Biochimica et Biophysica Acta, 1159 (1992) 81-93

81

© 1992 Elsevier Science Publishers B.V. All rights reserved 0167-4838/92/$05.00

BBAPRO 34282

Conformational analysis of a mitochondrial presequence derived from the F1-ATPase/3-subunit by CD and NMR spectroscopy Martha D. Bruch and David W. Hoyt 1 Departments of Pharmacology and Biochemistry, University of Texas Southwestern Medical Center, Dallas, TX (USA)

(Received 13 April 1992)

Key words: ATPase, F1-;/3-Subunit; Signal sequence; Circular dichroism Previous studies on mitochondrial targeting presequences have indicated that formation of an amphiphillic helix may be required for efficient targeting of the precursor protein into mitochondria, but the structural details are not well understood. We have used CD and NMR spectroscopy to characterize in detail the structure of a synthetic peptide corresponding to the presequence for the /3-subunit of F~-ATPase, a mitochondrial matrix protein. Although this peptide is essentially unstructured in water, a-helix formation is induced when the peptide is placed in structure-promoting environments, such as SDS micelles or aqueous trifluoroethanol (TFE). In 50% TFE (by volume), the peptide is in dynamic equilibrium between random coil and a-helical conformations, with a significant population of a-helix throughout the entire peptide. The helix is somewhat more stable in the N-terminal part of the presequence (residues 4-10), and this result is consistent with the structure proposed previously for the presequence of another mitochondrial matrix protein, yeast cytochrome oxidase subunit IV. Addition of increasing amounts of TFE causes the a-helical content to increase even further, and the TFE titration data for the presequence peptide of the F~-ATPase/3-subunit are not consistent with a single, cooperative transition from random coil to a-helix. There is evidence that helix formation is initiated in two different regions of the peptide. This result helps to explain the redundancy of the targeting information contained in the presequence for the F~-ATPase/3-subunit.

Introduction Over 90% of mitochondrial proteins are synthesized on cytoplasmic ribosomes and must be imported into the mitochondria [1]. In most cases, the presence of a N-terminal presequence is sufficient for targeting of the protein into the mitochondria, and this presequence, typically, is proteolytically cleaved in the mitochondria following transport (for reviews, see Refs. 1-3). Remarkably, when recombinant D N A methods are used to attach a mitochondrial presequence to a soluble protein not normally imported into the mitochondria, the resultant fusion protein usually is succesfully targeted into the mitochondria [4,5]. However, the mechanism by which presequences achieve such efficient targeting remains unclear. Although there is no obvious sequence homology among presequences for different mitochondrial proteins, several general fea-

Correspondence to (present address): M.D. Bruch, Department of Chemistry and Biochemistry, University of Delaware, Newark, DE 19716, USA. 1 Present address: Department of Biochemistry, University of Alberta, Edmonton, Canada T6G 252.

tures are shared. Mitochondrial presequences are largely hydrophilic, basic sequences of 10-70 amino acids. They contain a high content of both positively charged residues (e.g., arginine and lysine) and hydroxylated residues (e.g., serine and threonine), but do not typically contain any acidic residues. These presequences also contain nonpolar residues interspersed between the basic residues, so an amphiphilic structure is formed if the sequence folds into an a-helix or /3-sheet. These characteristics of mitochondrial presequences are in distinct contrast to those observed for N-terminal signal sequences which are essential for protein export across the endoplasmic reticulum, in eukaryotes, or across the cytoplasmic membrane, in prokaryotes (for review, see Refs. 6-9). Although both types of targeting sequences are rich in hydrophobic and positively charged residues, bacterial export signal sequences have fewer positively charged residues, and these are restricted to the N-terminal part of the sequence. Furthermore, these signal sequences contain a core of hydrophobic residues not observed in mitochondrial presequences. One intriguing question regarding the mechanism by which precursor proteins are imported into the mitochondria is how the targeting information in the prese-

82 quence is recognized and transmitted. The absence of strict sequence homology among presequences precludes recognition of a specific amino-acid sequence and suggests that specific, common structural features may be essential for efficient targeting. Indeed, recent results suggest that formation of a positively charged, amphiphilic structure is an essential, common feature of functional presequences [10-12]. Formation of this amphiphilic structure may be necessary to enable the presequence to penetrate the membrane, and direct interactions of isolated presequences with lipids has been observed [12-15]. However, formation of a specific structure also may be required for recognition by proteinaceous components of the import apparatus, e.g., MOM19 or MOM72 [3,16]. Elucidation of the exact nature of the interactions involved in import of precursor proteins can be assisted through determination of the specific details (e.g., length, location or stability) of the structural features required for functional mitochondrial presequences. Spectroscopic techniques such as circular dichroism (CD) and nuclear magnetic resonance (NMR) are ideally suited for structural characterization of small peptides, and these techniques have been applied to isolated presequences bound to dodecylphosphocholine micelles for two different mitochondrial matrix proteins. Endo et al. report that the isolated presequence for cytochrome oxidase subunit IV has an a-helical conformation only in the N-terminal half of the peptide [17], whereas Karslake et al. observe two helical regions connected by a turn or coil in the isolated presequence from rat liver alcohol dehydrogenase [18]. To gain further insight into general structural features required for efficient import, we report the detailed structural analysis, using CD and NMR spectroscopy, of a synthetic peptide corresponding to the presequence for another mitochondrial matrix protein, the /3-subunit of F1-ATPase. This peptide consists of 20 amino acids with the sequence MVLPRLYTATSRAAFKAAKQ; post-translational proteolytic cleavage in the precursor protein occurs between Lys-19 and Gin-20 [19]. The import of F1-ATPase /3-subunit into mitochondria has been studied extensively [1,2024], and the presequence is sufficient to localize the protein into mitochondria both in vivo and in vitro. Therefore, comparison of the structure of this presequence with the structures reported for cytochrome oxidase subunit IV and rat liver aldehyde dehydrogenase will help to identify essential, shared conformational features required for effective targeting. In addition to being useful for comparison purposes, the fl-subunit of F~-ATPase has several interesting properties which make the conformational analysis of this presequence particularly interesting. The isolated presequence of this protein binds to artificial membranes. However, in contrast to the behavior observed

for bacterial export signal sequences, the mitochondrial presequence peptide does not penetrate deeply into lipid bilayers nor disrupt the integrity of bilayer structures [15]. Furthermore, the mitochondrial targeting information for the F~-ATPase /3-subunit is contained redundantly in three regions of the precursor protein [23,24]. Two of these regions, 5-12 and 16-19, are within the cleavable presequence, while the third region, 28-34, is within the mature part of the protein. The presence of any one of three regions is sufficient to direct efficient import of F~-ATPase fl-subunit; import is blocked only when all three of these regions are deleted [23]. Hence, any structural requirements must still be satisfied when any one of these regions is present in the precursor protein. Previous studies on the isolated presequence for F1-ATPase /3-subunit using CD spectroscopy have shown that this peptide is essentially random in water, but SDS micelles or trifluoroethanol (TFE)/water mixtures induce formation of an a-helix [15]. We have chosen aqueous TFE as the solvent system for detailed conformational analysis for several reasons. First of all, the conformation is similar in TFE and in SDS micelles, and this suggests that aqueous TFE provides a reasonable model for the interfacial environment which would be experienced when a presequence encounters a membrane. Secondly, TFE has been shown to stabilize helix formation, but not indiscriminately. Previous studies have shown that TFE promotes helix formation only for regions which have a strong tendency to adopt a helical conformation [25-29]. Finally, this solvent system can easily be manipulated by changing the temperature a n d / o r the TFE concentration to explore helix propagation and stability. Previous work on isolated bacterial export signal sequences has shown a correlation between helix stability in aqueous TFE and export competence of the signal peptide [30,31], and the present study provides a direct comparison between the structural tendencies of mitochondrial presequences and signal sequences. In aqueous TFE, the entire peptide corresponding to the isolated presequence for the /3-subunit of F lATPase has a tendency to adopt a helical conformation. However, the most stable part of the helix is localized in the N-terminal part of the peptide, residues 4-10. These results suggest that the helical structure may be stabilized independently by two regions in the presequence, and this may be related to the fact that the presence of either of two regions in the presequence is sufficient for translocation of the /3-subunit of F~-ATPase into mitochondria. Materials and Methods

Peptide synthes&. The mitochondrial presequence peptide was synthesized using standard solid-phase

83 methods on an ABI 430A automated peptide synthesizer, t-Boc protected amino acids were purchased from ABI. The completed peptide was deprotected and reduced from the resin by H F cleavage. After evaporation of HF, the peptide was precipitated with diethyl ether and separated from the resin by successive washings with 50-100% acetonitrile (in water) with 0.1% added trifluoroacetic acid, followed by filtration. The lyophilized crude peptide was purified by reversephase H P L C using a Vydac phenyl column with a water/acetonitrile gradient (0.1% TFA). Purity was checked using an analytical column, and a single peak was observed in the H P L C trace. The sequence of the peptide was confirmed by liquid-phase sequencing on a AB1 477 protein sequencer using Edman degradation; no anomalies were observed. Finally, a Beckman 6300 amino acid analyzer was used to determine amino-acid composition, and the composition agreed well with the values expected for this sequence. Circular dichroism. All CD measurements were made on an AVIV model 60DS spectrophotometer, using a 5 mm cell at controlled temperatures of 5, 25 and 50°C. The temperature was regulated by a Hewlett-Packard 89100 A temperature controller. The peptide concentration (10-20 /xM) was determined by quantitative amino-acid analysis. The solvent system was a mixture of 2,2,2-trifluoroethanol (Aldrich, spectral grade) and 4.5 mM phosphate buffer (pH 4.2 and 7.0). All CD spectra were the average of three consecutive scans with a total duration of 9 s at each nanometer from 260 to 190 nm. All spectra were baseline corrected and smoothed. Nuclear magnetic resonance. All N M R measurements were made on a Varian VXR 500 spectrometer operating at a proton frequency of 500 MHz. Peptide concentration for N M R samples was approx. 3 mM. The solvent was a mixture of 2,2,2-trifluoroethanol-d 3 (Merck Isotopes) and unbuffered distilled water adjusted to pH 4 to reduce the exchange rate of the amide protons with the water. The T F E content was either 30% by volume (10 mol%) or 50% by volume (20 mol%). Two-dimensional total correlation spectroscopy (TOCSY) [32,33] and nuclear Overhauser effect spectroscopy (NOESY) [34,35] were performed using standard pulse sequences for these experiments. The water resonance was suppressed by preirradiation during the relaxation delay (1 s) and during the mixing period in the NOESY spectra (300 ms). For all 2D-NMR spectra, 256 free induction decays were obtained containing 2 K complex points each, and the matrix was zero-filled to 1 K × 1 K real points. The spectral width was 6000 Hz in both dimensions, and 64 transients were accumulated per t I value with a 1 s relaxation delay. The mixing time was 75 ms for TOCSY spectra and 300 ms for N O E S Y spectra. All data were collected in the

phase-sensitive mode using the method of States et al. [36] and were processed on a Sun 4 / 2 6 0 computer using software developed by Dennis Hare (Infinity Systems, Seattle, WA). All time-domain data were premultiplied by an apodization function consisting of either a gaussian or a sine-bell phase shifted by 25 ° prior to Fourier transformation. Two-dimensional spectra were symmetrized [37] to eliminate noise arising from image peaks associated with the large residual T F E methylene peak, and the symmetrized spectra are presented for clarity. However, all spectra were analyzed in both symmetrized and unsymmetrized form, and care was taken to distinguish symmetrization artifacts from real cross-peaks. Results

Circular dichroism Circular dichroism (CD) is a good technique for initial characterization of the overall structural features of a peptide. Previous work showed that the CD spectrum of the Fi-ATPase/3-subunit presequence peptide (Fl-presequence) exhibited a double minimum characteristic of an a-helical conformation [38] in SDS micelles at 25°C [15]. To explore the formation and propagation of this helical conformation, CD spectra were obtained as a function of T F E content in a T F E / w a t e r mixed solvent system. T F E is known to stabilize helical structures, but does not induce helix formation in regions of the peptide which have a low inherent helical propensity [25-31]. The T F E titration data (Fig. 1) show that the Ft-presequence is essentially random in pH 4 buffer with no T F E and that the helical content

40

20 ffl

"\ I

I o

-20

•. . . . . .

i

200

.......

i

•

|

220 240 Navelength (nm)

260

Fig. 1. CD spectra of the isolated presequence peptide corresponding to the /3-subunit of F]-ATPase. The spectra were obtained at 25°C with the following TFE concentrations (by volume) in pH 4 buffer: 0 (solid), 10 (dashed), 20 (dotted), 30 (solid), 40 (dashed), 50 (dotted), 60 (solid), 80 (dashed) and 95% (dotted).

84 systematically increases as the T F E concentration is increased from 0 to 95% TFE. This increase in helical content is evidenced by the steady increase in the absolute value of the mean residue ellipticity at 222 nm, 0222, and by the increase in the location of the low wavelength minimum from 199 to 209 nm [38]. Since the CD intensity at 222 nm is almost entirely due to a-helical structures, the percent a-helix can be estimated quantitatively from O222. For a 20-residue pep100__ tide which is 100% helical, ~9222-31320, so % ahelix = (100)(0222)/~9~ [39]. Based on this equation, the helical content of the Fx-presequence ranges from less than 5% in the absence of T F E to approx. 70% in 95% TFE. The T F E titration CD spectra have an isodichroic point at 204 nm (Fig. 1), and this implies that the conformation of this peptide in aqueous T F E can be described by a two state random coil ~ a-helix transition. This coil ~, helix transition can be characterized more definitively by plotting 0222 as a function of T F E content (Fig. 2a). This curve does not have the sigmoidal shape which is characteristic of a single, cooperative transition from random coil to a-helix. Instead, the CD intensity at 222 nm increases steadily as the T F E content is increased, with a somewhat slower rate of increase observed near 50% TFE. One problem with using 8222 tO monitor helical content is that the intensity of a CD spectrum is dependent on the peptide concentration, and concentration errors can dramatically alter 8222 values. To be sure the titration curve based on CD intensities is not misleading, we also made estimates of the helical content based on the shape of the CD spectra. The CD spectra can be fit to polylysine reference spectra to extract percentages of a-helix, fl-sheet, and random coil structures [40], but the use of basis spectra obtained on large polymers to fit spectra of small peptides can give misleading results. Alternatively, the shape can conveniently be described by two ratios of intensities in CD spectra, R1 and R2, and these parameters have been used successfully to precisely measure relative helical content in small peptides which can be characterized by a coil o helix transition [41]. R1 is defined as the ratio of ~gmax, the maximum intensity from 190 to 195 nm, to ~gmin, the minimum intensity from 195 to 210 nm. R1 is positive when the helical content is low and becomes more negative as the helical content is increased. R2 is defined as the ratio of 1~222 t o {~min, and this quantity increases to a limiting value near one as the helical content is increased. Both R1 and R2 are unaffected by errors in peptide concentration. Plots of R1 (Fig. 2b) and R2 (Fig. 2c) as a function of T F E concentration provide additional measures of the coil o helix transition. Although these two titration curves have somewhat different shapes than the curve obtained using 8222 as a measure of helicity, these titration

-5

-t0

®

-t5

-20

-25

P 0

t 20

'

I 40

:

: 60

:

: 80

: t00

volume Z TFE 0.5

0.0

-.5

-! .0

-I .5

-2.0 40

2O

6O

8O

I00

volume Z TFE

1.0

0.8

0.8

0.4

0.2

0.0

•

:

;

20

0

l

40 volume

,

t

, --÷-----~---

60

8O

I00

Z TFE

Fig. 2. CD parameters for estimation of the helical content of the isolated Ft-presequence peptide as a function of T F E concentration at 25°C (see text for details). The smooth curves for each parameter were obtained by interpolation using cubic splines [50]: (a), 0222 (relative error is +5%); (b), R1 (absolute error is _+0.05); (c), R2 (absolute error is _+0.02).

curves based on R1 and R2 also are inconsistent with a single, cooperative coil ,,-, helix transition. This transition can be explored further by examining

85

30

20

t0 0

-t0

/:.~

-t0

. / \/

2o0

2?0

2'o

Wavelength (rim)

260

200

220

240

2150

~avelength (rim]

Fig. 3. CD spectra of the F1-presequence in 50% TFE: (a), as a function of temperature in pH 4 buffer: 5°C (dashed), 25°C (solid) and 50°C (dotted); (b), as a function of pH: pH 7.0 (solid), pH 4.2 (dashed), pH 2.5 (dotted).

the temperature-dependence of the CD spectrum in 50% TFE (Fig. 3a). These CD spectra show that the helical content is increased as the temperature is decreased from 50 to 5°C; ~9222 increases and the low wavelength minimum shifts from 208 to 207 nm. The q u a n t i t i e s 8222, R1, and R2 as a function of temperature indicate that the overall helical content in 50% TFE at 5°C is nearly equal to the helical content in 80% TFE at 25°C, while the helical content in 50% TFE at 50°C is similar to the helical content in 30% TFE at 25°C. Hence, both the TFE content and the temperature can be manipulated readily to significantly alter the coil/helix equilibrium of the Fl-presequence in aqueous TFE. The previous discussion refers to peptide solutions in aqueous TFE at pH 4, since pH 4 is required to facilitate the NMR studies by reducing the amide exchange rates. However, since physiological pH is near 7, the conformational behavior of the peptide at neutral pH is of interest. The CD spectra of the Fl-presequence in 50% TFE as a function of pH (Fig. 3b) show that there is significantly more helix at pH 7.0 than at pH 4.2 or 2.5. The TFE titration data at pH 7 yield curves similar in shape to those obtained at pH 4, but the helical content is higher at each point on the curve for pH 7. This result is unexpected, since this sequence does not contain any residues with sidechain pK values between 4 and 7 in aqueous buffer. However, the pK values for ionizable groups may be quite different in aqueous trifluoroethanol than in aqueous buffers, and this may explain the unusual pH dependence observed in T F E / H 2 0 . For example, if the terminal NH~ group were neutralized more easily in T F E / H 2 0 , then the helix dipole would be stabilized at higher pH by neutralization of the charge at the amino terminus [42]. Alternatively, if the C-terminal C O O group were protonated more readily in aqueous TFE,

then protonation of the carboxyl group at lower pH may destabilize the helix due to disruption of interactions between positive charges near the C-terminal end and the negatively charged carboxyl group. Unfortunately, the exact cause of the higher helical content observed at pH 7 for this peptide cannot be pinpointed, since the effect of TFE on the pK values is not known. Nevertheless, the pH-dependence observed for this peptide demonstrates that the coil/helix equilibrium can be shifted easily in aqueous TFE. The pH dependence observed for the mitochondrial presequence peptide is in marked contrast to the pH independent nature of the CD spectra of the signal peptide for the exported protein, LamB, in 50% TFE [30,31]. Furthermore, the helical content of the LamB signal peptide in 50% TFE cannot be increased by further addition of TFE, whereas the helical content of the mitochondrial Fl-presequence continues to increase as the TFE concentration is increased beyond 50 vol.%. This difference in behavior for the two types of targeting sequences suggests that the mitochondrial F~-presequence forms a less stable helix than the LamB signal sequence in 50% TFE. CD spectroscopy is useful for assessment of the overall peptide conformation and the results discussed above demonstrate that the Fl-presequence in aqueous TFE can be described by a two-state equilibrium between a-helix and random coil structures. However, CD does not provide information about structural details, e.g., the location, length or stability of the helix. These details can be provided by NMR spectroscopy. Since the NMR studies require a much higher peptide concentration (approx. 3 mM) than that employed in the CD studies (10-20 tzM), it is important to ensure that the structure is not concentration dependent (as would be observed if the structure were stabilized by association or aggregation of the peptide). No differences were observed among CD spectra obtained over a concentration range of 10-85 IzM, so the peptide is probably behaving monomerically in this regime. Although this does not prove that the structure is the same at the concentration of 3 mM used in the NMR studies as at the lower concentrations ( < 100 /~M) used in the CD studies, most changes in structure observed at high peptide concentrations are due to association or aggregation of the peptide. No signs of association or aggregation were seen in the NMR samples. The solutions were clear and free-flowing and the NMR line-widths were as expected for the monomeric form of the peptide. Formation of dimers or other aggregates would result in significant line broadening in the NMR spectra, and this was not observed. Although the absence of aggregation does not eliminate the possibility of concentration-dependent effects on the structure, these results suggest that any concentration effects are probably small.

86 NMR methods NMR analysis of each peptide sample is accomplished in two stages. First, each line in the N M R spectrum is assigned to a specific proton in the peptide, using standard sequential assignment methods [43]. In brief, TOCSY spectra [32,33] are used to connect together protons belonging to the same amino acid through intra-residue J-coupling and the pattern of cross-peaks observed in the TOCSY spectrum allows the lines to be assigned to a specific amino acid or type of amino acid. Sequential assignments are then made with the aid of NOESY spectra [34,35], which connect together spatially close protons through both intra- and interresidue nuclear Overhauser effects. Once the line assignments are known, secondary structure elements can be identified through observation of the type and pattern of NOEs present in the NOESY spectrum. The nuclear Overhauser effect (NOE) is a through-space interaction between two spatially close protons which is dependent upon the inverse sixth power of the interproton distance, r - 6 and on the motional properties of the molecule [44]. Although it is difficult to predict exactly how close together two protons must be in order to see an appreciable N O E between them, previous studies indicate that, for a peptide the size of the Fl-presequence in aqueous TFE, no significant Overhauser effects are observed for interproton distances much longer than 4 .& [30,31]. A regular a-helix has several short interresidue distances which can be used to identify regions which adopt a helical structure. These include distances between successive amide protons, dNN(i,i + 1 ) = 2.8 A, the distance between H a and NH protons in successive residues, d,N(i,i + 1) = 3.5 A and the distance between H a of residue i and NH of residue i + 3, d~N(i,i + 3) = 3.4 ,~ [45]. N O E interactions often are observed in rigid helices for two slightly longer distances,od~N(i,i + 4) and dNN(i,i + 2), both of which are 4.2 A in a regular helix [45]. By contrast, only one interresidue distance is sufficiently short for observation of an appreciable NOE in a /3-sheet or extended structure: d,N(i,i + 1) = 2.2 A [44]. The interproton distances associated with a random coil are highly variable due to the amorphous, undefined nature of a random coil. However, the average distance in a random coil is anticipated to be similar to that observed for an extended structure, since, in a random coil, the peptide spends the majority of time sampling regions of conformational space corresponding to some sort of extended structure. An alternate measure of helix formation is provided by the amide vicinal coupling constant, 3JHN~. It has been shown that for proteins of known structure, the amide coupling constants typically are small (less than 5 Hz) for residues in an a-helix and large (greater than 8 Hz) for residues in a 0-sheet [46]. Amide coupling constants for residues in a random coil are expected to

be somewhat reduced from that observed for a/3-sheet, due to motional averaging. For a rigid helix, the entire network of NOE interactions corresponding to the short interproton distances described above will be observed and all of the amide coupling constants will be small. However, the Fl-presequence peptide is in dynamic equilibrium between helical and random coil conformations, and this equilibrium complicates the interpretation of the NMR data. The observed coupling constants are simply given by the population weighted average of coupling constants corresponding to the helical and random coil structures. Therefore, the observed 3JnN . is a measure of the population of the a-helical state; lower J-values are indicative of a higher population of helix. The N O E data are more difficult to interpret in the case of an a-helix ~ random coil equilibrium, since both the interproton distances and the motional properties of the peptide are different in the two states. For molecules in the slow-motion limit, faster motion geoerally results in smaller Overhauser effects. For an a - h e l i x / r a n d o m coil equilibrium, random coil structures are typically more flexible than a helix, while interproton distances which are short in an a-helix are longer in random coil structures. Hence, a decrease in the population of a-helix (increase in random coil) will result in a significant decrease in the magnitude of helical NOE interactions due to both an increase in the average value of r -6 and a decrease in the effective correlation time, r e , caused by increased motion. This decrease in the size of a specific type of interaction, e.g., a ( i ) / N H ( i + 3), may not effect all residues equally. The interproton distances for a real helix may differ significantly from the ideal distances calculated for a regular helix, and the details of the random coil structure may vary significantly from one residue to the next. Despite these ambiguities, the magnitude of the observed NOE interactions can be used as a rough gauge of the thermodynamic stability of an a-helix; a larger population of helical structures will result in larger helical NOE interactions. NMR results Line assignments. N M R analysis was performed on the Fl-presequence peptide under a variety of conditions which induce different populations of a-helix; the same sample was analyzed in 50% T F E at 5, 25, and 45°C *, in 30% T F E at 5 and 25°C, and in 0% T F E at 25°C (all at pH 4). Line assignments were made using the sequential assignment methods described in the

* NMR data could not be obtained at 50°C, due to loss of amide signals because of the fast amide exchange rates at this temperature. Therefore, CD data obtained at 50°C were compared to NMR data obtained at 45°C.

87

° I

i~1~

~ 00.-. 0 ~

1.'. f , •

~,m2

•

i

•

~:4.'W ~

4

o

If~

O:

4:4

nao

D_

e4

Ale

•

4~0

3:6

3:2

,7

•

2:0

2:4

"

2:0

1:6

•

1:2

ppm

Fig. 4. Amide region of the 2D TOCSY spectrum of the F~-presequencein 50% TFE at 25°C. The mixingtime was 75 ms. previous section [43]. Pattern recognition can be applied to the amide region of the TOCSY spectrum (Fig. 4) in order to assign each line to a specific type of amino acid. The NOESY data (Fig. 5) were used to make sequential assignments, and the assignments for the F~-presequence under a variety of conditions are summarized in Tables I and II. Structure determination. Now that the line assignments are known, the structure of the peptide can be analyzed in detail. The amide region of the NOESY spectrum of Ft-presequence in 50% T F E at 25°C (Fig. 5a) contains many N H / N H interactions indicative of a helix spanning from P4 to Q20 (for proline, H a is substituted for NH). In addition, two a ( i ) / N H ( i + 3) interactions are observed: T 8 / S l l and K16/K19 (Fig. 5b); the NOEs observed are summarized in Fig. 6a. Since symmetrized spectra processed with a phaseshifted sine-bell apodization function may contain artifacts and other misleading information, all data sets were processed using several different apodization

functions and analyzed with and without symmetrization. This procedure allowed us to distinguish real effects from artifacts and to obtain a better estimate of the relative intensities of NOEs. It is not possible to determine exactly how many N H ( i ) / N H ( i + 1) and a ( i ) / N H ( i + 3) interactions may be present due to extensive spectral overlap, but it is clear that some of the possible NOEs indicative of a helix are absent in the region from P4 to Q20. As discussed earlier, the presence of some, but not all, of the potential helical interactions suggests that the population of helix in this region, while significant, is considerably less than 100%. Of course, the magnitude of the observed NOEs depends on the experimental conditions employed, especially the mixing time. Previous studies on peptides near this length in aqueous T F E indicate that, under the experimental conditions used in these NOESY experiments, the entire network of helical interactions would be observed for a rigid helix, including N H ( i ) / N H ( i + 2) and a ( i ) / N H ( i + 4) interactions

It K I I I ~ / 1 1 --

~

[ ~ IID/II It A l l

14/16

[

E 0. ri_ O

d

.,.oo,,.0.,. x v2~'o ~ r

t',J

I/2

Q

- - -

8'0

7 '6 ppm

7~2

4.4

412

410

318

~16

[:,pm

Fig. 5. Expanded regions of the 2D N O E S Y spectrum of the Fi-presequence in 50% T F E in p H 4 water at 25°C. The mixing time was 300 ms. In the amide region (expansion on the left), the downfield peak is listed first in all labels. In the H ' ~ / N H region (expansion on the right), the H '~ proton is listed first in all labels. Intraresidue interactions are denoted with 'X'.

88 TABLE I

NMR chemical shifts for the mitochondrial presequence peptide corresponding to the fl-subunit of FvA TPase in 50% TFE at 25°C All chemical shifts are in ppm; estimated absolute error is _+0.02. Residue

NH

H~

H~

He

Met 1 Val 2 Leu 3 Pro 4 Arg 5 Leu 6 Tyr 7 Thr 8 Ala 9 Thr 10 Ser 11 Arg 12 Ala 13 Ala 14 Phe 15 Lys 16 Ala 17 Ala 18 Lys 19 Gin 20

8.20 7.83 7.86 7.64 8.15 7.96 8.08 7.92 7.88 7.84 7.80 7.95 7.80 7.83 7.85 7.74 7.73 7.91

4.13 4.20 4.54 4.31 4.13 4.20 4.26 3.93 4.11 4.04 4.16 4.13 4.10 4.06 4.32 4.04 4.15 4.19 4.24 4.32

2.12 2.02 1.62 2.30, 1.81 1.62 3.02 4.26 1.47 4.19 3.84, 1.87, 1.43 1.35 3.t4 1.84 1.39 1.38 1.80, 1.98,

2.49, 0.91 1.53 1.86, 1.68, 1.55 1.21 1.21 1.63 1.48, 1.33 2.18,

2.03

3.73 1.77

1.69 2.15

[30,31]. Therefore, the absence of some a(i)/NH(i + 3) interactions and the complete absence of NH(i)/NH(i + 2) and a(i)/NH(i + 4) interactions under these experimental conditions implies the existence of a significant population of random coil structure throughout

Ha

H"

0.85, 0.89 3.60, 3.78 3.16 0.82, 0.87

7.07 -

3.12

7.06

1.40

1.66

2.95

2.33

1.58 -

3.16, 3.12 -

2.54

1.96 1.64

the peptide, and this indicates that the helical structure is only marginally stable and is readily converted to a more random structure. When the temperature is lowered to 5°C, many more cross-peaks indicative of a-helix formation are

T A B L E II

Summary of amide and H" chemical shifts On ppm) for the Fl-presequence peptide in aqueous TFE Residue

50% T F E

30% T F E

25°C

Met 1 Val 2 Leu 3 Pro 4 Arg 5 Leu 6 Tyr 7 Thr 8 Ala 9 Thr 10 Ser 11 Arg 12 Ala 13 Ala 14 Phe 15 Lys 16 Ala 17 Ala 18 Lys 19 Gin 20

5°C

45°C

NH

H~

NH

H~

-

4.13

-

8.20 7.83 7.86 7.64 8.15 7.96 8.08 7.92 7.88 7.84 7.80 7.95 7.80 7.83 7.85 7.74 7.73 7.91

4.20 4.54 4.31 4.13 4.20 4.26 3.93 4.11 4.04 4.16 4.13 4.10 4.06 4.32 4.04 4.15 4.19 4.24 4.32

8.29 8.04 7.94 7.70 8.49 8.14 8.29 8.12 7.99 7.95 7.89 8.19 8.02 8.02 7.99 7.82 7.78 7.93

4.16 4.24 4.52 4.31 4.10 4.21 4.17 3.82 4.05 3.98 4.12 4.09 4.09 4.05 4.24 3.96 4.12 4.17 4.22 4.30

25°C

5°C

NH

H~

NH

H~

NH

H"

-

4.12

-

8.10 7.68 7.78 7.57 7.90 7.81 7.92 7.78 7.80 7.75 7.72 7.77 7.64 7.78 7.72 7.64 7.84 7.87

4.10 4.16 4.57 4.30 4.13 4.18 4.31 4.00 4.14 4.08 4.18 4.13 4.10 4.07 4.36 4.07 4.10 4.16 4.24 4.31

8.31 7.98 8.00 7.75 7.99 7.89 8.11 7.9 a 7.92 7.97 b 7.92 7.81 7.85 7.86 7.79 7.66 8.04

4.18 4.58 4.32 4.15 4.22 4.35 4.06 4.16 4.10 4.23 4.15 4.12 4.11 4.37 4.07 4.17 4.21 4.24 4.32

8.41 8.18 8.13 7.84 8.18 8.03 8.33 8.04 8.00 8.06 7.94 8.07 7.99 7.99 7.95 7.82 7.86 8.06

4.12 4.18 4.53 4.31 4.12 4.22 4.26 3.99 4.08 4.02 4.16 4.09 4.1 a 4.08 4.28 3.98 4.12 4.18 4.20 4.30

a E stimated absolute u n c e r t a i n t y f o r these values is + 0.05 ppm; all other values are + 0.02 ppm. b Could not be d e t e r m i n e d due to spectral overlap.

89 seen in the NOESY spectrum in 50% TFE. (Observed N O E interactions are summarized in Fig. 6b). An increased number of a ( i ) / N H ( i + 3) interactions are present at the lower temperature as well as some a ( i ) / N H ( i + 4) interactions. Furthermore, an extensive network of N H ( i ) / N H ( i + 2) interactions from R5 to Q20 are visible, and this pattern of cross-peaks is indicative of a more stable, rigid helix at the lower temperature. The N M R results suggest that the increase in helical content observed in the CD spectrum

a)

1 M VLP

when the temperature is lowered from 25 to 5°C is primarily due to stabilization of the a-helix in the entire region from P4 to Q20. Conversely, when the temperature is raised to 45°C, very few Overhauser effects indicative of an c~-helix are observed (summarized in Fig. 6c). Some N H ( i ) / N H ( i + 1) interactions are observed from P4 to T10, but no a ( i ) / N H ( i + 3) interactions are present. This suggests that the helix is largely unraveled at the higher temperature in 50% TFE; the portion of the helix which

5 10 15 R LYTATSRAAFKAAKQ

•

oooooo

oo?

20 N.(1)/N.(;* 1)

?o~o

I--a---I I--.a--q I : I- - -?- q i__.a---i

b)

1

5

10

a(0/N.(i° :J)

15

20

MVLPRLYTATSRAAFKAAKQ 000000000 00077000

NH ( i ) / N H ( i * 1) NH (1)/N H (;* 2)

I

I

t

I

I

I

1 I ~-.~- 4 t

I ,I

p--.~--I

p-~-~

I:

1

C

_j~

(]]

10

1

5

10

20

•

? •

15

MVLPRLYTATSRAAFKAAKQ OeeeO ? ? ? •

5

I~ (i)/N H (I • 4.)

•I

15

MVLPRLYTATSRAAFKAAKQ QOO OO? ?

l

e

5

a(L)/N.(;. =)

10

N . 0 ) / . . ( ; ° 1)

20

~

•

15

N.(0/WH(;. ~)

20

MVLPRLYTATSRAAFKAAKQ OOOOOO00

~OO07

•

•

NH (i)/NH(;