CXCL12 expression in intrahepatic cholangiocarcinoma is associated with metastasis and poor prognosis.

Intrahepatic cholangiocarcinoma is a rare malignant biliary neoplasm that causes a poor prognosis even after curative hepatectomy. Liver metastasis is the major recurrence pattern of intrahepatic cholangiocarcinoma; therefore, the prevention of liver metastasis is a desirable objective. The aim of this study is to identify gene(s) related to liver metastasis of intrahepatic cholangiocarcinoma and to examine the inhibitory effects on metastasis of intrahepatic cholangiocarcinoma by controlling such gene(s). We collected 3 pairs of intrahepatic cholangiocarcinoma frozen samples, and 36 pairs (primary and metastatic lesions) of intrahepatic cholangiocarcinoma formalin-fixed paraffin-embedded samples, from patients who underwent surgical resection at hospitals related to the Kyushu Study Group of Liver Surgery between 2002 and 2016. We carried out cDNA microarray analyses and immunohistochemistry to identify candidate genes, and evaluated one of them as a therapeutic target using human cholangiocarcinoma cell lines. We identified 4 genes related to liver metastasis using cDNA microarray, and found that CXCL12 was the only gene whose expression was significantly higher in liver metastasis than in primary intrahepatic cholangiocarcinoma by immunohistochemistry (P = .003). In prognosis, patients in the high CXCL12 group showed a significantly poor prognosis in disease-free (P < .0001) and overall survival (P = .0004). By knockdown of CXCL12, we could significantly suppress the invasive and migratory capabilities of 2 human cholangiocarcinoma cell lines. Therefore, CXCL12 might be associated with metastasis and poor prognosis in intrahepatic cholangiocarcinoma.

carbohydrate antigen 19‐9

cancer associated fibroblast

C‐X‐C motif chemokine 12

formalin‐fixed, paraffin‐embedded

intrahepatic cholangiocarcinoma

immunohistochemistry

keratin 83

liver metastasis

olfactomedin 4

regenerating islet‐derived protein 3 gamma

INTRODUCTION

Intrahepatic cholangiocarcinoma is a malignant biliary cancer that generally has a poor prognosis. Even after curative hepatectomy, its 5‐year overall survival rate remains 22%‐24%.1 One of the reasons for this is the high incidence of recurrence after curative hepatectomy. Therefore, patients with ICC need more effective and additional therapies, such as chemotherapy or radiotherapy before or after surgery, or both before and after surgery. Several clinical trials involving adjuvant or neoadjuvant therapy for ICC have been undertaken; however, there has been little evidence of their effectiveness to date.2 Although reresection is an effective treatment for limited recurrence cases,3 in order to improve patients’ prognosis, developing new treatment methods to reduce ICC metastasis after curative hepatectomy is an urgent issue.

Several molecular and biological studies on ICC metastasis have already been reported.4, 5, 6, 7, 8 As these studies were concerned with primary tumor or cholangiocarcinoma cell lines, they did not directly compare metastatic lesions to primary tumors using clinical samples. Therefore, how such proteins are expressed in metastatic lesions of ICC still remains unclear. In addition, in ICC, more than half of the instances of recurrence are LM9; therefore, being able to predict and prevent LM could lead to improved patient prognosis after curative hepatectomy.

The aim of this study is to find new therapeutic targets to suppress ICC metastasis using resected samples of primary and metastatic lesions of ICC.

MATERIALS AND METHODS

Patients and tissue samples

Intrahepatic cholangiocarcinoma frozen and FFPE samples were collected from patients who underwent surgical resection for ICC at hospitals affiliated with the Kyushu Study Group of Liver Surgery between 2002 and 2016. Three pairs of frozen samples and 127 patients’ FFPE samples were finally included in this study. Among the 127 patients’ samples, we could obtain 36 surgically resected metastatic FFPE samples in 30 primary samples. We obtained written informed consent from each patient, and the study procedure was approved by each institutional review board.

cDNA microarray

cDNA microarray analysis was carried out according to the Oncomics protocol using RNA extracted from the 3 pairs of frozen samples of primary and metastatic lesions of ICC, using an RNeasy Kit from Qiagen according to the manufacturer's protocol.

Cholangiocarcinoma cell lines

Human cholangiocarcinoma SSP‐25 cells and HuH‐28 cells were purchased from RIKEN Bioresource Center. SSP‐25 cells were grown in RPMI‐1640 medium supplemented with 10% FBS, and HuH‐28 cells were grown in Eagle's minimum essential medium supplemented with 10% FBS. The cells were maintained at 37°C in a humidified atmosphere containing 5% CO2.

Antibodies and siRNA

For IHC, mAbs against CXCL12 (MAB350) were obtained from R&D Systems, KRT83 (orb184603) from biorbyt, OLFM4 (#14369) from Cell Signaling Technology, and REG3G (ab198216) from Abcam. CXCL12 expression was transiently downregulated using a predesigned Silencer Select siRNA directed against CXCL12 from Thermo Fisher Scientific, and a nontargeting siRNA was used as a negative control. SSP‐25 and HuH‐28 were transfected with the annealed siRNA for 24 hours using Lipofectamine RNAimax from Thermo Fisher Scientific.

Real‐time RT‐PCR

RNA was isolated from the cultured cells using an RNeasy Kit from Qiagen according to the manufacturer's protocol. mRNA expression levels were determined by quantitative RT‐PCR using TaqMan probes (Roche), and the values were normalized to those of β‐actin. All quantitative RT‐PCR reactions were run using the LightCycler 480 System II (Roche Diagnostics). All data obtained using real‐time RT‐PCR were from experiments undertaken in triplicate, and the data are shown as the mean ± SE.

Immunohistochemistry

Paraffin‐embedded sections of tissues obtained from the ICC patients were deparaffinized and soaked in distilled water. Sample processing and IHC procedures were undertaken as described below. Endogenous peroxidase activity was blocked using 3% hydrogen peroxide. The sections were incubated with diluted antibodies (CXCL12, 1:50; KRT83, 1:200; OLFM4, 1:200; REG3G, 1:100) and detection was carried out with a biotin‐free HRP enzyme‐labeled polymer of the Envision Plus detection system (Dako). Positive reactions were visualized using diaminobenzidine solution, which was followed by counterstaining with Mayer's hematoxylin. All IHC staining was independently scored by 2 blinded pathologists, as follows: staining intensity was scored as 0 to 3 to indicate absent, weak, moderate, or strong expression, respectively. The percentage area of positive cells was scored as 0% to 100%. We calculated the IHC score according to the staining intensity score multiplied by the percentage area of positive cells, and then divided them into 2 groups by each median value of IHC scores.

Growth assay

We evaluated cell growth using a CCK‐8 Kit (Dojindo Molecular Technologies) according to the manufacturer's protocols. SSP‐25 and HuH‐28 cells were inoculated in a 96‐well plate at 3.0 × 103 cells in 100 μL/well and the plate was incubated overnight in a humidified incubator at 37°C with 5% CO2. We used SSP‐25 and HuH‐28 cells 24 hours after transfection with siRNA when evaluating under conditions that knocked down CXCL12. Each well of the plate also received 10 μL CCK‐8 solution at the indicated time points (0, 1, 2, and 3 days). Absorbance was measured at 450 nm using a microplate reader after incubating the plate for 1.5 hours. The absorbance of each sample was measured in triplicate.

Invasion assay

In vitro cell invasion assay was carried out using a BD BioCoat Matrigel Invasion Chamber (BD Biosciences) according to the manufacturer's protocol. Briefly, the invasion rate of tumor cells that migrated through Transwell inserts (8‐μm pore size) with a uniform layer of BD Matrigel basement membrane matrix was assessed. The ICC cells were seeded (SSP‐25, 2.0 × 104; HuH‐28, 3.0 × 104) into the upper chamber of the insert in 500 μL medium, and 750 μL medium in the lower well. After 48 hours of incubation (37°C, 5% CO2), the invading cells were fixed and stained. The number of invading cells in 3 predetermined fields (total magnification, ×100) was counted by independent investigators using a microscope. The mean of the number counted in each of the 3 fields was defined as the cell number. For each group, the cultures were carried out in triplicate.

Migration assay

Six‐well plates were imaged with a Keyence BZ‐X700 All‐in‐one Fluorescence Microscope equipped with a CO2‐ and temperature‐controlled chamber and time‐lapse tracking system (Keyence). Images were taken every 5 minutes for 24 hours and converted to movie files using a BZ‐X Analyzer (Keyence). The movies were analyzed for cell migration with the video editing analysis software VW‐H2MA (Keyence). We tracked 10 cells/well and the tracking data were subsequently processed with Microsoft Excel 2010 to create x‐y coordinate plots and distance measurements.

Statistical analysis

All experiments were undertaken in triplicate, and the data shown are representative of consistently observed results. Data are presented as the mean ± SD. The Mann‐Whitney U test was used to compare continuous variables between the 2 groups, and categorical variables were compared using the χ2 test. Survival curves were constructed using the Kaplan‐Meier method, and log‐rank tests were used to evaluate the statistical significance of differences. For all statistical analyses, we used JMP software (version 12; SAS Institute) and considered P values less than .05 were statistically significant.

RESULTS

CXCL12 identified as a key gene in LM of ICC by cDNA microarray and IHC

From the cDNA microarray, we extracted genes whose expression increased 2‐fold or more in LM compared with the primary tumor among the 3 pairs of frozen samples. Notably, the CXCL12 and KRT83 genes were most highly expressed in metastatic lesions compared with primary among these 114 genes after sorting by total score of fold changes (Figure 1A,B). We also extracted 162 genes whose expression decreased 2‐fold or more in LM compared with the primary tumor among the frozen samples (Figure 1C,D). REG3G was most weakly expressed in metastatic lesions compared with primary tumor. In addition, OLFM4 was reported as highly expressed in cholangiocarcinoma.10 Therefore, we chose REG3G and OLFM4 of the lower expressed genes on metastatic lesions.

Figure 1

cDNA microarray using 3 resected intrahepatic cholangiocarcinoma frozen samples. A, A total of 114 genes were commonly more highly expressed in liver metastasis than in primary lesions. B, and were most highly expressed among the 114 genes. C, 162 genes whose expression was commonly lower in liver metastasis than in primary lesions. D, was most weakly expressed among the 162 genes

Next, we evaluated the expression of these 4 genes by IHC using FFPE samples including 30 primary and 36 metastatic lesions. Typical images by each Ab are shown in Figures 2A and [Link], [Link], [Link]. We confirmed that CXCL12 was significantly highly expressed in metastatic lesions (P = .043); however, the other 3 genes were not significantly different (Figure 2B). In addition, we found that CXCL12 expression was higher in LM than in lung or lymph node metastasis (P = .003) (Figure 2C). Moreover, we evaluated CXCR4 and CXCR7 expression, which are receptors to CXCL12, and found there were no significant differences in CXCR4 and CXCR7 expression between primary and metastatic lesions (data not shown).

Figure 2

is highly expressed in liver metastasis (LM) lesions of intrahepatic cholangiocarcinoma by immunohistochemistry. A, expression according to intensity score 0 to 3. Scale bar = 500 μm. B, Only was expressed significantly more highly in metastatic lesions (M) than in primary lesions (P) (P = .043). C, expression was significantly higher in LM than primary lesions among several metastatic lesions (P = .003). LN, lymph node

CXCL12 expression in primary ICC correlates with poor prognosis and LM

We examined the association between CXCL12 expression in primary ICC and patients’ characteristics, prognosis, and LM. The following studies were carried out on CXCL12 expression in 127 primary ICCs. Of 127 patients, 66 expressed a high level of CXCL12 and 61 expressed a low level. In patients’ characteristics, CA19‐9 was significantly higher (P = .004) in the high CXCL12 group than in the low group. In addition, in tumor‐related factors, there were more patients with vascular invasion (P = .02), lymph node metastasis (P = .047), and adjuvant chemotherapy (P = .02) in the high CXCL12 group (Table 1). In prognosis, the patients in the high CXCL12 group had a significantly poor prognosis in both disease‐free (P < .0001) and overall survival (P = .0004) (Figure 3). Furthermore, with regard to LM, patients in the high CXCL12 group experienced significantly more frequent LM after resection (P = .0012) (Table 2).

Table 1

Comparison of clinicopathological characteristics and intrahepatic cholangiocarcinoma tumor‐related factors according to CXCL12 expression

Variable	CXCL12 expression		P value
	Low (n = 61)	High (n = 66)
Clinicopathological characteristic
Age (years)	65.0 ± 9.8	66.3 ± 10.2	.48
Gender (M/F)	36/25	44/22	.46
BMI (kg/m2)	23.2 ± 3.1	23.0 ± 3.8	.67
HBs‐Ag (+/‐)	7/54	6/60	.77
HCV‐Ab (+/‐)	13/48	9/57	.35
T‐bil (mg/dL)	0.96 ± 1.7	0.86 ± 0.38	.22
Alb (g/dL)	4.1 ± 0.4	4.1 ± 0.4	.89
PT (%)	97.2 ± 14.2	95.5 ± 14.6	.33
ICG R15 (%)	10.7 ± 6.7	10.4 ± 5.9	.90
Child‐Pugh (A/B)	60/1	63/3	.62
CEA (ng/mL)	2.9 ± 2.8	38.1 ± 258.4	.54
CA19‐9 (U/mL)	190 ± 942	2684 ± 9893	<.01
Tumor‐related factor
Gross type			.14
Mass‐forming	52	46
Periductal infiltrating	3	8
Mass‐forming + periductal infiltrating	6	11
Tumor size (mm)	34.5 ± 24.5	34.3 ± 25.2	.83
Tumor number (single/multiple)	53/8	57/9	.93
Tumor differentiation (well‐mod/poorly)	38/21	44/18	.56
Vascular invasion (yes/no)	21/40	36/29	.02
Lymph node metastasis (yes/no)	6/55	15/51	.047
UICC pStage (I‐II/III‐IV)	43/18	38/27	.19
Adjuvant therapy (yes/no)	8/53	21/45	.02

Alb, albumin; BMI, body mass index; CA19‐9, carbohydrate antigen 19‐9; CEA, carcinoembryonic antigen; F, female; HBs‐Ag, hepatitis B surface antigen; HCV‐Ab, hepatitis C virus Ab; ICG R15, indocyanine green retention rate at 15 min; M, male; PT, prothrombin time; T‐bil, total bilirubin.

Figure 3

Association between prognosis and expression in patients with intrahepatic cholangiocarcinoma. Patients with high expression had significantly shorter relapse‐free survival (P < .0001) and overall survival (P = .0004)

Table 2

Association between CXCL12 expression and liver metastasis in intrahepatic cholangiocarcinoma

	All	CXCL12 expression
	n = 127	Low (n = 61)	High (n = 66)
Liver metastasis (+)	50 (39.4)	15 (24.6)	35 (53.0)
Liver metastasis (−)	77 (60.6)	46 (75.4)	31 (47.0)

Data shown as n (%).

Knockdown of CXCL12 leads to reduced invasion and migration of ICC cell lines

We examined the suppression of metastatic potential by knockdown of CXCL12 using 2 human cholangiocarcinoma cell lines, SSP‐25 and HuH‐28. In the invasion assay, we confirmed that knockdown of CXCL12 (Figure 4A), and it significantly inhibited the invasive capabilities of SSP‐25 and HuH‐28 cell lines (Figure 4B). In addition, in the migration assay, we also confirmed that knockdown of CXCL12 significantly inhibited their migratory capabilities (Figure 4C). On the other hand, by knockdown of CXCL12, neither morphological change nor inhibitory effect on cell proliferation could be observed (data not shown).

Figure 4

Knockdown of suppresses invasion and motility of SSP‐25 and HuH‐28 cholangiocarcinoma cells. A, was suppressed by siRNA in both cell lines. B, Invasiveness of SSP‐25 and HuH‐28 was significantly suppressed by knockdown. C, Migration capabilities of SSP‐25 and HuH‐28 were significantly suppressed by knockdown. *P < .05; **P < .001; ***P < .0001. siCntl, control

DISCUSSION

This is the first report to investigate key molecule(s) in LM of ICC by cDNA analysis using clinically resected samples. We carried out this experiment using paired samples of primary and metastatic lesions, and could identify that CXCL12 was significantly highly expression in LM lesions compared to primary lesions by cDNA microarray and IHC. We also found that patients with high expression of CXCL12 in primary lesions had a higher incidence of LM and poor prognosis. In addition, we could confirm the inhibitory effects on invasiveness and migration capabilities of 2 human ICC cell lines by suppressing CXCL12 secretion by the cancer cell itself.

CXCL12, which is also known as stromal cell‐derived factor‐1, is a member of the C‐X‐C chemokine subfamily and a known ligand for the G protein‐coupled receptors CXCR4 and CXCR7.11, 12 Interactions between CXCL12 and CXCR4 or CXCR7 comprise a biological axis that affects growth, angiogenesis, and metastasis of cancers. Cancer cells themselves secrete CXCL12, as do CAFs and several organs such as lung, liver, bone, and brain.13 In other words, CXCL12 plays a role not only in paracrine but also in autocrine signaling through CXCR4 or CXCR7. There are several reports on paracrine function in cancer involving CXCL12 and CXCR4/CXCR7. Izumi et al14 found that CXCL12/CXCR4 activation by CAFs promoted the invasiveness of gastric cancer cells. In ICC, Ohira et al15 reported that the interaction of CXCL12 released from fibroblasts and CXCR4 expressed on ICC cells could be actively involved in ICC migration. Gentilini et al16 showed that ICC cell migration and survival were modulated by cross‐talk between CXCR4 and CXCL12 released by human hepatic stellate cells. Zhao et al17 also reported that interaction of CXCR4‐CXCL12 was associated with tumor formation, invasion, and migration of HuCCT‐1 cells using shCXCR4 and CXCL12. Thus, there are several reports concerning the paracrine interactions of CXCL12 in ICC. However, although Calinescu et al18 found that CXCR4‐CXCL12 autocrine positive feedback controlled glioblastoma progression, to our knowledge, there are no reports on autocrine signaling involving CXCL12 in ICC. Our study therefore has novelty in that we could suppress invasive and migratory capabilities by suppressing the ICC cells’ own CXCL12. As CXCL12 can bind to CXCR4 and CXCR7, if we target such receptors, we would need to block both receptors in order to suppress metastasis.

We also found that CXCL12 expression was significantly higher in LM than in lymph node and lung metastatic lesions, and patients with high expression of CXCL12 in the primary ICC were more likely to have significantly frequent LM. In addition, in our own experiments, CXCL12 tended to be more highly expressed in cholangiocytes than in other organs (lungs and lymph nodes) in IHC (data not shown). It is conceivable that LM is increased due to differences in secretion of CXCL12 in each organ, but to test this it will be necessary to further examine CXCL12 in each organ by quantification and animal experiments. Patients with high CXCL12‐expressing esophagogastric, lung, and pancreatic cancers had a poor prognosis compared those with low CXCL12 expression19; however, in ICC, the effect of CXCL12 expression on patients’ prognosis remains unclear. In addition, there is also no information available on the association between CXCL12 expression and metastatic lesions. Therefore, this study is the first to show an association between patients’ prognosis and CXCL12 expression in primary and metastatic lesions of ICC.

This study has 2 limitations. First, in the cDNA microarray, the number of pairs of clinical samples was small (n = 3); however, the pairs of primary and metastatic lesions of ICC frozen samples should be considered to be very valuable. Second, the study was undertaken using only clinical samples and cell lines. To confirm the inhibitory effect on LM by suppression of CXCL12 in vivo, animal experiments will be required.

In summary, CXCL12 was associated with invasion, migration, and metastasis in ICC, and might be a pivotal target that can improve prognosis in patients with ICC. Thus, additional studies are needed to evaluate this result as a potentially new therapeutic target or prognostic biomarker for ICC. In future, development of a multidisciplinary treatment strategy is expected to contribute to developing individualized therapeutic regimens in ICC.

DISCLOSURE

The authors declare that they have no conflict of interest.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.