Lipoprotein (a) Particles Characterization by Dynamic Light Scattering

Lp(a) is a novel cardiovascular risk factor resembling an LDL particle. It includes a copy of apolipoprotein (a) [apo(a)], whose molecular weight is dependent on the number of genetically encoded kringle IV type 2 (KIV-2) repeats and inversely related with Lp(a) plasma concentration and risk. The reason for this inverse relationship is unclear and, particularly, there are no data regarding the size of Lp(a) particles carrying apo(a) with different molecular weights. The aim of the present work was to explore if a relationship existed between apo(a) molecular weight and particles size in Lp(a) samples carrying 20, 25 and 28 KIV-2 repeats (K20, K25 and K28, respectively). Dynamic Light Scattering (DLS) measurements were performed on affinity-purified Lp(a). A preliminary finding was that particles were typically distributed into three different size groups instead of the single one expected. No difference in average particle size between Lp(a) carrying different apo(a) isoforms was found. However, the percentage of medium-sized particles in each sample was found to be inversely related to the number of KIV-2 repeats (R 2 =0.99), with a clear predominance in K20 (58.53%). These data deserve further investigations, as they might be potentially relevant to explain the pathogenic role of low molecular weight Lp(a) isoforms.


INTRODUCTION
Lipoprotein (a) [Lp (a)] is a plasma LDL-like particle [1] and it is an independent risk factor for atherosclerotic diseases because of its concentration-dependent pro-atherogenic, prothrombotic and antifibrinolytic properties [2,3]. Different epidemiological studies have suggested that Lp(a) could increase the risk of cardiovascular disease and ischemic stroke, especially if associated with other predisposing factors such as hypercholesterolemia, hypertension, diabetes mellitus and low HDL level [2].
Lp (a) is present in the arterial wall of atherosclerotic lesions and its accumulation involves recruitment of macrophages [4,5]. These cells incorporate the lipid component and are transformed into foam cells. In this environment, macrophages also release cytokines and growth factors that lead to the proliferation and migration of vascular smooth muscle cells from the intima-media and contribute to plaque formation. Lp(a) also has a direct effect on fibrinolytic factors, in particular, it stimulates the expression of PAI-1 in endothelial cells and it has been described as directly inhibiting plasmin by complexing and inactivating tissue-type plasminogen activator (tPA) [6].
Lp (a) composition is similar to that of LDL in terms of cholesterol, triglycerides, phospholipids, and apoB100. The unique and distinctive component of Lp (a) is the apolipoprotein (a) [apo (a)] glycoprotein, a member of the plasminogen gene family, with a strong structural homology with plasminogen [7]. Apo (a) is disulfide linked to the apoB100 of the LDL-like particle. It is known to be a very heterogeneous glycoprotein including domains referred to as kringle IV, kringle V, and the protease domain [8]. The apo(a) kringle IV domains can be classified into 10 types (KIV 1 -KIV 10 ) on the basis of amino acid sequence [9]. Kringle IV type 2 is present in a variable number of copies (from 3 to 48), which generates Lp(a) isoform size heterogeneity (more than 25 described) in humans [10][11][12], with the low MW species related to high plasma concentration and vice versa [13].
The size of the main lipoprotein classes is well defined [14], but recent evidence suggests that further classifications in class subtypes according to size can be relevant to determine the related risk [15]. This is probably related to the fact that different particle sizes correspond to a different lipid composition and also to a different capacity to overcome the endothelium and to be internalized and metabolized by target cells.
In the present study, we wanted to verify for the first time if affinity purification of plasma-derived Lp(a) combined with Dynamic Light Scattering (DLS) can represent a useful method to investigate the relationship between the apo(a) isoform, expressed as a number of KIV-2 repeats, and the size of affinity purified Lp(a) particles obtained from homozygous individuals.

Molecular Weight Determination of Apo(a) in Patients Sera
Plasma samples were kindly donated by the Immunohaematology and Transfusional Service of IRCCS San Matteo Foundation (Pavia, Italy), and were derived from healthy plasma donors aged between 18 and 60 years old. The apo (a) phenotype was analyzed by high-resolution phenotyping with sodium dodecyl sulfate agarose gel electrophoresis (SDS-agarose) under reducing conditions as outlined previously [13] with slight modifications. Briefly, 15 µl of EDTAplasma samples were pretreated with 30 µl of a reducing solution. The submarine electrophoretic run was performed on a 1% SDS-agarose gel and electrophoresis was carried out in tank buffer at 80 V and 0.04 A for 14 h. Reduced samples (20 µl) were applied into wells, at 3 cm from the cathode of the gel. The separated proteins were transferred onto a nitrocellulose membrane (Bio-Rad, Segrate, Italy) by a capillary blotting technique and tested with a rabbit polyclonal anti-human Lp(a) antiserum (DAKO, Glostrup, Denmark) diluted 1:500 in 1% BSA overnight. A peroxidase-conjugated goat anti-rabbit immunoglobulin (DAKO, Glostrup, Denmark, 1:1000 in TBS) was used as a secondary antibody. The membrane was developed with 50 ml of TBS, 500 µl of hydrogen peroxide, 30 mg of 4-chloro-1-naphthol (4CN) dissolved in 10 ml of cold methanol for 15 min. The reaction was then stopped washing with water. Relative band mobility was determined as referred to the mobility of apo(a) standard isoforms included in each blot (values: 35, 27, 23, 19, and 14 KIV-2 repeats; Immuno AG, Wien, Austria) and the number of KIV-2 repeats in each sample was thus calculated.

Determination of Lp(a) Concentration
Macra® Lp(a) Test Kit (Trinity Biotech) is a sandwich ELISA which specifically detects the apo(a) moiety of Lp (a). The monoclonal antibody was used to coat the wells of a microtiter plate and used to capture Lp(a) from the sample during a one-hour incubation at room temperature. After washing, a polyclonal anti-Lp(a) horseradish peroxidase (HRP) conjugate antibody was added and incubation performed at room temperature for 20 min. The plate was then washed and a chromogenic substrate for horseradish peroxidise (o-phenylenediamine) was provided to produce a coloured solution. After 20 min, the reaction was stopped with sulfuric acid and the concentration of Lp(a) mass (mg/dl) was quantitatively determined by comparison of the absorbance of the sample at 492 nm with a standard curve prepared with known concentrations of Lp (a). The standards contained Lp(a) in human plasma in a buffered solution and corresponded to 0, 5, 10, 20, 40 and 80 mg/dl. The kit intra-assay %CV was 1.4-7.0% and the inter-assay %CV 6.0-12.7%.

Size Measurement
Lp(a) samples were concentrated to 1 mg/ml on a 50 kDa cut-off Amicon Ultra-15 Centrifugal concentrator (Millipore) and then dialysed by diafiltration against PBS or refolding buffer using the same concentrator. Volume reduction to 500 µl, 1:30 dilution with buffer and centrifugation were repeated at least 3 times. Measurement of particle size was performed by a Malvern Zeta Sizer ZS90 using default settings. Buffer was used as a negative control. Measurements were repeated 3 times for each sample.

Statistical Analysis
The experiments were performed twice. Statistical analysis was performed by ANOVA and Tukey post-hoc analysis using the VassarStats server (http://vassarstats.net/).

Apo(a) Molecular Weight Determination
Among the plasmas tested in western blotting (of which 14 are shown in Fig. 1), four homozygous samples (single band phenotypes) with the following number of kringles were selected for further analysis: 20 (K20, two samples, of which one shown in Fig. 1), 25, 28 (K25, K28, one sample each). This prevalence of single band phenotypes (30.8%) is not too far from that present in white Americans (23.9%), while the expected prevalence of the specific phenotypes here detected is 1.51%, 1.36% and 1.9% in the same population [16].

Affinity Purification of Lp(a)
Lp(a) concentration in starting plasma samples was comprised between 40 and 140 µg/ml and purification yields were 3-5 µg/ml plasma (3.5-7.5%). Two independent purifications were repeated for each sample and sample integrity confirmed by western blotting.

Lp(a) Size
DLS allows a label-free measurement of particle size. Zeta-sizer measurements were performed on Lp(a) dialysed versus refolding buffer or PBS, in order to assure the best stability of the sample, in the first case, and more physiological conditions, in the second case. Very similar results were found in both cases and the data obtained with refolding buffer are here presented and discussed.
Unexpectedly, a composite profile including 3 peaks of different size was typically found, as summarized in Table 1 Along with size, we also measured a second parameter provided by the Zeta-sizer, peak intensity ( Fig. 1), expressed as a percentage of total signal and hence an indicator of the relative abundance of a particle species present in a given sample. Within small particles (blue in Fig. 2), the relative peak intensity was higher in K28 compared to K20 and K25 (P=.0019 and P=.0083, respectively).
No difference was detected in medium sized lipoproteins (red in Fig. 2, P=.17), while a higher proportion of large particles (green in Fig. 2) was detected (P=.02) in K28 compared to K20 (P=.025). Within each isoform, significant differences were also observed in relative peak intensities for K20 and K25 (Fig. 2, P=.000 for K20, P=.000 for K25), but not for K28, where the three species seemed to occur in roughly the same percentage (P=.21). Particularly, in both K20 and K25 the most represented particle size was the medium-sized one (58% for K20 and 46% for K25, respectively).
Interestingly, normalizing the intensity of medium and large particles by the one of small particles, the value for large ones is similar in the three apo(a) types, while medium particles greatly prevail in K20 and K25 samples compared with K28 (4 and 2.5 times, respectively, Fig. 3). Indeed, the inverse relationship between KIV-2 repeats number and percentage of medium particles has an R 2 of 0.99. The direct relationship between KIV-2 repeats number and percentage of large particles has an R 2 of 0.88. R 2 becomes only 0.68 for small particles.

DISCUSSION
In this work, the particle size of purified Lp(a) carrying three different homozygous apo (a) isoforms was evaluated by DLS, which allows lable-free measurements.
The first observation was a consistent heterogeneity of Lp (a) particles, demonstrated by the presence of three peaks instead of the single peak expected for a single particle species. Why did we observe several peaks? A possible interpretation is based on the isolation procedure we adopted. In fact, most of the described purification procedures to isolate Lp(a) include a gradient centrifugation step. This introduces a selection bias based on particle density leading to the isolation of a very homogenous fraction, which could however reduce the real complexity of the sample. In contrast, a single affinity chromatography step, like the one we applied, should capture all the plasma elements that can bind anti-Lp(a) antibodies. We can therefore hypothesise that three types of plasma particles bind to anti-apo(a) antibodies and the heterogeneity we detected could be a relevant physiological feature thus far undetectable with non-DLS techniques and using material derived from non-affinity based purification methods. Because of the relevance microvesicles (100-1000 nm in size) are acquiring in biology [17], this issue requires further investigations. Indeed, apo (a) has been detected in plasma microvesicles [18] and LDL particles have been shown to co-purify with them [19].
An alternative interpretation is that two of the three observed peaks could be artefacts deriving either from Lp(a) degradation or aggregation, the latter possibly due to the high Lp(a) concentration needed for DLS measurements. Stabilization of purified Lp(a) in solution, in fact, is not trivial, especially at high concentration and, in our experience, precipitation of purified Lp (a) in PBS can start occurring already at 80 µg/ml, a concentration 12.5x lower than the one required for DLS determinations.
Both hypothesis (aggregation and microvescicles) can explain the high SDs of large particles measurements and are not mutually exclusive. Because of this, they will be kept in mind in the following experiments.
A size for Lp(a) particles of 28.3±0.5 nm and 25 nm has been previously determined by nondenaturing native gel electrophoresis ( [20] and [14], respectively), while a smaller diameter can be deduced from the experiments by Fless et al. [21] (ca 20 nm). Interestingly, none of the sizes we determined for small, medium and large particles perfectly fits with these reported data, the closest values being those of the medium particle series (45.70-56.92 nm). The reason for this discrepancy may lay on two factors. One is the fact that DLS measures the hydrodynamic diameter of particles, which includes both the molecule and the adherent solvent layer. The other is that apo(a), which is a very hydrophilic molecule because of its highly glycosilated state, is often described as floating from the Lp(a) particle like a sort of tail, thus turning its approximate shape from that of an ideal sphere into that of an elliptical object (Fig. 4, [22]).
In DLS, the direct observations are the intensity fluctuations due to the diffusion of the particles, and this diffusion coefficient, interpreted as a hydrodynamic size using the Stokes Einstein equation, is compared to that of a hypothetical sphere moving with the same diffusion coefficient. Particularly, the direct intensity size distribution may inherently be weighed to larger sizes, due the fact that the scattering intensity is proportional to the sixth power of size. Under ideal conditions, the number distribution transformation of the DLS result should be very close to other measurements techniques, but, for polydisperse samples like ours, this can be more difficult to achieve. It is worth noticing that the hydrodynamic diameter of a particle is relevant in biological assays and in In vitro migration. According to our data, we can exclude a correlation, either direct or inverse, between apo(a) isoform and any of the three Lp(a) particle sizes we detected, which was the original aim of the present work. However, analysing peak intensity, a difference in the relative distributions of the three peaks was found within each isoform, with a higher prevalence of small and large particles in K28 compared to K20 and K25. Moreover, a perfect inverse correlation was found between the percentage of medium-sized particles present in each isoform and the number of KIV-2 repeats carried by apo (a). This behaviour might depend on the physico-chemical features of Lp(a) carrying different apo (a) isoforms and might have physiological and pathological relevance. In fact, it has been shown that, in patients, an LDL pattern that has more small dense LDL particles, called Pattern B (19.0-20.5 nm), equates to a higher risk factor for coronary heart disease than does a pattern with more of the larger and less-dense LDL particles (Pattern A, 20.6-22 nm) [23]. This is thought to be because the smaller particles are more easily able to penetrate the endothelium, whose normal gaps are 26 nm in diameter [23]. According to our data, the relative distribution by size could also be relevant for Lp (a).

CONCLUSIONS
In summary, a difference in average size was not detected in affinity-purified Lp(a) carrying apo(a) with different isoforms, but two serendipitous observations were reported. The first regards the co-existence of three particle types with different hydrodynamic diameters in Lp(a) samples purified by affinity chromatography. The second regards the different relative abundance of these species in Lp(a) samples with different apo (a) isoforms. Particularly, the highest prevalence of medium particles, whose size is the most compatible with the one thus-far described for Lp(a) by other techniques, was measured in K20.
In order to characterise the different particles we detected in DLS, affinity chromatography could be implemented with separation techniques apt to isolate and analyse homogenous Lp(a) particles subpopulations, such as highperformance liquid chromatography (HPLC), electron microscopy (EM), nuclear magnetic resonance (NMR), Size Exclusion Chromatography -Multi-angle Light Scattering (SEC-MALS) and Field-Flow Fractionation (FFF). Unfortunately, plasma samples from healthy donors have the great disadvantage of being very limited in available amounts. In order to circumvent this issue, we are also planning to produce apo(a) and Lp(a) of given molecular weights in recombinant form in order to have access to an indefinite source of homogenous samples. More effort is needed to characterize the role of Lp(a) as a particle in atherogenesis. The data here presented, despite their limitations mainly due to the original combination of techniques adopted, suggest for the first time that DLS analysis could be relevant to study the physiology and the pathogenetic mechanisms of Lp (a), which, if these results are validated, might need to be considered not as a single particle species, but as composed by multiple subclasses.

CONSENT AND ETHICAL APPROVAL
All the plasma samples were obtained from donors affiliated to the Immunoheamatology and