Reports GFAJ-1 Is an Arsenate-Resistant, ments provided additional evidence that GFAJ-1 is a phosphate-dependent organism, even when cultivated in the Phosphate-Dependent Organism presence of high arsenate concentrations. Next we addressed the question of Tobias J. Erb,1*† Patrick Kiefer,1* Bodo Hattendorf,2 Detlef Günther,2 Julia A. whether arsenate enters intermediates Vorholt1† 1Institute of Microbiology, ETH Zurich, Wolfgang-Pauli-Str. 10, 8093 Zurich, Switzerland. 2Laboratory of Inorganic Chemistry, ETH Zurich, Wolfgang-Pauli-Str. 10, 8093 Zurich, Switzerland. of the central carbon metabolism of GFAJ-1 and how this might affect the organism’s metabolome. For that purpose, cells of GFAJ-1 were grown on *These authors contributed equally to this work. the minimal medium described by †To whom correspondence should be addressed. E-mail: toerb@ethz.ch (T.J.E), vorholt@micro.biol.ethz.ch (J.A.V.) Wolfe-Simon et al. with 10 mM glucose (1) in the absence or presence of 40 mM arsenate and trace amounts of The bacterial isolate GFAJ-1 has recently been proposed to substitute arsenic for phosphorus to sustain growth. We have shown that GFAJ-1 is able to grow at low phosphate concentrations (1.7 μM), even in the presence of high concentrations of arsenate (40 mM), but lacked the ability to grow in phosphorus-depleted (<0.3 μM), arsenate-containing medium. High resolution mass spectrometry analyses revealed that phosphorylated central metabolites and phosphorylated nucleic acids predominated. A few arsenylated compounds, including C6 sugar arsenates, were detected in extracts of GFAJ-1, when GFAJ-1 was incubated with AsO43-, but further experiments showed they formed abiotically. Inductively coupled plasma mass spectrometry confirmed the presence of phosphorus and the absence of arsenic in nucleic acid extracts. Taken together, we conclude that GFAJ-1 is an arsenate- resistant, but still a phosphate-dependent bacterium. phosphate. Cells were collected in midlog phase through fast filtration and washed rapidly at 24°C on the filter to remove inorganic salts of the medium, before metabolites were extracted and analyzed by a high resolution mass spectrometry platform that allows detection of compounds in the femto/attomole range (11). Metabolite peaks were searched against a modified EcoCyc-metabolite database (12) that has been systematically extended by all the arsenylated and arsinylated metabo- lites theoretically possible (e.g., glu- The discovery of GFAJ-1, a Gammaproteobacterium that was claimed to cose-arsenate, glucose-arsenite), as well as permutations of mixed phos- use arsenic instead of phosphorus has challenged the universal role of phorylated, arsinylated or arsenylated species to detect modified bis- or phosphorus in biology (1), although the arguments used in the study tris-phosphate metabolites (e.g., different ADP-, or ATP-species). Feawere doubted (2–9). Here, we combine classical physiological experi- tures that matched a database entry with an accuracy <1 × 10−3 atomic ments with high-resolution mass spectrometry, as well as inductively mass units were assigned as potential metabolites, and their retention coupled plasma optical emission and -mass spectrometry (ICP-OES / time was compared to the known retention times of corresponding phos- ICP-MS) to provide evidence that GFAJ-1 is highly resistant to arsenate, pho-metabolite standards for further characterization. With the caveat yet still requires phosphate for growth. that the metabolome of GFAJ-1 might have been perturbed to some ex- To understand the physiological properties of GFAJ-1, we studied tent by the short washing step, we found that when GFAJ-1 was grown the organism’s phosphate-dependency in greater detail. Using reagents in the presence of arsenate, most core metabolites (nucleotides, sugar of the highest purity available for medium preparation we reduced the phosphates, etc.) were only detected in their phosphorylated, but not phosphorus-background in the minimal medium below the detection arsenylated form (Table 1 and tables S1 to S4). Moreover, the absolute limit (<0.3 μM phosphorus), which is an order of magnitude less than concentrations of most phospho-metabolites did not differ between detected by Wolfe-Simon et al., i.e., 2.7-3.2 μM phosphorus impurity GFAJ-1 cells grown in the presence or absence of arsenate (Table 1). (1). In the phosphorus-depleted medium (<0.3 μM) we made, no growth Notably, the levels of nucleotide-trisphosphates (ATP, CTP, GTP, UTP), of GFAJ-1 was observed; however, the amount of growth of GFAJ-1 as a measure for cellular energy status, were similar between both correlated with the amount of external phosphorus added in form of growth conditions, although some nucleotide-bisphosphates (ADP, CDP, phosphate, up to a concentration of 20 μM beyond which other nutrients GDP), but not their corresponding monophosphates, appeared elevated might become limiting (10) (Fig. 1, A and B). Notably, 1.7 μM phos- in GFAJ-1 cells grown in the presence of arsenate. The elevated nucleo- phate was sufficient to sustain growth of GFAJ-1 similar to that remain- tide-bisphosphate levels might result from a higher energy demand of ing as impurity in the arsenate-grown culture media published by Wolfe- GFAJ-1 cells when grown in the presence of arsenate (e.g., due to ATP- Simon et al. (1). From this observation we conclude that cultures in the dependent detoxification mechanisms, such as active export of arsenate), previous study might have grown on trace amounts of phosphate rather or might point to the formation of transient, instable nucleotide- than on the arsenate provided. Indeed, when GFAJ-1 was tested for bisphosphate-arsenate species, as proposed before (13). Nevertheless, growth, by optical density measurements and direct cell counts, on arse- our results indicated that the core-metabolism of GFAJ-1 is based on nate (40 mM) using phosphorus-depleted medium (<0.3 μM), no growth phospho-metabolites, independent of its growth condition without major was observed unless trace amounts of phosphate were added (Fig. 1C). perturbation of most core metabolite levels (except above discussed Inductively coupled plasma mass spectrometry (ICP-MS) was used to nucleotide-bisphosphates) as a consequence of arsenate addition to the follow the fate of trace phosphate during growth of GFAJ-1 in arsenate- medium, which strongly argues against the use of arsenate to replace containing medium. The concentration of phosphorus in the medium phosphate in GFAJ-1. supernatant decreased with time and became enriched in the cellular Yet, we observed hexose-arsenate in metabolome extracts of arse- fraction, in parallel with the growth of GFAJ-1, as shown by optical nate-grown cells of GFAJ-1 that co-eluted with the glucose-phosphate density and cell-number counting (Fig. 1D). The accumulation of phos- standard, as well as a potentially bisarsenylated hexose species (Table phorus in the cellular fraction during growth of GFAJ-1 in these experi- 1). However, the retention time of the latter deviated from that of fruc- / http://www.sciencemag.org/content/early/recent / 8 July 2012 / Page 1/ 10.1126/science.1218455 Fig. 1. Growth behavior of GFAJ-1 on minimal medium containing 10 mM glucose amended with different concentrations of phosphate and arsenate. Cell growth was followed by increase in optical density at 600 nm (OD600, open circles), or direct cell counting (closed diamonds). (A) Phosphate-dependent growth of GFAJ-1, as determined by OD600. Shown are triplicate growth curves of GFAJ-1 in the presence of 15.4 μM phosphate, as determined by elemental analysis (brown open circles); 12 μM phosphate, mixed from 15.4 μM phosphate medium and <0.3 μM phosphate medium (dark red open circles); 9 μM phosphate, mixed from 15.4 μM phosphate medium and <0.3 μM phosphate medium (red open circles); 4 μM phosphate, as determined by elemental analysis (light red open circles); 1.7 μM phosphate, as determined by elemental analysis (orange open circles); below 0.3 μM phosphate, as determined by elemental analysis (pale red-violet open circles); 40 mM arsenate, below 0.3 μM phosphate, as determined by elemental analysis (gray open circle). (B) Correlation of stationary phase OD600 and phosphate concentration in the medium. Shown are the results from three independently prepared batches of growth media and precultures. Red circles indicate the absence of arsenate in the growth medium. Black circles indicate the presence of 40 mM arsenate in the growth medium. Medium batch 1 (yellow filled circles); medium batch 2 (light blue filled circles); medium batch 3 (light green filled circles). (C) Phosphate-dependent growth of GFAJ-1 in the presence and absence of arsenate, as determined by OD600 and direct cell counting. Open circles indicate OD600 measurements, diamonds indicate direct cell counts. Shown are representative growth curves of GFAJ-1 cultures in the presence of 40 mM arsenate, 10 μM phosphate (black open circles, black closed diamonds); 10 μM phosphate, no arsenate (red open circles, red closed diamonds); 40 mM arsenate, phosphate below 0.3 μM (pale red-violet open circles); below 0.3 μM phosphate, no arsenate (gray open circles). (D) Dynamics of phosphorus during growth of GFAJ-1 in the presence of arsenate and limited amounts of phosphate. Shown are OD600 measurements (black open circles) and direct cell counts (black closed diamonds) of a GFAJ-1 culture in the presence of 40 mM arsenate and 10.4 μM phosphate, as well as the distribution of elemental phosphorus in the supernatant (red crosses), or the cellular fraction (boxed red crosses). Phosphorus concentrations were determined by ICP-MS. / http://www.sciencemag.org/content/early/recent / 8 July 2012 / Page 2/ 10.1126/science.1218455 Table 1. Selected core metabolites of GFAJ-1 cells actively growing on limiting amounts of phosphate in the absence or presence of 40 mM arsenate, and metabolic response upon incubation with 40 mM arsenate. Metabolites (fmol/μl extract) Cells grown with 9 μM P Cells grown with 9 μM P 40 mM As Medium treated filter Cells grown with 10 μM P 6 s As incubat. Cells grown with 10 μM P 360 s As incubat. Nucleotide-monophosphates AMP / dGMP 4.5 ±1.2 5.2 ±0.6 <0.3# CMP 1.3 ±0.1 1.4 ±0.3 <0.3# UMP 0.7 ±0.3 0.5 ±0.1 <0.3# Nucleotide-bisphosphates ADP CDP GDP UDP 58 ±16 141 ±27 <0.8# 15 ±1 27 ±5 <0.5# 14 ±2.4 31 ±0.9 <0.4# 2.1 ±0.2 4.5 ±1.1 <0.2# Nucleotide-trisphosphates ATP CTP GTP UTP 145 ±13 153 ±27 <1.2# 21 ±3 31 ±4 <0.2# 22 ±2 30 ±4 <0.7# 18 ±1 23 ±7 <0.3# Deoxynucleotide-phosphates dADP dTDP dTTP 1.2 ±0.1 3.7 ±1.2 <0.2# 2.4 ±0.2 6.1 ±2.1 <1.1# 32 ±3 30 ±10 <1.7# Sugar-phosphates Hexose-phosphate Hexose-bisphosphate Phospho-gluconate 13 ±3 13 ±3 <0.9# 18 ±3 18 ±4 <0.9# 3.6 ±0.1 2.9 ±1.6 <0.2# Nucleotide-sugar- phosphates UDP-hexose 42 ±2 55 ±11 <1.3# UDP-N-acetyl-hexosamine 29 ±2 37 ±4 <1.1# Other phospho-metabolites Methylerythritol-cyclo-P-P 5.2 ±3.0 7.2 ±0.7 <1.6# 2.9 ±<0.1 3.9 ±0.8 1.0 ±0.1 1.1 ±0.2 0.2 ±0.1 0.2 ±0.2 33 ±5 58 ±31 13 ±2 16 ±3 11 ±3 14 ±4 1.7 ±0.2 2.2 ±0.3 166 ±15 128 ±29 25 ±2 21 ±2 25 ±2 20 ±4 18 ±1 15 ±2 0.7 ±0.1 1.3 ±0.8 2.2 ±0.5 4.0 ±0.6 28 ±5 19 ±3 7.6 ±0.1 10 ±4 21 ±3 29 ±6 2.7 ±0.2 4.0 ±1.0 36 ±3 28 ±2 24 ±2 22 ±2 5.7 ±0.3 17 ±3 Arsenylated analogs (rel. levels) Hexose-arsenate <0.1 Hexose-bisarsenate <0.1 159 ±18 215 ±120 155 ±81 74 ±75 4 ±5 4 ±2 18 ±8 9 ±9 #Below detection limit. tose-1,6-bisphosphate, which makes it questionable that it represented an arsenate analog of the glycolytic intermediate. The origin of arsenylated hexoses in metabolome extracts of arsenate grown GFAJ-1 cells was subsequently addressed in more detail. Notably, hexose-arsenate as well as bisarsenylated hexose were also detected in control extracts of filters that had been mock-treated with arsenate amended, glucose containing, growth medium, indicating abiotic formation of these arsenylated hexoses (Table 1). This finding is in line with thermodynamic considerations and previous reports indicating spontaneous formation of glucosearsenate from arsenate and glucose in solutions of alkaline pH (14). Consequently, the bisarsenylated hexose species we detected is likely to be abiotically formed glucose-bisarsenate, or an unspecific glucosearsenate adduct rather than an arsenate analog of fructose-1,6bisphosphate. Although our experiments indicated that hexose-arsenate is preformed abiotically in the growth medium, we could not exclude the possibility that some might be produced biotically, e.g., through transient ADP-arsenate as suggested before (13). Hence, to reduce the background levels of abiotically formed hexose-arsenates in metabolome preparations, a more stringent washing step with water was included during cell collection. The modified protocol changed the metabolite pool sizes, particularly of hexose-phosphate, which decreased about tenfold (table S5). However, when GFAJ-1 cells, grown in the presence of arsenate, were extracted with this stringent protocol, we were not able to detect any arsenylated hexoses (table S5). Similarly, when 10 mM glucose was added to the washing solution to minimize depletion effects, we did not observe the formation of hexose-arsenates beyond small abiotic background levels in extracts of GFAJ-1 cells grown in the presence of arsenate (table S5). Although we did not find evidence for the biotic formation of hexose-arsenates, GFAJ-1 cells remained metabolically active in these experiments, as demonstrated by addition of 10 mM 13Clabeled glucose to the wash solution: about 80% of the hexose-phosphate pool and 60% of the fructose-1,6-bisphospate pool were exchanged with the 13C-label, even during stringent washing, indicating active glycolysis (table S6). These results suggest that the biotic formation of glucosearsenate is most likely negligible in GFAJ-1. / http://www.sciencemag.org/content/early/recent / 8 July 2012 / Page 3/ 10.1126/science.1218455 Table 2. Elemental analysis of nucleic acid preparations from GFAJ-1 grown on 10 μM phosphate in the absence or presence of 40 mM arsenate. lome results in general and the lack of evidence for (alpha-)arsenylated nucleotides in particular, we did not find evidence for significant incorporation of arsenic into GFAJ-1 grown nucleic acids. The amount of arsenic in nucleic acid Sample GFAJ-1 grown with 9 μM P with 9 μM P preparations was found to be below detection limit (Table 2), resulting in an arsenic to phosphorus ratio of less 40 mM As than 1 ‰, confirming that metabolism of GFAJ-1 is Nucleic acids (ng) 15,700 ±1,800 17,400 ±1,900 essentially based on phosphorylated compounds. In conclusion, our experiments indicate that GFAJ-1 Phosphorus in nucleic acids (ng) Arsenic in nucleic acids (ng) 934 ±13 <1# 1,043 ±8 <1# requires phosphorus for growth. We did not find evidence that arsenate can replace phosphate; however, in line with previous results, we confirm that GFAJ-1 is As/P molar ratio <0.001 <0.001 able to grow in the presence of very high arsenate and limiting phosphate concentrations (arsenate:phosphate #Below detection limit. ≈1:10,000), in contrast to other arsenate resistant strains that require much higher phosphate concentrations (typi- cally arsenate:phosphate ≈1:10-1:100, Aspergillus sp. To probe the metabolic plasticity of GFAJ-1, we analyzed its metab- P37 ≈1:1,000) (16–18). Phosphorus incorporation studies and metabo- olome in response to perturbation with arsenate in the absence of phos- lome analysis indicated that the core metabolism of GFAJ-1 is based on phate, and without pre-adaption of the organism to high-arsenate phosphorylated metabolites, even when cells are grown at high concen- concentrations. Cultures of phosphate pre-grown GFAJ-1 cells were trations of arsenate (40 mM) and low concentrations of phosphate (be- exchanged with medium that contained 40 mM arsenate, but no (i.e., low 10 μM). Feeding of arsenate to GFAJ-1 leads to the abiotic <0.3 μM) phosphate, and incubated for 6 and 360 s, respectively. To formation of some arsenylated compounds (hexose-arsenates); yet, most avoid depletion of glycolysis metabolite pool sizes that occurred during of the phosphorylated metabolite pools do not change upon arsenate stringent washing (see above), the standard washing protocol was ap- feeding. We note that the abiotic formation of hexose-arsenates observed plied, resulting in the detection of abiotically formed arsenylated hexos- by us might explain the results of Wolfe-Simon et al. in which the intra- es, as expected (Table 1). However, all other central metabolites were cellular accumulation of arsenate organo-esters by GFAJ-1 was indicat- only detected in their phosphorylated form and their pool sizes remained ed by secondary ion mass spectrometry, and x-ray analysis (1). virtually unchanged upon arsenate-incubation (Table 1). Most notably, However, the fact that these abiotically formed arseno-compounds ap- neither the levels of trisphosphate nucleotides (e.g., ATP), nor the levels parently were not metabolized by GFAJ-1 and that virtually no arsenic of glycolysis-derived phosphorylated metabolites (e.g., glucose- was detected in nucleic acid preparations of GFAJ-1 is in line with our phosphate, fructose-1,6-bisphosphate) indicating energy status as well as conclusions that GFAJ-1 is an arsenate-resistant, yet phosphate- metabolic activity of the cell changed (15). This suggested that, inde- dependent organism. The molecular basis for arsenate resistance in pendent of its pre-adaption to arsenate, the pool sizes of core metabolites GFAJ-1 might be the subject of further investigations, in particular given in GFAJ-1 are not affected by perturbation through arsenate or hexose- the finding that the concentrations of arsenic and phosphorous were arsenate. At the same time, our experiments also did not indicate that found within the same order of magnitude in the cellular fraction of GFAJ-1 possess a detectable downstream metabolism of these abiotical- growing GFAJ-1 cultures, despite the ratio of phosphate and arsenate in ly formed arseno-hexoses. the medium differing by four orders of magnitude (fig. S3). When our search for arseno-metabolites was extended beyond core metabolism, a total of six potentially arsenylated compounds were de- References and Notes tected in extracts of GFAJ-1, although most of them only at low abundance, and not in every experiment or replicate (table S7). Note that two of the corresponding phosphate analogs are also not known to be of biotic origin, according to the EcoCyc-metabolite database (12). In order to understand whether formation of these compounds was a feature specific to GFAJ-1 or might be a more general phenomenon, we examined the metabolome of Escherichia coli in response to arsenate perturbation. In contrast to GFAJ-1, E. coli is able to cope only with low arsenate con- 1. F. Wolfe-Simon et al., A bacterium that can grow by using arsenic instead of phosphorus. Science 332, 1163 (2011). doi:10.1126/science.1197258 Medline 2. S. A. Benner, Comment on “A bacterium that can grow by using arsenic instead of phosphorus”. Science 332, 1149-c (2011). doi:10.1126/science.1201304 Medline 3. D. W. Borhani, Comment on “A bacterium that can grow by using arsenic instead of phosphorus”. Science 332, 1149-e (2011). doi:10.1126/science.1201255 Medline 4. J. B. Cotner, E. K. Hall, Comment on “A bacterium that can grow by using centrations, when phosphate limited (16, 17), and has not been reported arsenic instead of phosphorus”. Science 332, 1149-f (2011). to incorporate arsenate into its biomass. However, similarly to GFAJ-1, some putatively arsenylated compounds were detected at low abundance in metabolome extracts of arsenate-incubated E. coli (table S7), indicating that the presence of such low abundance, putatively arsenylated compounds is not specific to GFAJ-1 and thus might not be of physiological relevance. Since these compounds are apparently unlinked to central carbon metabolism, they might also have formed abiotically. The apparent absence of arsenylated nucleotides and desoxynucleo- doi:10.1126/science.1201943 Medline 5. I. Csabai, E. Szathmáry, Comment on “A bacterium that can grow by using arsenic instead of phosphorus”. Science 332, 1149-b (2011). doi:10.1126/science.1201399 Medline 6. P. L. Foster, Comment on “A bacterium that can grow by using arsenic instead of phosphorus”. Science 332, 1149-i (2011). doi:10.1126/science.1201551 Medline 7. S. Oehler, Comment on “A bacterium that can grow by using arsenic instead of phosphorus”. Science 332, 1149-g (2011). doi:10.1126/science.1201381 tides in GFAJ-1 under all conditions tested raises the question of how Medline and whether arsenate would find its way into larger biomolecules (e.g., 8. R. J. Redfield, Comment on “A bacterium that can grow by using arsenic RNA, DNA). To answer this question, total nucleic acids were extracted from GFAJ-1 cells grown in the presence of 10 μM phosphate with or without 40 mM arsenate (fig. S1) and analyzed for their arsenic and phosphorus content, respectively, by ICP-MS. In line with the metabo- instead of phosphorus”. Science 332, 1149-h (2011). doi:10.1126/science.1201482 Medline 9. B. Schoepp-Cothenet et al., Comment on “A bacterium that can grow by using arsenic instead of phosphorus”. Science 332, 1149-d (2011). / http://www.sciencemag.org/content/early/recent / 8 July 2012 / Page 4/ 10.1126/science.1218455 doi:10.1126/science.1201438 Medline 10. J. Liebig, in Die Grundsätze der Agrikultur-Chemie mit Rücksicht auf die in England angestellten Untersuchungen. (Friedrich Vieweg und Sohn Publ Co., Braunschweig, Germany, 1855). 11. P. Kiefer, N. Delmotte, J. A. Vorholt, Nanoscale ion-pair reversed-phase HPLC-MS for sensitive metabolome analysis. Anal. Chem. 83, 850 (2011). doi:10.1021/ac102445r Medline 12. I. M. Keseler et al., EcoCyc: a comprehensive database of Escherichia coli biology. Nucleic Acids Res. 39, (Database issue), D583 (2011). doi:10.1093/nar/gkq1143 Medline 13. S. A. Moore, D. M. Moennich, M. J. Gresser, Synthesis and hydrolysis of ADP-arsenate by beef heart submitochondrial particles. J. Biol. Chem. 258, 6266 (1983). Medline 14. R. Lagunas, Sugar-arsenate esters: thermodynamics and biochemical behavior. Arch. Biochem. Biophys. 205, 67 (1980). doi:10.1016/00039861(80)90084-3 Medline 15. D. E. Atkinson, G. M. Walton, Adenosine triphosphate conservation in metabolic regulation. Rat liver citrate cleavage enzyme. J. Biol. Chem. 242, 3239 (1967). Medline 16. G. R. Willsky, M. H. Malamy, Effect of arsenate on inorganic phosphate transport in Escherichia coli. J. Bacteriol. 144, 366 (1980). Medline 17. G. R. Willsky, M. H. Malamy, Characterization of two genetically separable inorganic phosphate transport systems in Escherichia coli. J. Bacteriol. 144, 356 (1980). Medline 18. D. Cánovas, C. Durán, N. Rodríguez, R. Amils, V. de Lorenzo, Testing the limits of biological tolerance to arsenic in a fungus isolated from the River Tinto. Environ. Microbiol. 5, 133 (2003). doi:10.1046/j.14622920.2003.00386.x Medline Acknowledgments: We thank R. Oremland and J. S. Blum for providing strain GFAJ-1, as well as an anonymous reviewer for valuable suggestions and comments on our manuscript. T.J.E. was supported through an ETH Fellowship. Supplementary Materials www.sciencemag.org/cgi/content/full/science.1218455/DC1 Materials and Methods Figs. S1 and S3 Tables S1 to S8 References 27 December 2011; accepted 1 June 2012 Published online 8 July 2012 10.1126/science.1218455 / http://www.sciencemag.org/content/early/recent / 8 July 2012 / Page 5/ 10.1126/science.1218455