Tumor Budding-Derived CCL5 Recruits Fibroblasts to Promote Colorectal Cancer Progression Through CCR5-SLC25A24 Signaling

Background: Tumor buddings have been included in the routine diagnosis of colorectal cancer (CRC) and considered to be a tumor prognostic factor independent of TNM staging. This study aimed at identifying the contribution of tumor budding-derived C-C chemokine ligand 5 (CCL5) to tumor microenvironment (TME) through broblasts. Methods: Co-cultivation recruitment assays and human cytokine array were used to detect the main cytokine derived from CRC tumor buddings to recruit broblasts. siRNA transfection and inhibitor treatment were applied to investigate the effective receptor of CCL5 on broblasts. Transcriptome sequencing was performed to explore the mechanism inside broblasts when stimulated by CCL5. Stimulation with CCL5 in vitro, orthotopic xenograft mouse model and clinical specimens were designed to clarify the contribution of CCL5 to angiogenesis and collagen synthesis. Results: H&E and immunochemistry staining conrmed that CRC with tumor buddings in the invasive front was accompanied by more broblasts compared with CRC without tumor buddings. Further vitro study indicated that CCL5 derived from tumor buddings could recruit broblasts through CCR5 receptor on broblasts and positively regulate solute carrier family 25 member 24 (SLC25A24) expression in broblasts, which could activate pAkt-pmTOR signaling. Moreover, CCL5 can increase the number of α-SMA + CD90 + FAP - broblasts to promote tumor angiogenesis through enhancing the expression of VEGFA and making broblasts transdifferentiate into vascular endothelial cells. Meanwhile, CCL5 also can promote collagen synthesis through broblasts, thus contributing to tumor progression. Conclusions: In the invasive front of CRC, tumor-budding-derived CCL5 can recruit broblasts through CCR5-SLC25A24 signaling, further promoting angiogenesis and collagen synthesis through broblasts, eventually creating a tumor-promoting microenvironment. percent positivity Red staining 0 0 (0-5%), 1 (6–25%), 2 (26–50%), 3 (51–75%), and 4 (>75%). The staining intensity was scored on a 4-point scale: 0 (no staining), 1 (weak staining, light orange or 2 (moderate staining, medium orange or green), and 3 (strong staining, orange or green). Subsequently, the Sirius Red staining score was calculated through multiplying the percent positivity score and staining intensity Accordingly, the level of Sirius Red staining was dened as Low (0–4), Medium (5–8), and High (9–12).


Background
The relationship between tumor and microenvironment is like seed relying on soil. Recent studies have gradually concentrated on TME rather than only tumor itself. Cancer-associated broblasts (CAFs), as an indispensable part of tumor stroma, have attracted more attention and been widely concerned by scholars [1][2][3]. Evidences from clinical and basic researches have shown a strong association between CAFs and poor prognosis in several types of cancer, including breast cancer [4], uterine cervical cancer [5], lung cancer [6], cholangiocarcinoma [7] and CRC [8,9]. Concretely, the mutually crosstalk between broblasts and tumor can activate broblasts, which leading to tumor metastasis [8], therapy resistance [10], and immunosuppression [11]. However, the process of how broblasts initially become a tumorpromoting phenotype is still unclear.
Tumor budding is de ned as a single tumor cell or a cell cluster of up to 4 tumor cells, which is assessed in one hotspot (in a eld measuring 0.785 mm2) at the invasive front [12,13]. Moreover, there have been studies supporting a close link between tumor budding and a distinctive immune-suppressive 2h, the tissues were passed through 40-µm cell strainer to generate single-cell suspensions. Centrifuge the single cell suspensions at 1200 rpm/min for 12 min, discard the supernatant, and resuspend the pellet with medium. Primary cells were then plated at a density of 1 × 105 viable cells in 25-cm2 adherent asks in EMEM medium (ATCC, USA) with 10% FBS (Gibco, USA) in 5% CO2 at 37°C (trypsin digestion method). The tissues couldn't pass through the strainer were transferred to the 25-cm2 adherent asks.
Add EMEM medium (ATCC, USA) with 10% FBS (Gibco, USA) into the adherent asks after 24h when the tissue stuck to the bottom of the adherent ask and waiting for broblasts to crawl out of the tissues (improved tissue planting method).
Clinical specimens CRC clinical specimens were obtained from patients who were pathologically diagnosed with CRC at Shenzhen Hospital, Southern Medical University. The study was approved by the ethics committee of Shenzhen Hospital, Southern Medical University, China.
Co-culture recruitment assay Boyden Transwell chambers (Corning, 353097) were used following the instructions of the manufacturer. Brie y, 2 × 104 broblasts were incubated into the upper chambers. 1× 105 tumor cells or different concentration of CCL5 cytokine (Peprotech, 300-06-20) were incubated into the lower chamber. After 48h incubation, cells that successfully migrated were xed with 4% paraformaldehyde, stained with Hematoxylin and counted in 5 random visual elds using a light microscope.

Construction of stable cell lines
Lentivirus vector carrying the luciferase gene (Luc) and the human CCL5 sequence were obtained from GenePharma (Shanghai, China). An empty vector was used as control to CCL5 overexpression. According to the manufacturer's instructions, stable cell lines were established by transfection of LS174T and RKO with these lentivirus vectors.
Enzyme linked immunosorbent assay (ELISA) CCL5 levels in the conditioned medium (CM) of FHC and CRC tumor cells were measured by ELISA using a commercially available kit (Huamei Biotech, China), as described by the manufacturer's instructions. The results were expressed in pg/ml, and the standard curve is based on the measured OD value of the standard product.
RNA extraction and quantitative reverse transcription polymerase chain reaction (qRT-PCR) Total cellular mRNA was extracted using TRIzol (TaKaRa). Prime-Script RT Reagent Kit with gDNA Eraser (TaKaRa) was used to reverse-transcribed mRNA into cDNA. And SYBR Premix Ex Taq (TaKaRa) was used for qRT-PCR.

Immunoblot
Total proteins were isolated from cells using RIPA lysis buffer (FDbio, FD008),PMSF (FDbio, FD0100), Protease inhibitors (FDbio, FD1001) and protein phosphatase inhibitors (FDbio, FD1002). The concentration was determined using BCA protein assay kits (FDbio, FD2001). Total proteins were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene uoride (PVDF) membrane. The membranes were blocked with 5% skimmed milk or 5% BSA for 1h at room temperature and then incubated with primary antibodies at 4°C overnight. Subsequently, the membranes were incubated with goat anti-rabbit or anti-mouse secondary antibody (FDbio, FDR007 and FDM007). The protein was detected by ECL chemiluminescence solution (FDbio, FD8030) and visualized using chemiluminescence detection system (Universal Hood II, Bio-Rad). The intensity of the immunoblot bands was quanti ed with the NIH ImageJ software.

Immuno uorescence of cells and CRC tissue
Immuno uorescence of cells was performed, as previously described [23]. The steps of immuno uorescence of CRC tissue before incubation of primary antibody were the same as IHC. The following steps after incubation of primary antibody were the same as those of immuno uorescence of cells. Images were captured using the uorescence inverted microscope.
Orthotopic xenograft colorectal cancer mouse model BALB/C-nude mice (male, 3-5 weeks old) were purchased from the Gem Pharmatech Co., Ltd, Guangdong, China. All animal experiments were approved by the ethics committee of Shenzhen Hospital, Southern Medical University, China. 5X106 HCT116 cells were injected subcutaneously into the back of nude mice (n=3). After subcutaneous tumor formation, one of the tumor tissues was xed and stained with hematoxylin-eosin (H&E), and the remaining tumor tissues were peeled off and cut into 1mm3 tumor tissue pieces with ophthalmic scissors. Choose 5 pieces from the cut tissue pieces and bury them in the cecal serosal layer of nude mice by purse string suture (n=5). After 8 weeks, the ceca of nude mice were surgically removed after euthanasia, xed in 10% formalin, embedded in para n, and prepared into 2.5µm sections for H&E staining.
Human cytokine array Serum-free CMs from FHC and CRC tumor cells HCT-8, HCT116, HCT-15, SW620 were collected after incubation for 24h and ltered through a 0.22-µm mesh. The CM samples were added to antibody arrays against 1000 unique cytokines (RayBio, GSH-CAA-X00) and processed according to the manufacturer's protocol.

Transcriptome sequencing
Total mRNA was extracted from broblasts before and after CCL5 stimulation. Sequencing libraries were generated using NEBNext® UltraTM RNA Library Prep Kit for Illumina® (NEB, USA) following manufacturer's recommendations and index codes were added to attribute sequences to each sample.
The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumia) according to the manufacturer's instructions.

Bioinformatics analysis of relapse-free survival of CRC patients
The CRC microarray pro les GSE39582 was used to analyze the correlation between the expression of COL1 or COL3 and the relapse-free survival of patients. The chip platform used in this analysis was the Affymetrix Human Genome U133 Plus 2.0 Array. The mRNA values of COL1 or COL3 were divided into low and high expression groups according to the median. Thus, each group had mRNA values for 283 cases. Then, the survival curves of the two groups were obtained using the Kaplan-Meier method.

Matrigel angiogenesis experiment
First, high concentration Matrigel (BD, 354248) was diluted in half. 10µl diluted Matrigel was pipetted into each well of µ-Slide Angiogenesis Glass Bottom (Ibidi, 81506) and was allowed to polymerize for 30 minutes at 37 ℃. 1X104 broblasts before or after CCL5 stimulation for 24h with 50ul medium were incubated on high concentration Matrigel. After 2h, the broblasts were xed with 4% paraformaldehyde, stained with Hematoxylin and counted in 3 random visual elds using a light microscope. The capillary tubes were quanti ed by counting the number of lumens.
Statistical analysis SPSS software for Mac OS version 25.0 was used for statistical analyses. An unpaired two-tailed Student's t test was used to compare normally distributed data in two groups. Pearson's χ2 test and Spearman's correlation test were applied to analyze the correlation between the expression of CCL5 or Sirius Red staining and clinicopathological features. Log-rank test was performed in Kaplan-Meier survival curves. The correlated expression levels of CCL5 and Sirius Red staining in CRC tissue were analyzed by Spearman's correlation test. All data were expressed as the mean ± standard error of mean (SEM). P < 0.05 was considered signi cant. ns, no signi cance; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

CRC tumor cells can recruit broblasts around tumor buddings
Through the clinical observation of 195 cases of CRC tissue with H&E staining, we found that in the invasive front CRC tissue with tumor buddings was accompanied by more broblasts compared with CRC tissue without tumor buddings (Fig. 1a). Moreover, we veri ed broblasts around tumor buddings were highly expressed with α-SMA and CD90, but without the expression of FAP (Fig. 1b). It suggested that tumor buddings in the invasive front of CRC tissue were closely related to the heterogeneity of broblasts.
Recently, many studies have revealed that broblasts can be recruited by some cytokines secreted by tumor cells or other cells in the TME [7,[24][25][26]. To study the contribution of CRC tumor cells to broblasts, rstly, two types of broblasts were used, including normal colorectal broblast cell line CCD-18Co and primary normal colorectal broblast NF1. Primary normal colorectal broblasts were extracted by trypsin digestion or improved tissue planting method (Additional le 1: Figure S1A). To identify these broblasts, epithelial marker E-cadherin and broblast markers Vimentin and α-SMA were detected by immuno uorescence (Fig. 1c and Additional le 1: Figure S1B).
Secondly, co-culture recruitment assay was conducted, in which CRC tumor cells were incubated in the lower chamber while the broblasts were incubated in the upper chamber. The pore size of the co-culture upper chamber is 8µm, which is enough for broblasts to pass through (Fig. 1d). As shown in Fig. 1e, normal colorectal epithelial cell line FHC as well as CRC cell lines LS174T, RKO, DLD-1 and Caco2 had weak recruitment ability towards broblasts, while HCT-8, HCT116, HCT-15 and SW620 had stronger recruitment ability towards broblasts.
Next, we chose HCT116 for further vivo orthotopic xenograft colorectal cancer mouse experiments, which had stronger recruitment ability towards broblasts in vitro. Through the observation of H&E stained sections of tumors on mice cecum, we found that HCT116 also could recruit broblasts in vivo (Fig. 1f). Taken together, these data demonstrated that broblasts around tumor buddings could be recruited by CRC tumor cells.

CRC tumor cells in tumor buddings can recruit broblasts through CCL5
To further ascertain the key cytokine secreted by CRC tumor cells for broblasts recruitment, the CMs of FHC and 4 CRC tumor cell lines HCT-8, HCT116, HCT-15, SW620, which owned stronger recruitment ability to broblast, were detected on a human cytokine array (Fig. 2a). GO function enrichment analysis indicated that differential proteins found above were involved in various activities, including "positive regulation of cell migration" (Fig. 2b) and "extracellular matrix"(Additional le 2: Figure S2A), which suggested that cytokines secreted by CRC tumor cells could recruit other cells in tumor environment and might be correlated with broblasts since broblasts are the main source of extracellular matrix [1].
Among these proteins, CCL5 was the most up-regulated protein in the CMs of CRC tumor cells compared with FHC (Fig. 1c). Then we veri ed the mRNA and protein secretion levels of CCL5 in FHC and 8 CRC tumor cells, the results of which were entirely consistent with the initial ndings in the co-culture recruitment assay (Fig. 2d, e). Besides, CCD-18Co and primary normal colorectal broblasts secreted lower level of CCL5 than FHC (Fig. 2f). Accordingly, it was paracrine not autocrine of CCL5 that recruited broblasts. Next, we detected the expression of CCL5 in CRC tissues using IHC and found that CCL5 was highly expressed in tumor buddings among the invasive front (Fig. 2g).
Then, to detect whether CCL5 is the key cytokine in recruiting broblasts, we used the co-culture recruitment assay shown as Fig. 2h. Among different concentrations of CCL5, 40ng/ml CCL5 had the strongest recruitment ability towards broblasts (Additional le 2: Figure S2B) and this concentration of CCL5 generally had the ability to recruit CCD-18Co and human colorectal primary broblasts ( Fig. 2i and Additional le 2: Figure S2C).
Moreover, to further ascertain whether CCL5 that could recruit broblasts was secreted by CRC tumor cells, we constructed CCL5 down-regulating cell lines with siRNA, and constructed CCL5 overexpressing cell lines with lentivirus, and veri ed the CCL5 mRNA level and protein secretion level for these cell lines (Additional le 2: Figure S2D-G). Then, using the co-culture recruitment assay shown in Fig. 2j, we found that the ability of tumor cells to recruit broblasts was signi cantly weakened after CCL5 interference ( Fig. 2k and Additional le 2: Figure S2H), and the same ability was increased after CCL5 overexpression ( Fig. 2l and Additional le 2: Figure S2I). In this context, the data pointed out that CCL5 was highly expressed in CRC tumor buddings and can recruit broblasts from TME.

CCL5 recruits broblasts through CCR5 receptor
It has been well de ned that CCR1, CCR3, CCR4, CCR5, CD44 and GPR75 are probable receptors of CCL5 [18,19]. Accordingly, to explore whether CCL5 mediated broblasts recruitment through these receptors, rstly, we used siRNA to interfere with them in broblasts respectively and veri ed their interference e ciency (Fig. 3a, b). Next, we conducted the co-culture recruitment assay described as Fig. 3c, and surprisingly found that only downregulation of CCR5 in broblasts can signi cantly reverse the recruitment ability of CCL5 towards broblasts (Fig. 3d). Moreover, CCR1 inhibitor BX471 or CCR5 inhibitor Maraviroc were used to treat with broblasts respectively for blocking CCR1 or CCR5 before and during co-culture recruitment assay (Fig. 2e). Similarly, results showed that only Maraviroc treatment can signi cantly weakened the recruitment ability of CCL5 to broblasts (Fig. 3f). The above results strongly indicated that CCL5 participated in broblasts recruitment through CCR5 receptor.

CCL5 recruits broblasts through SLC25A24 inside broblasts
In order to explore the intracellular changes of broblasts after CCL5 stimulation, we performed transcriptome sequencing of broblasts before and after CCL5 stimulation (Fig. 4a). The top 20 genes were selected and veri ed in CCD-18Co cell lines and 4 primary broblasts with CCL5 stimulation (Additional le 3: Figure S3A), among which SLC25A24 was consistently up-regulated. We subsequently detected that the mRNA level of SLC25A24 was up-regulated in more broblasts with CCL5 stimulation (Fig. 4b). SLC25A24,also known as ATP-Mg2 + /phosphate carrier 1 (APC1), has a regulatory N-terminal domain containing EF-hand Ca2 + binding sites which can activate transport during cytosolic Ca2 + increases [27][28][29]. To preliminarily detect the protein expression and location of SLC25A24 in CRC tissue, we conducted immuno uorescence staining and found that SLC25A24 was highly expressed in the broblasts surrounding the tumor buddings in the invasive front (Fig. 4c). Then, we selected three broblasts to further verify that CCL5 can up-regulate the protein expression of SLC25A24 (Additional le 3: Figure S3B). Functionally, to study whether CCL5 recruited broblasts is dependent on SLC25A24, siRNAs of SLC25A25 were used in broblasts, and followed by CCL5 recruitment assay (Fig. 4d). Results showed that silencing of SLC25A24 in broblasts signi cantly reduced the recruitment ability of CCL5 (Fig. 4e-g and Additional le 3: Figure S3C). Moreover, immunoblot results demonstrated that the increased expression of SLC25A24 in broblasts after CCL5 stimulation relied on CCR5, indicating that CCL5 recruiting broblasts was SLC25A24-CCR5 dependent (Additional le 3: Figure S3D).
To further ascertain the pathway activated by CCL5-SLC25A24 in broblasts,KEGG signaling pathway enrichment analysis of above transcriptome sequencing was performed which identi ed that PI3K-Akt signaling pathway was the most relevant after CCL5 stimulation (Fig. 4h). Immunoblot results showed that CCL5 could positively regulate the phosphorylation of Akt and mTOR in broblasts (Fig. 4i, left  panel). Whereas silencing SLC25A24, phosphorylated Akt and mTOR decreased in broblasts even with CCL5 stimulation (Fig. 4i, right panel). These ndings suggested that CCL5 recruited broblasts through SLC25A24-pAkt-pmTOR inside broblasts. CCL5 contributes to the increase of α-SMA + CD90 + FAP − broblasts to promote tumor angiogenesis The above results in Fig. 1b proved that broblasts around tumor buddings were speci cally expressed with α-SMA + CD90 + FAP − . To further ascertain whether CCL5 is the main reason accounting for the increase of α-SMA + CD90 + FAP − broblasts, we performed immuno uorescence and found that the expression of α-SMA and CD90 were increased in broblasts with the stimulation of CCL5 (Fig. 5a).
VEGFA plays an irreplaceable role in coordinating vascular endothelial cells proliferation/survival, migration and invasion into the surrounding tissue, leading to formation of lumen-containing structures [30] and it has been reported that broblasts are the main source of VEGFA other than tumor cells [2]. Furthermore,studies also have shown that broblasts can transdifferentiate into vascular endothelial cells to facilitate tumor progression [31][32][33]. Next, to detect the function of CCL5 on angiogenesis through broblasts, immunoblots showed the expression of vascular endothelial markers FLI1, VE-cadherin, CD31 and VEGFA as well as the expression of α-SMA and CD90 were increased in broblasts stimulated with CCL5 (Fig. 5b). Functionally, we further found that the angiogenesis ability of broblasts was enhanced with CCL5 stimulation (Fig. 5c). Clinically, IHC staining was performed on serial sections from the same CRC tissue with 17 patients in total to detect the expression of CCL5, α-SMA, CD90, FAP and CD31, and found that in the invasive front, CCL5 was highly expressed in the tumor buddings, which were surrounded by accumulated α-SMA + CD90 + FAP − broblasts as well as increased angiogenesis (Fig. 5d).
In this context, we pointed out that tumor buddings secreted CCL5 to increase the number of α-SMA + CD90 + FAP − broblasts to promote tumor angiogenesis through increasing the expression of VEGFA and letting broblasts transdifferentiate into vascular endothelial cells.
CCL5 promotes collagen synthesis through broblasts contributing to tumor progression GO function enrichment analysis performed on above transcriptome sequencing indicated that CCL5 was also relevant to enrichment of extracellular matrix (Fig. 6a). It is well known that collagen type I (COL1) and collagen type III (COL3) are the main components of extracellular matrix and they are mainly synthesized and secreted by broblasts [2]. To clarify the function of CCL5 on collagen synthesis, we performed immunoblots which demonstrated that CCL5 could increase protein expression of COL1 and COL3 in broblasts in vitro (Fig. 6b).
Clinically, to further investigate the role of CCL5 plays in CRC, we detected the expression of CCL5 protein in 195 para n-embedded CRC tissues and 162 adjacent normal colorectal tissues. Results showed that CCL5 was signi cantly higher expressed in CRC tissues than normal tissues ( Fig. 6c and Additional le 4: Figure S4A). The correlation between CCL5 expression level and CRC clinical features was analyzed (Table 1), and further Spearman's correlation test showed that high expression of CCL5 was positively associated with high risk of the deep tumor invasion (r=0.244, Fig. 6d), lymph node metastasis (r=0.237, Additional le 4: Figure S4C), the presence of peri-intestinal cancer nodules (r=0.198, Additional le 4: Figure S4D) and advanced TNM stages (r=0.256, Additional le 4: Figure S4E). Moreover, to detect collagen distribution in 88 of the above 195 CRC tissues, Sirius Red staining of COL1 (reddish) and COL3 (greenish) was applied. As shown in Fig. 6e and Additional le 4: Figure S4B, the expression of COL1 and COL3 were increased in CRC tissue. The correlation between Sirius Red staining and CRC clinical features was also analyzed ( Table 2), and further Spearman's correlation test showed that its high distribution surrounding CRC tumor cells was also positively related to deep tumor invasion (r=0.431, Fig. 6f), lymph node metastasis (r=0.351, Additional le 4: Figure S4F), the presence of peri-intestinal cancer nodules (r=0.288, Additional le 4: Figure S4G) and advanced TNM stages (r=0.442, Additional le 4: Figure S4H). Intriguingly, we observed that high expression of CCL5 in tumor buddings in the invasive front was often accompanied by increased synthesis of collagen (Fig. 6g). Additionally, Spearman's correlation analyses revealed positive correlation between the expression levels of CCL5 in tumor cells and the collagen in surrounding area in serial sections from the same CRC tissue with 88 patients in total (r=0.317, Fig. 6h).
In vivo, we used HCT116, the CCL5 highly secreted tumor cell line, to establish orthotopic xenograft CRC mouse model and found that collagen synthesis was increased around the tumor cells in the invasive front (Fig. 6i). Furthermore, we used the CRC microarray pro les GSE39582 to analyze the relapse-free survival of COL1 and COL3 and found that higher expression of COL1 and COL3 tightly related to the poorer prognosis of CRC patients (Fig. 6j). Taken together, these data suggested that tumor buddings secreted CCL5 to promote collagen synthesis through broblasts, thus contributing to tumor progression.
Discussion TME is widely implicated in tumorigenesis because it harbors cancer cells that interacts with surrounding cells to foster the evolution of cancer [34,35]. Inside this microenvironment, broblasts have a strong tumor-modulating effect closely related to poor prognosis and recurrence of patients [36][37][38][39]. Despite their vital roles in CRC carcinogenesis, there is lack of speci c markers to de ne the heterogeneous of highly complicated broblasts populations. In our research, we found that CRC tumor buddings could highly secret CCL5, which could recruit broblasts through CCR5-SLC25A24 signaling and form a characteristic broblasts cluster with α-SMA + CD90 + FAP − expression located around the tumor buddings in the invasive front, furthermore facilitate tumor angiogenesis and collagen synthesis (Fig. 6k).
Tumor cells are heterogeneous, and tumor buddings consist of the most aggressive subgroups of tumor cells, which play a leading role in the process of tumor invasion [12,13]. However, there is no unique marker for tumor buddings, and it is unreliable to count them only by means of H&E staining when tumor cells are di cult to distinguish from reactive mesenchymal cells. In our research, CCL5 was found as a marker for tumor buddings which provides a concretely new evidence for the role of tumor budding in tumor progression. Consequently, the pathological diagnosis of tumor buddings in CRC can be more accurately diagnosed with the expression of CCL5. In addition, studies have shown that the high expression of CCL5 in colorectal cancer tumor cells can promote their proliferation [40]. The high secretion of CCL5 in colorectal cancer can also promote the apoptosis of CD8 + T cells through Tregulatory cells, thereby promoting tumor progression through immunosuppression [22]. Our research makes the role of CCL5 in the progression of CRC more comprehensive and also makes the network of tumor-microenvironment interaction in CRC clearer. Taken together, therapies targeting CCL5 may play a signi cant role in blocking the progression of colorectal cancer.
The heterogeneity of broblasts is also worth of attention. We have found a subgroup of broblasts induced by CCL5 with the biomarkers of α-SMA + CD90 + FAP − , which is mainly located around the tumor buddings. It contributes to multiple tumor-promoting effects, including tumor angiogenesis and collagen synthesis, which provides a new mechanism for tumor budding in tumor progression. It has been found in literature that the circulating tumor cells were detected in the circulation system together with broblasts [41][42][43]. It appears that the in ltration and metastasis of tumor-broblast clusters has a greater effect on the prognosis of tumor patients, and the tumor progression might be inhibited by blocking invasive tumor cells together with surrounding broblasts. And our research provides another strong evidence for the existence of congregated in ltration and metastasis of tumor cells and broblasts. In our research, tumor buddings can recruit broblasts through CCR5 receptor in the process of invasion and it is a vital step for further angiogenesis and collagen synthesis. CCR5 inhibitor Maraviroc, as a therapeutic drug for HIV, has been relatively mature in clinical treatment [44][45][46]. There is also study demonstrated that tumoral immune cells can be targeted effectively in CRC metastasis by anti-CCR5 therapy in cancer patients [47]. Consequently, CCR5 inhibitor can not only target immune cells but also broblasts in the invasive front in CRC, suggesting that it is very suitable for the treatment of CRC. Despite all these, further researches are also needed.
The SLC25 carrier family comprises 53 members, most members of which transport solutes across the inner mitochondrial membrane as part of a variety of distinct metabolic processes [48]. SLC25A24 belongs to a subgroup of short calcium-binding mitochondrial carriers (SCaMCs) with four paralogs in mammals: SCaMC-3/SLC25A23, SCaMC-1/SLC25A24, SCaMC-2/SLC25A25, and SCaMC-3like/SLC25A41 [49-51]. These carriers consist of a C-terminal domain containing six transmembrane helices homologous to the mitochondrial carrier proteins and an N-terminal domain with Ca2 + -binding EF hands conferring Ca2 + sensitivity to the carrier. As a mitochondrial inner membrane protein, SLC25A24 has been reported to be involved in the uptake and accumulation of adenine nucleotides [52]. In addition, SLC25A24 has also been found to be related to anti-oxidative stress in previous studies [53]. In this study, we found that SLC25A24 was highly expressed in broblasts in the invasive front which surrounded tumor buddings. The impact of these SLC25A24 + broblasts on tumor cells in turn is still unclear and remains to be explored.

Conclusions
In conclusion, our study suggested that in the invasive front of CRC, tumor-budding-derived CCL5 can recruit broblasts through CCR5-SLC25A24 signaling, further promote angiogenesis and collagen

Consent for publication
All authors have consented to publication of the results presented in this manuscript.

Availability of data and materials
All data presented or analyzed in this study are included either in this article or in the additional les.

Competing interests
The authors have no con icts of interest to declare.