Arthritogenic SKG T cells have a transcriptional program of activation and a repertoire pruned by superantigen

Purpose: The goals of this study were to study the transcriptome and TCR repertoire of naive CD4 T cells before the induction of arthritis in the SKG and wild-type (WT) mouse with eGFP-Nur77 as a reporter of antigen stimulation (SKGNur and WTNur). Methods: Bulk RNA sequencing was used to generate the mRNA profiles for the top 10% and bottom 10% of GFP expressing cells from three SKGNur and three WTNur BALB/c mice resulting in 12 samples overall (3 WT Low, 3 WT High, 3 SKG Low, 3 SKG High) using the Truseq Stranded mRNA kit followed by sequencing on a HiSeq 2500 sequencer. The sequence reads that passed quality filters were aligned with STAR and then analyzed with DESeq2. Results: We mapped about 38,000,000 million reads per sample to the mouse genome (mm10). The differential expression pairwise comparisons (WT Low v WT High, WT Low v SKG Low, SKG Low v SKG High, and WT High v SKG High) generated a collection of 991 differentially expressed genes with log2FC >1 and Benjamini-Hochberg adjusted p value < 0.05. Hierarchical clustering of differentially expressed genes uncovered six modules of gene expression which highlighted heterogeneity within the naive CD4 T cell compartment and transcriptional signatures associated with the SKG High samples. Methods: Single cell RNA gene expression and TCR sequencing libraries were generated for the top 10% and bottom 10% of GFP expressing cells from two SKGNur and two WTNur BALB/c mice resulting in 8 samples overall (2 WT Low, 2 WT High, 2 SKG Low, 2 SKG High) using the 5' 10x single cell 5"+V(D)J v1 kit and were sequenced on a NovaSeq 6000 sequencer. The reads were aligned using cellranger and analyzed using a combination of single cell tools including scanpy and scvelo. Results: We mapped about 785,000,000 sequence reads and about 379,000,000 sequence reads per sample for the gene expression and TCR libraries to the mouse genome (build mm10 with the addition of the eGFP transcript) and TCR reference (vdj GRCm38 v 3.1.0), respectively, using cellranger. We identified 31,054 transcripts across 101,869 cells and recovered paired TRA and TRB chains from more than 80% of cells. After filtering cells and genes by our quality control metrics, we identified 1119 highly variable genes which were used for PCA analysis to create a nearest neighbors graph for leiden clustering and UMAP projection. We identified 13 clusters that we collapsed into 9 cell sub-types. Within the cluster (T.4_Nr4a1) exclusively expressing our marker gene of interest, Nr4a1, we found that SKG High cells upregulated TCR signaling genes despite their impaired impaired TCR signaling ability due to a Zap70 mutation. Within the T.4Nr4a1 cluster we found two states of acute to chronic TCR signaling, marked by Egr2 and Tnfrsf9, and we used trajectory inference to uncover a continuum of cell states with these acute to chronic TCR signaling states as the endpoints. Interestingly, SKG High cells in the T.4 Nr4a1 cluster seemed arrested in the earlier more acute TCR signaling state suggesting they are unable to appropriately fully upregulate the chronic TCR signaling and more anergic cell state. We also uncovered TRBV restriction within the SKG High samples and specifically within the SKG High cells within the T.4 Nr4a1 cluster that seems to be driven by a failure of negative selection by endogenously expressed super antigens from MMTVs present in the BALB/c mouse. Thus the naive CD4 T cell compartment within the SKG High cells are poised to enter into a reactive immune response via both a unique transcriptomic and TCR repertoire which may underlie the mechanism of high arthritogenic potential of these cells. Overall design: Bulk RNA seq and single cell RNA and TCR sequencing of naïve CD4 T cells before arthritis induction in Nur77-eGFP SKG and wild-type mice

研究目的：本研究旨在探究以增强绿色荧光蛋白-Nur77（eGFP-Nur77）作为抗原刺激报告基因的SKG小鼠（SKGNur）与野生型（WT）小鼠在关节炎诱导前，其初始CD4 T细胞（naive CD4 T cell）的转录组（transcriptome）与T细胞受体库（TCR repertoire）特征。

方法1：本研究选取3只SKGNur小鼠与3只WTNur BALB/c小鼠，通过Truseq链特异性mRNA建库试剂盒构建文库，对其体内绿色荧光蛋白（GFP，Green Fluorescent Protein）表达量最高的10%与最低的10%细胞进行批量RNA测序（bulk RNA sequencing），最终获得12个样本（3个WT Low组、3个WT High组、3个SKG Low组、3个SKG High组），随后使用HiSeq 2500测序仪完成测序。对通过质量过滤的测序reads使用STAR比对软件进行比对，再通过DESeq2软件进行差异表达分析。

结果1：每个样本约3800万条reads比对至小鼠基因组（mm10版本）。通过4组两两差异表达分析（WT Low vs WT High、WT Low vs SKG Low、SKG Low vs SKG High以及WT High vs SKG High），共筛选得到991个log2倍数变化（log2FC）>1且Benjamini-Hochberg校正P值<0.05的差异表达基因（differentially expressed genes）。对差异表达基因进行层次聚类（hierarchical clustering）分析，共识别出6个基因表达模块，该结果揭示了初始CD4 T细胞亚群的异质性，以及与SKG High样本相关的转录特征。

方法2：本研究选取2只SKGNur小鼠与2只WTNur BALB/c小鼠，使用10x Genomics 5'单细胞5"+V(D)J v1建库试剂盒，对其体内GFP表达量最高的10%与最低的10%细胞构建单细胞RNA基因表达与T细胞受体测序（TCR sequencing）文库，最终获得8个样本（2个WT Low组、2个WT High组、2个SKG Low组、2个SKG High组），随后使用NovaSeq 6000测序仪完成测序。测序reads通过Cell Ranger分析软件进行比对，结合Scanpy与scVelo等单细胞分析工具完成后续分析。

结果2：通过Cell Ranger软件，分别将基因表达文库与TCR测序文库的每条样本约7.85亿条与约3.79亿条reads比对至小鼠基因组（添加eGFP转录本的mm10版本）与TCR参考序列（vdj GRCm38 v3.1.0）。本研究在101869个细胞中共识别出31054个转录本，且在超过80%的细胞中成功获取了配对的TRA与TRB链。根据质量控制标准过滤细胞与基因后，共得到1119个高可变基因（highly variable genes），通过主成分分析（PCA，Principal Component Analysis）构建最近邻图，用于Leiden聚类与UMAP降维投影。最终识别出13个细胞簇，并将其整合为9个细胞亚型。在仅表达目标标记基因Nr4a1的T.4_Nr4a1细胞簇中，尽管SKG High细胞因Zap70突变存在TCR信号通路受损的情况，但其仍上调了TCR信号相关基因。在T.4Nr4a1细胞簇中，我们发现了以Egr2与Tnfrsf9为标记的急性至慢性TCR信号传导两种状态，并通过轨迹推断（trajectory inference）揭示了以这两种状态为端点的连续细胞状态谱系。值得注意的是，T.4_Nr4a1细胞簇中的SKG High细胞似乎停滞在更早的急性TCR信号传导状态，这表明其无法充分上调慢性TCR信号通路及更偏向无反应性细胞状态（anergic cell state）。本研究还在SKG High样本中，尤其是T.4_Nr4a1细胞簇内的SKG High细胞中，发现了TRBV基因取用受限的现象，该现象可能由BALB/c小鼠体内内源性表达的小鼠乳腺肿瘤病毒（MMTVs，Mouse Mammary Tumor Viruses）超抗原（super antigens）导致的阴性选择（negative selection）缺陷所驱动。综上，SKG High细胞群中的初始CD4 T细胞凭借独特的转录组与TCR库特征，处于触发适应性免疫应答的就绪状态，这或许是此类细胞具备高致关节炎潜能（arthritogenic potential）的分子机制。

实验设计：对Nur77-eGFP标记的SKG小鼠与野生型小鼠在关节炎诱导前的初始CD4 T细胞进行批量RNA测序、单细胞RNA测序及TCR测序。