OF BIOLOGICAL CHEMISTRY
`THE JOURNAL
`0 1987 by The American Society for Biochemistry
`
`and Molecular Biology, Inc.
`
`Vol. 262, No. 23, Issue of August 15, PP. 10971-10979,1987
`Printed in U.S.A.
`
`Carbon 13 NMR Studies of Saturated Fatty Acids Bound to Bovine
`Serum Albumin
`I. THE FILLING OF INDIVIDUAL FATTY ACID BINDING SITES*
`
`(Received for publication, September 10,1986)
`
`David P. Cistola, Donald M. Small, and James A. Hamilton
`From the Biophysics Institute, Housman Medical Research Center, Departments of Medicine and Biochemistry,
`Boston University School of Medicine, Boston, Massachusetts 02118-2394
`
`13C NMR chemical shift and intensity results for a
`series of carboxyl “C-enriched saturated fatty acids
`(8-18 carbons) bound to bovine serum albumin (BSA)
`are presented as a function of increasing fatty acid
`(FA)/BSA mole ratio. Spectra for long-chain (212 car-
`bons) FA-BSA complexes exhibited up to five FA car-
`boxyl resonances, designated a, b, b’, c, and d. Only
`three resonances (peaks b, b’, and d ) were observed
`below 3:l FA*BSA mole ratio, and at r3:l mole ratio,
`two additional resonances were observed (peaks c and
`a). In a spectrum of 5:l stearic acid-BSA complexes,
`peaks b, b’, and d each represented approximately one-
`fifth, and peak c approximately two-fifths, of the total
`FA carboxyl intensity. Plots of total carboxyl/carbonyl
`intensity ratio as a function of FA-BSA mole ratio were
`linear up to 7-9 mole ratio. Deviation from linearity
`at mole ratios 27 was accompanied by the detection of
`crystalline unbound FA (as 1:l acidlsoap) by x-ray
`diffraction. In contrast to
`long-chain FA*BSA com-
`plexes, 13C NMR spectra of octanoic acid*BSA com-
`plexes yielded only one FA carboxyl resonance (peak
`c ) at FA-BSA mole ratios between 1 and 20. We con-
`clude: (i) peaks b, b’, and d represent FA bound to three
`individual high affinity (primary) long-chain FA bind-
`ing sites on BSA; (ii) peak c represents FA bound to
`several secondary long-chain (or primary short-chain)
`FA binding sites on BSA; (iii) peak a represents long-
`chain FA bound to an additional lower affinity binding
`site. We present a model that correlates the observed
`13C NMR resonances with individual binding site lo-
`cations predicted by a recent three-dimensional model
`of BSA.
`
`Utilization of circulating FFA’ by tissues is influenced not
`only by the avidity of FFA for and the blood flow through
`
`*This research was supported by United States Public Health
`Service Grants HL-26335 and HL-07291. This work was originally
`submitted in partial fulfillment of the degree of Doctor of Philosophy
`at Boston University (Cistola, 1985), and preliminary accounts of
`portions of this work have been published in abstract form (Cistola
`et al., 1983). The costs of publication of this article were defrayed in
`part by the payment of page charges. This article must therefore be
`hereby marked “advertisement” in accordance with 18 U.S.C. Section
`1734 solely to indicate this fact.
`The abbreviations used are: FFA, free (nonesterified) fatty
`acid(s); FA, fatty acid(s); BSA, bovine serum albumin; NOE, nuclear
`Overhauser enhancement, C8.0, octanoic acid; C12.0, dodecanoic (lauric)
`acid; Clr.o, tetradecanoic (myristic) acid; Cla.o, hexadecanoic (palmitic)
`acid; C18.0, octadecanoic (stearic) acid; C18.1, oleic acid. The use of the
`abbreviation “FA” or the numerical abbreviations for individual FA
`compounds is not meant to imply anything about the ionization state
`of the FA carboxyl group.
`
`each tissue, but also by the mole ratio of FFA to albumin in
`the circulation (Scow and Chernick, 1970; Spector and
`Fletcher, 1978). In normal human subjects, this ratio is vari-
`able and is elevated under certain metabolic or environmental
`conditions such as fasting (Frederickson and Gordon, 1958)
`and/or prolonged exercise (Have1 et al., 1967). Under certain
`pathological conditions, FFA/albumin ratios may be
`tran-
`siently or consistently elevated secondary to increased FFA
`mobilization (diabetic ketoacidosis, myocardial infarction,
`acute anxiety) or decreased circulating albumin (nephrotic
`syndrome, liver disease, familial hypoalbuminemia). It is con-
`ceivable that increased FFA production and/or decreased
`circulating albumin could result in abnormal partitioning of
`FFA into other components of the circulation (lipoproteins,
`blood cell membranes, endothelial cell membranes; Spector
`and Fletcher, 1978). This abnormal FFA partitioning might
`result in detrimental structural and/or functional alterations
`such as decreased neutrophil phagocytic and bacteriocidal
`activity (Hawley and Gordon, 1976), platelet aggregation
`(Hoak et al., 1970), and endothelial cell damage (Zilversmit,
`1976).
`As one approach to predicting FFA/albumin interactions
`at different mole ratios in uiuo, the FA binding properties of
`albumin have been extensively examined in vitro using several
`approaches (for a review, see Spector, 1975). Binding data
`obtained from equilibrium partitioning methods have been
`analyzed by the Scatchard model (Scatchard, 1949) or the
`stepwise association model (Klotz et al., 1946; Spector et al.,
`1971). The Scatchard analyses for long-chain FA bound to
`human (Goodman, 1958) and bovine (Spector et al., 1969)
`albumin have yielded the concept of three classes of FA
`binding sites with respect to relative affinities. In contrast,
`the stepwise association analyses assumed no grouping of
`binding constants into classes and suggested that this group-
`ing is somewhat arbitrary (Spector et al., 1971; Ashbrook et
`al., 1975). Second, mapping studies using peptide fragments
`et al., 1975),
`(King and Spencer, 1970; King, 1973; Reed
`chemical modifications (Koh and Means, 1979), or affinity
`labeling (Lee and McMenamy, 1980) have aided in the local-
`ization of ligand binding sites to general regions on
`the
`polypeptide sequence. However, pitfalls include possible dis-
`ruptive changes in protein conformation and binding proper-
`ties following fragmentation, inconsistencies in affinity label-
`ing, and a
`lack of specificity with chemical modification
`(Brown and Schockley, 1982). Third, spectroscopic studies
`using fluorescence (Sklar et al., 1977; Berde et al., 1979), ESR
`(Kuznetsor et al., 1975; Morrisett et al., 1975; Rehfield et al.,
`1978; Perkins et al., 1982), and NMR (Muller and Mead, 1973;
`Inoue et al., 1979) spectroscopy have yielded information
`concerning the physicochemical interactions of ligands and
`
`10971
`
`MPI EXHIBIT 1008 PAGE 1
`
`MPI EXHIBIT 1008 PAGE 1
`
`

`

`13C NMR of Fatty Acid-Albumin Interactions
`albumin. However, pitfalls include the need for fatty acids
`phoresis. I3C NMR spectra for C14.0.BSA complexes using monomeric
`or unfractionated BSA (obtained under identical conditions) were
`containing structure-perturbing spin-label probes (ESR) or
`indistinguishable. This result is consistent with our previous obser-
`conjugated double bonds (fluorescence). Also, 13C NMR at
`vations for C,,.,.BSA complexes (Parks et al., 1983).' Therefore,
`natural abundance has been hampered by a lack of sensitivity
`unfractionated BSA was used throughout this study.
`(Kragh-Hansen and Riisom, 1976).
`13C carboxyl-enriched (90%) fatty acids were purchased from KOR
`As an alternative approach, we have utilized 13C NMR
`Isotopes (Cambridge, MA) ( G o , Clz.o, c16.0, C18.,) and Merck Sharp
`spectroscopy with 13C-enriched fatty acids to investigate the
`and Dohme Isotopes (St. Louis, MO) (C14.0). Sample purity as deter-
`mined from thin layer chromatography (hexanediethyl ether:acetic
`interactions of biologically important FA with bovine albu-
`acid, 90:9:1) was >98%. In addition, no impurities were visualized by
`min. Carbon 13 enrichment greatly enhances spectral sensi-
`'H NMR for I3C-enriched FA samples dissolved in deuterated chlo-
`tivity and permits investigation of FA/albumin interactions
`roform.
`in the range of physiologically relevant FA/albumin mole
`Sample Preparation-BSA solutions (7%, w/v) were prepared using
`ratios. Using this approach, we have shown that the carboxyl
`doubly distilled deionized water. After adjusting the pH to 7.4, the
`chemical shift of oleic acid bound to BSA is highly sensitive
`solutions were centrifuged at 10,000 rpm for 30 min to remove trace
`amounts of particulate matter. The protein concentration was deter-
`to the FA binding environment on albumin; spectra revealed
`mined from the absorbance at 279 nm of 1:lOO dilutions using an
`multiple FA carboxyl resonances corresponding to multiple
`extinction coefficient of 6.67 for a 1% sample (Janatova et al., 1968).
`FA binding environments (Parks et al., 1983). In addition, we
`Crystals of 13C-enriched FA were dissolved in 2:l chloroform/meth-
`have utilized fatty acids with 13C enrichment in hydrocarbon
`anol, and the concentrations were determined by measuring dry
`chain carbons (C-3, (2-14) to probe the interactions of myristic
`weights on an electrobalance (Cahn model 25, Cerritos, CA). Stoichi-
`acid with BSA (Hamilton et al., 1984).
`ometric amounts of fatty acids in solvent were added to 10-mm NMR
`tubes, and the solvent was evaporated under N,. D20 (200 pl) and 1.2
`This paper presents 13C NMR results for a series of carboxyl
`eq of 1 N KOH were added, and samples were thoroughly mixed until
`I3C-enriched saturated fatty acids (8-18 carbons) bound to
`all FA crystals dissolved to form an optically clear micellar soap
`BSA as a function of increasing FA.BSA mole ratio. The
`solution. Hydrated BSA samples (1.8 ml, pH 7.4) were added to the
`results provide direct physicochemical information regarding
`soap solutions (0.2 ml) with continuous vortexing for several minutes,
`the order of filling and saturation of individual FA binding
`and samples were equilibrated and intermittently vortexed for 30
`min. Potassium stearate solutions formed a gel phase at room tem-
`sites with increasing FA.BSA ratio as well as the relative
`perature and had to be gently and briefly heated before BSA solutions
`occupation of individual sites at a given mole ratio. Second,
`were added. Sample pH was adjusted from pH 7.5-7.6 (following
`the results delineate differences between shorter chain and
`mixing) to 7.4 and samples were equilibrated at room temperature
`longer chain FA with regard to the number and type of FA
`(25 "C) for 8-12 h prior to I3C NMR experiments. The NMR results
`binding sites. Third, accompanying powder x-ray diffraction
`were independent of equilibration time.
`results provide information about the physical state of un-
`All pH measurements were made directly in the NMR tube using
`a pH meter (Beckman 3560, Fullerton, CA) equipped with a 29 cm x
`bound FA at high mole
`ratios. Finally, the results permit
`4 mm glass combination electrode (Markson MiraMark, Phoenix,
`direct correlation of observed NMR resonances with the FA
`AZ). Values measured before and after obtaining NMR spectra agreed
`binding sites predicted by recent models of BSA structure
`within 0.1 pH unit.
`(Brown and Shockley, 1982) and provide insights into the
`The final FA.BSA samples used for NMR contained from 0.5 to
`binding locations of FA at different FA/albumin ratios i n
`20 mol of FA/mol of BSA and 7% w/v protein. As reported elsewhere
`uiuo.
`(Cistola, 1985), I3C NMR spectra (FA as well as protein resonances)
`for C16.0.BSA complexes at 3.8, 7.5, and 11.4% w/v BSA (all at 5:l
`mol ratio, pH 7.4, 35 "C) were essentially identical. Hence, it is
`unlikely that noncovalent protein aggregation occurred over this
`concentration range. The salt concentrations (as determined by flame
`photometry) of 7% hydrated samples with no added salt were 7 mM
`(sodium) and 6 mM (potassium). No salt was added to FA.BSA
`samples in this study. Addition of KC1 to FA.BSA samples up to a
`final concentration of 0.1 M does not change 13C NMR results pro-
`vided that the sample temperature is kept below 38 "C (Cistola, 1985).
`Carbon 13 NMR Spectro~copy-~~C NMR spectra were obtained
`on a Bruker WP-200 NMR spectrometer at 50.3 MHz as described
`elsewhere (Hamilton and Small, 1981; Cistola et al., 1982). Internal
`D,O was used as a lock and shim signal. Chemical shift values were
`measured digitally with an estimated uncertainty of f O . l ppm. The
`chemical shift (6 = 39.57 ppm) of the narrow resonance from protein
`c-Lys/P-Leu carbons (Gurd and Keim, 1973) was used as an internal
`reference after calibrating this resonance against external tetrameth-
`ylsilane. To enhance spectral resolution in selected cases, the convo-
`lution difference method was used (Campbell et al., 1973). FA car-
`boxyl/BSA carbonyl intensity ratios were measured using the inte-
`gration routine provided in the Bruker DISNMR program. NMR
`sample temperatures were controlled to 34 "C and measured as de-
`scribed previously (Cistola et al., 1982). Spin-lattice relaxation times
`( Tl) were measured using a fast inversion recovery technique (Canet
`et al., 1975) and calculated using a three-parameter fitting routine
`(Sass and Ziessow, 1977). Nuclear Overhauser enhancements (NOE)
`were determined from comparisons of peak heights from spectra
`accumulated with broad-band and inverse-gated decoupling (Opella
`et al., 1976). For all spectral accumulations, pulse intervals were equal
`to the TI value of the largest FA carboxyl peak in the spectrum, and
`90 "C pulses (15 p s ) were used.
`
`EXPERIMENTAL PROCEDURES
`Materials-Essentially FA-free crystallized lyophilized BSA was
`purchased from Sigma (A-7511, lot 22F-9340). The supplier used
`method IV of Cohn et al. (1947) to recrystallize fraction V albumin
`and the charcoal defatting procedure of Chen (1967) to remove bound
`fatty acid. The content of bound fatty acid, as determined by gas-
`liquid chromatography, was C0.02 mol of FA/mol of BSA prior to the
`addition of FA. The protein content of the BSA sample was analyzed
`by sodium dodecyl sulfate-polyacrylamide gel electrophoresis; the gels
`were overloaded with sample in order to search for minor impurities.
`In addition to the major albumin band at 66,000 daltons, minor bands
`were observed at -120,000 (-5%), 55,000 (<1%), and 160,000 ( 4 % ) .
`These minor bands most likely corresponded to BSA dimers (Fried1
`and Kistler, 1970; Foster, 1977), a,-antitrypsin (Laurel1 and Jeppsson,
`1975), and immunoglobulins, respectively (Putnam, 1975; Peters,
`1975). Although apoprotein A-I is often present as a contaminant in
`commercial albumin preparations (Fainaru and Deckelbaum, 1979),
`no bands were observed at the appropriate molecular weight (28,000).
`The dimer/polymer content of Sigma A-7511 BSA (with added
`FA), as determined by column chromatography (Parks et al., 1983),
`was 25%. To determine whether NMR results were affected by the
`presence of disulfide-linked dimers and polymers, Sigma BSA was
`further fractionated by gel filtration chromatography. A 2-ml aliquot
`of hydrated BSA (100 mg/ml) was applied to a column of Sephadex
`G-150 (90 X 2.6 cm) equilibrated with 20 mM KC1, 0.1% NaN3 at
`4 "C. Fractions of 6 ml were collected at a flow rate of 7 ml/h. The
`protein concentration of each fraction was determined by its absorb-
`ance at 279 nm. The elution profile was essentially identical to one
`ly published (Morrisett et al., 1975). Fractions containing
`prcr .'I
`monomeric BSA were pooled
`and concentrated by ultrafiltration
`using Amicon (Danvers, MA) PM-10 filters. The final monomeric
`BSA sample contained 1.8 ml of 60 mg/ml BSA. No reformation of
`dimers/oligomers occurred in the concentrated monomeric fraction,
`as determined by sodium dodecyl sulfate-polyacrylamide gel electro-
`
`10972
`
`Note that in our previous study, a 90 X 2.5-cm column was used
`for preparative fractionation of monomeric BSA, rather than a 90 x
`1.5-cm column (incorrectly noted in Fig. 1 caption of Parks et al.,
`1983).
`
`MPI EXHIBIT 1008 PAGE 2
`
`MPI EXHIBIT 1008 PAGE 2
`
`

`

`I 3 C NMR of Fatty Acid-Albumin Interactions
`containing suspended crystalline
`X-ray Diffraction-For samples
`material (>7:1 FA. BSA), the material was pelleted by centrifugation
`at 10,000 rpm for 30 min at 30 "C. The pellet was transferred to 1-
`mm quartz capillary tubes (Charles Supper Co., Natick, MA), and
`the capillaries were placed in
`a sample holder kept
`at constant
`temperature (30 "C) by a circulating antifreeze/water bath. Nickel-
`(X = 1.5418 A) from a microfocus x-ray
`filtered CuKa x-radiation
`generator (Jarrell-Ash, Waltham, MA) was focused by a single nickel-
`coated mirror and further collimated by a Luzzati-Bar0 camera with
`slit optics. Low-angle powder x-ray diffraction patterns were recorded
`with a position-sensitive detector (Tennelec PSD-1100, Oak Ridge,
`TN) and a computer-based analysis system (Tracor Northern TN-
`1710, Middleton, WI).
`
`h
`
`10973
`
`711 - 511
`
`d
`
`,
`
`D'
`
`311 - 111
`
`111 - 011
`
`RESULTS
`NMR spectra at various C14.0-BSA mole ratios (at fixed
`pH, BSA concentration, ionic strength, and temperature) are
`shown in Fig. 1, A-E. The broad envelope centered at -176
`ppm represents carbonyl carbons of glutamine, asparagine,
`and the peptide backbone (Gurd and Keim, 1973) as well as
`aspartate carboxyl carbons (Shindo and Cohen, 1976) of BSA.
`The narrower resonances falling between 179 and 184 ppm
`primarily represent carboxyl carbons of 13C-carboxyl-enriched
`FA bound to BSA (Parks et al., 1983; Cistola et al., 1983;
`Hamilton et al., 1984), except for the resonance at
`180.9-
`181.1 ppm, which represents protein glutamate carboxyl car-
`bons (Shinodo and Cohen, 1976). Protein-free saturated FA
`(>lo carbons) exist as crystalline 1:1 acid/soap compounds a t
`pH 7.4 and 35 "C (Cistola et al., 1986), and crystalline phases
`do not give rise to high resolution NMR resonances. Further-
`more, none of the observed carboxyl resonances had chemical
`shifts coincident with those of soluble short-chain FA in water
`et al.,
`(without protein) under the same conditions (Cistola
`1982; Cistola, 1985). Hence, the observed carboxyl resonances
`do not represent unbound
`FA. In addition, none
`of these
`resonances represented BSA peaks that shifted into the car-
`boxyl region upon FA binding, since spectra
`of FA.BSA
`complexes using C14.0 with no 13C enrichment were identical
`to FA-free BSA samples (Fig. lA). In order to compare FA
`carboxyl peaks in different FA. BSA spectra, we have named
`analogous FA carboxyl peaks, based on their chemical shifts
`at pH 7.4 (and their ionization behavior; Cistola et al., 1987),
`as follows: peak a (183.7-184.1 ppm), peak b (182.5 ppm),
`peak b' (182.2-182.4 ppm), peak c (181.8-182.1 ppm), and
`peak d (180.4-180.7 ppm). In addition, we have named the
`glutamate carboxyl resonance as peak p r (180.9-181.1 ppm).
`b' was not observed for Clal .BSA
`It is notable that peak
`complexes (Parks et al., 1983) either because peak b' was not
`present or because it was not resolved from peaks b and c.
`For most of the FA.BSA spectra presented in this study,
`the intensitivies of individual FA carboxyl peaks could not be
`
`quantitatively measured as peak areas or peak heights because
`of closely overlapping FA or protein carboxyl resonances.
`Therefore, increases in FA carboxyl peak intensity with in-
`creasing mole ratio are represented qualitatively in the form
`of difference spectra. Digital subtraction of the upper spectra
`from the lower spectra yielded the difference spectra shown
`in the right column of Fig. 1 (F-Z). The difference spectra
`contained only those FA carboxyl peaks which increased or
`decreased in intensity or changed chemical shifts between the
`two different mole ratios. The intensities resulting from un-
`perturbed carboxyl or carbonyl resonances of BSA were sub-
`tracted out by this procedure.
`At 1:l mole ratio (Fig. l B ) , peaks b and b' were clearly
`visible, but peak d was difficult
`to distinguish because of
`closely overlapping protein glutamate resonance(s). However,
`subtraction of FA-free BSA (Fig. lA) from 1:1 C14.0. BSA (Fig.
`1B) revealed that peak d was present a t 1:1 mole ratio (Fig.
`
`188
`
`I82
`
`178
`
`I74
`
`chemical shlft
`(ppm)
`FIG. 1. Carboxyl/carbonyl region of ''C NMR spectra (A-
`E ) and difference spectra (F-I) for C14.,,.BSA complexes with
`different C14.0.BSA mole ratios at pH 7.4 and 34 "C. Difference
`spectra were obtained by digitally subtracting a spectrum at a given
`mole ratio from one at a higher mole ratio. This method removes
`BSA resonances from
`spectra and shows which FA carboxyl reso-
`nances increased between two corresponding FA.BSA mole ratios.
`The pairs of spectra which were
`subtracted are indicated by the
`dashed lines in the middle of the figure. The lower case letters above
`each peak indicate specific FA carboxyl resonances with characteristic
`chemical shifts (see "Results"). For all samples, the BSA concentra-
`tion was 7% (w/v). All spectra were recorded after 6,000 accumula-
`tions with a pulse interval of 2.0 s, 16,384 time domain points, and a
`spectral width of 10,000 Hz. Line broadening (3 Hz) was used in all
`spectrum; B, 1:l C14.n.BSA
`spectral processing. A, FA-free BSA
`spectrum; C, 3:l C14.0.BSA spectrum; D, 5:l CI4.,.BSA spectrum; E,
`7:l C1,n.BSA spectrum; F, difference spectrum, 1:l C14.n.BSA minus
`FA-free BSA; G, difference spectrum, 3:l C14.,.BSA minus 1:1 Clro.
`BSA; H, difference spectrum, 5:l CI4.,.BSA minus 3:l Cl4,,.BSA; I,
`difference spectrum, 7:l C14.0'BSA minus 5:l CI4.,.BSA.
`
`1F). Between 1:l and 3:l mole ratio (Fig. 1, B, C, and G),
`peaks b, b', and d increased, and peaks c and a appeared above
`2:l mole ratio (2:l spectrum not shown). Between 3:l and 5:l
`mole ratio (Fig. 1, C, D, and H), intensity increases occurred
`in all five FA carboxyl peaks. Between 5:l and 7:l mole ratio
`(and up to 13:l mole ratio), only peak c increased in intensity
`(Fig. 1, D, E, and I). Peak d decreased in intensity between
`5:l and 7:l mole ratio (Fig. 1Z).
`A plot of total carboxyl/carbonyl area ratio as a function of
`C14.0 .BSA mole ratio is shown in Fig. 2 (circles). The plot is
`linear up to 8-9:l mole ratio, above which point the sample
`downward
`became increasingly
`turbid (suspended crystals;
`pointing arrow in Fig. 2). The samples were centrifuged and
`yielded a transparent supernatant and a crystalline pellet.
`
`MPI EXHIBIT 1008 PAGE 3
`
`MPI EXHIBIT 1008 PAGE 3
`
`

`

`13C NMR of Fatty Acid-Albumin Interactions
`
`i
`
`i
`
`,""
`
`Mole Ratio FAIBSA
`FIG. 2. Plots of total "C NMR carboxyl intensity relative
`to total carbonyl intensity as a function of FA.BSA mole ratio
`for Cl,.o.BSA (o"-o) and Cla.o.BSA (A-A).
`The arrows
`indicate the mole ratio at and above which turbidity and crystalline
`precipitate were visualized in Clao.BSA (arrow pointing downward)
`and Clao.BSA (arrow pointing upward) samples. The dashed line
`indicates the expected (extrapolated) intensity ratio if essentially all
`FA were protein associated. FA not associated with protein under
`these conditions was crystalline and would not contribute to the total
`NMR carboxyl intensity. Note that the absolute values of the inten-
`sity ratios are arbitrary and have no direct stoichiometric meaning.
`
`Examination of the pellet by powder x-ray diffraction showed
`first order long spacings (41.0 f 0.9A) characteristic for
`crystalline 1:l potassium (or sodium) hydrogen dimyristate, a
`1:l acid/soap compound (Piper, 1929; Cistola et al., 1986).
`Therefore C14.0'BSA samples at high mole ratio contained
`unbound FA in the form of crystalline 1:l acid/soap. In
`general, long-chain fatty acids in water (>micromolar concen-
`trations) form 1:1 acid/soap crystals or fatty acid/soap lamel-
`lar liquid crystals between pH 7 and 10 (Small, 1986)?
`13C NMR spectra and difference spectra at various Cla.o.
`BSA mole ratios are shown in Fig. 3, A-I. At 1:l mole ratio,
`peaks b, b', and d were present (Fig. 3, B and F ) , although
`peak d was barely detectable. Peak d was much more clearly
`seen at 2:1 (spectrum not shown) and 31 (Fig. 3, C and G).
`Between 1:l and 3:l mole ratio, peaks b, b', and d increased
`in intensity (Fig. 3, B, C, and G ) . Between 3:l and 5:l mole
`ratio, peaks c and a became visible and all five FA carboxyl
`peaks increased in intensity (Fig. 3, C, D, and H), and between
`5:l and 7:l (spectrum not shown), peaks c, b/b', and a in-
`creased in intensity (Fig. 3, D, E, and I ) .
`A plot of total carboxyl/carbonyl area ratio as a function of
`Clao-BSA mole ratio (not shown) was linear up to about 7:l
`mole ratio. Deviation from linearity was accompanied by the
`visual appearance of sample turbidity (suspended crystals) at
`and above 7:l mole ratio. As with C14.0/BSA (see above), these
`crystals most likely represent unbound Ciao in the form of
`crystalline 1:l acid/soap.
`13C NMR spectra (but not difference spectra) for Clz,o. BSA
`complexes at four mole ratios are shown in Fig. 4. These
`spectra were very similar to corresponding spectra for c14.0'
`BSA complexes. At 1:l mole ratio, peaks b/b' and pr/d were
`visualized; at 3:1, three FA peaks were visualized (b, b', d).
`At 5:l and 7:l mole ratios, all five FA carboxyl peaks were
`distinguishable. Convolution difference spectra (not shown)
`permitted greater resolution of peaks b and b'.
`
`D. P. Cistola, J. A. Hamilton, D. Jackson, and D. M. Small (1987)
`Biochemistry, submitted for publication.
`
`",
`
`186
`
`I
`
`I
`170
`
`188
`
`I \
`
`h
`
`3/1 - 111
`
`170
`
`1
`I78
`I78
`174
`1 8 2
`182
`114
`chemical shift
`chemical shift (pprn)
`(pprn)
`FIG. 3. Carboxyl/carbonyl region of "C NMR spectra (A-
`E ) and difference spectra (F-I) for Cle.o.BSA complexes with
`increasing Cls.o.BSA mole ratio at pH 7.4, 34 OC. The lower
`case letters above each peak indicate specific FA carboxyl resonances
`with characteristic chemical shifts (see "Results"). Spectra were
`recorded after 6000 accumulations with a pulse interval of 2.7 s. All
`other spectral conditions and explanations are as described in the
`legend to Fig. 1.
`A plot of total carboxyl/carbonyl area ratio as a function of
`C12.0-BSA mole ratio (not shown) was essentially identical to
`that for C14.0+BSA (Fig. 2, circles). Deviation from linearity
`and the appearance of sample turbidity (crystal formation)
`occurred above 8-9:l mole ratio. Centrifugation and exami-
`nation of crystals by powder x-ray diffraction revealed first
`order long spacings (35.4 f 0.7 A) characteristic for crystalline
`1:l potassium (or sodium) hydrogen laurate, an acid/soap
`compound (Cistola et al., 1986).
`13C NMR spectra for C16.0. BSA complexes as a function of
`mole ratio (Cistola, 1985) are not shown here and were essen-
`tially identical to those for Cl8,, BSA complexes (Parks et al.,
`1983). A plot of total carboxyl/carbonyl area ratio as a func-
`tion of C16.0-BSA mole ratio is shown in Fig. 2 (triangles).
`Deviation from linearity and sample turbidity (crystals) ap-
`peared at and above 7:l mole ratio. The crystals most likely
`represented crystalline 1:l acid/soap, as was demonstrated by
`x-ray diffraction, for C12.,.BSA and C14,o'BSA samples.
`Although peaks b, b', and c closely overlapped in many FA.
`BSA spectra, peaks d and a were nearly completely resolved
`from the b/b'/c envelope, and their relative intensities (areas)
`could be quantitatively estimated (Table I). To determine the
`relative area of peak d, the area from the glutamate carboxyl
`resonance (peak p r ) had to be subtracted out (see legend to
`Table I). The results indicate that the relative intensities of
`peak d increased and peak a decreased, with increasing FA
`chain length. However, the relative intensity of the b/b'/c
`
`MPI EXHIBIT 1008 PAGE 4
`
`MPI EXHIBIT 1008 PAGE 4
`
`

`

`13C NMR of Fatty Acid-Albumin Interactions
`
`10975
`
`TABLE I
`Relative I 3 C NMR intensities of peaks a and d
`The areas of all resonances were measured by integration (see
`“Experimental Procedures”). The areas of peak d and the total FA
`carboxyl region were determined by subtracting out the contribution
`of BSA glutamate carboxyl resonances. This glutamate contribution
`(glutamate/carbonyl ratio) was determined from spectra of FA-free
`BSA accumulated under the same conditions as FA/BSA spectra. All
`three samples reported in this table contained 5:l mole ratio of FA/
`BSA. Ciao. BSA and C1s.O.BSA results were derived from spectra
`shown in Figs. 1D and 30, respectively, and C16.0.BSA results were
`derived from a spectrum not shown here (Cistola, 1985). The esti-
`mated uncertaintv is +lo%.
`
`dla ratio
`
`Ciao. BSA
`Cxo.BSA
`C1s.o.BSA
`
`0.09
`0.7
`2.3
`
`:I
`
`Total FA carboxyl intensity
`b + b ‘ + c
`a
`
`d
`
`2
`12
`19
`
`%
`
`20
`19
`8
`
`78
`69
`73
`
`4 IC
`
`4 / 1
`
`b
`
`‘
`I
`I
`186 1 8 2 1 7 8 1 7 4 1 7 0
`
`{
`
`‘
`
`I
`
`
`
`chemical shift (ppm)
`FIG. 5. Carboxyl/carbonyl region of “C NMR spectra for
`
`l-”C CB.O.BSA with increasing CS.~.BSA qple ratio at pH 7.4
`and 34 “C. The numbers at the right of each spectrum indicate the
`n\ole ratio of Cs.o’BSA. The narrow peaks labeled c are FA carboxyl
`resonances, and the very broad peaks centered at -176 ppm corre-
`spond to BSA carbonyl resonances. Spectra were recorded after 4000
`accumulations with a pulse interval of 2.0 s All other spectral
`conditions are as described in the legend to Fig. 1. The BSA concen-
`tration was 7.5% (w/v).
`
`182.1 ppm; this value suggested that this FA resonance was
`analogous to peak c in the other FA.BSA spectra. This
`correlation was also supported by NMR titration results
`which demonstrated that peak c was the only m e of the five
`observed FA carboxyl peaks to exhibit a titration shift be-
`tween pH 7.4 and 3.0 (Cistola et d., 1987). The chemical shift
`of peak c in C,,.BSA spectra increased with increasing mole
`ratio from 182.0 ppm (1:l and 3:l mole ratio) to 182.1 ppm
`(5:1), 182.2 ppm (7:1), 182.3 ppm (9:1), and 182.9 ppm (20:1).4
`The line widths of this resonance remained narrow ( 4 0 Hz)
`at all mole ratios studied.
`‘ Unbound Ca.0 under these conditions would have existed as mon-
`omers in solution, rather than crystalline or liquid-crystalline aggre-
`gates or micelles. For C8., BSA samples at >3:1 mole ratio, a progres-
`sive increase in the chemical shift of peak c (see “Results”) toward
`the chemical shift of monomeric aqueous C S . ~ (184.2 ppm) provided
`evidence that monomeric unbound Cs.0 was present at concentrations
`>>1 ~ L M and in rapid exchange (>>lo0 exchanges/s) with protein-
`bound C8.0.
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`I
`
`
`
`I
`
`
`
`1 8 2 1 7 8 1 7 4 1 7 0
`chemical shift (ppm)
`FIG. 4. Carboxyl/carbonyl region of “C NMR spectra for
`l-”C C12.0.BSA with increasing C12.,.BSA mole ratio at pH
`7.4 and 34 “C. The numbers at the right of each spectrum indicate
`the mole ratio of Clz.o.BSA. Spectra were recorded after 4000 accu-
`mulations with a pulse interval of 2.8 s. All other spectral conditons
`are as described in the legend to Fig. 1.
`
`FA
`envelope exhibited little or no change with increasing
`chain length. (Results
`for C,,,.BSA were not included be-
`cause peak d was not well resolved from the b-b’-c complex;
`Fig. 4.)
`13C NMR spectra for Ca.o.BSA complexes (Fig. 5 ) , unlike
`those for all other FA.BSA complexes studied, showed only
`one FA carboxyl peak at all mole ratios (up to 20:l). At 5:l
`mole ratio (pH 7.4), the chemical shift of this resonance was
`
`MPI EXHIBIT 1008 PAGE 5
`
`MPI EXHIBIT 1008 PAGE 5
`
`

`

`TABLE I1
`FA carboxyl Tl and NOE values for various FAIBSA complexes
`All measurements were made at pH 7.4 and 35 'C. The Tl value
`for the BSA c-Lys/p-Leu resonance (39.57 ppm) was 0.3 s for all
`samples shown in this table.
`BSA
`Sample,
`mole ratio
`concentration
`%, w/u
`7.5
`1.5
`1.5
`3.8
`1.5
`11.4
`7.5
`
`S
`
`C8.o. BSA, 7:l
`C12.0.BSAI 5:l
`C,d.o.BSA, 5:l
`Cl'.o.BSA, 5:l
`Cl'.o.BSA, 5:l
`C16.o.BSA, 5:l
`Cls.o.BSA, 5:l
`
`C
`b/b'/c
`b'lc
`b'lc
`b'lc
`b'lc
`b'lc
`
`2.2 f 0.1 1.6 f 0.2
`3.1 f 0.1
`2.5 f 0.1 1.7 f 0.2
`2.7 f 0.1 1.7 f 0.2
`2.8 f 0.1 1.7 f 0.2
`2.2 f 0.1 1.9 f 0.2
`2.9 k 0.1
`
`13C NMR of Fatty Acid-Albumin Interactions
`ratio, significant amounts o