hello，大家好，随着我们分析的深入，单细胞TCR和BCR的分析也要大量进入我们的视野了，之前也分享了一些有关TCR的文章，以后我们慢慢丰富这块的内容。关于TCR（BCR）分析其实有两个角度，一个是基因序列角度，一个是氨基酸序列角度，个人倾向于氨基酸序列角度，每个氨基酸的理化性质是固定的，更能从功能角度认识TCR与MHC的结合特性，做单细胞的同学不妨试一试。

先来学习一下基础知识

MHC（我们需要知道MHC的主要功能）

• 每个人的每个细胞表面都有一组特异的糖蛋白分子，简称MHC。它是由遗传决定的，不同人的MHC不同。各种哺乳动物都拥有主要组织相容性复合体（major histocompatibility complex, MHC)，人的MHC统称为HLA。

• 根据基因的位置和功能，主要组织相容性复合体分为三类，分别为MHC class I,MHC class II,MHC class III。

• MHC class I（MHC I）：位于一般细胞表面上，可以提供一般细胞内的一些状况，比如该细胞遭受病毒感染，则将病毒外膜碎片之氨基酸链（peptide）透过MHC提示在细胞外侧，可以供杀手CD8+ T细胞等辨识，以进行扑杀。

• MHC class Ⅱ（MHC Ⅱ）：大多位于抗原提呈细胞（APC)上，如巨噬细胞等。这类提供则是细胞外部的情况，像是组织中有细菌侵入，则巨噬细胞进行吞食后，把细菌碎片利用MHC提示给辅助T细胞，启动免疫反应。

• MHCIII类，位于MHCI和MHCII类基因区之间的一组基因。编码产物为某些血清补体成分，如人C2、Bf、C4A和C4B。其中我们主要关注的就是前两类。

TCR（BCR）的结构

图片.png

如图所示，BCR由两条重链和两条轻链连接而成，其中重链分为可变区(V区)、恒定区(C区)、跨膜区及胞质区；轻链则只有V区和C区。TCR每条肽链可分为V区、C区、跨膜区和胞质区。TCR分为两类，多数T细胞的TCR由α和β链组成, 少数TCR由γ和δ肽链组成。BCR和TCR有三个高变区CDR1、CDR2、CDR3，其中以CDR3变异最大，直定了抗原结合特异性，对于B细胞和T细胞多样性的研究也就集中在CDR3 区。

图片.png

如图所示，BCR的CDR3区包括轻链V、J基因片段和重链V、D、J基因片段，TCR的CDR3区包括α链V、J基因片段和和β链V、D、J基因片段。依据CDR3氨基酸序列的差异对BCR或TCR进行分析，可以揭示其clonotype分型、V-J基因丰度分析和相关免疫解析。

今天我们分享的文章是Repertoire analyses reveal TCR sequence features that influence T cell fate，里面有很多值得学习的地方，我们需要精读文献。

ABSTRACT

1、T cells acquire a regulatory phenotype when their T cell receptors (TCRs) experience an intermediate-high affinity interaction with a self-peptide presented on MHC.

2、Using TCR sequences from FACS-sorted human cells, we identified TCR features that shape affinity to these self-peptide-MHC complexes（TCR塑造MHC复合物的亲和力）.

3、CDR3β hydrophobicity（疏水性） and certain TRBV genes promote Treg fate（这一点还是很重要的，需要学习）。

4、found that within the tumor microenvironment clones exhibiting Treg-Tconv plasticity had higher TiRP than expanded clones maintaining the Tconv phenotype（肿瘤组织T细胞具有更高的调控潜能，我们后面会详细讲解）。

INTRODUCTION（这个地方对于我来说，知识量很大，免疫做的很少，需要补充）

1、During T cell development, regulatory T cells (Tregs) acquire their suppressive phenotype（抑制表型） when the affinity of their TCR to the peptide-MHC complex (pMHC) is intermediate high.

2、In most cases, randomly rearranged V, D, and J genes produce a TCR without enough affinity to pMHC, and so most developing T cells do not survive positive selection in the thymus（这一点我也是第一次知道，/(ㄒoㄒ)/~~）

3、On the other hand, TCRs with too strong affinity to pMHC result in apoptosis and negative selection for the expressing T cell（感觉有点像负反馈调节的意思）。

4、For the T cells that survive both positive and negative selection, however, a divergence in phenotype emerges: those whose TCRs have lower affinity to pMHC tend to become conventional T cells (Tconvs) and those whose TCRs have higher affinity tend to gain the Treg phenotype（胸腺保留了这两种T细胞类型，很智能的调控，Tconvs 和 Treg）。

5、Following thymic selection, a crucial prerequisite（先决条件） for the peripheral induction（外周感应，专业词汇真的多） of Tregs is suprathreshold affinity to pMHC, though other factors such as costimulatory signals exert additional influence.（这应该会介导我们的细胞免疫吧）。

6、The body of evidence that regulatory versus conventional T cell phenotypes are largely driven by TCR signal strength suggests that the developmental fate of CD4+ T cells may be influenced by sequence features of the TCR.（TCR信号强度驱动，这一点很关键，需要多多研究）。

7、Indeed, the degree of overlap in TCR sequence between Tregs and Tconvs is minimal compared to T cell samples of the same phenotype。（两种表型的TCR重叠程度很小）。

8、The distinguishing features of Treg and Tconv TCRs could shed light on the determinants of TCR strength, but the majority of extant work has focused on exact sequence matching rather than generalizable TCR sequence features.（但现有的大部分工作都集中在精确序列匹配上，而不是可概括的 TCR 序列特征，这一片的空间需要填补）。

接下来看看作者是怎么做的

图片.png

图注：(a) T cell receptor (TCR) β chain in complex with antigenic peptide（抗原肽） (red) and human MHC II molecules (brown). The TCR is colored by region: V-region (including CDR1β and CDR2β loops) in green, CDR3β middle region (CDR3βmr) in orange, and J-region in pink. We considered 798 candidate TCR sequence features and selected 612 nonredundant（非冗余） TCR features that best explained variance in T cell state. (b) To develop a TCR-intrinsic regulatory potential (TiRP) scoring system, we first split the discovery and replication cohorts into data for training,calibration, and testing. Each human figure represents an individuals’ TCR repertoire sample and is colored according to cohort. We fit logistic regression models for the discovery and replication cohorts separately, and combined the effect sizes for each TCR feature across the two cohorts via inverse variance-weighted meta-analysis (看来涉及到很深的算法知识，后面我们详细分解). We calibrated（校准） the P value threshold for including a TCR feature in TiRP based on held-out data from both cohorts. We then tested TiRP in held-out donors from both cohorts, as well as two independent cohorts of T cells sampled from the tumor microenvironment. (c) We then examined TiRP in mixed clones: groups of Tregs and Tconvs with the same TRB and TRA sequences observed in the same individual. These mixed clones likely represent lineages of T cells that have undergone peripheral conversions between the regulatory and conventional phenotypes, which include induced or iTregs (Tconv cells that have acquired a regulatory phenotype) and exTregs (Treg cells that have lost the regulatory phenotype). (d) We then investigated the drivers of TiRP by separately examining the two elements of the human Treg TCR ligand: the self-peptide and the human MHC II molecule.（原理看起来很难，我们需要慢慢来分解了）.

First, we derived a comprehensive collection of TCR features and tested them for differential abundance(差异丰度) between Tregs and Tconvs in two human cohorts of TCRβ chains from FACS-sorted T cells（对应上图a）。

第二步，From these results, we developed a Treg-propensity scoring system for the TCR (referred to as TCR-intrinsic regulatory potential or TiRP)（对应上图b，需要深刻的算法知识）。

第三步，Upon confirming its accuracy in two independent cohorts of T cells sampled from the tumor microenvironment, we used TiRP to examine Treg-Tconv plasticity of tumor-infiltrating clones（对应上图b，c,这个地方应该是我们的终极目标）。

第四步，Finally, to shed light on the etiology（病因） of the observed TCR sequence biases, we separately examined the two elements of the Treg TCR ligand: 1) the self-peptide and 2) the human MHC II molecule by calculating TiRP in 1) murine Tregs and 2) human memory Tconvs（对应上图d，感觉比较难）。

最后来一句总结，Our work reveals that CDR3β hydrophobicity（疏水性） promotes reactivity to self-peptides, while TRBV gene usage shapes the TCR’s general propensity（一般倾向） for activation in the context of human MHC II restriction.

Result

Defining independent features of the T cell receptor sequence（看一看标准）

1、The TCR is a membrane-anchored heterodimeric protein（膜锚定异二聚体蛋白） consisting of an α and a β chain。

2、Each of the two chains includes three highly variable peptide loops that protrude into the TCR-pMHC complex.（这个参考前面的TCR结构图）。

3、The most variable of these loops is the CDR3β region in the β chain which mediates recognition of specific antigens.

4、Because TRBV, TRBD, and TRBJ genes each encode regions of CDR3β, we anticipated that the CDR3β sequence would feature blocks of strongly correlated residues.（以此为特征最合适不过 ，很多分析都是测到CDR3序列）.

5、TCR的测序结构

图片.png

图注：(a) Probability of each amino acid in each CDR3β position depicted by a sequence logo, with a heatmap of normalized mutual information between each pair of CDR3β residues for the most frequent CDR3β length, 15 amino acids. Based on this mutual information structure, we partitioned the CDR3β sequence into a Vmotif within a V-region, a CDR3β middle region (CDR3βmr), and a Jmotif within a J-region. (b) Schematic showing TCRs of multiple lengths aligned to the TCR β chain structure. Three complementary-determining regions within the TCR β chain protrude as loops into the pMHC-TCR complex: CDR1β, CDR2β, and CDR3β. CDR1β and CDR2β are encoded by the TRBV gene, while CDR3β spans TRBV-encoded residues, random nucleotide insertion。

the V-region (IMGT position 1–107), CDR3β middle region、(CDR3βmr, p108–p112), and J-region (p113–p118)。虽然高度可变的 CDR3βmr 中的随机核苷酸插入掩盖了 TRBD 基因的身份，但种系编码的 V 和 J 区显示出序列保守性和高残基间互信息 。

6、Mutual information was concentrated at the flanking ends of CDR3β such that eight p104-p106 tripeptides (“Vmotifs”) and 42 p113-p118 pentapeptides (“Jmotifs”) accounted for >90% of observations.（这个区域最为关键,examine the V-, middle, and J- regions independently）.

结果2、Regulatory T cells use specific amino acids within the CDR3β middle region

1、We first examined the middle region of CDR3β (“CDR3βmr”) of Tregs (CD4+CD127-CD25+) and Tconvs (CD4+CD127+) in the discovery cohort.

2、计算每个氨基酸占据的 CDR3βmr 残基的平均百分比在供体之间产生了惊人一致的 Treg-Tconv 差异：苯丙氨酸 (F)、亮氨酸 (L)、色氨酸 (W) 和酪氨酸 (Y) 始终富含 Treg，而天冬氨酸(D) 和谷氨酸 (E) 始终富含 Tconv（研究很精细啊，值得深入学习）。

图片.png

3、Categorization of amino acids by physicochemical features（理化性质） showed that hydrophobic amino acids（疏水氨基酸） were enriched in Tregs, while negatively charged amino acids（带负电荷的氨基酸） were enriched in Tconvs.

4、To statistically assess these differences, we used nested conditional mixed effect logistic regression models(嵌套条件混合效应逻辑回归模型,后续方法我们会详细解释) which account for inter-individual differences such as those driven by HLA genotype and tissue source.

5、observed that 15 amino acids had an independent effect on Treg fate。To confirm that these effects were consistent across donors and clinical phenotypes, we estimated them in each of the 18 individuals and in T1D cases and controls separately 。 We found consistent effect sizes in all contexts（这个发现很有意思）.

6、然后运用该模型scores CDR3βmr 理化特征（之前是用于单个的氨基酸）,Physicochemical features did not capture as much information as amino acid percentages ，因此，继续研究基于氨基酸的 CDR3βmr 模型。

图片.png

7、We then ran a separate mixed effects model for each CDR3βmr position (IMGT p108 -112)（看来这个模型相当重要），testing whether the amino acid at the given position explained variance in T cell fate beyond that accounted for by the CDR3βmr amino acid percentages。每个位置确实传达了关于 Treg 命运可能性的额外信息，但这些位置特异性效应并不能解释像 CDR3βmr 的一般氨基酸组成一样多的变异

图片.png

注：Percent of explained variance by each TCR feature type,summing to 100% for each length of CDR3β. VGSR = V gene selection rate。CDR3βmr %AAs = percent composition of amino acids in the CDR3βmr。

结果3.CDR3β V and J regions explain variance in T cell state

1、We then examined the V-region of the TCR.先前的研究已经确定，MHC 基因座中的遗传变异决定了 TR(A/B)V 基因在repertoire中的使用频率 。

2、Interestingly, MHC polymorphisms explained far more variance in TRAV gene usage compared to TRBV，consistent with protein structure data demonstrating that TRAV contacts MHC at polymorphic sites while TRBV contacts MHC at conserved sites。（与蛋白质结构数据一致，表明TRAV在多态位点接触MHC，而TRBV在保守位点接触MHC。）假设 TRBV 编码残基的变异可能会改变 TCR 对这些保守 MHC 位点的亲和力，从而影响 T 细胞命运.

3、To test this hypothesis, we extracted sequence features from the V-region and tested their association with Treg fate using mixed effects logistic regression。通过模型比较，发现包含 TRBV 基因identity和 p107 的联合模型最能代表该区域，因为 58 个 TRBV 基因比 8 个 Vmotif 解释了更多的variance。

图片.png

注：Incremental variance explained by the addition of labeled TCR features to the V-region mixed effect logistic regression models.（通过将标记的 TCR 特征添加到 V 区域混合效应逻辑回归模型来解释增量方差。 ）

4、为了解释 TRBV 基因选择的个体间差异，我们为每个 TRBV 基因推导出胸腺选择参数 (VGSR) 作为协变量 .尽管控制了 VGSR，TRBV 基因identity继续解释了 T 细胞命运的显着差异，与参考基因 TRBV05-01 相比，三个 TRBV 基因将 Treg 命运的几率降低了 30% 以上 .(P = 1.3 x 10^-804,算法很深， 基础需要的也很多，理解起来有点费劲).

5、看一看V基因的结论，The consistency in TRBV gene effects across individuals suggests that their influence on Treg fate indeed occurs through interactions with conserved MHC residues, and is largely independent of MHC variability between individuals（个体间 TRBV 基因效应的一致性表明它们对 Treg 命运的影响确实是通过与保守的 MHC 残基相互作用而发生的，并且在很大程度上独立于个体之间的 MHC 变异性，具有普遍性）。

接下来是Jgene，We then examined the J-region.

1、然后我们检查了 J 区。 与 V 区相反，其中强 p104-p106 序列保守性将多个 TRBV 基因限制为相同的 Vmotif，D/J 连接处的可变核苷酸编辑导致与每个 TRBJ 基因相关的多个 Jmotif。

2、The 42 Jmotifs explained slightly more variance than the 13 TRBJ genes ，and so we proceeded with a joint model containing the Jmotif and p113 residue

图片.png

3、the most important TCR features for T cell fate determination were the TRBV gene identity and the percent composition of amino acids in the CDR3βmr。

图片.png

结果4、Treg TCRs are enriched for CDR1β apex positive charge and CDR3β middle region hydrophobicity（特征分析）.

1、We wanted to localize physicochemical effects underlying CDR3βmr residue enrichments to specific sequence positions.在CDR(1-3)β三个区域估计重要的理化性质、等电点、and volume in influencing Treg fate using a ridge regression model（又是一个模型，数学，难！！！）

2、Tregs were enriched for positively charged amino acids at p37 of CDR1β

图片.png

注：Estimated odds ratio (per standard deviation) for each physicochemical feature at each CDRβ(1-3) loop position; features with an estimate > 1 are positively associated with Treg fate while features with an estimate < 1 are negatively associated. Odds ratios denote the change in Treg odds per standard deviation increase in the given physicochemical feature at the given TCR position. For each CDR3β length, all effects were estimated jointly in an L2-regularized logistic regression with 10-fold cross-validation (Methods). Shown are the odds ratio estimates for each position-feature averaged across the six CDR3β lengths. Vertical lines denote the boundaries of each CDRβ loop

3、seven TRBV genes assessed in our models harbor a negatively charged residue at p37。all seven of these were significantly depleted for Tregs compared to the reference gene TRBV05-01, which has a positively charged Arginine (R) at p37

图片.png

4、CDR3βmr 在每个位置都具有正的疏水系数，At each of these positions, every standard deviation increase in hydrophobicity led to a 2.5% (L17,p113) – 6.3% (L12, p113) increase in odds of Treg fate。

图片.png

5、To directly visualize the amino acids associated with Treg fate, we generated sequence logo representation of the CDR3βmr based on differential amino acid usage at each position,hydrophobicity at p109 and p110 promotes the development of T cells that recognize self-antigens.

图片.png

6、this principle extends beyond p109-110 throughout the stretch of entropic CDR3β residues.

结果5、在独立数据集中再现 TCR 关联（上述模型在未知数据中的运用）

1、Despite a different distribution of tissue sources in this data set, the CDR3βmr amino acid percentage effects were nearly identical（下图a），Effects for individual TRBV genes, Jmotifs, and position-specific amino acid effects were also consistent with discovery（下图b）.

图片.png

注：Correspondence between the discovery and replication cohort odds ratios for all 612 nonredundant TCR features. Odds ratios for CDR3βmr percent composition amino acids are shown on the left; OR corresponds to the change in Treg odds associated with one standard deviation increase in the given CDR3βmr amino acid percentage. All other TCR features are shown on the right; OR corresponds to the change in Treg odds associated with the presence of the given TCR sequence feature compared to the reference feature.

结果6、Developing TiRP: a Treg-propensity scoring scheme for the TCR

1、for a given TCR, TiRP is the sum of Treg association effect sizes of independent sequence features in all three TCR regions（这个对单细胞数据不友好啊）。

2、tested TiRP score on the four discovery cohort donors and two replication，cohort donors whose repertoire data had been withheld from all former analyses。observed that a one standard deviation increase in TiRP in these held-out data resulted in a 23% increase in the odds of Treg status。

图片.png

注：Validation of TCR-intrinsic regulatory potential (TiRP) in held out donors of the discovery and replication dataset (n = 3,277,036 TCRs). Each standard deviation increase in TiRP was associated with a 23% increase in the odds of Treg status (OR: 1.231, 95% CI:1.227 – 1.235, P = 2.4 x 10-3248, LRT). Percentile points are colored by Treg:Tconv ratio ranging from blue (lowest) to purple (highest)。

结果7、运用TiRP helps to explain Treg-Tconv plasticity in the tumor microenvironment。（单细胞数据）

图片.png

1、与前面的研究结果一致

2、We next asked whether TiRP could help to explain regulatory T cell plasticity.It is well recognized that naïve Tconv thymic emigrants can be peripherally induced to adopt a regulatory phenotype. Conversely, some Tregs have been observed to lose FOXP3 expression and adopt a pro-inflammatory phenotype. Expanded T cell clones (possessing the same TCR) observed as both Tregs and Tconvs within the same donor (hereafter referred to as “mixed clones”) may represent lineages of T cells that have undergone such peripheral conversions. We hypothesized that the TiRP of these T cells may be intermediate, rendering them most susceptible to peripheral conversion.

3、单细胞数据的基础分析

图片.png

On average, 19.2% of expanded clones from the same donor were observed in both the Treg and Tconv state, including a few large clones with a relatively even balance。

图片.png

4、We next tested whether the TiRP score of mixed clones was in between that of purely Tconv and Treg clones (Methods). In the previously held-out bulk sequencing data, the TiRP scores of mixed clones were significantly greater than those of expanded Tconv clones and less than those of expanded Treg clones (上图f, mixed-Tconv difference = 0.03, P = 2.0 x 10-40; mixed-Treg difference = -0.29, P = 9.1 x 10-16). These single cell data confirmed that Tregs of mixed clones indeed exhibited greater FOXP3 expression than Tconvs within the same clonal expansion。As in the previously held-out bulk sequencing data, mixed clones in single cell data had intermediate TiRP scores which were significantly greater than the scores of expanded, pure Tconv clones(上图g).

5、来个结论To our knowledge, TiRP is the first metric to identify TCR-intrinsic, rather than TCR-extrinsic factors relevant to peripheral phenotypic conversion.

结果8、Separable components of TiRP: affinity to self-peptides and affinity to human MHC

1、conducted a principal components analysis (PCA) of TCR feature frequencies in the sorted samples of the replication dataset, in which all T cell states of interest were available。observed that the major axes of TCR sequence variation corresponded to T cell state, rather than donor HLA genotype or clinical phenotype。While our previous supervised modeling was designed to focus on Treg-Tconv differences, this approach recovered the importance of T cell state in an unsupervised manner.

图片.png

2、PC轴的意义,antigen-experienced (Treg and memory Tconv) versus naïve (PC1), and regulatory versus conventional (PC2).The axis dividing antigen-experienced from inexperienced samples (PC1) was most reliant on TRBV gene frequencies,while the axis dividing regulatory versus conventional samples (PC2) was most reliant on mean percent composition of amino acids in CDR3βmr and the CDR3βmr adjacent residue p113.

图片.png

3、Since TiRP is a weighted sum of TCR features from the V-, J- and middle regions, the score can be divided into three score components corresponding to these three regions. TiRP scoring by TCR region revealed that V-region-specific TiRP (vTiRP) and CDR3βmr-specific TiRP (mTiRP) indeed captured PC1 and PC2, respectively(Figure 7d-e, vTiRP – PC1 R = -0.86, P = 1.5 x 10-20, mTiRP – PC2 R = 0.85, P = 2.6 x 10-20).

图片.png

4、We next investigated possible biological drivers for vTiRP and mTiRP。The biological structure of the pMHC-TCR complex suggests that different regions of the TCR may promote Treg fate via particular affinities: MHC II mostly contacts the V-region of the TCR, while the self peptide is in closest contact with CDR3βmr。Thus, we hypothesized that vTiRP was driven by affinity to human MHC II, while mTiRP was driven by affinity to self peptides. To test this idea, we examined TiRP in two complementary datasets: 1) murine Treg TCRs, which recognize self-peptide but not human MHC, and 2) human memory Tconv TCRs, which recognize human MHC but not self-peptide.

图片.png

5、Applying our scoring system to murine data revealed that human TiRP was significantly elevated in murine Tregs compared to Tconvs。Because the self-peptide is the only consistent element of the Treg ligand between these species, the best explanation for such cross-species Treg TCR similarity is affinity to thymic self-peptides.

图片.png

图片.png

最后看看方法

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

图片.png

真的是相当难，看来读懂需要好几遍啊

生活很好，有你更好