https://orcid.org/0000-0001-8827-023X
Abstract
Chytridiomycosis, a lethal fungal disease caused by Batrachochytrium dendrobatidis (Bd), is responsible for population declines and extinctions of amphibians worldwide. However, not all amphibian species are equally susceptible to the disease; some species persist in Bd enzootic regions with no population reductions. Recently, it has been shown that the amphibian skin microbiome plays a crucial role in the defense against Bd. Numerous bacterial isolates with the capacity to inhibit the growth of Batrachochytrium fungi have been isolated from the skin of amphibians. Here, we characterized eight Acinetobacter bacteria isolated from the frogs Agalychnis callidryas and Craugastor fitzingeri at the genomic level. A total of five isolates belonged to Acinetobacter pittii,Acinetobacter radioresistens, or Acinetobactermodestus, and three were not identified as any of the known species, suggesting they are members of new species. We showed that seven isolates inhibited the growth of Bd and that all eight isolates inhibited the growth of the phytopathogen fungus Botrytis cinerea. Finally, we identified the biosynthetic gene clusters that could be involved in the antifungal activity of these isolates. Our results suggest that the frog skin microbiome includes Acinetobacter isolates that are new to science and have broad antifungal functions, perhaps driven by distinct genetic mechanisms.
Introduction
In recent decades, amphibian species have suffered drastic population declines and extinctions all over the world due to chytridiomycosis (Stuart et al. 2004, Scheele et al. 2019), a lethal disease caused by the chytrid fungus Batrachochytrium dendrobatidis (Bd; Berger et al. 1998), and the more recently described Batrachochytrium salamandrivorans (Martel et al. 2013). These pathogenic fungi penetrate the skin of amphibians as part of their life cycle and in turn alter several skin functions, especially those involved in maintaining the osmotic balance (Voyles et al. 2009).
The origins of these pathogenic fungi have been traced to East Asia, and their global expansion is explained by anthropogenic factors such as the international commercial trade of amphibians (Martel et al. 2014, O'Hanlon et al. 2018). Specifically, Bd affects a wide range of amphibians, with anurans (frogs and toads) being the most affected by this disease, particularly in Central America, South America, and Australia (Scheele et al. 2019). Nonetheless, not all amphibian species are equally susceptible to the disease, considering that some populations persist in Bd enzootic regions with no apparent population reductions (Daszak et al. 1999, Peterson et al. 2007, Rebollar et al. 2014, Rodríguez-Brenes et al. 2016). Bd infection severity can be explained by many factors, including variation in pathogen virulence and genetics (Campbell et al. 2019, Voyles et al. 2011), as well as host immunity variation (including the production of antimicrobial peptides; Grogan et al. 2018). In addition, studies from the last decade have shown that the amphibian skin microbiome plays a relevant role in the defense against Bd (Harris et al. 2009b, Bernardo-Cravo et al. 2020, Bates et al. 2018, Rebollar et al. 2016). The composition of the amphibian skin microbiome varies across amphibian species and populations (McKenzie et al. 2012, Belden et al. 2015, Costa et al. 2016), and in some cases, Bd susceptibility has been correlated with particular skin microbial configurations (Lam et al. 2010, Burkart et al. 2017). In fact, thousands of bacterial strains with the capacity to inhibit Batrachochytrium fungi have been isolated from the skin of numerous amphibian species around the world, and in some cases, the metabolites involved in this inhibitory process have been identified (Kearns et al. 2017, Woodhams et al. 2015, 2018, Loudon et al. 2014, Brucker et al. 2008a, 2008b, Becker et al. 2009). Specifically, the antifungal effect of many skin bacterial isolates has encouraged the search for bacterial candidates that could be used as probiotics to protect amphibians against chytridiomycosis (Harris et al. 2009a, 2009b, Kueneman et al. 2019). In addition, some of these skin bacteria also inhibit human pathogenic bacteria and fungi (Ramsey et al. 2015, Martin et al. 2019). Thus, the search for bacteria producing these kinds of compounds represents an opportunity to identify and develop new drugs to combat pathogens.
Recent studies of the skin microbiota of neotropical frogs showed that bacteria belonging to the Acinetobacter genus have high relative abundances in these microbiomes (Rebollar et al. 2016, 2018), and overall, this genus has been found in the skin microbiome of several amphibian species across the globe (Kueneman et al. 2019). Members of the Acinetobacter genus have been isolated from the skin of tropical amphibian species and have shown an ability to inhibit the growth of Bd in vitro (Rebollar et al. 2019). Even though Acinteobacter strains seem to be common on amphibian skin and possibly play a protective role against pathogens, to our knowledge, their similarity with known Acinetobacter species and their genomic content have not yet been described. Genomic exploration and functional analyses of amphibian skin bacteria will allow us to gain insights into the functions that these bacteria play as part of the amphibian microbiota. Moreover, these approaches can be useful to identify new species from underexplored sources, such as the amphibian skin.
In this work, we identified and characterized eight Acinetobacter isolates obtained from the neotropical frogs Agalychnis callidryas and Craugastor fitzingeri using genomic approaches. We evaluated their growth inhibition capacity in vitro against Bd and against the necrotrophic fungus Botrytis cinerea, an important phytopathogen that affects a wide variety of crops, especially grapes (Veloso and van Kan 2018). We identified biosynthetic clusters that could explain their antifungal capacity against both types of fungi. Moreover, we found that some of the isolates could not be identified as any of the known Acinetobacter species, suggesting that they are members of new species from this genus.
Materials and methods
Previous Acinetobacter collection and identification
Acinetobacter bacteria were collected, isolated, and identified as described previously (Rebollar et al. 2019). Briefly, 359 bacterial strains were collected from three species of tropical frogs: the tree frogs A. callidryas and Dendropshophus ebbracatus and the terrestrial frog C. fitzingeri from Gamboa, Panama, in September 2012. All bacterial isolation using R2A agar plates occurred during November and December 2012 at the Tropical Smithsonian Research Institute in Panama City, Panamá. Taxonomic identification took place at James Madison University: DNA of all isolates was extracted, the full-length PCR product of the 16S rRNA gene of each sample was obtained, and a fragment of the PCR product was Sanger sequenced using the universal primer 907R (Eurofins Louisville, KY). Taxonomic classifications were assigned by aligning all sequences to the SILVA database (Quast et al. 2013). All 359 Sanger sequences were deposited in NCBI with the accession numbers MK506363–MK506718 [see supplementary methods in Rebollar et al. (2019)] From the entire collection, eight bacterial isolates described in this study were assigned to the Acinetobacter genus (two from A. callidryas and six from C. fitzingeri) and showed range in their ability to inhibit Bd in vitro using challenge assays (Rebollar et al. 2019).
Genome sequencing
All Acinetobacter isolates were cultured overnight at 37°C and 250 rpm in 5 ml of Luria–Bertani (LB) liquid medium. Genomic DNA for each bacterial culture was extracted with the Genomic DNA purification kit (Thermo Fisher). After measuring the quantity and quality of all DNAs, short read library construction and sequencing (2 × 150 bp) were performed at Beijing Genomic Institute (China) using the BGISEQ-2000 platform. Adapter removal and quality trimming of the DNAseq reads were performed using Trim Galore v0.6.4 (https://github.com/FelixKrueger/TrimGalore).
Isolates that, according to our analyses, could be new Acinetobacter species were sequenced with long reads (Oxford-Nanopore Technologies) at the Unidad Universitaria de Secuenciación Masiva y Bioinformática (UNAM, México). Long-read libraries were constructed following the protocol “Genomic DNA by ligation” (SQK-LSK109; Oxford Nanopore Technologies). DNAs were not sheared and used directly from purification to library construction. Reads were obtained using a MinION device with the MinION R9.4.1 flow cell. Base calling was performed using Guppy software v4.0.14, and adapter sequences were removed using Porechop v0.2.4 (https://github.com/rrwick/Porechop).
Short-read assemblies were initially constructed with three programs: ABySS 2.0.1 (Jackman et al. 2017), SPAdes 3.9.0 (Bankevich et al. 2012), and Velvet 1.2.10 (Zerbino and Birney 2008), with various kmers. The best assembly obtained with each program was selected for further use with Metassembler 1.5 (Wences and Schatz 2015) to obtain a merged and optimized assembly of each isolate. Hybrid assemblies and circularization were obtained using trimmed short reads and raw long reads with Unicycler (v0.4.7; Wick et al. 2017). Assembly statistics were calculated with Quast (v5.0.2; Gurevich et al. 2013). The BUSCO tool suite (v5.1.2; Seppey et al. 2019) was employed to evaluate the completeness of each genome, and all were annotated using the NCBI Prokaryotic Genome Annotation Pipeline (Tatusova et al. 2016). All programs were used with the default options.
Accession numbers
All genomes were deposited at DDBJ/ENA/GenBank. New potential species genomes have the following accession numbers: Acinetobacter sp. C32I, CP098480–CP098481; Acinetobacter sp. C26G, CP098479; Acinetobacter sp. C26M, CP098478. Accession numbers for genomes of previously described species are: Acinetobacter pittii A45P, JAMOZH000000000; A. pittii A47H, JAMOZG000000000; Acinetobacter radioresistens C21A, JAMOZF000000000; A. radioresistens C30P; JAMOZE000000000; Acinetobacter modestus C23M, JAMOZD000000000.
Comparative analysis of genome sequences
The pairwise average nucleotide identity (ANI) based on MUMmer was calculated with pyANI (0.2.9; Pritchard et al. 2016) for the eight Acinetobacter genome sequences obtained in this work and the genomes of the 74 Acinetobacter type strains with a validly published name, plus 31 genomes from tentative species or true species without a validly published species name. All genome sequences used in this work are listed in the Alexander Nemec’s home page: apps.szu.cz/anemec/Classification.pdf. ANI values of ≥ 95% between genomes were used to allocate organisms to the same species according to Goris et al. (2007).
The core genomes of the amphibian skin Acinetobacter isolates were identified with GET_HOMOLOGUES (Contreras-Moreira and Vinuesa 2013) using the genome sequences of all Acinetobacter strains mentioned above. The GET_PHYLOMARKERS program was used to select single-copy genes without recombination signals and lacking evidence of horizontal transfer (Vinuesa et al. 2018). Additionally, all in-paralog genes were eliminated from the analysis (Vinuesa et al. 2018). A maximum likelihood phylogenetic tree was constructed with the codon alignments of the selected 240 genes passing these filters under the best-fitting substitution model using IQ-TREE (Minh et al. 2020). The inferred tree was visualized and annotated with iTOL (Letunic and Bork 2021).
In vitro antagonistic effect of Acinetobacter isolates against B. dendrobatidis growth
Bacterial cell-free supernatants (CFS) were obtained from a single liquid culture for each Acinetobacter isolate. All CFSs were tested against Bd using the 96-well challenge assay method described in Bell et al. (2013) and Rebollar et al. (2019) with modifications as follows: first, the bacterial CFSs were obtained by filtering each liquid culture grown for 3 days on a shaker (200 rpm) through a 0.22-μm filter. Then, Bd zoospores of the global panzootic lineage strain (JEL 423) were collected by flooding 7-day-old plate cultures (1% tryptone) and filtering the liquid to obtain only the zoospore stage of the fungus. Finally, the challenge assays were set up in 96-well microplates in which experimental wells contained 50 μl of Bd zoospores at a concentration of 2 × 106 ml−1 and 50 μl of a CFS sample. Since Bd may self-regulate its growth by tryptophol production (Verbrugghe et al. 2018), we did not grow bacterial isolates in coculture with Bd zoospores before the assay, as done in previous methods (Bell et al. 2013, Rebollar et al. 2019). Each of the CFS isolates was tested in triplicate along with the following controls: (1) 50 µl of 1% tryptone + 50 µl of Bd zoospores (positive control); (2) 50 µl of heat-killed Bd zoospores + 50 µl of 1% tryptone (heat-killed control); and (3) 100 µl of 1% tryptone only (negative control). Assay plates were incubated at 21°C, and growth was measured as the optical density (OD) at 492 nm on a spectrophotometer on days 0, 4, 7, and 10.
To calculate an inhibition score for each isolate, the slope of a regression of OD readings over time was calculated, and triplicate values from each replicate were averaged to generate a mean slope per isolate. Then, the mean slope of each isolate was divided by the mean slope of the positive control to determine the proportion of growth. Finally, the growth proportion was subtracted from 1 to determine the inhibition score, in which the values closer to 1 represent those isolates with higher inhibition capacity.
In vitro antagonistic effect of Acinetobacter isolates against B. cinerea growth
Each Acinetobacter isolate was grown in LB liquid medium until an OD of 0.6 was reached, while a spore suspension of B. cinerea strain B05.10 (kindly provided by Brigitte Mauch-Mani) was prepared as previously described (L’Haridon et al. 2011). To evaluate the inhibitory effect of each Acinetobacter strain against B. cinerea, 10 µl of each bacterial and spore suspension was placed on opposite sides of a Petri dish containing PDA medium and incubated at 24°C for 7 days. After this time, the effect on the fungus was evaluated by measuring the percentage of inhibition on B. cinerea mycelium radial growth. As a control, a Petri dish containing only the fungus without bacteria was incubated in parallel. Pictures were taken with a digital camera, and images were analyzed using ImageJ version 2 (NIH). For each Acinetobacter strain, two plates were analyzed per experiment, and three independent experiments were performed. All results are reported as the mean values (± SD). Tukey’s analysis (P < .05) was carried out to determine statistically significant differences between the control and each of the samples. GraphPad Prism version 9.2.0 software was used (GraphPad Software, San Diego, CA, USA, 2019). All the data analyzed were obtained from three independent experiments.
Prediction of biosynthetic gene clusters
Biosynthetic gene cluster (BGC) predictions of the Acinetobacter isolates were obtained using the online antiSMASH tool (version 6.0) with relaxed detection strictness (Blin et al. 2021). BCGs identified by this tool were searched in a repository for BGCs of known function based on experimental analysis called the MIBiG database (version 2.0; Kautsar et al. 2020). We searched the literature for experimental evidence that all BCGs matched with a specific entry of this database by at least 70% genes. We clustered the BGCs using BiG-SCAPE (Navarro-Muñoz et al. 2020). Finally, gene maps comparing the BCGs predicted in the genomes of our Acinetobacter collection and the MIBiG database-specific matches were made with the genome visualizer tool EasyFig (Sullivan et al. 2011).
Antibiotic susceptibility and antibiotic resistance genes
Considering that some Acinetobacter species can be opportunistic human pathogens, with variable degrees of antibiotic resistance (Howard et al. 2012), we explored antibiotic tolerance and the presence of antibiotic resistance genes in the eight Acinetobacter from this study. Antibiotic resistance genes were identified using RGI v5.2.0 and CARD 3.1.4 (Alcock et al. 2020). Antibiotic susceptibility profiles for the eight isolates were obtained using a Vitek® 2 automated system (BioMériux, Marcy l’Étoile, France), and the antibiotics considered were amikacin, cefepime, ceftriaxone, ciprofloxacin, colistin, doripenem, gentamicin, imipenem, meropenem, tigecycline, and β-lactam combination agents, ampicillin–sulbactam and piperacillin–tazobactam, according to the Clinical and Laboratory Standards Institute.
Phenotypic characterization of the new Acinetobacter species
To identify potentially new bacterial species, it is relevant to describe the phenotypic profiles of the strains including physiological/metabolic and structural features such as the following:
1. Physiological and metabolic characterization
The physiological and metabolic characteristics of the amphibian skin isolates not belonging to any of the validly named species were assessed using a genus-targeted set of in-house standardized tests (Krizova et al. 2015, Nemec et al. 2009) and compared with recently published data (Nemec 2022). Except for the temperature growth tests, the culture temperature was 30°C. The carbon assimilation tests were performed in a fluid mineral medium supplemented with a 0.1% (w/v) of each carbon source. The assimilation tests were interpreted after 6 days of culture. The haemolytic and gelatinase activities were examined after 3 days, and D-glucose acidification and temperature growth tests were performed after 2 days.
2. Cellular fatty acid profiles
The cellular fatty acids of isolates C32I and C26G were identified at “Colección Española de Cultivos Tipo; Universitat de València.” C26M was not described since the genomic analysis showed a high similarity with C26G. The isolates were grown in LB medium at 37°C for 24 h. Then, the cells were harvested, and the fatty acids were extracted following the protocols recommended in the MIDI Sherlock Microbial Identification System Manual. Fatty acid profiles were obtained using an Agilent 6850 gas chromatograph.
Results
Taxonomic characterization of the amphibian skin isolates
A total of eight Acinetobacter isolates were obtained from the skin microbiota of two neotropical frog species as part of a previous study (Rebollar et al. 2019). Isolates C21A, C23M, C30P, and C32I were obtained from different specimens of the terrestrial frog C. fitzingeri. C26G and C26M were isolated from the same individual. A45P and A47H were collected from two different individuals of the tree frog A. callidryas (Table 1). All of them were initially identified as belonging to the Acinetobacter genus through Sanger sequencing of the 16S rRNA gene (Rebollar et al. 2019).
Characteristics of the Acinetobacter isolates studied. G_SIZE, genome size in Mb. GI_Bd, percentage of growth inhibition of B. dendrobatidis. GI_Bc, percentage of growth inhibition of B. cinerea. Values in the last column followed by different letters indicate significant differences according to the Tukey’s test (P < .05). Asterisks indicate isolates that were obtained from the same individual.
| Isolate | Bacterial species | Amphibian host | G_SIZE | %GC | GI_Bd | GI_Bc |
|---|---|---|---|---|---|---|
| A45P | A. pitti | A. callidryas | 3.80 | 38.75 | 92.54 | 27.33 ± 14.6a |
| A47H | A. pitti | A. callidryas | 3.97 | 38.72 | 94.89 | 27.40 ± 15.0a |
| C21A | A. radioresistens | C. fitzingeri | 3.01 | 41.76 | 21.80 | 33.98 ± 10.1a |
| C23M | A. modestus | C. fitzingeri | 3.54 | 38.39 | 1.88 | 29.31 ± 9.55a |
| C30P | A. radioresistens | C. fitzingeri | 3.02 | 41.75 | 26.27 | 47.46 ± 7.8b |
| C26G* | Acinetobacter sp. | C. fitzingeri | 3.93 | 40.52 | 42.34 | 47.61 ± 15.6b |
| C26M* | Acinetobacter sp. | C. fitzingeri | 3.93 | 40.50 | 50.23 | 44.71 ± 3.31b |
| C32I | Acinetobacter sp. | C. fitzingeri | 4.19 | 41.32 | 43.93 | 58.41 ± 10.8b |
| Isolate | Bacterial species | Amphibian host | G_SIZE | %GC | GI_Bd | GI_Bc |
|---|---|---|---|---|---|---|
| A45P | A. pitti | A. callidryas | 3.80 | 38.75 | 92.54 | 27.33 ± 14.6a |
| A47H | A. pitti | A. callidryas | 3.97 | 38.72 | 94.89 | 27.40 ± 15.0a |
| C21A | A. radioresistens | C. fitzingeri | 3.01 | 41.76 | 21.80 | 33.98 ± 10.1a |
| C23M | A. modestus | C. fitzingeri | 3.54 | 38.39 | 1.88 | 29.31 ± 9.55a |
| C30P | A. radioresistens | C. fitzingeri | 3.02 | 41.75 | 26.27 | 47.46 ± 7.8b |
| C26G* | Acinetobacter sp. | C. fitzingeri | 3.93 | 40.52 | 42.34 | 47.61 ± 15.6b |
| C26M* | Acinetobacter sp. | C. fitzingeri | 3.93 | 40.50 | 50.23 | 44.71 ± 3.31b |
| C32I | Acinetobacter sp. | C. fitzingeri | 4.19 | 41.32 | 43.93 | 58.41 ± 10.8b |
Characteristics of the Acinetobacter isolates studied. G_SIZE, genome size in Mb. GI_Bd, percentage of growth inhibition of B. dendrobatidis. GI_Bc, percentage of growth inhibition of B. cinerea. Values in the last column followed by different letters indicate significant differences according to the Tukey’s test (P < .05). Asterisks indicate isolates that were obtained from the same individual.
| Isolate | Bacterial species | Amphibian host | G_SIZE | %GC | GI_Bd | GI_Bc |
|---|---|---|---|---|---|---|
| A45P | A. pitti | A. callidryas | 3.80 | 38.75 | 92.54 | 27.33 ± 14.6a |
| A47H | A. pitti | A. callidryas | 3.97 | 38.72 | 94.89 | 27.40 ± 15.0a |
| C21A | A. radioresistens | C. fitzingeri | 3.01 | 41.76 | 21.80 | 33.98 ± 10.1a |
| C23M | A. modestus | C. fitzingeri | 3.54 | 38.39 | 1.88 | 29.31 ± 9.55a |
| C30P | A. radioresistens | C. fitzingeri | 3.02 | 41.75 | 26.27 | 47.46 ± 7.8b |
| C26G* | Acinetobacter sp. | C. fitzingeri | 3.93 | 40.52 | 42.34 | 47.61 ± 15.6b |
| C26M* | Acinetobacter sp. | C. fitzingeri | 3.93 | 40.50 | 50.23 | 44.71 ± 3.31b |
| C32I | Acinetobacter sp. | C. fitzingeri | 4.19 | 41.32 | 43.93 | 58.41 ± 10.8b |
| Isolate | Bacterial species | Amphibian host | G_SIZE | %GC | GI_Bd | GI_Bc |
|---|---|---|---|---|---|---|
| A45P | A. pitti | A. callidryas | 3.80 | 38.75 | 92.54 | 27.33 ± 14.6a |
| A47H | A. pitti | A. callidryas | 3.97 | 38.72 | 94.89 | 27.40 ± 15.0a |
| C21A | A. radioresistens | C. fitzingeri | 3.01 | 41.76 | 21.80 | 33.98 ± 10.1a |
| C23M | A. modestus | C. fitzingeri | 3.54 | 38.39 | 1.88 | 29.31 ± 9.55a |
| C30P | A. radioresistens | C. fitzingeri | 3.02 | 41.75 | 26.27 | 47.46 ± 7.8b |
| C26G* | Acinetobacter sp. | C. fitzingeri | 3.93 | 40.52 | 42.34 | 47.61 ± 15.6b |
| C26M* | Acinetobacter sp. | C. fitzingeri | 3.93 | 40.50 | 50.23 | 44.71 ± 3.31b |
| C32I | Acinetobacter sp. | C. fitzingeri | 4.19 | 41.32 | 43.93 | 58.41 ± 10.8b |
To obtain a precise taxonomic assignment at the species level, we compared the genomic sequences of the eight isolates (Table 1 and Table S1, Supporting Information). We first calculated the pairwise ANI between the eight isolates versus the type strains of all known Acinetobacter species (Fig. 1). Isolates A47H and A45P were identified as A. pittii (both isolates with an ANI of 96.7% to NZ_KB849785.1) and have an ANI value of 99.9% between them. Isolates C21A and C30P were identified as A. radioresistens (ANI of 98.9% to NZ_KB849737.1) and have an ANI value of 98.9% between them. Isolate C23M was identified as A. modestus (ANI of 97.08% to NZ_KB849180.1). Interestingly, the three remaining isolates yielded ANI values < 92% to all known Acinetobacter species, suggesting that they are members of new species. In the case of isolate C32I the highest ANI value was 91.78% to A. proteolyticus (NZ_KB849177.1), whereas isolates C26M and C26G had ANI values of 89.90% and 89.78% to genomic species 15BJ (NZ_KE007344.1) and an ANI of 100% between them. Taking into account that these isolates represent the same strain, we will refer to them as C26G/C26M. To look deeper into the genomic content of the potentially novel species (C26M/C26G and C32I), we completed their genomes as explained in the Methods section (see Table S2, Supporting Information).
ANI among Acinetobacter relevant species based on entire genome sequences. Heat map reflects the degree of identity between genomes. An ANI above 95% between two genomes is an indication that they belong to the same species. The Acinetobacter isolates studied are shown in different colors.
To establish the phylogenetic position of these isolates within the Acinetobacter genus, we constructed a maximum likelihood phylogenetic tree using 240 core-genome single copy genes lacking recombination signals (Fig. 2). The ML tree showed that isolates C32I, C26M, and C26G are part of the haemolytic clade (Nemec et al. 2016) This clade includes A. beijerinckii,A. colistiniresistens,A. courvalinii,A. dispersus,A. gyllenbergii,A. haemolyticus,A. halotolerans,A. junii,A. modestus,A. parvus,A. proteolyticus,A. tjernbergiae,A. venetianus, and A. vivianii, and three provisional species without a formal status in the nomenclature (genomic sp. 6, 15BJ, and 16; Nemec 2022). In agreement with the ANI results, C26M and C26G were placed in the phylogram close to genomic 15BJ. Based on the genomic results, we suggest that strains C32I and C26M/C26G are members of two new Acinetobacter species.
Maximum likelihood phylogenetic tree constructed with single-copy genes of the core genome of the isolates studied in the context of the genus Acinetobacter. Moraxella catarrhalis CCRI-195ME was used as the outgroup. The novel isolates are shown in different colors. Yellow and green boxes indicate the species of the haemolyticus clade and the Acinetobacter calcoaceticus–A. baumannii complex, respectively. Clades with bootstrap values higher than 90% are marked with a gray circle.
Antibiotic resistance of Acinetobacter isolates
Acinetobacter isolates were susceptible to all antibiotics tested with the exception of strains A45P and A47H, which showed intermedium susceptibility to ceftriaxone under our specific conditions (Table S3, Supporting Information). Despite the susceptibility found in all isolates, we found genes linked to antibiotic resistance and efflux pumps in all of them. Specifically, we found that the two A. pittii isolates A47H and A45P each carry an intrinsic blaOXA-821, an ADC-18-encoding gene linked to cephalosporin resistance. Acinetobacterradioresistens C21A and C30P carry blaOXA-23-like, which is intrinsic to this species (Poirel et al. 2008). Acinetobacter sp. C26G/C26M possessed blaOXA-672, whereas A. sp. C32I had an intrinsic blaOXA-286 and two genes encoding aminoglycoside-modifying enzymes (aac(6’)-Ix and ant(3’’)-IIc).
Acinetobacter isolates inhibit the growth of B. dendrobatidis
The 96-well challenge assays showed that the CFS of seven (out of eight) Acinetobacter isolates inhibited the growth of Bd (Fig. 3). Acinetobacter pittii A47H and A. pittii A45P showed the strongest inhibition percentage of all isolates (94.8% and 92.5%, respectively), followed by A. sp C26G/C26M and A. sp C32I (50.23%, 42.33%, and 43.93%, respectively), which showed moderate inhibition. Acinetobacterradioresistens C30P and A. radioresistens C21A showed lower levels of inhibition (26.26% and 21.8%, respectively), and A. modestus C23M showed no inhibition (1.87%). These results indicate that most CFS produced by the Acinetobacter isolates (except C23M) showed Bd growth inhibition to varying degrees (Table 1).
Batrachochytrium dendrobatidis growth inhibition capacity of frog skin Acinetobacter isolates. Growth curves of B. dendrobatidis in the presence of Acinetobacter cell-free supernatants. Lines indicate average values of three replicates and vertical lines indicate standard deviations. Black dashed lines indicate negative controls and gray line indicates the positive control. Colored lines represent different Acinetobacter isolates.
Acinetobacter isolates inhibit mycelial growth of B. cinerea
Confrontation assays showed that all the Acinetobacter isolates were able to inhibit the mycelial growth of B. cinerea at different levels. Although mycelial growth was not fully inhibited by any of the bacterial strains, in all the treatments, B. cinerea growth was significantly reduced compared to the control, in which the fungus grew over the whole Petri dish (Fig. 4A). The % inhibition revealed that A. sp C32I, A. sp C26G, and A. radioresistens C30P had the strongest antagonistic activity against B. cinerea, with values of 58.41%, 47.61%, and 47.46%, respectively. However, isolates A. pittii A47H, A. pittii A45P, A. radioresistens C21A, A. modestus C23M, and A. sp C26M inhibited the fungus radial growth between 27.33% and 44.71% (Fig. 4B). Taken together, these results indicate that all Acinetobacter isolates inhibit the mycelial growth of B. cinerea (Table 1).
Botrytis cinerea growth inhibition capacity of frog skin Acinetobacter isolates. (A) Representative competition assays between B. cinerea and each Acinetobacter isolate. (B) Mycelium radial growth inhibition measurements of B. cinerea. Letters above each bar represent statistically significant differences based on Tukey’s pairwise tests (P < .05). Error bars represent mean values (± SD) of three independent experiments (n = 2).
BGC predictions
To identify potential antimicrobial BGCs within the Acinetobacter genomes we used antiSMASH (Blin et al. 2021). With this data mining tool, we found 40 BGCs within the eight Acinetobacter genomes (Table S4, Supporting Information). These predicted clusters have matches with a high percentage of sequence identity ranging from 58% to 100%, with BGCs already identified in other Acinetobacter species. The predicted secondary metabolites synthesized by the BGCs are arylpolyenes, betalactones, nonalpha polyamino acids, like e-Polylysin (NAPPA), nonribosomal peptides (NRPs), posttranslationally modified peptide products (RiPP-like), and siderophores (Fig. 5). The predicted BGCs were then matched to the MIBiG database, a repository that embraces more than 800 BGCs of known function with experimental validation, and can be used to infer products encoded by detected BGCs (Kautsar et al. 2020). With this tool, we found one BGC in the strains C26G/C26M, C23M, and C32I that shared 70% of the genes to a MIBiG cluster BGC0000295 characterized in A. haemolyticus (Figure S1, Supporting Information). This gene cluster is involved in acinetoferrin biosynthesis. It has been shown that this siderophore produced by A. calcoaceticus is capable of inhibiting the mycelium growth of Fusarium oxysporum, a plant pathogen (Maindad et al. 2014).The other predicted BCGs had a lower percentage of genes shared with MIBiG, suggesting that further research is required to experimentally characterize the natural products encoded by Acinetobacter species.
BGC predictions of the Acinetobacter isolates derived from the skin of frogs A. callidryas and C. fitzingeri. From left to right: phylogenetic relationships among the eight isolates, antifungal inhibitory properties of the isolates against B. dendrobatidis (Bd) and B. cinerea (Bc), presence/absence of BGCs colored by different metabolite types. BGCs are shown according to BiG-SCAPE clustering, and named based on its Cluster Blast compound hit (Table S4, Supporting Information).
Physiological and metabolic characteristics of taxonomically novel isolates
We explored the physiological and metabolic phenotypes of the novel Acinetobacter isolates C26G/C26M and C32I (Table S5, Supporting Information) in the context of the haemolytic clade (Nemec 2022). The isolates grew up to 37°C, hemolyzed sheep blood, hydrolyzed gelatin, did not acidified D-glucose, and utilized 15 (C26G/C26M) or 16 (C32I) of the 36 carbon sources tested. They could be differentiated from all the known members of the haemolytic clade by their growth on tryptamine combined with the inability to assimilate gentisate and putrescine. Unlike C26M/ C26G, C32I grew on β-alanine and malonate, but did not assimilate glutarate.
Only the cellular fatty acid profiles of the novel isolates C32I and C26G were obtained, considering that C26M and C26G are clonally related, as mentioned above. The major fatty acids of strain C32I were C18:1 ω9c (34.38%), C16:0 (26.24%), and C12:0 3OH (5.49%), and those of strain C26G were C18:1 ω9c (30.51%), C16:0 (22.49%), and C12:0 3OH (6.06%) (Figure S2, Supporting Information).
Discussion
Despite the enormous body of work showing that amphibian skin bacteria have antifungal capacities, very little is known about the genomic bases behind this trait. To our knowledge only two bacterial genomes, derived from the amphibian skin, have been fully sequenced and, in both cases, BGCs associated to antimicrobial functions have been identified (Bletz et al. 2019, Brunetti et al. 2022). Thus, we consider that genomic approaches can be extremely useful to describe (and perhaps predict) antimicrobial capacities which could be fundamental to unravel the role of these bacteria as host-associated symbionts. In this study, we focused on isolates from the Acinetobacter genus and we revealed the potential genetic mechanisms behind their antifungal capacities. Moreover, through comparative genomics and functional analyses we were able to identify two potentially new Acinetobacter species.
Acinetobacter is a common genus present in a wide range of ecological systems, such as freshwater and marine environments, soils, and sediments. Some of its members have been isolated in extreme environments, including polar regions (Jung and Park 2015, Opazo-Capurro et al. 2019). Others are capable of resisting desiccation and tolerating common disinfectants (Zeidler and Müller 2019). Many species of the genus have versatile metabolic capabilities that allow them to grow on long-chain dicarboxylic acids, polycyclic aromatic hydrocarbons, and other unconventional carbon sources (Premnath et al. 2021). A crucial observation regarding this bacterial genus is that it is frequently found as an endophyte of many plants (Rahal and Chekireb 2021, Wu et al. 2021) and as an important component of the microbiome of lower animals such as the Edwardsiella andrillae anemone (Murray et al. 2016), the coral Stylophora pistillata (Yang et al. 2017) or the snail Biomphalaria glabrata (Silva et al. 2013), and a wide range of vertebrates from fish to humans (Mendoza-Hoffmann et al. 2018, Ortega et al. 2021, Das et al. 2021, Antunes et al. 2014). It has been shown that Acinetobacter is a common component in the skin microbiome of many amphibians (Proença et al. 2021, Rebollar et al. 2019, Bates et al. 2018). Some Acinetobacter strains isolated from the skin of frogs produce secondary metabolites that have antifungal properties against B. dendrobatidis (Brucker et al. 2008b, Martin et al. 2019). To improve our knowledge of the role of Acinetobacter in the microbiomes of the skin of frogs, we characterized eight isolates obtained from the Neotropical frogs A. callidryas and C. fitzingeri. Based on their sequenced genomes, five of them were identified as known Acinetobacter species. However, three of them (C26G/C26M and C32I) did not belong to any of the known Acinetobacter species. A phylogenetic tree constructed with single copy core genome gene sequences showed that these three isolates were part of the haemolytic clade (Nemec et al. 2016). Some species of this clade have been isolated from environmental samples, but others have been found predominantly in human clinical specimens, such as A. hemolyticus (Castro-Jaimes et al. 2020) or A. colistiniresistens (Nemec et al. 2017). The members of this clade typically show β-hemolysis halos in blood–agar media, and sometimes they can also degrade gelatin. Isolates C32I and C26G/C26M show both of those activities. These results suggest that C32I and C26G/C26M are members of two new Acinetobacter species. To deeply explore the existence of new bacterial species, more isolates must be obtained and an evaluation of their intraspecific diversity would be a highly recommended taxonomical practice. Moreover, this exploration will allow selecting the best type-strain as the representative of the new species (de Lajudie et al. 2019, Oren and Garrity 2014).
A total of seven of the eight Acinetobacter isolates described here have some degree of antifungal activity against Bd, an observation that is congruent with the prevalence of this genus in frog skin microbiomes. Previous studies have shown that the Bdstrain JEL423 (used in this study) has a high resistance to bacterial secretions, produced by single bacterium or consortia, in contrast to other Bd strains (Antwis and Harrison 2018, Harrison et al. 2020).Thus, the Acinetobacter isolates analyzed here are likely producing strong antifungal compounds, but further studies are needed to corroborate this.
The antifungal capacity of the isolates was not restricted to B. dendrobatidis since they were also capable of inhibiting the growth of B. cinerea. However, the isolates that had the strongest effects on inhibiting B. dendrobatidis were different than those inhibiting B. cinerea. The search for BGCs showed that these isolates can potentially produce a similar spectrum of antimicrobial compounds: arylpolyenes, beta-lactones, siderophores, NRPs, NAPAA, and posttranslationally modified peptide products (RiPP-like). However, the specific molecules produced by the Acinetobacter isolates could differ since more than one cluster exists for each BGC type. For example, isolates A47H and A45P both have 100% match to siderophore_I gene cluster while the rest of the isolates match sideropohore_II gene cluster (matching between 92% and 100%). Thus, these two clusters (siderophore_I and II) are likely producing different types of siderophores. Moreover, it has been shown that different bioactive molecules may be produced by the same BGC depending on the environmental conditions (Martinet et al. 2019). Thus, to fully describe the functions associated to each BGC, studies describing the compounds produced by Acinetobacter isolates are needed (Liu et al. 2007).
It is important to note that the antifungal activities of the Acinetobacter isolates are not restricted to those present in the skin microbiome. Recently, it has been reported that the isolation and characterization of Acinetobacter strains belonging to different species and isolated from distinct sources also show antifungal activities. These strains have been studied to explore their potential use as biocontrol agents. For example, A. calcoaceticus AcDB3 was shown to inhibit growth of the plant pathogen F. oxysporum (Khalil et al. 2021), and A. baumannii LCH001 was capable of inhibiting the growth of the fungi Cryphonectria parasitica,Glomerella glycines,Phytophthora capsici,Fusarium graminearum,B. cinerea, and Rhizoctonia solani (Liu et al. 2007). These observations suggest that the frog skin microbiome includes Acinetobacter strains with broad antifungal functions and not only those with antifungal capacities against B. dendrobatidis. The presence of Acinetobacter strains on the skin of frogs could be linked to the protective effect that the skin microbiota has over their hosts, defending them not only from chytridiomycosis but also from a broader spectrum of pathogenic microbes.
Conflict of interest statement. None declared.
Acknowledgements
We would like to thank Myra C. Hughey and Daniel Medina for the bacterial isolation and field work in Panamá. We also thank Ángeles Pérez-Oseguera, Martha Torres (CCG) and Guadalupe Jiménez-Flores and Gerardo Ortiz-Segura for their technical assistance as well as to Ricardo Grande, Verónica Jiménez Jacinto, and Karel J. Estrada for their help as part of the Unidad de Secuenciación Masiva y Bioinformática (UNAM). We thank Martina Maixnerová (Laboratory of Bacterial Genetics, Prague) for the phenotypic testing.
Funding
This work was supported by the Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica (IN204421 to M.A.C., IN203720 to M.S., and IA200921 to E.A.R.), the El Consejo Nacional de Ciencia y Tecnología (CF-2019/373914 to E.A.R.) and the National Science Foundation (DEB-1136602 to Reid N. Harris, DEB-1136640 to Lisa K. Belden). Postdoctoral researchers were supported by the El Consejo Nacional de Ciencia y Tecnología (1200/94/2020 and 1200/224/2021 to E.B.L. CVU_469558 and CF-2019/373914 to M.D.B. CVU_30670). Francisco González-Serrano and Yordan J. Romero-Contreras are doctoral students from the Programa de Doctorado en Ciencias Biomédicas, Universidad Nacional Autónoma de México, and are supported by the El Consejo Nacional de Ciencia y Tecnología (CVU_856429 and CVU_745733, respectively).
