Plant pathogenic anaerobic bacteria use aromatic polyketides to access aerobic territory

See allHide authors and affiliations

Science  06 Nov 2015:
Vol. 350, Issue 6261, pp. 670-674
DOI: 10.1126/science.aac9990

Protecting against too much of a good thing

The slimy pink rot of potatoes is caused by the bacterium Clostridium puniceum, which cannot grow in the presence of oxygen. These bacteria produce a polyphenolic metabolite known as clostrubin that functions as an antibiotic. Shabuer et al. now show that the bacteria also use clostrubin to protect themselves from the aerobic environment of the potato tuber.

Science, this issue p. 670


Around 25% of vegetable food is lost worldwide because of infectious plant diseases, including microbe-induced decay of harvested crops. In wet seasons and under humid storage conditions, potato tubers are readily infected and decomposed by anaerobic bacteria (Clostridium puniceum). We found that these anaerobic plant pathogens harbor a gene locus (type II polyketide synthase) to produce unusual polyketide metabolites (clostrubins) with dual functions. The clostrubins, which act as antibiotics against other microbial plant pathogens, enable the anaerobic bacteria to survive an oxygen-rich plant environment.

The global food supply is weakened by infectious plant diseases and microbe-mediated decay during crop storage and transportation (1). Infections of potato (Solanum tuberosum), one of the top four food crops cultivated worldwide, result in losses amounting to 65 billion kg of food and $16 billion annually, of which 30 to 50% are caused by pectolytic bacteria (2, 3). In wet seasons and humid storage conditions, potato tubers are infected and decomposed by Clostridium puniceum that grow in the absence of oxygen (4). The potato tuber is, however, an oxygen-rich environment.

Of the many known plant pathogenic bacteria, C. puniceum is the only characterized representative of the diverse genus Clostridium (5). These obligate anaerobes are typically killed by normal atmospheric oxygen concentrations, yet they are frequently isolated from oxic plant matter (6). It has been argued that mixed bacterial infections could account for an anaerobic microenvironment in diseased tubers (7, 8). However, pure cultures of C. puniceum also cause potato slimy rot, manifested by formation of pink pigments by the bacterium (4).

To study the pathogenesis of this anaerobe, we reproduced potato slimy rot by stab-inoculating C. puniceum into potato tubers. Over 4 days, potato decomposition accompanied accumulation of pink pigment and slime (Fig. 1A). High-performance liquid chromatography (HPLC) analysis of the ethyl acetate extract showed two main peaks with ultraviolet (UV)/visible maxima at 530 nm (Fig. 1B). The HPLC profile resembled that derived from the soil anaerobe C. beijerinckii (9). By HPLC-HRMS (high-resolution mass spectroscopy) correlation using clostrubin A (1) as an authentic reference, we identified one of the main products of C. puniceum as 1, an aromatic polyketide (9). Obtaining the second main product, clostrubin B (2), as a pure compound was an arduous task because of its strong adhesive properties, low solubility, and propensity to appear in various mesomeric forms. We isolated pure 2 (4.0 mg) from 8.0 liters of cultured C. beijerinckii. Analyses including HRMS, UV spectroscopy, and nuclear magnetic resonance (NMR) spectroscopy (figs. S1 to S3, including 13C labeling experiments) revealed that 1 and 2 share the same aromatic polyketide backbone (Fig. 1C). However, extensive one- and two-dimensional NMR studies revealed that 2 features a sugar-like linear side chain connected to position C-2 (supplementary text and table S1).

Fig. 1 Structure, genetic basis, and antibiotic activity of pigments (clostrubins) formed during C. puniceum infection.

(A) Typical appearance of potato tubers stab-inoculated with pectinolytic clostridia. Scale bar, 1 cm. (B) HPLC profile of extract of diseased tubers showing clostrubin A (1) and clostrubin B (2) as main metabolites. (C) Structures of 1 and 2. (D) Minimum inhibitory concentrations (MICs) of 1, 2, and Cip (ciprofloxacin) against selected potato pathogenic bacteria (black, B. pumilus; gray, C. michiganensis; white, S. scabies). (E) Organization of the clostrubin biosynthesis gene clusters in C. puniceum (Cp) and C. beijerinckii (Cb). (F) Phylogeny of the principal polyketide synthase components (maximum likelihood algorithm). ClrA(p) and ClrB(p), ClrA and ClrB from C. puniceum; ClrA(b) and ClrB(b), ClrA and ClrB from C. beijerinckii. (G) Genetic evidence for successful gene disruption; WT, wild type. (H) Photograph showing appearance of wild-type (left) and ΔclrA mutant (right) cultures.

To investigate how clostrubins A and B affect plant disease, we tested for phytotoxic activity. Applied clostrubins had no effect and thus did not act as virulence factors. Because clostrubin A (1) showed strong antibacterial activities against the human pathogens MRSA (methicillin-resistant Staphylococcus aureus) and VRE (vancomycin-resistant enterococci) (9), we reasoned that the clostrubins might be produced by the plant pathogen to battle microbial competitors. We therefore tested the effect of clostrubins on some of the more common microbial pathogens of potato: soft rot (Bacillus pumilus), ring rot (Clavibacter michiganensis subsp. sepedonicus), and common scab (Streptomyces scabies) (10, 11). Both C. puniceum metabolites showed antibacterial activity against all tested potato pathogens, with minimal inhibition concentration (MIC) values in the nanomolar range (Fig. 1D and table S2). Thus, clostrubins are effective at removing competitors in a resource-limited niche (12). The microbial competitors tested were aerobic plant pathogens, whereas C. puniceum was described as a strictly anaerobic bacterium (4). We found, however, that C. puniceum is capable of growing in tubers in an aerobic environment. To unveil the genetic inventory for tolerance of reactive oxygen species (ROS) by the plant pathogen, we sequenced the genome of C. puniceum using Illumina MiSeq technology. By bioinformatic genome analysis, we found orthologs of typical superoxide dismutase, catalase, peroxidase, rubredoxin, and rubrerythrin genes (table S4), which have been described for various other anaerobes (1317).

C. puniceum produced stronger pigmentation when grown in the presence of oxygen. To study this relationship, we needed a clostrubin-negative mutant of C. puniceum. On the basis of the clostrubin structure, we reasoned that the most likely biosynthetic route would be via a type II polyketide synthase (PKS) (18). However, type II PKSs are characteristic of actinomycetes; only two compounds derived from type II PKSs have been identified in bacteria outside the actinomycetes (19, 20). Moreover, we were unable to find any type II PKS genes in more than 210 sequenced anaerobe genomes (21). In contrast, BLAST (22) and antiSMASH (23) analyses revealed that the C. puniceum genome harbors a gene locus (clr) coding for a putative type II PKS, including the minimal PKS [clrA, KSα (ketosynthase); clrB, CLF (chain length factor); clrC, ACP (acyl carrier protein)], cyclases, and modifying enzymes (Fig. 1E and table S3). The predicted PKS shows the highest homology to the resistomycin (rem) PKS, which produces a polycyclic ring system in the aerobe Streptomyces resistomycificus (24). For comparison, we also sequenced and analyzed the clostrubin biosynthesis gene clusters in two pigmented C. beijerinckii strains and found a nearly identical gene locus architecture (Fig. 1E). Phylogenetic analysis using the maximum likelihood algorithm showed that the ClrA and ClrB variants cluster most closely with KSα and CLF proteins, yet form a distinct clade (Fig. 1F and fig. S12).

To confirm that the clr locus is responsible for the production of clostrubins, we generated knockout mutants deficient in an essential PKS gene. We constructed two individual clrA-inactivated mutants of C. puniceum and a ΔclrA mutant of C. beijerinckii using the Clostron gene inactivation system (25), which uses a retro-homing bacterial group II intron to insert within a specific DNA sequence in the genome. Introduction of the knockout plasmid by conjugation from an Escherichia coli donor strain resulted in positive colonies on antibiotic-selective agar plates. Insertion of the introns was confirmed by polymerase chain reaction (Fig. 1G and figs. S6 and S9). The ΔclrA mutants lost pigmentation and were also incapable of producing clostrubins (Fig. 1H and figs. S5 and S7). Thus, the aromatic polyketide clostrubin is indeed the product of a type II PKS.

We compared the physiology and pathogenicity of the clostrubin-negative C. puniceum ΔclrA mutant to that of the wild type. We infected potato slices under anaerobic conditions with the same number of wild-type or ΔclrA mutant bacteria (2 × 108), then measured the amount of degraded plant matter. After 4 days, wild-type and mutant bacteria caused similar amounts of degradation (74% reduction in the weight of rinsed slices) (Fig. 2A, top). However, the situation changed under aerobic conditions (Fig. 2A, bottom). Growth of the ΔclrA mutant was radically impaired in the presence of air. In contrast, the wild-type strain grew well and produced pink pigments. Quantification of the pectinolytic activity in four independent experiments revealed that the ΔclrA mutant caused little tissue rot (less than 5% degradation) (Fig. 2B). Addition of clostrubins 1 and 2 (1 μl, 1 mg ml−1 in dimethyl sulfoxide) to tuber slices inoculated with the ΔclrA mutant restored the wild-type phenotype (Fig. 2, A and B).

Fig. 2 Impact of clostrubins on slimy rot disease.

Potato tuber infection assay was performed by measuring the degradation level of potato slices under anaerobic or aerobic conditions. (A) Potato slices were inoculated with 2 × 108 cells of wild-type, mutant, or mutant C. puniceum supplemented with solutions of 1 or 2. (B) Summary and statistical evaluation of experiments. Under anaerobic conditions, bars were not significantly different (P > 0.05). Under aerobic conditions, the absence of clostrubin reduced potato degradation to <5%. *P < 0.001 [Bonferroni’s means comparison test, two-way analysis of variance (ANOVA)]. Data (means ± SD) are representative of four independent experiments; each experiment was performed in six biological replicates with three tuber slices each.

Because the absence of clostrubins had no effect on the action of the plant pathogen under anaerobic conditions, we concluded that clostrubins are not needed for plant infection and tissue degradation. However, our findings imply that the clostrubins permit survival of the plant pathogen in an oxygen-rich atmosphere. To test this hypothesis, we investigated the fate of the C. puniceum strains under anaerobic and aerobic conditions. In liquid culture, growth curves of the wild type and the clostrubin-negative mutant were identical under anaerobic conditions but different in oxygenated (pO2 40%) liquid medium. Whereas the ΔclrA mutant cells died immediately, the clostrubin producer survived and only showed a delayed log phase relative to anaerobic conditions (Fig. 3A). The experiment was repeated at a smaller scale for chemical complementation of the mutant. The mutant culture supplemented with clostrubin grew even faster than the wild type, likely because clostrubin was already present at the point of inoculation (fig. S11). The same phenotypes were observed for the wild type and ΔclrA mutant of C. beijerinckii (fig. S8). These findings show that oxygen tolerance is conferred by the presence of clostrubin.

Fig. 3 Clostrubin mediates survival of C. puniceum cells in an oxygen-rich atmosphere.

(A) Growth curves of C. puniceum wild-type and ΔclrA strains in P2 media (fermenter) under anaerobic and aerobic (pO2 40%) conditions. Data are means ± SD. (B) Summarized results from live/dead flow cytometric assays [BacLight stain (ThermoFisher), FACS plots, and gating strategy for live/dead cell stain] of C. puniceum strains after 4 days of incubation on potato slices under aerobic conditions. Potato slices were inoculated with 2 × 108 cells of wild type, mutant, or chemically complemented mutant (solutions of 1 or 2). Five independent FACS experiments consisted of five biological replicates. *P < 0.001 (Bonferroni’s means comparison test, one-way ANOVA). Data are means ± SD. (C) Confocal fluorescence microscope image of dead ΔclrA cells (red, live/dead BacLight stain) recovered after 4 days of incubation on potato slices under aerobic conditions. (D to G) SEM images of C. puniceum cells from inoculated tubers after 2 days of incubation under aerobic conditions. (D) Wild type. (E) ΔclrA mutant; arrowhead highlights sporulating cell. (F) and (G) Chemically complemented mutant (with 1 and 2, respectively). Scale bars, 2 μm.

To investigate the impact of the polyphenols in the context of infection, we quantified live and dead cells in the potato infection assay using the plant pathogen. We recovered the bacterial lawns of the wild type, mutant, and chemically complemented mutant of C. puniceum from the infected potato slices and subjected them to live/dead staining followed by fluorescence-activated cell sorter (FACS) analysis (Fig. 3B and fig. S10). In the absence of oxygen, similar ratios of living cells were detected in wild-type (97% live) and mutant (90.4% live) cultures. In the presence of oxygen, the situation was markedly different. Whereas 57% of the wild-type cells survived, no living cells of the ΔclrA mutant could be detected. Indeed, it was not even possible to recover the appropriate number of mutant cells required for FACS analysis (1 × 106 cells ml−1). Live/dead staining followed by fluorescence imaging revealed that all ΔclrA mutant cells were dead (Fig. 3C). However, chemical complementation of the mutant with clostrubins restored the viability (44 to 59% living cells). The impact of clostrubin on survival was evident within 2 days of inoculation under aerobic conditions. Scanning electron microscopy (SEM) showed differences between the wild type and the ΔclrA mutant cells, most of which were distinct from vegetative cells (Fig. 3, D to G). Again, chemically complemented mutants were similar to the wild type.

It would be conceivable that clostrubins interfere with the plant wound response, which involves the generation of hydrogen peroxide. Under anoxic conditions, H2O2 would not be generated, thus explaining why clostrubins are expendable under this condition. However, this scenario is unlikely because clostrubin is essential for bacterial survival in synthetic liquid medium (Fig. 3A) and on autoclaved potato infusion agar (Fig. 4A). We also tested whether hydrogen peroxide affects the growth of the wild type and/or mutant, but no significant difference in growth between the wild type and mutant was observed (Fig. 4B). Thus, the effect of clostrubin cannot be attributed to inactivating ROS formed in the context of plant defense mechanisms. Furthermore, an established assay [2,2-diphenyl-1-picrylhydrazyl (DPPH)] for testing radical scavengers was negative for clostrubin. Other standardized ROS assays were infeasible for clostrubin because of the interference of the absorption spectrum with the colorimetric assay methods. As an alternative, we performed functional profiling using a range of antioxidants that cover diverse ROS-scavenging activities (Fig. 4C). All compounds were tested for their ability to complement the effect of clostrubin. We supplemented antioxidants to potato tubers inoculated with the ΔclrA mutant; cell growth was determined by the bacterial pectolytic activity. We found that vitamin C and cysteine complementation were most similar to the effect of clostrubin. In contrast, uric acid and tocopherol proved to be less potent. At present, it is unclear exactly how clostrubin functions in its antioxidant capacity, and the next challenge for this work is to solve the mechanism of clostrubin action.

Fig. 4 Evaluation of the antioxidative effect of clostrubin.

(A) Clostrubin production of C. puniceum wild-type strain on potato infusion agar after 4 days of incubation under aerobic conditions. (B) Survival of C. puniceum wild-type and ΔclrA mutant strains after exposure to H2O2, as assessed by optical density at 600 nm. Exponentially growing cells were incubated with different concentrations of H2O2 for 2 days; experiments were performed in three biological replicates. Data are means ± SD. Bars were not significantly different (P > 0.05). (C) Comparison of clostrubins with typical ROS-inactivating compounds. Potato tuber infection assay was performed by measuring degradation level of potato slices under aerobic conditions. Inoculation of potato slices was assessed with 2 × 108 cells of wild type, mutant, or mutant supplemented with 1 mM vitamin C, cysteine, riboflavin, uric acid, or tocopherol. Experiments were performed in six biological replicates with three tuber slices each; data are means ± SD. *P < 0.001, Bonferroni’s means comparison test (one-way ANOVA). The photos show representative degraded potato tubers and the negative control (upper left).

During the past decade, analyses of the enzymatic repertoire that confers oxygen tolerance upon obligate anaerobes have implicated enzymes that scavenge ROS (26). The comparison of clostrubin-positive and -negative clostridia shows that C. puniceum uses a different strategy to cope with an oxygen-rich environment, because the typical repertoire of ROS-inactivating genes is not sufficient for this plant pathogen to grow in the aerobic conditions of a potato tuber. Our data demonstrate that clostrubins promote survival of C. puniceum and C. beijerinckii under aerobic conditions. Thus, these clostridia are not obligate anaerobes but rather facultative anaerobes when producing clostrubins. By tolerating oxygen, the anaerobe gains access to an oxygen-rich plant environment.

Analysis of more than 210 published genome sequences of anaerobes did not reveal any clr-like gene loci. Thus, the involvement of aromatic polyketides in the aerobic growth of anaerobes is not a general phenomenon but a particular trait of individual clostrubin-positive strains. However, the presence of transposons (tnp) in vicinity of the clr locus indicates that the gene cluster is transferable to other bacteria.

In this context, it should also be highlighted that many bacteria use redox-active pigments, such as phenazines, as antibiotics to inhibit competitors (27) and which may also function as regulators that can even control community behavior (28). Notably, phenazine redox cycling may also enhance the anaerobic survival of pseudomonads (29), which is the reversal of the effect of clostrubins under aerobic conditions.

We have shown that clostrubin is essential for a pathogenic anaerobe to enter an oxygen-rich world. The clostrubins thus have a dual function in securing the niche of the plant pathogen, as they represent highly potent antibiotics against potential competitors in the aerobic environment. Clostrubin synthase may be a useful target for development of antibacterial therapeutics and plant-protective agents.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S13

Tables S1 to S4

References (3042)

References and Notes

  1. Acknowledgments: We thank A. Perner for MS measurements, H. Heinecke for NMR measurements, U. Knüpfer and M. Cyrulies for anaerobic fermentation support, K. Martin for assistance in sample preparation, S. Linde for recording micrographs, L. D. Halder for recording confocal fluorescence microscope images, and T. Stinear and T. Seemann for assistance with DNA sequencing and bioinformatic analyses. Supported by an Alexander von Humboldt Foundation fellowship (S.J.P.) and Deutsche Forschungsgemeinschaft grant SFB 1127 ChemBioSys. G.S., K.I., S.J.P., C.H., and the Leibniz Institute for Natural Product Research and Infection Biology (Hans Knoell Institute) have filed a patent application (WO 2015/113761 A1) that relates to the structures of clostrubins A and B and their biological activities. Author contributions: G.S., K.I., and C.H. designed the research; G.S., K.I., S.J.P., and C.H. prepared the manuscript; G.S. performed anaerobic/aerobic bacterial culture experiments, potato tuber infection experiments, and antibacterial assays and constructed the ΔclrA mutant; K.I. and S.J.P. isolated clostrubins; K.I. elucidated structures; S.J.P. performed full genome sequencing and bioinformatic analyses and constructed the C. beijerinckii ΔclrA mutant; H.-M.D. performed FACS analyses; and M.R. designed and supervised anaerobic cultivation in fermenters. Supplement contains additional data.
View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article