# 2.2.13 研究 | 2021-解析食管鳞癌化疗病人的单细胞转录组

## 前言

题目：Dissecting the single-cell transcriptome network in patients with esophageal squamous cell carcinoma receiving operative paclitaxel plus platinum chemotherapy

日期：2021-10-26

期刊：Oncogenesis

链接：[https://www.nature.com/articles/s41389-021-00359-2](#qian-yan)

## 一句话概括

分段式作文：5个常规治疗病人和5个化疗病人得到的单细胞数据，先初步分成上皮、基质和免疫，然后每一类继续细分，分别探讨

## 首先是初步的分群

> 三大群：上皮、基质和免疫，然后基质又细分一下：内皮和成纤维细胞；免疫又细分为T、NK、B、髓细胞、肥大细胞

选择的样本：5个病人underwent surgery alone (SA-ESCC) + 5个病人underwent preoperative paclitaxel plus platinum chemotherapy (NACT-ESCC)，总共得到113,581个细胞（63,837 SA-ESCC derived cells and 49,744 NACT-ESCC derived cells）

使用`SingleR`结合`CellMarker`的数据库以及其它文献的marker基因进行注释，得到了**epithelial**, **stromal** (fibroblasts and endothelial cells), and **immune** cells (T, NK, B, Myeloid, and Mast cells)

**图D**是细胞比例分布：NACT-ESCC 病人的免疫细胞（尤其是T、B细胞）比例很高，但它的上皮细胞和基质细胞比较少，说明NACT病人的宿主抗肿瘤免疫应答反应被激活

**图E**展示了几个通路的关键基因的热图：

* cytokines和NF-kB相关基因（TNFSF13B, IL6, and NKFB1）在monocytes上调，说明monocytes与食管癌发生相关
* hypoxia genes (e.g., LDHA, HIF1A, and EPO)在SA-ESCC病人中高于NACT-ESCC，说明化疗影响缺氧微环境

> Cytokines, nuclear factor-kB (NF-kB), and hypoxia signaling pathways play an essential role in tumorigenesis

![image-20211101101344645](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-021345.png)

## 然后看上皮

上皮细胞细分得到7个亚群：

* Tumor epithelial cell markers (EPCAM, MDK, and SOX4) 在cluster1、6中高表达
* non-malignant epithelial cell markers (KRT5 and KRT14)在cluster0、2、4、5高表达
* cluster3同时表达tumor和normal的上皮细胞marker

同时CNV也显示tumor epi的malignant scores要比normal epi高（**图C**）

另外，`scpred`R包发现基本没有误判的情况：

* 86% (3662/4245) of malignant epithelial cells were not identified as healthy epithelial cells
* 95% (6977/7324) of non-malignant epithelial cells were not identified as tumor epithelial cells

再次证明细胞分群注释的正确性

![image-20211101102815761](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-022815.png)

然后是**差异分析：**

* 预后差相关基因（SOX4 and MDK）在SA-ESCC-malignant epithelial cells富集
* 免疫相关基因（IGLC2, FABP5, and S100A2）在non-malignant epithelial cells上调
* SPRR3 and CEACAM6在NACT-malignant epithelial cells高表达

另外，non-malignant clusters也是存在异质性的，可以按起源继续分成2个cluster

然后是**拟时序分析：**

* 最开始是non-malignant epithelial cells derived from SA-ESCC patients
* 然后是 NACT-derived non-malignant epithelial cells
* 然后分叉：一个是SA-ESCC-malignant epithelial cell，一个是epithelial cells from different sources (NACT-ESCC or SA-ESCC patients)

最后看不同cluster的**通路活性差异**：

* TME formation related pathways (e.g., hypoxia and angiogenesis)主要在non-malignant epithelial cells
* ESCC development and progression (e.g., G2M-checkpoint and DNA-repair pathways)主要在 SA-ESCC malignant epithelial cell
* inflammatory response, and IL2/IL6 related signaling pathways 在NACT-ESCC malignant epithelial cells和non-malignant epithelial都有表达

#### SA-ESCC和NACT-ESCC的细胞通讯

在SA-ESCC中，fibroblasts与其他类型细胞联系密切

在NACT-ESCC中，免疫细胞（尤其是T细胞和单核细胞）作用明显

针对这两种类型，然后分别对SA-ESCC的基质细胞和NACT-ESCC的免疫细胞取top50差异基因做通路，发现：

* SA-ESCC的基质细胞主要调控extracellular matrix
* NACT-ESCC的T和单核细胞主要抑制肿瘤发展

![image-20211101104446457](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-024446.png)

接着做了代谢分析，发现：

* Oxidative phosphorylation, glycolysis/gluconeogenesis, and citrate cycle (TCA cycle) pathways在SA-ESCC病人的恶性与非恶性细胞都上调
* 而在NACT-ESCC中，只在非恶性细胞中上调
* NACT-ESCC has a unique TME compared to SA-ESCC
* **mitochondria** is a potential therapeutic target for SA-ESCC

![image-20211101103900818](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-023901.png)

## 接着看基质细胞（内皮+成纤维）

基质细胞包括了**19,977** and **6588** fibroblasts and endothelial cells (EDCs)

* 内皮细胞的maker选取了CLDN5 and RAMP2，又细分为8个cluster，整合为4个子细胞类型：immune EDCs (CCL5 and CXCL13), lymphatic EDCs (PDPN and PROX1), tumor EDCs (ACKR1 and POSTN), and vascular EDCs (PLVAP and SLC9A3R2)【图A】
* 成纤维细胞marker选取了C1R and COL1A2，得到11个cluster，整合为3个子细胞类型：COL14A1 matrix fibroblasts (COL14A1 and GSN), myofibroblasts (CTHRC1 and MMP11), and vascular smooth muscle cells (DES and MYH11)【图B】

### **看内皮细胞**

在图A发现：NACT-ESCC的tumor EDC虽然少于SA-ESCC，但它的immune EDCs却高，于是接下来做了tumor EDC和immune EDC的GSVA和差异分析【图C、D】，发现：

* metabolic pathways, including cholesterol homeostasis and fatty acid metabolism在NACT-ESCC的tumor EDCs中更高
* **cancer-promoting** genes (e.g., S100A family genes) were highly expressed in **SA-ESCC**
* **HSP** family genes were significantly elevated in the **NACT-ESCC**
* marker genes for **SA-ESCC** derived immune EDCs were implicated in biological regulation and **cell growth**
* marker genes for immune EDCs in the **NACT-ESCC** group were mainly involved in cellular and **immune responses**

> HSP family genes promote cancer growth and metastasis in several tumors

### **看成纤维细胞**

图B看到：

* COL14A1+ matrix fibroblasts 主要在 NACT-ESCC
* myofibroblasts 主要在 SA-ESCC

图F看到：

* POSTN and WNT5A 在 myofibroblasts高表达，和Wnt and Notch signaling pathways相关
* ACTA2在myofibroblasts and smooth muscle cells高表达

另外图G的GSVA也看到，Wnt and Notch signaling也是在myofibroblasts富集

![image-20211101105031858](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-025032.png)

## 接着看髓细胞

基于CD68 and LYZ基因，拿到16,305 myeloid cells，继续细分为monocytes, macrophages, dendritic cells, and undefined myeloid cells

NACT-ESCC的macrophage细胞相对少，但monocyte多

**图B**选了一个marker基因Secreted phosphoprotein1 (SPP1)，它与预后差相关，发现它主要在SA-ESCC的macrophage

**图C**想看从单核到巨噬细胞的转变过程：State 1 and 2包括了NACT-ESCC and SA-ESCC的巨噬细胞；但3和4主要就是NACT-ESCC

另外，GSVA看到anti-tumor (e.g., Interferon-gamma (IFN-gamma) and Interferon-alpha (IFN-alpha)) response pathways主要在NACT-ESCC富集；pro-tumor responses (e.g., KRAS and EMT pathways)在SA-ESCC富集，TNF-α-NF-κB pathway也是在SA-ESCC富集

**图E**用`SCENIC`做了转录因子分析：

* SA-ESCC中，NFIA, TWIST1, and MAFB是在macrophages中top3上调；RXRA, TGIF1, and XBP1在monocytes中上调
* NACT- ESCC中，SPI1 and KLF4在macrophages中上调

**图F**中显示：PDCD1LG2 (PD1-L2), CD274 (PD-L1), CD276 (B7-H3), and LAG3这些与抑制CD8+ T 细胞活性相关的基因，在macrophages中上调；促炎相关免疫检查点基因（IL1A and IL6）在monocytes上调

![image-20211101110603940](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-030604.png)

## 最后看B细胞和T细胞

根据CD79A and IGKC，拿到2999个B细胞，分成3个cluster，整合成2个亚型：Plasma B cells和follicular B cells

发现 plasma B cells在NACT-ESCC更多

图C做了富集分析，发现：

* NACT-ESCC的plasma B cells会富集 interferon（干扰素） response-related pathways, MYC, and allograft（异源移植） rejection pathways
* SA-ESCC的plasma B cells会富集unfolded protein responses and angiogenesis pathways

同样是转录因子分析：

* STAT3 were elevated in follicular B cells
* ATF3 were observed in plasma B cells

![image-20211101111834562](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-031834.png)

对于T细胞，根据CD3D and CD3E拿到36,706 T cells，细分为10个cluster，整合成6个亚型，发现CD8+ T在NACT-ESCC中更多，不过NKT cells, CD8/CD4 mixed Th, and CD4+ T cells数量减少

图C发现：

* naïve and co-stimulatory molecule markers在CD4+ T细胞上调
* Cytokines, effector molecules and inhibitory receptors 在CD8+ and NK T上调

图D计算了Bhattacharyya距离，用于估计NACT-ESCC and SA-ESCC的T细胞亚型相似性，发现它们的**差异主要来自CD8+ T细胞（图D）**

然后将 CD8+ T 继续细分（图E），得到10个cluster，整合成3个亚型：cytotoxic CD8+ T cells, exhausted CD8+ T cells, and naïve CD8+ T cells。在NACT-ESCC中，exhausted CD8+ T cells and naïve CD8+ T占比明显升高，表示化疗改变了ESCC的免疫微环境

![image-20211101112607678](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-032608.png)

之后又对 CD8+ T cells进行了拟时序分析，轨迹从naïve CD8+ T & cytotoxic CD8+ T cell开始，分叉之一是cytotoxic CD8+ T cells，另一个分叉是exhausted CD8+ T cells；

轨迹热图发现了3个时期：

* phase 1主要是Naïve CD8+ and cytotoxic CD8+ T cells，GZMK and IL7R高表达，HAVCR2 and LAG3 低表达；
* phase2 高表达STMN1, CENPE, and GNLY，包含了exhausted CD8+ and cytotoxic CD8+ T cel
* phases 3 主要包括exhausted CD8+ T cells

另外，发现了SA-ESCC和NACT-ESCC免疫检查点相关的差异：PDCD1LG2基因在NACT-ESCC的exhausted CD8+ T cells高表达，而在SA-ESCC中是在cytotoxic CD8+ T中高表达，说明SA-ESCC and NACT-ESCC在免疫治疗反应方面存在差异

![image-20211101124629150](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-11-01-044629.png)