Abstract
Rare individuals with hypomorphic inactivating mutations in the Huntington’s Disease (HD) gene (HTT), identified by CAG repeat expansion in the eponymous neurodegenerative disorder, exhibit variable abnormalities that imply HTT essential roles during organ development. Here we report phenotypes produced when increasingly severe hypomorphic mutations in Htt, the murine HTT orthologue (in HdhneoQ20, HdhneoQ50, HdhneoQ111 mice), were placed over a null allele (Hdhex4/5). The most severe hypomorphic allele failed to rescue null lethality at gastrulation, while the intermediate alleles yielded perinatal lethality and a variety of fetal abnormalities affecting body size, skin, skeletal and ear formation, and transient defects in hematopoiesis. Comparative molecular analysis of wild-type and Htt-null retinoic acid-differentiated cells revealed gene network dysregulation associated with organ development and proposed polycomb repressive complexes and miRNAs as molecular mediators. Together these findings demonstrate that the HD gene acts both pre- and post-gastrulation and possibly suggest pleiotropic consequences of HTT-lowering therapeutic strategies.
Author Summary The HTT gene product mutated in Huntington’s Disease (HD) has essential roles during normal organism development, however, still not fully predictable are the functional consequences of its partial inactivation. Our genetic study provides a comprehensive effects’ description of progressively stronger suppression of Htt gene, the murine HTT counterpart. The most severe Htt reduction leads to embryo lethality, while intermediate Htt dosages yield a variety of developmental abnormalities affecting body size, skin, skeletal and ear formation, and hematopoiesis. Complementing molecular analysis in differentiating cells depleted of a functional Htt gene further elucidates genes’ networks dysregulated during organ development and proposes chromatin regulators and short non-coding RNAs as key molecular mediators. Together these findings demonstrate that the HD gene acts both at early and later stages of development, thus possibly suggesting long-term consequences associated to the newest HD therapeutic strategies aimed at lowering the HTT gene product.
Introduction
Huntington’s Disease (HD) is a dominantly inherited neurodegenerative disorder characterized by motor, cognitive and behavioral signs, generally of mid-life onset [1]. HD is caused by an unstable CAG trinucleotide repeat expansion in the 4p16.3 gene HTT (previously HD) [2]. The size of the expanded repeat is strongly correlated with age at onset but genetic variants at other loci, including DNA maintenance genes involved in somatic repeat instability, can modify the rate of pathogenesis [3].
Though knowledge of the pathogenic rate driver is emerging from HD genetic studies, it is not yet evident which aspect of the HTT expansion mutation is harmful to the cellular targets whose dysfunction or demise contributes to the disorder. The observation that CAG repeat expansions at different unrelated genes yield distinct neurological disorders argues that the harmful entity may be at the level of the HTT-encoded protein (huntingtin), through some opportunity afforded by the mutant protein’s normal function [4]. Consistent with this hypothesis, HD CAG expansion homozygotes exhibit onset similar to HD heterozygotes [5] and HTT inactivating mutations do not produce HD. Instead, rare humans with a single functional HTT copy [6], and mice with a single functional HTT orthologue (Htt, previously Hdh), whether a wild-type or a CAG repeat knock-in allele [7-9], are unremarkable. By contrast, complete inactivation of both alleles is early embryonic lethal [7,10,11].
The essential role played by HTT early in development is hypothesized, from genetic studies in the mouse, to intersect with chromatin regulation, influencing for example, the histone methyltransferase polycomb repressive complex 2 (PRC2) [12-14], although the lethality of null alleles has hampered studies later in development [7]. However, a recent report of a family segregating two rare, apparently incompletely inactivating HTT mutations [15,16], emphatically demonstrates that HTT is also essential for proper development of the brain and possibly other organs, as compound heterozygote children exhibit variable neurodevelopmental features, including motor disturbance, hypotonia, abnormal cranial circumference and small stature, as well as early death (OMIM: 617435) [16]. These observations are generally consistent with findings from genetic studies with members of a hypomorphic allelic series of flox-pGKneo-in CAG repeat knock-in mouse lines. HdhneoQ20, HdhneoQ50 and HdhneoQ111 mice carry a mild, intermediate and severe hypomorphic neo-in allele, respectively [8,9,17]. Compound heterozygotes with combinations of these alleles, as well as homozygotes, exhibit a number of evident, variable abnormalities, involving brain development, movement deficits, decreased body size and reduced survival [9,17].
We have systematically evaluated each of the neo-in CAG repeat knock-in alleles for the ability to support proper development when placed over the same Hdhex4/5 null allele in order to assess the extent to which dosage of the HD gene mouse orthologue, below a single functional copy, is essential for normal development and to uncover processes sensitive to gene dosage. Our findings, augmented by molecular analysis of Hdhex4/5 embryonic stem cells (ESC) and retinoic acid differentiated cells, indicate that in addition to the brain, the proper development of several other organ systems requires huntingtin, as revealed by different developmental blocks that are by-passed at different dosages. Our data also highlight potentially responsible regulatory factors and networks that are sensitive to inactivating mutation of Htt.
Results
HdhneoQ20, HdhneoQ50, HdhneoQ111 hypomorphic mutation and Hdhex4/5 null rescue
HdhneoQ20, HdhneoQ50 and HdhneoQ111 mice were bred to Hdhex4/5 mice, with a targeted null allele [7], and their progeny were genotyped at weaning and at earlier stages, to assess the ability of each hypomorphic allele to rescue the early embryonic lethality at ~E7.5-8.0 imposed by complete gene inactivation. The results, summarized in Table 1, indicate the expected Mendelian ratio for progeny with wild-type alleles, at all ages, but revealed lethality for hypomorph mutation hemizygotes (with the Hdhex4/5 null allele) beginning at progressively earlier stages with increased severity of inactivation. The HdhneoQ20/Hdhex4/5 diplotype was recovered at the expected Mendelian ratio at E14.5 and embryos appeared normal, while the HdhneoQ50/ex4/5 was recovered at the expected Mendelian ratio at E10-12.5 but the embryos were smaller than wild-type. Finally, the HdhneoQ111/Hdhex4/5 diplotype was seen at E8.5, though all embryos had the abnormal sock-like appearance of Hdhex4/5/Hdhex4/5 null embryos just post-gastrulation, lacking head-folds (Figure 1A). HdhneoQ111/Hdhex4/5 embryos (non-resorbed) were not observed at later stages. HdhneoQ20/Hdhex4/5 and HdhneoQ50/Hdhex4/5 survivors were recovered at later stages but these died perinatally and at birth, respectively, displaying abnormalities that increased in severity with the gradient of inactivation HdhneoQ20/Hdhex4/5 < HdhneoQ50/Hdhex4/5: dome shaped cranium-to-overt exencephaly and mild-to-robustly decreased height and weight (Figure 1A-E).
HdhneoQ20/Hdhex4/5 and HdhneoQ50/Hdhex4/5 have deficits in multiple organ systems
In previous studies of Hdhex4/5/Hdhex4/5 null embryos, embryoid bodies (EBs), as well as cultured ESC and neuronal cells, we demonstrated that a functional Htt allele is required for proper PRC2 activity, during genome-wide deposition of H3K27me3 epigenetic chromatin marks [12,13]. We therefore surveyed HdhneoQ20/Hdhex4/5 and HdhneoQ50/Hdhex4/5 embryos, mainly at E18.5, to evaluate a number of organ systems whose development is reported to require appropriate PRC2 activity and/or proper activity of its co-regulator polycomb repressive complex 1 (PRC1). Skeletal elements, ear development, skin barrier formation and fetal liver hematopoiesis were assessed.
Skeleton
HdhneoQ20/Hdhex4/5 and HdhneoQ50/Hdhex4/5 embryos exhibited lumbar to sacral (L6 to the S1) vertebral transformation whose penetrance (20% and 30%, respectively) increased with the severity of inactivation. This abnormality was not observed in embryos with one wild-type allele (Figure 2A, Table 2). In addition, 50% of HdhneoQ20/Hdhex4/5 and 62.5% of HdhneoQ50/Hdhex4/5 embryos but none of the embryos with a wild-type allele displayed variable abnormalities of the sternum and xyphoid process (Figure 2B, Table 2), although the phenotypes were distinct. The HdhneoQ50/Hdhex4/5 xyphoid process was narrow, with reduced ossification, while the HdhneoQ20/Hdhex4/5 process was fenestrated. Furthermore, the tip of the HdhneoQ50/Hdhex4/5 sternum was bent inward, perhaps to accommodate a narrower thoracic cavity (Figure 2B, Table 2). The cervical vertebrae gap (C1 to C2 gap) was abnormally increased in 100% of HdhneoQ50/Hdhex4/5 embryos, perhaps secondary to the domed cranium and/or exencephaly (Figure 2C-D, Table 2). Variable hyoid bone and cartilage defects also increased with severity of the inactivating allele, such that 62% of HdhneoQ50/Hdhex4/5 and 10% of HdhneoQ20/Hdhex4/5 embryos had an abnormally short hyoid bone, greater horns and abnormal thyroid and cricoid cartilage morphology or spacing of the thyroid cartilage relative to the hyoid bone, respectively (Figure 2E, Table 2).
Ear
External ear and middle ear structures were also abnormal, increasing with the severity of the mutant allele: 100% of HdhneoQ50/Hdhex4/5 embryos had (bilateral) hypoplastic pinna and occasionally lack of the structure, while 33% of HdhneoQ20/Hdhex4/5 had (unilateral) hypoplastic pinna (Figure 3A-B, Table 3). Impact on middle ear structures further distinguished the inactivating alleles. Whereas all HdhneoQ20/Hdhex4/5 embryos had middle ear components, 75% of HdhneoQ50/Hdhex4/5 embryos lacked the tympanic ring, gonium, malleus, incus, and the third ossicle stapes, although Meckel’s cartilage was intact (Figure 3C, Table 3). The squamous bone was hypoplastic (Figure 3D, Table 3). For both inactivating mutations, inner ear structures were normal (data not shown). At E10.5, Gsc and Hoxa2 expression in HdhneoQ50/Hdhex4/5 embryos was abnormal. Gsc mRNA, detected in the first and second branchial arches in wild-type embryos, was restricted to the first branchial arch. Hoxa2 mRNA was reduced in intensity at the level of the branchial arches although intense staining was detected dorsally, suggesting a delayed or impaired migration of the neural-crest derived cells to colonize the arches, thereby possibly affecting the proper formation of ear structures (Figure 3E-F).
Skin barrier
E18.5 HdhneoQ20/Hdhex4/5 embryos, like embryos with a wild-type allele, excluded toluidine blue dye, demonstrating proper formation of the cornified layer that serves as the skin barrier, but 100% of the HdhneoQ50/Hdhex4/5 hypomorphs excluded dye from the dorsal, but not the ventral epidermis (Figure 4A). The four stratified layers of the epidermis (basal, spinous, granular, cornified) were present (data not shown). However, immunoblot analysis indicated decreased levels of mature filaggrin (27 kDa, epidermal granular layer marker [18]), profilaggrin processing intermediates (50-70 kDa), and of loricrin (epidermal cornified layer marker) (Figure 4B). HdhneoQ50/Hdhex4/5 ventral epidermis also exhibited decreased PCNA staining (S-phase cell cycle marker) and increased TdT-mediated dUTP nick-end labeling (TUNEL)-positive apoptotic cells, compared to wild-type (Figure 4C-D), consistent with failed differentiation.
Hematopoiesis
At E14.5, HdhneoQ20/Hdhex4/5 embryos were unremarkable but HdhneoQ50/Hdhex4/5 embryos had small, pale livers, contrasting with abnormal vasculature and extensive blood inclusions in the head (brain and ventricles) reported previously [9] (Figure 5A, Suppl. Fig. 1). Flow cytometry (FACS) of fetal liver cells showed decreased absolute cell number and lack of erythropoiesis (CD71hiTer119neg-lo proerythroblasts to CD71hiTer119+ basophilic erythroblasts to CD71+Ter119+ polychromatic erythroblasts to CD71−Ter119+ reticulocytes) in HdhneoQ50/Hdhex4/5 embryos (Figure 5B, C), concomitant with increased numbers of myeloid (Mac-1+Gr-1−) and B cell progenitors (B220+CD19−) (Figure 5B, C). Furthermore, the numbers of erythroid (CFU-E) and megakaryocyte (CFU-Mk) colony forming progenitors were decreased, along with a decrease in the total number of colony forming cells (Figure 5D), implying impaired expansion of HSCs/progenitor cells and defective erythropoiesis. The impairment was transient, as E18.5 HdhneoQ50/Hdhex4/5 livers did not show these hematopoietic deficits (Figure 5E-I). Analysis of thymus at this age also did not reveal deficits in the development of CD4+CD8+ double positive T cells suggesting normal T-lymphopoiesis, despite an apparently decreased number of thymic precursors (Figure 5E-I).
Hdhex4/5/ex4/5 developmental response to retinoic acid-induced differentiation
In comparisons of wild-type and Hdhex4/5/ex4/5 null embryonic stem cells (ESC), at baseline and after Embryoid Bodies (EB) retinoic acid (RA)-induced differentiation, we have found previously that complete Htt inactivation impairs PRC2 activity [12,13] and alters genome-wide deposition of epigenetic chromatin marks [12,13]. Since RA-differentiation of ESC provides a culture system with which to study not only neurogenesis but organogenesis [19-21] more broadly, we utilized this paradigm to identify transcriptional regulators responsive to Htt dosage that may play a role in the incomplete hypomorph rescue of the Hdhex4/5/ex4/5 null allele. Specifically, we performed differential gene expression analyses of genome-wide RNA and miRNA next generation sequencing data (Suppl. Table 1) generated from wild-type parental ESC, Hdhex4/5/ex4/5 ESC and from wild-type and Hdhex4/5/ex4/5 EB differentiated cells, treated with RA. The overall gene expression changes, upon RA-induced differentiation, were quantitatively and qualitatively similar for wild-type and Hdhex4/5/ex4/5 cells: 79% of the several hundred changed RA-responsive miRNAs were shared (Suppl. Fig. 2, Suppl. File S1) and 86% of the more than 9,000 mRNA gene expression changes were shared (Suppl. Fig 2, Suppl. File S1). Functional annotation enrichment analysis (clusterProfiler) [22] highlighted genes involved in organogenesis in a large differentiation up-regulated cluster (Cluster 1, Suppl. Fig. 2E), associated with terms such as Skeleton, Ossification, Ear, Skin and Neuron development (Suppl. Fig. 2, Suppl. File S2), as well as genes involved in subcellular processes in a smaller down-regulated class (Cluster 2) (Suppl. Fig. 2E), with enriched terms such as DNA metabolic process, Ribosome biogenesis, Cell Cycle (Suppl. Fig. 2, Suppl. File S2). We then determined the impact of Htt nullizigosity on development, assessing the initial pluripotent stage (wild-type versus Hdhex4/5/ex4/5 ESC) as well as the differentiated stage (wild-type versus Hdhex4/5/ex4/5 RA-induced) (Suppl. Fig.3, Suppl. File S1). The results disclosed a modest number of miRNAs (ESC: 149 up: 95 down; RA: 70 up; 86 down) and mRNAs (ESC: 699 up: 461 down; RA 1314 up; 1784 down), whose expression was sensitive to Htt inactivation. These were largely specific to a given developmental stage (Suppl. Fig.3, Suppl. File S1), forming four non-overlapping gene sets (ESC-up, ESC-down, RA-up and RA-down) (Suppl. Fig. 3, Suppl. File S2). Gene Ontology and pathway enrichment analysis revealed highly significant enrichment of the Htt-inactivation sensitive RA-down gene set in developmental pathways related to organ system deficits observed in the hypomorphic mice: ‘skeletal system development’, ‘generation of neurons’, ‘blood vessel morphogenesis’, ‘blood vessel morphogenesis’, “skin development” and “ear development” (Suppl. Fig. 3).
In addition, we examined the interplay between RA-differentiation and Htt-inactivation, generating four gene expression sets: up_up, up_down, down_up and down_down, depending upon whether expression of a particular gene or miRNA was significantly up- or down-regulated by RA-differentiation and further was up- or down-regulated by Hdhex4/5/Hdhex4/5 null mutation, respectively (Fig. 6A-B, Suppl. File S1). Interestingly, miRNAs were evenly distributed across these classes (Suppl. File S1), while the mRNA expression changes were largely in the up_down class (Figure 6C). With these RA-differentiation-Htt-null gene sets we performed pathways analyses to assess membership in: 1) available Gene Ontology biological processes, 2) manually curated lists of genes relevant to organ development found to be abnormal in hypomorphic mice (Materials and Methods, Suppl. Table 2 and Suppl. File S1), and, 3) lists of genes expressed in specific brain cell types (210 neuron genes, 144 astrocyte genes, 61 oligodendrocyte genes [23], specifically to assess the hypothesis that Htt-null mutation alters the neuron-glia developmental switch [24]. The results of Gene Ontology analysis showed significant enrichment, mostly for ‘up_down but also the ‘up_up gene sets, highly associated with the terms ‘Organ Morphogenesis’, ‘Nervous System Development’, ‘Skeletal System Morphogenesis’, ‘Blood Vessel Morphogenesis’, ‘Ear Development’ and “Skin Development” (Fig. 6D, Suppl. File S2). Analysis with our manually-curated lists of genes involved in body weight (bodyweight), skin differentiation (skin), skeleton development (skeleton), middle-ear development (middle-ear) and hematopoiesis revealed significant enrichment scores for all of these hypomorph phenotype-related pathways, with the majority of the genes belonging to the ‘up-down’ class being up-regulated by RA-differentiation and down-regulated by Htt-null mutation (Fig. 7A-C). Furthermore, this class also contained most of the differentially regulated astrocyte and oligodendrocyte specific genes, whereas most of the neuron-specific genes were in the up-up class, up-regulated by RA differentiation and further up-regulated by Htt-null mutation (Fig. Suppl. 4, Suppl. File S1). This differential impact on genes expressed in neurons and in the major macroglial cell types strongly implied an effect on neurogenesis that, by analogy, implied effects of Htt loss of function mutation on progenitor stem cells.
Integrative regulatory network analysis
Differentiation reflects the coordinated action of epigenetic regulators, transcription factors and post-transcriptional events among which those regulated by miRNAs on gene expression. Therefore, to determine whether Htt-inactivation may influence regulatory loops, thereby affecting the expression of downstream targets that may be involved in hypomophic mutation-associated mouse developmental phenotypes, we performed analysis with the above mentioned RA-differentiation-Htt-null gene sets to evaluate: 1) enrichment of targets of chromatin regulators in the ChEA experimental ChIP-seq database (https://www.ncbi.nlm.nih.gov/pubmed/20709693) and 2) potential miRNA regulators, by virtue of having a differentiation-Htt null expression pattern (down-down and down-up) that was opposite to that of their mRNA target genes (up-up and up-down) [25,26]. Consistent with previous reports, the results of the chromatin regulator analysis revealed a strong enrichment for targets of polycomb regulators (Fig. 8A) among the differentiation genes also regulated by Htt inactivating mutation, especially members of the PRC2 complex (Suppl. File S3). The second analysis revealed a set of 22 RA-differentiation-Htt-null responsive miRNAs whose known target genes’ expression pattern fulfilled our criteria (Suppl. File S1) (Fig. 8B-C, Suppl. File S4), with let-7b-5p and miR-329-3p the most significant for the down-down and down-up expression pattern classes, respectively. A functional enrichment analysis of the mRNAs regulated by our list of 22 miRNAs whose expression pattern fit our criteria (opposite to that of the binding miRNAs) confirmed enrichment of ‘up-up’ miRNA target genes in processes involved in ‘nervous system development, ‘axonogenesis’ and ‘synapse’, while the targets of ‘up-down’ miRNAs were enriched for terms related to ‘skeletal system development’, ‘organ morphogenesis’, ‘vasculature development’ ‘ear development’ and ‘skin development’ (Fig. 8B-C), implying a role for these regulators and target genes in the altered developmental phenotypes observed in mice with expression of the HD gene orthologue below a single functional allele.
Discussion
HD is a neurodegenerative disorder for which there is as yet no disease-modifying intervention. It is hoped that approaches that decrease expression of the mutant gene, or both the mutant and normal gene, will provide benefit as suggested by observations from pre-clinical studies in model systems [27,28]. However, there is a need to better understand HTT function, in order to forecast potential effects of HTT silencing and lowering strategies and also to understand precisely how HTT CAG expansion mutation harms vulnerable cell populations that drive the features of the disease.
Our assessment of dosage effects, in which different hypomorphic mutations of the mouse HD gene orthologue attempt to rescue a null allele, has revealed critical involvement of Htt throughout development, from conception. First, the severe HdhneoQ111 mutation, similarly to what previously shown for the homozygote null alleles, was characterize by a huntingtin level insufficient to properly initiate the organogenesis phase causing embryos to fail post-gastrulation, before head-folds become evident. Second, the intermediate HdhneoQ50 mutation, did make sufficient huntingtin to overcome this first block, but this level was insufficient later, such that embryos failed after E10.5-12.5 [9], with survivors revealing several needful organ systems: brain, skeleton, middle and external ear, and skin barrier acquisition and fetal liver erythropoiesis. Third, the dose of huntingtin from the milder HdhneoQ20 mutation was also wanting, but was sufficient for preventing most deficits observed with more severe mutations. Exceptions included the low penetrant external ear and sternum deficits and the enlarged ventricles noted previously [9,17] along with developmental delay after E14.5, although the small mice are born and can be viable (and are fertile) with assistance in the form of removal of normal littermates (with a wild-type allele) or hand-rearing [17]. These findings, taken together with previous observations that the cognate (neo-out) HdhQ20, Hdh50, HdhQ111 CAG repeat knock-in alleles (appropriately expressing endogenous mutant huntingtin with 20-, 50- and 111-glutamines) fully rescue the Hdhex4/5 null [9,29] clearly demonstrate that the equivalent of a single functional HD orthologue allele (regardless of CAG repeat size) is needed for proper development. Our findings complement and enrich previous observations of aberrant external ear, vasculature, brain development and neurological deficits in mice that are compound heterozygote for neo-in hypomorphic alleles [17] and are also consistent with the report of a family segregating two HTT inactivating mutations, where compound heterozygotes exhibit decreased survival, global developmental delay and neurological deficits [15,16]. Indeed, our results strongly suggest that one or both of the HTT [4469+1G>A] and [8156T>A] mutations segregating in this family is functionally hypomorphic rather than completely inactivating.
Our molecular analysis of wild-type ESC and ESC with the Hdhex4/5 null mutation, in a RA-morphogen-driven-paradigm [19-21], which we confirmed elicits broad effects on developmental genes, has nominated categories of regulators sensitive to huntingtin dosage that seem likely to play key roles in Htt function during organ system development. One important category is polycomb group proteins: PRC2 (Suz12, Eed, Jarid2 and Mtf2), implicated previously [12,13], as well as PRC1 (Bmi1, Ring1b), which are especially relevant for neurogenesis, hematopoiesis, skeleton and middle-ear formation [30-34]. The other major category is a group of miRNAs, among which Let-7b-5p and miR-129-3p, with known involvement in nervous system development, as well as skeletal, ear and skin development [35-37]. Precisely which processes these potential regulators orchestrate and how they may be disrupted by the lack of functional Htt in different organ systems remains to be investigated. However, several observations in severely hypomorphic embryos at different ages strongly suggest an impact at the level of progenitor stem cells implicating effectors of these regulators that are critical to cell adhesion or other aspects of cell-migration and can thereby influence lineage fate: 1) the pharyngeal arches were formed but the middle-ear fated Gsc-positive cells were located in the first arch, rather than in the second and third arches, implying delayed migration of these progenitor cells [38,39]; 2) ventral skin barrier formation was not permanently disrupted, only delayed at E18.5, suggesting aberrantly slow migration of epidermal precursors with delayed formation of a cornified layer and acquisition of barrier [40,41]; and 3) deficits in fetal liver hematopoiesis were transient, resolved by E18.5, consistent with delayed migration from the blood islands in the yolk sac, the site of embryonic hematopoiesis [42]. In addition, as reported previously [9,12,17,24,43], neurogenesis and brain development is abnormal in the absence of sufficient Htt. The results of our gene expression analysis support disruption of a process at the level of neural progenitor cells that influences the proportion of daughter neuronal and macroglial cell types, a process that normally involves the acquisition by radial glia of adherens junctions important for cell migration and proper specification, initially of neurons and later of astrocytes and oligodendrocytes [44].
It is evident that the inherited CAG repeat expansion mutation does not recapitulate the blatant effects produced by inheritance of two HTT inactivating mutations. HD mutation carriers, even homozygotes, are overtly indistinguishable from those who do not carry the mutation, until subtle changes presage the emergence of signs of the neurodegenerative disorder [45,46]. By contrast, though one active allele is sufficient [9], levels below a single allele’s worth of HTT function produce developmental consequences that our findings show can vary dramatically in scope and severity with dosage. Studies of decreased Htt dosage later in development in adults, rather than at conception, are more equivocal. Some reports show harmful consequences from neuron-specific knock-out, for example in [47], while a different strategy did not elicit harmful effects [48]. Our findings suggest that, if the harmful effects of the HD mutation are due to some opportunity for mischief that is provided by normal HTT function, then studies of the polycomb group protein genes and Let-7b-5p and miR-129-3p miRNAs as regulators of cellular adhesion may illuminate the biology that is disrupted by the HD mutation and, thereby, provide specific assays with which to evaluate therapeutics that aim to decrease or silence expression of the gene in individuals with the HD mutation.
Methods
Mice
All the mouse experiments were conducted in accordance with an IACUC approved protocol, through the MGH Subcommittee on Animal Research. The Hdhex4/5/ex4/5 and HdhneoQ20/Hdh+, HdhneoQ50/Hdh+, HdhneoQ111/Hdh+ alleles as well as the genotyping protocols have been described previously [6,7,9].
For skeletal preparations, embryos and newborns mice were eviscerated, fixed in ethanol, incubated in acetone and finally stained using alizarin red and alcian blue [49,50]. The in vivo epidermal barrier assay was performed on E18.5 embryos using a dye exclusion assay as reported[51]. Briefly, embryos were sacrificed, immersed in methanol solutions at different concentrations, then stained in 0.1% toluidine blue and washed several times with PBS again to remove the excess dye.
Immunohistological analyses
Staining with hematoxylin and eosin (H&E) and immunostaining was performed by fixing frozen brain or skin sections [cryostat] (LEICA CM3050S), sectioned at 6 μm, in 4% Paraformaldehyde, incubating with primary antibodies at 4 °C overnight, while secondary antibody were used following vendor instructions. Anti-Proliferating Cell Nuclear Antigen (PCNA) was from Santa Cruz Biotechnology, Inc. Slides were rinsed and mounted using Vectashield with 4,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories) for visualization of nuclei.
The DeadEnd™ Colorimetric TUNEL (TdT-mediated dUTP Nick-End Labeling) Assay (Promega) was used to detect apoptotic cells in situ accordingly to the vendor’s instructions.
Protein extraction and immunoblot analysis
Total protein lysates were prepared by pulverizing skin tissue and extracted using RIPA (Boston Bio-Products) lysis buffer with protease inhibitor mixture (Roche). After Bradford protein assay (BIORAD), fifty micrograms subjected to 10% SDS–PAGE, transferred to nitrocellulose membranes (Schleicher and Schuell) and incubated with the following primary antibodies: anti-filaggrin (Covance), anti-loricrin (Covance) and anti-alpha tubulin antibody (Santa Cruz Biotechnology, Inc).
Mouse and rabbit secondary antibodies were used and the specific protein bands were detected using the ECL Plus kit (Pierce) and autoradiographic film (Hyperfilm ECL; Amersham Bioscience).
Whole mount in situ hybridization
After dissection in PBS, embryos were fixed overnight in 4% paraformaldehyde at 4°C. RNA in situ hybridizations were performed as described previously (cold spring harbor lab method). Briefly, E11-11.5 mouse embryos were fixed overnight in 4% formaldehyde in PBS, then dehydrated in methanol and stored at −20°C until use. The digoxigenin-labelled RNA probes were used at 0.5 ug/ml. Alkaline phosphatase-conjugated anti-digoxigenin Fab fragments were used at 1:5000. Color reactions were carried out over time periods ranging from 2 hours to overnight. Embryos were mounted in 80% glycerol before being photographed. Three to four embryos were evaluated for each marker.
Flow cytometry
Fetal liver and fetal thymus were harvested from E14.5 and E18.5 Hdh+/Hdh+ and Hdhneo- inQ50/Hdhex4/5 embryos and single cell suspensions were obtained by mechanical disruption. Fetal liver cells were stained with Ter119, CD71, Mac-1, Gr-1 B220 and CD19, and thymocytes were stained with CD4 and CD8a cell surface markers, respectively. Flow cytometry was performed on a two-laser FACSCanto (BD Biosciences). FCS files were analyzed by FlowJo software (Tree Star). Antibodies used were CD4 (L3T4), CD8a (53-6.7), CD19 (1D3), Mac-1 (M1/70), B220 (RA3-6B2), Gr-1 (RB6-8C5), Ter119, CD71 (C2) (BD Pharmingen, eBioscience).
Colony forming cell (CFC) assays
Methylcellulose colony forming cell (CFC) assays was performed on fetal liver cells of E14.5 and E18.5 Hdh+/Hdh+ and Hdhneo-inQ50/Hdhex4/5 mice as previously described (Yoshida et al G&D 2008). Briefly, 5 × 104 cells were cultured with Methocult M3434 (Stem Cell Technologies) supplemented with hTPO (50 ng/ml) in 35-mm culture dishes (NUNC 174926) in duplicates. Colonies were scored from day 2 to day 17 according to the technical manual and previously described criteria [52], and were confirmed by May-Giemsa staining (Harleco) of cytospins (400 rpm, 5 min) of individual colonies. Cytokines were purchased from R&D Systems.
Cell culture
Wild-type and huntingtin null Hdhex4/5/ex4/5 mouse embryonic stem cell lines were described previously [9,12,53]. Differentiation was performed essentially as described in Bibel et al., 2007 [54]. In brief, ESCs were deprived of feeder cells for 4 passages, then 3 x (106) cells were used for formation of embryoid bodies (EBs). EBs were grown in non-adherent bacterial dishes (Greiner, Germany) for 8 days. Retinoic acid (5 uM, SIGMA) was added from day 4 to day 8 and medium was changed every other day. Subsequently, EBs were dissociated by trypsin digestion and plated on Poly-Ornithine (SIGMA) and laminin (SIGMA) coated plates to obtain RA-differentiated cells (RA-DIFF). Two hours after plating, RA-differentiated cells were collected for different analyses.
RNA Isolation and RNAseq, miRNAseq library preparation and sequencing
RNA was extracted from cell lines by using TRIzol reagent (Life technologies), following manufacturer’s instructions. All RNA samples were subjected to DNAse I treatment (Ambion). RNA sequencing was performed following the protocol described by the Broad Institute (Cambridge, MA)[55,56]. Briefly, poly-A-plus mRNA, was retro-transcribe to cDNA using a strand specific dUTP method, random hexamers and amplified by PCR using bar-coded DNA adaptors from Illumina. HiSeq2000 platform and 50bp pair-end (PE) was used for sequencing obtaining 50-75M reads/library.
Small RNA library preparation (mainly miRNAs) was obtained using the Illumina TruSeq Small RNA protocol (Illumina) according to manufacturer instructions and libraries were sequenced using single-end, 50bp reads on HiSeq2000 platform, obtaining 45-90M reads/library.
RNA-Sequencing and miRNA-Sequencing data analysis
For RNA-Seq, 50 bp paired-end reads were were aligned to the mouse genome (GRCm38.p4) with Tophat (version 2.0.14, default settings), using the Gencode M6 transcript annotation as transcriptome guide. All programs were used with default settings unless otherwise specified. Mapped reads (Suppl. Table 1) were assembled into transcripts guided by reference annotation (Gencode M6) with Cufflinks (version 2.2.1). Expression levels were quantified by HTSeq (Version 0.6.1, --mode intersection-nonempty) using Gencode M6 annotation. Normalization was performed using the TMM method implemented in edgeR. Differential expression was calculated using edgeR (dispersion 0.1, pval < 0.05, log2 fc >0.75, log2CPM > 0).
For miRNA-Seq, 50 bp single-end reads were trimmed against Illumina’s TruSeq Small RNA 3’ (TGGAATTCTCGGGTGCCAAGG) and 5’ (GUUCAGAGUUCUACAGUCCGACGAUC) adapters using cutadapt(v. 1.3) with parameters –e 0.1 –O 5 –m 15[57]. Next, trimmed sequence reads with length of 16-25bp were mapped to 1915 known mature miRNA sequences from miRBase (release 21) database [58], using BWA aln (v. 0.7.5a-r418) with parameter –n 1 [59]. Uniquely mapped reads (Suppl. Table 1) were identified by selecting alignments with non-zero mapping quality and “XT:A:U” tag using samtools (v 0.1.18)[60]. Normalization was performed using the TMM method (BWA and TMM were suggested previously[61]). Differential expression was calculated using edgeR (dispersion 0.1, pval < 0.05, log2 fc >0.75, log2CPM > 0)[62].
Computational analyses
Functional annotation of gene lists and enrichment analysis with Gene Ontology terms and KEGG or REACTOME pathways were performed with the clusterProfiler Bioconductor package. Neuron, astrocyte and oligodendrocyte markers were downloaded from Cahoy et al., 2008 [23], while lists of genes implicated in the different phenotypes were manually created based on multiple papers/web-resources [39, 63-68] (Suppl. Table 2). The significance of custom enrichments was measured with one-sided Fisher exact test.
Collections of murine miRNA targets were downloaded from miRTarBase (Release 6.1) [69]. For each miRNA – target mRNA couple, Pearson’s correlation values were calculated from standard normalized expression values. For each miRNA, the negative shift in the distribution of target correlation values was measured with one-sample one-sided Wilcoxon test (significance threshold: P < 0.05).
Acknowledgments
We thank Jolene Guide for technical support with mouse breeding and Ilaria Sanvido for assisting in the ES cell culture experiments. This work was supported by the National Institutes of Health (USA) (grant NS32765), the CHDI Foundation Inc. (MEM, JFG), an Anonymous Donor (MEM) and the University of Trento (MB). MB was a recipient of a Marie Skłodowska-Curie reintegration fellowship (the European Union’s Horizon 2020 research and innovation programme) under the grant agreement No. 706567.