EAS was designed to have a short peptide linker sequence sensitive to phosphorylation by ERK fused between ECFP and EYFP. ECFP and EYFP were chosen as a FRET pair because of the good spectral overlap between the emission of ECFP and the excitation of EYFP. We hypothesized that phosphorylation would result in changes in secondary structure of the linker peptide that would alter FRET efficiency for the protein (Figure 1A). Since it was unclear which factors would contribute to forming a suitable FRET sensor, we prepared several different constructs with peptide linker sequences containing ERK phosphorylation sites taken from Ets1 and Elk1 transcription factors, which are physiological substrates of the ERK MAPKs (Figure 1B). EAS-2, derived from mouse Ets1, has a rather large linker peptide, due to the discovery that the MAP Kinase binding site (D-domain) is approximately 90 amino acids downstream of the target serine 13. The linker sequences for EAS 3–5 were taken from human Elk1, which has been shown to have the DEF domain (FXFP amino-acid motif), a known MAPK binding site, near the C-terminus. These peptides also contain two Serine-Proline (SP) motifs (consensus MAPK phosphorylation sites) corresponding to serines 383 and 389 of Elk1, which have been shown to be targets of MAP kinases. Phosphorylation of these serines has also been shown to be directed specifically by the DEF domain 14. EAS-Neg is a negative control construct with a short peptide linker derived from Ets1, containing a consensus MAPK phosphorylation site, but no MAPK binding site.

We expected a change in energy transfer between ECFP and EYFP upon incubation with active ERK2 (pERK2) for EAS to be effective as a real time sensor. This would indicate that a phosphorylation event had affected the distance and orientation between the two fluorophores. Each EAS construct was incubated with pERK2 in a fluorimeter (30°C) and readings were taken prior to addition of ATP. Upon the addition of ATP, all EAS constructs, with the exception of EAS-Neg, showed a decrease of energy transfer indicated by an increase in donor (ECFP) emission (475 nm) with a concomitant decrease in acceptor (EYFP) emission (525 nm) after 40 minutes (Figure 2A). Therefore, a decrease in FRET efficiency correlates with phosphorylation of EAS by pERK2. The resultant change in ECFP/EYFP emission ratio is comparable to or exceeds that of similar FRET-based signal transduction sensors 456. Under the same conditions, EAS-Neg showed no change in FRET over the course of 30 minutes, confirming the specificity of ERK2 for EAS 2–5 and the requirement of a phosphorylation event to induce a change in FRET. Figure 2B shows the absolute change in ECFP/EYFP ratio over a 40 minute time course for each construct. EAS-3 and EAS-4 show the greatest absolute decrease in EYFP/ECFP emission ratio, with EAS-5 having an intermediate change in emission ratio. EAS-2 showed very little change in FRET. Phosphorylation was also confirmed by an in vitro kinase ³²P phosphorylation assay using recombinant pERK2 (Figure 2C). EAS 2–5 sensors were efficiently phosphorylated as compared to the non-specific control, EAS-Neg, and Myelin Basic Protein (MBP), a common target of activated MAPKs for in vitro studies. Based on the robust FRET signal change and radioassay results, further studies focused on EAS-3.

Mutants of EAS-3 were utilized to validate the structural features of EAS essential for decreased FRET efficiency upon incubation with pERK2. We refined EAS-3 by replacing the bulky N-terminal GST-purification tag (Glutathione-S-Transferase) with a Histidine-10 tag on the C-terminus. This reduced the possibility that the large purification tag would interfere with FRET efficiency changes. In addition, we mutated alanine 207 and alanine 487 to lysine. As previously reported, this prevents EGFP dimerization 15. This latter modification reduced co-purification of truncation products with full-length EAS (data not shown).

To further characterize EAS-3, we generated different constructs targeting critical residues, and analyzed their ability to serve as substrates for pERK2. Mutants EAS3-S286A, EAS3-S292A, and the double mutant EAS3-S(286,292)A eliminated consensus phosphorylation sites (Figure 2A). The dead binding domain mutant (EAS3-DBD) eliminated the DEF domain by replacing the two key phenylalanines with alanines, creating an AXAP motif (Figure 3A). As Figure 3B demonstrates, all five mutants have decreased phosphorylation compared to the wild type sensor. As expected, the EAS3-S(286,292)A mutant had the lowest level of phosphate incorporation due to absence of both serine-proline consensus phosphorylation sites required by MAP Kinases. A decrease in phosphorylation of EAS-DBD iss also consistent with the requirement of MAPKs to bind targets for efficient phosphorylation (reviewed in 16). Differences between wild-type EAS-3 and mutants are also reflected in change of FRET efficiency when treated with pERK2 in fluorimeter experiments (Figure 3C). These relative changes in FRET efficiency are consistent with the radioassay data.

Distinguishing the activation of different MAP kinases within the cell is essential since each MAPK pathway is activated by multiple mitogens and external environmental factors to varying degrees (reviewed in 17). Furthermore, there is extensive cross-talk between the different MAP kinase pathways. Detection methods must effectively isolate the signal of the target kinase from other family members to elucidate the contributions of these different pathways to a given cellular process.

The target peptide linkers were designed to impart specificity for ERK. Elk1 and Ets1 were chosen because of their seemingly specific interaction with ERK relative to p38 or SAPK/JNK (reviewed in 18). To determine whether EAS-3 acted as a specific substrate for ERK we performed in vitro phosphorylation assays to quantify the ability of pERK2, pp38, and pSAPK/JNK to phosphorylate EAS-3 (Figure 4). As shown in Figure 4A, EAS-3 is an approximately 2000-fold more specific target for pERK2 than pp38, and 50-fold more specific for pERK2 than pSAPK/JNK. Activities of pERK2, pp38 and pSAPK/JNK were confirmed by using known substrates specific for each kinase (Figure 4B). MBP was used in conjuction with pERK2, GST-ATF2 with pp38, and GST-c-Jun with pSAPK/JNK. Specificity of these kinases for EAS-3 was also confirmed by FRET in a fluorimeter (Figure 4C). It is clear that the change in FRET was dramatic when EAS-3 was incubated with pERK2, but showed little or no response when incubated with pp38 or pSAPK/JNK. This is consistent with the results found in 4A. Given the similarity of these kinases, this also suggests that EAS-3 would be specifically acted upon by pERK2 even in the presence of other activated intracellular kinases.

The demonstrated specificity of pERK2 for EAS-3 phosphorylation suggests that this sensor is a candidate for in vivo studies of ERK signaling. However, our preliminary experiments in NIH-3T3 cells indicated that EAS-3 was susceptible to intracellular phosphatase activity (data not shown). We surmise that this is due to the absence of a protective phospho-specific binding domain within the EAS-3 construct. Such binding domains have been crucial in the development of other FRET-based signaling sensors of kinase activity 23467. These binding domains, however, are taken from naturally occurring domains that are specific for each target sequence. Unfortunately, no known phospho-specific binding domain exists for Elk1 in the region of Ser383/389. Therefore, we are using a semi-rational approach to develop a phospho-specific binding domain that acts to protect the phosphorylated EAS-3 linker from intracellular phosphatases. This will enable us to adapt the EAS-3 sensor for use in live cells.

DNA coding for peptide linkers was amplified by PCR with primers designed with Bgl II and BamHI sites at the 5' and 3' ends, respectively. The products were cloned into pECFP-C1 (Clontech), and EYFP from pEYFP-N1 was subsequently cloned in frame to create EAS constructs. EAS's were cloned into pGEX-2T (Amersham) in frame with GST for expression and purification purposes. These GST-tagged constructs were used for initial fluorimetry experiment of EAS constructs. Other versions of EAS's were constructed using ECFP and EYFP with mutation A(207,487)K, which eliminates dimerization of GFPs, and cloned into pET21a (Novagen) with a Histidine-10 tag. These His10-tagged constructs were used for mutant and specificity experiments. EAS-3 linker mutants were made by Quickchange Site Directed Mutagenesis (Stratagene) to make either single or double point mutation within the peptide linker. For expression in NIH-3T3 cells, EAS's were cloned into the pECFP vector backbone without tags.

BL21(DE3) cells were transformed with EAS constructs and plated on LB agar supplemented with 100 μg/ml ampicillin. Colonies were picked into 4 ml LB-Amp and shaken overnight at 30°C. 250 ml LB was inoculated with starter culture, grown to A₆₀₀ of 0.6, and induced with 0.1 mM IPTG for 18 hours. Cells were spun down, lysed by french press, and cell fragments were spun down at 18,000 rpm in a Beckman JA-20 rotor for 30 min. Proteins were purified from lysates over Ni-NTA Superflow (Qiagen) or Glutathione Sepharose 4B (Amersham).

Activated kinases were purified as described 19. pERK2 was either purchased (NEB) or purified from bacteria transformed with a plasmid containing both constitutively active MEK1* and His-tagged ERK2. Phosphorylation of ERK2 by MEK1* occurred in bacteria prior to lysis. BL21(DE3) cells were electroporated with MEK1*/ERK2 construct and grown overnight at 37°C on LB-Amp agar plates. Several colonies were picked and incubated in 4 ml of TB supplemented with 100 μg/ml carbenicillin. The starter culture was added to 1L of TB, grown to A₆₀₀ of 0.35, and then induced with 0.25 mM IPTG for 12 hours at 30°C. Cells were harvested and lysed as above, and purified over Ni-NTA Superflow. Activated p38 and SAPK/JNK were purified from bacteria electroporated with two plasmids, one coding for constitutively active MEK kinase 4 (MEKK4*) and another coding for MEK4 and either His-tagged p38 or His-tagged SAPK/JNK. These constructs were grown and purified as above, except with both carbenicillin and kanamycin (50 μg/ml). Protein concentration and buffer exchange performed with Centriplus, Centricon, and Microcon ultrafiltration membranes (Millipore).

Radioactive kinase assays were performed in 25 μl in 12.5 mM MOPS pH 7.5, 12.5 mM β-glycerophosphate pH 7.3, 7.5 mM MgCl₂, 500 μM NaOrthovanadate, 500 μM NaF, and 9.7 nM DTT. Amounts of kinase and substrate added to reactions are indicated in figure legends. Finally, ATP supplemented with 0.2 nmol [γ-³²P]ATP (6000 mCi/mmol, Molecular Bioproducts), was added to a final concentration of 1 mM. Reactions were incubated at 30°C for 15 minutes, 8.3 μl of 4× protein sample buffer (SB) were added, and samples were boiled for 5 minutes. Samples were analyzed with 10% or 12% SDS-PAGE (BioRad Mini-Protean II) and transferred to a nitrocellulose membrane. The membrane exposed to a phosphoscreen and scanned on a Storm 860 (Molecular Dynamics). Quantitative densitometry of bands was performed using ImageQuant 5.0 (Molecular Dynamics).

Fluorimetry performed in a Shimadzu Spectrofluorophotometer RF-5301PC. Assays were performed in same buffer as above in a volume of 2.5 ml. The final concentration of EAS constructs was 250 nM and final concentration of pERK2 was 50 nM. The reaction was incubated in fluorimeter and warmed to 30°C. Excitation was set at 425 nm to excite ECFP and avoid excitation of EYFP and readings were taken in 0.2 nm increments. An initial reading was taken prior to ATP addition. To initiate reaction, ATP was added to a final concentration of 1 μM, a spinner was activated for 1 minute, and first reading taken at 2 minutes. EYFP to ECFP ratios were calculated by dividing EYFP peak emission by ECFP peak emission. Peak emissions were defined as an average intensity of EYFP between 524.6–525.4 nm and of ECFP between 474.6–475.4 nm.