读文献04-空转揭示结直肠癌fibro和巨噬两亚群互作

• Date : [[2022-10-31]]
• 公众号：北野茶缸子
• 参考：Single-cell and spatial analysis reveal interaction of FAP+ fibroblasts and SPP1+ macrophages in colorectal cancer | Nature Communications[1]
  • 10X单细胞联合空间转录组技术解析结直肠癌肿瘤微环境 - 知乎 (zhihu.com)[2]
  • 单细胞和空间转录组分析揭示结直肠癌中FAP+成纤维细胞和SPP1+巨噬细胞的相互作用 - 知乎 (zhihu.com)[3]
• IF：NC

数据

image-20221031110042388

5名非转移性肠癌患者（olonic adenocarcinoma (COAD), n = 2; rectal adenocarcinoma (READ), n = 3）：手术获取肿瘤样本和邻近正常组织，经质控后获得54103个细胞的单细胞转录组测序数据用于后续分析。

in which 29,481 cells were originated from adjacent non-malignant tissues and 24,622 cells from tumors

注释

The cells were classified into nine major cell types (Fig. 1b–d), including epithelial cells (n = 8940) identified by the expression of EPCAM and CDH110, T/ILCs cells (n = 17,420) which expressed the T-cell receptor (TCR) signaling mediators CD3E and CD3G15, B cells (n = 2998) marked by MS4A1 and CD79A11, plasma cells (n = 7252) identified by SDC1 and MZB1 expression25, myeloid cells (n = 4617) which were positive for CD14 and FCGR3A expression26, mast cells (n = 2781) defined by their classical markers KIT, IL1RL1,andMS4A2, endothelial cells (ECs; n = 2205) marked by PECAM1 and CDH511, mesenchymal stromal cells (MSCs; n = 7451) marked by COL1A1 and COL3A111, and glial cells (n = 439) marked by S100B and CDH2.

基础操作了：

image-20221031110336949

image-20221031110919399

对每个分群还进行了亚群注释：

image-20221031134738677

很好看的图。

此外，作者还利用随机森林去评价亚群注释：

image-20221031134808126

image-20221031134909939

ps：也就是利用随机森林，看利用二分train 的特征，能否预测出test 中对应的label。作者的代码里也提供了这部分。

signature 打分比较normal 与tumor

收集了MsigDB 上的一些通路：

image-20221031110957174

包括免疫、代谢、信号、增殖，看不同cell type tumor 与normal 对比后logFC。

he immune-related pathways,，suggesting the involvement of MSCs and ECs in the immune response against colorectal cancer.

These characteristics might reflect a stromal cell interaction localized in the hypoxic region of the tumor that links macrophages, MSCs, and ECs to remodel the CRC microenvironment. In addition, tumorinfiltrating myeloid cells and T/ILC exhibited greater enrichment of metabolism-related genes, including fatty acid metabolism, xenobiotic metabolism, bile acid metabolism, and cholesterol homeostasis, than those cells from normal mucosa (Fig. 2a), suggesting immunometabolism was reprogrammed in the CRC TME.

regulatory pathways of major cell types were shaped in the CRC TME

这里已经开始定位到 MSCs 了。

为了验证normal 和tumor 不同cell type 的表达区别，使用了更大的队列！这里可圈可点啊，去看bulk 反卷积后各个cell type 对比normal 和tumor 的差别吗？

这里为了验证cibersort 的性能，还做了验证，用两个tumor 做反卷积，看预测的另外三个sample 的cell type 构成：

image-20221031112328929

收集的队列真多呀：

image-20221031111917979

在bulk 里看比例的相关性还是比较有意思的。也是一个定位某些亚群的方法。

比较了一下不同cell type 比例的相关性：

image-20221031112439400

发现MSCs 和髓系细胞相关性很高，而且MSCs 的浸润和生存显著相关：

image-20221031112618865

FAP+ 成纤维细胞和肿瘤进展相关

正常与肿瘤对比的细胞类型差别说明微环境存在某些差异。

MSCs and fibroblast-like cells have long been suggested as a key stromal cell type involved in regulating tumorigenesis and the progression of cancer

间充质（MSCs）细胞亚群

image-20221031133441405

鉴定出了MSCs 的Fibro 的FAP+ fibroblasts 亚群。

还比较了N-T 各个亚群的占比：

image-20221031140328022

image-20221031140110217

The FAP+ fibroblasts (Diff = 42.4%, p = 0.0052), proliferating fibroblasts (Diff = 2.29%, p = 0.0099), and pericytes (Diff = 10.4%, p = 0.031) were markedly enriched in tumor tissue as compared to that from adjacent normal tissue, while NT5E+ fibroblasts (Diff = 14.7%, p = 0.0053), FGFR2+ fibroblasts (Diff = 19.3%, p = 0.015), ICAM1− telocytes (Diff = −6.53%, p = 0.0071), and MFAP5+ myofibroblasts (Diff = 9.84%, p = 0.0066) were enriched in tumor-adjacent normal tissue

使用上述队列做反卷积：

image-20221031111917979

发现FAP+ 无论在scRNA 还是bulk 当中都是显著富集的：

image-20221031140733870

而且FAP+ 还和生存有关、MSI、分期等有关：

image-20221031142251790

image-20221031140638543

接着用流式去佐证：

image-20221031143123552

image-20221031143208784

the infiltration of FAP+ fibroblasts were significantly increased, while NT5E+ and FGFR2+ fibroblasts were significantly decreased in CRC samples

这里为什么定位到FAP+ 的成纤维？仅仅是它的group差异？临床表型关联？应该也是生物学背景吧。

We next performed data integration between publicly available single-cell transcriptomics data of CRC MSCs and our own datasets to validate tumor-specific FAP+ fibroblasts

拟时序和转录因子分析

tumor-specific FAP+ fibroblasts likely originated from FGFR2+ fibroblasts or ICAM1+ telocytes

image-20221031144651381

pySCENIC 找了一些top 的转录因子：

image-20221031144928241

此外还提到了：

To check the gene expression of transcription factors (TFs) alone, we retrieved Genes encoding TFs from four TF-related public datasets: JASPAR87 (http://jaspar.genereg.net/[4]), DBD88 (http://www.transcriptionfactor.org/[5]), AnimalTFDB89 (http://bioinfo.life.hust.edu.cn/AnimalTFDB/[6]), and TF2DNA90 (http://www.fiserlab.org/tf2dna_db/[7]). We overlapped the TF genes with the DEGs quantified above, and determined the most specifically expressed TFs in each cluster.

比较了不同cell type 的top TF，Hypoxia is one of the most important characteristics of the TME, and a previous study demonstrated that TWIST1 might be regulated by hypoxia52,53. Indeed, the hypoxia-dependent HIF-1α signaling pathway was significantly enriched in FAP+ fibroblasts compared with other MSC subtypes

TWST1.

ps：这里结合模块化分析，也是一个思路。另外，可否比较normal 和tumor 相同cell type TF 调控的差别呢？

对FAP+ fibro 和其他间充质细胞比较，看差异基因富集通路：

image-20221031150124619

FAP+ fibroblasts were enriched in ECM-receptor interaction and focal adhesion-related pathways compared with either FGFR2+ fibroblasts or ICAM1+ telocytes (Supplementary Fig. 4i, j), implying the involvement of these cells in the ECM formation. We further demonstrated that genes highly expressed in FAP+ fibroblasts compared with all other MSCs subtypes were mostly the genes involved in ECM-receptor interaction, TNF signaling pathway, and fatty acid biosynthesis, suggesting that activation of these processes is involved in the commitment of FAP+ fibroblasts

肿瘤相关SPP1+ 巨噬细胞和肿瘤进展相关

定位到髓系，是因为之前的相关性结果吗？

image-20221031150312194

常规的细分亚群：

image-20221031150350369

We investigated the alterations in myeloid cell subtypes among adjacent tissues and tumor tissues and showed that macrophage and neutrophil populations were predominantly present in tumor tissues, while DCs were enriched in adjacent normal tissues：

image-20221031150745499

SPP1+ macrophages are tumor-specific macrophages, accounting for 11.6% of myeloid cells in tumor samples but only 0.68% of the myeloid cells in adjacent normal tissues：

且发现SPP1+ 浸润也和生存有关：

image-20221031151114545

接着都是类似FAP+ 的一套分析了：

image-20221031151323381

image-20221031151314221

FAP+ 与SPP1+ 浸润更坏预后

image-20221031150312194

AP+ fibroblasts and SPP1+ macrophages were mostly enriched in tumor tissue (Figs. 3b and 4b), and a high correlation between the infiltration of MSCs and myeloid cells was found in patients across 14 colorectal cancer datasets (Fig. 2b).

ps：从这样的相关性结果来定位某些亚群细胞，也是挺有意思的。

对细分亚群再去看相关性：

image-20221031151714766

发现同样关联到了这两个亚群。

发现两个亚群共同与更坏预后有关：

image-20221031151838320

去看这些双高样本和双低的差异富集分析：

image-20221031152032038

These samples also displayed highly enriched hypoxia gene set (Fig. 5c), which is consistent with the supposed hypoxic environment of tumors

蛋白水平同样：

image-20221031153534324

ps：这个可以用tcga 的蛋白数据库啊。

免疫印记发现两个细胞在一陀儿Immunofluorescent labeling demonstrated the close proximity of SPP1-positive and FAP-positive cells in CRC tissue：

image-20221031153702498

空转结果揭示SPP1+ 和FAP+ 的相互作用

一共四个切片，亚群注释：

image-20221031154912324

这里是如何注释的呢？

Spatial Transcriptomics (ST) slides were printed with two identical capture areas from four CRC patients. The capture of gene expression information for ST slides was performed by the Visium Spatial platform of 10x Genomics through the use of spatially barcoded mRNA-binding oligonucleotides in the default protocol. Raw sequencing reads of spatial transcriptomics were quality checked and mapped by Space Ranger v1.1. The gene-spot matrices generated after ST data processing from ST and Visium samples were analyzed with the Seurat package (versions 3.2.1) in R. Spots were filtered for minimum detected gene count of 200 genes while genes with fewer than 10 read counts or expressed in fewer than 3 spots were removed. Normalization across spots was performed with the LogVMR function. Dimensionality reduction and clustering were performed with independent component analysis (PCA) at resolution 1.1 with the first 30 PCs. Signature scoring derived from scRNA-seq or ST signatures was performed with the AddModuleScore function with default parameters in Seurat. Spatial feature expression plots were generated with the SpatialFeaturePlot function in Seurat (versions 3.2.1).

作者并没有说。貌似是各自做了降维分群注释。同时用到了sig score：

image-20221031163259037

这里直接将部分FAP+ fibroblasts/SPP1+ macrophages 分到了一起，是否合适呢？

病人中的各个细胞这两个sig 也都是显著相关的：

image-20221031155736495

且发现免疫细胞都在外围：

image-20221031155934084

说明FAP+ 和SPP1+ 可能会限制免疫细胞到肿瘤中央。

FAP+ fibroblasts and SPP1+ macrophages 互作

利用nichenet 细胞通讯鉴定：

In addition, FAP+ fibroblasts enhanced the proinflammatory activity of SPP1+ macrophages via the expression of TGF-β superfamily genes, TGFB1 and INHBA, and TGF-β induced expression of ACVRL1, ACVR1,orACVR1B in SPP1+ macrophages (Fig. 7a). Furthermore, FAP+ fibroblasts enhanced the recruitment of SPP1+ macrophages through the WNT5AFZD2 and CCL3-CCR5 pairs. Of note, FAP+ fibroblasts interacted with SPP1+ macrophages through RARRES2-CMKLR1

through the adhesive ligand-receptor pairs COL1A1/LAMA1ITGB1 (

image-20221031160526792

ps：这张图也不错啊，展示两个方向的结果。

FGFR2+ fibroblasts or ICAM1+ fibroblasts as reference cell types for ligand-receptor analysis. 这是为啥呢？

When evaluating the regulatory network of FAP+ fibroblasts on SPP1+ macrophages, THBS1+ macrophages were considered as reference receiver cells due to their potential differentiation trajectory to SPP1+ macrophages. Meanwhile, FGFR2+ fibroblasts and ICAM1+ telocytes were used as reference cells to check the regulatory potential of SPP1+ macrophages on FAP+ fibroblasts. Nichenet_output$ligand_activity_target_heatmap was used to plot Ligands regulatory activity.

这个所谓的ref 应该就是nichenetr/seurat_wrapper.md at master · saeyslab/nichenetr[8] 里的两种状态吧：

image-20221031162327437

探究哪些配受体可能影响它的分化。

We further determined the relative expression of RARRES2 across MSC clusters and found this gene to be expressed at a higher level in FAP+ fibroblasts than other MSC clusters

image-20221031162936896

Chemerin, encoded by RARRES2, was shown to be an independent risk factor for CRC and has the ability to affect macrophage polarization in the DSS-induced colitis model62,63. In addition, CMKLR1, encoding the receptor for chemerin, showed higher expression in both THBS1+ macrophages and SPP1+ macrophages

果然下面和拟时序关联了，Since RNA “velocity” analysis indicated that SPP1+ macrophages can be differentiated from THBS1+ macrophages (Fig. 4e), FAP+ fibroblasts might function as a driver, promoting SPP1+ macrophage differentiation through chemerin。

Furthermore, we found that chemerin levels were significantly higher in the plasma of CRC patients than in healthy donors (Fig. 7d), suggesting that chemerin may serve as a predictive marker for CRC.

此外，Genes of extracellular matrix (ECM)-related pathways were highly expressed in FAP+ fibroblasts and SPP1+ macrophages, suggesting that either FAP+ fibroblasts or SPP1+ macrophages may facilitate the generation of desmoplastic structures (Supplementary Figs. 4f and 6g).

we investigated whether SPP1+ macrophages promote the ECM remodeling ability of FAP+ fibroblasts.

In addition, TGFB1 encoding protein bound to receptors encoded by TGFBR3, ACVRL1, and TGFBR1 on FAP+ fibroblasts, whereas ligands encoded by IL1B or IL1A interacted with receptors encoded by IL1R1 or IL1RAP on FAP+ fibroblasts (Fig. 7g), resulting in the expression of target genes encoding collagen or matrix metallopeptidase in these cells (Fig. 7h). These targets are important components of desmoplastic reactions, and 35 of the 100 predicted targets encoded components of ECM pathways 关联到了与ECM 相关的基因上：

image-20221031163644994

接着构建了调控网络：

We further explored the signaling pathways among the top three ligands (encoded by TGFB1, IL1A, and IL1B) and ECM-related targets, and we found 40 downstream regulators, including HIF1A, AKT1, STAT1, and NFKB1, those are involved in signaling pathways connecting ligands secreted by SPP1+ macrophages and ECM-target genes that contribute to the formation of the desmoplastic region

用scMLnet 吗？

image-20221031164409965

Taken together, our findings indicate that FAP+ fibroblasts and SPP1+ macrophages form an interaction network supporting each other’s maintenance and function. These two cell types may play important roles in the remodeling of ECM, potentially promoting the formation of the desmoplastic region of the TME

还是需要背景啊。

FAP+ 和SPP1+ 浸润和免疫治疗耐受有关

Thetumorimmune microenvironment can generally be classified into immune-excluded, inflamed (referred to as “hot” tumors, which respond to immunotherapy well), and immune-desert types based on the infiltration of immune cells and CD8+ Tcel

其他组学研究：

image-20221031170251202

说明双高的肿瘤是suggesting that the micro-environmental characteristics of this certain type of tumors is immune-exclusive with high infiltration of both FAP+ fibroblasts and SPP1+ macrophages.

免疫治疗印证了这个结果：

image-20221031170446466

随想

还提供了示例代码！

youqiongye/CRC_scRNAseq at V1.0.0[9]

这个空转数据，用的有点儿鸡肋啊。工作量其实也不大，创新点还是使用了nichenet 这个工具吧。不过也都是常规用法啊。（这个空转结合拟时序结果的使用，还是值得学习的）

不过他喵的就nichenet 代码不给：

image-20221031171119811

感觉就是加了个空转的噱头。

参考资料

[1]

Single-cell and spatial analysis reveal interaction of FAP+ fibroblasts and SPP1+ macrophages in colorectal cancer | Nature Communications: https://www.nature.com/articles/s41467-022-29366-6#Abs1

[2]

10X单细胞联合空间转录组技术解析结直肠癌肿瘤微环境 - 知乎 (zhihu.com): https://zhuanlan.zhihu.com/p/533147797

[3]

单细胞和空间转录组分析揭示结直肠癌中FAP+成纤维细胞和SPP1+巨噬细胞的相互作用 - 知乎 (zhihu.com): https://zhuanlan.zhihu.com/p/538571181

[4]

http://jaspar.genereg.net/: http://jaspar.genereg.net/

[5]

http://www.transcriptionfactor.org/: http://www.transcriptionfactor.org/

[6]

http://bioinfo.life.hust.edu.cn/AnimalTFDB/: http://bioinfo.life.hust.edu.cn/AnimalTFDB/

[7]

http://www.fiserlab.org/tf2dna_db/: http://www.fiserlab.org/tf2dna_db/

[8]

nichenetr/seurat_wrapper.md at master · saeyslab/nichenetr: https://github.com/saeyslab/nichenetr/blob/master/vignettes/seurat_wrapper.md

[9]

youqiongye/CRC_scRNAseq at V1.0.0: https://github.com/youqiongye/CRC_scRNAseq/tree/V1.0.0