Microenvironmentally controlled secondary structure motifs of apolipoprotein A-I derived peptides

The structure of apolipoprotein A-I (apoA-I), the major protein of HDL, has been extensively studied in past years. Nevertheless, its corresponding three-dimensional structure has been difficult to obtain due to the frequent conformational changes observed depending on the microenvironment. Although the function of each helical segment of this protein remains unclear, it has been observed that the apoA-I amino (N) and carboxy-end (C) domains are directly involved in receptor-recognition, processes that determine the diameter for HDL particles. In addition, it has been observed that the high structural plasticity of these segments might be related to several amyloidogenic processes. In this work, we studied a series of peptides derived from the N- and C-terminal domains representing the most hydrophobic segments of apoA-I. Measurements carried out using circular dichroism in all tested peptides evidenced that the lipid environment promotes the formation of α-helical structures, whereas an aqueous environment facilitates a strong tendency to adopt β-sheet/disordered conformations. Electron microscopy observations showed the formation of amyloid-like structures similar to those found in other well-defined amyloidogenic proteins. Interestingly, when the apoA-I peptides were incubated under conditions that promote stable globular structures, two of the peptides studied were cytotoxic to microglia and mouse macrophage cells. Our findings provide an insight into the physicochemical properties of key segments contained in apoA-I which may be implicated in disorder-to-order transitions that in turn maintain the delicate equilibrium between both, native and abnormal conformations, and therefore control its propensity to become involved in pathological processes.


Introduction
Apolipoprotein A-I (apoA-I) is considered the major component of high-density lipoproteins (HDL) and plays a key role in reverse cholesterol transport, a process that removes excess cholesterol from the cell membranes of peripheral tissues and, therefore, works as a protection mechanism against the development of atherosclerosis [1].
Several HDL models that go from a discoidal to a spherical shape have suggested that the exposed N-and C-terminal domains of apoA-I interact with lipids [12][13][14][15], and shown that the presence of cholesterol and phospholipids determines whether the apoA-I structure is present in an open or closed conformation [6]. The highly hydrophobic C-terminal domain anchors apoA-I to membranes [16], whereas both the N-and C-terminal domains have been shown to play roles in receptordependent binding and lipid remodeling of discoidal HDL particles. Antibody binding and cross-linking studies have shown that the apoA-I N-and C-terminal domains interact with the scavenger receptor B1 (SRB1), responsible for the transfer of free cholesterol and cholesteryl-esters to the liver [17,18]. On the other hand, we have recently shown that several a-helical peptides derived from the Nand C-terminal domains of apoA-I are able to activate the plasma enzyme lecithin cholesterol acyltransferase (LCAT) [19]. While several a-helical structures of apoA-I and their interactions with proteins and lipids have been widely studied, other protein structural features of apoA-I, such as b-sheets, have not been systematically investigated. From the few studies performed, site-directed spinlabeling electron paramagnetic resonance spectroscopy (SDSL-EPR) has revealed the presence of a short b segment at both the N-and the C-terminal regions of the lipid-free form of apoA-I [7,20].
Although several of the conformational transitions in apoA-I are prone to be dependent on their microenvironment, little is known related to its capacity to acquire, depending on this microenvironment, a secondary structure susceptible to aggregation, such as globular forms and b-sheets.
The most hydrophobic and disordered segments of the N-and C-end regions of apoA-I were identified through in silico analysis performed for the complete apoA-I sequence. The physicochemical analysis included properties such as hydrophobic moment, charge, average hydrophobicity associated to steric zippers, theoretical average velocity of aggregation, and possible patterns including residues promoting membrane interactions, self-assembling and aggregation. According to these data, we studied the structural features of short peptides derived between Nand C-terminal domains of apoA-I. We identified several apoA-I modules that promote self-assembly and aggregation, and found that the positions of specific key aromatic residues may affect lipid binding. Our results shed light on the mechanisms that regulate localized conformational transitions that in turn might affect the way apoA-I interacts with HDL particles.

In silico analysis
The primary structure of apoA-I was placed into multiple algorithms to predict disorder-susceptible regions, hydrophobic clusters and aggregation-prone regions. The PONDR-FIT algorithm, a meta-predictor that joins the results of six programs (PONDR-VLXT, PONDR-VSL2, PONDR-VL3, FoldIndex, IUPred, and TopIDP) and forms an artificial consensus from these results, was used to predict conformational disorder [21]. Hydrophobic segments were predicted using the hydrophobic cluster analysis (HCA) server [22]. Regions prone to form amyloid fibrils and globular structures were predicted using the Zyggregator server [23]. Zyggregator uses an algorithm that considers patterns of hydrophobicity, as well as the polar and aromatic amino acid content of amyloidogenic proteins. The prediction of the aggregation rate was calculated using the equation of Dubay et al. [24], considering several factors that influence aggregation, such as pH, ionic force, the presence of specific amino acid sequences, net charge, and total hydrophobicity.

Circular dichroism spectroscopy
Far-ultraviolet (UV) circular dichroism (CD) spectra were recorded on an Aviv 62DS spectropolarimeter in a 0.1 cm quartz cell using an average time of 2.5 s and a step size of 0.5 nm over a wavelength range of 190-260 nm. Sample concentration was determined before each CD measurement and following baseline correction, ellipticity was converted to mean molar ellipticity (H, deg cm 2 dmol -1 ). Secondary structure content was calculated at 190-260 nm using the circular dichroism neural network (CDNN) based software [25].
Transmission electron microscopy and atomic force microscopy Samples of peptide solutions in water were collected after 0, 24, 48, 72, 96 h and 120 days of incubation at 4°C and observed using transmission electron microscopy (TEM) with a JEM-1200EX11 JEOL microscope (70 kV) and atomic force microscopy (AFM) performed with an AFM Digital Instruments/Veeco. Peptide solutions for TEM were deposited on carbon-coated transmission electron microscopy grids and stained with 2 % uranyl acetate. Aliquots of peptide solutions in water were deposited onto freshly cleaved mica, dried under laminar flow for 5 min, and visualized by AFM. Images were obtained using a Multimode microscope (Digital Instruments/Veeco) and a Nano Scope IIIa (Digital Instruments/Veeco) control system. Images (5.0 9 5.0 lm) were obtained in contact mode at a scan frequency of 2 Hz using silicon nitride (Si 3 N 4 ) AFM tips.
Viability assays and optical microscopy RAW (mouse macrophage, ATCC CRL-2467) and EOC cells (mouse microglia, ATCC TIB-71) were grown as previously described by us [26]. Macrophage and microglial cells were placed into 96-well plates at a density of 1 9 10 4 cells/well (100 lL/well) and incubated for 24 h at 37°C. Serial dilutions of aged peptide solutions (120 days at 4°C) were prepared in Opti-MEM reduced-serum medium (OptiMEM) without phenol red. Incubations were performed for 20 h, and cell viability estimated by measuring the cellular reduction of MTT (3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide) as previously described by us [26]. Cell images were observed by optical microscopy at 409 magnification, processed and stored as TIF using an Olympus IX71 microscope and the Image-Pro 3DS 6.0 software.

Discussion
In 1950, Karush [29] proposed that protein-ligand interactions stabilized the best-fitting members within an assembly of structures in equilibrium. Since then, numerous studies have demonstrated that disorder and flexibility in protein structure are important features in the understanding of protein function. In the case of apolipoproteins such as apoA-I, apoC-II, and apoE, it has been found that they correspond to intrinsically disordered proteins. As a consequence of their structural flexibility, the X-ray crystal structure of the C-terminal apoA-I has shown its propensity to destabilization as well as to be able to adopt different conformations when associated to lipids or in a lipid-free state [30,31]. The crystal structure of lipid-free apoA-I demonstrated that although it contains a largely helical four-segment bundle [31], when in solution, this segment adopts a molten globule conformation with its N-terminal domain completely disordered [7]. When apolipoproteins are exposed to air/water and lipid/ water interfaces, evident disorder-to-order conformational transitions take place that might have an important impact upon HDL function [32][33][34]. In this sense, we have previously shown that conformational transitions observed with a series of apoA-I derived peptides stabilize and improve the enzymatic activity of LCAT [19].
In order to predict which segments of apoA-I could present the greater propensity to develop disorder-to-order transitions, using the PONDR-FIT server, we have studied three segments that bind to lipids and acquire a helical conformation that might contribute to the stabilization of lipid-protein interaction. Within each of the disordered Nand C-terminal domains, small segments of hydrophobic bsheet structure are exposed and, therefore, proposed to interact with lipids [7,20]. Our analysis using the HCA server revealed three highly hydrophobic clusters within the helical structure of apoA-I, being the longest one found at the C-terminal domain. This result supports our previous reports and also suggests that lipid-free apoA-I first binds membrane lipids or surface lipids of lipoproteins through the C-terminal fragment. The presence of nonpolar core residues in the protein may be related with this phenomenon, which does not occur when proteins have an inadequate number of hydrophobic residues.
Based on the conformational transitions and cytotoxicity associated to the apoA-I derived peptides used in this work, we suggest that transitions leading to an a-helix formation in this protein at the hydrophilic/hydrophobic interface of membranes can be considered a key feature to explain cell toxicity.
Our study puts into perspective the fact that highly hydrophobic segments of apoA-I present the ability to develop secondary structure disorder-to-order transitions depending on the molecules to which it is associated. The association of these highly hydrophobic segments to specific types of lipid molecules could shift the equilibrium toward the consolidation of a-helical segments that would apparently warranty the normal function of the protein. In contrast, if these segments follow protein-protein interactions or are kept in highly hydrophilic environments, the possibility for the generation of localized pro-aggregation structures might disrupt the normal function of apoA-I.
The dynamic structure exhibited by apoA-I basically supported by intrinsically disordered exposed segments that undergo disorder-to-order and order-to-disorder conformational transitions might also explain the exchangeability properties shown by this family of apolipoproteins. When the protein is located in a highly hydrophilic media with its lateral segments exposed, these segments mostly show a disordered conformation and the permanence of the protein in plasma maintained. Nevertheless, when these segments start to get associated to lipid, there is a shift toward organized secondary conformational structures mostly a-helical structures that in a continuum tend to change the equilibrium, toward the formation and consolidation of lipid loaded particles that eventually give rise to the generation of HDL.
For many years, protein function had resided in the fact that well-ordered structures mostly through rigid tridimensional blocks were fundamental for understanding the way proteins work. Nevertheless, nowadays this concept has been surpassed when recognition has been made to phylogenetically advanced organisms that develop function through and important number of intrinsically disordered proteins and the concept of structural disorder, as a new form of secondary structure of proteins conceived. In this sense, an important number of diseases that in the past had been difficult to understand, during the present days they start to find an explanation in the anomalous folding of proteins [50]. Without a doubt, we can say that, in the near future many diseases with poorly understood origins not only will find a molecular explanation based on this phenomenon, but also in the way intrinsically disordered regions of proteins are modulated. ApoA-I aggregation properties. Lipid-poor apoA-I interacts with the ABCA1 receptor and produce discoidal structures following a process still under active investigation. Discoidal particles are transformed into spherical HDLs by the action of the LCAT enzyme [39]. Only the spherical forms of HDL can interact with the SRB1 receptor. Modified apoA-I can not interact properly with the ABCA1 receptor forming smaller abnormal discoidal HDLs [40]. A direct consequence of this condition is the exposure of highly unstructured apoA-I segments prone to enzymatic hydrolysis [41]. Peptides released by proteolysis might form amyloidogenic structures that can be organized first as micelle-like peptides that can evolve to form globular or protofibrillar structures depending on their residue composition [42,47]. Modified from [33]