High CTHRC1 expression may be closely associated with angiogenesis and indicates poor prognosis in lung adenocarcinoma patients

A total of 210 LUAD patients were included in this study; patients were diagnosed between 2003 and 2016 in the Department of Pathology of the First Affiliated Hospital of Sun Yat-sen University and followed up for between 6 and 125 months (median 46.0 months). The patients’ clinical and tumor pathological characteristics are shown in Table 1. Formalin-fixed and paraffin-embedded (FFPE) cancer surgical specimens were used to perform subsequent analyses. This study was approved by the institutional ethics committee of the First Affiliated Hospital of Sun Yat-sen University, and written informed consent was obtained from all patients.

Representative paraffin-embedded tissues were arrayed with a tissue-arraying instrument with 2.0-mm diameter core and were sectioned (4 um) for further analysis. CTHRC1 (Abcam, Cambridge, UK) was used in a 1:100 dilution [29, 30], VEGF (ZSGB-Bio, Beijing, China) and CD34 (ZSGB-Bio, Beijing, China) was used in ready to use dilution [31]. Samples were incubated with antibodies against CTHRC1 (Abcam, Cambridge, UK), VEGF (ZSGB-Bio, Beijing, China) and CD34 (ZSGB-Bio, Beijing, China). The protocol for the IHC staining of tumor tissues from humans was described previously [32]. Brown particles in the cytoplasm represent CTHRC1- or VEGF- positive staining. The expression intensities of CTHRC1 and VEGF were semiquantitatively evaluated according to the immunostaining intensity and positive cell distribution. The percentage of positive tumor cells was determined in at least three areas at 400× magnification and was averaged. The mean percentage was then assigned to one of five categories (Additional file 1: Fig. S1a–j): 0, no cancer cells stained; 1, 0–10% of cancer cells stained; 2, 11–50% of cancer cells stained; 3, 51–75% of cancer cells stained, 4, more than 75% of cancer cells stained. The intensity of immunostaining was scored as follows: 0, colorless; 1, tan; 2, brownish-yellow; and 3, dark brown. A weighted score was obtained by multiplying the positive cell percentage and staining intensity for each case. Microvessel density (MVD) was evaluated by the technique of Weidner et al. [33] and was based on the average CD34 positive cell count from IHC staining. Tumor slides were scanned first at low magnification (100×) to select three fields with the highest vascularization where the cell membrane of vascular endothelial cells was present and (or) there was brown staining, and then the microvessels were counted at high magnification (400×) (Additional file 1: Fig. S1k–o). Microvessels with a clearly defined lumen or a well-defined linear vessel shape were selected for counting and branching vessel structures were regarded as a single vessel. The mean value of three fields was considered as the microvessel density (MVD) for each case. Based on the receiver operative characteristic (ROC) analysis, the optimal cutoff value of CTHRC1, VEGF and MVD was confirmed: a staining index of 7.5 and 5 or greater was used to define tumors with high CTHRC1 and VEGF expression, respectively, and a staining index below 7.5 or 5 was defined as low expression; an evaluation of 28.5 or greater was used to define tumors with a high MVD, while an evaluation below 28.5 was used to define tumors with a low MVD [23, 34, 35].

JavaGSEA-3.0 was downloaded from the official website. Before performing the analysis, we downloaded the related Molecular Signatures Database (MSigDB) and obtained microarray data from 40 LUAD patients and from the microarray data of 514 LUAD cases from The Cancer Genome Atlas (TCGA) database (https://​www.​cbioportal.​org/​, accessed September 18, 2019). These two cohorts were divided into a high-CTHRC1 group and a low-CTHRC1 group based on the level of CTHRC1 expression. GSEA was then carried out to determine the functions or pathways that showed statistically significant, concordant differences between the two groups. A positive enrichment score (ES) and a normalized enrichment score (NES) indicate that the majority of genes in this gene set were positively associated with our predefined group statuses. A normalized P-value (NOM P-value) of < 0.05 was considered statistically significant.

Primary normal lung epithelial cells (BEAS-2B) were purchased from American Type Culture Collection (ATCC) and cultured in a keratinocyte serum-free medium (Invitrogen, Carlsbad, CA) supplemented with epidermal growth factor (EGF) (Invitrogen), bovine pituitary extract, and antibiotics (100 μg/mL streptomycin and 100 U/mL penicillin). Lung adenocarcinoma cell lines (A549, GLC-82, SPC-A1, PC9, H1299 and H1975) and non-adenocarcinoma cell lines (L78 and H460) were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, USA) supplemented with 10% fetal bovine serum (HyClone, San Angelo, TX, USA).

CTHRC1 shRNA vectors (GV248, CTHRC1-shRNA, and Control shRNA) and CTHRC1-overexpression vectors (GV358, CTHRC1, and Control vector) were purchased from GeneChem (Shanghai, China), and were manipulated according to the protocol provide by the manufacturer. The transfection efficiency of the template was detected by fluorescence microscope (Axio Observer Z1, Zeiss), western blotting and quantitative real-time PCR.

Western blotting was performed as previously described [32]. Targeted membranes were incubated with the antibodies for CTHRC1 (Abcam, USA), VEGFA (Abcam, USA) and Actin (Cell Signaling Technology, USA) in 5% milk/tris-buffered saline Tween-20 (TBST) at 4 °C overnight, and then were washed with TBST and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (Cell Signaling Technology, USA) for 1 h with shaking. After enhanced chemiluminescence (ECL) (Merck Millipore, Germany) reaction, the immunoreactive bands were observed through a Gel Imaging System (Syngene, USA).

Total RNA from tissue specimens and cells was extracted with TRIzol method Invitrogen, USA). RNA was reverse transcribed into cDNA using the two-step method with PrimeScript™ RT reagent kit with gDNA Eraser (Takara, China), according to the manufacturer’ s instructions. Then, qRT-PCR was performed with the SYBR^® Premix Ex TaqTM kit (Takara, China), according to the manufacturer’ s protocol. The following primers were used: CTHRC1, forward 5′-TGGACACCCAACTACAAGCA-3′ and reverse 5′-GAACAAGTGCCAACCCAGAT-3′; GAPDH, forward 5′-ACCCACTCCTCCACCTTTG-3′ and reverse 5′-CTCTTGTGCTCTTGCTGGG-3′. GAPDH was used as an internal reference, with the 2^−ΔΔCt method used for quantitation [36].

The clinical and histopathologic characteristics of the 210 LUAD patients are summarized in Table 1. The patient cohort consisted of 114 (54.3%) males and 96 (45.7%) females, and the mean age at diagnosis was 58.5 years (range, 33–81 years). Of the 210 patients with a known smoking status, most (79.5%) were nonsmokers. The tumor size was 2.67 ± 1.8 cm (range, 0.4–15 cm). Tumors were classified as well, moderately and poorly differentiated for 10 (4.8%), 181 (86.2%) and 19 (9.0%) patients, respectively. The number of patients with pathologic TNM stage I, II, III, and IV was 130 (61.9%), 25 (11.9%), 36 (17.1%) and 19 (9.0%), respectively.

Representative stained fields from histopathological slides for CTHRC1, VEGF and CD34 are displayed in Fig. 1a and Additional file 1: Fig. S1. As shown in Table 1, the CTHRC1 expression status in LUAD was significantly associated with the clinical stage (P < 0.001), T classification (P < 0.001), N classification (P < 0.001) and M classification (P < 0.001). In contrast, there were no significant correlations between CTHRC1 expression and other clinicopathologic characteristics.

Univariate analysis of clinicopathologic features demonstrated that age, clinical stage, T classification, N classification, M classification and CTHRC1 expression in LUAD were significantly correlated with overall survival (OS) (P < 0.05; Table 2), while clinical stage, N classification, M classification and CTHRC1 expression in LUAD were significantly correlated with progression-free survival (PFS) (P < 0.05; Table 2). According to the Kaplan–Meier analysis, the OS and PFS of patients in the CTHRC1^High group were significantly shorter than those in the CTHRC1^Low group (P < 0.001; Fig. 1b, c). Then, we performed a Kaplan–Meier survival analysis according to clinical stage, T classification, N classification and M classification and the data showed that clinical stage, N classification and M classification were significantly related to patient survival, and the survival time in the CTHRC1^High group was distinctly shorter than that in the CTHRC1^Low group (P < 0.01; Additional file 1: Fig. S2, Additional file 2: Table S1–S3).

As shown in Table 3, multivariate Cox regression analysis revealed that CTHRC1 expression was an independent prognostic marker for LUAD. The prognostic value of CTHRC1 expression was investigated according to clinical stage, T classification, N classification and M classification. Our results showed that the expression of CTHRC1 was strongly associated with the OS and PFS of patients in the clinical stage s I + II, clinical stages III + IV, T1–2, N0 and M0 subgroups (P < 0.05; Fig. 2). However, no significant difference was observed between the CTHRC1^High group and the CTHRC1^Low group in the T3–4, N1–3 and M1 subgroups, probably because there was a limited number of samples in these subgroups (Additional file 1: Fig. S3). We concluded that patients with tumors expressing high CTHRC1 levels had distinctly poor survival compared with patients with low CTHRC1 levels in the early or advanced stage subgroups (P < 0.05; Fig. 2a, b). Similarly, there was clearly shorter OS and PFS in patients with high CTHRC1 expression in the T1–2 subgroup (P < 0.05; Fig. 2c), N0 subgroup (P < 0.05; Fig. 2d) and M0 subgroup (P < 0.05; Fig. 2e). Therefore, CTHRC1 may act as an effective prognostic marker for LUAD patients.

We found that patients in the CTHRC1^High group expressed more VEGF (P < 0.001; Table 4) and had a higher MVD (P = 0.001; Fig. 3a, Table 4) than those in the CTHRC1^Low group. To elucidate the relationship between CTHRC1 expression and tumor angiogenesis, we utilized Gene Set Enrichment Analysis (GSEA) to determine whether a previously defined set of angiogenesis-associated genes showed statistically significant, concordant differences between the high-CTHRC1 group and the low-CTHRC1 group (ES = 0.556, NES = 1.442; P < 0.001; Fig. 3b) in our microarray data. The results showed that angiogenesis-associated genes were enriched in the high-CTHRC1 group, which indicated that the expression of genes related to the promotion of angiogenesis was more activated in LUAD patients with high CTHRC1 levels than in LUAD patients with low CTHRC1 levels. Additionally, in the TCGA cohort, we observed that the mRNA expression of CTHRC1 was positively associated with the expression of VEGF (r = 0.137, P = 0.002; Fig. 3c), and GSEA also showed that CTHRC1 expression remained linked with angiogenesis (ES = 0.635, NES = 1.838; P < 0.001; Fig. 3d). As shown in Fig. 3e, CTHRC1 and VEGFA was also differentially increased in six detected LUAD samples compared to corresponding non-cancerous tissues. These results further demonstrated that high CTHRC1 expression may be closely associated with tumor angiogenesis due to the upregulation of angiogenesis-related genes.

As shown in Fig. 4a, lung adenocarcinoma cell lines, including A549, GLC-82, SPC-A1, PC9, H1299 and H1975, exhibited higher CTHRC1 levels compared to that of primary human normal lung epithelial cells (BEAS-2B) and non-adenocarcinoma cell lines (L78 and H460). To investigate the angiogenesis marker expression on CTHRC1 enhancement or deregulation LUAD cells, we constructed stable clone of H1299-shCTHRC1 cells and A549-CTHRC1 cells by using lentivirus transfection. At 72 h after transfection with Lv-shCTHRC1 or Lv-CTHRC1 and their corresponding control virus, respectively, H1299 and A549 were observed by fluorescence microscopy. Green fluorescent protein (GFP) was expressed by more than 80% of the cells (Fig. 4b). After stable transfection and puromycin screening, CTHRC1 mRNA expression and protein levels in the H1299-Lv-shCTHRC1 group were significantly lower than that in the control group (0.147 ± 0.015 versus 1.066 ± 0.061, P < 0.001; Fig. 4b, c), while CTHRC1 mRNA expression and protein levels in the A549-Lv-CTHRC1 group were significantly higher than that in the control group (132.6 ± 7.540 versus 1.035 ± 0.039, P < 0.001; Fig. 4b, c). Furthermore, the protein levels of VEGFA also presented significantly higher in the A549-Lv-CTHRC1 group than in the control group (Fig. 4c). These data demonstrated that enhanced CTHRC1 expression can promote VEGF expression in LUAD in vitro.

Univariate analysis of clinicopathologic characteristics demonstrated that tumor angiogenesis markers (VEGF and MVD) were significantly associated with OS (P < 0.05; Table 2) and PFS (P < 0.05; Table 2). Based on our results described above, it is reasonable to simultaneously consider CTHRC1 expression and tumor angiogenesis markers as a predicted panel. Therefore, we established a predictive model that included CTHRC1, VEGF and MVD, and all patients were divided into three groups: patients were assigned to the low-risk group if the three biomarkers were low; to the moderate-risk group if more than one of the three biomarkers was high; and to the high-risk group if the three biomarkers were high.

According to the Kaplan–Meier survival analysis, the three survival curves were significantly different, and high-risk patients had a shorter OS and PFS than those in the low-risk group and moderate-risk group (P < 0.001; Fig. 5a, Additional file 2: Table S4). Similarly, in multivariable analysis, the predictive model was also confirmed as an independent predictor of OS and PFS (P < 0.05; Table 5). The predictive power for OS of the predictive model was better, with an area under the curve (AUC) of 0.740 (95% confidence interval [CI] 0.672–0.808; P < 0.001; Fig. 5b, c, Additional file 2: Table S5), than other single biomarkers, such as age, clinical stage, T classification, N classification, M classification, CTHRC1, VEGF and MVD, as shown by the ROC analysis. For PFS, although the predictive power of VEGF was the best, with an AUC of 0.704 (95% CI 0.632–0.776; P < 0.01; Fig. 5b, c, Additional file 2: Table S5), VEGF was not an independent predictor of prognosis (P > 0.05; Table 3). Therefore, as an effective predictor for PFS (P < 0.001; Table 5) in LUAD, the predictive model had the best predictive value, with an AUC of 0.691 (95% CI 0.620–0.763; P < 0.001; Fig. 5b, c, Additional file 2: Table S5).