Quantification of Binding of Small Molecules to Native Proteins Overexpressed in Living Cells

Abstract

The affinity and selectivity of small molecules for proteins drive drug discovery and development. We report a fluorescent probe cellular binding assay (FPCBA) for determination of these values for native (untagged) proteins overexpressed in living cells. This method uses fluorophores such as Pacific Blue (PB) linked to cell permeable protein ligands to generate probes that rapidly and reversibly equilibrate with intracellular targets as established by kinetic assays of cellular uptake and efflux. To analyze binding to untagged proteins, an internal ribosomal entry site (IRES) vector was employed that allows a single mRNA to encode both the protein target and a separate orthogonal fluorescent protein (mVenus). This enabled cellular uptake of the probe to be correlated with protein expression by flow cytometry, allowing measurement of cellular dissociation constants (K_d) of the probe. This approach was validated by studies of binding of allosteric activators to eight different Protein Kinase C (PKC) isozymes. Full-length PKCs expressed in transiently transfected HEK293T cells were used to measure cellular K_d values of a probe comprising PB linked to the natural product phorbol via a carbamate. These values were further used to determine competitive binding constants (cellular K_i values) of the non-fluorescent phorbol ester PDBu and the anticancer agent bryostatin 1 for each isozyme. For some PKC-small molecule pairs, these cellular K_i values matched known biochemical K_i values, but for others altered selectivity was observed in cells. This approach can facilitate quantification of interactions of small molecules with physiologically relevant native proteins.

Graphical Abstract

INTRODUCTION

Approximately 97% of oncology drug candidates that reach clinical trials are not approved by the FDA.¹ One factor thought to contribute to these low success rates is a poor understanding of the affinity and selectivity of small molecules for presumed target proteins in physiologically relevant living systems.² Although these affinities can often be measured with recombinant proteins,^3,4 purified proteins do not consistently mimic endogenous cellular proteins because biochemical and biophysical experiments do not precisely replicate the intracellular environment. Within cells, as many as 50% of proteins are post-translationally modified,⁵ and endogenous cellular proteins extensively assemble into complexes in specific subcellular environments. Other factors that can affect small molecule-protein interactions in living cells include off-target associations,² competition with endogenous factors,⁶ mechanisms of cellular uptake and efflux,^{7, 8} and xenobiotic metabolism.⁹ Consequently, cell-based assays of these interactions are needed to provide additional physiological relevance. These assays can establish selectivity for specific target proteins, determine the relationship between free drug in blood or tissues and pharmacodynamic effects, optimize drug pharmacokinetics, and select doses for preclinical and clinical studies. As a result, methods for quantification of engagement of drug targets by small molecules in living cells are valuable for drug discovery and development.

When proteins are exposed on the cell surface, radioligands,³ and fluorescent probes,^10–14 can often be used to analyze affinities of small molecules. However, for intracellular proteins, which comprise ~86% of the proteome,¹⁵ quantitative binding studies of living cells typically require sophisticated microscopy approaches,^16,17 which may have limited generality, or fusion of expressed proteins to protein or peptide tags.^18,19 One widely used approach is NanoBRET,²⁰ where small molecules are linked to fluorophores that accept energy from nanoluciferase. This enzyme is expressed in cells fused to a protein target of interest to detect binding of the fluorescent probe and evaluate competition by unlabeled small molecules. This method was recently used to characterize small molecule engagement of the kinases CDKL5,²¹ p38α²² and JAK3,²³ and the lysosomal ion channel TRPML1.²⁴ However, a drawback is that fusion of proteins to other proteins or peptides has the potential to affect their function.²⁵ Other approaches for studies of cellular target engagement²⁶ such as thermal shift,^{27, 28} DARTS,^{28, 29} HIPStA,³⁰ and chemical proteomic methods^31–34 have the advantage of not requiring tagging of proteins. However, many of these methods require lysis of cells for analysis, which can cause loss of equilibrium, reduce physiological relevance,³⁵ and can inactivate some proteins.³⁶ These and other cellular target engagement assays have been reviewed.^{37, 38}

We report a new method for quantification of engagement of native (untagged) intracellular proteins by small molecules in living cells (Figure 1). This method comprises a fluorescent probe cellular binding assay (FPCBA) that uses ligands of proteins linked to specific fluorophores such as Pacific Blue (PB).³⁹ PB was chosen because of its potential for cellular permeability and facile detection in cells by flow cytometry. This fluorophore has been shown to promote cellular efflux when linked to the small molecule paclitaxel,^40–42 and we hypothesized that this property might reduce non-specific binding of other linked compounds to cells, and facilitate quantification of higher affinity specific binding to overexpressed protein targets. To detect binding of small molecules to specific target proteins, we co-expressed proteins in cells with the yellow fluorescent protein mVenus.^{43, 44} This spectrally orthogonal fluorescent protein was investigated both fused to full-length target proteins and co-expressed as a separate marker with more physiologically relevant full-length untagged protein targets using an internal ribosomal entry site (IRES)⁴⁵ vector. In both cases, mVenus allowed ratiometric correlation of cellular fluorescence due to binding of the blue-fluorescent probe with expression of the target protein by flow cytometry. Because flow cytometry can precisely measure changes in cellular fluorescence, comparison of cells that overexpress a target protein with cells in the same population that lack expression allowed quantification of apparent affinities (cellular K_d values) of the probe for the target protein at equilibrium, as established by kinetic assays of probe uptake and efflux. These cellular K_d values further allowed quantification of cellular K_i values of non-fluorescent competitors for specific overexpressed target proteins.

Figure 1.

A fluorescent probe cellular binding assay of small molecule-protein interactions. Treatment of cells with cell-permeable ligands linked to fluorophores such as Pacific Blue under equilibrium binding conditions allows quantification of binding to expressed protein targets by flow cytometry in a homogeneous format without additional wash steps. Co-expression with the spectrally orthogonal fluorescent protein mVenus as a separate molecule using an IRES vector or fused to the protein target provides a ratiometric marker of target protein expression. This marker can be used to directly compare uptake of the probe by cells that do not express the target protein with cells that overexpress this protein in a single experiment. This method allows quantification of cellular K_d values of the probe for the target and quantification of cellular K_i values of non-fluorescent competitors.

To validate the FPCBA, we investigated small molecule allosteric activators of the Protein Kinase C (PKC) family of cytoplasmic serine-threonine kinases. These enzymes play key roles in signal transduction pathways that control cellular growth, differentiation, and apoptosis.⁴⁶ Most of these enzymes are activated by the second messenger diacylglycerol (DAG), a product of turnover of phosphatidylinositol, which binds regulatory C1 domains.⁴⁷ This binding induces a conformational change that expels an inhibitory pseudosubstrate peptide and causes translocation of PKCs to membranes where substrates are phosphorylated (Figure 2).⁴⁸ Although the C1 domain has tandem binding sites for DAG (C1a and C1b), only one of these sites engages the membrane at a time.^{49, 50} Signaling mediated by PKCs is later terminated by protein ubiquitination and degradation.⁵¹ Conventional PKC isozymes (α, βI, βII and γ) require DAG, Ca²⁺, and a phospholipid such as phosphatidyl serine (PS), whereas novel PKC isoforms (δ, ε, η and θ) require DAG and PS but not Ca²⁺ for activation. Other atypical PKCs (ζ and λ/ι) do not bind DAG and require only PS for activation. Mimics of DAG such as the natural products phorbol 12, 13-dibutyrate (PDBu)⁵² and bryostatin 1⁵³ (Figure 2) activate conventional and novel PKCs by binding their C1 domains with nanomolar affinities.^{54, 55} Small molecules that engage the ATP binding site of PKCs such as bisindolylmaleimide I (BIM1, GF109203X, Figure 2)⁵⁶ block the catalytic activity of PKCs and enhance interactions with allosteric activators by promoting the accessibility of C1 domains.⁵⁷ Early studies of activation of PKCs by phorbol esters suggested that they are oncogenes,^58–60 but some PKC activators such as bryostatin 1⁶¹ are potent anticancer agents.^{60, 62–64} Because mutations in PKCs in cancer are generally loss of function, recent studies classify most PKCs as tumor suppressors.^65–67 Using FPCBA, we show here that PDBu and bryostatin 1 exhibit altered selectivity for specific PKC isozymes in cells compared to biochemical assays, offering a novel approach for optimization of related anticancer agents. This method may facilitate studies of a wide range of small molecule-protein interactions in living cells.

Figure 2.

(A) Binding of PKC C1 domains to diacylglycerol (DAG), phorbol esters such as PDBu, or bryostatin 1 causes translocation of DAG-dependent PKCs to cellular membranes and initiation of signal transduction. (B) AlphaFold⁶⁸ model of full-length human PKCδ (ribbon diagram) bound to phorbol 13-acetate and bisindolylmaleimide I (sphere representations). Small molecules were overlaid on the AlphaFold model using X-ray structures 1PTR⁶⁹ and 2I0E⁷⁰). (C) Structures of the natural product PKC activators PDBu and bryostatin 1 and the ATP-competitive inhibitor bisindolylmaleimide I (BIM1).

RESULTS AND DISCUSSION

Design and synthesis of fluorescent phorbol carbamates as mimics of phorbol esters.

To investigate interactions of small molecules with PKC isozymes, we designed three phorbol carbamates (1–3) as fluorescent mimics of phorbol esters (Figure 3). PB-Phorbol (probe 1) includes the fluorinated Pacific Blue fluorophore,³⁹ probe 2 incorporates a non-fluorinated analogue derived from 7-hydroxycoumarin-3-carboxylic acid (7-OHCCA), and probe 3 substitutes the related 7-(diethylamino)coumarin-3-carboxylic acid (7-DCCA). The low molecular weight of these coumarin fluorophores enhances their potential for cellular permeability, and they were chosen because they are efficiently excited with violet lasers (405 nm) commonly found on confocal microscopes and flow cytometers. To maximize cellular permeability, these compounds include N-hexyl side chains that confer predicted hydrophobicities (cLogD_pH7.4 = 3.1 (1), 4.8 (2), and 6.2 (3)) greater than PDBu (cLogP = 1.6) but less than or comparable to bryostatin 1 (cLogP = 6.1, ChemAxon MarvinView 23.3.0 method), depending on the fluorophore. Probes 1–3 were synthesized from phorbol as shown in Scheme 1. Selective modification of the 13-OH of the trityl-protected derivative 4 to afford the protected 13-(4-nitrophenyl)carbonate 5 was established by 2D-NMR (Figure S21, Supporting Information). The PB fluorophore was prepared as previously described.⁷¹ Structurally related N-hexyl coumarin amides (20–22, Scheme 1) were synthesized as standards for optical spectroscopy. These standards were used to generate molar extinction coefficients (Figure S1, Supporting Information) for precise determination of concentrations of stock solutions of probes 1–3 by absorbance spectroscopy.

Figure 3.

Structures of fluorescent phorbol carbamates (1–3) as mimics of phorbol esters.

Scheme 1.

Synthesis of probes 1–3 (A), N-hexyl-N-dodecyl-coumarin amide precursors 6–8 (B), and N-hexyl coumarin amides 20–22 (C) as standards for optical spectroscopy.

Fluorescent phorbol derivatives 1–3 exhibit PKC-dependent cytotoxicity towards Jurkat lymphocytes.

As a preliminary assessment of their ability to activate PKCs, we evaluated the cytotoxicity of probes 1–3 towards HEK293T and Jurkat E6-1 cells. These probes and PDBu were found to be non-toxic towards HEK293T cells after treatment for 48 h at concentrations of ≤ 10 μM (Figure S2, Supporting Information). Due to activation of endogenous PKCs, phorbol esters are cytotoxic towards Jurkat lymphocytes,⁷² and PDBu exhibited high cytotoxic potency towards this cell line (IC₅₀ = 2 nM). This toxicity was completely blocked by addition of the PKC catalytic domain inhibitor BIM1 (2 μM, structure shown in Figure 1). BIM1 has previously been shown to protect Jurkat cells against apoptosis mediated by the PKC activator phorbol-12-myristate-13-acetate (PMA).⁷² Probes 1–3 were cytotoxic towards Jurkat lymphocytes but less potent (IC₅₀ = 300–600 nM) than PDBu. BIM1 reduced or eliminated this effect (Figure S2, Supporting Information), indicating that 1–3 activate PKC isozymes in this cell line.

Probes 1–3 translocate PKCβI-mVenus to cellular membranes analogous to phorbol esters.

To investigate interactions of probes 1–3 with specific PKCs, we generated mammalian expression vectors for eight full-length mouse isozymes.⁷³ Murine and human PKCs are highly homologous, exhibit very similar biochemical affinities for allosteric activators that bind C1 domains,⁵⁴ and amino acid sequences in the C1a-linker-C1b region for PKCβI, PKCγ, and PKCθ are identical in mice and humans. For PKCα, δ, ε, and η, the identities of mouse and human amino acids in this region range from 97.5% to 99.1%. These vectors were designed for expression of these enzymes either fused at their C-terminus to the monomeric yellow fluorescent protein mVenus^{43, 44} or as native (untagged) proteins co-expressed independently with mVenus using a bicistronic (IRES) vector that produces a single mRNA that encodes both proteins. This IRES vector was derived from a previously reported⁷⁴ variant that expresses the less fluorescent protein Venus. mVenus was chosen as a ratiometric marker of target protein expression because it is spectrally orthogonal to Pacific Blue and related 3-carboxycoumarins when these fluorophores are excited at 405 and 488 nm (optical spectra shown in the Supporting Information, Figure S1). When HEK293T cells were transiently transfected to express the PKCβI-mVenus fusion protein, and imaged by confocal microscopy, all three of these three blue-fluorescent probes (1–3) promoted translocation of this fusion protein to the plasma membrane (Figure 4), similar to other studies of translocation of PKC-GFP mediated by phorbol esters.⁷⁵ Confocal microscopy further revealed that these probes exhibited substantially higher blue fluorescence in transfected cells compared to adjacent non-transfected cells, with the highest overall fluorescence observed for probe 1 (Figure 4).

Figure 4.

Differential interference contrast (DIC) and confocal laser scanning micrographs of living HEK293T cells transiently transfected to express the PKCβI-mVenus fusion protein. The cytosolic yellow-fluorescent fusion protein in untreated cells (A) is translocated to cellular membranes upon treatment with probes 1 (B), 2 (C), and 3 (D) for 2 h (1 μM). White arrows point at representative non-transfected cells to illustrate enhanced uptake of the blue-fluorescent probes by nearby yellow-fluorescent transfected cells. [BIM1] = 0.5 μM. [DMSO] = 1%. Scale bars = 10 microns.

Probes 1–3 preferentially accumulate in cells that overexpress DAG-binding PKC isozymes.

Having established that 1–3 exhibit PKC-dependent cytotoxic activity and translocate PKCβI-mVenus to membranes of transiently transfected HEK293T cells, we investigated the accumulation of these probes in cells by flow cytometry. When HEK293T cells were transiently transfected with PKCβI-mVenus and suspended by treatment with the protease trypsin (Figure 5, panel A), a well-defined bimodal distribution of the population of yellow-fluorescent transfected and essentially non-fluorescent non-transfected cells (Figure 5, panel B, left) was observed. When these cells in suspension were additionally treated with blue-fluorescent 1 (1 μM, 2 h), and BIM1 (2 μM) to block PKC kinase activity, cells expressing high levels of PKCβI-mVenus showed greater median blue fluorescence from 1 compared to non-transfected cells (Figure 5, panel B, right). Similar results were obtained with native PKCβI overexpressed with mVenus as a separate protein using the IRES-mVenus vector (Figure 5, panel C). In contrast, this differential uptake of 1 was not observed with constructs encoding PKCζ, which does not bind phorbol esters (Figure S3, Supporting Information). These results are consistent with enhanced cellular uptake resulting from binding of 1 to overexpressed PKCβI-mVenus or native PKCβI in the cytoplasm.

Figure 5.

(A) The FPCBA method for analysis of binding of probes to proteins overexpressed in living cells. Co-expression with the yellow-fluorescent marker protein mVenus allows gating to define cell populations with high and low levels of the target protein for analysis of target-dependent cellular uptake of blue-fluorescent probes. (B) Analysis of cells expressing a PKCβI-mVenus fusion protein. (C) Analysis of cells expressing native PKCβI co-expressed with mVenus using an IRES vector. Cells were transiently transfected for 24 h, treated with probe 1 (1 μM) for 2 h at 37 °C (5% CO₂), and analyzed by flow cytometry without further wash steps at room temperature on a 96-well plate. The bimodal yellow fluorescence (Ex. 488 nm, left panels, FITC-A channel) from mVenus was gated as the lowest 20% of the population (P1 gate, non-transfected cells) and the highest 20% of population (P2 gate, highly transfected cells) to analyze differential accumulation of the blue-fluorescent probe 1 (Ex 405 nm, right panels, PB450-A channel) in cells. Compared to non-transfected cells (P1 gate), highly transfected cells (P2 gate) exhibit substantially higher blue fluorescence from uptake of probe 1. These gates were used to measure total binding of 1 to living cells overexpressing PKCβI and non-specific binding of 1 to non-transfected cells ([BIM1] = 2 μM, [FBS] = 4%, [DMSO] = 1%). Graphics in panel A were from the DataBase Center for Life Science (CC BY 4.0 license, unmodified).

To investigate the potential for quantitative analysis of binding to PKCs overexpressed in cells, we evaluated whether 1–3 could readily equilibrate with the intracellular target protein. Achieving equilibrium is necessary for accurate measurements of equilibrium dissociation constants (K_d), and measurements of K_d values under non-equilibrium conditions leads to a rightward shift in binding curves and underestimation of protein engagement.^{3, 4, 76, 77} Analysis of the kinetics of cellular uptake and efflux by confocal microscopy (Figure 6) revealed that the more polar probes 1 and 2 were rapidly taken up by transfected cells (uptake t_1/2(1) = 4 min; uptake t_1/2(2) = 9 min) with saturable kinetic profiles and half-times of less than 10 min. In contrast, the more hydrophobic probe 3 equilibrated much more slowly (uptake t_1/2(3) > 90 min). After treatment with these probes for 2 h, dilution with medium resulted in rapid loss of blue fluorescence for probes 1 and 2 (efflux t_1/2(1) = 3 min and (2) = 8 min) but much slower efflux of probe 3 (efflux t_1/2(3) > 90 min). This rapid loss of signal for probe 1 is consistent with active efflux previously observed^{40, 42} with other cell-permeable probes linked to Pacific Blue. This rapid efflux of 1 and 2 indicates that these probes reversibly interact with the expressed target protein, another key criterion for studies of equilibrium binding. Because probe 3 did not rapidly equilibrate with or efflux from transfected cells, we focused on probes 1 and 2 for further studies.

Figure 6.

Kinetics of uptake (A) and efflux (B) of probes 1–3 by transiently transfected living HEK293T cells. Adherent cells expressing PKCβI-mVenus were treated with probes and changes in blue fluorescence of transfected cells were analyzed in one-minute time intervals by confocal and widefield microscopy in the humidified incubation chamber (37 °C, 5% CO₂) of an Opera Phenix Plus imaging system. (A) Probes 1 and 2 (1 μM) were rapidly taken up by transfected cells, whereas probe 3 (1 μM) exhibited substantially slower uptake kinetics as determined by curve fitting with a one-phase association model. (B) To measure efflux, transfected cells were equilibrated with probes (1 at 400 nM; 2 and 3 at 1 μM) for 2 h at 37 °C (5% CO₂) followed by dilution by 9-fold with assay medium with the on-board liquid handler of the Opera Phenix. Analysis with a one-phase dissociation model revealed that probes 1 and 2 underwent rapid efflux, with half-lives of less than 10 min, whereas probe 3 exhibited much longer retention by cells. [BIM1] = 2 μM. [FBS] = 4%. [DMSO] = 1%. Ranges are shown as 95% confidence intervals from curve fitting.

Binding of probes 1 and 2 to overexpressed PKCβI can be quantified by flow cytometry.

For quantitative equilibrium binding studies, the fixed component (the overexpressed PKC protein in this case) generally needs to be maintained at a concentration per well near or below the K_d to avoid inaccuracies associated with ligand depletion.⁴ When the concentration of receptors per well is two-times higher than the dissociation constant (K_d or K_i), the theoretical rightward shift in a measured K_d or K_i value is a factor of 2.³ An advantage of flow cytometry for these types of binding studies is that even if receptors or enzymes are expressed at high levels, such as 10 μM per cell, cell densities can be adjusted so that the total concentration of receptors per well is near or below one nanomolar to avoid ligand depletion. Importantly, the concentration of this fixed component (expressed PKC) needs not be precisely known to determine accurate affinities provided that the total concentration per well is held constant below the K_d value, and both total and non-specific binding can be measured. To quantify the binding of 1 and 2 to PKCβI overexpressed in cells, we used flow cytometry to analyze the blue fluorescence of transfected and non-transfected cells as a function of probe concentration after equilibration for 2 h at 37 °C (Figure 7). Using homogeneous (no additional wash steps) assays, we directly compared binding of a PKCβI-mVenus fusion protein with its native PKCβI counterpart overexpressed separately from mVenus using a bicistronic IRES vector.

Figure 7.

Quantification of binding of probes 1 and 2 to PKCβI overexpressed living HEK293T cells in the presence and absence of the PKC catalytic domain inhibitor BIM1. Cells were transiently transfected to express PKCβI-mVenus fusion protein (A, B) or native PKCβI (IRES-mVenus, C, D) for 24 h, treated with 1 and 2 for 2 h at 37 °C (5% CO₂) to promote equilibration, and analyzed by flow cytometry without further wash steps at 22 ° C. Cellular K_d values were calculated by non-linear regression using a one-site total and non-specific binding model. Comparison of total binding (transfected cells) and non-specific binding (non-transfected cells) of 1 and 2 to PKCβI-mVenus revealed higher signal-to-background (S/B) for probe 1 compared to 2 and enhanced affinity and improved S/B when BIM1 was added. S/B was calculated at 0.625 μM probe as blue fluorescence of transfected (P2)/non-transfected (P1) cells. [BIM1] = 2 μM. [FBS] = 4%. [DMSO] = 1%. Ranges are shown as 95% confidence intervals from curve fitting. Error bars (SD, N = 2) were omitted when smaller than symbols.

As shown in Figure 7, in non-transfected living cells (the P1 gate shown in Figure 5) probes 1 and 2 exhibited low background levels of linear dose dependent blue fluorescence. This is consistent with our previous observations^40–42 of cellular efflux of PB-linked probes. This non-specific binding was expected to result from interactions of the probe with membranes, endogenous C1 domain-containing proteins, and other cellular biomolecules. This low background signal facilitated quantification of cellular K_d values, where the linear non-specific binding of probes to non-transfected cells was subtracted from total binding to highly transfected cells expressing PKCβI, as gated by the fluorescence of the mVenus marker protein (the P2 gate shown in Figure 5). Specific binding was analyzed using non-linear regression with a one-site total and non-specific binding model (GraphPad Prism). Because these biologically active probes were analyzed in living cells, the maximal concentration used for determination of cellular K_d values was generally limited to values that achieved ≥ 50% saturation of specific binding sites and allowed accurate curve fitting. This use of the lowest possible top concentration needed to achieve ≥ 50% saturation of binding reduced variability observed when higher concentrations of the probe were added to cells overexpressing some PKCs.

We further compared cells treated with and without BIM1 to reduce the kinase activity of PKCs and isolate/enhance the analysis of interactions of 1 and 2 with the regulatory C1 domains of these enzymes. As shown in Figure 7, treatment of cells with BIM1 increased the efflux of probes 1 and 2 (compare the linear non-specific binding curves), consistent with prior observations⁷⁸ that efflux transporters are downregulated by activation of some PKCs. Treatment with BIM1 enhanced the signal-to-background (S/B, calculated at 0.625 μM as blue fluorescence of transfected/non-transfected cells) and improved the apparent cellular affinity of these probes for PKCβI, especially when higher top concentrations that conferred greater PKC activation were used for curve fitting. Compared to probe 2, probe 1 exhibited superior cellular properties, with cellular K_d = 163 nM for PKCβI-mVenus (top concentration of 0.625 μM) and S/B = 4.0 (with BIM1) under these conditions. In contrast, the more hydrophobic probe 2 exhibited a somewhat higher affinity (cellular K_d = 91 nM) for PKCβI-mVenus but was associated with a substantially lower S/B of 2.0. Comparison of the PKCβI-mVenus fusion protein with native PKCβI (compare Figure 7 panels A and C) revealed that the untagged native protein exhibited ca. 2-fold higher affinity for 1 in the presence of BIM1. This native protein bound 1 with a cellular K_d of 94 nM calculated using a top concentration of 0.625 μM for curve fitting. However, we found that limiting the top concentration of 1 to 0.078 μM for curve fitting could be used to reduce activation of native PKCβI by this probe, minimizing perturbation of the biological system, while achieving ≥ 50% saturation of binding sites. Under these conditions, a more consistent cellular K_d of 55 nM was measured (data shown in Figure 8 and supporting information Figures S3, S7, and S8). Compared with probe 2, the greater S/B of probe 1 for native PKCβI (5.7 with BIM1) made probe 1 particularly attractive for further quantitative studies of engagement of PKC family members by small molecules.

Figure 8.

Specific and competitive binding of small molecules to murine PKC isozymes overexpressed in living HEK293T cells by flow cytometry. Specific binding curves for probe 1 were generated by subtraction of blue fluorescence of non-transfected cells (non-specific binding) from blue fluorescence of cells that overexpress specific PKC isozymes (total binding) using a single population of cells. Coexpression of mVenus as a fusion or independent protein provided a yellow-fluorescent marker of PKC expression for gating of transfected and non-transfected cell populations. (A, B) Binding of 1 to PKC-mVenus fusion proteins (A) and native PKCs (IRES-mVenus, B). (C, D) Competitive binding of PDBu (C) and bryostatin 1 (D) to native PKC isozymes (IRES-mVenus). HEK293T cells were transiently transfected with PKC expression vectors, treated with compounds for 2 h at 37 °C (5% CO₂) to promote equilibration, and analyzed by flow cytometry without further wash steps at 22 °C. For competition assays, concentrations of probe 1 were fixed within ca. 2-fold of measured cellular K_d values (cellular K_d values, cellular K_i values, and probe concentrations are shown in Table 1). [BIM1] = 2 μM. [FBS] = 4%. [DMSO] = 1%.

Because albumin in fetal bovine serum (FBS) binds small molecules⁷⁹ and can contribute to ligand depletion,⁷⁶ we additionally examined the effect of varying the percentage of FBS on the cellular K_d of probe 1 for native PKCβI overexpressed in HEK293T cells. Increasing the FBS from 1% to 10% in media decreased the cellular affinity of probe 1 by 3-fold (Figure S4, Supporting Information), but the difference between 1% FBS and at 4% FBS in media on affinity was less than 2-fold, and 4% FBS was chosen for cellular binding studies by FPCBA to maximize cellular viability and minimize ligand depletion. The use of media containing 4% FBS has been previously reported⁸⁰ for NanoBRET assays.

Quantification of binding of probe 1 to specific members of the PKC family in living cells.

Because of its superior cellular properties, probe 1 was used to measure binding isotherms for eight full length murine PKC isozymes overexpressed in live cells by flow cytometry (Figure 8 and Figure S3, Supporting Information). To focus on interactions of probe 1 with PKC C1 domains, cellular K_d values of this probe were determined in the presence of the PKC catalytic domain inhibitor BIM1 for both native (untagged) PKCs and PKCs fused to mVenus (Table 1). As expected from prior biochemical binding studies of PDBu,⁸¹ the PKCζ isozyme did not bind probe 1 and provided a negative control. The highest apparent affinities of 1 were observed for native PKCα (cellular K_d = 124 nM) and native PKCβI (cellular K_d = 55 nM), with affinities for other native isozymes ranging from 75 nM (PKCγ) to 997 nM (PKCθ). Because of the high biological activity of probe 1, increased affinities for some isozymes, such as native PKCβI, were observed when lower top concentrations of probe were used for binding studies (e.g. 0.078 μM for native PKCβI), and the lowest top concentrations of probe 1 that provided ≥ 50% saturation and the most consistent measurements based on independent replicates were used to determine the values shown in Table 1 and Figure S3. Fusion of mVenus to the C-terminus of PKCs appeared to modestly but consistently reduce the cellular affinity of probe 1 (Table 1).

Table 1.

Cellular and biochemical affinities of small molecules for overexpressed PKCs. Cellular affinities were measured with PKC-mVenus and native PKCs (IRES-mVenus) by treatment of transiently transfected cells with probe 1 and BIM1 (2 μM) for 2 h at 37 °C (5% CO₂) followed by analysis of living cells by flow cytometry at 22 °C. Cellular K_d values were calculated from the total and non-specific binding curves shown in Figure S3 and were chosen as representative based on at least two independent replicates. The cellular K_d value of probe 1 for PKCβI was measured independently of those shown in Figures 7 and S4 (Supporting Information). For competition with PDBu and bryostatin 1, cells overexpressing native PKCs were treated with probe 1 at a fixed concentration within ca. 2-fold of its K_d value for a given isozyme (100 nM for PKCα and PKCβI, 200 nM for PKCγ and PKCη, 400 nM for PKCδ, PKCε, and PKCθ). Previously reported biochemical K_i values for PDBu⁸⁶ and bryostatin 1⁸⁷ (right columns) were measured with [³H]PDBu at 4 °C. Values in parentheses represent 95% confidence intervals from curve fitting of representative data sets (N = 2).

Isozyme	Cellular Kd values for probe 1 (nM)		Cellular Ki values for native PKCs (nM)		Previously reported biochemical Ki values (nM)
	PKC-mVenus	Native PKCs	PDBu	Bryostatin 1	PDBu	Bryostatin 1
PKCα	301 (276–328)	124 (96–163)	22 (18–28)	4 (3–5)	15.1	0.81
PKCβI	121 (103–142)	55 (41–75)	15 (11–20)	2 (1–3)	8.8	2.2
PKCγ	453 (348–601)	75 (57–98)	18 (12–26)	2 (1–3)	13.8	2.2
PKCδ	733 (528–1075)	498 (320–830)	45 (37–54)	17 (12–24)	4.5	2.1
PKCε	634 (552–732)	273 (229–330)	43 (36–51)	11 (7–17)	6.2	3.0
PKCζ	No binding	No binding	No binding	No binding	No binding	No binding
PKCη	603 (498–741)	167 (142–196)	27 (22–35)	5 (3–7)	18.4	2.2
PKCθ	1264 (655–3107)	997 (547–2099)	49 (37–65)	8 (5–14)	28.8	1.5

Some proteins have been expressed in HEK293 cells at up to 5% of total cellular protein.⁸² To confirm that measured affinities were not affected by ligand depletion from high protein expression mediated by the strong CMV promoter and SV40 origin of replication present on pmVenus-N1 and pIRES-mVenus, which enhances replication of these plasmids in HEK293T cells, we measured intracellular protein concentrations by analyzing molecules of mVenus per cell using beads bearing standardized numbers of fluorophores (Figure S5, Supporting Information). Flow cytometry revealed that concentrations of PKC-mVenus proteins reached median levels of 5–33 μM in individual cells in the top 20% of the population (Supporting Information, Table S1). Because B_max values of binding of probe 1 (a measure of the number of functional binding sites per cell) were highly correlated with the fluorescence of mVenus fusion proteins (Figure S5B), these values were used to calculate concentrations of native overexpressed PKCs per cell (5–16 μM) by correlating B_max of binding of probe 1 to mVenus fusions and cognate untagged PKCs. These intracellular protein concentrations were within 2-fold of concentrations calculated by independent referencing of B_max values of probe 1 to blue-fluorescent bead standards (Figure S5A). EMCV IRES vectors have been shown to express the IRES-dependent second gene at 6% to 100% of the first gene,⁸³ and based on comparisons of yellow fluorescence and B_max values, mVenus expressed from the IRES-mVenus vectors ranged from 24% to 67% of the level observed in the mVenus fusion proteins. The total concentration of overexpressed PKC isozymes in media per well ([PKC] < 2 nM for 37,500 cells/150 μL) was far below the measured cellular K_d values for the probe (Supporting Information, Table S1).

For some PKCs, PDBu and bryostatin 1 exhibit altered selectivity in living cells compared to biochemical assays.

The C1 domains of specific PKC isozymes represent promising targets for drug discovery.⁸⁴ Although isozyme-specific peptides have been described,⁸⁵ small molecule activators such as PDBu and bryostatin 1 exhibit low selectivity for specific isozymes in biochemical assays.^{86, 87} To evaluate the selectivities of these small molecule activators in live cells, we used the cellular K_d values of probe 1 to quantify cellular K_i values of these non-fluorescent competitors for specific overexpressed native PKC isozymes. As shown in Figure 8 (panels C and D), these compounds reduced the blue fluorescence of cells treated with 1 in a dose-dependent manner. Using a Fit K_i model (GraphPad Prism) with the isozyme-specific cellular K_d values measured for this probe and its fixed concentration, we determined the cellular K_i values listed in Table 1. Because tracer concentrations substantially above K_d can cause ligand depletion, manifesting as a rightward shift of IC₅₀ curves,⁸⁸ we additionally measured K_i values of PDBu for the high affinity PKCβ1 using concentrations of probe 1 that range from near its cellular K_d (50 nM) to above its cellular K_d (400 nM). These experiments revealed that this range of concentrations of probe 1 caused a less than 3-fold fluctuation in measured cellular IC₅₀ and K_i values, consistent with low ligand depletion under these conditions (Figure S4, Supporting Information). Because these values were most consistent when the probe concentration was fixed within about 2-fold of the cellular K_d, concentrations of the probe were used in this range for each isozyme for competition experiments.

For most of the PKC isozymes investigated, the cellular K_i values for PDBu measured with probe 1 were within 2-fold of biochemical K_i values measured with radioactive PDBu.⁸⁶ This provides an important validation of this FPCBA method. However, more substantial differences were observed for others. As shown in Table 1, differences in biochemical versus cellular affinities of PDBu of ca. 2-fold (statistically insignificant) were observed for PKCα (K_i = 15.1 nM (biochemical) vs 22 nM (cellular)), PKCβI (K_i = 8.8 nM vs 15 nM), PKCγ (K_i = 13.8 nM vs 18 nM), PKCη (K_i = 18.4 nM vs 27 nM), and PKCθ (K_i = 28.8 nM vs 49 nM). In contrast, greater differences of 7–10-fold were observed for PKCδ (K_i = 4.5 nM vs 45 nM) and PKCε (K_i = 6.2 nM vs 43 nM). The approach used to gate live cells by flow cytometry (Figures S6 and S7) and representative data and non-linear regression parameters used to analyze binding of probe 1 to PKCβI and competition binding with PDBu (Figures S7 and S8) are shown in the supporting information.

Biochemical affinities of bryostatin 1 for PKCs are reported⁸⁷ to be highly similar for all isozymes (Table 1). These affinities span a 4-fold range from 0.81 nM to 3 nM, but five isozymes exhibit biochemical K_i values close to 2 nM. In contrast, cellular assays of binding of bryostatin 1 revealed a broader range of affinities for PKCs. Whereas overexpressed PKCα, PKCβI, PKCγ, PKCε, PKCη, and PKCθ bound bryostatin 1 in cells within 1–5 fold of reported biochemical values, bryostatin 1 bound PKCδ with 8-fold lower affinity (cellular K_i = 17 nM). This cellular/biochemical divergence for PKCδ is particularly of interest because the anticancer activity of bryostatin 1^{60, 62, 63} is reported to involve selective stabilization of this isozyme.⁶⁴

CONCLUSIONS

The FPCBA method described here allows quantitative studies of interactions of small molecules with native full-length proteins overexpressed in living cells. We show that these interactions can be analyzed by flow cytometry by quantifying changes in cellular fluorescence that result from specific binding of an optimized fluorescent small molecule probe to specific overexpressed protein targets. To develop this homogeneous assay, we analyzed the binding of allosteric activators to members of the PKC family of serine-threonine kinases. These proteins were overexpressed in HEK293T cells by transient transfection with mammalian expression vectors that encode both the specific target of interest and a fluorescent protein (mVenus) as a ratiometric marker of protein expression. Comparison of cellular uptake of an orthogonal blue-fluorescent probe with both transfected (high yellow fluorescence) and non-transfected (low yellow fluorescence) cells in the same population was used to analyze specific binding of the probe and non-fluorescent competitors to the intracellular target protein.

To validate the FPCBA, eight full-length PKC isozymes were overexpressed in HEK293T cells. These isozymes were investigated both directly fused at their C-terminus to the fluorescent protein mVenus and as native untagged isozymes expressed separately from mVenus using an IRES vector. Because this IRES vector uses a single mRNA to encode both proteins, it similarly provides a ratiometric fluorescent marker of protein expression. Key to this approach was the synthesis of cell permeable molecular probes that bind PKC C1 domains and incorporate coumarin fluorophores such as Pacific Blue that are spectrally orthogonal to mVenus, readily detected by flow cytometry, and show rapid kinetics of cellular uptake and efflux that promote equilibration.

To accurately measure dissociation constants in cell-based assays, systems must reach equilibrium.⁷⁷ Using kinetic assays of probe uptake and efflux, we confirmed that uptake of the probes PB-Phorbol (1) and 7-OHCCA-Phorbol (2) by HEK293T cells was sufficiently rapid (influx t_1/2 < 10 min) for equilibrium with the cytoplasm to be reached within 1–2 h, conditions that were designed to minimally perturb the living biological system. For binding assays, waiting at least five half-times assures ≥ 97% equilibration.³ These interactions were also rapidly reversible, with cellular efflux half-times of < 10 min when probes were diluted, establishing another important criterion for saturation binding assays. This efflux appears to involve specific transporters as evidenced by reduced cellular fluorescence when cells were treated with the PKC inhibitor BIM1, consistent with previous studies⁷⁸ of downregulation of efflux transporters mediated by PKCs. Other PB-linked probes have been shown to be substrates of the efflux transporters MDR1 and MRP2,⁴¹ and some engagement of active efflux pathways likely reduces non-specific binding to cellular biomolecules, facilitating detection of higher affinity specific binding in live cells. Active influx via transporters may also affect the cellular affinities of small molecules for proteins. Other factors that may affect cellular affinities include compound solubility and partitioning of compounds into cellular membranes. To maximize solubility, the hydrophobic probes and competitors investigated here were formulated with dimethyl sulfoxide (DMSO). Although this commonly used cosolvent can affect biological systems under certain conditions,⁸⁹ the rapid influx kinetics of probes such as 1 allow equilibration for cellular binding assays after short incubation times (≤ 2 h) to limit effects on cellular biology.

To maximize the assay signal window⁹⁰ and minimize variance, the physicochemical properties of probes were optimized for FPCBA. Probes 1–3 include an N-hexyl side chain because early studies of an analogue of 1 that linked the phorbol carbamate to the fluorophore via N,N’-dimethyl-1,12-dodecanediamine revealed that efflux was reduced and affinity for PKCs enhanced when a more hydrophobic N-hexyl side chain was installed. For these hydrophobic allosteric activators of PKC that bind both proteins and membranes, the polar Pacific Blue fluorophore proved to be more effective than two more hydrophobic coumarins. However, for more polar protein ligands, more hydrophobic fluorophores are likely to be required to optimize intracellular target engagement using these types of assays. Among probes 1–3, the best activity was observed for probe 1 where its moderately high lipophilicity (cLogD_pH7.4 = 3.1) was associated with rapid uptake and efflux kinetics and high specific binding to overexpressed PKCs.

Cellular affinities were measured by plotting the concentration of free ligand against the change in signal due to binding at equilibrium. The assumption in these experiments is that the concentration of ligand added to each well is equivalent to the free ligand concentration. However, if the total concentration of receptors per well is > 10% of the K_d or K_i of a small molecule, binding curves can shift rightward due to ligand depletion.³ When the concentration of receptors per well is twice the K_d or K_i, the theoretical rightward shift in the calculated K_d or K_i is a factor of two.³ In cell-based assays, this type of ligand depletion can often be controlled by limiting the density of cells per well to the lowest level where an adequate assay signal window is achieved.⁷⁷ For the FPCBA described here, these conditions were established as 37,500 cells/150 μL on a 96-well plate. Because FPCBA uses transient transfection, reducing the amount of transfected plasmid DNA or the duration of transfection could provide other strategies to control receptor numbers and minimize ligand depletion. To measure the concentrations of overexpressed PKC-mVenus fusion proteins, we analyzed intracellular concentrations of these enzymes by comparing the median fluorescence of fused mVenus per cell with beads bearing standardized numbers of fluorophores. For native overexpressed PKCs, we calculated median protein concentrations by using B_max values for binding of probe 1, a measure of intracellular binding sites, that were normalized to B_max values of cognate mVenus fusion proteins. Alternatively, median concentrations of native proteins in the top 20% of cells can be directly calculated from B_max values of probe 1 using blue-fluorescent bead standards. For competition assays, overexpressed PKCs were present at < 2 nM concentrations per well to minimize ligand depletion. Ligand depletion was additionally minimized by selecting 4% FBS in media for binding assays and using concentrations of probe fixed within ca. 2-fold of the K_d value of a given isozyme for competition assays.

A benefit of FPCBA is that the cellular K_d of the fluorescent probe can be used to determine cellular K_i values of competitors on the same microtiter plate. This reduces the day-to-day ca. 2-fold variance that is often associated with assays of living cells. Cellular K_d values of probe 1 for specific PKC isozymes were used with the fixed probe concentration to determine cellular K_i values for the competitors PDBu and bryostatin 1 by non-linear regression.

Bryostatin 1 was of particular interest because this marine natural product has been investigated in over 30 clinical trials as an anti-cancer agent, an anti-AIDS agent, and as a treatment of Alzheimer’s disease.⁶¹ Importantly, for competitive binding to overexpressed PKCα, PKCβI, PKCγ, PKCη, and PKCθ, cellular K_i values for PDBu were within 2-fold and for bryostatin 1 were within 5-fold of previously reported biochemical K_i values, supporting the validity of this method. Moreover, the cellular K_i values for binding of bryostatin 1 to PKCβI and PKCγ matched previously reported biochemical K_i values (2 nM). However, we found that the cellular K_i values measured for binding of PDBu and bryostatin 1 to PKCδ and PKCε diverged more substantially from previously reported biochemical K_i values. The greatest divergence was for PKCδ, which bound PDBu 10-fold less tightly and bryostatin 1 8-fold less tightly in cells than predicted from biochemical assays. This may relate to the inability of biochemical assays to precisely replicate the cellular environment. Optimization of cellular versus biochemical selectivity could be particularly important for the development of simpler analogues^{87, 91} of bryostatin 1, which is thought to manifest anticancer activity by preferentially stabilizing PKCδ.⁶⁴

The FPCBA provides a unique method to analyze interactions of small molecules with physiologically relevant native proteins overexpressed in live cells. This method has the potential to be extended to a wide range of target proteins that can be expressed appreciably above endogenous levels, and we will describe the use of this approach to study small molecule modulators of other protein families in due course. Future applications might include adaptation for high throughput screening by flow cytometry. These types of studies of target engagement may also help inform decisions regarding compounds to advance to clinical trials and influence concentration ranges chosen for evaluation. These values could help predict drug safety and offer starting points for the generation of new probes and drugs.

EXPERIMENTAL SECTION

Normalization of concentrations of fluorescent probes by absorbance spectroscopy.

Molar extinction coefficients (ε) determined for N-hexyl amides 20–22 (Figure S1, Supporting Information, were used to normalize the concentrations of DMSO stock solutions of 1–3 (1 mM) by absorbance spectroscopy. Using dry powders of pure 20–22, Beer’s Law plots of absorbance λ_max versus concentration in PBS (10% DMSO) were used to calculate the slope (ε) by linear least squares curve fitting. The ε values of 20 and 22 were analyzed at pH 7.4, whereas the ε value of the hydroxycoumarin derivative 21 was analyzed at pH 10 to assure complete deprotonation. 6,8-Difluoro-N-hexyl-7-hydroxy-2-oxo-2H-chromene-3-carboxamide (20, ε_{405 nm} = 29,000 M⁻¹ cm⁻¹, PBS pH 7.4, 10% DMSO) was used as a standard for probe 1. N-Hexyl-7-hydroxy-2-oxo-2H-chromene-3-carboxamide (21, ε_{405 nm} = 22,000 M⁻¹ cm⁻¹, PBS pH 10, 10% DMSO) was used as a standard for probe 2. 7-(Diethylamino)-N-hexyl-2-oxo-2H-chromene-3-carboxamide (22, ε_{425 nm} = 16,000 M⁻¹ cm⁻¹, PBS pH 7.4, 10% DMSO) was used as a standard for probe 3. Absorbance-normalized stock solutions of 1–3 in DMSO (1 mM) prepared fresh from ~ 1 mg of dry powder were typically stored at −20 °C for less than two weeks prior to bioassays. Over longer periods, decomposition of probes in DMSO may be observed.

Kinetic studies of equilibration of fluorescent probes in living cells.

The kinetics of cellular uptake and efflux of fluorescent probes were measured with a PerkinElmer Opera Phenix Plus High Content Imaging System. HEK293T cells were suspended with trypsin-EDTA (0.25%) and resuspended in fresh complete DMEM medium at 150,000 and 200,000 cells per mL. To a black 96-well PhenoPlate (PerkinElmer #6055302) was added 100 μL per well at each cell density. The plate was returned to the 37 °C (5% CO₂) incubator overnight. These cells were subsequently transfected with the plasmid encoding the PKCβ1-mVenus fusion protein. For transfection, plasmid DNA (1.5 μg) and X-tremeGENE HP (3 μL) were added to Opti-MEM I Reduced Serum Medium (150 μL) in a sterile Eppendorf tube (1.5 mL). This mixture was maintained at room temperature for 30 min to allow complex formation. This transfection solution (10 μL) was added to each well of the 96-well plate. The 96-well plate was returned to the incubator for 24 h. After this transfection for 24 h, the confluence of the transiently transfected cells on the 96-well plate was analyzed using the Phenix microscope, and a well with evenly distributed cells and a cell density of 40–60% was selected for compound treatment and kinetic analysis. For the kinetic uptake assay, cells were treated with probes 1–3 (1 μM) and BIM1 (2 μM) in assay medium (4% FBS, 1% DMSO). To achieve this, after 24 h of transfection, the culture medium with transfection reagents was replaced with 50 μL of assay medium and the plate was loaded on the Phenix and incubated at 37 °C (5% CO₂) for 30 min. Assay medium (150 μL) containing the probe (1.33 μM) and BIM1 (2.66 μM) was added to wells containing 50 μL of assay medium using the on-board liquid handling system of the Phenix. For the efflux assay, cells were equilibrated with probes (400 nM for 1 and 1 μM for 2 and 3) and BIM1 (2 μM) in assay medium (100 μL) for 2 h, 75 μL medium was removed, and to the remaining 25 μL of medium was added fresh assay medium (200 μL) containing BIM1 (2 μM) using the on-board liquid handling system to dilute the probe. Cells were imaged with a 20X water-immersion objective using four optical channels: bright field, digital phase contrast, mVenus (Ex. 488 nm, Em. 515–550 nm) and Pacific Blue (Ex. 405 nm, Em. 435–480 nm). For all kinetic assays, each probe was analyzed individually in a single well on a single plate. Pre-reading of 10 time points at 60 second intervals was followed by addition of the probe or medium containing BIM1 (2 μM) to a single well (7 seconds) and subsequent acquisition of images from six different optical fields every 60 seconds for up to 180 min. Image analysis used Harmony software (Perkin Elmer). Cells were selected by Find Cells (Digital Phase Contrast images with method P) and further selected by cell area (<1500 μm²). Selected cells were categorized by the fluorescent intensity (contrast) of mVenus as 4 groups: Ppl 1, Ppl2, Ppl3, and Ppl4, corresponding to increasing mVenus expression levels. The fluorescence intensity (contrast) of the Pacific Blue in cells with high expression of mVenus (Ppl3) in selected cells from different populations was analyzed as a function of time using TIBCO SpotFire (PerkinElmer). Further non-linear curve fitting with an exponential one-phase association or decay model (GraphPad Prism) was used to determine t_1/2 values.

Transient transfection of HEK293T cells and preparation of cells for determination of cellular K_d and K_i values

For analysis by flow cytometry, cells were plated on a 6-well plate at 500,000 cells/mL (2 mL) and incubated at 37 °C (5% CO₂) for 16 h to promote adherence (confluence = 90%). Plasmid DNA (1.5 μg) and X-tremeGENE HP (3 μL) were added to Opti-MEM I Reduced Serum Medium (150 μL) in a sterile Eppendorf tube (1.5 mL). The mixture was pipetted gently to mix and maintained at room temperature for 30 min to allow complex formation. To each well of the 6-well plate was added 150 μL of the transfection solution. The 6-well plate was returned to the incubator and incubated at 37 °C (5% CO₂) for 24 h. For subsequent treatment with compounds and analysis by flow cytometry, the medium was removed by vacuum aspiration and trypsin-EDTA was added (0.25%, 1 mL/well, for up to 5 min at room temperature). After the cells detached from the bottom of the plate, 4 mL of culture medium containing 10% FBS was added to quench the trypsin. The cells were resuspended with a pipette, transferred into 15 mL tubes, and pelleted by centrifugation (2000 RPM, 2 min). The supernatant was removed by vacuum aspiration, cell pellets were resuspended in assay medium (1 mL), mixed well using a 1 mL pipette, and additional assay medium (4 mL) was added and mixed well in closed tubes. To evaluate transfection efficiency and cell density, an aliquot of these cells (150 μL) was analyzed by flow cytometry and cell density adjusted to ~500,000 cells/mL using assay medium. These cells were used as 2X stock solutions of transiently transfected cells for compound treatment and flow cytometry. Concentrations of PKCs fused to mVenus and expressed independently using IRES-mVenus were analyzed by flow cytometry as shown in Table S1 (Supporting Information).

Determination of equilibrium cellular K_d values of probes and cellular K_i values of non-fluorescent competitors.

For all cellular binding assays, 2X stock solutions of transiently transfected cells in suspension, prepared as previously described (75 μL), were added to 2X stock solutions (75 μL) of compounds on 96-well plates. These plates were gently shaken for 2 min to mix prior to equilibration for 2 h at 37 °C (5% CO₂) followed by analysis of samples in duplicate by flow cytometry without further wash steps at 22 °C. Absorbance-normalized stock solutions were used to prepare fluorescent probes as 1 mM stocks in DMSO from dry powders. For determination of cellular K_d values using saturation binding assays, 12-point dose response curves were generated with 2-fold dilutions, where the highest final concentrations of fluorescent probes in assay medium (DMEM with 4% FBS) containing cells was 5 μM, but maximal top concentrations of 0.625 μM (1.25 μM for PKCθ) or lower were used for curve-fitting to the minimize the effects of probes on living cells. Serial 2-fold dilutions of 200X stocks of probes in DMSO were prepared in 0.2 mL tube strips to provide final concentrations of probes of 5 μM (dilution 1) to 9.8 nM (dilution 10), with additional 0.98 nM (dilution 11), and DMSO alone (1% DMSO, 0 nM probe, dilution 12) concentrations. On 96-well assay plates, 0.75 μL of these 200X serial dilutions of probes were aliquoted using a 12.5 μL multichannel digital pipette. To prepare 2X solutions of compounds, assay medium (74 μL) containing 2X BIM1 (4 μM) was added to the assay plate containing 200X stocks of probes in DMSO (0.75 μL). This plate was shaken for 2 min to mix the compounds well and incubated at 37 °C (5% CO₂) for an additional 10–20 minutes for further equilibration prior to addition of an equal volume of assay media containing transiently transfected cells. For determination of cellular K_i values using competition assays, a 10 mM stock solution of PDBu (LC laboratories #P-4822) in DMSO (1 mg in 198 μL DMSO) was diluted to 2 mM in DMSO. A 400 μM stock solution of bryostatin 1 (Sigma Aldrich #B7431) in DMSO (0.01 mg in 27.5 μL DMSO) was prepared. These solutions were serially diluted 3-fold with DMSO in 0.2 mL tube strips until dilution 11, where DMSO only was prepared as dilution 12. These 200X concentrated serial dilutions were aliquoted (0.75 μL) to a 96-well assay plate using a 12.5 μL multichannel digital pipette. Separately, binding assay medium with 2X concentrations of probe 1, BIM1, and competitors was prepared from 400X stock solutions in DMSO (e.g. 160 μM probe 1 and 800 μM BIM1 for assays with 400 nM probe 1). Assay medium (74 μL) containing 2X probe 1 (0.8 μM for assays with 400 nM probe 1) and 2X BIM1 (4 μM) were added to a 96-well assay plate seeded with the 200X solutions (0.75 μL) of PDBu or bryostatin 1 in DMSO. This plate was shaken for 2 min to mix the probes well and the samples were allowed to further equilibrate at 37 °C (5% CO₂) for 10–20 min. These assay plates, containing 75 μL/well (2X PDBu or bryostatin 1, 2X probe 1 (0.8 μM for assays with 400 nM probe 1), and 2X BIM1 (4 μM) in 2% DMSO) were removed from the incubator and a 2X solution of transiently transfected cells in suspension (75 μL) was added. This plate was gently shaken for 2 min to mix, followed by equilibration for 2 h at 37 °C (5% CO₂), without further wash steps prior to analysis. For analysis by flow cytometry, the final cell density was 250,000 cells/mL, the final [DMSO] was 1%, and the final [BIM1] was 2 μM, unless otherwise noted. Live cells were gated based on FSC-A and SSC-A guided by independent propidium iodide staining experiments as shown in Figure S6 (supporting information). The approximately bimodal population of transiently transfected and non-transfected living cells (FITC channel) was analyzed to measure total binding, gated as the top 20% of cells with the highest green fluorescence (P2 gate shown in Figure 5), and non-specific binding, gated as the bottom 20% of cells with the lowest green fluorescence (P1 gate shown in Figure 5). Every well of the plate was gated individually. Median PB450 blue fluorescence values for these highly transfected and non-transfected gates were plotted against the concentration of the fluorescent probe or competitor. To calculate the probe dissociation constant (K_d), a one-site total and non-specific binding model was used (GraphPad Prism). Total and non-specific equilibrium saturation binding curves for PB-Phorbol (1) binding to PKC-mVenus fusion proteins and native PKC isozymes in living HEK293T cells are shown in Figure S3 (Supporting Information). K_d values were calculated from curve fitting of binding isotherms using maximum concentrations of 1.25 μM (for PKCθ), 0.625 μM, or lower for probe 1. These values were chosen based on using the lowest top concentration of probe 1 that reliably provided at least 50% saturation of binding sites. For some PKCs, greater variance was observed when higher concentration data points were included, presumably due to the high biological activity of the probe. K_i values of PDBu and bryostatin 1 were measured using a fixed probe concentration within 2-fold of the cellular K_d of the probe measured for each specific PKC isozyme using a One site - Fit K_i model (GraphPad Prism).