Uncovering a Microbial Enigma: Isolation and Characterization of the Streamer-Generating, Iron-Oxidizing, Acidophilic Bacterium “Ferrovum myxofaciens”

Strain P3G was isolated on a modified overlay solid medium (“iFeo”) (12). This is essentially an “inorganic” medium, which contains only ferrous sulfate (25 mM) as an energy source. Acidiphilium cryptum (strain SJH) was included in the lower gel layer to remove traces of organic materials, as with most other overlay media. Plates were inoculated with disrupted acid streamers growing in a stream draining the abandoned Mynydd Parys copper mine in north Wales (Fig. 1A) (7, 13) and incubated at 20 or 30°C for up to 4 weeks. Other overlay media (Feo, FeSo, FeTo, and YE3o) (12) were also used to isolate acidophiles from the disrupted streamers.

Over 20 ferric iron-encrusted colonies, varying in size from <1 mm to ∼10 mm in diameter, which grew on iFeo medium were transferred into 200 μl of liquid medium (in 1-ml Eppendorf tubes) containing 5 mM ferrous sulfate, basal salts, and trace elements, pH 2.5 (14). Cultures in which the ferrous iron oxidized (indicated by a change in color to orange/brown and deposition of ferric iron) were transferred into small volumes (5 ml) of the same medium of progressively lower pH (to pH 2.1) to avoid hydrolysis and precipitation of ferric iron. In the absence of precipitates in the oxidized medium, small streamer-like growths were evident in some cases. Amplification of 16S rRNA genes using primers that were specific to “Ferrovum”-like isolates (see below) confirmed that the novel proteobacterium was present in many of the latter cultures, though not in those where small streamer growths were not evident. Culture purity was confirmed by terminal restriction fragment length polymorphism (T-RFLP) analysis, using restriction enzymes that separate “Ferrovum”-like bacteria from other acidophiles (13), and also by plating 50 μl of oxidized medium onto Feo medium (which supports the growth of other acidophilic iron oxidizers) and a solid medium containing 5 mM fructose and 0.02% (wt/vol) yeast extract (to test for the presence of heterotrophs, such as Acidiphilium and Acidocella spp.). One culture confirmed as pure (isolate P3G) was characterized, as described below.

Oxidation of ferrous iron by “F. myxofaciens” P3G in shake flask cultures containing 100 ml of 5 mM ferrous sulfate liquid medium (pH 2.1) and incubated at 30°C was monitored by measuring time-related changes in concentrations of ferrous iron, using the ferrozine assay (15). The ability of isolate P3G to use alternative inorganic electron donors was tested by harvesting ferrous iron-grown biomass and inoculating it into liquid medium (pH 3.0) containing elemental sulfur (0.1%, wt/vol), potassium tetrathionate (2.5 mM), or ammonium sulfate (10 mM). The higher initial pH used in these tests was because oxidation of these substrates generates acidity, in contrast to ferrous iron oxidation, which is a proton-consuming reaction. As well as monitoring any changes in pH, concentrations of sulfate (sulfur and tetrathionate cultures) and nitrite and nitrate (ammonium cultures) were measured at regular intervals, using ion chromatography. Autotrophic growth via oxidation of hydrogen was tested using a protocol described elsewhere (16). In brief, this entailed incubating “F. myxofaciens” P3G in liquid (5 mM ferrous sulfate) and solid (iFeo plates) media in a sealed jar with an atmosphere containing nitrogen, hydrogen, oxygen, and carbon dioxide. Control cultures containing only nitrogen, oxygen, and carbon dioxide were set up at the same time. Cultures were inspected after 2 and 4 weeks to look for differences in colony morphologies (switching to hydrogen causes colonies of iron oxidizers to become larger and cream/gray colored) and to determine whether there was enhanced growth in the presence of hydrogen. Finally, oxidative dissolution of pyrite (FeS₂) was tested by inoculating iron-grown “F. myxofaciens” P3G in a pH 2.5 liquid medium containing 1% (wt/vol) finely ground (<100-μm) pyrite (Strem Chemicals, Inc., MA) and monitoring changes in total soluble iron concentrations (by ion chromatography). All of the above-described tests were carried out in duplicate, and Acidithiobacillus ferrooxidans (ATCC 23270^T), which oxidizes ferrous iron, sulfur, tetrathionate, hydrogen, and pyrite, though not ammonium, was used as a positive control.

Liquid media containing either 100 μM or 5 mM ferrous sulfate, basal salts, and trace elements, adjusted to pH 2.1, were amended with the following potential organic substrates: glucose, fructose, citric acid, glutamic acid, glycerol (at 5 mM), and peptone or yeast extract (at 0.02%, wt/vol). Cultures were incubated at 30°C for 10 days, biomass was assessed visually (by comparing sizes of streamer growths), and yields were determined by measuring the protein contents of harvested cultures (17). Additionally, genes encoding the large subunit of type I (cbbL) and type II (cbbM) RubisCO were amplified and sequenced, as described below.

To assess whether isolate P3G could fix dinitrogen, a liquid medium containing 20 mM ferrous iron and nitrogen-free basal salts/trace elements was prepared and duplicate flasks were inoculated with an active culture of the iron oxidizer. The flasks were placed in a desiccator containing sulfuric acid to absorb any ammonia present in the atmosphere and incubated at 30°C. The oxidized cultures were subcultured (twice) into liquid medium of the same composition. Parallel cultures using standard inorganic N-containing basal salts were also set up. Genetic evidence for the capacity of P3G to fix nitrogen was obtained by amplifying a portion of the nifH gene, as described below.

“F. myxofaciens” P3G was grown in batch culture in a pH- and temperature-controlled 2-liter bioreactor (Electrolab, United Kingdom) with 20 mM ferrous sulfate. The reactor was aerated (1 liter min⁻¹) and stirred at 100 rpm. Concentrations of ferrous iron were determined at regular intervals, and growth rates of strain P3G at each preset condition were determined from the exponential phases identified in semilogarithmic plots of iron oxidized against time. As is the case with other acidophilic chemolithotrophic iron oxidizers such as Acidithiobacillus and Leptospirillum spp., ferrous iron oxidation was coupled to bacterial growth in actively growing cultures. To determine optimum growth temperature, the bioreactor was maintained at pH 2.4, and to determine optimum pH, the temperature was set at 25°C.

Specific rates of ferrous iron oxidation by “F. myxofaciens” P3G at different temperatures were determined using a protocol described elsewhere (18). In brief, the isolate was grown in 5-liter shake flasks on ferrous iron, and the macroscopic growths were harvested, washed, and homogenized in pH 2.2 basal salts. Aliquots of dispersed streamers were then inoculated into 10 ml of basal salts containing 1 mM ferrous sulfate in triplicate 100-ml flasks, which were incubated at 5 to 35°C. Changes in ferrous iron concentrations with time were recorded (using the ferrozine assay), and the protein concentration in the homogenized inoculum was determined using the Bradford reagent (17).

An approximate value of the water and ash contents and the relative amounts of protein and EPS produced by “F. myxofaciens” P3G was obtained using an acid streamer sample from Mynydd Parys where this bacterium accounted for >90% of the cells present (from T-RFLP analysis). One hundred grams of wet streamers was first dried at 105°C and then ignited at 450°C, to determine water and ash contents, respectively. Duplicate 1-g samples of the same acid streamer sample were dissolved in 20 ml of 0.5 M NaOH, and the protein and carbohydrate contents (as glucose equivalents) were determined using the Bradford and phenol (19) colorimetric assays, respectively. Parallel analysis of proteins and carbohydrates was carried out using harvested biomass of “F. myxofaciens” P3G. To identify the sugars present in EPS from both native acid streamers and a pure culture of “F. myxofaciens” P3G, EPS were first extracted and hydrolyzed (20) and monosaccharides were identified using ion chromatography (21).

“F. myxofaciens” P3G was grown in 5 mM ferrous sulfate liquid medium (pH 2.1) containing 0 to 100 mM ferric iron, copper, zinc, nickel, manganese(II), and aluminum or 0 to 1 mM sodium molybdate. Media containing up to 100 mM ferrous iron and no other transition metal were also prepared. Duplicate cultures were set up in 25-ml universal bottles and incubated at 30°C for up to 10 days, at which time they were tested for concentrations of residual ferrous iron.

“F. myxofaciens” P3G was grown in shake flasks by decanting oxidized medium, retaining the streamer-like growths, and replacing the medium with fresh medium containing 5 mM ferrous iron (pH 2.1). When the streamer biomass had reached an estimated 500 mg (wet weight), it was removed and washed in pH 2 basal salts solution, and DNA was extracted (22). This served as the template for amplification PCRs in which portions of the 16S rRNA gene (11) and the genes coding for the large subunit of type I (cbbL) and type II (cbbM) RubisCO (23) as well as the nifH genes (24) were amplified; all of the PCR products obtained were sequenced. Individual colonies on overlay plates were confirmed as strains of “Ferrovum” via a PCR that used a reverse primer (NitrosoR) that was specific to this iron-oxidizing acidophile (5′-TCC AGG TTA TTC GCC TGA AC-3′) in combination with the generic 27F primer (11).

The 16S rRNA gene sequence of “F. myxofaciens” P3G was analyzed, using BLAST, at the NCBI database (http://ncbi.nlm.nih.gov/BLAST) and added to a database of over 200,000 prokaryotic 16S rRNA gene sequences. The sequence was aligned together with other related sequences using ClustalX (25); gene sequence alignments were corrected manually, and alignment uncertainties were omitted in the phylogenetic analysis. Phylogenetic trees were generated with a subset of 45 nearly full-length sequences (>1,000 bp) by DNA parsimony, neighbor-joining, and maximum-likelihood analyses. In all cases, general tree topology and clusters were stable, and the reliability of the tree topologies was confirmed by bootstrap analysis using 1,000 replicate alignments. As the topologies of trees generated by all three methods were nearly identical, only the neighbor-joining tree is presented.

Newly determined sequences were deposited in GenBank under accession numbers HQ322122, HQ322123, and HM044161.

Numbers of ferric iron-encrusted colonies that grew on iFeo plates inoculated with disrupted Mynydd Parys acid streamers that appeared (from T-RFLP analysis) to be dominated by the unknown betaproteobacterium were about 1 to 3 orders of magnitude greater than those found on other overlay media. Most of these colonies were very small (<1 mm) and visible only after 2 to 3 weeks of incubation at both 20°C and 30°C. The larger (and less numerous) ferric iron-stained colonies were identified as mostly Acidithiobacillus ferrivorans (22). The specific 16S rRNA gene primers designed to identify “Ferrovum”-like bacteria were successful at differentiating between colonies of this acidophile and those of other iron-oxidizing bacteria. Many of the “Ferrovum”-like colonies that were transferred to liquid medium failed to grow, and most of those that did were subsequently found to be mixed cultures of “F. myxofaciens” and Acidiphilium spp. The latter were mostly A. cryptum SJH, which had presumably migrated from the gel underlayers, which is a common occurrence when plate incubation is protracted, as was necessary to obtain colonies of “Ferrovum.” Obtaining pure cultures of “F. myxofaciens” from liquid cultures that also contained Acidiphilium spp. was found to be extremely difficult, as numbers of planktonic-phase heterotrophic bacteria were usually far greater than those of the iron oxidizer, even when grown in ferrous sulfate medium, since most of the “F. myxofaciens” P3G cells were present in the small streamers. This meant that culture purification using serial dilution was a nonviable option. Alternative methods used to obtain pure cultures of other iron-oxidizing acidophiles, e.g., using the greater tolerance of acidithiobacilli to copper or of the leptospirilli to extreme acidity to eliminate heterotrophic contaminants (12), were also not applicable in the case of “F. myxofaciens.” Ultimately, the only approach that produced axenic cultures of “Ferrovum” was to transfer colonies from iFeo plates that were obtained directly from environmental samples into liquid medium. Due to the problems described above, very few colonies were found to be both viable and axenic, and therefore, large numbers (>50) of colonies needed to be individually processed. A pure culture of “F. myxofaciens” (strain P3G) was obtained from Mynydd Parys acid streamers, but only mixed cultures (with Acidiphilium-like heterotrophs) were obtained from other sites, including the Cae Coch mine, Wales (26); a water treatment plant at Nochten, Germany (27); and the Pyhäsalmi copper/zinc mine, Finland (C. M. Kay and D. B. Johnson, unpublished data). It was also necessary to subculture axenic cultures on a regular basis (every 5 to 7 days for active [30°C] cultures or monthly for those maintained at 4°C) to maintain their viability. Attempts to revive “F. myxofaciens” P3G from stored freeze-dried cultures, or from cultures frozen (at −70°C) in glycerol, glycine betaine, or dimethyl sulfoxide, were not successful.

In liquid medium containing ferrous iron as an electron donor, “F. myxofaciens” P3G produced small streamer-like growths in shake flask cultures, and planktonic-phase cells were also evident. Cells were nonmotile and oval in shape (1.0 ± 0.1 μm long by 0.6 ± 0.04 μm wide), and no endospores were observed. When grown in continuous culture in bioreactors, the streamers became attached to metal supports and other structures within the growth vessel and formed long filaments (28). When grown at pH >2.3, macroscopic growths of “F. myxofaciens” P3G became heavily encrusted with ferric iron precipitates as oxidation progressed, while in media containing ferrous iron poised initially at 2.1 (where ferric iron remained in solution), the gelatinous bacterial growths were free of such precipitates and off-white in color.

Electron microscopy revealed that large amounts of EPS were associated with bacterial cells in the small, streamer-like growths of “F. myxofaciens” P3G that grew in liquid media (Fig. 1B). Insufficient biomass of pure culture was available to determine accurately the water and ash contents of the small streamer-like growths produced in vitro. However, analysis of acid streamers from Mynydd Parys that were dominated by “Ferrovum”-like bacteria found that they consisted of 99% water and that the dry streamers had an ash content of 53%. Of the organic fraction of the streamer biomass, 31% was polysaccharide (glucose equivalent) and 17% was protein. Analysis of hydrolyzed EPS of both iron-grown “F. myxofaciens” P3G and Mynydd Parys acid streamers found the same monosaccharides to be present, with glucose identified as the major sugar and ribose, rhamnose, galactose and mannose present in smaller relative amounts. No acidic or basic sugar monomers were identified in hydrolyzed EPS of either native acid streamers or “F. myxofaciens” P3G streamers.

“F. myxofaciens” strain P3G was shown to be a highly specialized bacterium that grew only by using ferrous iron as an electron donor and oxygen as an electron acceptor. Repeated tests for growth on reduced sulfur, ammonium, hydrogen, and various organic carbon compounds all proved negative. The isolate was, however, able to catalyze the oxidative dissolution of pyrite, which was anticipated in view of its ability to generate ferric iron (the main oxidant of pyrite in acidic liquors). The rate of pyrite dissolution by “F. myxofaciens” P3G was much lower than that by the A. ferrooxidans type strain, and culture pH values did not fall so much (Fig. 2), probably because A. ferrooxidans (unlike “F. myxofaciens”) can oxidize reduced sulfur as well as ferrous iron. Only oxygen was used by “F. myxofaciens” as a terminal electron acceptor.

“F. myxofaciens” P3G could be maintained by subculturing it in liquid media containing only ferrous iron and basal salts; amending liquid media with different organic materials did not result in increased cell numbers and often inhibited iron oxidation, indicating that this acidophile, like many others, is an obligate autotroph. Successful amplification and sequencing of a portion of the large subunit type I (cbbL) RubisCO gene (GenBank accession number HQ322122) implied that isolate P3G fixes carbon dioxide using the Calvin-Benson-Bassham cycle, but despite varying conditions of PCR (including variations in Mg²⁺ and annealing temperature), no PCR product was obtained using primers that targeted the type II cbbM gene. “F. myxofaciens” P3G could be successfully transferred in liquid medium devoid of inorganic (and organic) nitrogen and under an ammonium-free atmosphere. This, together with the successful amplification of the nifH gene (GenBank accession number HQ322123) from DNA extracted from “F. myxofaciens” P3G, provided evidence that this novel iron-oxidizing acidophile is, like Acidithiobacillus and Leptospirillum spp., a diazotroph.

The optimum pH and temperature for growth of “F. myxofaciens” P3G were 3.0 and 32°C, respectively (Fig. 3), with lower limits of pH 2.0 and 4°C; no growth was observed at 37°C. On the basis of these data, “F. myxofaciens” P3G can be categorized as a psychrotolerant extreme acidophile. When grown under optimum conditions of temperature and pH, the culture doubling time of isolate P3G was 4 h (equivalent to a growth rate of 0.17 h⁻¹).

Specific rates of ferrous iron oxidation by “F. myxofaciens” P3G at different temperatures are shown in Fig. 4. These were slightly greater than those recorded for the iron-oxidizing acidithiobacilli at 30°C (18). At 5°C, specific rates of iron oxidation were 8% of that measured at the optimum temperature for this acidophile (30°C).

As is the case with many extreme acidophiles, “F. myxofaciens” P3G displayed tolerance to many (cationic) transition metals frequently found at elevated concentrations in mine-impacted environments, including iron (both ferrous and ferric), aluminum, manganese, copper, and zinc (Table 1). However, this acidophile was far more sensitive to nickel than were iron-oxidizing Acidithiobacillus spp. and, in common with most other acidophiles, was highly intolerant of the oxyanion molybdate.

The 16S rRNA gene sequence of “F. myxofaciens” P3G (GenBank accession number HM044161) was nearly identical (>99%) to those cloned from a variety of acidic mine water-impacted environments (Fig. 5). Sequence analysis of the 16S rRNA gene firmly placed “F. myxofaciens” P3G in the Betaproteobacteria (100% confidence using the RDP Classifier), but confidence in the assignment of isolate P3G to any lower rankings of this group was low (e.g., 54% for inclusion in the order Nitrosomonadales). The closest cultivated relatives of “F. myxofaciens” in the database are strains belonging to the ammonium-oxidizing genus Nitrosospira (93% identity). As highlighted in Fig. 5, the phylogenetic relationship (based on the 16S rRNA gene) of “Ferritrophicum radicicola” to other members of the Betaproteobacteria varied when using different algorithms to construct phylogenetic trees. While dendrograms constructed with parsimony and neighbor-joining methods suggested that “Ferritrophicum radicicola” was more closely related to the proposed order “Ferrovales,” calculations based on the maximum-likelihood algorithm placed “Ferritrophicum radicicola” in a cluster with sequences belonging to the orders “Gallionellales” and Rhodocyclales.

“F. myxofaciens” and related clone sequences form a monophyletic group within the Betaproteobacteria distantly related to classified orders (e.g., Nitrosomonadales) (Fig. 5). Based on their 16S rRNA gene sequences, environmental clones related to “F. myxofaciens” can further be subdivided into at least six monophyletic groups which might well represent further taxa within the class Betaproteobacteria. Signature nucleotides of the 16S rRNA gene sequences representing the proposed family “Ferrovaceae” and the order “Ferrovales” can be found in Escherichia coli nucleotide positions 258, 464, 559, 630, 721, 811 and 812, 817, 843, and 1109. Bootstrap values shown in Fig. 5 were relatively low for some of the nodes, indicating a lack of statistical significance of the respective branching points, which is probably due to minor differences in the primary structures of 16S rRNA genes of members of neighboring families. Also, most of the signature nucleotides are located in highly variable regions of the 16S rRNA genes, resulting in phylogenetic trees being based on regions which are subject to change. Despite that, similar branching of the orders and families in Fig. 5 occurs irrespective of the algorithm used, affirming their phylogenetic uniqueness (29).