Frameshift mutations of YPEL3 alter sensory circuit function in Drosophila

A frameshift mutation in Yippee-like (YPEL) 3 was recently found from a rare human disorder with peripheral neurological conditions including hypotonia and areflexia. The YPEL gene family is highly conserved from yeast to human, but their functions are poorly defined. Moreover, the pathogenicity of the human YPEL3 variant is completely unknown. To tackle these issues, we generated a Drosophila model of human YPEL3 variant by CRISPR-mediated In-del mutagenesis. Gene-trap analysis suggests that Drosophila YPEL3 (dYPEL3) is predominantly expressed in subsets of neurons, including nociceptors. Analysis on chemical nociception induced by allyl-isothiocyanate (AITC), a natural chemical stimulant, revealed a reduced nociceptive response in dYPEL3 mutants. Subsequent circuit analysis showed a reduction in the activation of second-order neurons (SONs) in the pathway without affecting nociceptor activation upon AITC treatment. Although the gross axonal and dendritic development of nociceptors was not affected, the synaptic contact between nociceptors and SONs were decreased by dYPEL3 mutations. Together, these suggest that the frameshift mutation in human YPEL3 causes neurological conditions by weakening synaptic connection through presynaptic mechanisms.

Moreover, the pathogenicity of the human YPEL3 variant is completely unknown. To tackle 23 these issues, we generated a Drosophila model of human YPEL3 variant by CRISPR-24 mediated In-del mutagenesis. Gene-trap analysis suggests that Drosophila YPEL3 25 (dYPEL3) is predominantly expressed in subsets of neurons, including nociceptors. Analysis 26 on chemical nociception induced by allyl-isothiocyanate (AITC), a natural chemical stimulant, 27 revealed a reduced nociceptive response in dYPEL3 mutants. Subsequent circuit analysis 28 showed a reduction in the activation of second-order neurons (SONs) in the pathway without 29 affecting nociceptor activation upon AITC treatment. Although the gross axonal and dendritic 30 development of nociceptors was not affected, the synaptic contact between nociceptors and 31 SONs were decreased by dYPEL3 mutations. Together, these suggest that the frameshift 32 mutation in human YPEL3 causes neurological conditions by weakening synaptic connection 33 through presynaptic mechanisms. 34

INTRODUCTION 35
YPEL3 belongs to the Yippee gene family that is composed of a number of genes present in 36 various eukaryotic species ranging from yeast to human (Hosono et al., 2004), which 37 suggests that they are involved in fundamental biological processes. However, only a 38 handful of studies have hinted at the biological roles of YPEL3. YPEL3 was initially identified 39 as a small unstable apoptotic protein because of its low protein stability and the ability to 40 induce apoptosis when overexpressed in a myeloid cell line (Baker, 2003). Subsequent 41 studies implicate YPEL3 as a tumor suppressor. YPEL3 expression correlates with p53 42 activity (Kelley et al., 2010). Overexpression and knockdown analyses suggest that YPEL3 43 suppresses the epithelial-to-mesenchymal transition in cancer cell lines by increasing 44 GSK3β expression (Zhang et al., 2016). Other studies have shown the role of YPEL genes 45 in development. The loss-of-function mutations of YPEL orthologs in ascomycete fungus 46 altered fungal conidiation and appressoria development (Han et al., 2018). In zebrafish, a 47 morpholino-mediated targeting of YPEL3 altered brain structures (Blaker-Lee et al., 2012). 48 Recently, a mutation in human YPEL3 was found in a patient with a rare disorder that 49 manifests a number of neurological symptoms (the NIH-Undiagnosed Diseases Program). 50 The mutation was caused by a duplication of a nucleotide in a coding exon of human 51 YPEL3, resulting in a frameshift and consequently a premature stop codon. The clinical 52 observation showed that the patient had normal cognition but manifested peripheral 53 symptoms, including areflexia and hypotonia. While these findings indicate significant 54 functions of YPEL3 in the peripheral nervous system, little is known about YPEL3's functions 55 in the nervous system. Furthermore, the pathogenicity of the identified YPEL3 mutation in 56 the nervous system is completely unknown. 57 In the present study, we generated a Drosophila model of the human condition caused by 58 the disease-relevant YPEL3 variant using CRISPR/CAS9-mediated in-del mutations. Our 59 gene-trap analysis suggests that subsets of neurons, including nociceptors, express the 60 Drosophila homolog of YPEL3 (dYPEL3). Subsequent analysis revealed reduced 61 nociceptive behavior in dYPEL3 mutants. Consistently, we found that dYPEL3 mutations 62 impaired the activation of second-order neurons (SONs) in the nociceptive pathway and 63 reduced the synaptic contact between nociceptors and these SONs. These findings suggest 64 that the identified human YPEL3 mutation presents its pathogenicity at neuronal synapses. 65

Generation of a disease-relevant variant of YPEL3 in Drosophila 67
Athough the discovery of a YPEL3 variant in a patient underscores the importance of YPEL3 68 in human health, whether this variant causes any defects in the nervous system is unknown. 69 There are five YPEL genes in human. YPEL1, 2, 3, and 4 are highly homologous to each 70 other (up to 96% identity at amino acid sequences), while YPEL5 has only ~40% homology 71 to the other members (Hosono et al., 2004). We found two YPEL homologs in Drosophila, 72 Yippee and CG15309, using an ortholog search (Hu et al., 2011). The predicted amino acid 73 sequences of CG15309 showed a 88% similarity (81% identity) to human YPEL3 ( Figure  74 1A), while that of Yippee showed a 65% similarity (53% identity) (Data not shown). Yippee 75 appears to be an ortholog of YPEL5 because it is more closely related to YPEL5 than 76 YPEL3 with 87% similarity and 73% identity to YPEL5 (Data not shown). Therefore, we 77 named CG15309 as dYPEL3. 78 The variant identified in the human patient introduces an extra nucleotide in the 79 middle of the coding exon, which produces a frameshift and consequently results in the 80 incorporation of the 37 ectopic amino acids followed by a premature stop codon ( Figure  81 1B). To generate a Drosophila model of the human variant, we took advantage of the 82 CRISPR/CAS9 technology to induce In-del mutations (Port et al., 2014). The entire coding 83 sequence of dYPEL3 is in a single exon. We designed a guide RNA that targets the middle 84 of the coding exon ( Figure 1C, top) and successfully isolated two dYPEL3 frameshift 85 mutants named dYPEL3 T1-6 and dYPEL3 T1-8 (Figure 1C, middle). dYPEL3 T1-6 has a 2-86 nucleotides deletion at 121 nucleotides downstream of a start codon, which generated a 87 premature stop codon at 153 downstream of start codon, while dYPEL3 T1-8 carries a 4-88 nucleotides deletion at 118 and generated a premature stop codon at 145 downstream of a 89 start codon. Similar to the human variant, the mutations introduced additional amino acids 90 followed by a premature stop codon ( Figure 1C, middle). The ectopic amino acids in 91 dYPEL3 T1-6 closely resemble those of the human variant ( Figure 1C, bottom panel). 92 93 dYPEL3 is expressed in subsets of neurons 94 We did not find any gross developmental defects in dYPEL3 T1-6 or dYPEL3 T1-8 flies.

95
Homozygotes were viable and fertile, and showed normal growth under standard culture 96 condition (data not shown). This raises the possibility that dYPEL3 is expressed in a subset 97 of cells in the body. Our efforts of generating antibodies against dYPEL3 failed in two 98 independent trials, precluding the use of immunostaining for identifying the cell types that 99 express dYPEL3. We thus took advantage of a GAL4 enhancer-trap line, CG15309-GAL4 100 (dYPEL3-GAL4) (Gohl et al., 2011), to study the expression pattern of dYPEL3 in flies. This 101 line contains a GAL4 insertion in the first intron of dYPEL3, which places the GAL4 under the 102 control of the endogenous dYPEL3 promoter and enhancers (Figure 2A, top). As a result, 103 the expression pattern of GAL4 represents that of dYPEL3. We expressed a membrane GFP 104 reporter (mouse CD8::GFP or mCD8::GFP) to visualize dYPEL3 expression pattern in 105 Drosophila larvae. A small number of cells in the larval central nervous system (CNS), 106 including the ventral nerve cord (VNC) and brain, were labeled by mCD8::GFP (Figure 2A). 107 These cells extended fine processes that cover most of the neuropil area in the larval CNS, 108 suggesting that they are neurons. To identify the cell types that express dYPEL3, dYPEL3-109 GAL4 > mCD8::GFP samples were co-immunostained with the neuron marker anti-Elav and 110 the glial marker anti-Repo ( Figure 2B). About 85% of cells that were labeled with dYPEL3-111 GAL4 were positive for Elav, but none was positive for Repo ( Figure 2C). This result 112 suggests that dYPEL3 is predominantly expressed in neurons, but not in glia. Interestingly, 113 dYPEL3-GAL4 also labeled a subset of sensory neurons, including the class IV da neurons 114 (nociceptors), class III da neurons and chordotonal neurons (both mechanosensors), but not 115 the class I da neurons (proprioceptors) (Figure 2B-ii and -iii). dYPEL3 was not expressed in 116 muscles nor epidermal cells (supplement of Figure 2).

118
The disease-relevant mutations of dYPEL3 cause defective nociceptive behavior 119 The human patient shows symptoms mainly in the peripheral nervous system (PNS), 120 including areflexia and hypotonia (the NIH-Undiagnosed Diseases Program). We thus 121 focused type control, dYPEL3 T1-6 , and dYPEL3 T1-8 . and found a significant reduction in nociceptive 133 rolling behavior in the dYPEL3 mutants ( Figure 3B). The extent of decrease in nociceptive 134 rolling was not different between the two mutant alleles of dYPEL3, which are almost 135 identical except for the sequences in the ectopic stretch of amino acids ( Figure 1C). This 136 suggests that the truncation of dYPEL3, but not the presence of the ectopic amino acid 137 sequences, is responsible for the observed phenotype. dYPEL3 T1-8 represents a simpler 138 version since it only has a few ectopic amino acid incorporation ( Figure 1C). Therefore, we 139 focused our analysis on dYPEL3 T1-8 for further analysis. 140 How does dYPEL3 mutation affect the sensory function? We first looked into whether 141 the dYPEL3 mutation affects the development of nociceptors. We expressed mCD8::GFP 142 specifically in nociceptors in wild-type and dYPEL3 T1-8 larvae using the nociceptor-specific 143 driver ppk-GAL4 (Grueber et al., 2007). The gross morphology and total length of dendrites 144 were not affected in dYPEL3 T1-8 ( Figure 4A). Next, we tested whether the presynaptic 145 terminals of nociceptors are defective in dYPEL3 mutants. To this end, a flip-out mosaic 146 experiment was performed to label single nociceptive presynaptic arbors (Yang et al., 2014). 147 The quantification of the total presynaptic arbors revealed that dYPEL3 T1-8 did not affect the 148 development of presynaptic arbors of nociceptors ( Figure 4B).

150
The disease-relevant mutations of dYPEL3 reduce the synpatic transmission from 151 nociceptors to their postsynaptic neurons 152 Next, we assessed the synaptic transmission from nociceptors to their postsynaptic neuron 153 Basin-4, a key second-order neuron in the nociceptive pathway (Ohyama et al., 2015). The . We found that the cumulative GCaMP 162 signals from Basin-4 neurons were signficantly decreased in dYPEL3 T1-8 mutants, as 163 compared to wild-type control ( Figure 5A, ~55% decrease). By contrast, GCaMP 164 measurement in nociceptor axon terminals showed that dYPEL3 mutations did not change 165 nociceptor activation by AITC ( Figure 5B).

167
The disease-relevant mutations of dYPEL3 reduce the synpatic contact between 168 nociceptors and their postsynaptic neurons 169 How do the dYPEL3 mutations reduce the nociceptor-to-Basin4 synaptic transmission? To 170 address this, we employed a synaptic-contact-specific GFP reconstitution across synaptic 171 partners (GRASP) technique, termed sybGRASP (Macpherson et al., 2015), to assess the 172 synaptic contact between the presynaptic terminals of nociceptors and the dendrites of 173 Basin-4 neurons. The GRASP technique utilizes two separate fragments of GFP molecule -174 split-GFP1-10 (spGFP1-10) and split-GFP11 (spGFP11), which can be detected by a 175 specific anti-GFP antibody only when the two fragments are in close proximity to reconstitute 176 a complete GFP. In sybGRASP, spGFP1-10 is fused to the synaptic vesice protein Synaptic vesicle exocytosis from presynaptic terminals exposes spGFP1-10 onto pre-182 synaptic cleft where it reconstitutes the functional GFP molecule by associating with 183 postsynaptic spGFP11 molecules. This technique has been used widely to visualize synaptic 184 contact between two identified neuron types. 185 The spGFP1-10 and spGFP11 were specifically expressed in nociceptors and Basin-186 4 neurons, respectively ( Figure 6A). The resulting GRASP signal was measured in each 187 segmental neuropil, and normalized by the spGFP1-10 intensity in wild-type and in 188 dYPEL3 T1-8 (Figure 6A, top). We detected a mild, but significant, decrease (23%) in the 189 GRASP signals in dYPEL3 T1-8 , as compared to those in wild-type control ( Figure 6A, bottom  190 right). This suggests that the synaptogenesis bewteen nociceptors and its synaptic target 191 Basin-4 is compromised by the dYPEL3 mutations.

192
While the nociceptors express dYPEL3 (Figure 3A), dYPEL3 does not seem to be 193 expressed in Basin-4 neurons, because dYPEL3-GAL4 was not expressed in  neurons labeled by a Basin-4-selective LexA ( Figure 6B) YPEL gene family is highly conserved across the eukaryotic species ranging from 207 yeast to human. Likewise, our homology analysis indicated a strikingly high sequece 208 homology between human and Drosophila YPEL3 (80% identity, Figure 1B). Interestingly, it 209 appears that the sequence homology extends even to the nucleotide level since the 210 analogous frameshift mutation gave rise to the generation of similar amino acid sequences 211 in the ectopic sequences in dYPEL3 T1-6 ( Figure 1C). Given such high sequence homology, 212 we envision that the functions of human YPEL3 and Drosophila YPEL3 are also conserved.

213
YPEL family can be subdivided into two categories. Human YPEL1, 2, 3, and 4 belong to 214 one with high homology with each other, while YPLE5 constitute a distinct family (Hosono et 215 al., 2004). In Drosophila, there is only a single homolog of human YPEL1 to 4, CG15309 216 ( Figure 1B). Because the tissue expression patterns of YPEL genes are complex in human 217 and mice (Hosono et al., 2004), the single YPEL gene makes Drosophila advantageous as a 218 model for studying YPEL3-induced pathogenesis. 219 In human and mice, YPEL3 is ubiqutiously expressed as based on RT-PCR method 220 (Hosono et al., 2004). Northern blot analysis on murine tissues indicated releative 221 enrichment of YPEL3 in brain and liver tissue (Baker, 2003). Our results based on a gene-222 trap Drosophila line indicates that dYPEL3 is expressed in subsets of neurons, but not in glia 223 (Figure 2B and C). The human patient exhibited multiple neurological sympotoms in the 224 PNS, but had normal cognition (the NIH-Undiagnosed Diseases Program). Interestinlgy, 225 dYPEL3-GAL4 was selectively expressed in nociceptors and mechanosensors in the PNS. 226 Furthermore, YPEL3 frameshift mutations reduced nociceptive behavior ( Figure 3B). This 227 suggests that the neurological symptoms in the human patient originates from the neural 228 tissues that normally express YPEL3.

229
Our results suggest that dYPEL3 is expressed in nociceptors, but not in their 230 postsynaptic target Basin-4 neurons (Figure 3A and 6B). Then, how does the YPEL3 231 frameshift mutation lead to neuronal pathogenesis? Our GCaMP experiments showed that 232 the activation of nociceptors by AITC was not altered in dYPEL3 T1-8 mutants ( Figure 5B). 233 Rather, dYPEL3 T1-8 mutations reduced Basin-4 responses to nociceptor stimulation ( Figure  234 5A). The gross neuronal development of nociceptors was not altered by the dYPEL3 235 muations (Figure 4). This suggests that the neurotransmission from nociceptors to their 236 projection neurons is altered by the dYPEL3 frameshift mutation. Interestingly, we found that 237 the synaptic contact between nociceptors and Basin-4 was reduced ( Figure 6A). Because 238 Basin-4 activation is central to nociceptive behavior (Ohyama et al., 2015), the reduced 239 synaptic transmission from nociceptors to Basin-4 is likely reponsible for the reduction in 240 nociceptive behavior in dYPEL3 mutants. It is intriguing that human patient has peripheral 241 symptoms of hypotonia and areflexia, both may arise from reduced synaptic transmission.

242
Our results suggest that nociceptors, but not Basin-4, express dYPEL3 ( Figure 3A and 6B), 243 implying that dYPEL3 mutations affected presynaptic function. It will be important to identify 244 neuron types that express YPEL3 in human in future studies. 245 How does YPEL3 frameshift mutation generate pathogenecity? The mutations in 246 human patient and in our Drosophila model introduce premature stop codons, which may 247 induce the non-sense mediated decay resulting in YPEL3 loss-of-function. However, the 248 analysis of gene structure indicates that the frameshift mutation may escape from the non-249 sense mediated decay, because the premature stop codons are present in the last coding 250 exons both in human and Drosophila YPEL3. This may lead to the generation of a truncated 251 version of YPEL3 proteins. If this is the case, the truncation, rather than the introduction of 252 ectopic amino acid sequences, may play a role in pathogenesis. This notion is supported by 253 our finding that both dYPEL3 T1-8 and dYPEL3 T1-6 altered the nociceptive behavior to the 254 same extent (Figure 3B).

255
The this is the first report on the pathogenecity that is caused by the frameshift mutation of 268 YPEL3. Our model will be instrumental for future investigations that may lead to effective 269 treatments for the disorders caused by YPEL3 mutations. 270

Drosophila melanogaster genetics 272
Drosophila strains were kept under standard condition at 25 ⁰C in a humidified chamber. The 273 following strains were used in the study: w 1118 (3605), ppk-GAL4 (Grueber et al., 2007), ppk-274 LexA ( resulting progeny were screened for the desired mutations by the genomic PCR of CG15309 289 following the Sanger sequencing. 290 291

AITC-induced nociceptive behavior 292
Allyl-isothiocyanate (AITC, Sigma-Aldrich) was prepared in DMSO, dissolved in water as 293 final 25 mM concentration, incubated on a rocker for 3 days before use. Fly embryos were 294 grown for five days in 12 hour light/dark cycle at 25 °C, humidified incubator. The third instar 295 larvae were moved to room temperature for an hour, gently scooped out of food, washed in 296 tap water, and placed on a grape-agar 24 well plate that is covered with 300 µl AITC solution 297 (25 mM). Both male and femle larvae were used for the test. The behavior was recorded 298 with a digical camera for 2 min and the number of larvae showing a complete rolling 299 behavior (minimum 360° rolling) and curling (curling plus any rolling that is under 360°) was 300 manually analyzed (Honjo et al., 2012). The experiments were paired for the wild-type 301 control (w 1118 ) and dYPEL3 homozygous mutant larvae. The experiments were repeated 302 three times in different days with different AITC preparation. All three trials were combined 303 for statistical analysis.

305
Calcium imaging 306 Live calcium imaging was done using GCaMP6f (Mutlu et al., 2012). Briefly, the wandering 307 third instar larvae -the wild-type control male or dYPEL3 T1-8 hemizygous were dissected in a 308 modified hemolymph-like 3 (HL3) saline (Stewart et al., 1994) (70 mM NaCl, 5 mM KCl, 0.5 309 mM CaCl2, 20 mM MgCl2, 5 mM trehalose, 115 mM sucrose, and 5 mM HEPES, pH 7.2). 310 Glutamate (10 mM) were added to the HL3 solution to prevent muscle contractions and 311 sensory feedback. The GCaMP signal was recorded in the entire volume of nociceptor axon 312 terminals or Basin-4 cell bodies. The live imaging was done with a Leica SP5 confocal  313 system equipped with an extra-long-working distance 25X water objective a 1 or 2 µm step-314 sizes. The membrane tdTomato proteins were expressed along with GCaMP6f and used as 315 an internal normalization control for both lateral and focus drifting. The basal GCaMP signal 316 was recorded for a duration of 30 sec to generate baseline fluorescence (F0), then the 317 samples were treated with AITC (25mM) in the HL-3 while continuous recording. The 3D 318 time-lapse images were collapsed to 2D time-lapse by using the maximum Z-projection in 319 the imageJ software. The region of interest was selected either in the axonal projection of 320 nociceptors or in the cell bodies of Basin-4. The ImageJ Time Series Analyzer plugin (NIH) 321 was used to quantify the fluorescence intensity of GCaMP6f. 322 323 Immunostaining 324 The immunostaining was done essentially as previously reported (Kim et al., 2013). Cy5 conjugated goat anti-chicken, Cy2 or Cy5 conjugated goat anti-mouse, Cy5 conjugated 330 goat anti-Rabbit, Cy3 conjugated goat anti-rat. The confocal imaging was done with a Leica 331 SP8 confocal system equipped with a 63X oil-immersion objective with 0.3 µm step-size. 332 The resulting 3D images were projected into 2D images using a maximum projection 333 method.

334
In order to report the relative synaptic contact between the nociceptors and their reconstituted GFP. Therefore, the mouse anti-GFP antibody was used to measure the 342 GRASP signal (anti-GRASP) and the polyclonal chicken anti-GFP antibody was used as an 343 internal control for normalizing the GRASP signal. The fluorescence images were acquired 344 to minimum signal saturation for quantitation. The mean fluorescence intensities of anti-345 GRASP and anti-split-GFP1-10 from each hemi-neuropil segment (segments 4, 5, and 6) 346 were measured from the confocal images. 347

Assessment of dendrite development in nociceptors 348
The membrane GFP, mCD8::GFP, was specifically expressed in nociceptors using ppk-349 GAL4 in a wild-type control (wt) and dYPEL3 frameshift mutants (dYPEL3 T1-8 ). The wild-type 350 control male or dYPEL3 T1-8 hemizygous larvae were used for the analysis. Total length of 351 dendrites was measured from the male larvae of wt and dYPEL3 T1-8 using Simple neurite 352 tracer plugin (Longair et al., 2011) in the ImageJ software. heat-shock inducible Flippase (FLP) was introduced either in a wild-type control (w 1118 ) or in 358 dYPEL3 T1-8 mutants along with ppk-GAL4. The three-day-old larvae grown in grape-agar 359 plate were heat-shocked for 15 min in 37 ⁰C water bath and allowed one more day of growth 360 at 25 ⁰C before dissected and processed for immunostaining and imaging. The wild-type 361 control male or dYPEL3 T1-8 hemizygous larvae were used for the analysis. The total 362 presynaptic arbor length was manually measured using the ImageJ software. Branches 363 shorter than 5 µm were excluded from the analysis. 364

559
(A) The sybGRASP technique was used to report the synaptic contact between nociceptors 560 and Basin-4. The spGFP1-10 and spGFP11 were expressed in nociceptors and Basin-4 561 respectively (bottom left). The resulting GRASP signal was visualized by anti-GRASP 562 antibody (top left, green, see materials and methods for details). The spGFP1-10 that is 563 expressed in nociceptor axon terminals was used as a normalization control (top right, 564 magenta). n = 36 for wt, n = 34 for dYPEL3 T1-8 . Data was presented as a violin plot. Mann-565 Whitney test. The homology between human YPEL3 and Drosophila CG15309