The processing of membrane-anchored signalling molecules and transcription factors by RIP (regulated intramembrane proteolysis) is a major signalling paradigm in eukaryotic cells. Intramembrane cleaving proteases liberate fragments from membrane-bound precursor proteins which typically fulfil functions such as cell signalling and regulation, immunosurveillance and intercellular communication. Furthermore, they are thought to be involved in the development and propagation of several diseases, such as Alzheimer's disease and hepatitis C virus infection. In this issue of the Biochemical Journal, Schrul and colleagues investigate the interaction of the endoplasmic reticulum-resident intramembrane cleaving SPP (signal peptide peptidase) with different type II oriented transmembrane proteins. A combination of co-immunoprecipitation experiments using wild-type and a dominant-negative SPP with electrophoretic protein separations under native conditions revealed selectivity of the interaction. Depending on the interacting protein, SPP formed complexes of different sizes. SPP could build tight interactions not only with signal peptides, but also with pre- and mis-folded proteins. Whereas signal peptides are direct substrates for SPP proteolysis, the study suggests that SPP may be involved in the controlled sequestration of possibly toxic membrane protein species in a proteolysis-independent manner. These large oligomeric membrane protein aggregates may then be degraded by the proteasome or autophagy.
The concept of regulated intramembrane proteolysis has emerged over the last few decades as a novel concept of cellular signalling. One group among the proteases capable of cleaving substrates within the phospholipid bilayer is the SPP (signal peptide peptidase) and its homologues, the SPPLs (SPP-like proteins) [1,2].
Three different families of proteases which are able to promote RIP (regulated intramembrane proteolysis) have been described. One of these is a group of metalloproteases comprising human site-2 protease and homologous proteins. The second family is represented by a group of serine proteases including Drosophila melanogaster rhomboid-1. The third group is formed by several aspartic proteases. They include the human PSs (presenilins) PS1 and PS2, which provide the catalytically active part of the γ-secretase, SPP , and four additional proteases which have been discovered by database searching. Because of their homology with the PSs and SPP, these additional proteases are referred to as PSHs (PS homologues) or SPPLs. The SPPLs include the proteins SPPL2a, SPPL2b, SPPL2c and SPPL3. SPP and SPPLs are found throughout eukaryotes, in fungi as well as in plants, animals and archaea [4,5]. SPP and SPPL3 are predominantly localized in the ER (endoplasmic reticulum). SPP was first described as an ER protease involved in the generation of self-peptides presented by HLA-E, suggesting an important role of SPP in the regulation of immunological surveillance .
PSs, SPP and SPPLs share a catalytically critical aspartate residue embedded within a GxGD motif that forms the catalytic centre of the proteases, together with the aspartate residue of the equally critical N-terminal YD motif. Therefore they can be characterized as GxGD proteases . In addition, a PALP motif which is reported to be obligatory for stabilization, complex formation and γ-secretase activities of presenilins is highly conserved in PSs, SPP and SPPLs . One of the most important differences between PSs and the other GxGD proteases is the opposite orientation of the catalytic domains inside the membrane. PSs are able to catalyse intramembrane cleavage of several type I anchored membrane proteins (e.g. β-amyloid precursor protein and Notch I), whereas SPP and SPPLs seem to show selectivity for type II oriented transmembrane segments [e.g. signal sequences, HCV (hepatitis C virus) core protein precursor] . The similarities between PSs and SPP suggest a common catalytic mechanism which is underlined by the observation that, in PS and SPP, mutations of the conserved aspartate residues in the YD and GxGD motif abolished proteolytic activity and some inhibitors were reactive towards both types of proteases . SPP probably sequentially cleaves its substrates at multiple sites within its transmembrane segment. Furthermore, presenilins require a number of cofactors to build a functional γ-secretase complex, whereas all SPP family members apparently do not require such proteins. It is suggested that short membrane stubs are substrates for proteolysis by SPP or SPPLs. Usually, a process termed ectodomain shedding generates such a stub after releasing a large extracellular domain. In the case of SPP, the separation of a signal peptide from a preproprotein by signal peptidase can be regarded as this initial ‘shedding-like’ event (Figure 1A). In addition to its suggested role in the processing of MHC class I molecules, SPP may have a function in protein dislocation from the ER. This plays an important role during HCMV (human cytomegalovirus) infection where SPP was found to associate with the HCMV immunoevasin US2. A decrease in SPP levels by RNA-mediated interference inhibited heavy-chain dislocation by US2, suggesting that SPP helps HCMV in an apparent protease-independent fashion to escape from immune detection. SPP also contributes to HCV infection since the intramembrane proteolysis of the HCV core protein through SPP is essential for the release from the ER membrane into the cytosol .
Model of the molecular events during proteolysis by SPP
The study by Schrul et al.  in this issue of the Biochemical Journal provides an important link to better understand the molecular events associated with proteolysis mediated by SPP activity and protein interactions modulating the function of this intramembrane protease. One of the central questions, where this paper provided some first answers, is how the selection of SPP substrates is performed and how these events are linked to the roles of SPP in dislocation from the ER and degradation. In an elegant set of overexpression experiments, the authors made use of the preprolactin-derived signal peptide as an established SPP substrate and analysed co-immunoprecipitation with wild-type SPP and a catalytically inactive SPP mutant. The mutant SPP, but not a similar mutant form of SPPL3, tightly bound to the substrate, thereby preventing its processing and degradation. Interestingly, the entire preprolactin bound both to the SPP mutant and the active protease, suggesting that such a SPP overexpression leads to an accumulation of the preproprotein independently of its proteolytic function. Also newly synthesized and misfolded proteins such as a truncated form of opsin, a G-protein-coupled receptor involved in vision, bound to both SPP and the SPP mutant. Normally, misfolded proteins are retained in the ER membrane before they are targeted to ER-associated degradation. The co-expression of SPP or of the SPP catalytically inactive mutant reduced the degradation of the misfolded opsin protein, suggesting that SPP contributes, in a protease-independent manner, to the stabilization of misfolded proteins.
It could be envisaged that other type II oriented transmembrane proteins possessing non-cleaved signal peptides behave like preprolactin. However, it could be demonstrated in similar kinds of experiments that the Ii (MHC class II-associated invariant chain) and RAMP4 (ribosome-associated membrane protein 4) neither interact with SPP nor become processed by this protease. These findings indicate that SPP requires a substrate to contain a specific overall topology as well as certain intrinsic features in their transmembrane regions.
Having determined that SPP bound to substrates and non-substrate proteins (referred to as client proteins by the authors) such as preproproteins and misfolded proteins, the question arose as to how the SPP complex is composed. Initial experiments using native electrophoretic separations revealed that both overexpressed SPP and endogenous SPP are found in rather stable high-molecular-mass complexes of 200, 400 and 600 kDa, and 250 kDa respectively. The exact protein composition of these complexes is still not clear since protein identification is technically demanding because of the low amounts obtained after immunoprecipitation. However, using co-transfection experiments combined with co-immunoprecipitation studies the preprolactin-derived signal peptide was found in the 200 kDa complex with the catalytically inactive SPP mutant, presumably representing a SPP dimer (Figure 1B). The preproprotein was predominantly found in the 600 kDa complex. In contrast misfolded proteins, such as the truncated opsin mutant were detected independently of their glycosylation status to a similar extent in all types of SPP complexes. It may well be that the smaller complexes containing SPP dimers dynamically homo-oligomerize into larger complexes thereby recruiting additional, as yet unidentified, proteins. These complexes could act as platforms for the dislocation of certain proteins from the ER membrane and the final disposal of client proteins (Figure 1C).
It remains to be determined at the molecular level how SPP can distinguish between substrate and non-substrate (client) proteins. On the basis of these experiments, it is assumed that the signal peptide plays a critical role which mediates the interaction with SPP. After cleavage of the signal peptide, it becomes a suitable SPP substrate, a process similar to the situation seen in the presenilin-mediated proteolysis where an initial cleavage event is required for intramembrane proteolysis to occur . The signal peptide alone is possibly not sufficient to determine the binding specificity to SPP. The exact structural arrangement of the transmembrane regions and the exposure of hydrophobic domains of certain type II oriented proteins may also contribute to the binding and recognition by SPP. In addition, a rapid oligomerization, as is known to occur in Ii and RAMP4, which do not bind to SPP, may prevent the interaction with SPP.
In a model, SPP may represent a site within the ER where both client and substrate proteins meet at different domains of the protease (Figure 1C). Whereas substrate proteins can enter the active site for proteolytic processing, client proteins cannot access this site, possibly because of steric hindrance by their usually rather large ectodomains. In addition specific features of the individual transmembrane segments and hydrophobic interfaces may contribute to the binding and complex formation of SPP. Such a scenario appears to be somewhat distinct from interaction studies investigating the presenilins. Possibly, the multi-protein complex nature of the γ-secretase complex allows an ordered sequential recognition, pre-selection and transfer of substrates to the catalytic core. SPP could, however, also act as a sort of scaffold to trigger the segregation of proteins destined for disposal. Such patches of membrane proteins may be subjected to degradation via autophagy within lysosomes or by the proteasome after cytosolic translocation. Although the studies by Schrul et al.  give a first and valuable insight into the complex SPP interaction network, more systematic investigations, ideally avoiding artefact-prone overexpression systems, are needed in order to clarify the exact protein-binding sites and to describe further the molecular events allowing discrimination between substrates and non-substrates as well as protein-binding partners in general. This will certainly deepen our understanding regarding the mode of catalysis of SPP and SPPLs and will help to elucidate the function of these proteins which may not be limited to their function as intramembrane proteases.
This work was supported by the Deutsche Forschungsgemeinschaft [grant number SFB 415], Interuniversity Attraction Poles Program IUAP P6/58 of the Belgian Federal Science Policy Office and DeZnit (European Union Framework Programme VI).