:Biochimica et Biophysica Acta, 1120(1992) 59-68 ;© 1992 Elsevier Science Publishers B.V. All rights reserved 0167-4838/92/$05.00
59
BBAPRO 34141
Structural analysis of an outer surface protein from the Lyme disease spirochete, Borrelia burgdorferi, using circular dichroism and fluorescence spectroscopy Louisa Lee France i, Jan Kieleczawa, John J. Dunn, Geoffrey Hfnu and John C. Sutherland Biology Department, Brookharen National Laboratory, Upton, NY (USA) ( Received 8 August 199! )
Key words: Spirochete: Lyme disease: Outer surface protein: Protein structure; Circular dichroism; Fluorescence spectroscopy
The etiological agent of Lyme disease is the tick-borne spirochete, Borrelia burgdorferi. A major antigen of B. burgdorferi is a 31 kDa lipoprotein called outer surface protein A (OspA). Recently, a truncated form of OspA (lacking 17 amino acids at the N-terminus) was cloned, expressed and purified in large quantities (Dunn, J.J., Lade, B.A. and Barbour, A.G. (1990) Protein Expression and Purification 1,159-168). The truncated protein (OspA-257) is water-soluble, retains the ability to bind antibodies from the sera of Lyme disease patients and may prove useful in development of a vaccine against Lyme disease. We have used far UV circular dichroism (CD) and fluorescence spectroscopy to characterize the secondary structure of and to study conformational changes in OspA-257. CD spectra from 260 to 178 nm predict five classes of secondary structure: a-helix (11%), anti-parallel /3-sheet (32%), parallel /3-sheet (10%), /3-turns (18%) and aperiodic structures (including 'random coil') (30%). Analysis of the primary sequence of OspA yielded the most likely sites for a-helical regions (residues 100-107, 121-134, 253-273) and for antigenic determinants (Lys-46, Asp-82, Lys-231). CD spectra of the native protein show little change from pH 3 to 11. Thermal denaturation curves, indicate that 'salt bridges' play a role in stabilizing the native protein. Both thermal and chemical denaturations that eliminate all secondary structure as judged by CD or fluorescence are reversible. Denaturation by guanidine-HC! (gdn-HCI) appears to be a cooperative, two-state transition, as indicated by a sudden change in the CD spectrum at ~ 0.75 M gdn-HCl, and an isodichroic point at 208 nm in all CD spectra measured from 0.0-1.75 M gdn-HCl.
Introduction The incidence of Lyme disease has reached epidemic proportions in some regions of North America, with at least 6000 new cases reported yearly [1]. The etiological agent is the tick-borne spirochete, Borrelia
i
Present address: Plum Island Animal Disease Center, Greenport, NY 11957, U.S.A. Abbreviations: OspA, outer surface protein A; CD, circular dichroism; gdn-HCI, guanidine hydrochloride; OspA-257, truncated OspA; At, A R, absorption of left and right circularly polarized light; AA, AL-AR; ~L, ~R, extinction coefficients for left and right circularly polarized light; A~, ¢ t - ¢ a. Correspondence: J.C. Sutherland, Biology Department, Brookhaven National Laboratory, Upton, NY 11973, U.S.A.
burgdorferi [2-5]. A major antigen of B. burgdorferi is a 273 amino acid (31 kDa) lipoprotein called outer surface protein A (OspA)', the primary sequence of which is known [6]. OspA is believed to be directed to the outer membrane of the spirochete by a hydrophobic signal sequence, consisting of 17 amino acid residues at the N-terminus [6]. Full-length OspA is poorly expressed in E. coli [7]. Recently a truncated form of OspA (OspA-257), in which the first 17 residues at the N-terminus have been replaced by a single alanine residue has been cloned, expressed and purified in large quantities [7]. A recent report [4] indicates that antibodies to a recombinant full-length OspA protein provide protection in mice against challenge by several strains of B. burgdorferi. Dunn et al. [7] showed that OspA-257 binds antibodies from the sera of Lyme disease patients. The higher expression and solubility of OspA257 are desirable properties for vaccine development.
0 Characterization of the structure of OspA-257 may assist the elucidation of antigenic sites and eventual development of a vaccine. We have used circular dichroism (CD) and fluorescence spectroscopy to characterize the secondary structure and study conformational changes in OspA-257. CD spectra for wavelengths less than 240 nm are sensitive to the net secondary structure of the polypeptide backbone. The best estimates of secondary structure are obtained when CD spectra are extended to at least 178 nm [8]. Five types of secondary structure can be predicted from such CD spectra: a-helix, anti-parallel g-sheet, parallel fl-sheet, g-turns (all types) and aperiodic structures (including 'random coil'). We have analyzed the CD of OspA-257 using a computer program developed by Johnson and his colleagues [8-10], which uses matrix techniques (singular value decomposition) and statistical procedures (variable selection) to predict net secondary structure. The program fits the experimental CD spectrum to a linear combination of orthogonal CD spectra derived from a library of up to 22 reference proteins, whose secondary structures are known from X-ray diffraction. In contrast to CD, fluorescence techniques yield information about the micro-environment surrounding aromatic amino acids. OspA-257 contains a single tryptophan residue, located at position 216 in the full-length protein (Trp-216), and can be selectively excited, using wavelengths around 295 nm. Tryptophan fluorescence emission is sensitive to the polarity of the local environment [11]. We have used time-resolved and steadystate fluorescence techniques to demonstrate that Trp216 is buried within native OspA-257 in a hydrophobic environment, and that denaturation of the protein (by increasing the temperature or adding > 1.25 M guanidine=HCi (gdn,HCl))results in Trp-216 becoming completely solvent-exposed (as expected for a 'random coil' configuration). The native conformation of OspA-257 appears to be stable despite the absence of disulfide bridges, particularly with respect to pH. The native protein has also been shown to be resistant to digestion by trypsin [7], although its combined lysine and arginine content is 17%. The protein refolds to its native conformation after thermal denaturation at 96°C for < 10 min. Similarly, if a sample of the denatured OspA-257 (containing 1.25 M gdn-HCI) is diluted to 0.42 M gdn-HCl, its fluorescence emission maximum returns to the value characteristic of the native protein. Our CD experiments indicate that chemical denaturation of OspA-257 is cooperative, and the presence of an isodichroic point at 208 nm indicates that the transition between the native and denatured conformations take place in a single step. That is, no significant accumulation of spectrally distinct intermediate conformations is detected.
Materials and Methods
Materials OspA-257 was expressed and purified as described by Dunn et al. [7], exhaustively dialyzed against doubly-distilled water and diluted with a selected buffer to a buffer concentration of 10 mM. Unless otherwise specified, samples contained 10 mM potassium ph0s~ phate buffer (pH 7). For the pH-dependeney measurements, the following buffer systems were used: sodium acetate/acetic acid, for pH 3-6; K2HPO4/KH2PO 4, for pH 7-8; glycine/NaOH, for pH 9-11. GuanidineHCI and all chemical reagents for buffers were purchased from Sigma. Protein concentrations were determined from absorption measurements (Lambda 3 Spectrophotometer, Perkin Elmer), using the extinction coefficient at 280 nm, e2s0 = 10600 [7]. Methods Spectra. CD spectra were measured with the vacuum UV CD spectrometer at Port U9B of the National Synchrotron Light Source at Brookhaven National Laboratory [12]. This instr-lment measures absorption and CD spectra simultaneously [13], The concentration of the protein was made as high as possible, without letting the absorption of sample plus buffer exceed .-, 1.0 in the wavelength range of interest. Spectra were measured from 330 to 175 nm, at intervals of 0.5 nm. A 'Gray' cell [14] with quartz windows and an adjustable path length (determined by the thickness of a mylar spacer) was used to provide an optical path length of either 25 or 75/~m. The spectrometer was calibrated using the CD and absorption values reported by Chen and Yang [15] of an aqueous solution of (+)-10-camphor sulfonic acid in a 1 mm cuvette. The ratio of the CD at 192.5 n m t o that at 290,5 nm was -2.01, in good agreement w i t h published values [8]. Estimates of secondary structure. CD spectra w e r e analyzed with the program VARSELEC [8-10], which uses a statistical procedure (variable selection) to systematically choose basis sets that fit the experimental spectra, according to criteria set by the operator. Analysis of OspA-257 was accomplished by eliminating one protein at a time from the reference set of 22 proteins, yielding 22 possible basis sets. Nine of the 22 possible basis sets met the following criteria for a sufficient fit: (1) the sum of the different fractions of secondary structure was greater than 0.95 and less than 1.05; (2) all fractions were required to be greater than -0.05; (3) the root mean square of the errors was required to be less than 0.09. The final prediction was obtained by averaging the parameters from the nine sufficient fits. Predictions of the secondary structure and of hydrophobic and hydrophi!ic domains of both OspA-257 and of the full-length protein were based on the pri-
61 m a r y sequences [6,7], using the program P R O T Y L Z E (Scientific and Educational Software, State Line, PA), which includes: the Chou-Fasman method [16-20] and the Garnier-Osguthorpe-Robson method [21] for prediction of secondary structure; the Kyte-Doolittle [22] and the Hopp and Woods [23] plots for prediction of hydrophobic and hydrophilic domains; and the method o f Janin et ai. [24] which indicates the probability that a particular residue is located on the protein surface. Fluorescence measurements. Steady-state fluorescence emission spectra were measured at room temperature (23°C) on a Perkin Elmer MPF-4 fluorescence spectrophotometer or a Spex 212 fluorolog fluorometer. Tryptophan was selectively excited, using 295 nm radiation. For steady-state measurements, the protein concentration was such that the optical density at 295 nm was _ Pv. with "propensities' averaged over a four-residue window) are indicated as boxes above the trace of P, in the top panel. The three regions, designated by filled boxes, (residues 100-107, 121-134 and 253-273), which constitute 16% of the residues in OspA-257, have P,, >_ 1.12, indicating that helix formation is likely. These regions are candidates for the 28 residues (11%) of OspA-257 predicted by CD to be in helical conformation. The method of Garnier et al. predicts a high probability of helical regions with similar, but slightly different, boundaries (residues 98-113, 121-136 and 256-273). The simplest interpretation of the CD data is that two of these three regions are, and one region is not, in a helical conformation. The Chou and Fasman analysis for/j-sheet sites is shown in the bottom panel of Fig. 3A. The solid boxes above the P~ trace represent regions that are predicted to be in /j-sheet conformation by the Chou-Fasman method and which also significantly overlap regions predicted to be l-sheets by the Gamier method (positions 36-42, 70-74, 119-124, 126-132, 204-213, 218223, 225-233). These residues comprise 19% ot the
E3 ~
•
mm
~
E3
c ~
v
. K
.,,,,. W
D
klY.LI.L 'VvI iV t,w'vk7 -]
o~L~.~,
K 3U
D
100
Hydrophilicity
40
D
E" VtSl WlVI K
,-4
K
VtV'? VvNv i_ I I I
150
200 W
"1"I D
K
250
Plots for OspA
u
q/
k
~-~vVV
D
O0
Y
K
~ W
V
'
K
,'~,~
~',
W
K
Fig. 3. Primary sequence analyses of secondary structure sites and regions of hydrophilicity. The horizontal axes represent the amino acid sequence, with position l at the N-terminus, and position 273 at the C-terminus, Hatched boxes indicate hydrophobic signal sequence. Letters indicate: Lys-46 and Lys-231, K: Asp-82, D; Trp-216, W. (A) Chou-Fasman analysis for sites of a-helices (upper plot) and /]-sheets (lower plot). (B) Kyte-Doolittle (K-D, upper plot), ltopp and Woods (H-W, middle plot), and Surface Exposure (S-E, lower plot) analyses for regions of hydmphilicity (above the horizontal axes) and hydrophobicity (below the horizonlal axes).
total number of residues in OspA-257. The Gamier method and the Chou-Fasman method each under-predict the /j-sheet content (17 and 29%, respectively), relative to the CD analysis (42%), as indicated in Table 1. The most likely sites for fl-sheet structures appear to include all 50 residues (19%) listed above. For the full-length protein, an additional region (around residues 2-12) is also predicted by both the Chou-Fasman and the Gamier methods to be fl-sheet. Both methods also predict fl-turn content. However, the Chou-Fasman method analyzes /j-turns separately from a-helices and/j-sheets, and consequently residues predicted to be in fl-turns may also be predicted to be in helices or sheets, causing the fraction total for this method (1.16) to exceed unity (see Table I). In addition, aperiodic structures ('other') are not predicted by this method. Although the Gamier method does require all secondary structures to be mutually exclusive, and 'random coil' predictions are included, it does not
64! require a minimum number of residues for nucleation of a particular structure. So, single residues can be erroneously predicted to be in helices or turns. Thus, we are reluctant to associate the fractions of the residue predicted by analysis of the CD to be in E-turn o~ random conformation with specific regions of the primary sequence. Three analyses of hydrophUic/hydrophobie regions of the full-length OspA are presented in Fig. 3B, including a Kyte-Doolittle plot (K-D, top panel), a Hopps-Woods plot (H-W, center panel), and a surface exposure plot (S-E, bottom panel). The hydrophobic signal sequence at the N-terminus (positions 1-17) is evident in all three plots (hatched boxes). There is evidence that intact B. burgdorferi cells bind at least one penicillin molecule per protein [30]. The amino acid sequence at the binding site in all known penicillin-binding enzymes is: Ser-Xxx-Xxx-Lys [31]. The primary sequence of OspA-257 shows three such sites, with the active serine at positions 43, 84 and 116. We assume that a functional penicillin binding site is surface-exposed. Hopp and Woods found that in 12 out of 12 proteins in which antigenic determinants were known, the region of greatest hydrophilicity was associated with a known antigenic determinant, while for the regions with the second and third highest indices of hydrophilicity, 50% were associated with antigenic determinants. In both the Hopp and Woods and Kyte-Doolittle plots of OspA, the most hydrophilic region is located at Lys-46 (hydrophilicity index= 2.53), which is three residues downstream from Ser-43 (hydrophilicity index = 1.00), and hence a potential penicillin binding site. The S-E plot indicates that Lys-46 is surface-exposed.
The second and third most hydrophilic sites contain Lys-231 (hydrophilicity index = 1.98) and Asp-82 (hydrophilieity index = 1.97), respectively. According to Hopp and Woods, the probability that each of these two regions is antigenic is ~ 0.50. Asp-82 is two residues upstream from the putative active serine at position 84 (hydrophilicity index = 1.80). Thus, serine84 also represents a potential penicillin binding site.
OspA-257 is stable from pH 3 to 11 Protein stability can be altered by changes in hydrogen ion concentration and pH titrations have been widely used to study such changes [32,33]. At extremes of pH, the increased density of like charges is expected to destabilize native structures [29]. Titration of OspA-257 over a broad range of pH shows that the native conformation of this protein, as reflected in the CD spectra, is stable from pH 3 to pH 11 (Fig. 1). The amplitude of the positive peak at 194
nm is invariant within experimental error, while the amplitude of the negative peak changes by ~ 13%. The relative stability of native OspA-257 with respect to pH may arise from the potentially large number of 'salt bridges' which appear likely to form (vide ante, Fig. 2). If a large fraction of charged residues are involved in 'salt bridges' at physiological pH, the destabilizing effect of high or low pH is screened, making the protein more stable with respect to pH. OspA-257 is stable orer a wide range o f salt concentrations The effect of electrostatic interactions on protein stability has been studied using ionic strength as a variable [29]. Increased ionic strength results in increased shielding of the charges on the protein. However, as pointed out by Dill [29], the shielding effect is only observed at low ionic strength ( 1 M, shielding is saturated. Absorption by salts confines studies of ionic strength effects on protein CD spectra to wavelengths above -- 200 nm, with consequent loss of half the information available in spectra measured to 178 nm [8]. We therefore turned to the fluorescence emission of Trp-216 to investigate the effect of ionic strength on OspA-257. It has been shown that Trp-216 is buried in a hydrophobic pocket in OspA-257 and is characterized by an emission maximum of 330 nm and a mean fluorescence lifetime of 1.12 ns in the native protein under physiological conditions (150 mM NaCl), while in the denatured protein (6 M gdn-HCl), the emission maximum shifts to 355 nm and the mean fluorescence lifetime increases to 1.67 ns, indicating that Trp,216 is com, pletely solvent,exposed (France, L.L., u n p u b l i s h e d data): We monitored both the emission m a ~ m u m and the fluorescence decay profile at increasing concentra' tions of NaCI from 0.0 to 1.0 M. The emission maximum showed no change, remaining constant at 330 nm. The mean lifetime increased only 6% from 1.11 ns (no NaCl) to 1.18 ns (1.0 M NaCl), and far less than the 49% increase resulting from denaturation by gdn-HCl. The individual fluorescence decay parameters and the mean lifetimes at each NaCl concentration are listed in Table II. The 6% increase in the mean lifetime evidently results from a similar increase in the lifetime of the fast component (~-j), perhaps indicating that the dominant component (~-~) arises from a ground state conformation in which Trp-216 is slightly more solvent-exposed. The fractional contribution, Oi, of each component to the total fluorescence yield, remains constant over the entire range of sodium chloride concentrations. Thus, the shielding effect at high ionic strength does not appear to alter significantly the hydrophobic environment of Trp-216 in the native protein.
65 360
!
i
!
!
i
E
-'"
c 350
E" E 345
//
:"
E 340
g 335 uJ
330
~-----&-T t
-~1~--~& I
20
-~"
I
I
Temperatuce.
I
I
I
1
°C
Fig. 4. Temperature-induced shifts in the steady-state fluorescence emission maximum of Trp-216 in OspA-257, excited at 295 nm. Measurements made as the temperature was increased, in the presence ( A ) and in the absence (zx) of 150 mM NaCI. Measurements made as the temperature was decreased, in the presence ( • ) and absence ( v ) t~f 150 mM NaCI.
Thermal denaturation and renaturation The emission maximum of a sample of OspA-257 was monitored as the temperature was increased in incremental stages from 23 to 80°C, and then decreased (incrementally) to 23°C, with 20 min incubation at each temperature (Fig. 4) before the spectrum was recorded. These experiments were repeated in the presence and in the absence of 150 mM NaCI, to compare denaturation and renaturation under physiological conditions (150 mM NaCI) with denaturation and renaturation under conditions used for CD experiments (no NaCl). The emission maximum under both conditions shifts from 330 nm at temperatures < 55°C, to 355 nm at temperatures > 74°C. As shown in Fig. 4, t h e midpoints of the wavelength shifts are roughly 59°C (150 mM NaCI) and 63°C (no NaCI). It appears that t h e protein is destabilized slightly by the presence of 150 mM NaCI. The midpoints for the wavelength shifts for the decreasing temperature curves are about 53°C (150 mM NaCI) and 55°C (no NaCl). Again, the presence of 150 mM NaCl appears to destabilize the pro-
tein slightly. The effect of ionic strength on protein stability depends on the type of electrostatic interactions which predominate in a given protein [29]. If classical electrostatic interactions (i.e., nonspecific charge repulsions) are important, then increasing the salt concentration will increase the shielding and stabilize the native protein. If specific intramolecular 'salt bridges' are important, then increasing salt concentrations (increased shielding) destabilizes the native protein, as observed for OspA-257. The renaturation curves are displaced 5 to 10 degrees towards lower temperatures compared to the corresponding denaturation data, indicating that the system was not in a completely reversible thermodynamic equilibrium. This could be due either to a tendency of the unfolded protein to form a structure that resists refolding or to the need for a longer holding period prior to recording the emission spectrum. As a corollary to the above experiment, eight independent protein samples were incubated at a selected temperature for 20 min (37, 46, 56, 65, 72, 81, 89 and 100°C), then returned to 23°C, and the emission maximum of each was measured. The fluorescence of a ninth sample that was maintained at 23°C for the same period was also recorded. The emission maximum for each sample was 330 nm, except for the protein incubated at 100°C, which showed an emission maximum of 332 nm. Thus, it appears that thermal denaturation is largely reversible, as monitored by the fluorescence emission of Trp-216. We measured the CD spectrum of the refolded OspA-257 after thermal denaturation (at 96°C) for different periods of time from zero to 60 min. Vials containing different protein samples were immersed in a thermostatted water bath for selected periods of time (zero to 60 min), after which each protein sample was placed on ice for > 60 min, and its CD spectrum was measured (at 23°C). We observed no significant change, within experimental uncertainty, in peak amplitudes (Ad194 nm), Ad213 nm)) nor in the crossing point (200 nm) for denaturation times _< 10 min at 96°C.
TABLE II
Tryptophan fluorescence parameters at different NaCI concentrations (from 0 to 1.0 M) [NaCI] (M)
¢1 (ns)
oq
@l
T2 (ns)
a2
@2
X~
(~') (ns)
0 0.0739 0.150 0.224 0.301 0.450 0.599 0.750 1.00
1.08 1.08 1.08 1.07 i.10 1.10 1.11 1.11 1.13
0.987 0.986 0.979 0.979 0.982 0.982 0.974 0.973 0.980
0.956 0.951 0.942 0.932 0.946 0.947 0.934 0.930 0.942
3.77 3.91 3.22 3.56 3.48 3.36 2.97 2.93 3.43
0.0129 0.0139 0.0203 0.0215 0.0178 0.0180 0.0256 0.0276 0.0200
0.0436 0.0486 0.0582 0.0681 0.0542 0.0530 0.0657 0.0697 0.0583
0.931 0.903 0.768 0.853 0.856 0.873 0.734 0.710 0.753
l. 11 I. 12 l. 12 i. 12 i. 14 I. 14 1.16 1.16 1.18
66 2C
•
i
•
i
,
i
'
i
,
r
,
i
.
,
,
[
•
I
'
J.
}
:
:
,
I 250
IC
~ oc -1C
•. . . "
-2C
-4i
•
I
. . . . . . . .
,
:
oo[ 2oE
-20 40 190
I l l 200 210
I I 220
Wovelencjt
l
I%
l l 230
240
nm
Fig. 5. Denaturation o f OspA-257. induced by increasing concentrations of guanidine-HCI (0.0 to 1.75 M), as probed by CD, U p p e r plot: 0,0 M (): 0.125 M ( ): 0.25 M. ( - - : - - ) : 0.50 M ( . . . . . . ); 0.75 M ( . . . . . . ). Center plot: 1.0 M ( ); 1.25 M ( - - - - - ) ; !.50 M ( . . . . ): 1.75 M ( . . . . . . ). Lower plot: 0.0 M ( ); 0.5 M ( - - - - - - ) : 0.75 M ( . . . . ): 1.0 M ( . . . . . . ); i.75 M ( . . . . . . ). The pH was maintained at 7 for these experiments.
Immersion for 20 min produced only a decrease of 40% in the positive peak (194 nm). However, after 30 min immersion, the CD spectrum showed large distortions, although the positions of the positive and negative peaks were similar to the native spectrum. After 60 min immersion, the CD spectrum consisted of a large negative peak at 200 nm (indicative of random coil ), a n d no characteristics of the native spectrum remained. The native conformation of OspA-257 thus appears to be energetically favorable as judged by the ability of the protein to refold after thermal denaturation at 96°C for < 10 rain. Furthermore, although the plots of emission maxima as a function of temperature (Fig. 4) suggest that intramolecular salt bridges play a role in stabilizing the native state, both steady-state and timeresolved fluorescence experiments show that the presence of < 1 M NaC! does not affect the native conformation enough to significantly alter the Trp-216 environment.
Denaturation by guanidine-HCl CD spectra were measured for OspA-257 at different concentrations of gdn-HCl from 0.0 to 1.75 M (Fig. 5). A superposition of five CD spectra, measured at 0.0, 0.125, 0.25, 0.50 and 0.75 M gdn-HCi, is shown in the upper plot. Spectra were truncated at 198 nm
because of the absorption of gdn-HCI at shorter wavelengths. The native conformation appears stable at [gdn-HCI] < 0.5 M. The 0.75 M gdn-HCI spectrum shows a large negative peak at 200 nm, indicating the presence of a significant fraction of random coil. In the center plot, the CD spectra measured at 1.00, 1.25, 1.50 and 1.75 M gdn-HC! are shown. An increase in the amplitude of the negative peak at about 200 nm from 1.00 M to 1.25 M, indicates that denaturation is almost 85% complete at 1.00 M gdn-HCI. No further spectral change takes place at [gdn-HCI] > 1.25 M. The transition from the native to the denatured state appears to be cooperative, i.e., denaturation occurs rapidly and suddenly, with little change occurring at gdn-HCI < 0.50 M or at gdn-HCl > 1.25 M. The lower plot in Fig. 5 demonstrates that CD spectra measured over the complete range of gdn-HCi concentrations (from zero to 1.75 M) pass through an isodichroic point at 208 nm. Although, for clarity, only five representative spectra are shown in the lower plot, all measured spectra pass through this point. isodichroic points in CD spectra have been predicted at ~ 204 nm for a helix-to-coil and at ~ 208 nm for a sheet-to-coil transition [34]. The isodichroic point observed for OspA-257 is thus consistent with that expected for a protein with low a-helical and high fl-sheet content (Table 1). The presence of an isodichroic point suggests that for the transition from the native to the denatured state, only two, spectraIiy distinct, major conformations are present (i.e., partially-folded intermediate conformations are not present in significant quantity), it follows that a CD spectrum measured at any intermediate concentration of denaturant will be a linear combination of the two basis CD spectra: the native state (0,0 M gdn-HCl), and the denatured state (1.75 M gdn-HCl). Thus, the fraction of native state present at any selected concentration M can be written: Ae(A, M)- AeD(A ) fN(M)=
ArN(,~)_ A ~ D ( A )
,
(4)
where fN is the fraction of the native state, AE(A, M ) is the observed differential extinction coefficient at wavelength A and denaturant concentration M, and eN(A) and Co(A) are the differential extinction coefficients at wavelength A for the native and denatured states, respectively. The cooperative nature of the unfolding transition for OspA-257 is illustrated in Fig. 6. The fraction fN was calculated at each concentration of gdn-HC! (from 0 to 1.75 M), using Eqn. 4 and values for A¢ measured at 201 nm. This wavelength was selected because the difference between the native and denatured CD spectra is large. As shown in Fig. 6, fN remains constant ( = 1.0) for M from 0.0 to 0.5. At 0.75 M and 1.0 M
67 100
- -
Acknowledgements
J
O80 f 06C Z 04C 02¢ ,O
OOC ooo
|
I
050
i
I
1.OO [Gdn
- HC[~,
i
!
150
i
2.00
M
Fig. 6. Fraction of native OspA-257 present at increasing concentrations of guanidine-HCl, as calculated from the CD spectra at 201 nm. .IN is calculated from Eqn. 4, using values for de measured at 201 nm for each concentration of gdn-HCI (0.0M to !.75 M).
gdn-HCl, fN is calculated to be 0.37, and 0.14, respectively. At [gdn-HCi] > 1.25 M, fN remained constant at = 0.0. Similar titration curves were obtained using CD values at 204, 210 and 218 nm (data not shown). The reversibility of denaturation by gdn-HC1 was tested, using steady-state fluorescence. A sample of OspA-257 was mixed with 1.25 M gdn-HCI, and its emission maximum was 355 nm, as expected for the denatured protein. The sample was then diluted 3-fold (to 0.42 M gdn-HCl), and its emission maximum was found to be 330 rim, characteristic of the native protein. Thus, both chemical and thermal denaturation appear to be completely reversible for OspA-257, under limiting conditions (< 1.25 M gdn-HCi and < 10 rain heating at 96°C, respectively). The significant stability of the native state of OspA257 with respect to pH and ionic strength, in addition to its ability to refold after denaturation by 1.25 M gdn-HCl or incubation at 96°C for _< 10 rain, presumably reflects structural features related to the protein's biological function. OspA is one of two major outer surface proteins which have been identified in Borrelia burgdorferi. These two proteins (OspA and OspB) have similar masses and show a large degree of sequence homology. Although these outer surface proteins may play an important role in the progression of Lyme disease, neither the specific function of the outer surface proteins nor the specific advantage of their stability for the survival of the spirochete are known. To our knowledge, this work is the first physical/ chemical structural analysis of an outer surface protein from the Borrelia genus. Experiments are in progress to investigate recombinant OspB, as well as the outer membrane proteins from related species of Borrelia. These experiments will help determine if the stability of native OspA is perhaps a general characteristic of the outer surface proteins of Borrelia, and may contribute to an understanding of their function and their ability to be recognized as immunogens.
We thank Ann Emrick, Denise Monteleone and John Trunk for assistance with the experiments and analysis of data, and Dr. Erwin London (State University of New York at Stony Brook) for the use of his Spex 212 fluorolog fluorometer. We also thank Dr. Curtis W. Johnson and his colleagues at Oregon State University for providing a copy of VARSELEC, the program for the analysis of protein secondary structure. Research supported by the Office of Health and Environmental Research and the Division of Energy Biosciences, Office of Basic Energy Sciences (JK and GH), United States Department of Energy (USDOE) and a grant from the National Institutes of Health (GM34662) to JCS. The National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory is supported by the Office of Chemical Research and The Office of Materials Research, USDOE. The circular dichroism spectrometer and fluorometer at port U9B of the NSLS are supported by the Office of Health and Environmental Research, USDOE.
References t Centers for Disease Control (1989) Morb. Mortal. Wkly. 38, 668. 2 Burgdorfer, W., Barbour, A.G., Hayes, S.F., Benach, J.L., Grunwaldt, E. and Davis, J.P. (1982) Science 216, 1317-1319. 3 Steere, A.C. (1989) N. Engl. J. Med. 321,586-596 4 Fikrig, E., Barthold, W.W., Kantor, F.S. and FlavelL R.A. (1990) Science 250, 553-556. 5 Persing. D.H., Telford III, S.R., Rys, P.N., Dodge, D.E., White, T.J., Malawista, S.E. and Spielman, A. (1990) Science 249. 14201422. 6 Bergstrom, S., Bundoc, V.G. and Barbour, A.G. (1989) Mol. Microbiol. 3, 479-486. 7 Dunn. J.J., Lade, B.A. and Barbour, A.G. (1990) Protein Expression and Purification 1, 159-160. 8 Johnson, C.W., Jr. (1990) Pro. Struct. Funct. Gen. 7, 4108-4116. 9 Hennessey, LP., Jr. and Johnson, W.C., Jr. (1981) Biochemistry 20, 1085-1094. l0 Manavalen, P. and Johnson, W.C., Jr. (1987)Anal. Biochem. 167, 76-85. 11 Eflink, M.R. and Ghiron, C.A. (1981)Anal. Biochem. 114, 119227. 12 Sutherland, J.C., Desmond, E.J. and Takacs, P.Z. (1980) Nucl. Instr. and Meth. 172, 195-199. 13 Sutherland, J.C., Keck, P.C., Griffin, K.P. and Takacs, P.Z. (1982) Nucl. Instr. and Meth. 195, 375-379. 14 Gray, D.M, Lane, D., Kuner, E., Vaughan, M. and Sutherland. J. (1984) Anal. Biochem. 136, 247-250. 15 Chen. G.C. and Y~ng, J.T. (1977)Analyt. Lett. 10. 1195-1207. 16 Chou, P.Y. and Fasman, G. (1974a) Biochemistry 13, 211-222. 17 Chou, P.Y. and Fasman, G. (1974b) Biochemistry 13, 222-245. 18 Chou, P.Y. and Fasman, G. (1978a) Adv. Enzymol. 47, 45-148. 19 Chou, P.Y. and Fasman, G. (1978b) Annu. Rev. Biochem. 47, 251-276. 20 Chou, P.Y. and Fasman, G. (1979) Biophys. J. 26, 367-384. 21 Gamier, J., Osguthorpe, D.J. and Robson, B. (1978) J. Mol. Biol. 120, 97-120. 22 Kyte, L and Doolittle, R.F. (1982)J. Mol. Biol. 157. 104-132.
68 23 Hopp. T.P. and Woods. K.R. (1981) Proe. Natl. Acad. Sci. USA 78, 3824-3828. 24 Janin, J.. Wodak, S.. Levitt, M. and Maigret, B. (1978) J. Mol. Biol. 125, 357-386. 25 Laws, W.R. and Sutherland. J.C. (1986) Photochem. Photobiol. 44, 343-348. 26 O'Connor. D.B. and Phillips. D. (1984) Time-Correlated Single Photon Counting, Academic Press. London. 27 Grinvald, A. and Steinberg, 1.Z. (1976) Bioehim. Biophys. Acta 427, 663-678. 28 Barlow, DJ. and Thornton, J.M. (1983) J. Mol. Biol. 168. 867-885.
29 Dill, K.A. (19911) Biochemistry 29, 7133-7155. 30 Urban. C, Rahal, J.J., Dattwyller, R.J., Gorcvic, P. and Luft, BJ. (1990) J. Bacteriol. 172, 6139-6141, 31 Duez, C.. Prion-Fraipont. C., Joris, B., Dusart, J., Urdea. M.S., Martial, J.A., Frere, J.M., and Ghuysen, J.M. (1987) Eur. J. Biochem. 162, 509-518. 32 Pace, C.N., Laurents. D.V. and Thomson, J.A. (1990) Biochemistry 29, 2564-2572. 33 Privalov, P.L (1979) Ads'. Protein Chem. 33, 167-241. 34 Greenfield, N. and Fasman, G.D. (1969) Biochemistry 8. 41fl84tl6.
59
BBAPRO 34141
Structural analysis of an outer surface protein from the Lyme disease spirochete, Borrelia burgdorferi, using circular dichroism and fluorescence spectroscopy Louisa Lee France i, Jan Kieleczawa, John J. Dunn, Geoffrey Hfnu and John C. Sutherland Biology Department, Brookharen National Laboratory, Upton, NY (USA) ( Received 8 August 199! )
Key words: Spirochete: Lyme disease: Outer surface protein: Protein structure; Circular dichroism; Fluorescence spectroscopy
The etiological agent of Lyme disease is the tick-borne spirochete, Borrelia burgdorferi. A major antigen of B. burgdorferi is a 31 kDa lipoprotein called outer surface protein A (OspA). Recently, a truncated form of OspA (lacking 17 amino acids at the N-terminus) was cloned, expressed and purified in large quantities (Dunn, J.J., Lade, B.A. and Barbour, A.G. (1990) Protein Expression and Purification 1,159-168). The truncated protein (OspA-257) is water-soluble, retains the ability to bind antibodies from the sera of Lyme disease patients and may prove useful in development of a vaccine against Lyme disease. We have used far UV circular dichroism (CD) and fluorescence spectroscopy to characterize the secondary structure of and to study conformational changes in OspA-257. CD spectra from 260 to 178 nm predict five classes of secondary structure: a-helix (11%), anti-parallel /3-sheet (32%), parallel /3-sheet (10%), /3-turns (18%) and aperiodic structures (including 'random coil') (30%). Analysis of the primary sequence of OspA yielded the most likely sites for a-helical regions (residues 100-107, 121-134, 253-273) and for antigenic determinants (Lys-46, Asp-82, Lys-231). CD spectra of the native protein show little change from pH 3 to 11. Thermal denaturation curves, indicate that 'salt bridges' play a role in stabilizing the native protein. Both thermal and chemical denaturations that eliminate all secondary structure as judged by CD or fluorescence are reversible. Denaturation by guanidine-HC! (gdn-HCI) appears to be a cooperative, two-state transition, as indicated by a sudden change in the CD spectrum at ~ 0.75 M gdn-HCl, and an isodichroic point at 208 nm in all CD spectra measured from 0.0-1.75 M gdn-HCl.
Introduction The incidence of Lyme disease has reached epidemic proportions in some regions of North America, with at least 6000 new cases reported yearly [1]. The etiological agent is the tick-borne spirochete, Borrelia
i
Present address: Plum Island Animal Disease Center, Greenport, NY 11957, U.S.A. Abbreviations: OspA, outer surface protein A; CD, circular dichroism; gdn-HCI, guanidine hydrochloride; OspA-257, truncated OspA; At, A R, absorption of left and right circularly polarized light; AA, AL-AR; ~L, ~R, extinction coefficients for left and right circularly polarized light; A~, ¢ t - ¢ a. Correspondence: J.C. Sutherland, Biology Department, Brookhaven National Laboratory, Upton, NY 11973, U.S.A.
burgdorferi [2-5]. A major antigen of B. burgdorferi is a 273 amino acid (31 kDa) lipoprotein called outer surface protein A (OspA)', the primary sequence of which is known [6]. OspA is believed to be directed to the outer membrane of the spirochete by a hydrophobic signal sequence, consisting of 17 amino acid residues at the N-terminus [6]. Full-length OspA is poorly expressed in E. coli [7]. Recently a truncated form of OspA (OspA-257), in which the first 17 residues at the N-terminus have been replaced by a single alanine residue has been cloned, expressed and purified in large quantities [7]. A recent report [4] indicates that antibodies to a recombinant full-length OspA protein provide protection in mice against challenge by several strains of B. burgdorferi. Dunn et al. [7] showed that OspA-257 binds antibodies from the sera of Lyme disease patients. The higher expression and solubility of OspA257 are desirable properties for vaccine development.
0 Characterization of the structure of OspA-257 may assist the elucidation of antigenic sites and eventual development of a vaccine. We have used circular dichroism (CD) and fluorescence spectroscopy to characterize the secondary structure and study conformational changes in OspA-257. CD spectra for wavelengths less than 240 nm are sensitive to the net secondary structure of the polypeptide backbone. The best estimates of secondary structure are obtained when CD spectra are extended to at least 178 nm [8]. Five types of secondary structure can be predicted from such CD spectra: a-helix, anti-parallel g-sheet, parallel fl-sheet, g-turns (all types) and aperiodic structures (including 'random coil'). We have analyzed the CD of OspA-257 using a computer program developed by Johnson and his colleagues [8-10], which uses matrix techniques (singular value decomposition) and statistical procedures (variable selection) to predict net secondary structure. The program fits the experimental CD spectrum to a linear combination of orthogonal CD spectra derived from a library of up to 22 reference proteins, whose secondary structures are known from X-ray diffraction. In contrast to CD, fluorescence techniques yield information about the micro-environment surrounding aromatic amino acids. OspA-257 contains a single tryptophan residue, located at position 216 in the full-length protein (Trp-216), and can be selectively excited, using wavelengths around 295 nm. Tryptophan fluorescence emission is sensitive to the polarity of the local environment [11]. We have used time-resolved and steadystate fluorescence techniques to demonstrate that Trp216 is buried within native OspA-257 in a hydrophobic environment, and that denaturation of the protein (by increasing the temperature or adding > 1.25 M guanidine=HCi (gdn,HCl))results in Trp-216 becoming completely solvent-exposed (as expected for a 'random coil' configuration). The native conformation of OspA-257 appears to be stable despite the absence of disulfide bridges, particularly with respect to pH. The native protein has also been shown to be resistant to digestion by trypsin [7], although its combined lysine and arginine content is 17%. The protein refolds to its native conformation after thermal denaturation at 96°C for < 10 min. Similarly, if a sample of the denatured OspA-257 (containing 1.25 M gdn-HCI) is diluted to 0.42 M gdn-HCl, its fluorescence emission maximum returns to the value characteristic of the native protein. Our CD experiments indicate that chemical denaturation of OspA-257 is cooperative, and the presence of an isodichroic point at 208 nm indicates that the transition between the native and denatured conformations take place in a single step. That is, no significant accumulation of spectrally distinct intermediate conformations is detected.
Materials and Methods
Materials OspA-257 was expressed and purified as described by Dunn et al. [7], exhaustively dialyzed against doubly-distilled water and diluted with a selected buffer to a buffer concentration of 10 mM. Unless otherwise specified, samples contained 10 mM potassium ph0s~ phate buffer (pH 7). For the pH-dependeney measurements, the following buffer systems were used: sodium acetate/acetic acid, for pH 3-6; K2HPO4/KH2PO 4, for pH 7-8; glycine/NaOH, for pH 9-11. GuanidineHCI and all chemical reagents for buffers were purchased from Sigma. Protein concentrations were determined from absorption measurements (Lambda 3 Spectrophotometer, Perkin Elmer), using the extinction coefficient at 280 nm, e2s0 = 10600 [7]. Methods Spectra. CD spectra were measured with the vacuum UV CD spectrometer at Port U9B of the National Synchrotron Light Source at Brookhaven National Laboratory [12]. This instr-lment measures absorption and CD spectra simultaneously [13], The concentration of the protein was made as high as possible, without letting the absorption of sample plus buffer exceed .-, 1.0 in the wavelength range of interest. Spectra were measured from 330 to 175 nm, at intervals of 0.5 nm. A 'Gray' cell [14] with quartz windows and an adjustable path length (determined by the thickness of a mylar spacer) was used to provide an optical path length of either 25 or 75/~m. The spectrometer was calibrated using the CD and absorption values reported by Chen and Yang [15] of an aqueous solution of (+)-10-camphor sulfonic acid in a 1 mm cuvette. The ratio of the CD at 192.5 n m t o that at 290,5 nm was -2.01, in good agreement w i t h published values [8]. Estimates of secondary structure. CD spectra w e r e analyzed with the program VARSELEC [8-10], which uses a statistical procedure (variable selection) to systematically choose basis sets that fit the experimental spectra, according to criteria set by the operator. Analysis of OspA-257 was accomplished by eliminating one protein at a time from the reference set of 22 proteins, yielding 22 possible basis sets. Nine of the 22 possible basis sets met the following criteria for a sufficient fit: (1) the sum of the different fractions of secondary structure was greater than 0.95 and less than 1.05; (2) all fractions were required to be greater than -0.05; (3) the root mean square of the errors was required to be less than 0.09. The final prediction was obtained by averaging the parameters from the nine sufficient fits. Predictions of the secondary structure and of hydrophobic and hydrophi!ic domains of both OspA-257 and of the full-length protein were based on the pri-
61 m a r y sequences [6,7], using the program P R O T Y L Z E (Scientific and Educational Software, State Line, PA), which includes: the Chou-Fasman method [16-20] and the Garnier-Osguthorpe-Robson method [21] for prediction of secondary structure; the Kyte-Doolittle [22] and the Hopp and Woods [23] plots for prediction of hydrophobic and hydrophilic domains; and the method o f Janin et ai. [24] which indicates the probability that a particular residue is located on the protein surface. Fluorescence measurements. Steady-state fluorescence emission spectra were measured at room temperature (23°C) on a Perkin Elmer MPF-4 fluorescence spectrophotometer or a Spex 212 fluorolog fluorometer. Tryptophan was selectively excited, using 295 nm radiation. For steady-state measurements, the protein concentration was such that the optical density at 295 nm was _ Pv. with "propensities' averaged over a four-residue window) are indicated as boxes above the trace of P, in the top panel. The three regions, designated by filled boxes, (residues 100-107, 121-134 and 253-273), which constitute 16% of the residues in OspA-257, have P,, >_ 1.12, indicating that helix formation is likely. These regions are candidates for the 28 residues (11%) of OspA-257 predicted by CD to be in helical conformation. The method of Garnier et al. predicts a high probability of helical regions with similar, but slightly different, boundaries (residues 98-113, 121-136 and 256-273). The simplest interpretation of the CD data is that two of these three regions are, and one region is not, in a helical conformation. The Chou and Fasman analysis for/j-sheet sites is shown in the bottom panel of Fig. 3A. The solid boxes above the P~ trace represent regions that are predicted to be in /j-sheet conformation by the Chou-Fasman method and which also significantly overlap regions predicted to be l-sheets by the Gamier method (positions 36-42, 70-74, 119-124, 126-132, 204-213, 218223, 225-233). These residues comprise 19% ot the
E3 ~
•
mm
~
E3
c ~
v
. K
.,,,,. W
D
klY.LI.L 'VvI iV t,w'vk7 -]
o~L~.~,
K 3U
D
100
Hydrophilicity
40
D
E" VtSl WlVI K
,-4
K
VtV'? VvNv i_ I I I
150
200 W
"1"I D
K
250
Plots for OspA
u
q/
k
~-~vVV
D
O0
Y
K
~ W
V
'
K
,'~,~
~',
W
K
Fig. 3. Primary sequence analyses of secondary structure sites and regions of hydrophilicity. The horizontal axes represent the amino acid sequence, with position l at the N-terminus, and position 273 at the C-terminus, Hatched boxes indicate hydrophobic signal sequence. Letters indicate: Lys-46 and Lys-231, K: Asp-82, D; Trp-216, W. (A) Chou-Fasman analysis for sites of a-helices (upper plot) and /]-sheets (lower plot). (B) Kyte-Doolittle (K-D, upper plot), ltopp and Woods (H-W, middle plot), and Surface Exposure (S-E, lower plot) analyses for regions of hydmphilicity (above the horizontal axes) and hydrophobicity (below the horizonlal axes).
total number of residues in OspA-257. The Gamier method and the Chou-Fasman method each under-predict the /j-sheet content (17 and 29%, respectively), relative to the CD analysis (42%), as indicated in Table 1. The most likely sites for fl-sheet structures appear to include all 50 residues (19%) listed above. For the full-length protein, an additional region (around residues 2-12) is also predicted by both the Chou-Fasman and the Gamier methods to be fl-sheet. Both methods also predict fl-turn content. However, the Chou-Fasman method analyzes /j-turns separately from a-helices and/j-sheets, and consequently residues predicted to be in fl-turns may also be predicted to be in helices or sheets, causing the fraction total for this method (1.16) to exceed unity (see Table I). In addition, aperiodic structures ('other') are not predicted by this method. Although the Gamier method does require all secondary structures to be mutually exclusive, and 'random coil' predictions are included, it does not
64! require a minimum number of residues for nucleation of a particular structure. So, single residues can be erroneously predicted to be in helices or turns. Thus, we are reluctant to associate the fractions of the residue predicted by analysis of the CD to be in E-turn o~ random conformation with specific regions of the primary sequence. Three analyses of hydrophUic/hydrophobie regions of the full-length OspA are presented in Fig. 3B, including a Kyte-Doolittle plot (K-D, top panel), a Hopps-Woods plot (H-W, center panel), and a surface exposure plot (S-E, bottom panel). The hydrophobic signal sequence at the N-terminus (positions 1-17) is evident in all three plots (hatched boxes). There is evidence that intact B. burgdorferi cells bind at least one penicillin molecule per protein [30]. The amino acid sequence at the binding site in all known penicillin-binding enzymes is: Ser-Xxx-Xxx-Lys [31]. The primary sequence of OspA-257 shows three such sites, with the active serine at positions 43, 84 and 116. We assume that a functional penicillin binding site is surface-exposed. Hopp and Woods found that in 12 out of 12 proteins in which antigenic determinants were known, the region of greatest hydrophilicity was associated with a known antigenic determinant, while for the regions with the second and third highest indices of hydrophilicity, 50% were associated with antigenic determinants. In both the Hopp and Woods and Kyte-Doolittle plots of OspA, the most hydrophilic region is located at Lys-46 (hydrophilicity index= 2.53), which is three residues downstream from Ser-43 (hydrophilicity index = 1.00), and hence a potential penicillin binding site. The S-E plot indicates that Lys-46 is surface-exposed.
The second and third most hydrophilic sites contain Lys-231 (hydrophilicity index = 1.98) and Asp-82 (hydrophilieity index = 1.97), respectively. According to Hopp and Woods, the probability that each of these two regions is antigenic is ~ 0.50. Asp-82 is two residues upstream from the putative active serine at position 84 (hydrophilicity index = 1.80). Thus, serine84 also represents a potential penicillin binding site.
OspA-257 is stable from pH 3 to 11 Protein stability can be altered by changes in hydrogen ion concentration and pH titrations have been widely used to study such changes [32,33]. At extremes of pH, the increased density of like charges is expected to destabilize native structures [29]. Titration of OspA-257 over a broad range of pH shows that the native conformation of this protein, as reflected in the CD spectra, is stable from pH 3 to pH 11 (Fig. 1). The amplitude of the positive peak at 194
nm is invariant within experimental error, while the amplitude of the negative peak changes by ~ 13%. The relative stability of native OspA-257 with respect to pH may arise from the potentially large number of 'salt bridges' which appear likely to form (vide ante, Fig. 2). If a large fraction of charged residues are involved in 'salt bridges' at physiological pH, the destabilizing effect of high or low pH is screened, making the protein more stable with respect to pH. OspA-257 is stable orer a wide range o f salt concentrations The effect of electrostatic interactions on protein stability has been studied using ionic strength as a variable [29]. Increased ionic strength results in increased shielding of the charges on the protein. However, as pointed out by Dill [29], the shielding effect is only observed at low ionic strength ( 1 M, shielding is saturated. Absorption by salts confines studies of ionic strength effects on protein CD spectra to wavelengths above -- 200 nm, with consequent loss of half the information available in spectra measured to 178 nm [8]. We therefore turned to the fluorescence emission of Trp-216 to investigate the effect of ionic strength on OspA-257. It has been shown that Trp-216 is buried in a hydrophobic pocket in OspA-257 and is characterized by an emission maximum of 330 nm and a mean fluorescence lifetime of 1.12 ns in the native protein under physiological conditions (150 mM NaCl), while in the denatured protein (6 M gdn-HCl), the emission maximum shifts to 355 nm and the mean fluorescence lifetime increases to 1.67 ns, indicating that Trp,216 is com, pletely solvent,exposed (France, L.L., u n p u b l i s h e d data): We monitored both the emission m a ~ m u m and the fluorescence decay profile at increasing concentra' tions of NaCI from 0.0 to 1.0 M. The emission maximum showed no change, remaining constant at 330 nm. The mean lifetime increased only 6% from 1.11 ns (no NaCl) to 1.18 ns (1.0 M NaCl), and far less than the 49% increase resulting from denaturation by gdn-HCl. The individual fluorescence decay parameters and the mean lifetimes at each NaCl concentration are listed in Table II. The 6% increase in the mean lifetime evidently results from a similar increase in the lifetime of the fast component (~-j), perhaps indicating that the dominant component (~-~) arises from a ground state conformation in which Trp-216 is slightly more solvent-exposed. The fractional contribution, Oi, of each component to the total fluorescence yield, remains constant over the entire range of sodium chloride concentrations. Thus, the shielding effect at high ionic strength does not appear to alter significantly the hydrophobic environment of Trp-216 in the native protein.
65 360
!
i
!
!
i
E
-'"
c 350
E" E 345
//
:"
E 340
g 335 uJ
330
~-----&-T t
-~1~--~& I
20
-~"
I
I
Temperatuce.
I
I
I
1
°C
Fig. 4. Temperature-induced shifts in the steady-state fluorescence emission maximum of Trp-216 in OspA-257, excited at 295 nm. Measurements made as the temperature was increased, in the presence ( A ) and in the absence (zx) of 150 mM NaCI. Measurements made as the temperature was decreased, in the presence ( • ) and absence ( v ) t~f 150 mM NaCI.
Thermal denaturation and renaturation The emission maximum of a sample of OspA-257 was monitored as the temperature was increased in incremental stages from 23 to 80°C, and then decreased (incrementally) to 23°C, with 20 min incubation at each temperature (Fig. 4) before the spectrum was recorded. These experiments were repeated in the presence and in the absence of 150 mM NaCI, to compare denaturation and renaturation under physiological conditions (150 mM NaCI) with denaturation and renaturation under conditions used for CD experiments (no NaCl). The emission maximum under both conditions shifts from 330 nm at temperatures < 55°C, to 355 nm at temperatures > 74°C. As shown in Fig. 4, t h e midpoints of the wavelength shifts are roughly 59°C (150 mM NaCI) and 63°C (no NaCI). It appears that t h e protein is destabilized slightly by the presence of 150 mM NaCI. The midpoints for the wavelength shifts for the decreasing temperature curves are about 53°C (150 mM NaCI) and 55°C (no NaCl). Again, the presence of 150 mM NaCl appears to destabilize the pro-
tein slightly. The effect of ionic strength on protein stability depends on the type of electrostatic interactions which predominate in a given protein [29]. If classical electrostatic interactions (i.e., nonspecific charge repulsions) are important, then increasing the salt concentration will increase the shielding and stabilize the native protein. If specific intramolecular 'salt bridges' are important, then increasing salt concentrations (increased shielding) destabilizes the native protein, as observed for OspA-257. The renaturation curves are displaced 5 to 10 degrees towards lower temperatures compared to the corresponding denaturation data, indicating that the system was not in a completely reversible thermodynamic equilibrium. This could be due either to a tendency of the unfolded protein to form a structure that resists refolding or to the need for a longer holding period prior to recording the emission spectrum. As a corollary to the above experiment, eight independent protein samples were incubated at a selected temperature for 20 min (37, 46, 56, 65, 72, 81, 89 and 100°C), then returned to 23°C, and the emission maximum of each was measured. The fluorescence of a ninth sample that was maintained at 23°C for the same period was also recorded. The emission maximum for each sample was 330 nm, except for the protein incubated at 100°C, which showed an emission maximum of 332 nm. Thus, it appears that thermal denaturation is largely reversible, as monitored by the fluorescence emission of Trp-216. We measured the CD spectrum of the refolded OspA-257 after thermal denaturation (at 96°C) for different periods of time from zero to 60 min. Vials containing different protein samples were immersed in a thermostatted water bath for selected periods of time (zero to 60 min), after which each protein sample was placed on ice for > 60 min, and its CD spectrum was measured (at 23°C). We observed no significant change, within experimental uncertainty, in peak amplitudes (Ad194 nm), Ad213 nm)) nor in the crossing point (200 nm) for denaturation times _< 10 min at 96°C.
TABLE II
Tryptophan fluorescence parameters at different NaCI concentrations (from 0 to 1.0 M) [NaCI] (M)
¢1 (ns)
oq
@l
T2 (ns)
a2
@2
X~
(~') (ns)
0 0.0739 0.150 0.224 0.301 0.450 0.599 0.750 1.00
1.08 1.08 1.08 1.07 i.10 1.10 1.11 1.11 1.13
0.987 0.986 0.979 0.979 0.982 0.982 0.974 0.973 0.980
0.956 0.951 0.942 0.932 0.946 0.947 0.934 0.930 0.942
3.77 3.91 3.22 3.56 3.48 3.36 2.97 2.93 3.43
0.0129 0.0139 0.0203 0.0215 0.0178 0.0180 0.0256 0.0276 0.0200
0.0436 0.0486 0.0582 0.0681 0.0542 0.0530 0.0657 0.0697 0.0583
0.931 0.903 0.768 0.853 0.856 0.873 0.734 0.710 0.753
l. 11 I. 12 l. 12 i. 12 i. 14 I. 14 1.16 1.16 1.18
66 2C
•
i
•
i
,
i
'
i
,
r
,
i
.
,
,
[
•
I
'
J.
}
:
:
,
I 250
IC
~ oc -1C
•. . . "
-2C
-4i
•
I
. . . . . . . .
,
:
oo[ 2oE
-20 40 190
I l l 200 210
I I 220
Wovelencjt
l
I%
l l 230
240
nm
Fig. 5. Denaturation o f OspA-257. induced by increasing concentrations of guanidine-HCI (0.0 to 1.75 M), as probed by CD, U p p e r plot: 0,0 M (): 0.125 M ( ): 0.25 M. ( - - : - - ) : 0.50 M ( . . . . . . ); 0.75 M ( . . . . . . ). Center plot: 1.0 M ( ); 1.25 M ( - - - - - ) ; !.50 M ( . . . . ): 1.75 M ( . . . . . . ). Lower plot: 0.0 M ( ); 0.5 M ( - - - - - - ) : 0.75 M ( . . . . ): 1.0 M ( . . . . . . ); i.75 M ( . . . . . . ). The pH was maintained at 7 for these experiments.
Immersion for 20 min produced only a decrease of 40% in the positive peak (194 nm). However, after 30 min immersion, the CD spectrum showed large distortions, although the positions of the positive and negative peaks were similar to the native spectrum. After 60 min immersion, the CD spectrum consisted of a large negative peak at 200 nm (indicative of random coil ), a n d no characteristics of the native spectrum remained. The native conformation of OspA-257 thus appears to be energetically favorable as judged by the ability of the protein to refold after thermal denaturation at 96°C for < 10 rain. Furthermore, although the plots of emission maxima as a function of temperature (Fig. 4) suggest that intramolecular salt bridges play a role in stabilizing the native state, both steady-state and timeresolved fluorescence experiments show that the presence of < 1 M NaC! does not affect the native conformation enough to significantly alter the Trp-216 environment.
Denaturation by guanidine-HCl CD spectra were measured for OspA-257 at different concentrations of gdn-HCl from 0.0 to 1.75 M (Fig. 5). A superposition of five CD spectra, measured at 0.0, 0.125, 0.25, 0.50 and 0.75 M gdn-HCi, is shown in the upper plot. Spectra were truncated at 198 nm
because of the absorption of gdn-HCI at shorter wavelengths. The native conformation appears stable at [gdn-HCI] < 0.5 M. The 0.75 M gdn-HCI spectrum shows a large negative peak at 200 nm, indicating the presence of a significant fraction of random coil. In the center plot, the CD spectra measured at 1.00, 1.25, 1.50 and 1.75 M gdn-HC! are shown. An increase in the amplitude of the negative peak at about 200 nm from 1.00 M to 1.25 M, indicates that denaturation is almost 85% complete at 1.00 M gdn-HCI. No further spectral change takes place at [gdn-HCI] > 1.25 M. The transition from the native to the denatured state appears to be cooperative, i.e., denaturation occurs rapidly and suddenly, with little change occurring at gdn-HCI < 0.50 M or at gdn-HCl > 1.25 M. The lower plot in Fig. 5 demonstrates that CD spectra measured over the complete range of gdn-HCi concentrations (from zero to 1.75 M) pass through an isodichroic point at 208 nm. Although, for clarity, only five representative spectra are shown in the lower plot, all measured spectra pass through this point. isodichroic points in CD spectra have been predicted at ~ 204 nm for a helix-to-coil and at ~ 208 nm for a sheet-to-coil transition [34]. The isodichroic point observed for OspA-257 is thus consistent with that expected for a protein with low a-helical and high fl-sheet content (Table 1). The presence of an isodichroic point suggests that for the transition from the native to the denatured state, only two, spectraIiy distinct, major conformations are present (i.e., partially-folded intermediate conformations are not present in significant quantity), it follows that a CD spectrum measured at any intermediate concentration of denaturant will be a linear combination of the two basis CD spectra: the native state (0,0 M gdn-HCl), and the denatured state (1.75 M gdn-HCl). Thus, the fraction of native state present at any selected concentration M can be written: Ae(A, M)- AeD(A ) fN(M)=
ArN(,~)_ A ~ D ( A )
,
(4)
where fN is the fraction of the native state, AE(A, M ) is the observed differential extinction coefficient at wavelength A and denaturant concentration M, and eN(A) and Co(A) are the differential extinction coefficients at wavelength A for the native and denatured states, respectively. The cooperative nature of the unfolding transition for OspA-257 is illustrated in Fig. 6. The fraction fN was calculated at each concentration of gdn-HC! (from 0 to 1.75 M), using Eqn. 4 and values for A¢ measured at 201 nm. This wavelength was selected because the difference between the native and denatured CD spectra is large. As shown in Fig. 6, fN remains constant ( = 1.0) for M from 0.0 to 0.5. At 0.75 M and 1.0 M
67 100
- -
Acknowledgements
J
O80 f 06C Z 04C 02¢ ,O
OOC ooo
|
I
050
i
I
1.OO [Gdn
- HC[~,
i
!
150
i
2.00
M
Fig. 6. Fraction of native OspA-257 present at increasing concentrations of guanidine-HCl, as calculated from the CD spectra at 201 nm. .IN is calculated from Eqn. 4, using values for de measured at 201 nm for each concentration of gdn-HCI (0.0M to !.75 M).
gdn-HCl, fN is calculated to be 0.37, and 0.14, respectively. At [gdn-HCi] > 1.25 M, fN remained constant at = 0.0. Similar titration curves were obtained using CD values at 204, 210 and 218 nm (data not shown). The reversibility of denaturation by gdn-HC1 was tested, using steady-state fluorescence. A sample of OspA-257 was mixed with 1.25 M gdn-HCI, and its emission maximum was 355 nm, as expected for the denatured protein. The sample was then diluted 3-fold (to 0.42 M gdn-HCl), and its emission maximum was found to be 330 rim, characteristic of the native protein. Thus, both chemical and thermal denaturation appear to be completely reversible for OspA-257, under limiting conditions (< 1.25 M gdn-HCi and < 10 rain heating at 96°C, respectively). The significant stability of the native state of OspA257 with respect to pH and ionic strength, in addition to its ability to refold after denaturation by 1.25 M gdn-HCl or incubation at 96°C for _< 10 rain, presumably reflects structural features related to the protein's biological function. OspA is one of two major outer surface proteins which have been identified in Borrelia burgdorferi. These two proteins (OspA and OspB) have similar masses and show a large degree of sequence homology. Although these outer surface proteins may play an important role in the progression of Lyme disease, neither the specific function of the outer surface proteins nor the specific advantage of their stability for the survival of the spirochete are known. To our knowledge, this work is the first physical/ chemical structural analysis of an outer surface protein from the Borrelia genus. Experiments are in progress to investigate recombinant OspB, as well as the outer membrane proteins from related species of Borrelia. These experiments will help determine if the stability of native OspA is perhaps a general characteristic of the outer surface proteins of Borrelia, and may contribute to an understanding of their function and their ability to be recognized as immunogens.
We thank Ann Emrick, Denise Monteleone and John Trunk for assistance with the experiments and analysis of data, and Dr. Erwin London (State University of New York at Stony Brook) for the use of his Spex 212 fluorolog fluorometer. We also thank Dr. Curtis W. Johnson and his colleagues at Oregon State University for providing a copy of VARSELEC, the program for the analysis of protein secondary structure. Research supported by the Office of Health and Environmental Research and the Division of Energy Biosciences, Office of Basic Energy Sciences (JK and GH), United States Department of Energy (USDOE) and a grant from the National Institutes of Health (GM34662) to JCS. The National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory is supported by the Office of Chemical Research and The Office of Materials Research, USDOE. The circular dichroism spectrometer and fluorometer at port U9B of the NSLS are supported by the Office of Health and Environmental Research, USDOE.
References t Centers for Disease Control (1989) Morb. Mortal. Wkly. 38, 668. 2 Burgdorfer, W., Barbour, A.G., Hayes, S.F., Benach, J.L., Grunwaldt, E. and Davis, J.P. (1982) Science 216, 1317-1319. 3 Steere, A.C. (1989) N. Engl. J. Med. 321,586-596 4 Fikrig, E., Barthold, W.W., Kantor, F.S. and FlavelL R.A. (1990) Science 250, 553-556. 5 Persing. D.H., Telford III, S.R., Rys, P.N., Dodge, D.E., White, T.J., Malawista, S.E. and Spielman, A. (1990) Science 249. 14201422. 6 Bergstrom, S., Bundoc, V.G. and Barbour, A.G. (1989) Mol. Microbiol. 3, 479-486. 7 Dunn. J.J., Lade, B.A. and Barbour, A.G. (1990) Protein Expression and Purification 1, 159-160. 8 Johnson, C.W., Jr. (1990) Pro. Struct. Funct. Gen. 7, 4108-4116. 9 Hennessey, LP., Jr. and Johnson, W.C., Jr. (1981) Biochemistry 20, 1085-1094. l0 Manavalen, P. and Johnson, W.C., Jr. (1987)Anal. Biochem. 167, 76-85. 11 Eflink, M.R. and Ghiron, C.A. (1981)Anal. Biochem. 114, 119227. 12 Sutherland, J.C., Desmond, E.J. and Takacs, P.Z. (1980) Nucl. Instr. and Meth. 172, 195-199. 13 Sutherland, J.C., Keck, P.C., Griffin, K.P. and Takacs, P.Z. (1982) Nucl. Instr. and Meth. 195, 375-379. 14 Gray, D.M, Lane, D., Kuner, E., Vaughan, M. and Sutherland. J. (1984) Anal. Biochem. 136, 247-250. 15 Chen. G.C. and Y~ng, J.T. (1977)Analyt. Lett. 10. 1195-1207. 16 Chou, P.Y. and Fasman, G. (1974a) Biochemistry 13, 211-222. 17 Chou, P.Y. and Fasman, G. (1974b) Biochemistry 13, 222-245. 18 Chou, P.Y. and Fasman, G. (1978a) Adv. Enzymol. 47, 45-148. 19 Chou, P.Y. and Fasman, G. (1978b) Annu. Rev. Biochem. 47, 251-276. 20 Chou, P.Y. and Fasman, G. (1979) Biophys. J. 26, 367-384. 21 Gamier, J., Osguthorpe, D.J. and Robson, B. (1978) J. Mol. Biol. 120, 97-120. 22 Kyte, L and Doolittle, R.F. (1982)J. Mol. Biol. 157. 104-132.
68 23 Hopp. T.P. and Woods. K.R. (1981) Proe. Natl. Acad. Sci. USA 78, 3824-3828. 24 Janin, J.. Wodak, S.. Levitt, M. and Maigret, B. (1978) J. Mol. Biol. 125, 357-386. 25 Laws, W.R. and Sutherland. J.C. (1986) Photochem. Photobiol. 44, 343-348. 26 O'Connor. D.B. and Phillips. D. (1984) Time-Correlated Single Photon Counting, Academic Press. London. 27 Grinvald, A. and Steinberg, 1.Z. (1976) Bioehim. Biophys. Acta 427, 663-678. 28 Barlow, DJ. and Thornton, J.M. (1983) J. Mol. Biol. 168. 867-885.
29 Dill, K.A. (19911) Biochemistry 29, 7133-7155. 30 Urban. C, Rahal, J.J., Dattwyller, R.J., Gorcvic, P. and Luft, BJ. (1990) J. Bacteriol. 172, 6139-6141, 31 Duez, C.. Prion-Fraipont. C., Joris, B., Dusart, J., Urdea. M.S., Martial, J.A., Frere, J.M., and Ghuysen, J.M. (1987) Eur. J. Biochem. 162, 509-518. 32 Pace, C.N., Laurents. D.V. and Thomson, J.A. (1990) Biochemistry 29, 2564-2572. 33 Privalov, P.L (1979) Ads'. Protein Chem. 33, 167-241. 34 Greenfield, N. and Fasman, G.D. (1969) Biochemistry 8. 41fl84tl6.
