Introduction

The organoarsenical 3-nitro-4-hydroxybenzene arsonic acid (roxarsone) has seen extensive use in the production of broiler chickens. It is added to feed to prevent coccidiosis (an infection by species of Eimeria, an apicomplexa protist) but also has the added benefit of stimulating growth and improving pigmentation (1). The majority of the arsenic is not retained by the chickens but excreted (2). According to the USDA, the United States produced 8.9 billion broiler chickens in 2005 (3), typically in concentrated animal feeding operations (CAFOs) (4). Although some commercial produc ers have voluntarily ended the practice, a significant percentage of these animals are still fed roxarsone, and the litter they produce can contain from 15 to 48 mg/kg of the organoarsenical (5-9). Approximately 90% of this litter is disposed of through land application (8), and the presence of arsenic presents a problem for waste management (10). While there has been some debate over whether this arsenic is getting into drinking water (4), an increase in soil arsenic concentration and the presence of arsenicals in house dust from homes (10.7-130 mg/kg) in farming communities surrounding CAFOs have been reported (10, 11). Thus the health risk posed by the use of organoarsenicals in CAFOs remains a real concern (10).

Although roxarsone is the major arsenic species in fresh litter, inorganic arsenate (As(V)) predominates in composted litter (6, 7). Such a species not only is more toxic but also can be readily leached into groundwater (12, 13). Previous studies using chicken litter slurries (7) or sewage sludge (14) have suggested that biological processes are responsible for the transformation of the roxarsone. Neither the organisms nor the biological processes involved, however, were identified. This investigation focused on identifying the populations, products, and processes involved in the biotransformation of roxarsone in chicken litter using enrichments and a pure culture of an arsenate-respiring species of Clostridium, strain OhILAs. The recently completed draft genome for Clostridium sp. strain OhILAs allowed for in depth analysis of potential biochemical pathways involved.

Materials and Methods

Incubations. The basal medium contained (in 1 L) 0.095 g of MgCl₂, 4.2 g of NaHCO₃, 0.5 g of yeast extract, and 10 mL each of trace elements and vitamin mix (15). The pH was adjusted to 7.3. For anaerobic incubations, the medium (100 mL) was dispensed in 125 mL Wheaton crimp seal bottles, degassed with oxygen-free N₂, sealed, and autoclaved. For aerobic incubations, the medium (100 mL) was dispensed in 250 mL flasks with cotton plugs. Lactate or fructose was added separately (final concentration 10 mM) from sterile, degassed with oxygen-free N₂ (omitted for aerobic experiments), stock solutions, each prepared in basal medium. Roxarsone, 3-amino-4-hydroxybenzene arsonic acid (3A4HBAA), 4-hydroxybenzene arsonic acid (4HBAA), and 2-nitrophenol stock solutions were also prepared and added separately (as described for the carbon sources) to a final concentration of 1 mM. The slurry was prepared by suspending 5 g of chicken litter (kindly provided by M. Schreiber, Virginia Tech) in 100 mL of sterile basal medium. For anaerobic enrichments, the slurry was degassed with sterile N₂ to create an anoxic condition. The medium (100 mL) was inoculated with either 1 mL of slurry (for the chicken litter enrichments) or 1 mL of the pure culture of Clostridium sp. strain OhILAs and incubated under ambient conditions (e.g., day/night cycle, 22 C) or at 37 C in the dark. The chicken litter enrichments were terminated at 300 h. Pure culture experiments were terminated at 192 h (roxarsone transformation) or 72 h (cell yield studies). Heat-killed enrichments incubated under the same conditions were used as controls. Aliquots were taken every 12 h for cell counts (1 mL) and product analyses (1 mL). For analytical measurements, the aliquots were filtered (0.22 m nylon, Fisher Scientific) and acidified by adding 200 L of 2 M sulfuric acid. Cell growth was monitored by measuring the optical density (600 nm) or by direct cell counts that were done on 0.2 m black polycarbonate filters (Poretics Inc.) stained with acridine orange (1% w/v) and observed on a Nikon Mikrophot SA fluorescence microscope. All experi ments were done in triplicate.

Molecular Analysis. At the termination of the incubations, the chicken litter enrichments were harvested by centrifuga tion. The DNA was extracted using alkaline lysis and phenol extraction. The 16S rRNA genes were amplified using the 344F and 907R primers (16), the library was constructed using the Topo cloning kit (Invitrogen, Carlsbad, CA), clones were screened using restriction fragment digests (EcoR1), and selected clones (CL-504, CL751, CL752, CL1001, CL1002, CL1006) were sequenced. The sequences for the clones and Clostridium sp. strain OhILAs have been deposited in GenBank under the accession numbers DQ250645-DQ250649. BLAST analysis was used to determine the closest known relatives with additional sequences obtained from GenBank. The sequences were aligned using CLUSTAL X and trees constructed with PAUP as reported (17). The tree includes 16S rRNA sequences that were amplified from the pig intestine (p-228-o5, p-478-o2) (18) and chicken cecum (CIAJG45)(19).

Genome sequence and draft annotation of Clostridium sp. strain OhILAs (ATCC BAA-1360) were provided by the DOE Joint Genome Institute. Genome analysis was done using the BLAST search engine provided on the Integrated Microbial Genomes website (http://genome.ornl.gov/microbial/clos/).

Analytical Method. Cultures were monitored daily for the disappearance of the yellow color indicating the loss of roxarsone. Roxarsone and its metabolites were identified by HPLC (Waters 2692 and 600), UV/vis spectroscopy (Perkin-Elmer Lamda 2), and mass spectrometry (Micromass ZMD, ESI-MS) (20, 21). The analytes were separated using either an ion exchange (BioRad Aminex-HPX-87H) or C18 (Phenominex aqua C18) column and detected with a UV detector (190 nm or 254 nm). For ion exchange chromatography the mobile phase was acidified with H₂SO₄ with a flow rate of 1.0-0.5 mL/min. For the reverse phase chromatography the mobile phase was acidified water containing methanol and methyl tributyl ammonium hydroxide which was used with a flow rate of 0.15 mL/min. Standard solutions of roxarsone, 3A4HBAA, 4HBAA, phenyl arsine oxide, salvarsan, phenol, monomethyl arsonic acid (MMA), dimethyl arsonic acid (DMA), sodium arsenate, sodium arsenite, 2-nitrophenol, and 2-aminophenol were prepared in the base media. 3A4HBAA, 4HBAA, 2-nitrophenol, and 2-aminophenol were separated using a C18 column. Under these conditions the retention times were as follows: roxarsone - 49.7 min, arsenite - 7.4 min, arsenate - 5.3 min, lactate - 15.2 min, acetate - 18.6 min, 3A4HBAA - 14.7 min. Liquid chromatography coupled with mass spectrometry (electrospray ionization) was used to confirm the presence of 3A4HBAA.

Computational Methods. All computations were performed using the Gaussian 03 software package (22) run with Windows. The geometries were obtained by first building the structure using HyperChem and then optimizing the structure using HF (with STO-3G basis set) with the final optimization done using Gaussian. In all cases, vibration frequencies were calculated to ensure that the optimized geometries represented local minima. In all calculations, Becke's three-parameter hybrid exchange functional (B3P86) (23) and Lee-Yang-Parr's nonlocal correlation functional (B3LYP) (24) were used. 6311gdp was used for C, H, and O, and 6-311+gdp was used for As. The electron densities for the molecular orbitals were generated using Gauss View.

Results

Roxarsone Transformation by Chicken Litter Enrichments. The initial experiments were done to determine if roxarsone could be transformed by chicken litter enrichments under strict anoxic conditions with lactate as a carbon source. Roxarsone (1 mM) was rapidly degraded (maximum rate of 6.0 × 10^-3 mM/h) and was no longer detectable after 204 h (Figure 1A). Its disappearance was concurrent with the oxidation of lactate to acetate (Figure 1B). Both the consumption of lactate and production of acetate exhibited a biphasic behavior. Between 12 and 168 h, lactate was consumed at a rate of 14.8 (±0.5) × 10^-3 mM/h, while acetate was produced at a rate of 11.8 (±0.52) × 10^-3 mM/h. After 168 h, the rate of the consumption of lactate increased to 90.8 (±22.5) × 10^-3 mM/h, and the rate of acetate production also increased to 235.1 (±25.6) × 10^-3 mM/h. While the rate of lactate consumption was increased by 6-fold, the acetate production increased by 20-fold. The bulk of lactate consumption and acetate production occur after 168 h suggesting different metabolic pathway(s) might be operative. Furthermore, the dramatic increase in the consumption of lactate and production of acetate occurred only after the roxarsone had been transformed. 3A4HBAA was detected beginning at 84 h and continued to increase until 204 h, reaching a final concentration of ~0.5 mM. Inorganic arsenate appeared after 108 h and continued to increase to ~0.3 mM as the incubations progressed. No biotransformation of roxarsone was discernible in aerobically grown chicken litter enrichment cultures despite a significant increase in biomass (Supporting Information). No loss of roxarsone was found in the heat-killed controls as well, and no photodegradation was observed (data not shown).

Figure 1 Transformation of roxarsone in chicken litter enrichments. (A) Consumption of roxarsone () and production of 3-amino-4-hydroxybenzene arsonic acid () and As(V) (·). (B) Consumption of lactate (), production of acetate (·), and cell growth as optical density at 600 nm (). (C) Phylogenetic tree, maximum parsimony (bootstrap values at the nodes), showing the affiliation of the four chicken litter enrichment clones (CL751, CL752, CL1001, CL1002) and Clostridium sp. strain OhILAs with Clostridium species.

After a short lag period, cell numbers increased exponentially, reaching stationary phase by 96 h (Figure 1B). Examination of the chicken litter enrichments by light microscopy revealed populations of small spore-forming rods of similar morphology. Molecular analysis indicated the presence of Clostridium species (Figure 1C). A total of 20 clones were obtained, and six clones were chosen for sequencing based on the restriction digests. The sequenced clones yielded four different 16S rRNA sequences all of which clustered with clostridial species (Figure 1C). Chicken litter enrichment clones CL1001 and CL751 were most closely related to Clostridium celerecrescens (99 and 94%, respectively), while the closest relatives of CL752 and CL1002 were Clostridium irregulare (99%) and Clostridium sp. IMSNU 40011 (99%), respectively.

Roxarsone Transformation by Clostridium sp. strain OhILAs. Clostridium sp. strain OhILAs is a low G+C gram positive bacterium that was isolated from anoxic sediments from the Ohio River (Pittsburgh, PA) and found capable of dissimilatory arsenate reduction (25). Initial studies with the pure culture showed that it could also transform roxarsone. Determination of the minimal inhibitory concentration revealed that 2.5 and 5 mM roxarsone inhibited growth, but growth yields were enhanced with 0.5 and 1 mM roxarsone (Supporting Information). As 1 mM was found to be optimal for transformation and growth, we used that concentration for our experiments.

A similar pattern of roxarsone transformation was observed for the pure culture of Clostridium strain OhILAs with lactate although the rate was faster (maximum rate 16 × 10^-3 mM/h) (Figure 2A). The disappearance of roxarsone was completed by 84 h and coincided with growth, disappearance of lactate, and production of acetate (Figure 2B). The amount of acetate produced (0.26 mmol) was slightly greater than the amount of lactate consumed (0.22 mmol). Unlike in the chicken litter experiments no biphasic behavior was observed. Lactate was consumed at a rate of 36.6 (±3.7) × 10^-3 mM/h, and acetate was produced at the rate of 45.7 (±3.3) × 10^-3 mM/h. Propionate was detected in the Clostridium sp. strain OhILAs cultures only after the bulk of the roxarsone had been depleted (data not shown). 3A4HBAA was detected by 36 h reaching a maximum concentration (0.6 mM) at 84 h, while As(V) first appeared at 48 h and reached its maximum concentration (0.3 mM) by 72 h (Figure 2A). Subsequent growth experiments conducted under the same conditions (e.g., basal medium with lactate) but with 3A4HBAA or 4HPAA showed cell growth (comparable to cells numbers with lactate alone) but no transformation of either compound (data not shown). As expected, Clostridium sp. strain OhILAs also grew on and transformed 2-nitrophenol (Supporting Information).

Figure 2 Transformation of roxarsone in cultures of Clostridium sp. strain OhILAs. (A) Consumption of roxarsone () and production of 3-amino-4-hydroxybenzene arsonic acid () and inorganic As(V) (·). (B) Consumption of lactate (), production of acetate (·), and cell growth as optical density at 600 nm (). (C) Cell yields (as optical density at 600 nm) on lactate alone (·) and lactate plus roxarsone ().

A separate set of experiments was conducted to investigate the metabolic role of roxarsone. Cultures of Clostridium sp. strain OhILAs were grown in basal medium with either lactate or fructose and with or without roxarsone. Whether lactate or fructose was used, the cell yields in the presence of roxarsone were ~10 fold greater than without. The results for lactate are shown in Figure 2C. This significant difference in cell yield in the presence of roxarsone suggests its reduction to 3A4HBAA might be linked to a respiratory process.

Electronic Structure of Roxarsone. Roxarsone has two functional groups (nitro and arsenate) that can act as an electron sink. To understand which group is preferred to accept reducing equivalents we conducted electronic structure calculations on roxarsone using DFT methodology. The molecular orbital diagram and the shapes of selected orbitals are shown in Figure 3. The highest occupied molecular orbital (HOMO) is energetically well separated from a couple of near degenerate orbitals (HOMO-1 and HOMO-2). An orbital below the HOMO-2 is also energetically well separated from these two. Shapes of these frontier orbitals reveal that the electron density of the HOMO is primarily localized on the benzene ring with antibonding interaction with other functional groups including the phenolato oxygen. In the three orbitals below the HOMO it is primarily on either the nitro or arsenate group. For example, HOMO-1 has electron density on the benzene as well as on the nitro group and small electron density on the terminal oxygen of the arsenate group. In contrast, the majority of the electron density lies on the arsenate group in HOMO-2. Because these are the filled frontier orbitals, they should be involved in any oxidative process. Thus, the electronic structure implies that in the case of oxidation the benzene ring will be oxidized first.

Figure 3 The electronic structure of roxarsone as calculated by density function theory (DFT) (see text for details). The molecular orbital energy (in eV) level diagram of the frontier orbitals shows the LUMO is well separated from the HOMO, and, in the inset boxes, shapes of the LUMO (top) and the HOMO (bottom) are shown. Note the significant contribution from the benzene ring into the HOMO, while the nitro group contributes substantially into the LUMO.

The first four unfilled virtual orbitals are also separated energetically. While caution must be exercised in dealing with the energy of an unfilled orbital, the shapes of the orbitals reveal interesting features. The lowest unoccupied molecular orbital (LUMO) is almost equally distributed among the nitro group and the benzene ring. The two orbitals above the LUMO (i.e., LUMO +1 and LUMO+2) have significant contribution from the benzene ring and the nitro-group with increasing participation from the arsenate group. The next orbital is dominated by the arsenate group. This suggests that the reduction of the arsenate group does not occur before the reduction of the nitro group.

Discussion

Our studies have demonstrated that roxarsone can be rapidly transformed into 3A4HBAA and inorganic arsenic by clostridial species, both in chicken litter enrichments and pure culture, under anaerobic conditions. This is in marked contrast to the study by Cortinas et al. in which no significant release of inorganic arsenic occurred during the formation of 3A4HBAA (14). We attribute this difference as possibly due to their use of wastewater sludge from a paper recycling plant as an inoculum resulting in a different microbial community (although the microbial composition in the sludge is yet to be reported). Regardless, clostridia are the dominant species in the chicken cecum (26, 27) and are a common component of the litter (28). The observation of only spore-forming rods under light microscopy and the limited diversity seen in the 16s rRNA clones suggest that clostridial species were the dominant organisms in the enrichment cultures. While it is possible that the enrichment medium was selective for clostridia, no chemical inhibitors were used, and additional nutrients were provided by the chicken litter slurry. In addition, it has been reported that roxarsone is only mildly inhibitory to methanogenesis (29).

Physiological Role of Roxarsone Transformation in Clostridium. The fact that we observed a similar pattern of roxarsone transformation with our pure culture, Clostridium sp. strain OhILAs, allowed for further investigation of the process. The reduction of the nitro group to an amine on the aromatic ring requires six electrons and suggests the following equation:

In our experiments with Clostridium sp. strain OhILAs, 0.22 mmol of lactate was oxidized. This should have yielded an equal amount of acetate and 0.88 mmol of [H]. The 0.083 mmol of roxarsone transformed should consume 0.49 mmol of [H]. The remainder of the reducing equivalents can be accounted for if the additional amount of acetate (~0.04 mmol) were produced via acetogenesis. Acetogenesis via the Lunjdahl-Woods pathway converts 2CO₂ and 8[H] to one acetate (30). Only ~0.06 mmol of 3A4HBAA was found in the cultures of Clostridium sp. strain OhILAs, but the remaining roxarsone can be accounted for by the amount of As(V) detected (~0.03 mmol).

The oxidation of lactate generates acetate, CO₂, and four reducing equivalents ([H]). In the acrylate pathway, as found in some clostridial species, the 4[H] are subsequently consumed by the conversion of two more molecules of lactate to two molecules of propionate (31). Lactate can also be used as an electron donor in anaerobic respiration, with the reducing equivalents shuttled to an electron-transfer chain ultimately reducing a suitable electron acceptor (31). Thus it is possible for roxarsone to serve as either a sink for the reducing equivalents generated during fermentation (with ATP generated by substrate level phosphorylation of acetyl-CoA) or a terminal electron acceptor linked to oxidative phosphorylation (31). Genomic analysis indicates that Clostridium sp. strain OhILAs should be capable of both the fermentation of lactate and glycolysis. All known components of the acrylate pathway (e.g., lactate dehydrogenase, pyruvate-ferredoxin oxidoreductase, phosphotransacetylase, acetate kinase) were identified in the genome, whereas components of the succinate-propionate pathway (e.g., malate dehydrogenase, fumarate hydratase) were missing. An almost complete glycolytic pathway is also present; however, glucokinase and glucose-6-phosphate isomerase are missing, indicating the organism should ferment fructose but not glucose.

Even though clostridia are considered primarily fermentative, a respiratory process cannot be ruled out. The recent description of chemiosmotic ATP generation coupled to the reduction of caffeate (a phenylacrylate ester) via a Na⁺ dependent F₁/F₀ ATPase provides a possible mechanism (30). Using the annotated genome of Clostridium sp. strain OhILAs we were able to determine that it has homologs of Na⁺ dependent F₁/F₀ ATPase, NADH dehydrogenase, and hydrogenase. The absence of genes encoding cytochromes and the presence of the biosynthetic pathway for membrane-bound corrinoids further suggest that the cells can generate a Na⁺-dependent proton motive force (30).

We also suggest that a similar process may be occurring in the chicken litter enrichments (i.e., roxarsone serving as an electron acceptor for anaerobic respiration). However, we are aware of the fact that these are mixed cultures and could include homoacetogenic species (e.g., Moorella thermoacetica) that possess a cytochrome-containing respiratory chain (30). Interestingly, the dramatic increase in the rate of lactate consumption and acetate production seen once the roxarsone was consumed suggests that roxarsone may inhibit other metabolic processes in the mixed populations. Both rates are much higher than the corresponding rate constants for the first process but significantly lower than the second set of rate constants obtained for the chicken litter experi ments. We attribute the initial oxidation of lactate in chicken litter to the reduction of roxarsone, similar to what we observed in Clostridium sp. strain OhILAs. We attribute the higher rate of roxarsone transformation observed for Clostridium sp. strain OhILAs to the fact that freshly prepared chicken litter slurries required additional incubation time to establish the microbial populations.

Reduction of the Nitro Group on Roxarsone. Our computational analysis indicates that any reductive pathway first must involve the reduction of the nitro group is consistent with both our experimental observations and the results of others (7, 14). Our analysis further suggests that any oxidative pathway would involve the oxidation of the benzene ring that may ultimately release inorganic arsenate. Interestingly, nitrophenol, the expected product from the direct cleavage of the arsenate group, has as yet been detected. One of the proposed pathways suggests the formation of 3A4HBAA followed by deamination to 4HBAA where the arsenate group remains bonded to the benzene ring that is subsequently released (14). In general, the aromatic carbon arsenic bond is expected to be stronger than that in aliphatic molecules (32). Therefore, it is unlikely that the release of the arsenate would involve a direct breaking of C-As bond in 4HBAA as suggested. We propose that the release of the arsenate group should involve ring cleavage.

The reduction of nitro groups on aromatic compounds by clostridia is well-known (33, 34). C. celerescence, the species most closely related to both chicken litter enrichment clones CL751 and CL1001, is capable of using hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) as a nitrogen source (35). Furthermore, the reduction of the nitro group has been shown to be dependent on the availability of reducing equivalents (36). Genomic analysis of Clostridium sp. strain OhILAs revealed the presence of two homologs of pentaerythritol tetranitrate reductase (NemA), several nitroreductases, carbon monoxide dehydrogenase, and hydrogenase, all which have been implicated in the reduction of nitro groups on aromatic compounds (33, 34). Which of these, if any, are involved is currently under investigation.

Environmental Significance. Roxarsone is highly water soluble, and therefore mobile, and can potentially partition between ground and surface waters. The composition of the chicken litter depends, among other things, on the availability of water and the duration of composting. The data presented here demonstrate that roxarsone can be transformed by clostridial species at rates much faster than those reported in earlier studies. There have been reports of 4HBAA, phenyl arsonic acid, and methylated species in addition to 3A4HBAA and inorganic arsenic (9); however, the processes by which these arsenic species are generated during composting have not as yet been identified (7, 37). More importantly, our study suggests that the microbial transformation of roxarsone that occurs in the litter may ultimately impact the arsenic composition in the soil and dust (11). While we have demonstrated, for the first time, that clostridial species are involved in simultaneous generation of inorganic arsenic and 3A4HBAA, we have only elucidated a part of the process. Thus further studies are warranted.

Acknowledgment

We gratefully acknowledge M. Schreiber for providing the chicken litter and A. Barchowsky, J. H. Enemark, W. M. Griffin, M. Lee, L. Ljundahl, B. K. Nicholson, G. M. Momplaisir, and R. S. Oremland for fruitful discussions and comments. This work was supported in part by the USGS-NIWR. L. Wormer was supported by a NASA Planetary Biology Internship.