Introduction

Careless management of agricultural and industrial activities can result in serious contamination of soils by metals (He et al. 2015; Sharma et al. 2007; Walker et al. 2003). If unchecked, this can pose significant risk to public health. Therefore, new biosafety and effective technologies intended to reduce such contamination are needed. A commonly used technology to remove metals from soil is phytoextraction: a phytoremediation method based on the application of hyperaccumulating plants that can decrease metal level in contaminated areas (Kumar et al. 1995). Hyperaccumulators are capable of sequestering extremely high levels of metals in their tissues. Although phytoextraction is an eco-friendly, low-cost method, it tends to have low efficiency because of slow growth of the plants and the low mobility and bioavailability of the metals in soil (Khan et al. 2000; Liu et al. 2020). Hence, recent years have seen growing interest in developing new phytoextraction efficiency approaches.

Recently, one promising technology for enhancing phytoextraction based on the use of plant-growth promoting bacteria (PGPB) to increase plant biomass production and tolerance to metals has been approved (Ahemad 2015; Kong and Glick 2017; Silambarasan et al. 2020). These PGPB include rhizosphere microorganisms inhabiting plant roots (PGPR) and endophytes inhabiting internal plant tissues without causing them any harm (PGPE). PGPB protect plants and promote their growth mainly by producing antibiotics, and phytohormones, and by inducing the Induced Systemic Resistance (ISR) system of the plant; they also support the dissolution of mineral nutrients, such as phosphorus or potassium, and support iron chelation (Olanrewaju et al. 2017; Gamalero and Glick 2011). PGPB can also stimulate metal uptake and bacteria resistance by various mechanisms, such as metal sorption (Kloepper et al. 1980), enzymatic reduction (Glick 2012), oxidation or extracellular precipitation via active efflux pumping (Alves et al. 2022; Bargaz et al. 2018; Kong and Glick 2017). The most commonly used PGPB species are Azospirillum, Azotobacter, Bacillus, Burkholderia, Pseudomonas or Rhizobium (Alves et al. 2022).

The use of PGPB to enhance phytoremediation may well be a common biotechnology in the near future. Various strains of PGPB have been tested. Wu et al. (2018) confirmed that the endophytic strain Buttiauxella sp. SaSR13 significantly enhanced cadmium accumulation in Sedum alfredii. Inoculation with this bacterium resulted in root elongation and, stimulated the secretion of organic acids and increased Cd uptake by S. alfredii compared to controls during a seven-day pot experiment.

Endophyte assisted phytoremediation has also been studied in Sedum alfredii by Zhang et al. (2013). The findings indicate that the tested Burkholderia sp., Sphingomonas sp., and Variovorax sp. strains significantly promoted Zn and Cd-extraction and had plant growth promoting properties. The experiment was conducted in pots for 60 days (Zhang et al. 2013). Similarly, Wang et al. (2023) revealed that inoculation of Miscanthus floridulus with an endophytic strain Bacillus cereus BL4 significantly strengthen Cd phytoremediation.

Nowadays, PGPB are commonly used in agriculture as bioinoculants. However, it is important to note that such PGPB may enhance the spread of antibiotic resistance genes (ARGs) in soil and plants because they themselves very often harbour ARGs (Chen et al. 2019; Zhang et al. 2020; Mahdi et al. 2022). Furthermore, ARGs can be located on mobile genetic elements (MGE) and they can be easily transferred among indigenous soil bacteria by horizontal gene transfer (HGT) (Arber 2014; Forsberg et al. 2012). This can represent a potential threat to public health because agricultural soil and agricultural plants act as huge reservoir and propagation hotspot of ARGs (Cadena et al. 2018; Tan et al. 2018; Forsberg et al. 2012; Zhang et al. 2015). Plants and their associated bacteria can absorb ARGs from soil and threaten human health (Zhang et al. 2011; Buchholz et al. 2011). Despite this, little research has been performed of the ARGs present in PGPB used in agriculture, and no description yet exists of the ARGs in endophytes inhabiting green parts of metallophytes.

It has been proposed that a regulatory framework is needed for new bacterial-based biofertilizers (Mahdi et al. 2022). This should include inter alia better characterization of new biofertilizers (genome mining) regarding their antibiotic resistance (AR) profile, ARG content and ARG transfer potential. Moreover, multidrug resistant strains or human pathogens should be excluded. It has also been suggested that standard criteria, regulations and quality control procedures for biofertilizer candidates should be established, so as to guarantee environmental and public health protection (Mahdi et al. 2022).

The present study describes the isolation and characterization of Armeria maritima subsp. halleri (Wallr.) Rothm. endophytes. It demonstrates that isolated Pseudomonas spp. endophytes were resistant to antibiotics and metal ions, and they harboured potential resistance genes. It also explores the possible resistance mechanisms present in the bacteria and attempts to explain the origin of the ARGs present in the isolated endophytes.

Materials and methods

Study site, sampling and soil physicochemical analysis

The studied area was located near the ZGH “Bolesław” mining and metallurgical plant in Bukowno village, in the south of Poland (50°16’40.7"N 19°28’13.8"E). ZGH “Bolesław” S.A. is a Polish company that has been operating since 1955 in Bukowno village, near Olkusz. Today, it is a modern mining and metallurgical complex, the main producer of zinc in Poland and a supplier of zinc to neighboring countries, mainly the Czech Republic, Slovakia, Austria and Hungary. In this plant zinc and lead ores are extracted and processed to produce electrolytic zinc, zinc alloys, sulfuric acid and zinc and lead concentrates.

Samples were taken during May 2015, during the flowering stage of the plants. The plant species selected for investigations was Armeria maritima subsp. halleri (Wallr.) Rothm. All collected plants were placed in polyethylene bags and transported to the laboratory in an ice cooler at 4 °C; all testing was performed within two days.

The total organic carbon, pH, calcium, magnesium, and metal content (Cr, Cu, Cd, Ni, Pb, Zn, Hg) were determined. Hg content was determined as described in DIN ISO 16,772. The other metals were tested according to the following: (ICP-OES/ICP-MS) – DIN EN ISO 11,885/DIN EN ISO 17294-2. pH was determined according to DIN EN ISO 10,390 and Total Organic Carbon (TOC) according to DIN EN ISO 15,936.

Isolation and purification of metal-tolerant bacteria

Any metal-tolerant endophytic bacteria were isolated using the Luria Bertani agar (LB) medium supplemented with filter-sterilized soluble salts of lead (CH3COO)2Pb (Pb2+) or zinc ZnSO4 (Zn2+) at a concentration of 20 mg/dm3. To isolate endophytic bacteria, the green parts of plants were separated and subjected to surface sterilization in sterile conditions under a laminar chamber (Goryluk et al. 2009). Before starting the procedure, the ends of the stem sections were secured against the inflow of sterilization agents. The first stage of sterilization was to rinse the plant fragments in 70% ethanol for about 60 s; these were then transferred to 2% mercury (II) chloride solution for 10 s and rinsed three times in distilled water. After surface sterilization, the plant material was homogenized. The obtained homogenates were diluted 10-fold and 100-fold, and 0.1 cm3 aliquots were plated on culture media. All plates were incubated at 30° C for 24–48 h. In order to determine the dry weight of the tested plants, each homogenate was poured onto a filter paper and weighed after complete drying. Based on these results, the numbers of colony forming units were then calculated per one gram of dry plant matter.

Individual bacterial colonies with different morphological characteristics were randomly selected and streaked on the LB agar medium supplemented with metal salts until pure cultures were obtained. A total of 100 bacterial isolates were selected for further studies and stored in 20% glycerol stock at -80 ºC.

Characterization of metal-tolerant bacteria

Identification

The morphological features of bacterial isolates (Gram staining) were recorded using light microscopy. Following this, biochemical analyses were performed, involved to determined oxydase and catalase activity, gelatin hydrolysis, citrate utilization, glucose fermentation and urease and fluoresceine production. All tests were prepared according to Bergey’s Manual of Systematic Bacteriology and isolates were identified to genus level (Bergey 1994).

Five out of 100 Gram-negative bacterial isolates with different morphologies were selected for further analysis. To identify the species, isolates were plated on LB agar and MALDI-TOF MS analysis was conducted by a commercial service (ALAB laboratory, Warsaw Poland). The standard Bruker interpretative criteria were applied. A score > 2.300 was used for certain species identification (Suppl. Tab. S1).

The toxic metal MIC assay

The Minimum Inhibitory Concentration (MIC) values were determined by the plate dilution method as adopted by Malik and Jaiswal (2000) with modifications. Luria Bertani LB medium supplemented with filter-sterilized soluble salts of (CH3COO)2Pb (Pb2+) and ZnSO4 (Zn2+) was used. The starting concentration for each metal was 10 mM. The inoculation was performed using 0.1 ml of bacterial suspension with a density of 106 CFU/ml. The MIC was taken as the lowest metal concentration that prevented the growth of the bacteria (Haroun et al. 2017). In this experiment E. coli 1655 strain was used as control (Spain and Alm 2003).

The antibiotic susceptibility test

Antibiotic susceptibility was determined by the disk diffusion method according to the European Committee on Antimicrobial Susceptibility Testing EUCAST version 11.0, valid from 2023-01-01. All 13 antibiotics recommended for Pseudomonas spp. were tested. Two additional antibiotics not included in EUCAST breakpoints were tested, viz. fosfomycin (50 µg) and streptomycin (25 µg), based on the presence of resistance genes detected by genome sequencing (see below). The diameter of bacterial growth inhibition zone around each of the antibiotic discs was interpreted according to the EUCAST criteria for Pseudomonas spp. If the antibiotic was not included in the standard, then a lack of any inhibition zone was interpreted as no susceptibility to the given antibiotic.

Isolation of resistance genes

The genomic DNA of the selected bacterial isolates was extracted according to Kpoda et al. (2018), and then stored at -20 °C for subsequent use. The genes coding for the efflux pump were identified using PCR amplification, while bla genes were isolated using multiplex PCR.

The efflux pump genes mexA and mex B of the Mex AB-OprM pump were amplified by PCR as described by Ugwuanyi et al. (2021). MexD, mexF and mexY genes of the MexCD-OprJ, MexEF-OprN, MexXY-OprM efflux pumps were amplified according to Poonsuk and Chuanchuen (2014). In addition, the czcA and czcR genes encoding components of the CzcCBA efflux pump were amplified using primers proposed by Perron et al. (2004). The types of ß-lactamase coding genes present were determined by multiplex PCR (Colom et al. 2003; Dallenne et al. 2010; Piotrowska et al. 2019). Four multiplex PCR assays were performed for the detection of bla genes: blaTEM, blaSHV and blaOXA genes (Multiplex I); blaCTXM genes (Multiplex II); blaVER, blaPES and blaGES genes (Multiplex III) and blaKPC, blaIMP and blaVIM genes (Multiplex IV).

All the PCR amplicons were gel-purified (Gel-Out kit, AA Biotechnology) and submitted for sequencing by a commercial service (Institute of Biochemistry and Biophysics, Polish Academy of Sciences) using ABI 3730 Genetic Analyzer, Applied Biosystems (BigDye v3.1 sequencing chemistry). Sequences of obtained gene fragments were searched against the National Center of Biotechnology Information (NCBI) using a local BLASTX program. A gene was designated as a resistanc gene if it shared at least 98% identity with other resistance gene in the database.

The whole genome sequencing and bioinformatic analysis

Genomic DNA from five selected bacteria was extracted using a Genomic Mini® kit (A&A Biotechnology) as described by the manufacturer. DNA concentration and quality were checked with the QubitTMfluorometer (Invitrogen) and bacterial genomes were sequenced by a commercial service (Institute of Biochemistry and Biophysics, Polish Academy of Sciences).

The whole genome sequencing WGS was performed on MiSeq platform (Illumina) with 300 bp paired-end reads (Supp. Tab. S4). Only high quality reads after filtering using fastp (https://github.com/OpenGene/fast) were taken for assembly step. The Unicycler version 0.4.8 assembly method was used. This Whole Genome Shotgun BioProject was deposited at DDBJ/ENA/GenBank under the accession number PRJNA886618.

Phylogenetic affiliation analysis

The genome sequences of each strain were uploaded to the Type Strain Genome Server (TYGS), i.e. a bioinformatic platform for digital, highly-reliable estimation of the relatedness of genomes based on DNA-DNA hybridization (DDH) (available at the website: http://tygs.dsmz.de) (Meier-Kolthoff et al. 2013). Additionally, a phylogenetic tree was constructed based on the RNA polymerase sigma factor RpoD (rpoD) gene (Banasiewicz et al. 2021; Girard et al. 2020). Briefly, 37 environmental-type Pseudomonas spp. strain rpoD genes were uploaded from the NCBI database (accession numbers listed in Supp. Table 2). The rpoD sequences (650 bp) were aligned using Clustal W software. The multiple sequence alignments were then used to create phylogenetic trees by the Neighbor Joining method with complete deletion of gaps, implemented in MEGA7 software (Kumar et al. 2016; Saitou and Nei 1987; Tamura et al. 2004). The evolutionary distances between sequences were computed using the Maximum Composite Likelihood method (Tamura et al. 2004), represented as the units of the number of base substitutions per site. The tree itself was drawn to scale, with branch lengths given in the same units as those of the evolutionary distances used to infer the phylogenetic tree.

Resistance genes screening

Genomes were annotated in the Rapid Annotation using Subsystem Technology (RAST) server (available at the website: http://rast.nmpdr.org) (Brettin et al. 2015). The annotation process enables the prediction of protein-coding genes, like ARG and HMRG, as well as other important elements, like direct and inverted repeats, insertion sequences, transposons and plasmids. ARGs were predicted using the Resistance Gene Identifier (RGI) application, available at the Comprehensive Antibiotic Resistance Database (CARD) (available at the website: http://card.mcmaster.ca/analyzer/rgi). In addition, Antibacterial Biocide and Metal Resistance genes Database (BacMet) was used to find resistance genes (Pal et al. 2014; available at the website: http://bacmet.biomedicine.gu.se). Finally, manual annotation was performed.

BLASTX analysis

The obtained gene fragments were searched against the National Center of Biotechnology Information (NCBI) using a local BLASTX program. A gene was designated as an ARG or MRG if it shared at least 98% identity with the best hit in the database.

Results

Physico-chemical properties of soil

At the test site, the total concentrations of Pb and Zn were 1100 mg/kg soil and 3620 mg/kg soil, respectively (Table 1). Hence, the tested soil was classified as highly contaminated (Trafas et al. 2006).

Table 1 Soil properties and metal concentrations
Full size table

Biochemical characterization of metal tolerant bacteria

The mean total count of metal-tolerant bacteria isolated from A. maritima endosphere varied from 5.73 log CFU/g of plant material on lead supplemented medium to 5.46 log CFU/g on the zinc supplemented material. All 100 isolates selected for further studies were Gram-negative, catalase positive and glucose fermentation negative. Some differences in urease and fluoresceine production were noted between selected isolates (Table 2). According to Bergey’s Manual of Determinative Bacteriology, the isolates were identified as Pseudomonas spp. MALDI-TOF-MS analysis failed to identify the isolates down to the species level (Supp. Tab. S1).

Table 2 Biochemical characteristics of endophytic metal-tolerant bacteria (positive result +, negative result -)
Full size table

The antibiotic susceptibility test, performed according to EUCAST, revealed different resistance profiles among selected isolates (Table 3, Supp. Tab. S3). All of the isolates were resistant to aztreonam (ATM), and meropenem (MEM) while four out of five isolates were resistant to ceftazidime (CAZ), cefepime (FEP) and streptomycin (S). Only two isolates showed resistance to imipenem (IPM). Regarding metal resistance, three isolates demonstrated a maximum MIC of 60 mM for Pb (II) (AM4, AM8, AM14) and one isolate a maximum MIC of 220 mM for Zn (II) (AM14; Table 3). The lowest MIC (30 mM) was observed for isolate Z18, for both metals tested.

Table 3 Antibiotic resistance profile and metal minimum inhibitory concentration (MIC) of endophytic bacteria isolated from Armeria maritima (R- resistant, S- sensitive)
Full size table

Resistance determinants detection

The tested bacteria were screened for genes encoding multidrug efflux pumps known to be common in various Pseudomonas strains, such as the Resistance Nodulation Cell Division family pump genes (RND). However, none of the Mex-type pumps genes were detected and only one of two tested CzcCBA system genes were detected in any tested strain. The PCR amplicons of the czcR gene were shorter (315 bp) than the expected czcR gene (880 bp) and sequence analysis did not confirm membership of any known resistance gene.

Regarding antibiotic resistance multiplex PCR amplify any selected variants of the bla genes.

Genome characterization of endophytic Pseudomonas sp. strains

Multidrug-resistant endophytic Pseudomonas sp. strains were sequenced on Illumina platform (AM4, AM8, AM14, Z13, Z18). The draft whole genome sequence length varied from 6.1 Mb to 7.4 Mb (Table 4, Supp. Tab. S4). Mean G + C content ranged from 60 to 61%. Annotation performed using RAST server predicted between 5,700 and 7,059 coding sequences. The analysis revealed the presence of between 374 subsystems with 65 RNA genes and 395 subsystems with 63 RNA genes. The entire Genome Shotgun project was deposited at DDBJ/ENA/GenBank under the following accessions: SAMN31135831, SAMN31136163, SAMN31136268, SAMN31137609, SAMN31137624 (Table 4). The version described in this paper is the first version of the WGS project.

Table 4 General features for Pseudomonas spp. strain draft genomes
Full size table

A whole-genome based taxonomic analysis based on DNA-DNA hybridization (in TYGS) found that isolate AM8 shared high homology with Pseudomonas marginalis species (dDDH above 80%; Supp. Tab. S5). In addition, the species demonstrating the greatest homology to the rest of the tested bacteria were, as follows: P. paracarnis (strain AM4, dDDH = 78.6%); P. koreensis (AM14, dDDH = 78.7%) and P. yamanorum (Z13, dDDH = 72.2%; Z18, dDDH = 70.7%) (Supp. Tab. S5). The rpoD phylogeny confirmed the species identification of three isolates, viz. AM4, Z13, and Z18; however, the species of AM8 and AM14 remain unclear (Fig. 1).

Fig. 1
figure 1

Neighbor-Joining phylogeny of rpoD partial gene sequences (650 bp), comprising type strains of 30 Pseudomonas species and 5 unknown Pseudomonas spp. (AM4, AM8, AM14, Z13, Z18). Escherichia coli ATCC11775 was used as an outgroup. The optimal tree with a sum of branch length = 1.73015111 is shown. The analysis involved 43 nucleotide sequences. Codon positions included were 1st + 2nd + 3rd + Noncoding. All positions containing gaps and missing data were eliminated. There were a total of 617 positions in the final dataset

Full size image

Prediction of metal resistance genes

The annotated draft genomes confirmed the presence of genes conferring resistance to arsenic, cadmium, chromium, cobalt, copper, lead, tellurium and zinc (Table 5). All tested endophytic bacteria genomes demonstrated different ars, cus and teh genes. Moreover, all the genomes carried copC, copD, copG and cueO. The Pseudomonas sp. AM8 genome lacked the cadA, cadR and czcD genes. The czcB, czcR and czcS genes were absent from the P. paracarnis AM4 and Pseudomonas sp. AM14 genomes, while chrA and copB genes were not detected in P. yamanorum species (Z13, Z18).

Table 5 Putative metal resistance genes detected in tested Pseudomonas sp. genomes (positive result +, negative result -)
Full size table

The detected genes encode components of various metal resistance mechanisms (Table 6). Gene arsB encodes an inner membrane polypeptide of the ArsAB efflux pump. ArsB confer resistance to As(III) and it provides sufficient arsenic resistance, even if the bacterium lacks ArsA (Bhattacharjee and Rosen 2007). Gene arsC encodes arsenate reductase, which reduces As(V) to As(III), and gene arsR encodes transcriptional regulator of the ars operon. Moreover, the arsH gene encodes a product that strengthens bacterial resistance to arsenate and arsenite; however, its function remains unclear (Rosen 2002).

Metal-translocating P-type ATPase genes, such as cadA, were present in the genomes of Armeria maritima endophytes. The CadA protein catalyzes the active efflux of zinc, cadmium and lead ions (Rossbach et al. 2000). Additionally, genes coding transcriptional regulators were detected in our studies, like cadC and cadR. Other P-type ATPase genes conferring resistance to copper were detected, such as copB, copC, copD and copG genes, which encode periplasmic proteins that bind and/or transport Cu ions. Genes confering other copper resistance mechanisms were also observed, such as cueO encoding multicopper oxidase, and the cusR and cusS genes encoding regulatory proteins of the CusCFBA system, involved in periplasmic copper detoxification. CusS is a sensor histidine kinase while CusR is a regulatory protein (Nies 1999).

Another resistance mechanism detected in bacteria was the chemiosmotic pump protein ChrA, a membrane transporter protein responsible for the efflux of chromium out of the cell cytoplasm. It is encoded by the chromium resistance gene chrA (Branco et al. 2008).

The CzcCBA system functions as a cation-proton antiporter by transferring Cd2+, Co2+ and Zn2+ ions out from the bacterial cell (Wang et al. 2017). The czcD gene encodes cation diffusion facilitator transporter CzcD, while czcB encodes one of the CzcCBA efflux transporter system as genes czcR and czcS encode the two-component regulatory system CzcRS. Those genes form part of the Czc system, comprising the CzcCBA transporter lying across the inner and outer membrane, regulated by the CzcRS two-component system, and the CzcD cation diffusion facilitator protein (CDF).

Additionally, the cusR and cusS genes encoding regulatory proteins of the CusCFBA system involved in periplasmic copper detoxification were detected. CusS is a sensor histidine kinase while CusR is a regulatory protein (Bondarczuk and Piotrowska-Seget 2013). The proteins of the CusCFBA efflux pump, CusA and CusB, were also detected (Bondarczuk and Piotrowska-Seget 2013).

Finally, the tehA and tehB genes, which may conferr resistance to tellurium, were detected. TehA encodes an internal membrane protein, while tehB encodes the membrane associated protein (Turner et al. 1995).

Table 6 Characterization of selected metal resistance genes
Full size table

Prediction of antibiotic resistance genes

Tested Pseudomonas genomes carried genes conferring different antibiotic resistance mechanisms (Table 7). All five tested genomes contained genes believed to encode components of efflux pumps, such as: acrA, acrB, macA, macB, mexT, pmpM, lysR, soxR, tolC. All sequenced genomes were found to contain genes encoding components of the antibiotic inactivation system, such as ampC, ampR, and those coding for MBL (metallo-ß-lactamase encoding gene) and the GNAT enzyme family. In addition, all tested genomes demonstrated point mutations in the gyrA and gyrB gene. The Pseudomonas yamanorum genomes lacked fosA and oprM genes, while those of the rest of the tested Pseudomonas spp. lacked oprN. The presence of genes believed to encode components of various Mex pumps varied among the tested Pseudomonas strains (Table 7). The following genes were detected: mexA, mexB, mexE, mexH, mexI and mexX.

Table 7 Putative antibiotic resistance genes detected in tested Pseudomonas sp. genomes (positive result +, negative result -)
Full size table

Genes coding the components of three antibiotic resistance mechanism types, viz. antibiotic inactivation, antibiotic target alteration and antibiotic efflux pumps, were identified in the tested genomes (Table 8). Five genes responsible for antibiotic inactivation were found in the tested genomes. Metallo-ß-lactamase genes (MBL), ampC and ampR genes confer resistance to ß-lactam antibiotics. Resistance to ß-lactams is largely mediated by enzymes called ß-lactamases encoded by bla genes. These enzymes are divided into four classes (A, B, C, D) and form two groups: metallo-ß-lactamases (MBL) and serine ß-lactamases. Metallo-ß-lactamases belong to Class B. Class B is divided in three subclasses: chromosomally-encoded genes of B2 or B3 subclasses, and MGE-encoded genes of the B1 subclass (Behzadi et al. 2020). In the present study, all tested genomes contained bla genes (Table 7). This result was consistent with the antibiotic resistance profile, as the tested bacteria were resistant to ß-lactam antibiotics (Table 3). A comparison of the sequences of identified genes with some other bla genes in the NCBI database confirmed that they coded for metallo-ß-lactamases. There is therefore a high probability that the bla genes belong to the chromosomally-encoded B2 or B3 subclass ß-lactamases.

Another antibiotic inactivation gene detected in the tested genomes was fosA. FosA encodes the FosA protein, and is the most common fosfomycin resistance mechanism in Gram-negative bacteria. FosA is a Mn2+, K+ dependent metalloenzyme that catalyzes the addition of glutathione to fosfomycin, thus resulting in antibiotic inactivation. In the tested genomes, the fosA gene was detected in Pseudomonas spp. AM4, AM8 and AM14 strains (Table 7). These results were in accordance with antibiotic resistance profile, as these three strains demonstrated a much narrower inhibition zone around the antibiotic disk (< 22 mm) compared to the other two strains (> 40 mm) (Supp. Table 3).

Finally, the GNAT-family gene was detected in tested genomes The gene encodes GCN5-related N-acetyltransferase (GNAT) responsible for acetylation and inactivation of aminoglycoside antibiotics (Burckhardt and Escalante-Semerena 2019).

Two genes responsible for antibiotic target alteration, gyrA and gyrB, were detected in all tested strains. Antibiotic target alteration is driven by single point mutations in gyrA and gyrB, which reduce the affinity between an antibiotic and its target. This is a very common mechanism of fluoroquinolone resistance detected in Gram-negative bacteria. Mutations in quinolone-resistance determining regions, such as gyrA or gyrB in DNA gyrase, are chromosomally encoded. Research indicates that high-level resistance to fluoroquinolones requires mutations in at least two genes with quinolone-resistance determining regions (Zhang et al. 2015). In the present study, all tested genomes contained single point mutations in the gyrA and gyrB genes. Our phenotypic antibiotic susceptibility test found the inhibition zone around fluoroquinolones (CIP, LEV) to be narrower for all strains (< 30 mm) than the susceptibility zone defined by EUCAST (> 50 mm) (Supp. Tab. S3).

The tested Pseudomonas genomes were found to encode numerous efflux pump components and regulatory protein genes. They included all genes of three efflux pumps, viz. MacAB-TolC, AcrAB-TolC and MexAB-OprM. Moreover, genes encoding components of other efflux pumps were detected: mexE and oprN of the MexEF-OprN pump, mexX and oprM of the MexXY-OprM pump, mexH and mexI of the MexGHI-OpmM pump.

The redox-sensitive protein SoxR is a global regulator of various efflux pump genes. It belongs to the MerR-family transcriptional regulators and it is common among both Gram-negative and Gram-positive bacteria. In enteric bacteria, SoxR mediates resistance to oxidative stress caused by nitric oxide or superoxide, and induces the expression of the soxS gene. SoxS activates the transcription of more than 100 genes encoding products that repair cellular damage. In nonenteric bacteria like Pseudomonas spp., SoxR directly activates the transcription of several multidrug efflux pump genes known to confer resistance to antibiotics (Park et al. 2006).

Table 8 Characterization of selected antibiotic resistance genes
Full size table

Discussion

Since the beginning of the XXI century, endophytes inhabiting metallophytic plants have gained increasing attention (El-Deeb et al. 2006; Idris et al. 2004; Stepanauskas et al. 2005; Ma et al. 2015); indeed, by the end of 2022, about 25 metallophytic plants had been tested for bacterial endophytes (Alves et al. 2022; Goryluk-Salmonowicz and Popowska 2019). All tested plants were found to harbor such endophytes, and all bacteria were resistant to high metal concentrations (He et al. 2013; Idris et al. 2004; Ma et al. 2015). Therefore, it was proposed that all metallophytic plants harbor endophytes resistant to metals. Even so, the presence and origin of ARGs in the bacteria, and their antibiotic resistance mechanisms, remain unclear. Seeing that endophytes isolated from hyperaccumulators have recently been used as beneficial biofertilizers, it is important to monitor the presence of ARGs in the genomes to prevent uncontrolled spread of ARGs in the environment.

Several studies have found endophytes inhabiting metallophytes to demonstrate antibiotic resistance. In 2005, Pseudomonas fluorescens and Microbacterium sp. endophytes isolated from Brassica napus (Stepanauskas et al. 2005) were found to be resistant to the antibiotics ampicilin, kanamycin and spectinomycin. In 2006, Enterobacter sp. endophytes resistant to ampicilin, kanamycin and tetracycline were isolated from Eichhornia crassipes (El-Deeb et al. 2006), while in 2015, a Stenotrophomonas sp. strain inhabiting Sedum plumbizincicola resistant to ampicilin, kanamycin and chloramphenicol was detected (Ma et al. 2015). While all these bacteria were found to be resistant to lead, zinc and cadmium, their antibiotic resistance genes were not researched.

In the present study, Pseudomonas spp. endophytes were isolated from the green parts of the hyperaccumulator plant Armeria maritima. The bacteria were resistant to ß-lactam antibiotics, fosfomycin, streptomycin and toxic metals, and demonstrated genes conferring possible resistance to arsenic, cadmium, chromium, cobalt, copper, lead, tellurium and zinc (Tables 5 and 6). Additionally, genes responsible for antibiotic inactivation, antibiotic target alteration, and genes coding efflux pumps were identified (Tables 7 and 8).

Bacteria are known to employ various mechanisms to provide metal resistance (Bruins et al. 2000; Ji and Silver 1995; Niño-Martínez et al. 2019), some of which were observed in the present study; for instance, some strains were found to harbor genes of the CzcCBA system, which are believed to confer resistance to cadmium, cobalt and zinc. Genes of the CzcCBA system can be detected in other Pseudomonas spp. endophytic genomes (Supp. Tab. S6). Previously, these genes were reported in the P. poae A2-S9 genome isolated from Panicum virgatum plant, the Pseudomonas sp.382 genome isolated from Paullinia cupana seeds (Xia et al. 2013, 2019; de Siqueira et al. 2018; Liotti et al. 2018) and the P. putida GM4FR genome isolated from the green parts of Festuca rubra L. (Wemheuer et al. 2016, 2017) (Supp. Tab. S6).

Interestingly, it has been proposed that the CzcRS two-component regulatory system may also be responsible for carbapenem antibiotic resistance (Wang et al. 2017). In the presence of Zn(II) ions, CzcS autophosphorylates and transmits a signal to the response regulator CzcR. CzcR up-regulates the expression of the CzcCBA efflux pump and represses the expression of the OprD porin responsible for the entry of carbapenem antibiotics (Perron et al. 2004). This is an example of a co-regulation system that act as a co-selection mechanism between the metal and antibiotic resistance mechanisms (Baker-Austin et al. 2006; Goryluk-Salmonowicz and Popowska 2019). It is possible that a co-regulation system also operates in the endophytes tested in our present study, as these were found to be resistant to both zinc ions and carbapenem antibiotics (Table 3).

Further, possible chromium resistance genes were detected, one of which codes for ChrA, a membrane potential dependent transporter. The gene has previously been detected in a variety of bacterial genera, including Arthrobacter spp., Bacillus spp., Lysinibacillus spp. and Pseudomonas spp. (He et al. 2010, 2011; Henne et al. 2009; Mondaca et al. 1998). The ChrA gene has also been found in environmental Pseudomonas genomes, like P. chlororaphis GP72 isolated from green pepper rhizosphere, P. viridiflava CDRTc14 isolated from the roots of Lepidium draba and P. fluorescens UM270 isolated from the roots of Medicago truncatula (Supp. Tab. S6) (Hernández-León et al. 2015; Samad et al. 2016; Liu et al. 2006; Shen et al. 2012). Interestingly, the chrA gene is commonly detected on plasmid or Tn DNA. In 1990, it was detected on the Pseudomonas aeruginosa plasmid pUM505, and on the Alcaligenes eutrophus pMOL28 (Cervantes et al. 1990; Nies et al. 1990). The ChrA protein was also found to be encoded by a gene detected on transposon TnOtChr from Ochrobactrum tritici (Branco et al. 2008). Similarly, genes encoding CzcCBA components can be located on mobile elements; these genes have been detected on the plasmid pMOL30 of C. metallidurans (Nies et al. 1990).

Finally, we identified genes associated with three copper resistance mechanisms (CusCFBA efflux system, P-type ATPases and multicopper oxidase CueO) and one zinc/cadmium/lead transporter (P-type ATPase CadA).Arsenic (ArsAB efflux pump), chromium (ChrA transporter) and tellurium (TehAB transporter) metalloid resistance genes were detected in all tested bacterial genomes (Table 5). These genes were also detected in other genomes of environmental Pseudomonas strains (Supp. Tab. S6).

The broad spectrum of metal tolerance demonstrated by the isolated endophytes makes them interesting candidates as bioremediation enhancing agents. However, this raises the important question of whether the presence of metal resistance genes promotes antibiotic resistance. Numerous studies have confirmed that metal-resistant environmental bacteria isolated from soil and water environments harbor antibiotic resistance genes (Barker-Reid et al. 2010; Cycoń et al. 2019; Forsberg et al. 2014; Su et al. 2020). There is a growing concern that metal contamination acts as selective agent in the spread of antibiotic resistance, especially when the metal resistance genes are located on mobile elements (Goryluk-Salmonowicz and Popowska 2019, 2022; Seiler and Berendonk 2012; Zhang et al. 2020; Baker-Austin et al. 2006; Perron et al. 2004). Therefore, the present study examined the locations of any potential resistance genes in the genomes of the tested bacteria.

The Pseudomonas spp. genomes were searched for genes encoding components of mobile genetic elements (MGE). Interestingly, numerous transposase genes were detected in the tested genomes. Transposases are needed for efficient transposition of insertion sequences (IS) or transposon DNA (Tn) (Beuzón et al. 2004). Transposases encoding genes of different Insertion Sequence (IS) families were detected, including IS5, IS66, IS110, IS200like, IS630, InsE, InsO and ISL3. However, it is unlikely that potential ARGs detected in the sequenced Pseudomonas genomes are a part of their mobilome, as the detected ARGs are commonly located on the chromosomes of other Pseudomonas spp. (CARD database). Further research is needed to confirm the presence of the detected ARGs on bacterial chromosomes. If so, the isolated endophytes can be used in laboratory experiments to evaluate their potential to promote plant growth and increase metal pollution remediation capacity. An interesting question is whether all detected potential metal resistance genes encode functional metal resistance proteins.

Conclusion

The present study demonstrated that Armeria maritima subsp. halleri (Wallr.) Rothm. is inhabited by Pseudomonas endophytes that are resistant to metals and antibiotics. Genome analysis confirmed the presence of genes conferring resistance to arsenic, cadmium, chromium, cobalt, copper, lead, tellurium and zinc. Genes encoding components of efflux pumps, extracellular sequestration proteins, P-type ATPases or metal detoxification proteins were detected. Moreover, the bacteria were resistant to antibiotics: streptomycin, fosfomycin and ß-lactams. Genes that may confer resistance to macrolides (MacAB-TolC efflux pump) and multidrug efflux pumps genes (AcrAB-TolC and MexAB-OprM) were identified, as well as some genes that may promote antibiotic inactivation and antibiotic target alteration.

The spread of antibiotic resistance genes (ARGs) in the environment is a global problem, and their main reservoir is considered to be soil. The bacteria inhabiting soil and plants are recipients of ARGs and hence form part of their transmission routes. Therefore, while the bacterial endophytes inhabiting hyperaccumulators may be beneficial for the host plant, they can also hasten the rise of antibiotic resistance.

In the available literature, little attention has been given to the problem of antibiotic resistance in bacteria used to promote plant growth in agriculture, and the resistance profiles of endophytes used in phytoremediation have not been addressed at al. Noteworthily, such biological control agents and biofertilizers form important parts of new agricultural management systems. Such growth in the number of biopreparations available on the market, and the consequent large-scale and long-term usage of biopreparations containing ARGs may further increase the dissemination of antibiotic resistance.