This article is mentioned in:

Introduction

Arthrogryposis-renal dysfunction-cholestasis (ARC) syndrome is a life-threatening autosomal recessive multisystem disorder caused by germline mutations in VPS33B-interacting protein, apical-basolateral polarity regulator (VIPAR) or vacuolar protein sorting 33 homolog B (VPS33B) (1). The principle clinical manifestations of ARC include renal tubular dysfunction, cholestasis, ichthyosis, central nervous system malformation and congenital joint contractures involving multiple organ systems (2,3). It has been recognized that ARC syndrome exhibits notable clinical variability, and the prognosis of this condition is particularly poor, with the majority of patients not surviving beyond the first year of life (4,5). Furthermore, there is currently no specific treatment for this syndrome.

Mutations in VPS33B are detectable in 75-77% of patients with a clinical diagnosis of ARC syndrome (3,6). A better understanding of the molecular pathology of this disorder is of vital importance for the development of an appropriate therapeutic regimen. VPS33B encodes a 617-amino-acid protein, which is a homolog of the class C yeast vacuolar protein sorting, and the VPS33B protein contains a Sec-1 domain involved in intracellular protein sorting and vesicular trafficking (7). It has also been reported that VPS33B is a downstream target gene of the hnf6/vhnf1 signaling pathway that is important for zebrafish biliary development (8). In addition, the VPS33B protein can interact with soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs), which are involved in vesicular exocytosis and synaptic transmission to facilitate vesicle targeting and fusion (9). Therefore, the interaction between the mutant protein expressed in VPS33B mutants and the SNAREs at the late endosomal stage may be impeded, leading to abnormal secretion of lamellar granules, and localization or accumulation of plasma proteins (2,10). Abnormal protein trafficking and impairment in the maturation of multi-vesicular bodies in megakaryocytes underlie the α-granule deficiency in a mouse model of VPS33B deficiency and in patients with ARC (11). The VPS33B-VIPAR complex may regulate apical-basolateral polarity via the Rab11a-dependent apical recycling pathway and the transcriptional regulation of epithelial cadherin (1). This complex also regulates the delivery of lysyl hydroxylase 3 into newly identified post-Golgi collagen IV carriers, which are essential for the modification of lysine residues in multiple collagen types (12).

In the study by Hanley et al (13), a murine model with a liver-specific deletion of VPS33B (VPS33B fl/fl-AlfpCre) was successfully established, as indicated by the abnormalities identified in mice, which were similar to those observed in children with ARC syndrome. Furthermore, the analysis of gene expression profiles provided an insight into the possible regulatory mechanisms responsible for ARC syndrome. However, only the gene expression and pathway analysis of the microarray data were performed in the aforementioned study. To further elucidate the molecular basis of ARC, the gene expression profiles deposited by Hanley et al (13) were downloaded in the present study in order to conduct weighted gene co-expression network analysis and to identify potential therapeutic drugs.

Materials and methods

Microarray data and preprocessing

The gene expression profile of GSE83192 (13), generated by the GPL16570 platform (Affymetrix Mouse Gene 2.0 Array; Thermo Fisher Scientific, Inc., Waltham, MA, USA), was downloaded from Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/). This dataset contained six liver tissue samples from liver-specific VPS33B knockout (VPS33Bfl/fl-AlfpCre) mice and four liver tissue samples from control (VPS33Bfl/fl) mice. The raw data preprocessing was conducted using the oligo package in R software (www.r-project.org), including conversion of the data format, filling-in of missing data with the median values (14), background correction using the MicroArray Suite method (15) and normalization of the sequencing data using the quantile method (16).

Differential expression analysis and hierarchical clustering

The Limma package (17) was used to perform differential expression analysis for normalized values. In addition, P-values were adjusted for the false discovery rate (FDR) via the method described by Benjamini and Hochberg (18). The thresholds for differentially expressed gene (DEG) screening were set at FDR<0.05 and |log2 fold change|>1. The expression values of screened DEGs were hierarchically clustered by the pheatmap package (19) in R to intuitively observe the differences in gene expression levels.

Identification of co-expression modules

The gene significance (GS) values, defined as the log of the P-value, indicated the difference in the mRNA expression between VPS33B knockout and control mice. The module significance (MS), defined as the mean value of GS for all genes in a given module, was calculated for each module to identify its connection with the disease status. Two representative co-expression modules with the highest MS values were selected since a higher MS value indicates a closer connection.

Construction and analysis of the protein-protein interaction (PPI) network

The DEGs in the two selected representative co-expression modules mentioned earlier were adopted for PPI network construction. The database Search Tool for the Retrieval of Interacting Genes/Proteins (22) was employed to collect pairwise PPIs among the DEGs. Cytoscape software (23) was applied for visualization of the interaction associations, and the Molecular Complex Detection (MCODE) plugin (24) was used to create the modules with the following parameters: Degree cut-off=2, node score cut-off=0.2, and K-core=2. BiNGO (25), another plugin of Cytoscape, was used to annotate module function with an adjusted P-value of <0.05.

Enrichment analysis and potential therapeutic drug identification

The Gene Ontology (GO) annotations of the PPI network were performed by GOstat (26) in three categories, namely biological process (BP), cellular component (CC) and molecular function (MF), with P<0.05. The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was conducted by the KEGG Orthology-Based Annotation System server (27), and P<0.05 was used as the cut-off criterion. Bioactive small molecules of putative relevance to the ARC syndrome were searched for using the Connectivity Map (CMAP) database, with the criteria set to |connectivity score|>0.9 and P<0.05 (28). A connectivity score closer to −1 implied that this small molecule may have a stronger therapeutic effect.

Results

DEG screening and hierarchical clustering

Subsequent to data preprocessing, 1,016 DEGs, including 768 upregulated and 248 downregulated genes, were identified in the VPS33B knockout mice as compared with the control mice. The expression values of screened DEGs were hierarchically clustered by pheatmap package, and the color contrast indicated that there were significant differences in gene expression between the VPS33Bfl/fl-AlfpCre and VPS33Bfl/fl mice (Fig. 1).

Figure 1 Differences in the expression values of differentially expressed genes between liver tissue samples from the VPS33B (VPS33Bfl/fl-AlfpCre) knockout mice and control (VPS33Bfl/fl) mice. Vps33b, vacuolar protein sorting 33 homolog B.

WGCNA analysis and PPI network construction

Based on the correlation coefficient between log (k) and log P (k), the value of the adjacency matrix was set to 18 in order to guarantee the scale-free topology of the co-expression modules (Fig. 2). A gene clustering tree with the cut-off height of 0.9 was then constructed (Fig. 3), and the MS values of all modules were >0.6 with a P<0.05 (Fig. 4). Two representative co-expression modules, including the green module (72 DEGs) and the turquoise module (247 DEGs), were selected. Subsequently, the PPI network was constructed according to the PPIs of DEGs in these two representative co-expression modules, and the constructed PPI network contained 71 nodes and 135 PPIs (Fig. 5).

Figure 2 Power value of the adjacency matrix. The red line represents the correlation coefficient between log (k) and log P (k) of 0.9.

Figure 3 Gene dendrogram of co-expression modules.

Figure 4 Module significance value of different co-expression modules.

Figure 5 Protein-protein interaction network and the three identified modules. The regular triangle and inverted triangle stand for upregulated and down-regulated differentially expressed genes, respectively. The lines indicate the interactions between genes, while the colors of the triangles represent different co-expression modules.

PPI network analysis

Three modules were identified from the constructed PPI network by MCODE (Fig. 5). The functional annotations revealed that the genes in module 1 were mainly involved in the positive regulation of biological processes (CD86, CD83, IL1B and TLR2; adjusted P=1.32×10−3) and positive regulation of cytokine production (CD83, IL1B and TLR2; adjusted P=5.25×10−5; Table I). The DEGs in module 2 were significantly enriched with respect to protein heterotrimerization (COL1A1 and COL1A2; adjusted P=2.85×10−6) and cellular component organization (COL1A1, COL1A2 and CD44; adjusted P=1.12×10−2). The four DEGs (ABCG8, ABCG5, ABCB4 and ABCG3) in module 3 were clearly associated with transport (adjusted P=1.49×10−3) and the establishment of localization (adjusted P=1.49×10−3). ABCG8 was observed to mainly participate in intestinal cholesterol absorption, lipid digestion, cholesterol efflux, intestinal absorption and sterol transport.

Table I Enriched GO terms for DEGs in the three identified modules from the protein-protein interaction network.

Table I

Enriched GO terms for DEGs in the three identified modules from the protein-protein interaction network.

GO ID	Adjusted P-value	Description	DEGs
Module 1
GO:0048518	1.32×10−3	Positive regulation of biological process	CD86, CD83, IL1B, TLR2
GO:0050789	1.82×10−2	Regulation of biological process	CD86, CD83, IL1B, TLR2
GO:0065007	2.22×10−2	Biological regulation	CD86, CD83, IL1B, TLR2
GO:0001819	5.25×10−5	Positive regulation of cytokine production	CD83, IL1B, TLR2
GO:0031349	5.25×10−5	Positive regulation of defense response	CD86, IL1B, TLR2
GO:0031347	1.54×10−4	Regulation of defense response	CD86, IL1B, TLR2
GO:0001817	1.54×10−4	Regulation of cytokine production	CD83, IL1B, TLR2
GO:0048584	2.31×10−4	Positive regulation of response to stimulus	CD86, IL1B, TLR2
GO:0002684	2.31×10−4	Positive regulation of immune system process	CD86, CD83, TLR2
GO:0080134	2.31×10−4	Regulation of response to stress	CD86, IL1B, TLR2
GO:0051240	2.31×10−4	Positive regulation of multicellular	CD83, IL1B, TLR2
		organismal process
GO:0002682	4.86×10−4	Regulation of immune system process	CD86, CD83, TLR2
GO:0048583	6.88×10−4	Regulation of response to stimulus	CD86, IL1B, TLR2
GO:0010033	1.76×10−3	Response to organic substance	CD83, IL1B, TLR2
GO:0002376	1.80×10−3	Immune system process	CD86, IL1B, TLR2
GO:0051239	3.31×10−3	Regulation of multicellular organismal process	CD83, IL1B, TLR2
GO:0042221	3.31×10−3	Response to chemical stimulus	CD83, IL1B, TLR2
GO:0048519	5.00×10−3	Negative regulation of biological process	CD83, IL1B, TLR2
GO:0048522	5.02×10−3	Positive regulation of cellular process	CD83, IL1B, TLR2
GO:0050896	1.10×10−2	Response to stimulus	CD83, IL1B, TLR2
Module 2
GO:0070208	2.85×10−6	Protein heterotrimerization	COL1A1, COL1A2
GO:0070206	3.99×10−5	Protein trimerization	COL1A1, COL1A2
GO:0051291	2.91×10−3	Protein heterooligomerization	COL1A1, COL1A2
GO:0051259	6.07×10−3	Protein oligomerization	COL1A1, COL1A2
GO:0070271	9.81×10−3	Protein complex biogenesis	COL1A1, COL1A2
GO:0006461	9.81×10−3	Protein complex assembly	COL1A1, COL1A2
GO:0016043	1.12×10−2	Cellular component organization	COL1A1, COL1A2, CD44
GO:0065003	1.12×10−2	Macromolecular complex assembly	COL1A1, COL1A2
GO:0048646	1.12×10−2	Anatomical structure formation involved in morphogenesis	COL1A1, CD44
GO:0043933	1.12×10−2	Macromolecular complex subunit organization	COL1A1, COL1A2
Module 3
GO:0006810	1.49×10−3	Transport	ABCG8, ABCG5, ABCB4, ABCG3
GO:0051234	1.49×10−3	Establishment of localization	ABCG8, ABCG5, ABCB4, ABCG3
GO:0051179	1.62×10−3	Localization	ABCG8, ABCG5, ABCB4, ABCG3
GO:0030299	4.77×10−3	Intestinal cholesterol absorption	ABCG8
GO:0044241	5.72×10−3	Lipid digestion	ABCG8
GO:0033344	1.43×10−2	Cholesterol efflux	ABCG8
GO:0050892	1.50×10−2	Intestinal absorption	ABCG8
GO:0006855	1.67×10−2	Drug transmembrane transport	ABCB4
GO:0015893	1.67×10−2	Drug transport	ABCB4
GO:0015918	1.67×10−2	Sterol transport	ABCG8

[i] GO, gene ontology; DEGs, differentially expressed genes.

Functional and pathway enrichment analysis of the PPI network

The functional enrichment analysis revealed that the DEGs in the PPI network were significantly correlated with 10, 11 and 11 GO terms in the BP, CC and MF categories, respectively (Table II). A total of 11 DEGs (ALCAM, ITGAX, CLDN4, CD44, ITGB8, CD34, CLDN6, BCL2, CDH1, CD2AP and SPP1) were mainly involved in cell adhesion (P=7.25×10−5). In addition, 7 DEGs (ALCAM, CD83, CD86, ITGAX, CD44, CD34 and TLR2) were significantly associated with the external side of the plasma membrane (P=4.44×10−4) and cell surface (P=4.44×10−3). Meanwhile, the DEGs in the PPI network were evidently associated with ATP-binding transport activity (ABCG8, ABCG5, ABCG3, ABCC5, ABCB4 and ABCA; P=4.71×10−3), cytokine activity (CCL2, CCL19, IL1B, CCL6 and SPP1; P=5.31×10−3), and cholesterol transporter and sterol transporter activities (ABCG8 and ABCG5; P=3.47×10−3). Furthermore, COL1A2 and COL1A1 were associated with collagen type I (P=9.57×10−3), fibrillar collagen (P=3.31×10−3) and platelet-derived growth factor binding (P=3.47×10−3).

Table II Functional annotations in the BP, CC and MF categories for DEGs in the protein-protein interaction network.

Table II

Functional annotations in the BP, CC and MF categories for DEGs in the protein-protein interaction network.

Term	Description	P-value	Count	DEGs
BP
GO:0007155	Cell adhesion	7.25×10−5	11	ALCAM, ITGAX, CLDN4, CD44, ITGB8, CD34, CLDN6, BCL2, CDH1, CD2AP, SPP1
GO:0022610	Biological adhesion	7.36×10−5	11	ALCAM, ITGAX, CLDN4, CD44, ITGB8, CD34, CLDN6, BCL2, CDH1, CD2AP, SPP1
GO:0006952	Defense response	7.02×10−5	10	NFKBIZ, TMEM173, CCL2, CD44, BCL2, TLR2, CCL19, IL1B, APAF1, TLR6
GO:0009611	Response to wounding	7.13×10−5	9	NFKBIZ, CCL2, CD44, BCL2, TLR2, CCL19, IL1B, TLR6, PLAUR
GO:0006955	Immune response	5.71×10−4	9	TMEM173, CCL2, TLR2, CCL19, IL1B, AF251705, TLR6, GBP3, CCL6
GO:0006954	Inflammatory response	2.84×10−4	7	NFKBIZ, CCL2, CD44, TLR2, CCL19, IL1B, TLR6 CD83, CHRM3, BCL2, TLR2, IL1B
GO:0051240	Positive regulation of multicellular organismal process	4.23×10−3	5
GO:0006631	Fatty acid metabolic process	6.48×10−3	5	HACL1, CYP4A32, ACNAT2, ALOX5AP, ACACB
GO:0016337	Cell-cell adhesion	1.52×10−2	5	CLDN4, CLDN6, BCL2, CDH1, CD2AP
GO:0044093	Positive regulation of molecular function	3.51×10−2	5	CHRM3, BCL2, TLR2, APAF1, TLR6
CC
GO:0009897	External side of plasma membrane	4.44×10−4	7	ALCAM, CD83, CD86, ITGAX, CD44, CD34, TLR2
GO:0045177	Apical part of cell	4.78×10−4	6	EPCAM, ABCG8, ABCG5, CDH1, ABCB4, SPP1
GO:0009986	Cell surface	3.32×10−3	7	ALCAM, CD83, CD86, ITGAX, CD44, CD34, TLR2
GO:0005584	Collagen type I	9.57×10−3	2	COL1A2, COL1A1
GO:0016324	Apical plasma membrane	1.20×10−2	4	EPCAM, ABCG8, ABCG5, ABCB4
GO:0000267	Cell fraction	2.34×10−2	8	ABCG5, CYP3A16, CYP4A32, CLIC5, BCL2, APAF1, ABCC5, ABCB4
GO:0034707	Chloride channel complex	2.39×10−2	3	CLIC5, ANO1, ANO6
GO:0030054	Cell junction	2.47×10−2	7	CHRM3, CLDN4, CLDN6, CDH1, PAK1, ABCB4, GRID1
GO:0016323	Basolateral plasma membrane	3.01×10−2	4	EPCAM, CD44, CDH1, PAK1
GO:0005583	Fibrillar collagen	3.31 ×10−2	2	COL1A2, COL1A1
MF
GO:0016887	ATPase activity	4.71×10−3	6	ABCG8, ABCG5, ABCG3, ABCC5, ABCB4, ABCA5
GO:0005125	Cytokine activity	0.53×10−3	5	CCL2, CCL19, IL1B, CCL6, SPP1
GO:0008009	Chemokine activity	9.65×10−3	3	CCL2, CCL19, CCL6
GO:0042379	Chemokine receptor binding	1.01×10−2	3	CCL2, CCL19, CCL6
GO:0005254	Chloride channel activity	2.59×10−2	3	CLIC5, ANO1, ANO6
GO:0031404	Chloride ion binding	2.90×10−2	3	CLIC5, ANO1, ANO6
GO:0005253	Anion channel activity	2.98×10−2	3	CLIC5, ANO1, ANO6
GO:0043168g	Anion binding	3.06×10−2	3	CLIC5, ANO1, ANO6
GO:0017127	Cholesterol transporter activity	3.47×10−2	2	ABCG8, ABCG5
GO:0048407	Platelet-derived growth factor binding	3.47×10−2	2	COL1A2, COL1A1
GO:0015248	Sterol transporter activity	3.47×10−2	2	ABCG8, ABCG5

[i] BP, biological process; CC, cellular component; MF, molecular function; DEGs, differentially expressed genes.

In total, 8 KEGG pathways were identified for the DEGs in the PPI network (Table III). The members of the adenosine triphosphate-binding cassette (ABC) family, such as ABCG8, ABCG5, ABCG3, ABCC5, ABCB4 and ABCA5, were significantly associated with the ABC transporters (P=1.12×10−5). Certain other DEGs, including ITGB8, COL1A2, COL1A1 and SPP1, were evidently associated with extracellular matrix (ECM)-receptor interactions (P=2.28×10−3) and focal adhesion (P=1.03×10−2). Finally, the enriched Toll-like receptor (TLR) signaling pathway was associated with CD86, TLR2, IL1B, TLR6 and SPP1 (P=4.31×10−3).

Table III Significantly enriched pathways for DEGs in the protein-protein interaction network.

Table III

Significantly enriched pathways for DEGs in the protein-protein interaction network.

Term	Count	P-value	DEGs
mmu02010: ABC transporters	6	1.12×10−5	ABCG8, ABCG5, ABCG3, ABCC5, ABCB4, ABCA5
mmu04514: Cell adhesion molecules	7	5.31×10−4	ALCAM, CD86, CLDN4, ITGB8, CD34, CLDN6, CDH1
mmu04512: ECM-receptor interaction	5	2.28×10−3	CD44, ITGB8, COL1A2, COL1A1, SPP1
mmu04620: Toll-like receptor signaling pathway	5	4.32×10−3	CD86, TLR2, IL1B, TLR6, SPP1
mmu04510: Focal adhesion	6	1.03×10−2	ITGB8, BCL2, COL1A2, COL1A1, PAK1, SPP1
mmu00640: Propanoate metabolism	3	1.74×10−2	ALDH1B1, ACACB, ACAT1
mmu00620: Pyruvate metabolism	3	3.12×10−2	ALDH1B1, ACACB, ACAT1
mmu00071: Fatty acid metabolism	3	3.71×10−2	CYP4A32, ALDH1B1, ACAT1

[i] DEGs, differentially expressed genes.

Potentially therapeutic small molecules

A total of six small molecules, including mercaptopurine, ikarugamycin, camptothecin, quinostatin, dexpanthenol and DL-thiorphan, were screened using the CMAP database (Table IV). The score for mercaptopurine was the lowest (connectivity score=−0.939), indicating that this small molecule may be a potential drug for ARC treatment.

Table IV Identification of small molecules with a potential therapeutic role in arthrogryposis-renal dysfunction-cholestasis syndrome using the Connectivity Map database.

Table IV

Identification of small molecules with a potential therapeutic role in arthrogryposis-renal dysfunction-cholestasis syndrome using the Connectivity Map database.

Name	Connectivity score	P-value
Mercaptopurine	−0.939	7.67×10−3
Ikarugamycin	−0.906	1.62×10−3
Camptothecin	0.902	1.82×10−3
Quinostatin	0.904	1.88×10−2
Dexpanthenol	0.920	4.00×10−5
DL-thiorphan	0.975	9.70×10−4

Discussion

ARC, mainly caused by mutations in VPS33B, is associated with abnormalities in polarized liver and kidney cells, resulting in a multisystem disorder (1). In the present study, the microarray data of liver tissue samples from liver-specific VPS33B knockout mice and control mice were comprehensively analyzed. The DEGs in two representative co-expression modules with the highest MS values were selected for PPI network construction via WGCNA analysis. Three further modules were identified from the PPI network and annotated.

The five DEGs in module 1 included CD86, CD83, IL1B, TLR2 and LGSF6. Pathway enrichment analysis of the PPI network demonstrated that CD86, TLR2, IL1B, TLR6 and SPP1 were significantly associated with the TLR signaling pathway (P=0.004317). The TLRs are part of the naive immune system and serve key roles in the elicitation of immune responses to microbes (29). It has been suggested that the VPS33B-VIPAR complex interacts with an active form of Rab11a (1). In addition, Rab11a-positive endosomes have been revealed to be important intermediates in the transport of TLRs (TLR2 and TLR4) and TLR adaptor molecules to phagosomes (30,31). In a study by Yu et al (32), deletion of Rab11a induced cytokine production and altered the intracellular distribution of TLRs, indicating that Rab11a contributes to intestinal host-microbial homeostasis through the sorting of TLRs. The data of the present study revealed that CD83, IL1B and TLR2 were significantly enriched with respect to the positive regulation of cytokine production (adjusted P=5.25×10−5). Thus, the identified DEGs, including CD83, IL1B, TLR2 and TLR6, may participate in the pathology of the ARC syndrome caused by mutations in VPS33B via the TLR signaling pathway and positive regulation of the cytokine production.

VPS33B serves a key role in the regulation of vesicle-to-target SNARE complex formation and subsequent membrane fusion (33). Furthermore, inhibition of SNARE-mediated membrane traffic disrupted the intracellular integrin trafficking that can provide a linkage between the ECM and the cytoskeleton (34,35). In the present study, COL1A2 and COL1A1 in module 2 were significantly enriched with respect to cellular component organization, ECM-receptor interaction and focal adhesion. It has also been demonstrated that loss of SNAP29 may cause alterations in the Rab11-expressing domains of the endocytic recycling compartment and the structure of focal adhesions, impairing endocytic recycling and cell motility (36). Taken together, the current study results provide evidence that mutations in VPS33B may disturb cellular component organization, ECM-receptor interactions and focal adhesion by regulating COL1A2 and COL1A1.

It has been reported that vesicles containing ABC transporters co-localize with Rab11a prior to their insertion into the canalicular membrane (37). In the present study, the four DEGs (ABCG8, ABCG5, ABCB4 and ABCG3) in module 3 were significantly involved in the ABC transporter pathway (P=1.12×10−5). The ABC transporters are necessary for the energy-dependent biliary secretion of bile acids, phospholipids, sterols (for instance, ABCG8 and ABCG5 are sterol transporters) and non-bile acid organic anions (38). Impaired bile acid transport at the canalicular membrane, associated with reduced amounts of ABC transporter proteins, may cause cholestasis (bile secretory failure) (39). The functional annotations for DEGs in the PPI network revealed that ABCG8 and ABCG5 were evidently associated with cholesterol transporter and sterol transporter activities. In addition, ABCG8 mainly participated in intestinal cholesterol absorption, lipid digestion, cholesterol efflux, intestinal absorption and sterol transport, according to the module annotations. Therefore, it may be speculated that mutations in VPS33B influence sterol absorption and transport by regulating ABCG8 and ABCG5.

In conclusion, the results of the present study strongly indicate that the DEGs in the three identified modules serve important roles in the pathogenesis of ARC caused by mutations in VPS33B. Furthermore, CD83, IL1B, TLR2 and TLR6 may participate in the pathology by influencing the TLR signaling pathway and positive regulation of cytokine production. The mutations in VPS33B may disturb the cellular component organization, ECM-receptor interaction and focal adhesion by dysregulation of COL1A2 and COL1A. Finally, sterol absorption and transport may also be impeded by mutations in VPS33B via the regulation of ABCG8 and ABCG5 expression.

Acknowledgments

Not applicable.

Funding

No funding was received.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors’ contributions

XHan, MZ and LZ searched and downloaded gene expression profile from the Gene Expression Omnibus database. MC, LS, JB, XHao and BY made substantial contributions to analysis and interpretation of microarray dataset. All authors read and approved the final manuscript.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References