The role of CNBP in brain atrophy and its targeting in myotonic dystrophy type 2

Abstract

Myotonic Dystrophy type 2 (DM2) is a multisystem disease affecting many tissues, including skeletal muscle, heart, and brain. DM2 is caused by unstable expansion of CCTG repeats in an intron 1 of a gene coding for cellular nuclear binding protein (CNBP). The expanded CCTG repeats cause DM2 pathology due to the accumulation of RNA CCUG repeats, which affect RNA processing in patients’ cells. We have previously shown that mutant CCUG repeats reduce CNBP protein in DM2 patients. Reducing Cnbp in Cnbp KO mouse model causes late skeletal muscle atrophy. In this study, we examined if the reduction of Cnbp affects the Central Nervous System (CNS). MRI and DTI analyses showed that total brain volume and grey matter are reduced in Cnbp KO mice, while mean, radial and axonal brain diffusivity is increased. The morphological changes in the brains of Cnbp KO mice are accompanied by reduced stereotypic behavior, anxiety and neuromotor defects. These findings suggest that the reduction of CNBP contributes to CNS pathology in DM2. Since CNBP stability is regulated by pAMPK-dependent phosphorylation, we examined protein levels of pAMPK in DM2 cells and found that the active pAMPK is reduced in DM2. Interaction of CNBP with pAMPK and stability of CNBP protein are also decreased in DM2. Our data show that a small molecule AMPK activator A769662 corrects CNBP stability and normalizes CNBP targets in DM2 fibroblasts. Thus, activators of AMPK could potentially be developed as therapeutics to correct CNBP and reduce muscle and brain atrophies in DM2.

Graphical Abstract

Graphical Abstract.

Introduction

Myotonic Dystrophy type 2 (DM2) (OMIM #602668) is a multisystemic disease, characterized by muscle atrophy and weakness, myotonia, cardiac defects, insulin resistance and decline of CNS function including brain atrophy, intellectual disability, daytime sleepiness, behavioral defects (avoidant personality), significant defects in executive functions; and problems with concentration [1–6]. DM2 affects 1 in 8000 people between 30 and 60 years of age in some regions of the world. DM2 is caused by an expansion of unstable CCTG repeats in the intron 1 of the gene encoding CCHC-type Zinc Finger Nucleic Acid-Binding Protein (CNBP) (also known as Zinc Finger Protein 9, ZNF9) [7]. The multisystemic DM2 phenotype is like the phenotype of DM1, a neuromuscular disease caused by unstable CTG repeats in the 3’ UTR of DMPK gene [8]. However, DM2 pathology affects mainly proximal muscles, whereas DM1 affects distal muscles [1, 9]. In DM1 and DM2, the mutant transcripts (CUG and CCUG) accumulate in patients’ cells, mis-regulating RNA metabolism via specific RNA-binding proteins [10–12]. Despite these similarities, there are several important differences between DM1 and DM2 pathogeneses. These differences include the lack of correlation between the length of CCTG repeats and severity of DM2; the lack of congenital and childhood forms in DM2; the incidence of DM2 patients with short expansions in the pre-mutation range (40–50 CCTG repeats) [13]; fewer MBNL1 aggregates in DM2 cells than in DM1 cells [14]; distinct or only partially overlapping with DM1 splicing patterns [14, 15] and a lack of correlation between specific splicing events in DM2 with the progression of muscle histopathology [16]. These differences and the lack of identical clinical phenotypes in DM1 and DM2 suggest that other factors downstream of the mutant CCUG repeats might be involved in DM2 pathophysiology. One of these candidates is the CNBP protein, encoded by the mutant CNBP gene.

CNBP is a multifunctional protein which is expressed in skeletal muscle and in non-muscle tissues. It has RNA- and DNA-binding activities and regulates many targets, mainly at the level of translation and transcription [17–26]. Several reports showed that CNBP, its activity or mRNA processing are altered in patients with DM2 [17–20, 27–31]. CNBP reduction is one of the primary events of the accumulation of the CCUG-containing mutant RNA which occurs at the same time as the accumulation of CCUG foci [29]. Some early studies did not find differences in CNBP expression in DM2 [32–34]; however, recent investigations of the role of CNBP in DM2 showed that the expanded CCUG repeats alter the processing of the mutant CNBP mRNA [27] and that CNBP protein is reduced in DM2 muscle biopsies [28].

The role of Cnbp in DM2 pathogenesis was studied independently of CCUG expansion in Cnbp KO mice. The first Cnbp KO mouse model showed some DM2-like phenotype in muscle, heart, and eye [35]. However, these mice had very strong underdevelopment, including mortality of homozygous mice. The phenotype of our Cnbp KO mouse model is much milder [19] than the one in the earlier generated Cnbp KO model [35]. Although surviving homozygous Cnbp KO mice (around 40%) in our strain are smaller than heterozygous mice, they live until 16–24 months. Heterozygous Cnbp KO mice develop normally [19]. Both homozygous and heterozygous Cnbp KO mice show muscle weakness, accompanied by progressive muscle atrophy, mimicking the slow progression of muscle pathology in DM2, worsening with age.

In this study, we investigated whether Cnbp KO mice show CNS abnormalities. Also, searching for the pathways correcting CNBP in DM2, we investigated whether pAMPK kinase, which regulates CNBP stability is altered in DM2 human cells and whether AMPK activator corrects CNBP stability in human cells derived from patients with DM2.

Results

Brain atrophy in Cnbp KO mice

Since patients with DM2 are characterized by late brain atrophy [2–6], we examined if we could find similar defects in Cnbp KO mice. Magnetic resonance imaging (MRI) was used to calculate brain volumes in Cnbp KO mice. This analysis showed a significant reduction of the cerebrospinal fluid, grey matter (GM) and total brain volumes in adult (6-month-old) homozygous Cnbp KO mice (females) vs matching WT mice (Fig. 1A–C). The same volumetric parameters were also reduced in 6-month-old heterozygous Cnbp KO mice, but only a reduction of the GM volume was close to significant value (P = 0.054). We expect that older heterozygous Cnbp KO mice might have stronger brain atrophy.

Figure 1.

(A–C) Volumetric analysis of brain morphology (CSF, GM, and total brain volumes) in the matched (females, 6 months) WT (n = 9), heterozygous (n = 11) and homozygous (n = 4) Cnbp KO mice by MRI. ^* and ^** are P values <  0.05 and < 0.01. The number of mice per group and SD are shown.

We next examined the status of white matter (WM) in Cnbp KO mice by DTI imaging. A region-of-interest (ROI) based assessment of manually drawn ROIs in the brain of WT and homozygous Cnbp KO mice is shown in Fig. 2A–F. We found that the mean diffusivity (MD) (P value <  0.0001), radial diffusivity (RD) (P value < 0.001) and axial diffusivity (ad) (P value < 0.001) are strongly increased in right fimbria (RF) of homozygous 6-month-old Cnbp KO mice (females) vs matching WT mice (Fig. 2A–C). Fractional anisotropy (FA) is reduced in the left external capsule, left fimbria, and in corpus callosum (CC) in homozygous Cnbp KO mice vs WT littermates (P values < 0.01) (Fig. 2D–F). These changes in diffusion metrics suggest a disturbance of the WM integrity in homozygous Cnbp KO mice.

Figure 2.

(A–C) The increase of mean diffusivity, radial diffusivity, and axial diffusivity in right fimbria in 6-month-old homozygous Cnbp KO mice (all females) vs matched WT mice. (D-F) the reduction of fractional anisotropy in left external capsule, left fimbria, and corpus callosum in 6-month-old homozygous Cnbp KO mice (females) vs matched WT mice. The number of analyzed mice is shown on the top. Standard deviations are also shown. ^**, ^*** and ^**** are P values <  0.01, < 0.001 and < 0.0001 correspondently. (G) TBSS analyses of RD (upper image) and MD (lower image) in the ROI of the brains of homozygous Cnbp KO vs WT mice. Arrows show examples of locations where homozygous Cnbp KO mice have higher RD (upper panel) or higher MD (lower panel) than WT matching mice. P-values from 0.05 to 0.001 are indicated by the bar intensity shown on the right. Images are arranged in sequential slices through the brain from ~ bregma −3 mm to bregma 1.3 mm.

A significant increase in MD in homozygous Cnbp KO mice vs WT mice was also found.

in CC, right external capsule (REC), left fimbria (LF) and right internal capsule (RIC) (P < 0.05 for all mentioned regions). In contrast, ad was increased in CC (P = 0.03) (not shown). RD was significantly (P < 0.05) increased in homozygous Cnbp KO mice vs WT mice in CC, REC, LEC and LF (data not shown).

To verify these findings in an operator-independent manner, voxel-wise statistical analysis of the diffusion data was carried out using tract-based spatial statistics (TBSS) [36], part of FSL [37], with minor modifications for mouse data. TBSS projects all the mouse FA data onto a mean FA skeleton (Fig. 2G, green) before applying voxel-wise cross-subject statistics to each derived diffusion parameter. This analysis confirmed the significant increase of RD (Fig. 2G, upper) and MD (Fig. 2G, lower) in 6-month-old homozygous Cnbp KO brains vs WT brains (Fig. 2G, red and yellow signals), suggesting the WM disturbance in Cnbp KO mice.

Behavioral dysfunction of Cnbp KO mice

The structural changes in the brains of Cnbp KO mice suggest that these mice might have cognitive and behavioral abnormalities. Therefore, we began examining the behavioral and neuromotor outcomes in young Cnbp KO mice using the Open Field Test. As shown (Fig. 3A–D), stereotypic behavior is significantly reduced in 1-month-old homozygous Cnbp KO mice (males) (n = 4) vs matching WT mice (n = 9). While stereotypic episode count, stereotypy time and stereotypy activity count are significantly reduced in homozygous mice vs WT mice, these outcomes are also reduced in heterozygous mice. Still, they were not significant or close to significant values likely due to very high variability of behavioral outcomes in both WT and heterozygous Cnbp KO mice. Stereotypy episode activity count was significantly reduced in young heterozygous Cnbp KO mice vs matching WT mice (Fig. 3D). This reduction was close to significant in homozygous Cnbp KO mice. There were no significant changes in neuromotor activity and anxiety, evaluated by the time spent in the center of the box during the OFT analysis in 1-month-old heterozygous or homozygous Cnbp KO mice (males). This could be due to the high variability of behavioral parameters in young WT and heterozygous Cnbp KO mice and the low number of homozygous mice in this strain since only around 40% of homozygous mice survived [19].

Figure 3.

Young Cnbp KO mice are characterized by atypical behavior. (A-D) stereotypic episode count, stereotypy time, stereotypic activity count and stereotypic episode activity count, measured by the OFT test during 5 min travel, are reduced in Cnbp KO mice. WT (n = 9), heterozygous (n = 18) and homozygous (n = 4) mice (males) were compared. (E) Total distance y axis is reduced in 1-month-old heterozygous Cnbp KO mice (females). WT (n = 10) and heterozygous (n = 21) mice were compared. (F) Stereotypic episode activity count is reduced in adult, 5-month-old heterozygous Cnbp KO mice (n = 3) vs WT littermates (n = 3) (males). ^** and ^{** *} are P values < 0.05 and < 0.01, respectively.

One-month-old heterozygous Cnbp KO females (n = 21) show some neuromotor problems, such as a reduction of total distance, y-axis, vs matching WT littermates (n = 10) (Fig. 3E). Like homozygous males, the number of 1-month-old homozygous females was low for the behavioral analysis. Heterozygous Cnbp KO mice continued to display a reduced stereotypic behavior in adulthood as stereotypic episode activity count was significantly diminished in 5-month-old heterozygous Cnbp KO mice (males) vs matching WT littermates (Fig. 3F).

One of the features of DM2 behavioral dysfunction is the appearance of anxiety. As noted, 1-month-old heterozygous and homozygous Cnbp KO mice showed no significant changes in the time spent in the center of the Open Field Box. Since behavioral outcomes in WT and heterozygous Cnbp KO mice vary from mouse to mouse in different families, we compared the time spent in the center of the box using the OFT test in the same groups of WT and heterozygous Cnbp KO mice at 1 month and at 5 months of age. As shown in Fig. 4A, there was no significant difference in the time traveled by 1-month-old WT and heterozygous Cnbp KO mice (males) in the center of the box using the Open Field test. The same group of WT mice at 5 months of age travelled longer in the center of the box, likely due to a better adaptation of mature WT mice. In contrast, in 5 months, heterozygous Cnbp KO mice travelled almost three times less in the center of the box compared to WT mice (Fig. 4A and C). The reduction of the time travelled in the center of the box by 5-month-old heterozygous Cnbp KO mice was not due to reduced total movement time because both WT and heterozygous Cnbp KO mice showed similar movement time during the OFT test (Fig. 4B). Thus, adult Cnbp KO mice develop anxiety.

Figure 4.

Anxiety in adult Cnbp KO mice. (A) the time spent in the center of the box and the movement time (B) during the OFT test was compared in the same groups of WT and heterozygous Cnbp KO mice at 1 and 5 months of age. (C) Examples of the heatmap tracking the movement of WT and matched heterozygous Cnbp KO mice during the OFT. ^* is P value < 0.05. The standard deviations and the number of mice per group are shown.

Cnbp KO mice show reduced neuromotor activities with age

While homozygous Cnbp KO mice have more severe skeletal muscle and brain phenotypes than heterozygous mice, the number of homozygous Cnbp KO mice is low, making it difficult to use homozygous mice for detailed behavioral studies during ageing. Since heterozygous Cnbp KO mice have the same levels of Cnbp as those in patients with DM2, we examined if anxiety and neuromotor outcomes are worsening with age in heterozygous Cnbp KO mice using the OFT. In these experiments, the same groups of WT and heterozygous Cnbp KO mice were analyzed at 5 and 8 months of age. As shown (Fig. 5A), the center time travelled during 5 min within the box during the OFT test is reduced by about 4.5-fold in 8-month-old heterozygous Cnbp KO mice vs 5-month-old heterozygous Cnbp KO mice. The center time value in the 8-month-old heterozygous Cnbp KO mice was adjusted to the reduction of the total movement time in the 8-month-old heterozygous group (Fig. 5B).

Figure 5.

Anxiety and neuromotor defects are increased in heterozygous Cnbp KO mice with age. Center time (A), movement time (B), ambulatory activity count (C), horizontal activity counts (D), rest time (E) and rest episode count (F) are worsening in 8-month-old heterozygous Cnbp KO mice vs the same group of heterozygous mice at 5 months of age. The center time in the 8-month-old heterozygous Cnbp KO mice, shown in a, was adjusted to the reduced total movement time in the same group of mice, shown in B. ^* and ^** are P values <  0.05 and < 0.01, respectively.

It is important that at younger age (1 and 5 months), heterozygous Cnbp KO mice have no or very minor neuromotor defects. However, at 8 months of age, some of these defects become significant. Consistent with the reduced movement time, 8-month-old heterozygous Cnbp KO mice need more rest time than 5-month-old mice (Fig. 5E). Respectively, the rest episode count is increased in 8-month-old heterozygous Cnbp KO mice vs 5-month-old mice (Fig. 5F). Ambulatory activity count and horizontal activity count are also reduced in 8-month-old heterozygous Cnbp KO mice compared to the same mice at 5 months of age (Fig. 5C and D). These data show that heterozygous Cnbp KO mice develop neuromotor defects with age. Anxiety is also increasing in these mice (Fig. 5A). These findings agree with the increased severity of the neuromuscular and CNS defects in older patients with DM2.

CNBP reduction in DM2 correlates with muscle weakness and with patients’ age

DM2 is an ageing disease affecting patients in their sixties [38, 39]. To determine if CNBP expression varies in patients of different ages, we examined CNBP levels in DM2 muscle samples from patients aged 33, 41, 53 and 56 years by Western blot assay. This analysis showed that CNBP levels are ~ 1.5–4-fold reduced in all DM2 muscle samples vs normal controls (Fig. 6A and B). The smallest reduction of CNBP was observed in a patient of 33 years of age (about 1.4-fold). However, in patients of 41–56 years of age the CNBP reduction was approximately 4-fold compared to normal control patients. A younger patient of 33 years (DM2–1) has mild muscle weakness, while the muscle weakness in patients DM2–2—DM2–5 of 41–56 years was moderate. Thus, although we have muscle biopsy from only one patient of 33 years of age (the number of DM2 patients in their thirties is low), it seems that there is a reverse correlation between age and muscle weakness on one side and the degree of CNBP reduction on another side (Fig. 6C). Therefore, in addition to CCTG repeats, age might contribute to the reduction of CNBP in DM2 patients, resulting in a stronger effect on the muscle and CNS defects with age.

Figure 6.

CNBP is reduced in DM2 muscle biopsies. (A) Western blot of total protein extracts from muscle biopsies (vastus lateralis) of normal controls of 31 and 45 years and DM2 patients of 33, 41, 53, 53 and 56 years. For CNBP, two film exposures are shown (short and long) because the CNBP signals in older DM2 patients (samples 2–5) are weak relatively CNBP signal in DM2 sample #1 from a younger patient (33 years). Re-probing of the membrane with antibodies to actin shows equal loading. Note that differences in the shape of the CNBP and actin bands are due to the differences of the molecular weights of CNBP (20 kD) and actin (42 kD). Coomassie staining shows protein integrity and shape of the bands in positions of 42 kD (actin) and 20 kD (CNBP). (B) Bar graphs for CNPB levels, adjusted to actin, shown in (A). Average values of CNBP signal for two normal controls and four DM2 patients of 41–56 years are shown at the top. ^**P value (<0.01) for the average CNBP signal in DM2 patients vs normal controls. (C) CNBP reduction in this group of patients correlates with the age and severity of muscle weakness.

The reduction of CNBP protein in DM2 myoblasts and fibroblasts correlates with the reduction of active pAMPK

A reduction of CNBP in DM2 patients and the appearance of late muscle atrophy, anxiety and neuromotor abnormalities in adult heterozygous Cnbp KO mice suggest that the correction of CNBP levels in DM2 can reduce atrophy in muscle and in the brain. We have previously shown that CNBP mRNA levels are unchanged in myoblasts and fibroblasts derived from patients with DM2 [29, 40], while CNBP protein is reduced [17–19, 29]. The reduction of CNBP protein in DM2 could be due to the reduced stability or reduced protein synthesis. It has been shown that CNBP stability in human brain cancer, medulloblastoma, is regulated by AMPK via phosphorylation at T173 [41]. AMPK is a critical energy sensor, regulating many processes, including cell metabolism, growth, glucose uptake and mitochondria homeostasis [42]. Important, that AMPK signaling is affected in DM1 [43, 44], a disease related to DM2. To test if AMPK interacts with CNBP in normal cells and if this interaction is reduced in DM2, the levels of total and active, phosphorylated at T172 pAMPK were measured in human normal and DM2 fibroblasts and myoblasts by Western blotting. We found that total levels of AMPK are comparable in normal and in DM2 fibroblasts and myoblasts; however, active pAMPK was undetectable in both DM2 fibroblasts and myoblasts (Fig. 7A and B). CNBP levels were decreased in the same DM2 human cell lines with reduced pAMPK. We found that active pAMPK strongly interacts with CNBP in normal fibroblasts, but its interactions with CNBP are undetectable in DM2 fibroblasts (Fig. 7A, bottom panel). Thus, we conclude that pAMPK is reduced in DM2 cells and that the reduction of pAMPK correlates with the reduction of CNBP. Interaction of pAMPK with CNBP is also reduced in cells derived from DM2 patients.

Figure 7.

(A) Western blot analyses of total and pAMPK and CNBP in normal and DM2 human fibroblasts (a, top panel) and normal and DM2 human myoblasts (B). B-actin is a loading control. Bottom panel in a: The same protein extracts from normal and DM2 fibroblasts, shown on the top panel (input) were used for the IP-western blot assay. pAMPK was immunoprecipitated and the levels of CNBP in the pAMPK-IPs were determined by western blot with antibodies to CNBP.

Small molecule activator of AMPK A769662 stabilizes CNBP in DM2 fibroblasts and corrects CNBP levels

Since levels of pAMPK and its interaction with CNBP are reduced in DM2 human cells (Fig. 7A and B) and because in medulloblastoma, CNBP stability is controlled by phosphorylation at T173 by AMPK [41], we thought that the reduction of pAMPK in DM2 cells also reduces CNBP stability. To examine this suggestion, we compared CNBP stability in normal fibroblasts and in DM2 fibroblasts using the inhibitor of protein synthesis cycloheximide, CHX (Fig. 8A and B). Normal and DM2 fibroblasts were maintained in the growth medium at about 80% density. CHX was added to the media and protein levels of CNBP were measured at 1, 2, 3 and 5 hours after addition of CHX. As shown in Fig. 8A and B, the half-life of CNBP in normal fibroblasts is about 4.75 h while the half-life of CNBP in DM2 fibroblasts is about 3 h. To examine if the reduced stability of CNBP in DM2 human fibroblasts can be corrected by the activator of AMPK, A769662 was added to DM2 fibroblasts, cells were treated with CHX, and CNBP levels were evaluated. We found that A769662 increases CNBP half-life in DM2 human fibroblasts. It is important that A76962 corrected CNBP to nearly normal levels. We suggest that the correction of CNBP is due to improved CNBP stability since normalization of CNBP levels occurs in the cells with blocked protein synthesis. These findings show that CNBP stability is reduced in DM2 human fibroblasts, and that the activation of AMPK by A769662 normalized CNBP stability.

Figure 8.

(A) Reduced CNBP stability in DM2 cells is corrected by the activator of AMPK. CNBP half-lives were measured in normal human fibroblasts and untreated and treated with A769662 DM2 fibroblasts. Protein extracts were analyzed by western blot in 0, 1, 2, 3 and 5 h after cycloheximide (CHX) addition (10 μg/ml). Since CNBP is reduced in DM2 cells, amounts of loaded proteins from DM2 cells were 3-fold higher than those from normal cells. B-actin is control. (B) Diagram shows CNBP levels in normal and untreated and A769662-treated DM2 cells (25 μM for 24 h) after CHX addition. (C) Normalization of CNBP-regulated pathways in DM2 fibroblasts treated with A769662. Western blot analysis of human fibroblasts from normal and untreated and treated with A769662 DM2 fibroblasts with abs to CNBP, PABP, RPS17 and β-actin (control). (D) A model suggests that in normal cells active pAMPK maintains CNBP stability likely through phosphorylation of CNBP at T173. As a result, CNBP levels are normal. In contrast, in DM2 cells, the mutant CCUG repeats via mechanism that remains to be identified reduce the pAMPK. As a result, unphosphorylated CNBP is unstable.

To examine if correction of CNBP has a positive effect on the molecular pathways regulating by CNBP, we examined CNBP targets, poly(A)-binding protein, PABP1, and ribosomal protein RPS17 [17], in normal and DM2 fibroblasts. As shown in Fig. 8C, a reduction of CNBP in DM2 correlates with a reduction of CNBP targets, PABP1 and RPS17. However, treating DM2 fibroblasts with the activator of AMPK A769662 (25 μM for 24 h) restored PABP1 and RPS17 levels. In summary, these data suggest that the expression of active pAMPK in normal cells maintains normal CNBP stability and normal levels of CNBP (Fig. 8D). However, in DM2 cells, a reduction of active pAMPK leads to reduced stability of CNBP, decreasing protein levels of CNBP.

Discussion

There are several critical results of our study. (1) We found that adult (6 months of age) homozygous Cnbp KO mice develop brain atrophy and have increased diffusivity in their brains (Figs 1 and 2). The grey matter atrophy is close to the significant value in adult heterozygous Cnbp KO mice. These findings show that CNBP function is critical for brain morphology. It has direct implications for DM2 since reduced levels of CNBP in human patients with DM2 might contribute to brain atrophy. This mechanism is in complete agreement with the primary role of CCUG repeats in DM2 since toxic CCUG RNA alone causes the reduction of CNBP, and the reduction of CNBP is an early event of CCUG RNA toxicity [29]. Important, that the degradation of the CCUG RNA leads to the recovery of the normal levels of CNBP [29]. In agreement with the morphological changes, Cnbp KO mice show behavioral defects. The adult (5 months of age) heterozygous Cnbp KO mice develop anxiety which is not observed at a young age. The anxiety is further worsening in older (8 months of age) heterozygous Cnbp KO mice. Since homozygous Cnbp KO mice have a more severe phenotype than heterozygous mice, abnormal behavior is already observed in young homozygous Cnbp KO mice (Fig. 3). It is important that neuromotor abnormalities (reduced ambulatory time and increased rest time) appear in heterozygous Cnbp KO mice at 8 months of age. We are planning to examine if older heterozygous Cnbp KO mice develop even stronger brain atrophy, anxiety and neuromotor defects. Unfortunately, the number of homozygous Cnbp KO mice in this strain is low due to increased mortality complicating the analysis of their behavioral features during ageing. It is important to note that, while we found anxiety and changes of neuromotor activities in Cnbp KO mice using OFT test, the cognition and memory in these mice need to be investigated further using additional tests.

Our data, connecting Cnbp with the development of brain atrophy, anxiety, and behavioral defects, show the significance of CNBP reduction in vivo. These data also suggest that CNBP correction might reduce not only muscle atrophy and weakness in DM2 but also brain atrophy. CNBP might be involved in developing other symptoms in DM2 because Cnbp KO mice show increased body fat [19]. Thus, CNBP reduction might play a role in metabolic balance and the development of type 2 diabetes in patients with DM2. DM2 is also associated with myotonia, cardiac defects and cataracts. We found only traces of myotonia in young Cnbp KO mice. While cardiomyocytes are smaller in young homozygous Cnbp KO mice (unpublished data), both heterozygous and surviving homozygous Cnbp KO mice live up to more than 16 months. Obvious cataracts were not found in our Cnbp KO model. However, further studies of aged (>12 months) Cnbp KO mice might identify additional phenotypic features reminiscent of the DM2 phenotype.

Other important results of our study include the identification of the reduction of active pAMPK in DM2 human cells (Fig. 7A and B) and the finding that the reduction of CNBP in DM2, caused by the mutant CCUG repeats [29] correlates with the reduction of active AMPK (Fig. 7A and B). We also found that CNBP protein stability is reduced in DM2 (Fig. 8A and B). The correction of AMPK using small molecule AMPK activator A769662 almost normalized CNBP stability (Figs. 8A and B). These findings suggest that the use of AMPK activators might be beneficial for DM2 patients. In addition to A769662, other AMPK activators should be tested in DM2 cells. There are many indirect AMPK activators which include metformin, quercetin, and resveratrol [45]. A769662 is a selective small molecule that directly and allosterically regulates AMPK via inhibition of dephosphorylation of T172 in AMPKα [45]. Although A769662 interferes with many pathways, it benefits mouse models for several diseases [45]. We are also planning to test a small molecule AMPK activator PXL770, which was applied in mouse models and in a human trial for other diseases [46–48]. To determine the effect of the small molecules AMPK activators on Cnbp stability in vivo and on DM2 phenotype, they must be evaluated in the mouse model for DM2, expressing CCUG repeats [49].

It is interesting that small molecule activators of AMPK show benefits in DM1 cell and mouse models. AICAR and resveratrol improve splicing and other defects in DM1 mice [43, 44], and metformin shows positive effects in DM1 models and DM1 clinical trial [50–52]. The AMPK activator quercetin corrects mis-splicing in DM1 and DM2 cells and myotonia in DM1 mice [53]. While mechanisms of the action of these AMPK activators in DM1/2 are not well understood, the findings described in this paper show that AMPK activators might correct CNBP levels and CNBP-dependent pathways in DM2.

There are additional questions that need to be addressed prior to examining the potential of AMPK activators for DM2 therapy. Since AMPK is a multifaceted kinase and might function differently depending on the cell environment, optimal doses of AMPK activators which normalize CNBP levels, avoiding CNBP and pAMPK over-expression should be determined first. Another important question is why pAMPK is reduced in DM2. We hypothesize that the mechanism of the reduction of pAMPK in DM2 might be related to the altered signaling pathways, caused by the mutant CCUG repeats. Since AMPK phosphorylation at T172 might be promoted by the LKB1 and the Ca2+/calmodulin-activated protein kinase (CaMKK2) [45], these kinases should be investigated in DM2 and CCUG-expressing cells to examine their possible effect on AMPK in DM2.

Materials and methods

Antibodies

The affinity-purified rabbit antibodies against a peptide from human CNBP were previously described [17]. Antibodies to AMPK and AMPK activator A76962 were from Cell Signaling. Antibodies to RPS17 and PABP were from Thermo Fisher Scientific.

Cnbp KO mice

Generation and analysis of Cnbp KO mice with disrupted Cnbp gene were described previously [19]. All mice were genotyped as described [19]. Matching (the same age and gender) WT, heterozygous and homozygous littermates were used as indicated in the text and in the Figure legends.

MRI, DTI, data acquisition and analysis

MRI and DTI analyses were performed by the Imaging Research Center at CCHMC. Mice were anesthetized with isoflurane via a nose cone, placed in a 38 mm inner diameter volume transmit/receive coil (Bruker, Billerica, MA), and positioned in the center of a 7 T Bruker Biospec (Bruker, Billerica, MA) small animal MRI. Their respiration rate was kept around 60 breaths per minute, and they were maintained between 36–38°C. After acquiring localizer images in 3 planes, a respiratory-gated, isotropic 3D T2-weighted spin echo data set was acquired for voxel-based morphometry (VBM) analysis using the following parameters: fast spin echo sequence, repetition time 1800 ms; echo time 80 ms; field of view 51.2 × 22 × 22 mm3; matrix 256 × 110 × 110. After shimming on the brain, an echo-planar-imaging-based diffusion tensor imaging sequence was acquired (parameters: repetition time 7500 ms; echo time 24 ms; 8 segments; 2 repetitions; 6 directions; b-value 850; 5 slices; 500 μm thick; axial orientation; field of view 25.6 × 25.6 mm2; matrix 114 × 114). Total scan time was about 1 h/mouse. VBM analysis was done using FSL after cropping the images in ImageJ to contain mainly the brain. The cropped images were brain extracted using FSL’s BET function to further minimize non-brain signals. The brain-extracted images were segmented using FSL’s FAST function into CSF, GM, and WM, the number of voxels in each segmented image were counted and converted to volumes based on the spatial resolution of the scan. These volumes were used for statistical analysis.

DTIStudio was used to generate parametric maps for various diffusion metrics (FA, ad, RD, and MD). The maps were imported into ImageJ, and regions-of-interest (ROIs) were drawn in the corpus callosum, external capsule, internal capsule, and fimbria. The diffusion metrics in those ROIs were extracted from the parametric maps. An automated analysis using the TBSS pipeline in FSL was also done. The DTI parametric maps were calculated in FSL, registered to a low-resolution anatomical image, registered to the 3D volumetric image acquired previously, and finally registered to a study specific template. The registered FA maps generated a skeleton map reflecting the WM tracts in the data. Voxel-wise statistics were calculated using FSL’s randomize algorithm on voxels within the skeleton for FA, ad, RD, and MD.

Open field test

Mice of different ages were examined in the Open Field Box using the Fusion software, v5.3 (Omnitech Electronics, Columbus, OH). Mice were placed in the box, and their behavior and neuromotor activity were analyzed for 5 min to examine various parameters, including total distance, total distance x-axis, total distance y-axis, horizontal activity count, ambulatory activity count, rest time, rest episode count, movement time, movement episode count, ambulatory time, ambulatory episodes count, stereotypy time, stereotypic episodes count, stereotypic activity count, stereotypic episode activity count, vertical episode count, vertical activity count, vertical movement time, center time legacy and center distance legacy. The test was performed in the same room protected from vibration and noise. The box was cleaned with 70% ethanol after each test.

Western blot analysis

Total protein extracts using RIPA buffer were prepared from the human skeletal muscle biopsies or from human cultured cells as indicated. Using human samples without direct interactions with patients is considered no human subjects research at CCHMC which did not require a human protocol. The quality of the proteins was examined by the Coomassie staining of the proteins separated by electrophoresis. Forty μg of proteins from each sample were loaded on the gel, separated by gel-electrophoresis, transferred onto the membrane, and probed with various antibodies. Antibodies to RBS17 and PABPC1 (ThermoFisher Scientific), AMPK (Cell Signaling Technologies) and β-actin (Santa-Cruz) were used as described in the manufacturers’ protocols.

Immuno-precipitation assay

Co-IP experiments were performed using total protein extracts from human cells as described previously [54].

Human cell culture

Primary human DM2 fibroblasts and myoblasts were obtained from the Muscle Tissue Culture Collection (MTCC) at the Friedrich-Baur-Institute (Department of Neurology, LMU Klinikum, Ludwig-Maximilians-University, Munich, Germany). All materials were obtained with the written informed consent of the donor. Ethical approval for this study was obtained from the ethical review committee at the Ludwig-Maximilians-University, Munich, Germany (LMU IRB Vote No. 45–14). The Protocol was updated in April 2022. Primary myoblasts were grown on 10 cm culture dishes at 60% density in the medium containing F10 medium (Gibco) with 15% fetal bovine serum (Hyclone), 1% Sodium Bicarbonate (Gibco), 5% defined supplemental calf serum (Hyclone), 1% L-glutamine (Gibco) and 1% penicillin/streptomycin (Gibco). Growth medium was changed every other day. Fibroblasts were grown in the medium containing DMEM (Gibco), 10% fetal bovine albumin (HyClone) and 1% penicillin/streptomycin (Gibco) at 80% density with growth medium change every two days.

Analysis of CNBP stability in DM2 fibroblasts

Normal and DM2 human fibroblasts were grown in 10 cm plates. To measure CNBP half-life, normal and DM2 human fibroblasts were treated with cycloheximide (CHX) (10 μg/ml) to block protein translation. Protein extracts were collected in 0, 1, 2, 3 and 5 h after addition of CHX and used for CNBP measuring by Western blot. CNBP levels were compared in the cells treated with CHX for different time periods and CNBP half-life was deduced. Where indicated, CNBP half-life was determined after treatments of DM2 fibroblasts with A769662.

Statistical analysis

The intensities of the protein bands in Western blot were determined by scanning densitometry relative to β-actin. Data were presented as a mean based on 3–6 repeats. Statistical analysis was performed using a two-tailed Student’s t-test. A P value <  0.05 was considered statistically significant.

 

Conflict of interest statement: The authors declare no conflict of interest.

Funding

LT has been supported by the Cincinnati Children’s Hospital Research Development Funds and National Institute of Health [5RO1NS115662 and 5RO1AR073379].