Emerging Themes in SecA2 Protein Export

Abstract

The conserved general secretion (Sec) pathway carries out most protein export in bacteria and is powered by the essential SecA ATPase. Interestingly, mycobacteria and some Gram positive bacteria possess two SecA proteins: SecA1 and SecA2. In these species, SecA1 is responsible for exporting the majority of proteins whereas SecA2 exports only a subset of substrates and is implicated in virulence. However, despite the impressive body of knowledge on the canonical SecA (SecA1), less is known concerning SecA2 function. Here, we review our current understanding of the different types of SecA2 systems and outline future directions for SecA2 studies.

Introduction

All bacteria have systems for exporting specific proteins out of the cytoplasm and into the cell envelope or extracellular environment, where they have roles in cell wall synthesis, nutrient acquisition, and other vital physiological processes. In bacterial pathogens, exported proteins also play critical roles in virulence. Most exported proteins are translocated across the cytoplasmic membrane by the conserved general secretion (Sec) pathway¹, which is present in all bacteria and is essential for viability.

The Sec pathway translocates unfolded proteins through a protein complex in the cytoplasmic membrane comprised of SecY, SecE, and SecG proteins. SecY forms the channel through which unfolded proteins pass². SecE is thought to stabilize the open SecY conformation necessary for translocation^3–6 while SecG increases export efficiency^{7, 8}. The SecYEG channel is used in two types of Sec export: post-translational and co-translational. In post-translational Sec export, which is the focus of this review, proteins translocate completely across the cytoplasmic membrane through SecYEG. In co-translational export, SecYEG works with the signal recognition particle (SRP) to insert integral membrane proteins into the cytoplasmic membrane⁹ or, in some cases, translocate proteins across the cytoplasmic membrane¹⁰. For integral membrane proteins, SRP recognizes transmembrane domains as they emerge from the ribosome during translation and targets them as ribosome-mRNA-nascent protein complexes to FtsY for delivery to SecYEG¹¹. A lateral gate in SecY is thought to then open and allow passage of transmembrane domains into the membrane¹².

In post-translational Sec export, the proteins destined for translocation across the cytoplasmic membrane are synthesized as preproteins that are distinguished from the larger pool of cytoplasmic proteins by the presence of N-terminal Sec signal peptides. Sec signal peptides have a positively charged N-terminus, hydrophobic core, and polar C-terminus containing a signal peptidase cleavage site¹³. In addition to the signal peptide, another requirement for Sec export is that the mature portion of the preprotein remains unfolded for competent passage through SecY. Some proteins are recognized and kept unfolded by cytoplasmic chaperones, such as SecB^{14, 15}, although other preproteins may be unfolded as they are translocated^{16, 17}.

A central component of the post-translational Sec pathway is the cytosolic SecA motor protein¹⁸, which has a vital role in targeting and powering preprotein transport through SecYEG^{19, 20}. Since the discovery of SecA in 1981²¹, Sec export has been the focus of extensive study. A combination of genetic, structural and biochemical studies including in vitro reconstitution have led to a relatively sophisticated understanding of the Sec pathway and SecA function in particular (Fig. 1a). SecA binds preproteins along a cleft that includes the SecA preprotein-crosslinking (PPXD) domain^22–26, and targets them to the SecYEG complex in the membrane through interactions with SecY^{27, 28} and membrane phospholipids²⁹. After delivery to SecYEG, the signal peptide of the preprotein inserts into SecY to stabilize an open SecY channel conformation^30–32. SecA then undergoes cycles of conformational changes coupled to ATP-binding and hydrolysis to drive preproteins through the SecY channel^{20, 33}. Several models have been proposed to explain how SecA powers preprotein insertion through SecY¹⁸. Nonetheless, most models propose that portions of SecA, including the IRA-1 (intramolecular regulator of ATP hydrolysis 1) two-helix finger, insert into SecY to promote forward preprotein motion through the channel. During or immediately following export, the signal peptide is removed by signal peptidases³⁴ and the mature domain of the protein adopts a folded conformation.

Figure 1. Models of Sec export.

a) Post-translational Sec export is powered by the essential SecA ATPase. SecA can be divided into two main structural domains: a motor domain that drives ATP hydrolysis and a specificity domain that interacts with the preprotein destined for export. Step 1, preproteins synthesized with N-terminal Sec signal peptides (hatched) are bound by cytoplasmic SecA along a cleft between the two domains. Cytoplasmic chaperones, such as SecB, aid in keeping some preproteins unfolded prior to export and can directly deliver these preproteins to SecA. Step 2, SecA delivers the preprotein to a membrane-spanning complex composed of SecY, SecE, and SecG. Here, the signal peptide inserts into SecY to help keep an open channel conformation. Step 3, SecA goes through rounds of conformational changes coupled to ATP hydrolysis to promote forward movement of the unfolded preprotein through the SecY channel. During these cycles, it is proposed that SecA inserts into the SecY channel, specifically by the IRA-1 two-helix finger (two helices). Step 4, during or shortly after translocation the signal peptide is removed by periplasmic signal peptidases (SP) and the protein then adopts its mature, folded conformation. The SecD, SecF, and YajC transmembrane proteins contribute to the efficiency of Sec export but are not shown. After translocation, the exported protein can remain associated with the cell envelope or be fully secreted into the extracellular environment. b) Biogenesis of surface glycoproteins by SecA2/SecY2 systems involves both glycosylation factors (green) and export machinery (yellow) that are distinct from the canonical Sec machinery. Serine-rich proteins are synthesized with N-terminal signal peptides (hatched). The accessory Sec proteins (Asps 1–3) promote SecA2/SecY2 export by unknown mechanisms, but could target preproteins to the translocase and/or serve as a scaffold for the export complex. Asp4 and Asp5 are putative accessory components of the SecY2 channel, but they are not present in all SecA2/SecY2 systems. SecA2/SecY2 export and glycosylation are likely coupled processes. As glycosyl groups (orange hexagons) are added to the preprotein by cytoplasmic glycosylation factors, including the core GtfA and GtfB glycosyltransferases, the SecA2 ATPase will energize transport of the unfolded preprotein through a channel formed by SecY2. SecA2/SecY2 substrates also contain a C-terminal cell wall anchoring domain, which targets the exported protein to the cell wall after SecA2/SecY2 export through the membrane. c) SecA2-only systems lack a SecY2 channel; therefore, SecA2 likely utilizes the canonical SecYEG channel for export. SecA2 could target a certain subset of preproteins to the Sec translocase that are otherwise overlooked or incompatible with SecA1. There are examples of proteins with N-terminal Sec signal peptides (pictured here) and proteins lacking signal peptides that require SecA2 for export. SecA2 may deliver some preproteins to SecYEG and/or associate with SecYEG as needed to export substrates. The ATPase activity of SecA2 likely drives export of preproteins through the SecYEG channel, possibly in conjunction with SecA1.

SecA, SecY, and SecE all have essential roles in Sec export and are required for cell viabilty^{35, 36}. The SecG component of SecYEG is not essential³⁷ but increases translocation efficiency, possibly by stabilizing the SecY/E complex⁷ or assisting the conformational changes of SecA⁸. Other non-essential membrane-bound proteins that increase Sec export efficiency are SecD, SecF and YajC^37–39.

SecA2

For years, it was thought that all bacterial species had a single SecA protein⁴⁰. However, we now know that a number of bacteria possess two SecA homologs: SecA1 and SecA2. The first example of a second SecA was revealed by the Mycobacterium tuberculosis genome sequencing project^{41, 42}. It is now recognized that two SecA proteins exist in all mycobacteria and a diverse, but small, set of Gram positive bacteria including Listeria, Staphylococcus, and some Streptococcus species⁴³. In bacteria with two SecAs, the two proteins are not interchangeable and each SecA has unique functions^{42, 44}. SecA1 is the name given to the SecA with higher sequence similarity to the well-studied SecA proteins of Escherichia coli and Bacillus subtilis. SecA1 is essential and is responsible for canonical Sec export, as described above^{42, 45–48}. Unlike SecA1, SecA2 is responsible for exporting a smaller set of proteins and often dispensable. Notably, proteins exported by SecA2 are linked to virulence in many bacterial pathogens including M. tuberculosis^49–51, Streptococcus gordonii⁵², Streptococcus parasanguinis⁵³, Staphylococcus aureus⁵⁴, and Listeria monocytogenes⁵⁵.

Currently, there are two types of SecA2 systems known to exist. Some bacteria with a SecA2 also have an accessory SecY2 protein. As a consequence, these SecA2/SecY2 systems appear to function largely independent of the canonical Sec machinery to export a set of proteins that are highly glycosylated and incompatible with the canonical SecA1/SecYEG^{56, 57} (Fig. 1b). There are also SecA2-only systems, so named because they lack a SecY2 or an obvious accessory membrane channel. SecA2-only systems likely function as part of the canonical Sec pathway, utilizing SecYEG^{45, 47} (Fig. 1c). Furthermore, the repertoire of proteins exported by SecA2-only systems is more diverse than that of SecA2/SecY2 systems.

While several published crystal structures for SecA (SecA1) proteins exist⁵⁶, including those of M. tuberculosis⁵⁷, B. subtilis⁵⁸ and E. coli⁵⁹, there is no structure available for any SecA2 protein. However, sequence alignments and structural modeling predict that most domains, including PPXD and IRA-1 mentioned above, are conserved between SecA (SecA1) and SecA2 (Fig 2). In addition, all SecA2 proteins have two nucleotide-binding domains (NBD1 and NBD2) which together constitute the DEAD (Asp-Glu-Ala-Asp)-like motor domain. The motor domain contains two ATP-binding Walker boxes and is responsible for ATP hydrolysis^{58, 60, 61}, suggesting that SecA2 proteins are functional ATPases. In fact, SecA2 from S. gordonii⁴⁴ and M. tuberculosis⁶² have demonstrated endogenous ATPase activity in vitro. Furthermore, SecA2 ATPase activity is shown to be required for accessory SecA2 protein export in mycobacteria and C. difficile^{45, 47, 62}.

Figure 2. Domain organization of accessory SecA2 proteins.

Sequence alignments and structural modeling suggest that most functional domains are conserved between SecA1 and SecA2 proteins. The crystal structure of the M. tuberculosis SecA1 protein depicted here represents a typical SecA1 protein with the corresponding colored domains outlined below⁵⁷. *The C-terminal domain (CTD) was not resolved in the M. tuberculosis crystal structure but is shown in the domain graphic. For comparison to SecA1, the predicted domain organization of M. tuberculosis SecA2 and Streptococcus gordonii SecA2 are included. SecA1 can be divided into two main structural domains, which are both composed of several subdomains^18,56. The DEAD (Asp-Glu-Ala-Asp)-like motor domain is responsible for the ATP hydrolysis^58,60,61 and consists of two nucleotide-binding folds: NBD1 and NBD2. NBD1 contains the two ATP-binding Walker boxes^122,123. The helical scaffold domain (HSD) connects the motor domain with the rest of the specificity domain, to allow coupling of preprotein-binding with ATP hydrolysis^124,125. Interactions between SecA1 and the preprotein map along a hydrophobic cleft formed by NBD1, the preprotein crosslinking domain (PPXD)^22–24, and portions of the HSD linker. Within the HSD region is a two-helix finger known as IRA-1, which is thought to assist inserting of the preprotein into the SecY channel during translocation. Conservation of the motor domain between SecA1 and SecA2 proteins confirms the observation that SecA2 proteins are ATPases^44,62. There are differences between SecA1 and SecA2 in the specificity domain that could affect substrate specificity and/or function. All SecA2 proteins are smaller than their SecA1 counterparts due to a truncation of the CTD and helical wing domain (HWD). Although not depicted in the M. tuberculosis SecA1 structure, the CTD was resolved in an E. coli SecA structure²³ and the C-terminal linker (CTL) within the CTD lies along the hydrophobic preprotein-binding cleft.

Even though SecA2 proteins of SecA2/SecY2 and SecA2-only systems likely function differently, it is interesting that all SecA2 proteins are smaller than their SecA1 counterparts due to a carboxyl-terminal domain (CTD) truncation, although the boundary of this truncation varies (Fig. 2). In E. coli, portions of the SecA CTD binds phospholipids²⁹, SecB^63–67, and Zn+^{29, 66}. One area of the CTD missing in all SecA2 proteins is the C-terminal linker (CTL), which lies within the preprotein-binding cleft and in E. coli, has been shown to influence substrate binding²³. In addition to the CTD truncation, the helical wing domain (HWD) is absent in the mycobacterial SecA2 and truncated in other SecA2 proteins. However, even in the canonical SecA1, the function of the HWD is not clear. The significance of the CTD and HWD truncations in SecA2 proteins awaits further studies.

Our understanding of the mechanisms of canonical Sec export is at an impressive level of detail¹, but by comparison our understanding of accessory SecA2 export is limited. Here, we review the current state of knowledge of SecA2 systems and provide models for both the SecA2/SecY2 and SecA2-only pathways. We also discuss gaps in our knowledge of SecA2 export and how these questions can be addressed by extending approaches used previously to study the general Sec pathway.

SecA2/SecY2 systems

Bacteria with SecA2/SecY2 systems share a conserved secA2/secY2 genomic locus, comprised of a suite of similarly arranged homologous genes (Fig. 3a). In addition to genes encoding SecA2 and SecY2, each locus contains a gene that encodes a large serine-rich glycosylated protein that is exported by the SecA2/SecY2 system, as well as glycosylation factors that modify this substrate, and additional export machinery with unknown functions. The SecA2/SecY2 systems that are found in a subset of Gram positive species include pathogenic Streptococcus gordonii⁶⁸, Streptococcus agalactiae⁶⁹, Streptococcus parasanguinis⁷⁰, Streptococcus pneuomoniae⁷¹, and Staphylococcus aureus⁷², although it should be noted that not all streptococcal species possess SecA2/SecY2 systems⁴³. Another notable Gram positive pathogen with a putative SecA2/SecY2 pathway is Bacillus anthracis. However, the B. anthracis secA2 is phylogenetically more distant from those of Streptococcus and Staphylococcus⁴³, which is also reflected by the dissimilar organization of the Bacillus secA2 locus (Fig. 3a). It will be interesting to determine if Bacillus SecA2/SecY2 export functions in the same manner as the Streptococcus and Staphylococcus systems described here.

Figure 3. Organization of accessory secA2 genomic loci.

a) Accessory SecA2/SecY2 systems are found in a diverse set of Streptococcus and Staphylococcus species, where they function in biogenesis of surface glycoproteins (depicted in purple). All of these loci contain the core secA2 and secY2 genes shown in red. Some Bacillus species, including B. anthracis, have putative SecA2/SecY2 systems, although the exported substrates of these systems are unknown and the secA2 and secY2 genes are separated in these cases. Genes encoding putative export machinery are shown in yellow while those encoding glycosylation machinery are shown in green. b) Examples of SecA2-only systems are found in mycobacteria, Clostridium, and Listeria. The secA2 loci of SecA2-only systems are not conserved and export a diverse set of substrates. In some cases, the genes encoding the exported substrates (purple) are found at the secA2 locus. However, this is not always the case and genes located elsewhere in the genome encoding SecA2-dependent proteins are not depicted. In addition to slpA, the C. difficile secA2 locus contains genes encoding eleven additional cell wall proteins that are putative SecA2 substrates, of which three are shown in light purple.

The current model of SecA2/SecY2 export suggests that these specialized systems exist to export the large serine-rich protein encoded in the secA2/secY2 locus. The serine-rich substrates have cleavable N-terminal signal peptides that are unusually long and mature domains that are heavily glycosylated⁴³. Examples of experimentally confirmed SecA2/SecY2-exported substrates include GspB of S. gordonii⁶⁸, Fap1 of S. parasanguinis⁷⁰, Srr1 of S. agalactiae⁶⁹, and SraP of S. aureus⁷². These substrates have roles related to bacterial adhesion to host tissues and/or biofilm formation^{53, 54, 68, 69, 73, 74}. Consequently many of these exported glycoproteins, and presumably their respective SecA2/SecY2 systems, are required for virulence^{52, 54, 69, 71}. In S. parasanguinis, the FimA adhesin is a second protein whose export is reported to depend on SecA2⁷⁰. However, FimA is not a serine-rich glycoprotein and the fimA gene is not at the S. parasanguinis secA2/secY2 locus. It is currently unknown whether FimA is a true SecA2/SecY2 substrate.

SecA2 is not essential for growth in these bacteria; however, SecA2 is absolutely required for export of their respective serine-rich substrates, suggesting a lack of functional redundancy with canonical SecA1. Mutations in secA2 abolish export of the serine-rich glycosylated substrates of S. gordonii⁶⁸, S. parasanguinis⁷⁰, S. agalactiae⁶⁹, and S. aureus⁷². Below, we discuss other genes in the secA2/secY2 loci that have also been analyzed for roles in glycosylation and/or export.

Glycosylation factors of SecA2/SecY2 systems

In export-defective SecA2/SecY2 mutants, the serine-rich substrate retained in the cytoplasm is glycosylated^{70, 75} indicating that the protein is modified by cytoplasmic glycosylation factors prior to export^{70, 75}. There are two core glycosyltransferases conserved in all SecA2/SecY2 systems, GtfA and GtfB (Gtf1 and Gtf2) (Fig. 3a). These enzymes are shown to be required for substrate glycosylation in S. gordonii⁷⁶, S. agalactiae⁶⁹, and S. parasanguinis^{77, 78}, and in S. parasanguinis a Gtf1-Gtf2 interaction is required for substrate glycosylation⁷⁷. Some SecA2/SecY2 systems include additional glycosylation factors that further modify the substrate prior to export, including Gly and Nss of S. gordonii⁷⁹, Nss (Gtf3) and GalT1-2 of S. parasanguinis^{80, 81}, and the GtfC-GtfH proteins of S. agalactiae⁶⁹. While not identical, the carbohydrate modifications on two SecA2/SecY2 exported proteins, Fap1 and GspB, are primarily composed of N-acetylglucosamine and glucose^{82, 83}. In S. parasanguinis, Fap1 is modified by O-linked glycosylation⁸³ but it is unclear if this type of linkage extends to the other SecA2/SecY2 substrates. Furthermore, while the sugar composition of the Fap1 and GspB glycoproteins is known, the number of linkage sites and structure of the glycosyl modifications remains unknown.

Export machinery of SecA2/SecY2 systems

In S. gordonii⁶⁸ and S. aureus⁷², secY2 mutations result in a loss of substrate export that is equivalent to the export defect exhibited by secA2 mutations, demonstrating that SecY2 is essential for accessory SecA2/SecY2 export in these systems. However, in S. parasanguinis⁷⁸, deletion of secY2 has only a modest effect on Fap1 export and the residual exported Fap1 species is incorrectly glycosylated⁷⁸. This result suggests that in the absence of SecY2 and full glycosylation, Fap1 export defaults to the canonical SecA1/SecYEG pathway. This result also suggests that in the S. parasanguinis SecA2/SecY2 system, export and glycosylation of Fap1 are coupled (as discussed further below).

There are additional proteins encoded by secA2/secY2 loci that are referred to as accessory secretion proteins (Asps) in S. gordonii or glycosylation accessory proteins (Gaps) in S. parasanguinis. All SecA2/SecY2 systems include the Asp1, Asp2, and Asp3 proteins (Gap1- 3). While Asp1 and Asp3 are predicted cytosolic proteins, Asp2 may be membrane localized⁸⁴. Some organisms, including S. gordonii, have the additional Asp4 and Asp5 (Fig. 3a). Asp4 and Asp5 are both required for GspB export in S. gordonii⁸⁵ and are predicted integral membrane proteins with sequence homology to B. subtilis SecE (17% identity) and SecG (15% identity), respectively. This sequence homology is intriguing, albeit limited, and suggests that Asp4 and Asp5 may be accessory components of a SecY2 membrane channel. However, the exact role of Asp4 and Asp5 in export remains to be defined. Also unknown is whether there are functionally equivalent proteins in SecA2/SecY2 systems lacking Asp4 and Asp5.

With regard to the role(s) of Asps 1–3, findings in different bacterial systems are not in complete agreement. In both S. aureus and S. gordonii, Asps 1–3 are clearly required for export of the respective serine-rich glycoproteins, SraP and GspB^{72, 79}. Additionally, in a glycosyltransferase-deficient S. gordonii background, export of the non-glycosylated GspB variant is still compromised in asp1, asp2, and asp3 mutants, indicating a role in export that is independent glycosylation for these Asp proteins⁸⁴. By contrast, deletion of either gap1 or gap3 in S. parasanguinis (the asp1 and asp3 homologs) has only a modest effect on export of Fap1, and the residual exported Fap1 protein has altered glycosylation^{86, 87}. On the basis of this result, the Asp1 and Asp3 homologs of S. parasanguinis were named Gap1 and Gap3 to reflect a proposed function in glycosylation. However, a secY2 deletion in S. parasanguinis results in a phenotype similar to that of the gap1 and gap3 mutants – export of an aberrantly glycosylated Fap1^{78, 86, 87}. It seems highly unlikely that SecY2 would have a direct role in protein glycosylation. Additionally, the S. gordonii and S. aureus studies of Asp1-3 in SecA2/SecY2 export are compelling. It is possible that the discrepancy in S. parasanguinis results is because Fap1 glycosylation and export are highly coupled processes. In this case, export defects of gap1, gap3 and secY2 mutants^{78, 86, 87} would indirectly affect Fap1 glycosylation to such an extent that the resulting altered Fap1 species is then compatible for export by the canonical SecA1/SecYEG system. However, at this time, a more direct role for Gaps in glycosylation cannot be ruled out.

While currently there is no clear understanding of the function of any of the Asp proteins, a network of interactions between Asp1, Asp2, Asp3 and SecA2 has been mapped in S. gordonii⁸⁴ and S. parasanguinis^87–89. Asp3 may be a central scaffolding protein in this network as it interacts with multiple members of the SecA2/SecY2 system in S. gordonii including Asp1, Asp2, SecA2, and itself⁸⁴. Interestingly, the S. gordonii Asp2 and Asp3 proteins also bind the GspB substrate prior to its glycosylation, which suggests a possible function for these proteins in delivering the substrate to an export/glycosylation complex⁹⁰. In S. parasanguinis, interactions between Gap1 (Asp1), Gap3 (Asp3), and SecA2 have also been identified^87–89. In addition, it was recently reported that Gap1 of S. parasanguinis stabilizes Gap3 to allow efficient Fap1 export⁸⁹.

Targeting proteins to the SecA2/SecY2 pathway

The serine-rich glycoproteins exported by SecA2/SecY2 systems have features that not only prevent their routing to the canonical SecA1/SecYEG pathway but promote their targeting to the SecA2/SecY2 pathway (Fig. 4). The characteristic glycosylation of these exported substrates is one such element^{69, 70, 75}. In addition to being important for protein stability^{69, 91}, glycosylation of these proteins also blocks their export by the canonical Sec pathway in both S. gordonii and S. parasanguinis^{91, 92}. For example, in the absence of secA2, the canonical SecA1/SecYEG pathway can export a stable, truncated GspB variant that is non-glycosylated. However, a glycosylated GspB protein cannot utilize the canonical Sec pathway and instead requires SecA2/SecY2 for export⁹¹.

Figure 4. SecA2/SecY2-targeting features.

SecA2/SecY2 preproteins have features for both targeting to the SecA2/SecY2 machinery for glycosylation/export, and for blocking export by the canonical Sec pathway. Most of these targeting features have been defined using the S. gordonii GspB protein, shown here. The mature domain of GspB can be divided into several domains. Two serine-rich repeats domains (SRR1 and SRR2) are glycosylated in the cytoplasm prior to export and this post-translational modification blocks GspB export by SecA1/SecYEG. GspB, like all SecA2/SecY2 preproteins has an unusually long N-terminal signal peptide that is required for export. The signal peptide has the same tripartite structure of Sec signal peptides including a positively charged N-terminus, hydrophobic core, and cleavage domain. Within the hydrophobic core of the signal peptide are three glycine residues that are required for GspB export though SecA2/SecY2, but these same glycines also inhibit export by SecA1/SecYEG, through unknown mechanisms. The accessory Sec transport (AST) domain is adjacent to the signal peptide and is required for export by SecA2/SecY2. Also depicted in the mature domain are the basic region and C-terminal cell wall anchoring domain (CWA).

As mentioned above, SecA2/SecY2 serine-rich proteins are glycosylated in the cytoplasm prior to export. This is in contrast to many other glycosylated Sec substrates in bacteria^{93, 94}, as well as the analogous eukaryotic Sec pathway where glycosylation occurs only after proteins are translocated from the cytosol into the endoplasmic reticulum lumen⁹⁵. However, there is evidence that some bacterial proteins in addition to SecA2/SecY2 substrates share the unusual property of being glycosylated prior to Sec export^96–98, such as the HMW1 adhesin of Haemophilus influenzae⁹⁶. But, there is likely a limit to the modifications that the SecA1/SecYEG system can handle, as the level and/or type of glycosylation of SecA2/SecY2 substrates appear incompatible with canonical Sec export. It will be interesting in the future to determine the degree and structure of glycosylation modifications that are incompatible with the SecA1/SecYEG. Additionally, it will be important to understand how these modifications are accommodated by SecA2/SecY2 systems. One possibility is that the pore size of the SecY2 channel is larger than that of SecY1⁹⁹ to allow passage of proteins with large glycan branches. A structure of SecY2 would be incredibly valuable for addressing these unknowns.

In addition to glycosylation, there are other features of SecA2/SecY2 substrates that dictate export by the accessory SecA2 pathway. The distinctive long signal peptides of GspB, Fap1, and presumably other SecA2/SecY2 substrates, are absolutely required for export^{91, 92, 100}. Furthermore, three glycine residues in the hydrophobic core of the GspB signal peptide promote SecA2/SecY2-dependent export¹⁰⁰. However, these same glycine residues also act along with glycosylation to block export by the canonical SecA1¹⁰⁰. The mechanisms by which these glycine residues act in preprotein targeting are currently unknown. Interestingly, these glycine residues are conserved in the signal peptides of most SecA2/SecY2 substrates and could represent a SecA2/SecY2-targeting element that is shared among diverse species.

Finally, there is also a region of approximately 20 amino acids at the start of the mature domain of GspB that is required for targeting this protein to the SecA2/SecY2 system¹⁰¹. Introduction of glycine residues into this region can disrupt this SecA2/SecY2-targeting domain, suggesting it may form an alpha-helix¹⁰¹. Currently, it is not known if this accessory Sec transport (AST) domain is a conserved feature of all SecA2/SecY2 substrates. However, the first 34 amino acids of the Fap1 mature domain are also required for SecA2-dependent export⁹². In addition to targeting to the SecA2/SecY2 translocase, the AST domain may also stabilize an open SecY2 channel conformation¹⁰¹.

Model for SecA2/SecY2 export

The current model of SecA2/SecY2 export is as follows (Fig. 1b). The distinctive glycosylation of the serine-rich proteins of SecA2/SecY2 systems is incompatible with export via the canonical SecA1/SecYEG pathway and demands a specialized export system. In a signal peptide-dependent manner, the SecA2/SecY2 preproteins are targeted to the SecA2/SecY2 machinery^{91, 92, 100}. Features of the mature domain, such as the AST¹⁰¹, may also be involved in targeting. In addition, Asp2 and Asp3 could contribute to translocase-targeting by binding the unmodified substrate⁹⁰.

Analogous to canonical Sec export, SecA2 likely uses cycles of ATP hydrolysis to drive unfolded, glycosylated preproteins through the SecY2 channel. In some bacteria, Asp4 and Asp5 may function like SecE and SecG, whereas SecA2/SecY2 systems lacking Asp4 and Asp5 could utilize the canonical SecE or SecG for export. In fact, there is some genetic evidence that SecY2 and SecG may function together in S. aureus¹⁰². Also akin to canonical Sec export, experiments using a slow-folding model protein suggest that SecA2/SecY2 preproteins must remain unfolded for passage through the SecY2 channel¹⁰¹.

At this time, there remain many unknowns regarding the mechanism of SecA2/SecY2 export and glycosylation. What is the sequence of preprotein interactions among the SecA2/SecY2 export and glycosylation machinery? Are glycosylation and export coupled? What are the features of SecA2/SecY2 glycosylation that necessitate export by the SecA2/SecY2 system? Does SecA2 function in targeting and/or translocation across the membrane? What is the role of each Asp protein, and do these roles vary among SecA2/SecY2 systems of different bacteria? In vitro glycosylation and translocation systems would help address questions about this new type of export system.

SecA2-only systems

Bacteria with SecA2-only systems lack a SecY2 homolog or obvious accessory membrane channel. The emerging model is that the SecA2 proteins of these systems work with the canonical SecA1/SecYEG translocase. Unlike secA2/secY2 loci, there is no conservation of gene content or organization at the secA2 genomic region for SecA2-only systems (Fig. 3b). In addition, there is a greater variety in the types of proteins exported by SecA2-only systems when compared to the category of glycosylated serine-rich proteins exported by SecA2/SecY2 systems. SecA2-only systems exist in all mycobacteria, including the human pathogen M. tuberculosis⁴², as well as some Gram positive bacteria such as L. monocytogenes¹⁰³, Corynebacterium glutamicum⁴⁸, and Clostridium difficile⁴⁷.

In mycobacteria^{42, 49, 104} and Listeria^{103, 105–108}, SecA2 is not essential for growth in liquid media and secA2 mutants are defective in the export of specific proteins. However, secA2 mutants of both M. tuberculosis and L. monocytogenes are attenuated for growth in infection models, indicating the importance of the respective SecA2 systems for exporting virulence factors^{49, 50, 55}. Additionally, secA2 mutants of both M. tuberculosis and L. monocytogenes elicit aberrant immune responses during infection, which has led to the use of these mutants in vaccination studies^{50, 109–111}. By contrast, SecA2 is essential for growth of Corynebacterium glutamicum⁴⁸, and Clostridium difficile⁴⁷.

SecA2-only exported substrates

Proteins exported by SecA2-only systems have been identified in the model organism M. smegmatis (a nonpathogenic mycobacterial species), M. tuberculosis, L. monocytogenes, and C. difficile. Some of these proteins have predicted N-terminal Sec signal peptides and some do not. In M. smegmatis there are two lipoproteins (Ms1704 and Ms1712) with predicted N-terminal Sec signal peptides that are exported to the cell wall in a SecA2-dependent manner^{45, 112}. Ms1704 and Ms1712 are homologous and both are predicted ABC type sugar-binding proteins of putative ABC transporters¹¹². However, it is important to note that not all mycobacterial lipoproteins require SecA2 for export¹¹².

In M. tuberculosis and L. monocytogenes, several proteins are reduced in exported fractions of secA2 mutant bacteria analyzed by 2D-PAGE, 3 and 17 respectively^{49, 55}. Of these proteins, only a few have been studied further and confirmed to be SecA2-dependent. In M. tuberculosis, one of these proteins is superoxide dismutase SodA, which notably lacks a predicted cleavable Sec signal peptide. There are likely additional SecA2-dependent proteins in M. tuberculosis because SecA2 is required to block phagosome maturation and the SecA2-dependent effectors of this process are unknown⁵¹.

In L. monocytogenes the p60 autolysin, which is a cell wall hydrolase that cleaves peptidoglycan, is a confirmed SecA2 substrate⁵⁵. The gene for p60 is positioned adjacent to secA2 in the genome (Fig. 3b); although, other SecA2-dependent proteins of Listeria are encoded elsewhere. An additional peptidoglycan-hydrolyzing autolysin NamA (MurA) of Listeria is also SecA2-dependent^{55, 106, 107}. However, unlike p60, NamA lacks a typical Sec signal peptide. MnSod superoxide dismutase is another protein lacking a predicted Sec signal peptide that is exported in a SecA2-dependent manner in L. monocytogenes¹¹³. This particular finding parallels the SecA2-dependence of SodA export in M. tuberculosis and suggests that other similarities may exist between the mycobacteria and Listeria SecA2-only systems. Also among the list of SecA2-dependent proteins identified in L. monocytogenes are four predicted lipoproteins with Sec signal sequences⁵⁵. One of these lipoproteins is a predicted ABC type sugar-binding protein⁵⁵, which is in the same family as the Ms1704 and Ms1712 substrates of M. smegmatis¹¹².

In C. difficile the S-layer protein SlpA,⁴⁷ which constitutes a proteinaceous lattice structure surrounding the Clostridium cell¹¹⁴ has been identified as being exported in a SecA2-dependent fashion. SlpA is a member of a larger family of 29 clostridial cell wall proteins (Cwp)¹¹⁵ that are implicated in host-pathogen interactions^116–118. CwpV is another protein shown to require SecA2 for export⁴⁷, suggesting that additional members of this Cwp family may be SecA2-dependent as well. In C. difficile, the secA2 gene is adjacent to slpA and the larger secA2 genomic region includes genes encoding 12 Cwps (Fig. 3b)⁴⁷. However, the gene encoding CwpV is notably located elsewhere in the genome⁴⁷. In C. difficile, both of the demonstrated SecA2-dependent proteins (SlpA and CwpV) contain predicted N-terminal Sec signal peptides¹¹⁵.

Export machinery of SecA2-only systems

SecA2-only systems lack an obvious alternative membrane channel and accessory export factors. An attractive idea is that SecA2 works with the canonical SecA1/SecYEG machinery either through cooperation with SecA1 or by sharing SecYEG. In support of this model, depletion of the essential SecA1 protein in M. smegmatis abolishes export of the SecA2 substrate Ms1712⁴⁵. The simplest interpretation of this experiment is that mycobacterial SecA2 export requires the canonical SecA1. However, it remains possible that SecA1 depletion in this experiment has an indirect effect on SecA2 export.

In mycobacteria and C. difficile, studies using ATPase-defective dominant negative SecA2 proteins are also consistent with a model where SecA2 works with the canonical Sec machinery^{45, 47}. Dominant negative proteins often exert their effect by forming nonfunctional complexes with their normal binding partners. In mycobacteria, overexpression of the dominant negative SecA2 inhibits growth⁴⁵. This result implies an interaction between SecA2 and proteins important to an essential process, with the essential SecA1/SecYEG machinery being a leading candidate. In C. difficile, expression of the corresponding dominant negative SecA2 also inhibits growth and over shorter time frames is shown to impact protein export⁴⁷. Importantly, overexpression of a dominant negative SecA1 in C. difficile reduces export of SecA2 substrates, possibly by blocking accessibility of the SecA2 substrates to the SecYEG channel⁴⁷. However, unlike in mycobacteria, depletion of SecA1 in C. difficile does not influence export of SecA2 substrates⁴⁷ suggesting that in Clostridium SecA2 works with SecYEG but not SecA1.

There has not been a similar investigation for a relationship between SecA2 and SecA1/SecYEG in Listeria, but recently it was shown that secretion of the SecA2-dependent proteins p60 and NamA depends on the DivIVA protein¹⁰⁵. The DivIVA protein is involved in localizing proteins to the cell poles and septa of bacteria¹¹⁹. Interestingly, GFP fusions to DivIVA, SecA2, p60, and NamA all localize to the septum in Listeria¹⁰⁵. Thus, it is possible that the SecA2-only system is specifically localized and DivIVA is required to either establish that localization pattern or deliver the SecA2-dependent proteins to the SecA2 machinery.

Targeting proteins to the SecA2-only pathway

The features defining exported substrates of SecA2-only systems have not yet received significant attention. The two M. smegmatis lipoproteins that require SecA2 for export contain typical N-terminal Sec signal peptides. While these signal peptides are required for export¹¹², they are not specific for targeting these proteins to SecA2 (Feltcher and Braunstein, unpublished). Therefore, there appears to be one or more features of the mature domains that determine SecA2-dependent export in mycobacteria. The mature domain of these SecA2-dependent substrates could possess post-translational modifications, as seen with the SecA2/SecY2 substrates, although there is currently no evidence for this possibility.

Model for SecA2-only Export

For SecA2-only systems, genetic studies support a model where SecA2 utilizes SecYEG to assist in Sec export of a certain class of preproteins^{45, 47}(Fig. 1c). These SecA2 substrates must have features, which are not understood at this time, that make them incompatible for Sec export without the assistance of SecA2. SecA2 of both mycobacteria and Clostridium primarily localize to the cytoplasm, while much of SecA1 is found at the membrane in these bacteria^{45, 47}. Thus, SecA2 could possibly function in the cytoplasm to recognize and target for export a specific subset of proteins that are otherwise overlooked or incompatible with the canonical SecA1. Alternatively, SecA2 could serve as an alternate motor protein that is necessary for translocation of certain proteins through SecYEG. In either case, SecA2 ATPase activity is required for export.

Application of an in vitro translocation system would help strengthen this developing model and address basic questions about the operation of SecA2-only systems. Is there a role for SecA1 during SecA2-only export? What prevents the substrates of SecA2-only systems from being exported solely by SecA1/SecYEG system and how does SecA2 recognize these proteins? Does SecA2 function in targeting and/or translocation?

Another unresolved issue is whether preproteins without recognizable Sec signal peptides, like the Sod proteins of M. tuberculosis and L. monocytogenes, are true SecA2 substrates or if the export of these proteins is indirectly affected by SecA2. For example, SecA2 could export a currently unknown protein with a signal peptide that is itself required for export of proteins like Sod. Still, it is also plausible that proteins without signal peptides are recognized by SecA2 and exported directly through the SecYEG channel. In support of this model, SodA of Rhizobium leguminosarum is exported in a SecA-dependent manner despite lacking a recognizable signal peptide¹²⁰. However, similar to M. smegmatis⁴⁵ and C. difficile⁴⁷ studies, an indirect role for the Sec system in exporting SodA of R. leguminosarum cannot be ruled out.

Future directions

While it is now clear that SecA2 systems are critical for exporting specific proteins, surprisingly little is known about how these export systems function. The canonical Sec pathway is a well characterized process resulting from years of extensive study; however, the acquisition of SecA2 proteins in some bacteria suggests limitations to SecA1 and SecYEG function. The same structural, biochemical and in vitro reconstitution analyses used to understand canonical Sec export will surely prove valuable in answering the many mechanistic questions that exist about the SecA2/SecY2 and SecA2-only systems.

Structural regions of difference between SecA1 and SecA2 proteins will be of obvious interest to investigate. It will also be informative to compare SecA2 structures from SecA2/SecY2 systems to those of SecA2-only systems. Surprisingly, the predicted domain organization of SecA2 proteins from SecA2/SecY2 systems and SecA2-only systems is similar, despite the known differences between these two types of SecA2 export pathways. Given the possibility that SecA2 proteins might be specifically adapted to accommodate binding of certain glycosylated preproteins or proteins lacking signal peptides, it will be particularly interesting to compare the architecture of SecA2 preprotein-binding clefts to the corresponding regions of SecA1 proteins. Also of interest will be the SecA2 IRA-1 two-helix finger domain^{120, 121}. The amino acid at the tip of the two- helix finger, which contacts the preprotein, is not conserved between SecA1 and SecA2 proteins¹²¹. This could reflect unique interactions between SecA2 proteins and their respective substrates or interactions with a unique SecY2 channel.

Finally, although there remains controversy about the multimeric state of the canonical SecA during export⁵⁶, it will be interesting to establish whether SecA2 forms and functions as dimers. Heterodimers of SecA1 and SecA2 could possibly explain the roles of both these SecA proteins in export of some mycobacterial substrates⁴⁵. In fact, SecA2 can bind SecA1 in S. parasanguinis⁸⁸, but it is unknown if this interaction has any biological function in export.

Accessory SecA2 systems make significant contributions to protein export in a diverse set of bacteria. Future research should apply the same depth of structural and biochemical studies to SecA2, as done for the canonical Sec pathway. A deeper understanding of accessory SecA2 systems will reveal additional mechanisms used by bacteria to transport proteins across the cytoplasmic membrane, and it is also likely to reveal unappreciated limitations of canonical Sec export.

Online Summary.

• The post-translational general secretion (Sec) pathway is powered by the essential SecA ATPase, which works with the SecYEG channel to translocate proteins across the cytoplasmic membrane.

• Mycobacteria and some Gram positive bacteria have two non-redundant SecA proteins: SecA1 and SecA2. SecA1 powers the essential canonical Sec pathway, while SecA2 exports of a limited set of proteins.

• SecA2 export systems have diverse and important roles in cell envelope biogenesis and bacterial pathogenesis.

• There are two types of SecA2 systems: SecA2/SecY2 systems and SecA2-only systems.

• SecA2/SecY2 systems appear to operate mostly independent of the canonical Sec machinery as dedicated transporters of serine-rich glycoproteins that function as adhesins. SecA2/SecY2 systems are encoded by a genomic locus that includes open reading frames for glycosylation and export machinery.

• SecA2-only systems do not contain a SecY2 or an obvious accessory membrane channel. Instead, SecA2-only systems appear to use the canonical SecYEG channel for exporting a diverse assortment of proteins.

• It is unknown how SecA2 functions in comparison to the well-studied canonical SecA (SecA1) proteins. The same structural, biochemical and in vitro reconstitution analyses used to understand canonical Sec export will surely prove valuable in answering the many mechanistic questions concerning SecA2/SecY2 and SecA2-only systems.

Glossary

In vitro reconstitution

A technique for studying biochemical processes in vitro. In vitro reconstitution of the Sec pathway involves incubation of preprotein and purified SecA with SecYEG-containing inverted membrane vesicles (IMV). Preprotein translocation through SecYEG into the IMV is monitored by the loss of preprotein sensitivity to protease.

Preproteins

Proteins synthesized with N-terminal Sec signal peptides for targeting to the Sec machinery.

Signal peptide

An N-terminal amino acid sequence present on preproteins. The signal peptide helps target specific proteins for export out of the bacterial cytoplasm. Sec signal peptides are comprised of a tripartite structure with a positively charged N-terminus, hydrophobic core, and signal peptidase cleavage site.

Glycosyltransferase

An enzyme that catalyzes the post-translational addition of carbohydrate mono- or oligosaccharides to an acceptor molecule. Some glycosyltransferases attach sugars to proteins, resulting in glycoproteins. The sugar composition and structure of glycosyl modifications can be diverse.

N-linked glycosylation

A type of glycosylation where a saccharide moiety is added to the to the amide nitrogen of asparagine.

O-linked glycosylation

A type of glycosylation where a saccharide moiety is added to the hydroxyl oxygen of serine or threonin.

Lipobox

An amino acid motif found in the signal peptidase cleavage site of bacterial lipoproteins. The lipobox motif contains an invariant cysteine that becomes the first amino acid of the mature protein after signal peptidase cleavage and is the site of diacylglycerol attachment.

Autolysin

Hydrolases that break the peptidoglycan matrix in the bacterial cell wall. Autolysins are important in bacterial cell growth and division.

S-layer

An ordered array of protein subunits that form a lattice structure surrounding the bacterial cell wall. S-layers exist in many Gram-positive and Gram-negative bacteria, as well as Archaea. S-layers serve as scaffolding structures for enzymes, contribute to cell surface adhesion, and act as virulence factors among other functions.

Phagosome maturation

Phagosomes are vacuoles that form around foreign particles or bacteria during phagocytosis by eukaryotic cells. The phagosome undergoes multiple maturation events and ultimately fuses with lysosomes resulting in a degradative phagolysosomal compartment.

Biographies

Meghan Feltcher

Meghan Feltcher obtained a B.S. in genetic engineering from Cedar Crest College. She is currently a Ph.D. candidate in the Department of Microbiology and Immunology at the University of North Carolina in Chapel Hill. Working in the laboratory of Dr. Miriam Braunstein, her graduate work is focused on defining mechanisms of SecA2 export in mycobacteria and identifying novel SecA2 substrates in M tuberculosis.

Miriam Braunstein

Miriam Braunstein is an associate professor in the Department of Microbiology and Immunology at the University of North Carolina at Chapel Hill. She obtained a B.S. in biology from Tufts University and went on to receive M.A. and Ph.D. degrees in molecular biology from Princeton University. Her Ph.D. research was conducted with Dr. James Broach and focused on the role of histone acetylation and the Sir2 deacetylase on gene expression in Saccharomyces cerevisiae. Her postdoctoral studies were conducted with Dr. William Jacobs Jr. at the Albert Einstein College of Medicine, where she began studying the protein export systems of M. tuberculosis. In 2001, she took an assistant professor position at the University of North Carolina Chapel Hill. Her laboratory studies conserved and specialized protein export systems of M. tuberculosis. Her research group is also working to identify M. tuberculosis proteins secreted during host infection and determine the functional significance of these proteins in virulence.