The Molecular Chaperone, Atp12p, from Homo sapiens

Work in Saccharomyces cerevisiae has shown that Atp12p binds to unassembled α subunits of F1 and in so doing prevents the α subunit from associating with itself in non-productive complexes during assembly of the F1 moiety of the mitochondrial ATP synthase. We have developed a method to prepare recombinant Atp12p after expression of its human cDNA in bacterial cells. The molecular chaperone activity of HuAtp12p was studied using citrate synthase as a model substrate. Wild type HuAtp12p suppresses the aggregation of thermally inactivated citrate synthase. In contrast, the mutant protein HuAtp12pE240K, which harbors a lysine at the position of the highly conserved Glu-240, fails to prevent citrate synthase aggregation at 43 °C. No significant differences were observed between the wild type and the mutant proteins as judged by sedimentation analysis, cysteine titration, tryptophan emission spectra, or limited proteolysis, which suggests that the E240K mutation alters the activity of HuAtp12p with minimal effects on the physical integrity of the protein. An additional important finding of this work is that the equilibrium chemical denaturation curve of HuAtp12p shows two components, the first of which is associated with protein aggregation. This result is consistent with a model for Atp12p structure in which there is a hydrophobic chaperone domain that is buried within the protein interior.


Work in
Saccharomyces cerevisiae has shown that Atp12p binds to unassembled ␣ subunits of F 1 and in so doing prevents the ␣ subunit from associating with itself in non-productive complexes during assembly of the F 1 moiety of the mitochondrial ATP synthase. We have developed a method to prepare recombinant Atp12p after expression of its human cDNA in bacterial cells. The molecular chaperone activity of HuAtp12p was studied using citrate synthase as a model substrate. Wild type HuAtp12p suppresses the aggregation of thermally inactivated citrate synthase. In contrast, the mutant protein HuAtp12p E240K , which harbors a lysine at the position of the highly conserved Glu-240, fails to prevent citrate synthase aggregation at 43°C. No significant differences were observed between the wild type and the mutant proteins as judged by sedimentation analysis, cysteine titration, tryptophan emission spectra, or limited proteolysis, which suggests that the E240K mutation alters the activity of HuAtp12p with minimal effects on the physical integrity of the protein. An additional important finding of this work is that the equilibrium chemical denaturation curve of HuAtp12p shows two components, the first of which is associated with protein aggregation. This result is consistent with a model for Atp12p structure in which there is a hydrophobic chaperone domain that is buried within the protein interior.
Atp12p was first identified in studies of Saccharomyces cerevisiae mutants that are respiratory-deficient due to a defect in mitochondrial F 1 assembly. The core structure of F 1 is a hexameric unit of alternating ␣ and ␤ subunits that surrounds a rod-shaped ␥ subunit (1). Yeast atp12 mutants fail to assemble the ␣ 3 ␤ 3 oligomer, and instead accumulate F 1 ␣ and ␤ subunits as large, insoluble aggregates in the matrix of the organelle (2). This particular phenotype is observed also in yeast mutants that lack a functional Atp11p, another molecular chaperone of the mitochondrial F 1 assembly pathway (3). There is also F 1 protein aggregation in yeast null mutants that are missing either the ␣ subunit or the ␤ subunit, as such strains accumulate the lone ␤ or ␣ subunit as aggregated proteins inside mitochondria (2). Because lack of the ␥ subunit does not produce aggregation of ␣ and ␤ subunits (4), it is believed that ␥ assembles into the F 1 structure after Atp11p-and Atp12pmediated steps have secured the formation of a soluble ␣␤ intermediate (3).
The amount of yeast Atp12p in the mitochondrial matrix is roughly 100 times smaller than the amount of F 1 protein (5). This molar ratio seems appropriate in consideration of the fact that Atp12p does not bind to ␣ subunits that are part of the F 1 oligomer but, instead, associates with unassembled ␣ subunits. Because unassembled F 1 proteins do not accumulate to a significant degree in the cell (6), the Atp12p concentration in mitochondria may be comparable with the concentration of free F 1 ␣ subunit protein. Therefore, the amount of Atp12p is not likely to be a limiting factor in mitochondrial F 1 biogenesis under normal conditions in the cell. However, reducing the level of active Atp12p in mitochondria is anticipated to correlate directly with a decreased flux through the F 1 assembly pathway. Such considerations underscore the important role of human Atp12p (HuAtp12p) and its potential ties to mitochondrial disease.
We have now developed a method to purify recombinant HuAtp12p in high yield from a bacterial expression system, and this has enabled us to expand the analysis of the physical and functional properties of the molecular chaperone beyond what was formerly achieved in studies of the yeast protein in mitochondrial samples. One significant aspect of the work has been to evaluate wild type HuAtp12p in comparison with a mutant form of the protein that harbors a Glu 3 Lys substitution at a conserved position in the amino acid sequence. In particular, we show that the wild type HuAtp12p behaves in vitro as a molecular chaperone, whereas the mutant HuAtp12p E240K is defective in this capacity. Despite this clear functional difference, the wild type and mutant proteins are reported to be similar with respect to sedimentation properties, intrinsic fluorescence under native conditions, cysteine accessibility, and protease sensitivity. Spectroscopic analysis of HuAtp12p unfolding in guanidine hydrochloride is also presented, which suggests the presence of a hydrophobic subdomain in the protein structure.

EXPERIMENTAL PROCEDURES
Plasmid Construction-Recombinant plasmids for the production of human Atp12p (HuAtp12p) in Escherichia coli were constructed in the His-tag vector pPROEX HTa (Invitrogen) and were designed to produce the mature form of the protein without the mitochondrial leader peptide. The DNA for wild type HuAtp12p was prepared using the PCR to amplify an 806-bp fragment inclusive of HuAtp12p codons 33-289, start and stop codons, and flanking restriction sites for cloning. The forward and reverse primers were, respectively, 5Ј-CCGGAATTCATGATCCC-GTCTCCAGCCCGGGCTTAC-3Ј and 5Ј-CCGAGGTACCGGTCACT-CCTTCAGGAGCTTGTGCTT-3Ј and are annotated to highlight the EcoRI and KpnI restriction sequences (underlines), ATG and TGA codons (double underlines), and HuAtp12p coding sequences (bold) contained within. The plasmid template was pCUPHUATP12/YEp (7). The PCR product and pPROEX HTa vector were each digested with a combination of EcoRI and KpnI, and the linear fragments were ligated and used to transform E. coli strain RRI. The resultant new plasmid (pPROEX /Atp12h) was used to prepare wild type HuAtp12p for this  work and also served as the template for PCR-mediated site-directed  mutagenesis via the Stratagene QuikChange method to introduce a  Glu 3 Lys change at position 240 in the recombinant protein. The two  complementary oligonucleotides used for mutagenesis were sensestrand primer 5Ј-GCCGTGCTGCTGTCTAGACTGAAGGAGGAGTAC-CAGATC-3Ј and antisense-strand primer 5Ј-GATCTGGTACTCCTCC-TTCAGTCTAGACAGCAGCACGGC-3Ј and encompassed HuAtp12p DNA sequence 697-735 with three silent mutations (underline) to create a XbaI site for screening purposes and a G 3 A (bold) mutation to convert Glu-240 to lysine. The PCR reaction (50 l) contained 20 mM Tris-HCl, pH 8.8, 10 mM KCl, 10 mM NH 4 (SO 4 ) 2 , 2 mM MgSO 4 , 0.1% Triton X-100, 0.1 mg/ml bovine serum albumin, 50 ng of pPROEX/ATP 12-h plasmid (DNA template), and 125 ng of each primer. After the reaction was incubated at 95°C for 2 min and then allowed to slowly cool to room temperature, 3 l of QuikSolution (Stratagene), 6 l of 2 mM deoxynucleotide triphosphate mixture, and 1 l of Pfu Turbo (2.5 units) were added, and the tube was incubated at 68°C for 5 min. Next, PCR was performed for 18 cycles (95°C for 50 s, 60°C for 50 s, 68°C for 11 min). The resultant mixture was digested with DpnI at 37°C for 1 h to destroy the original plasmid DNA template, after which 10 l was used to transform competent E. coli cells. Bacterial colonies harboring the mutant plasmid (called pPROEX/Atp12hE240K) were identified by restriction analysis in an EcoRI/XbaI digest and tested in small scale experiments for isopropyl ␤-D-thiogalactopyranoside-induced production of recombinant protein. One such expressing clone was selected and used for large scale production/purification of the mutant Atp12p protein (HuAtp12p E240K ). DNA sequencing confirmed that only the desired mutation is present in the plasmid insert of pPROEX/ Atp12phE240K.
Purification of Recombinant HuAtp12p Wild Type and Mutant Proteins-HuAtp12p and HuAtp12p E240K were overproduced in E. coli RRI cells and purified according to the following method. An overnight LB ampicillin (40 mg/liter) culture of plasmid-bearing cells was diluted 200 times into 1 liter of 2XYT (1% yeast extract, 1.6% tryptone, 0.5% NaCl) plus ampicillin, and the cells were grown at 37°C until A 600 nm ϭ 0.5, at which time the culture was supplemented to include 1 mM isopropyl ␤-D-thiogalactopyranoside and incubated overnight with shaking at 37°C. The next morning the cells were harvested by centrifugation (4000 ϫ g, 10 min, 4°C), resuspended in 40 ml of 20 mM Tris-HCl pH 7.5, and sonically irradiated in a 50-ml beaker for 30 s, 5 times, at maximum power output using a Branson Sonifer model 450. Phenylmethylsulfonyl fluoride was added to 1 M during cell breakage to prevent proteolysis, and lysozyme was added at 0.2 mg/ml to enhance breaking. After the addition of streptomycin sulfate to 1% final concentration, the disrupted cells were centrifuged at 145,000 ϫ g for 1 h at 4°C, and the resultant supernatant was filtered through a 0.45-m syringe disc, diluted 3-times with 20 mM Tris-HCl, pH 7.5, and applied to a DEAE column (8.7 ϫ 2.5 cm). The column was developed with a salt gradient in the same buffer that contained 0 -1 M NaCl. Column fractions containing HuAtp12p were pooled and dialyzed for 1 h against 20 mM Tris-HCl, pH 7.5, to lower the salt concentration, and the dialysate was loaded on a cobalt column (Talon resin, Clontech; 5.1 ϫ 1.5 cm) previously equilibrated with 20 mM Tris-HCl, pH 7.5. The cobalt column was washed with 50 ml of 20 mM Tris-HCl, pH 7.5, 40 mM imidazole, 0.3 M NaCl buffer, and His-tag HuAtp12p was eluted with 20 mM Tris-HCl, pH 7.5, 0.2 M imidazole, 0.3 M NaCl. Final preparations of HuAtp12 and HuAtp12 E239K were dialyzed overnight versus 20 mM Tris-HCl, pH 7.5, brought to 10 -15 mg/ml using Centricon 10 concentrators, flash-frozen in liquid nitrogen, and stored at Ϫ80°C.
Sedimentation Analysis-Sucrose solutions were prepared in 20 mM Tris-HCl, pH 8.0, 1 mM EDTA, 150 mM NaCl, 0.1% Triton X-100. Linear gradients (4.4 ml) of 7-20% (w/v) sucrose were overlaid on top of 0.4 ml of 80% (w/v) sucrose in 5-ml ultracentrifuge tubes. Protein samples (0.2 ml) applied to the gradients contained 4 g of purified recombinant HuAtp12p (wild type or mutant) in a mixture with molecular weight standards (M r horse hemoglobin ϭ 64,500) and (M r myokinase ϭ 21,000). The tubes were centrifuged in a Beckman SW55Ti rotor at 37,800 rpm for 18 h at 4°C, and gradient fractions (300 l) were collected from the bottom of the tube. The fractions were probed by Western analysis with a polyclonal antibody raised against yeast Atp12p and assayed for peak positions of hemoglobin and myokinase as previously described (8).
Aggregation Suppression Experiments-Measurements of thermally induced aggregation of citrate synthase (CS) 1 were carried out essentially as described (9). Porcine heart CS (Sigma) was diluted to 0.075 M in 1 ml of 40 mM HEPES, pH 7.5, buffer that was prewarmed to 43°C in a thermostatted cell of a SLM Aminco-Bowman Series 2 fluorescence spectrometer. Right angle light scattering was recorded over time at 43°C with excitation and emission wavelengths set at 465 nm and slit widths at 4 nm. HuAtp12p (wild type or mutant) or other additives were included in the incubation at the concentrations indicated for each experiment. CS concentration was determined using the extinction coefficient of 1.78 for a 1 mg/ml solution of the dimer at 280 nm (10).
Proteolytic Digestion-For analytical experiments 5 g of wild type or mutant HuAtp12p were mixed with serial dilutions of trypsin or chymotrypsin ranging from 0.01 to 3% in 20 l of 10 mM Tris-HCl, pH 7.5. Proteolysis was allowed to proceed for 30 min at 37°C, at which point the reactions were stopped with the addition of trichloroacetic acid to 10%. Precipitated protein was collected by centrifugation in a microcentrifuge, suspended in SDS gel buffer, and resolved in a 12% SDS-polyacrylamide gel. To identify the principal proteolytic fragment of highest molecular weight, a 100-g sample of HuAtp12p was first digested with either trypsin or chymotrypsin at 1% in 67 l, and after the reaction was quenched, the acid-precipitated protein was resolved in an SDS gel, and then transferred to polyvinylidene difluoride membrane (Schleicher & Schuell, WESTRAN®) as described (11). The amino-terminal residues of the immobilized proteolytic fragment were identified by Edman degradation at the protein sequencing facility of the Department of Genetics at the University of Georgia.
Chemical Denaturation Studies and Data Analysis-Wild type or mutant HuAtp12p (1 M) was incubated in 20 mM Tris-HCl, pH 8.0, 150 mM NaCl supplemented with increasing amounts of guanidine hydrochloride (GdnHCl) at 23°C for 3 h. Equilibrium was reached during this time period. The concentration of GdnHCl in each experimental sample (0 -5 M range) was determined from its refractive index (12). Intrinsic fluorescence emission spectra were recorded between 300 and 400 nm after excitation at 280 nm. Slit widths were set at 4.0 nm.
The red shift in intrinsic fluorescence emission spectra at increasing GdnHCl concentrations were quantified as the intensity-averaged emission wavelength, avg (13), calculated according to Equation 1, where i and I i are the emission wavelength and its corresponding fluorescence intensity at that wavelength. Base-line and transitionregion data for the first and second component of the HuAtp12p GdnHCl equilibrium denaturation curve were fitted to a two-state linear-extrapolation model (14) according to Equation 2, where ⌬G unfolding is the free energy change for unfolding at a given denaturant (GdnHCl) concentration, ⌬G H2O is the free energy change for unfolding in the absence of denaturant, m is a slope term that equates the change in ⌬G unfolding per unit concentration of GdnHCl, R is the gas constant (1.987 cal mol Ϫ1 K Ϫ1 ), T is the temperature (296.15 K), and K unfolding is the equilibrium constant for unfolding. The linear extrapolation model expresses the signal ( avg ) as a function of GdnHCl concentration according to Equation 3 (15), where y i is the observed signal, y N and y D are the base-line intercepts corresponding to, respectively, native and denatured proteins, m N and m D are the corresponding base-line slopes, [X] i is the denaturant concentration corresponding to the ith addition, m is the slope of a ⌬G unfolding versus [X] plot, R and T are the gas constant and temperature as defined above.
[D] 50% is the denaturant (GdnHCl) concentration at the midpoint of the transition, in which there is an equal amount of the folded and unfolded protein (K unfolding ϭ 1). Therefore, from Equation 2, it follows that The change in free energy of unfolding on mutation, ⌬⌬G H2O , between wild type and mutant HuAtp12 can be calculated from the difference between the ⌬G H2O values derived from Equation 4.
Miscellaneous Procedures-Citrate synthase activity was measured at 23°C following the procedure of Srere (16), with the exception that the buffer was 0.1 M Tris-HCl, pH 8.0. The number of free sulfhydryl groups in HuAtp12p was determined with the Ellman reagent, 5,5Јdithiobis(2-nitrobenzoic acid), using an extinction coefficient at 412 nm of 13,400 M Ϫ1 for the 5-nitro-2-thiobenzoate liberated in the reaction as previously described (17). SDS-polyacrylamide gel electrophoresis followed the method of Laemmli (18). Western analysis and detection of immuno-decorated Atp12p by chemiluminescence was as described (5). Unless indicated otherwise, protein was estimated using the procedure of Lowry et al. (19). Emission spectra for excited tryptophans in HuAtp12p were recorded between 300 nm and 400 nm for 1-ml samples of protein in 20 mM Tris-HCl, pH 7.5.

RESULTS
Purification of Recombinant HuAtp12p-The complete cDNA for HuAtp12p encodes a protein with 289 amino acids, which includes a mitochondrial targeting sequence at the amino terminus (7). Cleavage of the precursor protein to the mature form is estimated to occur at/near the junction of Thr-32 and Ile-33 following its passage into mitochondria (7). In previous work we have shown that the polypeptide fragment of HuAtp12p encompassing Ile-33 through Glu-289 closely approximates the true mature form of the protein since this fragment rescues the respiratory defect of a yeast ⌬atp12 deletion strain and binds to the F 1 ␣ subunit in a yeast two-hybrid screen (7). In the present work we have extended our studies of Atp12p function to experiments that explore features of the purified protein. The first step in this project has been to subclone codons 33-289 of HuAtp12p cDNA in a bacterial expression vector to create the plasmid pPROEx/Atp12ph (see "Experimental Procedures"). Induction of expression from pPROEx/Atp12ph leads to high production of His-tagged HuAtp12p (M r ϭ 32,770) in E. coli (Fig. 1, lane 1) of which 60 -70% partitions to the soluble fraction after centrifugation of sonically irradiated cells (Fig. 1, lane 2). The protein is purified to near homogeneity after successive chromatographic steps through DEAE and cobalt columns (Fig. 1, lane 3).
Previous work with yeast Atp12p identified a Glu 3 Lys mutation that inactivates the protein in vivo (8). To study the effect of this mutation in vitro we performed site-directed mutagenesis of the pPROEx/Atp12ph plasmid insert to convert the corresponding glutamic acid (Glu-240) of HuAtp12p to lysine (see "Experimental Procedures"). Highly pure preparations of the His-tagged mutant protein (HuAtp12p E240K ) (Fig. 1, lane 4) are obtained following the same method described above for wild type recombinant HuAtp12p purification.
Sedimentation Analysis of HuAtp12p-Samples of purified wild type or mutant HuAtp12p were centrifuged through 7-20% sucrose gradients that were formed on top of an 80% sucrose solution deposited at the bottom of the centrifuge tube. The behavior of both proteins relative to molecular weight standards was found to be similar with a single peak at the predicted location for a 32.7-kDa monomer (Fig. 2). The absence of immunologically detectable signal in fractions 2 and 3 of the gradient shows that there are no large aggregates of HuAtp12p present in the protein preparations, as these would have been trapped at the 20:80% sucrose interface (black arrow).
Molecular Chaperone Assays with HuAtp12p-We have proposed that the true substrate of Atp12p in mitochondria is a folded F 1 ␣ subunit monomer (5). Atp12p is believed to act by temporarily shielding a hydrophobic element of the ␣ subunit surface that will ultimately be buried at the non-catalytic site interface with an adjacent ␤ subunit of the assembled ␣ 3 ␤ 3 oligomer. To study this activity in vitro we initially tried to obtain recombinant F 1 ␣ subunit as a soluble monomer. Because there are no Atp12p homologs in E. coli and there is no evidence from yeast two-hybrid screens of mitochondrial Atp12p interaction with E. coli F 1 ␣, 2 our work was focused on constructing plasmids for the expression of yeast and bovine mitochondrial F 1 ␣. Unfortunately, forms of these recombinant proteins that accumulated either in the cytosol or in the periplasm (via a leader peptide) were not soluble; the former produced inclusion bodies, and the latter were found to be tightly associated with the membrane, requiring harsh alkaline conditions for extraction. 2 Ultimately, efforts to recover the ␣ subunit from urea-denatured inclusion bodies in a conformation that would be bound by Atp12p in vitro were not successful. 2 As for using surrogate protein substrates to study the molecular chaperone activity of HuAtp12p and assess the effect of mutations in vitro, we found that, for example, HuAtp12p did not inhibit the aggregation of reduced insulin b-chains, an assay that had proved particularly useful in assessing in vitro the molecular chaperone activity of yeast Atp11p (20). Instead, in our screen of potential substrates we identified mitochondrial CS as a suitable client protein for HuAtp12p. CS is a dimeric protein of 2 identical 48-kDa subunits that catalyzes the formation of citric acid and CoA from oxaloacetate and acetyl-CoA. In the absence of substrates CS is inactivated and rapidly aggregates after exposure to 43°C, a property that has made it popular as a model substrate for studies of molecular chaperone activity in vitro (9,(21)(22)(23)(24). We monitored thermally induced aggregation of CS at 43°C by measuring the apparent fluorescence due to light scattering at 465 nm of solutions containing the enzyme in the absence or presence of other proteins. The presence of wild type HuAtp12p in the mixture has different effects on the CS aggregation phenomenon depending on the molar ratio of molecular chaperone (monomer) relative to CS dimers. At low molar ratios (3:1 and below), the degree of sample light scattering is exacerbated over what is observed for solutions of CS alone (Fig. 3A, data shown for 1:1 and 2:1 ratios). Molar ratios of HuAtp12p:CS in the range of 4:1-7:1 provide conditions under which the rate of CS aggregation is reduced, although the final degree of light scattering is the same as in the absence of HuAtp12p (Fig. 3A, data shown for 4:1 and 5:1 ratios). Raising the level of HuAtp12p to 8-fold molar excess over CS reduces the final level of sample light scattering to ϳ40% of the maximum observed in the absence of effector protein (Fig. 3A, open triangles). Increasing the molar ratio of HuAtp12p:CS above 8:1 does not further improve the level of protection (data not shown). The amount of CS and HuAtp12p protein recovered in the particulate fraction after centrifugation of 1:1 and 8:1 protein mixtures exposed to 43°C is shown in the inset of Fig. 3A. Bovine serum albumin fails to prevent thermally induced aggregation of CS when the effector protein is present at 8-fold or even 16-fold molar excess relative to CS dimers (Fig. 3B). Companion studies examined HuAtp12p for an affect on preventing loss of CS enzymatic activity at 43°C. We found that the kinetics of CS inactivation is unaltered in the presence of the chaperone (Fig. 3C).
In contrast to the wild type protein, the mutant HuAtp12p E240K does not prevent the aggregation of CS at 43°C when present in an 8-fold molar excess relative to the enzyme dimer concentration in the cuvette (Fig. 4). Increasing the molar ratio to 16:1 increased the time necessary to observe maximal CS aggregation, but the final degree of light-scattering observed for the sample was the same as in the absence of HuAtp12p.
Physical Characterization of Wild Type and Mutant HuAtp12p-Several parameters were measured for wild type HuAtp12p and HuAtp12p E240K to determine whether gross alterations of the three-dimensional structure of the two proteins could explain the loss of activity in the mutant. Titration of free sulfhydryl groups with 5,5Ј-dithiobis(2-nitrobenzoic acid) gave similar results with both proteins; of the three cysteines in mature HuAtp12p, one is accessible to the reagent under non-denaturing conditions, whereas all three react with the probe after denaturation of either the wild type or mutant protein in 8 M urea. The tryptophan emission spectra of the wild type and mutant proteins were also similar, with no sig- Molar ratios refer to amount of "effector protein" relative to CS dimer. nificant difference in the amplitude or wavelength of maximal fluorescence (data not shown). We also examined the recombinant proteins for sensitivity to proteolytic digestion (Fig. 5). Wild type HuAtp12p (Fig. 5, upper panels) and HuAtp12p E240K (Fig. 5, lower panels) were exposed to increasing amounts of trypsin or chymotrypsin and then loaded on SDS-polyacrylamide gels to analyze the products of digestion. The digestion patterns for wild type and mutant HuAtp12p are essentially the same. Both proteases yield one principal digestion product of ϳ29 kDa, which correlates well with the size of the mature HuAtp12p protein encoded by the expression plasmid. Aminoterminal sequencing of the 29-kDa fragments showed that they result from either trypsin or chymotrypsin cleavage of 7 or 9 amino acids, respectively, downstream from the start of HuAtp12p-coding region (Fig. 5, bottom). Hence, the limited proteolysis only removes the tag sequence at the amino terminus of the recombinant proteins and does not disclose the presence of hypersensitive proteolytic sites in either the wild type or mutant proteins.
Chemical Unfolding Studies-GdnHCl denaturation curves were obtained for wild type and mutant HuAtp12p. Purified recombinant protein was incubated with increasing GdnHCl concentrations at 23°C for 3 h. Exposure to the chaotrope resulted in changes of the intrinsic fluorescence emission from the HuAtp12p proteins that were manifest as a red-shift in maximal emission wavelength. The intensity-averaged emission wavelength ( avg ), an integral measurement that is negligibly influenced by the noise, was calculated for each sample (Equation 1 under "Experimental Procedures") and plotted versus GdnHCl concentration (Fig. 6). Wild type and mutant HuAtp12p both show two distinct and unequal components in the equilibrium denaturation curve, suggesting that either HuAtp12p consists of two domains or that there is a stable unfolding intermediate. Light-scattering measurements indicate that there is protein aggregation at the concentrations of GdnHCl (0 -2.2 M) corresponding to the first component of the equilibrium denaturation curves (Fig. 6, inset) but not to the second (data not shown). Thermodynamic parameters associated with the two parts of the curve were calculated for HuAtp12p and HuAtp12p E240K (Table I)  FIG. 5. Protein gels of HuAtp12p proteins exposed to limited digestion with trypsin or chymotrypsin. Five micrograms of purified wild type (WT) HuAtp12p (upper panels) or mutant HuAtp12p E240K (lower panels) were incubated with increasing amounts of trypsin (TRP, left side) or chymotrypsin (CHY, right side) for 30 min, and the products of proteolysis from acid quenched samples were analyzed in 12% SDSpolyacrylamide gels as described under "Experimental Procedures." The position of full-length recombinant HuAtp12p proteins is marked by arrowheads on the left-hand side. Migration of molecular weight standards is shown on the right-hand side. The arrows point to the ϳ29-kDa fragments of trypsin and chymotrypsin digestion that were produced in preparative scale digestion with the proteases, transferred to polyvinylidene difluoride membrane, and sequenced at the amino terminus (see "Experimental Procedures). The sequence of the amino terminus, beginning with the initiator methionine a few residues upstream from the histidine tag of wild type and mutant HuAtp12p recombinant proteins is shown at the bottom of the figure. The amino acid residues that correspond to HuAtp12p sequence are underlined. The bent arrows show the first amino acid in the ϳ29-kDa fragments produced by digestion of HuAtp12p with 1% trypsin or 1% chymotrypsin. cating the mutant is less stable than the wild type) was calculated for this part of the curve. This value of ⌬⌬G H2O compares well with the range of stability changes (0.17-1.63 kcal/mol) observed for model proteins after the substitution of an alanine residue with all other amino acids (with the exception of proline) in either an ␣-helix or ␤-sheet (see Table 17.3 in Ref. 14). A change in m, which is a constant of proportionality correlating the value of ⌬G unfolding at any denaturant concentration to that of ⌬G H2O , can be easily appreciated between wild type and mutant in the slope of the transition region of the second component of the denaturation curve (Fig. 6). The smaller m value of the E240K mutant (1.47 kcal/mol M Ϫ1 ) versus the wild type (2.21/mol M Ϫ1 ) suggests that in the mutant there may exist a folding intermediate that is more exposed to solvent (14). DISCUSSION Previous studies from our laboratory measured the physical properties for Atp12p as it exists in the yeast mitochondrial matrix. We found that the protein is part of an oligomer of ϳ70 -80 kDa (25), that such oligomerization is dependent on sequence at the carboxyl terminus of the protein (8), that it is the oligomeric form of Atp12p, which is functional in vivo (8), and that the oligomeric state of Atp12p is dependent on the absence or presence of the F 1 ␣ subunit (5). It has not been possible to extend the studies of yeast Atp12p to in vitro analysis because the protein forms inclusion bodies after overproduction in bacteria, and there are doubts about the functional integrity of the refolded protein (8). In contrast human Atp12p (HuAtp12p) remains largely soluble when produced in E. coli, and this has opened the door to experiments that probe the structural and functional properties of this protein in greater detail. A principal objective of the current work has been to analyze a Glu 3 Lys substitution that knocks out Atp12p action in vivo (8). Site-directed mutagenesis was used to change the relevant glutamic acid of HuAtp12p (residue 240) to lysine, and the mutant and wild type recombinant proteins were purified from bacterial hosts (Fig. 1) and studied in parallel.
The Atp12p gene product is a monomeric species as illustrated by the sedimentation profile obtained for purified HuAtp12p (both wild type and E240K mutant) in experiments that were designed to resolve monomers from dimers in the linear portion of a sucrose gradient and to capture high molecular weight assemblies, if present, at the 20%:80% concentration interface (Fig. 2). On this basis we conclude that the oligomeric form of Atp12p observed in mitochondrial extracts is heterogeneous in nature. It is of interest that the molecular weight for Atp12p is shifted toward a lower value in samples that are deficient for the F 1 ␣ subunit (5). However, it remains to be determined if the change in sedimentation behavior reflects the absence of an Atp12p-␣ subunit complex or the loss of Atp12p binding to some other protein(s) under conditions in which the substrate for the chaperone (F 1 ␣) is missing. A possible candidate for an Atp12p binding partner is Fmc1p, a protein of yeast mitochondria that has been linked genetically to Atp12p function (26).
One major challenge to the study of molecular chaperones in a purified system is often the difficulty encountered in trying to reproduce in vitro the physical state of protein substrates that are bound by their chaperones in vivo. In our case the uncontrolled aggregation of recombinant yeast and bovine mitochondrial F 1 ␣ subunits makes them unsuitable as externally added substrates in assays with purified HuAtp12p. Instead, our approach has been to employ a model substrate whose aggregation in solution can be initiated under defined experimental conditions. A screen of model proteins that have been described in the chaperone literature identified CS as a suitable surrogate client for Atp12p. In the absence of bound substrate the porcine heart CS dimer rapidly loses activity and aggregates after exposure of the protein solution to 43°C (27). This property has been exploited for the study of many different chaperone proteins including Hsp90 (9,21), Hsp 31, (24), Hsp33 (23), and Hsp25 (22). Our work shows that wild type HuAtp12p reduces significantly the final degree of CS aggregation at 43°C when the chaperone is present in 8-fold molar excess relative to CS dimers (Fig. 3A). We have observed protection as high as 65% under such conditions. Molar ratios of 12p WT :CS on the order of 4:1 and 5:1 already provide conditions under which the rate of CS aggregation is substantially reduced. A curious finding has been that at low molar ratios of 12p WT :CS (1:1, 2:1, 3:1 (not shown)) the presence of the chaperone is associated with an exacerbated degree of sample light-scattering (Fig. 3A). Inspection of the insoluble material that is collected after incubation of CS with wild type HuAtp12p at 43°C shows that the amount of chaperone protein recovered in the particulate fraction actually correlates negatively with the amount of light scattering observed in the sample before centrifugation (Fig. 3A, inset). Hence, the increased light scattering is not due to increased HuAtp12p aggregation at low molar ratios of chaperone:CS. Instead, we suggest that at concentrations much below the threshold necessary for protection, HuAtp12p interacts non-productively with CS and promotes a higher extent of conversion of denaturing CS dimers into a light-scattering species than would occur in solutions of CS alone. In sharp contrast to the protection afforded by HuAtp12p, the inclusion of bovine serum albumin at 8:1 or 16:1 molar ratio relative to CS has no effect on thermally induced CS aggregation (Fig. 3B). Such results support the view that HuAtp12p behaves as a bona fide molecular chaperone in the CS aggregation assays. However, the inclusion of HuAtp12p at a level known to provide maximal protection against CS aggregation (8:1 molar ratio) did not change the kinetics of thermal inactivation of the enzyme (Fig. 3C). In this respect, HuAtp12p is similar to Hsp25, as also the latter protein prevents CS aggregation at 43°C without preventing inactivation (22). These results are interesting in view of the fact that Hsp90 protects against both thermally induced aggregation and inactivation of CS (9), an effect ascribed to the capacity of Hsp90 to stabilize the native form of CS through frequent binding and release of the early unfolding intermediates that are formed at high temperature (9). In contrast, Atp12p and Hsp25 may form more long-lived complexes with CS intermediates that might interfere with CS acquiring the native (active) conformation (see Ehrnsperger et al. (22) for a discussion of Hsp25). The idea of Atp12p acting as a trap for unassembled protein fits well with its proposed function in mitochondria, where it would hold  (3) and (4).
an aggregation-prone form of the F 1 ␣ subunit in a protected state until the conditions are ready for the next step in the assembly pathway. The phenotype of atp12 yeast mutants has led us to suggest that Atp12p serves a specialized role in the F 1 assembly pathway in mitochondria (3), as opposed to being a "general purpose" chaperone with a broad substrate clientele. In fact, recombinant HuAtp12p shows selectivity even with model proteins as it is completely ineffective in preventing the aggregation of reduced insulin, which is instead a surrogate client protein for Atp11p in vitro (20,28). At this time we do not know the common features between thermally denatured CS and unassembled F 1 ␣ subunits that enable these two proteins to be bound by HuAtp12p. Perhaps, during the thermal denaturation of dimeric CS, hydrophobic regions of the monomers surface that are hidden from solvent at the dimer interface produce an uncontrolled aggregation of the monomers, which is prevented by the shielding properties of HuAtp12p. We also do not have a good explanation for why the maximal effect of HuAtp12p requires that it be present in 8-fold molar excess over the CS dimer. It suffices, however, that the CS aggregation assay has provided us with a tool to assess Atp12p function in vitro and to examine the effect of mutations on Atp12p molecular chaperone activity. To this end we show that the E240K mutation severely alters the function of HuAtp12p. There was no protection by HuAtp12p E240K against CS aggregation at 43°C when the mutant protein was included in 8-fold molar excess (Fig. 4). At best, a 16-fold molar excess HuAtp12p E240K (Fig. 4) slows down aggregation in a manner that is comparable with what is observed with 4-or 5-fold molar excess wild type HuAtp12p (Fig. 3A). These results obtained in vitro are in agreement with the observation that the corresponding yeast mutant Atp12p (E289K) is inactive in vivo (8) and substantiate the point that our in vitro studies reflect the physiological function of Atp12p.
Wild type and E240K mutant HuAtp12p were found to be similar with respect to several measured physical parameters, including accessibility of cysteines to labeling with 5,5Ј-dithiobis(2-nitrobenzoic acid) under native and denaturing conditions (data not shown), intrinsic fluorescence spectra obtained under native conditions (data not shown), and sensitivity to limited trypsin and chymotrypsin digestion (Fig. 5). The latter work established that there is no proteolytic hot spot in HuAtp12p that could define the boundary between subdomains in the protein. However, the GdnHCl equilibrium denaturation curves for both wild type and mutant HuAtp12p were found to be biphasic (Fig. 6), suggesting the existence of a stable intermediate between the native and fully denatured protein, possibly associated with the unfolding of a subdomain of HuAtp12p. The first phase of the denaturation curves is also associated with some protein aggregation, as indicated by an increase in light scattering (Fig. 6, inset). This aggregation might be caused by the exposure of a hydrophobic subdomain that is buried within HuAtp12p under native conditions. We further speculate that this subdomain may be the "active" surface of the chaperone that shields the exposed hydrophobic surfaces of unassembled F 1 ␣ subunits (3) or thermally denatured CS (Fig. 3A). Under this point of view, the perturbation of HuAtp12p structure observed at low concentrations of GdnHCl may mimic a conformational switch that occurs naturally when HuAtp12p interacts with a substrate protein.
The second phase of the equilibrium denaturation curve of both the wild-type and mutant HuAtp12p was fitted to a twostate cooperative (all or none) transition between native and denatured states. The appreciable difference between wild type and mutant in the values of m (the slope around the curve mid-point) suggests the existence of a state R in equilibrium with the native state N. In the mutant E240K the equilibrium N 7 R may be shifted toward R, which could be a different conformation of the protein with a higher solvent exposure and a correspondingly lower value of m (14). The extremely low concentration of R at equilibrium may be the reason why the individual components of the GdnHCl equilibrium denaturation curves of HuAtp12p are fitted reasonably well by Equation 3, which describes a two-state cooperative transition. In this view the existence of two components in the equilibrium chemical denaturation curve of HuAtp12p and the different slope of the second component of this curve in the inactive mutant E240K are consistent with the R state being the conformation of the chaperone that binds to the hydrophobic surfaces of the target proteins. The Glu 3 Lys substitution might increase the probability of this state to occur but decrease its intrinsic chaperone activity.