search

ixxmu / mp_duty

抓取网络文章到github issues保存
https://archives.duty-machine.now.sh/
103 stars 30 forks source link

ccRCC-2：scRNA+scATAC #3125

Closed ixxmu closed 1 year ago

ixxmu commented 1 year ago

https://mp.weixin.qq.com/s/KyqJfWW7MDoFHT70mXLpeA

ixxmu commented 1 year ago

ccRCC-2：scRNA+scATAC by YuYuFiSH

ccRCC-2：scRNA+scATAC

Author: YuYuFiSH

Created time: August 3, 2022 7:55 PM


🍥文章：

Single-cell multiomics analysis reveals regulatory programs in clear cell renal cell carcinoma

Cell DIscovery  IF=30.96 Q1

🍥代码：

GitHub - Dragonlongzhilin/RenalTumor: Single-cell multi-omics analysis reveals regulatory programs in clear cell renal cell carcinoma

🍥数据集：

PRJNA768891


可以发现整个文章的基调，重点探索肿瘤，TAM，CD8+细胞差异的转录因子TF和靶向基因的调控关系

流程图

样本

  • 三名患者：配对的scRNA-seq和scATAC-seq
  • 一名患者：单独的scRNA-seq
  • cell carcinoma (ccRCC) ：透明细胞肾细胞癌 (ccRCC)

质控

  • scRNA-seq：38，097个细胞
  • scATAC-seq：21，272个细胞

1.  降维分群

scRNA注释

scRNA-seq数据集中鉴定了15种细胞类型

Tumor：CA9，NDUFA4L2

可以看到主要是5个淋巴系免疫亚群：

  • CD4+ (CD4, IL7R, CD3D, CD3E)
  • CD8+ T cells (CD8A, CD8B, CD3D, CD3E)
  • Treg (FOXP3, IL2RA)
  • natural killer (NK)/natural killer T (NKT) cells (KLRD1, GNLY)
  • B cells (MS4A1/CD20, CD79A)

以及5个髓系细胞

  • macrophages (CSF1R, CD68, CD163),
  • monocytes (S100A12, FCGR3A/CD16),
  • mast cells (TPSAB1, KIT)
  • Dendritic cell
  • Neutrophil

和之前研究一样，VCAM1− endothelial cells highly expressed genes regulating endothelial cell proliferation and vasculature development内皮细胞增值和血管发育, whereas the VCAM1+ endothelial cells highly expressed genes regulating immune cell chemotaxis and migration 免疫细胞的趋化和迁移

scATAC注释

关于scATAC注释，作者做了两种策略
  1. For our scATAC-seq data, we calculated the prediction scores by Seurat’s label-transfer algorithm and annotated cell clusters in a supervised manner 直接将scRNA的结果映射到scATAC中（d，g图）
  2. In parallel, we inspected the chromatin accessibility at the promoter regions for known marker genes and calculated their activity scores for assigning cell identities 气泡图，直接根据marker命名（h图）

Comparing the annotation results between these two annotation strategies, we found that most cell types were present in both datasets, which supported that scATAC-seq is comparable to scRNA-seq in the detection and annotation of cell types (Supplementary Fig. S1i)

使用Jaccard index 比较两种策略得到的细胞类型，仔细看到映射这种方法多得到一个树突细胞 Dendritic cell

补充：Jaccard 系数，又叫Jaccard相似性系数，用来比较样本集中的相似性和分散性的一个概率。Jaccard系数等于样本集交集与样本集合集的比值，即J=|A∩B|/|A∪B|。例如可以计算同批样本两种不同技术SMART-seq2和10x数据得到的cluster相似性。不过看Jaccard index，两种策略注释细胞的匹配度也太好了吧？

结合上面两种策略， scATAC-seq 注释得到12种细胞类型

2. 探索肿瘤细胞的异质性

肿瘤细胞特异marker CA9 的染色体可及性以及CNVs分析 (chromosome 3p loss or chromosome 5q gain)均证明细胞分型的准确性

f：Normalized chromatin accessibility profiles for each cell type at canonical marker genes. The promoter region is highlighted in gray with the gene model and chromosome position shown below

  • 免疫细胞是数量最多的细胞群，占总细胞的70%以上，而肿瘤细胞占群体的<10%（9.36%来自scRNA-seq数据;6.2%来自scATAC-seq数据）
  • 样本间的细胞类型也不同。综上，整合两种技术解释了ccRCC的异质性

3. 特异性调控元件 peaks

  • 数据：scATAC-seq
  • MACS2方法确定了212，326个peaks，约10.6%（22，682个peaks）存在细胞间差异，即定义为DARs（adjusted P < 0.05 and log2(fold change (FC)) > 0.25）

富集分析，主要探索肿瘤细胞上调DARs的功能

Gene ontology term enrichment analysis for distal(intergenic region) and proximal (promoter and gene body) upregulated DARs in tumor cells.

Gene set enrichment analysis (GSEA) for significantly upregulated genes (log2 (FC) ≥ 0.25 and adjusted P < 0.05) in tumor cells

4. 识别转录因子TF

TF基序活性（chromVAR，bias-corrected deviation scores）：进一步支持鉴定出的细胞类型

  • T-Box (e.g., EOMES and TBX5) TFs :NK/NKT and CD8+ T cell
  • SPI127 TF were enriched in macrophages,
  • CEBP TF family28 exhibited high activity in monocytes
  • GATA229 TF was specifically enriched in mast cells.

tumor-specific TFs

为了揭示参与肿瘤发生和发展的关键转录因子，我们设计了一种过滤策略来识别富含肿瘤细胞的高度特异性转录因子 （tumor-specific TFs）

  • 肿瘤细胞中49个特异的TF (log2(FC) > 4 in tumor cells and log2(FC) < 1 in any other cell types; adjusted P < 0.0001)
  • 分别按照 logfc，表达量绘制热图展示
  • km生存分析 (TCGA-KIRC，Patients with high average expression of tumorspecific TFs had shorter overall survival and disease-free survival than the patients with low average expression）

TF调控网络

再次挑选重点的TF（HOXC5，VENTX，ISL1和OTP）

挑选条件：

we selected four TFs (HOXC5, VENTX, ISL1, and OTP) whose expression levels were significantly associated with worse overall survival in the TCGA-KIRC dataset, and binding sites were located in accessible chromatin regions, which were specific for kidney cancer and identified by ATAC-seq (Fig. 3a and Supplementary Fig. S3a).

TF调控网络：

We identified their target genes whose promoters or linked candidate cis-regulatory elements (cCREs) 顺势调控模块were accessible and contained the TF binding motif in tumor cells.

接下来，探索

  • TF的靶基因富集情况，和在多个数据集中探索每种细胞类型的靶基因评分，发现在肿瘤细胞中均显著增加
  • 实验证明：敲除这些TF显着降低了肿瘤细胞增殖并增加了细胞死亡
  • 靶向TF的潜在药物（LINCS consortium）

5. 恶性转录程序

  • 数据：四个样品的scRNA-seq 3564个肿瘤细胞
  • Pairwise correlation analysis：猜测肿瘤细胞内有多种不同的转录状态
  • nonnegative matrix factorization (NMF)：揭示了11个intratumor programs，并分层聚类确定了两个样本间相关性高的Meta-program（图b），看到图中有一些相关性低的样本舍弃掉了
  • 探索每个programs间的相关性，和两个Meta-program间相关性，发现相关性低 (r = −0.2)，表明它们在肿瘤中的不同作用
  • 富集分析：有啥不同生物学效应
  • 生存分析：每个Meta-program的平均基因表达与患者预后关系（TCGA-KIRC）

  • 因为一个TF普遍能调控多个基因，现探索上一部分提到的top 30 TF和Meta-program关键基因的调控数量关系
  • 发现TF主要调控靶基因 VEGFA 和 SMIM24
  • 看看前面提到的四种肿瘤特异性TF和和Meta-program关键基因的调控关系，同样出现了VEGFA 和 FXYD2（肿瘤内第一大DEG）

6. CD8 T淋巴细胞异质性

  • CD8+ T 淋巴细胞（CD8+ T lymphocytes）在抑制肿瘤进展中起着至关重要的作用，因此提取scRNA-seq和scATAC-seq数据集中的CD8 T细胞亚群，并删除了少于100个细胞的簇
  • Tissue-resident CD8+ T cells 组织驻留T细胞：(CD69, ZNF683/Hobit, ITGAE/CD103, and ITGA1/CD49A)
  • one cluster (tissue-resident C2) exhibited high expression of effector molecules (TNF, IFNG, and GZMA)
  • the other (tissue-resident C1) was highly expressed naive/memory genes (IL7R, CCR7, and TCF7).
  • exhausted CD8+ T cells：PDCD1 and TOX
  • exhausted immediate-early genes (exhausted IEG)：HSPA1A, DNAJB1, JUNB, and ATF3

  • 使用各种gene sets 探索不同亚群的生物学功能：cytotoxic（细胞毒性）, exhaustion（细胞耗竭）等
  • Recent studies have reported that progenitor exhausted T cells control tumors more effectively than terminally exhausted T cells and respond better to anti-PD1 therapy
  • 对scRNA-seq数据中每个CD8 T细胞簇差异分析
  • 对DEG富集分析
    • Multiple T cell exhaustion-related pathways, such as PD-1, IL2-STAT546, and IL6-JAK-STAT347 signaling, were significantly enriched in the exhaustion cluster.
  • 每个亚群的TF

7. 巨噬细胞的降维聚类

（与上面t细胞分析类似思路）

肿瘤相关巨噬细胞 (TAM) 是肿瘤中骨髓细胞的主要群体，在肿瘤发生和耐药性中起着至关重要的作用。分别在scRNA-seq和scATAC-seq数据对其进行降维聚类


在scRNA数据中检查不同分群一些特定的MHC分子，富集分数，免疫检查点基因和共刺激分子的表达


但同样的通路，在scATAC数据中分析，也许数据的稀疏性呈现不同的趋势


随后，分析得到显著差异的TF，其中四个TF在两种数据中呈现一样的趋势

Four TFs (MEF2C, NFKB1, RUNX3, and ENO1) showed a substantial increase in both gene expression and activity in the corresponding clusters

接下来，作者把分析落脚点放在MEF2C，以往的研究表明，MEF2C 在促进 M1 巨噬细胞极化和诱导巨噬细胞死亡方面起着至关重要的作用

We identified MEF2C target genes using the same strategy described in the previous section and found that this gene regulated multiple TFs (e.g., FOXO1, NEU1, and NRP1) and chemokines (e.g., CCL3 and CCL3L1) (Fig. 6j) that have been demonstrated to be associated with phagocytosis and angiogenesis in macrophages

target gene——TF网络

生存分析

8. 细胞间通讯

  • scRNA-seq数据
  • 配体-受体互作图（CellPhoneDB，其中内皮细胞数量做多）
  • 肿瘤与TAM，CD8+ cell 的互作情况
  • 发现在ccRCC中尚未报道的新相互作用，如RPS19-C5AR1和LTB-LTBR