A single-cell transcriptome atlas during Cashmere goat ...morphogenesis mainly used the mouse as a...

A single-cell transcriptome atlas during Cashmere goat hair 1

follicle morphogenesis 2

Wei Ge a, Wei-Dong Zhang a, Yue-Lang Zhang a, Yu-Jie Zhenga, Fang Li a, Shan-He Wang a, 3

Jin-Wang Liu c, Shao-Jing Tan b, Zi-Hui Yan b, Lu Wang b, Wei Shen b, Lei Qu c, Xin Wang a 4

5

a Key Laboratory of Animal Genetics, Breeding and Reproduction of Shaanxi Province，6

College of Animal Science and Technology, Northwest A&F University, Yangling, Shaanxi 7

712100, China; 8

b College of Life Sciences, Qingdao Agricultural University, Qingdao 266109, China; 9

c Life Science Research Center, Yulin University, Yulin, Shaanxi 719000, China 10

11

12

¶ Correspondence and reprint requests to: 13

Prof. Xin Wang; E-mail: [email protected] 14

15

16

.CC-BY-NC-ND 4.0 International license(which was not certified by peer review) is the author/funder. It is made available under aThe copyright holder for this preprintthis version posted January 31, 2020. . https://doi.org/10.1101/2020.01.30.926287doi: bioRxiv preprint

https://doi.org/10.1101/2020.01.30.926287http://creativecommons.org/licenses/by-nc-nd/4.0/

Abstract: 17

Cashmere, also known as soft gold, is produced from the secondary hair follicles in 18

Cashmere goats and it’s therefore of significance to investigate the molecular profiles 19

during Cashmere goat hair follicle development. However, our current understanding of the 20

machinery underlying Cashmere goat hair follicle remains largely unexplored and 21

researches regarding hair follicle development mainly used the mouse as a research model. 22

To provides comprehensively understanding on the cellular heterogeneity and cell lineage 23

cell fate decisions, we performed single-cell RNA sequencing on 19,705 single cells from 24

induction (embryonic day 60), organogenesis (embryonic day 90) and cytodifferentiation 25

(embryonic day 120) stages of fetus Cashmere goat dorsal skin. Unsupervised clustering 26

analysis identified 16 cell clusters and their corresponding cell types were also 27

unprecedentedly characterized. Based on the lineage inference, we revealed detailed 28

molecular landscape along the dermal and epidermal cell lineage developmental pathways. 29

Notably, by cross-species comparasion of single cell data with murine model, we revelaed 30

conserved programs during dermal condensate fate commitment and the heterochrony 31

development of hair follicle development between mouse and Cashmere goat were also 32

discussed here. Our work here delineate unparalleled molecular profiles of different cell 33

populations during Cashmere goat hair follicle morphogenesis and provide a valuable 34

resource for identifying biomarkers during Cashmere goat hair follicle development. 35

36

Key Words: Single-cell transcriptome; Developmental trajectories; Hair follicle 37

morphogenesis 38



Introduction 39

Every year, more than 20,000 tons of Cashmere were generated in China and Cashmere 40

goat has become an important source of income for the people lived in north China (Scott 41

Waldron et al., 2014). Cashmere is the secondary hair follicles in Cashmere goats and forms 42

as early as the fetus stage and due to the commercial value of Cashmere (Ansari-Renani et 43

al., 2011, Geng et al., 2013), it’s therefore of great significance to reveal molecular 44

pathways during early hair follicle development in Cashmere goats. Besides, by using 45

mouse as a research model, research has demonstrated that molecular pathways during 46

early hair follicle morphogenesis play important roles in regulating the hair characteristics, 47

including hair fiber length, fineness and curvature (Duverger & Morasso, 2009), therefore, 48

further enhancing the significance of revealing the molecular pathways driving hair follicle 49

morphogenesis in Cashmere. However, due to the long-time duration of pregnancy (about 50

145~159 days) in Cashmere goats (AJ Ritar et al., 1989), researches focused on hair follicle 51

morphogenesis mainly used the mouse as a research model, while our current 52

understanding on hair follicle morphogenesis in Cashmere goat remains largely unknown. 53

Similar to murine hair follicle development, Cashmere goat hair follicle in uterus 54

development can also be divided into three main stages: induction stage (about embryonic 55

day 55 - 65), organogenesis (about embryonic day 85 - 95) and cytodifferentiation stages 56

(around embryonic day 115) (Zhang Y et al., 2006). In mice, the molecular underpinnings 57

underlying the induction stage has recently been comprehensively investigated by virtue of 58

the development of single-cell RNA sequencing technology, while the late two stages 59

remain not well-known (Saxena et al., 2019). According to what is known in mice, the 60



formation of placodes and dermal condensates (DC) are two marking events in the 61

induction stage and requires a conserved crosstalk between dermal and epidermal cell 62

populations, including Wnt/β-catenin signaling, Edar signaling and Fgf signaling (Chen et 63

al., 2012, Huh et al., 2013, Zhang et al., 2009). More recently, Mok et al., demonstrated that 64

murine DC formation can be further divided into three sub-stages using single-cell RNA 65

sequencing (scRNA seq) technology and characterized detailed transcriptome signature 66

genes during each substage, they also found that ECM/Adhesion signaling was vital for the 67

early DC fate commitment (Mok et al., 2019). The organogenesis stage is characterized by 68

the formation of dermal papilla (DP) cells, hair shaft and inner root sheath (IRS) and the 69

molecular pathways involved includes PDGFα signaling and Shh signaling (Karlsson et al., 70

1999, Ouspenskaia et al., 2016). For the cytodifferentiation stage, the differentiation of IRS, 71

hair shaft and keratinocyte become obvious and Eda signaling is demonstrated to play a 72

role (Duverger & Morasso, 2009, Millar, 2002). Noteworthy, the asynchronous 73

development of different hair follicles, including guard hair follicles (starts from E13.5), 74

awl, auchene hair follicles (starts from E15.5), and zigzag hair follicle (starts from E17.5) 75

in mice, is also a marking event during the cytodifferentiation stage (Schlake, 2007). 76

However, the molecular machinery underlying the asynchronous development of different 77

hair follicles remains not well-known (Chi et al., 2013, Driskell et al., 2009). 78

To preliminarily reveal molecular pathways involved during Cashmere goat hair 79

follicle morphogenesis, several groups have collected skin samples from fetus goat and 80

performed transcriptome sequencing analysis to reveal gene expression dynamics between 81

different time points (Gao et al., 2016, Ren et al., 2016). However, due to the lack of 82



conserved markers to label particular cell types within the hair follicles, most studies used 83

skin tissues to perform transcriptome sequencing analysis and generated the “equalized” 84

expression matrices, which is sometimes, hard to reveal the real scenario. The paucity of 85

information regarding the cell heterogeneity within the hair follicles has obviously become 86

the main obstacle in dissecting the hair follicle morphogenesis. scRNA seq has recently 87

became robust tool in dissecting cell heterogeneity and several groups have also 88

successfully used scRNA seq technology to reveal the molecular machinery underlying 89

murine hair follicle development (Ge et al., 2019, Gupta et al., 2019, Mok et al., 2019), 90

further emphasized its application prospect in hair follicle development-related researches. 91

Tackling the paucity of information regarding the cellular heterogeneity and molecular 92

pathway underlying key cell fate decisions during Cashmere hair follicle development. 93

Here, we reported a single-cell transcriptome landscape during Cashmere goat hair follicle 94

morphogenesis based on 19,705 single-cell transcriptional profiles. We successfully 95

identified different cell types during Cashmere goat hair follicle development and 96

delineated their cell type-specific gene expression profiles, which provides valuable 97

information for the identification of biomarkers and dissecting cellular heterogeneity during 98

Cashmere goat hair follicle development. Besides, cell lineage inference analysis provides a 99

comprehensively understanding of the molecular pathways underlying major cell lineage 100

fate decisions. Our study here provides a valuable resource for understanding Cashmere 101

goat hair follicle development, and will also have implications for future Cashmere goat 102

breeding. 103

104



Results 105

Single-cell sequencing and characterization of cellular heterogeneity during Cashmere 106

goat hair follicle morphogenesis 107

To provide in-depth insight into molecular profiles during Cashmere goat hair follicle 108

development and main cell fates transitions, we collected skin samples from E60, E90 and 109

E120 stage fetus Cashmere goat skin (Supplemental. Fig. S1a), which correspond to hair 110

follicle induction, organogenesis and cytodifferentiation stage and performed single-cell 111

RNA sequencing (Fig. 1a). We totally captured 7,000 single cells for each sample, and for 112

each sample we detected at least 16,000 genes and the genome mapping rate was higher 113

than 90% for all the samples (Supplemental. Fig. S1b). For quality control, we filtered cells 114

according to the number of genes detected (Supplemental. Fig. S1c) and retained 115

high-quality cells for downstream analysis. After quality control, we totally analyzed 116

19,705 single-cell transcriptome expression profiles from E60 (6,825 single cells), E90 117

(6,873 single cells) and E120 (6,007 single cells) stage fetus Cashmere goat back skin. 118

To dissect cellular heterogeneity, we next performed t-distributed stochastic neighbor 119

embedding (t-SNE) analysis and we totally identified 16 different cell clusters across three 120

developmental times (Fig. 1b,c and Supplemental. Table 1). By analyzing cluster-specific 121

expressed gene expression, we successfully identified the different cell types according to 122

their marker gene expression (Supplemental. Fig. S1d). Briefly, we found that cluster 1, 4, 6, 123

7 and 13 expressed high level of dermal cell lineage markers LUM and COL1A1 (Gupta et 124

al., 2019), while cluster 0, 2, 3, 5, 8, 9 and 12 expressed high level of epithelial lineage 125

markers KRT14 and KRT17 (Gu & Coulombe, 2007, Joost et al., 2016). Besides, we also 126



identified hair shaft clusters (LHX2 and MSX1, cluster 2, 5) (Yang et al., 2017), endothelial 127

cluster (KDR and PECAM1, cluster 11) (Detmar et al., 1998), DP cluster (SOX2 and SOX18, 128

cluster 13) (Driskell et al., 2009), pericyte cluster (ACTA2 and TPM2, cluster 15) 129

(Paquet-Fifield et al., 2009), muscle cell cluster (CNMD and ARS1, cluster 17) and 130

macrophage cell cluster (ALF1 and RGS1, cluster 16) (Lee et al., 2016). More importantly, 131

we further delineated transcriptional characteristics for each cell types and identified a 132

series of cell type-specific marker genes during Cashmere goat hair follicle development 133

(Fig. 1d), and it was also worth noting that many cell type-specific expressed marker genes 134

were also consistent with a murine scenario, such as dermal cell markers POSTN, DCN, 135

APOE, epithelial cell markers KRT14, KRT15, and DP cell markers SOX2, SOX18. 136

137

Defining dermal cell lineage and epidermal cell lineage developmental trajectory 138

along pseudoptime 139

After the characterization of different cell clusters, we then want to investigate major cell 140

fate transitions during hair follicle development. We, therefore, performed pseudotime 141

trajectory construction analysis on dermal and epidermal cell clusters (Fig. 2). Since we 142

have successfully characterized all cell clusters, we then selected dermal cell lineage cell 143

clusters (Fig. 2a, cluster 1, 4, 6, 7 and 12) and epidermal cell lineage cell clusters (Fig. 2b, 144

cluster 0, 2, 3, 5, 8, 9 and 11) to infer cell lineage developmental trajectory. For the dermal 145

cell lineage, pseudotime trajectory displayed 2 branch points (Fig. 2c), while the epidermal 146

cell lineage showed 3 branch points (Fig. 2d). Noteworthy, when the cells were color-coded 147

with their corresponding developmental time, they also showed a time-ordered pattern 148



along the pseudotime. As for the branch point in the dermal and epidermal cell populations, 149

based on the prior knowledge on cell lineage dynamics, during early hair follicle 150

development, the dermal cell fate involves DC fate commitment and DP fate commitment 151

(Fig. 2e) (Saxena et al., 2019), while epidermal cell fate involves matrix cell fate 152

commitment, hair shaft/IRS fate commitment and keratinocyte fate commitment (Fig. 2f) 153

(Forni et al., 2012, Millar, 2002, Schmidt-Ullrich & Paus, 2005), it’s therefore different 154

branch points may represent the process of cell fate decisions. 155

156

Delineating developmental pathway during DC fate commitment 157

After dermal cell lineage trajectory inference, we firstly focused on the first branch point on 158

the dermal cell pseudotime trajectory to reveal the first dermal cell fate decision. By 159

analyzing gene expression dynamics along pseudotime, we observed 2,679 differentially 160

expressed genes at the end of cell fate commitment (Supplemental. Table 2) and gene 161

functional enrichment analysis revealed that these genes enriched GO terms of “tissue 162

morphogenesis, response to growth factor and cell morphogenesis involved in 163

differentiation” (Fig. 3a). A comparison of overlapped GO terms between each gene set 164

showed that they shared substantial GO terms (Supplemental. Fig. 2). Of particular notice, 165

we observed a series of canonical murine dermal condensate cell markers such as APOD, 166

LUM and APOE (Mok et al., 2019) in those differentially expressed genes, and the 167

pseudotime expression pattern of DC markers APOD, SOX18, CTNNB1 and SOX2 was 168

increased along pseudotime (Fig. 3b), therefore, we termed the first branch point as DC fate 169

commitment. 170



To our knowledge, no reports yet have delineated cellular heterogeneity and 171

transcriptional landscape during Cashmere goat hair follicle morphogenesis, and most 172

researches regarding hair follicle development have been performed on the murine model. 173

To gain in-depth insight into machinery driving DC fate commitment, we then compared 174

DC fate signature genes with recently reported murine DC fate commitment signature 175

genes (Fig. 3c) and observed 729 (about 14.5% of all) overlapped genes between Cashmere 176

goat and murine DC signature genes (Supplemental. Table 3). Noteworthy, based on 177

single-cell RNA sequencing on E15.0 dorsal skin, Mok et al. recently demonstrated that 178

murine DC fate commitment can be divided into pre-DC, DC1 and DC2 stage at a more 179

detailed level (Mok et al., 2019). By analyzing murine DC marker genes at different stages, 180

we similarly found that DC signature genes of Cashmere goat also showed chronological 181

expression patterns along pseudotime (Fig. 3d). Briefly, murine pre-DC markers DKK1 and 182

LEF1 were elevated prior to DC1 and DC2 marker expressions, such as PRDM1, TRPS1, 183

INHBA and RSPO3, it’s therefore plausible that DC fate commitment in Cashmere goat 184

may also involve different stages. 185

186

Delineating DP cell heterogeneity along pseudotime 187

After defining DC fate commitment, we then focused on the next branch point. We firstly 188

compared differentially expressed gene expression between the two branches and observed 189

that cell fate 1 elevated canonical DP marker genes, such as APOD, SOX18, and enriched 190

GO terms of “tissue morphogenesis and epidermis development”, while cell fate 2 elevated 191

genes such as CENPW, TOP2A and enriched GO terms of “mitotic cell cycle process and 192



DNA-dependent DNA replication” (Fig. 4a,b and Supplemental. Fig. 3a,b). To gain an 193

in-depth understanding of the differences between the two branches, we further identified 194

differentially expressed genes between the two branches using another differential analysis 195

method (Wilcox, Wilcoxon Rank Sum test). Consistent with Monocle analysis, cell fate 1 196

also showed higher expression of APOD, SOX18 and IGF1, while cell fate 2 showed 197

elevated expression of CENPW, TOP2A and DCN (Fig. 4c). GO enrichment analysis 198

similarly revealed that differentially expressed genes in cell fate 1 enriched GO terms of 199

“tissue morphogenesis and epithelial cell differentiation”, while cell fate 2 enriched GO 200

mainly related to the regulation of cell cycle (Supplemental. Fig. 3c). 201

To dissect the heterogeneity within DP cells, we further performed 202

immunofluorescence analysis on E120 Cashmere goat back skin tissues and observed 203

different staining patterns between primary hair follicles (PF) and secondary hair follicles 204

(SF) (Fig. 4d). For the BMP2 staining, we found that BMP2 specifically expressed in the 205

matrix cells surrounding DP cells, while in the secondary hair follicle, BMP2 positive cells 206

could be found in both DP cells and surrounding matrix cells. Besides, we found that 207

PCNA, a cell cycle-related marker, was expressed mainly in surrounding matrix cells in the 208

primary hair follicle, while in the secondary hair follicles, we observed high expression of 209

PCNA both in DP cells and surrounding matrix cells. Taken together, our analysis here 210

demonstrated that different hair follicles showed different gene expression patterns in 211

Cashmere goat, which may also plausible for the asynchronous development of different 212

hair follicles. 213

214



Delineating developmental pathway during the first epidermal cell fate decision 215

After revealing dermal cell fate decisions, we then focused on the epidermal cell clusters. 216

Monocle pseudotime trajectory inference analysis revealed that epidermal cells showed 217

three different branch points. We then firstly focused on the first branch point and analyzed 218

differentially expressed gene dynamics along pseudotime. Based on k-means clustering, we 219

observed four distinct gene clusters. As expected, we observed a series matrix cell markers 220

at the end of pseodotime, including HOXC13, KRT25 and KRT71, thus deciphering a 221

matrix cell fate commitment (Fig. 5a). Gene functional enrichment analysis showed that 222

matrix cell expressed signature genes enriched GO terms of “keratinocyte differentiation, 223

epidermis development, and skin epidermis development” (Fig. 5b), while comparison of 224

GO terms between different gene set reveals that gene set 1,2 differs to that of gene set 3,4 225

(Supplemental. Fig. 4b). 226

To gain in-depth insight into molecular profiles during matrix cell fate commitment, we 227

further analyzed the signature gene expression pattern along pseudotime. For the gene set 4, 228

namely matrix cell fate, they enriched a series of keratin family genes, such as KRT25, 229

KRT27, KRT84, and their expression was increased along pseudotime (Fig. 5c). For the 230

gene set 3 and 2, we found that they transiently elevated expression of LEF1, SBSN, SOX18, 231

SOX9, and enriched GO terms of “cardiac epithelial to mesenchymal transition, 232

mesenchymal cell differentiation and skin development, morphogenesis of an epithelium, 233

respectively. For the gene set 1, they enriched genes such as VIM, LUM and COL1A1 and 234

both showed decreased expression along pseudotime. Immunohistochemistry staining 235

analysis also confirmed that LEF1 and CTNNB1 were expressed in the upper epidermis, 236



which was also consistent with a murine scenario (Polakis, 2001, Tsai et al., 2014). 237

238

Delineating cell fate decisions during hair shaft and IRS cell fate commitment 239

After deciphering matrix cell fate commitment in the first epidermal trajectory point, we 240

next focused on the next branch point. By analyzing differentially expressed genes along 241

pseudotime, we found that cell fate 1 enriched canonical hair shaft markers, including SHH, 242

VDR, and HOXC13, while cell fate 2 enriched canonical IRS markers, such as SOX9, 243

KRT14, SBSN and LEF1 (Fig. 6a) (Yang et al., 2017). Our immunofluorescence results also 244

confirmed their expression in the Cashmere hair follicles (Supplemental. Fig. 5a). For the 245

pre-branch, we observed genes such as LUM, COL1A, OGN, SOX18 and they all showed 246

decreased expression along pseudotime (Fig. 6b). For the hair shaft fate (cell fate 1), we 247

also observed elevated expression of DCN, TOP2A, CENPW and H2AFZ and enriched GO 248

terms of “mitotic cell cycle process and DNA replication” (Supplemental. Fig. 5 a,b). For 249

the IRS cell fate (cell fate 2), we observed elevated expression of PRDM1, KRT1, SOX9, 250

KRT14 and enriched GO terms of “supramolecular fiber organization and tissues 251

morphogenesis”. 252

Besides, we also compared our identified hair shaft and IRS signature genes with our 253

recently identified murine hair shaft and IRS signature genes (Ge et al., 2019). Interestingly, 254

we only observed about ~8.8% overlapped IRS signature genes between mouse and goat 255

(Fig. 6c and Supplemental. Table 4), and these overlapped genes mainly consisted of 256

Keratin family genes, such as KRT14, KRT17, KRT79. As for overlapped hair shaft 257

signature genes, we only observed ~4.6% genes (Fig. 6d), including MSX1, HOXC13, 258



APOD, LHX2, SHH, and VDR. To further validate our analysis, we compared the 259

expression of PCNA and VDR between E90 Cashmere goat skin and E16.5 mouse skin 260

tissues, and the immunohistochemistry showed that they showed similar expression 261

patterns during hair follicle development (Fig. 6e). These data together demonstrate that 262

hair shaft and IRS specification may require a conserved program during cell fate decisions. 263

264

Revealing the developmental pathway during keratinocyte cell fate commitment 265

After delineating the first two epidermal cell lineage cell fate decisions, we then focused on 266

the last branch point. Pseudotime trajectory analysis also revealed two different branches, 267

and we then analyzed pseudotime gene expression dynamics between the two branches (Fig. 268

7a). Analyzing differentially expressed genes along psusotime, we found that cell fate 1 269

enriched genes such as VDR, BMP4, STAR, KRT85 and KRT14, while cell fate 2 enriched 270

genes such as BMP2, SHH, CUX1 and ETV5. Noteworthy, KRT14 has been identified as a 271

marker for keratinocyte both in humans and mice (Green et al., 2003, Joost et al., 2016), we, 272

therefore, termed this fate as keratinocytes. To gain in-depth insight into their 273

corresponding cell type, we performed gene functional enrichment analysis for each gene 274

set and analyzed pseudotime GO enrichment dynamics (Fig. 7b and Supplemental. Fig. 6a). 275

The result showed that the pre-branch mainly enriched genes related to “regulation of 276

response process”, while for the cell fate 1, the signature genes mainly enriched GO terms 277

of “development differentiation of epidermal”. Interestingly, for cell fate 2, these 278

differentially expressed genes mainly enriched in GO terms of “regulation of cell cycle” 279

and showed lower expression of KRT14 (Fig. 7c), it’s therefore plausible that keratinocyte 280



differentiation at this stage is not synchronized. To confirm such hypothesis, we performed 281

immunofluorescence analysis of VDR, PCNA, BMP2, CTNNB1 and LEF1, and the results 282

showed that the expression of VDR and BMP2 in the outer layer epidermis was not 283

homogeneous, with some cell clusters showed higher expression while some cell 284

populations showed lower expression (Fig. 7d). For the pre-branch enriched CTNNB1, it 285

was uniformly expressed in the interfollicular epidermis. Besides, the expression pattern of 286

PCNA and LEF1 was even more significant and they were partially expressed in the 287

epidermis. Taken together, these results together emphasize that keratinocyte differentiation 288

in Cashmere goat is asynchronous and requires different gene expression profiles. 289

290

Discussion 291

The development of scRNA seq in recent years has been demonstrated as a robust tool for 292

the developmental biologist to investigate organogenesis and has provided us with 293

unparalleled insight into mammalian development. In the past decades, the number of 294

papers using scRNA seq based technology has increased exponentially and scRNA has also 295

been awarded Science's 2018 Breakthrough of the Year (Angerer et al., 2017, Pennisi, 296

2018). Here, we successfully constructed single-cell atlas during Cashmere goat hair 297

follicle development. Based on the downstream analysis, we provided unparalleled insight 298

into the cellular heterogeneity and major cell fate decisions during the Cashmere hair 299

follicle in uterus morphogenesis. As far as we have known, this is the first study to 300

comprehensively delineating molecular profiles of various cell types and revealing major 301

cell fate decisions during Cashmere hair follicle development. More importantly, by 302



analyzing cluster-specific gene expression profiles, our data here provides a valuable 303

resource for identification of markers for future studies in Cashmere skin tissues and also 304

provides an in-depth understanding of the Cashmere hair follicle development. 305

In the current study, we totally analyzed 19,705 single-cell transcriptional profiles from 306

three different points, which can represent major cell types during hair follicle development. 307

To dissect cellular heterogeneity during Cashmere goat hair follicle development after tSNE 308

analysis, we analyzed cluster-specific expressed signature genes for each cluster to infer 309

their corresponding cell types. It’s worth noting that all the cell markers used in the current 310

study were referenced from the murine model due to the paucity of information regarding 311

marker gene expression during Cashmere goat hair follicle development. However, our 312

study here showed that substantial murine cell type-specific biomarkers are identical to the 313

Cashmere goat. Besides, by comparison of cell type-specific signature genes between 314

Cashmere goat and mouse (Fig. 3c and Fig. 6c,d), we found that the transcriptome 315

similarity decreased along pseudotime (revealed by overlapped signature genes in this 316

study). Briefly, about 729 DC signature genes overlapped between Cashmere goat and 317

mouse, and murine DC signature genes at different stages (pre-DC, DC1 and DC2) showed 318

similar expression pattern along pseudotime in Cashmere goat (Mok et al., 2019). While for 319

the late-stage IRS and hair shaft cell population in the late stage of hair follicle 320

development, the percentage of overlapped genes decreased (126 and 93 overlapped genes, 321

respectively). Similarly, by comparing gene expression across 7 different species from brain, 322

heart, ovary, kidney, testis and liver, researches have demonstrated that organ-specific 323

molecular profiles are more similar in early development and become more distinct during 324



development (Cardoso-Moreira et al., 2019), it’s therefore plausible that hair follicle 325

development in Cashmere goat and mouse may also consistent with such theory. 326

Similar to the murine model, we also observed differential gene expression profiles in 327

different DP cell populations from different hair follicles (primary hair follicles vs 328

secondary hair follicles), which was also consistent with our recent findings on mice (Ge et 329

al., 2019). Besides, an in-depth comparison of differentially expressed genes revealed that 330

different DP cell populations require different transcriptional profiles, and it’s, therefore, 331

plausible that the different induction signals may be involved during the asynchronous 332

development of hair follicles. Consistent with such hypothesis, researches found that SOX2 333

was specifically expressed in guard hair follicles, but not zigzag hair follicles, while SOX18 334

was demonstrated to regulate zigzag hair follicle morphogenesis (Driskell et al., 2009, 335

James et al., 2003, Pennisi et al., 2000). Although our study here also demonstrates the 336

asynchronous development of different hair follicles in Cashmere goat, the detailed 337

machinery underlying such phenomenon was not investigated here and future studies may 338

focus on such topics, which will definitely provide us new insight into the hair follicle 339

biology and hair follicle regeneration. 340

Based on single-cell pseudotime trajectory inference, our study here also highlighted 341

the underappreciated cell fate decisions during hair follicle development. Different from 342

our previous studies performed in mice, our pseudotime lineage trajectory of epidermal cell 343

lineage showed three different branch points while murine epidermal cell lineage trajectory 344

showed two branch points. by analyzing the gene expression profiles of the first two branch 345

points, they actually showed similar cell fate decisions, while the additional branch point in 346



Cashmere goat mainly comes from keratinocyte differentiation. Such differences in 347

pseudotime trajectory may be partially explained by the differences in the development of 348

hair follicles across species, for example, E120 stage Cashmere goat showed obvious hair 349

fiber on the surface of the skin, while it was not until 6-7 postnatal day that the hair follicles 350

in the mouse skin surface become visible. The difference in the pseudotime trajectory 351

reveals that hair follicle development was heterochrony between Cashmere goat and mouse, 352

which was frequent when comparing the development of specific organs across species. 353

Finally, our dataset here provides an important resource for understanding the cellular 354

heterogeneity and major cell fate decisions during Cashmere goat hair follicle development. 355

For the first time, the detailed transcriptional landscape of different cell populations was 356

delineated here at single-cell resolution, which provided valuable resources for the 357

identification of biomarkers. Besides, our trajectory inference analysis here successfully 358

recapitulated major cell fate decisions during Cashmere goat hair follicle development, 359

which enabled us to comprehensively study the detailed developmental pathways involved 360

during Cashmere goat hair follicle morphogenesis, and will also have implications for the 361

Cashmere goat breeding work in animal husbandry. 362

363



Materials and methods 364

Experimental Animals 365

All the experimental Shaanbei White Cashmere goats involved in this study were obtained 366

from the Shaanbei Cashmere Goat Engineering Technology Research Center of Shaanxi 367

Province and were fed with Cashmere goat feeding standard (DB61/T583-2013) of Shaanxi 368

Province. All pregnant goats were prepared using artificial insemination and all the 369

experimental procedures involved goats in this study were approved by the Experimental 370

Animal Manage Committee of Northwest A&F University. 371

Single-cell suspension preparation 372

The goat fetus at desired dates was isolated using cesarean operation when the pregnant 373

goats were anesthetic with the compound ketamine. The skin tissues (0.5 cm × 0.5 cm) 374

were isolated from the fetus back skin and were immediately transferred to the ice-cold 375

DMEM/F12 media (Gibco, Beijing, China) with 50 U/ml penicillin and 50 mg/ml 376

streptomycin (HyClone, Beijing, China). After washing three times with DMEM/F12 to 377

remove contaminative blood cells, the skin tissues were then dissociated into single cells 378

prior to sequencing. For E60 and E90 skin tissues, the obtained skin tissues were firstly 379

incubated with 2 mg/ml collagenase IV (Sigma, St Louis, MO, USA) for 30 min at 37°C, 380

and then the skin tissues were mechanically dissociated into single-cell suspensions with a 381

1ml pipette tip. For E120 skin tissues, the skin tissues were firstly cut into ~3 mm skin 382

pieces and then incubated with 2 mg/ml collagenase IV for 30 min. After incubation, the 383

hair follicles within the skin tissues were isolated with a pair of precise forceps and the 384



pooled hair follicles were further dissociated into single cells with TypLE Express (Gibco, 385

Grand Island, NY, USA) for 30 min at 37°C. The obtained single-cell suspensions were 386

then washed three times with PBS supplemented with 0.04% BSA (Sigma, St Louis, MO, 387

USA) and were filtered with a 40 μm cell strainer (BD Falcon, BD Biosciences, San Jose, 388

CA, USA) to remove debris and cell aggregations. For each stage, the samples were 389

obtained from at least 2 different goat fetus and for each fetus goat, the single-cell 390

suspension was prepared separately until finally pooled together prior to single-cell 391

barcoding. 392

Single-cell library construction and sequencing 393

Single-cell library was constructed using 10x Genomics’ Chromium Single Cell 3’ V3 Gel 394

Beads Kit (10x Genomics, Pleasanton, CA, USA) according to the manufacture’s 395

instructions. After cell counting, the single-cell suspension was adjusted to 1000 cells/μl 396

and about 7,000 cells were obtained for each stage. The single-cell barcoding procedure 397

was performed using a 10x Genomics Chromium barcoding system (10x Genomics, 398

Pleasanton, CA, USA) according to the manufacturer's guide. After single-cell library 399

construction, the Illumina HiSeq X Ten sequencer (Illumina, San Diego, CA, USA) was 400

used to sequence and 150 bp pair ended reads were generated. 401

10x Genomics single-cell RNA sequencing data processing 402

The obtained raw sequencing files were processed with standard CellRanger (v2.2.0) 403

pipeline according to the 10x Genomics official guide (https://www.10xgenomics.com/cn/). 404

The produced raw base call files were firstly transformed into fastq files by using 405



Cellranger mkfastq function. The goat ARS1 reference genome downloaded from ensemble 406

was used as a reference genome (https://asia.ensembl.org/Capra_hircus/Info/Index). 407

Cellranger count function was used to perform mapping, filtering low-quality cells, 408

barcoding counting and UMI counting. 409

After the standard Cellranger pipeline, the generated gene expression matrice files were 410

then analyzed with Seurat (V2.3.4) package according to the official user guide 411

(https://satijalab.org/seurat/vignettes.html). Quality control was performed using FilterCells 412

function and cells with detected genes less than 200 and genes expressed less than 3 cells 413

were filtered. After normalization and data scaling, the different datasets were integrated by 414

using RunMultiCCA function. tSNE was used to perform dimension reduction analysis and 415

different cell clusters were identified by using FindClusters function. The cluster 416

specifically expressed genes were analyzed with FindAllMarkers function and with the 417

parameter “min.pct = 0.25, thresh.use = 0.25”. 418

Single-cell pseudotime lineage trajectory reconstruction 419

Single-cell lineage reconstruction analysis was performed using Monocle (V2) packages 420

according to the online tutorial (http://cole-trapnell-lab.github.io/monocle-release/docs/). 421

The monocle object was constructed from Seurat object with newCellDataSet function, and 422

Seurat determined variable genes were used as ordering genes to order cells in pseudotime 423

along a trajectory. Dimension reduction was performed using DDRTree methods. To 424

analyze differential gene expression between different cell branches, BEAM function was 425

used and differentially expressed genes were identified with q-val < 1e-4. Branch-specific 426

gene expression heatmap was plotted with plot_genes_branched_heatmap function and 427



different gene set was calculated according to k-means clustering. 428

Immunohistochemistry staining analysis 429

The immunofluorescence or enzyme horseradish peroxidase (HRP)-based 430

immunohistochemistry analysis procedure was performed as we previously described (Ge 431

et al., 2017, Liu et al., 2017). For immunofluorescence analysis, the paraffin-embedded skin 432

tissues were firstly deparaffinized in xylene and further rehydrated in ethanol solutions. 433

Antigen retrieval was performed in 0.01 M sodium citrate buffer at 96°C. Following a 434

permeabilization procedure in 0.5 M Tris-HCI buffer supplemented with 0.5% TritonX-100 435

(Sorlabio, Beijing, China) for 10 min, the slides were then blocked with 3 % BSA and 10 % 436

donkey serum (Boster, Wuhan, China) in 0.5 M Tris-HCI buffer for 30 min. The primary 437

antibodies diluted in the blocking buffer were incubated with the slides at 4°C overnight. 438

The next morning, the corresponding secondary antibodies were then added in the slides 439

and incubated at 37°C for 1 h. DAPI was used to stain the nuclei and the pictures were 440

taken using Nikon AR1 confocal system (Nikon, Tokyo, Japan). For HRP based 441

immunohistochemistry analysis, following antigen retrieval, the samples were firstly 442

incubated with 3% H2O2 for 10 min at room temperature to remove endogenous peroxidase 443

activity. The primary antibodies were incubated 4 °C overnight, and the corresponding 444

HRP-labeled secondary antibodies were added in the next morning for 40 min at room 445

temperature. After that, peroxidase substrate DAB (Zsbio, Beijing, China) was used for 446

chromogenic reaction with hematoxylin used to stain nuclei. The slides were finally 447

mounted with neutral resins and pictures were captured with an Olympus BX51 microscope 448

imaging system (Olympus, Tokyo, Japan). All the primary and secondary antibodies used in 449



this study were listed in Supplementary Table 5. 450

451

Data availability 452

The single-cell RNA sequencing data used in this research is deposited in NCBI GEO 453

databases under accession number: GSE144351. 454

455

Acknowledgments 456

This work was supported by the National Natural Science Foundation of China (31972556 457

and 31671554). 458

459

460



References 461

AJ Ritar, S Salamon, PD Ball, O'may P (1989) Ovulation and fertility in goats after intravaginal device-PMSG 462

treatment. Small Ruminant Research 2: 323-331 463

Angerer P, Simon L, Tritschler S, Wolf FA, Fischer D, Theis FJ (2017) Single cells make big data: New challenges 464

and opportunities in transcriptomics. Current Opinion in Systems Biology 4: 85-91 465

Ansari-Renani HR, Ebadi Z, Moradi S, Baghershah HR, Ansari-Renani MY, Ameli SH (2011) Determination of 466

hair follicle characteristics, density and activity of Iranian cashmere goat breeds. Small Ruminant Research 95: 467

128-132 468

Cardoso-Moreira M, Halbert J, Valloton D, Velten B, Chen C, Shao Y, Liechti A, Ascencao K, Rummel C, 469

Ovchinnikova S, Mazin PV, Xenarios I, Harshman K, Mort M, Cooper DN, Sandi C, Soares MJ, Ferreira PG, 470

Afonso S, Carneiro M et al. (2019) Gene expression across mammalian organ development. Nature 571: 471

505-509 472

Chen D, Jarrell A, Guo C, Lang R, Atit R (2012) Dermal beta-catenin activity in response to epidermal Wnt 473

ligands is required for fibroblast proliferation and hair follicle initiation. Development 139: 1522-33 474

Chi W, Wu E, Morgan BA (2013) Dermal papilla cell number specifies hair size, shape and cycling and its 475

reduction causes follicular decline. Development 140: 1676-83 476

Detmar M, Brown LF, Schon MP, Elicker BM, Velasco P, Richard L, Fukumura D, Monsky W, Claffey KP, Jain RK 477

(1998) Increased microvascular density and enhanced leukocyte rolling and adhesion in the skin of VEGF 478

transgenic mice. The Journal of investigative dermatology 111: 1-6 479

Driskell RR, Giangreco A, Jensen KB, Mulder KW, Watt FM (2009) Sox2-positive dermal papilla cells specify 480

hair follicle type in mammalian epidermis. Development 136: 2815-23 481

Duverger O, Morasso MI (2009) Epidermal patterning and induction of different hair types during mouse 482

embryonic development. Birth Defects Res C Embryo Today 87: 263-72 483

Forni MF, Trombetta-Lima M, Sogayar MC (2012) Stem cells in embryonic skin development. Biol Res 45: 484

215-22 485

Gao Y, Wang X, Yan H, Zeng J, Ma S, Niu Y, Zhou G, Jiang Y, Chen Y (2016) Comparative Transcriptome Analysis 486

of Fetal Skin Reveals Key Genes Related to Hair Follicle Morphogenesis in Cashmere Goats. PloS one 11: 487

e0151118 488

Ge W, Tan S-J, Wang S-H, Li L, Sun X-F, Shen W, Wang X (2019) Single-cell transcriptome profiling reveals 489

dermal and epithelium cell fate decisions during embryonic hair follicle development. bioRxiv 490

Ge W, Zhao Y, Lai FN, Liu JC, Sun YC, Wang JJ, Cheng SF, Zhang XF, Sun LL, Li L, Dyce PW, Shen W (2017) 491

Cutaneous applied nano-ZnO reduce the ability of hair follicle stem cells to differentiate. Nanotoxicology: 492

1-34 493

Geng R, Yuan C, Chen Y (2013) Exploring differentially expressed genes by RNA-Seq in cashmere goat (Capra 494

hircus) skin during hair follicle development and cycling. PloS one 8: e62704 495

Green H, Easley K, Iuchi S (2003) Marker succession during the development of keratinocytes from cultured 496

human embryonic stem cells. Proceedings of the National Academy of Sciences of the United States of 497

America 100: 15625-30 498

Gu LH, Coulombe PA (2007) Keratin function in skin epithelia: a broadening palette with surprising shades. 499

Current opinion in cell biology 19: 13-23 500

Gupta K, Levinsohn J, Linderman G, Chen D, Sun TY, Dong D, Taketo MM, Bosenberg M, Kluger Y, Choate K, 501

Myung P (2019) Single-Cell Analysis Reveals a Hair Follicle Dermal Niche Molecular Differentiation Trajectory 502

that Begins Prior to Morphogenesis. Developmental cell 48: 17-31 e6 503

Huh SH, Narhi K, Lindfors PH, Haara O, Yang L, Ornitz DM, Mikkola ML (2013) Fgf20 governs formation of 504



primary and secondary dermal condensations in developing hair follicles. Genes & development 27: 450-8 505

James K, Hosking B, Gardner J, Muscat GE, Koopman P (2003) Sox18 mutations in the ragged mouse alleles 506

ragged-like and opossum. Genesis 36: 1-6 507

Joost S, Zeisel A, Jacob T, Sun X, La Manno G, Lonnerberg P, Linnarsson S, Kasper M (2016) Single-Cell 508

Transcriptomics Reveals that Differentiation and Spatial Signatures Shape Epidermal and Hair Follicle 509

Heterogeneity. Cell Syst 3: 221-237 e9 510

Karlsson L, Bondjers C, Betsholtz C (1999) Roles for PDGF-A and sonic hedgehog in development of 511

mesenchymal components of the hair follicle. Development 126: 2611-21 512

Lee SB, Shim S, Kim MJ, Shin HY, Jang WS, Lee SJ, Jin YW, Lee SS, Park S (2016) Identification of a distinct 513

subpopulation of fibroblasts from murine dermis: CD73(-) CD105(+) as potential marker of dermal fibroblasts 514

subset with multipotency. Cell Biol Int 40: 1008-16 515

Liu JC, Lai FN, Li L, Sun XF, Cheng SF, Ge W, Wang YF, Zhang XF, De Felici M, Dyce PW, Shen W (2017) Di 516

(2-ethylhexyl) phthalate exposure impairs meiotic progression and DNA damage repair in fetal mouse 517

oocytes in vitro. Cell death & disease 8: e2966 518

Millar SE (2002) Molecular mechanisms regulating hair follicle development. The Journal of investigative 519

dermatology 118: 216-25 520

Mok KW, Saxena N, Heitman N, Grisanti L, Srivastava D, Muraro MJ, Jacob T, Sennett R, Wang Z, Su Y, Yang LM, 521

Ma'ayan A, Ornitz DM, Kasper M, Rendl M (2019) Dermal Condensate Niche Fate Specification Occurs Prior 522

to Formation and Is Placode Progenitor Dependent. Developmental cell 48: 32-48 e5 523

Ouspenskaia T, Matos I, Mertz AF, Fiore VF, Fuchs E (2016) WNT-SHH Antagonism Specifies and Expands Stem 524

Cells prior to Niche Formation. Cell 164: 156-169 525

Paquet-Fifield S, Schluter H, Li A, Aitken T, Gangatirkar P, Blashki D, Koelmeyer R, Pouliot N, Palatsides M, Ellis 526

S, Brouard N, Zannettino A, Saunders N, Thompson N, Li J, Kaur P (2009) A role for pericytes as 527

microenvironmental regulators of human skin tissue regeneration. The Journal of clinical investigation 119: 528

2795-806 529

Pennisi D, Gardner J, Chambers D, Hosking B, Peters J, Muscat G, Abbott C, Koopman P (2000) Mutations in 530

Sox18 underlie cardiovascular and hair follicle defects in ragged mice. Nature genetics 24: 434-7 531

Pennisi E (2018) Chronicling embryos, cell by cell, gene by gene. Science 360: 367 532

Polakis P (2001) More than one way to skin a catenin. Cell 105: 563-6 533

Ren H, Wang G, Chen L, Jiang J, Liu L, Li N, Zhao J, Sun X, Zhou P (2016) Genome-wide analysis of long 534

non-coding RNAs at early stage of skin pigmentation in goats (Capra hircus). BMC genomics 17: 67 535

Saxena N, Mok KW, Rendl M (2019) An Updated Classification of Hair Follicle Morphogenesis. Experimental 536

dermatology 537

Schlake T (2007) Determination of hair structure and shape. Seminars in cell & developmental biology 18: 538

267-73 539

Schmidt-Ullrich R, Paus R (2005) Molecular principles of hair follicle induction and morphogenesis. Bioessays 540

27: 247-61 541

Scott Waldron, Colin Brown, Komarek. AM (2014) The Chinese Cashmere Industry A Global Value Chain 542

Analysis. Development Policy Review 32: 589-610 543

Tsai SY, Sennett R, Rezza A, Clavel C, Grisanti L, Zemla R, Najam S, Rendl M (2014) Wnt/beta-catenin signaling 544

in dermal condensates is required for hair follicle formation. Developmental biology 385: 179-88 545

Yang H, Adam RC, Ge Y, Hua ZL, Fuchs E (2017) Epithelial-Mesenchymal Micro-niches Govern Stem Cell 546

Lineage Choices. Cell 169: 483-496 e13 547

Zhang Y, Yin J, C L (2006) Study on development of skin and hair follicle from fetal inner mongolian Arbas 548

cashmere goats. ACTA VETERINARIA ET ZOOTECHNICA SINICA, 37: 761 549



Zhang Y, Tomann P, Andl T, Gallant NM, Huelsken J, Jerchow B, Birchmeier W, Paus R, Piccolo S, Mikkola ML, 550

Morrisey EE, Overbeek PA, Scheidereit C, Millar SE, Schmidt-Ullrich R (2009) Reciprocal requirements for 551

EDA/EDAR/NF-kappaB and Wnt/beta-catenin signaling pathways in hair follicle induction. Developmental cell 552

17: 49-61 553

554 555 556



Figure Legends 557

Fig. 1 Dissecting cellular heterogeneity during Cashmere goat hair follicle 558

development. (a) Overall experimental design. (b) tSNE plot of all single cells labeled with 559

developmental time. Cells from the different developmental points were color-coded with 560

different colors. (c) tSNE plot of all single cells labeled with cell types according to their 561

marker gene expression. Different colors represent different cell clusters and the cell 562

number for each cluster was listed in the bracket. (d) Dot plot of representative marker 563

genes for different cell clusters. The color intensity represents its expression level, and the 564

dot size represents the positive cell percentage. 565

566

Fig. 2 Delineating dermal and epidermal cell lineage pseudotime developmental 567

trajectory. (a) Dermal cell lineage highlighted in the tSNE plot. (b) Developmental 568

trajectory of dermal cell lineage along pseudotime. Cells were color-coded with cell types 569

identified by Seurat (left panel) and developmental time (right panel), respectively. (c) 570

Epidermal cell lineage highlighted in the tSNE plot. (d) Developmental trajectory of 571

epidermal cell lineage along pseudotime. Cells were also color-coded with cell types (left 572

panel) and developmental time (right panel), respectively. (e) Main dermal cell lineage 573

decisions during hair follicle morphogenesis. (f) Main epidermal cell lineage fate decisions 574

during hair follicle morphogenesis. 575

576

Fig. 3 Revealing molecular profiles during DC fate commitment. (a) Pseudotime 577



expression heatmap during DC fate commitment. The four gene sets were determined by 578

k-means clustering according to their expression pattern and GO terms for each gene set 579

were listed in the right panel. (b) Immunofluorescence analysis of BMP2, CTNNB1, LEF1 580

and K15 in E60 Cashmere goat skin tissues. Scale bars = 50 μm. (c) Veen diagram 581

illustrating overlapped signature genes between E60 Cashmere goat DC cells and murine 582

E13.5 DC signature genes. (d) Visualizing murine DC signature genes of different stages 583

along pseudotime in Cashmere goat. The murine signature DC markers were listed in the 584

top panel and their corresponding expression pattern in Cashmere goat was listed in the 585

lower panel. Cells were color-coded according to their cluster identify. 586

587

Fig 4. DP cells from primary hair follicles and secondary hair follicles in Cashmere 588

goat showed distinct gene expression profiles. (a) Heatmap illustrating dynamic gene 589

expression profiles during DP cell fate commitment. The gene expression pattern for each 590

gene set was listed in the middle panel, and the top 5 enriched GO terms for each gene set 591

were listed in the right panel. (b) Pseudotime expression of cell fate 1 enriched genes 592

APOD, SOX18 and cell fate 2 enriched genes CENPW, TOP2A. Cells were colored-coded 593

with their corresponding cluster identity. (c) Volcano plot illustrating cell fate 1 and cell fate 594

significantly enriched differentially expressed genes. (d) Immunofluorescence analysis of 595

BMP2, K15, VDR and PCNA in the primary hair follicles and secondary hair follicles from 596

E120 Cashmere skin sections. Scale bars = 50 μm. 597

598



Fig. 5 Delineating matrix cell fate decision along pseudotime. (a) Heatmap 599

demonstrating dynamic gene expression patterns during matrix cell fate commitment. The 600

differentially expressed genes were divided into 4 different gene sets based on k-means 601

clustering. (b) Gene functional enrichment analysis of signature genes in the 4 different 602

gene sets identified in Fig. 5a. (c) The pseudotime expression pattern of representative 603

signature genes from each gene set. Cells were color-coded according to their cluster 604

identity. (d) Immunofluorescence analysis of LEF1 and CTNNB1 in E60 Cashmere skin 605

tissues. Scale bars = 50 μm. 606

607

Fig. 6 Revealing pseudotime gene expression dynamics during IRS and hair shaft cell 608

fate commitment. (a) Heatmap illustrating pseudotime gene expression pattern of 609

differentially expressed genes during IRS and hair shaft cell development. Cell fate 1 610

depicts IRS fate, while cell fate 2 depicts hair shaft fates. (b) Pseudotime expression pattern 611

of representative marker genes during IRS and hair shaft cell fate commitment. Cells were 612

color-coded with their corresponding cluster identity and the solid line depicts cell fate 1, 613

while the dashed line depicts cell fate 2. (c) Veen diagram demonstrating overlapped 614

murine and Cashmere goat IRS signature genes. The representative overlapped genes were 615

listed in the corresponding boxes. (d) Comparison of murine hair shaft signature genes with 616

Cashmere goat hair shaft signature genes. The representative overlapped genes were listed 617

in the rectangular boxes. (e) Comparison of PCNA and VDR expression in E90 Cashmere 618

goat and E16.5 mouse skin tissues. Scale bars = 50 μm. 619

620



Fig. 7 Dissecting the asynchronous development of keratinocyte. (a) Heatmap 621

illustrating gene expression dynamics during keratinocyte differentiation. (b) GO terms 622

corresponding to the gene set in Figure 7a. The arrow indicates the elongation of 623

pseudotime. (c) Expression of cell fate 1 signature genes KRT14, KRT17 and cell fate 2 624

signature genes TOP2A, PCNA along the pseudotime trajectory. (d) Immunofluorescent 625

staining analysis of VDR, PCNA, CTNNB1, BMP2, LEF1 and K15 in E90 Cashmere goat 626

skin tissues. Scale bars = 50 μm. 627

628

Supplementary Figure Legends 629

Supplementary Fig. 1 Single-cell dataset quality control and representative marker 630

expression across all single cells. (a) Morphology of fetus Cashmere goat at E60, E90 and 631

E120 and its corresponding skin structure revealed by HE staining. (b) Quality matrices 632

revealed by CellRanger of all datasets used in this study. (c) Comparison of the number of 633

genes detected and the number of UMI detected in each cell. (d) Evaluating key cell type 634

markers across all single cells in the tSNE plot. 635

636

Supplementary Fig. 2 Gene function enrichment of differentially expressed genes 637

during DC fate commitment. (a) The network of enriched GO terms for different gene 638

sets during DC fate commitment. Each dot represents one GO terms and different colors 639

represent different gene sets. (b) Circos plot demonstrating the number of overlapped GO 640

terms between different gene sets. 641



Supplementary Fig. 3 Gene function enrichment of differentially expressed genes 642

during DP cell fate commitment. (a) The network of enriched GO terms for different gene 643

sets during DP fate commitment. Each dot represents one GO terms and different colors 644

represent different gene sets. (b) Circos plot demonstrating the number of overlapped GO 645

terms between different gene sets. (c) Comparison of enriched GO terms between two 646

different cell fate. 647

648

Supplementary Fig. 4 Comparison of enriched GO terms during matrix cell fate 649

commitment. (a) Heatmap illustrating gene set specific enriched GO terms. (b) Circos plot 650

demonstrating the number of overlapped GO terms between different gene sets. 651

652

Supplementary Fig. 5 Validation of hair shaft/IRS marker expression and comparison 653

of enriched GO terms. (a) Immunohistochemistry analysis of hair shaft marker HOXC13 654

and IRS marker SOX9 expression in Cashmere goat skin. Scale bars = 50 μm. (b) The 655

network of enriched GO terms for different gene sets during hair shaft and IRS fate 656

commitment. Each dot represents one GO terms and different colors represent different 657

gene sets. (c) Heatmap illustrating gene set specific enriched GO terms. 658

659

Supplementary Fig. 6 Gene functional enrichment and representative gene expression 660

along pseudtotime. (a) The network of enriched GO terms for different gene sets during 661

keratinocyte differentiation. Each dot represents one GO terms and different colors 662



represent different gene sets. (b) Psedotime expression pattern of representative cell fate 1 663

and cell fate 2 signature genes. 664

665

Supplementary Tables 666

Supplementary table 1: All cluster specific differentially expressed genes. 667

Supplementary table 2: DC signature genes for different gene sets. 668

Supplementary table 3: DC signature genes comparasion between goat and mice. 669

Supplementary table 4: IRS and hair shaft signature genes comparasion between goat and 670

mice. 671

Supplementary table 5: All antibodies used in this study. 672

673



E60E90E120

Figure 1

(a)

E60

E90

E120

Fetus Cashmere goat skin biospy Single cell barcoding

Beads Oil

Sequencing & Dissecting heterogeneity

(b) (c) Cell type (cells)

(d)

CD

H19

JSR

P1

ARSI

CN

MD

L1C

AM

−1 2

e

0 25 50 750

1

23

4

5

67

89

11

1

111

LUM

COL1

A1

POST

N

DCN

APO

E

KRT1

5

KRT1

7

SOX9

KRT1

4

KLF5

RSP

O2

APO

D

SOX1

8

SOX2

PCO

LCE2

PLVA

P

VWF

PEC

AM1

KDR

DEP

P1

KRT1

0

KRT1

SBSN

KRTD

AP

LRAT

TBX2

ACTA

2

TPM

2

EBF2

ABC

C9

LCP1

FCER

1A

RG

S1

RG

S10

LTC

4S

tSNE 1

tSN

E 2

Clu

ster

ID

dermal epithelial DP endothelial keratinocyte pericyte macrophagy muscle

Developmental time

tSNE 1

tSN

E 2

0

1

2

3

4

5

6

7

8

9

1

1

1

1

1

1



1 2

Component 1

Com

pone

nt 2

Figure 2

(a)1

4

1 2

Component 1

Com

pone

nt 2

Cell type (cells)Developmental trajectory Developmental time

（）

（）

（）

Component 1

Com

pone

nt 2

1

2

3

Cell type (cells)Developmental trajectory

（）

（）

（）

Dermal lineage

Epidermal lineage

tSNE 1

tSN

E 2

tSNE 1

tSN

E 2

Component 1

Com

pone

nt 2

1

2

3

Developmental time

(c)

(b) (d)

(e)

Undiff. DC

Fibro.

DP

(f)

Undiff.

Matrix Hair shaft Keratinocyte



Figure 3

(a)

Pseudotime

FbAPODHES1LUMDCNVIMSOX18SOX9APOESOX2LHX2COL1A1HOXC13JUNPDGFADKK1TPX2NOTCH3

PTNLCKCORINCENPCCCDC66SFPQ

SYNRGBDKRB2AUTS2MRPL16ACADLHUNKWDR36CAB39LANGPT1PRXL2APURBKLHL24UBE2J2NDUFS5

0 10 20 30 40

0.0 2.5 5.0 7.5

0 2 4 6

0 2 4 6

tissue morphogenesisresponse to growth factor

cell morphogenesis involved in differentiationepidermis development

GO (BP) -Log(P)

negative regulation of cell cycle

positive regulation of organelle organization

negative regulation of protein modification

DNA-dependent DNA replication

Wnt signaling pathway

regulation of type I interferon production

viral translational termination-reinitiation

multicellular organismal homeostasis

proteasomal protein catabolic process

nucleobase-containing compound catabolic

intracellular receptor signaling pathway

anatomical structure homeostasis

(b)

1951（38.8%）

2345（46.7%）

729（14.5%）

Cashmere goat DC signature genes

E13.5 murine DCsignature genes

TRPS1

PRDM1

0 25 50 75 100

13

10

13

10

Pseudotime

Expr

essi

on

RSPO3

INHBA

0 25 50 75 100

13

1030

13

1030

Pseudotime

Expr

essi

on 1

13

467

(c)

LEF1

DKK1

0 25 50 75 100

13

10

0.51.0

3.05.0

Pseudotime

Expr

essi

on

Pre-DC DC1 DC2

Lef1, Dkk1, Twist, Fst Prdm1, Trps1, Sox18, Foxd1 Inhba, Rspo3, Foxp1, Hey1

Pseudotime

(d)

Dc

-3

3Exp.

1

2

3

4

SOX18

APOD

0 25 50 75 100

13

1030

1

10

100

Pseudotime

Expr

essi

on

SOX2

CTNNB1

0 25 50 75 100

0.51.0

3.05.0

13

1030

Pseudotime

Expr

essi

on



Cell Fate 1 vs Cell Fate 2

Figure 4

(a)

-3

3Exp. 0 10 20

0 20 40 60

0 5 10 15 20

0 1 2 3

tissue morphogenesisepidermis development

epithelial cell differentiationneuron projection morphogenesis

GO (BP)

mitotic cell cycle process

DNA-dependent DNA replication

cellular response to DNA damage stimulus

microtubule cytoskeleton organization

regulation of cell morphogenesis

citrulline metabolic process

protein-containing complex disassembly

protein localization to membrane

extracellular matrix organization

skeletal system development

blood vessel development

cellular response to growth factor stimulus

Pre-Branch Cell Fate 1 Cell Fate 2

Pseudotime

-Log(P)Signature gene expression

1,53

91.

077

559

201

BMP2

/K15

/DAP

I

Primary

BMP2

/K15

/DAP

I

Secondary

VDR

/PC

NA/

DAP

I

VDR

/PC

NA/

DAP

I

(b) (c) (d)

SOX18

APOD

0 25 50 75 100

13

1030

1

10

100

Pseudotime

Expr

essi

on

1

13

467

TOP2A

CENPW

0 25 50 75 100

13

1030

13

1030

Pseudotime

Expr

essi

on

SOX18 SOX2

FGF7KLF2

H2AFZ

CENPWLUM

KIF20B TOP2A

VIM

DLK1

0

50

100

150

200

−2 20

avg_logFC

log1

0(p_

val_

adj)

Significant None



Figure 5

(a)KRT25KRT71KRT85KRT27KRT35KRT84AHCYFGF21HBP1HPRT1CLTBGANTIMP2KCNK7CTSVTXN

POSTNVIMCOL1A1LUMACTC1COL1A2OGNTOP2A

KRT28PADI3SBSNKRT14KRT19LRATKRT23KRT73BMP4LEF1KRT4KRT1KRT17KRT10PDGFASOX9HES1SOX18

keratinocyte differentiationepidermis development

skin epidermis developmentregulation of keratinocyte differentiation

GO (BP) -Log(P)

cardiac epithelial to mesenchymal transition

mesenchymal cell differentiation

epithelial to mesenchymal transition

BMP signaling pathway

skin development

morphogenesis of an epithelium

negative regulation of cell proliferation

keratinocyte proliferation

cell division

skeletal system development

ossification

regulation of endothelial cell proliferation

0.0 2.5 5.0 7.5

0 2 4

0 5 10 15

0 5 10 15

Pseudotime

Undiff. Matrix

KRT27

KRT25

0 25 50 75 100

1e+01

1e+03

1e+05

1e+021e+051e+08

Expr

essi

on

SBSN

LEF1

0 25 50 75 100

1

3

10

1

10

100Exp

ress

ion

POSTN

LUM

0 25 50 75 100

1

10

100

110

1001000

Exp

ress

ion

SOX9

SOX18

0 25 50 75 100

13

1030

13

1030

Exp

ress

ion

gene

set 1

gene

set 2

gene

set 3

gene

set 4

Pseudotime

Cell clusters：0 2 3 5 8 9 12

(b) (c)

-3

3Exp.

4 3 2 1

LEF1/DAPILEF1 DAPI

E60

skin

(d)

CTNNB1/DAPICTNNB1 DAPI

E60

skin

epidermal

dermal

epidermal

dermal

epidermal

dermal

epidermal

dermal

epidermal

dermal

epidermal

dermal



Figure 6

(a)

Pseudotime

Cell fate 2

-3

3Exp.

ALCAMCOL1A2LUMOGNVIMCOL1A1SOX18VCANFSTAPODDKK1DLK1NR3C1CDH11MICU3ACADSLRCH3

H2AFZDCNTOP2APCNASHHVDREGFRLCKDCTETV5DCKSLF1MT3SNRPB

SBSNKRT4KRT1KRT14SOX9HES1PRDM1ELF1

MSX1KRT35ID3ID1AHCYHOXC13LEF1APOOUSP16RPS9MARS

Cell fate 1

OGN

LUM

0 25 50 75 100

1

10

100

13

1030

Expr

essi

on

TOP2A

DCN

0 25 50 75 100

1

10

100

13

1030

Expr

essi

on

SOX18

COL1A1

0 25 50 75 100

1

10

100

13

1030

Expr

essi

on

H2AFZ

CENPF

0 25 50 75 100

13

1030

13

1030

Expr

essi

on

Cell clusters：0 2 3 5 8 9 12

Pseudotime

Cell fates：fate 1fate 2

Pre-branch(b)

581 (40.3%)

306 (21.3%)

57 (4%)

427 (29.7%)

69 (4.8%)

Cashmere goat IRSsignature genes

Murine IRS-1signature genes

Murine IRS-2signature genes

Cell fate 1

KRT79, SBSN, KRT17,KRT7, KRT79, DUSP1

KRT14, KRT5, KRT1,KRT10, KRT77, KLF5,KLF4, KLF3 ...

(c)

Murine hair shaftsignature genes

491(23.9%)

389(18.9%)

1,085(52.7%)

26（1.3%）

67(3.3%)

Goat hair shaft HS-1

Goat hair shaftHS-2

MSX1，KRT35, HOXC13APOO, BMP4 ...

VDR, CUX1, SHH,LHX2, PCNA, DCN...

(d)

1

2

3

4

(e) E90 goat E16.5 mouse

PCN

A

VDR

E90 goat E16.5 mouse



1

2

3

KRT14

Figure 7

(a) Cell fate 2

-3

3Exp.

KRT85KRT35KRT14KRT84KRT28VDRBMP4STAR

Cell fate 1 Pre-branch

BMP2SHHCUX1ETV5IGSF8NARSURI1DNLZ

SBSNKLK10KRT17SOX9FSTPARM1LAMA3

CTNNB1POSTNH2AFZTIMP2KIF23TUBB2ABGNDLK1APOEVCANVIMASPHFMR1

regulation of cell cycledevelopment differentiationepidermal cell population proliferation

intermediate filament organization

GO enrichment dynamics along pseudotime

endocytosis virus host

lipid metabolic process

process metabolic cellular

folding 'de novo'

start

cell fate 1 cell fate 2

regulation response process

1

2

3

PCNA

log10( + 0.1)

1

2

3

TOP2A

component 1

com

pone

nt 2

Cel

l fat

e 1

Cel

l fat

e 2

1

2

3

KRT17

log10( + 0.1)

BMP2/K15/DAPI LEF1/PCNA/DAPI

Hi Hi

low low

low low

Hi Hi

VDR/PCNA/DAPI CTNNB1/DAPI

(b)

(c) (d)

1

2

3

4


A single-cell transcriptome atlas during Cashmere goat ...morphogenesis mainly used the mouse as a...

Documents

Transcript of A single-cell transcriptome atlas during Cashmere goat ...morphogenesis mainly used the mouse as a...