Optical control of MAP kinase kinase 6 (MKK6) reveals that it has divergent roles in pro-apoptotic and anti-proliferative signaling

The selective pressure imposed by extrinsic death signals and stressors adds to the challenge of isolating and interpreting the roles of proteins in stress-activated signaling networks. By expressing a kinase with activating mutations and a caged lysine blocking the active site, we can rapidly switch on catalytic activity with light and monitor the ensuing dynamics. Applying this approach to MAP kinase 6 (MKK6), which activates the p38 subfamily of MAPKs, we found that decaging active MKK6 in fibroblasts is sufficient to trigger apoptosis in a p38-dependent manner. Both in fibroblasts and in a murine melanoma cell line expressing mutant B-Raf, MKK6 activation rapidly and potently inhibited the pro-proliferative extracellular signal–regulated kinase (ERK) pathway; to our surprise, this negative cross-regulation was equally robust when all p38 isoforms were inhibited. These results position MKK6 as a new pleiotropic signal transducer that promotes both pro-apoptotic and anti-proliferative signaling, and they highlight the utility of caged, light-activated kinases for dissecting stress-activated signaling networks.

The selective pressure imposed by extrinsic death signals and stressors adds to the challenge of isolating and interpreting the roles of proteins in stress-activated signaling networks. By expressing a kinase with activating mutations and a caged lysine blocking the active site, we can rapidly switch on catalytic activity with light and monitor the ensuing dynamics. Applying this approach to MAP kinase 6 (MKK6), which activates the p38 subfamily of MAPKs, we found that decaging active MKK6 in fibroblasts is sufficient to trigger apoptosis in a p38-dependent manner. Both in fibroblasts and in a murine melanoma cell line expressing mutant B-Raf, MKK6 activation rapidly and potently inhibited the pro-proliferative extracellular signal-regulated kinase (ERK) pathway; to our surprise, this negative cross-regulation was equally robust when all p38 isoforms were inhibited. These results position MKK6 as a new pleiotropic signal transducer that promotes both pro-apoptotic and anti-proliferative signaling, and they highlight the utility of caged, light-activated kinases for dissecting stress-activated signaling networks.
Cells respond to their dynamically changing and chemically diverse surroundings through highly regulated, intracellular signal transduction networks. Some cellular responses, such as growth and cell cycle progression, may be viewed as shortterm, whereas responses such as terminal differentiation and programmed cell death (apoptosis) may be viewed as decisive and cell fate-determining, with limited or no plasticity. Core modules in signal transduction networks include the mitogenactivated protein kinase (MAPK) cascades, notably the extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase (JNK), and p38 pathways in mammals (1,2). In each of these cascades, three kinases are sequentially activated: a serine/threonine kinase classified as a MAPK kinase kinase (MAPKKK or M3K) phosphorylates and activates a dual-specificity MAPK kinase (MAPKK or M2K), which in turn phosphorylates and activates the MAPK. Whereas the ERK cascade controls cell proliferation in response to growth factor stimulation (3,4), JNK and p38 are considered stress-activated protein kinases (SAPKs) that commonly promote apoptosis and other responses to environmental insults and inflammatory cytokines (5)(6)(7). Consistent with these opposing roles that influence normal tissue development, homeostasis, and immune responses, aberrations in MAPK signal transduction are found in the vast majority of human cancers and are broadly implicated in tumorigenesis (8,9).
Parsing the roles of specific nodes in a signal transduction network, many of which are protein kinases, is plagued by multiple challenges that are often unacknowledged, because traditional methods frequently fail to address them. For one, cytokines and environmental stressors generally activate multiple pathways simultaneously, making it difficult to isolate the effect of a particular signaling protein. Adding to the difficulty of this issue are the natural interactions between canonical pathways (i.e. cross-talk). For example, the MAPK cascades affect one another by influencing the expression of dual-specificity phosphatases that dephosphorylate and thus deactivate the various MAPKs with differing specificities (10,11). Another, more specific example is found in the regulation of ERK signaling by p38. This interaction was first suggested by the evidence that ERK1/2 co-immunoprecipitates with p38 (12); activation of p38 by constitutively active MKK3 resulted in potent inhibition of ERK and its upstream MAPKK, MEK. In another study, arsenite treatment induced p38 activation and prevented phorbol ester-triggered activation of MEK and ERK, and this inhibitory effect was attributed to enhanced activity of protein phosphatase 2A (PP2A) (13). Activation of p38 induces dephosphorylation of MEK and apoptosis in various cell types (14,15).
In such studies, pharmacological inhibitors are commonly used to help deduce a protein's function(s), but such an approach only probes loss of function and often lacks complete kinase specificity (16). To target gain-of-function, cells are typically transfected with a constitutively active variant of the signaling protein, but this approach allows ample time for the cell to alter its gene expression profile. This issue is especially problematic for a pathway that mediates stress responses, as its activation can introduce selective pressure to adapt; obviously, this issue is further magnified for pathways that directly and rapidly promote cell death, such as the p38 signaling cascade (17,18).
Various strategies have been implemented to address the limitations outlined above, most commonly by inducing the association or dissociation of a protein-protein interaction via the addition of a small-molecule dimerizer or through light exposure (19)(20)(21)(22). Typically, this strategy involves the translocation of a signaling protein to the plasma membrane to enhance its activity or the release of a protein from a sequestering interaction (23)(24)(25)(26). Although induction or release of membrane This article contains supporting information. * For correspondence: Jason M. Haugh, jason_haugh@ncsu.edu. association is often sufficient to promote signal transduction, it is not a definite "on" switch for catalytic activity. In this study, we develop and apply a photodecaging strategy to rapidly introduce active MKK6, a MAPKK specific to the p38 pathway (27,28), in cells. This optogenetic approach, site-directed unnatural amino acid mutagenesis for the introduction of light-removable protecting groups-so-called "caging groups"-has enabled the development of genetically encoded caged kinases (29)(30)(31)(32)(33)(34), phosphatases (35), and other light-activatable proteins (34,36). With the ability to rapidly generate active MKK6 in cells, we have studied the dynamics of the p38 pathway in relation to p38-mediated apoptosis and the cross-regulation of the ERK pathway. In normal fibroblasts, we found that MKK6 activation is sufficient, and that p38 activation is necessary, for triggering the various phases of apoptosis (caspase activation, cytochrome c release, and cell death) within hours of light-triggered decaging. Consistent with reports on apoptotic responses to inflammatory cytokines and environmental stressors (37)(38)(39), the timing of the critical cytochrome c release varied widely in the cell population. MKK6 activation also rapidly induces potent inhibition of the ERK pathway; surprisingly, we discovered that this negative cross-talk is essentially p38-independent. This finding was replicated in a murine melanoma line with mutant B-Raf, and optical triggering of MKK6 in these oncogene-addicted cells abrogated ERK-dependent cell division. These results position MKK6 as a pleiotropic signal transducer that both promotes cell death and inhibits pro-proliferative signaling, and they demonstrate the utility of caged kinases as unique tools for dissecting signaling networks-stress-activated pathways in particular.

Results
Genetically encoded, caged MKK6 isolates kinase-dependent signaling through optical control To parse the functions of a stress-activated signaling pathway in cells, we developed a genetically encoded, optically con-trolled MKK6. The incorporation of the caged lysine (CK) (Fig.  1A) through site-directed, unnatural amino acid mutagenesis has enabled the development of light-activatable proteins. This approach utilizes an orthogonal aminoacyl-tRNA synthetase/ tRNA pair to selectively incorporate an unnatural amino acid (UAA) into proteins, in response to a recoded UAG amber codon, introduced at a desired site into a gene of interest. Supplementing the protein biosynthetic machinery of cells and animals with an engineered Methanosarcina barkeri pyrrolysyl-tRNA synthetase (pylRS) and its cognate tRNA CUA (pylT) enables the incorporation of a wide range of UAAs, including photocaged amino acids (36). Photocaged amino acids, such as the CK (31), contain a light-removable protecting group that imposes steric demand on an enzyme active site and alters the electronic and nucleophilic characteristics of amino acid side chains. This renders the protein of interest inactive until a brief light exposure (e.g. 365-nm irradiation) removes the caging group and generates the protein's WT catalytic site. This is a generalizable approach for the study of protein kinases (32,40,41), because of the conserved lysine residue that coordinates ATP and catalyzes phosphotransfer in almost all protein kinases (42). Site-specific incorporation of CK at the conserved Lys 82 of constitutively active MKK6 (43) renders the kinase enzymatically inactive, until light exposure removes the caging group (Fig. 1B). Expression of caged MKK6 in mammalian cells was achieved by introducing pylRS, eight expression cassettes encoding pylT, and constitutively active MKK6-K82TAG (referred to hereafter as caged MKK6) from two plasmids (Fig. 1C). In HEK293T cells cotransfected with the plasmids and supplemented with or without 2 mM CK, caged protein expression was detected only in the presence of the UAA (Fig. 1D).
MKK6 is a dual-specificity kinase that activates p38 by phosphorylation ( Fig. 2A) (27,44). To confirm the optically controlled signaling function of caged MKK6, NIH 3T3 fibroblasts were co-transfected with plasmids expressing caged MKK6 and a plasmid encoding a p38 kinase activity reporter construct, which is exported from the nucleus upon phosphorylation (45), Figure 1. Construction of genetically encoded caged MKK6. A, structure of the photocaged lysine (CK), with the light-removable caging group shown in red. B, caging of Lys 82 in MKK6 blocks the ATP-binding site until the caging group is removed through UV exposure, restoring the native, constitutively active kinase (Protein Data Bank entry 3FME). C, two-plasmid system to express caged MKK6 in mammalian cells. The TAG mutation is introduced at the Lys 82 position of MKK6. Two other mutations (S207E and T211E) transform the kinase into a constitutively active form. pylRS for CK (CKRS) is driven by a CMV promoter, and FLAG-caMKK6 K82TAG is driven by an EF1a promoter. U6, an RNA polymerase III promoter, controls the expression of pylT on both plasmids in the form of four (43) repetitive cassettes. CMVe (CMV 59 enhancer fragment) is coupled with each U6 promoter on one of the plasmids for higher pylT yield. D, Western blotting shows CK-dependent expression of caged MKK6 in HEK 293T cells. and the transfected cells were supplemented with or without CK (2 mM). After washing the cells and replacing the medium with imaging buffer, reporter localization was monitored by live-cell epifluorescence microscopy. In cells expressing caged MKK6 and incubated with CK, a brief (,2-s) exposure to UV light (DAPI excitation filter) rapidly induced p38 kinase activity, as shown by the live-cell reporter (45) (Fig. 2, B and C), confirming that MKK6 photoactivation and activation of p38 (the canonical MKK6 substrate) was substantial. No such UVinduced translocation responses were observed in the negative controls: cells incubated without CK (Fig. 2C) or in cells incubated with CK but expressing mCherry in place of constitutively active MKK6 (Fig. S1, A-C). These controls indicate that p38 is not activated by a brief UV exposure alone or by decaging of lysines that might have been incorporated in an inert protein or the endogenous proteome. The translocation response in caged MKK6-expressing cells was promptly and completely reversed following the addition of the p38a/b-specific inhibitor SB 239063 (46) (10 mM; Fig. 2D). The extent and kinetics of the p38 reporter response elicited by MKK6 decaging were comparable with those elicited by the chemical stressor, anisomycin (Fig. 2E). We also confirmed that brief UV exposure does not induce significant translocation of the analogous JNK activity reporter, either with or without expression of caged MKK6  Fig. S2. Taken together, these results establish an experimental system in which canonical MKK6 ! p38 signaling is readily and specifically induced by a brief and benign exposure to low-energy UV light.

Decaging of MKK6 induces p38-dependent apoptosis
Stress-activated signaling pathways orchestrate pronounced cellular responses, such as autophagy and apoptosis, which are difficult to study in traditional genetics experiments. JNK and p38 signaling have been strongly implicated in apoptosis of normal and cancer cells through a variety of mechanisms (5)(6)(7)(47)(48)(49)(50). We asked whether an acutely active MKK6 upstream of p38 is sufficient to switch on the apoptotic program. Within hours after brief UV exposure of serum-starved fibroblasts coexpressing caged MKK6 and the p38 activity reporter, morphological changes associated with apoptosis, namely shrinkage, blebbing, and nuclear condensation or fragmentation (51, 52), with plasmids to express caged MKK6 and p38KTRClover, serum-starved, and briefly exposed to UV light (DAPI excitation filter, 365 nm, ,2 s) as indicated. B, time-course montage of representative cells, before and after MKK6 light activation. Scale bar, 10 mm. C, quantification of p38 activity in cells grown in medium supplemented with or without CK (2 mM) (n = 10 each). D, cells (n = 12) were briefly exposed to UV light and then treated with the p38a/b-specific inhibitor SB 239063 (10 mM). E, analysis of NIH 3T3 cells (n = 27) transiently expressing p38KTRClover and treated with anisomycin (50 ng/ml). F, NIH 3T3 cells (n = 11) coexpressing caged MKK6 and the JNK activity reporter JNKKTRClover were briefly exposed to UV light and then treated with anisomycin (50 ng/ml). The solid lines in C-F show the mean kinase translocation reporter translocation, and the shaded regions show the 95% confidence intervals. All data are representative of at least three independent experiments.
were observed in a significant fraction of cells ( Fig. 3A and Movie S1). Although the timing of cell death varied greatly, as seen in other studies (39,53), the tendency of each cell to undergo apoptosis correlated well with the intensity and duration of p38 activity (Fig. 3B). To quantify apoptotic responses more precisely, we monitored the release of fluorescent protein-tagged cytochrome c from mitochondria, a hallmark of apoptosis (54,55). This event manifests as an abrupt transition from punctate to diffuse fluorescence ( Fig. 3 (C and D) and Movie S2). Optical activation of MKK6 induced cytochrome c release in a substantial fraction of the cells within ;5 h; in contrast, almost none of the cells responded within that time period when CK was not supplied (Fig. 3E). Whereas detection of cytochrome c release stereotypically lagged MKK6 decaging by hours, we observed accumulation of a fluorescent cleavage product reporting caspase-3/-7 protease activity (56) much earlier (Fig. S3).
To assess the role of p38 in the apoptotic response to MKK6 decaging, we pretreated cells with SB 239063, the pan-p38 in-hibitor BIRB 796 (57), or a DMSO vehicle control prior to UV exposure, and cytochrome c localization was monitored. For each cell, the time of the punctate-to-diffuse transition was recorded, and cumulative distributions of cytochrome c release were constructed. Control cells were not transfected with caged MKK6 but were incubated with CK as usual. The results show that p38 inhibition delayed the onset of cytochrome c release, almost to the same extent as in cells lacking caged MKK6 (Fig.  3F). Collectively, these results demonstrate that light activation of MKK6 through active site decaging is sufficient to trigger the hallmarks of apoptosis-caspase activation, release of cytochrome c, and cell morphology changes-and that the response is largely, if not completely, p38-dependent.
Photoactivation of MKK6 reveals p38-independent cross-regulation of ERK The optogenetic system developed here also affords the opportunity to elucidate how individual kinases influence crosstalk between signaling pathways. MKK6 signaling is thought to flow exclusively through activation of p38, which has been implicated in negative cross-regulation of the Raf/MEK/ERK kinase cascade (12,13,15). To assess this pathway dynamically, we used an ERK kinase translocation reporter (45). A robust translocation response was observed in serum-starved NIH 3T3 cells stimulated with platelet-derived growth factor-BB (PDGF; 1 nM), and the response was reversed by the addition of the MEK inhibitor U0126 (20 mM) (Fig. 4 (A and B) and Movie S3). To assess pathway cross-talk, we co-expressed caged MKK6 and the ERK reporter. Cells were incubated with a moderate concentration of serum (3% fetal bovine serum (FBS)) to foster basal ERK activity; upon optical decaging  Optical control of MKK6 signaling of active MKK6, the ERK reporter was rapidly imported to the nucleus, indicating an acute reduction of ERK kinase activity and net dephosphorylation of the reporter construct.
After the subsequent addition of PDGF, ERK activity was only partially and transiently restored ( Fig. 4C and Movie S4). Using compartment-specific FRET biosensors (58), we confirmed that MKK6 decaging reduced ERK activity in both the nucleus and the cytosol (Fig. S4). When cells were not supplied with CK, or in cells incubated with CK but expressing mCherry in place of MKK6-K82TAG, UV exposure had no discernible effect on ERK activity, whereas the subsequent addition of a MEK inhibitor dramatically reduced ERK reporter translocation, as expected (Fig. S5). In a different stimulation protocol, the cells were serum-starved and then stimulated with PDGF to elicit ERK activation. In this context, decaging of active MKK6 also reduced ERK reporter translocation (Fig. 4D).
To further test this newly discovered role of MKK6 in the inhibition of ERK signaling through pathway cross-talk, we performed the MKK6-decaging experiment and subsequently added the p38 inhibitor SB 239063. Surprisingly, no significant recovery of ERK activity was observed; inhibition of p38 did not even partially reverse the negative regulation of ERK induced by light activation of MKK6 (Fig. 5 (A and B) and Movie S5). Faced with this evidence, we considered the possibility that the negative regulation of ERK could be p38-dependent, but inherently irreversible. To rule this out, we performed the experiment with the treatments in reverse order. With p38 inhibited by SB 239063, we observed a significant increase in ERK activity, consistent with the known crossregulation (59); however, even with p38 activity blocked, subsequent decaging of MKK6 abrogated ERK activity as before (Fig. 5C). We repeated these experiments using the pan-p38 inhibitor BIRB 796, at a concentration (10 mM) reported to block all p38 isoforms (57) and obtained similar results (Fig.  S6). These findings establish that active MKK6 rapidly and potently abrogates the ERK pathway in a p38-independent manner, in parallel with the previously established p38-dependent mechanisms (Fig. 5D).
To corroborate the results obtained with caged MKK6 and its dependence on MKK6 activity, we co-expressed the ERK reporter with various MKK6 variants (43) and assessed ERK activity in the presence of 3% FBS (Fig. 6). Whereas expression of a kinase-dead MKK6 (K82A) had no apparent effect on ERK activity, expression of WT MKK6 or constitutively active MKK6 (S207E/T211E) diminished ERK reporter translocation whether or not p38 was inhibited (BIRB 796; Fig. 6A). Quantification of these data show that these comparisons are significant (Fig. 6B). Immunoblotting was performed to confirm that transfection of kinase-competent MKK6 yielded phosphorylation of p38 and of the p38 substrates, ATF-2/7, and that the latter response was blocked by BIRB 796 (Fig. S7). For the aforementioned conditions where ERK was inhibited, PDGF stimulation elicited little or no increase in ERK signaling in the majority of the cells (Fig. S8).
Whereas introduction of active MKK6 clearly regulates ERK kinase activity, as observed through the application of multiple biosensors, we did not find evidence implicating  perturbation of the Raf/MEK/ERK phosphorylation cascade. MKK6 decaging did not perturb the nuclear localization of a GFP-tagged ERK2 construct, and transfection of kinaseactive MKK6 did not elicit a reduction in MEK or ERK phosphorylation measured by immunoblotting (Fig. S9).

Light-activated MKK6 inhibits ERK signaling and abrogates mitosis in oncogene-addicted cells
To confirm that MKK6-mediated regulation of ERK is not cell line-specific, we repeated the experiments of MKK6 decaging followed by p38 inhibition (as in Fig. 5 (A and B)) in serumstarved PBT-2460 cells, a genetically engineered murine melanoma line with constitutively elevated ERK signaling due to expression of active B-Raf (V600E) (60,61). As seen in untransformed fibroblasts, light-activated MKK6 inhibited ERK signaling in PBT-2460 cells, even with p38 inhibited (Fig. 7 (A and B) and Movie S6; see Fig. S10 for the corresponding controls).
Unlike untransformed cells, PBT-2460 cells proliferate in low serum (3% FBS), but they are addicted to V600E B-Raf signaling; their proliferation is abrogated by inhibition of MEK ( Fig.  7C and Fig. S11). In the same set of experiments, PBT-2460 proliferation was modestly reduced by inhibition of p38, consistent with reports that p38 is required for efficient G 2 -M transition (62). In live-cell imaging experiments, PBT-2460 cells with stable expression of the ERK kinase translocation reporter and transfected with caged MKK6 were irradiated and monitored for ;16 h. When the cells had not been incubated with CK beforehand, none exhibited a detectable decrease in ERK activity upon UV treatment, and 58% of the observed cells divided (Fig. 7D). In cells that had been incubated with CK, 22% of the cells exhibited a rapid drop in ERK activity following UV treatment (consistent with the transfection efficiency for PBT-2460). ERK activity remained depressed in those cells, and none of them divided, during the 16-h post-light exposure. Among the cells in this cohort that did not show a rapid drop in ERK activity, a substantial fraction divided ( Fig. 7E and Movie S7). The same relationship between ERK signaling and mitosis was observed when the experiment was performed in the presence of the p38 inhibitor BIRB 796, except with somewhat reduced overall proliferation consistent with the results shown in Fig.  7C (Fig. 7F). These observations confirm that p38-independent regulation of ERK signaling by rapid introduction of active MKK6 prevents mitosis of cells addicted to the ERK pathway. As a whole, this study demonstrates how optical control of kinase function through unnatural amino acid mutagenesis with photocaged lysine can be used to study the dynamic regulation of normal and dysfunctional signaling networks.

Discussion
Protein kinases form the backbone of signal transduction networks, and several genetic and proteomic approaches have been developed and refined to study their regulation and function (63)(64)(65). Considering that signaling networks are rife with cross-talk and feedback interactions, isolating the function of a particular signaling protein is challenging. The general, genetically encoded photodecaging approach applied here to MKK6 allows for introduction of a particular kinase activity in an otherwise quiescent background in a dynamic fashion. It combines the complete specificity of genetic tools with the temporal precision of pharmacological inhibitors. In contrast to the latter, abrupt gain-of-function is achieved, enabling studies that are otherwise difficult to perform. Given that MKK6 phosphorylates and thus activates the p38 SAPKs, which are functionally associated with apoptosis in many cell types, the abrupt introduction of activity is absolutely necessary, because the longterm activation of stress-activated signaling inherently forces selection and/or adaptation within the cell population. Indeed, in cells transfected with either WT or constitutively active MKK6 (;60% transfection efficiency), we found that the cells initially survived in high-serum growth medium, but after 48 h, the cell population showed substantial cell death relative to empty-vector control (Fig. S12).
In the context of apoptosis, simple transfection of active MKK6 or p38 only allows one to evaluate overall cell death; it is The solid line shows the mean kinase translocation reporter translocation, and the shaded regions show the 95% confidence intervals. C, proliferation assay of murine melanoma PBT-2460 cells in low-serum (3% FBS and essential and nonessential amino acids) medium containing DMSO vehicle (0.1%, v/v), MEK1/2 inhibitor U0126 (10 mM), or either p38 inhibitor BIRB 796 (10 mM) or SB 239063 (10 mM). Proliferation is quantified as the -fold change in cell density after 48 h, and the error bars show mean 6 S.D. (n = 3 independent experiments); ****, p , 10 24 ; **, p , 0.01 (unpaired t test). D, PBT-2460 cells stably expressing ERKKTRClover were transiently transfected with caged MKK6, but CK was withheld. After 24 h, cells were switched to imaging buffer supplemented with nonessential amino acids and DMSO vehicle control (0.1% v/v), irradiated with UV on our microscope, and monitored for ;16 h. Cells were classified according to whether they did/did not show translocation of the ERK reporter following UV treatment and whether they did/did not divide during the experiment, and the numbers and percentages of cells are tabulated (pooled from three independent experiments). E, same as D except that 2 mM CK was supplied following transfection. F, same as E except the imaging buffer contained the pan-p38 inhibitor, BIRB 796 (10 mM). not possible to study the dynamics of pre-and post-mitochondrial apoptotic processes, nor is it possible to cleanly isolate the effect of a single kinase by incubating cells with pro-apoptotic factors or stressors. Here, we found that decaging active MKK6 in a quiescent background was sufficient to induce a complete and kinetically staged apoptotic response, which required the activation of endogenous p38. The early activation of p38 following MKK6 decaging was accompanied by cleavage of a caspase-3/7 biosensor. This early emergence of executioner caspase activity is consistent with the reported ability of p38 to mediate activation of the initiator caspase-8, either in concert with or independent of death receptor signaling (66,67).
Loss of mitochondrial integrity, marked by release of cytochrome c into the cytoplasm, is a pivotal occurrence during apoptosis that was typically observed within several hours of MKK6 activation through light-induced decaging. The timing of this event varied broadly in the cell population, consistent with natural apoptotic responses to pro-death signals and environmental insults. The aforementioned induction of caspases-3 and -7 likely contributes to the late-stage response, considering that recombinant caspase-3 is sufficient to induce cytochrome c release from isolated mitochondria, accompanied by cleavage of the anti-apoptotic regulator, Bcl-2 (55). In parallel, p38mediated inhibition of Bcl-2 and the related Bcl-xL (68) and/or potentiation of the pro-apoptotic regulators Bax and Bim (69,70) might also contribute to late-stage loss of mitochondrial membrane potential and release of cytochrome c.
Our experiments also showed that after a reasonably consistent p38 activation response following MKK6 decaging (Fig.  2C), cells typically showed a further, sudden increase in p38 activity after a variable lag (Fig. 3B). This delayed increase in p38 activity reporter translocation was not due to collapse of the nuclear envelope, and it was often transient. The phenomenon has hallmarks of an excitable system and could be driven by positive feedback, which might or might not be intertwined with the apoptotic cascade. For example, activation of caspase-3 has been implicated in p38 activation through cleavage of the upstream kinase MEKK1 to produce a constitutively active, truncated form (71,72). Another pivotal player in the p38 and apoptotic signaling networks is the tumor suppressor p53 (73,74), which responds to p38-catalyzed phosphorylation (75,76) and can enhance p38 activity through increased reactive oxygen species production (77). Future studies will reveal the dynamics of these intertwined networks; the general approach demonstrated here offers a systematic way to do so.
Decaging MKK6 also rapidly and potently abrogated ERK activity, theretofore maintained by a moderate concentration of serum. Although negative cross-talk among MAPK cascades has been described for some time (12,78,79), and despite strong indications that p38 enhances the activity of PP2A as a regulator of the Raf/MEK/ERK pathway (13, 80), we unexpectedly found that acute activation of MKK6 rapidly mediates the inhibition of ERK1/2 in a p38 kinase activity-independent fashion (as demonstrated through conditional blocking of p38 activity). In the context of B-Raf-transformed cells, which are reliant on ERK signaling for proliferation, this regulation blocked cell division. Because our understanding of MKK6 function is currently limited to its activation of p38 kinase activities by phosphorylation, these results force a broader view of MKK6 as a signaling branch point affecting pro-death and anti-proliferative pathways by mechanisms that are apparently different from those previously thought. We assert that a deeper investigation of the mechanisms by which MKK6 regulates the ERK pathway will be necessary, as might be a general re-examination of MAPKK functions.
Regarding the mechanism of ERK1/2 regulation, we did not detect significant MKK6-mediated inhibition of the upstream pathway at the level of ERK localization or MEK/ERK phosphorylation. Although stasis of ERK localization or MEK/ERK phosphorylation can belie changes in the pathway input (81), we are inclined to consider other mechanisms. An intriguing observation in that regard is that an MKK6-based docking-site peptide inhibits in vitro ERK2 activity with potency comparable with that of MEK1/2-based peptides (82). If active MKK6 were able to sequester ERK1/2, presumably MKK6 would be in competition with MEK1/2; however, protein scaffolds may bias MEK-ERK interactions toward inactive ERK (83), indirectly biasing the hypothetical MKK6-ERK interactions toward active ERK.
Our results using caged MKK6 illustrate how acute and precise activation of a single kinase can affect signaling and cell response outcomes. MKK6 activation blocked pro-proliferative signaling while promoting apoptosis, suggesting that MKK6 orchestrates a definitive switch in cell priorities and fate. Such cell fate switches are likely to vary across contexts of cell type and environmental conditions and are critically important for understanding outcomes of targeted cancer therapies, for example. Here, in the context of cancer, we have scratched the surface of the application of genetically encoded, caged kinases, which could be used to probe dynamics of signaling networks with aberrant or reprogrammed kinomes (84).

Materials and methods
Cell culture and other commercial reagents HEK293T cells and NIH 3T3 mouse fibroblast cells were acquired from American Type Culture Collection. PBT-2460 tumor cells (60, 61) were a gift from James E. Bear (University of North Carolina, Chapel Hill, NC, USA). Cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% FBS, 2 mM L-glutamine, and the antibiotics penicillin and streptomycin in a 37°C, 5% CO 2 incubator. All cell culture reagents were purchased from Thermo Fisher Scientific/Invitrogen. Transient transfections were performed using Lipofectamine and PLUS reagents (Thermo Fisher Scientific

CK
The photocaged lysine amino acid was synthesized as reported previously (31).
The following live-cell fluorescence reporters were obtained from Addgene: kinase translocation reporters for ERK (pLenti-CMV Puro DEST ERKKTRClover; Addgene no. 59150), p38 (pLentiPGK Puro DEST p38KTRClover; Addgene no. 59152) and JNK (pLentiPGK Blast DEST JNKKTRmRuby2; Addgene no. Transient transfection, live-cell imaging, and image analysis NIH 3T3 cells plated at 50-60% confluence were transiently transfected with Lipofectamine 3000 and P3000 reagent in normal growth medium with no antibiotics, following the manufacturer's protocol. Where applicable, CK (2 mM) was added to the growth medium immediately after the transfection. 24 h after transient transfection, cells were replated on fibronectin (10 mg/ml)-coated glass-bottom dishes (MatTek) and were allowed to spread in the same growth medium for 3-4 h. For serumstarved conditions, the cells were incubated for an additional 4 h in imaging buffer (20 mM HEPES, 125 mM NaCl, 5 mM KCl, 1.5 mM MgCl 2 , 1.5 mM CaCl 2 , 10 mM glucose, and 2 mg/ml fatty acid-free BSA, pH 7.4) supplemented with 0.1% FBS and penicillin-streptomycin-glutamine. The buffer was maintained at 37°C during experiments, and mineral oil was layered on top to prevent evaporation.
Epifluorescence microscopy was performed with a modified microscope (Axioskop 2 FS) equipped with a 50-watt mercury lamp, 340/0.8 NA and 360/1.0 NA Achroplan water-dipping objectives (Carl Zeiss), a Hamamatsu ORCA ER chargecoupled device camera, and MetaMorph image acquisition software (Universal Imaging). The microscope is outfitted with an automated multistage and CRISP autofocus control system (Applied Scientific Instrumentation). Band-pass excitation fil-ters were 350/50 nm for UV decaging, 436/20 nm for Cerulean (FRET reporters), and 488/10 nm for Clover and EGFP constructs. Band-pass emission filters were 480/40 and 535/30 nm for Cerulean and Venus, respectively (FRET reporters) and 525/50 nm for Clover and EGFP constructs. UV irradiation was from the mercury lamp passed through a DAPI filter. The duration of exposure (1-2 s) was controlled using a shutter (Uniblitz). All filters were obtained from Chroma Technology.
The mean intensity of an acellular region of each image was considered as background and was subtracted from the intensity of each pixel before further analysis. For nucleocytoplasmic translocation analyses, nuclear and cytosolic masks were generated for each cell using MATLAB-based Microscopy Image Browser software (87). Mean nuclear and cytosolic intensities were calculated for each frame, accounting for modest photobleaching during the experimental time period. To quantify cytochrome c release, a custom MATLAB code was written to track each individual cell across consecutive image frames, and the entropy (a measure of randomness commonly used to quantify image texture) of each cellular region of each frame was calculated using the entropy function. For FRET biosensors, the cellular regions were segmented from the background, and the ratio of acceptor (Venus) to donor (Cerulean) fluorescence intensities was calculated for each frame. For all live-cell fluorescence measurements, the value for each cell was normalized by the value prior to stimulation. For cell proliferation measurements, equal densities of cells were plated in each dish, and one dish was used to measure the initial cell density; 48 h after treatment, cell density was measured for each condition. Nuclei were stained with Hoechst 33258 (1 mg/ml) before imaging at 310 magnification. Images of 10-15 fields/dish were acquired to estimate cell density. The -fold change in cell density was calculated using average cell density of all imaging fields of each condition relative to that of the initial condition. For the cell division assay, individual cells were tracked using Lineage Mapper (88), which automatically detects cell division events. The data were reviewed by eye to confirm the events.
Immunoblotting SDS-PAGE, wet electrophoretic transfer to polyvinylidene difluoride membranes, and immunoblotting were performed by standard methods. After primary and horseradish peroxidase-conjugated secondary antibody incubations and intermittent washes, SuperSignal West Femto Maximum Sensitivity substrate (Pierce/Fisher Scientific, catalog no. PI34095) was applied, and chemiluminescence was imaged using a digital imaging system (Syngene G:BOX, Synoptics Group).

Data Availability
Data not contained in the paper are available by request from the corresponding author (Jason M. Haugh, jason_haugh@ ncsu.edu).