Complete high-density lipoproteins in nanoparticle corona Erik Hellstrand 1 , Iseult Lynch 2 , Astra Andersson 3 , Torbjo ¨ rn Drakenberg 1 , Bjo ¨ rn Dahlba ¨ ck 3 , Kenneth A. Dawson 2 , Sara Linse 1,2 and Tommy Cedervall 1,2 1 Biophysical Chemistry, Lund University, Sweden 2 School of Chemistry and Chemical Biology, University College Dublin, Ireland 3 Clinical Chemistry, University Hospital, Malmo ¨ , Lund University, Sweden Nanoparticles entering any biological fluid will imme- diately be covered by a corona of biomolecules. The corona confers biological identity to the nanoparticles, as it is this that interacts with the cellular machinery, thereby determining the nanoparticle destiny in and impacts on organisms. Previously, research has focused on the protein composition of the corona, but many other biomolecules, such as carbohydrates, nucleic acids, and lipids, can also be contained in the corona and play important biological roles that are different from those of proteins. It is therefore essential to extend the characterization of the corona to include other biomolecules, as each plays a vital role in cellular functionality and signaling. Complete characterization of the nanoparticle biomolecule corona may be the key to classifying nanoparticle risk and predicting impacts. The nanoparticles used in this study are copoly- mers of N-isopropylacrylamide (NIPAM) and N-t- butylacrylamide (BAM) at four ratios (85 : 15, 75 : 25, 65 : 35, and 50 : 50) and three sizes (70, 120 and 200 nm in diameter). These nanoparticles are well characterized, and can be prepared monodisperse with defined size and hydrophobicity (through the NIPAM ⁄ BAM ratio), making them highly suitable as model nanoparticles in biophysical and biochemical studies. In addition, these polymer particles have been suggested as drug delivery vessels Keywords apolipoprotein; HDL; lipids; nanotoxicology; transport pathways Correspondence K. A. Dawson, School of Chemistry and Chemical Biology, University College Dublin, Belfield, Dublin 4, Ireland Fax: +353 1 716 2415 Tel: +353 1 716 2447 E-mail: kenneth@fiachra.ucd.ie T. Cedervall, Department of Biophysical Chemistry, Lund University, Box 124, SE-22100 Lund, Sweden Fax: +46 46 222 4116 Tel: +46 46 222 8240 E-mail: tcedervall@yahoo.com (Received 4 March 2009, revised 8 April 2009, accepted 15 April 2009) doi:10.1111/j.1742-4658.2009.07062.x In a biological environment, nanoparticles immediately become covered by an evolving corona of biomolecules, which gives a biological identity to the nanoparticle and determines its biological impact and fate. Previous efforts at describing the corona have concerned only its protein content. Here, for the first time, we show, using size exclusion chromatography, NMR, and pull-down experiments, that copolymer nanoparticles bind cholesterol, tri- glycerides and phospholipids from human plasma, and that the binding reaches saturation. The lipid and protein binding patterns correspond clo- sely with the composition of high-density lipoprotein (HDL). By using frac- tionated lipoproteins, we show that HDL binds to copolymer nanoparticles with much higher specificity than other lipoproteins, probably mediated by apolipoprotein A-I. Together with the previously identified protein binding patterns in the corona, our results imply that copolymer nanoparticles bind complete HDL complexes, and may be recognized by living systems as HDL complexes, opening up these transport pathways to nanoparticles. Apolipoproteins have been identified as binding to many other nanoparti- cles, suggesting that lipid and lipoprotein binding is a general feature of nanoparticles under physiological conditions. Abbreviations BAM, N-t-butylacrylamide; HDL, high-density lipoprotein; HSA, human serum albumin; LDL, low-density lipoprotein; NIPAM, N-isopropylacrylamide; SR-BI, scavenger receptor class B type 1; VHDL, very high-density lipoprotein; VLDL, very low-density lipoprotein. 3372 FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS with potential for controlled release [1]. The identity of proteins in the hard corona on these nanoparticles is well established, and several biophysical parame- ters, including stoichiometries, exchange rates, and enthalpies of the interactions, have been determined [2–4]. Surprisingly, most of the identified proteins are apolipoproteins and other proteins associated with lipoprotein particles. Serum albumin is also present, but has a much higher dissociation rate and lower affinity than the apolipoproteins, and will therefore be replaced over time by apolipoproteins in plasma. The selective binding of apolipoproteins raises many interesting questions that this article begins to address. For instance, it is not known whether the nanoparticles bind only the apolipoproteins or the complete lipoprotein particles, and, if the latter is true, whether a specific lipoprotein particle is selec- tively targeted. Additionally, we are interested in knowing whether the binding is mediated by pro- teins, lipids, or both, and whether the bound protein or lipoprotein particle retains its receptor binding and enzymatic activity. Lipids in blood are transported via lipoprotein parti- cles, of eight to several hundred nanometers in diame- ter, containing lipids and proteins. Triglycerides and cholesterol esters are found in the core of these lipo- protein particles, surrounded by proteins and a mono- layer of phospholipids. The proteins in the lipoprotein particles are mainly apolipoproteins with a range of structural and functional properties. The different clas- ses of lipoprotein particles can be distinguished from one another by the composition of apolipoproteins and lipids. HDL is a heterogeneous population of lipo- protein particles [5]. Apolipoprotein A-I is the main protein component of HDL. In blood, apolipopro- tein A-I recruits phospholipids and cholesterol to form discoidal HDL particles that mature into larger, spher- ical HDL particles. Apolipoprotein A-II and apolipo- protein E are present in subfractions of HDL. Additionally, proteins involved in lipid metabolism are also associated with HDL (but not with other lipopro- tein particles), including lecithin:cholesterol acyltrans- ferase, cholesterol ester transfer protein, and paraoxinase, which are involved in cholesterol meta- bolism and transport. Here we have investigated the binding to copolymer nanoparticles of both lipids and proteins from whole plasma and from isolated lipoprotein fractions, using size exclusion chromatography, gel electrophoresis, NMR, and enzymatic assays. We have investigated in detail the lipid and protein binding pattern, and the results imply that the copolymer nanoparticles bind HDL lipoprotein particles. Results Lipid binding to nanoparticles determined by size exclusion chromatography Size exclusion chromatography was carried out to establish whether lipids are associated with copolymer nanoparticles, as shown in Fig. 1. This technique has been previously used to determine interactions of pro- teins with nanoparticles, as the elution behavior of the proteins is shifted upon interaction with the nanoparti- cles [3]. Copolymer nanoparticles with composition 50 : 50 NIPAM:BAM and diameters of 120 or 200 nm were incubated with plasma, pelleted by centrifugation, washed, and finally redispersed in buffer before being loaded onto a Sephacryl S-1000 SF column. The pres- ence of cholesterol in the eluted fractions was deter- mined by an enzymatic assay. The results (Fig. 1) show that cholesterol coelutes with nanoparticles of different size at their respective elution volumes. This clearly shows that plasma cholesterol associates with the copolymer nanoparticles. Lipids identified by NMR spectroscopy after extraction NMR spectroscopy was used to detect and identify the lipids bound to the copolymer nanoparticles in plasma. NIPAM ⁄ BAM 50 : 50 copolymer nanoparticles were mixed with plasma, and unbound plasma lipids were removed by repeated washing ⁄ centrifugation steps. Lipids were then extracted from the nanoparticles using chloroform ⁄ methanol extraction, and analyzed by NMR spectroscopy (Fig. 2A). A substantial amount of nanoparticles ended up in the chloroform phase, disturbing the spectra and making quantifica- 0 0.2 0.4 0.6 0.8 1.0 010203040 Fraction number 120 nm 200 nm Cholesterol/a.u. Fig. 1. Cholesterol travels with nanoparticles in size exclusion chro- matography. Copolymer particles, 50 : 50 NIPAM ⁄ BAM, with diam- eters of 120 or 200 nm were mixed with human plasma. Unbound plasma lipids were removed by centrifugation. The nanoparticle pel- lets were washed and resuspended in buffer before being loaded onto a Sephacryl S-1000 SF column. Each fraction was analyzed for cholesterol by enzymatic assay. Arrows indicate the elution volume of the nanoparticles. E. Hellstrand et al. Complete high-density lipoproteins in nanoparticle corona FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS 3373 tion difficult. However, 1 H 1D-NMR spectra of lipids extracted from nanoparticles incubated in plasma consistently showed signals specific for cholesterol at 0.68 p.p.m. and for triglyceride at 4.1–4.3 p.p.m., as shown in Fig. 2A. Lipids extracted directly from plasma (Fig. 2B) gave the same peaks as in the spec- trum of lipids extracted from nanoparticles. The peaks were confirmed by reference spectra and through 2D total correlation spectroscopy (TOCSY) NMR experi- ments. In humans, about 80% of the phospholipid content in lipoprotein complexes is phosphatidylcho- line, which could be identified in the extract from plasma from the nitrogen-attached methyl groups visible at 3.4 p.p.m., but not in the extract from nano- particles incubated in plasma. However, as we have shown that lipids bind to the nanoparticles (both by size exclusion chromatography and by enzymatic assay), an enzymatic kit for detection of phosphatidyl- choline was used to verify that phosphatidylcholine is indeed bound to the nanoparticle pellet before extrac- tion. The kit detected phosphatidylcholine in the pellet of 50 : 50 NIPAM ⁄ BAM but not in the pellet of the less hydrophobic 65 : 35 NIPAM ⁄ BAM. The lower affinity for the less hydrophobic nanoparticles corre- lates well with the results from protein, cholesterol and triglyceride binding studies, where binding is seen only with the more hydrophobic 50 : 50 particles, and serves as a negative control (see below). This means that phospholipids are adsorbed onto the nanoparticle but are not detected in the NMR spectrum of the extract, probably as a result of being bound to the nanoparticles also in the chloroform phase and there- fore not being visible, owing to slow tumbling. Surface characteristics are important for lipid binding The surface hydrophobicity of the copolymer nano- particles can be varied by changing the ratio of the two comonomers. After incubation in plasma and repeated washing ⁄ centrifugation, considerable amounts of cholesterol were identified in the pellets of 50 : 50 NIPAM ⁄ BAM nanoparticles (Fig. 3A). In comparison, the less hydrophobic 65 : 35 NIPAM ⁄ BAM nanoparti- cles bound very little cholesterol (Fig. 3A). This shows that the amount of lipids bound to the nanoparticles is dependent on the hydrophobicity of the nanoparticle surface. This behavior is similar to the hydrophobicity dependence in protein binding reported previously [2]. The hydrophobicity dependence was also confirmed in NMR experiments on extracts from 65 : 35 NIPAM ⁄ BAM nanoparticles, showing much lower or no signals from cholesterol and triglyceride as com- pared with extracts from 50 : 50 NIPAM ⁄ BAM nano- particles. The same behavior was seen for phospholipid binding as described in the previous paragraph. 0123456 7 8 0 2 4 6 8 ×10 7 4.14.34.5 0 8 16 ×10 5 0.50.70.9 0 8 14 ×10 6 Extract from plasma p.p.m. Arbitrary units H OCOR 1 H H H H OCOR 2 OCOR 3 ROCO CH 3 p.p.m. 012 34 5 6 78 0 1 2 ×10 9 4.14.3 4.5 5 15 25 ×10 6 0.50.70.9 5 9 13 ×10 7 Extract from 200 nm 50 : 50 incubated in plasma Arbitrary units A B Fig. 2. Lipids identified by NMR spectroscopy after extraction. (A) 1 H-NMR spectrum of lipids extracted from 200 nm 50 : 50 NIPAM ⁄ BAM copolymer nanoparticles incubated in plasma. (B) 1 H-NMR spectrum of lipids extracted from plasma. Magnifications to the left: Double quartet of peaks from the outer four hydrogen protons in the glycerol part of triglycerides. Magnifications to the right: Single peak from cholesterol and cholesterol esters at 0.68 p.p.m. from the methyl group positioned between the hexago- nal and pentagonal carbon rings in cholesterol and cholesterol ester. Complete high-density lipoproteins in nanoparticle corona E. Hellstrand et al. 3374 FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS Copolymer nanoparticles of even lower hydrophobicity (75 : 25 and 85 : 15 NIPAM ⁄ BAM) were also tested, and the amount of bound cholesterol was at back- ground level (data not shown). However, these nano- particles disperse more readily in aqueous buffers, which makes any detailed comparison difficult. The low amount of lipids bound to the less hydrophobic nanoparticles serves as a good negative control for the lipid binding detected on the 50 : 50 NIPAM ⁄ BAM co- polymer nanoparticles. Lipid binding is surface area dependent Copolymer nanoparticles (50 : 50 NIPAM ⁄ BAM) of two sizes, 120 and 200 nm in diameter, were incubated in plasma, and the bound lipids were separated from free lipids by repeated centrifugation and washing. The two nanoparticles produce very similar pellets, but the surface area is 1.7 times greater in the 120 nm nano- particle pellet than in the 200 nm nanoparticle pellet. One milligram of nanoparticles corresponds to approx- imately 1.1 · 10 12 (1.8 · 10 )12 mol) 120 nm particles or 2.4 · 10 11 (4.0 · 10 )13 mol) 200 nm particles. A comparison of the amount of bound lipids in a nonsat- urated system shows that 0.5 mg of 120 nm nanoparti- cles binds about 1.4–1.9 times more cholesterol and triglycerides than 0.5 mg of 200 nm nanoparticles, as shown in Table 1. Consequently, the amount of lipids bound depends on the total surface area rather than on the pellet volume or the number of nanoparticles. In control experiments using plasma without nanopar- ticles, only minute amounts of lipids are detected, and these are subtracted from the values reported in Table 1. An additional factor that may influence the cholesterol binding is the surface curvature ⁄ particle size, as the naturally occurring lipoprotein complexes range in size from 8–10 nm to 100 nm, and the amount of cholesterol associated with each differs, as shown in Table 1. Lipid binding by copolymer nanoparticles reaches saturation If nanoparticles bind discrete lipoprotein particles, the binding may reach saturation. This was tested in experiments with increasing amounts of plasma added to a constant amount of 200 nm 50 : 50 NI- PAM ⁄ BAM nanoparticles. The cholesterol levels were determined by enzymatic assay, as shown in Fig. 3B. After a steep increase of the amount of bound choles- terol, a plateau was reached at about 40% plasma, indicating saturation. This is approximately the same percentage of plasma as that at which protein satura- tion was reached in protein adsorption experiments at similar particle concentrations, suggesting a coupled binding behavior [2]. At saturation, there is approxi- mately 20 nmol cholesterol per mg 200 nm nanoparti- cles, which corresponds to 50 000 cholesterol or cholesterol ester molecules per nanoparticle, or 60 lg HDL per mg nanoparticles (assuming 3.08 wt% cho- lesterol and 17.6 wt% cholesterol ester in HDL). Using a radius of 5 nm and a density of 1.14 gÆmL )1 for HDL, this can be estimated to be 400 HDL molecules per 200 nm particle, or one-quarter of the theoretical maximum coverage in one layer. The same experiments 0 0.2 0.4 0.6 50 : 50 65 : 35 nmol·mg –1 nmol·mg –1 0 10 20 020406080 Plasma conc. / % NIPAM : BAM AB Fig. 3. Surface hydrophobicity is important, and the binding is sur- face limited. (A) Particle surface hydrophobicity is important for plasma lipid binding. Copolymer nanoparticles, 200 nm 50 : 50 or 65 : 35 NIPAM ⁄ BAM, were incubated with human plasma. Unbound lipids were separated from the particles by centrifugation. The particle pellets were washed three times, and the amounts of cholesterol and triglyceride were determined by standard enzymatic assays. (B) Plasma cholesterol binding by 200 nm 50 : 50 NIPAM ⁄ BAM nanoparticles reaches saturation at approximately 20 nmoL cholesterol per mg nanoparticles, which corresponds to 50 000 cholesterol or cholesterol ester molecules per nanoparticle. Copolymer nanoparticles, 0.5 mg, were incubated with increasing amounts of human plasma in a constant volume. Unbound lipids were separated from the nanoparticles by centrifugation. The parti- cle pellets were washed three times, and the amount of choles- terol was determined by standard enzymatic assay. Table 1. Amount and ratio ± standard deviation of lipids on 50 : 50 NIPAM ⁄ BAM copolymer particles with two different diameters and at two plasma concentrations. One milligram of nanoparticles con- tains approximately 1.8 · 10 )3 nmol of 120 nm particles or 4.0 · 10 )4 nmol of 200 nm particles, and the surface area is 1.7 times larger in 1 mg of 120 nm particles than in 200 nm particles. Particles Cholesterol (nmolÆmg )1 particles) Triglyceride (nmolÆmg )1 particles) Molar ratio of cholesterol ⁄ triglyceride 120 nm 50 : 50, 33% plasma 11.1 ± 1.6 3.4 ± 0.1 3.2 ± 0.6 120 nm 50 : 50, 67% plasma 17.9 ± 0.9 6.1 ± 0.6 2.9 ± 0.4 200 nm 50 : 50, 33% plasma 5.9 ± 0.1 2.5 ± 0.3 2.4 ± 0.3 200 nm 50 : 50, 67% plasma 11.0 ± 0.4 4.0 ± 0.4 2.7 ± 0.4 E. Hellstrand et al. Complete high-density lipoproteins in nanoparticle corona FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS 3375 were performed with 70 and 120 nm nanoparticles, but complete saturation could not be reached, owing to the larger particle surface area. Fewer nanoparticles in each sample will lead to small pellets that are difficult to handle in a reproducible way. The cholesterol ⁄ triglyceride ratio is increased in the nanoparticle lipid corona The molar ratio of cholesterol and triglyceride bound to the nanoparticles was established for the 120 and 200 nm 50 : 50 NIPAM ⁄ BAM nanoparticles at two plasma concentrations (Table 1). To ensure that there were no differences in the experimental routine, the pellets were split into two equal parts in the last wash before the cholesterol and triglyceride levels were mea- sured. The cholesterol ⁄ triglyceride molar ratios varied between 2.4 and 3.2, but were within experimental error for all conditions. The cholesterol ⁄ triglyceride molar ratio measured in the same plasma was 1.5. Thus, the cholesterol ⁄ triglyceride ratio was increased by a factor of 2 following interaction with and binding to the nanoparticles as compared with the ratio in plasma, indicating that specific lipoprotein particles were targeted. The specificity is further analyzed in Table 2, where the ratios from Table 1, after conver- sion to mass ratios, are compared to ratios for the different lipoprotein classes. The approximate amount of protein was determined by comparing bound apoli- poprotein A-I with known amounts of apolipopro- tein A-I by SDS ⁄ PAGE, as shown in Fig. S1. Table 2 also shows the apolipoprotein pattern estimated from SDS ⁄ PAGE and compares it with the different lipo- protein classes. Bound protein and lipids from purified lipoprotein particle fractions Three fractions of lipoprotein particles ) chylomi- crons + very low density lipoprotein (VLDL), low-den- sity lipoprotein (LDL), and HDL – were obtained from human plasma by ultracentrifugation in a salt gradient. The HDL fraction was further fractionated into HDL and very high-density lipoprotein (VHDL) fractions. The proteins in the final four fractions were visualized by SDS ⁄ PAGE (Fig. 4A, lanes 1–4), and the proteins bound to 50 : 50 NIPAM ⁄ BAM 200 nm copolymer nanoparticles from each lipoprotein particle fraction are shown in Fig. 4A, lanes 5–8. The main proteins in the chylomicron + VLDL fraction (Fig. 4A, lane 1) were (in size order) apolipoprotein B-100 and ⁄ or apolipopro- tein B-48 (apolipoprotein B-100 is not separated from its truncated variant apolipoprotein B-48 in this sys- tem), human serum albumin (HSA), apolipoprotein E, and apolipoprotein A-I. The same proteins were present on the nanoparticles from this fraction (Fig. 4A, lane 5), but the relative apolipoprotein E and apolipoprotein A-I as compared with apolipoprotein B-100 were much greater on the nanoparticles, indicating preferential binding of apolipoprotein A-1 and apolipoprotein E. In the LDL fraction (Fig. 4A, lane 2), the main pro- tein was apolipoprotein B-100, as expected, but visible amounts of albumin, apolipoprotein E and apolipo- protein A-I were also present. The relative amount of apolipoprotein B-100 was much less on the copolymer nanoparticles incubated in the LDL fraction (Fig. 4A, lane 6), indicating that lipoprotein particles with apoli- poprotein E or apolipoprotein A-I preferentially bind to the copolymer nanoparticles. In the HDL and Table 2. Protein and lipid composition of lipoprotein particles, and the biomolecule corona around the 200 nm 50 : 50 NIPAM ⁄ BAM nano- particles following incubation in plasma. The lipoprotein compositions are from several references collected by LipidBank (http://lipidbank.jp). Chylomicron VLDL LDL HDL 2 HDL 3 Nanoparticle corona Mass ratio of protein, cholesterol, and triglyceride Protein ⁄ cholesterol (gÆg )1 ) 0.3–0.7 0.6 0.4–0.6 1.3–2.1 2.5–6.1 1.3–2.4 Protein ⁄ triglyceride (gÆg )1 ) 0.01–0.02 0.2 1.7–4.2 5.8–13 7.5–18 2.2–3.6 Cholesterol ⁄ triglyceride (gÆg )1 ) 0.03 0.3 3.4–7.5 3.2–8.7 1.5–6 1.5–2.0 Mass percentage of apolipoprotein contributions Apolipoprotein A-I + 65 62 70 Apolipoprotein A-II 10 23 10 Apolipoprotein A-IV Trace 10 Apolipoprotein E 4 13 4 1 10 Apolipoprotein B-48 5–20 37 Apolipoprotein B-100 100 Apolipoprotein C 70–80 50 12 5 Complete high-density lipoproteins in nanoparticle corona E. Hellstrand et al. 3376 FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS VHDL fractions (Fig. 4A, lanes 3 and 4), the major proteins were apolipoprotein A-I and HSA. On the copolymer nanoparticles incubated in the HDL and VHDL fractions (Fig. 4A, lanes 7 and 8), apolipo- protein A-I dominated. Previous studies of proteins bound to the 50 : 50 NIPAM ⁄ BAM copolymer nanoparticles in human plasma did not identify apolipoprotein B in the hard corona [2]. Here we observed, in small amounts, apo- lipoprotein B on nanoparticles incubated in the lipo- protein fractions in which apolipoprotein B is the dominating protein (chylomicrons, VLDL, and LDL). The ratio of apolipoprotein B to apolipoprotein A-I was significantly lower on the nanoparticles relative to the respective lipoprotein fraction, indicating that apoli- poprotein B-100 or LDL bind to the copolymer nano- particles with lower affinity than apolipoprotein A-I or HDL. Size exclusion chromatography was used to further study this competition. NIPAM ⁄ BAM 50 : 50 200 nm copolymer nanoparticles were mixed with lipo- protein particle fractions and, after a washing step, par- ticles and their bound proteins were loaded onto a Sephacryl S-1000 column. The lipoprotein particles did not affect the elution volume of the nanoparticles, and free LDL and HDL clearly eluted separately from the nanoparticles (Fig. 4B). Eluted nanoparticles were pel- leted, and the associated proteins were separated by SDS ⁄ PAGE (Fig. 4C). Only apolipoprotein A-I was present on the eluted copolymer nanoparticles mixed with the LDL fraction, indicating that apolipopro- tein A-I–HDL has a much greater binding affinity than apolipoprotein B-100–LDL for the copolymer nanopar- ticles. As expected, only apolipoprotein A-I was recov- ered from the nanoparticles after mixing with the HDL or VHDL fractions. No proteins could be seen in SDS ⁄ PAGE from the nanoparticles incubated with the B 0 0.2 0.4 0.6 0 50 100 150 200 Elution volume/(mL) A 280 Apo A-I Apo A-I LDL-enriched fraction HDL-enriched fraction B-100 B-100 C A LDL HDL 8 6 4 2 0 Cholesterol/(µg) Loaded Eluted LDL HDL Cholesterol/(a.u) 1 0 B-100 HSA Apo A-I M r 130 72 36 28 17 D 96 150141132123114105 96 150141132123114105 18765432 Elution volume/(mL) LP-fractions Adsorbed on NP Fig. 4. Fractionated lipoproteins (LP) and their binding to copolymer particles. (A) SDS ⁄ PAGE (15% gel) of lipoprotein fractions and nanoparti- cles (NP) incubated in lipoprotein fractions. Lanes 1–4: density fractions from human blood enriched in chylomicron + VLDL, LDL, HDL, and VHDL, respectively. Lanes 5–8: proteins adsorbed to 200 nm 50 : 50 NIPAM ⁄ BAM copolymer nanoparticles incubated in the density frac- tions loaded in lanes 1–4, respectively. Bound proteins were separated from unbound proteins by centrifugation, and desorbed by SDS ⁄ PAGE loading buffer. (B) Size exclusion chromatography of lipoprotein fractions enriched in LDL or HDL and on 200 nm 50 : 50 copoly- mer nanoparticles incubated in the same fractions, separated by centrifugation. Open circles: LDL fraction. Open triangles: HDL fraction. Filled circles: nanoparticles incubated in LDL fraction. Filled triangles: nanoparticles incubated in HDL fraction. (C) SDS ⁄ PAGE on bound pro- teins at the elution volumes from size exclusion chromatography corresponding to the elution volumes of copolymer particles. (D) Amounts of cholesterol loaded in the size exclusion experiments in (A) with LDL and HDL fractions as compared with the relative amounts that coelute with the nanoparticles. E. Hellstrand et al. Complete high-density lipoproteins in nanoparticle corona FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS 3377 chylomicron–VLDL fractions, probably because the amounts bound to the nanoparticles were too small. In all cases of lipoprotein particles binding to the copoly- mer nanoparticles, apolipoprotein A-I was identified, suggesting that the binding is mediated by apolipopro- tein A-1, although a similar role for apolipoprotein A-IV and apolipoprotein E cannot be excluded. The relative amounts of cholesterol on the nanopar- ticles mixed with the HDL or LDL fractions were determined after gel filtration. The volume of HDL or LDL fraction mixed with the nanoparticles was the same in each experiment, which means that there was 6.5 times more cholesterol available in the samples incubated with the LDL fraction than in those incu- bated with the HDL fraction. Nevertheless, there was 1.5 times more cholesterol on the eluted nanoparticles mixed with HDL than on the nanoparticles mixed with LDL, as shown in Fig. 4D. This shows that the nano- particles bind both proteins and lipids with high speci- ficity, supporting the conclusion that HDL rather than LDL binds to the copolymer nanoparticles. Discussion This is, to our knowledge, the first time that lipids have been detected in the biomolecular corona sur- rounding nanoparticles in human plasma. Moreover, we have found that intact HDL particles bind to nano- particles. We show, with three different approaches, that the plasma lipids, cholesterol and triglycerides, are present on 50 : 50 NIPAM ⁄ BAM copolymer nanopar- ticles incubated in plasma. First, cholesterol elutes together with nanoparticles in size exclusion chroma- tography, and the elution position of cholesterol depends on the size of the nanoparticles. Second, nanoparticles preincubated in plasma were extracted with a mixture of chloroform and methanol. In the extract, cholesterol and triglyceride were detected with NMR spectroscopy. Third, cholesterol, phospholipids and triglyceride were detected by enzymatic assay on nanoparticles following incubation with human plasma and separation of unbound lipids by centrifugation. The amount of bound lipids depends on the surface area presented by the nanoparticles and not on the pellet size following centrifugation. Less hydrophobic nanoparticles (65 : 35, 75 : 25 and 85 : 15 NI- PAM ⁄ BAM) bind no or minute amount of lipids in any of these methods, and therefore provide an excel- lent negative control. Lipoprotein particles can be distinguished from one another by the identity and amount of proteins, and by the amount and ratio of cholesterol and triglycerides. We have previously characterized the protein profile and shown that it includes apolipoproteins and enzymes [2]. The identified apolipoproteins and enzymes found in that study correspond to the proteins found mainly in HDL and chylomicrons, which further strengthens the present results. Furthermore, the protein ⁄ cholesterol and protein ⁄ triglyceride ratios correspond well with the ratios in HDL (Table 2), but not with the ratios in larger lipoprotein particles. The protein ⁄ triglyceride and cho- lesterol ⁄ triglyceride ratios are in the lower range, imply- ing that a small number of triglyceride-rich lipoprotein particles, like chylomicrons, also bind to the nanopar- ticles. Chylomicrons, like HDL, contain apolipo- protein A-I, which is identified as the main candidate for mediating binding of lipoprotein complexes to the nanoparticles. In experiments with plasma fractions enriched in different lipoprotein classes, the 50 : 50 NIPAM ⁄ BAM nanoparticles show high specificity for apolipoprotein A-I and bind lipids from the plasma fraction enriched in HDL with higher affinity than those from the plasma fraction enriched to LDL. In conclusion, the results strongly suggest that the 50 : 50 NIPAM ⁄ BAM copolymer nanoparticles in plasma are associated with intact lipid-loaded apolipoprotein A-I-containing lipoprotein particles, preferentially HDL. Apolipoproteins have also been detected on other nanoparticles, e.g. polystyrene, solid lipid nano- particles, and carbon nanotubes, raising the possibility that binding of intact lipoprotein particles extends to other classes of nanoparticles [6–15]. We may speculate a little on the role of the lipopro- tein biomolecular corona in determining the destiny of nanoparticles that enter the bloodstream. The size of the nanoparticles used in this study is of the same order as that of large lipoprotein particles, and there is a possibility that the relative curvature between the nanoparticles and lipoproteins favors the binding of small HDL over larger lipoproteins, although no such tendency can be seen in the results in Table 2. An interesting aspect of the equal size of the nanoparticles and the lipoproteins is, however, that it may open the door to the lipoprotein transport system for the lipo- protein-coated nanoparticles. Lipoprotein particles are known to bind selectively to receptors expressed in organs and cells. There are several receptors described that mediate lipid transport and endocytosis of LDL. In the field of nanomedicine, this has led to numerous publications exploring the possibility of using LDL in drug delivery systems [16,17]. The most common receptor is scavenger receptor class B type 1 (SR-BI), which mediates the bidirectional lipid transfer between VLDL, LDL and HDL and cells [5]. SR-BI is expressed mainly in the liver and steroidogenic glands, but also in brain, intestine, and placenta, and in cells Complete high-density lipoproteins in nanoparticle corona E. Hellstrand et al. 3378 FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS such as macrophages and endothelial cells [5]. Another possible receptor is cubilin, which has been shown to bind apolipoprotein A-I and HDL, and to mediate endocytosis of HDL [18,19]. Cubilin is expressed in the proximal tubule in kidney and in epithelial cells in yolk sac and intestine [20,21]. As the copolymer nanoparti- cles bind HDL, it is possible that there will be recep- tor-mediated uptake and enrichment of the nanoparticles in organs and cells rich in SR-BI and cubillin. Thus, the discovery that nanoparticles can bind intact lipoprotein complexes offers a new window on nanomedicine, as nanoparticles may also hitch a lift on existing cellular lipidic transport pathways. We are currently investigating the biological fate of these lipoprotein-binding nanoparticles in vitro. We have, for the first time, detected lipids in the bio- molecular corona surrounding nanoparticles and char- acterized the lipid binding. The interaction with lipoprotein particles is highly specific, and several experimental findings suggest that the copolymer nano- particles (50 : 50 NIPAM ⁄ BAM) bind complete and intact lipoprotein particles with high specificity for HDL. It is possible that lipoprotein particle binding is a common feature of nanoparticles in a general sense, which makes the mechanism of the binding and the implications for nanoparticle fate and impacts in vivo important topics for future study. Experimental procedures Copolymer nanoparticles NIPAM ⁄ BAM copolymer nanoparticles of diameter 70, 120 and 200 nm and with several different ratios of the comono- mers (85 : 15, 25 : 75, 65 : 35 and 50 : 50 NIPAM ⁄ BAM) were synthesized in the presence of SDS as described previ- ously, although higher SDS concentrations were used in the present work, resulting in similarly sized nanoparticles [22]. The procedure for the synthesis was as follows: 2.8 g of monomers (in the appropriate w ⁄ w ratio) and 0.28 g of cross- linker (N,N-methylenebisacrylamide) was dissolved in 190 mL of MilliQ water with either 0.8 g of SDS (for the 70 nm nanoparticles) or 0.32 g of SDS (for the 200 nm nano- particles), and degassed by bubbling with N 2 for 30 min. Polymerization was induced by adding 0.095 g ammonium persulfate initiator in 10 mL of MilliQ water and heating at 70 °C for 4 h [23]. The nanoparticles were extensively dialyzed against MilliQ water for several weeks, with the water being changed daily, until no traces of monomers, crosslinker, initiator or SDS could be detected by proton NMR (spectra were acquired in D 2 O using a 500 MHz Varian Inova spectrometer). The nanoparticles were freeze- dried and stored in the refrigerator until used. Human plasma, and buffers Blood was drawn from a healthy individual into tubes with EDTA or heparin, and centrifuged at 14 000 g for 30 min. The supernatants from several vials were combined, and aliquots of 400 lL were stored at – 80 °C. Before each experiment, plasma aliquots were centrifuged at 14 000 g to remove possible aggregates. Enzymatic determination of triglycerides and cholesterol Copolymer nanoparticles were dispersed on ice in NaCl ⁄ P i ⁄ EDTA (10 mm phosphate, 150 mm NaCl, pH 7.5, 1 mm EDTA) and mixed with various amounts of plasma. After 1 h of incubation on ice, the mixtures were heated to 23 °C to promote aggregation of the nanoparticles. The samples were centrifuged at 14 000 g for two minutes, and the nano- particle pellets were saved and washed three times with NaCl ⁄ P i ⁄ EDTA. The amount of bound triglycerides was measured by adding 150 lL of a 4 : 1 mixture of Free Glyc- erol Reagent (Sigma, Stockholm, Sweden) and Triglyceride Reagent (Sigma) to the pellets. The nanoparticle pellets were dispersed, and incubated at 37 °C for about 30 min. After incubation, the nanoparticles were pelleted by centrifugation at 14 000 g for two minutes, and the absorbance of the super- natant was measured at 495 nm. The reagents cleave the fatty acids from the triglycerides, and the amount of free glycerol is quantified and used as a reporter of the initial amount of triglycerides, so a standard curve of glycerol was used in each experiment. The amount of bound cholesterol was deter- mined using an Amplex Ò Red Cholesterol kit (Invitrogen, Stockholm, Sweden). The nanoparticle pellets were resus- pended in 50 lL of cholesterol kit reaction buffer and 50 lL of the cholesterol kit working solution. After 30 min of incu- bation at 37 °C, the nanoparticles were pelleted by centrifu- gation at 14 000 g for two minutes, and the fluorescence of the supernatant was measured. The cholesterol kit deter- mines the concentration of total cholesterol, including the part that is present in the lipoprotein particle core as esters with fatty acids. Size exclusion chromatography of nanoparticles and bound plasma lipids Copolymer nanoparticles of 0.5 mL (10 mgÆmL )1 ) were mixed with 0.4 mL of NaCl ⁄ P i ⁄ EDTA or with 0.4 mL of human plasma, and incubated on ice. After 1 h, the samples were heated to 23 °C to promote aggregation of nanoparticles, allowing pelleting by centrifugation. The pellets were washed with 1 mL of NaCl ⁄ P i ⁄ EDTA and dispersed in 0.5 mL of NaCl ⁄ P i ⁄ EDTA on ice. The mixture was loaded onto a 30 · 1.5 cm Sephacryl S-1000 SF column and eluted with NaCl ⁄ P i ⁄ EDTA at a flow rate of E. Hellstrand et al. Complete high-density lipoproteins in nanoparticle corona FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS 3379 0.8 mLÆmin )1 , and 1.7 mL fractions were collected. The elu- tion profile of the nanoparticles was obtained by recording the scattering at 280 nm in a UV spectrometer after aggre- gation of the nanoparticles at 37 °C. To obtain the elution profiles of cholesterol or triglyceride, the nanoparticles in each fraction were pelleted by centrifugation at 14 000 g for two minutes, and the amount of lipid was determined by enzymatic assays as described above. In the experiments with fractionated lipoprotein particles, 1 mL of copolymer nanoparticles (10 mgÆmL )1 ) was mixed with 0.5 mL of lipoprotein particles, and incubated and washed as described above. The mixture was loaded onto a 95 · 1.5 cm Sephacryl S-1000 column and eluted with NaCl ⁄ P i ⁄ EDTA. To analyze the bound proteins, 2 mL frac- tions from each nanoparticle and lipoprotein fraction were spun down, and the proteins were desorbed by SDS ⁄ PAGE loading buffer and separated by SDS ⁄ PAGE (15% gel). Extraction of lipids and detection by NMR Plasma (800 lL) was incubated with 800 lLof10mgÆ mL )1 200 nm 50 : 50 copolymer nanoparticles in NaCl ⁄ P i ⁄ EDTA on ice for 1 h, and then for 30 min at room temperature. The nanoparticles were then harvested by centrifugation at 14 000 g for two minutes, and washed three times with 800 lL of NaCl ⁄ P i ⁄ EDTA. Plasma (800 lL) or nanoparti- cle pellets, resuspended in 800 lL of NaCl ⁄ P i , were extracted against 1.2 mL of 0.5 m KH 2 PO 4 , 6 mL of CHCl 3 , and 2 mL of MeOH. The chloroform phase was evaporated, and the remaining material was dissolved in 600 lL of deuterated (99.8%) chloroform. 1 H ⁄ 1D spectra were recorded at 25 °C using a 600 MHz Varian Unity Ino- va spectrometer. The chemical shift was referenced to the residual chloroform signal (d 7.26). Enzymatic determination of phosphatidylcholine Plasma (200 lL) was incubated with 200 lLof10mgÆmL )1 120 nm 50 : 50 or 65 : 35 copolymer nanoparticles in NaCl ⁄ P i ⁄ EDTA on ice for 1 h, and then for 30 min at room temperature. The nanoparticles were then harvested by centri- fugation at 14 000 g for two minutes, and washed three times with 400 lL of NaCl ⁄ P i ⁄ EDTA. The pellet was analyzed with the kit Phospholipids B no. 990-54009E (Wako, Neuss, Germany) in a reaction volume of 1.5 mL. Purification of lipoprotein particle fractions from plasma Lipoprotein particle fractions were purified as described by Schumaker and Puppione [24]. Lipemic citrate plasma was ultracentrifuged repeatedly at 147 000 g in an Optimal L-70K Beckman centrifuge with a Ti 701 rotor, for 25 h at 12 °C. Before each centrifugation, the density was adjusted with 5 m NaCl and saturated NaBr, both containing 0.04% EDTA. After each centrifugation step, a lipoprotein fraction was collected from the top of the centrifuge tubes. The cor- responding densities from which the fractions were collected are: 1.0068, 1.068, 1.21 and 1.25 gÆmL )1 for chylomicron + VLDL, LDL, HDL, and VHDL, respectively. All fractions were dialyzed against NaCl ⁄ P i ⁄ EDTA. The cholesterol and the triglyceride concentrations were determined to be 1.5 and 6.8 mm in the chylomicron + VLDL fraction, 15 and 2.6 m m in the LDL fraction, and 2.4 and 0.73 mm in the HDL fraction, respectively. In the VHDL fraction, the cholesterol concentration was 0.073 mm, but the triglyceride concentration was too low to be determined. Acknowledgements This work was funded in part by the EU FP6 projects NanoInteract (NMP4-CT-2006-033231), BioNano- Interact SRC and SIGHT (IST-2005-033700-SIGHT), the Swedish Research Council (VR), and Science Foundation Ireland. References 1 Salvati A, Soderman O & Lynch I (2007) Plum-pudding gels as a platform for drug delivery: understanding the effects of the different components on the diffusion behavior of solutes. J Phys Chem B 111, 7367– 7376. 2 Cedervall T, Lynch I, Foy M, Berggad T, Donnelly SC, Cagney G, Linse S & Dawson KA (2007) Detailed iden- tification of plasma proteins adsorbed on copolymer nanoparticles. Angew Chem Int Ed 46, 5754–5756. 3 Cedervall T, Lynch I, Lindman S, Berggard T, Thulin E, Nilsson H, Dawson KA & Linse S (2007) Under- standing the nanoparticle-protein corona using methods to quantify exchange rates and affinities of proteins for nanoparticles. Proc Natl Acad Sci USA 104, 2050–2055. 4 Lindman S, Lynch I, Thulin E, Nilsson H, Dawson KA & Linse S (2007) Systematic investigation of the ther- modynamics of HSA adsorption to N-iso-propylacryla- mide ⁄ N-tert-butylacrylamide copolymer nanoparticles. Effects of particle size and hydrophobicity. Nano Lett 7, 914–920. 5 Zannis VI, Chroni A & Krieger M (2006) Role of apoA-I, ABCA1, LCAT, and SR-BI in the biogenesis of HDL. J Mol Med 84, 276–294. 6 Blunk T, Hochstrasser DF, Sanchez JC, Muller BW & Muller RH (1993) Colloidal carriers for intravenous drug targeting-plasma-protein adsorption patterns on surface-modified latex-particles evaluated by 2-dimen- sional polyacrylamide-gel electrophoresis. Electrophore- sis 14 , 1382–1387. Complete high-density lipoproteins in nanoparticle corona E. Hellstrand et al. 3380 FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS 7 Gessner A, Lieske A, Paulke BR & Muller RH (2002) Influence of surface charge density on protein adsorption on polymeric nanoparticles: analysis by two-dimensional electrophoresis. Eur J Pharm Biopharm 54, 165–170. 8 Gessner A, Waicz R, Lieske A, Paulke BR, Mader K & Muller RH (2000) Nanoparticles with decreasing surface hydrophobicities: influence on plasma protein adsorption. Int J Pharm 196, 245–249. 9 Goppert TM & Muller RH (2003) Plasma protein adsorption of Tween 80- and poloxamer 188-stabilized solid lipid nanoparticles. J Drug Target 11, 225–231. 10 Goppert TM & Muller RH (2005) Polysorbate-stabi- lized solid lipid nanoparticles as colloidal carriers for intravenous targeting of drugs to the brain: comparison of plasma protein adsorption patterns. J Drug Target 13, 179–187. 11 Goppert TM & Muller RH (2005) Protein adsorption patterns on poloxamer- and poloxamine-stabilized solid lipid nanoparticles (SLN). Eur J Pharm Biopharm 60, 361–372. 12 Gref R, Luck M, Quellec P, Marchand M, Dellacherie E, Harnisch S, Blunk T & Muller RH (2000) ‘Stealth’ corona-core nanoparticles surface modified by polyeth- ylene glycol (PEG): influences of the corona (PEG chain length and surface density) and of the core composition on phagocytic uptake and plasma protein adsorption. Colloids Surf B Biointerfaces 18, 301–313. 13 Luck M, Paulke BR, Schroder W, Blunk T & Muller RH (1998) Analysis of plasma protein adsorption on polymeric nanoparticles with different surface character- istics. J Biomed Mater Res 39, 478–485. 14 Muller RH, Ruhl D, Luck M & Paulke BR (1997) Influence of fluorescent labelling of polystyrene particles on phagocytic uptake, surface hydrophobicity, and plasma protein adsorption. Pharm Res 14, 18–24. 15 Salvador-Morales C, Flahaut E, Sim E, Sloan J, Green MLH & Sim RB (2006) Complement activation and protein adsorption by carbon nanotubes. Mol Immunol 43, 193–201. 16 Chung NS & Wasan KM (2004) Potential role of the low-density lipoprotein receptor family as mediators of cellular drug uptake. Adv Drug Deliv Rev 56, 1315–1334. 17 Rensen PC, de Vrueh RL, Kuiper J, Bijsterbosch MK, Biessen EA & van Berkel TJ (2001) Recombinant lipo- proteins: lipoprotein-like lipid particles for drug target- ing. Adv Drug Deliv Rev 47, 251–276. 18 Hammad SM, Stefansson S, Twal WO, Drake CJ, Fleming P, Remaley A, Brewer HB & Argraves WS (1999) Cubilin, the endocytic receptor for intrinsic factor–vitamin B-12 complex, mediates high-density lipoprotein holoparticle endocytosis. Proc Natl Acad Sci USA 96, 10158–10163. 19 Kozyraki R, Fyfe J, Kristiansen M, Gerdes C, Jacobsen C, Cui SY, Christensen EI, Aminoff M, de la Chapelle A, Krahe R et al. (1999) The intrinsic factor–vitamin B-12 receptor, cubilin, is a high-affinity apolipopro- tein A-I receptor facilitating endocytosis of high-density lipoprotein. Nat Med 5, 656–661. 20 Sahali D, Mulliez N, Chatelet F, Dupuis R, Ronco P & Verroust P (1988) Characterization of a 280-kd protein restricted to the coated pits of the renal brush-border and the epithelial-cells of the yolk-sac – teratogenic effect of the specific monoclonal-antibodies. J Exp Med 167, 213–218. 21 Seetharam B, Christensen EI, Moestrup SK, Hammond TG & Verroust PJ (1997) Identification of rat yolk sac target protein of teratogenic antibodies, gp280, as intrinsic factor cobalamin receptor. J Clin Invest 99, 2317–2322. 22 Wu X, Pelton RH, Hamielec AE, Woods DR & McPhee W (1994) The kinetics of poly(n-isopropyl- acrylamide) microgel latex formation. Colloid Polym Sci 272, 467–477. 23 Lynch I, Dawson KA & Linse S (2006) Detecting cryp- tic epitopes created by nanoparticles. Sci STKE 2006, pe14. 24 Schumaker VN & Puppione DL (1986) Sequential flota- tion ultracentrifugation. Methods Enzymol 128, 155–170. Supporting information The following supplementary material is available: Fig. S1. Binding of apolipoprotein A-I to 200 nm 50 : 50 NIPAM ⁄ BAM nanoparticles monitored by SDS ⁄ PAGE on a 15% polyacrylamide gel. This supplementary material can be found in the online version of this article. Please note: Wiley-Blackwell is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corre- sponding author for the article. E. Hellstrand et al. Complete high-density lipoproteins in nanoparticle corona FEBS Journal 276 (2009) 3372–3381 ª 2009 The Authors Journal compilation ª 2009 FEBS 3381 . is expressed mainly in the liver and steroidogenic glands, but also in brain, intestine, and placenta, and in cells Complete high-density lipoproteins in nanoparticle. by apolipoproteins in plasma. The selective binding of apolipoproteins raises many interesting questions that this article begins to address. For instance,