Dysbiosis of skin microbiome and gut microbiome in melanoma progression

Cutaneous Melanoma (CM) is a malignant skin cancer originating from epidermal melanocytes [1, 2]. Even though it is less common than other skin cancers, CM is more lethal due to its high metastatic potential [2–4]. Considering the aggressiveness of this disease, it is important to identify the risk factors associated with melanoma development to improve diagnosis and treatment methods of this serious skin cancer. Different risk factors are associated with melanoma development: besides the genetic predisposition, such as the familial history of melanoma or other skin cancers and type of melanocytic nevi, other environmental factors, particularly the sun and UV radiation exposure, increase the risk of melanoma development [4–6].

Recently, the host microbiome is considered as a new component of the tumour microenvironment that influences tumour cell metabolism and plays a role in the cancer pathogenesis and treatment response [7, 8]. The commensal microbes interact directly with the cancerous cells of the tumour tissue, in which they are residing. Indirect effects could occur when the tumour development is affected by the metabolites of the microbiome from another location or by the administration of probiotics in the host diet [9].

Many observations suggested that the skin commensal microbiome may promote skin immunity and confer the host defence including the protection against skin inflammatory disorders, infections, wounds and skin cancer [10–19]. Several studies have suggested a potential role of the microbiota in skin carcinogenesis [20]. A reduced rate of skin cancer was observed in germ-free rats [21]. Similarly, different microorganism-associated molecular patterns (MAMPs) were identified as the trigger of receptors responsible for the inflammatory response that leads to tumour development, while skin inflammation in response to tumour promoters was reduced in mice lacking receptors [22–24].

Several studies have investigated the gut-skin connection in various skin diseases including skin cancer [25–27]. The gut-skin axis indicates that the gut microbiota and its metabolites can have a critical role in the development or prevention of skin cancers, including melanoma. In human, gut microbiota and its metabolites may have a mechanistic impact on antitumor immunity and immunotherapy in patients with advanced melanoma [28–37]. Also, it has been reported that the administration of selected strains from commensal intestinal microbiota may establish anti-tumour immunity and restrict melanoma growth in germ-free WT mice [38].

In order to investigate changes in the gut and skin microbiota composition during melanoma development, swine models are highly suitable due to the high similarities with human in terms of skin and gastrointestinal anatomy and physiology, genetics, immunology and pathophysiology of many human diseases [39–41]. The Melanoma-bearing Libechov Minipig (MeLiM) is a unique large animal model of hereditary melanoma [42]. In the MeLiM strain, the melanomas occur only in animals with pigmented skin, as the white pigs lack melanocytes in their skin. The MeLiM piglets are born with several nodular melanomas or the melanomas develop postnatally. The melanomas mostly invade into deep dermis and subcutaneous fat (corresponding to Clark level IV to V in human staging). At the age of 8 to 12 weeks, the spontaneous regression of melanomas starts to occur in the majority of animals, which is characterized by lymphocyte infiltration, and flattering and colour fading of tumours due to the replacement of the tumour by fibrous tissue [43–46]. However, in approximately 20% of piglets, the melanoma progression develops which is characterized by tumour growth and spread by metastases [47], mainly into the lymph nodes and lungs, cachexia and animal growth retardation. The MeLiM model enables to study the melanoma spontaneous development without any therapeutic interventions, which is not possible in human.

The aim of the present study was to assess the association between the diversity, composition and function of the skin and faecal microbiome and melanoma development and to compare the bacterial composition in such entities using high throughput sequencing. The samples were collected from multiple sites (inner melanoma tissue, melanoma surface, healthy skin and stool) of experimental piglets at different ages throughout melanoma progression and melanoma spontaneous regression (experimental scheme is presented in Fig. S). Findings could contribute to the characterization of skin and gut microbiome composition and modification, as well as functional mechanisms following melanoma progression. 1

Skin and stool microbiome samples were collected from MeLiM piglets with melanoma progression (n = 10) and spontaneous regression (n = 10), as well as from crossbreds of MeLiM and white strain with black skin and several melanomas with regressive disease course (n = 4) and were compared to control MeLiM x white crossbreds with melanoma-free white skin (n = 10).

A total of 13,747,282 sequences were obtained from different samples. The mean sequence length was 275 bp. The metagenomic analysis of each microbiota (faecal and cutaneous) was performed separately.

Several studies have reported the association of commensal microbiota of human or animal models with cancers, mostly focused on colorectal cancer [48–51], in addition to gastric [52, 53], liver [54], pancreatic [55, 56], lung [57–59], breast [60, 61] and bladder [62, 63] cancers. Generally, many reports suggested that microbiota induces carcinogenesis and other ones supported that microbiota play protective roles against cancer [64–67]. Recent reviews focused on the importance of the microbiome in skin cancer research and explored the crosstalk between the immune system and the skin microbiota in health and diseases (including cancer) [6, 11, 12, 68]. The profound reliance of the skin immune system on its resident microbiota for both host defence and tissue repair led to the integration of the skin microbiota and its metabolites as an intrinsic regulator of immune responses in the tissue microenvironments [13, 69–71]. The interactions between skin immune cells and microbiota are not only within the local microenvironment but also the skin immune system was stimulated by metabolites of microbes from other body sites (e.g. gastrointestinal tract) [9]. The gut microbiota is involved in cancer and is associated with anticancer therapy efficacy. Recently, several studies illustrated that the gut microbiota and the treatment with faecal microbiota transplantation (FMT) promoted the responses to anti-PD-1 immunotherapy and restored the tumour microenvironment in patients with advanced melanoma [28–30, 32, 34, 35, 72, 73].

In this study, the association between melanoma development and the changes in the bacterial composition of the gut and the skin microbiome were explored. The samples were collected from melanoma tissue, melanoma surface and healthy skin of porcine models: MeLiM piglets with melanoma progression or melanoma spontaneous regression and crossbred piglets with melanoma spontaneous regression or healthy (white) controls. Pigs are valuable large animal model due to similar anatomy and physiology, including metabolism and nutritional requirements to human. Human and pigs have been previously shown to share the major bacterial phyla (Firmicutes and Bacteroidetes) in both their gut and skin microbiome. Nonetheless, differences between human and pig have been found at the bacteria species level, which are expected to be caused mainly by the environmental factors, nutrition and age [74–76]. The pig breeding under uniform conditions enables to minimize the influence of environment and nutrition on the microbiome.

There are few studies about the link between skin cancer and skin microbiota in skin diseases [77–80]. The first work which studied the relationships between the human skin microbiome and melanoma has shown that the skin microbiome diversity was lower in patients with melanoma than in patients with melanocytic nevi. However, the authors did not detect the association between the cutaneous microbiome and melanoma [81]. Recently, the significant association of the skin microbiome in patients with acral melanoma was investigated at different stages [82]. In our previous study, we showed a significant difference between the healthy skin microbiome and melanoma tissue microbiome using DGGE method [83]. Similarly, in the present study, using high throughput sequencing of the 16Sr RNA gene, the metagenomic analysis revealed differences in skin microbiome of healthy skin, melanoma surface and melanoma tissue. Alpha diversity showed that the diversity in the healthy skin microbiome was significantly higher than in the melanoma microbiome. Also, the diversity of skin microbiome and faecal microbiome was significantly higher in crossbred white piglets (control animals without tumours) than in MeLiM piglets. The high diversity and richness of host microbiota are generally related to the host health stat [84–86]. Moreover, beta diversity based on Bray Curtis distance showed dissimilarities between microbial communities from multiple sites. The differences between regressive melanoma microbiome and progressive melanoma microbiome were not significant. However, a dynamic shift in the gut and cutaneous microbiome was explored in piglets with melanoma regression following the age, while the microbiota diversity was stable in MeLiM piglets with melanoma progression. That indicates that both the cutaneous microbiome and intestinal microbiome have a strong correlation with the melanoma process by age.

The dominant bacterial phyla in porcine skin microbiota were Firmicutes, Actinobacteria, Proteobacteria and Bacteroidetes in addition to Fusobacteria, which was abundant in the melanoma tissue microbiome of MeLiM piglets. The porcine faecal microbiota was mainly dominated by Firmicutes and Bacteroidetes, besides Fusobacteria, Proteobacteria and Actinobacteria, which were abundant in the faecal microbiota in MeLiM piglets. At the genus level, Lactobacillus, Clostridium, Corynebacterium1, Terrisporobacter and Enterococcus were associated with healthy skin of crossbred piglets whereas the relative abundance of Fusobacterium, Staphylococcus, Trueperella, Streptococcus and Bacteroides were discriminately higher in the melanoma microbiome. Consistent with our previous findings [86], Fusobacterium necrophorum subsp. necrophorum (18,3%) was the most abundant species in melanoma tissue microbiome of MeLiM piglets besides Staphylococcus hyicus (8,3%), Trueperella pyogenes (7,1%) Streptococcus (uncultured bacterium) (4,3%) and Staphylococcus chromogenes (2,7%), while they were considerably low or absents in the healthy skin microbiome. The relative abundance of Lachnospiraceae (9,7%), Bacteroides (4,2%), Escherichia-Shigella (2,4%) and Fusobacterium (1,6%) (Fusobacterium necrophorum subsp. necrophorum and Fusobacterium gastrosuis) were significantly discriminant bacterial genera in the faecal microbiota of MeLiM piglets, while Prevotella 9 (8,3%), Prevotellaceae NK3B31 group (4,4%), Lactobacillus (5,8%) and Feacalibacterium (1,9%) were the most discriminant bacterial genera in the faecal microbiota of crossbred piglets.

Fusobacterium was associated with pathogenesis in both human and livestock infections [87, 88]. Two subspecies of F. necrophorum are recognized. The subsp. necrophorum is more frequently present animal infections, while subsp. funduliforme was isolated from clinical human infections and their virulence is determined by secreting leukotoxin. In humans, F. necrophorum is responsible for Lemierre syndrome, which begins as bacterial pharyngitis and rapidly progresses to septic thrombophlebitis of the jugular vein [87–90]. Fusobacterium nucleatum was enriched in colorectal carcinoma. It was considered as a risk factor for the progression and severity of pancreatic and colorectal cancers [91, 92]. The active invader species F. nucleatum and F. periodonticum can independently invade host cells. Fusobacterium nucleatum colonized breast cancer tissue and colorectal cancer tissue, promoted tumour growth and caused cancer progression by inducing immunosuppression using extracellular adhesion and invasion molecule Fusobacterium adhesion (FadA) [93–98]. Gur et al. have demonstrated that the direct interaction of the Fap2 protein of F. nucleatum with the immune cells inhibitory receptor TIGIT (T cell immunoglobulin and ITIM domain) protected melanoma tumours bounded with F. nucleatum from NK cell cytotoxicity and T-cell activity [98]. Kalora et al. have identified 11 HLA-bound peptides derived from F. nucleatum, Staphylococcus aureus and Staphylococcus capitis inside melanoma tumour cells that elicited the immune response [99, 100].

Staphylococcus hyicus is the major causative agent of piglets’ exudative epidermitis [101]. Staphylococcus chromogenes has been identified as a frequent cause of bovine mastitis and intramammary infections [102, 103]. Interestingly, colonization of mice with Staphylococcus epidermidis, a skin commensal bacteria producing 6-N-hydroxyaminopurine (6-HAP), has reduced the incidence of ultraviolet-induced tumours [10]. Trueperella pyogenes causes diverse diseases in animals like mastitis, liver abscesses and pneumonia and it is rarely a cause of infection in humans [104–106].

Lactobacillus and Corynebacterium species have been shown to produce immunostimulatory metabolites leading to anti-cancer effects. Lactobacillus johnsonii was conducted to an immune-stimulatory effect by producing inosine which is a modulator of response to immune checkpoint blockade therapy and strongly enhanced the antitumor capacities of T cells in different tumour models including colorectal cancer, bladder cancer, and melanoma, by inducing Th1 differentiation through the inosine-A2AR-cAMP-PKA pathway [107]. A recent study showed that extracellular vesicles derived from Lactobacillus rhamnosus GG showed direct anti-tumour effects on hepatic cancer cell growth [108]. Also, the oral administration of lipoteichoic acid from Lactobacillus rhamnosus decreased the number of UV-induced skin tumours in SKH-1 hairless mice [109] and the administration of Lactobacillus acidophilus may reduce the incidence of colorectal cancer in rat models [110]. The antitumor effect of Corynebacterium parvum has been demonstrated since a very early time [111–113], and it was used as an immunostimulant and antitumor agent. Indeed, the intratumoral injection of Corynebacterium granulosum and Corynebacterium parvum in hamster melanoma showed regression of the tumour and reduction in the number of lung metastases [114, 115]. Lipton et al. indicated a decreased relapse rate and prolonged survival in patients in stage II, but not in stage I, of malignant melanoma treated with C. parvum when compared with BCG treated patients [116]. Nonetheless, no significant results were observed during the administration of Corynebacterium parvum followed by chemotherapy in patients with metastatic malignant melanoma compared with the group receiving chemotherapy alone [117]. Though, Corynebacterium sp was clinically associated with the progression of acral melanoma [82]. In our results, LDA detected two biomarkers genera from Corynebacteriaceae family: Corynebacterium1, which was discriminately abundant in healthy skin microbiome (Fig. S5a, S5b), (Table S3) and Corynebacterium, which was associated with melanoma progression (Fig. 4b), (Table S3).

Among the other bacteria affecting the immune response in cancer, E. coli produces colibactin, which may induce adenocarcinomas in human colorectal cancer patients [118]. Bacteroides fragilis led to promote colon tumorigenesis by overstimulating immune responses via T helper 17 (Th17) cells in mouse colorectal cancer model [119]. Clostridium species may suppress tumour growth in the liver and melanoma by restoring antitumour immunity [120, 121]. Gut microbiome enriched with Faecalibacterium was correlated with increased Immune Checkpoint Inhibitors response and improved immunotherapy response in mouse models and humans with metastatic melanoma [30, 122]. In a study of metastatic melanoma treated with anti-PD1 immune checkpoint blockade, the patients have reacted differently, responded and non-responded. The diversity of the faecal microbiome of the responders' patients was higher with increased abundance of the Ruminococcaceae and Faecalibacterium, while an increased abundance of Bacteriodales and a much lower bacterial diversity were observed in the non-responders’ microbiomes. In animal models, FMT of human gut microbiome enriched in Faecalibacterium to germ-free mice with melanoma showed reduced tumour growth and increased immune response in the tumour microenvironment [30]. The decrease of the relative abundances of opportunistic pathogens in skin microbiota and faecal microbiota of piglets with melanoma regression revealed the maturation of host microbiota following the age when the bacterial composition shifted from dysbiosis to the “healthy” balanced microbiota.

Functional prediction pathways results suggested the potential role of host microbiota in melanoma development. Numerous pieces of evidence have been demonstrated that metabolic disorders involved in carcinogenesis and can be a target to treat cancer [123, 124]. Sporulation and Bacterial motility proteins pathways were significantly higher in melanoma-free skin of piglets (Fig. S8). It was shown that Flagellin, the structural protein subunit of the bacterial flagellum, played a role as an adjuvant, immunomodulator, anti-tumour agent (in melanoma, colon, breast, lungs and prostate cancers) and radioprotective agent [125–127]. Salmonella Typhimurium flagellin stimulates NK cells to produce interferon-γ (IFN-γ) [128]. The flagellin genes (fliC) were detected in Clostridium chauvoei, Clostridium haemolyticum, Clostridium novyi types A and B, and Clostridium septicum [129]. Bacterial flagellin enhanced the antitumor response of the activated CD8 + T cells via TLR5 activation. Consequently, the perforin and granzyme proteins were secreted by activated CD8 + T cells and efficiently killed tumour cells [128]. Additionally, a significant reduction in tumour mass was observed after injection of flagellin into human colorectal tumours xenografted into nude mice [130]. Importantly, the vaccination of mice with syngeneic B16-OVA melanoma–derived plasma membrane vesicles engrafted with flagellin-related peptides 9Flg or 42Flg induced a dramatic inhibition of tumour growth and metastasis and resulted in complete tumour regression in lungs [131]. Clostridium novyi-NT (non-toxic) is a highly mobile spore-forming organism. Promising antitumour responses in both canine and human clinical studies were described after intratumoral injection of Clostridium novyi-NT spores [132]. Moreover, clostridial spores were considered as an ideal delivery vehicle for anti-cancer agents due to their selective germination in the hypoxic regions of solid tumour, their wide and fast dispersion throughout the tumour and their stability to oxygen and harsh conditions which allowed the immune system to recognize and destroy cancer cells efficiently [133].

Transporters, ABC transporters, Ribosome, Other ion-coupled transporters_Unclassified, Pyrimidine metabolism, Purine metabolism, Replication, Recombination and repair proteins, Aminoacyl-tRNA biosynthesis, Lipopolysaccharide biosynthesis proteins, Lipopolysaccharide biosynthesis, Bacterial secretion system and Ribosome biogenesis were highly predicted pathways affected by microbiota in MeLiM melanoma tissue (Fig. S8). Recently, ribosome synthesis was designated as a new target in cancer therapy. Moreover, recent research indicated the potential role of ribosomes compositions in tumorigenesis. The increase in ribosome biogenesis was noted in cancer cells which led to an elevation in protein synthesis and unrestrained growth [134]. The production of lipopolysaccharide (LPS) was potentially promoted by Fusobacterium which were abundant in melanoma tissue. LPS from Fusobacterium led to skin inflammation and Shwartzman reaction in rabbits [135]. LPS may promote inflammation via TLR4-mediated NF-kB activation and the production of different inflammatory factors, such as TNF-α, IL-6, and IL-1β [136, 137]. Many studies indicated the capacity of LPS to be involved in the progression of various cancers: breast cancer via a ‘MyD88-BLT2’-linked signalling cascade [138], prostate cancer by activating the NF-κB pathway [139], gastric cancer through the LPS-NF-κB-PD-L1 axis [140], and oesophageal cancer [141]. A recent study demonstrated the co-stimulation with Trueperella pyogenes pyolysin and LPSs induced autophagy and ATF6-dependent mechanism in endometrium stromal cells [142].

Amino acid metabolism is linked with tumour progression due to their indispensable role in cancer growth, cancer immunity and the tumour microenvironment [143, 144] and therapeutic means for targeting amino acid metabolism were suggested [145]. A potential antitumor effect of a combination of ascorbic acid, lysine, proline, arginine and green tea extract was investigated on human colon cancer cells HCT 116 in vivo (xenograft into male nude mice). Histological studies showed that the mixture supplementation strongly suppressed the growth of tumours by inhibiting Matrix metalloproteinases expression and invasion without toxic effects [146]. Tumour associated myeloid cells have the ability to suppress the protective anti-tumour immune response by targeting arginine metabolism and reducing arginine levels by producing arginases [147, 148]. In line with these findings result from our previous metabolomic study, where a highly significantly decreased level of arginine in plasma of MeLiM pigs with progression was the most striking difference compared to pigs with spontaneous regression [149]. Up-regulation of proline in melanoma cells compared to melanocytes was observed by de Ingeniis et al. [150]. An inhibition of the gene ALDH18A1 encoding pyrroline-5-carboxylate synthase regulating proline biosynthesis in melanoma cells significantly decreased cultured melanoma cell viability and tumour growth [151, 152].

The predicted functional pathways affected by microbiota were significantly different between regressive melanoma microbiome and progressive melanoma microbiome. It is known that tumour cells accumulate several mutations and changes in metabolic pathways that might lead to cancer proliferation and metastasis. Ribosomes, DNA repair and recombination proteins, Pyrimidine metabolisms, Purine metabolisms were significantly higher in melanoma progression while Bacterial motility proteins. Flagellar assembly, Sporulation and Bacterial chemotaxis were significantly higher in melanoma spontaneous regression (Fig. S9). Consistent with our findings, many studies have assessed the interplay of pyrimidine metabolism in tumorigenesis [153, 154]. Comprehensively, the pyrimidine metabolic rate-limiting enzymes were highly expressed in lung, breast, colon, liver, stomach, and bladder cancer and played a key role in tumour cell proliferation [155]. Pyrimidine analogues acting as antimetabolites are used in cancer treatment for decades [156, 157]. Purines are basic components of nucleotides in cell proliferation, thus impaired purine metabolism is associated with the progression of cancer [158]. Notably, a high amount of purine metabolites has been noticed in tumour cells [159]. It was illustrated that Escherichia coli was able to target lung cancer cells using chemotaxis towards the biochemical factors secreted by carcinoma cells [160]. In faecal microbiota of MeLiM piglets, fatty acids biosynthesis and lipid biosynthesis metabolism were significantly associated with melanoma progression (Fig. S6d). Cancer cells as proliferative active cells require also lipids and fatty acids for cell growth, division, proliferation and survival. Deregulated lipid metabolism is an important metabolic phenotype of cancer cells [161]. Different mechanisms of fatty acids synthesis promoting tumour progression and metastasis were explored [162, 163]. Upregulations of several phosphatidylcholines were previously observed in MeLiM plasma of pigs with melanoma progression, compared to spontaneous regression [148]. A phospholipid derivative Lysophosphatidic acid (LPA) induces chemotaxis of melanoma cells and LPA degradation by melanoma cells forms a gradient in the tumour microenvironment that drove their spreading [164].

Ribosomes, DNA repair and recombination proteins and Pyrimidine metabolisms pathways, which were associated with melanoma progression in melanoma tissue microbiome, were significantly lower in the faecal microbiota of MeLiM piglets compared to crossbred piglets (Fig. S) and they were significantly lower in the faecal microbiota of MeLiM piglets with melanoma regression at the age of 8 weeks compared to MeLiM piglets with melanoma regression at the age of 10 and 12 weeks (Fig. S, S). However, membrane transport pathways (Transporters, phosphotransferase system (PTS), ABC transporters, and others ion-coupled transporters unclassified) were significantly enriched in melanoma tissue microbiome, in the faecal microbiota of MeLiM piglets compared to crossbred piglets and they were significantly higher in the faecal microbiota of MeLiM piglets with melanoma regression at the age of 8 weeks (Fig. S, S, S, S, S). Throughout the age, membrane transport pathways significantly decreased in the melanoma regression group while no significant difference was detected in the melanoma progression group in this pathway. 10 11 12 8 9 10 11 12

Alterations of membrane transport pathway were associated with several human diseases including cancer [141, 165, 166] and caused severe functional influences: increase intestinal permeability, alteration of the balance of substances between both sides of gut mucosa and destruction of the intestinal mucosal barrier. In addition, there is new evidence of the impact of the intestinal microbiome on skin physiology and microbiome [25, 167]. In case of disrupted intestinal barriers, intestinal bacteria and their metabolites have been reported to diffuse in the bloodstream, reach the skin, and disturb skin homeostasis [25, 27]. Based on those observations, our results lead up to deduce that in MeLiM piglets the intestinal permeability was higher and allowed the migration of the intestinal bacteria and its metabolites throughout the blood to gain access to the melanoma microenvironment and enhance tumour growth and proliferation. However, following the age, the intestinal permeability of piglets with melanoma regression was decreased and consequently, the transit of intestinal bacteria and its metabolites from the gut to the skin was restricted which was conducted to melanoma regression.

Depth studies suggested that gut F. nucleatum originated from the oral cavity [168, 169]. Several reports have demonstrated that Fusobacteria can migrate from the oral cavity to other districts of the body via blood circulation or lymphatic circulation systems to cause serious infections in various organs of humans including the head, neck, respiratory tract, brain, colon, liver and lymph nodes to cause metastatic abscess formation [170–175]. Also, recent studies suggested the translocation of Fusobacteria from colorectal cancer cells to metastatic sites attached with the primary tumour cells via Fap2, as part of metastatic tissue colonization [170, 175]. These observations indicate the importance of tumour microbiota as essential components of the tumour microenvironment.

According to our knowledge, this is the first study exploring the association between gut and skin microbiome changes and melanoma progression. The implementation of MeLiM piglets as an animal model in melanoma progression might be a promising approach. Significant differences were observed in bacterial composition, relative abundances, richness and diversity indexes, and potential functional pathways in skin and gut microbiome of MeLiM piglets with melanoma progression compared to healthy piglets (controls). Fusobacterium and Bacteroides were the common potential biomarkers identified in both gut and skin microbiome in MeLiM piglets with melanoma progression.

The comprehensive analysis of the gut-skin axis is essential to understand the bidirectional cross-talk between the gut and skin tumour microbiome and is regarded as an exciting field of research, with promising therapeutic, dermatologic and cosmetic applications. Moreover, the evaluation of the functions of distinct bacteria (biomarker) and their metabolites in host physiology and cancer progression can provide novel insights into the underlying mechanisms and pathways to enhance the efficacy of both anticancer therapy and cancer prevention. Furthermore, a meta-analysis might be potentially expanded to include fungi and viruses in order to fully exploit the interaction network and potential functional between the gut and skin microbiome and melanoma development.

We would like to thank Jitka Klucinova and Jaroslava Sestakova for their excellent technical assistance. The authors are also grateful for the human resources support by the "PPLZ" programme of CAS.

All authors contributed significantly to the manuscript. Study design: CM, HKS, JM; methodology: CM, HKS, JC, VH, JM; animal handling and monitoring: HKS, JC, VC, AP, VH; sampling: HKS, JC, VC, AP, VH; NGS experiment: CM, JM; statistical analyses: CM, JM; writing—original draft preparation: CM, HKS; writing—review and editing: all authors; supervision: HKS, JM; project administration: CM, HKS, JM; funding acquisition: HKS, JM. All authors have read and agreed to the published version of the manuscript.

This study was supported by the Ministry of education, youth and sports from the Operational Programme Research, Development and Education projects Center for Tumor Ecology—Research of the Cancer Microenvironment Supporting Cancer Growth and Spread (reg. No. CZ.02.1.01/0.0/0.0/16_019/0000785), that provided sources for animal experiments and sample collection, and the Project Excellence (No. CZ.02.1.01/0.0/0.0/15_003/0000460), that provided sources for molecular biology analyses. The funding bodies played no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.