Bacteria Boost Mammalian Host NAD Metabolism by Engaging the Deamidated Biosynthesis Pathway

Abstract

Nicotinamide adenine dinucleotide (NAD), a cofactor for hundreds of metabolic reactions in all cell types, plays an essential role in metabolism, DNA repair, and aging. However, how NAD metabolism is impacted by the environment remains unclear. Here, we report an unexpected trans-kingdom cooperation between bacteria and mammalian cells wherein bacteria contribute to host NAD biosynthesis. Bacteria confer resistance to inhibitors of NAMPT, the rate-limiting enzyme in the amidated NAD salvage pathway, in cancer cells and xenograft tumors. Mechanistically, a microbial nicotinamidase (PncA) that converts nicotinamide to nicotinic acid, a precursor in the alternative deamidated NAD salvage pathway, is necessary and sufficient for this protective effect. Using stable isotope tracing and microbiota-depleted mice, we demonstrate that this bacteria-mediated deamidation contributes substantially to the NAD-boosting effect of oral nicotinamide and nicotinamide riboside supplementation in several tissues. Collectively, our findings reveal an important role of bacteria-enabled deamidated pathway in host NAD metabolism.

Keywords: NAMPT inhibitors; cancer cells; deamidated NAD synthesis; germ-free mice; host-microbe interaction; microbial nicotinamidase; mycoplasma; nicotinic acid; oral nicotinamide riboside supplementation.

Figure 1.. Mycoplasma infection confers human cells with resistance to NAMPT inhibitors

(A) NAD biosynthesis pathway. NAR: nicotinic acid riboside; NAMN: nicotinic acid mononucleotide; NAAD: nicotinic acid adenine dinucleotide; NMN: nicotinamide mononucleotide; NR: nicotinamide riboside; NADS: NAD Synthetase; NAPRT: Nicotinic Acid Phosphorybosyltransferase; NMNAT: Nicotinamide Mononucleotide Adenylyltransferase. (B) Drug screen in H1299 cells. WT and E2F1 KO H1299 cells were treated with 1 μM compounds from the bioactive compound library and viability was measured 48 hours later by CellTiter-Glo assay. E2F1 KO cells were subsequently found to be infected with mycoplasma. (C) Mycoplasma infection confers resistance to NAMPT and proteasome inhibitors but sensitizes to fludarabine. CRC119 cells incubated with a supernatant from a mycoplasma-infected culture or a control medium, and treated with 100 nM STF118804, 1 μM STF31, 50 nM MLN2238 or 1μM fludarabine or control for 68 hours. Cell viability was measured by CellTiter-Glo assay (n=3, values are normalized to the corresponding DMSO controls and expressed as mean ±SD, *p<0.05). (D) Mycoplasma hyorhinis confers mammalian cells with resistance to STF118804. CRC119 cells or CRC119-Hyor cells chronically infected with Mycoplasma hyorhinis were seeded in 12-well plates. Next day, cells were treated with 100 nM STF-118804. Seventy-two hours later cells were washed and adherent cells were stained with crystal violet. (E) Elimination of mycoplasma sensitizes cells to NAMPTi-induced toxicity. Clean and mycoplasma-infected CRC119 cells were treated with 1 μg/ml doxycycline 25 μg/ml plasmocin, 400 μg/ml gentamicin or with medium control for 24 hours, then co-treated with antibiotics and 100 nM STF118804 or DMSO control for an additional 48 hours. Cell viability was measured by CellTiter-Glo assay (n=3, values are normalized to the corresponding DMSO controls and expressed as mean ±SD, *p<0.05). (F) Mycoplasma infection of tumors attenuates the transcriptional response to STF118804 treatment. Clean (no bacteria) or M. hyorhinis-infected (Mycoplasma) HCT116 cells were xenografted into nude mice, then mice were treated with 15 mg/Kg STF118804 or vehicle control as described in STAR Methods. Tumor transcriptomes were profiled by RNA-seq. Heatmap represents the relative expression levels of 4183 genes differentially expressed between vehicle-and STF188804-treated clean tumors following hierarchical clustering of the genes (n=4, cut-offs: mean expression>20, two-fold change and p<0.05). (G-I) Mycoplasma infected tumors are resistant to STF118804-induced inhibition of proliferation. Clean (no bacteria) or M. hyorhinis-infected (Myco) HCT116 cells were xenografted into nude mice. Eighteen days later, mice were treated with 30 mg/Kg STF118804 or vehicle control twice daily for nine days. Xenograft tumors were stained for a cell proliferation marker, Ki67 (G) and percentage of Ki67-positive area in each tumor section was calculated (H) (n=6–7, values are expressed as mean ±SEM). (I) Tumor weights (n=13–16, values are expressed as mean ±SEM,n. s., not significant). Bars in (G), 1 mm. See also Figures S1 and S2.

Figure 2.. Mycoplasma infection confers host cells resistance to NAMPT inhibitors by preventing NAD depletion

(A) Mycoplasma prevent NAMPTi-induced NAD and ATP depletion. Total cellular NAD (NADH + NAD⁺) and ATP levels were measured in uninfected (no bacteria) or mycoplasma-infected CRC119 cells treated with 100 nM STF118804 (STF) (n=3, values are normalized to the corresponding time zero controls and expressed as mean ±SD, *p<0.05). (B) Mycoplasma prevent NAMPTi-induced inhibition of glycolysis (ECAR) and oxidative phosphorylation (OCR). Uninfected clean or mycoplasma-infected CRC119 cells were treated with 100 nM STF118804 (STF) for 48 hours. The basal extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) were measured using Seahorse instrument (n=7–8, values are expressed as mean ±SD, *p<0.05). (C) Mycoplasma prevent NAMPTi-induced mitochondrial NAD depletion. Relative levels of mitochondrial total NAD were measured after 24 hours treatment with 100 nM STF118804 (STF) or DMSO control (n=3, values are normalized to the uninfected DMSO control and expressed as mean ±SD, *p<0.05). (D) Mycoplasma prevent NAMPTi-induced loss of mitochondrial membrane potential. Mitochondrial membrane potential loss (% TMRE-negative cells) was measured by flow cytometry after 24 hours treatment with 100 nM STF118804 (STF) or DMSO control. (n=3, values are expressed as mean ±SD, *p<0.05). See also Figure S2C.

Figure 3.. Mycoplasma induce deamidated NAD precursors and prevent NAMPTi-induced NAD and energy depletion

(A) Mycoplasma increase cellular levels of deamidated NAD precursors. Clean (no bacteria) or M. hyorhinis-infected (myco) CRC119 cells were treated with either 100 nM STF118804 or with DMSO control for 24 hours and relative levels of 330 metabolites were determined by LC-MS. Volcano plot of metabolites from mycoplasma-infected cells (Myco_DMSO) vs. uninfected cells (no bacteria_DMSO). (B) Mycoplasma increase deamidated NAD precursors inside the cells and in the medium upon STF118804 treatment. The log ratios of the relative abundance of metabolites in clean and infected cells are represented by color scale. Metabolites in parenthesis were not detected in our LC-MS analysis. (C) Relative levels of NAD⁺ and the top differential metabolites (NAAD and NAR) (n=3, values are normalized to the uninfected DMSO control and expressed as mean ±SD, *p<0.05). (D) Mycoplasma prevent NAMPTi-induced inhibition of energy metabolism. Relative changes in the levels of glycolysis and TCA cycle metabolites following STF18804 treatment are shown for uninfected (left panel) and mycoplasma-infected cells (right panel). The log ratios of the relative abundance of metabolites in indicated pathways in clean and infected cells are represented by color scale. Metabolites in parenthesis were not detected in our LC-MS analysis (n=3, values are expressed as mean ±SD). See also Table S3 for the complete metabolomics data. See also Figure S3.

Figure 4.. Bacteria rescue NAMPTi-induced toxicity through nicotinamidase PncA

(A) E. coli protect cells from NAMPTi-induced toxicity. CRC119 cells were cultured with or without E. coli in the presence of 1 μg/ml gentamycin to prevent bacterial overgrowth. Cells were treated with 100 nM STF118804, 50 nM FK866, or with DMSO control for 42 hours (n=3, values are normalized to the corresponding DMSO controls and expressed as mean ±SD, *p<0.05). (B) E. coli protect cells from NAMPTi-induced toxicity through PncA. CRC119 cells were treated with 100 nM STF118804 or DMSO control for 24 hours, then treated with control medium, live WT E. coli, or pncA KO E. coli for additional 21 hours in the presence of 1 μg/ml gentamycin to prevent bacterial outgrowth (n=3, values are normalized to the uninfected DMSO control and expressed as mean ±SD, *p<0.05). (C) E. coli-provided protection from NAMPTi require NAM. E. coli were incubated in EBSS medium with or without nicotinamide (NAM) for 3 hours, then removed by filtering through 0.2 μm filter. The conditioned media (CM) were added for an additional 21 hours to CRC119 cells that were pre-treated for 24 hours with 100 nM STF118804 (n=3, values are normalized to the DMSO control and expressed as mean ±SD, *p<0.05). (D) NA protects cells from NAMPTi-induced toxicity. CRC119 cells were treated with 100 nM STF118804 with or without 100 μM nicotinic acid (NA) for 48 hours (n=3, values are normalized to the DMSO control and expressed as mean ±SD, *p<0.05). (E) Overexpression of PncA protects cells from NAMPTi-induced toxicity. CRC119 transfected with a control vector or a construct expressing E. coli pncA were treated with 100 nM STF118804 (STF) or DMSO control for 72 hours (n=3, values are normalized to the Vector DMSO control and expressed as mean ±SD, *p<0.05). (F-H) Blocking the deamidated NAD biosynthesis abolishes bacteria-provided protection from NAMPTi. (F) CRC119 cells (no bacteria, E.coli) or CRC119-Hyor cells chronically infected with Mycoplasma hyorhinis were treated with 100 nM STF118804, 1 mM of HNA or their combination for 66 hours. Control or E. coli-containing media were added to CRC119 cells during the last 24 hours of drug exposure. (n=3, values are normalized to the corresponding DMSO controls and expressed as mean ±SD, *p<0.05). (G) CRC119 cells or CRC119-Hyor cells chronically infected with Mycoplasma hyorhinis were transfected with control siRNA (siControl) or two independent siRNAs against NAPRT (siNAPRT #1 and siNAPRT #2) and treated with 100 nM STF118804 or DMSO control for 48 hours. (n=3, values are normalized to the corresponding DMSO controls and expressed as mean ±SD, *p<0.05). (H) CRC119 cells were transfected with control siRNA (siControl) or two independent siRNAs against NAPRT (siNAPRT #1 and siNAPRT #2) and treated with 100 nM STF118804 or DMSO control for 48 hours. Control or E. coli-containing media were added during the last 24 hours of drug exposure. (n=3, values are normalized to the uninfected siControl DMSO control and expressed as mean ±SD, *p<0.05). Relative cell numbers in panels (A)-(H) were measured by CellTiter-Glo assay. (I) Bacteria augment incorporation of NAM into metabolites in the deamidated NAD salvage pathway and NAD. CRC119 cells were infected with E. coli and treated with 100 nM STF118804 or DMSO control for 24 hours in the presence of 5 mg/l NAM labeled with four deuterium atoms (D4 NAM) on the pyridine ring. The absolute concentration of the indicated unlabeled (m+0) and labeled metabolites (m+3/m+4) was measured by LC-MS as described in STAR Methods (n=3, values are expressed as mean ±SD, *p<0.05). (J) The flux of labeled metabolites through the NAD pathways in the different conditions is shown in blue based on the data in (I). The font size of each metabolite is roughly proportional to its concentration. See also Figure S4.

Figure 5.. Gut microbiota is critical to incorporate dietary NAM into metabolites in the deamidated NAD salvage pathway and NAD in vivo

(A) Schematic of the experiment. C57BL/6J mice were treated with either regular water (Reg) or antibiotics-containing water (Abx) for 12 days to deplete gut microbiota. They were then gavaged with 80 mg/kg of D4 NAM or with PBS control, and dissected three hours later. (B) Gut microbiota is required to convert NAM into NA in colonic lumen. Relative abundance of unlabeled (m+0) and labeled (m+3/m+4) NAM and NA in colonic luminal content of indicated mice were measured by LC-MS (n=5–6 mice/group, values are expressed as mean ±SEM, *p<0.05). (C) Gut microbiota is important to incorporate dietary NAM into metabolites in the deamidated NAD salvage pathway and hepatic NAD in mice. Relative abundance of unlabeled (m+0) and labeled (m+3/m+4) NAD pathway metabolites in colons, small intestines, and livers were measured by LC-MS (n=5–6 mice/group, values are expressed as mean ±SEM, *p<0.05). (D) Bacterial pncA is required to incorporate oral NAM into metabolites in the deamidated NAD salvage pathway in mice. Germ-free (GF) C57BL/6NTac mice repopulated with WT or pncA KO E. coli were orally gavaged with 80 mg/kg of D4 NAM and dissected three hours later. The relative abundance of unlabeled (m+0) and labeled (m+3 and m+4) NAD pathway metabolites in colon and liver were measured by LC-MS (n=3 mice/group, values are expressed as mean ±SEM, *p<0.05). (E) Gut microbiota systemically boosts the flux of dietary NAM into NAD through the deamidated pathway in vivo. The log ratios of the relative abundance of indicated labeled metabolites are represented by a color scale based on data in (B) and (C). Metabolites in parenthesis were not analyzed in this LC-MS analysis (n=5–6 mice/group, values are expressed as mean ±SEM). See also Figure S5A–S5C.

Figure 6.. Oral NAM is incorporated into NAD primarily through gut microbiota-enabled deamidated NAD biosynthesis pathway in several mouse tissues

(A) Dietary NAM is incorporated into colon and hepatic NAD primarily through the deamidated pathway in regular mice. Microbiota-proficient regular mice (Reg) were gavaged with control saline, 80 mg/Kg amide-¹⁵N labeled NAM (¹⁵N NAM), or 80 mg/Kg D4 NAM, then dissected three hours later. Relative abundance of unlabeled (m+0), and ¹⁵N labeled (m+1), or D4/D3 labeled (m+3/m+4) in colons and livers of indicated mice were measured by LC-MS (n=4 mice/group, values are expressed as mean ±SEM). (B) Schematic of the [¹⁸O,¹⁵N,¹³C-amide]nicotinamide (¹⁸O,¹⁵N,¹³C-NAM) gavage experiment. The right panel shows the PncA-mediated reactions that lead to the indicated molecular weight changes. (C) Oral NAM is incorporated into NAD primarily through the deamidated NAD salvage pathway in colon, liver, and kidney. Regular mice were gavaged with either saline or 80 mg/Kg ¹⁸O,¹⁵N,¹³C-NAM. The concentrations of unlabeled (m+0) and labeled NAD pathway metabolites synthesized via the deamidated (m+3 and m+1), and amidated (m+4) pathways in indicated tissues were measured by LC-MS 3 hours after gavage (n=3–4 mice/group, values are expressed as mean ±SEM). (D) NAMN is the major deamidated NAM precursor in the portal blood. Concentrations of the indicated NAD pathway metabolites in blood collected from the portal vein in the experiment described in (B), (C) (n=3 mice/group, values are expressed as mean ±SEM). (E) Oral NAM supplement is incorporated into NAD primarily through gut microbiota-enabled deamidated NAD salvage in several mouse tissues. Data from the experiment in (B) is presented as a percentage of labeled NAD synthesized via the deamidated (m+3 and m+1) and amidated (m+4) pathways for both 4 mg/Kg and 80 mg/Kg doses of ¹⁸O,¹⁵N,¹³C-NAM (n=3–4 mice/group). (F) The relative fractions of the labeled NAD metabolites synthesized via the amidated pathway (Red, m+4) or via the deamidated pathway (Black, m+1 and m+3) after oral gavage of 80 mg/Kg of ¹⁸O,¹⁵N,¹³C-NAM are graphed proportionally based on the data in (C, D). The font size of each metabolite is roughly proportional to its tissue concentration. Gut lumen content was not analyzed in this experiment and is presented here schematically based on data from a similar experiment presented on Figure 5B. See also Figures S5D–S5G.

Figure 7.. Gut microbiota plays an important role in incorporation of dietary NR into deamidated precursors and NAD in vivo

(A) Schematic of the experiment. Regular (Reg) and germ-free (GF) C57BL/6NTac mice were gavaged with 185 mg/kg of pyridyl-¹⁵N-labeled NR chloride or with saline control, and dissected one or four hours later. (B) Gut microbiota converts NR into NAM, NA, and NAR in the colonic lumen. Relative abundance of unlabeled (m+0) and labeled (m+1) NR, NAM, and NA in the colonic lumen content of indicated mice were measured by LC-MS four hours after NR gavage (n=3–4 mice/group, values are expressed as mean ±SEM, *p<0.05). (C) Gut microbiota is important to incorporate dietary NR into metabolites in both amidated and deamidated NAD biosynthesis pathways and into NAD in colons. Relative abundance of unlabeled (m+0) and labeled (m+1) NAD pathway metabolites in colons were measured by LC-MS 4 hours after NR gavage (n=3–4 mice/group, values are expressed as mean ±SEM, *p<0.05). (D) Gut microbiota-produced amidated and deamidated NAD precursors from dietary NR are detectable in the portal blood. Relative abundance of unlabeled (m+0) and labeled (m+1) NAD pathway metabolites in the portal blood were measured by LC-MS four hours after NR gavage (n=3–4 mice/group, values are expressed as mean ±SEM, *p<0.05). (E) Gut microbiota is important to incorporate dietary NR into hepatic NAD through both amidated and deamidated pathways. Relative abundance of unlabeled (m+0) and labeled (m+1) hepatic metabolites in NAD metabolic pathways were measured by LC-MS 4 hours after NR gavage (n=3–4 mice/group, values are expressed as mean ±SEM, *p<0.05). (F) Gut microbiota boosts the flux of dietary NR into NAD through both amidated and deamidated pathway in vivo. The log ratios of the relative abundance of indicated labeled metabolites are represented by a color scale (n=3–4 mice/group). See also Figures S6 and S7.