Background

Subo Merino (SBM) is a superfine wool-producing sheep breed in China. The average wool fiber diameter is 17–19 µm, which surpasses the standard textile count of 80 Nm [1] and has a far-reaching impact on the fine wool sheep industry. The growth and development of sheep wool is controlled by hair follicles (HFs), which are tiny organs attached to the skin that have a complex morphology, complex structure and periodic growth [2]. HFs are composed of multiple cells with very intricate interactions and are involved in the regulation of HF development, growth, regeneration and differentiation. The development of wool follicles has been described in detail for the Merino and it is well established that no new follicles are initiated after birth [3,4,5,6,7]. The first stage in follicle development is the proliferation of epidermal cells to form a placode beneath which an aggregation of dermal cells occurs and the two cell formations grow down together into the dermis. Progressively, the dermal cells move into the epithelial bud to form the pre-papilla and finally the epithelial bulb cells envelop the pre-papilla as the follicle lengthens and descends into the dermis. The stages as they occur in Merino sheep is first follicles formed are the primary follicles (PFs) followed by secondary follicles (SFs) and then secondary-derived follicles (SD) that branch from the SFs [4]. In Merino sheep, fibres from the SD constitute the bulk of the fleece. The first follicles to be initiated in the sheep fetus PFs are visible from 75 days of gestation and are producing a fibre by 90 days of gestation [3]. SFs do not appear until approximately 85 days of gestation. Some of these follicles will begin to branch (SD) at around 105 days [8]. HFs fully mature after birth; therefore, the number of HFs does not increase after birth. In du Cros et al. [9] description of the localization of epidermal growth factor immunoreactivity in sheep skin during wool follicle development, it was found that immunoreactivity was restricted to the periderm and intermediate layers of fetal epidermis at 55 d of gestation, when the first wave of wool follicles are initiated. This particular distribution persisted during subsequent development but never became associated with the basal cells of the epidermis. At approximately 105 d of gestation, however, reactions were detected in the outer root sheath as the follicles matured and in the differentiating cells of the sebaceous glands. Hutchison and Mellor [6] study the effects of maternal nutrition on the initiation of secondary wool follicles in foetal Scottish Blackface sheep. They found initiation of SFs usually takes place between about 95 and 135 days of gestation. Severe underfeeding during the first half of this period did not significantly inhibit the initiation of SFs, but severe underfeeding during the latter half of this period resulted in a significantly lower number of SFs and this number was not increased by refeeding ewes to a high level between 132 days and term. They concluded that SFs initiation is most affected by maternal undernutrition between about 115 and 135 days.

The differentiation of HFs is regulated by a variety of signaling pathways, including the bone morphogenetic protein (BMP), transforming growth factor beta (TGF-β) and Wnt signaling pathways [10, 11]. The specific expression patterns of these molecules in dermal papillae or stromal cells determine their functions during differentiation. However, knowledge regarding the corresponding cellular and molecular mechanisms is limited. After the primary HFs form in the sheep fetus, branches of the primary HFs form secondary HFs [12, 13]. Therefore, it is important to understand the associated molecular gene regulation mechanisms. The wool quality and commercial value of fine wool sheep are determined by the structure and characteristics of their HFs. This branching can be extensive and determines the final follicle population density in which about 80% of the follicles are SD follicles and several wool fibres emerge at the skin surface from the same orifice [14]. To improve the wool yield and wool quality of these sheep, it is necessary to study the factors affecting the formation of HFs and to deeply understand the molecular regulatory mechanisms of HF development. The processes of HF cell development and differentiation are regulated by a variety of genes and multiple signaling pathways; thus, identifying the major genes regulating the development and differentiation of HF cells has become the focus of research.

Wool fiber fineness, fiber length, wool bending, wool strength and hair flexibility determine not only the differences between wool products and other textile fibers but also the craft value of wool textile products [24], GLI2, KRT25 [25, 26], LAMA5 [27, 28], LRP4 [29, 30], SOSTDC1 [31, 32], TGFβ2 [33, 34], TMEM79 [35], BMP7 [36, 37], WNT10A [38], ZDHHC21 [39], SOX10 [40, 41], ITGB4 [42], KRT14 [43], and ITGA6 [44], which are associated with the development of the epidermis and HF. They are also related to the development of the epithelium. Functional enrichment analysis showed that the DEGs were significantly enriched in negative regulation of the canonical Wnt signaling pathway, HF development, negative regulation of the BMP signaling pathway, establishment of the skin barrier, and positive regulation of epidermal cell differentiation and skin development, highlighting the central roles of these DEGs in hair morphogenesis. Notably, the fate of HFs is affected by typical Wnt/β-catenin signaling, BMP signaling, the TGF-β signaling pathway, the PI3K-Akt signaling pathway, and the Hippo signaling pathway.

Embryonic HF development and postnatal hair growth rely on intercellular communication within the epithelium and between epithelial and mesenchymal cells [50]. A homozygous missense mutation within KRT25 that causes autosomal recessive woolly hair in humans, which is consistent with findings in mutant mice. The identification of a KRT25 mutation as a cause of woolly hair in humans [51, 52]. All-trans-retinoic acid could inhibit hair follicle growth via inhibiting proliferation and inducing apoptosis of DPCs partially through the TGFβ2/Smad2/3 pathway [33]. Reddy et al. [24] found that expression of FZD1 in the placode correlates with expression of WNT10A and WNT10B in the placode. They suggest that canonical WNT signaling is likely activated in the placodes and DC of develo** hair follicles by WNT10A and WNT10B, expressed in and secreted from epithelial cells, binding to FZD1, expressed in the placode epithelium and DC. Expression of FZD1 is detected in the DP and Mx of anagen follicles, and could potentially interact with DP and Mx cells, in particular WNT10A. In postnatal HFs in full anagen, FZD1 express in the ORS. At the germ and bulbous peg stages, WNT10A and WNT10B are expressed continuously in follicular epithelium during these later stages of morphogenesis. At the bulbous peg stage, WNT10A and WNT10B are strongly expressed in a cone of epithelial cells surrounding the DP. In rat whisker HFs WNT10A was expressed in the ORS, IRS, Mx and HS of anagen follicles [53].

Fig.9
figure 9

a Diagrams of HF development in Merino sheep. b Correspondence between different cell types and DEGs. Epi epidermal, Pc placode, DC dermal condensate, DP dermal papilla, Mx matrix, SG sebaceous glands, SwG Sweat gland, AMP Arrector pili muscle, Mc melanocytes, ORS outer root sheath, IRS inner root sheath, PF primary follicle, SF secondary follicle, SD secondary-derived follicle, HS hair shaft

Full size image

The characteristics of HF development and postpartum regeneration are significantly altered in microanatomy and cell viability experiments. HF development is controlled by a variety of signaling pathways, transcription factors and epigenetic regulatory factors (including miRNAs) [54]. Some studies based on HF gene regulatory networks have shown that the Wnt [10], TGF-β [55, 56], MAPK [57], Shh [58], BMP [59], PI3K-Akt [77, 78] is determined by controlling FDR (false discovery rate), and the corrected P-value is Q-value. At the same time, we calculated the differential expression multiple (fold change) according to the FPKM value. The screening conditions of differential genes are as follows: Q-value ≤ 0.05 and fold change ≥ 2 were considered differentially expressed between the adjacent comparison groups (comparisons: G2/G1, G3/G2 G4/G3, G5/G4, G6/G5, and G6/G1) [79]. The expression patterns of DEGs were analyzed by systematic clustering to explore the similarities and relationships between the different libraries. Furthermore, the DEGs were subjected to K-means clustering using the Euclidean distance method associated with complete linkage on the BMK Cloud platform (https://www.biocloud.net/).

Functional enrichment analysis and gene annotation

Functional annotation was performed using the GO (http://geneontology.org) and KEGG (http://www.genome.ad.jp/kegg/) databases [80] based on GO and KEGG pathways. These analyses were conducted for the DEGs identified in each tissue at each stage of HF development, which were significantly enriched in dynamic expression patterns. The BPs and metabolic pathways significantly associated with the gene lists were determined based on their FDR [81]. Functional evidence was obtained on the basis of the relationships between the significant GO terms (FDR < 0.05) and the DEGs [82]. The Database for Annotation Visualization and Integrated Discovery (DAVID) 6.8 (https://david.ncifcrf.gov/tools.jsp) was used to perform functional annotation of the genes that were significantly enriched in the expression patterns [83].

Construction of a PPI network

Following the integration of the protein information in the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) (https://string-db.org/) database with the DEGs, a PPI network of the identified DEGs was established. PPI network analysis was performed using STRING and Cytoscape software V3.8.0 (http://www.cytoscape.org/). First, STRING was utilized to analyze the correlation coefficients between genes. Interacting pairs with confidence scores > 0.5 were selected to build the PPI network, and the Matthews correlation coefficient (MCC) algorithm was used to calculate hub genes and the selected top 120 genes were visualized using Cytoscape software.

K-means analysis and WGCNA

DEGs were clustered with the R package K-means function, where K = 10 within the cluster package according to the Euclidean distance. WGCNA [84, 85] was applied to the FPKM expression data. A coexpression network was constructed with a beta value of 4. We calculated the coefficients of gene dissimilarity, performed hierarchical clustering of the genes and then determined the gene modules by the dynamic tree cut method. Through clustering analysis, modules close to each other were merged into new modules. Before WGCNA, we identified and filtered the selected gene set. We removed low-quality genes with an unstable impact on the results to improve the accuracy of network construction. The filtering criteria in this study were as follows: for each gene, the maximum count value in all samples was > 50, and the count was > 20 in at least 16 samples. The modules were functionally annotated using DAVID. Highly connected genes in each module, which are also known as hub genes, may play important roles in the module. Hub genes are conserved to a certain extent and are at the core of the gene coexpression network. These genes can act as genetic buffers to reduce the impacts of other gene mutations. We identified the top 150 hub genes in the modules that were most closely related to HF development differences, that is, the 150 genes with the highest connectivity in the modules, and used Cytoscape software to map the gene–gene interaction network to visualize the gene relationships.

Validation of RNA-seq data

Several differentially expressed mRNAs involved in HF development were selected and confirmed by RT–PCR with GAPDH as an internal reference. The primers used for RT–PCR are listed in Additional file 11: Table S7. Total RNA from the samples used for high-throughput RNA-seq was isolated and converted into cDNA using a PrimeScript™ RT Reagent Kit with gDNA Eraser (TaKaRa, Japan). RT–PCR was carried out on a CFX96™ Real-Time System (Bio–Rad, USA) using a TB Green Premix Ex Taq™ kit (TaKaRa, Japan) according to the manufacturer’s instructions. The thermal cycling conditions used in RT–PCR were 95 °C for 30 s followed by 40 cycles of 95 °C for 5 s and 60 °C for 30 s. A reaction volume of 20 μL was used for RT–PCR according to the manufacturer’s protocol. The specificity of the SYBR Green PCR signal was confirmed by melting curve analysis. The RT–PCR experiments were performed in triplicate, and the average Ct value was used for further analysis. The 2−ΔΔCt method was used to determine the relative mRNA abundance.