# 2.2.12 研究 | 2021-单细胞转录组揭示乳头状甲状腺癌起始与发展

## 前言

题目：Single-cell transcriptomic analysis of the tumor ecosystems underlying initiation and progression of papillary thyroid carcinoma

日期：2021-10-18

期刊：nature communications

链接：<https://www.nature.com/articles/s41467-021-26343-3>

## 一句话概括

文章关注肿瘤微环境：先初步分群，然后依次看了几个重要组成部分：thyrocytes、immune细胞群、基质细胞（fibroblasts 、endothelial cells）

## 使用的数据

11个病人的23个组织，得到158,577个细胞

![image-20211026100416227](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-020416.png)

## 初步分群结果

得到**6个主要的细胞群：**

* T/natural killer (NK) cells (CD3D, CD3E, CD3G, CD247)
* B cells (CD79A, CD79B, IGHM, IGHD)
* thyrocytes (TG, EPCAM, KRT18, KRT19)
* myeloid cells (LYZ, S100A8, S100A9, CD14)
* fibroblasts (COL1A1, COL1A2, COL3A1, ACTA2)
* endothelial cells (PECAM1, CD34, CDH5, VWF)

![image-20211026100646905](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-020647.png)

接下来重点关注免疫细胞（T, NK, B, and myeloid cell ）的细分，得到**22个亚群**：

* T细胞可以根据CD4和CD8的表达量继续细分，得到7 clusters for CD4+ cells & 4 clusters for CD8+ cells
* NK, B, and myeloid cells同样细分得到 2, 3, and 6 subsets

并且每个亚群都有各自的组织偏好性（图e）

其中，CD4-c6中的FOXP3上调，CD8-c4中的PDCD1上调，分别对应了 regulatory CD4+ T cells (Treg) 和 exhausted CD8+ T cells (Tex)，它们主要存在于皮下和肿瘤组织中；另外CD8-c3 中的GNLY上调，对应了效应T细胞，主要存在于肿瘤组织

这几种免疫细胞均在肿瘤组织中富集，说明了宿主免疫应答与肿瘤免疫逃逸共存

![image-20211026101100992](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-021101.png)

## 看甲状腺细胞群的异质性

拿thyrocytes这群细胞继续细分，得到9个亚群，其中1、2亚群主要是para-tumor，可以代表non- malignant thyrocytes；3-9则是malignant thyrocytes

![image-20211026102825082](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-022825.png)

接下来就用3种方法证明这样划分是对的：

![image-20211026103431062](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-023431.png)

* 首先是相关性（图c），发现1、2最相关
* 然后（图d）计算每个亚群的thyroid differentiation score (TDS)。TDS用来评价分化程度，其中1、2最高，并且照应了para-tumor最高的情况
* 最后使用TCGA的PTC样本bulk转录组，构建一个分类器，达到97% sensitivity and 96% specificity ；再将这个分类器用在这个单细胞数据，发现与推测的组织类型相似度达到了：95%, 97% and > 98% of c01, c02, and c03–09 cells

于是，下面就认为**malignant (c03-c09)** and **non-malignant (c01, c02)** thyrocytes

探索了恶性vs非恶性、恶性肿瘤 vs恶性淋巴、恶性肿瘤 vs恶性皮下，发现：**TMSB4X** as a suggestive biomarker that potentially involves in PTC initiation and progression

![image-20211026103921664](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-023922.png)

#### 既然分开了恶性和非恶性，那就先研究非恶性的两个亚群

之前图d看到，c1和c2的分化指标虽然比其他群高，但二者还是有差距，于是就开始探索这两群非恶性细胞的差异

看到c1到c2会有中间态，然后看到c1=》c2=》c3-c9的过程中，TMSB4X也确实逐渐升高

推测：**c2这个群可能不是完全的正常状态，是出于正常和恶性之间的过渡态**

![image-20211026104550993](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-024551.png)

通过实验验证了c2的属性是：cancer-primed nature of these outwardly normal but **transcriptionally altered** premalignant cells, which provide both seeds and soil for the eruption of malignant growths，将c2归结为premalignant

最后对比了c1和c2，对PKHD1L1这个基因感兴趣，推测 PKHD1L1 might function as a tumor suppressor gene in multiple solid tumors

![](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-025250.png)

## 接下来重点看恶性的细胞——发育轨迹

正常的c1和premalignant的c2在右上角，作为发育起源（normal-cell-initiated State 1），看看三个发育状态有什么区别：

![image-20211026110343493](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-030344.png)

* state1有一个明显的转录因子SOX9，作用是branching folliculogenesis（卵泡生成） of normal thyroid gland，于是state1的名称是follicular-like thyrocyte

![image-20211026110630820](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-030631.png)

* State 2 was basically a half-half mixture of tumor and LN-metastatic thyrocytes ![image-20211026110928873](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-030929.png)
* State 3, featured by the **lowest TDS score and RAS score**, and **highest** **BRAF** score, contained the vast majority of RAIR subcutaneous metastatic thyrocytes；另外state3高表达GATA2, MYC, SOX4（去分化相关的转录因子）

## 拿发育轨迹的结果做点什么

得到了480 pseudotime-associated genes (PAGs)，因为它们主要就是在甲状腺上皮细胞中，再结合2个bulk转录组数据，可以用来更新BRAF-/RAS-like 分子亚型方法。发现和之前的分型方法一样，**BRAF-like tumors也是存在很强的异质性，将其分成2个亚型**（BRAF-like-A and BRAF-like-B），发现BRAF-like-B更加特别：associated with TCV pathology ； lower TDS scores；advanced staging；significantly compromised DFS；

![image-20211026123054006](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-043054.png)

GSEA结果发现BRAF-like-B在免疫相关通路活性更高

![image-20211026123531415](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-043531.png)

并且三个亚型RAS-like, BRAF-like-A, BRAF-like-B分别对应了三种表型：follicular-like, p-EMT-like, dediff-like

![image-20211026123856784](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-043856.png)

## 接下来看基质细胞——CAFs

> 其中重点关注cancer-associated fibroblasts (CAFs)

初步分群的fibroblasts这群，表达了CAF的marker：VIM, S100A4, ACTA2 (α-SMA), and PDGRFA

因此可以直接把fibroblasts成为CAF，然后把CAF分群，一开始分成4群，其中cluster 0, 1 and 3可以当成myofibroblastic CAFs (myoCAF)；cluster 2是inflammatory subtype (iCAF) ，它的marker基因是CFD, PLA2G2A, CCDC80

![image-20211026124527891](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-044528.png)

正式有了iCAF和myoCAF，再看和分型的关系，发现CAFs主要存在于BRAF-like中

接着做了iCAF和myoCAF的细胞通讯，发现：

* regulation of cellular immunity is an important function for **iCAFs** in PTC
* **myoCAFs** tends to exert mechanical and chemical influence on tumor progression rather than through direct cell communications, as described in other solid tumors

![image-20211026124310210](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-044310.png)

## 再看基质细胞——内皮细胞（endothelial cells, ECs）

根据已有的marker，将EC分成5个类型：arterial, venous, lymphatic, immature and tip

* arterial 高表达 arterial development and remodeling (**FBLN5, GJA5, JAG1**) and smooth muscle contraction (**PPP1R14A**)
* venous 高表达 VWF (von Willebrand factor) and genes associated with leukocyte recruitment (**ACKR1**) or adhesion (**SELE**)
* Lymphatic 高表达 canonical marker **LYVE1** and a chemokine ligand **CCL21**
* immature 高表达 Notch signaling and target genes (**JAG1, HES1, ID1, ID2, ID3**), and genes involving in barrier integrity (**ENG, PLVAP, HSPG2, APLNR**), which may resemble the stalk-like cells
* tip 高表达 cell migration **(NRP1, ENPP2**), adhesion (**THY1**) and vessel formation (**FLT1**, also called VEGFR1; **KDR**, also called VEGFR2; **NRP1**, also called VEGF165R)；并且tip高表达一些转录因子，比如 endothelial migration and sprouting, such as ZEB1, HOXB5 and STAT family

图h看到，只有lymphatic在正常的甲状腺组织中富集，其他均存在于原发或转移癌中

最后的细胞通讯看了EC和免疫细胞，发现：

* lymphatic 和免疫细胞通过atypical chemokine receptor 2 (ACKR2)进行关联【它用来调控chemokine availability】（在附图）
* venous, immature and arterial ECs 和免疫细胞通过 Intercellular Adhesion Molecule 1 (ICAM1) 关联 （在附图）
* tip 和免疫细胞通过key angiogenic VEGF-VEGFR signalings 关联（图i）

![image-20211026125434470](https://jieandze1314-1255603621.cos.ap-guangzhou.myqcloud.com/blog/2021-10-26-045435.png)
