Structural features of tRNALys favored by anticodon nuclease as inferred from reactivities of anticodon stem and loop substrate analogs.

The bacterial tRNA(Lys)-specific PrrC-anticodon nuclease efficiently cleaved an anticodon stem-loop (ASL) oligoribonucleotide containing the natural modified bases, suggesting this region harbors the specificity determinants. Assays of ASL analogs indicated that the 6-threonylcarbamoyl adenosine modification (t(6)A37) enhances the reactivity. The side chain of the modified wobble base 5-methylaminomethyl-2-thiouridine (mnm(5)s(2)U34) has a weaker positive effect depending on the context of other modifications. The s(2)U34 modification apparently has none and the pseudouridine (psi39) was inhibitory in most modification contexts. GC-rich but not IC-rich stems abolished the activity. Correlating the reported structural effects of the base modifications with their effects on anticodon nuclease activity suggests preference for substrates where the anticodon nucleotides assume a stacked A-RNA conformation and base pairing interactions in the stem are destabilized. Moreover, the proposal that PrrC residue Asp(287) contacts mnm(5)s(2)U34 was reinforced by the observations that the mammalian tRNA(Lys-3) wobble base 5-methoxycarbonyl methyl-2-thiouridine (mcm(5)s(2)U) is inhibitory and that the D287H mutant favors tRNA(Lys-3) over Escherichia coli tRNA(Lys). The detection of this mutation and ability of PrrC to cleave the isolated ASL suggest that anticodon nuclease may be used to cleave tRNA(Lys-3) primer molecules annealed to the genomic RNA template of the human immunodeficiency virus.

The active form of the core polypeptide PrrC has not been purified to homogeneity. Yet PrrC certainly harbors the AC-Nase function, judged from the manifestation of ACNase activity by E. coli (8) and mammalian cells overexpressing PrrC (9). Moreover, a tRNA Lys anticodon recognition site has been pinpointed in a cluster of selected PrrC mutations, two adjacent members of which (D287Y, S288P) alter the cleavage site specificity (10). Other mutations directed into this site (D287H, D287Q, or D287N) alter the specificity so that tRNA Lys forms with a hypomodified wobble base, otherwise poor substrates, are rendered more reactive than the natural. These compensatory effects suggest the existence of a specific interaction between the Asp 287 residue of PrrC and the modified wobble base (11).
Comparing bacterial and mammalian tRNA Lys substrates of ACNase has suggested that the specificity determinants reside in the anticodon loop and lower part of the anticodon stem (9). These determinants must include the anticodon sequence, judged from the following observations. First, the anticodon of tRNA Lys resembles the anticodons of most secondary substrates cleaved when PrrC is overexpressed (10). Second, most single anticodon base substitutions in unmodified tRNA Lys abolish ACNase reactivity. Third, a substrate analog with a tRNA Lys anticodon transplanted in an otherwise tRNA Arg sequence is as reactive as the wild type tRNA Lys sequence (11).
The anticodon stem and loop (ASL) domain of E. coli tRNA Lys contains three modified bases that profoundly affect its conformation (12). They include the doubly modified wobble base 5-methylaminomethyl-2-thiouridine (mnm 5 S 2 U34), 6-threonyl carbamoyladenosine 3Ј to the anticodon (t 6 A37) and the pseudouridine of the lower base pair of the stem (⌿39). Comparing the NMR solution structures of a synthetic, fully modified tRNA Lys ASL and hypomodified counterparts has suggested that mnm 5 s 2 U34 and t 6 A37 rigidify the anticodon in a predominantly A-RNA stacked conformation. In addition, t 6 A37 modestly destabilizes the base pairing interactions of the ASL while mnm 5 s 2 U34 and ⌿39 counteract this effect (12). It is conceivable that these conformational effects influence ACNase reactivity. However, some of the modifying groups could also interact with PrrC directly (11). Although the ASL seems the prime target of ACNase, weaker interactions with other substrate regions, e.g. the acceptor domain, are not excluded. This is suggested by the observations that substituting the discriminator base A73 partially inhibits ACNase and truncating the ACCA 3Ј-overhang relaxes the cleavage site specificity (11). * This work was supported in part by grants from the Israeli National Science Foundation, the Israeli Ministry of Science Canadian-Israeli Cooperation Fund, and work at the University of Utah was supported by National Institutes of Health Grant GM55508. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Supported in part by postdoctoral fellowship of the Israeli Council for Higher Education.
These results indicate that the tRNA acceptor domain helps position ACNase for the native substrate and that alternative RNA-protein interactions may form in the ACCA truncated mutants as well as for A73 mutants.
Here we report that ACNase efficiently cleaves a synthetic tRNA Lys ASL containing the base modifications of E. coli tRNA Lys . Comparing it with various hypomodified and unmodified analogs revealed opposing effects of the base modifications on ACNase reactivity. Correlating these data with the observed contributions of the base modifications to the ASL solution structure (12,14,15) leads us to suggest that ACNase prefers substrates where the anticodon nucleotides assume a stacked A-RNA conformation and where the base pairing interactions in the ASL stem are relatively destabilized. We also show that the PrrC D287H mutation renders human tRNA Lys3 relatively more reactive than the natural substrate. This finding and the inhibitory effect of the side chain of the tRNA Lys-3 modified wobble base (5-methoxycarbonylmethyl-2-thiouridine, mcm 5 s 2 U34) reinforce the previously proposed interaction between PrrC residue Asp 287 and the side chain of the E. coli tRNA Lys wobble base (11). This result and the ability to cleave isolated ASLs raise the interesting prospect of directing ACNase against tRNA Lys-3 annealed to the genomic RNA template in the priming complex of the human immunodeficiency virus.

EXPERIMENTAL PROCEDURES
Materials-All ASL oligonucleotides used in this study are listed in Table I. The ASL oligonucleotides 1-15 in Table I were chemically synthesized. The methods have been described for oligonucleotides 1, 3, 4, 6, and 7 (13) oligonucleotides 14 and 15 (16) and the syntheses of oligonucleotides 8 -11 will be described elsewhere. 2 ASL oligonucleotides 16 -23 were transcribed in vitro using T7 RNA polymerase according to an established procedure (17). Those containing inosine (ASLs 20 -22) were transcribed using GMP as primer and ITP instead of GTP. The in vitro transcribed ASLs were dephosphorylated by calf intestinal alkaline phosphatase. All ASLs were 5Ј-end labeled using T4 polynucleotide kinase. Bovine tRNA Lys-3 was a gift from Dr. Roland Marquet, CNRS Strasbourg. Purified E. coli tRNA Lys labeled with 32 P at phosphate 33p34 was prepared by ligating fragments 1-33 and 34 -76 of tRNA Lys purified from T4-infected E. coli prr ϩ cells (7). T7 RNA polymerase and T4 polynucleotide kinase were purchased from U. S. Biochemical Corp., T4 RNA ligase from New England Biolabs, [␥-32 P]ATP from Amersham Biosciences, Inc., poly(U) and trimethylamine-N-oxide from Sigma. Calf intestinal alkaline phosphatase and Protease Inhibitor Mixture Tablets (Complete Mini) were purchased from Roche Molecular Biochemicals and DNA oligonucleotides from Invitrogen.
ACNase Assays-E. coli tRNA Lys and mammalian tRNA Lys-3 were assayed as substrates of wild type and mutant alleles of ACNase using crude (S-150) fractions essentially as described (11). Assays involving ASLs required in general the removal of nonspecific nucleases. Therefore, they were performed with a partially purified enzyme fraction isolated from E. coli K38:pGP1-2:pRRC11-D222E cells by Superdex 200 gel filtration in a procedure to be detailed elsewhere. 3 Additional protection against nonspecific degradation was afforded by including in the reaction mixture carrier poly(U), increasing the trimethylamine-N-oxide concentration to 1.5 M and lowering the reaction temperature from 10 to 0°C. The standard reaction mixtures (10 l) contained 5 l of the partially purified core ACNase (D222E allele, Superdex 200 peak activity fraction of ϳ270 kDa) containing 6 ng of PrrC, 36 mM NH 4 Cl, 6 mM Tris-HCl buffer, pH 7.5, 9 mM MgCl 2 , 3 mM 2-mercaptoethanol, 50 -100 ng of poly(U), 1.5 M trimethylamine-N-oxide, 6% glycerol, protease inhibitor mixture (Complete, Mini, Roche Molecular Biochemicals, diluted according to manufacturers instructions), 2-3 fmol of 5Ј-32 P-labeled ASL or E. coli tRNA Lys labeled with 32 P at the ACNase cleavage junction. The products were deproteinized and separated by denaturing polyacrylamide-urea gel electrophoresis. Each assay was replicated 3-4 times. Kinetic analyses were performed with E. coli tRNA Lys and several of the ASL substrates found most reactive (listed in Table II). The kinetic analyses entailed measurements of initial reaction rates during incubation times reaching 120 min at 1-50 fmol substrate concentrations. Determinations of cleavage extents were carried out with all the ASL substrates following 3 h incubation. The amounts of product were determined by counting the radioactivity or by densitometric tracing of the gel autoradiogram using Hewlett Packard ScanJet 3p and TINA software (Raytest Isotopenmessgerä te GmbH), compatible with the TINA-PCBAS and TIFF files of the scanner.

RESULTS
Specific and Efficient Cleavage of the Fully Modified tRNA Lys ASL by ACNase-An ASL 17-mer containing the three modified bases of E. coli tRNA Lys (henceforth the reference ASL, Fig. 1), various hypomodified and unmodified derivatives and ASLs containing modified bases of mammalian tRNA Lys3 (Table I) were examined as ACNase substrates. The reference ASL and other chemically synthesized ASLs (ASLs 1-15) matched the mammalian tRNA Lys-3 sequence, which differs from that of E. coli tRNA Lys in the upper three base pairs of the anticodon stem (compare the left and right panels in Fig. 1). However, the upper portion of the stem was considered nonessential for recognition by ACNase (9). The synthetic ASLs 5, 12, and 13 contained an extra 3Ј-dT residue since a labile linker was utilized for the chemical synthesis of the ASLs containing mcm 5 s 2 U and ms 2 t 6 A. 2 The unmodified ASL 15 featuring the mammalian tRNA Lys3 sequence was also chemically synthesized. Unmodified ASLs 16 -23 were transcribed in vitro and all of them contained a 5Ј terminal G residue dictated by the constraints of transcription with T7 RNA polymerase (17). The in vitro transcribed ASL 16 corresponded in sequence to E. coli tRNA Lys (Fig. 1b). ASL 19 resembled the E. coli tRNA Val-3 ASL sequence (19) except for a reversed 5Ј to 3Ј terminal GC base pair. The DNA templates of ASL 19 and of two other ASLs with arbitrary GC-rich stem sequences (ASLs 17 and 18) were also transcribed into inosine-rich versions (ASLs 20 -22). ASL 23 was a mutant of E. coli tRNA Lys sequence in which A37 was replaced by G.
Incubation of the 5Ј-32 P-labeled reference ASL with a partially purified ACNase preparation under conditions that minimize nonspecific degradation ("Experimental Procedures") yielded a heptanucleotide-like product missing from the ACNase-negative control (Fig. 2, compare lanes 2 and 3 with lane 4). This product had a slightly reduced electrophoretic mobility compared with a heptanucleotide containing a (2Ј) 3Ј-terminal phosphate within a marker ladder generated by partial alkaline hydrolysis of the reference ASL (Fig. 2, compare lanes 1 and 4). This retardation of the ACNase product could be accounted for by the lower negative charge of the 2Ј:3Ј-cyclic phosphate end-group expected to be found in the product. Incubating the ACNase product with T4 polynucleotide kinase, endowed with 2Ј:3Ј-cyclic-phosphodiesterase and 3Ј-phosphatase activities (7,20), shifted the product to the position of the 3Ј-dephosphorylated heptanucleotide marker (compare lanes 5, 6), consistent with removal of the cyclic phosphate end group. Hence, ACNase appeared to cleave the ASL similar to the natural substrate, targeting the same site and producing the same cleavage termini.
Kinetic comparison of the reference ASL and full sized tRNA Lys (Table II) yielded respective K cat (for a PrrC-D222E hexamer) of ϳ0.04 min Ϫ1 and ϳ0.02 min Ϫ1 and K m of 2.6 ϫ 10 Ϫ7 M and 0.8 ϫ 10 Ϫ7 M. A 2-fold faster cleavage of the ASL was also demonstrated by digesting the two substrates in the same reaction mixture (Fig. 3).
Contributions of ASL Base Modifications to ACNase Reactivity-In vitro transcribed, unmodified tRNA Lys is less reactive than the natural, fully modified ACNase substrate, suggesting a role for at least some of the base modification in substrate recognition and/or reactivity. The tRNA Lys base modifications relevant to ACNase reactivity could be confined to the ASL domain, judged by the high reactivity of the reference ASL (Fig.  2, Table II). The importance of at least one of the wobble base modifications to ACNase reactivity has been previously suggested by inefficient cleavage of hypomodified tRNA Lys lacking the mnm 5 U34 modification (both in vivo and in vitro) or s 2 U34 (in vivo) and suppression of the wobble base lesions by certain PrrC Asp 287 replacement mutations (11). To investigate the contribution of all the ASL base modifications to ACNase reactivity under more defined conditions, we compared the reference ASL with the analogs listed in Table I in an in vitro ACNase assay. Apparent K m and K cat values were obtained for the most reactive ASLs (Table II). Weaker substrates were characterized by extents of cleavage under standard assay conditions (Table I). Below we use these data to assess the contributions of individual base modifications to ACNase reactivity, comparing matched ASL pairs differing by the presence of a given base modification.
The most pronounced effect was exerted by the bacterial t 6 A37 and mammalian ms 2 t 6 A37 modifications. Hypomodified ASLs containing either as the only modified base were cleaved to a 17-18-fold greater extent than the corresponding unmodified ASL. These two partially modified ASLs resembled the reference ASL in K m but had at least 2-fold larger K cat values (Table II). Hence, the mnm 5 s 2 U34 and/or ⌿39 base modifications could attenuate ACNase reactivity of the reference ASL.
The specific contributions of the wobble base modifications to ACNase reactivity were evaluated similarly. The extent of cleavage of the mnm 5 s 2 U34/⌿39-ASL was 3-4-fold larger than that of the s 2 U34/⌿39-ASL (Table I). This suggested that the mnm 5 U34 side chain exerts a positive effect on ACNase reactivity. In contrast, the mcm 5 U34-ASL and mcm 5 s 2 U34-ASL, both containing the side chain of the tRNA Lys-3 wobble base, were less reactive than the corresponding unmodified ASL (Table I). The opposing effects of mnm 5 U34 and mcm 5 U34 were underscored by the large difference in the extents of cleavage between the mnm 5 s 2 U34/⌿39 and mcm 5 s 2 U34/⌿39 containing ASLs (7.2 versus less than 0.2%, respectively). The negative effect of the mcm 5 U34 was further demonstrated by the different kinetic values of the ms 2 t 6 A37/⌿39/3Ј-dT ASL (K m ϭ 3.0 Ϯ 0.6 ϫ 10 Ϫ8 M, K cat ϭ 0.014 min Ϫ1 ) and mcm 5 s 2 U34/ms 2 t 6 A37/ FIG. 1. Composition of the tRNA ASLs. The sequence of the reference ASL is derived from mammalian tRNA Lys-3 , which differs from E. coli tRNA Lys in the upper three bases of the anticodon stem. The mammalian tRNA Lys-3 isoacceptor contains the modified wobble base 5-methoxycarbonylmethyl-2-thiouridine (mcm 5 s 2 U or U9), the modified base 3Ј to the anticodon is 2-methylthio-6-threonylcarbamoyl-adenosine (ms 2 t 6 A or A9), and the C-nucleoside pseudouridine (⌿) stabilizes the A31-⌿39 base pair. The single E. coli tRNA Lys isoacceptor has a related modification pattern with 5-methylaminomethyl-2-thiouridine (mmm 5 s 2 U or U8), 6-threonylcarbamoyl-adenosine (t6A or A7), and pseudouridine (⌿). The modified nucleoside abbreviations are those originally described by Sprinzl et al. (29). The updated nucleoside modification data base is maintained by McCloskey and co-workers (30). ⌿39/3Ј-dT ASL (K m ϭ 6.0 Ϯ 2.0 ϫ 10 Ϫ8 M, K cat ϭ 0.004 min Ϫ1 ). Thus, mcm 5 s 2 U34 reduced the K cat by more than 3-fold and the binding affinity by ϳ2-fold and the overall catalytic efficiency ϳ7-fold. On the other hand, adding mnm 5 s 2 U34 to the t 6 A/⌿39 background increased K cat by more than 3-fold and reduced the affinity ϳ2-fold (K m ϭ 2.6 Ϯ 0.6 ϫ 10 Ϫ7 M, K cat ϭ 0.044 min Ϫ1 versus K m ϭ 1.7 Ϯ 0.7 ϫ 10 Ϫ7 M, K cat ϭ 0.144 min Ϫ1 , respectively). Hence, the negative changes in catalytic efficiency con-ferred by mcm 5 s 2 U34 may be ascribed to the side chain rather than 2-thiol group. This inference is supported by the similar inhibitions caused by mcm 5 U34 and mcm 5 s 2 U34 when introduced into the unmodified background (Table I, compare ASLs  8 and 10 with 15). It is noteworthy that the lack of the s 2 U34 modification from E. coli tRNA Lys had no detectable effect on the cleavage efficiency in vitro although in vivo this lesion was synthetically lethal with PrrC and severely inhibited tRNA Lys cleavage (11). The fully modified E. coli tRNA Lys mnm 5 s 2 U34/t 6 A37/⌿39-ASL was cleaved with a 10-fold higher K cat and exhibited a 3-4-fold weaker binding compared with mcm 5 s 2 U34/ms 2 t 6 A37/ ⌿39/3Ј-dT-ASL carrying the base modifications of tRNA Lys-3 (Table II). However, confounding this comparison is the large negative effect of the extra 3Ј-dT residue in the latter substrate. Namely, against the ms 2 t 6 A37/⌿39 backdrop, the extra 3Ј-dT reduced the K m and K cat each by about an order of magnitude (Table II).
The presence of ⌿39 instead of U39 rendered the ASL in certain cases less reactive. The unmodified ASL was cleaved to a 2-fold higher extent than the ⌿39-containing counterpart. Similarly, introducing ⌿39 into ASLs containing mcm 5 U34 or FIG. 2. ACNase specifically cleaves the reference ASL. The 5Ј-32 P-labeled reference ASL was incubated with partially purified AC-Nase (Superdex 200 fraction) and the products were separated by denaturing 20% polyacrylamide-urea gel electrophoresis as such (lane 4) or after treatment with T4 polynucleotide kinase (lane 5). A marker "ladder" (lane 1) was obtained by partial alkaline hydrolysis of the ASL. Part of the digest was further treated with polynucleotide kinase that effectively dephosphorylates tri-and higher oligonucleotides (lane 6). The gap between the heptamer and octamer markers in lanes 1 and 6 is due to the presence of the basic mnm 5 s 2 U34 residue in the octamer. Numbers on the right are chain length in nucleotides; numbers on the left marked by an asterisk represent the slower migrating 3Ј-OH containing oligonucleotides. Lanes 2 and 3 are respective controls of untreated ASL or ASL incubated with inactivated ACNase (Superdex fraction preincubated 30 min at 37°C).  3. E. coli tRNA Lys and the reference ASL as ACNase substrates. The two substrates were incubated in the same reaction mixture. RNA extracted from aliquots withdrawn at the indicated times was separated by denaturing polyacrylamide-urea gel electrophoresis. ASL frag, heptanucleotide cleavage product of the ASL. mcm 5 s 2 U34 reduced their activity more than 5-fold (Table I). However, ⌿39 had no apparent effect on either kinetic parameter when introduced in the ms 2 t 6 A37/3Ј-dT context and over the t 6 A37 background it even reduced the K m Ͻ3-fold (Table II).
ASL Reactivity and Stem Stability-Unmodified ASLs with stems (underlined) corresponding in sequence to mammalian tRNA Lys3 (5Ј-UCAGACUUUUAAUCUGA-3Ј) or E. coli tRNA Lys (5Ј-GUUGACUUUUAAUCAAC-3Ј) were similarly reactive as ACNase substrates, confirming the assumption that the top three base pairs are not critical for ACNase specificity (9). Arbitrary, all GC stems of the respective ASLs 17 and 18 (5Ј-GGCGGCUUUUAACCGCC-3Ј and 5Ј-GCCGGCUUU-UAAGCGGC-3Ј) abolished the reactivity. A similar effect was exerted by the GC-rich stem of ASL 19, resembling that of E. coli tRNA Val-3 except for a reversed 5Ј to 3Ј terminal pair (5Ј-GCACCCUUUUAAGGUGC-3Ј). Replacing the G residues of these ASLs with inosine residues (except for the 5Ј-terminal) restored the reactivity almost to the level of the unmodified reference ASLs in the case of the two arbitrary stem sequences. However, the tRNA Val-3 -like I-rich ASL was an order of magnitude less reactive.
A PrrC Mutation Renders tRNA Lys-3 More Reactive Than the Natural Substrate-The D287H, D287N, or D287Q mutations have rendered hypomodified E. coli tRNA Lys forms containing s 2 U34 or mnm 5 U34 more reactive than the natural substrate, opposite to the preference of wild type PrrC (11). These effects, attributed to some interaction of Asp 287 with the wobble base, have also raised the prospect of designing new ACNase cleavage specificities (10), especially for a differently modified wobble base such as that of mammalian tRNA Lys-3 . In fact, whereas wild type PrrC or the pseudo-wild type allele D222E (11) cleaved E. coli tRNA Lys 2-3 faster than mammalian tRNA Lys-3 , D287H featured the opposite preference (Fig. 4).

DISCUSSION
The tRNA Lys ASL as a Minimal Essential Substrate of ACNase-The reference ASL with the three modified bases of E. coli tRNA Lys was cleaved by ACNase twice as fast as the natural substrate but featured a 3-4-fold reduced affinity to the enzyme, suggesting that the ASL domain contains all or nearly all of the ACNase recognition determinants. The small differences between the ASL and full-sized tRNA in K cat and K m are attributed to greater conformational freedom of the ASL. Such flexibility could facilitate attainment of a productive transition-state conformation by the ASL but weaken initial binding to the enzyme. However, the weaker binding of the ASL also leaves open possible interactions between PrrC and other portions of tRNA Lys such as the acceptor region. Namely, mutating the A73 discriminator base inhibits ACNase and truncating the 3Ј-terminal ACCA sequence relaxes the cleavage site specificity (11). These outcomes could be explained by a weak interaction between PrrC and the acceptor region of tRNA Lys . Accordingly, the A73 discriminator could contribute to substrate specificity, directly or by counter-selecting other tRNAs. We also speculate that truncation of the substrate to contain just the ASL eliminates these interactions that cause misalignment and therefore allows for specific and efficient recognition of the cleavage site.
Structural ASL Features Favored by ACNase-Comparing the NMR solution structures of the reference ASL and hypomodified counterparts has suggested that individual base modifications play distinct and partially opposing roles in transforming the disordered, unmodified tRNA Lys ASL into the highly ordered native structure (12). Thus, t 6 A37 improves stacking with A38, and strengthens the interaction of U33 N3H with the 35p36 phosphate, a hallmark of the anticodon loop U-turn. On the other hand, t 6 A37 partially destabilizes the ASL stem base pairing, reducing the T m by about 2°C. The wobble base side chain mnm 5 U34 also stabilizes the anticodon base stacking and U-turn conformation but partially counteracts the base pair disrupting effect of t 6 A37. Destabilization of the stem and/or the C 32 -A 38 base pair (21) by t 6 A37 also opposes the effect of the ⌿39 modification (12,14,15). Here we have shown that the ASL base modifications also exert distinct effects on ACNase reactivity (Tables I and II). The underlying causes of these complicated and opposing effects may be understood by considering the contributions of the individual base modifications to the ASL solution structure, as described above. Such a correlation leads us to propose that ACNase favors substrates where the anticodon nucleotides are held in a helical A-RNA conformation and the U-turn conformation is retained but the ASL stem base pairing interactions are modestly destabilized. This assumption is backed by the following arguments. First, t 6 A37 or ms 2 t 6 A37 that dramatically enhanced the ACNase reaction rate also stabilize the anticodon A-RNA and the loops U-turn conformations and destabilize the anticodon stem and bifurcated C 32 -A 38 base pair (12,14), suggesting that ACNase favors these structural features. In agreement, ⌿39, which inhibited ACNase activity in certain cases, exerts the opposite effect on ASL base pairing interactions. As for the mild positive effect of mnm 5 U34 on ACNase reactivity, it could reflect a balance between a positive contribution to ACNase reactivity of the enhanced anticodon stacking by this modification and a negative contribution due to the stabilized base pairing. However, part of the ACNase-stimulating effect of mnm 5 U34 may be attributed to a direct interaction with Asp 287 , the anticodon-recognizing residue of PrrC (an issue to be elaborated below). Abolition of ACNase reactivity by the GC-rich stems, but not the IC-rich, reinforces the notion that base pairing stability of the ASL negatively affects the reactivity. As already mentioned, the higher K cat and lower K m of the reference ASL compared with the full-sized tRNA Lys may be attributed to the fact that the upper base pairs of the stem are more constrained within the intact tRNA structure. Hence, the destabilization of the anticodon stem, needed for the productive interaction with ACNase, may be inhibited with intact tRNA Lys . However, as we have seen for the ASLs, the more rigid conformation of the intact tRNA could promote tighter initial binding to the enzyme. FIG. 4. Mutation D287H renders tRNA Lys-3 more reactive than E. coli tRNA Lys . The two tRNA substrates were incubated in the standard reaction mixture containing the indicated PrrC alleles. The products were separated by denaturing polyacrylamide urea gel electrophoresis as described in the legend to Fig. 3. The ratios between the reactivities of the two substrates were calculated from the initial rates of cleavage. EC, E. coli tRNA Lys ; Lys3, tRNA Lys-3 .
Transplanting the tRNA Lys UUU anticodon within an otherwise tRNA Val-3(UUU) sequence does not elicit detectable ACNase reactivity whereas tRNA Arg(UUU) is highly reactive (11). The difference between the two chimerical tRNA sequences has been attributed to ill-defined ACNase determinants found outside the anticodon loop and shared by tRNA Lys and tRNA Arg but missing from tRNA Val-3 . However, in view of the current data, a new interpretation becomes apparent. Namely, the failure to cleave tRNA Val-3(UUU) may reflect the greater stability of its GC-rich anticodon stem (4 GC pairs) and the reversed purine-pyrimidine configuration of the two base pairs at the bottom of the stem. The importance of this configuration is suggested by the weaker reactivity of the IC-rich version of the tRNA Val-3 -ASL, compared with the counterparts with arbitrary stem sequences that shared the tRNA Lys / tRNA Arg configuration. Presumably, the reversed base pair configuration further stabilizes the stem by enhancing G 39 -A 38 base stacking.
A Distinct Wobble Base Side Chain Causes the Reduced Reactivity of tRNA Lys-3 -The relative ACNase reactivities of natural substrate, the completely unmodified version and mutants thereof as well as of the mammalian tRNA Lys isoacceptors have suggested the following order of wobble base preference of ACNase: mnm 5 s 2 UϾUϾCϾmcm 5 s 2 U (11). However, such an order has been derived from eclectic data collected in vivo and in vitro and comparisons of substrates differing in more that just their wobble bases. The current comparison uses defined in vitro assay conditions of matched ASLs differing only in a wobble base modification (Tables I and II). These results confirm the opposing contributions of the bacterial mnm 5 U34 and mammalian mcm 5 U34 side chains to ACNase reactivity, whereas both modifications similarly influence the overall ASL conformation (12,24). Therefore, the inhibitory effect of mcm 5 U34 on ACNase reactivity suggests a specific interaction between residue Asp 287 of PrrC and the mnm 5 U34 side chain of the natural substrate (11). Since replacing mnm 5 U34 with mcm 5 U34 reduces both K m and K cat (Table II), the Asp 287mnm 5 U34 interaction could influence both substrate binding and the catalytic step.
The detection of PrrC mutations that alter the substrate cleavage specificity and compensate for a missing wobble base modification (11) raised the possibility that some of these mutations would also cause ACNase to prefer mammalian tRNA Lys-3 over the natural substrate. One interest in such an outcome stems from the role human tRNA Lys-3 plays as the primer tRNA of reverse transcription in human immunodeficiency viruses (22,23). There are several characteristics of the tRNA Lys-3 primer that raise the prospect of using ACNase as a model for developing anti-HIV therapeutics. The tRNA Lys-3 primer is indifferent to viral genetic drift (25) and the tRNA anticodon interaction with the HIV A-loop is critical for transcription initiation (26). Provided that the tRNA anticodon is accessible to ACNase in the primer-template complex, it may be possible to tailor ACNase to discriminate between the free form of tRNA Lys-3 and the annealed tRNA. Detection of ACNase derivatives more proficient in cleaving tRNA Lys-3 would constitute a step toward the goal of developing a model system for therapeutics targeted at the primer-template complex. Since the reduced reactivity of tRNA Lys-3 is likely due to the presence of mcm 5 U34 instead of mnm 5 U34 in the natural substrate, it was expected that tRNA Lys-3 will be rendered more reactive by some of the PrrC mutations that compensate for the absence of mnm 5 U34 (11). This expectation was confirmed by the in vitro behavior of the D287H allele, which cleaved tRNA Lys-3 relatively faster than wild type and pseudowild type alleles, whereas the latter two alleles were relatively more reactive with E. coli tRNA Lys (Fig. 4). Mechanistically, the reversal of substrate preference may be explained in that the mnm 5 U34-Asp 287 and mcm 5 U34-D287H pairs form saltbridges or hydrogen bond interactions of opposite polarities that are vital for substrate binding and/or reactivity. Interestingly, among the PrrC D287 replacement mutants expressed in mammalian cells D287Q showed the highest activity with tRNA Lys-3 , cleaving it to a severalfold higher extent than wild type PrrC. 4 The efficient cleavage of the isolated ASL domain by ACNase also suggests that partially melted tRNA Lys-3 annealed to the HIV-1 genomic RNA will be recognized by ACNase since the native ASL conformation would be retained in the primer-template complex (27,28). This expectation has been recently confirmed. 5