Fibrinogen Alpha Chain Knockout Promotes Tumor Growth and Metastasis through Integrin–AKT Signaling Pathway in Lung Cancer

This article is featured in Highlights of This Issue, p. 941

Lung cancer is the leading cause of cancer-related deaths around the world (1, 2). About 80% to 85% of lung cancers are non–small cell lung cancers (NSCLC), including lung adenocarcinoma (LUAD, 40% of lung cancers) and lung squamous cell carcinoma (LUSC, 25%–30% of lung cancers; ref. 3). The majority of lung cancers are diagnosed at advanced stages and are inoperable (4). However, biologic risk factors of lung cancer aggressiveness and metastasis remain elusive. Fibrinogen is an extracellular matrix protein involved in blood clot formation, but also a key biologic factor associated with tumor angiogenesis and metastasis (5, 6). Fibrinogen is composed of fibrinogen alpha chain (FGA), beta chain (FGB), and gamma chain (FGG) encoded by a compact gene cluster, and each chain contributes two copies to the functional fibrinogen hexamer joined by disulfide bridging (7, 8). Fibrinogen is expressed primarily in hepatocytes (9) and mutations in any of the three genes (FGA, FGB, and FGG) cause dysfibrinogenemia. Specifically, FGA mutations can lead to hereditary systemic amyloidosis (10).

Early studies identified the role of fibrinogen and related fragments in tumor angiogenesis and metastasis. Fibrinogen and its breakdown products modulate the overall angiogenic potential of the solid tumors (6). Specifically, fibrinogen binds growth factors to stimulate endothelial cells and promotes an angiogenic phenotype (6). Fibrinogen is also cleaved by thrombin to form fibrin in conjunction with growth factors, extracellular matrix (ECM) proteins, and integrin α5β3 to promote angiogenesis (6). In animal models, lung metastasis after intravenous injection of lung carcinoma and melanoma cell lines is substantially reduced in fibrinogen-deficient mice (11). Recent clinical studies revealed that pretreatment of plasma fibrinogen is associated with poor disease-free survival in various cancers, including lung cancer (12). However, the degradation of fibrinogen yields fragments that affect angiogenic and metastatic processes. Fibrinogen fragments, caused by the degradation of FGB, have been shown to inhibit endothelial cell migration and tubule formation (13, 14). Of note, FGA interacts with HBsAg to promote apoptosis in HepG2 cells (15). Thus, fibrinogen and its polypeptide chains or yielded fragments may play different roles in tumor angiogenesis and metastasis.

Gene knockout (KO) for different parts of the fibrinogen molecule is now warranted to elucidate their role in angiogenesis and metastasis. A recent study used a genome-scale CRISPR screening library with 67,405 single guide RNAs (sgRNAs) to mutagenize a nonmetastatic mouse cell line of lung cancer (16). Once the mutant cells are transplanted into immunocompromised mice, resulting metastases are generated quickly. Enriched sgRNAs in lung metastases and late-stage primary tumors were found to target a small set of genes, suggesting specific loss-of-function mutations drive tumor growth and metastasis (16). Individual sgRNAs and a small pool of 624 sgRNAs that target the top scoring genes from the primary screen dramatically accelerate metastasis (16). Of note, mouse Fga is one of the most frequent targets with enriched sgRNAs in metastatic lung tumors compared with that in primary tumors (16). Human FGA encodes 610 amino acid residues, which is a plasma glycoprotein with a crucial role in the coagulation cascade through its conversion to fibrin (7). In the current study, to address the role of FGA in tumor growth and metastasis of lung cancer cells, we generated an FGA KO in two LUAD cell lines A549 and H1299 using CRISPR/Cas9 genome editing. Using these cell models, we investigated the effect of FGA on tumor growth and metastasis as well as in underlying signaling pathways.

Human LUAD cell lines A549 and H1299, breast cancer cell lines MBA-MB-231 and MCF7, prostate cancer cell lines LNCaP, PC3, and DU145, and hepatocellular carcinoma cell line HepG2 were obtained from the ATCC. Cells freshly amplified and frozen after obtention from the ATCC were used every 5 months. Cell line was authenticated by examination of morphology and growth characteristics and was confirmed to be Mycoplasma free. Cells were maintained in DMEM supplemented with 10% FBS (Thermo Fisher Scientific) and cultured for less than 6 months. Specific primary antibodies for Western blots or IHC were used to detect the following proteins: FGA, Integrin α5, CK5, CK8, Ki67, E-cadherin, Vimentin, BCL2, BCL-XL, MCL1, cleaved caspase-3, AKT, p-AKT^T308, p-AKT^S473, S6, p-S6^S235/236, 4EBP1, p-4EBP1^S65, and p-4EBP1^T37/46 as shown in Supplementary Table S1. Western Blotting Detection Kit was purchased from Millipore. Recombinant human FGA (Zeye Biotechnology), mutant recombinant human FGA (Cloud-Clone Corp.), and Fibrinogen (Sigma) were used for the treatment of cells. pCMV3-FGA-Flag vector was ordered from SinoBiological (Cat #: HG16000-CF, Wayne, PA) used for the overexpression of FGA in A549 cells.

For FGA KO, the single guide RNAs (sgRNA) were designed using the online CRISPR design tool (Benchling, https://benchling.com). The exon 2 region of FGA was selected to be targeted by CRISPR/Cas9 genome editing. A ranked list of sgRNAs was generated with specificity and efficiency scores. The pair of oligos for two targeting sites was annealed and ligated to the Bbs I-digested pSpCas9(BB)-2A-GFP (PX458) vector (Addgene) referencing a previously published protocol (17, 18). The pX458 plasmids containing each target sgRNA sequences were transfected into cells with Lipofectamine 3000 (Thermo Fisher Scientific). After flow cytometry sorting with GFP, 100 GFP⁺ cells were seeded into each well of a 96-well plate. After the selection of single colonies, FGA KO colonies were determined by Sanger sequencing with isolated genomic DNA, and FGA expression levels in each clone were validated by Western blot analysis. All sgRNAs were accessed using the online, off-target searching tool (Cas-OFFinder; http://www.rgenome.net/cas-offinder; ref. 19). To avoid an off-target effect, potential off-target regions were selected and subjected to PCR and Sanger sequence analysis. As previously described, the sgRNAs and primers for CRISPR design are shown in Supplementary Table S2 (18).

Cells were seeded into 12-well plates at a density of 1.5 × 10⁴ cells/well and were grown in complete medium containing 10% FBS. The viable cells were stained by 0.4% trypan blue solution (Sigma), and the cells were counted in triplicate every day using a hemocytometer as described previously (20).

After starvation of cells for 24 hours, 10⁵ cells with 200 μL serum-free DMEM were seeded into the upper chamber in Transwell chamber (8-μm pore size; Millipore), and 500 μL DMEM with 10% FBS was added into the lower chamber. After 24 hours, nonmigrated cells on the filter side of the upper chamber were cleansed with a cotton swab, and the polycarbonate membrane on the Transwell chamber was fixed with 10% formalin 800 μL for 15 minutes, rinsed with PBS 3 times, and stained with 50 μL DAPI for 10 minutes in the dark. The Transwell membrane was covered with cover glass by Fluoromount G (Thermo Fisher Scientific). The migrated cells were counted under an immunofluorescent microscope.

Three hundred cells/well were seeded into 6-well plates. After colony formation for 12 days, the plates were washed twice with cold PBS buffer, fixed with 4% paraformaldehyde for 10 minutes, and then stained with 0.2% (w/v) crystal violet for 30 minutes. The colonies were quantified by using the software of Image J.

Cells are harvested and pipetted well to become single-cell suspension in complete culture media in 1 × 10⁶/mL. A mixture of 0.9 mL 4% soft agar (Sigma) with 4.1 mL prewarmed 10% FBS DMEM was added into a 60-mm culture dish to make the bottom layer. The top layer contained 3 × 10⁴ cells in 3 mL of 10% FBS DMEM and 0.36% agar. The soft agar colony dish was marked and placed at a 37°C incubator for 3 weeks.

Apoptosis was assessed by flow cytometry based on cell binding to Annexin V (BD Biosciences). For apoptosis induction by FGA, cells were treated with 100 μg/mL recombinant human FGA for 1 hour.

Western blotting was performed as described previously (21, 22). For coimmunoprecipitation, cells were lysed in ice-cold buffer [20 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 1 mmol/L EDTA, and 1% NP-40] supplemented with complete protease inhibitors (Sigma) on ice for 10 minutes. Lysates were aliquoted into two tubes and incubated with the designated antibody or an appropriate IgG control for 16 hours at 4°C. Protein A/G agarose (Thermo Fisher Scientific) was used to precipitate antibody–protein complexes (23)

The ABC detection system (Vectastain Elite ABC Kit, Vector Labs) was used for immunostaining according to the manufacturer's protocol as described previously (21, 22). The results were determined to be negative if <10% of cells within tumor areas were stained or positive if 10% to 100% were stained. The percentage of positive tumor cells per slide (10%–100%) was multiplied by the dominant intensity pattern of staining (1, weak; 2, moderate; 3, intense); therefore, the overall score ranged from 10 to 300 H-scores (24). All slides were examined by two pathologists in a blinded fashion.

For tumor growth, wild-type (WT) and FGA KO A549 cells (2 × 10⁶ cells in 200 μL PBS) were injected subcutaneously into the right flanks of immunodeficient BALB/c nude mice 8 weeks old. Xenograft tumor size was measured every other day and a tumor volume formula was used (volume = (width (2) × length)/2) for caliper measurements. Mice were sacrificed at week 8 after tumor cell injection, and metastatic sites were checked by histologic analysis. All animal experiments were conducted in accordance with accepted standards of animal care and approved by the Institutional Animal Care and Use Committee of Harbin Medical University Cancer Hospital (Harbin, China).

A total of 1 × 10⁴ control A549 WT cells or KO cells were implanted intravenously into 8-week-old immunodeficient BALB/c nude mice. At 4 weeks after implantation, the mice were euthanized for histologic examination and expression analysis. The number of surface lesions over all lobes of the liver and lungs was scored before pathologic analysis. Tumor burden in the lungs was quantified in two-step sections from each lobe (lung left two lobes and right three lobes) in a blinded fashion by calculating the area of tumor tissue as a percentage of the total tissue area as described previously (23).

Fifty formalin-fixed and paraffin-embedded human lung cancer specimens were obtained from the Harbin Medical University Cancer Hospital. The tumor specimens were collected from 50 patients with lung cancer who underwent primary surgery between January 2012 and June 2018. All had histologically confirmed lung cancer with information on the histologic type and tumor stage (AJCC, American Joint Committee on Cancer) and grade (Supplementary Table S3). This study, involving the use of human lung tumor specimens, was approved by the Institutional Review Board (IRB) of the Harbin Medical University Cancer Hospital. For all specimens, written informed consent was obtained from all subjects in accordance with the requirements of the IRB.

The Cancer Genome Atlas (TCGA) Data Portal was used to download the data from samples of LUAD, lung squamous cell carcinoma (LSCC), and normal lung controls. The TCGA data analysis was performed using cBioPortal (http://www.cbioportal.org; refs. 25, 26) for genetic alteration analysis, UALCAN (http://ualcan.path.uab.edu/index.html; ref. 27) for gene expression and survival analysis, and MethHC (https://omictools.com/methhc-tool; ref. 28) for DNA methylation analysis. Gene-level normalized expression data were used in Partek Genomic Suite (PGS) for additional normalization, statistics, and annotation. FDR corrections (Benjamini–Hochberg methods) were applied to test multiple hypotheses.

Continuous variables were summarized using mean, SD, and median values. In samples with normal distributions, the means of the variables were compared using a two-tailed t test between two groups. In samples with nonnormal distributions, the medians of the variable between two groups were compared by a Mann–Whitney U test. ANOVA, one- and two-way, were used to test for overall differences, followed by Dunnett post hoc test for differences between groups. All data were entered into an access database using Excel 2016 and analyzed with SPSS (version 24; IBM), and StatView (version 5.0.1, SAS Institute Inc.).

We performed a genetic analysis of FGA in human lung cancers with the most commonly used TCGA dataset for LUAD and LSCC and other public multiple datasets for small cell lung cancer (SCLC). As shown in Supplementary Fig. S1A–S1C, in these datasets with more than 1,000 cases, genetic alterations of FGA were present in 4% of LUAD cases, 5% of LSCC cases, and 5% of SCLC cases. The genetic alterations of FGA mainly compose of gene deletions and mutations, including several truncating mutations and few gene amplification. Furthermore, we analyzed the mRNA expression of FGA and its relationship with patient survival in the TCGA dataset. Our analysis showed a significant 2.8-fold decreased mRNA expression of FGA in LUAD tissues compared with normal lung tissues (Supplementary Fig. S2A). Although 27-fold decreased mRNA expression of FGA was also evident in LUSC tissues compared with normal lung tissues, there was no statistical significance (Supplementary Fig. S2B). Of note, survival analysis showed that high mRNA expression of FGA was likely to be associated with poor prognosis for patients with LUSC but not LUAD (Supplementary Fig. S2C and S2D). In addition, DNA hypermethylation in the promoter region of FGA was evident in both LUAD and LUSC as compared with normal lung tissue controls (Supplementary Fig. S3A) and was negatively correlated with mRNA expression of FGA in LUAD but not LUSC (Supplementary Fig. S3B and S3C). These data suggest that genetic alterations of FGA are most likely to be an infrequent event, but a low mRNA expression is a common event in human lung cancers, which may be through DNA hypermethylation in the promoter region of FGA in LUAD.

Fibrinogen is generated primarily in hepatocytes (9), but it is also synthesized and secreted from epithelial cells, such as a LUAD cell line A549 (29) and breast cancer cell line MCF-7 and MDA-MB-231 (30). We next examined the protein levels of FGA in multiple human cancer cell lines. As shown in Fig. 1A, the expression level of FGA protein was the highest in hepatocellular carcinoma cell line HepG2, and the median expression was found in two LUAD cell lines A549 and H1299 but not in two breast cancer cell lines MBA-MB-231 and MCF7 and three prostate cancer cell line LNCaP, PC3, and DU145. Furthermore, using CRISPR/Cas9 genome editing, we knocked out FGA in A549, and H1299 cells, respectively, and the FGA KO cells were confirmed by Sanger sequencing (Supplementary Fig. S4) and Western blotting (Fig. 1B). In A549 and H1299 cells, cell proliferation and colony numbers were increased in FGA KO cells compared with that in WT cells (Fig. 1C–G). Likewise, cell migration and invasion were increased in FGA KO cells by transwell migration assay (transferred cell numbers, P < 0.001, KO vs. WT; Fig. 1H and I) and matrix invasion assay (colony spheroid area, P < 0.001, KO vs. WT; Fig. 1J–M), respectively. In addition, to test whether FGA KO-increased cell migration and invasion are related to the epithelial–mesenchymal transition (EMT), we further analyzed the expressions of EMT markers by Western blotting. As shown in Fig. 1N, expressions of cytokeratin (CK5 and CK8), and E-cadherin were reduced in FGA KO cells compared with that in WT cells, suggesting an increased EMT by FGA KO in LUAD cells.

We next determined the effect of FGA on apoptosis of A549 and H1299 cells. Although a decreased apoptosis was observed in FGA KO cells compared with that in FGA WT cells, no statistical significance was found (Fig. 2A and B), To address whether FGA induces apoptosis, we added FGA (10 μg/mL) into the culture medium to treat FGA KO A549 and H1299 cells, respectively. At 6, 12, and 24 hours after treatment with FGA, apoptosis was gradually elevated upon FGA stimulation in both FGA KO A549 and H129 cells (Fig. 2C and D), suggesting that FGA induces apoptosis in LUAD cells. BCL-2 family-regulated activation of caspase along with apoptosis in cancer cells contain several different signaling pathways (31). To elucidate the molecular mechanism underlying FGA-mediated cell apoptosis, expressions of BCL-2 family members and related proteins, such as BCL2, BCLXL, MCL1, and cleaved caspase-3 were determined by Western blot analysis in A549 cells. As shown in Fig. 2E, expressions of BCL2 and MCL1 were gradually decreased but BCLXL was not changed after FGA treatment for 12 hours in FGA KO A549 cells, whereas expression of cleaved caspase-3 was also gradually increased after FGA treatment for 24 hours in both FGA KO A549 and H1299 cells (Fig. 2F), suggesting FGA-induced apoptosis in LUAD cells.

We first measured the secreted protein levels of FGA in the culture medium of human A549 cells using ELISA. FGA in culture medium was dramatically reduced in FGA KO cells compared with that in WT cells (Fig. 3A). To determine the effect of FGA on cell proliferation, we added the recombinant human FGA (10 μg/mL) into the culture medium of A549 FGA KO cells and found a strong inhibition of cell proliferation by FGA (Fig. 3B), respectively. Cell colony assays further identified similar effects of FGA KO on tumor growth (Fig. 3C and D). Likewise, we observed similar effects of FGA on the suppression of cell migration (Fig. 3E and F). Furthermore, we used FGA WT and FGA KO cells, respectively, to coculture with FGA KO cells separated by Transwell chambers, and then counted migrated cells in the lower chamber (Fig. 3G). As shown in Fig. 3H, a decreased number of transferred FGA KO cells was observed by transwell migration assay in FGA KO cells cocultured with FGA WT cells as compared with those with FGA KO cells. Likewise, an increased FGA in the culture medium was also confirmed by ELISA in FGA KO cells cocultured with FGA WT cells (Fig. 3I). These results implicate a suppressive role of FGA in cell growth and migration of LUAD cells. Next, we treated the WT A549 cells with fibrinogen (10 μg/mL) or fibrinogen (10 μg/mL) plus FGA (10 μg/mL), respectively. Significant induction of cell proliferation was observed in the A549 cells after fibrinogen treatment, but this induction was partly reduced by addition of FGA (Fig. 3J). Likewise, this observation was confirmed by cell colony assay (Fig. 3K and L) and transwell migration assay (Fig. 3M and N). These results suggest an opposite or competitive role of fibrinogen and FGA in cell growth and migration of LUAD cells.

To investigate the FGA-mediated molecular mechanism in LCAD cells, we performed a coexpression analysis of FGA in human LUAD using the TCGA dataset. The mRNA expression levels of FGA were positively correlated with that of 192 genes and negatively correlated with that of 156 genes (Spearman correlation coefficient r ≥ 0.3 or r ≤ −0.3, P < 0.05; Fig. 4A; Supplementary Table S4). Of note, mRNA expression levels of FGA were highly coexpressed with that of FGG (r = 0.92, P < 0.001; Fig. 4B). Next, using these coexpression genes, we performed the KEGG pathway enrichment analysis. The top 3 enriched KEGG pathways, including focal adhesion, PI3K–AKT signaling, and riboflavin metabolism pathways, were significantly associated with FGA expression in human LUAD (adjusted P < 0.05; Fig. 4C).

PI3K–AKT signaling plays a critical role in tumorigenesis of NSCLC (32, 33). Consequently, we determined the effect of FGA on the top 2 signaling pathway in A549 cells. As shown in Fig. 4D, although expression levels of total AKT were not changed, both p-AKT^T308 and p-AKT^S473 were dramatically increased in FGA KO A549 cells compared with in FGA WT A549 cells. Likewise, as critical downstream effectors of AKT–mTOR signaling, p-4EBP1 and p-S6 were simultaneously upregulated after FGA KO in A549 cells (Fig. 4D), identifying an FGA loss-induced AKT–mTOR signaling. In addition, we used FGA to treat both FGA WT and KO A549 cells. Western blot analysis revealed that FGA did not induce p-AKT in WT cells, whereas FGA suppressed phosphorylation of both AKT^T308 and AKT^S473 in FGA KO cells (Fig. 4E). However, in FGA KO A549 cells, phosphorylation of AKT^T308 was increased in treatment with both recombinant integrin 5α and FGA as compared with those with FGA alone (Supplementary Fig. S5), suggesting that integrin 5α may compete with FGA to block the FGA-mediated suppression of AKT activation in LUAD cells.

Focal adhesion is the top ranking pathway associated with FGA expression in human LUAD (Fig. 4C). In focal adhesion, integrins are α/β heterodimeric adhesion glycoprotein receptors that regulate a wide variety of dynamic cellular processes, including cell growth, migration, and phagocytosis (34), through major downstream signal pathways, such as PI3K–AKT signaling pathway, in lung cancer progression (35–37). Fibrinogen is a ligand for integrin α5β1 on endothelial cells (38). Thus, FGA may regulate PI3K–AKT signaling through integrins in LUAD cells. To test this possibility, we immunoprecipitated FGA from A549 cells and probed them with an anti-integrin α5 mAb. As shown in Fig. 4F, anti-FGA coprecipitated integrin α5, whereas anti-integrin α5 coprecipitated FGA. In previous studies, crystal structure analysis identified the extracellular segment of integrin α5β3 in complex with an Arg-Gly-Asp (RGD) sequences (39), and functional analysis demonstrated that integrin α5β3 binds two specific RGD sequences (amino acids 114–116 and 590–593) of FGA (40). We used a mutant recombinant human FGA (amino acids 124–214) without the RGD sequences to test the binding of mutant FGA to integrin α5 in FGA knockout A459 cells. As shown in Supplementary Fig. S6A and S6B, there was no specific binding of mutant FGA to integrin α5 in the cells. In addition, we treated the WT A549 cells with fibrinogen or fibrinogen plus FGA, respectively. Western blot analysis revealed that fibrinogen induced phosphorylation of both AKT^T308 and AKT^S473 in A549 cells, whereas FGA dramatically suppressed the p-AKT in the cells regardless of fibrinogen treatment (Fig. 4G), indicating that FGA-mediated suppression of p-AKT may be independent to fibrinogen in LUAD cells. These data suggest that FGA inhibits PI3K–AKT signaling through a direct interaction of FGA with integrin 5α in LUAD cells (Fig. 4H).

To determine the effect of FGA on tumor growth in vivo, FGA WT and KO A549 cells were subcutaneously injected, respectively, into both male and female immunodeficient BALB/c nude mice. Xenograft tumor growth was faster in mice with FGA KO A549 cells compared with in WT A549 cells (Fig. 5A and B) up to 4 weeks after injection. Likewise, tumor weights were increased in mice with FGA KO A549 cells than in WT A549 cells at day 28 (Fig. 5C). Increased protein expression of Ki67 but decreased protein expression of E-cadherin was evident in FGA KO xenograft tumors compared with the WT xenograft tumors (Fig. 5D–F). Likewise, protein expressions of p-AKT^S473 were also increased in FGA KO xenograft tumors compared with the WT xenograft tumors (Fig. 5D). In addition, we conducted a xenograft assay with FGA overexpressed A459 cells in both male and female immunodeficient nude mice. As shown in Supplementary Fig. S7A–S7C, tumor growth and weights were decreased in mice with FGA overexpressed A549 cells as compared with those with WT A459 cells.

Pulmonary metastases of the A549-derived LUAD xenograft tumors have been observed in nude mice (41). However, we did not observe lung metastasis in the nude mice at 6 weeks after subcutaneous injection with FGA WT or KO A549 cells. Thus, to test the role of FGA in tumor metastasis in vivo, we intravenously injected FGA WT or KO A549 cells into male and female immunodeficient BALB/c nude mice. At 4 weeks after injection, a significant increase in the tumor number and burden of lung metastases were observed in the mice injected with FGA KO cells as compared with those with WT A549 cells (Fig. 5G–I). Significant increases of Ki67 and p-AKT^S473 in the xenografted metastatic tumors were also detected in FGA KO cells (Fig. 5J and K). These data suggest that FGA KO promotes A549 cell growth and colonization in vivo.

We next evaluated, by IHC, the protein expressions of FGA and p-AKT^S473 and their relationship in 50 human primary lung cancer tissues, including LUAD, LSCC, lung adenosquamous carcinoma, and SCLC (Supplementary Table S3). The protein expression of FGA in tumor cells was found in 36% (18/50) of total cases, including 41% (9/22) of LUAD, 32% (6/19) LSCC, and 29% (2/7) SCLC cases (Supplementary Table S3; Supplementary Fig. S7A). However, expression of p-AKT^S473 was found in 66% (33/50) of total cases, including 59% (13/22) of LUAD, 79% (15/19) LSCC, and 57% (4/7) SCLC cases (Supplementary Table S3; Fig. 6A). H-score quantitative analysis showed a negative correlation of protein expressions of FGA with p-AKT^S473 (Fig. 6B). Furthermore, no expression of FGA or expression of p-AKT^S473 was likely to be associated with poor 5-year disease-free survival, but these differences were not statistically significant (Fig. 6C and D). However, expression of FGA without expression of p-AKT^S473 was significantly associated with a better disease-free survival as compared with no expression of FGA with the expression of p-AKT^S473 (Fig. 6E). In addition, the expression of FGA was not associated with the histologic type and tumor stages and grades (Supplementary Table S3). These data suggest that the downregulation of FGA with the upregulation of p-AKT is likely to be a poor prognostic factor in human lung cancer.

On the basis of our bioinformatics analysis, genetic alterations of FGA are unlikely to be a frequent event in human lung cancers. However, our expression pattern analysis showed a low expression of FGA in human lung cancer tissues, including LUAD and LUSC. Of note, DNA hypermethylation in the promoter region of FGA is correlated with low expression of FGA in human LUAD tissues, suggesting an epigenetic mechanism in the transcriptional regulation of FGA in LUAD. Furthermore, in our experimental data, FGA KO promotes but the administration of FGA inhibits LUAD cell growth, migration, and invasion, as well as tumor colonization in the lung. Our functional analysis revealed the FGA-mediated regulation of tumor growth and metastasis through apoptosis and EMT is also involved in the integrin–AKT signaling pathway in LUAD cells in vitro and xenograft tumor model in vivo. These data suggest that FGA plays a suppressive role in the growth and metastasis of LUAD cells.

Fibrinogen, fibrin, and their degradation products are involved in blood clotting, inflammation, angiogenesis, and tumor metastasis (6, 42). Fibrinogen is thought to originate from exudation of plasma fibrinogen and subsequent deposition into the tumor stroma and is converted to fibrin polymers, resulting in the inflammatory response within the tumor microenvironment (TME) (43). The fibrin matrix may help induce angiogenesis to promote tumor growth and metastasis, while fibrinogen depletion can result in a reduction in tumor colonization in the lung (11, 44, 45). Various fibrinogen-derived peptide fragments also modulate the migration, proliferation, and differentiation of endothelial cells to affect tumor growth and metastasis (6). The products of fibrin degradation (E and D fragments) can stimulate the proliferation, migration, and differentiation of endothelial cells, contributing to tumor vasculature, progression, and metastasis (42). However, as polypeptide chains of fibrinogen, FGA-derived fragment (a 15-amino acid peptide) inhibits endothelial cell migration, adhesion, and tubule formation to reduce tumor growth (46). Likewise, FGB-derived fragment (a beta 43–63-amino acid peptide) is an inhibitor of activated endothelial cells and reduces tumor vascularization but induces the formation of tumor necrosis (14). Thus, fibrinogen, components, and their derivatives appear to play different roles in endothelial cells within the TME. Currently, in tumor cells, the role of fibrinogen, components, and their derivatives are not sufficiently understood. Fibrinogen can directly be synthesized and secreted by breast cancer cells and assembles into the extracellular matrix and reduces cancer cell migration (5). In lung cancer cells, fibrinogen augments tumor cell proliferation through interaction with fibroblast growth factor-2 (47) but blocks tumor cell migration (48). In the current study, FGA KO induces proliferation, migration, and EMT of LUAD cells in vitro and promotes LUAD xenograft tumor growth and lung metastasis in vivo, but administration of FGA inhibits LUAD cell growth, migration, and invasion, supporting an inhibiting role of FGA against LUAD cells. In addition, fibrinogen binds integrin α5β3 through two specific RGD sequences of FGA but not FGB or FGG (39, 40). Although FGA KO promotes tumor growth and metastasis through integrin α5, FGB or FGG unlikely has similar effects on LUAD.

The mechanism by which functional roles are different between fibrinogen and FGA in tumor growth and metastasis remains unknown. In the current study, we observed a significant induction of cell proliferation and migration in LUAD cells by fibrinogen treatment, but this induction was partly reduced by the addition of FGA, suggesting a potential competition between fibrinogen and FGA in cell growth and migration of LUAD cells. Likewise, fibrinogen induced phosphorylation of both AKT^T308 and AKT^S473 in LUAD cells, but this induction was blocked by the addition of FGA in a fibrinogen-independent manner. These data suggest that FGA may compete with fibrinogen to inhibit LUAD cell growth and migration through AKT signaling, but also has a fibrinogen-independent effect on AKT signaling through a direct interaction of FGA with integrin α5 in LUAD cells. However, the mechanism by which fibrinogen and FGA interact or compete against tumor cells or TME cells in vivo remains to be elucidated by further studies.

Many soluble secretory proteins released from cancer cells into the extracellular space are involved in inflammation and angiogenesis during tumor growth and metastasis (49, 50). As a secretory protein, fibrinogen binds integrins (e.g., α5β1, α2bβ3, and α5β3; ref. 51) in endothelial cells to promote tumor growth and metastasis. Fibrinogen has critical roles in tumor metastasis by facilitating the adhesion of pancreatic tumor cells to endothelial cells and transendothelial migration and extravasation (52). However, fibrinogen polypeptide chains, FGA, FGB, and FGG are downregulated during EMT of lung cancer cells (53). In HepG2 cells, knockdown of FGA promotes cell apoptosis, suggesting an FGA-mediated inhibition of apoptosis in cells, and expression of BCLXL and MCL1 are likely to be decreased, but BCL2 is increased in the FGA knockout-downed cells (15). However, FGA-derived fragment induces apoptosis and blocks tube formation of endothelial cells in gastric cancer, suggesting an apoptotic role of FGA in antiangiogenesis (46). In the current study, the administration of FGA induces apoptosis through downregulation of BCL2 and MCL1 but not BCLXL in LUAD cells. Of note, FGA directly binds to integrin α5 and stimulates AKT–mTOR signaling, suggesting a functional FGA–integrin–AKT axis in LUAD cells. Likewise, in TCGA dataset, survival analysis also showed an opposing role of FGA between patients with liver cancer and renal cancer (Supplementary Fig. S8). Thus, the role of FGA in apoptosis is likely to be different between various cell types, but the underlying mechanism remains to be elucidated by further studies.

In conclusion, FGA may play a suppressive role in LUAD cells to inhibit tumor growth and metastasis through induction of apoptosis and inhibition of EMT. In LUAD cells, FGA can bind integrin α5 and reduce phosphorylation of AKT, leading to an inhibition of mTOR signaling. Administration of FGA may provide a new therapeutic approach to inhibit LUAD cell growth and metastasis. However, FGA may also affect TME cells in vivo, such as endothelial cells, leading to the various roles of FGA in tumor growth and metastasis.

No potential conflicts of interest were disclosed.

Conception and design: R. Liu

Development of methodology: S. Gao, W.-H. Yang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Wang, Y. Zhang, Y. Liu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Cui, S. Wang, J.H. Bae, R. Liu

Writing, review, and/or revision of the manuscript: X. Cui, R. Liu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Wang, L.S. Qi, L. Wang

Study supervision: R. Liu

Other (carried out the experiments): G. Zhang, S. Gao

We thank Dr. Jonathan Leavenworth for editorial assistance in preparing this manuscript. This work was supported by grants from the Mike Slive Foundation for Prostate Cancer Research (L. Wang and R. Liu), the Breast Cancer Research Foundation of Alabama (L. Wang), and the Mercer University Seed Grant (W.H. Yang). Results are based, in part, upon data generated by the TCGA Research Network: http://cancergenome.nih.gov/.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.