Abstract

Introduction

Tambjamines are a family of yellow-pigmented alkaloid natural products that possess a number of useful properties including antimicrobial, antifungal, antimalarial, and anti-tumour activities.1–3 Compounds in this family contain a methoxy-bipyrrole core with a variable imine moiety, consisting of decarboxylated amino acid side chains or linear alkyl chains (general structure in Fig. 1).4–8 Numerous tambjamines have been isolated from marine and terrestrial organisms and many have also been produced synthetically.1,4,6–11 Identifying new tambjamine natural products is appealing due to their exceptional bioactivity profile, much of which arises from the bipyrrole core, a structural feature shared with the well-studied prodiginine family.12–14 This rigid, aromatic bipyrrole structure allows for DNA intercalation and the formation of ion coordination complexes.12,15,16

Open in a separate window

Fig. 1

General tambjamine and prodiginine structures.

The ability to coordinate ions depends heavily on the orientation of the compound; in prodiginine compounds, rotation around the methenyl bond produces an α (E) and a β (Z) rotamer.17,18 This bond restriction also exists for tambjamines, thus similar rotameric forms are possible (Fig. 2). The β rotamer is believed to be the more bioactive and pharmaceutically relevant conformation for both molecule classes. Its ability to coordinate ions is thought to be the source of multiple bioactivities, including chloride transmembrane transport to induce apoptosis through cytosol acidification and copper dependent oxidative cleavage of DNA.12,16,18,19

Open in a separate window

Fig. 2

General tambjamine structure in the α and β rotamer conformations.

Stabilizing the β rotamer by introducing bulky imine moieties to restrict bond rotation is a tactic used to enhance the bioactivity of these compounds; this strategy has been effectively employed to increase antimalarial potency of synthetic tambjamine analogues.2 The lipophilicity of the imine sidechain also influences the anion affinity and transmembrane transport of tambjamines.20 Manipulating these structural characteristics of tambjamines can easily be done synthetically to interrogate structure–activity relationships, but it is also valuable to consider the naturally occurring derivatives that have come about through evolution.

Most naturally derived tambjamines have been isolated from marine invertebrates such as bryozoans and nudibranchs, however it is postulated that these compounds are actually produced by microbial symbionts.4–7 Thus far, only two tambjamine analogues have been identified from bacteria; BE-18591 from the terrestrial Streptomyces sp. BA 18591, and tambjamine YP1 from marine bacterium Pseudoalteromonas tunicata.8,9 In 2007, Burke et al. linked tambjamine YP1 to its biosynthetic gene cluster in P. tunicata (Fig. 3), and assigned putative functions to all biosynthetic enzymes based on similarity to prodiginine biosynthetic genes and domain homology to known proteins (functional annotation can be found in Table S1^†).21 This annotation allowed us to identify a very similar cluster in Pseudoalteromonas citrea using the antiSMASH cluster comparison tool.22P. citrea displays characteristics indicative of tambjamine production, such as a lemon-yellow colouring and anti-fouling properties, making it a likely producer.23,24

Open in a separate window

Fig. 3

The tam cluster in P. tunicata and the product tambjamine YP1.9,21

Interestingly, the tam cluster in both P. tunicata and P. citrea contains a predicted oxidase enzyme, TamC.21,22 This enzyme displays a high degree of similarity (34% identity) to the Rieske oxygenase, RedG, responsible for the cyclization of streptorubin B in Streptomyces coelicolor A3(2).25,26 RedG catalyzes a carbon–carbon bond formation between the pyrrole ring C of undecylprodigiosin and an internal CH₂ unit in the alkyl chain.25,27 The presence of TamC in the tam cluster suggests the possible construction of a cyclized product. In the initial identification of tambjamine YP1 from P. tunicata, a potential cyclized variant was briefly mentioned by Burke et al., but to date no structural data has been reported.9,21

In this work, we have extracted and purified the predicted cyclized tambjamine YP1, herein called tambjamine MYP1, from P. citrea, and elucidated the macrocyclic structure using NMR spectroscopy and X-ray crystallography. We also compare the apparent pK_a of our cyclic compound with a linear counterpart to investigate the implications this unique structural feature may have on ion coordination due to its rotational restraint and cage-like structure.

Results and discussion

Optimization studies on the growth and extraction of P. citrea determined that the bacteria displayed the highest intensity of yellow-pigmentation when grown on a solid marine agar rather than in a liquid culture. The isolation of the putative tambjamine was successfully completed from 3.2 L of solid media cultures of P. citrea extracted using ethyl acetate (EtOAc). The concentrated crude extract was purified by flash Si-chromatography and subsequent RP-HPLC using a C8 column, to obtain a yellow solid with a final yield of 0. 03% purified tambjamine from the dry cell mass.

High-resolution mass spectrometry using FTMS-ESI in positive ion mode generated an [M + H]⁺ ion at m/z = 354.25532 corresponding to a molecular formula of C₂₂H₃₂ON₃ (calc 354.25399, Δ = 3.74 ppm). The reported [M + H]⁺ of tambjamine YP1 is m/z 356.27,9 the difference of approximately 2 mass units suggesting that the extracted tambjamine has one additional degree of unsaturation compared to tambjamine YP1. This mass difference is a good indication that a cyclic tambjamine compound is produced by P. citrea.

To determine the structure of the new tambjamine and particularly the points of cyclization, we performed a number of 1-D and 2-D NMR experiments in chloroform-d (CDCl₃). Based on the NMR data we have proposed the structure shown in Fig. 4 and all ¹H and ¹³C chemical shifts measured are displayed in Table 1. When we compare shifts for our new molecule to corresponding literature values for the linear tambjamine YP1 variant we can confirm the two molecules have very similar structures.9 However, there are key differences; no proton signal was detected at the 5′ pyrrole position of the putative cyclic tambjamine marking this as one point of cyclization. The proton and carbon atoms at position 19 have a large downfield shift from the linear tambjamine YP1 chemical shift values, and the ¹H signal only integrates for two protons. These pieces of evidence give the second connection point for cyclization. The remaining spectral data remain consistent with the values for the linear compound, except the ¹³C chemical shifts for quaternary carbons 5′ and 2′ could not be identified with certainty due to the small quantity of purified material.

Open in a separate window

Fig. 4

Structure of tambjamine MYP1 and selected 2D NMR correlations.

To confirm the NMR structure and cyclization points and to visualize the overall conformation of the cyclic tambjamine YP1 compound, we carried out X-ray crystallography. The purified compound readily co-crystallized in a complex with a formate anion and a formic acid molecule, stabilized through hydrogen-bonding interactions (Fig. S6^†). For clarity, the formic acid molecule is omitted from the thermal ellipsoid drawing in Fig. 5. The full asymmetric unit, all refinement parameters, and additional view angles can be found in Fig. S7 and Table S2.^†

Open in a separate window

Fig. 5

X-ray structure of tambjamine MYP1. Thermal ellipsoids are drawn to 30% probability (crystal structure was deposited in the Cambridge Structural Database (CSD), deposition number 1892787).

The solved crystal structure agrees with our NMR-based structure. The interatomic distance between C5′ and C19 is 1.502 ± 0.004 Å; the approximate distance of a C–C bond, validating this point of macrocyclization. Additionally, the crystal structure confirms the stereochemistry of the cis-alkene at C10–C11 with a torsional angle between C9–C10–C11–C12 of 1.3 ± 0.4°.

It is interesting to note that the linear compound, tambjamine YP1, originally isolated from P. tunicata was not detected from P. citrea in this work. This is surprising as it appears both P. citrea and P. tunicata produce the same cyclic tambjamine.9,21

This discrepancy in production leads to questions about the potential differences in biological and chemical properties of the analogues produced by very similar gene clusters. One such difference is the rotameric conformation and by extension ion binding properties of the linear and cyclic compounds.

Due to chemical and bioactivity similarities between the tambjamine and prodiginine families of molecules, we turned to the abundance of literature characterizing prodiginine structural features for guidance. In the prodiginine family it is known that the α rotamer is formed at low pH, due to favourable hydrogen bonding between the pyrrolium ion and the methoxy resulting in a high pK_a of 8.23 ± 0.03 for one family member.17,28,29 Whereas, the β rotamer of the same molecule has a lower pK_a of 5.4 ± 0.2 resulting from hydrogen bonding between pyrrole rings and is stabilized at a high pH.17,28,29 The β rotamer can also be formed at low pH when an anion such as chloride is present, raising the pK_a to facilitate ionic interactions.12,28 Because the two rotamers have quite different pK_a values, a molecule's apparent pK_a reflects the contributions of each rotamer to the overall population.28 The propensity to form each rotamer can be estimated with the apparent pK_a of the molecule, with higher pK_as suggesting a preference for α rotamer and lower pK_as suggesting a preference for the β rotamer.28 One caveat that should be noted is binding with anions such as chloride ions may cause shifts between the rotamer populations.28 As stated in the introduction, the β rotamer of prodiginines appears to be more biologically active and so variants that prefer taking this conformation are more likely to display strong bioactivity profiles.12,16,18,19

Far less is known about the tambjamine system, however, it is also likely able to form both the α and β rotamers. Experimental evidence of the β rotamer has been reported and although no pK_a values have been determined for a specific rotamer conformation,8,9 an apparent pK_a of 10.06 has been measured for tambjamine E.15 By analogy to the prodiginine system we would expect tambjamines that preferentially form the β rotamer to possess greater bioactivities. To investigate the potential biological impact of the macrocyclization present in tambjamine MYP1, we compared the apparent pK_a of tambjamine MYP1 with the linear and unsaturated derivative, tambjamine BE-18591.

Tambjamine BE-18591 was synthesized from 4-methoxy-2,2′-carbaldehyde and dodecylamine following previously described protocols.30 The protonation state of each compound was monitored using their absorption spectra in 1:1 (v/v) MeCN/H₂O and 0.1 M NaCl following the same approach as Melvin et al.18 The absorbance maxima of each compound changes depending on its protonation, allowing the apparent pK_a to be measured by the absorbance shift as the pH increases. BE-18591 shifts from 400 nm in low pH to 368 nm as the pH increases, whereas tambjamine MYP1 shifts from a maximum at 420 nm to 385 nm as the pH increases (spectra can be found in Fig. S8^†). As we do not know the population of each rotamer contributing to the overall pK_a, the value measured is the apparent pK_a of each compound.

The pK_a of tambjamine MYP1 was calculated to be 10.4 ± 0.3, in comparison to a pK_a of 9.9 ± 0.2 for the linear tambjamine BE-18591 (calculations in eqn S1^†). These values are similar to those measured for tambjamine E, as described above. In our experiments, there appears to be a slight increase in pK_a for the cyclic molecule, however the difference is small. Perhaps both molecules have very similar populations of the two rotamers and so cyclization may not cause a large increase in the β conformation population. Additionally, the macrocycle might form a cage-like structure around the chloride ion, similar to what we see in the crystal structure around formate (Fig. 5) restricting anion disassociation as the pH increases and favouring imine protonation. Due to the large size of the macrocycle the strain associated with cyclization may not be pronounced, thus it would be interesting to investigate the properties of synthetic tambjamines with smaller macrocycles. If the existing macrocycle is large enough to rotate to the α form it may be able to assume a helical conformation, which could have intriguing structural and biological implications. Further work will be needed to probe the bioactivity of macrocyclic tambjamine structures and their limitations.

The macrocyclization at the 5′ position bears resemblance to cyclizations seen in prodiginine analogues including streptorubin B, metacycloprodigiosin, and the marineosins.14,25,27 The cyclization of these compounds are all catalyzed by Rieske non-heme iron-dependent oxygenases, which use radical chemistry to form C–C bonds.25,27 In all of the characterized examples of prodiginine cyclization, the radical is formed on an internal CH₂ of an alkyl chain, this instigates bond formation with either a pyrrole carbon or the methenyl carbon.25,27 In our cyclized tambjamine the C–C bond is initiated at the terminal methyl of the alkyl chain, forming a much less stable radical. This means that the putative oxidase, TamC, catalyzes a very difficult proton abstraction from the alkyl chain to produce the radical. Similar cyclization connectivity is also present in the compound cyclononylprodiginine isolated from Actinomadura madurae,31 however the biosynthesis of this compound is yet to be studied. To our knowledge, there are no known Rieske oxygenases that catalyze a comparable reaction, thus TamC presents an intriguing target for future studies in this family of enzymes.

Conclusion

We have described the isolation, crystallization, and identification of tambjamine MYP1 from Pseudoalteromonas citrea; this is the first macrocyclic tambjamine compound identified from nature. To probe if the macrocyclic structure favours formation of the putatively more bioactive β rotamer we determined the apparent pK_a of the cyclic tambjamine and compared it that of a linear analogue. The small difference between the two pK_a values suggests the rotamer populations of the cyclic and acyclic molecules are very similar and that this quite large macrocycle does not drastically change the conformational preference. Future investigations could focus on properties of tambjamines with smaller, more strained macrocycles and characterization of the proposed cyclase enzyme (TamC) could shed light on the functional capacity of the Rieske oxygenase enzyme family.

Experimental

Conflicts of interest

There are no conflicts to declare.

Supplementary Material

Acknowledgments

Footnotes

References