Bioengineering of lanthipeptides in Escherichia coli: assessing the specificity of lichenicidin and haloduracin biosynthetic machinery

Research in Microbiology 165 (2014) 600e604 www.elsevier.com/locate/resmic

Brief note

Bioengineering of lanthipeptides in Escherichia coli: assessing the specificity of lichenicidin and haloduracin biosynthetic machinery T. Caetano a,*, J. Barbosa a, E. M€ oesker b, R.D. Su¨ssmuth b, S. Mendo a a

Department of Biology and CESAM, University of Aveiro, 3810 Aveiro, Portugal b Institut fu¨r Chemie, Technische Universit€at Berlin, 10623 Berlin, Germany Received 3 May 2014; accepted 10 July 2014 Available online 21 July 2014

Abstract The lichenicidin and haloduracin biosynthetic machinery specificity was investigated in vivo in Escherichia coli. Unlike previous reports using different hosts, it was found that the biosynthetic machineries of lichenicidin and haloduracin are highly specific to their dedicated peptide precursors. Likewise, the substitution of lichenicidin structural genes by chimeras of lichenicidin leader sequences and haloduracin core peptides did not yield mature haloduracin peptides. Despite these restrictions, it was found that the bifunctional enzyme HalT was able to process and export lichenicidin peptides. These findings corroborate the promiscuity of LanT enzymes reported for other lantibiotics, such as nukacin ISK-1 and lacticin 481. © 2014 Institut Pasteur. Published by Elsevier Masson SAS. All rights reserved.

Keywords: Bacillus; Bacteriocins; Chimeras; In vivo; Lantibiotics

1. Introduction Considering the increasing ineffectiveness of classical antimicrobials against some clinically relevant Gram positive bacteria, lanthipeptides with potent antimicrobial activity are currently under investigation as viable alternatives to antibiotics presently commercialized [1e5]. These compounds, also known as lantibiotics, are gene encoded, representing an advantage over non-ribosomal or polyketide antibiotics to implement bioengineering strategies [6]. Lanthipeptides result from extensive post-translational modifications of their precursor peptides to reach a mature and active form. Therefore, it is essential to understand the flexibility and/or specificity of the biosynthetic machinery responsible for this processing [7e10].

* Corresponding author. Department of Biology, University of Aveiro, 3810 Aveiro, Portugal. Tel.: þ351 234 370 780; fax: þ351 234 426 408. E-mail address: [email protected] (T. Caetano).

Lichenicidin is a class II lanthipeptide composed by two peptides (Blia and Blib) that act synergistically to inhibit the growth of several Gram positive bacteria, including Staphylococcus aureus and Listeria monocytogenes [11,12]. Previously, the study of lichenicidin biosynthesis in Escherichia coli revealed that the modification modifying enzymes, LicM1 and LicM2, tolerate various amino acid substitutions within the LicA1 and LicA2 core peptides [13]. Most excitingly, analogs of active Blia and Blib containing non-canonical amino acids were produced in vivo in E. coli [14]. Among the two-component peptide lantibiotics already characterized, lichenicidin is more related to haloduracin (Hala and Halb), which is produced by Bacillus halodurans C-125 from modification of HalA1 and HalA2 by the dedicated enzymes HalM1 and HalM2, respectively [15,16]. The LicA1 and LicA2 precursor peptides have similarities of approximately 52% and 50% with HalA1 and HalA2, respectively. Considering their corresponding enzymes dedicated to modification and transport, it was found that: i) LicM1 and LicM2 each display 35% similarity with modifying enzymes HalM1 and

http://dx.doi.org/10.1016/j.resmic.2014.07.006 0923-2508/© 2014 Institut Pasteur. Published by Elsevier Masson SAS. All rights reserved.

T. Caetano et al. / Research in Microbiology 165 (2014) 600e604

HalM2, and ii) LicT shares 50% similarity with its counterpart in the haloduracin biosynthetic cluster, HalT. Regarding their sequence and the functional similarities of haloduracin and lichenicidin, the present study aims to assess the flexibility of the modifying enzymes (LanM) and transporter (LanT) to produce fully active peptides, using E. coli as the producer organism. 2. Results and discussion To achieve our purpose, the system based on the strains produced in the previous study performed by Caetano et al. (2011) was used. Briefly, five different fosmids were obtained by deleting the genes licA1, licA2, licM1, licM2 or licT from the parental pLic5 fosmid containing the complete lichenicidin (lic) biosynthetic gene cluster (Fig. 1A) and transformed into E. coli BL21Gold (Table 1). Subsequently, chemically competent cells of each of these strains were transformed with a complementation plasmid, containing the deleted gene, or its counterpart in the haloduracin (hal ) biosynthetic gene cluster (Table 1; Fig. 1A). The antibacterial activity was assessed by deferred antagonism assays against the indicator strain Micrococcus luteus ATCC 9341, as previously published for the mutagenesis studies of Blia and Blib [13]. Moreover, the presence of lichenicidin and/or haloduracin peptides was investigated in the culture supernatant extracts by LC-ESI-MS, as described by Caetano et al. [17]. As stated above, two-peptide lantibiotics act synergistically to inhibit other bacteria. Accordingly, the gene-inactivation strains constructed in this study presented no antibacterial activity, due to the lack of one or both of the complementary lichenicidin peptides. Nonetheless, the bioactivity was restored after their transformation with plasmids expressing each one of the complementary lic genes. In addition, the molecular masses of Blia and Blib were detected by LC-ESIMS in supernatant extracts of all of the strains (Table 1). These results demonstrated that the trans complementation system herein used was fully functional. The proteins essential for the biosynthesis of haloduracin were also successfully expressed with E. coli BL21 (DE) and the pET-system in previous studies [15,18]. However, when the same experiment was

601

conducted by replacing the missing lic genes by their counterparts of the haloduracin biosynthetic cluster, bioactivity was only restored in the presence of the halT gene (BLic5DTphalT; Table 1). This phenotype indicated that fully active Blia and Blib were produced and secreted by strain BLic5DTphalT, which was confirmed by the detection of their molecular masses using LC-ESI-MS (Table 1). The bifunctional transporters/proteases of class II lanthipeptides like LicT or HalT, commonly referred as LanT, are characterized by a dual role: i) removal of the leader peptide and ii) secretion of the modified and active core peptide [19]. Thus, our results show that HalT is able to replace LicT in these two biosynthetic steps in lichenicidin production. This observation is corroborated by a previous report, whereupon the lacticin 3147 transporter LtnT was able to proteolytically process and transport nisin as well as the hormone angiotensin, when fused to the leader peptide of lacticin LtnA2 [20]. However, in the present study, HalT was able to play its role in the absence of haloduracin leader peptides. A similar outcome was also reported by Nagao et al. (2007), which described the processing of the lantibiotic nukacin ISK-1 by the transporter of lacticin 48s1 (LctT) in the Lactococcus lactis NZ9000 strain. However, the inhibition radius of the Lic5DTplicT (0.68 mm ± 0.04) and Lic5DTphalT (0.45 mm ± 0.05) colonies demonstrate that HalT protein should not be as efficient as LicT. The analysis of the four leader sequences recognized by HalT (LicA1, LicA2, HalA1 and HalA2), revealed that LicA1 and HalA1 share the highest similarity (53%), while LicA1 and HalA2 share the lowest (29%). Together, these results suggest that LanT proteins are extremely flexible regarding the amino acid residues of the leader peptides that they recognize. Focusing only in the proteolysis reaction, it was determined that the Glu residue of the consensus sequence ELXXBX (B ¼ V, L or I), characteristic of class II leader peptides, is a possible recognition site [21]. In fact, this residue is found in all of the leader sequences processed by HalT (Fig. 1). Interestingly the consensus sequence of LicA1 and HalA1 is longer than expected (ELX5(X)BXX; Fig. 1B). Therefore, the distance between the conserved residue Glu and the double-glycine motif should not affect the recognition and removal of leader peptides by LanT proteins. In addition, it

Fig. 1. Schematic representation of the lichenicidin and haloduracin gene clusters (A) and ClustalW alignment of the leader peptides of mersacidin (MrsA), lichenicidin (LicA1 and LicA2), haloduracin (HalA1 and HalA2), lacticin 481 (LctA) and nukacin ISK-1 (NukA). The consensus sequence characteristic of the leader peptides of class II lanthipeptides (ELXXBX (B ¼ V, L or I)) is highlighted with a black box (B).

602

T. Caetano et al. / Research in Microbiology 165 (2014) 600e604

Table 1 List of gene-inactivation strains used in this study with the respective genes deleted, complemented (with the pET24-a(þ) plasmid) and strains obtained. The bioassay and LC-ESI-MS results for each strain are also included. þ and e represent the presence or absence of antibacterial activity, respectively. The masses represented the following peptides: Blia (3250 Da), Blib (3020 Da), Hala (2331 Da) and Halb (3043 Da). Knockout strain

Deleted gene

Gene complemented

Final strain

Colony bioassay

BLic5DA1

licA1

licA1

BLic5DA1plicA1

þ

halA1

BLic5DA1phalA1



licA1halA1

BLic5DA1pchimA1



licA2

BLic5DA2plicA2

þ

halA2

BLic5DA2phalA2



licA2halA2

BLic5DA2pchimA2



licA2-NDVNPE-halA2

BLic5A2pchimA2.1



licM1

BLic5DM1plicM1

þ

halM1

BLic5DM1phalM1



licM2

BLic5DM2plicM2

þ

halM2

BLic5DM2phalM2



licT

BLic5DTplicT

þ

halT

BLic5DTphalT

þ

BLic5DA2

BLic5DM1

BLic5DM2

BLic5DT

licA2

licM1

licM2

licT

was demonstrated that HalT is able to cleave after all the double GG motifs known, since LicA1, LicA2, HalA1 and HalA2 have the motifs GG, GA and GS as proteolysis sites (Fig. 1B). Previous studies demonstrated that the leader peptide is not essential for post-translational modification by the LanM proteins, but can greatly increase the efficiency of modification [22]. Thus, the ability of LicM1 and LicM2 to correctly modify the haloduracin peptides was herein investigated by complementing BLic5DA1 and BLic5DA2 strains with the plasmids encoding the HalA1 and HalA2 precursor peptides, respectively (Table 1). A similar approach was used to perceive if HalM1 and HalM2 enzymes were able to modify the lichenicidin peptides by complementing BLic5DM1 and BLic5DM2 strains with the plasmids encoding, respectively, the modifying enzymes HalM1 and HalM2 (Table 1). As mentioned above, none of these strains presented antibacterial activity, neither were the molecular masses of the respective haloduracin or lichenicidin peptides detected by the LC-ESIMS analysis of their extracts. Thus, HalM and LicM enzymes were unable to process appropriately lichenicidin and haloduracin peptides, respectively. These results demonstrated that LanM enzymes are not capable of processing peptides from distinct closely related biosynthetic pathways under in vivo conditions. These observations are in agreement with the reports of Nagao et al. (2007) wherein i) the lacticin 481 mature peptide was not produced when its structural gene

Mass spectra (Da) Expected

Detected

3250 3020 2331 3020 2331 3020 3250 3020 3250 3043 3250 3043 3043 3250 3250 3020 3250 3020 3250 3020 3250 3020 3250 3020 3250 3020

3250 3020 e 3020 e 3020 3250 3020 3250 e 3250 e e 3250 3250 3020 e 3020 3250 3020 3250 e 3250 3020 3250 3020

(lctA) was expressed in a nukacin ISK-1 producer and ii) where nukacin ISK-1 was not produced with the replacement of NukM enzyme by LctM (modification enzymes of nukacin and lacticin 481, respectively), presenting 55% of similarity. On the contrary, in vitro assays showed that LctM is able to process NukA substrate into a 4-fold dehydrated peptide. In this case and after proteolytic removal of the leader sequence, the compound showed weak activity, indicating that the pattern of nukacin's thioether rings was not correctly formed [23]. In the absence of a proper cyclization, the peptides can be more susceptible, for instance, to proteolysis-mediated degradation [24]. This could explain the differences obtained for in vivo and in vitro studies involving the functionality of the LctM enzyme directed by distinct leader peptides. It was previously reported that a mutation in the Leu residue of the abovementioned ELXXBX conserved sequence can strongly affect the efficiency of dehydration in lacticin 481 biosynthesis [22,23]. However, the inefficiency of LicM and HalM proteins to correctly modify haloduracin and lichenicidin percursor peptides cannot be addressed to this residue, since it is present in all leader sequences of LicA1, LicA2, HalA1 and HalA2 (Fig. 1B). Also, it has been proposed that the optimal activity of LanM proteins is dependent on the recognition of the secondary structure of the leader peptides. The analysis of this feature revealed that all herein discussed leader peptides possess a a-helix conformation, with some spatial differences on coil and b-sheet occurrence. Therefore, it is also possible

T. Caetano et al. / Research in Microbiology 165 (2014) 600e604

603

Fig. 2. Amino acid sequences of lichenicidin (LicA1 and LicA2) and haloduracin (HalA1 and HalA2) precursor peptides and the chimeric peptides LicA1LHalA1P (A), LicA2LHalA2P and LicA2L-NDVNPE-HalA2P (B) used in this study.

that these patterns are crucial signals assisting in the correct assembly of thioether rings by LanM proteins. Apparently, the amino acid similarities between the precursor peptides of lichenicidin and haloduracin were not sufficient to attain the functionality of LanM enzymes. With that in mind, the production of haloduracin core peptides (Hala and Halb) directed by LicA1 and LicA2 leader sequences was attempted. To achieve this, the two chimeric genes LicA1LHalA1P and LicA2LHalA2P (Fig. 2), containing lichenicidin leader peptides fused to the haloduracin core peptides, were synthesized by GeneArt®. A third chimera (LicA2L-NDVNPE-HalA2P) was obtained by site-directed mutagenesis, where the first six amino acids of the HalA2 C-terminally from the GG-double motif were replaced by their counterparts of the LicA2 peptide (Fig. 2). Such an approach considered a particular biosynthetic feature of Blib and Halb peptides, which require the removal of the N-terminal hexapeptide after the LanT-mediated proteolysis at the GG-site. In Blib biosynthesis, this reaction is performed by the protease LicP, encoded in the lic cluster [13]. However, until now, a corresponding LanP protease was not identified in the haloduracin gene cluster. The above mentioned three chimeric genes were cloned into pET24-a(þ) vector and transformed either in the strains BLic5DA1 or BLic5DA2 (Table 1). The strains obtained were inactive against M. luteus and the expected molecular masses of haloduracin peptides were not detected in their supernatant extracts. Thus, it is concluded that even in the presence of the lichenicidin leader sequences, LicM1 and LicM2 does not enable the synthesis of correctly modified haloduracin core peptides. These results differ from previous reports, since active class II lantibiotics from Streptococcus spp. were produced using a chimera consisting of their core peptides fused to nisin's leader peptide. This was accomplished in the presence of the biosynthetic machinery of nisin [25]. Also, LctM was shown to catalyze in vitro the dehydration and cyclization of nukacin ISK-1, mutacin II and ruminococcin A precursor peptides when fused to LtcA leader sequence [23]. Thus, this study shows a lack of unifying rules to understand occurrence or absence of post-translational modification of precursor peptides. In vivo conditions by expression in host strains show, that not all core peptides can be fully processed when fused to a leader peptide recognized by a given LanM enzyme. In vitro assays commonly show a significantly greater substrate tolerance. Thus, it would be interesting to

further test the dehydration and cyclization patterns of the lichenicidin and haloduracin chimeric peptides after modification by LicM1 and LicM2 using an in vitro reconstituted biosynthesis system. This would help to understand whether the observed results are the outcome of an incomplete modification of the peptides followed by rapid degradation in the producer host, or if there is a real biosynthetic incompatibility at the enzymatic level. Acknowledgments T^ania Caetano was supported by Fundaç~ao para a Ci^encia e Tecnologia, POPH, European Union and Medinfar Pharmaceuticals SA grants SFRH/BDE/15559/2005 and SFRH/BPD/ 77900/2011. This work was supported by Coli4Lan project (FCOMP-01-0124-FEDER-027569), funded by Fundaç~ao para a Ci^encia e a Tecnologia I.P. (PIDDAC) and by Fundo Europeu de Desenvolvimento Regional e FEDER, through the COMPETE e Programa Operacional Fatores de Competitividade (POFC). This project also supported Joana Barbosa with the BI/UI88/6285/2013 grant. Additionally, the work was also supported by European Funds through COMPETE and by National Funds through the Portuguese Science Foundation (FCT) within project PEst-C/MAR/ LA0017/2013. The work of R.D.S. was supported by the cluster of Excellence “unifying concepts in catalysis” funded by the German Research Council (DFG). Finally, the authors would like to thank you Joanna Krawczyk for the support in chimera's design. References [1] Arnison PG, Bibb MJ, Bierbaum G, Bowers AA, Bugni TS, Bulaj G, et al. Ribosomally synthesized and post-translationally modified peptide natural products: overview and recommendations for a universal nomenclature. Nat Prod Rep 2013;30:108e60. Available from: http:// www.ncbi.nlm.nih.gov/pubmed/23165928. [2] Asaduzzaman SM, Sonomoto K. Lantibiotics: diverse activities and unique modes of action. J Biosci Bioeng 2009;107:475e87. [3] Yoganathan S, Vederas JC. Fracturing rings to understand lantibiotics. Chem Biol 2008;15:999e1001. http://dx.doi.org/10.1016/ j.chembiol.2008.10.001. [4] Prieto ML, O'Sullivan L, Tan SP, McLoughlin P, Hughes H, O'Connor PM, et al. Assessment of the bacteriocinogenic potential of marine bacteria reveals lichenicidin production by seaweed-derived Bacillus spp. Mar Drugs 2012;10:2280e99. http://dx.doi.org/10.3390/md10102280. [5] Field D, Connor PMO, Cotter PD, Hill C, Ross RP. The generation of nisin variants with enhanced activity against specific gram-positive

604

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

T. Caetano et al. / Research in Microbiology 165 (2014) 600e604 pathogens. Mol Microbiol 2008;69:218e30. Available from: http://www. ncbi.nlm.nih.gov/pubmed/18485077. Field D, Hill C, Cotter PD, Ross RP. The dawning of a “Golden era” in lantibiotic bioengineering. Mol Microbiol 2010;78:1077e87. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21091497. Li B, Sher D, Kelly L, Shi Y, Huang K, Knerr PJ, et al. Catalytic promiscuity in the biosynthesis of cyclic peptide secondary metabolites in planktonic marine cyanobacteria. Proc Natl Acad Sci U S A 2010;107:10430e5. http://dx.doi.org/10.1073/pnas.0913677107. Nagao J, Aso Y, Shioya K, Nakayama J, Sonomoto K. Lantibiotic engineering: molecular characterization and exploitation of lantibioticsynthesizing enzymes for peptide engineering. J Mol Microbiol Biotechnol 2007;13:235e42. http://dx.doi.org/10.1159/000104749. Krawczyk B, Ensle P, Mu¨ller WM, Su¨ssmuth RD. Deuterium labeled peptides give insights into the directionality of class III lantibiotic synthetase LabKC. J Am Chem Soc 2012;134:9922e5. Available from: http://dx.doi.org/10.1021/ja3040224. Krawczyk JM, V€ oller GH, Krawczyk B, Kretz J, Br€onstrup M, Su¨ssmuth RD. Heterologous expression and engineering studies of labyrinthopeptins, class III lantibiotics from actinomadura namibiensis. Chem Biol 2013;20:111e22. Available: http://www.sciencedirect.com/ science/article/pii/S1074552112004279. Dischinger J, Josten M, Szekat C, Sahl H-G, Bierbaum G. Production of the novel two-peptide lantibiotic lichenicidin by Bacillus licheniformis DSM 13. PLoS One 2009;4:e6788. Available from: http://www.pubmedcentral.nih. gov/articlerender.fcgi?artid¼2727956&tool¼pmcentrez&rendertype¼abs tract. Shenkarev ZO, Finkina EI, Nurmukhamedova EK, Balandin SV, Mineev KS, Nadezhdin KD, et al. Isolation, structure elucidation, and synergistic antibacterial activity of a novel two-component lantibiotic lichenicidin from Bacillus licheniformis VK21. Biochemistry 2010;49:6462e72. Available from: http://www.ncbi.nlm.nih.gov/ pubmed/20578714. Caetano T, Krawczyk JM, M€osker E, Su¨ssmuth RD, Mendo S. Heterologous expression, biosynthesis, and mutagenesis of type II lantibiotics from Bacillus licheniformis in Escherichia coli. Chem Biol 2011;18:90e100. Available from: http://www.ncbi.nlm.nih.gov/pubmed/ 21276942. Oldach F, Al Toma R, Kuthning A, Caetano T, Mendo S, Budisa N, et al. Congeneric lantibiotics from ribosomal in vivo peptide synthesis with noncanonical amino acids. Angew Chem Int Ed Engl 2012;51:415e8. http://dx.doi.org/10.1002/anie.201106154. McClerren AL, Cooper LE, Quan C, Thomas PM, Kelleher NL, van der Donk WA. Discovery and in vitro biosynthesis of haloduracin, a two-

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

component lantibiotic. Proc Natl Acad Sci U S A 2006;103:17243e8. http://dx.doi.org/10.1073/pnas.0606088103. Lawton EM, Cotter PD, Hill C, Ross RP. Identification of a novel twopeptide lantibiotic, haloduracin, produced by the alkaliphile Bacillus halodurans C-125. FEMS Microbiol Lett 2007;267:64e71. http:// dx.doi.org/10.1111/j.1574-6968.2006.00539.x. Caetano T, Krawczyk JM, M€osker E, Su¨ssmuth RD, Mendo S. Lichenicidin biosynthesis in Escherichia coli: licFGEHI immunity genes are not essential for lantibiotic production or self-protection. Appl Environ Microbiol 2011;77:5023e6. Available from: http://www.pubmedcentral. nih.gov/articlerender.fcgi? artid¼3147384&tool¼pmcentrez&rendertype¼abstract. Shi Y, Yang X, Garg N, van der Donk WA. Production of lantipeptides in Escherichia coli. J Am Chem Soc 2010;133:2338e41. Available from: http://dx.doi.org/10.1021/ja109044r. Knerr PJ, van der Donk WA. Discovery, biosynthesis, and engineering of lantipeptides. Annu Rev Biochem 2012;81:479e505. Available from: http://www.ncbi.nlm.nih.gov/pubmed/22404629. Kuipers A, Meijer-Wierenga J, Rink R, Kluskens LD, Moll GN. Mechanistic dissection of the enzyme complexes involved in biosynthesis of lacticin 3147 and nisin. Appl Environ Microbiol 2008;74:6591e7. http:// dx.doi.org/10.1128/AEM.01334-08. Furgerson Ihnken LA, Chatterjee C, van der Donk WA. In vitro reconstitution and substrate specificity of a lantibiotic protease. Biochemistry 2008;47:7352e63. http://dx.doi.org/10.1021/bi800278n. Levengood MR, Patton GC, van der Donk WA. The leader peptide is not required for post-translational modification by lacticin 481 synthetase. J Am Chem Soc 2007;129:10314e5. Available from: http://dx.doi.org/10. 1021/ja072967þ. Patton GC, Paul M, Cooper LE, Chatterjee C, van der Donk WA. The importance of the leader sequence for directing lanthionine formation in lacticin 481. Biochemistry 2008;47:7342e51. http://dx.doi.org/10.1021/ bi800277d. Chatterjee C, Patton GC, Cooper L, Paul M, van der Donk WA. Engineering dehydro amino acids and thioethers into peptides using lacticin 481 synthetase. Chem Biol 2006;13:1109e17. Available from: http:// www.sciencedirect.com/science/article/pii/S1074552106003395. Majchrzykiewicz JA, Lubelski J, Moll GN, Kuipers A, Bijlsma JJE, Kuipers OP, et al. Production of a class ii two-component lantibiotic of Streptococcus pneumoniae using the class I nisin synthetic machinery and leader sequence. Antimicrobial Agents Chemother 2010;54:1498e505. Available from: http://aac.asm.org/content/54/4/ 1498.abstract.