Molecular characterization of native and recombinant apolipoprotein A-IMilano dimer. The introduction of an interchain disulfide bridge remarkably alters the physicochemical properties of apolipoprotein A-I.

The disulfide-linked dimer of apolipoprotein A-IMilano (A-IM/A-IM), a natural Arg173-->Cys variant of apoA-I, was purified from carriers' plasma and produced in Escherichia coli. The recombinant A-IM/A-IM is identical to native A-IM/A-IM, by mass spectrometry, SDS-polyacrylamide gel electrophoresis, and isoelectric focusing. Lipid-free A-IM/A-IM undergoes concentration-dependent self-association similar to apoA-I, but at all concentrations apoA-I is more self-associated than A-IM/A-IM. Far-ultraviolet CD spectra of A-IM/A-IM reveal a highly alpha-helical structure predicted to be approximately 65% in the lipid-free and approximately 78% in the lipid-associated states, versus 43 and 73% for apoA-I. A significant loss of alpha-helix occurs below pH 3.5 and above pH 10 in both apoA-I and A-IM/A-IM; A-IM/A-IM constantly shows a higher alpha-helical content than apoA-I over the entire pH range (1.7-12.8), suggesting that hydrophobic forces stabilize the interaction between the two A-IM chains. Indeed, and differently from apoA-I, the alpha-helical content of A-IM/A-IM is minimally affected by solvent ionic strength. The aromatic side chains in both lipid-free and lipid-bound A-IM/A-IM are immobilized in a more asymmetric and hydrophobic environment than in lipid-free apoA-I, the conformation of A-IM/A-IM being instead similar to that achieved by apoA-I following interaction with lipids. The present findings prove that rA-IM/A-IM is structurally identical to the native protein; the conformation of A-IM/A-IM is remarkably different from that of apoA-I, thus possibly explaining some of the peculiar functional properties of the apoA-IMilano dimer.

excess cholesterol is removed from peripheral tissues and transported to the liver for final elimination (1).
ApoA-I actively modulates the reverse cholesterol transport by acting at various stages in the process. Reconstituted HDL, or naturally occurring pre-P-migrating HDL, containing apoA-I as the only protein component, are more efficient acceptors of cell cholesterol than those containing other exchangeable apolipoproteins (2, 31, and lipid-free apoA-I can mediate cellular cholesterol eBw by forming p r e -P -~L -l i k e particles in the extracellular space (4). ~s t e r i~c a t i o n of cell-derived cholesterol, the second step in reverse transport, is also remarkably dependent on the ability of apoA-I to activate the lecithin: cholesterol acyltransferase enzyme (5). Finally, HDL can directly transport the lecithin:cholesterol acyltransferase-derived cholesteryl esters to the liver, in a process possibly mediated by recognition o f apoA-I (6). ApoA-I is a single polypeptide chain, composed of 243 amino acids (7). Studies of apoA-I structure have mainly focused on secondary structure because of the high percentage of a-helix and the motif of amphipathic helices, which is believed to mediate protein-lipid interaction in the lipoproteins (8). The most striking feature of the apoA-I molecule is the presence of internal repeat units of 11 or 22 amino acids, with the periodicity of an a m p h~p~~~c a-helix (9). This structure allows for the main biological activities of apoA-I, i.e. ~ipid-binding (8), lecithin: cholesterol acyltransferase activation (51, stimulation of cholesterol efllw from lipid-loaded cells (101, and interaction with the putative HDL receptor (11). The amino-terminal end of the protein has significantly less a-helix than the COOH-terminal fragment (residues 149-243), which is organized in highly packed antiparallel amphipathic helices, and has the highest affinity for lipids (12,13). Less is known about the tertiary structure of apoA-I; a structural model has been recently proposed, with a domain of interacting a-helical segments in the carboxyl terminus and a globular domain in the amino terminus 113-15). Physicochemical (14,16), monoclonal antibody (17, 18f, and limited proteolysis studies (13) suggest that a hinge domain may exist in the apoA-I molecuIe (91, possibly involved in 1ecithin:cholesterol acyltransferase activation (5,19,20) and promotion of cell cholesterol e M w (21).
Before use, A-IM/A-IM was dissolved in 20 rn phosphate buffer, pH 7.4, containing 6 M GdnHCl and extensively dialyzed against 20 m M phosphate buffer, pH 7.4. ApoA-I was purified from normal human plasma, essentially as described previously (27).
Expression and P~r i~c a~i o n of rA-I~/A-I'~rA-IM/A-I, was expressed in the E. coli K12 derivative strain BC 0050 (xyl-7, ara-14, T4RZ) and excreted to the growth medium. The expression vector was derived form pTrc99A (28). The A-I, cDNA was prepared by site-directed mutagenesis of apoA-I cDNA (29) and placed immediately downstream of an OmpA signal sequence derivative, where the second to last amino acid was changed from Gln to Asn. Two transcription terminators were placed downstream of the A-I, gene and the ampicillin resistance marker of pTrc99A was replaced with a kanamycin resistance marker.
Fermentation was performed at 30 "C and pH 7.0 in a fed-batch bioreactor of 75 liters as in Forsberg et al. (30) with the following exceptions. The medium contained 5 g/liter yeast extract and 0.05 gfliter kanamycin. The culture was grown until the absorbance at 600 nm was 0.6. A-I, synthesis was then induced by adding 0.5 m M isopropyl-l-thio-~-D-g~actopyranoside, and the temperature was raised to 37 "C. Six hours after induction, the concentration ofA-I, in the supernatant, determined by radioimmunoassay (Apolipoprotein A-I RIA, Pharmacia Biotech Inc.), was -2 gfliter.
Cells were separated by centrifugation at 3,000 rpm for 20 min, and the medium was diluted with distilled water to a conductivity below 10 mS/cm prior to loading onto a Q-Sepharose FF (Pharmacia Biotech Inc.) column, equiiibrated with 0.02 M sodium phosphate, pH 8.0. The column was washed and then eluted with 0.4 M NaCl in the same buffer. A-I,containing fractions were applied to a phenyl-Sepharose FF gel column, equilibrated with 0.02 M phosphate buffer, pH 7.5, containing 0.75 M ammonium sulfate. After washing with 0.02 M phosphate buffer, pH 7. phate buffer, pH 7.8, containing 1% mannitoI and lyophiiized. Before use, rA-IM/A-IM was dissolved in 20 nm phosphate buffer, pH 7.4, containing 6 M GdnHCl and extensively dialyzed against 20 m M phosphate buffer, pH 7.4.
Characterization of rA-IJA-IrrA-IM/A-IM (250 pg) was digested with endoproteinase Lys-C (Boehringer Mannheim) (2 pg in 50 p1 of water) for 18 h at 37 "C in 200 p1 of 25 m M Tris-HC1, pH 8.5,l D EDTA. The resulting peptides were separated by reverse-phase HPLC, using a Brownlee Aquapore butyl column developed at 40 "C with a gradient of 5 4 0 % acetonitrile in water, containing 0.25% pentafluoropropionic acid. Each peptide was sequenced by automated Edrnan degradation, and the molecular mass determined by a BioIon 20, "' Cf plasma desorption time-of-flight mass spectrometer (Applied Biosystems). A 59residue COON-terminal fragment (H2: residues 185-243) was prepared by cleavage with hydroxylamine. Lyophilized rA-I,/A-I, (480 pg) was dissolved in 1 ml of cleavage solution cont~ning 2 M hydro~lamine, 3 M GdnHCl, 0.2 M NaOH, and 2 nm EDTA, pH 9.4, and incubated for 5 h at 40 "C. The COOH-terminal fragment was purified by reverse-phase HPLC, using a YMC-Pak Protein column, eluted with a gradient of 10-60% acetonitrile in water, containing 0.25% pentafluoropropionic acid.
For determination of COOH-terminal amino acids, rA-I,/A-IM or its COOH-te~inal H2 fragment (4 nmol) were dissolved in 85 pl of 0.21 M acetic acid, containing 0.05% Brij 35 and 6 nmol of norleucine as an internal standard. Carboxypeptidase P (Sigma) (15 pg in 15 pl of water) was added, and the solution was incubated at 20 "C. Samples (25 pl) were withdrawn after 0, 1,2, and 4 min, and the digestion was stopped by adding 10 pl of 50% trifluoroacetic acid. The released amino acids were analyzed by the Picotag method (Waters). The NH,-terminal amino acid sequence was determined using a Hewlett-Packard GlOOOA protein sequencer or a Milligen Biosearch Prosequencer type 6600, with Sequelon AA membranes (Millipore~illigen).
The molecular mass of the intact protein was determined using positive electrospray mass spectrometry (VG Quattro, Fisons Instruments). The samples were applied in 50% methanol, 1% acetic acid at a flow rate of 5 pl/min. Amino Acid Analysis and Protein Concentration-Protein concentration of stock solutions was assayed by amino acid analysis, performed on a Beckman 6300 amino acid analyzer after acidic hydrolysis of samples in 6 M HCI for 45 min at 155 "C.
Spectroscopy-Circular dichroism (CD) spectra were recorded with a Jasco J500A spectropolarimeter at the constant temperature of 25 "C. Cells were 0.01 and 0.1 cm for the region below 250 nm, and 1.0 cm above 250 nm. Molar mean residue ellipticity (0) was expressed in degrees~cm2~dmol", and calculated as: where cobs is the observed ellipticity in degrees, 115 is the mean residue molecular weight of the proteins, I is the optical path length in centimeters, and c the protein concentration in gramdml. All of the spectra were base line-corrected. The u-helical content was calculated by the method of Chang et al. (31).
The ultraviolet absorption spectra were measured with a Jasco Uvidec-610 spectrophotome~r equipped with on-line digital data processor. Differential absorption spectra were calculated by subtraction of the base-line spectrum stored in memory. Topographical location of tyrosyl residues was investigated by second-derivative analysis, according to Ragone et al. (32). The a and b values were calculated as the peak to trough distances between the minima at 283 and 290.5 nm, and the maxima at 287 and 295 nm, respectively. The fractional tyrosine exposure was then expressed as OL = frr*)/(r, -rJ, where r is the ratio a / b , measured from the second-derivative ~b s o~t i o n spectra, re is the ratio for the protein dissolved in 6 M GdnHC1, and r, is the ratio calculated for a model protein solution containing the same tyrosindtryptophan ratio dissolved in ethylene glycol to simulate residues completely buried in the protein interior.

RESULTS AND DISCUSSION
Characterization ofrA-l,t,/A-I,,,--The HPLC profile of purified rA-Ih,/A-IM showed three major peaks (Fig. 1). Mass spectrometry analysis revealed that the peak eluting at -21 min corresponds to the full-length product, while the other two peaks are disulfide-linked A-I, dimers containing one (peak at 19 min) or two (peak at 17 min) NH,-terminally truncated monomers (see below). The full-length protein was further purified by reversephase HPLC, eliminating most of the truncated monomers (Fig. 2, lane C); this product was used in following experiments. The rA-IM/A-I>, co-migrated with the native protein on SDS-PAGE in both non-reduced and reduced conditions (Fig. 2). No protein thiols were detected by the Ellman's ( 3 6 ) assay in preparations of either native or recombinant A-IM/A-IM, indicating that the purified proteins were entirely in the disulfide form. Two-dimensional gel electrophoretic separation of apoA-I revealed three major and three minor isoproteins (Fig. 3 ) (37). Reduced rA-Ihl/A-IXl, as well as native A-ISl (38), showed the same isoproteins, hut with one less positive charge compared with normal apoA-I, as expected from the amino acid suhstitution in the mutant apolipoprotein.
The amino acid composition of both A-Isl/A-Is, and rA-IM/A-Is, was in close agreement with that deduced from the DNA sequence (39). The primary structure of rA-I>l, as well a s t h e sequencing strategy, is presented in Fig. 4. Determination of the NH,-terminal sequence showed, besides the correct translation product, shorter NH,-terminal sequences in preparation containing truncated rA-13,. The molecular mass (6,530 2 3 Da), amino acid composition and sequence of the COOH-terminal H2 fragment, together with the direct determination of the COOH-terminal Gln by carboxypeptidase P, confirmed t h a t t h e COOH terminus of rA-IM was correct. The rA-Ixl primary sequence was deduced from direct sequencing of the peptides generated by specific cleavage with lysyl and aspartyl endoproteinases. Their molecular mass deviated from theoretical values by less than 0.2%. The direct estimation of the rA-IM/A-Isl mass by electrospray ionization mass spectrometry gave a Cross-linking Studies-Cross-linking with DMS has been repeatedly used to evaluate the self-association of soluble apolipoproteins (35, 40). Cross-linking of apoA-I gives five prominent bands, identified as monomer ( M , 28,000) through pentamer ( M , 140,000) (35). The cross-linking pattern is sensitive to protein concentration, monomers and dimers predominating at the lower values (~0.5 mg/ml), and tetramers and pentamers becoming major species at higher concentrations ( 3 5 ) . Cross-linking experiments with both native and recombinant A-IM/A-IM gave superimposable patterns and showed a similar formation of high molecular weight forms (Fig.  5A). Cross-linked A-IM/A-13, gave four prominent bands; some high molecular weight material was seen at high protein concentration, but was negligible compared to the total area ofthe prominent bands (Fig. 5A ) Wavelength, nm the apoA-I dimer ( M , 56.000). The hand co-mimating with the apoA-I tetramer f M , 112,000, was identified as a cross-linked species made of two A-I\,/A-I\, molecules. Notahly. no hands co-migrating with the apoA-I trimer and pentamrr wrrr present in cross-linked A-I\,/A-I>,, while the two high molecular weight bands were tentatively identified as cross-linked specirs containing three and four A-13,/A-I\, moleculrs. As previously shown for apoA-I (35). the A-I\,/A-I\, cross-linking pattern was sensitive to protein concentration (Fig. 5A ). The pcrccntage of apoA-I and A-IM/A-Is, oligomers. calculated from the

SDS-
PAGE gels assuming that the amount of dye hound per unit weight is constant for all species, is plotted as function of protein concentration in Fig. 513. At all concentrations, thr extent of self-association was significantly lower for A-I\,/A-I,, than apoA-I. Some lipid-binding protein , 4 1 4 9 ) . which exist as up and down cr-helical bundles (44, 451, do not self-associatr in solution and remain monomeric up to very high protein concentrations. Thus, it seems reasonahle to speculntr that &I\,/ A-I, may have a molecular architecture intermediatr hrtween those of self-associating and monomeric glohular apolipoproteins. Circular Dichroism Studies-The secondary and trrtiary structures of native A-I\,/A-I,, and rA-I\,/A-I\,, in their lipid-frre and lipid-associated states, were examined hy CD spectroscopy. A protein concentration of 0.1 mg/ml, a t which both apoA-I 46) and A-I\,/A-I>, (Fig. 5 ) are essentially monomrric. was initially used. In the far-ultraviolet.
rA-I,,/A-I\, and the natitve protein gave superimposable CD spectra (Fig. 6 1 . Thr CD spectra are indicative of a highly cr-helical structurr. exhihiting negative troughs a t 208 and 222 nm, togethrr with a positive band at 190-192 nm. ApoA-I gave a similar spectrum, hut thr values of negative ellipticity for the troughs at 208 and 222 nm were lower than in A-I\,/A-I\, (Fig. 6  a-helical content of -75% being close to the maximum helical potential deduced from the apoA-I primary sequence (9).
The higher a-helical content of A-I,/A-I, versus apoA-I in water may result from pro~in-protein interactions between the a-helices of A-IM, brought in close proximity by the disulfide bond, i.e. in a way not different from that of self-associated apoA-I (46). The CD spectrum of apoA-I was indeed significantly affected by protein concentrations over the 0.1-1.3 mglml range (Table I), as reported by others (35,46); at low protein concentrations, where the monomer species is predominant, apoA-I displayed a 43% a-helical content, which rose up to 63% when the protein concentration was increased into the region where self-association is significant. Notably, the a-helical content ofA-1,IA-I, was less influenced by protein concentration (Table I), confirming the lower tendency of A-I,/A-I, to self-associate deduced from cross-linking studies. The a-helical content of self-associated apoA-I was almost identical to that of covalently linked A-I,IA-I, (64-66%), suggesting that a similar conformational change occurs in apoA-I when two polypeptide chains are brought in close proximity, either by increasing the protein concentration or by introducing a covalent bond.
To monitor the influence of lipid on apolipoprotein conformation, apoA-I, A-IM/A-I,, and rA-IH/A-IM were mixed with DMPC at 25 "C. The co-incuba~on resulted in a complete clearing of the turbid lipid dispersion, indicating dissolution of the large phospholipid multilamellar liposomes and the formation of stable lipid-protein complexes (27). Binding to DMPC caused an increase in the a-helical content of all proteins (Fig, 6, Table  I). The induction of a-helix in A-IM/A-IM was significantly lower compared to apoA-I, bringing the spectra of lipid-bound apoA-I and A-IM/A-IM in close proximity (Fig. 6).
AI1 together, the far-W CD data demonstrate that a maximum a-helical content is reached with both apoA-I and A-IM/A-IM upon self-association, lipid-binding, or exposure to trifluoroethanol (46,47,48).
In other studies, we investigated the effect of solvent pH on the stability of lipid-free apoA-I and A-1,fA-I, solutions. Plots of [@I, as a function of pH are shown in Fig. 7. The ellipticity of both proteins markedly decreased at pH values below 4 and above 10. ApoA-I and A-IdA-I, were insoluble at pH between 4.5 and 5.5, and 3.5 and 5.5, respectively. Similar pH-depend- Protein samples were adjusted with respect to pH over the range 1.5-12.8 and CD spectra recorded at 25 "C. The final protein concentration was 0.1 mg/ml. ent changes in the secondary structure of chicken and human apoA-I have been previously reported (42,48). Deprotonation of lysine residues and protonation of glutamates and aspartates at extreme pH values causes the disruption of intra-and interhelix ion pair and charge-dipole interactions, as well as charge repulsions, resulting in the loss of secondary structure. It should be noted that A-IMIA-IM constantly displayed a significantly higher [0]222 than apoA-I, over the entire pH range (Fig.  71, the pH-induced decrease in secondary structure being similar to that of apoA-I. It can be argued that ionic interactions are only partially responsible for the increased secondary structure of A-IM/A-IH versus apoA-I, hydrophobic forces being possibly involved in the stabilization of the A-I, dimer. This conclusion is supported by the observation of a different effect of solvent ionic strength on apoA-I and A-I,/A-I, secondary structure (Table I). While the a-helical content of A-IdA-IM was largely unaffected by ionic strength, apoA-I showed a 3 and 11% increase in a-helical content upon addition of 50 and 400 m~ NaCl. Thus, while an increase in the polarity of the medium strength affected the hydrophobic interactions between the amphipathic helical segments of apoA-I, favoring protein self-association (46), a similar structuring effect was not found with A-IMIA-IM. This observation is suggestive that hydrophobic interactions are i n~t~m e n~l in the higher secondary structure of A-IM/A-Ihf versus apoA-I.
The tertiary structure of lipid-free A-IMIA-IM and rA-I,/A-I, was studied by near-ultraviolet CD spectroscopy (Fig. 8). The overall features of the two spectra were almost coincident, and remarkably different from that of apoA-I. The shoulder at 264 nm is assigned to phenylalanyl residues (49). The shape of the A-IM/A-I, spectra indicates that the aromatic residues are immobilized in a highly asymmetric, hydrophobic environment. The association of A-IM/A-I, with DMPC did not result in significant changes in the CD spectrum (Fig, S) and a shoulder at approximately 286 nm are detectable (Fig. 8). It is of interest that these bands are reversed in sign and shifted to 283 and 290 nm following association of apoA-I with DMPC (Fig. 8). The almost identical shapes of the t~t o p h a n and tyrosine bands in A-I,!A-I, and lipid-associated apoA-I indicate that comparable changes in apoA-I conformation are induced either by protein-protein interactions between two apoA-I chains, brought in close proximity by a disulfide bond as in A-IM/A-IM, or by association with amphiphile (46,47,50), i.e. in good agreement with the far-ultra~olet CD data.
Spectroscopic Studies-Qrosine exposure in apoA-I, A-IM/A-I,, and rA-I,/A-I, was determined by second-derivative ultraviolet spectroscopy, according to Ragone et al. (32). The absorption and second-deriva~ive spectra recorded for the two A-I, dimers are, again, almost identical; those of rA-I,/A-I, are shown in Fig. 9. In all tested conditions, i.e. proteins in the lipid-free and lipid-associated states or denatured in 6 M GdnHCl, the ultra~olet-abso~tion spectra revealed absorption maxima at 280 nm with a shoulder at 290.2 nm. The second-derivative spectra demonstrated characteristic peaks at 287 and 295 nm with accompanying troughs at 283 and 290.5 nm (Fig. 91. The peak r ratios calculated for apoA-I were 1.10 in phosphate buffer, 0.74 upon association with DMPC and 1.34 in 6 M GdnHC1. These values correspond to a fractional tyrosine exposure (cy) of 0.88 for the native protein, 0.46 for the lipidbound protein and 1.10 for the unfolded state, or exposure of6, 3, and 7 tyrosine residues, respectively. The peak r ratio determined for A-I,/A-I, and rA-I,fA-I, in the denatured state, was of LO7 and 1.01, respectively. These values are s i~~c a n t l y lower than expected from the TyriTrp ratio in the protein sequence, and indicate that two out of the 14 tyrosine residues of A-&/A-X, are buried in the denatured proteins. The peak r ratio decreased to 0.88 and 0.87 in native proteins; association with DMPC did not result in further changes in the peak r ratio (0.82 and 0.85). These values correspond to a fractional tyrosine exposure (a) of -0.60 for native and lipid-boun~A-I,/A-IM, or exposure of 8.2 tyrosine residues. This could be an overestimate of the actual number of tyrosine residues which are exposed in the native A-IM/A-I, molecule, because of the sensiti~ty of the second-derivative a b s o~t i o n spectrum to changes in the electronic state of buried tyrosine residues (32).
Altogether, the near-W CD and the second-derivative UV data demonstra~ that in A-I,/A-~, the two A-I, monomers tightly fold upon themselves, mainly through apposition of the hy~rophobic sites of the amphipathic helical segments. Such a folded structure would drastically reduce the f l e~b i l~t y of the A-I, monomers thus severely affecting their functional properties.
~~p~i c a t i o n s for the F u~c t i o~~ ~o~p h o l o g~ of A-ItdA-Ir The present data indicate that A-IM/A-IM possesses molecular properties that are unique compared to those of normal apoA-I and monomeric A-I,. The -+ Cys substitution in A-IM occurs at the surface of the molecule, in the middle of one of the amphipathic helical segments of apoA-I 127). By analogy with the x-ray crystal structure of the NH&erminal domain of human apoE (451, and according to the amphipathic helix theory of apolipoprotein structure (91, the guanidinium group of Arg17* in apoA-I is salt-bridged with the oppositely charged carboxylate group of Glu'@. The loss of the salt bridge in monome~c A-I, (27) results in a lower a-helical content and increased flexibility in the interaction of the apolipoprotein with lipids (27). By contrast, the present findings indicate that the introduction of a disulfide bridge in A-IM/A-IM results in facilitated interhelix interactions, with an increased secondary structure, and a more folded tertiary structure.
The present concept of apoA-I is that of a highly flexible protein, which exists in an equilibrium between a lipid-free form and a bound lipoprotein form. By analogy with the molecular structure of other apolipoproteins (44, 45), lipid-fme apoA-I should form up and down bundles of amphipathic helices. Following binding to lipids, apoA-I spreads on the lipid surface (15, 171, the hydrophobic interactions in the interior of the lipid-free protein being substituted by interactions with lipids. In view of these unique properties, it is not surprising that the introduction of the disulfide bridge in A-IM/A-IM can result in major changes in molecular organization. While the definition of the three-dimensional structure of A-IM/A-IM awaits the availability of quality crystals for structure determination by x-ray crystallographic methods, the present data suggest that the disulfide bridge in the middle of a n amphipathic lipid-binding helix would drastically affect the ability of apoA-I to readily convert from a lipid-free to a lipidbound state, and vice versa. The kinetics of association with lipids, and the ability to desorb from lipoproteins would be remarkably affected. While lipid binding investigations are presently being carried out in this l a b~r a t o r y ,~ previous studies have shown that A-IM/A-IM confers to carriers' HDL an increased in vitro resistance against modifications induced by the interaction with other lipoproteins, enzymes and lipid-transfer proteins (25). The delayed in vivo catabolism of A-IM/A-IM compared to monomeric and normal apoA-I is also indicative of a tighter binding of A-IM/A-IM to lipids, lipid-free apoA-I being catabolized at a faster rate than in a lipid-bound form (26).
The low degree of structural adaptability of A-IM/A-IM would also affect the physical properties of stable lipid-protein complexes. Indeed, a careful characterization of plasma lipoproteins in the carriers has clearly shown the presence of small, A-IM/A-IM-containing HDL, which are absent in normal human plasma (51). On the other hand, the peculiar molecular characteristics ofA-I,/A-I, may confer to this protein properties not shared by normal apoA-I; indeed, A-IM/A-IM is capable to activate plasminogen, by interacting with binding sites on the plasminogen molecule, which apoA-I can not do (52). Finally, the availability of large amounts of recombinant A-IM/A-IM, structurally identical to the native protein, will allow testing in vivo whether the unique molecular properties of A-IM/A-IM result in an improved activity against atherosclerosis development (53).