Cutting Edge: Dysregulated Endocannabinoid-Rheostat for Plasmacytoid Dendritic Cell Activation in a Systemic Lupus Endophenotype

Systemic autoimmunity in systemic lupus erythematosus (SLE) is characterized by humoral autoreactivity against nuclear Ags and immune-complex deposition leading to pathologies involving multiple organs, namely skin, joints, vessels, CNS, and kidneys (1). Due to complex genetic factors associated with the disease and a dominant systemic type I IFN response critically linked to the pathogenesis, SLE is a prototypical member of polygenic IFNopathies (2). Type I IFNs produced by the plasmacytoid dendritic cells (pDCs) have been found to be a crucial event in SLE pathogenesis (3, 4). Genome-wide association studies have linked multiple genomic loci with SLE susceptibility, although direct pathogenetic role of very few of them have been reported, apart from the genes directly linked to induction and response of type I IFNs (5, 6). A recent genome-wide association study had reported association of a novel gene, α/β-hydrolase domain-containing 6 (ABHD6), with SLE susceptibility (7).

ABHD6 is a serine hydrolase that catalyzes hydrolysis of endocannabinoids, the endogenous ligands for the G-protein–coupled cannabinoid receptors (CB1 and CB2), which include 2-arachidonyl glycerol (2-AG) and anandamide (AEA) (8). The role of ABHD6, as the major hydrolytic enzyme for 2-AG, was first documented in microglia cells in the CNS (9). Expression of ABHD6 with a similar function in macrophages was also reported (10). In macrophages, ABHD6-mediated hydrolysis of 2-AG prevented 2-AG–driven inhibition of TLR activation; thus, ABHD6 activity was found to have a proinflammatory role. Systemic immunomodulatory role of cannabinoid receptor signaling is widely reported, and a link with autoreactive inflammation is also suggested (11). Apart from modulation of endocannabinoids, the substrate of ABHD6, function of this enzyme is also reported in various other physiological contexts, namely receptor recycling in the CNS (12), insulin secretion by pancreatic β cells (13), and lipolysis in metabolic tissues (14). Accordingly, deregulation of this enzyme has been implicated in various clinical contexts of CNS disorders and metabolic disorders (13–15), but any pathogenetic role of ABHD6 in immunological disorders remains unidentified. In this study, we aimed at exploring functional link of this gene with type I IFN induction and SLE pathogenesis.

Healthy individuals (n = 76) and SLE patients (n = 90) were recruited for the study; patients were recruited at the Department of Rheumatology at the Institute of Postgraduate Medical Education and Research, Kolkata, India (Supplemental Fig. 1A, 1B). Peripheral blood samples were collected from SLE patients and healthy donors, after taking written informed consent, in accordance with the Declaration of Helsinki and as recommended and approved by the institutional review boards of both the institutes (Institute of Postgraduate Medical Education and Research and Council of Scientific and Industrial Research–Indian Institute of Chemical Biology, Kolkata, India).

PBMCs were isolated from whole blood. PBMCs were either kept in TRIzol reagent for subsequent RNA isolation or cultured overnight in RPMI 1640 with 10% FBS in a 96-well plate at a density of 0.2 million cells per well and were treated with agonists and/or inhibitors as indicated. Collected plasma was stored at −80°C for future use. In some cases, pDCs were isolated from PBMCs by magnetic immunoselection using BDCA-4 MicroBeads (Miltenyi Biotec) and cultured overnight in RPMI 1640 with 10% FBS in 96-well plates at a density of 40,000 cells per well. pDCs were stimulated with 150 nM of CpGA (for IFN-α induction) or 500 nM of CpGB (for TNF-α induction), in the presence of the 2-AG (10 μM), WWL70 (25 μM), or other inhibitors as indicated in figure legends. Cell supernatants were collected 18 h poststimulation.

CB2 gene was knocked down in freshly isolated pDCs by nuclear transfection following the manufacturer’s protocol (Amaxa Lonza 4D nucleofector kit; Lonza) using CB2 small interfering RNA (siRNA) or eGFR siRNA as control. After 24 h, cells were plated in 96-well plate and treated as indicated. Cells and supernatant were harvested 18 h poststimulation.

Total RNA was isolated from cells using TRIzol reagent (Invitrogen). Quantitative PCR was carried out for the specified genes normalized against 18S rRNA as the housekeeping gene.

A total of 100 μl of plasma sample or cell supernatant was mixed with 300 μl of liquid chromatography–mass spectrometry –grade acetonitrile + 0.1% formic acid (1:4 dilution) and incubated on ice for 15 min with intermittent vortexing, followed by centrifugation at 14,000 rpm for 10 min at 4°C. The supernatant was extracted in a fresh tube and used for subsequent mass spectrometry of 2-AG on LTQ Orbitrap XL. The output was analyzed by Thermo Xcalibur software.

ELISA was done to quantify IFN-α and TNF-α from PBMC and pDC culture supernatants, using anti-human IFN-α ELISA kit and anti-human TNF-α ELISA kit (Mabtech).

Statistical analyses of all data sets were done on GraphPad Prism 5.0 software. Data were compared between groups using paired, unpaired Student t test or Mann–Whitney U test as specified in figure legends, and correlations were done using Spearman rank correlation as specified in respective figure legends.

An overexpression of ABHD6 in SLE patients was suggested in the previous study, which first identified polymorphisms of ABHD6 as the PXK-locus associated predisposition for SLE (7). To examine this, we recruited a cohort of SLE patients and checked the expression of ABHD6 in PBMCs and found that there is significant overexpression of ABHD6 in SLE patients (n = 90, female = 86, male = 4), as compared with healthy controls (n = 76, female = 32, male = 44) (Fig. 1A). A recent report suggested that female hormones can increase the expression of ABHD6 when added to in vitro cultured leukocytes (16); however, there was no difference in ABHD6 expression in PBMCs between female and male healthy controls in our cohort (Supplemental Fig. 1C). The data also revealed two subgroups of SLE patients based on expression of ABHD6—taking median expression in healthy controls as the threshold, we categorized SLE patients into ABHD6^High and ABHD6^Low groups (Fig. 1A). Of note here, ABHD6 expression did not show any correlation with the disease activity score SLE Disease Activity Index (SLEDAI) in our patient cohort (Supplemental Fig. 1D) and there was no significant difference between the ABHD6^High and ABHD6^Low groups in terms of either SLEDAI or Damage Index scores (Supplemental Fig. 1E, 1F).

Enriched expression of IFN signature gene (ISG) in the peripheral blood is a surrogate marker for all IFNopathies which closely correlate with disease activity in SLE (17, 18). In our cohort, the peripheral blood ISG was not correlated with disease activity (viz. SLEDAI) (Supplemental Fig. 1G), perhaps because the SLE patients in our cohort were not treatment naive. But average of relative expression of six major ISGs (viz. IRF7, TRIP14, MX1, XIAPAF1, ISG15, and IFI44L) was significantly correlated with ABHD6 expression (Fig. 1B), and the ABHD6^High group was found to have significantly higher value compared with the ABHD6^Low group (Fig. 1C). We also found that in the ABHD6^High group of SLE patients, ABHD6 expression in PBMCs strongly correlated with average expression of those six ISGs, whereas in the ABHD6^Low group this was not seen (Fig. 1D, 1E). This led us to hypothesize that overexpression of ABHD6 in the ABHD6^High SLE patients may play a regulatory role on systemic type I IFN induction, leading to ISG enrichment in peripheral blood.

To validate that ABHD6 may have a regulatory role on systemic type I IFN induction, we explored if WWL70, a selective ABHD6 inhibitor (19), can affect IFN-α induction from healthy PBMCs in response to the bona fide TLR9 ligand CpGA oligonucleotides. Indeed, WWL70 interfered with CpGA-induced IFN-α induction in PBMCs in a dose-dependent manner (Fig. 2A). TLR9 expression in humans is largely restricted to pDCs and B cells (20), and pDCs are the major type I IFN producers among the peripheral blood immune cells (21, 22). Accordingly, TLR9 ligand–induced IFN-α from PBMCs is mostly due to TLR9 activation in pDCs (data not shown). On analysis of a public database on human tissue specific transcriptome, we found that among the hematopoietic cells represented in that database, dendritic cells had the highest expression of ABHD6 (Fig. 2B). Also, when we assessed ABHD6 expression in isolated pDCs and compared it with the same in corresponding total PBMCs, it was evident that ABHD6 expression is enriched in pDCs (Fig. 2C). The other serine hydrolases that have similar activity as that of ABHD6 (viz. FAAH, MGLL, and ABHD12) (23) had a much lower expression in pDCs (Fig. 2D).

This made us explore whether WWL70 can also abrogate IFN-α induction in purified pDCs ex vivo stimulated with CpGA. Interestingly, there was no such inhibition documented at the same concentration of WWL70 that inhibits IFN-α induction from PBMCs (Fig. 3A). We reasoned that these apparently counterintuitive data were due to lack of direct regulatory effect of ABHD6 on this pathway, and the inhibition of IFN-α induction in response to TLR9 activation was, rather, driven by some ABHD6 substrate, which is available in PBMC culture but not in purified pDC culture—ABHD6 regulates the pathway through affecting the relative abundance of its substrate. The endocannabinoid 2-AG is the established endogenous substrate for ABHD6 (23, 24), so we repeated the same experiment in presence of exogenous 2-AG, considering the possibility that pDC-intrinsic sourcing of 2-AG may not be sufficient. Indeed, in the presence of exogenous 2-AG, ABHD6 inhibition by WWL70 abrogated IFN-α induction from purified pDCs stimulated with CpGA (Fig. 3B). Thus, we concluded that ABHD6, expressed in pDCs, regulates the abundance of the endocannabinoid 2-AG, which in turn relieves a steady-state resistive effect of 2-AG on TLR activation in pDCs and IFN-α induction. It is established that TLR activation in human pDCs also leads to induction of proinflammatory cytokines other than IFN-α (e.g., TNF-α). But induction of these other cytokines have a different signaling regulation downstream of the endosomal TLRs. Activation and nuclear translocation of IRF7 leads to IFN-α induction, whereas activation of NF-κB lead to induction of cytokines like TNF-α (25, 26). We found that the 2-AG–driven inhibition did not affect TNF-α induction in pDCs (Fig. 3C); thus 2-AG has a specific resistive influence on IFN-α induction from pDCs. We also explored if IFN-α has any influence on expression of ABHD6 in human pDCs as well, which revealed no such regulation (Supplemental Fig. 1H).

As ABHD6 can regulate 2-AG abundance and in SLE patients there is an overexpression of ABHD6, we explored if plasma 2-AG abundance was also different when SLE patients are compared with healthy controls. Plasma from both groups of patients were processed for estimation of 2-AG using liquid chromatography–tandem mass spectrometry (Supplemental Fig. 2). But we found there was no significant difference in plasma 2-AG levels between healthy controls and SLE patients (Fig. 4A). This apparent discrepancy may be explained by the fact that plasma abundance of 2-AG is not regulated solely by its catabolism driven by serine hydrolases like ABHD6 expressed on immune cells—this is also influenced by dynamic pathways leading to 2-AG synthesis, not only in circulating immune cells but also in other tissues in the body. Interestingly, we found that in the ABHD6^High SLE patients there was a notable, although not quite statistically significant, negative correlation between ABHD6 expression in PBMCs and plasma 2-AG levels (Fig. 4B), whereas in the ABHD6^Low patients no such relationship was evident (Fig. 4C). We concluded that overexpression of ABHD6 on immune cells has the potential of regulating local, possibly even systemic, levels of 2-AG in the ABHD6^High SLE patients.

For immune cells, like pDCs, that express ABHD6, this local control of 2-AG abundance is of great relevance as that can affect the cellular functions. To validate if ABHD6, expressed on pDCs, can indeed influence local 2-AG abundance, we added exogenous 2-AG to purified pDC cultures ex vivo and measured residual 2-AG in the culture supernatant after 4 h, in absence and presence of WWL70, the ABHD6 inhibitor. We found a significant increase in 2-AG retention in purified pDC cultures when ABHD6 is inhibited (Fig. 4D), indicating that pDC-intrinsic ABHD6 activity does regulate local 2-AG abundance and thus disrupts its resistive tuning of TLR activation and type I IFN induction.

We also found that in purified cultures with exogenous 2-AG, pDCs from SLE patients (n = 21) depleted significantly more 2-AG in 4 h, as compared with healthy donors (n = 12) (Fig. 4E). We also compared ABHD6 expression in purified pDCs from SLE patients (n = 12) and healthy donors (n = 11) and found significant overexpression of ABHD6 as well as a similar subgrouping of patients between ABHD6^High and ABHD6^Low groups (Fig. 4F), just like we found with ABHD6 expression in total PBMCs (Fig. 1A). In fact, the extent of 2-AG hydrolysis in purified pDC cultures was found to be strongly correlated with ABHD6 expression levels in the pDCs in the ABHD6^High group (Fig. 4G). This led us to conclude that ABHD6 expression in pDCs can indeed control local abundance of the endocannabinoid 2-AG, thus affecting the 2-AG–driven resistive influence on IFN-α induction in response to TLR9 activation.

After we identified that local presence of 2-AG can inhibit IFN-α induction in pDCs on TLR9 activation, we wanted to explore the molecular interface responsible for this resistive effect of 2-AG on pDC activation. 2-AG has been previously reported to affect cellular functions in different cell types through a myriad of receptors (Fig. 5A), for example, through engagement of cycloxygenase-2 (COX2) in myeloid cells by the 2-AG hydrolysis product arachidonic acid (10), through peroxisome proliferator-activated receptor α (PPARα) in mesenchymal stem cells (27), and the specific cannabinoid receptors CB1 and CB2 (28). We found that the 2-AG–driven resistive effect on pDC activation could not be reversed by the COX2-specific inhibitor valdecoxib (Fig. 5B) or the PPARα-specific inhibitor MK866 (Fig. 5C), so we hypothesized that resistive influence of 2-AG in pDCs may be mediated by the cannabinoid receptors. When we checked for expression of the two cannabinoid receptors in purified human pDCs, we found that CB2 is the more abundant cannabinoid receptor in human pDCs (Fig. 5D). Accordingly, the CB1-specific small molecule antagonist SLV319 could not reverse the 2-AG–driven inhibition of pDC activation (Fig. 5E), whereas the CB2-specific antagonist SR144528 could significantly reverse this effect of 2-AG (Fig. 5F). To gather more evidence in support of critical involvement of CB2 receptors in mediating the effect of 2-AG on pDCs, we targeted the CB2 gene in pDCs. We found that siRNA-mediated knockdown of CB2 in human pDCs could again significantly reverse the resistive effect of 2-AG on IFN-α induction (Fig. 5G).

Thus, we identified a hitherto unknown rheostat mechanism that controls IFN-α induction from human pDCs in response to TLR activation, which is driven by the endocannabinoid 2-AG working on pDCs through CB2 receptors (Fig. 5H). Cellular abundance of the 2-AG–catabolizing enzyme ABHD6 disrupts this rheostat mechanism, leading to supraphysiologic IFN-α induction from human pDCs in response to TLR activation. This pathway contributes to the systemic type I IFN induction associated with an SLE endophenotype characterized by higher expression of ABHD6 on the circulating immune cells, specifically pDCs. This study also establishes hitherto unexplored promising therapeutic targets, namely ABHD6 and CB2, for SLE treatment.

The authors have no financial conflicts of interest.