Recent studies have revealed novel molecular mechanisms by which innate monocytic cells acutely recognize and respond to alloantigen with significance to allograft rejection and tolerance. What remains unclear is the single-cell heterogeneity of the innate alloresponse, particularly the contribution of dendritic cell (DC) subsets. To investigate the response of these cells to exposure of alloantigen, C57BL/6J mice were administered live allogenic BALB/cJ splenic murine cells versus isogenic cells. In parallel, we infused apoptotic allogenic and isogenic cells, which have been reported to modulate immunity. Forty-eight hours after injection, recipient spleens were harvested, enriched for DCs, and subjected to single-cell mRNA sequencing. Injection of live cells induced a greater transcriptional change across DC subsets compared with apoptotic cells. In the setting of live cell infusion, type 2 conventional DCs (cDC2s) were most transcriptionally responsive with a Ccr2+ cDC2 subcluster uniquely responding to the presence of alloantigen compared with the isogenic control. In vitro experimentation confirmed unique activation of CCR2+ cDC2s following alloantigen exposure. Candidate receptors of allorecognition in other innate populations were interrogated and A type paired Ig-like receptors were found to be increased in the cDC2 population following alloexposure. These results illuminate previously unclear distinctions between therapeutic infusions of live versus apoptotic allogenic cells and suggest a role for cDC2s in innate allorecognition. More critically, these studies allow for future interrogation of the transcriptional response of immune cells in the setting of alloantigen exposure in vivo, encouraging assessment of novel pathways and previously unexamined receptors in this setting.
Experimental models of transplantation have identified and developed strategies of donor-specific induction of immunological tolerance mediated by the infusion of donor cellular Ag. In one such protocol, recipients receive an infusion of apoptotic donor cells 7 d before and 1 d after allogeneic pancreatic islet cell transplantation, resulting in indefinite graft survival (1, 2), which has since been successfully translated to a nonhuman primate model (3). Additional murine studies of skin and heart transplantation also reveal a prolongation of graft survival with apoptotic donor cell treatment (4–6). Alternative protocols use the infusion of “live” donor cells in combination with costimulation blockade, namely anti-CD40L, at the time of transplantation. This protocol has been repeatedly shown to produce a long-term and robust form of tolerance to murine islet (7, 8), skin (9), and cardiac allografts (7, 10, 11), as well as nonhuman primate models of skin (12) and kidney (13) transplantation. Yet, the molecular response to the aforementioned donor cell alloinfusion remains poorly understood in contexts of both tolerance induction and basic cellular physiology of immune cells.
Although the alloresponse has long been attributed to the adaptive immune system and, more specifically, T cells and their ability to recognize alterations in peptide-MHCs (14, 15), recent work has challenged this dogma. Innate monocytic cells have been shown to mount an alloimmune response in the absence of T, B, and NK cells through “nonself” recognition (16), which is required to induce an optimal graft rejection response (17). These findings causes us to wonder whether similar mechanisms of innate recognition could be co-opted for the promotion of tolerance. The means by which monocytic innate sensing occurs is only just beginning to be elucidated. Investigations have already identified multiple receptor/ligand pairs, including SIRPα/CD47 (18) and type A paired Ig-like receptor (PIR-A) with MHC class I (MHC I) (19), as playing an integral role in the acute and memory alloresponse of this population of innate monocytes.
It is unclear whether other cells of the myeloid immune system are able to innately discriminate between “self” and nonself, especially as it relates to conventional dendritic cells (cDCs). cDCs serve as professional APCs able to stimulate an efficient proliferative or immunomodulatory T cell response through TCR–MHC engagement in combination with a costimulatory or tolerogenic cDC signal (20). Yet, whether and how cDCs are able to identify the presence of nonself Ag cargo to upregulate costimulatory or immunosuppressive molecules remain open questions. Thus, in the setting of current studies by our group and others during immune tolerance (4) and recent findings implicating an innate immune response to the allograft (16–19), our primary research objective in the present study was to, in an unbiased fashion, identify specific transcriptional response signatures of cDC populations after exposure to infusion of donor alloantigen. The experimental controls of this question afforded additional broader opportunities to also examine how apoptotic cells in general are immunomodulatory.
Materials and Methods
C57BL/6J (B6) mice were used as wild-type controls and were bred in the Northwestern Center for Comparative Medicine. BALB/cJ (Balbc) mice were purchased from The Jackson Laboratory (strain no. 000651). Mice were maintained in a pathogen-free temperature- and humidity-controlled environment and kept on a 12-h light/12-h dark cycle. All mice used for experiments were between 2 and 3 mo of age. Animal studies were conducted in accordance with an Institutional Animal Care and Use Committee–approved protocol at Northwestern University.
For all infusion experiments, B6 mice received retro-orbital injections of 5 × 107 cells taken from full splenic extracts of Balbc or B6 mice that were washed, RBC lysed, twice filtered, and resuspended in 100 μl of PBS for infusion. In experimentation requiring apoptotic cells, splenic cells were exposed to UV radiation for 8 min followed by a 2-h incubation at 37°C before use, which has been shown to induce early apoptosis (21–23).
PKH-labeled splenic cell infusion
Spleens of Balbc mice were harvested as described in Splenocyte infusion. Prior to infusion of B6 recipient mice, cells were labeled with PKH67 green florescence utilizing a general cell membrane labeling kit (MilliporeSigma). Spleens of infusion recipient mice were then harvested and processed for flow cytometry at respective time points.
Splenic harvest and DC enrichment for single-cell sequencing
Spleens of recipient mice were harvested 48 h after cellular infusion (three biological replicates per infusion condition) and subjected to collagenase D (300 U/ml) and DNase I (30 U/ml) digestion in HBSS at room temperature for 20 min, followed by the addition of 1 mM EDTA for 5 min. Spleens were then mechanically pressed through a 40-μm filter, pooled by condition, RBC lysed, and filtered again. Total viable cell numbers were determined by trypan blue staining. Cells were resuspended at 108 million cells/ml in EasySep buffer (STEMCELL Technologies). Each condition was separated into two distinct enrichment procedures and were independently subjected to CD45+ or pan-DC magnetic enrichment kit protocols (STEMCELL Technologies) according to the manufacturer’s instructions. Enriched cell suspensions were counted and repooled by condition at a 1:1 ratio for single-cell library preparation. Splenic cell harvest was optimized such that time from mouse euthanasia to single-cell library preparation was minimized and cells were maintained on ice throughout processing.
Single-cell library preparation and sequencing
Single-cell mRNA sequencing libraries were prepared utilizing the 10x Genomics Chromium Next GEM single-cell 3′ library and gel bead kit (v3.1) pipeline following the manufacturer’s protocols. A targeted number of 9000 cells from the DC-enriched cell suspension were loaded per experimental condition. RNA quality was confirmed utilizing Northwestern’s NUSeq Core Facility Agilent Bioanalyzer high-sensitivity chip and KAPA library quantification kits for the Illumina platform (KAPA Biosystems). Libraries were sequenced on the HiSeq platform (Illumina) to a read depth of ∼25,000 reads per cell by Novogene.
Single-cell RNA sequencing analysis and visualization
Raw fastq files were analyzed using Cell Ranger version 4.0.0 (10x Genomics) (24). Barcode–gene matrices from the Cell Ranger pipeline were analyzed using the Seurat R package (v4.3.0) (25). Low-quality cells were removed when they expressed <500 unique molecular identifiers, <250 genes, or had a ratio of mitochondrial reads >0.2. After quality control, a total of 18,483 individual cells across the five conditions were available for downstream bioinformatics analysis. Cell cycle heterogeneity was evaluated by calculation of cell cycle phase scores based on known markers and was regressed during data preprocessing (26). The number of analyzed cells per condition are as follows: control (C), 1131; apoptotic alloantigen (AA), 793; apoptotic isoantigen (AI), 1056; live alloantigen (LA), 536; live isoantigen (LI), 1257. Normalization and variance stabilization were performed utilizing sctransform (v0.3.5) (27). Single-cell datasets across all conditions were subsequently integrated utilizing cross-dataset pairs of biological state “anchors” to minimize batch effects and allow for comparative single-cell RNA sequencing (scRNA-seq) analysis of immune cell clusters across conditions (28). In this methodology, each condition (C, AA, AI, LA, LI) is viewed as a distinct experiment producing datasets that contain subsets of cells with shared biological states that are referred to as anchors. To identify such anchor cells, dimensionality reduction and mutual nearest neighbor identification across each dataset are performed across conditions, and canonical correlation analysis identifies common sources of variation between the datasets. This strategy thus allows for accurate comparison of cell clusters between experimental conditions. Following integration, principal component analysis was performed to reduce dimensionality, and subsequent applications employed a dimension of 40 principal components when necessary. Uniform manifold approximation and projection (UMAP) was used to visualize and unbiasedly cluster cells. Identification of clusters was initially performed using the unbiased SingleR algorithm, which compares experimental data to reference transcriptomic datasets of pure cell populations (29). SingleR identifications were confirmed through assessment of canonical immune cell markers that were conserved across conditions. Differential expression tests to identify differentially expressed genes (DEGs) were performed utilizing R package DESeq2 (v1.38.2) (30).
Pathway enrichment analysis and trajectory inference
DEGs with adjusted p values <0.1 were input into g:Profiler to identify enriched molecular pathways within specific experimental processes. Gene sets from Gene Ontology biological process were used (31).
Splenic flow cytometry
Spleens were harvested, digested, RBC lysed, filtered, and counted as previously described in Splenic harvest and DC enrichment for single-cell sequencing. Cells were then incubated in the dark, on ice, with Zombie Aqua fixable viability live/dead staining (BioLegend) for 15 min, Fc Block (BioLegend) for 15 min, and labeled with fluorescently conjugated Abs for 30 min. When necessary, cells were fixed in 1% perfluoroalkoxy alkane and stored overnight at 4°C. Flow cytometry was performed on a BD LSRFortessa X-20 cell analyzer and data were analyzed with FlowJo software. cDCs were identified as MHC class II (MHC II)highCD11chigh cells and further delineated into subsets whereby type 1 cDCs (cDC1s) were identified as XCR1highCD172low and type 2 cDCs (cDC2s) as XCR1lowCD172high. Flow cytometry Abs are listed in Supplemental Table I.
Bone marrow–derived DC coculture
Bone marrow–derived DCs were cultured as previously described (32). In brief, bone marrow cells from femur and tibia of B6 mice were cultured in petri dishes of RPMI 1640 cell media containing 20% FBS, 1% penicillin-streptomycin, and 1% l-glutamine supplemented with 200 ng/ml recombinant Flt3. On days 4–5 of culture, a half media change was performed. Experimentation utilizing in vitro DCs occurred on days 7–9 of culture. For coculture, allogenic (Balbc) or isogenic (B6) splenic cells were isolated from spleens as previously described and added to Flt3 DC cultures at a 1:1 ratio. Cells were cocultured for 48 h before harvest and analysis by flow cytometry.
Statistical analyses were performed with GraphPad Prism (GraphPad Software, La Jolla, CA). Comparisons between two groups were performed using a two-tailed, unpaired t test with 95% confidence interval. For comparisons of more than two variables, ANOVA was used with a 95% confidence interval. Data are presented as mean ± SEM.
The datasets generated and analyzed for this study have been uploaded into National Center for Biotechnology Information Gene Expression Omnibus under the number GSE223921 (https://www.ncbi.nlm.nih.gov/geo/).
Canonical splenic DC subsets with transcriptional heterogeneity are observed by single-cell mRNA sequencing following exposure to alloantigen
To determine how allogenic donor cell infusion reprograms the acute immune response, we used B6 mice as recipients of the following i.v. infusion conditions: saline (C), AI cells, LI cells, AA cells, and LA cells. Comparisons between live and apoptotic cells were included in consideration of prior transplantation studies that infused either cell type (1–3, 7, 10, 11). All i.v. infused cells were obtained from full splenic extracts of either isogenic B6 or allogenic Balbc mice and therefore consist predominantly of lymphocytes where previous reports identify B and T cells to make up 60 and 30% of murine splenic cells, respectively (33). Forty-eight hours after infusion, spleens of recipient mice were subjected to CD45+ or DC enrichment and scRNA-seq using the 10× Genomics platform (Fig. 1A). This time point was chosen to allow sufficient exposure between host and donor cells for transcriptional responses while simultaneously avoiding confounding results by inadvertently sequencing infused donor cells. This was confirmed through a time course evaluation of i.v. infused membrane-labeled allogenic cells in which <0.5% of donor cells were still present in the spleen at 48 h (Supplemental Fig. 1). Relatedly, we previously determined internalized apoptotic RNA dissipates approximately within 6 h (21). Following filtering and quality control of cells subjected to scRNA-seq (Supplemental Fig. 2), 18,483 individual cells across the five conditions were available for downstream analysis. We performed data integration of the infusion conditions and used UMAP dimensionality reduction analysis to cluster cells. We identified canonical cellular markers that were conserved across conditions to determine cluster identity (Fig. 1B, 1C). We further confirmed appropriate cluster identification using an unbiased assignment algorithm, SingleR (29), in combination with an open-source reference database, ImmGen (34) (Fig. 1D).
Higher resolution clustering of DC-identified clusters from Fig. 1C revealed two cDC1, four cDC2, and one plasmacytoid DC cluster (Fig. 2A–C). Expression of canonical DC marker Itgax (CD11c) and transcription factor Flt3, required for homeostatic DC development and maintenance in the spleen (35), were used to confirm DC identity (Fig. 2B). Expression levels of DC subset–defining markers including Xcr1, Sirpα, and Siglech (36, 37) were assessed to determine cDC1, cDC2, or plasmacytoid DC identity, respectively (Fig. 2B). Marker genes expressed across all conditions were established for individual cDC clusters, and transcripts found to be differentially elevated in individual clusters were used to further define cDC1 and cDC2 populations. Thus, cDC1 clusters were identified by their relative expression of cystatin C (Cst3), an inhibitor of cathepsin S, which plays a role in MHC class II Ag presentation whereby increases in cystatin C yields decreases in MHC II surface expression (38, 39) (Fig. 2B, 2C).
cDC2 clusters were similarly identified by their differential expression of conserved marker genes Socs2, Ffar2, Ccr2, and Lars2 (Fig. 2B, 2C). Notably, all cDC2 clusters showed minimal expression of Xcr1 and elevated levels of Sirpa; however, note that the cluster denoted “Lars2+ cDC2” was small in number across all conditions with few cluster-defining genes aside from Lars2, a mitochondrial leucyl-tRNA synthetase, and calcium-binding proteins S100a4 and S100a6. It is unclear whether this cluster is a unique population of cDC2s or more simply a polarization state of cDC2s. However, given their unique expression of Lars2 compared with other cDC2 clusters or cycling cells, we chose to annotate them as shown. We also assessed whether infusion conditions affected the proportion of splenic DC clusters present. Saline control and apoptotic cell infusions resulted in very similar proportions of all Seurat-identified DC subset clusters present. However, slight differences can be seen in the proportions of splenic DC populations sequenced following live cell infusion (Fig. 2D). Infusion of LA cells appears to result in an increase in the proportion of cycling cells, which is a cluster that includes DCs from all Seurat-identified clusters expressing increased markers of cellular proliferation including Top2a, Pclaf, and Mki67. This could be indicative of an increased immunological response following alloantigen stimulation, although this result was not further evaluated.
Ccr2+ cDC2 cluster displays transcriptional response to cellular infusion of live allogenic cells
To identify the presence of DC populations that respond to allogenic Ag, we performed differential gene expression analysis following LA cell infusion compared with saline (C) infusion utilizing DESeq2. The DESeq2 package takes into account variability between replicates (when available) in addition to applying shrinkage of log fold change data when gene counts are low and thus data are less stable. As a result, DEG identification is focused on strength of gene expression and allows for increased interpretability of results (30). Comparisons of gene expression between each experimental condition and control (C versus LA, C versus LI, C versus AA, C versus AI) in addition to isogenic versus allogenic (LI versus LA, AA versus AI) were performed within each identified DC cluster seen in Fig. 2A. We quantified the number of DEGs with an adjusted p value <0.10 for each infusion condition by sub cluster. We then identified which genes were unique to certain infusion conditions or shared, likely as the result of similar cellular processes. By utilizing the total number of DEGs identified for a single DC cluster while comparing each experimental condition to the control, we then calculated the percentage of unique DEGs expressed within each cell cluster following the different experimental condition. This analysis revealed increased percentages of unique DEGs within cDC2s compared with cDC1s following LA infusion (Fig. 3A). The Ccr2+ cDC2 subcluster displayed the largest percent change of unique DEGs following LA infusion as well as a higher number of unique DEGs following live isogenic infusion compared with both apoptotic cell infusion conditions (Fig. 3A, 3C). The number of analyzed Ccr2+ cDC2s in each condition are shown in Supplemental Fig. 3. We then began to look closer at the identity of Ccr2+ cDC2 DEGs following LA infusion (Fig. 3B). To understand the potential influence of these unique DEGs found in Ccr2+ cDC2s after LA infusion, we performed pathway analysis by g:Profiler (31) and discovered multiple pathways indicative of an immune response, including “Ag processing and presentation,” “regulation of innate immune response,” and “T cell–mediated cytotoxicity,” as well as “type I IFN–mediated signaling” (Fig. 3D). It was intriguing that the second highest percentage of unique DEGs within the Ccr2+ cDC2 subset occurred following live isogenic infusion (Fig. 3C). In evaluating these DEGs both individually and by pathway analysis, we discovered that they were predominantly related to translation and nonspecific biosynthetic processes (Supplemental Fig. 3).
In seeking to further evaluate the response of the Ccr2+ cDC2 subset following cellular infusion, we assessed the expression of a variety of known markers for DC activation across all experimental conditions (Fig. 3E). Interestingly, Ccr2+ cDC2s appeared to have the highest expression levels of Cd40 and Cd86 transcripts following LA infusion, which was not observed following either allogenic or isogenic apoptotic cell infusion. Additionally, genes associated with Ag presentation by MHC I molecules, including B2m and H2-D1, were also more highly expressed following live cell infusion. Similar patterns of expression were seen in genes that play roles in cell migration and motility (Pfn1) (40), immune cell metabolism (Cox7c) (41), and regulation of DC activation (Irf7) (42) (Fig. 3E).
CCR2+ cDC2s uniquely upregulate markers of activation following exposure to alloantigen compared with isoantigen
Given the observed activation response transcriptionally within the Ccr2+ cDC2 subset in the presence of alloantigen, we sought to identify whether such a response could be observed at the protein level. We chose to use an isolated and controlled in vitro experimental model that would minimize effects of other immune cell populations. Culture of bone marrow–derived cells in the presence of Flt3 ligand has been shown to yield high proportions of DCs that are biologically similar to those found in vivo (32). Thus, following derivation of Flt3 DCs from B6 bone marrow, DCs were cocultured with allogenic (Balbc) cells, isogenic (B6) splenic cells, or media controls for 48 h and harvested for flow cytometric analysis (Fig. 4A). Because splenic cells added to culture were congenic (CD45.1+), these cells could be excluded from analysis (Fig. 4B). A population of CCR2+ cDC2s was identified and these cells were subsequently assessed for expression of canonical DC activation markers across all conditions. We observed a significant increase in the expression of CD40+ and IL-12+ following allogenic coculture compared with both media control and isogenic coculture within the CCR2+ cDC2 population (Fig. 4C, 4D). These findings are in agreement with the transcriptional response described previously (Fig. 3E), although we did not observe an increase in CD86 surface expression as was predicted transcriptionally.
Given this unique transcriptional and protein level response of CCR2+ cDC2s in the setting of alloantigen exposure, we began to question whether CCR2+ cDC2s express any receptors known to function in innate immune sensing and allorecognition. PIR-As, as receptors to MHC I, have been identified as critical for innate memory by macrophages and monocytes to allogenic cells (19). Whether PIR-As are present on other innate immune cells or function in a similar manner is, to our knowledge, unknown. PIR molecules exist as a family of receptors whereby engagement of MHC I with PIR-A provides an activating signal, whereas engagement with type B PIR (PIR-B) results in an inhibitory signal (43). Interestingly, Ccr2+ cDC2s displayed elevated gene expression of Pira2, which encodes PIR-A, following both apoptotic and live allogenic infusion conditions, which was not reciprocated following saline control or isogenic infusions. Pirb gene expression, which encodes PIR-B, remained consistent across all experimental settings (Fig. 4E). Expression of Pira2 was also observed in other single-cell sequencing (scSEQ)–identified cDC2 clusters following LA infusion, specifically Ffar2+ cDC2s and Socs2+ cDC2s. Interestingly only Ffar2+ cDC2s were seen to express Pira2 within the control condition (data not shown).
We next sought to validate these scSEQ findings indicating increases in Pira mRNA after LA infusion at the protein level, specifically within the cDC2 population. Thus, we infused live allogenic cells compared with saline control in B6 mice and after 72 h performed flow cytometry of recipient splenic cells. Within the cDC2 population we identified a significant increase in PIR-A/B mean fluorescence intensity (MFI) following LA infusion (Fig. 4F). These data indicate that PIRs may be one such receptor by which cDC2s and, more specifically, the CCR2+ cDC2 population are able to recognize and respond to alloantigen.
Through use of scRNA-seq technology, our findings shed (to our knowledge) new light on the innate immune response to alloantigen at a granularity not previously appreciated in more traditional bulk mRNA sequencing approaches. Such findings are especially relevant in the context of current experimental transplantation tolerance strategies that frequently employ infusion of donor-derived cells for tolerance induction (1, 3, 5, 8, 12). Additionally, recent elegant studies have identified key cell and molecular mechanisms by which innate myeloid cells, namely macrophages and monocytes, recognize alloantigen (16–19). Although our study focuses on and newly implicates cDCs, specifically the heterogeneous cDC2 population, as unique innate responders to alloantigen, this work will also allow for interrogation of the alloantigen transcriptional response of other innate cellular populations in vivo that can be easily identified within the larger scSEQ dataset generated.
As professional APCs within the immune system, cDCs are already known to be a critical player in promoting allograft rejection (44) or acceptance (2). However, whether an early innate response by this specific cell population mediated by allorecognition contributes to downstream graft outcome is unknown. Given the finding by Lakkis and colleagues (17) that such nonself recognition by monocytes in cardiac allografts initiates transplant rejection, further exploration of this phenomenon in other innate populations such as cDCs is warranted. Our analysis uncovered a transcriptionally heterogeneous population of cDC2s at both steady state and after infusion conditions, as has been previously reported (45). Some cDC2-identified clusters expressed higher levels of genes typically associated with monocytes, including Csf1r and Ccr2. However, analysis of Flt3 (35) and Zbtb46 (46) expression, known to distinguish cDCs from monocytes and macrophages, in addition to flow cytometric experimentation was able to confirm cDC identity of this CCR2-expressing MHC II and CD11chigh alloresponsive population.
Our in vitro coculture experimentation offers a controlled and easily modifiable setting to assess the specific interactions of cDCs with allogenic or isogenic immune cells that can be identified through the use of congenic markers, namely CD45.1 and flow cytometric Abs. The increase in CD40 and IL-12 of CCR2+ cDC2s that is more profound after allogenic coculture compared with isogenic or control conditions seems to suggest a distinct ability of this cellular population to recognize and respond to nonself Ag. Relatedly, the lack of increase in some activation markers such as CD86 within this population may point to differential expression of cDC activation markers across subsets in this setting, as has been previously reported (47). Our scSEQ experimentation revealed upregulation of a known allorecognition receptor, PIR-A, within a Ccr2+ cDC2 subcluster following exposure to alloantigen but not isoantigen at the transcriptional level in vivo. Flow cytometric analysis is likewise suggestive of a similar increase in PIR-A at the protein level in the greater cDC2 population. Notably, PIR-A was tested as one previously identified candidate receptor known to play a role in allorecognition, but it is entirely possible this is not the primary mechanism by which this cDC2 population is transcriptionally responsive following alloantigen exposure.
Future studies are required to test the functional significance of these findings. It is noteworthy that there are certain limitations to our scSEQ experimentation, namely that we chose to perform the infusions in fully immunocompetent mice. Murine models lacking T and B cells, such as SCID or RAG−/−, have been reported to contain fewer DCs with varying functional deficits (48, 49). Therefore, to understand the transcriptional response of physiologically relevant DCs, a wild-type B6 mouse was used as the infusion recipient. As such, we cannot exclude the possibility of signaling that may occur between innate DCs and other adaptive immune cells in our reported results. However, our findings serve as intriguing evidence to prompt additional investigation of the functional role of the cDC2 population in the context of allorecognition during transplantation.
In some tolerogenic transplantation strategies, infusion of donor apoptotic cells has been used with significant effect (1, 3). It is well established that ingestion of apoptotic cells by phagocytes results in a potent immunologic response that is regulatory in nature in numerous biological states, including myocardial infarction (50), chronic inflammatory lung disease (51), and induction of tolerance (52). Surprisingly, our findings did not reveal significant differences in the transcriptome of DCs exposed to an apoptotic cellular infusion. However, our study does not rule out whether such transcriptional responses were induced in DCs at alternative time points or in other locations, such as peripheral lymph nodes or the liver. We were likewise surprised to see higher proportions of unique DEGs across DC subsets following live isogenic infusion given the paradigm of this experimental control is the infusion of additional self immune cells. However, it has been shown that adoptive transfer of 104–105 isogenic TCR transgenic T cells is able to modify the endogenous immune response (53). Additionally, murine models of transplantation where donor organs differ in only three amino acids in the β-chain of MHC class II molecules will result in chronic allograft rejection in the absence of immunosuppression (54, 55). It is therefore not unreasonable to hypothesize that minor differences between donor and recipient B6 mice who received the live isogenic cell infusion were sufficient to induce transcriptional differences within the analyzed DC populations but not strong enough to be recognized by host immune cells for subsequent removal.
Additional studies are necessary to further identify the role of innate allorecognition by cDC2s in the immune response to a foreign allograft. PIR-A serves as one probable receptor to be involved in such nonself recognition by cDC2s, although additional candidate molecules and alternative mechanistic pathways should be considered and investigated, as conserved and redundant mechanisms are likely. Utilization of newly developed methodologies to track cellular interactions in vivo with intercellular enzymatic labeling, also known as LIPSTIC (56), could be employed through creation of reporter and enzymatic constructs of allorecognition receptor/ligand pairs to identify individual cells that have engaged these receptors and their subsequent transcriptional programming and immunologic response. One could imagine that such detailed knowledge may provide opportunities to circumvent pathways of alloreactivity and immune activation through targeted delivery of receptor blocking Abs that would “mask” the foreign organ from the host innate immune system, delaying or mitigating rejection episodes in their entirety. Thus, our studies contribute to an important body of work seeking to understand basic cellular physiology of an alloresponse that plays a critical role in the survival of a transplanted organ.
The authors have no financial conflicts of interest.
We thank Connor Lantz and Mallory Filipp for helpful bioinformatic discussions, Dr. Mathew DeBerge and Dr. Kristofor Glinton for experimental consultations, Dr. Xunrong Luo and Dr. Maria-Luisa Alegre for invaluable theoretical and practical knowledge, and members of the Thorp Laboratory for related conversations and feedback.
This work was supported by the National Institutes of Health Grants 1F30HL162456-01A1 and 1T32GM144295 and an American Heart Association predoctoral fellowship (903851) (to S.L.S.). E.B.T. is supported by National Institutes of Health Grant R01HL139812. This study was also supported by the Sidney and Bess Eisenberg Memorial Endowment.
The online version of this article contains supplemental material.
The datasets presented in this article have been submitted to the Gene Expression Omnibus repository (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE223921) under accession number GSE223921.
type 1 cDC
type 2 cDC
differentially expressed gene
mean fluorescence intensity
- MHC I
MHC class I
- MHC II
MHC class II
type A paired Ig-like receptor
type B PIR
single-cell RNA sequencing
uniform manifold approximation and projection