A comprehensive spatiotemporal map of dystrophin isoform expression in the developing and adult human brain

Abstract

Mutations in the dystrophin gene (DMD) cause the severe muscle-wasting disease Duchenne muscular dystrophy (DMD). Additionally, there is a high incidence of intellectual disability and neurobehavioural comorbidities in individuals with DMD. Similar behavioural abnormalities are found in mdx dystrophic mouse models. Unlike muscle, several dystrophin isoforms are expressed in the human brain, but a detailed map of regional and cellular localisation of dystrophin isoforms is missing. This is crucial in understanding the neuropathology of DMD individuals, and for evaluating the translatability of pre-clinical findings in DMD mouse models receiving genetic therapy interventions. Here, we provide a comprehensive dystrophin expression profile in human brains from early development to adulthood. We reveal expression of dp427p2, dp427c, dp427m and dp40 isoforms in human embryonic brains, not previously reported. We also detected dp427p2 expression and developmental regulation in human brain across the lifespan. In addition we showed by in situ hybridisation that dp140 was greatly downregulated in adult brains. Importantly, our data also demonstrate expression of DMD transcripts in human motor neurons and co-expression of different dystrophin isoforms within single neurons in both developing and adult brains. Finally, we show localisation of DMD transcripts with GAD1+ GABAergic-associated transcripts in neurons including cerebellar Purkinje cells and interneurons, as well as in the majority of neocortical and hippocampal SLC17A7+ glutamatergic neurons, suggesting a role for dystrophin in signalling at the neuronal inhibitory and excitatory synapses.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s40478-025-01996-z.

Introduction

Dystrophin (DMD) is the largest gene in the human genome. It consists of 79 exons located on the X chromosome and produces multiple transcripts, whose expression is regulated in a cell-tissue-specific fashion by seven promoters [1, 2]. In addition, a set of splice variants is synthesised under the regulation of multiple splicing elements and polyA sites at the DMD locus [2] (Fig. 1a).

Fig. 1.

Human DMD transcripts and RNA labelling dynamics at the 5′ and 3′ ends of the DMD gene protein. Isoforms derived from the DMD gene can be classified into full-, intermediate- and short-length according to their mass in kDa and are all expressed in the human nervous system. DMD exons are shown with colours matching the encoded dystrophin functional domains. ASP (alternative splicing products) are generated by alternative splicing and distinct polyA-addition signals (a). The DMD 5′ probe detects dp427c, dp427m and dp427p and the DMD 3′ probe detects all the mature DMD transcripts except dp40 (b). Images show the variable size of the RNAscope red punctate dots detected using DMD 5′ and DMD 3′ probes (c). Scale bars = 50µm

Therefore, several dystrophin protein isoforms can be produced, and they are classified according to their molecular masses as full-length (dp427m, muscle; dp427c, cerebral; dp427p, of which there are two variants, dp427p1 and dp427p2, Purkinje cells), intermediate-length (dp260; dp140; dp116, mainly expressed in retina, brain and Schwann cells, respectively) and short-length (dp71 and dp40, both more ubiquitously expressed) [3–5].

Loss of the muscle-specific dp427m isoform is the cause of the rare muscle-wasting disorder, Duchenne muscular dystrophy (DMD, OMIM #310200), that affects 1 in about 3500 live male births, and results in progressive muscle degeneration leading to premature death because of heart and pulmonary failure [6, 7]. Hence the role of dystrophin, and particularly dp427m, has been extensively investigated and is best understood in muscle [8–10].

However, it is becoming more and more recognised that DMD patients (50–70%) display central nervous system (CNS) co-morbidities, and dystrophin isoforms have been shown to be expressed in the brain [11–18].

It has been suggested that intellectual disability and behavioural comorbidities in DMD patients, including autism spectrum disorder, attention-deficit/hyperactivity disorder and epilepsy, are significantly influenced by the DMD mutation site, hence on which isoform expression is lost in the brain [13, 15, 16, 19, 20]. In-vivo studies in mice demonstrated that full-length DMD isoforms (dp427) are critical players in GABAergic transmission by contributing to postsynaptic gamma-aminobutyric acid A receptor (GABAARs) clustering [21].

Their deficiency in the mdx23 mouse model, which lacks dp427, was associated with longitudinal changes in brain morphology, resulting in progressive cognitive impairments [22]. Similarly, a significant reduction of cortical GABA_A receptors in DMD patients using position emission tomography, as compared to age-matched normal subjects, was observed and consistent with prefrontal cortex dysfunction [23].

GABAergic and glutamatergic transmission as well as behavioural and cognitive phenotypes were shown to be impaired also in the mdx52 mouse model, deficient in both dp140 and dp427 [24, 25]. In accordance with the mouse phenotype, DMD patients lacking dp427 and dp140 were reported to present reduced cognitive function, high internalising and externalising symptoms, and increased conditioned startle response [13, 14] alongside a reduction in the grey matter and total brain volume detected by MRI [26, 27]. This reduction was greater in patients lacking both dp427 and dp140 than dp427 only [26]. Finally the lack of dp71 clearly contributes to the severity of the neural phenotype in human studies, with the mean IQ of 55 in dp71 deficient patients [28]. A behavioural phenotype also characterises the dp71-null mouse model, where a selective increase in excitatory transmission at glutamatergic synapses on Purkinje neurons was observed [29].

Despite a clear correlation between deficiency of different isoforms and resulting phenotypes in the human and mouse models [30], there remain significant knowledge gaps regarding the etiopathogenesis underlying CNS co-morbidities in DMD. Current information on expression of DMD isoforms in the human brain is limited to the mining of transcriptomic data from publicly available databases and key limitations include lack of temporal resolution, absence of isoform and exon-specific data [12, 31, 32]. In order to understand the role of DMD in behaviour and cognition, and whether its expression is required for normal development, it will be crucial to generate a comprehensive map of the regional and cellular distribution of dystrophin isoforms in developing and adult human brains. It will also be important to establish whether human neural cells express single or multiple dystrophin isoforms.

Hence our aim was to systematically assess the expression and localisation of the DMD isoforms during development and in the adult human brain. We firstly used quantitative reverse transcription polymerase chain reaction (RT-qPCR) across multiple brain regions. We then employed 5′/3′ and isoform-specific DMD in situ hybridisation (ISH) probes to map DMD transcripts at the regional, cellular and sub-cellular levels. Using multiplex ISH probes, we studied the co-expression of DMD isoforms and their co-localisation with excitatory and inhibitory neurons. Specific antibodies were used to map dystrophin protein expression. Our studies provide the first comprehensive spatial and temporal expression patterns of the dystrophin isoforms of developing and adult, pathology-free human brain.

Materials and methods

Human tissues

Human tissue procedures were conducted in accordance with the regulations with informed consent in accordance with the UK Human Tissue Act 2006 for study participation under ethical approval (NRES Committee London—Fulham, UK; REC: 18/LO/0822). The human embryonic and foetal material was provided by the Joint MRC/Wellcome Trust (Grants MR/006237/1, MR/X008304/1 and 226202/Z/22/Z) Human Developmental Biology Resource (https://www.hdbr.org).

Tissue bank provided brain tissue from several donors at 17 (41 days) and 23 (56 days) Carnegie stages (CS) and 10, 14, 15, 19, and 20 post-conception weeks (PCW). Details of the subjects and samples cohorts are indicated in Supplementary Table 1.

Adult human brains: FFPE and frozen post-mortem brain specimens from four ‘non-neuromuscular’ pathology-free subjects (Supplementary Table 1) were selected and procured from the Edinburgh Brain Bank (Ethic Ref 20/PR/0582). All subjects showed no significant abnormalities after histopathological examination of the brain. Tissue samples were obtained from the Edinburgh Brain Bank (EBB) and BRAIN UK, which is supported by Brain Tumour Research and has been established with the support of the British Neuropathological Society and the Medical Research Council [33].

Three distinct samples cohorts were available: (1) frozen and FFPE specimens from embryonic and foetal subjects; (2) five ‘matching’ brain areas from the same adult subjects available both frozen and FFPE and; (3) twelve remaining adult brain areas available FFPE only.

Transcript analysis

RNA isolation

According to the manufacturer's instructions, RNeasy Mini Kit (Qiagen, Cat. No. 74104) and RNeasy FFPE Kit (Qiagen, Cat. No. 73504) were used to isolate RNA from frozen and fixed sections respectively. To dissociate the sections, RLT buffer was added, and 5 mL syringes with a 25 gauge needle were used for disruption.

Total RNA concentration was measured using the NanoDrop 2000 spectrophotometer (Thermo Scientific, Cat. No. ND-2000). cDNA was generated by reverse transcription from 200 ng (frozen) and 100 ng (FFPE) of RNA using the High-Capacity RNA-to-cDNA™ Kit (Thermofisher, Cat. No. 4387406) according to the manufacturer’s instructions. Next, cDNA was pre-amplified with the TaqMan™ PreAmp Master Mix (Thermofisher, Cat. No. 4391128) according to the manufacturer’s instructions.

Gene expression analysis by RT-qPCR

Subsequently, DMD expression was detected by RT-qPCR with the TaqMan™ Fast Advanced Master Mix (Thermofisher, Cat. No. 4444557) and a set of TaqMan isoform-specific probes (Thermofisher Cat.No 4332079). A total volume of 5µL of 1:5 diluted preamplified cDNAwas used for each RT-qPCR reaction.

Relative expression was calculated by 2^−ΔCt formula and normalised to the brain-specific RPL13 housekeeping gene. RPL13 was selected as a housekeeping gene following a comparison with other CYC1, GAPDH, and UBE2D2 genes (data not showed), demonstrating its higher consistency of expression in human brain samples, in line with previous studies. With the aim of comparing samples of different nature and therefore taking into account the variability caused by tissue fragmentation, 2^−ΔCt data from different sample groups were then converted to a common scale to fit within a 0–100 range, and visualised on truncated violin plots generated using the Prism 9 software (GraphPad Software LLC, Cat. No. 140000).

To mitigate the RNA fragmentation-related effects resulting from tissue fixation [34], a combined polyA/random hexamers reverse transcription and a cDNA preamplification steps were introduced for maximising the RT-qPCR sensitivity in FFPE samples [35, 36]. Despite our efforts to design probes and primers amplifying short amplicons (between 60 and 77bp) aimed at making highly sensitive RT-qPCR assays, a longer amplicon length was required for dp427m (88bp), dp427p1 (81bp) and dp40 (116bp) amplification. Consequently, when FFPE and frozen RT-qPCR results from the same area (and same subject) were compared, we detected a reduction of dp427m and dp427p1 expression in the FFPEs areas. This was likely due to the larger amplicon sizes compared to the other isoforms and in particular the full-length dp427c (66bp). The impact of RNA fragmentation was more evident with the 116bp amplicon of dp40 which was detected in frozen samples only. This is in line with previous reports showing that smaller amplicons produce more comparable cycle threshold (Ct) between FFPEs and frozen samples because less prone to fragmentation [35–37].

In situ hybridisation

RNAscope assays

The RNAscope in situ hybridisation (ISH) assay with its high specificity and sensitivity allows for the resolution of single mRNA transcripts which typically appear as discrete punctate/particulate foci within the cell cytoplasm and/or nuclear domains (Fig. 1c).

We assessed the prevalence and spatial distribution of DMD transcripts within cellular subtypes across different regions in the developing and adult human brain. Specifically, we employed a DMD 5′ probe targeting region (E2-E10) recognising all nascent and mature full-length transcripts (dp427m/dp427c/dp427p) and a DMD 3′ probe targeting region (E63-E75) recognising all mature dystrophin transcripts (dp427, dp260, dp140, dp116, dp71/dp40) except dp40 (Fig. 1b). Appropriate positive and negative controls were employed in parallel to confirm RNA preservation and non-specific labeling (Fig. S7). Brain FFPE and frozen samples were pre-treated according to the existing manufacturer’s instructions and incubated with probes designed by ACD Bio (Advanced Cell Diagnostics Inc., Newark, CA).

Singleplex RNAscope assays (Advanced Cell Diagnostics Inc, Cat. No. 322360) relying on RNAscope DMD 5′ and DMD 3′ probes were performed according to the manufacturer’s protocol. Probes are listed in Supplementary Table 2.

Before being processed for in situ assays, selected control sections were incubated against commercially available RNAscope UBC (ubiquitin C) and PPIB (peptidylprolyl isomerase B) probes to confirm RNA preservation. In addition, incubations with PBS/probe diluent were performed to test the extent of the background signal (negative control). Tissue staining was performed according to the manufacturer's instructions.

Basescope assays

Singleplex Basescope assays (Advanced Cell Diagnostics Inc., Cat. No. 323900​) combined with a set of singleplex probes were employed to deconvolute the RNAscope signal and study the spatial expression pattern of dp427m, dp427c, dp427p2, dp140, dp71, and dp40 in selected brain areas. Tests were performed according to the manufacturer’s protocol (Advanced Cell Diagnostics Inc., Cat. No. 323900-USM). Probes are listed in Supplementary Table 2. A few steps were modified for embryonic/foetal tissue to avoid tissue digestion. Specifically, the target retrieval was performed for 8 min in boiling water; tissue was treated with Protease III for 30 min at 40° in high-humidity conditions. Before being processed for in situ assays, selected control sections were incubated against a commercially available Basescope UBC (ubiquitin C) probe to confirm RNA preservation and PBS/probe diluent to test the extent of the background signal (negative control). Tissue staining was performed according to the manufacturer's instructions.

Tissue imaging and analysis

Whole section images were acquired within 24h at 40× magnification on a NanoZoomer S60 slide scanner (Hamamtsu, Cat. No. C13210-01).

RNAscope: In situ hybridisation signal for 5′ and 3′ DMD transcripts was assessed by a neuropathologist in consecutive sections by carefully aligning the digital images and mapping the region of interest (ROI) for each anatomical area, primarily within morphologically identifiable neuronal elements. Scoring was based on a subjective assessment of the proportion of cells expressing transcripts within the ROI. Neurons were marked as positive irrespective of the number of transcripts (dots) expressed, and the size or staining intensity of the dots. The data was plotted on a heatmap showing the following colour code: white: unprocessed; light green = sporadic (0–5% positive cells); green = low expression (5–20% positive cells); intense green = moderate expression (20–50% positive cells); orange = frequent (50–70% positive cells); red = strong expression (> 70% positive cells); grey: undetermined.

Singleplex basescope signal semi-quantification: anatomical regions for the processed brain area were manually annotated by a neuropathologist. For each annotated region, a set of images (from three randomly selected fields of view) was exported, and the in situ signal was manually scored by three independent operators as indicated in the manufacturer guidelines. We then averaged the results obtained by the three operators and plotted the data into heatmaps. Specifically, we quantified the number of cells displaying at least one dot (positive) and calculated the P+ parameter, representing the percentage of positive cells out of the total cells. Heatmap colour code: white = unprocessed; light green = low expression; green = medium expression; orange/red = high expression; grey = undetermined.

Duplex Basescope was performed in a qualitative fashion to detect the co-expression of multiple DMD cellular transcripts by probes targeting the following isoforms: dp427m (channel 1-green), dp427c (channel 2-red), dp140 (channel 1-green and channel 2-red), dp71 (channel 2-red), UBC (channel 1-green) and PPIB (channel 2-red).

Protein analysis

Protein extraction

Protein samples were prepared on ice for extraction after adding 10 μL of extraction buffer consisting of 50 μL of protease and phosphate inhibitor mini tab (Pierce, Cat. No. A32959) to 950uL of T-PER tissue protein extraction reagent (Thermo Scientific, Cat. No. 78510) for each mg of tissue. Tissue was then homogenised using Fisher brand 850 Homogeniser (Fisher scientific, Cat. No.15–340-169) for 20 s at 10,000 and placed in an ultrasonic ice bath (Branson Ultrasonics, Cat. No. 2510-DTH) for 5 min. The homogenates were then centrifuged for 15 min at 14,000g at 4 °C, the supernatant was then centrifuged again at 14,000g at 4 °C for 5 min and kept at –70 °C. Protein extracted from the samples were quantified using RC DC Protein Assay Reagents Package (Bio-Rad Laboratories Ltd, Cat. No. 5000120).

Capillary Western immunoassay (Wes)

Simple western blot (WES, Bio-techne) was used to quantify Dystrophin levels in protein lysates from brain samples. Ab154168 (abcam) was used as the antibody with a concentration of 1:1500, which recognises all the isoform except dp40. A total of 2 μg of protein for each sample was prepared for WES analysis following protocols for the 66–440 kDa separation module (Bio-techne, Cat. No. SM-W005) and for the total protein detection module (Bio-techne, Cat. No. DM-TP01). Each dystrophin protein signal was normalised to the total protein detection signal.

Protein localisation by immunohistochemistry (IHC)

Tissue was obtained from HDBR or Brain tissue bank as formalin-fixed paraffin-embedded (FFPE) blocks or 5 µm thick paraffin sections on microscope slides. After cutting, slides were placed in an oven or hotplate for 10 min. Then, they were loaded onto the Ventana Discovery Ultra machine. First, three cycles of deparaffinisation were performed for 4 min each using EZ Prep at 65 °C. Heat Induced Epitope Retrieval (HIER) is performed using Citrate-based buffer pH 6.0 at 95 °C for three cycles (8min, 12min, 20min). Slides were washed three times using a reaction buffer (Tris-based containing ~ 2% Brij 35 and ~ 4% Acetic Acid). An amplification kit was used for primary anti-dystrophin antibodies. Two primary anti-mouse antibodies were used, one toward the C-terminus (1:150), detecting all dystrophin isoforms except dp40 (Leica Biosystems, NCL-DYS2), and the second against the rod domain (1:150) detecting longer dystrophin isoforms (NCL-DYSA, Leica Biosystems). The primary antibodies incubated for 60 min involved heating at RT. A secondary antibody was performed using OmniMap anti-Mouse HRP (Roche diagnostics, Cat. No. 760-4310), and then a Chromomere DAB kit (Biocare Medical Cat. No. ACT500) was used for nuclear staining. Finally, sections were mounted after three times washing for 2 min each and left to dry overnight.

Statistical analysis

Significant changes in transcript levels in brain tissue for dystrophin expression were assessed by one-way analysis of variance (ANOVA) to compare the differences between multiple groups (Supplementary Table 3) when 3 or more biological replicates were available. Statistical analysis was performed using the Prism 9 software (GraphPad Software LLC, Cat. No. 140000).

The p-value < 0.05 was reported as significant **p ≤ 0.01, ***p ≤ 0.001 and ****p ≤ 0.0001.

Results

Dystrophin isoform expression in human brains from embryo to adult

To gain insights into the regulation of dystrophin isoforms in developing and adult human brains, we first assessed their expression by RT-qPCR. Specifically, we quantified regional expression of dp427c, dp427m, dp427p1, dp427p2, dp260-1, dp260-2, dp140, dp116, dp71/40 (sharing the first unique exon) and dp40 isoforms in embryonic, foetal, and adult frozen and/or formalin-fixed and paraffin embedded (FFPE) post-mortem pathology-free brain samples. RT-qPCR data showing isoform brain expression from embryo to adult are summarised in Fig. 2 and S1a.

Fig. 2.

Expression patterns of the main DMD isoforms in developing and adult brain areas assessed by RT-qPCR. Truncated violin plots showing the 2^−ΔCt values (converted to a 0–100 range) of the DMD isoforms in prenatal and adult human brains. Gene expression was normalised to the brain specific RPL13 housekeeping gene. Yellow lines indicate the median while each black dot represents a processed sample. Pcw postconceptional weeks, CS Carnegie stage, FFPEs Formalin Fixed Paraffin Embedded tissue samples

Among the full-length isoforms, the Purkinje dp427p1 transcripts were hardly detectable or absent in developing and adult brains. Dp427p2 showed high expression in the embryonic and foetal brain, including cerebellar and extra-cerebellar regions, but was significantly downregulated in the adult brain. The cortical dp427c isoform was highly expressed in the developing brain from the earliest embryonic stage studied (CS17) and its expression was maintained in most areas of adult brain, with the highest levels observed in hippocampus, amygdala, and several neocortical areas including the prefrontal cortex. Moderate to high levels of expression were also observed in the cerebellum. Dp427m transcripts were highly expressed throughout development and showed variable expression levels (low/moderate ) in the adult brain.

The dp260-1 splicing isoform was undetectable at all stages, whereas a low expression of dp260-2 was detected in some developing and adult brain regions (Fig. S1a).

Among the intermediate isoforms, dp140 was highly expressed in the embryonic and early foetal brains without regional specification, but not in the adult brain, where it became largely restricted to the cerebellum with high expression levels; lower levels of expression were detected in the dorsolateral prefrontal cortex, striate cortex and inferior frontal gyrus. Low dp116 expression was detected in most of the frozen regions analysed.

Both short isoforms dp71/dp40, showed high expression levels in the embryonic and early foetal brains, and continued to remain high in the adult brain, with little regional specification.

When the 5′UTR-dp71 and the 3′UTR-dp40 were targeted to specifically assess dp40 transcripts, low dp40 expression was detected in both embryonic/foetal and adult frozen specimens, but not in adult fixed tissues (Fig. S1b), where higher RNA fragmentation occurs as explained in Materials and Methods.

In summary, all the DMD isoforms apart from dp427p1 and dp260-1 are consistently expressed in the human brain, with changes in dp427p2 and dp140 expression suggesting developmental regulation of these isoforms.

RNAscope in situ hybridisation assay to 5′ and 3′ regions of DMD mRNA allows spatial mapping of dystrophin transcripts in the developing and adult human brain

Labeling with the DMD 5′ probe, targeting the full-length isoforms, showed a consistent pattern across all neuronal populations in the form of a discrete, large intranuclear particulate signal, suggestive of transcriptional ‘hot spots’ containing multiple nascent transcripts (Fig. 1c). A few neurons also displayed smaller, punctate foci in the nucleus and/or cytoplasm. Labeling with the DMD 3′ probe, targeting all the isoforms apart from dp40, showed 1–3 large discrete intranuclear foci, and multiple smaller, mostly cytoplasmic foci, suggestive of mature transcripts (Fig. 1c).

Figure 3a shows a heatmap summarising the semi-quantitative values from the RNAscope signal scored by the pathologist in the adult brain, as detailed in the Materials and Methods section.

Fig. 3.

In situ hybridisation in the human brain. Heatmap representing the DMD 5′ and DMD 3′ RNAscope signal scores in each adult ROIs (a). Heatmap representing the singleplex Basescope P+ (% Positive cells/total cells) across different adult brain areas (b). Colour code: white: unprocessed; light green: low expression; green: moderate expression; orange/red: high expression; grey: undetermined; S1-4: subjects processed for each developmental stage

To identify changes in DMD localisation with development, section from embryos and foetuses at developmental stages between CS17, CS23 and PCW19 were hybridised with either the DMD 3′ or the DMD 5′ probe (Figs. 4, 5, 6 and Fig. S2).

Fig. 4.

DMD transcript detection by RNAscope in embryonic human brain at Carnegie stage 17 (CS17). Top left image shows a low magnification of a CS17 embryo sagittal section. Individual plates show representative images of brain sections from the regions indicated in the CS17 embryo hybridised either with the DMD 5′ or DMD 3′ probe. Note extensive DMD 3′ signal in all brain regions, whereas the DMD5′ signal is either not detectable or detected only in a few cells, with the exception of the trigeminal ganglion where several positive cells are observed. As positive control, representative images of FFPE muscle sections hybridised either with the DMD 5′ or the DMD 3′ probe are shown; note extensive signal with both probes (red dots). Inserts show representative puncta at a higher magnification. Scale bars are: embryo = 1 mm; images on left column = 10 µm; telencephalon = 25 µm; all other images = 50 µm

Fig. 5.

DMD transcript detection by RNAscope in embryonic human brain at Carnegie stage 23 (CS23). The left image shows a low magnification of a CS23 embryonic brain coronal section highlighting main structures at this stage of development. Right panels show representative images of telencephalon, hippocampus and cerebellum hybridised either with the DMD 5′ or DMD 3′ probe. Note extensive DMD 3′ signal in all brain regions with representative puncta shown at higher magnification in the inserts; the DMD 5′ signal is very sparse (telencephalon and cerebellum) or absent (hippocampus). CP choroid plexus, H hippocampus, LV lateral ventricle, 4V 4th ventricle. Scale bars are: CS23 embryo = 2 mm telencephalon and cerebellum = 100 µm; hippocampus = 250 µm

Fig. 6.

DMD transcript detection by RNAscope in a 19 weeks post-conception (PCW19) foetal human brain. Left images show a low magnification of PCW19 hippocampus and cerebellum with localisation of regions shown on the left indicated. Individual plates on the right show representative sections of hippocampus and cerebellum hybridised either with the DMD 5′ or DMD 3′ probe. Signal from both probes is detected in the dentate granule cells, CA1 and subiculum of the hippocampus, as well as in all cerebellar layers: external granular layer (EGL), molecular layer (MOL), Purkinje cell layer (PCL) and granule cell layer (GCL). Some representative puncta are shown at higher magnification in the inserts. Scale bars are: low magnification hippocampus and cerebellar cortex = 1 mm and = 500 µm, respectively; all other images = 40 µm

In the brain region studied in CS17 embryos, with the exception of the trigeminal ganglion, more extensive and ubiquitous labeling was observed with the DMD 3′ probe than with the DMD 5′ probe, suggesting higher expression of shorter-length DMD isoforms than long ones (Fig. 4). The DMD 3′ signal was particularly abundant across the neuroepithelium of telencephalon, hippocampus, cerebellum, pons and medulla, and was the only signal detected in the optic vesicle (Fig. 4 and Fig. S2a). DMD 5′ signal was localised to neuroepithelial cells in pons and medulla, suggesting expression of full-length DMD isoforms in these areas. The hybridisation pattern observed suggests that at CS17 DMD is predominantly expressed in neural progenitor cells and early migratory neurons. The CS17 skeletal muscles expressed both DMD 5′ and DMD 3′ transcripts, providing a positive control for both probes (Fig. 4).

At the late embryonic stage CS23, DMD 3′ labeled transcripts were widely distributed in the telencephalon, hippocampus and cerebellum as well as in amygdala, cerebral cortex, basal ganglia, medulla and retina (Figs. 5, S2b and S4a). DMD 5′ labeled full-length transcripts were more widely expressed at CS23 than at CS17 in all regions studied apart from the hippocampus (Figs. 5 and S2b), but the DMD 3′signal remained the more robust one. The CS23 muscle continued to show positive signal with both probes (Fig. S2b).

At foetal stages, PCW14 and PCW19, DMD 3′ labeled transcripts were widely distributed in the hippocampus and cerebellum as well as in other brain regions (Figs. 6, S2c-d and S4a). At PCW14, DMD 5′ labeled transcripts were detected in the cerebellum, but not in cortical areas such as the cingulate gyrus and paracentral lobe (Figs. S2c and S4a). By PCW19, the DMD 5′ signal had become more widely expressed and, unlike at earlier stages, was detected also in the hippocampus, especially in the granule cells of the dentate gyrus and in the CA1 area, and in the subiculum, as well as in the cerebellum, where it was particularly striking in Purkinje cells and deep nuclei, with large puncta observed (Figs. 6 and S2d).

In the adult cerebellar cortex, DMD 5′ and DMD 3′ labeled transcripts were noted in the molecular and Purkinje cell layers (Figs. 7b, S3a and S4b), whereas only DMD 3′ labeled transcripts were seen in the granule cell layer (Figs. 7d, S3a and S4b), suggesting abundant expression of shorter mature transcripts in these cells (dp140/dp71). Both DMD 5′ and DMD 3′ labeled transcript were seen in the dentate nucleus (Fig. 7c, f). In the adult brain DMD transcripts were abundant in the hippocampus proper CA1-CA3, CA4/hilar region and the dentate fascia (Figs. 7g–m and S3a). While both DMD 5′ and 3′ transcripts were localised within the amygdala, subregional (nucleus-wise) specification could not be ascertained due to suboptimal orientation of the blocks (Fig. S3a). High expression was detected in the prefrontal cortices and inferior frontal gyrus (Fig. 8).

Fig. 7.

Expression of DMD 5′ and 3′ labeled transcripts in the adult cerebellum (a–f) and hippocampus (g-m). Low power overview of the cerebellum (a). In the molecular layer, scattered small cells, likely interneurons (e, red circles) express 3′ labeled transcripts. The large Purkinje cells (red circles) abundantly express 5′ (b), and 3′ (d) labeled transcripts. The granule cell layer preferentially expresses 3′ labeled transcripts (b, d). Neurons in the dentate nucleus (red circles) also express both 5′ (c) and 3′ (f) labeled transcripts. Low power overview of the hippocampus (g). There is frequent expression DMD transcripts in the granule cells of the dentate gyrus [5′ (h), 3′ (k)], CA3/hilus [i (5′), l (3′) and CA1 [j (5′), m (3′)]. WM white matter, ML molecular layer, PC Purkinje cell layer, GL granule cell layer, DN dentate nucleus, DGC dentate granule cells. Scale bars are: g = 1 mm; a = 2.5 mm; c, h and i = 100 µm; l and m = 50 µm; all other images = 25 µm

Fig. 8.

Expression of DMD 5′ and 3′ labeled transcripts in the adult dorsolateral prefrontal cortex (a–g), primary motor cortex (h, i), brain stem (j–l), pons (m-o) and medulla (p-v). Low power overview of the hexalaminar dorsolateral prefrontal cortex shows expression of 5′ and 3′ labeled transcripts in all layers (red circles). Higher magnification depicts examples of 5′ transcripts expressed in pyramidal cells of layers III (b), V (c), and polymorphic neurons in layer VI (d), and 3′ labeled transcripts expressed in pyramidal cells of layers III (f) and V (g). Within layer V of the precentral gyrus (primary motor cortex) the large pyramidal Betz cells (upper motor neurons) express both 5′ (h) and 3′ (i) labeled transcripts. A small cell, likely interneuron, expresses 3′ labeled transcript (i, black arrow). Overview of the mid brain substantia nigra, pars compacta (j). Higher magnification view within the region of interest (j, red rectangle) shows expression of both 5′ (k) and 3′ (l) labeled transcripts in the large dopaminergic pigmented neurons. Low power overview of the pons (m). Higher magnification shows neurons in the pontine nuclei expressing both 5′ (n) and 3′ (o) labeled transcripts. Low power view of the medulla sectioned at the level of the inferior olivary nucleus. Neurons within the hypoglossal nerve nucleus preferentially express 3′ labeled transcripts (s) over 5′ labeled transcripts (q). Neurons within the inferior olivary nucleus express both 5′ (r) and 3′ (t) labeled transcripts. Scale bars are: m and p = 2.5 mm; a and e = 250 µm; u = 100 µm; all other images = 50 µm

Moreover, DMD 5′ and DMD 3′ labeled transcripts were widely expressed across several neocortical regions: primary motor cortex, dorsolateral and ventromedial prefrontal cortex, inferior frontal gyrus, superior temporal gyrus, angular gyrus, calcarine/striate cortex and anterior cingulate gyrus (Fig. 8a–i and S3a). The expression pattern was similar across these regions with pan-cortical localisation of transcripts. Within the cortices, most transcripts were neuronal, including majority of pyramidal cells (Figs. 8b, c and S3a) and a small proportion of the 5′ transcripts localised to oligodendroglia-like cells in the cortex and white matter (Fig. S3a). Both DMD 5′ and DMD 3′ labeled transcripts were localised to the upper motor neurons, the giant pyramidal cells of Betz (Fig. 8h, i). DMD 3′ and DMD 5′ labeled transcripts were observed in several other brain areas such as Pons, Medulla and hypoglossal nerves (Fig. 8m–v).

Basescope DMD isoform-specific, singleplex and multiplex in situ hybridisation assays reveal isoform diversity and coexpression within neuronal, glial and vascular domains across key brain regions

Singleplex Basescope: As RNAscope did not provide information on expression and localisation of specific DMD isoforms, we used singleplex Basescope assays to localise dystrophin isoforms, dp427c, dp427m, dp140, dp71, and dp40. We could not study dp427p2 due to lack of specificity of the labelling probe. Expression of these isoforms was semi-quantified in adult brains and shown as a heatmap (Fig. 3b), and examples of cellular localisation are shown in Figs. 9 and S3b. In the adult brain, dp427c transcript was widely expressed across all neocortical regions (Fig. 9w), and also in the hippocampus (Fig. 9r-s), amygdala (Fig. S3b) and cerebellum (Figs. 9c, g, j and Sb3). Dp427m was consistently expressed within the leptomeningeal (Fig. 9l) and cortical blood vessels (Fig. S3b) containing a muscular coat (most likely in the smooth muscle cells).

Fig. 9.

Expression of DMD isoform-specific transcripts in the adult cerebellum (a–p), hippocampus (q–s, u, v), and primary visual (striate) cortex (t–x). Low power overview of the cerebellum (a). Three layered cerebellar cortex (b). Purkinje cells expressing dp427c (c) and dp71 (d) transcripts. Granule cells expressing dp40 (e) and dp140 (f) transcripts. Scattered small cells, likely interneurons within the molecular layer expressing dp427c (g) and dp71 (h) transcripts. Neurons within the dentate nucleus (i) expressing dp427c (j) and dp71 (k) and in the white matter (m) expressing dp427m (n) and dp71 (o). A leptomeningeal blood vessel expressing mural dp427m transcripts (l). A parenchymal capillary-sized small blood vessel shows dp71 expression in its wall (p). Low power overview of the hippocampus (q). Dentate granule cells express dp427c (r) and dp71 (u) transcripts. Pyramidal cells in CA3 express dp427c (s) and dp71 (v) transcripts. Overview of the striate cortex (t). Cortical pyramidal cells within layer III expressing dp427c (w and dp140 (x) transcripts. ML molecular layer, PCL Purkinje cell layer, PC Purkinje cell, GL granule cell layer, GC granule cells, DN dentate nucleus, DG dentate gyrus. Scale bars are: a = 2.5 mm; q = 1 mm; t = 0.5 mm; b, i, m = 100 µm; all other images = 10 µm

Dp140 transcripts were expressed in the dorsolateral and ventromedial prefrontal cortices and the striate cortex (Fig. S3b). Dp71 was detected in the ventromedial orbitofrontal cortex with transcripts identified in pyramidal cells.

In the adult hippocampus, the granule cells in the dentate gyrus showed moderate to frequent expression of dp427c (Figs. 9r and S3b), dp140 and dp71 (Figs. 9u and S3b) and low levels of dp427m and dp40 transcripts. The hilum/CA4, CA1-CA3 and subiculum showed moderate expression of dp427c and dp71 (Figs. 9s–v and S3b), and low level of dp427m transcripts.

The amygdala showed low levels of dp427c and dp140 expression (Fig. S3b). Very few neurons expressed dp427m (Fig. S3b). Sparse dp427c transcripts were noted in the basal ganglia. Few oligodendroglial cells expressed dp427m in the white matter (Fig. S3b). The thalamus showed moderate expression of dp427c (Fig. S3b). Dp427m signal was noted mostly in oligodendroglia, and rarely in neurons.

In the midbrain, dystrophin transcripts were mainly present in the substantia nigra pigmented neurons, specifically, dp427c, with occasional neurons expressing dp427m. The dorsal midbrain and the midbrain tegmentum showed occasional dp427c and dp427m in the Edinger Westphal nuclei (Fig. S3b), oculomotor nerve nuclei and the mid-line raphe neurons.

The pontine tegmentum, including the pigmented neurons of the locus coeruli, the periventricular grey, the midline raphe nuclei and the pontine reticular formation showed modest expression of dp427c and dp140 transcripts. Similarly in the basis pontis, there was moderate expression of dp427c, dp427m, dp140 and dp71 transcripts in the pontine nuclei. Dp427m expression was also noted in the oligodendroglial cells within the white matter tracts (corticopontine, corticospinal and pontocerebellar fibres) (Fig. S3b).

In the medulla, sparse dp427c transcripts were localised to the medullary tegmentum (hypoglossal nuclei and reticular formation) and the inferior olivary nucleus. Dp427m was expressed in oligodendroglial cells in the pyramidal tracts and the emerging rootlets of the vagus nerve.

The adult cerebellum showed few dp40 and dp71 (Fig. 9h) transcripts expressed within the molecular layer (likely in interneurons), Purkinje cells, dentate nucleus (Fig. S3b) and granule cells. Dp71 transcripts were also seen within the parenchymal capillary walls (Fig. 9p) and the pia-glial membrane (glia limitans), but not within the leptomeningeal blood vessels. In the cerebellum, dp140 transcripts were detected in a moderate number of granule cell layer (Fig. 9f). A small number of molecular layer neurons, Purkinje cells and dentate neurons also expressed dp140 transcripts. Moderate numbers of Purkinje cells and neurons in the dentate nucleus and few interneurons in the molecular layer expressed dp427c (Fig. 9c, j). In contrast, there was no dp427c signal in the granular cell layer. Dp427m expression was restricted to sparse basket cells within the Purkinje cell layer.

Duplex Basescope: In addition, duplex Basescope with dp427m/dp427c, dp140/dp427c, dp427m/dp140 and dp427m/dp71 probes was used to further assess isoform localisation in the adult brain, and establish whether more than one dystrophin isoform may be expressed in the same cell (Figs. 10 and S5).

Fig. 10.

Detection of co-expressed DMD isoforms in developing and adult human brain areas by duplex Basescope. Representative images of developing brain samples hybridised with the dp140 (green) and dp427c (red) probes. Note some cells express both isoforms (a). Representative images of adult brain samples hybridised with four different probe combinations: dp427m(green)/dp427c(red); dp427m(green)/dp140(red); dp427m (green)/dp71(red); dp140 (green)/dp427c(red) (b). Red and green arrows indicate RNA transcripts (dots) co-expressed in the same cell. The top left inset shows the co-expression events at a higher magnification. bv blood vessels, Pcw postconceptional weeks, CS Carnegie stage; Hippocampus CA Cornu Ammonis. Scale bars = 10 µm

In the adult brain, neuronal dp140/dp427c co-expression was seen in the, hippocampus, pons, striate cortex and dorsolateral prefrontal cortex while rare neuronal events were noted in pyramidal cells across all hippocampal subsectors and in the amygdala (Fig. 10b). Co-expression of the two full length dp427c/dp427m isoforms was seen in one of the midbrain’s raphe neurons (Fig. 10b). Unexpectedly, moderate dp427m signal was noted in oligodendrocytes, both within the cerebellar white matter, the cortex and the underlying white matter across all neocortical areas (Figs. 10b) and S5a).

Due to the limited number of prenatal sections available, we elected to study CS17, CS23 and 14 pcw sections for duplex BaseScope with the dp140/dp427c isoforms as they were the most relevant neuronal transcripts previously detected by RT-qPCR. In these sections, we detected a robust dp140 expression (Fig. 10a), consistent with the RT-qPCR data and in line with previous literature findings, indicating dp140 abundancy during the early stages of life. At embryonic CS stages 17 and 23 Dp140 transcripts were most abundant and widely expressed in the developing CNS, including the telencephalic cortical plate, dorsal hippocampus, thalamic, mesencephalic, cerebellar and brain stem neuroepithelia. Dp140 transcripts were noted in the subventricular germinal neuroepithelium as well as the early migrating neurons. Co-expression of dp140/dp427c isoforms was also detected in the CS17 embryonic pons and during the foetal 14pcw stage in cerebellum, hippocampus and cingulate gyrus (Fig. 10a). Dp427c transcripts were also ubiquitous, but fewer compared to Dp140, except in the mesencephalon where they were seen in abundance.

At 14 pcw, the developing cerebellum showed preponderance of Dp427c transcripts in the Purkinje cell zone, whereas Dp140 transcripts were more frequent in the external granule cell layer and the dentate nucleus. Expression of Dp71 and Dp40 was not assessed in the embryonic and foetal samples.

Altogether, results from singleplex and duplex Basescope assays confirmed that dp427c is abundantly expressed in neurons across all brain regions from early development to adulthood. Dp427m is expressed mainly in the leptomeningeal blood vessels, and small number of oligodendroglial cells and neurons (Fig. S5a–c). Dp140 is highly expressed in embryonic and foetal brains and continues to be expressed at low to moderate levels in the adult forebrain, cerebellum and the brain stem. Dp71 is widely expressed in neurons, as well as the parenchymal microvasculature and the glia limitans. The adult human hippocampus (dentate granule cells), cerebellum (Purkinje cells) and pons (neurons in the pontine nuclei) show the greatest diversity of dystrophin isoform expression.

DMD expression in excitatory and inhibitory neurons

Due to constraints in multiplex immunohistochemistry in FFPE tissues, we used duplex RNAscope comprising dual labeling of DMD 5′ and DMD 3′ probes with GAD1 as a proxy for GABAergic inhibitory neurons and SLC17A7 as a proxy for glutamatergic excitatory neurons. In the 14 pcw cerebellum, the migrating Purkinje cells coexpressed GAD1 and DMD 3′ transcripts (Figs. 11a, b). In contrast, in the granular cell layer, few cells coexpressed SLC17A7 and DMD 3′ transcripts (Fig. 11c, d). In the adult cerebellum, this distinction was even more pronounced, with coexpression of DMD 5′ and GAD1 transcripts in the molecular layer interneurons (Fig. 11k), Purkinje cells (Fig. 11e) and majority of cells in the dentate nucleus (Fig. 11f), whereas DMD 3′ and SLC17A7 transcripts were coexpressed mainly in the granule cells (Fig. 11i) and a small number of neurons in the dentate nucleus.

Fig. 11.

Detection of DMD isoforms, GABAergic and glutamatergic neurons in developing and adult human brain areas by duplex RNAscope. Representative image of experiments performed in 14pcw cerebellum (a–d). Representative images of the results from the adult samples (e–n). Green dots indicate RNAscope DMD 5′ and DMD 3′ transcripts. Red signal indicates GAD1 (GABAergic) or SLC17A7 (glutamatergic) transcripts. Black arrows indicate migrating Purkinje neurons (a) external granular layer (c) GAD1- or low GAD1+ Purkinje neurons (e, g). The green arrow shows sporadic DMD 3′ signal in d. Red arrows indicate GAD1+ Purkinje neurons (e, g) and GAD1+ dentate neurons (f, h) in the cerebellum. Scale bars are: a, e, l = 50 µm; all other images = 10 µm

In the adult hippocampus, an overwhelming majority of dentate neurons and pyramidal cells (CA1-CA3) coexpressed DMD3′ and SLC17A7 transcripts (Fig. 11l, n), and a small number of cells, most likely interneurons coexpressed DMD 3′ and GAD1 transcripts (Fig. 11m). A similar pattern was evident in the neocortical areas examined, with most pyramidal cells coexpressing DMD 3′ and SLC17A7 transcripts (Fig. S6b–f). Notably, there was no evidence of labeling in the subcortical and/or deep white matter with any of the probes above.

Immunohistochemical localisation of the dystrophin protein in the developing and adult human brain

In order to establish whether there was concordance of dystrophin transcript and protein expression, developing and adult brains were immunostained with the antibodies to dystrophin, DYS2, that reacts with the C-terminus, hence detects all isoforms apart from dp40, as well as DYSA. The latter antibody recognises an undisclosed epitope in the rod domain, but previously characterised in mice, and shown to detect both full-length (dp427) and shorter (dp71) isoforms of the dystrophin protein [30].

At stage CS17, DYS2 reactivity was detected in the cerebellum, pons, and medulla, whereas DYSA staining was negative (Fig. 12a), indicating that the longer isoforms are not present, or their expression is below the sensitivity of the method, at this early developmental stage. Similarly, only DYS2 reactivity was observed in the CS23 cerebellum (Fig. 12b), whereas weak DYSA reactivity was detected in the CS23 trigeminal ganglia (Fig. 12b). At the foetal stage examined, 14pcw, both cerebellum and hippocampus showed some DYS2 staining but very weak DYSA reactivity, suggesting predominant expression of shorter isoforms (Fig. 12c).

Fig. 12.

Dystrophin detection by immunohistochemistry in developing and adult brain regions. Developing brains stained either with the monoclonal antibody DYSA (dystrophin N-Terminus) or DYS2 (dystrophin C-Terminus) (a–c). Adult brain sections stained with DYSA (d). Adult upper and lower motor neurons stained with DYSA and RNAscope DMD 5′ and DMD 3′ probes (e–f). Pcw postconceptional weeks, CS Carnegie stage. Scale bars are: a = 1 mm (overview) and 50 µm; d = 100 µm; b and c = 50 µm all other images = 10 µm

In contrast, in adult brains DYSA strongly stained several neuronal cell types. Dystrophin expression across all neocortical regions was pan-cortical (Fig. 12d), sometimes with accentuated labelling in layer II and layer V/VI. The expression pattern was more readily discernible within the pyramidal cells, partly owing to the size of these neurons. Moderate to strong labelling was observed within the neuronal cell body, with accentuated punctate distribution along the cell surface and the dendritic trees, whereas the axons were negative. There was strong dystrophin expression in the upper motor neurons, so-called giant pyramidal cells of Betz, within the precentral gyrus (Fig. 12e).

In the hippocampus, dystrophin immunoreactivity was noted within the dentate granule cells, hilar/CA4 region, Ammon’s horn (CA1–CA3) and the subiculum, restricted to the neuronal somata and their dendritic trees (Fig. 12d). Within the amygdaloid complex the presence of large amounts of lipofuscin pigment confounded ascertainment of dystrophin signal within the neurons (Fig. 12d).

In the cerebellum, Purkinje cells displayed a distinct punctate labelling pattern at the surface of the cell bodies and their dendritic arbour (Fig. 12d). Small neurons within the molecular layers were also labelled as were a significant proportion of the cells in the granular cell layer. In the midbrain, weak labelling was seen in the oculomotor nerve nuclei. The presence of abundant neuromelanin in the substantia nigra pigmented neurones confounded the assessment for dystrophin expression. Strong labelling were noted within the neurons (pontine nuclei) and the surrounding neuropil in the pontine base. In the medulla, the neuropil within the inferior olivary nuclear ribbon showed a strong DYSA signal. Unexpectedly, there was strong labelling of the axons within the vagus and hypoglossal nerves. The respective nuclei displayed dystrophin expression in the surrounding neuropil. The anterior horn cells in the spinal cord showed robust dystrophin expression within the neuronal cell bodies and dendrites (Fig. 12e). Of note, these motor neurons expressed both 5′ and 3′ labeled DMD transcripts (Fig. 12e, f).

Dystrophin and DMD expression within the non-neural cell types

In the CS23 choroid plexus, only RNAscope DMD 3′ labeling was seen (Fig. 13a). In contrast, both probes labelled the choroid plexus at 14pcw, but antibody reactivity was clearly detected only with DYS2 (Fig. 13b). At CS23, only the RNAscope DMD 3′ signal was noted in the blood vessel wall with corresponding exclusive DYS2 immunoreactivity. At 14 pcw stage, the vessel walls showed expression of both RNAscope DMD 5′ and DMD 3′ labeled transcripts, and immunoreactivity for both DYSA and DYS2 antibodies (Fig. 13e). In the adult brain, the leptomeningeal, penetrating arteries and arterioles showed predominant expression of RNAscope DMD 5′ labeled transcripts localised to the vascular smooth muscle cells (Fig. 13f, g). At the isoform level, dp427m was consistently expressed in the smooth muscle coats of the leptomeningeal and cortical penetrating arteries and arterioles (Fig. 13f, g). Dp71 transcripts were noted in the parenchymal capillary walls and the glia limitans (Fig. 13h, i).

Fig. 13.

Dystrophin protein and transcript expression in brain vasculature. Embryonic and foetal choroid plexus and blood vessels stained with DYSA and DYS2 antibodies and hybridised with the RNAscope DMD 5′ and DMD 3′ probes (a–e). Adult blood vessels stained with DYSA antibody and processed for RNAscope or Duplex Basescope (f–i). Note positive signal detected with duplex Basescope (Dp427m-green/Dp71-red), RNAscope DMD 5′ and DMD 3′ probes, and the DYSA antibody within the tunica media of a medium-sized leptomeningeal artery (f) and in a cortical penetrating arteriole (g) in the ventromedial prefrontal/orbitofrontal cortex. Note also expression of the Dp71 transcript and DYSA immunoreactivity in the cerebellar cortical parenchymal capillary (h) and cerebellar pial–glial membrane (glia limitans) at the cortical surface (i). Each red arrow indicates RNAscope or duplex Basescope (dp71) signal (red dots) while green arrows indicate duplex Basescope (dp427m) signal (green dots). Scale bars are: f, g, h, i = 25 µm; all other images = 10 µm

Developmental regulation of dystrophin protein isoforms in the human brain assessed by capillary western blot

As immunohistochemical cellular localisation of different dystrophin isoforms in FFPE sections was hindered by the lack of isoform-specific dystrophin antibodies, analysis of protein isoform expression and semi-quantification was carried out by WES using a C-terminal antibody (Ab154168) in protein extracts from frozen developing and adult brains (Fig. 14). Protein peaks with the molecular weights of dp427, dp140, and dp71 were detected in all the brain areas from developing and adult samples, while dp116 was detected in 3 out of 5 adult brain areas (cerebellum, amygdala, and dorsolateral prefrontal cortex).

Fig. 14.

Detection of dystrophin in developing and adult human brain regions by capillary western blot. Relative dystrophin protein concentration (Dp427) in brain/brain regions from embryos (n = 2), foetuses (n = 15) and adults (n = 12) detected by the C-terminal antibody Ab154168. Pcw postconceptional weeks, CS Carnegie stage

During development, apart from CS17, dp140 was the predominant isoform detected in all brain regions assayed with an average relative protein concentration (RPC) of 0.58, followed by dp71 (average RPC = 0.25), and dp427 appeared to be the less expressed isoform (average RPC = 0.11), with the exception of CS17 brain (RPC = 0.6). In contrast, dp71 was the predominant isoform in all adult brain regions, and dp140, though expressed throughout was greatly reduced as compared to developing brains, as particularly evident in the cerebellum (RPC = 0.73 in developing cerebellum versus 0.07 in adult). Dp116 was selectively expressed in adult cerebellum (RPC = 0.24), amygdala (average RPC = 0.32) and dorsolateral prefrontal cortex (RPC = 0.28), but not in cingulate and frontal gyrus nor in embryonic/foetal samples. Although this technique did not allow us to pinpoint which of the long isoforms was expressed, it showed that dp427 was expressed, though to different extents, in all adult brain regions examined.

Together, good concordance between protein and transcript isoform expression was observed, with changes in dp140 and dp71 protein expression from developing to adult brain closely mirroring transcript changes, with the former being higher prenatally and the latter postnatally. However a degree of discordance between the levels of some protein and transcript was noted especially in some adult brain areas, potentially due to non-specific tissue contamination in the frozen specimen processed for WES.

Discussion

This study provides a comprehensive analysis of isoform-specific dystrophin expression in developing and adult pathology-free human brains combining a variety of quantitative methods (RT-qPCR and immunoblot), in situ hybridisation and immunohistochemistry techniques that deliver, for the first time, a comprehensive neuroanatomical, cellular and sub-cellular context to the dystrophin expression profile.

Our RNAscope studies, that allow for spatial localisation of the DMD transcripts, provide novel information on the complexities of DMD expression in the human brain. The DMD 5′ signal appeared as both large juxtanuclear/intranuclear dots, suggestive of transcriptional ‘hotspots’ containing multiple full-length nascent and mature transcripts, whereas the DMD 3′ signal consisted of multiple small cytoplasmic dots, indicating the presence of nascent and mature transcripts as previously described in the dog by Hildyard et al. [38].

Our results show that in the human brain the DMD 3′ signal was robust throughout the embryonic and foetal developmental stages, whereas the DMD 5′ signal increased from early to late development. This is consistent with the work by Doorenweerd et al., but it should be noted that the study did not cover embryonic stages of development, nor localised the transcripts at the cellular level [12]. Here we show that DMD transcripts were already abundant at an early developmental stage, CS17 (6 weeks of gestation) and mainly localised to neuroepithelial cells. This is consistent with DMD expression in the developing dog brain [38], suggesting an important and conserved role of DMD in early neural development. Significantly, the dystrophin interacting protein, dystroglycan, has been suggested to play a role in the maturation of the mouse perinatal neural stem cell niche, partly via regulation of Notch signalling, an important pathway in neural development [39, 40]; therefore, it is tempting to speculate that dystrophin may modulate Notch signalling in the human neuroepithelium.

Furthermore, the human hippocampus and cerebellar Purkinje cell layer also showed high expression of DMD transcripts. The complete absence of DMD 5′ signal in the granule cells of the developing and adult cerebellum was striking, supporting a cell-specific specialised role for the shorter isoforms, as suggested in dog cerebellum [38, 41]. The granular layer of the cerebellum appears to express mainly shorter isoforms, as in the dog, whereas both short and long isoforms are detected in the Purkinje cells. Altogether, our RNAscope data showed that DMD transcripts are ubiquitously expressed in the developing and adult human brain.

When isoform-specific DMD transcripts were spatially mapped by Basescope and quantified by RT-qPCR, dp427c showed consistent high expression across most regions in the developing and the adult brain. This is in contrast to the study by Doorenweerd et al. showing low-level expression across human brain development [12]. Also in contrast to previous findings reporting no expression of dp427p in the developing and adult human brain [4, 12], we found this isoform highly expressed (RT-qPCR) in the embryonic and foetal brains and still detectable, though at low levels, in some adult brain regions. Interestingly expression of two dp427p variants was detected by RNA protection assay by Holder et al. [4] in adult brains. It is conceivable that the differences observed across studies are due to different techniques used or probes targeting different dp427p variants. It should also be noted that dp427p2 is believed to translate a truncated protein, which may or may not have a function, whereas dp427p1 translates a full length protein [4, 42].

Like dp427p, expression of dp427m was high during development, but more variable in the adult brain. Most of the transcripts were expressed in the leptomeningeal and/or the parenchymal blood vessels within the smooth muscles. Unexpectedly, as not previously reported, consistent expression of dp427m, though at very low-levels, was observed in neurons and oligodendroglial-like cells across different brain regions.

Dp140 was highly expressed across all brain regions in the embryonic and foetal brain. In the adult brain its expression was significantly downregulated, with sparse to few transcripts noted in the cerebellum (Purkinje cells, granule cells and dentate neurons), hippocampus (dentate granule cells and hilar neurons), amygdala and across all neocortical areas (mostly in pyramidal neurons). Similarly, prior investigations reported a developmental regulation of dp140 characterised by high transcripts expression in foetal samples which remain stable in cerebral cortex and cerebellum in adults [12]. Our findings also support the hypothesis that the dp140 isoform is pivotal during development and persists in the adult brain in areas likely to undergo adult neurogenesis, such as the subgranular zone in the dentate gyrus of the hippocampus, as previously proposed [12, 38].

In keeping with earlier findings, dp71/dp40 expression was consistently high during development and persisted in the adult brains across all regions. Transcripts were identified within neurons, parenchymal capillary walls (likely astroglia), and the glia limitans. Low-to-moderate dp40 expression was noted throughout development and persisted in the adult brain.

In addition, we have demonstrated for the first time co-expression of two different DMD isoforms within the same neuron by duplex Basescope assays. This can be observed from the early developmental stages to adulthood across all anatomical brain regions. However, neurons co-expressing DMD isoforms represent less than approximately 2% of the total neuronal population.

It should be noted that gene co-expression studies suggest distinct roles for dp427, dp140 and dp71-related dystrophin signalling in the brain. Our study has also shown that the granule cells of the dentate gyrus in the hippocampus, neurons in the pontine nuclei, and the cerebellar granular cells displayed the most diversity in DMD isoform expression.

With regard to protein expression our western blot results showing good agreement with the RT-qPCR data, confirmed for the first time that the dp140 protein was the predominant isoform expressed in embryonic brains, whereas dp71 was more abundant in adult brains.

Dp427 was consistently expressed throughout development and in adult life. Unexpectedly, we detected expression of dp116 both at the transcript and protein level in developing and adult brains, mainly in the cerebellum and dorsolateral prefrontal cortex. This indicates that dp116 expression may not be restricted to the Schwann cells as previously suggested [12]. We also showed by RT-qPCR that dp260-1 is the only DMD isoform not expressed in the CNS.

Substantial presence of dystrophin in the brain was further supported by immunolabeling with the DYSA antibody showing dystrophin expression mainly in the somata and proximal dendrites of a diverse population of neurons, including pyramidal cells in the neocortex and hippocampus, granule cells of the dentate gyrus, Purkinje cells, granule cells and dentate neurons in the cerebellum, and neuronal populations in the deep grey nuclei (basal ganglia and thalamus), brain stem and cervical cord (anterior horn cells).

The use of duplex RNAscope assays allowed us to study cellular coexpression of dystrophin and markers of GABAergic (GAD1) and glutamatergic (SLC17A7) neurons circumventing limitations of other approaches (e.g. poor reactivity of dystrophin antibodies in FFPE brain sections and lack of isoform-specific dystrophin antibodies), and to demonstrate dystrophin expression in both GAD1+ inhibitory interneurons and SLC17A7+ excitatory neurons in all neocortical regions, cerebellum and the hippocampus. Furthermore, the extensive co-expression of GAD1+ and DMD 3′ transcripts observed in the 14pcw developing Purkinje layer suggest that dystrophin might play a crucial role in the cerebellar GABAergic circuitry maturation. Our data also demonstrate a switch from these isoforms to full length dystrophin in adult GAD1+ Purkinje neurons, as indicated by significant expression of DMD 5′ transcripts in these neurons. Transient expression of non-full length isoforms in GAD1+ Purkinje neurons is interesting, as Dp71-null mice display alterated glutamatergic transmission and cognitive deficits [28], and is consistent with a realatively early onset of developmental defects leading to at least some of the neural deficit observed in DMD patients.

The dystrophin isoform brain localisation reported here provides an important underpinning to better understand the common brain involvement in DMD patients.

Notably, although impacted by limitations mainly related to small sample size, clinical studies showed that proximal and distal DMD mutations are linked to type and severity of different brain co-morbidities [13, 43, 44].

In most patients, distal mutations are associated with more severe intellectual disability mainly linked to the loss of dp140 and dp71 [13, 45–47] leading to reduced intelligent quotients and working memory index scores [13] along with poorer academic performances [48].

In contrast, proximal mutations affecting only dp427 have been associated with neuropsychiatric manifestations ranging from an increased startle response, indicative of enhanced anxiety and authistic spectrum and attention deficit hyperactivity disorders, without significantly affecting intelligence quotient [13, 14, 49].

DMD has always been regarded as a primary muscle disease. There are no reports of dystrophin expression within the motor systems of the central nervous system in mouse models or in human tissues to date, to the best of our knowledge. In this context, our finding that DMD transcripts (5′ and 3′) were present in the upper motor neurons of the primary motor cortex, and the anterior horn cells in the spinal cord in the adult brain is interesting, as it suggests a role for dystrophin in synaptic signalling in the upper and lower motor neurons. An absence of one or more isoforms in these motor neurons could contribute to the motor deficits seen in DMD.

Our findings on dystrophin expression in brain blood vessels both during development and in adults are consistent with its previously reported expression in muscle blood vessels; interestingly, impaired angiogenesis was observed in muscle of DMD patients and animal models [50–53]. In addition, consistent with a possible vascular phenotype in DMD patient brain is the reduced perfusion observed in the mdx mouse brain [54] and impaired cerebral blood flow in DMD patients demonstrated by magnetic resonance imaging studies [55].

We are aware that our study has a number of limitations. Although we presented a temporally resolved dystrophin expression profile, the numbers of human brain samples in each age-group are very small. Information regarding dystrophin expression in the crucial post-natal and paediatric age-group is missing, because pathology-free brains for assessment in this age group were not available. Inherent variability (biological and technical) within and in between samples must be taken in to account while interpreting the data. In particular, accurate quantification of the in situ hybridisation signal was hampered by the sheer number and complexity of neuronal and glial elements in tissue sections, further compounded by the lack of clear cell boundaries, and often saturation of the particulate chromogenic signal within individual neurons.

Given the rapid advancement of single-cell and single-nucleus RNA sequencing technologies as well as single cell proteomics, it will eventually be possible to unravel the complexity of the brain dystrophin cellulome/proteome/interactome. However, validation of such findings in human pathology-free brains as well as DMD brains will remain challenging due to the aforementioned reasons.

Conclusions

In summary, we have shown that DMD and dystrophin isoforms are abundantly expressed in the human brain across the lifespan and are developmentally regulated, and localised them at single-cell resolution. We have demonstrated for the first time co-expression of different DMD isoforms across several key brain areas in the developing and adult human brain at the regional and single-cell level.

We also have shown for the first time ubiquitous expression of dystrophin within different glutamatergic neuronal populations in the developing human brain. This finding should prompt future studies investigating the molecular pathology of dystrophin signalling at the neuronal excitatory synapse. In addition, we have provided novel information on DMD gene expression in GABAergic and glutamatergic neuronal populations in the human brain. The granular localisation of dystrophin reported here will also underpin therapeutic developments aimed at restoring dystrophin functionality in the CNS, hence reducing the burden of brain comorbidities and improving the quality of life of DMD patients.

Author contributions

FC, DC, RA, SS and AA carried out the experiments. FC, DC, SS and RP performed semi-quantitative scoring. RP and FC designed the analysis. FC performed data analysis and wrote the manuscript. FC, PF, RP and JM edited the manuscript. FM, PF and RP contributed to the conception of the project. FM obtained funding.

Funding

Sarepta Therapeutics Grant 557462.

Availability of data and materials

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Human tissue procedures were conducted in accordance with the regulations with informed consent in accordance with the UK Human Tissue Act 2006 for study participation under ethical approval (NRES Committee London—Fulham, UK; REC: 18/LO/0822). The human embryonic and foetal material was provided by the Joint MRC/Wellcome Trust (Grants MR/006237/1, MR/X008304/1 and 226202/Z/22/Z) Human Developmental Biology Resource (https://www.hdbr.org). Adult human brains were selected and procured from the Edinburgh Brain Bank (Ethic Ref 20/PR/0582). Tissue samples were obtained from the Edinburgh Brain Bank (EBB) and BRAIN UK, which is supported by Brain Tumour Research and has been established with the support of the British Neuropathological Society and the Medical Research Council.

Consent for publication

Not applicable.

Competing interests

As COI, FM is an investigator in Sarepta, Genethon, Roche clinical trials. He is the PI of the Sarepta grant supporting this work (funding to UCL), and has participated to advisory boards and/or symposia for Sarepta, Roche, Dyne therapeutics, Wave and Entrada.