Ectonucleotidase Expression on Human Amnion Epithelial Cells: Adenosinergic Pathways and Dichotomic Effects on Immune Effector Cell Populations

Stem cells have gained acceptance in cell-based therapies, and their “off-the-shelf” availability has been recognized as being essential for the emerging field of regenerative medicine. The placental membrane in direct contact with the fetus is a tissue of fetal origin (amnion) and has been used in a variety of disorders (1, 2). Amnion tissue consists of cuboidal epithelial cells firmly adherent to a thick basal membrane and an abundant extracellular matrix with interspersed fibroblast-like cells. To our knowledge, our group was the first to identify and describe the nature of stem cells of human amnion epithelial cells (hAEC) (3). We developed efficient techniques to isolate hAEC with epithelial characteristics and a heterogeneous expression of membrane surface Ags (4). The hAEC phenotype includes classic epithelial adhesion molecules, such as EpCAM (CD325) and integrin subunits (CD29 and CD49f). On the contrary, markers routinely used to define and identify stromal cells are not detectable (4, 5). The surface markers commonly observed on embryonic stem cells (e.g., globoseries glycolipids SSEA-3 and -4 and keratansulfate-associated Ags TRA 1-60 and 1-81) are consistently expressed, perhaps contributing to their multilineage differentiation potential (1, 3, 6). Unlike pluripotent cells, hAEC are not immortal; they do not express telomerase and maintain a normal karyotype, without resulting in tumor formation upon transplantation (3, 7). hAEC have shown preclinical and clinical efficacy (1, 8). Over the past years, hAEC have been suggested as a useful cell type for the treatment of different life-threatening models of inborn error of diseases (9–11). The results on transplants confirmed that hAEC are capable of a functional maturation into hepatic cells in addition to their characteristic immune-modulatory and anti-inflammatory properties (12). In fact, the same cells maintain immunosuppressive properties in animal models of autoimmune/inflammatory disease (13–15). Their safety and the absence of immune rejection were confirmed after transplantation in immune-competent mouse models as well as in human volunteers and patients (16–18).

Mismatches in HLA expression are recognized by immune cells and are generally followed by rejection. The characteristic expression of noncanonical HLA-Ib (i.e., HLA-G) was proposed as a possible explanation for long-term engraftment and survival of hAEC in allogenic or xenogenic settings (12). Among immunosuppressive molecules, adenosine (ADO) has been recently identified to play an important role in different physiological and pathological settings. ADO is produced from the catabolism of mono- and di-nucleotides (ATP and NAD⁺) and from their byproducts (ADP, ADP ribose [ADPR], and AMP). The canonical pathway of ADO production starts from CD39, an ecto-nucleoside triphosphate diphosphohydrolase that converts ATP to ADP and then into AMP. The latter molecule is then converted to ADO by the 5′-nucleotidase CD73 (19). CD203a/PC-1 was recently proposed as a hinge ectoenzyme for its ability to convert both ADPR (generated from NAD⁺ by CD38) or ATP to AMP, subsequently metabolized by CD73 in ADO (20, 21).

Adenosinergic ectoenzymes were identified on different cell populations, such as immune effectors (22) and tumor cells (22), conferring immunosuppressive properties to these cells. Indeed, increased ADO concentrations are measured in several pathological conditions, such as tissue injury, ischemia, hypoxia, inflammation, trauma, and cancer (23). ADO can inhibit immune effector cells through ADO receptors (ADOR) (24, 25). Specific ligation of ADORA2a and ADORA2b on T lymphocytes impairs cytotoxicity and cytokine release (23, 26, 27), whereas engagement of ADORA2a on NK cells inhibits cytotoxicity and IFN-γ release (23). Moreover, ADO dampens macrophage activation through ADORA2a ligation and induces M2 polarization via ADORA2b. Lastly, through the ligation of ADORA2b, ADO inhibits maturation and differentiation of dendritic cells, leading to a defective Ag presentation (23).

Our study is aimed at characterizing the immunoregulatory properties of primary hAEC, primarily focusing on the level of expression of different ectonucleotidases and the role played by these molecules in the adenosinergic pathways responsible for immunomodulatory properties.

Discarded placentae from uncomplicated cesarean resections at 37–42 wk of gestational age were obtained from Karolinska University Hospital (Stockholm, Sweden) after the approval by the Institutional Review Board. Placental hAEC were isolated as previously described (4). Briefly, amnion membrane was surgically removed from the surface of the placenta and washed to remove blood cells. TrypLE (Life Technologies) digestion (30 min) was specifically performed to release epithelial cells without altering cell surface Ags, as previously described (3, 4). Viability after isolation ranged from 90 to 97%, whereas after cryopreservation it ranged from 85 to 94%. hAEC were seeded on Gelatin 0.2% and maintained for few days (7). Standard culture media consisted of DMEM supplemented with 10% FBS, 1 mM nonessential amino acids, 4 mM l-glutamine, 55 μM 2-ME (all from Sigma-Aldrich), and antimycotic/antibacterial. After discarding nonadhering cells, adherent cells were cultured in fresh medium until confluency. When specified, 10 ng/ml recombinant human epidermal growth factor (rhEGF; Immunotools) was supplemented to the cells in adhesion for 3 d. hAEC were detached using Trypsin/EDTA solution (Sigma-Aldrich) and then used for further experiments.

Peripheral blood (PB) samples were obtained from healthy donors of the Blood Bank of Istituto Giannina Gaslini (Genoa, Italy). CD4⁺ T cells, B cells, and NK cells were isolated by incubating the whole blood samples with RosetteSep Human CD4⁺ T cells, B cells, or NK cells enrichment mixture, respectively (StemCell Technologies), following manufacturer’s protocol. Untouched CD4⁺ T cells, B cells, or NK cells were then obtained by Ficoll-Hypaque (Sigma-Aldrich) density gradient centrifugation.

Evaluation of surface expression of ectoenzymes was performed on hAEC by means of anti-CD39 PE-Cy7 mAb (eBiosciences) and a panel of mAbs generated in our laboratory and FITC or allophycocyanin conjugates by Aczon (Bologna, Italy): anti-CD38 (clone IB4), anti-CD73 (clone CB73), anti-CD203a/PC-1 (clone 3E8, kindly provided by J. Goding), and anti-CD157 (clone SY/11B5). FITC- or allophycocyanin-conjugated irrelevant isotype-matched mAbs were purchased from Beckman Coulter. Expression of ADO deaminase (ADA)/CD26 complex was evaluated using PE-conjugated anti-CD26 mAb (clone CB26), locally generated.

Expression of ADOR was evaluated on hAEC as well as on CD4⁺ T and B lymphocytes and NK cells isolated from five normal donors (see above), using rabbit polyclonal Abs anti-ADORA1 (LifeSpan BioSciences), anti-ADORA2a, and anti-ADORA2b (Thermo Scientific). PE-conjugated goat anti-rabbit Ig (Beckman Coulter) was used as secondary reagents.

CD40L expression was evaluated using PE-conjugated anti-CD40L mAb (BD Biosciences). IL-21 production was evaluated on hAEC cultured for 4 h at 37°C in the presence of 10 ng/ml Brefeldin A (Sigma-Aldrich), using PE-conjugated anti-IL21 mAb (BioLegend).

Cells were run on a Gallios cytometer and analyzed using Kaluza software (Beckman Coulter). Data were shown as the percentage of positive cells or mean relative of fluorescence intensity (MRFI), calculated as mean fluorescence obtained with specific mAb normalized to mean fluorescence obtained with irrelevant isotype-matched mAb.

Adherent hAEC (3 × 10⁵ cells) maintained at 37°C and 5% CO₂ in 24-well plates (Costar Corning) were exposed to PBS supplemented with 0.1% glucose and in the presence (or absence) of AMP, ATP, or NAD⁺ (50 μM). Supernatants were collected after 5, 30, or 45 min, and acetonitrile (ACN; Sigma-Aldrich) was immediately added to supernatants at a 1:2 ratio to stabilize ADO. Samples were then centrifuged at 12,000 × g, and supernatants were collected and stored at −80°C until use. The presence of ADO, AMP, ATP, and NAD⁺ was investigated by HPLC analysis.

Chromatographic analysis was performed with an HPLC System (Beckman Gold 126/166 NM; Beckman Coulter) equipped with a reverse-phase column (Synergi Fusion C18, 5 μm; 150 × 4.5 mm; Phenomenex). The nucleotides and nucleosides were separated using a mobile-phase buffer (pH 5.1) (0.125 M citric acid and 0.025 M KH₂PO₄) along with 8% ACN over 10 min at a flow rate of 0.8 ml/min. UV absorption spectra were measured at 254 nm. HPLC-grade standards used to calibrate the signals were dissolved in 0.2 μm of sterile-filtered AIM V serum-free medium (Invitrogen, Paisley, U.K.) (pH 7.4) and injected in a buffer volume of 20 μl. The retention times (in min) of standards were as follows: AMP, 2.15; NAD⁺, 2.8; and ADO, 5.56. Peak integration was performed using Gold software (Beckman Coulter). ACN-treated hAEC supernatants were evaporated by Speed-Vac (Eppendorf), reconstituted in mobile-phase buffer, and assayed by HPLC.

The qualitative identity of HPLC peaks was confirmed by comigration of known reference standards. The presence of ADO was also confirmed by spiking standard (50 μM ADO), followed by chromatography. Quantitative measurements were inferred by comparing the peak area of samples with calibration curves for peak areas of each standard compound. Product concentrations were expressed as pmol/3 × 10⁵cells.

hAEC were pretreated with the alkaline phosphatase inhibitor levamisole (2 mM; Sigma-Aldrich). The enzyme immunocytochemical (EICC) staining was then performed by incubating human samples with 2 mM Pb(NO₃)₂ in the absence (blank) or presence of nucleotide substrates. AMP (2 mM) was used to assess CD73 activity, thiamine pyrophosphate (TPP, 4 mM) was used to assess CD203a activity, and 500 μM ATP (all from Sigma-Aldrich) was used to measure CD39 activity. Specific inhibitors of the enzymatic activities have been also added: 100 μM POM-1 (sodium polyoxotungstate, CD39 inhibitor), 300 μM β-γ methyl ATP (CD203a inhibitor), and 100 μM APCP (CD73 inhibitor) (all from Sigma-Aldrich). The activity of the alkaline phosphatases, the nonspecific tissue background, was assessed by adding 8 μM β-glycerophosphate (β-GP) (Sigma-Aldrich) as a substrate, with or without its inhibitor (2 mM Levamisole; Sigma-Aldrich). The assessment of the percentage of EICC-positive cells was determined by two independent pathologists on both sets of slides. Analyses were performed using ImageJ software (https://imagej.net).

Cell proliferation was assessed using CFSE dilution assay. Briefly, CD4⁺ T cells, B cells, or NK cells were stained with 1 μg/ml CFSE (Invitrogen), incubated for 15 min at 37°C, washed, and then cultured in RPMI medium supplemented with 10% FBS. Cells were kept at 37°C and 5% CO₂, alone or in the presence of specific stimuli. T cells were treated with beads coated with anti-CD3/anti-CD28 mAb (T cell activation/expansion kit; Miltenyi Biotec), and B cells were stimulated with 2.5 μg/ml CpG 2006 (TIB Biomol), 100 ng/ml CD40-L, and 20 ng/ml rhIL-4 (both from Immunotools). NK cells were cultured by adding 20 ng/ml rhIL-15 (Immunotools). Stimulated cells were cultured in 96-well flat-bottom plates (Costar Corning) in the presence or absence of hAEC (at hAEC/effector cell ratios ranging from 1:1 to 1:8). The same experiments were performed in the presence or absence of human fibroblasts (Lonza), tested as control. In some experiments, hAEC were treated before culture for 30 min with the following specific inhibitors of ectoenzymes: 10 μM kuromanin (28) (a CD38inhibitor; Sigma-Aldrich), 300 μM α-β methyl ADP (29) (APCP, a CD73 inhibitor; Sigma Aldrich), and 300 μM β-γmethyl ATP (30) (a CD203a/PC-1 inhibitor; Sigma-Aldrich). The optimal concentration of each inhibitor has been already determined in previous studies (22). hAEC were used with effector cells at 1:1 hAEC/effector cell ratio. In another set of experiments, effector cells (T and NK cells) were treated before culture for 30 min with the following specific antagonists of ADOR: 100 nM 8-cyclopentyl-1,3-dipropylxanthine (31) (DPCPX, a specific antagonist of A1 receptor; Sigma-Aldrich), 100 nM ZM 241385 (31, 32) (a specific antagonist of A2a receptor; Sigma-Aldrich), and 1 μM MRS 1706 (32) (a specific antagonist of A2b receptor; Tocris).

Effector cells were harvested after 6 d, washed, and then stained with PE-conjugated anti-CD4, anti-CD19, or anti-CD56 mAbs (Beckman Coulter). After additional washes, cells were run on a Gallios cytometer, and CFSE dilution was analyzed gating on PE⁺ cells, using Kaluza software (Beckman Coulter). Data were expressed as the percentage of proliferating cells.

PB B lymphocytes were cultured for 5 d in RPMI 1640 medium supplemented with 10% FBS in the presence (or absence) of hAEC (1:1 B cells/hAEC ratio). Apoptosis level was measured on B lymphocytes by flow cytometry using PE–Annexin V Apoptosis Detection Kit I (BD Pharmingen), following manufacturer’s protocol. Alive cells were detected as Annexin V⁻/7AAD⁻ cells. Dead cells were identified as Annexin V⁺/7AAD⁺ cells and shown as the percentage of total cells.

B lymphocytes or CD4⁺ T cells were isolated from PB of three different normal donors and cultured in 24-well plates (Costar Corning) at 500,000 cells per well in RPMI supplemented with 10% FBS, in the presence or absence of hAEC (1:1 lymphocytes/hAEC ratio). In some experiments, specific polyclonal stimuli for B or T lymphocytes were added to the culture medium (see above). After 5 d, cells were harvested and washed with staining buffer. B lymphocytes were stained with the following Abs: FITC-conjugated anti-CD19 (Immunotools), PE-conjugated anti-CD24 (Immunotools), and allophycocyanin-conjugated anti-CD38 (clone IB4, conjugated with allophycocyanin). T lymphocytes were stained with Treg Detection Kit (CD4/CD25/CD127; Miltenyi Biotec), following manufacturer’s protocol, and identified as CD4⁺CD25^highCD127^low cells. Regulatory B cells (Breg) were identified as CD19⁺CD24^highCD38^high cells. Cells were then washed, resuspended in staining buffer, and then subjected to flow cytometric analysis using Gallios cytometer and Kaluza software. Results are expressed as the percentage of positive cells.

Statistical analysis was performed using Prism 5.03 software (GraphPad Software). Gaussian distribution of data was analyzed using the Kolmogorov–Smirnov test. The Student t test or Mann–Whitney U test was used to compare data, depending on data distribution. The significance ranges as follows: *p < 0.05 (significant), **p < 0.005, and ***p < <0.0005.

The presence of different ectoenzymes on the surface of hAEC was measured by flow cytometry on cells obtained from 17 different donors. All the samples showed high levels of CD38, CD203a/PC-1, CD39, CD73, and CD157 (Fig. 1A). Inset shows a representative FACS analysis. Data are summarized in Supplemental Table I. All molecules were expressed by the whole cell population, with the exception of CD38, expressed only by a fraction of hAEC (percentage of positive cells ± SD: 47.89 ± 7.9). No difference was observed between freshly isolated and cryopreserved hAEC (data not shown). All the analyzed primary amnion-derived cells are equipped with a complete ectoenzymatic machinery, able to produce ADO through the canonical as well as the alternative pathway.

Because hAEC are cultured in the presence of rhEGF, we asked whether these molecules could alter the expression of the adenosinergic ectoenzymes. To address this issue, we performed an independent experiment on four different batches of hAEC, cultured for 3 d in the presence (or absence) of rhEGF. As shown in Fig. 1B, rhEGF significantly upregulates the expression of CD73, whereas the expression of CD38, CD39, and CD203a/PC-1 was not affected. Inset shows a representative staining for CD73 expression on hAEC treated or not treated with rhEGF.

One question was whether these cells are simply a producer or if they may also intervene in ADO metabolism. Thus, we analyzed the expression of ADOR and ADA on hAEC. A constitutive expression of ADOR, ADORA1, ADORA2a, and ADORA2b was detected on hAEC. Moreover, hAEC express surface CD26, the carrier of ADA that metabolizes ADO to inosine (Fig. 1C). Inset shows a representative experiment. Data are summarized in Supplemental Table I.

The expression of ADOR and CD26 was also analyzed on resting and activated immune effector cells. Data obtained were consistent with previous reports (33, 34) and are summarized in Supplemental Table I and in Supplemental Fig. 1. Indeed, resting T lymphocytes expressed low levels of all ADOR and CD26. The expression of all ADOR was significantly upregulated by activated lymphocytes, whereas the expression of CD26 was not influenced by T cell activation. Resting B lymphocytes commonly express low levels of ADORA1, ADORA2a, and CD26, whereas ADORA2b is highly expressed. This pattern of expression was not significantly affected by B cell stimulation.

Resting NK cells expressed high levels of ADOR and CD26. At variance with T lymphocytes, such expression was significantly downregulated in NK cells after activation. Again, CD26 expression was not significantly affected by activation.

For functional analysis of the distinct ectoenzymes, hAEC were challenged with the substrates specific for each ectoenzyme by testing the production of the final substrate. Data are summarized in Table I. Exposure of hAEC to AMP (substrate of CD73) was followed by the production of ADO. AMP conversion is a time-dependent process: ADO production is detectable within 5 min and reaches two peaks after 30 and 45 min. The conversion of AMP into ADO resulted in a significant reduction of the initial AMP concentration, both at 30- and 45-min incubation (Fig. 2A). To a lesser extent, hAEC can also produce ADO from ATP and NAD⁺ in a time-dependent manner. As shown in Fig. 2B, ADO produced by ATP conversion was detectable after 5 min and significantly increased after 30 and 45 min. Accordingly, the initial ATP concentration was reduced after 30 and 45 min. Similarly, ADO produced by NAD⁺ conversion was detected after 5 min and significantly increased after 30 and 45 min. Again, the initial concentration of NAD⁺ decreased after 30 and 45 min (Fig. 2C).

The roles of CD39 and CD203a/PC-1 in the production of ADO were assessed by performing experiments similar to those described above, in the presence of CD39 inhibitor POM-1. ADO production was detected after 15 min. As expected, POM-1 did not affect ADO production from AMP (mediated by CD73) (Fig. 2A). In contrast, the production of ADO from ATP (mediated by CD39 or CD203a/PC-1 in cooperation with CD73) decreases after exposure to POM-1 (Fig. 2B). These findings suggested that both CD39 and CD203a/PC-1 are involved in the process. Lastly, POM-1 treatment does not influence ADO production starting from NAD⁺ (Fig. 2C), being the processes mediated by CD38 and CD73 in sequence. All these data are summarized in Table I.

The ability of hAEC to produce ADO from different substrates was further confirmed in situ by EICC. Cells were counterstained with H&E (Fig. 3A1, 3A2) to confirm epithelial morphology. ADO production was detected on all hAEC samples after exposure to substrates specific for each ectonucleotidase activity: ATP for CD39 (Fig. 3A3), AMP for CD73 (Fig. 3A5), and TPP for CD203a/PC-1 (Fig. 3A7). Positive controls (endogenous phosphatases) were assessed using β-GP substrate (Fig. 3A9), whereas negative controls were obtained using the same substrate in the presence of levamisole (Fig. 3A10). ADO production was reduced in the presence of specific inhibitors: POM-1 (CD39 inhibitor, Fig. 3A4); APCP (CD73 inhibitor, Fig. 3A6), and β-γ-methyl ATP (CD203a/PC-1 inhibitor, Fig. 3A8). The analysis of the reactivities obtained in the EICC tests showed a significant decrease of the overall positive cells following incubation with specific inhibitors (Fig. 3B). Namely, ATP positivity was reduced by ∼80% after POM-1 incubation. AMP positivity was reduced by ∼85% following APCP incubation. TPP positivity was fully abolished following pretreatment with β-γ-meATP. Tetramisole incubation reduced the reactivity by ∼65% (Fig. 3C).

The functional relevance of different ectoenzymes on the surface of hAEC to their immunomodulatory activity was tested on activated human T lymphocytes. T cell proliferation assay was performed in the presence or absence of hAEC at different E:T cell ratios.

CD4⁺ T cells were stimulated with beads coated with anti-CD3 and anti-CD28 mAbs. As shown in Fig. 4A, the proliferation of CD4⁺ T lymphocytes was significantly reduced in the presence of hAEC at 1:1, 1:2, 1:4, and 1:8 hAEC/CD4 ratios. Inset shows a representative experiment. Additional experiments were performed using hAEC and human fibroblasts, tested as negative control. As shown in Supplemental Fig. 2A, T cell proliferation was inhibited in the presence of hAEC but not in the presence of human fibroblasts, at 1:1, 1:2, 1:4, and 1:8 ratios. These data proved unambiguously that hAEC-mediated inhibition of T cell proliferation was not due to a competition for nutrients and represented a specific feature of hAEC.

Next, we performed additional experiments in the presence of specific inhibitors of adenosinergic ectoenzymes. T cell proliferation was significantly reduced in the presence of hAEC, at a 1:1 ratio, and the inhibition was partially reverted in the presence of APCP, β-γ methyl ATP, and POM-1. Addition of kuromanin (CD38 inhibitor) did not significantly rescue the functional effects (Fig. 4B). Inset shows a representative experiment.

Because rhEGF upregulated CD73 expression, we investigated whether the same treatment could influence the hAEC-mediated inhibition of T cell proliferation. To this aim, independent experiments were performed using four different hAEC samples exposed (or not) to rhEGF. As shown in Fig. 4C, T cell proliferation (percentage of proliferating cells ± SD: 97 ± 1.2) was significantly inhibited by untreated hAEC at 1:1, 1:2, 1:4, and 1:8 hAEC/CD4 ratios. This inhibition was significantly increased using hAEC cultured in the presence of rhEGF (as compared with untreated hAEC), at 1:1 and 1:2 but not at 1:4 and 1:8 hAEC/CD4 ratios. Inset shows a representative experiment. All these results are summarized in Supplemental Table II.

To dissect the contribution of each ADOR to hAEC-mediated inhibition of T cell proliferation, we performed additional experiments using specific antagonists of A1 (ZM241385), A2a (MRS1706), or A2b (DPCPX) ADOR. As shown in Fig. 4D, T cell proliferation was significantly inhibited by hAEC at a 1:1 ratio. This inhibition was significantly reverted by pretreating T cells with specific antagonists of A1 and A2b (but not of A2a) ADOR. Thus, ADO produced by hAEC inhibits T cell proliferation mainly through A1 and A2b ADOR.

B cells were stimulated using CpG 2006, CD40L, and IL-4 in the presence (or absence) of hAEC. As shown in Fig. 5A, B cell proliferation was increased in the presence of hAEC, added at 1:1, 1:2, 1:4, and 1:8 hAEC/B cell ratios. The effects were further increased in the presence of APCP and β-γ methyl ATP. Addition of POM-1 resulted in a partial reversion of the hAEC effect. Inset shows a representative experiment. Results are summarized in Supplemental Table II.

NK cells were stimulated with IL-15 in the presence (or absence) of hAEC. As shown in Fig. 5B, NK cell proliferation was significantly reduced in the presence of hAEC at a 1:1 ratio. The effects were even more apparent at 1:2, 1:4, and 1:8 hAEC/NK cell ratios. Because a linear decrease of NK cell proliferation was not observed, we have performed additional experiments at lower ratios (1:8 to 1:64 hAEC/NK cell ratios, Supplemental Fig. 2B).

Lastly, NK cell proliferation was further decreased in the presence of APCP and POM-1, whereas β-γ methyl ATP did not induce statistically significant effects. Inset shows a representative experiment. Results are summarized in Supplemental Table II.

Additional experiments were performed using specific antagonists of ADOR. As shown in Fig. 5C, NK cell proliferation was significantly inhibited by hAEC at a 1:1 ratio. NK cell proliferation was partially but significantly restored by pretreating these cells with specific inhibitors of A1 and A2A ADOR but not using a specific inhibitor of A2B ADOR. These data suggested that NK cell proliferation was dampened by ADO produced by hAEC, mainly through the interaction with A1 and A2A ADOR.

Because B cell proliferation was increased in the presence of hAEC, we tested whether hAEC may rescue unstimulated B cells from apoptosis through the expression of activating receptors (e.g., CD40L) or secretion of cytokines (e.g., IL-21), both crucial for signaling in B lymphocytes. Data are summarized in Table II.

As shown in Fig. 6A, the percentage of viable resting B cells was significantly increased in the presence of hAEC as compared with the same cultures in the absence of hAEC. Consequently, the percentage of dead cells was significantly reduced in the presence of hAEC. Inset shows a representative experiment.

IL-21 production and CD40L expression were tested on six different batches of cryopreserved hAEC. The results obtained indicate that most hAEC produce IL-21 (percentage of positive cells ± SD: 75 ± 10), whereas only a small portion of these cells express CD40L (6.0 ± 3.1%) (Fig. 6B). Inset shows a representative staining with anti-IL-21 mAb performed on six different batches of hAEC.

The picture of the influence of hAEC on immune cells was completed by assessing whether hAEC are able to induce the expansion of B and T regulatory cells in vitro. Data are summarized in Table II.

The percentage of CD4⁺CD25^highCD127^low regulatory T cells (Treg) was significantly increased in the presence of hAEC. The percentage of Treg was significantly upregulated in the presence of hAEC, also when T lymphocytes were activated with polyclonal stimuli. Similar results were obtained with CD19⁺CD24^highCD38^high Breg. The percentage of Breg was significantly increased in the presence of hAEC, starting from samples of PB unstimulated or activated B lymphocytes (Fig. 6C). A representative experiment performed on resting T and B lymphocytes is shown in Fig. 6D.

Human amnion epithelial cells are a relatively new source of multipotent cells, and a phenotypic map of the molecules expressed by these cells is not completed. However, different groups highlighted the presence of immunomodulatory molecules, among which are noncanonical HLA class I molecules (12) or 5′-ectonucleotidase enzyme (CD73), commonly referred to as characteristic of mesenchymal stromal cells (35).

The experience acquired from studying tolerogenic environments, such as placenta, prompts investigations of the components of the adenosinergic pathways (21, 22). ADO, a potent agent involved in inflammation and immunosuppression, is generated in the extracellular milieu using ATP, metabolized by CD39 to AMP and then to ADO by CD73 (36). CD39 and CD73 have been previously identified on some tumor cells but are also expressed by regulatory cell subsets and immune effectors, such as T, B, and NK cells (37). The pyridine nucleotide NAD⁺ may also lead to ADO production through the concerted action of CD38, whose product ADPR is metabolized by CD203a/PC-1 to AMP (21), later degraded by the bottleneck ectoenzyme CD73. CD38/CD203a/CD73 is an alternative pathway, confirmed to be present in human placenta and cornea (38, 39) and in pathological environments, such as the one of myeloma (40).

Full-term placenta provides cells characterized by a homogenic coexpression of adenosinergic ectoenzymes. We have been able to confirm that hAEC are endowed with ectoenzymes of the canonical (CD39, CD73) as well as of the alternative pathway (CD38, CD203a/PC-1, CD73). Both pathways are functional and have a role in the immune modulation mediated by hAEC. The specific effect on all identified ectoenzymes was assessed using specific inhibitory compounds, as previously tested and titrated (28, 30, 41). Another original observation stands on the presence of ADOR on hAEC, suggesting a circuit of production and use of ADO in these cells. ATP and NAD⁺ were released into the extracellular space, and ADO thereby generated play roles in the control of proliferation and apoptosis of different sets of immune cells. Indeed, ADO has been shown to inhibit dendritic, T, and NK cells, and to promote the generation of Treg and M2 macrophages in vitro (23, 42). Different studies addressed an important role of ADO in the release of immunomodulatory molecules by macrophages (43–45).

Results presented in this article support the hypothesis that ADO generated extracellularly by hAEC is involved in the inhibition of T cell proliferation. Indeed, a selective inhibition of molecules of both adenosinergic pathways and blockade of A1 and A2B ADOR using selective antagonists led to a partial recovery in T cell proliferation. Because ADOR antagonists failed to completely reverse the effects of extracellular ADO, we may hypothesize the existence of an intracellular uptake of ADO. Indeed, extracellular ADO homeostasis is maintained by a bidirectional transport through equilibrative nucleoside transporters located in the plasma membrane in the direction of concentration gradients. ADO entering cells through equilibrative nucleoside transporters is phosphorylated by a low-Km ADO kinase into AMP, which may accumulate inside the cell and activate specific AMP-dependent protein kinase (46). In addition, extracellular ADO also mediates signals through membrane-anchored G protein–coupled ADOR (47). Finally, the partial block of ADOR may increase the concentration of extracellular AMP, favoring the mononucleotide internalization either through a unknown receptor or by gradient concentration, inducing a positive effect on the AMPK/mTOR/p70S6K/rpS6 protein axis. Consequently, it is possible to observe an increased proliferation and thus a partial reversal of the inhibition of T cell proliferation. Thus, the effects observed on T lymphocytes are at least in part mediated by both pathways of ADO production, but other immunosuppressive molecules, such as HLA-G, likely play additional roles (12) (F. Morandi, S. Strom, and R. Gramignoli, manuscript in preparation).

In this study, we demonstrated that hAEC promote an increase of Treg number in vitro, and such an effect might be secondary, at least in part, to the production of ADO from hAEC. Indeed, it has been demonstrated that ADO may promote an expansion of Treg after engagement of ADORA2a (48). Our data suggest that hAEC-derived ADO may be involved in the reduction of T cell responses in vivo, as observed in murine models of experimental autoimmune encephalomyelitis after treatment with hAEC (15). Indeed, hAEC induced a decrease of IL-17A and an increase of the number of Treg in blood and Th2 cells in peripheral lymphoid organs (15).

Notably, we measured a different effect on human NK cells. NK cell proliferation was slightly influenced by hAEC presence and mostly at low NK/hAEC ratios (1:4 to 1:64). Because NK cells express ADOR at a high level, it is likely that a large number of hAEC is necessary to quickly saturate all the ADOR, particularly A1 and A2A receptors that are mainly involved in such inhibition (as witnessed by experiments performed using selective antagonists). Moreover, NK cells express high levels of the CD26/ADA complex. ADA allows the conversion of ADO into inosine, neutralizing the ADO inhibitory effects.

The results obtained after coculturing hAEC with B lymphocytes delineate another scenario. In the presence of polyclonal stimuli, hAEC significantly increased B cell proliferation, which is further increased after a selective inhibition of the CD73 and CD203a/PC-1 ectoenzymes. B cell proliferation, per se, remains unaltered in the presence of hAEC (data not shown). These observations suggest that ADO produced by hAEC may counterbalance B cell activation by acting on ADORA1, which guides cell differentiation, survival, and proliferation by decreasing cAMP (49). B cell activation and survival can be supported by other molecules expressed or released by hAEC. Out of these, NAD⁺ and ATP permeate activated hemichannels and mediate paracrine effects in cells by increasing the intracellular Ca²⁺ levels (50). Furthermore, extracellular nucleotide concentrations can potentially favor (or suppress) the local immune responses, depending on their concentration as well as the relative abundance of ectonucleotidases and P2 receptor subtypes on immune cells (51, 52). Such biphasic effects of extracellular nucleotides could be partially explained in light of the peculiar impact of inhibitors of ectonucleotidases on the increased B cell proliferation observed after cocultures with hAEC. B cells express different levels of ADOR in comparison with T lymphocytes (prevalently, ADORA2b). Differently from T lymphocytes, the expression of ADOR was not increased upon activation of B cells, suggesting a different effect of hAEC-derived ADO. Indeed, the lack of CD26 expression (a molecule witnessing the function of ADA) may lead to an increase of pericellular ADO and, consequently, increased effects on B cells. Another original observation is that cocultivation of hAEC with resting peripheral B lymphocytes is followed by a rescue from apoptosis of the latter cells. CD40L is normally expressed by activated T lymphocytes, and it is effective in overcoming apoptosis induced by cross-linking of Ag receptor (53). CD40L is detectable on follicular helper T (54) and dendritic cells (55) and has been reported supporting B cell maturation in the germinal center. However, this ligand is expressed in <4% of hAEC. Another molecule involved in the control of B cell differentiation is IL-21, which can modulate proliferation, differentiation, and apoptosis in B lymphocytes, depending on costimulatory signals (56). Most of the hAEC analyzed express intracellular IL-21, suggesting that they are able to support the in vitro survival of resting B lymphocytes as well as B cell proliferation in the presence of polyclonal stimuli. These data should be taken into account in hAEC-based treatments of autoimmune diseases related to B cell activation because preclinical experiments in animal models were successful (13–15).

Finally, in this study we also demonstrated an expansion of Bregs driven by hAEC. The effective role of ADO in such context is still to be elucidated, taking into account that ADO induces the expansion of CD39^highCD73⁺ Bregs through ADORA1/A2a (57). A reasonable inference is that ADO may induce the expansion of CD19⁺CD24^highCD38^high Bregs as well.

In conclusion, hAEC constitutively express a complete set of cell surface nucleotidases, previously scattered on different cells, and were constitutively identified for the first time, to our knowledge, on a single cell type. The products of such ectoenzymes modulate important functions of immune effector cells. Such effects of hAEC in vitro on immune effector cells are summarized in Fig. 7. These conclusions are relevant in regenerative medicine, in which the adoption of therapeutic protocols based on hAEC infusion in immunocompetent recipients, as well as in patients with immune-related disorders, may constitute an effective treatment. Large numbers of hAEC can be isolated by full-term placenta, readily banked, and potentially available “off the shelf” in medical centers worldwide, similarly to what happens for bone marrow cells. For such reasons, the Karolinska Institute is in the process of creating the first GMP bank of human amnion-derived cells. Regulatory approvals, a preliminary step to start the transfer to patients with liver and other diseases, are ongoing.

F. Morandi was the recipient of a fellowship from Fondazione Umberto Veronesi.

The authors have no financial conflicts of interest.