Acid-induced Reversible Unfolding of Mitochondrial Aspartate

THEJOURNALOF BIOLOC~CAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 269,No. 35,Issue of September 2, pp. 21990-21999, 1994 Printed in U.5.A.

Acid-induced Reversible Unfolding of Mitochondrial Aspartate Aminotransferase* (Received for publication, March 30, 1994)

Antonio Artigues, Ana Iriarte, andMarino Martinez-Carrion$ From the Division of Molecular Biology ~.and Biochemistry, School of Biological Sciences, University of Missouri, Kansas City, Missouri 64110-2499

Theacid-induced reversible unfolding of several forms of the mitochondrial isoenzymeof mammalian aspartate aminotransferase, including its precursor form, has been characterized under equilibrium conditions.A minimum of two transitions can be detected forthe holoenzyme (pyridoxal form). One transition takes place at pH 3.6 and corresponds to the monomerization of the dimeric protein. The second transition is centered at pH 3.3 and represents the disappearance of much of the tertiary and secondary structures. Thepresequence peptide in the precursor protein does not affect the equilibria nor the rateof unfolding in the pH range from 7.5 to 2.0. The presence of the cofactor, pyridoxal B’-phosphate, stabilizes the protein against acid denaturation. At pH 2.0, the protein retains significant amountsof secondary structure (26% a-helix,20% p-structure). Increasing the ionic strength at pH 2.0 results in significant changes in the secondary structure of the unfolded protein that acquires some of the characteristics ascribed to a compact molten globule. Accordingto the circular dichroism spectra these changes are characterized by an increase in P-structure, although Fourier transform infrared spectroscopy analysis indicates that this increase in 6-structure is due mostly to the formation of intermolecular P-sheet as a consequence of protein aggregation. The formation of high molecular weight aggregates has been confirmed by analytical ultracentrifugation. Following neutralization of the acid-unfolded state at low ionic strength both mature and precursor proteins refold to their native active state (>SO% yield). By contrast the compact state present at pH 2.0 and high ionic strength is unable to recover its activity following neutralization. Thus, this compact state does notappear to represent an intermediate in the folding pathway of the protein, but rather a dead end product of aggregation, which may reflect the intrinsic tendencies of the unfolded protein to oligomerize at intracellular salt concentrations unless controlledby factors such as chaperones present in the cellular environment.

to reinforce this concept. The analysis of the intermediates in the unfolding and refolding pathway is expected to give information regarding the mechanisms of protein folding, as they are assumed to representa population of intermediate states within the folding pathway of the protein. In other words, the processes are reversible. More recently, however, there has been an increasing number of reports on the possible participation of a group of proteins knownas molecular chaperones in the process of protein folding as itoccurs within thecell. Thus, questions arise as to therole of these proteins and theirmechanism of action in vivo,particularly since theydo not appear to affect the pathway or rate of folding, but rather to limit offpathways reactions,including those leading to aggregation (for recent reviews see Craig (1993) and Mathews (1993)). Part of the problem in understanding such interactions resides is a basic lackof knowledge of the unfolding states of proteins in the absence of chaperones. In the specific case of proteins destined to be imported into mitochondria there are additional issues of concern. Many of these proteins contain an N-terminal presequence or signal peptide whose role in protein folding is largely undetermined either under in vitro or in vivo conditions. This is the case for the mitochondrial aspartate aminotransferase (mAspAT),’ a complex protein composed of two identical subunits each containing a molecule of pyridoxal 5“phosphate (PLP) bound to Lys258at the active site (Jansonius and Vincent, 1987). This cytosol as a precursor protein with enzyme is synthesized in the an additional 29-residue presequence peptide which is needed for targeting and transport to the mitochondria, where the form, folds to its native strucprotein is processed to its mature ture, binds PLP, and becomes active (Altieri et al., 1989). All these processes may be regulated in vivoby several cytosolic or mitochondrial chaperones. However, refolding after guanidine hydrochloride denaturation of mitochondrial rat liver mAspAT, in its mature and precursor forms (Reyes et al., 1993), and related proteins such as the smaller Escherichia coli AspAT (Herold and Kirschner, 1990; Leistler et al., 1992; Herold and Leistler, 1992), and thecytosolic pig heart AspAT (synthesized without a presequence) (West and Price, 19891, can be achieved successfully in vitro in the absenceof cellular factors. The mechanismby which complex proteins fold is a problem On the other hand, when analyzingfolding the of mAspAT in that has eluded clarification, even though it is generally ac- cell-free extracts, it is apparentthat the folding and interaction cepted that the intrinsic information necessary for folding is of the nascentprotein with cytosolic chaperones, such as hsp70, contained within the protein aminoacid sequence (Epstein et differs between the precursor and matureforms of the protein al., 1963). The ability of purified denatured proteins to refold (Lain et al., 1994). However, the presequence peptide per se is spontaneously upon removalof the denaturing agent continues

* This work was supported

by National Institutes of Health Grants GM-38341and GM-38184. The costsof publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordancewith18 U.S.C. Section 1734 solely to indicate this fact. .$ To whom correspondence should be addressed. Tel.: 816-235-5246; Fax: 816-235-5158.

The abbreviations used are: mAspAT, mitochondrialaspartate aminotransferase; AspAT, aspartate aminotransferase; &PAT,cytosolic aspartate aminotransferase;pmAspAT, precursor mitochondrialaspartate aminotransferase; DPPC, 1,2-dipalmitoyl-sn-glycero-3-phosphocholine; DPPG, 1,2-dipalmitoyl-sn-glycero-3-phophoglycerol; FT-IR, Fourier transform infrared spectroscopy; GdnHC1, guanidine hydrochloride; PAGE, polyacrylamide gel electrophoresis; PLP, pyridoxal 5’phosphate; PMP, pyridoxamine 5”phosphate.

2 1990

Acid Unfolding of Mitochondrial Aspartate Aminotransferase insufficient to retard protein folding in these cellular extracts as a signal peptide fused to the homologous cytosolic isoenzyme (cAspAT) has no effect on the folding of the protein (Mattingly et al., 1993a).Interestingly, although the two isoenzymes refold at comparable rates after denaturation with guanidine hydrochloride, the folding of nascent cAspAT newly synthesized in a cell-free translation system is markedly faster than thatof the mitochondrial isoenzyme (Mattingly et al., 1993b).Thus a better understanding of the i n vitro reversible unfolding reaction of these isoenzymes as well as of the structure of their unfolded states may provide the necessary background to approach the analysis of the interactions of the isoenzymes with cellular components that may regulate their distinct rates of folding i n vivo. During synthesis of a polypeptide chain i n vivo there is vectorial emergence of the protein from the ribosome, N-terminal end first, whereas under i n vitro conditions the whole molecule is unfolded to an experimentally predetermined degree. However, preparations of proteins denatured under controlled conditions are widely considered to represent suitable models to analyze critical events on the folding pathway of nascent proteins. Previous denaturation and renaturation studies of mAspAT in the presence of guanidine HC1 as the unfolding reagent2 (Reyes et al., 1993), produced solutions of unfolded proteins with high viscosity, which made difficult the experimental determination of some physicalparameters of the protein or, more importantly, conductrefolding experiments diluting the unfolded protein into cellular extracts to reproduce more i n vivo like conditions. Furthermore, the presence of the denaturant produced direct interferences with somespectroscopic techniques, and because there is a direct interaction between denaturant and protein, some physical parameters may reflect protein-denaturant interaction rather than intrinsic parameters of the protein. On the other hand, the use of pH changes to study the denaturation process of a protein obviate the above inconveniences, and upon dilution in a suitable buffer for refolding experiments, the unfolding agent, H', becomes itself part of the native system. The mechanism of acid unfolding of proteins has been described for monomeric proteins such p-lactamase, apomyoglobin (Gotoet al., 1990a, 1990b; Gotoand Fink, 1990; Fink et al., 1991; Barrick and Baldwin, 19931, melittin (Ramaligan et al., 1992), cytochrome c (Stellwagen and Babul, 1975; Dyson and Beattie, 1982),bovine serum albumin (Sadler and Tucker, 19931, and ribonuclease A (Scholtz and Baldwin, 1993), and their structures at acidic pH have been characterized byCD and fluorescence spectroscopy. Whereas the unfolding of these monomeric proteins is limited to the destruction of the tertiary and secondary structures, the acid denaturation of a more complex protein such as the dimeric PLP-dependent mAspAT involves several additional steps, including the perturbation of intersubunit interactions and the release of the cofactor, in addition t o the unfolding of each subunit. The use ofmAspAT as a model t o study the unfolding of oligomeric proteins has several advantages. First of all, the crystallographic structure of the protein is known at high resolution (Jansonius and Vincent, 1987; McPhalen et al., 1992), and since the isolated precursor protein is catalytically active, there are no anticipated significant structural differences between &PAT and pmAspAT. Second, since the protein is a homodimer, the identity between both subunits facilitates the study of its denaturation and renaturation transitions, without the possible difficulties involved in thestudy of heterodimers of highly oligomeric proteins. Finally, after its translation in the

21991

cytoplasm, mAspAT has t o be transported to the mitochondrial matrix, and the analysis and characterization of the possible cytosolic and mitochondrial factors involved in the regulation and control of these events, particularly the maintenance of the protein in apartially unfolded import-competent conformation, has been initiated (Mattingly et al., 1993a, 199313; Lain et al., 1994).Furthermore, duringthe import process, the protein will prevailing in the external face be exposedto the local acidic pH of the energized mitochondria, as a result of the vectorial pumping of protons associated with electron transport through the respiratory chain (Mitchell, 1979). Suchan acidic environment may contribute t o maintaining the conformational state amenable for translocation through the membrane, as proposed for the release of the retinol from the carrier protein to target cells (Bychkova et al., 1992). In this paper we describe the characterization of the reversible acid-unfolding equilibrium of various forms of mAspAT through a variety of spectroscopic and other analytical procedures, which was unattainable with guanidine hydrochloride as the denaturing agent. We report the detection of different new unfolding transitions corresponding to the monomerization of the protein, release of PLP, and destruction of both tertiary and secondary structures of the monomers. We have characterized the structureof the acid-unfoldedprotein and the origin of the acid-resistant residual secondary structure, more prominent at high ionic strength, isdiscussed as resulting from the interaction of the highly charged unfolded protein with salt ions in the media. EXPERIMENTAL PROCEDURES

Materials-Aspartic acid, cysteine sulfinic acid, and a-ketoglutaric acid were purchased from Sigma. D,O and DC1 were purchased from MDC Isotopes. DPPC and DPPG were purchased from Avanti Polar Lipids, Inc. All other reagents wereof the highest purity available. Purification of pmAspAT and the preparation of mAspAT were carried out as described previously (Altieri et al., 1989; Mattingly et al., 1993a). The concentration ofAspAT was determined spectrophotometrically from the absorbance at 280 nm using an extinctioncoefficient of E F =~1.40. The purified protein solution was finally concentrated by ultrafiltration (Amicon centricon, 30,000 molecular weight cutoff) and dissolved in 2 mM Tris-HC1, pH 7.4, buffer to a final protein concentraverified by SDS-PAGE, tion of 4 mg/ml. The integrityof the proteins was according to Laemmli (1970). Preparation of the apoenzyme was done as described previously (Reyes et al., 1993), by addition of a 4 M excess of cysteine sulfinate over total protein (PLP form) to convert bound the coenzyme to the pyridoxamine form. Preparationof the pyridoxamine 5"phosphate (PMP) form wasaccomplished by addition of a 5 M excess of PMP to a solution of apoenzyme in 2 mM Tris-HC1, pH 7.5, buffer, followed by incubation for 30 min at room temperature and extensive dialysis against the same buffer to eliminate the excess of unbound pyridoxamine. The reducedform of the enzyme (pyridoxyl-aminotransferase) was prepared by the method of Cheng and Martinez-Carrion (19721, except that sodium cyanoborohydryde was used instead of sodium borohydride to reducethe internal Schiff base between PLP and Lys258. at 37 "C using aspartate and The enzyme activity was measured a-ketoglutarate as substrates, ina coupled assay with malate dehydrogenase as described previously (Mattingly, 1993b). The isolated holoenzyme characteristicallyhad a specific activity of about 250 pmoV mg.min.Theactivity of the apoenzymewasassayedafter a brief incubation in 2 mM Tris-HC1, pH 7.5, containing 5PMPLP. Under these Conditions, any possible reactivation due to the change of pH should be insignificant, since the binding of the PLP is extremely fast, and the activity was measured within less than 1 min of PLP addition. Over 85% of the original activity was recovered upon reconstitution of the apoenzyme. Acid Unfolding of mAspAT-Acid unfolding of pmAspAT and mAspAT was achieved by decreasing the pH of a solution of enzyme, typically at 0.1 mg/ml in 2 mM Tris-HC1, pH 7.4, either by slow addition of a small volume of 25 mM HCl or by making a dilution of the stock enzyme in 2 mM Tris-HC1 previously acidified at the desired pH. The T. Wu, A. Iriarte, andM. Martinez-Carrion, manuscript in prepara- enzyme was then incubated at room temperature for 2 h, which was tion. shown t o be long enough toaccomplish complete changein the param-

21992

Acid Unfolding Mitochondrial of Aspartate Aminotransferase

eters being followed. pH measurements were performed both before and TABLEI after data acquisition with a Radiometer pH Meter 26 equipped witha n Dunsition midpoints for the acid-induced unfolding of &PAT Ingold microelectrode, and measurementsof pH agreedbefore and after Midpoint of transitions were obtained after curve fitting of the experthe experiment within 0.05 units. For FT-IR experiments all solutions imental data to asigmoid function. were made using D,O instead of H,O, and the pH of all solutions was Reduced PLPPMPadjusted by addition of diluted DCI. Corrected values for pH were calParameter ApO-mAspAT mAspAT mAspAT mAspAT = pD + 0.4, wherepD is the direct culated according to the equation pH Activity 5.3 4.6 measurement from the pH meter. 3.6 NA" ND ND 3.7 ND Analytical Ultracentrifugation-Determination of the sedimentation Sedimentation coefficient Formation of hybrids ND ND 3.7 ND velocity coefficient of mAspAT at different pH values was done in a ND ND 3.6 -3.0 Beckmann XL-A analytical ultracentrifuge. Samples previously equili- CD, 278 nm NA ND NA -3.0 brated at the desired pHin 2 mM Tris-HC1, a t a protein concentrationof CD, 330 nm nm CD, 440 NA NA 3.6 NA 0.1 mg/ml, were centrifuged at 20 "C and 6,000 rpm usinga n An-60 Ti emission, y, 3.4 3.3 3.3 3.1 analytical rotor. Sedimentation coefficient, s, was calculated from the Fluorescence CD, 221 nm 3.3 3.3 3.3 3.1 second derivatives of the migrationof the absorbance boundaryprofiles at 280 nm, using the softwareprovided with the instrument. NA, not applicable. ND, not determined. Circular Dichroism-Circular dichroism measurements were made a t room temperature, with a Jasco 700 spectropolarimeter, using a pH values, as described before. After refolding, the aggregated material bandwidth of 2 nm and scan speed of 50 nndmin. Each sample was measured five times, and the scans were automatically averaged. The was pelleted by centrifugation and then a n equal volume of DPPG: values of molecular ellipticity, (B), were obtained by using the expres- DPPC (3:Uliposomes solution was added tothe supernatant and incubated atroom temperature for 20 min to allow the precursor-containing sion: (0) = (O/lO)m/lc, where m is the mean residue molecular weight (1141, 1 is the path length of the sample solution, c is the protein con- dimers (either homo- or heterodimers) tobind. The bound proteins can centration in g/ml, and 0 is the direct value from the instrument. The be separated from the mature homodimers remaining in solution by centrifugation in a Beckman airfuge ultracentrifuge equipped with an units of (0) are degree cm2 dmol". Fluorescence Spectroscopy-Intrinsic fluorescence spectra of the pro- A-100 30" rotor, a t 60 p.s.i. for 20 min. The pellet, containing the hybrid buffer and tein were takena t room temperature with a SLM 8000 C Aminco spec- and theprecursor homodimers, wasdissolved in SDS sample analyzed by 12% SDS-PAGE. The silver-stained gels were quantified trofluorimeter, using the photon counting mode. The excitation wavecollected between 300 and 400 with a Protein & dna imageware systems (PDI) densitometer. Lipolength was280 nm, and the spectra were somes were preparedfrom solutions of DPPGDPPC (3:lmolar ratio) in nm with a resolution stepof 0.5 nm. FT-ZR Spectroscopy-Infrared spectra were obtained at room temper- chloroform dried under vacuum as a thin film in a glass tube. m e r ature ina Sinus 100 (Mattson Instruments) Fourier transform infraredaddition of refolding buffer (100 mM Hepes, 0.1 mM EDTA, 1 m~ 1,4spectrophotometer equipped with a liquid nitrogen cooled mercury-cad- dithiothreitol, 5 PM PLP, pH 7.5)containing 0.2 M NaCI, the lipid mixture was suspended by sonication using a ultrasonic processor W-385 mium-telluride detector. Calcium fluoride windows (I-mm thickness) and 0.056 mm Teflon spacers were employed in all experiments. To (Heat Systems Ultrasonics,Inc.) at 1-min intervalsfor a totalof 10 min a t 4 "C. The final 1ipid:protein molar ratio used was 20,OOO:l. eliminate spectral contributions from atmospheric water vapor, the spectrometer and cell chamber were continuously purged with dry niRESULTS trogen. Protein concentrations used werein the rangeof 15-20 mg/ml. Infrared spectra of the different protein solutions were collected as pH-dependent Inactivationof Aspartate Aminotransferase described previously (Sanchez-Ruiz et al., 1986). A total of 500 interThe acid-induced inactivation of aspartate aminotransferase ferograms were added and Fourier-transformed t o obtain spectra of samples. The speed of the moving mirror was 2.0 cnds, and data were was studied by measuring the enzyme activity of samples inacquired in the forward direction of the mirror, the total scan time was cubated a t different pH values at room temperature for 2 h, a 14 min. The spectrum of the solvent was measured under the same time that was established be to sufficient to reach equilibrium. conditions. Final corrected spectra were obtained after subtraction of The transitionmidpoints for the loss in activity as a functionof absorbance due to air and buffer according to the previously published pH were pH 3.6 for the PLP enzyme (Table I), 4.6 for the PMP method 1 of Sanchez-Ruiz and Martinez-Carrion (1988) and Sanchezform of the enzyme, and 5.3 for the apoenzyme (Fig. IA). In Ruiz et al. (1991), being the criteria for air and solvent subtraction obtaining clearly defined base lines.No smoothing was applied to any addition, the transition curves for these forms of the enzyme also differ in their degree of cooperativity as can be inferred spectra. Resolution EnhancementofFT-IR Spectra-Resolution enhancement from the markedly higher slope of the transition for the PLP of the spectra was achieved using the SpectraCalc software (Galactic enzyme as compared with those of the less stable forms PMP Industries Corp.) installed on a 486DX2 50 MHz PC. Second derivatives and apoenzyme. That is, the transition for the PLP enzyme of corrected spectra were obtained using the Savitsky-Golay deconvooccurs in a significantly narrower pH range. These results lution method(Savitsky-Golay, 1964), using 12-15 points, which allows for easy determinationof the position and numberof components pres- agree with several previous observations, indicating that the at~the presence of the Schiff base bond between PLP andL Y S ' ~ ent under the infrared amide I band. Fourier transform spectra were obtained usinga n average widthof 4 nm andBezel apodization function active site stabilizes the enzyme toward thermal (Iriarteet al., with a n enhancement factor K of 1.9. To determine the contributionof 1984) or GdnHC1-induced denaturation.' The transition mideach component band to the overall band contour iterative curve-fitting points were independent of the protein concentration used in was done usinga nonlinear least squares curve-fitting procedure, which is optional t o the SpectraCalc package. The number of bands and their the range 0.01 to 1 mg/ml (data not shown), indicating that approximate position were takenfrom the derivative and deconvolved dissociation of the dimer into monomers is not the rate-limiting spectra. Initial valuesfor peak height and widths were estimated visu- step of the inactivation reaction. The time course of inactivaally from the deconvolved spectra. A Gaussian band profile was used as tion is dependent on the reaction temperature. As expected, the a parameter for the envelope of the individual band components. During is raised, so that inactivation rate increases as the temperature fitting, the height, width, and position of all bands were varied continu- at 10 "C the half-life for inactivation of the PLPenzyme at pH ously. Assignment of individual stretching frequencies tospecific struc3.0 is 15 min, whereas it is only 5 min at 20 "C. On the other tural components was done as described previously (Byler and Susi, hand, no significant differences were detected between either 1988). the transition midpoints or the inactivation rate constants for Formation of Hybrid Dimers between pmAspAT a n d mAspAT-The formation of precursor-mature heterodimers upon refolding of a mix- the precursor and mature forms of the enzyme (Fig. IA). ture of unfolded mAspAT and pmAspAT was analyzed to monitor the The acid-induced unfolding of either precursor or mature dissociation step during unfolding. The determination of the relative forms at pH 2.0 and higher isa reversible process. As observed amount of hybrid dimers produced relied on the ability of the precursorwith the guanidinehydrochloride-denatured protein (Reyes et containing dimers to bind to negatively charged phospholipids vesicles (Martinez-Carrion et al.,1990). The assayinvolved the renaturationof al., 1993), upon dilution into refolding buffer (100 mM Hepes, a n equimolar mixture of mAspAT and pmAspAT unfolded a t different 0.1 mM EDTA, 1 mM 1,4-dithiothreitol, 5 p~ PLP) at pH 7.5, the

Acid Unfolding of Mitochondrial Aspartate Aminotransferase 7r"---A 6

21993

. . .

0

81

0

5 V 0

.>:

20

20 40

4

b'

O d J .

2

4

6

. . ...

0

3

PH FIG.1. pH-dependent inactivation of mAspAT. Equilibrium transition for the inactivation of pyridoxal (Ok,pyridoxamine (Ob, and and pyridoxal-pmAspAT 4 ) after 2 h of incubation a t apo-mAspAT (0) room temperature. Experimental data points were adjusted by iterative curve fitting to a sigmoid function to determine the transition midpoints.

60

'i

2

80

0

00.

O

1 1

enzyme denatured a t pH 2.0 refolds t o form the native active dimer with -80% yield, based on recovery of activity (data not shown).

2

3

4

100 5

6

7

8

PH

B

Dissociation of the Native Dimer Protein incubated at pH 3.0 3.7 7.5 Analysis by Ultracentrifugation-The sedimentation velocity , . . . .of the PLPform of rat liver mAspAT was measuredat different pH values under equilibrium conditions (Fig. 2). Between pH prnAspAT 7.4 and 3.8 the enzyme has a sedimentation coefficient of 5.7 s, " which is identical to that reported for the mAspAT from pig rnAspAT heart (Felis and Martinez-Carrion, 1970) and is in the range found for globular proteins with equivalent molecular weight (Tanford, 1967). A sharp transition takesplace a t pH 3.7 (Table I), a n d apH t 3.6 the enzyme has a sedimentation velocity of 1.9 s, which is in the rangefound for smaller proteins withmolecular weight of about 20,000. This transition may result only not FIG.2. Monomerization occurs in a very narrow pH range. A, from an apparent change in the molecular weight of the pro- acid-unfolding monomerization transition ofmAspAT determined by tein, due to monomerization of the native dimers, but, since the analytical ultracentrifugation(open symbols) and mature/precursorhybridization (closed symbols). Samples incubated at the indicated pH sedimentation coefficient is much smaller than that expected were centrifuged a t 60,000 rpm as described under "ExperimentalProfrom the molecular weight of the monomer (43,0001, may also cedures.'' Sedimentation coefficients are expressed in Svedberg units. 0 that a t pH 3.7 and 3.8. represent an increase in thehydrodynamic volume of the pro- and D represent the two different species coexist tein a s a consequence of partial unfolding of the dissociated B , SDS-PAGE analysis of the subunit composition of the liposomebound protein obtained upon incubation of refolding mixtures of premonomers. During the transition in the pH range between 3.6 cursor and maturemAspAT with acidic liposomes. The ratio of precurand 3.8, two different species can be detected (Fig. 2, V, O), sor tomaturesubunitswasdetermined by densitometry after of hybridizawhich have sedimentationcoefficients intermediate to those ob- visualization by silver staining and plotted as percentage tained for the fully monomeric and dimeric sates of the protein. tion as a function of original unfolding pH (A, closed symbols) considering the ratio found a t low pH a s 100%.See "ExperimentalProcedures" Although several factors may influence the sedimentation and text for further details. coefficient (Ralston, 1993; Laue etal., 1992;), when adilute and homogeneous solution of protein is sedimenting in an aqueouswill accumulatebehindthefastersedimentation boundary. buffer, the value obtained close is to so, the ideal sedimentation Consequently, theareasunderthe concentration gradient coefficient. This is true in our case at extremes pH values, curves are no longer a measure of the concentration of each where only dimers or fully denatured monomers are present in component, leading to the determination of lower and higher solution. However, during themonomerization transition of the coeffkients for the faster and slower components of the system, protein, where both dimers and monomers (partly or fully de- respectively. Another important factor to consider in thiscase is natured) are present in equilibrium, the sedimentationcoeffi- the existence of a monomer-dimer equilibrium that will prodcient is influenced not only by the heterogeneity and diffusion uce, accordingt o the lawof mass action,a different distribution of the species in equilibrium that will tend to spread out the of monomers and dimers throughout the sedimentation gradisedimentation boundary, but alsoby the concentrationdepend- ent, which will lead to the determination of an average sedience of the sedimentation coefficient through the cell that can mentation coefficient depending on the relative concentration lead to self-sharpening of boundaries. This effect can lead to of monomers and dimers in a particular segmentof the cell. The determination of a higher sedimentationcoefficient for the slow complexity of the system increaseswhen we consider not only moving species (Schachman, 1959). According tothe well the monomer-dimer equilibrium, but also the possible equilibknown Johnson-Ogston effect (Johnson andOgston, 19461, in a ria existing between partly or fully unfolded monomers or two component system, the faster component will sediment in dimers. Since the sedimentation velocity coefficient can be inthe usual way, but the measureds value will be determined by fluenced by some or all of the above factors, the possibility of the sum of the concentration of the two components under the the existence of unfolded monomers in equilibrium with unfaster sedimentation boundary, whereas the slower component folded dimers cannot be discarded from this data.

-

I-.

21994

Acid Unfolding of Mitochondrial Aspartate Aminotransferase

Analysis by Hybridization Techniques-The monomerization transition was further studied by analyzing the formation of heterodimers between precursor and mature forms of the enzyme denatured at different pH values by a novel procedure. The assay takes advantage of the ability of the precursor to bind to negatively charged liposomes (Martinez-Carrion et al., 1990). When an equimolecular solution of mature and precursor forms of the enzyme that haspreviously been incubated in the pH range 4-2 is neutralizedby dilution into the appropriate refolding buffer, they refold and form active native dimers. If the unfolding condition results in dissociation of the dimers, a distribution of mature homodimer:hybrid:precursor homodimer of 1:2:1 can be expected in the refolding reaction, assuming there is no preference (kinetic control) t o form homodimers. Since only the precursor containing dimers bind to liposomes through the presequence peptide, mAspAT: pmAspAT heterodimers as well as precursor homodimers can be resolved from mAspAT:mAspAT homodimers by addition of liposomes (DPPGDPPC, 3:l).Upon centrifugation, the precursor-containing dimers bound to vesicles will precipitate, whereas the unbound mature homodimer will remain in solution. Because of the higher molecular weight of the precursor subunits, the relative content of precursor and mature subunits associated with the pellet can be analyzed by SDS-PAGE (Fig. 21, and from these values a percentage of hybridization was calculated takingthehighest mAspAT/pmAspAT ratio observed a t pH 2.0 as 100%. Above pH 4.0 only precursor homodimers are found associated with the pelleted liposomes. The changes in theprecursor/ mature ratiobound t o liposomes as a function of pH are directly related to the increase in the fraction of heterodimeric species present after incubation at low pH and reflect the monomerization of the two protein forms. The resultsshow that a sharp transition takes place between pH 3.8 and 3.6 with a midpoint that coincides with that determinedby analytical ultracentrifugation (Table I), indicating thatformation of heterodimers and therefore monomerization of the protein occurs in a sharp transition centered at pH 3.7. The maximum mature/precursor ratheoretical ratio of 1:2 to be tio obtained, 1:1,is higher than the expected from a random association of monomers of the two protein forms. This discrepancycan be explained by: ( a )a lower yield of pmAspAT homodimers due to slightpreference for dimerization of hybrids or mature homodimers or ( b )unspecific binding of the refolded precursor protein to the walls of the containers (Mattinglyet al., 1993b).

330

1

2

3

4

5

6

7

8

PH FIG.3. Equilibrium unfolding transitions of apo (0) and pyridoxy1 (0) &PAT were examined by monitoring changes in intrinsic fluorescence. Samples at 0.4 mg/ml were incubated in 2 mM Tris-HCI, at the pH indicated, as described under “Experimental Procedures.” The emission maxima were determined as a function of pH, and the experimental data points were adjusted by iterative curve fitting to a sigmoid function (lines)to determine the transitionmidpoints. Inset, fluorescence emissionspectra of native apoenzyme (truce 1 ) and holoenzyme (trace 2 ) at pH 7.5, and denatured (apo or holo) enzymeat pH 2.0 (truce 3 ) .Fluorescence spectra were recordedat 20 “C, using A, = 280 nm.

Unfolding of the holoenzyme releases the coenzyme from the active site, and therefore the fluorescence spectra of acid-unfolded apo- and holoenzymes are identical. The pH dependence of the red shiftof the tryptophanylemission maximum clearly shows that the presence of pyridoxal 5”phosphate covalently bound at the active site increases the stability of the enzyme against acid denaturation (Fig. 3). The midpoint of the transitions was determined and is pH 3.4 for apoenzyme, 3.3 for pyridoxal-mAspAT and pyridoxamine-mAspAT, and 3.1 for pyridoxal-mAspAT after reduction of the Schiff base with NaBH, (pyridoxyl-mAspAT).Also noticeable is the significantly sharper transition observed for the unfolding of the holoenzyme which reflects a higher degree of cooperativity of the process.

Coenzyme Release and Loss of the Active Site Architecture An assessment of the tertiary structure of the active site pocket can be madedirectly in AspAT due to thepresence of a natural reportergroup, the coenzyme PLP. The circular dichroism spectra of native apo- and holo-mAspAT have been previpH-induced Unfolding of mAspAT ously characterized (Martinez-Carrion et al., 1970a). In the The quantum yield and the position of the emission maxi- n e a r - W region of the spectrum at pH 7.4, the spectra show mum of a protein’s fluorescence spectrum upon excitation at three positive overlapping bands in the aromatic region cen280 nm are highly sensitive to the environment of its trypto- tered at about 267, 278, and 285 nm (Fig. 4). These bands are associated with the asymmetry in the environment of aromatic phanyland tyrosylresidues. These two parametershave amino acids in the tertiary structure of the protein and in proved to be very informative in the study of the structural changes occurring during unfoldinghefolding of proteins particular at the active site. Upon binding of pyridoxal 5’-phos(Murry-Brelier and Goldberg, 1988; Goto and Fink, 1989; Bis- phate andformation of the internal aldimide bond, the protein muto et al., 1992; Elove et al., 1992; Bychkova et al., 1992; Chen elicits positive pH-dependent bands withmaxima at 360 nm at et al., 1992; Reyes et al., 1993). The sequence of aspartate pH 7.4 and at 440 nm at pH 4.5. As the pH of the samples is aminotransferase contains 6 tryptophanyl residues (Mattingly progressively reduced, the band corresponding to the bound et al., 1987), one of which (Trp140)is located at theactive site of coenzyme decreases and eventually disappears. The initial dethe enzyme (Jansonius and Vincent, 1987). The fluorescence crease of the 360 nm bandbetween pH7.5 and 4 is concomitant spectra of native apo- and holo-mAspAT (or its precursor) show to the appearance of a band at 440 nm and is related to the protonation of the Schiff base (Martinez-Carrion et al., 1970b). emission maxima at 337 and 335 nm, respectively, but the quantum yield of the holoenzyme is considerably lower due to Further decrease of the solution pH results in the disappearthe quenching of the Trp140fluorescence by the bound PLP (Fig. ance of the 440 nm band as well as of the aromatic bands 3; Reyes et al., 1993). After acid unfolding of the apoenzyme at centered at 278 nm. Both of these transitions show a similar pH 2.0, and probably as a consequence of the increasingexpo- midpoint at about pH 3.6 which coincides with those obtained sure of tryptophanyl residues towater, the emission maximum for the monomerization and inactivation of the enzyme. Thus, is shifted t o 352 nm, and the quantumyield is highly reduced. the simultaneous release of the coenzyme and perturbation of

21995

Acid Unfolding of Mitochondrial Aspartate Aminotransferase Apoenzyme

0.20 -4 0.10

10

-8

0 -10

0.00 ' U G J

-0.10 0.20

0 -

0

Holoenzyme

2

v

w 3

PH

0.00

0.10

FIG.5. Equilibriumunfoldingtransitionsexaminedby CD spectroscopy. Changes in dichroicity at 440 nm (A) and at 222 nm ( B ) as a function of pH values are: apo (O), pyridoxamine (V), pyridoxal (O), and pyridoxyl(*) forms of &PAT. Conditions for sample preparation and spectra measurement were as described in Fig. 4.The midpoint for each of the transitions was calculated by iterative curve fitting of the experimental data to a sigmoid function.

PLP-LysZ5*Schiff base also inducesa noticeable increase in the resistance to acid-induced unfolding as reflected in a sharper 0 0.00 222 nm transition with a midpoint at pH 3.1. Asimilar behavior was observed when following the overall unfolding of the pro-10 'W tein by monitoring the intrinsic fluorescence properties. -0.10 Analysis by FT-IR Spectroscopy-To determine thesecondary 200 250 225 400 300 500 structure content of a protein under a defined set of conditions, Wavelength (nm) it is convenient to use independent and complementary techFIG.4.Circular dichroism spectra for the apo-, pedoxal-, and niques. CD signals are subject to interferences due t o differpyridoxyl-&PAT. Samples were incubated in 2 mM Tns-HCl at pH ences in rotatory strength of certain residues, and in addition 7.5 (truce l ) ,3.5 (truce 21, 3.0(truce 3), and 2.0 (truce 4 ) . Conditions for contributetothe unfolding of the proteins were as described under "Experimental Pro- elements of tertiarystructure mayalso cedures." Measurements in the 260-550 nm region of the spectrum far-UV region of the CD spectrum (Buck et al., 1993; Surewicz were performedusing a cell width of 10 mm, and protein concentration et al., 1993). Therefore, we have made use of the conformational of 0.4 mg/ml, whereas for the analysis in the far-UV region of the sensitivity of the amide I infrared absorbance band to further spectrum (190-260 nm) the protein concentration was 0.1 mg/ml, and study the secondary structure of the protein under acidic conthe light path was 1mm. Molar ellipticities were obtained as described under "Experimental Procedures." The far-UV spectrum of the PLP- ditions. This band resultsprimarily from stretching vibrations &PAT denatured in 4 M GdnHCl is also included (5) for comparison of C=O groups in peptide bonds. The interference of the strong purposes. infrared water absorbance (maximum -1650 cm") can be easily eliminated by using D,O instead of H,O as solvent. the active site three-dimensional architectureoccurs at 0.3 pH The broad amide I infrared bandof native holo-mAspAT (Fig. units higher than the overall loss of tertiary structure and 6 A ) shows a maximum at 1652 cm" and a shoulder at 1636 seems tobe the immediate cause of loss of activity. After reduc- cm". While the position of the bandmaximum is characteristic tion of the Schiff base, the PLP band shifts to 340 nm and of proteins with a high content of a-helix structure(Byler and becomes pH-insensitive between pH4.5 and 8.5. The transition Susi, 1988; Surewicz et al., 19931, further structural details are midpoint for this form of the enzyme is significantly lower than obscured by the overlapping of the various components of amthat of the holoenzyme (approximately 3.0, Table I), indicating ide I bands thatreflect different elements of protein secondary a stabilizing effect of the irreversible bound coenzyme on the structure. The presence of these distinct components can be disruption of the active site structure. easily detected as minima in the second derivative of the spectrum, which for the spectrum of native mAspAT shows the Perturbation of Secondary Structure presence of two major components at 1658 and 1636 cm", Analysis by Circular Dichroism-In the far-UV region of the which correspond to themaximum and shoulder of the original spectrum at pH 7.4 all forms of the native enzyme show iden- amide I infrared band and represent a- and p-structure, retical curves with negative minima at 208 and 220 nm, and a spectively. The minima at 1679 and 1626 cm" are also related positive maximum at 192 nm, characteristicof the high a-helix to p-structure, whereas the component at 1669 cm" represents content of the protein(Fig. 4). For the pyridoxamine, pyridoxal, p-turns. Theweak band at 1609 cm" corresponds to amino acid and apoenzyme forms, the intensity of these bands progres- side chains vibrations (tyrosine and arginine). sively decreases between pH 4.0 and 2.0 (Fig. 4) showing a Acid unfolding of the protein at pH 2.0 results in a prohighly cooperative transition, with a midpoint at pH 3.3 (Fig. nounced alteration of the amide I band contour (Fig. 6B). The 5B). Simultaneously a new negative band centered at 200 nm, maximum of the bandis shifted t o 1650 cm", and theshoulder characteristic of the presence of random coil conformation, ap- at 1636 cm" disappears. The second derivative shows that the pears and reaches its maximum intensity at pH 2.0. These spectrum is composed of several individual bands. The two results indicate that the last event involved in theacid-unfold- major components are located at 1650 and 1646 cm" and coring ofmAspAT seems t o be the destruction of much of the respond, respectively, to a-helix and random coil. The other secondary structure of the protein. However, at pH 2.0 the components correspond to different elements of p-structure unfolded protein sample containssignificant amounts of resid- (1662,1628, and 1615 cm") and p-turns (1685 and 1672 cm"), ual secondary structure (Fig. 4), in contrast t o the GdnHCl and the last band corresponds, again, to the weak amino acid unfolded protein (Fig. 4, spectrum 5;Reyes et al., 19931, which side chains vibrations (1603 cm"). appears to exist mostly in a random coil conformation in the To determine the contribution of each infrared band compopresence of 4 M denaturant. Incidentally, the reduction of the nent t o the overall structure of the protein we used the com10

21996

Acid Unfolding of Mitochondrial Aspartate Aminotransferase element of p-turns at neutral pH, at 1679 cm", is replaced following denaturation by two components at 1685 and 1672 cm", and its total content increases. In summary, the secondary structure of protein at pH 2.0 is extensively modified.

Structural Changes Znduced by KC1 at Low p H It hasbeen reportedfor several proteins that, after unfolding a t low pH, an increase inionic strength results in theirrefolding to a compact A state havingsignificant amounts of secondary structure and little, if any, tertiary structure (Goto et al., 1990a, 1990b). These properties are similart o those ascribed to the "molten globule" intermediate state in the folding pathway of proteins (for reviews see Kim and Baldwin (1982, 1990)and Seckler and Jaenicke(1992)). Increasing theionic strength of a solution of mAspAT at pH 2.0 induces an increase in the fluorescence intensity and a blue shift of the emission maximum (data notshown), indicativeof a shift of tryptophanyl residues to a more hydrophobic environment. Fig. 7 shows the far-UV 1700 1650 1600 1700 1650 1600 CD spectra of acid unfolded mAspAT at pH 2.0 in the absence Wavenumber (cm") and in the presence of increasing concentration of KCl, ranging FIG.6. Infrared amide I band regionof mAspAT. Original (top), from 10 to100 mM. The characteristic band centered a t 200 nm, second derivative (middle), and deconvolved spectra (bottom) of native that reveals the presence of unordered structures in theacidprotein in 2 mM Tris-HC1, pH 7.4 (A), and of the acid-unfolded state at unfolded protein, decreases in a salt concentration-dependent pH 2.0 ( B ) .Sample preparation, spectra measurement, and manipulafashion with theconcomitant increase in ellipticity in the210tion were as described under "Experimental Procedures." Dotted lines represent individual band components obtained by iterative curve fit- 220 nm region, suggesting that in thisprotein the presence of ting, and dashed lines represent the envelope of the amido I band chloride anions induces the formation of secondary structure in contour obtained by addition of these gaussian bands. the acid-unfolded state. Addition of 100 nm KC1 t o protein samples unfolded at pH 2.0 with HC1 caused an increase in putational procedure of band narrowingby Fourier self-decon- ellipticity at 222 nm of about 50% relative to the difference volution. This procedure decreases the widths of the infrared between native protein (pH 7.5) and acid-unfolded protein (pH bands, thus allowing for increased separation and better iden- 2.0). tification of the overlapping component bands present under Analysis of the infrared spectrum in theamide I region for the composite amide I contour. Since this manipulation does acid-unfolded mAspAT also shows that significant changes are not affect the integrated intensities (areas) of the individual induced by the presence of increasing concentrations of salt component bands, the areas of the discrete bands can be cal- (Fig. 8). Raising the KC1 concentration at pH 2.0 causes a culated by iterative curve fitting, and, assuming that molar the substantial decrease in a-helix and a parallel increase in p-conabsorptivities of the individual amide I bands are equal, it is formation (Table 11).Furthermore, two new infrared bandscenpossible to estimate the relative content of each element of tered at 1616 cm" and 1780 cm-' appear in theFT-IR spectrum secondary structure (Surewicz et al., 1993). The results of the whose intensity increase as the salt concentration is raised. iterative curve fitting analysisof the Fourier transform spectraSimilar infraredabsorbance bands havebeen observed for therto the individual components bands previously identified are mally unfolded proteins (Surewicz et al., 1990), including shown in Fig. 6 (bottom).The relative amounts of each struc- ASPA AT.^ The exactsignificance of these two components is not tural component for the native andunfolded enzymes are sum- completely understood but in thermallyunfolded proteins they marized in Table 11, where, for comparison, the content of sec- have been interpreted as representing theformation of interondary structure obtained from crystallographic analysis is molecular P-sheet as a result of aggregation of the denatured also included (Jansonius and Vincent, 1987). The two sets of protein after the exposure of internal hydrophobic regions to data are inclose agreement, in spiteof the varying conditions, water (Surewicz et al., 1990). The development of this structure crystal uersus solution and high (FT-IR) uersus low (CD) protein in theacid-unfolded protein upon increasing salt concentration concentration. Thus, the assignmentof structure to each indi- probably reflects the formation of a similar interchain p-convidual infrared component band and themethod used for curve formation structure. The tendency of the so-called A state to fitting of the Fourier transform infrared spectra are reasona- aggregate is well documented (Arakawa et al., 1987; Goto and ble, in particularfor a series of comparative experiments under Fink, 1989). Furthermore, determination of the sedimentation changing solutionconditions. As expected, the acid-induced un- velocity coefficient by analytical ultracentrifugation gave an folding of this protein results in a large alteration of its sec- unusually high s value of 7.1 for the acid-unfolded mAspAT at ondary structure, as is shown mainly by a significant decrease pH 2.0 in the presence of 100 rn KCl, clearly indicating that in thea-helix content (34% decrease) and by the appearanceof the salt-induced structural changes of the acid-unfolded prorandom coil conformations (25%). However, in agreementwith tein result in theformation of higher order aggregates. It is interesting tonote that, whereas the protein unfolded a t the CD results, thedeconvolved amide I of the FT-IR spectrum of the protein at pH 2.0 also shows the presence of residual pH 2.0 and low ionic strength readily refolds with high yield elements of secondary structure (26% of a-helix, 22.2% of (-80%) to its native and active conformation upon neutralizap-sheet, Table 11).Although the total content of p-sheet does not tion, the presence of even small concentrations of salt in the change, detailed inspection of the individual components of unfolded mixtures precludes protein reactivation when shifted p-structure shows that there is also extensive modification of t o pH 7.5 (data not shown). The inability to refold is not the refolding buffer, as addition this structural element. The individual component bands of result of the presence of salt in the p-sheet at 1679 and 1636 cm-l at pH 7.5 disappear and are replaced by new components a t 1662 and 1615 cm", and the P. Rodriguez, A. Iriarte, and M. Martinez-Carrion, manuscript in band at 1626 cm" is slightly shifted at 1628 cm-l. The single preparation.

21997

Acid Unfolding of Mitochondrial Aspartate Aminotransferase

TABLE I1 FT-ZR-derived structure of mAspAT in the native and acid unfolded states Band curve fitting of the deconvolved spectra of &PATwasperformedusing a nonlinear least squares curve-fitting procedure in the SpectraCalc program. The number of bands andtheir assignmentswere taken from the second derivative and deconvolved spectra. The individual structural assignment of each band was done according to Bylerand Susi(1988). Native

Acid-unfolded, pH = 2.0

solution,Crystallographic InStructure data”

26.0 22.2

.5 3.6

a-Helix 60 p-Sheet 22.7 12.6 p-Turn Inter+ Random coil 1.7 Unknown

pH 1.5’

50.0 14.0 22.0 14.0

No KC1 added

10 mM KC1

12.9 33.5 24.5 1.0 13.3 25.7

18.1 23.1 22.3 11.5 24.0 1.3

20 mM KC1

8.9 32.9 12.7 19.4 14.4 13.7 0.7 1.1

50 mM KC1

100 mM KC1

9.8 31.6 25.2 18.4 14.1 0.9

Jansonius and Vincent (1987). This work.

,pH7.5

1

KCI, mM

I

!

loo

lo

Wavelength (nm) pH 7.5 FIG.7. Effect ofthe KC1 on the far-UV circular dichroism spectra of acid-unfoldedpyridoxal-mAspAT. Protein in 2 mM Tris-HC1 1700 1680 1660 1640 1620 1600 at either pH 7.5(. . . .), pH 2.0(- - - -1, or pH 2.0in the presence of 10, 20,50, and 100 m~ KCl. All samples at a protein concentration of 0.1 mglml were incubatedat the desired pH for 2 h at room temperature to Wavenumber (crn”) achieve equilibrium. Measurements were made at room temperature in a Jandel 700 spectropolarimeter,with a bandwidth of 2 nm;each FIG.8.Effect ofKC1 on the infrared amide I band region of the samplewasmeasuredfive times and the scanswereautomatically FT-IR spectrumof acid-unfoldedmAspAT. The deconvolvedspectra averaged. The values of molar ellipticity,(e), were obtained as described of &PAT in 2 mM Tris-DC1 buffers are shown for the native state at under “ExperimentalProcedures.” pH 7.5 and the acid-denaturedstate at pH 2.0,either in the absence or in the presence of 10, 20, 50, and 100 mM KC1. Sample preparation, of equivalent concentrations of KC1 to the refolding media of measurement, and data manipulation were made as described under “Experimental Procedures.” protein that had been unfolded at pH 2.0 and low ionic strength had no effect on the rateor yield of refolding (data not shown). Rather, these results suggest that in fhe presence of ions the folding in a process that, according to the transition curves unfolded protein adopts an irreversible conformational state obtained following differentconformational parameters,is that cannot be considered as a productive intermediate in the highly cooperative and takes place within a very narrow pH folding pathway despite its apparent higher content of second- range. However, because of the nonidentityof the middle points ary structure elements relative t o that of the acid-unfolded of the different transitionsmonitored, the acid denaturation of state (at low ionic strength). mAspAT does not appear t o be a simple two-steptransition, in which only two forms of the protein are present in equilibrium, DISCUSSION but rather a multistep process during which several intermeAcid-induced Denaturation Equilibrium Studies-Both the diate populations coexist in equilibrium. Between pH 4.0 and mature and precursorforms of mAspAT are stable on incuba- 3.5, small changes in the tertiary andsecondary structures of tion between pH 7.0 and 4.0, as they retain more than 85%of the proteinbegin t o appear; yet, the initiation of the red shift of their original activity, and their sedimentation velocity coeffi- the maximum emission of protein intrinsic fluorescence, and cient of about 5.6 indicates that the proteins remainas dimers. the beginningof reduction of the minima (at 221 and208 nm) No significant alterations are observed either in the fluores- and maximum (at 188 nm) in theCD spectrum, indicatesthat cence or the circular dichroism spectra of the proteins other the protein adopts a more relaxed conformation. The enzymatic than thewell documented pH-dependent changes inellipticity activity of the protein decreases, reaching50% of the original observed in the visible near-UV regionof the spectrum related value at pH 3.6, in a transition that parallels the destruction of t o the different protonation states of the internal aldimide the active site structure, as indicatedby the simultaneous dis(Martinez-Carrion et al., 1970a, 1970b). appearing of the strong Cotton effect (centered at 440 nm, in Between pH 4.0 and 2.0 mAspAT undergoes extensive un- this rangeof pH) associated with bound cofactor as well as of all

21998

Unfolding Acid

of Mitochondrial Aspartate Aminotransferase

dichroicity in the aromatic region (transition midpoint at pH 3.6). The inactivation transition also coincides with the dissociation of the dimers into monomers as shown by the sharp transition observed a t pH 3.7 in the sedimentation velocity coefficient. The latter event was corroborated by the formation of heterodimers between mAspAT and its precursor form from populations that were induced to undergo monomer/dimer equilibria. Since the active site of the enzyme is composed of elements of both subunits, the coincidence in the midpoint of the transitionscorresponding to changes in Cotton effect and s value is of no surprise. Between pH 3.5 and 2.0 the protein experiences further loss of tertiary andsecondary structure. Thedestruction of tertiary structure can be followed by monitoring the blue shift in the emission maximum of protein intrinsic fluorescence, arising from the exposure of tryptophan residues to a more hydrophilic environment. This transition (midpoint at pH 3.3) coincides with the highly cooperative transition for the disappearanceof most of the secondary structure, characterized by dramatic alterations in thefar-UV CD spectrum, including the development of a n intense negative band at 200 nm indicative of the presence of random coil structure. It appears that after the initial perturbation of the active site region and dissociation of the subunits, themonomers are highly unstable and in a very narrow pH range lose simultaneously and with high cooperativity both their tertiary and secondary levels of organization. The presence of PLP bound to the active site of the enzyme increases its stability toward unfolding at low pH, in particular with regard to perturbations in the active site region. This is clearly shown by the different transition curves obtained for the inactivation of the apoenzyme, PMP, and PLPforms (inactivation midpoints at pH 5.3, 4.5, and 3.6, respectively). These results are in agreement withprevious observations by differential scanning calorimetry (Iriarte et al., 19841, showing that the presence of the Schiff base linkage betweenPLP andactive site Lys258increases the thermal stability of the protein. Recent studies on the unfolding of mAspAT by GdnHCl are also consistent with the higher stabilityof the PLP enzyme (Reyes et al., 1993). As mentioned above, the inactivation transition of the PLP enzyme centered at pH 3.6 is related t o dimer dissociation, perturbation of the active sites, and release of the cofactor. Therefore, the unfolding transitions of the apo, PMP, or PLP enzymes associated with the overall loss of tertiary and secondary structures areidentical, because they occur at lower pH values and afterdissociation of the coenzyme, so that they are not affected by the presence or absence of the cofactor at the active site. Yet, if PLP is not allowed to diffuse from its binding site as when irreversibly attached to the protein through reduction of the internal Schiff base linkage, then the equilibrium transition is shifted further to pH 3.1, confirming the active site in dramatic role of covalent attachment of PLP to the the stabilization of mAspAT. Structure of mAspAT at p H 2.0-Circular dichroism studies have shown that GdnHCl unfolded proteins have basically a random coil conformation, in contrast to proteins denaturedat acidic pH which may retain significant amounts of secondary structure (Aune et al., 1967; Fink et al., 1991). The far-UV CD spectrum of mAspAT at pH 2.0 is characterized by an intense negative band at 200 nm together with residual ellipticity at 221 nm, implying that under these conditions the protein retains detectable amounts of secondary structure and cannotbe considered a fully random coil. Similar results havebeen previously described for other acid denatured proteins, including the pz subunit of tryptophansynthase(Muny-Brelierand Goldberg, 19881, horse apomyoglobin and cytochrome c (Goto et al., 1990a, 1990b). From these studies it has been suggested

that the structure remaining at these low pH environments may not be residual native-like structure that theacid unfolding has failed to disrupt, but rather newly formed organizations, and consequently with little resemblance to native structure (Murry-Brelier and Goldberg, 1988; Goto et al., 1989; Buck et al., 1993). We, thus, explored these possibilities under acidic conditions by the use of a complementary technique, FT-IR, which can provide quantitative estimatesof the structuralcontent of the protein. Despite the difficulties described in the literature for the quantitative estimation of the secondary structure of proteins from their infrared spectra (Surewicz et al., 1993), the values we obtained from the integrated areasof the bandcomponents after curve fitting of the Fourier deconvolved infrared amide I band of the protein in the native state were in relative good agreement with those obtained from crystallographic studies (Jansonius and Vincent, 1987). Similar analysis of the FT-IR spectra of the acid-unfolded protein showed that, although the presence of a 25.7% of random coil-like structure indicates that the protein is extensively denatured, it still contains significant amounts of a-helix (26%) and P-sheet structures (22.2%). Comparison of the FT-IR deconvolved spectra of native and acidunfolded enzymes also showsthe formation of what appears to be p-turn (1685 and 1672 cm") and p-structure bands (1662 and 1615 cm") in theacid-unfolded state that are absent in the native state.Since the vibration position of the infrared bands is highly susceptible t o the microenvironment of the secondary structures, theorigin of these bandscould be either a change in the solvent accessibility of the peptide bonds or to the formation of new types of structure. The latter possibility can be explained according to the previously described mechanism for the acid unfolding of proteins (Tanford, 1970; Goto et al., 1990a, 1990b), which considers that acid denaturation of a protein is achieved by the bindingof protons to its titratable groups, and as they become positively charged theresulting repulsion forces between them lead to the disruptionof the native structure of the protein. When the protein is fully protonated, the addition of more acid leads to the formation of electrostatic interactions between the anions and the positively charged groups in theprotein, which in turnmay decreasethe repulsive forces and indirectly increase the intrinsichydrophobic forces that may cause the protein to acquire a more compact structure. This last possibility is strengthened by the fact that increasing concentrations of KC1 also seemto induce a conformational change (discussedbelow) that leads to aggregation of the protein. The information available, however, does not allow t o establish whether the secondary structure remaining in the acid unfolded state of mAspAT is residualnative-like structure oris partly or completely formed de novo as a result of interactions of the ion components of the acid used to induce unfolding, H' and C1-, with amino acid residues. Nevertheless, at pH 2.0 the protein not only has become enzymatically inactiveand haslost its cofactor, but there is also disappearance of all dichroicity in the visible region of the spectrum and a red shift of the tryptophans fluorescence, all of which are consistent with hydrophobic segments of the protein becoming exposed to a more hydrophilic ambient. Below pH 3.3 the protein seems to exist largely in a monomeric form, and its sedimentation coefficient (s = 1.9) suggests that the monomers are largely unfolded, since this sedimentation Coefficient is significantly lower than that expected for a (compact) globular protein with the molecular weight of the subunitof mAspAT (47,000). Thus theacid-denatured mAspAT at pH 2.0 represents a highly unfolded monomeric state of the protein. Increasing theionic strength of a solution of mAspAT at pH

Unfolding Acid

of Mitochondrial Aspartate Aminotransferase

21999

2.0 induces significant alterationsinthe fluorescence and Arakawa, T., Hsu, Y.-R., and Yphantis, D. A. (1987) Biochemistry 26, 5428-5432 Aune, K. C., Salahuddin, A., Zarengo, M. H., and Tanford,C. (1967) J. B i d . Chem. f a r - W CD spectra of the protein that arecompatible with some 242,4486-4489 of the features ascribed for a compact molten globule. However, Barrick. D., and Baldwin, R. L. (1993)Biochemistry 32,3790-3796 the shape of the far-WCD spectrum is not consistent with the Bismuto, E., Sirangelo, I., and Irace, G. (1992)Arch. Biochem. Biophys. 298,624629 formation of native-like secondary structure, but rather sug- Buck, M., Radford, S., and Dobson, C. (1993) Biochemistry 32,669678 gests an exclusive increase in the content of p-structure. Fur- Bychkova, V. E., Berni, R., Rossi, G . L., Kutyshenko, V. P., and Ptisyn, 0. (1992) Biochemistry 31, 756G7571 thermore, the analysis by iterative curve fitting of the Fourier D. M., and Susi, H. (1988)J . Ind. Microbiol. 3, 73-88 deconvolved amide I band of the acid unfolded mAspAT in the Byler, Chen, H. M., Markin, V. S., and Tsong, T. Y. (1992)Biochemistry 31,12369-12375 presence of increasing concentrations of KC1 reveals an inter- Cheng, S., and Martinez-Carrion,M. (1972) J. Biol. Chem. 247,6597-6602 esting new feature. The increase in ionic strength causes the Craig, E. (1993) Science 260, 1902-1903 appearance of bands at 1616 and 1780 cm" which have been Dyson, H. J., and Beattie, J. K. (1982) J. Bid. Chem. 257,2267-2273 Elove, G. A,, Chaffotte, A. F., Roder, and Goldberg, M. E. (1992) Biochemistry 31, assigned to formation of an intermolecular p-sheet structure, 6876-6883 similar t o that describedfor thermally unfolded proteins Epstein, C. J., Goldberg, R. F., and Anfinsen, C. B. (1963) Cold Spring Harbor Symp. Quant. B i d . 28, 439446 (Surewicz et al., 1990).The intensity of these bands increases N., and Martinez-Carrion, M. (1970) Biochem. Biophys. Res. Comrnun. 40, in a saltconcentration-dependent fashion. We propose that the Felis, 932-940 development of this structure for this acid-unfolded protein Fink, A. L., Calciano, L. J., Goto, Y., and Palleros, D. (1991) in Conformation and Forces in Protein Folding (Nall, B. T., and Dill, K. A., eds) pp. 169-174, Ameriupon increasing salt concentration can be explained by the can Association for the Advancement of Science, Washington, D. C. same mechanism previously presented for the acid unfoldingof Goto, Y., and Fink, A. L. (1989)Biochemistry 28, 945-952 mAspAT. That is, the further reduction of the electrostatic re- Goto, Y., and Fink, A. L. (1990) J. Mol. B i d . 214, 803-805 pulsion between the positively charged polypeptide molecules Goto, Y., Takahashi, N., and Fink, A. L. (1990a) Biochemistry 29, 3480-3488 as a consequence of the interaction between chloride ions and Goto, Y., Calciano, L. J., and Fink,A. L. (1990b)Proc. Natl Acad. Sei. U. S. A . 87, 573-577 protonated positively charged residues of the protein can facili- Herold, M., and Kirschner, K. (1990) Biochemistry 29, 1907-1913 tate polypeptide hydrophobicinteractions that will lead to ag- Herold, M., and Leistler, B. (1992) FEBS Lett. 308, 2G29 gregation through formation of intermolecular p-sheets (Goto Iriarte, A., Relimpio, A. M., Chlebowski, J. F., and Martinez-Carrion,M. (1984) in Chemical and Biological Aspects of Vitamin B, Catalysis, Part E: Metabolism, and Fink, 1990; Goto et al., 1990a, 1990b; Fink et al., 1991). In Structure and Function of Dansaminases (Evangelopoulos, A. E., ed) pp. 107agreement with these results, the acid-unfolded protein in the 115, Alan R. Liss, Inc., New York presence of 100 mM KC1 has an unusually high s value of 7.1, Jansonius, J. N., and Vincent, M. G. (1987) in Biological Macromolecules and Assemblies (Jurnak, F., and McPerson, A., eds) Vol. 3, pp. 187-285, John Wiley clearly indicating that these salt-induced structural changes & Sons, Inc., New York may be due to aggregation of the protein. Johnson, J. P., and Ogston, A. G. (1946) Duns. Faraday SOC. 42, 789 The existence of transitions between different states for acid- Kim, P. S., and Baldwin, R. L. (1982)Annu. Reu. Biochem. 5 1 , 4 5 9 4 8 9 P. S., and Baldwin, R. L. (1990)Annu. Reu. Biochem. 59,631-660 denatured proteins, described mostly according t o far-UV CD Kim, Laemmli, U. K. (1970) Nature 227,680-685 data formonomericacid-unfolded proteins (Goto and Fink, Lain, B., Iriarte, A,, and Martinez-Carrion, M. (1994) J. Bid. Chem. 269, 1559015596 1989; Goto et al., 1990a, 1990b; Fink et al., 19911, is in contrast T. M., Shah, B. D., Ridgeway, T. M., and Pelletier, S. L. (1992) in Analytical with our results, because those transitions seem t o imply for- Laue, Ultracentrifugation in Biochemistry and Polymer Science (Harding, S. E., Rowe, mation of native-like secondary structure. The difference in A. J., and Horton,J. C., eds)pp. 90-125, Royal Society of Chemistry, Cambridge behavior between mAspAT and monomeric proteins may be due Leistler, B., Herold, M., and Kirschner, K. (1992) Eur J . Biochem. 205, 603-611 either to the nature of the monomeric proteins used in those Mathews, C. R. (1993) Annu. Reu. Biochem. 62,653-683 Martinez-Camon, M., Tiemeier, D. C., and Peterson,D.L. (1970a)Biochemistry 9, studies or more likelyto intrinsic properties of mAspAT, which 2574-2582 is adimer in itsnative state. The salt-induced compact state of Martinez-Carrion, M., Kuczenski, R., Tiemeier, D. C., and Peterson, D. L. (1970bj J. Bid. Chem. 245, 799-805 mAspAT possibly reflects the intrinsic tendency of mAspAT to Martinez-Carrion, M., Altieri, F., Iriarte, A,, Mattingly, J., Youssef, J., and Wu, T. oligomerize. Moreover,although at low ionic strength acid-un(1990) Ann. N . I:Acad. Sci. 586, 346356 folding ofmAspAT is reversible, the salt-induced aggregated Mattingly, J. R., Jr., Rodriguez-Berrocal. F. J., Gordon, J., Iriarte, A,, and MartinezCamon, M. (1987) Biochem. Biophys. Res. Commun. 149,859-865 form of the protein at pH 2.0 is unable to recover its activity J. R., Jr., Iriarte, A,, and Martinez-Carrion, M. (1993a) J. B i d . Chem. after neutralization. Thus, it isunlikely that thecompact state Mattingly, 268,26320-26327 present at acidic pH and high ionic strength represents an Mattingly, J. R., Jr., Youssef, J., Iriarte, A., and Martinez-Carrion. M. (1993b) J . Biol. Chem. 268, 3925-3937 intermediate in the refolding pathway of mAspAT. It is also C. A,, Vincent, M. G., and Jansonius, J. N. (1992) J. Mol. Bid. 226, concluded that formation of compact denatured states at low McPhalen, 495-517 pH upon addition of salts not always involves formation of Mitchell, P. (1979) Science 206,1148-1159 native-like secondary structure. Rather, at least with the un- Murry-Brelier, A., and Goldberg, M. E. (1988)Biochemistry 27, 7633-7640 G. (1993)Introduction toAnalyticu1 Ultracentrifugution, pp. 23-37, Beckfolded monomers of dimeric mAspAT, it seems to represent a Ralston, man Instruments Inc., Fullerton, CA dead-end aggregation process. Ramaligan, IC, Aimoto, S., and Bello, J. (1992) Biopolymers 32, 981-992 These observations have implications to protein folding un- Reyes, A., Iriarte, A., and Martinez-Carrion,M. (1993) J . Biol. Chem. 268,2228122291 der in vivo conditions, where nascent unfolded proteins in the Sadler, P. J., and Tucker, A. (1993) Eur J. Biochem. 212, 811-817 presence of the high salt concentrations of the intracellular Sanchez-Ruiz, J. M., and Martinez-Camon,M. (1986) Biochemistry 25,2915-2920 environment may be at risk of forming misfolded or dead-end Sanchez-Ruiz, J. M., and Martinez-Carrion,M. (1988)Biochemistry 27,3338-3342 structures perhapsnot far different from the denatured states Sanchez-Ruiz, J. M., Iriarte, A., and Martinez-Carrion, M. (1991) Arch. Biochem. Biophys. 286, 38-45 under in vitro unfolding conditions. Recent studies in our lab- Savitsky, A., and Golay, M. J. E. (1964) Anal. Chem. 36,1627-1639 oratory (Mattingly et al., 1993a) suggest that intracellular fac- Schachman, U. K. (1959) Ultracentrifugation in Biochemistry, Academic Press, New York tors such as molecular chaperones could mediate folding events Scholtz, J. M., and Baldwin, R. L. (1993) Biochemistry 32, 4604-4608 by preventing formation of these high salt-induced condensed Seckler, R., and Jaenicke, R. (1992) FASEB J. 6, 2545-2552 states and thusbe part of the salvage mechanism to productive Stellwagen, E., and Babul, J. (1975) Biochemistry 14,5135-5140 folding. We are currently pursuing a detailed analysis of these Surewicz, W. K., Leddy, J. J., and Mantsch, H. H. (1990) Biochemistry 31, 810G 8111 processes. Surewicz, W. K., Mantsch, H. H, and Chapman, D. (1993) Biochemistry 32, 389REFERENCES Altieri, F., Mattingly, J . R., Jr., Rodriguez-Berrocal. F. J., Youssef, J., Iriarte, A,, Wu, T., and Martinez-Carrion, M. (1989) J . Bid. Chem. 264,47824786

394 Tanford, C. (1967) in Physical Chemistry of Macromolecules, pp. 364-390, John Wiley & Sons, Inc., New York Tanford, C. (1970) Adu. Protein Chem. 24, 1-95 West, A., and Price, C. (1989)Biochem. J. 261, 189-196