HACE1 deficiency leads to structural and functional neurodevelopmental defects

Identification of novel HACE1 mutations in SPPRS patients

Family 1 originating from Saudi Arabia had 1 affected female individual, patient 9, born to healthy, consanguineous parents (figure e-1A, links.lww.com/NXG/A148). The patient presented with similar clinical symptoms as previously reported cases of SPPRS, including severe intellectual disability, developmental delay, and inability to sit or speak by the age of 5 years.⁶ Cranial MRI revealed microcephaly and brachycephaly, as well as hypoplastic corpus callosum and likely brainstem abnormality, small sella with ectopic neurohypophysis, and mild ventriculomegaly (figure e-1B). Additional clinical findings include mild facial dysmorphia, skeletal abnormalities, and ulnar deviation of the wrists and small feet (figure e-1C). In an unrelated consanguineous family 2 originating from Turkey, born to a healthy father and mother with Hashimoto disease are one female (patient 10) and one male (patient 11), 4 and 6 months of age at initial admission (Figure e-1D). During pregnancy, the mother was euthyroid and both children were carried to term. Patients had variable clinical symptoms in comparison with patient 9 and those previously reported, including hypotonia, small feet, enlarged head circumference, inverted and wide spaced nipples, facial dysmorphic features, and strabismus (figures e-1, E and F). Considering the pronounced hypotonia in all 3 patients, ataxia was difficult to determine. For detailed clinical symptoms of all patients, please refer to table.

In these 3 patients, mutations in HACE1 were identified by exome sequencing and verified by Sanger sequencing: patient 9 carries a homozygous c.625C>T; p.Q209* HACE1 mutation in exon 8, NM_020771.3:c.625C>T, p.Q209* (chr6:104 797 018 on the hg38 build); patients 10 and 11 carry a homozygous c.994C>T; p.R332* HACE1 mutation in exon 11, NM_020771.3:c.994C>T, p.R332* (chr6:104 791 584) (figure 1A). Both are nonsense mutations predicted to result in a truncated HACE1 protein product with no catalytic activity. Parents in both families were determined to be heterozygous carriers. Figure 1A also indicates all previously reported HACE1 mutations in SPPRS patients.^5,6 Thus, we have identified three new cases with biallelic HACE1 mutations associated with marked and diverse neurologic as well as non-neurologic abnormalities, diagnosed to be SPPRS.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Histologic analysis of brain regions in neonatal and young Hace1 knock-out (KO) mice

(A) Schematic representation of HACE1 protein and locations of mutations of previously published (in gray) and novel mutations reported here (red arrows, black letters). (B) H&E-stained sections of wild-type (WT) and Hace1 KO mice at E18.5 reveal no major structural abnormality across the regions studied, with slight ventriculomegaly in the Hace1 KO brain as compared with controls, left panel (yellow arrows). (C) H&E-stained coronal sections of the newborn brain of WT and Hace1 KO mice at P0.5 reveal ventricular dilation in the Hace1 KO brain (yellow arrows) and attenuation of the corpus callosum (black arrows). (D) H&E-stained sections of WT and Hace1 KO mice at P14.5. Cortical lamination, hippocampal formation, and the cerebellar vermis are comparable between WT and Hace1 KO in all 3 age groups studied. All brain regions are labeled in the figure and magnifications are as indicated. A = ankyrin domain; H&E = hematoxylin and eosin; HECT = homologous to the E6-AP carboxyl terminus; MID = middle region.

Hace1-mutant mice exhibit brain morphologies similar to SPPRS patients

To determine whether Hace1 KO mice develop any clinical symptoms of SPPRS patients, we first analyzed structural and cellular features of mutant mouse brains using 15.2 T MRI and/or histologic staining at different developmental stages. Histologic examination of mouse brains at embryonic day (E) 18.5, postnatal day (P) 0.5, and P14.5 revealed no evident defects cortical lamination, cerebellar foliation, hippocampal architecture, or cellular aberration and density in any developmental period examined (figure 1, B–D). However, enlarged ventricles and hypoplastic corpus callosum (figure 1, B and C) were noted in Hace1 KO animals as compared with WT littermates, as early as E18.5 and P0.5, respectively.

Western blot analyses of 8-week-old mouse brain lysates revealed HACE1 to be expressed throughout the adult mouse brain, which was absent in the brains of Hace1 KO mice (figure 2A). Total volume and morphology of the Hace1 KO adult mouse brain as compared with WT littermates measured by MRI and/or histologic examination revealed no whole-brain volumetric or cellularity differences as compared with WT littermates (figure 2, B and C). Review of histologic sections also revealed no abnormality in cortical laminar organization and cellular density in the adult brain (figure 2D). Detailed analysis of different brain regions via MRI also showed no significant differences in the volumes of the cortex, colliculi, thalamus, and putamen, comparing adult WT and Hace1 KO brains, with a small decrease in the volume of the olfactory bulb (data not shown). The adult cerebellar volume and foliation as well as cortical laminar organization were comparable between Hace1 KOs and their control WT littermates (figure 2, E and F). Hippocampal volume was significantly decreased as measured by MRI (figure 2G); however, there were no evident aberrations in cellularity and architecture on microscopic examination of hippocampi in adult animals (figure 2H).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Morphology of the Hace1 KO mouse brain

(A) Western blot analysis depicting HACE1 abundance in the indicated 8–12-week-old male mouse brain regions of wild-type (WT) and Hace1 knock-out (KO) mice. (B) Pseudocolored 3D reconstruction of lateral MRI images of representative whole mouse brains, and quantification (mean values ± SEM) showing no difference in total brain volumes in 8–12-week-old male WT and Hace1 KO littermates. Orange—cerebellum, yellow—hippocampus, red—corpus callosum, light blue—thalamus, purple—colliculi, turquoise—olfactory bulbs, and dark blue—putamen; WT n = 12 and KO n = 9; p value as indicated in the bar graph; unpaired Student's t test. (C) LFB-CV–stained coronal sections of whole WT and Hace1 KO adult mouse brains. Ventriculomegaly (red arrows) is evident in the KO brain as compared with the WT littermate. (D) LFB-CV–stained coronal sections of the adult cortex show no defects in cellularity or lamination (laminae I to VI) in Hace1 KO mice. (E) Pseudocolored (orange) 3D reconstructed MRI images of representative lateral views of WT and Hace1 KO littermate cerebelli showing no significant volumetric differences in Hace1 KO vs WT control littermates. Right, quantification of volumetric differences; mean values ± SEM; WT n = 12; KO n = 9; p value as indicated in the figure; unpaired Student's t test. (F) Selected LFB-CV–stained sagittal sections of the cerebellum showing no apparent differences between WT and Hace1 KO brains. (G) Pseudocolored (yellow) 3D reconstruction of MRI images of representative lateral view of hippocampi of WT and Hace1 KO mice shows a significant reduction in hippocampal volume in Hace1 KO brains; quantification is shown in the right panel; mean values ± SEM; WT n = 12 and KO n = 9; *p ≤ 0.05; unpaired Student's t test. (H) LFB-CV–stained coronal sections of the hippocampus show no notable morphological difference in cellularity between WT and Hace1 KO brains. Scale bars as labeled in the images. CA = cornu ammonis; CC = corpus callosum; cereb = cerebellum; Ctx = cortex; DG = dentate gyrus; hipp = hippocampus; hypo = hypothalamus; LFB-CV = luxol fast blue - cresyl violet; LV = lateral ventricle; stria = striatum; WM = white matter.

Of interest, volumetric MRI reconstructions revealed a significant reduction in whole-brain white matter (figure e-2, A–D, links.lww.com/NXG/A149) in Hace1 KO adult mice as compared with their WT littermates, substantiated by myelin basic protein staining and volumetric MRI analysis (figure e-2, B–D), similar to findings in reported SPPRS patients^5,–,7 and patient 9 described here (figure 1C and table). Also, noted was a reduction of the anterior commissure in Hace1 KO mice as compared with WT littermates (figure e-2C). In addition, in line with findings in SPPRS patients, as well as adult and developing mouse brains, we observed enlarged ventricles in Hace1 KO mice as compared with control littermates (figure 2C). These data show that Hace1 KO mice phenocopy key structural features associated with SPPRS patients.

HACE1-deficient mice exhibit locomotion and associative learning disabilities

To assess what, if any, behavioral defects can be detected in Hace1 KO mice, we subjected cohorts to an extensive battery of behavioral assays. The most notable clinical features of SPPRS are intellectual disability and ataxia^5,–,7; we therefore focused on hippocampal- and cerebellar-dependent tasks, principal brain regions regulating those behaviors in mice. Hace1 KO mice generated in our group previously are generally healthy, viable, and fertile.² First, we found no significant differences in anxiety levels in Hace1 WT and KO littermates analyzed by the OFT and the elevated plus maze (figure e-3, A and B, links.lww.com/NXG/A150); however, we noted a reduction in the distance traveled during the testing (figure 3A). These data indicated that while our mice exhibited normal anxiety levels, there may be a locomotion defect. Therefore, we tested Hace1 KO mice for any locomotion and gait defects. WT and Hace1 KO mice were tested in the accelerating rotarod task, a standard assay to determine locomotion and gait defects.¹⁶ While Hace1 KO mice and control littermates had similar latencies to fall off the rotating rod in the first trial, Hace1 KO mice exhibited shorter latencies to fall on all subsequent trials (figure 3B). We further analyzed the mice on a ladder rung walking task, which enables the observer to analyze coordinated front and hind paw placement on each individual rung of the ladder.¹⁷ While Hace1 KO animals were capable of performing the task without falling off the apparatus and showed similar percentages of correct placement over all limbs analyzed, percent of major missteps made were significantly higher as compared with WT littermates. Both groups of animals had similar placement and correction scores, while WT animals had a higher percentage of partial placement on rungs (figure 3C). Thus, Hace1 KO mice exhibit locomotion and gait deficits.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Behavioral analysis of Hace1 KO mice reveals deficits in specific locomotion, learning, and memory tasks

(A) Open-field maze (OFM) and elevated plus maze (EPM) showed slight and significant reductions in the distances Hace1 KO mice had traveled as compared with their WT littermates, respectively; data are shown as mean values ± SEM; WT n = 10 and KO n = 10; *p ≤ 0.05 or as indicated in the bar graphs; unpaired Student's t test. (B) Hace1 KO mice show significantly shorter latencies to fall off the accelerating rotarod; mean values ± SEM; WT n = 19 and KO n = 21; *p ≤ 0.05; 2-way ANOVA with Sidak's multiple comparison test. (C) Ladder rung walking task illustrates a significant increase in the percentage of total misses, deep slips, and/or slight slips (0–2 score) in Hace1 KO mice compared with WT littermate controls; mean values ± SEM; WT n = 12 and KO n = 9; *p ≤ 0.05; unpaired Student's t test. No significant differences were noted in the 3–4 score (replacement and correction), 5 score (partial placement), or 6 score (correct placement). Bar graphs depict mean values of all paws ± SEM; p values are indicated in figures; WT n = 12 and KO n = 9; unpaired Student's t test. (D) Significant deficiency in long-term contextual memory as measured by auditory fear conditioning was detected in Hace1 KO mice as compared with WT littermates (left panel); mean values ± SEM; WT n = 16 and KO n = 13; *p ≤ 0.05; unpaired Student's t test. Cued fear conditioning was not significantly different in Hace1 KO mice as compared with WT littermates; mean values ± SEM; WT n = 16; KO n = 13; p value = 0.21; unpaired Student's t test. (E) Hace1 KOs show significant reduction in acoustic startle inhibition at 110 and 120 dB as compared with WT littermates. Mean values ± SEM; WT n = 14; KO n = 11; *p ≤ 0.05; ***p ≤ 0.001; 2-way ANOVA with Sidak's multiple comparison test. ANOVA = analysis of variance; KO = knock out; WT = wild type.

The second notable clinical feature of SPPRS patients is intellectual disability (table).^5,–,7 To determine whether learning and memory processes are altered in Hace1 KO animals, we analyzed mutant mice in various hippocampal-dependent behavioral tasks. In the Morris water maze, KO mice were able to swim normally and there were no apparent differences in latencies between WT and Hace1 KO littermates to find the visible platform (not shown), in escape latencies during any of the training days (figure e-3C, links.lww.com/NXG/A150), nor in short- or long-term memory (figure e-3C, right panels). Likewise, there were no differences in T-maze between WT and Hace1 KO animals (figure e-3D). However, subjecting the animals to cued and contextual fear conditioning revealed significantly reduced freezing responses in the contextual aspect of the task in Hace1 KO mice compared with WT littermates (figure 3D).

In addition, Hace1 KO mice exhibited significantly enhanced startle amplitude responses or reduced startle inhibition, in the acoustic startle task at presentation of 110- and 120-dB startle sounds as compared with their WT littermates (figure 3E). These results also suggest Hace1 KO animals do not have any hearing deficiencies. PPI was reduced for all prepulse intensities presented (80, 85, 90, and 95 dB); however, this reduction was not significant when comparing Hace1 KO animals with WT controls (figure e-3E, links.lww.com/NXG/A150). These data indicate that, besides locomotion defects, associative learning is altered in Hace1 KO mice, paralleling behavioral findings in human SPPRS patients.

Hace1 deficiency results in altered hippocampal synaptic transmission and reduced synapse numbers

We and others have previously shown that altered synaptic transmission and plasticity at CA3-CA1 Schaffer collaterals can underlie deficits of contextual fear memory.^22,24 We tested whether synaptic transmission and plasticity of this pathway are affected in Hace1 KO mice. Recordings of field excitatory post synaptic potentials in response to CA3-CA1 synapse stimulation (schematic of the set-up in figure 4A) revealed an upregulation of excitatory synaptic transmission in hippocampal slices from Hace1 KO mice (figure 4, A and B). Further analysis showed that the increase in synaptic transmission at the CA3-CA1 synaptic pathway was most likely due to a postsynaptic effect of the Hace1 deletion because neither the axonal fiber volley (figure 4C) nor the paired-pulse ratio, an indirect measure of the presynaptic release probability,²⁵ was affected (interstimulus interval 50 ms, paired-pulse ratio: WT 1.70 ± 0.066, n = 19, Hace1 KO 1.66 ± 0.036, n = 20, p = 0.64, 2-population t test, not shown). We then probed long-term potentiation (LTP), a cellular correlate of memory formation,²⁶ of this synaptic pathway and found that LTP is significantly reduced in acute hippocampal slices from Hace1 KO mice compared with WT littermates (figure 4, D and E). Of note, nerve conduction velocity was not affected in Hace1 KO recordings as compared with controls (data not shown), in line with the report that nerve conduction velocity is unaffected in SPPRS patients.⁶

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 Altered long-term potentiation and reduced hippocampal synapse numbers in Hace1 KO mice as compared with WT controls

(A) Upper left panel, schematic fEPSP recording configuration from acute hippocampal slices: stimulation (red bolt) of CA3-CA1 Schaffer collaterals (green), the fEPSP extracellular recording electrode in the CA1 region (white cone, ep = extracellular pipette). Synaptic transmission at CA3-CA1 hippocampal synapses is increased in Hace1 KO mice compared with WT littermates. fEPSP slopes were normalized to axonal fiber volleys to account for variations in axonal stimulation efficacy between slice preparations and animals. Analysis was performed over a range of stimulus intensities. Average fEPSP slopes were significantly higher in Hace1 KO slices; mean values ± SEM; WT n = 19 and KO n = 20; *p ≤ 0.05; 2-population Student's t test. (B) Representative traces of fEPSPs in response to CA3-CA1 synapse stimulation obtained from WT (black trace) and Hace1 KO (red trace) mice. Field EPSP (black arrow) recordings were normalized to fiber volleys (gray arrow) to account for varying stimulation efficiencies across experiments and preparations. Time scale of recording is indicated. (C) No significant difference was found between fiber volley amplitudes (mean values ± SEM) between the indicated genotypes; WT n = 19 and KO n = 20. (D) Long-term potentiation (LTP) of synaptic transmission was induced by HFS after a 10-minute baseline recording. fEPSP slopes were normalized to their baseline levels and plotted over time. LTP was reduced in Hace1 KO animals; mean values ± SEM; WT n = 19 and KO n = 18; **p ≤ 0.01; 2-population Student's t test. (E) Representative fEPSP traces illustrating LTP experiments displayed in panel (D). Solid lines represent baseline fEPSPs, and dashed lines represent fEPSPs 30 minutes after LTP induction (Hace1 WT: black, upper panel; Hace1 KO: red, lower panel). (F) Representative 3D composite of z-stacked ×63 magnified (zoom factor of 2) confocal images of hippocampal CA1 pyramidal cell distal dendrites of Hace1^+/+;Thy1-GFP and Hace1^−/−;Thy1-GFP mice. Uncolored dendritic segments are on the left, synaptic spines are pseudocolored in magenta on the right panels; scale bar is 2.5 μm. Right bar graphs show quantifications of spine numbers; mean values ± SEM; WT n = 3 and KO n = 3; **p ≤ 0.01; unpaired Student's t test. fEPSPs = field excitatory post synaptic potentials; KO = knock out; WT = wild type.

Because we observed hippocampal-dependent contextual fear learning deficiency and dysfunctions in electrophysiologic properties of pyramidal cells within the CA1 region of the hippocampus, we analyzed morphological properties of pyramidal neurons in this brain region. To this end, we crossed Hace1 KO mice with Thy1-GFP-M mice to drive neuronal GFP expression, which allowed us to visualize neuronal morphology, including spines along dendrites of CA1 pyramidal cells.²⁷ Quantification of selected dendritic segments indeed revealed a significant reduction in spine numbers in Hace1 KO neurons as compared with hippocampal neurons from WT littermates (figure 4F). Whether such reduction in the synapse number is also present throughout the brain and might explain defective locomotion in Hace1 KO mice needs to be determined. These data indicate that, as a correlate to hippocampus-dependent impaired associate learning, HACE1 deficiency leads to a deficit of synaptic plasticity and a marked reduction of synapses in the hippocampus.

Elevated levels of active Rac1 in Hace1 KO mouse brains and SPPRS patient–derived fibroblasts

To determine molecular alterations downstream of HACE1 deficiency, we focused on Ras-related C3 botulinum toxin substrate 1, RAC1, a well-characterized substrate for HACE1-mediated proteasomal degradation.^8,9 Indeed, Western blot analysis showed that RAC1 protein abundance was increased in Hace1 KO brains when compared with WT littermates (figure 5A). In addition, Cyclin D1 abundance and ROS levels, as downstream read-outs for RAC1 activity,¹² were likewise increased in Hace1 KO brain regions as compared with controls (figure 5, A and B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5 Hace1 KO mouse brains and SPPRS patient–derived fibroblasts have elevated Rac1 and ROS levels

(A) Western blot analysis of the cortex (ctx.), hippocampus (hipp.) and cerebellum (cereb.) from WT and Hace1 KO mice revealed a strong upregulation of total RAC1 abundance (lower band, arrow) and Cyclin D1 protein in the absence of HACE1, in all indicated brain regions tested. Actin was used as a loading control. (B) ROS (detected by DHE staining) is elevated in the hippocampus of Hace1 KO mice as compared with WT littermates; ×20 magnification. (C) Fibroblasts harvested from SPPRS patients (P6, P7, and P10) have no detectable HACE1 protein product as compared with normal donor fibroblasts (ND) as revealed by Western blotting using an antibody recognizing the HACE1 C-terminus; however, patient fibroblasts express elevated total abundance of RAC1 and Cyclin D1 protein product. GAPDH was used as a loading control. (D) SPPRS patient–derived fibroblasts exhibit elevated abundance of active RAC1 in comparison with normal donor fibroblasts, as measured by the Active RAC1 Pull-Down and Detection Kit. (E) Left panel, FACS analysis of SPPRS patient–derived fibroblasts incubated with CellROX deep red reagent revealed significantly more ROS in patient cells as compared with normal donor (ND) fibroblasts. Right panel, quantification of ROS production in the indicated patient and control fibroblasts; mean values ± SD; n = 7 for ND, P6, and P7 fibroblasts, and n = 3 biological replicates for patient 10 fibroblasts; *p ≤ 0.05; unpaired Student's t test. (F and G) Confluent SPPRS patient P6, P7, P10 and ND-derived fibroblasts were subjected to scratch assays, and migration was monitored. Panel (F) shows a quantification of fibroblast migration rates; mean values ± SD; n = 3; **p ≤ 0.01 and ***p ≤ 0.001 comparing patient fibroblasts with the ND controls; unpaired Student's t test. Representative images are shown in (G). KO = knock out; ROS = reactive oxygen species; SPPRS = spastic paraplegia and psychomotor retardation with or without seizures; WT = wild type.

The role of RAC1 has been extensively studied in the context of mouse brain development.^10,11 Of interest, perturbation of RAC1 levels, either overexpression or downregulation, results in decrease in dendritic spines in the mouse brain, similar to our Hace1 KO mice.^10,28 To determine whether RAC1 is also changed in HACE1-mutant SPPRS patients, we performed Western blotting analysis of SPPRS patient–derived fibroblasts, isolated from patients 6, 7, and 10 (table). Of note, in these patients, the mutations result in undetectable HACE1 protein expression (figure 5C), resembling our Hace1 KO mice. In all patients' fibroblasts, total abundance of RAC1 and Cyclin D1 was increased as compared with fibroblasts from HD controls (figure 5C). It is important that increased abundance of active GTP-bound RAC1 was detected in all 3 patient fibroblasts (figure 5D). As a functional consequence of increased RAC1 activity, ROS upregulation was also evident in SPPRS patient–derived fibroblasts (figure 5E). In addition, in a scratch assay, patient fibroblasts migrated significantly faster, as compared with control fibroblasts (figure 5, F and G), confirming enhanced RAC1 activity.^29,–,32 These data show that loss of HACE1 expression results in increased levels of active RAC1 in both Hace1 KO mouse brains and HACE1-mutant SPPRS patient–derived fibroblasts.