LAMC1, Upregulated by TGFβ in Tumor Cells, Contributed to The Formation of Inammatory Cancer-Associated Fibroblasts Via NF-kB/CXCL1/STAT3 in Esophageal Squamous Cell Carcinoma.

Background: The tumor microenvironment (TME) consists of a variety of cells that interact with each other through cytokines. As an important member of the TME, cancer-associated broblasts (CAFs) play an important role in the development of tumor cells, which is inuenced by the heterogeneity of CAFs. Transforming growth factor β (TGFβ) not only plays a dual role in the progression of tumor cells directly but also inuences tumor cells by regulating the heterogeneity of CAFs. Methods: we explored oncogenes regulated by TGFβ, which were involved in signaling molecules and interactions in the TME. We analyzed sequencing data of TCGA and GSE53625, as well as ESCC cell lines with or without TGFβ1 stimulation, and then we focused on laminin subunit gamma 1 (LAMC1). The upregulation of LAMC1 after TGF-β1 stimulation was examined by western blot (WB), quantitative real time PCR (qRT-PCR) and Chromatin immunoprecipitation (ChIP). We performed gain-of-function and loss-of-function assays to examine the effect of LAMC1 on proliferation and migration of ESCC cells. CAFs were isolated and cocultured with ESCC cells. And conditional medium of shLAMC1 ESCC cells and CAFs with different treatments were collected. RNA-seq of those cells were also performed. Luminex liquid suspension chip detection, ELISA, WB, qRT-PCR and rescue experiments were carried out to reveal the interaction of between ESCC cells and CAFs. Results: LAMC1 was highly expressed in affecting the Moreover, could be upregulated through and SP1 synergistic activation. Further experiments showed that LAMC1 would and of tumor mainly Akt/NF-κB/MMP9 and MMP14. LAMC1 would CXCL1 mainly through CXCR2/pSTAT3. The conditioned medium of iCAFs promoted the proliferation and migration of tumor cells. Conclusions: Our study identied the mechanism by which upregulation of LAMC1 by TGFβ in tumor cells not only promoted ESCC progression but also indirectly induced carcinogenesis by stimulating CXCL1 secretion and promoting the formation of iCAFs. This suggests that LAMC1 could be a potential therapeutic target and prognostic marker for ESCC.

therapeutic target in many previous studies, but the results were disappointing 3,9,34 . An important reason may be the heterogeneity of CAFs 16,33 . Similar to tumor cells, different subgroups of CAFs exist among different tumors and within the same tumor 5,18,29 . According to the different functions of CAFs in tumors, CAFs are divided into tumor-promoting CAFs, tumor-suppressing CAFs and neutral CAFs in gastrointestinal tumors18. In pancreatic ductal adenocarcinoma, CAFs are classi ed into distinct in ammatory CAFs (iCAFs) and myo broblasts (myCAFs), which are not only different in their transcriptional pro le but also signi cantly differ in their tumor distribution location and effect on tumors 5,33 . In breast cancer, the speci c subgroup of CD10 + GPR77 + CAFs can maintain the stemness of tumor cells by secreting IL6 and IL8 and causing chemotherapy resistance. Targeting this speci c CAFs subgroup can reverse chemotherapy resistance in breast cancer 40 . The heterogeneity of CAFs is in uenced by paracrine and other pathways in other cells, such as IL1-induced LIF expression and downstream JAK/STAT activation to generate iCAFs 5,39 .
The regulation of CAFs heterogeneity by TGFβ is controversial. The TGFβ signaling pathway has both oncogenic and anticancer effects in ESCC 25,35,36 . In the early stage, tumor growth is inhibited by reduced TGFβ responsiveness. However, it later also promoted tumor invasion and metastasis. This "switch" in carcinogenesis may be related to the absence of adaptor proteins, such as β2-spectrin 25 . This contradictory effect of TGFβ not only exists in tumor cells but also regulates the phenotype of CAFs. The exosomes secreted by tumor cells contain TGFβ and induce the transformation of broblasts into activated CAFs in bladder cancer 39 . In prostate cancer, bone marrow-derived mesenchymal stem cells secrete TGFβ, which promotes the conversion of normal broblasts into CAFs with tumorigenicity 46 . However, tumor cell-secreted TGFβ antagonizes the activation of IL1 signaling and inhibits the transformation of myCAFs into iCAFs, which plays an anticancer role in pancreatic ductal adenocarcinoma 5 .
Overall, the tumorigenic effect of TGFβ on tumor cells has dual roles: it can affect tumor cells directly and indirectly through TME cell cross-talk, especially between tumor cells and CAFs 37 . We explored oncogenes regulated by TGFβ1, which are involved in signaling molecules and interactions in the TME. Through bioinformatics analysis, we hypothesized that laminin subunit gamma 1 (LAMC1) is an oncogene that is upregulated by TGFβ1 and participates in cell-to-cell signal transduction in the TME in ESCC. Many previous studies have shown that high expression of LAMC1 promotes tumor progression and can be used as a prognostic biomarker in many cancers, such as hepatocellular carcinoma, colorectal cancer, and endometrial cancer 21,52 . LAMC1 is involved in many carcinogenic effects within tumor cells. For example, miR-29b-3p negatively regulated LAMC1 to inhibit melanoma invasion 4 . LAMC1 can also promote the Warburg effect by upregulating PKM2 in hepatocellular carcinoma 50 . This study explored the carcinogenic effect of LAMC1 in ESCC from two aspects: direct effect on tumor cells and indirect effect through the TME, especially CAFs. Overall, we hypothesized that high expression of LAMC1, upregulated by TGFβ1, affected the prognosis of ESCC patients, and LAMC1 was involved in signaling molecules and interaction pathways in the tumor microenvironment. This study focused on exploring the tumorigenic mechanism of LAMC1 and its role in the microenvironment, especially in the interaction of tumor cells and CAFs.

Patients and tumor samples
A para n-embedded ESCC microarray containing 55 ESCC tissues and 50 adjacent tissues was purchased from Outdo Biotech (catalog no. HEsoS105Su01) for immunohistochemistry; 12 fresh ESCC tissues were obtained at our hospital for isolation of CAFs in 2018.
Isolation of CAFs and coculture system As described in our previous study, homogeneous CAFs were isolated from fresh tumor tissue and were identi ed using the cellular immuno uorescence marker αSMA 8 . Homogeneous CAFs were obtained for further analysis. All CAF cells used in the experiment were grown for no more than 10 passages. For the coculture system, CAFs were seeded in a 24-well plate with or without SB225002 in the medium, and the ESCC cells were placed in the upper chamber with a 0.4 µm pore size (Corning, USA). Cells were cocultured for 48 h, and then the RNA and proteins of CAFs were extracted.
Collection of conditioned medium (CM) As previously described, the CM was collected after cells had been cultured in serum-free medium for 24 h and was centrifuged at 1,000 × g for 5 min 8 . The CM was only concentrated 40-fold for western blot via Centricon Centrifugal lter (Millipore, USA). The CM used to stimulate cells was sterile ltered and diluted 1 time with the medium.

Drugs
For KYSE30 and KYSE450 cell treatments, recombinant TGFβ1 (R&D, USA) was used at a nal concentration of 10 ng/ml unless otherwise speci ed. TNFα (PeproTech, China) was used at nal concentrations of 10 ng/ml. Treatment periods were 24 h unless otherwise speci ed. We used 10 µM SB505124 and 10 µM JSH-23 (Selleck) to inhibit TGFβ signaling and NF-κB signaling, respectively. These inhibitors were administered to the cells 30 min before any other treatments. And 5uM MK-2206 2HCI(Selleck) was used to selectively inhibited Akt phosphorylation for 24 h. CAFs were treated with 10 ng/ml recombinant CXCL1 (PeproTech, China) for 24 h. SB225002 (Selleck, USA) was added 1 h prior to inhibiting CXCL1/CXCR2.

Tumor xenograft experiment
As previously described, 1 × 10 6 LAMC1 knockdown or overexpression KYSE30 cells and control cells were subcutaneously injected into the anks of BALB/c nude mice to establish tumor xenografts (6 mice per group). In another experiment, MRC-5 cells were stimulated with 20 ng/ml TGFβ1 for 4-5 days before further animal experiments. A total of 1 × 10 6 WT KYSE30 cells alone or 5 × 10 5 WT KYSE30 cells admixed with 5 × 10 5 MRC-5 cells (or rCXCL1-pretreated MRC-5 cells) were resuspended in equal 0.2 ml of PBS and then subcutaneously injected into the anks of mice to establish tumor xenografts (6 mice per group). Subgroups of mice were treated with SB225002 (1 mg/kg) intraperitoneally once every two days when the tumors reached 5 mm in diameter. The tumor volume was calculated by the formula V= (L◊W 2 )/2. Three or four weeks later, all BALB/c nude mice were sacri ced, and the tumors were excised and weighed.

Lung colonization assay
As described previously, cells were injected into female NOD-SCID mice through the tail vein. A total of 1 × 10 6 sh-1, sh-vec, LAMC1 and vector KYSE30 cells were injected (6 mice per group) 26 . The mice were sacri ced seven weeks later, and the lungs were excised and xed with 4% polysorbate, followed by embedding in para n for hematoxylin and eosin (H&E) staining. The number of lung surface metastatic nodes was calculated by gross and microscopic examination as previously described.

Statistical analysis
Prism GraphPad version 6.0, SPSS, GSEA, R script were used. Correlations between mRNA expression levels were analyzed using Pearson's correlation coe cient. A chi square test was performed to determine the relationship between clinicopathological variables and LAMC1 expression. Overall survival (OS) curves were analyzed by the Kaplan-Meier method and log-rank tests. The signi cant differences between different groups were analyzed using a two-tailed t-test. Data are presented as the mean ± standard deviation (SD). Differences were considered signi cant at P < 0.05 and are indicated as ****P < 0.0001, ***P < 0.001, **P < 0.01, and *P < 0.05 (ns, not signi cant) Results 1. LAMC1 expression was upregulated by TGFβ through synergistic activation of SMAD4 and SP1 and predicted a poor prognosis in ESCC.
Much of the previous research has established that TGFβ plays an important role in the tumor microenvironment, especially in cell-to-cell signaling 25 . We sought to identify genes regulated by TGFβ1, and those genes were found to be involved in signaling molecules and interaction pathways of the tumor microenvironment in ESCC. Furthermore, these genes themselves could in uence the prognosis of patients with ESCC. We made these observations through the following process. First, by analysis of our mRNA microarray data (GSE53625), we found that 4130 genes were upregulated in cancer tissues compared with adjacent tissues (Log2FC > 0.5, FDR < 0.001), and 238 genes were signi cantly (p < 0.05) associated with poor prognosis (Fig S1A). Second, by using Pearson's correlation analysis of GSE53625 and TCGA data, we found that 1609 genes were positively correlated with TGFβ1 in ESCC cancer tissues (r > 0.15, FDR < 0.05) (Fig. 1A, S1B). At the same time, in our previous study, RNA-seq was performed on TGFβ1-treated and untreated ESCC cells26, and we found that a total of 3084 genes were upregulated (Log2FC > 0, FDR < 0.05) (Fig. 1B). Combining the two results, we found that a total of 625 genes were coexpressed with TGFβ1 and upregulated by TGFβ1 (Fig. 1C). Enrichment analysis of these genes (KOBAS)47 (FDR < 0.0001) revealed that a total of 11 pathways were involved in environmental information processing, only two pathways of which, cytokine-cytokine receptor interaction and ECMreceptor interaction, were included in signaling molecules and interaction pathways (Fig. 1D, Fig S1C). A total of 31 genes were enriched in these two pathways. The 31 genes are positively regulated by TGFβ1 and are involved in signaling and cellular interactions in the TME. Overlap of the genes associated with prognosis showed that LAMC1 was the only one of the 31 genes that affected the prognosis of ESCC patients (Fig. 1E). Accordingly, at the protein level, LAMC1 was more highly expressed in cancer tissues than in para-cancer tissues, as demonstrated by IHC staining (Fig. 1F) and was also associated with low OS (Fig. 1G) and tumor stage (Supplementary Table 1). Therefore, LAMC1 can be used as an independent prognostic marker for ESCC.
Compared with that in the control group, the expression of LAMC1 was increased at the protein and mRNA levels in KYSE30 and KYSE450 cells after TGFβ1 treatment, which was time-and concentrationdependent in ESCC cells (Fig. 1H, Fig S2A, B). Additionally, to determine whether the TGFβ signaling pathway is responsible for the expression of LAMC1, we used the TGFβ receptor inhibitor SB505124 to eliminate the effect of TGFβ1 on LAMC1. The results showed that SB505124 could reverse TGFβ1induced LAMC1 expression in KYSE30 and KYSE450 cells (Fig. 1I, J). These results suggest that TGFβ signaling is responsible for the induction of LAMC1 transcription. By TRANFAC and JASPER database prediction, we speculated that the transcription factors SMAD4 and SP1 synergistically induced LAMC1 transcription. A ChIP assay was performed using anti-SMAD4 and anti-SP1 antibodies, and we found that TGFβ1 led to a signi cant increase in the enriched LAMC1 promoter sequence, suggesting that SMAD4 and SP1 were recruited to the promoter of the LAMC1 gene by TGFβ1 treatment (Fig. 1K). Additionally, by measuring proteins of chromatin fractions with antibodies against SP1 and SMAD4, we found that the expression of SMAD4 and SP1 were increased in each other's chromatin fraction (Fig. 1L). Furthermore, we conducted knockdown SP1 or SMAD4 ESCC cells respectively, and conducted ESCC cells that combined knockdown SP1 and SMAD4 (Fig. 1M, N, Fig S3 A, B). The expression of LAMC1 was decreased in those cells, which could not be rescued by TGFβ1 treatment (Fig. 1O-Q). This result suggested that the transcription factors SP1 and SMAD4 together induced the transcription of LAMC1. Taken together, these results showed that LAMC1 would be directly regulated by the TGFβ/SMAD4-SP1 signaling pathway.

LAMC1 promoted the proliferation and migration of ESCC cells in vitro and in vivo.
To evaluate the tumorigenic effect of LAMC1 on ESCC, we constructed KYSE30 and KYSE450 cell lines with stable knockdown or overexpression of LAMC1 ( Fig. 2A, B). shLAMC1 in KYSE30 and KYSE450 cells inhibited cell proliferation. Accordingly, overexpression of LAMC1 promoted cell proliferation (Fig. 2C,). And shLAMC1 in KYSE30 and KYSE450 cells promoted apoptosis (Fig. 2D). Furthermore, we also found that overexpression of LAMC1 could promote ESCC cell migration, while shLAMC1 inhibited migration (Fig. 2E).
In vivo, we established a xenograft tumor mouse model by subcutaneous inoculation or intravenous tail injection of KYSE30 cells transfected with shLAMC1, sh-vec, overexpression-LAMC1 and control vector. Consistent with the results of the in vitro experiments, the tumor volume and weight in the overexpressing LAMC1 group were signi cantly increased compared with those in the control group. The shLAMC1 group exhibited the opposite pattern ( Fig. 2G-J). The number of pulmonary metastasis nodules in the groups showed similar results (Fig. 2K, L). 3. The positive effect of LAMC1 on the migration in ESCC cells mainly via the Akt/IKKα/NF-κB/MMP9-MMP14 pathway.
We performed mRNA sequencing in KYSE30 and KYSE450 cells with sh-1 LAMC1 or sh-vec to explore downstream signaling pathways responsible for the aggressiveness of ESCC. Gene set enrichment analysis (GSEA) suggested that LAMC1 knockdown could affect the apoptosis pathway, NF-κB pathway, and cytokine and chemokine pathways (Fig. 3A). We detected IKKα phosphorylation of Akt, IKKα and p65 levels in shLAMC1 and LAMC1-overexpressing ESCC cells and found that expression of LAMC1 was positive correlated with phosphorylation of Akt, IKKα and p65 (Fig. 3B, C). Matrix metalloproteinases (MMPs) play an important role in tumor cell invasion and metastasis and are common downstream regulators of NF-κB-mediated cell metastasis 17 . We detected MMP2, MMP9, MMP10, MMP13, and MMP14. The results showed that the expression of MMP9 and MMP14 was in accordance with the phosphorylation of Akt, IKKα and p65 in shLAMC1-expressing and LAMC1-overexpressing ESCC cells (Fig. 3B, C). Furthermore, through cell immuno uorescence, we also found the expression of NF-κB (p65) were decreased in the nucleus in knockdown LAMC1 cells than the controls (Fig. 3D). Additionally, TNFα, as an activator of the NF-κ B pathway, could restore the expression of phosphorylation of IKKα, NF-κB, MMP9 and MMP14 in knockdown LAMC1 (Fig. 3E). Accordingly, the Akt phosphorylation selective inhibitor MK-2206 and the NF-κB nuclear translocation inhibitor JSH-23 both could reversed the high expression of phosphorylation of Akt, IKKα, NF-κB, MMP9 and MMP14 in overexpression LAMC1 cells (Fig. 3F).TNFα, also, reversed the inhibitory migration of shLAMC1 ESCC cells (Fig. 3G, S4 A). Accordingly, JSH-23abrogated the promotion of migration in LAMC1-overexpressing ESCC cells (Fig. 3H, S4 B I).
Cleaved caspase-9, cleaved caspase-3, and cleaved PARP levels increased in shLAMC1 KYSE30 and KYSE450 cells compared with the control group after cisplatin treatment (Fig. 4A), while they decreased in LAMC1-overexpressing cells compared with vector only cells (Fig. 4B). Furthermore, previous studies have shown that the NF-κB pathway can regulate cell anti-apoptosis via caspase 19 . We found that cleaved caspase9, cleaved caspase-3, and cleaved PARP expression in shLAMC1 cells was decreased by TNFα stimulation (Fig. 4C). Accordingly, after treatment with Akt phosphorylation selective inhibitor MK-the opposite results in proliferation and expression of cleaved caspase and PARP (Fig. 4D). Furthermore, JSH-23 also restored the positive effect on proliferation of overexpression LAMC1 cells (Fig. 4E), and TNFα could restore the negative effect on proliferation and the positive effect on apoptosis of shLAMC1 KYSE30 and KYSE450 cells (Fig. 4F, G).

CXCL1 would be regulated by LAMC1 mainly via NF-κB activation.
Increasing attention has been paid to the role of the TME in solid tumors. Cytokines and chemokines, as tumor-promoting factors, often play a role in intercellular signaling. By GSEA enrichment analysis of mRNA-seq data of shLAMC1 and sh-vec ESCC cells, knockdown of LAMC1 affected the cytokine and chemokine signaling pathways (Fig. 3A). Moreover, enrichment analysis of RNA-seq data of ESCC tissues (GSE53625) revealed the same effect of LAMC1 (Fig. 5A). We hypothesized that LAMC1 is involved in the regulation of signaling molecules and interactions in the TME via enrichment analysis of genes positively regulated by TGFβ1. All the above ndings suggest that LAMC1 may be involved in regulating the secretion of cytokines or chemokines. First, we detected a total of 48 cytokines and chemokines in the conditioned medium (CM) of shLAMC1 cells using the Bio-Plex Pro Human Chemokine Panel 48-plex kit. We found that the expression of CXCL1, IL8, and MIF increased in sh-vec cells compared with sh-1 cells at a higher concentration (Fig. 5B). We also detected the expression of the three cytokines in concentrated CM and found that only CXCL1 was regulated by LAMC1 both in KYSE30 and KYSE450 cells (Fig. 5C). Considering those results, we speculate that LAMC1 may upregulate CXCL1, and it was con rmed in knockdown LAMC1 and overexpression LAMC1 cells by ELISA (Fig. 5D). Thus, CXCL1 could be identi ed as a downstream target of LAMC1.
As predicted on the Cistrome website 53 , NF-κB can upregulate CXCL1 through transcriptional activation (Fig. 5E). In addition, we determined that LAMC1 may activate the NF-κB pathway, so we speculated that LAMC1 upregulates CXCL1 through NF-κB transcriptional activation. To verify whether NF-κB is responsible for CXCL1 expression, we used ELISA and WB to detect CXCL1 secretion of knockdown LAMC1 cells with or without TNFα stimulation and that of overexpression LAMC1 cells with or without MK-2206 2HCI and JSH-23. As expected, lower CXCL1 secretion by LAMC1 knockdown cells could also be increased by TNFα, and MK-2206 2HCI and JSH-23 could reverse the high expression of overexpression LAMC1 cells (Fig. 5F, G) And at the mRNA levels, expression of CXCL1 in shLAMC1 cells also could be revered by TNFα (Fig. 5H).
In vivo, the expression of LAMC1 was associated with that of CXCL1 at the RNA level (Fig. 5I), and the expression of CXCL1 was higher in cancer tissues than in adjacent tissues (Fig. 5J), but it did not affect the prognosis of patients, especially OS (Fig S4E).

CXCL1 secreted by ESCC tumor cells promoted the transformation of CAFs into in ammatory CAFs.
CAFs are heterogeneous cells with different subtypes, such as iCAFs and myCAFs 33 . The two subgroups not only have signi cant differences in their transcriptional pro les but also have different effects on tumor cells. myCAFs are contractile and can remodel the stroma, while iCAFs are characterized by a secretory phenotype and regulate tumor cells and other cells in a paracrine manner. CAFs were isolated from fresh tumor tissue and cultured in vitro and were identi ed by αSMA expression by cell immuno uorescence detection (Fig S5A). To explore whether ESCC tumor cell secreted CXCL1, upregulated by LAMC1, in uences CAF heterogeneity, we performed mRNA-seq in the following cells: CAFs treated with PBS, CAFs with recombinant CXCL1 (rCXCL1) treatment and CAFs with sh-vec CM treatment. We found that in CAFs treated with rCXCL1 or sh-vec CM, some gene clusters of iCAFs, including cytokines (CSF2, VEGF, etc.), chemokines (CXCL2, CXCL3, CXCL5, etc.) and interleukins (IL6, IL7, etc.), were upregulated than CAFs with PBS stimuli. In addition, gene clusters of myCAFs, such as COL1A1 and COL4A1, were downregulated (Fig. 6A). Moreover, compared with analysis of the controls, GSEA of CAFs with rCXCL1 or sh-vec CM treatment con rmed the upregulation of the cytokine/chemokine signaling and the regulation of STAT cascade, especially phosphorylation of STAT3 pathway. But the smooth muscle contraction pathway was downregulated (Fig. 6B, C).
In order to verify the above sequencing results, we detected iCAF markers (IL1, IL6, LIF, CSF3) and myCAF markers (Acta2, Ctgf) in CAFs treated with CM from shLAMC1 ESCC cells, CAFs cocultured with shLAMC1 ESCC cells, and controls. Since in ammatory markers are secreted, the expression of these proteins was detected in the concentrated CM of CAFs. At the protein and mRNA level, we found that CAFs cocultured with sh-vec ESCC cells or treated with the CM of sh-vec ESCC cells had higher expression of in ammatory markers than CAFs cocultured with shLAMC1 ESCC cells (Fig. 6D, E) or treated with CM from shLAMC1-1 ESCC cells (Fig. 6F, G). But the myCAF markers (Acta2, Ctgf) had lower expression in CAFs cocultured with sh-1 ESCC cells or treated with the CM of sh-1 ESCC cells than controls at protein level, which were not obvious at mRNA level (Fig. 6G). To verify whether CXCL1 is the main driver of this effect, we directly treated CAFs with rCXCL1 and obtained a similar result (Fig. 6H, I).
The CXCL1 common receptor is CXCR2, and combined with the above results of GSEA, we suggest that the CXCR2/pSTAT3 pathway may be responsible for this effect. We found that CXCR2 and pSTAT3 expression in CAFs was upregulated at the RNA and protein levels after rCXCL1 treatment (Fig. 7A, S5B). Additionally, SB225002, an inhibitor of CXCR2, reversed the upregulation of the in ammatory markers, CXCR2 and pSTAT3 and the downregulation of αSMA by rCXCL1 in CAFs (Fig. 7D-F). In addition, SB225002 decreased the changes in these proteins, especially in ammatory markers, in CAFs cocultured with shLAMC1 ESCC cells and CAFs treated with CM from shLAMC1 ESCC cells (Fig. 7B-C, G-L). All the above results suggest that CXCL1, regulated by LAMC1 and secreted by tumor cells, promotes the transition of CAFs into iCAFs via CXCR2/pSTAT3.

iCAFs, induced by CXCL1, promoted ESCC progression in vivo and in vitro.
Furthermore, to explore the effect of iCAFs induced by CXCL1 on the proliferation of tumor cells, we compared the proliferation of tumor cells after coculture with CAFs stimulated by PBS or rCXCL1 together with or without SB225002, and we found that CAFs with rCXCL1 promoted WT ESCC proliferation, which could be reversed by SB225002 (Fig. 8A). Additionally, to explore the in uence of iCAFs on migration, WT ESCC cells with different treatments were divided into four groups: the control group (WT ESCC cells treated with PBS) and WT ESCC cells treated with CM secreted by CAFs that were pretreated with rCXCL1 or PBS (CM-CAFs-PBS, CM-CAFs-pretreatCXCL1) in the presence or absence of SB225002. We found that both CM-CAFs-pretreatCXCL1 and CM-CAFs-PBS could promote the proliferation and migration of ESCC cells, and CM-CAF-pretreatCXCL1 had a stronger promoting effect than CM-CAFs-PBS, while SB225002 could also reverse the effect (Fig. 8B, C). In vivo, the tumor volume and weight of WT KYSE30 cells mixed with rCXCL1-pretreated CAFs were larger than those of cells mixed with CAFs without rCXCL1 pretreatment. SB225002 also weakened this effect (Fig. 8D, E). The migration markers MMP9 were measured by IHC and western blot in xenograft tumor tissue. Tumor with rCXCL1 pre-treated CAFs had higher expression of MMP9 than that of tumor with PBS pre-treated CAFs, which could be decreased by SB225002 (Fig. 8F, G).

Discussion
As in other types of tumors, high expression of LAMC1 can promote ESCC tumor cell proliferation and migration and has a poor prognosis, which can be used as a biomarker. We found that TGFβ1 regulates the expression of LAMC1 via SMAD4/SP1 synergic activation. Previous studies have shown that LAMC1 is regulated by SP1 in hepatocellular carcinoma 13,20,32 . The mechanism by which SMADs and SP1 synergistically activate genes has also been reported previously 48 .
The positive effect of LAMC1 on the phosphorylation Akt mediated NF-κB activation in ESCC, could promote the downstream processes of anti-apoptosis, pro-migration and secretion of CXCL1. However, the speci c mechanism of the positive effect of LAMC1 on phosphorylation Akt mediated NF-κB activation remains to be further studied. Many previous studies also con rmed the high expression of NF-κB in ESCC26. NF-κB expression was increased in mouse models of ESCC with p120-catenin knockdown.
NF-κB signals are activated by regulating upstream mediators, such as upregulating the transcription factor Id-1 or downregulating the tumor suppressor Nkx2-825. And TGFβ-induced long noncoding RNAs repressed NF-κB signals 26 . We found that shLAMC1 ESCC cells had higher expression of cleaved caspase 9, caspase 3, and PARP. The apoptosis pathways are mainly divided into endogenous and exogenous pathways, and the activation of caspase 9 and caspase 8, which are apoptotic initiators, represent endogenous and exogenous apoptosis initiation. After being cleaved and activated, caspase enzymes, such as caspase 3, are activated in a cascade manner, and cytoskeletal and nuclear proteins, such as PARP, are cleaved to promote apoptosis. LAMC1 could promote ESCC migration mainly by upregulating MMP9 and MMP14 downstream of NF-κB. In ESCC, some MMPs are upregulated to promote invasion, such as MMP2, MMP7, and MMP9 12,17 . In addition, LAMC1 and MMPs are both components of the ECM and are related to ECM remodeling. ECM remodeling plays an important role in the development of tumors, especially in invasion 6 . Other components of the ECM have also been reported in ESCC, such as bronectin, proteoglycan dermatan sulfate and hyaluronan 43 . LAMC1 could promote CXCL1 secretion through the transcriptional activation of NF-κB. As a transcription factor, NF-κB can regulate cytokine production 41 . Activation of the NF-κB pathway is often considered an important link between the in ammatory microenvironment and tumor development 41 . The cytokines downstream of NF-κB in ESCC have been reported to be IL-8 and IL1 2, 10, 24 . IL1 promotes tumor invasion, tumor-mediated immunosuppression, and tumor stem cell self-renewal28. IL1 also plays a role in inducing iCAFs 5 .
Many previous studies of LAMC1 have indicated that highly expressed LAMC1 can serve as a biomarker for a variety of tumors 21 , but less attention has been paid to the effect of LAMC1 on the heterogeneity of CAFs. We determined that ESCC tumor cells secrete CXCL1 and promote iCAF activation. In addition, we veri ed that TGFβ1 upregulated LAMC1. Furthermore, we found that CXCL1 expression was increased with TGFβ1 treatment, which could be eliminated by TGFβ receptor SB (Fig S2A-C). And TGFβ1 was positively associated with CXCL1 based on GSE53625 data (Fig S2D), Consistent with our ndings, in previous studies, CXCL1 expression increased in cocultures of CAFs and oral squamous cell carcinoma cells with IL1β, promoting tumor invasion and CAF activity 45 . In addition, CAF-secreted CXCL1 can induce the progression of ESCC tumor cells 51 . It is worth mentioning that CXCR2 inhibitors showed a better response to rCXCL1 on CAFs than tumor CM, suggesting that LAMC1 may also in uence other factors contributing to the formation of iCAFs. Whether LAMC1 in uences the heterogeneity of CAFs by affecting other factors or exosome secretion in tumor cells may require further study. Furthermore, CXCL1 increased the expression of in ammatory markers in CAFs and decreased the expression of myo broblast markers. In accordance with previous studies, iCAFs stimulated by rCXCL1 and sh-vec CM also presented a secretory phenotype that can interact with other cells in a paracrine manner and have tumor-promoting functions 33 . Due to its secretory phenotype, we speculate that iCAFs induced by CXCL1 may be involved in cancer-associated systemic effects 33 . CXCL1 activates phosphorylation of STAT3 in CAFs, which leads to an increase in many in ammatory mediators. Similarly, in pancreatic ductal cell carcinoma, tumor cell-secreted IL1 induces the formation of iCAFs in a cascade involving increased LIF expression and activation of JAK/STAT signaling 5 . It has also been con rmed that CAFs can reprogram cytokines secretion by other cells 39 . Moreover, it is worth mentioning that CAFs stimulation by either rCXCL1 or tumor CM did not induce proliferation (Fig S4A). This further indicated that the carcinogenic effect of CAFs was not through their own proliferation but indirectly regulated the phenotype of tumor cells through paracrine signaling 14, 31 .
We con rmed that TGFβ1 acted on tumor cells, causing a series of changes: upregulation of LAMC1, phosphorylation of NF-κB, secretion of CXCL1, phosphorylation of STAT3 in CAFs, and nally induction of the formation of iCAFs, that is, tumor-promoting CAFs. Previous studies have shown that TGFβ signaling plays a dual role in CAFs 46 . It can activate CAFs, such as through the conversion of NFs into CAFs and broblasts into tumor-promoting CAFs 39,46 . However, TGFβ1 secreted by tumor cells inhibits iCAF formation by antagonizing IL1 signaling activity and prevents myCAF conversion to iCAFs by blocking the JAK/STAT pathway 5 . The reason for these dual functions may be due to the balance between TGFβ signaling and STAT signaling in CAFs 5, 7, 27 .