Site-Directed Mutagenesis and Structural Studies Suggest that the Germination Protease, GPR, in Spores of Bacillus Species Is an Atypical Aspartic Acid Protease

ABSTRACT Germination protease (GPR) initiates the degradation of small, acid-soluble spore proteins (SASP) during germination of spores of Bacillus and Clostridium species. The GPR amino acid sequence is not homologous to members of the major protease families, and previous work has not identified residues involved in GPR catalysis. The current work has focused on identifying catalytically essential amino acids by mutagenesis of Bacillus megaterium gpr. A residue was selected for alteration if it (i) was conserved among spore-forming bacteria, (ii) was a potential nucleophile, and (iii) had not been ruled out as inessential for catalysis. GPR variants were overexpressed in Escherichia coli, and the active form (P41) was assayed for activity against SASP and the zymogen form (P46) was assayed for the ability to autoprocess to P41. Variants inactive against SASP and unable to autoprocess were analyzed by circular dichroism spectroscopy and multiangle laser light scattering to determine whether the variant's inactivity was due to loss of secondary or quaternary structure, respectively. Variation of D127 and D193, but no other residues, resulted in inactive P46 and P41, while variants of each form were well structured and tetrameric, suggesting that D127 and D193 are essential for activity and autoprocessing. Mapping these two aspartate residues and a highly conserved lysine onto the B. megaterium P46 crystal structure revealed a striking similarity to the catalytic residues and propeptide lysine of aspartic acid proteases. These data indicate that GPR is an atypical aspartic acid protease.

During the first minutes of germination of spores of Bacillus species, 10 to 20% of the spore's total protein is degraded to amino acids (26). The majority of the degraded protein is a group of small, acid-soluble spore proteins (SASP), some of which (␣/␤ type; SASP-A and -C in Bacillus megaterium) bind to the spore's DNA and contribute to its resistance against heat and UV radiation (14,22,23,26). Another SASP subtype (␥ type; SASP-B in B. megaterium) does not bind spore DNA but rather, along with the ␣/␤ type SASP, serves as an important amino acid reserve in the mature spore (4). Germination protease (GPR) initiates SASP degradation by cleaving these proteins at one or two sites, depending on the SASP subtype, thereby freeing the spore's DNA and ultimately providing amino acids for metabolism and protein synthesis in spore outgrowth (26).
GPR is a sequence-specific endoprotease that is synthesized as an inactive zymogen (P 46 ) during sporulation at about the same time as its SASP substrates (26). Just prior to the completion of sporulation, GPR undergoes an autoprocessing reaction which removes an N-terminal propeptide of 5 to 15 amino acids, depending on the species (13,21). This process results in an enzyme (P 41 ) that is active in vitro but does not cleave SASP in vivo due to the relatively low water content in the core of the developing and dormant spore (7,13,21). In vitro, autoprocessing is triggered by pyridine-2,6-dicarboxylic acid (dipicolinic acid [DPA]) and decreases in pH and hydra-tion level (6). The latter two conditions mimic those within the developing spore at the time of P 46 autoprocessing, which also parallels the accumulation of an enormous depot of DPA by the developing spore (6-8, 19, 21). Analysis of GPR autoprocessing in various spo mutants is consistent with this DPA accumulation being essential for autoprocessing (19), while further in vitro studies showed that 50% dimethyl sulfoxide (DMSO) and a mildly acidic pH are sufficient to cause GPR zymogen autoprocessing (8), suggesting that autoprocessing requires some type of conformational change in P 46 .
GPR is primarily homotetrameric, both as P 46 and as P 41 , and is only active as such (12). The only detailed structural information on GPR is from the crystal structure of B. megaterium P 46 (18), and consequently the details of structural changes that occur during the P 46 to P 41 conversion are not known. Conversion of P 46 to P 41 is not accompanied by a large change in GPR's secondary structure, as the circular dichroism (CD) spectra of P 46 and P 41 are very similar (11) (see below). However, structural changes do take place in this conversion, as P 41 's sensitivity to trypsin digestion and thermal unfolding as well as the reactivity of the single cysteine residue in B. megaterium P 41 are markedly higher than with P 46 (8). An alternative explanation is that the removal of the propeptide directly causes increased dissociation of P 41 monomers by disrupting the monomer-monomer interface. However, most of GPR's propeptide is necessary neither for the maintenance of inactivity nor for structural stability, as a B. megaterium GPR zymogen variant lacking up to 10 propeptide amino acids autoprocesses normally (16).
Although GPR is clearly an endoprotease, analysis of GPR's primary sequence by the program ScanProsite (3) indicates that this enzyme does not contain any of the consensus sequences common to serine, aspartic acid, cysteine, or metalloproteases. GPR also does not appear structurally similar to the recently established glutamate/glutamine protease family, Eqolisin (2). GPR's active site has been localized to the N-terminal two-thirds of P 41 , since the trypsin-mediated removal of the C-terminal onethird of the protein does not eliminate activity (15). Site-directed mutagenesis of B. megaterium gpr has also ruled out a number of residues as contributing to catalysis, including four serines and GPR's only cysteine (15). In addition, a variety of studies on the potential metal dependence of GPR have provided no evidence that it is a metalloprotease (15), although it has been suggested that a metal ion (or ions) plays a role in substrate recognition by GPR based on structural homology modeling of GPR with another protease (17).
As indicated above, the available data are clearly not sufficient to allow the assignment of GPR to one of the five classes of proteases. We have undertaken site-directed mutagenesis of potential nucleophilic residues in GPR in order to identify the catalytic amino acids of this novel protease and provide insight into P 46 inactivity and autoprocessing.

MATERIALS AND METHODS
Bacteria and plasmids. Both the full-length B. megaterium gpr gene (gpr-46) (28) and a truncated form encoding only P 41 (gpr-41) (7) have been cloned and subsequently subcloned into the pET23a vector (Novagen/EMD Biosciences, Inc., San Diego, CA). All mutations in gpr-46 and gpr-41 were generated in the pET23a vector (Novagen) carrying the appropriate B. megaterium gpr gene. The pET23a plasmid was chosen for its C-terminal His 6 tag that allowed a straightforward and high-yield purification of the various forms of GPR by use of Ni-nitrilotriacetic acid agarose beads (see below). Plasmids were maintained in XL1-Blue Escherichia coli cells (Stratagene, Cedar Creek, TX). BL21(DE3) pLysS E. coli cells (Stratagene) were used for overexpression of P 46 and P 41 variants.
Site-directed mutagenesis. The Quickchange system (Stratagene) was used to create all mutants. The composition of PCR mutagenesis reactions run on an Applied Biosystems GeneAmp PRC system 9700 thermocycler were as follows: 5 l of 10ϫ Pfu Turbo polymerase buffer (Stratagene); 125 nM each of dATP, dTTP, dGTP, and dCTP; 125 ng of each primer; 10 ng plasmid pET23a carrying either gpr-46 or gpr-41; 1 l DMSO; H 2 O to 49 l; and 1 l (2.5 U) Pfu Turbo polymerase (Stratagene). Following PCR, the reactions were incubated with 1.5 l (9 U) of DpnI (Invitrogen, Carlsbad, CA) for 1.5 h to selectively digest the methylated parent plasmids, and the resulting PCR products were analyzed by 1% agarose gel electrophoresis. Productive reactions were transformed into XL1-Blue competent E. coli cells with selection for resistance to ampicillin (100 g/ml), and successful mutagenesis was confirmed by sequencing of plasmid DNA. Mutated plasmids were transformed into BL21(DE3) pLysS E. coli cells with selection for resistance to ampicillin (100 g/ml) and chloramphenicol (20 g/ml) for protein overexpression.
Site-directed deletion. Deletion mutations of the putative inhibitory loop in P 46 were generated by a modification of the standard Quickchange (Stratagene) procedure as described previously (29). In this modified procedure, complementary primers are constructed to contain the DNA sequence flanking the region to be deleted rather than a single base change to mediate an amino acid substitution as described above. Each primer was then added to separate reaction mixtures, and a single round of PCR was carried out. After the extension period, the reactions were pooled and the remainder of the PCR was conducted as described above.
Protein overexpression and purification. BL21(DE3) pLysS E. coli cells carrying gpr mutants were grown overnight at 37°C in 3 ml of Luria-Bertani medium (NaCl, 5 g/liter; tryptone, 10 g/liter; yeast extract, 5 g/liter; and 1 ml 1 N NaOH) containing 100 g/ml ampicillin and 20 g/ml chloramphenicol, with shaking. Fifty milliliters of Luria-Bertani medium with the latter antibiotics was inoculated with the overnight culture and grown at 37°C with shaking, and GPR expression was induced at an optical density at 600 nm of 0.7 by the addition of isopropyl-␤-D-thiogalactopyranoside to 1 mM. After incubation for a further 1.5 h at 37°C with shaking, cultures were centrifuged (10 min; 3,220 ϫ g), the pellet was resuspended on ice in 1.5 ml lysis buffer (10 mM Tris-HCl [pH 7.4], 5 mM CaCl 2 , 10% glycerol) plus 40 mM imidazole, and cells were broken by sonication in the presence of ϳ70 mg of 0.1-mm glass beads. The rest of the procedure was carried out in the cold room at ϳ4°C. The broken cells were centrifuged (10 min; 3,000 ϫ g), and the supernatant fluid was added to a 1.5-ml Ni-nitrilotriacetic acid agarose (QIAGEN, Valencia, CA) column equilibrated with lysis buffer plus 40 mM imidazole. The column was washed with 5 ml lysis buffer plus 40 mM imidazole and then eluted with 5 ml lysis buffer plus 50 mM imidazole and finally with 5 ml lysis buffer plus 150 mM imidazole. The last two eluates were combined and concentrated to 1 ml in a 15-ml Centriprep centrifugal filter unit (10 kDa molecular mass cutoff) (Millipore, Bedford, MA), and the concentrated protein was dialyzed overnight against lysis buffer without imidazole.
Activity and autoprocessing assays. GPR activity was assayed on a mixture of SASP partially purified from B. megaterium spores by dry rupture in a dental amalgamator (Wig-L-Bug), followed by extraction of SASP with cold 1% acetic acid, dialysis against cold 1% acetic acid, and passage through a DEAE-cellulose column in 1% acetic acid as described previously (25). P 41 activity was determined by incubating GPR (routinely 540 nM) with B. megaterium SASP (2.25 mg/ml) in 40 l 50 mM Tris-HCl (pH 7.4), 5 mM CaCl 2 , 10% glycerol for 30 min at 37°C and monitoring SASP degradation by polyacrylamide gel electrophoresis (PAGE) at low pH as described previously (24) (see Results). Activity against SASP was also assayed at higher concentrations of GPR variants to rule out low levels of residual activity. Autoprocessing of the GPR zymogen was carried out at a GPR concentration of 4 M in 20 l of 100 mM 2-[N-morpholino]ethanesulfonic acid (pH 6.2), 5 mM CaCl 2 , 10% glycerol, 50% DMSO (8). Samples were incubated at 30°C overnight, diluted in 2ϫ sodium dodecyl sulfate (SDS)-PAGE loading buffer (24), and analyzed by 8.5% SDS-PAGE to resolve P 46 and P 41 (6,8,21) (see Results). These conditions allowed for detection of ϳ3% autoprocessing.
Analytical methods. CD spectroscopy was done with a Jasco spectropolarimeter (J-715) at 20°C and a protein concentration of 10 M in lysis buffer. Each spectrum was obtained by averaging three scans between 190 and 270 nm, and raw data were converted to molar ellipticity. Multiangle laser light scattering (MALLS) was detected with a TriStar MiniDAWN detector (Wyatt Technology, Santa Barbara, CA) by using the buffer system 10 mM Tris-HCl (pH 7.4)-5 mM CaCl 2 and a Superdex 200 10/300 gel filtration column (Amersham Biosciences, Piscataway, NJ).

RESULTS
Site-directed mutagenesis. B. megaterium GPR was chosen for this mutagenesis study because the genes encoding both the P 46 and P 41 forms of this protein have been cloned (28), the purified proteins have proven amenable to in vitro studies (8), and the crystal structure of B. megaterium P 46 has been solved (18). Amino acids in B. megaterium GPR were chosen for change if they fit three criteria. Each altered residue must (i) be a potential nucleophile (arginine, asparagine, aspartate, glutamate, glutamine, histidine, lysine, serine, threonine, or tyrosine), (ii) be conserved among 23 spore-forming bacteria for which there are gpr sequence data, and (iii) not be in regions of GPR whose removal does not eliminate GPR activity (Fig. 1). The latter criterion eliminated some conserved and potentially nucleophilic residues in the C-terminal one-third of the protein, as they can be removed from B. megaterium P 41 without loss of activity, although the C-terminally truncated protein is unstable (15). Similarly, the first 21 amino acids of P 46 can be removed without loss of P 41 activity (11). Some residues that appeared to be good candidate nucleophiles were also changed even if they were not completely conserved, to protect against missing an important residue as the result of a genome sequencing error. Residues selected for variation were changed to amino acids that were expected to maintain the general shape and, where possible, the polarity of the original residue while not supporting nucleophilic attack. Multiple amino acid substitutions at a single position were employed in two cases. K223 was changed to alanine, glutamine, histidine, and methionine, while D127 was changed to alanine, glutamate, glutamine, and serine.
The aspartate at position 127 was altered to multiple residues because the initial substitution, D127N, resulted in disruption of the variant's quaternary structure (Table 1). D127's importance in GPR's activity only became evident once D127A and D127E were generated, as these variants were well structured, unable to autoprocess as P 46 , and inactive against SASP as P 41 . Of particular note is the result that substituting a glutamic acid residue for the aspartate at position 127 resulted in a well-structured and inactive variant ( Table 1), suggesting that this position is both essential for catalysis and exquisitely sensitive to structural perturbations.
Activity and autoprocessing of GPR variants. Multiple lines of evidence support the hypothesis that the processing of P 46 to P 41 is an autocatalytic event and that the same catalytic site mediates both autoprocessing and activity against SASP (1, 6, 7). It was, however, possible that a given variant could be active against SASP as P 41 but be unable to autoprocess as P 46 , or vice versa. Consequently, variants were assayed for both the ability to autoprocess as P 46 and for activity against SASP as P 41 (Table 1; Fig. 2A and B). Note that SASP-A and SASP-C comigrate when run on a polyacrylamide gel at low pH ( Fig.  2A). In Fig. 2B, more protein was loaded in lanes 1 and 2 than in lanes 3 to 6 to show that there was no autoprocessing in the absence of DMSO. Loading less protein revealed only P 46 and no P 41 (data not shown).
While all variants capable of autoprocessing as P 46 were active against SASP as P 41 , there were some variants capable of cleaving SASP when expressed as P 41 but unable to autoprocess (Table 1). Further analysis of these variants is beyond the scope of the current work, but it is likely that the amino acid substituted in each of these variants is located in a position important for the conformational changes necessary for autoprocessing and so may provide useful fodder for future study.
Of the 27 residues altered in this work, only the substitution of D127 and D193 resulted in a P 41 that was inactive against SASP and a corresponding P 46 that was unable to autoprocess to P 41 (Table 1). These data provide further strong support for the hypothesis that GPR utilizes a single catalytic site to mediate both its autoprocessing and activity against SASP and, further, that D127 and D193 are GPR's only catalytically essential amino acids.
CD spectroscopy. While the inactivity of variants resulting from the substitution of D127 and D193 suggested that these residues are essential for P 46 autoprocessing and P 41 activity against SASP, it was possible that these amino acid substitutions rendered the enzyme inactive by disrupting its threedimensional structure rather than by replacing a catalytically essential residue. To address this issue, CD spectroscopy was carried out to determine the overall secondary structure of these inactive variants in comparison to that of wild-type GPR. CD spectra were also obtained for a number of active GPR variants to ensure that simply changing a noncatalytically essential residue did not alter the protein's CD spectrum (Table  1; Fig. 3). It was clear from these data that no amino acid substitution in either P 46 or P 41 noticeably disrupted the protein's overall secondary structure, although deletion of the putative inhibitory loop may have caused misfolding of secondary structural elements (Table 1). This analysis does not rule out local perturbations in structure or even large changes in regions of the protein lacking defined secondary structure. However, it does provide support for the hypothesis that altering D127 or D193 results in nonautoprocessing, inactive GPR variants because these are catalytically essential residues and not because they are important in folding of GPR's secondary structural elements.
MALLS. CD spectroscopy addressed the possibility that an amino acid substitution might effect folding of secondary structural elements of GPR, thereby resulting in an inactive variant. Another structural consideration important for GPR activity is its oligomeric state. Previous work has shown that P 46 and P 41 are active only as tetramers (12,24), and so it is important to address the possibility that an amino acid substitution might cause a variant to be inactive against SASP and unable to autoprocess because of disruption of quaternary structure.
Use of gel filtration and particle size determination by MALLS showed that wild-type P 46 was nearly 100% tetrameric, while P 41 was largely tetrameric but also exhibited a significant percentage of lower-molecular-mass species (Table  1; Fig. 4). These results are consistent with previous data that have suggested P 41 is less stable and more prone to dissociation than its zymogen precursor (8,15). The data collected for the variants subjected to this analysis indicated that the oligomerization of GPR is sensitive to amino acid substitution, as at least one P 41 variant (D127N) formed little tetramer while others (e.g., D127E and D193N) were more tetrameric than wild-type P 41 (Table 1). However, the hypothesis that alteration of D127 and D193 each cause GPR inactivity by virtue of their being catalytically essential is supported by these data, since variants at both positions (D127E and D193N, both as   FIG. 1. Residues changed in B. megaterium GPR. The amino acid sequence of B. megaterium P 46 is shown, with every tenth position indicated by a dot below the corresponding amino acid given in the single-letter code. Altered amino acid positions are represented by large, bold letters, and the catalytically essential aspartic acid residues (D127 and D193) are highlighted and boxed in red. The putative inhibitory loop (residues 213 to 227) is surrounded by apposed arrows, and K223 is highlighted and boxed in blue. Arrow a indicates the peptide bond cleaved in the conversion of P 46 to P 41 , and arrow b indicates the peptide bond cleaved when P 41 undergoes a further autoprocessing reaction cleaving P 41 to P 39 , which is still active against SASP. The underlined amino acids comprise GPR's autoprocessing recognition sequence. Arrow c indicates the site of trypsin cleavage that removes the C-terminal one-third of the protein while not affecting P 41 's activity against SASP.  Fig. 4). Three-dimensional analysis of D127, D193, and surrounding structures. While site-directed mutagenesis identified D127 and D193 as the only two amino acids essential for GPR's autoprocessing and activity against SASP, these data needed to be checked against GPR's three-dimensional structure. D127 and D193 were, therefore, mapped onto the crystal structure of P 46 and found to be in sufficiently close proximity to participate together in catalysis ( Fig. 5B and Fig. 6). The argument for classification of GPR as an aspartic acid protease is further strengthened by the three-dimensional relationship of D127 and D193 with a nearby lysine (K223), since the relative orientations of these three residues are strongly reminiscent of many aspartic acid protease zymogen active sites, particularly that of pepsinogen (5). Figure 5 illustrates the relative orientations of the catalytic aspartic residues of pepsinogen (27) (Fig. 5A) and GPR (18) (Fig. 5B).

DISCUSSION
Previous work has determined that GPR is a sequencespecific tetrameric endoprotease but has failed to identify its catalytic residues. Two previous modeling studies have suggested very different residues as involved in GPR catalysis (9,17). The first study utilized results of molecular docking simulations to suggest that three glutamic acid residues (E19, E26, and E202) take part in binding and hydrolysis of SASP (9). One of these (E19) had previously been ruled out as a residue essential for catalysis, as its removal by trypsin did not affect the activity of B. megaterium P 41 (15). The current work shows that the P 41 variants E26Q and E202Q also are fully active (Table 1).
Another group has suggested, based on analysis of threedimensional homology, that aspartate residues D127 and D193 and lysine K223 contribute to substrate recognition by coordinating a divalent metal ion ostensibly provided by SASP (17). This hypothesis was consistent with the fact that D127, D193, and K223 have a similar three-dimensional relationship with the surrounding protein fold and with one another, compared with the two acidic residues and a histidine responsible for metal-mediated substrate recognition and cleavage by bacterial hydrogenase maturation protease (HybD) (17). There are, however, at least four lines of evidence inconsistent with this hypothesis. There is no evidence that SASP binds any metal (P. a Assays of activity, autoprocessing, CD spectroscopy, and MALLS on wild-type P 41 and P 46 and all variants were done as described in Materials and Methods. b Active, no significant difference (Ϯ 20%) in activity relative to that of wild-type P 41 . Inactive, at least 80-fold less active than wild-type P 41 . Partially active, Ϸ50% activity relative to that of wild-type P 41 .
c Folded, no significant difference between a variant's CD spectrum and that of the corresponding wild-type GPR (P 41 or P 46 ) (see Fig. 3). Misfolded, significant difference between a variant's CD spectrum and that of wild-type GPR.
d Data are reported as ratios of the percent tetramer (T) to dimer (D) to monomer (M). These values may not add up to 100% because most proteins did not completely give well-defined populations of homogeneous particle size. (See Fig. 4.) e Active, generation of P 41 after overnight incubation under autoprocessing conditions. Inactive, no P 41 generated. Wild-type P 46 typically autoprocesses to 50% completion overnight, and the sensitivity of this assay allows detection of Ϸ3% autoprocessing. Inactive/inactive, P 46 variant was unable to autoprocess to P 41 and was inactive against SASP as P 46 .
f ND, not done. Setlow, unpublished data), and GPR's activity is not inhibited by metal ion chelators nor have metal ions been found associated with either P 46 or P 41 , other than the Ca 2ϩ used to stabilize the protein (15). Finally, altering K223 to multiple residues (alanine, glutamate, histidine, and methionine) or simply deleting it had no effect on GPR's activity against SASP or autoprocessing (Table 1). While GPR is an aspartic acid protease by virtue of its two catalytically essential aspartates, at least four characteristics set it apart from other members of this protease class. First, GPR's primary amino acid sequence is not homologous to any known protein. Second, three-dimensional homology modeling suggests that GPR has a fold similar to HybD (17) but not to any aspartic acid protease for which there is an atomic resolution structure. Third, GPR is a homotetramer with each subunit containing a full catalytic site. This is in contrast to other aspartic acid proteases, most of which are either homodimers, where each monomer provides one catalytic aspartate, or monomers containing a single active site. Finally, and perhaps most interestingly, GPR appears to differ significantly from all known proteases in the way it maintains zymogen inactivity. While there are multiple mechanisms utilized by proteases to maintain zymogen inactivity (reviewed in reference 11), P 46 appears to make use of a novel variation on at least one, and possibly two, common themes.
Many proteases, including some aspartic acid proteases, maintain zymogen inactivity by placing their propeptide over the future active site, thereby sterically blocking substrate access and rendering the protease inactive (e.g., pepsinogen). Another common method of maintaining zymogen inactivity is to hold the active-site residues in a conformation incompatible with enzymatic activity while leaving the active site accessible (e.g., proplasmepsin [10]). The crystal structure of P 46 suggests that GPR maintains inactivity by at least one, and possibly both, of these strategies. Figures 1 and 5 illustrate the relative positions of GPR's propeptide, catalytic aspartic acid residues, and the putative inhibitory loop in the primary amino acid sequence and in three-dimensional space, respectively. Consistent with previous propeptide deletion analysis (16), the crystal structure of P 46 suggests that the majority of the propeptide does not interact with the rest of the protein in general or with the active site in particular. Rather, examination of the P 46 structure suggests that a loop composed of residues 213 to 227 physically blocks access to GPR's catalytic aspartic acid residues. Further, K223 appears to interact closely with both putative catalytic aspartic acid residues in a manner reminiscent of pepsinogen's structure (5), and measurement of the distance between C ␦ of D127 and that of D193 (4.4 Å) indicates that they may be held in an inactive conformation. However, this determination is complicated by the low resolution of the P 46 crystal structure (crystals diffracted to 3.0 Å [18]). Finally, the putative inhibitory loop of GPR has a pI significantly higher than that of the protein as a whole (10.0 and 4.9, respectively) (20). These latter values are in good agreement with that of pepsinogen's propeptide compared to that of pepsinogen as a whole (pI values for the propeptide and whole protein of 10.3 and 4.2, respectively). Given these structural considerations and similarities between GPR's putative inhibitory loop and pepsinogen's propeptide, we hypothesize that the loop formed by residues 213 to 227 functions to maintain GPR zymogen inactivity by steric occlusion. We further hypothesize that the conditions that induce P 46 to autoprocess to P 41 cause a conformational change such that this loop no longer fully inhibits the catalytic residues, allowing the propeptide cleavage site to swing down into the active site and thereby mediating self-cleavage. An examination of the P 46 structure (Fig. 5) lends support to this hypothesis as follows: (i) the putative inhibitory loop does not contain any secondary structural elements, and so it could change conformation without altering GPR's CD spectrum; (ii) this loop's ends are in close proximity to each other, suggesting that the loop could swing away from the surface of the protein, exposing GPR's active site without disrupting structural elements in close proximity along the polypeptide backbone; and (iii) D127 and D193 are ϳ15 Å from the autoprocessing cleavage site, putting them within striking distance to mediate autocleavage, as this distance is comparable to the analogous interresidue distances of pepsinogen (ϳ20 Å). The alternative hypothesis that one monomer's propeptide is cleaved by a neighboring monomer is unlikely, given the large distance between one monomer's active site and another monomer's propeptide cleavage site. Specifically, the active site of any monomer relative to another monomer's propeptide cleavage site is (i) spatially separated (30 to 45 Å, depending on the monomer pair examined) and (ii) structurally restrained from traversing this large distance separating them (Fig. 6).
Additionally, previous work is consistent with the N terminus of GPR being able to interact with its active site, since incubation of P 41 under autoprocessing conditions results in a further autocatalytic cleavage reaction that removes seven additional amino acids, producing a smaller species termed P 39 (7) (Fig. 1).
An inhibitory loop such as the one hypothesized here is apparently unprecedented among aspartic acid proteases in that it is not part of the zymogen's propeptide but rather remains integral to the enzyme's polypeptide backbone following autoprocessing. If the loop were shown to maintain P 46 inactivity, it would set GPR apart from all known aspartic acid proteases and, indeed, perhaps even from all known proteases in general. To investigate this possibility in more detail, variants of this putative inhibitory loop were generated as both P 46 and P 41 and assayed for the ability to autoprocess and for activity against SASP. However, none of these variants were found to autoprocess or to be active against SASP, most likely due to misfolding as judged by CD and MALLS (Table 1).
GPR's catalytic residues have now been identified as D127 and D193. Questions still remain, however, regarding accessory active site amino acids, residues important for autoprocessing but not activity, and the mechanistic details of GPRmediated proteolysis of SASP and P 46 autoprocessing. A full understanding of these issues will likely need to await atomicresolution structure determination of P 41 in combination with biochemical studies.