Biochemical Characteristics, Adhesion, and Cytotoxicity of Environmental and Clinical Isolates of Herbaspirillum spp.

Abstract

Herbaspirillum bacteria are best known as plant growth-promoting rhizobacteria but have also been recovered from clinical samples. Here, biochemical tests, matrix-assisted laser deionization–time of flight (MALDI-TOF) mass spectrometry, adherence, and cytotoxicity to eukaryotic cells were used to compare clinical and environmental isolates of Herbaspirillum spp. Discrete biochemical differences were observed between human and environmental strains. All strains adhered to HeLa cells at low densities, and cytotoxic effects were discrete, supporting the view that Herbaspirillum bacteria are opportunists with low virulence potential.

TEXT

Herbaspirillum bacteria are nonfermentative, Gram-negative, generally curved or sometimes helical bacilli (1, 2) that can be found in several environmental niches as free-living (3, 4) or plant-endophytic (5–7) organisms.

The main interest in these bacteria is due to the ability of some species to fix nitrogen (8–10) and to promote plant growth (6) through transfer of fixed nitrogen (5, 11–15) and/or phytohormone production (16). However, Herbaspirillum rubrisubalbicans is a pathogen for some sugarcane varieties (17, 18) and Herbaspirillum seropedicae strains Os34 and Os45 inhibit the growth of rice seedlings and induce a hypersensitive response in tobacco leaves (19, 20).

Herbaspirillum bacteria are rarely associated with human infections. The first reports of the isolation of Herbaspirillum from human infection sites were made in the 1980s. Later, isolates previously described as CDC group EF-1, containing mainly strains of clinical origin, were denominated Herbaspirillum species 3 (21). More recently, however, the possible role of Herbaspirillum as a human pathogen has received more attention. Herbaspirillum spp. have been recovered from the blood of patients with cystic fibrosis (22), leukemia (23, 24), and cellulitis and bacteremia (25) and from sputum (22). They have been also found in arterial walls of aortic aneurysms (26). The isolates were identified as Herbaspirillum by 16S rRNA gene sequencing, and at least three species were identified: H. seropedicae, Herbaspirillum huttiense, and Herbaspirillum frisingense (22, 25); additionally among these, two subspecies (Herbaspirillum huttiense subsp. huttiense and Herbaspirillum huttiense subsp. putei) and three distinct lineages were characterized (22, 27). All these isolates have been misidentified as Ralstonia, Burkholderia cepacia complex, Cupriavidus pauculus, or Ochrobactrum anthropi (22, 24) or could not be identified (22, 23, 25) by phenotypic identification methods used in the clinical laboratory.

There is still little information available on Herbaspirillum in the clinical microbiology literature (1, 2, 22–25), and data on possible pathogenicity-associated characteristics are lacking. Furthermore, environmental strains might be a source of infection for humans (23). Therefore, we compared the biochemical profiles, the abilities to use carbon sources, and the mass spectrometry patterns of environmental isolates (8 strains) and human isolates (7 strains) of Herbaspirillum spp. Additionally, the bacteria were tested by in vitro biological assays for adhesion, hemolytic activity, and cytotoxicity.

The clinical isolates studied were H. frisingense AU14559, H. huttiense subsp. huttiense AU11883, H. huttiense subsp. putei AU13384, H. seropedicae AU14040, and Herbaspirillum lineage 1 strain AU14775, lineage 2 strain AU13965, and lineage 3 strain AU3926 (22), and the environmental strains were H. frisingense GSF-30 (28), H. huttiense subsp. huttiense IAM 14941 (3, 4), H. huttiense subsp. putei IAM 15032 (4), H. seropedicae Z67 (8) and SmR1 (29), H. rubrisubalbicans M4 (8), Herbaspirillum lusitanum P6–12 (30), and Herbaspirillum hiltneri N3 (7). These isolates were selected to represent the Herbaspirillum species recovered from human clinical samples and their environmental counterparts, a phytopathogenic strain, and other plant-associated species not found in clinical samples.

Bacteria were cultured in tryptone soy broth (TSB) in a rotary shaker (160 rpm) for 18 h or in tryptone soy agar (TSA) (Merck, Darmstadt, Germany). Incubation was at 36 ± 1°C for clinical isolates or at 30 ± 1°C for environmental isolates, unless described otherwise. Bacteria were maintained in skim milk (31) at −20°C and NFbHP malate semisolid medium (32) at room temperature.

Conventional biochemical tests used were Gram stain; growth on TGY agar (33) at 25, 30, and 36°C for 24 h and at 42°C for 24 to 48 h; oxidase and catalase activity; motility; o-nitrophenyl-β-d-galactopyranoside (ONPG); growth on MacConkey, SS, and cetrimide agar; gelatin, urea, and esculin hydrolysis; citrate; nitrate reduction; gas from nitrate; indole production; d-glucose fermentation; d-glucose, sucrose, lactose, glycerol, trehalose, arabinose, maltose, rhamnose, mannitol, xylose, and fructose oxidation; lysine and ornithine decarboxylase; arginine dihydrolase; phenylalanine deaminase; ability to grow on NaCl (0 to 6%); polymyxin B (300 U) resistance; and hemolysis activity on sheep blood agar (1, 34, 35). Two independent assays were performed, and tests were checked up to the 7th day of incubation, except for growth on TGY agar at different temperatures.

The Biolog GN2 microplate (Biolog, Hayward, CA, USA) was used to assess carbon source usage. Bacterial suspensions with 58% transmittance at 590 nm were prepared in 0.85% sterile saline (36). Incubation was at 36 ± 1°C for human isolates and at 30 ± 1°C for environmental isolates. Three independent assays were performed, and the results were determined visually.

Nitrogenase activity of the clinical isolates of Herbaspirillum was determined through the reduction of acetylene to ethylene as measured by gas chromatography (37) in two independent assays. H. seropedicae SmR1 was used as a positive control.

Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) analyses were performed on a Bruker Autoflex II MALDI-TOF spectrometer (Bruker, Bremen, Germany) in linear positive mode with delayed ion extraction (20 kV as accelerating voltage). Spectra were obtained with an average of 1,000 laser shots (accumulation of 10 data sets of 100 shots at different spot positions), and mass-to-charge ratio (m/z) ranging from 3,000 to 20,000 was analyzed. Data were acquired using the AutoXecute tool of FlexControl software 3.0 (Bruker-Daltonics) and peak resolution in the range of 500 to 600. External calibration was performed using insulin (5,734.51 m/z), ubiquitin I (8,565.76 m/z), and cytochrome c (12,360.97 m/z). Raw data were converted into peak lists using FlexAnalysis 3.0 (Bruker-Daltonics, Bremen, Germany). Cell extracts (38, 39) were spotted on three different positions of the MALDI plate, generating at least three spectra per strain. Matrix was 10 mg/ml α-cyano-4-hydroxycinnamic acid in 50% acetonitrile-2.5% trifluoroacetic acid. The peak lists of each strain were analyzed with SPECLUST (40; available at http://bioinfo.thep.lu.se/speclust.html), to generate a consensus peak list containing only the m/z values detected in at least two of the replicates using an error window of ±5 m/z. Then, the consensus lists were compared to determine common peaks among the strains. Escherichia coli ATCC 25922 was used as an outgroup.

Hemolytic activity was determined as described by Scheffer et al. (41) with some modifications. Briefly, all strains were grown at 36°C, and after 6, 18, and 24 h of incubation, aliquots of 0.2 ml of each culture were taken, added to 0.2 ml of type O Rh⁺ human erythrocyte suspension (10⁸ cells/ml), and maintained at 36°C for 2 h. The mixtures were centrifuged, and the hemoglobin level in the supernatant was determined at 540 nm. The hemolysis is represented as a percentage of that of the erythrocyte suspension treated with TSB containing 0.1% Triton X-100. The erythrocyte suspension in TSB was the negative control. Two independent assays were performed in duplicate.

To perform the cytotoxicity and adherence assays, all Herbaspirillum strains were grown in TSB at 36°C. The cytotoxic activity of Herbaspirillum strains was tested on African green monkey kidney cells (Vero, ATCC CCL-81) and cervical adenocarcinoma cells (HeLa, ATCC CCL-2) using the MTT [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide] assay (42). Cells were maintained in Dulbecco's modified Eagle's medium (DMEM) (Gibco, Grand Island, NY, USA) containing 10% fetal bovine serum, at 36°C in a 5% CO₂ atmosphere. Briefly, cells (1 × 10⁴) were seeded into 96-well plates and incubated for 24 h and then 70 μl of cell-free culture supernatant filtrates (0.2 μm) of Herbaspirillum strains grown in TSB was added. After incubation at 36°C for 24 h, 70 μl of 0.5 mg/ml MTT (Sigma) in phosphate-buffered saline (PBS) was added to each well and the plate was incubated for a further 2 h. Then, 100 μl of 0.1 M HCl in anhydrous isopropanol was added. The absorbance was measured at 570 nm. Control cells were treated with TSB. Adherence assays were performed as described by Scaletsky et al. (43, 44) with some modifications. Briefly, 2 × 10⁵ HeLa cells were added in 24-well plates containing a coverslip on the bottom and incubated with DMEM for 24 h at 36°C under a 5% CO₂ atmosphere. After this period, cells were washed with PBS and then 0.4 ml of DMEM was added. Aliquots of 0.1 ml of Herbaspirillum suspensions with turbidity equal to a McFarland standard of 0.5 were added to HeLa cells and incubated for 3 h at 36°C. The wells were then washed three times with PBS; cells were then fixed with formaldehyde 2% for 2 min and stained with May-Grünwald stain. Tests were made in the presence or absence of 2% d-mannose added to DMEM. Two independent assays were performed in duplicate.

Statistical analyses of hemolysis and cytotoxicity assays were performed using one-way analyses of variance (ANOVAs) and the Tukey test for average comparisons via the GraphPad InStat 5.0 program.

All strains were found to be glucose-nonfermenting, Gram-negative, straight, curved, or helical bacilli. All were positive for the following tests: catalase; oxidase; urea hydrolysis; citrate; motility; growth on TGY agar at 25, 30, and 36°C; MacConkey agar and nutrient broth in the absence or presence of 1 or 2% NaCl; and production of acid from d-glucose, glycerol, arabinose, mannitol, and xylose. All strains grew on sheep blood agar without hemolysis. Most strains were resistant to polymyxin B (growth up to the margin of the disc; no inhibition zone). However, H. frisingense GSF-30, H. huttiense subsp. huttiense IAM 14941, H. seropedicae Z67, and H. lusitanum P6–12 presented inhibition zones from 10 to 11.5 mm in sheep blood agar, close to the value of 13.5 mm observed for Pseudomonas aeruginosa ATCC 27853, used as a control.

All strains were negative for indole, esculin, phenylalanine deaminase, lysine and ornithine decarboxylase, gelatin hydrolysis, and gas from nitrate. No growth was observed on SS and cetrimide agar and nutrient broth with 4, 5, and 6% NaCl. No strain produced acid from sucrose and maltose.

Tests for which variability was observed among the strains are presented in Table 1. Discrete differences were observed between environmental strains and human isolates of the same species. H. huttiense subsp. huttiense strains had distinct susceptibility to polymyxin B, the environmental strain IAM 14941 being susceptible. On the other hand, weakly positive reactions for acid from trehalose and growth at 42°C were observed only for the clinical isolate AU11883. H. huttiense subsp. putei strains were distinguished only by weak growth at 42°C observed for the clinical isolate AU13384.

TABLE 1.

Phenotypic variability among Herbaspirillum strains

Test	Result for straina
	H. huttiense subsp. huttiense		H. huttiense subsp. putei		H. seropedicae			H. frisingense		H. rubrisubalbicans M4	H. hiltneri N3	H. lusitanum P6–12	Herbaspirillum lineage 3 AU3926	Herbaspirillum lineage 2 AU13965	Herbaspirillum lineage 1 AU14775
	IAM 14941	AU11883	IAM 15032	AU13384	Z67	SmR1	AU14040	GSF-30	AU14559
Growth at 42°C	−	(+)	−	(+)	−	+	+	−	(+)	+	−	−	+	+	(+)
Nitrate reduction	−	−	−	−	+	+	+	+	+	+	−	−	+	+	−
ONPG	+	+	+	+	+	+	+	+	+	+	−	−	+	+	+
Arginine dihydrolase	+	+	+	+	+	+	+	+	+	−	−	+	−	+	+
3% NaCl	−	−	−	−	−	−	−	+	−	−	−	−	−	+	−
Lactose	±	±	−	±	−	−	−	(+)	±	−	−	−	−	−	+
Trehalose	−	(+)	−	−	−	−	±	−	−	−	−	−	−	−	−
Rhamnose	−	−	−	−	(+)	−	+	−	−	−	−	+	−	(+)	−
Fructose	±	(+)	(+)	(+)	(+)	(+)	(+)	+	(+)	+	(+)	+	+	(+)	+

^aSymbols: −, negative; +, positive, (+), weakly positive; ±, test inconclusive.

The environmental strain GSF-30 of H. frisingense was distinguished from the human isolate AU14559 by its susceptibility to polymyxin B and growth in 3% NaCl. The latter in turn presented weak growth at 42°C under the conditions tested. H. seropedicae Z67 was distinguished from the clinical isolate by its susceptibility to polymyxin B and lack of growth at 42°C, while the production of acid from rhamnose differentiated H. seropedicae strain AU14040 from SmR1, which was negative in this test.

When the results of the conventional tests were compared with the data available in texts of clinical microbiology (1, 2, 35), it was observed that these strains share some characteristics with Burkholderia cepacia complex, Burkholderia pseudomallei, Ochrobactrum, Achromobacter, Ralstonia, and Pseudomonas-like group 2, confirming the difficulty in identifying Herbaspirillum in the clinical laboratory and partially explaining erroneous identification (22). However, most of the results are compatible with Pseudomonas-like group 2 and Herbaspirillum species 3 (1).

Furthermore, all the human isolates of Herbaspirillum analyzed were unable to fix nitrogen as tested by the acetylene reduction assay, which is also a characteristic of Herbaspirillum species 3 (21). The endophytic H. seropedicae strain SmR1, used as a positive control, showed nitrogenase activity of 8 nmol of ethylene/min/mg of protein.

H. frisingense GSF-30 and H. rubrisubalbicans M4 strains also had acetylene reduction activity, whereas H. lusitanum P6–12 did not. Recently, genome sequence analysis of the latter strain showed that it does not possess nif genes (45). This result suggests that testing of nitrogenase activity, which is not part of the clinical laboratory routine, may help in distinguishing the human isolates from diazotrophic environmental strains such as H. seropedicae, H. rubrisubalbicans, and H. frisingense (21, 28, 29).

Biolog GN2 microplate tests showed that all strains analyzed tested were positive or weakly positive for 29 sources: Tween 80, l-arabinose, d-arabitol, l-fucose, d-galactose, d-glucose, d-mannitol, d-mannose, methyl pyruvate, cis-aconitic acid, citric acid, formic acid, d-galacturonic acid, d-gluconic acid, β-hydroxybutyric acid, α-ketobutyric acid, α-ketoglutaric acid, d,l-lactic acid, propionic acid, d-saccharic acid, succinic acid, bromosuccinic acid, succinamic acid, l-asparagine, l-aspartic acid, l-glutamic acid, l-proline, l-pyroglutamic acid, and glycerol.

All strains were negative for α-cyclodextrin, glycogen, N-acetyl-d-galactosamine, d-cellobiose, gentiobiose, α-d-lactose, maltose, d-melibiose, β-methyl-d-glucoside, d-psicose, d-raffinose, sucrose, d-trehalose, turanose, γ-hydroxybutyric acid, α-ketovaleric acid, sebacic acid, glycyl-l-glutamic acid, l-histidine, hydroxy-l-proline, l-ornithine, d,l-carnitine, urocanic acid, inosine, uridine, thymidine, putrescine, 2,3-butanediol, and α-d-glucose-1-phosphate. The substrates for which distinct results were found among the strains are indicated in Table 2. These results indicate that there is variability among these strains and also that, except for H. frisingense AU14559, apparently the human isolates are able to metabolize a larger number of substrates (∼30 substrates) than the environmental isolates (metabolizing ∼20 substrates). The metabolic differences observed in human isolates may be associated with their ability to colonize human hosts, since to thrive in new environments pathogens requires metabolic pathways, which are absent in their less-virulent counterparts (46).

TABLE 2.

Carbon source usage among the Herbaspirillum strains

Source	Result for straina
	H. huttiense subsp. huttiense		H. huttiense subsp. putei		H. seropedicae			H. frisingense		H. rubrisubalbicans M4	H. hiltneri N3	H. lusitanum P6–12	Herbaspirillum lineage 3 AU3926	Herbaspirillum lineage 2 AU13965	Herbaspirillum lineage 1 AU14775
	IAM 14941	AU11883	IAM 15032	AU13384	Z67	SmR1	AU14040	GSF-30	AU14559
Dextrin	−	(+)	−	(+)	−	−	(+)	−	(+)	(+)	−	−	(+)	(+)	(+)
Tween 40	(+)	(+)	−	(+)	−	−	(+)	−	(+)	(+)	−	−	(+)	(+)	(+)
N-Acetyl-d-glucosamine	(+)	+	+	+	+	+	+	+	(+)	−	(+)	(+)	+	+	+
Adonitol	(+)	+	+	+	+	+	+	+	−	(+)	−	(+)	+	+	+
i-Erythritol	+	+	+	+	−	−	−	−	(+)	(+)	−	−	+	+	−
d-Fructose	(+)	+	+	+	+	+	+	+	(+)	−	−	(+)	+	(+)	+
m-Inositol	−	−	−	−	+	+	+	−	−	−	−	−	+	+	−
Lactulose	(+)	+	−	(+)	−	−	−	(+)	−	−	−	−	−	−	(+)
l-Rhamnose	−	−	−	−	+	+	+	−	−	−	−	+	−	+	−
d-Sorbitol	+	+	+	+	+	+	+	+	+	(+)	+	−	+	+	+
Xylitol	+	+	+	+	+	+	+	+	+	(+)	+	−	+	+	+
Acetic acid	(+)	+	(+)	(+)	+	+	+	+	+	−	+	+	+	+	(+)
d-Galactonic acid lactone	(+)	(+)	−	(+)	−	(+)	(+)	(+)	(+)	−	−	(+)	(+)	(+)	+
d-Glucosaminic acid	−	+	−	+	−	(+)	+	(+)	(+)	−	(+)	+	(+)	(+)	(+)
d-Glucuronic acid	+	+	(+)	(+)	−	+	+	(+)	(+)	(+)	+	(+)	+	+	+
α-Hydroxybutyric acid	(+)	(+)	−	(+)	−	(+)	(+)	(+)	(+)	(+)	(+)	(+)	(+)	(+)	(+)
p-Hydroxy-phenylacetic acid	+	+	+	+	+	(+)	+	+	+	(+)	−	−	+	+	+
Itaconic acid	−	+	(+)	+	+	(+)	+	+	+	(+)	(+)	(+)	+	+	(+)
Malonic acid	(+)	(+)	−	(+)	(+)	(+)	(+)	(+)	(+)	−	(+)	(+)	(+)	(+)	(+)
Quinic acid	+	+	+	+	+	+	+	−	−	(+)	−	−	+	+	+
Glucuronamide	(+)	+	(+)	(+)	−	+	+	−	−	−	(+)	(+)	+	+	(+)
l-Alaninamide	(+)	(+)	−	(+)	−	(+)	+	−	−	−	(+)	(+)	(+)	(+)	(+)
d-Alanine	(+)	+	−	+	−	(+)	+	+	+	−	−	(+)	(+)	+	(+)
l-Alanine	(+)	(+)	−	+	−	(+)	(+)	+	+	−	(+)	(+)	+	(+)	(+)
l-Alanyl-glycine	(+)	(+)	−	+	−	−	(+)	−	−	−	−	−	(+)	−	(+)
Glycyl-l-aspartic acid	−	(+)	−	−	−	−	−	−	−	−	−	−	−	−	−
l-Leucine	(+)	(+)	−	+	−	(+)	(+)	(+)	(+)	−	−	(+)	(+)	(+)	(+)
l-Phenylalanine	−	−	−	−	−	−	−	−	−	−	(+)	−	−	−	−
l-Proline	+	+	+	+	+	+	+	+	+	−	+	+	+	+	+
d-Serine	−	(+)	−	−	−	−	(+)	(+)	(+)	−	−	−	(+)	(+)	−
l-Serine	−	(+)	−	(+)	−	(+)	(+)	−	−	−	−	(+)	(+)	−	(+)
l-Threonine	(+)	(+)	−	+	−	(+)	(+)	(+)	−	−	−	(+)	(+)	(+)	(+)
γ-Aminobutyric acid	+	+	(+)	+	+	+	+	+	+	(+)	−	−	+	+	+
Phenylethylamine	−	−	−	−	−	(+)	−	−	−	−	−	−	−	−	−
2-Aminoethanol	+	+	+	+	+	+	+	+	+	(+)	−	−	+	+	+
d,l-α-Glycerol phosphate	(+)	(+)	(+)	(+)	+	+	+	+	(+)	(+)	−	−	(+)	+	(+)
Glucose-6-phosphate	−	−	−	−	−	+	+	−	−	(+)	−	−	+	(+)	(+)

^aSymbols: −, negative; +, positive; (+), weakly positive.

MALDI-TOF analysis showed one peak at m/z 6,701.54 shared by all Herbaspirillum strains tested; no marker was found to distinguish the clinical and environmental strains. The dendrogram (Fig. 1) generated with the mass spectral data indicates the presence of two main groups of Herbaspirillum strains, one containing only H. hiltneri and H. lusitanum. The other strains are in 2 subgroups, one containing only H. rubrisubalbicans. Interestingly, within the second subgroup environmental and clinical isolates of H. frisingense and H. huttiense cluster together. For H. seropedicae, the plant-associated strains Z67 and SmR1 group together whereas AU14040 clusters separately with lineages 2 (AU13965) and 3 (AU3926). Furthermore, Herbaspirillum lineage 1 seems closely related to H. huttiense.

FIG 1.

Clustering of Herbaspirillum strains based on MALDI-TOF mass spectra. The dendrogram was obtained using the relative distances calculated from the MALDI-TOF patterns of Herbaspirillum strains by SPECLUST. E. coli ATCC 25922 was used as an outgroup, and the scale indicates the relative distances.

Production of toxins is a hallmark of several pathogens. However, when the Herbaspirillum strains were tested for the ability to produce hemolysin toxin, none of the strains caused lysis of human erythrocytes under the assay conditions in any of the culture periods (6, 18, or 24 h). However, it was observed that the culture filtrates of H. frisingense strains AU14559 and GSF-30, H. rubrisubalbicans (M4), H. lusitanum (P6–12), and H. seropedicae strains Z67 and SmR1 decreased markedly the number of viable HeLa cells, as assessed by the MTT assay (not shown). In contrast, the viability of Vero cells was not affected (not shown), suggesting that the cytotoxic effect of Herbaspirillum strains may be lineage specific. Strains of H. hiltneri, H. huttiense subsp. huttiense, and H. huttiense subsp. putei and the clinical isolates of H. seropedicae and Herbaspirillum lineages 1, 2, and 3 did not show cytotoxic activity against HeLa and Vero cells under the test conditions.

All the Herbaspirillum strains tested showed low adherence to HeLa cells (Fig. 2), with only about 30% of HeLa cells infected. Bacteria were observed on the cells and also on the surface of the coverslips (Fig. 2). Herbaspirillum adherence on HeLa cells was not affected by the presence of 2% d-mannose, suggesting that this monosaccharide is not involved in the cell receptor recognition. In addition, exposure of HeLa cells to Herbaspirillum induced some cytopathogenic changes such as cell retraction, cytoplasmic vacuolation, and nuclear chromatin condensation (Fig. 2). These effects were discrete compared to those caused by known pathogens such as Aeromonas and Escherichia coli pathotypes, which induced intense damage to HeLa cells (not shown). The results suggest that Herbaspirillum spp. have low virulence potential.

FIG 2.

Adherence of Herbaspirillum strains to HeLa cells. (A) Control (HeLa cells incubated with TSB only); (B) Herbaspirillum lineage 3 (AU3926); (C) H. huttiense subsp. huttiense (AU11883); (D) H. huttiense subsp. putei (AU13384); (E) Herbaspirillum lineage 2 (AU13965); (F) H. seropedicae (AU14040); (G) H. frisingense (AU14559); (H) Herbaspirillum lineage 1 (AU14775); (I) H. frisingense (GSF-30); (J) H. hiltneri N3; (K) H. huttiense subsp. huttiense (IAM 14941); (L) H. huttiense subsp. putei (IAM 15032); (M) H. rubrisubalbicans (M4); (N) H. lusitanum (P6–12); (O) H. seropedicae (SmR1); (P) H. seropedicae (Z67). Weak cytopathogenic effects can be observed: nuclear condensation (thick black arrow), cell retraction (white arrow), and perinuclear vacuolation (thin black arrow). The very wide gray arrows point to the magnifications of dotted squares.

To our knowledge, this is the first study comparing characteristics of human and environmental/endophytic strains of Herbaspirillum and the effects of interaction of these bacteria with eukaryotic cells. Only discrete phenotypic differences between human and environmental strains were observed. Furthermore, Herbaspirillum strains have low virulence in vitro, supporting their role as opportunistic human pathogens.