Metal Ion Complexes of EDTA as Solutes for Density Gradient

Sep 28, 2005 - In the study reported here, we study the nature of the metal ion complexes of EDTA as solute systems for analysis of lipoproteins by de...
0 downloads 5 Views 555KB Size
Download PDF
Anal. Chem. 2005, 77, 7054-7061

Metal Ion Complexes of EDTA as Solutes for Density Gradient Ultracentrifugation: Influence of Metal Ions Jeffery D. Johnson, Natalie J. Bell, Erin L. Donahoe, and Ronald D. Macfarlane*

Department of Chemistry, Texas A&M University, College Station, Texas 77843

In the study reported here, we study the nature of the metal ion complexes of EDTA as solute systems for analysis of lipoproteins by density gradient ultracentrifugation (DGU) by varying both the complexing metal ion and the counterion. Specifically, the sodium and cesium salts of complexes of Bi/EDTA, Pb/EDTA, Cd/EDTA, Fe/ EDTA, and Cu/EDTA were chosen for this study. We show that useful gradients can be formed within a few hours beginning with a homogeneous solution. Data are presented that provide insight into the nature of how these gradients are formed from these complexes and how the selection of a specific complex can be used to enhance particular regions of the lipoprotein density profile for clinical studies. We also examine the use of equilibrium sedimentation theory to correlate the measured density profiles generated by these complexes with their molecular weight. Recently, the cesium salt of bismuth EDTA was described as a novel solute for forming a self-generating density gradient from a homogeneous solution during ultracentrifugation of serum samples for the separation and quantitation of lipoproteins.1 Its ease of synthesis, ready solubility in water, and ability to selfgenerate a density gradient that spans the density range of lipoproteins (1.00-1.20 g/mL) in a short amount of time makes it a very useful solute system for lipoproteomics and offers an alternative to earlier methods that use KBr as the density gradient forming solute.2,3 An appealing feature about solutes involving the EDTA ligand is that the molecular weight can be modified by changing the complexing ion and counterion, thus effectively modifying the resulting density gradient. In this study, we investigate the metal ion/EDTA solute system in more detail, studying the influence of changing both the complexing ion and the counterion on the density gradient that is formed. In particular, we focus on comparisons between the sodium and cesium counterions and on complexing ions spanning an atomic weight range from 55 to 209 amu. Here, we measure the density profiles of the EDTA complexes of Cu, Fe, Cd, Pb, * To whom correspondence should be addressed. Phone: 979-845-2021. Fax: 979-845-8987. E-mail: [email protected]. (1) Hosken, B. D.; Cockrill, S. L.; Macfarlane, R. D. Anal. Chem. 2005, 77, 200-207. (2) Swinkels, D. W.; Hak-Lemmers, H. L.; Demacker, P. N. J. Lipid Res. 1987, 28, 1233-1239. (3) Bozoky, Z.; Fulop, L.; Kohidai, L. Fur. Biophys. J. 2001, 29, 621-627.

7054 Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

and Bi using Na+ and Cs+ counterions and compare the lipoprotein particle density profiles obtained with these solute systems. A series of complexing ions spanning a large range of atomic weights were chosen to understand the link between molecular weight of the solute and shape of the resulting density gradient. The comparison of the Pb/EDTA and Bi/EDTA solute systems also gives some information on the influence of the charge of the complexing ion on the lipoprotein separation. In our postseparation analysis of lipoprotein subfractions in capillary electrophoresis, the strong ultraviolet absorbance of the complexes can interfere with analyses, but there is a possibility that this property could be used to our advantage, since the specific wavelength of this absorbance is linked to the complexing metal ion. In addition, we were curious about the influence of different metal ion complexes of EDTA solutions on the fluorescence properties of the fluorophore that we use for a lipoprotein marker. Since the binding properties of this fluorophore to lipoproteins have been previously studied, it is important to determine what effect, if any, the different metal ion complexes of EDTA will have on quantification based on the fluorescence yield.4,5 EXPERIMENTAL SECTION Materials. The fluorophore 6-((N-(7-nitrobenz-2-oxa-1,3-diazol4-yl)amino)-hexanoyl) sphingosine, NBD (C6-ceramide), was obtained from Molecular (Eugene, OR). Iron EDTA monosodium salt, copper EDTA monosodium salt, EDTA, and cesium carbonate were from Sigma Aldrich (St. Louis, MO). Sodium carbonate was from J. T. Baker (Phillipsburg, NJ). Deionized water used in all experiments was from a Milli-Q water purification system (Millipore, Bedford, MA). Blood Draw. The serum for this study was obtained from a normolipidemic subject following a 12 h fast via blood draw into a 9.5-mL Vacutainer treated with polymer gel and silica activator (366510, Becton Dickinson Systems, Franklin Lakes, NJ). The serum was separated from the red blood cells by centrifugation at 3200 rpm for 20 min at 5 °C and separated into 250-µL aliquots. Synthesis of EDTA Complexes. The various EDTA complexes were synthesized from H4EDTA; the appropriate alkali carbonate; and the heavy metal oxide, hydroxide, carbonate, or oxycarbonate using a procedure similar to that described else(4) Schmitz, G.; Mollers, C.; Richter, V. Electrophoresis 1997, 18, 1807-1813. (5) Henriquez, R. R.; Johnson, J. D.; Farwig, Z. N.; Chandra, R.; Macfarlane, R. D. Anal. Chem. 2005, in preparation. 10.1021/ac0509657 CCC: $30.25

© 2005 American Chemical Society Published on Web 09/28/2005

where.6,7 The reagents were combined stoichiometrically in 100 mL of DI H2O, followed by a 2-h reflux, yielding a clear solution. Sodium carbonate or cesium carbonate was then added to the clear solution to bring the final pH range to 6-7. The final solution volume was reconstituted to 100 mL to account for evaporation during reflux to give stoichiometric solutions with a final concentration of 0.200 M. Preparation of EDTA Complexes for Density Calibration. A portion of the 0.200 M stock EDTA complex solutions was concentrated to a second stock solution at 0.400 M for calibration purposes. From this new stock solution, a set of dilutions were made to cover the dynamic range of lipoprotein densities expected from an ultracentrifuge spin. Measurements were performed on each dilution at room temperature to obtain the refractive index, density, and absorbance. Measurement of Solution Properties. The measurement of the density of each dilution was performed gravimetrically by recording the weight of a 1000-µL aliquot. A Bellingham & Stanley refractometer (60/DR, Lawrenceville, GA) was used to monitor the refractive index of 20-µL aliquots of each solution at room temperature. A Biotek Instruments µ-Quant reader (MQX200, Winooski, VT) was used to measure the absorbance spectrum in the UV-visible range. Ultracentrifugation of the Salts. A 1000-µL volume of the metal ion EDTA salt, at a concentration of 0.200 M in deionized water, was transferred to a 1.5-mL, thick-walled, polycarbonate ultracentrifuge tube. These solutions were spun for 6 h at 120 000 rpm and 5 °C in a Beckman Optima TLX-120 Ultracentrifuge equipped with a 30° fixed angle TLA 120.2 rotor. Fluorescence Labeling of Serum Lipoproteins. The flourophore NBD (C6-ceramide) was reconstituted to a density of 1 g/mL by addition of DMSO and subsequently added to a sample of serum.2 The density gradient forming solute was then added, and the mixture was allowed to incubate for 30 min at room temperature. Imaging. A fluorescence imaging system was developed to give increased sensitivity and versatility to the lipoprotein fingerprinting method reported previously.8 The light source used was a Fiber-Lite MH-100 Illuminator, (MH100A, Edmund Industrial Optics). The camera used was a digital color microscope camera (S99808, Optronics, Goleta, CA). The camera and light source were placed orthogonally to each other on an optical bench, and a slit (1 cm × 4 cm) was placed 8 cm away from the tube holder and suspended by a post/post holder to collimate the excitation beam. A gain of 1.0000 and an exposure time of 15.8 ms were chosen using the accompanying MicroFire camera software. A blue-violet excitation filter (BG-12, Schott, Edmund Industrial Optics) with a bandwidth centered at 407 nm and a yellow emission filter (OG-515, OEM, Edmund Industrial Optics) with a bandwidth centered at 570 nm were used and were chosen to match the NBD (C6-ceramide) excitation and emission. Measurement of Density Profile. A 20-µL volume Pipetman pipet (P-20, Rainin Instruments, Woburn, MA) was fixed to a ring (6) Davidovich, R. L.; Logvinova, V. B.; Kaidalova, T. A. Russ. J. Coord. Chem. 1998, 24, 399-404. (7) Davidovich, R. L.; Gerasimenko, A. V.; Logvinova, V. B. Russ. J. Inorg. Chem. 2001, 46, 1518-1523. (8) Macfarlane, R. D.; Hosken, B. D.; Farwig, Z. N.; Espinosa, I. L.; Myers, C. L.; Cockrill, S. L. Lipoprotein fingerprinting method. U.S. Patent 6,753,185, 2004.

stand holder mounted on top of a stainless steel positioning lift (E36-283, Edmund Industrial Optics). This apparatus was used to position the pipet tip in specific locations in the ultracentrifuge tube containing freshly spun samples containing only the EDTA salt solutions. Serum was not added to these samples for this portion of the experiment. The Microfire camera was used to record the vertical pixel location of the pipet tip at the sampling position from tube depths ranging from 10 to 34 mm. Details of converting these pixel values to vertical dimensions are described in a later portion of the Experimental Section. Aliquots of the salt solution were taken in ∼2-mm increments until the bottom of the ultracentrifuge tube was reached, yielding 10-11 aliquots of 20 µL per ultracentrifuge tube. Each of these aliquots was used in the determination of their respective refractive indices. Preparation of Serum Samples for UC Spin. Into a 1.5-mL Eppendorf tube, 1100 µL of 0.200 M solution of the metal ion EDTA salt, 60 µL of serum, and 10 µL of 2 mg/mL NBD (C6ceramide) in DMSO were added, and the tubes were vortexed at 1400 rpm for 1 min and allowed to stain for a period of 30 min. A 1000-µL aliquot of this solution was transferred to a 34- × 11-mm polycarbonate, open-top, centrifuge tube and spun as previously described. This procedure was repeated for each of the eight solute systems being studied. Establishing the Tube Coordinate. When an image of the tube is generated by Microfire, it is displayed as an 1800 × 1200 matrix of pixels. These pixel values were converted to tube coordinates by imaging previously measured lines etched onto the surface of an ultracentrifuge tube. Origin 7.0 (Microcal Software Inc., Northampton, MA) was used for the analysis of all images and calibrations by measuring the intensity values over a horizontal range of 10 pixels near the middle of the tube, calculating their average, and plotting these average intensity values against their respective tube depth. Origin’s peak-fitting module was used to determine the centroid of these calibrated lines from the camera image. Generation of the Lipoprotein Density Profile. Following an ultracentrifuge spin at the previously described settings, 200 µL of DI H2O was carefully layered atop the freshly spun solutions to provide a separation of the VLDL band from the meniscus at the top of the solution. The Microfire camera system was used to record images of the ultracentrifuge tube, and Origin software was used to convert the pixel values into intensities that could be plotted against the tube coordinate, thereby establishing a lipoprotein density profile. RESULTS AND DISCUSSION The long-term objective of this project was to develop a library of aqueous metal-ion complexes of EDTA that can be used to generate a family of density gradients in the ultracentrifuge for highlighting features of the serum lipoprotein particle density profile for clinical studies. The regions of the density profile located between VLDL and LDL and between LDL and HDL are particularly important for this application. The density profile can be altered by three approaches: adjusting the rotor speed of the centrifuge, varying the initial concentration of the solute, or using the molecular weight of the solute as the variable. In this study, we investigated the latter approach. We first determined the viability of this method by ascertaining whether the particular solute systems generated interpretable lipoprotein density profiles. Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

7055

Table 1. Calibration of Density vs Refractive Index salt

slope (m)

y-intercept (y0)

R2

CsBiEDTA NaBiEDTA Cs2PbEDTA Na2PbEDTA Cs2CdEDTA Na2CdEDTA Na2CuEDTA NaFeEDTA

5.4548 4.7647 6.7665 5.016 6.612 4.4776 4.0663 3.3795

-6.2502 -5.3525 -8.0089 -5.6811 -7.8024 -4.9596 -4.4191 -3.5019

0.9954 0.9896 0.9956 0.993 0.9961 0.9904 0.9901 0.9885

Each of the solute systems reported in this study did, but to varying degrees. Following the protocol established in our initial studies using CsBiEDTA, we then measured the density profiles generated by these solutes, as described in the Experimental Section, and studied the relevant solution properties of each of the complexes. Physical Properties of Aqueous EDTA Complexes. Upon completion of the syntheses of the various solutes as previously described, solutions with densities ranging from 1.04 g/mL for NaFeEDTA to 1.12 g/mL for Cs2PbEDTA were obtained for solutions at a concentration of 0.200 M. It was previously reported that the CsBiEDTA solution has a solubility in water of 1.351 g/mL, which is considerably higher than the maximum density of the various lipoprotein subclasses and lipid-free serum proteins.1 In each case, the solubility of each of these complexes is higher than the maximum density of proteins expected in an ultracentrifuge spin. A link was established between the concentration, density, and refractive index of each solution at varying solute concentrations to better understand their physical properties and to serve as a means of measuring the density profile that is generated after an ultracentrifuge spin. It was found that there was a linear relationship linking these three physical properties over the density range of interest, resulting in a useful calibration of solution density versus refractive index for each solution. Because of the small sample volume that is required, it was decided that measuring the refractive index of aliquots of the solute at varying ultracentrifuge tube depths was the most efficient and precise method of determining the density profile. This relationship is given by the following equation,

F ) mη + y0

(1)

where F is the solution density, m is the slope of the curve, η is the refractive index of the solution, and y0 is the y-intercept. In each case, this relationship was found to be linear over the density range of interest (1.00 to 1.30 g/mL). These results are summarized in Table 1. The slopes of the density/refractive index relationship correlated linearly with the molecular weight of the solute. Comparisons of Self-Generating Density Gradient Profiles. The density profiles generated in the ultracentrifuge tubes were measured by sampling the gradient at specific depths and determining their refractive indices, as described in the Experimental Section. The shape of the resulting profile is dependent on both the complexing metal ion and the counterion in each of the EDTA complexes. Figure 1 depicts the manner in which the 7056

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

density profile and resulting lipoprotein profile is affected by altering the complexing ion among Bi3+, Pb2+, and Cd2+ and by altering the counterion between Cs+ and Na+. In this figure, the intensity of fluorescence of the fluorescently labeled lipoproteins is plotted against the vertical tube coordinate and is overlaid by a curve depicting the measured density of the EDTA complex as a function of the vertical tube coordinate. In these lipoprotein profiles, there are five clearly defined areas: chylomicron/ meniscus, VLDL, LDL, HDL, and protein regions. As described in the Experimental Section, 200 µL of DI H2O is layered atop the freshly spun solution in order to separate the chylomicron/ meniscus region from the VLDL region, which would otherwise be collapsed into a single visible peak. This separation occurs because the density of chylomicrons is less than that of water, and the density of VLDL is slightly higher. Since light from the fluorescence of the stained lipoproteins can be scattered from the meniscus of the solution, this effect adds to the intensity in the chylomicron/meniscus region. There is also a sharp distinguishable peak observed near 28 mm in 1B, 1D, 1E, and 1F that occurs as a result of light scattering from the seam that joins the cylindrical portion of the ultracentrifuge tube with the curved portion at the bottom during fabrication. Various features become obvious when a comparison is made among each of the spun solutions at the same initial concentration of 0.200 M. Solute Systems of Bi/EDTA. In EDTA complexes containing bismuth, as seen in Figure 1A and B, a near-baseline separation is achieved among all five regions, although the separation between VLDL and LDL is greater when using CsBiEDTA, and the separation between LDL and HDL is greater when using NaBiEDTA. In addition, since the density profile is not as steep for the NaBiEDTA solution, the resulting LDL and HDL subclasses tend to be expanded more than when using CsBiEDTA as the density gradient generating medium. This proves that simply changing the counterion from cesium to sodium can have a large impact on the resulting density gradient. Solute Systems of Pb/EDTA. When Pb2+ is used as the complexing metal ion, as opposed to Bi3+, two counterions of Cs+ or Na+ will be associated with the EDTA complex, resulting in compounds of Cs2PbEDTA and Na2PbEDTA that have a higher molecular weight than the corresponding bismuth complexes. In the cases seen in Figure 1C and D, the shape of the density gradient results in an HDL region that is more expanded relative to that seen in Figure 1A and B, and the LDL region is forced to a lower position on the tube coordinate scale (i.e., closer to the meniscus), since the gradient is more linear for the Pb/EDTA complexes than the Bi/EDTA complexes. Solute Systems of Cd/EDTA. As in the case of the Pb/EDTA systems, when cadmium is used as the complexing metal ion, two counterions will associate with the EDTA complex. The result is that the Cd/EDTA complexes have a molecular weight between the Bi/EDTA and Pb/EDTA complexes. In the Cs2CdEDTA solute system seen in Figure 1E, the slope of the density gradient is less steep than the Cs2PbEDTA system, causing the HDL and protein regions of the lipoprotein profile to shift to a higher position on the tube coordinate scale. There is essentially no difference in the LDL region of the lipoprotein profiles between these two solute systems, but the intensity of the VLDL peak in the Cs2CdEDTA system is elevated. The density gradient formed

Figure 1. Density and lipoprotein profiles. By using 0.200 M solutions of (A) CsBiEDTA, (B) NaBiEDTA, (C) Cs2PbEDTA, (D) Na2PbEDTA), (E) Cs2CdEDTA, and (F) Na2CdEDTA, various features are enhanced for a particular serum sample. Gradients were formed by centrifuging 0.200 M solutions for 6 h at 120 000 rpm and 5 °C.

by Na2CdEDTA in Figure 1F, however, is relatively linear, resulting in two noticeable alterations to the lipoprotein profile. Since the density of this complex does not exceed 1.10 g/mL, the HDL contained in the sample is not separated from the higher density protein region. Additionally,, the LDL region of the profile is elongated to such an extent that three discernible subclasses can be seen. Solute Systems of Na2CuEDTA and NaFeEDTA. The low molecular weight complexes of Na2CuEDTA and NaFeEDTA yield density gradients that are shallower than the solution of Na2CdEDTA, thereby making them impractical for separating the

different lipoprotein subclasses. In addition, the NBD-ceramide intensities for the lipoprotein subclasses are considerably smaller than for the other solute systems due to fluorescence quenching. The only regions of the lipoprotein profiles that are visible are the chylomicron/meniscus and VLDL regions, where the concentrations of these EDTA complexes are significantly lower in the region at a low tube coordinate. Fluorescence quenching occurs as a result of binding of the NBD fluorophore to Fe3+ ions.9 Binding of Fe3+ to the NBD fluorophore competes with binding (9) Ramachandram, B.; Samanta, A J. Phys. Chem. A 1998, 102, 10579-10587.

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

7057

Table 2. Effect of Solute on Density of Lipoproteins solute

LDL density (g/mL)

HDL density (g/mL)

CsBiEDTA NaBiEDTA Cs2PbEDTA Na2PbEDTA Cs2CdEDTA Na2CdEDTA

1.0588 1.0396 1.0571 1.0480 1.0574 1.0491

1.1473 1.1179 1.1296 1.1204 1.1378 1.1002

to the EDTA ligand and can only be reversed at low pH with an excess of EDTA present.10 Since Cu2+ is also a transition metal with empty d orbitals, quenching occurs here as well, although to a lesser extent, since the copper ion is more tightly bound to the EDTA ligand.11 This result is particularly interesting because it indicates that these paramagnetic ions must be in close proximity to the NBD-ceramide fluorophores that are imbedded in the hydrophobic domains of the lipoproteins. The Sodium/Cesium Effect. It can be seen that solutions of complexes containing the Cs+ counterion yield a steeper density profile than those containing the Na+ counterion in each case due to the inherently higher molecular weight of these complexes. It was also noticed that when a serum sample is spun in a solute containing Cs+, the corresponding density of the lipoprotein subclasses are higher than those spun in a solute containing the Na+ ion. The results of this effect can be seen in Table 2. These lipoproteins are macromolecular anions at a pH of 7, so the Na+ and Cs+ ions will form an ionic atmosphere around them and increase their density relative to their natural state. Since the Cs+ ion is heaver than the Na+ ion, the density of the lipoprotein can be greatly altered by substituting the counterion of the EDTA complex. These findings suggest that the ionic atmosphere configuration is retained under the influence of the forces operating during ultracentrifugation as the lipoproteins migrate to their isopycnic positions. Application to Lipoprotein Density Profiling. During developmental work, various ultracentrifugation rotor speeds and times were studied to determine the optimal conditions for separating lipoprotein subclasses. The final rotor speed of 120 000 rpm was chosen to speed up the equilibration process between sedimentation and diffusion of the particles contained in the ultracentrifuge tube and yield a sufficient separation of the lipoprotein molecules.1 The two primary methods of modifying the lipoprotein density profiles discussed here, however, involve altering the concentration of the gradient-generating solute or by changing the solute system altogether. In the case of the latter, it becomes possible to tailor the lipoprotein profiles to enhance specific areas, such as the region between VLDL and LDL or the region between LDL and HDL. An example of this application would be in the study of Lp(a) and Apo C-1 enriched HDL that are located between the LDL and HDL regions of the lipoprotein profile; Na2PbEDTA and Na2CdEDTA, Figure 1D and F, respectively, would give the best separation for this region of interest. Theory of Density Gradient Formation. During ultracentrifugation, there are two primary forces at work that oppose each (10) Lytton, S. D.; Mester, B.; Libman, J.; Shanzer, A.; Cabantchik, Z. I. Anal. Biochem. 1992, 205, 326-333. (11) Martell, A. E.; Smith, R. M. Critical Stability Constants; Plenum: New York, 1974; Vol. I, pp 204-211.

7058

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

Table 3. Density Profile Equations for Each Solute System salt

A

B

y-intercept (y0)

R2

CsBiEDTA NaBiEDTA Cs2PbEDTA Na2PbEDTA Cs2CdEDTA Na2CdEDTA Na2CuEDTA NaFeEDTA

0.000 39 0.000 43 0.015 78 0.012 16 0.019 37 0.017 20 0.010 58 0.003 79

4.463 20 4.875 21 10.215 80 12.008 42 12.236 43 17.539 58 16.319 34 11.059 98

1.028 40 1.017 73 0.984 60 1.006 14 0.985 98 1.005 11 1.011 42 1.011 39

0.986 39 0.991 79 0.989 47 0.987 49 0.982 72 0.991 80 0.991 88 0.993 53

other: sedimentation of molecules as a result of the centrifugal force’s being applied by the spinning rotor, and diffusion. As equilibrium is reached between these two forces, a stable density gradient is generated. The density profile data can be fitted to an exponential function given by the following equation,

F ) Ae(x/B) + y0

(2)

where F is the solution density, x is the tube coordinate, A and B are constants, and y0 is the y-intercept. The results for each of the eight solute systems are seen in Table 3. By plotting the natural logarithm of the concentration ratios of the density gradient forming solute versus the square of their relative positions in the ultracentrifuge tube, linear relationships are seen that correlate strongly with the molecular weight of the solute. This relationship is given by the following equation,

Mw )

ln(c2/c1)(2RT) 2

(r2 - r12)(1 - vF)ω2

(3)

where c2/c1 represents the concentration ratio between two points in an ultracentrifuge tube at radii of r2 and r1, Mw is the molecular weight of the solute, ν is the partial specific volume of the solute, F is the solute density, ω is the angular velocity of the rotor, R is the gas constant, and T is the temperature.12 For simplification purposes, it is possible to substitute the solution density for the c2/c1 term and (tube coordinate)2 for the (r22 - r12) term. Figure 3 demonstrates how the molecular weight of the sodium salts of Bi/EDTA, Pb/EDTA, Cd/EDTA, and Fe/EDTA affect the slope of this relationship. The slopes of the lines generated from the density profiles of solutes with a higher molecular weight, such as NaBiEDTA and Na2PbEDTA, were higher than those with a low molecular weight, such as NaFeEDTA. The cesium complexes followed the same trend, but they had higher slopes than their corresponding sodium salts. The partial specific volume of each solute plays an important role in determining the slope of the density gradient that is generated, with lower values yielding steeper gradients. Values for the molar properties for the complexes of Na2CuEDTA, NaFeEDTA, Na2CdEDTA, and Na2PbEDTA have been reported previously.13,14 When the similar partial molar volumes of these solutes are divided by their (12) Atkins, P. Physical Chemistry, 6th ed.; W. H. Freeman & Co.: New York, 1998; pp 686-689. (13) Hovey, J. K.; Tremaine, P. R. J. Phys. Chem. 1985, 89, 5541-5549. (14) Hovey, J. K.; Hepler, L. G. Inorg. Chem. 1988, 27, 3442-3446.

Figure 4. Slope of density (eq 3) vs MW of solute. By comparing the slope calculated from eq 3 with the MW of the solute systems, a linear relationship is seen.

Figure 2. Fluorescence quenching. The lipoprotein profiles for (A) Na2CuEDTA and (B) NaFeEDTA show that the fluorescence of the NBD (C6-ceramide) fluorophore is quenched by these solute systems.

Figure 3. Plot of ln(density) vs coordinate2. The molecular weights of NaBiEDTA -9-, Na2PbEDTA -b-, Na2CdEDTA -2-, and NaFeEDTA -[- alter the slope of this relationship.

respective molecular weights, it becomes clear that those with the highest molecular weight will have the lowest partial specific volume, which correlates to a steeper density gradient. The relatively high partial specific volumes of the Na2CuEDTA and NaFeEDTA complexes relative to the other complexes explain why such shallow-density gradients are generated from an ultracentrifuge spin. Data for the partial molar volumes of various cations reveal that the Cs+ ion has a larger partial molar volume

than the Na+ ion, which would indicate that a complex containing the Cs+ counterion, (i.e., CsBiEDTA) should have a higher partial molar volume than its corresponding complex containing the Na + counterion (i.e., NaBiEDTA).15,16 However, since the contribution of the mass of the counterion to the molecular weight of the solute is much greater in this case than the contribution seen in the partial molar volume, it is expected that a complex containing the Cs+ counterion would have a smaller partial specific volume than the corresponding complex containing the Na+ counterion. The net effect is a steeper density gradient for the EDTA complexes containing the Cs+ counterion. Since each of the solute systems in this study were observed to display a similar relationship, it is also possible to predict the slope of the resulting density gradient on the basis of the molecular weight of the solute alone by plotting the slopes of eq 3 against the molecular weight of the solute system, as seen in Figure 4. The close fit of this line suggests that our data agree with the trend expected from theory. Absorbance Properties. The use of the ultracentrifuge serves a dual role of not only providing a useful lipoprotein profile, but also serving as a means of preseparating the lipoprotein subclasses prior to analysis by capillary electrophoresis and mass spectrometry. It was previously known that the Bi/EDTA complexes displayed a mobility similar to that of LDL molecules at physiological pH, and its strong UV absorbance proved to be problematic due to spectral interference. The absorbance properties of these solute systems were studied to gain understanding about how they may influence the analysis of lipoprotein subclasses by capillary electrophoresis following an ultracentrifuge spin. With the exception of the copper and iron complexes, the various solutions are transparent in the visible region but absorb in the ultraviolet region. Interestingly, the maximum absorbance wavelength of each solution is strongly dependent on the metal ion that is chelated to the EDTA ligand. Figure 5 demonstrates the relative differences in absorbance between the various EDTA complexes and particularly emphasizes the comparatively low absorbance of the cadmium complexes. The concentration of these (15) Hovey, J. K.; Hepler, L. G., Tremaine, P. R. J. Solution Chem. 1986, 15 (12), 977-987. (16) Barta, L.; Hepler, L. G. J. Phys. Chem. 1989, 93, 5588-5595.

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

7059

Figure 5. Absorbance spectra of the solutes. UV-visible spectra were measured for 0.400 mM solutions containing complexes of (A) CsBiEDTA -2-, NaBiEDTA -b-, (B) Cs2PbEDTA -2-, Na2PbEDTA -b-, (C) Cs2CdEDTA -2-, Na2CdEDTA -b-, (D) Na2CuEDTA -2-, and NaFeEDTA -b-.

solutions was sufficiently high to illustrate this point, but much more dilute solutions were analyzed for Beer’s Law relationships. Complexes containing Bi3+ have a maximum absorbance near 210 nm and at 264 nm, whereas those containing Pb2+ have a maximum absorbance near 210 and 241 nm. Cadmium-based complexes have only a single maximum absorbance located between 204 and 206 nm. The copper and iron complexes have a very broad absorbance that encompasses the majority of the ultraviolet spectrum. A set of dilutions in the linear range for each stock solution was made to determine the molar absorptivity of each EDTA complex by Beer’s Law. These results can be seen in Table 4. It can be seen that complexes containing the same complexing ion have very similar absorption properties, though the complexes containing cadmium have an approximately 10fold reduction in molar absorptivity from the others. The absorptivity of various other complexes containing the Bi3+, Pb2+, and Cd2+ ions as a result of charge transfer between the ligand and metal have been examined elsewhere.17-20 This effect is seen (17) Oldenberg, K.; Vogler, A. J. Organomet. Chem. 1996, 515, 245-248. (18) Oldenberg, K.; Vogler, A. Z. Naturforsch. 1993, 48b, 1519-1523.

7060 Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

Table 4. Molar Absorptivity at the Maximum Absorbance Wavelength for Each Solute salt

wavelength (nm)

 (L cm-1 mol-1)

CsBiEDTA NaBiEDTA Cs2PbEDTA Na2PbEDTA Cs2CdEDTA Na2CdEDTA Na2CuEDTA NaFeEDTA

264 264 241 241 206 204 250 256

7889 7855 7527 7517 780 717 2219 7029

because the Bi3+ and Pb2+ ions are capable of undergoing charge transfer between the ligand and metal, whereas this process is presumably less favorable for the Cd2+ ion due to its filled d shell. Since the Cd/EDTA complexes are doubly charged and have a relatively low absorptivity as opposed to the much higher absorp(19) Busenlehner, L. S.; Cosper, N. J.; Scott, R. A.; Rosen, B. P.; Wong, M. D.; Giedroc, D. P. Biochemistry 2001, 40, 4426-4436. (20) Busenlehner, L. S.; Apuy, J. L.; Giedroc, D. P. J. Biol. Inorg. Chem. 2002, 7, 551-559.

tivity of the singly charged Bi/EDTA complexes, it is currently perceived that they may serve as a better choice for a densitygradient-forming solute for both their ability to separate lipoprotein subclasses effectively and for their potentially favorable absorbance and electrophoretic properties. CONCLUSIONS We have shown that the density gradient that is formed by ultracentrifugation of EDTA complexes can be tailored by altering the complexing ion and counterion. Pairing this versatility with the relative ease of synthesizing the different EDTA complexes makes it possible to generate a density gradient that results in a favorable separation of lipoprotein subclasses. By linking equilibrium sedimentation theory with the measured density profiles, it should also be possible to predict the resulting density gradient

from the use of new solute systems from this family of complexes. As a result, we have added a valuable preparative tool that offers extended flexibility to our lipoprotein fingerprinting method. ACKNOWLEDGMENT This work was supported by the NIH Heart, Blood, & Lung Institute (HL 68794). We thank Brian Hosken for his expertise and assistance in the initial stages of this study and Mike Nowlin for his valuable contributions in the development of the method currently used to digitally image the lipoprotein subclasses.

Received for review June 1, 2005. Accepted August 16, 2005. AC0509657

Analytical Chemistry, Vol. 77, No. 21, November 1, 2005

7061