Ca2+/CaM dependent protein kinase II (CaMKII)α and CaMKIIβ hub domains adopt distinct oligomeric states and stabilities

Abstract

Ca²⁺/calmodulin‐dependent protein kinase II (CaMKII) is a multidomain serine/threonine kinase that plays important roles in the brain, heart, muscle tissue, and eggs/sperm. The N‐terminal kinase and regulatory domain is connected by a flexible linker to the C‐terminal hub domain. The hub domain drives the oligomeric organization of CaMKII, assembling the kinase domains into high local concentration. Previous structural studies have shown multiple stoichiometries of the holoenzyme as well as the hub domain alone. Here, we report a comprehensive study of the hub domain stoichiometry and stability in solution. We solved two crystal structures of the CaMKIIβ hub domain that show 14‐mer (3.1 Å) and 16‐mer (3.4 Å) assemblies. Both crystal structures were determined from crystals grown in the same drop, which suggests that CaMKII oligomers with different stoichiometries likely coexist. To further interrogate hub stability, we employed mass photometry and temperature denaturation studies of CaMKIIβ and CaMKIIα hubs, which highlight major differences between these highly similar domains. We created a dimeric CaMKIIβ hub unit using rational mutagenesis, which is significantly less stable than the oligomer. Both hub domains populate an intermediate during unfolding. We found that multiple CaMKIIβ hub stoichiometries are present in solution and that larger oligomers are more stable. CaMKIIα had a narrower distribution of molecular weight and was distinctly more stable than CaMKIIβ.

1. INTRODUCTION

Ca²⁺/calmodulin‐dependent protein kinase II (CaMKII) is a serine/threonine kinase that is crucial to memory formation, cardiac pacemaking, and fertilization. The unique multimeric structure of CaMKII is driven by the C‐terminal hub domain. The N‐terminal kinase domain is tethered to the hub domain through a regulatory segment and flexible linker. The hub domains assemble as two stacked rings, which form the core of the CaMKII holoenzyme (Bhattacharyya et al., 2019; Chao et al., ²⁰¹¹). CaMKII is highly conserved across species, with the exception of the linker region, and is expressed by four paralogous genes in humans (α, β, δ, and γ) (Sloutsky & Stratton, 2020; Tombes et al., ²⁰⁰³). CaMKII is a metazoan protein, and homologs of full‐length CaMKII have been found in choanoflagellates (Bhattacharyya et al., 2016). Homologs of the CaMKII hub domain have also been found in bacteria and plants (McSpadden et al., 2019).

Structural studies have shown that each ring of the hub domain has 6 to 8 subunits, forming oligomers with 12 to 16 total subunits (Bhattacharyya et al., 2016; Hoelz et al., ²⁰⁰³; Kolodziej et al., ²⁰⁰⁰; McSpadden et al., ²⁰¹⁹; Morris & Torok, ²⁰⁰¹; Rellos et al., ²⁰¹⁰; Rosenberg et al., ²⁰⁰⁶). To date, the human and mouse CaMKII hub domains of α, δ, and γ have been crystallized as either 12‐ or 14‐mer complexes (Bhattacharyya et al., 2016; Hoelz et al., ²⁰⁰³; Rellos et al., ²⁰¹⁰). A CaMKII hub‐like protein from Chlamydomonas reinhardtii has been crystallized as an 18‐mer complex, which indicates the possibility of higher stoichiometries (McSpadden et al., 2019). Upon comparative analysis, it has been proposed that different stoichiometries are facilitated by the changes in beta sheet curvature and/or additional intrachain hydrogen bonds (McSpadden et al., 2019). A recent study revealed a putative model of the first 16‐mer complex of a CaMKIIβ holoenzyme using negative stain electron microscopy (EM) (Buonarati et al., 2021). The same study showed that both CaMKIIα and CaMKIIβ holoenzymes form mostly 12‐mers along with a lower population of 14‐mers.

Mass photometry (MP) measures the molecular weight of single particles in solution (Sonn‐Segev et al., 2020; Young et al., ²⁰¹⁸). This technique has been used to show that CaMKIIα holoenzymes in solution cluster around a molecular weight of a 10 to 12‐mer (Lucic et al., 2023; Torres‐Ocampo et al., ²⁰²⁰), whereas the CaMKIIα hub domains form larger species, around 16‐mers (Torres‐Ocampo et al., 2020). The CaMKIIα hub domain is significantly more stable than the CaMKIIα holoenzyme in terms of temperature stability and oligomerization (Leurs et al., 2021; Torres‐Ocampo et al., ²⁰²⁰). Another study measured the melting temperature of the monomeric Caenorhabditis elegans CaMKII variant to be roughly 63°C (Hoffman et al., 2011). Overall, the stability of CaMKII variants has been largely understudied.

Here, we report two human CaMKIIβ hub domain crystal structures at <3.5 Å with 14‐ and 16‐mer stoichiometries. We use MP and circular dichroism (CD) to interrogate the stoichiometry and stability in solution of the CaMKIIβ hub domain, and we compared this to the CaMKIIα hub. It has been previously shown that the hub domain impacts CaMKII activity (Roman Sloutsky et al., 2020), so it is important to consider distinctions between variants in understanding how stoichiometry and hub dynamics impact CaMKII regulation.

2. RESULTS

2.1. Two CaMKIIβ hub stoichiometries from a single crystallization drop

Initial CaMKIIβ hub crystals were canoe shaped (Figure 1a). A 14‐mer hub structure was determined at 3.1 Å resolution using a data set collected from a crystal on the right side of the drop (Figure 1a,b, PDB: 7URW). The space group (C222₁) and unit cell dimensions (104.24, 183.19, 108.37) of this structure are comparable to a recently published CaMKIIα hub 14‐mer (PDB: 7REC) (Buonarati et al., 2021). Crystal packing and oligomer dimensions (112 Å × 112 Å x 60 Å, Figure 1b,c; Table 1) of both structures are nearly identical.

FIGURE 1.

Two CaMKIIβ hub domain stoichiometries from one crystallization drop. (a) Protein crystals shown at the two‐day timepoint. The arrow indicates the crystal used for data collection of crystal structure presented in (b). A 14‐mer hub domain crystal structure (PDB code: 7URW) shown (b) top‐view and (c) side‐view. (d) Protein crystals with different morphology appeared in the same drop at the four‐day timepoint. The arrow indicates the crystal used for data collection of crystal structure presented in (e). A 16‐mer CaMKIIβ hub domain crystal structure (PDB code: 7URZ) shown (e) top‐view, (f) side‐view. (g) Overlay of tetramers from 14‐mer (green, PDB: 7URW) and 16‐mer (magenta, PDB: 7URZ) structures. On the left, a tetramer of the hub domains is shown. Structures are aligned along the core α‐helix and the first β‐sheet of the domain in the box. The inset shows an orientation of the subunit where the curvature is highlighted, as evidenced by the offset of residues highlighted as sticks for clarity. Increased beta sheet curvature facilitates larger ring formation. CaMKII, Ca²⁺/calmodulin‐dependent protein kinase II.

TABLE 1.

The 14‐ and 16‐mer CaMKIIβ hub domain crystal structures.

	14‐mer CaMKIIβ hub domain (PDB:7URW)	16‐mer CaMKIIβ hub domain (PDB:7URZ)
Data collection
Space group	C2221	P4 2212
Cell dimensions
a, b, c (Å)	104.24, 183.19, 108.37	81.34, 81.34, 180.04
α, β, γ (o)	90, 90, 90	90, 90, 90
Resolution (Å)	50–3.1 (3.15–3.1)	60–3.45 (3.53–3.47)
R merge	0.291 (1.454)	0.368 (1.361)
Mean I/σI	6.2 (1.67)	4.5 (2.02)
Completeness (%)	98.5 (95.4)	99.2 (99.0)
Redundancy	6.2 (5.5)	7.3 (7.5)
CC1/2	0.962 (0.567)	0.992 (0.650)
CC*	0.990 (0.851)	0.998 (0.888)
Refinement
Resolution (Å)	37.8–3.11 (3.186–3.11)	37.09–3.45 (3.537–3.45)
Unique reflections	17,703 (1200)	7921 (578)
R work/R free (%)	25.7/30.02	35.44/40.74
No. of atoms
Protein	6789	3910
Water	‐	‐
Ligand	‐	‐
Ramachadran plot
In preferred regions (%)	97.54	95.16
In allowed regions (%)	1.75	4.44
Outliers (%)	0.70	0.40
B factors
Protein	74.74	108.92
Water	‐	‐
Ligand	‐	‐
R.M.S. deviations
Bond lengths (Å)	0.0046	0.0022
Bond angles (°)	1.3269	1.1924

Abbreviation: CaMKII, Ca²⁺/calmodulin‐dependent protein kinase II. R.M.S., roote mean squared, CC., Pearson correlation coefficient

Two days later, crystals with a different morphology (tetragonal bipyramid with 4‐fold symmetry), as well as many microcrystals, appeared in the same drop (Figure 1d). From these crystals, we determined a 16‐mer structure at 3.4 Å resolution (PDB: 7URZ). There are four hub domains in the asymmetric unit cell with a P4 symmetry. The space group (P4₂2₁2) and unit cell dimensions (81.34, 81.34, 180.04) are distinct from the tetradecameric structure. Comparing this 16‐mer hub structure to the hub of the 16‐mer holoenzyme EM structure (Rosenberg et al., 2006), the outer diameter measurements are comparable (128 Å × 125 Å × 62 Å, Figure 1e,f; Table 1).

In the structures presented here, the curvature of the central β‐sheets changes with stoichiometry, as described previously (Figure 1g) (McSpadden et al., 2019). The difference in angle between β‐sheets in CaMKIIα 12‐mer compared with a 14‐mer is ~8.6° (Bhattacharyya et al., 2016). The difference in angle is closer to 3° when we compared the 14‐ and 16‐mer CaMKIIβ hubs. We would expect that as ring size increases, the angle between subunits will not change as drastically. The 16‐mer is subtly larger in all dimensions compared to 14‐mer (Figure 1c,f). Comparison of the inside cavities shows that the 16‐mer structure has an elliptical cavity (36 Å × 40 Å), whereas the 14‐mer structure cavity is circular (25 Å × 25 Å, Figure 1b,e).

The C terminus of each hub domain extends from the final β‐sheet (β6), which faces the central pore of the donut (Figure S1a). The construct used for the CaMKIIβ hub structures presented here includes all C‐terminal residues, but the last 3 to 5 residues are only well resolved in 1 out of 11 unique chains between the two structures. We compared an additional 14‐mer CaMKIIβ hub structure deposited at higher resolution (2.6 Å, PDB code: 7URY) (Saha et al., 2023), where the last three to five residues are resolved in five out of seven unique chains. We observe that these terminal residues fold back toward the central pocket of their own hub domain (Figure S1a). Two proline residues facilitate turns at the C terminus (Figure S1b,c).

2.2. CaMKIIβ hub stability

Since multiple stoichiometries can be formed in crystals, we sought to interrogate stoichiometry and its effects on stability in solution. We performed differential scanning calorimetry on the CaMKIIβ hub, which showed an apparent melting temperature (T _m) of 102.2°C (Figure S2a). We then used CD to measure protein unfolding, and, as expected from the DSC data, a standard temperature melt to 100°C did not completely unfold the protein (Figure S2b). We performed the temperature melt in the presence of 2 M guanidine hydrochloride (GdnHCl) and observed two distinct, cooperative unfolding transitions (Figure 2a). This process was irreversible, so all values reported here are phenomenological (Figure S3a,b). The first transition (N ⇌ I) had a T _m of 56.5°C and a slope of 0.18 ± 0.04 cal mol⁻¹°C⁻¹. The second transition (I ⇌ U) had a T _m of 81.8°C and a slope of 0.35 ± 0.06 cal mol⁻¹°C⁻¹ (Table 2). We also performed guanidine melts on the CaMKIIβ hub (Figure 2b). The oligomer is fully unfolded at 4.5 M GdnHCl and evidently transitions through an intermediate, but these data were not easily fitted to a 2‐ or 3‐state equation.

FIGURE 2.

CaMKIIβ hub unfolding. (a) Temperature melt of CaMKIIβ hub oligomer in the presence of 2 M guanidine hydrochloride (GdnHCl). Data points represent average circular dichroism (CD) signal at 218 nm (n = 9). Data are fit to a 3‐state non‐linear regression. Vertical dashed lines indicate temperatures at which mass photometry (MP) measurements were performed (c). (b) Guanidine melt of CaMKIIβ hub oligomer. Blue data points represent average CD signal at 218 nm (n = 9). (c) Representative MP histogram for CaMKIIβ hub samples (n = 8) that have been incubated in 2 M GdnHCl and heated to the indicated temperature. Each count indicates a single molecule. (d) Guanidine melt of CaMKIIβ hub dimer mutant (F585A + L622M). Data points represent average CD signal at 218 nm (n = 9). Data are fit to a 2‐state non‐linear regression. CaMKII, Ca²⁺/calmodulin‐dependent protein kinase II.

TABLE 2.

Thermodynamic parameters for N ⇌ I and I ⇌ U transitions.

	Slope (kcal mol−1 K−1)	∆G app (kcal mol−1)	T m (°C)
N ⇌ I	0.18 ± 0.04	10 ± 2	56.5 ± 1.9
I ⇌ U	0.35 ± 0.06	28 ± 5	81.8 ± 1.4

Note: Mean and standard deviation for the calculated slope, ∆G _app, and T _m for the two transitions observed in the temperature melt; n = 9.

2.3. Defining the unfolding mechanism

We utilized mass photometry (MP) to determine the oligomeric state of the CaMKIIβ hub during the unfolding process. We performed MP measurements of samples collected along the unfolding trajectory as shown in Figure 2a, sampling temperatures representing each state along the unfolding pathway in the presence of 2 M GdnHCl (25°C, 40°C, 70°C, and 97°C) (Figure 2c; Figure S4a–c). Since MP is performed at significantly lower concentrations, we verified that a dilution series maintains the same populations with and without GdnHCl (Figure S5a,b). At all temperatures, there is a small population around 75 kDa. Control measurements indicated that this is partially contributed by guanidine and likely partially contributed by lower‐order species or unfolded monomers (Figure S4d). Before the first transition (25°C and 40°C), there are two distinct CaMKIIβ hub populations that represent ~12‐ and ~14‐mers. The lower mass peaks are centered around 191 ± 16.5 kDa (25°C) and 193 ± 15.6 kDa (40°C), which correspond to 12‐mers (180 kDa) along with a small population of 14‐mers (210 kDa). The higher mass peaks are centered around 218 ± 19.2 kDa (25°C) and 230 ± 17.7 kDa (40°C), which correspond to a mixture of 14‐mers (210 kDa) and 16‐mers (240 kDa). Notably, there is a population shift from 218 kDa to 230 kDa between 25°C and 40°C, indicating that the larger species are selected for at higher temperatures. After the first unfolding transition (70°C), the lower mass oligomeric peak disappears and a single peak with a mass of 226 ± 16.0 kDa is observed, representing a mixture of 14‐ and 16‐mers. After the second unfolding transition (97°C), no oligomeric peaks remain. We cannot measure whether the proteins are all monomeric at 97°C (the resolution of the Refeyn One MP cannot detect individual hub monomers of 15 kDa), but it is likely that all hub domains have fully unfolded because the CD shows all secondary structure was melted at this temperature (Figure 2a; Figure S3b).

Since MP can distinguish larger species but not smaller ones, we wanted to use a different technique to assess the stability of a smaller hub unit. There is mounting evidence that upon dissociation, CaMKII releases vertical dimer units (Sarkar et al., 2017; Stratton et al., ²⁰¹⁴). We designed a double mutant (F585A + L622M) that disrupts the lateral interfaces in the CaMKIIβ hub domain to form predominantly vertical dimers. A single mutation has not sufficed to eliminate all oligomerization in solution (Bhattacharyya et al., 2016; Sarkar et al., ²⁰¹⁷). Size‐exclusion chromatography paired with multiangle light scattering confirmed that the double‐mutant we created (herein referred to as dimer) migrated as a single peak and had a calculated mass of around 36 kDa, whereas the wild‐type (WT) hub domain had two mass peaks: 207 and 176 kDa (Figure S6a). This confirmed our double‐mutant forms only dimers in solution. Additionally, the CD signatures of the dimer and the WT hub overlap well, indicating that these mutations do not affect monomer folding (Figure S6b). We performed a guanidine melt on the dimer, which gave a C _m of 0.63 M GdnHCl (Figure 2d). We directly compared the unfolding of the CaMKIIβ hub oligomer to that of the hub dimer and observed that the oligomer starts to unfold around 90°C, whereas the dimer is completely unfolded by 60°C (T _m = 51.7) (Figure S2b,c).

2.4. CaMKIIα hub displays differences in stability compared with CaMKIIβ hub

CaMKIIα and CaMKIIβ hub domains share 77% sequence identity—but do they act similarly in solution? In MP measurements, the CaMKIIα hub formed a single population centered at a 16‐mer size (240 kDa), as opposed to the two populations observed for the CaMKIIβ hub (Figures S7 and S8). We performed temperature denaturation melts on the CaMKIIα hub in the presence of 2 M GdnHCl. This showed two unfolding transitions, with T _m values of 71.8°C and 85.2°C (Figure 3a), both higher than those of the CaMKIIβ hub unfolding (see Figure 2a). We performed MP on the CaMKIIα hub samples at temperatures before and after each transition (Figure 3b). The single population of the CaMKIIα hub decreased in number across each transition (from 60°C to 90°C), indicating oligomer dissociation and unfolding. However, there is no clear change in stoichiometry of the oligomer population across transitions nor an emergence of new populations that would directly account for the two separate transitions.

FIGURE 3.

CaMKIIα hub unfolding. (a) Temperature melt of CaMKIIα hub oligomer with 2 M guanidine hydrochloride (GdnHCl). Blue data points represent average circular dichroism (CD) signal at 218 nm (n = 6). Data are fit to a 3‐state non‐linear regression. Vertical dashed lines indicate temperatures at which mass photometry (MP) measurements were performed (b). (b) Representative MP histogram for CaMKIIα hub samples (n = 6) that have been incubated in 2 M GdnHCl and heated to the indicated temperature. Each count indicates a single molecule. Note that the 25°C population (blue) is overlapping with the 40°C population (yellow). CaMKII, Ca²⁺/calmodulin‐dependent protein kinase II.

3. DISCUSSION

The CaMKII holoenzyme is assembled by the hub domain, which oligomerizes to form a donut‐like structure. The N‐terminus of the hub domain forms a helix which connects to the catalytic domain in the full‐length holoenzyme on the outside of the donut (Figure S1a). If the C‐terminus was extended, it is unclear where these residues would travel to—this is an important factor to consider with fusion proteins and tags added at the C‐terminal end of CaMKII. Of note, CaMKIIδ has an extended C‐terminus, which will likely have distinct effects on hub dynamics and stability. Each hub domain harbors a central pocket, which is lined with arginine residues. A previous study determined a structure of the CaMKIIα hub domain bound to an analog of γ‐hydroxybutyrate (5‐HDC) in this arginine pocket and showed that this compound bound specifically to CaMKIIα (Leurs et al., 2021). Other hub structures also trap molecules in this pocket, for example the other 14‐mer CaMKIIβ hub structure (PDB code: 7URY) shows malic acid bound in the pocket (Saha et al., 2023). The specificity for 5‐HDC is striking, and rather remarkable, given the high sequence conservation between CaMKII variants. A major difference between the hub domains is that in CaMKIIα, there are three arginines in the pocket, whereas in CaMKIIβ, there are only two, and the third is substituted with cysteine (Figure S1b,c). Another difference is the sequence of the five C‐terminal residues (Figure S1c), where both segments would be flexible but the sequences are different. As noted, our constructs have full‐length C‐termini compared with other CaMKIIα structures which are terminated early; this may also affect stoichiometry and it will be important to interrogate this further.

We were intrigued to capture two distinct oligomeric states grown within the same crystallization drop, suggesting that these species coexist and/or interconvert. Although the biological implications of stoichiometry are not clear, it is possible that these allowable conformations lend CaMKII its unique properties such as subunit exchange (Bhattacharyya et al., 2016; Stratton et al., ²⁰¹⁴). A recent study has shown that CaMKIIβ has different activation properties compared with CaMKIIα (Roman Sloutsky et al., 2020). CaMKIIα requires higher concentrations of Ca²⁺/calmodulin for activation, whereas CaMKIIβ was active at much lower concentrations. This difference may be attributed to different stoichiometric states.

Both CaMKIIα and β hub domains are extremely stable, requiring temperatures of over 90°C for unfolding (Torres‐Ocampo et al., 2020). In fact, in order to observe unfolding in an achievable temperature range, we needed to add denaturant (2 M GdnHCl). Surprisingly, both hub domains unfolded in a 3‐state reaction, but these reactions are distinct from one another despite the high sequence similarity between these domains (Figure S8a). CaMKIIβ hub undergoes its first unfolding transition at a lower temperature compared with CaMKIIα. From MP measurements, we interpret this first transition as the 12‐mer oligomers unfolding, while the 14/16‐mer oligomers remain (Figure S8a–e). The second transition is the remaining 14/16‐mer oligomers dissociating and unfolding. We interpret this to mean that the 14/16‐mer population is more stable; the 14/16‐mer has more buried interfaces, suggesting that this stability can be attributed to favorable enthalpic contributions provided by the higher oligomeric state. The CaMKIIα unfolding mechanism appears to be more nuanced, as the oligomeric state (16‐mer) does not obviously change during unfolding. A possible explanation for the first unfolding transition (71°C) is that the oligomeric ring cracks at one vertical interface, creating a spiraled ring which has been previously reported (Bhattacharyya et al., 2016). The second transition would then be all spiraled oligomers dissociating and unfolding. It is also notable that the CaMKIIα transitions (closed to open ring/dissociation and unfolding) are closer energetically (Δ13°C) compared with the transitions for CaMKIIβ (Δ26°C). The second transition for CaMKIIα and CaMKIIβ hubs is overlapping (Figure S8a), it is possible that full dissociation and unfolding are the same energetically for both variants.

The molar ellipticity of the hub dimer we created is the same as that of the oligomer, so any change in CD signal observed during the denaturation is likely attributed to monomer unfolding rather than disassembly into dimers. Both hub domains displayed a large native baseline slope over the first 40–60 degrees of the experiment. We interpret this slope as oligomers (regardless of stoichiometry) disassembling into dimers and unfolding immediately (Figure 4a), which is reflected in the MP measurements where the oligomeric peak(s) counts decrease steadily over this temperature range (Figures 2c and 3b). We cannot rule out precipitation occurring during the measurement; however, we do not observe any large species in MP. We were unable to fit the GdnHCl melt of the CaMKIIβ hub domain to a three‐state curve, as the intermediate was not well sampled under these conditions. However, the native baseline only begins to slope after ~0.5 M GdnHCl. This corresponds with the C_m of the CaMKIIβ hub dimer (~0.6 M), indicating that dimers are unfolding at this point, likely because of oligomer dissociation. There are two visible transitions, one around 3 M GdnHCl and a second around 4.5 M GdnHCl. These transitions likely represent the same processes we visualized in the temperature + GdnHCl denaturation melts.

FIGURE 4.

Schema of proposed mechanism of unfolding for CaMKIIβ hub oligomer during temperature melts in the presence of 2 M guanidine hydrochloride (GdnHCl). (a) At low heat, some oligomers disassemble into dimers regardless of stoichiometry. The dimers immediately unfold. (b) At high heat, 12‐mers first disassembly and unfold in a cooperative transition. The 14 ‐and 16‐mers disassemble and unfold in a second cooperative transition. CAMKII, Ca²⁺/calmodulin‐dependent protein kinase II.

It is clear from both crystallography and EM studies that CaMKII holoenzymes and hub domains can adopt different stoichiometries. This adds to the complexity of this enzyme, which can form heterooligomeric complexes, and also adopt mixed oligomeric states of heterooligomers. The physiological relevance of these differences remains unclear, but will be important to interrogate moving forward. It is possible that the allowance of multiple stoichiometries facilitates subunit exchange between CaMKII holoenzymes (Bhattacharyya et al., 2016; Stratton et al., ²⁰¹⁴). Our study has provided clear evidence that, despite nearly identical sequences, CaMKIIα and β hub domains adopt different stoichiometries in solution and have distinct stabilities (Figure S8). This provides a framework for assessing the differences observed in activation and regulation between CaMKIIα and β holoenzymes and how they may be related to dynamics and stability. Before addressing such questions, it would be necessary to determine whether full‐length CaMKII holoenzymes exhibit the same stoichiometries and disassembly properties as those observed here in hub domain holoenzymes.

4. MATERIALS AND METHODS

4.1. Protein purification

The CaMKIIβ full‐length hub domain (residues 534–666) was expressed in Protein deficient B strain (BL21 DE3) cells with a N‐terminal hexahistidine tag. The cells were induced at 18°C by adding 1 mM Isopropyl β‐D‐1‐thiogalactopyranoside (IPTG) and grown overnight. The cells were resuspended in Buffer A (25 mM Tris–HCl pH 8.5, 150 mM KCl, 40 mM imidazole, and 10% glycerol), supplemented with commercially available protease inhibitors (0.5 mM benzamidine, 0.2 mM 4‐(2‐aminoethyl)benzenesulfonyl fluoride hydrochloride, 0.1 mg/mL trypsin inhibitor, 0.005 mM leupeptin, 1 μg/mL pepstatin), 1 μg/mL DNase/50 mM MgCl₂ were added, and the cells were then lysed. All subsequent steps were performed at 4°C using an ÄKTA pure chromatography system (GE). Cell lysate was loaded onto 5 mL His Trap FF Ni Sepharose column (GE) and eluted with 50% Buffer B (25 mM Tris–HCl pH 8.5, 150 mM KCl, 1 M imidazole, 10% glycerol). The protein was desalted from excess imidazole using a HiPrep 26/10 Desalting column using Buffer C (25 mM Tris–HCl pH 8.8, 150 mM KCl, 1 mM EDTA, 2 mM DTT, and 10% glycerol). The hexahistidine tag was cleaved with Prescission protease overnight at 4°C. Proteins were eluted from HiTrap Q‐FF with a step‐gradient from 0% to 17% using Buffer BQ (25 mM Tris–HCl pH 8.5, 1 M KCl, 10% glycerol). Eluted proteins were concentrated and further purified using a Superose 6 size‐exclusion column equilibrated with gel filtration buffer (25 mM Tris–HCl pH 8.0, 150 mM KCl, 1 mM Tris(2‐carboxyethyl)phosphine (TCEP), 10% glycerol). Fractions (>95% purity) were pooled, concentrated, flash frozen in liquid nitrogen, and stored at −80°C until needed.

The CaMKIIβ hub dimer variant was designed by modifying the wild type CaMKIIβ hub domain via two mutations: F585A and L622M (residue numbering based on full‐length, non‐spliced CaMKIIβ). The variant was expressed in BL21 DE3 cells with an N‐terminal hexahistidine tag and purified using the same process as the wild type CaMKIIβ hub. During the size‐exclusion step, a Superdex 75 size exclusion column was used.

4.2. Crystallization, data collection, and structure determination

Protein crystals of 14‐ and 16‐mer CaMKIIβ hub domain were grown in the same drop using hanging drop vapor diffusion at 20°C. The drops contained 1 μL of the well solution and 1 μL of 1.9 mM hub domain. The well solution was 0.1 M imidazole pH 7, 0.15 DL‐Malic acid pH 7 and 22% Polyethylene glycol (PEG) methyl ether 550. Crystals were frozen directly from the well solution. Diffraction data were collected at a wavelength of 1.5418 Å using a Rigaku MicroMax‐007 HF X‐ray source, which was coupled to a Rigaku VariMax HF optic system (UMass Amherst). The X‐ray data were collected at 100 K. The datasets were integrated, merged scaled using HKL‐2000. The structure was solved by molecular replacement with Phaser (McCoy et al., 2007) using single CaMKIIγ hub domain (PDB: 2UX0) (Rellos et al., 2010) as the search model. Model building was performed using Coot and the refinement was carried out with Refmac5 (Murshudov et al., 2011).

4.3. Guanidine melts

Two solutions of 10 μM CaMKIIβ hub were prepared, containing either 0 M GdnHCl or 6 M GdnHCl. All protein dilutions were performed using a modified gel filtration buffer suitable for CD (5 mM Tris–HCl pH 8.0, 150 mM KCl, 1 mM TCEP). A stock 6.5 M GdnHCl solution was used (6.5 M GdnHCl, 5 mM Tris–HCl pH 8.0, 150 mM KCl, 1 mM TCEP) to prepare GdnHCl solutions. The two solutions were titrated together to give an array of 10 μM CaMKIIβ hub solutions with a range of GdnHCl concentration. For the CaMKIIβ hub oligomer, these solutions ranged from 0 M to 6 M GdnHCl at steps of 0.33 M. For the CaMKIIβ hub dimer, these solutions range from 0 M to 1.15 M in 0.05 M steps.

Following the titration, protein samples were incubated for 20 min (CaMKIIβ hub oligomer) or 50 min (CaMKIIβ hub dimer). After equilibration, the extent of unfolding in each solution was measured on a Jasco J1500 CD spectrophotometer using 1 mm cuvettes. This was measured using CD signal. For each sample, the average CD at 218 nm was recorded over 30 s at 25°C. Additionally, full spectrum scans (195–260 nm) were recorded for CaMKIIβ hub oligomer and dimer samples in 0 M GdnHCl. The raw CD signal (mdegs) was converted to molar ellipticity ($\Theta$) for the full spectrum scans via the following conversion: $\Theta$ = mdegs × M/(10 × L × C), where M is average molecular weight in g/mol, L is path length of the cell, and C is concentration in grams per liter.

Using a non‐linear regression, data were fitted to a two‐state curve (Equation (1)) in Graph Pad Prism. The parameter values were used to calculate C _m values (C _m = m4/m5).

$$\Theta =\frac{{\Theta }_{\mathrm{N}}+{\Theta }_{\mathrm{I}}\times {\mathrm{e}}^{\frac{-\left({\mathit{\Delta G}}_{{\mathrm{H}}_2\mathrm{O},\mathrm{NI}}^o-m_{\mathrm{NI}}\times \text{[GdnHCl]}\right)}{\mathit{RT}}}+{\Theta }_{\mathrm{U}}\times {\mathrm{e}}^{\frac{-\left({\mathit{\Delta G}}_{{\mathrm{H}}_2\mathrm{O},\mathrm{NI}}^o-m_{\mathrm{NI}}\times \text{[GdnHCl]}\right)}{\mathit{RT}}}\times {\mathrm{e}}^{\frac{-\left({\mathit{\Delta G}}_{{\mathrm{H}}_2\mathrm{O},\mathrm{IU}}^o-m_{\mathrm{IU}}\times \text{[GdnHCl]}\right)}{\mathit{RT}}}.}{1+{\mathrm{e}}^{\frac{-\left({\mathit{\Delta G}}_{{\mathrm{H}}_2\mathrm{O},\mathrm{NI}}^o-m_{\mathrm{NI}}\times \text{[GdnHCl]}\right)}{\mathit{RT}}}+{\mathrm{e}}^{\frac{-\left({\mathit{\Delta G}}_{{\mathrm{H}}_2\mathrm{O},\mathrm{NI}}^o-m_{\mathrm{NI}}\times \text{[GdnHCl]}\right)}{\mathit{RT}}}\times {\mathrm{e}}^{\frac{-\left({\mathit{\Delta G}}_{{\mathrm{H}}_2\mathrm{O},\mathrm{IU}}^o-m_{\mathrm{IU}}\times \text{[GdnHCl]}\right)}{\mathit{RT}}}}$$ (1)

4.4. Temperature melts

Protein samples of 60 μM CaMKIIβ hub and 2 M GdnHCl were prepared at a final volume of 215 μL. CD measurements were performed using a Jasco J1500 CD spectrophotometer. Protein samples in a 1 mm cuvette were heated from 20°C to 100°C at a rate of 0.67°C per minute. Every 1°C°C, the CD signal at 218 nm was recorded over 30 s. Full spectrum CD scans (205 to 260 nm) were taken every 10°C. Eight temperature melt replicates were performed. Using a nonlinear regression, data were fitted using Equation (2) in GraphPad Prism, where N is the native protein, I is the intermediate, U is the unfolded protein, NI is the first transition, and IU is the second transition. The average m4, m5, m6, and m7 values were calculated over all replicates. These values were used to calculate the melting temperature for each transition (T _m,1 = m4/m5 and T _m,2 = m6/m7) and unfolding energies (m4 × RT and m6 × RT, respectively) (T = 297 K).

(2)

4.5. Mass photometry

All mass photometry measurements were performed on a Refeyn One MP mass photometer. All protein solutions were prepared in a filtered buffer (25 mM Tris–HCl pH 8.0, 150 mM KCl). Before each experiment, the buffer alone was measured as a control. The buffer was considered clean if the hit count was below 75. For each experiment, a standard mass calibration curve was calculated using Thyroglobulin (Tg) at 25 nM (final concentration on the coverslip), apoferritin (440 kDa) at 3.75 nM, and Bovine serum albumin (BSA) (66.5 kDa) at 15 nM; 5 μL of standard protein was added to 15 μL of MP buffer on a coverslip, and the MP was measured for 60 s.

For each run, a 200 μL solution of 2 μM protein and 2 M GdnHCl was prepared. The GdnHCl solution was filtered using a 0.22 μm filter before use. The sample was heated to the chosen temperatures using a thermocycler. The sample was held at each temperature for 90 s, at the end of which 6 μL were taken and immediately diluted onto the coverslip with 14 μL of MP buffer to perform an MP measurement over 60 s. Mass histograms of each run were generated using Refeyn's DiscoverMP software (n = 6 for CaMKIIɑ hub, n = 8 for CaMKIIβ hub).

4.6. Size‐exclusion chromatography multiangle light scattering

Size‐exclusion chromatography (SEC) coupled with multiangle light scattering (MALS) was performed using an Agilent 1260 HPLC system, a DAWN HELEOS II MALS detector, and an Optilab‐T‐rEX refractive index detector. The SEC column used was a Tosoh TSKGel G3000 SWxl (particle size 5 μm, 30 cm length, calibration range 10–500 kDa) and was equilibrated with ethanol, distilled water, and finally gel filtration buffer (25 mM Tris–HCl pH 8.0, 150 mM KCl, 1 mM TCEP, 10% glycerol). The SEC‐MALS was conducted at a constant flow rate of 0.9 μL/min with an injection volume of 50 μL to ensure accurate separation. Protein samples of CaMKIIβ hub oligomer and dimer were injected at a concentration of 262 mg/mL. The column temperature was maintained at 25°C throughout the analysis to prevent temperature‐induced variations. Data were acquired and processed using ASTRA. The scattering data were analyzed based on the Zimm plot method to determine the molecular weight of the samples. The SEC‐MALS system was calibrated using BSA (2 mg/mL) to ensure accurate determination of molecular weights.

AUTHOR CONTRIBUTIONS

Margaret M. Stratton: Conceptualization; investigation; funding acquisition; writing – original draft; methodology; writing – review and editing; formal analysis; project administration; supervision; visualization. Can Özden: Conceptualization; investigation; writing – original draft; methodology; validation; visualization; writing – review and editing; formal analysis; data curation. Sara MacManus: Conceptualization; investigation; writing – original draft; methodology; validation; visualization; writing – review and editing; formal analysis; data curation. Ruth Adafia: Investigation; methodology; validation; visualization; data curation. Alfred Samkutty: Visualization; data curation. Ana Torres Ocampo: Visualization; data curation. Scott C. Garman: Formal analysis; visualization; supervision; methodology.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Data S1. Supporting Information.

Özden C, MacManus S, Adafia R, Samkutty A, Torres‐Ocampo AP, Garman SC, et al. Ca2+/CaM dependent protein kinase II (CaMKII)α and CaMKIIβ hub domains adopt distinct oligomeric states and stabilities. Protein Science. 2024;33(4):e4960. 10.1002/pro.4960