The roles of the RIIβ linker and N-terminal cyclic nucleotide-binding domain in determining the unique structures of the type IIβ protein kinase A: a small angle x-ray and neutron scattering study

Abstract

Protein kinase A (PKA) is ubiquitously expressed and is responsible for regulating many important cellular functions in response to changes in intracellular cAMP concentrations. The PKA holoenzyme is a tetramer (R2:C2), with a regulatory subunit homodimer (R2) that binds and inhibits two catalytic (C) subunits; binding of cAMP to the regulatory subunit homodimer causes activation of the catalytic subunits. Four different R subunit isoforms exist in mammalian cells, and these confer different structural features, subcellular localization, and biochemical properties upon the PKA holoenzymes they form. The holoenzyme containing RIIβ is structurally unique in that the type IIβ holoenzyme is much more compact than the free RIIβ homodimer. We have used small angle x-ray scattering and small angle neutron scattering to study the solution structure and subunit organization of a holoenzyme containing an RIIβ C-terminal deletion mutant (RIIβ(1-280)), which is missing the C-terminal cAMP-binding domain to better understand the structural organization of the type IIβ holoenzyme and the RIIβ domains that contribute to stabilizing the holoenzyme conformation. Our results demonstrate that compaction of the type IIβ holoenzyme does not require the C-terminal cAMP-binding domain but rather involves large structural rearrangements within the linker and N-terminal cyclic nucleotide-binding domain of the RIIβ homodimer. The structural rearrangements are significantly greater than seen previously with RIIα and are likely to be important in mediating short range and long range interdomain and intersubunit interactions that uniquely regulate the activity of the type IIβ isoform of PKA.

Keywords: Cyclic AMP (cAMP); Intrinsically Disordered Protein; Protein Domain; Protein Dynamic; Protein Kinase A (PKA); Protein Structure; Small Angle Neutron Scattering (SANS); Small Angle X-ray Scattering (SAXS).

FIGURE 1.

Small angle x-ray scattering of RIIβ(1–280) holoenzyme as compared with full-length RIIβ holoenzyme. A, I(q) versus q plots of RIIβ(1–280) holoenzyme formed with hydrogenated C subunit (open squares) and with deuterated C subunit (open diamonds). The scattering data were binned, and the error bars reflect the S.D. of the data within the bin. The I(q) versus q plot of full-length RIIβ holoenzyme (from Vigil et al. (31)) is shown for comparison (open circles). The I(q) scale is arbitrary, and the plots have been offset to facilitate comparison. B, Guinier plots of RIIβ(1–280) holoenzyme formed with hydrogenated C subunit (open squares) and deuterated C subunit (open diamonds) as compared with full-length RIIβ holoenzyme (open circles). The I(q) scale is arbitrary and has been offset to facilitate comparison. The line shown with each dataset is a linear fit to the data in the range from q_min to q² = 0.0013. This value of q² corresponds to q × R_g = 1.3. C, pairwise length distance distribution curves of RIIβ(1–280) holoenzyme formed with hydrogenated C subunit (open squares) and deuterated C subunit (open diamonds) as compared with full-length RIIβ holoenzyme (open circles). The area under each curve has been scaled to reflect the square of molecular weight of the particle. Estimated errors are equal to or smaller than the symbols.

FIGURE 2.

Small angle neutron scattering of type IIβ(1–280) holoenzyme. A, I(q) versus q plots of a neutron scattering solvent contrast series using type IIβ(1–280) holoenzyme containing deuterated C subunit. The solid green lines tracing through the scattering data points at each solvent contrast point are predicted scattering curves obtained from the type IIβ(1–280) holoenzyme structural model shown in Fig. 4A. The dashed black lines represent the range of predicted scattering curves of the 75 best-fit structural models. arb. units, arbitrary units. The scattering data were binned, and the error bars reflect neutron counting statistics and the S.D. of the data within the bin. B, P(r) function of RIIβ(1–280) homodimer in the holoenzyme complex (filled circles) was extracted from the SANS solvent contrast series using MULCh. The P(r) of free RIIβ(1–280) homodimer determined by SAXS (open circles) is shown for comparison. The size of the symbols is comparable to or larger than the estimated errors. C, P(r) functions of C subunits in the type IIβ(1–280) holoenzyme complex (open squares). The P(r) curve of the C subunits in the type Iα holoenzyme (open circles (25)) is shown for comparison.

FIGURE 3.

Small angle x-ray scattering of RIIβ(1–280) homodimer. A, I(q) versus q plots of free RIIβ(1–280) homodimer in the absence (open triangles) and presence (filled triangles) of cAMP. The scattering data were binned, and the error bars reflect the S.D. of the points within the bin. The I(q) scale is arbitrary, and the plots have been offset to facilitate comparison. B, Guinier plots of free RIIβ(1–280) homodimer in the absence (open triangles) and presence (filled triangles) of cAMP. The I(q) scale is arbitrary and has been offset to facilitate comparison. The line shown with each dataset is a linear fit to the data in the range from q_min to q² = 0.0013. This value of q² corresponds to q × R_g = 1.3. C, P(r) functions for free RIIβ(1–280) homodimer in the absence (open triangles) and presence (filled triangles) of cAMP. The P(r) function of full-length RIIβ homodimer (from Ref. 41) is shown for comparison. The area under each curve has been scaled to reflect the square of molecular weight of the particle. Estimated errors are equal to or smaller than the symbols.

FIGURE 4.

Best-fit models of type IIβ(1–280) holoenzyme based on neutron scattering with solvent contrast. Models were generated as described under “Experimental Procedures” using data from the SANS solvent contrast series shown in Fig. 2. The models were generated using the D/D domain structure of RIIα (PDB: 2HWN) and the RIIβ(104–280)-C(14–350) heterodimers from the structure of type IIβ holoenzyme (PDB: 3TNP). The linker regions are modeled as a single cylinder to simulate the volume occupied by this region of the protein. The chains in the R subunit domains are colored with N termini in blue and C termini in red. The C subunits are colored gray, and the volume corresponding to the linkers is depicted as a red cylindrical cloud. A–C, the model with the best χ-square fit to the SANS data is shown (χ² = 0.815) panel A with two other independently determined models shown in panels B (χ² = 0.817) and C (χ² = 0.821). The predicted neutron scattering curves from the model in panel A are shown as the solid green lines in Fig. 2 as compared with the experimental neutron scattering. D, the structure is the crystal structure of the type IIβ tetramer formed with full-length RIIβ (PDB: 3TNP); it is arbitrarily oriented relative to the models. Molecular graphics for this figure were prepared with the UCSF Chimera package. Chimera is developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco (supported by National Institutes of Health Grant P41-GM103311 through the NIGMS) (42).

FIGURE 5.

Comparison of the 75 best-fit models. A, the 75 best-fit models were compared by overlaying all of the models based on the centers of mass of the C subunits (boxes), the CNBD-A of the R subunits (spheres toward the periphery), and the D/D domain of the R subunits (spheres near the center). The models were aligned by rotating them around the z axis such that the y-coordinates of the centers of mass of the R subunits are all 0.00. Three orthogonal views of the overlaid models are shown. The lowest χ² model (Fig. 4A) is shown in red, and the other models are shown in transparent shades ranging from blue to gray with the lowest χ² models being blue and the highest χ² models being gray. B, root mean squared deviation (RMSD) relative to the lowest χ-squared model as a function of χ² value. The points with the three lowest χ² values are labeled as A, B, and C and correspond to the models shown in Fig. 4, A–C, respectively. The solid line is a linear regression fit to the data and shows that there is only a weak correlation (r² = 0.1784) between χ² and root mean squared deviation to the lowest χ² model. Root mean squared deviation values were calculated by the program DAMCLUST (43) using the all atom, P2 symmetry, and handedness options.