COMMENTARY

COMMENTARY

Unraveling the inner workings of respiratory
arsenate reductase
John F. Stolza,1 and Partha Basub,1

It began back in 1994 with a short note in the journal
Nature, about a curious bacterium from the Aberjona
watershed, strain MIT-13, that could grow on arsenic (1).
Arsenic resistance had been well established, as it had
been found in many clinical species like Staphylococcus
aureus and Escherichia coli. Arsenite oxidation linked to
resistance was also known. However, this was different—
the organism, later to be named Sulfurospirillum
arsenophilum (2), could couple the oxidation of lactate
to the reduction of arsenate [As(V)] for growth. More
significantly, incubation experiments revealed dissolution and reduction of arsenic from the sediments. This
suggested that such organisms could be responsible for
mobilizing arsenic in aquifers, resulting in the poisoning
of millions of people worldwide. As more researchers
began investigating microbial arsenic metabolism, it became apparent that not only was there a robust biogeochemical cycle, but arsenic played a role in the evolution
of life on Earth (3). Thus began the journey to decipher
how these organisms were capable of harnessing the
energy from arsenic oxyanions. Now, Glasser et al. (4)
have found the holy grail by solving the crystal structure
of the respiratory arsenate reductase, Arr, from Shewanella sp. ANA-3. In doing so, they answer several questions, including the type of iron sulfur clusters in both
ArrA and ArrB, the coordinating ligand to the molybdenum (cysteine, as had been predicted), the nature of
the catalytic pocket, the binding of arsenic to the
reaction site, and the mechanism of substrate transformation. Furthermore, they propose a reasonable solution to the preferred electron flow in these bidirectional
enzymes.
The respiratory arsenate reductase, Arr, was established both biochemically and through gene knockout
studies (5–7). Once key conserved regions in the protein sequence were recognized, more examples were
found in both Bacteria and Archaea. Although the operon structure varied, with different anchoring subunits (e.g., ArrC, CymA), chaperone proteins (ArrD),
and regulatory elements, the core enzyme, consisting
of the large catalytic subunit ArrA and the smaller

S

S
- e+ e-

S

S

S

S

S

e-

+ eS

Fig. 1. Redox noninnocence of dithiolene unit in the pyranopterin cofactor.

iron–sulfur cluster subunit ArrB, was highly conserved
(8). Interestingly, the need for both subunits for enzyme activity was found regardless of the organism
(5–7). So one of the first mysteries solved in the current
work was the significance of the fourth iron–sulfur cluster [4Fe4S] in ArrB (the binding site for the electron
transfer partner, in this case, methyl viologen).
Another important feature was in the large subunit.
Like other members of the DMSO reductase family of
enzymes, the catalytic Mo center of ArrA is coordinated
by two equivalents of the pyranopterin cofactor (present as the guanine dinucleotide). In addition to the
four sulfur atoms from the pyranopterin cofactor, the
Mo center is coordinated by a cysteine sulfur (C193)
similar to DMSO reductase with its serinato oxygen
coordination (9). In addition, a long Mo–O bond has
been detected that is presumably due to the reduction
of a Mo=O unit in the synchrotron radiation. This completes six coordination sites about the Mo center with a
distorted octahedral overall geometry as opposed to
trigonal prismatic structures found in DMSO reductase
and polysulfide reductase (PsrA). The composition of
the first coordination sphere is very similar to that of
polysulfide reductase (10) and the initial structure
of periplasmic nitrate reductase (NapA) (11) that has
subsequently been revised (12). Thus, compositionally
the Mo center is similar to other members of the DMSO
reductase family.
A related enzyme arsenite oxidase, Aio, catalyzes
the oxidation of arsenite. The crystal structures of Aio
from two different organisms, Rhizobium sp. NT-26
and Alcaligenes faecalis, have shown no coordination
of a protein-based ligand to the Mo center (13, 14). It

a

Department of Biological Sciences, Duquesne University, Pittsburgh, PA 15282; and bDepartment of Chemistry and Chemical Biology, Indiana
University–Purdue University Indianapolis, Indianapolis, IN 46202
Author contributions: J.F.S. and P.B. wrote the paper.
The authors declare no conflict of interest.
Published under the PNAS license.
See companion article on page E8614.
1
To whom correspondence may be addressed. Email: stolz@duq.edu or basup@iupui.edu.
Published online August 27, 2018.

www.pnas.org/cgi/doi/10.1073/pnas.1812841115

PNAS | September 11, 2018 | vol. 115 | no. 37 | 9051–9053

has long been suggested that different amino acid coordination
sets the substrate specificity. Reactivity toward other substrates
has been examined for a select few enzymes in this class. For example, NapA from Paracoccus denitrificans cannot reduce arsenate
and ArrA from Bacillus selenitisreducens MLS 10 or Shewanella strain
ANA 3 cannot reduce nitrate, even though their immediate coordination environments are very similar. Even more striking is that the
related bidirectional enzyme system, the ArxAB complex, whose
coordination environment should be very similar to ArrAB, can carry
out the reverse of the physiological reaction, oxidation of arsenite to
arsenate, while AioAB whose physiological function to oxidize arsenite cannot reduce arsenate (15). These differences in reactivity cannot be completely understood considering the nature of the first
coordination sphere alone. The current paper (4) discusses the presence of a second binding site that may be used for phosphate
binding. The presence of a second binding site provides an excellent opportunity in understanding the substrate transformation.
One of the intriguing aspects of the crystallographic studies is
that the conformation of the pyranopterin cofactor changes upon
binding of the substrate, arsenate. The S–C–C–S fragment exhibits a small but definable deviation from planarity upon substrate binding that has been interpreted as due to a reduction
of the dithiolene fragment. Dithiolene ligands are well known
for their redox noninnocent behavior. In the fully reduced state,
they can undergo two one-electron oxidations yielding the fully
oxidized dithione form; during the process, the S–C–C–S torsion
angle can change (Fig. 1). The conformational changes reported
in ArrAB structures underscore the involvement of pyranopterin
cofactor in the reaction process.
The dithiolene is not the only redox active unit in the
pyranopterin cofactor, as the pterin itself can exhibit a variety of
redox states from the fully reduced tetrahydro to the partially
reduced dihydro to the fully oxidized state (16). The conformation
of the pyranopterin cofactor provides a hint to the redox state
with the tetrahydro state being more distorted than the dihydro.
The conformational analyses of the cofactor in crystallographically
characterized enzymes have provided a basis for establishing the
redox states in those enzymes (17, 18). In this case, the conformation change indicates the role of the pyranopterin cofactor, at
least the P pterin (the pterin distal from the partner [4Fe4S] prosthetic group) in electron transfer.
The substrate transformation of Arr occurs at the Mo center,
and the proposed mechanism involves a direct oxygen atom
transfer reaction from the substrate to the reduced Mo(IV) center
(4). The reaction scheme is consistent with that proposed for
DMSO reductase (9). The catalytic center is regenerated by two
one-electron transfer reactions from the ArrB subunit. The crystal
structure of the ArrAB complex provides a roadmap for the electron transfer pathway in the enzyme. The smaller ArrB subunit
shows the arrangement of four [4Fe4S] clusters that are electron
transfer partners of the Mo center.
To date, no EPR signal attributable to a Mo(V) species has been
reported for Aio complexes. The electrochemical behavior of this
system examined by voltammetry indicates that the thermodynamically uphill redox step at the Mo center is an obligate twoelectron process. In contrast, the ArrAB complex reported here

exhibits EPR signals due to Mo(V) species by reducing the fully
oxidized enzyme with a mixture of dithionite, arsenate, and methyl
viologen. The Mo(V) species exhibit near axial EPR features. A
small hyperfine structure was detectable in ENDOR and pulsed
EPR experiments conducted in the Q-band. The small hyperfine
was interpreted as due to the β-proton present in coordinating
cysteine, not to a Mo–OH species. This suggests that the Mo–OH
intermediate is not accumulating in an appreciable quantity during
the experiment. With a higher concentration of methyl viologen,
a Mo–arsenite species was observed in the EPR experiments.
The EPR signals due to the iron–sulfur clusters were observed as

The current work by Glasser et al. marks a
milestone in the structure/function of arsenic
enzymes. It has established a clear mechanistic
difference between the classic arsenite oxidase
(Aio) and respiratory reductase (Arr) and provided
a model from which to investigate Arx.
expected. The observation of the EPR signals strongly supports
the notion that mechanistically Aio and Arr behave differently.
Microbial activity can have a significant impact on the toxicity
and mobility of arsenic (3). Although arsenate respiring bacteria
are typically grown in media containing millimolar amounts of
As(V), in the environments in which they live the concentration is
often micromolar or the arsenic is bound in the sediment. Interestingly, all three enzymes that have been linked to respiration
(e.g., Aio, Arr, and Arx) are topologically located on the outer side
of cytoplasmic membrane, facing into the periplasmic space. Arr
has both high affinity (Km of 44.6 μM) and fast kinetics (kcat of
∼10,000 s−1), more than compensating for the less than optimal
orientation for the generation of a proton gradient. Glasser et al.
surmise that the rate-limiting step is either other components of
the electron transfer chain (e.g., CymA, ArrC), or the diffusion rate
of solid-phase arsenic (4). Interestingly, AioAB from Rhizobium sp.
NT-26 also functions near the diffusion control limit (kcat/Km ∼ 108),
while Aio from A. faecalis is less efficient (kcat/Km ∼ 106) (9). This
suggests that the respiratory enzymes function at a higher efficiency than the resistance enzymes. It will be interesting to see
whether the arsenite oxidases involved in light-dependent arsenite oxidation (e.g., Aio in Chloroflexus aurantiacus, Arx in Ectothiorhodospira spp.) behave similarly.
It has indeed been a remarkable journey over the past two and
a half decades, one in which our understanding of arsenic in the
environment and in microbiology has changed dramatically. The
current work by Glasser et al. marks a milestone in the structure/
function of arsenic enzymes. It has established a clear mechanistic
difference between the classic arsenite oxidase (Aio) and respiratory reductase (Arr) and provided a model from which to investigate
Arx. Furthermore, their advancement in overexpression of native
proteins should prove extremely useful in the investigation of
complex, multisubunit, metalloenzymes.

1 Ahmann D, Roberts AL, Krumholz LR, Morel FM (1994) Microbe grows by reducing arsenic. Nature 371:750.
2 Stolz JF, et al. (1999) Sulfurospirillum barnesii sp. nov. and Sulfurospirillum arsenophilum sp. nov., new members of the Sulfurospirillum clade of the e
Proteobacteria. Int J Syst Bacteriol 49:1177–1180.
3 Oremland RS, Stolz JF (2003) The ecology of arsenic. Science 300:939–944.

9052 | www.pnas.org/cgi/doi/10.1073/pnas.1812841115

Stolz and Basu

4 Glasser NR, Oyala PH, Osborne TH, Santini JM, Newman DK (2018) Structural and mechanistic analysis of the arsenate respiratory reductase provides insight into
environmental arsenic transformations. Proc Natl Acad Sci USA 115:E8614–E8623.
5 Krafft T, Macy JM (1998) Purification and characterization of the respiratory arsenate reductase of Chrysiogenes arsenatis. Eur J Biochem 255:647–653.
6 Afkar E, et al. (2003) The respiratory arsenate reductase from Bacillus selenitireducens strain MLS10. FEMS Microbiol Lett 226:107–112.
7 Saltikov CW, Newman DK (2003) Genetic identification of a respiratory arsenate reductase. Proc Natl Acad Sci USA 100:10983–10988.
8 Stolz JF, Basu P, Santini JM, Oremland RS (2006) Arsenic and selenium in microbial metabolism. Annu Rev Microbiol 60:107–130.
9 Hille R, Hall J, Basu P (2014) The mononuclear molybdenum enzymes. Chem Rev 114:3963–4038.
10 Jormakka M, et al. (2008) Molecular mechanism of energy conservation in polysulfide respiration. Nat Struct Mol Biol 15:730–737.
11 Dias JM, et al. (1999) Crystal structure of the first dissimilatory nitrate reductase at 1.9 Å solved by MAD methods. Structure 7:65–79.
12 Najmudin S, et al. (2008) Periplasmic nitrate reductase revisited: A sulfur atom completes the sixth coordination of the catalytic molybdenum. J Biol Inorg Chem
13:737–753.
13 Anderson GL, Williams J, Hille R (1992) The purification and characterization of arsenite oxidase from Alcaligenes faecalis, a molybdenum-containing hydroxylase.
J Biol Chem 267:23674–23682.
14 Warelow TP, et al. (2013) The respiratory arsenite oxidase: Structure and the role of residues surrounding the rieske cluster. PLoS One 8:e72535.
15 Richey C, et al. (2009) Respiratory arsenate reductase as a bidirectional enzyme. Biochem Biophys Res Commun 382:298–302.
16 Basu P, Burgmayer SJN (2011) Pterin chemistry and its relationship to the molybdenum cofactor. Coord Chem Rev 255:1016–1038.
17 Rothery RA, Stein B, Solomonson M, Kirk ML, Weiner JH (2012) Pyranopterin conformation defines the function of molybdenum and tungsten enzymes. Proc Natl
Acad Sci USA 109:14773–14778.
18 Sparacino-Watkins C, Stolz JF, Basu P (2014) Nitrate and periplasmic nitrate reductases. Chem Soc Rev 43:676–706.

Stolz and Basu

PNAS | September 11, 2018 | vol. 115 | no. 37 | 9053