Scholarly article on topic 'Topological Plasticity of Enzymes Involved in Disulfide Bond Formation Allows Catalysis in Either the Periplasm Or the Cytoplasm'

Topological Plasticity of Enzymes Involved in Disulfide Bond Formation Allows Catalysis in Either the Periplasm Or the Cytoplasm Academic research paper on "Biological sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Journal of Molecular Biology
OECD Field of science
Keywords
{DsbB / VKOR / "topology inversion" / "rational design" / "transmembrane proteins"}

Abstract of research paper on Biological sciences, author of scientific article — Feras Hatahet, Lloyd W. Ruddock

Abstract The transmembrane enzymes disulfide bond forming enzyme B (DsbB) and vitamin K epoxide reductase (VKOR) are central to oxidative protein folding in the periplasm of prokaryotes. Catalyzed formation of structural disulfide bonds in proteins also occurs in the cytoplasm of some hyperthermophilic prokaryotes through currently, poorly defined mechanisms. We aimed to determine whether DsbB and VKOR can be inverted in the membrane with retention of activity. By rational design of inversion of membrane topology, we engineered DsbB mutants that catalyze disulfide bond formation in the cytoplasm of Escherichia coli. This represents the first engineered inversion of a transmembrane protein with demonstrated conservation of activity and substrate specificity. This successful designed engineering led us to identify two naturally occurring and oppositely oriented VKOR homologues from the hyperthermophile Aeropyrum pernix that promote oxidative protein folding in the periplasm or cytoplasm, respectively, and hence defines the probable route for disulfide bond formation in the cytoplasm of hyperthermophiles. Our findings demonstrate how knowledge on the determinants of membrane protein topology can be used to de novo engineer a metabolic pathway and to unravel an intriguingly simple evolutionary scenario where a new “adaptive” cellular process is constructed by means of membrane protein topology inversion.

Academic research paper on topic "Topological Plasticity of Enzymes Involved in Disulfide Bond Formation Allows Catalysis in Either the Periplasm Or the Cytoplasm"

Featured Article

Topological Plasticity of Enzymes Involved in Disulfide Bond Formation Allows Catalysis in Either the Periplasm Or the Cytoplasm

Feras Hatahet and Lloyd W. Ruddock

Department of Biochemistry, Linnanmaa Campus, University of Oulu, 90570 Oulu, Finland

Correspondence to Lloyd W. Ruddock: Lloyd.ruddock@oulu.fi

http://dx.doi.org/10.1016/j.jmb.2013.04.034

Edited by I. B. Holland

Feras Hatahet and Lloyd W. Ruddock

Abstract

The transmembrane enzymes disulfide bond forming enzyme B (DsbB) and vitamin K epoxide reductase (VKOR) are central to oxidative protein folding in the periplasm of prokaryotes. Catalyzed formation of structural disulfide bonds in proteins also occurs in the cytoplasm of some hyperthermo-philic prokaryotes through currently, poorly defined mechanisms. We aimed to determine whether DsbB and VKOR can be inverted in the membrane with retention of activity. By rational design of inversion of membrane topology, we engineered DsbB mutants that catalyze disulfide bond formation in the cytoplasm of Escherichia coli. This represents the first engineered inversion of a transmembrane protein with demonstrated conservation of activity and substrate specificity. This successful designed engineering led us to identify two naturally occurring and oppositely oriented VKOR homologues from the hyperthermophile Aeropyrum pernix that promote oxidative protein folding in the periplasm or cytoplasm, respectively, and hence defines the probable route for disulfide bond formation in the cytoplasm of hyperthermophiles. Our findings demonstrate how knowledge on the determinants of membrane protein topology can be used to de novo engineer a metabolic pathway and to unravel an intriguingly simple evolutionary scenario where a new "adaptive" cellular process is constructed by means of membrane protein topology inversion.

© 2013 Elsevier Ltd. All rights reserved.

Legend: Disulfide bond formation in the periplasm of prokaryotes is catalyzed by the transmembrane proteins DsbB and VKOR. Engineered topology inversion of DsbB or natural inversion of VKOR in hyperthermophiles allows disulfide bond formation activity in the cytoplasm. The background picture is used with kind permission of the National Oceanic and Atmospheric Administration Office of Ocean Exploration and Research.

0022-2836/$ - see front matter © 2013 Elsevier Ltd. All rights reserved.

J. Mol. Biol. (2013) 425, 3268-3276

Introduction

The catalyzed formation of structural disulfide bonds in prokaryotes is usually limited to the periplasm. Two transmembrane proteins, thiol-disulfide oxidoreductase disulfide bond forming enzyme B (DsbB) and vitamin K epoxide reductase (VKOR), are responsible for de novo disulfide bond formation in this compartment [1]. DsbB has a helix-bundle architecture [2], composed of four transmembrane helices (TMHs), two periplasmic loops that include four essential catalytic cysteines and one cytoplasmic loop connecting TMH2 and TMH3. The N- and C-termini of DsbB are in the cytosol. This topology of DsbB is necessary for its function in transferring oxidizing equivalents, from a quinone [3,4] to periplasmic and secreted polypeptides, via the periplasmic thiol-disulfide oxidore-ductase disulfide bond forming enzyme A (DsbA) [2,5]. Hence, Escherichia coli strains deficient in either DsbB or DsbA show defects in oxidative protein folding in the periplasm. In contrast to what occurs in the periplasm, structural disulfide bonds do not readily form in the cytoplasm of most prokaryotic organisms due to the existence of reducing pathways, thioredoxin/thioredoxin reduc-tase and glutathione/glutathione reductase [6], and the absence of a de novo disulfide bond formation catalyst [7]. However, some hyperthermophiles do form structural disulfide bonds in cytoplasmic proteins [8,9]. One enzyme in this catalyzed system has been previously identified, protein disulfide oxidore-ductase (PDO) [9,10]. However, PDO is not a catalyst of de novo disulfide bond formation; rather, pDo probably acts as an intermediary similar to DsbA in the periplasm [5] or protein disulfide isomerase in the endoplasmic reticulum of eukaryotes [11].

Since no sulfhydryl oxidases have been identified in prokaryotes, it would seem to be a sensible assumption that a transmembrane DsbB or VKOR would be the de novo catalyst for disulfide bond formation in prokaryotic systems that generate structural disulfide bonds in cytoplasmic proteins such as hyperthermophiles. However, all reported family members of both proteins have their active sites located on the periplasmic side of the membrane and it is not known if they would be able to function with inverted topology.

Approximately 25% of all predicted open reading frames are expected to encode integral membrane proteins [12]. All transmembrane proteins adopt a specific topology, relative to the membrane lipid bilayer, which is related to function. The topology of helix-bundle transmembrane proteins, such as DsbB and VKOR, is thought to be mainly governed by the positive-inside rule: that there is an abundance of positively charged amino acids, lysine and arginine (K + R), on the cytoplasmic side of the membrane [13]. When the charge bias on either side of the

membrane is not decisively in favor of a particular orientation, a protein can have mixed orientations in the membrane, that is, dual topology, as established for a few E. coli proteins including the drug transporters CrcB and EmrE [14,15]. During evolution, transmembrane segment orientation can be altered by changing the charge distribution [16,17]. Additionally, it is possible to mimic evolution by manipulating the charge bias to control the topology of a membrane protein [18]. However, while topology has previously been inverted, including with retention of biological activity, it has not yet been demonstrated that any topology-manipulated proteins retain biological specificity.

Here we asked whether E. coli DsbB could be engineered to flip its topology in the membrane while retaining its DsbA and quinone-dependent disulfide bond forming activity, such that oxidative protein folding could happen in the cytoplasm. We found that DsbB exhibits a remarkable topological plasticity and robustness in cross-compartment catalysis of disulfide bond formation. That prokaryotic transmembrane catalysts of disulfide bond formation could catalyze disulfide bond formation in the cytoplasm leads us to identify a naturally evolved, topologically inverted catalyst of disulfide bond formation in hyperthermophilic archaea. We hypothesize that, by gene duplication of VKOR and subsequent shifting, the balance of positively charged residues that some hyperthermophilic archaea have evolved a pathway to support disulfide bonds formation in cytoplasmic proteins, an adaptation that increases protein stability at extreme temperature.

Results

DsbB as a model for membrane protein topogenesis

Examination of whether it is possible to invert DsbB topology while retaining its biological activity requires a specific assay to quantify biological activity on both sides of the inner membrane. Alkaline phosphatase (pPhoA, where p indicates periplasmic) is a natural substrate for the DsbB-DsbA system that contains two disulfide bonds that are essential for its phosphatase activity [19]. Hence, endogenous pPhoA exhibits low activity (0.21 |jM product/min) when expressed in a AdsbBstrain, with a circa 15-fold increase in active pPhoA (3.20 |M product/min) upon co-expression of wild-type DsbB (see Tables S1 and S2 for strains and plasmids used in this study). Similarly, expression of alkaline phosphatase devoid of its periplasm targeting sequence (denoted cPhoA for cytoplasmic PhoA) does not result in the production of active protein (0.09 |M product/min) in the reducing environment

of the cytoplasm unless an active catalyst of de novo disulfide bond formation is present [7]. We therefore reasoned that pPhoA and cPhoA can be used as reporters for periplasmically or cytoplasmically localized disulfide bond forming activity of DsbB as long as DsbA was present in the appropriate compartment to mediate electron transfer, for example, with endogenous DsbA in the periplasm and by co-expression of a cytoplasmic version of DsbA (cDsbA).

There is limited evidence for a function of the cytoplasmic regions of DsbB or for cytoplasmic factors in the function of DsbB. Therefore, our strategy for the inversion of topology of DsbB was based on the systematic reduction of the number of lysine and arginine residues that normally reside in the cytoplasm (Fig. 1). Wild-type DsbB has nine (K + R) in the cytoplasm and five (K + R) on the same face as the active site in the periplasm (DsbB9/5; see Table S2 for schematics of the DsbB mutants), as well as one R in the middle of TMH3. Point mutations K68Q (DsbB8/5) or K68Q/R72N (DsbB7/5) removing one or two of the cytoplasmic positive charges results in no change in cPhoA or pPhoA activity (Fig. 2) and hence no change in topology (Fig. 1a). While deletion of the C-terminus of DsbB (AK169-R178, DsbB4/5) reduces periplasmic activity by 30% with no observed cytoplasmic activity, a further point mutation R5N (DsbB3/5a) or K68Q (DsbB3/5b), which result in the removal of six (K + R) residues from the side opposite to where the active site is located, conferred low but significant cPhoA activity (Fig. 2). This corresponds to a partial

topology inversion of these mutant DsbBs. Combining R5N/K68Q (DsbB25) or making the additional R72N substitution in this C-terminal truncated background (DsbB1/5) increased cPhoA activity substantially compared with wild-type DsbB, indicating major topology inversion toward the cytoplasm. The gain of cytoplasmic activity was concomitant with a decrease in periplasmic activity to around 46% of wild-type for DsbB1/5 (Fig. 2). Hence, DsbB 2/5 and DsbB 1/5 can simultaneously catalyze oxidative protein folding in two different cellular compartments indicating mixed or dual topology of these mutants (Fig. 1b).

Relocalization of DsbB activity across the membrane with conservation of specificity

In order to force DsbB toward a fully inverted topology, we fused either the first or the last TMH from E. coli maltose transporter MalF [20] to our truncated and mutated DsbB constructs (Fig. 1c). A C-terminal fusion of the terminal TMH from MalF, which has three (K + R) after the TMH, after F168 of AK169-R178 R5N/K68Q DsbB (DsbBHf8) and the subsequent R72N mutation (DsbBH5/8), resulted in the formation of active proteins but had a relatively small effect on shifting the topology toward inversion compared with DsbB2/5 or DsbB1/5 (Fig. 2). In contrast, replacement of the N-terminus of a similar C-terminal truncated DsbB with a fusion of the first TMH from MalF, which has four (K + R) prior to the TMH, to give H0DsbB3/9 resulted in a protein that displayed low cPhoA and pPhoA activity. However,

Fig. 1. Original and engineered membrane topology of DsbB. (a) The active-site cysteines (yellow circles) of wild-type DsbB are periplasmic. Electrons (red arrows) flow from folding proteins in the periplasm to DsbA, then to DsbB and then to quinones. (b) Reducing the number of lysine and arginine residues (light-blue circles) that are naturally on the cytoplasmic side of DsbB induces partial topology inversion so that it can catalyze disulfide bond formation in the periplasm and the cytoplasm providing that DsbA is expressed in both compartments. (c) Addition of an N-terminal TMH fusion to the DsbB mutants allows oxidative folding to occur only in the cytoplasm.

со cr DO

CO CT СП

со ст □3

со ст ГО

ст ГО

со ст CD

w CT DD

О со er Ш X

ст ГО X

Fig. 2. The activity of DsbB can be inverted across the membrane. DsbB mutants were expressed under an arabinose promoter in pLysSBAD in the AdsbB strain FSH8. The periplasmic activity of DsbB mutants was determined by measuring the activity of pPhoA (white bars) after growth in low-phosphate minimal media while the cytoplasmic activity was determined by expressing cPhoA and cDsbA under a lac promoter in LB media and measuring cPhoA activity (black bars). Simultaneous high pPhoA and cPhoA activities indicate dual topology. N-terminal fusion of the first TMH, H0(Nin) of MalF (M1-G38), to these mutants results in inverted topology, while C-terminal fusions of the last TMH,

H5(Cin) of MalF (G478-D515), do not. Data represents mean relative activity (%) ± SD (n = 4). The maximal endogenous pPhoA activity was 3.20 ^M product/min, the maximal exogenous cPhoA activity was 8.96 ^M product/min. The cPhoA (0.09 ^M product/min) and pPhoA (0.21 ^M product/min) activities from the negative control, the vector without DsbB, have been subtracted. Schematic diagrams of the DsbB mutants can be found in Table S2.

со cr

со ty DP

<л ст ГО

mi ni I mi: ли

a single additional mutation, K68Q in L3 (HoDsbB279), leads to a dramatic increase in cytoplasmic activity (Fig. 2), suggesting topological instability of H0DsbB3/9. A further point mutation, R72N giving H0DsbB1/9a, showed minimal pPhoA activity, but very high cPhoA activity, above the vector negative control (Fig. 2). A similarly charged species with K68N instead of K68Q (H0DsbB1/9b) also showed minimal pPhoA activity but had increased cPhoA activity, presumably linked to changes in structure. These results suggest that, for DsbB, N-terminal topological signals are stronger than those that are C-terminal. This is in contrast to what was observed for the drug transporter variant EmrE-TMH5 [21] where a single positively charged amino acid can change the variants topology even when placed at the C-terminal region.

H0DsbB1/9b did not restore DsbB-dependent motility (Fig. S1) that, combined with the PhoA activity data, suggests a complete relocalization of its disulfide forming activity from the periplasm to the cytoplasm, that is, full inversion of topology (Fig. 1c).

As a control for non-specific effects that may be associated with recombinant expression of a transmembrane protein, subcellular fractionation of E. coli expressing H0DsbB1/9b, cDsbA and cPhoA was performed. This showed that PhoA activity is retained mostly in the cytoplasm and that the subcellular localization of internal markers such as ß-galactosidase and ß-lactamase was as expected (Table 1).

Wild-type DsbB uses quinone and DsbA as substrates. To examine whether our inverted DsbB proteins also use quinone, we expressed one in E.

Table 1. Subcellular fractionation of BL21(DE3) E. coli strains expressing H0DsbB1/9b

Activity (mAU/min), Activity (mAU/min),

Marker Fraction cPhoA + cDsbA co-expression cPhoA co-expression

ß-Galactosidase Total 9.43 ± 0.01 6.29 ± 0.32

ß-Galactosidase Cytoplasmic 9.88 ± 1.02 6.11 ± 0.50

ß-Galactosidase Periplasmic 0.00 ± 0.00 0.00 ± 0.00

ß-Lactamase Total 35.1 ± 0.8 34.9 ± 1.4

ß-Lactamase Cytoplasmic 0.92 ± 0.08 1.04 ± 0.02

ß-Lactamase Periplasmic 29.6 ± 1.6 30.6 ± 1.1

cPhoA Total 26.2 ± 3.9 1.77 ± 0.03

cPhoA Cytoplasmic 20.3 ± 3.8 0.24 ± 0.01

cPhoA Periplasmic 0.84 ± 0.22 0.59 ± 0.17

Absorbance of hydrolyzed ONPG (0.4 mg/ml), Nitrocefin (0.1 mg/ml) and PNPP (1 mg/ml) was measured over time at 420, 486 and 410 nm, respectively. Data represent mean ± SD (n = 2).

Fig. 3. The activity of inverted DsbB retains the same specificity and has the same mechanism as the wild-type protein. (a) Efficient oxidative folding in the cytoplasm catalyzed by H0DsbB1/9a requires an electron transport chain. cPhoA activity was determined in the E. colistrain An384 (AubiA AmenA) with 0.04 mM or 2 mM quinone precursor hydroxybenzoic acid (HBA) or in the AdsbB strain FSH8. Data represent mean relative activity (%) to the maximal exogenous cPhoA activity (8.96 ^M product/min) ± SD (n = 4). (b) The activity of inverted DsbB requires expression of cDsbA. Expression of cPhoA ± cDsbA from T7 promoter of pET23 in the E. coli strain BL21(DE3) pLysS with or without pre-expression of H0DsbB1/9b from an arabinose promoter in pLysSBAD. For these experiments, induction of H0DsbB was with 0.5% arabinose. (c) The activity of inverted DsbA requires the active-site Cys46. Expression of cPhoA + cDsbA from pLysSBAD (induction with 0.5% arabinose) with pre-induction of wild-type DsbB, H0DsbB1/9b or the active-site mutant H0DsbB1/9b C46A from the T7 promoter of pET23 (using 10 ^M IPTG). For (b) and (c), the data represent the mean relative activity (%) to the maximal cPhoA activity (2.31 ^M product/min) ± SD (n = 2). For all panels, the cPhoA activity from the negative control, the vector without DsbB, has been subtracted.

coli strain AN384 (AubiA420 AmenA401) [22], which is deficient for ubiquinone and menaquinone biosynthesis, and tested for cPhoA activity. Consistent with the results for wild-type DsbB [23], H0DsbB1/9a failed to catalyze cytoplasmic disulfide bond formation in AN384 (Fig. 3a), indicating that quinone dependence is retained with DsbB inversion. Similarly, inverted DsbB (H0DsbB1/9b) shows the same dependence on DsbA as an intermediary as the wildtype protein does (Fig. 3b). Furthermore, H0DsbB1/9b cytoplasmic activity was dependent on Cys46, one of the four essential catalytic cysteine residues [24] in DsbB (Fig. 3c). Hence, inversion of DsbB occurs with conservation of activity and substrate (DsbA and quinone) specificity.

Naturally inverted topology and adaptation to extreme conditions

Disulfide bond formation in bacteria and archaea is catalyzed either by DsbB or by the transmembrane enzyme VKOR [1,25]. As per DsbB, VKOR family members have four TMHs that form a helix bundle (Fig. 4a). That DsbB showed topological plasticity and that cytoplasmic disulfide bonds are readily formed once catalysts for their formation are co-localized (results above and our previous published results using soluble sulfhydryl oxidases [7]) sug-

gested to us that the occurrence of naturally occurring topologically inverted catalysts of disulfide bond formation is evolutionarily plausible.

To test our hypothesis, we systematically searched within the DsbB and VKOR families for members that are predicted to have their catalytic residues in the cytosol. Using global membrane protein topology prediction algorithms (SCAMPI [26]), we searched 1042 DsbB and 243 VKOR bacterial and archaeal homologues from Pfam [27]. While no naturally occurring DsbB members with predicted inverted topology were found, five species containing VKOR homologues that are predicted to have cytoplasmic localization of their catalytic sites were identified, all of which belong to the hyperther-mophilic crenarchaeon phylum (Table 2). Each of these species had two VKOR homologues in their genome of similar size and number of TMHs (Fig. S2) but that are predicted to possess opposite topologies. Two of these organisms, Aeropyrum pernix and Pyrobaculum aerophilum, have been shown to have cytoplasm proteins with disulfide bonds and have the strongest correlation between cytoplasmic disulfide abundance and hyperthermo-philicity [8]. To test the topology prediction, we expressed ApVKORo (active-site cysteines outside; i.e., in the periplasm) and ApVKORi (active-site cysteines inside; i.e., in the cytoplasm) from the

(8.96 ^M product/min) activity ± SD (n = 4). The cPhoA and pPhoA activities without DsbB or VKOR, have been subtracted.

Fig. 4. A. pernix has two copies of VKOR that have different topologies. (a) Comparison of the topology of DsbB and VKOR [1]. While both comprise of a core four-helix bundle, VKOR from some species have an additional TMH. A naturally inverted VKOR whose active site is located toward the cytoplasm is identified here. The active-site cysteines are indicated by black circles. (b) When expressed in the AdsbB E. coli strain FSH8 under the pBAD promoter, ApVKORo and ApVKORi can catalyze periplasmic and cyto-plasmic disulfide bond formation, respectively, as determined by the activities of pPhoA (white bars) and cPhoA (black bars). (c) ApVKORi dependence on cDsbA. Cells harboring vectors encoding cPhoA or cPhoA + cDsbA were cotrans-formed with empty vector or with ApVKORi. Data represent mean relative activity (%) to the maximal endogenous pPhoA (3.20 ^M product/min) or exogenous cPhoA from the negative control, the vector

hyperthermophile A. pernix in E. coliand tested them for disulfide bond formation in the cytoplasm and the periplasm. Expression of ApVKORo induced disulfide bond formation only in pPhoA, whereas ApV-KORi induced disulfide bond formation only in cPhoA (Fig. 4b), with ApVKORi activity being dependent on co-expression of cDsbA (Fig. 4c). This indicates that both A. pernix VKOR family members can catalyze disulfide bond formation but that they have opposite membrane orientation of their active sites with one catalyzing disulfide formation in the periplasm and the other in the cytoplasm.

Table 2. Species containing VKOR homologues that are predicted to have cytoplasmic localization of their catalytic sites

Accession Predicted topology

Organism number of active site

A. pernix Q9Y922 Periplasmic

A. pernix Q9YB70 Cytoplasmic

P. aerophilum Q8ZV09 Periplasmic

P. aerophilum Q8ZXF9 Cytoplasmic

Pyrobaculum arsenaticum A4WKP9 Periplasmic

P. arsenaticum A4WI60 Cytoplasmic

Pyrobaculum calidifontis A3MSV5 Periplasmic

P. calidifontis A3MVJ3 Cytoplasmic

Pyrobaculum islandicum A1RVU0 Periplasmic

P. islandicum A1RVC9 Cytoplasmic

Each of the hyperthermophiles is predicted to have two VKOR homologues with opposite topologies.

Discussion

To invert the topology of transmembrane proteins, the charge distribution of K + R must be changed by either deletion or insertion. Here inversion of DsbB by such mutagenesis allows oxidation of cDsbA and thereby catalyzes disulfide bond formation in the cytoplasm. The mutations made in DsbB are located throughout the protein, and the extent of topology inversion, as seen by activation of cPhoA, increases as the topological bias of K + R changes. However, it appears that topological confusion can occur for TMH2 in H0DsbB3/9, as evidenced by defects in both periplasmic and cytoplasmic DsbB activity and the fact that activity is regained by the replacement of a single charged residue N-terminal on either side of TMH2. While the overall charge difference is in favor of cytoplasmic localization of active sites for the H0DsbB3/9 construct, local charge discrimination of orientation is not satisfied for TMH2 as it is flanked at opposite ends by two positively charged residues R50 (essential forquinone interaction [28]) and K68. This might give rise to topological frustration of TMH2, a phenomenon that has been observed previously for some diagnostic hybrids of E. coli Leader peptidase [29]. Therefore, while the overall transmembrane charge difference may favor a particular orientation, appropriate local charge distribution around individual TMHs is also required to prevent local topology defects.

While it might seem paradoxical that oxidizing and reducing pathways can co-exist in the cytoplasm of E. coli and other organisms, this is not at odds with what is known about disulfide bond formation in other organelles such as the endoplasmic reticulum and the periplasm, where native disulfide bond formation and disulfide reduction occur concomi-tantly [30-32]. In particular, the dependency of catalyzed disulfide bond formation by inverted DsbBs or VKORs on cDsbA minimizes potential futile and deleterious engagement of reducing and oxidizing pathways by a mechanism reminiscent to the situation in the periplasm.

The emergence of new metabolic pathways, by evolution or through protein engineering, may reach the same conclusions, with ApVKORi also requiring an intermediary protein to function in the cytoplasm of E. coli. In our experiments, this was cDsbA, but in the cytoplasm of A. pernix, the most likely intermediary is PDO [9,10]. Preliminary data suggest that ApPDO can mediate electron transfer between ApVKORi and cPhoA in the cytoplasm of E. coli but that this process is inefficient (data not shown). This low efficiency may arise due to the reconstitution experiment being performed in E. coli at 37 °C while the optimal growth temperature for A. pernix is 95 °C.

The molecular mechanism by which ApVKOR evolved to have two distinct VKOR family members with different localized disulfide bond formation activities appears to be a gene duplication event followed by the redistribution of positively charged amino acids across the sequence of one copy by deletion and mutational substitution. The appearance of two copies of VKOR might reflect differential regulatory and/or efficiency requirements for disulfide bond formation in the cytoplasm and periplasm, something that could not be met by a single VKOR with dual topology.

To date, there have been few cases where membrane proteins were rationally designed to attain specific topology (e.g., Refs. [16] and [21]). However, our findings present the first example where the enzymatic activity, including substrate specificity, of a membrane protein is inverted across the membrane. The apparent topological dynamism of DsbB reported here makes it an invaluable model to advance our understanding of membrane protein topogenesis and evolution. We have in essence replicated and thereafter discovered an evolutionary event that most likely augmented cytoplasmic disulfide bond formation and adaptation to extreme conditions in some extreme thermophiles [8].

Materials and Methods

Strains and DNA constructs

Standard molecular biology techniques and P1 trans-duction [33,34] were used for the construction of strains

and expression vectors as described in Supplementary Materials and Methods. All E. coli strains and plasmids used in this study are listed in Tables S1 and S2. Schematics of the DsbB mutants can be found in Table S2.

For expression in BL21(DE3), cytoplasmic DsbA (cDsbA; A20-K208, with an N-terminal MHHHHHHM-tag) was cloned into pVD80 [7], a pET23a derived plasmid that encodes for cytoplasmic PhoA (cPhoA; R22-K271, with an N-terminal MHHHHHHM-tag) to give a bicistronic plasmid pFH258 with expression of both genes under a single T7 promoter.

Expression in other E. coli strains under a lac promoter/ repressor was based on using a modified version of pMALc2x (New England Biolabs). First, a Spel site was introduced upstream of the start of MBP gene to give pFH313. The bicistronic DNA fragment coding for cPhoA and cDsbA was cloned from pFH258 into pFH313 to generate pFH314. This replaces the original ribosome binding site and MBP gene of pFH313 with cPhoA and cDsbA each preceded by a ribosome binding site carried along from pFH258.

For expression of DsbB and VKOR variants, we wanted a low-copy plasmid that is compatible for cotransformation with pFH258 or pFH314. We used pLysS (Promega) as the backbone and cloned into it pBAD/araC as described previously [35]. The gene encoding DsbB was amplified by colony PCR from the E. coli strain BL21 and cloned into this vector and into pET23. The longer DsbB variant in the database (AAA23711.1), encoding a 178-amino-acid protein, was chosen as the template. The shorter variant is 176 amino acids in length and lacks the first two amino acids Met and Ile. Lysine and arginine residues in cytoplasmic loops were mutated individually (R5N, R12N, K68N, K68Q and R72N) or in combination along with the active-site cysteine (C46A) using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturers' instructions. The first and last TMHs of E. coli MalF (M1-G38) and (G478-D515) were amplified by PCR and cloned in-frame into vectors harboring DsbB variants. This adds the two amino acids PW or GS before the start or after the end of DsbB variant, respectively. ApVKORo and ApVKORi were amplified by PCR from the genomic DNA of A. pernix (DSM 11879, German Collection of Microorganisms and Cell Cultures) and cloned into the pLysSBAD vector. All plasmids were sequenced to ensure no PCR or cloning errors prior to use.

Alkaline phosphatase activity

Localization of DsbB activity was inferred by measuring the activity of alkaline phosphatase in FSH8 (which is AdsbB and ara+) transformed with pLysSBAD harboring different mutants of DsbB and also by monitoring motility on soft agar.

For endogenous pPhoA, activity cells were grown in 5-ml fresh Mops media {Mops salts [34], 50 Mg/ml thiamine, 1 mM MgSO4, 0.5% glycerol, 0.2% N-Z-Case plus (Sigma), 0.0001% arabinose and 35 Mg/ml chloramphenicol} without addition of phosphate in order to induce chromosomal expression of PhoA. After 14 h of growth, cells were collected by centrifugation and resuspended in lysis buffer [50 mM Tris (pH 8.0), 50 mM N-ethyl malei-mide, 20 Mg/ml DNase and 0.1 mg/ml egg white lysozyme] to give a suspension equivalent to OD600 of 1 based on the final OD600 of the culture and then frozen.

For exogenous cPhoA activity in the E. colistrain FSH8, 5-ml pre-cultures, containing suitable antibiotics, were used to seed a 20-ml culture of LB in a 125-ml flask to an OD600of0.05. This was then grown at 37 °C,200 rpm, until an OD600 reached 0.4 when expression from pLysSBAD was induced with 0.5% arabinose and 0.08% glucose. Thirty minutes later, expression from the pMal plasmid was induced with 1 mM IPTG and cells were harvested after 3 h. The rationale behind including 0.08% glucose is that expression of the transmembrane protein DsbB can be toxic causing growth cessation [36]. The addition of glucose brings on catabolite repression of the pBAD promoter [37] and thereby reduces expression of DsbB. The combination of arabinose and glucose at specific ratios proved to be very useful to regulate expression of DsbB from the pBAD promoter without causing growth defects (Table S3). Harvested cells were resuspended in lysis buffer as above and frozen. A similar protocol was used for exogenous cPhoA activity in the E. coli strain BL21(DE3) except that the DsbB constructs were induced first with either 10 mM IPTG (pET23) or 0.5% arabinose (pLysS-BAD), with cPhoA and cDsbA being induced 30 min later with 1 mM IPTG (pET23) or 0.5% arabinose (pLysSBAD).

Alkaline phosphatase activity for cPhoA and pPhoA samples was measured in a continuous assay by monitoring AA410 upon the hydrolysis of 4-nitrophenyl phosphate [0.1% (w/v) in 1 M Tris (pH 8.0)] with a plate reader using 5 M of cleared lysate and 195 M of substrate in a 96-well plate at 25 °C. aA410 values were corrected to a pathlength of 1 cm and converted into changes in product concentration using a molar extinction coefficient of 18,300 M-1 cm-1 [38].

Motility assay

A single colony from JW5182 strain carrying DsbB constructs was stabbed into a motility plate [M63 salts, 1 mM MgSO4, 50 Mg/ml thiamine, 0.4% glycerol, 0.1% N-Z-Case plus (Sigma), 0.3% (w/v) agar, 0.0001% arabinose and appropriate antibiotics]. Motility halos were examined after 48 h growth at 30 °C.

Subcellular fractionation

For cell fractionation studies, none of the AdsbB strains could be used as they all lack p-galactosidase. Instead, BL21(DE3) cells, which are lacZ+ harboring the plasmid pFH272 encoding H0DsbB1/9b with either pFH258 (encoding cPhoA and cDsbA) or pVD80 (encoding cPhoA), were grown in LB as above except that cells were collected 2 h post IPTG induction. One milliliter of culture was spun down, and the cell pellet was resuspended in 0.5 ml of 20% sucrose buffer containing 33 mM Tris (pH 8.0), 1 mM ethylenediaminetetraacetic acid and 50 mM N-ethyl mal-eimide and left at room temperature for 10 min. The cells were then spun down at 4000g for 5 min at +4 °C, the supernatant was discarded and the cells were resus-pended in ice-cold 5 mM MgSO4 to give an OD600 equivalent of 5 and left on ice for 20 min. A 100-M sample was removed and marked as total, and another 100 M was spun down at 10,000g for 10 min; the periplasmic content was removed to a new tube. The pellet was resuspended to the original volume with 5 mM MgSO4,20 Mg/ml DNase, 0.1 mg/ml lysozyme and 50 mM Tris (pH 8.0) added to all fractions and then freeze-thawed twice. Subsequently

activities of p-galactosidase (cytoplasmic marker), p-lactamase (periplasmic marker) and alkaline phosphatase were determined. p-Galactosidase was assayed [33] using ortho-nitrophenyl-p-D-galactopyranoside (Sigma) as a substrate. p-Lactamase was assayed [39] using Nitro-cefin (Calbiochem) as a substrate. Alkaline phosphatase activity was determined as above.

Bioinformatic analysis

Accession numbers of PF02600 and PF07884 family members were collected from Pfam and used to retrieve full sequences from UniProt. Sequences that appeared to lack catalytic cysteines were excluded, the remaining sequences were run on SCAMPI [26] to predict transmembrane regions and topology and selected sequences were further analyzed by TOPCONS [40]. We further examined all VkOr homologues from thermophiles by BLASTing ApVKORi within Archaea. Sequences were aligned using ClustalW2 and drawn with Bioedit.

Acknowledgements

We would like to Kenji Inaba for a generous gift of materials. This work was financially supported by the Academy of Finland and the University of Oulu. A patent application has been filed.

Supplementary Data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jmb.2013.04.034

Received 4 March 2013; Received in revised form 29 April 2013;

Accepted 30 April 2013 Available online 28 June 2013

Keywords:

DsbB; VKOR; topology inversion; rational design; transmembrane proteins

Present address: F. Hatahet, Department of Microbiology and Immunobiology, Harvard Medical School, Boston,

MA 02115, USA.

Abbreviations used:

PDO, protein disulfide oxidoreductase; TMH, transmembrane helix; VKOR, vitamin K epoxide reductase.

References

[1] Dutton RJ, Boyd D, Berkmen M, Beckwith J. Bacterial species exhibit diversity in their mechanisms and capacity for protein disulfide bond formation. Proc Natl Acad Sci USA 2008;105:11933-8.

[2] Inaba K, Murakami S, Suzuki M, Nakagawa A, Yamashita E, Okada K, et al. Crystal structure of the DsbB-DsbA complex reveals a mechanism of disulfide bond generation. Cell 2006;127:789-801.

[3] Bader MW, Xie T, Yu CA, Bardwell JC. Disulfide bonds are generated by quinone reduction. J Biol Chem 2000;275: 26082-8.

[4] Takahashi YH, Inaba K, Ito K. Characterization of the menaquinone-dependent disulfide bond formation pathway of Escherichia coli. J Biol Chem 2004;279:47057-65.

[5] GuilhotC, JanderG, Martin NL, Beckwith J. Evidence that the pathway of disulfide bond formation in Escherichia coli involves interactions between the cysteines of DsbB and DsbA. Proc Natl Acad Sci USA 1995;92:9895-9.

[6] Prinz WA, Äslund F, Holmgren A, Beckwith J. The role of the thioredoxin and glutaredoxin pathways in reducing protein disulfide bonds in the Escherichia coli cytoplasm. J Biol Chem 1997;272:15661-7.

[7] Hatahet F, Nguyen VD, Salo KE, Ruddock LW. Disruption of reducing pathways is not essential for efficient disulfide bond formation in the cytoplasm of E. coli. Microb Cell Fact 2010;9:67.

[8] Beeby M, O'Connor BD, Ryttersgaard C, Boutz DR, Perry LJ, Yeates TO. The genomics of disulfide bonding and protein stabilization in thermophiles. PLoS Biol 2005;3:e309.

[9] Pedone E, Limauro D, Bartolucci S. The machinery for oxidative protein folding in thermophiles. Antioxid Redox Signaling 2008;10:157-69.

[10] Ren B, Tibbelin G, de Pascale D, Rossi M, Bartolucci S, Ladenstein R. A protein disulfide oxidoreductase from the archaeon Pyrococcus furiosus contains two thioredoxin fold units. Nat Struct Biol 1998;5:602-11.

[11] Hatahet F, Ruddock LW. Protein disulfide isomerase: a critical evaluation of its function in disulfide bond formation. Antioxid Redox Signaling 2009;11:2807-50.

[12] Wallin E, von Heijne G. Genome-wide analysis of integral membrane proteins from eubacterial, archaean, and eukary-otic organisms. Protein Sci 1998;7:1029-38.

[13] von Heijne G, Gavel Y. Topogenic signals in integral membrane proteins. Eur J Biochem 1988;174:671-8.

[14] Daley DO, Rapp M, Granseth E, Melen K, Drew D, von Heijne G. Global topology analysis of the Escherichia coli inner membrane proteome. Science 2005;308:1321-3.

[15] Nasie I, Steiner-Mordoch S, Gold A, Schuldiner S. Topolog-ically random insertion of EmrE supports a pathway for evolution of inverted repeats in ion-coupled transporters. J Biol Chem 2010;285:15234-44.

[16] von Heijne G. Control of topology and mode of assembly of a polytopic membrane protein by positively charged residues. Nature 1989;341:456-8.

[17] Lolkema JS, Dobrowolski A, Slotboom DJ. Evolution of antiparallel two-domain membrane proteins: tracing multiple gene duplication events in the DUF606 family. J Mol Biol 2008;378:596-606.

[18] Rapp M, Seppala S, Granseth E, von Heijne G. Emulating membrane protein evolution by rational design. Science 2007;315:1282-4.

[19] Derman AI, Beckwith J. Escherichia coli alkaline phospha-tase fails to acquire disulfide bonds when retained in the cytoplasm. J Bacteriol 1991;173:7719-22.

[20] Froshauer S, Green GN, Boyd D, McGovern K, Beckwith J. Genetic analysis of the membrane insertion and topology of MalF, a cytoplasmic membrane protein of Escherichia coli. J Mol Biol 1988;200:501-11.

[21] Seppala S, Slusky JS, Lloris-Garcera P, Rapp M, von Heijne G. Control of membrane protein topology by a single C-terminal residue. Science 2010;328:1698-700.

[22] Wallace BJ, Young IG. Role of quinones in electron transport to oxygen and nitrate in Escherichia coli. studies with a ubiA-menA- double quinone mutant. Biochim Biophys Acta 1977;461:84-100.

[23] Kobayashi T, Kishigami S, Sone M, Inokuchi H, Mogi T, Ito K. Respiratory chain is required to maintain oxidized states of the DsbA-DsbB disulfide bond formation system in aerobi-cally growing Escherichia coli cells. Proc Natl Acad Sci USA 1997;94:11857-62.

[24] Jander G, Martin NL, Beckwith J. Two cysteines in each periplasmic domain of the membrane protein DsbB are required for its function in protein disulfide bond formation. EMBO J 1994;13:5121-7.

[25] Goodstadt L, Ponting CP. Vitamin K epoxide reductase: homology, active site and catalytic mechanism. Trends Biochem Sci 2004;29:289-92.

[26] Bernsel A, Viklund H, Falk J, Lindahl E, von Heijne G, Elofsson A. Prediction of membrane-protein topology from first principles. Proc Natl Acad Sci USA 2008;105:7177-81.

[27] Finn RD, Tate J, Mistry J, Coggill PC, Sammut SJ, Hotz HR, et al. The Pfam protein families database. Nucleic Acids Res 2010;38:D211-22.

[28] Inaba K, Takahashi YH, Ito K, Hayashi S. Critical role of a thiolate-quinone charge transfer complex and its adduct form in de novo disulfide bond generation by DsbB. Proc Natl Acad Sci USA 2006;103:287-92.

[29] Gafvelin G, von Heijne G. Topological "frustration" in multi-spanning E. coli inner membrane proteins. Cell 1994;77:401-12.

[30] Ushioda R, Hoseki J, Araki K, Jansen G, Thomas DY, Nagata K. ERdj5 is required as a disulfide reductase for degradation of misfolded proteins in the ER. Science 2008;321:569-72.

[31] Chakravarthi S, Bulleid NJ. Glutathione is required to regulate the formation of native disulfide bonds within proteins entering the secretory pathway. J Biol Chem 2004;279:39872-9.

[32] Cuozzo JW, Kaiser CA. Competition between glutathione and protein thiols for disulphide-bond formation. Nat Cell Biol 1999;1:130-5.

[33] Miller JH. A Short Course in Bacterial Genetics. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1992.

[34] Sambrook J, Russell DW. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory; 2001.

[35] Nguyen VD, Hatahet F, Salo KE, Enlund E, Zhang C, Ruddock LW. Disruption of reducing pathways is not essential for efficient disulfide bond formation in the cytoplasm of E. coli. Microb Cell Fact 2011;10:1.

[36] Bader M, Muse W, Zander T, Bardwell J. Reconstitution of a protein disulfide catalytic system. J Biol Chem 1998;273:10302-7.

[37] Miyada CG, Stoltzfus L, Wilcox G. Regulation of the araC gene of Escherichia coli: catabolite repression, autoregulation, and effect on araBAD expression. Proc Natl Acad Sci USA 1984;81:4120-4.

[38] Ulrich H, Bergmeyer E. Methods of Enzymatic Analysis. New York, NY: Academic Press; 1974.

[39] O'Callaghan CH, Morris A, Kirby SM, Shingler AH. Novel method for detection of ß-lactamases by using a chromo-genic cephalosporin substrate. Antimicrob Agents Che-mother 1972;1:283-8.

[40] Bernsel A, Viklund H, Hennerdal A, Elofsson A. TOPCONS: consensus prediction of membrane protein topology. Nucleic Acids Res 2009;37:W465-8.