Introduction

In the frame of our efforts to generate prototype biocontained strains exhibiting geneticand trophic isolation and expanded biological functions1,2 we aimed toexpand our previous attempts to engineer ribosomally synthesized and post-translationalmodified peptides (RiPPs) by ribosomally introducing ncAAs into their sequences.Thereby, we are pursuing Xenobiology with the aim to implement various man-made chemicalsyntheses in living cells. Whereas Synthetic Biology mainly works with naturallyoccuring building blocks and a canonical chemistry, Xenobiology uses non-naturalbuilding blocks and non-canonical chemistries3.

Currently, the development of alternative biological systems with radically alteredgenetic codes implies massive genome engineering4. However, approachesaiming at the generation of cell factories as platforms are still immature, as theygenerally suffer from synthetic lethal mutations, codon reversions and dramaticallydecreased fitness during the genome assembly process5. On the other hand,widely used orthogonal pairs are not as active and accurate as natural aminoacyl-tRNAsynthetases with related cognate tRNAs6. Our alternative strategy forexperimental genetic code evolution towards changes in its biochemistry and to achievebiocontainment relies on the global substitution of canonical amino acids with ncAAsassisted with simple metabolic engineering7,8. Recently, we described along-term evolution experiment which led to the reassignment of all 20,899 Trp codons inthe genetic code of the bacterium Escherichia coli2. Cultivationof the E. coli strain in defined synthetic media resulted in the generation ofthe bacterial strain MT21, which is capable of proteome-wideTrp → [3,2]Tpa substitutions in response to all TGGcodons in the genome. These evolved bacteria with their new-to-nature amino acidcomposition are capable of robust growth in the complete absence of the canonical(natural) amino acid Trp (Fig. 1a,b)2,9.Previously, we and others have applied various methods, aiming to engineer RiPPs byribosomally introducing ncAAs into their sequences in vitro and in vivo,exploiting the natural biosynthetic pathways1,10,11,12. Nevertheless,supplementation-based incorporation (SPI) only allows for the insertion of isostericanalogues of cAAs, the structural diversity of which is restricted by the promiscuity ofthe respective tRNA and aminoacyl-tRNA synthetase and limited by the use of auxotrophicstrains13,14,15. Expanding the structural complexity of the ncAAregardless of the amino acid to be replaced, can be achieved by stop-codon-suppression(SCS) or reassignment of a sense codon but requires the design of new pairs oforthogonal tRNA and the corresponding aminoacyl-tRNA synthetases8,16,17,18,19,20,21,22 and genetic modifications such asintroduction of the respective codon in the addressed gene and removing of suppressortRNAs or release factor 1 for improved yields23,24,25,26. Herein wereport the use of fully adapted E. coli MT21 as a platform for production ofncAAs-containing small-molecule-type antibiotic peptides, which undergo massivepost-translational modifications, being only recently addressed in the frame of singleprotein/peptide recombinant production by using standard expression strains1,10,11. The transfer of xenobiological concepts and ideas to peptideswith antibiotic properties opens up a new structural space for various compound classesand thus possibly altered or enhanced bioactivities. Peptide natural products, which areribosomally synthesized and post-translationally modified peptides (RiPPs) comprise ofvarious subgroups, e.g. lanthipeptides27,28,29,30, microviridins31,32, lasso peptides33, or linear azole containingpeptides34,35 with various characteristic structural features36. We apply the assembly of the otherwise toxic amino acidl-β-(thieno[3,2-b]pyrrolyl)alanine ([3,2]Tpa)37 (Fig. 1b) to an evolved E. coli strainwhich carries the gene cluster for the heterologous production of the congenericlantibiotic lichenicidin. Lichenicidin is a two-component lantibiotic originating fromBacillus licheniformis38. The two peptides, Bliαand Bliβ, are assumed to act synergistically on the cell wall ofGram-positive bacteria in a manner that has been similarly described for othertwo-component lantibiotics39,40,41,42. In this scenario, theα-peptide binds to the peptidoglycan precursor lipid II, and theβ-peptide is subsequently recruited to the cell wall to induce poreformation43,44,45. The lichenicidin gene cluster (liccluster, 15 kb) comprises of 14 genes (see Supplementary Fig. S1)46, of which onlysix are essential for heterologous expression of the lichenicidin peptides(Bliα and Bliβ) in E. coli47. Productionof Bliα and Bliβ includes a number of biosynthetic steps (Fig. 1c): subsequent to the ribosomal biosynthesis, anintramolecular crosslinking occurs between dehydrated Ser or Thr and Cys residues toform the diamino diacid lanthionine (Lan) or methyllanthionine (MeLan), respectively.These modifications provide structural stability and rigidity, making lanthipeptidesparticularly attractive compounds as potential novel antibiotics48. ThelicA1 and licA2 structural genes each code for the 72-mer linearprecursor peptide of Bliα and Bliβ, respectively. Twosequence-specific modifying enzymes interact with the leader sequence in thecorresponding precursor peptide and catalyze the thioether formation in the core regionof the respective peptides46. A specific membrane transporter protein,carrying a peptidase domain, removes a large portion of the leader sequence prior to theexport of the peptide from the cell. An N-terminal hexapeptide remains covalently boundto the modified β-peptide and is not removed until the peptide istranslocated outside of the cell, keeping the peptide inactive during the transport. Anextracellular protease cleaves off the remaining part of the leader peptide and releasesthe active peptide (Fig. 1c)46.

Figure 1
figure 1

Strategy and prerequisites for the production of congeneric, ribosomallysynthesized peptides in emancipated E. coli cells.

(a) Evolutionarily adapted E. coli cells are cultivated indefined minimal medium until residual Trp is consumed and cells solely growon [3,2]Tpa. (b) The Trp (2) progenitor indole (1) isreplaced by 4H-thieno[3,2-b]pyrrole (Tp) (3), which inturn is converted into [3,2]Tpa (4) by the tryptophan synthetase.(c) Scheme of the biosynthesis of Bliβ. The linearprecursor peptide is translationally synthesized from the corresponding gene(indicated in black) (I). Dehydrations and thioether bridges areenzymatically installed (II–III) (residues depicted in lightgrey) and the modified peptide is exported via a specific transporter (IV).Extracellularly, a protease activates the peptide by removal of anN-terminal hexapeptide (V). Dhb, didehydrobutyrine; Dha, didehydroalanine;Obu, 2-oxobutyryl; Abu, aminobutyrate.

For the assembly of the Trp-congener [3,2]Tpa (4) we chose theβ-peptide of lichenicidin, because it naturally carries one Trp in position9 of the core peptide (see Supplementary Fig.S2). Another advantage is that it is a genetically manageable RiPP system,which can be applied in the heterologous E. coli host49. Accordingto our approach, by cultivating the evolutionarily adapted strain E. coliMT21(DE3) in minimal medium containing a defined set of amino acids with4H-thieno[3,2-b]pyrrole (Tp) (3) replacing indole (1)(Fig. 1b) will increase the selective pressure in favor oftranslational incorporation of the Trp analogue over Trp into the protein (Fig. 1a). The challenging aspect of our approach is that all of thepreviously described biosynthetic steps must be able to incorporate [3,2]Tpa globallyinto all biosynthesized proteins, including those of the post-translationalmachinery.

Results

Cells of strain E. coli MT21(DE3) were transformed with the plasmidpRSFDuet-1_TPM2A2 (see Supplementary Fig. S1), which carries the required genes forBliβ production in E. coli49. The resulting strainE. coli MT21.1 HPβ was used to express the congenericBliβ carrying [3,2]Tpa. The cells were first cultivated in LB medium asa starter culture and subsequently washed and cultivated in minimal mediumcontaining Tp as a precursor for [3,2]Tpa synthesis, until the remaining Trp wasconsumed (Fig. 1a). Taking the biosynthetic pathway ofBliβ into consideration, we assumed that only the fully processedpeptides are exported from the cell and we expected all active peptides to beexclusively located in the culture supernatant. Consequently, the peptides wereextracted from the supernatant by addition of n-butanol. Indeed, we detectedthe doubly([M + 2H]2+ = 1514.17),triply([M + 3H]3+ = 1009.78)and quadruply([M + 3H + Na]4+ = 763.33)charged molecular masses of the congeneric peptide by HPLC-MS (Fig.2a,b). In order to verify the incorporation of [3,2]Tpa intoBliβ, we additionally performed MS/MS experiments, which confirmed thespecific mass shift of 6 Da (indole[Mcalc = 117.06Da] → 4H-thieno[3,2-b]pyrrole[Mcalc = 123.01 Da]) in the A-ring ofBliβ, thus replacing Trp in the peptide (Fig. 2c).In order to assess the specificity, efficiency and the robustness of the expressionsystem we again analyzed the supernatant extracts by means of ESI-MS. When theadapted cells were cultivated in minimal medium with indole as source for Trpsynthesis, wild type Bliβ was produced (Fig. 3a).If both, indole and Tp are present in the culture medium, indole is preferablyconverted into Trp and used for ribosomal synthesis of the peptides (data notshown). When the adapted cells were cultivated in minimal medium supplemented withTp, the exclusive production of congenericBliβ([3,2]Tpa9) was observed (Fig.3b), which exemplifies the robustness of the expression system by notallowing the production of the wild type Bliβ. To assess the bioactivityof this new-to-nature compound, the concentration was determined by massspectrometric analysis (see SupportingInformation). Dried extracts from a cultivation of the same strain in amedium supplemented with indole, instead of Tp, contained wild typeBliβ. We measured the amount of Bliβ proportional to theamount of Bliβ([3,2]Tpa9) produced by the straincultivated in NMM19 + Tp andNMM19 + indole, respectively and observed a 2-fold decreasein production of the congener compared to the wild type (data not shown). In generalthe peptide yields were much lower than that previously reported49,which can be attributed to the limitations of the non-optimal culture medium (NMM19)and genetic modifications necessary for this experimental setup. Considering thedifferences in the production of Bliβ peptides, we adjusted the amountsof Bliβ and Bliβ([3,2]Tpa9) to 0.5μM and used both in an antimicrobial agar diffusion assay againstMicrococcus luteus (Fig. 4). As expected, theseparate testing of the wild type peptides Bliα and Bliβ didnot show any antibacterial effect, while the combination of both peptides resultedin a clear halo indicating antimicrobial bioactivity. Interestingly, the congenericpeptide Bliβ([3,2]Tpa9) did not show a decrease inbioactivity, suggesting that the introduction of [3,2]Tpa does not influence theoverall structure of the peptide, nor does it negatively affect the interaction withBliα.

Figure 2
figure 2

LC-ESI-MS analytics of congenericBliβ([3,2]Tpa9).

(a) Total ion chromatogram ofBliβ([3,2]Tpa9) extracted withn-butanol. (b) Mass spectrum ofBliβ([3,2]Tpa9)([Mcalc + 2H]2+ = 1514.17,[Mcalc + 3H]3+ = 1009.78)with annotated fragment masses. (c) HR-ESI-MS2analysis of wild type Bliβ([M + 2H]2+ = 1511.18Da) and congeneric Bli([3,2]Tpa9)([M + 2H]2+ = 1514.17Da). Characteristic mass shifts of 6 Da due to incorporation of [3,2]Tpa asa surrogate for Trp are indicated in red color.

Figure 3
figure 3

Relative abundance of wild type Bliβ compared to congenericBliβ([3,2]Tpa9).

E. coli MT21(DE3) cells were cultivated in NMM19 medium supplementedwith (a) indole and (b) 4H-thieno[3,2-b]pyrrole(Tp). Peptides were quantified by HPLC-ESI-MS analysis.

Figure 4
figure 4

Antimicrobial activity of lichenicidin peptides.

Bliα. Bliβ andBliβ([3,2]Tpa9), indicated asBliβ*, were tested separately (concentration0.5 μM) and in equimolar (1:1) combinations againstthe indicator strain Micrococcus luteus (DSM-1790). The assay wasperformed in triplicate.

In this study, we firmly prove our working hypothesis, that the application ofadapted strains is not only suitable for the expression of a one single protein butalso encompasses the possibility for the production of new-to-nature bioactivesecondary metabolites, which are synthesized via complex biosynthetic pathways,requiring a relaxed substrate specificity of the PTM machinery for the alteredpeptide sequence. Moreover, we could demonstrate and confirm the versatileapplicability of the complex biosynthesis of lichenicidin, that involves theinteraction and catalytic reactions of several proteins, with regard to the exchangeof an amino acid with a particular surrogate, beyond techniques aiming at amino acidexchange that have been addressed so far.

Discussion

Reprogrammed cells or proteins equipped with synthetic structures are currentlyusually considered as useful tools for academic research or small applications.However, this engineering can even have practical importance when applications suchas bioremediation (in open systems) biocatalysts or peptide-based drugs (closedsystems) are considered50. Furthermore, the incorporation of variousncAAs into the proteome51 or in some E. coli essentialgenes4,5 can be envisioned as a promising biosafety approach: aslong as the ncAAs is absent from the medium, no bacterial growth is possible. Thisis an important prerequisite for biocontainment which is still not fully achieved inour MT21 strain. Namely, it should be noted that 20,899 TGG codons are onlytrophically reassigned (i.e. the meaning of a codon is redefined throughout thewhole translational machinery for the evolved cells only in the defined syntheticmedium). That means the supplementation of cells in such a medium with the canonicalsubstrate Trp reverses them to ‘natural’ ones as they stillfavor the incorporation of the canonical building block. To achieve anutrient-independent reassignment (i.e. ‘real’ codonreassignment) for the all genome TGG codons in E. coli – anexperimental strategy for biocontainment needs to be developed and executed.

Nonetheless, for the first time we have provided a solid proof-of-principle for theapplication of an evolutionarily adapted E. coli strain in production ofnew-to-nature modified lantibiotics. For future bioengineering purposes, our systemand its improved versions will doubtlessly provide a manifold of opportunities todesign various novel ribosomally synthesized compounds. State-of-the-art modifiedlanthipeptides are produced (semi-) synthetically52,53,54,55, andcurrently are limited to only a few applications in a healthcare setting56,57. However, with our methodology we could open up the opportunityto incorporate non-canonical amino acids, enabling us to push forward the invivo diversification of difficult-to-synthesize RiPPs. Recent reports on thedevelopment of super-pathogens58 emphasize the unabated need for newantibiotics, which can circumvent naturally arising host defense mechanisms59,60. Hence, the engineering of lantibiotics with chemicalstructures, rarely occurring in Nature, is a necessary approach to fill the void indeveloping new antimicrobial compounds61.

Methods

Cloning

The plasmid pSTB7, carrying the trpBA gene originating from Salmonellatyphimurium which is required for conversion of indole into tryptophanwas described earlier2. Additionally, we used the vectorpRSFDuet-1_TPM2A2 which carries four genes required forBliβ production in E. coli49. Forcompatibility reasons we exchanged the kanamycin resistance gene ofpRSFDuet-1_TPM2A2 (See Supplementary Fig. S1) for an ampicillin (amp) resistance byheterologous recombination applying the arabinose-inducibleλ-recombinase system (a kind gift from Dr. Bertolt Gust,Tübingen)62. The ampR gene was amplifiedfrom pET-Duet-1 (Novagen) using the primers AK163(5′-TTCAAATATGTATCCGCTCATGAGACAATAACCC-3′) and AK164(5′-TGTGCGCGGAACCCCTATTTGTTTATTTTTCTAAATACATTCAAATATGTATCCGCTCATGAATTAATTCTTACCAATGCTTAATCAGTGAGGCACC-3′).

Strains

The initial strain used for evolutionary adaption to the non-natural amino acidTpa was E. coli K12 W3110 (CGSC#7679). The generation of thus Trpemancipated strain has been published earlier2 and will only besummarized in brief: the genes for the Trp biosynthesis pathway were deleted(∆trpLEDCBA) and substituted by trpBA on anextrachromosomal plasmid pSTB7. Hence, Trp-synthetase, the gene product oftrpBA, enables the strain to convert indole into Trp, facilitating tofeed the strain either with indole or indole analogues. Adaptation to the indolederivative 4H-thieno[3,2-b]pyrrole (Tp) finally gave rise to thestrain E. coli MT21 which continuously could feed on this substrate. Asthe expression system for lichenicidin requires a T7-polymerase, cells weretransformed with a λDE3-lysogenization kit (Novagen, MerckMillipore). The resulting MT21(DE3) cells were transformed with plasmidpRSFDuet-1_TPM2A2(amp).

Culture Conditions

500 μL of an overnight culture in LB-medium werecollected and washed twice in NMM19 medium (7.5 mL 1 M(NH4)2SO4, 68 mL1 M KH2PO4, 31 mL 1 MK2HPO4, 1.7 mL 5 M NaCl,20 mL 1 M glucose, 1 mL 1 MMgSO4, 1 mL Ca2+ (1 mgmL−1), 1 mL Fe2+(1 mg mL−1), 1 mL traceelements, ad 1 L deionised H2O, supplemented with 19canonical amino acids solution, whereupon Trp has been substituted by4H-thieno[3,2-b]pyrrole (Tp). Chemical synthesis of Tp hasbeen described earlier2. After the second wash the cells wereused for inoculation of a 50 mL culture ofNMM19 + [3,2]Tp (NMM19 medium supplemented with0.1 mM of the indole surrogate Tp). The cultures were incubated at37 °C, 200 rpm until they reached stationaryphase. The procedure was repeated for another selection round. From the second50 mL culture a total of 10 L of main expression culturewas inoculated. The main cultures were incubated until optical density was 0.2at OD600. Gene expression was induced by addition of0.5 mM IPTG (f.c.) and cultures were incubated at30 °C, 160 rpm for 20 h. Forproduction of wild type lichenicidin the strains E. coli HPαand E. coli HPβ were cultivated as described earlier49.

Peptide Extraction

Cultures were harvested by centrifugation and supernatant was collected as fullyprocessed congeneric peptides were expected to be exported from the cell. 1/5volume of nBu was added to the supernatant and incubated shaking. DriednBu extracts were dissolved in 70% ACN and precipitated in ice-coldacetone for 16 h. Pure Bliα and Bliβ wereisolated as described earlier49.

Mass Spectrometric Analysis and Quantification

LC-ESI-MS and LC-ESI-MS2 experiments were performed on anESI-LTQ-Orbitrap (Thermo Scientific). For chromatographic separation a Grom-Sil120 ODS-5 ST(100 mm × 2 mm,5 μm) column (GRACE) was used with an Agilent 1260 HPLCsystem. A gradient starting at 5% solvent B, increasing to 100% solvent B over10 min, then held at 100% solvent B for 3 min, then over0.1 min to 5% solvent B followed by 3.9 min isocratic at5% solvent B was applied with a flow rate of 0.2 mL min-1 (solventA: H2O + 0.1% HFo, solvent B:ACN + 0.1% HFo). MS2 spectra wereobtained from an IDA Top2 scan using HCD(CE = 35 eV). For quantificationLC-ESI-MS/MS using multiple reaction monitoring (MRM) analytics were performedon an ESI-Triple-Quadrupole LC-MS 6460 with a preceding Agilent 1290 InfinityHPLC system (Agilent Technologies). A GRACE Grom-Sil 300 Octyl-6 MB(2 × 50 mm,3 μm) column was applied for an acetonitrile gradientstarting at 5% B, then increasing to 20% B in 0.5 min, then to 70% Bin 4 min, and finally to 100% B in 0.2 min, followed bya 1.3 min isocratic hold on 100% B. The flowrate was0.5 mL min−1. For quantitation oflichenicidin peptide yields the[M + 3H]3+ adducts of the peptideswere used as precursor ions. For MRM the mass transitions for Bliβm/z 1007.8 → 1302.0, and m/z1007.8 → 265.1 and forBliβ([3,2]Tpa9) m/z1009.5 → 1304.5, and m/z1009.5 → 265.1 were used. Peptideconcentrations were compared to a standard curve from purified wildtypeBliβ (see Supplementary Fig.S4).

Antibacterial Assay

Antibacterial activity was assessed in Mueller Hinton Broth Agar Plates (Difco)against Micrococcus luteus DSM-1790 at a final concentration of 0.02OD600. Supernatant extracts from cultures expressingBliβ or Bliβ([3,2]Tpa9) were analyzed bymass spectrometry on an ESI-Triple-Quadrupole with respect to compoundconcentration and compared to a standard curve. The respective compound wasdiluted to a final concentration of 0.5 μM and mixedwith equal amounts of purified Bliα in 70% ACN and applied to a5 mm well on the plate. Inhibition zones were determined after18 h incubation at 30 °C.

Additional Information

How to cite this article: Kuthning, A. et al. Towards Biocontained CellFactories: An Evolutionarily Adapted Escherichia coli Strain Produces aNew-to-nature Bioactive Lantibiotic Containing Thienopyrrole-Alanine. Sci.Rep.6, 33447; doi: 10.1038/srep33447 (2016).