Cyclic AMP-dependent Protein Kinase Binding to A-kinase Anchoring Proteins in Living Cells by Fluorescence Resonance Energy Transfer of Green Fluorescent Protein Fusion Proteins*

A-kinase anchoring proteins tether cAMP-dependent protein kinase (PKA) to specific subcellular locations. The purpose of this study was to use fluorescence resonance energy transfer to monitor binding events in living cells between the type II regulatory subunit of PKA (RII) and the RII-binding domain of the human thyroid RII anchoring protein (Ht31), a peptide containing the PKA-binding domain of an A-kinase anchoring protein. RII was linked to enhanced yellow fluorescent protein (EYFP), Ht31 was linked to enhanced cyan fluorescent protein (ECFP), and these constructs were coexpressed in Chinese hamster ovary cells. Upon excitation of the donor fluorophore, Ht31·ECFP, an increase in emission of the acceptor fluorophore, RII·EYFP, and a decrease in emission from Ht31·ECFP were observed. The emission ratio (acceptor/donor) was increased 2-fold (p < 0.05) in cells expressing Ht31·ECFP and RII·EYFP compared with cells expressing Ht31P·ECFP, the inactive form of Ht31, and RII·EYFP. These results provide the first in vivo demonstration of RII/Ht31 interaction in living cells and confirm previous in vitro findings of RII/Ht31 binding. Using surface plasmon resonance, we also showed that the green fluorescent protein tags did not significantly alter the binding of Ht31 to RII. Thus, fluorescence resonance energy transfer can be used to directly monitor protein-protein interactions of the PKA signaling pathway in living cells.

There is now increasing evidence for clustering and compartmentalization of signaling enzymes with their activators and target proteins. This targeting of enzymes with their substrates may promote specificity of signaling events (1)(2)(3). To better understand the physiological role of scaffolding proteins, such as A-kinase anchoring proteins (AKAPs), 1 which target cyclic AMP-dependent protein kinase (PKA) (2,4), it is essential to monitor in real time the interactions between the components of this signaling complex in living cells. However, to date such measurements have not been reported for the PKA pathway.
AKAPs and PKA co-localize in cells. This was first shown by the redistribution of the type II regulatory (RII) and the catalytic subunits of PKA from the cytosolic to the particulate fraction following overexpression of AKAP75 in HEK293 cells (5). In studies in HEK293 cells, S-AKAP84 targeted RII to mitochondria (6). In rat cardiac myocytes, endogenous AKAP100 co-localized with RII in the region of the junctional sarcoplasmic reticulum/transverse tubule (7). A functional role of AKAP targeting of PKA is further indicated by studies in which Ht31 peptide, a competitive inhibitor of AKAP/PKA binding, was introduced into AKAP-overexpressing cells. For example, transfection of RINm5F pancreatic beta cells with Ht31 caused redistribution of RII from a perinuclear to a diffuse cytoplasmic localization and prevented cAMP-dependent insulin secretion (8). Similarly, in cardiac myocytes transfected with AKAP79, microinjection of Ht31 impaired the PKA-dependent increase in L-type Ca 2ϩ current (9). These studies demonstrate a role for AKAPs in the regulation of PKA distribution and in the specificity of PKA function.
To date, the binding of RII with full-length AKAPs or with the peptide Ht31 has been demonstrated in vitro by RII overlay assay (10 -12), by band shift assays on nondenaturing gels (12,13), by equilibrium dialysis measurements (12,13), and recently by solution NMR measurements of the N-terminal dimerization domain of RII with Ht31 (14). Equilibrium dialysis measurements (12) and surface plasmon resonance 2 showed that Ht31 binds RII with nanomolar affinity. Although these in vitro approaches provide evidence for high affinity RII/AKAP or RII/Ht31 interactions in vitro, the results presented here extend these studies by measuring real-time interactions between these molecules in living cells.
Fluorescence resonance energy transfer (FRET) was used to analyze the interaction between RII⅐EYFP and Ht31⅐ECFP expressed in CHO cells. We used enhanced cyan fluorescent protein (ECFP) and enhanced yellow fluorescent protein (EYFP) (16) to construct the fusion proteins, RII⅐EYFP, Ht31⅐ ECFP, and Ht31P⅐ECFP (the inactive, prolinated Ht31 analog). A significant increase in RII⅐EYFP fluorescence emission was observed in cells expressing RII⅐EYFP/Ht31⅐ECFP compared with control cells expressing RII⅐EYFP/Ht31P⅐ECFP thus dem-onstrating, for the first time, binding of RII to Ht31, the RIIbinding domain of an AKAP, in living cells.

Construction of Vectors for the Expression of GFP-tagged Proteins-
The cDNAs for RII␣, Ht31, and Ht31P were gifts from John Scott (Vollum Institute, Howard Hughes Medical Institute, Portland, OR). The construct for Ht31, used in the FRET experiments, encodes residues 418 -718 of the human thyroid RII anchoring protein (8). The RII-binding domain, residues 493-515 (12), is included in this peptide. The construct for Ht31P contains residues 418 -718 from the same AKAP but with proline substitutions at positions 498 and 507 that prevent Ht31P from binding RII (12). The cDNA encoding ECFP or EYFP (gifts from Roger Tsien, Howard Hughes Medical Institute, University of California, San Diego, CA) was ligated into EGFP-N1 (CLON-TECH) using AgeI and BsrGI to form pECFP or pEYFP, respectively.
The cDNAs for RII, Ht31, or Ht31P were amplified by polymerase chain reaction to remove the stop codon and add restriction enzyme sites for BglII, XhoI, and BamHI. The cDNA encoding Ht31 or Ht31P was excised using BglII and XhoI and ligated into pECFP upstream of ECFP to create pHt31⅐ECFP or pHt31P⅐ECFP, respectively. The cDNA encoding RII was excised using XhoI and BamHI and ligated into pEYFP upstream of EYFP to create pRII⅐EYFP. Thus, ECFP or EYFP was expressed at the C terminus of each of the proteins of interest. Constructs were verified by DNA sequencing.
Transfection and Selection of Positive Cells-CHO cells were stably transfected with pECFP, pEYFP, pHt31⅐ECFP, or pHt31P⅐ECFP using Lipofectin (Life Technologies, Inc.) according to the manufacturer's protocol. Cells were subcultured 1:5 into selection media (0.5 mg/ml G418) 48 h posttransfection. Stable transfectants were amplified and sorted by fluorescence-activated cell sorter analysis using excitation wavelengths of 410 nm for ECFP-expressing cells and 480 nm for EYFP-expressing cells. For the FRET experiments, cells stably transfected with Ht31⅐ECFP or Ht31P⅐ECFP were plated on 30-mm glassbottom dishes and then transiently transfected with RII⅐EYFP using Lipofectin, similar to the methods described by Miyawaki et al. (16) in their FRET study.
Western Blot Analysis-Western blot analysis was used to verify that transfected CHO cells expressed the appropriate fusion proteins. Adherent cells were released by trypsin-EDTA treatment, washed with phosphate-buffered saline, pelleted, and resuspended in ice-cold lysis buffer (20 mM HEPES, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 10% glycerol, pH 7.4) containing protease inhibitors (30 g/ml aprotinin, 10 g/ml leupeptin, 30 g/ml pepstatin, 1 mM phenylmethylsulfonyl fluoride) and lysed using a freeze-thaw method. Proteins from the cell lysate (10 g/lane) were separated by SDS-polyacrylamide gel electrophoresis using a 10% gel, transferred to nitrocellulose, and probed with an anti-GFP antibody (monoclonal, Roche Molecular Biochemicals) that recognizes both ECFP or EYFP mutants. Densitometry was performed using NIH Image Software to determine differences in expression levels of Ht31 and Ht31P and to determine whether equal levels of RII⅐EYFP were transiently expressed in cells stably expressing Ht31 or Ht31P.
Fluorescence Microscopy-Cells were plated on glass coverslips to 95% confluency, washed with phosphate-buffered saline, and fixed using ice-cold methanol. Cells containing the RII⅐EYFP fusion protein were viewed directly using a fluorescein isothiocyanate filter. Immunofluorescent techniques were used to view cells expressing ECFP or an ECFP fusion protein. These cells were fixed, blocked with 3% bovine serum albumin in phosphate-buffered saline, and probed with a primary antibody to GFP and a fluorescein isothiocyanate-conjugated secondary antibody (Jackson ImmunoResearch). Coverslips were then mounted onto slides, and cells were viewed using a fluorescein isothiocyanate filter. Although an indirect method was used for viewing ECFPexpressing cells, both methods can identify the location of respective fusion proteins.
Preparation of Recombinant RII␣ and Synthetic Peptides-Purified recombinant RII␣ was prepared as described previously (7). Recombinant RII protein was biotinylated (RII B ) using a BAC-SulfoNHS reaction (Sigma) according to manufacturer's instructions. Ht31 and Ht31P (residues 493-515 of the full-length protein) were biotinylated at the N terminus (Ht31 B , Ht31P B ) to allow detection of RII binding events 2 (17). All peptides were synthesized by the Molecular Biotechnology Core Facility at the Cleveland Clinic Foundation.
Surface Plasmon Resonance-Surface plasmon resonance was used to investigate the effects of adding a GFP derivative to the C termini of RII and Ht31. Measurements were taken using the BIAcore 3000 in-strument. Streptavidin (SA) sensor chips and HBS buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM EDTA, 0.005% Surfactant P20) were purchased from Pharmacia Biosensor AB. The concentration of RII⅐EYFP in CHO cell lysates was determined by enzyme-linked immunosorbent assay using purified RII and anti-RII antibody (Upstate Biotechnology) as described previously (18). Samples of RII, RII⅐EYFP, Ht31⅐ECFP, or Ht31P⅐ECFP (analytes) were diluted in HBS buffer. For determining RII/Ht31 and RII⅐EYFP/Ht31 interactions, samples of RII or RII⅐EYFP (0.125-1.0 g/ml) were injected at 25°C at a flow rate of 10 l/min over the active SA surface on which Ht31 B (ligand) had been immobilized to ϳ30 response units or over a control surface (control peptide, immobilized Ht31P B ). 2 For determining Ht31⅐ECFP/RII interactions, samples of Ht31⅐ECFP or Ht31P⅐ECFP were injected over the active RII surface (immobilized RII B ) or control surface (SA surface blocked with free biotin). In the absence of ECFP, Ht31 or Ht31P peptides are too small to give a significant increase in response units, so we were unable to compare binding of the GFP-tagged Ht31 with the native peptide. Therefore, RII/Ht31 interactions were determined by passing RII over immobilized Ht31 B and not vice versa. Flow-rate variation experiments confirmed that both SA surfaces (RII B and Ht31 B ) were not limited by mass transport effects. Kinetic constants were calculated by global fitting of the data to a 1:1 Langmuir binding model after subtraction of the control surface, using BIAevaluation software, version 3.0, according to the BIAcore pseudo-first order rate equation as described by Zakhary et al. 2 FRET Analysis of Cells That Contain Ht31⅐ECFP or Ht31P⅐ECFP and RII⅐EYFP-FRET experiments were performed 48 h after transient transfection of Ht31-or Ht31P-expressing CHO cells with RII⅐EYFP. Each experiment was performed on 15-20 cells at ϳ80% confluency. Fluorescence was observed with an Olympus IX-70 inverted fluorescence microscope and recorded using the Felix software imaging system (Photon Technology International). Using this system, excitation was set by a monochromator to the peak excitation wavelength for ECFP, 440 Ϯ 5 nm (using a 455DCLP dichroic mirror). Emitted light was measured at 480, the peak emission wavelength for ECFP, and at 535 nm (D480/30 and D535/25 emission filters), the peak emission wavelength for EYFP. Background fluorescence was determined for each plate of cells by taking readings in a region of the plate that contained no cells, using the same size field of view. These readings were subtracted from each data point. We did not subtract the autofluorescence from untransfected cells because cell densities differed between dishes. Data were expressed as either the raw fluorescence intensity or as the ratio of fluorescence intensity at 535 nm/480 nm.

Expression of Fluorescent Fusion Proteins in Transfected
CHO Cells-Western blot analysis was used to verify that transfected CHO cells contained the appropriate fusion proteins (Fig. 1). The presence of immunoreactive bands confirmed that cells expressed either the 65-kDa Ht31⅐ECFP or Ht31P⅐ ECFP fusion proteins and/or the 80-kDa RII⅐EYFP fusion protein. Densitometric scanning of the blots revealed only a 2% difference in expression levels between Ht31⅐ECFP and Ht31P⅐ ECFP in stably transfected cells (n ϭ 4). Lysate was also collected from five plates of cells that were stably transfected with either Ht31⅐ECFP or Ht31P⅐ECFP and then transiently transfected with RII⅐EYFP to determine whether cells stably expressing Ht31⅐ECFP or Ht31P⅐ECFP transiently expressed equal levels of RII⅐EYFP. Proteins from the cell lysate were separated by SDS-polyacrylamide gel electrophoresis, and GFP fusion proteins were identified using an anti-GFP antibody (Fig. 1). Densitometric analysis of RII⅐EYFP revealed that cells stably transfected with Ht31P⅐ECFP expressed slightly higher levels of RII⅐EYFP (an 8% difference) than cells stably expressing Ht31⅐ECFP.
In addition to Western blot analysis, fluorescence microscopy was used to directly visualize the appropriate fusion protein in the transfected cells. ECFP and EYFP were expressed throughout the cytosol (data not shown). However, RII⅐EYFP appeared throughout the cytosol and in the perinuclear region ( Fig. 2A). Ht31⅐ECFP-expressing cells showed a similar pattern to those transfected with RII⅐EYFP but with more fluorescent signal being detected in the cytosol (Fig. 2B). The latter results are similar to the findings of Lester et al. (8), who demonstrated that Ht31 localized to the cytosol of transfected RINm5F cells.
Determination of Kinetic Parameters by Surface Plasmon Resonance-Equilibrium binding constants (K d ) were determined for the interaction of RII⅐EYFP and Ht31⅐ECFP with their respective binding partners. There was no significant difference in the K d values for Ht31 and RII (11.0 Ϯ 0.2 ϫ 10 Ϫ9 M; n ϭ 3) or Ht31 and RII⅐EYFP (11.6 Ϯ 0.8 ϫ 10 Ϫ9 M; n ϭ 3). The K d for Ht31⅐ECFP binding to RII was 4.3 ϫ 10 Ϫ9 M. The lower K d for Ht31⅐ECFP binding to RII, as compared with the K d values for interactions between Ht31 and RII or between Ht31 and RII⅐EYFP, may be due to the different experimental conditions, because Ht31⅐ECFP binding to RII was measured using an SA surface containing immobilized RII, whereas RII⅐EYFP/Ht31 binding was measured on an SA surface containing immobilized Ht31. A representative sensorgram of RII⅐ EYFP binding to immobilized Ht31 peptide is shown in Fig. 3. Ht31P⅐ECFP did not bind RII, indicating that proline substitution of the amphipathic helix disrupts RII interaction with this fusion protein as it does for Ht31P without the ECFP tag (results not shown).
FRET Analysis of Ht31P⅐ECFP Interactions with RII⅐EYFP-In CHO cells expressing Ht31⅐ECFP or Ht31P⅐ECFP plus RII⅐ EYFP, FRET was observed as a decrease in emission at 480 nm (the peak emission wavelength for ECFP) and as an increase in emission at 535 nm (the peak emission wavelength for EYFP). This indicates that the 480-nm emission from the donor fluorophore (Ht31⅐ECFP) excited the acceptor fluorophore (RII⅐ EYFP) resulting in an increase in emission at 535 nm. There was a dramatic increase in fluorescence emitted at 535 nm in cells expressing Ht31⅐ECFP and RII⅐EYFP, because the intensity of the 535-nm fluorescence from these cells was 5 times greater than native CHO cells and 2 times greater than control cells expressing Ht31P⅐ECFP and RII⅐EYFP (Fig. 4). Emission at 480 nm was reduced by 34% in Ht31⅐ECFP/RII⅐EYFP expressing cells when compared with control CHO cells expressing Ht31P⅐ECFP and RII⅐EYFP. Cells expressing RII⅐EYFP alone displayed background levels of fluorescence when excited at 440 nm, because emission at 480 or 535 nm was reduced by an average of 97 or 84%, respectively, when compared with cells expressing both Ht31⅐ECFP and RII⅐EYFP (data not shown). Interestingly, when cells expressing Ht31⅐ECFP were excited at 440 nm, emission at 535 nm was only 18% lower than emission from cells expressing Ht31P⅐ECFP and RII⅐EYFP (data not shown).
To control for different densities of cells on each plate, the ratio of fluorescence at 535 nm/480 nm was used to compare FRET between cells expressing RII⅐EYFP and Ht31⅐ECFP or RII⅐EYFP and Ht31P⅐ECFP. The ratio of fluorescence was over 2 times greater in cells expressing Ht31⅐ECFP and RII⅐EYFP compared with cells transfected with Ht31P⅐ECFP and RII⅐EYFP (n ϭ 18) (Fig. 5). These results are consistent with the findings of Miyawaki et al. (16) who report an ϳ1.5-fold ratio increase with the ECFP/EYFP pair when their "yellow cameleon" (ECFP and EYFP linked together with calmodulin  (18). R o , the critical Forster radius, is the distance at which FRET is 50% efficient (20). For the ECFP⅐EYFP pair, R o is reported to be 50 Å (19). Thus, significant energy transfer should only occur between Ht31⅐ECFP and RII⅐EYFP if these two species are ϳ10 -50 Å apart. We observed energy transfer between Ht31⅐ ECFP and RII⅐EYFP; therefore we can conclude that the Ht31⅐ ECFP and RII⅐EYFP expressed in the CHO cells are sufficiently close for interaction between these molecules to take place.
N-terminal dimerization of RII is required for RII docking to AKAPs (21,22) and for RII binding to Ht31 (23). Therefore, our results imply that RII⅐EYFP associates to form dimers in the CHO cells. Because Ht31 can bind to the N termini of dimerized RII in vitro (23,24), the ECFP tag at the C terminus of RII should not interfere with AKAP (or Ht31) binding to RII. We tested this hypothesis using surface plasmon resonance. We assessed whether significant changes in K d occurred as a result of adding a GFP derivative to the C terminus of these molecules. Our results showed that the affinity of RII for Ht31 was not significantly altered by fusion of EYFP to the C terminus of RII. Likewise, taking into consideration that these experiments were performed using two different SA surfaces, the K d for Ht31⅐ECFP binding to immobilized RII was similar to that of RII binding to immobilized Ht31. Thus, our results demonstrate that not only can we successfully use FRET to measure the interaction between molecules of RII and the RII-binding domain of AKAPs in intact cells but also that the addition of the 27-kDa ECFP or EYFP fluorescent tag to the C terminus of the expressed proteins does not significantly alter this intercellular interaction. Thus, we can conclude that the C-terminal tag does not affect the interaction between Ht31 and RII.
In previous FRET studies, two different pairs of GFP derivatives were used (25)(26)(27). We used the ECFP⅐EYFP pair rather than enhanced blue fluorescent protein with enhanced GFP. One disadvantage of the ECFP⅐EYFP pair is that the amplitude of the change in the emission ratio is reduced, compared with the enhanced blue fluorescent protein-enhanced green fluorescent protein pair, because the ECFP emission is relatively broad and overlaps the EYFP emission spectrum (27). However, the ECFP⅐EYFP pair has improved photostability (16,19). In our study, overlap of ECFP emission into the EYFP channel, referred to as "cross-talk" by Gordon et al. (28), may account for elevated FRET measurements in cells expressing the negative control Ht31P⅐ECFP and RII⅐EYFP or in cells expressing Ht31⅐ECFP alone. Also, autofluorescence, believed to be primarily from flavoproteins with peak excitation at 488 nm and peak emission at ϳ520 nm (15), may have contributed to background fluorescence in both the RII⅐EYFP/Ht31⅐ECFP and the RII⅐EYFP/Ht31P⅐ECFP-expressing cells. Thus, there is likely to be a contribution both from FRET and from autofluorescence to the 535/480-nm emission ratio for both sets of transfected cells.
In summary, the results of this study provide the first direct demonstration of binding of RII to the RII-binding domain of an AKAP in intact cells, thereby demonstrating the physiological relevance of previous measurements of RII/Ht31 interactions performed in vitro (12). Our findings indicate the potential of FRET, between expressed GFP fusion proteins, as a powerful tool for measuring PKA/AKAP interactions in the intracellular environment.
Acknowledgments-We acknowledge Dr. Jianjie Ma for helpful discussions and Russ Desnoyer for help in creating the fluorescent protein fusion constructs. We would also like to thank Dr. Dianne Perez for critical review of the manuscript. The Becton-Dickinson FACSVantage used to perform sorting of EYFP-expressing cells was purchased through a generous gift from the Keck Foundation. ECFP-expressing cells were sorted using an Elite ESP at the Cancer Research Center at Case Western Reserve University. Polymerase chain reaction products were sequenced, and peptides were made by the Molecular Biotechnology Core facility at the Cleveland Clinic Foundation.