Patients with Polr3-related leukodystrophy express mutant forms of Pol III in all their cells. We sought to generate mice with a similarly broad cellular distribution of a Pol III enzyme bearing the Polr3a W671R/G672E mutation. To circumvent the embryonic lethality of the mutation, we employed a ubiquitously expressed tamoxifen-inducible CAGGCre-ER transgene to knockin the Polr3a mutant allele postnatally in diverse tissues (Hayashi and McMahon, 2002). Five tamoxifen injections were administered every other day starting at postnatal day 28 (P28, Figure 1A and B). Recombination frequencies determined by flow cytometry or fluorescence microscopy using a dual tdTomato-EGFP reporter were ~60% in liver, ∼70% in cerebrum, ∼80% in cerebellum and kidney and ~90% in heart (Figure 1—figure supplement 1). Tamoxifen-induced knockin mice (denoted as Polr3a-tamKI mice) have significantly lower body weights than tamoxifen-injected WT controls. Body weight differences appear in both sexes at P34 and become progressively larger as Polr3a-tamKI mice lose weight while the control mice gain weight over time (Figure 1C). These differences are due in part to the smaller body size of Polr3a-tamKI mice post-injection (Figure 1D), suggesting a growth defect. Notably, a growth defect was reported previously for Polr3a-Olig2KI mice (formerly Polr3a-cKI mice, Merheb et al., 2021), consistent with the reduced stature of many patients with Pol III-related leukodystrophy (Watt et al., 2023). However, in contrast to the current model, Polr3a-Olig2KI mice gain weight, albeit more slowly than controls, from adolescence to adulthood despite their smaller size (Merheb et al., 2021). Thus, the loss of body weight in Polr3a-tamKI mice demonstrates a more severe failure to thrive phenotype, in line with the widespread expression of the Polr3a mutation.

We conducted a broad histopathological analysis of the effect of the Polr3a mutation in adolescent (P42) and adult (P90) mice. Gross morphology was assessed by Hematoxylin and Eosin (H&E) staining and in the CNS, myelin and neurons were examined using Luxol Fast Blue (LFB) and Nissl, respectively. LFB staining of coronal sections of the cerebrum revealed impaired myelin deposition in Polr3a-tamKI cortex, hippocampus and thalamus with qualitatively greater differences in the cortex at P90 (Figure 1E). In contrast, LFB- and Nissl-stained sagittal sections of the cerebellum showed no differences in myelin or granular layer neurons, respectively, and no difference in Purkinje cell density (Figure 1—figure supplement 2A, B). The absence of a myelination defect in the cerebellum was previously noted in Polr3a-Olig2KI mice (Merheb et al., 2021). Non-CNS tissues (>20 examined) were histologically normal except for the pancreas. Polr3a-tamKI mice exhibit pronounced and severe exocrine pancreatic atrophy at P42 and P90, respectively (Figure 1F). This condition is characterized by insufficient production of digestive enzymes (Kleeff et al., 2017) and correlates with the lower body weight of Polr3a-tamKI mice, the marked reduction in subcutaneous and visceral adipose tissue at P90, and the poor health of the mice at this age. In contrast to the paucity of body fat at P90, echoMRI measurements at P42 showed that body fat as a percentage of total body weight was comparable to WT (Figure 1—figure supplement 2C). To assess whether the exocrine pancreatic defect might extend to the endocrine pancreas, glucose homeostasis was examined in adolescent mice. Blood glucose concentrations increased upon feeding, as expected, and were not significantly different between Polr3a-tamKI and WT mice in either the fasted or the refed state (Figure 1—figure supplement 2D). Similarly, a glucose tolerance test showed comparable responses (Figure 1—figure supplement 2E). These data suggest that endocrine pancreatic functions controlling glucose homeostasis are not significantly impaired at P42 by the Polr3a mutation.

A behavioral spectrometer with video tracking and pattern-recognition software was used to quantify ~20 home cage-like behaviors in a longitudinal study of behavior. Testing began at P42 and was continued at weekly intervals (Figure 1B). No differences were observed in anxiety-like behavior, which was assessed using metrics that measure the natural tendency of mice to avoid open spaces (Figure 2—figure supplement 1A–C). However, overall locomotor activity was decreased in Polr3a-tamKI mice as indicated by the reduced track length at P42, P49, P56, and P63 (Figure 2A). Polr3a-tamKI mice also spent less time running and more time standing still at all time points compared to WT mice, consistent with a locomotor deficit (Figure 2B and C). Risk assessment was measured by orientation behavior according to three discrete metrics that can be summed to provide an overall evaluation of this behavior (Brodkin et al., 2014). Individual orientation behaviors were significantly lower in Polr3a-tamKI mice at all time points compared to WT mice, except for Orient-shuffle at P42 where there was a trend towards lower performance in the knockin mice (Figure 2—figure supplement 1D–F). The sum of these behaviors shows that Polr3a-tamKI mice spend less time assessing risk at all time points (Figure 2D). Exploratory behavior was investigated via three rearing motions performed when the mice are standing on their hind legs. Polr3a-tamKI mice spent less time performing these motions at most timepoints and a tendency towards diminished performance at other times (Figure 2—figure supplement 1G–I). When all three rearing metrics were combined, Polr3a-tamKI mice showed lower levels of exploration compared to WT mice at all timepoints although the data did not reach significance at P49 (Figure 2E). Grooming behavior was evaluated by measuring the time each animal spent attending to a specific region of their body (Figure 2F and Figure 2—figure supplement 1J, K). The Polr3a-tamKI mice spent less time grooming their legs and back compared to WT mice starting at P49 and progressing to P63. Head grooming was also reduced at P63. In summary, the findings reveal behavioral phenotypes in Polr3a-tamKI mice at the earliest time point in the study, two weeks after introduction of the mutation, consistent with neurobehavioral deficits (Brodkin et al., 2014). As none of the affected behaviors were apparent in Polr3a-Olig2KI mice (Merheb et al., 2021), the data further demonstrate that the mutation has a more severe impact on animal behavior when its expression is induced postnatally in many cell types than when it is induced during embryonic development only in the Olig2 lineage.

To investigate the molecular changes to Pol III transcript levels caused by the Polr3a mutation and any secondary effects on the Pol II transcriptome, we initially focused on the cerebra of adult mice at P75. Precursor tRNA levels reflect Pol III transcription activity since these nascent RNAs are rapidly processed into mature sized tRNA by the removal of 5’ leader, 3’ trailer and, in some instances, intronic sequences (Berg and Brandl, 2021). The levels of four precursor tRNAs, determined either by northern blotting for intron sequences or by RT-qPCR with precursor-specific primers, were reduced ~threefold in Polr3a-tamKI cerebra (Figure 3A–C). Mature tRNAs and other Pol III transcripts have relatively long half-lives and thus their levels reflect steady state abundance rather than synthesis (Nwagwu and Nana, 1980). Nonetheless, Northern blotting of four mature tRNAs showed that all were lower in Polr3a-tamKI cerebra with two of them reduced to ~60% of the WT (Figure 3A and B).

Despite the defect in precursor tRNA synthesis and the reduction in mature tRNAs, the Pol III transcriptome was not universally affected (Figure 3A–C). Transcripts from genes with a tRNA-like Type II promoter (e.g. 7SL, BC1 and Mvg1, aka Vault RNA) showed a decrease in abundance to between 45% and 60% of the WT level (Figure 3C). Transcripts from genes with a Type III promoter (e.g. U6, Rmrp, Rpph1, and 7SK RNAs), were variably decreased: Rpph1 and 7SK RNAs were decreased to 70% and 56%, respectively, while U6, U6atac and Rmrp RNAs were unaffected. The level of 5S rRNA, the only RNA produced using a Type I promoter, was elevated (Figure 3C).

Mouse models of neurodegeneration with underlying defects in the translation machinery induce an adaptive response to cellular stress known as the integrated stress response (ISR; Abbink et al., 2019; Ishimura et al., 2016; Spaulding et al., 2021; Terrey et al., 2020). Canonical activation of the ISR is initiated by stress-activated protein kinases which phosphorylate eIF2α to globally attenuate bulk mRNA translation initiation, increase translation of uORF-containing mRNAs including transcription factors such as ATF4 and induce a stress response transcription program (Costa-Mattioli and Walter, 2020). We measured the expression of 10 ATF4-regulated ISR target genes to evaluate whether the ISR had been activated in Polr3a-tamKI cerebra. A two to sixfold upregulation was measured for seven of these genes, two others were induced at lower levels and one was unaffected (Figure 3D). These results point to altered translation in Polr3a-tamKI cerebra.

Bulk RNAseq was used to broadly profile the transcriptional changes in Polr3a-tamKI cerebra at P75. Differentially expressed (DE) genes (p adj. <0.05, log2FoldChange >|0.58|) were biased towards upregulation; 83% of DE genes were increased in the Polr3a mutant (Figure 3E, Supplementary file 1). We compared our DE genes with a list of 774 ISR genes that show ATF4-dependent induction following tunicamycin treatment of mouse embryo fibroblasts (Torrence et al., 2021) and found 192 genes upregulated in Polr3a-tamKI cerebra (Figure 3E, Figure 3—figure supplement 1A, Supplementary file 1). Among these were key ISR regulators, such as uORF-containing transcription factors (Atf3, Atf5 and Ddit3, aka Chop), pro-apoptotic targets of Ddit3 (Chac1 and Bbc3), and negative ISR regulators (Ppp1r15a, aka Gadd34, and Trib3) that function to antagonize the response (Figure 3E). Interestingly, expression of the transcription factor ATF6 and its canonical ERAD targets (Pdia4, Pdia6, Edem1, Grp78 aka Hspa5, Grp94 aka Hsp90b1, Xbp1, and Canx) were not affected suggesting that ISR induction in the Polr3a mutant does not involve an ER stress-mediated Unfolded Protein Response (UPR, Supplementary file 1; Hillary and FitzGerald, 2018). This was confirmed by showing that the relative levels of inactive unspliced and active spliced forms of Xbp1, a critical transcription factor for the UPR, were unchanged (Figure 3—figure supplement 1B).

Metascape pathway enrichment of upregulated DE genes identified activation of the immune system and related processes as major functional categories (Figure 3F, Supplementary file 1). These data point to activation of microglia (see below) and the upregulation of signaling pathways that promote inflammation and cell death by multiple mechanisms (Butovsky and Weiner, 2018). For example, pyroptosis is indicated by the upregulation of caspases 1 and 4, the inflammasome sensor NLRP3 and its adapter protein, PYCARD, interleukin-1β and the plasma membrane pore-forming protein, Gasdermin D, among others (Supplementary file 1). The most enriched functional categories among genes down-regulated in Polr3a-tamKI cerebra were cholesterol biosynthesis and electron transport (Supplementary file 1). All mitochondrially encoded transcripts were also down-regulated in the Polr3a mutant (Figure 3E). In contrast, nuclear-encoded mitochondrial genes were mostly unaffected although 24 were found in the upregulated gene set (Supplementary file 1, Figure 3—figure supplement 1C). These observations suggested a loss of mitochondrial DNA. Indeed, qPCR analysis showed that the ratio of two mitochondrial genes (Mt-Rnr2 and Mt-Nd1) to a nuclear gene (Hk2) was substantially reduced in Polr3a-tamKI cerebra (Figure 3—figure supplement 1D). Thus, the loss of mitochondrial genomes underlies the reduction in mitochondrially encoded RNA detected by RNAseq. Hmgcr and Sqle were among numerous sterol biosynthetic genes down-regulated in Polr3a-tamKI cerebra (Figure 3—figure supplement 1E, Supplementary file 1). These enzymes function in rate-limiting steps in de novo cholesterol synthesis, suggesting dysfunctional cholesterol production. Cholesterol synthesis in the brain occurs predominantly in astrocytes and oligodendrocytes, accounts for almost one quarter of total body cholesterol and cannot be compensated by the periphery since cholesterol uptake is blocked by the blood-brain barrier (Saher and Stumpf, 2015). Astrocytes synthesize cholesterol for export to neurons while oligodendrocytes (OLs) synthesize 70–80% of brain cholesterol that resides in myelin (Barber and Raben, 2019; Li et al., 2022). Impairment of cholesterol biosynthesis in either cell type can result in cognitive and behavioral phenotypes (Li et al., 2022) and thus may contribute to the behavioral defects noted in Polr3a-tamKI mice.

We queried the Polr3a-tamKI DE genes against consensus sets of mouse brain cell type gene expression signatures for OLs, neurons, astrocytes, microglia and endothelial cells (McKenzie et al., 2018). Twenty-one percent of Polr3a-tamKI DE genes (p-adj <0.05) could be assigned to one of these five cell types with microglia and neurons constituting the bulk of the cell-type-specific genes (Supplementary file 2). OL and neuronal gene expression was predominantly down-regulated suggesting a reduction in these cell populations while microglia were predominantly elevated (Figure 4A–C, Supplementary file 2). Immunohistochemistry confirmed that these changes in cell type gene expression signatures correlated with changes in cell-specific protein markers (Figure 4D–H). Myelination in the mouse begins at birth and is largely complete by weaning, although it continues with decreasing frequency through adulthood (Hammelrath et al., 2016; Nishiyama et al., 2021). Thus, under our treatment regimen, substantial myelination has occurred by P28 when the Polr3a mutation is introduced. This, together with the long half-life of myelin (Barnes-Vélez et al., 2023) contributes to a high background level of myelin. Accordingly, the mean fluorescence intensity of myelin basic protein (MBP) staining in the corpus callosum of adult Polr3a-tamKI mice was not significantly diminished relative to WT. However, the density of MBP staining was reduced as was the number of OLs staining positive with the CC1 antibody (Figure 4D and E). In the cerebral cortex, the mean fluorescence intensity of NeuN staining and the number of NeuN positive neuronal nuclei was lower in Polr3a-tamKI mice (Figure 4F) consistent with the decrease in neuron-specific gene expression. Additionally, we observed reactive microglia in Polr3a-tamKI cerebra, as evidenced by the increased cell density and altered morphology (enlarged cell bodies and shorter, thicker processes) of Iba1-stained cells in the cortex and particularly in the striatum (Figure 4G and H). Together, these data indicate that behavioral deficits, induction of the ISR, loss of mitochondrial genomes, an elevated immune response and neurodegeneration are causally linked to the loss of Pol III transcription in Polr3a-tamKI cerebra.

To assess whether the changes observed in Pol III and Pol II transcripts in P75 cerebra occurred in other tissues, we examined cerebellum, heart, kidney, and liver from the same animals. The levels of three precursor tRNAs in the cerebellum, heart, kidney were reduced ~threefold, comparable to the changes seen in cerebra (Figure 3—figure supplement 2A). However, in the liver, where the efficiency of CAGGCre-ER recombination is lower (Figure 1—figure supplement 1C; Hayashi and McMahon, 2002), only one of the three precursor tRNAs was reduced (Figure 3—figure supplement 2A). The levels of two mature tRNAs were variably affected, being unchanged in liver and cerebella, trending lower in heart and reduced in kidney to ~70% of WT (Figure 3—figure supplement 2A). Among other Pol III transcripts, tissues other than cerebra showed few changes: 7SL and Rmrp RNA levels were reduced in cerebella but otherwise no reductions were detected (Figure 3—figure supplement 2B). Similarly, there was little indication of an ISR response in the tested tissues other than cerebra (Figure 3—figure supplement 2C). These data imply that the diminished activity of Pol III in these tissues is above the threshold necessary for ISR induction.

To separate causal changes in Pol III activity from secondary stress responses that could down-regulate Pol III transcription, we profiled both the Pol II and Pol III transcriptomes of WT and Polr3a-tamKI cerebra at P42, two weeks after induction of recombination. Similar to the P75 time point, Polr3a-tamKI pre-tRNA levels were reduced ~threefold and three out of four mature tRNAs were reduced to ~60% of the WT level (compare Figure 5A–C with Figure 3A–C). No Pol III transcripts other than tRNAs were decreased at P42 (Figure 5C). Thus, decreased production of precursor and mature tRNA is the earliest detectable Pol III defect in Polr3a-tamKI cerebra.

Analysis of the ATF4-regulated ISR gene panel revealed that fewer genes were induced in Polr3a-tamKI cerebra at P42 and the response was less robust than at P75 (compare Figure 5—figure supplement 1A and Figure 3D). Sesn2, Mthfd2, Chac1, Ald1l2, and Asns were up-regulated twofold or less indicating partial activation of the ISR. Bulk RNAseq showed fewer significant DE genes at P42 (181 genes) with 68% up-regulated and minimal overlap with the P75 DE gene set (Figure 5D, Figure 5—figure supplement 1B, Supplementary file 3). Although no GO terms were enriched, the P42 DE gene set contained a small number of ISR genes (20 genes), the majority of which are ATF4-regulated and showed increased expression (Figure 5—figure supplement 1B, Supplementary file 3). Similarly, a small number of up-regulated genes were associated with the immune response GO bioprocess (21 genes at P42 versus 582 genes at P75, Supplementary file 3). Additionally, the expression of mitochondrially-encoded genes was unaffected at P42, as was mitochondrial genome content (Figure 5—figure supplement 1C, Supplementary file 3). Consistent with these data at P42 and in contrast to the findings at P75, Iba1-staining did not reveal any increase in the density of microglia in the cerebral cortex or the striatum (Figure 5—figure supplement 1F, G). Together these data indicate a progression of stress and immune response programs with time in Polr3a-tamKI cerebra.

The effect of the Polr3a mutation on the tRNAome in P42 cerebra was assessed by QuantM-tRNA-seq (Pinkard et al., 2020). This method exploits the conserved 3’-CCA end of mature tRNAs using a splint-ligation approach to prepare DNA libraries for sequencing. Prior to sample workup, a synthetic tRNA spike-in based on E. coli tRNA^Gln was added for normalization of tRNA reads to total input RNA. Since mitochondrially encoded mRNAs and mitochondrial DNA content were comparable between WT and Polr3a-tamKI cerebra at P42 (Figure 5—figure supplement 1C, Supplementary file 3), mitochondrially encoded tRNAs were also used for normalization. Total nuclear-encoded tRNA reads normalized with either or both standards were similar and significantly lower in Polr3a-tamKI cerebra suggesting an ~25% reduction in total tRNA abundance (Figure 6A, Figure 5—figure supplement 1D, E). Differential expression (DE) analysis indicated that the majority of tRNA isodecoders (i.e. tRNAs that share the same anticodon but have different body sequences) were significantly lower in the Polr3a mutant (58%, 108/187 tRNAs with p-adj <0.05, Supplementary file 5). DE was observed across the entire read depth which spans 5 orders of magnitude and included tRNAs whose sequences could be uniquely mapped as well as redundantly encoded tRNAs (Figure 6B). Compared to most elongator tRNAs, the three tRNAi^Met isodecoders were less dramatically affected, reduced to 86–88% of the WT level. The level of n-TRtct5 (aka tRNA-Arg-TCT-4–1) an abundant neuron-enriched tRNA that is mutated in the C57BL/6 J background and causes synthetic phenotypes when combined with ribosome rescue factor mutants (Ishimura et al., 2014; Terrey et al., 2020), was unchanged in the Polr3a mutant. Thus, the cerebral phenotypes of Polr3a-tamKI mice are unlikely to result from synthetic effects of the n-TRtct5 and Polr3a mutations.

Grouping tRNA isodecoders into their isoacceptor families illustrates that most members of each family were reduced in the Polr3a mutant (Figure 6C). Indeed, when tRNA read counts were combined into their respective decoder (i.e. anticodon) groups, all decoders were reduced in the Polr3a mutant, and for 30 decoders (60%), the reductions were significant (Figure 6F, Supplementary file 5). To assess how these changes affect the composition of the tRNA population (i.e. the tRNA profile), each decoder was plotted as a percentage of the total tRNA population for the Polr3a mutant versus WT (Figure 6E). This revealed changes in the relative proportions of some tRNAs. For example, tRNA^Val(AAC) and tRNA^Tyr(GTA) were proportionally less abundant than tRNA^His(GTG), and seven other tRNAs, respectively. These changes coupled with the overall decrease in tRNA abundance have the potential to change the kinetics of tRNA binding to the ribosome (decoding speed) as well as the kinetic competition with near-cognate tRNAs for ribosome binding. The most severely affected decoders were tRNA^Val(AAC) (to 48% of the WT level), tRNA^Tyr(GTA) (to 50%), tRNA^Sup(TCA) (to 53%), tRNA^Ser(AGA) (to 63%) and tRNA^Gln(CTG) (to 64%). In mouse, tRNA^Ser(AGA) and tRNA^Val(AAC) decode 41% and 42% of their respective codons (accounting for 3^rd base wobble), tRNA^Gln(CTG) decodes 75% of glutamine codons and tRNA^Tyr(GTA) decodes 100% of tyrosine codons. These findings suggest that the decrease in tRNA decoders is likely to negatively affect the tRNA decoding potential at the earliest stage of disease pathogenesis.

An initial analysis of the Pol III transcriptome in Polr3a-tamKI cerebella at P42 revealed comparable findings to the cerebra with decreases in several precursor and mature tRNAs and no effects on other Pol III transcripts (compare Figure 6—figure supplement 1A, B with Figure 5B and C). QuantM-tRNA-seq indicated that total tRNA abundance was reduced in cerebella by ~23% in the Polr3a mutant (Figure 6—figure supplement 1D), similar to cerebra, and the expression of tRNA isodecoders was broadly decreased (Figure 6—figure supplement 1E, F). However, a comparative DE analysis showed that fewer tRNAs (30 vs 108) and fewer tRNA decoders (11 vs 30) achieved statistical significance in Polr3a-tamKI cerebella compared to cerebra (Figure 6G and H, Supplementary file 6) and the distribution of fold change values spanned a narrower range and was shifted towards smaller differences relative to WT (Figure 6D). In addition, the tRNA profile was minimally altered in cerebella (Figure 6—figure supplement 1G). These differential effects on the tRNA population in cerebella versus cerebra are consistent with bulk RNA-seq data from cerebella which showed few significant DE genes (13 genes, p-adj <0.05, log2Fold Change >|0.58|, Supplementary file 4) and limited overlap with cerebral DE genes (Figure 6—figure supplement 1C). The minimal effect of the Polr3a mutation on the Pol II transcriptome of cerebella at P42 and the absence of ISR induction at P75 suggests that any reduction in tRNA decoding potential is insufficient to cause significant pathology in this tissue.

All experiments involving mice were performed using protocols (00001373 and 00001376) approved by the Institutional Animal Care and Use Committee of the Albert Einstein College of Medicine (AECOM). Mice were housed in a barrier facility at 22 °C on a 12 hr light/dark cycle with constant access to food (PicoLab Mouse Diet 20) and water. C57BL/6J-Polr3a^tm1Iwil mice were described previously (Merheb et al., 2021) and were bred to B6.Cg-Tg(CAG-cre/Esr1*5Amc/J) mice (CAGGCre-ER, #004682) and to B6.129(Cg)-Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Luo}/J mice (tdTomato-EGFP, #007676). Whole-body inducible Polr3a mutant mice and Cre-minus WT controls were obtained by breeding male mice heterozygous for both the floxed Polr3a allele and Cre recombinase to female mice homozygous for the floxed Polr3a allele. The breeding of mice to assess recombination used females homozygous for the tdTomato-EGFP reporter. Experimental animals (knockin mutants and WT controls) were treated with tamoxifen (6 mg/40 g body weight, five injections i.p.) every other day starting at P28. Unless otherwise indicated, all experiments were performed with male and female mice and the data were pooled for analysis.

Single-cell suspensions of dissected cerebrum and cerebellum were prepared using an adult brain dissociation kit (Miltenyi). Flow analysis was performed on a LSRII-U flow cytometer (BD Biosciences) with single color controls to establish gating prior to sample analysis (Merheb et al., 2021). For liver, heart and kidney, fixed, sectioned tissue was mounted with ProLong diamond antifade with DAPI (Invitrogen) and imaged using a 3D Histech p250 high-capacity slide scanner. Manual cell counts were performed in ImageJ (five sections/tissue, >150 cells/section).

Images of Nissl-stained sagittal sections of cerebellum were viewed using CaseViewer software for manual counting of Purkinje cells in lobes III, IV/V, VI/VII, and VIII. The open polygon tool was used to measure the length of the Purkinje layer for calculating Purkinje cell density.

Necropsy and staining of tissues with H&E, LFB, and Nissl was performed in the Histopathology and Comparative Pathology Facility at AECOM. Images were acquired using a 3D Histech p250 high-capacity slide scanner.

Behavioral studies were conducted during the light cycle using a Behavioral Spectrometer (BiObserve) equipped with video-tracking, infrared beams and a vibration-sensitive floor to record motion in the X, Y, and Z planes and to measure the frequency and magnitude of vibrations in a 40 cm² arena. Locomotor activity was tracked and stereotyped behaviors were measured algorithmically using Viewer Software (BiObserve) (Brodkin et al., 2014). Specifically, locomotor classes of behavior include Run, Trot, Walk, and Still (not moving) metrics. Locomotor activity - open field-like behavior is defined by Track Length (total distance travelled in the arena). Exploratory classes of behavior include Rearing and Orientation behaviors. Rearing movements include Rear Climb (e.g. trying to climb walls), Rear Bob (moving body in an up/down vertical motion) and Rear Sniff (head movement in a nodding arc) activities. Orientation movements include Orient Shuffle (repositioning of all 4 feet and facing a different direction), Orient Creep (repositioning 1–3 steps more or less in the same direction) and Orient Sniff (moving less than a body length with no head movement) motions. Grooming behaviors are defined by Scratches (involving the hind legs) and grooming of specific body parts (Genital, Tummy, Back, Leg, Head, Face, Nose, and Paw). Anxiety-like behavior is defined by Center visits and Center duration which score the number and duration of visits to a 15 cm² central area of the arena and Center track (total distance travelled in the central area). Individual mice were placed in the instrument for a total of 9 min and recorded in 3-min intervals (3 recordings in total). Behaviors are reported after 6 min (two recordings) and 9 min (all three recordings). Approximately, equal numbers of mice of both genders (five WT females, four WT males and three KI females, two KI males) were examined at weekly intervals on postnatal days 42, 49, 56, and 63. The data for each genotype at each age were combined for analysis. Significance was calculated by multiple t tests and one-way ANOVA in GraphPad Prism.

Mice at P63-P66 were transcardially perfused with 4% paraformaldehyde (PFA), brains were removed and postfixed overnight in 4% PFA at 4 °C before storage in 25% sucrose. Brains were embedded in OCT medium, cut into 10-μm-thick matched sections and stored at –20 °C. Sections were post-fixed in 4% PFA for 10 min at room temperature and rinsed in PBS 3x5 min before permeabilization and blocking in 1% Triton X-100, 10% goat serum, 10% BSA, PBS for 2 hr at room temperature prior to staining. Slides were incubated overnight at 4 °C with primary antibodies, MBP (Abcam Cat# ab7349, RRID:AB_305869; 1:100), CC1 (Millipore Cat# OP80, RRID:AB_2057371; 1:20) or NeuN (Abcam Cat# ab104224, RRID:AB_10711040; 1:100) in 0.1% Triton X-100, 1% goat serum, 10% BSA, PBS and then washed with PBS 3x5 min. Secondary antibodies (1:1000) were conjugated to Alexa-Fluor 488 (Invitrogen A28175), Alexa-Fluor 568 (Invitrogen A21144) or Alexa Fluor 633 (Invitrogen A21094). For Iba1 staining, antigen retrieval (0.025% Tween-20 in 10 mM citrate buffer pH 6.0) and non-specific blocking (10% donkey serum) was performed before incubation overnight with primary antibody (Wako Chemicals RRID:AB_839504; 1:250) and detection with IgG conjugated to Alexa 488 (Invitrogen A11008). After washing in PBS, slides were mounted with ProLong diamond antifade with DAPI (Invitrogen). Images were captured using a 3D Histech P250 High Capacity Slide Scanner and Z-stack images were acquired using a Leica SP8 inverted DMi8 confocal microscope (40 X N.A.1.3).

LIF micrographs were analyzed using Volocity software. Composite images were generated and then split into individual channels. The threshold for the green, red and magenta channels was set uniformly across all sections according to signal intensity. Mean fluorescent intensity and area values were reported directly from Volocity software. CC1 stained cells were counted at the midline of the corpus callosum and NeuN stained cells were counted in the motor cortex. Colocalization of CC1 and MBP signals was interrogated in the merged image throughout the Z-stack and confirmed in the individual split channels. Quantification of Iba1 straining was performed manually in areas covering ~1.3 mm² of the cortex and striatum.

Tissues (50–100 mg, freeze-clamped and flash frozen in liquid nitrogen) were homogenized into TRIzol Reagent. RNA was reprecipitated before quantification. DNA recovered using the TRIzol DNA isolation protocol was used in mitochondrial genome content assays (Quiros et al., 2017). Total RNA (2 µg) was used for cDNA synthesis with SuperScript III (SSIII, Invitrogen). qPCR followed the Lightcycler 480 SYBR Green I Master mix product manual for 384 multiwell plates (Roche). All primers are listed in Supplementary file 7. Each sample was run in triplicate with four to six biological replicates. ΔΔCt values were calculated using two reference genes (GAPDH, β-actin and/or γ-tubulin) (Taylor et al., 2019).

Total RNA (5 µg) was resolved by denaturing polyacrylamide electrophoresis before electrophoretic transfer to Nytran N or SPC membranes (Cytiva), and hybridization to [³²P]-end-labeled oligonucleotide probes at 37 °C (Bonhoure et al., 2015). Transcripts detected by phosphorimaging were quantified using ImageQuant software, normalized to U3 snRNA, and expressed as a fraction of the mean WT value. Oligonucleotide probes are listed in Supplementary file 7.

Total RNA was mixed with a synthetic spike-in tRNA based on E. coli tRNAGln (Supplementary file 7). The spike-in tRNA was denatured and refolded (80 °C 2 min with a 0.1 °C/s ramp to 20 °C in 0.5 mM EDTA), and then added to 2.5  µg of total RNA from P42 cerebra (0.375 ng spike-in) or P42 cerebella (0.750 ng spike-in) prior to tRNA library preparation. The RNA mixture was deacylated (37 °C for 45 min in 20  mM Tris-HCl pH 9.0, RNaseOUT [1  µg/µL, Invitrogen]) and then 3’-end dephosphorylated and 5’-end phosphorylated using T4 polynucleotide kinase (NEB). tRNA library preparation used SuperScript IV following the QuantM-tRNA-seq protocol (Pinkard et al., 2020). Library multiplexing, sequencing (Illumina NextSeq500, single-end 150 bp reads) and mapping was performed by the Center for Epigenomics/Computational Genomics Core at AECOM. Library reads were trimmed and mapped, as specified previously (Pinkard et al., 2020), to a custom look-up table containing the high confidence set of nuclear-encoded cytosolic tRNAs (gtRNAdb grcm38/mm10), mitochondrially encoded tRNAs (RNAcentral) and the spike-in tRNA sequence. tRNA reads  > 10 were used for further analyses. Decoder families were created by summing reads for all tRNAs with the same anticodon sequence. tRNAseq raw read counts for tRNAs and tRNA decoder families were normalized to the mitochondrial tRNA and spike-in tRNA read counts using the estimateSizeFactors control gene function in DESeq2 (Love et al., 2014). Differential expression and significance were determined using the likelihood ratio test (LRT) in DESeq2. Data visualization and plotting was performed in Prism (v9, GraphPad).

Poly(A) selected mRNA libraries were generated from total RNA (RINs >8.9, Agilent Bioanalyzer), sequenced (Illumina PE150) and processed by Novogene. Differential expression and significance was determined using DESeq2 in R studio for all read counts >10 using default settings and the Wald test. DE was defined as adjusted p-value <0.05. Downstream data visualization and plotting were performed using pheatmap in R and Prism (v9, GraphPad).

Statistical analyses were performed using Prism (v9 GraphPad) or Excel. Two-tailed unpaired Student’s t tests were used for experiments with two conditions. Behavioral testing used multiple unpaired t tests and one-way ANOVAs. In molecular analyses, all samples represent biological replicates (no samples were pooled). Significance was determined at p < 0.05. * p<0.05, ** p<0.01, *** p<0.001. Data are presented as mean  ± SD or ± SEM as indicated.