Growth arrest and DNA-damage-inducible, beta (GADD45b)-mediated DNA demethylation in major psychosis

Abstract

Aberrant neocortical DNA methylation has been suggested to be a pathophysiological contributor to psychotic disorders. Recently, a growth arrest and DNA-damage-inducible, beta (GADD45b) protein-coordinated DNA demethylation pathway, utilizing cytidine deaminases and thymidine glycosylases, has been identified in the brain. We measured expression of several members of this pathway in parietal cortical samples from the Stanley Foundation Neuropathology Consortium (SFNC) cohort. We find an increase in GADD45b mRNA and protein in patients with psychosis. In immunohistochemistry experiments using samples from the Harvard Brain Tissue Resource Center, we report an increased number of GADD45b-stained cells in prefrontal cortical layers II, III, and V in psychotic patients. Brain-derived neurotrophic factor IX (BDNF IXabcd) was selected as a readout gene to determine the effects of GADD45b expression and promoter binding. We find that there is less GADD45b binding to the BDNF IXabcd promoter in psychotic subjects. Further, there is reduced BDNF IXabcd mRNA expression, and an increase in 5-methylcytosine and 5-hydroxymethylcytosine at its promoter. On the basis of these results, we conclude that GADD45b may be increased in psychosis compensatory to its inability to access gene promoter regions.

Figure 1

Proposed mechanism of activity-dependent DNA demethylation. Following depolarization GADD45 protein binds to a methylated promoter region proximal to an acetylated histone (a). GADD45 recruits a deaminase (DA), which converts 5-methylcytosine (5MC) to thymine leading to a T:G mismatch (b). GADD45 recruits a DNA glycosylase (GLY), which removes thymine from the T:G mismatch. Thymine is later replaced with an unmethylated cytosine (c) (Guidotti et al, 2011).

Figure 2

Exon–intron structure and alternative transcripts of human GADD45 genes. Exons are shown as boxes and introns are shown as lines. Despite the three GADD45 genes being located on different chromosomes there are substantial similarities in terms of gene structure and protein composition. All three proteins are 17–18 kDa in mass and highly acidic. They share 56% protein sequence identity. In (a), are the three known splice variants of GADD45a. Isoform 1 contains four exons, while the other two splice variants are missing the second or third exon respectively. In (b), are the three splice variants of GADD45b. Isoform CRA_a and CRA_b differ based on their start sites. Another splice variant, referred to here as ‘GADD45b, Alt' includes the first intron (gray box), and translation begins at the second exon. This isoform is noted on the NCBI database (AAT38867.1) and in (Ying et al, 2005). In (c), is the gene structure of GADD45g.

Figure 3

GADD45b mRNA and protein expression are increased in psychosis. In post-mortem parietal cortical sections obtained from the Stanley Foundation Neuropathology Consortium (SFNC) mRNA was measured using qRT-PCR with β-actin and transferrin receptor (p90, CD71) (TFRC) as controls. Expression was calculated via geometric averaging of multiple internal control genes (Vandesompele et al, 2002). GADD45b (p=0.034) and BDNF IXabcd (p=0.016) mRNA significantly differ between nonpsychotic subjects (NPS) (n=13) and psychotic patients (PP) (n=19) (a). In SFNC samples, GADD45b protein (Santa Cruz, SC-133606) normalized to β-actin (Sigma-Aldrich #A5316) is higher in PP (n=26) compared with NPS (n=15) (p=0.04) (b). Representative immunoblot (c). Measurements were made for each subject in the linear range of 1.2, 0.6, and 0.3 μg of total protein with good agreement at all protein concentrations (molecular weight markers, MW). ^*p<0.05.

Figure 4

Immunohistochemistry and in situ hybridization results. In (a), is a representative immunohistochemistry with specific GADD45b antibodies using samples from the Harvard Brain Tissue Resource Center cohort, prefrontal cortex (BA9). Note more stained cells in cortical layers II, III, and V in psychotic patients (PP) compared with nonpsychotic subjects (NPS) (scale bar: 100 μm). In (b), we find significant increases in GADD45b positive neurons among PP (n=30) vs NPS (n=25) in layers II (p=0.007), III. (p=0.01), and V (p=0.02) (b). In (c), is a representative in situ hybridization experiment showing presence of BDNF IXabcd transcript in layer III and V cortical pyramidal neurons. Inset is a magnified cortical pyramidal neuron. (Scale bar: 100 μm; inset scale bar 10 μm). Error bars represent SEMs. ^*p<0.05; ^**p<0.01.

Figure 5

GADD45b is expressed in pyramidal and GABAergic cells, and not highly expressed in glial cells, in the post-mortem prefrontal cortex. In (a), is immunofluorescence showing colocalization of GADD45b (red) with a marker for GABAergic cells, GAD65/67 (Millipore, AB1511) (green). In (b), is immunofluorescence showing colocalization of GADD45b (red) with a marker for pyramidal cells, VGLUT2 (Abcam, ab79157) (green). In (c), is immunofluorescence showing colocalization of GADD45b (red) with a marker for glial cells, S100β (Sigma, S2532) (green).

Figure 6

GADD45b differentially binds to the BDNF IXabcd promoter in psychotic patients (PP) vs nonpsychotic subjects (NPS) in parietal cortical samples from the Stanley Foundation Neuropathology Consortium. In (a), is a representative gel from human brain chromatin immunoprecipitation (ChIP). Left gel lane reveals BDNF IXabcd PCR product in GADD45b (Aviva Systems Biology, ARP48346_P050) immunoprecipitated sample, center lane is negative control (NTC) (normal rabbit IgG; Santa Cruz, sc-2027) and right lane is input. In (b), is a western blot demonstrating that the antibody (Aviva Systems Biology, ARP48346_P050) used in ChIP experiments produces a band using protein from a wild-type (WT) mouse brain, but not from a GADD45b knockout (KO). In (c), is a western blot of human parietal tissue with 1.2, 0.6, and 0.3 μg of total protein indicating one major band for the GADD45b antibody. In (d), are the locations of human BDNF regions studied (Pruunsild et al, 2007). In ChIP experiments, we measured GADD45b binding and performed methylated DNA immunoprecipitation (MeDIP) experiments in the region surrounding the BDNF IXabcd transcription start site and within the exon BDNF IXd (blue lettering and blue lines below figure). Underlined is a transcribed portion of BDNF IXabcd. In red are the CpG sites in the ChIP areas studied. In italicized orange lettering is the location of the BDNF IXabcd transcript measured in RT-PCR experiments. The downward arrow represents the start of the shared 3′ exon used by other mRNA transcripts. Primer locations are in larger font size. In ChIP assays, GADD45b (Aviva, ARP48346_P050) is significantly more bound to a region of the BDNF IXabcd (p=0.013) promoter in NPS (n=15) compared with PP (n=26) (e). By contrast there is no difference in binding in a region of the BDNF IXd exon well downstream of the transcription start site (+1185 to +1305). Numbers represent pull down (ChIP) minus negative control as percent input. Error bars represent SEMs. ^*p<0.05.

Figure 7

Increased 5-methylcytosine (5MC) and 5-hydroxymethylcytosine (5HMC) at the BDNF IXabcd promoter in parietal cortical samples from psychotic patients (PP) from the Stanley Foundation Neuropathology Consortium. In (a), is a dot blot indicating specificity of 5MC and 5HMC antibodies. 75 ng of either unmethylated, 5MC, or 5HMC standards (Active Motif, 55 008) were spotted onto two membranes. The membrane on the left was incubated with 5MC antibody (Ab) (Diagenode, MAb-081-100), while the right membrane was incubated with 5HMC Ab (Active Motif, 39769). In (b), methylated DNA immunoprecipitation (MeDIP) assays reveal significantly more 5MC (Diagenode, MAb-081-100) at the BDNF IXabcd (p=0.044) promoter in PP (n=26) compared with nonpsychotic subjects (NPS) (n=15). In (c), MeDIP experiments using anti-5HMC antibody reveal increased 5HMC at the BDNF IXabcd (p=0.040) promoter in psychotic patients. Numbers represent pull down (ChIP) minus negative control as percent input. Error bars represent SEMs. ^*p<0.05.