Ordered Opening of LDL Receptor Binding Domain of Human Apolipoprotein E3 Revealed by Hydrogen/Deuterium Exchange Mass Spectrometry and Fluorescence Spectroscopy

Abstract

Apolipoprotein E3 (apoE3) is an exchangeable apolipoprotein that plays a critical role in cholesterol homeostasis. The N-terminal (NT) domain of apoE3 (residues 1-191) is folded into a helix bundle comprised of 4 amphipathic α-helices: H1, H2, H3 and H4, flanked by flexible helices N1 and N2, and Hinge Helix 1 (Hinge H1), at the N-and C-terminal sides of the helix bundle, respectively. The NT domain plays a critical role in binding to the low density lipoprotein receptor (LDLR), which eventually leads to lowering of plasma cholesterol levels. In order to be recognized by the LDLR, the helix bundle has to open and undergo a conformational change. The objective of the study was to understand the mechanism of opening of the helix bundle. Hydrogen/deuterium exchange mass spectrometry (HDX-MS) revealed that apoE3 NT domain adopts several disordered and unfolded regions, with H2 exhibiting relatively little protection against exchange-in compared to H1, H3, and H4. Site-directed fluorescence labeling indicated that H2 not only has the highest degree of solvent exposure but also the most flexibility in the helix bundle. It also indicated that the lipoprotein behavior of H1 was significnatly different from that of H2, H3 and H4. These results suggest that the opening of the helix bundle is likely initiated at the flexible end of H2 and the loop linking H2/H3, and involves movement of H2/H3 away from H1/H4. Together, these observations offer mechanistic insight suggesting a regulated helix bundle opening of apoE3 NT domain can be triggered by lipid binding.

1. Introduction

ApoE3 is an exchangeable apolipoprotein that plays a crucial role in cholesterol and triglyceride homeostasis by virtue of its ability to serve as a ligand for the low density lipoprotein receptor (LDLR) family of proteins [1,2]. These receptors mediate endocytosis of the lipoprotein particles thereby decreasing plasma cholesterol and triglycerides levels [3]. ApoE3 also bears the ability to promote efflux of cholesterol from macrophages in atherosclerotic lesions, and to mediate retrograde transport of cholesterol from macrophages to the liver [4].

ApoE3 is composed of 299 amino acids (∼34 kDa) that are folded into a 22 kDa N-terminal (NT) domain (residues 1-191) and a 10-kDa C-terminal (CT) domain (residues 201-299). The two domains are linked by a protease sensitive flexible segment (192-200). The NT domain houses the LDLR binding sites, while the CT domain bears high affinity lipid binding sites, mediates protein-protein interaction that facilitates apoE3 tetramerization and efficiently promotes ABCA1-mediated cholesterol efflux [4]. X-ray crystallographic analysis of apoE3(1-191) at 2.5 Å resolution [5] and of apoE3(1-183) by NMR [6] reveal a long helix bundle (65 Å) comprised of 4 anti-parallel amphipathic α-helices (labeled H1-H4). H1 and H2 are linked by a short helix 1′ (H1′) that is roughly perpendicular to the helix bundle. H4 harbors several positively charged residues between 134 and 150, which display a large region of positive electrostatic potential; this segment, along with Arg172 [7] and additional residues between 174 and 183 [8], plays a significant role in LDLR binding.

As an exchangeable apolipoprotein, apoE3 exhibits the capability to exist in lipidfree and lipid (lipoprotein)-bound states. From a functional standpoint, it is recognized that lipid binding of the NT domain is an a priori requirement for apoE3 to elicit LDLR binding activity [9]. Disulfide bond engineering coupled to lipid binding studies suggest that lipid-association involves an initial interaction between protein and lipids, followed by opening of the four-helix bundle to unveil the hydrophobic interior [7] . This is analogous to that observed for insect apolipophorin III, which involves lipid-triggered opening of a five-helix bundle [10]. Fluorescence resonance energy transfer [11–13][11–13], parallax depth quenching [14] and Attenuated Total Reflectance Fourier Transformed Infrared [15] spectroscopic analyses indicate that interaction of the NT domain helix bundle of apoE3 with phospholipids results in a dramatic conformational change. This involves re-configuration of the lipid-free helix bundle to a belt-like organization of α-helices circumscribing a discoidal bilayer of phospholipids. In this state, the helical axes are oriented perpendicular to the fatty acyl chains of the phospholipid. The transition between lipid-free and lipid-bound forms is a complex process, and likely represents a key regulatory strategy in lipoprotein metabolism.

Despite evidence that suggest lipid-triggered conformational change in apoE3, the details regarding the mode of helix bundle opening is not known. In the current study, we address the issue of helix bundle opening by applying hydrogen/deuterium exchange mass spectrometry (HDX-MS) and site-directed fluorescence spectroscopy to assess amide-backbone structural dynamics and the order of opening and lipoprotein binding of the helices.

2. Materials and methods

2.1. Design and generation of single Cys constructs

Single Cys constructs were designed by examining the X-ray crystal structure of human apoE3(1-191) (PDB ID: 1NFN) and the NMR structure of apoE3(1-183) (PDB ID: 2KC3) using the Chimera software [16]. The introduction of Cys allows covalent attachment of fluorophores containing sulfhydryl-reactive functional groups. The selected sites allow sampling of individual helices by monitoring the fluorescence behavior of the attached probe. Throughout this study apoE3(1-191) encompassing the entire LDLR binding domain and a hexa-His tag at the N-terminal was used and referred to as apoE3 NT domain. Wild type (WT) apoE3 bears a single endogenous Cys at position 112; to generate single Cys constructs for fluorescence studies, C112 was replaced by serine (with no significant change in function or protein fold) [17], and designated as apoE3C112S(1-191). Single Cys constructs were generated in apoE3C112S(1-191)/pET22b vector using the QuikChange II Site Directed Mutagenesis kit (Agilent Technology, Stratagene, La Jolla, CA) to substitute a Cys at desired locations on H1, H2, H3 and H4 (A29C, A62C, A102C and V161C, respectively). The plasmids were sequenced to confirm the presence of the desired mutations.

2.2. Expression, isolation and purification of WT and single Cys apoE3 NT domain variants

E. coli cells were transformed with the pET-22b(+) expression vector encoding WT or single Cys constructs of apoE3 NT domain and bearing ampicillin resistance. The recombinant proteins were over-expressed, isolated in the presence of 4M guanidine HCl (GdnHCl) and purified using a Ni²⁺-affinity matrix (Hi-Trap chelating column, GE Healthcare, Piscataway, NJ) as described previously [18]. Protein purity was verified by SDS-PAGE analysis using a 4-20% acrylamide gradient gel under reducing conditions. Protein concentration was determined in a Nano-Drop 2000/2000c spectrometer (Thermo Fisher Scientific, Wilmington, DE) using the molar extinction coefficient at 280 nm of 27,960 M⁻¹ cm⁻¹. All proteins were unfolded in 6M GdnHCl and freshly refolded by dialysis against buffer at 4 °C for 48 h with 3 changes.

2.3. AEDANS labeling

The purified single Cys apoE3 NT domain variants (5 mg protein in 10 mM ammonium bicarbonate, pH 7.4) were treated with 6 M guanidine HCl (GdnHCl) (Applied Biosystems, Eugene, OR), 2-fold molar excess of Tris(2-carboxyethyl)phosphine hydrochloride (TCEP) (Sigma-Aldrich, St. Louis, MO) and 5-fold excess of 5-((((2-Iodoacetyl)amino)ethyl)amino)napthalene-1-sulfonic acid (IAEDANS) (Invitrogen Molecular Probes, Eugene, OR) for 2 h at 37°C. The labeled proteins were dialyzed extensively against 10 mM ammonium bicarbonate, pH 7.4. The extent of labeling was calculated using the molar extinction coefficients of apoE3(1-191) and that of IAEDANS at 340 nm (5,700 M⁻¹cm⁻¹).

2.4. Circular dichroism (CD) spectroscopy

The secondary structure of labeled apoE3 NT domain variants was assessed by CD spectroscopy on a Jasco J-810-150S spectropolarimeter at 24 °C. Far-UV CD scans were recorded between 185 nm and 260 nm in 10 mM sodium phosphate buffer, pH 7.4 using protein concentrations of 0.2 mg/ml in a 0.1 cm path length cuvette. The CD profiles were the average of four independent scans recorded with a response time of 1s and bandwidth of 1 nm. The molar ellipticity ([θ]) in deg. cm²dmol⁻¹ at 222 nm was obtained using the equation:

$$[\theta {]}_{222\text{nm}}=\text{MRW}(\theta )∕10(l)(c)$$ (1)

where MRW is the mean residue weight (obtained by dividing molecular weight by the number of residues), θ is the measured ellipticity at 222 nm (in degrees), l is the cuvette path-length (in cm) and c is the protein concentration (in g/ml). The percent α-helix content was calculated as described previously [19].

2.5. Fluorescence measurements

Fluorescence emission intensity (FI) measurements of AEDANS-apoE3 NT domain variants were carried out in a Perkin-Elmer LS55B fluorimeter at 25 °C. Fluorescence emission spectra were recorded in 10 mM sodium phosphate buffer, pH 7.4, containing 150 mM NaCl (phosphate buffered saline, PBS) in the absence or presence of 6M GdnHCl between 350 and 600 nm following excitation at 340 nm, at a scan speed of 50 nm/min (3 nm excitation and emission slit widths); an average of 10 scans were recorded. The samples were treated with TCEP to reduce any residual S-S bonds prior to analysis. Fluorescence polarization (FP) measurements of AEDANS labeled apoE3 NT domain variants were carried out in a Perkin-Elmer LS55B fluorimeter at 25 °C with an integration time of 0.1 s and slit width of 5 nm. The excitation and emission wavelengths were 340 nm and 480 nm, respectively.

2.6. HDX-MS

HDX was initiated by adding 10-fold deuterated buffer (10 mM sodium phosphate buffer, pD 7.4, 25°C) to WT apoE3(l-191) (0.1 mg/ml in 10 mM sodium phosphate buffer, pH 7.4) as described previously [20]. Briefly, the exchange-in reaction was arrested with cold deuterated phosphate buffer, pD 2.5 at various time points, and samples were flash frozen with liquid nitrogen until further use. Samples were injected into a Poroszyme Immobilized Pepsin Cartridge, 2.1 mm × 30 mm (Thermo Fisher Scientific, Wilmington, DE) and the resulting peptides separated on a nanoAcquity UPLC BEH C18 column (1.7 μm, 1 × 100 mm) (Waters, Milford, MA). The column was held at ∼0°C and a flow rate of 50 μL/s was used. The gradient and elution conditions employed were as described previously [20]. Mass spectral analyses were conducted using a Synapt G1 electrospray ionization Quadrupole time-of-flight mass spectrometer (Waters, Milford, MA) operated in the MS^E mode [21]. Back exchange was minimized by lowering the HDX workflow to ∼0°C during analysis. Overall, the HDX experimental platform employed in the current study is similar in design as that described by Wales et al [22].

The deuterium level of each peptide was analyzed and calculated as described previously [20, 21]. Briefly, the centroid mass of each peptide at different time points was derived by HDX Browser software version 1.2.4 (Waters, Milford, MA) in combination with HX-Express software [23]. In the present study, we focused on qualitatively characterizing the HDX patterns and mechanisms of each peptide without correction for back-exchange [24]. Therefore, the deuterium levels were reported as determined and expressed as relative deuterium level %D:

$$\%D=\frac{M_D-M}{N}\times 100$$

Where M is the centroid mass of the undeuterated peptide, M_D is the centroid mass of the deuterated peptide, N is the total number of exchangeable amide hydrogens in each peptide.

2.7. Lipid binding analysis

In order to assess helix bundle opening, lipid binding analysis of AEDANS-apoE3 NT domain variants was performed using human low-density lipoprotein (LDL, Sigma Aldrich, St. Louis, MO) as a model system [25]. In this assay, the ability of apolipoproteins to prevent induced aggregation of LDL is used an indicator of their lipid binding ability. About 50 μg of LDL was treated with 200 μg of AEDANS-apoE3 NT domain variants in 50 mM Tris-HCl, pH 7.4, containing 150 mM NaCl and 2 mM CaCl₂ in a 96-well plate. The reaction was carried out at 37 °C in a Varioskan LUX fluorescence plate reader (Thermo Fisher, Grand Island, NY) equipped with a SkanIt Software 4.1 for Microplate Readers RE. LDL aggregation was initiated by addition of 160 mU of phospholipase C (PLC, Sigma Aldrich, St. Louis, MO). Two control reactions were included: one with LDL alone with no added PLC or apolipoproteins, and a second with LDL and PLC alone in the absence of apolipoproteins. All reactions were set up in triplicates. The absorbance at 340 nm (and 600 nm) was monitored every 6 min over a period of 2 h; in addition, the fluorescence emission spectra of each sample were recorded between 420 and 560 nm following excitation at 340 nm (excitation band width set at 5 nm, and measurement time at 50 ms) at each time point.

3. Results

The expressed WT apoE3(1-191) has a total of 208 residues including the 17-residue extension at the N-terminal end (MHHHHHHGLVPRGSIDP) (numbered −17 to −1) bearing a hexa-His tag and a protease cleavage site. The entire sequence bears 6 Pro, reducing the number of backbone N-H groups to 202. A total of 35 peptides (Figure S1) covering 92% amino acid sequence of the apoE3 NT domain were generated in this study. Fifteen out of 35 peptides were selected for further data analysis based on their consistently high quality of mass spectral data which allowed determination of the relative deuterium level at different time points.

Table S1 and Figures S2, S3 summarize the HDX-MS data of the fifteen selected peptides for apoE3 NT domain. The HDX mechanism (Figure S2) combined with the HDX patterns (Figure S3) of each peptide allows us to make inferences about the HDX behavior for most segments of this domain. HDX behavior is governed by solvent exposure, disorder and/or conformational flexibility. Generally, there are two different HDX mechanisms – EX1 and EX2 (refer to Supplementary Data for detailed explanation). Protein refolding rates under EX1 kinetics are significantly slower than deuterium exchange rates; in contrast, proteins undergoing EX2 kinetics have refolding rates that are significantly faster than deuterium exchange rates. In addition, the bimodal isotope pattern detected in the mass spectra of labeled protein/peptide is an indication of EX1 mechanism while a single binomial isotopic distribution is a characteristic feature of a protein/peptide region that undergoes EX2 mechanism [26]. Figure S2 illustrates the mass spectra of each labeled peptide under different time points, with 13 out of the 15 peptides demonstrating complex HDX behavior in which EX1 and EX2 are blurred under most HDX time frames probed. For example, peptides 15-30 and 133-154 exhibited distinct bimodal isotopic distribution (EX1 kinetics) as illustrated by the recordings at 0.5 min with distributions centered around a lower mass and a higher mass (Figure 1A). Bimodal distributions are indicative of the presence of protein population with drastically different protein conformational stabilities of which at least some of the populations conform to EX1-type behavior (Figure 1A). On the other hand, peptides that resulted in mass spectral data displaying a binomial pattern were interpreted as being associated with EX2 mechanism (Figure 1A). Overall, the time evolution of the isotope patterns observed suggest that these regions undergo exchange according to mixed EX1 and EX2 kinetics. Only two peptides, namely 74-78 and 105-123, displayed exchange behavior consistent with EX2 kinetics (Figure 1A).

Figure 1.

Representative mass spectra of tryptic peptides of apoE3 NT domain residues 1–191 demonstrating different HDX mechanism. A. Mass spectra of peptides encompassing residues 15–30 and 133–154, which undergo both EX1 and EX2 HDX kinetics, and peptides bearing residues 74–78 and 105–123, which undergo EX2 kinetics predominantly. Deuteration times from bottom to top: 0, 0.5, 1, 5, 10, 15, 30, 60 and 120 min. B. X-axis is log scale based on time (seconds). Kinetics of HDX patterns of peptides demonstrating disparate exchange-in characteristics.

Figure S3 summarizes the composite plots of the relative deuterium levels of each peptides vs different HDX incubation times. In this work, we limit our interpretation of the HDX data to exchange categories based only on the time-dependent trends of the observed deuterium levels for each peptide. Peptides 15-30, 38-45, 52-60, 94-104, and 133-154 showed relative deuterium levels that gradually increased over exchange time (Figure 1B, Figure S3, Table 1); this time course behavior is usually displayed by conformationally flexible regions that experience diverse folding/unfolding events and inconstant solvent exposure, which collectively cause a certain level of deuterium exchange retardation. In contrast, high relative deuterium levels independent of exchange time often is observed for highly solvent exposed and/or disordered regions which render little protection against deuterium exchange-in. Multiple regions of apoE3 NT domain exhibited this type of exchange behavior, including 1-14, 46-87, 124-133 and 154-187. These regions typically reached deuterium levels over 50% right after 30s D₂O exposure (Figure 1B, Figure S3). The peptide 105-123 covering most part of the sequence of H3 was observed to have the lowest deuterium level with little time dependence; it’s relative deuterium level barely reached 20% after 120 min exposure to D₂O (Figure 1B).

Table 1.

Exchange dynamics of peptides derived from peptic digestion of apoE3 NT domain residues 1–191 following HDX.

Amino Acid Position		Sequence	HDX pattern observed a	Exchange Category b
1	14	KVEQAVETEPEPEL	Mixed	Fast
15	30	RQQTEWQSGQRVELA#L	Mixed	Slow
38	45	RWVQTLSE	Mixed	Slow
46	51	QVQEEL	Mixed	Fast
52	60	LSSQVTQEL	Mixed	Slow
61	73	RA#LMDETMKELKA	Mixed	Fast
74	78	YKSEL	binomial	Fast
79	87	EEQLTPVAE	Mixed	Fast
94	104	SKELQAAQA#RL	Mixed	Slow
105	123	GADMEDVCGRLVQYRGEVQ	binomial	Slow
124	133	AMLGQSTEEL	Mixed	Fast
133	154	LRVRLASHLRKLRKRLLRDAD	Mixed	Slow
154	160	DLQKRLA	Mixed	Fast
162	176	YQAGAREGAERGLSA	Mixed	Fast
177	187	IRERLGPLVEQ	Mixed	Fast
Amino Acid Position		Secondary structure from NMR data*
1	5	Loop N
6	9	Helix N1
12	22	Helix N2
26	40	Helix 1
41	44	Loop 1
45	52	Helix 1'
53	54	Loop2
55	79	Helix 2
80	88	Loop3
89	125	Helix 3
126	130	Loop4
131	164	Helix 4
165	167	Loop 5
168	180	Hinge Helix 1
190	199	Hinge Helix 2

^*Data were adapted from [28], PDB ID: 2L7B. The locations of single Cys substitution in apoE3 NT domain used in subsequent fluorescence analysis (A29C, A62C, A102C, and V161C) are denoted with “#”

^aHDX pattern classification according to Figure S2;

^bfast exchange: ≥ 50% deuterium content right after 30 sec with less time-dependency (Figure S3).

To further evaluate the conformational dynamics at the helix level (Table 1, analyzed further in Discussion) in apoE3 NT domain, a fluorescence spectroscopy approach was employed; AEDANS was covalently attached to a single Cys that was substituted at specified sites on helices H1, H2, H3 and H4, Figure 2.

Figure 2.

Overlay of HDX-MS data onto NMR structure of apoE3. The HDX-MS data were overlaid on the NMR structure of monomeric apoE3 (PDB ID: 2L7B) (only residues 1–191 are shown) to visualize the exchange dynamics at the peptide level. The color coding scheme is as follows: blue represents slow HDX regions, red represents regions that displayed high conformational dynamics resulting in high deuterium levels under the experimental conditions employed. Grey represents segments for which there is no data. The figure also shows amino acid residues involved intramolecular interaction between H1 and H2, and, between H3 and H4 in purple. Residues involved in intramolecular interaction between H2, 3, and 4 and other helices and loops are shown in yellow. Sites probed with AEDANS in the four helices are shown in green.

AEDANS served as a molecular beacon reporting on the microenvironment and mobility of the site it was attached to, and therefore that of the particular helix. The following single Cys variants were overexpressed in E. coll and purified: A29C, A62C, A102C, and V161C (Figure S4, Table 1). Where possible, the selected sites bore conservative substitutions (Ala to Cys) and were facing the aqueous environment allowing access to labeling. The proteins were labeled with AEDANS under reducing and denaturing conditions as described previously [20], and dialyzed for renaturation, achieving labeling stoichiometry between 0.9 and 1.2. The changes in [θ], wavelength of maximal fluorescence emission (λ_max) and FP for each variant were monitored in the absence and presence of 6M GdnHCl.

Figure 3 shows the far UV CD scans of the labeled variants, revealing negative bands as troughs at 208 and 222 nm, which are signature spectral features of α-helical structures. In addition, the α-helical content was calculated to be 54 ± 5% for all variants (Table 2), similar to that observed for WT protein (56 ± 2%), and consistent with the range reported by others (53-64%) [29-31]. This indicates that the labeled proteins were predominantly helical and that the labeling did not significantly affect the overall secondary structure.

Figure 3.

Far UV CD spectra of AEDANS-apoE3 NT domain single Cys variants. Far UV CD spectra of 0.2 mg/ml AEDANS-apoE3 NT domain variants in 10 mM sodium phosphate buffer, pH 7.4 and 10-fold molar excess of TCEP were recorded with a response time of 1s and bandwidth of 1 nm using a 0.1 cm path length cuvette. An average of 4 independent scans is shown.

Table 2.

Spectral characteristics of AEDANS-apoE3 NT domain single Cys variants

AEDANS-apoE3 NT domain single Cys variants	α-helical content Mean± SD (n=3)	λmax in 0 M GdnHCl (nm)	*λmax Shift (nm)	FP values in 0 M GdnHCl (n=3)
A29C	51 ± 7%	475	21	0.114 ± 0.001
A62C	52 ± 8%	488	8	0.100 ± 0.006
A102C	57 ± 2%	481	15	0.111 ± 0.006
V161C	57 ± 7%	475	21	0.116 ± 0.002

^*λ_max shift = λ_max in 6 M GdnHCl - λ_max in 0 M GdnHCl

In the next step, changes in the probe microenvironment and tertiary fold of the labeled variants were assessed. Figure 4 shows the fluorescence emission spectra of AEDANS-apoE3 NT domain variants; the λ_max of variants was between 475 and 488 nm (Table 2). As expected a variant bearing AEDANS in H2 (A62C) displayed a highest red-shifted λ_max 488 nm (being the most solvent exposed of the 4 tested sites). The difference in λ_max between the variants was attributed to variations in polarity of the localized microenvironment. The minor variation in FI between the variants was attributed to small differences in labeling stoichiometry. In the presence of 6M GdnHCl, the λ_max for all variants shifted to 496 ± 1 nm, resulting in a shift that varied between 8 and 21 nm (Table 2).

Figure 4.

Fluorescence emission spectra of AEDANS-apoE3 NT domain single Cys variants. Fluorescence emission spectra of ∼0.05 mg/ml AEDANS-apoE3 NT domain variants (A29C, A62C, A102C and V161C) were recorded at 24 °C following excitation at 340 nm at a scan speed of 50 nm/min (3 nm excitation and emission slit widths); an average of 10 scans were recorded.

Further, the mobility at helix level in apoE3 NT domain was monitored by FP measurements. The FP value of a molecule is proportional to its rotational relaxation time and the magnitude of FP is dependent on the hydrated volume of the probe or probe conjugated to protein. Implicit in this process is the concept that segments with defined secondary structure such as α-helix or β-strand would be relatively more rigid and show higher FP values, while unstructured flexible segments would be more mobile with lower FP values. The expectation was that AEDANS attached to different single Cys sites on apoE3 NT domain variants would monitor and report on mobility of different local segments, similar to that reported by other researchers who noted differences in probe mobility depending on its location on rigid secondary structures versus flexible loop segments [32-34].

The theoretical values of FP can range from 0 to 0.5, with precise and reliable values obtainable with high confidence (FP ± 0.002) [35]. Given this narrow numerical range, it was important to check if FP values change in a predictable manner within this range under the conditions employed in this study; to accomplish this, polarization measurements of AEDANS-WT apoE3 NT domain (bearing an endogenous single Cys at position 112) were carried out under different conditions with known effects, Table 3. In buffer, AEDANS-WT apoE3 NT domain displayed an FP value of 0.19. Upon treatment with 6M GdnHCl, which was expected to completely unfold the NT domain, a lowered FP value (0.07) was noted. In contrast, treatment with 50% trifluoroethanol, which is expected to induce an α-helical structure (increase in molar ellipticity by ∼ 25% compared to WT apoE3 NT) [14] (and therefore further rigidity) in unstructured segments with a propensity for helix formation, yielded an FP value of 0.23. Lastly, using established protocols [14, 36], AEDANS-WT apoE3 NT domain was complexed with phospholipids to yield reconstituted high density lipoprotein (rHDL) particles with molecular mass of ∼ 600 kDa, a slow-tumbling species (and an increase in molar ellipticity by ∼ 25% compared to WT apoE3 NT) [14]. The FP value for AEDANS-WT apoE3 NT domain on rHDL was 0.27.

Table 3.

FP values are predictable indicators of mobility

Sample	FP
AEDANS-WT apoE3 NT domain	0.19
AEDANS-WT apoE3 NT domain + 6M GdnHCl	0.07
AEDANS-WT apoE3 NT domain + 50% trifuoroethanol	0.23
AEDANS-WT apoE3 NT domain on reconstituted HDL	0.27

Having established the trend in FP values for the system under study, the FP values of different AEDANS-labeled single Cys variants of apoE3 NT domain were determined (Table 2). AEDANS at position A62C (monitoring H2) displayed significantly lower (P-value < 0.05) FP value (0.100 ± 0.006) compared to those located at positions A29C, A102C and V161C, monitoring H1, H3 and H4, respectively (0.114 ± 0.00, 0.111 ± 0.006, and 0.116 ± 0.002). The higher FP value (0.19) for AEDANS-WT apoE3 NT domain (bearing an endogenous Cys at position 112) compared to AEDANS located on substituted Cys on H1, H2, H3 or H4 (0.10-0.11) is likely due to the location of Cys112 as shown in the NMR structure (PDB ID: 2L7B) where it is completely buried in the protein interior and protected from solvent (which also resulted in difficulty of AEDANS labeling); this was one of the reasons for introducing a Cys at position 102 to monitor H3.

Lastly, in an effort to mimic a functional state that is expected to lead to structure opening, the AEDANS- labeled variants were exposed to LDL that had been treated with PLC. In this assay, PLC cleaves the phosphocholine group and generates apolipoprotein binding sites on the LDL surface in the form of diacylglycerol. In the absence of a functional apolipoprotein, DAG causes packing defects that leads to aggregation of LDL, which is elicited as increase in turbidity at 340 nm. However, if a functional apolipoprotein is present when the hydrophobic DAG patch is generated, the protein binds to the DAG surface and protects LDL against aggregation; this would be manifested as a delay or prevention of aggregation and onset of turbidity [25]. The rationale for employing this assay is that the binding of AEDANS-apoE3 NT domain would involve opening of the helix bundle to allow the hydrophobic faces of the amphipathic helices H1, H2, H3 and H4 to interact with the exposed DAG surface. Thus, by additionally monitoring the fluorescence characteristics of AEDANS located on different helices as a function of time, we can obtain complementary information to the mass spectral data to check if there are differences between the helices in terms of their lipid binding behavior.

Figure S5, Panel A shows change in absorbance at 340 nm as a function of time. In the absence of added apolipoprotein, a steady increase in absorbance was noted indicative of increased PLC-induced LDL aggregation with time. All 4 AEDANS-labeled variants were effective in affording protection against aggregation as evidenced by the minimal change in absorbance. This indicates that substitution of cysteine residue at the different locations and covalent attachment of AEDANS group did not alter the ability of apoE3 NT domain to undergo helix bundle opening and elicit lipoprotein binding. It also indicates that the condition under which the mass spectrometric and fluorescence data of the lipid free protein were collected was functionally relevant. Further, the change in relative fluorescence intensity (RFI) was followed simultaneously as a function of time, Figure S5, Panel B. There was no significant change in the RFI of AEDANS located at A29C in H1. In contrast, AEDANS on A62C, A102C and V161C experienced a significant (and similar) decrease in RFI as a function of time indicative of changes in the microenvironment of the probe on H2, H3 and H4 upon lipid binding.

4. Discussion

The ability of apoE3 to act as an anti-atherogenic protein hinges on its role in mediating binding to the family of proteins that leads to receptor mediated endocytosis, thereby lowering plasma levels of cholesterol and triglycerides. The LDLR binding ability of apoE3 is elicited predominantly in its lipid-associated state, in which the NT domain adopts a significantly altered conformation compared to the four-helix bundle organization in the lipid free state [9, 37]. The isolated NT domain is monomeric as noted by sedimentation equilibrium and gel filtration studies [31,38].

The CT domain of apoE3 is also rich in amphipathic α-helices, mediates protein-protein interaction to form dimers, tetramers and oligomers, is less stable and bears a significantly higher lipid binding affinity compared to the NT domain, and likely initiates lipid binding [1]. Thus, lipid binding of apoE3 would involve dissociation of the tetrameric protein and replacement of helix-helix interaction by helix-lipid interaction. It likely occurs in a two-step process, with the CT domain initiating a fast lipid binding followed by a slower binding of the NT domain [39]. This would mean that there are at least two different lipid associated states, in both of which the CT segment is lipid associated: in one the NT domain is in a lipid-free state, while in the other, the NT domain is lipid-associated [40, 41]. In the former, the amphipathic α-helices in the NT domain would remain folded in the 4-helix bundle organization, while in the latter, the helix bundle would undergo a conformational opening [7] to facilitate binding of the hydrophobic interior to the lipid surface. The helix bundle opening appears to be a regulatory step in lipid metabolism since the lipid-associated (but not the lipid-free) state of the NT domain would initiate LDLR binding and cellular uptake of the lipoprotein particle. However, the mechanistic details of the initiation of opening remain unclear. The present study addresses this issue using isolated NT domain of apoE3 to see if the initial step involves a collapse of the helix bundle (like opening an umbrella), or an ordered opening about a hinge (like opening a book) with two pairs of helices moving away from each other. Both HDX-MS and site-directed fluorescence spectroscopy analysis suggest opening about a hinge as discussed below.

The hydrophobic sides of the four, main amphipathic α-helices (H1 – H4) of the helix bundle face the protein interior making stable helix-helix contacts, while the polar sides face the aqueous environment as seen in the NMR structure of apoE3(1-183) at 1.2 Å resolution [42] and X-ray structure of apoE3(1-191) at 2.1 Å resolution [5]; a short helix (H1′), which is almost perpendicular to the helix bundle axis, links H1 and H2. NMR studies reveal the presence of additional helices not noted by X-ray analysis, likely due to their flexibility, such as helices N1 and N2 spanning residues 6-9 and 12-22 respectively, and Hinge H1 spanning residues 168-180; both appear to make extensive contacts with the helix bundle (residues 24-164).

A recent study of intact tetrameric apoE3 by HDX-MS [43] found the helix bundle organization to be broadly similar to that of the monomeric apoE3 studied by NMR. These authors noted a stable NT domain helix bundle and a long CT helix with markedly reduced stability for apoE3. In the current study, HDX-MS data of the isolated NT domain (collected under native conditions in the absence of denaturant at pH 7.4) indicate that apoE3 NT domain adopts numerous disordered and unfolded regions contributing to a relatively high level of conformational flexibility. We propose that apoE3 NT’s flexibility likely impedes formation of crystals suitable for high resolution X-ray structure analysis. Likely, this is the reason that there is a limited number of high resolution structures related to apoE3 in the protein data bank to date.

Moreover, analysis of the HDX mechanism indicated the presence of EX1/EX2 mixed kinetics in 13 out of 15 selected peptides, and we interpret EX1 as HDX MS signature for partial unfolding [44]. Interestingly, other researchers also reported that 27 out 43 peptides generated from the N-terminal segment (residues 1-191) of full length apoE3 elicited EX1/EX2 kinetics [43], indicating high occurrence of mixed HDX kinetics. Due to the challenges fitting mixed EX1/EX2 HDX kinetic mechanism, we classified (Figure S2) 15 peptides into two groups based on their experimentally observed HDX pattern: those that exhibit EX1/EX2 mixed HDX profile and those that exhibit binomial HDX profile (Table 1). The exchange-in plots of peptides (Figure S3) can also be split into two groups: peptides showing deuterium levels ≥ 50% right after 30 sec with less time-dependency were classified as fast HDX group; those showing a gradual exchange over time were classified as slow HDX group (Table 1). The peptide 105-123 located in H3 showed the lowest relative deuterium level with little change over time, staying at 20% from 1 min to 120 min. This observation suggested that at least a segment of the amide hydrogens in this region were shielded against solvent exposure during the HDX time frame probed; therefore, we classified this peptide as a slow HDX group.

The NMR structure of full length monomeric apoE3 [28] was used to visualize the deuterium exchange dynamics at the peptide level in the apoE3 NT domain (Figure 2); only residues 1-191 are shown for clarity. Most parts of H1, H3, and H4 showed slow exchange (blue) compared to other regions of the protein (red) (Figure 2); this is consistent with the NMR structure, which indicate that the slow-exchanging segments are stabilized by various intramolecular interactions such as hydrogen bonding and salt bridges (depicted as purple and yellow amino acids in Figure 2). Further, the HDX-MS data of the C-terminal end of H2 indicated that this region showed relatively little protection against deuterium exchange-in compared to H1, H3 and H4. This is consistent with observations from a recent study that focused on isoform-specific differences in helical structure, stability, and dynamics of apoE3 and apoE4 by HDX-MS [37], where it was observed that peptides located in H2 (for example 74-78) elicited lower protection than those in H1, H3 or H4. Taken together, it appears that the flexibility of H2 noted in full length apoE3 is recapitulated in the isolated NT domain as well. Other studies revealed that helices H2 and H3 are linked by a flexible loop, termed 80s loop, rich in acidic residues (79EEQLTPVAEET89), which was originally proposed by Segelke et al [45] as initiation sites for lipid binding based on crystallographic data. The current HDX data indicated that peptide 79-87 exhibited little protection against deuterium exchange-in, providing experimental evidence for the flexibility noted in the structural analysis.

Fluorescence studies offered further insights about the degree of solvent exposure and mobility at a localized level on specific helices. AEDANS on H2 presented the highest λ_max (488 nm) compared to that on H1 (475 nm), H3 (481 nm), and H4 (475 nm), which resulted in the smallest λ_max difference (8 nm) between native and denatured conformation of apoE3 NT domain (Table 2). This observation suggested that the microenvironment of AEDANS on A62C on H2 is relatively polar and solvent exposed. Further, AEDANS located on H1, H3 and H4 had similar FP values (ranging from 0.111 to 0.116) under native conditions, reflecting that they were 14% more rigid than that on H2 (with FP value of 0.100), Table 2, suggesting that H2 bears the least rigidity among the 4 helices. With regard to H1, its stability may be attributed to aromatic stacking interaction between Trp residues at positions 20, 26, 34 or 39. By the same token, AEDANS at position 29 may exhibit some degree of stacking interaction with one or more of the Trp [46], thereby exaggerating the stability of H1 (also discussed in terms of unique lipid binding behavior of H1 below).

Overall, the results of the current study allow us to rationalize the lipid binding interaction of apoE3 with a focus on the NT domain. Taking HDX-MS and fluorescence data together, we propose a two-step opening of the four-helix bundle in apoE3 NT domain during lipid-binding (Figure S6), which starts with a closed conformation (A, Figure S6). Lipid binding in apoE3 CT domain likely induces disruption of the interactions between NT and CT domains, further facilitating dissociation of the CT and hinge domains from the NT domain (Priming step, B, Figure S6). Because of the high mobility/flexibility of H2, the large degree of conformational flexibility at the 80s loop between H2 and H3 as shown in our HDX data and in reports by other researchers [45], and the relatively higher stability of H4 provided by additional interaction with Loop N, Helix N2, and Loop 4 (amino acids colored in yellow in Figure 2), we suggest that the opening of the four-helix bundle is initiated by the flexible end of H2, resulting in disruption of interactions between H1 and H2 (Step 1, C, Figure S6). This is also schematically represented in Figure 5. The conformational changes of H2 likely disrupts interactions between H3 and H4, finally leading to opening about a hinge with H2 and H3 moving away from H1 and H4 optimizing the exposure of the LDLR-binding domain located on H4 (Step 2, D, Figure S6). The model of apoE3 NT helix bundle opening proposed in the present study provides additional details at the molecular level than the previous model, which also suggested helix bundle opening involves H2 and H3 moving away from HN1, HN2, H1 and H4 [28].

Figure 5.

Proposed model of initiation of lipid binding of apoE3 NT domain based on HDX-MS and lipoprotein binding data. Lipid-free apoE3 NT domain is schematically depicted in closed helix bundle conformation with secondary structural elements labeled as shown in Table 1. Initiation of lipid binding at the flexible end of H2 and Loop 3 (red) likely triggers helix bundle opening (as predicted by mass spectrometric data), resulting in disruption of interactions between helices. Lipoprotein binding data suggest that the binding behavior of H1 is significantly different from that of H2, H3 and H4. This may be due to H1 (and a segment close to the N-terminal end of the protein) remaining unbound (Bottom left) or to the microenvironment around H1 remaining unaltered compared to lipid-free state (Bottom right).

Lastly, the functional data on lipoprotein binding of apoE3 NT domain in the current study offered supporting information for the unique behavior of H1. The results suggest that although the NT domain interacted and offered protection against LDL aggregation regardless of the probe location, the binding behavior of H1 (which showed no change in RFI with time) is significantly different from that of the other helices (which showed decrease in RFI with time). This may be interpreted as H1 maintaining interaction(s) with HN1 and HN2 during lipid association with the microenvironment around A29 remaining unchanged, Figure 5. Alternately, it is possible that the entire NT domain binds lipids with the exception of H1 or a segment close to the N-terminal end of the protein. Further studies are needed to understand the lipid-associated conformation of the receptor binding domain of apoE3, which plays a critical role in maintaining plasma lipid homeostasis.

5. Conclusions

In summary, the current HDX-MS studies of the NT domain of apoE3 bearing LDLR binding sites showed that H2 has relatively little protection against exchange-in compared to H1, H3 and H4, and therefore exhibits a higher extent of solvent exposure. Independently, fluorescence analysis indicated that H2 is the least rigid of the four helices in the helix bundle. Overall, the apoE3 NT four-helix bundle opening is likely initiated at the flexible regions (C-terminal end of H2 and the 80s loop between H2 and H3) followed by H2 and H3 moving away from H1 and H4. Further studies involving a combination of site-directed fluorescence labeling and time-resolved measurements of apoE3 may offer additional insights into the lipid binding mechanism at the molecular level.

Highlights.

• ApoE3 LDL receptor binding domain undergoes ordered opening of helix bundle

• Helix H2 showed relatively less protection against deuterium exchange-in compared to helices H1, H3 and H4

• Site-directed fluorescence spectroscopy suggest that H2 is relatively flexible of the 4 helices in the 4-helix bundle

• ApoE3 NT domain 4-helix bundle opening possibly involves movement of H2/H3 away from H1/H4

• Lipoprotein binding behavior of H1 is significantly different from that of H2, H3 and H4

Abbreviations

apoE3:

Apolipoprotein E3

CD:

Circular dichroism

CT:

C-terminal

GdnHCl:

Guanidine hydrochloride

FI:

Fluorescence intensity

FP:

Fluorescence polarization

HDL:

High density lipoprotein

HDX-MS:

Hydrogen/deuterium exchange mass spectrometry

IAEDANS:

5-((((2-Iodoacetyl)amino)ethyl)amino)napthalene-1-sulfonic acid

LDL:

Low density lipoprotein

LDLR:

Low density lipoprotein receptor

NT:

N-terminal

PLC:

Phospholipase C

TCEP:

Tris(2-carboxyethyl)phosphine hydrochloride

WT:

Wild type