Occurrence of aetokthonotoxin producer in natural samples – A PCR protocol for easy detection

Abstract

Cyanobacteria are well known producers of bioactive metabolites, including harmful substances. The recently discovered “eagle killer” neurotoxin aetokthonotoxin (AETX) is produced by the epiphytic cyanobacterium Aetokthonos hydrillicola growing on invasive water thyme (Hydrilla verticillata). The biosynthetic gene cluster of AETX was previously identified from an Aetokthonos strain isolated from the J. Strom Thurmond Reservoir, Georgia, USA. Here, a PCR protocol for easy detection of AETX-producers in environmental samples of plant-cyanobacterium consortia was designed and tested. Three different loci of the AETX gene cluster were amplified to confirm the genetic potential for AETX production, along with two variable types of rRNA ITS regions to confirm the homogeneity of the produceŕs taxonomic identity. In samples of Hydrilla from three Aetokthonos-positive reservoirs and one Aetokthonos-negative lake, the PCR of all four loci provided results congruent with the Aetokthonos presence/absence detected by light and fluorescence microscopy. The production of AETX in the Aetokthonos-positive samples was confirmed using LC-MS. Intriguingly, in J. Strom Thurmond Reservoir, recently Hydrilla free, an Aetokthonos-like cyanobacterium was found growing on American water-willow (Justicia americana). Those specimens were positive for all three aet markers but contained only minute amounts of AETX. The obtained genetic information (ITS rRNA sequence) and morphology of the novel Aetokthonos distinguished it from all the Hydrilla-hosted A. hydrillicola, likely at the species level. Our results suggest that the toxigenic Aetokthonos spp. can colonize a broader array of aquatic plants, however the level of accumulation of the toxin may be driven by host-specific interactions such as the locally hyper-accumulated bromide in Hydrilla.

Graphical Abstract

1. Introduction

Cyanobacteria are an ancient group of photosynthetic prokaryotes known for their extraordinary repertoire of specialized bioactive metabolites, including potent hepato-, neuro-, and cytotoxins (Tidgewell et al., 2010). In aquatic environments, the growing incidence of planktic cyanobacterial blooms accumulating cyanotoxins has been linked to ecosystem changes induced by human activities. The symptoms of aquatic ecosystem deterioration include eutrophication and pollution (Wurtsbaugh et al., 2019), climate change (rising average temperatures and CO₂ levels) (Huisman et al., 2018), and the spread of non-native species (Sukenik et al., 2012). Altogether, these are predicted to increase the occurrence of harmful cyanobacterial species (Paerl and Barnard, 2020).

The research focus has extended to non-planktic cyanobacteria and an incredible variety of bioactive compounds have been documented in benthic and soil filamentous cyanobacteria (Costa et al., 2012, Hrouzek et al., 2016). Focusing on neurotoxins, numerous genera of freshwater benthic cyanobacteria were found to produce anatoxin-a, which is responsible for recurrent animal fatalities (Quiblier et al., 2013) as well as saxitoxins (Wood et al., 2020) and ß-methylamino-L-alanine (BMAA) whose toxic and bioaccumulation potential is well documented in marine food webs (Costa, 2016, Davis et al., 2019).

Furthermore, symbiotic (endophytic) terrestrial cyanobacteria are assumed to accumulate neurotoxic agents transferred and biomagnified along a food chain. An intriguing case is the neurodegenerative disease amyotrophic lateral sclerosis-parkinsonism dementia complex, observed in human population of Guam since the 1950s (Koerner, 1952), and tentatively linked to the neurotoxic cyanobacterial BMAA. According to the hypothesis, BMAA is transferred through a food chain starting with symbiotic cyanobacteria in cycas (Cycas micronesica) roots, through flying foxes consuming their fruits and seeds, and eventually impacting the local human population (Kurland, 1988, Cox et al., 2003).

Recently, the novel cyanobacterial neurotoxin aetokthonotoxin (AETX), a peculiar pentabrominated biindole alkaloid implicated in fatal Vacuolar Myelinopathy (VM), was discovered by our team (Breinlinger et al., 2021). This neurodegenerative disease was first recorded in 1994 during an outbreak of bald-eagle poisonings at De Gray Lake in Arkansas, USA (Thomas et al., 1998), and later confirmed in aquatic animals of multiple phyla connected through a shared food web (Birrenkott et al., 2004, Dodd et al., 2016). AETX was experimentally confirmed to be produced by the true branching heterocytous cyanobacterium Aetokthonos hydrillicola Wilde and Johansen 2014 (Breinlinger et al., 2021), growing epiphytically on the bottom side of water thyme (Hydrilla verticillata (L. f.) Royle) leaves (Wilde et al., 2014). The production of AETX is dependent on bromide (Br⁻) availability, and likely linked to its hyper-accumulation by the host plant (Breinlinger et al., 2021). However, the origin of the increased Br⁻ levels at affected localities is not well understood and may include industrial pollution such as coal power plants and bromine production from brine (Good and VanBriesen, 2017; Vainikka and Hupa, 2012) as well as herbicide treatment (Glomski et al., 2005) and other anthropogenic sources.

H. verticillata is considered strongly invasive in most of the southeastern USA. While its global range of distribution includes vast areas of the temperate and the subtropical zones on both north and south hemispheres (Rojas-Sandoval, 2022), its symbiotic relationship with A. hydrillicola was reported thus far only from artificial reservoirs in southeastern USA (Breinlinger et al., 2021). Hence, the poisonings by AETX might be a consequence of multiple combined anthropogenic stressors (pollution, herbicide treatment, introduction of non-native species), possibly with some specific conditions present in these semi-artificial environments. However, a future large-scale expansion of A. hydrillicola still cannot be excluded. Hydrillaś expansion in the US reservoirs has been monitored since the 1970s (Haller and Sutton, 1975) and mostly unsuccessful attempts to mitigate its spread are almost as old (Haller, 2014; Gerrin et al., 2022). Unfortunately, the reservoirs not only host abundant wildlife but also serve as resources of drinking water to humans. Therefore, regular monitoring of A. hydrillicola (accompanied by assessment of Br⁻ and AETX levels) is highly advisable to predict the possible threat of further VM outbreaks.

The growing concern of toxic cyanobacteria, especially in freshwater resources, has stimulated the development of standardized methods for their routine detection (Jaramillo and ÓShea, 2019). These include, in particular, analytical assessment using LC-MS techniques, accompanied by enzymatic activity assays (Gaget et al., 2017b). In parallel, the cyanotoxin producers can be detected in environmental samples using PCR and qPCR-based methods (Al-Tebrineh et al., 2012), including highly sensitive loop-mediated isothermal amplification (Ramya et al., 2018). The successful detection of cyanotoxin biosynthetic genes can provide early-warning information about the toxigenic potential of natural cyanobacterial populations even in the absence of the actual cyanotoxin production.

The present study aimed to design a simple and fast PCR protocol suitable for routine monitoring of the AETX biosynthetic gene cluster (BGC) in environmental samples of plant-cyanobacterium consortia. The cluster consisting of six genes was previously identified in two A. hydrillicola strains isolated from one of the VM outbreak sites (Breinlinger et al., 2021). For five of the genes, their predicted biosynthetic function was successfully reconstituted in vitro (Adak et al., 2022). AetA encodes a FAD-dependent halogenase, aetB a cytochrome P450, aetD a nitrile synthase, aetE a tryptophanase, and aetF another FAD-dependent tryptophan halogenase. The function of aetC remains unknown (Adak et al., 2022). The unique sequence of fine-tuned mechanism of bromination, biaryl coupling, and carbonitrile synthesis is assumed to offer sufficient specificity in its nucleotide sequence for design of a highly specific PCR protocol.

2. Material and methods

2.1. Environmental sampling

The months of October and November 2021 were selected as an appropriate sampling period based on previously reported seasonal VM occurrence in fall and winter (Wiley et al., 2007). Four artificial reservoirs in the area of occurrence of A. hydrillicola in Georgia, USA, were chosen for sampling: Covington Reservoir (COV), Longbranch Reservoir (LB), Tussahaw Reservoir (TUS) and Lake Sinclair (SIN) (Fig. 1, Table 1). At each lake, three independent sampling positions were selected (Fig. 1). The biomass of Hydrilla was collected using a custom sampling device consisting of a rake on a rope. At each position, five replicate samples of Hydrilla biomass were collected into a 3.785 L (= 1 gallon) collecting plastic bag, which altogether made fifteen individual samples per lake. In three of the sampled lakes (COV, LB and TUS) Hydrilla was expected to co-occur with A. hydrillicola, the remaining lake (SIN) was expected to host Hydrilla without A. hydrillicola based on previous observations and was therefore applied as a negative control in the present study. The fifth visited lake was J. Strom Thurmond reservoir (THU), from which the original taxonomic description of A. hydrillicola was done, and the model strain called Thurmond2011 was isolated. No Hydrilla was found in this lake, however three specimens of American water willow (Justicia americana (L.) Vahl) with macroscopically visible epiphytic cyanobacterial colonies were collected for further investigation.

Fig. 1.

Localities examined in present study. The detailed map of the collection sites of individual lakes.

Table 1.

Environmental samples investigated in the present study, with summarized results of the molecular and chemical analyses. N – serial number of the sample applied as a unique identifier throughout the present study. Peak area – AETX relative peak area in the MS analysis (normalized to sample biomass dry weight).

N	Sample ID	Positive PCR	LC-MS	Peak area (mg−1)	N	Sample ID	Positive PCR	LC-MS	Peak area (mg−1)
1	COV-H1–1	A,B,C, ITS1,2	+	3.33 × 106	33	TU-H1–3	A,B,C, ITS1,2	+	2.43 × 107
2	COV-H1–2	A,B,C, ITS1,2	+	1.35 × 107	34	TU-H1–4	A,B,C, ITS1,2	+	2.61 × 107
3	COV-H1–3	A,B,C, ITS1,2	+	1.47 × 107	35	TU-H1–5	A,B,C, ITS1,2	+	8.15 × 107
4	COV-H1–4	A,B,C, ITS1,2	+	4.04 × 106	36	TU-H2–1	A,B,C, ITS1,2	+	1.53 × 107
5	COV-H1–5	A,B,C, ITS1,2	+	1.20 × 106	37	TU-H2–2	A,B,C, ITS1,2	+	2.51 × 105
6	COV-H2–1	A,B,C, ITS1,2	+	1.39 × 107	38	TU-H2–3	A,B,C, ITS1,2	+	2.40 × 106
7	COV-H2–2	A,B,C, ITS1,2	+	5.28 × 106	39	TU-H2–4	A,B,C, ITS1,2	+	7.79 × 105
8	COV-H2–3	A,B,C, ITS1,2	+	1.49 × 107	40	TU-H2–5	A,B,C, ITS1,2	+	1.56 × 106
9	COV-H2–4	A,B,C, ITS1,2	+	9.07 × 105	41	TU-H3–1	A,B,C, ITS1,2	+	2.62 × 107
10	COV-H2–5	A,B,C, ITS1,2	+	1.97 × 107	42	TU-H3–2	A,B,C, ITS1,2	+	2.50 × 107
11	COV-H3–1	A,B,C, ITS1,2	+	6.42 × 105	43	TU-H3–3	A,B,C, ITS1,2	+	2.73 × 107
12	COV-H3–2	A,B,C, ITS1,2	+	2.11 × 106	44	TU-H3–4	A,B,C, ITS1,2	+	1.30 × 107
13	COV-H3–3	A,B,C, ITS1,2	+	2.60 × 105	45	TU-H3–5	A,B,C, ITS1,2	+	1.72 × 107
14	COV-H3–4	A,B,C, ITS1,2	-	-	46	SIN-H1–1	-	-	-
15	COV-H3–5	A,B,C, ITS1,2	+	1.93 × 105	47	SIN-H1–2	-	-	-
16	LB-H1–1	A,B,C, ITS1,2	+	6.14 × 107	48	SIN-H1–3	-	-	-
17	LB-H1–2	A,B,C, ITS1,2	+	1.59 × 108	49	SIN-H1–4	-	-	-
18	LB-H1–3	A,B,C, ITS1,2	+	4.03 × 107	50	SIN-H1–5	-	-	-
19	LB-H1–4	A,B,C, ITS1,2	+	4.58 × 107	51	SIN-H2–1	-	-	-
20	LB-H1–5	A,B,C, ITS1,2	+	3.36 × 107	52	SIN-H2–2	-	-	-
21	LB-H2–1	A,B,C, ITS1,2	+	7.50 × 107	53	SIN-H2–3	-	-	-
22	LB-H2–2	A,B,C, ITS1,2	+	5.83 × 107	54	SIN-H2–4	-	-	-
23	LB-H2–3	A,B,C, ITS1,2	+	3.30 × 107	55	SIN-H2–5	-	-	-
24	LB-H2–4	A,B,C, ITS1,2	+	4.06 × 107	56	SIN-H3–1	-	-	-
25	LB-H2–5	A,B,C, ITS1,2	+	6.52 × 107	57	SIN-H3–2	-	-	-
26	LB-H3–1	A,B,C, ITS1,2	+	1.09 × 108	58	SIN-H3–3	-	-	-
27	LB-H3–2	A,B,C, ITS1,2	+	1.03 × 107	59	SIN-H3–4	-	-	-
28	LB-H3–3	A,B,C, ITS1,2	+	5.70 × 107	60	SIN-H3–5	-	-	-
29	LB-H3–4	A,B,C, ITS1,2	+	3.44 × 107	61	THU-J1–1	A,B,C, ITS2	+	1.32 × 105
30	LB-H3–5	A,B,C, ITS1,2	+	2.80 × 107	62	THU-J1–2	A,B,C, ITS2	+	3.16 × 105
31	TU-H1–1	A,B,C, ITS1,2	+	2.04 × 107	63	THU-J1–3	A,B,C, ITS2	+	4.21 × 104
32	TU-H1–2	A,B,C, ITS1,2	+	1.11 × 107

The biomass was stored on ice and immediately transported to the laboratory. In the laboratory, a subsample for microscopic analysis and taxonomic identification was taken from each sample. Hydrilla leaves from approximately ten stems from each Hydrilla sample (and an adequate amount of Justicia specimens containing a mixture of plant parts with visible epiphytic cyanobacteria) were collected and transferred to a 50 mL whirl-pack. Subsequently, the biomass was freeze-dried and stored at −80°C.

2.2. Light microscopy observations

The presence of A. hydrillicola on the Hydrilla leaves was first confirmed by light and fluorescent microscopy as described in Wilde et al. (2014), using the fresh biomass. The detailed observation of specimens collected at THU (cyanobacterial growths on Justicia) was conducted using an Olympus BX-100 microscope equipped with DIC optics. Photographs were made using the Olympus DP-72 digital camera and CellSens software at 400 × magnification (Fig. 2). The morphology of examined specimens was compared to the original species description of A. hydrillicola (Wilde et al., 2014).

Fig. 2.

Novel Aetokthonos sp. growing epiphytically on Justicia americana. A: submerged vegetation of J. americana in J. Strom Thurmond Reservoir. B-C: Macroscopically visible colonies of Aetokthonos sp. D-E: Epifluorescent micrographs of Aetokthonos sp.; the scalebars represent 100 μm. F-K: Morphology of Aetokthonos sp. growing on rhizoids (F,I), leaves (G,J) and stems (H,K) of J. americana. The scalebars represent 50 μm. Long lateral branches are clearly visible (C,D,F,J).

2.3. Strains, cultivation, DNA extraction

Two available cultured strains of A. hydrillicola were used as positive controls containing the AETX BGC: Thurmond2011 and CCALA 1050. The first was found to be contaminated by a high number of heterotrophic bacteria, the latter was almost axenic at the time of analysis. Another set of 32 AETX-nonproducing cyanobacterial strains across multiple taxonomic orders (Strunecký et al., 2022) was compiled to test the specificity of PCR primers (Table S1). All strains were cultivated in liquid BG-11 medium (Rippka et al., 1979) in 50 mL Erlenmeyer flasks at 18°C under constant irradiance of 21 μmol m⁻² s⁻¹ photosynthetically active radiation and a 16/8 h light/dark cycle. The biomass was harvested and either dried on silica-gel for 4 days, or directly kept frozen at −80 °C prior to DNA extraction.

In total, biomass of 34 single-species cyanobacterial strains and 63 specimens of lyophilized biomass of natural samples of approximatly 35 mg of dried Hydrilla leaves (or Justicia specimens) were prepared for DNA extraction. Samples were homogenized in 2 mL Eppendorf tubes using a Retsch MM 200 mill (Retsch, Haan, Germany) and wolfram carbide beads (2 mm diameter) shaken at maximum frequency (25 s⁻¹) for 7 minutes. Subsequently, total gDNA extraction was performed on the whole mixture of homogenized plant and cyanobacterial biomass using NucleoSpin Soil Mini kit (Macherey Nagel, Düren, Germany) according to manufacturers’ instructions (including the additional glass bead-beating step), using the lysis buffer S1 and Enhancer SX options.

2.4. Primer design, PCR amplification and sequencing

The PCR protocols were developed based on the previously published BGC of AETX (aetA-F, GenBank accession no. MT225528.1) derived from the sequenced genomes of the two A. hydrillicola strains used as positive controls.

Primer design was conducted using the Primers tool in Geneious Prime 2022.1.1 software to achieve sequence specificity and optimal melting temperatures, and check for the absence of hairpin and dimer formation at working temperatures. Altogether, 18 primer combinations scattered throughout the whole BGC were tested on the set of 32 AETX-nonproducing cyanobacterial strains (Table S1) to evaluate the specificity of the individual primer pairs towards the AETX BGC. The three best performing primer pairs were selected (Table 2) based on PCR product specificity and high amplification yield, each amplifying a different locus of the BGC: the genes aetA, aetB, and the region aetC-D, respectively (Fig. 3). Specific primers amplifying the Aetokthonos rRNA ITS region were designed (Table 2) based on the data obtained from cloned rRNA operon sequences of the strain Thurmond2011 with comparison of ITS sequences of closely related taxa.

Table 2.

Primers used in present study. Tm responds to the melting temperatures predicted by Geneious Prime software. Position represents the site of primer annealing counted from the beginning of the AETX gene cluster or from the beginning of the Aetokthonos hydrillicola ITS of the type 1 rRNA operon.

Name	Sequence (5´ to 3´)	Gene	Tm	Position (bp)	Band length (bp)
AETA-F	ACACGAACCACACTTTGCCA	aetA	57.3	0813 – 0832	599
AETA-R	CGCTTTCCATTCGCTCGTG	aetA	58.8	1412 – 1394	599
AETB-F	TAAGACCATCACAGAAGCTT	aetB	54.0	2343 – 2362	608
AETB-R	TGTCAGACTCACTTAGATCG	aetB	54.0	2931 – 2950	608
AETCD-F	TCTGGGCTTGAATTTTTCTG	aetC-D	54.0	3929 – 3910	530
AETCD-R	TTTGATAGTAGTCGCTCGAA	aetC-D	54.0	4439 – 4420	530
A-ITS1-F	GACCTAACCCATCACAAAAC	ITS1	54.2	17 – 36	362
A-ITS1-R	CTTTCTAGCTTGACCACTTTA	ITS1	53.2	362 – 382	362
A-ITS2-F	TCATTACCCCAACTATTCGT	ITS2	53.7	−101 – −81	360
A-ITS2-R	CCTACACTTTCCTATCTGCT	ITS2	53.9	240 – 59	360

Fig. 3.

Gene map of the aetokthonotoxin biosynthetic gene cluster (aetA-F) and the rRNA operon of Aetokthonos hydrillicola Thurmond2011 with annealing sites of the PCR primers designed in this study.

All PCR reactions had a total volume of 25 μL and consisted of 12.5 μL of Plain PP Master Mix Combi (Top Bio, Prague, Czech Republic), 10 pmol of the forward and the reverse primers, 0.5 μL of DNA, and sterile PCR water filling up the remaining volume.

First, the integrity of DNA and content of cyanobacterial DNA was verified in the cultured strains by PCR amplification of the 16S rRNA and ITS region, using universal cyanobacterial primers VFR1 and VFR2 (Boyer et al., 2001). The applied thermocycler settings were: initial denaturation (94°C, 5 min), followed by 40 cycles of denaturation (94°C, 1 min), annealing (54°C, 1 min) and elongation (72°C, 2 min), and a terminal elongation step (72°C for 10 min). PCR product of the Aetokthonos strain Thurmond2011 was cloned according to a standard procedure (Johansen et al., 2017) to obtain both variants of the rRNA operon. These data were used as a background for design of Aetokthonos-specific ITS primers for both rRNA operon variants.

Amplifications of all three aet loci and the Aetokthonos-specific ITS were accomplished with the following thermocycler settings: initial denaturation (94°C, 5 min), 40 cycles of denaturation (94°C, 30 s), annealing (60°C, 30 s) and elongation (72°C, 40 s), and terminal elongation (72°C for 10 min). Additionally, the sensitivity of the aet BGC primers was tested on a sample set of dilution series of the DNA isolated from the almost axenic Aetokthonos strain CCALA 1050 (Fig. S1–F). ITS PCR products of Justicia (THU) specimens were also cloned (Johansen et al., 2017) to avoid neglecting minor content of A. hydrillicola or another Aetokthonos species. Raw nucleotide sequences were assembled using the de novo method in Geneious Prime, all alignments and pairwise p-distance analysis of nucleotide sequences was performed in the same software. The obtained sequences were deposited in GenBank – NCBI under the accession numbers OQ325307, OQ325311–OQ325313, and OQ338813–OQ338815, in case of identical sequences, only a single representative was uploaded.

2.5. Chemical analysis

Approximately 10 mg of lyophilized Hydrilla leaves or Justicia specimens were mechanically homogenized with a spatula in a reaction tube, resuspended in 3 mL of methanol, and subsequently treated with an ultra sonotrode (Bandelin Sonificator 250, 2 min, 90 % duty cycle) to break up cell walls. To prevent degradation, the samples were maintained on ice during sonification. Samples were placed on an overhead shaker for 0.5 h, centrifuged, and the supernatant was collected. The residues were extracted two more times (1 min sonification, 0.5 h overhead shaking) with the same volume of methanol (9 mL of methanol per 10 mg of dry biomass in total). Supernatants from respective samples were combined and dried using a vacuum centrifuge.

LC-MS analysis was carried out on a Phenomenex Kinetex C18 column (2.1 × 50 mm, 2.6 μm particle size) using a standard gradient from 5–100% (v/v) MeCN in water (+0.1 % formic acid each) in 14 min with a flow rate of 0.4 mL min⁻¹. Final conditions were maintained for an additional 6 min. ESI parameters of general experimental procedure: HRESIMS² data were acquired on a Q Exactive Plus mass spectrometer equipped with a heated ESI Interface coupled to an UltiMate 3000 HPLC system (both Thermo Fisher Scientific, Waltham, MA, USA). The following parameters were used for the data acquisition: pos. mode, ESI spray voltage 4.5 kV; neg. mode, ESI spray voltage 2.5 kV, scan range 50 – 2000 m/z.

The acquired data set was analyzed with the software Freestyle^™ (Thermo Fisher Scientific, Version 1.6). The presence or absence of AETX was determined by comparing the EIC of the m/z range 649.6 – 649.7 against an in-house AETX standard.

For semiquantitative analysis, the area under the AETX peak was determined. The obtained values were normalized to the dry weight of extracted biomass from each sample. Statistical single factorial ANOVA analysis was performed using the software GraphPad Prism (Dotmatics, Version 6.01). The alpha level was set to 0.05 and a Turkey’s multiple comparison test was used to compare the reservoirs against each other.

3. Results

The specimens of invasive Hydrilla with epiphytic cyanobacterium were found growing abundantly throughout three independent freshwater reservoirs in Georgia (COV, TUS, LB), an additional reservoir (SIN) contained Hydrilla without macroscopically visible A. hydrillicola colonies. Intriguingly, Aetokthonos sp. were found growing on the surface of native Justicia in THU, a previous VM site where Hydrilla has been recently eradicated.

3.1. Microscopic observations

The presence of A. hydrillicola on the bottom surface of Hydrilla leaves was first confirmed by light- and epifluorescent microscopy and the observations were documented. As expected, the samples from COV, TUS, LB, contained A. hydrillicola, whereas SIN was Aetokthonos-free. Three specimens of Justicia from THU hosted a diverse epiphytic cyanobacterial community on the surface of their leaves, stem and rhizomes. The microbial community was dominated by a true branching heterocytous cyanobacterium, highly reminiscent of A. hydrillicola. However, it differed from A. hydrillicola in particular morphological traits, most remarkably by its disproportionately long lateral branches, with gradually tapering and elongating cells (Fig. 2).

3.2. PCR detection of aet genes

Among the set of tested PCR primers for the AETX BGC, three pairs (Table 2) spanning the genes aetA-D (Fig. 3) were selected based on PCR product specificity and high amplification yield to be the best candidates for routine monitoring. These primers unambiguously amplified a single PCR product in positive controls as well as Aetokthonos-positive environmental Hydrilla samples (Fig. 4), and no PCR product was obtained in the array of negative control cyanobacterial strains (Table S1, Fig. S1). The primers exhibited high sensitivity, successfully amplifying the target loci at gDNA concentrations in the final extract at least as low as 5.10⁻⁴ ng.μL⁻¹ (Fig. S1–F). All the specimens from COV, TUS, LB, were detected as positive in all three tested loci of the AETX BGC, which was confirmed by sequencing of the PCR products. The same applied to the specimens of Justicia from THU, yielding PCR products with identical nucleotide sequences of the amplified aet loci to the Hydrilla samples (Fig. 4, Table 1).

Fig. 4.

Gel electrophoresis (1.5% agarose, 110 V for 45 min, 3.5 μL per sample) of PCR products obtained using primers newly designed in this study. L = ladder (GeneRuler DNA Ladder Mix, Thermo Fisher Scientific). A1 = Aetokthonos hydrillicola CCALA 1050, A2 = Aetokthonos hydrillicola Thurmond2011 applied as positive controls. C = negative control. Numbers 1–63 represent individual natural samples (Table 1). A: aetA gene amplified by primer combination AETA-F and AETA-R. B: aetB gene amplified by primer combination AETA-F and AETA-R. C: aetC – aetD region amplified by primer combination AETCD-F and AETCD-R. D: ITS region of the ribosomal operon type 2 of Aetokthonos hydrillicola amplified using primer combination A-ITS2-F and A-ITS2-R. Exact expected sizes of the particular amplified bands are in Table 2.

3.3. ITS rRNA analysis

The presence of two different rRNA operon variants and identical ITS sequences within these variants confirmed the genetic homogeneity of A. hydrillicola growing on Hydrilla across the three reservoirs.

In Justicia samples from THU, both primer combinations amplified just one of the ITS variants, a homologue of the A. hydrillicola ITS2 operon. The ITS1 primer pair product was very weakly amplified; thus, sequence verification was needed (Fig. S1–A). Therefore, both ITS1 and ITS2 PCR products of sample 63 were analyzed by sequencing a clone library. All 46 obtained sequences of THU specimens were identical to each other (with rare single nucleotide polymorphisms as an artefact of cloning) but they slightly differed from the Hydrilla-hosted A. hydrillicola sequences, found to be only 92.14 % identical (Fig. S2). No sequence with a higher identity to A. hydrillicola was detected among the clones. The results indicated the presence of a unique, although closely related, genotype.

3.4. Analysis of AETX content

LC-MS analysis confirmed the presence of AETX in the biomass of the plant-cyanobacteria samples from all four water bodies (COV, TUS, LB, THU) containing Aetokthonos (Table 1), whereas the toxin was not detected in samples from the negative control lake (SIN). Only a single sample of all forty-five Aetokthonos-positive and PCR-positive samples was detected as AETX-negative (COV-H3–4). As the semi-quantitative analysis showed a relatively moderate presence of AETX across the samples obtained from lakes COV and TUS (Fig. S3), concentration of the toxin in this sample was perhaps under the detection limit. In the lake LB, LC-MS analysis indicated approximately ten-fold higher amounts of AETX compared to COV and TUS (Fig. S3). Interestingly, in the samples of Justicia hosting a cyanobacterial community including the Aetokthonos morphotype and PCR-positive for aet genes, relatively low AETX production was detected (Fig. S3–4).

4. Discussion

The recently discovered cyanotoxin AETX, responsible for the fatal neurodegenerative animal disease VM, represents a significant concern for wildlife protection and possibly public health (Breinlinger et al., 2021). However, reliable molecular methods to monitor its producer have not yet been developed. Following the results of this study, a set of three pairs of reliable and highly sensitive PCR primers are provided (Table 2) to amplify the biosynthetic genes of AETX from environmental samples. The PCR protocols were found to be specific enough to selectively amplify the AETX loci in the presence of excess plant biomass (when whole leaves with the cyanobacterial biofilm are processed) as well as in a mixed cyanobacterial and microbial community growing on the various plant parts in case of THU samples of Justicia (Fig. 2). The PCR tests were negative in a set of AETX-nonproducing cyanobacterial strains across various phylogenetic lineages (Table S1). Furthermore, the sensitivity of the PCR primers (up to 5.10⁻⁴ ng.L⁻¹) was far below typical DNA extraction yields from environmental samples (Gaget et al., 2017a), even when assuming that cyanobacterial DNA is less abundant compared to the excess plant DNA (Fig. S1–F). Therefore, we suggest that the presented method is simple and accurate enough to be directly applied in routine field monitoring by researchers as well as environmental protection agencies and water management authorities.

The sequence analysis of the obtained PCR products demonstrated the genetic uniformity of A. hydrillicola growing on Hydrilla at three independent reservoirs, and also two cultured A. hydrillicola strains. The sequences were identical in all amplified regions, including four AETX biosynthetic genes and two paralogues of the rRNA ITS. Thus, it is reasonable to expect that the aet genes are likely going to be specifically and reliably amplified at other potential VM sites in the USA.

Detection of AETX in the sampled biomass, as provided by LC-MS analysis, was in good correlation with the PCR results – all PCR-negative samples were AETX-negative, and all but one PCR-positive samples contained detectable amounts of AETX (Table 1). As a side product of testing the new primers, the presence of the toxigenic genotype and production of AETX at three Hydrilla sites previously not studied by molecular or analytical methods (COV, LB, TUS) was discovered. Thus, PCR screening of plant-cyanobacterial associations for aet genes may serve as a good predictor for the occurrence of VM in the future, especially if concentrations of available Br⁻ in the water column, sediment, and the host plant tissue are determined simultaneously (Breinlinger et al., 2021). VM was observed in LB previously (Wilde at al., 2014), therefore it is not surprising that aet genes were successfully detected here (Fig. 4) and relatively high concentrations of AETX in the Hydrilla/ Aetokthonos samples were confirmed (Fig. S3–4).

Molecular detection of cyanotoxin producers using (q)PCR has been widely applied to identify potentially toxigenic species at low concentrations in natural samples, and even under toxin non-producing conditions (Gaget et al. 2017b). At THU, Hydrilla was recently completely eradicated by a successful biomanipulation strategy using herbivorous sterile grass carp (Ctenopharyngodon idella) (Gerrin at al. 2022). Remarkably, in the current study, cyanobacteria morphologically similar to A. hydrillicola were found growing as a component of a cyanobacterial community colonizing the surface of Justicia (Fig. 2).

This putatively novel Aetokthonos material was very similar in the creeping part of the thallus to A. hydrillicola but additionally displayed long erect branches with elongated cells (Fig. 2). Even though true-branched cyanobacteria are known to exhibit extreme morphological plasticity (during their life cycle or in response to environmental conditions), hindering their taxonomic interpretation (Casamatta et al., 2020), the morphological differences detected here could be significant enough to establish a novel species. In the absence of a cultured strain of the Justicia’s epibiont, it was difficult to conclusively assess whether it represents a novel species or just a morphotype of A. hydrillicola growing on a different host plant. However, the absence of the type 1 rRNA operon (or a sequence difference sufficient to avoid amplification by the Aetokthonos specific ITS PCR protocol) may indicate substantial genetic diversification. Moreover, the results of sequencing the rRNA ITS2 locus of Aetokthonos sp. from Justicia revealed a genotype related to the canonical A. hydrillicola rRNA ITS2 sequence but exhibiting a pairwise p-distance of 7.86 % (Fig. S2). In cyanobacterial taxonomy, such a level of diversity in the homologous ITS regions has recently been considered as an indicator of variability among two different species (Osorio-Santos et al., 2014). Notably, in the Scytonemataceae family that includes Aetokthonos (Strunecký et al., 2022), rare examples of high intragenomic rRNA divergence, content of particular operons (and their total number) were previously documented (Johansen et al., 2017).

Irrespective of the taxonomic questions, Aetokthonos growing on Justicia contained all the monitored AETX loci in our PCR analysis, exhibiting identical nucleotide sequences. It is worrying that the toxigenic Aetokthonos genotype can colonize other host plants than Hydrilla, including the indigenous aquatic flora of the USA. Since Aetokthonos was never reported or sequenced outside the USA (Breinlinger et al., 2021), it is likely a native cyanobacterium that existed here before the introduction of Hydrilla, and we should not be surprised to find members of this genus on other submersed substrates in the future.

Importantly, AETX production was found to be very low in the Justicia specimens colonized by Aetokthonos sp. (Table 1, Fig. S3–4). The cyanobacterium was relatively less abundant in the samples of Justicia plant compared to Hydrilla samples, furthermore, the PCR amplification was less pronounced (Fig. 4). Another cause for the low measured AETX concentrations might be that the biosynthesis of AETX is dependent on sufficient supply of dissolved Br⁻. Although THU is one of the previous VM sites (Wilde et al., 2014), and Br⁻ is expected to be available there, the possible explanation may lie in different physiological traits of the host plants. It was shown that Hydrilla can hyper-accumulate Br⁻ from the sediment (Breinlinger et al., 2021), and the locally increased Br⁻ levels on the surface of its leaves may therefore be the trigger for AETX production. To test this hypothesis, following studies should further focus on quantification of Br⁻ in water, sediments, and biomass of the host plants (including Justicia). However, we can already suggest that elimination of bromide pollution likely represents the most direct and suitable prevention against VM.

Data availability

Sequencing data from this study is available at the NCBI under accession numbers OQ325307, OQ325311–OQ325313 and OQ338813–OQ338815.

Data availability

The sequencing data are available online as explained and other data are available on request.