An important aspect of bacterial physiology and pathogenicity is the ability to move proteins out of the cytoplasm, across membranes and very often out of the cell completely. Gram-negative bacteria export proteins across the inner membrane to the periplasm using the Sec or Tat machineries and have a variety of specialized secretion machineries for secreting proteins across the outer membrane (OM) to the exterior of the cell, or even into host cells (e.g. Type I–VI secretion systems). Certain proteins are also permanently localized to the inner or outer membranes.
Dickeya sp., formerly known as Erwinia chrysanthemi, is an enteric phytopathogen causing soft rot disease of a variety of economically significant plants such as potato, maize and carnation. Dickeya and other members of the soft rot erwinias cause soft rot disease primarily by the co-ordinated production of large amounts of multiple secreted plant cell wall-degrading enzymes (PCWDEs), the most important of which are the pectinases, including multiple isoforms of pectate lyase, polygalacturonases and pectin methylesterase (Toth et al., 2003). In Dickeya and Erwinia carotovora (another soft rot erwinia), the major PCWDEs are secreted by a Type II secretion (T2S) system known as Out. T2S systems provide a key conduit for the secretion of virulence factors in many Gram-negative pathogens, and secretion of PCWDEs by the Dickeya Out system, comprising OutB–M, OutO and OutS, is a well-studied model for T2S (Sandkvist, 2001). T2S is a two-step pathway in which proteins are exported to the periplasm via the normal Sec or Tat export machinery (see below) and are then secreted to the external environment by the T2S apparatus. The T2S system is a complex multiprotein complex that spans the cell envelope and is related to the Type IV pilus assembly machinery (Filloux, 2004). It includes an OM pore complex composed of a single major protein, secretin, an inner membrane-energizing platform and ‘pseudopilins’, believed to form a pilus-like structure in the periplasm. The secretion motif for T2S substrates is poorly defined but is believed to involve recognition of folded structural motifs.
Proteins are usually transported across the inner (cytoplasmic) membrane into the periplasm by one of two general export pathways. The Sec pathway, which is the predominant route of protein transport, recognizes proteins bearing N-terminal signal peptides and exports them in an unfolded conformation. In contrast, the Tat system transports folded proteins, many of which bind redox cofactors, which are targeted for export by N-terminal signal peptides containing an almost invariant twin-arginine motif. In addition to very many soluble periplasmic and secreted proteins, the vast majority of integral inner and outer membrane proteins, as well as lipoproteins, are substrates of the Sec system. The Tat system does assemble a small number of monotopic inner membrane proteins that are anchored either by means of non-cleaved signal anchor sequences (e.g. the Rieske iron-sulphur proteins of bacteria and plants) or by short C-terminal hydrophobic stretches (e.g. the small subunits of formate dehydrogenases and hydrogenases) (Berks et al., 2005). More recently, Tat has also been shown to export some lipoproteins (Gimenez et al., 2007).
The Gram-negative OM contains bilayer-spanning integral proteins as well as lipoproteins that are anchored in the membrane by fatty acid modification of N-terminal cysteine residues. Most integral OM proteins (OMPs) have a β-barrel architecture, formed from amphipathic β-strands that traverse the membrane a number of times. β-Barrel OMPs are exported across the inner membrane by the Sec system and are maintained in an unfolded state by the action of periplasmic chaperones, SurA and Skp, prior to interaction with the Bam complex, which facilitates their integration into the OM (Bos et al., 2007). The high-resolution structure of Wza revealed that OMPs can also span the bilayer by means of amphipathic α-helices that form a so-called α-barrel. However, the assembly pathway for this latter class of membrane protein is currently unknown (Whitfield and Naismith, 2008).
The biogenesis of OM lipoproteins utilizes a mechanism distinct from that of integral OMPs. After transport across the inner membrane by either the Sec or Tat pathways, lipoprotein precursors are modified by attachment of a diacylglyceryl moiety to the conserved cysteine of the lipobox, rendering the modified precursors substrates for the lipoprotein-specific signal peptidase II. In Gram-negative bacteria, lipoproteins are subsequently further modified by the attachment of a fatty acid to the now N-terminal cysteine residue. This amine-linked fatty acid modification is necessary (but probably not sufficient) for recognition of the fully matured lipoprotein by the Lol machinery, which transports lipoproteins from the periplasmic face of the inner membrane to the periplasmic face of the OM. The Lol system comprises the ABC transporter, LolCDE, the periplasmic chaperone, LolA, which is responsible for shuttling mature lipoproteins across the periplasm, and the OM-docking protein LolB, which facilitates lipoprotein insertion into the inner leaflet of the OM. Lipoproteins that are surface-exposed (with the exception of those in spirochaetes such as Borrelia burgdorferi) are transported to the outer leaflet of the OM by a T2S system (Bos et al., 2007).
In this issue of Molecular Microbiology, Ferrandez and Condemine describe a new class of OMP and a novel means of OM targeting, dependent on Tat and a new T2S system, which acts on the pectin lyase, PnlH, in Dickeya dadantii (Ferrandez and Condemine, 2008; Fig. 1). The authors identified a cluster of genes apparently encoding a second T2S system, in addition to the well-characterized out gene cluster, from the genome sequencing project of D. dadantii 3937. This cluster contained 11 genes, homologues of all the Out proteins except OutB, H and O, and was named stt (second type two). A gene encoding a candidate substrate for Stt was identified immediately downstream of the stt genes: pnlH, encoding a homologue of the E. carotovora pectinase, pectin lyase (Pnl). Use of reporter gene fusions revealed that both stt and pnlH were expressed at very low levels under normal laboratory conditions, but were derepressed in an hns mutant and at elevated growth temperatures. An hns background was used throughout the study to provide easily detectable levels of PnlH. Surprisingly, PnlH was not detectable in the cell supernatant (unlike Out-secreted pectinases, for example), instead it was localized to the OM fraction and shown, by imunofluorescence microscopy, to be located on the outer surface of the cell. In either an stt mutant or in Escherichia coli (lacking Stt), PnlH was still localized in the OM but could not be detected on the outside of the cell. Thus the Stt system is responsible for localizing PnlH to the outer face of the OM, although it is not responsible for initially targeting the protein to the OM.
Examination of the N-terminal sequence of PnlH revealed the presence of a well-conserved Tat signal sequence, but lacking any obvious signal peptidase cleavage site. The authors hypothesized that this N-terminal region might represent an uncleaved Tat signal sequence acting as an OM anchor. PnlH was confirmed to be translocated across the inner membrane by the Tat system as it was confined to the cytoplasmic fraction and unable to reach the periplasm or OM in a tat mutant of either D. dadantii or E. coli. Moreover, mutation of the conserved RR motif in the Tat signal of PnlH to RG also impaired membrane localization of the protein. N-terminal sequencing of PnlH purified from the OM of E. coli revealed that no processing of the N-terminus had occurred, consistent with the idea that the N-terminus can act as an uncleaved signal anchor.
In order to address the question of whether the N-terminal sequence is sufficient to direct all of the targeting steps of PnlH, hybrid protein fusions of the N-terminus of PnlH to a cytoplasmic protein (KdgR) and the mature domains of a periplasmic (Bla) and an Out-secreted (PemA) protein were constructed. All fusions were localized in the OM of E. coli and D. dadantii, confirming that the N-terminus of PnlH is sufficient to address proteins to the OM. Importantly, however, none of the fusions were detectable on the outer surface of D. dadantii cells, demonstrating that the signal for secretion by Stt is present in regions of the protein other than the N-terminus, as is the case for the substrates of other ‘typical’ T2S systems (Filloux, 2004). In reciprocal experiments, the N-terminus of PnlH was replaced by a classical Tat signal sequence (from HybO) and a classical Sec signal sequence (from PemA). The classical Tat signal directed PnlH to the periplasm (not the OM) and the Sec signal was also unable to direct proper secretion of PnlH, confirming that the N-terminus of PnlH represents a new type of Tat signal sequence, able to target proteins to the OM.
The work of Ferrandez and Condemine is exciting because it identifies a new type of OMP, PnlH, whose transmembrane domain closely resembles that of a standard inner membrane protein, such as the Rieske subunit of the cytochrome bc1 complex. This raises some very significant questions, for example: How is PnlH distinguished from other inner membrane proteins to allow differential targeting to the OM? What is the process by which such a transmembrane protein is extracted from the inner membrane and how might it then be targeted to the OM? Genome analysis suggests that it is unlikely that E. coli encodes any similar Tat-dependent OMPs; however, this study clearly demonstrates that E. coli also has the capacity to target PnlH to its OM. This gives a strong indication that the OM-targeting mechanism is conserved between Dickeya and E. coli. One possibility is that the mechanism by which PnlH is targeted to the OM is related to the route by which other classes of OMPs are targeted, for example, the α-barrel proteins and/or some OM secretins, as membrane integration of the secretin PulD has been shown recently to be independent of the BamA/YaeT protein (Collin et al., 2007).
The other key mechanistic question raised by this study is that of how the Stt T2S system is able to ‘flip’ PnlH from the periplasmic face of the OM to its final, surface-exposed location. As the authors note, the conformation of the N-terminal signal anchor in the OM is unknown; in the transition of the soluble domain from the periplasm to the cell surface, the orientation of the signal anchor may be inverted, or it may remain the same and PnlH may be somehow ‘folded back’ through the OM to reach the surface. By analogy with other T2S systems, it seems most likely that Stt moves PnlH across the OM through a channel formed by the secretin, SttD. Interestingly, the stt cluster does not contain an OutO homologue, and does not require OutO itself for function. OutO and other T2S system prepilin peptidases are required for processing of pseudopilin subunits prior to their polymerization into a pilus-like structure. A functional prepilin peptidase activity is required for T2S, but in several cases the protein is ‘borrowed’ from a Type IV pilus assembly system (e.g. Pseudomonas aeruginosa PilD) (Filloux, 2004). Hence the Stt system of D. dadantii might also utilize a Type IV pilus peptidase.
The findings of this study also have interesting implications for the pathogenicity and plant interaction of Dickeya, and for protein secretion in other organisms. Unlike other pectinases in Dickeya, PnlH is not secreted using the Sec and Out systems; rather, it uses Tat and the novel Stt system. Similarly, in E. carotovora, Pnl is Out-independent (and lacks a Sec signal), but has been suggested to be released from the cell by stress-induced cellular lysis (Coulthurst et al., 2008). This implies that Pnl is somehow incompatible with the Sec–Out route, perhaps because it requires cytoplasmic factors for folding. The stt system of D. dadantii appears to be horizontally acquired, as it is located between two tRNA genes, has an atypical GC content relative to the rest of the genome and is not present in the related E. carotovora ssp. atroseptica. It is not yet clear whether there are substrates for Stt other than PnlH in D. dadantii. One way of identifying such substrates might be to look for Tat signal sequences in proteins lacking any apparent signal peptidase cleavage site.
An important question raised by this work is the role and importance of Stt and PnlH in the infection and disease process of D. dadantii. While the impact of pnlH and stt inactivation on plant pathogenicity is yet to be reported, it is hard to imagine that the acquisition of a new T2S system and at least one cognate pectinolytic enzyme would occur without it conferring some selective advantage in a plant host. Pectin lyase, which, unlike pectate lyases, is able to cleave methylated pectin, is most likely to play a role in plant cell wall breakdown. An intriguing question is why this PCWDE is anchored to the cell, rather than released to the supernatant, as are the other secreted PCWDEs of Erwinia. Perhaps localizing PnlH in close proximity with proteins downstream of Pnl in pectin breakdown, e.g. the oligogalacturonate-specific OM porin, KdgM, or the pectin methylesterase, PemB, might facilitate efficient pectin catabolism. It could also be speculated that the role of PnlH might include, or even consist entirely of, a role in adhesion of the bacterial cells to pectin-containing plant tissue by binding of the anchored protein to pectin.
PnlH might act early in the infection process, as it is not regulated by KdgR, a transcriptional repressor mediating induction of pectinase genes only in the presence of plant cell wall breakdown products (Toth et al., 2003). In E. carotovora, expression of pnl is upregulated in response to DNA-damaging stress via the RdgAB regulators, and Pnl is proposed to be released by stress-induced lysis (Liu et al., 1996). Perhaps, by analogy, expression of D. dadantii pnlH is activated in response to certain stress signals, consistent with the increased expression observed at increased temperature. In addition, pnl in E. carotovora has been shown recently to be regulated by an intercellular signalling process known as quorum sensing, in common with many other virulence factors, emphasizing the potential importance of Pnl to Erwinia pathogenicity (Liu et al., 2008). The finding that the stt genes were only expressed at very low levels in laboratory conditions suggests that the Stt T2S system is only elaborated in response to plant-derived signals. Regulation of Stt is distinct from that of the Out system of D. dadantii, as expression of stt was not dependent on the normal regulators of pectinase synthesis and secretion, KdgR and PecS (Hugouvieux-Cotte-Pattat et al., 1996). Stt provides another example of co-regulation of T2S systems with their cognate substrates in Erwinia. The expression of both stt and pnlH was Hns- and temperature-dependent, similar to the KdgR- and PecS-dependence of secreted pectinases and the Out system in Dickeya, and the recently observed QS-dependence of the Out system and its PCWDE substrates in E. carotovora (Liu et al., 2008).
Finally, this study raises the exciting possibility that this new mechanism of protein targeting is important in other bacteria, including pathogens. The use of a Tat signal as an uncleaved OM anchor is likely to be a general mechanism, as the heterologous host, E. coli, was able to successfully integrate proteins bearing the D. dadantii N-terminus into its OM. It remains to be seen how widely other Stt-like T2S systems, which transfer an integrated OM substrate to the outer face of the OM, may be distributed.
In conclusion, this work has identified a novel route by which Gram-negative bacteria can target proteins to the outer surface of the bacterial cell, utilizing the Tat translocase and a novel T2S system. This targeting mechanism is likely to play a role in the plant interaction of the phytopathogen, D. dadantii, and may also be important in other pathogenic bacteria.