Loss of alcohol dehydrogenase 1B in cancer-associated fibroblasts: contribution to the increase of tumor-promoting IL-6 in colon cancer

Abstract

Background

Increases in IL-6 by cancer-associated fibroblasts (CAFs) contribute to colon cancer progression, but the mechanisms involved in the increase of this tumor-promoting cytokine are unknown. The aim of this study was to identify novel targets involved in the dysregulation of IL-6 expression by CAFs in colon cancer.

Methods

Colonic normal (N), hyperplastic, tubular adenoma, adenocarcinoma tissues, and tissue-derived myo-/fibroblasts (MFs) were used in these studies.

Results

Transcriptomic analysis demonstrated a striking decrease in alcohol dehydrogenase 1B (ADH1B) expression, a gene potentially involved in IL-6 dysregulation in CAFs. ADH1B expression was downregulated in approximately 50% of studied tubular adenomas and all T1-4 colon tumors, but not in hyperplastic polyps. ADH1B metabolizes alcohols, including retinol (RO), and is involved in the generation of all-trans retinoic acid (atRA). LPS-induced IL-6 production was inhibited by either RO or its byproduct atRA in N-MFs, but only atRA was effective in CAFs. Silencing ADH1B in N-MFs significantly upregulated LPS-induced IL-6 similar to those observed in CAFs and lead to the loss of RO inhibitory effect on inducible IL-6 expression.

Conclusion

Our data identify ADH1B as a novel potential mesenchymal tumor suppressor, which plays a critical role in ADH1B/retinoid-mediated regulation of tumor-promoting IL-6.

Introduction

Colorectal cancer (CRC) is the fourth most deadly cancer with almost 900,000 deaths annually worldwide [1]. Over the past decades, it has been demonstrated that the tumor microenvironment (TME) supports tumor growth and immunity escape [2–4]. However, the various mechanisms by which the TME promotes cancer are far from being understood.

The importance of TME-restricted tumor-promoting inflammation to cancer initiation, progression and metastasis is well established [5, 6]. The tumor stroma of CRC consists of several cell types, of which cancer-associated fibroblasts (CAFs) are abundant. CAFs are a critical component of the TME contributing to tumor-promoting inflammation [7, 8]. In contrast to normal colonic stromal cells (known as fibroblasts and myofibroblasts, N-MFs), CAFs are biologically more diverse whose origin is reported to be from mesenchymal cell types- fibroblasts, pericytes, endothelium (or mesothelium), adipocytes, from engraftment of mesenchymal stem cells or fibrocytes of bone marrow origin, and from epithelial-to-mesenchymal transition [9–16]. CAFs display vast heterogeneity in transcriptomic and secretory profiles and these differences are suggested to be linked with the differences in their capacity to support tumor growth and immunity escape [2–4]. For example, recent lineage-tracing studies revealed that MCAM⁺ACTA2⁺CAFs may shape a tumor-promoting microenvironment via increased chemotaxis of myeloid cells [17].

We and others previously reported that CAFs, distinct from N-MFs, support tumor-promoting inflammation [16, 18, 19]. This is in part due to increased IL-6 production by CAFs [16, 19]. IL-6 is expressed at high levels in the tumor microenvironment and is a major mediator of inflammation [20]. Increased IL-6 production by CAFs has been shown to promote tumor cell growth, cancer stem cells, and metastasis in colon cancer [16, 19, 21]. A study in gastric cancers demonstrated that IL-6 is a CAF-specific secretory protein that, via paracrine signalling, confers chemo-resistance to the tumor cells [22]. Therefore, targeting IL-6 is proposed to be an attractive strategy to overcome stroma-induced resistance to chemotherapy for GI cancers [19]. The critical role of TLR4-NFkB signalling in the production of IL-6 by CAFs has been demonstrated [23–25]. However, the normal IL-6 regulatory networks that are impaired in CAFs and responsible for the abnormal IL-6 overexpression remain unclear.

Using a transcriptomic approach in the present study, we observed that several genes linked to the regulation of IL-6 expression or its downstream signalling are abnormally expressed in CAFs. Further, we identified alcohol dehydrogenases 1B (ADH1B) as a potentially novel regulator of IL-6 in MFs and showed that ADH1B is dramatically suppressed in CAFs. Our data suggest that ADH1B is a negative regulator of IL-6 and show that MFs are the major source of ADH1B in a normal colon. Further, a dramatic decrease in ADH1B expression by CAFs contributes to the upregulation of LPS-inducible IL-6. We demonstrate here that ADH1B plays an important role in retinoic acid regulation of inducible IL-6 production by stromal cells and, thus may be a novel target for cancer therapy.

Methods

Antibodies and biological reagents

Fluorochrome-conjugated murine anti-human α-smooth muscle actin (α-SMA, clone 1A4) monoclonal antibodies (Abs) were purchased from Sigma Aldrich (St Louis, MO). Rabbit anti-human ADH1B and β-actin polycolonal Abs, as well HRP-labelled anti-rabbit secondary Abs were purchased from Cell Signalling (MA, USA). Zenon Mouse Rabit IgG labelling kits were purchased from Life Technology (Grand Island, NY). Retinol (RO) and all-trans retinoic acid (atRA) were purchased from Sigma-Aldrich. Ultra-pure LPS from Escherichia coli was purchased from InvivoGen (San Diego, CA).

Human tissues and cells

Human samples were collected from patients undergoing colectomy for colon cancer under IRB-approved human protocols at the University of Texas Medical Branch at Galveston, TX. In addition, polyp and normal tissues were also collected under IRB-approved protocols at the University of Utah GI Tissue Bank. For MF isolation, full-thickness fresh human colonic tissue samples were obtained from discarded colonic surgical resection material from both the CRC tumor and its adjacent, uninvolved normal colonic tissue. T1–T4 tumors were used in this study. The demographic data, cancer stage and cancer grade for the samples used in this study are presented in Table 1. HPV = positive rectal tumors were excluded from this study. Normal colonic MFs (N-MFs) and carcinoma-associated fibroblasts (CAFs), were isolated according to the protocol of Mahida et al. [26], which is routinely used in our laboratory [27]. The purity of isolated CD90⁺ MFs (98–99%) was confirmed by flow cytometry, as previously described [27]. Cell culture studies were performed with primary MF isolates at passages 4–10. Cells were cultured at 37 °C in 5% CO₂ atmosphere in complete Modified Eagle Medium (MEM) as described previously [27].

Table 1.

Demographic and clinicopathological information on tissue samples.

Samplea	Age	Sex	Cancer stage (TMN data)	Cancer grade
CA-8	70	M	T1N0MX	MODERATELY DIFFERENTIATED COLONIC ADENOCARCINOMA
CA-61,b	74	F	T4N0MX	MODERATELY DIFFERENTIATED COLONIC ADENOCARCINOMA
CA-4	49	M	T4N1M1	MODERATELY DIFFERENTIATED INVASIVE COLONIC ADENOCARCINOMA
CA-1	76	M	T3N0MX	MODERATELY DIFFERENTIATED INVASIVE MUCINOUS COLONIC ADENOCARCINOMA
CA-52,b	61	F	T3N1MX	MODERATELY DIFFERENTIATED, INVASIVE ADENOCARCINOMA OF THE RECTUM
CA-7	51	F	pT4N1MX	MODERATELY DIFFERENTIATED INVASIVE COLONIC ADENOCARCINOMA
CA-23,b	59	M	T2N0MX	ULCERATED MODERATELY DIFFERENTIATED INVASIVE COLONIC ADENOCARCINOMA
NL-2	59	M	T2N0MX	ULCERATED MODERATELY DIFFERENTIATED INVASIVE COLONIC ADENOCARCINOMA
NL-1	60	F	n/a	VILLOUSE ADENOMA
NL-3	56	F	n/a	VILLOUSE ADENOMA
NL-51,b	74	F	T4N0MX	MODERATELY DIFFERENTIATED COLONIC ADENOCARCINOMA
NL-62,b	61	F	T3N1MX	MODERATELY DIFFERENTIATED, INVASIVE ADENOCARCINOMA OF THE RECTUM
NL-73,b	36	M	T2N0MX	MODERATELY DIFFERENTIATED COLONIC ADENOCARCINOMA

^aCancer-associated fibroblasts were isolated from colonic tissue at tumor site and referred in this table and further in the text as CAFs. Normal myo-/fibroblasts (N-MFs) were isolated from normal margin of tumor (distanced at least 5 cm from tumor) and referred as NL.

^n,bBoth N-MFs and CAFs were isolated from the same patient.

Microarray hybridisation

Total cellular RNA was extracted using RNAqueous (Ambion, Austin, TX, USA) according to the manufacturer’s recommendations. RNA was quantitated using a NanoDrop ND-1000 (NanoDrop Technologies, Wilmington, DE, USA). RNA integrity was assessed by visualisation of 18S and 28S RNA bands using an Agilent BioAnalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). Total RNA extracted from the samples was processed using the RNA labelling protocol described by Ambion (MessageAmp aRNA Kit Instruction Manual) and hybridised to Affymetrix Gene Chips (HGU133 Plus 2.0 arrays).

Microarray data analysis

Analysis was performed to identify genes (as probe sets) that were differentially expressed between myofibroblasts isolated from normal (N; n = 4) and colonic adenocarcinoma (CA; n = 5) tissues. Probe sets are surrogate genes in this context. Total RNA was isolated from confluent cultures and subjected to microarray analysis using Affymetrix HG-U133 plus 2.0 arrays. Data were deposited in a public repository; accession code number is 10.6084/m9.figshare.20685118.v2. Specific versions of the software used are recorded along with the code. Probe level data from 9 Affymetrix CEL files were normalized using the MAS5 algorithm then probe set, signal intensity, detection p-value, and detection status (present, absent or marginal) were exported using Affymetrix GCOS 1.4 (service pack 2) software. Only present probe sets were kept in the data set and summarised into gene expression levels. Differential gene expression analysis was performed with limma (v 3.50.0) between cancer and normal samples [28]. Upregulated and downregulated genes were submitted to gorilla for gene ontology: biological processes enrichment analysis [29]. Genes and gene ontology terms that had p-value < 0.05 were deemed significant. Key genes and gene ontology terms were expertly selected for illustration. Differentially expressed genes were further analyzed using the gene ontology (GO) Database and Cytoscape genetic interaction networks for their association with terms and pathways.

Analysis of publicly available colorectal cancer single-cell RNAseq data set

A publicly available single-cell RNAseq data set for cellular landscape and intercellular interactions in colorectal cancer was retrieved from https://colorectal-cancer.cells.ucsc.edu (UCSC [30]). The data set was analysed using the UCSC Cell browser interactive tool. We generated plots showing cell origin: tumor or normal tissue, and log-normalized gene expression levels in cells. Details on data processing methods are in the manuscript and UCSC Cell browser website listed above.

Quantitative RT-PCR

Total RNA was isolated using Qiagen RNAeasy mini isolation kits according to the manufacturer’s procedure (Qiagen, Germany). RNA analysis was performed according to the two-step RT real-time PCR protocol. Briefly, cDNA was prepared using Superscript Reverse Transcriptase (Life Technologies Inc.) according to the manufacturer’s instructions. The appropriate assays-on-demand™ gene expression FAM™ and VIC™ labelled primer/probe mixes were purchased from Thermo Fisher Scientific. Human β-actin and 18S genes were used as housekeeping genes. FastStart Universal Probe Master mix (Roche Diagnostic USA, Indianapolis, IN) was used to prepare a PCR mix according to the manufacturer’s instructions. The reactions were carried out in a 20 μL final volume using a BioRad Q5 real-time PCR machine according to the following PCR protocol: 2 min at 50 °C, 10 min at 95 °C (1 cycle) and 15 s at 95 °C and 1 min at 60 °C (40 cycles).

Immunostaining and confocal microscopy

Frozen human colonic step-sections were fixed and immunostained as described previously [16]. Samples were then mounted in SlowFade® Gold antifade reagent with DAPI (Life Science Technology, CA). Confocal microscopy was performed with a Zeiss LSM 510 confocal microscope (Carl Zeiss, Thornwood, NY). The Corrected Total Cell Fluorescence was used for quantitative analysis as previously described [31].

Effect of retinoids on MFs

Because of the significant light sensitivity of retinoids, these experiments were undertaken under yellow light. N-MF ad CAF were incubated with RO (2.5 µM) or atRA (2.5 µM) purchased from Sigma-Aldrich in the presence or absence of 1 µg/mL LPS. After 18 h, supernatants were collected for cytokine determination and RNA was isolated from cells as described above.

Primary MFs transfection with siRNA

Small interfering RNA (siRNA) technology was used to knockdown expression of ADH1B in N-MFs as described previously [32]. Negative siRNA controls were included in each experiment. Stealth siRNAs, siRNA probes to the conserved domains of ADH1B, and siRNA control probes were purchased from Thermo Fisher Scientific (USA). Transfection of MFs was performed using the Human Dermal Fibroblast Nucleofector kit according to the manufacturer instruction (Lonza, Allendale, NJ). ADH1B downregulation efficiency by each specific siRNA set was monitored by real time RT-PCR and Western Blot.

Western blot analysis

Total cellular proteins were extracted by lysing the cells from 12 well plates with 300 µL of Laemmli sample buffer (Sigma-Aldrich, MO, USA) under orbital agitation for 30 min at room temperature. Expression of ADH1B and β-actin (housekeeping) proteins was assessed using MF cytoplasmic extracts in 4-20% gradient SDS-polyacrylamide precasted gels (Biorad, CA, USA). Proteins were transferred to Immuno-blot PVDF membrane^® (Biorad, CA, USA) using a Transblot-turbo transfer system^® (Biorad, CA, USA). The membrane was blocked for 2 h at room temperature with 5% BSA. Blots were probed by overnight incubation at 4 °C with specific antibodies directed against ADH1B (1/1000 dilution) or β-actin (1/5000 dilution). The blots were subsequently incubated 2 h at room temperature with horseradish peroxidase-conjugated secondary antibody (1/2,000 dilution) and revealed by an Amersham ECL Western Blot detection reagent (Global Life Sciences Solutions USA LLC, Marlborough, MA, USA). Image acquisition and measurement of relative density were processed with Gel Doc EZ Imager^® coupled to Image Lab^® version 5.2.1 (Biorad, CA, USA).

Statistical analysis

Unless otherwise indicated, the results were expressed as the mean ± SEM of data obtained from at least three independent experiments done with duplicate sets in each experiment. Unless otherwise indicated differences between means were evaluated by one-way analysis of variance using Student’s t test for multiple comparisons. Values of P < 0.05 were considered statistically significant.

Reporting summary

Further information on experimental design is available in the Nature Research Reporting Summary linked to this paper.

Results

Changes in CAFs transcriptome impacts immune responses and metabolic processes

We and other previously demonstrated that CAFs and N-MFs preserved their differential activity in culture [16]. Thus, first we analyzed transcriptomic changes in cultured CAFs (n = 5) compared to N-MFs (n = 4) using an Affymetrix RNA microarray. Transcriptome analysis revealed significant changes in the gene expression profile between cultured CAFs and N-MFs (Fig. 1a). A total of 1201 genes (611 upregulated and 590 downregulated) were differentially expressed between sample groups with p value < 0.05 (Supplementary Table 1). Some highly upregulated and downregulated genes were previously described to have abnormal expression at tumor site, and in particular in CAFs in different cancers. This upregulated gene list includes insulin-like growth factor-binding protein 7 (IGFPB7), Chitinase-3-like-1 (CHI3L1), and Triosephosphate isomerase (TPI1) [33–35]. Among the top genes previously reported to be downregulated in CAFs were Complement factor D (CFD), Complement C1s (C1S), and Collagen 14A1 (Col14A1) [36, 37].

Fig. 1. Gene expression profiling of MFs in colonic mucosa: IL-6 gene network is altered in CAFs when compared to N-MFs.

a Volcano plot of differential gene expression of microarray expression data from N-MF (n = 4) and CAFs (n = 5) samples. Differentially-expressed genes(p value < 0.05) are shown in red (upregulated) and blue (downregulated). b Gene Ontology (GO) biological processes enrichment analysis of microarray data revealed a number of processes enriched among up- and downregulated genes between N- and C-CMFs. c Network of differentially expressed genes between CAFs and N-MFs. Upregulated genes are shown in blue, downregulated in green, and gene interaction network is shown around IL-6 (orange). Directions of predicted gene–gene interactions are shown with arrows. d UMAP of cell types in colorectal cancer patients from single-cell RNA-seq data set, showing tissue of origin. Data set is from Lee et al. (GEO:GSE132465). Inset shows the expression of ADH1B and IL-6 mRNAs in stromal compartment.

Associated GO terms demonstrated significant changes in the network of genes involved in the immune responses, angiogenesis, regulation of cytokine signalling network and metabolism occur in CAFs (Fig. 1b). These data are in agreement with previous reports showing that metabolomic and angiogenic properties are altered in CAFs to support tumor growth and progression [19, 38].

Transcriptomic profile of the IL-6 network is changed in CAFs when compared to normal stromal cells

We and others have previously demonstrated that an increase in basal and inducible IL6 production by CAFs is among key factors promoting tumor progression in colon cancer [16, 19], but causes of this increase remain unknown. Thus, next we analyzed global changes in the IL-6 gene regulatory network in CAFs. Transcriptome analysis demonstrated that the IL-6 regulatory network is altered in CAFs when compared to N-MFs with upregulated expression of several genes involved in the IL-6 signalling activation in CAFs when compared to N-MFs (Table 2 and Fig. 1a, c). Among the top IL-6 network genes upregulated in CAFs, besides IL-6, were ACVRL1, CHI3L1, CADM1, ENO2, LEF1, LIF, NOTCH3, TNFSF4, VCAN. Among the downregulated genes involved in the IL-6 network were CFD, CEBPD, CXCL12, GPX3,IL-33, IL-1R1, and IRF2. The link between the above genes and regulation of IL-6 expression was previously well established. Complement factor D (CFD) was among the highest downregulated gene in the IL-6 network. While its direct role in IL-6 regulation in cancer requires further investigation, it is reported to be involved in the upregulation of the cleavage of the IL-6 promoting complement C3a [39–41]. Surprisingly, among the genes that were dramatically downregulated in CAFs was Alcohol Dehydrogenase 1B (ADH1B), an enzyme which was also indirectly linked to IL-6 regulatory pathway by a machine learning algorithm (Fig. 1c). However, our in-depth literature search did not find a clear association or mechanistic reports about the role of ADH1B in IL-6 regulation. Thus, next we focused on the evaluation of the role of ADH1B in the regulation of IL-6.

Table 2.

Expression fold change (FC) of top genes involved in IL-6 network in CAFs versus N-MFs.

	Abbreviation	Protein name	Function/involving in regulation/interaction with IL6	Reference	FC
Upregulated in CAFs	CHI3L1	Chitinase-3-like protein 1	Upregulates production IL-6 by added rCHI3L1 in breast carcinoma.	Cohen et al. [34]	4.90E + 05
	TNFSF4	Tumor necrosis factor (ligand) superfamily, member 4	Signalling through OX40L upregulates IL-6.	Burgess et al. [56]; Karulf et al. [55]; Lei et al. [54]	1.50E + 03
	LEF1	Lymphoid enhancer-binding factor 1	TCFs/LEF signalling positively regulate IL-6 expression in astrocytes.	Robinson et al. [58]	5.00E + 02
	NOTCH3	Neurogenic locus notch homologue protein 3	Functional Notch3 is essential for IL-6 to maintain high levels of carbonic anhydrase, an enzyme associated with hypoxia survival and cancer invasiveness.	Sansone et al. [62]	1.40E + 02
	ENO2	Enolase 2 (Gamma, Neuronal)	Induces the expression of glycolysis-related metabolic genes (ENO2, HK2, and PFKFB3) and thereby increased lactate production in CAF cells that may upregulate IL-6.	Kim et al. [99]	1.20E + 02
	CADM1	Cell adhesion molecule 1	CADM1 is required for Tax to activate the canonical NF-κB pathway and downstream IL-6.	Hartsough et al. [60]; Pujari et al. [59]	5.90E + 01
	ACVRL1	Activin receptor-like kinase 1, Serine/threonine-protein kinase receptor R3	ALK1-co-expressed genes were highly enriched in IL6/JAK/STAT3 pathways.	Bocci et al. [100]	2.70E + 01
	IL6	Interleukin 6	IL-6 in CAFs activates the signal transducer and activator of transcription 3 (STAT3) in both a paracrine and an autocrine manner.	Nguyen et al. [101]	2.00E + 01
	VCAN	Versican proteoglycan	Versican is involved in an inflammatory tumor microenvironment and induces the expression of IL-6.	Hope et al. [102]	1.80E + 01
	LIF	Leukaemia inhibitory factor	Autocrine LIF is required for IL-6 regulation in human fibroblasts.	Nguyen et al. [101]	1.20E + 01
Downregulated in CAFs	CFD	Complement factor D	Enhances cleavage of complement C3a, which is known to induce IL-6 production.	Takabayashi et al. [39]; Song et al. [41]	7.80E-06
	ADH1B	Alcohol dehydrogenase 1B (class I), beta polypeptide	Enzyme metabolising ethanol, retinol, and other aliphatic alcohols.		1.30E-04
	CEBPD	CCAAT/enhancer-binding protein delta	A nuclear factor that specifically binds to an IL1-responsive element in the IL-6 gene.	Akira et al. [66]	1.30E-03
	LAMA4	Laminin subunit alpha-4	Downstream target of IL-6 pathway. Blockade of IL-6/sIL-6R signalling downregulates LAMA4.	Zegeye et al. [103]	2.60E-03
	CXCL12	Stromal cell-derived factor 1	CXCL12 enhanced IL-6 in synoviocytes.	Nanki et al. [69]	3.60E-03
	eIF3K	Eukaryotic translation initiation factor 3k	eIF3 is acting as a regulator of cap-dependent transcript-specific translation through direct binding to defined RNA structural elements. In non-mammal organisms eIF3K downregulates LPS induced IL-6 through inhibition of NFkB.	Chen et al. [104]; Lee et al. [105]	4.00E-03
	IL33	Interleukin 33	IL-1 family of cytokine member, regulates production of IL-6.	Shao et al. [71]	7.10E-03
	GPX3	Glutathione peroxidase 3	One of the key defensive enzymes against oxidative damages to host cells, downregulated in several types of cancers.	Zhang et al. [106]	8.10E-03
	IL1R1	Interleukin-1 receptor type 1	IL-1R1 inhibition using recombinant IL1 receptor antagonist significantly reduces stromal-derived IL6 in IL-1α stimulation.	Dosch et al. [107]	1.60E-02
	IRF2	Interferon regulatory factor 2	Many primary human cancers, including lung, colon, breast, prostate, and others, frequently downregulated IRF2 that leads to immune evasion through decreased MHC class I antigen presentation and increased PD-L1 expression.	Kriegsman et al. [108]	2.30E-01

Expression of ADH1B is downregulated in cultured CAFs and in situ in colon cancer

Our data indicated that expression of ADH1B mRNA was dramatically decreased in primary cultured CAFs (Fig. 1a and Table 2). To eliminate potential “in vitro” artefacts, we first analyzed publicly available recent single-cell (sc) RNAseq data of colon cancer published by Lee et al. [30]. scRNAseq data analysis demonstrated that in colon cancer tumors ADH1B was downregulated in the myofibroblasts population and this was associated with upregulation of IL-6 (Fig. 1d). This study was in agreement with our previous observations that cultured CAFs have an α-SMA⁺ vimentin⁺, but desmin-negative phenotype that corresponds to myofibroblasts in situ [42, 43]. Downregulation of ADH1B in cultured CAFs when compared to adjacent N-MFs was confirmed at the mRNA and protein levels by qRT-PCR and western blot, respectively (Fig. 2a, b).

Fig. 2. Colonic myofibroblasts (MFs) are the major expressers of ADH1B in normal mucosa and this expression is lost in CAFs.

ADH1B mRNA levels and ADH1B protein expression measured, respectively, by RT-qPCR (a) and western blot (b) in cultured MFs are significantly downregulated in CAFs compared to the matched N-MF pair, t test was used for statistical analysis, n = 10 (a) and n = 5 (b) per group, ****p < 0.05. c High-power resolution confocal images from representative sections of eight matched sets of normal and CRC tumor tissue from the same subject are shown. In these experiments, immunostaining of CRC tumor and its surrounding normal colonic tissue sections was performed, followed by confocal microscopy. DAPI was used to stain cell nuclei (grey); activated MFs were detected by anti-α-SMA mAb (green) and stained for ADH1B with mAb (blue). A light blue colour on merged images indicates co-localisation of α-SMA and ADH1B, and thus the expression of ADH1B by MFs. In the normal tissue, ADH1B was largely colocalised with α-SMA expressed by activated fibroblasts, while no significant expression of ADH1B by the tumor cells was observed. Relative density of ADH1B staining in tumor tissues over matched normal colonic tissue, t test was used for statistical analysis, n = 8, ****p < 0.001. d ADH1B mRNA expression is drastically decreased at early stage of colon cancer. A drastic downregulation of ADH1B expression was observed in nearly one-half of tubular adenoma polyps (TA), but not hyperplastic polyps, and this reduction was present in essentially all colonic T1–T4 adenocarcinomas, real time RT-qPCR analysis. Non-parametric test with Dunn’s correction for multiple comparison was used for statistical analysis. The means ± SEM are shown, n = 5–16 per group, *p < 0.01; ****p < 0.0001.

Similar observations were made in situ (Fig. 2c), in which we identified MFs by expression of α-SMA. We observed that ADH1B expression was mainly co-localised within N-MFs in normal tissues (formation of cyan-white colour on merged images, Fig. 2c). By contrast, expression of ADH1B was essentially lost in the tumor stroma (Fig. 2c).

To investigate the stage of neoplastic transformation where the loss of ADH1B gene expression occurs, we analyzed ADH1B mRNA expression in tubular adenomas, which represent an early neoplastic transformation stage, and at the colon cancer carcinoma (T1–T4) stage and compared these to the ADH1B level in normal colonic tissue. We also determined the expression of this gene in benign hyperplastic polyps. We observed (Fig. 2d) that ADH1B levels in tubular adenomas were statistically significantly reduced, when compare to biopsies from healthy controls and hyperplastic polyps. However, more than half of the tubular adenomas had only slight decrease in ADH1B expression while the remainders showed dramatic decrease or compete loss of ADH1B expression similar to that observed in the colonic adenocarcinomas. Taken together, these results demonstrate that activated MFs are the major cell site of ADH1B expression in normal colonic mucosa, and that ADH1B expression is dramatically reduced in CAFs and in the colonic tissues somewhere during the neoplastic to carcinogenic progression.

ADH1B negatively regulates LPS-inducible IL-6 in N-MFs

Next, we evaluated whether ADH1B regulates IL-6 production in N-MFs using ADH1B-specific siRNA. The efficiency of ADH1B expression knockdown was confirmed at the mRNA and protein levels by real time qRT-PCR and Western blot analysis (Fig. 3a, b). We observed that silencing of ADH1B by specific siRNA, but not control siRNA, in N-MFs resulted in a dramatic increase in LPS-induced IL-6 expression and secretion (Fig. 3c, d, respectively). In this study, LPS was chosen as an activator of the TLR4 signalling pathway, which is known to be involved in the increase in tumor promoting IL-6 in CAFs in intestinal tumorigenesis [25]. These results suggest that the decreased ADH1B within MFs may be among the key factors leading to the tumor promoting inflammatory activity of CAFs via higher production of LPS-inducible IL-6.

Fig. 3. ADH1B substrate RO is unable to downregulate LPS-induced IL-6 expression in CAFs.

Silencing of ADH1B gene in N-MFs reproduced a pro-inflammatory CAFs phenotype. Cultured fibroblasts have been isolated from normal human colonic biopsies and ADH1B gene was silenced using specific siRNA. ADH1B-silenced cells and control cells were treated for 18 h with or without LPS. Silencing efficiency was confirmed by measuring a ADH1B gene expression by RT-qPCR and b ADH1B protein expression by western blot. c IL-6 mRNA level and d IL-6 production, measured by RT-qPCR and singleplex, respectively, are significantly increased in ADH1B silencing group treated with LPS. Cultured fibroblasts were isolated from normal or adenocarcinoma human biopsies and treated for 18 h with or without LPS and in the presence of RO (2.5 µM) or atRA (2.5 µM). RT-qPCR measurements of IL-6 mRNA levels normalized to β-actin in e N-MFs and f CAFs. Both atRA, the most active product of RO metabolism, and RO, the substrate for ADH1B, suppressed LPS-induced IL-6 expression in N-MFs. In contrast, a lack of Adh1b expression in CAFs appeared to prevent the suppression of LPS-induced IL-6 expression by RO, one-way ANOVA analysis, n = 5 per group *p < 0.05; **p < 0.01, ***p < 0.001. g Silencing of ADH1B expression in N-MFs impairs these cells capacity to regulate LPS-inducible IL-6 through RO and reproduced a pro-inflammatory CAFs phenotype. Gene-specific or control siRNA was used to knockdown ADH1B in N-MFs that were then treated with LPS in the presence or absence of RO or atRA. RT-qPCR measurements of IL-6 mRNA levels normalized to β-actin in N-MFs. Two-way ANOVA analysis was performed, n = 7 per group, *p < 0.05.

ADH1B substrate retinol is unable to downregulate LPS-induced IL-6 expression in CAF

ADH1B is a key enzyme in retinoid metabolism which metabolizes retinol (RO) to retinaldehyde, the transient metabolite, which is then metabolized to atRA, a final byproduct of retinol metabolism [44, 45]. Retinoids are known to inhibit tumor-promoting IL-6 production [46], however their use for treatment of solid tumors has yielded controversial results [47]. Thus, next we analyzed the impact of the ADH1B reduction in CAFs on retinoid-mediated inhibition of LPS-inducible IL-6 production and compared this to IL-6 production in N-MFs that exhibit high levels of ADH1B. Cultured MFs were treated for 18 h with the ADH1B substrate RO or its final by-product atRA. We observed that both atRA and RO downregulate LPS-inducible IL-6 expression and secretion in ADH1B^high N-MFs (Fig. 3e). However, while addition of atRA downregulated LPS-induced IL-6 mRNA expression and protein secretion in ADH1B^lowCAFs, no decrease was observed after addition of RO to CAFs (Fig. 3e, f). Finally, silencing of ADH1B by specific siRNA, but not control siRNA in N-MFs resulted in disruption of the RO, but not RA-mediated downregulation of the LPS-inducible IL-6 expression in N-MFs to a similar level observed in CAFs (Fig. 3g).Taken together, these results suggest that a disruption of the vitamin A pathway through the abrogation of ADH1B expression in CAFs contributes to the tumor-promoting inflammation is, at least in part, through alteration of atRA, the negative regulator of IL-6 expression.

Discussion

CAFs are known to be important modulators of tumorigenesis. Being a major cellular component of colon cancer, CAFs promote tumor growth by supporting cancer cell survival, growth, invasion, and cancer stem cell expansion [14, 16, 48–50]. CAFs have a major impact on the generation of a pro-inflammatory milieu [16, 51], where the critical role of IL-6 is well established [16, 52]. We have previously demonstrated that in colon cancer, CAFs are the major source of IL-6 at least in T2–T3 stage tumors [16]. Herein, we further extend this study and demonstrated that the network of genes that regulate IL-6 signalling are dysregulated in CAFs derived from colonic adenocarcinomas. Further we identified ADH1B as an unappreciated suppressor of tumor promoting IL-6 expression.

Gene expression profiling demonstrated that multiple genes are differentially expressed in CAFs when compared to normal stromal cells. Moreover, the expression profile of genes in N-MFs and CAFs from our study correlate with those from public scRNAseq data sets of normal and tumor-associated stroma previously published by Lee at al. [30]. Among the genes implicated in the IL-6 regulatory network and also upregulated in colon cancer CAFs were CHI3L1, which is reported to be involved in the positive regulation of IL-6 in breast carcinoma [34, 53]; TNSF4 (OX40L) known to upregulate IL-6 expression when bind to OX40 receptor [54–57]; LEF1, which is reported to positively regulate IL-6 expression in astrocytes [58]; CADM1 that upregulates NF-κB activity and down-stream IL-6 expression [59–61]. NOTCH3 was also upregulated in CAFs and its interplay with IL6 signalling has been reported to contribute to cancer progression [62].

Among the IL-6 network genes dramatically downregulated in CAFs was CFD, a factor critical to the complement component C3a cleavage in the alternative complement pathway [40, 41]. C3a initially was reported to have a dual (positive and negative) regulatory role on endotoxin-inducible IL-6 production in peripheral mononuclear cells [39]. However more recent reports indicate that C3a is mostly involved in enhancing IL-6 production [63, 64]. Further study is needed to understand the role of CFD-mediated regulation of IL-6 in cancer. CEBPD, also known as NF-IL6, is a nuclear factor also suggested to have a dual regulatory function on the expression of IL-6 [65, 66]. Surprisingly, we also observed downregulation of CXCL12 (also known as SFD-1) and IL-33. Both molecules have previously been suggested to have cancer-promoting role with upregulation of IL-6 [67–70]. However, a decrease of IL-33 was also reported to increase IL-6 in inflammatory pulmonary diseases [71]. Further studies are needed to understand the impact of the downregulation of these molecules on the IL-6 signalling network in CAFs and its role in the development of colon cancer.

Our data also suggest that the ADH1B gene is likely to play an unappreciated role as a suppressor of tumor promoting IL-6. ADH1B belongs to a ADH Class I alcohol dehydrogenases, which catalyze NAD-dependent oxidation of alcohols to aldehydes, including retinol [72, 73]. ADH1B gene mutations and functional polymorphisms have been linked various types of cancer such as oesophageal [74], head and neck [75], colorectal [76], and breast [75–78]. Further, downregulation of ADH1B in colon cancer [79, 80] and ovarian cancer [81] has been reported, However, it is not clear at which stage of the neoplastic transformation the loss of stromal cell ADH1B occurs. Our data suggest that decreased ADH1B expression is specific to neoplastic transformation of colonic tissue since no significant decrease was observed in hyperplastic polyps. Additionally, our data suggests that this ADH1B defect occurs somewhere along the neoplastic transformation of tubular adenoma to carcinoma. Remarkably, recent studies demonstrated a significant role of ADH1B in cellular differentiation [82, 83]. The mechanism of the loss of ADH1B expression in colon cancer, particularly in CAFs is unknown, but there have been suggestions that loss of ADHs in different cancers can occur through transcriptional repression and DNA methylation [84–86].

Our novel findings are that myofibroblasts are the major cellular site of ADH1B expression in normal colonic mucosa, and that its level is dramatically reduced in CAFs. Further, our data demonstrate that ADH1B within MFs acts as a suppressor of LPS-inducible IL-6 expression in N-MFs. Two recent studies [87, 88] suggested that ADHs and downstream aldehyde dehydrogenases (ALDHs) may be among previously unappreciated regulators of IL-6, although these studies did not specifically identify stromal cell ADH1B as the locus of action. Also, overexpression of ADH1C in the epithelial cell line NCM460 results in decreased production of IL-6 induced by a mix of inflammatory cytokines [87] and an increased production of IL-6 was observed in mice lacking ALDH2 in alcoholic liver injury model [88].

The pathophysiological contribution of the decreased ADH1B expression to cancer progression remains unclear. However, it is known that ADH1 dehydrogenases, including ADH1B, convert retinol to retinaldehyde which is in turn metabolized to atRA which is required for several crucial physiological processes [89, 90]. A decrease in the retinol metabolzsing alcohol and downstream aldehyde dehydrogenases was observed in the AOM-DSS murine model of colitis-associated colon cancer, and treatment with the downstream product atRA reduced the tumor burden in this model [91]. A therapeutic anti-tumor effect of atRA has been shown for breast cancer [92], multiple myeloma [93], and acute myeloid leukaemia [94]. While a direct role of ADH1B in the regulation of IL-6 is unknown, ADH1B may convert retinol to retinaldehydes, which are metabolised to atRA, and retinoids are known to negatively regulate IL-6 production [46, 95–97]. Our data suggest that this enzyme is critical to the retinoid-mediated control over the TLR4 mediated increase of IL-6 within MFs in colonic mucosa, and that this control is lost in CAFs. Our data also indicate that normal intestinal fibroblasts are an important source of ADH1B, and thus, retinol metabolism under homoeostasis. This finding gains importance through recent studies by Buechler et al. [98] supporting the ability of Wt1+ stromal cells (fibroblasts) to regulate large cavity-resident macrophage homoeostasis in a contact-independent manner through the metabolism of retinol and the release of atRA. Thus, it is possible that fibroblasts could also regulate macrophages in the normal colonic stroma through retinol metabolism. In the cancer microenvironment, absence of ADH1B expression and conversion of retinol into atRA in CAFs could further drastically disturb macrophage functions and contribute to the increase of tumor promoting inflammation in CRC tumors.

In conclusion, our data demonstrated that in colon cancer a network of genes which regulates IL-6 expression is altered in CAFs favouring the upregulation of tumor promoting cytokines. Our data identified ADH1B as a novel metabolic suppressor of tumor promoting IL-6 overexpression. Decrease/loss of ADH1B in MFs during the adenoma-carcinoma sequence contributes to disruption of the retinol-mediated suppression of tumor promoting IL-6 and to the increase of this cytokine in neoplastic tissue. These findings further strengthen the notion that modulation of the tissue microenvironment by stromal myofibroblasts (CAFs) is an important factor during colorectal carcinogenesis and should be considered in the development of novel, targeted therapies.

Author contributions

RV, MC and RCM: conceptualisation, data acquisition, analysis, manuscript preparation; NSM: data analysis; JT, MS, PJ, JIS, PAA: data acquisition and analysis, clinical material support; BAL: data analysis and manuscript revision for important intellectual content; EJB, DWP and IVP: study concept and design, data analysis, manuscript preparation and revision for important intellectual content, funding acquisition.

Funding

The authors acknowledge grant support from: NIDDK (1R01DK103150 and R56 DK55783-10A1), NCI (1R01CA127229-01A2 NCATS (KL2TR000072 and ILTR000072), NCATS (TR000071), NCI (3R01-CA97959), NCI R01CA207051.

Data availability

Accession code for Bulk RNAseq data is 10.6084/m9.figshare.20685118.v2.

Competing interests

This research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Ethics approval and consent to participate

Not applicable.

Supplementary information

The online version contains supplementary material available at 10.1038/s41416-022-02066-0.