Aeropyrum Pernix Membrane Topology of Protein VKOR Promotes Protein Disulfide Bond Formation in Two Subcellular Compartments

RESEARCH ARTICLE Hibender et al., Microbiology 2017;163:1864–1879 DOI 10.1099/mic.0.000569

Aeropyrum pernix membrane topology of protein VKOR promotes protein disulfide bond formation in two subcellular compartments

Stijntje Hibender,1† Cristina Landeta,2† Mehmet Berkmen,3 Jon Beckwith2 and Dana Boyd2,*

Abstract Disulfide bonds confer stability and activity to proteins. Bioinformatic approaches allow predictions of which organisms make protein disulfide bonds and in which subcellular compartments disulfide bond formation takes place. Such an analysis, along with biochemical and protein structural data, suggests that many of the extremophile Crenarachaea make protein disulfide bonds in both the cytoplasm and the cell envelope. We have sought to determine the oxidative folding pathways in the sequenced genomes of the Crenarchaea, by seeking homologues of the enzymes known to be involved in disulfide bond formation in bacteria. Some Crenarchaea have two homologues of the cytoplasmic membrane protein VKOR, a protein required in many bacteria for the oxidation of bacterial DsbAs. We show that the two VKORs of Aeropyrum pernix assume opposite orientations in the cytoplasmic membrane, when expressed in E. coli. One has its active cysteines oriented toward the E. coli periplasm (ApVKORo) and the other toward the cytoplasm (ApVKORi). Furthermore, the ApVKORo promotes disulfide bond formation in the E. coli cell envelope, while the ApVKORi promotes disulfide bond formation in the E. coli cytoplasm via a co-expressed archaeal protein ApPDO. Amongst the VKORs from different archaeal species, the pairs of VKORs in each species are much more closely related to each other than to the VKORs of the other species. The results suggest two independent occurrences of the evolution of the two topologically inverted VKORs in archaea. Our results suggest a mechanistic basis for the formation of disulfide bonds in the cytoplasm of Crenarchaea.

INTRODUCTION organisms, disulfide-bonded proteins are restricted to non- cytoplasmic compartments (i.e. the cell envelope in bacteria Archaea represents the third domain of life and consist of and the endoplasmic reticulum and intermembrane space of two major subdivisions, one being the Crenarchaeota that mitochondria in eukaryotes) given to the presence of reduc- includes species of microorganisms that can survive in the ing pathways in the cytosol. Since these non-cytoplasmic highest known growth temperatures. These organisms have compartments are oxidizing, this ‘environment’ per se was developed unusual adaptations of the cell envelope to originally considered sufficient to promote disulfide bond extreme conditions. The archaeal cell envelope consists of formation. Conversely, the reducing ‘environment’ of the an S-layer (surface layer) cell wall composed of glycopro- cytoplasm was thought to be sufficient to prevent disulfide teins, in some cases associated with the cytoplasmic mem- bond formation. brane [1]. However, two lines of research have raised questions about Structural disulfide bonds link cysteines in a protein and these explanations for the restricted location of proteins result from enzymatically catalysed oxidative processes. The with disulfide bonds. First, genetic and biochemical studies covalent bond formed by the oxidation of two cysteines showed that the bacterial cell envelope and the endoplasmic results in stabilization of the folded protein and, in most reticulum contain enzymes that are essential for the efficient cases, contributes to its thermostability. For most formation of covalent bonds between the cysteine residues

Received 15 June 2017; Accepted 24 October 2017 Author affiliations: 1Faculty of Science, University of Amsterdam, Postbus 94216, 1090 GE Amsterdam, The Netherlands; 2Department of Microbiology and Immunobiology, Harvard Medical School, Boston, MA 02115, USA; 3New England Biolabs, Ipswich, MA 01938, USA. *Correspondence: Dana Boyd, [email protected] Keywords: VKOR; archaea; Disulfide bond formation; membrane topology. Abbreviations: DSB, Disulfide bond formation; IPTG, Isopropyl b-D-1-thiogalactopyranoside; MIC, Minimal Inhibitory Concentration; PDO, Protein Disul- fide Oxidoreductase; PNPP, para-Nitrophenyl-phosphate; VKOR, Vitamin K epoxide Reductase; X-Gal, 5-Bromo-4-chloro-3-indolyl galactopyranoside; XP, 5-Bromo-4-chloro-3-indolyl phosphate; Dss, signal sequence less; b-Galdbs, disulfide-bond sensitive b-Galactosidase. LacZ fused to the membrane protein MalF, localizes LacZ in the periplasm making it sensitive to disulfide bond formation MalF-LacZ102. †These authors contributed equally to this work. One supplementary table and two supplementary figures are available with the online version of this article.

000569 ã 2017 The Authors Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521864 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879 of proteins [2–4]. In the absence of these enzymes, disulfide d-Proteobacteria use VKOR rather than DsbB to re-oxidize bonds were greatly diminished, indicating that the oxidizing DsbA [25]. Using the same approach, we determined that nature of the compartment alone did not suffice for their among Crenarchaea there is a significant bias for an even formation in vivo. Second, the absence of disulfide-bonded number of cysteines in the cytoplasm and cell envelope. proteins in the cytoplasm was shown not to be due simply Notably, some organisms have two VKOR homologues, one to its reductive environment. Genetic manipulation of with a strong prediction for oriention to the periplasm and E. coli allows efficient disulfide bond formation in the cyto- the other with either a clear or weak prediction for orienta- plasm catalysed by oxidized thioredoxins with reducing tion to the cytoplasm. In addition, among those that have a pathways still present [5–9]. In addition, certain viruses VKOR homologue facing the cytoplasm some also have a encode their own enzymes which, when expressed in the protein disulfide oxidoreductase (PDO) homologue as well cytoplasm of eukaryotic cells after viral infection, promote as a cytoplasmic DsbA-like protein. Amongst the VKORs disulfide bond formation [10]. In other words, the presence from different archaeal species, the pairs of VKORs in each of an oxidative catalytic system for creating disulfide bonds, species are much more closely related to each other than to when expressed in the cytoplasm, can function even though the VKORs of the other species. We chose Aeropyrum the cytoplasm is still a reducing environment. These latter pernix VKOR homologues for further study due to their findings raised the possibility that some organisms might unambiguous topology predictions and strong bias towards normally create disulfide bonds in their cytoplasmic pro- even-numbered cysteines in their cytoplasmic proteome. teins. In fact, Yeates et al. subsequently presented evidence We aimed to demonstrate the membrane topology of these that certain members of the Crenarchaea do create disulfide two proteins using membrane fusion with alkaline phospha- bonds in many of their cytoplasmic proteins [11–13]. Fur- tase in E. coli. We demonstrated that Q9YB70 (ApVKORi) ther support came from crystallization of Crenarchaeal orients its second pair of cysteines (C-Xn-C) to face the cytoplasmic proteins whose X-ray structures revealed the cytoplasm and Q9Y922 (ApVKORo) has its C-Xn-C motif presence of disulfide bonds, e.g. adenylosuccinate lyase facing the periplasm of E. coli. These results are consistent from Pyrobaculum aerophilium [14] and cystathionine b- with independent studies [26] that also reported the synthase from P. aerophilium [15]. Conclusive evidence inverted topology of ApVKOR proteins. We also showed came from cytoplasmic proteins called protein disulfide- that ApVKORi is able to interact with ApPDO and cyto- bond oxidoreductases (PDO), which were shown to catalyse plasmic E. coli DsbA (EcDsbA) to oxidize cytoplasmic alka- both the oxidation and isomerization of disulfide bonds in line phosphatase. Likewise ApVKORo interacts with vitro [16, 17]. periplasmic EcDsbA, but not ApPDO, to oxidize periplasmic flagellar protein as well as periplasmic b-galactosidase. In the E. coli cytoplasm there are no known enzymes that We also inverted ApVKORo by rational design to obtain a introduce disulfide bonds. Conversely, in the cytoplasm protein oriented to the cytoplasm by mimicking an evolu- there are two pathways, thioredoxin and glutaredoxin [18, tionary step that might have happened in the past. This 19], that maintain proteins in the reduced state when they inverted protein (ApVKORoflip) contains eight mutations. become oxidized as part of its catalytic state [6, 20, 21]. On Our studies suggest a mechanistic basis for the formation of the other hand, the formation of disulfide bonds is cata- disulfide bonds in the cytoplasm of Crenarchaea. lysed in the periplasm of E. coli and is performed by a set of Dsb proteins. DsbA is a thiol-disulfide oxidoreductase containing two cysteines separated by two amino acids RESULTS (CXXC motif) [2]. DsbB is a membrane protein that re- Bioinformatic analysis of crenarchaeal disulfide oxidizes DsbA to start a new catalytic cycle [22]. DsbB has bond-forming enzymes four catalytically active cysteines, the first pair embedded We have described the use of a bioinformatic approach, in the membrane being in a CXXC motif and interacting originally developed by Mallick et al. [12], to identify bac- with quinones [23]. The second pair of cysteines faces the teria with compartments in which disulfide bonds are periplasm and is a C-X -C motif known to interact with n formed [25]. This bioinformatic analysis is based on the DsbA [23, 24]. observation that most proteins in oxidizing compartments However, hyperthermophilic archaea have a bias for an have a significantly higher proportion of proteins with an even number of cysteines in their intracellular proteome, even number of cysteines than would be expected at which represents a greater abundance of disulfide-bonded random. proteins in this compartment [12]. This suggested that We applied this cysteine-counting approach to the archaea intracellular disulfide bonds are likely to be a result of selec- using the fraction of cysteine-containing exported or cyto- tive pressure for thermostable proteins [12]. plasmic proteins with even numbers of cysteine, as well as Using a bioinformatic analysis to determine a significant the fraction if cysteine was distributed according to the bias for even numbers of cysteine in proteins, together with random model in that set of proteins. These fractions were data on the presence of homologues of known disulfide used to calculate the z-score, which is the number of stan- bond-forming enzymes, our group has previously found dard deviations of the actual data from the mean obtained that Actinobacteria, Cyanobacteria and some members of with the random model. Values above 2.57 indicate that

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521865 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879 proteins have a significant bias for even numbers of cys- Secondly, we looked for the presence of homologues known teines, i.e. there is a probability of 0.01 % (99.99 % confi- to be involved in disulfide bond formation pathways in the dence) that the genome would have a z-score of >2.57 if cell envelope, i.e., DsbA, DsbB and VKOR (Fig. 1). We also the cysteines were distributed at random (Fig. 1 in green, included cytoplasmic protein PDO in the predictions since Table S1, available in the online version of this article). We this has been proposed to be a catalyst in the disulfide find that 22 of the 48 Crenarchaea in our dataset have a bond-formation pathway in the cytoplasm of thermophiles significantly higher frequency of cytoplasmic proteins with [16, 17]. Predicted cytoplasmic PDOs are found in most of an even number of cysteines than expected at random the Crenarchaea in our dataset (Table S1). A close homo- (Table S1). In addition, for 29 of these 48 Crenarchaea, the logue of DsbA is present in most Sulfolobales species and in predicted exported proteins also have a high frequency of Hyperthermus butylicus DSM5456. In contrast to bacterial proteins with an even number of cysteines (Fig. 1 in purple, DsbAs, the Sulfolobales homologues are predicted to be Table S1). This finding suggests that these organisms make cytoplasmic while those of H. butylicus are predicted to be disulfide bonds in proteins both in the cytoplasm and in exported by a single transmembrane helix near the N-termi- the extracellular compartment and therefore should have nus. None of the Crenarchaea have DsbB homologues, nor two distinct pathways for disulfide bond formation, one do they have homologues of PDO that are predicted to be localized to the cytoplasm and the other to the cell extracytoplasmic. In addition, VKOR homologues are found envelope. in 24 of these Crenarchaea (Fig. 1, Table S1). Eleven of these

Fig. 1. A ‘Tree of Life’ diagram showing representative archaeal organisms and the presence of DsbA, DsbB, VKORo, VKORi and PDO homologues as well as the cysteine fraction count in the periplasmic or cytoplasmic proteome. See Material and Methods and Table S1 for the complete dataset of organisms analysed in this study. The phylogenetic tree was generated using the Interactive Tree of Life (iTOL) web server v3.5.4 (http://itol.embl.de) by modifying the default tree [61].

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521866 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

24 species have two VKORs, 9 of which are predicted to oxidizing cytoplasm [7, 30]. We observed that fusions H66 make disulfide bonds in both cytoplasmic and extracyto- (strains DHB7748 and DHB7771) and A121 (strains plasmic compartments. DHB7769 and DHB7774) of ApVKORi show low levels of activity while L96 (strains DHB7749 and DHB7772) and Topcons-single and Phobius predict the membrane topol- G147 (strains DHB7750 and DHB7775) show high numbers ogy of the two A. pernix VKOR unambiguously. One is pre- in both strains (Fig. 2b, left). This indicates that the C-X -C dicted to have its active face oriented toward the outside of n motif of ApVKORi is oriented towards the cytoplasm. On the cell (VKORo) and the other toward the cytoplasm the other hand, fusions A55 (strains DHB7751 and (VKORi). While this project was ongoing, Hatahet and DHB7760), V60 (strains DHB7752 and DHB7761) and Ruddock [26] arrived to this conclusion by inverting E. coli G116 (strains DHB7753 and DHB7763) of ApVKORo show DsbB to face the cytoplasm. These authors demonstrated high activity while F89 (strains DHB7755 and DHB7762) that ApVKORo is able to interact with EcDsbA to oxidize and T146 (strains DHB7754 and DHB7764) show lower alkaline phosphatase in the E. coli periplasm. Similarly, activity (Fig. 2b, right), which indicates that ApVKORo has ApVKORi is able to reoxidize cytoplasmic EcDsbA, which its C-X -C motif facing the periplasm. in turn oxidizes cytoplasmic alkaline phosphatase [26]. n Thus, in this study we aimed to determine the membrane ApVKORi and ApVKORo are functional proteins in topology of these two proteins and their interactions with E. coli ApPDO. In order to determine the functionality of VKORs in cyto- ApVKORi is oriented to the cytoplasm and plasmic disulfide bond formation, we expressed them in an ApVKORo to the periplasm of E. coli E. coli strain that can be used to test for disulfide bond formation in the cytoplasm. This strain contains a plasmid In order to determine the topology of the two A. pernix pro- encoding a signal sequence-less alkaline phosphatase teins, we cloned the two VKORs under an IPTG-inducible (pAD495), which is localized to the cytoplasm. Since disul- promoter and expressed them in E. coli. We then assessed fide bond formation ordinarily does not occur in the cyto- their level of expression and likely membrane topology plasm, the alkaline phosphatase does not fold properly and using the alkaline phosphatase fusion method. This is enzymatically inactive. However, this enzyme can be acti- approach, widely used to determine membrane protein vated by altering the cytoplasm so that disulfide bonds form topology [27], is based on the properties of alkaline phos- in it. This activity can be detected by the ability of the active phatase which is active in its regular location, the periplasm alkaline phosphatase to replace other cytoplasmic phospha- of E. coli, but inactive when forced to remain in the cyto- tases such as serine-1-phosphatase (SerB). Thus, while a plasm by deletion of its signal sequence. Accordingly, alka- - strain lacking serB (DHB7787 strain) is auxotrophic for ser- line phosphatase is active when it is fused to a periplasmic ine, if alkaline phosphatase is active in the cytoplasm, it can domain of a membrane protein, but inactive when fused to replace the missing biosynthetic phosphatase and restore a cytoplasmic domain. This method has also been applied to prototrophy [5]. We expressed the two A. pernix VKORs M. tuberculosis VKOR, yielding results consistent with - into a serB strain in which the signal sequence-less alkaline physiological, genetic and structural studies on that protein phosphatase is also expressed, and looked for growth on [28]. We previously described the optimal approach for minimal medium lacking serine. using alkaline phosphatase fusions, in which the enzyme is fused to the carboxy-terminus of hydrophilic domains of Since the archaeal cytoplasmic oxidant ApPDO has been membrane proteins [29]. We thus fused alkaline phospha- proposed to be the direct catalyst of disulfide bond forma- tase to the ApVKORi-C’terminus of residues His66, Leu96, tion in the archaeal cytoplasm, it is a reasonable candidate Ala121 and Gly147, and to the ApVKORo-C’terminus of as a partner of ApVKORi. We generated a plasmid that residues Ala55, Val60, Phe89, Gly116 and Thr146. These expresses ApPDO (pDHB7833) and ApPDO with either constructs were then assayed for alkaline phosphatase activ- ApVKORi (pDHB7836) or ApVKORo (pDHB7837) from ity on XP plates and quantified by hydrolysis of PNPP the same promoter. We also constructed a similar plasmid substrate. that encodes either ApVKORi (pCL113) or ApVKORo (pCL114) with cytoplasmic EcDsbA (signal sequence-less) The results of alkaline phosphatase fusion to the two in an operon. VKORs yielded results that are consistent with the membrane topologies predicted by Topcons-single and Phobius, The expression of the two VKORs gave consistent results either qualitatively on XP plates (Fig. 2a) or quantitatively in regard to our results for membrane topology, and con- (Fig. 2b), and these changes are not due to lack of alkaline firmed that ApVKORi is oriented toward the cytoplasm phosphatase expression except for the F89 construct in and ApVKORo to the periplasm. ApVKORi restored strain DHB4, presumably because cytoplasmic alkaline growth on minimal medium to a serB- strain when the phosphatase is unstable and hence degraded (Fig. S1). We plasmid also expressed ApPDO or EcDsbA (Fig. 3a, strains measured alkaline phosphatase activity of the fusions using phoA+, DHB7854 and CL680, respectively) and growth two background strains, wild type (DHB4) and a double- was dependent on the expression of cytoplasmic alkaline mutant gor- trxB- (FA113) that lacks the reducing pathways phosphatase (Fig. 3a, strains phoA-, DHB7860 and CL690, required to stabilize alkaline phosphatase given its more respectively). In the absence of either ApPDO or EcDsbA

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521867 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

Fig. 2. ApVKORi is oriented to the cytoplasm and ApVKORo to the periplasm. (a) XP-minimal media plates showing the membrane topology of ApVKORi and ApVKORo using fusions with alkaline phosphatase (phoA) in E. coli strain DHB4. The amino acids shaded dark grey denote where alkaline phosphatase was fused. Catalytic cysteines are indicated by diamonds. Strains were streaked in M63 glucose minimal media with 50 µM XP and 400 µM IPTG to induce expression of the respective ApVKOR. Protter was used to visualize proteins according to Topcons prediction. (b) Alkaline phosphatase activity determined for each membrane fusion in two background strains, DHB4 (wild type) and FA113 (gor- trxB-). Amino acid positions indicate where alkaline phosphatase was fused. Data represent the average and standard deviation of two independent experiments.

there was no growth of the strain expressing only ApV- ApVKORi pair showed low levels (92 Miller units). In KORi, and the expression of only ApPDO did not restore contrast to these results, ApVKORo did not restore the growth to strain serB- (data not shown). We observed that growth of a serB- strain whether expressed with cyto- the pair ApVKORi–EcDsbA grew better on media lacking plasmic EcDsbA or ApPDO (Fig. S2c, strains DHB7855 serine than the ApPDO-ApVKORi pair, as indicated by and CL683, respectively). Consistent with its orientation colony size. We also confirmed these results by quantify- in the membrane, it would not be expected to restore ing alkaline phosphatase activity in these strains (Fig. 3a), growth. where the ApVKORi–EcDbsA pair showed high alkaline To assess the disulfide bond formation functionality of the phosphatase activity (629 Miller units) and the ApPDO– two A. pernix VKORs in extra-cytoplasmic proteins, we

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521868 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

Fig. 3. ApVKORi is functional in the cytoplasm of E. coli and ApVKORo in the periplasm of E. coli. (a) Cytoplasmic functionality of archaeal proteins was tested for complementation of E. coli DserB growth (either DssphoA+or DssphoA-) on minimal media lacking serine. Strains were serially diluted on M63 media with 0.2 % glucose and 1 mM IPTG to induce expression of indicated proteins either in the absence (upper) or presence (lower) of 50 µg mlÀ1 serine. phoA activity (Miller units) was determined in log-phase cultures grown in M63 media with 0.2 % glucose and 19 amino acids, excepting cysteine and methionine. Numbers represent the average ±SD of at least three independent experiments. (b) Periplasmic functionality was tested as complementation of motility in E. coli strains DdsbB and DdsbADdsbB. One colony of each strain was stabbed on 0.3 % agar M63 media with 0.2 % glucose and 400 µM IPTG to induce expression of indicated proteins. (c) Periplasmic functionality was tested as b-galactosidase activity of strains lacking dsbB and expressing b-Galdbs. Strains were streaked on M63 minimal media plates with 0.2 % glucose, 400 µM IPTG and 60 µg mlÀ1 X-Gal.

expressed these in two E. coli tester strains. We showed pre- (Fig. 3b, bottom). The second strain that we used to confirm viously that mycobacterial VKOR can promote extra-cyto- periplasmic functionality is an E. coli strain that expresses plasmic disulfide bond formation, replacing E. coli protein b-Galdbs [31]. This strain expresses b-galactosidase in the DsbB in the re-oxidization of DsbA [25]. This complemen- periplasm, rendering its native cysteines sensitive to the for- tation is readily detected by the restoration of bacterial mation of disulfide bonds. Hence, the lack of b-galactosi- motility, which is dependent on the disulfide-bonded pro- dase activity indicates the formation of disulfide bonds in tein FlgI. Thus, when we expressed ApVKORo in a dsbB- the periplasm whereas the presence of b-galactosidase activ- strain (DHB7867 strain) it restored motility (Fig. 3b, top), ity indicates the absence of the disulfide bond-forming sys- while ApVKORi (strain DHB7866) was unable to do so tem. We transformed the two VKOR homologues into the (Fig. S2a). Furthermore, this motility is dependent on dsbB- strain expressing b-Galdbs and then tested its activity EcDsbA and cannot be replaced by expression of ApPDO on X-Gal plates. We observed that the strain expressing

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521869 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

ApVKORo with the wild-type EcDsbA (strain DHB7874) in the phoA fusions of ApVKORo, fusion G116 of ApV- formed white colonies (Fig. 3c), indicating an active disul- KORoflip shows lower activity whereas fusion T146 shows fide bond-forming system, whereas ApVKORi with wild- higher activity than that observed for wild-type protein. type EcDsbA (strain DHB7873 strain) formed dark blue - To confirm these results we used a method analogous to colonies similar to strain dsbB and thus was unable to b restore disulfide bond formation in the periplasm (Fig. S2b). alkaline phosphatase fusion by making use of -lactamase b When the pair ApPDO–ApVKORo was co-expressed, the fusions to determine membrane topology [34]. -Lactamase strain formed pale blue colonies (Fig. 3c, strain DHB7878), folds in the periplasm and thereby becomes active [20]. The which may be explained by the levels of expression in this targets of b-lactamases are b-lactam antibiotics, such as construct since ApVKORo is expressed in an operon after ampicillin, and these are exported out of the cytoplasm into ApPDO, likely reducing the expression levels of ApVKORo. the cell envelope to degrade b-lactam antibiotics. When b- The same was observed in the size of the halo using this lactamase is fused to a periplasmic domain it confers resis- construct (Fig. 3b). tance to ampicillin. On the other hand, when it is fused to a cytoplasmic domain it remains sensitive to ampicillin. Thus, These results are again consistent with our results for topology can be analysed by finding which b-lactamase membrane topology and cytoplasmic functionality using - fusions to a membrane protein confer ampicillin resistance serB auxotrophy. ApVKORi can only function in oxidizing to strains expressing them. A low MIC indicates cyto- cytoplasmic proteins and ApVKORo can only oxidize periplasmic location of b-lactamase and a higher MIC indicates plasmic proteins in E.coli and, along with the topological periplasmic location. We observed that fusion G116 of analysis, this strongly suggests that these two proteins play ApVKORoflip shows low MIC whereas fusion T146 demon- roles in disulfide bond formation in two different compart- strates high MIC (Fig. 4c). These results are opposite to the ments. While some caution should be expressed, as these MICs observed for the wild-type protein, i.e. G116 fusion of studies were conducted in E. coli only, they are consistent ApVKORo shows high MIC and T146 shows lower MIC with a picture in which the two VKORs are both responsible (Fig. 4c), indicating that ApVKORoflip has an inverted for regenerating one of the active sulfhydryl oxidases, per- topology to the wild-type protein. haps PDO in the cytoplasm and an unidentified oxidase in the cell envelope. These latter proteins can then promote After confirming that the mutant ApVKORoflip shows an formation of disulfide bonds in numerous substrate proteins inverted topology, we then tested for complementation of in their respective compartments. growth on strain serB- (expressing DssphoA), combining it Mutations in ApVKORo lead to an inverted topology with either cytoplasmic EcDsbA or ApPDO. However, none of these combinations conferred growth on minimal media A striking property of the Crenarchaeal VKOR protein tree lacking serine. One possibility for the lack of phenotype was presented in Fig. 4(a) is that, in certain organisms, the two the difference in the interaction of the mutated VKOR with VKORs with cytoplasmic- and extra-cytoplasmic-facing the partner, which in the case of ApVKORo has not yet active sites do not constitute two separate branches. Rather been identified. it appears that one of the pairs of oppositely oriented VKOR species in each of the Crenarchaeal Orders represents two independent instances of gene duplication and divergence, DISCUSSION one in each Order. For instance, the closest homologue of ApVKORo is the inward-facing ApVKORi. Similarly Vul- To summarize, we have shown ApVKORi to be functional canisaeta and Caldivirga species VKORs appear to have in combination with either E. coli DsbA or A. pernix PDO diverged from a common ancestor at the Order or Family in oxidizing alkaline phosphatase in the E. coli cytoplasm, level [32]. and thereby to support the growth of a serine auxotroph without serine. In contrast, ApVKORo was not active in These observations raise the possibility that an inward- these tests. Additionally, ApVKORo, but not ApVKORi, has facing VKOR could have evolved from the outward-facing been shown to be capable of fulfilling the role of DsbB in VKOR, and this could have happened twice independently two tests of periplasmic disulfide bond formation, motility in the Crenarchaea. Therefore, this means that it is possible and oxidation of periplasmic b-galactosidase. Finally, to invert a VKOR facing the periplasm to the opposite side ApPDO has been shown to be functional as a disulfide oxi- of the membrane. We tested this possibility by mutating the dase in activating alkaline phosphatase in the E. coli protein ApVKORo, which faces the periplasm, to invert its cytoplasm. orientation to face the cytoplasm. In order to do this we changed the distribution of positively charged amino acids Our results suggest that multiple VKOR gene duplications that are important for topological orientation in the mem- within the Crenarchaea family, followed by topological brane [33]. Thus, we created eight mutations in ApVKORo divergence, account for their distribution in Crenarchaea (pDHB7786) shown in Fig. 4(b) and determined the topol- rather than independent inheritance of genes that diverged ogy by measuring alkaline phosphatase activity in two of the prior to their evolution. An example of topological diver- fusions of ApVKORoflip, 116 (strain DHB7815) and 146 gence within Proteobacteria has been noted previously (strain DHB7816) (Fig. 4b). Opposite to what we observed [35, 36].

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521870 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

Fig. 4. Mutating eight residues in wild-type ApVKORo to invert its orientation on the membrane. (a) Phylogenetic tree of VKOR proteins of Archaea members; circles denote the instances of gene duplication and divergence. i: inward-facing VKOR, o: outward-facing VKOR. See Methods, genomic analysis and construction of tree. (b) XP minimal media plate showing membrane topology of ApVKORoflip using fusions with alkaline phosphatase (phoA) in E. coli strain DHB4. Amino acids highlighted in dark grey show the eight mutations created to invert the wild-type protein. The amino acids shaded light grey denote where alkaline phosphatase was fused. Catalytic cysteines are indicated by diamonds. Strains were streaked in M63 glucose minimal media with 50 µM XP and 400 µM IPTG to induce expression of ApVKOR fusions. (c) Comparison of topologies between wild-type ApVKORo and inverted ApVKORoflip. The activity of alkaline phosphatase fusions was determined in strain DHB4. b-lactamase activity was measured as sensitivity to ampicillin (Amp). Num- bers represent the average of at least two independent experiments.

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521871 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

Our phylogenetic tree suggests that the evolution of the We can conceive of several explanations for this failure to second VKOR in some of the Crenarchaea occurred after detect potential VKORo substrates bioinformatically in separation of genera. When one protein evolves from these cases. 1) There could be DsbA-like proteins, exported another in bacteria, it appears often to be the case that the thioredoxins that were not detected in our analysis as second protein will be spread amongst bacteria. The fact exported. This mis-assignment of localization could be due that there are two archaeal species where the same evolu- to the presence of a novel type of export system that does tionary event for VKOR has taken place independently not use the classical signal sequences and can export a suggests either 1) greater separation of archaeal species and DsbA-like protein. Alternatively, there may be variation in possibly more limited lateral transfer than in bacteria and/ the properties of the archaeal signal sequences that still or 2) that it is relatively straightforward for a membrane allows interaction with the Sec export system, but which protein to evolve into its inverted form simply by changing interferes with proper prediction of localization of particular a few amino acids so as to affect the charge distribution proteins. 2) Even though the ApVKORo is active on around the transmembrane segments. Hatahet and Rud- EcDsbA, it is able to recognize a substrate that exhibits a dock indicated that the latter could be the case [26]. One fold different from that of the thioredoxin superfamily, con- could even imagine that an earlier version of the VKOR trary to our assumption. 3) Crenarchaeal VKORo proteins did not have an ideal distribution of charges and had a may have a diverse set of thioredoxin-like substrates that dual orientation in the membrane. Duplication and further are not all members of one subfamily, such as DsbA or selection for activity could have converted this to a VKOR PDO. dedicated to functioning in a single compartment. Mixed Another unresolved issue is that, although most Crenarch- topological variation of certain individual membrane pro- aea have PDO and some have a cytoplasmically oriented teins has been reported [36, 37]. VKORi, some do not have a preference for even numbers of While our results do not establish which are the direct sub- cysteines in their cytoplasmic proteins. This might suggest strates of the A. pernix VKORi, we present data that are that these organisms have the apparatus for making protein consistent with previous in vitro studies suggesting that disulfide bonds in a small subset of proteins, or do not make PDO could be the protein sulfhydryl oxidase in the cyto- proteins with disulfide bonds. However, as we discussed plasm. We found that ApVKORi, when expressed with previously, there are examples of bacteria (such as Chlamy- ApPDO, permits disulfide bond formation, although not diales) that do have DsbA and DsbB homologues but do not very efficiently as observed with cytoplasmic EcDsbA. This have a high bias for a even ratio of cysteines in their low efficiency could be due to factors that influence the exported proteins and yet still make disulfide bonds [25]. activity of either ApVKORi or ApPDO in the E. coli cyto- These examples can be explained in some cases by the small plasm. These factors include influence of the reductive number of substrate proteins requiring disulfide bonds. pathways in the E. coli cytoplasm on the redox state of the Thus, even though these organisms do make proteins with two proteins, lack of presence of the appropriate quinone disulfide bonds, the number of proteins is too small to affect for oxidation of ApVKORi, and limitations in the sub- the even:odd ratio and the overall odd:even cysteine ratio strates on which ApPDO can act. Alternatively, the low remains close to one. Hence, those Crenarchaea that have activity of oxidation of alkaline phosphatase could indicate analogous properties may also produce small numbers of that there is a more efficient oxidase than PDO present in disulfide-bonded proteins. A. pernix. Some of the organisms that contain genes for a cytoplasmic In the case of ApVKORo, some of the Crenarchaea have PDO do not have a cytoplasmic-facing VKORi (or DsbB) homologues of DsbA while many others, including A. per- homologue. We suggest that either there must be another nix, do not. We have found no exported thioredoxin-like partner for PDO in these organisms or that the PDO may protein that is generally conserved among Crenarchaea with be so susceptible to oxidation that it can function in the VKORo, which would be predicted candidates for substrates absence of an enzyme dedicated to that purpose. A DsbA of VKORo. Nevertheless, we consider it highly likely that with such properties has been proposed to function in disul- the substrates of VKORo in archaea will be thioredoxin fide bond formation in S. aureus, but this is an exported lipoprotein that can be re-oxidized by small-molecule oxi- family members. In E. coli the counterpart to VKOR, the – DsbB protein, has substrate specificity restricted to DsbA dants [40 42]. Together, our results suggest a possibly greater complexity and diversity of mechanism for disulfide and certain other thioredoxin family members. When bond formation in the Crenarchaea. expressed in E. coli, other bacterial extracytoplasmically oriented VKORs also oxidize periplasmic DsbA. Even the Even though the two VKORs are initiating disulfide bond human VKORc1 oxidizes thioredoxin family members in formation in two different compartments, they are presum- the endoplasmic reticulum [38, 39]. Furthermore, PDO, the ably both using quinones in order to be maintained in the likely substrate of VKORi, is a member of the thioredoxin oxidized state. If this is the case, then both are able to use superfamily. Thus, all available information on these pro- membrane-embedded quinones even though they exhibit teins indicates that both DsbBs and VKORs are restricted to opposite membrane topologies. This likely suggests the thioredoxin family members. hydrophobic nature of quinones, which allows them to

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521872 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879 freely move within the cytoplasmic membrane. Oxidation were constructed from the trimmed alignments with by quinones has been observed in other cytoplasmic-facing PhyML [50] or FastTree [51]. Trees were visualized with membrane proteins, such as ArcB [43]. either Dendroscope [52] or FigTree (http://tree.bio.ed.ac. uk/software/figtree/). We point out that related proteins with opposite topologies, as is the case with the Aeropyrum VKORs, pose a problem Analysis of ApVKORi and ApVKORo membrane for topology predictors like Polyphobius or Topcons (as topology opposed to Topcons Single) that use homology aming to collect a group of related proteins in anticipation of using Topological predictions were made with Topcons Single their sequences to produce a more accurate prediction for [53] (http://topcons.net), Phobius [54] (http://phobius.sbc. the query protein. That approach cannot be used in such su.se) and LipoP [55] (http://www.cbs.dtu.dk/services/ cases. LipoP/). We used the alkaline phosphatase fusion method as an in vivo approach to analyse topology [27]. Based on METHODS the bioinformatic predictions of transmembrane segments and hydrophilic domains, we fused DNA corresponding to Bacterial strains and growth conditions the gene encoding alkaline phosphatase lacking its signal The strains and plasmids used in this study are listed in sequence to sites in the protein that correspond to the pre- Table 1, and the primers in Table 2, respectively. All strains dicted carboxy-terminus of hydrophilic regions [29]. PhoA were grown at 37 C in NZ media or M63 minimal media fusions were constructed as described previously [28] using À supplemented with 0.2 % glucose, 50 µg ml 1 leucine and pDHB7764 and pDHB7666 as PCR templates and the ApV- À 50 µg ml 1 isoleucine for all DHB4 derivative strains. The KOR oligonucleotides listed in Table 2 as primers. PCR antibiotic concentrations used were: ampicillin 100 or 25 µg products were cloned into DHB7743, a derivative of À À ml 1 (chromosomal), kanamycin 40 µg ml 1, chloramphen- pDHB5747 in which the oligonucleotides phoA_dN- À – icol 10 µg ml 1 and tetracycline10 µg ml 1. co_trp_f and phoA_dNco_trp_r were used with Quick Change mutagenesis to remove an NcoI site on the phoA Genomic analysis gene. These fusions (Table 1, plasmids) were then integrated Genomic sequences for 349 non-redundant archaea were into the chromosome of E. coli strain DHB4 via the bacteri- downloaded on 28 January 2017 from NCBI. Counting of ophage lInCh, resulting in single-copy expression [56] and cysteines predicted to be in different subcellular locations P1 transduced to FA113 (gor- trxB-) [30]. and calculation of z-scores were done as described previously [25], except that in addition to a prediction for each protein Alkaline phosphatase activity was then determined in two using Phobius and LipoP as previously, we performed a sec- ways using these constructs. First, to qualitatively asses the ond prediction using Topcons Single and LipoP (references membrane topology of each protein, membrane fusions were streaked in M63 agar minimal media plates containing below). The even-cysteine frequencies, which were similar in À À 0.2 % glucose, 50 µg ml 1 leucine, 50 µg ml 1 Isoleucine, all cases, were averaged for calculation of the z-score. The À1 À1 cytoplasmic even-fraction z-score and the z-score for and the 100 µg ml carbenicillin, 50 µg ml XP (5-bromo-4-chloro exported segments of transmembrane proteins were used to indolyl phosphate, Sigma) and 400 µM IPTG (Isopropyl-b- assess the redox state of the two compartments. PFAM mod- D-1-thiogalactopyranoside, Enzo) to induce expression of els [44] and Hmmer2 [45] were used to search for DsbA, the membrane fusion. Topcons software was used to predict PF01323.12, DsbB, PF02600.8 and VKOR, PF07884.6. PDOs the topology and Protter [57] (http://wlab.ethz.ch/protter/ start/) was used to visualize the predicted topology. Plates were identified using a hidden Markov model constructed from an alignment of 23 Archaeal PDOs obtained by PSI- were incubated at 30 C for three days. Second, alkaline Blast [46, 47]. A family of eight proteins with membrane phosphatase activity was assayed as described before [58]. topology and cysteine topography similar to VKORs, but Briefly, an overnight culture was diluted 1 : 50 in M63 minimal media with 400 µM IPTG to induce the expression of that are not close homologues, have been included (WP_012185561.1, WP_013335825.1, WP_013604921.1, the membrane fusion. Cells were grown at 37 C with aera- – WP_075059700.1, WP_054844240.1, WP_054849719.1, tion to an OD600 of 0.2 0.5, after which 10 mM Iodoaceta- WP_066792976.1 and WP_069806451.1). These were identi- mide was added to the cultures to block reduced cysteines fied by collecting Psi-Blast homologues of a very weak and prevent air oxidation in PhoA. Cells were washed and VKOR homologue from the NCBI non-redundant protein re-suspended in 1M Tris, pH 8, with SDS and chloroform to data set. An HMM model was constructed from an align- lyse the cells. After centrifugation for 10 min at 13 000 r.p. ment of these and used to probe the downloaded archaeal m., samples were diluted 1 : 10 with 1M Tris, pH 8, and the genomes. All of these are predicted as inward facing, and are reactions were started by adding 20 µl of 0.4 % p-nitrophenyl phosphate (PNPP, Sigma). Samples were then incu- found in Thermoproteales with predicted oxidizing cytoplasms. bated at 37 C until the appearance of a yellow colour indicative of PNPP hydrolysis. The reaction was stopped by Construction of trees the addition of 200 µl of 1M K2HPO4. Finally, OD420 was Alignments of protein sequences were done with Muscle determined and Miller units were calculated as described [48]. Alignments were trimmed with Gblocks [49]. Trees before [58].

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521873 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

Table 1. List of strains and plasmids used in this study

Name Genotype Ref.

E. coli strains MC1000 F- araD139 D(ara leu)7697 DlacX74 galU galK rpsL [62] DHB4 MC1000 DphoA PvuII phoR DmalF3 F’ [lac+ (lacIQ) pro] [27] FA113 DHB4 gor522… miniTn10 (Tcr) trxB :: Km (Kmr) supp [7] HK295 MC1000 Dara714 leu+ [23] HK320 HK295 DdsbB [23] HK329 HK295 DdsbA DdsbB Kadokura H. HK325 HK295 DdsbB MalF-LacZ102 (Kanr) Kadokura H. DHB5059 DHB4 phoA+ pDHB60 (Ampr) This study DHB7802 FA113 phoA+ pDHB60 (Ampr) This study Alkaline phosphatase fusions DHB7800 DHB4 pDHB60 (Ampr) This study DHB7801 FA113 pDHB60 (Ampr) This study DHB7748 DHB4 pDHB7748 (ApVKORi-H66phoA fusion, Ampr) This study DHB7749 DHB4 pDHB7749 (ApVKORi-L96phoA fusion, Ampr) This study DHB7769 DHB4 pDHB7769 (ApVKORi-A121phoA fusion, Ampr) This study DHB7750 DHB4 pDHB7750 (ApVKORi-G147phoA fusion, Ampr) This study DHB7771 FA113 pDHB7748 (ApVKORi-H66phoA fusion, Ampr) This study DHB7772 FA113 pDHB7749 (ApVKORi-L96phoA fusion, Ampr) This study DHB7774 FA113pDHB7769 (ApVKORi-A121phoA fusion, Ampr) This study DHB7775 FA113pDHB7750 (ApVKORi-G147phoA fusion, Ampr) This study DHB7751 DHB4 pDHB7751 (ApVKORo-A55phoA fusion, Ampr) This study DHB7752 DHB4 pDHB7752 (ApVKORo-V60phoA fusion, Ampr) This study DHB7755 DHB4 pDHB7755 (ApVKORo-F89phoA fusion, Ampr) This study DHB7753 DHB4 pDHB7753 (ApVKORo-G116phoA fusion, Ampr) This study DHB7754 DHB4 pDHB7754 (ApVKORo-T146phoA fusion, Ampr) This study DHB7760 FA113 pDHB7751 (ApVKORo-A55phoA fusion, Ampr) This study DHB7761 FA113 pDHB7752 (ApVKORo-V60phoA fusion, Ampr) This study DHB7762 FA113 pDHB7755 (ApVKORo-F89phoA fusion, Ampr) This study DHB7763 FA113pDHB7753 (ApVKORo-G116phoA fusion, Ampr) This study DHB7764 FA113 pDHB7754 (ApVKORo-T146phoA fusion, Ampr) This study DHB7815 DHB4 pDHB7815 (ApVKORoflip-G116phoA fusion, Ampr) This study DHB7816 DHB4 pDHB7816 (ApVKORoflip-T146phoA fusion, Ampr) This study b-Lactamase fusions DHB7827 DHB4 pDHB7827 (ApVKORo-G116-bla fusion Kanr) This study DHB7828 DHB4 pDHB7828 (ApVKORo-T146-bla fusion Kanr) This study DHB7829 DHB4 pDHB7829 (ApVKORoflip-G116-bla fusion Kanr) This study DHB7830 DHB4 pDHB7830 (ApVKORoflip-T146-bla fusion Kanr) This study Serine-auxotrophs with DssphoA DHB7787 DHB4 D(deoD-serB)zjj::Tn10(Tcr) pAD495 (DssphoA, Cmr) This study DHB7838 DHB7787 pDHB7664 (ApVKORi, serB-, DssphoA, Cmr, Ampr) This study DHB7839 DHB7787 pDHB7666 (ApVKORo, serB-, DssphoA, Cmr, Ampr) This study DHB7840 DHB7787 pDHB7786 (ApVKORoflip, serB-, DssphoA, Cmr, Ampr) This study DHB7841 DHB7787 pMER77 (serB-, DssphoA, Cmr, Ampr) This study DHB7854 DHB7787 pDHB7836 (ApPDO, ApVKORi, serB-, DssphoA, Cmr, Ampr) This study DHB7855 DHB7787 pDHB7837 (ApPDO, ApVKORo, serB-, DssphoA, Cmr, Ampr) This study DHB7856 DHB7787 pDHB7833 (ApPDO, serB-, DssphoA, Cmr, Ampr) This study DHB7892 DHB7787 pDHB7884 (ApPDO, ApVKORoflip, serB-, DssphoA, Cmr, Ampr) This study CL680 DHB7787 pCL113 (ApVKORi, DssdsbA, Ampr) This study CL683 DHB7787 pCL114 (ApVKORo, DssdsbA, Ampr) This study CL666 DHB7787 pCL109 (ApVKORoflip, DssdsbA, Ampr) This study Serine-auxotrophs without DssphoA DHB7675 DHB4 D(deoD-serB)zjj :: Tn10(Tcr) This study

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521874 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

Table 1. cont.

Name Genotype Ref.

DHB7846 DHB7675 pDHB7664 (serB-, ApVKORi, Ampr) This study DHB7847 DHB7675 pDHB7666 (serB-, ApVKORo, Ampr) This study DHB7848 DHB7675 pDHB7786 (serB-, ApVKORoflip, Ampr) This study DHB7849 DHB7675 pMER77 (serB-, Ampr) This study DHB7860 DHB7675 pDHB7836 (ApPDO, ApVKORi, serB-, Ampr) This study DHB7861 DHB7675 pDHB7837 (ApPDO, ApVKORo, serB-, Ampr) This study DHB7862 DHB7675 pDHB7833 (ApPDO, serB-, Ampr) This study DHB7894 DHB7675 pDHB7884 (ApPDO, ApVKORoflip, serB-, Ampr) This study CL690 DHB7675 pCL113 (ApVKORi, DssdsbA, Ampr) This study Motility-test strains DHB7866 HK320 pDHB7664 (ApVKORi, dsbB-, Ampr) This study DHB7867 HK320 pDHB7666 (ApVKORo, dsbB-, Ampr) This study DHB7868 HK320 pDHB7786 (ApVKORoflip, dsbB-, Ampr) This study DHB7869 HK320 pMER77 (dsbB-, Ampr) This study DHB7870 HK320 pDHB7836 (ApPDO, ApVKORi, dsbB-, Ampr) This study DHB7871 HK320 pDHB7837 (ApPDO, ApVKORo, dsbB-, Ampr) This study DHB7872 HK320 pDHB7833 (ApPDO, dsbB-, Ampr) This study DHB7896 HK320 pDHB7884 (ApPDO, ApVKOR2flip, dsbB-, Ampr) This study DHB7910 HK329 pDHB7666 (ApVKORo, dsbA-, dsbB-, Ampr) This study DHB7911 HK329 pMER77 (dsbA-, dsbB-, Ampr) This study DHB7912 HK329 pDHB7672 (ApPDO, dsbA-, dsbB-, Ampr) This study DHB7913 HK329 pDHB7670 (ApPDO, ApVKORo, dsbA-, dsbB-, Ampr) This study b-Galactosidase-test strains DHB7873 HK325 pDHB7664 (ApVKORi, dsbB-, MalF-LacZ102, Kanr, Ampr) This study DHB7874 HK325 pDHB7666 (ApVKORo, dsbB-, MalF-LacZ102, Kanr, Ampr) This study DHB7875 HK325 pDHB7786 (ApVKORoflip, dsbB-, MalF-LacZ102, Kanr, Ampr) This study DHB7876 HK325 pMER77 (dsbB-, MalF-LacZ102, Kanr, Ampr) This study DHB7877 HK325 pDHB7836 (ApPDO, ApVKORi, dsbB-, MalF-LacZ102, Kanr, Ampr) This study DHB7878 HK325 pDHB7837 (ApPDO, ApVKORo, dsbB-, MalF-LacZ102, Kanr, Ampr) This study DHB7879 HK325 pDHB7833 (ApPDO, dsbB-, MalF-LacZ102, Kanr, Ampr) This study DHB7897 HK325 pDHB7884 (ApPDO, ApVKORoflip, dsbB-, MalF-LacZ102, Kanr, Ampr) This study Genomic DNA Aeropyrum pernix K1 [63] Plasmids r pDHB60 Derived from the PTAC expression vector pKK223-3 (Pharmacia Biotech, Inc.) with M13mp10 polylinker (Amp ) [56] pTrc99a Trc promoter, pBR322 origin (Ampr) pDSW204 Promoter down mutation in À35 of pTrc99a (Ampr) [64]

pMER77 pDSW204 C-term FLAG3 [65] pWP101 Vector for b-lactamase fusion (Kanr) [34] r pAD495 pACYC ori, pTac-phoAD2-22 (DssphoA, Cm ) Derman A. pDHB7664 pMER77-ApVKORi (Ampr) This study pDHB7666 pMER77-ApVKORo (Ampr) This study pDHB7836 pMER77-ApPDO and ApVKORi (Ampr) This study pDHB7837 pMER77-ApPDO and ApVKORo (Ampr) This study pDHB7833 pMER77-ApPDO (Ampr) This study pDHB7786 pTrc99a-ApVKORoflip (8 mutations, Ampr) This study pDHB7884 pMER77-ApPDO and ApVKORoflip (Ampr) This study r pCL113 pTrc99a-ApVKORi and EcdsbAD2-19 (DssdsbA, Amp ) This study r pCL114 pTrc99a-ApVKORo and EcdsbAD2-19 (DssdsbA, Amp ) This study r pCL109 pTrc99a-ApVKORoflip and EcdsbAD2-19 (DssdsbA, Amp ) This study Plasmids with alkaline phosphatase fusions pDHB7748 pDHB60 ApVKORi-H66phoA fusion, (Ampr) This study pDHB7749 pDHB60 ApVKORi-L96phoA fusion (Ampr) This study

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521875 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

Table 1. cont.

Name Genotype Ref.

pDHB7769 pDHB60 ApVKORi-A121phoA fusion (Ampr) This study pDHB7750 pDHB60 ApVKORi-G147phoA fusion (Ampr) This study pDHB7751 pDHB60 ApVKORo-A55phoA fusion, (Ampr) This study pDHB7752 pDHB60 ApVKORo-V60phoA fusion, (Ampr) This study pDHB7755 pDHB60 ApVKORo-F89phoA fusion, (Ampr) This study pDHB7753 pDHB60 ApVKORo-G116phoA fusion, (Ampr) This study pDHB7754 pDHB60 ApVKORo-T146phoA fusion, (Ampr) This study pDHB7815 pDHB60-ApVKORoflip-G116phoA fusion, (Ampr) This study pDHB7816 pDHB60-ApVKORoflip-T146phoA fusion, (Ampr) This study Plasmids with b-lactamase fusions pDHB7827 pWP101-ApVKORo-G116-b-lactamase fusion, (Kanr) This study pDHB7828 pWP101-ApVKORo-T146-b-lactamase fusion, (Kanr) This study pDHB7829 pWP101-ApVKORoflip-G116-bÀlactamase fusion, (Kanr) This study pDHB7830 pWP101-ApVKORoflip-T146-bÀlactamase fusion, (Kanr) This study

Immunoblot analysis primers Cl245–246. Then, ApVKORi or ApVKORo was – – E. coli cells with different plasmids were grown overnight in inserted at NcoI-KpnI using primers Cl254 255 or Cl247 NZ-rich medium supplemented with ampicillin. Cultures 256, respectively. In this way, EcDsbA was expressed from were diluted 1 : 100 in fresh medium with 1 mM IPTG. Pro- the same Trc promoter as ApVKORs and translated from teins were precipitated with trichloroacetic acid (TCA) from the SD sequence added to Cl245 primer. All plasmids were mid-log cultures and proteins were pelleted and washed verified by sequencing. with acetone. The protocol of Guilhot et al. [59] was fol- Assays for functioning of VKORs in disulfide bond lowed to prepare the samples for standard Western blot formation analysis. SDS-PAGE was then performed. Proteins were Motility electrotransferred to a nitrocellulose membrane in 25 mM Constructs were transformed into E. coli strains DdsbB Tris base/192 mM glycine/20 % methanol. Membranes were (HK320) and DdsbADdsbB (HK329). One colony of each immunoblotted with anti-DsbA or anti-His rabbit serum. strain was stabbed on 0.3 % agar M63 media with 0.2 % glu- Immunodetection was performed according to the ECL cose and 400 µM IPTG to induce expression. protocol (Amersham) using streptavidin-horseradish peroxidase. b-Galactosidase activity To perform a more stringent test where more disulfide Cloning of ApVKORi, ApVKORo, ApPDO and EcDsbA bond formation is needed, the same constructs were trans- dbs Plasmids expressing ApVKORi (pDHB7664), ApVKORo formed into a DdsbB strain expressing b-gal fusion (pDHB7666) and ApPDO (pDHB7833) were cloned (HK325). Strains were streaked on M63 minimal media À1 between NcoI-KpnI of pMER77. First, PCR products ampli- plates with 0.2 % glucose, 400 µM IPTG and 60 µg ml fied from A. pernix chromosomal DNA were cloned into X-Gal. pMER77 (FLAG-tag sequence of pMER77 was not Serine auxotrophy expressed in these constructs) using primers ApVKORi_f, Cytoplasmic alkaline phosphatase, when active, is able to ApVKORo_f, Nco_ApPDO_f and Nco_SDApPD_r. substitute for serine-1-phosphate phosphatase (the last Plasmids expressing bicistronic ApPDO-ApVKORi enzyme of serine biosynthesis) [5]. Thus, plasmids were (pDHB7836) and ApPDO-ApVKORo (pDHB7837) were transformed into a strain lacking serB and expressing cyto- constructed amplifying ApPDO from pDHB7833 using pri- plasmic alkaline phosphatase (strain DHB778). Growth was tested on M63 0.2 % glucose minimal media supplemented mers Nco_ApPDO_f and Nco_SDApPD_r and inserting À À with 50 µg ml 1 leucine and 50 µg ml 1 isoleucine either in this fragment at the NcoI site of plasmids pDHB7664 and À1 pDHB7666, respectively. Correct gene orientation was con- the presence or absence of serine (50 µg ml ). firmed by sequencing. In this way ApVKORs were Mutagenesis of ApVKORo to invert the protein expressed from the same trc promoter as ApPDO and 204 Macro-priming was used to construct the mutations within translated with the Shine–Dalgarno (SD) sequence added to ApVKORo. The primers used are listed in Table 2. Primers Nco_SDApPD_r primer. P1 (ApVKORo_f) and P2 (pTrc_r) were used to amplify the Plasmids expressing ApVKORi-EcDsbA (pCL113) and entire region between restriction sites NcoI and XmaI. ApVKORo-EcDsbA (pCL114) were constructed by insertion M1 (E110K_L111K_r, R83S_R84S_r, V27K_S28K_r, first at BamHI-XbaI sites EcdsbAD2–19 into pTrc99a using V60K_H61K_r) and M2 (E110K_L111K_f, R83S_R84S_f,

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521876 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

Table 2. List of primers used in this study

Primer ID Sequence Restriction site

ApVKORi_f CCCCCCATGGTTGAGGCGAGACTGC NcoI ApVKORo_f CCCCCCATGGGGTCGAGGCTTTCGCATG NcoI ApVKORoflip_G116_r CCCCTCCGGACCAGCGACGAACTTTTTTAG ApVKORi_A121_r CCCCTCCGGAGCCTTGGCAACCCTCACCTCAAGG ApVKORo_G116_r CCCCTCCGGACCAGCGACGAACAGTTCTAG ApVKORo_T146_r CCCCTCCGGAGTGTGGCCACCAATCCGC ApVKORo_E110K_L111K_f GTTCCCTACCTGGTTTATCTAAAAAAGTTCGTCGCTGGTGCTGTCTGC ApVKORo_E110K_L111K_r GCAGACAGCACCAGCGACGAACTTTTTTAGATAAACCAGGTAGGGAAC ApVKORo_R83S_R84S_f GCCCTGGCTTATTGGATTAGCTCGTCTCGCATCTTTCTAATC ApVKORo_R83S_R84S_r GATTAGAAAGATGCGAGACGAGCTAATCCAATAAGCCAGGGC ApVKORo_V27K_S28K_f ATTGCTGGCTATTTCGAGTACAAGAAGGGCTCTGGGGTTTGTGAGATT ApVKORo_V27K_S28K_r AATCTCACAAACCCCAGAGCCCTTCTTGTACTCGAAATAGCCAGCAAT ApVKORo_V60K_H61K_f GAGGCGGTTGTATTTGGCGTTAAGAAGCTGAGTGTTTTAGCGCCAGTA ApVKORo_V60K_H61K_r TACTGGCGCTAAAACACTCAGCTTCTTAACGCCAAATACAACCGCCTC AtoBlaNhe_r AGGACCTAGGTTGTGTATAAGAGTCCGG AtoBlaSpe_r AGGAACTAGTTTGTGTATAAGAGTCCGG MluMalF_f TAAGACGCGTCAGCGTCGCATCAGG Nco_ApPDO_f CCCCCCATGGCTAGGTATTACGTGC NcoI Nco_SDApPD_r ACCCCATGGGTCTGTTTCCTTTATAATC NcoI pTrc_f CAAGGCGCACTCCCGTTCTGG pTrc_r GTTCTGATTTAATCTGTATCAGGC TnPhoA_long_r AATATCGCCCTGAGCAGCCC Cl254 ACAGACCATGGGGGTTGAGGCGAGACTGCTAGA NcoI Cl255 CGGGGTACCTTAACCTCCCAGGATCTTTGCGGT KpnI Cl247 ACAGACCATGGGGTCGAGGCTTTCGCATGC NcoI Cl256 CGGGGTACCCTAGGTGTGGCCACCAATCC KpnI Cl245 CCGGGGATCCAGGAGATATAACATGGCGCAGTATGAAGATGGT BamHI Cl246 AGTCTCTAGATTATTTTTTCTCGGACAGATATTTCACTGTATCAGC XbaI

V27K_S28K_f, V60K_H61K_f) primers overlap but are opposite strands. Two PCR reactions were run, one combin- Funding information ing P1 and M1 and the other with P2 and M2. These two This work was supported by grants to J. B. from the National Institute products that overlap were used as templates for the final of General Medical Sciences, GMO41883 and a pilot grant from Har- vard Catalyst, the Harvard Clinical and Translation Science Center (NIH reaction with primers P1 and P2. PCRs were purified, cut grant 1 UL1 RR 025758-01 and financial contributions from participat- with NcoI and XmaI restriction enzymes (New England ing institutions). S. H. was supported by a scholarship from the Univer- Biolabs, USA) and cloned into a restricted pDHB60 plas- sity of Amsterdam. J. B. is an American Cancer Society Professor. mid. All mutations were then combined by PCR. Conflicts of interest Membrane topology of ApVKORoflip The authors declare that there are no conflicts of interest. To determine membrane topology, fusions phoA G116 and T146 were made to the ApVKORoflip. Fusions were made References using primers ApVKORo_f with either ApVKORo_T146_r 1. Albers SV, Meyer BH. The archaeal cell envelope. Nat Rev Microbiol 2011;9:414–426. or ApVKORo_flip_G116_r. Constructs were made as 2. Bardwell JC, McGovern K, Beckwith J. Identification of a protein described above. Plasmids were verified by sequencing and required for disulfide bond formation in vivo. Cell 1991;67:581– then transformed into DHB4 (wild type) to measure alka- 589. line phosphatase activity. 3. Frand AR, Kaiser CA. The ERO1 gene of yeast is required for oxidation of protein dithiols in the endoplasmic reticulum. Mol Cell b-Lactamase fusions and activity 1998;1:161–170. Alkaline phosphatase fusions (G116 and T146) were con- 4. Tu BP, Ho-Schleyer SC, Travers KJ, Weissman JS. Biochemical basis of oxidative protein folding in the endoplasmic reticulum. verted to b-lactamase fusions as described previously [34] Science 2000;290:1571–1574. using primers MluMalF_f, AtoBlaNhe_r and AtoBlaSpe_r. 5. Derman AI, Prinz WA, Belin D, Beckwith J. Mutations that allow The MIC of ampicillin necessary to kill strains expressing disulfide bond formation in the cytoplasm of Escherichia coli. the fusions was determined as previously described [60]. Science 1993;262:1744–1747.

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521877 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

6. Stewart EJ, Aslund F, Beckwith J. Disulfide bond formation in the 26. Hatahet F, Ruddock LW. Topological plasticity of enzymes Escherichia coli cytoplasm: an in vivo role reversal for the thiore- involved in disulfide bond formation allows catalysis in either the doxins. Embo J 1998;17:5543–5550. periplasm or the cytoplasm. J Mol Biol 2013;425:3268–3276. 7. Bessette PH, Aslund F, Beckwith J, Georgiou G. Efficient folding 27. Manoil C, Beckwith J. A genetic approach to analyzing membrane of proteins with multiple disulfide bonds in the Escherichia coli protein topology. Science 1986;233:1403–1408. cytoplasm. Proc Natl Acad Sci USA 1999;96:13703–13708. 28. Wang X, Dutton RJ, Beckwith J, Boyd D. Membrane topology and 8. Hatahet F, Nguyen VD, Salo KE, Ruddock LW. Disruption of reduc- mutational analysis of Mycobacterium tuberculosis VKOR, a protein ing pathways is not essential for efficient disulfide bond formation involved in disulfide bond formation and a homologue of human in the cytoplasm of E. coli. Microb Cell Fact 2010;9:67. vitamin K epoxide reductase. Antioxid Redox Signal 2011;14:1413– 9. Lobstein J, Emrich CA, Jeans C, Faulkner M, Riggs P et al. SHuf- 1420. fle, a novel Escherichia coli protein expression strain capable of 29. Boyd D, Traxler B, Beckwith J. Analysis of the topology of a mem- correctly folding disulfide bonded proteins in its cytoplasm. Microb Cell Fact 2012;11:1. brane protein by using a minimum number of alkaline phosphatase fusions. J Bacteriol 1993;175:553–556. 10. White CL, Senkevich TG, Moss B. Vaccinia virus G4L glutaredoxin is an essential intermediate of a cytoplasmic disulfide bond path- 30. Michaelis S, Inouye H, Oliver D, Beckwith J. Mutations that alter way required for virion assembly. J Virol 2002;76:467–472. the signal sequence of alkaline phosphatase in Escherichia coli. 1983;154:366–374. 11. Ren B, Tibbelin G, de Pascale D, Rossi M, Bartolucci S et al. A protein disulfide oxidoreductase from the archaeon Pyrococcus 31. Froshauer S, Green GN, Boyd D, McGovern K, Beckwith J. Genetic furiosus contains two thioredoxin fold units. Nat Struct Biol 1998;5: analysis of the membrane insertion and topology of MalF, a cyto- 602–611. plasmic membrane protein of Escherichia coli. J Mol Biol 1988;200: 501–511. 12. Mallick P, Boutz DR, Eisenberg D, Yeates TO. Genomic evidence that the intracellular proteins of archaeal microbes contain disul- 32. Hatahet F, Boyd D, Beckwith J. Disulfide bond formation in pro- fide bonds. Proc Natl Acad Sci USA 2002;99:9679–9684. karyotes: history, diversity and design. Biochim Biophys Acta 2014; 1844:1402–1414. 13. Beeby M, O’Connor BD, Ryttersgaard C, Boutz DR, Perry LJ et al. The genomics of disulfide bonding and protein stabilization in 33. Boyd D, Beckwith J. The role of charged amino acids in the locali- thermophiles. PLoS Biol 2005;3:e309. zation of secreted and membrane proteins. Cell 1990;62:1031– 1033. 14. Toth EA, Worby C, Dixon JE, Goedken ER, Marqusee S et al. The crystal structure of adenylosuccinate lyase from Pyrobaculum 34. Prinz WA, Beckwith J. Gene fusion analysis of membrane protein aerophilum reveals an intracellular protein with three disulfide topology: a direct comparison of alkaline phosphatase and b-lac- bonds. J Mol Biol 2000;301:433–450. tamase fusions. J Bacteriol 1994;176:6410–6413. 15. King NP, Lee TM, Sawaya MR, Cascio D, Yeates TO. Structures 35. Sa€af€ A, Johansson M, Wallin E, von Heijne G. Divergent evolution and functional implications of an AMP-binding cystathionine b- of membrane protein topology: the Escherichia coli RnfA and RnfE synthase domain protein from a hyperthermophilic archaeon. homologues. Proc Natl Acad Sci USA 1999;96:8540–8544. J Mol Biol 2008;380:181–192. 36. Rapp M, Granseth E, Seppal€ a€ S, von Heijne G. Identification and 16. D’Ambrosio K, Pedone E, Langella E, De Simone G, Rossi M et al. evolution of dual-topology membrane proteins. Nat Struct Mol Biol A novel member of the protein disulfide oxidoreductase family 2006;13:112–116. from Aeropyrum pernix K1: structure, function and electrostatics. 37. Chen YJ, Pornillos O, Lieu S, Ma C, Chen AP et al. X-ray structure J Mol Biol 2006;362:743–752. of EmrE supports dual topology model. Proc Natl Acad Sci USA 17. Pedone E, Limauro D, D’Alterio R, Rossi M, Bartolucci S. Charac- 2007;104:18999–19004. terization of a multifunctional protein disulfide oxidoreductase 38. Schulman S, Wang B, Li W, Rapoport TA. Vitamin K epoxide from Sulfolobus solfataricus. Febs J 2006;273:5407–5420. reductase prefers ER membrane-anchored thioredoxin-like redox 18. Aberg A, Hahne S, Karlsson M, Larsson A, Ormö M et al. Evidence partners. Proc Natl Acad Sci USA 2010;107:15027–15032. for two different classes of redox-active cysteines in ribonucleo- 39. Hatahet F, Blazyk JL, Martineau E, Mandela E, Zhao Y et al. tide reductase of Escherichia coli. J Biol Chem 1989;264:12249– 12252. Altered Escherichia coli membrane protein assembly machinery allows proper membrane assembly of eukaryotic protein vitamin 19. Zheng M, Aslund F, Storz G. Activation of the OxyR transcription K epoxide reductase. Proc Natl Acad Sci USA 2015;112:15184– factor by reversible disulfide bond formation. Science 1998;279: 1718–1722. 15189. 40. Dumoulin A, Grauschopf U, Bischoff M, Thöny-Meyer L, Berger- 20. Pollitt S, Zalkin H. Role of primary structure and disulfide bond € formation in beta-lactamase secretion. J Bacteriol 1983;153:27– Bachi B. Staphylococcus aureus DsbA is a membrane-bound lipo- 32. protein with thiol-disulfide oxidoreductase activity. Arch Microbiol 2005;184:117–128. 21. Derman AI, Beckwith J. Escherichia coli alkaline phosphatase fails to acquire disulfide bonds when retained in the cytoplasm. 41. Kouwen TR, van der Goot A, Dorenbos R, Winter T, Antelmann H J Bacteriol 1991;173:7719–7722. et al. Thiol-disulphide oxidoreductase modules in the low-GC Gram-positive bacteria. Mol Microbiol 2007;64:984–999. 22. Bardwell JC, Lee JO, Jander G, Martin N, Belin D et al. A pathway for disulfide bond formation in vivo. Proc Natl Acad Sci USA 1993; 42. Heras B, Kurz M, Jarrott R, Shouldice SR, Frei P et al. Staphylo- 90:1038–1042. coccus aureus DsbA does not have a destabilizing disulfide. A new paradigm for bacterial oxidative folding. J Biol Chem 2008;283: 23. Kadokura H, Beckwith J. Four cysteines of the membrane protein 4261–4271. DsbB act in concert to oxidize its substrate DsbA. Embo J 2002;21: 2354–2363. 43. Alvarez AF, Rodriguez C, Georgellis D. Ubiquinone and menaquinone 24. Inaba K, Takahashi YH, Fujieda N, Kano K, Miyoshi H et al. DsbB electron carriers represent the yin and yang in the redox regulation elicits a red-shift of bound ubiquinone during the catalysis of of the ArcB sensor kinase. J Bacteriol 2013;195:3054–3061. DsbA oxidation. J Biol Chem 2004;279:6761–6768. 44. Sonnhammer EL, Eddy SR, Durbin R. Pfam: a comprehensive 25. Dutton RJ, Boyd D, Berkmen M, Beckwith J. Bacterial species database of protein domain families based on seed alignments. exhibit diversity in their mechanisms and capacity for protein Proteins 1997;28:405–420. disulfide bond formation. Proc Natl Acad Sci USA 2008;105:11933– 45. Eddy SR. Hidden Markov models. Curr Opin Struct Biol 1996;6: 11938. 361–365.

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521878 On: Tue, 19 Dec 2017 17:28:39 Hibender et al., Microbiology 2017;163:1864–1879

46. Altschul SF, Madden TL, Schaffer€ AA, Zhang J, Zhang Z et al. 57. Omasits U, Ahrens CH, Müller S, Wollscheid B. Protter: interactive Gapped BLAST and PSI-BLAST: a new generation of protein data- protein feature visualization and integration with experimental base search programs. Nucleic Acids Res 1997;25:3389–3402. proteomic data. Bioinformatics 2014;30:884–886. 47. Altschul SF, Gertz EM, Agarwala R, Schaffer€ AA, Yu YK. PSI- 58. Brickman E, Beckwith J. Analysis of the regulation of Escherichia BLAST pseudocounts and the minimum description length princi- coli alkaline phosphatase synthesis using deletions and phi80 ple. Nucleic Acids Res 2009;37:815–824. transducing phages. J Mol Biol 1975;96:307–316. 48. Edgar RC. MUSCLE: multiple sequence alignment with high accu- 59. Guilhot C, Jander G, Martin NL, Beckwith J. Evidence that the racy and high throughput. Nucleic Acids Res 2004;32:1792–1797. pathway of disulfide bond formation in Escherichia coli involves interactions between the cysteines of DsbB and DsbA. Proc Natl 49. Talavera G, Castresana J. Improvement of phylogenies after Acad Sci USA 1995;92:9895–9899. removing divergent and ambiguously aligned blocks from protein sequence alignments. Syst Biol 2007;56:564–577. 60. Broome-Smith JK, Spratt BG. A vector for the construction of translational fusions to TEM b -lactamase and the analysis of pro- 50. Guindon S, Gascuel O. A simple, fast, and accurate algorithm to tein export signals and membrane protein topology. Gene 1986;49: estimate large phylogenies by maximum likelihood. Syst Biol 341–349. 2003;52:696–704. 61. Ciccarelli FD, Doerks T, von Mering C, Creevey CJ, Snel B et al. 51. Price MN, Dehal PS, Arkin AP. FastTree 2–approximately maxi- Toward automatic reconstruction of a highly resolved tree of life. mum-likelihood trees for large alignments. PLoS One 2010;5: Science 2006;311:1283–1287. e9490. 62. Casadaban MJ, Chou J, Cohen SN. In vitro gene fusions that join 52. Huson DH, Richter DC, Rausch C, Dezulian T, Franz M et al. Den- an enzymatically active beta-galactosidase segment to amino-ter- droscope: an interactive viewer for large phylogenetic trees. BMC minal fragments of exogenous proteins: Escherichia coli plasmid Bioinformatics 2007;8:460. vectors for the detection and cloning of translational initiation sig- 53. Bernsel A, Viklund H, Hennerdal A, Elofsson A. TOPCONS: con- nals. J Bacteriol 1980;143:971–980. sensus prediction of membrane protein topology. Nucleic Acids 63. Kawarabayasi Y, Hino Y, Horikawa H, Yamazaki S, Haikawa Y et al. Res 2009;37:W465–W468. Complete genome sequence of an aerobic hyper-thermophilic cren- 54. Kall€ L, Krogh A, Sonnhammer EL. A combined transmembrane archaeon, Aeropyrum pernix K1. DNA Res 1999;6:83–101. topology and signal peptide prediction method. J Mol Biol 2004; 64. Weiss DS, Chen JC, Ghigo JM, Boyd D, Beckwith J. Localization of 338:1027–1036. FtsI (PBP3) to the septal ring requires its membrane anchor, the Z 55. Rahman O, Cummings SP, Harrington DJ, Sutcliffe IC. Methods ring, FtsA, FtsQ, and FtsL. J Bacteriol 1999;181:508–520. for the bioinformatic identification of bacterial lipoproteins 65. Hemmis CW, Berkmen M, Eser M, Schildbach JF. TrbB from con- encoded in the genomes of Gram-positive bacteria. World J jugative plasmid F is a structurally distinct disulfide isomerase Microbiol Biotechnol 2008;24:2377–2382. that requires DsbD for redox state maintenance. J Bacteriol 2011; 56. Boyd D, Weiss DS, Chen JC, Beckwith J. Towards single-copy 193:4588–4597. gene expression systems making gene cloning physiologically rel- evant: lambda InCh, a simple Escherichia coli plasmid-chromosome shuttle system. J Bacteriol 2000;182:842–847. Edited by: S-V. Albers and F. Sargent

Five reasons to publish your next article with a Microbiology Society journal 1. The Microbiology Society is a not-for-profit organization. 2. We offer fast and rigorous peer review – average time to first decision is 4–6 weeks. 3. Our journals have a global readership with subscriptions held in research institutions around the world. 4. 80% of our authors rate our submission process as ‘excellent’ or ‘very good’. 5. Your article will be published on an interactive journal platform with advanced metrics.

Find out more and submit your article at microbiologyresearch.org.

Downloaded from www.microbiologyresearch.org by IP: 128.103.149.521879 On: Tue, 19 Dec 2017 17:28:39