Specific induction of the unique GPR15 expression in heterogeneous blood lymphocytes by tobacco smoking

Purpose: In the peripheral blood, it has been shown that smoking is, to date, the only specific condition leading to an increase in GPR15+ T cells. We therefore aimed to characterize GPR15-expressing blood T cells in more detail. Materials and Methods: The whole transcriptome by RNAseq as a proxy for protein expression was analyzed in GPR15+ and GPR15-T cells. A deep immuno-phenotyping was conducted for identification of T cell subtypes. Results: The expression of GPR15 seemed to be unique, not concomitantly accompanied with expression of another protein. According to different T cell subtypes there is no single cell type prominently represented in GPR15+ T cells. The individually different proportions of GPR15+ cells among each GPR15-expressing T cell subtypes in blood were strongly associated with chronic smoking. Indeed, the frequency of GPR15+ T cell subtypes can be effectively used as a highly convincing biomarker for tobacco smoking. Conclusions: While the chronic smoking-induced enrichment of GPR15+ T cells in blood might indicate a systemic inflammation, by the widespread presence in different T cell subtypes, GPR15 could feature a general impact on maintaining the systemic homeostasis to putatively prevent harm from smoking.


Introduction
The recent identification of the chemoattractant ligand GPR15L for the G protein-linked receptor GPR15-expressing T cells suggested an interfering role of GPR15-GPR15L interaction in mucosal and cutaneous inflammation (Ocon et al. 2017). The GPR15L, encoded by C10orf99, was found to be an upregulated marker gene for psoriatic pathogenesis in skin (Guo et al. 2014). However, in colon the constitutively expressed GPR15L was only modestly impaired be inflammation or the presence of microbiota.
In human lymphocytes the GPR15 receptor serves as a chemokine receptor for the immunodeficiency virus type 1 and 2 (Deng et al. 1997). In mice, it mediates the homing of lymphocytes to the sites of inflammation in the large intestine (Kim et al. 2013) and of fetal dendritic epidermal T cells (DETCs), overexpressing GPR15, from the thymus to the skin where the GPR15 expression is rapidly downregulated on skin resident DETCs (Lahl et al. 2014). Although the lymphocyte recruitment maintains intestinal immune homeostasis, it also contributes to inflammation (Nguyen et al. 2015). With the presence of GPR15+ regulatory T cells in mice colitis and their absence in human colitis, indeed, there are serious differences between human and mice (Fischer et al. 2015). This hampers the extrapolation of findings from mice to human. Nevertheless, while GPR15+ T cells are enriched in inflamed tissue, such as colon, their frequency in blood remains unchanged.
Similarly, gene expression of GPR15 was not changed in blood of subjects with psoriatric skin compared to healthy subjects (Coda et al. 2012). In contrast to above mentioned chronic systemic inflammatory disorders, chronic smoking is to date the only strong inducer of GPR15+ lymphoid cells in blood . The adaptive immunological process of GPR15 activation in lymphoid cells has not yet been resolved.
Their excess in blood was not evoked by a suggested smoking-induced disturbed homeostasis of the lung (Bauer et al. 2017). Smoking-induced expression of the GPR15 gene in blood was recently assumed to indicate its potential role in chronic inflammatory pathologies (Koks et al. 2015). Because of both the strong association between epigenetic hypomethylation at CpG site cg19859270 (annotated to GPR15) with cellular GPR15 protein expression  and linear correlation between duration and degree of smoking and time since cessation with degree of hypomethylation at cg19859270 A c c e p t e d M a n u s c r i p t (Ambatipudi et al. 2016), moreover at protein level there are strong evidences for the number of GPR15-expressing T cells in blood as a high-sensitive and high-specific biomarker for the degree of systemic disturbed homeostasis by chronic tobacco smoking.
With respect to the recently emerged therapeutic interest in GPR15L-GPR15 interaction for management of chronic diseases, such as autoimmune disorders, the present study aimed at characterization of GPR15+ T cells in blood in more detail. We show that the GPR15 is a unique surface receptor which is not accompanied concomitantly by another cellular protein and not restricted to a specific T cell subtype. Additionally we show the different impact on T cell subtypes of chronic smoking as, to date, the only specific inducer of an excess of GPR15+ T cells in blood.

Clinical Significance
 The physiological role of GPR15 remains elusive.
 Proportion of GPR15-expressing T cells in peripheral blood is an unspecific indicator for pathological processes, such as inflammatory bowel disease, but is a highly specific and sensitive biomarker for chronic tobacco smoking.
 Smoking behavior should be considered for suggested therapeutic intervention of the GPR15-GPR15L interaction in autoimmune diseases.

Subjects
Pseudonymous blood samples from volunteers (22 non-smokers and 23 chronic active smokers) were obtained from the blood bank at the University of Leipzig (Figure 1).
Smoking behaviour (yes or no), age and gender were recorded via questionnaires. Smoking behavior was rechecked by cotinine level in plasma using the Cotinine direct ELISA Kit according to manufacturer's instruction (DRG Instruments GmbH, Marburg, Germany) Cotinine level exceeding the sensitivity level of the assay (1 ng/ml) was considered as smoker. All participants gave their written informed consent. The study received approval A c c e p t e d M a n u s c r i p t by the Ethics Committee of the University of Leipzig (reference number 079-15-09032015).

Analysis of surface proteins
GPR15 surface expression concomitant with markers of differentiation on lymphocytes was analysed by flow cytometry. Briefly, 100 µl of blood specimen was incubated with mouseanti-human-GPR15 antibody (1:500; R&D Systems, Wiesbaden-Nordenstadt, Germany) supplemented with 5% goat serum for 1 h. After washing in PBS/1% fetal calf serum (FCS) the GPR15 was stained with R-phycoerythrin-labelled goat-anti-mouse IgG2b (1:500, 1 h; Biozol, Eching, Germany) following an additional wash step. Thereafter, cells were incubated with 5% mouse serum for 30 min following a double-staining step for leucocyte A c c e p t e d M a n u s c r i p t

Lymphocyte separation
Peripheral Blood Mononuclear Cells (PBMCs) were obtained by density-gradient centrifugation using Ficoll-Paque (GE Healthcare, Solingen, Germany). GPR15+CD3+ and GPR15-CD3+ T cells were separated by flow cytometric cell sorting after antibody staining of PBMCs for CD3 and GPR15 surface proteins as indicated above. Flow-cytometric cell sorting was performed at the laboratory of cytometry of the Core Facility at the University of Leipzig.

Gene expression by quantitative PCR
For semi-quantitative PCR (qPCR), total RNA of separated leucocytes was prepared by using peqGold RNA Pure (peqlab, Erlangen, Germany) according to manufacturer's instructions. The cDNA synthesis was carried out with 1 µg of RNA by using ImProm-II TM Reverse Transcription System (Promega, Mannheim, Germany). Intron-spanning primers were designed and UPL probes were selected by the Universal Probe Library Assay Design Center (http://qpcr.probefinder.com/organism.jsp) (Table S4). Gene expression of 38 randomly selected genes was measured using the 96.96 Dynamic Array or FLEXsix Integrated Fluidic Circuits (IFCs) (Fluidigm, San Francisco, CA, USA). Gene expression values were determined by using the 2 -∆∆CT method (Livak and Schmittgen 2001) with GAPD and GUSB as reference genes and normalized to the lowest measured value.

Transcriptome-sequencing
Total RNA was extracted as described previously (Hackermuller et al. 2014 (Illumina, San Diego, CA, USA) on a HiSeq2000 system. Adapters were clipped from the sequencing reads using cutadapt (v 1.5) (Martin 2011). Reads were mapped to the human genome version hg19 using TopHat2 (v2.0.13) (Kim et al. 2013). Read counts for Gencodeannotated transcripts were determined using htseqcount (v 0.6.1p1) (Anders et al. 2015) with mode intersection-strict. Counts were normalized and transformed using R and the variance stabilizing transformation of the DESeq2 package (Love et al. 2014) from Bioconductor.
Raw sequencing data with appropriate experimental information is available in the NCBI Gene Expression Omnibus (GEO) repository under the accession number GSE114837 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE114837)

Statistics
Statistical significance of parametric distributed values was calculated with Student's t-test.
Otherwise, the nonparametric Mann-Whitney U test and Wilcoxon-test were applied for comparison. Boxes in figures indicate the 25 and 75% percentile, whiskers the non-outlier range. Both sensitivity (proportion of positives that are correctly identified as such) and specificity (proportion of negatives that are correctly identified as such) as statistical measures of a binary classification test were calculated based on a cut-off (mean plus one standard deviation) of each T cell subtype. All p-values <0.05 were considered to be significant. For multiple testing, the significance level was adjusted according to A c c e p t e d M a n u s c r i p t

GPR15 expression in T cells is uniquely, not accompanied by another protein
To better describe the physiological role of this cell type we checked first whether expression of other proteins concomitantly follow the GPR15 expression in T cells. As a proxy approach for that, flow cytometric sorted CD3+GPR15+ T cells from peripheral blood were compared with CD3+GPR15-T cells by RNA sequencing (Table S1, Figure 2).
The GPR15 expression was not concomitantly accompanied by expression of another gene neither in the similar range of transcripts (TPM, transcript per million) nor by similar fold change of expression. Additionally, intraindividual differences of randomly selected differentially expressed genes (n, 38) between GPR15-and GPR15+ T cells did not correlate with GPR15 expression (Table S2). It can therefore be assumed, that the expression of GPR15 at protein level is uniquely, not specifically accompanied by another protein.
Additionally, comparing particular differences of transcript expression by RNAseq with protein expression we could confirm the above assumption. Analyzing surface receptors which were used for identification of specific T cell subtypes, GPR15+ differed from GPR15-T cells by the proportion of receptor-expressing cells (Figure 3, Table S3). A c c e p t e d M a n u s c r i p t
A c c e p t e d M a n u s c r i p t

Smoking induces an excess in each of the GPR15-expressing T cell subtypes in blood
Independent on smoking behavior, GPR15 was nearly absent in CD45RA high naive T cells and MAIT cells but occurred in all other analyzed T cell subtypes ( Figure 5). The abundance of GPR15+ cells differently distinguished in T cell subtypes compared to the whole CD3+ T cell population (NS, 5.1% [confidence interval, 3.9% -6.3%]; S, 13.8% In contrast to the strong smoking-dependent increase in proportion of GPR15+ cells for most of the analyzed differential T cell subtypes, the cell count of T cell subtypes was similar between non-smoker and smoker, except for the slightly higher proportion of CD3+ [5.2% -12.3%, p=0.013], respectively)( Figure S2).

Proportion of GPR15+ T cells in blood is the best specific and sensitive cellular biomarker of smoking
As depicted in figure 5, the analysis of proportion of GPR15+ cells among T cell subtypes in peripheral blood can be used as a biomarker to distinguish chronic active smoker from non-smoker. By using a cut-off of mean ± one standard deviation, the most specific and sensitive differentiation into non-smoker and chronic smoker was given for Th17 (sensitivity, 1; specificity, 0.91). In contrast, smoking-influenced proportion of MAIT in blood was much less specific to stratify for smoking (sensitivity, 1; specificity, 0.23).
A c c e p t e d M a n u s c r i p t

Discussion
For the first time, we recently have reported T cells in peripheral blood as the main GPR15+ leucocyte evoking excessively by chronic smoking . To get more insight into the physiology as well as pathological importance of GPR15+ T cells we first performed transcriptome analyses by RNAseq and semiquantitative PCR as proxy approaches to identify proteins concomitantly expressed with GPR15. Starting with RNAseq, this approach was done exclusively with separated cell subtypes from one healthy volunteer. Since the focus of this work was not the subject-or exposure-triggered interindividual differences in gene expression, the RNAseq analysis of separated T cells on one subject was sufficient to show co-expressed genes/proteins. Based on the whole transcriptome, the GPR15 surface protein expression seemed to be clearly unique, not accompanied by co-expression of another intracellular or surface protein. This result was substantiated in a greater cohort by (i) a smoking-independent absent correlation between intraindividual GPR15 expression and randomly selected differentially expressed genes and by (ii) the findings from analyses of surface receptors that, unlike GPR15, differences in gene expression between GPR15-and GPR15+ T cells exclusively relied on different proportions of gene-expressing cells but not on a changed cellular abundance of transcripts in each cell.
In blood, we found the distribution and proportion of GPR15+ cells were strikingly influenced by the differentiation state of T cells. The vast majority of about 80% of GPR15+ T cells consisted of diverse subtypes of central memory T cells (T CM ). A less frequent occurrence of GPR15 was found in cells with effector-like features, with an absence in effector-like MAIT. In contrast to the replicated absence of GPR15+ Treg in the human colon mucosa (Nguyen et al. 2015, Fischer et al. 2015 with the concluding statement about a probable impossibility of GPR15 expression in Treg cells (Nguyen et al. 2015), we and other clearly have shown that proportions of both natural (nTreg (Ocon et al. 2017, Fischer et al. 2015 and active nTreg) and further Tregs (Tr1, Th3) in blood were able to express GPR15. Irrespective of the presence of GPR15+ cells among nearly all analyzed T cell subtypes (Th1,Th2,Th9,Th17,Th22 et al.), the most abundant GPR15+ T A c c e p t e d M a n u s c r i p t cells were Th17. This overrepresentation should be the reason for an increased IL-17 secretion of GPR15+ CD4+ T cells in healthy PBMCs in comparison to GPR15-T cells (Adamczyk et al. 2017).
To date, the physiological mechanism to release GPR15+ lymphoid cells into blood remains unknown. First changes in proportion of GPR15+ cells among T cell subtypes of blood were reported exclusively for Treg in patients with ulcerative colitis (Fischer et al. 2015). However, due to the marginal (difference between number of GPR15+ Treg in non-UC and UC patients, <10%) and interindividual changes the proportion of GPR15+ Treg cannot serve as a clinical biomarker for UC. Chronic tobacco smoking is to date the only known specific inducer for an excess of GPR15+ cells in blood . We now have shown that this excess was not restricted to any GPR15+ expressing T cell subtype in blood. The extent of GPR15+ cells individually reached values of up to 80% in single T cell subtypes, especially in Th17. A degree of proportion similar found after HIV-1 infection of a CD4+ T-cell clone (PM1, up to about 60% of infected cells) in vitro (Kiene et al. 2014) or in seemingly natural occurring CD25+CD127 high FoxP3-Th in mucosa of the colon (Nguyen et al. 2015).
Because of the strong association of GPR15 expression with epigenetic DNA methylation at CpG site cg19859270 (annotated to GPR15), it can be assumed from epigenome-wide association studies (EWAS) that the excess of GPR15+ cells in blood increases linearly with increasing smoking dosage (pack years, (Ambatipudi et al. 2016)). In our study we could not prove this expectation because of missing data about smoking dosage. A sustained T cell stimulation critically dependents on a continuous supply of antigencarrying DCs (Lanzavecchia and Sallusto 2000). Therefore, it was recently postulated that the tobacco smoking-induced disturbance of homeostasis, primarily in the lung, might be the origin to imprint T cells by activated antigen-presenting dendritic cells (DC). However, a smoking-independent disturbed homeostasis by different lung pathologies did not evoke an excess of GPR15+ T cells in blood in never-smoking patients (Bauer et al. 2017).
Similarly, a mucosal inflammation did not affect the content of GPR15+ effector memory Th cells in inflamed colon mucosa (Nguyen et al. 2015). Whether GPR15 induction is A c c e p t e d M a n u s c r i p t influenced by water-soluble compounds of cigarette smoke remains speculative, but could not have been confirmed by a simple exposure of PBMCs to an aqueous cigarette smoke extract in an artificial in-vitro approach .
Apart from the cellular response on smoking by augmentation of GPR15+ T cells, it has long been known that smoking impairs myeloid (Friedman et al. 1973, Petitti andKipp 1986) and lymphoid cell (Moszczynski et al. 2001, Ammitzboll et al. 2017) count in whole blood. With respect to T cell subtypes we could confirm the recently described smokinginduced reduction of mucosal associated invariant T cell (MAIT) content in blood (Ammitzboll et al. 2017). Interestingly, that one T cell subtype which not expressed GPR15. However, because of several MAIT-depleting adverse physical conditions in peripheral blood (Hinks et al. 2015, Serriari et al. 2014, Bottcher et al. 2018) the abundance of MAIT was less specific in comparison to the proportion of GPR15+ T cells to discriminate subjects for smoking behavior. Since the content of MAIT was reduced both in blood and liver tissue in non-infectious liver pathologies, their reduction was consequently proposed to be rather a sign of exhaustion induced by chronic activation, a mechanism seemingly similar evoked by chronic smoking. In contrast, the smoking-induced excess of GPR15+ T cells in blood is probably more exposure-specific, therefore applicable as a convincing biomarker for chronic smoking reaching maximal sensitivity and specificity of 1 each (Bauer et al. 2017). In this study we obtained only a suboptimal specificity of 0.91 since we could not discriminate non-smoker from never smoker by missing questionnaire for that.
Because of the migratory activity of seemingly each GPR15-expressing T cells in blood, including Treg, Tmemory, Teffector, towards GPR15L (Ocon et al. 2017), it can be presumed that, acting as a homing receptor, a similar diverse pattern of GPR15+ T cell subtypes in peripheral blood occurs at the site of homing, such as epidermis or colon mucosa, with an expected slightly overrepresentation of Th17 cells. The prominent presence of Th17 at sites of tissue inflammation such as psoriasis and inflammatory bowel A c c e p t e d M a n u s c r i p t disease (Korn et al. 2009) as well as high proportion of IL-17-secreting GPR15+ memory Th cells in UC mucosa would confirm this assumption (Nguyen et al. 2015). However, similar to the differentiation state of T cells the microenvironments should decisively influence GPR15 expression and cell type specific homing. (i) The higher, compared to normal, extent of GPR15+ cells in blood by chronic smoking, (ii) the repressed GPR15 expression of Treg in human colon mucosa but not in blood, or (iii) the induced GPR15 expression under artificial polarizing conditions in vitro (Nguyen et al. 2015) underline the role of tissue-specific microenvironment to influence the abundance of GPR15+ cells.
Irrespective of GPR15, the final role of Th17 is decisively triggered by the local microenvironment. It may generate either non-pathogenic or pathogenic Th17 cells.
Interestingly, under non-pathogenic conditions the GPR15 was about 1.9-fold overexpressed in Th17 cells (Lee et al. 2012), what might account for a doubling of abundance of GPR15+ Th17 cells under non-pathogenic conditions.
Additionally, since GPR15-deficient mice were prone to develop severe large intestine inflammation (Kim et al. 2013) a similar preventive role of GPR15-expressing cells was herewith documented. Altogether, GPR15 might be considered more as a sign of protection from harm by inflammation than a harm by itself. In healthy chronic smoking subjects, therefore, the pronounced GPR15 expression in blood might, irrespective to be an indicator of adverse effects, be important to prevent harm from smoking.

Conclusion
In this report we highlighted the strong impact of chronic smoking on the imprinting of T cells for GPR15 occurring high-frequently in vascular circulation. Whether smokinginduced enrichment of GPR15+ lymphocytes in blood or appropriate tissue has a healthpreventing effect of smoking for patients suffering from ulcerative colitis (Bridger et al. 2002) or, in opposite, has a health-threatening effect of smoking for patients suffering from incident psoriasis (Naldi 2016) could not be resolved in this study. Indeed, severe large intestine inflammation was rescued by the transfer of GPR15-sufficient Treg indicating that GPR15 plays a role in mucosal immune tolerance, at least in mice (Kim et al. 2013).
A c c e p t e d M a n u s c r i p t Therefore, whether the GPR15L-GPR15 interaction may be an option for systemic therapeutic intervention needs to be proven in human. In mice, first evidence was given by systemic application of a GPR15L in form of the fifth region of epidermal growth factorlike domain of TM (TME5) that alleviated an acute graft-versus-host disease by its proposed anti-inflammatory impact ).
However, because of the naturally occurring GPR15+ lymphocytes in blood of healthy subjects and the packyears-dependent increased presence of GPR15+ lymphocytes in blood of healthy smokers we suggest that GPR15+ lymphocytes in blood of smokers constitute rather a sign of adaptation to prevent harm particularly from tobacco smoking. Therefore, it should be proven whether therapeutic intervention of the GPR15-GPR15L interaction will be applicable in smokers, too.