Microarray analysis for differentially expressed genes of patients undergoing total knee arthroplasty with ischemia preconditioning.

BACKGROUND: Ischemia preconditioning (IPC) has been proved as a powerful method of protecting tissues against ischemia reperfusion insults. We aimed to elucidate the mechanism of IPC in ischemia reperfused tissues.
METHODS: GSE21164 containing 16 muscle biopsies taken from the operative knee of four IPC-treated patients and four control at the onset of surgery (T¿=¿0) and 1 h into surgery (T¿=¿1) undergoing primary total knee arthroplasty was downloaded from the Gene Expression Omnibus (GEO) database. Differentially expressed genes (DEGs) between IPC group and control were screened with Limma package in R language. KEGG pathway enrichment analysis was performed by the DAVID online tool. Meanwhile, potential regulatory microRNAs (miRNAs) for downregulated DEGs and targets of transcription factors for upregulated DEGs were screened out. Based on the above DEGs, protein-protein interaction (PPI) networks were constructed by the STRING software.
RESULTS: Significantly upregulated DEGs at T1 were mainly enriched in asthma and p53 signaling pathway. Meanwhile, significantly enriched transcriptional factor NOTCH1 at T1 and GABP at T0 were obtained. Moreover, miRNA analysis showed that targets of miR141/200a were enriched in downregulated DEGs both at T0 and T1. Mostly, RPA1 and JAK2 in PPI network at T1 were with higher degree.
CONCLUSIONS: In our study, obtained DEGs, regulatory transcriptional factors, and miRNA might play a vital role in the protection of ischemia reperfusion injury. This finding will provide a deeper understanding to the mechanism of IPC.

Background

Many surgical procedures involve prolonged ischemia of organs or tissues which can lead to sever postoperative complications, including dysfunction and necrosis [1]. In order to prevent ischemia-reperfusion injury, Murry et al. first documented the protective effect of ischemia preconditioning (IPC) in 1986 [2]. IPC has been proved as an extremely powerful method of protecting tissues against subsequent sustained ischemia insults [3] when it is firstly subjected to short bursts of ischemia and reperfusion. This method is universally applicable in modulating ischemia-reperfusion injury to tissues including the myocardium [4], brain [5], liver [6], lung [7], kidney [8], intestine [9], and skeletal muscle [10]. IPC is thought to provide protection by inducing tissues’ tolerance to ischemia, therefore reducing oxidative stress [11], inflammation, and apoptosis [12].

The complex mechanism of protection through IPC has partially been demonstrated. Studies have shown that release of signaling molecules such as adenosine, bradykinin, reactive oxygen species (ROS), catecholamines, and opioids can trigger protective response through various cell surface G-protein coupled receptors [13-16]. The released agonists then might activate protecting signaling pathway. Kinases such as protein kinase C [17], PI-3 K [18], tyrosine kinase [19], and MAPK kinase [20] play a vital role in the signaling pathway. IPC exerts a protective effect through upregulating heat shock proteins, reducing oxidative stress and inhibition of apoptosis in tissue injury [21].

In recent studies, microarray analysis for IPC effect was utilized to assess the gene expression after ischemia reperfusion. Murphy and colleagues screened differentially expressed genes (DEGs) > 1.5-fold and performed gene ontology analysis from GSE21164 [22]. Chunxiao Li also searched related motif and phosphorylation sites for significant DEGs [23] using same gene data. With the same limitation, they did not perform deep analysis for GSE21164. As a result, the mechanism of IPC protection in humans has not fully been elucidated.

For better understanding the effect of IPC, we carried out deep analysis in additional to DEGs screening and pathway analysis to explore molecular mechanism of protective effect of IPC. First of all, we predicted transcriptional factor in upregulated DEGs and evaluated microRNA (miRNA) targets in downregulated DEGs in our study. In addition, we constructed protein-protein interaction (PPI) network based on up- and downregulated DEGs at the onset of surgery (T = 0) and 1 h into surgery (T = 1).

Materials and methods

Microarray data

The gene expression profile of GSE21164 based on the platform GPL570 Affymetrix Human Genome U133 Plus 2.0 Array was obtained from National Center of Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/). A total of eight patients undergoing total knee arthroplasty were randomized to IPC treatment group (n = 4) and control group (n = 4). Patients in the IPC group received IPC stimulus which consisted of three 5-min periods of tourniquet insufflation on the lower operative limb and interrupted by 5-min periods of reperfusion prior to total knee arthroplasty. Comparisons for gene expression of muscle biopsies taken from the operative leg between the IPC group and the control both at the onset of surgery (T = 0) and 1 h into surgery (T = 1) were performed for future analysis.

Date preprocessing and DEGs screening

For each sample of GSE21164, the original expression datasets were converted into recognizable format and standardized using robust multi-array analysis (RMA) method [24]. Then, the RMA signal values were subjected to log2 transformation. To identify DEGs between the IPC group and the control, the empirical Bayes test (implemented in Linear Models for Microarray Data (Limma) package [25]) was applied in our analysis. P values were determined by Student’s t test. Log 2 |fold change (FC)| > 0.26 (|FC| > 1.2) and P < 0.05 were chosen as the cutoff criteria for screening DEGs.

Pathway enrichment analysis and prediction of transcriptional factors and regulatory miRNAs

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotation for DEGs was carried out using Database for Annotation, Visualization and Integrated Discovery (DAVID) [26] online tool. P < 0.05 was selected as the cutoff criterion. For prediction of the key transcriptional factors that regulated the expression of the upregulated DEGs, ChAE (ChIP Enrichment Analysis) [27] tool was utilized under the criteria of P < 0.005. Furthermore, Fisher’s exact test was applied to identify the significant miRNAs which targeted downregulated DEGs in TargetScan [28] database with P < 0.005.

Construction of PPI network

Building interaction network is beneficial for understanding of protein functions because most proteins play an important role in biological activity in the form of complexes not work alone. STRING software [29] was utilized to establish the interaction networks for significant up- and downregulated DEGs. Then, Cytoscape [30] was employed to explore and plot PPI network. Node degree is the number of interactions for a node, in other words, the number of other nodes which connected with directly. We calculated the node degree with significance in PPI network.

Results

DEGs identification and KEGG pathway enrichment analysis

Based on the cutoff criteria of |FC| > 1.2 and P < 0.05, we screened 372 up- and 227 downregulated DEGs from IPC group at the onset of surgery (T = 0). Besides, at 1 h into surgery (T = 1), 419 up- and 383 downregulated DEGs were obtained.

The KEGG pathway enrichment analysis was performed for both up- and downregulated DEGs at T0 and T1. Upregulated DEGs at T0 were mainly enriched in aminoacyl-tRNA biosynthesis. At T1, upregulated genes were mostly enriched in Soluble NSF Attachment Protein receptor (SNARE) interactions in vesicular transport, asthma, graft-versus-host disease, melanogenesis, and p53 signaling pathway, while downregulated genes at T1 were significantly enriched in chemokine signaling pathway (Table 1).

Table 1

Results of KEGG pathway enrichment analysis of up- and downregulated genes both at the onset of surgery (   = 0) and 1 h into the surgery (   = 1)

Term	P value	Genes
T = 0 upregulated DEGs
hsa00970: aminoacyl-tRNA biosynthesis	3.045E − 02	FARS2, CARS2, DARS2, SARS2
T = 1 upregulated DEGs
hsa04130: SNARE interactions in vesicular transport	4.998E − 03	SNAP29, STX5, BNIP1, VAMP5, GOSR2
hsa05310: asthma	1.579E − 02	HLA-DQB1, IL3, TNF, HLA-DQA1
hsa05332: graft-versus-host disease	3.461E − 02	HLA-DQB1, IL6, TNF, HLA-DQA1
hsa04916: melanogenesis	3.554E − 02	FZD9, WNT1, ADCY1, CREB3L4, CALML6, FZD5
hsa04115: p53 signaling pathway	3.631E − 02	CCNB1, BBC3, BAI1, RPRM, GADD45B
hsa05200: pathways in cancer	3.827E − 02	FZD9, WNT1, FOS, IL6, RXRB, JUN, STAT5B, ARNT2, FZD5, MYC, MMP1, CTNNA3
hsa00970: aminoacyl-tRNA biosynthesis	3.933E − 02	TARS2, FARS2, MARS2, EARS2
T = 1 downregulated DEGs
hsa04062: chemokine signaling pathway	1.919E − 02	PLCB4, GNAI1, TIAM1, CCR3, CCL19, GRK4, JAK2, CCL16, CXCL10
hsa04730: long-term depression	3.554E − 02	PLA2G4A, PLCB4, GNAI1, ITPR1, ITPR2
hsa04070: phosphatidylinositol signaling system	4.421E − 02	PLCB4, PLCG2, INPP4B, ITPR1, ITPR2
hsa00140: steroid hormone biosynthesis	4.976E − 02	AKR1C2, CYP7A1, HSD17B3, AKR1C1

Significant transcriptional factors for upregulated DEGs

Considering that most changes in gene expression are controlled by upstream regulatory transcriptional factors, we searched on ChEA for transcriptional factors that could regulate the 791 upregulated genes identified both at T0 and T1. A total of 12 transcriptional factors at T0 and 13 at T1 with statistical significance were found probably to modulate upregulated DEGs (Table 2). Furthermore, GA repeat-binding protein (GABP), vitamin D receptor (VDR), forkhead box P1 (FOXP1), and FLI1 were most significant at T0, while FLI1, forkhead box P3 (FOXP3), FOXP1, and NOTCH1 at T1. Ignoring the overlapping transcriptional factor, targets of GABP at T0 and NOTCH1 at T1 were significantly enriched in upregulated DEGs.

Table 2

Enriched targets of transcriptional factors in upregulated differentially expressed genes both at the onset of surgery (   = 0) and 1 h into the surgery (   = 1)

Transcriptional factor	P value
T = 0
GABP-19822575	3.5453E − 10
VDR-23849224	4.39418E − 10
FOXP1-21924763	5.58129E − 09
FLI1-21571218	6.78555E − 09
FOXP3-21729870	1.16903E − 08
ETS1-20019798	1.39977E − 08
XRN2-22483619	5.46758E − 08
E2F4-21247883	5.49972E − 06
DCP1A-22483619	4.81578E − 05
RUNX1-21571218	0.000157662
TTF2-22483619	0.000349486
TP63-22573176	0.000439082
T = 1
FLI1-21571218	6.78838E − 11
FOXP3-21729870	2.40847E − 10
FOXP1-21924763	5.84899E − 09
NOTCH1-21737748	1.56938E − 06
XRN2-22483619	5.20959E − 06
MITF-21258399	2.10127E − 05
VDR-23849224	2.1157E − 05
TTF2-22483619	3.85815E − 05
ATF3-23680149	4.62329E − 05
GATA1-21571218	7.06282E − 05
RUNX1-21571218	7.27917E − 05
HNF4A-19822575	0.000155321
DCP1A-22483619	0.000221014

Note: The number following gene symbol is the PubMed ID of the article which provided the Chip-seq data.

Significant DEGs-related miRNAs

The screening of significant DEGs-related miRNAs for downregulated DEGs was performed by Fisher’s exact test. As shown in Table 3, predicted targets of miR-141/200a family were obviously enriched in downregulated DEGs both at T0 and T1.

Table 3

Enriched potential regulatory microRNAs for targeting downregulated differentially expressed genes both at the onset of surgery (   = 0) and 1 h into the surgery (   = 1)

miR-141/200a targets	P value
T = 0
ASTN1, ATP8A1, CHKA, ENAH, GLS, LRRTM2, PDCD4, PEG3, RGAG1, SCN9A, SNAP91, STRN, STX2, THRB, TIAM1, TRHDE, WWTR1	0.0071
T = 1
AMPD2, APOLD1, CDC25B, ENAH, FAT3, FBXL2, GPC6, GPM6B, KALRN, LENG8, MAP4K4, MARK1, MYH10, PEG3, PLCB4, PTPRD, RBM47, SOX11, THRB, TIAM1, TTLL7, VASH2, VCAN, VGLL3, WWTR1	0.0076

We mapped DEGs to the STRING database and screened significant interactions with score larger than 0.4. By integrating these relationships, we constructed interaction networks among interactive proteins at T0 (Figure 1) and T1 (Figure 2). Several proteins including replication protein A1 (RPA1), MYC, Janus kinase 2 (JAK2), and cyclin B1 (CCNB1) were screened from PPI network based on their highest degree, shown in Table 4.

Figure 1

Protein-protein interaction network for significant differentially expressed genes at T0 (a-b). The network includes two small nets a and b. Orange circle means upregulated DEG, and blue circle means downregulated DEG. The line indicates one pair interaction between different proteins.

Figure 2

Protein-protein interaction network for differentially expressed genes at T1 (a-b). The network includes two small nets a and b. Orange circle means upregulated DEG, and blue circle means downregulated DEG. The line indicates one pair interaction between different proteins.

Table 4

Degree of differentially expressed genes in protein-protein interaction network both at the onset of surgery (   = 0) and 1 h into the surgery (   = 1)

Gene	Regulated	Degree
RPA1 (T = 1)	Up	8
MYC (T = 1)	Up	7
JAK2 (T = 1)	Down	7
CCNB1 (T = 1)	Up	6
JUN1 (T = 1)	Up	5
TP63 (T = 1)	Up	5
CHAF1A (T = 1)	Up	4
FOXM1 (T = 1)	Up	4
PRPF4 (T = 1)	Up	4
RUVBL2 (T = 0)	Up	4
FBXO25 (T = 0)	Up	4
STX8 (T = 1)	Up	4

Discussion

IPC is a universal method to reduce ischemia reperfusion injury in several tissues and organs [31]. In this study, we screened DEGs from biopsies of four control and four IPC-treated patients who were subjected to total knee arthroplasty surgery in order to gain insight into the molecular mechanism of IPC in protection against ischemia reperfusion injury. A number of DEGs were identified between IPC group and the control at T0 and T1.

Moreover, KEGG pathway enrichment analysis showed that upregulated DEGs were significantly enriched in aminoacyl-tRNA biosynthesis both at T0 and T1, SNARE interactions in vesicular transport and p53 signal pathway at T1. As shown in previous studies, the expression of ARS (aminoacyl-tRNA synthetases) coding genes and genes involved in metabolism and transport of amino acids were upregulated after ischemia but decreased after IPC treatment [32]. Previous report showed that p53 was a caspase inhibitor, which has been known to stimulate the disruption of mitochondria and widely used in studies of apoptosis [33]. P53 expression has also been shown to inhibit apoptosis induced by tumor necrosis factor-α (TNF-α) in the liver [34,35]. Furthermore, reperfusion-induced hepatic apoptosis could be decreased by IPC through lowering TNF-α levels and modulating the caspase dependent pathway [36]. Notably, emerging evidences have pointed out that Golgi-SNARE GS28 (Golgi SNAP receptor complex member 1) forms a complex with p53, and thus affect the stability and activity of p53 [37]. In addition, GS28 may enhance cells to DNA-damage-induced apoptosis through inhibiting the ubiquitination and degradation of p53 [37]. Taken together, IPC might render protection against reperfusion-induced injury at cellular, organ, and systemic level partly through aminoacyl-tRNA biosynthesis, SNARE interactions in vesicular transport and p53 signaling pathways.

To identify transcriptional factor upon Chip-seq gene profile, we performed ChEA analysis in upregulated DEGs both at T0 and T1. Results suggested that NOTCH1 remarkably regulated overexpressed DEGs at T1 but not at T0. Recent studies have shown that NOTCH1 assumed a fundamental role in the mechanisms of cerebral ischemia injury [38] through eliciting protective effects against ischemia injury by decreasing neuronal apoptosis in mice [39]. IPC-induced NOTCH1 signaling could activate the endogenous neuroprotective components and decrease the ischemic-reperfusion injury at the early phase after stroke [40]. In addition, NOTCH1 signaling protected against ischemia-reperfusion injury partly though PTEN/Akt-mediated anti-nitrative and anti-oxidative effects [41]. Remarkable miR-141/200a was member of miR-200 family which was originally associated with the inhibition of cancer invasion [42] or olfactory neurogenesis [43]. Lee's group indicated that miR-141/200a were upregulated early after IPC and they were neuroprotective mainly by improving neural cell survival via proly1 hydroxylase 2 (PHD2) silencing and subsequent HIF-1α (hypoxia-inducible factors-1α) stabilization [44,45]. To sum up, transcriptional factor NOTCH1 and miR-141/200a might regulate effects of IPC.

To well understand functions of apparent DEGs, we mapped them to STRING software and obtained PPI networks at T0 and T1. It is well known that DNA damage and DNA response related proteins were revealed to play crucial role during ischemia reperfusion injury in brain and heart. In network at T1, the dual-specificity tyrosine-(Y)-phosphorylation regulated kinase 2 (DYRK2), interacted with RPA1, regulated p53 to induce apoptosis in response to DNA damage [46]. As a result, we suggested that RPA1 which had the highest degree might play a role in IPC protection via interacting with DYRK2. Meanwhile, JAK2 known as a member of Janus kinase signal transducers and activators of transcription (JAK-STAT) pathway which protected against ischemia-reperfusion injury by decreased number of apoptotic cardiomyocytes, improved functional recovery, and reduced infarct size in the early phase of IPC [47]. When isolated working rat hearts were subjected to ischemia, tyrosine phosphorylation of JAK2 and STAT3 immediately increased after IPC stimulus as well as 2 h after reperfusion [48]. Specially, ischemia-reperfusion could activate JAK2 and recruit STAT3, resulting in transcriptional upregulation of inducible cyclooxygenase-2 (COX-2) and nitric oxide synthase (iNOS), which then mediated the infarct-sparing effects of the late phase of preconditioning [49]. However, in our research, JAK2 has been shown to be downregulated in the PPI network which was inconsistent with previous demonstration. Thus, JAK2 might also participate in other pathway to regulate IPC effect when tissue was subjected to ischemia-reperfusion. Further studies will be necessary to elucidate the mechanisms for JAK2 signaling in IPC treatment. In our study, MYC, CCNB1, and JUN1 with higher degree in PPI network were newly reported associated with IPC protection.

Conclusion

Our study shed new light on the mechanism of IPC effect and treatment to diseases due to ischemia reperfusion. Based on the screened significant DEGs, p53 signaling pathway, transcriptional factor NOTCH1, and miR141/200a might regulate IPC protective effect to ischemia reperfusion injury. Importantly, significant DEGs RPA1 and JAK2 in PPI network might participate in the protection of IPC. Although these evidences were obtained, further research will be indispensable to explore the mechanisms of previous mentioned genes. On the other hand, experimental verifications in further research were needed to support our results.