J Chem Ecol (2008) 34:1422–1429 DOI 10.1007/s10886-008-9555-7
Amphibian Chemical Defense: Antifungal Metabolites of the Microsymbiont Janthinobacterium lividum on the Salamander Plethodon cinereus Robert M. Brucker & Reid N. Harris & Christian R. Schwantes & Thomas N. Gallaher & Devon C. Flaherty & Brianna A. Lam & Kevin P. C. Minbiole
Received: 17 July 2008 / Revised: 24 August 2008 / Accepted: 25 September 2008 / Published online: 24 October 2008 # Springer Science + Business Media, LLC 2008
Abstract Disease has spurred declines in global amphibian populations. In particular, the fungal pathogen Batrachochytrium dendrobatidis has decimated amphibian diversity in some areas unaffected by habitat loss. However, there is little evidence to explain how some amphibian species persist despite infection or even clear the pathogen beyond detection. One hypothesis is that certain bacterial symbionts on the skin of amphibians inhibit the growth of the pathogen. An antifungal strain of Janthinobacterium lividum, isolated from the skin of the red-backed salamander Plethodon cinereus, produces antifungal metabolites at concentrations lethal to B. dendrobatidis. Antifungal metabolites were identified by using reversed phase high performance liquid chromatography, high resolution mass spectrometry, nuclear magnetic resonance, and UV-Vis spectroscopy and tested for efficacy of inhibiting the pathogen. Two metabolites, indole3-carboxaldehyde and violacein, inhibited the pathogen’s growth at relatively low concentrations (68.9 and 1.82 μM, respectively). Analysis of fresh salamander skin confirmed the presence of J. lividum and its metabolites on the skin of
Electronic supplementary material The online version of this article (doi:10.1007/s10886-008-9555-7) contains supplementary material, which is available to authorized users. R. M. Brucker : C. R. Schwantes : T. N. Gallaher : K. P. C. Minbiole (*) Department of Chemistry and Biochemistry, James Madison University, Harrisonburg, VA 22807, USA e-mail:
[email protected] R. M. Brucker : R. N. Harris : D. C. Flaherty : B. A. Lam Department of Biology, James Madison University, Harrisonburg, VA 22807, USA
host salamanders in concentrations high enough to hinder or kill the pathogen (51 and 207 μM, respectively). These results support the hypothesis that cutaneous, mutualistic bacteria play a role in amphibian resistance to fungal disease. Exploitation of this biological process may provide longterm resistance to B. dendrobatidis for vulnerable amphibians and serve as a model for managing future emerging diseases in wildlife populations. Keywords Janthinobacterium lividum . Plethodon cinereus . Batrachochytrium dendrobatidis . Beneficial bacteria . Violacein . Indole-3-carboxaldehyde
Introduction Nearly one third of amphibian species are threatened with extinction, due in part to the spread of the emerging infectious disease chytridiomycosis (Berger et al. 1998; Stuart et al. 2004; Rachowicz et al. 2006). This disease is linked to the decline of many frog populations, especially in tropical areas not affected by habitat loss (Lips 1999; Lips et al. 2006). Chytridiomycosis is caused by the chytrid fungus Batrachochytrium dendrobatidis and can lead to tissue erosion, hyperkeratosis, moderate hyperplasia, weight loss, and death (Berger et al. 1998). Some species, such as the bullfrog Rana catsbeiana and the Eastern red-backed salamander Plethodon cinereus, persist despite infection (Daszak et al. 2004; unpublished data). Interestingly, survival in the presence of the pathogen can vary within a species; such is the case for Rana muscosa, the mountain yellowlegged frog (Briggs et al. 2005; Woodhams et al. 2007b). An explanation of the differences in susceptibility remains
J Chem Ecol (2008) 34:1422–1429
elusive but may involve differences in innate immune responses (Woodhams et al. 2007a). An additional hypothesis is that a mutualism exists between the amphibian host and cutaneous antifungal bacteria symbionts and is effective on some individuals and species, but not in others (Fig. 1; Harris et al. 2006; Lauer et al. 2007, 2008; Woodhams et al. 2007b). We suspected that the inhibitory effects of the antifungal symbionts are due to secondary bacterial metabolites as opposed to resource competition. Previously, we tested the hypothesis of inhibitory metabolite production with a strain of Lysobacter gummosus isolated from the skin of the salamander P. cinereus. The bacterium inhibited the fungus in challenge assays and produced the antifungal metabolite 2,4-diacetylphloroglucinol in vitro at concentrations that could inhibit the pathogen (Brucker et al. 2008). Continuing our investigations of antifungal metabolites, a bacterium closely related to Janthinobacterium lividum based on DNA sequence similarity was also examined. This bacterium is commonly found on P. cinereus as well as the four-toed salamander, Hemidactylium scutatum, and it strongly inhibited pathogenic fungi in vitro (Lauer et al. 2007). We hypothesized that it, too, produces antifungal metabolites that act to reduce or prevent colonization of B. dendrobatidis on amphibians. Furthermore, we sampled the skins of wild salamanders to determine whether antifungal metabolites were present in inhibitory concentrations. Fig. 1 Model of amphibianbacteria-fungi interactions. The arrows indicate response or causation between the organisms. aAntifungal metabolite produced by L. gummosus. b Antifungal metabolite produced by J. lividum
1423
Methods and Materials Isolation and Culturing of J. lividum As per published method (Harris et al. 2006; Lauer et al. 2007, 2008; Brucker et al. 2008), individuals of the salamander species P. cinereus were rinsed twice to remove transient bacteria and swabbed with sterile swabs. Pure bacterial cultures were obtained, genomic DNA was extracted, and a portion of the 16S rRNA gene was amplified by using 357F and 907R primers (Lauer et al. 2007). DNA obtained from the pure culture of J. lividum was sequenced by Agencourt (Beverly, MA, USA). A consensus sequence of approximately 1,400 bp was obtained by aligning the forward and reverse sequences’ amplicons. The sequence was compared with the reference organisms by BLAST search (http://www.ncbi.nlm.nih.gov/ blast) using the GenBank database in order to confirm the identification of J. lividum (99% match). Co-cultures of J. lividum and B. dendrobatidis (JEL 215 strain) were grown on 1% tryptone agar plates and incubated for 48 h. The plates were checked for the presence of a zone of fungal inhibition around the bacterial colony to reconfirm the bacterium’s antifungal properties (Lauer et al. 2007). Additionally, 100 ml co-cultures were made in liquid 1% tryptone medium to obtain adequate amounts of metabolites for chemical analysis (Brucker et al. 2008). Concurrently, cocultures were compared to monocultures of the organisms to discriminate metabolites that are not produced by J. lividum (unpublished data). To inoculate the cultures, J. lividum was transferred from agar plates into 100 ml of 1% tryptone and
1424
J Chem Ecol (2008) 34:1422–1429
1 ml of a solution containing zoospores (the motile infective stage of B. dendrobatidis; 1–2×106 zoospores/ml). All cultures were incubated for 72 h in a Lab-Line incubator shaker at room temperature (∼23°C). After incubation, cultures were centrifuged at 5,000 rpm for 12 min. The supernatant was kept frozen until organic extraction; the pellet was discarded.
control wells contained the fungus with the solvent DMSO; the negative control wells contained heat-killed zoospores. An exception to our previous published procedure was the reduction of the solvent DMSO to 1% by volume. Indole-3carboxaldehyde treatments and controls were replicated 16 times, while the violacein assay treatments and controls were replicated eight times.
Organic Extraction and Antifungal Metabolite Isolation To extract organic metabolites from the cultures, the supernatant was thawed and extracted ×4 with a 1/3 volume of ethyl acetate (EtOAc). The combined organic layers were dried over Na2SO4, filtered, and evaporated in vacuo. The crude samples (3–15 mg) were kept frozen at −20°C, under nitrogen, until further analysis. A total of eight samples were extracted. Each crude sample was dissolved in high performance liquid chromatography (HPLC)-grade methanol (2–3 ml). RP-HPLC analysis (Agilent Technologies, 1200 series, Wilmington, DE, USA) was then used to determine the retention times of the antifungal standards as well as to analyze the components of bacterial samples. The HPLC diode array detector was programmed to record absorbance at 220, 270, and 310 nm. Samples were injected (50 μl) into the HPLC equipped with a C18 reverse phase column (5 μm; 4.6×150 mm; Agilent Technologies, Wilmington, DE, USA) and eluted at 1 ml/min. The initial eluent, 10% acetonitrile/water (v/v, both acidified with 0.1% acetic acid), ran for 2 min. This was followed by a linear gradient to 100% acetonitrile (acidified with 0.1% acetic acid), over an 18-min period. This final solvent was eluted for another 3 min. Fractions were collected, and those that demonstrated antifungal activity were purified further and identified.
Statistical Analysis A Dunnett’s t test was used to test the null hypothesis of “no inhibition” by comparing the amount of fungal growth in each metabolite concentration treatment to the positive control treatment that had fungi without metabolites. Pair-wise comparisons were made with Tukey’s HSD procedure, which holds experiment-wide error at a maximum of 0.01.
Nuclear Magnetic Resonance and High Resolution Mass Spectrometry Analysis of Antifungal Metabolites 1H nuclear magnetic resonance (NMR) spectra were performed in CDCl3 (indole-3-carboxaldehyde) and DMSO-d6 with an aliquot of CDCl3 (violacein) on a Bruker Advance 600 MHz NMR spectrophotometer. Samples were sent to the Mass Spectrometry Facility at Harvard University (Cambridge, MA, USA) for high resolution mass spectrometry (HR-MS) analysis (Agilent 6210 Time-of-flight LC/MS, ESI source, positive mode). Confirmation of Antifungal Metabolites and Their Inhibitory Activity To determine the minimum inhibitory concentration (MIC) of the compounds required to inhibit the growth of B. dendrobatidis, an in vitro assay was performed according to the procedures previously detailed (Brucker et al. 2008). In brief, concentration of metabolites was varied in 96-well microtiter plates, and fungal growth was determined by measuring optical density. The positive
Extraction of Metabolites from Skin Samples Seven wildcaught P. cinereus were collected from the James Madison University Arboretum (Harrisonburg, VA, USA) during the months of November and December of 2007. Each individual was handled separately with sterilized gloves and housed overnight, in separate sterile containers at 14° C. Salamanders were euthanized with gaseous CO2. A 2.3–4.4-cm 2 portion of skin was excised from the shoulders to the hips, and the surface area of the skin was determined by using the program ImageTool 3.0 (distributed by the University of Texas Health Science Center at San Antonio, Texas. May 2002). The skin components were then extracted with HPLC-grade methanol (4×5 ml). The organic solvent was evaporated, and the resulting sample was then constituted with 200 μl of HPLC-grade methanol before being injected onto the HPLC, as per above. Approximate concentrations were determined by computing a standard curve of the metabolites. Animals were collected by permit from the Virginia Division of Game and Inland Fisheries. Our animal care protocol was approved by JMU’s Animal Care and Use Committee. Sampling for Naturally Occurring J. lividum Sampling and analysis of cutaneous bacteria follow methods published in detail by us elsewhere (Lauer et al. 2007). Briefly, transient bacteria were removed from salamander skin by rinsing them twice in sterile water and then swabbing their ventral and lateral sides with a wet sterile cotton swab (Harris et al. 2006; Lauer et al. 2007, 2008). The presence of J. lividum was tested by using polymerase chain reaction or denaturing gradient gel electrophoresis (DGGE; see published methods (Lauer et al. 2007)). Typically one bacterial species is indicated by one band (sequence type) on DGGE gels, although heterogeneity in ribosomal RNA copy number in some species can cause a species to appear as two or more bands. In one lane, we placed the amplification products from a
J Chem Ecol (2008) 34:1422–1429
1425
pure culture of J. lividum. Other lanes of the DGGE contained a culture-independent assay of bacterial diversity on salamander skins. If one band in these lanes migrated to the same position as the band from the pure culture of J. lividum, then we concluded that J. lividum was present on the skin. Evaluating Mucus Depth of P. cinereus The mucous depth of P. cinerus was calculated by using three salamanders collected from local populations in order to determine metabolite concentration on salamander skins. The salamanders were sedated and euthanized using carbon dioxide. Skin samples were removed using a double-headed scalpel, which was assembled by using two size 10 scalpel blades taped parallel to each other. The resulting double-headed scalpel had a parallel 1–1.5 mm span between the blades and was used to cut and remove three 1–1.5-mm thick cross-sections, per salamander, at three locations: on the tail, above the hind legs, and below the fore limbs. Scalpel blades were replaced between each cut. Each crosssectional sample was rotated 90° and immediately laid transversely onto a glass microscope slide. To avoid distortion of the sample from the weight of the cover slip, two platforms were cemented on either side of the sample comprised of 7–10 cover slips, with a final cover slip spanning the sample field. Drops of a 1% saline solution were added to the sample to prevent dehydration of the sample. Two to three drops of methylene blue were added to the saline solution to provide contrast for the background and translucent mucus layer. The slide was placed on an inverse microscope within half an hour of construction. Five out of the nine slides had undamaged, viable, cross-sectional samples. From these five samples, the depth of the mucus layer was measured at five random locations per sample, using microscopy measurement software (Nikon Imaging Systems Elements, Tokyo, Japan).
Results Compounds in the culture medium produced by J. lividum that were fractioned around 9.07 min and 11.15 min via HPLC (“Electronic Supplementary Material” SI Fig. 1) inhibited fungal growth. Further HPLC purification of these bioactive fractions led to the isolation of two antifungal metabolites whose structures were identified as violacein and indole-3-carboxaldehyde (Fig. 2) via NMR and HR-MS analysis (“Electronic Supplementary Material” SI Table 1). Optical density was used to assess fungal growth, and inhibitory growth assays against B. dendrobatidis were performed to determine the MIC for each of the metabolites. Increasing concentrations of violacein and indole-3-carboxaldehyde reduced fungal growth (violacein: F=73.32, df=1,
Fig. 2 Chemical structures of indole-3-carboxaldehyde and violacein, the two antifungal metabolites produced by J. lividum isolated from both the redbacked salamander (P. cinereus) and the four-toed salamander (H. scutatum)
180, P
