Vanadyl(IV) conalbumin. I. Metal binding site configurations

Vanadyl(IV) Conalbumin. Site Configurations*

I. Metal Binding

J. David Casey and N. Dennis Chasteen Deparfment of Cizemistry. University of New Hampshire

ABSTRACT Electron paramagneticresonance(epr) and ultravioletdifferencespectroscopyof vanadyl conalbumin indicatea bindingcapacity of two vanadylions, VO2+, per protein molecule

in the pH 8-11 range; the biding capacity drops in the pH 6-8 range with an apparent pK,’ = 6.6. Iron-saturated conalbumin does not bind vanadyl ions, which suggestscom-

mon binding sites for iron and vanadium.Ultravioletdifference spectioscopy indicates 2-3 tyrosinesare involvedin the biding of each metal ion; pH titrationsshow that three protons are releasedper vanadyl ion bound by conalbumin. Room and liquid nitrogen temperatureX-band (ca. 9.2-9.5 gHz) epr spectra show that the vanadyl ion binds in three magneticallydistinct environments(A, B, and CJ that arise from interconvertiile metal site configwations.These confi~tions are probably examplesof conformational substratesof the protein Q-band (ca 34 gIGI epr spectraresolve the spectralfeatures more clearly and show that two conf&urations (A and B) have axially symmetric epr parameters but angles of noncoincidenceof 12” and 8”, respectively,between the z components of the g and nuclear hyperfiie tensors. The third (C) configurationhas rhombic magnetic symmetry and a 6” angle of noncoincidence. These observations demonstratethat the metal sites are of low symmetry and are flexible in theirgeometry about the metal_ The isotropicg and nuclear hyperfine tensor values and the line widthsused in computer-simulatedepr spectra are consistent with four oxygen or three oxygen and one nitrogen donor atoms biding equatoriallyto the VOz+ group. The apparentstability constant indicatesthat vanadyl ion bids to conalbuminapproximatelytwelve ordersof magnitudemore weakly than iron to human serotransferrin but still sufficientIystrongly to overcomehydrolysis.

INTRODUCTION Human serotransferrin (HST) and conalbumin (CA) from egg white are similar in molecular weight of ca. 80,000 [l, 21, amino acid content [3], metal binding Address reprint requeststo: N. D. Chasteen,Departmentof Chemistry, Universityof New Hampshire,Durham, NH, 03824, USA. l This materialconstitutespart of the dissertationof J. D. C. submittedto the GraduateSchool of the University of New Hampshireat Durham in partial t%ilhnent of the requirementsfor a Ph.D. degreein In0rganicChemistry~ Jounrnlof Inorgnni Biochemk@~13,111-126 (1980) Q EIsevierNorth fIoI!and, Inc., 1980 52 VanderbiltAve., New York, New York 10017

111 0162-0134/80/060111-16302.25

J. David Casey and N. Dennis Chasteen

112

capacity of two iron atoms per protein molecule [2,4-6] , and they appear to originate from the same ancestral protein [7]. Several other properties of these proteins are also similar [3,8,9] _ While the primary function of human serotransferrin is the transport of metabolic iron, the role of conalbumin is less certain [3]. It may be important in protecting the developing chick embryo by denying bacteria essential trace metals [lo-l21 _ Although the transferrins may be involved in the transport of metals other than iron, to date only the binding constants for Fez+ are known [13,14]. Our interest in vanadium binding to conalbumin stems in part from the fact that VOa* epr has been used very successfully to study other transferrins. Furthermore, this metal is now known to be an essential element for mammals [15] and may play a role in iron metabolism as well. It is firmly established that the normal growth and development of rats and chicks on vanadium-deficient diets is retarded [16,17]. Rats show elevated iron levels in blood and bone, and chicks have increased hematocrits [I 8] . Recent studies with rats administered vanadium intravensously in different forms show that the metal becomes associated with the plasma serotransferrin fraction (probably as V02+) and eventually with the transfer-r-inand fen&n fractions from liver [ 191. Other studies employing epr spectroscopy have demonstrated that vanadium in the liver exists as protein-bound V02+ [20] _In vitro experiments with erythrocytes demonstrate that the cells take up metavanadate, VOd3-, which undergoes reduction and ultimately binds to hemoglobin as V02+ [21]. V02+ epr spin probe studies of the metal and anion binding sites of I-IST have revealed several features not known previously [22-291. IR particular, it has been demonstrated that anion binding is required for V02* to bind to HST, as in the case with Fes+, and that the anion is directly linked to the metal [24,26]. In physiological media, the anion is bicarbonate or carbonate. Moreover, the two metal sites are distinguishabIe in V02+ epr spectrum and exist in different conformations (configurations) at physiological pH [29] _At pH 10, the two sites become spectroscopically equivalent, having the same configuration. Changes ‘a conformation appear to be important to the function of the transferrins (for reviews of this topic see Refs. 8 and 9). Our distinction between conformation and configuration is presented in the Discussion section. Because HST and CA have evolved separately, it is important to know whether significant differences exist in their metal and anion binding properties. We have undertaken a detailed study of VO 2+ binding to CA to determine (1) the stoichiometry of the metal-protein complex, (2) the binding constant; (3) the role of the anion in vanadium binding, (4) the magnetic symmetry of the metal sites, (5) the nature of the protein ligands, and (6) whether the metal sites of CA exist in different configurational states with properties similar to those of HST. This report and the following one present results demonstrating that vanadium and anion binding to CA and HST differ in several aspects. The implications of these findings are discussed. EXPERIMENTALPROCEDURE Apoconalbumin was prepared by the method of Woodworth and Schade [30], subject to the modification of using carboxymethylated Sephadex C-50 cation exchange resin in the cohmn chromatography step [31]. The purity of the apoconalbumin was verified by anodic disc gel electrophoresis on 5%, 7%, and 10% polyacrylamide gels in 2.5 x IO- 3 F tris, 2 X 10-2 F glycine, pH 9.5. A principal band with very faint secondary bands is observed in each type of gel. Atomic absorption spectroscopy indi-

Vanadyl Conalbumin.

I.

113

cated the residual iron of the apoconalbumin varied from 0% to 3% of the total iron binding capacity in different preparations of the protein. The dialysis procedure of Price and Gibson [32] was used to remove this iron, employing dialysis tubing with a cutoff of 12,000. A 034 mM conalbumin solution was dialyzed twice for 6-hr periods against 30-fold volumes of 0.1 F sodium acetate, 0.1 F citric acid, pH 4.7. The apoprotein reversibly denatures at a lower pH [33]. Most of the citrate and acetate were removed by dialysis against two 30-fold volumes of distilled, deionized water for 30 hr each. The residual citrate was removed by two dialyses against thirtyfold volumes of 0.1 F sodium perchlorate, pH 5.4, for ten hours each. A final dialysis against three 30-fold volumes of distilled deionized water for 6 hrs each gave an apoconalbumin solution, pH 4.7, which was free of iron. The solution was then neutralized and lyophilized; the white, solid apoconalbumin was stored at 4’C. All regents used for the dialysis were extracted with a 0.001% (by weight) solution of diphenylthiocarbazone in distilled carbon tetrachloride to remove heavy metal impurities [34] _ A second set of extractions with distilled carbon tetrachloride removed traces of diphenylthiocarbazone. The concentrations of apoconalbumin solutions were determined spectrophotometrically at 280 nm using an Ezzm of 12.0 [6] and a molecular weight of 76,000 r2,333 Stock solutions of aqueous vanadyl ion were prepared from VOS04 nHa0 crystals (Fisher Scientific Co.). The vanadyl ion concentration was determined spectrophotometrically using a molar absorptivity of 18.0 f 0.2 M-l cm-l at 750 nm ]35]. Solutions of conalbumin were purged with moist nitrogen gas for 30 min prior to the addition of vanadyl ion to minimize solution evaporation while preventing air oxidation of the vanadium_ In those experiments requiring close monitoring of the pH, a procedure similar to that of Fitzgerald and Chasteen [36] was used. A Radiometer pHM 26 expanded scale pH meter equipped with a Radiometer GKZ321C calomelglass combination electrode was used to obtain pH readings to L-O.01 pH units. Reproducible pH readings were possible on 100 JLIsamples. The amount of free acid inherent in solid VOSO, nHz0 was determined by titration of 0.4 X 10-a F VOSO,-oxalic acid, 1:2, with standardized 0.100 N NaOH to pH 7 under a nitrogen atmosphere_ The equivalents of base required in excess of 4 per VOSO* were attributed to free acid; this was typically 0.2 equivalents per V02+. The number of protons released per VO a+ bound by conalbumin was determined by pHrestoration. A 9.4 X low3 F VOSO, solution was added to 0.5 mM conalbumin, pH 8-9, in increments of 0.60 VOa+ per conalbumin up to 3.0 VOa+ per conalbumin. Following a S-min equilibration period with each increment of V02+ added, the protein solution was titrated back to the original pH with standardized 0.010 N NaOH. The contribution due to free acid was subtracted to obtain the number of protons released per V02+ bound to the protein. Ultraviolet difference spectra were recorded on a dual-beam Car-y Model 14 spectrophotometer. Three-milliliter volumes of nitrogen-flushed 0.025 mM apocon&m~ in 0.01 F NaHC03 were placed in two stoppered, nitrogen-flushed, l-cm cuvettes. Successive increments of 0.94 X 10m2 F VOSO, were added to one cuvette followed by shaking and equilibration for 5 min. The difference spectrum relative to apoconalbumin was recorded from 240 to 310 nm. The solution was buffered at pH 9.5 by the bicarbonate. Concentrations of O-2-0.5 mM conalbumin were used in the epr experiments. Normally solutions were made ten fold molar excess in bicarbonate relative to conalbumin

114

J. David Casey and N. Dennis Chasteen

with 0.90 F NaHCOa. The pH of the solution was easily controlled with 0.5 N HCl or NaOH. epr spectra were recorded at X-band (9.5 gHz) and Q-band (35 gHz) frequencies_ A Varian E-4 or Varian E-9 spectrometer was used in the X-band mode. Samples were placed in quartz Bat cells for room-temperature X-band spectra and in quartz tubes (approximately 4 mm o-d., 3 mm i-d.,) for X-band spectra recorded at 77 K. Both types of ce& were capped and flushed with nitrogen prior to the introduction of vanadyl-containing solutions. Careful tuning of spectrometer for quantitative measurements has been described (35). The quartz tube was placed in a quartz, liquid-nitrogen Dewar insert for the low-temperature spectra. The effect of nitrogen bubbling on the spectrometer stability was minimized by the method of Chasteen [37]. A speck of diphenylpicrylhydrazyl, g = 2.0036, was taped on the outside of the flat cell or Dewar inert as a g-mark for spectra recorded on the E-4 spectrometer. On the E-9 spectrometer Varian pitch, g = 2.0028, was used as a standard in a TE;oa mode rectangular dlual cavity. For Q-band spectra the Varian E-9 spectrometer was fitted with high field pole caps, an E-266 cylindrical cavity, and an E-110 Q-band microwave bridge. All Q-band spectra were recorded at approximately 100 K on samples in quartz capillary tubes (approximately 2 mm o-d., 0.8 mm i-d.) which had been flushed with nitrogen prior to the sample introduction. The line position of the vanadyl conalbumin complexes were taken from the X-band epr spectra recorded at 77 K and were used in second-order perturbation equations [38] to obtain initial estimates of g values and vanadium nuclear hyperfme splittings that were then refined by computer simulation of the epr spectra with a modified version of a program developed by white and Belford [39] for S = l/2 systems [27] _ Noncoincidence of the g tensor axes with the nuclear hyperfime tensor axes was suggested by simulations of Q-band spectra. The simulation programs were run on a DEC10 computer equipped with a CALCOMP plotter. A Gaussian line-shape function was used with a 12-G linewidth for the absorption curve at half-height of the x, y, and z magnetic components of all three magnetic environments.

Stoichiometry

of Binding

The ultraviolet difference spectrum of vanadyl conalbumin has maxima at approximately 250 and 290 nm at pH 9.5. A spectrophotometric titration of apoconalbumin with VO*+ ion while monitoring these bands shows breaks at nominally 2 VO*+ per conalbumin (Figure 1); this suggests that there are two vanadium binding sites per conalbumin. Absorption maxima at 250 and 290 run are commonly observed for various transition metal complexes with the transferrins [6,40,41] and are believed to be largely due to the ionization of coordinating tyrosines in the specific metal binding sites. By using AE = 10,000 at 250 nm and AE = 2310 cm -l M--l at 290 nm [41] we calculate from the 250- and 290-run intensities that 1.7 and 2.7 tyrosines, respectively, are ionized per V02+ bound. This result suggests that 2-3 tyrosines are ligated to the metal in the vedyl-protein complex. The room-temperature, X-band, epr spectrum of VOa+ ion in conalbumin solution is anisotropic, indicating binding of the VCP+ ion to a slowly tumbling macromolecule.

115

Vanadyl Conalbumin. I.

A

.80

-60

-20

0

1.0

3-O

2.0

VO/CA

FIGURE 1. Absorbance differenceat 250 and 290 run of vanadylconalbuminversusapoconalbumin as a function of the VO2*/conalbumin ratio. Protein concentration25 X 10-S M in 0.01 F NaHC03, pH 9.5.

A similar spectrum has been reported for (VOs+js-HST [22] _ epr spectrometric titrations of conalbumin in the pH range 8-l 1 (Figure 2) with the vanadyl ion while monitoring the strong central line in the spectrum showed intensity breaks at 192.0 VOa+ per conalbumin, a result consistent with the ultraviolet difference spectral measurements at pH 9.5. No room-temperature vanadyl epr signal is observed when a VOe+ ion is added to iron-saturated conalbumin When excess ferric ion is added as FeCls, the vanadyl ion is displaced from the protein with a resulting loss of the vanadyl epr signal. This establishes that V02+ and Fea+ most likely compete for the same binding sites. Binding Constants A study of V02+ binding to the protein below pH 8.9 is summaried in Table 1. Since the data are not sufficiently refined to distinguish a small difference in binding constant between the two sites, we have assumed that the binding of the VOa+ ion to the protein is random, i.e., K,’ = 4K2’, where K1’ and K2’ are the apparent binding constants for binding of the first and second V02+ ions, respectively. K

,

_

WO**-W

’ - p02+]

[CA] ’

IWO- )2-W K2’ = po2+]

1 Unbound VO2+ ions form epr-silentspeciesin the pH 6-l 1 range [42]_

[V02+_CA]’

i

J. David Casey and N. Dennis Chasteen

116

0

1.0

PHa.0

0

PHs3

0

PHS.4

.

PWlOD

0

PWll.0

I

20

3.0

VO/CA FIGURE 2. Normalized ratio of the epr first derivative signal height divided by the spectrometer gain setting as a function of the VO2+/conalbumin ratio at varied PH. The most intense signal of the room temperature, X-band spectrum was used. The pH 9.3 and 9.4 studies were done in lOOfold molar excess HCOS- relative to conalbumin whiIe the other studies were in tenfold moIar excess HCOe-_ Signalto gain ratios were normalized for each pH hecause the conalbumin concentrations and positioning of the epr flat cell holders were different at each pII

In the pH range of these experiments, the concentration of free V02* ion is given by v@+] = &,/[OH-] 2, where Ksp = 1.l X 1O-22 Ma [42] _ From equation 6 of Ref. 14, with R = 4K2’/KI’ = 1, we calcuIate the KI’ values listed in Table 1 for different vahresof pH_ A plot of log KI’ versuslog [H*] is linear(correlation coefficient = 0.99) with a slope of 29, indicating that three protons are released per VOe+ bound by the protein. Ibis result is confirmed by pH restoration experiments above pH 8.0 (see experimental) showing an average, for sixteen determinations, of 3.0 f 0.4 protons released per V02+ bound. This compares with a value of 2.7 5~0.1observed for VOe+ binding to HST [29] Accordingly we write the true binding constant KI as K

= 1

~02’-W p02']

[@I

3

[CA]

-

From the pH 6.0-7.23 data we calculate a mean KI value of (4.5 -C1.3) X lo-13 Ms_ Increasing the concentration of HCOa- from 2.9 mM to 0.1 M does not increase vanadium binding_ Other observations discussed in the following paper [43] also suggest that VW+ binding to conalbumin is not facilitated by HC03-_

Vanadyl Conalbumin. I.

TABLE

117

1. VO2+ Binding CapacityO PH

% Saturationb

8.10

100

7.23 6.70 6.46 6.26 6.00

74 67 35 31 18

1.4 x 8.2 X 8.2 x 3.2 x 4.4x

109 107 106 10‘5 105

=Concentrations of protein, VOSO4, and NaHC03 were 0.29, 058, and 2.9 mM respectively_The VOSO4 was added to the protein-bicarbonate solution above pH 8 and equilibratedfor 10 min before lowering the pH with OS N HC!ito the valuelisted. b The percent saturation of the binding sites of conalbuminas determined from the peak-to-peak height of the first-derivativeroom temperaturesolution epr spectrumat X ban& f Calculatedfrom equation6 of Ref. 14.

The data in Table 1 was also cast in the form of a Henderson-Hasselbalch

pH = p&

equation,

+ m log [(occupied sites)/(unoccupied sites)] . A linear regression yields m =

0.95 and pK,’ = 6.6 (correlation coefficient = 0.95). While it is tempting to associate the loss of VO2+ binding with the protonation of a key functional group with a p&I = 6.6, it can be shown that K,’ is actually a composite of other constants, i.e.,

where KW = [H+] [OH-],

I&,

= [V02+]

,= K SSSDC

[OH-]

2, and

bowled sW [I-I? 3

=2.5

x

lo_1

3

[unoccupied sites] [v02+]

Actually, the intrinsic pH dependence of VO2+ binding to conalbumin, i.e., the ease with which coordinating functional groups are protonated, is reflected in the value of K PMOC’ (or K1)_ In retrospect, the same can be said of V02+ binding HST in which an apparent pK,’ = 6.6 was also observed [29] _ Frozen Solution Spectra The 77-K, X-band spectra are useful in characterizing the binding sites of conalbumir Figures 3 and 4 show X-band spectra of divanadyl CA at pH 8,10, and 1 l_ A pH ! spectrum has been presented elsewhere 1281. The designations A, B, and C show thret magnetic environments of the bound VO2+ ion. These designations are arbitrary and not necessarily related to labels previously used in studies of vanadyl HST [22,25,28] _ The numerical notation (e-g_, (-5/2),) refers to the MI value of the nuclear spin state of the parallel and perpendicular lines, M1 = *7/2, -C5/2, s/2, *l/2. The A, B, and C

118

J. David Casey and N. Dennis Chasteen

pn 10

I

3.80 6

FIGURE 3. First derivative X-band epr spectra of divanadyl conalbumin. F’rotein concentration 0.28 mbf in 28 X 10-S F NaHCOa at stipulated pH values. Microwavefrequencyapproximately

9.2 GHz, modulation amplitude5 G at 100 kHz, 77 K Note that the perpendicubrresonanceof A. B, and C, overlap.

resonance lines are also apparent in the corresponding demonstrating that

room temperature

spectra,

they are not an artifact of freezing the sample.

The peak heights of the (-7/2),,) lines of the A, B, and C resonances, scaled to 100, are pIotted as a function of pH in Figure 5. The three resonances have nearly equal linewidths and the relative concentrations are therefore proportional to the peak heights [a]_ Within the range pH 8 to 11 the relative concentrations can be directly converted to the absolute concentrations of bound V02+ ion since the binding capacity of conalbumin is constant at approximately 2.0 V02+ per CA Figure 5 indicates that all three resonances are prominent between pH 8.5 and 10.7. Below pH 8.5 the C

resonance is weak and above pH 10.7 the B resonance dominates. The interchange of the A, B, and C intensities is reversible within the range of pH 8 to 11. The epr spectrum is only dependent on the pH at which it is recorded, provided the VOa+ ion is added between pH 8 and 11_ Figures 6 and 7 show the Q-band spectra recorded at pH 8,10, and 11. The pH 10 spectrum (Figure 6) reflects approximately equal contributions of the A, B, and C environments and has many well-resolved features_ The pH 8 spectrum (Figure 6) has most of the same features; as in the X-band mode, however, they are not as well re-

Vanadyl

Conalbumin.

I.

119

FIGURE 4. First-derivative experimental (top) and simulated (bono& X-band epr spectra of divanady1 conalbumin Conditions as in Figure 3 except pH = 11. Simulation (see Experimental section) is for the B confmration only and includes an angIe of noncoincidence in the xz plane of 8”.

solved. At pH 1 I the Q-band spectrum (Figure 7) shows that A and C contribute but B is dominant. The A and C contributions are not as obvious in the pH 11, X-band spectrum (Figure 4) The magnetic symmetry and epr parameters of the three environments were obtained with the aid of computer-simulated spectra. The C features are most prominent in the pH 10 X- and Q-band spectra and in the pH 10 spectra of mixed metal conalbumin containing Fea+ and V02+ [43]. The doublet features that are marked C, and C,, in Figures 3 and 6 are characteristic of rhombic tensors [45,46]. The B environment was easiest to study because it dominates the X- and Q-band spectra at pH 11. Only one set of absorbances is identified in the perpendicular region, which is typical of an axial complex. These features are marked Bl and compared tith the simulated spectra in Figures 4 and 7. The A resonances were difficult to characterize because they do not dominate in any experimental spectra. Once the B and C features were identified, the remaining ones were assigned to A. They are also characteristic of axial symmetry and labeled AU and Al in Figures 3 and 6.

J. David Casey and N. Dennis Chsteen

120

; 50 ‘-______._3 I I .--‘\ ‘. -0 z , _. 40* ----__._ -\- ,’ --..S___ . . L, / .’ --.-_.. I;-“<. ._.zp~-_,-’2, 30A .’ -1.*-._-1 .’ 5-.-\.

20- __.___-------__<

---_\ =-lrc,

IO7.0

I

I

I

8.0

9.0

10.0

I 11.0

PH

FIGURE 5. The relativepeak heights of the A, B, and C (-7/2) resonancesof divanady1conalbumiu as a fimction of pH in the range 8-11. Conditions as in Figure 3. Vanadyl ion bindingdecreasesbekw pH 8 CabIe 1).

Close comparison of the simulated and experimental Q-band spectra indicate that

of the g-tensor and nuclear hype&me tensor are noncoincident. This situation has been discussed for (VO2+),-HST by White and Chasteen [27] and will not be presented in detail here. The most characteristic observation is that the spacing between the Mr = -7/Z and MI = -S/2 parallel lines in the experimental spectrum is larger than in the simulated spectrum. Angles of noncoincidence of 12”, 8”, and 6” in the xz plane were found for A, B, and C, respectively, to give the best computer fits to the experimental spectra. Angles of this magnitude do not noticeably affect the perpendicular regions of the Q-band simulations (Figure 7) nor are the parallel or perpendicular regions of the X-band simulations (Figure 4) significantly altered. The epr parameters are the axes

listed in Table 2.

DISCUSSION Tyrosine

Ionizalion

The ultraviolet difference spectrum of the apoconalbumin at alkaline versus neutral pH shows maxima at 245 and 295 nm with an absorbance ratio for A24r,:A29 5 of 4.5: 1 .O 163 while the Azso :A2ao ratio for vanadyl conalbumin is 2.4: I .O,suggestingprocesses in addition to tyrosine ionization. Tan and Woodworth [6] converted the two absorbance maxima from difference spectra of various metal-conalbumin complexes into the number of tyrosines ionized per bound metal ion. They obtained similar values from each of the two bands, but our data for vanadyl conalbumin converts of 1.7 Tyr per V02+ at 250 mn and 2.7 at 290 ML An analogous situation is observed for vanadyl

Vanadyl Conalbumin. 1.

121

FIGURE 6. First-derivative Q-band epr spectraof div-dyl conalbumin_Solution conditionsas in Figure 3. Microwave frequency approximately 34 GHz, modulation amplitude 5 G at 100 kHz, temperatureapproximately 100 kHz

HST in which the 246 nm maximum converts to 1.8 Tyr per V02+ and the 296 nm to 2.4.2 The exposure of buried tryptophan functional groups to polar solvent during binding and subsequent absorbance increase in the 290-run region could be responsible for these observations [47] _Confonnational changes during met& binding has been observed in the formation of ferric conalbumin [48-501. maximum

Metal Binding

The decreased V02+ binding below pH 8 occurs over a different pH range than for Fe3* [2, 51, 521. A dependence of binding at different pH ranges for different metal ions could be important if conalbumin controls the unbound metal ion content as the egg matures; the pH of the egg white is 7.6 at laying but increases to 9-9.5 within two days [3]. Based on the reported K,’ value for Fe3+-HST [13], Fe3* binds 10x2 more strongly to HST than VO 2+ does to CA at pH 6.7. Nominally three protons are released per metal bound for either Fea* or VOe+. In the case of (‘V02+)&IST, only one site loses its binding capacity for VOa+ in the pH range 6-O-7.0, whereas both sites of (V02+)2-CA do so below pH 8.0. More2 In a prior report, these values were mistakenly interconverted [ 231.

J. David Casey and N. Dennis Chasteen

122

FIGURE 7. First derivation experimental (top) and simulated
over, bicarbonate is required for VO 2+ to bind to HST 1291 but apparently is not for binding to CA l%ds point will be addressed in the following report [43]. Properties

Configurational

The literature Woodworth

indicates that conforruational states exist for conalbumin. Tan and subjected apoconalbumin and various dimetal conalbumin com-

[40]

TABLE 2. epr ParametersO Confi&uration

gz

Azb

gx

Ax”

gy

Ayb

go=

AoC

A

1.940 1.937 1939

17L2 170.7 163.9

1.972 1.975 l-978

58.6 56.6 52.3

1.972 1.975 1.974

58.6 56.6 59.2

1.961 1.962 1.964

96.1 94.6 91.8

B c

a A, B, and C have angles of noncoincidence equal to 12”) 8O, and 6”) respectively, in the xz plane. Hyperfine coupling constants are g units of l@ cm-k 6 A,, A,, and A, are the symbols for the components of the nuclear hyperfine interaction along the x, y, and z axes and are to be distinguished from the symbols, AU and AL used to designate resonances of VO2+ in the A configuration ofconalbumin
Vanadyl Conalbumin. I.

123

plexes to wide ranges of temperature, denaturant concentration, and pH while monitoring changes in the fluorescent intensity of the tryptophan functional groups which were related to reversible conformational changes in the protein. Price and Gibson [53] explained the altered epr spectra of ferric conalbumin and ferric HST in perchlorate ion solution as a conformational change induced by perchlorate, a chaotropic agent [53]. Differences in physical properties [48], hydrogen exchange with solvent [48], and resistance to proteolysis [50] indicate that ferric conalbumin differs in macromolecular conformation from the apoprotein. It is probable that A, B, and Care three configmations of the metal binding sites.3 Multiple binding site configurations that are pH dependent have also been postulated for vanadyl HST 1291. In HST, the protonation of a pK, = 10.0 functional group is associated with a configurational transition; however, the reversible changes in the metal site configurations of conalbumin above pH 10 shown in Figure 5 cannot be explained by the ionization of a single functional group. In general the pH dependence of the epr spectrum of conalbumin is quite different from that of HST. The A, B, and C signals of conalbumin reflect the sensitivity of epr in detecting differences in the localized binding configurations that are probably related to subtle variations in the macromolecular conformation corresponding to conformational substrates of nearly equal thermodynamic stability. Detection of conformational substates in proteins [55] supports the concept of dynamic rather that static m,lcromolecules.

Magnetic Parameters and the Metal Sites The g and nuclear hyperfme coupling constants can be used to obtain information about the ligand atoms bound to vanadium, the ligand field symmetries, and bonding parameters. A correlation between the isotropic g-factor [jr,,) and the isotropic nuclear .hyperfme values (A,) as done for HST (27) suggests that the equatorial ligands in the A, B, and C configurations are four oxygen or possibiy three oxygen and one nitrogen donor atoms. The 12-G linewidths used in the epr simulation program are somewhat larger than those used for human serum transferrin [27] and are consistent with nitrogen bonding which is usually characterized by line broadening in vanadyl complexes. Postulates of metal interactions with tyrosine [40, 56-581 and histidine [14,59,60] support this deduction. The angles of noncoincidence for the A, B, and C configurations and the rhombic epr parameters for the C configurations indicate that the binding sites are of low symmetry. The unexpected observation of axially symmetric g-tensors for the low symmetry A and B configurations can be explained by a mechanism in which the 3d,z metal orbital is mixed into the ground state 3d,a_,2 orbital 1613. The variation in epr parameters from one configuration to another could be attributed to a change in geometry of the ligand field about the metal without a change in ligands. Parameter variations of similar magnitude have been observed for cis and rrans a An alternate, but Iess likely, explanation of the data is that A, B, and C correspond to different “orientations” of the VOz+ ion in a rigid bindingsite such that the disposition of the fmt coordination sphere ligands about the metal is different in each case. Because the A, B, and C epr signalshave comparable intensities, these “‘orientations” would have to have nearly equal thermodynamic stabilities.This would not be expected given the probable low symmetry of the binding site and the fact that the strengths of ligand bindingin the axial and equatorial positions of the VOa+ ion are very different.

J_ David Casey and N. Dennis Chasteen

124

vanadyl(IV) bis(benzoylacetonate) [62] and for square planar and trigonal bipyramidal vanadyl tartrate complexes [63] _ CaIcuIation of a2, the population of the s&2-,.2 orbital containing the unpaired electron, followed the procedures outlined by Belford and Hitchman [62]. Values of (Y* = 097,0.97, and 0.92 were obtained for the A, B, and C configurations respectively and indicate that the orbital is largely non-bonding as is usually the case for VOa+ complexes 162]_ The somewhat lower value of on for the lower symmetry C configuration possibly reflects increased “in-plane” n bonding with the protein ligands brought about a marked change in geometry about the metal. The C configuration also has the smallest isotropic hyperfiie coupling constant, A o, indicating a greater electron spin delocalization onto the ligands. Low symmetry allows 4S mixing into the ground state which can further reduce the value of A0 [64].

CONCLUSION We propose a conformational substrate model for V02* binding to CA in which the three low symmetry metal site configurations probably difGer from one another in the geome’tical arrangement of the ligands about the metal. Studies reported in the subsequent paper 1433 show that both metal sites exist in all three configurations_ The observed noncoincidence of the g, and A, tensor components is possibly due to the presence of a fifth ligand atom not coordinated perfectly trans to the vanadyl oxygen or by a large distortion in the equatorial plane of the square pyramidal or square bipyramidal coordination geometry normally found f@r VO*+ complexes_ It is difficult to say whether these previously unknown configurations have a physiological role. It is clear, however, that considerable flexibility exists in the metal sites of this protein_ This property is probabIy a necessary one if CA is involved in sequestering (or transporting in the case of hen serotransferrin) a diversity of metal ions with different chelation requirements. Although the metal binding constant for V02+ is considerably weaker than that for Fea’, it is sufficiently large to overcome hydrolysis of VOa* at physiological p& The hydrous oxides of Fea+ are much more insoluble than those of V02+ and therefore a larger affiity of the protein for iron is required to solublize this metal. Since conalbumin may be a chelator of metal ions in general, it would be of interest to establish whether other metal ions also follow this trend. The authors

s&h to

thank Dr. John WiIbkms of the Universi~

ments on the manuscripf_ Thid work was supported Irz-tute of Health

of Btikfoifor~~LWUI heIp&I com-

in poll by Grant Gbf'20194fiom

the NationaI

REFERENCES 1. IL G. Mann,W. W. Fish,A-C. Cox, andC. Tanford,Biochen&try 9,1348

2. R. C_W&nerapd I. Weber,J.Am Chem_.Soc. 75,5094 (1953). 3. R. E:Feeney andS. K. Komatsu, Stnrcfure a@ Bon&g 1,150 (1966). 4. C. B. Lanrell and B. Ingelman,Acta C?mn.Scnnd. 1,770 (1947). 5_ C_ IC.Luk, Biochemirhy IO, 2838 (1971).

(1970).

6. A. T- Tan and R- C. Woodworth, Biochtvn&by 8,3711(1969).

7. 3. BIuar&Deconinck,P. L. Mason, P. A. Osinski,and J. F. Heremans,i?ioc?&n_ Bsopirvs Acre 365,311(1974).

125

anadyl Conalbumin. I.

8. M. Llinas, Srmcture and Bonding 17,135

(1973).

9. N. D. Chasteen, Coord. C7zem. Rev. 22,1(1977). 10. hf. Sussman, in Iron in Biochemistry andMedicine,A. Jacobs andhi. Worwood, Eds, Academic 11. 12. 13. 14. 15.

Press, Inc., New York, 1974, p_ 650. R. E. Feeney and D. A. Nagy, J. Bacteriil. 64,629 (1952). J. A. GarIIdi and H. G. Bayne, J. Food Sci 27,57 (1962). P. Aisen, A. Leibman, and J. Zweir, J. Biol. Chem. 253,193O (1978). R. Aasa, B_ G. MaImstrom, P. Saltman, and T. Vanngard, Biochim Biophys. Acfa 75, 203 (1963). L. L. Hopkins Jr., H. L. Cannon, A. T. Miesch, R. M. Ross, and F. H. Nielsen, Geochem. En-

v*n 16. 17. 18. 19. 20.

L. K. F. E. J.

2,93

(1977).

L. Hopkins Jr. and H. E. Mohr, Fed. Proc., Fed. Amer- Sec. Ekp_ BioL 33,1773 (1974). Schwarz and D. B. Milne, Science 174,426 (1971). H. Nielsen and D. A. Onerich, Fed. Proc. 32,929 (1973). Sabbioni, E. Ma&ante, L. Amantini, and L. Ubertalli, Bioinorg. Chem. 8,503 (1978). L. Johnson, H. J. Cohen, and K. V. RajagopaIan, Biochem Biophys Res Commun. 59,

940 (1974). L.C. CantIey, Jr., andP. Aisen,J. Biol. Chem. 254,178l (1979). J. C. Cannon and N. D. Chasteen, Biochemistty 14,4573 (1975). J. C. Cannon and N. D. Chasteen, in Proteins of Iron Storageand TransportinBiochemistry andMedicine, R. R. Crichton, Ed., North-Holland PubIishiig Co., Amsterdam, 1975. p_ 67. 24. R_ F_ Campbell and N. D. Chasteen, J. BioL Chem. 252,5996 (1977). 16,560 (1977). 25. D. C. Harris, Biochemistry 26. D. C!. Han-is and M. H. Gelb, unpublished. 27. L. K. White and N. D. Chasteen, J. Phys_ Chem. 83,279 (1979). 28. N. D. Chasteen, R. F. Campbe& L. K. White, and 1. D. Casey in Proteins of Iron Metabolisnz. E. G. Brown, P. Aisen, J- Fielding, and R. R. Crichton, Eds., Grune and Stratton, Inc., New York, 1977, p. 187. 29. N. D. Chasteen, L. K. White and R. F. Campbell, Biochemistry 16,363 (1927). 82,78 (1959). 30. R. C. Woodworth and A. L. Schade, Arch. Biochem. Biophys 31. P. Aisen, G. Lang, and R. C. Woodwcrth,J. Biol. Chem. 248,649 (1973). 32. E_ M. Price and J. F. Gibson, Bibchem. Biophys. Res. Cbmmun. 46,646 (1972). 33_ Y_ Yeh, S. Iwai, and R. E. Feeney, Biochemistry 18,883 (1979). 34. E. B. Sandell. in Colorimehic Determination of lzaces of Metals 3rd ed., Interscience Publishers, Inc., New York, 1959, p. 139. 35. J. J. Fitzgerald and N. D. Chasteen, Anal. Biochem 16,170 (1974). (1974). 36. J. J. Fitzgerald and N. D. Chasteen, Biochemistry 13,4338 37. N. D. Chasteen, J, Magx Res. 27,349 (1977). 38. B. Bleaney,PhiZos. Msg. 42,441(1951). 39_ L. K. White and R. L. BeIford, J. Am. them. Sot. 98,4428 (1976). 40. A. T. Tan and R. C. Woodworth,X Polymer Sci Part C, No. 30,599 (1970). 41. B. Teuwissen, P. L. Masson, P. OsinskI, and J. F. Heremans, Eur. .L Biochem. 31,239 (1972). 42. J. FrancavilIa and N. D. Chasteen, Inorg. Chem. 14,286O (1975). 43. J. D. Casey and I& D. Chasteen, J. Itwrg_ Biochem. 13,127 (1980). 44. I. A. Weil and H. G. He&t, J. Chem. Phys. 38,281 (1963). 45. C. P. Stewart and A. L. Porte, J. Chem. Sot. Dalton Bans. 1972,1661. 46. H. J. StokIosa, J. R. Wasson, and B. J. McCormick,Inorg. Chem 13,592 (1974). 47_ V- S. Ananthanarayanan and C. C. Bigelow, Biochemistry 8,3717 (1969). 8,877 (1969). 48. T. F. Emery, Biochemistry 49. R. A. Fuller and D. R. Briggs,J. Am Chem. Sot. 78.5253 (1956). 50. P. R. Azair and R. E. Feeney, J. Bioi Chem. 232,293 (1958). 51. J. Ross, S. Kochwa, and L. R. Wasserman, Biochim. Biophys. Acta 154, ?O (1968). 52. A. N. Lestas,Brit- L Haematologv 32,341 (1976). 53. E. M. Price and J. F. Gl%son, J BioZ. Chem. 247,803 (1972). 54. Y. Hatefi and W. G. Hanstein,&oc_ Nat. Acad. Sci 62,1129 (1969). 55. H. Frauenfelder,G. A. Pet&o, and D. Tsemoglov,Natwe 280,558 (1979). 56. J. L. PhIllips andP. Azari, Arch. Bibchem. Biophys. 151,445 (1972). 21. 22. 23.

126

J. David Casey and N. Dennis Chasteen

R. Prados, R. K. Boggess,R. B. Martin, and R. C. Woodworth,Bioinorg. Chn 4,135 (1975). 58. R C. Woodworth, K. G. MoraBe, and R. J. P. Williams,Bkhemistry 9,839 (1970). 59. Y. Tomimatsu, S. Kid, and J. R. Scherer, Biochent. Biophys. Res. CZbmw~ 54,1067 (1973). 60. R. C. Woodworth, R. J. P. Williams, and B. M. Ah&i, inptofeimof IionMeraboEsm, E. G. Brown, P. Aisen, J_ Fielding, and R. R. Crichton, Eds., Grune and Stratton,Inc., New York, 1977, p_ 211. 61. M_ A. Hitchman,C. D. Olson, and R. L. Belford,J, C71enz. Phys. SO,1195 (1969). 62 bf. A- Hitchmau and R_ L_ Belford, Inorg. them 8,958 <1969). 63. M. E. McILwain,R. E. Tapscott, and W. F. Coleman,J. M&n. Res. 26,35 (1977). 64. L. J_ Boucher, E. C. Tymn, and T. F. Yen in Ei’ecrom Spitz Resmmce ofMeraI Gmzpkxes, T. F_ Yen, Ed, plenum Press,New York, 1969,p_ 111. 57.

ReceivedJv

31.1980; revisedFebrumy 19.1980