Methods Optimization for the Expression and Purification of Human Calcium Calmodulin-Dependent Protein Kinase II Alpha

Calcium/calmodulin-dependent protein kinase II (CaMKII) is a complex multifunctional kinase that is highly expressed in central nervous tissues and plays a key regulatory role in the calcium signaling pathway. Despite over 30 years of recombinant expression and characterization studies, CaMKII continues to be investigated for its impact on signaling cooperativity and its ability to bind multiple substrates through its multimeric hub domain. Here we compare and optimize protocols for the generation of full-length wild-type human calcium/calmodulin-dependent protein kinase II alpha (CaMKIIα). Side-by-side comparison of expression and purification in both insect and bacterial systems shows that the insect expression method provides superior yields of the desired autoinhibited CaMKIIα holoenzymes. Utilizing baculovirus insect expression system tools, our results demonstrate a high yield method to produce homogenous, monodisperse CaMKII in its autoinhibited state suitable for biophysical analysis. Advantages and disadvantages of these two expression systems (baculovirus insect cell versus Escherichia coli expression) are discussed, as well as purification optimizations to maximize the enrichment of full-length CaMKII.


25
The CaMKII protein family is composed of four similar genes identified as α, β, γ, and δ (1). While the isoforms γ, 26 and δ are found ubiquitously in eukaryotic tissues, the α and β isoforms are found predominantly in the brain, making up 1-2% 27 of total protein in the hippocampus, and play a role in the regulation of synaptic plasticity in neurons (3). In the mammalian 28 brain CaMKII is found as a heterogeneous oligomer, or holoenzyme, of twelve or fourteen subunits assembled from a mix of 29 α and β isoforms (4-7). Each individual subunit includes a kinase domain incorporating a regulatory segment, an 30 oligomerization or hub domain, and a linker region joining the two domains whose length varies among isozymes. A number 31 of works have focused on the structural assembly of CaMKII and its mechanism of activation (1,6,7,11-15). It has been shown 32 that the close proximity of neighboring subunits within the holoenzyme enables rapid inter-subunit cross-phosphorylation 33 resulting in enzyme activation kinetics that are cooperative (13, 16). Transmission electron microscopy, crystallography, and 34 small-angle x-ray scattering have been used to reveal the morphology of subunit assembly and estimate the range of movement 35 of the tethered kinase domains (6,7,11,12,(17)(18)(19)(20). To reduce the complexity of understanding these multi-subunit mechanisms, 36 these works have relied upon the recombinant expression of CaMKII where all subunits are of a single isozyme in the 37 autoinhibited state. For high resolution structural studies, highly-enriched, autoinhibitied, monodisperse oligomeric CaMKII 3 38 enzyme is needed. Further, like many other kinases, CaMKII is labile and shows decreasing activity and aggregation over time 39 (21,22). Thus, the conditions of purification and handling time must be optimized to ensure the recovery of the intact and 40 functional CaMKII in high yield.

41
Many of the methods reported in the literature for generating purified recombinant CaMKII protein use bacterial 42 expression or baculovirus insect expression, and these systems have employed non-tagged and polyhistidine tag fusion 43 constructs. Table 1 lists the different purification strategies and yields (when given) that have been reported to date.

50
The Soderling lab first reported expression of recombinant wild type rat CaMKIIα using a baculovirus-insect 51 expression vector system (BEVS  a CaMKII phospho-antibody 22B1 showed that CaMKII was not phosphorylated at Thr286 (Fig 1D). analysis on BEVS CaMKII confirmed that the protein was not phosphorylated at Thr286 (Table S1), Thr305 or Thr306 (Table   203 S2), but did show phosphorylation at the NLS serine residues in 53% of the fragment sequences (Table S3).

204
Three additional parameters for the baculovirus/insect expression vector system were evaluated to determine the effect 205 on protein yield: seed density, virus concentration, and the use of an expression additive to supplement the media (Fig 2). Western blots detected with primary antibody anti-CaMKII 6G9 followed by IRDye 680RD (Fig 2A). Both flasks appeared to 210 produce a similar amount of full-length CaMKII, yet the expression seeded with 2 x 10 6 cells/mL showed increased truncation 211 or degradation products. showed that BPA enabled a greater amount of full-length CaMKII and a slightly lighter band of truncated products (Fig 2C).  purification that is not present in the phosphocellulose purification. This is attributed to some of the CaM-Sepharose resin 299 accidentally being washed away when decanting the wash buffer from the resin after centrifugation. While this is not a common 300 occurrence, we note that care must be taken when decanting after the wash and elution steps in batch reactions to maximize 301 protein recovery. A representative set of Mono S / CaM-Sepharose blots that do not have this artifact is shown in Fig S2. 302

303
In the previous sections we screened the growth and expression conditions in bacterial and insect systems to produce 304 maximal full-length CaMKII in either system. We also optimized a two-stage purification system to enrich full-length CaMKII 305 from cell lysate. With these steps optimized, we compared the yields of bacterial and BEVS CaMKII produced by this same 306 two-stage chromatography method using bacterial and insect pellet masses of approximately 5 g each (Fig 6). CaMKII as desired. Next, we characterized the assembly of recombinant CaMKII into oligomers using three different methods.

320
First, a native polyacrylamide gel (7.5%) loaded with two-step purified CaMKII and stained with gel code blue showed a 14 321 primary diffuse band just above the 720 kDa marker (expected mass is 648 kDa) (Fig 7A). A second faint band can be seen 322 slightly higher which may represent the phosphorylation of the NLS. The second characterization method was to isolate 323 CaMKII species using a Superose 6 size-exclusion column. Elution produced a predominant peak of dodecameric CaMKII at 324 approximately 12 mL that coincides with a thyroglobulin standard peak at 660 kDa ( Fig 7B) and also a minor flow through 325 peak of higher masses that are likely to be aggregates.

360
Phosphorylation and other post-translational modifications are a concern for BEVS, and for CaMKII, we found the NLS serine 361 residues are phosphorylated during expression. Still, we found that the critical kinase-activation residues Thr286 and 362 Thr305/Thr306 are not phosphorylated and do not require a de-phosphorylation step to yield autoinhibited (inactive) CaMKII 363 when expressed in the BEVS system.

364
The first step compares protein expression quantity and contamination produced in clarified lysate between the best 365 protocols of each expression system. Our method for measuring CaMKII is by Western blot using the monoclonal antibody 366 6G9, which recognizes the N-terminal kinase domain of CaMKII. Since both full-length and partially translated protein will 367 contain some or all of the kinase domain, the 6G9 antibody is useful to detect even truncated protein species. We also use a 368 second phospho-specific antibody to detect the subpopulation of CaMKII that is phosphorylated at Thr286.

369
In the bacterial expression system, a qualitative comparison between full-length and truncated protein detected over 370 the expression time course shows that they are similar in amount in the first 24 hours, but the detection of these truncations 371 declines substantially over longer time periods (Fig 1A), likely due to proteasome activity or potentially sequestration and 372 excretion from the cells. Full-length CaMKII also appears to decrease after 18 hours, but not as markedly as the truncations.

373
Thus, a longer expression time improves the relative purity of full-length protein over truncated protein. However, if a 374 purification scheme can remove the truncations, then it is more beneficial to stop expression where full-length protein is 375 maximized. Our conclusion is that a lower expression temperature, gentle induction, and expression for 18 h yields a maximal 376 amount of full-length CaMKII provided that the large number of truncations can be removed efficiently.

377
In the BEVS, fewer truncations than full-length CaMKII are detected by Western blot, and these truncations do not 378 appear to degrade over time. Thus, it is helpful to optimize the initial conditions to limit truncations and increase yield. These 379 initial conditions include two interdependent factors: seeding density and infection concentration. The current study shows that 380 a 1% v/v infection concentration and culture at 1 x 10 6 cells/mL density allows for early-stage infection and cell doubling while 381 still providing enough nutrients to produce protein in late-stage infection (Fig 2A). Infection at a higher seed density (2 x 10 6 382 cells/mL) appears to increase degradation products without increasing full-length protein, likely because of the stress of cell 383 density and increased nutrient consumption. Interestingly, the increased degradation products are also seen when a culture of 384 1 x 10 6 cells/mL density is infected with 0.1% v/v virus concentration (Fig 2B). The low infection concentration allows for 385 more than one round of cell doubling and repeated virus amplification until late-stage infection is reached towards the end of 386 cell viability in the media. However, an initial virus concentration of 5% v/v produces a lower yield, because cell replication is 387 halted, and protein expression begins at or near the initial seeding density. Our conclusion is that seeding BEVS Tni cells at a 388 density of 1 x 10 6 cells/mL infected with 1% v/v produced a maximal ratio of full-length of CaMKII to truncations when 389 harvested at 72 h. Additionally, the use of BPA 24 h post-infection appears to produce an enhanced amount of full-length 390 protein expression with no relative increase in truncations ( Fig 2C).

391
It is important to note that the doublet band found at 50 kD in the BEVS Western blots is likely due to our CaMKII 392 gene isoform B (Q9UQM7-2) that contains a short nuclear localization sequence (NLS). This addition inserts 12 residues 393 (KKRKSSSSVQLM) into the linker region that connects the kinase and hub domains. Proteomic analysis reveals that while 394 Thr286 and Thr305/THr306 are not phosphorylated during expression in BEVS, the four NLS serine residues (Ser332 -Ser335) 395 are phosphorylated in 52% of the sample which can be seen as a distinct second band in SDS-PAGE. The phosphorylated 396 protein will have slightly lower mobility due to the increase in negative charge (47).

398
Having optimized full-length CaMKII expression in both systems, the next goal is to choose a purification strategy 399 that maximizes enrichment while limiting the handling time to minimize protein degradation. Some publications report the use