The catalytic (C) subunit of protein kinase A functions both in the cytoplasm and the nucleus. A major charge variant representing about one third of the enzyme in striated muscle results from deamidation in vivo of the Asn2 residue at the conserved NH2-terminal sequence myrGly-Asn-Ala (Jedrzejewski, P.T., A. Girod, A. Tholey, N. König, S. Thullner, V. Kinzel, and D. Bossemeyer. 1998. Protein Sci. 7:457–469). Because of the increase of electronegativity by generation of Asp2, it is reminiscent of a myristoyl-electrostatic switch. To compare the intracellular distribution of the enzymes, both forms of porcine or bovine heart enzyme were microinjected into the cytoplasm of mouse NIH 3T3 cells after conjugation with fluorescein, rhodamine, or in unlabeled form. The nuclear/cytoplasmic fluorescence ratio (N/C) was analyzed in the presence of cAMP (in the case of unlabeled enzyme by antibodies). Under all circumstances, the N/C ratio obtained with the encoded Asn2 form was significantly higher than that with the deamidated, Asp2 form; i.e., the Asn2 form reached a larger nuclear concentration than the Asp2 form. Comparable data were obtained with a human cell line. The differential intracellular distribution of both enzyme forms is also reflected by functional data. It correlates with the degree of phosphorylation of the key serine in CREB family transcription factors in the nucleus. Microinjection of myristoylated recombinant bovine Cα and the Asn2 deletion mutant of it yielded N/C ratios in the same range as encoded native enzymes. Thus, Asn2 seems to serve as a potential site for modulating electronegativity. The data indicate that the NH2-terminal domain of the PKA C-subunit contributes to the intracellular distribution of free enzyme, which can be altered by site-specific in vivo deamidation. The model character for other signaling proteins starting with myrGly-Asn is discussed.
An increase in cellular cAMP represents a rapid early metabolic response to a variety of external stimuli. Cyclic AMP activates the cyclic AMP–dependent protein kinase A (PKA), an abundant multifunctional enzyme. By phosphorylation of key substrates, PKA transmits exogenous signals regulating a large number of cellular processes in the cytoplasm as well as in the nucleus (for review see Walsh and Van Patten 1994). Although different signaling pathways are propagated through the same protein kinase, the character as well as the time frame of cellular responses may differ widely. Whereas cAMP-triggered metabolic steps in the cytoplasm are initiated within less than a minute, cAMP-stimulated nuclear events proceed at a much slower pace with the maximum occurring only after ∼30 min (Hagiwara et al. 1993). Control of the intracellular distribution of the active enzyme is undoubtedly of great importance (for reviews see Lohmann and Walter 1984; Nigg 1990).
In the absence of cAMP, PKA exists largely as an inactive tetrameric holoenzyme consisting of two regulatory (R) subunits composed of a dimer and two catalytic (C) subunits (for review see Krebs and Beavo 1979). The holoenzyme is primarily localized in the cytoplasm or bound to specific cytoplasmic organelles via anchoring proteins, depending on the type of R-subunit (for review see Scott 1991; Hubbard and Cohen 1993). As sufficient amounts of cAMP become available, up to 4 mol of the nucleotide are bound to each R-subunit dimer, which in turn liberates two molecules of monomeric C-subunit. Evidence has been presented that the free C-subunit can translocate to the nucleus (Nigg et al. 1985; Meinkoth et al. 1990; Nigg 1990), although a classical nuclear localization sequence has not yet been identified. Moreover, it has been shown that cAMP-stimulated transcription is rate-limited by the nuclear entry of the enzyme (Hagiwara et al. 1993).
Quantitative data on nuclear entry have been obtained previously by microinjection of recombinant catalytic subunit Cα (mouse sequence), and were explained by diffusion (Harootunian et al. 1993). Native C-subunit isolated from tissues carries a long chain fatty acid at the NH2 terminus, mostly myristate, which is absent from the recombinant form (Carr et al. 1982). From a comparison of recombinant with the native catalytic subunit it was assumed that myristoylation of the native enzyme was not obligatory for intracellular function, including nuclear entry of the enzyme (Clegg et al. 1989; Meinkoth et al. 1990).
Recent evidence shows that in the native catalytic subunit, an increase in electronegativity can occur close to the NH2 terminus through a posttranslational mechanism. Preparations of the catalytic subunit from striated muscle of several mammalian species, although homogenous by SDS gel electrophoresis, have been separated into two chromatographic fractions, A and B, in many laboratories on the basis of charge differences. These are not due to differences in the molar phosphorylation rate (Kinzel et al. 1987). Based on earlier observations (Hotz et al. 1989), it recently has been shown by electrospray mass spectrometry and sequence analysis of the C-subunit from four different mammalian species (Jedrzejewski et al. 1998) that fraction B contains the catalytic subunit gene products Cα and Cβ in the encoded form with Asn at position two in both cases. Fraction A, which represents consistently about one third of routine enzyme preparations, contains Cα and Cβ at the same ratio as fraction B but exclusively in the Asp2 form. The fact that Asp2 is not encoded in Cα and Cβ, and even not allowed in substrates of the N-myristoyl transferase (NMT; for review see Towler et al. 1988), argues for a posttranslational conversion of Asn2 to Asp2 by deamidation, which is responsible for the 0.4-pH unit difference of the isoelectric values of the fractions A and B. Evidence has been presented that indicates that deamidation of Asn2 occurs in vivo, i.e., represents a natural modification (Jedrzejewski et al. 1998) that appears to be regulated since the deamidated fraction of C-subunit from striated muscle of different mammalian species represents reproducibly one third from the total. The regulation of deamidation of Asn2, however, remains to be elucidated.
The significance of the Asn2 to Asp2 conversion in the C-subunit is unknown. Effects on the enzyme activity in the test tube, its reassociation with R-subunits (Kinzel et al. 1987), or its association with protein kinase inhibitor (Van Patten et al. 1988) were not observed so far. This prompted us to ask if such a conserved deamidation of the C-subunit might influence the localization of the enzyme in living cells. We determined the nuclear/cytoplasmic distribution of both forms of porcine and bovine C-subunit after microinjection into the cytoplasm of cultured cells. The data show that the nuclear/cytoplasmic ratio approached by the Asp2 form of the enzyme is lower than that obtained with the encoded Asn2 form. Accordingly, the phosphorylation of substrate in the nucleus (CREB family transcription factors) through the deamidated form is smaller than with the encoded form. It seems that the partial deamidation of the C-subunit at the conserved Asn2 is aimed at modulating the intracellular distribution as well as the phosphorylation capacity of free enzyme among the major cellular compartments.
Materials and Methods
Catalytic Subunit of the cAMP-dependent Kinase Type II from Pig and Bovine Hearts
The catalytic subunit of the cAMP-dependent kinase type II from pig and bovine hearts was prepared as described earlier (Nelson and Taylor 1981; Kinzel et al. 1987; Bossemeyer et al. 1993) with minor modifications including a scaling up. In brief, hearts were homogenized in 50 mM potassium phosphate buffer, pH 6.5, containing 1 mM EDTA, 2 mM DTT, and the following protease inhibitors: 0.5 mM PMSF, 0.2 mM TPCK, 0.5 mg/l leupeptin, 0.7 mg/l pepstatin A, 0.5 mg/l E64, 0.5 mM N-NH2-capronic acid, and 0.5 mg/trypsin inhibitor. The homogenate was cleared by centrifugation and filtration through glass wool and bound to DE-52 cellulose (Whatman). The DE cellulose was washed with 100 mM potassium phosphate buffer, pH 6.5, and the holoenzyme was eluted by 300 mM potassium phosphate buffer, pH 6.5. After dilution of the eluate (1:5) the holoenzyme was again bound to DE cellulose. After washes with potassium phosphate buffer (50 mM, pH 6.5, and 15 mM, pH 6.1), the C subunit was eluted by cAMP (0.2 mM in buffer A, 15 ml potassium phosphate, 1 mM EDTA, 0.1 mM DTT) and immediately bound to a small column with CM-52 cellulose (Whatman). The C-subunit was eluted by a gradient 0–60% buffer B (100 mM Mops, 150 mM KCl, l mM DTT, 0.l mM EDTA, pH 7) for the first 20 ml; 60–100% buffer B for the next 30 ml. Fractions of 10 ml were collected. From 4-kg hearts, typically, 15–20 mg of electrophoretically homogenous C subunit were obtained.
Separation of Fraction A and B
Up to 2 mg of C-subunit diluted 1:2 by H2O was separated on a Mono S HR 5/5 column (FPLC-System; Pharmacia). The gradient was programmed to be 0–22% buffer B (10 mM bis-Tris propane, 1 M LiCl, pH 8.5) 0–12 ml; 22–26% buffer B 12–25 ml; 26–100% buffer B 25–26 ml; 100% buffer 26–50 ml. Flow rate was 0.75 ml/min. The two peaks, at ∼23% buffer B and 55% buffer B representing enzyme fraction A and B (see Fig. 1), were collected. Separation of fraction A and B was verified by isoelectric focusing. Protein kinase activity was assayed as described previously (Kinzel et al. 1987); the fractions had an equal specific enzyme activity. Details of the analysis of fraction A and B by mass spectrometry have been described elsewhere (Jedrzejewski et al. 1998).
Labeling of Enzyme in Fraction A and B with FITC or with TRITC
Labeling of fraction A and B with FITC or with TRITC (both from Pierce Chemical Co.) was carried out essentially according to Adams et al. 1991. The buffers of the protein solution (∼0.5 mg/ml) were substituted by 25 mM bicine, 0.1 mM EDTA, pH 8.0, using centrifuge concentrators (Centricon; Amicon). The catalytic subunits were labeled with 0.3 mM FITC or 0.5 mM TRITC in the presence of 8 mM MgC12 and 5 mM ATP to prevent inactivation of kinase activity. Labeling was allowed to proceed for 30 min at room temperature, and then quenched by addition of 5 mM glycine for 15 min. Excess dye was removed by passing each protein solution through a PD 10 column (Pharmacia) equilibrated with 25 mM potassium phosphate buffer, pH 6.8, 0.5 mM EDTA, l mM DTT, 10% glycerol. The labeled C-subunits were eluted with 3.5 ml of the above buffer and adjusted to 1 mg/ml protein using Centricon and Minicon microconcentrators for microinjection. The dye to protein stoichiometry was determined by measuring absorbance at 495 nm for protein-bound fluorescein (assuming an extinction coefficient of 65 × 103 M−1cm−1), at 552 nm for protein-bound rhodamine (assuming an extinction coefficient of 72 × 103 M−1cm−1), and measuring the protein concentration according to Bradford 1976. FITC/protein ratios for fraction A and B were ≥ 0.9 mol FITC per mole of C-subunit as indicated in Results. The TRITC/protein ratio was ≥ 0.4 mol TRITC per mole of C-subunit. Dye conjugation had no significant effect on the specific enzyme activity. The activity of labeled enzymes could be inhibited by protein kinase inhibitor and by regulatory subunit.
Expression and Purification of Recombinant myr-rCα and Mutants
Recombinant bovine catalytic Cα (Wiemann et al. 1992) was expressed using vector pT7-7 in the Escherichia coli strain BL21(DE3) and purified by affinity chromatography and FPLC separation on a Mono S column as described (Girod et al. 1996), except that the vector pBB131 for the expression of yeast NMT was coexpressed together with Cα (Duronio et al. 1990, Jedrzejewski et al. 1998), leading to virtually complete myristoylation of the catalytic subunit. The specific enzyme activity equaled that of native enzyme obtained from heart muscle. Site-directed mutagenesis was performed according to Kunkel 1985; the fragment was characterized by sequencing and cloned into the expression vector pT7-7, coexpressed with yeast NMT and purified as above.
Polyclonal antibodies against the C-subunit were raised in rabbits by two intramuscular injections (3- wk interval) of 0.4–0.5 mg C-subunit (isolated from bovine heart coupled to the same amount of hemocyanin by the use of glutaraldehyde 0.5%, 15 min at room temperature) and incomplete Freund adjuvants (1:1). After ∼2 mo, the serum reacted with C-subunit at a 4 × 104 dilution in an ELISA. C-subunits from bovine and porcine origins were equally recognized by the antiserum. This was expected on the basis of the high degree of sequence homology (Buechler et al. 1989; Jedrzejewski et al. 1998). For some experiments, antibodies were purified by affinity chromatography using immobilized bovine C-subunit (3 mg coupled to 2 ml CNBr Sepharose; Pharmacia). Polyclonal rabbit antibodies against CREB and CREB-(p)Ser133 were a gift of Dr. W. Schmid (Division of Molecular Biology of the Cell I, German Cancer Center, Heidelberg, Germany). Recombinant CREB protein for in vitro phosphorylation was a gift from D. Gau (Division of Molecular Biology of the Cell I, German Cancer Center, Heidelberg, Germany).
Electrophoresis of fraction A and B in the presence of SDS was done in 11% polyacrylamide gels. The proteins were subsequently blotted to an Immobilon-P membrane. For immunostaining of proteins after Western blotting, the membrane was first soaked in TBS with 0.05% Tween 20 for 2 h, followed by incubation for at least 2 h with antibodies in TBS/0.05% Tween 20 (TBS/T). The blot was washed three times in TBS/T and incubated with peroxidase-coupled secondary antibody for 2 h. After another wash, the blot was finally stained according to Kobayashi and Tashima 1989.
Isoelectric Focusing of the C Subunit
Isoelectric focusing of the C-subunit was done in 1-mm ampholine gels, pH 3.5–9.5 (Pharmacia), prefocused at 8 W until 500 V was reached. 1–2 μg of the different C-subunits (labeled or unlabeled) diluted to 10 μl by water was spotted ∼3 cm away from the anode on the gel. The gel was run for 2 h at 1,500V, 8W, and stained by Coomassie blue. 2 μg myoglobin of horse and whale were used as pH markers.
Cell Culture and Microinjection
Mouse NIH 3T3 and HeLa cells were cultured as previously described (Kinzel et al. 1988; Pepperkok et al. 1988, Pepperkok et al. 1993a,Pepperkok et al. 1993b). For microinjection experiments, cells were plated onto glass coverslips (10 × 10 mm) or in glass bottom Microwell petri dishes (Mattek) and grown for 2 d (70% confluency). For microinjection, cells were transferred into carbonate-free culture medium containing 5% FCS and 20 mM Hepes. Microinjection was performed at room temperature using a Zeiss automated injection system (AIS; Ansorge and Pepperkok 1988). About 0.1 pl of sample was introduced into the cytoplasm of the cells representing 5–10% of the cell volume. The C-subunit samples to be microinjected were adjusted to identical concentration for fraction A and B in individual experiments as given in Results. Subsequently, microinjected cells were transferred into prewarmed carbonate-buffered culture medium and further incubated at 37°C and 5% CO2 for various periods as indicated.
Immunofluorescence and Fluorescence Microscopy
Cells microinjected with fluorescently labeled C-subunits were fixed with 3.5% paraformaldehyde for 20 min on ice, washed with PBS/BSA, and mounted onto glass slides using 6 μl of Mowiol. In some experiments, injected cells plated on glass bottom microwells were directly transferred to the microscope for quantification of C-subunit fluorescence in living cells. Cells injected with unlabeled C-subunit were fixed with ice-cold 100% methanol for 4 min on ice. After fixation, they were washed extensively with PBS containing 0.1% BSA (PBS/BSA), incubated with polyclonal rabbit anti–PKA-AB (dilution 1:500) for 30 min at room temperature, subsequently washed with PBS/BSA, and incubated for another 30 min with rhodamine-conjugated goat anti–rabbit antibodies (Sigma Chemical Co.; dilution 1:100). Coverslips were finally washed in PBS/BSA and mounted onto glass coverslips as described above.
For immunostaining of CREB or phospho-CREB, injected cells were fixed with 3.5% paraformaldehyde as described above, permeabilized with 0.1% Triton X-100 in PBS for 4 min, and subsequently stained with anti–CREB or anti–phospho-CREB antibodies for 20 min, followed by incubation with ALEXA488-conjugated anti–rabbit antibodies (Molecular Probes). Coinjected mouse IgGs were stained with Alexa568-conjugated anti–mouse antibodies (Molecular Probes).
Images of stained cells were recorded either on a Zeiss inverted fluorescence microscope (model 135TV; Axiovert) equipped with a cooled slow scan CCD camera (model CH250; 1,317 × l,035 pixels; Photometrics) controlled by a Macintosh Quadra 900, or on a similar microimaging system recently described (Pepperkok et al. 1993a,Pepperkok et al. 1993b). Images were further processed and figures were arranged with the software package IPLab spectrum V3.1 (Signal Analytics Corp.) and Adobe Photoshop 3.0.
Quantification of compartment-specific cellular fluorescence was performed as described (Lorenz et al. 1993; Pepperkok et al. 1993a,Pepperkok et al. 1993b). Nuclear and cytoplasmic fluorescence intensities were determined separately in single cells and from these values was subtracted the local background intensity measured in neighboring uninjected cells by a background staining correction method described in Pepperkok et al. 1993a. The ratio of the corrected nuclear and cytoplasmic fluorescence intensities (N/C) was determined individually for each cell measured. In all experiments, C-subunit fractions, A and B, from the same preparation were compared directly.
Porcine PKA C-subunit Fraction A and B
To obtain the major fractions A and B of the C-subunit, porcine catalytic subunit from the heart (Fig. 1, inset, lane 2) was chromatographed as described in Materials and Methods and shown in Fig. 1. The fractions were characterized by mass spectrometry as reported in detail elsewhere (Jedrzejewski et al. 1998). The A/B ratio was ∼1:2, and the measured masses did not differ significantly: 40835.5 ± 1.6 D for fraction A; and 40833.5 ± 1.6 D for fraction B. The molar phosphate content was 1.91 ± 0.17 in fraction A and 1.98 ± 0.14 in fraction B, indicating two phosphorylation sites. For further characterization of both fractions, tryptic digests were analyzed for the myristoylated NH2-terminal peptides by μLC-ESI-MS and by sequence analysis through tandem mass spectroscopy (summarized in Table). Fraction B contains the isozymes Cα and Cβ (∼2.5:1) in the encoded form with Asn2; fraction A contains the isozymes exclusively in the Asp2 form with traces of isomers, which is indicative of a deamidation reaction by the β-aspartyl shift mechanism (for review see Wright 1991). Superimposable data were obtained in the same study with C-subunit fractions A and B from bovine heart (Jedrzejewski et al. 1998). Physical separation of Cα and Cβ is presently not possible. The data in Table indicate that Cα and Cβ were deamidated to a comparable extent. No further differences between fractions A and B were detectable. Therefore, it is reasonable to assume that it is the Asn2 to Asp2 conversion that causes the 0.4-pH unit difference in their isoelectric values and, thus, their separation on cation exchange resins.
Microinjection of Dye-conjugated C-subunit Fractions
For microinjection in the first set of experiments, porcine fractions A and B were labeled with fluorescein (FITC) according to Adams et al. 1991, after which the enzymes retained >97% of their specific activity as measured by phosphorylation of histone. The labeled enzymes were microinjected at room temperature into the cytoplasm of subconfluent NIH 3T3 fibroblasts that were fixed either immediately (1-min values) with paraformaldehyde or after incubation at 37°C for the indicated intervals. The injection volume (∼0.1 pl) was in the order of one tenth of the cell volume. Thus, the amount of enzyme injected (1 mg/ml in the injection needle equivalent to ∼25 μM) exceeded the cellular concentration (estimated in the range of 0.2–0.8 μM in organ tissues; Hofmann et al. 1977; Lohmann and Walter 1984).
Microinjected cells first emitted a strong fluorescence in the cytoplasm, followed by a subsequent increase in the nucleus. Cells fixed within 10 min after injection of the enzyme exhibited no morphological changes. However, beyond 10 min, injected cells started to develop morphological characteristics (see below) as if they had been treated with permeating cAMP derivatives or with agents causing an increase of cellular cAMP, thus, indicating a functional enzyme. No morphological changes were observed when cells were injected with purified mouse or rabbit IgGs in the injection buffer (data not shown).
For quantification of the cellular distribution of fractions A and B after microinjections, the cells were further incubated in the presence of 8-bromo-cAMP (1 mM; Byus and Fletcher 1982) to prevent reassociation with R-subunits. Quantification was obtained from images digitized with a slow scan cooled CCD camera at 12-bit resolution as described in Materials and Methods. At least 50 cells per group and time point were analyzed and the ratio of nuclear/cytoplasmic fluorescence (N/C) was determined. The results, given in Fig. 2, show that between 5 and 10 min after injection, the N/C ratios for A and B diverged and reached ∼2.8 and 4.8, respectively; i.e., the N/C ratio obtained with fraction B exceeded that of fraction A ∼1.7-fold. After 10 min, the curves seem to reach a plateau lasting for another 50 min at least, ruling out the possibility that the differences observed here for N/C ratios simply reflect different cytoplasmic to nuclear transport kinetics for fractions A and B. The data in Fig. 2 show that microinjected fraction B reaches a significantly higher nuclear concentration than microinjected fraction A. Moreover, the stability and duration of the plateau phase obtained with both enzyme fractions argues against a measurable interconversion of fraction B into fraction A, i.e., against a significant deamidation during the experiment.
To see if the intracellular distribution of the FITC-conjugated fraction A and B was restricted to NIH 3T3 fibroblasts or represented a more general phenomenon, the fluorescent enzyme fractions were also injected into the cytoplasm of HeLa cells held subsequently in the presence of 8-bromo-cAMP. Again, the N/C ratios of fraction A and B reflected those obtained with mouse cells (data not shown), indicating that the observed phenomenon was not restricted to the particular cell system used.
If the minor molecular differences between fraction A and B are responsible for a differential intracellular distribution, the type of dye conjugated might influence the distribution as well. Therefore, fraction A and B were conjugated with a different dye, rhodamine (TRITC), and microinjected into the cytoplasm of NIH 3T3 cells held subsequently in the presence of 8-bromo-cAMP. The N/C ratios obtained with the TRITC-conjugated fraction A and B 10 min after injection in comparison with the data obtained with FITC-conjugated enzyme fractions are shown in Fig. 3. TRITC-conjugated fraction B reaches an N/C ratio 2.3 times larger than TRITC-conjugated fraction A. The comparable relationships of the N/C ratios indicate that the enzyme distribution approached a similar balance independent of the type of dye conjugated.
The possibility that the charge differences between both enzyme fractions could contribute to the differential translocation behavior led to a control of the isoelectric values of conjugated enzymes. Labeling with rhodamine as well as fluorescein lowered the isoelectric point of the respective enzymes by >0.4 pH units. The data obtained with FITC-conjugated C-isozymes are illustrated in Fig. 4 (lanes 4 and 5). Lane 1 shows the unlabeled electrophoretically homogenous C-subunit preparation composed of the two charge variants. Isolated native fraction A (lane 2) and fraction B (lane 3) focus at about pH 7.1 and 7.5, respectively. (The isoelectric values of Cα and Cβ isozymes were indistinguishable.) Occasionally, fraction B may contain traces of the more acidic A fraction because of some spillover on the column. Upon FITC conjugation of fraction A (lane 4) and fraction B (lane 5) (labeling ratio 0.9, respectively, 1.3 mol FITC/mol C-subunit), the major amount of protein is shifted in both cases towards a more acidic value. In the case of fraction B (lane 5), it should be noted that the major portion of the labeled enzyme is more acidic than the unlabeled fraction A, and that a further protein band appears at an even lower pH, possibly representing enzyme with 2 mol FITC bound per mole.
Microinjection of Native C-subunit Fractions
To evaluate further to what extent the dye modification might contribute to the observed subcellular distribution, native catalytic subunit fractions A and B were used for microinjection. Detection was carried out using antibodies after the enzymes had equilibrated intracellularly. Antibodies distinguishing specifically between fraction A and B are not available. Therefore, the rationale behind the following experiments was that upon microinjection, the cellular concentration of the particular fraction injected would be high enough to be dominantly detected above the endogenous cellular level. The antiserum reacted with C-subunit at a 4 × 104 dilution. The reactivity of the antiserum against separated fractions A and B (tested in ELISA assays) was equal. In Western blots of SDS gels, small amounts of enzyme in fractions A and B were similarly well detectable by the antiserum. Together, these data indicated that fractions A and B had a comparable affinity to the antibodies.
Purified fractions A or B (both at 4 mg/ml) were microinjected into the cytoplasm of NIH 3T3 fibroblasts. They were incubated for 30 min at 37°C, fixed with methanol, incubated with affinity-purified antibodies, and stained with the fluorescent secondary antibody. As shown in Fig. 5, for the antibody dilutions used, the microinjected cells showed a strong fluorescent signal. In contrast, little fluorescence (<25% of the signal of injected cells) was obtained in noninjected cells (marked by stars in Fig. 5). Discrimination between injected and noninjected cells was facilitated by the increased appearance of growth cones (Fig. 5A and Fig. B, arrows) in cells injected with either fraction A or B. At time points longer than 30 min after microinjection, rounding up of the cells was observed (not shown). Similar morphological changes upon microinjection of the C-subunit into living cells have been reported by others (e.g., see Lamb et al. 1988, Lamb et al. 1989). They seem to indicate that the injected enzyme is functional. In all injected cells, a significant fluorescence signal was obtained in the cytoplasm and the nucleus. For both microinjected fraction A and B, an increased nuclear fluorescence was observed (Fig. 5).
After injection of native fraction A and B (4 mg/ml each), the nuclear and cytoplasmic fluorescence in injected cells was quantified as before. To prevent association with cellular R-subunits, 5 mM cAMP (in the pipette) was coinjected. The N/C ratios of the experiments are shown in Fig. 6. The 5-min values are almost identical. However, between 5 and 10 min, a pronounced increase of the N/C ratio was seen with fraction B surpassing that obtained with fraction A by ∼1.5-fold. 30 min after injection, the N/C value for fraction B was 1.6-fold larger than that with fraction A. The time course and the relative distribution of fractions A and B mirror the results obtained with fluorescent probes (Fig. 2 and Fig. 3). Taken together, the data with native fractions A and B from porcine origin suggest that deamidation of Asn2 is sufficient to attenuate the nuclear concentration of the C-subunit. Moreover, the results indicate that dye conjugation as such did not profoundly alter the intracellular behavior of C-subunit isozymes.
The experiments described so far were done with the porcine enzyme. To see how an enzyme from a different species behaves, native fractions A and B from the bovine heart were prepared (the mass of fraction A was 40,858.4 ± 1.6 D; that of fraction B 40,855.7 ± 1.6 D) and injected (4 mg/ml) into NIH 3T3 cells (together with 5 mM cAMP in the pipette). After a 10-min resp., 30 min incubation at 37°C cells were fixed, incubated with the affinity-purified antibodies, and stained with the secondary antibody as above. The N/C ratios obtained (Table) reproduce those obtained with porcine enzymes; the N/C ratio for fraction B exceeded that of fraction A at 10 and 30 min by ∼1.8-fold. Comparison of the 10- and 30-min values shows again that the equilibrium was nearly complete after ∼10 min. That the data with the bovine enzyme is so similar to that obtained with porcine enzyme implies that this phenomenon is independent of the origin of the enzyme.
A survey of the experiments carried out during this study, including all the repeats, showed that the N/C ratios (including 10- and 30-min values) obtained with FITC- and TRITC-labeled as well as unlabeled enzyme ranged for fraction A from 1.9 to 2.9; and for fraction B from 3.7 to 6.8. The variation appeared to be independent of the detection method (even though the unlabeled enzyme was injected at a four times larger concentration than the labeled enzymes). It should be noted that the two enzyme fractions from the same preparation were always compared within the same experiment, and that the individual results reached a comparable level within the particular range of a group. Accordingly, the apparent nuclear fluorescence level of fraction B surpassed that of fraction A 1.5–2.3 times, i.e., the B/A ratio of the N/C values varied comparably little, between 1.5 and 2.3.
To evaluate the possibility that the apparent differences of the N/C ratios obtained with the fractions A and B were caused by a preferential degradation of one of the fractions within a particular subcellular compartment, the total cellular amount of enzyme detectable by immunofluorescence was determined, and the results obtained for A and B were compared with each other. The data (Fig. 7) indicate that fractions A and B do not decay to a different extent. Injected dye-conjugated C-subunit fractions A and B decayed at a similar rate (not shown). Therefore, the N/C ratios indeed seem to reflect a different intracellular distribution.
Microinjection of Recombinant C-subunit
Since the enzyme isolated from porcine as well as bovine heart in fraction B is composed of ∼70% Cα and 30% Cβ isozymes as indicated in Table, a control experiment with recombinant Cα was carried out. For this purpose, the bovine Cα sequence (Wiemann et al. 1992) was coexpressed in E. coli together with NMT to obtain myristoylated recombinant bovine Cα (Duronio et al. 1990; Jedrzejewski et al. 1998). The enzyme was purified by affinity chromatography followed by FPLC chromatography on a Mono S column. The doubly phosphorylated form was injected (4 mg/ml together with cAMP) into the cytoplasm of NIH 3T3 cells and allowed to equilibrate at 37°C for 10 min. The N/C ratio detected by indirect immunofluorescence was 4.8 ± 0.7; i.e., well within the range of values obtained with fraction B in native and dye-conjugated forms.
A control with a recombinant enzyme mimicking the Asp2 consensus form in fraction A was not possible since the NH2-terminal sequence Gly-Asp-X unlike Gly-Asn-X is not a substrate of NMT (Towler et al. 1988). Coexpression of the rCαAsn2Asp mutant with NMT in E. coli did not yield myristoylated enzyme (Jedrzejewski et al. 1998). Moreover, a procedure for specifically deamidating Asn2 of Cα in vitro is not available yet. Therefore, one is limited in this respect to the fraction A isolated from tissue. To evaluate a potential role of Asn2 for the intracellular distribution of the enzyme, Asn2 was deleted from bovine Cα, and the enzyme was coexpressed with NMT in E. coli. The resulting protein was myristoylated; the doubly phosphorylated form was isolated, concentrated, and injected into the cytoplasm of NIH 3T3 cells in the presence of cAMP. After a 10-min incubation at 37°C, the cells were fixed and the N/C ratio was analyzed by the use of antibodies. This deletion mutant reached 5.5 ± 0.5, a value well within the range of data obtained with fraction B, indicating that Asn2 per se is not a determinant for the nuclear/cytoplasmic distribution of the enzyme.
Functional Significance of C-subunit Deamidation
The cyclic AMP–responsive transcription factor CREB and related transcription factors (for simplicity subsequently referred to as CREB) can be regulated by phosphorylation of a key serine (Ser133 in CREB) within a conserved subdomain by a number of protein kinases including PKA (for reviews see De Cesare et al. 1999; Shaywitz and Greenberg 1999). To test if the microinjected enzyme fractions exhibit catalytic activity in vivo according to their differential intracellular distribution, the phosphorylation of the key serine residue was analyzed by the use of antibodies that specifically recognize the phospho form of this epitope, i.e., CREB-(p)Ser133. For comparison, in vitro studies with recombinant CREB were carried out. In the test tube, the phosphorylation of recombinant CREB by both fractions did not differ as the Western blot analysis of electrophoresed samples revealed (data not shown), thus, supporting earlier observations that different PKA substrates were phosphorylated in vitro by both enzyme fractions to the same extent (Kinzel et al. 1987). For the in vivo phosphorylation of the CREB enzyme, fractions A and B (1 mg/ml) were microinjected in the cytoplasm of NIH 3T3 cells; after a 25-min incubation at 37°C, the cells were fixed. Microinjected cells were identified by the detection of coinjected IgGs (Fig. 8B and Fig. D) and simultaneously stained with primary anti–CREB and anti–CREB-(p)Ser133 antibodies (Fig. 8A and Fig. C). The CREB signal as such was not altered by the microinjection of both enzyme fractions (data not shown). The phospho-CREB signal in cells injected with fraction B was stronger than in cells injected with the enzyme fraction A (Fig. 8A versus C). The quantified phospho-CREB–specific signals in the nuclei of cells injected with different amounts of the enzyme fractions are plotted as a percentage of the maximal response seen (Fig. 8 E). The data demonstrate that up to an enzyme concentration of 2 mg/ml (in the microinjection pipette) CREB phosphorylation by fraction B appears to be more efficient during 25 min than by fraction A. At 3.5 mg/ml, the 25-min period is sufficient for fraction A to phosphorylate CREB to the same extent as fraction B, possibly indicating substrate limitation. In contrast to the phosphorylation mediated by fraction B, that effected by fraction A exhibits an almost linear concentration dependence up to 1 mg/ml. In this range, the largest differences in the phosphorylation of CREB by fraction A and B were observed. It should be noted that the two smallest concentrations injected (0.25 and 0.5 mg/ml) lead to and represent enzyme levels that are within the physiological range (Hofmann et al. 1977; Lohmann and Walter 1984). A small concentration was chosen to test the time course of CREB phosphorylation by both enzyme fractions (Fig. 8 F). For this purpose, NIH 3T3 cells were microinjected with 0.5 mg/ml of fraction A or B at time zero, and incubated at 37°C for various periods as indicated before they were fixed and stained for phospho-CREB analysis. The data show again that fraction B causes a more pronounced phospho-CREB appearance than fraction A. The 25-min values agree with those seen with the same concentration in the previous analysis (Fig. 8 E). The onset of CREB phosphorylation by enzyme fraction B appears to be faster than that effected by fraction A. The intensity of CREB phosphorylation peaks at ∼20–25 min after the injection of both enzyme fractions with a subsequent takeover of dephosphorylation. Taken together, the data on CREB phosphorylation obtained with both enzyme fractions mirror the differential intracellular distribution of the encoded and the deamidated forms of the PKA catalytic subunit, and illustrate at the same time functional consequences of Asn2 deamidation.
The biological significance of the NH2-terminal domain of the PKA C-subunit, a segment comprising ∼40 residues that precedes the catalytic domain, is unknown. In the crystal structure, it forms an amphipathic α-helix (Veron et al. 1993) and reaches from the large to the small lobe of the enzyme opposite the catalytic cleft (Knighton et al. 1991; Bossemeyer et al. 1993). The NH2 terminus myrGly-Asn-Ala is conserved among the isozymes Cα, Cβ, and Cγ, the latter being expressed in the testis (Beebe et al. 1990). We have shown previously that Cα and Cβ are prone to deamidation at Asn2, since about one-third of the C-subunit isolated from striated muscle of four different mammalian species (pig, cattle, rabbit, and rat) are in the myrGly-Asp-Ala configuration, which is an NH2-terminal sequence not so far present in protein data banks. There is evidence that deamidation of the C-subunit occurs in vivo (Jedrzejewski et al. 1998). The physiological significance of the Asp2 form is certainly not limited to the fact that it is more resistant towards a specific protease isolated from kidney brush border membranes than the Asn2 form (Kinzel et al. 1987). The finding that after microinjection, the Asn2 form reaches a significantly larger nuclear/cytoplasmic ratio than the Asp2 form constitutes evidence that the NH2 terminus of the C-subunit is critical for the intracellular distribution of the enzyme. This difference between the encoded and the deamidated form has been observed regardless of whether the C-subunit was labeled with FITC, TRITC, or was in unlabeled form. The N/C ratio in the plateau phase with the Asn2 form was ∼1.5–2.3 times greater than that observed with the Asp2 form. The difference in the intracellular distribution appears to be independent of the mammalian species, as enzymes from two different sources behave in the same manner, and the molecular parameters from two further mammalian species reproduce those seen with the former (Jedrzejewski et al. 1998). This apparent independence of the data from the cell type used suggests a general mechanism designed to discriminate between the two forms of the enzyme.
The relatively large amounts of microinjected C-subunit fractions required to measure the intracellular distribution may question the potential physiological significance of the results. However, the significance is supported by data on the site-specific phosphorylation of nuclear CREB, a reaction known to activate the transcription factor, elicited by the small enzyme amounts injected. The differential phosphorylation by fraction A and B is particularly well demonstrable at the small enzyme amounts, which yield intracellular concentrations in the physiological range. Under these conditions, nuclear substrate is probably not a limiting factor. The encoded, Asn 2 form of the PKA C-subunit is more efficient in that respect than the deamidated, Asp2 form. Since in vitro both enzyme fractions were shown to phosphorylate PKA substrates (Kinzel et al. 1987), including CREB equally effective, the data indicate that the differences observed in vivo are based on the uneven intracellular distribution; i.e., the larger nuclear level of the encoded C-subunit. The distribution data and the degree of protein phosphorylation in the nucleus correlate. Therefore, the differential distribution of both enzyme fractions observed at high concentrations is unlikely to be caused by this fact, but appears to reflect a physiological relevance.
The difference in the N/C ratios observed for fraction A and B might suggest in the first instance that it is the increase in electronegativity that causes a decrease of the nuclear enzyme level of fraction A. A simple charge difference as the driving force for the variant distribution of the Asn2 and the Asp2 form is unlikely, however, because of the fact that FITC labeling renders fraction B even more acidic than the unlabeled fraction A (Fig. 4), yet it, nevertheless, reaches larger N/C values than the latter. Therefore, it seems that a negative charge alone is not sufficient to explain the decrease in the nuclear level of the enzyme, but that it must probably be added at a specific site.
The Asn2 residue per se does not appear to constitute a determinant for the intracellular enzyme distribution as its deletion does not alter the N/C ratio of the deletion mutant when compared with the wild-type enzyme. It appears that Asn2 plays a role as a deamidation site only, conserved in Cα, Cβ, and Cγ and, thus, as a potential source for increasing electronegativity at the NH2 terminus, and perhaps with the further consequence of facilitating Ser10 autophosphorylation. Since a negatively charged residue next to Gly1 inhibits N-myristoylation (Towler et al. 1988), deamidation appears to be the only way to achieve the specific sequence myrGly1-Asp2 observed in fraction A.
The gain in electronegativity close to the NH2-terminal myristoyl moiety in proteins is reminiscent of myristoyl-electrostatic switch proteins (for review see, McLaughlin and Aderem 1995) in which electronegativity is achieved through phosphorylation aimed at influencing the interaction of these proteins with membranes. This is usually mediated by the myristoyl moiety partitioning into the lipid bilayer with positively charged residues interacting with acidic phospholipid head groups. Thus, it is possible that a lipophilic entity of membranous or proteinaceous nature may participate in the recognition and in the determination of the intracellular distribution of the C-subunit. It has been shown that the NH2 terminus of the PKA C-subunit in chimeric pp60v-src can partially substitute for the wild-type NH2 terminus in conferring membrane binding in vivo (Silverman and Resh 1992). However, this observation with the myristoylated NH2 terminus of the C-subunit within the foreign structural context of another protein may not necessarily reflect its natural role in PKA. A PKA-specific significance of its myristoylated NH2 terminus may be indicated by the finding that the rate of C-subunit binding to liposomes in vitro is dramatically increased on interaction with the regulatory subunit type II (RII) but not with type I (Gangal et al. 1999). Even these PKA-specific observations, however, are unlikely to play a role since the experiments in our study were carried out under saturating amounts of cAMP and with excess of free C-subunit sufficient to guarantee subunit dissociation and to exclude a role of R-subunit isoforms.
Results published so far on the cytoplasmic to nuclear translocation of the free C-subunit are compatible with a model in which the enzyme can traverse nuclear pores in either direction by diffusion (Harootunian et al. 1993). A functional nuclear localization signal has not been identified yet in C-subunits. In agreement with a diffusion model, the distribution is not particularly influenced by the concentration of the C-subunit. In our study, the unlabeled enzymes were injected at a four times larger concentration than the labeled enzymes. The N/C ratios of the plateau, however, reach comparable values characteristic for a specific isoform, indicating that it is not saturable, at least not within this range. However, the observation that both forms of the enzyme after ∼10 min reach an equilibrium which is different for the deamidated and the nondeamidated forms, seems to require supplementary assumptions for explanation.
Deamidation of Asn2 might, of course, result in changes of the higher ordered structure of the NH2 terminus. No noticeable structural differences are apparent between the high resolution structure of the fraction A enzyme (Bossemeyer et al. 1993) and the recombinant bovine (Engh et al. 1996) or mouse (Knighton et al. 1991) enzyme; however, the immediate area around Asn2 is not well resolved in the ternary structure (Bossemeyer et al. 1993; Zheng et al. 1993). Comparative secondary structure studies of synthetic myristoylated hexadecapeptides representing the bovine Cα NH2 terminus either in the Asn2 and the Asp2 form provided no evidence for significant structural differences at this level (Tholey, A., D. Bossemeyer, V. Kinzel, and J. Reed, unpublished data). This still does not mean that a different structure or flexibility might not occur. It could be induced by interaction with cellular target(s), a possibility indicated indirectly by RII-induced binding of C-subunits to liposomes in vitro, which was not seen with RI (Gangal et al. 1999). The charge shift to a more acidic character of the NH2 terminus could be sufficient to result in a preference for cytoplasmic structures that would retain fraction A to a certain extent in the cytoplasm, or a more favorable environment for the less acidic fraction B in the nucleus. The described differences in the sensitivity of fraction A and B to a specific protease in vitro (Kinzel et al. 1987) suggests that such differences are in principle recognizable at the molecular level. A differential export from the nucleus (Schmidt-Zachmann et al. 1993), mediated in the case of the C-subunit by protein kinase inhibitor protein (PKI) (Fantozzi et al. 1994), represents another possibility that requires further attention.
The fact that the different N/C ratios for fraction A and B are seen in the presence of saturating amounts of cAMP seems to argue against R-subunits as a retaining principle. The interaction of enzyme fractions A and B with RI or RII appeared to be indistinguishable in vitro (Kinzel et al. 1987). However, it is not excluded that the affinity of type II R-subunits for C-subunits in situ may be modulated by association with anchor proteins (Scott 1991; Hubbard and Cohen 1993) in such a way that the isozymes are bound to a different extent. A preferential binding of cell-endogenous PKI to the Asp2 form of the enzyme seems unlikely, however, as the cause for differential nuclear uptake since both enzyme fractions appeared to interact equally well with PKI, at least in vitro (Kinzel et al. 1987; Van Patten et al. 1988). Again, the isozymes might bind to cytoplasmic substrates to a different extent. The finding that the phosphorylation by fraction A and B of standard substrates including CREB in vitro was identical does not reflect the situation in the living cell (Lamb et al. 1988, Lamb et al. 1989; this study).
The example of the C-subunit may be typical of cases where proteins fulfill nuclear as well as cytoplasmic functions. Such proteins may adapt intermediate states of intracellular distribution by different mechanisms. One of them seems to be site-specific deamidation, as shown in this study. The circumstances in cells that control the partial deamidation of certain proteins are unknown and may even differ among candidate proteins; those for the C-subunit are unknown as well. However, they may play an important role in those cells or tissues in which the C-subunit is required to fulfill specific cytoplasmic functions. A preferentially deamidated state of the C-subunit may explain the lack of a significant redistribution of the enzyme to the nucleus upon treatment of certain cells with 8-bromo-cAMP (Murtaugh et al. 1982). The occurrence of comparable fraction A/fraction B ratios in muscle tissues of different mammalian species (Kinzel et al. 1987; Jedrzejewski et al. 1998) seems to point to a physiological significance of this modification. An example of protein-mediated deamidation of a cellular protein at a specific glutamine residue caused by a bacterial toxin has been described recently (Flatau et al. 1997; Schmidt et al. 1997). Moreover, several myristoylated proteins involved in signal transduction share the NH2-terminal myr-Gly-Asn sequence and, thus, may be candidates for partial deamidation (listed by Jedrzejewski et al. 1998).
Although the isoAsp2 form of the enzyme cannot be evaluated separately, its presence in trace amounts in fraction A is worth mentioning in the context of this study. The occurrence of isoAsp2 in the C-subunit fraction A points to a mechanism of deamidation through the so-called β-aspartyl-shift mechanism (for review see Wright 1991). It involves the formation of a cyclic succinimide intermediate that opens to yield preferentially the isoAsp form (to a lesser extent the Asp form) (for reviews see Bornstein and Balian 1977; Aswad and Johnson 1987; Geiger and Clarke 1987; Lura and Schirch 1988; Artigues et al. 1990), which carries the original Cβ-methylene of Asn2 now in the peptide backbone. The inverse ratio of isoAsp/Asp in tissues, with a clear-cut surplus of the Asp form, is probably due to the action of a ubiquitous repair enzyme, protein carboxyl methyltransferase, which methylates the atypical α-carboxyl group of l-isoAsp residues (Aswad 1984; Murray and Clarke, 1989) as a part of a mechanism aimed at the reconstitution of the regular peptide backbone. For a more detailed discussion of this relation with respect to the C-subunit see Jedrzejewski et al. 1998. Cases of site-specific nonenzymatic deamidation of proteins have been reported, which appear to be involved in aging and subsequent degradation of proteins. However, in certain proteins, this type of modification appears to be part of their physiological program (for review see Wright 1991). The conserved Asn2 deamidation of the PKA C-subunit with signs of ample repair (to yield predominantly the Asp 2 form) may belong to the latter group. This appears to be supported by the differential in vivo phosphorylation of the CREB family transcription factor(s) effected by the encoded and the deamidated C-subunit.
In conclusion, this study with living cells allows for the first time to consider different physiological functions of the major C-subunit isozymes of PKA (generated by deamidation of Asn2) on the basis of their differential distribution within cells, and points to a potential physiological significance of the NH2-terminal domain of this enzyme, and possibly of other proteins with a similar NH2 terminus for which the C-subunit may represent a model system. Further studies are aimed at the elucidation of the control of deamidation and of the molecular mechanism responsible for the altered intracellular distribution of Asn2 deamidated catalytic subunit.
We thank Dr. Wolfgang Schmid and Daniel Gau (both from Division of Molecular Biology of the Cell I, German Cancer Research Center, Heidelberg, Germany) for anti-CREB and anti–CREB-(p)Ser133 antibodies as well as for recombinant CREB protein, Dr. Jennifer Reed for critically reading the manuscript and semantic assistance, and Angelika Lampe-Gegenheimer (both from Department of Pathochemistry, German Cancer Research Center, Heidelberg, Germany) for expert secretarial assistance.
The work was supported by the Deutsche Forschungsgemeinschaft.
Abbreviations used in this paper: C, catalytic; CREB, cAMP-responsive element binding protein; N/C, nuclear/cytoplasmic fluorescence; NMT, N-myristoyl transferase; PKA, protein kinase A or cAMP-dependent protein kinase; PKI, protein kinase inhibitor; R, regulatory.