Expression profiles uncover the correlation of OPN signaling pathways with rat liver regeneration at cellular level

Introduction

The secreted glycol-phosphoprotein 1 (OPN), as a member of Th1 cytokines, has been revealed to participate in the occurrence and development of multiple liver diseases. In obesity-induced hepatic steatosis, OPN could activate natural killer T (NKT) cells and neutrophils by interacting with its receptors, triggered the infiltration of neutrophils, and then aggravated liver damage (Kiefer et al., 2011). In the progress of hepatic fibrosis, OPN was able to affect hepatic stellate cells (HSCs) activation, and then accelerate the expression of collagen, matrix metalloproteinase (MMP), and cytokines (Honsawek et al., 2010; Sabo-Attwood et al., 2011; Urtasun et al., 2012). Moreover, the over-expression of OPN was closely related to metastasis and prognosis of hepatocellular carcinoma (HCC), and OPN depletion by siRNA could inhibit formed colony numbers and antagonize metastasis potential for HCC cell lines (Lin et al., 2011). Two studies from one Japanese lab together revealed that OPN protein was presented in the regenerating hepatocytes and bile ductular structures in the livers of patients with fulminant hepatic failure, and then surmised that OPN could activate hepatic stem cells during liver regeneration (LR) (Arai et al., 2004, 2006). Taken together, OPN has been shown to regulate inflammation, cell activation, proliferation, and migration through targeting several hepatic cell types.

It is well known that liver is composed of at least eight hepatic cell types, including hepatic parenchymal cells (hepatocytes, HCs, and biliary epithelial cells, BECs) and non-parenchymal cells such as HSCs, Kupffer cells (KCs), sinusoidal endothelial cells (SECs), etc. One of our previous studies confirmed that the mRNA level of OPN was elevated at 2–72 h in isolated HCs, KCs, pit cells (PCs), and dendritic cells (DCs) from regenerated livers after partial hepatectomy (PH), and its highest expression even exceeded control levels by 31-fold, indicating that OPN might be associated with the activation and proliferation of various hepatic cells during LR (Wang and Xu, 2010a). On the other hand, it is generally acknowledged that OPN could perform its functions through PI3K/Akt, MAPK, and NF-κB signaling pathways upon binding to a number of integrins receptors (Sodek et al., 2000; Denhardt et al., 2001; Chakraborty et al., 2006; Urtasun et al., 2012), or through PI3K/Akt (Chakraborty et al., 2006; Wang and Denhardt, 2008) and JAK/STAT (Behera et al., 2010) signaling pathways when interacting with CD44. Figure 1 depicts the main downstream signal pathways activated by OPN. Therefore, the influences of OPN and its related signaling pathways on kinds of hepatic cells in LR need to be further defined. In this study, OPN-mediated signaling pathway was first summarized into five branches according to several common databases and related literatures. Then, based on the OPN signaling network mentioned above, this study used Rat Genome 230 2.0 chip to assess the expression changes of OPN signaling pathway-related genes in eight kinds of hepatic cells during LR and applied Ingenuity Pathway Analysis (IPA) software to clarify the possible regulatory effects of OPN on eight hepatic cell types at transcriptional level.

Materials and methods

Results

Identification and microarray detection of eight hepatic cell types during LR

The procedures for isolation, identification, and microarray detection of eight hepatic cell types were indicated as a flow chart in Figure 2. Briefly, four rats in every group (totally 19 groups) were anesthetized at corresponding recovery time points after PH, respectively, and the scattering of mixed liver cell suspension was prepared according to the conventional two-step method with 0.025% collagenase. Then, hepatocytes were isolated by gradient centrifugation with 60% Percoll, and sorting of other seven cell types except hepatocytes required magnetic bead method in addition to Percoll gradient centrifugation. For each type of liver cell at one time point, a few purified cells were identified by immunocytochemical and Western blotting analysis, and other cells were used to detect gene expression profiles by microarray with each sample repeated three times. Totally, the required chips in this study were 19 (groups) × 8 (cell types) × 3 chips. Immunocytochemical analysis elucidated that the eight hepatic cell types were highly purified as described in our previous studies (Chen et al., 2010a; Wang and Xu, 2010b). Furthermore, above eight kinds of isolated hepatic cells were also identified by Western blotting using the corresponding protein markers. As displayed in Figure 3, except LYZ and ED2 in KCs, other marker proteins were clearly detected by their corresponding antibodies. Obviously, ALB and CD103 were strongly positive in HCs and DCs, respectively, with thick bands of proteins. On the whole, the expressions of marker proteins in every cell type did not exhibit significant changes at 10 time points after rat PH, suggesting that the above eight hepatic cell types were prepared with high purity.

Validation of chip results by qRT-PCR

qRT-PCR method was applied to assay the expression changes of several marker genes including g6pc for HCs, ggt1 for BECs, oc2 for OCs, vim for HSCs, cd14 for SECs, cd68 for KCs, cd161a for PCs, and cd40 for DCs in order to confirm the reliability of microarray. Moreover, several common functional genes including jun, ttr, icam1, apoe, myc, pcna, trim24, and b2m were also selected to detect their expression changes in the above eight kinds of hepatic cell types, respectively. It meant that one marker gene and one functional gene were designed for a corresponding hepatic cell type. Furthermore, among the eight selected functional genes, jun and myc were downstream transcription factors in OPN-mediated signaling pathway. Figure 4 illustrated that there were no remarkable differences in the expression changes of target genes including jun in HC and myc in SEC by qRT-PCR and gene chip detection.

Expression profiles of OPN signaling pathway-associated genes in eight hepatic cell types during LR

In this study, the module of “Pathway and function” in IPA software and several database including GenMAPP, KEGG, BIOCARTA, QIAGEN, and Biocompare, etc. were mainly relied on to build the interaction network of OPN signaling pathway. The results showed that five branches including 184 genes were involved in OPN-mediated signaling pathway. After detected by microarray, a total of 123 genes were found to be associated with LR. Among them, 93 genes were significantly changed in PCs, 98 in SECs, 96 in HSCs, 75 in KCs, 101 in DCs, 111 in BECs, 63 in OCs, and 86 in HCs (Supplementary Table 1). Furthermore, the numbers of significantly changed OPN signaling pathway-related genes at 10 time points were counted in eight hepatic cell types respectively, and the result revealed that the expression peaks of OPN signaling pathway-related genes occurred at different time points in various cell types and that of PCs and SECs were at 6 h, HSCs and KCs at 12 h, DCs and BECs at 24 h, OCs at 30 h, and HCs at 72 h (Table 2), indicating various cell types may play distinct roles during liver regeneration. To comprehensively compare the expression differences of OPN pathway-related genes in eight hepatic cell types, we employed H-means clustering to classify the above 123 genes, and the results indicated that the number of up-regulated genes in HCs, PCs, OCs, and BECs were more than that of the down-regulated, and the above four cell types were clustered in one major branch of the tree diagram. However, the majority of genes were down-regulated in other four cell types and, thereby, KCs and SECs, DCs, and HSCs were clustered together, respectively (Figure 5). Therefore, HCs, PCs, OCs and BECs, KCs and SECs, DCs, and HSCs shared the most similar patterns of genes, respectively, which may indicate the close correlations and great differences in functions among various hepatic cell types during rat LR.

Table 2. Overview of the numbers of OPN signaling pathway-involved genes at 10 recovery time points in eight hepatic cell types

	Recovery time points (h) after partial hepatectomy
Cell types	2	6	12	24	30	36	72	120	168h	Total
PCs	28	50	25	33	38	38	49	37	27	93
SECs	28	49	28	28	31	32	24	31	30	98
HSCs	57	16	58	26	18	43	22	26	25	96
KCs	19	28	37	34	31	30	20	19	26	75
DCs	24	43	41	64	37	53	22	22	23	101
BECs	48	63	74	74	45	21	20	50	26	111
OCs	19	22	23	20	25	23	17	24	23	63
HCs	27	35	39	33	37	32	57	36	24	86

• PCs, pit cells; SECs, sinusoidal endothelial cells; HSCs, hepatic stellate cells; KCs, Kupffer cells; DCs, dendritic cells; BECs, biliary epithelia cells; OCs, oval cells; HCs, hepatocytes. Dark gray fields represent the expression peaks in every cell type with the most number of significantly-changed genes.

Function enrichment analysis of OPN signaling pathway-associated genes in eight hepatic cell types during LR

The probe numbers of OPN signaling pathway-related genes and their corresponding expression extrema in eight liver cell types were entered into IPA software for core analysis and comparison analysis among individual liver cell types, and then H-means were used for clustering the bifunctions with z-score value > 2. The results suggested that apoptosis, cell death, and necrosis were extensively attenuated in eight liver cell types, especially in KCs and SECs; cell proliferation, activation, migration and movement, inflammation response, and chemotaxis were remarkably augmented in HCs, PCs, OCs, and BECs but significantly inhibited in HSC; skeleton reconstruction, cell spreading, and RNA transcription were promoted to various degrees in most of cell types. Based on the similarities of GO categories presented by the dendrogram, HCs, PCs, OCs, and BECs were clustered together, while KCs and SECs, DCs, and HSCs were in one cluster, respectively, because the majority of physiological categories were inhibited (Figure 6). Notably, the dendrograms presented by Figures 5 and 6 resembled each other, indicating that the differences in OPN signaling pathway, regardless of reflected by gene expression profiles or physiological activities, were consistent among eight kinds of hepatic cells.

Regulatory roles of each branch of OPN signaling pathway in eight hepatic cell types

To obtain preliminary insights into the potential mechanisms of OPN signaling pathway in LR, LR-related genes in each branch of OPN signaling pathway were respectively collected (Table 3), and their expression extreme values were entered into IPA software for “Core analysis” and “Comparison analysis” and, thereby, the functional heatmaps of each branch were predicted and obtained in eight hepatic cell types (Figure 7). Obviously, JAK/STAT signaling branch mainly promoted survival and viability of HCs and KCs but inhibited inflammation and cell migration in HSCs and apoptosis and death in most cell types. ERK1/2 branch principally participated in boosting proliferation of most cell types except HSCs, survival of most cell types except KCs, HSCs and DCs and inflammation in HCs and PCs but inhibited apoptosis of most cell types including PCs. Overall, the regulatory roles of JNK branch were similar to that of ERK1/2, though JNK branch was weaker than ERK1/2 in promoting cell proliferation and inhibiting apoptosis. Remarkably, NF-κB branch inhibited apoptosis and death in most cell types and almost all physiological processes in HSCs but reinforced proliferation, survival, activation, and inflammation and its related cell movement in HCs, PCs, OCs, and BECs. RhoA-ROCK branch mainly took part in skeleton reconstruction, thereby strengthening movement and migration of most cell types, while strongly inhibiting the physiological processes in HSCs.

Table 3. Overview of five branches of OPN signaling pathway-involved genes in eight hepatic cell types. The abbreviations represent the same hepatic cell types as displayed in Table 2

	Cell types
OPN-mediated pathways	HCs	BECs	OCs	HSCs	SECs	KCs	PCs	DCs
JAK/STAT pathway	11	10	9	11	8	11	5	12
ERK1/2 pathway	32	46	21	37	42	27	37	42
JNK pathway	28	39	15	32	34	22	32	33
NF-kB pathway	32	41	23	33	39	31	37	41
RhoA-ROCK pathway	50	62	39	55	55	44	53	55
Total	86	111	63	96	98	75	93	101

Discussion

OPN, as a highly phosphorylated chemokine-like protein, was closely associated with the occurrence of various liver diseases via promoting inflammation, cell activation, proliferation, and migration. Moreover, OPN was even found to induce the activation of hepatic stem cells during liver regeneration (LR) (Arai et al., 2004, 2006). LR is considered as a compensatory hyperplasia process, including cell proliferation and activation, and is usually accompanied by inflammation and immunity (Markiewski et al., 2006). After all, the liver is composed of various cell types including HCs, BECs, HSCs, SECs, KCs, PCs, and DCs, except for hepatic stem cells. On the other hand, OPN could function via PI3K/Akt, MAPK, and NF-κB signaling cascades. Therefore, whether OPN or OPN-mediated signaling pathway could modulate other cell types during LR is still unknown.

To uncover the possible regulatory effects of OPN signaling pathway on various hepatic cell types during LR, this study isolated and purified eight hepatic cell types from regenerating livers, detected the expression profiles of OPN-mediated signaling pathway-related genes in above-mentioned liver cell types, and applied IPA software to predict the enriched physiological activities. As a result, the physiological processes possibly regulated by OPN signaling pathway were significantly diverse in different cell types, and OPN signaling pathway may be closely related to HCs, OCs, BECs, and PCs but has less regulatory effect on SECs, KCs, HSCs, and DCs during rat LR.

Several researches have confirmed that OPN could inhibit apoptosis and promote survival of NKT cells, neutrophiles, macrophagocytes, and hepatoma carcinoma cells (Wang and Denhardt, 2008; Kiefer et al., 2010; Wu et al., 2012). In this study, OPN was up-regulated in almost cell types with greater than 10-fold overexpression in HCs, KCs, PCs, and DCs, and the expressions of anti-apoptosis genes Bcl2 and Bcl2l1 were also reinforced in most cell types, which is partially consistent with the report of Patouraux et al. (2014). Figures 6 and 7 illustrated that apoptosis, necrosis, and cell death regulated by OPN were significantly attenuated, suggesting that OPN might inhibit apoptosis in rat regenerative livers. Moreover, several physiological processes including cell proliferation, activation and survival, cytoskeleton reorganization, cell movement and migration, inflammation reaction, and chemotaxis were to different degrees activated in HCs, OCs, BECs, and PCs (Figure 6). HCs and BECs are hepatic parenchymal cells, whereas OCs, as one kind of hepatic stem cells, could be induced to proliferate rapidly and differentiate into BECs and HCs when liver was severely damaged or liver cell proliferation ability was restricted (Tanimizu and Mitaka, 2014). The transforming relationships mentioned above could partly explain the similar activities of OPN signaling pathway in HCs, OCs, and BECs. PCs, as liver-specific NKT cells, were essential for hepatic defense response (Xu et al., 2011c). OPN signaling pathway could promote secretion of various cytokines, and then modulate inflammation and immune reaction (Wang and Denhardt, 2008). Furthermore, there were several reports demonstrating that OPN could bind on the surface of immune cells and modulate chemotaxis of these cells (Koh et al., 2007; Vetrone et al., 2009). Therefore, OPN signaling pathway may be very important for PCs. In this study, the majority of OPN signaling pathway-related genes were found to be up-regulated and, thereby, cell activation, migration, invasion and chemotaxis were predicted to be strongly enhanced in PCs, suggesting OPN may have great effects on PCs during LR.

The expression peak of OPN signaling pathway-related genes appeared at 6 h in PCs, but it exhibited at 72 h in HCs (Table 2), which may be caused by the different dominant functions of OPN specifically in the above two cell types. The most important role of OPN signaling pathway in PCs was to promote inflammation response (Locatelli et al., 2013), cytoskeleton reorganization, and spreading, which was similar to the report of Xu et al. (2011c). However, its role was to inhibit apoptosis and promote cell proliferation in HCs (Figure 6), which agreed with the viewpoint of Patouraux et al. (2014). JAK/STAT signaling cascade was necessary for cell survival (Wu et al., 2014). However, the effects of JAK/STAT branch on HCs and PCs were significantly different during liver regeneration with enhanced cell survival and activation in HCs and little regulatory role for PCs (Figure 7), which was consistent with the result of Chen et al. (2010b). Surprisingly, the mitosis in BEC was evidently enhanced, but the proliferation of BECs was not activated as the former, which deserved further experimental studies.

SECs were similar to KCs in terms of expression profiles of OPN pathway-related genes (Figure 5). In addition, the OPN-modulated physiological activities including cell proliferation and survival, cell spreading, invasion, and migration were slightly increased in both of the above two cell types (Figures 6 and 7). The obvious difference between SECs and KCs was that the activation of ERK1/2 and JNK branches in SECs was stronger than that in KCs and, thereby, ERK1/2 and JNK branches may play critical roles in promoting the survival and proliferation of SECs (Figure 7). Hang et al. (2012) found ERK signaling pathway could facilitate the survival and proliferation of SECs, which is partly in accordance with our result. On the other hand, cell survival regulated by JAK/STAT branch was more activated in KCs than that in SECs during LR, whereas Chen et al. (2010b) showed that IFN-γ-induced JAK/STAT pathway genes were down-regulated particularly in KCs, and then might lead to the attenuation of apoptosis induction in KCs during rat LR. The above results inferred that different ligands (OPN and IFN-gamma) could activate JAK/STAT signaling pathway but have distinct effects on apoptosis in KCs during LR. Moreover, cell invasion in KCs, modulated by ERK1/2, NF-κB, and Rock-RhoA branches, was stronger than that of SECs. This result may to some extent confirm the fact that OPN could activate KCs and then promote inflammation (Yang et al., 2011). Furthermore, it seemingly supported the conclusion that the activation of KCs and its related inflammation were mainly regulated by NF-KB, ERK, and JNK branches (Wu et al., 2012a).

Furthermore, gene expression profiles of HSCs and DCs resembled each other, accompanied by more down-regulated genes than the up-regulated, but the inhibition of physiological processes in HSCs was stronger than that in DCs. Inflammation, cell chemotaxis, invasion, and migration in HSCs were significantly inhibited through JAK/STAT, NF-κB, and Rock-RhoA branches. However, Seth et al. (2014) demonstrated that OPN could mediate the activation of HSCs in alcoholic liver diseases. Thus, it is inferred that OPN signaling cascade might inhibit inflammation in HSCs during LR, which was not contrary to our previous report that HSCs had suppressed immunity/inflammation after PH in rats (Xu et al., 2011b). With respect to the effects of OPN on DCs, Shao et al. (2014) revealed that OPN had a pro-chemotactic sequence for DCs and could promote the migration of DCs. On the contrary, Umemura et al. found that OPN suppressed the DC migration to lymph nodes in immune response (Begum et al., 2007). One of our previous studies also indicated that the attenuated ability of producing chemokines and proinflammatory factors for DCs may lead to its weak inflammatory activities during LR (Xu et al., 2011a). In this study, inflammation response, chemotaxis, and migration in DCs were significantly inhibited through JAK/STAT, ERK1/2, and JNK branches, suggesting OPN signaling pathway might be responsible for the reduced inflammation and cell migration in DCs during LR.

Conclusions

In summary, to our knowledge it is the first time we discussed the possible relationship between OPN signaling pathway and rat liver restoration at cellular level. In our study, high-throughput analysis was utilized to measure and compare the expression pattern of OPN signaling pathway genes among eight hepatic cell types. The study findings demonstrated that the majority of OPN pathway-related genes were up-regulated in HCs, PCs, OCs, and BECs but down-regulated in other four cell types including SECs, KCs, DCs, and HSCs. In addition, functional enriched analysis indicated that OPN signaling pathway may promote cell proliferation, activation, migration, and inflammation in HCs, OCs, BECs, and PCs, and slightly boost proliferation and migration of SECs and KCs, but inhibit inflammation response and chemotaxis in SECs and KCs and almost all physiological processes in DCs and HSCs. It is supposed that our study may provide some data for modulating liver regeneration through OPN-mediated signaling pathway at cellular level. However, the conclusion is drawn mainly based on the microarray data and bioinformatics analysis, and there is no direct cause–effect relationship between gene profiling and the conclusion. Therefore, in order to prove that those changes in eight hepatic cell types are truly dependent upon OPN, experiments of OPN overexpression or knockdown and one or more cell types depletion are needed to further analyze the regulatory effects of OPN signaling pathway on regenerating livers in future.