Skip to main content
Download PDF

Human-associated bacteria adopt an unusual route for synthesizing 3-acetylated tetramates for environmental adaptation

Microbiome volume 11, Article number: 97 (2023) Cite this article

Abstract

Background

Tetramates or tetramic acid-containing compounds (TACs) are a group of bioactive natural products featuring a pyrrolidine-2,4-dione ring acknowledged being closed via Dieckmann cyclization. The cariogenic Streptococcus mutans strains bearing a muc biosynthetic gene cluster (BGC) can synthesize mutanocyclin (MUC), a 3-acetylated TAC that can inhibit both leukocyte chemotaxis and filamentous development in Candida albicans. Some strains can also accumulate reutericyclins (RTCs), the intermediates of MUC biosynthesis with antibacterial activities. However, the formation mechanism of the pyrrolidine-2,4-dione ring of MUC and the distribution of muc-like BGCs along with their ecological functions has not been explored extensively.

Results

We demonstrated that a key intermediate of MUC biosynthesis, M-307, is installed by a hybrid nonribosomal peptide synthetase-polyketide synthase assembly line and its pyrrolidine-2,4-dione ring is closed via an unprecedented lactam bond formation style. Subsequent C-3 acetylation will convert M-307 to RTCs, which is then hydrolyzed by a deacylase, MucF, to remove the N-1 fatty acyl appendage to generate MUC. Distribution analysis showed that the muc-like BGCs distribute predominantly in human-associated bacteria. Interestingly, most of the muc-like BGCs possessing a mucF gene were isolated from human or livestock directly, indicating their involvement in alleviating the host’s immune attacks by synthesizing MUC; while those BGCs lacking mucF gene distribute mainly in bacteria from fermented products, suggesting that they tend to synthesize RTCs to compete with neighboring bacteria. It is noteworthy that many bacteria in the same habitats (e.g., the oral cavity) lack the muc-like BGC, but possess functional MucF homologues to “detoxify” RTCs to MUC, including several competitive bacteria of S. mutans. We also comparably studied the distribution of TAS1, a fungal enzyme responsible for the production of phytotoxic tenuazonic acids (TeAs), a class of 3-acetylated TACs with similar structure but distinct biosynthetic mechanism to MUC, and found that it mainly exists in plants or crops.

Conclusions

The in vivo and in vitro experiments revealed that the pyrrolidine-2,4-dione ring of MUC is closed via lactam bond formation, which may be adopted by many TACs without 3-acyl decorations. Besides, we found that muc-like BGCs are widespread in human-associated bacteria and their shapes and main products can be influenced by the habitat environment and vice versa. By comparing with TeAs, we provided thought-provoking insights into how ecological and evolutionary forces drive bacteria and fungi to construct a common 3-acetylated pyrrolidine-2,4-dione core through different routes, and how the biosynthetic processes are delicately controlled to generate diverse 3-acetylated TACs for environmental adaptation.

Video Abstract

Introduction

Tetramates or tetramic acid-containing compounds (TACs), featuring a pyrrolidine-2,4-dione ring, represent a large group of natural products from various terrestrial and marine organisms, including bacteria, fungi, and sponges [1,2,3]. TACs exhibit fascinating biological activities ranging from antibacterial, antifungal, antiviral, cytotoxic, to phytotoxic due to its significant structural diversities [2,3,4,5]. In recent years, hundreds of natural TACs have been isolated, while varied chemical synthesis strategies have been developed to expand the structural diversity of TACs [6, 7], and Spirotetramat (brand name Movento®), a synthetic TAC derivative possessing lipid biosynthesis inhibition activity, has been developed into an insecticide for crop protection [8].

In the past decade, investigations of TAC biosynthesis have revealed a universal Dieckmann cyclization scenario for the pyrrolidine-2,4-dione ring formation [9] (Fig. 1a). Most characterized TAC assembly lines are hybrid polyketide synthase (PKS)/non-ribosomal peptide synthetase (NRPS) enzymes with the C-terminal NRPS module containing at least three domains: a condensation (C) domain, an adenylation (A) domain, and a peptidyl carrier protein (PCP) domain. The NRPS C domain condenses the upper substrate, an acyl carrier protein (ACP) activated β-ketoacyl group, with a PCP-tethered amino acid to form a linear N-(β-ketoacyl)-amino acid intermediate, which is then offloaded and cyclized by a C-terminal reductase (R) [10], a thioesterase (TE) domain [11], or a free Dieckmann cyclase [12,13,14] via Dieckmann cyclization to give a pyrrolidine-2,4-dione ring (Fig. 1a, Additional file 1, Fig. S1).

Fig. 1
figure 1

Two different mechanisms of the pyrrolidine-2,4-dione ring formation in TACs. a Representative TACs produced using Dieckmann cyclization mechanism. b The lactam bond formation mechanism proved in this study

Full size image

Most TACs are acylated at the C-3 position of the pyrrolidine-2,4-dione ring, including some 3-acetylated TACs with a short C-3 acetyl modification. Pyrrolidine-2,4-dione ring installation in the structurally simple 3-acetylated TACs is not so “simple.” In addition to the general PKS/NRPS Dieckmann cyclization mechanism for 3-acetylated TACs, e.g., cyclopiazonic acid and Speradine A [15, 16] (Additional file 1, Fig. S1), NRPS/PKS assembly lines are proposed to be responsible for the biosynthesis of several 3-acetylated TACs, including Tenuazonic acids (TeAs), mutanocyclin (MUC), and reutericyclin (RTC) (Fig. 1), wherein the latter two were assumed to be installed through a non-Dieckmann mechanism [17, 18].

TeAs, represented by l-tenuazonic acid (l-TeA), are a group of 3-acetylated TACs isolated from Alternaria and some other phytopathogenic fungal species [19, 20]. They exhibit a broad spectrum of biological properties, including antitumor, antibacterial, and antiviral activities in addition to significant phytotoxicity [21, 22]. The biosynthesis of l-TeA has been well elucidated in the rice fungal pathogen Pyricularia oryzae. A hybrid NRPS/PKS enzyme TAS1 condenses the PCP-tethered l-isoleucine with an acetoacetyl-CoA to generate a PCP-tethered N-acetoacetyl-l-isoleucine intermediate, which is then offloaded and cyclized by the C-terminal KS domain to close the pyrrolidine-2,4-dione ring via a Dieckmann reaction [23] (Fig. 1a).

The anti-infiltration agent MUC from Streptococcus mutans, a human oral caries causative agent, is the C-5 (R)-configured enantiomer of iso-tenuazonic acid, one of the TeA isomers [24]. The proton-ionophore antibiotic RTC is the N-1 decenoic acylated MUC discovered in sourdough-isolated Limosilactobacillus reuteri strains [25]. Their biosynthetic gene clusters (BGCs, muc for MUC, and rtc for RTC) are shared with five conserved core genes [17]. Using muc BGC as an example, the core genes encode a NRPS/PKS assembly line (MucD/MucE) and three proteins (MucA/MucB/MucC) responsible for acetylation (Fig. 2). Moreover, RTC and its analogues with varied N-1 fatty acyl groups could be isolated from S. mutans B04Sm5 together with MUC, which was proposed to be generated by detaching the N-1 acyl group of RTCs by the deacylase, MucF [18]. Overall, previous studies have suggested that MUC and RTCs can be synthesized by the same NRPS/PKS machinery, and their biosynthetic pathways have been inferred [17, 18]. However, the formation mechanism of the pyrrolidine-2,4-dione ring in MUC and RTCs has not been proven yet.

Fig. 2
figure 2

The different biosynthetic mechanisms of MUC and TeAs and their diverse ecological roles in hosts. MUC and RTC isolated from bacteria share a pyrrolidine-2,4-dione ring backbone with TeAs produced by fungi. MUC and RTC are generated through lactam bond formation differently from the prevalent Dieckmann cyclization mechanism as in TeAs. Strains containing the muc gene cluster, such as human oral bacteria S. mutans, can produce both the immunomodulator (MUC) and antibacterial agent (RTC), while the rtc gene cluster without mucF originally discovered in sourdough-generated L. reuteri can only produce RTC. Different Alternaria strains isolated from plants or insects can produce diverse TeAs with phytotoxic or antibacterial activities through the incorporation of different precursors

Full size image

The biosynthetic mechanisms and evolutionary relationships of those 3-acetylated TACs (MUC, RTCs, and TeAs) with similar structures but different biological properties attracted our attention for further investigation. The functions of MucD and MucE were dissected in this study, and we provided both in vivo and in vitro evidence showing that in addition to Dieckmann cyclization, the pyrrolidine-2,4-dione ring could be closed via lactam bond formation. Additionally, we found that many human-associated bacteria possess functional MucF homologues capable of “detoxifying” the antibacterial RTCs to MUC. The distributions and organizations of related BGCs and mucF-like genes were analyzed to understand this phenomenon. Generally, interesting insights into how bacteria and fungi evolve different mechanisms to install 3-acylated TAC skeletons and how they control the biosynthetic processes delicately to synthesize different compounds with diverse bioactivities to adapt to their habitats and influence their “neighbors” was provided.

Results

Production of MUC in S. mutans UA140

In the previous study, we successfully expressed the muc BGC from S. mutans 35 in a model strain S. mutans UA159 [24]. However, MUC production in S. mutans UA159 UM was quite low (Fig. 3a), making it difficult to study its biosynthesis. To overcome this issue, we employed S. mutans UA140 [26], another genetically amenable S. mutans strain, as a host for MUC production. The muc BGC was cloned into the same genome locus of S. mutans UA140 as the producer S. mutans 35 using the NabLC technique, and the mucA-E operon was put under the control of a xylose-inducible promoter xylS1p to generate S. mutans UM (Additional file 1, Fig. S2, S3). When the S. mutans UM cells were cultured in ASS (Artificial Saliva Substitutes) medium with 0.4% xylose, stable and high-level production of MUC was detected through HPLC (Fig. 3a). Therefore, this expression system was used in the following MUC biosynthesis studies.

Fig. 3
figure 3

In vivo and in vitro characterization of MucD and MucE. a HPLC analysis of muc gene cluster heterologous expression strain S. mutans UM and the muc gene mutants. UM-, no xylose induction; UM + , xylose-induced. b HPLC analysis of M-307 feeding experiments of S. mutans UM ΔmucD and ΔmucE. c HPLC analysis of the MucE assays using d-M-283-SNAC as a substrate

Full size image

Characterization of mucD

The NRPS/PKS assembly line of MUC consists of two tri-domain proteins, including MucD (an NRPS with C-A-PCP domains) and MucE (a PKS with KS-ACP-TE domains). To study the role of mucD, we constructed its in-frame deletion mutant S. mutans UM ∆mucD (Additional file 1, Fig. S4a, S5). Next, using multiple sequence alignment and phylogenetic analysis, the N-terminal C domain of MucD (MucD-C) was carefully analyzed, suggesting that it belongs to the starter C domain clade with conserved essential active site residues (Additional file 1, Fig. S6). Two S. mutans UM strains(mucD*(H135V) and mucD*(D139N)) with point mutations at their active site residues were constructed to verify the function of MucD (Additional file 1, Fig. S4b, S5). MUC production was blocked in all three mutants (Fig. 3a). Further careful inspection of the HPLC profiles revealed that a small peak at 39 min disappeared in those three mutants. Remarkably, inactivation of mucE did not influence the production of the peak at 39 min. According to this result, the peak represents an early offloaded intermediate from the NRPS assembly line. High resolution-mass spectrometry (HR-MS) of the compound (named M-283) resulted in an m/z of 284.2227 for [M + H]+ ion, matched well with a decenoic acyl-Leu structure (C16H30NO2, calcd 284.2220 for [M + H]+ ion) (Additional file 1, Fig. S7a). It was verified to be a mixture of (E)-dec-2-enoyl-d-leucine and (E)-dec-2-enoyl-l-leucine with a ratio of about 9:1 by comparing its (S)-1-(naphthalen-2-yl)ethan-1-amine derivatives with the chemically synthesized standards (Additional file 1, Fig. S7b). Previous precursor isotope labeling studies showed that only l-Leu could be incorporated into MUC [18]. Thus, the tri-domain NRPS, MucD, may catalyze the condensation of a decenoic acyl-CoA with an l-Leu and an epimerization.

Characterization of mucE

To probe the function of MucE, a mucE in-frame deletion mutant (S. mutans UM ∆mucE) was constructed (Additional file 1, Fig. S4c, S5), which lost the capacity to synthesize MUC but could still produce M-283 (Fig. 3a). Based on the structures of MUC and M-283, we proposed that MucE-KS condenses the MucD-PCP tethered M-283 and the MucE-ACP tethered malonyl group, after which the MucE-TE offloads the elongated linear chain and closes the pyrrolidine-2,4-dione ring through the lactam bond formation to give the deacylated RTC (named M-307). To test this hypothesis, M-307 was chemically synthesized and fed to S. mutans UM ∆mucD or ∆mucE. It resulted that both mutants could restore the production of MUC (Fig. 3b), indicating that M-307 is an intermediate formed by MucD and MucE.

Subsequently, we expressed MucE as an N-His6 tagged protein in Escherichia coli BAP1 containing the sfp gene, which encodes a phosphopantetheinyl transferase (PPTase) converting apo-MucE to its active holo-form post-translationally. Meanwhile, PKsC, a Malonyl-CoA: ACP transacylase (MAT) with high substrate flexibility toward the ACP domain of different trans-AT PKSs, was also expressed and purified for loading CoA-activated malonyl group onto MucE-ACP. The N-acetylcysteamine forms of (E)-dec-2-enoyl-d-leucine (d-M-283-SNAC) and (E)-dec-2-enoyl-l-leucine (l-M-283-SNAC) were synthesized to mimic their PCP tethered state. Then, the catalytic activity of Holo-MucE was tested in assays with PKsC, malonyl-CoA, and d-M-283-SNAC or l-M-283-SNAC. M-307 was generated only in the case of d-M-283-SNAC (Fig. 3c, Additional file 1, Fig. S8), suggesting that MucE can catalyze the Claisen condensation between d-M-283-SNAC and the ACP-tethered malonyl group, release the elongated chain, and close the pyrrolidine-2,4-dione ring via lactam bond formation.

Both mimic substrate d-Leu-SNAC and putative product M-155 were synthesized to check whether MucE could accept PCP-tethered d-Leu. The incubation of d-Leu-SNAC with holo-MucE, PKsC, and malonyl-CoA failed to generate M-155 or any other product (Additional file 1, Fig. S9), verifying that the N-1 fatty acyl initiation catalyzed by MucD-C is necessary for MUC assembly.

Genes mucF–J are not necessary for MUC biosynthesis in S. mutans UM

It was proposed that M-307 can be modified by MucABC to add a C-3 acetyl group to generate RTC, which is then converted to MUC by an unusual peptidase encoded by mucF. Unexpectedly, the production of MUC in S. mutans UM ∆mucF–J (Fig. 3a), a mutant with the mucF–J gene cassette being deleted (Additional file 1, Fig. S4d, S5), was not influenced. This suggests that the five genes are not essential for the production of MUC in S. mutans UM. MucG and MucH are TetR/AcrR family transcriptional regulators, MucI is a DHA2 family efflux MFS transporter permease subunit [27], and MucJ is a putative multi-drug export protein (Additional file 1, Table S1). Since the mucA–E operon was under the control of a xylose-inducible promoter in S. mutans UM, it was not surprising that the regulatory genes were unnecessary. The functions of MucI and MucJ may be compensated by other exporters. However, the peptidase activity of MucF should be essential for MUC biosynthesis. To probe the function of mucF35 from S. mutans 35, we first cloned it into pET28a and expressed this membrane protein in E. coli. MUC production could be detected when the cell lysate was incubated with RTC (Fig. 4), verifying the RTC deacylase activity of MucF35. Additionally, we conducted an in vivo assay in which the codon-optimized mucF35 was inserted into pEXT06 (containing mucA–E for synthesizing RTCs in E. coli), and the E. coli strain harboring this plasmid gained the ability to produce MUC as anticipated (Fig. 4a). These results implied that there are other proteins in S. mutans UM ∆mucF–J capable of rescuing the loss of MucF. A BLASTP search of the S. mutans UA140 genome for MucF-like proteins resulted in a homologous protein, MucF140 that shares 55.9% identity with MucF35. When MucF140 was characterized in vivo using the same assay as that for MucF35, it could act as an efficient RTC deacylase (Fig. 4), which explains why S. mutans UM ∆mucF–J can still synthesize MUC.

Fig. 4
figure 4

Characterization of MucF and its homologous proteins as deacylase. a LC–MS analysis of MucF and its homologous proteins in converting RTC to MUC. MUC was detected. b LC–MS analysis of MucF and its homologous proteins in converting RTC to MUC. RTC was detected. The cell lysate of E. coli Rosetta 2(DE3)pLysS/pET28a supplemented with RTC was used as the negative control for in vitro assays; E. coli BAP1::mucA-E (E. coli BAP1 containing pEXT06), which can produce RTC and does not contain MucF, was used as another negative control for in vivo assays. The function of MucF35 was verified both in vitro and in vivo; MucF140 represents MucF homologous proteins from non-MUC-producing S. mutans strains; MucFSwi and MucFSgo were derived from Scardovia wiggsiae and S. gordonii respectively, representing MucF homologous proteins from oral bacteria other than S. mutans. See the LC–MS results of other MucF homologues in Additional file 1, Fig. S11, S12

Full size image

Distribution of the muc-like BGCs and the mucF-like genes

Some S. mutans with the muc BGC, like the strain B04Sm5, can synthesize both RTCs and MUC. RTCs are antibiotics that inhibit the growth of many Gram-positive bacteria, including the competitors of S. mutans in the oral cavity [18], while MUC acts as an immunomodulator to protect bacteria from the host’s immune attack and a regulator of filamentous development in Candida albicans [28], an opportunistic fungal pathogen. The fact that S. mutans UA140, which does not produce RTCs, has a MucF-like peptidase capable of hydrolyzing RTCs to MUC implies that some bacteria can take advantage of the chemical weapons produced by bacteria harboring the muc or rtc BGC, which aroused our interests to determine those “smart” bacteria by distribution investigation of the muc-like BGCs and the mucF-like genes.

Distribution analysis of the muc-like BGCs was conducted by searching the core genes of muc BGC, including mucA-E, which resulted in 63 strains belonging to Firmicutes (60 strains) and Actinobacteria (3 strains) (Additional file 1, Table S2). Strains of the former phylum include 37 S. mutans (14% of the 264 sequenced strains) and 23 other Bacilli; strains of the latter phylum include two Bifidobacterium, and one strain belonging to Bifidobacteriaceae Family. Significantly, most of the 63 strains were isolated from the human body (40 strains) or human-related environments (18 strains), such as fermented products, livestock, and crops. Further phylogenetic analysis of the nucleotide sequences spanning mucA-E could group those BGCs into four types (Fig. 5, Additional file 1, Fig. S10). Type I BGCs are distributed exclusively in the genus Streptococcus (42 strains) of which 38 strains are from the human body, including the S. mutans from the oral cavity (32 strains), ICU (2 strains), infective endocarditis tissues (2 strains), duodenal aspirate (1 strain), and the S. sp. HMSC068F04 strain from sputum. Additionally, four Streptococcus strains were either isolated from the livestock or the oral cavity of primates. The type I BGCs all share the same genetic organization as the muc BGC (Fig. 5b), and the presence of the mucF homologues suggests MUC could be the main product. Type II BGCs can be found in Limosilactobacillus and Lactiplantibacillus genus (10 strains, Additional file 1, Table S2), with eight strains isolated from fermented foods or animal feed, including four L. reuteri strains from industrial sourdough continuously used in the same factory (Fig. 5b). The other two Lactobacillaceae strains were isolated from Odontesthes bonariensis and Lion-tailed macaque. Most type II BGCs lack the mucF-like gene. This suggests that their products are RTCs retaining the N-1 fatty acyl decoration. Type III BGCs were observed in the Lactococcus hircilactis DSM 28,960 from goat milk and three Bifidobacteriaceae strains from human and fermented food. They have some additional insertions inside the mucA-E and the PPTase gene operon. The seven muc-like BGCs from the Bacillus and Priestia strains comprised Type IV BGCs, where the homologous gene of mucD is split into two genes. Apart from Bacillus cereus IZSPB_BC561A (isolated from ice cream) and Priestia megaterium 22–2 (unknown source), the other five Bacillus strains were obtained from plants. Type III and IV BGCs lack the mucF-like gene and may also synthesize RTCs. Notably, except for the clusters from S. mutans, transposase or recombinase encoding genes could be frequently observed in the vicinity of the muc-like BGCs. This implies the involvement of horizontal gene transfers (HGTs) to promote their spreading.

Fig. 5
figure 5

Phylogeny, organization, and distribution of the muc-like BGCs. a The Maximum likelihood tree of nucleotide sequences covering mucA-E from the 63 muc-like BGCs. Strains belonging to the same species were plotted as one branch; see the detailed tree showing information of all strains in Additional file 1, Fig. S9. The four types of muc-like BGCs were represented by different colors. b Organization and distribution of the four types of muc-like BGCs. The hosts of strains are indicated by black icons and text on the right. Transposases or recombinases at the adjacent position are marked with red bars

Full size image

A subsequent search using MucF35 as a probe resulted in 111 MucF-like proteins existing in 362 strains (some mucF-like genes from different strains dictate the same protein) (Fig. 6a, Additional file 1, Table S3). Significantly, most hits (312 strains) do not have the other genes for MUC biosynthesis, and some genera, such as Actinomyces, Granulicatella, Lachnospira, Oribacterium, and Scardovia, have never been reported to produce either MUC or RTCs. Even for S. mutans, the main producer of MUC and RTCs, only 37 of the 255 strains with a mucF homologue possess muc BGC, the other 218 strains may share competitive advantages by only expressing the MucF-like proteins. Further habitat analysis of those strains revealed that about 90% of them (329 strains) are from the human body and more than 70% (254 strains) are directly from the oral cavity. Interestingly, a small number of the Streptococcus competitors of S. mutans in the oral cavity, e.g., Streptococcus gordonii (3 of 85 strains) and Streptococcus mitis (2 of 174 strains), have also acquired the mucF homologues for defense.

Fig. 6
figure 6

Distribution of the MucF homologues characterized in this study. a The Venn diagram shows the number of mucF homologues containing strains from the human body (blue shade) and the oral cavity (red shade) or with a muc-like BGC (yellow shade). Strains with mucF homologues but without the muc-like BGCs were considered as “smart” bacteria. Oral cavity* represents the oral cavity of humans and two primate animals. b The diagram shows the proportion and number of “smart” bacteria in different species, with some known competitors of S. mutans marked in red font. MucF35 and the other 17 MucF homologous tested for deacylase activity in this work were marked with red dots (from strains containing a muc-like BGC) or circles (from strains without muc-like BGC). Details of each protein are described in Additional file 1, Table S3

Full size image

Sixteen of the MucF-like proteins were selected, synthesized, and tested using the same assays as those for MucF35, including six from non-mutans Streptococcus strains with the muc BGC and 10 from strains without the cluster to verify their functions (Fig. 6b, Additional file 1, Fig. S11, S12, Table S3). In the latter group, we specifically chose two MucF-like proteins from Actinomyces isolated from human dental plaque and liver abscess, respectively, one from Scardovia wiggsiae, a causative agent for early childhood dental caries, and one from S. gordonii, a known competitor of S. mutans. Results showed that all 16 MucF homologues could hydrolyze RTC into MUC regardless of whether they were from strains with or without the muc BGC, strongly supporting a wide distribution of the “smart” bacteria that have gained the ability to survive under the stress of RTCs in certain human-related environments.

Discussion

Bacterial origin MUC and RTCs are 3-acetylated TACs structurally similar to TeAs from fungi. As aforementioned, the pyrrolidine-2,4-dione core of TeAs and many other TACs are formed via Dieckmann cyclization. It was shown that MucD starts MUC biosynthesis by establishing M-283 whose d-isomer is further elongated and cyclized by MucE to form the pyrrolidine-2,4-dione moiety via a lactam bond formation mechanism to generate M-307 in this study. A clear difference between TACs synthesized via these two ways is that the pyrrolidine-2,4-dione moiety installed via Dieckmann cyclization bears an innate 3-acyl chain from its β-ketoacyl precursor. In contrast, those installed via lactam bond formation have no C-3 decoration unless added by tailoring enzymes (Fig. 2). Many naturally occurring TACs do not actually possess C-3 acyl appendage. They either have no C-3 decoration (e.g., belamide A [29], eliamid [30], or sintokamides [31]) or possess a simple C-3 methyl group (e.g., palauimide [32] and tetrapetalones [33]) (Additional file 1, Fig. S13). Further biosynthetic investigations of those natural products may reveal more TACs constructed via lactam bond formation. It was proposed that MucABC further acetylates M-307 to generate RTCs, which are finally converted to MUC by the MucF-catalyzed N-1 deacylation for MUC biosynthesis.

As mentioned above, RTCs are ionophore antibiotics capable of inhibiting the growth of many Gram-positive bacteria; MUC exhibits no antibiotic activity but can inhibit leukocyte chemotaxis and relieve the immune pressure from the host [22]. It can also suppress the hyphal formation and virulence of C. albicans [28], a familiar commensal existing in many body sites, such as the human oral cavity. The analysis of the 63 strains containing muc-like BGCs strongly supported that ecological adaptation is a driving force shaping the BGCs and their products. All type I BGCs with mucF are from Streptococcus strains dwelling in humans, primates, and livestock, which face the host immune pressure and drastic competition from neighboring microbes. Since zoonotic transmission of microbes including Staphylococcus aureus, Campylobacter spp., and L. reuteri has been documented [34,35,36], we cannot preclude the possibility that these muc-like BGCs containing strains isolated in primates and livestock may have been transmitted from humans. Strains harboring type II–IV BGCs are mainly isolated from fermented products, crops, and plants. Similarly, since some strains of L. reuteri and Lactiplantibacillus plantarum can permanently or transiently colonize the intestinal tract of humans and animals, they may also have the opportunity to obtain muc-like BGCs through HGT from bacteria like Streptococcus strains dwelling in humans. The missing of mucF in most BGCs of these strains may suggest that they no longer need to deal with immune pressure and, therefore, mainly synthesize the antibacterial product RTCs. Notably, some S. mutans strains can produce both RTCs (to inhibit the growth of a plethora of oral microbes, including Rothia mucilaginosa and some Streptococcus strains) and MUC (to modulate the host environment) [18], which may be achieved through the delicate control of the expression and/or catalytic activity of MucF. This phenomenon indicates that even within S. mutans strains containing the muc BGCs, there are differences in their ability to produce RTC/MUC. One limitation of this study is that we have not verified the function of muc-like BGCs in mixed communities. Exploring the impact of muc BGC-containing strains on colonization and abundance of S. mutans populations through in vitro oral biofilm or animal models will help us further understand the ecological role of muc BGCs.

Considering that about 2/3 of the 63 strains are human-associated bacteria, and most of the others are from fermented food, livestock, and crops, it was proposed that human beings play an essential role in spreading the muc-like BGCs. Such HGT events from human commensals (e.g., S. mutans) to closely related bacteria in human-related environments are supported by the presence of transposase or recombinase genes flanking the BGCs from strains other than S. mutans. After that, selection pressures from habitats of the receptor strains will reshape the BGCs, guiding them to produce MUC or RTCs. The four RTCs producing L. reuteri strains repeatedly isolated from industry sourdough of the same bakery may be the result of such an adaptation process (Fig. 2). It was reported that the bread dough prepared with 6–24% sourdough fermented with the RTCs producing L. reuteri strain could inhibit the bread spoilage caused by Bacillus and extend shelf-life [37]. Since Bacillus spp. can form heat-resistant endospores that are difficult to destroy while baking [38, 39], the thermal stable RTCs further highlight their beneficial effects. Therefore, the Lactobacillaceae strains lacking the mucF gene tend to be artificially selected for high-quality fermented food preparation. Such HGT events between S. mutans and Lactobacillales from human-related environments are not rare. S. mutans UA140 and Lactococcus lactis KF147 (from mung bean sprouts) share a similar TnSmu2 gene cluster capable of improving the host’s oxidative stress resistance capacity [40]. Research on the homologous BGCs shared by human commensals and microbes from human living environments may uncover more cases of similar adaptive evolution.

An interesting phenomenon is the widespread distribution of mucF-like genes in human-associated microorganisms. They dwell mainly in the same habitats as the bacteria containing a muc-like BGC, especially in the oral cavity. Moreover, all 11 tested MucF homologues from strains without the muc-like BGC worked well as RTC deacylases, indicating that those “smart” bacteria can benefit from the neighbor-made chemical weapons by hydrolyzing the toxic RTCs to protective MUC. Notably, among the 264 genome-sequenced S. mutans strains, 255 possessed a mucF homologue of which 37 contain the muc BGC and 218 do not, which may help the S. mutans species outcompete the oral flora and promote dental caries development. The mucF homologue can also be found in S. wiggsiae, which is another caries-associated agent that can exist in teeth plaque together with S. mutans [41, 42], implying a potential synergistic relationship between the two species. Meanwhile, a small group of the competitive bacteria of S. mutans, like several S. gordonii and S. mitis strains, also acquire the mucF-like gene during the long co-existence and combat with S. mutans in the oral cavity. Generally, the wide spread of mucF homologues in oral bacteria reveals how the muc BGC, a result of adaptive evolution in several S. mutans strains, further influences their “neighbors” in the same community, whether they are “friends” or “foes.” A similar phenomenon was observed in a recent work about the ecological role of colibactin [43], which can induce DNA damage and prophage induction. The clbs gene in its BGC encodes a cyclopropane hydrolase and plays a protective role. It was revealed that clbs homologues can be found in and protect non-colibactin-producing E. coli strains along with members of the human and honey bee gut microbiome.

TeAs from fungi are very similar to MUC in structure. Specifically, iso-TeA (5S) and MUC (5R) are a pair of enantiomers although they are biosynthesized through totally different routes. It was found that the TeA producers, e.g., Alternaria brassicicola and Alternaria raphani, could produce d-allo-TeA with 5R configuration as a minor product [44]. This implies that some fungi can synthesize MUC, which raises the question: why do fungi produce TeAs instead of MUC? To understand the impact of habitat environment on the stereochemical preference of TAS1, we analyzed the distribution of TeA synthetase using the TAS1 from P. oryzae 70–15 as a probe, which resulted in 97 retrievals spreading in two fungus phyla, Basidiomycota (12) and Ascomycota (85) (Additional file 1, Table S4). Considering that the phytotoxicity of TeAs may help fungi colonize and infect plants, it is not surprising that many TAS1-containing fungi, including many phytopathogenic fungal species like Aspergillus, Pyricularia, and Fusarium, are from plants or crops worldwide. Additionally, ten TAS1-containing fungi were from insects (e.g., Popillia japonica, Spodoptera litura, and Locusta migratoria manilensis) and/or their nests. Specifically, it was shown that l-TeA, AITeA, and d-allo-TeA produced by Alternaria strains from beehive pollens could protect bee brood from American foulbrood by inhibiting the pathogenic bacterium Paenibacillus larvae [44]. However, no TAS1-containing fungus has been isolated from a niche facing the immune pressure from vertebrates, although this result may be constrained by the number of fungi isolated and sequenced in vertebrates currently. MUC and TeAs with similar structures present a good example of how bacteria and fungi adopt different biosynthetic routes to generate enantiomers with distinct bioactivities to help the producers adapt to their habitats.

Conclusively, we showed that MUC closes the pyrrolidine-2,4-dione ring via lactam bond formation and proposed that such a non-Dieckmann cyclization mechanism may be adopted by many TACs without 3-acyl decoration. Comparisons of the producers, biological roles, and biosynthetic mechanisms of MUC, RTCs, and TeAs are summarized in Fig. 2. The figure depicts an interesting case of convergent evolution, in which ecological and evolutionary forces drive bacteria and fungi to assemble a common 3-acetylated pyrrolidine-2,4-dione core through totally different manners. In addition, it shows how fungi and bacteria can diversify the structures and activities of natural products with this general skeleton to adapt to environments and even influence the shape of their “neighbors.” It unveils a corner of the everlasting evolution and screening process of life and may encourage the use of 3-acetylated TACs to improve oral health and inspire the synthetic biology efforts to further expand the chemical space of TACs.

Conclusions

In this study, we demonstrate that the muc BGC produces mutanocyclin (MUC), an immunomodulator, and reutericyclins (RTCs), the antibacterial agent, through lactam bond formation. Interestingly, this differs from the usual Dieckmann cyclization observed in phytotoxic TeAs produced by TAS1, despite these compounds sharing the pyrrolidine-2,4-dione ring backbone. The muc-like BGCs mainly exist in human-related bacteria, with many of their “neighbors” acquiring MucF homologues detoxifying RTCs, while TAS1 mainly exists in fungi from plants or corps. Strains originating from diverse habitats produce 3-acetylated TACs through distinct biosynthetic routes, resulting in varying activities. These findings provide potential evidence of how ecological and evolutionary forces shape the behavior of microorganisms.

Materials and methods

Bacterial strains, plasmids and culture conditions

Bacterial strains and plasmids used in this study are listed in Additional file 1, Table S5. S. mutans strains were grown anaerobically on BHI agar plates or in BHI broth (BHI; OXOID LTD., Basingstoke, England) at 37 °C. The modified ASS (artificial saliva substitutes) medium was prepared as described [22] and used specifically for basic fermentation and feeding experiments. BHI plates containing 4 mg/mL p-Cl-Phe were used for the selection of p-Cl-Phe-sensitive colonies. For the selection of antibiotic-resistant colonies, BHI plates were supplemented with erythromycin (12.5 μg/mL) and LB plates were supplemented with chloramphenicol (27 μg/mL) and kanamycin (50 μg/mL).

General DNA manipulations

All PCR primers used in this study were synthesized in Genery Co., Ltd. (Shanghai, China) and listed in Additional file 1, Table S6. Common DNA sequencing and genome sequencing of S. mutans B30 was performed by Novogene (Beijing, China). The genes encoding different MucF homologous proteins were codon optimized and synthesized by Tsingke Biotechnology Co., Ltd. (Beijing, China). PCR reactions were carried out with Taq DNA polymerase (TransGene, Beijing, China) or Phusion® HF DNA polymerase (Thermo Fisher scientific, MA) according to the manufacturers’ instructions. Golden Gate cloning and overlapping PCR were performed following the literature procedures [45]. Preparation of genomic DNAs of different S. mutans strains was performed as described [22]. For natural transformation experiments, cells were maintained in Todd-Hewitt medium (Becton–Dickinson, Sparks, MD) supplemented with 0.3% yeast extract. A detailed natural transformation procedure was described previously [46].

Heterologous expression of the muc gene cluster in S. mutans UA140

The muc gene cluster (mucA–J) locates at a genomic island in S. mutans 35, and sequences flanking this genomic island are well conserved in S. mutans UA140. Therefore, the muc gene cluster was cloned to the same genome locus of S. mutans UA140 using the flanking sequences as homologous arms via the NabLC technique (Additional file 1, Fig. S2). To construct the recipient strain, the 2.1-kb IFDC2 cassette [47] that contains the erythromycin resistance gene (ermAM) and the counter-selective marker gene (pheS*) was PCR cloned with primer pair ldh-F/erm-R using pIFDC2 as a template. The 0.6-kb left capture arm (CAL) and the 0.6-kb right capture arm (CAR) were PCR amplified from S. mutans UA140 genomic DNA with primer pairs UA140-upF/UA140-upR-erm and UA140-dnF-ldh/UA140-dnR, respectively. The three fragments were mixed as the template for an overlapping PCR with primer pair UA140-upF/UA140-dnR to generate a 3.3-kb fragment CAL-IFDC2-CAR, which was then transformed to S. mutans UA140. Colonies resistant to erythromycin on BHI plates were selected and PCR-verified with primers UA140-upF1/UA140-dnR1 as the recipient strain S. mutans UA140-RS/MUC. Genomic DNAs of S. mutans 35 were then transformed to S. mutans UA140-RS/MUC via natural transformation, and the colonies that could grow on BHI plates containing 4 mg/mL p-Cl-Phe were selected as positive transformants. The positive transformants that could not grow on BHI plates with erythromycin were picked and PCR-verified using primers UA140-upF1/orfJ, bacA-F/bacA-R, orfE-F/orfE-R, and orfA/UA140-dnR1 as the S. mutans UM* (Additional file 1, Fig. S2).

To insert the xylose-inducible promoter xylS1P, the 1.4-kb fragment containing the xylS1P promoter was amplified from pZX9 [48] using primer pair XylRR-BsaI/XylOR-BsaI. The 2.1-kb IFDC2 cassette was amplified using pIFDC2 as a template with primer pair ermR-BsaI/ldhF-BsaI. Two 0.7-kb and 0.6-kb fragments flanking the insertion site were PCR cloned with primers orfA-upF/orfA-upR2-BsaI and orfA-dnF2-BsaI/orfA-dnR. The four fragments were ligated by Golden Gate cloning and transformed into S. mutans UM*. Colonies growing on BHI with erythromycin were selected and PCR-verified as S. mutans UM*/IFDC2-xylS1P with primers orfA-upF/orfA-dnR. The IFDC2 cassette was then removed by counterselection via homologous recombination. The 0.7-kb and 2.0-kb fragments flanking the IFDC2 cassette were PCR cloned with primers orfA-upF/orfA-upR3-BsaI and XylRR1-BsaI/orfA-dnR using S. mutans UM*/IFDC2-xylS1P genomic DNAs as a template. The two amplicons were ligated using Golden Gate cloning and transformed into S. mutans UM*/IFDC2-xylS1P. Colonies that were resistant to BHI plates containing 4 mg/mL p-Cl-Phe were selected and PCR-verified using primers orfA-upF/orfA-dnR as S. mutans UM (Additional file 1, Fig. S3).

Construction of the muc gene in-frame deletion or point mutated strains

Construction of the mucD gene in-frame deletion mutant (S. mutans UM ∆mucD) was completed via two rounds of homologous recombination (Additional file 1, Fig. S4a). The 0.8-kb fragment and 0.9-kb fragment flanking the mucD gene was PCR amplified with primers DUF/DUR and DDF/DDR using the genomic DNA of S. mutans UM as the template. The 2.1-kb IFDC2 cassette was PCR amplified with primer pair ldhF-BsaI/ermR-BsaI using pIFDC2 as a template. The two fragments flanking mucD and the IFDC2 cassette were ligated via Golden Gate cloning and transformed to S. mutans UM. Colonies resistant to erythromycin on BHI plates were selected and PCR-verified with primer pair DSupF/DSdnR as the mucD disrupted mutant, in which mucD was replaced with the IFDC2 cassette. The elimination of the IFDC2 cassette was achieved via the second round of homologous recombination. Briefly, the 0.8-kb fragment and 0.9-kb fragment flanking the IFDC2 cassette were PCR amplified with primers DUF/DUR and DDF2/DDR, ligated via Golden Gate cloning, and transformed to the mucD disrupted mutant. Colonies that could grow on BHI plates supplemented with p-Cl-Phe were selected, PCR-verified as S. mutans UM ∆mucD with primer pair DSupF/DSdnR (Additional file 1, Fig. S5), and confirmed by sequencing.

In vivo point mutagenesis of the C-domain of MucD was carried out using the same strategy as that of S. mutans UM ∆mucD to generate S. mutans UM mucD*(H135V) and mucD*(D139N), and different point mutations were introduced by the PCR primers at the second round of homologous recombination (Additional file 1, Fig. S4b). The 0.8-kb fragment and 1.0-kb fragment flanking mucD-C were PCR amplified with primer pairs CdHIUF/CdHIUR and CdHIDF/CdHIDR (to generate H135V mutation) and CdDIUF/CdDIUR and CdDIDF/ CdDIDR (to generate D139N mutation). The 2.1-kb IFDC2 cassette was amplified with primer pair ldhF-BsaI/ermR-BsaI. The three fragments were ligated and used for the first round of homologous recombination. For the removal of IFDC2 and the introduction of C domain point mutation, two 0.8-kb and 1.0-kb fragments were amplified with primer pairs CdHIUF/CdHUR and CdHDF/CdHIDR (to generate H135V mutation) and CdDIUF/CdDUR and CdDDF/CdDIDR (to generate D139N mutation). The MucD-C domain point mutations were PCR verified using primer pair CdmF/CdmR and confirmed by sequencing (Additional file 1, Fig. S5).

S. mutans UM ∆mucE was constructed in a similar manner, except that overlapping PCR technique, instead of Golden Gate cloning, was used in ligation of the two homologous fragments with the IFDC2 cassette (Additional file 1, Fig. S4c). The 0.8-kb and 0.6-kb fragments flanking mucE were amplified using primer pairs EUF/EUR and EDF/EDR, respectively. The 2.1-kb IFDC2 cassette was PCR amplified with primer pair ldhF/ermR. The mucE disrupted mutant was PCR verified using primer pair ESupF/ESdnR. For the IFDC2 cassette elimination, the 0.8-kb left homologous arm was amplified using primer pair EUF/EUR2, and the 0.6-kb right arm was amplified using primer pair EDF2/EDR. The successful construction of S. mutans UM ∆mucE was PCR verified with primer pair ESupF/ESdnR and confirmed by sequencing (Additional file 1, Fig. S5).

The mucF–J locus in-frame deletion mutant S. mutans UM ∆mucF–J was constructed via two rounds of homologous recombination similar to the constructions of S. mutans UM ∆mucE (Additional file 1, Fig. S4d). The 0.8-kb and 0.7-kb fragments flanking the mucF–J locus were PCR amplified with primer pairs JUF/JUR and FDF/FDR, respectively. The 2.1-kb IFDC2 cassette for counterselection was amplified using primer pair ldhF-MucJ/ermR1-MucF. The two fragments and the IFDC2 cassette were assembled by overlapping PCR to generate a 3.6-kb fragment for the construction of the mucF–J disrupted mutant, which was PCR-verified with primer pair MucJ-upF/MucF-dnR. For the IFDC2 cassette elimination, the 0.8-kb and 0.7-kb fragments were PCR amplified with primer pairs JUF/JUR2 and FDF2/FDR, ligated via overlapping PCR, and transformed to replace the IFDC2 cassette to generate S. mutans UM ∆mucF–J, which was PCR verified with primer pair MucJ-upF/MucF-dnR and confirmed by sequencing (Additional file 1, Fig. S5).

Production of MUC

S. mutans UM and all the mutants were inoculated into 100 mL ASS medium with 0.4% xylose and cultured statically at 37 °C for 48 h. The cell pellet was discarded after centrifugation (5000 × g, 30 min) and the supernatant was extracted twice with 100 mL EtOAc. The combined organic layers were evaporated in vacuum and re-dissolved in 1 mL methanol for HPLC detection. HPLC detection of MUC was carried out on a Shimadzu HPLC system (Shimadzu, Kyoto, Japan) using a C18 column (4.6 × 250 mm, 5 μm, Apollo, Alltech, Lexington, Kentucky, USA) developed with solvent A (H2O with 0.1% (v/v) formic acid) and B (acetonitrile) at a flow rate of 1.0 mL/min. The percentage of acetonitrile was kept at 5% over 0–5 min, changed from 5 to 40% over 5–25 min, from 40 to 100% over 25–40 min, and kept at 100% over 40–45 min.

For the feeding experiments, S. mutans UM ∆mucD or ∆mucE was inoculated into 100 mL ASS medium with 0.4% xylose and M-307 (0.03 mM). After being cultured statically at 37 °C for 48 h, the supernatant was treated and analyzed under the same conditions as the fermentation broth of S. mutans UM.

Expression and purification of MucE and PKsC

The 3.2-kb mucE gene was amplified as three fragments with primer pairs T7F/mucER1, mucEF2/mucER2, and mucEF3/T7R and assembled via overlapping PCR. After sequencing verification, the fragment was digested with NdeI/XhoI and inserted into the same sites of pET28a to generate pET28a-MucE, which was then transformed into E. coli BAP1 to generate E. coli BAP1/pET28a-MucE. A single transformant was inoculated into LB with 50 μg/mL kanamycin and cultured overnight at 37 °C, 220 rpm. The overnight seed was used to inoculate the same medium at 1:100 dilution and incubated at 37 °C, 220 rpm until OD600 reached 0.6. After the addition of 0.1 mM isopropyl-β-thiogalactoside (IPTG), the cells were further cultured at 16 °C, 180 rpm for 16–18 h. Purification of N-His6 tagged MucE was performed with the Ni–NTA affinity column at 4 °C following the manufacturer’s instructions.

Similarly, the 0.9-kb fragment containing the gene encoding PKsC was PCR cloned with primer pair PKsCF/PKsCR from Bacillus subtilis sp. 168 and verified by sequencing. The plasmid pET28a-PKsC was constructed by digesting the fragment with BamHI/EcoRI and inserting it into pET28a. Then the plasmid was transformed into E. coli BL21 (DE3) to afford E. coli BL21 (DE3)/pET28a-PKsC. Protein expression and purification of N-His6-tagged PKsC were carried out using the same procedures as those for MucE. Measurements of protein concentrations were performed with the Bradford assay using bovine serum albumin as a standard. The purified proteins were stored at − 80 °C in 20 mM Tris–HCl (pH 7.5) with 20% glycerol.

Enzymatic assays of MucE

A holo-MucE assay was conducted at 37 °C for 1 h in 200 μL reaction mixture containing 1 μM holo-MucE, 1 μM PKsC, 100 mM Tris–HCl (pH7.5), 0.1 mM Malonyl-CoA, and 0.25 mM d-M-283-SNAC or l-M-283-SNAC. The reaction mixtures were extracted with 200 μL EtOAc twice, evaporated in vacuum and dissolved in 200 μL methanol for HPLC detection at a wavelength of 254 nm. HPLC detection of the MucE enzymatic assays was carried out using a C18 column (4.6 × 250 mm, 5 μm, Apollo, Alltech, Lexington, Kentucky, USA) on a Shimadzu HPLC system developed with solvent A (H2O with 0.1% (v/v) formic acid) and B (acetonitrile) at a flow rate of 1.0 mL/min. The percentage of acetonitrile was kept at 20% over 0–5 min, changed from 20 to 90% over 5–25 min, and from 90 to 100% over 25–30 min.

Expression and activity testing of the MucF homologues

For in vitro characterization, the 0.5-kb mucF35 gene was amplified with primer pair pET-mucF-F/pET-mucF-R and inserted into pET28a at the NcoI/XhoI sites to generate pET28a-MucF35, which was then verified by sequencing and transformed into E. coli Rosetta 2(DE3)pLysS to construct E. coli Rosetta 2(DE3)pLysS/pET28a-MucF35. A single transformant was inoculated into TB broth (Thermo Fisher Scientific, USA) with 50 μg/mL kanamycin, 50 μg/mL chloramphenicol, and cultured overnight at 37 °C, 220 rpm. The overnight culture was used to inoculate in 1 L TB broth with the same antibiotics at 1:1000 dilution and incubated at 37 °C, 220 rpm until OD600 reached 0.6. The cells were then induced with 0.5 mM IPTG, cultured at 18 °C for a further 16 h, harvested by centrifugation, and re-suspended in 10 mL reaction buffer (50 mM Tris–HCl, 150 mM NaCl, 10% glycerol, pH 8.0) supplemented with 0.5 mg/mL lysozyme and 0.5 mM PMSF. After sonication at 4 °C, 96 μL lysate was mixed with 4 μL RTC solution (6.6 mM, 80% EtOH) and incubated at 37 °C for 30 min. The reaction was terminated by the addition of 1 μL acetic acid, extracted twice with 200 μL EtOAc, then evaporated in vacuum and dissolved in 200 μL methanol for LC–MS analysis. E. coli Rosetta 2(DE3)pLysS/pET28a was treated in the same procedure as a negative control. The encoding gene of MucFSgo was codon optimized and inserted into pET28a and its function was verified using the same method as E. coli Rosetta 2(DE3)pLysS/pET28a-MucF35.

For in vivo testing, mucF35 was codon optimized, synthesized, and inserted into pEXT06 (containing mucA–E for synthesizing RTCs in E. coli) to generate pEXT06-MucF35 and transformed into E. coli BAP1 to generate E. coli BAP1::mucA–E/MucF35. A single transformant was inoculated into LB with 1% glucose, 50 μg/mL chloramphenicol, and cultured at 37 °C, 220 rpm for 6 h. The culture was then inoculated into 50 mL LB with 1% glucose, 50 μg/mL chloramphenicol at 1:50 dilution, and incubated at 37 °C, 220 rpm until OD600 reached 0.4–0.6. Expression of protein was induced by the addition of IPTG at a final concentration of 0.2 mM and further cultured at 30 °C, 220 rpm for 12–14 h. After harvesting cells by centrifugation, 500 μL ddH2O was added to 0.1 g pellets, mixed with glass beads, and vortex 3 times for 15 min at room temperature. The lysates were then extracted twice with an equal volume of EtOAc (added with 1% acetic acid), evaporated in vacuum, and dissolved in 200 μL methanol for LC–MS analysis. E. coli BAP1::mucA–E (E. coli BAP1 containing unmodified pEXT06) was treated in the same procedure as a negative control. The construction of expression vectors and functional verification of other MucF homologues were also processed in the same way as the in vivo testing of mucF35 (except for mucFSgo whose insertion into pEXT06 failed; therefore, its function was characterized in vitro as described above).

Agilent 1290/6470 Triple-Quadrupole LC/MS system (Santa Clara, CA, USA) was used to analyze the reactions. The LC/TQ used a Jet Stream electrospray ionization source operated in the positive ionization mode. Samples were analyzed using a C18 column (2.1 × 100 mm, 2.7 μm, Agilent Technology Inc, Santa Clara, CA, USA) developed with solvent A (H2O) and B (acetonitrile) both with 0.1% (v/v) formic acid at a flow rate of 0.3 mL/min. The percentage of acetonitrile was changed from 10 to 100% over 0–15 min, kept at 100% over 15–20 min, and equilibrated at 5% over 20–25 min.

Chemical syntheses and separation of d/l-M-283-SNAC

The detailed synthesis procedures of d-M-283, l-M-283, d-M-436, l-M-436, M-307, M-283-SNAC, d-Leu-SNAC, and M-155 were described in Additional file 1. For the synthetic compounds, LC–MS analyses were performed on an Agilent 1260/6460 Triple-Quadrupole LC/MS system (Santa Clara, CA, USA) with an electrospray ionization source using a C18 column (4.6 × 250 mm, 5 μm, Apollo, Alltech, Lexington, Kentucky, USA). HR-ESI–MS was performed on an Agilent 1260 HPLC/6520 QTOF-MS instrument (Santa Clara, CA, USA). NMR spectra were recorded on a Bruker-500 NMR spectrometer (Billerica, MA). For the comparison between (S)-1-(naphthalen-2-yl)ethan-1-amine derivatives of isolated M-283 and synthesized standards d/l-M-436, HPLC detection was carried out on a Shimadzu HPLC system using a C18 column (4.6 × 250 mm, 5 μm, Apollo, Alltech, Lexington, Kentucky, USA) developed with solvent A (H2O with 0.1% (v/v) formic acid) and B (acetonitrile) at a flow rate of 1.0 mL/min. The percentage of acetonitrile was kept at 40% over 0–3 min, changed from 40 to 70% over 3–6 min and from 70 to 76% over 6–21 min, kept at 76% over 21–25 min, changed from 76 to 90% over 25–26 min and from 90 to 100% over 26–28 min, and kept at 100% over 28–34 min.

The synthetic M-283-SNAC is a mixture of two enantiomers. Chiral separation of M-283-SNAC was performed on an Agilent 1260 series HPLC system using a Chiralpak AD-H column (4.6 × 250 mm, 5 μm, Daicel, Tokyo, Japan) developed with solvent A (isopropanol) and B (hexane) at a flow rate of 1.0 mL/min. The percentage of hexane was kept at 95%. Circular dichroism (CD) measurements of the two enantiomers in methanol were then carried out on a Chirascan spectrometer (Applied Photophysics Ltd., UK) using quartz cells with a path length of 1 mm at 25 °C. Scans were conducted between wavelengths of 190 and 400 nm with a bandwidth of 1 nm, 1-nm step resolution, 100 nm/min scan speed, and 2-s response time. Assignment of the two enantiomers were performed by comparing their CD spectra with the calculated electronic circular dichroism (ECD) spectra of d-M-283-SNAC and l-M-283-SNAC (Additional file 1, Fig. S22), which were carried out with Conflex 7.0a, Gaussian9, and Shermo 2.3 as described [49].

Sequence and phylogenetic analyses

BLASTP search (https://blast.ncbi.nlm.nih.gov/Blast.cgi) with default parameters (nr database, BLOSUM62 matrix, expect threshold = 0.05, word size = 6) was used to assign protein functions in the muc gene cluster (Additional file 1, Table S1) or to identify homologous proteins. Specifically, MucF-like proteins were identified using the sequence of MucF35 as a query, and results with coverage above 90%, and identity above 50% were retrieved. AntiSMASH (v5.1.2) [50] (https://antismash.secondarymetabolites.org/#!/start) was used to analyze the genome sequences of strains containing the MucF-like proteins to investigate whether they have the muc gene cluster (Additional file 1, Table S3). TAS1-like proteins were identified using the protein sequence of TAS1 from P. oryzae 70–15 as a probe, and results with coverage above 80%, and identity above 30% were retrieved. The genome sequences of corresponding strains were also analyzed by antiSMASH to confirm the integrity of TAS1 and the results were manually checked (truncated sequences or sequences containing unrelated domains were discarded; considering that the TAS1 homologous sequences in plant genomes may be contamination from fungi, three sequences from Quercus suber HL8 were excluded) (Additional file 1, Table S4).

For the phylogenetic analysis of MucD-C, BGCs containing different subtypes of C domains were downloaded from MIBiG (v2.0) [51] (https://mibig.secondarymetabolites.org/) and analyzed by HMMER (v3.3.2) [52] (https://www.ebi.ac.uk/Tools/hmmer/) to obtain the C-domain sequences. All of the sequences were provided as a source data file. Multiple sequence alignments were generated using MUSCLE [53] (https://www.ebi.ac.uk/Tools/msa/muscle/) with default parameters and manually trimmed. The output file was converted to.phy format by Mesquite (v3.61) and used as the input file for the phylogenetic analysis carried out using PhyML 3.0 [54] (http://www.atgc-montpellier.fr/phyml) with SMS (v1.8.4). The phylogenetic tree was visualized using iTOL (v6) [55] (https://itol.embl.de/).

To explore the distribution of muc-like BGC, Cblaster (v1.3.12) [56] was performed using the MucA–J protein sequences from S. mutans 35 as queries with the following parameters: cblaster search -qf mucAJ.fasta -mi 20 -p plot.html -o summary.csv -s session.json. The results of all the retrieved gene clusters were extracted through the extract_clusters parameter, clusters containing mucA–E homologous genes were manually checked, and the corresponding genomes were further analyzed by antiSMASH. Multiple nucleotide sequence alignments for the 63 resultant BGCs (ranging from mucA to mucE) were generated using MAFFT [53] (https://www.ebi.ac.uk/Tools/msa/mafft/) with default settings. Phylogenetic analysis of the resulting output file was carried out similarly to MucD-C.

Availability of data and materials

Data supporting the findings of this work are available within the paper and the Additional file 1. The genome sequence of S. mutans B30 has been deposited into GenBank with the accession number JAGEVH000000000.

References

  1. ZygonRoyles BJ. Naturally occurring tetramic acids: structure, isolation, and synthesis. Chem Rev. 1995;95:1981–2001.

    Article  Google Scholar 

  2. Mo X, Li Q, Ju J. Naturally occurring tetramic acid products: isolation, structure elucidation and biological activity. RSC Adv. 2014;4:50566–93.

    Article  CAS  Google Scholar 

  3. Schobert R, Schlenk A. Tetramic and tetronic acids: an update on new derivatives and biological aspects. Biorg Med Chem. 2008;16:4203–21.

    Article  CAS  Google Scholar 

  4. Jiang M, Chen S, Li J, Liu L. The biological and chemical diversity of tetramic acid compounds from marine-derived microorganisms. Mar Drugs. 2020;18:114.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Petermichl M, Schobert R. 3-Acyltetramic acids: a decades-long approach to a fascinating natural product family. Synlett. 2017;28:654–63.

    Article  CAS  Google Scholar 

  6. Ishida T, Kobayashi R, Yamada T. Novel method of tetramic acid synthesis: silver-catalyzed carbon dioxide incorporation into propargylic amine and intramolecular rearrangement. Org Lett. 2014;16:2430–3.

    Article  CAS  PubMed  Google Scholar 

  7. Chen S, Qiang S. Recent advances in tenuazonic acid as a potential herbicide. Pestic Biochem Physiol. 2017;143:252–7.

    Article  CAS  PubMed  Google Scholar 

  8. Brück E, et al. Movento®, an innovative ambimobile insecticide for sucking insect pest control in agriculture: biological profile and field performance. Crop Prot. 2009;28:838–44.

    Article  Google Scholar 

  9. Mo X, Gulder TA. Biosynthetic strategies for tetramic acid formation. Nat Prod Rep. 2021;38:1555–66.

    Article  CAS  PubMed  Google Scholar 

  10. Kakule TB, Sardar D, Lin Z, Schmidt EW. Two related pyrrolidinedione synthetase loci in Fusarium heterosporum ATCC 74349 produce divergent metabolites. ACS Chem Biol. 2013;8:1549–57.

    Article  CAS  PubMed  Google Scholar 

  11. Antosch J, Schaefers F, Gulder TA. Heterologous reconstitution of ikarugamycin biosynthesis in E. coli. Angew Chem Int Ed. 2014;53:3011–4.

    Article  CAS  Google Scholar 

  12. Mo X, et al. Cloning and characterization of the biosynthetic gene cluster of the bacterial RNA polymerase inhibitor tirandamycin from marine-derived Streptomyces sp. SCSIO1666. Biochem Biophys Res Commun. 2011;406:341–7.

    Article  CAS  PubMed  Google Scholar 

  13. Olano C, et al. Deciphering biosynthesis of the RNA polymerase inhibitor streptolydigin and generation of glycosylated derivatives. Chem Biol. 2009;16:1031–44.

    Article  CAS  PubMed  Google Scholar 

  14. Wu Q, Wu Z, Qu X, Liu W. Insights into pyrroindomycin biosynthesis reveal a uniform paradigm for tetramate/tetronate formation. J Am Chem Soc. 2012;134:17342–5.

    Article  CAS  PubMed  Google Scholar 

  15. Liu X, Walsh CT. Cyclopiazonic acid biosynthesis in Aspergillus sp.: characterization of a reductase-like R* domain in cyclopiazonate synthetase that forms and releases cyclo-acetoacetyl-L-tryptophan. Biochemistry. 2009;48:8746–57.

    Article  CAS  PubMed  Google Scholar 

  16. Tokuoka M, et al. Cyclopiazonic acid biosynthesis gene cluster gene cpaM is required for speradine A biosynthesis. Biosci Biotechnol Biochem. 2015;79:2081–5.

    Article  CAS  PubMed  Google Scholar 

  17. Lin XB, et al. Genetic determinants of reutericyclin biosynthesis in Lactobacillus reuteri. Appl Environ Microbiol. 2015;81:2032–41.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Tang X, et al. Cariogenic Streptococcus mutans produces tetramic acid strain-specific antibiotics that impair commensal colonization. ACS Infect Dis. 2020;6:563–71.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Rosett T, Sankhala R, Stickings C, Taylor M, Thomas R. Studies in the biochemistry of micro-organisms. 103. Metabolites of Alternaria tenuis Auct.: culture filtrate products. Biochem J. 1957;67:390.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Schlegel B, Schmidtke M, Dörfelt H, Kleinwächter P, Gräfe U. (–)-Terpestacin and l-tenuazonic acid, inducers of pigment and aerial mycelium formation by Fusarium culmorum JP 15. J Basic Microbiol. 2001;41:179–83.

    Article  CAS  PubMed  Google Scholar 

  21. Miller FA, et al. Antiviral activity of tenuazonic acid. Nature. 1963;200:1338–9.

    Article  CAS  PubMed  Google Scholar 

  22. Gitterman C. Antitumor, cytotoxic, and antibacterial activities of tenuazonic acid and congeneric tetramic acids1. J Med Chem. 1965;8:483–6.

    Article  CAS  PubMed  Google Scholar 

  23. Yun C-S, Motoyama T, Osada H. Biosynthesis of the mycotoxin tenuazonic acid by a fungal NRPS-PKS hybrid enzyme. Nat Commun. 2015;6:1–9.

    Article  Google Scholar 

  24. Hao T, et al. An anaerobic bacterium host system for heterologous expression of natural product biosynthetic gene clusters. Nat Commun. 2019;10:1–13.

    Article  Google Scholar 

  25. Ganzle MG, Holtzel A, Walter J, Jung GN, Hammes WP. Characterization of reutericyclin produced by Lactobacillus reuteri LTH2584. Appl Environ Microbiol. 2000;66:4325–33.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Qi F, Chen P, Caufield PW. The group I strain of Streptococcus mutans, UA140, produces both the lantibiotic mutacin I and a nonlantibiotic bacteriocin, mutacin IV. Appl Environ Microbiol. 2001;67:15–21.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Baker JL, Tang X, LaBonte S, Uranga C, Edlund A. mucG, mucH, and mucI modulate production of mutanocyclin and reutericyclins in Streptococcus mutans B04Sm5. J Bacteriol. 2022;204: e0004222.

    Article  PubMed  Google Scholar 

  28. Tao L, et al. Streptococcus mutans suppresses filamentous growth of Candida albicans through secreting mutanocyclin, an unacylated tetramic acid. Virulence. 2022;13:542–57.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  29. Simmons TL, McPhail KL, Ortega-Barría E, Mooberry SL, Gerwick WH. Belamide A, a new antimitotic tetrapeptide from a Panamanian marine cyanobacterium. Tetrahedron Lett. 2006;47:3387–90.

    Article  CAS  Google Scholar 

  30. Höfle G, et al. Isolation, biological activity evaluation, structure elucidation, and total synthesis of eliamid: a novel complex I inhibitor. Chem Eur J. 2012;18:11362–70.

    Article  PubMed  Google Scholar 

  31. Sadar MD, et al. Sintokamides A to E, chlorinated peptides from the sponge Dysidea sp. that inhibit transactivation of the N-terminus of the androgen receptor in prostate cancer cells. Org Lett. 2008;10:4947–50.

    Article  CAS  PubMed  Google Scholar 

  32. Bai W-J, Jackson SK, Pettus TR. Mild construction of 3-methyl tetramic acids enabling a formal synthesis of palau’imide. Org Lett. 2012;14:3862–5.

    Article  CAS  PubMed  Google Scholar 

  33. Komoda T, Sugiyama Y, Hirota A. Biosynthesis of tetrapetalones. Org Biomol Chem. 2007;5:1615–20.

    Article  CAS  PubMed  Google Scholar 

  34. Algammal AM, et al. Methicillin-Resistant Staphylococcus aureus (MRSA): one health perspective approach to the bacterium epidemiology, virulence factors, antibiotic-resistance, and zoonotic impact. Infect Drug Resist. 2020;13:3255–65.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Man SM. The clinical importance of emerging Campylobacter species. Nat Rev Gastroenterol Hepatol. 2011;8:669–85.

    Article  CAS  PubMed  Google Scholar 

  36. Li F, et al. A phylogenomic analysis of Limosilactobacillus reuteri reveals ancient and stable evolutionary relationships with rodents and birds and zoonotic transmission to humans. BMC Biol. 2023;21:1–17.

    Article  Google Scholar 

  37. Li Z, Siepmann FB, Rojas Tovar LE, Chen X, Ganzle MG. Effect of copy number of the spoVA(2mob) operon, sourdough and reutericyclin on ropy bread spoilage caused by Bacillus spp. Food Microbiol. 2020;91: 103507.

    Article  CAS  PubMed  Google Scholar 

  38. Pepe O, Blaiotta G, Moschetti G, Greco T, Villani F. Rope-producing strains of Bacillus spp. from wheat bread and strategy for their control by lactic acid bacteria. Appl Environ Microbiol. 2003;69:2321–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Fangio MF, Roura SI, Fritz R. Isolation and identification of Bacillus spp. and related genera from different starchy foods. J Food Sci. 2010;75:M218–21.

    Article  CAS  PubMed  Google Scholar 

  40. Golomb BL, Yu AO, Coates LC, Marco ML. The Lactococcus lactis KF 147 nonribosomal peptide synthetase/polyketide synthase system confers resistance to oxidative stress during growth on plant leaf tissue lysate. MicrobiologyOpen. 2018;7: e00531.

    Article  PubMed  Google Scholar 

  41. Tanner A, et al. Cultivable anaerobic microbiota of severe early childhood caries. J Clin Microbiol. 2011;49:1464–74.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Kressirer CA, et al. Scardovia wiggsiae and its potential role as a caries pathogen. J Oral Biosci. 2017;59:135–41.

    Article  PubMed  PubMed Central  Google Scholar 

  43. Silpe JE, Wong JW, Owen SV, Baym M, Balskus EP. The bacterial toxin colibactin triggers prophage induction. Nature. 2022;603:315–20.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. Gallardo GL, Peña NI, Chacana P, Terzolo HR, Cabrera GM. L-Tenuazonic acid, a new inhibitor of Paenibacillus Larvae. World J Microbiol Biotechnol. 2004;20:609–12.

    Article  CAS  Google Scholar 

  45. Engler C, Gruetzner R, Kandzia R, Marillonnet S. Golden gate shuffling: a one-pot DNA shuffling method based on type IIs restriction enzymes. PLoS ONE. 2009;4: e5553.

    Article  PubMed  PubMed Central  Google Scholar 

  46. Xie Z, Qi F, Merritt J. Cloning-independent plasmid construction for genetic studies in streptococci. J Microbiol Methods. 2013;94:77–82.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  47. Xie Z, Okinaga T, Qi F, Zhang Z, Merritt J. Cloning-independent and counterselectable markerless mutagenesis system in Streptococcus mutans. Appl Environ Microbiol. 2011;77:8025–33.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  48. Xie Z, Qi F, Merritt J. Development of a tunable wide-range gene induction system useful for the study of streptococcal toxin-antitoxin systems. Appl Environ Microbiol. 2013;79:6375–84.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Wang D, et al. Assigning the stereochemical structures of aurantinin A and B with the assistance of biosynthetic investigations. Org Chem Front. 2022. https://doi.org/10.1039/D2QO01251K.

    Article  Google Scholar 

  50. Blin K. antiSMASH 5.0: updates to the secondary metabolite genome mining pipeline. Nucleic Acids Res. 2019;47:W81–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  51. Kautsar SA, et al. MIBiG 2.0: a repository for biosynthetic gene clusters of known function. Nucleic Acids Res. 2020;48:D454–8.

    PubMed  Google Scholar 

  52. Potter SC, et al. HMMER web server: 2018 update. Nucleic Acids Res. 2018;46:W200–4.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. Madeira F, et al. Search and sequence analysis tools services from EMBL-EBI in 2022. Nucleic Acids Res. 2022;50:W276–9.

    Article  PubMed  PubMed Central  Google Scholar 

  54. Guindon S, et al. New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0. Syst Biol. 2010;59:307–21.

    Article  CAS  PubMed  Google Scholar 

  55. Letunic I, Bork P. Interactive Tree of Life (iTOL) v5: an online tool for phylogenetic tree display and annotation. Nucleic Acids Res. 2021;49:W293–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  56. Gilchrist CL, et al. cblaster: a remote search tool for rapid identification and visualization of homologous gene clusters. Bioinforma Adv. 2021;1:1–10.

    Article  Google Scholar 

Download references

Acknowledgements

We thank Dr. Jinwei Ren and Dr. Wenzhao Wang from IM, CAS for technical supports. We are grateful for the support from mass spectrometry facilities in Shenzhen Bay Laboratory.

Funding

This study was supported in part by the MOST of China (2020YFA0907703), China NSFC (32025002), Biological Resources Program, CAS (KFJ-BRP-009), Transformation Program of Scientific and Technological Achievements of Inner Mongolia (2020CG0012), Guangdong Province’s Pearl River Recruitment Program of Talents (2021QN02Y855), and Shenzhen Bay Laboratory Funds (21230051 and SZBL2021080601007).

Author information

Author notes

  1. Yuwei Zhang, Ge Liao, Min Wang, and Zhao Zhang contributed equally.

Authors and Affiliations

  1. State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, Beijing, 100101, China

    Yuwei Zhang, Min Wang, Zhao Zhang, Yuqin Song, Dacheng Wang, Tingting Hao, Jie Feng & Yihua Chen

  2. University of Chinese Academy of Sciences, Beijing, 100049, China

    Yuwei Zhang, Tingting Hao, Jie Feng & Yihua Chen

  3. Institute of Chemical Biology, Shenzhen Bay Laboratory, 518132, Shenzhen, China

    Ge Liao & Xiaoyu Tang

  4. School of Biotechnology and Health Sciences, Wuyi University, Jiangmen, 529020, Guangdong, China

    Min Wang

  5. Renal Division, Peking University First Hospital, Peking University Institute of Nephrology, Beijing, 100034, China

    Zhao Zhang

  6. Department of Marine Pharmacy, College of Food and Pharmaceutical Sciences, Ningbo University, Ningbo, 315211, China

    Liwei Liu

  7. Department of Pediatric Dentistry, Peking University School and Hospital of Stomatology, Beijing, 100081, China

    Bin Xia

  8. Central Laboratory Peking University School and Hospital of Stomatology, Beijing, 100081, China

    Yixiang Wang

Authors
  1. Yuwei Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Ge Liao
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Min Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Zhao Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Liwei Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Yuqin Song
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Dacheng Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Tingting Hao
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Jie Feng
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Bin Xia
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Yixiang Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Xiaoyu Tang
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Yihua Chen
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

Y.Z., G.L. and Z.Z. performed most of the experiments. M.W. determined the structures of related compounds and conducted chemical syntheses. Y.Z. and Z.Z. performed the analyses with help from L.L and Y.S. D.W performed the ECD calculation. Z.Z. constructed and analysed the mutant strains with help from T.H. Y.C. and X.T. conceived the project. Y.C., Y.Z. and Z.Z. wrote the paper, with input from X.T., Y.W., B.X. and J.F. The author(s) read and approved the final manuscript.

Corresponding authors

Correspondence to Xiaoyu Tang or Yihua Chen.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1.

Supplementary Tables, Supplementary Figures.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhang, Y., Liao, G., Wang, M. et al. Human-associated bacteria adopt an unusual route for synthesizing 3-acetylated tetramates for environmental adaptation. Microbiome 11, 97 (2023). https://doi.org/10.1186/s40168-023-01548-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s40168-023-01548-4

Keywords

Microbiome

ISSN: 2049-2618

Contact us