Genome Mining Reveals the Largely Untapped Biosynthetic Potential of Archaeal RiPPs

Our previous genomic analysis indicated that archaea contain a relatively small number of BGCs per genome compared to the extensively studied and well-documented bacterial counterparts.^[31] They encode a broad range of SM classes, such as RiPPs, terpenes, nonribosomal peptides (NRPs), and polyketides (PKs). Our initial fermentation-based discovery identified a new lanthipeptide from Haloarchaea, marking the first discovery of lantibiotic in the archaeal domain.^[31] In the current study, we further applied the antiSMASH 7.0 tool^[34] to examine 11644 archaeal genomes from the NCBI database to delve deeper into the biosynthetic landscape, diversity, and novelty of archaeal RiPPs. From 3451 representative archaeal genomes, 7731 BGCs were identified, mainly from phyla Halobacteriota (3379, 43.7%), Thermoplasmatota (1359, 17.6%), Methanobacteriota (1271, 16.4%), Thermoproteota (962, 12.4%), and Nanoarchaeota (221, 2.9%) (Figure 1a and Supplementary Data 1). Among them, RiPPs (3599, 46.6%), Terpenes (1990, 25.7%), and NRPs (581, 7.5%) emerged as dominant BGCs in the domain of Archaea (Figure 1a). These BGCs exhibited significant novelty compared to known BGCs from MIBiG,^[35] indicating a largely unexplored source for natural product discovery.

Phylogenetic distribution analysis unveiled a remarkable diversity within RiPP BGCs, widely distributed across the Asgard, TACK, and DPANN superphyla and predominantly within Euryarchaeota (Figure 1b). Among these abundant RiPP classes are YcaO-related RiPPs (thioamitides and thiopeptides), rSAM-dependent enzyme-modified RiPPs (ranthipeptides and sactipeptides), lanthipeptides, and lassopeptides (Figure 1b). Although RiPP BGCs are widely distributed throughout different phyla/classes, the scarcity of identified archaeal RiPP products hinders us from elucidating their unique chemical structure and putative ecological functions. Unlike other archaeal RiPP BGCs, class II lanthipeptide BGCs, defined by the presence of the lanthipeptide synthase LanM, are exclusively distributed in Haloarchaea (refer to class Halobacteria) within our dataset, hinting at a niche-specific association (Figure 1b, Supplementary Data 1 and 2). Consequently, we constructed an updated library in silico to decipher the chemical landscape of archaeal lanthipeptides. This library contains 353 putative precursors (LanAs) and 110 lanthipeptide synthase LanMs from 102 archaeal lanthipeptide BGCs (Supplementary Data 2). Among these, 67 deduplicated archaeal LanMs, exclusively present in Haloarchaea, are mainly from the genera Halorussus (26, 38.8%) and Haloferax (14, 20.9%), followed by Haladaptatus (5, 7.5%), Halobacteriales (5, 7.5%), and Natrinema (5, 7.5%) (Figure S1a and Supplementary Data 2). Moreover, 196 unique LanA sequences have been categorized into 157 families through sequence similarity network (SSN) analysis (Figure 1c), highlighting untapped chemical space for lanthipeptide discovery.

Co-Evolved Archaeal LanMs and LanAs Exhibit Environmental Adaptation

A collective analysis of the phylogenetic tree, amino acid composition bias, SSN, and sequence logo indicates distinct chemical signatures of archaeal lanthipeptides compared to their bacterial counterparts (Figure 1 and Figures S1 and S2). The phylogenetic tree of archaeal LanMs highlighted their evolutionary divergence from bacterial LanMs (Figure S1b and Supplementary Data 2). The tree of archaeal LanAs exhibited two primary clades: one encompassing our previously characterized LanAs, Alnα and Alnβ, and another unexplored clade that offers prospects for further investigation (Figure S2). Additionally, archaeal LanAs are predicted to be shorter,^[31] with a medium length of 51 amino acids (aa) compared to 58 aa for bacterial LanAs (Mann–Whitney U test, P < 0.001, Supplementary Data 2). Furthermore, archaeal LanAs comprise a higher portion of C and S instead of T residues than bacterial LanAs (Figure 1c,d), which are crucial conserved residues for thioether ring formation in lanthipeptide biosynthesis. More importantly, the unique amino acid composition bias observed in archaeal LanAs and LanMs, as opposed to their bacterial counterparts, supports the hypothesis that they underwent eco-evolution in response to hypersaline environments (Figure 1d,e). Specifically, archaeal LanMs and LanAs evolved to over-represent D, V, R, and P residues while under-representing K, I, N, Q, M, L, W, and F residues (Figure 1d,e). The increased abundance of D/E residues and the reduced portion of K residue is linked to the “salt-in” strategy of Haloarchaea, enabling interactions with cations in their high-salinity environment.^{[17, 36]} Furthermore, the reduced occurrence of hydrophobic amino acids (I, L, M, W) may increase random-coil structure formation, thereby preventing protein aggregation or inactivation in low-water environments. The unusually high prevalence of basic R residues in archaeal LanMs and LanAs may represent a distinct evolutionary signature and potential functional determinant of archaeal lanthipeptides. To our knowledge, this distinctive feature has not been observed in any other reported halophilic enzyme systems or proteomic studies to date.^[36] These results imply that archaeal LanAs, co-evolving with LanMs as a part of environmental adaptation (Figure 1f), may enhance the chemical diversity of lanthipeptides exclusive to Haloarchaea, thereby playing significant ecological roles in halophilic environments.

Heterologous Expression-based Discovery of Canonical Archaeal Lanthipeptides

Moving beyond the traditional method of natural product discovery through wild-type cultivation, our initial attempts to heterologously express archaeal lanthipeptide BGCs in the model bacterial hosts (Escherichia coli and Bacillus subtilis) were unsuccessful. This result was associated with our evolutionary and amino acid bias analysis, suggesting a possible unique biosynthetic characteristic in archaeal lanthipeptides. To solve this problem, we utilized heterologous expression in haloarchaea, Haloferax volcanii H1424, using the pTA1228 vector (Figure 2a, Table S2 and Supplementary Data 3).^{[37, 38]} The efficacy of this procedure was confirmed by the successful expression of our previously validated lanthipeptide BGCs,^[31] alnα and alnβ, which resulted in the production of corresponding analogs of the natural lanthipeptides, archalan α2 (7), β1 (9), and β2 (10) (Figure 2b,c and Figures S3 and S4). It marks the first successful validation of archaeal BGCs via heterologous expression (Figure 2), laying the groundwork for subsequent expression of additional archaeal lanthipeptide BGCs.

Based on this approach, we selected eleven additional BGCs for investigation and detected putative lanthipeptide signals in the crude extracts of nine of the corresponding recombinant strains. Among these, seven BGCs were successfully linked with their canonical lanthipeptide products, including sallans from sal BGC, ciblans from cib1 and cib2 BGCs, pellan from pel BGC, archalan γ from alnγ BGC, medlan B from medb BGC, and litlans from lit BGC (Figures 1c and 2, and Figures S5–S36). It is worth mentioning that sallans, pellan, and litlans were absent in their corresponding wild-type strain fermentation extracts, highlighting the effectiveness of the archaeal heterologous expression system in unearthing cryptical archaeal metabolites.

To pinpoint low-yield BGC-encoded peptides within the metabolomic data, we mapped the calculated mass data, which was predicted based on lanthipeptide post-translational modification rules,^[33] with the high-resolution mass spectrometry (HRMS) data to identify corresponding hits (Figure 2a). For instance, HRMS analysis of the crude extracts from the recombinant strain harboring the BGC sal (Halorussus salinus YJ-37-H) exposed low-molecular-weight peptidic signals (Figure S5). The MS1 matching process revealed that a hit of [M + 2H]²⁺  =  379.6688 (2) matched the calculated mass data well of “CTRYSF” with one dehydration (calculated [M + 2H]²⁺ = 379.6682), linking this low-yield peptide to precursor SalA (Figure 2a and Figure S5). Furthermore, the MS/MS data confirmed a C-S crosslink between Cys1 and Ser5 (b₅, -H₂O, -18 Da) in 2 (Figure S5). Additionally, sallan A1 (1, observed [M + 2H]²⁺ = 437.1816, calculated [M + 2H]²⁺ = 437.1816, Δ  = 0.0 ppm) was predicted to be an analog of 2, differing by an extra D residue at the N-terminus (y₂, “FD” residue, Figure S5). More importantly, 2D nuclear magnetic resonance (NMR) data of purified 2 confirmed the thioether ring in Cys1-Ser5 (δ_H = 2.96, 3.31 ppm to δ_C = 35.9 ppm and δ_H = 3.07, 3.24 ppm to δ_C = 34.3 ppm) (Figure 3, Figure S6 and Table S3), which was consistent with the result of metabolomic analysis. The stereochemical analysis confirmed that all unmodified amino acids of 2 were of L configuration and the Lan residue was of DL (2S,6R) configuration (Figure 3a, Figure S7, and Table S4). Notably, 2 with only six amino acids represents the smallest natural lanthipeptide to date.

Ciblan A1 (3, BGC cib1, Haladaptatus cibarius JCM 19505) and ciblan A2 (4, BGC cib2, H. denitrificans DSM 4425) were predicted with a similar modification on their corresponding core peptides (Figure S8). The modification includes bicyclic ring structure and one unsaturated amino acid Dhb residue (Figure 2), evidenced by a series of b ions (b₆/b₅ for the first ring and b₉/b₈ for the second ring in ciblan A1/A2) and z₂ fragment in MS/MS analysis (Figure S8). The NMR data of purified 3 exhibited the presence of Dhb residue (C_α = 131.0 ppm, C_β = 130.9 ppm, H_β = 6.66 ppm, C_γ = 13.4 ppm, H_γ = 1.86 ppm), along with two thioether rings in Ser2-Cys6 (δ_H = 2.97, 2.92 ppm of Ala6 to δ_C = 35.0 ppm of Ala2) and Cys7-Thr9 (δ_H = 2.98 ppm of Abu9 to δ_C = 41.2 ppm of Ala7) (Figure 3a, Figure S9 and Table S3). The NMR-verified structure further supported the deductions from the MS/MS analysis (Figure S8). The stereochemical analysis confirmed that all unmodified amino acids of 3 were of L configuration and the thioether rings were of DL-Lan (2S,6R) and DL-MeLan (2S,3S,6R) configurations (Figure 3a, Figure S7, and Table S4).

The full desulfurization and MS/MS analysis of pellan (5, BGC pel, H. pelagicus RC-68) deduced the presence of two dehydrations within residues Thr2, Cys4, Ser7, and Cys12, and proved intertwined ring structure in the core peptide of PelA (STGCYISDKPVCI, Figure S10). The partial desulfurization and dithiothreitol reduction products of archalan γ (6, BGC alnγ, H. salinus YJ-37-H) supported one methylation on Cys1, a disulfide bond between Cys1 and Cys8 and C-S crosslinking in Ser2-Cys3 and Thr5-Cys9 (CSCYTRICCDIDS, Figures S11–S17). This structure expands the limited number of lanthipeptides possessing the smallest thioether ring formed by the adjacent Cys and Ser/Thr residues.^[39] Archalan α2 (7, BGC alnα, H. salinus YJ-37-H) and medlan B (8, BGC medb, H. mediterranei ATCC 33500) contained two thioether rings, supported by their key MS fragments of b₄ and b₉ ions (Figure S3), in line with the previously reported structure of archalan α.^[31] Archalan β1–2 (9 and 10, BGC alnβ, H. salinus YJ-37-H) and litlan A1-2 (11 and 12, BGC lit, H. litoreus JCM31109) were deduced to possess thioether crosslinking patterns similar to the previously reported archalan β.^[31] Their putative structures feature intertwined thioether rings, with MS/MS fragments originating from the N/C-terminal residues outside the rings (Figures S4 and S18).

The identified structures of archaeal lanthipeptides emphasize the advantages and feasibility of utilizing heterologous expression in conjunction with biosynthetic rule-guided metabolomic analysis to unveil the latent chemistry in Archaea, particularly those with limited yields. Furthermore, the chemical compositions of mature archaeal lanthipeptides demonstrate that they are typically shorter than bacterial lanthipeptides, such as nukacin ISK-1 and haloduracins (Figure 2d). Despite being short in length, their bicyclic and intertwined crosslinking patterns and modifications (e.g., disulfide bond, methylation, dehydration, hydroxylation) are as intricate as those found in bacterial lanthipeptides.

Characterization of Unique Archaeal Lanthipeptides Featuring Unprecedented Maturation

Apart from the classical lanthipeptides, we detected non-canonical peptide signals originating from the meda and lar BGCs (Figure 3 and Figures S19–S29). These signals were identified in the extracts of both wild-type and recombinant strains, confirming their biosynthetic connection to these BGCs. Unlike classical lanthipeptides, which can be precisely characterized through MS/MS analysis to predict core peptide sequences and modifications, these peptide signals provided only short amino acid fragments lacking consistent patterns. For example, an “FD” fragment from the C-terminus of LarA (CPTCDF), a “Y” residue also from C-terminus, and an “S” residue from N-terminus (STCTY) were identified (Figure 3b and Figure S24).

To fully characterize their structures, 3.2 mg of 13 (observed [M + 3H]³⁺ = 415.4769, calculated [M + 3H]³⁺ = 415.4772, Δ  = 0.7 ppm) from the wild-type strain, 3.5 mg of 14 (observed [M + 2H]²⁺ = 631.2194, calculated [M + 2H]²⁺ = 631.2198, Δ  = 0.6 ppm), and 2.0 mg of 15 (observed [M + 2H]²⁺ = 652.2301, calculated [M + 2H]²⁺ = 652.2251, Δ  = 7.7 ppm) from the recombinant strain were purified, which were named medlan A1 (13), larlan A2 (14), and larlan A5 (15), respectively. Further NMR spectroscopy analysis revealed that these lanthipeptides comprised two antiparallel peptide chains crosslinked by thioether bonds, thereby defining them as a novel lanthipeptide subfamily (Figure 3b, Figures S20–S22 and Table S3). These crosslinked patterns expose diamino-dicarboxylic termini with certain amino acids within the thioether rings being partially cleaved, leading to the name of DADC lanthipeptides (Figure 3b). In addition, medlan A2 (16), larlans A1 (17), A3 (18), and A4 (19) are analogs of medlan A1 (13) or larlan A2 (14) and deduced to exhibit variable N/C-terminal residues extending outside the thioether rings, possibly due to hydrolysis by unknown aminopeptidases (Figures S23–S29). The three MeLan residues in medlans and larlans increased the complexity of the stereochemical analysis, necessitating the construction of single mutations of Thr to Ser in precursor LarA. To assign the absolute configuration of each MeLan residue in larlan A2, 0.3 mg of each mutant (T2S and T4S) were purified for advanced Marfey's analysis and compared the results with four MeLan standards (Figure S7 and Table S4).^{[40, 41]} Stereochemical analysis confirmed that the thioether rings formed between Thr2 and Cys10, Thr9 and Cys3 within larlan A2 were of LL-MeLan (2R,3R,6R) configuration, whereas the ring between Thr4 and Cys7 was of DL-MeLan (2S,3S,6R) configuration (Figure 3b and Figure S7).

To find out whether other lanthipeptides also employ a similar maturation strategy, the MS/MS data of both wild-type and recombinant strains were meticulously analyzed. The metabolomic profile of H. salinus YJ-37-H, harboring six lanthipeptide BGCs, displayed mass signals with a fragmental pattern resembling larlans and medlans (Figures S30–S36). A representative compound of 20 (observed [M + 2H]²⁺ = 453.6800, calculated [M + 2H]²⁺ = 453.6796, Δ  = 0.9 ppm) was obtained for NMR (1.0 mg) and advanced Marfey's analysis (Figures S31 and S32). As anticipated, archalan β3 (20), derived from BGC alnβ, shares the same characteristics as DADC lanthipeptides. The putative structures of other derivatives, archalans β4–7 (21–24), were also detected and predicted by LC-MS/MS with additional modifications or N/C-terminal stretches of residues outside the rings (Figures S33–S36). The advanced Marfey's analysis revealed that the Lan residues of archalan β3 were of LL (2R,6R) configurations, and all unmodified amino acids were of L configurations, except for the achiral Gly residue (Figure 3b, Figure S7, and Table S4).

Identifying this particular lanthipeptide subfamily in Archaea highlights the existence of unique biosynthetic pathways and enzymes that are distinct from bacterial counterparts. To characterize the minimal essential genes for efficient biosynthesis of larlan A2, the lar BGC was selected for in vivo reconstitution. Co-expression of larA and larM produced trace amounts of larlan A2 (Figure 3c), suggesting the potential contributions from generic peptidases in the host. Furthermore, co-expressing them with the unannotated gene larH significantly improved the yield (Figure 3c and Supplementary Data 3), indicating that LarH may act as a regulator, cofactor, chaperone, or possess other unknown function in larlan A2 biosynthesis. In another case, although classical and DADC lanthipeptides were detected in the wild-type strain, the recombinant strain only yielded classical lanthipeptides of archalan β1–2, despite the incorporation of the putative peptidase from alnβ BGC (Figure 3 and Figure S4). This outcome suggests the involvement of extraneous proteases encoded outside the BGC within the genome of the wild-type strain, pointing toward a complex biosynthetic pathway for DADC lanthipeptides. It likely involves additional hydrolysis steps compared to classical lanthipeptides, possibly facilitated by unidentified proteases that cleave amino acids within the thioether rings (Figure 3d). The exclusive identification of this lanthipeptide subfamily within Archaea showcases the uniqueness of archaeal biosynthetic pathways and enzymatic machinery.

Archaeal Lanthipeptides Exhibit Anti-Archaeal and Motility Regulation Activities

Given the chemical diversity and novelty of archaeal lanthipeptides, we next wanted to investigate their ecological functions. Seven lanthipeptide crude extracts obtained from heterologous expression and seven purified lanthipeptides were collected for antimicrobial activity screening against nine haloarchaeal and six bacterial strains (Figure S37a–c). The same gradient fractions from the host containing an empty vector were tested for comparison to exclude the potential influence of halocin produced by the heterologous host.^[42] Specifically, crude extracts from the heterologous expression of BGCs medb, cib2, and alnα-γ exhibited significant anti-haloarchaeal activity at 100 µg mL⁻¹ (Figure S37a). The lanthipeptides from BGCs alnα and medb, archalan α (25), archalan α2 (7), and medlan B (8), were analogs exhibiting a similar anti-haloarchaeal spectrum (Figure S37a,b). The purified archalan α (25) demonstrated strong inhibition against H. larsenii JCM 13917, H. cibarius DSM 19505, H. amylolyticus JCM 18367, and Halomicroarcula salina JCM 18369 at 100 µg mL⁻¹ (Figure S37b). Notably, 25 exhibited a minimum inhibitory concentration of 6.25 µg mL⁻¹ (5 µM) against H. salina JCM 18369 (Figure S37c). The anti-haloarchaeal activity of archaeal lanthipeptides may be attributed to the characteristics of their habitats, which predominantly host Haloarchaea.^[43]

To study the in vivo function of archaeal lanthipeptides, stab-inoculation screens were conducted to evaluate their impact on host motility. Surprisingly, the motility of three recombinant strains was either positively or negatively affected (Figure 4a). Strains containing BGCs cib2 and pel, responsible for the production of ciblan A2 (4) and pellan (5), exhibited a significantly larger motility zone than the control (Figure 4b, Figure S37 and Supplementary Data 4). To address whether the motility-regulatory function of lanthipeptides depends on the growth regulation of the recombinant strain, we evaluated this relationship through time-course experiments and growth kinetics analysis. The results indicated that cib2 and pel BGCs significantly activated the host's motility without stimulating growth (Figure S37). Conversely, BGC alnα suppressed motility through the antagonistic activity of archalan α (25) and α2 (7), which could impede the host's growth (Figure S37). These findings point to a new ecological role of lanthipeptides, which may reduce the host's reaction time in response to stimuli by regulating motility activity.^[44-46]

A comprehensive investigation combining morphological and transcriptomic analysis was conducted to elucidate the mechanisms by which archaeal lanthipeptides enhance motility. Microscopic analysis of recombinant strains demonstrated that lanthipeptides could elevate the frequency of rod-shaped cell morphology, a characteristic associated with hypermotility (Figure 4c–e and Supplementary Data 5).^[44-46] Additionally, the transcriptomic analysis revealed that the archaellum gene flgA1, a major gene responsible for the motility of H. volcanii,^[47] was upregulated in strains possessing BGCs cib2 or pel (Figure 4f,g and Supplementary Data 6). These findings support the hypothesis that lanthipeptides may promote motility by inducing rod-shaped cell morphology and upregulating the expression of archaellin (Figure 4h).