|Home | About | Journals | Submit | Contact Us | Français|
Microorganisms use chemical inactivation strategies to circumvent toxicity caused by many types of antibiotics, yet in all reported cases; this approach is limited to enzymatically facilitated mechanisms that each target narrow ranges of chemically related scaffolds. The fungal-derived shikimate analogues, pericoxide and pericosine A, were identified as chemoreactive natural products that attenuated the antagonistic effects of several synthetic and naturally derived antifungal agents. Experimental and computational studies suggested that pericoxide and pericosine A readily react via SN2′ mechanisms against a variety of nucleophilic substances under both in vitro aqueous and in situ coculture conditions. Many of the substitution products from this reaction were highly stable and exhibited diminished toxicities against environmental fungal isolates including the Tolypocladium sp. strain that produced pericoxide and pericosine A.
The fungal-derived shikimate analogues, pericoxide and pericosine A, have been identified as archetypal chemoreactive natural products that neutralize the antagonistic properties of chemically-diverse, synthetic and naturally-derived antifungal agents and toxins.
Antibiosis is a type of antagonistic chemical exchange that lies at the heart of many microbe-microbe interactions. The regularity with which the genes responsible for biosynthesizing antibiotics/toxins are encountered in genomes lends support to the idea that natural product ‘chemical warfare’ is a widespread and perhaps essential feature of microbial community structure and function. In many cases, the direct detection of antibiosis within native microbial communities remains a challenge; however, the biological consequences of these antagonistic interactions – antibiotic and toxin resistance mechanisms – are abundant among both antibiotic producing and non-producing bacteria and fungi. The prevalence of antibiotic resistance genes throughout natural microbial populations suggests that antibiosis has long been an evolutionary driver in the refinement and proliferation of antibiotics.
Several types of antibiotic resistance mechanisms are recognized, including (i) membrane impermeabilization, (ii) expulsion via efflux, (iii) antibiotic inactivation, and (iv) modification of cellular targets. Concerning the antibiotic inactivation strategies, all currently known systems share two key features: the processes are enzymatically driven and the enzymes target narrow ranges of structurally related antibiotics. In this report, we provide evidence for a chemically driven (non-enzymatic) toxin inactivation system employed by a soil ascomycete. This process was serendipitously discovered shortly after we initiated further chemical studies of the shikimate-PKS-NRPS metabolite, maximiscin (1), which our group identified from Tolypocladium sp. Salcha MEA-2 (T1). These results led us to determine that the shikimate portion of 1 is incorporated via a substitution reaction involving a chemoreactive precursor metabolite. This natural product is reactive toward a broad range of exogenous antibiotic/toxic chemicals. Herein we describe the unique balance of electrophilic promiscuity and chemical stability exhibited by the chemoreactive Tolypocladium metabolite and its analogue, as well detail how they likely serve to protect the fungus from antibiosis.
Before discussing these new discoveries, it is important to draw the reader’s attention to the fact that during the initial stages of our follow-up studies about the production of 1, we were confronted with data that were at odds with our original hypothesis of the metabolite’s absolute configuration (details of these studies are provided in the supporting information). The new results led us to realize that during the original VCD experiments, in which 1 was held for hours in DMSO with warming, the compound had rearranged into a new isomeric species, isomaximiscin (2) (Figure 1A). Despite this setback, the finding proved to be auspicious because it (i) provided an early clue regarding the remarkable process leading to the formation of 1, (ii) enabled us to couple the spectroscopically-derived absolute configuration results for 1 to data reported for synthetically prepared 3 (Figure 1A), and (iii) led to our development of new investigational tools (i.e., 13C isotopic labeling analysis and ECD measurements; refer to supporting information) that proved useful for probing the reactivity spectrum of the precursors of 1 with a range of antibiotics and toxins.
Continuing with our exploration of fungus T1, a notable attribute of its behavior was the consistent production of secondary metabolites in response to the presence of cocultured microbial species. Further studies examining additional fungal coculture scenarios confirmed the robustness of this response (Table S17). For example, UPLC-ESIMSn analysis of a coculture consisting of T1 and Penicillium sp. P1 provided evidence for a new compound that yielded ions at m/z 464.2301 ([M+H]+) and m/z 278.1763 ([M+H]+) (Figures 2E and 2F). Curiously, the difference between these two major ions (Δ m/z 186.0538) was identical to the neutral loss due to cleavage of the shikimate analogue moiety, which we had previously detected during the MSn analysis of 1 (Figures 2A and 2D). This alerted us to the possibility that the new coculture metabolite might also contain a shikimate analogue moiety. Scale-up preparation, purification, and structure determination revealed that the new compound was structurally related to 1 and was named pseudomaximiscin A (4) (Figure 2F). Similar to 1, incubation of 4 in DMSO-d6 led to its isomerization resulting in a ~1:1 equilibrium mixture containing its diastereomeric product, pseudomaximiscin B (5) (Figure 3).
Intrigued by the discovery of 4, we reexamined the MSn data from the fungal coculture experiments and noted that in addition to the recurring Δ m/z 186 for several new metabolites, a second neutral loss of Δ m/z 204 (Figure 2B) was apparent. Using these two parameters to filter the MSn data, a neutral loss event of Δ m/z 204 was identified for two metabolites that were generated when fungus T1 was cocultured with Penicillium P2 (Figure 2H). Scale-up production yielded the new compounds, mycophenolic acid 3-O-pericosine (6) and mycophenolic acid 16-O-pericosine (7). Whereas 6 was optically active ([α]20D -126), 7 was not, indicating that it was a racemic mixture. Deliberate probing of all samples by UPLC-ESIMSn and ion-selective MS revealed that the likely non-shikimate precursor metabolites of 4, 6, and 7 were coming from the coculture partners (Figure 3). For example, fungus P1, was determined to be the source of metabolite PF1140 (8), whereas fungus P2 made mycophenolic acid (9). Therefore, compounds 4, 6, and 7 were proposed to be chimeric metabolites made from the union of 8 or 9 with a yet undetermined chemoreactive compound from fungus T1.
To determine how the T1 culture facilitated this process, an experiment was conceived using P1-derived metabolite 8 as ‘bait’ in a chemoassay-guided process meant to uncover the origins of the shikimate analogue incorporation (Scheme 1). Initially, purified 8 was mixed with dialysate prepared from one-week-old T1 culture broth. LAESIMS monitoring of the reaction revealed that compound 4, which yielded a [M+H]+ quasi molecular ion peak at m/z 464.2301 was detectable with dialysate prepared both with large (1000 kDa) and small (0.5–1 kDa) average-molecular-weight cutoff membranes. This suggested that the reaction could occur in a cell-free environment by means of a non-enzymatic process. Next, T1 cultures were successively extracted with EtOAc and n-butanol. Whereas the remaining aqueous layer was inactive, samples from both organic layers were able to generate 4 upon the addition of 8. Subsequent chemoassay-directed HPLC fractionation led to the purification of two shikamate analogues, including the new epoxide metabolite, pericoxide (10), from the EtOAc extract, as well as the known chlorinated compound, (+)-pericosine A (11), from the n-butanol extract. These results led us to determine that 10 and 11 were the probable precursors to the non-enzymatic formation of 1 in T1 cultures, not the 6-OH analogue as we had previously implicated.
With 10 and 11 identified as candidates for the formation of the cocultured-derived hybrid metabolites, the chain of antecedence linking the two compounds was called into question. UPLC-ESIMSn analysis of the MeOH extract of the cell lysate of T1 revealed that neither 10, 11, nor any other hybrid metabolites (e.g. 1) were detectable intracellularly, implying that both 10 and 11 were either formed extracellularly or sequestered and secreted from the cells upon their formation. Further examining the 13C-labeled 10 and 11 (prepared by feeding fungus T1 [U-13C6]-D-glucose), it was determined that both compounds were present in the spent culture broth as single enantiomers exhibiting “type B” 13C-labeling pattern (Figure 1A and supporting information). Treatment of 13C-labeled 10 with NaCl in ddH2O yielded 11 seemingly via an SN2 mechanism (Figure 4A). ECD analysis showed that 11 prepared both from 10, as well as directly from the fungal culture broth bore the same absolute configuration (Figure S126). These results implied that 11 was non-enzymatically produced from 10 within the T1 culture. This hypothesis was tested by preparing fungal culture broth for T1 using Millipore water and observing that the yield of 11 was strongly correlated with the quantity of NaCl or other Cl− sources that were added to the culture medium (Figure S124). Though the origin of 10 remains undetermined, a hint of the process involved in its formation can likely be gleaned from the biosynthesis of the chorismic acid derivative, cyathiformine A, (Figure S130).
The roles that 10 and 11 might play in the production of 1 were tested by assessing their reactivities toward pyridoxatin (12). The production of 1 was found to occur in Millipore water upon addition of 12 to both 10 and 11. Manipulation of selected reaction conditions (i.e., solvent, temperature, and catalyst) confirmed that epoxide 10 was generally more reactive toward 12 than its halohydrin counterpart 11. In all cases, enantiomerically pure 1 was obtained as the product indicating that a selective SN2′ mechanism was involved in the formation of 1 under both synthetic, as well as in situ culture conditions (Figure 4B).
To test the promiscuous reactivity of 10 and 11 toward other compounds, T1 cultures were treated with a panel of substrates that included chemically diverse functional groups: hydroxamic acids, phenols, carboxylic acids, alcohols, alkenes, amides, and amines (Table S14). Candidate products from each reaction were purified and their structures confirmed by HRESIMS and multidimensional NMR. In addition, T1 cultures were supplied with [U-13C6]-D-glucose so that the labeling patterns of the resulting products (Figure 3) would afford insights regarding the probable substitution mechanisms involved in their formation (i.e., SN1, SN2, or SN2′) (Figure S125). Additionally, the absolute configuration of the C-6′ position in each product was determined by comparing its experimental and theoretical ECD data (Figure S126). Based on these analyses, the stereoselectivities of the coupling reactions (particularly anti- vs syn-SN2′ mechanisms) were systematically assessed (Figure S125; details of the structure determination for the new compounds are provided in the supporting information).
The results demonstrated that 10 and 11 were decidedly reactive toward diverse chemical targets. Stereoselective SN2′ (anti-SN2′ for 10 and syn-SN2′ for 11) reaction processes were observed involving all the hydroxamic acids, including PF1140 (8), ciclopirox (13), and SAHA (15), to give optically active products 4, 14, and 16, respectively. A similar SN2′ reactivity pattern was observed involving the reaction of 10 and 11 with phenol-containing [3-OH of mycophenolic acid (9)] and secondary-amine-containing [anisomycin (17)] substrates. In contrast, reaction with the carboxylic acid moiety of 9 yielded racemic 7 (Figure S26). The 5′R*6′R* relative configuration of the product was assigned based on an examination of its 13C NMR chemical shifts in comparison with DFT calculated data (Figures S12–S14, details see SI). Whether product mixture 7 arose from competing reaction processes or a rearrangement remains unknown.
The primary amine tryptamine (19) was also administered to T1 resulting in the formation of products 20 and 21 and an unexpected novel product named mallimiscin (22) (Figure 3). Distinct JH-5′,H-6′ values aided in determining the 5′,6′-relative configuration (trans >9 Hz, cis ~5 Hz) of the products. Whereas 20 was obtained as an enantiomerically pure product, the diastereomers (21) were a racemate (Figure S27). Compound 22 was seemingly formed via a Maillard reaction of 20 with D-glucose, based on its 13C-labeling pattern, ROESY correlations, JH,H coupling constants, and ECD calculation (Figures 3, S126, and S127, see SI for details).
To better understand the selectivity exhibited by 10 and 11 for several of the substrates, DFT calculations were employed to determine the energies of the transition states for syn-SN2′, anti-SN2′ and SN2 reactions of 10 and 11 with the model hydroxamic acid 23. The β-hydroxycarbonyl hydroxy group in 23 was estimated to have a pKa of ca. 7–9 indicating that its anionic form may be present as the reactive nucleophile in aqueous media. The computational results (Figure 5) were in quantitative agreement with the experimentally observed regio- and stereoselectivities of the reactions of hydroxamic acids with both 10 and 11. For the reaction of 23 with 11, the syn-SN2′ transition state (and its corresponding activation energy) was energetically the lowest by a substantial margin. In contrast, the anti-SN2′ pathway was favored in the reaction involving epoxide 10 by a smaller margin, dependent on the computational method. Notably, each of the transition states identified in these calculations suggest that tautomerization occurs between the C-2 carbonyl and N-OH groups.
The observed (and calculated) exclusively syn-SN2′ (for allyl-X) or anti-SN2′ (for vinyl oxirane) selectivity detected among some of these reactions has been found in other vinylogous nucleophilic substitution reaction systems, but there are also many exceptions, depending on the substrate, nucleophile and solvent. Numerous factors have been invoked to explain the observed selectivity in such reactions, including frontier orbital and coulombic interactions, conformational and steric effects, specific Lewis acid-base and H-bonding interactions, and solvation. The origin of the remarkable and divergent stereoselectivities in the reactions of these two allylic substrates is presently unclear and its elucidation will require additional experimental and computational investigation.
Reflecting on the potential biological roles of electrophilic 10 and 11, we noted that the natural product precursors made by T1’s partner fungus had possessed antifungal activities. This prompted us to ask the question what effect the addition of the shikimate analogue moiety had on the bioactivities of compounds 8, 9, 12, and the other nucleophilic substances that we tested. A panel of fungi including four Tolypocladium spp. (T1–T4), four Penicillium spp. (P1–P4) and Aspergillus niger (A1) were selected for assessment. The test revealed a distinct trend in which the antifungal activities of the substrates were greatly diminished or abolished following the addition of the shikimate analogue moiety (Figure 6A). For example, both the P1-derived antifungal 8 and the synthetic antifungal 13 showed growth inhibition against all of the fungal strains with MIC values in the range of 1–50 μM, whereas their adducts, 4 and 14, exhibited an average >5-fold decrease in potencies. To test the capacity of 11 to block the toxicity of 13 in real time, fungal cultures were preincubated with 11 and then treated with varying doses of 13. This regimen of preadministering 11 afforded up to an 8-fold decrease in the MIC of 13 (Figure 6B). Our prior time-course studies examining the production of 10 and 11 showed that these metabolites rapidly accumulated in T1 cultures after 96 hours. We had also noted that the addition of NaNO3 to the culture medium thoroughly abolished the formation of 10 and 11. Using this information, we determined that fungus T1 was equally sensitive to the antifungal activities of 8 or 13 in both freshly prepared normal and NaNO3-supplemented media during a 3-day test window. However, when 5-day-old T1 culture broth was used to prepare the test medium, the non-NaNO3-supplemented T1 culture became increasingly resistant to 8 and 13 (Figure 6C).
In summary, these studies provide evidence for a new chemically facilitated mode of toxin resistance exhibited by a soil ascomycete. Whereas previously reported resistance mechanisms involving antibiotic modification depended on the enzymatic modification of target substrates, fungal metabolites 10 and 11 functioned as electrophilic warheads that were reactive to a wide variety of natural and synthetic organic substances. The actions of these metabolites appear to limit the deleterious effects of antibiotics/toxins against their microbial targets. Further studies investigating the steric and stereoelectronic regulation of this process are expected to provide inspiration for the creation of new biologically compatible toxin scavenging materials.
**We appreciate the thought-provoking comments offered by Drs. Jason Clement and Will Gutekunst concerning reaction mechanisms. We also thank Prof. Dr. Peng Liu (U. Pittsburgh) for helpful discussions regarding the computational modeling. This work was funded in part by grants from the National Institute of General Medical Sciences (R01GM107490) and National Institute of Allergy and Infectious Diseases (R01AI085161) of the National Institutes of Health (R.H.C.).
Dr. Lin Du, Department of Chemistry and Biochemistry, Natural Products Discovery Group, and Institute for Natural Products Applications and Research Technologies, University of Oklahoma, Norman, OK 73019-5251 (USA)
Dr. Jianlan You, Department of Chemistry and Biochemistry, Natural Products Discovery Group, and Institute for Natural Products Applications and Research Technologies, University of Oklahoma, Norman, OK 73019-5251 (USA)
Prof. Dr. Kenneth M. Nicholas, Department of Chemistry and Biochemistry, Natural Products Discovery Group, and Institute for Natural Products Applications and Research Technologies, University of Oklahoma, Norman, OK 73019-5251 (USA)
Prof. Dr. Robert H. Cichewicz, Department of Chemistry and Biochemistry, Natural Products Discovery Group, and Institute for Natural Products Applications and Research Technologies, University of Oklahoma, Norman, OK 73019-5251 (USA)