Skip to main content

In vivo characterization of the activities of novel cyclodipeptide oxidases: new tools for increasing chemical diversity of bioproduced 2,5-diketopiperazines in Escherichia coli



Cyclodipeptide oxidases (CDOs) are enzymes involved in the biosynthesis of 2,5-diketopiperazines, a class of naturally occurring compounds with a large range of pharmaceutical activities. CDOs belong to cyclodipeptide synthase (CDPS)-dependent pathways, in which they play an early role in the chemical diversification of cyclodipeptides by introducing Cα-Cβ dehydrogenations. Although the activities of more than 100 CDPSs have been determined, the activities of only a few CDOs have been characterized. Furthermore, the assessment of the CDO activities on chemically-synthesized cyclodipeptides has shown these enzymes to be relatively promiscuous, making them interesting tools for cyclodipeptide chemical diversification. The purpose of this study is to provide the first completely microbial toolkit for the efficient bioproduction of a variety of dehydrogenated 2,5-diketopiperazines.


We mined genomes for CDOs encoded in biosynthetic gene clusters of CDPS-dependent pathways and selected several for characterization. We co-expressed each with their associated CDPS in the pathway using Escherichia coli as a chassis and showed that the cyclodipeptides and the dehydrogenated derivatives were produced in the culture supernatants. We determined the biological activities of the six novel CDOs by solving the chemical structures of the biologically produced dehydrogenated cyclodipeptides. Then, we assessed the six novel CDOs plus two previously characterized CDOs in combinatorial engineering experiments in E. coli. We co-expressed each of the eight CDOs with each of 18 CDPSs selected for the diversity of cyclodipeptides they synthesize. We detected more than 50 dehydrogenated cyclodipeptides and determined the best CDPS/CDO combinations to optimize the production of 23.


Our study establishes the usefulness of CDPS and CDO for the bioproduction of dehydrogenated cyclodipeptides. It constitutes the first step toward the bioproduction of more complex and diverse 2,5-diketopiperazines.


2,5-diketopiperazines (2,5-DKPs) are a large class of molecules characterized by a 2,5-DKP ring, resulting from the condensation of two α-amino acids [1]. Originally isolated from natural products more than 100 years ago, they have then been largely studied and developed in medicinal chemistry for their noteworthy biological activities [1]. 2,5-DKPs exhibit a large spectrum of chemical structures, ranging from simple cyclic dipeptides (CDPs) to polycyclic compounds carrying various chemical modifications. Their biosynthesis has been explored recently with the discovery of dedicated biosynthetic pathways, revealing the diversity of enzymes involved in the assembly and the tailoring of the CDP core [2,3,4,5]. In the context of synthetic biology, the characterization and the exploitation of enzymes of the 2,5-DKP biosynthetic pathways appear to be an efficient means to access a novel chemical diversity of potentially bioactive 2,5-DKPs [6].

Cyclic dipeptide oxidases (CDOs) are tailoring enzymes found in 2,5-DKP biosynthetic pathways that depend on cyclodipeptide synthases (CDPSs) [2, 7]. CDPSs catalyse the first step of these pathways by using aminoacyl-tRNAs (aa-tRNAs) to synthesize CDPs. Then tailoring enzymes introduce chemical modifications into the CDPs, including the dehydrogenation of the Cα-Cβ bonds catalysed by CDOs [2, 4, 8]. Four CDO-containing pathways have been unravelled, revealing the activities of the corresponding CDOs: AlbA/B of the albonoursin biosynthetic pathway from Streptomyces noursei, Ndas_1146/1147 of the nocazines biosynthetic pathway from Nocardiopsis dassonvillei, Gut(BC)24309 of the guanitrypmycins biosynthetic pathway from Streptomyces monomycini, and PcmB/C of the purincyclamide biosynthetic pathway from Streptomyces chrestomyceticus [9,10,11,12] (Fig. 1). Furthermore, incomplete data have been obtained for a fifth CDO-containing pathway encoded in Nocardiopsis prasina, with CDPS-Np synthesizing cYF and cYY and CDO-Np catalysing the tetra-dehydrogenation of these CDPs [13, 14]. CDOs are comprised of two subunits of approximately 21 and 11 kDa, referred to hereafter as CDOA and CDOB, respectively. CDOA and CDOB are encoded by two genes that overlap by 20-30 nucleotides, a feature believed to be important for CDO production as attempts to produce recombinant CDO from two independent monocistronic operons have systematically failed [9, 10]. CDOA, CDOB, and a flavinic cofactor assemble into an active high molecular weight heteropolymeric CDO that uses molecular oxygen to catalyse CDP dehydrogenation [10, 15].

Fig. 1
figure 1

Activities of CDPSs and CDOs in biosynthetic pathways. The names of the final products of the pathways are indicated. Dashed arrows indicate hypothetical subsequent tailoring steps in the biosynthetic pathway. The activities of PcmA (CDPS) and PcmB/C (CDO) involved in purincyclamide biosynthesis are similar to those of GutA24309 and Gut(BC)24309, respectively, and purincyclamide is identical to guanitrypmycin A2-1 [11, 12]

Studies focusing on CDO specificity and the utilization of CDOs as 2,5-DKP tailoring enzymes are scarce. Preliminary studies on CDO specificity showed that they can accept CDPs other than those synthesized by their associated CDPSs, with the CDPs having an aromatic side chain being better substrates [15,16,17]. Additionally, the group of Kanzaki reported the conversion of phenylahistine, a prenylated CDP, to its dehydro- derivative by the CDO from Streptomyces albulus KO-23, showing the acceptance of more complex CDP substrates by CDOs [18]. Recently, AlbA/B, Ndas_1146/1147, and CDO-Np were investigated for the conversion of a large set of chemically synthesized cyclodipeptides and analogues [14]. The authors used either bioconversion using recombinant Escherichia coli bacteria that overproduce CDO or in vitro conversion using a bacterial crude extract containing recombinant CDO. The CDOs exhibited a strong preference for CDP containing l-Phe or l-Tyr. Compounds with a benzodiazepine di-one ring instead of the 2,5-DKP ring were not converted, suggesting that the 2,5-DKP ring may be an important substrate determinant for CDO activity. This recent study also showed differences in specificity between CDOs, highlighting the interest of having a wide variety of CDOs for modifying a large range of CDPs. Thus, investigating the activity of novel CDOs should provide new tools for CDP chemical diversification, while allowing a better understanding of the biosynthetic pathways the CDOs are associated with.

Over the last 10 years, the number of CDPSs has exploded, whether identified in genomes as putative enzymes or biochemically characterized. Several hundreds non-redundant putative CDPSs have been identified in genomes [5, 8, 19, 20], and the activities of approximately 120 CDPSs have been described by following the appearance of CDPs secreted into the culture supernatants of E. coli bacteria overexpressing a CDPS [10, 13, 17, 19, 21,22,23,24,25,26,27]. A global analysis of 721 cyclodipeptide biosynthetic gene clusters showed the association of a CDPS gene with CDO genes in 42 of them [20]. Furthermore, many of these associations are original in terms of the biosynthetic gene cluster composition or the activity of the associated CDPS, suggesting novel activities for the corresponding CDOs. The straightforward production of active CDPSs and CDOs in E. coli led us to implement a completely recombinant approach to characterize the activity of novel CDOs. Here, we describe the bioinformatic analysis of the CDO subunits, the characterization of the activities of seven previously undescribed CDOs in E. coli, and the evaluation of their usefulness for the bioconversion of various CDPs produced by CDPSs in combinatorial approaches using E. coli as a host organism.


Bioinformatic analysis of putative CDO subunits encoded in genomes

We used bioinformatic tools at the Enzyme Function Initiative website to obtain a global view of CDOs encoded in genomes. We collected 1610 protein sequences homologous to CDOA-Snou11455 (AlbA) from the UniProt database and generated a protein sequence similarity network [28] (SSN; Fig. 2a). CDOA-Snou11455 and CDOA-Ndas43111 (CDOA subunits of the two most-studied CDOs) are clustered with a first set of CDOA sequences sharing over 52% sequence identity (nodes coloured in green in Fig. 2a). Examination of the genomic environment of the corresponding cdoA genes showed the association of CDOB- and CDPS-encoding genes. The SSN also displays a second set of sequences linked to the previously identified CDOA proteins by their edges (nodes coloured in salmon in Fig. 2a). Percentage identity between sequences of the first and second set are between 30 and 42%. Inspection of the genomic environment of the genes encoding sequences of this second set revealed the presence of a cdoB gene, but the absence of an associated CDPS gene. We continued our study focusing on the first set of CDOA sequences.

Fig. 2
figure 2

Diversity of CDOs. a A sequence similarity network was constructed with 1610 sequences homologous to CDOA-Snou11455, which are represented as nodes. Edges between nodes indicate sequence identity > 26.5%. The nodes corresponding to CDOA-Snou11455 and CDOA-Ndas43111 are shown in red. Nodes corresponding to CDOA subunits, for which the gene is associated in the genome with cdoB and cdps genes are shown in green. Salmon-coloured nodes correspond to CDOA subunits linked by an edge to CDOA subunits coloured in green or red. b Phylogenetic tree of 32 selected sequences of CDOA subunits (left panel) and CDOB subunits (right panel). The main product of the activity of the CDPS associated with the corresponding CDO in the biosynthetic gene cluster is indicated as cXX, X being one l-amino acid. Subunit names of CDOs of S. noursei ATCC 11455, S. monomycini NRRL B‐24309, S. varsoviensis NRRL B‐3589 and N. dassonvillei DSM 43111 are written in red. Clades encompassing subunits of CDOs of S. noursei ATCC 11455 and N. dassonvillei DSM 43111 are highlighted by rectangles coloured in yellow and blue, respectively. Subunit names of CDOs investigated in this study are coloured in green

Using CDOA-Snou11455 and CDOA-Ndas43111 sequences as an entry point, we collected 30 additional CDOA sequences from the NCBI database, together with their associated CDOB sequences, thus constituting two sets of 32 sequences (Additional file 1: Table S1). Phylogenetic trees of CDOA and CDOB sequences were built using tools of the EMBL-EBI website (Fig. 2b). The sequences of subunits of CDO-Snou11455 and CDO-Ndas43111 are located within two clades (Fig. 2b). To expand the diversity of studied CDOs, we selected CDOs for which the subunit sequences are located outside these two clades (Fig. 2b).

Co-expression of putative CDOs with their associated CDPSs results in the production of dehydrogenated cyclodipeptides

At the beginning of this study, two of the selected CDOs, CDO2-Aoli43269 and CDO-Said5739, were found in biosynthetic gene cluster (BGC) with CDPSs of unknown activity, CDPS2-Aoli43269 and CDPS-Said5739 (Additional file 1: Table S2). We investigated the activity of these two CDPSs as previously reported [19, 23]. CDPS2-Aoli43269 synthesized exclusively cWW. Three different cyclodipeptides were detected in culture supernatants of bacteria overexpressing CDPS-Said5739, the most abundant being cWL (97% according to its peak area on the chromatogram recorded at 214 nm), the other cyclodipeptides being cWF (3%), and cFL (detected only by ionic current). This is consistent with a recent characterization of the activity of DmtB3 of S. aidingensis (100% identical to CDPS-Said5739) [29].

We characterized the activities of the novel CDOs by co-expressing the CDPS and CDO genes from the same BGC in E. coli and examining the resulting 2,5-DKPs in SPE-treated culture supernatants by LC–MS/MS (Fig. 3). The five main CDPs produced by CDPS-Npot45234 (cFA, cFM, cFY, cFL, and cFF) were recovered and compounds with m/z values corresponding to dehydrogenated CDPs were also detected (Fig. 3b). The predicted ∆cFF (m/z 293, RT = 26.1 min) was the most abundant, shown by the corresponding peak on the UV chromatogram. Most of the other predicted dehydrogenated CDPs were produced in very low amounts. CDPS1-Aoli43269 and CDPS-Scat2342 produced mainly cYF (Fig. 3c, d). The co-expression of the cognate CDO led to the detection of two compounds predicted to be ∆cYF (m/z 309) in both cases, with RTs of 21.4 and 22.3 min, respectively. The compounds characterized by a similar RT in each sample displayed similar fragmentation patterns (Additional file 2: Fig. S1). The co-expression of CDPS and CDO of S. rimosus led to cWY and two compounds corresponding to the predicted di-dehydrogenated CDPs at m/z 348 (Fig. 3e). The main product of the CDPS-Ssp5123 activity, cWP, was observed in low amounts and a large peak containing the predicted ∆cWP at m/z 282 was observed on the UV chromatogram (Fig. 3f). The co-expression of the enzymes of S. aidingensis led to a large peak on the UV chromatogram at m/z 298, corresponding to the predicted ∆cWL, and the CDP cWL at m/z 300 corresponding to a smaller peak (Fig. 3h). Finally, we detected no predicted dehydrogenated CDPs in samples resulting from the co-expression of CDPS2 and CDO2 of A. oligospora, although cWW, the product of CDPS2-Aoli43269 activity, was clearly produced in high amounts (Fig. 3g).

Fig. 3
figure 3

Detection of CDPs and their derivatives in culture supernatants of E. coli bacteria overexpressing CDPS and CDO. Culture supernatants of E. coli bacteria overexpressing no enzyme (a) or a CDPS/CDO couple from one pathway (b, N. potens; c, pathway 1 of A. oligospora; d, S. catenulae; e, S. rimosus; f, S. sp. NRRL 5123; g, pathway 2 of A. oligospora; h, S. aidingensis) were collected, treated by SPE, and analysed by LC–MSMS. UV traces recorded at 214 nm are shown between 16 and 31 min. The Y axes of the UV214 traces were set from 20 to 280 mAU. Recovered CDPs and the m/z values of their MH+ ions are indicated in grey. Stars indicate CDPs that have not been previously described for the characterized CDPSs. The m/z values of detected MH+ parent ions corresponding to dehydrogenated CDPs (loss of 2 and 4 atomic mass units) are shown in brown. The experiment was performed three times with similar results. Compounds chosen for purification and NMR characterization are indicated in black bold Arabic numbers in brackets (15)

We performed LC-HRMS analysis of the SPE-treated culture supernatants to obtain a more accurate mass of the predicted dehydrogenated products. The 2,5-DKPs detected upon co-expression of a CDPS and a CDO originating from the same pathway are summarized in Table 1. The potential dehydrogenated CDPs recovered by LC-HRMS were measured with high mass accuracy (∆m/z ranged from 0 to 0.86 ppm of the [M + H]+ parent ions) and precision (two independent experiments, each injected twice). Thus, the LC-HRMS data confirmed the presence of dehydrogenated CDPs suggested by the LC–MS/MS data, except for the compound at m/z 348 (predicted ∆cWY) produced by the combination of CDPS1-Aoli43269 and CDO1-Aoli43269 (Fig. 3c). Some of the other differences between the LC–MS/MS and LC-HRMS data are related to the number of isoforms found for one dehydrogenated CDP (in particular for ∆cFY).

Table 1 LC-HRMS analysis of SPE-treated supernatants of bacterial cultures expressing a CDPS/CDO couple originating from the same operon

Altogether, these results show that all but one studied CDOs (CDO2-Aoli43269 excepted) catalyse the efficient dehydrogenation of the major cyclodipeptide(s) synthesized by the associated CDPS in a BGC, albeit with different efficiencies.

Structural characterizations of bio-produced 2,5-DKPs reveal dehydrogenation site preferences for CDOs

We bio-produced and purified five predicted dehydrogenated CDPs (compounds 15 in Fig. 3). We structurally characterized them by 1H, 13C, and 15N NMR spectroscopy (Additional file 1: Tables S3–S4 and Additional file 2: Fig. S2–S11). The analysis of 2D 1H, 13C heteronuclear multiple bond correlation (HMBC) experiments enabled us to identify the site of desaturation of di-dehydrogenated CDPs. Furthermore, the Z configuration of the double bond could be unambiguously established through the measurement of 3J (1H, 13C) coupling constants, showing a cis relationship between the alkene proton and the carbonyl group, together with the observation of through space ROE correlations between the amide proton and side-chain protons of the desaturated amino acid (Additional file 1: Table S4).

The compounds 1 and 2 resulted from the co-expression of CDPS1-Aoli43269 and CDO1-Aoli43269 and were predicted derivatives of cYF (Fig. 3c). NMR data showed that 1 corresponds to the Z isomer of cY∆F and 2 the Z isomer of c∆YF (Additional file 1: Tables S3–S4 and Additional file 2: Fig. S2–S5). Two predicted di-dehydrogenated cWY were observed upon co-expression of CDPS-Srim3904 and CDO-Srim3904 (Fig. 3e). We purified 3, but did not succeed in purifying the other compound with a [M + H]+ parent ion at m/z 348. NMR data indicated that 3 is the Z isomer of cW∆Y (Additional file 1: Tables S3–S4 and Additional file 2: Fig. S6–S7). Compound 4 was the only predicted cWP derivative detected in LC–MS/MS analysis of samples obtained upon co-expression of CDPS-Ssp5123 and CDO-Ssp5123 (Fig. 3f). NMR characterization showed that 4 contains a Cα-Cβ double bond on the prolyl moiety of cWP (Additional file 1: Tables S3–S4 and Additional file 2: Fig. S8–S9). Co-expression of CDPS-Said5739 and CDO-Said5739 resulted in the production of 5 as the unique CDP derivative (Fig. 3h) which was shown by NMR to correspond to the Z isomer of cW∆L (Additional file 1: Tables S3–S4 and Additional file 2: Fig. S10–S11).

In vivo evaluation of CDOs for CDP dehydrogenation in combinatorial engineering experiments

Combinatorial engineering of natural product biosynthetic pathways consists of co-expressing enzymes from different pathways in a suitable host to create unnatural biosynthetic pathways and achieve greater chemical diversity. We evaluated CDOs using such an approach by co-expressing them with CDPSs known to synthesize different CDPs. Eighteen CDPSs were selected for the variety of CDPs they synthesize (Additional file 1: Table S2). We co-expressed each of them with each of eight selected CDOs (AlbA/B, Ndas_1146/1147, CDO-Npot45234, CDO1-Aoli43269, CDO-Scat2342, CDO-Srim3904, CDO-Ssp5123, and CDO-Said5739) in E. coli and evaluated the production of CDPs and their dehydrogenated derivatives in culture supernatants by LC–MS/MS. CDPs detected in the CDPS+/CDO– control experiments (Additional file 1: Table S5) were in accordance with previously published CDPS characterizations [19, 23], except for CDPS-Mmed1, which was found to synthesize cLM, in addition to cLL and cLF. Data for the detection of the dehydrogenated CDPs are reported in Additional file 1: Table S5 and summarized in Fig. 4. Two groups could be identified in terms of substrate acceptance in vivo, corresponding to broad-spectrum (AlbA/B, Ndas_1146/1147, CDO-Npot45234, and CDO1-Aoli43269) and narrow-spectrum CDOs (CDO-Scat2342, CDO-Srim3904, CDO-Ssp5123, and CDO-Said5739). In addition to accepting a larger spectrum of substrates, broad-spectrum CDOs produced tetra-dehydrogenated CDPs more frequently and in greater amounts than narrow-spectrum CDOs. In terms of the nature of the converted CDPs, CDOs prefer CDPs with at least one hydrophobic side chain and conversion was never or poorly detected for cAE, cAA, cGN, cAP, cGV, cPP, cCC, cWA, and cWS. Except for cWA and cWS, these CDPs are comprised of amino acids carrying small side chains. We also observed different rates of conversion of CDPs, whether they were produced in high amounts or not (Additional file 1: Table S5 and Fig. 4). For example, cWW produced by CDPS-Scat8057 (UV214 peak area = 18,590) was converted upon co-expression of five different CDOs, whereas its conversion was not detected when it was produced by CDPS-Srim3904 (UV214 peak area = 306). The same was true for cLL, cFL, and cYY. Finally, the number of dehydrogenated derivatives detected for a given m/z value ranged from one to three, as previously observed for characterized CDOs [14, 18, 30], and were initially differentiated by their retention time.

Fig. 4
figure 4

Efficiency of CDOs for in vivo dehydrogenation of CDPs. One CDPS and one CDO were co-expressed in E. coli and dehydrogenated CDPs were detected in culture supernatants by LC-MS/MS. CDPSs are indicated on the left by the suffix used to design them in Additional file 1: Table S2. The CDPs they synthesize are also indicated with percentages representing the proportion of each CDP (ratio of UV peak areas) as observed in an experiment without CDO expression (Additional file 1: Table S9). Δ and ΔΔ designate compounds with a [M + H]+ parent ion with a m/z value corresponding to di-dehydrogenated and tetra-dehydrogenated CDP, respectively. Co-expressed CDOs are indicated at the top on the right. The level of detection of each compound is indicated according to data in Additional file 1: Table S5 by a colour and pattern style (top left). When several compounds were detected for one m/z value, their number is indicated, and data reported in this figure correspond to those of the most highly produced compound. The best conditions for producing derivatives of the major CDPs produced by CDPSs are indicated by red rectangles

One interesting point concerns the utility of these CDOs to produce a desired derivative in large quantities. The best combination(s) to produce the derivatives of the major CDPs produced by CDPSs are highlighted in Fig. 4 (red rectangles). Six of the eight tested CDOs were used in these combinations, two of them belonging to the so-called narrow-spectrum group. Notably, conversion of cWW into ∆cWW was best upon co-expression of CDO-Ssp5123, whereas the associated CDPS-Ssp5123 did not synthesize cWW. In most cases, several CDOs may be used for the conversion of one type of CDP. Certain CDPs (cLL, cWL, cWP, cWY, and cYY) are even very good substrates for several CDOs. Conversely, the production of significant amounts of dehydrogenated cPM was observed only upon co-expression of AlbA/B. Our combinatorial biosynthesis results provide the basis for the choice of CDOs to be used for the bioproduction of 2,5-DKPs in E. coli.


CDOs catalyse the dehydrogenation of the Cα-Cβ bond of CDPs in 2,5-DKP natural product biosynthetic pathways. Although the number of CDO-containing pathways identified in genomes has exploded in the last years, the number of studied CDOs has remained limited. We characterized the activity of seven previously undescribed CDOs to expand our knowledge of these pathways and evaluated the usefulness of CDOs for CDP modification in combinatorial engineering experiments to increase the toolkit of enzymes for 2,5-DKP chemical diversification.

We assessed the activities of the seven original CDOs by co-expressing each in E. coli with the associated CDPS in the BGC and characterizing the products by LC–MS/MS, LC-HRMS, and NMR spectroscopy. Six of the seven studied CDOs efficiently catalysed CDP dehydrogenation. Mass spectrometry and NMR data clearly support the production of cW∆Y by CDO-Srim3904, cW∆P by CDO-Ssp5123, cW∆L by CDO-Said5739, and cY∆F and c∆YF by CDO1-Aoli43269 (Fig. 3, Table 1, Additional file 2: Fig. S2–S11). LC–MS/MS and LC-HRMS data showed that CDO-Scat2342 produced cY∆F and c∆YF, as the retention times and MS2 spectra of the two compounds were very similar to those of cY∆F and c∆YF produced by CDO1-Aoli43269 (Fig. 3, Table 1, Additional file 2: Fig. S1). Concerning CDO-Npot45234, the dehydrogenation of several cFX cyclodipeptides is supported by LC-HRMS data, cF∆F being produced in larger amounts based on the UV and EIC peak areas measured in LC–MS/MS and LC-HRMS chromatograms, respectively (Fig. 3 and Table 1). Finally, we did not observe a significant conversion of cWW by CDO2-Aoli43269, as di- and tetra-dehydrogenated cWW were only detected in trace amounts by LC-HRMS. We cannot exclude that CDO2-Aoli43269 may have little activity under our experimental conditions, thus resulting in the extremely low conversion of cWW. However, inspection of the genomic environment of CDPS2-Aoli43269 shows the presence of a cytochrome P450 (P450) gene (Additional file 2: Fig. S12). Many P450s have been described in 2,5-DKP biosynthetic pathways, and this encoded P450 may be involved in the modification of cWW before its conversion by CDO.

Our data on the active CDOs provide details on several biosynthetic pathways in which they are involved (Additional file 2: Fig. S12). CDO-Srim3904 is probably part of a guanitrypmycin biosynthetic pathway (Fig. 1). Its activity is similar to that of Gut(BC)24309, and the P450 and methyl transferase encoded by genes surrounding the CDO-Srim3904 gene share high sequence identity with the P450 GutD24309 and the methyl transferase GutE24309 involved in guanitrypmycin biosynthesis (83% over 80% of the entire sequence of GutD24309 and 79% over 86% of the entire sequence of GutE24309, respectively) [11]. The genes of CDO-Said5739 and CDO-Ssp5123 are colocalized with three genes encoding a terpene cyclase (TC), a CDPS, and a phytoene synthase-like prenyl transferase (PT). A similar organization, but lacking the CDO genes, is also observed in Streptomyces youssoufiensis OUC6819. Recently, the TC/CDPS/PT proteins were shown to direct the synthesis of diketopiperazine-terpenes, called pre-drimentines and drimentines [29]. In these pathways, the tryptophanyl-containing CDP cWX synthesized by the CDPS is converted to pre-dimentine by the PT, which adds a farnesyl moiety to the indole ring. Then, the TC catalyses cyclisation of the farnesyl group, resulting in drimentines. A diversity of drimentines was observed, resulting from the various CDPs synthesized by each CDPS, i.e. cWL by CDPS-Said5739, cWP by CDPS-Ssp5123, and cWX by the CDPS of S. youssoufiensis [29]. Our results showing the activity of CDO-Said5739 and CDO-Ssp5123 on cWL and cWP, respectively, suggest that more complex drimentines can be obtained. Furthermore, the presence of methyl transferase genes in the vicinity of the TC/CDPS/PT genes could also account for the chemical diversification of drimentines. Insights into the corresponding biosynthetic pathways of the CDPS/CDO pairs from A. oligospora (pathway 1) and S. catenulae are more limited. The activities of these two pairs are similar, with the production of di- and tetra-dehydrogenated derivatives of cYF. Examination of the genomic environment allows identification of genes encoding potential 2,5-DKP tailoring enzymes that are clearly different between the two organisms (Additional file 2: Fig. S12). However, in the absence of other data, it is difficult to predict the fate of these compounds in the biosynthetic pathways, except that the end products are probably different.

We investigated the utility of eight CDOs for in vivo CDP diversification in combinatorial experiments with 18 CDPSs. We detected significant conversion of 14 of the 27 different CDPs synthesized by CDPSs in vivo into dehydrogenated derivatives (Fig. 4). Globally, CDOs show a preference for CDPs bearing large hydrophobic side chains, such as Leu, Phe, Tyr, or Trp. CDPs carrying small or polar side chains are poor substrates, and cLL, cFL, cFF, cFY, cLW, cWY, and cYY are relatively good substrates for CDOs. We defined broad- and narrow-spectrum CDOs according to the range of converted CDPs. These differences in in vivo activity may arise from the catalytic properties specific to each CDO, as well as from other uncontrolled factors in our experiments, such as the amount of active CDO produced in bacteria or the bioavailability of the CDPs. Furthermore, this narrow-/broad-spectrum distinction does not predetermine the usefulness of a particular CDO. For example, we observed the best conversion rates for cLM (Mmed1), cWW (Scat8057), and cWP (Ssp1868) upon co-expression of narrow-spectrum CDOs. Certain CDPs were efficiently converted by a restricted number of CDOs, such as cPM (Sjap) and cEL (Ppro5), which are good substrates for only AlbA/B, or cWW, for which the conversion to the dehydrogenated derivative was best observed with two narrow-spectrum CDOs. Except for CDO-Scat2342, of which its co-expression did not induce a large range of dehydrogenated derivatives, all other tested CDOs may be useful in generating novel dehydrogenated compounds, thus highlighting the interest of evaluating diverse enzymes for desired biotransformation.

We observed an effect of the level of production of a CDP on its possible conversion by a CDO. Indeed, several highly-produced CDPs gave rise to dehydrogenated derivatives with several CDOs (cLL produced by Mmed1, cFL produced by Rgry, cWW produced by Scat8057), whereas the same compounds produced by other CDPSs, but at a lower level (cLL produced by Ssp5053, cFL produced by Ssp5053, cWW produced by Srim3904), were not transformed by any CDO or only poorly transformed (Fig. 4 and Additional file 1: Table S5). Furthermore, unconverted CDPs were observed in most experiments with co-expression of a CDO (Additional file 1: Table S5), indicating that higher levels of dehydrogenated derivatives could be potentially produced. The CDPS/CDO couple must be chosen carefully, depending on the desired transformation, but synthetic biology approaches aimed at reducing substrate escape by gene fusion or enzyme scaffolding could be developed to optimize the biosynthetic pathway from amino acyl-tRNAs to dehydrogenated CDPs [31].

Recently, S. M. Li et al. assessed the activity of three CDOs, AlbA/B, Ndas_1146/1147, and CDO-Np from Nocardiopsis prasina (CDO-NspL17 in Fig. 2 and Additional file 1: Table S1) on chemically synthesized CDPs in either in vitro assays using CDO-containing cell-free extracts or in feeding experiments to E. coli bacteria expressing each CDO [14]. The authors mentioned the difficulty of the three CDOs to generate ∆Trp-containing derivatives [14]. In spite of the larger number of CDOs tested in this study and their distribution on the phylogenetic tree, we observed similar results, with few ∆Trp-containing derivatives. Indeed, the various ∆cWL, ∆cWY, and ∆cWP produced by several CDPS/CDO combinations had similar LC–MS/MS characteristics (retention time and MS2 fragmentation pattern), indicating that they correspond to the compound identified by NMR spectroscopy as the Z-isomer of the cW∆X derivative (Fig. 4, Additional file 1: Table S5). ∆Trp-containing derivatives were observed only upon cWW conversion, the most efficient conditions being with CDO-Ssp5123 co-expression. Intriguingly, CDO-Ssp5123 can produce ∆Trp derivatives of cWW, but very few or no ∆Trp derivatives of cWL, cWY, or cWP. Our full biosynthetic strategy and the mutasynthetic approaches developed by Li et al. provide two different options for the derivatization of CDPs, each offering advantages and drawbacks. A major benefit of the mutasynthetic approaches resides in the large panel of testable CDPs, provided that they are chemically synthesizable. However, we recently described the in vivo incorporation of unnatural amino acids into CDPs by CDPSs and the production of prenylated CDPs by reprogrammed E. coli bacteria, thus enlarging the panel of biosynthesized 2,5-DKPs [32, 33]. Our study lays the foundations to evaluate the in vivo activity of CDOs in such more complex unnatural 2,5-DKPs biosynthetic pathways.



We homogenized the nomenclature of the CDPS and CDO proteins for clarity. We denoted the two CDO subunits CDOA and CDOB. Furthermore, we added a suffix to indicate the native producing organism. For example, CDPS-Srim3904 and CDO-Srim3904 denote the CDPS and CDO originating from Streptomyces rimosus strain NRRL WC-3904. The two CDO subunits are denoted CDOA-Srim3904 and CDOB-Srim3904. We used the terms CDPS1/CDO1 and CDPS2/CDO2 to distinguish between the enzymes of two CDPS- and CDO-containing pathways found in the same organism. The names of the CDPS and CDO belonging to previously studied BGCs are conserved, i.e. AlbC and AlbA/B of Streptomyces noursei ATCC 11455 [9], Ndas_1145 and Ndas_1146/1147 of Nocardiopsis dassonvillei DSM 43111 [10], GutA24309 and Gut(BC)24309 of Streptomyces monomycini NRRL B-24309 [11], and CDPS-Np and CDO-Np of Nocardiopsis prasina [13, 14].

Throughout the text, cXX refers to CDPs constituted by l-amino acids, X representing each amino acid in the one-letter code. The dehydrogenation of the Cα-Cβ bond is indicated by the symbol ∆. For example, c∆YW refers to cyclo(dehydrotyrosyl-l-tryptophanyl).


Protein Sequence Similarity Network. We used facilities at the Enzyme Function Initiative-Enzyme Similarity Tool (EFI-EST) website to create a protein sequence similarity network (SSN) [28]. First, a set of 1610 protein sequences was retrieved from the Uniprot database using CDOA-Snou11455 (AlbA) as the entry and an E-value of 5. Then the SSN was generated using an E-value of 10−30. Finally, the SSN was visualized and edited using Cytoscape version 3.5.1. In this SSN, two nodes (each representing one protein) are linked by an edge if the two protein sequences have more than 26.5% sequence identity.

Phylogenetic Tree calculation. We searched the NCBI protein database for CDOA and CDOB sequences using CDOA-Snou11455 (AlbA) as the entry (June 2017). We selected CDOA sequences for which the CDOA gene is in proximity of a CDPS gene and a CDOB-encoding gene. We thus constituted two sets of 32 CDOA and CDOB subunits (Additional file 1: Table S1). The CDOA and CDOB phylogenetic trees were calculated using bioinformatic tools at EMBL-EBI ( Two multiple sequence alignments of the CDOA and CDOB sequences were generated using Clustal Omega set with default settings. Alignments were then sent to Simple Phylogeny for tree calculation using the UPGMA clustering method (default setting parameters). Finally, trees were edited using iTol (

DNA manipulation and plasmids

DNA was manipulated according to standard procedures [34]. Molecular biology enzymes were purchased from New England Biolabs. Bacteria of E. coli strain DH5α were used for cloning procedures and plasmid propagation (Invitrogen). They were grown in LB medium at 37 °C unless otherwise stated. When needed, ampicillin and kanamycin were used at 200 µg/ml and 30 µg/ml, respectively. Competent cells were prepared and transformed according to a high-efficiency transformation protocol [35]. Plasmids were purified using the GenElute™ Plasmid MiniPrep kit or the GenElute™ HP Plasmid MidiPrep kit (Sigma-Aldrich). DNA fragments were purified from agarose gels using NucleoSpin Gel and PCR Clean-up (Macherey–Nagel). For DNA sequencing, plasmids were dialyzed against water using 0.025 µm VSWP membrane filters (Millipore) and sent to Eurofins Genomics in Mix2Seq kit tubes.

CDPSs were expressed from pIJ196 derivatives (ColE1 replicon, ampicillin resistance) under the control of the PT5-lacO promoter [23]. Most of the pIJ196-CDPS plasmids used in this study have been described previously and the characteristics of the encoded CDPSs are given in Additional file 1: Table S2 [19, 23]. We constructed pIJ196-CDPS1-Aoli43269 and pIJ196-CDPS-Said5739. Briefly, the synthetic genes were obtained from Life Technologies SAS according to the sequences described in Additional file 1: Table S6. They were cloned in pIJ196 as previously described and the final constructs were sequenced to verify the promoter region and the cloned gene [23].

CDOs were expressed from pIJ194 derivatives (RSF replicon, kanamycin resistance) under the control of the PT7-lacO promoter [33]. CDOA and CDOB genes overlap in genomes, and previous studies have shown that the expression of CDOA and CDOB from different cistrons is detrimental to CDO production [9, 10]. Thus, the CDO genes used in this study corresponded to the native genes with introduction of a NcoI restriction site 5′ and a XhoI restriction sites 3′ for cloning and elimination of undesired restriction sites (Additional file 1: Table S7). The gene of CDO-Snou11455 (AlbA/B) was obtained by PCR SOEing to eliminate a NcoI site in the CDOA gene. Briefly, PCR1 and PCR2 were performed using Phusion DNA polymerase, plasmid pSL150 as matrix, and oligonucleotides and the PCR conditions described in Additional file 1: Table S8 [9]. After purification from an agarose gel, the products of PCR1 and PCR2 were used as matrices in PCR3 to obtain the final DNA fragment (Additional file 1: Table S8). After purification from an agarose gel, it was digested with NcoI and XhoI and the digested DNA fragment cloned between the NcoI and XhoI sites of pIJ194. The sequence of the promoter region and cloned DNA of pIJ194-CDO-Snou11455 was verified. Synthetic genes encoding the other CDOs studied herein were obtained from Life Technologies SAS, according to Additional file 1: Table S7, and were provided cloned in a supplier vector. After digestion with NcoI and XhoI, we purified the CDO gene-containing DNA fragment from an agarose gel and cloned it between the NcoI and XhoI restriction sites of pIJ194. Sequences of the pIJ194-CDO plasmids were verified.

Expression of CDPS and CDO genes in E. coli bacteria

Expression of CDPS and CDO genes for 2,5-DKP production was performed in E. coli bacteria of strain BL21AI (Invitrogen). Competent cells of BL21AI were prepared using the frozen storage III procedure described by Hanahan [36]. We used a two-transformation procedure to introduce pIJ194 derivatives and pIJ196 derivatives into BL21AI. First, plasmid pIJ194 and its pIJ194-CDO derivatives were used to transform BL21AI bacteria using 30 µg/ml kanamycin. Second, competent cells of transformants were prepared according to Hanahan after growing in LB plus kanamycin and used for transformation with pIJ196 and its pIJ196-CDPS derivatives. Double transformants were selected at 37 °C on LB plates containing 30 µg/ml kanamycin and 200 µg/ml ampicillin. They were used to inoculate M9 minimal medium containing 0.5% glucose plus the corresponding antibiotics, trace elements, and vitamins [23]. After 20 h growth at 37 °C, we used the starter cultures to inoculate prewarmed M9 minimal autoinducing medium, containing 0.05% glucose, 0.5% glycerol, 0.2% lactose, and 0.02% arabinose plus trace elements and vitamins, at a 1:100 volume ratio [37]. After 3 h growth at 37 °C in a rotary shaker (210 rpm), production cultures were transferred to 20 °C in a rotary shaker for 44 h. For co-expression of the CDPS and CDO genes from the same biosynthetic pathway, we used 5-ml starter cultures in 50 ml Falcon tubes and 25-ml production cultures in 250 ml Erlenmeyer flasks. For combinatorial engineering experiments with co-expression of CDPS and CDO genes from different pathways, starter cultures and production cultures were made in 24 well-plates (deep well, round bottom; Dutscher) containing 2 ml medium per well and covered with a sterile porous membrane (VWR).

For large-scale production of 2,5-DKP, we performed 350-ml cultures in 3 L Erlenmeyer. The media composition, inoculation ratio, and growth conditions were preserved, except for transfer to 20 °C that occurred after 3 h growth at 37 °C.

The produced 2,5-DKPs were detected in and recovered from culture supernatants as described below.

LC–MS/MS analysis

CDPs and dehydrogenated CDPs were found in the culture supernatants upon expression of CDPS and CDO in E. coli bacteria. Production cultures were acidified to 2% formic acid to stop growth and the culture supernatants were recovered by centrifugation. For the co-expression of CDPS and CDO genes from the same biosynthetic pathway, 5 ml acidified supernatant was treated by solid phase extraction (SPE) using Strata™-X polymeric sorbent following the manufacturer’s instructions (30 mg per tube, Phenomenex). After washing with 5% methanol, elution was performed with 500 µl methanol and the samples conserved at 4 °C in Eppendorf tubes sealed with parafilm until LC–MS/MS analysis. For the co-expression of CDPS and CDO genes from different biosynthetic pathways for combinatorial engineering experiments, acidified supernatants were analysed by LC–MS/MS.

LC–MS/MS analyses were performed on an Elute HPLC (Bruker Daltonics) coupled via a split system to an AmaZon SL ion-trap mass spectrometer set in positive mode (Bruker Daltonics). Samples were loaded onto an Excel 3 C18-PFP column (150 × 4.6 mm, 3 µm, 100 Å, ACE) equilibrated in 0.1% formic acid in water (solvent A) at 0.6 ml/min. After 5 min in solvent A, a 0–50% linear gradient of solvent B (0.1% formic acid in 90:10 CH3CN:H2O) in A was applied over 20 min at the same flow rate.

HRMS analysis

SPE-treated supernatants were analysed by LC-HRMS. Two microliters of the eluates in methanol were injected as technical duplicates in an Ultimate 3000 (Dionex) and separated on a C18 column (3.5 μm, 1 mm × 150 mm, Zorbax-Microbore SB-C18, Agilent Technologies) at a flow rate of 50 μl/min with a 50 min linear gradient from 100% solvent A (2% CH3CN, 0.1% formic acid in H2O) to 80% solvent B (98% CH3CN, 0.1% formic acid in H2O). The LC was coupled to an LTQ Orbitrap XL (Thermo Fisher Scientific) operating in positive-ion mode. All MS spectra were acquired on the Orbitrap with the following parameters: m/z range, 100–1500; resolution, 100000; AGC target, 2 × 105; maximum injection time, 400 ms; lockmass enabled (m/z 279.15909).

Raw files were opened with Thermo Xcalibur Qual Browser (v. 3.0.63) and manually interpreted. Observed m/z values were averaged over the chromatographic peak. Retention time (RT) was reported at the apex of the peak. Reported extracted ion chromatograms were used as is. All species considered in this study were systematically searched in every raw file.

Purification of 2,5-DKPs

Large-scale cultures for the production of 2,5-DKPs were acidified with formic acid (final concentration 2%) and centrifuged to recover the supernatants. Acidified supernatants were loaded onto SPE columns (Strata™-X Giga, 1 g per tube, Phenomenex) conditioned according to the manufacturer’s instructions. A maximum of 500 ml of supernatant was applied per tube. After washing with 5% methanol, elution was performed with 10 ml methanol. Purification of the 2,5-DKPs was continued by semi-preparative HPLC (Hitachi, LP1100/LP3101) using a Purospher Star RP-18e column (250 × 10 mm, 5 µm, VWR) and two solvent conditions (solvent A: 0.1% formic acid; solvent B: 90:10 CH3CN:H2O in 0.1% formic acid) at a flow rate of 4.75 ml/min. Details of the chromatographic conditions optimized for each purified compound are given in Additional file 1: Table S9. Identification of the fraction containing the desired compound, pooling of fractions, lyophilization, and the evaluation of purity were performed as previously described [33].

NMR experiments

NMR experiments were recorded on a Bruker Avance III spectrometer equipped with a TCI cryoprobe and operating at a 1H frequency of 500.3 MHz. Spectra were recorded at 26 °C in DMSO-d6 (Eurisotop). 1H, 13C, and 15N resonances were assigned through the analysis of 1D 1H, 1D 13C DEPTQ, 2D 1H-1H COSY, 2D 1H-1H ROESY, 2D 1H-13C HSQC, 2D 1H-13C HMBC, and 2D 1H-15N HMBC. 1H and 13C chemical shifts were referenced to the DMSO solvent signal (δ 2.50 and 39.5 ppm, respectively) and 15N chemical shifts were referenced indirectly. NMR experiments were processed and analysed using the Bruker Topsin 3.5 program. NMR spectra and NMR data for cY∆F, c∆YF, cW∆Y, cW∆P, and cW∆L are presented in Additional file 2: Fig. S2–S11 and Additional file 1: Table S3. Heteronuclear 3JHβ-CO coupling constants (Additional file 1: Table S4) were measured on a 2D 1H-13C HSQMBC IPAP experiment [38].


E. coli is a widely used genetic background for the production of a large range of natural products. Here, we added the production of various dehydrogenated 2,5-DKPs from carbon sources to the E. coli catalogue. We used the concomitant recombinant expression of CDPSs and CDOs, which allowed the recovery of dehydrogenated 2,5-DKPs directly in the culture supernatants. After the characterization of the activities of six novel CDOs, we implemented a combinatorial engineering approach based on two sets of enzymes consisting of 18 CDPSs and 8 CDOs. Among the 144 combinations, we identified the best pairs for the production of the highest levels of many dehydrogenated 2,5-DKPs. This work constitutes the first step toward the bioproduction of more complex 2,5-DKPs.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.







Biosynthetic gene cluster


Cyclodipeptide oxidase


CDO subunit A


CDO subunit B


Cyclic dipeptide


Cyclodipeptide synthase


Retention time


Solid phase extraction


Sequence similarity network


  1. Borthwick AD. 2,5-Diketopiperazines: synthesis, reactions, medicinal chemistry, and bioactive natural products. Chem Rev. 2012;112:3641–716.

    Article  CAS  Google Scholar 

  2. Belin P, Moutiez M, Lautru S, Seguin J, Pernodet JL, Gondry M. The nonribosomal synthesis of diketopiperazines in tRNA-dependent cyclodipeptide synthase pathways. Nat Prod Rep. 2012;29:961–79.

    Article  CAS  Google Scholar 

  3. Borgman P, Lopez RD, Lane AL. The expanding spectrum of diketopiperazine natural product biosynthetic pathways containing cyclodipeptide synthases. Org Biomol Chem. 2019;17:2305–14.

    Article  CAS  Google Scholar 

  4. Giessen TW, Marahiel MA. The tRNA-dependent biosynthesis of modified cyclic dipeptides. Int J Mol Sci. 2014;15:14610–31.

    Article  CAS  Google Scholar 

  5. Canu N, Moutiez M, Belin P, Gondry M. Cyclodipeptide synthases: a promising biotechnological tool for the synthesis of diverse 2,5-diketopiperazines. Nat Prod Rep. 2019.

    Article  Google Scholar 

  6. Giessen TW, Marahiel MA. Rational and combinatorial tailoring of bioactive cyclic dipeptides. Front Microbiol. 2015;6:1–11.

    Article  Google Scholar 

  7. Aravind L, de Souza RF, Iyer LM. Predicted class-I aminoacyl tRNA synthetase-like proteins in non-ribosomal peptide synthesis. Biol Direct. 2010;5:48.

    Article  CAS  Google Scholar 

  8. Moutiez M, Belin P, Gondry M. Aminoacyl-tRNA-utilizing enzymes in natural product biosynthesis. Chem Rev. 2017;117:5578–618.

    Article  CAS  Google Scholar 

  9. Lautru S, Gondry M, Genet R, Pernodet JL. The albonoursin gene cluster of S. noursei: biosynthesis of diketopiperazine metabolites independent of nonribosomal peptide synthetases. Chem Biol. 2002;9:1355–64.

    Article  CAS  Google Scholar 

  10. Giessen TW, von Tesmar AM, Marahiel MA. Insights into the generation of structural diversity in a tRNA-dependent pathway for highly modified bioactive cyclic dipeptides. Chem Biol. 2013;20:828–38.

    Article  CAS  Google Scholar 

  11. Liu J, Xie X, Li S-M. Guanitrypmycin biosynthetic pathways imply cytochrome P450-mediated regio- and stereospecific guaninyl transfer reactions. Angew Chemie Int Ed. 2019;58:11534–40.

    Article  CAS  Google Scholar 

  12. Shi J, Xu X, Zhao EJ, Zhang B, Li W, Zhao Y, et al. Genome mining and enzymatic total biosynthesis of purincyclamide. Org Lett. 2019;21:6825–9.

    Article  CAS  Google Scholar 

  13. Brockmeyer K, Li SM. Mutations of residues in pocket P1 of a cyclodipeptide synthase strongly increase product formation. J Nat Prod. 2017;80:2917–22.

    Article  CAS  Google Scholar 

  14. Mikulski L, Schäfer J, Brockmeyer K, Kraut R, Li S. Comparative studies on similarities and differences of cyclodipeptide oxidases for installation of C–C double bonds at the diketopiperazine ring. Appl Microbiol Biotechnol. 2020;104:2523–36.

    Article  CAS  Google Scholar 

  15. Gondry M, Lautru S, Fusai G, Meunier G, Ménez A, Genet R. Cyclic dipeptide oxidase from Streptomyces noursei. Eur J Biochem. 2001;268:1712–21.

    Article  CAS  Google Scholar 

  16. Kanzaki H, Imura D, Nitoda T, Kawazu K. Enzymatic conversion of cyclic dipeptides to dehydro derivatives that inhibit cell division. J Biosci Bioeng. 2000;90:86–9.

    Article  CAS  Google Scholar 

  17. Giessen TW, von Tesmar AM, Marahiel MA. A tRNA-dependent two-enzyme pathway for the generation of singly and doubly methylated ditryptophan 2,5-diketopiperazines. Biochemistry. 2013;52:4274–83.

    Article  CAS  Google Scholar 

  18. Kanzaki H, Yanagisawa S, Kanoh K, Nitoda T. A novel potent cell cycle inhibitor dehydrophenylahistin–enzymatic synthesis and inhibitory activity toward sea urchin embryo. J Antibiot. 2002;12:1042–7.

    Article  Google Scholar 

  19. Gondry M, Jacques IB, Thai R, Babin M, Canu N, Seguin J, et al. A comprehensive overview of the cyclodipeptide synthase family enriched with the characterization of 32 new enzymes. Front Microbiol. 2018;9:46.

    Article  Google Scholar 

  20. Skinnider MA, Johnston CW, Merwin NJ, Dejong CA, Magarvey NA. Global analysis of prokaryotic tRNA-derived cyclodipeptide biosynthesis. BMC Genomics. 2018;19:45.

    Article  Google Scholar 

  21. Meng S, Han W, Zhao J, Jian XH, Pan HX, Tang GL. A six-oxidase cascade for tandem C−H bond activation revealed by reconstitution of bicyclomycin biosynthesis. Angew Chemie Int Ed. 2018;57:719–23.

    Article  CAS  Google Scholar 

  22. Gondry M, Sauguet L, Belin P, Thai R, Amouroux R, Tellier C, et al. Cyclodipeptide synthases are a family of tRNA-dependent peptide bond-forming enzymes. Nat Chem Biol. 2009;5:414–20.

    Article  CAS  Google Scholar 

  23. Jacques IB, Moutiez M, Witwinowski J, Darbon E, Martel C, Seguin J, et al. Analysis of 51 cyclodipeptide synthases reveals the basis for substrate specificity. Nat Chem Biol. 2015;11:721–7.

    Article  CAS  Google Scholar 

  24. James ED, Knuckley B, Alqahtani N, Porwal S, Ban J, Karty JA, et al. Two distinct cyclodipeptide synthases from a marine actinomycete catalyze biosynthesis of the same diketopiperazine natural product. ACS Synth Biol. 2016;5:547–53.

    Article  CAS  Google Scholar 

  25. Liu J, Yu H, Li SMM. Expanding tryptophan-containing cyclodipeptide synthase spectrum by identification of nine members from Streptomyces strains. Appl Microbiol Biotechnol. 2018;102:4435–44.

    Article  CAS  Google Scholar 

  26. Seguin J, Moutiez M, Li Y, Belin P, Lecoq A, Fonvielle M, et al. Nonribosomal peptide synthesis in animals: the cyclodipeptide synthase of Nematostella. Chem Biol. 2011;18:1362–8.

    Article  CAS  Google Scholar 

  27. Patteson JB, Cai W, Johnson RA, Santa Maria KC, Li B. Identification of the biosynthetic pathway for the antibiotic bicyclomycin. Biochemistry. 2018;57:61–5.

    Article  CAS  Google Scholar 

  28. Gerlt JA, Bouvier JT, Davidson DB, Imker HJ, Sadkhin B, Slater DR, et al. Enzyme function initiative-enzyme similarity tool (EFI-EST): a web tool for generating protein sequence similarity networks. Biochim Biophys Acta Proteins Proteomics. 2015;1854:1019–37.

    Article  CAS  Google Scholar 

  29. Yao T, Liu J, Liu Z, Li T, Li H, Che Q, et al. Genome mining of cyclodipeptide synthases unravels unusual tRNA-dependent diketopiperazine-terpene biosynthetic machinery. Nat Commun. 2018;9:4091.

    Article  Google Scholar 

  30. Kim MC, Cullum R, Machado H, Smith AJ, Yang I, Rodvold JJ, et al. Photopiperazines A-D, photosensitive interconverting diketopiperazines with significant and selective activity against u87 glioblastoma cells, from a rare, marine-derived actinomycete of the family Streptomycetaceae. J Nat Prod. 2019;82:2262–7.

    Article  CAS  Google Scholar 

  31. Agapakis CM, Boyle PM, Silver PA. Natural strategies for the spatial optimization of metabolism in synthetic biology. Nat Chem Biol. 2012;8:527–35.

    Article  CAS  Google Scholar 

  32. Canu N, Belin P, Thai R, Correia I, Lequin O, Seguin J, et al. Incorporation of non-canonical amino acids into 2,5-diketopiperazines by cyclodipeptide synthases. Angew Chemie Int Ed. 2018;57:3118–22.

    Article  CAS  Google Scholar 

  33. Dubois P, Correia I, Le Chevalier F, Dubois S, Jacques I, Canu N, et al. Reprogramming Escherichia coli for the production of prenylated indole diketopiperazine alkaloids. Sci Rep. 2019;9:9208.

    Article  Google Scholar 

  34. Sambrook J, Russell DW. Molecular cloning: a laboratory manual. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press; 2001.

    Google Scholar 

  35. Inoue H, Nojima H, Okayama H. High efficiency transformation of Escherichia coli with plasmids. Gene. 1990;96:23–8.

    Article  CAS  Google Scholar 

  36. Hanahan D. Techniques for transformation of E. coli. In: Glover DM, editor. DNA Cloning, vol. I. Oxford: IRL Press; 1985. p. 109–36.

    Google Scholar 

  37. Studier FW. Protein production by auto-induction in high density shaking cultures. Protein Expr Purif. 2005;41:207–34.

    Article  CAS  Google Scholar 

  38. Gil S, Espinosa JF, Parella T. IPAP-HSQMBC: measurement of long-range heteronuclear coupling constants from spin-state selective multiplets. J Magn Reson. 2010;207:312–21.

    Article  CAS  Google Scholar 

Download references


The authors would like to thank Robert Thai for his advice for the LC–MS/MS analyses.


This work was supported by the CEA, the CNRS, and one grant from the French National Research Agency (ANR-16-CE29-0026).

Author information

Authors and Affiliations



MG, OL, and PB designed the study; FLC, MB, IC, LM, and PB performed the research experiments; FLC, IC, LM, MM, NC, MG, OL and PB analysed data; PB wrote the first draft of the manuscript and all the authors contributed to and approved the final version; OL and PB obtained financial support. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Pascal Belin.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

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

Supplementary information

Additional file 1: Table S1.

Database information relative to CDOA and CDOB subunits. Table S2. Characteristics of the CDPSs encoded in the pIJ196-CDPS plasmids used in this study. Table S3. NMR data for cYΔF, cΔYF, cWΔY, cWΔP, and cWΔL. Table S4. 3JHβ-CO coupling constants and key ROEs observed for the dehydroaminoacid. Table S5. Detection of CDPs and their derivatives in the supernatants of cultures of bacteria expressing CDPSs alone or in combination with diverse CDOs. Table S6. Database information and sequence data relative to previously uncharacterized CDPSs. Table S7. Sequences of the genes encoding CDOs. Table S8. Oligonucleotides and PCR conditions for the construction of pIJ194-CDO-Snou11455. Table S9. HPLC conditions for the purification of 2,5-DKPs.

Additional file 2: Fig. S1

. Mass spectra of predicted ΔcFY (m/z 309). Fig. S2. 1D 1H NMR spectrum (with presaturation of residual water) and 1D 13C NMR DEPTQ spectrum of cYΔF in DMSO-d6. Fig. S3. 2D 13C-1H HSQC spectrum and 2D 13C-1H HMBC spectrum of cYΔF in DMSO-d6. Fig. S4. 1D 1H NMR spectrum (with presaturation of residual water) and 1D 13C NMR DEPTQ spectrum of cΔYF in DMSO-d6. Fig. S5. 2D 13C-1H HSQC spectrum and 2D 13C-1H HMBC spectrum of cΔYF in DMSO-d6. Fig. S6. 1D 1H NMR spectrum and 1D 13C NMR DEPTQ spectrum of cWΔY in DMSO-d6. Fig. S7. 2D 13C-1H HSQC spectrum and 2D 13C-1H HMBC spectrum of cWΔY in DMSO-d6. Fig. S8. 1D 1H NMR spectrum and 1D 13C NMR DEPTQ spectrum of cWΔP in DMSO-d6. Fig. S9. 2D 13C-1H HSQC spectrum and 2D 13C-1H HMBC spectrum of cWΔP in DMSO-d6. Fig. S10. 1D 1H NMR spectrum and 1D 13C NMR DEPTQ spectrum of cWΔL in DMSO-d6. Fig. S11. 2D 13C-1H HSQC spectrum and 2D 13C-1H HMBC spectrum of cWΔL in DMSO-d6. Fig. S12. Genomic environment of the CDO genes and predicted biosynthetic pathways.

Rights and permissions

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

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Le Chevalier, F., Correia, I., Matheron, L. et al. In vivo characterization of the activities of novel cyclodipeptide oxidases: new tools for increasing chemical diversity of bioproduced 2,5-diketopiperazines in Escherichia coli. Microb Cell Fact 19, 178 (2020).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: