Analytical performance of a sandwich enzyme immunoassay for pre 1-HDL in stabilized plasma

We have established an immunoassay for pre 1HDL (the initial acceptor of cellular cholesterol) using a monoclonal antibody, MAb55201. Because pre 1-HDL is unstable during storage, fresh plasma must be used for pre 1HDL measurements. In this study, we describe a method of stabilizing pre 1-HDL, and evaluate the analytical performance of the immunoassay for pre 1-HDL. Fresh plasma was stored under various conditions with or without a pretreatment consisting of a 21-fold dilution into 50% (v/v) sucrose. Pre 1-HDL concentration was measured by immunoassay. In nonpretreated samples, pre 1-HDL decreased significantly from the baseline after 6 h at room temperature. Although pre 1-HDL was more stable at 0 C than at room temperature, it increased from 30.2 8.5 (SE) to 56.5 5.5 mg/l apolipoprotein A-I (apoA-I) ( P 0.001) in hyperlipidemics, and from 18.4 1.2 to 37.9 3.3 mg/l apoA-I ( P 0.001) in normolipidemics after 5-day storage. After 30-day storage at 80 C, pre 1-HDL increased from 29.0 4.0 to 38.0 5.7 mg/l apoA-I ( P 0.001) in hyperlipidemics, whereas it did not change in normolipidemics. In pretreated samples, pre 1-HDL concentration did not change significantly under any of the above conditions. Moreover, pre 1-HDL concentrations determined by immunoassay correlated with those determined by native two-dimensional gel electrophoresis (n 24, r 0.833, P 0.05). An immunoassay using MAb55201 with pretreated plasma is useful for clinical measurement of pre 1-HDL. —Miida, T., O. Miyazaki, Y. Nakamura, S. Hirayama, O. Hanyu, I. Fukamachi, M. Okada. Analytical performance of a sandwich enzyme immunoassay for pre 1HDL in stabilized plasma. J. Lipid Res. 2003. 44: 645–650. Supplementary key words hyperlipidemia • lecithin:cholesterol acyltransferase • apolipoprotein A-I • two-dimensional gel electrophoresis A large number of studies have demonstrated that a high level of HDL is a negative risk factor for coronary artery disease (1–5). These observations are consistent with the hypothesis that HDL removes excess cholesterol from peripheral tissues (6). Clinically, HDL-cholesterol and apolipoprotein A-I (apoA-I) are used as markers for the quantity of HDL particles in plasma (7–9). However, the various HDL subfractions differ in their ability to remove cellular cholesterol (10–12). Therefore, researchers have sought a method to specifically measure these subfractions (13–17). One of these subfractions, pre 1-HDL, is more anti-atherogenic than other subfractions. Pre 1-HDL contains apoA-I, phospholipid, and a small amount of free cholesterol (11). In cell culture systems, pre 1-HDL captures free cholesterol from cell membranes within a few minutes (11, 18–19). This property of pre 1-HDL is similar to that of reconstituted lipid-poor HDL (20), an experimental model of nascent HDL. Although pre 1-HDL comprises only a small proportion of the total apoA-I in healthy subjects (14, 21–23), its level changes significantly in coronary artery disease (14, 24), obesity (23), and hyperlipidemia (21, 25–27), as well as in patients treated with lipid-lowering therapies (21, 27). Pre 1-HDL concentrations are currently measured in a limited number of clinical laboratories by native twodimensional (2D)-gel electrophoresis (14, 23–27). This method is not well suited for widespread use in clinical laboratories because it is expensive, time-consuming, and requires a significant amount of technical skill. Another obstacle encountered in the clinical laboratory is that pre 1-HDL is unstable during storage (14, 28), so that measurements should be made only on fresh plasma samples. We have recently established a sandwich enzyme immunoassay using a monoclonal antibody (MAb55201) Abbreviations: 2D-gel, two-dimensional gel; DTNB, 5,5 -dithiobis(2-nitrobenzoic acid). 1 To whom correspondence should be addressed. e-mail: miida@med.niigata-u.ac.jp Manuscript received 5 July 2002 and in revised form 29 November 2002. Published, JLR Papers in Press, December 16, 2002. DOI 10.1194/jlr.D200025-JLR200 methods by gest, on O cber 0, 2017 w w w .j.org D ow nladed fom

Pre ␤ 1-HDL contains apoA-I, phospholipid, and a small amount of free cholesterol (11).In cell culture systems, pre ␤ 1-HDL captures free cholesterol from cell membranes within a few minutes (11,(18)(19).This property of pre ␤ 1-HDL is similar to that of reconstituted lipid-poor HDL (20), an experimental model of nascent HDL.Although pre ␤ 1-HDL comprises only a small proportion of the total apoA-I in healthy subjects (14,(21)(22)(23), its level changes significantly in coronary artery disease (14,24), obesity (23), and hyperlipidemia (21,(25)(26)(27), as well as in patients treated with lipid-lowering therapies (21,27).Pre ␤ 1-HDL concentrations are currently measured in a limited number of clinical laboratories by native twodimensional (2D)-gel electrophoresis (14,(23)(24)(25)(26)(27).This method is not well suited for widespread use in clinical laboratories because it is expensive, time-consuming, and requires a significant amount of technical skill.Another obstacle encountered in the clinical laboratory is that pre ␤ 1-HDL is unstable during storage (14,28), so that measurements should be made only on fresh plasma samples.We have recently established a sandwich enzyme immunoassay using a monoclonal antibody (MAb55201) that is specific for pre ␤ 1-HDL (15).At present, this immunoassay is open to researchers.A Japanese commercial laboratory (SRL, Tachikawa, Japan) has been measuring pre ␤ 1-HDL concentrations in clinical or experimental samples using this assay.In this study, we describe a method for stabilization of pre ␤ 1-HDL by pretreatment, and we evaluate the analytical performance of the immunoassay for pre ␤ 1-HDL measurement.

Study protocol
Blood samples were obtained from 20 volunteers (12 men and 8 women, aged 26 to 68 years), or 24 outpatients (13 men and 11 women, aged 22 to 76) at our institutions.The former group consisted of 10 normolipidemics and 10 hyperlipidemics (four hypercholesterolemics, three hypertriglyceridemics, and three combined hyperlipidemics).The latter group consisted of 11 normolipidemics, and 13 hyperlipidemics (six hypercholesterolemics, three hypertriglyceridemics, and four combined hyperlipidemics).All subjects willingly gave informed consent before entering into this study.The study protocol was approved by the ethical committee of our institution.Serum cholesterol concentrations were 3.83-7.03mmol/l in the volunteers and 3.34-7.27mmol/l in the outpatients; triglyceride concentrations were 0.45-4.10mmol/l in the volunteers and 0.55-5.34mmol/l in the outpatients.
Venous blood was drawn from subjects who had fasted for at least 12 h.The blood was immediately mixed with an anticoagulant (EDTA-K 2 , 1 g/l) in a glass tube, and chilled on ice water.Plasma was separated at 0 Њ C, placed into screw-capped tubes, and stored under different conditions before assaying for pre ␤ 1-HDL.In some experiments, LCAT activity was inhibited by the addition of 5,5 Ј -dithio-bis(2-nitrobenzoic acid) (DTNB) to a final concentration of 2 mmol/l.In some experiments, whole blood was stored under the indicated condition, and plasma was separated later.Some fresh plasma samples were pretreated by 1:20 (v/v) dilution with 50% sucrose solution (stabilizing buffer), mixed thoroughly, and stored under the same conditions as the nonpretreated samples.
The pre ␤ 1-HDL levels of all subjects were measured by the sandwich enzyme immunoassay technique described below (15).For some outpatients, pre ␤ 1-HDL levels were also determined by native 2D-gel electrophoresis.The pre ␤ 1-HDL concentrations obtained by these two methods were then compared.

Sandwich enzyme immunoassay
First, a 96-well plastic plate was coated with a monoclonal antibody (MAb 55201) specific for pre ␤ 1-HDL (Daiichi Pure Chemicals, Tokyo, Japan, miyazaki-o@daiichichem.co.jp) (15).Then, plasma was diluted 2,121-fold with 1% BSA-PBS.Diluted plasma (sample) or purified human apoA-I (standard) was then added to each well and incubated at room temperature for 1 h.The wells were washed three times with 0.1% BSA-PBS, and the adsorbed pre ␤ 1-HDL was incubated with a horseradish peroxidasecoupled secondary antibody (goat anti-human apoA-I polyclonal antibody).The wells were washed again with 0.1% BSA-PBS, and o -phenylenediamine and H 2 O 2 in citrate buffer were added to the wells.The amount of pre ␤ 1-HDL in the wells was determined by measurement of the absorbance at 492 nm.The absorbance curves in the diluted samples were parallel to those obtained with the purified apoA-I.The coefficient of variation was 3.1% to 5.3% for individual analytical runs, and 4.9% to 9.1% between different analytical runs.
In the preliminary experiments, we confirmed that this immunoassay measures most apoA-I of the small HDL fraction as pre ␤ 1-HDL (15).Plasma lipoproteins were separated by gel chromatography with a FPLC system (Amersham Pharmacia Biotech).Then, we determined apoA-I and pre ␤ 1-HDL concentrations in each fraction by two immunoassays.Either polyclonal anti-apoA-I antibody or MAb55201 was used as a capture antibody.The high pre ␤ 1-HDL peak was detected in the small apoA-I-containing lipoprotein with molecular mass less than 67 kDa (15).In this fraction, pre ␤ 1-HDL/apoA-I ratio was 80-90%.

Native 2D-gel electrophoresis
Fresh plasma was separated in the first dimension by electrophoresis on 0.75% agarose gels, and then in the second dimension by electrophoresis on gradient (2% to 15%) polyacrylamide gels (14,21,(25)(26)(27)(28).The separated HDL subfractions in the polyacrylamide gels were then electroblotted onto nitrocellulose membranes.The membranes were probed with a 125 I-labeled goat anti-human apoA-I polyclonal antibody and autoradiographed.Using the autoradiogram as a guide, we clipped the HDL subfractions out of the membranes and determined the amount of radioactivity in the subfractions by ␥ spectrometry.The relative concentration of each subfraction was expressed as a percentage of the total apoA-I concentration.The absolute concentration was calculated by multiplying the relative concentration of each HDL subfraction and plasma apoA-I level that was measured by turbidity immunoassay (ApoAuto AI, Daiichi Pure Chemicals).The reproducibility of the native 2D-gel technique has been described in our previous publications (14,21,(25)(26)(27)(28).

Statistical analyses
We used StatView (version 5.0J) for statistical analyses.All values were expressed as mean Ϯ SE.Changes in an individual subject were analyzed using Student's paired t -test.The correlation between the two methods was evaluated using Pearson's correlation coefficient.We considered the difference between two groups to be statistically significant if the P value was less than 0.05.

Effects of storage temperature and duration on pre ␤ 1-HDL concentration
Storage temperature affected the measured pre ␤ 1-HDL concentration in the normolipidemic and hyperlipidemic subjects.At room temperature (24 Њ C), the measured pre ␤ 1-HDL concentration decreased as a function of time ( Fig. 1 , open and closed squares), whereas at 4 Њ C, the measured pre ␤ 1-HDL concentration increased slowly with time (Fig. 1, open and closed circles).When plasma was stored on ice water, however, no change in apparent pre ␤ 1-HDL concentration was observed for at least 4 h (Fig. 1, open and closed triangles).
Neither an LCAT inhibitor nor blood cells were observed to stabilize the concentration of pre ␤ 1-HDL at room temperature.When DTNB was added to plasma, the pre ␤ 1-HDL concentration did not decrease, but rather increased significantly during storage ( Fig. 2 , open and closed circles).When samples were stored as whole blood (Fig. 2, open and closed squares), a decrease in measured pre ␤ 1-HDL concentrations occurred at nearly the same rate as for stored plasma samples (Fig. 2, open and closed triangles).
A sucrose-stabilizing buffer completely inhibited changes in measured pre ␤ 1-HDL concentration during storage in both normolipidemic and hyperlipidemic subjects.As described above, samples stored at 4 Њ C without pretreatment exhibited increases in pre ␤ 1-HDL, so that the mean pre ␤ 1-HDL concentration at day 5 was almost double the baseline level ( Fig. 3 , open and closed circles).Although pre ␤ 1-HDL levels appeared stable for 4 h when samples were stored on ice water, pre ␤ 1-HDL levels did increase over longer storage periods (Fig. 3, open and closed triangles).In contrast, samples stored at 4 Њ C after pretreatment exhibited no change in pre ␤ 1-HDL concentration for 5 days (Fig. 3, open and closed squares).

Effect of freezing on measured pre ␤1-HDL concentration
When nonpretreated samples were stored at Ϫ80ЊC for 30 days, the mean pre␤1-HDL concentration increased significantly in the hyperlipidemic subjects (Fig. 4, right open and hatched bars).However, pre␤1-HDL did not increase significantly in the normolipidemic subjects (Fig. 4, left open and hatched bars).In contrast, no samples that were pretreated prior to freezing exhibited a significant change from baseline levels (Fig. 4, closed bars).We also tested samples that were stored at Ϫ20ЊC for 30 days.In these samples, the measured pre␤1-HDL concentrations of fresh and frozen samples did not differ significantly (n ϭ 6; 94.8 Ϯ 7.0% of the baseline level).

Comparison of sandwich enzyme immunoassay and native 2D-gel electrophoresis techniques
In this comparison, the outpatient samples were used, because since they exhibited a broader range of pre␤1-HDL concentrations than did those from the healthy volunteers, and were thus better suited to determining the applicability of these techniques to a clinical situation.Pre␤1-HDL concentrations determined by immunoassay using pretreated samples had strong positive correlations with those determined by native 2D-gel electrophoresis using fresh plasma (Fig. 5).However, values determined by immunoassay were lower than those determined by native 2D-gel electrophoresis.

DISCUSSION
These results clearly indicate that our immunoassay technique using MAb55201 with pretreated samples is a precise and reproducible method for pre␤1-HDL mea-Fig.1. Changes in pre␤1-HDL concentration during storage at different temperatures.Plasma was obtained from 10 normolipidemic (open symbols) and 10 hyperlipidemic volunteers (closed symbols: hypercholesterolemia, n ϭ 4; hypertriglyceridemia, n ϭ 3; combined hyperlipidemia; n ϭ 3).Plasma was stored at the indicated temperatures for 6 h.Samples were taken at intervals, and pre␤1-HDL concentrations were immediately determined at each sampling point by immunoassay as described in Materials and Methods.
Accurate measurement of pre␤1-HDL levels in stored samples requires awareness of the possibility of intercon-version of pre␤1-HDL and ␣-HDL during storage.The conversion of pre␤1-HDL to ␣-HDL is promoted by LCAT (15,28), whereas the reverse reaction is promoted by many factors, including CETP (21,29), PLTP (30)(31), hepatic lipase (27,32), and serum amyloid A protein (33)(34).During a 90 min incubation at 37ЊC, about 80% of pre␤1-HDL is converted into ␣-HDL in normolipidemic subjects (28).This phenomenon can be blocked by inhibition of LCAT or by coincubation with certain types of cells, such as fibroblasts and macrophages (28).In our previous study, blood cells did not prevent pre␤1-HDL from converting to ␣-HDL (28); indeed, the apparent pre␤1-HDL concentrations in whole blood and plasma decreased by similar amounts when stored at room temperature (Fig. 2).
Storage of plasma at 4ЊC reduces the rate at which pre␤1-HDL is converted to ␣-HDL (Fig. 1).However, apparent pre␤1-HDL levels increased gradually for samples stored at 4ЊC for long periods (Figs. 1, 3).This finding strongly suggests that pre␤1-HDL is generated from ␣-HDL under these conditions.Jaari et al (31) transiently overexpressed human PLTP in mice by an adenovirus-mediated method, and then incubated plasma from these mice at 37ЊC with an LCAT inhibitor (iodo-acetate).They found that total pre␤-HDL in these samples increased as a function of time, and that PLTP activity was positively correlated with the ability to generate pre␤-HDL during the 37ЊC incubation.Therefore, pre␤1-HDL may be generated by PLTP activity during long storage at 4ЊC.PLTP activity may be also involved in the increase in pre␤1-HDL in the presence of DTNB (Fig. 2).
When pretreated with 50% sucrose, pre␤1-HDL is very stable during storage in both normolipidemic and hyperlipidemic subjects.In pretreated samples, pre␤1-HDL levels did not change at all for 5 days at 4ЊC (Fig. 3).Moreover, pretreated samples can be frozen and stored for up   to 30 days at either Ϫ20ЊC or Ϫ80ЊC with no apparent effect on pre␤1-HDL levels (Fig. 4).In our experiments, pre␤1-HDL levels in frozen samples from normolipidemic subjects did not change significantly over time, but pre␤1-HDL levels in frozen samples from hyperlipidemic subjects increased significantly over time.We have tested many reagents, including protamine, protamine sulfate, sodium fluoride, phenylmethylsulfonylfluoride, aprotinin, benzamidine, sodium azide, and surfactants for their ability to stabilize pre␤1-HDL.However, none of these treatments worked as well as the 50% sucrose solution (data not shown).Pretreatment with sucrose solution is an inexpensive, simple, and reliable method.Since pretreated samples are stable at Ϫ20ЊC, samples can easily be stored in a Ϫ20ЊC freezer until assayed.
The immunoassay with pretreated samples offers other advantages over the native 2D-gel electrophoresis method for measuring pre␤1-HDL.First, the immunoassay requires much less time to complete.According to our protocol (14,21,(25)(26)(27)33), native 2D-gel electrophoresis takes about 1 day (2 h and 20 h for agarose and gradient polyacrylamide gel electrophoresis, respectively), electrophoretic transfer and immunoblotting take another day, and exposure to X-ray film takes another 1 to 2 days.In contrast, the immunoassay can be completed within 3 h.Second, the immunoassay does not require special technical skills, significant space, or use of radioisotopes.Finally, many samples can be measured simultaneously by the immunoassay technique, whereas only four samples can be analyzed in one 2D-gel run.
It should be noted that pre␤1-HDL concentrations as determined by immunoassay were decreased by almost two thirds compared with those determined by native 2Dgel electrophoresis (Fig. 5).Such discrepancy may result from two possible reasons, that is, underestimation of apoA-I of pre␤1-HDL by immunoassay or overestimation of apoA-I of pre␤1-HDL by native 2D-gel electrophoresis.In our previous study, we added MAb55201 to fresh plasma, and kept it on ice for 5 min.In the analysis using native 2D-gel electrophoresis, all plasma pre␤1-HDLs reacted with MAb55201 to form pre␤1-HDL/antibody complexes (15).As described in Materials and Methods, MAb55201 measures most apoA-I of the small HDL subfraction as pre␤1-HDL.Although pre␤1-HDL may expose less epitopes of apoA-I than delipidated apoA-I, insufficient reactivity of MAb55201 is not likely to be the main cause of such a big difference between immunoassay and 2D-gel electrophoresis.
Determination of pre␤1-HDL concentrations by native 2D-gel is highly dependent on the techniques and antibodies used for the assays.In fact, in the earlier studies of normal subjects, 2D-gel determination of mean pre␤1-HDL concentrations yielded values that varied from 1.3 Ϯ 0.8% to 9.2 Ϯ 3.1% of plasma apoA-I values (11,14,21,23), demonstrating that results from 2D-gel are highly variable.Among institutions measuring pre␤1-HDL concentrations, conditions for 2D-gel electrophoresis are not really standardized.For example, the electrophoresis temperature and time for the 2D-gel range from 0 to 10ЊC, and from 1.5 to 20 h.As we have shown in the present study, plasma pre␤1-HDL concentration increases steadily, even at 0ЊC (Figs. 1, 3).These results suggest that significant amounts of pre␤1-HDL may be newly generated during native 2D-gel electrophoresis.The mechanical friction through the gradient polyacrylamide gels might cause some dissociation of pre␤1-HDL from ␣-HDL.In addition, it is conceivable that the relative concentrations of minor HDL subfractions are somewhat overestimated due to their background radioactivities.In ␥ spectrometry, we cannot avoid measuring background counts around the spots of HDL subfractions.The ratio of background to proper count must be higher in minor HDL subfractions than in major HDL subfractions.The high ratio probably results in the increase in pre␤1-HDL concentration determined by native 2D-gel electrophoresis.Thus, the MAb55201-based ELISA may be more suitable than electrophoresis for standardization of pre␤1-HDL concentrations.
In summary, an immunoassay using MAb55201 with pretreated samples is a precise and reproducible method for measuring pre␤1-HDL levels in plasma samples.This immunoassay may be used clinically to aid in detection of subjects at risk for atherosclerotic disorders.

Fig. 4 .
Fig. 4. Effect of pretreatment on pre␤1-HDL stability during storage at -80ЊC.Plasma was obtained from 10 normolipidemic (left bars), and 10 hyperlipidemic volunteers (right bars: hypercholesterolemia, n ϭ 4; hypertriglyceridemia, n ϭ 3; combined hyperlipidemia; n ϭ 3) Plasma was frozen with (closed bars) or without pretreatment (hatched bars), and stored at -80ЊC.The baseline pre␤1-HDL concentrations in fresh plasma (open bars) were measured by immunoassay on the same day that blood was obtained.The frozen samples were thawed 30 days later, and pre␤1-HDL concentrations were determined by immunoassay.