferative abilities of PID-MSCs were analyzed and compared with those of HD-MSCs, we found that they did not differ (Fig. 1). Moreover, PID-MSCs displayed a life-span in vitro similar to that of HD-MSCs and regularly entered into replicative senescence. These data indicate that, when a congenital defect of the immune system is present, MSCs, as fundamental components of the HSC supportive niche, do not seem to display intrinsic defects in these in vitro biological characteristics.
Neither HD-nor PID-MSCs expressed CYBB and WAS at the protein and gene levels, in line with data on hematopoietic cells, where these genes are preferentially expressed by mature cells and very low o not expressed by hematopoietic progenitors 34,35 . For what concerns ADA-MSCs, the expression of the ADA gene, which is known to be ubiquitously expressed, was found to be reduced as compared with HD-MSCs; accordingly, also the levels of ADA enzymatic activity were lower in ADA-MSCs as compared with HD-MSCs.
We showed that PID-MSCs retain immune-regulatory properties on adaptive immunity typical of HD-MSCs, being able to inhibit both T-and B-cell proliferation, as well as plasma-cell generation (Figs 2 and 3). This preserved functionality in vitro of PID-MSCs might support the hypothesis of a non-direct involvement of mesenchymal progenitors in the pathophysiology of these diseases, although we cannot exclude that primary PID-MSCs in vivo may behave differently. Only ADA-MSCs and SCID-MSCs displayed a significantly reduced ability to inhibit T-cell proliferation when tested at the MSC:PBMC ratio 1:10, which was almost completely abolished for ADA-MSCs. We also evaluated the ability of MSCs to influence some aspects of the innate immunity by studying their capacity to modulate monocyte functions, as it has been reported that HD-MSCs are able to inhibit differentiation and maturation of monocyte-derived DCs 31,33,36 . We observed that both HD-and CGD-MSCs were able of efficiently preventing differentiation of monocytes into iDCs in a dose-dependent manner, whereas WAS-, ADA-and SCID-MSCs displayed a reduced ability, which was very evident in SCID-MSCs and completely abolished in ADA-MSCs (Fig. 4A). The impaired capacity of ADA-and SCID-MSCs to manage T-cell responses at high MSC:PBMC ratios and to inhibit differentiation of iDCs might be explained by the lack of stimuli by T/B and, in some cases also NK cells, in the BM microenvironment of these patients, due to their underlying immune defect. In case of ADA-MSCs, these reduced functional abilities might also be directly attributed, at least in part, to the impaired enzymatic activity in ADA-MSCs, as well as to their reduced expression of the ADA gene.
Moreover, we analyzed the expression level of pro-and anti-inflammatory molecules after priming of PID-MSCs with TLR agonists, as a model to test the ability of MSCs to respond to an inflamed microenvironment. In fact, Waterman et al. 32 , reported that MSCs are able to polarize into two distinct phenotypes following specific TLR stimulation, the pro-inflammatory MSC1 and the anti-inflammatory MSC2 phenotypes, resulting in different immunomodulatory effects and secretomes. The short incubation time that we employed and the minimal TLR agonist concentrations used are intended to mimic the gradient of danger signals that endogenous MSCs may encounter in the surrounding microenvironment, especially in clinical situations in which infections and/ or inflammation are frequently present, such as in PID patients 29 . Comparison of unstimulated MSCs, derived from PID patients and HDs, versus both MSC types primed with TLR agonists indicate that all MSCs were activated, after the challenge, to synthetize chemokines/soluble factors involved in inflammation, independently from the patient or donor origin. However, when PID-MSCs were compared with HD-MSCs, we found that and key osteogenic markers BMP2 (early), RUNX2 (early), ALPL (late) and SPP1 (late) (D) on HD-and PID-MSCs after culture in differentiation medium. mRNA levels of the adipogenic marker were quantified on day +21 of the culture, whereas mRNA levels of osteogenic markers were quantified both on day +10 and on day +21 of the culture to reveal the expression of both early and late osteogenic genes by the 2 −ΔΔCt method after normalization with respect to GAPDH. Results are expressed as fold-change relative to HD-MSCs. Each bar represents the mean ± SEM of multiple experiments (each point was performed in duplicate and repeated independently at least 3 times for 7 HD-MSC, 5 CGD-MSC, 6 WAS-MSC, 5 ADA-MSC and 5 SCID-MSC sample). P values lower than 0.05 were considered to be statistically significant (compared with HD-MSCs: *p < 0.05; **p < 0.01; ***p < 0.001). patient-derived cells displayed an altered gene expression profile in vitro of both pro-and anti-inflammatory soluble factors. This altered secretory phenotype may reflect an impaired ability of PID-MSCs to sense and respond to the surrounding microenvironment. Indeed, the frequent ongoing infections and extensive inflammation present in PID patients may influence and modify the basal activation status of MSCs, ranging from hyper-activation with increased secretion of soluble factors to exhaustion of their secretory ability and reduced molecule production. For example, a higher expression of the anti-inflammatory factor HGF in WAS-, ADA-and SCID-MSCs may reflect the attempt of MSCs to counterbalance the underlying inflammatory situation with the aim to re-establish tissue homeostasis 37 . On the other hand, the contemporary high levels of the pro-inflammatory molecule CXCL10 in TLR3-primed CGD-MSCs may indicate their inability to handle the hyper-inflammatory status typically present in CGD patients. In light of this altered secretory phenotype, also the absent/impaired ability of ADA-, SCIDand WAS-MSCs to influence maturation of monocytes may be due to modifications in PID-MSC ability to sense the surrounding microenvironment and to respond by producing proper amounts of soluble factors.
We also tested PID-MSCs ability to differentiate in vitro into adipocytes and osteoblasts. While histological staining for both adipose and bone tissue resulted positive for all PID-MSCs, when RT-PCR for key differentiation markers was performed we found differences between HD-and PID-MSCs. In particular, we found that PPARγ expression was significantly reduced in MSCs obtained from all patients, as compared with HD-MSCs, indicating an impaired ability to form adipose tissue. As regards the osteogenic differentiation, while CGD-MSCs seemed to display a reduced ability to form bone, as compared with HD-MSCs, WAS-, ADA-and SCID-MSCs expressed significantly higher levels of some osteogenic markers (ALP and SPP1 for WAS-MSCs; BMP2, RUNX2 and ALP for ADA-and SCID-MSCs), as compared with HD-MSCs (Fig. 6). HGF, which is a nutrient factor secreted by MSCs and important for tissue development and regeneration, has been reported to promote in vitro osteogenic differentiation of human MSCs 38 and to stimulate MSC-mediated osteogenic regeneration 39 . Based on these observations, the increased expression of HGF that we have found in SCID-, ADA-and WAS-MSCs may correlate with their increased capacity to differentiate into osteoblasts. Meanwhile, the reduced ability of CGD-MSCs to express early bone markers, together with their altered secretory phenotype, may be explained by the impairment in some of their functional abilities, in line with inherent defects also in the HSC compartment in this disease. Indeed, persistent chronic inflammation in CGD patients has been reported to be associated with hematopoietic proliferative stress and consequent decreased functional activity of their HSCs 40 .
In conclusion, our study provides a comprehensive characterization of BM-derived MSCs obtained from PID patients before allogeneic transplantation/HSC-GT procedure. Our results indicate that PID-MSCs, which represent a key component of the HSC niche, maintain some of the main biological properties of HD-MSCs, as well as the characteristic ability to modulate adaptive immune responses. For what concerns the ability to interact with innate immunity and to sense and respond to the surrounding microenvironment, PID-MSCs seem to be impaired in vitro in some of their functions. Whether these defects reflect an intrinsic alteration of mesenchymal compartment in these diseases, which might also negatively influence the biology and functions of the HSCs residing in the BM niche, or are secondary to the altered microenvironment deserves further investigation. For example, in case of ADA-MSCs, this might at least in part reflect an intrinsic defect due to the fact that the expression of the ADA gene, as well as its specific enzymatic activity, is defective in ADA-MSCs, as compared with HD-MSCs. Indeed, an inherent defect of osteoblast functions and a reduced capacity to support in vitro and in vivo hematopoiesis has been already reported for the BM microenvironment of ADA-deficient mice 41 . Our observations may have clinical implications for the design of MSC-based supportive strategies in the context of HSC-GT both with the aim to facilitate engraftment of gene-corrected HSCs and in view of the potential need to correct also mesenchymal progenitors, especially in those diseases in which an impaired BM stroma may be present. This concept has been previously highlighted by Jacome et al. 42 , who reported that the lentiviral transduction of unselected Fanconi Anemia BM cells could mediate an efficient phenotypic correction of both hematopoietic progenitor cells and MSCs, which are known to be defective in this disease 43 .

PID patients and healthy donors.
Thirty-three children were included in the study: 7 had CGD (median age 5 years, range 8 months-12 years), 15 WAS (median age 3 years, range 6 months-12 years), 6 ADA-SCID (median age 2 year, range 10 months-5 years) and 5 SCID other than ADA (3 RAG-1 deficiency and 2 with γ-chain defect; median age 2 year, range 6 months-3 years). Patients were diagnosed at: (i) Department of Pediatric Haematology-Oncology, Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS) Bambino Gesù Children's Hospital, Rome; (ii) University Department of Pediatrics, Unit of Immune and Infectious Diseases of IRCCS Bambino Gesù Children's Hospital, Rome; and (iii) San Raffaele-Telethon Institute for Gene Therapy (SR-TIGET), Pediatric Immunohematology, San Raffaele Scientific Institute, Milano. Patient BM was collected, after obtaining parental informed consent, during diagnostic procedures or central venous line placement according to the San Raffaele Hospital-approved research protocol for patho-physiology studies in immune-deficient patients. All experiments were performed in accordance with relevant guidelines and regulations. MSCs were isolated and expanded ex-vivo from BM aspirates of patients and from residual BM cells of 15 pediatric HDs (median age 8 years, range 2-15 years), who donated BM for hematopoietic cell transplantation at Bambino Gesù Children's Hospital, used as controls. Patient and HD characteristics are summarized in Table 1. HD-derived peripheral blood mononuclear cells (PBMCs) were isolated from buffy-coats obtained from the Unit of Immuno-Hematology and Transfusion Medicine, Bambino Gesù Children's Hospital, Rome, using a Ficoll-Paque density gradient.

Characterization of BM-derived PID-and HD-MSCs. CFU-F ability and proliferative capacity.
Fibroblast colony-forming unit (CFU-F) formation was assessed by examining the cultures at day +7; the clonogenic efficiency was calculated as the number of colonies per 10 6 MNCs initially seeded for HD-, CGD-, WASand ADA-MSCs. Population doublings (PDs) were determined at each passage for each MSC sample by using the formula log 10 (N)/log 10 (2) where N means cells harvested/cells seeded; PDs were calculated for HD-, CGD-, WAS-, ADA-and SCID-MSCs; results were expressed as PD from passage (P) 1 to P5. Western-Blot (WB) for Gp91phox and WAS protein (WASp) expression. Cells were lysed in RIPA buffer (Thermo Scientific, Rockford, IL) containing phosphatase (PhosSTOP phosphatase inhibitor cocktail tables, Roche) and protease inhibitors (Protease inhibitor cocktail tables; Roche) or in an in-house prepared Laemly Buffer (WASp) for 30 minutes on ice. Lysates were collected and stored at −20 °C. Lysates and supernatants were separated on 10% Mini-Protean TGX Gels (BioRad) with TRIS buffer and transferred into Trans-Blot Turbo Transfer membranes (BioRad). Membranes were blocked for half an hour with Blocking Buffer (TBS containing 0.05% Tween 20 and 5% low fat milk). Membranes were then incubated with primary antibody diluted in blocking buffer for 1 hour at room temperature or overnight at 4 °C. The following antibodies were used: mouse anti-human NOX2 (1:500; Sanquin Blood Supply Foundation), mouse anti-tubulin (1:3000, Sigma), rabbit anti-Human WASP (1:500; Clone H250, Santa Cruz) and mouse anti-Human GAPDH (1:500; MAB374, Millipore). After 3 washing steps, membranes were incubated with horseradish peroxidase-linked secondary antibody diluted in blocking buffer for 1 hour at room temperature (goat anti-mouse or anti-rabbit immunoglobulins/HRP, 1:3000, Dako Denmark A/S). After 3 further washing steps, proteins of interest were detected using Peroxidase substrates (GE Healthcare Life Sciences).
ADA enzymatic activity. ADA enzymatic activity was analyzed as reported by Carlucci et al. 45 in ADA-MSCs (3 patients) and HD-MSCs (3 HDs).
In vitro T-cell proliferation assay with phytohemagglutinin. PBMCs were purified by conventional Ficoll separation from heparinized samples obtained from HDs. Cells were processed and used immediately after collection. Before MSC:PBMC co-culture, MSCs were stimulated with IFNγ (10 ng/ml; R&D Systems, In vitro B-cell proliferation assay with CpG. Also  (1 µg/ml, Sigma-Aldrich) were used as agonists for TLR4 and TLR3, respectively, in order to activate MSCs. Briefly, 1 × 10 5 MSCs were plated on 6-well plates and grown to 60-70% confluency in complete culture medium prior to the start of the experiments. TLR-agonists were added to fresh medium for 1 hour; thereafter, MSCs were washed twice in complete medium without the TLR-agonists and incubated for 18 hours before collecting their RNA for gene expression assay.

PID-MSC differentiation capacity into osteoblasts and adipocytes.
The osteogenic and adipogenic differentiation capacity of MSCs was assessed at P2-P4, as previously described 29 . To detect osteogenic differentiation, cells were stained for alkaline phosphatase (AP) activity using Fast Blue (Sigma-Aldrich) and for calcium deposition with Alizarin Red (Sigma-Aldrich). Adipogenic differentiation was evaluated through the morphological appearance of fat droplets stained with Oil Red O (Sigma-Aldrich).