Galnt17 loss-of-function leads to developmental delay and abnormal coordination, activity, and social interactions with cerebellar vermis pathology

Abstract

GALNT17 encodes a N-acetylgalactosaminyltransferase (GalNAc-T) protein specifically involved in mucin-type O-linked glycosylation of target proteins, a process important for cell adhesion, cell signaling, neurotransmitter activity, neurite outgrowth, and neurite sensing. GALNT17, also known as WBSCR17, is located at the edge of the Williams-Beuren Syndrome (WBS) critical region and adjacent to the AUTS2 locus, genomic regions associated with neurodevelopmental phenotypes that are thought to be co-regulated. Although previous data have implicated Galnt17 in neurodevelopment, the in vivo functions of this gene have not been investigated. In this study, we have analyzed behavioral, brain pathology, and molecular phenotypes exhibited by Galnt17 knockout (Galnt17^−/−) mice. We show that Galnt17^−/− mutants exhibit developmental neuropathology within the cerebellar vermis, along with abnormal activity, coordination, and social interaction deficits. Transcriptomic and protein analysis revealed reductions in both mucin type O-glycosylation and heparan sulfate synthesis in the developing mutant cerebellum along with disruption of pathways central to neuron differentiation, axon pathfinding, and synaptic signaling, consistent with the mutant neuropathology. These brain and behavioral phenotypes and molecular data confirm a specific role for Galnt17 in brain development and suggest new clues to factors that could contribute to phenotypes in certain WBS and AUTS2 syndrome patients.

Graphical Abstract

INTRODUCTION

GALNT17 encodes one of twenty mammalian UDP-GalNAc:polypepetide N-acetylgalactosaminyltransferase (GalNAc-T) enzymes, which catalyze mucin-type O-glycosylation, or the transfer of GalNAc from UDP-GalNAc to a hydroxyl group of serine (Ser) or threonine (Thr) in targeted proteins. Mammalian GalNAc-T enzymes are highly similar in amino acid sequences and molecular function, but differ widely in expression patterns indicating tissue-specific activities (Bennett et al., 1999; Cheng et al., 2002; Peng et al., 2010). Mucin-type O-glycosylation plays a critical role in a wide variety of cellular pathways, including cell adhesion and signaling (Ferreira et al., 2018; Tsuiji et al., 2003; van den Steen et al., 1998).

GalNAc-T activity also plays roles in development of the brain. For example, GALNT2 loss-of-function (LOF) is associated with a complex syndrome including developmental delay, intellectual disability, epilepsy and autistic features in human patients and similar phenotypes in rodent models (Zilmer et al., 2020), and Galnt3 functions in neurogenesis (Xu et al., 2016). Furthermore, in the Drosophila nervous system, GalNAc-T substrates play key roles in synapse organization (Itoh and Nishihara, 2021). Like those genes, Galnt17, a Y-subfamily member, shows an evolutionarily conserved pattern of developmental expression in the brain, with broad neuronal expression throughout embryonic development and in adult mammals, the highest levels of expression in cerebellum and hippocampus (Nakamura et al., 2005; Weisner et al., 2019). Y-subfamily members including GALNT17 do not glycosylate canonical GalNAc-T substrate proteins such as mucins, but may act on a limited number of protein substrates or serve as chaperones to regulate the activities of other GalNAc-Ts (Li et al., 2012; Nakayama et al., 2012). At the cellular level, GALNT17 has been shown to regulate membrane trafficking and the formation of lamellipodia , predicting potential roles in neurite development and neuron function (Nakayama et al., 2012).

The GALNT17 gene was initially named WBSCR17 because it is located at the distal edge of the critical region that is commonly deleted or duplicated in Williams-Beuren Syndrome (WBS) patients (Merla et al., 2002). Although GALNT17 is not located within the WBS critical region (WBSCR) per se, genes flanking the WBSCR loop together in nuclear chromatin and interact with Autism Susceptibility Locus 2 (AUTS2), the closest distal neighbor of GALNT17, in a long-range regulatory domain (Gheldof et al., 2013). Long-range interaction between WBSCR and AUTS2 was also suggested by another group, which showed that mouse Caln1, Galnt17 and Auts2 sit within a conserved topologically associating domain (Engmann et al., 2017), and our recent data showed significant co-expression of the neighboring genes (Weisner et al., 2019). This regulatory relationship could help explain why WBS and AUTS2 syndrome patients exhibit so many similar phenotypes, including abnormal social behavior, Attention Deficit Hyperactivity Disorder, developmental delay, speech impairment, and stereotypic movements (Beunders et al., 2015; van der Aa et al., 2009). Interestingly, Autism spectrum disorder (ASD) and WBS patients also display similar neuropathological phenotypes in cerebellum and hippocampus. In particular, both ASD and WBS patients display size changes in the cerebellar vermis (Courchesne et al.,2011; Schmitt et al., 2001), which is thought to be involved in social interaction (Al-Afif et al., 2013; Kelly et al., 2020), and in the hippocampus, where both ASD and WBS patients display abnormal synaptic activity (Gilbert and Man, 2017; Meyer-Lindenberg et al., 2005). The overlapping pathologies suggest the importance of these two brain regions in shared ASD and WBS phenotypes, and further support a possible functional connection between the neighboring sets of genes.

Together these data have suggested a role for GALNT17 in brain development. To test this hypothesis, we investigated behavior, brain structure, and molecular features of Galnt17 knockout (KO) mice, revealing a role in development of the cerebellar vermis and providing new clues to molecular mechanisms that may underlie certain WBS and AUTS2-linked phenotypes.

MATERIALS and METHODS

Animals and tissue collection

Galnt17 KO heterozygous (Galnt17^−/+) mouse sperm samples were purchased from the Knockout Mouse Project (KOMP) repository (http://velocigene.com/komp/detail/16112; official allele name Galnt17^{tm1.1(KOMP)Vlcg}) and used to generate several litters of mutant offspring at the University of Illinois Transgenic mouse facility by in vitro fertilization of C57BL/6J females. The resulting Galnt17^−/+ animals were then interbred to obtain Galnt17 KO homozygous (Galnt17^−/−) pups. Surprisingly, although live homozygous mice were generated on a closely related strain (C57BL/6NCrl) for basic testing in the International Mouse Phenotyping Consortium (IMPC; https://www.mousephenotype.org), homozygous pups were not born live on the C57BL/6J in repeated litters. To obtain live animals we therefore bred the heterozygous mice to C3H/He mice to create a C57BL/6J X C3H/He F1 (B6C3) F1 hybrid background, and crossed them to obtain, homozygous, heterozyogous and wild type (WT) littermates. Genotyping primers and protocols were followed according to the KOMP website. Animals of each genotype were measured for weight at postnatal day 5 (P5), P7, P9, P35, and P60; tested for developmental delay from P5-P16; tested for coordination at P28 and P42; and administered behavioral tests from P42-P56. The animals were maintained under standard conditions (12h light/dark cycle, group housed); All mouse phenotypic tests were conducted in the light cycle, between one and five hours before lights-off. This study strictly followed the Guide for the Care and Use of Laboratory Animals of the Use Committee of the University of Illinois (Animal Assurance Number: A3118–01; approved IACUC protocol number 18240).

Mouse phenotypic tests

Developmental delay tests

Developmental delay tests included two experimental designs: 1) Righting reflex - the pups were flipped and the time for them to turn back to their original position was recorded. Different scores were assigned each day according to their performance. Score 0 = pups can’t right themselves within 15 seconds; score 1 = pups turn back between 10 to 15 seconds; score 2 = the first day of righting within 10 seconds; score 3 = the second day of righting within 10 seconds. 2) Eye opening was examined, and different scores were assigned to different stages of the eye-opening process, score 0 = both eyes are closed; score 1 = one eye is open; score 2 = both eyes are open.

Coordination test

Coordination test was to test the ability of mice to balance and walk along the ledge of the normal cage (Guyenet et al., 2010). The protocol was modified to test mice with two different conditions: (1) with the cage placed flat on the surface of the table and (2) after tilting the cage up at 22.5 degrees. The time for the mice to complete walking along one long arm was measured.

Novel Object Recognition (NOR) test

Novel Object Recognition (NOR) test was previously described (Botton et al., 2010). In short, there are 3 test sections: habituation, training, and retention; each section is completed 24 hours apart. Mice had 10 minutes to explore the empty box in the habituation section, 10 minutes to explore two novel LEGO toys in the training section, and 5 minutes to explore an old and a novel LEGO toy in the retention section. The exploratory index was calculated as the exploration time on the novel object versus the total exploration time as a percentage.

Horizonal activity

Horizonal activity was quantitated as total traveled distance in a 10-minute habituation section in NOR test in an empty 35 cm × 30 cm paper box using ToxTrac software (Rodriguez et al., 2018).

Vertical activity

Vertical activity was examined by measuring rearing behavior, including free rearing (in which the animal rears on its hind feet without support) and wall rearing (in which the animal balances its front paws against a wall). The number of times a mouse exhibited free and wall rearing was observed within 10 minutes of habituation section in NOR test.

Three chamber test

Three chamber test was performed to examine social preference (SP) (Naviaux et al., 2013). A Plexiglas box was divided into 3 chambers: object chamber (left), middle chamber, and social chamber (right). A LEGO toy was placed in the object chamber, and a sexually immature juvenile female mouse (P21-P28) was placed in the social chamber. Both the LEGO toy and the immature juvenile female mouse were covered by a small stainless steel wire cup. The test mouse was first given 10 minutes to freely explore the empty box. Secondly, both the LEGO toy and juvenile female mouse were placed into each chamber when the test mouse was removed from the empty box after 10 minutes of exploration time. Finally, the test mouse was placed in the middle chamber again and was given 10 minutes to explore each chamber. Each juvenile female mouse was used for at most 3 tests. A SP percentage was calculated as the ratio of exploration time in the social chamber to the exploration time in both chambers.

Food burying test

Food burying test was carried out to assess olfactory function of the mice (Yang and Crawley, 2009) . Each test mouse was singly housed the day before the test and food was withheld for 18–24 hours. On the test day, each mouse was transferred to a new cage containing 3 cm deep solid corncob bedding and given 5 minutes to explore. Then, the test mouse was removed, and one food pellet was embedded 1.5 cm deep. The test mouse was put back to the cage, and the time for the test mouse to locate the food pellet was measured.

Tail suspension test

Tail suspension test was performed to determine mobility of a mouse (Mori et al., 2005). The mouse was suspended by the tail 30 cm above the testing station. The movements of forelimbs and hindlimbs were observed for each test mouse.

Histopathology

Hematoxylin and eosin stained paraffin slide preparation and scanning with a NanoZoomer Digital Pathology System (Hamamatsu) instrument were previously described (Weisner et al., 2019). For cerebellum, two sections from each cerebellar vermis and hemisphere region were chosen to measure the thickness of the external granular layer (EGL) and granule cell layer (GL) in each lobule and fissure using the “Ruler” function in NanoZoomer Digital Pathology Software, NDP.view 2. To quantitate the number of Bergmann glia (BG), two 50 μm lines were randomly drawn parallel to each fissure on glial fibrillary acidic protein (GFAP) stained cerebellar confocal images, and the number of intersects of each 50 μm line and GFAP+ BG processes was quantitated. A TUNEL assay was performed using the DeadEnd Colorimetric TUNEL System (Promega, G7360) to measure the number of dead cells, and stained paraffin slides were scanned by a NanoZoomer Digital Pathology System (Hamamatsu).

Immunohistochemistry (IHC)

IHC was carried out as previously described (Weisner et al., 2019). Briefly, cerebellar slides were blocked with Antibody Diluent Reagent Solution (Life Technologies), washed with Phosphate-buffered saline (PBS), stained with primary antibodies at an optimal ratio (listed below) and incubated overnight at 4°C. Slides were then washed and incubated with secondary antibody (1:200) for 1 hour at room temperature. Finally, slides were washed again, stained with the nuclear marker, Hoechst 33342 (1:10,000), mounted, and scanned under a Zeiss LSM 710 Confocal Microscope or Leica TCS SP8 Confocal Microscope. For bromodeoxyuridine (BrdU) injection experiments, P14 pups were injected with 0.05 mg/g BrdU 24 hours, 8 hours and 4 hours prior to tissue harvesting. The BrdU-injected paraffin embedded cerebellar slides were incubated with 1M HCL for 1 hour at 37°C followed by de-waxing steps and neutralized with 0.1M sodium borate buffer pH8.5 for 10 minutes at room temperature. Antibody staining was then carried out as described above. Primary antibodies: GFAP (Invitrogen 180063, 1:400), BrdU (Abcam ab6326, 1:400), Calbindin (CalB; Sigma C9848, 1:400), beta-III Tubulin (Tuj1; Neuromics CH23005, 1:200), Phospho-H3 (Millipore 06–570, 1:500); secondary antibodies: Thermo-Fisher Scientific, Alexa Fluor 488, Alexa Fluor 594.

Western blot/Lectin blot

Nuclear and non-nuclear protein extracts were prepared from dissected tissues from individual animals using a NucBuster Protein Extraction Kit (Millipore) and concentration was measured on a NanoDrop 1000 Spectrophotometer. Protein extracts were run on 10% acrylamide gels and transferred to hydrophobic polyvinylidene difluoride (PVDF) membranes. For detection of glycoproteins, the membrane was incubated with 1.5% bovine serum albumin (BSA) for at least 1 hour at room temperature for blocking and incubated with a Vicia Villosa (VVA; Vector Laboratories, B-1235–2) lectin in 1.5% BSA for 45 minutes at room temperature. After washing with 0.2% TBST, membranes were incubated with HRP conjugated secondary antibodies in 1.5% BSA for 30 minutes at room temperature. For normal western blot, membranes were blocked with 5% non-fat milk, washed with 0.2% PBST, and incubated with primary and secondary antibodies in 1% non-fat milk for overnight at 4°C and 2 hours at room temperature, respectively. Finally, membranes were treated with ECL enhancer (ThermoFisher) and visualized with a FluorChem E System (ProteinSimple). The blots were then stripped with stripping buffer (ThermoFisher) for 45 minutes at room temperature, washed with 0.2% TBST or PBST, and incubated with loading controls. Band intensities were quantified using ImageJ-Fiji software v2.1.0, with each normalized to the signal of standardized control antibodies used on the same blots, as described in the figures and text. Statistical significance of difference between normalized mutant and WT intensities was assessed using the ANOVA single-factor test. Primary antibodies: HS (Millipore MAB2040), Beta-Actin (Santa Cruz sc-47778), Lamin B1 (Abcam ab16048).

Quantitative RT-PCR (qRT-PCR)

qRT-PCR was completed as described by Weisner, Chen et al., (2019). Cerebellum and hippocampus were harvested from P14 animals, and RNA was purified from the tissues using Trizol (Invitrogen) extraction followed by RNase-free DNase I (NEB) treatment and RNA Clean & Concentrator Kit (Zymo Research). cDNA was then synthesized from 2 ug RNA with Random Primer Mix (NEB) and M-MuLV Reverse Transcriptase (NEB). qRT-PCR was then carried out using custom-designed primers sets specific to Galnt17 with Power SYBR Green PCR master mix (Applied Biosystems) in a final volume of 10 μl. Reactions were run on an Applied Biosystems QuantStudio Flex 6 thermocycler, and expression values were normalized relative to the Pgk1 control in each sample and compared using standard methods (Schmittgen and Livak, 2001). Two custom-designed primer sets for Galnt17 transcripts were: 1) Galnt17 Exon 3–4 forward 5’-CCTGGTGGATGACAACAGTGA-3’; reverse 5’-CCTTCTCGCTTTTTGATTGCGTA-3’. 2) Galnt17 Exon 8–10 forward 5’-CACTATCGCCTATGGGGAGC-3’; reverse 5’-AACCCTTCCTTGGTGTAGCG-3’.

RNA-seq

Three P14 Galnt17^−/− mutant and three P14 WT male RNA samples were prepared as previously described (Weisner et al., 2019), and RNA-seq libraries were generated using Illumina TruSeq RNA Library Prep Kits. Samples were sequenced as 150 bp paired end reads (≥ 23 million read pairs per sample) on Illumina Hi-Seq 4000 sequencer through the GENEWIZ sequencing service (South Plainfield, NJ). STAR v2.7.5a was used to map reads (mm10/GRCm38 genome and Ensembl v102 annotation) and quantify gene expression (Dobin et al., 2013), and differential expression was determined using EdgeR v3 (Robinson et al., 2009) as described in Saul and Seward et al. (Saul, Seward et al., 2017). For functional analysis, we used ToppCluster annotation tool (Kaimal et al., 2010) with default statistical parameters, focusing on genes identified as having at least 1.5-fold change (FC) in mutants compared to WT controls with FDR less than 0.05.

Statistical analysis

One-way ANOVA and Tukey’s Honestly Significant Difference (HSD) test were conducted on mouse phenotypic tests exclude tail suspension test to compare between WT, Galnt17^−/+ and Galnt17^−/− animals, and Tukey’s HSD test p-value was mentioned in the Results section. T-Test was applied on cerebellar histopathological studies to compare the actual data points from WT and Galnt17^−/− mice, and those data points were used to calculate mean WT and mutant for mutant/WT ratio. To obtain standard error (SE) for mutant/WT ratio, the following formula was used: SE_ratio = mutant/WT × [(SEmutant/mean mutant)+(SE_WT/mean WT)]^1/2.

Data availability

Sequencing data from this project have been deposited in the Gene Expression Omnibus (GEO) database under accession number GSE180040.

RESULTS

Generation and validation of Galnt17 knock-out mutants

To investigate the functions of Galnt17 in brain development, we obtained heterozygous KO mouse sperm from the KOMP and mice as described in Methods. We validated the KO by measuring expression of Galnt17 in WT, Galnt17^−/+ and Galnt17^−/− hippocampus and cerebellum with qRT-PCR, using two different primer sets that targeted exon 3–4 and 8–10, respectively (Figure 1A; see methods for primer sequences). These two primer sets were designed because both Merla et al. (2002) and the public annotation (Refseq, as visualized in the UCSC browser (https://genome.ucsc.edu/index.html) have suggested that the human GALNT17 gene includes a downstream alternative promoter that would not be affected by the KOMP exon 1 KO design. Although there was no evidence that such an alternative promoter exists in mouse, we wanted to confirm that the Galnt17 mutants we had obtained were truly null for Galnt17 expression in brain. We generated RNA from P14 cerebellum and hippocampus, the two brain regions with highest levels of expression according to published studies (Nakamura et al., 2005; Weisner et al., 2019). The qRT-PCR result confirmed that Galnt17 expression is around 50% lower in Galnt17^−/+ mutants than in WT mice, and completely ablated in the both cerebellum and hippocampus of Galnt17^−/− animals, with both primer sets giving consistent results (Figure 1B). The data indicated that the KO allele does not express Galnt17, including any possible alternative isoforms for this gene.

Figure 1. Expression of Galnt17 in Galnt17 KO mutants using qRT-PCR.

(A) Locations of custom-designed primer sets targeting different exons of Galnt17. Primer set 1 targets exon 3 and 4 of Galnt17; primer set 2 targets exon 8 and 10 of Galnt17. (B) The expression of Galnt17 is around 50 % lower in Galnt17^−/+ mutants and is completely ablated in Galnt17^−/− mice compared to WT controls in the cerebellum and hippocampus at P14 for both primer sets. Galnt17 mRNA expression level was normalized relative to Pgk1 control in WT and mutants. n=3 for each genotype and brain region. HC, hippocampus; CB, cerebellum.

Galnt17 loss-of-function leads to developmental delay and adult deficits in motor coordination

Developmental delay is a phenotype observed in both WBS and AUTS2 syndrome patients (Beunders et al., 2013; Pober, 2010). We therefore characterized basic developmental phenotypes in Galnt17^−/+ and Galnt17^−/− mice and WT animals, including timing of development of the righting reflex and eye opening from P5 to P17. IMPC has reported that Galnt17^−/− and WT animals do not differ in weight as young adults (measured weekly from 4–16 weeks), but did not measure body weight during earlier postnatal development. We measured animals of all three genotypes beginning at earlier developmental stages (P5, P7, and P9) as well as adulthood (P35, P60) and discovered that Galnt17^−/+ mice weighed less than age-matched WT mice at P5 (p<.001), P7 (p<.001), P9 (p<.001) and P35 (p<.01), although there was no weight difference between heterozygous mice and WT at P60 (Figure 2A, B). In contrast, Galnt17^−/− mutants weighed significantly less at all the time points compared to age-matched WT mice (Figure 2A, B; P5: p<.01, P7: p<.001, P9: p<.01, P35: p<.001, P60: p<.001). Galnt17^−/+ animals were similar in weight to WT mice at P60, matching what IMPC has reported. However, in our experiments, Galnt17^−/− mutants still weighed less than age-matched WT mice at this time point. Focusing next on other developmental indicators, we found no difference in the development of righting reflex in both Galnt17^−/+ or Galnt17^−/− compared to WT pups (not shown). However, eye opening was delayed in both Galnt17^−/+ and Galnt17^−/− compared to WT pups, with Galnt17^−/− pups fully opening their eyes one day later than Galnt17^−/+ pups (Figure 2C).

Figure 2. Postnatal growth and development of Galnt17 KO mutants.

(A) Body weights were measured at P5, P7 and P9 for each genotype. Galnt17^−/+ (light gray) and Galnt17^−/− (dark gray) mutants were lighter in weight at all time points compared age-matched WT (white) mice. (B) Body weights were measured in adult stages for each genotype. Galnt17^−/− mutants were lighter than WT mice at all the time points. Galnt17^−/+ mutants were lighter at P35, and caught up to WT mice at P60. n, WT = 8–17; Galnt17^−/+ = 9–12; Galnt17^−/− = 12–17. (C) The development of eye opening was measured from P9 to P17, and scores were assigned based on different stages of eye opening development. Galnt17 KO mutants started opening their eyes later and took more time to fully open their eyes than WT (light gray) mice. Galnt17^−/− (black) mutants fully opened their eyes one day later than Galnt17^−/+ (dark gray) mutants. n, WT =13; Galnt17^−/+ = 10; Glant17^−/− = 11. Tukey’s Honestly Significant Difference (HSD) test, *p<0.05; **p<0.01; ***p<0.001.

The IMPC did report that young adult Galnt17^−/− mice display decreased vertical activity, suggesting a role for the gene in motor coordination and control. To confirm and extend these observations, we evaluated activity for adult (6–8 week old) animals, observing horizontal and vertical activity levels. Both Galnt17^−/+ and Galnt17^−/− mice exhibited normal horizontal activity compared to WT (not shown). However, whereas Galnt17^−/− mutants showed normal wall rearing, but reduced levels of free rearing when compared to the WT mice (Figure 3A; p<.05), Galnt17^−/+ did not differ from WT mice in either rearing type (Figure 3A). The lighter weight and developmental delay observed in Galnt17^−/− mutants suggested a similarity to AUTS2 syndrome and WBS phenotypes, while the reduced levels of vertical activity suggested abnormal coordination in the homozygous mutants, which is a common phenotype observed in WBS patients (Pober, 2010).

Figure 3. The assessments of coordination on Galnt17 KO mutants.

(A) Number of free and wall rearing was quantitated to determine the vertical activity in each genotype. Galnt17^−/− (dark gray) mutants had reduced free rearing compared to WT (white) mice. WT and Galnt17 KO mutants exhibited similar number of wall rearing. n, WT = 9; Galnt17^−/+ = 11; Galnt17^−/− = 13. (B) Walking on the ledge test was performed on 6 weeks old animals, and walking time was measured under two different conditions (see methods). Galnt17^−/− mutants walked more slowly than WT and Galnt17^−/+ (light gray) mutants for both conditions. However, Galnt17^−/+ mutants walked faster than WT mice on the tilted cage. (C) When weight was taken into account for the walking time, Galnt17^−/− mutants required more time than WT and Galnt17^−/+ mice for both conditions. The walking time for Galnt17^−/+ mutants became insignificant compared to WT mice when cage was tilted up. n, WT and Galnt17^−/− = 8; Galnt17^−/+ = 7. Tukey’s HSD test,*/# p<0.05, **/## p<0.01, ***/### p<0.001.

To further investigate motor coordination, we performed the “walking on the ledge” test (Guyenet et al., 2010) on P28 and 6 weeks old (P42) adult animals under two conditions: (1) with the cage placed flat on the surface of the table and (2) after tilting the cage up at 22.5 degrees. In both conditions, Galnt17^−/− mutants took a significantly longer time to walk on the ledge of the cage than did WT and Galnt17^−/+ mice at two time points (Figure 3B for adult; flat condition - WT to Galnt17^−/−: p<.01, Galnt17^−/+ to Galnt17^−/−: p<.001; 22.5 degrees - WT to Galnt17^−/−: p<.01, Galnt17^−/+ to Galnt17^−/−: p<.001; P28 not shown). In contrast, adult Galnt17^−/+ mice walked along the ledge of the tilted cage more quickly than the WT controls (Figure 3B; p<.05). However, studies have been shown that weight is highly correlated with coordination in rodents (Mao et al., 2015), and when we took weight into account as a factor in walking time in Galnt17 KO and WT animals, the significance of the result for Galnt17^−/− mutants was increased (flat condition – WT to Galnt17^−/−: p<.001, Galnt17^−/+ to Galnt17^−/−: p<.001; 22.5 degrees – WT to Galnt17^−/−: p<.001, Galnt17^−/+ to Galnt17^−/−: p<.001), but the faster walking for Galnt17^−/+ mutants compared to age-matched WT mice became insignificant (Figure 3C). These results, together with the reduced vertical activity displayed by Galnt17^−/− mutants, indicated coordination problems in the homozygous mutants, although motor coordination was normal in Galnt17^−/+ mice.

Impaired social behavior and reduced exploratory activity in Galnt17 knockout mutants

Since both AUTS2 region mutations and WBS are associated with abnormal social behavior (Beunders et al., 2013; van der Aa et al., 2009), we also tested social behavior in the Galnt17 KO mutant mice. For this purpose, we used the three-chamber test, in which the test mouse was allowed to choose to interact with a novel object (a LEGO toy) in object chamber, or a juvenile female mouse in social chamber (Naviaux et al., 2013) (Figure 4A). Galnt17^−/− animals spent significantly less time in the social chamber (55% of SP) and more time in the object chamber (45% of SP) compared to either Galnt17^−/+ or WT animals (Figure 4B; WT to Galnt17^−/−: p<.001, Galnt17^−/+ to Galnt17^−/−: p<.001). Since impaired olfaction can affect the results of this social test, we also conducted a food burying test (Yang and Crawley, 2009) and confirmed that olfactory function was normal in animals of both mutant genotypes (not shown). Finally, since both WBS and AUTS2 syndrome patients display intellectual disability and learning deficits (Beunders et al., 2013; Pober, 2010), we tested learning and memory using the NOR test (Botton et al., 2010). The results indicated that Galnt17^−/− and Galnt17^−/+ animals have normal learning and memory (Figure 4C). However, Galnt17^−/− mice exhibited significantly reduced exploratory activity in the NOR test compared to WT mice (Figure 4D; WT to Galnt17^−/−: p<.05).

Figure 4. Impaired social behavior and reduced exploratory activity in Galnt17^−/− mutants.

(A) Three-chamber test setup. The left chamber is an object chamber, and the right chamber is a social chamber. The test mouse is placed in the middle chamber. (B) Social behavior was examined through three-chamber test. Galnt17^−/− mutants spent more time in object chamber and less time in social chamber compared to WT and Galnt17^−/+ mice. (C) NOR test was carried out to determine learning and memory ability for each genotype, and it showed Galnt17 KO mutants had normal learning and memory like WT mice. (D) The exploratory activity was determined in training section of NOR test, and Galnt17^−/− (dark gray) mutants showed reduced object exploration time compared to WT (white) and Galnt17^−/+ (light gray) animals. n, WT = 8–9; Galnt17^−/+ = 8–10; Galnt17^−/− = 9. Tukey’s HSD test, *p<0.05; ***p<0.001.

Together, these behavioral tests show that Galnt17^−/− mutants displayed abnormal motor coordination, impaired social behavior and reduced exploratory activity - all phenotypes related to abnormalities observed in AUTS2 syndrome and WBS patients (Beunders et al., 2013; Beunders et al., 2015; Pober, 2010; Tamada et al., 2010). Since the cerebellum plays a key role in these phenotypes (Al-Afif et al., 2013; Caston et al., 1998), and Galnt17 is highly and widely expressed in the developing cerebellum (Nakamura et al., 2005; Weisner et al., 2019), we hypothesized that cerebellar development and function might be compromised in Galnt17^−/− mutants.

Galnt17^−/− mutants display delayed development and reduced thickness of the cerebellar vermis in adulthood

We investigated this possibility by conducting a histopathological evaluation of Galnt17 KO cerebellum. Of the two major cerebellar regions, the hemispheres are responsible for higher cognitive function, whereas the vermis is involved in emotional processing and motor movements (Koziol et al., 2014; Turner et al., 2007). Both the vermis and hemispheres are composed of specific lobules and fissures (Figure 5A for hemispheres; Figure 6A for vermis) and the precise timing of lobulation and foliation is carefully regulated, with formation of each lobule occurring at different time points (Sudarov and Joyner, 2007). The development of each region requires the coordinated differentiation, migration, and interaction between granule cells (GCs), Purkinje cells (PCs) and BG; deficits in the differentiation of any one of these cell types leads to significant cerebellar abnormalities (Rahimi-Balaei et al., 2018). We used hematoxylin and eosin staining and IHC to analyze cerebellar vermis and hemisphere sections at certain positions (Figure 5A for hemisphere; Figure 6A for vermis) in one week- (P7), two week- (P14) and five week-old (P35) WT and Galnt17^−/− mice. We quantitated the thickness of the EGL at P7 and P14, and GL for all three time points, and determined the number of BG at P7 and P14.

Figure 5. Normal hemispherical development in Galnt17^−/− mutant cerebellum.

(A) A diagram showing the cell layer thickness was analyzed at certain position, and WT hemisphere images at P14 and P35. (B) Mutant/WT ratio of cell layer thickness of GL in each lobule at P14 (dark gray) and P35 (black), and EGL in each fissure at P14 (light gray). Galnt17^−/− mutants had similar thickness of EGL and GL at P14, and GL at P35 compared to WT mice. Standard error for Mut/WT was calculated using following formula: SE_ratio = Mut/WT × [(SE_mut/mean Mut)+(SE_WT/mean WT)]^1/2. n, WT = 3–4; Galnt17^−/− = 3–5. Scale bar = 1 mm. Pr, primary; S, simple lobule; sp, superior posterior; CI, lobule crus I; itc, intercrural; CII, lobule crus II; ans, ansoparamedia; PML, paramedian lobule; pp, prepyramidal; COP, copula pyramids; Mut, mutant; EGL, external granule cell layer; GL, granule cell layer.

Figure 6. Delayed development of cerebellar vermis in Galnt17^−/− mutants.

(A) A cerebellum diagram showing the vermis position for histological analysis, and WT vermis images at P7, P14 and P35. Scale bar = 1 mm. (B) Galnt17^−/− mutants had increased BrdU+/Phospho-H3+ cells in lobules III, IV/V, VI and IX compared to WT at P14. (C) Cell layer thickness of EGL was measured at P7 and P14, and thickness of GL was measured at P7, P14 and P35. Galnt17^−/− mutants had decreased EGL in lobule IV/V at P7 (dashed black line), increased EGL in lobules VI, VIII and IX at P14 (dashed dark gray line), and decreased GL in lobules IV/V and VI at P14 (dark gray line) and P35 (light gray line). Lobule VII was not yet formed at P7, thus it was 0 for thickness of EGL and GL. (D) The comparison image of GL and EGL in lobule VI between WT and mutant at P14. Scale bar = 50 μm. (E) Galnt17^−/− Mutants had increased Bergmann glia in fissures prc, pr and pp at P7 (black line), and in fissures pr, sec and pl at P14 (gray line). Foliation of fissure itc was not completed at P7, thus the number of BG was 0. n, WT=Galnt17^−/− = 3–5. Standard error for Mut/WT was calculated using following formula: SE_ratio = Mut/WT × [(SE_mut/mean Mut)+(SE_WT/mean WT)]^1/2, and statistical analysis was conducted using actual data points from WT and mutant animals. t-Test, *p<0.05; **p<0.01. Pr, primary; S, simple lobule; CI, lobule crus I; CII, lobule crus II; PML, paramedian lobule; COP, copula pyramids; prc, precentral; pc, pre-culminate; pr, primary; itc, intercrural; pp, prepyramidal; sec, secondary; pl, posterolateral; Mut, mutant; EGL, external granule cell layer; GL, granule cell layer.

To identify pathology that might be associated with the behavioral phenotypes, we first looked at cerebellar hemispheres in Galnt17^−/− animals. The proliferation peak of granule cell precursors (GCPs) is around P7; those newly generated GCPs migrate from the EGL to GL along the BG processes (Behesti and Marino, 2009; Espinosa and Luo, 2008). This migration activity is completed by P21, so there is no EGL in normal P35 mice (Espinosa and Luo, 2008; Rahimi-Balaei et al., 2018). Although we cannot rule out subtle differences in hemisphere development, we found no difference between WT and Galnt17^−/− animals in the thickness of EGL or GL at P14 (Figure 5B), or in the GL at P35 (Figure 5B) in multiple slides from hemispheric regions in multiple animals (n=3 of each stage). Our observations thus indicated that hemispheric development is normal in Galnt17^−/− mutants.

In contrast, we observed that the thickness of the EGL was decreased in vermis lobule IV/V at P7 (t(4)=5.03, p<.01), which is the proliferation peak for GCPs, and increased in vermis lobule VI, VIII and IX at P14 of Galnt17^−/− mice (Figure 6C and 6D; VI: t(5)=−6.67, p<.01, VIII: t(3)=−5.06, p<.05, IX: t(5)=−3.59, p<.05). Furthermore, the thickness of the GL in Galnt17^−/− mutants was decreased in vermis lobule IV/V and VI at both P14 and P35 (P14 - IV/V: t(5)=3.78, p<.05, VI: t(4)=3.85, p<.05; P35 - IV/V: t(3)=4.30, p<.05, VI: t(2)=5.45, p<.05), although no difference was observed in these lobules at P7 compared to age-matched WT mice (Figure 6C and 6D). These data suggested that differentiation and migration of GCPs are delayed in the mutant vermis.

Since we identified pronounced abnormalities in the vermis, we counted dividing cells specifically in the vermal regions. We injected animals with BrdU and counted the numbers of BrdU+/phospho-histone H3+ (Phospho-H3+) cells in the EGL at P14. The hypothesis regarding delayed differentiation was supported by detection of increased numbers of BrdU+/Phospho-H3+ cells specifically in the EGL of lobules IV/V, VI and IX at P14 (Figure 6B; IV/V: t(5)=−6.79, p<.01, VI: t(4)=−3.40, p<.05, IX: t(3)= −3.99, p<.05), when proliferation should be subsiding.

Additionally, since BG guide GCPs to migrate from ECL to GL and BG might therefore influence the vermal pathology we observed, we examined BG numbers in the vermal region. Indeed, compared to WT controls, Galnt17^−/− mutants had increased numbers of BG the precentral (prc), primary (pr) and prepyramidal (pp) fissures at P7 (Figure 6E; prc: t(3)=−3.81, p<.05, pr: t(4)= −2.78, p<.05, pp: t(3)= −5.63, p<.05), and in the pr, secondary (sec) and posterolateral (pl) fissures at P14 (Figure 6E; pr: t(5)= −4.61, p<0.01, sec: t(6)= −2.86, p<.05, pl: t(5)= −5.51, p<.01). These fissures correspond to the lobules that have increased thickness of EGL at P7 and P14, and decreased thickness of GL at P35 in the mutant mice (Figure 6C). On the other hand, the thickness of GL in lobules VIII and IX did not correlate with the increased number of BG in P35 mutants; we speculate that this is because lobulation and foliation of these two lobules occur at earlier time points than in lobules IV/V and VI, and that by P14, lobules VIII and IX have had more time to overcome the delayed maturation of both GCPs and BG.

To monitor PCs, we stained P60 WT and Galnt17^−/− cerebellum slices with Tuj1 and CalB antibodies. This analysis revealed abnormal morphology of primary branches of PC dendrites in the vermal region of the Galnt17^−/− adult brains. Specifically, the primary branches were 1) thinner, 2) not linear (Figure 7A; arrow heads) and 3) bifurcated (Figure 7A; arrows) in mutants compared to WT PCs at P60 (Figure 7A). Thinner primary branches were also observed in P14 Galnt17^−/− mutants (not shown). CalB staining also showed reduced bundling of PC axons in P14 mutant axonal tract (Figure 7B); the axonal tracts in P14 Galnt17^−/− animals were also significantly narrower in lobule IV/V compared to that of age-matched WT mice (Figure 7C; t(2)=7.08, p<.05). In contrast, we found no difference in PC dendrite or axon morphology between WT and Galnt17^−/− animals within the hemispheric regions (not shown). Together, these histopathological studies indicate that deviant growth and maturation of GCPs and BG together with abnormal PC dendritic and axonal development contribute to delayed vermal development and subsequently, reduced size of lobules IV/V and VI in the cerebellar vermis of mutant adults.

Figure 7. Defects in Purkinje cell morphology and forelimb clasping in Galnt17^−/− mutants.

(A) Tuj1 (left) and CalB (right) staining showed the morphology of Purkinje cells in WT (top) and Galnt17^−/− (bottom) animals. Galnt17^−/− mutants lost thick primary branches (arrowheads), and the primary branches were bifurcated (arrows) compared to WT mice at P60. (B) Galnt17^−/− mutants (bottom) showed fewer CalB (green) marked axons in axonal tract in lobule IV/V compared to WT mice (top) at P14. (C) Width of axonal tract was measured, and it showed mutants (dark gray) have narrower axonal tract than WT (light gray) in lobule IV/V at P14. n=3 for each genotype. t-Test, *p<0.05. (D) Galnt17^−/− mutants (right) showed forelimb clasping (arrowhead) during tail suspension test compared to WT (left) mice in adult. n=5 animals for each genotype were scored for clasping (1) or no clasping (0) with all mutant animals scoring as 1 and all WT animals scoring as 0 in this test. Scale bar = 50 μm.

Brain pathology studies have documented decreased anterior and posterior vermal volume in both WBS and ASD patients (Courchesne et al., 2011; Schmitt et al., 2001), and patients with lesions in the vermal, paravermal and hemispheric lobules IV/V and VI suffer from upper limb ataxia (Schoch et al., 2006). The vermal pathology thus led us to ask whether Galnt17 mutant mice might display signs of forelimb ataxia. To test limb movements, we carried out a tail suspension test (Mori et al., 2005) on adult Galnt17^−/− and WT mice (n=5 for each genotype). The WT animals never displayed this phenotype, but each of the test Galnt17^−/− mice displayed clear and consistent forelimb clasping under these conditions. This simple test thus suggests that the vermal abnormalities are indeed associated with forelimb ataxia in Galnt17^−/− mice (Figure 7D).

Global gene expression reveals molecular pathways affected by Galnt17 loss-of-function

To gain a clearer view of the relationship between Galnt17 LOF and the behavioral and neuropathological phenotypes exhibited by the mutant animals, we compared global gene expression in P14 WT and Galnt17^−/− cerebellum using RNA-seq. Consistent with qRT-PCR and as expected, Galnt17 was detected as one of the most highly down-regulated genes (Table 1). In addition, RNA-seq identified hundreds of other significant differentially expressed genes (DEGs; defined as FDR ≤ 0.05). Down-regulated gene groups showed the most significant functional enrichments for distinct gene ontology (GO) categories and pathways (Table 2).

Table 1.

Down-regulated or up-regulated differentially expressed genes detected in Galnt17−/− compared to wild type P14 cerebellum.

Down-regulated					Up-regulated
Gene name	logFC	logCPM	Pvalue	FDR	Gene name	logFC	logCPM	Pvalue	FDR
Gm10503	−7.32	0.19	3.56E-37	2.36E-34	Gm6560	0.57	2.91	3.58E-05	1.75E-03
Rps18-ps4	−4.44	2.02	3.11E-79	7.91E-76	Gm10925	0.57	5.46	1.70E-12	3.12E-10
Rps18-ps6	−4.29	1.85	5.34E-68	1.17E-64	Gm11223	0.57	2.61	1.52E-04	5.57E-03
Galnt17	−4.11	3.14	1.28E-141	4.88E-138	Ccdc61	0.58	2.04	1.34E-03	3.24E-02
Rps2-ps10	−4.06	3.17	2.65E-145	1.35E-141	Gm13456	0.58	4.00	1.17E-08	1.25E-06
Gm15772	−3.82	3.32	6.69E-146	5.11E-142	Rbm24	0.58	1.85	1.91E-03	4.13E-02
Gm45104	−3.69	1.09	1.15E-38	8.37E-36	Wdfy1	0.58	4.49	6.89E-11	9.82E-09
Tdg-ps2	−3.61	0.44	7.95E-26	3.28E-23	Gm13340	0.59	7.06	5.86E-11	8.51E-09
Gm12751	−3.45	0.17	1.02E-20	3.25E-18	170004 7M11Rik	0.59	2.91	1.03E-05	5.91E-04
Mfrp	−3.34	0.49	4.10E-23	1.61E-20	mt-Ts1	0.59	2.93	1.68E-05	8.98E-04
Rpl15-ps6	−3.19	0.76	4.30E-26	1.82E-23	Ptgds	0.60	6.91	8.82E-12	1.48E-09
Gm8615	−3.17	−0.26	7.21E-16	1.67E-13	Ubb-ps	0.60	4.12	7.15E-10	9.17E-08
Rps3a3	−3.16	3.16	2.40E-105	7.32E-102	Gm31763	0.61	1.58	2.25E-03	4.64E-02
Sostdc1	−3.02	1.43	1.51E-33	8.53E-31	Gm12183	0.61	3.11	2.38E-06	1.59E-04
Tmem72	−3.01	1.07	1.07E-26	4.93E-24	6430511E19Rik	0.62	1.61	1.94E-03	4.17E-02
Ttr	−2.95	8.37	3.42E-175	5.22E-171	mt-Co2	0.62	5.57	1.54E-14	3.23E-12
Hmga1b	−2.95	1.98	2.87E-47	3.37E-44	Srrm4os	0.63	2.04	2.78E-04	9.08E-03
Kcne2	−2.95	0.46	3.16E-19	9.10E-17	Gm29216	0.63	5.27	2.32E-14	4.78E-12
Gm14719	−2.80	0.10	1.64E-15	3.57E-13	Gm6210	0.63	2.24	2.36E-04	7.86E-03
Aqp1	−2.74	1.18	1.05E-25	4.23E-23	1190007I07Rik	0.63	1.57	1.52E-03	3.53E-02
Cldn1	−2.56	−0.06	4.32E-12	7.67E-10	Gm10222	0.63	1.62	2.24E-03	4.63E-02
Gm7335	−2.50	0.16	4.14E-14	8.32E-12	Mir5125	0.64	1.90	4.78E-04	1.43E-02
Folr1	−2.44	1.18	9.57E-22	3.40E-19	Gm15427	0.64	1.74	1.23E-03	3.04E-02
Kl	−2.37	1.87	2.00E-30	1.02E-27	Dnah7b	0.64	2.35	6.77E-05	2.85E-03
Calml4	−2.37	−0.04	1.80E-11	2.83E-09	Trip6	0.64	1.91	4.18E-04	1.27E-02
Lamp5	−2.35	0.52	1.78E-15	3.83E-13	Col19a1	0.65	2.04	2.33E-04	7.79E-03
F5	−2.35	−0.14	1.01E-10	1.42E-08	Col27a1	0.65	3.17	1.60E-07	1.41E-05
Nrgn	−2.30	3.24	2.44E-67	4.65E-64	Lypla1	0.65	3.42	1.85E-08	1.89E-06
Tmem181c-ps	−2.29	3.04	3.01E-59	4.59E-56	Gria4	0.65	7.20	5.06E-13	9.41E-11
Gm15446	−2.20	2.66	3.51E-40	2.68E-37	Hba-a2	0.66	2.37	5.88E-05	2.56E-03
Car12	−2.12	1.19	1.30E-17	3.41E-15	Gm18807	0.66	1.46	1.59E-03	3.65E-02
Slc4a5	−2.02	1.48	3.11E-19	9.10E-17	Gm13436	0.66	1.48	2.38E-03	4.81E-02
Sst	−1.99	2.14	1.11E-27	5.28E-25	2610035D17Rik	0.67	2.26	6.03E-05	2.61E-03
Rps26-ps1	−1.97	1.17	1.23E-16	3.03E-14	Myo7a	0.67	2.64	6.65E-06	4.05E-04
Ecrg4	−1.97	2.44	1.22E-31	6.41E-29	H3f3a-ps1	0.67	1.30	2.27E-03	4.65E-02
Kcnj13	−1.89	0.68	1.95E-11	3.03E-09	Cyp4f14	0.67	1.55	9.63E-04	2.50E-02
Mndal	−1.88	2.97	1.20E-45	1.22E-42	Eef1a1-ps1	0.68	2.96	7.75E-07	5.72E-05
Aass	−1.77	−0.01	1.10E-07	9.75E-06	Myh6	0.69	1.43	8.96E-04	2.35E-02
P4ha3	−1.76	0.52	2.62E-09	3.05E-07	Gm4076	0.69	2.25	4.36E-05	2.06E-03
Ifi203-ps	−1.73	−0.50	6.24E-07	4.81E-05	Gm9616	0.69	1.54	1.31E-03	3.19E-02
Gm6166	−1.73	1.19	1.26E-13	2.46E-11	Fam228b	0.69	1.44	9.78E-04	2.53E-02
Eif3j2	−1.72	1.22	1.82E-13	3.52E-11	Gm7172	0.69	2.19	6.71E-05	2.84E-03
Ifi203	−1.72	3.17	1.38E-43	1.24E-40	Gm9843	0.69	1.51	1.15E-03	2.89E-02
Gm7993	−1.71	0.53	9.61E-10	1.20E-07	Eef2-ps2	0.70	1.71	5.11E-04	1.50E-02
Gm14292	−1.67	−0.44	5.83E-06	3.61E-04	Sycp3	0.70	1.83	2.22E-04	7.49E-03
CT010467.1	−1.64	8.60	4.72E-61	8.00E-58	Tpt1-ps3	0.70	3.08	9.06E-08	8.28E-06
Erdr1	−1.62	1.57	3.26E-14	6.64E-12	mt-Atp6	0.72	5.22	4.47E-18	1.24E-15
Rps7-ps3	−1.61	1.81	1.23E-16	3.03E-14	Gm9824	0.72	1.46	7.96E-04	2.17E-02
Atp6v0c-ps2	−1.60	3.76	2.86E-48	3.97E-45	Ttll3	0.72	3.52	1.59E-10	2.19E-08
Kcnc2	−1.50	1.01	1.25E-09	1.54E-07	Gm9493	0.73	1.24	1.45E-03	3.40E-02
Cldn2	−1.50	1.13	5.29E-10	6.85E-08	Metap1d	0.73	3.75	3.75E-12	6.78E-10
Thsd7b	−1.50	−0.48	4.41E-05	2.08E-03	Zfhx2	0.73	6.51	7.76E-18	2.11E-15
Col8a1	−1.49	1.30	6.41E-11	9.23E-09	Gm48239	0.73	1.07	1.69E-03	3.78E-02
Myo5b	−1.47	−0.01	5.56E-06	3.47E-04	Gabra2	0.74	4.51	1.03E-16	2.62E-14
Baiap3	−1.47	2.20	3.53E-17	9.13E-15	Gm12346	0.74	2.99	4.47E-08	4.32E-06
Rpl15-ps2	−1.47	0.69	3.45E-08	3.38E-06	Rpl39-ps	0.74	1.18	1.23E-03	3.03E-02
Npr3	−1.44	−0.45	5.47E-05	2.45E-03	Gm50432	0.75	1.66	2.39E-04	7.95E-03
Ndufa12-ps	−1.43	1.30	1.33E-10	1.84E-08	Rpl26	0.77	5.74	1.51E-21	5.14E-19
Prlr	−1.43	−0.29	6.61E-05	2.80E-03	Gask1b	0.77	2.40	1.51E-06	1.07E-04
Hs3st2	−1.42	0.44	6.24E-07	4.81E-05	Zfhx2os	0.77	3.72	4.88E-13	9.20E-11
Htr2c	−1.37	2.79	4.34E-21	1.44E-18	Eps8l1	0.78	1.94	3.00E-05	1.48E-03
Mid1	−1.36	1.38	4.13E-10	5.38E-08	Xlr3b	0.78	1.72	9.16E-05	3.66E-03
Rasgrf2	−1.36	1.75	3.78E-12	6.78E-10	Gm6789	0.78	2.53	5.92E-07	4.66E-05
Gm47283	−1.33	3.35	6.46E-28	3.18E-25	Pgam1-ps2	0.78	0.89	2.48E-03	4.95E-02
Pcna-ps2	−1.33	0.73	2.86E-07	2.43E-05	Tma7-ps	0.79	1.41	2.45E-04	8.12E-03
Rnps1-ps	−1.33	1.20	1.00E-08	1.10E-06	Paqr6	0.79	2.87	1.00E-08	1.10E-06
Slit1	−1.32	0.37	7.99E-06	4.69E-04	Gm13341	0.79	5.09	5.71E-21	1.85E-18
Ndst4	−1.30	1.21	1.43E-08	1.49E-06	Hapln2	0.79	2.70	5.91E-08	5.53E-06
Gm24187	−1.30	1.44	1.76E-09	2.12E-07	Gm10250	0.79	1.57	1.29E-04	4.84E-03
Mir6236	−1.25	1.44	4.80E-09	5.47E-07	Prrg1	0.80	1.32	2.96E-04	9.55E-03
Meis2	−1.25	1.61	2.01E-09	2.39E-07	Gpi-ps	0.80	1.66	6.93E-05	2.89E-03
Abca4	−1.25	1.00	3.55E-07	2.92E-05	mt-Nd4l	0.81	2.80	1.86E-08	1.89E-06
Enpp2	−1.24	7.25	2.26E-41	1.92E-38	Gm16185	0.81	2.12	2.19E-06	1.50E-04
Gm15459	−1.22	4.90	3.14E-47	3.43E-44	Olfml2a	0.81	0.85	1.21E-03	3.01E-02
Sertm1	−1.20	0.19	6.54E-05	2.78E-03	Ppp1ccb	0.81	0.76	1.92E-03	4.14E-02
Pla2g4e	−1.17	2.80	1.86E-16	4.50E-14	Gm42664	0.81	2.18	1.50E-06	1.06E-04
Wdr86	−1.16	0.30	1.09E-04	4.21E-03	Tubb4b-ps1	0.82	1.09	6.56E-04	1.84E-02
Col8a2	−1.16	1.43	9.71E-08	8.76E-06	AC163703.1	0.82	1.63	5.57E-05	2.46E-03
Msx1	−1.15	1.63	1.33E-08	1.41E-06	Gm21092	0.82	1.18	3.98E-04	1.23E-02
Gm14150	−1.15	2.20	3.52E-11	5.32E-09	2610507I01Rik	0.83	1.25	2.33E-04	7.79E-03
Lbp	−1.12	1.29	4.38E-07	3.52E-05	C4b	0.84	1.17	3.10E-04	9.91E-03
Tcf7l2	−1.12	2.07	1.67E-10	2.28E-08	Gm15421	0.84	0.69	1.83E-03	3.99E-02
Wfikkn2	−1.12	0.54	6.54E-05	2.78E-03	Gm6863	0.84	1.69	2.75E-05	1.37E-03
Trpv4	−1.09	0.21	4.84E-04	1.44E-02	Csf2ra	0.84	3.11	2.61E-11	3.98E-09
Col25a1	−1.06	0.21	4.81E-04	1.44E-02	Rps16-ps2	0.85	1.24	1.93E-04	6.69E-03
Gm3375	−1.06	1.60	2.61E-07	2.23E-05	Nox1	0.85	1.42	5.67E-05	2.49E-03
Elfn1	−1.05	0.32	3.79E-04	1.18E-02	Malat1	0.85	7.04	1.11E-21	3.84E-19
Chrna4	−1.05	1.70	4.85E-08	4.63E-06	Gm9385	0.86	1.30	1.55E-04	5.63E-03
Mrpl27-ps	−1.04	0.44	1.06E-04	4.09E-03	AC160336.1	0.86	1.14	2.77E-04	9.06E-03
Gm5900	−1.04	−0.28	2.17E-03	4.53E-02	Gm12254	0.86	0.64	1.90E-03	4.13E-02
Gm23935	−1.03	6.32	1.90E-34	1.11E-31	Gm2614	0.86	0.81	1.02E-03	2.61E-02
Gm50209	−1.03	0.15	7.42E-04	2.06E-02	Gm43759	0.87	0.58	1.17E-03	2.93E-02
Tox2	−1.02	0.79	7.71E-05	3.18E-03	Gm38393	0.88	2.53	1.06E-08	1.15E-06
Arg1	−1.02	0.02	1.63E-03	3.71E-02	Gm17066	0.88	0.69	1.05E-03	2.68E-02
Gm29650	−1.02	2.02	1.50E-08	1.55E-06	Gm4828	0.89	1.34	8.46E-05	3.44E-03
Grem2	−1.01	0.19	7.54E-04	2.09E-02	Ctxn3	0.89	0.87	2.99E-04	9.60E-03
Sema3b	−1.01	0.67	1.26E-04	4.76E-03	Gm8203	0.89	0.56	1.25E-03	3.07E-02
Npnt	−1.00	2.17	3.89E-09	4.47E-07	Gm28439	0.90	4.12	5.35E-20	1.63E-17
Gm24270	−1.00	2.09	1.34E-08	1.41E-06	Gm12669	0.90	0.66	8.86E-04	2.33E-02
Slc35f4	−0.99	0.79	1.31E-04	4.88E-03	Rec8	0.91	3.66	9.30E-18	2.49E-15
Rps15a-ps7	−0.99	1.16	3.41E-05	1.68E-03	Gm11478	0.91	1.41	3.46E-05	1.70E-03
Pcdha11	−0.97	0.18	1.25E-03	3.07E-02	Gm43843	0.91	0.38	1.73E-03	3.83E-02
Hhipl1	−0.97	0.63	2.96E-04	9.55E-03	Gm37486	0.92	0.52	9.94E-04	2.57E-02
H2-Bl	−0.97	0.73	2.15E-04	7.27E-03	Gm10819	0.92	0.73	5.72E-04	1.66E-02
Camkv	−0.96	1.89	2.95E-07	2.48E-05	Gm6485	0.92	1.16	9.34E-05	3.69E-03
Eif2s3y	−0.96	5.70	1.08E-32	5.91E-30	Atp5pb-ps	0.93	1.16	6.36E-05	2.72E-03
Gm36445	−0.96	0.37	9.03E-04	2.35E-02	Ybx1-ps2	0.93	0.22	2.31E-03	4.72E-02
AA465934	−0.95	2.62	3.59E-11	5.37E-09	Gm8717	0.95	0.30	1.88E-03	4.10E-02
BC022960	−0.95	1.85	7.45E-07	5.55E-05	Gm8451	0.95	0.74	3.75E-04	1.17E-02
Igfbp2	−0.94	4.59	1.24E-26	5.55E-24	Hspd1-ps3	0.96	1.03	9.95E-05	3.87E-03
Gm11868	−0.94	0.57	5.66E-04	1.65E-02	Gm16238	0.97	0.49	8.85E-04	2.33E-02
Gchfr	−0.93	0.96	1.73E-04	6.16E-03	Rpl18-ps2	0.97	0.10	2.35E-03	4.78E-02
Zfp729a	−0.92	3.41	4.68E-15	9.93E-13	Gm12966	0.98	0.51	4.89E-04	1.45E-02
Vgf	−0.91	3.51	1.21E-15	2.71E-13	Gm2962	0.99	0.40	6.76E-04	1.89E-02
Ngef	−0.91	1.71	2.79E-06	1.84E-04	Gm6627	0.99	0.65	1.97E-04	6.75E-03
Zkscan16	−0.91	0.76	3.74E-04	1.17E-02	Gm20900	1.00	0.65	2.74E-04	9.02E-03
Uty	−0.91	4.25	3.12E-22	1.14E-19	Gm6192	1.00	1.13	3.84E-05	1.87E-03
Kdm5d	−0.90	4.85	1.68E-26	7.34E-24	Etnppl	1.00	1.03	2.45E-05	1.24E-03
Zcchc12	−0.89	2.25	1.05E-07	9.34E-06	mt-Nd3	1.00	2.31	2.21E-09	2.59E-07
Hccs	−0.89	4.37	2.96E-22	1.10E-19	Gm37310	1.00	−0.05	2.32E-03	4.74E-02
Fam163b	−0.89	1.42	4.41E-05	2.08E-03	Gm11407	1.02	1.73	4.03E-07	3.28E-05
Pcdh8	−0.89	1.24	8.83E-05	3.54E-03	Gm12696	1.02	0.06	1.64E-03	3.71E-02
Fmn1	−0.87	1.45	5.27E-05	2.38E-03	Gm10073	1.03	0.86	6.07E-05	2.62E-03
Dynlt1b	−0.85	3.13	1.66E-11	2.64E-09	Gm5087	1.03	0.53	1.42E-04	5.26E-03
Bc1	−0.85	1.45	7.01E-05	2.91E-03	H2-D1	1.04	5.10	1.42E-35	8.69E-33
Folh1	−0.85	0.78	1.10E-03	2.78E-02	Wsb2-ps	1.05	−0.06	2.13E-03	4.47E-02
Tuba1c	−0.84	3.99	2.43E-16	5.79E-14	Gm7308	1.05	1.19	8.04E-06	4.70E-04
Rgs11	−0.83	1.30	1.61E-04	5.81E-03	Actr3-ps	1.05	0.59	1.81E-04	6.37E-03
Prdm16	−0.82	0.83	1.53E-03	3.53E-02	Gm49405	1.05	−0.10	1.39E-03	3.31E-02
Plvap	−0.82	2.08	2.39E-06	1.59E-04	AC149090.1	1.07	5.55	6.97E-41	5.60E-38
Tmem54	−0.82	1.04	5.66E-04	1.65E-02	Gm5436	1.07	0.55	1.33E-04	4.97E-03
Rps3a2	−0.82	1.34	1.55E-04	5.63E-03	Gm2830	1.08	0.57	1.09E-04	4.20E-03
Clic6	−0.81	2.50	2.33E-07	2.00E-05	Rps4x-ps	1.09	−0.21	1.67E-03	3.75E-02
Tpbgl	−0.81	1.12	4.91E-04	1.46E-02	Gm3531	1.10	1.26	2.09E-06	1.44E-04
Tafa2	−0.81	0.89	1.00E-03	2.58E-02	Arhgap6	1.12	0.06	3.91E-04	1.21E-02
Gm8210	−0.80	1.26	3.22E-04	1.02E-02	Gm9774	1.13	0.35	1.42E-04	5.26E-03
Nptx2	−0.78	1.59	1.43E-04	5.26E-03	Gm10654	1.15	0.06	3.01E-04	9.62E-03
Slc2a12	−0.77	2.06	1.19E-05	6.79E-04	Rpl7-ps7	1.16	0.09	3.14E-04	9.97E-03
Slc6a11	−0.75	5.18	4.42E-20	1.38E-17	Gm9794	1.17	1.24	4.38E-07	3.52E-05
Ifi30	−0.75	1.62	1.97E-04	6.75E-03	Gm28437	1.17	5.95	2.95E-45	2.81E-42
Hs3st4	−0.73	1.68	1.66E-04	5.95E-03	Gm5644	1.17	−0.20	6.86E-04	1.91E-02
Enc1	−0.73	3.93	3.19E-13	6.09E-11	Gm37630	1.18	−0.15	8.46E-04	2.28E-02
Tpbg	−0.72	2.36	7.17E-06	4.28E-04	H2-T24	1.18	1.20	1.92E-07	1.68E-05
Resp18	−0.72	2.23	1.51E-05	8.20E-04	Gm5846	1.19	0.13	2.12E-04	7.22E-03
Lars2	−0.71	5.84	1.49E-18	4.20E-16	Gm10053	1.19	1.09	9.56E-07	6.85E-05
Uaca	−0.70	2.51	7.51E-06	4.46E-04	Cox6c2	1.20	0.32	9.19E-05	3.66E-03
Mab21l2	−0.69	1.52	5.92E-04	1.70E-02	mt-Co3	1.20	5.98	2.11E-47	2.68E-44
Slit2	−0.69	1.48	9.77E-04	2.53E-02	Gm13292	1.21	−0.28	6.48E-04	1.83E-02
Chrm2	−0.69	1.44	8.96E-04	2.35E-02	Gm14303	1.22	0.67	7.71E-06	4.54E-04
Fam81a	−0.69	2.31	2.18E-05	1.12E-03	Gm13339	1.24	1.85	2.64E-10	3.57E-08
4933404O12Rik	−0.69	2.00	1.13E-04	4.33E-03	Gm5526	1.24	0.49	1.57E-05	8.49E-04
Rpl3-ps1	−0.69	2.65	2.43E-06	1.61E-04	Gm2223	1.26	−0.09	2.01E-04	6.86E-03
Tshz2	−0.68	2.31	2.29E-05	1.17E-03	Gm16418	1.27	0.36	1.99E-05	1.03E-03
Il33	−0.68	1.85	2.20E-04	7.44E-03	Tubb4b-ps2	1.27	−0.11	2.80E-04	9.12E-03
Pcp4l1	−0.68	4.02	5.81E-12	1.01E-09	Gm10224	1.27	0.01	1.17E-04	4.45E-03
Pard6b	−0.68	1.51	1.03E-03	2.63E-02	Gm9392	1.28	0.14	4.73E-05	2.19E-03
Gpc4	−0.68	1.37	1.41E-03	3.35E-02	Gm6030	1.31	−0.07	8.70E-05	3.50E-03
Thrb	−0.67	2.58	8.99E-06	5.21E-04	Gm12481	1.31	−0.02	7.86E-05	3.23E-03
Scg5	−0.66	6.11	1.40E-15	3.10E-13	Gm6252	1.33	0.07	5.51E-05	2.45E-03
Zswim4	−0.65	4.16	1.01E-11	1.66E-09	Gm4708	1.39	−0.31	9.78E-05	3.82E-03
Tceal7	−0.65	2.30	6.13E-05	2.63E-03	Sspo	1.40	−0.20	5.43E-05	2.44E-03
Ide	−0.64	5.33	9.85E-16	2.24E-13	Gm27016	1.41	0.06	1.76E-05	9.32E-04
Cd82	−0.64	2.37	6.93E-05	2.89E-03	H2-K2	1.45	−0.10	1.82E-05	9.55E-04
Rcn1	−0.64	3.50	2.06E-08	2.08E-06	Cpsf4l	1.45	0.63	1.03E-07	9.23E-06
Sema3f	−0.64	1.46	2.34E-03	4.75E-02	Rpl9-ps11	1.53	0.26	9.44E-07	6.83E-05
Sulf1	−0.63	2.05	4.07E-04	1.25E-02	A330076C08Rik	1.54	0.20	6.90E-07	5.21E-05
Igsf1	−0.61	2.44	9.71E-05	3.80E-03	Gm5805	1.56	−0.07	7.06E-06	4.23E-04
Nptxr	−0.61	7.02	5.24E-12	9.19E-10	Gm28438	1.66	1.70	4.90E-16	1.15E-13
Ddx3y	−0.61	5.31	6.13E-14	1.21E-11	Gm36989	1.73	−0.20	6.78E-07	5.17E-05
Hbegf	−0.61	4.42	2.08E-11	3.21E-09	Muc3a	1.81	2.90	1.27E-36	8.10E-34
Gm50394	−0.60	3.53	7.13E-08	6.56E-06	Rpsa-ps12	1.88	−0.03	6.43E-08	5.98E-06
Ngfr	−0.60	2.18	4.32E-04	1.31E-02	mt-Atp8	2.10	2.59	2.69E-37	1.87E-34
Egr1	−0.60	2.91	1.40E-05	7.68E-04	Gm27000	2.19	0.32	8.46E-12	1.43E-09
Tmem179	−0.59	2.92	1.36E-05	7.51E-04	S100a9	2.30	−0.52	7.71E-09	8.58E-07
Chrm3	−0.57	1.95	1.76E-03	3.88E-02	Gm26384	2.56	−0.46	2.98E-10	3.95E-08
					Gm15682	2.84	0.61	2.61E-19	7.82E-17

Genes reported were detected at false discovery rate (FDR) 0.05 and absolute value of fold change (FC) > 1.5. MGI Gene name, log fold change (log FC), log counts per million reads (log CPM), P values and Benjamani-Hochberg corrected false discovery rates (FDR) are listed for each gene.

Table 2.

Functional categories enriched with DEGs in Galnt17^−/− P14 cerebellum.

Down-regulated category	-logPfdr	Example genes
response to growth factor	>10	Arg1,Egr1,Folr1,Slit2,Ngfr,Sulf1,Tcf7l2
transmembrane receptor protein serine/threonine kinase signaling pathway	5.8	Egr1,Folr1,Grem2,Igsf1,Msx1,Npnt,Sostdc1,Tcf7l2, Wfikkn2,Zcchc12
heparan sulfate proteoglycan metabolic process	5.7	Hs3st2,Hs3st4,Ndst4,Sulf1,Tcf7l2
neurotransmitter binding	5.3	Chrm2,Chrm3,Chrna4,Htr2c,Slc6a1
regulation of axon guidance	5.2	Sema3b, Sema3f, Slit1, Slit2,Zswim4
hormone binding	5.2	Chrm2,Chrm3,Chrna4,Ide,Prlr,Thrb,Ttr
modulation of chemical synaptic transmission	5.1	Baiap3,Chrm2,Chrna4,Egr1,Htr2c,Myob,Ngfr,Nptx2,Nptxr,Nrgn,Rasgrf2,Vgf
synaptic transmission, cholinergic	5.2	Chrm2,Chrm3,Chrna4,Htr2c,Ngfr
regulation of synapse organization	4.8	Camkv,Gpc4,Myo5b,Ngef,Nptxr,Pcdh8,Sema3f, Slit1,Tpbg
negative regulation of axon guidance	4.7	Sema3b,Sema3f,Slit1,Slit2
Up-regulated category		Example genes
positive regulation of intrinsic apoptotic signaling pathway	5.0	Bmf,Nox1,Rpl26,S100a9
regulation of integrin biosynthetic process	4.5	Nox1,S100a9

Category enrichment and -log p fdr values were determined using the ToppCluster tool(Kaimal et al., 2010). The analysis was conducted with all DEGs reported in Table 1. Only the most highly significant and non-redundant gene ontology (GO) and Pathway categories detected by ToppCluster are shown

More specifically, the down-regulated DEGs selected for higher levels of FC between mutants and WT ≥1.5 FC) were very highly enriched in functional categories related to neuron differentiation and nervous system development, especially terms related to axon guidance, cholinergic signaling, and synaptic organization (Table 2). These enrichments were accompanied by the down-regulation of genes associated with heparan sulfate (HS) biosynthesis, which is central to axon guidance and synapse development (Condomitti and de Wit, 2018; Maeda et al., 2011). Hormone signaling and neurotransmitter binding were also highlighted as being enriched in down-regulated genes. Genes involved in cerebellar vermis development including Mid1 and Folr1 (Lancioni et al., 2010; Ohba et al., 2013) were also included as prominent DEGs, along with genes with key roles more generally in cerebellar development, including neuropeptide gene, Vgf (Mizoguchi et al., 2019)and the beta-subunit of thyroid hormone receptor Thrb. Thrb is notable because hypothyroidism leads to stunted growth and branching of PC dendrites as well as the delayed maturation of GCs, including delayed migration of GCs from the ECL (Koibuchi, 2013). More generally, down-regulation of Ser-Thr protein kinase and growth factor signaling pathways was also indicated. On the other hand, the DEGs up-regulated in Galnt17^−/− cerebellum were enriched in categories related to apoptotic signaling and integrin biosynthesis.

To test predictions derived from the transcriptomic analysis, we first used the TUNEL assay in sections from WT and mutant vermis. The results confirmed the presence of increased numbers of apoptotic GCs in P14 Galnt17^−/− EGL compared to age-matched WT animals, specifically in vermal lobules VI and IX (Figure 8A; VI: t(5)=−2.95, p<.05, IX: t(11)=−2.65, p<.05) - the same two lobules that showed increased numbers of dividing cells (Figure 6B) and ECL thickness (Figure 6C). Additionally, we confirmed the prediction of reduced HS synthesis in the mutants by measuring HS levels in WT and Galnt17^−/− P14 proteins in a western blot (Figure 8B). We separated nuclei from all other cellular components together, including cytoplasm, membrane fractions, and organelles, before protein preparation to enrich each fraction and analyze them independently. The results confirmed significantly decreased expression of HS proteoglycans in the non-nuclear fraction of cerebellar protein lysates in Galnt17^−/− compared to WT mice (measured as Mean±SE = 0.55±0.17 of WT levels in replicate samples taken from individual animals; F1,3=12.60, p<.05).

Figure 8. Confirming predicted functional consequences of Galnt17 mutation in the cerebellum.

(A) RNA-seq data suggested increased levels of apoptosis in P14 Galnt17^−/− cerebellum compared to P14 WT mice. TUNEL Assays confirmed this prediction and revealed that apoptotic cells were increased specifically within the EGL of vermal lobules VI and IX. The graph shows the ratio of TUNEL-positive cells in mutant (Mut) compared to wild type (WT) in a series of stained slides from 2–4 animals of each genotype, as described in Methods. Standard error for Mut/WT was calculated using following formula: SE_ratio = Mut/WT × [(SE_mut/mean Mut)+(SE_WT/mean WT)]^1/2, and statistical analysis was conducted using actual data points from WT and mutant animals. t-Test, *p<0.05. (B) Western blots confirmed that Galnt17^−/− animals had decreased expression of heparan sulfate (HS) in cerebellar non-nuclear proteins compared to WT mice. The low intensity bands detected in the wild-type and Galnt17^−/− mouse nuclear fractions may be non-specific due to the low cross-reactivity of the anti-HS antibody to DNA. In both panels B and C, loading control for non-nuclear protein was Actin, and Lamin-B1 was the loading control for nuclear proteins. (C) Decreased levels of O-GalNAcylation were identified in specific Galnt17^−/− non-nuclear and nuclear cerebellar protein bands using a VVA lectin blot. One nuclear and one non-nuclear protein band, each marked with an asterisk, were excised from gels and sent for LC-MS analysis as described in the text.

GALNT17 loss of function leads to abnormal O-glycosylation profiles in mutant cerebellum

Our next goal was to provide initial clues to the mechanisms through which the loss of Galnt17 function could generate this cerebellar phenotype. Because GALNT17 is a GalNAc-T, abnormal glycosylation of specific proteins, either directly as enzyme substrates or secondarily, through chaperone activity, should provide the primary driver of both the molecular and brain-pathology phenotypes. To determine whether Galnt17 LOF leads to abnormal patterns of mucin-type O-glycosylation, we used the VVA lectin (which specifically binds to the O-GalNAc attachment to Ser or Thr on target proteins) to probe western blots; the results identified protein bands with significantly reduced VVA signal in the mutant compared to WT cerebellar samples (Figure 8C). Mucin-type O-glycosylation is typically associated with membrane-associated and secreted proteins, although GALNT17 has been reported to display low activity toward typical mucin-type target proteins and may have strict and limited or unusual target specificity (Nakamura et al., 2005). In addition, although GALNT17 targets would be expected to be membrane-associated or secreted, published reports have identified nuclear proteins as targets of certain GalNAc-T proteins (Cejas et al., 2019; Deng et al., 2018).

We therefore used standard nuclear isolation procedures to collect two protein fractions from mutant and WT cerebellum, as described above. Somewhat to our surprise, we identified distinctive patterns of VVA lectin-positive (VVA+) proteins in both the nuclear and non-nuclear fractions. Furthermore, there were specific bands in each fraction that were VVA+ in WT but not in the mutant samples (Figure 8C). For example, in replicate samples, a nuclear fraction protein band of ~85 kDa was detected with VVA lectin in both mutant and WT nuclear fractions but was reduced in mutants to approximately half the intensity of WT controls (Mean±SE = 0.49±0.26; F1,2=20.75, p<.05). Although the protocol we used enriches for nuclear proteins, these preparations can be contaminated with ER proteins (Kato and Kurokawa, 1967), and we cannot rule out the possibility that this putative GALNT17 target protein is an ER protein instead. A ~75 kDa protein band in the non-nuclear fraction was similarly reduced in the mutants compared to WT (Figure 8C). These differentially detected proteins represented potential candidates as glycosylation targets of GALNT17.

We reasoned that only the most abundant glycoproteins could likely produce such significant loss in signal intensity. However, the most abundant O-glycosylated proteins might still be relatively minor components of those protein size-fractions overall. Indeed, mass spectrometric analysis of proteins extracted from VVA differentially stained protein bands identified only very abundant proteins (Synapsin I, ~75 kDa band in the non-nuclear fraction; and HSP90A/B from ~85kDa band in the nuclear fraction (Table 3)) that are unlikely GalNAc-T targets (not shown). Therefore, substantial additional work will be required to explicate the identities of the direct or indirect substrates of GALNT17.

Table 3.

Top scoring protein matches for proteins found within the cytoplasmic ~75 kDa and nuclear ~85 kDa protein bands detected as differentially glycosylated in VVA lectin blots of Galnt17^−/− and wild type cerebellar proteins.

SwissProt Accession	Score	Number of significant matches	Number of significant sequences	emPAIa	Description/Gene name
Non-nuclear 75kDa- Wild type
SYN1_MOUSE	515	23	10	0.78	Synapsin-1/ Syn1
K2C1_MOUSE	201	6	2	0.14	Keratin/ Krt1
K1C10_MOUSE	183	5	4	0.34	Keratin/ Krt10
Non-nuclear 75kDa- Galnt17−/−
SYN1_MOUSE	922	40	14	1.22	Synapsin-1/ Sny1
DPYL1_MOUSE	416	14	6	0.72	Dihydropyrimidinase-related protein 1/ Crmp1
K1C10_MOUSE	365	10	5	0.44	Keratin/ Krt10
Nuclear 85kDa- Wild type
HS90B_MOUSE	2929	85	23	3.36	Heat shock protein HSP 90-beta/ Hsp90ab1
HS90A_MOUSE	2755	72	22	2.48	Heat shock protein HSP 90-alpha/ Hsp90aa1
ACON_MOUSE	738	31	15	1.10	Aconitate hydratase/ Acon
Nuclear 85kDa- Galnt17−/−
HS90B_MOUSE	3727	109	35	7.46	Heat shock protein HSP 90-beta/ Hsp90ab1
HS90A_MOUSE	3031	90	28	3.71	Heat shock protein HSP 90-alpha/ Hsp90aa1
ACON_MOUSE	818	30	16	1.32	Aconitate hydratase/ Acon

Output from the Mascot 2.7 analysis tool, with SwissProt matches ranked by score and cutoff of false discovery rate ≤ 0.01.

^a._emPAI = 10^PAI − 1, which is an approximation for relative protein abundance in a mixture (Ishihama et al., 2005).

DISCUSSION

As the direct downstream neighbor of the AUTS2 locus in both humans and mice, GALNT17 is likely included in the long-range regulatory relationship that includes AUTS2 and WBS genes (Gheldof et al., 2013); this idea is supported by observations that Galnt17 and Auts2 are co-expressed in a wide variety of cell types in the hippocampus and cerebellum throughout postnatal life (Weisner et al., 2019). These and other findings raised the possibility that information regarding the in vivo function of Galnt17 in developing brain could be directly relevant to certain WBS or AUTS2 syndrome phenotypes, and possibly other neurodevelopmental disorders as well.

Here we show that Galnt17 LOF has significant effects on cerebellar development, resulting in developmental delay and behavioral deficits. The fully penetrant expression of these WBS and AUTS2 syndrome-related phenotypes makes the Galnt17^−/− mutant mouse a useful tool for examining the molecular mechanisms. Curiously, although Galnt17 is expressed throughout the cerebellum, LOF for this gene had a dramatic effect on the maturation of the cerebellar vermis, a brain region important to motor coordination, social behaviors and emotional processing (Al-Afif et al., 2013; Koziol et al., 2014; Turner et al., 2007). In Galnt17^−/− adults, we observed decreased thickness of the GL in certain vermal lobules in adults, together with abnormal development of vermal BG and PC abnormalities including reduced dendritic growth, axon defasciculation, and a narrower PC-derived axonal tract in those same lobules. While different genetic backgrounds could show more widespread pathology, the data indicates that Galnt17 is particularly important to development of this vermal region.

These cellular pathologies were coordinated with significant changes in gene expression in the Galnt17^−/− cerebellum at P14, a critical age for cerebellar development (Rahimi-Balaei et al., 2018). The gene expression changes revealed the coordinated disturbance of genes and signaling pathways controlling cerebellar development (Lancioni et al., 2010; Mizoguchi et al., 2019; Ohba et al., 2013), coupled with a robust down-regulation of genes encoding enzymes involved in the synthesis of the GAG, HS. HS plays crucial roles in brain development, including neural stem cell proliferation, differentiation, cell migration, axon pathfinding, synaptogenesis and plasticity (Condomitti and de Wit, 2018; Hu, 2001; Kantor et al., 2004; Maeda et al., 2011), and HS dysregulation leads to abnormal brain morphology and plasticity, with clear downstream effects on behavior (Mizumoto et al., 2014). There are several sulfotransferases that determine the sulfation patterns of HS; these sulfotransferases are expressed in PCs and GCs at different developmental time points to synthesize GAGs that recognize different partner proteins (Yabe et al., 2005). Genes encoding HS sulfotransferases (Hs3st2, Hs3st4, and Ndst4) as well as HS binding proteins (e.g. Slit1 and Slit2) were down-regulated in Galnt17^−/− cerebellum, indicating a substantial reduction of HS-mediated pathways. Importantly, Slit protein functions include self-avoidance of PC dendrites (Gibson et al., 2014), migration of GCs (Ypsilanti et al., 2010), and axon pathfinding and fasciculation (Jaworski and Tessier-Lavigne, 2012; Yeo et al., 2004), all of which were identified as abnormal in Galnt17^−/− mice. Loss of HS synthesis and signaling by HS binding partners could thus underlie several aspects of the Galnt17^−/− vermal pathology, disrupting development of both PCs and GCs. The initial driver of these molecular, brain, and behavioral phenotypes should be the failure to O-glycosylate specific GALNT17 substrate proteins or possibly, the substrates of GalNAc-Ts regulated by GALNT17 chaperone activities. Mucin type O-glycosylation influences the stability, processing, and functions of many types of proteins, including secreted components of signaling pathways and their receptors (van den Steen et al., 1998) and proteins involved in synaptic structure and organization (Itoh and Nishihara, 2021). Indeed, VVA lectin profiles presented here suggested that loss of GALNT17 protein leads to significant defects in O-glycosylation of some relatively abundant cerebellar proteins, some of which may be involved in regulation of HS synthesis, neurite outgrowth and synpase organization as well. A full understanding of Galnt17^−/− mutant phenotypes will ultimately depend upon identifying the protein’s substrates or its chaperone clients in developing brain, a quest that will be the subject of future work.

The multitude of cell types that are impacted by Galnt17^−/− LOF in the cerebellum leaves other important questions for further inquiry. Specifically, GCs, BG, and PCs are all affected in Galnt17 mutants and the GALNT17 protein is expressed in each. However, these cell types are interdependent, and whether the glycosylation defects are centered in one of those cell types or occur independently and cell-intrinsically for combined effect is a lingering question. A combination of single-cell sequencing methods and Cre-Lox technologies will allow us to address this question in future studies. Nevertheless, the present study is the first to implicate Galnt17 for a role in development of the cerebellar vermis, with impact on motor coordination as well as motivational and social behaviors. Together with our recent findings, the data are also consistent with the idea that Galnt17 may contribute to phenotypes associated with the co-regulated Williams-Beuren and AUTS2 syndromes and suggest that its functions could be relevant to other neurodevelopmental disorders as well.

Highlights.

• Galnt17 knockout mice display behavioral deficits and cerebellar vermis pathology

• Mutants show delayed granule cell migration and abnormal Purkinje cell processes

• RNA-seq indicated primary defects in neurite outgrowth and synaptic functions

• Data showed reduced mucin-type O-glycosylation and heparan sulfate in mutants

• The data identify a novel role in neurite outgrowth and synaptogenesis