A silencer repressing redundant enhancer activities revealed by deleting endogenous cis-regulatory element of ebony in Drosophila melanogaster

The spatiotemporal regulation of gene expression is essential to ensure robust phenotypic outcomes. Pigmentation patterns in Drosophila are formed by the deposition of different pigments synthesized in the developing epidermis and the role of cis-regulatory elements (CREs) of melanin biosynthesis pathway-related genes is well-characterized. These CREs typically exhibit modular arrangement in the regulatory region of the gene with each enhancer regulating a specific spatiotemporal expression of the gene. However, recent studies have suggested that multiple enhancers of a number of developmental genes as well as those of yellow (involved in dark pigment synthesis) exhibit redundant activities. Here we report the redundant enhancer activities in the cis-regulatory region of another gene in the melanin biosynthesis pathway, ebony, in the developing epidermis of Drosophila melanogaster. The evidence was obtained by introducing an approximately 1 kbp deletion at the endogenous primary epidermis enhancer (priEE) by genome editing. The effect of the priEE deletion on pigmentation and on the endogenous expression pattern of a mCherry-tagged ebony allele was examined in the thoracic and abdominal segments. The expression level of ebony in the priEE-deleted strains was similar to that of the control strain, indicating the presence of redundant enhancer activities that drive the broad expression of ebony in the developing epidermis. Additionally, the priEE fragment contained a silencer that suppresses ebony expression in the dorsal midline of the abdominal tergites, which is necessary for the development of the subgenus Sophophora-specific dark pigmentation patterns along the midline. The endogenous expression pattern of ebony in the priEE-deleted strains and the reporter assay examining the autonomous activity of the priEE fragment indicated that the silencer is involved in repressing the activities of both proximal and distant enhancers. These results suggest that multiple silencers are dispensable in the regulatory system of a relatively stable taxonomic character. The prevalence of other redundant enhancers and silencers in the genome can be investigated using a similar approach. Author summary Genes are expressed at the right timing and place to give rise to diverse phenotypes. The spatiotemporal regulation is usually achieved through the coordinated activities of transcription-activating and transcription-repressing proteins that bind to the DNA sequences called enhancers and silencers, respectively, located near the target gene. Most studies identified the locations of enhancers by examining the ability of the sequence fragments to regulate the expression of fused reporters. Various short enhancers have been identified using this approach. This study employed an alternative approach in which the previously identified enhancer that regulates expression of ebony (a gene involved in body color formation) was deleted in a fruitfly, Drosophila melanogaster, using the genome-editing technique. The knockout of this enhancer did not affect the transcription level of the gene to a large extent. This indicated the presence of transcription-activating elements with redundant functions outside the deleted enhancer. Additionally, the transcription of ebony at the midline of the abdomen, which is repressed in the normal flies, were derepressed in the enhancer-deleted flies, which indicated that the deleted enhancer fragment contained a silencer that negatively regulates multiple enhancer activities in a spatially restricted manner.


73
The spatiotemporal regulation of gene expression during the development of organisms 74 results in diverse phenotypes. The cis-regulatory elements (CREs) are nucleotides that differentially 75 modulate the transcription levels of specific genes typically in an allele-specific manner. The most 76 common CREs, enhancers and silencers, are located within a certain distance from the transcription 77 start sites of the target gene and contain binding sites for the transcription activators or repressors 78 [1,2]. Conventionally, the CREs comprising enhancers and silencers, which are unique to each 79 expression unit, are considered as subsets of modular structures in the upstream region of a gene [3-80 body color of the priEE-deleted strains was similar to that of the control strain (Fig 2A). 162 The pigmentation intensity in the females of the two priEE-deleted strains (w 1118 ; e Δ1088priEE 163 and w 1118 ; e Δ1089priEE ), the control strain (w 1118 ; +), and an ebony null mutant (e 1 ) was compared. The 164 percent darkness at the specific positions of the thoracic center and the fourth abdominal tergite (A4) 165 was measured (10 flies per strain) (Fig 2B-C). The pigmentation scores were significantly different 166 between the strains (thorax, F 3, 36 = 733.59, P < 10 -15 , one-way analysis of variance (ANOVA); 167 abdomen, F 3, 36 = 94.919, P < 10 -15 , one-way ANOVA). The thoraces of the two priEE-deleted 168 strains showed significantly but only slightly darker pigmentation than those of the control strain 169 ( Fig 2B). The abdominal pigmentation in the w 1118 ; e Δ1088priEE strain was slightly lighter than that in 170 the control strain, but that in the w 1118 ; e Δ1089priEE strain was not significantly different from the 171 control strain (Fig 2C). These subtle changes in pigmentation intensity suggest that the priEE 172 deletion perturbs ebony transcription to some extent. However, the pigmentation in the thorax and 173 abdomen of the two priEE-deleted strains was markedly lighter than that in the thorax and abdomen 174 of e 1 (Figs 2B-C). This indicated that deletion had limited effects on the overall transcription level 175 regulation. Hence, the priEE was shown to be dispensable for driving the transcription of ebony in 176 the developing epidermis. 177 Also, unexpectedly, the deletion of the priEE affected the pigmentation pattern. In 178 particular, the deletion of the priEE resulted in the loss of a dark pigmentation line along the dorsal 179 midline of the abdominal tergites (Fig 2A). This indicated that ebony expression is suppressed in the 180 midline area and that the priEE fragment is necessary for this suppression. 181 To directly investigate the expression sites of this gene in the abdomen, the abdominal 182 epidermis of the mCherry knocked-in strains, w 1118 ; e::mCherry and w 1118 ; e Δ1077priEE ::mCherry (Fig  183   1C), were subjected to fluorescence confocal microscopy (Figs 3A-B). The abdominal pigmentation 184 of w 1118 ; e Δ1077priEE ::mCherry was consistent with the priEE-deleted strains without mCherry (w 1118 ; 185 e Δ1088priEE and w 1118 ; e Δ1089priEE ) (data not shown), which suggested that the catalytic function of 186 Ebony in the pigmentation synthesis pathway is not disrupted upon fusion with mCherry. 187 Endogenous ebony exhibited a broad epidermal expression pattern (as expected from the light 188 pigmentation), and a suppressed expression at the posterior stripe region of each tergite (A1-6 of a 189 female and A1-4 of a male) and tergite-wide suppression at A5 and A6 in males (Figs 3A,D,and E). 190 These expression patterns were consistent with those reported in previous studies [17,34]. As 191 predicted from the pigmentation scores in Fig 2A, ebony expression in w 1118 ; e Δ1077priEE ::mCherry 192 was similar to that in the control strain except that the expression at the dorsal midline was not 193 suppressed (Figs 3B,D,and E). These results indicate that the enhancer activity resides outside the 194 primary enhancer and that redundant enhancer element(s) is present in the surrounding genomic 195 region. The expression of ebony has been reported in other body regions [26,[37][38][39] but the priEE 196 has not been indicated to drive expression in tissues other than the developing epidermis [26]. As 197 expected, the expression patterns of ebony did not markedly change in other tissues upon deletion of 198 the priEE (Fig S3). 199 The results indicated that the deleted priEE fragment contained a silencer element for the 200 dorsal midline as well as an epidermal enhancer element. In order to confirm that the deleted priEE 201 contains both elements, a GFP reporter assay was performed. The priEE fragment was fused to an 202 enhanced GFP (EGFP) gene (priEE-EGFP) with a minimal Hsp70 promoter and transformed into 203 two attP strains (VK00033 and VK00037). The confocal images of GFP from priEE-EGFP 204 transformed to a third chromosome landing site in VK00033 (Figs 3C and E) and a second 205 chromosome landing site in VK00037 ( Fig S4) indicated that the priEE autonomously drives the 206 epidermal expression except at the flanking regions of the dorsal midline. Rebeiz et al. [26] reported 207 that the 0.7 kbp core element (included in the approximately 970-bp priEE) drove a similar 208 expression pattern. The pattern clearly showed that a dorsal midline silencer is present in the priEE 209 fragment and that it can silence the activity of the proximal enhancer element within the priEE 210 fragment, which drives the broad expression of ebony in the developing abdominal epidermis.

Redundant enhancer activity resides in the regulatory region of ebony 215
The removal of the endogenous priEE of ebony using the CRISPR-Cas9 system did not 216 cause a drastic darkening as observed in the null mutant (e 1 ), although a slight perturbation of the 217 pigmentation intensities in the thoracic and abdominal segments was observed (Fig 2). A strong 218 negative correlation between the darkness of body pigmentation and the expression level of ebony in 219 the developing epidermis has been repeatedly detected in strains sampled from the natural 220 populations of D. melanogaster [24][25][26][27][28][29]. Thus, the dark pigmentation intensity of the cuticle is 221 suggested to be a sensitive indicator of local changes in the expression level of ebony. Therefore, the 222 lack of a large increase in dark pigmentation in strains with the priEE deletion indicated that 223 transcription of the gene was not largely disrupted. Further, these findings indicated that the 224 transcriptional activation of ebony in the developing epidermis can also be driven by the endogenous 225 sequences outside the priEE, suggesting the presence of redundant enhancer elements in the 226 surrounding genomic region. The complex arrangement of multiple CREs may be a reason for the 227 scarcity of polymorphisms association with pigmentation intensity or gene expression level within or 228 near the enhancer element across worldwide populations [32]. 229 The locations of the redundant elements have not been determined. The results of a 230 previous reporter assay revealed that no fragments other than those including the priEE segment 231 were detected within the approximately 10 kbp regulatory region that contains the 5' intergenic 232 region and the first intron [26]. Therefore, redundant enhancer elements are likely to be located 233 elsewhere. However, unlike the recent reporter assay conducted with the yellow regulatory region 234 [23], many regions were tested using relatively large fragments (> 2 kbp), which may contain cryptic 235 enhancers that are repressed by surrounding sequences in their native genomic context. Therefore, 236 the possibility of the presence of redundant enhancer elements within the approximately 10 kbp 237 regulatory region cannot be ruled out. Some secondary enhancers are reported to be shadow 238 enhancers that are more than 20 kbp away from the transcription start site [8]. Thus, there is a need 239 for extensive search to elucidate the detailed spatial arrangement of CREs. Nevertheless, the 240 advantage of deleting an endogenous enhancer, a strategy employed in this study, is the rapid 241 capturing of redundant enhancer activity in the native genomic context. Such knockout assays using 242 endogenous genome editing may reveal more cases of redundant enhancer activities in the 243 perturbations [10,12,39] or define sharp boundaries for gene expression [11,13]. The transcriptional 251 activation of ebony by multiple enhancers appears to be largely overlapping but may not be 252 . This is consistent with the results of a previous study, which showed that the locations 287 of these silencers are not within the deleted fragment [26]. The authors revealed that the male 288 silencer was located approximately 1.5 kbp upstream of the transcription start site, and the stripe 289 silencer was located within the first intron. Whether these CREs recruit long-range and/or 290 short-range repressors or not would be an intriguing question for obtaining a comprehensive picture 291 of the regulatory system of this gene. A similar approach to remove the putative silencer region can 292 be effective for the purpose. 293 At the molecular level, omb, dpp, and wg, are reported to be involved in the formation of 294 sexually monomorphic pigmentation patterns in the abdomen of D. melanogaster, and dpp, which is 295 expressed at the dorsal midline is essential for the formation of dark pigmentation along the midline 296 [46][47][48]. Additionally, dpp is known to activate the BMP signaling pathway, which regulates the 297 transcription of numerous genes through a downstream transcription factor Mad (reviewed in [49]). 298 Kopp et al. [46] showed that Mad 12 clones at or near the dorsal midline promoted the loss of dark 299 pigmentation, which suggested that Dpp signaling contributes to pigmentation. Furthermore, an 300 RNAi screening revealed that 48 transcription factors, including Mad, are involved in abdominal 301 pigmentation [50]. Therefore, it is likely that the suppression of ebony by the silencer is regulated 302 through the Dpp signaling pathway. 303 304

Derepression of ebony is sufficient to diminish a taxonomic character 305
In the genus Drosophila, the pigmentation pattern of the abdominal midline is one of the 306 traits used to classify the subgenus Sophophora, which includes D. melanogaster, and the subgenus 307 Drosophila. With some exceptions, the pigmentation stripes on the abdominal tergites of the 308 subgenus Sophophora are mostly connected or expanded anteriorly at the dorsal midline forming a 309 distinct dark area along the midline as in D. melanogaster ( americana has a dark body color with uniformly dark abdominal tergites, whereas D. novamexicana 320 exhibits a light pigmentation along the abdominal midline [51], which is a typical pattern of the 321 subgenus. A recent study used reciprocal hemizygosity testing to demonstrate that the difference in 322 abdominal midline pigmentation intensity between the two species was due to ebony [52]. The 323 authors showed that ebony is required for the development of light pigmentation along the dorsal 324 midline in wild-type D. novamexicana. It has not been demonstrated whether the interspecific 325 differences of ebony resides in the cis-regulatory region or not. However, the study also suggests that 326 ebony suppression might be a key factor for determining this taxonomically important trait. 327 In this study, a single silencer was sufficient to suppress the activities of multiple 328 enhancers in the cis-regulatory region of ebony. This long-range repression eliminates the need for 329 an acquisition of repressor-binding sites for individual enhancer elements. A study of yellow, which 330 is also expressed in the developing epidermis, from three different species revealed a contrasting 331 picture of frequent evolutionary acquisition and loss of short-range repressor binding sequences [23].

Construct for reporter gene assay 401
The sequence of the priEE of ebony was amplified from Cas-0002-iso ( Fig 1B) using the 402 following primers with restriction enzyme digestion sites: 403 5'-CGGgaattcGGGCAAAGCAGGGTGAATA-3' (EcoRI site) and 404 5'-ACTgcggccgcTCACAGGGACTTATGGGAAA-3' (NotI site). These primers were designed to 405 amplify most of the priEE knocked out sequences including the whole e_ECR0.9 [29] and 406 e_core_cis [28] elements ( Fig S5). The amplified product and the pEGFP-attB vector with a minimal 407 Hsp70 promoter (Drosophila Genomics Resource Center) were digested with EcoRI and NotI. The 408 PCR product was cloned into the multi-cloning site of the vector. 409 410

Deletion strains generated by CRISPR-Cas9 431
The deletion strains were established by crossing gRNA expression strains with 432 Cas-0002-iso or Cas-0002-iso_e::mCherry (the mating scheme shown in Fig S2). Deletions (Dels) 433 occur in the germline cells of G1. Twelve G1 males were crossed one by one with several TBX-0010 434 virgin females. Eight G2 males sampled from the progenies of each G1 male were subjected to PCR 435 screening. DNA samples extracted from the mid-legs of G2 males were amplified using the primers 436 e_-5029F (5'-CGTGTGCCTGATCGCTAGA-3') and e_-3391R 437 (5'-ACTCGTGCCTTACTTAATCTGAA-3'), which were designed to amplify the target region. The 438 G2 individuals were screened by subjecting the amplicons to electrophoresis using a 1% agarose gel.

Quantification of pigmentation intensity 450
At 5-7 days after eclosion, females were placed in 10% glycerol in ethanol at 4ºC for 1 h. 451 Next, the flies were rotated in 10% glycerol in phosphate-buffered saline (PBS) at room temperature 452 for 1 h after removing the head, legs, and wings. The images of the dorsal body of flies soaked in 453 10% glycerol in PBS were captured using a digital camera (DP73, Olympus) connected to a 454 stereoscopic microscope (SZX16, Olympus). The same parameters (exposure time, zoom width and 455 illumination) and reference grayscale (brightness = 128; ColorChecker, X-rite) were applied for 456 capturing all images. White balance was corrected using the white scale (Brightness = 255; 457 ColorChecker, X-rite) with cellSens Standard 1.6 software (Olympus). Pigmentation intensity was 458 measured in manually selected areas of the thorax and abdomen (Figs 2 B  The raw measurement data are in Table S1. The data were analyzed using one-way 463 ANOVA, followed by Tukey HSD post-hoc test. Statistical analyses were performed using R version

Confocal microscopy 467
The adult flies were dissected 4 h after eclosion and the abdomen, wings, front legs, and 468 halteres were collected in PBS. The dorsal abdominal cuticle and epidermis were separated from the 469 rest of the abdomen. The fat body, internal organs, and genitalia were gently removed. The head of 470 adult females collected at 4-4.5 h after the light was turned on was dissected in PBS and the intact 471 brain was obtained. Each brain sample was fixed in 4% paraformaldehyde for 1 h and washed with 472

PBS for 1 h after fixation. 473
Each specimen was mounted with VECTASHIELD Mounting Medium with DAPI (Vector 474 Laboratories) and imaged under a C2 plus confocal microscope (Nikon). Max intensity images were 475 composited from the XY overlapping images (abdomen: 12 images, wing: 10 images) with 1 μm 476 wide Z-stacks using the NIS Elements AR 4.50.00 software. The following laser wavelengths were 477 applied for obtaining images: 488 nm activation wavelength and 509 nm imaging wavelength for 478 EGFP imaging; 561 nm activation wavelength and 620 nm imaging wavelength for mCherry. The 479 identical parameters of C2 plus settings (HV, offset, laser power, pinhole size, scan size, scan speed, 480 scan direction, and zoom) were applied for imaging the same fluorescence in the same tissue 481 (mCherry or EGFP). No further corrections were applied. 482             Different letters indicate significant differences between strains (P < 0.05; one-way analysis of 710 variance, followed by Tukey HSD post-hoc test). Error bars denote standard error.  CRISPR-Cas9-based genome editing was performed by crossing the Cas-0002-iso with the guide RNA (gRNA) expression strain (G0). The progenies from the G1 cross were screened for the presence of deletions. Homologous deletions were achieved by the crosses in G3 and G4. y 2 was removed by the crosses in G4 to G7 because it interferes with ebony in the pigment biosynthesis pathway. Deletion strains for mCherry fluorescence observation were established using the same scheme, except Cas-0002-iso_e::mCherry was used instead of Cas-0002-iso (G0). Control strains (w 1118 ; +; + and w 1118 ; +; e::mCherry) were established with the same crosses using TBX-0002 instead of the gRNA expression strain and using + instead of the Del genotype for the G2 cross.    The guide RNA (gRNA) target sequences are indicated above the sequence. Bases in bold indicate the 947bp sequence of e_ECR0.9 from Takahashi and Takano-Shimizu [29] (969-bp sequence in the control strain (Cas-0002-iso-derived w 1118 ; + strain)). Bases in blue indicate the 961-bp sequence of e_core_cis sequence from Miyagi et al. [28] (975 bp in the control strain (Cas-0002-iso-derived w 1118 ; + strain)). Shaded bases indicate the sequence fragment (1,047 bp) used for the reporter assay in this study.