Fragile T-stem in Disease-associated Human Mitochondrial tRNA Sensitizes Structure to Local and Distant Mutations*

Mutations in human mitochondrial isoleucine tRNA (hs mt tRNA Ile ) are associated with cardiomyopathy and opthalmoplegia. A recent study showed that opthal-moplegia-related mutations gave rise to severe decreases in aminoacylation efficiencies and that the defective mutant tRNAs were effective inhibitors of aminoacylation of the wild-type substrate. The results suggested that the effectiveness of the mutations was due in large part to an inherently fragile mitochondrial tRNA structure. Here, we investigate mutant tRNAs associated with cardiomyopathy, and a series of rationally designed second-site substitutions introduced into both opthalmoplegia- and cardiomyopathy-related mutant tRNAs. A source of structural fragility was uncovered. An inherently unstable T-stem appears susceptible to misalignments. This susceptibility sensitizes both do-mains of the L-shaped tRNA structure to base substitutions that are deleterious. Thus, the fragile T-stem makes the structure of this human mitochondrial tRNA particularly vulnerable to local and distant mutations. The canonical tRNA cloverleaf folds into a two-domain L-shaped structure. One domain contains the amino acid attach-ment site and the 12-base pair acceptor-T C C minihelix. The other domain of 10 base pairs contains the dihydrouridine and anticodon-stem and -loop. The primary structures of hs mt tRNAs 1 differ significantly The cleavage of the GST tag by incubation with thrombin (16) and subsequent FPLC (Amersham Phar- macia Biotech) purification yielded a protein of activity comparable with that of the fusion protein. Aminoacylation Assays— Aminoacylation assays were carried out at 37 °C in reaction mixtures containing 10 m M HEPES, pH 7.5, 75 m M NH 4 Cl, 0.05 m M EDTA, 10 m g/ml bovine serum albumin, 10 m M MgCl 2 , 195 m M isoleucine, 5 m M [ 3 H]isoleucine, 2 m M ATP, 25 n M enzyme, and 2 m M tRNA as each assay, tRNAs were annealed in 10 m M Tris, pH 7.5, 0.5 m M EDTA, and 4 m M MgCl 2 by incubation at 70 °C for 5 min and 25 °C for 15 min. Concentrations of tRNA stock solutions were determined by quanti-tation of absorbance at 260 nm using an extinction coefficient of 9100 M 2 1 (mononucleotide) cm 2 1 (18). Assays were otherwise performed as described (11). Secondary Structure Calculations— Secondary structural analysis was performed using MFOLD Parameters default settings, except suboptimality, varied from 25 to 50%. cloverleaf-like structures were considered.

Mutations in human mitochondrial isoleucine tRNA (hs mt tRNA Ile ) are associated with cardiomyopathy and opthalmoplegia. A recent study showed that opthalmoplegia-related mutations gave rise to severe decreases in aminoacylation efficiencies and that the defective mutant tRNAs were effective inhibitors of aminoacylation of the wild-type substrate. The results suggested that the effectiveness of the mutations was due in large part to an inherently fragile mitochondrial tRNA structure. Here, we investigate mutant tRNAs associated with cardiomyopathy, and a series of rationally designed second-site substitutions introduced into both opthalmoplegia-and cardiomyopathy-related mutant tRNAs. A source of structural fragility was uncovered. An inherently unstable T-stem appears susceptible to misalignments. This susceptibility sensitizes both domains of the L-shaped tRNA structure to base substitutions that are deleterious. Thus, the fragile T-stem makes the structure of this human mitochondrial tRNA particularly vulnerable to local and distant mutations.
The canonical tRNA cloverleaf folds into a two-domain Lshaped structure. One domain contains the amino acid attachment site and the 12-base pair acceptor-T⌿C minihelix. The other domain of 10 base pairs contains the dihydrouridine and anticodon-stem and -loop. The primary structures of hs mt tRNAs 1 differ significantly from cytoplasmic tRNAs (1). Despite their analogous function in the decoding of genetic information, mt tRNAs are generally shorter than cytoplasmic tRNAs and feature higher numbers of AU pairs. In some cases, entire structural elements are deformed or even missing from the canonical cloverleaf, although a truncated L-shaped structure can still be made. The minimized structures of hs mt tRNAs appear to be susceptible to point mutations, as errors in the corresponding mt genes are associated with disease. Over 70 pathology-related mutations in mt tRNA genes are known (2). The impact of the base substitutions on the structure and function of hs mt tRNAs is difficult to predict a priori, given the limited information available concerning the properties of these molecules.
Within the gene for hs mt tRNA Ile , eight different point mutations are correlated with pathologies (3)(4)(5)(6)(7)(8)(9)(10). The genetic errors identified are generally associated with two diseases, cardiomyopathy (3-7) and opthalmoplegia (8 -10). The mutations can be classified according to the types of structural defects induced. The opthalmoplegia mutations are associated with one specific type of structural defect, CA mispairs, whereas the cardiomyopathy mutations are more varied.
The opthalmoplegia-related mutants of tRNA Ile are poor substrates for aminoacylation (11). Studies of the reactivity of these molecules with the hs mt isoleucyl-tRNA synthetase (Il-eRS) revealed kinetic defects that impeded the aminoacylation reaction. Because the binding affinities of the mutated tRNAs for the enzyme were not affected, the mutants were effective inhibitors of the charging of wild-type tRNA. The aminoacylation of these mutant tRNAs was restored with compensatory mutations that reintroduced base pairing. These results prompted the proposal that the structure of the hs mt tRNA Ile was inherently fragile and, thus, small perturbations introduced by the pathogenic mutations were magnified into large losses in function and the creation of effective inhibitors.
Here, using aminoacylation to probe structure-function properties of hs mt tRNA Ile , the effects of a subset of cardiomyopathy-associated mutations are investigated. An A59G mutation in the TC-loop (hereafter called the T-loop) significantly decreased the aminoacylation efficiency of this molecule. In contrast, a C62U mutation (that introduced an AU pair in place of a CA pair within the T-stem), increased the aminoacylation efficiency. Because the C62U mutation both stabilized the Tstem and improved aminoacylation, we considered the possibility that the T-stem had an inherent fragility that affected the entire molecule. With this in mind, we set out to study systematically the effects of specific, rationally designed manipulations of the T-stem to elucidate structural properties that cause the hs mt tRNA Ile to be susceptible to pathogenic mutations. These investigations were aimed at not only exploring the local consequences of mutations in the T-stem and loop, but also at seeing whether these mutations affected distant parts of the tRNA.

MATERIALS AND METHODS
Preparation of tRNA Transcripts-Plasmids encoding tRNA transcripts were generated by ligating overlapping oligonucleotides into a pUC18 plasmid (12) between a BamHI and PstI cleavage site. The tRNA constructs were placed between a T7 RNA polymerase promoter sequence and a BstN1 cleavage site and adjoined at the 3Ј-end with a hammerhead ribozyme derivative to increase the transcription yields obtained with this sequence (13).
T7 RNA polymerase was purified from Escherichia coli as described (14). Plasmids encoding the tRNA constructs were linearized with BstN1 and incubated with T7 RNA polymerase for 2 h at 37°C in the presence of 40 mM Tris, pH 8, 10 mM MgCl 2 , 5 mM dithiothreitol, 0.01% * This work was supported in part by the National Institutes of Health (NIH) Grant GM15539 and by a fellowship from the National Foundation for Cancer Research. 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.
§ Recipient of an NIH postdoctoral fellowship. Present address: Dept. of Chemistry, Boston College, Chestnut Hill, MA 02467. ** Recipient of an operating grant from the Medical Research Council of Canada and a fellowship from le Fonds de la Recherche en Santé du Québec.
Triton X-100, 8% polyethylene glycol 8000, 50 g/ml bovine serum albumin, and 1 mM spermidine. DNase I was added to digest the template, and reactions were extracted with 25:24:1 phenol, pH 4.5, chloroform, isoamyl alcohol. After ethanol precipitation, reactions were resuspended in 50 mM Tris, pH 8, and 10 mM MgCl 2 and incubated at 55°C for 30 min to effect ribozyme cleavage. The tRNA transcripts were purified by 12% polyacrylamide gel electrophoresis and electroeluted. After ethanol precipitation, tRNAs were desalted using a G-50 spin column (Amersham Pharmacia Biotech). There were no detectable differences in the efficiency of transcription or transcript length when mutations were introduced into hs mt tRNA Ile .
The purification protocol employed produced highly active tRNA transcripts. Aminoacylation assays performed with high concentrations of enzyme (ϳ 1 M) were used to assess the percentage of tRNAs that could be charged with isoleucine, and for both wild-type and mutant tRNAs, between 80 and 100% of the tRNA quantitated by uv-vis absorbance as described below could be aminoacylated. These experiments indicated that the 3Ј-end of this tRNA was transcribed with high fidelity.
Expression and Purification of Human Mitochondrial IleRS-The gene for human mitochondrial IleRS was cloned as described (15). The cloned gene was inserted into a yeast expression vector containing a glutathione S-transferase affinity tag and the GAL4 promoter. The fusion protein was expressed under the control of a GAL4 promoter in Saccharomyces cerevisiae in strain INVSc (Invitrogen, Carlsbad, CA). After cell lysis performed in a French press, human mitochondrial IleRS⅐GST was batch purified by isolation on glutathione-agarose (Amersham Pharmacia Biotech) and subsequent elution with glutathione. This protocol yielded protein samples with Ͼ95% purity, as judged by SDS-polyacrylamide gel electrophoresis. The cleavage of the GST tag by incubation with thrombin (16) and subsequent FPLC (Amersham Pharmacia Biotech) purification yielded a protein of activity comparable with that of the fusion protein.
Concentrations of tRNA stock solutions were determined by quantitation of absorbance at 260 nm using an extinction coefficient of 9100 M Ϫ1 (mononucleotide) cm Ϫ1 (18). Assays were otherwise performed as described (11).
Secondary Structure Calculations-Secondary structural analysis was performed using MFOLD (19). Parameters were used in default settings, except for percent suboptimality, which was varied from 25 to 50%. Only cloverleaf-like structures were considered.

RESULTS
Aminoacylation Efficiency of a Cardiomyopathy-related Mutant tRNA-Mutants of hs mt tRNA Ile associated with cardiomyopathy ( Fig. 1) that involve substitutions in the T-stem and loop were generated by in vitro transcription and tested as substrates for the cognate recombinant hs mt IleRS. (Previous studies showed that transcripts of hs mt tRNA Ile were relatively robust substrates for the recombinant enzyme, Ref. 11.) A tRNA containing an A59G mutation that alters an unpaired base in the T-loop had a significantly decreased rate of aminoacylation (Fig. 2). A similar decrease in aminoacylation for this A59G mutant was observed in a previous study employing a crude mitochondrial extract (20).
Interestingly, the T-loop is generally not a locus for the tRNA identity determinants (21). Thus, we considered the possibility that this substitution perturbed the secondary structure by shifting the base pairing so that the loop was reduced in size by three nucleotides (Fig. 3).
Aminoacylation of a Second Cardiomyopathy-related Mutant tRNA-Another cardiomyopathy-associated mutation, C62U, replaces the CA pair in the T-stem with a UA pair. Secondary structure calculations revealed that this substitution decreased the number of stable alternate folds competing with the canonical cloverleaf structure. Indeed, increased aminoacylation activity was observed with the C62U substitution, suggesting that the introduction of an additional Watson-Crick base pair in the T-stem stabilized the cloverleaf structure and enhanced aminoacylation by locking out inactive conformations (Fig. 3). The enhanced charging seen with the C62U substitution suggests that, in this case, the associated cardiomyopathy is not because of a defect in aminoacylation.
Hs mt tRNA Ile Constructs with Altered T-stems-To further investigate the role of base-pairing patterns within the T-stem in the stabilization of the tRNA conformation required for function, different base pairs were substituted for the wild-type C62:A52 pair. In addition to the C62U mutant, two additional mutant tRNAs were synthesized. A C62G/A52C double mutant was made to test the effect of a GC pair, and an A52G mutant was generated to investigate the effect of a CG pair in the 52:62 position.
The C62G/A52C construct (with a G62:C52 pair) displayed slightly higher levels of aminoacylation compared with the C62U mutant tRNA Ile (with a U62:A52 pair, Fig. 4). The increase in activity for these two variants over the wild-type tRNA therefore results mainly from the presence of a base pair at this position. The highest activity is achieved with the G62: C52 pair that is predicted to be most stable. That the identity of the Watson-Crick pair at the 62:52 position is not important for aminoacylation is consistent with this location in the struc- ture of tRNA Ile not being a major contact point for the synthetase in the bacterial system (22,23).
Secondary structure calculations revealed that neither the U62:A52 nor the G62:C52 pairs altered the alignment of bases in the T-stem. The creation of a C62:G52 base pair is predicted to change the pairing such that two unpaired bases are introduced between the acceptor-and T-stems, and the size of the T-loop is reduced by one nucleotide (Fig. 3). Consistent with these predictions, the A52G mutant tRNA Ile (with a C62:G52 pair) was charged at low levels compared with the substrate containing this same base pair in the opposite orientation (Fig.  4).
Rescue of Pathology-related tRNA Mutants with a Stabilized T-stem-Having established the sensitivity to mutation and the fragility of the T-stem (particularly the capacity to form alternative structures), we wondered whether a stabilized stem could compensate for the deleterious effects of other pathologyrelated tRNA mutations. To address whether the impact of the A59G mutation on aminoacylation was related to the lability of the T-stem, this pathogenic mutation was incorporated into a tRNA Ile variant containing a C62U mutation. The substitution of C62 stabilizes the T-stem (by creating a U62:A52 pair) and, as described above, increases aminoacylation relative to the wild-type hs mt tRNA Ile .
Indeed, the C62U mutant was more resistant to the negative effect of the A59G mutation on charging (Fig. 5). Instead of the 80% decrease observed with the wild-type CA-containing Tstem, only a 10% decrease in charging was observed with an U62:A52 T-stem. The loss of charging caused by the A59G mutation in the wild-type construct therefore appears to reflect a greater tendency for the molecule to misfold. This tendency can be counteracted by strengthening the T-stem. Consistent with the notion of misfolding contributing to the effects of the A59G mutation, the secondary structure calculated for this mutant tRNA Ile differs from that of the wild-type structure. The most stable structure resembles the canonical cloverleaf, but the T-loop is shortened by 4 nucleotides and 2 nucleotides are forced out of the T-stem into the junction between this helix and the acceptor stem (Fig. 3). With the introduction of the C62U substitution, the wild-type secondary structure is regained (Fig. 3, far right).
The ability of a stabilized T-stem to rescue the aminoacylation and restore the structure of the cardiomyopathy-related A59G mutant prompted us to examine the effect of this structural element on the functional defects caused by the opthalmoplegia-related mutations: U12C, U27C, G40A. Local structural perturbations in AU-rich stems were previously suggested as the source of attenuated reactivities displayed by these mutants (11). By reexamining these mutations in constructs where the T-stem is stabilized, the contribution of distal structural elements to the deleterious impact of the opthalmoplegia mutations could be evaluated.
Aminoacylation efficiencies were compared for wild-type and mutant tRNAs by calculating the ratios of initial rates for tRNAs containing a C62:A52 (wild-type) versus an U62:A52 base pair. In all cases, the decreases in aminoacylation efficiency seen with the U12C, U27C, G40A, and A59G mutant tRNAs are less severe with a secondary C62U mutation. As previously discussed, even the charging of wild-type tRNA is enhanced. For three of the pathology-associated mutants (U12C, U27C, A59G), the context provided by the U62:A52 substitution enhanced aminoacylation even more than when the U62:A52 substitution was placed in the wild-type tRNA. Positions 12 and 27 lie in the second domain of the L-shaped tertiary structure, whereas position 59 is in the first, acceptor-TC minihelix domain. Thus, these results show that the consequence of a fragile T-stem extends beyond the minihelix to the opposite domain, and thereby suggest an important domain-domain communication. DISCUSSION In the case of the opthalmoplegia-related mutations, previous work showed that these mutant hs mt isoleucine tRNAs were severely defective in aminoacylation (11). The effects of the mutations on the tRNA structure left the mutant tRNAs sufficiently ordered to be excellent inhibitors of aminoacylation of wild-type hs mt tRNA Ile . Thus, in the potentially heteroplasmic environment of the mitochondria (24,25) with mixtures of wild-type and mutant tRNAs, the mutant tRNAs could in principle arrest protein synthesis by virtue of their ability to act as inhibitors.
In contrast, whereas the A59G cardiomyopathy-associated mutation attenuates significantly aminoacylation, the C62U substitution actually improves the efficiency of charging. This improvement is plausible because the substitution strengthens the proper alignment of the T-stem by replacing an A-C pair with A-U (Fig. 3). That the aminoacylation efficiency with this mutant is actually increased through the stabilization of a canonical tRNA structure emphasizes that the nature of the connection between a pathology and a specific mutation may be different for each mutation. The higher activity of the more stable C62U mutant tRNA Ile raises the possibility that the inherent fragility of the wild-type hs mt tRNA structure may be important for function. For example, among other possibilities, a flexible, relatively loose structure may be required for optimal activity with the mt translation apparatus (23).
The inherent weakness of the structure hs mt tRNA Ile gives rise to a significant degree of domain-domain communication. Because of the capacity for "slippage" in the secondary structure caused by base pair misalignments (Fig. 3), the stability of one domain is reliant on the integrity the other. (Whereas interdomain communication is prevalent in this tRNA possessing a minimized structure, it has also been detected in other tRNAs, Ref. 26.) In particular, the slipped secondary structures can disrupt loose tertiary interactions and thereby make one domain particularly sensitive to perturbations in the other.
Although the cause-and-effect basis for the opthalmoplegias and cardiomyopathies associated with mutations in hs mt tRNAs has not been conclusively established, these pathologies have nonetheless provided the motivation and rationale for investigating in more depth the structural features of hs mt tRNAs. In particular, were it not for the cardiomyopathy-associated mutations studied here, the fragility of the T-stem of hs mt tRNA Ile might not have been appreciated. This fragility is manifested by the ease with which single point mutations can lead to serious realignments of the T-stem loop (Fig. 3) that, in turn, have consequences that are global (Fig. 6). It is these global effects that may be most significant for pathology because, for example, they may influence not only aminoacylation, but also the ability of the tRNA to function in subsequent steps of protein synthesis or to be a substrate for processing or modification enzymes.