Chemical rescue of I-site cleavage in living cells and in vitro discriminates between the cytomegalovirus protease, assemblin, and its precursor, pUL80a.

Chemical rescue is an established approach that offers a directed strategy for designing mutant enzymes in which activity can be restored by supplying an appropriate exogenous compound. This method has been used successfully to study a broad range of enzymes in vitro, but its application to living systems has received less attention. We have investigated the feasibility of using chemical rescue to make a conditional-lethal mutant of the cytomegalovirus (CMV) maturational protease. The 28-kDa CMV serine protease, assemblin, has a Ser-His-His catalytic triad and an internal (I) cleavage site near its midpoint. We found that imidazole can restore I-site cleavage to mutants inactivated by replacing the critical active site His with Ala or with Gly, which rescued better. Comparable rescue was observed for counterpart mutants of the human and simian CMV assemblin homologs and occurred in both living cells and in vitro. Cleavage was established to be at the correct site by amino acid sequencing and proceeded at approximately 11%/h in bacteria and approximately 30%/h in vitro. The same mutations were unresponsive to chemical rescue in the context of the assemblin precursor, pUL80a. This catalytic difference distinguishes the two forms of the CMV protease.

sequential cleavage at its maturational (M) site and then its release (R) site to yield a mature form, called assemblin (10,23,24). Among the CMV-type herpesviruses (members of the ␤-herpesvirus family, Ref. 25), the assemblin homolog is additionally cleaved at an internal (I) site (15,26) (e.g. see Fig. 1D), converting it to a two-chain form that remains active (27)(28)(29). Although these self-processing steps are coupled with capsid maturation, little is understood about their regulation.
To further investigate this enzyme and the relationship of its selfprocessing to capsid assembly and maturation, we considered creating a mutant controlled by a rationally designed small-molecule switch. Such "chemical rescue" mutants have an inactivating amino acid substitution that can be functionally compensated by an exogenous compound (30,31). Given that two catalytic triad members of the herpesvirus protease are histidines and that their imidazole group is well tolerated by cells as a free compound (32), we targeted these residues for chemical rescue. Because the goal of creating a virus carrying this mutation requires its growth and chemical rescue in live cells, we began our study using a cell-based transfection assay system.
Our approach was to express specific mutants of assemblin and its precursor in transfected mammalian cells and in transformed bacteria, establish whether their proteolytic activity could be restored by imidazole added to the culture medium, and reproduce the effect in vitro. Our results show that imidazole enabled I-site cleavage of mutant assemblin in living cells and in vitro but that the effect was not seen when the same mutations were tested in the assemblin precursor. Although this finding challenges the feasibility of creating an imidazole-regulated protease mutant of CMV, it reveals interesting and potentially important structural and/or enzymatic differences between assemblin and its precursor.

EXPERIMENTAL PROCEDURES
Cells, Plasmids, and Expression-Human endothelial kidney (HEK, line 293; American Type Culture Collection, Manassas, VA) carcinoma cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (HyClone, Logan, UT) and 100 units each of penicillin and streptomycin per ml. HEK cells growing at ϳ70% confluence in 24-well plates (Falcon Multiwell, BD Biosciences) were transfected using FuGENE 6 (Roche Applied Science) with RSV.5(neo) expression plasmids according to the manufacturer's instructions. Transfected cells were collected into 75 l of protein sample buffer and frozen at Ϫ80°C until analyzed.
Plasmids encoding wild-type simian CMV (SCMV) assemblin, its precursor (pPR), and mutants S118A, H47A, H47Q, and H142A have been described elsewhere (10,24). The H47G mutants of SCMV assemblin and pPR were made by site-directed mutagenesis (His-47 codon CAC 3 GGC; QuikChange Kit, Stratagene, La Jolla, CA) followed by subcloning a small mutation-containing fragment of the PCR product into plasmids encoding the wild-type SCMV proteins (10). The corresponding HCMV assemblin and pPR mutants were made similarly, by changing the His-63 codon from CAC to GCC (H63A) or to GGC (H63G) and subcloning the mutant fragment into plasmids encoding wild-type HCMV assemblin or pPR (28). The intended mutant sequence of all constructs was verified by the Johns Hopkins University Biosynthesis and Sequencing Facility.
Bacteria (BL21(DE3)) were transformed with the appropriate plasmid and induced for protein expression with 1 mM isopropyl 1-thio-␤-D-galactopyranoside. Following overnight (ϳ18 h) incubation at 16°C in the presence or absence of 80 mM imidazole, the bacteria were collected by centrifugation and further processed as specified.
Recovery of Mutant Assemblin by Affinity Chromatography-In preparation for immobilized metal affinity chromatography (IMAC), bacterial pellets collected (6,000 ϫ g for 20 min at 4°C) from 100-ml cultures were suspended in 10 ml of lysis buffer (50 mM NaPO 4 , 300 mM NaCl, pH 8.0) and ruptured using a French press (ϳ14,000 p.s.i.), and the lysates were clarified by ultracentrifugation (100,000 ϫ g for 30 min at 4°C).
H63G mutant assemblin tagged with an amino-terminal His/Trp purification handle (MHWHWHW-) (33) was affinity-selected from the clarified lysate onto nickel-nitrilotriacetic acid Superflow beads (Qiagen, Valencia, CA), washed with lysis buffer, and eluted using 50 mM EDTA (to avoid exposure to imidazole). The two or three 0.5-ml fractions containing the most H63G were combined and concentrated 3-fold by centrifugal filtration (catalog no. 3491.2, Orbital Biosciences, LLC., Topsfield, MA), leaving the protein in the same elution buffer or exchanging the elution buffer with cleavage buffer (500 mM Na 2 SO 4 , 10 mM dithiothreitol, 10% glycerol, 100 mM MOPS, pH 7.2), as specified.
In vitro cleavage assays contained IMAC-recovered protein in 1ϫ cleavage buffer, 50 mM EDTA, and 80 mM imidazole, each adjusted to pH 8.0. Optimization experiments showed that EDTA in the range of 2-50 mM had little effect on the reaction, imidazole in the range of 20 -100 mM gave Ն15% better rescue than higher or lower concentrations, and incubation at 16°C or room temperature (ϳ24°C) gave better cleavage than at 37 or 45°C.
SYPRO-R-stained proteins were detected by transillumination at 300 nm and recorded and quantified with a Kodak EDAS290 or Gel Logic 200 imaging system (Eastman Kodak Co., Rochester, NY).
Amino Acid Sequence Analysis-In preparation for amino acid sequencing, proteins were separated by SDS-PAGE in a 4 -12% gradient gel and electrotransferred to a polyvinylidene difluoride membrane, as described above. Proteins of interest were located by their contrast with the white membrane after it had been dried and then wetted in 10% methanol, 7% acetic acid. The A c bands in several adjacent lanes containing replicate samples were cut from the membrane and subjected to sequencing by automated Edman degradation (model 492 protein sequencer, Applied Biosystems, Inc., Palo Alto, CA).

RESULTS
Experiments to test the toxicity of imidazole showed that HEK cells grow for at least 3 days in medium containing 40 mM imidazole and remain adherent for 3 h in 80 mM imidazole and that isopropyl 1-thio-␤-D-galactopyranoside-induced bacteria tolerate 80 mM imidazole for at least 24 h at 16°C. I-site Cleavage by His-47 Mutants of Simian CMV Assemblin Rescued by Imidazole-As an initial screen for the effect of imidazole on assemblin activity, wild-type simian CMV assemblin was compared with previously described mutants having an Ala substitution for one of the catalytic triad residues, Ser-118, His-47, or His-142 (24). One day after transfection, the maintenance medium was replaced with fresh medium or with fresh medium containing imidazole. Two days later the medium was removed, and the cells were solubilized in protein sample buffer 2 and subjected to SDS-PAGE followed by Western immunoassay to detect assemblin and its cleavage products.
In the absence of imidazole the phenotypes were as expected (24): (i) wild-type assemblin (PR n ) was partially cleaved to its products A n ϩ A c (Fig. 1A, lane 1, and D); (ii) mutation of the essential serine nucleophile (S118A) or the critical histidine (H47A) prevented I-site cleavage (Fig.  1A, lanes 3 and 5); and (iii) mutation of His-142 (H142A) had comparatively little effect on I-site cleavage (Fig. 1A, lane 7). The presence of imidazole had essentially no qualitative effect on the I-site cleavage of wild-type assemblin or the S118A and H142A mutants (Fig. 1A, lanes 2, 4, and 8). Encouragingly, however, when H47A was expressed with imidazole, the A n and A c fragments diagnostic of I-site cleavage were detected ( Fig. 1, A, lane 6, and C, lane 4), consistent with chemical rescue of that mutation.
Because the nature of the substituting amino acid can influence the extent to which activity is restored (37), two additional substitutions for His-47 were tested for comparison. The bulkier glutamine (H47Q) partially restored I-site cleavage but with apparently reduced specificity. In the absence of imidazole it enabled I-site cleavage (i.e. presence of A n and A c ) and additional cleavages resulting in a group of proteins less evident in the other preparations (Fig. 1A, lane 9; indicated by brace). This pattern was similar in the presence of imidazole (Fig. 1A, lane 10), and the H47Q mutant was not analyzed further. The smaller glycine substitution (H47G), like H47A, was inactive without imidazole (Fig. 1B, lane 1). However, in the presence of imidazole, I-site cleavage by the H47G mutant dramatically exceeded that of the H47A mutant ( Fig. 1, B, lanes 2 and 3, and C, compare lane 4 with lane 5). Calculations based on phosphorimaging measurements indicate that 36% of wild-type assemblin, 3% of the H47A mutant, and 28% of the H47G mutant was cleaved to A n ϩ A c in the presence of imidazole. Thus, the percentage of assemblin cleaved by the H47G mutant was nearly 10-fold greater than that of the H47A mutant and 78% greater than that of the wild-type SCMV enzyme.

I-site Cleavage by His-63 Mutant of Human CMV Assemblin
Rescued by Imidazole-To extend this finding to the human CMV (HCMV) homolog, with which most structural and enzymatic characterizations have been performed, we tested the counterpart mutants of HCMV assemblin. Because HCMV assemblin expresses less well than its SCMV homolog in transfected mammalian cells (28), it was expressed in bacteria, where relatively high concentrations are obtained routinely (15,26).
Cultures of transformed cells were induced to express either wildtype HCMV assemblin or the H63A and H63G mutants, and duplicate samples of each were incubated overnight at 16°C with or without imidazole. The bacteria were then collected, solubilized, and subjected to SDS-PAGE and protein staining. As expected, wild-type HCMV assemblin yielded the A n and A c products of I-site cleavage (15, 26), whether or not imidazole was present ( Fig. 2A, lanes 1 and 6). Consist-FIGURE 1. SCMV H47A responds to imidazole rescue of I-site cleavage. A, HEK cells were transfected with plasmids encoding wild-type (Wt) SCMV assemblin or the mutants S118A, H47A, H142A, and H47Q. Nontransfected cells (Mock) were included as controls. The cells were treated with no (0 mM) or 20 mM imidazole (Imid), solubilized, and subjected to SDS-PAGE and Western immunoassay using antisera specific for the amino (N2) and carboxyl (C2) amino acid sequences of SCMV assemblin. A collage prepared from a phosphorimage of the resulting membrane is shown. Brace to right of lane 10 indicates group of proteins described under "Results." Abbreviations to left of collage are for assemblin (PR n ) and its amino (A n ) and carboxyl (A c ) cleavage fragments (see D). B, shown here is a phosphorimage of a Western immunoassay performed, as described for A, on lysates of HEK cells transfected with a plasmid encoding mutant H47G-assemblin. C, wild type and several mutants of SCMV assemblin were expressed in HEK cells with 20 mM imidazole in the culture medium. Lysates of the cells were prepared and subjected to SDS-PAGE and Western immunoassay, as described for A. Shown here is a collage of the lanes of interest prepared from a phosphorimage of the processed membrane. Abbreviations are as in A. D, schematic of simian CMV assemblin (PR n ) and internal (I)-site cleavage products A n and A c . Protein molecular weights (ϫ10 Ϫ3 ) are in parentheses; the amino (N2) and carboxyl (C2) peptide sequences used to prepare antisera are indicated as shaded boxes. Positions of the catalytic triad amino acids and the internal and release (R) cleavage sites (arrows) are also indicated. FIGURE 2. I-site cleavage by HCMV H63G mutant restored by imidazole. A, bacteria expressing wild-type (Wt), H63A, or H63G HCMV assemblin were grown in the presence (ϩ) or absence (Ϫ) of 80 mM imidazole, collected, and analyzed by SDS-PAGE followed by staining with SYPRO-R. Protein molecular weight (ϫ10 Ϫ3 ) markers are shown to the right of lane 7; abbreviations in the left margin are for specific fragments derived from HCMV assemblin (PR n ) by cleavage at its I and C sites, as shown in C. B, the H63A (A) and H63G (G) mutants were expressed in bacteria with (ϩ) or without (Ϫ) 80 mM imidazole. Bacteria collected from 1 ml of each culture were solubilized and analyzed in lanes 1-4. H63G assemblin was recovered from a 100-ml culture by IMAC. 50 l of the preparation was made 98 mM in imidazole by adding 1 l of 5 M imidazole, and 50 l was treated in parallel without imidazole. After incubation for ϳ20 h, each sample was combined with protein sample buffer 2 and subjected to SDS-PAGE followed by staining with SYPRO-R (lanes 5 and 6) and quantification, all as described under "Experimental Procedures." Molecular weight markers and abbreviations are as described for A and C. C, schematic of HCMV assemblin showing internal (I), cryptic (C), and release (R) cleavage sites and the resulting fragments and their molecular weights (ϫ10 Ϫ3 ) in parentheses. Catalytic triad residues are His-63, Ser-132, and His-157. ent with the transfection experiments performed with SCMV assemblin, neither of the His-63 mutants showed I-site cleavage in the absence of imidazole ( Fig. 2A, lanes 2 and 3), and there was little cleavage by the H63A mutant even with imidazole added (Fig. 2A, lane 5). In contrast, the H63G mutant expressed with imidazole yielded readily detected A n and A c bands, consistent with restored I-site cleavage (Fig. 2A, lane 4). Quantification of the stained proteins ( Fig. 2A, lanes 4 and 6) indicated that 34% of the H63G mutant and 72% of the wild-type assemblin was cleaved to A n ϩ A c with imidazole present.
The band just below wild-type assemblin ( Fig. 2A, lanes 1 and 6) is likely to be the C n product of cleavage at the cryptic (C) site ( Fig. 2C; reactive with anti-A n and co-migrates with cloned C n during SDS-PAGE, data not shown) (23), but this remains to be verified by sequence analysis. It accounted for ϳ9% of the total assemblin-related protein (i.e. PR n ϩ C n ϩ A n ϩ A c ). A discernable band at the same position in the preparation of H63G expressed with imidazole suggests that there may also be some rescue of C-site cleavage (e.g. Fig. 2A, lane 4, white dots). The origin of the band migrating close to the 6-kDa marker in both preparations of wild-type assemblin is unknown; however, sequence analysis established that it has the same amino-terminal sequence as A c (data not shown), compatible with it being the 7.2-kDa A c Ј fragment predicted if assemblin were cleaved at both its I and C sites (Fig. 2C). It was not detected in any of the H63A or H63G preparations ( Fig. 2A,  lanes 2-5).
Sequence Confirmation of I-site Cleavage-Bacterial expression of the HCMV protease gave sufficient amounts to confirm the cleavage site by direct sequence analysis. Lysates of bacteria expressing wild-type HCMV assemblin or the H63G mutant in the presence of imidazole were subjected to SDS-PAGE in five adjacent lanes each. The A c bands were located and subjected to amino acid sequence analysis, all as described under "Experimental Procedures." Both the wild-type and H63G A c fragments gave the predicted, unambiguous NЈ-ATSLSG sequence, establishing that cleavage resulting from imidazole rescue is at the I site, VEA2ATSLSG.
Imidazole Rescues I-site Cleavage by H63G Mutant of Human CMV Assemblin in Vitro-Rescue of the HCMV H63G mutation by imidazole was also tested in vitro. Bacterially expressed H63G assemblin was affinity-selected from a clarified cell lysate by IMAC, and the bound protein was displaced from the beads with EDTA in lysis buffer, which was replaced with cleavage buffer by centrifugal filtration, as described under "Experimental Procedures." Half of the sample was incubated with 98 mM imidazole and half without for ϳ20 h at room temperature, and the samples were subjected to SDS-PAGE followed by protein staining. For comparison, bacteria expressing the H63G or H63A mutants, in the presence or absence of imidazole, were directly solubilized in protein sample buffer 2 and analyzed in parallel.
An image of the resulting stained gel showed little, if any, I-site cleavage of affinity-selected H63G in the absence of imidazole (Fig. 2B, lane  6) but readily detected A n and A c cleavage products with imidazole added (Fig. 2B, lane 5). The bacterial lysates included for reference gave results similar to those in the preceding experiment: no I-site cleavage for either mutant in the absence of imidazole, little if any cleavage for the H63A mutant even with imidazole, and obvious I-site cleavage by the H63G mutant in the presence of imidazole (Fig. 2B, lanes 1, 2, 4, and 3, respectively). Calculations based on measurements made from this gel indicate that, in the presence of imidazole, 46% of H63G assemblin was cleaved to A n ϩ A c in bacteria and 30% in vitro.
Time Course of Chemical Rescue in Bacteria and in Vitro-The rate of I-site cleavage resulting from imidazole rescue in bacteria was determined by measuring the appearance of the A n and A c fragments after adding imidazole to an isopropyl 1-thio-␤-D-galactopyranoside-in-duced culture that had been grown overnight at 16°C. At the indicated times after adding imidazole, samples were removed from the culture and volumetrically adjusted, according to their A 600 to have the same amount of bacteria as the t ϭ 0 sample. The bacteria were then collected by centrifugation (16,000 ϫ g, 2 min, room temperature), solubilized in protein sample buffer 1, heated in a boiling water bath for 3 min, and subjected to SDS-PAGE followed by staining with SYPRO-R. Quantification of the progressive decrease in assemblin and increase in A n ϩ A c , evident over the time course (Fig. 3A), showed that cleavage was fastest (ϳ11%/h) and approximately linear from 0 to 60 min and then slowed to ϳ1%/h between 2 and 24 h, at which point ϳ47% of the assemblin had been cleaved (Fig. 3C).
We also determined the time course of chemical rescue in vitro. H63G assemblin was prepared by IMAC and exchanged into cleavage buffer, as described under "Experimental Procedures." 200 l of the resulting H63G preparation was made 98 mM in imidazole by adding 4 l of a 5 M stock, and the reaction was continued at room temperature for 3 days. At the times indicated in Fig. 3B, 20-l samples were removed from the incubation mixture, combined with 20 l of protein sample buffer 1, and frozen at Ϫ80°C until analyzed. The samples were further diluted with 20 l of protein sample buffer, heated 3 min in a boiling water bath, and subjected to SDS-PAGE followed by protein staining with SYPRO-R (Fig. 3B). Calculations of the relative amounts of assemblin and A n ϩ A c at each time point showed that 36% of the assemblin

. Time course of I-site cleavage by HCMV H63G assemblin (PR n ) treated with imidazole in bacteria and in vitro.
A, H63G assemblin was expressed overnight at 16°C in an 80-ml culture of bacteria. Half of the culture was then made 80 mM in imidazole, half was left untreated (data not shown), and incubation was continued at 16°C. Samples were taken from both cultures at the time points indicated, and a normalized amount of bacteria from each was subjected to SDS-PAGE. Shown here is an image of the resulting gel after protein staining with SYPRO-R. Abbreviations to the right of the image are as in Fig. 2A. B, IMAC-recovered H63G assemblin was made 98 mM in imidazole, and samples of the reaction mixture were taken at the indicated times. Shown here is an image of the SYPRO-R-stained proteins following SDS-PAGE. C, the percentage of assemblin cleaved to A n and A c in bacteria was calculated from the gel shown in A, as the ratio of (A n ϩ A c )/(Asbln ϩ A n ϩ A c ). D, the percentage of assemblin cleaved to A n and A c in vitro was calculated from the gel shown in B as described above.
had been cleaved to A n ϩ A c within the first hour, 77% after 24 h, and 86% after 3 days (Fig. 3D; day 2 and 3 data points not shown).
To confirm that the cleavage observed with in vitro chemical rescue was at the I site, the putative A c band was recovered and subjected to sequence analysis, as described above. The first six cycles of sequencing gave ATSLSG, as predicted for the amino end of authentic A c .

Counterpart Mutations in Assemblin Precursor Show No
Chemical Rescue-We next tested for chemical rescue of the same mutations in the context of the protease precursor (pPR), the primary translation product of the protease gene made in virus-infected cells, the additional carboxyl sequence of which contains the herpesvirus group conserved maturational (M) and release (R) cleavage sites (Fig. 4C). As done with assemblin, we first tested the H47A mutant of SCMV pPR (originally called pNP1, "precursor nested protein 1," Ref. 38) by expressing it in transfected HEK cells, with 0, 20, or 40 mM imidazole added to the maintenance medium the day after transfection. Wild-type pPR and inactive mutant S118A-pPR were expressed in parallel as controls. Three days after transfection the cells were solubilized, and the samples were analyzed by phosphorimaging following SDS-PAGE and Western immunoassay.
Wild-type pPR was cleaved at the M site to yield PR, at the R site to yield assemblin (PR n ), and at the I site to yield A n and A c (Fig. 4A, lanes  1-3). None of these proteins diagnostic of pPR M-, R-, and I-site cleavage was detected in the S118A-or H47A-pPR mutants (Fig. 4A, lanes  4 -9). Although the H47A-pPR sample contained comparatively more immunoreactive material smaller than pPR, the amount was not increased by imidazole and is attributed to nonspecific proteolysis. A similar experiment was performed using the H47G-pPR mutant and a 14-cm gel to obtain better resolution of the M-and R-site cleavage products. It can be seen more easily here that none of the products resulting from M-and R-site cleavage (i.e. PR and PR c ϩ PR n , respectively) was present in the H47G-pPR preparations (Fig. 4B, lanes 1 and  2). The bands indicated by dots above pPR, PR, and PR c are phosphorylated forms of each (Fig. 4B) (39,40). The A n and A c fragments are not retained in these gels. . Imidazole fails to rescue counterpart mutations of SCMV assemblin precursor, pPR. A, HEK cells transfected with plasmids encoding wild type (Wt) or the S118A or H47A mutants of the SCMV precursor protease were grown in 0, 20, or 40 mM imidazole, lysed in protein sample buffer 2, and analyzed by Western immunoassay (with a mixture of anti-N2 and anti-C2) following SDS-PAGE. Shown here is a phosphorimage of the resulting membrane. Abbreviations to the left of the image are as in Fig. 1A; asterisk above PR (M-site-cleaved pPR) indicates position of protease precursor, pPR (see B and C); dashes are for alignment with abbreviations in margin. B, HEK cells were transfected with wild-type or H47G mutant pPR or S118A assemblin and analyzed as described for A. Western immunoassay was performed with a mixture of anti-N2 and anti-N1 (detects PR c , see C). Shown here is a phosphorimage of the resulting membrane; dots denote phosphorylated forms of pPR, PR, and PR c (39,40), and dashes are to align positions PR and PR c . C, shown here is a schematic of SCMV pPR illustrating the internal (I), release (R), and maturational (M) cleavage sites, fragments resulting from their cleavage (lines), and their abbreviations and molecular weights (ϫ10 Ϫ3 ) (in parentheses). Protein abbreviations are as in legend to Fig. 1D. PR is the large fragment resulting from M-site-cleaved pPR. PR c is the carboxyl portion of R-site-cleaved PR. FIGURE 5. Imidazole fails to rescue counterpart mutations of HCMV assemblin precursor, pPR. A, bacteria transformed with plasmids encoding the wild type (Wt) or the H63A, H63G, or S132A mutants of HCMV pPR were grown in the absence (Ϫ) or presence (ϩ) of imidazole, lysed in protein sample buffer 2, and analyzed by Western immunoassay (using anti-N2) following SDS-PAGE. Shown here is a phosphorimage of the resulting membrane. Abbreviations to the left of the image are as in Fig. 2A; pPR is the protease precursor (see also B). Dashes between lanes 3-4 and lanes 5-6 indicate position of PR n . B, shown here is a schematic of HCMV pPR illustrating the internal (I), release (R), and maturational (M) cleavage sites and fragments resulting from their cleavage and their abbreviations and molecular weights (ϫ10 Ϫ3 ) (in parentheses). Protein abbreviations are for the M-site cleavage products of pPR (PR and Tail), the R-site cleavage products of PR (PR n and PR c ), and the I-site cleavage products of PR n (A n and A c ).
We repeated the experiment using the HCMV assemblin precursor expressed in bacteria to reproduce the conditions that yielded rescue of I-site cleavage for HCMV assemblin. Western immunoassays of the resulting samples showed that little of the wild-type pPR or PR remained uncleaved and that most assemblin (PR n ) was likewise cleaved to A n and A c (Fig. 5A, lanes 1and 2; A c not detected by anti-N2). The patterns for the H63A-and H63G-pPR mutants showed none of the diagnostic fragments resulting from M-, R-, or I-site cleavage, most notably A n from I-site cleavage, and were essentially indistinguishable from that of the catalytically dead S132A-pPR mutant, with or without imidazole (Fig.  5A, lanes 3-8).
Thus, under the same conditions that imidazole restored I-site cleavage activity to the His mutants of SCMV and HCMV assemblin, we found no evidence that it restored any cleavage activity to corresponding mutants of SCMV and HCMV pPR. We were unable to test for imidazole rescue of pPR in vitro, as was done for assemblin, because of "sticky" domains within the PR c sequence that interfere with eluting pPR from chromatography media under nondenaturing conditions (data not shown).

DISCUSSION
We have tested the feasibility of using chemical rescue to create a conditional-lethal mutant of the HCMV serine maturational protease, pUL80a, and unexpectedly found that the approach works for CMV assemblin but not its precursor. We initiated the study using assemblin, the 28-kDa structurally and enzymatically well characterized proteolytic portion of the larger precursor enzyme, and measured activity by monitoring cleavage at its internal (I) site. The critical His (SCMV, His-47; HCMV, His-63) of the active site triad was targeted for mutagenesis, and imidazole was used to restore activity. Inactivating substitutions of Ala or Gly were rescued for I-site cleavage in living cells and in vitro (Figs. 1 and 2). However, when the same mutations were tested for imidazole rescue in the precursor form of the enzyme, we detected no cleavage at any of its four sites (Figs. 4 and 5).
Comparable results were obtained with mammalian, insect (data not shown), and bacterial cells, indicating the applicability of imidazolemediated rescue to a broad range of living systems, an approach also exploited recently in studies of the protein tyrosine kinase, Csk (41), and the influenza M protein ion channel (42). Although the observed rescue in living cells could have alternate explanations (e.g. because of imidazole metabolite), its reproducibility in vitro using affinity-selected mutant assemblin supports the conclusion that it is due to a direct influence of imidazole on the enzyme. We also considered that the observed imidazole-mediated cleavage might be a general property of all herpesvirus assemblin homologs or that failure to rescue the precursor might be peculiar to the CMV homolog. We found, however, that the counterpart mutants of the herpes simplex virus assemblin and precursor homologs did not respond to chemical rescue (data not shown from expression of H43A and H43G HSV assemblin and pPR in HEK cells and in bacteria), consistent with a requirement for factors specific to CMV assemblin (e.g. I site, I-site loop, or both).
Of the four amino acid side chains known to be involved directly in catalysis by assemblin (e.g. HCMV Ser-132, His-63, His-157, Arg-166), only that of the critical His-63 was compensated functionally by imidazole. Failure of imidazole to rescue mutations of the serine nucleophile and His-157 was not surprising, given the structural dissimilarity of imidazole to serine and the nonessential nature of His-157. More noteworthy was its inability to restore activity to the oxyanion hole Arg (data from transfection experiments performed with the SCMV counterpart mutants, R151A and R151G assemblin, see supplemental Fig. S1), considering that this residue is essential for assemblin activity (43) and that there is precedent for mutations of catalytic arginines in other enzymes responding to chemical rescue by imidazole (44 -46).
The comparatively better rescue of Gly substitutions than Ala for the critical His in both SCMV and HCMV assemblin is attributed to the smaller size and greater structural flexibility of Gly, allowing imidazole a better functional fit at the mutant His position (37). This interpretation is consistent with (i) the inability of imidazole to enhance activity of a mutant containing the larger Gln substitution (Fig. 1A, lanes 9 and 10) and (ii) results showing no rescue by larger compounds structurally related to imidazole (data not shown from in vitro assays done with the four compounds listed under "Experimental Procedures"). The capacity of H47Q to enable limited I-site cleavage by assemblin and production of additional cleavage products (Fig. 1A) indicates that Gln can partially substitute for the catalytic His and suggests that its substitution alters cleavage specificity.
The inability of imidazole to restore activity to mutants in the protease precursor was unexpected, considering how effectively it rescues the same mutants of assemblin under the same conditions. The difference was not due to a low concentration (e.g. insufficient for activation by dimerization) or inactivity of pPR (e.g. protein insolubility), as determined by comparison with other pPR mutants that were expressed in comparably high amounts in mammalian, insect, and bacterial cells and were active (Figs. 4 and 5; data not shown from studies in progress to purify active pPR). It is also unlikely that the difference was due to a reduced accessibility of imidazole to the pPR active site, given that much larger molecules (e.g. its 38-kDa substrate, pUL80.5) gain access. Instead, we interpret this finding to reveal a structural difference between the two forms of the enzyme, with implications for the design and screening of protease inhibitors.
It has been suggested that dimerization of assemblin results in a small positional shift of the oxyanion hole Arg that results in activation (47)(48)(49). It has also been suggested that without R-site cleavage, conformation of the pPR sequence destined to form the carboxyl terminus of assemblin would be different (14) and possibly unable to participate in the interactions and changes, which accompany dimerization and activation. Thus, one explanation of our finding is that imidazole rescue of I-site cleavage works only when the active site residues are in the relative orientation they assume in the activated assemblin dimer, which may be different in the precursor. Sensitivity of the imidazole rescue effect to such small changes is compatible with the Ͼ10-fold improved I-site cleavage observed when Gly was substituted for the critical His rather than Ala (e.g. Figs. 1 and 2).
It is also possible that the I-site substrate of pPR has comparatively reduced access to the active site. The crystal structure of assemblin places the I site within an unstructured loop in the vicinity of the active site (11)(12)(13)(14). Extension of the loop toward the interior of the protease would situate the I site in proximity to the catalytic residues. This structural feature could give the I site privileged access to the catalytic residues, perhaps bypassing the normal mode of substrate binding. Constraints on movement of the loop in pPR could restrict its access to the active site, eliminating the proximity that may be required for chemical rescue.
The inability of imidazole to restore either M-or R-site cleavage to the H47G SCMV or the H63G HCMV pPR mutants presents a major challenge to developing the imidazole-dependent mutant virus that we had envisioned because cleavage at these two sites is likely to be essential for the replication of all herpesviruses (50). On the other hand, the possibility that the protein fold of the precursor may differ sufficiently within the nascent capsid to enable rescue of the H63G mutant during virus replication is not formally ruled out.
Our finding that imidazole can restore I-site cleavage to mutants in assemblin but not in its precursor distinguishes the two forms of the CMV protease and reinforces other evidence for enzymatic differences between them (51,52). We suggest that the difference reflects a conformational change in the active site environment and that the imidazole rescue effect may provide a useful tool for evaluating whether and how the conformation and mechanism of the enzyme are influenced by changes that accompany its processive maturational cleavage and folding.