Despite numerous relevant studies, the CAF-related treatment and prognosis of LUSC remained unclear. Further understanding of the molecular mechanisms underlying occurrence and development is necessary for clinical practice. Several studies focused on CAFs in LUSC patients in the last 10 years. CAFs had been reported to promote the recruitment of CCR2+ monocytes by CCL2, suppress CD8+ T-cell proliferation, and IFNγ production 31. Suzuki et al. 6 also claimed that PDPN+ CAFs contributed to the immunosuppressive microenvironment with higher expression of TGFB1. Despite the evidence of enhancing tumor growth, mouse models also validated their effects on squamous cell carcinoma lung metastasis with TGF-β activation 32. By intergrading the data from GEO and TCGA, we uncovered the close relationship between higher CAF and stromal scores and worse OS in LUSC patients for the first time. Drug response also differs according to CAF score, which could better guide clinical medication. Thus, the field of novel molecular targets in CAF-targeted therapies for LUSC patients was newly grounded.
We mined highly related hub gene modules by applying WGCNA and multiple computational algorithms into CAF and stromal co-expressed networks and then constructed the three-gene- (EFEMP2, MICAL2, CRISPLD2) CAF model. At the same time, the reported genes were found to be highly correlated with CAF infiltrations and CAF markers. Pathway enrichment analysis revealed the main enrichment in the aspect of ECM. Moreover, cytological and histologic evidence supported the three-hub-gene CAF risk model. As for the treatment response, 5-fluorouracil and afatinib are preferable for those with low CAF risk score, which highlight the clinical application of the CAF risk model.
As a unique family of redox enzymes, MICALs destabilize F-actin in cytoskeletal dynamics. They could mediate Semaphorin3A-NRP2 response to VEGFR1 in rat ECs and enter the p130Cas interactome in reaction to VEGF in HUVEC 33. Previous evidence had shown MICAL2 was strongly expressed in metastasizing cancer cells, especially at the invasive front and in the neo-angiogenic vasculature 34. MICAL2 was also involved in angiogenesis and vascular development pathways based on whole-genome gene expression profiling 33. MICAL2 expression levels were positively correlated with lymphatic metastasis and shorter overall survival in lung adenocarcinoma (LUAD) patients, which mainly involved the AKT and myosin-9 pathways 35.
Moreover, MICAL-L2 was reported as a key regulator of c-Myc deubiquitination and stability, promoting NSCLC cell proliferation 36. Clinically, MICAL2 inhibitors like CCG-1423 were proven to be effective in impeding endothelial cell migration and angiogenesis 37. It is promising that signaling molecules targeted at MICAL2 in tumor and immune cells will have bright therapeutic implications.
During lung development, CRISPLD2 plays an essential role in epithelial morphogenesis 38. CRISPLD2 inhibited the expression of pro-inflammatory mediators in lung fibroblasts and epithelial cells in early life and adulthood 39. It can also bind lipopolysaccharide (LPS) to prevent LPS from binding to target cells, reduce TNF-α and IL-6 production, and protect the body against endotoxin shock 40. Clinical evidence confirmed that it was down-regulated during sepsis and could be regarded as a potential biomarker 41,42.CRISPLD2 was found as a part of cancer stemness-related prognostic signature in colon adenocarcinoma 43. It was also reported highly expressed by cancer-associated myofibroblasts in cholangiocarcinoma tissue 44. Consequently, CRISPLD2 plays a role in tumor progression and inflammation, thus suggesting high prognostic value in this model.
Consistent with current knowledge, ECM receptor interaction, MAPK, and TGF-β signaling pathway performed essential roles in tumor development in the respective CAF. CAF and ECM were popular research hotspots in recent years 45. CAFs influence cancer cell behavior directly and indirectly, including via secreted molecules and adhesions between cells 31. Expressed on all tumor and stromal cell types, the integrin family regulates tumor proliferation, survival, and paracrine cross-talk between cells. Integrin α3β1 binding to laminin-332 promotes CAF differentiation and function 46. Integrin α11β1 regulates cancer stromal stiffness and promotes tumorigenicity and metastasis in non-small cell lung cancer 47. The tumor-inhibition and drug-delivery function of integrin-targeting drugs like Knottin was confirmed already 48. Former research pointed out that αVβ8, αVβ5, and α6β4 were three targets in LUSC 49. Our findings offer promising extracellular-reversion strategies targeting the integrin to regulate CAF activation.
Besides, the MAPK signaling pathway was vital in lung cancer proliferation and invasion. Vascular cell adhesion molecule-1 (VCAM-1) secreted from CAFs enhances lung cancer growth and invasion by activating the MAPK signaling pathway through receptor α4β1 integrin 50. In a paracrine manner, leptin produced by CAF promotes the proliferation and migration of NSCLC cells via PI3K/AKT and MAPK/ERK1/2 signaling pathways 51. Inhibitors of this pathway have shown effectiveness in clinical trials with manageable side effects 52. PD 098059, as the first MEK inhibitor described, inhibited MEK with an IC50 ~10 nmol/L, specifically 53. Other inhibitors of the RASRAF-MEK-ERK/MAPK signaling pathways like Sorafenib, Vemurafenib, and Dabrafenib have undergone phase II/III clinical trials with promising results 54–57.
In addition, the TGF-β pathway should not be ignored. Fibroblasts and myofibroblast-bearing integrins (i.e., αvβ3, αvβ5) could induce TGF-β from ECM, activating the TGF-β pathway, exerting broad effects on tumor pathology 58,59. ED-A-containing fibronectin could also enhance the association of the latent TGF-β 60. TGF-β activates the myofibroblast phenotype in CAFs, producing pro-metastatic chemokines (e.g., SDF-1/CXCL12) 61. Bintrafusp alfa, a bifunctional fusion protein targeting TGF-β and PD-L1, received encouraging efficacy and manageable tolerability in patients with NSCLC in a Phase 1 Trial 62. More excitingly, Belagenpumatucel-L, a TGF-beta2 antisense gene-modified allogeneic tumor vaccine, behaved well in a Phase II trial with NSCLC patients63.
To ensure robustness and avoid overfitting, we adopted four bioinformatics methods to quantify CAF infiltrations and examined the three vital genes in CCLE and IHC databases. Further basic and clinical data are still in need to confirm this model. Our CAF risk model could give a hint to guide basic research, drug development, and clinical follow-up.
However, some limitations should be mentioned in this study. To begin with, as a retrospective bioinformatic analysis rooted in two public cohorts, the model needs to be cross-validated in a large-scale population. Additionally, molecular and animal experiments are required to verify the clinical and biological value in practice. Nevertheless, our findings still provide novel knowledge for future CAF-related studies at LUSC.