Method for Lipoprotein (a) Density Profiling by BiEDTA Differential

Dec 14, 2005 - College Station, Texas 77843-3255, and Scott & White Memorial Hospital and Texas A&M University Health Science. Center, Temple, Texas 7...
0 downloads 0 Views 189KB Size
Download PDF
Anal. Chem. 2006, 78, 438-444

Method for Lipoprotein(a) Density Profiling by BiEDTA Differential Density Lipoprotein Ultracentrifugation I. Leticia Espinosa,† Catherine J. McNeal,‡ and Ronald D. Macfarlane*,†

Laboratory for Cardiovascular Chemistry, Department of Chemistry, Texas A&M University, College Station, Texas 77843-3255, and Scott & White Memorial Hospital and Texas A&M University Health Science Center, Temple, Texas 76508

In this article, we demonstrate the analytical power of linking density gradient ultracentrifugation with affinity separations. Here we address some of the analytical challenges in the study of lipoprotein(a), (Lp(a)). The mean density distribution of Lp(a) was determined by a differential density lipoprotein profile (DDLP). For DDLP, the lipoprotein density distribution of a serum sample with elevated Lp(a) levels was determined by ultracentrifugation using a BiEDTA complex as a density gradient. Lp(a) was removed from a second aliquot of the same serum sample by carbohydrate affinity using wheat germ agglutinin (WGA). WGA was demonstrated to have high specificity for Lp(a) in a serum sample. This sample was ultracentrifuged to obtain a lipoprotein density distribution in the absence of Lp(a). A DDLP was obtained after subtracting the Lp(a)-depleted lipoprotein density profile from the untreated lipoprotein density profile. The DDLP methodology reported herein gives relevant information of the lipoproteins in serum such as density, isoform, and subclass characteristics. Lp(a) was quantitatively isolated from serum with a recovery efficiency of 82%. Lp(a) was purified by ultracentrifugation. Lp(a) retained its inherent density (1.086 g/mL) and immunoreactivity. The major outcome of this research was the effectiveness of using affinity separations coupled with density ultracentrifugation for the isolation of pure Lp(a) from serum and its isoform characterization based on density by DDLP. Cardiovascular disease is the number one cause of death in America. A better understanding of the emerging risk factors for cardiovascular disease (CVD) is needed in order to determine the steps toward its control. One of such emerging risk factors is lipoprotein(a), (Lp(a)).1-4 Berg discovered Lp(a) in 1963.5 Since then much progress in the study on the structural and functional properties of Lp(a) has been * To whom correspondence should be addressed. E-mail: macfarlane@ mail.chem.tamu.edu. Phone: 979-845-2021. Fax: (979) 845-8987. † Texas A&M University. ‡ Scott & White Memorial Hospital and Texas A&M University Health Science Center. (1) Kostner, K. M.; Kostner, G. M. Curr. Opin. Lipidol. 2002, 13, 391-396. (2) Scanu, A. Curr. Artherosclerosis Rep. 2003, 5, 106-113. (3) Kronenberg, F.; Utermann, G. Lipoprotein(a). In Enclyclopedia of Endocrine Diseases; San Diego, Academic Press, 2004; pp 188-196.

438 Analytical Chemistry, Vol. 78, No. 2, January 15, 2006

made. Understanding the role of Lp(a) in the development of atherosclerotic cardiovascular disease is necessary for the eventual development of clinical therapies. The complexity of the Lp(a) molecule sets a challenge for the determination of the risk it represents for the cardiovascular system. The existence of up to 34 apolipoprotein(a), (apo(a)), isoforms as well as the inverse relationship between apo(a) molecular mass/atherogenicity, raises demands for the development of efficient methods for apo(a) phenotyping.6 This paper addresses some of the analytical challenges in using Lp(a) as a CVD marker. These analytical challenges include its isolation, characterization, and quantitation. Lp(a) is a low-density lipoprotein (LDL)-like particle formed by the covalent link of the unique glycoprotein apo(a) to one molecule of LDL’s apolipoprotein B-100 (apoB-100).7-9 Apo(a) isoform heterogeneity results in a wide range of apo(a) molecular weights from 280 000 to 800 000 10 and a density range of 1.05 to 1.1 g/mL.11 The concentration and size of Lp(a) are genetically determined and inversely related, the smaller isoforms being the more atherogenic.3 Standardized methods for Lp(a) isolation are not available. The most common method for the isolation of Lp(a) from plasma is sequential ultracentrifugation.12-14 Gel filtration and affinity chromatography on heparin-Sepharose15 or CNBr-activated lysineSepharose 8 are two methods used to purify isolated Lp(a). The process of isolating Lp(a) by ultracentrifugation and further (4) Homma, Y. J. Atherosclerosis Thromb. 2004, 11, 265-270. (5) Berg, K. Acta Pathol. Microbiol. Scand. 1963, 59, 369-382. (6) Marcovina, S. M.; Albers, J. J.; Gabel, B.; Koschinsky, M. L.; Gaur, V. P. Clin. Chem. 1995, 41, 246-255. (7) Kostner, K. M.; Kostner, G. M. Drugs News Perspect. 2002, 15, 69-75. (8) Weisel, J. W.; Nagaswami, C.; Woodhead, J. L.; Higazi, A. A. R.; Cain, W. J.; Marcovina, S. M.; Koschinsky, M. L.; Cines, D. B.; Bdeir, K. Biochemistry 2001, 40, 10424-10435. (9) Gerglund, L.; Ramakrishnan, R. Arteriosclerosis Thromb. Vasc. Biol. 2004, 24, 2219-2226. (10) Marcovina, S. M.; Zhang, Z. H.; Gaur, V. P.; Albers, J. J. Biochem. Biophys. Res. Commun. 1993, 191, 1192-1196. (11) Utermann, G.; Menzel, H. J.; Kraft, H. G.; Duba, H. C.; Kemmler, H. G.; Seitz, C. J. Clin. Invest. 1987, 80, 458-465. (12) Gabel, B. R.; Koschinsky, M. L. Biochemistry 1998, 37, 7892-7898. (13) Kang, C.; Dominguez, M.; Loyau, S.; Miyata, T.; Durlach, V.; Angles-Cano, E. Arteriosclerosis Thromb. Vasc. Biol. 2002, 22, 1232-1238. (14) Edelstein, C.; Mandala, M.; Pfaffinger, D.; Scanu, A. M. Biochemistry 1995, 34, 16483-16492. (15) Fless, G. M.; Rolih, C. A.; Scanu, A. M. J. Biol. Chem. 1984, 259, 1147011478. 10.1021/ac050962u CCC: $33.50

© 2006 American Chemical Society Published on Web 12/14/2005

purification is tedious and time-consuming. A more efficient way for Lp(a) isolation from serum is much needed. Apo(a) is highly glycosylated. Carbohydrates are responsible for 25-40% of apo(a) mass.10,16 Sialic acid content in apo(a) is six times more than in an LDL particle.17 Apo(a) is very hydrophilic due to its high degree of O-glycosylation.18 Following studies of deglycosylated apo(a), Utermann et al. concluded that there is not relation between apo(a) glycosylation and its heterogeneity.11 Taking advantage of the high degree of glycosylation of Lp(a) and the lectin wheat germ agglutinin (WGA) specificity for sialic acid, Seman et al. 16 achieved a rapid separation of Lp(a) from other plasma lipoproteins. Lp(a) cholesterol content was measured on the fraction separated by WGA. Lp(a) recovery after extraction from serum by WGA was reported to be 64%.19 The objective of the study reported here was to develop a rapid method for the separation, purification, and density measurement of Lp(a) from serum using a procedure that is isoform independent. This objective was met by taking advantage of Lp(a)’s glycosylation and using a novel density gradient ultracentrifugation method for lipoprotein separation.20,21 This method can also be used as part of a separation procedure for further Lp(a) characterization. EXPERIMENTAL SECTION Chemicals and Materials. N-Acetyl-D-glucosamine, phosphatebuffered saline (PBS) tablets, agarose coupled wheat germ agglutinin (Catalog No. L-1882), phosphorylase b, cross-linked molecular weight markers (Catalog No. 9012-69-5), and 2-mercaptoethanol were purchased from Sigma (St. Louis, MO). Ethylenediaminetetracetic acid monosodium bismuth salt (NaBiEDTA) was from TCI America (Portland, OR). The cesium bismuth EDTA (CsBiEDTA) was synthesized in our laboratory as reported previously.20 NBD C6-ceramide (NBD) was from Molecular Probes (Eugene, OR). L-Proline was purchased from Dickinson & Co. (Franklin Lakes, NJ). The 3-8% NuPAGE Trisacetate SDS-PAGE minigels, Novex Tris-acetate SDS running buffer (20×), NuPAGE SDS sample buffer (4×), nitrocellulose membranes (0.2-µm pore size), WesternBreeze Chromogenic Western Blot Detection Kit, and NuPAGE transfer buffer (20×) were purchased from Invitrogen (Carlsbad, CA). Polyclonal antiapoB antibody was purchased from Chemicon International (Catalog No. AB742, Temecula, CA). Blood Draw. Blood from normolipidemic subjects was drawn into Vacutainer-brand series collection tubes following a 12-h fast (7 mL, sterile interior, with STT gel and clot activator; Beckton Dickinson Vacutainer Systems, Franklin Lakes, NJ). Serum was separated from red blood cells by centrifugation at 3200 rpm for 10 min at 4 °C to separate it from red blood cells. The supernatant (serum) was aspirated from red blood cells, separated into 200µL aliquots, and used immediately or stored at -86 °C until used. (16) Seman, L. J.; Jenner, J. L.; McNamara, J. R.; Schaefer, E. J. Clin. Chem. 1994, 40, 400-403. (17) Ehnholm, C.; Garoff, H.; Renkonen, O.; Simons, K. Biochemistry 1972, 11, 3229-3232. (18) Deng, G.; Chow, D.; Sanyal, G. Anal. Biochem. 2001, 289, 124-129. (19) Gaw, A.; Brown, E. A.; Gourlay, C. W. D.; Bell, M. A. Br. J. Biomed. Sci. 2000, 57, 13-18. (20) Hosken, B. D.; Cockrill, S. L.; Macfarlane, R. D. Anal. Chem. 2005, 77, 200-207. (21) Johnson, J. D.; Bell, N. J.; Donahoe, E. L.; Macfarlane, R. D. Anal. Chem. 2005, 77, 7054-7061.

Lp(a) Extraction from Serum. Lp(a) was removed from serum by carbohydrate affinity according to the method developed by Seman et al. with some modifications.16 Briefly, 50 µL of serum, 100 µL of PBS-200 mM proline, and 50 µL of agarose-WGA were mixed together. The homogeneous mixture was incubated for 30 min at room temperature under continued mixing in the M-60 orbital shaker (Labnet International Inc., Edison, NJ). After incubation, the WGA-Lp(a) complex was sedimented by a slow centrifugation (5 min at 6000 rpm). A control sample (Lp(a) serum) was prepared by following the Lp(a) extraction protocol (above) in the absence of WGA. Sample Preparation for Ultracentrifugation. After Lp(a) was extracted from serum, the remaining serum (Lp(a)-depleted serum) was separated from the WGA-Lp(a) complex for further analysis. Lp(a) and Lp(a)-depleted serum samples were ultrafiltrated to achieve similar sample composition. Briefly, the samples were transferred into the top chamber of a 10 000 molecular weight cutoff filter (Microcon YM-10, Millipore, Bedford, MA) and ultrafiltrated two times for 8 min at 10 000 rpm, replacing the volume displaced with density gradient solution (NaBiEDTA or CsBiEDTA). Samples in the top chamber (Lp(a) and Lp(a)depleted serum) were recovered by inverting the filter into a new reservoir and centrifuging for 10 min at 10 000 rpm. Lp(a) Recovery from WGA-Lp(a) Complex. Lp(a) was recovered from the WGA-Lp(a) complex by incubation of the complex in PBS-200 mM N-acetyl-D-glucosamine following the protocol established by Seman et al. with some modifications.16 Briefly, 300 µL of PBS-200 mM proline was added to the tube containing the sedimented WGA-Lp(a) complex. The sample was well mixed and transferred to the top chamber of a 45-µm filter (Ultrafree-MC 0.45 µm centrifugal filter units, Millipore, Bedford, MA). The sample was rinsed twice for 90 s at 6000 rpm with PBS200 mM proline to drain any unbound lipoproteins from the complex. Then 300 µL of PBS-200 mM N-acetyl-D-glucosamine were added to the top chamber, and filter contents were briefly mixed. The homogeneous samples were incubated for 30 min at room temperature and continued mixing in the M-60 orbital shaker (Labnet International Inc., Edison, NJ). After incubation, the sample was spun for 90 s at 6000 rpm, and the filtered solution containing recovered Lp(a) in PBS-200 mM N-acetyl-D-glucosamine collected. The buffer was removed and replaced by density gradient solution (NaBiEDTA or CsBiEDTA) by ultrafiltration. Lipoprotein Ultracentrifugation. Lipoproteins were separated according to their embedded density as described previously.20,21 Lp(a) and Lp(a)-depleted serum samples were individually mixed with 10 µL of 2 mg/mL NBD and enough BiEDTA density gradient solution to obtain a final sample volume of 1200 µL. The samples were allowed to stand for 30 min at 5 °C in order to equilibrate NBD C6-ceramide uptake by lipoproteins. A 1000µL volume of each sample was transferred to an ultracentrifuge tube (1.5-mL, thick-walled, polycarbonate, Beckman-Coulter, Palo Alto, CA). The tubes were positioned in a TLA120.2 rotor (Beckman-Coulter, Palo Alto, CA). A BiEDTA density gradient was formed by ultracentrifugation for 6 h.20,21 Recovered Lp(a) Ultracentrifugation. Recovered Lp(a) (from the WGA-Lp(a) complex) in NaBiEDTA or CsBiEDTA was mixed with 10 µL of 2 mg/mL NBD C6-ceramide and enough Analytical Chemistry, Vol. 78, No. 2, January 15, 2006

439

Figure 1. Lipoprotein density profile in 10% NaBiEDTA density gradient (‚‚‚). Panel A shows the lipoprotein density profile for a serum sample with elevated Lp(a) (111 mg/dL). Panel B demonstrates the lipoprotein density profile resulting from the serum sample in panel A after removal of Lp(a) by WGA. Panel C shows an overlay of lipoprotein density profiles in panels A and B. Panel D is the Lp(a) DDLP from where Lp(a) density is obtained. It shows the differential profile between the profiles in panels A and B. The differential Lp(a) density for this sample is 1.086 g/mL.

density gradient medium to obtain a total sample volume of 1200 µL. Ultracentrifugation was carried out following the procedures described above. Digital Imaging and Analysis. After centrifugation, the tubes were layered with 150 µL of deionized water. Tubes were imaged and analyzed as reported previously.20 Lipoprotein density profiles obtained for Lp(a) and Lp(a)-depleted serum samples in duplicate were averaged and a mean Lp(a) serum profile and a mean Lp(a)-depleted profile obtained. Differential Lipoprotein Density Profiling. Lp(a)’s mean density was obtained after differentially comparing the Lp(a) serum profile and the Lp(a)-depleted serum profile. Briefly, the mean Lp(a)-depleted serum profile was subtracted from the mean Lp(a) serum profile and the difference graphed as a differential density Lp(a) profile (DDLP). Lp(a) Sample Storage after Ultracentrifugation. After ultracentrifugation and imaging, the tubes containing Lp(a) were slowly frozen in liquid nitrogen following the protocol described elsewhere.22 Lp(a) fractions were collected by cutting the tube at (22) Liu, M.-Y.; McNeal, C. J.; Macfarlane, R. D. Electrophoresis 2004, 25, 29852995.

440

Analytical Chemistry, Vol. 78, No. 2, January 15, 2006

the appropriate positions with a 16-in. scroll saw (model 1672; Dremel, Racine, WI). The cut positions were determined from the profile obtained from the digital imaging and analysis (above). Electrophoresis and Immunoblotting. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed on a 3-8% Tris-acetate gradient slab minigel (Invitrogen, Carlsbad, CA). Samples were prepared for electrophoresis by mixing 15 µL of sample with 3 µL of sample buffer and 2 µL of 10% 2-mercaptoethanol. Samples were reduced by incubation for 7 min in boiling water. After electrophoresis, the gel was stained with BioSafe Coomassie Blue (BioRad, Hercules, CA). When necessary, samples were electroblotted onto nitrocellulose and apo(a) or apoB revealed by immunoblotting using Invitrogen’s WesternBreeze Chromogenic Western Blot Immunodetection Kit and goat antiapo(a) (DiaSorin, Stillwater, MN) or goat polyclonal anti-apoB (Chemicon Int., Temecula CA) as primary antibody. Lp(a) Concentration. Lp(a) concentration in serum and Lp(a) concentration in recovered samples were determined using DiaSorin’s Lp(a) SPQ kit (DiaSorin) adapted to analysis in a microplate reader. This assay for Lp(a) is designed for the quantitative determination of Lp(a) levels in human serum or

plasma by immunoprecipitin analysis. Briefly, 5 µL of serum sample or recovered Lp(a) was placed in a well of a microplate and mixed with 300 µL of DiaSorin’s Lp(a) diluent solution. Background absorbance of the diluted samples was read at 340 nm (absorbance A). A 50-µL volume of DiaSorin’s Lp(a) antibody reagent, diluted 2:3 in phosphate buffer, was added to each plate well containing diluted sample and mixed thoroughly. Samples were incubated for 10 min at 37 °C and individual absorbances at 340 nm read (absorbance B). A calibration curve was obtained following the same procedure but using DiaSorin’s Lp(a) calibration set in lieu of serum and subtracting the background (absorbance A) from the immunoprecipitated samples (absorbance B). Lp(a) concentration was calculated according to the linear regression obtained from the Lp(a) calibration curve. RESULTS AND DISCUSSION Serum Lipoprotein Profile. (A) Lp(a) Serum. Figure 1A is an example of the mean lipoproteins density profile obtained for the Lp(a) serum (control sample). The serum used in this figures contained elevated Lp(a), (111 mg/dL). The profile presents characteristic lipoprotein density distribution for lipoproteins obtained by ultracentrifugation using 10% NaBiEDTA as density gradient. Each of the peaks in Figure 1A corresponds to one lipoprotein class: very-low-density lipoprotein (VLDL), LDL, and high-density lipoprotein (HDL). The peak in Figure 1A that is between the LDL-HDL density region was identified as Lp(a). The artifact in the HDL peak is a result of an optical effect due to the tube shape and should not be mistaken as an HDL's inherent characteristic. The Lp(a) serum profile was run in duplicate and averaged to achieve better statistics and ease the comparison between Lp(a) and Lp(a)-depleted serum profiles. Using this protocol to study several Lp(a)-containing samples, we have observed that the Lp(a) mean density of the samples analyzed varies, sometimes being closer to the LDL density and sometimes being closer to the HDL density. This wide range of Lp(a) densities is attributable in part to the existence of different Lp(a) isoforms within samples.15 Lp(a) density heterogeneity can be also attributable to metabolic changes such as triglyceride levels in serum and LDL levels.24 This metabolic Lp(a) density heterogeneity is best observed when the same subject is studied over a long period of time. Lp(a) density results presented herein are not significantly affected by LDL serum levels or variations in triglyceride levels. (B) Lp(a)-Depleted. Lp(a) was removed from a second aliquot of the same serum sample from where the Lp(a) serum profile was obtained (above). Lp(a) was removed from serum by carbohydrate affinity using WGA, following the procedures described in the Experimental Section. Figure 1B shows the mean lipoprotein density profile for the Lp(a)-depleted serum sample. Compared to what is observed in Figure 1A (Lp(a) serum), in Figure 1B, the peak assigned to Lp(a) is missing from the lipoprotein profile. All other lipoproteins in serum remained intact with the same density profile after the incubation with WGA, leaving the lipoprotein profile essentially unaffected. Only the Lp(23) Henriquez, R. R.; Johnson, J. D.; Farwig, Z. N.; Macfarlane, R. D. Anal. Chem, Manuscript in preparation. (24) Nakajima, K.; Hinman, J.; Pfaffinger, D.; Edelstein, C.; Scanu, A. M. Arteriosclerosis Thromb. Vasc. Biol. 2001, 21, 1238-1243.

Figure 2. Lp(a) density profile in 10% NaBiEDTA density gradient (‚‚‚). Lp(a) recovered from serum after being extracted by WGA. Panel A: Inherent density of Lp(a) recovered from the WGA-Lp(a) complex (bottom profile in panel A) remains unaltered when compared to Lp(a)’s density in the serum lipoprotein density profile (top profile in panel A). Panel B: Lower profile is the Lp(a) profile obtained with an 18.5ms exposure time. A more detailed Lp(a) profile was obtained after increasing the exposure time to 100 ms (top profile in panel B). The mean density for this recovered Lp(a) is 1.086 g/mL.

(a) density region was affected where the Lp(a) peak is missing from the Lp(a)-depleted lipoprotein density profile. (C) Differential Density Lipoprotein Profile, DDLP. Figure 1C shows an overlay of the lipoprotein profiles from Figure 1A and B. In this figure, it is clear that only the Lp(a) peak was removed when the sample was treated with WGA (bottom profile in Figure 1C). The density gradient formed is very sensitive to mixture composition. During sample preparation, the final mixture may contain other components such as PBS-200 mM proline or NaCl from WGA. The presence of these components in the NaBiEDTA solution slightly changes the density gradient formed. This effect was minimized by keeping the composition of both samples’ (Lp(a) and Lp(a)-depleted) serum almost identical and replacing the PBS buffer in the sample with NaBiEDTA by ultrafiltration. A differential density lipoprotein profile was obtained by subtracting the mean Lp(a)-depleted serum profile (Figure 1B) from the mean Lp(a) serum profile (Figure 1A). These profiles were obtained after staining the same amount of serum (Lp(a) and Lp(a)-depleted) with the same amount of NBD. In both cases, the stain is at saturation level to guarantee that VLDL, LDL, and Analytical Chemistry, Vol. 78, No. 2, January 15, 2006

441

Figure 3. Identification of Lp(a) recovered from serum by WGA after ultracentrifugation. Panel A is the Lp(a) density profile in a 10% NaBiEDTA density gradient (‚‚‚) showing the places from where protein fractions were collected for analysis. Panel B is the reduced Western blot against apoB and apo(a) for those fractions. Apo(a) was only detected in the fraction corresponding to the main Lp(a) peak (fraction 2).

HDL take up the same amount of NBD independently of the presence or absence of Lp(a). This makes it possible for an accurate comparison between density profiles.23 The DDLP is shown in Figure 1D. This profile corresponds to the density profile of Lp(a). The major peak observed in this profile represents the density region from where Lp(a) was removed. Other artifacts in the profile in Figure 1D resulted from the slight shift in the density gradient between samples and should not be mistaken as Lp(a) features. With the DDLP (Figure 1D), and knowing the density gradient formed by the BiEDTA solute, the mean density of the Lp(a) removed can be calculated, being 1.086 g/mL for this particular sample. Analysis of the area under the curve of the Lp(a) peak in the Lp(a) serum profile (Figure 1A) and the area under the curve of the differential Lp(a) peak (Figure 1D) results in an overall Lp(a) extraction efficiency of 90%. These high Lp(a) extraction findings were confirmed using DiaSorin’s Lp(a) kit. The initial Lp(a) concentration of serum (111 mg/dL) was compared to the Lp(a) concentration in the recovered Lp(a) sample (from the WGA-Lp(a) complex, 90.8 mg/dL) and undetectable Lp(a) in the Lp(a)depleted serum. This resulted in an Lp(a) recovery efficiency of 82%. Recovered Lp(a). Lp(a) extracted from serum was recovered from the WGA-Lp(a) complex formed by incubation of the sample in PBS-200 mM N-acetyl-D-glucosamine. The high concentration of N-acetyl-D-glucosamine in this buffer interacts with WGA, releasing the intact Lp(a) back into solution. Ultrafiltration was used to exchange the PBS buffer with the BiEDTA density gradient solution used for serum lipoproteins separation. As mentioned earlier, removing the PBS buffer and having the Lp(a) sample in a density gradient solution, the mean density obtained for the sample is not affected by the introduction of a third component to the density gradient. The density gradient formed is then similar to the density gradient used for the serum lipoprotein separation (Figure 2A, top profile). 442

Analytical Chemistry, Vol. 78, No. 2, January 15, 2006

The density of recovered Lp(a) is correlated with Lp(a)’s inherent density when both density gradients are identical. Figure 2A demonstrates the feasibility of this comparison. An aliquot of the sample in the top profile was treated with WGA following the procedures explained in the Experimental Section. Lp(a) was recovered from the WGA-Lp(a) complex, mixed with 10% NaBiEDTA, and stained with NBD, and the density gradient was generated by ultracentrifugation. The lipoprotein density profile for this sample is represented by the bottom profile in Figure 2A. A single peak at the reported density range for Lp(a) was obtained.11 More importantly, both Lp(a) peaks (lipoprotein density profile on the top and recovered Lp(a) density profile on the bottom, Figure 2A) are at the same density. This shows how Lp(a) is recovered intact from serum after treatment with WGA. Lp(a) isolated and recovered by the method presented herein retained its inherent density and properties. This result confirms the fact that Lp(a) is effectively extracted from serum by WGA and that its density is accurately calculated from the Lp(a) density profile. The Lp(a) mean density for this particular sample is 1.086 g/mL. This density is the same as the density obtained from the DDLP. When the exposure time of the imaging camera was increased from the normal 18.5 (bottom profile in Figure 2B) to 100 ms (top profile in Figure 2B), the intensity of the peak was increased and more details in the profile are detected. For example, there is a shoulder present in the Lp(a) peak under what corresponds to the LDL density region. This lipoprotein moiety was not evident when the whole serum lipoproteins were profiled (top profile, Figure 2A), but after obtaining a more detailed picture of the density profile of the recovered Lp(a) (top profile, Figure 2B), these details are now evident. Another detail that was visible when the exposure time of the camera was increased is the presence of other proteins at high density. These proteins presumably represent glycosylated serum proteins that were extracted by WGA along with Lp(a).

Figure 4. Coomassie Blue-stained SDS-PAGE gel demonstrating the advantage of using ultracentrifugation for the purification of Lp(a) extracted from serum by WGA. Lane 1 contains molecular weight markers (phosphorylase b, cross-linked). Lane 2 represents all proteins extracted from serum by WGA, including Lp(a). Lane 3 presents all proteins present in the Lp(a) fraction collected after extraction from serum by WGA and separation from other serum proteins by ultracentrifugation.

The presence of Lp(a) in the main peak obtained was confirmed by Western blot. Fractions for the Lp(a) peak shoulder (fraction 1), Lp(a) peak (fraction 2), and serum proteins (fraction 3) were collected by the freeze-cut technique.22 Figure 3A shows the positions from where these fractions were collected. Western blot with immunodetection confirmed the identity of pure Lp(a) in fraction 2 (Figure 3B) and the absence of free apo(a) in the high-density region (fraction 3, Figure 3B). The identity of the shoulder peak under the LDL density region was also determined by Western blot. There was the possibility

that this shoulder may contain a different Lp(a) isoform. Fractions 1 and 2 (from Figure 3A) were immunoblotted against anti-Lp(a) and anti-apoB (Figure 3B). The negative result of fraction 2 against anti-Lp(a) demonstrates that this lipoprotein removed by WGA from serum is not another Lp(a) isoform in this particular sample. Proteins in the high-density region (fraction 3) were not assayed against anti-apoB. Ultracentrifugation of the recovered Lp(a) sample serves a dual purpose. First, it is an efficient way to determine the Lp(a) density and could eventually elucidate the presence of Lp(a) isoforms masked under other lipoprotein peaks in the serum lipoprotein profiles. Second, Lp(a) is further purified by ultracentrifugation. This result is demonstrated in the Coomassie Blue-stained gel in Figure 4. High molecular weight markers (phosphorylase b crosslinked) were used to determine the approximated molecular weight of the proteins in the SDS-PAGE gel (lane 1). Lane 2 in Figure 4 contains all proteins extracted from serum by WGA. Lane 3 consists of Lp(a) extracted from serum by WGA and separated from other serum proteins by ultracentrifugation. The Lp(a) fraction purified by this ultracentrifugation method recovered by the freeze-cut technique,22 resulted in a single band with an approximate molecular weight of 330 000. The results reported here are further substantiated in Figure 5. The Lp(a) and Lp(a)-depleted serum lipoprotein profile presented in this figure belong to serum samples from normolipidemic subjects with no detectable Lp(a) (measured with DiaSorin’s Lp(a) kit). The VLDL, LDL, and HDL peaks in Figure 5A remained unchanged after the serum was incubated in WGA. This result confirms the high specificity of WGA for Lp(a). A more complex case is presented in Figure 5B. This sample contains a feature (shoulder) in the HDL region that makes it interesting for study. It was known from the immunoassay that this sample did not contain Lp(a). However, when the serum was incubated with WGA, there was a difference detected in the lipoprotein density profile in the HDL region. Most likely, in the absence of Lp(a), WGA complexes with other glycosylated lipoproteins. In this case, the feature removed from the HDL peak suggests that a glycosylated HDL subclass might be present in this sample. These results are consistent with the suggestion by

Figure 5. 10% NaBIEDTA lipoprotein DDLP profiles (‚‚‚) for non-Lp(a) serum samples showing the specificity of WGA for Lp(a). Panel A shows a normal lipoprotein density profile where lipoproteins were unaffected by WGA. Panel B is the lipoprotein density profile suggesting that a glycosylated HDL subclass was removed by WGA in the absence of Lp(a).

Analytical Chemistry, Vol. 78, No. 2, January 15, 2006

443

Seman et al. concerning a competitive binding of WGA to Lp(a) in the presence of other lipoproteins.16 Lp(a) Characterization. The purity and apparent molecular weight of the recovered Lp(a) fraction obtained after ultracentrifugation (sample in Figure 2B) was determined by SDS-PAGE followed by Western blotting (Figures 3 and 4). The Lp(a) fraction was reduced in an SDS-containing buffer and electrophoresed. After staining the gel with Coomassie Blue, a single band at an approximated molecular weight of 330 000 was obtained (Figure 4, lane 3). Further characterization analysis such as SDS-PAGE molecular weight phenotyping, electrophoretic mobility by capillary electrophoresis, and mass spectrometry25 can be performed using this pure Lp(a) sample. CONCLUSIONS This study demonstrates the analytical power of linking ultracentrifugation with affinity separations. The primary objectives of this study were to develop an analytical technique for the rapid separation, purification, and measurement of the density of Lp(a) from serum. Lp(a) was effectively removed from serum following a simple protocol using WGA that demonstrated high affinity for the carbohydrates in Lp(a). The lipoprotein density profile of a serum sample containing high levels of Lp(a) was compared with the lipoprotein density profile of the same serum sample from where Lp(a) was removed. The difference between these profiles was reported as a DDLP. Lp(a)’s mean density was determined (25) Huby, T.; Schroeder, W.; Coucet, C.; Chapman, J.; Thillet, J. Biochemistry 1995, 34, 7385-7393.

444

Analytical Chemistry, Vol. 78, No. 2, January 15, 2006

from the DDLP. We concluded that this DDLP methodology gives relevant information such as Lp(a) density and isoform characteristics that are important for the assessment of Lp(a) as a marker for cardiovascular disease. Lp(a) was quantitatively removed from serum and recovered with more than 80% recovery efficiency. Lp(a) recovered from serum by the method described herein retained its inherent density (1.086 g/mL) and immunoreactivity. The pure Lp(a) obtained can be used for further Lp(a) characterization and isoform phenotyping with analytical techniques such as SDSPAGE, capillary electrophoresis, and mass spectrometry after limited proteolysis by thermolysin.25 The procedures described herein are relevant in a clinical setting for the detection of Lp(a) isoforms based on density. ACKNOWLEDGMENT We especially thank Erika L. Brooks for her invaluable contribution for this research. The technical assistance of Sarah A. Tilford for the preparation of the figures for this article is greatly appreciated. We also thank Richa Chandra for her helpful comments and support. This work was supported by the NIH, Heart, Lung and Blood Institute (HL 068794).

Received for review June 1, 2005. Accepted October 19, 2005. AC050962U