A GTP:AMP Phosphotransferase, Adk2p, in Saccharomyces cerevisiae

Adenylate kinases participate in maintaining the homeostasis of cellular nucleotides. Depending on the yeast strains, the GTP:AMP phosphotransferase is encoded by the nuclear gene ADK2 with or without a single base pair deletion/insertion near the 3′ end of the open reading frame, and the corresponding protein exists as either Adk2p (short) or Adk2p (long) in the mitochondrial matrix. These two forms are identical except that the three C-terminal residues of Adk2p (short) are changed in Adk2p (long), and the latter contains an additional nine amino acids at the C terminus of the protein. The short form of Adk2p has so far been considered to be inactive (Schricker, R., Magdolen, V., Strobel, G., Bogengruber, E., Breitenbach, M., and Bandlow, W. (1995) J. Biol. Chem. 270, 31103–31110). Using purified proteins, we show that at the physiological temperature for yeast growth (30 °C), both short and long forms of Adk2p are enzymatically active. However, in contrast to the short form, Adk2p (long) is quite resistant to thermal inactivation, urea denaturation, and degradation by trypsin. Unfolding of the long form by high concentrations of urea greatly stimulated its import into isolated mitochondria. Using an integration-based gene-swapping approach, we found that regardless of the yeast strains used, the steady state levels of endogenous Adk2p (long) in mitochondria were 5–10-fold lower compared with those of Adk2p (short). Together, these results suggest that the modified C-terminal domain in Adk2p (long) is not essential for enzyme activity, but it contributes to and strengthens protein folding and/or stability and is particularly important for maintaining enzyme activity under stress conditions.

intracellular AMP level, which serves as a highly sensitive sensor of cellular energy status. Under stress conditions, increased levels of AMP trigger the AMP-activated protein kinase cascade to switch on catabolic pathways that generate ATP while switching off anabolic processes that consume ATP (5,6). Adenylate kinases therefore occupy a central position in many cellular functions under normal as well as stress conditions. Although ATP-specific adenylate kinases from different organisms have been extensively studied, not much is known about the GTP-specific isoforms.
Adenylate kinase in Escherichia coli is indispensable for growth (7). In mammalian cells, at least three (AK1, AK2, and AK3) isoforms of adenylate kinase exist (8 -11). AK1 and AK2 use Mg 2ϩ ATP as the high energy phosphate donor, whereas AK3 is a GTP:AMP phosphotransferase and uses GTP instead of ATP as a phosphoryl donor. AK1 and AK3 are localized to the cytosol and mitochondrial matrix, respectively. In contrast, AK2 resides in the cytosol and mitochondrial intermembrane space (2)(3)(4)(12)(13)(14). Basic cellular and mitochondrial functions are highly conserved from yeast to human (15), and in agreement with this notion, the yeast Saccharomyces cerevisiae also contains three isoforms of adenylate kinase localized to the cytosol and/or mitochondria. As in the mammalian isoforms, all three yeast isozymes are nuclearly encoded. In yeast, the major isoform Adk1p/Aky2p (ϳ24 kDa) is homologous to mammalian AK2 and displays a dual localization; most (Ͼ90%) of the protein is localized to the cytoplasm with a small portion in the mitochondrial intermembrane space (16 -20). The yeast equivalent of mammalian AK1 is the cytosolic Ura6p (21). This yeast protein was identified as a multicopy suppressor of the respiratory-deficient phenotype of adk1 mutants.
The gene ADK2/AKY3 (originally called PAK3) was isolated from the S. cerevisiae strain DL1 by a polymerase chain reaction using degenerate oligonucleotide primers corresponding to conserved regions of the nucleoside monophosphate kinases (22,23). The corresponding protein Adk2p (225 amino acids) showed 45% identity with both E. coli adenylate kinase and human GTP:AMP phosphotransferase (AK3) and was localized to the matrix side of the mitochondrial inner membrane. A disruption of ADK2 did not produce any detectable phenotype. Moreover, plasmid-borne ADK2 did not restore growth of a temperature-sensitive adk1 E. coli mutant at the non-permissive temperature (39°C). Based on these results, it was concluded that the DL1 version of ADK2 codes for a non-functional yeast protein (22,23). These studies, however, could not rule out the possibility that the DL1 version of Adk2p was not inherently inactive; rather it was inactivated at 39°C (which is significantly higher than the usual yeast growth temperature of 30°C) and thus failed to complement the adk1 E. coli mutant at the non-permissive temperature.
Interestingly, random mutagenesis of non-functional ADK2 led to the identification of functional alleles as judged by complementation of the growth phenotype of the adk1 E. coli mutant and GTP:AMP phosphotransferase activity. In each case, the gain of activity was accompanied by a ϩ1 frameshift mutation near the 3Ј end of the coding region of ADK2, causing a change in three C-terminal residues and the addition of nine amino acids at the C terminus of Adk2p (Adk2p (long); see Fig.  1A). It was therefore concluded that the lack of activity of the DL1 version of Adk2p (the short form) was due to the absence of the C-terminal domain resulting from a single base pair deletion near the 3Ј end of the open reading frame (23).
The "so-called" non-functional allele or Adk2p (short) was found to be characteristic of most S. cerevisiae strains (except D273-10B) that are commonly used for laboratory or brewery purposes. In contrast, the alleles carrying the insertion and encoding the active enzyme were found in the more distantly related species Saccharomyces diastaticus, Saccharomyces capensis, and the wine yeasts Malaga and Bordeaux. Since publication of these findings in 1995 (23), Adk2p has not received much attention, perhaps because of its presumed non-functionality in most laboratory strains. Here we have investigated the contribution of the modified C-terminal domain in Adk2p (long) to the molecular properties of the enzyme. Contrary to the existing knowledge, we show that it is not essential for enzyme activity. Rather, it contributes to protein folding and/or stability and plays a critical role in maintaining enzyme activity under harsh conditions.

MATERIALS AND METHODS
Constructs and Yeast Strains-The genes encoding Adk2p (short, accession number AY558457) and Adk2p (long; accession number AY949617) were amplified by PCR from the yeast genomic DNA isolated from BY4741 (Invitrogen) and D273-10B (ATCC 24657), respectively. The PCR products were digested with NdeI and XhoI and cloned into the same sites of pSP64T (24) and pET21b (Novagen, Madison, WI). The former was used for cell-free synthesis of proteins. The latter introduced a His 6 tag in-frame at the C terminus of the protein and was used for bacterial expression. All constructs were verified by sequencing.
Using the PCR-based transplacement cassette approach (25), the native ADK2 gene in the strains BY4741 and D273-10B/A/H/U (26) was replaced with the corresponding tagged version of the gene that introduced three hemagglutinin tags in tandem at the C terminus of the short and long forms of Adk2p, respectively. Likewise, modified BY4741 and D273-10B/A/H/U strains were generated that express C-terminally triple hemagglutinin-tagged long and short forms of Adk2p, respectively. The correct integration of the tagged gene was verified by PCR analysis of the genomic DNA.
Bacterial Expression, Purification, and Renaturation of Proteins-BL21 (DE3) cells (Stratagene, La Jolla, CA), carrying the plasmid pET21b/Adk2p (short) or pET21b/Adk2p (long), were cultured in M9 medium supplemented with 0.1 mg/ml ampicillin at 37°C. Cells were switched to 30°C prior to induction of proteins by the addition of isopropyl-1-thio-␤-D-galactopyranoside to 1 mM. For some experiments, radiolabeled proteins were desired; hence, protein induction was carried out in the presence of 25 Ci/ml [ 35 S]Met (1175 Ci/mmol, PerkinElmer Life Sciences). Following incubation at 30°C for 3 h, cells were fractionated into soluble fractions and inclusion bodies (IB). 1 Both short and long forms of Adk2p were purified essentially as described (27)(28)(29).
Briefly, inclusion bodies were solubilized with 50 mM Tris/HCl, pH 8.0, containing 8 M urea, and proteins with the His 6 tag were purified on Ni-NTA-agarose (Qiagen, Chatsworth, CA) in the presence of 8 M urea. The elution was done with 50 mM Tris/HCl, pH 8.0, 8 M urea, 0.4 M imidazole. The eluate was dialyzed against buffer A (10 mM Hepes/ KOH, pH 7.5, 6 mM MgCl 2 , 50 mM KCl) containing 6 M urea and stored in aliquots at Ϫ80°C until further use. For renaturation of the ureadenatured short and long forms of Adk2p, proteins were diluted 10-fold with buffer A containing 1% Triton X-100 but no urea. Following incu-bation at 25°C for 10 min, samples were centrifuged at 15,000 ϫ g for 10 min at 4°C. The supernatant fractions containing the renatured ("Renat") short or the long form of Adk2p will be called S IB/Renat and L IB/Renat , respectively. Adk2p (long) was also purified from the soluble fraction of the bacterial lysate as described above except that no urea was used. The purified protein was stored at Ϫ80°C in the presence of 20% glycerol until further use. Such a preparation will be referred to as "native" Adk2p (long) or L Nat . Protein concentrations of the native and "renatured samples were determined using the Micro BCA protein assay kit (Pierce).
GTP:AMP Phosphotransferase Activity-Unless otherwise stated, the assay mixture (20 l) contained Adk2p in 20 mM Hepes/KOH, pH 7.5, 0.6 M sorbitol, 0.1 mg/ml bovine serum albumin, 5 mM KCl, 6 mM MgCl 2 , 1 mM dithiothreitol, 0.06 M urea, 0.1% Triton X-100, 1 mM AMP, 1 mM unlabeled GTP plus 1 Ci of [␣-32 P]GTP (ϳ3000 Ci/mmol, PerkinElmer Life Sciences). Following incubation at 30°C for 15 min, the reaction was stopped by the addition of acetic acid to a final concentration of 10 mM. One-tenth of the reaction mixture was spotted onto polyethyleneimine cellulose thin layer chromatography plates (Sigma) and dried with a hair dryer. The plates were then developed with 1 M LiCl 2 and 1 M acetic acid (30), air-dried, and exposed to film for 15-60 min.
Spots on the autoradiogram were quantitated using the FluorChem 8000 imaging software package (Alpha Innotech Corp., San Leandro, CA). For each sample, the densitometric ratio of GDP/(GDPϩGTP) was calculated. The activity of the sample with the highest ratio was considered 100%, and the relative activity of other samples was then normalized. For experiments in Fig Proteolytic Stability-Radiolabeled S IB/Renat or L IB/Renat (4 g/ml) was incubated with increasing concentrations of trypsin (2-20 g/ml) in buffer A containing 1% Triton X-100 and 0.48 M urea for 30 min on ice. Trypsin was inactivated with the addition of phenylmethylsulfonyl fluoride (1 mM), and samples were analyzed by SDS-PAGE and autoradiography.
Mitochondrial Protein Import-The plasmids pSP64T/Adk2p (short) and pSP64T/Adk2p (long) were linearized with BamHI. Transcription was carried out using the Ribomax-SP6 kit, and 35 S-labeled proteins were synthesized in reticulocyte lysate using the manufacturer's protocol (Promega, Madison, WI). For import experiments, the postribosomal supernatant containing the newly synthesized radiolabeled protein was used either directly (native) or after treatment with urea (urea-denatured). The urea treatment was performed as described (24). Briefly, an aliquot of postribosomal supernatant (20 l) was mixed with saturated ammonium sulfate (40 l) and incubated on ice for 30 min. Samples were centrifuged, and the pellet was solubilized in 20 mM Hepes/KOH, pH 7.5, containing 8 M urea.
The procedure for isolation and purification of mitochondria has been described elsewhere (31). Import reactions were performed essentially as described (24,29,32,33). Briefly, native or urea-denatured Adk2p (short or long) was incubated with mitochondria (100 g of proteins) in the presence of ATP (4 mM) and GTP (1 mM) at 30°C for 30 min. As a control, mitochondria were pretreated with valinomycin (5 g/ml) to dissipate the membrane potential. For urea-denatured proteins, the final urea concentration in the import reaction was 0.16 M. Following import, reaction mixtures were treated with trypsin (0.2 mg/ml) for 30 min on ice. To inactivate trypsin, samples were diluted with a buffer containing a mixture of protease inhibitors. Mitochondria were reisolated and washed with 10% trichloroacetic acid. Samples were analyzed by SDS-PAGE and autoradiography.

RESULTS
We focused on Adk2p corresponding to two representative yeast strains, BY4741 and D273-10B. The former is very popular because mutant strains deleted for individual non-essential genes are commercially available in this background. The latter is widely used for studying various aspects of mitochondrial biogenesis. ADK2 was amplified by PCR from the genomic DNA of these two strains and sequenced. The ADK2 open reading frame in BY4741 codes for a protein of 225 amino acids and will be referred to as Adk2p (short). In D273-10B, a cytosine residue was found to be inserted into the codon 223 of ADK2. This results in changes of three C-terminal residues (RNY to PKL) and the addition of nine amino acids (LN-LITSKFG) at the C terminus of the protein; this extended form will be called Adk2p (long) (Fig. 1A).
To study the molecular properties of the short and long forms of Adk2p, we took advantage of recombinant proteins. Both forms were individually expressed in bacteria with a C-terminal His 6 tag. The C-terminal tag was preferred over an Nterminal tag because most mitochondrial precursor proteins contain the targeting information at their N termini, and a C-terminal His 6 tag usually does not interfere with the import and functions of mitochondrial proteins (27,28). Adk2p (short) was found to be sequestered in the inclusion bodies (Fig. 1B,  lanes 3-5). In contrast, Adk2p (long) was distributed equally between the soluble fraction and inclusion bodies (Fig. 1B,  lanes 8 -10). Short and long forms of Adk2p from inclusion bodies were solubilized with 8 M urea and purified to homogeneity by Ni-NTA-agarose chromatography (Fig. 1B, lanes 6 and  12, respectively). The soluble form of Adk2p (long) was purified exactly the same way (Fig. 1B, lane 11) except that urea was omitted from all buffers; this purified preparation will be referred to as native Adk2p (long) or L Nat .
Proteins from the inclusion bodies were purified in the presence of high concentrations of urea and hence considered denatured. These proteins must therefore be allowed to renature before they can be used for enzymatic studies. To achieve this goal, urea-denatured proteins were diluted 10-fold with buffer containing a non-ionic detergent (Triton X-100) and centrifuged to determine whether they remained in solution (27,28). In the absence of detergent, proteins were aggregated and found exclusively in the pellet fractions (data not shown). In 1% Triton X-100, a major portion (Ͼ80%) of both short and long forms of Adk2p remained in the supernatant fractions (Fig.  1C). The soluble short and long forms of Adk2p thus obtained (purified from inclusion bodies and subsequently renatured in Triton X-100) will be called S IB/Renat and L IB/Renat , respectively.
To measure adenylate kinase activity, most previous studies used a spectrophotometric coupled assay system for the determination of nucleoside diphosphates with pyruvate kinase and lactate dehydrogenase (for example, Ref. 34). This method is quite sensitive and reliable at physiological temperatures. However, at higher temperatures, the coupled enzymes may be inactivated, and thus the results may not truly reflect adenylate kinase activity. To examine GTP:AMP phosphotransferase activity of Adk2p, we therefore used an assay that does not depend on other enzymes. We modified a highly sensitive radioactive assay that is normally used to monitor hydrolysis of nucleotides (30). Native Adk2p (long) was incubated with [␣-32 P]GTP in the absence or presence of unlabeled AMP at 30°C, and samples were analyzed by thin layer chromatography followed by autoradiography and densitometric quantitation of the autoradiogram. The conversion of [␣-32 P]GTP to [␣-32 P]GDP was observed only in the presence of the enzyme and AMP (Fig. 2A). As little as 50 ng/ml native Adk2p (long) could be easily assayed (data not shown). Using this method, both L IB/Renat and S IB/Renat were found to be active. The relative activity of L IB/Renat was very similar to that of the native enzyme (L Nat ) and was ϳ1.6 times the value calculated for S IB/Renat . The specific activity of native Adk2p (long) at 30°C was calculated to be 100 mol of GTP utilized (or GDP formed) ⅐ min Ϫ1 ⅐ (mg of protein) Ϫ1 . The activity of all three forms of Adk2p (L Nat , L IB/Renat , and S IB/Renat ) was specific for GTP, as they failed to efficiently use ATP as the high energy phosphate donor (Fig. 2B). No significant formation of [␤-32 P]ADP was detected in the presence of Adk2p, [␥-32 P]ATP, and unlabeled AMP. Adk1p is an ATP:AMP phosphotransferase, and bacterially expressed and purified Adk1p (data not shown) served as a positive control for these assays. These results suggest that the presence of the modified C-terminal domain of Adk2p (long) may slightly enhance its GTP:AMP phosphotransferase activity at 30°C, but this domain is not essential for activity or for conferring specificity to GTP as the preferred phosphate donor. For the purpose of comparisons, most experiments in the following sections were performed with L IB/Renat and S IB/Renat because both were purified from inclusion bodies and renatured in an identical manner.
Protein function often relies upon a delicate balance between protein stability and flexibility. ATP-specific adenylate kinases from different organisms vary greatly with respect to their optimum temperatures for activity and their stability against FIG. 1. Bacterial expression, purification, and renaturation of the short and long forms of Adk2p. A, ADK2 was amplified by PCR from genomic DNA isolated from the yeast strains BY4741 and D273-10B and sequenced. Compared with the ADK2 of BY4741, a cytosine residue was introduced into the codon 223 of ADK2 in D273-10B. Adk2p (short) and Adk2p (long) represent the corresponding proteins in BY4741 and D273-10B, respectively. B, BL21 (DE3) cells carrying the plasmid pET21b/Adk2p (short or long) were incubated in the absence (U, uninduced) or presence (I, induced) of isopropyl-1-thio-␤-D-galactopyranoside. The overexpressed Adk2p (short)-His 6 was found in the inclusion bodies with very little, if any, of the protein in the soluble fraction (Sol). Adk2p (long)-His 6 was distributed equally between the soluble fraction and inclusion bodies. Proteins were purified on Ni-NTA-agarose; P IB and P Sol indicate proteins purified from inclusion bodies and the soluble fraction, respectively. Samples were analyzed by SDS-PAGE and Coomassie Blue staining. The molecular mass (kDa) of the protein markers (M) are indicated. C, purified short and long forms of Adk2p from stocks in 6 M urea (T, total) were diluted 10-fold with buffer containing 1% Triton X-100 and centrifuged to obtain supernatant (S) and pellet (P) fractions. Samples were analyzed by SDS-PAGE and Coomassie Blue staining. thermal denaturation (20,(35)(36)(37)(38). Such studies for the GTPspecific isozymes are lacking, and we therefore assayed activity of the short and long forms of Adk2p at different temperatures. S IB/Renat exhibited optimum activity at 30°C in agreement with the optimum growth of most wild type yeast strains at this temperature. It was, however, extremely sensitive to thermal denaturation and was inactivated with an increase of 5°C from the optimum temperature (Fig. 3, left panel). The situation with Adk2p (long) was strikingly different. L IB/Renat displayed maximum activity over a wide temperature range (35-55°C) with 66% of the maximum activity at 30°C. A residual activity of ϳ16% was observed even at 65°C (Fig. 3, right panel). Thus, compared with Adk2p (short), Adk2p (long) is much more resistant to thermal denaturation. An increase of ϳ25°C in thermal stability is quite remarkable and suggests that the presence of the modified C-terminal domain in Adk2p (long) is critical for protein stability and/or flexibility.
To further examine protein stability, we tested Adk2p activity in the presence of increasing concentrations of a protein denaturant such as urea. The activity of S IB/Renat was greatly reduced in urea concentration as low as 0.6 M and was completely abolished in 1.6 M urea (Fig. 4A, left panel). In contrast, the activity of L IB/Renat remained unaffected in the presence of 3.6 M urea and was only moderately affected in a urea concentration as high as 4.6 M. A complete inactivation of L IB/Renat was observed at 5.6 M urea (Fig. 4A, right panel). These data further substantiate that Adk2p (long) is much more stable than Adk2p (short); the presence of the modified C-terminal domain in Adk2p (long) may contribute to and strengthen the folding of the enzyme. We also compared the protease sensitivity of the short and long forms of Adk2p. As shown in Fig. 4B, left panel, S IB/Renat was completely degraded by the lowest concentration (2 g/ml) of trypsin tested. In contrast, a significant portion of L IB/Renat remained undigested even at the highest concentration (20 g/ml) of trypsin examined (Fig. 4B, right panel). These results suggest that Adk2p (long) is tightly folded and thus resistant to degradation by trypsin. The absence of the modified C-terminal domain in Adk2p (short) prevents the protein from being tightly folded.
Adk2p is synthesized on cytoplasmic ribosomes and is imported into the matrix side of the mitochondrial inner membrane. It is well established that proteins must be at least partially unfolded prior to import into mitochondria (27, 39 -41). Because the short and long forms of Adk2p appear to differ in their folding status, we investigated their import into isolated mitochondria (Fig. 5A). 35 S-Labeled native proteins were synthesized in reticulocyte lysate and incubated with isolated mitochondria under import conditions. Samples were treated with trypsin and analyzed by SDS-PAGE and autoradiography. Adk2p (short) was efficiently imported and remained protected from external trypsin (Fig. 5A, left panel, top row). Note that unlike most proteins that are targeted to the matrix, Adk2p does not contain a cleavable targeting signal; hence, there was no change in the molecular mass of the protein after import. Import was strictly dependent on a membrane potential because no protease-protected Adk2p molecules were observed in the presence of valinomycin, which dissipates the membrane potential. Unlike in the case of native Adk2p (short), native Adk2p (long) failed to be imported (Fig. 5A, left panel, bottom  row). However, urea-denatured Adk2p (long) was imported as efficiently as Adk2p (short) (Fig. 5A, right panel). In agreement with these in vitro findings are the in vivo observations that regardless of the strain background, the steady state endogenous levels of Adk2p (short) in mitochondria were 5-10-fold greater than those of Adk2p (long) (Fig. 5B). These results suggest that the tight folding of Adk2p (long) may not allow its efficient import into mitochondria. We conclude that the modified C-terminal domain in Adk2p (long) plays important roles in protein folding and stability.

DISCUSSION
Depending on the yeast strains, the GTP:AMP phosphotransferase exists as either Adk2p (short) or Adk2p (long) in the mitochondrial matrix. This is because of a single base pair deletion/insertion near the 3Ј end of the open reading frame of the gene ADK2 that encodes the enzyme. The long form contains a change of three C-terminal residues and the addition of nine amino acids at the C terminus of the protein (Fig. 1A). Our data show that at the physiological temperature for yeast growth (30°C), both the short and long forms of bacterially expressed and purified Adk2p are functional, the latter being ϳ1.6-fold more active (Fig. 2A). Both forms of Adk2p use GTP as the preferred phosphate donor over ATP (Fig. 2B). The modified C-terminal domain in Adk2p (long) is therefore not essential for enzyme activity or for conferring GTP specificity. It, however, contributes to and strengthens protein folding and hence is critical for protein stability under harsh conditions. The role of the modified C-terminal domain in the tight folding of Adk2p (long) was documented by four independent lines of evidence: resistance to thermal inactivation (Fig. 3), resistance to urea denaturation (Fig. 4A), resistance to protease degradation (Fig. 4B), and inability to be imported into isolated mito-FIG. 2. Both short and long forms of Adk2p are enzymatically active. A, the GTP:AMP phosphotransferase activity of native Adk2p (long) or L Nat was compared with the activities of L IB/Renat and S IB/Renat . L Nat was purified from the soluble fraction of the bacterial lysate under non-denaturing conditions. The other two samples (L IB/Renat and S IB/ Renat) indicate the long and short forms of Adk2p, respectively, that were purified from inclusion bodies in the presence of urea and subsequently renatured in the presence of Triton X-100. All three proteins were used at a final concentration of 0.5 g/ml in a total volume of 20 l, and the assays were performed at 30°C for 15 min. One-tenth of each reaction mixture was analyzed by thin layer chromatography followed by autoradiography. B, the ATP:AMP phosphotransferase activity was examined exactly as in A except that unlabeled and ␣-32 Plabeled GTP was replaced with unlabeled and ␥-32 P-labeled ATP. The ATP-specific adenylate kinase Adk1p was used as a positive control at a final concentration of 0.5 g/ml. chondria without prior unfolding by high concentrations of urea (Fig. 5A). In agreement with these in vitro findings, geneswapping experiments show that regardless of the yeast strains used, the steady state levels of endogenous Adk2p (long) in mitochondria are 5-10-fold lower compared with those of Adk2p (short) (Fig. 5B).
Previous studies have concluded that Adk2p (short) is nonfunctional (22,23). This was based on two negative results. First, Adk2p (short) failed to restore growth of a temperaturesensitive adk1 E. coli mutant at the non-permissive temperature (39°C). Our data showing that Adk2p (short) is inactivated at 35°C (Fig. 3, left panel) now explains the earlier results. Second, bacterially expressed and purified Adk2p (short) with an N-terminal His 6 tag did not show any enzymatic activity (23). This issue is difficult to explain because details for these experiments are not available from the published papers. We find that bacterially expressed Adk2p (short) is sequestered in inclusion bodies (Fig. 1B). In previous studies, some of the conditions for protein purification might have been incompatible with the enzyme activity. We would like to emphasize that the position of the His 6 tags, whether at the N or the C terminus of the protein, does not influence enzyme activity. For example, Adk2p (short) with the N-terminal His 6 is as active as Adk2p (short) with the C-terminal His 6 at 30°C (data not shown). Furthermore, the presence or absence of the C-terminal His 6 tag does not influence the tight folding of Adk2p (long). For example, Adk2p (long) with the His 6 tag is tightly folded as judged by its resistance to urea denaturation and degradation by trypsin (Fig. 4, L IB/Renat ). Native Adk2p (long) synthesized in reticulocyte lysate does not contain a tag, but it also appears to be tightly folded as judged by its inability to be imported unless denatured by urea (Fig. 5A).
ATP-specific adenylate kinases from different organisms have been extensively studied (7, 36 -38, 42-50). In contrast, studies with GTP:AMP phosphotransferases have so far been confined mostly to their mitochondrial localization and substrate specificity and kinetic characterization of various mutants (9,10,12,23,34). The molecular properties of the GTP: AMP phosphotransferase, particularly in the context of protein folding and thermal tolerance, have not been studied. S. cerevisiae D273-10B and more distantly related species such as S. diastaticus and S. capensis contain the long form of Adk2p (23), which may be advantageous for these strains to cope with various stress conditions. Data presented here will certainly help in elucidating Adk2p functions under normal as well as stress conditions in different yeast strains. The major finding of this report is that the presence of the modified C- terminal   FIG. 3. Adk2p (long) is quite resistant to thermal inactivation. The GTP:AMP phosphotransferase activity of S IB/Renat and L IB/Renat was assayed at different temperatures. Both proteins were used at a final concentration of 2.5 g/ml. For details, see legend for Fig. 2A.   FIG. 4. Adk2p (long) is resistant to urea denaturation and degradation by trypsin. A, the activity of S IB/Renat and L IB/Renat was assayed at 30°C in the presence of increasing urea concentrations. Both proteins were used at a final concentration of 2.5 g/ml. For details, see legend for Fig. 2A. B, radiolabeled S IB/Renat and L IB/Renat were treated with increasing concentrations of trypsin for 30 min on ice. Samples were analyzed by SDS-PAGE and autoradiography.
FIG. 5. Native Adk2p (long) is not efficiently imported into mitochondria. A, 35 S-labeled short and long forms of Adk2p were synthesized in a reticulocyte lysate cell-free translation system. In one set of experiments, postribosomal supernatants containing the newly synthesized proteins were used directly (Native). In another set, proteins were treated with 8 M urea prior to their use in import reactions (Urea-denatured). Import assays were performed with purified mitochondria in the presence of ATP and GTP. Valinomycin was included as indicated. Before and after trypsin treatment, samples were analyzed by SDS-PAGE and autoradiography. Lanes 1 and 5 indicate 50% of the proteins used for import assays. B, using an integration-based gene-swapping approach, BY4741 and D273-10B strains were generated that expressed either Adk2p (short) or Adk2p (long) with a C-terminal triple hemagglutinin tag. Mitochondria (150 g of proteins) isolated from these strains were analyzed by immunoblotting using anti-hemagglutinin (53) and anti-Tom40p (33) antibodies. Tom40p is a mitochondrial outer membrane marker protein (28,33) and served as a loading control.
domain not only facilitates the tight folding of Adk2p but also confers thermal stability. Adk2p (long) shows maximum activity over a wide temperature range (35-55°C) with 66% of the maximum activity at 30°C. The resistance of Adk2p (long) to urea denaturation and protease degradation and its activity over a wide temperature range (15-55°C) make it an attractive model for studying the role of protein flexibility in enzyme activity. At this stage, we predict that the modified C-terminal domain in Adk2p (long) participates in stabilization mechanisms such as hydrophobic interactions, hydrogen bonds, and salt bridges. Molecular modeling studies are currently in progress in our laboratory to test this hypothesis.
Adenylate kinase in E. coli and Adk1p in S. cerevisiae use ATP as the preferred phosphate donor. Adk2p (long) can rescue the growth phenotype of the temperature-sensitive adk1 E. coli mutant, but it cannot complement the petite phenotype of the ⌬adk1 yeast mutant (23). Although Adk1p is localized to the cytosol and mitochondrial intermembrane space (20), Adk2p resides in the mitochondrial matrix (23). Together, these observations suggest that the compartmentalized activities of adenylate kinases in yeast are critical for cellular/mitochondrial functions. It is interesting to note that in the context of both submitochondrial localization and thermal stability, human GTP:AMP phosphotransferase (AK3) resembles the yeast Adk2p (long). Like yeast Adk2p (long), human AK3 is also localized to the mitochondrial matrix and can complement the temperature-sensitive adk1 E. coli mutant (10).
Both ADP and GTP are primarily synthesized in the cytosol. They diffuse across the mitochondrial outer membrane and cross the inner membrane barrier into the matrix via the ADP/ ATP (51) and GDP/GTP (52) carriers. AMP is generated in the mitochondrial matrix through several hydrolytic reactions such as turnover of DNA or RNA (23). Although Adk2p has been proposed to participate in oxidative metabolism (22), its precise role in the mitochondrial matrix has not been established. The mitochondrial matrix is the major site of ATP synthesis; yet Adk2p in the matrix is specific for GTP. The enzyme catalyzes the reversible reaction GTP ϩ AMP N GDP ϩ ADP, and we propose that it serves as a backup for the synthesis of ADP or GTP depending on the metabolic conditions. This way mitochondria do not have to solely depend on the cytosolic supply of ADP and GTP, thereby minimizing disruptions in critical processes such as oxidative phosphorylation and organellar protein synthesis. The function of Adk2p, however, is not essential for cell viability under normal laboratory conditions. Under these conditions, there may be a sufficient supply of ADP and GTP from the cytoplasm to the mitochondrial matrix of ⌬adk2 cells. It would be interesting to determine whether mitochondrial functions in the ⌬adk2 strain become severely impaired when the cytosolic supply of ADP and/or GTP to the matrix falls below a threshold level.