Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 https://doi.org/10.1186/s13018-019-1182-1

RESEARCH ARTICLE Open Access Identify differential expressions in fatty infiltration process in rotator cuff Pengfei Hu1,2 , Lifeng Jiang1,2 and Lidong Wu1*

Abstract Background: Rotator cuff tears are one of the most frequent upper extremity injuries and lead to pain and disability. Recent studies have implicated fatty infiltration in rotator cuff is a key failure element with the higher re- tear rates and poorer functional prognosis. Therefore, we investigated the differential expression of key in each stage of rotator cuff tear. Methods: A published expression profile was downloaded from the Omnibus database and analyzed using the Linear Models for Microarray Data (LIMMA) package in R language to identify differentially expressed genes (DEGs) in different stages of injured rotator cuff muscles. (GO) functional and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed to annotate the function of the DEGs. Finally, PPI network and module analysis were used to identify hub genes. Results: A total of 1089 fatty infiltration-related DEGs were identified, including 733 upregulated and 356 downregulated genes, and GO analyses confirmed that fatty infiltration was strongly associated with inflammatory response, aging, response to lipopolysaccharide, and immune response. Significantly enriched KEGG pathways associated with these DEGs included the phagosome, cell adhesion molecules, tuberculosis, and osteoclast differentiation. Further analyses via a PPI network and module analysis identified a total of 259 hub genes. Among these, Tmprss11d, Ptprc, Itgam, Mmp9, Tlr2, Il1b, Il18, Ccl5, Cxcl10, and Ccr7 were the top ten hub genes. Conclusions: Our findings indicated the potential key genes and pathways involved in fatty degeneration in the development of fatty infiltration and supplied underlying therapeutic targets in the future. Keywords: Rotator cuff tears, Fatty infiltration, DEGs, GO annotation, KEGG pathway, Module analysis

Background Surgery is an effective solution for patients with large Rotator cuff tear (RCT) is one of the most common and full-thickness RCT. Currently, surgical procedures musculoskeletal disorder in orthopedics department. include mini-open and all-arthroscopic rotator cuff re- Earlier in 1991, the prevalence of shoulder disorders in pair. According to the follow-up results, there is no sig- people aged 70 and above was 21% [1]. According to a nificant difference in clinical prognosis between the two survey performed by Hiroshi in 2012, the prevalence of methods [4]. However, the failure rate was reported RCT was 22.1% was in the general population, and the ranged from 23 to 77% in arthroscopically repaired prevalence was significantly increased with advanced age massive rotator cuff tears, which characterized as the [2]. The disease is characterized as injury of one or more size of tear was more than 5 cm or the tear involving tendons/muscles in rotator cuff, which can cause pain, more than two cuff tendons [5–7]. The size of the de- stiffness, muscle weakness, and functional limitation. fect, tendon retraction, muscular atrophy, and degree of Common risk factors for RCT include advancing age, fatty degeneration were regarded as failure elements and overuse of shoulder, trauma, and shoulder arthritis [3]. resulted in higher rate of no-heal and re-tear [8, 9]. Fatty infiltration was introduced and staged by Goutallier et * Correspondence: [email protected] al. in 1994 and was viewed as a key factor responsible 1 Department of Orthopedic Surgery, Second Affiliated Hospital, School of for the higher re-tear rates and poorer functional prog- Medicine, Zhejiang University, 88 Jiefang Road, Hangzhou, Zhejiang 310009, People’s Republic of China nosis [10, 11]. Fatty degeneration of the infraspinatus Full list of author information is available at the end of the article

© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 2 of 10

was reported as the most independent predictor on out- lmFit, eBayes, and topTable functions, was used for pair- comes following rotator cuff repair [12]. Chung et al. wise comparison of DEGs [16]. P <0.05andabs(log2)fold also found fatty infiltration of the infraspinatus is a sig- change (FC) > 1 were used as the cutoff criteria. nificantly positive correlated factor with healing failure rate after arthroscopic repair of a massive rotator cuff tear [6]. Another cohort study performed by Gladstone GO annotation and KEGG pathway enrichment analyses et al. demonstrated that both muscle atrophy and fatty of DEGs infiltration of the rotator cuff muscles cast significant GO annotation is a classic method used in bioinformat- impacts on regulating cuff repair and functional recovery ics analyses to describe the normal, in vivo biological [13]. Therefore, an increasing large number of research role of genes or gene products [17]. KEGG, a collection groups are focusing on the specific mechanism of fatty of databases representing our knowledge on the molecu- infiltration. lar interaction, reaction, and relation networks, was used Till now, the key genes and pathway involved in the to clarify the potential role of the DEGs [18]. GO func- fatty infiltration of rotator cuff tear still unclear. Ren et tional analyses encompassing biological processes (BP), al. identified differentially expressed genes and signaling cell components (CC), and molecular functions (MF) pathways involved in rotator cuff tear through bioinfor- was performed using the Database for Annotation, matics analysis [14]. Their study focused on the different Visualization and Integrated Discovery (DAVID ver. 6.8), genes between torn tendon samples and intact samples. with P < 0.05 used as a cutoff. In the present study, to explore differential gene expres- sion of the differential stage of RCT, we examined one published gene dataset available in the Gene Expression PPI network construction and hub gene identification Omnibus (GEO) database. Differential expression of A PPI network for the DEGs identified above was built genes during each developmental stages of massive rota- based on The Search Tool for the Retrieval of Interact- tor cuff tear was examined by comprehensive bioinfor- ing Genes (STRING, http://string-db.org) and Cytoscape matics analyses. Then Gene Ontology (GO) and Kyoto (ver. 3.5.1) [19]. In the PPI network, genes served as Encyclopedia of Genes and Genomes (KEGG) pathway “nodes” and the line segment between two nodes repre- enrichment analyses were performed, and a –pro- sented associated interactions. The color of the lines be- tein interaction (PPI) network was used to screen for tween genes indicates the degree of interaction. The crucial genes and pathways in the pathophysiological Centiscape plugin was used to determine the degree of processes of fatty infiltration. Chief sub-PPI networks connectivity for each node in the PPI network. Genes are explored by MCODE. with a degree >10 were defined as hub genes. Further- more, module analysis was performed by using the Methods MCODE plugin (version 1.3) to explore the most im- Expression profile dataset portant clustering modules in the huge PPI network (de- Gene expression profiles for rats with a massive tear of gree cutoff = 5, k-core = 2, node score cutoff = 0.2, and the rotator cuff at different development stages max. Depth rom Seed: 100). (GSE103266) was obtained from the GEO database. In this experiment, rats were established by a bilateral full-thickness supraspinatus tear and suprascapular Results neurectomy and divided into four groups according to Data distribution analyses and DEG screening the period after operation: 0 days (un-operated controls), In GSE103266, a total of 17147 genes were detected in 10 days, 30 days, and 60 days post-injury. The count data 16 samples. The expression values (ranging from 0 to 20 in GSE103266 was normalized by DESeq2 package in R log2 (TPM)) and the distributions were similar between software and presented in log2(TPM) [15]. The quality the four groups (Fig. 1a, b). H-cluster analyses and prin- of gene expression data was analyzed and visualized cipal component showed that samples were easily using the ggplot2 package of R software for each sample. grouped into different groups (Fig. 1c, d). Confirmed by This research has been approved by the IRB of the au- these data distribution analyses, further bioinformatics thors’ affiliated institutions. analyses could be performed based on the available data. Following data processing using limma, we identified Differential gene expression analyses 2877 DEGs (1554 upregulated, 1323 downregulated) in We compared the different three stages of rats (10 days, 30 group 0 day vs. group 10 days (Fig. 2b), 2743 DEGs (1711 days, and 60 days after operation) with un-operated con- upregulated, 1032 downregulated) in group 0 day vs. group trols to identify the DEGs. The Linear Models for Micro- 30 days (Fig. 2c), and 2199 DEGs (1439 upregulated, 760 array Data (LIMMA) (limma) package, which includes downregulated) in group 0 day vs. group 60 days (Fig. 2d). Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 3 of 10

AC

BD

Fig. 1 Distribution analyses of gene expression. a Distribution of gene expression levels in GSE103266. b Distribution of genes expression of 16 samples. c Cluster analyses for all samples. d Principle component analyses of each samples

Intersection between the DEGs (Fig. 5a). The top five significantly enriched KEGG path- We identified a total of 1089 DEGs common across each ways were the HTLV-I infection, phagosome, cell adhesion of the different stages by intersecting the three compari- molecules, tuberculosis, and osteoclast differentiation. De- sons above (Fig. 3a), including 733 upregulated and 356 tailed information regarding each of the DEGs involved in downregulated genes (Additional file 1: Table S1). Hier- the KEGG pathway are listed in Fig. 5b. archical clustering of the identified DEGs is displayed as a heatmap in Fig. 3b. PPI network construction and module analysis GO annotation and KEGG pathway enrichment analyses We constructed a putative PPI network map for the GO annotation and KEGG pathway enrichment analyses overlapping DEGs using the STRING database and were performed to better understand the functional signifi- Cytoscape(Fig. 6). After degree calculation, a total of 259 cance of the DEGs. DEGs mainly partcipated in biological hub genes were identified with the degree more than 10, processes (BP) regarding response to drug, inflammatory as listed in Additional file 2: Table S2. Among these, response, aging, response to lipopolysaccharide and im- Tmprss11d, Ptprc, Itgam, Mmp9, Tlr2, Il1b, Il18, Ccl5, mune response. Significant cellular component (CC) terms Cxcl10, and Ccr7 were the top ten hub genes according revealed DEGs were principally enriched in extracellular to the MCODE degree score. exosome, extracellular space, and cell surface. Finally, the A total of six modules were separated from the PPI top five molecular functions (MF) identified were protein network with both MCODE score ≥ 5 and nodes ≥ 5 homodimerization activity, calcium ion binding, receptor (Fig. 7). KEGG analysis showed that the main pathways binding, heparin binding, and carbohydrate binding of these modules were associated with cytokine-cytokine (Fig. 4a). All significant biological processes-enriched en- receptor interaction, rheumatoid arthritis and toll-like triesareshowninahistograminFig.4b. KEGG analyses of receptor signaling pathway, osteoclast differentiation, the DEGs revealed mainly enrichment for ten terms metabolic pathways, fatty acid degradation, and so on. Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 4 of 10

A B 3000 The number of up gene is 1554 The number of down gene is 1323

150 2000

Group

DOWN

value 100 change UP DOWN DEGs Counts 1000 NOT UP

log10 p 50

0 10 days 30 days 60 days 0 50 510 log2 fold change

CDThe number of up gene is 1711 The number of up gene is 1439 The number of down gene is 1032 The number of down gene is 760

80 150

60 value 100 change value change DOWN DOWN NOT 40 NOT UP UP log10 p log10 p 50 20

0 0 50 5 50 5 log2 fold change log2 fold change Fig. 2 a Overview of DEGs in three comparisons. b Volcano plots displaying of group 0 day vs. group 10 days. c Volcano plots displaying of group 0 day vs. group 30 days. d Volcano plots displaying of group 0 day vs. group 60 days

AB

Fig. 3 a Venn diagram of DEGs across different stages. b Hierarchical clustering analyses of DEGs common to all groups Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 5 of 10

A extracellular exosome cellular response to cytokine stimulus response to nicotine positive regulation of phosphatidylinositol 3−kinase signaling neutrophil chemotaxis transmembrane receptor protein tyrosine kinase signaling pathway cell cycle arrest positive regulation of T cell proliferation response to cytokine ossification wound healing

osteoblast differentiation BP positive regulation of ERK1 and ERK2 cascade cellular response to lipopolysaccharide cytokine−mediated signaling pathway response to estradiol −log10(PValue) positive regulation of cell migration response to organic cyclic compound innate immune response immune response 12.5 response to lipopolysaccharide aging 10.0 inflammatory response response to drug 7.5

5.0 extracellular matrix proteinaceous extracellular matrix external side of plasma membrane CC cell surface extracellular space extracellular exosome

fibronectin binding non−membrane spanning protein tyrosine kinase activity glycoprotein binding carbohydrate binding MF heparin binding receptor binding calcium ion binding protein homodimerization activity 0 50 100 150 200 Count B BP Term Count PValue FDR Gene response to drug 61 2.15E-08 3.98E-05 TSPO, ADCY1, APOBEC1, MMP9, AQP7, MMP2, TGFB1, TGFB2, B2M, CCNE1, CASP3, CDKN2A, APOD, PDE4A, GPX3, GNPAT, IL1B, LOX, MYC, MAP2K6, NDUFA10L1, GSTT1, CYP2E1, MMP13, MMP12, PRKCB, NCAM1, CD86, ABCB1B, LCK, HSD11B2, COL1A1, ERBB3, ERBB2, MGMT, ABCA1, CCL5, TIMP2, MDK, ARG1, ACSL1, OXCT1, FN1, GSTA1, DNMT3A, BGLAP, GNAO1, AK1, IL1RN, ATP1A3, NCKAP1L, IGF2, AK4, TP73, PSMB9, CDKN1A, CYBB, SFRP1, SFRP2, ADRA1B, IGFBP2

inflammatory response 51 3.29E-13 6.08E-10 ELF3, CD8A, S100A8, AIF1, IL18, TLR2, IL17RE, NFKB2, TLR5, CXCR3, SGMS1, TLR7, TLR8, TGFB1, CXCL10, CD44, IL1B, SYK, CIITA, HYAL1, C5AR1, C4A, C4B, PDPN, TNFRSF14, NLRP3, PRKCQ, CCR7, SERPINA3N, CCR5, NGFR, C3, CCR1, MAPKAPK2, TAC4, CCL5, CCL7, SLC11A1, KLKB1, CSF1R, SPP1, SELP, HCK, CHI3L1, PTGFR, TP73, CD180, LAT, CYBB, CXCL13, IGFBP4

aging 43 3.67E-08 6.78E-05 TSPO, LDLR, MMP9, COL3A1, CNP, ITGB2, NFKB2, TIMP2, CCL5, MOG, MMP2, TGFB1, TIMP1, TGFB2, ARG1, CDKN2A, APOD, CDKN2B, APOE, DMD, RGN, CIITA, NOX4, GSTA1, DNMT3A, EEF1A1, BGLAP, GNAO1, SOCS2, BCL2A1, IGFALS, SERPING1, NCAM1, PRKCQ, CD86, LRP1, CCR5, PENK, SERPINF1, TFRC, LCK, CP, IGFBP2

response to lipopolysaccharide 40 3.42E-08 6.33E-05 TG, ALDOA, S100A8, MMP9, TLR2, ACP5, FASLG, CNP, NFKB2, MAPKAPK2, MMP3, CCL5, TRIB1, CXCL10, SLC11A1, ARG1, CASP3, IL10RA, CSF2RB, IL1B, C2, LBP, LOXL1, SFTPB, SELP, C5AR1, CFB, IL1RN, TNFRSF14, PTGFR, SERPINA3N, CCR7, CD86, CCR5, PENK, CXCL13, ABCB1B, SLPI, NGFR, MGST2 immune response 38 8.77E-08 1.62E-04 LST1, RT1-DB2, RT1-DB1, CCR1, TLR2, FASLG, OAS2, RT1-BA, CD74, RT1-DA, CXCL10, ZAP70, RT1-CE10, IL1B, NFIL3, CD28, RT1-CE2, FYB, RT1-CE1, RT1-CE6, RT1-CE4, SMAD6, RT1-S2, TNFRSF14, CTSS, SECTM1B, VAV1, CTSW, RT1-A2, LAT, RT1-A1, PLSCR1, CCR7, CCR5, FCGR2B, CXCL13, SLPI, NGFR innate immune response 35 1.24E-06 2.30E-03 TNFAIP8L2, FGR, S100A8, BLK, TLR2, KLRK1, TRIM14, NFKB2, OAS2, TLR5, C1S, TLR7, TLR8, B2M, TMEM173, CLEC4A, FCER1G, LBP, MATK, SYK, CSF1R, TEC, ITK, C4A, LGALS3, HCK, SERPING1, C1QA, CYBB, LCK, SLPI, CLEC7A, JAK3, CLEC5A, SRMS response to organic cyclic compound 33 2.14E-05 3.96E-02 LUM, MMP9, SLC37A4, MGMT, ACP5, CCL5, TGFB1, CCNE1, CASP3, ACSL1, CDKN2A, UGT1A7C, CD44, CDKN2B, GPX3, IL1B, SCN5A, SELP, BGLAP, GNAO1, IL1RN, RXRG, GSTT1, IGF2, MMP14, MMP12, PRKCQ, CD83, CDKN1A, G6PD, SFRP1, CCND2, ABCB1B positive regulation of cell migration 29 4.25E-06 7.86E-03 FGR, AIF1, CCR1, ACP5, MMP3, LRRC15, CCL5, MMP2, CCL7, TGFB1, CXCL10, TGFB2, SEMA7A, PDGFC, FGF1, CSF1R, FN1, PDPN, PIK3CD, MYO1F, MMP14, SNAI2, CORO1A, CXCL16, ROR2, LAMC2, SEMA4D, COL1A1, GRB7

response to estradiol 28 8.59E-06 1.59E-02 BID, LDLR, C3, MMP9, POSTN, MMP3, SHH, TGFB1, TGFB2, CCNE1, CASP3, FCGR1A, IL1B, CD4, MYC, GHR, DNMT3A, SOCS2, AK1, ESR1, IGF2, PTGFR, FCGR2B, PENK, ABCB1B, CCND2, COL1A1, IGFBP2 cytokine-mediated signaling pathway 27 1.50E-08 2.77E-05 EDN2, IL21R, IL17RE, LRRC15, GREM2, IL17RA, CD44, IL10RA, CSF2RB, IL1B, CSF2RA, IFNLR1, CSF1R, GHR, LRRC4, FLRT3, FLRT2, FLRT1, SOCS2, FLT3, IL1RN, PTPRN, CISH, BGN, IRF5, JAK3, IL22RA2 cellular response to lipopolysaccharide 26 2.62E-06 4.84E-03 PPARD, TSPO, IL18, MMP9, KLRK1, TLR5, ABCA1, SGMS1, ANKRD1, CCL5, CXCL10, B2M, ARG1, TICAM2, IL1B, LBP, FN1, MRC1, EPAS1, NLRP3, CD180, PLSCR1, CD86, CCR5, FCGR2B, CXCL16 positive regulation of ERK1 and ERK2 cascade 26 1.81E-05 3.35E-02 GPR183, C3, CCR1, TLR2, CCL5, CD74, TGFB1, CCL7, C1QTNF3, CD44, NDRG4, SEMA7A, RASGRP1, IL1B, PDGFC, GPNMB, CSF1R, NOX4, C5AR1, CHI3L1, ESR1, NTSR2, SLAMF1, SERPINF2, TREM2, ST5 osteoblast differentiation 21 4.35E-06 8.04E-03 IBSP, BGLAP, IL18, MRC2, SOX2, ITGA11, IGF2, LRRC17, ACTN3, SNAI2, MMP13, PENK, SFRP1, SEMA7A, COL1A1, TMEM119, GPNMB, RUNX2, CHRD, TP53INP2, SPP1 wound healing 21 1.06E-05 1.96E-02 PPARD, FGF7, LTBP1, S100A8, ERBB2, COL3A1, POSTN, IGF2, MUSTN1, MMP3, MMP12, TGFB1, TIMP1, TGFB2, SLC11A1, NOTCH2, CASP3, IL1B, COL1A1, LOX, FN1 ossification 19 3.56E-06 6.58E-03 IBSP, BGLAP, PDLIM7, MMP9, MYF5, MGP, ACP5, IGF2, MMP14, MMP13, SLC26A2, CACNA1S, IGSF10, MYOG, COL1A1, RUNX1, COL11A1, FN1, SPP1 response to cytokine 19 4.14E-06 7.64E-03 GNAO1, AIF1, COL3A1, ACP5, NFKB2, MAPKAPK2, MMP3, CCL5, TIMP2, MMP2, TIMP1, TGFB2, CCNE1, CORO1A, SERPINA3N, CDKN2B, CXCL16, XCR1, GHR positive regulation of T cell proliferation 17 2.48E-07 4.59E-04 ITGAL, PTPRC, AIF1, NCKAP1L, CCL5, NCK2, PRKCQ, CD86, CORO1A, TFRC, CD46, IL1B, CD4, JAK3, SASH3, SPN, CD28 cell cycle arrest 17 1.58E-06 2.93E-03 WNT10B, CGRRF1, AK1, SOX2, RPRM, TP73, TGFB1, TGFB2, MLF1, MFN2, NOTCH2, CDKN1A, CDKN2A, CDKN2B, GADD45A, MYC, TP53INP1 transmembrane receptor protein tyrosine kinase 17 6.42E-06 1.19E-02 ITK, FGR, FLT3, ERBB3, ERBB2, HCK, BLK, ALK, DOK1, LCK, ROR2, JAK3, SRMS, MATK, SYK, CSF1R, TEC signaling pathway neutrophil chemotaxis 16 2.36E-07 4.37E-04 C5AR1, S100A8, LGALS3, EDN2, SLC37A4, NCKAP1L, ITGB2, CCL5, VAV1, CCL7, ITGAM, TGFB2, IL1B, FCER1G, SPP1, SYK positive regulation of phosphatidylinositol 3- 14 2.61E-05 4.82E-02 SELP, PPARD, WNT16, FGR, SERPINA12, ERBB2, HCLS1, PDGFC, SEMA4D, CCL5, PRR5L, CD28, HCST, TGFB2 kinase signaling response to nicotine 14 2.61E-05 4.82E-02 ALDOA, CASP3, DHFR, PENK, MMP9, HMOX1, CHRNA5, IGF2, ATP1A2, CHRNA1, MMP2, CHRNE, SLC7A11, CHRNG cellular response to cytokine stimulus 12 1.37E-05 2.53E-02 PLSCR1, CD86, CCR7, FLT3, MMP9, HCLS1, DPYSL3, JAK3, CCL5, MMP2, IL17RA, CSF1R Fig. 4 a Histogram of Gene Ontology (GO) functional classification of DEGs. GO terms including biological processes (BP), cellular components (CC), and molecular functions (MF). b Top BP terms and their corresponding genes in GO functional enrichment analyses

Discussion group. And the severity of fatty infiltration was closely The mechanism of fatty infiltration of the rotator cuff is linked with the shoulder function [20]. Another massive complex and still unclear. Using a mouse model con- rotator cuff tear model caused by a full-thickness tenect- structed by unilateral supraspinatus and infraspinatus omy of the right supraspinatus and infraspinatus in rats tendon tear, Wang et al. demonstrated atrophy and fatty showed that the muscle fiber force is decreased and ex- infiltration of muscle exist in the 6-week delayed repair pressions of fibrosis and lipid accumulation related Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 6 of 10

A

Rheumatoid arthritis

Viral myocarditis

Staphylococcus aureus infection −log10(PValue) Antigen processing and presentation

Natural killer cell mediated cytotoxicity 7

Osteoclast differentiation 6

Tuberculosis 5

Cell adhesion molecules (CAMs)

Phagosome

HTLV−I infection

0 10203040 B Count KEGG Term Count PValue FDR Gene HTLV-I infection 39 1.04E-05 1.36E-02 ITGAL, TSPO, ADCY1, WNT16, RT1-DB1, LOC100360552, SPI1, PPP3R1, ITGB2, NFKB2, RT1-BA, RT1-DA, TGFB1, TGFB2, MSX3, CDKN2A, CDKN2B, POLE2, RT1-CE10, PIK3R5, MYC, TERT, RT1-CE2, WNT10B, CD3G, RT1-CE1, RT1-CE4, PIK3CD, RT1-M2, RT1-CE14, FZD4, RT1-A2, RT1-A1, CDKN1A, ADCY9, CCND2, LCK, JAK3, TP53INP1 Phagosome 35 3.18E-08 4.15E-05 MSR1, C3, RT1-DB1, TLR2, ITGB2, RT1-BA, RT1-DA, ITGAM, FCGR1A, COMP, TAP1, RT1-CE10, DYNC2H1, TUBB6, THBS2, TUBA1B, ATP6V0D2, THBS4, RT1- CE2, MRC1, RT1-CE1, RT1-CE4, NCF4, MRC2, RT1-M2, RT1-CE14, CTSS, RT1- A2, RT1-A1, ATP6V1C2, CORO1A, TUBA8, FCGR2B, TFRC, CLEC7A Cell adhesion molecules (CAMs) 33 2.13E-08 2.77E-05 ITGAL, CD8A, CD8B, RT1-DB1, ITGB2, CDH2, RT1-BA, RT1-DA, ITGAM, ITGB7, RT1-CE10, CD22, CD4, NEGR1, SELPLG, SPN, CD28, RT1-CE2, LRRC4, SELP, PTPRC, RT1-CE1, RT1-CE4, RT1-M2, RT1-CE14, NFASC, NRXN1, RT1-A2, NCAM1, RT1-A1, SIGLEC1, CD86, CNTN2 Tuberculosis 31 9.25E-07 1.21E-03 BID, C3, RT1-DB1, IL18, TLR2, PPP3R1, ITGB2, RT1-BA, CD74, RT1-DA, TGFB1, ITGAM, TGFB2, CASP3, CALML3, FCGR1A, IL10RA, FCER1G, IL1B, LBP, CAMK2A, ATP6V0D2, SYK, CIITA, MRC1, MRC2, CTSS, CORO1A, FCGR2B, CALM3, CLEC7A Osteoclast differentiation 26 3.42E-07 4.46E-04 SPI1, PPP3R1, ACP5, NFKB2, TGFB1, TGFB2, BTK, FCGR1A, LILRA5, IL1B, PIK3R5, MAP2K6, TYROBP, SYK, CSF1R, TEC, NCF4, PIK3CD, TAB2, SIRPA, CTSK, CYBB, FCGR2B, LILRB3, LCK, TREM2 Natural killer cell mediated cytotoxicity 22 5.30E-07 6.91E-04 BID, ITGAL, PIK3CD, PPP3R1, KLRK1, FASLG, ITGB2, NCR1, VAV1, PRKCB, HCST, LAT, CASP3, RAC2, LCK, ZAP70, FCER1G, PIK3R5, KLRD1, SH3BP2, SYK, TYROBP Antigen processing and presentation 20 7.54E-06 9.83E-03 CIITA, RT1-CE2, RT1-CE1, CD8A, CD8B, RT1-CE4, RT1-DB1, RT1-CE14, RT1- M2, CTSS, RT1-BA, RT1-DA, CD74, B2M, RT1-A2, RT1-A1, TAP1, RT1-CE10, CD4, KLRD1 Staphylococcus aureus infection 18 1.56E-08 2.04E-05 SELP, ITGAL, C5AR1, C4A, MASP1, CFB, C3, RT1-DB1, ITGB2, C1S, RT1-BA, ITGAM, RT1-DA, C1QA, FCGR2B, FCGR1A, C2, SELPLG Viral myocarditis 18 3.72E-05 4.85E-02 RT1-CE2, BID, ITGAL, RT1-CE1, RT1-CE4, RT1-DB1, RT1-CE14, RT1-M2, ITGB2, RT1-BA, RT1-DA, RT1-A2, RT1-A1, CD86, CASP3, RAC2, RT1-CE10, CD28 Rheumatoid arthritis 18 3.72E-05 4.85E-02 ITGAL, IL18, RT1-DB1, TLR2, ACP5, ITGB2, RT1-BA, MMP3, CCL5, RT1-DA, TGFB1, TGFB2, ATP6V1C2, CD86, CTSK, IL1B, ATP6V0D2, CD28 Fig. 5 a Histogram of KEGG pathways enrichment in DEGs. The graph displays only significantly enriched KEGG terms (P < 0.05), with darker red indicating greater significance. b Individual KEGG terms and their corresponding genes are shown for each group genes were increased [21]. A further study performed by grade 3 fatty degeneration showed significant improve- Liu et al. demonstrated that the main cellular sources in- ment after arthroscopic rotator cuff repair compared to volved in fatty infiltration were Tie2 + progenitor cells those patients who graded as Goutallier stage IV after and fibro/adipogenic progenitor cells [22]. These three nearly 3 years of follow-up [24]. However, a successful studies were consistent with the clinical reports and re- repair of rotator cuff would not bring a significant im- searches. In a prospective study aimed to massive rotator provement of fatty infiltration and atrophy of the rotator cuff tears patients younger than 65 years, 67.9% were cuff [13]. So, studies focused on the cellular and molecu- demonstrated with a stage 3 or 4 fatty infiltration of the lar basis of fatty infiltration are of great significance in supraspinatus muscle and 47.3% were suffered from a the prevention and treatment of fatty infiltration in rota- stage 3 or 4 fatty infiltration of the infraspinatus muscle tor cuff tear. [23]. And patients with a high degree of fatty infiltration In the present study, a bioinformatics analysis was exhibited poorer prognosis after surgery. Patients with used to analyze an available database downloaded from Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 7 of 10

Fig. 6 Protein–protein interaction (PPI) network construction. Genes with yellow circles represent hub genes with a degree of interaction ≥ 10. A total of 259 DEGs were identified as fatty infiltration-related hub genes

GEO. According to our results, a total of 1089 DEGs At present, there are only a few theories and research were identified as the key mediators in the pathophysio- results about the mechanism of fatty infiltration in rota- logical processes of fatty infiltration. Moreover, we also tor cuff. Tmprss11d, a member of type II transmem- performed GO annotation and KEGG pathway enrich- brane serine protease (TTSP) family, was highly ment analyses to potential biological functions and path- expressed in respiratory epithelium and proved to exert ways involved in fatty infiltration. In addition, 259 hub an important role in the host defense system [25]. Till genes including Tmprss11d, Ptprc, Itgam, Mmp9, Tlr2, now, the role of TMPRSS11D in the musculoskeletal Il1b, Il18, Ccl5, Cxcl10, and Ccr7 and six modules with system was not clear. Cao et al. reported higher expres- important functions were identified from the PPI sion of TMPRSS11D is closely linked with the severity of network. the cancer because of its ability to disintegrate the Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 8 of 10

A

BC

Fig. 7 a The significant six modules in the PPI network with both MCODE score ≥ 5 and nodes ≥ 5. b Histogram of KEGG pathways enrichment for these six modules. c Individual KEGG terms and their corresponding genes extracellular matrix [26]. TMPRSS11D was also demon- fat infiltration compared to the wildtype MMP-13 (+/+) strated to have a relationship with fat metabolism in the mice group. These interesting results indicated that liver [25]. Therefore, it is an important research direc- MMP-13 exert a significant role in rotator muscle fatty tion of the role of TMPRSS11D in the degeneration of infiltration and may be a potential target for treatment tendon and lipid metabolism in the future. MMP family [28]. MMP inhibitor was helpful in the improvement of was also identified as a major player in rotator cuff tear. tendon-to-bone healing in a rotator cuff repair model Protein expressions of MMP-9 and MMP-13 were not- [29]. Moreover, Tlr2, another identified hub gene in our ably increased in rotator cuff specimens obtained from study, was highly expressed in rotator cuff tendon in the surgery patients [27]. However, in MMP-13 (−/−) knock- first 2 weeks following injuries [30]. Tlr2 was viewed as out mice group, researchers found a higher amount of an important mediator in NLRP3 inflammasome. The Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 9 of 10

lowered activity of NLRP3 leads to a decrease in the fat Additional file 2: Table S2. Hub-genes identified by CentiScape (degree accumulation in hepatocytes [31]. So, the molecular me- ≥ 5) among 1089 DEGs. (DOCX 35 kb) diators associated with NLRP3 inflammasome was worth studying in the pathology of fatty infiltration in rotator Abbreviations cuff tendons. Taken together, hub genes identified in our BP: Biological processes; CC: Cell components; DEGs: Differentially expressed genes; GEO: Gene Expression Omnibus; GO: Gene ontology; KEGG: Kyoto research by comprehensive bioinformatic analyses were Encyclopedia of Genes and Genomes; LIMMA: Linear Models for Microarray meaningful and worthy of deep-going study in the data; MF: Molecular functions; PPI: Protein–protein interaction; RCT: Rotator future. cuff tear We also performed an intensive module analysis based Acknowledgements on the PPI network in this study. In chronically rotator Not applicable. cuff muscles, increased gene expressions of PPAR-γ, PLD1, miR-27a and lipid synthesis were obversed. And Funding None ECM degradation- and remodeling-related genes such as MMP2, MMP9, and MMP14 were also overexpressed in Availability of data and materials the pathological process of fibrosis [32]. Blocking of p38 The data was freely downloaded from the public GEO database.

MAPK signaling, an important regulator of adipogenesis, Authors’ contributions reduced the fat accumulation by nearly half and adjusted WLD conceived the design of the study. HPF and JLF participated in the the expressions of adipogenesis- and matrix design of this study and performed the statistical analysis. They both contributed to the collection of background information. HPF drafted the accumulation-related genes [33]. Our KEGG pathway manuscript. All authors read and approved the final manuscript. enrichment analyses showed that the functions of mod- ule 4 and 5 were mainly associated with metabolic path- Ethics approval and consent to participate ways and fatty acid degradation. And DEGs such as Not applicable. This paper does not involve research on humans. Acsl1, Acsl4, Acsl6, Lipg, Lpin1, Mboat2, Pnpla3, Consent for publication Ppap2a, Ccbl1, Impdh1, and Nt5c1a were functionality All authors read and approved the final manuscript and consented to associated with these pathways. publication.

A limitation of this study is that the stage of fatty infil- Competing interests tration in the rat models in GSE103266 is unclear. Be- The authors declare that they have no competing interest. sides, there are some anatomical differences between human and rat. Therefore, the pathophysiological Publisher’sNote process of fatty infiltration in rat is different from hu- Springer Nature remains neutral with regard to jurisdictional claims in man. We still need to conduct validation experiment to published maps and institutional affiliations. proof our speculation in the future. Author details 1Department of Orthopedic Surgery, Second Affiliated Hospital, School of Medicine, Zhejiang University, 88 Jiefang Road, Hangzhou, Zhejiang 310009, Conclusion People’s Republic of China. 2Orthopedics Research Institute of Zhejiang ’ The current study identified a series of DEGs in each stage University, Hangzhou, People s Republic of China. of massive rotator cuff tear. Using a series of bioinformat- Received: 12 March 2019 Accepted: 30 April 2019 ics analyses, hub genes including Tmprss11d, Ptprc, Itgam, Mmp9, Tlr2, Il1b, Il18, Ccl5, Cxcl10, and Ccr7 References were identified. Further GO and KEGG pathway analysis 1. Chard MD, Hazleman R, Hazleman BL, King RH, Reiss BB. Shoulder disorders suggests potential mechanisms through which these hub in the elderly: a community survey. Arthritis Rheum. 1991;34:766–9. genes mediate fatty infiltration pathogenesis in rotator cuff 2. Minagawa H, Yamamoto N, Abe H, Fukuda M, Seki N, Kikuchi K, et al. Prevalence of symptomatic and asymptomatic rotator cuff tears in the tear. In addition, module analysis highlights the important general population: from mass-screening in one village. J Orthop. 2013;10: roles of metabolic pathways and fatty acid degradation in 8–12. the regulation of fatty infiltration. Our results indicated 3. Sayampanathan AA, Andrew TH. Systematic review on risk factors of rotator cuff tears. J Orthop Surg (Hong Kong). 2017;25:2309499016684318. the potential key genes and pathways involved in fatty de- 4. Oliva F, Piccirilli E, Bossa M, Via AG, Colombo A, Chillemi C, et al. I.S.Mu.L.T - generation and supplied therapeutic targets to improve Rotator cuff tears guidelines. Muscles Ligaments Tendons J. 2015;5:227–63. clinical outcomes for patients who suffer from chronic ro- 5. Lo IK, Burkhart SS. Arthroscopic revision of failed rotator cuff repairs: technique and results. Arthroscopy. 2004;20:250–67. tator cuff tears and fatty infiltration. 6. Chung SW, Kim JY, Kim MH, Kim SH, Oh JH. Arthroscopic repair of massive rotator cuff tears: outcome and analysis of factors associated with healing failure or poor postoperative function. Am J Sports Med. 2013;41:1674–83. Additional files 7. Gerber C, Fuchs B, Hodler J. The results of repair of massive tears of the rotator cuff. J Bone Joint Surg Am. 2000;82:505–15. 8. Rashid MS, Cooper C, Cook J, Cooper D, Dakin SG, Snelling S, et al. Additional file 1: Table S1. 1089 DEGs in each of the different stages of rotator cuff tear. (DOCX 148 kb) Increasing age and tear size reduce rotator cuff repair healing rate at 1 year. Acta Orthop. 2017;88:606–11. Hu et al. Journal of Orthopaedic Surgery and Research (2019) 14:158 Page 10 of 10

9. Valencia AP, Lai JK, Iyer SR, Mistretta KL, Spangenburg EE, Davis DL, et al. 33. Wilde JM, Gumucio JP, Grekin JA, Sarver DC, Noah AC, Ruehlmann DG, et al. Fatty infiltration is a prognostic marker of muscle function after rotator cuff Inhibition of p38 mitogen-activated protein kinase signaling reduces fibrosis tear. Am J Sports Med. 2018;46:2161–9. and lipid accumulation after rotator cuff repair. J Shoulder Elbow Surg. 10. Goutallier D, Postel JM, Bernageau J, Lavau L, Voisin MC. Fatty muscle 2016;25:1501–8. degeneration in cuff ruptures. Pre- and postoperative evaluation by CT scan. Clin Orthop Relat Res. 1994;304:78–83. 11. Montgomery SR, Petrigliano FA, Gamradt SC. Failed rotator cuff surgery, evaluation and decision making. Clin Sports Med. 2012;31:693–712. 12. Oh JH, Kim SH, Ji HM, Jo KH, Bin SW, Gong HS. Prognostic factors affecting anatomic outcome of rotator cuff repair and correlation with functional outcome. Arthroscopy. 2009;25:30–9. 13. Gladstone JN, Bishop JY, Lo IK, Flatow EL. Fatty infiltration and atrophy of the rotator cuff do not improve after rotator cuff repair and correlate with poor functional outcome. Am J Sports Med. 2007;35:719–28. 14. Ren YM, Duan YH, Sun YB, Yang T, Tian MQ. Bioinformatics analysis of differentially expressed genes in rotator cuff tear patients using microarray data. J Orthop Surg Res. 2018;13:284. 15. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:550. 16. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47. 17. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25:25–9. 18. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28:27–30. 19. Szklarczyk D, Franceschini A, Wyder S, Forslund K, Heller D, Huerta-Cepas J, et al. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic Acids Res. 2015;43:D447–52. 20. Wang Z, Liu X, Davies MR, Horne D, Kim H, Feeley BT. A mouse model of delayed rotator cuff repair results in persistent muscle atrophy and fatty infiltration. Am J Sports Med. 2018;46:2981–9. 21. Gumucio JP, Korn MA, Saripalli AL, Flood MD, Phan AC, Roche SM, et al. Aging-associated exacerbation in fatty degeneration and infiltration after rotator cuff tear. J Shoulder Elbow Surg. 2014;23:99–108. 22. Liu X, Ning AY, Chang NC, Kim H, Nissenson R, Wang L, et al. Investigating the cellular origin of rotator cuff muscle fatty infiltration and fibrosis after injury. Muscles Ligaments Tendons J. 2016;6:6–15. 23. Berhouet J, Collin P, Benkalfate T, Le Du C, Duparc F, Courage O, et al. Massive rotator cuff tears in patients younger than 65 years. Epidemiology and characteristics. Orthop Traumatol Surg Res. 2009;95:S13–8. 24. Burkhart SS, Barth JR, Richards DP, Zlatkin MB, Larsen M. Arthroscopic repair of massive rotator cuff tears with stage 3 and 4 fatty degeneration. Arthroscopy. 2007;23:347–54. 25. Sales KU, Hobson JP, Wagenaar-Miller R, Szabo R, Rasmussen AL, Bey A, et al. Expression and genetic loss of function analysis of the HAT/DESC cluster proteases TMPRSS11A and HAT. PLoS One. 2011;6:e23261. 26. Cao X, Tang Z, Huang F, Jin Q, Zhou X, Shi J. High TMPRSS11D protein expression predicts poor overall survival in non-small cell lung cancer. Oncotarget. 2017;8:12812–9. 27. Jacob J, Eisemon E, Sheibani-Rad S, Patel A, Jacob T, Choueka J. Matrix metalloproteinase levels as a marker for rotator cuff tears. Orthopedics. 2012;35:e474–8. 28. Liu X, Ravishankar B, Ning A, Liu M, Kim HT, Feeley BT. Knocking-out matrix metalloproteinase-13 exacerbates rotator cuff muscle fatty infiltration. Muscles Ligaments Tendons J. 2017;7:202–7. 29. Bedi A, Kovacevic D, Hettrich C, Gulotta LV, Ehteshami JR, Warren RF, et al. The effect of matrix metalloproteinase inhibition on tendon-to-bone healing in a rotator cuff repair model. J Shoulder Elbow Surg. 2010;19:384–91. 30. Thankam FG, Roesch ZK, Dilisio MF, Radwan MM, Kovilam A, Gross RM, et al. Association of inflammatory responses and ECM disorganization with HMGB1 upregulation and NLRP3 inflammasome activation in the injured rotator cuff tendon. Sci Rep. 2018;8:8918. 31. Yang G, Lee HE, Lee JY. A pharmacological inhibitor of NLRP3 inflammasome prevents non-alcoholic fatty liver disease in a mouse model induced by high fat diet. Sci Rep. 2016;6:24399. 32. Gumucio J, Flood M, Harning J, Phan A, Roche S, Lynch E, et al. T lymphocytes are not required for the development of fatty degeneration after rotator cuff tear. Bone Joint Res. 2014;3:262–72.