Fig. 1. DM1 (CUG)_n expansion mutation. (A) The structure of DMPK mRNA is illustrated with the positions of the DMPK coding region (stippled box), the 3′-UTR CUG repeat region (black box) and the poly(A) tail [(A)_n] indicated. Also highlighted are CUG repeats corresponding to normal, pre-mutant and mutant DM1 RNAs. (B) (CUG)_n RNAs showing that CUG repeats ≥20 spontaneously fold into dsRNA hairpins while (CUG)_≤10 RNAs are primarily single stranded. (C) Multiple 38–45 kDa proteins crosslink preferentially to DMPK 3′-UTR RNAs with large CUG repeats. DMPK 3′-UTR RNAs containing either 6, 54 or 90 (6s, 54s, 90s) CUG repeats, or an antisense transcript containing six CAG (6as) repeats, were labeled with [α-³²P]UTP, incubated in HeLa nuclear extract, photocrosslinked, digested with RNase and fractionated by SDS–PAGE. A bracket indicates the position of the EXP proteins. Sizes are indicated in kilodaltons. (D) Crosslinking of EXP proteins to dsCUG RNAs is proportional to the number of CUG repeats. RNAs composed of 11, 20, 35, 74 or 97 CUG repeats were labeled with [α-³²P]UTP and used for photocrosslinking analysis. The relative amount of EXP crosslinking was quantified by PhosphorImager analysis and is illustrated together with a 35 kDa protein that binds preferentially to shorter CUG repeats.

Fig. 2. EXP proteins are novel dsCUG-binding factors. (A) EXP proteins crosslink directly and preferentially to large CUG repeat RNAs. Photocrosslinking was performed as in Figure 1C except that RNAs were synthesized in the presence of [α-³²P]GTP and transcribed from pCTG₁₀, pCTG₅₄, pCAG₁₀ or pCAG₅₄ which do not contain DMPK sequences. The bracket indicates EXP protein position. (B) EXP proteins do not crosslink to the HIV-1 TAR RNA. Plasmids containing either 90 CUG repeats [(CUG)₉₀], the TAR RNA sequence (TAR) or a mutant TAR (mTAR) were transcribed in the presence of [α-³²P]UTP and photocrosslinking performed as described in (A). The first three lanes indicate binding activity seen with each labeled RNA. In the last two lanes, labeled (CUG)₉₀ was mixed with a 500-fold molar excess of unlabeled (CUG)₉₀ or unlabeled TAR RNA prior to addition to HeLa nuclear extract. (C) EXP proteins are not immunologically related to CUG-BP1 or hnRNP C proteins. Immunopurifications of EXP crosslinked proteins were performed using mAbs 3B1 (anti-CUG-BP1) and 4F4 (anti-hnRNP C).

Fig. 3. Purification of HeLa cell EXP proteins. (A) Purification scheme. The initial purification step included digestion of HeLa cell nuclear extract (HeLa NE) with micrococcal nuclease (MN) followed by DEAE Fast-Flow Sepharose chromatography and collection of the flow-through (FT). Final fractions are EXP1 eluate fraction 1 (eF1) and EXP2 eluate fraction 1 (eF1). (B) Label transfer activities of various fractions (top panel). Fractions were incubated with ³²P-labeled (CUG)₉₀ RNA followed by photocrosslinking analysis. The bottom panel is a silver-stained SDS–polyacrylamide gel showing the protein composition of each fraction. (C) Silver-stained SDS–polyacrylamide gel showing purified EXP1 and EXP2 fractions. Proteins from the EXP1 (bracket) and EXP2 (asterisk) regions indicated in (B) were excised and refractionated by SDS–PAGE. (D) Anti-hEXP42 polyclonal antibodies recognize two minor proteins in HeLa nuclear extract and a single protein in normal and DM1 lymphoblast nuclear extracts. An immunoblot using the anti-CUG-BP1 mAb 3B1 is also shown as a protein loading control. (E) EXP activity is immunopurified with anti-hEXP42 antibodies. Crosslinking reactions with (CUG)₁₀ and (CUG)₅₄ RNAs and HeLa nuclear extracts were performed as described in Figure 2A except that in vitro transcription was performed with [α-³²P]UTP. Immunopurifications were performed using either the anti-hEXP42 antiserum, a non-immunized control antiserum, or mAb 3B1 (anti-CUG-BP1). The last lane is a crosslinking reaction using ³²P-labeled (CUG)₉₀ with 50 ng of purified His-tagged hEXP42 in the absence of HeLa nuclear extract.

Fig. 4. EXP proteins are homologous to the Drosophila muscleblind family. (A) Alignment of the human hEXP42/MBNL, hEXP40/KIAA0428 and hEXP35 proteins with the largest of the four muscleblind proteins, mblB. Identical amino acid positions are highlighted by black boxes while similarities are shaded. (B) Illustration showing the organization of the four C₃H motifs (C₃H I–IV) and a potential transmembrane domain (TM) in the EXP proteins. (C) Poly(A)⁺ RNA (2 µg/lane) blot analysis of human tissues showing high EXP expression in cardiac and skeletal muscle. Lanes were standardized for β-actin levels, and sizes are indicated in kilobases.

Fig. 5. Characterization of EXP proteins in mouse tissues and during myoblast differentiation. (A) Immunoblot analysis using anti-hEXP42 antibodies showing that the EXP proteins are expressed in muscle and eye. Equal protein loads (15 µg/lane) were confirmed by Coomassie Blue staining (data not shown). (B) EXP expression is induced during myoblast differentiation. Immunoblot showing myosin heavy chain (αmyosin HC), mEXP42 (αhEXP42) and CUG-BP1 (αCUG-BP1) protein levels prior to (Day 0) and during (Days 1–5) myoblast differentiation. (C) Phase and immunofluorescence microscopy showing two examples of multinucleated myotubes, present on Day 4 following induction of myoblast differentiation, as well as undifferentiated myoblasts.

Fig. 6. EXP proteins accumulate in the nucleus of DM1 cells. (A) Cell immunofluorescence of normal and DM1 myoblasts using anti-EXP polyclonal antibodies. Phase contrast microscopy highlights cell position while DNA was detected with DAPI. The arrow indicates one of the EXP-enriched foci. (B) FISH analysis with a Cy3-labeled (CAG)₁₀ probe (red), which detects DMPK (CUG)_n expansions in DM1 cell nuclei (shown by DAPI staining). (C) Cell immunofluorescence, using anti-EXP antibodies, of normal fibroblasts infected with a MyoD adenovirus (normal fibroblast + MyoD) or DM1 fibroblasts either uninfected (DM1 fibroblast – MyoD) or infected (DM1 fibroblast + MyoD). Arrows indicate several sizes of foci enriched in hEXP42.

Fig. 7. RNA dominance model for DM1 pathogenesis. The processing and nucleocytoplasmic export of DMPK transcripts from the normal and mutant alleles are illustrated. This model proposes that the normal RNA-binding sites for the EXP proteins are distributed in both the nucleus and cytoplasm.