Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`JOURNAL OF
`
`Inorganic
`Biochemistry
`
`www.elsevier.com/locate/jinorgbio
`
`Structure of carbohydrate-bound polynuclear iron oxyhydroxide
`nanoparticles in parenteral formulations
`
`Dina S. Kudasheva a,b, Jriuan Lai a,b,c, Abraham Ulman a,b,c, Mary K. Cowman a,b,*
`
`a Othmer Department of Chemical and Biological Sciences and Engineering, Polytechnic University, Six Metrotech Center, Brooklyn, NY 11201, USA
`b Herman F. Mark Polymer Research Institute, Polytechnic University, Six Metrotech Center, Brooklyn, NY 11201, USA
`c The NSF MRSEC for Polymers at Engineered Interfaces, Polytechnic University, Six Metrotech Center, Brooklyn, NY 11201, USA
`
`Received 21 January 2004; received in revised form 28 April 2004; accepted 11 June 2004
`Available online 11 September 2004
`
`Abstract
`
`Intravenous iron therapy is used to treat anemia associated with chronic kidney disease. The chemical structures of parenteral
`iron agents have not been characterized in detail, and correlations between structure, efficiency of iron delivery, and toxicity via
`catalysis of oxygen-derived free radical creation remain to be established. In this study, two formulations of parenteral iron have
`been characterized by absorption spectroscopy, X-ray diffraction analysis (XRD), transmission electron microscopy (TEM), atomic
`force microscopy (AFM), and elemental analysis. The samples studied were VenoferÒ (Iron Sucrose Injection, USP) and FerrlecitÒ
`(Sodium Ferric Gluconate in Sucrose Injection). The 250–800-nm absorption spectra and the XRD patterns showed that both for-
`mulations contain a mineral core composed of iron oxyhydroxide in the b-FeOOH mineral polymorph known as akaganeite. This
`was further confirmed for each formulation by imaging using TEM and AFM. The average core size for the nanoparticles, after
`dialysis to remove unbound or loosely bound carbohydrate, was approximately 3 ± 2 nm for the iron–sucrose, and approximately
`2 ± 1 nm for the iron–gluconate. Each of the nanoparticles consists of a mineral core, surrounded by a layer of bound carbohydrate.
`The overall diameter of the average bead in the dialyzed preparations was approximately 7 ± 4 nm for the iron–sucrose, and 3 ± 1
`nm for the iron–gluconate. Undialyzed preparations have particles with larger average sizes, depending on the extent of dilution of
`unbound and loosely bound carbohydrate. At a dilution corresponding to a final Fe concentration of 5 mg/mL, the average particle
`diameter in the iron–sucrose formulation was approximately 22 ± 9 nm, whereas that of the iron–gluconate formulation was
`approximately 12 ± 5 nm.
`Ó 2004 Elsevier Inc. All rights reserved.
`
`Keywords: Parenteral formulations; Iron oxyhydroxide; Iron–carbohydrate complex; AFM; Iron–sucrose injection; Sodium ferric gluconate;
`Akaganeite
`
`1. Introduction
`
`Parenteral iron formulations containing iron com-
`plexed to monosaccharides or disaccharides are among
`the products currently used for the treatment of anemia.
`
`* Corresponding author. Tel.: +1 718 260 3054; fax: +1 718 260
`3125.
`E-mail address: mcowman@poly.edu (M.K. Cowman).
`
`0162-0134/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved.
`doi:10.1016/j.jinorgbio.2004.06.010
`
`Yet, the chemical structures of these materials have not
`been characterized in detail, and correlations between
`structure and complex stability, efficiency of iron deliv-
`ery, and toxicity via catalysis of oxygen-derived free rad-
`ical creation remain to be established. This study
`represents a first step toward complete characterization
`of the structures of this subclass of modern therapeutic
`iron–carbohydrate compounds.
`Iron(III) oxyhydroxide is formed in aqueous solu-
`tions by the dissolution of a suitable iron salt, such as
`ferric chloride, at low pH, followed by neutralization
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1052 - Page 1
`
`

`

`1758
`
`D.S. Kudasheva et al. / Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`or basification [1–3]. The aquated iron ions are hydro-
`lyzed initially to the mononuclear iron hydroxides
`Fe(OH)(H2O)5 and Fe(OH)2(H2O)4. Formation of
`Fe(OH)3(H2O)3 leads to the growth of a polynuclear
`(containing multiple linked iron atoms) iron oxyhydrox-
`ide with the approximate composition FeOOH. This
`mineral is insoluble unless stabilized by the presence of
`a water-soluble chelating agent.
`Carbohydrates can be appropriate chelating agents for
`the stabilization of iron oxyhydroxide nanoparticles in
`colloidal suspension. Certain low molecular mass neutral
`carbohydrates, such as sucrose [4], or fructose [5] present
`multiple hydroxyl groups in a suitable array to chelate
`iron, although the binding is inherently weak in neutral
`aqueous solution. Polymeric carbohydrates such as dex-
`tran may also be used to stabilize nanoparticles because
`they present a large number of hydroxyl groups, which
`can cooperatively chelate the surface of iron oxyhydrox-
`ide nanoparticles. For a neutral carbohydrate, the chela-
`tion to iron is enhanced at high pH, because the hydroxyl
`groups may become deprotonated, thus acquiring a neg-
`ative charge, and interacting more strongly with the cati-
`onic iron ion [6]. At neutral pH,
`inherently anionic
`carbohydrates such as gluconate or partially oxidized
`polysaccharides are more effective nanoparticle stabiliz-
`ers. In these cases, a carboxyl group provides the negative
`charge over a broad pH range. Its chelation to iron can
`also drive the deprotonation of nearby hydroxyl groups,
`further enhancing the complex stability [4,7].
`The structures of complexes formed between iron
`nanoparticles and polysaccharides have been studied
`previously, because iron–dextran complexes [8,9] and
`iron–polymaltosate complexes [10–12] (where polymal-
`tosate was the commercial name for partially hydro-
`lysed and oxidized starch) have been used for
`decades in the treatment of anemia. In a number of
`cases, the nanoparticle cores have been identified as
`iron oxyhydroxide. X-ray diffraction data are most
`consistent with the akaganeite (b-FeOOH) polymorph
`(polymorphs are solids with the same chemical com-
`position but different crystal structures) [13–18]. Ultra-
`violet–visible
`(UV–VIS)
`absorption
`spectroscopy
`[5,17–19] and Mo¨ ssbauer spectroscopy [13–16,20] have
`shown the presence of octahedrally coordinated high-
`spin Fe(III)
`ions. Extended X-ray absorption fine
`structure (EXAFS) studies [14] confirm the coordina-
`tion of iron by 6 oxygen atoms, at an Fe–O distance
`of 1.95 A˚ , and the location of a somewhat disordered
`shell of iron ions at a distance of about 3.05 A˚ . The
`iron oxyhydroxide crystallite dimensions estimated
`from the broadening of X-ray diffraction peaks are
`generally about 1–5 nm in diameter [9,14,17]. Electron
`microscopy has shown that the dimensions of the core
`(which may be larger than the size of the crystalline
`portion) can range from 3 nm diameter spheres [9]
`to ellipsoidal particles with dimensions of up to
`
`5 · 34 nm [21]. The mineral core is stabilized by inter-
`actions with the polysaccharide, presumably through
`the hydroxyl and/or carboxyl groups of the carbohy-
`drate units, the latter having been generated by partial
`oxidation of the polysaccharide. The overall particle
`size is highly dependent on the nature of the polysac-
`charide and the mode of complex formation. An older
`commercial
`iron–dextran complex had a reported
`overall diameter of about 12–13 nm [9], whereas iron
`complexes with j-carrageenan are reported to form
`25–50 nm aggregates of 10–20 nm particles [17,18].
`The apparent complex molecular mass, based on elu-
`tion position in gel permeation chromatography
`(GPC) chromatograms, may range from 72 kDa
`(ImferonÒ)
`(InFedÒ)
`to 90 kDa
`to 265 kDa
`(DexferrumÒ).
`Information about the structure of iron complexes
`with low molecular mass carbohydrates has been pre-
`dominantly derived from studies of soluble mono- or
`di-nuclear iron complexes. Weakly stable mononuclear
`complexes are formed by carbohydrates with struc-
`tures that allow at least
`three hydroxyl groups to
`interact simultaneously with a single iron ion. For
`example, sucrose or its component sugar fructose
`can bind iron weakly at neutral pH [4–6]. There is
`much less information about complexes containing
`polynuclear iron species complexed with monosaccha-
`rides or disaccharides. Studies of ferric fructose sys-
`tems suggested that the fructose may serve primarily
`to coat iron-containing particles at neutral pH. Inter-
`estingly, when the pH is raised, the formation of a tet-
`ra-deprotonated sugar moiety was proposed to break
`down the polynuclear iron to the mononuclear ferric
`fructose complexes. It was also noted that gluconate,
`with its carboxyl group, was much more effective in
`breaking down the particles [5]. For iron sucrose, ear-
`lier studies suggest the presence of polynuclear iron
`complexed with sucrose,
`in which the mineral core
`has been proposed to be a 2-line ferrihydrite [22]
`rather than an iron oxyhydroxide.
`Some of the questions that may arise in the analy-
`sis of polynuclear iron–carbohydrate complexes in-
`clude:
`(1)
`the composition and polymorphic form
`and degree of crystallinity of the iron mineral compo-
`nent, (2) the size and shape of the nanoparticles, (3)
`the stability of the nanoparticles, (4) the location of
`the saccharide component within the particle, (5) the
`molar ratio of iron to saccharide, and (6) the mode
`of binding between the iron and saccharide compo-
`nents,
`including the extent of saccharide deprotona-
`tion in the complex. In the present work, we address
`some of these questions by examining two different
`iron–carbohydrate complexes with respect to the min-
`eral environment of the iron, the nanoparticle size,
`and the particle stability upon dilution or separation
`from unbound carbohydrate by dialysis.
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1052 - Page 2
`
`

`

`D.S. Kudasheva et al. / Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`1759
`
`2. Materials and methods
`
`The parenteral iron formulations used in this study
`were an iron–sucrose (VenoferÒ, Iron Sucrose Injection,
`USP, lots 1644 and 2323A), from American Regent,
`Inc., Shirley, NY, and an iron–gluconate in sucrose
`(FerrlecitÒ, Sodium Ferric Gluconate Complex in Su-
`crose Injection, lots 1F541 and 1C239), from Watson
`Pharmaceuticals, Inc., Corona, CA. VenoferÒ was sup-
`plied as the iron–sucrose complex in 5 mL vials contain-
`ing 100 mg of elemental iron in 30% sucrose solution (20
`mg Fe/mL) at a pH of about 10.5–11.1. The apparent
`average molecular mass is stated in the package insert
`to be 34–60 kDa. FerrlecitÒ was obtained as the iron–
`gluconate complex in 5 mL ampules containing 62.5
`mg of elemental iron as a sodium salt of a ferric gluco-
`nate complex in 20% sucrose (12.5 mg Fe/mL) at a pH
`of about 7.7–9.7. The apparent average molecular mass
`is stated in the package insert to be 350 ± 23 kDa. Some
`samples, where noted in the text, were subjected to dial-
`ysis to remove low molecular mass components. Dialysis
`of samples for absorption spectroscopy was performed
`against deionized water, at a pH of about 5.4, in Slide-
`A-Lyzer Dialysis Cassettes (Pierce, Rockford, IL) made
`of low-binding regenerated cellulose membrane, molecu-
`lar weight cutoff (MWCO) 10,000. Cassette capacity was
`0.5–3 mL. Larger volumes of dialyzed samples were
`needed for AFM and TEM studies; in that case dialysis
`against deionized water at neutral pH, or with pH ad-
`justed with KOH to match the sample pH, was per-
`formed in tubes made of Spectra/Por Molecularporous
`Regenerated Cellulose Membrane (Spectrum Laborato-
`ries
`Inc., Rancho Domingues, CA) with MWCO
`6–8,000. Dialysis time for both trials was 20–48 h. Dia-
`lyzed samples were sent to Huffman Laboratories, Inc.
`(Golden, CO) for elemental analysis to determine the
`amount of iron, carbon and sodium.
`Absorption spectra were recorded on a Perkin–Elmer
`Lambda 800 UV/VIS absorption spectrophotometer.
`The spectra were analyzed with UV WinLab software
`v3.00.03. The spectra were recorded in the range of
`800–200 nm for original non-diluted samples and for di-
`luted or dialyzed samples. This instrument has the abil-
`ity to analyze samples with relatively high absorbance.
`Its photometric linearity at an absorbance (A) of 3
`is ± 0.006 A units and at A of 2 is ± 0.002 A units.
`The photometric accuracy at A of 2 is ± 0.003 A units
`and stray light
`in the range of 220–370 nm is
`<0.00008% T. The photometric range is ± 7 A. For
`the undiluted samples, cylindrical quartz cells of path
`length 0.001 or 0.01 cm were employed; for all other
`samples a cylindrical quartz cell of path length 0.01
`cm was used. The reference was water. All spectra were
`recorded as absorbance vs. wavelength of light in nm.
`Molar extinction coefficients were calculated from the
`absorbances at wavelengths of 300 and 470 nm, based
`
`on the molar concentration of iron in each sample.
`The molar iron concentrations were calculated from
`the stated weight concentration of iron in each sample.
`X-ray powder diffraction (XRD) patterns were re-
`1 in the range
`corded at a scanning rate of 0.008° min
`2h from 0° to 70° using a Philips X-ray diffractometer
`with Cu Ka radiation (k = 1.5418 A˚ ). Original and dia-
`lyzed samples were prepared by freeze-drying prior to
`analysis. The average iron mineral crystallite size, D,
`was calculated from the observed XRD profiles using
`the peak full width at half maximum height (b) of the se-
`lected diffraction peak, using the Scherrer formula [23]:
`D ¼ Kk=b cos h;
`where K is the Scherrer constant, with a value of 0.94.
`For recorded crystalline patterns, diffraction angles
`and interplanar spacings obtained were compared with
`the values compiled in the JCPDS-ICDD (Joint Com-
`mittee on Powder Diffraction Standard-International
`Center for Diffraction Data-copyright PSI Interna-
`tional) reference cards.
`Electron microscopic studies of the iron-core crystal
`size and morphology for the iron–carbohydrate samples
`were carried out by using a Phillips CM-12 Transmis-
`sion Electron Microscope (100 keV). All preparations
`were deposited one night before imaging onto a carbon
`stabilized Formvar-coated copper grid (400 mesh), pur-
`chased from Ted Pella, Inc., Redding, CA. Approximate
`grid hole size is 42 lm, and the thickness of Formvar
`films stabilized with carbon is 5–10 nm. All attempts
`to image the original undiluted liquids were not success-
`ful, because the resulting film on a grid was too thick for
`electrons to penetrate. Consequently, the iron–carbohy-
`drate complexes were dialyzed and the grid was dipped
`into the sample and then the excess liquid was removed
`with a very gentle flow of compressed nitrogen.
`Atomic force microscopy (AFM) was used to deter-
`mine size and shape of iron–carbohydrate complexes.
`The instrument used was a MultiMode Scanning Probe
`Microscope with Nanoscope IIIa controller and a type
`EV scanner (Veeco Instruments, Inc). The scanner was
`calibrated in the x–y plane using the 1 lm etched grid,
`and in the z-direction using the 200 nm height calibra-
`tion standard. The samples for imaging were prepared
`in two ways. In the first procedure, samples were diluted
`2 to 80-fold with deionized water prior to analysis. In
`the second procedure, samples were dialyzed against
`deionized water prior to analysis. In both cases, 4 lL
`of sample was put on a freshly cleaved mica surface,
`then after 20–30 s the surface was rinsed with 100 lL
`of deionized water to get rid of unbound particles, and
`the surface was dried by a gentle flow of nitrogen. The
`AFM imaging technique used was Tapping ModeTM.
`TESP etched silicon cantilever probes of 125 lm nomi-
`nal length were used, at a drive frequency of approxi-
`mately 240–280 kHz. We generally use an RMS (root
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1052 - Page 3
`
`

`

`1760
`
`D.S. Kudasheva et al. / Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`mean square) voltage of approximately 2.5–3 V, and ad-
`just the setpoint for optimum image, which is generally
`about 1 V below the RMS voltage. Both height and
`amplitude information were recorded at a scan rate of
`2–3 Hz, and stored in a 256 · 256 pixel format. Images
`were processed using the Nanoscope version 4.43r8 soft-
`ware. For optimum clarity in visual presentation of the
`nanoparticle images in the figures, flattening of first or-
`der was employed unless stated otherwise. For images to
`be used in measuring heights, the only image processing
`was zero order flattening. These images were analyzed to
`determine height and diameter of the particles observed.
`Although the height measurement is considered accu-
`rate, the apparent particle diameter is broadened by
`the finite dimensions of the probe tip. A correction for
`the tip broadening effect has been made following the
`procedure of Margeat et al. [24]. The apparent diameter
`measured at half height, S, is related to the real diame-
`ter, D, and the tip radius, R, according to the following
`
`equation [24]:
`S ¼ 2 RD þ D2
`4
`
`1=2
`
`:
`
`
`
`on the same day and with a single tip, and should be rep-
`resentative of the effect of dilution on the particle diam-
`eter. The relative sizes of particles in different iron
`formulations were confirmed by analysis of all samples
`on the same day, using a single tip, and running a cali-
`bration image of DNA on the same occasion.
`
`3. Results and discussion
`
`3.1. Absorption spectroscopy
`
`For Fe(III) in a high spin state octahedrally coordi-
`nated to oxygen, several characteristic absorption bands
`are expected. At 250–390 nm, a band of high intensity
`arises from oxo-metal charge transfer transitions, as
`has been reported for Fe(III)–glucose at 320 nm and
`for Fe(III)–sucrose at 370 nm [19]. Low intensity bands
`at 470–500 nm and 560–580 nm arise from d to d tran-
`sitions of octahedral d5 complexes of Fe(III). For low
`spin Fe(III), the latter bands would be of high intensity.
`The presence of Fe(II) would result in a band at 420–450
`nm for the d–d transition of octahedral d6 Fe(II), and a
`high intensity band at 800–900 nm for an aquated-Fe(II)
`complex [25].
`The spectral shapes of the studied parenteral iron for-
`mulations show a weak absorption band at about 470–
`500 nm and a strong band about 300 nm. Fig. 1 shows
`the absorption spectra for the two samples after dilution
`to the same nominal concentration of 5 mg Fe/mL, to
`facilitate spectral shape comparison. The 470–500 nm
`band is attributed to the d–d transition of octahedral
`high spin d5 Fe(III). The 300 nm band shows that the
`iron is chelated to oxygen, and is due to the oxo-metal
`
`The tip radius was estimated by imaging DNA, which
`has a real chain diameter of 2 nm. For one AFM tip
`used, the measured apparent diameter of DNA at half
`height was 11.8 nm, resulting in an estimated tip radius
`of 16.8 nm. A second tip had an estimated radius of 14.8
`nm. The appropriate tip radius was used to estimate cor-
`rected diameters in our images, although we may expect
`some minor variation for a given tip as a function of use.
`Thus the estimated particle diameters are subject to
`uncertainty on this basis. The relative diameters for a gi-
`ven series of samples differing in dilution were measured
`
`4.0
`
`3.5
`
`3.0
`
`2.5
`
`2.0
`
`1.5
`
`1.0
`
`0.5
`
`Absorbance
`
`0.0
`250
`
`300
`
`350
`
`400
`
`550
`500
`450
`Wavelength (nm)
`
`600
`
`650
`
`700
`
`750
`
`800
`
`Fig. 1. Absorption spectra of iron–sucrose (dashed line), and iron–gluconate (solid line) diluted to nominal iron concentrations of 5 mg/mL and
`analyzed in an optical cell of 0.01 cm path length.
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1052 - Page 4
`
`

`

`D.S. Kudasheva et al. / Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`1761
`
`Table 1
`Characterization of iron–carbohydrate complexes by absorption spectroscopy
`
`Sample
`
`Iron–sucrose
`
`Iron–gluconate
`
`Fe Conc. (mg/mL)
`
`Cell path length (cm)
`
`e300 (M
`
`1 cm
`
`1)
`
`e470 (M
`
`1 cm
`
`1)
`
`e300/e470
`
`20
`20
`5
`
`12.5
`12.5
`5
`
`0.001
`0.01
`0.01
`
`0.001
`0.01
`0.01
`
`2990
`
`2720
`
`2600
`
`2670
`
`380
`390
`
`290
`250
`
`7.9
`
`7.0
`
`9.0
`
`10.6
`
`charge transfer absorption band. There is no evidence
`for the existence of low spin state Fe(III) or for aquated
`Fe(II). The shapes of the spectra are essentially the same
`for both iron–carbohydrate samples; both formulations
`showed characteristic absorption spectra expected for
`Fe(III) complexes in an octahedral d5 high spin state.
`The position of the d–d transition band around 500
`nm has been shown to depend on the distance between
`two neighboring Fe atoms [26]. Thus, in the case of
`face-sharing octahedra of an iron oxide hematite, this
`band is above 500 nm. For the edge- and corner-sharing
`octahedra and hence larger Fe–Fe distances in iron oxy-
`hydroxides this transition is weaker and lies at shorter
`wavelengths of 470–499 nm [27]. Based on the fact that
`both iron formulations have this band at approximately
`the same position, we can conclude that the nature of
`the iron core is the same and both spectra are compati-
`ble with the expected ferric oxyhydroxide mineral form.
`The molar extinction coefficients of the absorption
`bands for each sample were closely similar (Table 1).
`The peak at 300 nm was best quantitated in undiluted
`samples analyzed in an optical cell with a path length
`of 0.001 cm. Calculated extinction coefficients at 300
`1 cm
`1 for the
`nm were approximately 2600–3000 M
`two samples. This compares with reported values of
`6660 for a different iron–sucrose preparation [19] and
`1700 for iron–carrageenan [17]. The shoulder at 470–
`
`500 nm was best quantitated in undiluted sample ana-
`lyzed in an optical cell with a 0.01 cm path length.
`The extinction coefficient at 470 nm was approximately
`1 cm
`1 for the two samples.
`300–400 M
`The stability of the iron core structure in each of the
`iron formulations was assessed by dilution and/or dialy-
`sis. Dilution to a concentration of 5 mg Fe/mL, or dial-
`ysis against water resulting in a similar dilution, had no
`significant effect on the spectral shape or molar extinc-
`tion coefficients of the iron absorption bands (Table 1).
`
`3.2. Transmission electron microscopy (TEM)
`
`TEM micrographs of dialyzed iron–sucrose and dia-
`lyzed iron–gluconate samples at equivalent magnifica-
`tions are compared in Fig. 2. The beadlike iron
`mineral cores appear to be fairly uniform in size for a gi-
`ven sample. This uniformity reflects the manufacturing
`processes, and results in particles with a narrow range
`of sizes. The comparison of the images of two samples
`shows that the iron–gluconate particles have cores that
`are generally slightly smaller (Fig. 2(a)) than those of
`iron–sucrose (Fig. 2(b)).
`Measurement of the diameters of the electron dense
`particles is used to estimate the diameter of the iron min-
`eral core in a given sample. For iron–sucrose, the results
`indicate a mineral core size distribution ranging from
`
`Fig. 2. Transmission electron microscopy image of (a) dialyzed iron–gluconate complex and (b) dialyzed iron–sucrose complex. The bar is 20 nm.
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1052 - Page 5
`
`

`

`1762
`
`D.S. Kudasheva et al. / Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`approximately 1.0–6.5 nm. The average size of the core is
`3 ± 2 nm (n = 20). For iron–gluconate, the mineral core
`diameter ranges from approximately 0.9–3.5 nm, and
`the average diameter of the core is 2 ± 1 nm (n = 20).
`
`3.3. X-ray powder diffraction
`
`The chemical structure of the iron core was evaluated
`by means of X-ray diffraction (XRD) analysis. XRD is
`considered to be the most reliable way to identify a par-
`ticular iron polymorph because it is based on the long
`range order of the atoms [3]. The diffractograms ob-
`
`tained for lyophilized samples of original, nondialyzed
`preparations were compared with reference data for
`different iron oxyhydroxide mineral phases, similar in
`composition and structure (Table 2). Both iron–
`carbohydrate samples give a crystalline pattern of a sin-
`gle phase whose peak positions and intensities match
`with reported JSPDS-ICDD reference data for iron
`(III) oxyhydroxide akaganeite b-FeOOH. The X-ray re-
`sults are not consistent with the ferrihydrite structure
`found in ferritin [28]. In addition, two peaks in the dif-
`fractograms of both samples in the small angle region
`(11.5° and 20°) belong to sucrose. From the analysis
`
`Table 2
`Interplanar spacings in Angstroms and relative strength of reflections (from JCPDS-ICDD cards for sucrose and iron minerals and from recorded
`difractograms for original nondialyzed formulations)
`
`Iron–sucrose (angle)
`7.59 (11.5°)
`
`Iron–gluconate (angle)
`7.54 (11.7°)
`
`6.91 (13.1°)
`
`4.68 (17.1°)
`
`3.61 (22.3°)
`
`3.61 (22.3°)
`
`2.64 (33.5°)
`
`2.56 (35.1°)
`
`2.34 (38.3°)
`2.28 (39.6°)
`
`1.76 (51.8°)
`
`1.64 (55.5°)
`1.52 (61.1°)
`
`1.49 (62.3°)
`
`2.69 (33.3°)
`
`2.555 (35.13°)
`
`2.35 (38.2°)
`
`1.75 (52.01°)
`
`1.66 (55.2°)
`
`1.50 (61.8°)
`1.49 (62.3°)
`
`Sucrose
`
`7.58s
`
`6.94m
`6.73m
`
`4.71s
`4.52s
`4.25m
`
`3.59s
`3.53m
`
`2.88w
`
`2.35m
`2.339w
`
`2.028w
`1.971w
`
`Hematite
`
`Goethite
`
`Ferrihydrite
`
`Akaganeite
`
`7.467s
`
`5.276m
`
`3.728w
`
`3.333s
`
`2.634m
`
`2.550s
`
`2.356w
`2.295m
`
`1.954m
`
`1.755w
`
`1.643m
`1.515w
`1.503w
`1.489w
`1.485w
`1.445m
`
`4.98w
`
`4.18s
`
`2.69m
`
`2.58w
`
`2.45s
`
`2.25w
`2.19m
`
`2.50s
`
`2.21s
`
`1.96s
`
`1.72m
`
`1.72m
`
`1.51m
`
`1.48s
`
`1.42w
`1.393w
`1.36w
`1.345w
`1.317w
`1.29w
`
`3.68m
`
`2.70s
`
`2.52s
`
`2.21m
`
`1.84m
`
`1.69s
`
`1.49m
`
`1.45m
`
`1.055w
`1.033w
`0.960w
`0.9581w
`0.8436w
`
`s, strong; m, medium; and w, weak.
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1052 - Page 6
`
`

`

`D.S. Kudasheva et al. / Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`1763
`
`220
`
`200
`
`180
`
`160
`
`140
`
`120
`
`100
`
`80
`
`60
`
`40
`
`20
`
`Intensity
`
`5
`
`15
`
`25
`
`35
`Degrees 2-Theta
`
`45
`
`55
`
`65
`
`Fig. 3. XRD patterns of iron–sucrose (top) and iron–gluconate (bottom).
`
`of the diffractograms in Fig. 3, it can be also concluded
`that all peaks broadened evenly, without the presence of
`a selective peak broadening, i.e., there was no preferen-
`tial deformation plane due to the complexation with su-
`crose or gluconate.
`It should be mentioned that XRD patterns of both
`formulations appear to have a very weak diffraction
`peak
`from the
`(0 0 2)
`plane
`(diffraction
`angle
`
`2h = 61.15°), which would indicate the presence of Cl
`
`anions. The presence of Cl
`is in accord with the tun-
`nel-containing akaganeite structure.
`On the XRD pattern of iron–sucrose sample after
`pH-matched dialysis, the sucrose peaks at 11.5° and
`22° are almost eliminated (Fig. 4, top spectrum). How-
`
`lifting of the area of their presence
`ever, an overall
`around 20° confirms that there is still some sucrose in
`the sample and that the sucrose is not uniformly or-
`dered. Dialysis did not destroy the iron mineral core,
`since the main reflection attributed to the akaganeite
`polymorph remains strong after dialysis.
`For iron–gluconate, the effect of removing unbound
`sucrose was more noticeable (Fig. 4, bottom spectrum).
`In addition to the apparent disappearance of the sucrose
`peaks at 11.5° and 22°, the pattern of the dialyzed sam-
`ple has a newly visible akaganeite peak at 26.8°. That
`difference in the peak appearance between the two dia-
`lyzed samples occurs despite the fact that the size of
`the iron–gluconate core is slightly smaller than that of
`
`200
`
`180
`
`160
`
`140
`
`120
`
`100
`
`Intensity
`
`80
`
`60
`
`40
`
`20
`
`5
`
`15
`
`25
`
`35
`Degrees 2-Theta
`
`45
`
`55
`
`65
`
`Fig. 4. XRD patterns of dialyzed iron–sucrose (top) and dialyzed iron–gluconate (bottom).
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1052 - Page 7
`
`

`

`1764
`
`D.S. Kudasheva et al. / Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`iron–sucrose, as demonstrated by TEM. On the basis of
`overall core size, the peaks in the XRD pattern of iron–
`gluconate should be less pronounced. Thus, the differ-
`ence in the diffractogram appearance may indicate more
`uniform crystallites in the iron–gluconate than in the
`iron–sucrose, after dialysis to remove unbound and
`loosely bound carbohydrate.
`Our FTIR studies (unpublished data) of different iron
`oxyhydroxide polymorphs also confirmed that the iron
`core of both formulations has the chemical structure
`of iron (III) oxyhydroxide akaganeite b-FeOOH.
`The average crystallite size, D, was estimated from
`the diffraction peak from the (2 1 1) plane (diffraction an-
`gle 2h = 35.1°), and gave a similar result for both formu-
`lations.
`In the case of nondialyzed samples,
`the
`crystallite size for iron–sucrose was estimated to be 4.6
`nm, while for iron gluconate it was 4.1 nm. For the
`iron–sucrose sample after pH-adjusted dialysis (Fig. 4),
`D calculated for the same peak gave a smaller size of
`3.4 nm. The average crystallite size for the dialyzed
`iron–gluconate was 3.5 nm. These values are slightly
`higher than the core diameters estimated for dialyzed
`samples by transmission electron microscopy. Consider-
`ing the difficulty in accurately measuring the diffraction
`peak width at half height, the TEM data are considered
`more reliable.
`
`3.4. Atomic force microscopy
`
`Imaging the iron–carbohydrate complexes was per-
`formed for samples applied to a mica surface, followed
`by brief rinsing to remove unbound material, including
`excess soluble low molecular mass carbohydrates. The
`images of the two iron formulations showed large num-
`bers of beadlike nanoparticles spread over the scanning
`surface.
`The arrangement of particles on the mica surface is
`different from the arrangement on the surface of the
`TEM grid, which is quite expected because the surfaces
`are different. AFM mica is hydrophilic, whereas the
`TEM grid is covered with carbon, which makes it highly
`hydrophobic. Thus the particles are more uniformly dis-
`tributed on the mica surface then on the carbon coated
`TEM grid.
`Fig. 5 shows 3 lm · 3 lm scans of each iron formu-
`lation sample at an iron concentration of 5 mg/mL.
`The surfaces showed residual excess sucrose or sucrose
`and gluconate surrounding the nanoparticles. The bead-
`like particles, which include both the mineral core and
`the bound carbohydrate, appear to be similar in size
`for iron–sucrose and iron–gluconate.
`Images obtained at higher resolution are preferred for
`more accurate measurements of the nanoparticle sizes.
`Fig. 6 shows 1 lm · 1 lm scans of nanoparticles from
`each preparation type, analyzed at comparable iron con-
`centrations after dilution with water. The distributions
`
`Fig. 5. Atomic force microscope height images of: (a) iron–sucrose
`complex, (b) iron–gluconate complex in sucrose, after dilution to an
`iron concentration of 5 mg/mL. Image size 3 lm · 3 lm, with a color
`scale covering a height range of 15–20 nm (darker features being lower
`in height than lighter ones).
`
`of particle heights measured from the images in Fig. 6
`are given in Table 3 for iron–sucrose and Table 4 for
`iron–gluconate. An equal number of particles (75) were
`analyzed at each concentration for each preparation
`type.
`At a concentration of about 5 mg Fe/mL, the iron–
`sucrose complexes (Fig. 6(a) and Table 3) appear as
`beads with an average height of about 3.5 nm, and a cor-
`rected width of approximately 22 nm. Analyzed at the
`same 5 mg Fe/mL concentration, the iron–gluconate in
`sucrose complex (Fig. 6(e) and Table 4) had an average
`height of approximately 3.1 nm, and an average appar-
`ent particle width of approximately 12 nm.
`For an incompressible spherical bead, the height and
`width measurements should both be equal to the bead
`diameter. Two factors contribute to the mismatch be-
`tween vertically measured height and horizontally meas-
`ured diameter. The first is a flattening effect caused by
`the tapping impact of the probe. The carbohydrate com-
`ponent of the beads is subject to compression and dis-
`
`PGR2020-00009
`Pharmacosmos A/S v. American Regent, Inc.
`Petitioner Ex. 1052 - Page 8
`
`

`

`D.S. Kudasheva et al. / Journal of Inorganic Biochemistry 98 (2004) 1757–1769
`
`1765
`
`observed upon dilution to 2 mg Fe/mL, was primarily
`caused by the fact that the particles were deeply embed-
`ded in a sucrose residue covering the surface for that
`particular sample (Fig. 6(b)). Thus, the measured height
`in that case is really only the height of the upper part of
`the particles above the sucrose. That phenomenon can
`also explain the observation that upon the next dilution
`to 0.8 mg/mL the average measured height increased
`again to 3 nm (Fig. 6(c)), in contrast with the logical
`expectation that upon dilution the particle size would
`decrease. Upon the final dilution to 0.66 mg/mL the
`average measured height stayed nearly constant at 3
`nm. We conclude that the height of the particles is rela-
`tively stable with dilution. The fact that the average
`height of a whole iron–sucrose particle approximately
`coincides with the average diameter of the iron mineral
`core from TEM images indicates the severity of the car-
`bohydrate shell compression and/or displacement by the
`AFM t