Assembling the anaerobic gamma-butyrobetaine to TMA metabolic pathway in Escherichia fergusonii and confirming its role in TMA production from dietary L-carnitine in murine models

ABSTRACT

Trimethylamine-N-oxide (TMAO) is a major pro-atherogenic and pro-thrombotic metaorganismal molecule produced through the initial conversion of the dietary L-carnitine and other precursors into trimethylamine (TMA). We recently identified a dual-microbe anaerobic pathway for the metabolism of L-carnitine into TMA, in which the widely distributed cai operon in Enterobacteriaceae converts L-carnitine into gamma-butyrobetaine (γBB), followed by the degradation of γBB into TMA by the relatively rare gamma-butyrobetaine utilization (gbu) gene cluster present in Emergencia timonensis and few other related microbes. Studies of this pathway in animal models, however, have been limited by the lack of single microbes harboring the whole L-carnitine→γBB→TMA transformation pathway. Such recombinant microbes would both serve as a tool to further prove the contribution of this pathway to gut microbial TMA production and for future in vivo studies investigating the diet linkage to cardiovascular disease and the involvement of the TMAO pathway in this linkage. Here, we recapitulate the whole pathway in a single microbe by cloning the E. timonensis gbu gene cluster into Escherichia fergusonii, which naturally harbors the cai operon. We then show that the native E. timonensis GroES/GroEL-like proteins are needed for the proper functioning of the gbu cluster at 37°C. Finally, we demonstrate that inoculating germ-free mice with this recombinant E. fergusonii strain is sufficient to raise serum TMAO to pathophysiological levels upon dietary L-carnitine supplementation. The recombinant E. fergusonii strain developed will be a useful tool to facilitate future studies on the role of anaerobic gut microbial L-carnitine metabolism in cardiovascular and metabolic diseases.

IMPORTANCE

The key atherosclerotic TMAO originates from the initial gut microbial conversion of L-carnitine and other dietary compounds into TMA. Developing therapeutic strategies to block gut microbial TMA production needs a detailed understanding of the different production mechanisms and their relative contributions. Recently, we identified a two-step anaerobic pathway for TMA production from L-carnitine through initial conversion by some microbes into the intermediate γBB which is then metabolized by other microbes into TMA. Investigational studies of this pathway, however, are limited by the lack of single microbes harboring the whole pathway. Here, we engineered E. fergusonii strain to harbor the whole two-step pathway and optimized the expression through cloning a specific chaperone from the original host. Inoculating germ-free mice with this recombinant E. fergusonii is enough to raise serum TMAO to pathophysiological levels upon L-carnitine feeding. This engineered microbe will facilitate future studies investigating the contribution of this pathway to cardiovascular disease.

OBSERVATION

Trimethylamine-N-oxide (TMAO) is a key pro-atherogenic and pro-thrombotic molecule produced through a metaorganismal pathway, in which the gut microbes first degrade trimethylamine (TMA) containing nutrients, including L-carnitine and phosphatidylcholine, into TMA (1 – 3). TMA is then converted by host flavin-containing monooxygenases into TMAO (4). Previously, our group proposed a novel, anaerobic multi-step gut microbial pathway for TMA production from dietary L-carnitine (3, 5 – 7). In this pathway, dietary L-carnitine is first converted into gamma-butyrobetaine (γBB) by microbes possessing the cai operon, which is widely distributed among members of the family Enterobacteriaceae, including Escherichia, Citrobacter, Proteus, and others (8, 9). The intermediate γBB is then converted into TMA by another group of microbes that include Emergencia timonensis and a handful of closely related species belonging to the Clostridia class (6, 7).

In recent studies, we reported that the novel gbu (gamma-butyrobetaine utilization) gene cluster in E. timonensis plays a critical role in the conversion of γBB to TMA and the abundance of gbu is increased in the setting of a diet rich in red meat as a protein source (7). In parallel, studies from another group also identified the gbu gene cluster as contributing to the gut microbial transformation of L-carnitine-derived γBB into TMA (10). Human dietary intervention experiments have shown that the fecal abundance of the gbuA gene (the first gene in the gbu gene cluster) not only is enhanced with increased dietary red meat but also is reduced following conversion to either a white meat or a non-meat isocaloric diet (7). Furthermore, it was previously shown that the red meat diet increases TMA and TMAO production from L-carnitine but not choline (11). A red meat-rich diet is a known contributor to cardiovascular disease risks (12 – 14) and has been similarly linked to risks for alternative diseases including colorectal cancer (15) and other cancer types (16). The ability to recapitulate the multi-step multi-microbial L-carnitine → γBB → TMA transformation in murine models using a single microbe possessing both cai and gbu gene clusters would therefore serve as a tool to further investigate the role of dietary red meat and the TMAO pathway in animal models of disease (17). Thus, in the present study, we aimed to investigate whether combining the cai and gbu gene clusters into a single genetically tractable microbe will allow this engineered microbe to convert dietary L-carnitine into TMA (and eventual TMAO generation) in murine models.

Given the poor colonization of E. timonensis in the murine host we observed in our previous study (7), and the current unavailability of the needed genetic tools to manipulate E. timonensis or the other related microbes harboring the gbu cluster, we opted to clone the gbu gene cluster, instead, into one of the microbes capable of performing the first step (L-carnitine → γBB). In previous studies employing combinatorial cloning and functional analyses (7), we showed that when the four gbu genes (gbuA, B, C, and E) were cloned from E. timonensis into a pET vector and transformed in Escherichia coli BL21/DE3 strain, TMA production from γBB occurred only at temperatures of 18°C or below, while no production was observed at 37°C in recombinant E. coli. Given that within the human gut, E. timonensis efficiently converts ɣBB into TMA at body temperature (37°C), this indicates that one or more additional proteins (presumably a chaperone) are needed for the proper folding or stability of the gbu proteins at 37°C. Identifying and expressing this additional chaperone, therefore, is needed to enable investigational studies of this metaorganismal pathway in a recombinant microbe within a murine host.

In the current study, to address this unmet need, we first generated a knock-in mutant of Escherichia fergusonii ATCC 35469 (E. fergusonii mutant 86G) in which a synthetic operon containing gbuA, B, C, and E was inserted downstream of the native CaiE gene (Fig. 1A and B). This strain was chosen because it lacks the aerobic CntA/B pathway, which can convert the L-carnitine directly into TMA (18). E. fergusonii, however, harbors choline TMA lyase enzyme (CutC/D), which converts choline into TMA (19). Therefore, to ensure that the mutant E. fergusonii is unable to utilize other pathways for TMA production, the gain-of-function mutation was generated in a ∆cutC E. fergusonii mutant background (E. fergusonii mutant 25A). The synthetic operon expression was derived by the synthetic promoter J23100 (BBa_J23100, Biobricks). Incubating this mutant with d₉-γBB under aerobic conditions confirmed that the four gbu genes were sufficient to produce d₉-TMA from d₉-γBB at 18°C, while no TMA was produced at 37°C (Fig. 1C). These results further confirmed that a molecular chaperone is needed for proper folding of one or more of the gbu cluster proteins at 37°C. Transforming and expressing each of the five commercial chaperone plasmids pG-KJE8, pGro7, pKJE7, pGTf2, and pTf16 (20, 21), which harbor different combinations of the E. coli chaperone genes dnaK, dnaJ, grpE, groES, groEL, and tig, did not result in TMA production at 37°C (data not shown). Examining the gbu cluster neighborhood in E. timonensis SN18, we could not identify a nearby gene likely to serve this function, so we next explored the gbu gene cluster neighborhood in other bacterial strains capable of producing TMA from γBB, including Agathobaculum desmolans, Eubacterium minutum, and Clostridia bacterium UC5.1–1D1 (7). Interestingly, we found a predicted groEL-like chaperonin that is associated with the gbu cluster in those three strains (Fig. 1D).

Fig 1.

Cloning and expression of the gbu gene cluster and groES/groEL–like chaperone genes from E. timonensis into E. fergusonii. (A) Scheme of the strains constructed in this study with their genotypes. (B) The gbu genes were cloned downstream of the caiE gene in E. fergusonii chromosome. (C) Harboring pGro plasmid enabled E. fergusonii 86G strain (∆cutC, PJ23100-gbuABCE) but not the control E. fergusonii 25A strain (∆cutC) to convert γBB to TMA at 37^ᴏC aerobically (n = 3). Reported are P-values from Welch t-test. (D) Gene neighborhood analysis for the gbu gene cluster showing the presence of a nearby chaperone groEL-like gene in A. desmolans, E. minutum, and Clostridia sp. UC5.1–1D1, but not in E. timonensis. (E) Map of the pGro plasmid. (F, G, and H) Time-wise TMA production from different substrates by E. fergusonii wild-type and mutant strains under anaerobic conditions. All strains were cultured anaerobically at 37°C in Luria-Bertani (LB) media supplemented with 120 µM of each of d₃-L-carnitine, d₉- γBB, and d₆-choline. Samples were taken at time intervals for LC-MS/MS analyses (n = 5). Bars represent Mean ± SE.

Blasting this chaperonin protein (locus tag: T363DRAFT_02358) from A. desmolans against E. timonensis SN18 found it to be 57% identical and 76% similar to the E. timonensis GroEL-like protein (WP_067536048). Likewise, E. timonensis GroEL-like protein showed 73% identity and 84% similarity to the corresponding locus tag of this chaperonin protein (Ga0349467_1214) in E. minutum. E. timonensis GroEL-like protein also showed 58% identity and 77% similarity to the corresponding locus tag of this chaperonin protein (Ga0100572_11685) in Clostridia bacterium UC5.1–1D1. Therefore, we next cloned the E. timonensis operon containing the putative groEL-like gene together with the upstream groES-like gene in plasmid (pGro; Fig. 1E) and transformed pGro plasmid into both E. fergusonii 25A (∆cutC) and E. fergusonii 86G (∆cutC, gbuA,B,C,E+) to make E. fergusonii 25A/pGro and E. fergusonii 86G/pGro mutants. After incubating both transformed strains with d₉-γBB, we found that E. fergusonii 86G/pGro strain was capable of performing the γBB → TMA transformation at 37°C, indicating that this chaperone complex from E. timonensis is sufficient for proper folding of the gbu cluster (gbuA, B, C, and E) proteins (Fig. 1C).

In E. fergusonii and other related organisms, the cai operon, responsible for converting L-carnitine into γBB, is induced under anaerobic conditions and in the presence of L-carnitine (9). As expected, when the culture media were supplemented with d₃-L-carnitine and upon incubating under anaerobic conditions, the E. fergusonii 86G/pGro strain was capable of not only converting d₉-γBB into d₉-TMA but also converting the supplemental d₃-L-carnitine into d₃-TMA through the d₃-γBB intermediate, while it is still impaired in converting d₆-choline into d₆-TMA (Fig. 1F through H; Fig. S1). As E. fergusonii is a facultative anaerobe, the growth and consequently TMA production from d₉-γBB by E. fergusonii 86G/pGro strain, were reduced under anaerobic conditions (Fig. 1G) when compared to TMA produced by this strain under aerobic conditions (Fig. 1C). To further increase TMA production by E. fergusonii, the J23100 promoter was replaced with the strong tac promoter to create the mutant E. fergusonii 32D/pGro. As expected, this mutant produced more of the respective TMA isotopologues from both d₃-carnitine (making d₃-TMA; Fig. 1F) and d₉-γBB (making d₉-TMA; Fig. 1G), albeit at the expense of slower growth rate (Fig. S2).

In additional studies, we tested the performance of these recombinant E. fergusonii mutants in murine models. Germ-free mice were first colonized with E. fergusonii 86G/pGro (which completes the L-carnitine to TMA transformation at 37°C) through oral gavage, while control mice were gavaged with the E. fergusonii 25A/pGro strain (which lacks the gbu gene cluster). The mice were then supplemented with either L-carnitine or γBB in the drinking water, and serum TMAO levels were measured following stabilization of the intestinal colonization 1 week later (Fig. 2A). We confirmed a similar degree of colonization between these two strains as determined by fecal bacterial load (Fig. S3A and B). On average, the 86G/pGro colonized mice supplemented with L-carnitine had TMAO levels around 17 µM (Fig. 2B), while those supplemented with γBB had an average TMAO level of only 1.8 µM (Fig. S4) indicating that direct feeding of L-carnitine but not γBB is effective in raising the serum TMAO level in 86G/pGro colonized mice. The mechanism(s) accounting for this difference are not clear but may reflect enhanced absorption (or metabolic transformation) of γBB in the proximal intestines, limiting the delivery of γBB to the more distal intestines.

Fig 2.

Colonization of germ-free mice with recombinant E. fergusonii expressing gbuABCE cassette and groES/groEL–like chaperone genes raises circulating TMAO levels following L-carnitine supplementation. (A) Schematic for the animal experiment flow. The gnotobiotic mice inoculated with control E. fergusonii 25A/pGro strain or the gain-of-function mutant (either E. fergusonii 86G/pGro or E. fergusonii 32D/pGro) were fed L-carnitine in the drinking water (at 1.3%) for 1 week, and serum TMAO levels were measured. Both groups were then gavaged with a single bolus containing 22.4 micromoles each of d₆-choline, d₃-L-carnitine, and d₉-γBB, and serum TMAO levels were measured at different time points after gavage. (B–E) The experiment was performed using gnotobiotic mice harboring the E. fergusonii 86G/pGro strain for the test group (n = 5) and the 25A/pGro strain for the control group (n = 5). (F–I) The same experiment was repeated using E. fergusonii 32D/pGro strain for the test group (n = 5) while using again E. fergusonii 25A/pGro for the control group (n = 5). Reported are P-values from the Mann Whitney test. Bars represent mean ± SE.

To confirm that TMAO originates from the dietary L-carnitine $\to$ γBB $\to$ TMA pathway rather than other sources, L-carnitine administered mice were gavaged with a single bolus containing d₃-L-carnitine, d₉-γBB, and d₆-choline, and the rate of substrate degradation (Fig. S5A through F) and accumulation of the corresponding TMAO isotopologues in the serum (Fig. 2C through E) and urine (Fig. S6A through C) was monitored over 24 h. As expected, no d₆-TMA (from d₆-choline) was detected in either the serum (Fig. 2E) or the urine (Fig. S6C) of either group, as the cutC gene is knocked out in both the control and the gain-of-function mutant. For the E. fergusonii 86G/pGro gain-of-function mutant, both d₃-TMA and d₉-TMA were detected in urine, albeit at low levels (Fig. S6A and B), while only d₉-TMA but not d₃-TMA was detected in serum (Fig. 2C and D). The absence of d₃-TMA but not d₉-TMA in the serum could be due to the competition of the gavaged d₃-L-carnitine with the L-carnitine supplemented (1.3%) in the drinking water (for L-carnitine transporting and metabolizing enzymes in E. fergusonii).

In a subsequent experiment, mice were colonized with E. fergusonii 32D/pGro mutant (∆cutC, Ptac-gbuA,B,C,E+/pGro) vs the E. fergusonii 25A/pGro strain (control). Both strains again showed similar degrees of colonization after 1 week (Fig. S3C). As shown in Fig. 2F, TMAO levels were dramatically increased in the 32D/pGro colonized mice, reaching an average of around 28 µM in serum. Similarly, both groups of mice were gavaged with a single bolus of d₆-choline, d₃-carnitine, and d₉-γBB to permit the simultaneous tracing of the rate of substrate degradation (Fig. S5G through L) and source of the different cognate TMA isotopologues. Both d₃-TMAO and d₉-TMAO (but not d₆-TMAO) accumulated in the serum (Fig. 2G through I) and urine (Fig. S6D through F) of the 32D/pGro colonized mice, indicating that 32D/pGro strain is effective in producing TMA within the murine gut and is relatively superior to 86G/pGro strain in TMA production both in vitro and in vivo (Fig. 2B and F; Fig. S6).

In conclusion, we successfully developed a gain-of-function system in E. fergusonii to recapitulate the microbial anaerobic pathway for the conversion of L-carnitine into TMA. Genetic engineering of E. fergusonii to harbor both the cai and gbu gene clusters together with the required GroES/EL-like chaperone resulted in recombinant commensal mutants (86G/pGro and 32D/pGro) that could complete both the L-carnitine to γBB and γBB to TMA transformations at 37°C, allowing the recapitulation of the pathway in gnotobiotic mice. Finally, these engineered strains will enable us and others to move closer to the principles of Koch’s Postulates and should prove useful in studies aimed at exploring the impact of a red meat-rich diet on the L-carnitine→→TMA conversion in murine models of cardiometabolic diseases. Importantly, the recombinant E. fergusonii strains we engineered constitutively express all the necessary genes for the conversion of γBB into TMA, which eliminates the need for the induction of this pathway in vivo through exogenous chemical inducers.

DATA AVAILABILITY

GraphPad prism nine was used to generate all the figures and for statistical analyses. All source data for figures included in the manuscript were deposited as GraphPad Prism files in Zenodo repository (https://zenodo.org/record/8189569).

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mbio.00937-23.

Supplemental Information. mbio.00937-23-s0001.pdf.

Supplemental methods and Fig. S1–S6.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.