Báo cáo khoa học: Electron-transfer subunits of the NiFe hydrogenases in Thiocapsa roseopersicina BBS pptx

The operon of the membrane-associated hydrogenase, HynSL, has an unusual gene arrangement: between the genes coding for the large and small subunits, there are two open reading frames, n

Thiocapsa roseopersicina BBS

Lı´via S Pala´gyi-Me´sza´ros1, Judit Maro´ti2, Do´ra Latinovics1, Tı´mea Balogh1, E´va Klement3,

Katalin F Medzihradszky3, Ga´bor Ra´khely1,2and Korne´l L Kova´cs1,2

1 Department of Biotechnology, University of Szeged, Hungary

2 Institute of Biophysics, Biological Research Centre, Hungarian Academy of Sciences, Szeged, Hungary

3 Proteomics Research Group, Biological Research Centre, Hungarian Academy of Sciences, Szeged, Hungary

Hydrogenases are metalloenzymes that catalyse the

reversible oxidation of molecular hydrogen according

to the reaction: H2M 2H++ 2e) They can catalyse

the reaction in both directions in vitro, but usually

either evolve or oxidize (take up) H2 in vivo The

hydrogenases can be classified according to the metal

content of their active centre: NiFe, FeFe or Fe

hydrogenases [1] The core of an NiFe hydrogenase

consists of a small electron-transfer subunit and a

large catalytic subunit Additional proteins are required

for post-translational maturation of the hydrogenase polypeptides and for connection of the core dimer to other bioenergetic⁄ redox processes of the cells These accessory hydrogenase-related proteins typically partici-pate in metallocentre assembly and the transcriptional regulation of the hydrogenases, and some seem to have

an electron-transfer function [1,2] The accessory genes are often located in the close vicinity of hydrogenase structural genes, but may also be found scattered in the genome Numerous microorganisms contain more

Keywords

electron transfer; haem, cytochrome b;

iron–sulfur protein; NiFe hydrogenase;

Thiocapsa roseopersicina

Correspondence

K L Kova´cs, Department of Biotechnology,

University of Szeged, H-6726 Szeged,

Ko¨ze´pfasor 52, Hungary

Fax: +36 62 544352

Tel: +36 62 544351

E-mail: kornel@brc.hu

(Received 31 August 2008, revised 6

October 2008, accepted 29 October 2008)

doi:10.1111/j.1742-4658.2008.06770.x

Thiocapsa roseopersicina BBS contains at least three different active NiFe hydrogenases: two membrane-bound enzymes and one apparently localized

in the cytoplasm In addition to the small and large structural subunits, additional proteins are usually associated with the NiFe hydrogenases, con-necting their activity to other redox processes in the cells The operon of the membrane-associated hydrogenase, HynSL, has an unusual gene arrangement: between the genes coding for the large and small subunits, there are two open reading frames, namely isp1 and isp2 Isp1 is a b-type haem-containing transmembrane protein, whereas Isp2 displays marked sequence similarity to the heterodisulfide reductases The other membrane-bound (Hup) NiFe hydrogenase contains the hupC gene, which codes for a cytochrome b-type protein that probably plays a role in electron transport The operon of the NAD+-reducing Hox hydrogenase contains a hoxE gene In addition to the hydrogenase and diaphorase parts of the complex, the fifth HoxE subunit may serve as a third redox gate of this enzyme The physiological functions of these putative electron-mediating subunits were studied by disruption of their genes The deletion of some accessory pro-teins dramatically reduced the in vivo activities of the hydrogenases, although they were fully active in vitro The absence of HupC resulted in a decrease in HupSL activity in the membrane, but removal of the Isp1 and Isp2 proteins did not have any significant effect on the location of HynSL activity Through the use of a tagged HoxE protein, the whole Hox hydrogenase pentamer could be purified as an intact complex

Abbreviation

tat, twin arginine transport.

than one hydrogenase Each enzyme has a specific

phys-iological function, e.g NAD+ reduction, electron

removal, H2 recycling for energy conservation, etc [3]

The electrons derived from H2 oxidation are used for

the reduction of the central quinone pool or terminal

electron acceptors, such as fumarate, NO3)or SO4) It

is noteworthy that, in spite of their specific expression

and physiological role, one enzyme can take over the

function of another to some extent [4]

Thiocapsa roseopersicina BBS belongs to the family

of purple sulfur photosynthetic bacteria, the

Chromati-aceae [5] During anoxygenic photosynthesis, this

bac-terium requires reduced sulfur compounds (e.g S2), S0

or S2O3 )) as electron sources for CO2fixation

T roseopersicinaproduces at least three NiFe

hydro-genases (Hyn, Hup and Hox) and contains the genes

of the so-called regulatory hydrogenase (HupUV) [6]

However, their physiological roles are still unclear

Both the HynS and HupS subunits have a ‘tat’-type

(‘twin arginine transport’) signal sequence; they are

therefore transported through the membrane by the

‘tat’ system [7] and are anchored to the membrane on

the periplasmic side The Hox enzyme has no signal

for transport across the membrane Hyn hydrogenase

(formerly Hyd [8]) is a membrane-bound bidirectional

enzyme which has remarkable stability under extreme

conditions; it is extracted from the photosynthetic

membrane as the catalytically active HynSL dimer [9]

The gene arrangement of the hyn operon is unusual:

the genes of the small and large subunits are separated

by a 2-kbp intergenic region In this section, two open

reading frames, isp1 and isp2, have been recognized

[8] The putative Isp1 and Isp2 gene products exhibit

remarkable similarity to the DsrK and DsrM subunits,

respectively, of the dissimilatory sulfite reductase

com-plex [10] Isp1 harbours few transmembrane domains,

and a putative b-type haem-binding site has been

pre-dicted by in silico analysis In contrast, the putative

Isp2 is a cytoplasmic enzyme resembling the

hetero-disulfide reductases [8] Similar gene structures can be

found in only a few bacteria, e.g in Chromatium

vinosum[10] (Accession No U84760), Aquifex aeolicus

[11], Aquifex pyrophilus [12] and an Archaeon,

Acidi-anus ambivalens [13], but their physiological role has

not been clarified so far

The other membrane-bound hydrogenase of

T roseopersicina, HupSL, is encoded in the

hupSLCD-HIR operon [14] It belongs to the group of uptake

NiFe hydrogenases which recycle H2 produced by the

nitrogenase complex [15] As a consequence of the

periplasmic location of the NiFe hydrogenases [16], H2

oxidation leads to the formation of a proton gradient

which is used for ATP synthesis [9] Next to the hupSL

genes encoding for the small and large hydrogenase subunits, the operon contains the hupC gene In Wolli-nella succinogenes, strong evidence has been provided that HupC, a b-type cytochrome [1], can transfer electrons from the NiFe hydrogenase to the quinones [17] Hence, HupC may link the electron transfer from Hup hydrogenases to the quinone pool

The third (Hox) hydrogenase has been partially purified from the soluble fraction of the cells [18] The genomic structure of the hox operon suggests a hetero-pentameric enzyme (HoxEFUYH) The HoxFU subunits are usually the NAD+-reducing part of the complex, and the HoxYH subunits are responsible for hydrogenase activity [19] Recently, a similar enzyme has been purified and partially characterized from a closely related strain, Allochromatim vinosum [20] Hox hydrogenases are composed of at least four subunits; the HoxYH and HoxFU dimers form the hydrogenase and diaphorase catalytic cores, respectively [19] In sev-eral cases, additional subunits have also been identi-fied In the Hox enzyme of Ralstonia eutropha (which was purified as a heterotetrameric enzyme for many years), a new subunit was discovered, and the compo-sition HoxFUYHI2 was suggested [21] In cyanobacte-ria and the phototrophic bactecyanobacte-ria T roseopersicina and

A vinosum, the heterotetrameric Hox enzyme is sup-plemented by a HoxE subunit, which is unrelated to the HoxI protein [18,20,22] In T roseopersicina, it has been shown previously that in-frame deletion of the hoxE gene impairs Hox activity in vivo, although the remaining part of the complex (HoxFUYH) still shows unaltered H2-dependent NAD+-reducing activity

in vitro [18] However, the roles of HoxE and the Hox complex are still not fully understood

In this article, we show that the various

hydrogenas-es use distinct electron-transfer subunits and routhydrogenas-es Deletion of the HupC, Isp1,2 and HoxE proteins clearly reveals their physiological relationships to their respective hydrogenases Affinity purification of the HoxE-tagged protein under mild conditions confirms the heteropentameric structure of this complex

Results Isp1 and Isp2 are expressed proteins The in silico analysis of the intergenic region of the hynS and hynL genes indicated two open reading frames It has been established that the hynS-isp1-isp2-hynL region is cotranscribed [23] In order to confirm that isp1 and isp2 are really coding regions, the hynS-isp1-isp2-hynL* genes were cloned behind a T7 promoter (see Experimental procedures) The genes of

the construct were expressed in the Escherichia coli

BL21(DE3) host, and the bands corresponding to the

calculated molecular masses of Isp1 (24.6 kDa) and

Isp2 (48.4 kDa) could be clearly identified (data not

shown) The small and large subunits were also

detected This means that all the translational signals

necessary for the expression of the HynSL and Isp

subunits are functionally present in the construct, and

are recognized by the translational apparatus of

E coli The coexpression of the Hyn and Isp subunits

suggests that they probably form a functional

complex

Isp1 and Isp2 are required for the in vivo function

of Hyn hydrogenase

The solubilized and purified Hyn hydrogenase

con-tained only the HynSL subunits [23] The role of the

Isp proteins in T roseopersicina is unknown, but

com-putational analysis has shown that Isp1 is a b-type

haem-containing transmembrane electron carrier,

whereas Isp2 seems to be a redox Fe–S-containing

pro-tein If these subunits are involved in the electron flow

from⁄ to the hydrogenase, their removal would abolish

the hydrogenase activity in vivo, where the endogenous

electron donors⁄ acceptors must be used

Therefore, a double isp1-isp2 in-frame mutant was

constructed in the T roseopersicina GB2131 (DhoxH,

DhupSL) strain (ISP12M, see Experimental

proce-dures) The hydrogenase activities were measured both

in vivo (without the addition of an artificial electron

carrier) and in vitro (in the presence of redox viologen

dyes) The data in Table 1 unequivocally prove that

the in vivo H2-producing activity of the isp1,2 mutant

strain is completely lost and the in vivo H2 uptake activity is dramatically decreased relative to the control GB2131 (DhoxH, DhupSL) strain containing all the functional gene products of the Hyn operon A single Isp1 in-frame deletion mutant was also constructed (ISP1M) Mutation of the Isp1 protein brings about the same phenotype as the deletion of both Isp1 and Isp2 (Table 1) Some remaining in vivo H2 uptake activity of the Hyn hydrogenase can be detected in both mutants, which suggests an alternative, less effec-tive electron-transfer pathway

In the in vitro measurements, in which benzyl-violo-gen was used as an artificial electron acceptor, the H2 uptake activity was not influenced by the lack of Isp1

or Isp1,2 proteins (Table 1) On the one hand, these and the in silico results confirm that the Isp proteins play an essential role in the H2 reduction and oxida-tion ability of Hyn hydrogenase in its natural environ-ment, but the lack of these subunits has no effect on the hydrogenase activity in the artificial assay On the other hand, this also means that the Isp1,2 proteins do not affect the post-translational maturation and expression level of the Hyn enzyme A trivial rationali-zation of these observations is that the lack of Isp1 or Isp1,2 proteins results in blockade of the electron flow from⁄ to Hyn hydrogenase under physiological condi-tions

As the computational analysis implies that Isp1 is

an integrated membrane protein, it is plausible to assume that the HynSL dimer is anchored to the mem-brane through the Isp1 protein Accordingly, we inves-tigated the localization of Hyn hydrogenase in the Isp mutant strains Unexpectedly, the in vitro H2 uptake measurements on the various cellular fractions indi-cated that a similar proportion of Hyn hydrogenase remained in the membrane fraction in the presence and absence of the Isp proteins (Table 2) This is surprising, as our protein purification experiments demonstrated that the HynSL subunits are only loosely associated with the membrane and can be easily

Table 1 Activities of Hyn hydrogenase in vivo and in vitro in the

presence and absence of the Isp1 and Isp2 proteins The results

are given as percentages of the level for GB2131 The cultures

were grown on Pfennig’s medium with 4 gÆL)1of Na 2 S 2 O 3 The

val-ues are normalized to bacteriochlorophyll content The GB112131

strain (DhupSL, DhoxH, DhynS-isp1-isp2-hynL) and the M539 strain

(hypF mutant) containing no active NiFe hydrogenase served as

negative controls.

Strain

Relative H 2

production

in vivo

Relative

H2uptake

in vivo

Relative

H2uptake

in vitro GB2131

(DhupSL, DhoxH)

100 ± 10.2 100 ± 13.7 100 ± 10.0 ISP1M

(DhupSL, DhoxH, Disp1)

0.00 29.5 ± 9.8 100.4 ± 9.2 ISP12M t

(DhupSL, DhoxH, Disp12)

0.0 35.6 ± 5.9 116.2 ± 8.3

Table 2 Location of Hyn hydrogenase with and without the Isp1,2 proteins (see description in Table 1).

Strain

Relative uptake activity

in vitro Membrane fraction

Soluble fraction

ISP1M (DhupSL, DhoxH, Disp1) 106.2 ± 33.4 102.7 ± 4.0 ISP12M (DhupSL, DhoxH, Disp12) 112.4 ± 0.3 113.9 ± 6.8

washed off [23] The strength of the HynSL–membrane

interaction apparently does not depend on the presence

or absence of the Isp1 protein

The expression of the Hup enzyme depends on

the thiosulfate content of the medium

With a view to examining the function of the HupC

protein in T roseopersicina, a DhynSL, DhoxH

(GB1131) strain was created (see Experimental

proce-dures) This strain is suitable for the measurement of

Hup hydrogenase activity alone, without the

contribu-tions of Hyn and Hox hydrogenases However, under

standard growth conditions, i.e in the presence of

4 gÆL)1 Na2S2O3, only very low HupSL hydrogenase

activity was detected in the DhynSL, DhoxH (GB1131)

strain It was postulated that this concentration of

thiosulfate resulted in a redox potential in the cells,

which downregulated the activity of HupSL

hydro-genase (as an uptake, electron-donating enzyme) To

test this hypothesis, the expression level and in vitro

activity of the Hup hydrogenase were measured in cells

grown in the presence of various amounts of

thiosul-fate The data in Table 3 clearly illustrate that the

lower the thiosulfate concentration in the medium, the

higher the Hup hydrogenase activity both in vivo and

in vitro The effects of the thiosulfate content on the

expression level of the hupSL genes were additionally

monitored by quantitative RT-PCR The data in

Table 4 reveal that a decrease in the thiosulfate

con-tent of the medium from 4 to 2 gÆL)1resulted in a

dra-matic (> 16-fold) increase in the hupSL mRNA level

These data suggest that, when Hup is the only active

hydrogenase in the cell, its activity strongly depends

on the thiosulfate content of the medium, and changes

in the activity primarily correlate with the expression

level of the enzyme Hence, as a practical consequence,

the subsequent experiments on Hup activity were

per-formed with samples grown in the presence of 2 gÆL)1

thiosulfate

HupC is an electron-transfer subunit of Hup hydrogenase

To establish the function of HupC, its gene was deleted in-frame in the DhynSL, DhoxH (GB1131) strain, and the HupSL activities were compared both

in vivoand in vitro

The in vivo H2 uptake activity of Hup hydrogenase was substantially decreased in the DhupC (HCMG4) strain At the same time, the in vitro activity was twice

as high as that of the strain harbouring HupC (Table 5) A comparison of the hupSL mRNA levels

of the cells containing or lacking the hupC gene per-ceptibly revealed that a loss of the hupC gene had a positive effect on the transcription level of the hupSL genes (Table 4) To check that the effect was really linked to the loss of HupC, a complementation experi-ment was performed by introducing an expression

Table 3 In vivo and in vitro H2 uptake activities of the GB1131

(DhynSL, DhoxH) strain grown photoautotrophically (Pfennig’s) at

various Na2S2O3concentrations The results are given as

percent-ages of that for the sample grown with 1 gÆL)1of Na2S2O3.

Concentration of

Na 2 S 2 O 3 (gÆL)1)

Relative H2uptake activity

Table 4 Relative mRNA levels of the hup operon in the presence (GB1131) and absence (HCMG4) of the hupC gene at various

Na2S2O3 concentrations The cultures were grown on Pfennig’s medium with 2 or 4 gÆL)1 of Na2S2O3 The mRNA levels were determined by quantitative RT-PCR and the results are given as percentages of the level for GB1131 The values are normalized to the total RNA content.

Strain

4 gÆL)1

Na 2 S 2 O 3

2 gÆL)1

Na 2 S 2 O 3

GB1131 (DhynS-isp1-isp2-hynL, DhoxH)

100.0 ± 0.0 1650.0 ± 44.5 HCMG4

(DhupC, DhynS-isp1-isp2-hynL, DhoxH)

300.0 ± 20.0 2700.0 ± 102.8

Table 5 Activities of Hup hydrogenase in vivo and in vitro in the presence (GB1131, pMHE6C HCMG4) and absence (HCMG4) of the HupC protein The cultures were grown on Pfennig’s medium with

2 gÆL)1of Na 2 S 2 O 3 The hydrogenase activity values are normalized

to the bacteriochlorophyll content The results are given as a percent-age of the level for GB1131 The GB112131 strain (DhupSL, DhoxH, DhynS-isp1-isp2-hynL) and the M539 strain (hypF mutant) containing

no active NiFe hydrogenase served as negative controls.

Strain

Relative H 2 uptake activity

GB1131 (DhynS-isp1-isp2-hynL, DhoxH)

100.0 ± 2.6 100.0 ± 14.5 HCMG4 (DhupC,

DhynS-isp1-isp2-hynL, DhoxH)

40.4 ± 5.5 198.9 ± 5.5 pMHE6C HCMG4 (DhupC,

DhynS-isp1-isp2-hynL, DhoxH, pMHE6C)

68.3 ± 10.3 231.2 ± 33.5

cassette containing the hupC gene driven by the crt

promoter (see Experimental procedures) Table 5

shows that the plasmid-borne HupC (pMHE6C⁄

HCMG4 in Table 5) partially ( 50%) restored the

Hup hydrogenase activity in vivo

It is plausible to assume that HupC serves as a

membrane anchor for Hup hydrogenase [24] In

con-trast with the findings on Hyn hydrogenase, the lack

of HupC substantially reduced the Hup hydrogenase

activity in the membrane fraction, i.e 98% of the

activity was lost in the hupC deletion mutant The

hydrogenase activity in the soluble fraction was also

decreased; therefore, it is unlikely that HupSL was

released from the membrane and accumulated in the

cytoplasm The lower total activity in the HupC-minus

cell fractions might be explained by the lower stability

of the HupSL enzyme in the absence of HupC in the

disrupted and fractionated cells relative to the

wild-type (Table 6) It is noteworthy that the HupSL

activ-ity was significantly higher in the soluble than in the

membrane fraction in the pMHE6C⁄ HCMG4 (HupC

complementing) strain (Table 6) The plasmid-borne

HupC could possibly restore the stability of HupSL,

although the majority of the activity remained in the

soluble fraction

These data suggest that HupC has no role in the

maturation process of HupSL hydrogenase, but

influ-ences the in vivo activity and the expression level of the

HupSL enzyme Taken together with the findings of

computational analysis, the HupC protein serves an

electron-transport role in T roseopersicina and

proba-bly forms a functional complex with the small and

large hydrogenase subunits in vivo

Purification of Hox hydrogenase

In T roseopersicina, the cytoplasmic Hox hydrogenase

is coded by the hoxEFUYH operon The enzyme

contains hydrogenase (HoxYH) and diaphorase (Ho-xEFU) subunits [18] The diaphorase subunits of the Hox-type hydrogenases exhibit significant sequence similarities to three subunits (NuoEFG) of NADH:ubiquinone oxidoreductase [18,25] The in-frame deletion of the hoxE gene led to the complete loss of Hox activity in vivo, whereas the enzyme was fully active in vitro [18] This suggests that HoxE may function in vivo as an electron-transfer protein Thus, HoxE would offer a third channel for the electrons in addition to the hydrogenase and diaphorase catalytic centres To test whether the HoxE protein forms a functional complex with the HoxFUYH subunits, its FLAG-tagged form was expressed from a pMHE6 expression vector [26] under the control of the

T roseopersicina crtpromoter (pMHE6HoxE Table 7) HoxE was purified by affinity chromatography via the FLAG-tag under very mild conditions in order to pre-serve the protein–protein interactions (see Experimen-tal procedures) The proteins eluted from the affinity column were separated on SDS-polyacrylamide gel and analysed by MALDI-TOF-MS Each subunit of the HoxEFUYH enzyme complex was easily identified, indicating that HoxE is physically associated with the other (HoxFUYH) subunits (Fig 1)

Discussion Hydrogenases are widespread in the microbial world The actively expressed hydrogenases must have a dedi-cated physiological role within the cells For the in vivo function, the catalytic dimers of NiFe hydrogenases must be connected to other oxidoreductases directly or via electron-transfer subunits In this study, attempts were made to identify the redox partners and the elec-tron-channelling subunits of all three hydrogenases in the cells

In T roseopersicina, there are at least three NiFe hydrogenases (HynSL, HupSL and HoxYH) with distinct properties and different functions The HypF accessory protein is required for the maturation of every NiFe hydrogenase, and disruption of the hypF gene therefore results in the hydrogenase-minus pheno-type [27] However, the hydrogenase-less cells showed virtually identical growth properties as the wild-type under standard growth conditions Special growth con-ditions, i.e photoautotrophic in the presence of H2 and only 0.005% Na2S, were identified, in which the presence of each hydrogenase was important, including the Hup enzyme being essential for H2-dependent growth (data not shown) This indicates that the Hup enzyme has a direct connection to the central redox, i.e quinone, pool Nonetheless, the real redox partners

Table 6 Location of HupSL hydrogenase with (GB1131, pMHE6C

HCMG4) and without (HCMG4) the HupC subunit (see description

in Table 5).

Strain

In vitro relative uptake activity Membrane

fraction

Soluble fraction GB1131 (DhynS-isp1-isp2-hynL,

DhoxH)

100.0 ± 12.9 36.0 ± 8.5 HCMG4 (DhupC,

DhynS-isp1-isp2-hynL, DhoxH)

1.8 ± 0.24 11.3 ± 1.8 pMHE6C HCMG4 (DhupC,

DhynS-isp1-isp2-hynL,

DhoxH, pMHE6C)

47.3 ± 0.9 115.0 ± 7.1

and⁄ or electron channels of each hydrogenase remain

poorly understood

The genes coding for the HynSL enzyme are

sepa-rated by two open reading frames, which have been

shown to code for real proteins, Isp1 and Isp2 Both

proteins have been demonstrated to be important for

the function of the HynSL enzyme in vivo, but neither

for its in vitro activity or expression Therefore, they

probably play an electron-transfer role from⁄ to the

Hyn enzyme The heterodisulfide reductase homologue

Isp2 is probably an oxidoreductase; its redox substrate

still remains to be identified We conclude that the

Hyn enzyme is indirectly linked to the central

redox⁄ bioenergetic processes via the Isp1,2 proteins

and an unknown redox substrate

A direct coupling of the HupC protein to the uptake

HupSL hydrogenase was demonstrated in this study

Deletion of the hupC gene resulted in reduced and

enhanced activities of HupSL in vivo and in vitro,

respectively As HupC is supposed to react with

qui-nones directly [17], the reduced in vivo activity stems

from the obstruction of the electron flow from the

hydrogenase Consequently, HupC is suggested to be

the third subunit of the Hup complex, catalysing the

H2-dependent reduction of quinones This is in line with the observation that HupSL hydrogenase is essential for H2-dependent growth under the above-mentioned growth conditions

The expression level of the HupSL enzyme was upregulated both by disrupting the HupC subunit and

by decreasing the thiosulfate content of the medium It

is assumed that both processes lead to a more oxidized quinone pool, as the disrupted HupC cannot transfer the electrons from HupSL, and thiosulfate serves as reducing power for the photosynthetic carbon fixation via the central quinone pool [28] The redox status of the quinone pool may influence HupSL expression: the increased electron requirement is reflected in a higher expression level of the electron-donating Hup hydro-genase

Interestingly, removal of the transmembrane elec-tron-transfer subunits of the Hyn and Hup enzymes gave rise to distinct effects on the locations of their corresponding hydrogenases The lack of Isp1 did not change the membrane association of the Hyn enzyme, whereas the elimination of HupC led to detachment of

Table 7 Strains and plasmids Indicated strains and plasmids are from Stratagene, La Jolla, CA, USA.

Strain or plasmid Relevant genotype or phenotype

Reference

or source Thiocapsa roseopersicina

Escherichia coli

XL1-Blue MRF¢ D (mcrA)183, D (mcrCB-hsdSMR-mrr)173, endA1, supE44, thi-1, recA1, gyrA96, relA1

lac [F¢ proAB lacIqZDM15 Tn10 (Tet r

)]c

Stratagene BL21 (DE3) F)ompT gal dcm lon hsdSB(rB)mB)) k(DE3) [lacI lacUV5-T7 gene 1 ind1 sam7 nin5] Stratagene Plasmids

pMHE6crtKm Km r , mob + , expression vector containing the promoter region of crtD gene [26]

pISP1M Km r , in-frame up- and downstream homologous regions of isp1 in pK18mobsacB This work pISP12M Kmr, in-frame up- and downstream homologous regions of isp1-isp2 in pK18mobsacB This work

the HupSL enzyme from the membrane The absence

of HupC also resulted in a destabilization of HupSL

A similar phenomenon has been described for HupC

in Rhodobacter capsulatus [24] In T roseopersicina, the

plasmid-borne HupC restored the stability and

mem-brane association of HupSL, but a significant amount

of enzyme remained in the soluble fraction

Controver-sial data have been published in the literature with

regard to the membrane anchoring role of HupC

Deletion of HoxW, the HupC homologous protein in

R eutropha, resulted in detachment of the hydrogenase

from the membrane [29] In contrast in

Pseudomo-nas hydrogenovora, the location of the HupSL dimer in

the hupC mutant strain did not change [30]

It has been shown previously that HoxE is required

for the in vivo function of the Hox hydrogenase [18]

Here, we have demonstrated that HoxE fulfills this

role in association with the other subunits of the Hox

hydrogenase Purification of the affinity-tagged HoxE

under mild conditions resulted in copurification of the

four other (HoxFUYH) subunits We observed that

the hydrogenase dimer dissociated from the HoxEFU

trimer relatively easily (data not shown) A similar

finding has been published recently for the A vinosum Hox hydrogenase [20] This means that the enzyme com-plex has three gates for electron flow: one for H2 oxid-ation⁄ proton reduction, one for the NAD+⁄ NADH redox reaction and one functioning as an electron channel via the HoxE subunit This makes the potential physiological function of this hydrogenase more complex, as the Hox hydrogenase has a potential to be associated with various metabolic pathways involving redox changes

In cyanobacteria, the Hox hydrogenase was initially suggested to have a relationship to the respiratory complex [25], but evidence challenging this idea was later published [31] A valve role of the Hox enzyme was suggested for the low-potential electrons generated during photosynthesis [32] The three gates for electron flow are in line with the valve hypothesis However, depending on the sulfur source, the Hox enzyme is able to produce H2either under illumination or in the dark [33], and thus its physiological function cannot

be restricted to photosynthetic electron flow, but the respiratory and fermentative processes should also be considered

Experimental procedures Bacterial strains and plasmids

The strains and plasmids are listed in Table 7 The T roseo-persicina strains were grown photoautotrophically in Pfennig’s medium under anaerobic conditions in liquid cultures with continuous illumination at 27–30C for 4–5 days [27] The acetate-supplemented (2 gÆL)1) plates were solidified with 7 gÆL)1 of Phytagel (Sigma, St Louis,

MO, USA) [34] The plates were incubated in anaerobic jars

by means of the AnaeroCult (Merck, Darmstadt, Germany) system for 2 weeks The E coli strains were maintained on LB-agar plates Antibiotics were used in the following con-centrations (mgÆL)1): for E coli, ampicillin (100), kanamycin (25), tetracyclin (20); for T roseopersicina, kanamycin (25), streptomycin (5) and gentamycin (5)

Expression of the hynS-isp1-isp2-hynL* genes

of T roseopersicina in E coli using the T7 promoter⁄ RNS polymerase system

The hynS-isp1-isp2-hynL* gene products were produced from pTSH2⁄ 8 [8] in the E coli BL21(DE3) strain This construct contains the native promoter⁄ regulatory region and, additionally, the complete hynS, isp1 and isp2 genes and truncated hynL (denoted by *) The incomplete hynL did not interfere with the outcome of the experiments Expression of the genes was induced by isopropyl

thio-b-d-P T7

P crtD

RBS

HoxE

FLAG-StrepII

kDa

120

100

3 2

1

M

85

70

50

40

HoxF

HoxU

HoxE HoxE-TAG

HoxH

30

25

20

Phenylalanyl t-RNA synthetase β subunit

Phenylalanyl t-RNA synthetase α subunit

Fig 1 The HoxFUYH subunits copurifiy with the tagged HoxE

sub-unit during the course of affinity chromatography (A) Scheme of

the cassette used to express tagged HoxE in T roseopersicina

(PcrtD, carotenoid promoter; PT7, T7 promoter; RBS,

ribosome-bind-ing site; M, marker) (B) Protein patterns of the elution fractions

separated by SDS-PAGE The soluble fraction of T roseopersicina

cells expressing tagged HoxE was loaded onto an Anti-Flag affinity

column, the resin was washed, and the bound proteins were eluted

three times by Flag peptide (for details, see Experimental

proce-dures) (1, 2, 3 indicate the elution fractions) The bordered bands

were cut and analysed by mass spectrometry.

galactoside, and monitored by the incorporation of

l-[35S]methionine into the proteins synthesized [35] The

samples were separated in an SDS-polyacrylamide gel and

analysed by a Phosphor Imager (Phosphor Imager 445

SI, Molecular Dynamics, Uppsala, Sweden)

Conjugation

Conjugation was carried out as described previously [27]

Deletion of the isp1,2 genes

The in-frame deletion constructs were derived from the

pK18mobsacB vector [36] The upstream region of the

isp1,2 genes was amplified with the otsh14r (5¢-GAT

product was cloned into the polished BamHI site of pUC19

[37], yielding pUNSBamHI

To clone the downstream region, another PCR was

per-formed with the isp1o7 (5¢-TCGCACGCTGGTACAA

CGGG-3¢) and isp2o2 (5¢-ACCAGGTGCTCGGCGAT

CAT-3¢) primers This fragment was cloned into the

XbaI-digested and blunted pUNSBamHI vector (pUS2) The

2502 bp EcoRI fragment of pUS2 was ligated with the

EcoRI fragment of pK18mobSacB, yielding pISP12M

The plasmid was transformed into the E coli S17-1(kpir)

strain, and then conjugated into the T roseopersicina

GB2131 strain as described previously [27] The single

recombinants selected through their kanamycin resistance

were grown in liquid medium The double recombinants

were selected on 3% sucrose-containing plates The

sucrose-resistant and kanamycin-sensitive colonies were

selected, and the genotype was confirmed by Southern

blotting and hybridization (ISP12M)

Deletion of the isp1 gene

The upstream homologous region was taken from the

pUNSBamHI vector The downstream homologous region

was amplified with the isp1o8 (5¢-AGCTGACGCACATCT

TCACG-3¢) and isp2o7 (5¢-GGTGAGACCGACCACCG

GGA-3¢) primers The product was cloned into the

BamHI-cleaved and polished pUNSBamHI construct (pUS3) The

EcoRI fragment of pUS3 was cloned into pK18mobSacB

(pISM1⁄ 3) The construct was conjugated into the GB2131

strain and the double recombinants were selected as

described below

Deletion of the hupC gene

For deletion of the hupC gene, the pHCD1 and pHCD2

in-frame deletion constructs were created as follows The

upstream region of hupC was amplified with the ohup20

(5¢-CGAGCAGGCCAAGTATTC-3¢) and ohup19 (5¢-TGT TGGTCAGGCGGATCT-3¢) primers, and the 836 bp PCR product was cloned into the SmaI-digested pK18mobsacB (pHCD1) The downstream region was amplified with the ohup21 (5¢-GGCGGATGTTCAAGGACG-3¢) and ohup22 (5¢-TCGACCACGACACTGAAG-3¢) primers The 800 bp fragment obtained was cloned into the PstI-digested polished pHCD1 (pHCD2) This construct was conjugated into the T roseopersicina GB1131 strain, yielding the HCMG4 strain The double recombinants were selected and the genotypes were confirmed as described above

Construction of HupC-expressing plasmid

The hupC gene was amplified with the ohupc1 (5¢-CATAT GTCGCGAGCTGCGTCGCG-3¢) and ohupc2 (5¢-AAGCT TTGGCCGATCGTCCTTGAACAT-3¢) primers containing NdeI and HindIII recognition sites The 777 bp PCR prod-uct was inserted into the EcoRV-digested pBluescripSK+ (pBtC) The 777 bp NdeI-HindIII-digested fragment was ligated into the corresponding sites of pMHE6crtKm [26], resulting in pMHE6C

RNA isolation

For RNA isolation, T roseopersicina was grown in 60 mL

of liquid medium in a hypovial to A600 nm= 1–1.5; 15 mL

of culture was centrifuged at 15 000 g for 2 min, the pellet was suspended in 300 lL of SET buffer [20% sucrose,

50 mm EDTA (pH 8.0) and 50 mm Tris⁄ HCl (pH 8.0)] and 300 lL of SDS buffer was added [20% SDS, 1% (NH4)2SO4, pH 4.8]; 500 lL of saturated NaCl was added next, the sample was centrifuged at 20 000 g for

10 min and the clear supernatant was transferred into a new tube 2-Propanol (70% of the total volume of the supernatant) was added to the solution and the mixture was centrifuged at 20 000 g for 20 min The pellet was washed twice with 1 mL of 70% ethanol The dried pellet was suspended in 20 lL of diethylpyrocarbonate-treated water

DNase I treatment

DNase treatment took place in the presence of 10· reaction buffer with MgCl2 (Fermentas, Burlington, Canada) and DNase I (RNase-free, Fermentas) at 37C for 1 h The reaction was inactivated by heat at 65C for 10 min in the presence of EDTA (Fermentas)

Reverse transcription and quantitative real-time PCR

For reverse transcription, the Omniscript Reverse Trans-criptase Kit (Qiagen, Hilden, Germany) was used according

to the manufacturer’s instructions One microgram of

DNase I-treated total RNA was added to a master mix

[10· buffer RT, dNTP mix (0.5 mm of each dNTP), reverse

primer (0.2 lm), RNase inhibitor (10 units⁄ reaction),

Omniscript Reverse Transcriptase (2 units⁄ reaction) in

diethylpyrocarbonate-treated water] on ice in a final volume

of 20 lL The reaction mixture was then incubated at

37C for 60 min

Reverse transcription was initiated from the huprto2

(5¢-CGCTTGAGCCGATTCTGAACAT-3¢) primer specific

for the hupL gene The cDNA produced during reverse

transcription was used as a template for quantitative PCR,

which was performed using the ohupSRT1 (5¢-GGA

CAAGGGCAGCTTCTATCA-3¢) and ohupSRT2 (5¢-CG

CATTGGCCTCGATACC-3¢) primers located in the hupS

gene PCR was carried out and the products were measured

with an Applied Biosystems (Foster City, CA, USA) 7500

real-time PCR instrument PCR was performed in a total

volume of 25 lL, including 1 lL of cDNA, 12.5 lL of

Power SYBR Green PCR Master Mix (Applied

Bio-systems), forward and reverse primers (12.5 pmol of each)

and 9 lL of nuclease-free water The following programme

was applied: 95C for 10 min; 95 C for 15 s and 60 C

for 1 min for 40 cycles; 95C for 15 s; 60 C for 1 min;

95C for 15 s; 60 C for 15 s A calibration curve was

gen-erated using sixfold dilutions of pKK48 plasmid DNA

(containing the sequence of the hupS gene) in the 100 to

0.001 ngÆlL)1concentration range

Activity measurements

The hydrogenase activities of the various mutants were

measured both in vivo and in vitro In all experiments, the

HypF mutant (lacking any NiFe hydrogenase activity) and

the GB112131 strain (DhoxH, DhupSL,

DhynS-isp1-isp2-hynL) were used as negative controls

In vitro H2uptake activity measurements

The samples were suspended in 2 mL of 20 mm potassium

phosphate buffer containing 0.4 mm of oxidized

benzyl-viologen The cuvettes were closed with SubaSeal rubber

stoppers The gas phase was flushed with H2 and the H2

uptake activity was measured spectrophotometrically at

600 nm and 60C

In vivo hydrogen evolution activity

measurements

Cultures (60 mL) were grown in 100 mL hypovials; the gas

phase was then flushed with N2 after inoculation and the

H2 produced was measured gas chromatographically [27]

on day 6

In vivo H2-uptake activity measurements

Medium (60 mL) was inoculated into 100 mL hypovials; the gas phase was flushed with N2 and 5 mL of pure H2 was injected into the bottles The cultures were grown under illumination and the H2content of the gas phase was measured gas chromatographically on day 6

Preparation of membrane and soluble fractions

of T roseopersicina

The membrane fractions were prepared from 50 and

110 mL cultures for Hyn and Hup measurements, respec-tively The cells were harvested by centrifugation at 7000 g, suspended in 1 mL of 20 mm potassium phosphate buffer (pH 7.0) and broken by sonication [Bandelin Sonopuls (Berlin, Germany) HD2070 ultrasonic homogenizer; at 85% amplitude six times for 10 s] The broken cells were centri-fuged at 15 000 g for 10 min The debris (sulfur globules and intact cells) was discarded and the supernatant was centrifuged at 100 000 g for 1.5 h The pellet was washed, resuspended in 800 lL of potassium phosphate buffer (pH 7.0) and used as membrane fraction The supernatant was regarded as the soluble fraction

Measurements of bacteriochlorophyll content

The bacteriochlorophyll content was estimated using a methanol extraction procedure, as described previously [38] The absorption of the samples was measured at 772 nm; the extinction coefficient was 8.41 g)1ÆLÆcm)1 The in vivo and in vitro activities were normalized to the bacteriochlo-rophyll content of the samples

Construction of the double-tagged hoxE gene

For the construction of an expression system capable of producing the HoxE protein of T roseopersicina fused with tandem FLAG-tag-Strep-tag II at the C-terminus, a 501-bp fragment was amplified from the pTCB4⁄ 2 clone [8] using the TCHO32 (5¢-CATATGAGTCTGCAGCAAGCCA-3¢) and TCHO33 (5¢-AAGCTTGGTCAGCTCCTCGAGC-3¢) primers and cloned into the SmaI site of pBluescript SK+

pBtHoxE was ligated into the NdeI-HindIII-digested pMHE6crtKm vector (pMHE6HoxE) The construct was confirmed by sequencing and conjugated into the T roseo-persicinaGB1121 strain

Purification of Hox hydrogenase

Four grams of cell paste from a GB1121⁄

pMHE6HoxE-Km culture were suspended in 5 mL of NaCl⁄ Tris [50 mm

Tris (pH 7.4) and 150 mm NaCl] The sample was sonicated

with a Bandelin Sonopuls HD2070 ultrasonic homogenizer

(at medium mode, amplitude 2.4 times for 10 s) The cell

debris and sulfur crystals were removed by centrifugation

(27 000 g, 10 min) The supernatant was incubated with

300 lL of ANTI-FLAG M2 affinity resin (Sigma) at 4C

for 2 h with gentle shaking The matrices were washed

seven times with 1.5 mL of NaCl⁄ Tris For elution, the

slurry was washed twice with 100 lL and once with 50 lL

of NaCl⁄ Tris with FLAG-peptide (200 lgÆmL)1) Aliquots

were collected and the samples were analysed by

SDS-PAGE

SDS-PAGE and protein staining

SDS-PAGE and silver staining of proteins were performed

as described by Ausubel et al [39]

Identification of proteins by MALDI-TOF-MS

Coomassie blue-stained gel bands were cut out and

analy-sed by MALDI-TOF-MS, as described previously [26]

Bioinformatics tools

Protein sequences in the various databases were compared

with the blast (P, X) programs (http://www.ncbi.nih.nlm

gov), the peptide mass fingerprints and the power spectral

density spectra; a database search was performed using the

National Center for Biotechnology Information protein

database with Protein Prospector MS-Fit and MS-Tag,

respectively (http://prospector.ucsf.edu/)

Acknowledgements

The contribution of Drs B D Fodor and A´ T

Kov-a´cs in the early phase of this work is gratefully

acknowledged This work was supported by EU

pro-jects HyVolution FP6-IP-SES6 019825 and FP7

Col-laborative Project SOLAR-H2 FP7-Energy-212508,

and by domestic funds (GOP-2007-1.1.2,

Asbo´th-DAMEC-2007⁄ 09, Baross OMFB-00265⁄ 2007 and

KN-RET-07⁄ 2005)

References

1 Vignais PM & Billoud B (2007) Occurrence,

classifica-tion, and biological function of hydrogenases: an

over-view Chem Rev 107, 4206–4272

2 Vignais PM & Colbeau A (2004) Molecular biology of

microbial hydrogenases Curr Issues Mol Biol 6, 159–188

3 Cammack R, Frey M & Robson R (2001) Hydrogen

as a Fuel: Learning from Nature Taylor & Francis,

London

4 Laurinavichene TV, Ra´khely G, Kova´cs KL & Tsygan-kov AA (2007) The effect of sulfur compounds on H2

evolution⁄ consumption reactions, mediated by various hydrogenases, in the purple sulfur bacterium, Thiocapsa roseopersicina Arch Microbiol 188, 403–410

5 Bogorov LV (1974) The properties of Thiocapsa roseo-persicina, strain BBS, isolated from an estuary of the White Sea Microbiologia 43, 326–332

6 Kova´cs KL, Kova´cs A´T, Maro´ti G, Me´sza´ros LS, Bal-ogh J, Latinovics D, Fu¨lo¨p A, Da´vid R, DorBal-ogha´zi E

& Ra´khely G (2005) The hydrogenases of Thiocapsa roseopersicina Biochem Soc Trans 33, 61–63

7 Sargent F, Stanley NR, Berks BC & Palmer T (1999) Sec-independent protein translocation in Escherichia coli.A distinct and pivotal role for the TatB protein

J Biol Chem 274, 6073–6082

8 Ra´khely G, Colbeau A, Garin J, Vignais PM & Kova´cs

KL (1998) Unusual organization of the genes coding for HydSL, the stable [NiFe] hydrogenase in the photo-synthetic bacterium Thiocapsa roseopersicina BBS

J Bacteriol 180, 1460–1465

9 Kova´cs KL & Bagyinka C (1990) Structural properties, functional states and physiological roles of hydrogenase

in photosynthetic bacteria FEMS Microbiol Rev 87, 407–412

10 Dahl C, Ra´khely G, Pott-Sperling AS, Fodor BD, Tak-a´cs M, To´th A, Kraeling M, Gy}orfi K, Kova´cs A´T, Tusz J et al (1999) Genes involved in hydrogen and sulfur metabolism in phototrophic sulfur bacteria Microbiol Lett 180, 317–324

11 Brugna-Guiral M, Tron P, Nitschke W, Stetter KO, Burlat B, Guigliarelli B, Bruschi M & Giudici-Orticoni

MT (2003) [NiFe] hydrogenases from the hyperthermo-philic bacterium Aquifex aeolicus: properties, function, and phylogenetics Extremophiles 7, 145–157

12 Lu J, Ra´khely G, Kova´cs KL, Xiao C & Zhou P (2001) Identification and cloning of partial mbh2 gene cluster

of hyperthermophile Aquifex pyrophilus Wei Sheng Wu Xue Bao 41, 674–679

13 Laska S, Lottspeich F & Kletzin A (2003) Membrane-bound hydrogenase and sulfur reductase of the hyper-thermophilic and acidophilic archaeon Acidianus ambivalens Microbiology 149, 2357–2371

14 Colbeau A, Kova´cs KL, Chabert J & Vignais PM (1994) Cloning and sequencing of the structural (hupSLC) and accessory (hupDHI) genes for hydroge-nase biosynthesis in Thiocapsa roseopersicina Gene 140, 25–31

15 Colbeau A, Kelley BC & Vignais PM (1980) Hydro-genase activity in Rhodopseudomonas capsulata:

relationship with nitrogenase activity J Bacteriol 144, 141–148

16 Kova´cs KL, Bagyinka C & Serebriakova LT (1983) Distribution and orientation of hydrogenase in various photosynthetic bacteria Curr Microbiol 9, 215–218