1. Introduction
Prostate cancer was the third leading cause of cancer-related death worldwide in 2018 [
1]. It has a long incubation period before clinical manifestation and is typically diagnosed in men, 50 years old or older [
2]. Genetic as well as environmental factors, especially dietary patterns, have been reported to be involved in prostate cancer development [
3,
4]. A long latency and strong environmental conditions make prostate cancer an ideal target for nutritional intervention. Furthermore, a number of diet-derived bioactive compounds, such as lycopene, vitamin D, selenium, and indole-3-carbinol, have been reported to exert protective effects on prostate cancer development [
5]. However, the precise mechanisms by which diet or diet-derived compounds protect against prostate cancer remain unclear.
Cruciferous vegetables, including broccoli, cabbage, and cauliflower, are inversely associated with prostate cancer incident. It has been reported that consuming three or more serving of cruciferous vegetables per week could reduce the risk of developing prostate cancer by 40% [
6]. Due to such potential benefits, the identification of natural, anti-carcinogenic bioactive compounds found in cruciferous vegetables has gained widespread attention. Indole-3-carbinol (I3C) is derived from naturally occurring glucosinolates in cruciferous vegetables. It is rapidly converted into a range of metabolites under the acidic stomach environment, among which 3,3-diindolylmethane (DIM) is the most prominent compound [
7]. These indoles are known for their chemopreventive effects on hormone-dependent cancers such as prostate cancer [
8]. It has been shown that I3C/DIM inhibit proliferation, migration, and invasion of prostate cancer cells in vitro by targeting a wide spectrum of signal pathways governing hormonal homeostasis, cell cycle progression, apoptosis, and DNA repair [
9,
10,
11]. Several nuclear transcription factor-mediated pathways related to these biological processes have been reported to be modulated by I3C/DIM [
12]. However, the precise mechanisms by which I3C or DIM may work to protect against prostate cancer remain unclear and warrant further elucidation.
Recently, emerging evidence supports work on the trillions of bacteria residing in the human gastrointestinal (GI) tract form a complex ecological community, commonly known as the gut microbiota [
13]. Published literature also supports that the gut microbiota considerably impacts on host health by modulating immune response, intestinal homeostasis, nutrients processing, energy harvesting, and resistance to pathogens [
14,
15,
16]. Many factors, including dietary, have been implicated in the establishment and development of gut microbiota in individuals [
17,
18]. Moreover, alteration of the gut microbiota may also arise in diverse clinical situation such as obesity, inflammatory bowel disease, and cancer [
19]. A recent study by Golombos et al. [
20] indicated that biologically significant differences exist in the gut microbial composition of men with prostate cancer compared to benign controls. Higher relative abundance of
Bacteriodes massiliensis was seen in the prostate cancer cases compared to the controls. In contrast,
Faecalibacterium prausnitzii and
Eubacterium rectalie had higher relative abundance among the control subjects. In addition, a study by Frugé et al. reported that Gleason score was positively associated with the phylum Deferribacteres (
p < 0.032) and several Proteobacteria taxa. Gleason score was also positively associated with Clostridium (
p < 0.005) and inversely associated with
Blautia (
p < 0.049) [
21]. These studies suggest a possible link between the gut microbiota and development of prostate cancer. More importantly, a critical modulatory roles of dietary bioactive compounds on the gut microbiota has emerged as an important factor in the etiology, prevention, and therapy of various diseases, including cancer [
13,
22]. However, literature related to the effects of diet or diet-derived compounds such as I3C on the gut microbiota are scarce. Thus, it is unclear whether a relationship between dietary ingestion of I3C, gut microbiota, and prostate cancer development exists.
The present study seeks to address the deficiency in the literature and test the hypothesis that the prostate cancer protective effects of I3C against prostate cancer may be associated with modulation of the gut microbiota. In this study, the impact of I3C on gut microbiota composition was investigated in a mouse prostate cancer xenograft model using a metagenomics approach. Furthermore, we sought to delineate the complex interactions among I3C, gut microbiome, and prostate cancer development using co-occurrence network analysis.
2. Materials and Methods
2.1. Animals and Diet
Male athymic nude mice (BALB/c nu/nu, 20–22 g, 6–8 weeks old) were purchased from Charles River (Frederick, MD, USA) and were individually housed in filter-top cages at the Beltsville Human Nutrition Center’s animal facility under 12-h light cycle. An acclimation period of 1 week prior to treatment feedings enabled mice to adapt to their environment. There were nine animals in each group. Animals were then randomized into two experimental feeding groups: (i) control diet, (ii) control diet with 1 µmol I3C/g diets. Animals were fed diet and filtered water ad libitum. Food consumption and body weights were recorded weekly. This study was approved by the USDA Animal Research Advisory Committee, Beltsville Area Animal Care and Use Committee protocol #12-030.
2.2. Tumor Xenograft Model
LNCaP (ATCC CRL-1740), an androgen-dependent human prostate carcinoma cell line, was grown in Roswell Park Memorial Institute 1640 medium (RPMI 1640) supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 µg/mL streptomycin purchased from Invitrogen (Carlsbad, CA, USA). Cultures were incubated in an atmosphere composed of 5% CO2 at 37 °C.
After a two-week dietary intervention, all mice were inoculated with LNCaP cell suspension via subcutaneous injection into the right and left flanks. Cell suspension consisted of LNCaP cells at a density of 2 × 10
6 with 50 µL of phosphate-buffered saline (PBS) and an equal volume of Matrigel (BD Biosciences, Mansfield, MA, USA). Mice remained on their respective diets for 7 weeks following inoculation. Injection sites and tumor volume (cm
3) were monitored and measured weekly for palpable tumor growth and biological efficacy: (cm
3) = 0.523 × [length (cm) × width
2 (cm
2)] [
23,
24].
2.3. Fecal DNA Extraction and Analysis
The mouse fecal pellet was homogenized with Precellys (Bertin Technologies, Saint-Aubin France) at 7500 rpm for 1 min. Then fecal DNA was extracted using a QIAamp DNA Stool Mini Kit from Qiagen (Valencia, CA, USA) following manufacturer’s protocol with modification [
25]. DNA was eluted from the column with 100 μL nuclease-free water. The concentration of DNA elution was determined by its absorbance at 260 nm, followed by serial dilutions to the final concentration of 10 ng/μL. Bacterial groups were quantified by quantitative PCR using Applied Biosystems 7900T Real-Time PCR System (Applied Biosystems™, Forest City, CA, USA) and the group-specific primers are shown in
Table 1. Real-time PCR was performed on Applied Biosystems 7900T Real-Time PCR System using 10 μL SYBR
® Green Real-Time PCR Master Mix (Applied Biosystems™, Forest City, CA, USA), 0.25 μL 500 nM custom-made oligo primers, 4.5 μL water, and 5 μL 10 ng/μL DNA [
26,
27]. The relative expression levels of bacteria in fecal samples were calculated from threshold cycle values.
2.4. Cecal DNA Extraction, Amplicon Library Construction and Sequencing
Cecal samples were obtained at the time of sacrifice and quick-frozen in liquid nitrogen, then stored at −80 °C until needed. Total DNA was extracted from cecum contents using QIAamp Fast DNA Stool Mini Kit (Qiagen, Valencia, CA, USA) with some modifications. The hypervariable regions V3-V4 of the 16S rRNA gene were amplified from 20 ng of total DNA with a sample-specific barcode using 2.5 units of AccuPrimeTM Taq DNA Polymerase (Invitrogen, Carlsbad, CA, USA) in a 50 µL reaction buffer containing 200 nM primers, 200 nM dNTP, 60 mM Tris-SO
4, 18 mM (NH
4)
2SO
4, 2.0 mM MgSO
4, 1% glycerol, and 100 ng/µL bovine serum albumin (New England Biolabs, Beverly, MA, USA). PCR was performed under the following cycling conditions: initial denaturing at 95 °C for 2 min followed by 20 cycles of 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 60 s. Amplicons were purified using Agencourt AMPure XP bead kits (Beckman Coulter, Fullerton, CA, USA) and quantified using a High Sensitivity DNA Kit (Agilent, Santa Clara, CA, USA). The purified amplicons from each sample were pooled in equal molar ratios and further spiked with approximately 25% of whole-genome shotgun libraries prepared using an Illumina TruSeq DNA sample prep kit (Illumina, San Diego, CA, USA) with a compatible adaptor barcode to enhance sequence diversity during the first few cycles of sequencing for better cluster differentiation. The final concentration of the library pool was determined using a Bio-Analyzer high-sensitivity DNA chip kit (Agilent, USA). The library pool was sequenced using an Illumina MiSeq Reagent Kit and Illumina MiSeq sequencer [
28,
29].
2.5. Sequence-Based Microbiome Analysis
The sequence data were preprocessed using MiSeq Control Software (MCS) version 2.4.1 (Illumina, San Diego, CA, USA). Raw sequences were first analyzed using FastQC version 0.11.2 (Babraham Institute, Cambridge, UK) to check basic statistics, such as %GC, per base quality score distribution, and sequences flagged as poor quality. The four maximally degenerate bases (“NNNN”) at the 5′ end of the read pair, which were designed to maximize the diversity during the sequencing run of first four bases for better identification of unique clusters and improve base-calling accuracy, were then removed. The presence of forward and reverse PCR primers at the 5′ and 3′ ends of each sequence read was scanned and the reads without primers sequences were discarded. The chimeric reads were also removed from the dataset. The processed pair-end reads were then merged using PandaSeq version 2.8 (University of Waterloo, Waterloo, ON, Canada) to generate representative complete nucleotide sequences (contigs) using default parameters. The overlapping regions of the pair-end read were first aligned and scored, and reads with low score alignments and high rate of mismatches were discarded.
QIIME pipeline (version 1.9.1, University of Colorado Boulder, Boulder, CO, USA) was used to analyze the 16S rRNA gene sequences. The sequences with ≥97% identity were binned into operation taxonomic unit (OTU) according to a “closed reference” protocol. GreenGene database (version 13.8) was used for taxonomy assignment (
greengenes.lbl.gov). PyNAST (version 1.2.2, University of Pennsylvania School of Medicine, Philadelphia, PA, USA) was used for sequence alignment. The microbial diversity in the murine intestine was analyzed using QIIME pipeline based on the OTU table. OTU relative abundance values were then analyzed using the LEfSe algorithm to identify taxa that displayed significant differences between two biological conditions. -The LEfSE uses the linear discriminant analysis (LDA) to estimate the - size of each differentially abundant feature.
2.6. Network Construction and Analysis
The network analysis (control and I3C treatment) was performed by Random-Matrix theory (RMT)-based pipeline, as described by Zhou et al. [
30]. The network construction was based on the OTU abundance table. The OTUs that were detected in less than 50% of all samples were excluded for further analysis. A similarity matrix, which measures the degree of concordance between the abundance profiles of individual OTUs across different samples [
30], was then obtained by using Pearson correlation analysis of the abundance data [
31]. The threshold values used in this study were 0.87 for the control group and 0.92 for I3C group. The fast-greedy modularity optimization procedure was used for module separation. The within-module degree (Zi) and among-module connectivity (Pi) were then calculated and plotted to generate a scatter plot for each network to gain insights into the topological roles of individual nodes in the network. The Olesen classification approach was used to define node topological roles [
32]. A Mantel test was performed to measure the relationship of the network topology and physiological traits by calculating OTU significance and node connectivity, as described by Zhou et al. [
30]. Finally, the network was visualized using Cytoscape version 3.1.0 (National Institute of General Medical Sciences, Bethesda, MD, USA).
4. Discussion
The results from the present study confirmed our hypothesis that dietary ingestion of I3C can modulate the gut microbiota. The major and novel observations from our study reside in observing an effect of dietary I3C on the gut microbiome and the association of such changes to tumor development. More specifically, ingesting I3C altered bacteria interaction, which appeared to be associated with promotion of tumor growth. These results support the notion that I3C and its metabolite DIM may protect against prostate cancer development indirectly through modulation of the gut microbiota.
The effect of dietary I3C on the gut microbiota appeared to be more subtle than we expected. There were no significant differences in microbial diversity, as indicated in alpha diversity analyses (
Table S1). I3C treatment did elicit a significant impact on the specific group of bacteria affiliated with phylum Deferribacteres, which was detected at the phylum level analysis as well as at species level. Additionally, non-hieratical analysis based on OTU level further identified changes in selected bacteria within the Firmicute phylum. However, there were no consistent changes to Firmicutes at species level, which were either down- or upregulated by I3C/DIM. Only
M. schaedleri, which belongs to the Deferribacteres phylum, was enriched by I3C treatment. Perhaps the most dramatic changes caused by I3C/DIM are related to the modulation of bacterial interaction. Module 8 of the network interaction within the control group was found to be strongly associated with tumor growth, and altered by I3C intake. Hence, we speculated that in the case of prostate cancer prevention, specific interactions between the bacterial community may be more important than overall and absolute changes to bacteria population. It is possible that the gut microbiota adapts to its intestinal environment, such as through microbe-detected compounds changes, and forms a coordinated response without dramatically altering the bacterial composition or diversity. We inferred the gut microbiota may have sufficient plasticity to handle dietary changes. The causal effects of these changes on prostate carcinogenesis warrant further validation.
Correlation of specific bacteria to prostate cancer or the effect of I3C/DIM on prostate cancer may prove to be complex. The published literature related to gut microbiota and prostate cancer development is relatively limited and may lack validation. A recent study in 22 men with prostate cancer indicated that Gleason Score was positively associated with the phylum Deferribacteres, several Proteobacteria taxa, and Bacteroidetes. Gleason Score also was positively associated with
Clostridium but inversely associated with
Blautia, both from phylum Firmicutes [
21]. We found Bacteroidetes phylum, through network analysis, was associated with tumor growth but not Proteobacteria. We did observe downregulation of Firmicutes via fecal PCR analysis. However, upregulation of Deferribacteres through metagenomic analysis would indicate that this phylum may be negatively associated with tumor growth. These findings may be a result of specie differences in mice and humans, or due to limited understanding of how bacteria are regulated by their environments. For example, specific bacteria may be more sensitive to environmental factors than others. Dietary I3C resulted in an enrichment of
M. schaedleri in our study. It has been known that
M. schaedleri, belonging to the Deferribacteres phylum, is an abundant habitant in the intestinal mucus layer of rodents but not humans. We suspect that this bacterium may be sensitive to I3C or its metabolite DIM. Interestingly, in Frugé’s article, Deferribacteres also responded to fruits and vegetables [
21], and supported the notion that this phylum may be particularly sensitive to dietary changes. Additional studies are needed to delineate the importance of the changes observed in gut microbiota in prostate carcinogenesis, as well as the role and possible protective effects of diet-derived components.
M. schaedleri, as identified in our study as having responded to I3C/DIM-induced changes, reduced nitrate and modified the mucosal gene expression of its host. This bacterium also has specialized systems for scavenging oxygen and reactive oxygen species during inflammation. Additionally, it was reported to be positively correlated with serum leptin levels in a study of diet-modified obesity [
34]. Hence, based on the known beneficial effect of
M. schaedleri in different diseases, the I3C/DIM-induced increase of
M. schaedleri may contribute to prostate cancer prevention and thus warrants further study.
Gut microbiota forms a complex community that is not only related to the number and the abundance of species, but also involves the interactions among microbial taxa. The network analysis of microbial co-occurrence patterns provided us with a new perspective to understand the structure of complex microbial communities, potential microbial interactions, and their ecological roles. Our data showed that while the topological properties of the global networks appear to be similar (
Table S2), the network composition was substantially differed between the control and I3C group (
Figure 2). The networks from each of the respective diet groups had the same number of modules (8,
Figure 2), which suggested that the interaction between the bacteria might be limited in the gut microbiota. However, few modules were functionally equivalent or shared similar node compositions. Furthermore, the topological roles of the nodes differed in these two networks. Each network had a distinct set of module hubs and connectors (
Figure 3), which likely reflected habitat heterogeneity or trophic specialization under different dietary treatments.
As mentioned above, only one module (module 8) was significantly and positively correlated with tumor growth in the control group (
r = 0.8,
p = 0.009). All nodes in this module belong to the Bacteroidetes, Bacteroidales S24-7 family. This positive correlation between module eigengenes and tumor growth was disrupted by the I3C treatment. None of the other modules showed significant associations with tumor growth in the I3C group. S24-7 is an abundant member in the gastrointestinal tracts of animals. Different environmental conditions were able to alter the relative abundance of S24-7. For example, the abundance of S24-7 increased in diabetes-sensitive mice fed a high-fat diet [
35]. Additionally, multiple studies suggested that some members of S24-7 are targeted by the innate immune system, implicating that it may be involved in host–microbiota interactions that influence gut function and host health [
36,
37]. Our findings provide further evidence that S24-7 may also be a pathological marker in prostate cancer and warrant further validation.
As mentioned above, the changes in cecal bacteria are variable even for bacteria from the same phylum, such as Firmicutes. Some bacteria from the Firmicutes phylum were upregulated and others were downregulated. In contrast to these changes, the result obtained from the PCR analysis of fecal samples showed that the Firmicutes in feces were determined to be downregulated. Although the cecal microbiota is quantitatively different from fecal microbiota, we consider this discrepancy may be due to the PCR analysis being skewed toward capturing changes in dominant species. Nonetheless, the significance of such changes/differences warrants further elucidation.
Given that cancer may modulate the gut microbiota [
38], it is possible that inoculation of tumor cells can alter the gut microbiota. Therefore, it is possible that I3C/DIM may influence the changes resulting from tumor inoculation. Additional experiments are necessary to test this hypothesis. Moreover, it is probable that I3C-related changes in microbiota may not lead to tumor changes. Additional experiments that utilize fecal transplantation in gnotobiotic animals [
39] may be used to determine whether microbiome changes elicited by I3C/DIM can affect tumor growth.