Complement receptor 2 and IL-8 production identifies in adults and neonates naïve T cells recently arising from the thymus

The adaptive immune system utilizes multiple mechanisms linked to innate immune cell functions to respond appropriately to pathogens and commensals. Here we discover further aspects of this connectivity by demonstrating that naive T cells as they emerge from the thymus express complement receptor 2 (CR2), the bacterial pathogen recognition receptor TLR1 and an enzyme that deactivates bacterial lipopolysaccharide (AOAH) and following activation during tissue immunesurveillance secrete the anti-microbial cytokine IL-8. CR2+ naive cells and a novel subset of IL-8-producing CR2+ memory cells are abundant in children but decrease with age. The ability of CR2, which is also a receptor for Epstein-Barr Virus (EBV), to identify recent thymic emigrants will facilitate assessment of thymic function during aging and aid investigations of multiple clinical areas including generation of T cell lymphomas caused by EBV

The maintenance of a diverse, naïve T cell repertoire arising from the thymus (recent thymic emigrants: RTEs) is critical for health 1 .Recent studies have reported a unique naïve CD4 + T cell subset in neonates and early childhood characterized by IL-8 production 2,3 .Here we demonstrate that IL-8 production is a characteristic feature of RTEs in adults, children and neonates and that a hallmark of these cells is the expression of complement receptor 2 (CR2) and the preferential production of IL-8 after activation.Although decreasing in number with age due to thymic involution and homeostatic expansion of naïve CD4 + T cell in humans 4 , CR2 + RTEs persist into old age, have the highest levels of T cell receptor excision circles (TRECs), co-express complement receptor 1 (CR1), TLR1 and produce IL-8 upon TCR stimulation.We have observed these phenotypes in the vast majority of cord blood naïve CD4 + T cells and in newly-generated naïve T cells appearing during reconstitution of the immune system in adults depleted of T cells through treatment for multiple sclerosis.A memory subset of CR2 + CD4 + T cells expressing high levels of CR1 and producing IL-8 following activation was also discovered, consistent with the hypothesis that RTE-specific gene expression confers a functional competence retained by particular memory T cells possibly because of their complement-dependent reactivity to pathogens.We suggest that assessing CR2 expression on naïve CD4 + T cells will give a measure of thymic function during aging of the immune system and in a number of clinical situations including bone marrow transplantation 5 , HIV infection 6 , and immune reconstitution following immune depletion 7 or chemotherapy 8 .Thymic involution decreases naïve CD4 + T cell production with age in humans and is compensated for by the homeostatic expansion of naïve cells that have emigrated from the thymus earlier in life 1,4 .Naïve T cells that have undergone decades of homeostatic expansion show reduced T cell receptor diversity that potentially negatively impacts host defence 9 .CD31 (PECAM-1) expression identifies cells that have divided more often in the periphery (CD31 − ) from those that have not (CD31 + ), although CD31 + T cells still divide with age as evidenced by the dilution of TRECs 10 and, therefore, better markers of undivided naïve T cells are clearly needed 11 .As we and others have reported 12,13 , CD25 is an additional marker of naïve T cells that have expanded in the periphery.CD31 + CD25 − naïve CD4 + T cells predominate at birth and contain the highest content of TRECs as compared to their CD31 − and CD25 + counterparts 12 .
Here, we assessed the heterogeneity of four naïve CD4 + T cell subsets defined by CD31 and CD25 expression in neonates and adults in a population study of 391 donors (Fig. 1a, Supplementary Fig. 1a).CD31 + CD25 − naïve CD4 + T cells decreased with age and this decrease was compensated for by the homeostatic expansion of three subsets of naïve T cells: CD31 + CD25 + , CD31 − CD25 − and CD31 − CD25 + .As expected 14 , the proportion of both naïve CD4 + and CD8 + T cells declined with age (Supplementary Fig. 1b).To define markers associated with the least expanded naïve subset, we performed statistically powered, genome-wide RNA analysis of FACS-purified naïve CD4 + T cells from 20 adults sorted into four subsets based on CD31 and CD25 expression (Supplementary Fig. 1c).
Principal component analysis of differentially expressed genes amongst the four subsets showed a clear separation between the groups, particularly between CD31 + and their CD31 − counterparts (Supplementary Fig. 1d).Genes upregulated in CD31 + CD25 − naïve cells as compared to the CD31 − CD25 − subset included two genes not normally associated with T cells: AOAH, which encodes the enzyme acyloxyacyl hydrolase that inactivates LPS and is highly expressed in innate immune cells 15 , and CR2, which encodes a cell surface protein that binds C3d and other complement components 16 and is also a receptor for EBV in humans 17 (Fig. 1b) (Supplementary Tables 1A-E).CR2 is expressed by follicular dendritic cells and B cells, and in the latter case, lowers their activation threshold when complement and antigen are both present 18 .CR2 expression has also been reported on CD8 + and CD4 + foetal T cells 19 and thymocytes 20 .Genes upregulated in expanded CD31 − CD25 − cells as compared to CD31 + CD25 − cells are consistent with the occurrence of activation and differentiation events during the homeostatic proliferation of this naïve subset and include a gene encoding an interferon-induced intracellular DNA receptor PYHIN1, CTLA4 and IL2RB, as well as transcription factors such as IRF4, PRDM1 (encoding BLIMP-1) and MAF.Expression of two genes, TOX (a transcription factor reported to regulate T-cell development in the thymus 21 ) and CACHD1, an uncharacterized gene that may encode a protein that regulates voltage-dependent calcium channels, was downregulated when CD31 + CD25 − cells either lost expression of CD31 (Fig. 1b) or upregulated CD25 (Supplementary Fig. 1e).
CR2 expression analysed at the protein level verified the microarray results with the CD31 + CD25 − naïve CD4 + T cell subset having the highest proportion of cells positive for CR2, with the proportion decreasing with age (Fig. 1c, Supplementary Fig. 2a).The reduction of CR2 + naïve cells by age was more evident when considered out of total CD4 + T cells (Supplementary Fig. 2b).The CR2 + fraction of the CD31 + CD25 − naïve CD4 + subset has the highest CR2 density.A similar pattern of decreasing levels of CR2 with age was also observed on naïve CD8 + T cells (Supplementary Fig. 2c).The proportion and density of CR2 expression was greatly reduced in the three naïve CD4 + populations previously shown to have a reduced TREC content when compared to CD31 + CD25 − naïve CD4 + cells 12 .
To determine whether CR2 is a molecular marker of CD31 + CD25 − naïve CD4 + T cells that have proliferated the least in the periphery since emigrating from the thymus, we sorted CR2 hi , CR2 low and CR2 − CD31 + CD25 − naïve CD4 + T cells from four adult donors (Fig. 1d) and CR2 + and CR2 − cells from three additional adult donors (Supplementary Fig. 2d) and assessed TREC levels.Sorted CD31 − naïve CD4 + T cells were used as a comparator (Fig. 1d) since the loss of CD31 expression from naïve cells is associated with a substantial loss of TRECs 4,10,12 .CR2 + cells had more TRECs than CR2 − cells in all cases.Where cell numbers were sufficient to separate the CR2 hi and CR2 low naïve CD4 + T cells, the CR2 low cells had undergone one or two more cell divisions than the CR2 hi cells indicating that high CR2 expression on naïve CD4 + T cells identifies cells that have divided the least since leaving the thymus (Fig. 1d).These observations explain why naïve CD4 + T cells isolated only according to CD31 expression show an age-dependent loss of TRECs 4,10 .
To further test the hypothesis that CR2 expression on naïve T cells defines RTEs throughout life rather than being specific to cells generated during the neonatal period, we monitored newly generated naïve CD4 + T cells in eight multiple sclerosis patients depleted of T and B lymphocytes using a monoclonal antibody specific for CD52 (alemtuzumab) 7 (Fig. 2, Supplementary Fig. 3a).
Twelve months after depletion, all eight patients had more naïve CD4 + (Fig. 2a) and CD8 + (Supplementary Fig. 3b) T cells expressing CR2 as compared to baseline demonstrating that de novo RTEs produced in the adult thymus are also defined by CR2 expression.Interim time points were available from most patients (Fig. 2b, Supplementary Fig. 3c) showing that when the first few naïve CD4 + T cells were detected after depletion (3-9 months post-treatment), they were essentially all CR2 + with a density of CR2 equalling that seen in cord blood (Supplementary Fig. 2a).This observation was independent of whether a patient had good or poor naïve CD4 + T cell reconstitution overall.
A potential utility of CR2 is, therefore, to assess the functionality of the human thymus.Along these lines we compared the frequency of CR2 + cells within the CD31 + CD25 − naïve CD4 + T cell subset prior to lymphocyte depletion with the ability of the thymus to reconstitute the naïve CD4 + T cell compartment, as a proportion of total CD4 + T cells.The two patients that failed to reconstitute their naïve T cell pool to pre-treatment levels by 12 months had the lowest levels of CR2 + T cells within the CD31 + CD25 − naïve CD4 + T cell subset (18% and 17%) prior to treatment (Fig. 2c, Supplementary Fig. 3c).In contrast patients who reconstituted their naïve CD4 + T cell pool to or above baseline by To evaluate the potential function of CR2 + naïve CD4 + cells, we compared RNA isolated from sorted CR2 + and CR2 − CD31 + CD25 − naïve CD4 + T cells ex vivo and after activation (Fig. 3a, Supplementary Tables 2-4).This strategy revealed a transcriptional signature that reinforced the uniqueness of the CR2 + RTEs.Complement receptor 1 (CR1) and TLR1 were more highly expressed in CR2 + cells, and coexpression at the protein level for CR1 and CR2 was observed in CD31 + CD25 − naïve CD4 + T cells from healthy controls (Fig. 3b) and MS patients reconstituting their naïve cells (Supplementary Fig. 4a).
Co-expression of CR1 and CR2 was also observed on naïve CD8 T cells (Supplementary Fig. 4b).
Following activation, IL8 was more highly expressed in CR2 + cells whereas IL2, IL21, LIF and IFNG were more highly expressed in CR2 − cells.TNF, LTA and IL23A were highly upregulated with activation but there was no difference between the subsets.IL8 upregulation was of particular interest since it has been termed a phenotype of neonatal naïve cells 2 .We therefore measured IL-8 and IL-2 protein production from sorted CR2 + versus CR2 − cells after activation and verified the RNA results showing that IL-8 is preferentially produced by the CR2 + subset whereas the opposite is the case for IL-2 (Fig. 3c).Because CR2 rapidly disappears from the surface of cells activated in vitro, it prevents the analysis of cytokine production using this marker after stimulation.Therefore, we examined IL-8 production in CD4 + T cells stratified by CD45RA expression from cord blood, MS patients reconstituting their naïve T cell compartment (<1 year after treatment) and MS patients in whom naïve T cells had undergone homeostatic expansion (>10 years after treatment) (Fig. 3d).
Notably all MS patients with high levels of RTEs that were tested (N=4) had a high proportion of their naïve CD4 + T cells producing IL-8 at similar levels as that of cord blood naïve CD4 + T cells.On the other hand, patients with naïve T cells that had expanded (N=3) produced much less IL-8.IL-8 production by naïve CD4 + T cells positively correlated with their CR2 expression confirming the ability of CR2 + RTE to produce IL-8.Our results are compatible with those from van den Broek et al.
showing reduced IL-8 production in children following neonatal thymectomy 3 .
When analysing IL-8 production by CD4 + T cells we noted that in healthy controls (Fig. 3c) and in MS patients who were more than 10 years past lymphocyte depletion (Fig. 3d), a fraction of activated CD45RA − memory CD4 + T cells produced IL-8.We therefore hypothesized that memory T cells can be generated from IL-8 producing CR2 + naïve T cells.CR2 expression was observed on a proportion of central and effector memory CD4 + T cells and Tregs expressed the lowest levels of CR2 (Fig. 4a, Supplementary Fig. 5a).CR2 expression by memory cells was correlated with CR2 expression by naïve T cells (Fig. 4b) and was age dependent (Fig. 4c) suggesting that with age the CR2 + memory .CC-BY-NC-ND 4.0 International license not peer-reviewed) is the author/funder.It is made available under a The copyright holder for this preprint (which was .http://dx.doi.org/10.1101/059535doi: bioRxiv preprint first posted online Jun.17, 2016; cells either lose CR2 expression or contract due to competition with other CR2 − memory cells.CR2 + memory CD8 + T cells were also observed and were similarly age dependent (Supplementary Fig. 5b).
As seen with the equivalent CD4 + naïve T cell subsets, RNA analysis of sorted CD4 + CR2 + and CR2 − central memory T cells (Supplementary Table 6) showed CR1 to be the most differentially expressed gene (Fig. 4d), a phenotype confirmed at the protein level (Fig. 4e).CR2 + memory cells produced higher levels of IL-8 after activation as compared to CR2 − memory cells, and unlike naïve CR2 + cells (Fig. 3c), all memory cells producing IL-8 produced IL-2 (Fig. 4f).Amongst the differentially expressed genes between the CR2 − and CR2 + memory cells, a gene of particular note is complement factor H (CFH) (Fig. 4d), which is upregulated in both CR2 − and CR2 + memory cells as compared to naïve cells but has 2.7-fold higher levels in CR2 + memory cells compared to CR2 − memory cells.
Shared expression between CR2 + naïve and CR2 + memory T cells ex vivo are highlighted in Supplemental Fig. 6 and include a potentially relevant transcription factor for CR2 + CD4 + cells, ZNF462, which is expressed 13.0-fold higher on CR2 + versus CR2 − naïve cells and 7.2-fold higher on CR2 + versus CR2 − memory cells.Four genes, ADAM23, ARHGAP32, DST and PLXNA4, shared by the two CR2 + subsets may enhance migratory properties that augment host surveillance 22 .
Our results demonstrate that relatively recent development in the thymus, not absolute age, confers a unique innate phenotype to naïve CD4 + T cells that includes expression of complement receptors, TLR1 and a group of genes, including those encoding the transcription factors HELIOS (encoded by IKZF2) and ZNF462, that appear to program the RTEs to preferentially secrete IL-8 and have reduced production of other cytokines following activation (Supplementary Table 5).We propose that the presence of CR1 and CR2 in naïve CD4 + as well as CD8 + T cells contribute to host defence (Supplementary Fig. 6), possibly in a manner similar to that provided by these same molecules on B and follicular dendritic cells 18 .Aspects of the innate signature of RTEs are retained by a subset of memory T cells that express both CR1 and CR2 and secrete IL-8 upon activation, suggesting there is selection because of these functional attributes.Gibbons and colleagues 2 have characterized IL-8 as a "proinflammatory immunoprotective cytokine of neonatal T cells" that recruits neutrophils and costimulates γδ T cells to produce IFN-γ, whereas we propose that this description should be expanded to include all CR2-expressing naïve CD4 + RTEs and a subset of memory CD4 + T cells.Our findings complement those of Liszewski et al. who have shown that human T cells express and process C3 to components that can bind CR2 23 .Furthermore, a complement-regulatory protein, CD46, influences cytokine production from human CD4 + T cells 24 .
. CC-BY-NC-ND 4.0 International license not peer-reviewed) is the author/funder.It is made available under a The copyright holder for this preprint (which was .http://dx.doi.org/10.1101/059535doi: bioRxiv preprint first posted online Jun.17, 2016; As humans age, we have shown that with homeostatic expansion naïve T cells differentiate into naïve cells that have reduced expression of CR1 and CR2 and no longer produce IL-8 with activation.
These attributes could contribute to the reduced host defence seen in some older individuals 1 , although there is marked heterogeneity at the rate of loss of RTEs amongst people.Further analyses of T cells that secrete IL-8 and the functional effects of EBV, which binds to CR2 17 , on CR2 + T cells are ongoing.We suggest that loss of CR2 expression on naïve cells with homeostatic expansion contributes to the increasing severity of EBV infection with age 25 .We also propose that the expression of CR2 on CD31 + CD25 − naïve CD4 + T cells defines cells that have divided the least in the periphery since emigrating from the thymus and that the abundance of this subset reflects the functional age of the thymus.The ability of CR2 to be a surrogate marker of TREC levels in naïve cells will be useful in a number of clinical settings where thymic function is assessed [5][6][7][8] .

ACKNOWLEDGMENTS:
This research was supported by the Cambridge NIHR BRC Cell Phenotyping Hub.In particular, we wish to thank Chris Bowman and Anna Petrunkina Harrison for their advice and support in cell sorting.We thank Sarah Dawson, Pamela Clarke, Meeta Maisuria-Armer and Gillian Coleman for their help in processing blood samples and Jane Kennet and Katerina Anselmiova for coordinating and obtaining blood samples from the Investigating Genes and Phenotypes Associated with Type 1 Diabetes study.We thank Karen May for obtaining blood samples from MS patients and Thaleia Kalatha for determining sample availability from patients.We thank Emma Jones for the use of the NanoString Instrument and Sarah Howlett for editorial review of the manuscript.We gratefully acknowledge the participation of all NIHR Cambridge BioResource volunteers, and thank the NIHR Cambridge BioResource centre and staff for their contribution.We thank the National Institute for  The copyright holder for this preprint (which was .http://dx.doi.org/10.1101/059535doi: bioRxiv preprint first posted online Jun.17, 2016; (REC reference: 12/LO/0393, EudraCT number: 2011-005606-30).Additional MS patients treated with alemtuzumab greater than 10 years before analysis consented to long-term follow-up (CAMSAFE REC 11/33/0007).
Whole blood and PBMC immunostaining.Blood samples were directly immunophenotyped within 5 hours following donation.Samples were blocked for 10 min with mouse IgG (20 μg/ml), stained for 40 min at room temperature with appropriate antibodies and then lysed with freshly prepared 1X BD FACS Lysing Solution (BD Biosciences).After lysis of red blood cells, samples were washed with BD CellWASH (BD Biosciences).Finally, the samples were fixed with freshly prepared 1X BD CellFIX (BD Biosciences).The samples were stored at 4 °C in the dark until analysis using a BD Fortessa flow cytometer.PBMC samples, prepared as previously described 26 , were blocked for 10 min, stained for 1 hour at 4°C, washed twice and fixed as described for peripheral blood immunophenotyping except for intracellular staining when surface-stained cells after the wash-step were placed in FOXP3 Fix/perm buffer (eBioscience).Phenotyping panels are detailed in Supplemental Table 7. CD25 detection sensitivity was increased 26 by simultaneous application of two anti-CD25 monoclonal antibodies labelled with the same fluorochrome (clones 2A3 and M-A251, BD Biosciences).Antibody concentrations used were based on the manufacturer's instructions as well as on optimization studies.Appropriate isotype controls and fluorescence-minus-one conditions were used during the development of staining panels.
Flow cytometry and data analysis.Immunostained samples were analysed on a BD LSRFortessa cell analyzer and data were visualized using Flowjo (TreeStar).
Cryopreserved PBMC.PBMC isolation, cryopreservation and thawing were performed as previously described 26 .Briefly PBMC isolation was carried out using Lympholyte (CEDARLANE).PBMCs were cryopreserved in heat-inactivated, filtered human AB serum (Sigma-Aldrich) and 10% DMSO (Hybri-MAX, Sigma-Aldrich) at a final concentration of 10 × 10 6 /ml and were stored in liquid nitrogen.Cells were thawed in a 37 °C water bath for 2 min.PBMCs were subsequently washed by adding the cells to 10 ml of cold (4 °C) X-VIVO (Lonza) containing 10% AB serum per 10 × 10 6 cells, in a drop-wise fashion.PBMCs were then washed again with 10 ml of cold (4 °C) X-VIVO containing 1% AB serum per 10 × 10 6 cells.
T cell subset purification by cell sorting and DNA isolation.CD4 + T cells (RosetteSep Human CD4 + T Cell Enrichment Cocktail, STEMCELL Technologies) were washed and immediately incubated with .CC-BY-NC-ND 4.0 International license not peer-reviewed) is the author/funder.It is made available under a The copyright holder for this preprint (which was .http://dx.doi.org/10.1101/059535doi: bioRxiv preprint first posted online Jun.17, 2016;     antibodies to surface molecules (Supplementary Table 7) for 40 minutes at 4 °C, washed and followed by sorting on a FACSAria Fusion flow cytometer cell sorter) into X-VIVO medium (Lonza) containing 5% human AB serum (Sigma).In order to isolate DNA, sorted cell subsets were checked for purity and DNA was isolated using a DNA extraction reagent (QIAGEN).TREC assay.TREC assay was performed as described previously 12 .Briefly, a quantitative PCR assay was purchased from Sigma-Genosys for the signal joint TCR excision circle (sjTREC) that arises through an intermediate rearrangement in the TCRD/TCRA locus in developing TCRαβ + T lymphocytes.An assay for the gene encoding albumin was used to normalise the data.For each sample, 24 ng of DNA was incubated in duplicate with both primers (700 nM), probe (150 nM) and 12.5 μl TaqMan mastermix (Applied Biosystems) and processed using the Applied Biosystems™ 7900HT Fast Real-Time PCR System.sjTREC data were calculated with application of the ΔΔCt method; sjTREC content for each subset was graphed as the proportion of sjTREC content present in the most naïve population sorted: CD31 + CR2 high naïve CD4 + T cells in Fig. 1d and CD31 + CR2 + naïve CD4 + T cells in Supplementary Fig. 2d.T cell activation.FACS-purified T cell subsets or total CD4 + T cells (RosetteSep) were stimulated with either anti-CD3/CD28 beads (Life Technologies) at 3 cells per bead overnight or cell stimulation reagent (PMA and Ionomycin, eBioscience) in the presence of protein transport inhibitors (eBioscience) for six hours at 37 °C in 96-well U-bottom plates.IL-8 + /IL-2 + T cells were identified with a staining panel shown in Supplementary Table 7.
Microarray gene expression analysis.Total RNA was prepared from cell subsets isolated by sorting using TRIzol reagent (Life Technologies).Single-stranded cDNA was synthesised from 200 ng of total RNA using the Ambion WT Expression kit (Ambion) according to the manufacturer's instructions.
Labelled cDNA (GeneChip Terminal Labelling and Hybridization Kit, Affymetrix) was hybridized to a 96 Titan Affymetrix Human Gene 1.1 ST array.RNA sequencing yielded on average 35.9 million paired-end reads per library.Maximum likelihood transcript read count estimates for each sample were obtained with Kallisto v0.42.5 31 , using Ensembl Release 82 32 as a reference transcriptome.Gene expression estimates were derived by aggregating all their constituent transcript read counts, which were then employed to perform a paired differential expression analysis using limma v3.28.5 29 .Analyses were performed using a FDR of 0.05%.A missing FDR is reported for genes that did not contain at least 2 counts per million (CPM) in at least 2 samples.Data from RNA-seq are deposited with European Nucleotide Archive, http://www.ebi.ac.uk/ena, accession number EGAS00001001870.

Statistical analysis of flow cytometry data interrogating T-cell subsets.
Statistical analyses of the percentage of cells expressing CR2 and CR1 were performed and presented using Prism 5 software (Graphpad.com).Comparisons between cell subsets were performed using a paired Student's t-test.P < 0.05 was considered significant, error bars show the SD of the samples at each test condition.
We included samples from 389 individuals in the analysis of flow cytometry data versus age.
Regression was done using a non-parametric method, LOESS 33 .The grey zones define a 95% confidence interval for each regression line.Statistical tests were implemented using Prism software and R software (http://www.R-project.org).

Health
Research and NHS Blood and Transplant.We thank the NIHR/Wellcome Trust Clinical Research Facility.This work was funded by the JDRF (9-2011-253), the Wellcome Trust (091157) and the National Institute for Health Research (NIHR) Cambridge Biomedical Research Centre.CW was supported by a Wellcome Trust grant (089989).The research leading to these results has received funding from the European Union's 7th Framework Programme (FP7/2007-2013) under grant agreement no.241447 (NAIMIT).The study was supported by the European Union's Horizon 2020 Research and Innovation Programme under grant agreement 633964 (ImmunoAgeing).The Cambridge Institute for Medical Research (CIMR) is in receipt of a Wellcome Trust Strategic Award (100140).