Impact of donor extracellular vesicle release on recipient cell “cross‐dressing” following clinical liver and kidney transplantation

In several murine models of transplantation, the “cross‐dressing” of recipient antigen presenting cells (APCs) with intact donor major histocompatibility complex (MHC) derived from allograft‐released small extracellular vesicles (sEVs) has been recently described as a key mechanism in eliciting and sustaining alloimmune responses. Investigation of these processes in clinical organ transplantation has, however, been hampered by the lack of sensitivity of conventional instruments and assays. We have employed advanced imaging flow cytometry (iFCM) to explore the kinetics of allograft sEV release and the extent to which donor sEVs might induce cross‐dressing following liver and kidney transplantation. We report for the first time that recipient APC cross‐dressing can be transiently detected in the circulation shortly after liver, but not kidney, transplantation in association with the release of HLA‐bearing allograft‐derived sEVs. In liver transplant recipients the majority of circulating cells exhibiting donor HLA are indeed cross‐dressed cells and not passenger leukocytes. In keeping with experimental animal data, the downstream functional consequences of the transfer of circulating sEVs harvested from human transplant recipients varies depending on the type of transplant and time posttransplant. sEVs released shortly after liver, but not kidney, transplantation exhibit immunoinhibitory effects that could influence liver allograft immunogenicity.


| INTRODUC TI ON
Induction of immune responses to major histocompatibility complex (MHC) mismatched allografts has been traditionally considered to result from the migration of graft-derived antigen presenting cells (APCs), or "passenger leukocytes," to secondary lymphoid organs where they directly present intact donor MHC molecules to recipient alloreactive T cells. This model has recently been called into question, with mounting data from both vascularized and non-vascularized animal transplant models showing that in the early posttransplant period few if any such passenger donor leukocytes are found in secondary lymphoid organs. [1][2][3] In contrast, within hours of transplantation, a much larger number of recipient APCs in the graft-draining lymph nodes or spleen carry donor-type MHC molecules on their surface. 1,3,4 The transfer of intact donor MHC to recipient APCs is known as "cross-dressing" and is partly mediated by allograft-derived extracellular vesicles (EVs) released into the circulation. 1,3 EVs are nanosized membranous particles released by most cell subsets, including graft parenchymal cells, endothelium, and passenger leukocytes. Owing to their small size and their capacity to transport a variety of biomolecules, they function as key mediators of intercellular communication, relayed across a spectrum of biofluids and tissue types. Among their surface protein cargo, EVs carry intact MHC and peptide/MHC complexes, which confers to them the capacity to activate T cells. [5][6][7] Although their direct allostimulatory capacity is weak and requires high EV concentrations, 5,6,8 the potential for such peptide/MHC complexes to elicit alloresponses can be markedly enhanced if transferred to APCs. [8][9][10][11][12] Among the range of EV subtypes, those most widely reported to play a key part in the cross-dressing of APCs are "exosomes," small EVs (50-200 nm) of endosomal origin, exhibiting characteristic morphological features, and thought to bear a characteristic complement of tetraspanin protein surface markers. In recent years, it has become evident that commonly employed exosome isolation techniques in fact isolate heterogenous populations of small EVs with the size characteristics of exosomes, not necessarily of endosomal origin. In the interests of definitional accuracy, such isolates are increasingly referred to as small EVs (sEVs). Though no single marker is known to uniquely identify the exosomal fraction of such isolates, proteomic analyses of sEV subpopulations have identified CD63 as a candidate -thus sEVs bearing CD63 are putatively designated exosomes. [13][14][15] In most experimental transplant models, allo-MHC cross-dressing has been shown to be a highly immunogenic phenomenon. The capacity of cross-dressed dendritic cells (DCs) to elicit alloreactive T cell responses in vitro was first described by Herrera and colleagues 16 and subsequently confirmed in vivo by the same group in a mouse skin transplant model. 4 More recently, Benichou and colleagues showed that injection into naïve mice of allogeneic EVs or of recipient cells cross-dressed in vitro with donor EVs was sufficient to elicit a donor-specific inflammatory alloresponse in vivo. 3 Furthermore, Morelli and colleagues demonstrated that upon acquiring donor EVs, recipient DCs became activated, stimulated alloreactive T cells, and promoted allograft rejection. The immunogenicity of allo-MHC cross-dressed DCs is not restricted to secondary lymphoid organs and is also apparent in the allograft itself. Thus, in rodent models of islet and kidney transplantation, rejection was preceded by the engagement of effector T cells with cross-dressed graft-infiltrating host DCs. 17 However, in contrast to these observations, in a model of spontaneous tolerance following MHC-mismatched murine liver transplantation, recipient intrahepatic DCs cross-dressed with donor sEVs markedly suppressed alloreactive host T cell responses, 18 suggesting that the outcome of recipient APC cross-dressing may vary depending on the type of allograft.
The characterization of the kinetics of allograft sEV release and elucidation of the extent to which donor sEVs induce recipient APC cross-dressing following clinical organ transplantation have been hampered by the lack of sensitivity of conventional flow cytometric instruments. We recently developed a method by which circulating small EVs, including exosomes, can be characterized and quantified using advanced imaging flow cytometry (iFCM). 13 We report here the results of applying this technique to investigate the kinetics of donor-derived, CD63 bearing, sEV release in the clinical contexts of liver and kidney transplantation and describe for the first time the development of recipient APC cross-dressing in clinical transplantation. Our findings confirm that following liver, but not kidney, transplantation, the majority of circulating cells exhibiting donor MHC are cross-dressed cells and not passenger leukocytes. This is, however, a transient phenomenon, which is no longer detectable beyond the first weeks posttransplant regardless of the liver allograft status.

| Study population and design
The study was approved by the North of Scotland Research Ethics Committee (REC Ref: 15/NS/0062) and all participants provided written informed consent. Peripheral blood samples were collected from liver (LTx) and kidney (KTx) transplant recipients immediately before transplantation and at posttransplant days 1, 4, 10, and 30.
Additional sequential peripheral blood specimens were collected from stable LTx >3 years posttransplant who experienced an episode of histologically confirmed rejection following attempted complete immunosuppression withdrawal (clinicaltrials.gov, NCT02498977).
All participants were human leukocyte antigen (HLA) genotyped by polymerase chain reaction sequence-specific oligonucleotide probes (PCR-SSOP, Luminex, Austin, TX) at the Clinical Transplantation Laboratory at Guy's Hospital London.

| Circulating sEV and peripheral blood mononuclear cell (PBMC) isolation
Peripheral blood was collected following standard procedures that minimize contamination by platelets and platelet-derived vesicles. 19 Following cubital vein venepuncture, 3 mL of blood was discarded before collection of 9 mL into BD Vacutainer ® K3-EDTA-coated collection tubes (Becton, Dickinson, Franklin Lakes, NJ). Tubes were inverted gently 5 times and blood was allowed to sit at room temperature for 30 minutes. Whole blood was centrifuged at 400 g (Heraeus Megafuge 40R with 195 mm 7500-3180 rotor, ThermoFisher Scientific, Waltham, MA) for 10 minutes at 20°C to remove cells. The plasma layer was collected and centrifuged again at 5000 g for 10 minutes at 20°C. The resulting platelet-poor plasma (PPP) was aliquoted and stored at −80°C.
small EVs (sEVs) were isolated from PPP by size-exclusion chromatography (SEC) using CellGS Exo-Spin TM Mini Columns according to manufacturer's instructions, and as previously validated. 13,20,21 Specifically, PPP was thawed and centrifuged at 16 000 g for 30 minutes (Sorval Legend Micro 21R equipped with 7500-3424 rotor, ThermoFisher Scientific) to remove large particles and cell fragments, and supernatant transferred to a new micrcentrifuge tube and set on ice. CellGS columns were equilibrated 15 minutes prior to use. Using a micropipette, the preservative buffer on top of the column was discarded, the outlet plug removed, and 200 μL of 0.22 μm-filtered phosphate-buffered saline (fPBS) was added to the top of the column. The column was then centrifuged for 10 seconds at 50 g (Centrifuge 5430 R, equipped with FA-45-24-11-HS rotor, Eppendorf, Hamburg, Germany). 0.1 mL PPP was applied to the column and centrifuged at 50 g for 60 seconds.
The column was then transferred to a new 1.5 mL collection tube, 200 μL fPBS applied to the top, and elution of sEVs performed by a final centrifugation step at 50 g for 60 seconds. Confirmation of sEV isolation was performed as previously described ( Figure S1).

| Circulating sEV analyses by ImageStream x
Multispectral imaging flow cytometric acquisition of sEVs was performed using Amnis ImageStream x MKII (ISx, EMD Millipore, Seattle, WA) as previously described. 13 ISx fluidics were set at low speed, sensitivity set to high, magnification at 60x, core size 7 μm, and the "Hide Beads" option unchecked before every acquisition in order to visualize speed beads in analyses. The ISx was equipped with the following lasers run at maximal power to ensure maximal sensitivity: 405 nm (120 mW), 488 nm (200 mW), 561 nm (200 mW), and 642 nm (150 mW). Upon each start-up, the instrument calibration tool ASSIST ® was performed to optimize performance and consistency. Two channels (Ch01 and Ch09) were set to brightfield (BF), permitting spatial coordination between cameras. Channel 12 was set to side-scatter (SSC), and further fluorescence channels were used for antibody detection as required.
To avoid the risk of coincident particle detection, sEV samples were not run at concentrations greater than 10 10 objects/mL. 22 sEV labeling was performed as previously described. 13

| PBMC analyses by ImageStream x
PBMCs were incubated with viability dye-v450 (eBioscience); according to manufacturer's instructions, and Fc receptor block was Colocalization analyses were performed using the Similarity Bright Detail Score (SBDS) feature in IDEAS ® as previously described. 13,[27][28][29] ISx enables quantitative analysis of the degree of colocalization between fluorophores on a pixel-by pixel basis by comparing digital images captured in each of its image detection channels. The Similarity Bright Detail R3 algorithm within IDEAS ® produces a score (SBDS) serving as a measure of the degree of colocalization between these. Following staining cells were fixed in fixation buffer (BioLegend).

| PBMC analyses by conventional flow cytometry
Conventional flow cytometry was performed using LSR Fortessa flow cytometer (BD) and analyzed using FlowJo software v7.6 (Tree Star,Ashland, OR).

| Cross-dressing of monocyte-derived DC and CD8 T cell proliferation assays
Monocyte-derived DCs were co-cultured for 24 hours in serum-free media (X-VIVO TM 15, Lonza, Basel, Switzerland) with sEVs freshly isolated as described previously from HLA-A2 positive and HLA-A2 negative healthy volunteers; 5 x 10 5 DCs were co-cultured with 50 μl sEV isolate, and successful cross-dressing of DCs confirmed by flow cytometric analysis of HLA-A2 ( Figure 4A-B). This approach achieves consistent cellular cross-dressing, as we have previously described. 13 Harvested DCs were immediately cultured for a further 5 days with 10 5 CD8 T cells at a 1:1 or 1:2 ratio. CD8 T cells were

| Statistical analysis
Statistical analyses were performed by GraphPad Prism v7.0 Software. Student's t and Mann-Whitney tests were used for comparisons between two groups as appropriate, and analysis of variance (ANOVA; with Tukey's posttest) to compare more than two groups. Paired samples were compared by nonparametric Wilcoxon tests (*P < .05, **P < .01, ***P < .001, and ****P < .0001).

| Host cells cross-dressed with donor HLA molecules are highly prevalent following liver transplantation
In order to evaluate the presence of cross-dressed recipient cells following liver transplantation, we enrolled 8 liver and 3 kidney transplant recipients who exhibited HLA class I mismatches amenable to discrimination using available fluorescent anti-HLA antibodies (Table 1). Conventional flow cytometry was used to assess the presence of circulating passenger leukocytes (displaying donor-but not recipient HLA) and cross-dressed recipient cells (displaying both donor-and recipient HLA) following kidney and liver transplantation. but their overall numbers were significantly lower than those of crossdressed cells ( Figure 1C). Advanced imaging flow cytometry was performed by ImageStream x in order to achieve visual corroboration of the cross-dressing of recipient APCs with donor HLA molecules. As compared to recipient HLA, which is diffusely present, donor HLA was seen as discrete spots on the recipient cell, a pattern in keeping with our previously published in vitro analyses of sEV uptake kinetics. 13

| Circulating sEVs bearing donor HLA peak early following liver transplantation and are no longer detectable at late time points after transplantation
Next, we investigated whether circulating EVs of donor origin could be detected following liver or kidney transplantation. To do this, we used advanced imaging flow cytometry -an approach that allows sensitive multiparameter characterization of nanosized particles. 13,22,23 Small EVs were isolated from plasma by size-exclusion chromatography as previously described ( Figure S2). SEC offers significant advantages over alternative methods of sEV isolation. These include a reduced risk of sEV damage during isolation, relatively low co-precipitation of nonvesicular contaminants, the capacity to extract sEVs from low-volume clinical samples, and the reduction of user-variability through the use of commercial SEC columns.
Although no single marker can serve to uniquely identify exosomes, comprehensive proteomic analyses of sEV subtypes identified CD63 as among the most suitable. 14 Thus, isolated sEVs bearing exosomal marker CD63 (putative exosomes) were assessed for their expression of donor HLA (Figure 2A

| Circulating sEVs bearing donor HLA are not detected in recipients undergoing allograft dysfunction due to rejection at late time points posttransplant
To determine whether donor sEVs release can be detected in circumstances of liver allograft dysfunction taking place beyond the   Table S1).

F I G U R E 2
Circulating donor-derived CD63 + small extracellular vesicles (sEVs) bearing PD-L1 peak early following liver transplantation (LTx). Advanced imaging flow cytometry was used for the multiparametric analysis of sEVs using Amnis ImageStream x (ISx). (A) Polystyrene calibration beads of known size demonstrate the capacity for small particle acquisition by ISx. Speed beads and their aggregates serve as internal calibrators for the cytometer and

| sEVs derived from liver but not kidney recipients transiently inhibit allogenic T cell responses
We sought to investigate the extent to which cross-dressing influences human APC immunogenicity, both in healthy individuals and in the setting of clinical transplantation. First, we employed HLA-A2 negative monocyte-derived dendritic cells (DCs) from healthy individuals that had been cross-dressed in vitro with plasma-derived HLA-A2 positive sEVs harvested from an allogeneic healthy control ( Figure 4A,B).
Following culture with allogeneic sEVs, DCs displayed HLA-A2 expression on their surface and acquired the capacity to elicit proliferation of syngeneic CD8 + T cells ( Figure 4C). Next, we conducted experiments in which monocyte-derived DCs from healthy controls were cultured for 24 hours with plasma-derived sEVs harvested from allogeneic liver (n = 7) and kidney (n = 3) recipients on days 1 and 10 posttransplant. DCs were then washed and replated for 5 days with allogeneic third-party CD8 T cells. sEVs isolated on day 2 post-livertransplantation ( Figure 5A), but not those harvested from day 1 kidney recipients ( Figure 5B), significantly inhibited T cell proliferation (7.4% ± 3.1% vs 22.4% ± 2.4% as compared to non-cross-dressed DCs; P < .001; Figure 5A). The inhibition of CD8 T cell proliferation observed when employing day 1 liver transplant recipient sEVs was reduced when DCs were cultured with day 10 sEVs, although the effect was still significant in comparison with non-cross-dressed DCS ( Figure 5A). In contrast, circulating sEVs from kidney transplant recipients did not induce significant changes in third-party CD8 T cell responses ( Figure 5B). To investigate the effect of sEVs isolated from liver transplant patients on DCs, we analyzed the changes in the expression of costimulatory molecules (CD40 and CD86) in DCs cultured with sEVs for 24 hours. DCs cultured with sEVs isolated from liver transplant recipients at day 1 following transplantation expressed significantly less CD40 than DCs cultured with sEVs at day 10 (P = .005), or DCs without sEVs. This was not observed in kidney transplant recipients ( Figure S3A). In addition, DCs cultured with sEVs from liver transplant recipients produced less IL-6 and more IL-10 when compared to the use of sEVs from kidney transplant recipients ( Figure S3B). Altogether, these data suggest that sEVs from liver transplant patients have a heightened capacity to induce an inhibitory phenotype in DCs.  The mechanistic basis for APC cross-dressing is an area of ongoing

D I SCLOS U R E
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.