Ecophysiology and geochemistry of microbial arsenic oxidation within a high arsenic, circumneutral hot spring system of the Alvord Desert


  • Present Address: Stephanie A. Connon, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA.

  • Editor: Max Häggblom

Correspondence: Timothy S. Magnuson, Biological Sciences, Idaho State University, 650 Memorial Dr, Pocatello ID 83209, USA. Tel.: +208 282 5014; fax: +208 282 4570; e-mail:


Microbial metabolism of arsenic has gained considerable interest, due to the potential of microorganisms to drive arsenic cycling and significantly influence the geochemistry of naturally arsenic-rich or anthropogenically arsenic-polluted environments. Alvord Hot Spring in southeastern Oregon is a circumneutral hot spring with an average arsenic concentration of 4.5 mg L−1 (60 μM). Hydrogeochemical analyses indicated significant arsenite oxidation, increased pH and decreased temperature along the stream channels flowing into Alvord Hot Spring. The dynamic range of pH and temperature over the length of three stream channels were 6.76–7.06 and 69.5–78.2 °C, respectively. Biofilm samples showed As(III) oxidation ex situ. 16S rRNA gene studies of sparse upstream biofilm indicated a dominance of bacteria related to Sulfurihydrogenibium, Thermus, and Thermocrinis. The lush downstream biofilm community included these same three groups but was more diverse with sequences related to uncultured OP10 bacterial phylum, uncultured Bacteroidetes, and an uncultured clade. Isolation of an arsenite oxidizer was conducted with artificial hot spring medium and yielded the isolate A03C, which is closely related to Thermus aquaticus based on 16S rRNA gene analysis. Thus, this study demonstrated the bacterial diversity along geochemical gradients of temperature, pH and As(III): As(V), and provided evidence of microbial arsenite oxidation within the Alvord Hot Spring system.


Oxidoreduction reactions of arsenic are a common biogeochemical occurrence found in various terrestrial, groundwater, and hydrothermal environments. Arsenic is especially prevalent in hydrothermal waters which are often rich in various species of arsenic, including arsenate (inline image, pH 7) and arsenite (H3AsO3, pH 7). While arsenic concentrations in geothermal waters typically range from 1–10 mg L−1(Christensen et al., 1983; Stauffer & Thompson, 1984; Ballantyne & Moore, 1988; Sakamoto et al., 1988; Criaud & Fouillac, 1989; Tanaka, 1990; Yokoyama et al., 1993; Wilkie & Hering, 1998; Koch et al., 1999; Langner et al., 2001; Arnórsson, 2003; Aiuppa et al., 2006), concentrations have been reported to be as high as 150 mg L−1 in Yellowstone National park, USA (Langner et al., 2001) and 50 mg L−1 in the El Tatio geothermal field, Chile (Ellis & Mahon, 1977; Romero et al., 2003). Arsenite, As(III), can be the sole or primary arsenic species in hot anaerobic source waters where it is rapidly converted to arsenate, As(V), due to microbial oxidations (Jackson et al., 2001; Langner et al., 2001).

Microbial oxidation of As(III) to As(V) was first reported in 1918 (Green, 1918). Since this time, many arsenite oxidizing organisms have been discovered and a wide variety of metabolisms linked to arsenite oxidation have been reported. Several heterotrophic strains have been shown to oxidize arsenite without gaining energy from this reaction, which is often considered to be a detoxification mechanism of the cell due to the far more toxic nature of arsenite compared with arsenate (Anderson et al., 1992; Gihring & Banfield, 2001; Gihring et al., 2001; Salmassi et al., 2002, 2006; Santini et al., 2002). Arsenite oxidizing strains have also been isolated that derive energy from the oxidation of arsenite to arsenate including aerobic chemoautotrophic isolates (Ilyaletdinov & Abdrashitova, 1981; Santini et al., 2000, 2002; Donahoe-Christiansen et al., 2004), anaerobic chemoautotrophic isolates (Oremland et al., 2002; Aguiar et al., 2004; Rhine et al., 2006), and an aerobic heterotrophic strain (vanden Hoven & Santini, 2004). Enzyme systems and corresponding genes have been identified in a number of arsenic-oxidizers, which has led to a better understanding of the process, at least in pure cultures (Santini et al., 2000, 2002; Santini & vanden Hoven, 2004). The evolution of these genes/enzymes has been investigated, and it is surmised that energy-generating and detoxification arsenic transformations have long been utilized by microorganisms (Lebrun et al., 2003). Published field and laboratory studies in other systems such as acid hot springs in Yellowstone National Park (Jackson, 2001; Donahoe-Christiansen, 2004) and circumneutral Hot Creek, CA (Salmassi, 2006) have shed light on microbial arsenic transformation processes, and provided a framework for the present study. The primary motivation of our work was to determine the ecophysiology and biogeochemistry of arsenic oxidation in a very distinct (with regards to pH, temperature, and dissolved arsenic) unstudied system, and compare our results with findings in other arsenic-rich ecosystems.

Our research was carried out in the Alvord Basin hydrothermal system, which consists of more than 200 individual features. Three main hydrothermal areas are found in the Alvord Basin: Mickey, Alvord and Borax Hot Springs. All three hydrothermal areas are of circumneutral pH and have significant arsenic content, making them an excellent natural laboratory for the study of microbially-mediated arsenic transformations in a neutral-pH, high-temperature environment. We chose Alvord Hot Spring for more detailed study, due to its high levels of arsenic compared with other springs in the basin. This feature has numerous source vents, seeps and springs along an exposed fault scarp, which all flow into a common pool and then out into a single channel that ultimately disappears into a dry lake bed. We have focused our research efforts on three of the source springs that flow into the common pool. Aqueous geochemical data (Koski & Wood, 2004) indicate that Alvord spring waters have significant concentrations of arsenic (4.5 mg L−1; 57.0 μM), boron (27.5 mg L−1; 2.5 mM), sulfate (227 mg L−1; 2.4 mM), chloride (858 mg L−1; 24.2 mM), sodium (1240 mg L−1; 53.9 mM), and bicarbonate (1258 mg L−1; 20.6 mM). Primary potential electron donors detected in this system include organic carbon, hydrogen, arsenite, and sulfide, while potential electron acceptors include sulfate, arsenate, and oxygen. Visually, source spring biofilms are colorless and of low overall biomass, which transition to luxuriant orange biofilm mats (significantly more biomass) within 0.5 m from the source. These observations led to the hypothesis that arsenic transformation detected in the aqueous phase of these hot circumneutral outflow channels is microbially-mediated, and that combined biogeochemical, biochemical, and molecular analyses would provide data to support this hypothesis. In order to test the hypothesis, we investigated the microbial communities involved in the arsenite oxidation in this thermal ecosystem with three different approaches; (1) Geochemical analyses to determine the chemical speciation, phase association, and quantity of inorganic arsenic in the spring water and sediments, (2) Molecular ecological studies to determine the diversity of uncultured bacteria that might be involved in arsenite oxidation, and (3) Isolation and physiologic characterization of an arsenite oxidizer with artificial hot spring minimal medium.

Materials and methods

Study site

Alvord Hot Spring is c. 30 miles north of Fields, Oregon, along the shores of Alvord Dry Lake. Three source spring channels, A1, A2 and A3 were chosen for detailed analysis (Fig. 1). Each channel was between 100 and 250 cm long, and each showed the characteristic colorless low biomass upstream biofilm and luxuriant orange downstream biofilm. The upstream biofilm (UB) samples consisted of black pebbles at the hot spring's source that were coated with a thin colorless biofilm, and the downstream biofilm (DB) samples consisted of thick orange mats. Detailed information about Alvord Hot Spring can be found at

Figure 1.

 Spring A3. The distinct colorless and orange biofilm areas are evident. Hydrothermal waters emerge from the small pool at lower left, and flow down the channel to a large pond (top of image). Scale bar is c. 50 cm.

In situ temperature and pH

Temperature and pH were measured using an OAKTON pH 6 Acorn Series pH meter with temperature probe that automatically corrects the pH for temperature effects (OAKTON Instruments, Vernon Hills, IL). An amber glass pH probe was attached to the meter (Cole Parmer, Vernon Hills, IL). Triplicate measurements were taken at each sampling point.

Arsenic species characterization

Water samples were collected using serological pipettes at 25 cm intervals along the outflow channel. Samples were then filtered using a 0.2 μm cellulose syringe filter (Corning Inc., Corning, NY) and acidified by adding Optima HCl (Fisher Scientific, Pittsburg, PA) to 4% final concentration. Samples were kept in the dark at 4 °C until analyzed at the University of Idaho using a Varian model VGA-77 hydride generator in conjunction with a Perkin Elmer Optima 3000XL inductively coupled plasma-atomic emission spectrometer (HG-ICP-AES) using previously published methods (Müller, 1999). In the presence of sodium borohydride (NaBH4) and at acidic pH, As(III) forms a volatile hydride complex much more readily than As(V). Therefore, the detection of As(III) is possible at low NaBH4 concentrations with minimal interference (<2%) from As(V). The sample, NaBH4, and 10 M HCl are simultaneously pumped to the hydride generator, through a gas/liquid separator, and the resulting hydride is carried to the ICP-AES via argon gas, limiting possible matrix effects. As(III) concentrations are measured using 13.2 mmol L−1 NaBH4. Samples were analyzed without any additional preparation other than filtration and preservation at the time of sample collection. Total arsenic concentrations were measured using 158.4 mmol L−1 NaBH4 after total arsenic reduction by the addition of 20 mmol L−1l-cysteine. As(V) concentrations were calculated as the concentration of total arsenic minus the concentration of As(III). Organic arsenic species are not considered in the analysis. Total accuracy was c. 2% for As(III) determination and c. 6% for the determination of total arsenic.

X-ray diffraction and fourier transform infrared spectroscopy (FT-IR)

X-ray diffraction patterns of pulverized streambed samples covered with biofilm were collected on an X2 Advanced Diffraction System (Scintag Inc.) employing CuKα radiation between 5 and 90° 2θ with a step size of 0.01° and a count time of 6 s. Fourier transform Infrared spectra were collected in transmission mode to avoid Reststrahlen scattering. Typically 1 mg of dried material was mixed with 100 mg of spectroscopic grade KBr powder (Spectra-Tech Inc., Shelton, CT) and pressed into a 13 mm pellet. Pellets were stored for 24 h in a drying oven before spectral collection on an FT-IR Magna 860 spectrometer (Thermo-Nicolet) equipped with a liquid-N2 cooled Ge detector and KBr beam splitter. Spectra were collected with a spectral resolution of 2 cm−1.

Molecular determination of bacterial communities in hot spring biofilms

DNA purification

Total microbial DNA was purified from microbial cells collected from channel A3 UB, which consisted of black pebbles at the hot spring's source coated with a thin colorless biofilm, and the DB, which consisted of thick orange mats in the downstream channel. Channel A3 was randomly chosen. Samples were aseptically collected with sterile forceps into sterile 50-mL centrifuge tubes and placed immediately on dry ice for transport into storage at −80 °C. The DNA was purified from c. 1 g samples using the UltraClean Soil DNA Kit (MoBio, Solana Beach, CA) according to the manufacturer's directions.

Clone library construction and restriction fragment length polymorphism (RFLP) screening

In order to examine bacterial diversity present in the hot spring, 16S rRNA gene clone libraries were made from channel A3 UB and DB. The PCR reaction mixture contained 0.025 U Taq μL−1 of reaction and 1 × PCR buffer (MBI Fermentas, Hanover, MD), 5% acetamide, 3.0 mM Mg2+, 220 μM dNTP (MBI Fermentas), and 200 nM of each primer 8F (5′-AGA GTT TGA TCM TGG CTC AG-3′) and 1492R (5′-GGY TAC CTT GTT ACG ACT T-3′), taken directly or modified from (Lane, 1991). The amplification conditions were 35 cycles at 94 °C denaturation for 30 s, 55 °C annealing for 1 min and 72 °C extension for 2 min with a final extension of 10 min. Amplification products were cloned using the TOPO TA Cloning Kit with TOP10 Chemically Competent Escherichia coli according to the manufacturer's instructions (Invitrogen, Carlsbad, CA). Each library had 96 clones, which were screened by the restriction enzymes HaeIII and Bsh1236I (MBI Fermentas) in two separate reactions. Each reaction consisted of 0.25 μL of enzyme, 5 μL of unpurified PCR reaction, 3.75 μL water and 1 μL 0.1 M MgCl2 incubated at 37 °C for 2 h. All 96 corresponding digestions from each library were run together on a 3% Nusieve 3 : 1 agarose (Cambrex, East Rutherford, NJ) large format gel with ‘100-bp Gene Ladder’ (MBI Fermentas) size marker. Each library was screened independently of the other and digestion patterns were screened separately for each enzyme for a total of four analyses. Both patterns, HaeIII and Bsh1236I, were taken together to determine the uniqueness of a clone. All clones, except one from the DB library, had an expected insert size of c. 1500 bp and thus the final libraries were 96 and 95 clones for UB and DB, respectively. Rarefaction calculations were performed on the RFLP pattern diversity using the program analytic rarefaction 1.3 ( This program uses the analytical approximation algorithm of (Hurlbert, 1971) with reformulations by (Tipper, 1979), and estimates 95% confidence intervals as described by (Heck et al., 1975).

Clone library sequencing

The most commonly occurring clones, those with RFLP patterns that occurred more than once in the library, were selected for sequencing. The entire clone insert was sequenced with the majority of the sequence receiving 2 × coverage; small regions received 1 × , 3 × or 4 × coverage. Sequencing primers included primers specific for the Invitrogen TOPO TA vector flanking regions M13F, M13R, T3 and T7 as well as 16S rRNA gene specific internal primers, including regions based on E. coli numbering that included 338, 515, 700, 1100 and 1392. All sequencing was done on an ABI 3100 automated sequencer (Applied Biosystems, Foster City, CA) at Idaho State University's Molecular Research Core Facility. Sequences were assembled using vector nti software (Invitrogen, Carlsbad, CA). Chimeric sequences were screened out of the libraries using two methods. First, independent phylogenetic trees were constructed with separate 5′ and 3′ halves of each sequence, and if the two halves branched in different locations in the tree they were considered to be chimeric sequences. Second, the sequences were analyzed using the RDP check_chimera program accessed from website supported by Michigan State University (Maidak et al., 2001). However, check_chimera is of limited use to detect chimeras of sequences that have no close homologs in the reference library.

Phylogenetic analysis

Sequences were aligned and masked in arb (Ludwig et al., 2004). Only unambiguously alignable nucleotide positions were kept for subsequent phylogenetic analysis as described by (Lane, 1991). Phylogenetic analyses were performed using paup* 4.0 β 10 (Swofford, 2002) and included 884 nucleotide positions. The tree topology was inferred by neighbor-joining using the Jukes and Cantor model to estimate evolutionary distances. Bootstrap proportions were determined from 1000 resamplings. Three archaeal sequences, Sulfolobus (D14876), Methanococcus (M36507) and clone pJP27 (L25852) were used as outgroups. The percent similarity of unmasked sequences was determined using the distance matrix tool in arb.

Isolation and characterization of an arsenite-oxidizing bacterium

Isolation of arsenite-oxidizing bacteria from downstream orange biofilm was conducted with a synthetic hot spring medium. A basal solution was first developed to mimic the aqueous geochemistry of the hydrothermal waters to which electron acceptors and donors could be added. The basal solution was made by combining three separate solutions (A, B and C). Solution A (pH 7.0) contained (per L) 80.1 mg H3BO3, 341.0 mg NaCl, 100.0 mg NH4Cl, 54.7 mg KCl, 6.7 mg KH2PO4, 3.0 mg MgCl2 7H2O, 2.6 g HEPES (to control pH), 10 mL Wolfe's vitamin solution, and 10 mL Wolfe's modified mineral (Wolin et al., 1963) solution. Solution B contained 55 g L−1 NaHCO3 and solution C contained 5.3 g L−1 CaCl2. Each solution was autoclaved separately before combining. Ten milliliters each of solution B and C were then added to 1 L of solution A. To this basal solution was added yeast extract (0.1%) as a source of carbon, and sodium arsenite, As(III), to 5 mM final concentration. Aerobic tubes of 10 mL of this synthetic hot spring medium were inoculated in the field with 1.0 mL of biofilm/water slurry, and immediately placed in thermos bottles at c. 65 °C for transport to the laboratory. Samples were never exposed to temperatures lower than 50 °C, because it is reasonable to assume that cultivability depends at least in part on the temperature at which organisms are maintained after sampling and before cultivation. Tubes were then placed in a 65 °C incubator and monitored for growth. Tubes showing turbidity were subcultured onto the same synthetic hot spring medium. Bacterial isolates were obtained by serial dilution to extinction preformed a minimum of three successive times. One isolate, A03C, was identified as a Thermus sp. by 16S rRNA gene identification. All growth experiments of this isolate were preformed at 65 °C on the above medium with a higher concentration of yeast extract (0.5%) and supplemented with 2 mM As(III).

Detection of arsenic-oxidizing enzyme activity in Thermus sp. A03C

Cultivation and enzyme extraction

Thermus sp. A03C cells were harvested by centrifugation at 5000 g for 20 min, and resuspended in an extraction solution [20 mM Tris-HCl pH 7.5, 0.1 M NaCl, 10% glycerol, 1% deoxycholate, 1X Sigma Protease Inhibitor cocktail (Sigma-Aldrich, St Louis, MO)]. Enzymes were extracted by glass-bead homogenization (400 μm beads, 3–1 min beating cycles) of the cell material, followed by centrifugation at 5000 g to remove cell and mineral debris. The crude total cell extract (soluble and membrane proteins) containing arsenite oxidase activity was then resolved on polyacrylamide gels with a Tris-phosphate native gel system (described below).

Protein electrophoresis

Protein electrophoresis was performed using a Pharmacia MiniVE system (Amersham Biosciences, Piscataway, NJ) with 10 cm × 10 cm plate format. Proteins were separated using a gel/buffer system with the following final composition: Stacking gel-62.5 mM Tris-phosphate pH 6.8, 4.5% acrylamide/bisacrylamide, 0.1% deoxycholate; separating gel-125 mM Tris-HCl pH 8.8, 5.0% acrylamide/bisacrylamide, 0.1% deoxycholate; Upper and lower chamber buffers-25mM Tris, 50 mM glycine pH 8.5, 0.1% deoxycholate. Samples had the following final composition (per 40 μL): 50 μg protein, 32 mM Tris-phosphate pH 6.8, 1.0% deoxycholate, 0.01 mg mL−1 bromophenol blue, 10% (w/v) glycerol. Electrophoresis was carried out at 15 mA constant current for about 2.5 h, the time required for the bromophenol blue to migrate to the bottom of the gel.

Arsenite oxidase zymography

After electrophoresis, the resolved proteins were stained for arsenite oxidase in a solution of 0.1 M Tris-HCl pH 7.5, 1 mM sodium arsenite, 69 μM dichloroindophenol, 0.2mM quinacrine hydrochloride (a flavoprotein inhibitor), and 0.2 mM 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide (MTT). Blue bands on a clear background were indicative of arsenic oxidase activity. Control reactions (no arsenite added to reaction mixture) were carried out in order to confirm that the bands visualized were products of arsenite oxidase and not nonspecific tetrazolium reduction.

Measurement of ex situ arsenite oxidation activity

An ex situ approach based on (Salmassi et al., 2006) was used to assess biofilm-associated arsenite oxidation in DBs of Alvord Hot Spring channel A3. Reaction mixtures (10.0 mL total volume in glass screw-cap culture tubes) contained filter-sterilized hydrothermal water amended with 1.0 mM sodium meta-arsenite. Biofilm samples (about 1 g of material) from the thick orange DB were introduced into the reaction tube, and samples were taken at t=0 and intervals thereafter. Control reactions received no biofilm material. Reactions were incubated in a 70 °C hot spring pool with the caps slightly ajar to allow for aerobic conditions. Reaction samples (50 μL each) were immediately transferred from reactions into 50 μL 5 mM HCl (nonoxidized sample) and 50 μL 5 mM HCl plus 5 mM potassium periodate (oxidized samples). Arsenic determinations performed according to Cummings et al. (1999).

Sequences submitted to DDBJ/EMBL/GenBank

All sequenced clones have been submitted to the DDBJ/EMBL/GenBank databases under accession numbers DQ645237DQ645260. The 16S rRNA gene sequence of the Thermus sp. isolate A03C has been submitted under accession number EF204914.


Aqueous geochemistry

Three source water outflow channels, A1, A2 and A3, were measured for temperature, pH, total arsenic and arsenic speciation along the channel (Fig. 2). The source waters of the three channels emerged at a temperature range of 74–78 °C, decreasing along the flow path. The pH of the source waters of the three channels ranged from 6.77–6.81, and increased along the flow path. Total arsenic averaged 4.5 mg L−1 (60 μM) in all three channels and remained constant while the ratio of As(III) to As(V) decreased along the source channels.

Figure 2.

 Temperature, pH, and As(III) : As(V) ratio along flow path of three source channels. Temperature, pH and arsenic were measured at 25 cm intervals along each channel. Squares, spring A1; Triangles, spring A2; Circles, spring A3. (a) Variation in temperature (open symbols) and pH (solid symbols) as a function of stream path distance. (b) Arsenic ratio (solid symbols) and total As (open symbols) as a function of stream path distance. Error bars shown for all data points.

Spectroscopic analysis of mineral phases

X-ray diffraction was used to identify crystalline phases in field samples, particularly to determine whether crystalline arsenic-bearing minerals were present in either of the collected biofilms. Neither specimen has significant amounts of crystalline arsenic-bearing minerals, and instead is largely composed of clays, silaceous materials, and iron oxides, suggesting that arsenic transformations are occurring primarily in the aqueous phase or resulting in the formation of only amorphous, noncrystalline arsenic-precipitates or adsorbates.

FT-IR was used to confirm the presence or absence of adsorbed arsenic species and presence of Fe-oxides in the absence of significant contributions to the X-ray diffractograms of As- and Fe-bearing minerals. FT-IR transmission spectra of the two materials (UB and DB) are presented in Fig. 3. Peaks due to As–O vibrations are evident in both spectra. The spectrum collected from UB exhibits a peak at 794 cm−1, attributed to the As–O symmetric stretch vibration of As(III) sorbed to Fe- and Al-oxides, and the peak at 637 cm−1 is also attributed to sorbed As(III) (Goldberg & Johnston, 2001). There is some evidence of an As(III) contribution to the spectrum collected from DB at 798 cm−1, consistent with the 794 cm−1 peak observed for UB. The peak observed for DB at 875 cm−1 is most likely due to As–O vibrations of nonsurface complexed As(V) species (Goldberg & Johnston, 2001) and this putative As(V) vibration is not observed in the spectra collected from UB. These data provide evidence for limited arsenic adsorption onto the silaceous materials, and again suggest that the transformation reactions are occurring primarily in the aqueous phase, the significance of which is the easy accessibility of arsenic species to the resident microbial biofilms.

Figure 3.

 Fourier Transform Infrared (FT-IR) spectra of biofilm material collected from the UB and DB. Peaks at c. 1000–1200 cm−1 and 475 cm−2 are due to SiO2 vibrations, the peak in the downstream spectrum at 875 cm−1 (indicated by the arrow) is most likely due to As(V)–O vibrations. The position of this peak is consistent with bands observed for nonsurface-complexed As–O bonds of As(V) absorbed to amorphous iron oxides.

Microbial diversity assessment

Two 16S rRNA gene bacterial clone libraries were made from PCR amplicons obtained from DNA extracted from DB and UBs of channel A3. Phylogenetic analyses with sequences from GenBank were used to determine the phylogenetic identity of the clones, Fig. 4. All sequenced clones were closely related to clones and isolates recovered from thermophilic environments. The diversity of the UB was relatively low, 47% of clones within the library grouped within a clade related to Sulfurihydrogenibium, 15% were related to Thermus and 2% to Thermocrinis. The DB library was more diverse, including clones within the same Sulfurihydrogenibium related clade that dominated the UB library (11%), Thermocrinis (11%), Thermus (7%), a clade within division OP10 (5%) as well as an uncultured Bacteroidetes clade (6%) and a clade of uncultured bacteria that groups with Acido-bacteria but without bootstrap support (2%). Rarefaction analyses, which was based on the RFLP patterns of the clone libraries, confirmed that the UB library had reduced diversity compared with the DB (Fig. 5), and neither library showed curve saturation indicating that full coverage of the microbial diversity was not obtained.

Figure 4.

 Phylogenetic tree of 16S rRNA genes illustrating affiliations of clones and isolate, in bold, obtained from Alvord Spring A3. Scale bar corresponds to 0.01 substitutions per nucleotide position. Dashed line indicates short sequences that were added, with a close approximation, into the tree based on a full phylogenetic analysis with the shorter sequence constraint of these three sequences, 685 unmasked positions; bootstrap values from this short sequence analysis shown in grey italics. Clone A3DB-E10 is grouped with the Acidobacteria, however, this designation is not supported by bootstrap analysis.

Figure 5.

 Rarefaction curves for the different RFLP patterns from 16S rRNA gene clone libraries A3-UB (dotted line) and A3-DB (solid line). The expected number of RFLP patterns is plotted vs. the number of clones. The faint lines represent 95% confidence intervals for each curve.

Ex situ As(III) oxidation by biofilms

Figure 6 shows the results of As(III) oxidation in the aqueous phase by the microbial community from a DB sample collected from channel A3. As(III) oxidation is clearly catalyzed by the biofilm, because total aqueous arsenic did not change appreciably during the course of the experiment indicating that the arsenic stayed in solution and was not absorbed onto the biofilm material. Approximately 90% of the added arsenic was oxidized within 2.5 h. A control with no biofilm added showed no As(III) oxidation.

Figure 6.

 Oxidation of arsenite by DB material from Alvord Hot Spring channel A3. Reactions were performed in duplicate, and the points represent average values. The supplemented As(III), almost completely oxidized by 2.5 h, was not adsorbed onto the biofilm as indicated by a steady state of total arsenic.

Isolation and identification of an arsenite-oxidizing organism

Strain A03C was obtained using a rationally-designed medium based on the observed aqueous geochemistry of the hydrothermal waters. The 16S rRNA gene sequence of A03C is c. 99% similar to clones acquired from UB and DB, constituting 15% and 7% of the clones in these libraries, respectively. A03C is related to Thermus isolates from a variety of hydrothermal areas around the world. There is a 98% sequence similarity of the 16S rRNA gene to Thermus sp. HR13 (Gihring & Banfield, 2001), Thermus aquaticus strain YT-1, Thermus sp. YSPID A.1 (Saul et al., 1993), and Thermus sp. AC-7 (Nold & Ward, 1995). Thermus aquaticus has been found capable of oxidization of arsenic, while Thermus sp. HR13 was found capable of both oxidization and reduction of arsenic (Gihring et al., 2001). Strain A03C also has the ability to oxidize arsenite to arsenate, with a rate of about 0.05 mmol L−1 min−1 (Fig. 7). This rate is comparable to that observed for the As(III)-oxidizing Thiomonas strain B2 from acid mine drainage in France (Bruneel et al., 2003) and Hydrogenophaga strain YED6-4 from Hot Creek, CA (Salmassi et al., 2006). Chemolithoautotrophic growth with As(III) as sole electron donor was not observed in strain A03C, nor was it observed in a related Thermus isolate (Gihring & Banfield, 2001). The tolerance of strain A03C to arsenite is high, as it is routinely maintained on medium containing 5 mM arsenic. Protein zymography identified an arsenite-dependent oxidase activity in cell extracts of Thermus A03C (Fig. 8).

Figure 7.

 Arsenite oxidation by strain A03C isolated from UB material (Spring A3). Cultures were grown aerobically on synthetic hot spring medium, with 0.5% yeast extract and 2.0 mM As(III). Triplicate cultures were incubated at 60°C in the dark.

Figure 8.

 Zymographic detection of As(III)-specific oxidase activity in cell extracts of Thermus A03C. + and − designations refer to reactions with and without As(III), respectively.


Aqueous and mineral-phase geochemistry

Aqueous geochemistry reveals a well-defined conversion of As(III) to As(V) forming a decreasing As(III) gradient along the stream flow path. Abiotic transformation of arsenic within this distance is highly unlikely because abiotic As(III) oxidation under these types of conditions proceeds extremely slowly (Cherry et al., 1979; Eary & Schramke, 1990). Furthermore the microbial community in the thick orange DB showed active arsenite oxidation ex situ, consistent with geochemical measurements. The transition from thin colorless UB to luxuriant thick orange DB was a well-defined line that often formed a V shaped tongue that pointed downstream in the channel. This transition was complete at 75 cm downstream from the source in channel A1, 28 cm in A2, and 39 cm in A3 and was not specifically temperature or As(III) dependent as no correlation with temperature or As(III) concentration could be determined.

X-ray spectroscopy and FT-IR analyses of the two distinct biofilm materials indicated that the solid phase arsenic found in this system consists of noncrystalline or amorphous forms rather than a crystalline mineral form. FT-IR analysis of the mineral phases has also revealed the presence of amorphous Fe-oxides in both the UB and DB material and confirms limited presence of absorbed As(V) within the DB. However, the total dissolved arsenic concentration does not significantly decrease over the stream distance of this study indicating that adsorption of arsenic into the biofilm is limited, and that microbial transformations of arsenic are occurring at the aqueous phase-biofilm interface.

Bacterial diversity

Bacterial diversity assessment indicates that both UB and DBs are dominated by the same three groups of organisms, within the Deinococcus-Thermus and Aquificae phyla. The genus Thermus contains several known arsenite-oxidizing bacteria, consistent with geochemical observations showing rapid oxidation of As(III) to As(V) (Gihring & Banfield, 2001; Gihring et al., 2001). The phylum Aquificae, which includes many known chemolithoautotrophic microorganisms able to grow at high temperatures, is represented by two clades one related to Thermocrinis and the other to Sulfurihydrogenibium. Members of the genus Sulfurihydrogenibium have been shown to both reduce arsenate and oxidize arsenite in thermal environments (Takai et al., 2002, 2003; Aguiar et al., 2004). Thermocrinis is a chemolithoautotrophic genus whose species can grow at high temperatures with hydrogen, sulfur, or thiosulfate as the electron donor (Huber et al., 1998; Eder & Huber, 2002). The DB also included clones from three uncultured clades with in the phyla OP10, Bacteroidetes and a tentative designation of Acidobacteria. Clone A3DB-E10 had very low sequence identity to existing sequences in GenBank with the exception of a few highly similar clone sequences that together formed an uncultured clade, which branched within the Acidobacteria but without bootstrap support. The neighbor-joining phylogenetic analysis shown in Fig. 4 included one sequence from each of the six Acidobacteria clades as described by (Barns et al., 1999). Most parsimonious analysis of this dataset (data not shown) also failed to give bootstrap support to this uncultured clade within the Acidobacteria. More sequence data from organisms that cover the span between the more distantly related and similar sequences currently available would undoubtedly allow a more accurate assessment of its phylogenetic placement within the Bacterial domain.

Numerous unsuccessful attempts were made to amplify archaeal DNA from Alvord Hot Spring using a variety of archaeal specific primer sets that were successful in amplifying positive control archaeal DNA (data not shown). And while not conclusive, this indicates a potential paucity of archaeal species in this hot spring system, which has also been noted in at least two Yellowstone thermal springs (Reysenbach et al., 1994; Hugenholtz et al., 1998). The microbial oxidation of arsenic in the Alvord Hot Spring outflow channel appears to be more likely catalyzed by bacteria rather than archaea.

Activity measurement, and cultivation of arsenic-oxidizing organisms

When Thermus A03C was evaluated for its ability to oxidize arsenite, the rates and conversions were similar to those reported for other arsenic-oxidizing isolates (Bruneel et al., 2003; Salmassi et al., 2006). Strain A03C typically oxidizes 90% of the As(III) added to cultures over a 50 h period (0.05 mmol L−1 min−1). Strain A03C does not appear to couple arsenite oxidation to energy transduction (data not shown), consistent with the arsenite-oxidizing strain Thermus HR13 (Gihring & Banfield, 2001). Because a high proportion of the clones from this system fell within the Thermus clade and because we readily obtained an arsenic-oxidizing Thermus isolate from this community, we conclude that Thermus-like bacteria (of which strain A03C is one) are responsible for arsenic oxidation in the outflow channels, although other arsenic-oxidizing microorganisms may also oxidize arsenite. Finally, zymography analysis clearly shows As(III)-dependent oxidase activity from cell extracts of Thermus A03C.

Summary and concluding remarks

From a comparative standpoint, our overall results are similar to those found in a high arsenic acid-chloride-sulfate system in Yellowstone National Park (55–62 °C) (Jackson et al., 2001; Donahoe-Christiansen et al., 2004) and a circumneutral creek that contains thermal inputs in California (Salmassi et al., 2006), in that arsenic oxidation occurs biotically, but there are major differences in community composition and geochemistry between these systems. Biofilm from Hot Creek in California, a system closer to mesophilic temperatures, was dominated by Beta- and Gammaproteobacteria, where Hydrogenophaga was implicated as an arsenite oxidizer incubated at a temperature of only 30 °C (Salmassi et al., 2006). The acidic thermal system in Yellowstone yielded archaeal 16S rRNA gene sequences that coincided with the start of arsenite oxidation (Jackson et al., 2001) and an arsenite oxidizing bacterium, a Hydrogenobaculum sp. of the phylum Aquificae, was recovered (Donahoe-Christiansen et al., 2004). The uniqueness of Alvord Hot Spring as a circumneutral, high-temperature hot spring with high levels of arsenic warranted further investigation. Our hypothesis, that microbial activity is responsible for arsenic transformations in the uniquely circumneutral high temperature Alvord Hot Spring, is supported by four lines of evidence: (1) significant transformation of arsenic from arsenite to arsenate forms an arsenite gradient along the stream channels, while total aqueous arsenic remains constant, and mineralogical studies confirm limited presence of absorbed As(V) onto mineral phases in the arsenic-oxidation zone; (2) substantial populations of bacteria phylogenetically related to known arsenic-oxidizing organisms have been detected in the stream channel communities; (3) the microbial biofilm community itself shows active As(III) oxidation activity using an ex situ assay preformed in the field immediately upon collection, (4) Thermus strain A03C, isolated from the biofilm, is 99% similar to a 16S rRNA gene sequence that dominated the clone libraries and zymography analysis showed this strain was capable of As(III)-dependent oxidase activity.

To begin to understand a natural microbial system and the reactions therein requires a suite of complementary interrogation tools. The assessment of the microbial community by 16S rRNA gene is informative but it often fails to predict key enzymatic pathways and reactions, and cannot be relied upon to describe the complex ecophysiologic processes of these systems. We have presented here a multidisciplinary strategy of aqueous- and solid-phase geochemical assessment, community diversity characterization, and microbial activity detection to more fully understand the arsenic-oxidizing microbial communities of the Alvord Desert Basin. Future work will involve the application of this new zymography method to detect and isolate arsenic oxidizing enzymes that are involved in geochemical transformation and mineral biogenesis. This method has the possibility of permitting the detection, recovery and purification of arsenic oxidizing enzymes directly from the environment without necessitating the isolation of microbial species that are recalcitrant to laboratory cultivation methods. Future work will also be enhanced by the development of functional gene probes and primers for the arsenic oxidase genes to further characterize the arsenic oxidizing capacity of the Alvord Basin hydrothermal microbial communities. The combination of geochemical, biochemical, and microbiological techniques has allowed us to gain a more comprehensive understanding of the active microbial functions within this ecosystem.


We thank the Alvord Ranch and the U.S. Bureau of Land Management for granting permissions to the field site. This work was funded by an NSF-EPSCoR Research Infrastructure Improvement Grant (Award Number 0132626) to the State of Idaho (J. Shreeve, PI), and by the NASA-Idaho Space Grant Consortium and NASA-EPSCoR (to T.S. Magnuson). We would like to thank the Molecular Research Core Facility-ISU (Michelle Andrews and Erin O'Leary-Jepsen) for DNA sequencing, and Jennifer Hinds and Jerry Fairley (University of Idaho) for assembling and maintaining the website about the Alvord Basin hot springs area geology and biogeochemistry which can be found at