Portal vein ligation alters coding and non-coding gene expression in rat livers

Portal vein occlusion increases the resectability of initially unresectable liver cancer by inducing hypertrophy in non-occluded liver lobes. However, the mechanisms of how portal vein occlusion induces hepatic hypertrophy remain unclear. A cDNA microarray was used to identify the gene expression signatures of ligated (LLLs) and non-ligated lobes (NLLLs) at different time points after portal vein ligation (PVL). The results of a bioinformatics analysis revealed that LLLs and NLLLs displayed different gene expression profiles. Moreover, the expression levels of both coding and non-coding RNA were different between LLLs and NLLLs at different time points after PVL. A Series Test of Cluster analysis revealed that the No. 22 and No. 5 expression patterns, which showed altered expression at 24 h and maintained this altered expression over the following 14 d, had the lowest P value and the largest number of differentially expressed genes in both LLLs and NLLLs. The results of a GO analysis showed the activation of hypoxia pathways in LLLs and the activation of cell proliferation and cell cycle pathways in NLLLs, suggesting the involvement of these pathways in PVL-induced hepatic hypertrophy and regeneration. These results provide insight into the molecular mechanisms underlying hepatic hypertrophy and regeneration induced by portal vein occlusion, and they identify potential targeting pathways that can promote the clinical application of PVL in liver cancer therapy. displayed unique gene (coding and non-coding) expression profiles assessed by cluster analysis based on cDNA microarray data. In the first 24 h after PVL, the expression of most genes was altered. The results of a subsequent GO analysis of these differentially expressed genes showed enrichment of hypoxia, cell proliferation and apoptosis pathways. The present study identified genes involved in PVL and proposes the pathways through which PVL induces hepatic hypertrophy and regeneration. These results may provide further insight into the hepatic hypertrophy and regeneration induced by PVL at the molecular level and potential targeting pathways to promote the clinical application of PVL in liver cancer therapy. further analysis showed that both coding and non-coding genes were differentially expressed in LLLs and NLLLs after PVL and that the expression levels changed over time. In addition, gene expression patterns assessed by STC showed that the largest number of genes was altered during the first 24 h at all three time points and that these expression levels were maintained for the following two weeks. The GO analysis showed that pathways involved in hypoxia, cell cycle and cell proliferation were activated in the liver lobes after PVL. Our findings provide a systemic view of the molecular alterations in liver lobes after PVL and provide insight into the mechanism underlying PVL-induced atrophy and hypertrophy.


Introduction
Complete hepatic tumor resection is the main option for curative treatment of liver malignancies and provides patients with a chance for long-term survival (Utsunomiya et al. 2014). Up to 70% of patients are unsuitable for resection because of insufficient remnant liver volume, which always leads to liver failure and increased postoperative morbidity after major liver resection (She and Chok 2015). Portal vein occlusion via embolization (PVE) or ligation (PVL) induces hypertrophy in non-occluded liver lobes and increases the resectability of an inadequate functional remnant liver volume (Siriwardana et al. 2012). It has been reported that PVE is associated with a minimal mortality rate in patients receiving extended hepatectomy (Shindoh et al. 2014). However, the mechanisms of how PVE/PVL induces contralateral hepatic hypertrophy are still poorly understood.
In patients, PVE has been reported to induce apoptosis in the embolized lobe and proliferation in the non-embolized lobe, which may involve the altered expression of transforming growth factor-alpha (TGF-α) and transforming growth factor-beta (TGF-β) in livers after PVE (Kusaka et al. 2006). Changes in portal flow and cytokine expression after PVE/PVL were also reported in animal models. A reduction in left liver flow and an increase in right liver flow were observed when left portal vein stenosis was performed-results that were further verified by the dilation of the portal branches in the non-embolized lobe (Kawai et al. 2002;Rocheleau et al. 1999). Cytokines and transforming growth D r a f t 4 factors, such as TGF-α and IL6, in both serum and remnant livers were all elevated after PVE/PVL (Garcia-Perez et al. 2015;Kusaka et al. 2004).
Hepatic regeneration was promoted by perioperative treatment with HGF, suggesting that the altered genes were involved in liver atrophy and hypertrophy after PVE/PVL (Mangieri et al. 2016). Moreover, the expression of activin beta family members (A, C and E) was elevated in the early stage of PVL, indicating the involvement of activin signaling in liver regeneration after PVL (Tashiro 2009). Another study reported that mitochondrial DNA replication and transcription were both enhanced in the non-ligated lobe, which increased the energy supply for liver regeneration (Shimizu et al. 1995). These findings suggest that a complex signal transduction network is activated in livers after PVE/PVL to induce contralateral hepatic hypertrophy and to prevent post-hepatectomy liver failure (Yokoyama et al. 2007). Although many factors have been reported to be involved in PVE/PVL-induced hepatic hypertrophy and liver regeneration, a systematic investigation of the gene expression profiles in livers after PVE/PVL using high-throughput platforms, such as cDNA microarrays, is lacking.
In the present study, we established a rat model of PVL and identified the gene expression signatures of ligated and non-ligated lobes at different times after PVL using cDNA microarrays. We found that PVL induced atrophy in ligated liver lobes (LLLs) and hypertrophy in non-ligated liver lobes (NLLLs) in rats. At different times after PVL (24 h, 7 d and 14 d), both LLLs and NLLLs D r a f t displayed unique gene (coding and non-coding) expression profiles assessed by cluster analysis based on cDNA microarray data. In the first 24 h after PVL, the expression of most genes was altered. The results of a subsequent GO analysis of these differentially expressed genes showed enrichment of hypoxia, cell proliferation and apoptosis pathways. The present study identified genes involved in PVL and proposes the pathways through which PVL induces hepatic hypertrophy and regeneration. These results may provide further insight into the hepatic hypertrophy and regeneration induced by PVL at the molecular level and potential targeting pathways to promote the clinical application of PVL in liver cancer therapy.

Animals
Male Sprague-Dawley rats, 6-8 weeks old and 140~220 g in weight, were obtained from the Experimental Animal Center of the Second Military Medical University. All rats were maintained in a pathogen-free facility with alternate light and dark conditions and were given standard laboratory rodent chow. All animal experiments (SCXK(HU)2013-0016) were performed in accordance with the institutional guidelines for animal care.

PVL model
The PVL model was established as previously reported (Dhar et al. 2015).
Briefly, abdominal anesthesia was administered using 10% chloral hydrate at a 0.035 ml/kg body weight dose. The abdomen was opened via a midline D r a f t 6 incision, and blood vessels including the portal branches and arteries were identified with the help of an operating microscope. The portal vein of the left lobe, the middle lobe and the hepatic papillary was ligated with a double surgical suture to block the blood supply. The PVL model was complete when the color of the right lobe remained light brown, and the other lobes turned dark brown. At 24 h, 3 d, 7 d and 14 d after PVL, the animals were sacrificed by cervical dislocation. Whole livers were dissected, the total liver weight was measured, and then the individual lobes were weighed separately. Liver tissues were kept in liquid nitrogen for future use.

Microarray analysis
LLLs and NLLLs were obtained from rats at different time points (24 h, 7 d and 14 d) after PVL. Normal liver tissues were used as the controls. RNA was isolated from these tissues using TRIzol® Reagent (Invitrogen Life Technologies, Carlsbad, CA) and sent to KANGCHEN Corporation for microarray analysis using the Rat 4 x 44K LncRNA expression array. Data were extracted using Agilent Feature Extraction software. Normalization and further data analysis were performed using Agilent GeneSpring GX v11.5.1 software.

Identification of differentially expressed genes
The random-variance model (RVM) F-test was applied to identify differentially expressed genes between the normal tissues and LLLs or NLLLs from different time points (24 h, 7 d and 14 d). After the significance analysis D r a f t and FDR analysis, we selected the differentially expressed genes according to their P value threshold (P<0.05) (Clarke et al. 2008).

Series test of cluster
A Series Test of Cluster (STC) was used to analyze the dynamic expression of genes in LLLs and NLLLs at different time points, as previously reported (Su et al. 2013).

Gene ontology
To analyze the differentially expressed genes at the functional level, GO enrichment analyses were performed using the DAVID online tool. P<0.05 was considered statistically significant (Liang et al. 2016).

Statistical analyses
Statistical analyses were performed with SPSS 13.0 for Windows (SPSS).
Quantitative variables were analyzed using a t-test. P<0.05 was considered statistically significant.

PVL induced atrophy in LLLs and hypertrophy in NLLLs in rats
The rat PVL model was established by ligating the portal vein of the left lobe, the middle lobe and hepatic papillary. All the rats survived the procedure and recovered during the postoperative period. At different time points (24 h, 3 d, 7 d and 24 d) after the ligation of the portal vein, liver tissues from sacrificed rats (n=6) were weighed. The results showed that the highest whole liver weight (WLW)/body weight (BW) ratio occurred at 7 d after PVL; no significant D r a f t 8 difference was found at 24 h or 3 d ( Figure 1A). The NLLL weight (NLLLW)/BW ratio increased along with the time after PVL, whereas the LLL weight (LLLW)/BW ratio decreased ( Figure 1B). Similar results were obtained when the NLLLW/WLW and LLLW/WLW ratios were analyzed ( Figure 1C). These results indicated that PVL induced hypertrophy in NLLLs and atrophy in LLLs in rats.

PVL induced gene expression changes in rat livers
To investigate gene expression in NLLLs and LLLs at different time points after PVL (0 h, 24 h, 7 d and 14 d), tissues were obtained and sent for cDNA microarray analysis. Multiple genes were found to be differentially expressed in LLLs compared with the corresponding NLLLs at all three time points ( Figure   2A (Table 1). The expression of genes contained in these two profiles was altered at 24 h, and these expression levels were maintained for the following two weeks. For long non-coding RNAs, 9 profiles in LLLs and 10 profiles in NLLLs were significantly changed ( Figure 4C and D). Similar to the mRNA results, 9 profiles were found in both groups, and the two most significant patterns (No. 22 and No. 5) included the largest number of differentially expressed genes (Table 1). In addition, the expression patterns of long non-coding RNAs included in No. 22 and No. 5 were similar to the mRNA data. These results indicated that liver tissues expressed different genes after PVL and that the largest number of genes was altered during the first 24 h after PVL.

Hypoxia, cell proliferation and apoptosis pathways enriched by GO
D r a f t analysis As different gene expression patterns were identified after PVL, we intended to investigate the biological pathways involved in this process. The genes in profiles No. 22 and No. 5 from the STC analysis, which were the most changed profiles, were uploaded to the DAVID online software to identify overrepresented GO categories. The results showed that genes in profile No.
22 from LLLs were significantly enriched in gene transcription, regulation and response to hypoxia for biological processes (BP); DNA and protein binding for molecular function (MF); and the cytoplasm and nucleus for cell components (CC) ( Table 2). The genes in profile No. 5 from the LLLs were significantly enriched in BP, including cell cycle and regulation of apoptotic process as well as those presented in profile No. 22 (Table 2). For MF and CC, the enriched categories were same as those found in profile No. 22 (Table 2)  Collectively, these data suggested that PVL induced the hypoxia response and inhibited gene transcription and protein catabolic processes in LLLs and activated DNA replication and cell division in NLLLs.

Discussion
In the present study, we investigated gene expression in rat liver lobes after PVL using cDNA microarrays. Unique coding and non-coding gene expression profiles were found in both LLLs and NLLLs. Moreover, we found

PVL induced atrophy in ligated liver lobes and hypertrophy in non-ligated liver lobes
A) The whole liver weight (WLW)/ body weight (BW) ratio at different times (n=6) after PVL.
C) The LLLW/WLW and NLLLW/WLW ratios at different times (n=6) after PVL. **P<0.01   D r a f t 21 C) The 9 expression profiles of non-coding genes that were statistically significant in the LLLs over time (0 h, 24 h, 7 d and 14 d).
D) The 10 expression profiles of non-coding genes that were statistically significant in the NLLLs over time (0 h, 24 h, 7 d and 14 d).