ARCHIVES

OF BIOCHEMISTRY

AND

BIOPHYSICS

Vol. 289, No. 1, August 15, pp. 53-61, 1991

The Structure of Human Acidic Fibroblast Factor and Its Interaction with Heparin

Growth

Robert A. Copeland,*,t Hanlee Ji,* Anne J. Halfpenny,?,’ Robert W. Williams,$ Wayne K. Herber, * Kenneth A. Thomas,* Mark W. Bruner,* James A. Ryan,* Dorothy Shigeko

Marquis-Omer,* Gautam Sanyal,* Robert Yamazaki,* and C. Russell Middaugh*>’

Karen C. Thompson,*

D. Sitrin,*

*Merck Sharp & Dohme Research Laboratories, P.O. Box 2000, Rahway, New Jersey 07065 and West Point, Pennsylvania TDepartment of Biochemistry & Molecular Biology, The University of Chicago, Cummings Life Science Center, 920 East 58th Street, Chicago, Illinois 60637; and $Department of Biochemistry, Uniformed Services University of Health Sciences, 4301 Jones Bridge Road, Bethesda, Maryland 20814.

Received January

19486;

23, 1991, and in revised form April 5, 1991

The secondary and tertiary structure of recombinant human acidic fibroblast growth factor (aFGF) has been characterized by a variety of spectroscopic methods. Native aFGF consists of ca. 55% &sheet, 20% turn, 10% ahelix, and 15% disordered polypeptide as determined by laser Raman, circular dichroism, and Fourier transform infrared spectroscopy; the experimentally determined secondary structure content is in agreement with that calculated by the semi-empirical methods of Chou and Fasman (Chou, P. Y., and Fasman, G. C., 1974, Biochemistry 13, 222-244) and Garnier et al. (Garnier, J. O., et al., 1978, J. Mol. Biol. 120,97-120). Using the Garnier et al. algorithm, the major secondary structure components of aFGF have been assigned to specific regions of the polypeptide chain. The fluorescence spectrum of native aFGF is unusual in that it is dominated by tyrosine fluorescence despite the presence of a tryptophan residue in the protein. However, tryptophan fluorescence is resolved upon excitation above 295 nm. The degree of tyrosine and tryptophan solvent exposure has been assessed by a combination of ultraviolet absorption, laser Raman, and fluorescence spectroscopy; the results suggest that seven of the eight tyrosine residues are solvent exposed while the single tryptophan is partially inaccessible to solvent in native aFGF, consistent with recent crystallographic data. Denaturation of aFGF by extremes of temperature or pH leads to spectroscopically distinct conformational states in which contributions of tyrosine and tryptophan to the fluorescence spectrum of the pro’ Current address: Genetics Institute, One Burrt Road, Andover, MA 01810. ’ To whom correspondence should be addressed at Merck Sharp & Dohme Research Laboratories, WP26-331, West Point, PA 19486. 0003.9861/91$3.00

Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

tein vary. The protein is unstable at physiological temperatures. Addition of heparin or other sulfated polysaccharides does not affect the spectroscopic characteristics of native aFGF. These polymers do, however, dramatically stabilize the native protein against thermal and acid denaturation as determined by differential scanning calorimetry, circular dichroism, and fluorescence spectroscopy. The interaction of aFGF with such polyanions may play a role in controlling the activity of this growth factor in vivo. o 1991 Academic POW+ IUC.

The heparin-binding family of growth factors induces the proliferation of a wide variety of mesodermal and neuroectodermal cells. The two best characterized members of this family are the acidic and basic fibroblast growth factors (aFGF and bFGF). Both of these proteins have been isolated from a number of mammalian sources and have recently been cloned and expressed in procaryotic hosts. Although aFGF and bFGF share significant sequence identity (l), the two proteins are distinguishable by their isoelectric points and the marked stimulation of aFGF (but not bFGF) mitogenic activity by heparin (2, 3). Heparin also appears to protect both proteins from inactivation by high temperature, acidic conditions, and proteolysis (4-8). The molecular basis of protection from denaturation and the stimulation of aFGF activity are unknown, although it has been suggested that heparin acts to stabilize the native conformations of the proteins (4). 3 Abbreviations used: aFGF, acidic fibroblast growth factor; bFGF, basic fibroblast growth factor; SDS/PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis. 53

54

COPELAND

Recently, aFGF has been shown to accelerate wound healing in animal models, suggesting a potential role for this protein as a therapeutic agent (3). There is thus great interest in understanding the structural and mechanistic basis for aFGF activity. The high-level expression of aFGF in Escherichia coli now permits isolation of sufficient quantities of this protein to begin addressing these issues. In the present report we describe the secondary and tertiary structure of native human aFGF (molecular weight -15,900) and its interaction with heparin and other polyanions. We have made use of spectroscopic probes of the single tryptophan (W107) and eight tyrosine (Y8, Y15, Y55, Y64, Y74, Y94, Y97, and Y125) residues of human aFGF to assess the structure of the native protein and to follow conformational transitions induced by thermal and acid denaturation. We have also determined the secondary structure of the native protein by a combination of spectroscopic (CD, FTIR, and laser Raman) measurements and semi-empirical calculations. The results of these studies are discussed in terms of the overall structure of the native protein and the possible effects of heparin-aFGF interactions on the in viva activity of this growth factor. MATERIALS

AND

METHODS

Recombinant human aFGF was isolated from transformed E. coli cells as previously described (9). Several alternative final isolation steps were employed for different batches of cells, all resulting in aFGF of similar purity (>99%), specific mitogenic activity, and spectroscopic properties; the details of these isolation methods will be described in a subsequent paper (10). Final sample purity was assessed by SDS/PAGE using silver staining. The biological activity of the purified protein was monitored as the mitogenic response of BALBc/3T3 cell lines and followed by uptake of tritiated thymidine (9). Protein concentration was determined spectrophotometrically employing an extinction coefficient of 18.64 mM-’ cm-’ at 277 nm based on the difference in absorbance of the eight tyrosine residues of aFGF at pH 7.4 and 13 (11). Sulfated polysaccharides were purchased from Sigma and used as received. All other reagents were the highest grades available. CD spectra were measured with either an AVIV 62 DS or Jasco 5500 spectropolarimeter. Samples of aFGF (2-5 pM in 0.05 M phosphate buffer, pH 6.5) were contained in 0.1 cm pathlength cells and the cell temperature was controlled by circulating a water/ethylene glycol mixture through the cell holder from a circulating water bath. Protein secondary structure was estimated from the CD spectra using the method of Yang et al. (12). Laser Raman spectra of aFGF were obtained with 514.5 nm excitation from an argon ion laser as previously described (13). The sample temperature at the laser focus was maintained at 14-15°C throughout the experiment. Spectra were obtained with an integration time of 2 s/cm-’ and a bandpass of 5 cm-‘. Four scans in the 800-1800 cm-’ region were collected and added together, and an additional 10 scans were collected in the amide I region (1500-1800 cm-‘) and summed for secondary structure analysis. The final spectra were baseline corrected, and the contributions from solvent and aromatic amino acids were subtracted by iteration with digitally stored model spectra. The resulting corrected spectra were then used to estimate the secondary structure of aFGF as described by Williams (13). Infrared spectra were measured with a Digilab FTSGO FTIR spectrometer at 2 cm-l resolution with a 0.1 mm cell. A protein concentration of 10 g/liter was used with D,O (0.15 M NaCI) as the solvent. Spectra were partially deconvoluted using parameters of K = 2 and c = 12

ET AL. and secondary structure was estimated by the method of Byler and Susi (14). Fluorescence spectra were obtained with either a Perkin-Elmer 65040, Spex Fluorolog-2 or an SLM-SOOOC spectrofluorometer. Bandpasses from 1-8 nm were employed and temperature was controlled as described above (vide supra). Sample absorptivities were usually maintained below 0.1 between 280 and 300 nm to minimize interferences from inner filter effects (15). Fluorescence lifetimes were measured by the multifrequency phasemodulation technique employing an ISS GREG fluorometer (16-18) equipped with a 300W Xenon-arc lamp. A cross-correlation frequency of 40 Hz was used and para-terphenyl was employed as the reference (T = 1.05 ns). A protein concentration of 1.0 g/liter was used in quartz cuvettes of pathlength 0.4 and 1 cm for absorption and emission, respectively. Data were analyzed by the method of least squares as a sum of exponentials (18). The degree of fluorescence quenching of aFGF by potassium iodide (KI) was determined by recording the 280-nm excited fluorescence spectrum of a 6.2~PM protein solution in 150 mM NaCl, 10 mM NaP04 (CM80 or RCMS), pH 7.4, after sequential addition of O-1000 ~1 of a 3 M KI solution containing 5 mM Na2S203 (in the same buffer). The fluorescence intensities were measured as peak heights at the band maximum and were corrected for the dilution effect of the added KI solution. In a similar manner, acrylamide quenching experiments were performed by titration of a 6.2~PM solution of aFGF with a 1 M acrylamide solution. These results were corrected for inner filter effects and dilution due to the added acrylamide (15). Neither iodide nor acrylamide at the highest concentrations examined induced a detectable shift in the fluorescence emission spectrum of aFGF. Solvent perturbation absorption spectra were recorded on a Beckman DU-70 uv-visible spectrophotometer using the method of Herskovitz and Laskowski (19). The perturbant used was 20% (v:v) ethylene glycol. Second derivative absorption spectra of native and denatured aFGF were also obtained with the Beckman spectrophotometer (vide supra). The number of exposed tyrosine residues was calculated as described by Ragone et al. (20) using the parameters A, B, and C for ethylene glycol from Table II of their study. Differential scanning calorimetry was performed with a Hart 7708 calorimeter. A scan rate of 600/h was employed using a protein concentration of 1 g/liter.

RESULTS

The tertiary structure of aFGF was examined by a variety of spectroscopic techniques. The growth factor contains eight tyrosines and a single tryptophan. The fluorescence emission spectrum of aFGF at room temperature (and below) is dominated by a single peak at ca. 306 nm (Fig. 1, solid line). A marked trailing region is observable in the 320-370 nm region, but no evidence of characteristic tryptophan emission is detected at these temperatures. The excitation spectrum of the 306-nm aFGF emission manifests bands of almost equal intensity at 277 and 282 nm (Fig. 2). When the excitation spectrum for the relatively weak 350-nm emission is monitored, the 282-nm band begins to predominate and a distinctive tryptophan shoulder begins to appear near 290 nm. As the temperature of the sample is raised, the ca. 306nm emission peak gradually shifts to 309 nm and decreases substantially in intensity; concomitantly, a second peak begins to appear at ca. 350 nm. This effect is illustrated by the dashed line in Fig. 1 which represents the fluorescence spectrum of aFGF at 60°C. At high temper-

HUMAN

ACIDIC

FIBROBLAST

GROWTH

55

FACTOR

L

x.=

250

260

270 Waveieng~h

i

o290

FIG. 1. phosphate pH 3.5 at 60°C (-

’ 310

330 350 Wavelength

I I



370 (nm)

390

200

290

300

(nm)

FIG. 2. Excitation spectrum of aFGF monitored with emission at 306 (----) and 350 (- - - - - -) nm. The solution conditions are as described in Fig. 1.

410

Fluorescence spectra of recombinant human aFGF in 0.05 M butfer, pH 6.5 at 25°C (----), in 0.05 M phosphate buffer, 25°C (- * -. * -. -1, and in 0.05 M phosphate buffer, pH 6.5 at - - -1. The samples (7-8 tiM) were excited at 280 nm.

atures, the excitation spectrum of the 350-nm emission is tryptophan-like (not ihustrated). These changes are irreversible. Removal of the high-temperature sample from the cuvette and subsequent scanning of the empty cuvette shows significant 350-nm fluorescence, indicating that the high temperature species is at least partially adsorbed to the interior euvette surface (CD measurements of aFGF also showed evidence of protein adsorption onto the cells; vide infra). When the aFGF solution pH is brought below 4, the fluorescence spectrum is again affected. As shown by the dot-dashed line in Fig. 1, at pH 3.5 aFGF gives rise to an emission spectrum with maxima at both 309 and 345 nm. In an attempt to directly detect tryptophan emission, aFGF at pH 7.2 and 6.3”C was selectively excited at increasing wavelengths (Fig. 3) employing narrow excitation and emission resolutions of 1 and 2 nm, respectively. Upon excitation at 290 nm, emission in the 330-370 nm region is markedly increased relative to the dominant 306-nm emission seen with excitation at lower wavelengths. A distinct shoulder becomes8 evident near 340 nm with 292 nm excitation, and typical tryptophan emission spectra with a peak at approximately 340 nm are produced when aFGF is excited 3295 nm (Fig. 3). Similar results were obtained for aFGF samples containing a threefold weight excess of heparin (data not shown). The excited state lifetime of aFGF was determined using excitation at 275 and 295 nm (Table I). When the sample was excited at 275 nm and emission at 306 nm was monitored, using an interference filter, the best-fit of the phase and modulation data was obtained for a twocomponent analysis. The major component contributed 67% to the total fluorescence intensity and had a lifetime

of 1.8 ns. The second component had a lifetime of 0.4 ns. Heparin, added in threefold weight excess, did not alter these results. When aFGF was excited at 295 nm and emission above 345 nm was monitored using a Schott WG345 filter, we obtained lifetimes of 3.3 and 0.2 ns for the major and minor components with contributions of 65 and 35%, respectively, in a two-component (best-fit) analysis. The average lifetimes for excitations at 275 and 295 nm were 1.4 and 2.3 ns, respectively. As a probe of the accessibility of aFGF aromatic side chains, the quenching of the intrinsic fluorescence of the protein by extrinsically added compounds was examined. Figure 4 shows a Stern-Volmer plot employing iodide and acrylamide as negatively charged and neutral quenching

da 300

320

340

360

Wavelength

380

400

420

(nm)

FIG. 3. Fluorescence spectra of aFGF for different excitation wavelengths as indicated on the figure. Excitation and emission slit widths were 1 and 2 nm, respectively. aFGF concentration was 1 mg/ml in 10 mM phosphate buffer at pH 7.2, containing 100 mM sodium chloride. The sample temperature was 63°C. The pathlength for light absorption was 4 mm for 280, 290, and 292 nm, and 10 mm for 295 and 297 nm. The fluorescence intensities for excitation at 296,292,295, and 297 nm were multiplied by the factors indicated on the figure in parentheses following the excitation wavelength.

56

COPELAND

agents, respectively. Acrylamide is a very effective quencher of aFGF fluorescence with a Stern-Volmer quenching constant (K,, = kq7,) of approximately 9 M-l. The corresponding & is 6.4 X 10’ M-‘s-’ when the average lifetime of 1.4 ns is used. Iodide is much less efficient with a value of KS, - 1 M-‘. However, this was not simply an ionic strength effect, as sodium chloride added up to a concentration of 1.2 M had a negligible effect on the emission maximum or intensity (k,, - 0.05 M-l). The reduction in fluorescence by iodide can be used to estimate the fractional exposure of tyrosine residues if we assume that the 306-nm band of aFGF arises primarily from tyrosine side chains. Lehrer (21, 22) has shown that plots of Fo/ F vs l/[Q] intercept the y-axis at a point that approximates l/fa where fa is the mole fraction of accessible fluorophore. As shown in the inset to Fig. 4, this procedure yields a value for fa of -0.82 which corresponds to 6.6 of aFGF’s eight tyrosine residues being solvent accessible in the native state of the protein. We have attempted to assess the solvent accessibility of the eight tyrosine residues of aFGF by a variety of other methods. Figure 5 shows the solvent perturbation difference spectrum obtained with 20% (v:v) ethylene glycol for native aFGF. This spectrum is similar to that seen for N-acetyl-L-tyrosine ethyl ester (19) and shows little, if any, contribution from tryptophan. The number of exposed tyrosine and tryptophan residues can be calculated from the following set of simultaneous equations (20): kg2

= aAeg2(Trp)

[II

+ bAc2g2(Tyr)

A~ZST = a&~ (TV) + bh87 (Tyr),

PI

where Atze7 and AtZg2 are observed extinction coefficient differences for the protein at 287 and 292 nm, a and b are the number of exposed tryptophan and tyrosine residues, respectively, per mole of protein and Acx(Trp) and

TABLE

ET AL. 31

//

00

4

6

6

l/p] 0

0

I 0.2

I

I 0.4

1

I 0.6

0

I 0.6

10

12

M '

I 1.0

I 1.2

IQ1M FIG. 4. Stern-Volmer plot for the quenching of aFGF (0, A) and iodide in (a, n ) in the presence (0, A) and of a 3~ excess of heparin. The inset shows a modified plot of the iodide data. Experiments were performed buffered saline, pH 7.4, at 10°C.

by acrylamide absence (A, n ) Stern-Volmer in phosphate-

Acx(Tyr) are the extinction coefficient differences for Nacetyl-L-tryptophan ethyl ester and N-acetyl-L-tyrosine ethyl ester at wavelengths X nm (values taken from Herskovitz and Sorensen) (23). Analysis of the aFGF spectrum in this manner indicates that 7.3 tyrosine and 0.3 tryptophan residues are solvent exposed in the native conformation of the protein. The fractional solvent exposure of Trp 107 estimated in this way agrees well with that predicted by Gimenez-Gallego et al. (24) using the method of Rose et al. (25). We have also estimated the number of exposed tyrosine residues of aFGF from the amplitude differences at 278-282 and 290-293 nm in the

I

Fluorescence Lifetimes of aFGF in the Absence and Presence h er

2

of Heparin”

71

72

Ligand

(nm)

Emission filterb

(ns)

(ns)

fi

f2

(ns)

XZd

None Heparin’ None

275 275 295

306BP13 306BP13 WG345

1.81 k 0.24 1.89 -t 0.19 3.33 -+ 0.12

0.43 f 0.20 0.47 t 0.13 0.23 f 0.05

0.67 f 0.12 0.64 i 0.09 0.65 + 0.02

0.33 0.36 0.35

1.35 1.38 2.25

5.25 2.68 1.69

7,” I’

a All measurements were made at 8°C and at pH 7.2. Modulation frequencies in the range of 30 to 180 MHz were used. Data were analyzed by a least-squares method. In each case the best fit was obtained for a 2-component (biexponential) analysis. b 306BP13 filter was centered at 306 nm and had a bandpass of 13 nm. WC345 was a 345 nm cut-off Schott glass filter. c 7.” (average lifetime) was calculated as T,, = rlfi + r2f2, where 7, and r2 were the lifetimes of the individual components whose fractional contributions to the total fluorescence intensity were fi and j.. ’ Frequency independent errors of 1.0” in phase angle and 0.01 in modulation were used. There was no effect of changing these windows on the values of T and f. e Heparin was added in threefold excess by weight over aFGF.

HUMAN

ACIDIC

FIBROBLAST

GROWTH

57

FACTOR

4 a

600 250

310 Wavelength

(nm)

FIG. 5. Solvent perturbation absorption spectrum of aFGF using 20% (xv) ethylene glycol as the perturbing solvent. The sample concentration was 7.3 pM in 0.05 M phosphate buffer, pH 6.5, at 25°C.

second derivative absorption spectrum of this protein following the method of Ragone et al. (20), and find that 92% (i.e., 7.4 of 8 residues,) of the tyrosines are solvent accessible (data not shown). Figure 6 shows the laser Raman spectrum of native aFGF. The pair of bands at 826 and 852 cm--l in the spectrum are due to a Fermi re;sonance doublet. This doublet arises from the interaction of a totally symmetric ring breathing mode, vI, (at ca. 850 cm-‘) with the first overtone of a degenerate ring mode vIGa (at ca. 416 cm-‘; 2 XV 16aN 830 cm-‘). The vIga fundamental is not Raman active; however, its first overtone contains a totally symmetric component which i:s Raman active and can effectively mix with u1 (26). Siamwiza et al. (27) have shown that the relative intensity of these bands (1850~830) is a sensitive indicator of the tyrosine environment within proteins. These workers define three tyrosyl environments: (A) buried residues which act as proton acceptors in strong hydrogen bonds and are characterized by I 850/830 = 2.50; (B) buried residues which act as proton donors in strong hydrogen bonds and are characterized by I 850/830 = 0.30; and (C) solvent-exposed residues which simultaneously act as proton acceptors and donors in hydrogen bonds and are char,acterized by 1850,830 = 1.25. The Fermi doublet bands of aFGF have an intensity ratio of 1.47, indicating that the majority of the tyrosyl I 852,828 residues are solvent exposed. While no unique ratio of A, B, and C type tyrosyl residues can be determined from Fig. 6, one combination which fits the Raman data and is consistent with the other spectroscopic data is 6.6 solvent exposed residues (type C) and 1.4 buried residues involved as proton acceptors in hydrogen bonds (type A). The inference that the majority of tyrosine residues in aFGF are solvent exposed is also supported by pH titration of the protein. To avoid interference from any pH

1000

1200 Frequency

1400 shift

1600 km-')

1600

FIG. 6. Uncorrected laser Raman spectrum of aFGF in the BOO-1800 cm-i region, illustrating the location of the amide I and III vibrational band envelopes. Data were obtained at a protein concentration of 2.5 mM in 0.05 M sodium phosphate, pH 6.5, at 15’C.

induced absorption changes from Trp 107, the difference spectra between aFGF at pH 7.0 and the experimental pH values were measured. The extent of tyrosine ionization was quantified from the amplitude difference (A A A) between 275 and 295 nm, the minimum and maximum in the difference spectrum of tyrosinate minus tyrosine. A least squares fit of these data to the Henderson-Hasselbach equation yielded a pK, value of 10.2 -+ 0.2. This pK, value agrees well with that expected for solvent exposed tyrosine (-10) (28). The far uv circular dichroism spectrum of aFGF is shown in Fig. 7. The spectrum is characterized by positive

, 200

210

, / 240 220 230 Wavelength (nm)

I 250

FIG. 7. Far uv circular dichroism of human aFGF. Spectra are shown at a protein concentration of 5 pM in 0.05 M sodium phosphate, pH 6.5 at 15°C (-) and 70°C (------).

58

COPELAND

ET AL.

FIG. 9. Two-dimensional based on the computational P-sheet; cylinder, a-helix; structure.

Wave Numbers FIG. 8. Curve fitting of the amide I region of aFGF in D,O (10 g/ liter) using a mixed Gaussian/Lorentzian function after Fourier selfdeconvolution. The relative areas of the peaks are 1618 cm-.‘; 1.2% (14%), 1628 cm-‘; 18.5% (79%), 1637 cm-‘; 28.9% (64%), 1646 cm-‘; 23.1% (42%), 1653 cm-‘; 11.8% (17%), 1663 cm-‘; 8.0% (0.4%), 1669 cm-‘; 3.7% (4%), 1676 cm-‘; 2.8% (11%) and 1685 cm-r; 2.0% (96%). The values in parentheses indicate the percentage of the Gaussian function used for each peak. Secondary structure content was estimated from the relative areas of the fitted bands assuming the following assignments: LY(1653 cm-‘), fl (1628, 1637 and 1676 cm-r), random (1646 cm-‘), and turns (1663, 1669, 1685 cm-‘) (14).

ellipticity at 226-228 nm and an intense negative band at 205 nm. Analysis of the spectrum by the method of Yang et al. (12) suggests a secondary structure content of approximately 60% P-sheet, 15% turn, 20% random structure, and less than 5% a-helix. Analysis of the Raman amide I and III vibrational band envelopes (Fig. 6) by the method of Williams (13) yields values of 50-56% P-sheet, 17-21% turns, 9-14% disordered structure, and 15-18% o-helix. The deconvoluted infrared spectrum of aFGF in the amide I region is dominated by bands at 1637, 1646, and 1663 cm-’ (Fig. 8). Analysis of secondary structure by the method of Byler and Susi (14) yields values of 51% (/I), 14% (T), 23% (R) and 12% (cy), consistent with the Raman results. Analysis of the amino acid sequence of aFGF by the computational methods of Garnier et al. (29) and Chou and Fasman (30) produces secondary structure estimates of 52% P-sheet, 28% turn, 9% disordered structure, and 11% a-helix, in reasonable agreement with the spectroscopic results. The good agreement between the various secondary structure estimates obtained in this study provides compelling evidence that the native structure of aFGF is predominately P-sheet. Using the Garnier et al. (29) algorithm, we have assigned P-sheet structure to specific regions of the aFGF polypeptide; these calculations result in the two-dimensional model of aFGF secondary structure which is illustrated in Fig. 9.

secondary structure model for human aFGF method of Garnier et al. (26). Legend: arrows, U-tubes, p-turns; straight lines, disordered

One of the more distinguishing features of aFGF compared to bFGF is its dependence on the presence of heparin for full biological activity. We, therefore, examined the effects of heparin on the conformation and stability of aFGF. At room temperature and physiological pH, addition of heparin to native aFGF had no effect on the spectroscopic properties of the protein as measured by CD, fluorescence, and FTIR (after subtraction of the spectral contribution of heparin) or the excited state lifetimes shown in Table I. The effects of temperature on the fluorescence emission at 350 nm and on the CD ellipticity maximum at 227 nm are shown in Figs. 10B and A, respectively. Similar changes are seen when decreases in fluorescence at 306 nm and increases in ellipticity at 205 nm are plotted. The midpoint of the thermally induced aFGF structural alterations (T,) is seen at ca. 35 and 42°C (k2’C) for the fluorescence and CD changes, respectively. Addition of a threefold weight excess of heparin shifts this midpoint to 55°C (fluorescence) and 60°C (CD).

0

20

40

60

80 0 20 40 Temperature (“C)

60

80

100

FIG. 10. Effect of temperature on the circular dichroism ellipticity maximum at 227 nm (A) and the fluorescence intensity at 350 nm of aFGF in the absence of sulfated polysaccharide (0) and in the presence of a threefold weight excess of heparin (O), pentosan polysulfate (M), and chondroitin sulfate (A). All experiments were performed in 0.05 M sodium phosphate, pH 6.5 at a protein concentration of 5 PM. Only representative data points are shown for clarity.

HUMAN

ACIDIC FIBROBLAST

These changes are independent of protein concentration over the l-40 PM range examined. Titration of aFGF with heparin revealed that 50% of the temperature shift was produced by a weight ratio of 0.1-0.2 heparin to aFGF with no further increases in T, at heparin ratios above 3. Pentosan polysulfate and dextran sulfate produced an effect similar to that of hep,arin while chondroitin sulfate had no effect on the thermal stability of the protein. Similar results were obtained u;sing differential scanning calorimetry to monitor the thermal stability of aFGF. As shown in Fig. 11, an unfolding transition at ca. 42°C is seen for aFGF in the absence of heparin. As heparin is added, this endotherm is shifted to higher temperature, with the maximum effect observed at a threefold weight excess of heparin. The effect of heparin on the low pH fluorescence transition of aFGF was also investigated. At pH 7.0 the fluorescence intensity ratio for aFGF at 3501 308 nm is ca. 0.25, with or without heparin present. At pH 3.5 this intensity ratio increases to -1.0 in the absence of heparin (Fig. l), but remains -0.3 when heparin is present, suggesting that heparin also protects aFGF from acid denaturation. DISCUSSION

Secondary Structure As has been determined for many small growth factors, the structure of aFGF appears to be composed mainly of P-sheet lacking significant (r-helix content. The CD spectra of the related proteins bFGF (31, 32), interleukin 16 (33), and epidermal growth factor (34, 35) are similar to that of aFGF, suggesting a related secondary structure motif for these proteins. Although epidermal growth factor displays no sequence homology with aFGF, interleukin lc~ and p as well as bFGF have 19,25, and 55% identical residues, respectively (24). Furthermore, the Amide I FTIR spectrum of bFGF is very similar to aFGF (36). This is also consistent with the recent X-ray crystal structures for interleukin 1.~ and lfi (37, 38) which indicate that both proteins co:ntain a P-barrel comprised of 55-60% of the polypeptides. Finally, a recent low resolution X-ray crystallographic study of aFGF and bFGF indicate that both proteins contain 12 antiparallel pstands accounting for 50-60% of the structure (39) in agreement with these solution spectroscopic results. Tyrosine and Tryptophan

Environments

The fluorescence spectra of most tryptophan-containing proteins are dominated by emission from the indole ring system, even for proteins with high tyrosine content (15). In contrast, the fluorescence spectrum of aFGF reflects mainly tyrosine-like fluorescence, with a maximum at ca. 306 nm. Two simple nonmutually exclusive hypotheses can be advanced to explain this observation: (i) the emission

59

GROWTH FACTOR

a

(051) (3.O:l)

I

I

I

I

I

40

50

60

70

Temperature

I

(“C)

FIG. 11. Effect of heparin concentration on the DSC endotherm of aFGF. Experiments were performed at a protein concentration of 1 g/ liter with the molar ratio of heparin to aFGF varied from (a) 0, (b) 0.2: 1, (c) 0.3:1, (d) 0.5:1, to (e) 3.0~1.

maximum of the tryptophan residue may be shifted into the envelope of the tyrosine peak and (ii) the tryptophan emission could be dramatically quenched relative to tyrosine emission but positioned at a more usual energy (320-355 nm). We favor the latter explanation for a variety of reasons. Selective excitation of aFGF at increasingly longer wavelength reveals the presence of characteristic tryptophan-like emission near 340 nm suggesting a partially buried tryptophan. The excited state lifetime of this emission is significantly longer than that observed at shorter wavelength. Since the lifetime of tryptophan residues in proteins (2-8 ns) is usually significantly longer than that for tyrosine side-chains (l-2 ns) (40), this observation is consistent with the 340-nm emission arising from the single tryptophan residue in aFGF. Although different emission bands of a single fluorophore can decay with different lifetimes, tyrosine emission above 340 nm is very weak especially for excitation at 295 nm. Our fluorescence spectra and lifetime data for excitation at and above 295 nm thus suggest that tryptophan is the origin of this long wavelength band of aFGF fluorescence. Two other sources of the 340-nm emission can be postulated. Fluorescence in the region could be due to emission from tyrosinate moieties. There is no evidence from spectrophotometric pH titrations, however, of the requisite abnormally titrating tyrosine. This at least suggests that a ground state tyrosinate residue is not responsible for the fluorescence. Furthermore, fluorescence from excited state ionization of tyrosines is extremely rare in proteins and in many cases, its postulated presence was probably due to tryptophan contamination (40). A small amount of unfolded or partially unfolded aFGF could also account for the presence of 340 nm fluorescence, since we observed that the unfolding of aFGF by temperature, pH,

60

COPELAND

and denaturing agents results in fluorescence in this spectral region. This explanation seems unlikely as well since the presence of heparin, which dramatically stabilizes the native state, has no effect on the 340-nm peak at low temperature. Thus, unlike azurin where 306-nm emission akin to that seen with aFGF has been attributed to a strikingly blue-shifted tryptophan (41, 42), the unusual fluorescence of this growth factor can be most simply explained by a quenching of the single tryptophan relative to the usual tyrosine emission. We presume that this weak Trp emission is simply hidden in the extended spectral tail at X > 330 nm when excitation at wavelengths < 290 nm is employed. Interestingly, bFGF displays a fluorescence spectrum which is virtually identical to that of aFGF. The fluorescence from the single tryptophan of interleukin I/3 is also significantly quenched at pH values above 7.0 (30), suggesting further structural similarities in the indole environment of these related proteins. In this regard, the single tryptophan residue of aFGF is contained within one of the few tetrapeptide sequences that is well conserved between the fibroblast growth factors and interleukin lfl (39). Surprisingly, this region of the molecule may be on the surface of the protein since it has been implicated in both receptor and heparin binding (39). Inspection of the crystal structure of both acidic and basic FGF finds two basic residues (His 102 and Lys 105) near the partially exposed indole ring. These residues are conserved between the two structures. It is possible that the positive charges on these sidechains either stabilize the ground state or destabilize the excited state dipoles resulting in quenching of fluorescence emission. In contrast to the present results, Jaye et al. (43) reported a fluorescence spectrum for human aFGF with a band maximum at 345 nm. The band was quenched and shifted to lower wavelength in the presence of heparin. One possible explanation for this discrepancy is that the aFGF isolated by Jaye et al. was conformationally altered, with induction of structure more similar to the heparinassociated aFGF employed in these studies. The results of our estimates of tyrosine exposure appear to converge on values of seven solvent exposed residues and one inaccessible tyrosine for native aFGF. The crystal structure of aFGF (39) shows only Tyr 97 extensively internalized suggesting this residue as the most likely candidate for the buried tyrosine. Interactions

with Heparin

Heparin has a dramatic stabilizing effect on the native structure of aFGF. In the absence of heparin, native aFGF is relatively unstable at physiological temperature (35 45”(Z), undergoing large structural changes and consequent loss of mitogenic activity. Addition of a threefold weight excess of this sulfated polyanion, however, increases the melting temperature of the protein by as much as 20°C. Likewise, heparin appears to stabilize native

ET AL.

aFGF against denaturation by low pH or structure-disrupting solutes. Recent evidence suggests that in vivo aFGF exists primarily as a complex with extracellular polyanions (44-46) and should thus be present naturally in a stable form. Enzymatic release of aFGF from the complex might then produce a relatively unstable protein, thus providing a mechanism for rapid termination of its biological activity. The precise nature of the interaction of aFGF with heparin remains to be elucidated. Human aFGF contains several regions with three or more tightly clustered basic residues (residues 9-12, 35-41, 100-105, 112-119, and 122-128) which are potential sites of heparin binding (47). In bovine aFGF, Lys 119 has been implicated in heparin binding on the basis of chemical modification studies (48). Examination of the crystal structure of aFGF suggests that residues 105-128, which contain seven basic residues, are a likely site of heparin interaction, although the region from amino acids 25-69 cannot be excluded from involvement (39). The molecular weight heterogeneity for the heparin used in the present study precludes an accurate estimate of the stoichiometry of interaction between aFGF and heparin. Nevertheless, titration of the thermal stabilizing effect of low molecular weight heparin (4000-6000 Da) and pentosan polysulfate (1500-5000 Da) indicates that saturation occurs at low molar ratios of polyanion to protein. It, therefore, appears that at most a few heparin molecules bind per aFGF. Binding of heparin to aFGF does not appear to induce significant structural changes in the growth factor. Rather, the polyanions appear to exert their stabilizing effect by preferential binding to the native conformation of the protein. This complex-induced stabilization of the native conformation of aFGF is not unique to heparin. A number of sulfated polysaccharides show a similar effect, although aFGF does display some specificity in its interactions with these molecules. Thus, while heparin, pentosan polysulfate, and dextran sulfate all achieve similar levels of stabilization, chondroitin sulfate has no effect on the thermal stability of aFGF. Furthermore, the sidechains of Lys 112, Lys 118 and Arg 122 appear to be sites of anion (IrCli-) binding in a heavy atom derivative of aFGF employed in crystallographic analysis (39). This low specificity may allow aFGF to complex with a variety of sulfated molecules in vivo, thus providing alternative means of stabilizing the growth factor under normal cellular conditions. ACKNOWLEDGMENTS We thank Donald Hupe for his encouragement and the use of his fluorometer. Preliminary absorption spectra were taken in Dr. T. Y. Lin’s laboratory. Preliminary Raman and CD spectra were obtained in the laboratories of Professors Thomas G. Spiro and Franklyn G. Prendergast. The contribution of R.W.W. was funded in part by the Strategic Defense Initiative Organization Medical Free Electron Laser Program at USUHS.

HUMAN

ACIDIC

FIBROBLAST

GROWTH

61

FACTOR

J. L. (1982) J. Raman Spectrosc. 13, 21-24.

REFERENCES

26. McHale,

1. Burgess, W. H., and Maciag, T. (1989) Annu. Reu. Biochem. 58, 575-606. 2. Gimenez-Gallego, G., Conn, G., Hatcher, V. B., and Thomas, K. A. (1986) Biochem. Biophys. Res. Commun. 138, 611-617. 3. Thomas, K. A. (1987) FASEB J. 1,434-440. 4. Gospodarowicz, D., and Cheng, J. (1986) J. Cell Physiol. 128,475484. 5. Rosenbart, T. K., Johnson, W. V., Friesel, R., Clark, R., and Maciag, T. (1988) Biochem. Biophys. Zi!es. Commun. 152, 432-440. 6. Sakesla, O., Moscatelli, D., Sommer, A., and R&in, D. B. (1988) J. CellBiol. 107,743-751. 7. Mueller, S. N., Thomas, K. A., DeSalvo, J., and Levine, E. M. (1989) J. CellPhysiol. 140, 439-448. 8. Sommer, A., and Rifkin, D. El. (1989) J. Cell. Physiol. 138, 215220. 9. Linemeyer, D. L., Kelly, L. J., Menke, J. G., Gimenez-Gallego, G., DeSalvo, J., and Thomas, K. .4. (1987) Biotechnology 5, 960-965. 10. Yamazaki et al. (in preparation). 11. Demchenko, A. P. (1986) Ultraviolet Spectroscopy of Proteins, Springer-Verlag, Berlin. 12. Yang, J. T., Wu, C.-S.C., and Martinez, H. M. (1986) in Methods in Enzymology, (Hirs, C. H. W., and Timasheff, S. N., Eds.), Vol. 130, pp. 208-296, Academic Press, San Diego. 13. Williams, R. W. (1986) in Methods in Enzymology, (Hirs, C. H. W., and Timasheff, S. N., Eds.), Vol. 130, pp. 311-331, Academic Press, San Diego. 14. Byler, D. M., and Susi, H. (19186) Biopolymers 25, 469-487. 15. Lakowicz, J. R. (1983) Principles of Fluorescence Spectroscopy, pp. 341-381, Plenum, New York. 16. Gratton, E., Jameson, D. M.: and Hall, R. D. (1984) Annu. Reu. Biophys. Bioeng. 13,105-124. 17. Alcala, J. R., Gratton, E., and

The structure of human acidic fibroblast growth factor and its interaction with heparin.

The secondary and tertiary structure of recombinant human acidic fibroblast growth factor (aFGF) has been characterized by a variety of spectroscopic ...
1MB Sizes 0 Downloads 0 Views