B/nch/m/ca¢t ~ a Acta, ! 115(1991} 145-150 1991 ~ Science PublishersB.V. All t'ighlsr e ~ ' e d 0304-4165/91/$03.50
145
BBAGEN 23625
Iron K-edge absorption spectroscopic investigations of the cores of ferritin and haemosiderins P a t r i c i a M a c k l e ~, C . D a v i d G a r n e r
~, R o b e r t a J. W a r d 2 a n d T i m o t h y J. P e t e r s 2
! Departnum~of C~.,~ni~ry, The Uniz~,sity. Manchester (U.KJ and 2 Department o f Clinical Bioclu'miffry. King's College School of Medicine and Dentistry, Uniz'ersityof London, London (U.IC)
(Rece~'ed 27 Augus! 1991)
Key words: Fernfin; l-laemosiderin,EXAFS The extended X-ray absorption fine structure (EXAFS) associated with the iron K-edge has been measured and interpreted for ferritin a n d haemosiderin extracted h ~ m horse spleen, and haemosiderin extracted from the livers of h u m a n s with treated primav8 haemochromatosis, a n d from the spleens of humans with treated secondary haemochromatosis. For ferritin, the data are consistmot with, on average, each iron atom being in a n environment comprised of approx, six ox)gen atoms a t 1.93 4- 0.02 A, approx. 1.5 irono atoms at 2.95 ___0.02 A and approx. 1.1 iron atoms at 3.39 _.+0.02 A, with a further shell of oxygens at approx. 3.6 A. Iron in horse spleen haemosiderin is in a n essentially identical local environment to t h a t in horse spleen ferritin. In contrast, the EXAFS data for primary haemechromatnsis haemosiderin indicate that the into-oxide core is amorphous; only a single shell of approx, six oxygen atoms at approx. 1.94_ 0.02 ~k being apparent. Secondary haemochromatosis haemosiderin shows an o r d e r ~ structure with approx. 1,4 iron atoms a t both 2.97 ___0.02 a n d 3.34 _ 0.02 .A. This arrangement of iron atoms is similar to that in horse spleen h a e m o s i d e ~ but the first oxygen shell is split with approx. 2.9 atoms at 1.90 _.+0.02 A and approx. 2.7 at 2.03 + 0.02 A, indicative of substantial structural differences between secondary haemorhcomatosis haemosiderin a n d horse spleen haemosiderin.
lalredaction Iron plays an essential role in many important biological processes such as oxygen transport, electron transfer, D N A synthesis and cellular growth and developmenL However, large quantifies of free iron cannot be stored under physiological conditions since iron(Ill) ions would rapidly precipitate as iron(Ill) oxide, from which iron would be poorly bioavailable. Therefore, nature has evolved two iron-storage proteins, ferritin and haemosiderin, in which iron can be sequestered and mobilised when required. The predominant iron protein, ferritin, has been extensively studied [1-6]. The molecules consist of a hollow proteinaceons shell which surrounds a core of iron(Ill) oxyhydroxide with smaller amounts of associated phosphate, leading to a composition which is typically approx. (FeOOH) s
Correspondence: C.D. Garner, Department of Chemistry,The University, Manchester,M13 9PL. U.K.
(FeOePO2H2). Previous extended X-ray absorption fine structure (EXAFS) data for ferritin [3-6] have indicated that iron in ferritin is bound to approximately six oxygen atoms at approx. 1.9 A and, beyond this shell, iron atoms are present an average distance of approx.
3.1
Clinical iron overload is due to either a recessively inherited abnormality in the intestinal uptake of iron, leading to enhanced absorption of dietary iron (primary haemochromatosis) [7] or to multiple blood transfusions for congenital anaemias, e.g., beta thalassaemia (secondary haemochromatosis) leading to progressive accumulation of iron as there is no effective excretory mechanism for excess tissue iron [8]. In both of these conditions the predominant form of stored iron is haemosiderin, and not ferritin as in normal individuals [9]. Little is known about the biochemical and biophysical properties of haemosiderin, but 57Fe M6ssbauer spectroscopy and electron diffraction [10,12] have shown that there are at least three different types of iron core, depending on the origin of the haemosi~erin.
146 Experimental procedures Biochemical methods
.,i~
Cadmium-free horse spleen ferritin (100 rag/roD was obtained from Sigma (Poole, U.K.). Haenmsiderin was extracted, from the spleens of horse and of human patients with treated secondary haenmchromato~is and from the livers of patients with treated primary haemochromatosis, by ultra-centrifugation c o m b i n e d with chaotrop~c solutions of high density potassium iodide [13]. X-ray absorption spectra were recorded in transmission mode at station 7.1 of Daresbury Synchrotron Radiation Source, operating at 2 G e V and an average current of 150 mA using a S ~ I I I ) monochromator. Unless stated otherwise, samples were maintained at a temperature of 80 K. EXAFS data were analysed using the single-scattering spherical wave approximation and phaseshifts were derived from ab initio calculations, as descn'bed elsewhere [14.15]. The ability of these phaseshifts to produce accurate F e - O and F e - F e distances in EXAFS simulations was checked by measurements of the iron K-edge EXAFS for haematite (a-Fe,O3) [16], magnetite ( F e 3 0 ~) [17] and y-Fe20~ [18]; interpretation of these spectra gave distances in excellent agreement with those determined by diffraction methods. Results a n d Discuss,on
Ferritin Fig. 1 shows the k3-weighted EXAFS associated with the iron K-absorption edge of horse-spleen fer-
o 3
4
5
6
7
~
,iJ (?.2
X-ray absorption spectroscopy
9
:0
TT
:2
13 14
15
k (Aq)
Fig. 1. Iron K-edge k3-weighted EXAFS of horse spleen ferrilin recorded at 80 K ( ) compared to the simulationof these data by backseattering from atoms with parameters as summarised in Table I ( - - - - - - ) .
j
\
.?
Radiai D~tar~-e~ ) Fig. 2. Fourier transformsof data presented in Fig. L ritin and the corresponding Fourier transform is d~played in F i ~ Z There is a strong F_~LAFS signal out to approx. 10 A - t in k-space where it suddenly damps. This profile is consistent with the results of other investigations [3-6]. Electronic spectra [19,20] hm~ suggested octahedral oxygen coordination about iron in ferritin, so the first shell of atoms about the iron was fitted with six oxygens at approx. 1.93 ~ . With no other independent information available for the structure of the outer shells, the difference method was used to analyse the nature of their contn'butions to the EXAFS. The calculated EXAFS from six oxygens at 1.93 was subtracted from the experimental data to give a residual spectrum representing only the backscattering from the outer shells. Fig. 3 shows the experimental and the best fit to this difference spectrum with a split shell of 1_5 iron atoms at 2.95 A and 1.1 iron atoms at 3_39 ~ . It is the destructive interference of the [3~'~kscattering from these two iron shells which damp the high k-region. At low k in the difference spectrum there is a poor fit between the calculated and the experimental data which suggests the presence of another low Z backscatterer (oxygen). When such a shell is included at approx. 3.6 A the simulation of the low k-region is improved significantly (Fi~ 4). This backscattering is approximately ~ out of phase with that from the iron shell at 3_39 A and, therefore, the two contributions to the EXAFS are highly correlated as seen from the change in the occupancy of the iron shell, from 1.1 to 4.1, upon inclusion ofothe backscattering from a shell of six oxygens at 3.57 A. Consequently, the results only indicate the presence of oxygen at approx. 3.6 A (Table I). Thus, we agree with the conclusions of pre~ous investigations [3-6] that iron in ferritin is bound to six
147 oxygen atoms at 1.93 + 0.02 ,A. Also, we concur with the latest [5,6] interpretations of a split iron shell centred at approx. 3.1 ~,, rather than a single iron shell as reported previously [3,4]. However, by improving the sigual-to-noise ratio of the data and increasing the range over which data were collected, from 10 to 15 ,~-I, we have been able to provide a more accurate assessment of the F e - F e distances im'olved. The distances favoured from our results, viz. 2.95 + 0.02 and 30.39 + 0.02 ,~, as o 0 ~ to 3.00 + 0.02 and 3.52 + 0.02 A (or approx. 3.83 A if oxygen rather than iron) [5,6], are important in the context of the medium-range atomic order in ferritin. The total occupation number o f the iron shell at approx. 2.6 is lower than that (approx. 6.5) previously reported [5,6]. The signifcance of this difference is not clear, as occupation numbers are usually not well determined by analysis of EXAFS and the effects o f the inclusion of backscattering from the oxygen shell at approx. 3.6 A noted above. Other reasons for this difference could include our more extended data range a n d / o r differences in the procedures adopted for data analysis. EXAFS measurements were performed on liquid and freeze-dried ferritln at room temperature and rapidly frozen to approx. 80 K by immersion in liquid nitrogen, to investigate the possibility of a change in structure o f the Fe core with temperature a n d / o r physical state. Apart from the slightly reduced amplitude at room temperature, there was no change in the EXAFS signal, indicating that the structure of the Fe core is invariant with a change in these conditions. This conclusion contrasts with the results of Heald et al. [4] who reported the iron K-edge EXAFS of ferfitin at 293, 210, 160, 115 and 80 IC They found that the F e - F e distance o f approx. 3.3 ~, increased slightly with decreasing temperature. As observed above, the interpretation of the F e - F e separations from the EXAFS data recorded by Heald et al. [4] should he viewed with caution in the light of the improvement achieved in the studies reported herein. Heald et al. also argued that the level o f disorder of the inner oxygen shell increases considerably on cooling. These interpretations led them
3
2
I
_! -2 -3 3
to
5
6
7
8
9
ii k (A-l)
10
12
13
11,
15
Fig. 3. Differenogof the IroB K-PxJgek3-w¢ightedEXAFSof horsespleen ferritin recorded at 80 K less the EXAFS from six oxygen atoms at 1.93 P, ( ), compared to the EXAFS produced by back.q:aneringfor 1.5 Fe atoms at 2.95 A (2o.2 =0.018 ,~2) and I.i Fe atomsat 3.39 A (2o,z = 0.011 A2) (-- - - --).
to propose that a phase change occured on cooling. If this was the case, the difference between our conclusions and theirs may have its origin in the rapid cooling of our sample compared to the slow cooling of ferritin by Heaid et al. Horse-spleen haemosiderin
The EXAFS spectra of horse-spleen haemosiderin is virtually indistinguishable from that of ferritin which suggests that the short- and medium-range order of the iron core is essentially the same in both (see Table !). 57Fe MOssbauer studies have suggested that the iron oxide cores of ferritin and horse spleen haemosiderin are structurally similar [10-12]. Both exhibit superparamagnetic behaviour with haemosiderin having a lower blocking temperature, consistent with a smaller core size as confirmed by electron microscopy [11].
TABLE ! Com~rison of the parameters obCair,ed * * from the mudy~ of the EXAFS associated wilh the iron K-edgeof horse spleenferritin and haemosiderin
Eo = 2135 eV; energyrange 13-847 eV. Errors in bond distancesare consideredto be +0.02 A Atom O Fe Fe • • *
Ferrifin
Haemosiderin
N
R (,A)
2o'z (Az) *
N
R (A)
2
145
BBAGEN 23625
Iron K-edge absorption spectroscopic investigations of the cores of ferritin and haemosiderins P a t r i c i a M a c k l e ~, C . D a v i d G a r n e r
~, R o b e r t a J. W a r d 2 a n d T i m o t h y J. P e t e r s 2
! Departnum~of C~.,~ni~ry, The Uniz~,sity. Manchester (U.KJ and 2 Department o f Clinical Bioclu'miffry. King's College School of Medicine and Dentistry, Uniz'ersityof London, London (U.IC)
(Rece~'ed 27 Augus! 1991)
Key words: Fernfin; l-laemosiderin,EXAFS The extended X-ray absorption fine structure (EXAFS) associated with the iron K-edge has been measured and interpreted for ferritin a n d haemosiderin extracted h ~ m horse spleen, and haemosiderin extracted from the livers of h u m a n s with treated primav8 haemochromatosis, a n d from the spleens of humans with treated secondary haemochromatosis. For ferritin, the data are consistmot with, on average, each iron atom being in a n environment comprised of approx, six ox)gen atoms a t 1.93 4- 0.02 A, approx. 1.5 irono atoms at 2.95 ___0.02 A and approx. 1.1 iron atoms at 3.39 _.+0.02 A, with a further shell of oxygens at approx. 3.6 A. Iron in horse spleen haemosiderin is in a n essentially identical local environment to t h a t in horse spleen ferritin. In contrast, the EXAFS data for primary haemechromatnsis haemosiderin indicate that the into-oxide core is amorphous; only a single shell of approx, six oxygen atoms at approx. 1.94_ 0.02 ~k being apparent. Secondary haemochromatosis haemosiderin shows an o r d e r ~ structure with approx. 1,4 iron atoms a t both 2.97 ___0.02 a n d 3.34 _ 0.02 .A. This arrangement of iron atoms is similar to that in horse spleen h a e m o s i d e ~ but the first oxygen shell is split with approx. 2.9 atoms at 1.90 _.+0.02 A and approx. 2.7 at 2.03 + 0.02 A, indicative of substantial structural differences between secondary haemorhcomatosis haemosiderin a n d horse spleen haemosiderin.
lalredaction Iron plays an essential role in many important biological processes such as oxygen transport, electron transfer, D N A synthesis and cellular growth and developmenL However, large quantifies of free iron cannot be stored under physiological conditions since iron(Ill) ions would rapidly precipitate as iron(Ill) oxide, from which iron would be poorly bioavailable. Therefore, nature has evolved two iron-storage proteins, ferritin and haemosiderin, in which iron can be sequestered and mobilised when required. The predominant iron protein, ferritin, has been extensively studied [1-6]. The molecules consist of a hollow proteinaceons shell which surrounds a core of iron(Ill) oxyhydroxide with smaller amounts of associated phosphate, leading to a composition which is typically approx. (FeOOH) s
Correspondence: C.D. Garner, Department of Chemistry,The University, Manchester,M13 9PL. U.K.
(FeOePO2H2). Previous extended X-ray absorption fine structure (EXAFS) data for ferritin [3-6] have indicated that iron in ferritin is bound to approximately six oxygen atoms at approx. 1.9 A and, beyond this shell, iron atoms are present an average distance of approx.
3.1
Clinical iron overload is due to either a recessively inherited abnormality in the intestinal uptake of iron, leading to enhanced absorption of dietary iron (primary haemochromatosis) [7] or to multiple blood transfusions for congenital anaemias, e.g., beta thalassaemia (secondary haemochromatosis) leading to progressive accumulation of iron as there is no effective excretory mechanism for excess tissue iron [8]. In both of these conditions the predominant form of stored iron is haemosiderin, and not ferritin as in normal individuals [9]. Little is known about the biochemical and biophysical properties of haemosiderin, but 57Fe M6ssbauer spectroscopy and electron diffraction [10,12] have shown that there are at least three different types of iron core, depending on the origin of the haemosi~erin.
146 Experimental procedures Biochemical methods
.,i~
Cadmium-free horse spleen ferritin (100 rag/roD was obtained from Sigma (Poole, U.K.). Haenmsiderin was extracted, from the spleens of horse and of human patients with treated secondary haenmchromato~is and from the livers of patients with treated primary haemochromatosis, by ultra-centrifugation c o m b i n e d with chaotrop~c solutions of high density potassium iodide [13]. X-ray absorption spectra were recorded in transmission mode at station 7.1 of Daresbury Synchrotron Radiation Source, operating at 2 G e V and an average current of 150 mA using a S ~ I I I ) monochromator. Unless stated otherwise, samples were maintained at a temperature of 80 K. EXAFS data were analysed using the single-scattering spherical wave approximation and phaseshifts were derived from ab initio calculations, as descn'bed elsewhere [14.15]. The ability of these phaseshifts to produce accurate F e - O and F e - F e distances in EXAFS simulations was checked by measurements of the iron K-edge EXAFS for haematite (a-Fe,O3) [16], magnetite ( F e 3 0 ~) [17] and y-Fe20~ [18]; interpretation of these spectra gave distances in excellent agreement with those determined by diffraction methods. Results a n d Discuss,on
Ferritin Fig. 1 shows the k3-weighted EXAFS associated with the iron K-absorption edge of horse-spleen fer-
o 3
4
5
6
7
~
,iJ (?.2
X-ray absorption spectroscopy
9
:0
TT
:2
13 14
15
k (Aq)
Fig. 1. Iron K-edge k3-weighted EXAFS of horse spleen ferrilin recorded at 80 K ( ) compared to the simulationof these data by backseattering from atoms with parameters as summarised in Table I ( - - - - - - ) .
j
\
.?
Radiai D~tar~-e~ ) Fig. 2. Fourier transformsof data presented in Fig. L ritin and the corresponding Fourier transform is d~played in F i ~ Z There is a strong F_~LAFS signal out to approx. 10 A - t in k-space where it suddenly damps. This profile is consistent with the results of other investigations [3-6]. Electronic spectra [19,20] hm~ suggested octahedral oxygen coordination about iron in ferritin, so the first shell of atoms about the iron was fitted with six oxygens at approx. 1.93 ~ . With no other independent information available for the structure of the outer shells, the difference method was used to analyse the nature of their contn'butions to the EXAFS. The calculated EXAFS from six oxygens at 1.93 was subtracted from the experimental data to give a residual spectrum representing only the backscattering from the outer shells. Fig. 3 shows the experimental and the best fit to this difference spectrum with a split shell of 1_5 iron atoms at 2.95 A and 1.1 iron atoms at 3_39 ~ . It is the destructive interference of the [3~'~kscattering from these two iron shells which damp the high k-region. At low k in the difference spectrum there is a poor fit between the calculated and the experimental data which suggests the presence of another low Z backscatterer (oxygen). When such a shell is included at approx. 3.6 A the simulation of the low k-region is improved significantly (Fi~ 4). This backscattering is approximately ~ out of phase with that from the iron shell at 3_39 A and, therefore, the two contributions to the EXAFS are highly correlated as seen from the change in the occupancy of the iron shell, from 1.1 to 4.1, upon inclusion ofothe backscattering from a shell of six oxygens at 3.57 A. Consequently, the results only indicate the presence of oxygen at approx. 3.6 A (Table I). Thus, we agree with the conclusions of pre~ous investigations [3-6] that iron in ferritin is bound to six
147 oxygen atoms at 1.93 + 0.02 ,A. Also, we concur with the latest [5,6] interpretations of a split iron shell centred at approx. 3.1 ~,, rather than a single iron shell as reported previously [3,4]. However, by improving the sigual-to-noise ratio of the data and increasing the range over which data were collected, from 10 to 15 ,~-I, we have been able to provide a more accurate assessment of the F e - F e distances im'olved. The distances favoured from our results, viz. 2.95 + 0.02 and 30.39 + 0.02 ,~, as o 0 ~ to 3.00 + 0.02 and 3.52 + 0.02 A (or approx. 3.83 A if oxygen rather than iron) [5,6], are important in the context of the medium-range atomic order in ferritin. The total occupation number o f the iron shell at approx. 2.6 is lower than that (approx. 6.5) previously reported [5,6]. The signifcance of this difference is not clear, as occupation numbers are usually not well determined by analysis of EXAFS and the effects o f the inclusion of backscattering from the oxygen shell at approx. 3.6 A noted above. Other reasons for this difference could include our more extended data range a n d / o r differences in the procedures adopted for data analysis. EXAFS measurements were performed on liquid and freeze-dried ferritln at room temperature and rapidly frozen to approx. 80 K by immersion in liquid nitrogen, to investigate the possibility of a change in structure o f the Fe core with temperature a n d / o r physical state. Apart from the slightly reduced amplitude at room temperature, there was no change in the EXAFS signal, indicating that the structure of the Fe core is invariant with a change in these conditions. This conclusion contrasts with the results of Heald et al. [4] who reported the iron K-edge EXAFS of ferfitin at 293, 210, 160, 115 and 80 IC They found that the F e - F e distance o f approx. 3.3 ~, increased slightly with decreasing temperature. As observed above, the interpretation of the F e - F e separations from the EXAFS data recorded by Heald et al. [4] should he viewed with caution in the light of the improvement achieved in the studies reported herein. Heald et al. also argued that the level o f disorder of the inner oxygen shell increases considerably on cooling. These interpretations led them
3
2
I
_! -2 -3 3
to
5
6
7
8
9
ii k (A-l)
10
12
13
11,
15
Fig. 3. Differenogof the IroB K-PxJgek3-w¢ightedEXAFSof horsespleen ferritin recorded at 80 K less the EXAFS from six oxygen atoms at 1.93 P, ( ), compared to the EXAFS produced by back.q:aneringfor 1.5 Fe atoms at 2.95 A (2o.2 =0.018 ,~2) and I.i Fe atomsat 3.39 A (2o,z = 0.011 A2) (-- - - --).
to propose that a phase change occured on cooling. If this was the case, the difference between our conclusions and theirs may have its origin in the rapid cooling of our sample compared to the slow cooling of ferritin by Heaid et al. Horse-spleen haemosiderin
The EXAFS spectra of horse-spleen haemosiderin is virtually indistinguishable from that of ferritin which suggests that the short- and medium-range order of the iron core is essentially the same in both (see Table !). 57Fe MOssbauer studies have suggested that the iron oxide cores of ferritin and horse spleen haemosiderin are structurally similar [10-12]. Both exhibit superparamagnetic behaviour with haemosiderin having a lower blocking temperature, consistent with a smaller core size as confirmed by electron microscopy [11].
TABLE ! Com~rison of the parameters obCair,ed * * from the mudy~ of the EXAFS associated wilh the iron K-edgeof horse spleenferritin and haemosiderin
Eo = 2135 eV; energyrange 13-847 eV. Errors in bond distancesare consideredto be +0.02 A Atom O Fe Fe • • *
Ferrifin
Haemosiderin
N
R (,A)
2o'z (Az) *
N
R (A)
2
