ww.nature.com/scientificreports/ Predicted KEGG Pathways-NAF from HC vS BC Beta-Glucuronidase(k01195) u010 Lipopolysaccharide ble 0.006 olycyclic aromatic Nitrogen metabolism NAF from HC NAF from BC -og2(p-value) Figure 5. PICRUSt-predicted metagenomes between NAF from HC vS BC. (a) Negative log 2 of adjusted pvalues from Kruskal-Wallis test comparing the relative abundances of 18 selected KEGG pathways in NAF from HC(n=6)vs BC(n=6). The NAF microbiome was normalized by copy numbers before the PICRUSt metagenomic inference. Relative abundances were normalized to a threshold of 26359 counts(lowest of the samples)before the Kruskal-Wallis test. (b)Strip-plot comparing the relative abundance of the KEgg gene Beta-Glucuronidase between NAF from HC vS BC. Beta-Glucuronidase is the only predicted KEGG ortholog hat contributes to the"Flavone and Flavonol Biosynthesis" pathway Beta-Glucuronidase is predicted by a omposite of 8 OTUs shown in Table 2. predictive of Beta-Glucuronidase levels(Table 1)were different fro Enterobacteriaceae family members and some Firmicutes genera". e potential role of bacterial Beta-Glucuronidase in relation to breast cancer. F ng microbiome by functional gene analysis may be more physiological otu differences based on taxonomic affiliations Implication of Beta-Glucuronidase in breast cancer is not a new idea. One way estrogen is excreted is by con ugating it with glucuronic acid, making it water soluble for excretion. When glucuronidate-conjugated estrogen reaches a site rich with Beta-Glucuronidase from microorganisms or from inflammation(human lysosomes con tain Beta-Glucuronidase), the enzyme reverses the conjugation and leaves biologically active estrogen to accu mulate at the site 2. In a rat model for breast cancer, inhibition of Beta-Glucuronidase with calcium D-glucarate educes breast cancer incidence by lowering endogenous levels of estradiol 3. Some have even postulated the use of calcium D-glucarate supplementation to reduce the cancer risk of some individuals". There are no previous reports of elevated Beta-Glucuronidase in breast ductal fluid. Our preliminary finding provokes an intriguing hypothesis that bacterial Beta-Glucuronidase may deconjugate glucuronidate-conjugated estrogen present in breast tissue Apart from metabolic activity, microbes may contribute to the multifactorial susceptibility to breast can cer. The epithelial cells that line the breast ducts act as a barrier to the outside world and, unless breached, eparate the NAF microbiome from that of the tissue. In addition, the ductal epithelium has the capability to sense microbial-derived signals through Toll-Like Receptors(TLrs)and NOD-like receptors(NLRs)expressed at their surfaces. While previous studies have demonstrated that the activation of the TLR2-MyD88 pathway in mammary and gastrointestinal epithelial cells leads to tumor growth and promotes carcinogenesis, 6, TLR5 activation with flagellin inhibits cancer cell growth and mediates potent anti-tumor activity in breast cancer6. Interestingly, Sphingobium yanoikuyae is motile with a single polar flagellum", providing a ligand for the TlR5 pathway. Further work is warranted to investigate the potential interaction of the breast ductal microbiome and the immune response as it relates to breast cancer. This study was limited to the microbiome obtained from ducts that produced NAF, while ducts that did produce NAF were not studied. In any given individual, NAF was produced by one or more ducts, and NAF from multiple ducts can vary within an individuals. Future investigations should focus on individual ducts from nor mal subjects and subjects with limited early breast disease such as ductal carcinoma in situ(DCiS). This would provide valuable insight into microbiome variations between malignant and healthy ducts and their relation breast cancer In conclusion, this study demonstrates the presence of microbes in NAE, shows that the microbiome of NAF from HC and BC are significantly different, identifies particular organisms that are differentially present in HC and BC, and provides a metabolic insight into possible mechanisms for the association between breast microflora d malignancy. Since the treatment information for all BC was not available, we cannot rule out the possibility that another factor BC have in common, such as radiation or other therapy for breast cancer could have contrib uted to our finding. Further studies should explore how bacteria are associated with breast cancer. SCIENTIFIC REPORTS 6: 28061 DO1: 10.1038/srep28061www.nature.com/scientificreports/ Scientific Reports | 6:28061 | DOI: 10.1038/srep28061 6 predictive of Beta-Glucuronidase levels (Table 1) were different from those reported in colon cancer, which are Enterobacteriaceae family members and some Firmicutes genera40. This preliminary finding hints at the potential role of bacterial Beta-Glucuronidase in relation to breast cancer. Finally, this analysis suggests that defining the microbiome by functional gene analysis may be more physiologically relevant than simply measuring individual OTU differences based on taxonomic affiliations. Implication of Beta-Glucuronidase in breast cancer is not a new idea. One way estrogen is excreted is by con￾jugating it with glucuronic acid, making it water soluble for excretion41. When glucuronidate-conjugated estrogen reaches a site rich with Beta-Glucuronidase from microorganisms or from inflammation (human lysosomes con￾tain Beta-Glucuronidase), the enzyme reverses the conjugation and leaves biologically active estrogen to accu￾mulate at the site42. In a rat model for breast cancer, inhibition of Beta-Glucuronidase with calcium D-glucarate reduces breast cancer incidence by lowering endogenous levels of estradiol43. Some have even postulated the use of calcium D-glucarate supplementation to reduce the cancer risk of some individuals44. There are no previous reports of elevated Beta-Glucuronidase in breast ductal fluid. Our preliminary finding provokes an intriguing hypothesis that bacterial Beta-Glucuronidase may deconjugate glucuronidate-conjugated estrogen present in breast tissue. Apart from metabolic activity, microbes may contribute to the multifactorial susceptibility to breast can￾cer2 . The epithelial cells that line the breast ducts act as a barrier to the outside world and, unless breached, separate the NAF microbiome from that of the tissue. In addition, the ductal epithelium has the capability to sense microbial-derived signals through Toll-Like Receptors (TLRs) and NOD-like receptors (NLRs) expressed at their surfaces45. While previous studies have demonstrated that the activation of the TLR2-MyD88 pathway in mammary and gastrointestinal epithelial cells leads to tumor growth and promotes carcinogenesis45,46, TLR5 activation with flagellin inhibits cancer cell growth and mediates potent anti-tumor activity in breast cancer36. Interestingly, Sphingobium yanoikuyae is motile with a single polar flagellum47, providing a ligand for the TLR5 pathway. Further work is warranted to investigate the potential interaction of the breast ductal microbiome and the immune response as it relates to breast cancer. This study was limited to the microbiome obtained from ducts that produced NAF, while ducts that did not produce NAF were not studied. In any given individual, NAF was produced by one or more ducts, and NAF from multiple ducts can vary within an individual48. Future investigations should focus on individual ducts from nor￾mal subjects and subjects with limited early breast disease such as ductal carcinoma in situ (DCIS). This would provide valuable insight into microbiome variations between malignant and healthy ducts and their relation to breast cancer. In conclusion, this study demonstrates the presence of microbes in NAF, shows that the microbiome of NAF from HC and BC are significantly different, identifies particular organisms that are differentially present in HC and BC, and provides a metabolic insight into possible mechanisms for the association between breast microflora and malignancy. Since the treatment information for all BC was not available, we cannot rule out the possibility that another factor BC have in common, such as radiation or other therapy for breast cancer could have contrib￾uted to our finding. Further studies should explore how bacteria are associated with breast cancer. Figure 5. PICRUSt-predicted metagenomes between NAF from HC vs. BC. (a) Negative log2 of adjusted￾pvalues from Kruskal-Wallis test comparing the relative abundances of 18 selected KEGG pathways in NAF from HC (n=6) vs. BC (n=6). The NAF microbiome was normalized by copy numbers before the PICRUSt metagenomic inference. Relative abundances were normalized to a threshold of 26359 counts (lowest of the samples) before the Kruskal-Wallis test. (b) Strip-plot comparing the relative abundance of the KEGG gene “Beta-Glucuronidase” between NAF from HC vs. BC. Beta-Glucuronidase is the only predicted KEGG ortholog that contributes to the “Flavone and Flavonol Biosynthesis” pathway. Beta-Glucuronidase is predicted by a composite of 8 OTUs shown in Table 2