Fractionation of human serum lipoproteins and simultaneous enzymatic determination of cholesterol and triglycerides

Analytica Chimica Acta 654 (2009) 85–91

Contents lists available at ScienceDirect

Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca

Fractionation of human serum lipoproteins and simultaneous enzymatic determination of cholesterol and triglycerides Rashid Nazir Qureshi, Wim Th. Kok ∗ , Peter J. Schoenmakers Polymer-Analysis Group, van’t Hoff Institute for Molecular Sciences, University of Amsterdam, Nieuwe Achtergracht 166, 1018WV Amsterdam, The Netherlands

a r t i c l e

i n f o

Article history: Received 4 March 2009 Received in revised form 24 June 2009 Accepted 25 June 2009 Available online 1 July 2009 Keywords: Asymmetric Flow Field-Flow Fractionation Lipoproteins Enzymatic determination Cholesterol Triglycerides

a b s t r a c t A method based on Asymmetric Flow Field-Flow Fractionation (AF4) was developed to separate different types of lipoproteins from human serum. The emphasis in the method optimization was on the possibilities to characterize the largest lipoprotein fractions (LDL and VLDL), which is usually not possible with the size-exclusion chromatography methods applied in routine analysis. Different channel geometries and flow programs were tested and compared. The use of a short fractionation channel was shown to give less sample dilution at the same fractionation power compared to a conventional, long channel. Different size selectivities were obtained with an exponential decay and a linear cross flow program. The ratio of the UV absorption signal to the light scattering signal was used to validate the relation between retention time and size of the fractionated particles. An experimental setup was developed for the simultaneous determination of the cholesterol and triglycerides distribution over the lipoprotein fractions, based on enzymatic reactions followed by UV detection at 500 nm. Coiled and knitted PTFE tubing reactors were compared. An improved peak sharpness and sensitivity were observed with the knitted tubing reactor. After optimization of the experimental conditions a satisfactory linearity and precision (2–3% rsd for cholesterol and 5–6% rsd for triglycerides) were obtained. Finally, serum samples, a pooled sample from healthy volunteers and samples of sepsis patients, were analyzed with the method developed. Lipoprotein fractionation and cholesterol and triglyceride distributions could be correlated with the clinical background of the samples. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Coronary heart disease (CHD), one of the major causes of death in the world, is strongly related to the human lipoprotein metabolism [1,2]. Lipoproteins are complex agglomerates of different sizes, composed of a variety of different types of molecules in their core and on their surface. Triglycerides, lipids and cholesteryl esters form the hydrophobic core of the spherical lipoprotein particles; this core is surrounded by an amphiphilic shell of free cholesterol, phospholipids and apoproteins. Lipoproteins can be classified according to their density and size into high-density lipoprotein (HDL), lowdensity lipoprotein (LDL), and very low-density lipoprotein (VLDL) and into various subclasses of these. HDL and LDL, being considered as anti-atherogenic and atherogenic in nature, have drawn most attention of researchers. LDL consists of relatively large particles, rich in cholesterol that transport triglycerides and cholesterol from the liver to all cells and tissues. HDL on the other hand removes cholesterol from cell membranes and the walls of blood vessels and delivers it to the liver for excretion. It has been well established that

∗ Corresponding author. Tel.: +31 20 5256539. E-mail address: [email protected] (W.Th. Kok). 0003-2670/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2009.06.060

an increase in total cholesterol, LDL-cholesterol and triglycerides in human blood are positive risk factors for CHD, while an increase in HDL-cholesterol is considered as an inverse risk factor [3]. Hence, the quantitative analysis of cholesterol and triglycerides associated to the various (sub)classes of lipoproteins is extremely important in clinical laboratories, because of their predictive association not only with CHD [4] but also with sepsis [5], liver dysfunction and cancer [6]. Since decades assays using enzymatic reagents are considered to be the most reliable methods for the selective and quantitative determination of total cholesterol and triglycerides in serum samples [4]. The enzymatic assays for the two clinically important analytes can be carried out in a single cuvette sequentially [7,8] and automated assays through flow-injection analyses (FIA) have been proposed [9–14]. However, the total serum levels of cholesterol and triglycerides are nowadays regarded as indicators with a limited diagnostic value, and data on the distribution of the lipids over the lipoprotein (sub)fractions are preferred [4]. Originally, fractionation was performed by ultracentrifugation, the technique that provided the classification system for lipoproteins [15]. However, ultracentrifugation is a lengthy and laborious method. A method for the fractionation of lipoproteins based on size-exclusion chromatography (SEC) or gel permeation chromatography was developed that

86

R.N. Qureshi et al. / Analytica Chimica Acta 654 (2009) 85–91

could separate HDL, LDL and VLDL in less than 60 min [16]. The cholesterol and triglyceride distribution over the lipoprotein fractions could be determined with enzymatic methods applied on collected fractions, but on-line quantification of one of the analytes associated to the different (sub)fractions of lipoproteins after their separation by SEC is also possible with a post-column enzymatic reaction setup [17,18]. Finally, Usui et al. showed that both the cholesterol and the triglyceride content of the separated fractions can be determined on line, by splitting of the column effluent and using two post-column reaction systems simultaneously [19]. Although the method of SEC with post-separation enzymatic determination of cholesterol and triglycerides eliminated laborious and time consuming procedures, there are also some limitations. First, the selectivity of the SEC separation is limited, especially at the high-MW (VLDL) side, even when multiple columns in series are used [16]. Moreover, undesired interaction between proteins and the stationary phase may result in adsorption or entanglement, which reduces recovery and separation [20]. As an alternative separation method for lipoproteins Flow Field-Flow Fractionation (F4) has been proposed [21–25]. F4 is a separation technique that enables the separation of (biological) macromolecules without mechanical or shear stress on their native conformational structures [26]. In F4 macromolecules are separated based on their diffusion coefficients by the combined action of an axial flow stream of a carrier liquid in a thin channel and a cross flow applied in the perpendicular direction [27]. In the asymmetrical variant (AF4) one of the walls of a flat channel is permeable for the carrier liquid and the cross flow is created as part of the inlet flow [28–30]. Compared to the original symmetrical system (with two porous walls), AF4 offers the advantages of a simplified instrumental setup and diminished sample dilution [31]. In the work presented here, we studied the coupling of AF4 for the size-based fractionation of lipoproteins to the dual enzymatic system for the simultaneous determination of the triglycerides and cholesterol distribution over lipoproteins fractions. For the fractionation different cross flow programs were tested and different channel geometries were compared. The emphasis in the method optimization was on the possibilities to characterize the largest lipoprotein fractions (LDL and VLDL), which is usually not possible with the SEC methods applied in routine analysis. For optimization of the enzymatic reaction systems kinetic parameters were studied such as the reaction temperature, time, flow rates and geometries of reactor tubing. Finally, the optimized system was tested by analyzing serum samples of healthy persons and of patients of the university hospital.

buffer (100 mmol L−1 ), 4-chlorophenol (4-CP, 5 mmol L−1 ), sodium chloride (2.3 mmol L−1 ) and Triton X 100 (1.5 mmol L−1 ). For triglyceride measurements R2 contained lipase (≥1000 U L−1 ), peroxydase (≥1700 U L−1 ), glycerol-3-phosphate oxydase (≥3000 U L−1 ), glycerol kinase (GK) ≥660 U L−1 , PAP (0.5 mmol L−1 ) and sodium adenosine triphosphate (ATP, 1.3 mmol L−1 ). Buffer R1 contained PIPES (100 mmol L−1 ), magnesium chloride (9.8 mmol L−1 ) and 4CP (3.5 mmol L−1 ). The reagents were stored at 2–5 ◦ C in the dark after preparation. Standard solutions of cholesterol and triglycerides were prepared by diluting in carrier solution. A pooled serum sample of healthy donors (80 volunteers, male and female) and samples from two sepsis patients were obtained from the university hospital (AMC, Amsterdam). The samples had been collected under non-fasting conditions. Samples were stored at −20 ◦ C before analysis. 2.2. Apparatus An Agilent 1100 series degasser and a 1200 HPLC series isocratic pump were coupled with an Eclipse2 AF4 separation system (Wyatt Technology Europe GmbH, Dernbach, Germany) to carry out fractionation. Separation channels with 350-␮m spacers of trapezoidal shape of different dimensions were used, and a regenerated cellulose membrane with a molar mass cut off of 10 kDa. Samples were injected with a 6-port valve with a 20 or 100 ␮L loop. A DAWN-DSP MALS detector (Wyatt) was used in series with UV detector (Applied Biosystems, USA). ASTRA software (Wyatt Technology) version 4.9 was used to handle signals from the detectors. For the simultaneous enzymatic determination of lipoprotein–cholesterol and lipoprotein–triglycerides the main channel flow of 0.6 mL min−1 was split into two lines by a MicroSplitter P-460 (Upchurch Scientific Inc., Oak Harbor WA, USA). The enzymatic reagents were pumped in the separate lines using two HPLC pumps (Applied Biosystems, USA) with a flow rate of 0.2 mL min−1 . The reagents were placed in amber glass bottles immersed in melting ice. PTFE tubing of 0.5 mm ID (Supelco, USA) of lengths 5.1 and 10.2 m was used to make coiled and knitted enzymatic reactors. Different lengths of reactors correspond to different reaction times, i.e., 2 and 4 min. Reactors were placed in thermostated bath at 37 ◦ C. Two Spectroflow 757 UV detectors (Applied Biosystems, USA) were used at 500 nm for detection. The complete system is shown schematically in Fig. 1. 3. Results and discussion 3.1. Optimization of the cholesterol and triglycerides measurements

2. Experimental 2.1. Chemicals and solutions All reagents used were of analytical grade. HDL (25.8 mg protein per mL) and LDL (5.0 mg protein per mL) standards were purchased from Sigma Aldrich (Saint Louis, USA). Phosphate buffer saline (PBS, 138 mM sodium chloride, 2.7 mM potassium chloride, and 10 mM phosphate buffer salts at pH 7.4) was used as carrier solution. It was prepared in doubly distilled water and filtered with 0.22 ␮m GV Millipore filters before use. Cholesterol CHOD-PAP and Triglycerides GPO kits (Biolabo SA, Maizy, France) were used for the quantitative determination of cholesterol and triglycerides. The enzymatic reagents were prepared by mixing an enzymes solution (R2) with a buffer solution (R1). For the cholesterol measurements R2 contained cholesterol oxydase (≥100 U L−1 ), cholesterol esterase (≥170 U L−1 ), peroxydase (≥1200 U L−1 ), 4-amino antipyrine (PAP, 0.25 mmol L−1 ), and PEG 6000 (167 ␮mol L−1 ). The buffer solution R1 contained phosphate

Determination of cholesterol and triglycerides by the CHOD and GPO methods involves different sequences of enzymatic reactions as shown in Table 1. In the final step of both sequences, hydrogen peroxide reacts with PAP and 4-CP to form an imine derivative Table 1 Enzymatic reactions for cholesterol and triglycerides determination. Cholesterol 1. Cholesterol esters 2. Cholesterol + O2

cholesteryl esterase

−→

cholesteryl oxidase

3. H2 O2 + 4-CP + PAP

−→

peroxidase

−→

cholesterol + free fatty acids

cholestenone + H2 O2

quinoneimine (pink) + H2 O

Triglycerides lipase

1. Triglycerides −→ glycerol + free fatty acids 2. Glycerol + ATP

glycerol kinase

−→

3. Glycerol phosphate + O2 4. H2 O2 + 4-CP + PAP

glycerol phosphate + ADP

glycerol phosphate oxidase

peroxidase

−→

−→

dihydroxyacetone phosphate+ H2 O2

quinoneimine + H2 O

R.N. Qureshi et al. / Analytica Chimica Acta 654 (2009) 85–91

87

Fig. 1. Schematic diagram of the setup. CS, carrier solution; D+P, degasser and pump; I, injection port; C, AF4 channel; P1 and P2, pumps for enzymatic reagents; E-TG, enzymatic reagent for triglycerides; E-CH, enzymatic reagent for cholesterol; M, MALS detector; R1 and R2, reactors; T, thermostat; D1–D3, UV detectors; W, waste bottle.

Fig. 2. Reaction kinetics of the enzymatic reactions of triglycerides and cholesterol. UV absorbance at 500 nm recorded against time at 37 ◦ C.

with maximum absorption at 500 nm. To study the kinetics of the enzymatic reactions under the conditions employed in the on-line system, the reagent solution was appropriately diluted with carrier solution (in a 2:3 volume ratio). A standard solution of cholesterol or triglycerides was added in a thermostated cuvette (37 ◦ C), and the absorbance at 500 nm was recorded against time. Results are shown in Fig. 2. With the triglyceride standard the absorption increased sharply in the first 100 s and then became constant, while for cholesterol the reaction took about 180 s for completion. On the other hand, the final sensitivity for cholesterol was almost double compared to that for triglycerides. For the on-line reaction systems, different tubing diameters can be considered for the reactors. To facilitate mixing of the enzyme solution with the effluent from the AF4 channel, and to minimize additional band broadening in the reaction coils, using narrow tubing would be preferred. However, with the length of tubing necessary to obtain the required reaction time, backpressures can be

obtained that are not compatible with the AF4 channel and the detectors before the split. For instance, when a reaction time of 4 min is assumed to be appropriate, 40 m of tubing with an internal diameter of 0.25 mm would be required, and the backpressure would be approximately 25 bar. Therefore, PTFE tubing with an ID of 0.5 mm was used. The reaction conditions were optimized with a flow-injection (FIA) system, i.e., with bypassing of the AF4 channel. Standard solutions were injected (20 ␮L) into a carrier solution flowing at a rate of 0.6 mL min−1 , and after splitting the flow was merged with the two reagent flows of 0.2 mL min−1 each. Pieces of tubing with lengths of 5.1 and 10.2 m were used, giving hold-up times of 2 and 4 min, respectively. Initially coiled reactors were tested. For the standard triglycerides, a reaction time of 2 min was apparently not enough, since the average peak area increased when a 10.2 m long reactor was used instead of 5.1 m (see Table 2). For cholesterol peak areas were almost the same with both reactors. However, strong peak broadening was observed for cholesterol and a low repeatability of the peak area. Since the repeatability for cholesterol improved when a longer reaction coil was used, the problems encountered may be attributed to incomplete mixing. Therefore, homemade knitted reactors were tested as an alternative for the reaction coils. With these reactors the peak shape for cholesterol was clearly improved and satisfactory peak area repeatability could be obtained. Relative standard deviations for the peak area were in the order of 2–3% for cholesterol and 5–6% for triglycerides. The dual enzymatic system with knitted reactors of 10.2 m lengths was calibrated with mixed standards of triglyceride and cholesterol of 0.2, 0.4, 0.6, 0.8 and 1 g L−1 concentrations (Fig. 3). Linearity was achieved for both cholesterol and triglycerides with as linear regression equations y = 254.96x + 5.46 (r2 = 0.991) and y = 111.06x + 1.44 (r2 = 0.996) respectively. Cross-reactivity was not observed; cholesterol did not give a signal in the triglyceride reaction and vice versa. 3.2. HDL and LDL separation and molecular mass determination The AF4 system was optimized with standard mixtures of HDL and LDL (obtained from human serum). Relevant lipoproteins in

Table 2 Optimization of the reactors for the on-line enzymatic reaction systems. Reactor

Cholesterol

Type

Length (m)

Average Coiled

5.1 10.2

19.4 19.0

Knitted

5.1 10.2

19.2 24.2

Triglycerides

Peak area

Peak width

Peak area

Peak width

SD

Sigma (s)

Average

SD

Sigma (s)

4.5 1.7

24 31

7.5 10.7

0.3 0.6

17 19

0.4 0.7

19 18

8.5 9.9

0.4 0.6

14 15

88

R.N. Qureshi et al. / Analytica Chimica Acta 654 (2009) 85–91

Fig. 3. Calibration curves for cholesterol and triglycerides with FIA.

human serum differ strongly in size, from typically 5 to 12 nm for HDL particles to over 80 nm for VLDL. In certain clinical cases (e.g., sepsis) even chylomicrons with a size in the micrometer range can be found in serum. Therefore, a fractionation method for lipoproteins should cover a wide size range. A simple way to expand the size window of AF4 is by applying a cross flow program, with a decrease of the cross flow rate during the run [32]. Different cross flow programs can be used, resulting in different selectivities over the time window. In a time-delayed exponential decay (TDE) program, the cross flow is first kept constant over a certain period, and next decreased exponentially with a time constant equal to the delay time [33]. An advantage of this method is that it results in a linear relation between retention time and the logarithm of the molecular mass for a certain type of analyte [34]. We used a TDE program with a 7 min delay/decay time for the separation of HDL and LDL. In the constant cross flow mode, the mixture of standards could be separated in 30 min. The cross flow program not only reduced the analysis time but also improved the sensitivity for the late eluting LDL. Initially, the fractionations were carried out with a conventional, 25 cm long channel. A satisfactory separation of HDL and LDL was obtained with a channel flow of 1 mL min−1 and a cross flow of 2 mL min−1 . However, when the fractionation system is to be coupled to the dual enzymatic detection setup, relatively high reagent flows (of 0.4 mL min−1 ) have to be maintained to obtain the recommended enzyme concentrations in the solution mixtures after merging of the reagent flows with the split channel effluent. Therefore, with the goal to diminish the consumption of the (expensive) enzymatic reagents, a miniaturized version of the AF4 channel was tested. With this 12 cm long channel virtually the same fractionation could be obtained with a channel flow of 0.6 and a cross flow of 1.2 mL min−1 . Only a small shift of retention times was observed. The fractograms obtained with the two channels are compared in Fig. 4. An additional advantage of the short channel was a decreased sample dilution. With the same sample volume (20 ␮L) peak heights for the UV and the light scattering detectors were higher by a factor of approximately 1.65 when the 12 cm channel is compared to the 25 cm channel. From the retention times of the lipoprotein species, using standard AF4 theory and the Stokes formula, the hydrodynamic diameters of the particles can be estimated. For HDL a (peak top) diameter of 7 nm was calculated, while for LDL an average particle size of 20 nm was found. With the TDE program some high-MW lipoprotein particles were observed at 35 min as the cross flow approached zero that were not eluting with a constant cross flow. This high-MW fraction could be an impurity in the standards, or agglomerates formed during storage, sample preparation or focusing.

By comparing the UV signal at 280 nm with the 90◦ light scattering response of the MALS, an indication can be obtained of the molar masses of HDL and LDL. The UV signal can be regarded as proportional to the (protein) concentration, while the scattering signal is also directly proportional to the molecular mass of the macromolecules. Therefore, the ratio of these two signals is proportional to the molecular mass of an eluting fraction, independent of the concentration. For an absolute determination of molar masses, the response factor of the UV detector and the dn/dc value (for the light scattering signal) should be known. Since such data were not available, only the relative sizes of fractions could be determined. Fig. 5a shows the UV and light scattering signals obtained with the fractionation of the standard HDL and LDL mixture, and – on a logarithmic scale – the relative molecular masses calculated from these signals. The average molar mass for the LDL peak is 20 times higher than that for the HDL peak. This is in good agreement with literature data that give values of 150–300 kDa for HDL and 3–5 MDa for LDL [22]. Moreover, it also agrees well with the difference found for the hydrodynamic size of the particles. The size distribution of the HDL and LDL particles is also reflected in Fig. 5. The Astra software gave polydispersity values for HDL and LDL of 1.07 and 1.09, respectively. Fig. 5b shows the fractogram obtained with the pooled serum sample of healthy volunteers, obtained under the same experimental conditions. The serum was injected without any pretreatment. Clearly, most of the (UV) signal will be coming from serum proteins

Fig. 4. Effect of channel length on the separation of HDL and LDL standards using a TDE cross flow program. For experimental details see text.

R.N. Qureshi et al. / Analytica Chimica Acta 654 (2009) 85–91

89

Fig. 5. Molar mass estimation from UV and light scattering signals. Dotted lines: light scattering response at 90◦ angle; solid lines: UV response at 280 nm. (a) HDL and LDL standards and (b) serum sample.

other than lipoproteins. For instance, the large peak at 6–7 min is, according to its retention time, from albumin. The peak at 12 min can be assigned, as was already pointed out in Ref. [23], to IgG. A regular increase of log MW (estimated from the ratio of the light scattering and the UV signals) with retention time was seen over the fractogram until the end of the flow program. At the high-MW end of the fractogram, in the region where the VLDL fraction is expected to elute, some low abundance compounds exhibit a strong light scattering signal. Still, it is clear that specific lipoprotein fractions cannot be identified in a serum sample on the basis of the 280 nm UV signal and/or the light scattering responses alone. 3.3. Cholesterol and triglyceride measurement in serum The experimental AF4 setup was coupled with the dual enzymatic reactor system discussed above, to get profiles of cholesterol and triglycerides associated with different types of lipoproteins in serum. The short (12 cm) channel was used with a channel flow of 0.6 mL min−1 and two different cross flow programs. The first cross flow program was the TDE program discussed above, with an initial cross flow of 1.2 mL min−1 and a delay/decay time of 7 min. In the alternative program the cross flow was decreased linearly between 7 and 27 min from the original value of 1.20 mL min−1 to a value of 0.01 mL min−1 . The cross flow was maintained for another 10 min at this low value. Fig. 6 shows the two cross flow programs used in this study schematically. From theory described in the literature [35] the relation between retention time and hydrodynamic

Fig. 6. Time-delayed exponential decay and linear cross flow programs used for the fractionation of serum samples, and the calculated relation between retention times and particle diameters. Solid lines: TDE program; dotted lines: linear program.

diameter was calculated for the two programs. The results of these calculations are included in Fig. 6. The figure shows that the linear program provides more selectivity in the 20–50 nm range than the TDE program. On the other hand, with the linear program most of the selectivity is lost for particles >60 nm, while with the TDE program such large particles can still be fractionated. Fig. 7a shows the fractogram obtained for the pooled serum sample from healthy volunteers. The sample appeared to contain 4.2 mM of total cholesterol and 0.9 mM of triglycerides, values that are within the normal range for healthy persons. The first peak at approximately 1 min is the void peak, possibly containing some remaining free cholesterol. The next two cholesterol-containing peaks had hydrodynamic sizes of 4.2 and 8 nm, which corresponds to the size of HDL particles. A large fraction of the total cholesterol was bound to fractions with particle sizes ranging from 15 to 110 nm, which can be identified as LDL and VLDL. The LDL- and VLDL-cholesterol fractions were not clearly resolved using this cross flow program. Most of the triglycerides were present in larger particles with an average diameter of 53 nm. To increase the separation power of the system in the LDL–VLDL range the other cross flow program was tested. In Fig. 7b the cholesterol and triglyceride profiles are shown as obtained with the linear program. The cholesterol is now seen to be distributed over the two HDL fractions, an LDL fraction with a mean particle size of approximately 30 nm, and VLDL particles with an average size >100 nm. Part of the cholesterol elutes after the linear decrease of the cross flow. Triglycerides are seen to be mainly present in the VLDL particles. The cholesterol and triglyceride patterns obtained for the serum samples of sepsis patients are strongly deviating (Fig. 7c–f). Total cholesterol was low for both samples, and the absence of HDLcholesterol is typical. For one patient total triglycerides was higher, for the other lower than in the polled serum. With the first patient sample, a large part of the cholesterol and most of the triglycerides were found to be bound to VLDL. Fig. 7d shows that the linear program is less suitable for this type of samples, since most of the lipoproteins elute after the end of the program without further fractionation. With the TDE program (Fig. 7c) it could be observed that cholesterol and triglycerides were present partly in VLDL particles with a size around 60 nm (the peak at 21 min) and partly in even larger particles with an average size of 110 nm. The other patient sample showed a clearly different pattern. Most cholesterol and triglycerides were bound in two fractions, with sizes of around 30 and 60 nm. The total concentrations of cholesterol and triglycerides, integrated over all separated fractions, could be compared with the data on total concentrations obtained with standard enzymatic tests in the hospital. The comparison is shown in Table 3. There is a fair

90

R.N. Qureshi et al. / Analytica Chimica Acta 654 (2009) 85–91

Fig. 7. Fractionation patterns of cholesterol and triglycerides in serum samples of healthy volunteers (a and b) (pooled) and sepsis patients (c–f). Cross flow program: (a, c, and e) TDE; (b, d, and f) linear decay. Experimental conditions: injection volume 20 ␮L; channel flow rate 0.6 mL min−1 ; enzyme reagent flow rates 0.2 mL min−1 ; UV detection at 500 nm. Table 3 Total cholesterol and triglycerides values measured directly and by integration of the AF4 signals. Serum sample

Pooled Patient 1 Patient 2

Total cholesterol (mM)

Total triglycerides (mM)

Direct

AF4

Direct

AF4

4.5 1.6 1.1

4.1 1.3 0.6

1.2 1.0 0.3

1.0 1.6 0.3

agreement between the values obtained. 4. Conclusions The present study has shown that AF4 is a versatile technique for the fractionation of human serum lipoproteins. The experiments in which the UV signal at 280 nm and the light scattering intensities

were compared (for lipoprotein standards and for the fractionation of serum proteins in general) showed a regular increase of molecular size with retention time. This can be regarded as a confirmation of the assumption for AF4 that the retention time can be taken as a measure for the size of the fractionated sample components. The fractionation could easily be coupled on-line to a dual enzymatic reaction system for the selective simultaneous detection of cholesterol and triglycerides. The use of a short AF4 channel resulted in a higher sensitivity, by a decrease of the dilution of the sample compounds, and in a lower consumption of the (expensive) enzymatic reagents. Still, the mixing of the channel effluent with the reagent solutions was not trivial. To obtain maximum detection sensitivity reactors had to be used that provide a reaction time of 4 min, which is longer than expected on basis of batch experiments. Also it was shown that with knitted reactors (instead of conventional coils) post-fractionation peak

R.N. Qureshi et al. / Analytica Chimica Acta 654 (2009) 85–91

broadening could be decreased and a better repeatability could be obtained. The proposed method can be beneficial for diagnostic studies and for routine analyses in clinical laboratories. The various lipoprotein species present in human serum can be separated in less than 40 min. It was shown that the separation selectivity can be tuned by using a cross flow program appropriate for the specific clinical application. A linear cross flow program gives a high size selectivity in the LDL–VLDL range. With a TDE program an increased selectivity is obtained for the very large lipoprotein species found in serum samples of for instance sepsis patients. Acknowledgements Thanks are due to Dr. J.H.M. Levels (Academic Medical Center, Department of Experimental Vascular Medicine, Amsterdam) for providing the serum samples and for good advise on lipoprotein analysis. Wyatt Technology Europe GmbH (Dernbach, Germany) is acknowledged for instrumental support. Dr. Ch. Johann (Wyatt) and Prof. Dr. P. Reschiglian (University of Bologna) are thanked for helpful discussions. R.N.Q. acknowledges the Higher Education Commission, Pakistan for a research scholarship. References [1] W.P. Castelli, R.J. Garrison, P.W.F. Wilson, R.D. Abbott, S. Kalousdian, W.B. Kannel, JAMA 256 (1986) 2835–2838. [2] W.B. Kannel, J.D. Neaton, D. Wentworth, H.E. Thomas, J. Stamler, S.B. Hulley, M.O. Kjelsberg, Am. Heart J. 112 (1986) 825–836. [3] A.R. Sharrett, C.M. Ballantyne, S.A. Coady, G. Heiss, P.D. Sorlie, D. Catellier, W. Patsch, Circulation 104 (2001) 1108–1113. [4] G.R. Warnick, Clin. Chem. Lab. Med. 38 (2000) 287–300. [5] O. Murch, M. Collin, C.J. Hinds, C. Thiemermann, Intensive Care Med. 33 (2007) 13–24.

91

[6] V. Michel, Z.F. Yuan, S. Ramsubir, M. Bakovic, Exp. Biol. Med. 231 (2006) 490–504. [7] F. Zoppi, Clin. Chem. 31 (1985) 2036–2039. [8] A. Sharma, J.D. Artiss, B. Zak, Clin. Biochem. 20 (1987) 167–172. [9] A. Krug, A.A. Suleimant, G.G. Guilbauit, R. Kellner, Enzyme Microb. Technol. 14 (1992) 313–316. [10] L.C. Wu, C.M. Cheng, Anal. Biochem. 346 (2005) 234–240. [11] A.N. Araujo, J.A.M. Catita, J.L.F.C. Lima, E.A.G. Zagatto, J. Fresenius, Anal. Chem. 360 (1998) 100–103. [12] A.N. Araujo, J.A.M. Catita, J.L.F.C. Lima, J. Farmacol. 54 (1999) 51–55. [13] A. Krug, R. Gobel, R. Kellner, Anal. Chim. Acta 287 (1994) 59–64. [14] J.M. Fernandez-Romero, M.D.L. DeCastro, M. Valcarcel, J. Pharm. Biomed. Anal 5 (1987) 333–340. [15] R.J. Havel, H.A. Eder, J.H. Bragdon, J. Clin. Invest. 34 (1955) 1345–1353. [16] M. Okazaki, Y. Ohno, I. Hara, J. Chromatogr. 221 (1980) 257–264. [17] M. Okazaki, K. Shiraishi, Y. Ohno, I. Hara, J. Chromatogr. 223 (1981) 285–293. [18] I. Hara, K. Shiraishi, M. Okazaki, J. Chromatogr. 239 (1982) 549–557. [19] S. Usui, Y. Hara, S. Hosaki, M. Okazaki, J. Lipid Res. 43 (2002) 805–814. [20] P. Reschiglian, A. Zattoni, B. Roda, L. Cinque, D. Parisi, A. Roda, F. Dal Piaz, M.H. Moon, B.R. Min, Anal. Chem. 77 (2005) 47–56. [21] P. Li, J.C. Giddings, J. Pharm. Sci. 85 (1996) 895–898. [22] P. Li, M. Hansen, J.C. Giddings, J. Liq. Chromatogr. Relat. Technol. 20 (1997) 2777–2802. [23] D.C. Rambaldi, A. Zattoni, S. Casolari, P. Reschiglian, D. Roessner, C. Johann, Clin. Chem. 53 (2007) 2026–2029. [24] G. Yohannes, M. Sneck, S.J. Varjo, M. Jussila, S.K. Wiedmer, P.T. Kovanen, K. Oorni, M.L. Riekkola, Anal. Biochem. 354 (2006) 255–265. [25] I. Park, K.J. Paeng, Y. Yoon, J.H. Song, M.H. Moon, J. Chromatogr. B 780 (2002) 415–422. [26] P. Reschiglian, M.H. Moon, J. Proteomics 71 (2008) 265–276. [27] J.C. Giddings, F.J.F. Yang, M.N. Myers, Science 193 (1976) 1244–1245. [28] A. Litzen, K.G. Wahlund, J. Chromatogr. 476 (1989) 413–421. [29] K.G. Wahlund, A. Litzen, J. Chromatogr. 461 (1989) 73–81. [30] K.G. Wahlund, J.C. Giddings, Anal. Chem. 59 (1987) 1332–1339. [31] A. Litzen, K.G. Wahlund, Anal. Chem. 63 (1991) 1001–1007. [32] K.G. Wahlund, H.S. Winegarner, K.D. Caldwell, J.C. Giddings, Anal. Chem. 58 (1986) 573–578. [33] W.W. Yau, J.J. Kirkland, Sep. Sci. Technol. 16 (1981) 577–605. [34] R.H. Zhu, W. Frankema, Y.L. Huo, W.T. Kok, Anal. Chem. 77 (2005) 4581–4586. [35] P.S. Williams, J.C. Giddings, R. Beckett, J. Liq. Chromatogr. 10 (1987) 1961–1998.