PKA inhibitor (PKI) fragment (6–22) amide was purchased from Tocris (Cat. No. 1904). Full length mouse PKIα was synthesized from L amino acids and then purified with high-performance liquid chromatography (HPLC) by the Koch Institute Biopolymers and Proteomics Facility. Mouse PKIα shares 74 out of the 76 amino acids with human PKIα, and the two differing amino acids are not in the 6–22 region.

Kinase assays were conducted in cell-free, in vitro assays with purified kinases with the Thermo Fisher Scientific SelectScreen Kinase Profiling Service. CamK1, DAPK1, and NUAK1 (ARK5) were screened with the Adapta Universal Kinase Assay, and the rest of the kinases were screened with the Z′-LYTE Peptide Kinase Assay (Rodems et al., 2002) (Figure 1B). All kinases used were human proteins expressed in and purified from Sf9 insect cell culture. Only one lot of kinase was used for each assay. Kinase activity was determined and reported on the lot-specific Certificate of Analysis at Thermo Fisher web site. The kinase concentration used in the assay was adjusted to achieve ∼25% substrate phosphorylation, with 10–50% as the allowable range in the 0% inhibition control. The % inhibition value was calculated with the formula specified in the assay manuals.

All the raw data can be found in Supplementary Table S1. Dose-response curves were fit using variable-slope nonlinear regression (4 parameters) in GraphPad Prism (GraphPad Software). Kruskal-Wallis followed by Dunn’s Multiple Comparison Test was used to evaluate statistical significance in Supplementary Figure S1.

To investigate whether PKI has effects on kinases beyond PKA, we screened the effects of PKI on a panel of kinases. There are about 510 kinases encoded in the mouse genome, and screening all of them would be too costly. Therefore, we limited our selection of kinases to screen via a tiered system (Figure 1A). First, among the 510 kinases, 322 can be screened for activity enhancement or inhibition via either Adapta Universal Kinase Assay or the Z′-LYTE Peptide Kinase Assay. These are in vitro, cell-free assays using purified kinases and therefore suitable for measuring direct kinase action. Second, since PKA is a serine/threonine kinase, we deduced that PKI would most likely target other serine/threonine kinases. Thus, we limited our screen to the 191 serine/threonine kinases among the 322 kinases that can be screened. Third, among the 191 serine/threonine kinases, we reasoned that PKI would mostly act on two kinase groups: the AGC kinases of which PKA is a member, and the CamK kinases since these share substrate similarities with PKA. In total, among the 191 serine/threonine kinases, 81 kinases belong to the AGC and CamK groups. Finally, since PKI plays important roles in neuronal function by regulating synaptic plasticity, and small peptide of PKI has been extensively used as pharmaceutical agent to inhibit PKA activity in the cortex, striatum, and hippocampus, we further narrowed down to 55 kinases that are expressed in these three brain regions based on RNA sequencing data (Cembrowski et al., 2016; Saunders et al., 2018; Zeisel et al., 2018). Therefore, we screened the effects of PKI on 55 kinases, which were selected based on the availability of kinase screen, kinase groups, and expression pattern of these kinases.

Fifty-two kinases were screened with the Z′-LYTE Peptide Kinase Assay (Figure 1B) (Rodems et al., 2002) and three kinases were screened with the Adapta Universal Kinase Assay. In the Z′-LYTE assay, the kinase substrate is flanked by a donor and acceptor fluorophore. Phosphorylation of the substrate prevents subsequent cleavage by a site-specific protease, and thus enables Förster Resonance Energy Transfer (FRET) between the two fluorophores. In contrast, kinase inhibition results in no phosphorylation, and the substrate can therefore be cleaved by the protease, resulting in no FRET. The degree of FRET is therefore correlated with kinase activity. For three of the 55 kinases [CamK1, death-associated protein kinase 1 (DAPK1), and AMPK-related protein kinase 5 (ARK5)], Z′-LYTE assays were not available, and Adapta Universal Kinase Assay was used instead. Adapta is a fluorescence-based immunoassay that detects ADP produced by kinase activity. In summary, two cell-free kinase assays were used to screen for the effect of PKI on fifty-five serine/threonine kinases.

PKI (6–22) amide was used to for the initial screen because this short peptide is a potent inhibitor of PKA and has been widely used to inhibit PKA activity in numerous studies. We first determined with a high concentration of PKI (6–22) amide (5 µM) the kinases whose activities are PKI sensitive at all and therefore merit further studies with dose response curves. As expected, 5 µM PKI (6–22) amide effectively inhibited PKA activity by 85%. PKI (6–22) amide had no effect on the majority of other kinases. However, at 5 μM, PKI (6–22) amide inhibited five other kinases by more than 10%. Surprisingly, PKI (6–22) amide (5 µM) enhanced the activity of 20 kinases by more than 10% (expressed in Figure 2 as negative inhibition). Notably, the eight most facilitated kinases were all isoforms of the PKC family. The amount of facilitation ranged from 59 to 93% (Figure 2).

Since we were surprised by the facilitating effect of PKI (6–22) amide on the PKC family, we set out to disambiguate the potential effect of PKI (6–22) amide on PKC in the kinase reaction, on the protease reaction (development reaction), or on fluorescence itself. PKI (6–22) amide did not induce any significant fluorescence readout change when added in or after the development reaction, but did induce a large change when added in the kinase reaction for PKC epsilon (Supplementary Figure S1). These control experiments indicate that PKI (6–22) amide indeed facilitated PKC activity.

To determine the concentration at which PKI (6–22) amide exerts effects, the dose-response curves of PKI (6–22) amide on activities of 11 selected kinases were measured. These kinases included 1) PKA, 2) all five other kinases whose activity is inhibited by more than 10% at 5 μM, and 3) five selected kinases whose activity is increased by more than 10% at 5 µM – three of these are PKC isoforms. The IC50 of PKI (6–22) amide for PKA was 0.61 nM, comparable to the value determined from previous studies (Glass et al., 1989a) (Figure 3A). PKI (6–22) amide altered the other kinase activities at much higher concentrations (Figures 3A,B). However, at the PKI concentrations often used in pharmacology experiments (1–10 µM), there were clear effects on several kinases. For example, at 5 μM, PKI (6–22) amide facilitated PKCα and PKCζ by more than 50%. Even at 1.7 µM, PKI (6–22) amide inhibited CamK1 activity by 30%, and facilitated PKCα activity by 33%.

Full length PKI may display different potency and specificity compared with the short peptide. Therefore, although the screen with PKI (6–22) amide above was useful in determining the specificity of PKI as pharmaceutical agents, it is important to perform dose-response curves with full length PKI to assess whether endogenous PKI may alter activities of kinases other than PKA. Full length mouse PKIα protein was synthesized and purified by High-Performance Liquid Chromatography (HPLC), and was subsequently used to determine the dose response curve of seven kinases (Figure 4).

PKIα inhibited PKA activity with an IC50 of 0.11 nM, with higher potency than PKI (6–22) amide (Figure 4). PKIα did not have significant effect on CamK1 or p70S6K (Figure 4). In addition, PKIα also facilitated multiple PKC isoforms and ROCK1 at much higher concentrations (1–5 µM) (Figure 4). At 5 μM, full length PKIα facilitated PKCα, PKCβII, PKCζ, and ROCK1 by 25, 36, 42, and 24% respectively.

We found that in addition to inhibiting PKA activity, PKI also facilitated the kinase activity of multiple PKC isoforms. This was surprising because previous studies found little influence of PKI on PKC (Day et al., 1989; Smith et al., 1990). In principle, this could be due to PKI interfering with 1) the protease reaction, 2) the fluorescence detection, or 3) the FRET response. Interference with the protease reaction was minimal, because PKI had very little effect on the control assay with unphosphorylated cleavable substrates (PKI + no ATP vs. no ATP conditions in the kinase reaction). In addition, adding PKI in the protease development reaction for PKC did not result in a significant FRET change (Supplementary Figure S1). Similarly, neither full length PKIα nor PKI (6–22) amide is autofluorescent, as evidenced by the test compound fluorescence interference control with no kinase/peptide mixture. Finally, PKI did not have any significant effects on FRET itself, because the FRET readout was not significantly different when PKI was added after the development reaction. These data indicate that PKI facilitated the kinase activity of multiple PKC isoforms.

Several technical differences may explain the differences between our results and previous studies that did not find significant influence of PKI on PKC. First, these studies differed in the form of PKI used: Smith et al. (Smith et al., 1990) used PKI-tide whose sequence is IAAGRTGRRQAIHDILVAA, whereas we used PKI (6–22) amide whose sequence is TYADFIASGRTGRRNAI (bolded amino acids are the shared fragment, and differences in sequences within the shared fragment are underlined). Second, the forms of PKC used are different: we used purified human recombinant PKC isoforms expressed in insect cells, whereas previous studies used partially purified PKC without clear distinction of isoforms (Day et al., 1989; Smith et al., 1990). Finally, although previous results did not reach statistical significance, Day et al. (Day et al., 1989) showed a mild enhancement of PKC activity by full length PKI in a cell-based assay with PKI transfection.

PKI (6–22) amide and other short peptides of PKI are widely used in pharmacological experiments to inhibit PKA activity. They are often used at between 1 and 10 μM, either intracellularly through a whole-cell recording pipette, or applied in the bath in cell-permeant myristoylated forms. Here, we found that PKI (6–22) amide, at these concentrations, also inhibits CamK1, and facilitates the activity of multiple PKC isoforms, possibly acting as a positive allosteric modulator. Future directions will involve testing the effects of PKI (6–22) amide on kinase activity in a cell-based system. Importantly, for any users of PKI to perturb PKA functions, these additional targets of PKI need to be taken into account to design experiments and interpret results.

Despite the additional targets of PKI identified in our study, PKI seems more specific than the other commonly used PKA inhibitors, H89 and KT 5720 (Murray, 2008). This may be due to their different mechanisms of action. PKI binds to the catalytic subunit of PKA with high affinity by serving as a pseudosubstrate, whereas H89 and KT 5720 competitively inhibits ATP binding on the PKA catalytic subunit. The off-target effects of these reagents have partial overlap (Murray, 2008). For example, PKBα is inhibited by all three compounds, whereas PDK1 activity is facilitated by PKI (6–22) amide but inhibited by KT5720. Similarly, H89 inhibits ROCK2 potently but PKI has little effect on ROCK2. Since all three reagents are commonly used as pharmacological inhibitors of PKA, their non-selective actions need to be carefully considered to interpret experimental results.

We found that full length PKIα, one of the endogenously expressed isoforms, enhanced the activity of ROCK1 and multiple PKC isoforms at high concentrations. This raises the interesting possibility that PKI acts as a molecular switch to regulate the balance between PKA and PKC activity. However, the validity of this hypothesis depends on the endogenous concentration of PKIα, which is hard to determine, partially because the local concentrations of PKIα within a subcellular compartment may be higher than what can be determined with traditional biochemical methods. Our study therefore lays the foundation for future explorations on how endogenous PKI impacts cellular signaling and confines it in space and time through non-classical targets.