Structural And Molecular Genetic Insight Into A Widespread Sulfur Oxidation Pathway

We proved with aΔdsrE mutant experiment that the cytoplasmic α2β2γ2-structured protein DsrEFH is absolutely essential for the oxidation of sulfur stored in the intracellular sulfur globu

Structural and Molecular Genetic Insight into a

Widespread Sulfur Oxidation Pathway

Christiane Dahl1, Andrea Schulte1, Yvonne Stockdreher1,

1Institut für Mikrobiologie &

Biotechnologie, Rheinische

Friedrich-Wilhelms-Universität

Bonn, Meckenheimer Allee 168,

D-53115 Bonn, Germany

2Department of Chemistry,

University of California,

Berkeley, CA 94720-5230, USA

3Physical Biosciences Division,

Lawrence Berkeley National

Laboratory, Berkeley, CA 94720,

USA

4College of Pharmacy,

Ewha Womans University,

Seoul 120-750, Korea

Received 16 July 2008;

received in revised form

27 September 2008;

accepted 3 October 2008

Available online

15 October 2008

Many environmentally important photo- and chemolithoautotrophic bac-teria accumulate globules of polymeric, water-insoluble sulfur as a transient product during oxidation of reduced sulfur compounds Oxidation of this sulfur requires the concerted action of Dsr proteins However, individual functions and interplay of these proteins are largely unclear We proved with

aΔdsrE mutant experiment that the cytoplasmic α2β2γ2-structured protein DsrEFH is absolutely essential for the oxidation of sulfur stored in the intracellular sulfur globules of the purple sulfur bacterial model organism Allochromatium vinosum The ability to degrade stored sulfur was fully regained upon complementation with dsrEFH in trans The crystal structure

of DsrEFH was determined at 2.5 Å resolution to assist functional assign-ment in detail In conjunction with phylogenetic analyses, two different types of putative active sites were identified in DsrE and DsrH and shown

to be characteristic for sulfur-oxidizing bacteria Conserved Cys78 of A vinosum DsrE corresponds to the active cysteines of Escherichia coli YchN and TusD TusBCD and the protein TusE are parts of sulfur relay system involved in thiouridine biosynthesis DsrEFH interacts with DsrC, a TusE homologue encoded in the same operon The conserved penultimate cys-teine residue in the carboxy-terminus of DsrC is essential for the inter-action Here, we show that Cys78 of DsrE is strictly required for interaction with DsrC while Cys20 in the putative active site of DsrH is dispensable for that reaction In summary, our findings point at the occurrence of sulfur transfer reactions during sulfur oxidation via the Dsr proteins

© 2008 Elsevier Ltd All rights reserved

Edited by M F Summers

Keywords: DsrEFH; dissimilatory sulfur oxidation; crystal structure; anoxygenic phototrophic sulfur bacteria YchN fold; dissimilatory sulfite reductase

Introduction

Reduced sulfur compounds such as sulfide and

thiosulfate are oxidized by a large and diverse group

of prokaryotes, including the phototrophic sulfur

bacteria, the thiobacilli, and other chemotrophic

sulfur bacteria and some thermophilic archaea

Typi-cally, these sulfur compounds are oxidized to sulfate,

but in many cases, globules of polymeric,

water-insoluble sulfur accumulate as a transient product The sulfur can be deposited outside of the cell as is the case for green sulfur bacteria On the other hand, purple sulfur bacteria of the family Chromatiaceae store sulfur globules inside the cells They have this trait in common not only with a large number of environmentally important free-living chemotrophic sulfur oxidizers such as Beggiatoa, Thioploca, or magnetotactic bacteria but also with sulfur-oxidizing bacterial symbionts of marine animals such as Riftia pachyptila or Olavius algarvensis It is very important

to note that the sulfur resides in the bacterial peri-plasm in the purple sulfur bacterial model organism Allochromatium vinosum and in many if not all other bacteria forming intracellular sulfur globules.1,2

Bio-*Corresponding author E-mail address:

dhshin55@ewha.ac.kr

Abbreviations used: MTM, Methanothermobacter,

Thermotoga, and Moorella; PDB, Protein Data Bank

Available online at www.sciencedirect.com

0022-2836/$ - see front matter © 2008 Elsevier Ltd All rights reserved.

chemical data, genetic studies with A vinosum, and

genome comparisons indicate that in all these

orga-nisms as well as in green sulfur bacteria and

thio-bacilli, a complicated pathway is at work, involving

transport of sulfur carrier molecules from outside the

cells or the periplasm into the cytoplasm and

re-quiring the presence of many different enzymes

including sulfite reductase (DsrAB).2,3

In A vinosum, several proteins encoded in the dsr

gene cluster (Fig 1a) have been shown to be

essential for further oxidation of stored sulfur to

the end product sulfate.4–8 The Dsr proteins are

either cytoplasmic or membrane-bound It is

pro-posed that sulfur is transported into the cytoplasm

in a persulfidic form, possibly as glutathione amide

persulfide.3,6,9–11Once in the cytoplasm, the sulfane

sulfur has to be made available to sulfite reductase,

which oxidizes it to sulfite The siroheme-containing

sulfite reductase specifically interacts with the

membrane-bound electron-transporting DsrMKJOP

complex7 that may feed electrons into

photosyn-thetic electron transport Such a pathway would be

analogous to that postulated for dissimilatory

sulfate-reducing bacteria,12operating in the reverse

direction DsrC, a protein with two conserved

carboxy-terminal cysteine residues (Cys100 and

C111), has been discussed to be involved in electron

transfer between DsrAB and DsrMKJOP via

thiol-disulfide switches.3,6,7 Recently, it has been shown

that the DsrC protein from the sulfate reducer

Desulfovibrio vulgaris can be bound in a cleft between

DsrA and DsrB with the cysteine corresponding to

Cys111 A vinosum DsrC reaching the distal side of

the active-site siroheme On this basis, it has been

proposed that DsrC is involved in the catalytic

reaction as a product-binding protein and that a

persulfide of DsrC is a crucial intermediate in the

reduction of sulfite.13 The protein DsrEFH occurs

exclusively in sulfur oxidizers.9 In Escherichia coli,

the DsrEFH-related protein TusBCD and the DsrC

homologous protein TusE are firmly established

parts of a sulfur relay system during thiouridine

biosyntheses.14 On this background, the recently

documented interaction of A vinosum DsrEFH and

DsrC led to the suggestion of an alternative model

for intracellular sulfur oxidation implying DsrEFH

and DsrC as parts of sulfur trafficking between

persulfidic sulfur imported into the cytoplasm and

sulfite reductase.6

DsrEFH is a soluble, cytoplasmic α2β2γ2 -struc-tured holoprotein with an apparent molecular mass

of 75 kDa.7The polypeptides DsrE, DsrF, and DsrH are homologous to each other (Fig 3) DsrE and DsrF are the prototypes of a family of conserved domains (Pfam 02635.11, COG 1553, COG 2044, COG 2923) DsrH is the prototype of yet another family of conserved proteins found in bacteria and archaea (Pfam04077.6; COG 2168) However, DsrH also fits into the DsrE/F family Structural infor-mation on representatives of the DsrH family of proteins is available through the work of Shin et al

on YchN from E coli,15 Gaspar et al on Tm0979 from Thermotoga maritima,16 Christendat et al on MTH1491 from Methanobacterium thermoautotro-phicum,17and Numata et al on E coli TusBCD.18 In contrast to DsrEFH and TusBCD, all others form homooligomers YchN is present as two rings of trimers, MTH1491 as a trimer, and Tm0979 as a dimer Except Tm0979, all of these proteins harbor conserved cysteine residues in a probable active-site region

In our effort to further dissect the functions of the proteins encoded at the A vinosum dsr locus and to test the existing models for the dsr-encoded sulfur oxidation pathway, we firstly constructed an

A vinosum mutant with an in-frame deletion of dsrE, complemented the dsrEFH genes in trans, and studied the resulting phenotypes regarding sulfur oxidation Secondly, we determined the three-dimensional structure of DsrEFH by X-ray crystal-lography Furthermore, we determined the site of interaction with DsrC via site-directed mutagenesis

of putative active-site cysteines in DsrE and/or DsrH

Results

Biological significance of DsrEFH

In order to examine the importance of DsrEFH for sulfur oxidation, we first deleted the complete dsrEFH genes However, the resulting A vinosum mutant turned out to be genetically unstable, most probably due to the deletion of the promoter of the constitutively expressed dsrC present in dsrF.7,8The dsrC gene cannot be stably deleted from A vinosum,

Fig 1 Schematic overview of the dsr locus of A vinosum Genes that have been proven to be individually essential for sulfur oxidation by in-frame deletion mutagenesis4,5are shown in black Absolute requirement of DsrE is proven in this study DsrN (light gray), a probable siroamide synthase providing the prosthetic group for DsrAB sulfite reductase, is important though not absolutely essential.5The dsrC gene is marked with an asterisk In-frame deletion of this gene leads

to a genetically unstable mutant, indicating that dsrC is indispensable in A vinosum even in the absence of reduced sulfur compounds.6

indicating that its product is essential for central

metabolic pathways in this organism.6 Therefore,

we deleted solely dsrE, leaving the promoter for dsrC

intact The A vinosumΔdsrE mutant was genetically

stable even after prolonged incubation In order to

examine the phenotype of A vinosum ΔdsrE, we

cultivated the strain photolithoautotrophically in

batch culture with sulfide as electron source As

expected for a classical purple sulfur bacterium,19

the sulfide concentration immediately decreased

and intracellular sulfur was formed The rate of

sulfide oxidation to sulfur was unaffected in the

ΔdsrE mutant During the oxidation of sulfide to

sulfur of oxidation state zero, two different

poly-sulfides are formed as intermediates by A vinosum

wild type.20The formation of both polysulfides was

not affected in theΔdsrE mutant (not shown) In the

wild type, stored sulfur is further oxidized to sulfate

when sulfide is depleted.8 In contrast, the ΔdsrE

mutant was completely unable to oxidize the

accumu-lated sulfur (Fig 2a) Another unambiguous indicator

for the inability of the mutant strain to oxidize

stored sulfur was the complete lack of sulfate as the

final product of sulfur oxidation (Fig 2b)

Further-more, an accumulation of intermediates en route to

sulfate, for example, sulfite, was discounted by

HPLC analysis Complementation of dsrEFH in trans

completely restored the mutant sulfur oxidation to

that of the wild type (Fig 2a) Sulfate was again

formed as the final product (Fig 2b) This

experi-ment confirmed that the observed phenotype was

indeed caused by the specific loss of dsrE and

consequently a lack of the DsrEFH protein Growth

under photoorganoheterotrophic conditions was

not influenced in the A vinosum ΔdsrE strain In

summary, the phenotype of the studied single-locus

dsrE mutant clearly demonstrates the vital

impor-tance of the DsrEFH protein for the oxidation of

stored sulfur in A vinosum

DsrEFH-related proteins are widespread and

form distinct groups

A whole array of bacteria and also some archaea

contain dsrEFH homologous genes located

immedi-ately adjacent to each other The function of the

encoded proteins is probably variable: An in silico

alignment of DsrEFH fusion proteins revealed that

they can be subdivided into distinct groups

char-acterized by the number and position of conserved

cysteine residues in putative active-site regions

(Fig 3) A vinosum DsrEFH belongs to a group of

deduced proteins, which contain conserved cysteine

residues only in DsrE (Cys78) and DsrH (Cys20)

Notably, the organisms containing these proteins

include Thiobacillus denitrificans, the green sulfur

bacteria, and the magnetotactic bacteria All of these

organisms are well-known sulfur-oxidizing bacteria

In addition, genome sequence data have been

claimed to suggest that Methylococcus capsulatus,

another member of this group, is also capable of

chemolithotrophic sulfur oxidation.21 In all of the

established sulfur oxidizers mentioned above, the

DsrEFH proteins are encoded in immediate vicinity

of other dsr genes (including dsrAB encoding dissimilatory sulfite reductase) similar to the situa-tion in A vinosum.7From this observation, we con-clude that these proteins form a homogeneous phy-siological group We can therefore state that the presence of one conserved putative active-site cysteine in each, DsrE and DsrH, is a common and typical property of DsrEFH proteins from sulfur-oxidizing bacteria In all these organisms, the dsrEFH genes are situated close to a dsrC gene

In order to investigate whether the different groups of DsrEFH-related proteins are also phylo-genetically distinct, we performed neighbor-joining, parsimony, likelihood, and Bayesian analyses of concatenated DsrE, F, and H sequences The occur-rence of conserved cysteine residues fits well with

Fig 2 Sulfur accumulation and oxidation (a) and sulfate production (b) by A vinosum wild type (▵),

A vinosumΔdsrE (▪), and A vinosum ΔdsrE+dsrE (●) Cells were grown photolithoautotrophically in batch cul-ture in the presence of 2 mM sulfide Sulfide is not com-pletely recovered as sulfate due to loss of gaseous H2S during sampling.8Sulfide and the polysulfides formed as intermediates during the formation of sulfur globules from sulfide are not shown for clarity Sulfite and thiosulfate were not detected during degradation of sulfur globules in any of the cultures Protein concentrations at the onset and at the end of the experiments were 50 and

70 μg, respectively Representative growth experiments for each strain are shown

the position of the respective protein in the

phylo-genetic tree (Fig 4), supporting the notion that

different physiological functions correlate with the

presence of certain conserved cysteine residues As

expected, the DsrEFH-related proteins of known

sulfur oxidizers are affiliated to each other

Surpris-ingly, this clade cannot be regarded as monophyletic

since the proteins of Methanothermobacter,

Thermo-toga, and Moorella (MTM clade) are nested within

this clade, regardless of method or data set used

(Fig 4) Only neighbor-joining trees show the MTM

clade as a sister group to the sulfur-oxidizing

pro-karyotes Since the MTM proteins are very different

at the sequence level from all other proteins

in-cluded in this analysis, we suppose this to be a result

of a phenomenon called“long branch attraction”.22

As a control and in order to exclude independent

evolution of the dsrE, dsrF, and dsrH genes, detailed

phylogenies for the single genes were also

calcu-lated The resulting trees were similar but not

absolutely identical Some variations were observed

concerning the positions of subbranches; however,

such minor differences also occurred when the

different methods (Bayesian analysis, maximum

parsimony, and neighbor joining) were compared

In all cases, the same major groups were observed;

that is, the proteins from the sulfur oxidizers always

group together

It has to be noted that the members of the genus

Pseudomonas as well as Shewanella oneidensis,

Chro-mohalobacter salexigens, Oceanobacter sp., M

capsula-tus, and Idiomarina loihiensis possess a dsrC-like gene

directly adjacent to the dsrEFH-related genes This

points at functional linkage between DsrEFH-like

and DsrC-like proteins in these organisms Bioche-mical studies in E coli14 and A vinosum6,7 already provided evidence for such an interaction The presence of a conserved cysteine residue in DsrH might point at a so far unidentified specific reaction partner

Quality of the model and overall structure of DsrEFH

In an attempt to gain more insight into the function

of A vinosum DsrEFH, its X-ray crystal structure was determined There were three α2β2γ2-structured DsrEFHs (three heterohexamers) in the asymmetric unit In the final refined models to 2.5 Å resolution, all residues of DsrE and all residues of DsrH except the first methionine are included However, the first four residues and the eight residues between 103 and 110 are undefined in the electron density map of DsrF The average B-factors for main-chain and side-chain atoms are 32.1 and 35.6 Å2, respectively Table 1 summarizes refinement statistics All residues except Tyr21 of DsrE and Tyr21 of DsrF lie in the allowed region of the Ramachandran plot produced with PROCHECK.24In the crystal structure of TusBCD, a structural homologue of DsrEFH, His13 of TusC, which corresponds to the tyrosines (Tyr21) of DsrE and DsrF, also lies in the disallowed region of the Ramachandran plot

The Cα trace of the atomic model of the DsrEFH structure is shown inFig 5a Each of DsrE, DsrF, and DsrH consists of a single domain with a three-layer (αβ)-sandwich architecture The α2β2γ2-structured hexamer (Fig 5b) found in the asymmetric unit is

Fig 3 Sequence comparison among some of YchN fold members The three-dimensional structures of the listed members are known Abbreviations are as follows: YchN, YchN from E coli; 1X9A, PDB ID of Tm0979; 1L1S, PDB ID of Mth1491 The“-” represents a gap, “⁎” denotes identical residues, “:” indicates highly conserved residues, and “.” denotes less highly conserved residues The blue character H represents a sequence belonging toα-helices, a green G for 310 -helices, a pinkβ for β-strands, and a black L for loops The conserved cysteine residues reported in the YchN family15

are marked yellow

a biological form confirmed by analytical

size-exclusion column and dynamic light-scattering

ex-periments.27 The heterohexamer has approximate

dimensions of 75 Å × 60 Å × 55 Å

Oligomeric forms of DsrEFH Interestingly, all the subunits of DsrEFH have a YchN fold first found in E coli YchN.15 One of the

Fig 4 Bayesian tree of in silico fusions of DsrEFH-related proteins Node significances are given as posterior probabilities (first values) for the Bayesian analyses and as bootstrap support for the neighbor-joining (second values) and maximum parsimony analyses (third values) Names of organisms containing putative dsr operons similar to that of

A vinosum are printed in bold letters In the left part of the figure, the presence of conserved cysteine residues is indicated

by bars: DsrE1, residue 78; E2, residue 81; F1, residue 80; F2, residue 83; and H, residue 20 (numbering according to the respective A vinosum protein) In the outer left lane, the presence of a bar indicates that dsrC and dsrEFH genes are located

in vicinity to each other in the respective organism's genome Accession numbers of the DsrE-like proteins are given after the organism names Genes for DsrF and DsrH are generally found in the immediate vicinity of dsrE genes The accession numbers for the DsrE-like proteins of Moorella thermoacetica, T maritima, and M thermoautotrophicum are ZP_00330986, NP_228789, and AAB 85834, respectively

structural characteristics of the YchN fold is that a

large portion of the fold is involved in

oligomeriza-tion Oligomerization is accompanied by an increase

in the buried surface areas from∼550 to ∼1800 Å2

per monomer during hexamerization (Table 2)

Various charged interactions contribute to

stabiliza-tion of each subunit and various oligomeric forms

(Table 2) The trimeric form (Fig 5a) is stabilized

mainly by hydrophobic interactions between

sub-units Interestingly, many aromatic side chains are

involved in these interactions: (1) Phe3, Phe23, and

Phe128 of DsrE and Phe33, Tyr85, and Phe88 of

DsrH at the interfaces of DsrEH; (2) Tyr40 of DsrE

and Phe108, Phe235, Phe134, and Phe136 of DsrF

at the interfaces of DsrEF The interactions among

the last β-strands (β5s) of each subunit in the

center of the trimer also play a role to stabilize the

trimeric form through a water-mediated

interac-tion as first shown in the E coli YchN structure.15

Therefore, the trimeric DsrEFH forms a very stable

structure

In the hexameric form, there are two different

types of contacts at the interfaces of trimers (Fig 5b):

(1) DsrE–DsrE′ interaction and (2) DsrF–DsrH′ or

DsrH–DsrF′ interactions These types of contacts

result in unique subunit symmetry also found in the

TusBCD structure.18 Most of the contacts between

hemispheres are governed by the hydrophilic

inter-actions Therefore, many water molecules are found

between the interfaces of trimers At the core of the

hexamer, several charged residues (His14 and

Asp130 of DsrE and Lys9, Arg14, and Glu34 of

DsrH) form an ion pair network Since these residues are close to each other, a closed cavity is not formed at the core of the hexamer unlike the case

of E coli YchN.15The dimensions of the hexamer are

∼75 Å along the pseudo-triad axis and ∼60 Å across the 2-fold axis The lack of hydrophobic interactions between the two trimers may result in disassembly

of the hemispheres depending on surrounding con-ditions influencing these ionic interactions, such as

pH, salt, or protein concentration

Identification and mutational analysis of putative active sites in DsrEFH

Considering the E coli YchN structure, there should be a putative active site at the beginning of the H3 helix of each subunit.15DsrE has a conserved cysteine (Cys78) corresponding to the active cys-teines of E coli YchN at this position (Fig 6a and b) However, nonconserved residues, Asp83 and Gly63, are found in this position in the case of DsrF and DsrH, respectively Therefore, only DsrE has the putative active cysteine similar to that of E coli YchN The putative active site is formed between interfaces of DsrE and DsrF with a depth of∼11 Å and a width of ∼6 Å×14 Å where the highly con-served Tyr40 is also present

There is a large cleft on the equatorial interface lined up by two DsrF and two DsrH (Figs 5b and 7a) Generally, a long L3 loop of the YchN fold contributes L3–L3 loop interactions during hex-amerization on the outer equatorial surface.15

However, DsrH has a shortened L3 loop (Fig 5b), which results in the formation of a big cleft with a depth of ∼11 Å and a width of about 10 Å×30 Å

on the surface of DsrFH (Fig 7a) The bottom of the cleft is lined up by two L1 loops (Fig 5b) Interestingly, highly conserved residues, His5 and Trp101 of DsrH, are constellated around Cys20 of DsrH in this pocket As shown in Fig 4, Cys20 of DsrH is a conserved cysteine Therefore, these fea-tures strongly support that this big cleft may be another putative active site In summary, the DsrEFH structure reveals two different types of putative active sites, which is different from the case

of the E coli YchN structure where the six interfaces among the adjoining subunits contain the same recessed cavities along the equatorial surface of the hexamer

Recently, it has been shown that DsrEFH interacts with DsrC from A vinosum and that this interaction

is strictly dependent on the presence of the penul-timate conserved cysteine residue of DsrC.6We now set out to identify the site of interaction in the DsrEFH protein and exchanged either one or both putative active-site cysteine residues (Cys78 of DsrE and Cys20 of DsrH) to serine Interaction of the proteins was assessed using a band-shift technique under nondenaturing conditions (Fig 8) It clearly appeared that Cys78 of DsrE is absolutely required for the interaction, as interacting bands are not formed when Cys78 of DsrE alone or both Cys78 of DsrE and Cys20 of DsrH are mutated to Ser The

Table 1 Refinement parameters

Crystal parameters and refinement statistics

Cell dimensions 56.6 Å × 183.1 Å

× 107.8 Å, β=99.6°

Volume fraction of solvent (%) 42.6

Total number of residues 2124

Total non-H atoms 17,203

Number of Se atoms 30

Number of water molecules 418

Average temperature factors (Å 2 )

Resolution range of reflections

Amplitude cutoffa(σ) 0.0

Free R-factor (%) 25.6

Stereochemical ideality

Ramachandran plot (%)

Residues in most favored regions 91.8

Residues in additional allowed

regions

7.0 Residues in generously allowed

regions

0.7 Residues in disallowed regions 0.5

a Sigma cutoff in CNS during refinement 23

interaction is not prevented by the exchange of

Cys20 of DsrH to Ser

Structural differences between DsrEFH and

TusBCD coincide with divergent cellular

functions

Apart from the described similarities, striking

structural differences between DsrEFH and TusBCD

are also apparent In the primary structure, DsrEFH

has a much lower aliphatic index (89.02) than

TusBCD (105.64) The difference is caused by the

higher alanine content in TusBCD (13.5%) compared

with DsrEFH (9.2%) The high aliphatic index also results from the reduced content of charged residues

in TusBCD Actually, the total number of charged residues (aspartate, glutamate, lysine, and arginine)

is 96 in DsrEFH and 65 in TusBCD Therefore, the contribution of charged residues to structure and function may be severely decreased in the case of TusBCD

The tertiary structure difference is prominent bet-ween DsrF and TusC, though the core structure of both subunits is conserved DsrF has a large in-sertion on L5 (the undefined loop including residues

103 and 110), which is not present in the TusC

Fig 5 Crystal structure of DsrEFH (a) A Cα trace of DsrEFH DsrE, DsrF, and DsrH are represented by green, magenta, and yellow, respectively Every 20th residue is numbered and represented by a dot The N-terminus (Met1 of DsrE, Val5 of DsrF, and Ser2 of DsrH) and C-terminus (Asp130 of DsrE, F136 of DsrF, and Leu102 of DsrH) are labeled The figure was generated by MOLSCRIPT.25(b) Loop interactions in the second putative active pocket of hexameric DsrEFH L1 and L3 loops are labeled to indicate L1–L1 and L3–L3 interactions (see the text) The figures were generated using the program RIBBONS.26

sequence (Fig 3) The quaternary structure

differ-ence is also visible in the second putative active site

on the surface of DsrFH The size of the pocket of

DsrFH (Fig 7a) is much larger than that of TusBC

(Fig 7b) In this pocket, TusBC has a glutamate

(Glu85 of TusC) and a tyrosine (Tyr45 of TusB)

instead of an alanine (Ala89 in the case of DsrF) and

a valine (Val46 of DsrH) These bigger substitutions

narrow down the size of the pocket of TusBC In

addition, the conserved cysteine residue (Cys20 of

DsrH) found in the pocket of DsrFH is not present in

the sequence of TusB Instead, TusC has a conserved

cysteine (Cys79) in the corresponding pocket of

TusBC.18 Interestingly, Cys79 of TusC is part of a

CXXC motif similar to that of YchN Therefore, the

molecular function of the second putative active site

of DsrEFH is thought to be quite different from that

of TusBCD

Discussion

Here, we have shown via in-frame deletion

muta-genesis and complementation that DsrEFH is an

essential and central player in oxidative sulfur

meta-bolism More specifically, it is an absolutely essential

component for the oxidation of stored sulfur via the

Dsr system, a pathway occurring not only in

photo-trophic but also in many chemophoto-trophic

sulfur-oxidizing bacteria

Our structural characterization clearly places

DsrEFH into the YchN family Based on the

struc-ture of E coli YchN,15 several broad putative

molecular functions such as peroxiredoxin,

oxidor-eductase, and hydrolase have originally been

proposed for members of that family In addition,

an involvement in sulfur metabolism has been

suggested for MTH1491 from M

thermoautotrophi-cum mainly on the basis of gene arrangement.17As

outlined above, our comparison of primary to

quaternary structures of DsrEFH and E coli TusBCD

revealed a number of common characteristics (Fig

6) and the sulfur transferase activity of TusBCD14

finally provided a more direct clue to understand

the possible molecular function of DsrEFH TusBCD

is involved in 2-thiouridine biosynthesis in bacterial

tRNAs In a first step, the cysteine desulfurase IscS

obtains sulfur fromL-cysteine and this sulfur is then transferred to the protein TusA (former gene name

in the E coli genome: yhhP) Sulfur-activated TusA then binds to TusBCD (encoded by the former genes yheLMN), in which TusD accepts sulfur from the TusA persulfide TusD transfers persulfide sulfur to TusE (encoded by the former yccK), which finishes 2-thiouridine biosynthesis with the aid of MnmA The residues comprising the putative active site of DsrE are exactly identical with those of TusD (Fig 6b) It is therefore well possible that the principal molecular function of A vinosum DsrE and E coli TusD is conserved and a sulfur trans-ferase activity of DsrEFH through the conserved cysteine located on the primary active site appears likely This assumption is strongly supported by our finding that the active-site cysteine of DsrE is abso-lutely essential for interaction in vitro with DsrC (Fig 8) As mentioned above, TusD in TusBCD forms persulfide and then transfers persulfide sulfur

to TusE Not surprisingly, TusE is a homologue of DsrC.8,14 Since TusE and DsrC also share a con-served carboxy-terminal cysteine, their molecular role may be similar too A persulfide of the penul-timate conserved cysteine of DsrC has been pro-posed to be an important intermediate not only during the production of sulfite by the reverse-acting sulfite reductase of A vinosum6 but also during the reduction of sulfite in a sulfate reducer.13 Besides the many common characteristics, impor-tant differences between DsrEFH and TusBDC are also apparent Structural as well as phylogenetic analyses of DsrEFH revealed two different types of putative active sites, one involving Cys78 of DsrE and the other involving Cys20 of DsrH The pre-sence of these two sites clearly differentiates DsrEFH from the enterobacterial TusBCD Furthermore, the large insertion on loop L5 and the different surface structure of DsrF may have an influence on selecting partner molecules of DsrEFH, which is known not only to interact with DsrC6but also to form a super-complex with other Dsr proteins.7In summary, the observed differences may contribute to the diver-gent cellular functions of DsrEFH and TusBCD, that

is, sulfur oxidation versus thiouridine biosynthesis,

of the two proteins though their subunits, and over-all structures are conserved

Table 2 Comparison between DsrEFH and TusBCD

Oligomeric

state

Subunit

interaction

Total surface area (Å2)

Area buried per monomer (Å2)

No of salt bridgea

Subunit interaction

Total surface area (Å2)

Area buried per monomer (Å 2 )

No of salt bridgea

Sequence identity (%) and rmsd (Å)b

Dimer DsrEF 10,383 577 15 TusCD 10,109 483 8

Trimer DsrEFH 12,930 1026 19 TusBCD 12,714 872 16

Hexamer (DsrEFH) 2 21,156 1874 44 (TusBCD) 2 21,032 1604 37

a The number of salt bridges within 3.2 Å distance.

b Structural comparison with the combinatorial extension method [ http://cl.sdsc.edu/ce.html ].

In conclusion, the results presented here indicate

an important and so far largely neglected role of

sulfur transfer reactions during sulfur oxidation via

the Dsr proteins The structural characteristics of

DsrEFH support the recently suggested model that

suggests DsrEFH and DsrC to be agents transferring

sulfur to dsrAB-encoded sulfite reductase.6DsrE of DsrEFH could be the acceptor for persulfide sulfur originating from sulfur stored in the periplasm In analogy to the related E coli protein and according

to our mutational analysis, DsrC may accept a sulfur atom from Cys78 of DsrE As pointed out before,6

Fig 6 Structural superposition of DsrEFH (green) with TusBCD (red) (a) An asterisk indicates a location of a putative active site of DsrE and TusD The figure is created by the program PyMOL (red, negative; blue, positive).28(b) A structural superposition of residues comprising the putative active sites of DsrEFH and TusBCD The residues' numbers are the same for both structures

Fig 7 Diagrams of the electronic surface potential of DsrEFH and TusBCD The molecular surface around the second putative active pocket is drawn The figure is created by the program GRASP (red, negative; blue, positive; white, uncharged).29(a) DsrEFH, (b) TusBCD