Jul 5, 2015 - from J. T. Baker Chemical Co., Hepes buffer from Research Organics, ... were acquired on a Digital Equipment Corporation MINC-23 labo-.
Vol. 260, No. 13, Issue of July 5, pp. 7975-7979.1985 Printed in U.S.A.
THEJOURNAL OF BIOLOGICALCHEMISTRY 0 1985by The American Society of Biological Chemists. Inc.
Fe(II1)*ATPComplexes MODELS FOR FERRITIN AND OTHER POLYNUCLEAR IRONCOMPLEXESWITHPHOSPHATE* (Received for publication, December 5,1985)
Azzam N. Mansour@, Carl Thompsonq, Elizabeth C. TheilJI**,N. Dennis Chasteenll**, and Dale E. Sayers$ From the Departments of $Physics and 11 Biochemistry, North Carolina State University, Raleigh, North Carolina 27695 and the VDepartment of Chemistry, University of New Hampshire, Durham, New Hampshire 03824
Polynuclear ironcomplexes of Fe(II1) and phosphate described here may form in vivo either as normal comoccur in seawater andsoils and in cells where the iron ponents of intracellular ironmetabolism or during iron core of ferritin, the iron storageprotein, contains up excess where theconsequent alteration of free nucleoto 4500 Fe atoms in a complex with an average com- tide triphosphatepools couldcontribute tothe observed position of (FeO .OH)sFeO.OPOsH2. Although phos- toxicity of iron. phate influences the size of the ferritin core and thus is known about the the availability of stored iron, little nature of the Fe(II1)-phosphateinteraction. In the present study,Fe-phosphate interactions wereanalyzed in Iron and phosphate are required by virtually all terrestrial stable complexes of Fe(II1).ATP which, in thepolynu- life forms. Spontaneous formation of polynuclear iron comclear iron form, hadphosphate at interior sites. Such plexes with phosphate can alter the bioavailability of either Fe(II1).ATP complexes are important not only as nutrient inoceans and soils (1,Z). Polynuclear iron-phosphate models but also because they may play a role in intra- interactions also occur in ferritin, a complex of protein, hycellular iron transport and in iron toxicity; the com- drous ferric oxides, and phosphate with the iron core inside plexes were studiedby extended x-ray absorption fine structure, EPR,NMR spectroscopy, and measurement the hollow protein. Ferritin provides an intracellular reservoir of proton release. Mononuclear iron complexes exhib- of iron, overcoming the extremely low solubility of Fe(II1) M (3)) at physiological conditions. The iron core of iting a g‘ = 4.3 EPR signal were formed at Fe:ATP ferritin has an average composition of (FeO.OH)&’eO. , polynuclear iron complexes (Fe 2250, ratios ~ 1 : 3and EPR silentat g’= 4.3) were formed at anFe:ATP ratio OP03Hz (reviewed in Ref. 9) and may contain upto 4500 iron of 4:1. NoNMR signals due to ATP were observed atoms. Whether the phosphate is bound only to theiron or to when Fe was in excess (Fe:ATP = 4:l). Extended x- the protein as well is not clear, since the procedures used to ray absorption fine structure analysis of the polynu- remove iron from ferritin are known to cleave serine phosclear Fe(II1). ATPcomplex was able to distinguishan phate (4, 5). Specific sites for serine phosphorylation on the 0 apoprotein by bovine skeletal muscle catalytic subunit have, Fe-P distanceat 3.27 A in addition to the octahedral at 1.95 A and 4-5 Fe atoms at 3.36 A. The Fe-0 and however, been demonstrated (6). More is known about phosFe- phate in thenative core, although knowledge of its role is far Fe-Fe distancesare thesame as in ferritin, and the P distance is analogous to that in anothermetal-ATP from complete. The location of the phosphate in the microcomplex. An observable Fe-P environment in such a crystalline iron core appears to be restricted to disordered or large polynuclear iron clusteras the Fe(II1).ATP (4: 1) amorphous regions since formation of the crystalline regions complex indicates that the phosphate is distributed of the core in ferritin, reconstituted from iron and the apothroughout rather than merely on the surface, incon- protein, is little, if any, influenced by the presence or absence trast to earlier models of chelate-stabilized iron clus- of phosphate (7). Only 60% of the phosphate in the ferritin ters. Complexes of Fe(II1) and ATP similar to those core may be exchanged by alkali or removed in solutions of magnesium ammonium citrate (S), suggesting a more intimate *This research was supported in part by National Institutes of association with iron than thatof adventitiously bound phosHealth Grant AM20251 (E. C. T., D. E. S.), the North Carolina phate (8,9). Such iron-phosphate interactionscould berelated Agricultural Research Service (E. C. T.), and National Institutes of Health Grant GM20194 (N. D.C.). The work reported herein was to the heterogeneity of iron environments observed during partially carried out at the Standford Synchrotron Radiation Labo- the reductive release of iron from ferritin (10). Experimental ratory, which is supported by the Department of Energy Office of manipulation of the conditions of core formation shows that Basic Energy Services and theNational Institutes of Health Biotech- changing the phosphate concentration alters the apparent nology Research Program, Division of Research Resources. Contri- size (magnetic domain) (11) and polydispersity of the iron bution from the Department of Biochemistry, School of Agriculture core (12). and Life Sciences and School of Physical and Mathematical Sciences. During reconstitution of ferritin from apoferritin and ferPaper 9642 of the Journal Series of the North Carolina Agricultural Research Service, Raleigh, NC 27650.The costa of publication of this rous ion, Fe(1I) is oxidized to Fe(II1) followed by hydrolysis article were defrayed in part by the payment of page charges. This and formation of the polynuclear iron complex; iron release article must therefore be hereby marked “advertisement” in accord- in vitro and possibly in uiuo requires reduction of iron(II1) in ance with 18 U.S.C. Section 1734 solelyto indicate this fact. the core (reviewedin Ref. 9). The participation of the protein 8 Present address: Naval Surface Weapons Center, White Oak, in core formation mainly occurs inthe early stages (13) Silver Springs, MD 20910. ** To whom correspondence should be sent: Department of Bio- through the formation of an initiation complex and small chemistry, Box 7622, North Carolina State University, Raleigh, NC Fe(II1) clusters bound to the protein (14, 15). The metal27695-7622. protein interaction presumably orients growth of the iron 7975
7976
Fe(III).ATP Complexes
polymer toward the hollow center of the protein. Single or the solid FePO,.2H10, with distances obtained from crystallographic multiple (512) sites of initiation may be selected in individual data (21) as a modelfor the nearest neighbor interaction. As a apoferritin molecules, as judged by the presence of single, consistency test of the data, the structure of ferritin was determined for the sample in solution, using FePO, .2H20as a model, withresults large, or multiple crystallites (16). equivalent to those previously obtained (20). Phosphate may influence the propertiesof the iron-protein Electron paramagnetic resonance spectraof samples in calibrated initiation complex, leading to the observed changes in the EPR tubes were measured at 77 K on a Varian E-4 spectrometer properties of the core in the presence of phosphate (11, 12). employing a TE,, rectangular cavity and a dewar insert. Spectra In order to understand more completely the nature of iron- were acquired on a Digital Equipment Corporation MINC-23 labophosphate interactions in ferritin and its initiation complex ratory computer and double integratedto determine spin concentrations. FeEDTA- (0.1 M EDTA, 1 mM Fe3+, pH 7.8) was used as an and in other polynuclear iron-phosphate complexes, we have intensity standard for the g' = 4.3 signal and 0.5 mMCuSO., 0.1 M undertaken a study of models for mononuclear and polynu- HC1 in glycerin/H20 1:3 for the g' = 2 signal. Infrared spectra of clear Fe(II1)-phosphate interactions, including Fe(II1). ATP lyophilized 4:l Fe(III).ATP samples in KBr pellets were measured on a PerkinElmer 283B infrared spectrophotometer, and optical complexes. Such complexes are of particular interest since they could easily form in vivo and, following the release of spectra of solutions were measured on a Cary 219A dual-beam recording spectrophotometer.The brown lyophilized samples redissolve iron from transferrin, may participate in intracellular iron in water to form the characteristic reddish brown solution of the transport and deposition in ferritin. In addition, Fe(II1) stable cluster. 31P NMR spin-lattice relaxation times were determined by complexes may contribute to irontoxicity by altering thesize the inversion recovery method (19) at 28 'C using a JEOL FXSOQ of the nucleotide triphosphate pool. A combination of chem- spectrometer operatingat 36.2 MHz for phosphorus-31. Solutions of ical, EXAFS,' and EPR, and NMR spectroscopy and proton 0.1 M ATP in 0.1 M Hepes, pH 7.4, withiron concentrationsof 0, 25, release studies shows that both mononuclear and polynuclear 50, and 200 PM were employed. Fe(II1) .ATP complexes readily form in solutionat physiologRESULTS AND DISCUSSION ical pH, depending on Fe:ATP the ratio. Mononuclear (Fe(II1) was stabilized by excess ATP (Fe:ATP = 1:3). Polynuclear Chemical and Spectroscopic (EPR, NMR, IR, and Optical) species, formed a t a n Fe:ATP ratio of4:1, consisted of a Properties-Although complexes betweenadenine nucleotides cluster of 250 or more ironatomswiththeapproximate and Fe(II1) have been previously reported (22-24), the nature formula [Fe(OH)g:5 1. EXAFS analysis showed that the av- of the complexes formed underphysiological conditions have erage Fe atom in thepolynuclear complex had 0 (1.95 A), P been littlestudied.For example, 1:l complexesbetween (3.27 A), and Fe(3.36 A) neighbors. Thus, in contrast to otherFe(II1) and AMP or ATP areformed in the pH range 1-2 but polynuclear iron complexes stabilized by anions, phosphate are veryinsoluble (22, 23). In the present study, complex appeared distributed throughout the cluster rather than on formation in the pH range 7-9 was studied using EPR and the surface. NMR spectroscopy. EPR spectra of the mononuclear Fe(II1) .ATP complexes MATERIALS AND METHODS in the pH range 7-9 exhibit a g' = 4.3 signal (Fig. 1, inset), FePOl. 2H20 was obtained from Alfa Inorganics, Fe(N03)3.9H20 indicating high-spin Fe3' in a coordination environment of from J. T. Baker ChemicalCo., Hepes buffer from Research Organics,rhombic symmetry (25). Fig. 1 shows the dependence of the EPR spectral amplitudeof the Fe:ATP ratio as ferric chloride ribose 5-phosphate from Boehringer Mannheim, 2'-deoxyadenosine triphosphate from P-L Biochemicals, and Pentex horse spleen ferritin is added to a solution of ATP. At least three types of comfrom Miles Laboratories. All other materials were Sigma products plexes are formed, as evidenced by discontinuities in the curve and used without further purification. ATP solutionswere standard- at Fe:ATP ratios of 1:3, 2:1, and 4:l (Fig. 1).A similar curve ized spectrophotometrically, = 15.4 X lo3 "' cm" at 259nm (17). Iron(II1)-ATP complexeswereformedby the slow addition of a is obtained with 2'-dATP. At Fe:ATP ratios of 1:3 and below, commercial 0.179 M FeC&standard solution in 1 M HCl to 3.3 D M all of the iron can be accounted for by the EPR signal, as ATP in 0.1 M Hepes/Na buffer, pH 7.0, with stirring using 1.00 N judged by the value of the double integral. Thus, within the NaOH to maintain the pH. In experimentsto quantitate the number precision of the S = 5/2 spin concentration determination, of protons released upon iron(II1)-ATPcluster formation, a freshly which is of the orderof +15%, only monoatomic iron species prepared aqueous solutionof 0.150 M Fe(NO&, standardizedby the arepresent.The reduction in EPR intensity beyond a n thiocyanate method (18),was added to 6.2 mM ATP in 0.1 M Hepes/ ratio of 1:3 indicatesthat polynuclear iron-ATP Fe:ATP Na buffer, pH7 or 9, until an Fe:ATP ratio of 4:l was obtained. The species are being formed (Fig. 1). The iron(II1)-ATP clusters total amount of 1.00 N NaOHrequired to maintain the pHwas recorded. The free acid inthe ferric nitrate solution was shownto be formed in solutions havingFe:ATP ratios of 4:1, where the g' negligible by addition of an equimolar amountof Na2EDTAto com- = 4.3 signal is virtually absent(Fig. I), have reasonably large plex the iron, followed by potentiometric titration with two equiva- molecular weights (see below) and are the primary focus of lents of1.00 N NaOH to the end point. pH was measured with a Radiometer model PHM26 pH meter equipped with a Radiometer the present study. The mononuclear complex was further examined by 31P GK2321Cglass/Ag-AgC1 combinationelectrode. Ultrafiltration or to the coordinating dialysis experimentswere carried out with Amicon membranes having NMR spectroscopy in an attempt identify molecularweightexclusionsizes of 500 (UM05), 10,000 (PMlO), phosphate groups. Spin-lattice relaxation times were mea30,000 (PM30), and 50,000 (XM50), and with Spectraportype 6 sured on solutions containing 0.1 M ATP and concentrations dialysis tubing havinga molecular weight exclusion sizeof 2,000. of Fe(II1) from 0 to 200 PM in 0.1 M Hepes, pH 7.4. The EXAFS and XANES (x-ray absorption near edge structure) measurements were made at Stanford Synchrotron Radiation Laboratory,paramagnetic contribution l/Tlp to the relaxation rate was found to be a linear function of the iron concentration as in the fluorescence mode, under parasitic conditions (1.8 GeV8.17 mA on the Wiggler beam line IV-2). Liquid samples were measured expected if the rate of ligand exchange is fast relative to the in cells as previously described (20), and FePO,. 2H20 was measured difference between the resonance frequencies Aw of free and as a powder. Measurementsof the polynuclear Fe(II1).ATP (41) (12 complexed ATP (26, 27). The normalized paramagnetic conmM Fe(II1))complex were collected overa 6.5-h period.The EXAFS data were analyzed as previously described (20) using the data from tributions, l/fT,,, to the relaxation rate for the CY,8, and y nuclei were determined to be 91,170, 185 andHz, respectively, at 28 "C. These values are approximately 1000-fold smaller ' The abbreviations used are: EXAFS, extended x-ray absorption for the analogous S = 5/2 Mn.ATP fine structure; XANES, x-ray absorption nearedge structure; Hepes, thanthosereported complex (26, 27). In the latter complex, the paramagnetic 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.
7977
FIG. 1. Dependence of the amplitude of the g' = 4.3 EPR signal (inset) on the Fe(II1):ATP mole ratio. Species formed at Fe:ATP mole ratios of 1:3,2:1, and 4:l are indicated. Conditions: Fe(II1) added as 0.179 M FeC4 in 1 M HCl to 3.39 mM ATP in 0.1M Hepes/ Na buffer, pH 7, maintained with 1 M NaOH. EPR instrument settings: field set = 1500 G, scan range = 2000 G, microwave power = 20 milliwatts, klystron frequency = 9.145 GHz, modulation amplitude = 10 G, modulation frequency = 100 kHz, receiver gain = 500, scan time = 8 min, and time constant = 0.3 a.
Fe (m)/ATP Ratio contribution to therelaxation arisesfrom the electron-nuclear dipole interaction involving the coordinating phosphate groups (26,27).Iron(II1) complexes are known to have ligand exchange rates several orders of magnitude smaller than those of Mn(I1) complexes (28).Since it appeared likely that l/fTl, in Fe .ATP was largely influenced by the ligand exchange rate, we pursued this line of investigation no further. At an Fe(II1):ATP ratio of 4:l and an ATPconcentration of 11 mM, 31PNMR signals due to ATP were not observed (1500transients, 2000 Hz sweep width). However, weak NMR signals of Pi and ADP in equal amounts, corresponding to a concentration of 1% that of the ATP,were observed. The lack of an ATP NMRspectrumis presumably due to paramagnetic broadening of the NMR lines caused by the polynuclear iron cluster into which essentially all of the phosphate is incorporated. A broad EPR signal centered at g' = 2 having a line width A H p p -1000 G was observed for the polynuclear Fe(II1) .ATP (4:l)solution at 77 K and was associated with iron retained after ultrafiltration with a PM-30 membrane. Quantitation of the spin concentration associated with g' = 2 signal indicates that thesignal accounts for only 0.8% of the iron present if it arises from a S = 5/2 spin state and 9.3% if it arises from a S = 1/2 spin state. Thus,most of the iron(II1) exists as EPR silent species at 77 K. Ferritin samples are also generally EPR silent even though the core has amagnetic moment (29). Optical absorption measurementsat 420 nm indicated a molar absorptivity/iron of 322 cm" M" for the solution of the polynuclear Fe(II1) .ATP (4:l)complex, which is close to the value of 362 cm" M" for a ferritin sample containing an average of 1400 Fe atoms/protein molecule. The solutions of polynuclear Fe(II1)-ATP (41) complex arestable,lasting several months before any precipitation is evident at iron concentrations of 12 mM; similar solutions with an iron concentration as greatas 45 mM have been prepared. To assess the approximate size of the polynuclear iron (111)ATP (4:l) cluster,ultrafiltration and dialysis experiments were carried out using membranes of various types with^ a range of selective porosities (see "Materials and Methods"). Although membrane separationsarenot expected to give accurate molecular weights since separation is influenced by factors such as charge and shape as well as size, their use nevertheless gives some indication of the molecular weight range of the cluster. Nearly complete retention of the iron
was observed with a membrane retaining molecules 2500, whereas nominally 5% of the iron passed through the membrane which selected for MI > 10,000. The value increased to 18 f 2% when a membrane retaining molecules of M I 2 50,000was used.Although the datasuggest some polydispersity of size, it is clear that most (282%) of the iron occurred in clusters with approximate molecular weights in excess of 50,000;a M,of 50,000corresponds to approximately 250 iron atoms/cluster. The polynuclear Fe(III).ATP (4:l)complexwas further characterized by measuring the number of protons released per iron(II1) bound at pH7 and 9. A value of 2.5 k 0.1 H'/Fe from four measurements was obtained, which is in good agreement with proton release measurements for the formation of other polynuclear iron complexes having the formula [Fe(III)(OH)gF5] (30). Anionic chelators such as citrate or carboxylate groups in dextran have been shown to provide the necessary charge stabilization in such polynuclear ironoxy clusters; presumably phosphate plays such a role in the Fe(111) ATP complex. In order to determine which parts of the ATP molecule are important in forming the soluble iron complexes, tripolyphosphate, pyrophosphate, orthophosphate, AMP, and ribose 5phosphate were individually tested at Fe:ligand ratios of 4; none formed soluble complexes with iron(II1). However, solutions of equimolar amounts of AMP and pyrophosphate or dATP were readily substituted for ATP in forming soluble complexes, suggesting that the adenine moiety facilitated cluster formation? Infraredspectra of the polynuclear Fe(II1) .ATP (4:l)complex and of free ATP showed that iron complexation caused a 60% decrease in intensity of the P-0 stretch at 1250 cm" and a shift in the 1700 cm" bending mode of the -NH2 group of adenine to beneaththe -OH bending mode at 1650 cm", suggesting the involvement of both of these groups in iron binding. Coordination through the 3"oxygen of ribose is also possible. Structural Properties of the Polynuclear Fe(III)-ATP (4:l) Complex Determined by X-ray Absorption Spectroscopy-In order to determine the site of P in the Fe(III).ATP (4:l) complex and to confirm the polynuclear iron environment suggested by the proton release, NMR and EPR studies (Fig. When equimolar concentrations of ribose 5-phosphate and pyrophosphate are employed (iron:ligand = 41,a precipitate forms which slowly dissolves over a 48-h period.
Fe(III,, ATP Complexes
7978
l), the XANES and EXAFS for iron were compared for the ATP complex, native ferritin, and FePO,. 2H20. From the stoichiometry of 1 (FeO-OPO3H2):8Fe0.OH in ferritin, we expect the contribution of Fe-P interactions to be sufficiently low that the x-ray absorption spectrum canbe considered to derive only from Fe-0 and Fe-Fe interactions,for the purpose of the EXAFS analyses in this study (see also Ref. 20). Visual inspection of the XANES (Fig. 2) for the polynuclear Fe(III).ATP (4:l)complex compared to ferritin and FePO, indicates the presence of an octahedral environmentfor iron, with higher shell contributions apparent in the shape of the curve on the high energy side of the edge. Fourier transformation of the EXAFS data (Fig. 3)clearly shows the influence of atoms in the second shell for all three samples. However, the second shell is much more complex for ferritin and the polynuclear Fe(II1).ATP (4:l)complex than for FeP04.2H20. Data were analyzed using the first Fe-0 shell from each standard as a model to fit the data from the other standard as well as from the polynuclear Fe(II1)-ATP (4:l)complex. Consistentresults were obtained using eitherferritin or FeP04.2H20 asthe model. The iron in each standard is surrounded by six oxygen atoms at an average distance of 1.95 A (Table I). In contrast, regardless of the model used, coordination number and Debye-Waller factor ( A 2 , Table I) were always larger for the polynuclear Fe(II1)-ATP complex than for ferritin orFeP04. 2Hz0,suggesting more disorder or heterogeneity in theFe(II1)-ATP cluster. Such a result would be obtained if N in the purine ring and/or 0 from the sugar were coordinated to iron; alteration of the NH2 bending mode was indicated by the IR spectrum (see above) and the complexing properties of sugars for iron are well known (e.g. Ref. 30). Analysis of the second shell in the Fourier transform (Fig. 3) could best be accomplished by assuming the presence of two different atoms in the polynuclear Fe(II1).ATP (4:l) complex. Such a hypothesis is justified by three observations: the phase shift of the back transformed data from the second shell of the complex was distinct compared to ferritin (Fig. 44; the two-shell fit decreased the variance between the
200 100 W
0 3 0 e200 -I
a
$100 W
L o 2200 -I
W
5 0 0 0
4
0
FIG. 3. Fourier transforms and modeling of EXAFS spectra of Fe(II1)-ATP (4:l)and model compoundsand transforms of the data in Fig. 2. The transforms were performed with k weighting over a range of 3-10 A”. a, FeP04.2H20b, Fe(II1)-ATP (41);c, horse spleen ferritin.
TABLE I EXAFS parameters of polynuclear Fe(III).ATP (4:l) and model compounds FePOI.2H20 fit with ferritin as a model had 6.1 oxygen atoms at 1.95 A. Ferritin fit with FePOl.2Hz0 had 5.7 oxygen atoms at 1.99 A. The second shell of polynuclear Fe(II1) .ATP (41)was fitted with a model of four P at 3.33 8, [FePO,. 2HzO] and six Fe at 3.34 A (ferritin). Values of N and A 2 are certain to within 20%. A 2 is the value needed to achieve the best fit (see Ref.20) relative to the standard; positive values may berelated to more disorder and negative values to more order (see Ref. 20). Fe(II1).ATP N r Ad X lo+’ (4:l)
A
W
-.2(3 3 0.2+
-
First shell (0) Second shell (P) Second shell (Fe) Third shell (0)
7.7
1.3 3.8 44..100
1.95 2.53 f 0.02 3.27 f 0.02 3.36 f 0.03 f 0.03 -3.15
A 2
-6.19 -5.14
experimental and the fitted curves by a factor of 200- and 2500-fold compared to thesingle shell fits using Fe-Fe or Feo.ot P interactions alone, respectively (Fig. 4); and the Fourier a transforms of the actual data and the transforms generated from the model were verysimilar for both the magnitude and imaginary part of the data (Fig. 5). The analysis showed that the second shell of the polynuclear Fe(III).ATP (4:l)complex consisted of 1.3 (f0.3) P at 0.03.27 A and 3.8 (f0.7)Fe at 3.36 A (Table I). A third layer of - 1 0 atoms was detected at 4.00 A. The interatomic distances are in good agreement with Fe-Fe and Fe-0 distances for the 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 iron in ferritin and the metal-P distance in Mn(I1). ATP (27, 0 100 0 5 4.1 I O 15 E k(A 31). Several observations suggest that the interaction between FIG. 2. XANES and EXAFS of polynuclear Fe(II1)-ATP (4:l) and model compounds. Spectra were obtained on powders Fe(II1) and P represent average iron atomsdistributed [FeP04.2HzO] or solutions (ferritin, polynuclear Fe(II1)-ATP (41)) throughout the large (2250) polynuclear iron cluster, rather under “Materials and Methods.” Left, XANES; right, EXAFS: a and than only those iron atoms on the surface of the polynuclear b, FeP04.2 HzO; c and d, Fe(II1)-ATP (4:l);e and f , horse spleen ferritin. a, c, and e, note the “pip” preceding the main edge, charac- complex. First, the absence of any g’ = 4.3 signal in the EPR spectrum indicates that the solution isnot a mixture of teristic of octahedral ferric complexes (20). J
Q
e l b”
,
Fe(III) ATP Complexes
s- 0.02
7979
in ATP notwithstanding; consideration of hydrolysis is cogent not only because of the well known hydrolytic properties of W metal phosphate complexes (e.g. Refs. 32-34) but also because 0.0 0.0 either ATP or a mixture of AMP and pyrophosphate forms a soluble, polynuclear iron complex. 4 L 8-0.02 -0.02 The ability to detect both the Fe-P andFe-Fe interactions W in the EXAFS analysis of polynuclear Fe(II1).PO4complexes, 0.02 -20.0 ln such asthe polynuclear Fe(II1). ATP (4:l) complex described U W here, permits the examination of the influence of phosphate 0.0 -29.0 0 on the iron environment in initiationcomplexes formed from J apoferritin and small amounts of Fe(II1). More importantly, w-0.02 -28.0 the existence of soluble Fe(II1).ATP complexes at physiolog4 6 8 loo-( 4 6 8 10 ical pH suggests that such complexes could exist in uiuo and k(A may provide a means for intracellular iron transport after FIG. 4. Curve fitting the second shell of the Fe(III)*ATP iron dissociation from transferrin. Even if such iron nucleo(4:l)complex. a, single-shell fit using Fe-P model; c, single-shell fit tide triphosphate complexes do not function in normal iron using Fe-Fe model; b, two-shell fit with both Fe-P and Fe-Fe as transport, theexistence of iron nucleotide triphosphate commodels; note that the two-shell fit decreased the variance between plexes in abnormal conditions of iron excess as in transfuthe experimental and fitted curves by 200- to 2500-fold compared to the single-shell fits; d, a comparison of the second peak phase shift sional iron overload or hemochromatosis could alter the size plotted uersus k for horse spleen ferritin and polynuclear Fe(II1)-ATP of the pool of available nucleotide triphosphates and contribute to thetoxicity of iron. (4:l); --- ferritin, -Fe(II1)-ATP (4:l). 0.02
2-
8
$
1
1
1
1
1
1
1
1
1
1
1
200
200
9 100
100
W
-I
a
f o
0
W
F >- 1 0 0
-100
4 W
a -200
-200
.
.. -,
FIG. 5. Back Fourier transform using parameters derived from Of the EXAFS data Of the polynuclear Fe(III)*ATP4:l complex. The data used are in Table I. a, transform of the experimental EXAFS data; b, back transform produced using the two-shell model. The dotted lines are the imaginary part of the transform.
monoatomic Fe(II1) complexes and large polynuclear complexes. Second, no evidence be Obtained for unbound ATP using 31PNMR spectroscopy, although it ispossible that resonances from free ATP were relaxation-broadened beyond detection through ligand exchange. Finally, the cluster size is too large to account for Fe-PO4 interaction merely on the surface; at least 70% Of the Fe atoms have to be coordinated (through 0 ) to P in order to detect the Fe-P distance in the EXAFS analysis. Thus, thephosphate appears to be distributed throughout the Polynuclear iron cluster.3 Whether all of the phosphate in the polynuclear Fe(II1).ATP (4:l) complex is still ATPor whether some (metal catalyzed) hydrolysis had occurred during complex formation cannot be determined from the data, theeffect of iron on the IR of NH, It should be noted that in the native ferritin core, asubstantial amount of p also appears to reside in interior and/or inaccessible positions of the core (8) (see also discussion in Ref. 9).
Acknowledgments-We are grateful to Dr. Forest Rennick for helpful discussion andtoJoann Fish for help in preparing the manuscript. 1. 2. 3. 4. 5. 6. 7. 8. 9.
10. 11. 12. 13. 14. 15. 16. 17. 18.
19. 20.
REFERENCES Berner, R. A. (1973) Earth Plunet Sci. Lett 18.77-86 Juo, A. S. R., and Fox, R. L. (1971) Soil Sci. 124,370-376 Biederman, G., and Schindler, P. (1957) Acta Chem. Scand. 11,731-740 Grant, C. T., and Taborsky, G. (1966) Biochemistry 6,544-555 Rosenstein, R. W., and Taborsky, G. (1970) Biochemistry 9,649-657 Ihara, K.,Maeguchi,K.,Young, C. T., and Theil, E.C. (1984) J. Biol. Chem. 2 6 9 , 278-283 Harrison, P. M., Fischbach, F. A., Hoy, T. G., and Haggis, G. A. (1967) Nature 216,1188-1190 Granick, S., and Hahn, P. F. (1944) J. Biol. Chem. 166,661-669 Theil, E. C. (1983) in Advances in Inorganic Biochemistry (Theil, E.C., Eichhorn, G. L., and Marzilli, L. G., eds) Vol. 5, pp. 1-38, Elsevier/ North-Holland, New York Jones, T., Spencer, R., and Walsh, F. (1978) Biochemistry 17,4011-4017 Williams, J. M., Danson, D. P., and Janot, C. (1978) Phys. Med. Biol. 23, 835-851 Treffry, A. and Harrison, P. M. (1978) Biochem. J. 171,313-320 Treffy, A., Sowerby, J., and Harrison, P. M. (1979) FEES Lett. 1 0 0 , 3336 Sayers, D.E., Theil, E. C., and Rennick, F. J. (1983) J. Biol. Chem. 2 5 8 , 14076-14079 Chasteen, N. D., and Theil, E. C. (1982) J. Biol. Chem. 267,7672-7677 Massover, W. H., and Cowley, J. M. (1973) Proc. Natl. Acad. Sci. U. s. A. 70.3847-3851 Kalckar, J. H. M. (1947) J. Biol. Chem. 167,445-459 Sandell, E. B. (1959) ColorimetricDetermination of Truces of Metals, p. 524, Interscience Inc., New York Farrar, T. C., and Becker, N.D. (1971) Pulse and Fourier Transform NMR, pp. 20-22, Academic Press, New York Theil, E.C., Sayers, D. E., and Brown, M.A. (1979) J. Biol. Chem. 264,
Moore 8132-8134 p, B, (1966) Am, Mineral, 61 168176 Terro;, A., and Moreno, V. (1983) In&. Chim. Acta 80, 213-216 Goucher, C. R., and Taylor,J. F. (1964) J. Biol. Chem. 239,2251-2258 Bartlett G. R. (1976) Biochrm. BW h s Acta 70 1063-1970 Scullan& M.I., White, L. K., a n f d a s t e e n , N: D. (1982) J . M q n . Res. 47, 383-397 26. Sternlicht, H., Shulman, R. G., and Anderson, E. W. (1965) J . C k m . Phys. 43,3123-3132 27. Brown, F. F., Campbell, I. D.,Henson, R., Hirst, C. W. J., and Richards, R. E. (1973) Eur. J. Biochem. 38,54-58, and references therein 28. Kustin, K., and Swinehart, J. (1970) Prog. Inorg. Chm. 13, 107-153 f O ~ ; l ~ and ~ ~Spire, ~ ~T,,G., (1973) 21, 22. 23. 24 25:
ii: ~
~
~
~
~
g
~
Bwinor% Chep. 2,311-327, and references therein 31. Belli, M., cafatl, A., Blanconl, A., Burattini, E., Mobilio, S., Natoli C. R. Palladino, L., and Reale, A. (1983) Il N u 0 0 Cimento 2D(N5), 128i-1304 32. Jones, D. R., Londoy, L. F., and Sargeson, A. M. (1983) J . A ~them. , sot. 1 0 6 , 7327-7336 33. Harrowfield, J. M., Jones D. R., Lindoy, L. F., and Sargeson, A. M. (1980) J. Am. Chem. SOC.102.7733-7741 34. Taborsky G. (1983) in Advances in Inor anic Btochemistry (Theil, E. C. Eichhokn, G. L., and Marzilli, L. G., e%) Vol. 5, pp. 235-279, Elsevier: New York
~
