key: cord-102720-ka95resa
authors: Aguilar, César; Verdel-Aranda, Karina; Ramos-Aboites, Hilda E.; Morales, Marco Antonio; Licona-Cassani, Cuauhtémoc; Barona-Gómez, Francisco
title: Convergent evolution of Streptomyces protease inhibitors involving a tRNA-mediated condensation-minus NRPS
date: 2020-10-27
journal: bioRxiv
DOI: 10.1101/2020.10.26.356543
sha: 
doc_id: 102720
cord_uid: ka95resa

Small peptide aldehydes (SPAs) with protease inhibitory activity are natural products typically synthesized by nonribosomal peptide synthetases (NRPS). SPAs are widely used in biotechnology, as therapeutic agents, they are physiologically relevant and regulate development of the natural hosts. During genome evolutionary analysis of Streptomyces lividans 66 we identified an NRPS-like biosynthetic gene cluster (BGC) that lacked a condensation (C) domain but included a tRNA-Utilizing Enzyme (tRUE) belonging to the leucyl/phenylalanyl (L/F) transferase family. This system was predicted to direct the synthesis of a novel SPA with protease inhibitory activity, called livipeptin. Following genome mining and phylogenomic analyses we confirmed the presence of tRUEs within diverse Streptomyces genomes, including fusions with a C-minus NRPS-like protein. We further demonstrate functional cooperation between these enzymes and provide the biosynthetic rules for the synthesis of livipeptin, expanding the known universe of acetyl-leu/phe-arginal SPAs. The L/F-transferase C-minus NRPS productive interaction was shown to be tRNA-dependent after semisynthetic assays in the presence of RNAse, which contrasts with leupeptin, an acetyl-leu-arginal SPA that we show to be produced by Streptomyces roseous ATCC 31245 via a tRUE-minus BGC with multiple complete NRPSs. Thus, livipeptin and leupeptin are the result of convergent evolution, which has driven the appearance of unprecedented biosynthetic logics directing the synthesis of protease inhibitors thought to be at the core of Streptomyces colony biology. Our results pave the way for understanding this Streptomyces trait, as well as for the discovery of novel natural products following evolutionary genome mining approaches. Abstract importance Convergent evolution in microbiology is believed to be highly recurrent yet examples that have been comprehensively characterized are scarce. Proteases inhibition by small peptide aldehydes is at the core of many microbiological processes, both within the cell and during colony development, and in microbial ecology. Here we report the biosynthetic foundations of leupeptin, the main Streptomyces protease inhibitor, and of livipeptin, a protease inhibitor produced by Streptomyces lividans. Although these peptides belong to the same chemical class, here we show that their biosynthetic routes result from convergent evolution, as they involve unrelated biosynthetic mechanisms, including the recruitment of a tRNA-utilizing enzyme that functionally replaces the condensation domain of a nonribosomal peptide synthetase during livipeptin biosynthesis. Thus, these results pave the way for understanding Streptomyces protease inhibitors as a trait and provide unprecedented knowledge for genome mining of natural products and synthetic biology where proteases inhibition is desirable.

Small peptide aldehydes (SPAs) are metabolites with protease inhibitory activity. Their production has been reported in several bacterial species belonging to the phyla Actinobacteria, Cyanobacteria and Firmicutes, as well as in fungal species belonging to the families Aspergillaceae and Apiosporaceae (Sabotič and Kos 2012) . SPAs molecular weight ranges between 300 and 900 Da, and they are characterized by unique chemical features, including (i) N-terminal groups capped with acyl groups or with ureido-amino acid groups, giving rise to acylated or aminoalkyl ends with a terminal carboxylic group; and (ii) a terminal aldehyde group derived from the carboxyl terminal modification of the peptide chain by a reductive process (Bullock et al. 1996) . Biosynthetically, SPAs have been shown to be produced by nonribosomal peptide synthetases (NRPS) with a reductase (R) domain, which mediates the release of the nascent peptide, and at the same time, the formation of the aldehyde group.

These reactions lead to the chemical warhead by which SPAs covalently interact with serine or cysteine proteases active site residues, giving place to hemiacetals or hemithioacetals, and irreversibly inhibiting proteases enzymatic activity (Brayer et al. 1979; Wlodawer et al. 2001) .

Based on their functional groups, SPAs can be divided into two sub-classes: (i) those with a terminal group protected by an acyl group, including flavopeptin, tyrostatin, tyropeptin, nerfilin, strepin, leupeptin, thiolstatin, acetyl-leu-arginal and bacithrocins A-D (Figure 1); and (ii) those in which the Nterminal group binds to an ureido moiety that is attached to an amino acid by means of an amide linkage, such as chymostatin, MAPI, GE20372, antipain and elastatinal (Umezawa et al. 1970; Suda et al. 1972; Tatsuta et al. 1973; Watanabe et al. 1979; Stefanelli et al. 1995; Murao and Watanabe 2014) . The chemical configuration of SPAs allows alteration of the peptide chain, wherein the ureido group acts as an adapter that changes the order of the peptide, from N-terminal to C-terminal into C-terminal to Cterminal, making them distinct from traditional ribosomal peptides. Amongst SPAs, the most widely used is the acetyl-leu/phe-arginal metabolite leupeptin, which was discovered by Umezawa and co-workers 4 more than fifty years ago Kondo et al. 1969 ). Yet, not much is known about the genetic basis and concomitant mechanisms driving the chemical structural diversity and biomolecular specific activities of leupeptin, and overall, of SPAs. The predicted biosynthetic system includes (i) an NRPS-like protein, lacking both canonical condensation (C) and thioesterase (TE) domains, but with one adenylation (A) domain, a peptidyl carrier protein (PCP) and a reductase (R) domain (Sli0883 gene); (ii) a homolog of a leucyl/phenylalanyl (L/F) tRNA transferase (Sli0884 gene), a tRNA-utilizing enzyme hitherto never implicated in natural products biosynthesis; and (iii) an N-acyltransferase (Sli0885 gene). These enzymes are located within the SLP3 linear plasmid, a mobile genetic element of S. lividans absent from the industrial strain TK24 (Cruz- Morales et al. 2013 ). Our previous proposal implicated these genes in the synthesis of an N-acylated leu -arginal dipeptide, which we called livipeptin. Since this prediction suggests common structural features with previously described SPAs, such as leupeptin Kondo et al. 1969) , thiolstatin (Murao et al. 1985; Kamiyama et al. 1994) , bacithrocins (Kamiyama et al. 1994 ) and acetyl-leu-arginal Kondo et al. 1969) , it was hypothesized that livipeptin could have protease inhibitory activity. 5 Here, we characterize S. lividans 66 putative livipeptin's BGC following a combined phylogenomics, genome mining and synthetic biology approach. As predicted, our results show that this unprecedented biosynthetic system directs the synthesis of livipeptin, which shows strong protease inhibitory activity. Livipetin was found to be chemically related to bacithrocins and thiolstatin, reported previously to be produced by Brevibacillus laterosporus (Kamiyama et al. 1994) ; and to leupeptin, whose BGC in Streptomyces is also identified herein. Interestingly, leupeptin and livipeptin were shown to be synthesized by unrelated pathways, demonstrating convergent evolution within Streptomyces and with other taxonomically unrelated leupeptin-producing proteobacteria (Li et al. 2020) . Genetic and semisynthetic chemical experiments in the presence of RNAse, as done previously for other tRUEs (Ortega and van der Donk 2016) , confirmed the involvement of an L/F transferase, typically involved in protein tagging during proteolysis (N-end rule) within an essential process carried out by many organisms that regulate the half-life of proteins and morphological differentiation (Leibowitz and Soffer 1969; Varshavsky 2011) , during livipeptin biosynthesis.

We anticipate that the discovery of this tRUE, which functionally interacts with a C-minus NRPS, will pave the way for exploiting this unprecedented biosynthetic logic to uncover novel natural products, and will assist in the development of auxiliary synthetic biology tools to inhibit proteolysis (Cruz-Morales et al. 2015a & 2015b . Similar implications are envisaged for leupeptin once its BGC is fully characterized. Moreover, our results are the first step towards investigating the evolution and role of protease inhibitors during Streptomyces development, colony biology and ecology, an old hypothesis that remains to be experimentally tested (Chater et al. 2010 ).

The presence of an L/F transferase homolog in the livipeptin BGC suggested that this enzyme family could have been recruited for the synthesis of a natural product (NP). This possibility encouraged us to mine for L/F transferases in the context of NP biosynthesis, with an emphasis in Actinobacteria, as close homologs of Sli0884 linked to Sli0883 could not be found beyond this phylum. Thus, combined EvoMining (Sélem-Mojica et al. 2019) and CORASON (Navarro-Muñoz et al. 2020 ) phylogenomic analyses of actinobacterial L/F transferases allowed us to identify 137 L/F transferase homologs within 1,246 good quality and well annotated actinobacterial genomes. Remarkably, despite the fact that L/F transferases are known to play a central role as housekeeping enzymes involved in proteolytic metabolism (Mogk et al. 2007) , only 11% of the actinobacterial genomes investigated include a homolog of this gene, suggesting that these organisms may have alternative proteolytic tagging strategies, such as the SsrA (tmRNA) tagging system (Braud et al. 2006 ).

Both the EvoMining and CORASON output revealed two discrete clades, consistent with one function possibly devoted to housekeeping metabolism, and a second function predicted to be involved in NP biosynthesis. The large branch includes 104 enzymes from diverse Actinobacteria species, but without a conserved gene neighborhood (Supplementary Figure S1) , supporting a single-enzyme housekeeping biochemical function. The second clade is less populated yet shows a larger degree of gene order conservation amongst its 33 entries. These enzymes are mostly from the genus Streptomyces, but also from Nocardiopsis (1), Kitasatospora (3), Streptacidiphilus (6) and Actinopolyspora (1), with a Frankia non-conserved entry at its root (Figures 2A and 2B) . This so-called small Streptomyces conserved clade consists of L/F transferases probably recruited into NP biosynthesis, as revealed by EvoMining in the first instance, and independently confirmed with the use of antiSMASH (Blin et al. 2019 ). These analyses also revealed potentially translational fusions between the L/F transferase tRUE 7 and the C-minus NRPS-like genes (shown with an asterisk, Figure 2B ). As in S. lividans, the latter were always found to encode for an adenylation domain predicted to have specificity towards an arginine residue, a peptidyl carrier protein (PCP) and a reductase (R) domain. The domain organization confirmed above resembles that of a canonical NRPS involved in the synthesis of other protease inhibitors (Winn et al. 2016) , but with the L/F transferase replacing (and thus possibly playing a similar role to) the expected condensation domain needed for formation of an amide bound. The R domain could be functionally equivalent to that in flavoprotein, an Streptomyces SPA of a different class, whose biosynthesis involves an NRPS with a R domain responsible for reductive release of the nascent peptide (Chen et al. 2013) . Within the small Streptomyces sub-clade, the livipeptin BGC branch includes hits from four other Streptomyces species, and one from Actinopolyspora erythreae JPMV. Yet, despite these species are from different genera, their BGCs share high sequence similarity and conserved gene order. This includes two other biosynthetically potential elements: an Omethyltransferase and a butyryl-CoA dehydrogenase (Sli0886 and Sli0887 in S. lividans). All together, these observations support our previous prediction related to the synthesis of the putative protease 8 inhibitor livipeptin in S. lividans (Cruz-Morales et al. 2013) , but also includes leucine in addition to phenylalanine as an alternative biosynthetic precursor ( Figure 3A ).

Owing to the predicted chemical structure of livipeptin we hypothesized it may chemically classify as an acetyl-leu-arginal, which includes other related molecules , the bacithrocins and thiolstatin, previously isolated from Brevibacillus laterosporus (Kamiyama et al. 1994) , and leupeptin Kondo et al. 1969) (Figure 1) . Therefore, using the L/F transferase as a beacon, we mined the sixty-one Brevibacillus spp. genome sequences available at the time of our analyses, without positive results. Likewise, after genome sequencing and mining of the leupeptin-producing strain S. roseus ATCC 31245 (Pridham et al. 1965) we were unable to find evidence of a homologous livipeptin BGC. Instead, we identified a BGC rich in NRPSs with A domains specific to diverse aminoacids, such as L-thronine and L-serine (Leup1), L-isoleucine (Leup7) and L-ornithine (Leup9). Additionally, homologs of argininosuccinate lyase, cysteine synthase, and threonine kinase, previously implicated in the biosynthesis of diaminobutyric acid (DABA) were also found ( Figure 3B and Supplementary Figure S2 ). Although we hypothesized this BGC to be implicated in the synthesis of leupeptin it is unclear how exactly its biosynthesis proceeds. 

These observations suggest that despite their chemical similarities, leupeptin and livipeptin BGCs are evolutionarily unrelated, as they follow different biosynthetic logics for their synthesis. While for leupeptin further analysis is need for postulating a complete NRPS biosynthetic pathway, including its biosynthetic precursors (Figure 3B ), for livipeptin we propose a biosynthetic mechanism involving a small assembly line that catalyzes amide bond formation between arginine (A domain encoded by the NRPS-like gene) and phenylalanine or leucine residues provided by their cognate aminoacyl-tRNAs as substrates (tRUE of the L/F transferase family). A reductive release mechanism leading to the aldehyde moiety of livipeptin would also be part of this pathway. The resulting dipeptide aldehyde may undergo acetylation catalyzed by the function of an N-acetyltransferase (Sli0885). These three enzymes may represent the minimal biosynthetic core necessary to synthesize a metabolite with protease inhibitory activity. However, additional methylation (Sli0886) and further dehydrogenation (Sli0887) could also be in place ( Figure 3A ). In the absence of transcriptional regulatory genes from both BGC, how these enzymes potentially interact to provide a regulatory mechanism controlling protease inhibitory activity, as previously suggested in Streptomyces species (Kim and Lee 1995; Kim and Lee 1996) , remains unknown.

To identify the livipeptin metabolite(s) potentially produced by Sli0883-5, proposed to have protease inhibitory activity, a S. lividans 66 mutant was constructed (S. lividans ΔLvp, Sli0883-5 minus; Table 1 ). S. lividans strains 66 (wt) and ΔLvp (mutant) grown in modified R5 medium were used for the identification of livipeptin after comparative HPLC analysis of the aqueous extracts of the resulting cultures. These conditions were suboptimal for comparative chromatographic analysis, as they resulted in far too many metabolites, but only one differential peak eluting at a retention time of 4.5 min with a 10 distinctive absorption at 280 nm. The compound associated with this peak could be isolated after extensive HPLC optimization and fractionation ( Figure 4A) , and high-resolution MS analysis of the isolated compound(s) showed that this peak corresponds to a metabolite with an m/z of 365.18 (Supplementary Figure S3) . Further experiments with an MS ion trap targeted toward this mass reproducibly lead to a distinctive fractionation that includes a derived ion with an m/z of 347 ( Figure   4B ), equivalent to loss of a water molecule, in addition to ions with m/z values of 203 and 185 ( Figure   4C , see also Methods).

The abovementioned compound is close to the theoretical molecular formula of thiolstatin (C17N5O3H21), but only if incorporation of L-phenylalanine over L-leucine, together with L-arginine, is considered. Indeed, MS 2 data obtained coincides with previously reported data for other acetyl-leuarginal metabolites Kondo et al. 1969) , the bacithrocins and thiolstatin produced by Brevibacillus laterosporus (Kamiyama et al. 1994) . Moreover, although more distantly, these data confirm the expected chemical connection via leucine residues between livipeptin with leupeptin (185 and 203 fragments, see Table 1 ). Therefore, we concluded that livipeptin is equivalent to thiolstatin and bacithrocin D, despite our inability to identify a homologous locus in the currently available Brevibacillus genome sequences. This could be due to the lack of taxonomic resolution and detailed information about the Brevibacillus laterosporus strain (Kamiyama et al. 1994) , leading to misclassification of the original thiolstatin-and bacithrocin-producing strain(s). After genome mining analyses (data not shown) we favor this explanation rather than the existence of an analogous livipeptin BGC in species belonging to Brevibacillus, although a different biosynthetic route yet to be identified cannot be ruled out.

In order to identify the leupeptin BGC, we first confirmed by HPLC-MS the ability of S. roseus ATCC 31245 to produce this metabolite. The metabolite found was consistent with the leupeptin standard from a commercial source. We then sequenced and mined the genome of this organism, which consists of 7.8 Mb encoding for 6,765 ORFs and 27 putative BGCs. As mentioned previously, our efforts failed to identify a livipeptin BGC in this streptomycete. Yet, we identified alternative BGCs that could account for the synthesis of leupeptin (Cruz-Morales et al. 2015; Cruz-Morales, et al. 2016) . A combined EvoMining, CORASON and antiSMASH analysis gave rise to a hypothesis that a BGC with five NRPSs and a diaminobutyric acid (DABA) biosynthetic cassette (Supplementary Figure S2 ) could direct the synthesis of leupeptin. Full functional characterization of this BGC is beyond the scope of this report, but to confirm the involvement of this locus in leupeptin biosynthesis we constructed an insertional mutant by targeting the leupA gene with a suicidal plasmid, leading to a strain termed S. roseus LeuP1 (see Table 1 and Methods). This strain was unable to synthesize leupeptin when compared with the wild type parental ATCC 31245 strain. Moreover, heterologous expression of pESAC13-A constructs (Tocchetti et al. 2018) bearing the putative leupeptin BGC in E. coli lead to production of leupeptin. The MS 2 fragmentation pattern of isolated leupeptin coincides with a commercial standard and with some fragments of livipeptin spectral data (185 and 203, see Table 1 and Supplementary Figure S4 ).

Our results therefore establish a chemical relationship between livipeptin and leupeptin as acetylleu/phe-arginal metabolites, despite involving distinct biosynthetic pathways, and potentially different precursors. While leupeptin is proposed to be synthesized by nine genes (including five NRPSs and a subcluster potentially involved in the synthesis of the non-proteinogenic amino acid DABA), livipeptin is produced by a C-minus NRPS-like protein functionally linked with a tRUE L/F transferase, plus an acyltransferase. On one hand, the identification of distinct biosynthetic pathways to produce very similar NPs highlights the increasing versatility observed among NRPS-derived natural products in terms of their biosynthetic logics and enzyme interactions within multifunctional protein templates. On the other hand, the occurrence of convergent evolution in natural products biosynthesis is a recurring theme (Fischbach 2009 ), although its significance and rate remains to be determined (Chevrette et al. 2020) . It is nevertheless tempting to speculate that the degree of convergent evolution correlates with chemical scaffolds that contribute towards the fitness of the host organisms, as previously suggested for protease inhibitors (Chater et al. 2010; Guo et al. 2017; Li et al. 2020 )

To further characterize livipeptin, and to be able to determine its presumed protease inhibitory activity, we used as positive controls commercial leupeptin and antipain against trypsin and papain, respectively. For leupeptin, three different sources with identical results were used: (i) commercial standard produced after bacterial fermentations, (ii) S. roseus ATCC 31245 and (iii) E. coli cultures expressing pESAC13-A constructs (Tocchetti et al. 2018) , namely pESAC_810, harboring the presumed leupeptin BGC (Table 1) . For livipeptin, we opted for the use of synthetic Sli0883-5 genes optimized for E. coli and their expression from the pAsk vector (Adams et al. 2014 ). The resulting expression plasmid, pLvp, was used to transform E. coli C41. The resulting strain, C41 Lvp, was cultivated in minimal medium. The conditions used for these experiments allowed us to easily perform a chromatographic comparative analysis of aqueous extracts, without the residual metabolites found in the S. lividans native system.

As above, comparison of the HPLC profiles of E. coli expressing pLvp and E. coli bearing the empty plasmid pAsk showed a differential peak at 4.5 min absorbing at 280 nm ( Figure 4A ). This metabolite shows up in the E. coli cultures with less background than that seen in the S. lividans samples ( Figures 4B and 4C ). Yet, this E. coli data also failed to identify the presumed L-leucine livipeptin analog with a mass of 313.39 (m/z), but did rule out the WT S. lividans-specific mass 487 (m/z) as related to livipeptin, given that it could not be detected within the E. coli pLvp spectra. Whether this mass corresponds to a bigger version of livipeptin was further ruled out after inspection of the MS 2 spectra of the 487 (m/z) metabolite (Supplementary Figure S3) . Thus, although these data do not confirm that all 13 three Sli0883-5 genes are needed for synthesis of livipeptin, it unequivocally establishes a link between the lvp genes Sli0883-5 and a putative SPA with protease inhibitory activity. Previous data suggests that thiolstatin has weak inhibitory activity against serine proteases, such as trypsin, but potent proteolytic inhibitory activity towards cysteine proteases, such as papain. These observations actually explain the use of the prefix 'thiol' in its name (Murao et al. 1985) . Unfortunately, the absence of thiolstatin or bacithrocins standards hampered our ability to compare the activity of these metabolites with that of livipeptin. Thus, we focused on HPLC-purified products from S. lividans 66 and E. coli C41 pLvp cultures. The results obtained from these experiments are included in Table 1 and shown in Figure 5A .

Although strong inhibitory activity equivalent to leupeptin and antipain could be found for the HPLC-purified metabolite(s) from the two experimental sources investigated, the material obtained from the E. coli heterologous system showed less activity, especially towards papain. This may contradict 14 early data on thiolstatin activity (Murao et al. 1985; Kamiyama et al. 1994 ), but given that the overall conditions of these experiments are not equivalent, this is an unfair comparison and we cannot explain this difference. Interestingly, as can be seen in Figure 4 , the extracts from S. lividans 66 have a larger metabolite diversity than that from E. coli. Similar to leupeptin (Suzukake et al. 1980) , under the rich Streptomyces growth conditions, livipetin could be produced and co-purified as a molecular cluster, leading to synergistic protease inhibitory activity. A similar situation has been previously shown for other biomolecular activities of natural products (Gutiérrez-García et al. 2017) . Indeed, commercial high quality leupeptin isolated from S. roseus cultures includes several related species, which contrasts with the less active synthetic leupeptin (McConnell et al. 1990 ). Indeed, the HPLC-purified metabolite(s) obtained from the S. lividans mutant ΔLvp shows some residual inhibitory activity against both proteases ( Figure 5A) .

In addition to codon usage optimization for E. coli expression, the pLvp plasmid was designed such that each of the three lvp genes (Sli0883-85) could be excised and the plasmid circularized with different-yet-compatible cohesive restriction enzyme, easing construction of expression plasmids with all possible combinations (Supplementary Figure S5) . In addition to pLvp (pSli3-4-5), as shown in Table 2 , five different constructs were obtained: three single-gene plasmids (pSli3, pSli4 and pSli5) and two double-gene plasmids (pSli3-4 and pSli4-5). The resulting plasmids were confirmed after DNA sequencing and used in assays consisting of mixtures of different cell-free extracts expressing the plasmids, such that all cis and trans gene-expression possible combinations could be addressed. This synthetic biology strategy allowed us to evaluate the potential functional impact that co-expression of two different enzymes could have, when compared with independent enzyme expression, providing a sense of the predicted enzyme-enzyme interactions between the C-minus NRPS-like protein, the L/F-transferase and the N-acetyl transferase, and its relationship with protease inhibitory activity (Figure 5B and 5C).

These experiments showed that only the three-genes combinations, obtained after independent expression (in trans) from independent plasmids or co-expression (in cis) from the same plasmid, yielded livipeptin synthesis ( Figure 5C) . Unexpectedly, in trans expression of Sli0883, Sli0884 and Sli0885 provided the highest yield of livipeptin. The combination Sli0883,5 and Sli0884 was not tested as the former plasmid was not obtained, whereas all combinations with only two or one gene did not lead to production of livipeptin ( Table 2 ). More importantly, as expected, reconstituted livipeptin showed protease inhibitory activity, similar to that found in the S. lividans native system ( Figure 5A ) and the heterologous E. coli system ( Figure 5B ) against trypsin, and to a lesser extent, papain. Although the reason of these differences remains unknown, these results establish the Sli0883-5 genes as the minimal biosynthetic core for the synthesis of livipeptin with protease inhibitory activity.

This experimental layout also provided us with the opportunity to test the hypothesized aminoacyl-tRNA dependency of the L/F transferase tRUE. In vitro synthesis of livipeptin with the gene combinations containing the entire cluster was redone. However, for these experiments, the cell-free extracts were pre-incubated with RNase prior to generating the reaction mixture. As shown in Figure 5C and Table 2 , HPLC analysis shows that the addition of RNase suppresses the formation of livipeptin, which can be explained by the degradation of the aminoacyl-tRNA substrate. This approach has been previously adopted during characterization of the tRNA-dependent lantibiotic dehydratase NisB, involved in nistatin biosynthesis (Ortega et al. 2014 ). As we could not detect protease inhibitory activity in these cell-free extracts, Sli0883-5 genes are strictly necessary for livipeptin synthesis. These results add to the increasing number of reports that implicate central metabolism tRUEs in natural products biosynthesis (Hong et al. 2004; Garg et al. 2008; Zhang et al. 2011; Belin et al. 2012; Bougioukou et al. 2013; Ortega et al. 2014 ). (Ichetovkin et al. 1997; Watanabe et al. 2007; Fung et al. 2011; Zhang et al. 2011; Ortega et al. 2014) , the catalytic mechanism and residues involved in substrate-assisted catalysis (Watanabe et al. 2007; Fung et al. 2011; Zhang et al. 2011 ) and in the innovative role of tRNA-dependent enzymes in pacidamycin and lantibiotic biosynthetic pathways (Zhang et al. 2011; Ortega et al. 2014 ). However, the role of an L/F tRNA transferase in peptide bond formation during biosynthesis of a SPA, in association with an NRPS-like protein, has not been shown until now. The combinatorial potential provided by the interaction between A domains and tRUEs warrants further investigation in many ways. For instance, we envisage exploitation of this discovery for untapping novel natural products diversity through genome mining efforts, as well as for the development of synthetic biology approaches targeting proteolysis in different settings (Hines et al. 2008) , such as within the present Covid19 pandemic crisis. al. 2020) algorithms were used as previously. S. roseus ATCC31245 was obtained from the ATCC collection, and its genomic DNA was extracted using common protocols (Kieser et al. 2000 ) and sequenced at the genomic sequencing facilities of Langebio, Cinvestav-IPN (Irapuato, Mexico), using an Illumina MiSeq platform in paired-end format with read lengths of 250 bases and insert length of 800 bases. In total, 721 Mbp of sequence was obtained. The raw reads were filtered using Trimmomatic (Bolger et al. 2014 ) and assembled with velvet (Zerbino and Birney 2008), obtaining a 7.8 Mb assembly in 165 contigs with a coverage of 95 X and a GC content of 72 %. This assembly was annotated using RAST (Aziz et al. 2008) , antiSMASH (Weber et al. 2015) and EvoMining (Sélem-Mojica et al. 2019).

The genome includes 7,834,742 bp, and it was assembled in a total of 234 contigs and deposited in NCBI under the genome accession number NZ_LFML01000000.

For the identification and characterization of livipeptin in S. lividans 66, a mutant deficient for the Sli0883-5 genes was constructed. These genes were replaced by the apramycin resistance cassette aac(3)IV marker in-frame within a pESAC13-A construct. A 1.5 Kb region flanking the Sli0883-5 genes replaced by the apramycin cassette was then amplified by PCR and cloned into the plasmid pWHM3, which contains a thiostrepton resistance gene. pWHM3 is an unstable Streptomyces vector that is lost after some rounds of cultivation of the transformed strain without selection (Vara et al. 1989 ). Double crossovers after integration of the pWHM3 Sli0883-5:: aac(3)IV construct were screened after apramycin resistance (50 μg/mL) and thiostrepton (25 μ/mL) sensitivity. The genotype of several transformants was confirmed by PCR, leading to S. lividans 66 ΔLvp used for experimentation. The synthetic livipeptin BGC, employing the E. coli usage codon, was obtained from GeneScript (New Jersey, USA). The design of the genetic construction includes restriction sites flanking each gene, as follows: NdeI-Sli0883-EcoRI-Sli0884-XbaI-Sli0885-BglII-HindIII. Different combinations of these genes were cloned into pAsk (Adams et al. 2014) , resulting in six different plasmids (Table 1 and Supplementary Figure S5) .

The Streptomyces roseus leupA mutant was constructed following an insertional mutagenesis strategy using as parental strain ATCC 31245. For this, a leupA 640 bp fragment was PCR amplified and cloned into pCR2.1-TOPO using a TA cloning kit (Invitrogen, Carlsbad, USA). The resulting suicide 19 plasmid, termed pLeupA, was introduced into S. roseus via protoplasts fusion (Kieser et al. 2000) . The transformants were selected with kanamycin (50 μ/mL) and the genotype of the insertional mutant S. For production of leupeptin, S. roseus ATCC 31245 was grown on shake flask cultures containing a media designed by us, as follows: glucose 3 g, NH4NO3 0.5 g, MgSO4 (7H2O) 0.5g, KCl 0.05 g, Lleucine 0.75 g, L-arginine 0.75 g, glycine 0.75 g, casamino acids 0.1 g, yeast extract 0.4 g per liter.

Cultures were incubated at 30°C for 48 h prior to supernatant analysis. Heterologous leupeptin production was performed in shake flask cultures using E. coli DH10B bearing pESAC_810, in M9 20 minimal media (Glucose 4 g/L plus casamino acids 2 g/L) supplemented with apramycin (50 μg/ mL) for selection of the cosmid at 30°C for 48 h.

The cell extract of E. coli C41 pGroEL/GroES was prepared following a previously described method (Kigawa et al. 2004 ) with modifications. E. coli C41 pGroEL/GroES was inoculated in 50 mL de LB medium and grown at 37°C to an OD600 of 0.5 for induction with 20 ng/µl of anhydrotetracycline.

After induction, the culture was grown at 20 °C for 12 h. Cells were harvested for (6000 rpm, 10 min, 4°C) and the pellet was washed three times by resuspension in S30 buffer (10 mM Tris-acetate buffer pH 8.2, 14 mM magnesium acetate, 60 mM potassium acetate, 1 mM dithiothreitol (DTT), 0.3 mM EDTA, 0.3 mM MgCl2) (Ortega et al. 2014 ) followed by centrifugation (6000 rpm, 10 min, 4°C). Cells were then resuspended in 1 mL of S30 buffer per gram of wet cells and lysed with a Sonicator. The cell lysate was centrifuged twice (13 000 rpm, 45 min, 4°C) and the supernatant was dialyzed four times against 50 volumes of S30 buffer (without DTT) using Amycon Ultra-15 (MerckMillipore) tubes for dialysis with a molecular mass cutoff of 10 kDa. The cell extract was then centrifuged (4 000 rpm, 10 min, 4°C) and the supernatant was frozen and stored in 1 mL samples at -80°C for future use.

Supernatant of cultures and in vitro reaction mixtures were evaporated to dryness. The dry residues were dissolved in a 0.1 volume of HPLC grade H2O and injected into a C18 Discovery 504955 Supelco column with a particle size of 5 μm connected to a HPLC-Agilent 1200 equipped with a diode array detector and a fraction collector. The mobile phase comprised a binary system of eluent A, H2O, and eluent B, 100% MeOH. The run consisted of H2O/MeOH gradient (0-5 min: 0% B; 5-35 min: 10% B; 35-45 min: 100%. Differential peaks were detected between the wild-type and mutant strains (or empty vectors) by monitoring absorbance at a wavelength of 280 nm, and the selected fractions were collected for bioactivity assays and MS analysis in an ion trap LTQ Velos mass spectrometer (Thermo 21 Scientific, Whatam, USA). MS/MS analysis of selected ions was performed with a collision energy of 20 eV.

Fractions collected in HPLC analysis were analyzed in vitro using a fluorometric assay with excitation at 340 nm and emission at 420 nm. The chromogenic compound Nα-Benzoyl-DL-arginine 4nitroanilide hydrochloride (BApNA, Sigma-Aldrich) was used as substrate for trypsin and papain ATP (5 mM) and NADPH (5 mM). The assay was incubated at 30°C for 5 h, centrifugated to remove insoluble material (13,000 rpm, 5 min, 25°C). In addition, the cell free extract (10 µL) was treated with RNase in the presence of CaCl2 (100 µM). The activity was calculated in percentage using as 100% the relative fluorescence units (RFU) of the proteases without inhibitors at the end of the reaction (20 min).

The slope of the curve (initial rate) of the time progress of the reaction in each experiment was also calculated by triplicate.

. a HPLC retention time (RT) and mass spectrometry (MS, MS 2 ) data with common fragments between livipeptin and leupeptin shown in bold. NI, not identified (see Figure 4 and its associated Supplementary Figures S3 and  S4 for spectral analysis). b T.I.A. and P.I.A., trypsin and papain inhibitory activity, respectively, which was recorded as activity curves for 20 minutes, and reported as a percentage of the activity found for leupeptin and papain, respectively (see Figure 5A ). Data was generated after three independent experiments, leading to standard deviations measured as relative fluorescence units (RFU). d Reported activity is from commercial leupeptin of microbial source with ≥ 90% purity (Sigma, No. L2884). e pESAC_810 construct (or pLeup) is a pESAC13-A derivative (Tocchetti et al. 2018) containing the predicted leupeptin BGC, including the leup1-9 genes. a The reaction mixtures were prepared with E. coli cell-free extracts (c. f. e.) with or without addition of RNase. b Genotype provided is for plasmids harboring the S. lividans 66 genes Sli0883-5. NI, livipeptin could be not identified. As in Figure 5 , Sli0883-5 genes are shown only with their final digit, i.e. 3, 4 and 5. A comma (,) and a hyphen (-) are used to denote expression in trans or in cis, respectively. Trypsin (T. I. A.) and papain (P. I. A.), HPLC and MS analysis, was determined and expressed as in Table 1 . respectively. Standard deviations shown were calculated from three independent experiments (Table 1) B. Activity reconstitution using a synthetic biology approach of livipeptin BGC. Metabolites, proteases and controls are as in panel A (data not shown). Only the three three-genes rection mixtures, irrespective of their genetic organization (expression in trans or in cis), showed protease inhibitory activity. Cell-free extracts with one-or two-genes constructs did not show inhibitory activity ( Table 2) . RNase (dotted bars) eliminates trypsin and papain inhibitory activity in around 87% and 37%, respectively. C. S li5 Sli3 , Sli 4, S li5 Sli3 , Sli 4, S li5 % RFU at 20 min Sli4, Sli5 Sli3, Sli5 

Promiscuous and adaptable enzymes fill "holes" in the tetrahydrofolate pathway in Chlamydia species

Leupeptins, new protease inhibitors from Actinomycetes

The RAST server: Rapid Annotations using Subsystems Technology

The nonribosomal synthesis of diketopiperazines in tRNA-dependent cyclodipeptide synthase pathways

0: updates to the secondary metabolite genome mining pipeline

Trimmomatic: a flexible trimmer for Illumina sequence data

Revisiting the biosynthesis of dehydrophos reveals a tRNA-dependent pathway

Effect of SsrA (tmRNA) tagging system on translational regulation in Streptomyces

Crystallographic and kinetic investigations of the covalent complex formed by a specific tetrapeptide aldehyde and the serine protease from Streptomyces griseus

Peptide Aldehyde Complexes with Wheat Serine Carboxypeptidase II: Implications for the Catalytic Mechanism and Substrate Specificity

The complex extracellular biology of Streptomyces

Proteomics Guided Discovery of Flavopeptins: Anti-proliferative Aldehydes Synthesized by a Reductase Domain-Containing Nonribosomal Peptide Synthetase

Evolutionary dynamics of natural product biosynthesis in bacteria

The genome sequence of

Streptomyces lividans 66 reveals a novel tRNA-dependent peptide biosynthetic system within a metalrelated genomic island

Genetic system for producing a proteases inhibitor of a small peptide aldehyde type. USA patent No. US20180265545A1

Recapitulation of the evolution of biosynthetic gene clusters reveals hidden chemical diversity on bacterial genomes

Phylogenomic analysis of natural products biosynthetic gene clusters allows discovery of arseno-organic metabolites in model Streptomycetes

Antibiotics from microbes: converging to kill

An alternative mechanism for the catalysis of peptide bond formation by L/F transferase: substrate binding and orientation

Investigations of valanimycin biosynthesis: elucidation of the role of seryl-tRNA

Discovery of reactive microbiota-derived metabolites that inhibit host proteases

Phylogenomics of 2,4-Diacetylphloroglucinolproducing Pseudomonas and novel antiglycation endophytes from Piper auritum

Proteasome inhibition by fellutamide B Induces nerve growth factor synthesis

Characterization of an inducible vancomycin resistance system in Streptomyces coelicolor reveals a novel gene (vanK) required for drug resistance

Substrate recognition by the

Leucyl/Phenylalanyl-tRNA-protein transferase

Bacithrocins A, B and C, novel thrombin inhibitors

MIBiG 2.0: a repository for biosynthetic gene clusters of known function

Practical Streptomyces Genetics

Preparation of Escherichia coli cell extract for highly productive cell-free protein expression

Regulation of production of leupeptin, leupeptin-inactivating enzyme, and trypsin-like protease in Streptomyces exfoliatus SMF13

Trypsin-like protease of Streptomyces exfoliatus SMF13, a potential agent in mycelial differentiation

Isolation and characterization of leupeptins produced by Actinomycetes

A soluble enzyme from Escherichia coli which catalyzes the transfer of leucine and phenylalanine from tRNA to acceptor proteins

Making and breaking leupeptin protease inhibitors in pathogenic Gammaproteobacteria

New leupeptin analogs: synthesis and inhibition data

The N-end rule pathway for regulated proteolysis: prokaryotic and eukaryotic strategies

Novel thiol proteinase inhibitor, thiolstatin, produced by a strain of Bacillus cereus

Novel microbial alkaline protease inhibitor, MAPI, produced by Streptomyces sp

A computational framework to explore large-scale biosynthetic diversity

Structure and mechanism of the tRNA-dependent lantibiotic dehydratase NisB

New Insights into the biosynthetic logic of ribosomally synthesized and post-translationally modified peptide natural products

Comparison of some dried holotype and neotype specimens of Streptomycetes with their living counterparts

Microbial and fungal protease inhibitors-current and potential applications

EvoMining reveals the origin and fate of natural product biosynthetic enzymes. Microb

GE20372 factor A and B: New HIV-1 protease inhibitors, produced by Streptomyces sp. ATCC 55925

Antipain, a new protease inhibitor isolated from Actinomycetes

Biosynthesis of leupeptin. III. Isolation and properties of an enzyme synthesizing acetyl-L-leucine

The structure of chymostatin, a chymotrypsin inhibitor

Large inserts for big data: artificial chromosomes in the genomic era

Chymostatin, a new chymotrypsin inhibitor produced by actinomycetes

Cloning of genes governing the deoxysugar portion of the erythromycin biosynthesis pathway in Saccharopolyspora erythraea (Streptomyces erythreus)

The N-end rule pathway and regulation by proteolysis

Protein-based peptidebond formation by aminoacyl-tRNA protein transferase

Structural elucidation of alpha-MAPI, a novel microbial alkaline proteinase inhibitor, produced by Streptomyces nigrescens WT-27

AntiSMASH 3.0-A comprehensive resource for the genome mining of biosynthetic gene clusters

Recent advances in engineering nonribosomal peptide assembly lines

Inhibitor Complexes of the Pseudomonas Serine-Carboxyl Proteinase

Velvet: Algorithms for de novo short read assembly using de Bruijn graphs

tRNA-dependent peptide bond formation by the transferase PacB in biosynthesis of the pacidamycin group of pentapeptidyl nucleoside antibiotics

We thank the HPLC and MS Units of Cinvestav-IPN, Irapuato, for technical support. We also thank Pablo Cruz-Morales for strains construction and Nelly Selem-Mojica for bioinformatics support.