Muscle-regulated expression and determinants for neuromuscular junctional localization of the mouse RIalpha regulatory subunit of cAMP-dependent protein kinase

Abstract

In skeletal muscle, transcription of the gene encoding the mouse type Ialpha (RIalpha) subunit of the cAMP-dependent protein kinase is initiated from the alternative noncoding first exons 1a and 1b. Here, we report that activity of the promoter upstream of exon 1a (Pa) depends on two adjacent E boxes (E1 and E2) in NIH 3T3-transfected fibroblasts as well as in intact muscle. Both basal activity and MyoD transactivation of the Pa promoter require binding of the upstream stimulating factors (USF) to E1. E2 binds either an unknown protein in a USF/E1 complex-dependent manner or MyoD. Both E2-bound proteins seem to function as repressors, but with different strengths, of the USF transactivation potential. Previous work has shown localization of the RIalpha protein at the neuromuscular junction. Using DNA injection into muscle of plasmids encoding segments of RIalpha or RIIalpha fused to green fluorescent protein, we demonstrate that anchoring at the neuromuscular junction is specific to RIalpha subunits and requires the amino-terminal residues 1-81. Mutagenesis of Phe-54 to Ala in the full-length RIalpha-green fluorescent protein template abolishes localization, indicating that dimerization of RIalpha is essential for anchoring. Moreover, two other hydrophobic residues, Val-22 and Ile-27, are crucial for localization of RIalpha at the neuromuscular junction. These amino acids are involved in the interaction of the Caenorhabditis elegans type Ialpha homologue R(CE) with AKAP(CE) and for in vitro binding of RIalpha to dual A-kinase anchoring protein 1. We also show enrichment of dual A-kinase anchoring protein 1 at the neuromuscular junction, suggesting that it could be responsible for RIalpha tethering at this site.

Figure 1

E box-dependent basal activity and MyoD transactivation of the mouse RIα Pa promoter. (A) Structure of the 5′ region of the mouse RIα gene, containing five alternative promoters (Pa to Pe). Untranslated first exons are given as open boxes. The solid box represents coding sequences. Bent arrows indicate transcription start sites upstream of the first exons. Solid bars underline the region corresponding to the Pa and Pb promoters, which are active in adult skeletal muscle; the open bars indicate the other RIα promoters (Pc, Pd, and Pe). Arrowheads show the positions of N boxes. (B) Schematic representation of the Pa-nLacZ construct, containing the Pa promoter upstream of the β-gal reporter gene to which a nuclear localization signal has been added (nLacZ). The E and N boxes are represented as open and solid rectangles, respectively, and the consensus binding sites for ubiquitous transcription factors are indicated as circles. Sequences of the E and N boxes and the Sp1-binding site are underlined, and mutations introduced into the E boxes are given. (C) Effects of the mutations on basal and MyoD-enhanced expression of the β-gal reporter gene. (Left) Normalized β-gal activities of total extracts of injected, nonregenerating tibialis anterior muscles expressed as mean values relative to the activity of nonmutated Pa-nLacZ (set to 100). Numbers of injected muscles are indicated in parentheses. (Right) Normalized β-gal activities of NIH 3T3 fibroblast cells cotransfected with the β-gal reporter constructs and different amounts of a MyoD expression vector (solid, shaded, and open histograms correspond to 0, 0.5, and 2 μg, respectively). Activities relative to Pa-nLacZ (set to 100) are mean values (SD of two independent experiments with duplicate samples).

Figure 2

Proteins binding to the E1 and E2 boxes of promoter Pa. NIH 3T3 fibroblast and C2/7 myotube nuclear extracts were incubated with radiolabeled, double-stranded oligonucleotides encompassing the E1 box (A), the E2 box or mckR (B), and E1 + E2 (C). Complexes were resolved in 4% acrylamide gels. Competitions were performed with a 10- and 50-fold molar excess of unlabeled oligonucleotide MLP (USF-binding site) or mckR (MyoD-binding site). Supershifted complexes resulting from preincubation with either MyoD, USF1 (U1), or USF2 (U2) antibodies are indicated by arrows.

Figure 3

Dependence on USF of Pa basal and MyoD-mediated activity. NIH 3T3 cells were cotransfected with Pa-nLacZ and different amounts (0.05, 0.5, and 2 μg) of expression vectors encoding truncated forms of USF1 either lacking the DNA-binding (DbT1) or the transactivation domain (TDU1). To each sample, 2 μg of the MyoD expression vector (Lower) or the vector lacking the MyoD-coding region (Upper) was added. β-Gal activities, expressed relative to activity in the absence of MyoD and truncated USF proteins (set at 100), are mean values of two experiments.

Figure 4

Accumulation of RIα-GFP fusion proteins and of D-AKAP1 at the NMJ. (A) The dimerization domain of RIα, but not of RIIα, is able to direct NMJ localization. Plasmids encoding full-length or truncated RIα-GFP or RIIα(1–44)-GFP fusion proteins were injected into tibialis anterior muscles. Whole fibers expressing GFP were dissected and stained with tetramethylrhodamine B isothiocyanate-α-bungarotoxin to reveal the NMJ (open arrowheads). A fiber negative for GFP is shown in Lower Right. The arrow points to a nucleus in the multinucleated muscle fiber. (B) Confocal microscopy images of an adult muscle NMJ double-stained with anti-D-AKAP1 antibody and FITC-α-bungarotoxin. The photos show a single optical slice. (Bar = 5 μm.)

Figure 5

Structural determinants of RIα required for its NMJ enrichment. Percentage of coincidence of tetramethylrhodamine B isothiocyanate-α-bungarotoxin and GFP fluorescence was obtained with plasmids encoding full-length RIα(pRIα-GFP), N-terminal sequences of RIα(residues 1–91 or 1–81) or of RIIα (residues 1–44), and single amino acid substitutions of full-length RIα (C18A, V22A, I27A, F54A). The number of GFP-positive fibers examined for each construct is indicated in parentheses.