Murine apolipoprotein serum amyloid A in solution forms a hexamer containing a central channel

Abstract

Serum amyloid A (SAA) is a small apolipoprotein that binds to high-density lipoproteins in the serum. Although SAA seems to play a role in host defense and lipid transport and metabolism, its specific functions have not been defined. Despite the growing implications that SAA plays a role in the pathology of various diseases, a high-resolution structure of SAA is lacking because of limited solubility in the high-density lipoprotein-free form. In this study, complementary methods including glutaraldehyde cross-linking, size-exclusion chromatography, and sedimentation-velocity analytical ultracentrifugation were used to show that murine SAA2.2 in aqueous solution exists in a monomer-hexamer equilibrium. Electron microscopy of hexameric SAA2.2 revealed that the subunits are arranged in a ring forming a putative central channel. Limited trypsin proteolysis and mass spectrometry analysis identified a significantly protease-resistant SAA2.2 region comprising residues 39-86. The isolated 39-86 SAA2.2 fragment did not hexamerize, suggesting that part of the N terminus is involved in SAA2.2 hexamer formation. Circular-dichroism spectrum deconvolution and secondary-structure prediction suggest that SAA2.2 contains approximately 50% of its residues in alpha-helical conformation and <10% in beta-structure. These findings are consistent with the recent discovery that human SAA1.1 forms a membrane channel and have important implications for understanding the 3D structure, multiple functions, and pathological roles of this highly conserved protein.

Fig 1.

Probing the oligomeric structure of SAA2.2. (A) GCL of SAA2.2 (0.1 mg/ml). The last lane (0.6*) shows cross-linking of a refolded SAA2.2 sample after it was unfolded in 6 M urea. (B) Size-exclusion HPLC of SAA2.2 at 0.1 and 0.01 mg/ml. Molecular mass standards (MW stds.) are BSA (67 kDa), ovalbumin (43 kDa), superoxide dismutase (32 kDa), horse myoglobin (19 kDa), and ribonuclease A (14 kDa). A 10-fold dilution of the SAA2.2 sample, from 0.10 to 0.01 mg/ml (10–1.0 μM), shows the same retention time, suggesting a submicromolar-to-nanomolar dissociation constant (K_d).

Fig 2.

SAA2.2 exists as a mixture of hexamer–monomer in equilibrium. (A) Sedimentation-velocity profiles of a 0.27 mg/ml sample of SAA2.2. Scans for analysis were recorded every 1 min; for clarity only eight representative scans 3 min apart are shown. The velocity data exhibit two resolvable boundaries, one corresponding to the monomer and the second boundary corresponding to hexamer. (B) Analysis of the sedimentation profiles in A using the time-derivative method (dc/dt) reveals the presence of two sedimenting species with an average sedimentation coefficient of 1.5 (±1) and 5.4 S (±1). The solid line represents a two-species fit of the data (o) using a Gaussian function.

Fig 3.

EM and projection structure of SAA2.2. Micrographs revealed doughnut-shaped particles that were homogeneous in size. (1) Representative projection average (177 particles) of the particle with a diameter of ≈8 nm and a central pore with a diameter of ≈2.5 nm. (2) Projection average (204 particles) that indicated a sixfold symmetry, which was applied to generate the sixfold-symmetrized average shown in 3. The side length of 1–3 is 20 nm. The circles in the raw image indicate bright spots that might represent monomeric SAA2.2.

Fig 4.

Time course of limited trypsin digestion of SAA2.2 (0.4 mg/ml) analyzed by SDS/PAGE (A) and RP-HPLC (B–F). The labels on RP-HPLC plots are based on the MS results (Fig. 5 and Table 1). MW stds., molecular mass standards.

Fig 5.

Liquid chromatography/MS analysis of the SAA2.2 fragments generated after 30 min of trypsin proteolysis.

Fig 6.

Secondary-structure analysis of SAA2.2 and SAA39–86. (A) Structure prediction of SAA2.2 using the PROF algorithm (26) yields the probability of each residue to be in α-helix (pH), β-sheet (pE), or loop/other (pL) structure. (B) Far-UV CD spectra (solid lines) of SAA2.2 (0.1 mg/ml) and SAA39–86 (0.17 mg/ml) were deconvoluted by using the programs CONTINLL, CDSSTR, and SELCON3 (18). The average secondary structures obtained for SAA2.2 and SAA39–86 were 54% α-helix, 7% β-structure, and 39% other and 17% α-helix, 31% β-structure, and 52% other, respectively. The reconstructed CD spectra from the deconvoluted secondary structure are in excellent agreement with the experimental spectra.