THE JOURNAL
`OF BIOLOGICAL CHEMISTRY
`0 1987 by The American Society for Biochemistry and Molecular Biology, Inc.
`
`Vol. 262, No. 23, Issue of August 15, pp. 10980-10985,1987
`Printed in W.S.A.
`
`Carbon 13 NMR Studies of Saturated Fatty Acids Bound to Bovine
`Serum Albumin
`11. ELECTROSTATIC INTERACTIONS IN
`
`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
`
`“C NMR chemical shift results as a function of pH
`for a series of carboxyl “C-enriched saturated fatty
`acids (8-18 carbons) bound to bovine serum albumin
`(BSA) are presented. For octanoic acid bound to BSA
`(6:1, mol/mol), the chemical shift of the only FA car-
`boxyl resonance (designated as peak c), plotted as a
`function of pH, exhibited a complete sigmoidal titration
`curve that deviated in shape from a corresponding
`theoretical Henderson-Hasselbach curve. However, FA
`carboxyl chemical shift plotted as a function of added
`HC1 yielded a linear titration curve analogous to those
`obtained for protein-free monomeric fatty acid (FA) in
`water. The apparent pK of BSA-bound octanoic acid
`was 4.3 f 0.2. However, the intrinsic pK (corrected
`for electrostatic effects resulting from the net positive
`charge on BSA) was approximately 4.8, a value iden-
`tical to that obtained for monomeric octanoic acid in
`water in the absence of protein. For long-chain FA
`(212 carbons) bound to BSA (6:1, mol/mol), chemical
`shift titration curves for peak c were similar to those
`obtained for octanoic acid/BSA. However, the four
`additional FA carboxyl resonances observed (desig-
`nated as peaks a, b, b‘, and d ) exhibited no change in
`chemical shift between pH 8 and 3. For C14.,,.BSA
`complexes (3:l and 6:1, mol/mol) peaks b’ and a exhib-
`ited chemical shift changes between pH 8.8 and 11.5
`concomitant with chemical shift changes in the t-car-
`bon (lysine) resonance. In contrast, peaks c and d ex-
`hibited no change and peak b only a slight change in
`chemical shift over the same pH range. We conclude:
`(i) the carboxyl groups of bound FA represented by
`peaks a, b, b’, and d were involved in ion pair electro-
`static interactions with positively charged amino acyl
`residues on BSA; (ii) the carboxyl groups of bound FA
`represented by peak c were not involved in electro-
`static interactions with BSA; (iii) the similarity of the
`titration curves of peak c for BSA-bound octanoic acid
`and long-chain FA suggested that short-chain and
`long-chain FA represented by peak c were bound to
`the same binding site(s) on BSA; (iv) bound FA repre-
`sented by peaks b’ and a (but not d or b) were directly
`adjacent to BSA lysine residues. We present a model
`which correlates NMR peaks b, b’, and d with the
`
`*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, 1985a), and preliminary accounts of
`portions of this work have been published in abstract form (Cistola,
`1985b). 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.
`
`putative locations of three individual high-affinity
`binding sites in a three-dimensional model of BSA.
`
`In biological systems, free (nonesterified) fatty acids often
`associate with macromolecules or macromolecular assemblies.
`For example, FFA’ in the circulation readily associate with
`albumin and, under certain physiological or pathological con-
`ditions, may associate with lipoproteins, platelets, red blood
`cells, and neutrophils (Spector and Fletcher, 1978). In order
`to predict the physical states formed by FFA in these complex
`systems, the ionization state of the FA carboxyl group must
`be known (Small, 1968, 1986; Cistola, 1985a, 1985b; 1986).
`Knowledge of the ionization behavior of bound FFA would
`aid in understanding not only the mechanism of binding of
`FFA to a given component of the circulation but also the
`mechanism of partitioning of FFA between these components
`and the uptake of FFA by tissue parenchymal cells. However,
`determination of the ionization behavior of FFA in systems
`containing multiple ionizable groups and complex macromo-
`lecular aggregates is difficult using conventional potentiomet-
`ric methods and simply assumptions about pK, values, and
`more specific methods, such as NMR titration, must be used.
`The sensitivity of NMR chemical shifts to the ionization
`state of chemical groups is well established (Jardetzky and
`Jardetzky, 1958) and has been extensively utilized to follow
`the ionization behavior of specific residues in amino acids and
`proteins (for a review, see Jardetzky and Roberts, 1981). Of
`the various NMR resonances which exhibit titration shifts,
`I3C carboxyl resonances are among the most useful because
`of their large titration shift range and resolution from ali-
`phatic, aromatic, and even carbonyl carbons.
`This paper presents 13C NMR titration results for a series
`of carboxyl I3C-enriched saturated fatty acids (8-18 carbons)
`bound to bovine serum albumin. NMR chemical shift titration
`curves at low pH (pH 8-3) indicated differences between
`high-affinity and low-affinity binding sites with respect to the
`ionization behavior of bound FA and the involvement of FA/
`BSA electrostatic interactions in individual FA binding sites.
`NMR chemical shift titration curves at high pH (pH 7-12)
`monitored the ionization behavior of lysine €-ammonium
`groups and the effects of lysine ionization on the carboxyl
`
`The abbreviations used are: FFA, free nonesterified fatty acid($;
`FA, fatty acid(s); BSA, bovine serum albumin; Ce.0, octanoic acid;
`C1z.o, dodecanoic (lauric) acid; C14.0, tetradecanoic (myristic) acid;
`C,,,, hexadecanoic (palmitic) acid; C1e.o, octadecanoic (stearic) acid;
`6, chemical shift; A& total change in chemical shift. 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 carboxyl group.
`
`10980
`
`MPI EXHIBIT 1009 PAGE 1
`
`MPI EXHIBIT 1009 PAGE 1
`
`

`

`DH
`
`"I IN ncI ADDED
`FIG. 1. "C NMR chemical shift titration curves for 1.6 mM
`[ l-'sC]Ca.o in water at 40 "C. A, FA carboxyl chemical shift plotted
`as a function of pH. The circles represent experimentally measured
`values and the triangles theoretically calculated (Henderson-Hassel-
`bach) values.' B, FA carboxyl chemical shift plotted as a function of
`added 1 N HC1. The lines represent a least squares fit of the data
`points ( r = 0.99).
`
`chemical shifts of bound FA in individual binding sites. These
`results permitted further
`correlation of FA
`carboxyl reso-
`nances with the putative locations of FA binding sites in a
`three-dimensional model of BSA (Brown and Schockley, 1982;
`Cistola et al., 1987).
`
`EXPERIMENTAL PROCEDURES
`All materials and sample preparation procedures were described in
`detail in the accompanying paper (Cistola et al., 1987). 13C NMR
`spectra of FA. BSA complexes (3:l and 6:l mole ratio) were recorded
`as a function of decreasing or increasing pH using instrumentation
`and techniques described elsewhere (Cistola et al., 1982a, 1987). All
`FA used in this study were 90% 13C enriched at the carboxyl carbon
`position.
`
`13C NMR Ionization Behavior of Fatty Acid-Albumin Complexes
`10981
`obtained for shorter-chain carboxylic acids (Cistola et al.,
`1982a). Fig. 1B presents the same chemical shift data plotted
`as a function of added HCl rather than pH. A linear decrease
`in carboxyl chemical shift was observed between 0.4 and -3.2
`p1 of added HC1 with break points at or near the ionization
`end points. This linear titration curve is analogous to those
`obtained for water-miscible carboxylic acids in the absence of
`protein (Cistola et al., 1982a, 1982b).
`13C NMR titration curves for C8.,.BSA complexes ( 6 1 mol
`ratio) as a function of pH and added HCl are shown in Fig.
`2, A and B, respectively. The chemical shifts of the only FA
`carboxyl peak observed (peak c; Cistola et al., 1987) exhibited
`a complete sigmoidal titration curve (Fig. 2 A , circles) but
`deviated somewhat from the corresponding theoretical Hen-
`derson-Hasselbach curve2 (Fig. 2 A , triangles). The apparent
`pK, derived from the experimental curve, was 4.3 k 0.2. The
`chemical shift versus microliter HC1 (Fig. 2B) plot exhibited
`a linear decrease in chemical shift from 35 to 147 p1 of added
`HCl, with break points at the FA ionization end points, and
`suggested an ionization behavior analogous to protein-free FA
`monomers in water (Fig. 1B and Cistola et al., 1982a).
`13C NMR spectra for C14.0.BSA complexes (6:l mole ratio)
`at selected pH values are shown in Fig. 3. At pH 7.1 (Fig. 3A),
`four partially resolved FA resonances (peaks a, b, b', and c;
`Cistola et al., 1987) were observed. Below pH 7.1, only peak c
`exhibited chemical shift changes; it shifted upfield with de-
`creasing pH from 181.94 ppm (pH 7.1) to 177.53 ppm (pH
`2.9). In contrast, peaks a, b, and b' exhibited no chemical
`shift changes with decreasing pH. (However, peak a decreased
`in intensity below pH 5 and peaks b and b' decreased in
`intensity below pH 4.)
`Fig. 4 displays chemical shift titration curves for CI4.,, . BSA
`complexes. For peak c, a plot of FA carboxyl chemical shift
`as a function of pH exhibited a sigmoidal titration curve (Fig.
`4A, solid circles) which deviated from the calculated theoret-
`ical Henderson-Hasselbach curve' (Fig. 44, triangles). The
`apparent pK, estimated from the experimental curve, was 4.1
`f 0.2. In contrast, peaks a, b, and b' exhibited little or no
`chemical shift changes with decreasing pH (Fig. 44, open
`circles). Fig. 4B shows a linear dependence of the chemical
`shift of peak c on the quantity of added HC1 (solid circles),
`with break points at the FA ionization end points, analogous
`to data for Cao bound to BSA (Fig. 2B) and protein-free C,,
`(Fig. 1B). In contrast, FA carboxyl peaks a, b, and b' showed
`little or no chemical shift changes with added HC1 (Fig. 4B,
`open circles). Thus, with respect to ionization behavior (as
`followed by 13C NMR), C14.,. BSA spectra exhibited two types
`of FA carboxyl peaks: one type (peak c ) exhibited complete
`titration curves similar to those for protein-free FA in water,
`and the second type (peaks b, b', and a ) exhibited little or no
`chemical shift change with decreasing pH or added HC1.
`
`RESULTS
`As a basis for understanding the more complex NMR
`titration curves for FA.BSA complexes, NMR titration curves
`for protein-free aqueous [l-'3C]C8.0 were determined (Fig. 1).
`The concentration of CS., used (1.7 mM) was far below the
`critical micelle concentration of potassium octanoate (400
`mM; Mukerjee and Mysels, 1970) and slightly below the
`solubility limit of fully protonated octanoic acid (2.2 mM; Bell,
`1973). The samples exhibited with no visible
`turbidity or
`phase separation at any pH value studied. Therefore, Cao was
`apparently in monomeric solution throughout the titration.
`Two types of plots are shown in Fig. 1. In Fig. lA, the
`carboxyl chemical shift of Cao was plotted as a function of
`pH. The experimental points (circles) exhibited a sigmoidal
`curve and were essentially consistent with the calculated
`points (triangles) for a theoretical Henderson-Hasselbach ti-
`tration curve.2 The pK derived from the experimental curve
`was 4.8 f 0.1; this value is essentially the same as those
`' The theoretical Henderson-Hasselbach curve was calculated from
`an NMR version of the Henderson-Hasselbach equation
`
`where amax represents the carboxyl chemical shift for fully ionized FA;
`the chemical shift for fully protonated FA, and 8, the chemical
`shift at a given pH value. Equation 1 was derived from the more
`commonly used form of the Henderson-Hasselbach equation
`pH = pK + log[A-]/[HA]
`(2)
`by equating the following expressions for the fractions of ionized FA
`present (Cistola et al., 1982a)
`[A-I/([HAl + [AI) = (8 - 8min)/(8max - 8 m d
`(3)
`and by solving for [A-]/[HA]. LA-] and [HA] represent the concen-
`trations (or activities) of ionized and protonated FA, respectively. Bv
`substituting an experimentally derived pK value into Equation i, pH
`values can be calculated for a given 6 value.
`
`$
`
`t
`
`
`
`PI tN HCI ADDED
`BH
`FIG. 2. "C NMR chemical shift titration curves for [l-''C]
`Cs.o.BSA, 6:l mol ratio, at 35 OC. A, FA carboxyl chemical shift
`plotted as a function of pH. 0--0, experimentally determined curve;
`A- - -A, theoretically calculated Henderson-Hasselbach curve.' B.
`FA carboxyl chemicalshift plotted
`as a function of added HCl.
`
`MPI EXHIBIT 1009 PAGE 2
`
`MPI EXHIBIT 1009 PAGE 2
`
`

`

`10982
`
`C
`
`bl "1\
`
`
`
`I3C NMR Ionization Behavior
`
`of Fatty Acid-Albumin
`
`Complexes
`
`C
`
`b b' I\
`
`FIG. 5. "C NMR chemical shift titration curves for [1-13C]
`C12.0.BSA, 6:l mol ratio, at 35 "C. Symbols are defined as de-
`scribed in the legend to Fig. 4. A, FA carboxyl chemical shifts as a
`function of pH. B, FA carboxyl chemical shifts plotted as a function
`of added HCI.
`
`A
`
`A
`
`d4 3.3
`
`b-t
`
`.n 2.9
`170
`
`188 n 182 T
`
`-
`
`178
`
`-
`7
`
`,
`
`.
`.
`174 7"-
`~
`
`
`7
`
`T
`
`.
`
`
`
`. . 170 \
`
`"
`
`;
`
`r
`
`I88
`
`182
`178
`174
` (PPM)
`CHEMICAL SHIFT IPPM)
`CHEMICAL W T
`spectra for
`FIG. 3. Carboxyl/carbonyl region of "C NMR
`[1-'3C]C14.0~BSA, 6:l mol ratio, at different pH values at
`34 "C. Spectra were recorded after 6,000 accumulations with a pulse
`interval of 2.5 s, 16,384 time domain points, and a spectral width of
`10,000 Hz. Identical scaling and processing factors (including 3.0-Hz
`line broadening) were used for each spectrum. The lower case letters
`above certain peaks indicate specific FA carboxyl resonances with
`characteristic chemical shifts at pH 7.4 (Cistola et al., 1987).
`
`179.0-
`
`178.0
`3.0
`
`'
`
` '
`
`I
`4.0
`
`
`
`'
`
`1
`6.0
`
` 1
`
`1
`7.0
`
`8
`
`
`80
`
`
`
`120
`
`4
`6.0
`40
`pl 1N HCl
`PH
`FIG. 6. "C NMR chemical shift titration curves for [1-13C]
`Cle.o.BSA, 6:l mol ratio, at 35 "C. A, FA carboxylehemical shift
`plotted as a function of pH. B, FA carboxyl chemical shift plotted as
`a function of added HCI. The FA carboxyl peaks became undetectable
`below pH 4.0 (above 100 pl of HCI).
`
`I
`
`0
`
`.--.,
`
`plot of chemical shift as a function of pH for peak c (Fig. 5A,
`solid circles) was sigmoidal with an apparqt pK, of 4.1 k 0.2
`but deviated from the calculated Henderson-Hasselbach curve
`(Fig. 5A, triangles). A plot of chemical shift as a function of
`added HC1 for peak c (Fig. 5B, solid circles) was linear ( r =
`-0.99) with break points at the FA ionization end points,
`analogous to data for C14.0'BSA (peak c; Fig. 3), Cao. BSA
`(peak c, Fig. 2), and protein-free C8., (Fig. 1).
`13C NMR spectra for Cl6.,-BSA and C18,,.BSA complexes
`(not shown) were similar to those for C14.0. BSA and C,,,.
`BSA except for one major difference: for C16.,.BSA and C18,,.
`BSA, peak c became undetectable below pH 3.9. Therefore,
`complete titration curves for peak c could not be generated.
`However, a partial titration curve for c16.0. BSA (peak c) is
`shown in Fig. 6. As with C14.O'BSA and C12.0.BSA, only peak
`c exhibited chemical shift changes with decreasing pH or
`added HCl for C16.,-BSA (Fig. 6) and Clao.BSA (not shown).
`Peaks a, b, and d yielded no chemical shift changes with
`decreasing pH or added HC1.
`Two C14.,.BSA samples (3:l and 6:l mole ratio) were ti-
`trated with 1.0 N KOH from pH 7.4 to 11.9. At 3:l mole ratio,
`essentially all of the c14.0 was bound to the three high-affinity
`binding sites on BSA (represented by peaks b, b', and d;
`Cistola et oL, 1987). The chemical shift of the e-carbon of
`lysine increased with increasing pH beginning at pH 8.8
`concomitant with a decrease in chemical shift of FA carboxyl
`peak b' above 8.8 (Fig. 7). FA carboxyl peak b exhibited a
`
`PH
`p l l N HCl
`FIG. 4. 13C NMR chemical shift titration curves for [l-"C]
`CI4.,.BSA, 6:l mol ratio, at 35 "C. 0--0, experimentally de-
`termined curves for peaks a, b, and b' (see Fig. 3);
`experi-
`mentally determined curves for peak c (Pig. 3); A- - -A, theoretically
`calculated Henderson-Hasselbach curves for peak c.' The experimen-
`tal data points were derived from the NMR spectra shown in Fig. 3
`as well as other spectra not shown. A, FA carboxyl chemical shifts
`plotted as a function of pH. E, FA carboxyl chemical shifts plotted
`as a function of added HCI.
`
`13C NMR spectra (not shown) of Clz.o.BSPv complexes (6:l
`mole ratio) obtained at pH values between 8.0 and 2.8 were
`very similar to those for C14,o.BSA complexes. The corre-
`sponding NMR chemical shift titration curves for C12.0 e BSA
`complexes are shown in Fig. 5. One FA carboxyl peak I(c)
`exhibited complete titration curves with decreasing pH (Fig.
`5A) or added HCl (Fig. 5B). In contrast, peaks b and b'
`exhibited no change in chemical shift over this pH range. A
`
`MPI EXHIBIT 1009 PAGE 3
`
`MPI EXHIBIT 1009 PAGE 3
`
`

`

`13C NMR Ionization
` Behavior
`
`-
`
`i ----.-I
`
`t
`
`181.0
`
`180.5
`
`5
`
`1
`
`
`1
`7.0 8.6
`
`1
`7.8
`
`
`
`1
`
`1
`
`1
`
`1
`
`
`
`
`1
`
`1
`
`1
`
`1
`1
`zT4
`10.2 11.0
`V H
`FIG. 7. "C NMR chemical shift titration curves at high pH
`for [l-"C]C,,.,.BSA, 3:l mol ratio, at 34 "C. The NMR sample
`was titrated with 1 N KOH from pH 7.4 to 11.9. External tetrameth-
`ylsilane in a capillary was used as a chemical shift reference. Vertical
`error bars represent estimated uncertainties in chemical shift deter-
`minations. A , chemical shift of €-carbon (lysine) resonance as a
`function of pH. B, carboxyl chemical shift of FA carboxyl peaks b, b',
`and d (cf. Fig. 3) and BSA glutamate carboxyl peak p r (Cistola et al.,
`1987) as a function of pH.
`
`Complexes
`of Fatty Acid-Albumin
`ing protein residues in individual FA binding sites.
`Based on studies of spin-label FA analogues or detergents
`bound to albumin (Morrisett et al., 1975), it has been sug-
`gested that hydrophobic interactions are the dominant mech-
`anism by which FA bind to albumin (Spector, 1975). For alkyl
`sulfate detergents, binding affinities for albumin increase as
`the alkyl chain length increases (Karush and Sonnenberg,
`1949). Aminoazobenzene, an uncharged derivative of methyl
`orange, binds nearly as well as methyl orange itself (Klotz
`and Ayers, 1952). For organic anions, there is a large positive
`entropy change for the first mole bound but only a small
`enthalpy change (Klotz and Urquhart, 1949), suggesting that
`binding is primarily an entropy effect resulting from a release
`of water when the ligand protein complex forms (Spector,
`1975).
`However, there is also evidence that electrostatics play at
`least a minor role in ligand/albumin interactions. Based on
`indirect evidence, it has long been thought that FA bound to
`albumin are anionic at pH 7.4 (Ballou et ab, 1945). This
`conclusion was directly confirmed by the 13C NMR titration
`curves presented previously for oleic acid (Parks et al., 1983)
`and myristic acid (Hamilton et al., 1984) along with curves
`presented in this study, at least for bound FA represented by
`relative absorption of an
`NMR peak c. Furthermore, the
`anionic dye (methyl orange) bound to BSA was altered over
`a pH range in which the €-ammonium group of lysine residues
`would be expected to ionize (Klotz and Walker, 1947). In
`addition, studies using spin-label analogues of fatty acids have
`shown that methyl ester analogues bind with lower affinity
`than carboxylate ion analogues (Morrisett et al., 1975). How-
`ever, few studies have utilized native FA, and little informa-
`tion is available about the relative importance of hydrophobic
`and electrostatic interactions in individual binding sites. The
`NMR titration results presented in this study address both of
`these concerns.
`In titration from pH 8 to pH 3 (Figs. 2-6), FA bound to
`BSA exhibited two types of ionization behavior. The first
`type (peak c only) was characterized by NMR titration curves
`analogous to those obtained for protein-free monomeric FA
`in water. In contrast, the second type of ionization behavior
`(peaks a, b, b', and d ) was characterized by no changes in FA
`carboxyl chemical shift with decreasing pH.
`The ionization behavior of peak c was analogous to protein-
`free FA monomers in water according to three criteria: A6
`values, apparent pK values, and the shape of NMR titration
`curves. First, the A6 value for FA.BSA complexes (except
`c8.0.BSA) was 4.7 ppm, identical to that for aqueous protein-
`free C8.0 (Fig. 1) and aqueous short-chain carboxylic acids
`(Cistola et al., 1982a). Second, the experimentally determined
`apparent pK values of FA bound to BSA (pK = 4.1-4.3) were
`somewhat lower than values obtained for protein-free mono-
`meric FA in water (pK = 4.7-4.9; Fig. 1 and Cistola et al.,
`1982a). However, after correction of these pK, values for
`electrostatic effects3 resulting from the net positive charge on
`
`10983
`
`1
`
`
`
`~
`11.8
`
`1
`
`
`
`small but significant increase with increasing pH, and peaks
`d andpr (glutamate carboxyl) demonstrated no chemical shift
`change with increasing pH. In spectra of 6:l mole ratio
`complexes, in which peaks a and c were observed (in addition
`to peaks b, b', and d), the chemical shift of peak a decreased
`above pH 8.8 (data not shown) in a manner similar to peak
`b' (Fig. 7).
`
`DISCUSSION
`As shown in the accompanying paper (Cistola et al., 1987),
`13C NMR spectra of 13C-enriched saturated long-chain FA
`bound to BSA at pH 7.4 contained multiple FA carboxyl
`resonances corresponding to multiple FA binding sites. Three
`of these resonances (b, b', and d ) were correlated with the
`three high-affinity long-chain FA binding sites deduced from
`Scatchard analyses of binding Qata (Spector et al., 1969) and
`from Brown and Schockley's (1982) three-dimensional model
`of BSA. In the latter model, NMR peaks b, b', and d appar-
`ently represented long-chain FA bound to binding sites 1-C,
`3-C, and 2-C, respectively (Cistola et al., 1987). In contrast,
`the FA carboxyl resonance labeled c was correlated with the
`nonspecific sites such as these located in subdomains 1-AB,
`2-AB, and/or 3-AB of Brown and Schockley's model. Peak a
`represented a low-affinity (secondary) long-chain FA binding
`site in subdomain 2-AB. In this study, the ionization behavior
`of ['3C]carboxyl-enriched FA bound to BSA was determined
`by 13C NMR spectroscopy in order to assess whether ion pair
`interactions occur between FA carboxyl groups and neighbor-
`
`Intrinsic pK values (pKi,,) for FA (peak c) bound to BSA were
`estimated using the equation
`= pH + 0.868wZ
`where the term 0.868wZ takes into account the electrostatic interac-
`tions between the net charge ( Z ) of BSA at a given pH value and
`dissociating hydrogen ions (Tanford et al., 1955). Values for the
`electrostatic interaction factor (w) can be determined empirically or
`calculated from an equation derived from Debye-Huckel theory (Tan-
`ford et al., 1955). We used values of 0.052 and 0.065, respectively, in
`these calculations. The net charge Z was determined from the equa-
`tion
`
`MPI EXHIBIT 1009 PAGE 4
`
`MPI EXHIBIT 1009 PAGE 4
`
`

`

`10984
`
`b
`
`- 1-c
`
`Site
`
`- 3-c
`b '
`
`Site
`
`13C NMR Ionization Behavior of Fatty Acid-Albumin Complexes
`BSA at pH 4.1, the resulting intrinsic pK values (4.5-4.8)
`resonances. Since conformational changes have not been de-
`were essentially the same as values for protein-free mono-
`tected over this pH range (Foster, 1977), this explanation is
`unlikely. (ii) FA carboxyl ionization may have given rise to
`meric FA in water. Third, the shape of NMR chemical shift
`titration curves for peak c were very similar to those obtained
`the observed chemical shift changes in peak 6' and also
`for aqueous protein-free FA (Fig. 1). Although plots of chem-
`affected nearby lysine residues. However, the pK values of
`ical shifts (peak c) as a function of pH yielded titration curves
`carboxyl groups are generally much lower than the pH range
`which deviated from idealized Henderson-Hasselbach ioni-
`being considered here. In addition, the chemical shift of peak
`b' exhibited a much smaller A6 value than, and changed in
`zation behavior, chemical shift versus added HCI exhibited a
`the opposite direction of, that expected for FA carboxyl ioni-
`linear decrease analogous to that obtained for protein-free
`monomeric FA in water (Fig. 1 and Cistola et al., 1982a).
`zation. Therefore, this second explanation is highly unlikely.
`Thus, changes in FA chemical shifts were sensitive only to
`The most plausible explanation for changes in the molec-
`changes in FA ionization state, rather than the
`carboxyl
`ular environment of the FA carboxyl (between pH 8.8 and
`group's molecular environment (e.g. alterations caused by
`11.3) is ionization of a basic amino acid residue such as lysine,
`BSA conformational changes). The latter would have been
`tyrosine, or arginine. Arginine guanidinium groups have ex-
`characterized by a nonlinearity, a change in slope of chemical
`tremely high intrinsic pK values (>13) in small molecules
`shift titration curves, or a A6 unequal to the expected value.
`such as guanidine and substituted guanidines (Cohn and
`The linear plots also suggested that the deviation from Hen-
`Edsall, 1943; Neivelt et al., 1951) and pK values >12 in
`derson-Hasselbach ionization behavior shown in Figs. 2 A , 4A,
`proteins such as lysozyme and BSA (Tanford et al., 1955).
`and 5A arose in the pH term, not the chemical shift term, of
`For tyrosine phenolic hydroxyl groups in BSA, the experi-
`Equation I.* The most likely explanation for anomalous pH
`mentally observed pH value at the ionization midpoint (un-
`values in the carboxyl ionization region is the unmasking of
`corrected for electrostatic effects resulting from the net charge
`BSA carboxylate groups during the N-F conformational tran-
`of the protein) was 11.5 (Tanford and Roberts, 1952; Decker
`sition below pH 5 (Foster, 1977). Apparently, the N-F tran-
`and Foster, 1967). Since the observed effect on FA occurs
`sition does not give rise to anomalous FA carboxyl chemical
`between pH 8.8 and 11.5, these amino acids probably would
`shift values for the bound FA represented by peak c. However,
`not be responsible. In contrast, €-ammonium groups in BSA
`this transition does result in a loss of intensity of FA carboxyl
`have a pH midpoint for lysine ionization of approximately
`peaks (Fig. 3).4
`10.7 (Tanford et al., 1955). Chemical shift changes in the t-
`The similarities between the ionization behavior of FA
`carbon of lysine occurred over this pH range (Fig. 7A). There-
`bound to BSA (peak c) and the ionization behavior of protein-
`fore, we conclude that lysine ionization occurs between pH
`free FA monomers in water suggested that the FA carboxyl
`8.8 and 11.5 and affects the chemical environment of adjacent
`groups of bound FA represented by peak c were freely acces-
`FA carboxyl groups represented by peak b' (Fig. 7).
`sible to the aqueous solvent and, therefore, free of specific
`Taken together, the titration results for different FA allow
`electrostatic interactions with side chain residues on BSA. In
`contrast, the lack of NMR titration shifts
`for bound FA
`represented by peaks a, b, b', and d suggested that FA carboxyl
`groups in these binding sites were solvent-inaccessible and
`were probably involved in ion pair electrostatic interactions
`with basic amino acid residues that line the mouths of several
`of the putative FA binding sites on BSA (Brown and Schock-
`ley, 1982).
`In an attempt to determine which positively charged BSA
`side chain residues interact with the carboxylate groups of
`bound FA (represented by peaks a, b, b', and d ) , CM.~.BSA
`complexes (3:l and 6:l mole ratio) were titrated with KOH
`from pH 7.4 to pH 12.2. For 3:l mole ratio complexes (Fig.
`7), peak b' and the €-carbon resonance (lysine) exhibited
`substantial chemical shift changes between pH 8.8 and 11.3.
`In contrast, peak b exhibited only slight chemical shift
`changes and peaks d andpr no chemical shift changes between
`pH 8.8 and 11.3. There are three possible explanations for
`these results. (i) C,,,o. BSA may have undergone gross confor-
`mational change(s) over this pH
`range which could have
`altered the local molecular environment around
`lysine c-
`carbons and FA carboxyl carbons. Spectral changes charac-
`teristic of protein unfolding and/or peptide cleavage did occur
`at pH values >11.5. However, these changes did not occur
`between pH 8.8 and 11.3. Alternatively, a localized change in
`BSA structure might have affected only certain portions of
`the molecule and, hence, only certain of the observed NMR
`Z = 96 - r -
`where r was taken to be 80 and +cl (the moles of bound chloride ions/
`mol of BSA) 6 (Carr, 1953; Steinhardt and Reynolds, 1969). Final
`estimatedp& values were 4.6-4.7 for long-chain FA.BSA complexes
`and 4.8-4.9 for C8.,,-BSA complexes.
`The effect of BSA conformational changes on 13C NMR results
`will be addressed in detail in a future paper.
`
`d-1
`Hslir 3C.X
`0 8 - Hiss,i Thr,=- GIus3;-/
`Helix 3C.Y
`FIG. 8. Schematic diagram depicting putative locations of
`bound FA molecules in three high-affinity long-chain FA
`binding sites on BSA. The nomenclature used for a-helices and
`the amino acid sequences in the putative binding sites are
`from
`Brown and Schockley (1982). We propose that the FA represented
`by NMR peak b' is bound to site 3-C, the FA represented by peak b
`at site I-C, and the FA represented by peak d at site 2-C (see also
`Cistola et al., 1987).
`
`- d
`
`Site
`2-c
`
`@
`Qly,,~ Hi8=3- Cvs,,- C y s , d / H.lia 2A.v
`(Asp - Louzc Lou,,;- Q W , d /
`e =
` Hdix 2AB.Z
`93-1
`@&&&,= Swwj/ HJlx 2C.X
`
`BoundFA
`
` pro,,^ Q l u , ~ Tyr,-
`
`Ahm-/
`
`W i x 2C.Y
`
`Qlu,-
`
`pm,,
`
`(D
`L w . ~ l h r a d /
`Matw+/
`
` Hdir 3A.V
`Halix 3AB.Z
`
`(b.rn; Q&--F&
`9 3 C -1
`
`Bound FA
`
`Pro -&,&,g&,,"J/
`
`Qlu,, Aspw- AhU- Valu-/
`Ahc Oly,s"----J/
`cSsrc,,
`
`OY e
`
`la&,n&,U**au+/
`roln- Tyr,,~ Pho,~ TW,~+/
`
`Hslix 1 A.V
`H d i r 1 AB.2
`
`FA
`
`
`
`H*lix 1c.x
`Hdir 1 C.Y
`
`e?C -- Bound
`
`MPI EXHIBIT 1009 PAGE 5
`
`MPI EXHIBIT 1009 PAGE 5
`
`

`

`10985
`
`13C NMR Ionization Behavior of Fatty Acid-Albumin Complexes
`vidual binding sites on other proteins (e.g. intracellular fatty
`speculative correlation of BSA binding sites and NMR FA
`acid binding proteins, a-fetoprotein, etc.)
`carboxyl peaks. Inspection of the amino acid sequence at the
`mouths of the putative high affinity FA binding sites in Brown
`REFERENCES
`and Schockley's model (1982) reveals clusters of three basic
`Ballou, G. A., Boyer, P. D., and Luck, J. M. (1945) J. Biol. Chem.
`residues (Fig. 8). Binding sites 1-C and 2-C contain the triplets
`159, 111
`His-Arg-Arg (amino acids 143-145 and 334-336, respectively),
`Bell, G. H. (1973) Chem. Phys. Lipids 10,l-10
`and site 3-C contains the triplet Lys532-His533-Lys534. Of the
`Brown, J. R., and Schockley, P. (1982) in Lipid-Protein Interactions
`three NMR peaks which represent high affinity binding
`(Jost, P. C., and Griffith, 0. H., eds) Vol. 1, pp. 26-68, John Wiley
`(peaks b, b', and d; Cistola et al., 1987), only one (peak b')
`and Sons, New York
`Carr, C. W. (1953) Arch. Biochem. Biophys. 46, 417-423
`exhibits chemical shift changes which correspond with lysine
`Cistola, D. P. (1985a) Ph.D. thesis, Boston University School of
`ionization. Therefore, we propose that peak b' represents FA
`Medicine
`carboxyl carbons adjacent to the Lys-His-Lys cluster in sub-
`Cistola, D. P. (198513) Biophys. J. 47, 252a
`domain 3-C. The remaining high-affinity FA peaks ( b and d ) ,
`Cistola, D. P., Small, D. M., and Hamilton, J. A. (1982a) J. Lipid
`Res. 23, 795-799
`therefore, represent FA carboxyl carbons adjacent to His-Arg-
`Cistola, D. P., Small, D. M., and Hamilton, J. A. (1982b) Biophys. J.
`Arg clusters in sites 1-C and
`2-C (not necessarily in that
`37,204a
`order). Further inspection of the amino acid sequences in
`Cistola, D. P., Atkinson, D.,