J. Mol. BioE. (1992) 227, 1182-1191

Domain Structure and Domain-Domain Interactions in the Carboxy-terminal Heparin Binding Region of Fibronect:in Valery Novokhatny’, Frederick Schwarz2, Donald Atha and Kenneth Ingham’? ‘Holland Laboratory, American Red Cross 15601 Crabbs Branch Way, Rockwille, MD 20855, U.S.A. YYenter for Advanced Research in Biotechnology Rockville, MD 20855, U.S.A. 3Biotechnology

National Division, Gaithersburg,

(Received I6 April

Institute of Standards MD 20899, U.S.A.

and Technology

1992; accepted 26 June 1992)

The domain structures and stabilities of fragments isolated from the so-called ‘hen 2’ region of plasma fibronectin have been investigated by differential scanning calorimetry (DSC) and fluorescence spectroscopy. The 30 kDa hep-2A fragment contains three type III modules (III,, to III,,), whereas the 40 kDa hep-2B fragment contains four such modules (IIT,, to III,,). Melting of these fragments at neutral pH was irreversible and accompanied by rapid aggregation. In contrast, melting was completely reversible in 50 mnil-glycine at pM 2.7, where DSC measurements revealed the presence of three independently folded domains in 30kDa hep-2A and four in 40 kDa hep-2B. That each domain represented a single module was confirmed by measurements with four single-module subfragments, all of which melted reversibly, even at neutral pH. At neutral pH in the presence of 6 M-urea, 30 kDa hep-2A could be resolved by melted reversibly in a sharp peak from which only two transitions deconvolution. Only the larger of these was stabilized by heparin and was assigned to Upon isolation, module III,, melted at lower temperature than in modules III,, and III,,. the parent fragment where it is stabilized through an interaction with module III,,. We of fibronectin constitute conclude that all type III modules in the hep-2 region and III,, form a highly co-operative structure independently folded domains. Modules III,, through functionally significant interactions that can be disrupted with acid or su%cient concentrations of urea or guanidinium chloride.

Keywords: fibronectin;

heparin;

denaturation;

I. Introduction

7 Author to whom all correspondence should be addressed. 1182 ~OS.OO/O

type

III

domain

heparin. The interaction of fibronectin with heparin is dominated by the so-called hep-2 domains, located in the C-terminal third of each polypeptide chain (Benecky et al., 1988; Ingham el (xl., 1990). Functionally active fragments containing these domains can be isolated from proteolytic digests of the parent molecule. The 30 kDa hep-%A fragment contains three type III modules, IIIIZm14. 1-n the parent protein, these are followed by a variably spliced V region that occurs only in the heavy A chain of plasma fibronectin and which is removed by proteolysis. This V region is absent in the light B chain, such that the corresponding 40 kDa hep-2B domain contains four type III modules, III,,-r5. Although these fragments are loosely referred to as domains, primarily because of their resistance to proteolysis, their folding properties have not been

Fibronectin is a large glycoprotein that occurs on cell surfaces, in the connective tissue matrix and in extracellular fluids (Hynes, 1990). In plasma, it is a dimer consisting of very similar subunits of molecular weight 250,000 daltons, held together near their C termini by disulfide bonds. Each subunit is composed of 29 or 30 motifs or modules, of three different types, each of which is also found in other proteins. These modules are gathered in groups to form functional domains, which are specialized for binding to cellular receptors and other macromolecules such as fibrin, collagen and

0022%2836/92/201182-10

calorimetry;

0 1992 Academic Press Limited

Domain

Structure

characterized. For example, it is not known whether each type III module contributes an independently folded unit linked to its neighbors by short connecting strands or if these modules interact with each other to form larger blocks that would fold and unfold together in a co-operative all-or-none fashion. Differential scanning calorimetry (DSCT) affords a powerful approach for investigating this question (Privalov, 1979, 1982; Privalov & Potekhin, 1986; Sturtevant, 1987). DSC measurements with whole fibronectin are difficult to interpret because of the presence of multiple overlapping transitions too numerous to resolve (Wallace et al., 1981; DeCarreira and 1985; Niedzwiadek et aE., 1988). Spectral Castellino, measurements with a series of fragments spanning the entire molecule confirmed the existence of at least three separate regions of the molecule melting between 60°C and 70°C (Ingham et al., 1984). Recent DSC measurements with a more extensive collection of fragments indicate that fibronectin contains at least 12 independent structural domains in each polypeptide chain (Tatunashvili et al., 1990). This study emphasized the need for further investigation of smaller fragments to fully understand the domain structure of this important molecule in the context of its modular composition. Our laboratory thus initiated an effort to elucidate the domain structure of, and domain-domain interactions in, fibronectin through detailed studies of ever smaller fragments. Studies of the 42 kDa fragment from the gelatin binding region revealed strong interactions between some of the six constituent type I and type II modules, all of which were shown to be independently folded (Litvinovich et al., 1991) The type 1 and II modules are each encoded by a single exon, whereas the type III modules tend to be encoded by two exons (Hynes, 1990). An easily isolated fragment containing a single type III module, module was recently shown to contain a single III,, independently folded domain flanked by unstructured regions (Litvinovich et al., 1992). A recent n.m.r. study of the secondary structure of module III10 expressed in yeast indicates that this module should also be independently folded (Baron et al., 1992). The folding status of type III modules from other regions of fibronectin or from other proteins has not been established. In the present study, DSC and fluorescence methods have been used to investigate the isolated hep-2 fragments and subfragments obtained from them. It is shown that type III modules 12 to 15 all constitute independently folded domains. at least some of which interact at neutral PH.

2. Materials

and Methods

Porcine mucosal heparin, tris[hydroxymethyl]aminomethane (TRIS) and urea were obtained from Sigma t Abbreviations used: DSC, differential scanning calorimetry; n.m.r., nuclear magnetic resonance;

SDS/PAGE, electrophoresis.

sodium dodecyl sulfate/polyacrylamide

gel

qf Fibronectin

IP83

Chemical Co., St Louis, MO$. Ultrapure GdmCl was purchased from USB. All other chemicals were of reagent grade or better. Fibronectin was purified from freshfrozen human plasma or from a byproduct of coagulation factor VIII by affinity chromatography on gelatinSepharose according to method A of Miekka et al. (1982). The 30 and 40 kDa hep-2 fragments were generated with thermolysin and isolated by affinity chromatography on gelatinand heparin-Sepharose and hydroxy-apatite chromatography according to Borsi et al. (1986) as previously described (Ingham et al., 1990). Subfragments were prepared by digestion of the parent fragments with pepsin, trypsin or Lys-C followed by affinity chromatography on heparin-Sepharose and further purification by ion-exchange and/or size-exclusion chromatography. Details of the preparation and characterization of subfragments are described in a separate publication (Ingham et nl., 1992). All fragments were more than 95% pure as judged by SDS/PAGE, exclusion chromatography on Superose-12 or Superdex-75 (Pharmacia) and amino terminal sequence analysis. The hep-2A fragment is comprised of modales III,,, III,, and III,, and was assumed to have a carboxy terminus at Gln1875 (Pande et al., 1987) based on the numbering system of Skorstengaard et al. (1986). hep-2B contains an additional type III unit, III,,, and was assumed to terminate at Gly2075. The subfragments used in this study are as follows: 8 kDa III,,, N terminus at Ala1597; 14 kDa III,,, N terminus at Glu1691; 12 kDa III,,, Iv terminus at Ser17’71; 8 kDa III,,, h’ terminus at Ala1994; 23 kDa IIIi3-r4, P; terminus at Asp1676; 20 kDa ?u’ terminus at Ser1771. The molecular weights III 14-15, given here are estimated from the relative mobility in SDS/PAGE (Ingham et al., 1992). Protein concentrations are based on molar extinction coefficients published elsewhere (Ingham et al., 1992). DSC measurements were made with a DAXlVI-4N instrument (KPO Biopribor, Puschino, Russia) at a heating rate of 1 deg.C/min. Protein concentrations varied from 0.3 mg/ml to 1.1 mg/ml in the absence of denaturants and 1.6 mg/ml to 2.0 mg/ml in the presence of urea or GdmCl. The fragments were dialyzed extensively against the given solvent. The heating curves were corrected for an instrumental baseline obtained by heating the dialysate. Melting temperatures, calorimetric and van’t Hoff enthalpies were determined using software provided by Microcal. Excess heat capacity curves were generated from the experimental curves by the straight line method or by the method of splines, using software provided by Microcal. Deconvolution of the excess heat capacity functions was performed by the recurrent procedure of Friere & Biltonen (1978), also with software from Microcal. The relative error of t’he total enthalpy values is estimated at + 10% and that of the t, values at &- 0.2 deg.C. The corresponding errors in the parameters of individual transitions obtained by deconvolution of curves representing multiple domains is estimated at +15% and Fl deg.C. Fluorescence monitoring of thermal transitions was accomplished with an SLM 8000C fluorometer controlled with a personal computer. Temperature was controlled with a Keslab programmable water bath and monitored by the computer via an Omega model 670 thermocouple thermometer. Two emission monochromators were $ In no case does the identification of commercial instruments or materials imply a recommendation or endorsement by the National Institute of Standards and Technology.

N. Novokhatny

%184

.5 7 _____________ ---__--.-__----’ t 30 40 50 60 70 80 90 loo Temperature (‘C) Figure 1. At physiological pH, melting of fibronectin hep-2 fragments is irreversible and accompanied by a precipitation. Melting of 30 kDa hep-2A fragment as monitored by DSC (panel A) and by the intrinsic fluorescence ratio (panel B, continuous line) and 90” light scattering (panel B; broken line) in 0.02 M-Tris, 0.15 M-N&I, (pH 7.4; TBS). Samples were heated at 1 deg.C/min. Note that the turbidity occurs immediately, on the same timescale as the unfolding, indicating that the unfolding process is not at equilibrium under these conditions.

employed to monitor changes in the ratio of fluorescence intensities at 350 nm and 320 nm during unfolding. Changes in this parameter provide a sensitive method for detecting melting transitions and assessing the degree of reversibility. However, to the extent that quantum yields of the native and denatured states are different, this parameter is not linearly related to the degree of unfolding and the t, values obtained with it may be less accurate than those obtained by calorimetry. Light scattering measurements were made concurrently by driving the excitation monochromator back and forth between 280 nm; to obtain the fluorescence ratio, and 360 nm for measurement of the scattering intensity at 350 nm; the offset of 10 nm was to keep the scattering signal on scale and avoid overloading the photomultiplier. Samples were approx. 61 mg/ml in TBS (0.02 ivr-tris(hydroxymethy1) 915 M-NaC1, containing aminomethane (PH 7.4) 50 mu-citrate or glycine buffers at reduced pH, and 50 rnxr-phosphate containing different amounts of urea or GdmCE.

3. Results (a) Irreversible

denaturation

at neutral

Preliminary DSC measurements with 40 kDa hep-2 fragments in TBS at pH sharp irreversible endotherms with ratios near 0.5. An example is shown in

pH

both 30 and 7.4 yielded AH,,,/AH,, Figure 1 (a),

et al.

where curve 1 represents the first heating and curve 2 the second heating for the same sample. The measured specific enthalpy under these conditions was much lower than expected for globular proteins of this size. Examination of the samples after cooling indicated that the protein had formed a white precipitate, suggesting that exlothermic effects of aggregation could have diminished the observed heat effect. This would be expected if aggregation of the denatured protein occurred rapidly, on a time-scale comparable to the rate of heating. The fluorescence spectrum of both fragments undergoes a large shift in maximum wavelength from approximately 325 nm to above 350 nm upon denaturation. When heated at Ii deg.C/min in TBS at pH 7.4, the ratio of fluorescence intensity at 350 nm to that at 320 nm underwent a, sharp transition with a midpoint between 70°C and 75°C (Fig. l(b), continuous line). When the right-angle light scattering was monitored simultaneously with the fluorescence ratio, the appearance of turbidity was found to be coincident with the unfolding process as monitored by the change in fluorescence (Fig. l(b), broken curve). The transitions were absolutely irreversible under these conditions, as well as in several other buffers at this pH. Furthermore, the midpoint of the transition was quite sensitive to the heating rate, shifting downward by approximately 2 deg.C when he,ated at 10 deg.C/h (not shown). Thus, it was not possible to obtain equilibrium data with these fragments under physiological conditions. We therefore conducted further spectroscopic measurements in an effort to determine optimum conditions for calorimetric investigations. (b) Reversible

unfolding

at low pH

The effects of pB on the melting of 40 kDa hep-2B, as monitored by fluorescence ratio is shown in Figure 2. The ratio itself is not significantly affected by pH; the curves in Figure 2 have been arbitrarily adjusted on the vertical scale to fa,cilitate The melting is absolutely irreversible comparison. at pH 4.0 or above in 50 mM-Citrate buffer, as evidenced by the fact that both the light scattering signal (not shown) and the fluorescence ratio remain elevated upon cooling. At lower pH, melting becomes reversible to an extent that depends somewhat on the buffer and salt composition. The best results were obtained in 50 mM-glycine where, at pH 3.2 or below, reversiblity is essentially complete, i.e. no increase in light scattering is detected and the fluorescence ratio returns to a value close to the original upon cooling (lower curves, Fig. 2). The latter buffer was therefore chosen for subsequent DXC measurements. Figure 3 presents excess heat capacity curves obtained at pH 2.7 in 50 mM-glycine buffer for the hep-2 fragments and several subfragments derived from them. All of the endotherms obtained under these conditions were highly reversible in that samples remained visibly clear aft,er heating and

Domain Xtructure of Fibronectin

20

40

60

80

IL85

100

Temperature (“C) Figure 2. hep-2 fragments melt reversibly below pH 40. This Figure shows the melting of the 40 kDa hep-XB fragment as monitored by the change in fluorescence buffer

ratio while heating at approx. 1 deg.C/min. The conditions were TBS (pH 7.4); 50 rnM-Na citrate (pH 60 to pH 3.0); 50 mM-glycine (pH 2.5). The vertical scale applies only to the lower curve; the remaining curves were adjusted upwards to facilitate comparison.

similar curves were obtained when the samples were heated a second time, Also shown in Figure 3 are the results of deconvolution analysis of those fragments containing more than one module. The top panel corresponds to the 40 kDa hep-2B fragments, which contains four type III modules, TII,,~,,; the endotherm is deconvoluted into four transitions. Panel B of Figure 3 corresponds to the 30 kDa hep-2A fragment, which is essentially identical to hep-2B except for the absence of module III,,. The endotherm for this fragment is deconvoluted into three transitions. The endotherms of the 23 and 20 kDa bimodular fragments, III,,~,, and III,,~,J are shown in panels C and D to be deconvoluted into two transitions. Panels E to G show the results for three single-module fragments, III, 2, III,, and all of which are well described by a single III,,, two-state transition. The individual modules melt in a similar temperature range as the parent fragments, suggesting the absence of strong interactions between them. Module III,, is the most stable; its t, and AH correlate well with those of the highesttemperature transition in the 40 and 20 kDa fragments that contain it (Table 1). Modules III,, and

20

30

40

50

60

70

80

90

loo

Temperature (“C) Figure 3. Each type III

module within the hep-2 region comprises an independently folded domain, as shown by deconvolution analysis of DSC curves obtained with 30 and 40 kDa hep-2 fragments and subfragments at pN 2.7 in 50 rniv-glycine. The number of transitions exactly corresponds to the number of modules in each case. The modular composition of the various fragments is shown schematically in each panel. Each melting curve appears as 2 nearly superimposable lines representing the experimental and best fit functions.

~1114 are the least stable to the lowest temperature

D. Although

module

III,,

and probably correspond transitions in panels A to was not isolated in quan-

1186

N. Novokhatny

et al.

Table 1 Thermodynamic

parameters

of melting

of hep-2 fragments

1 Pi-&&i hep-213 hep-2A lI113m16 lI1,,.,5 III,2 III,, III, 5 11111,

2

pH 2.7

3

kn

AH

L

AH

40

AH

fm

63.2 63.6 63.6 61.0

75 73 68 67

66.1 66.5

76 74

6%6 68.5 70.0

69 70 gg

73.1

6%

64.0 64.1

Ii 73.4

73

66 -~

66 72.0$

domain-domain

731

740

63

Total enthalpy (k&/molt) 282 216 137 133 73 66 63

--

et al. 1992)

tities sufficient for DSC measurements, it would appear to be responsible for the highest transition in panel R and second highest in panel A. This assignment was supported by fluorescence measurements on 14 kDa III,, which revealed a reversible melting curve with a t, and AEB,, close to the corresponding transition in the parent fragments (Table 1). From these results we conclude that all four of type III modules under consideration form independent domains that do not interact in the parent fragments under these conditions. interactions

Close inspection of the curves in Figure 2 indicates that the fluorescence-detected melting transition of 40 kDa hep-2B is noticeably steeper at pH 3.0 than that at pH 2.5, even though the reversibility is still quite good. It appears that the melting process is more co-operative at the higher pH, suggesting the presence of domain-domain interactions. To explore this situation, we compared 25

Temperature ( “Cl Figure 4. DSC profiles

4

__~ AH

From data in Fig. 3; t, values in “C. t 1 cal = 4.184