The Abd-B cis-regulatory landscape contains regulatory hotspots
Classical studies identified several regions in the genome that are bound by DNA binding factors via DNAseI hypersensitive sites (Moshkin et al. 2001; Tchurikov et al. 2011; Sung et al. 2016). With the advent of ChIP-seq and subsequent studies involving detailed spatial-temporal binding of the transcription factors (TFs), the information indicating functional relevance of a particular element became more evident. However, the available resources fail to classify the transcription factors functionally, thereby causing a limitation in predictability of the functioning of a cis-regulatory region. Towards this, we manually curated the available datasets for prominent transcription factors interacting at the Abd-B cis-regulatory landscape (Chr 3R: 16,865,212..16,974,406, V6.0) and classified the factors based on their primary function during embryogenesis (Fig 1A, supplemental data SD1.1).
Although, many TFs have multiple roles across different stages of development and in different tissues, TFs were broadly classified into different categories depending on their primary role in activation or expression of Abd-B during development. These include transcription factors involved in A-P and D-V axis formation, neurogenesis, organogenesis, gene regulation including Polycomb/trithorax group factors, and multiple functions (Fig. 1A). We found around 60% of the Abd-B locus to be occupied by TFs that are involved in AP axis formation. These include bicoid (bcd), caudal (cad), eve-strips (eve), fushi-tarazu (ftz), giant (gt), hunchback (hb), invected (inv), knirps (knr), Kruppel (Kr), paired (prd) and sloppy paired 1 (slp1) (Epstein et al. 1997; Kimelman and Martin 2012). Similarly, factors such as dorsal (dl), medea (med) and twist (twi) that are involved in D-V axis formation (Decotto and Ferguson 2001; Stein and Stevens 2014) also occupied around 59% of the Abd-B cis-regulatory landscape (CRL). Organogenic transcription factors such as daughterless (da) are involved in multiple processes like oogenesis neurogenesis, myogenesis and cell proliferation (Cummings and Cronmiller 1994; Brown et al. 1996; Smith et al. 2002; Zarifi et al. 2012), and, occupied almost a quarter (26.8%) of the Abd-B CRL. A significant region (20.94%) was occupied by factors involved in gene regulation including Trithorax-like (Trl), Pleiohomeotic (Pho), Insulator Class I factors (CP190, CTCF or BEAF32), transvection factor zeste (z), or chromatin organizer Early binding complex (Elba) and bric-a-brac 1 (bab1) (Mishra et al. 2001; Mishra et al. 2003; Nègre et al. 2010; Négre et al. 2011). Neurogenic factors such as Chinmo (Chn), Dichaete (D), senseless (sens) and zfh1 occupied ~15% of the landscape (Lai et al. 1991; Russell et al. 1996; Acar et al. 2006; Narbonne-Reveau et al. 2016). Multifaceted transcription factors with dynamic roles such as disconnected (disco), jumeau (jumu), scribbler (sbb) and Schnurri (Shn) are bound at 4% of the Abd-B CRL (Fig. 1B, supplemental data SD1.2) (Marty et al. 2000; Prokopenko et al. 2000; Pyrowolakis et al. 2004; Robertson et al. 2004; Dey et al. 2009; He et al. 2019).
Upon categorization of the TFs as described above, interesting patterns started emerging. For instance, a majority of the region (~70%) bound by the TFs involved in AP axis was also bound by the factors involved in DV axis formation (Fig. 1E). It is possibly for this reason, that a homeotic mutation associated with deletion of CRM or boundary transforms the entire segment into a copy of another antero-posteriorly as well as dorso-ventrally. This could also mean that the Hox genes, therefore, act as the central co-ordinators of overall development across multiple axes. Notedly, we could identify unique regions where AP (~18%) and DV (~13%) axis players bind in a mutually exclusive manner (Fig 1C, supplemental data SD1.3). Future studies targeting these regions specifically might help in understanding the co-ordination of Hox across multiple axes of development.
Most importantly, there were several regions in the Abd-B CRL that showed overlapping binding of TFs involved in all the above categories. We considered these regions as “regulatory hotspots” (Fig 1A and 1D, supplemental data SD1.4). Since a plethora of previous studies worked towards functionally dissecting the Abd-B cis-regulatory landscape, we mapped the experimentally validated elements such as boundaries, P/TREs, repressor elements, initiator elements, promoter targeting sequences and enhancers. Interestingly, a majority of the experimentally validated regions map with the regulatory hotspots identified in the study. There were two cases that appeared particularly interesting upon mapping (Fig 1A, encircled elements). A region in iab5 called i53 (Chr 3R: 16872095..16874092) is previously shown to act as a co-activator for iab6 in A6. However, available ChIP data did not show any known factor binding to the region in i53 except for a narrow region bound by twi and sbb. Conversely, a region in iab7 shows a regulatory hotspot (Chr 3R: 16909151..16910533) that is not experimentally elucidated till date. These incidences suggest the presence of as yet undiscovered elements or transcription factors regulating the expression of associated Hox genes in Drosophila melanogaster.
Targeting regulatorily inert sites for FRT knock-in
The extensive mapping of the Abd-B locus helped us in identifying regions that are devoid of regulatory hotspots, binding of known transcription factors or experimentally validated regulatory elements (Fig 1). Based on the mapping, we identified several such regions flanking boundary and bordering CRMs that are regulatorily inert. Since, the main aim of our study involves changing the number and direction of modules without disturbing their overall architecture, the regulatorily inert sites were chosen as curations to search optimal targets via CRISPR/Cas9. Each of these regions were targeted with FRT sites using ~1Kbp homology arm via CRISPR/Cas9 dependent homology directed repair. The orientation of FRT sites was carefully chosen to generate multiple combinations of the modules upon recombination. These include lines with single FRTs such as MP_F_i5 with the insertion of FRT between MCP and iab5 with an orientation towards iab5; F6_F_i6 having an insertion of FRT between Fab6 and iab6; and F8_F_i8 with the FRT between Fab8 and iab8. These lines would be used to generate deletion or duplication of the CRMs. To generate an inversion of the CRMs, a line with two FRTs in opposite orientation was also created. The two FRTs were inserted between MCP and iab5, and, Fab8 and iab7 (MP_F_i5, F8_F_i7), Fig 1E-F. The aforementioned transgenic lines were used to generate the combinations of CRMs mentioned in the study.
Additionally, a couple of lines with FRTs positioned between iab6 and Fab6 (i6_F_F6), Fab8 and iab7 (F8_F_i7), and, a dual FRT line with one FRT between iab6 and Fab6 and the other FRT between Fab8 and iab7 (i6_F_F6, F8_F_i7) were also generated (Supplemental Fig S1)
Since the entire region is under tight regulation, we intended to alter the internal CRM as minimum as possible and therefore refrained from using any phenotypic marker. Insertion of FRTs at various regions was validated via PCR. Briefly, primers were designed eccentric to the FRT insertion site to amplify screening regions of 500-700 bp length. The FRT site also contains Xba-I restriction endonuclease cleavage site. Therefore, the amplicon digested by Xba-I indicates the presence of FRT sites at desired locus which is later confirmed by sequencing (Supplemental Fig S1).
The single FRT lines can be repurposed to generate targeted deletions and duplications of the CRM-boundary combinations in the locus. Alternatively, the single FRT lines can also be used as landing platforms for various transgenic assays (Golic and Lindquist 1989; Horn and Handler 2005; Phan et al. 2017). The double FRT with the FRTs in the same orientation can be used to generate deletion in -cis or duplication in -trans(Golic et al. 1997). The duplicated line can further be used to generate subsequent modules of the cis-regulatory landscape by repurposing along with single or double FRTs. On the other hand, the double-FRT line with the two FRTs in opposite orientation can be used to generate inversion of the locus to re-arrange the order of the CRMs and boundary (Golic et al. 1997). The FRT lines were utilized to generate modular re-arrangement of the CRMs in the Abd-B locus. A summary of all the FRT lines and genomic rearrangements generated in the study is presented in Fig. 1E-F.
Deletion of iab5-Fab6 renders predicted phenotype
Since a majority of studies in the BX-C are centred around deletion mutations, we generated a fly completely lacking iab5 and Fab6. The fly was thus left with only three modules influencing Abd-B instead of four in a wild-type condition. The deletion spanned ~15 Kbp, and, included whole of iab5 and Fab6 along with adjacent PRE (Chr 3R: 16,872,062..16,887,207, R6.0; Fig 2A).
To obtain the line lacking iab5 and Fab6, we crossed hsFlp; MP_F_i5 with hsFlp; F6_F_i6 (Fig 2A). Both lines express Flippase recombinase (FLP) under the influence of heat-shock inducible promoter (hs). The trans-heterozygous progenies with one allele with an FRT between MCP and iab5 while the other with an FRT between Fab6 and iab6 were given heat shock at 37oC for 90 min at an interval of 24 hrs. Heat shocks were given from the late second instar larval stages until the late pupal stages of Drosophila development. We observed that this time window is the most efficient for the generation of a transgene as it marks the beginning of gametogenesis in flies. FLP activity was confirmed as described previously (Pignoni et al. 1997); see methods and supplementary information (Supplemental Fig. S2) for details.
Once the heat-shocked animals eclosed (G1), they were mated with third chromosome balancers to prevent genetic recombination of target loci. The progenies from these crosses (G2) were pooled in batches of 10 and were again crossed with third chromosome balancers. Once enough activity was observed in the vial, the genomic DNA of the flies were screened via PCR for the desired deletion. The PCR products of flies showing amplification of deletion specific regions were confirmed via Sanger sequencing (Supplemental Fig. S3).
We observed an anteriorization of A5 and A6 into copies of A4 in the deletion mutants Del_i5-F6. The A5 and A6 in wild-type males are characterized by complete melanization in the dorsal abdomen whereas, the A4 has melanization in the posterior end (Fig 2B). The homozygous fly for the deletion of iab5-Fab6 shows a transformation of A5 to A4 characterized by the loss of melanization. Additionally, the A6 was also transformed into a copy of A4 characterized by the loss of melanization and appearance of ventral bristles which are otherwise absent in a wild-type males (Fig. 2B-C). The anteriorization of A5 and A6 into A4 was also observed in females as shown in Fig. 2D-E.
Previously, iab5 was shown to contain redundant enhancers that are required for the activation of iab6 in A6 (Postika et al. 2021b). Since the deletion we generated encompassed all the regions mentioned by Postika et al., we see a similar phenotype with a clean transformation of A5 and A6 into copies of A4. The deletion of iab5 and Fab6 reduces the modularity of the Abd-B cis-regulatory landscape with just three CRMs (iab6, iab7 and iab8) instead of four along with intermittent boundaries and renders expected phenotype.
Duplication of iab6-Fab8 alters modularity and renders variable phenotype
The deletion of iab5-Fab6 causes loss of function of enhancers present in iab6. Therefore, without disrupting iab5, we duplicated the remaining CRMs, iab6 and iab7 through Fab8 (34.776 Kbp; Chr 3R: 16,887,208..16,921,983, R6.0) to enhance the number of modules present in the cis-regulatory landscape of Abd-B. Thus, each CRM is still flanked by a combination of boundary-PRE as shown in Fig. 3.
To obtain a fly with proposed duplication, we genetically crossed the flies with FRT insertions between Fab6-iab6 and Fab8-ia8 to obtain a trans-heterozygous line. A brief schematic of flies used and the duplication upon recombination is depicted in Fig. 3A. For details, see methods and Supplemental Table S9-S10.
Next, to validate the duplication event molecularly, the genomic DNA of prospective recombinants was isolated and amplified by primers specific to the duplication locus (F8i8_ScrF and F6i6_ScrR). The intact loci at Fab6-iab6 and Fab8-iab8 were amplified as controls. Only the flies possessing the desired duplication showed an amplified product (Supplemental Fig. S6C). Genomic DNA from CS flies was used as a control. Unlike the duplication locus that was amplified only in the positive recombinants, the Fab6-iab6 and Fab8-iab8 loci were amplified in both the duplication and CS genomic DNA. The products were later confirmed by sequencing (Supplemental Fig. S4C-D).
In the Dp_i6-F8 flies, the combination of CRM-boundary is intact. Thus, we expected a mixed gain and loss of functions in A6 and A7. Since iab6 and iab7 are regulators of Abd-B, an effect could also have been obtained in A5 and A8. The extra copies of iab6 and iab7 were modularly flanked by Fab6 and Fab7 akin to the wild-type scenario. Surprisingly, we observed a variety of phenotypes with perturbations ranging from disturbances in A1 to a normal-looking fly similar to wild-type (Fig. 3B-H). We classified the phenotypes into six different categories as follows: Category A (Cat-A) mutants show the most extreme phenotypes with all abdominal segments from A1 to A8 severely perturbed. Several flies have disturbances in left-right or dorsal-ventral patterning. These flies are often sterile (Fig 3C). Cat-B animals showed disturbances in segments A5 to A8 (Fig. 3D). The rationale for this category is the role of iab6 and iab7 to regulate Abd-B that defines A5-A8. Any segment perturbed anterior to A5 was not counted in this category, and instead, was noted in the previous category, A. The Cat-C mutants displayed disturbances in A6 as well as A7 (Fig. 3E). Since the duplicated iabs are responsible for providing identities to A6 and A7, flies having perturbations in only these segments were considered a separate category. Cat-D mutants had a perturbation in only one of the abdominal segments. Often these perturbations were restricted to A6 or A7 (Fig. 3F). In the next category, Cat-E mutants display a slight perturbation of dorsal melanization and often rendered “wavy” segmental boundaries (Fig. 3G). The Cat-F mutants show a wild-type phenotype with no observable change in the cuticle or ventral bristle patterning (Fig. 3H).
Both males and females showed the mentioned pattern. However, females were better to observe via imaging owing to clear demarcations of seven abdominal segments followed by genitalia. Also, the cuticles of these animals were extremely brittle, and therefore, the abdomen was directly imaged under the stereomicroscope (Fig. 3B-H; see methods for details).
All the classes of mutants were confirmed to have the same genotype molecularly. Further, the inbreeding of mutants from all the phenotypes, including Cat-F (normal segments), rendered progenies with varying phenotypes again. The development of progenies with inconsistent phenotypes is true for heterozygous and homozygous flies from these categories. Hence, we selected only category F adults and set up a breeding experiment to check the distribution of progeny with different phenotypes.
We observed almost half the population for homozygous (45.24%) and heterozygous (41.35%) progenies were normal (Cat-F). In contrast, the remaining categories had a smaller number of flies in the population (Fig. 3I). Interestingly, ~25% heterozygous (24.67%) and ~17% homozygous (16.52%) flies showed phenotypes grouped under category A. The heterozygous flies showed 1.5 times a greater number of progenies belonging to Cat-A than homozygous (Fig 3I). Our results suggest that more heterozygous flies tend to be severely affected by the duplication of CRMs (Supplementary Data SD2).
Thus, while the deletion of iab5-Fab6 maintains independent development of segments, duplication of modules keeping genomic modularity intact (Dp_iab6-Fab8) disrupts the segmental boundary in a manner that requires further investigations to understand the BX-C better.
The inversion of iab5 through iab7 provides mixed and altered identities of the associated segments, A5, A6 and A7
The CRMs of the BX-C are present in a spatially colinear manner. The three CRMs, iab5, iab6 and iab7 regulate Abd-B to provide identities to A5, A6 and A7 respectively. The specific ordering of the CRMs can be consequential towards co-regulation of the associated gene (Maeda and Karch 2015; Mateo et al. 2019; Hajirnis and Mishra 2021). One way to test this hypothesis is to invert the order of these regions. Towards this, we generated a 46.520 kb (3R: 16,872,062..16,918,581) inversion of iab5 through iab7 (Fig 4).
We generated a fly having two FRTs in the opposite orientations inserted between two loci: MCP-iab5 and iab7-Fab8. The FRT inserted between MCP-iab5 was oriented towards iab5, while the FRT inserted between iab7 and Fab8 was oriented towards iab7. These flies were crossed with a third chromosome balancer, and the heterozygous larvae were given heat shock as described earlier (Supplemental Table S11-S12). The eclosing adults post-heat shock (G0) were crossed with balancers in anticipation that their gametes had undergone the desired recombination (Fig. 6A). The progenies (G1) emerging from these crosses were further crossed with third chromosome balancers. The G1 flies, after three to four days of mating, were sacrificed for molecular screening.
To molecularly screen for the inversion event, the forward primer for MCP-iab5 locus (Mi5_ScrF) and the forward primer for iab7-Fab8 locus (i7F8_ScrF) were repurposed. The forward primer for the iab7-Fab8 locus is now used as a reverse primer for MCP-iab7 recombined locus. Similarly, the recombined locus of iab5-Fab8 was screened by repurposing the reverse primer to screen FRT at MCP-iab5 (Mi5_ScrR) locus with the reverse primer to screen FRT at the iab7-Fab8 region (i7F8_ScrR) as explained in Supplemental Fig. S5. A specific product upon repurposing the primers is obtained only for the flies that had an inversion event. The endogenous loci of MCP-iab5 and iab7-Fab8 were amplified only in the CS fly and not in the inversion line. The recombined loci were further confirmed by sequencing (Supplemental Fig. S5).
An expected outcome of this inversion was a wild-type phenotype for the mutant animal if all the elements are perfectly exchangeable. The other direct possibility was that the “code” for segmental identity resides in the order of CRMs, and therefore reversing their order might reverse the identities of abdominal segments. In this case, we expect a fly that forms A4, followed by A7, A6 and A5 in the anteroposterior order. However, we noted flies with different phenotypes for different segments (Fig 4B-E).
We must also consider that the order of CRMs was reversed, and their relative positioning was also exchanged. For instance, the native position of iab5 now has the presence of iab7 while iab5 replaces iab7 as it is juxtaposed next to Fab8(Fig. 4A). On the contrary, the iab6 is in the same position as before but is present in an opposite orientation. The re-orientation of iab6 also includes the change in directionality of flanking boundaries Fab6 and Fab7. Notedly the boundaries function in an orientation-dependent manner (Zhou et al. 1996; Hogga and Karch 2002; Postika et al. 2018). Therefore, altering the relative positioning of boundaries can also contribute to the phenotypes obtained.
The A5 segment of the flies shows partial gain and loss of function. The numbers and density of ventral bristles decrease, indicating a gain of function to A6 (Fig 4C-D). Occasionally, several males also show partial loss of melanization in A5, indicating partial transformation into A4. The mixed gain and loss of function phenotypes indicate a cell-type-specific behavior of the modules within a particular segment of the mutant animal. The ventral A5 also develops a sclera similar to A6, indicating slight posteriorization (Fig. 4D). The PREs associated with the boundaries in the BX-C provide temporal and cell-type-specific regulation of the adjacent CRMs (Ringrose and Paro 2004). The absence of any major PRE associated with a boundary near iab5 partly explains the incomplete anterior- and posteriorization upon inversion.
Next, even though iab6 is in the same position relative to wild-type condition, it shows a partial gain of function transformation into A7 as indicated by narrowing of the dorsal melanization and sternite in the males (Fig. 4C). The effect is also clearly observed in A6 of the females wherein the ventral bristles attain the identity of the A7 (Fig. 4D-E). Notedly, although the iab6 remains in the same relative order concerning adjoining boundaries and CRMs, the direction of the module is opposite. It has been previously shown that the direction of the BX-C boundaries including Fab6 and Fab7 is crucial to performing the insulator bypass activity (Zhou et al. 1996; Postika et al. 2018). In the inversion mutant, since both the boundaries are inverted, iab6 is misregulated. The studies on orientation were done individually on different elements. The rearrangement in the inversion locus is a complex interplay between many of them. Therefore, the mutants should be probed deeper, especially for the chromatin conformations of the CRMs to understand the interaction of different elements upon inversion.
Further, the A7 shows an opposite transformation when compared to A6. A distinct dorsal melanized segment is present in males in the position of A7, thus indicating anteriorization of the segment into a copy of A6 (Fig. 4C). However, the segment is not completely transformed into A6. A similar effect is evident in the A7 of adult females by the appearance of extra bristles and broadening of the sclera (Fig. 4E). The male flies also have a genital rotation of varying degrees and are sterile.
As mentioned earlier, an inversion such as this includes multiple disruptions of the regulatory landscape of the associated gene and although individual elements have been dissected earlier, the complex interplay of multiple elements is interesting to probe in such scenarios.
Discussion
Homeotic genes or Hox code for sequence-specific transcription factors that define the anterior-posterior body axis of a developing bilaterian embryo (Hajirnis and Mishra 2021). The Hox genes were discovered in Drosophila melanogaster wherein they display a striking property of spatial collinearity (Lewis 1978). That is, the genes are arranged in a complex and, the genes that are present at one end of the complex are responsible for the development of the anterior part of an embryo while the genes present at the other end are responsible for the development of the posterior regions. Most of the previous studies were focused on discovering number and order of Hox across multiple organisms but very few studies were focused on understanding the cis-regulatory landscape of these genes (McGinnis et al. 1984; Monteiro and Ferrier 2006; Shippy et al. 2008; Ho et al. 2009; Noordermeer et al. 2011; Pascual-Anaya et al. 2012; Pascual-anaya et al. 2013; Janssen et al. 2014; Hrycaj and Wellik 2016; Soediono 2016; Szabó and Ferrier 2018; Hajirnis and Mishra 2021).
The posterior Hox complex in Drosophila melanogaster, the bithorax complex (BX-C) presents us with a unique opportunity to study the relevance of the higher-order arrangement of the cis-regulatory modules (CRMs) of the homeotic genes (Maeda 2006). The complex also provides us with a unique locus to study the specific effects of rearrangement of the cis-regulatory modules without affecting the genes(Ho et al. 2009; Hajirnis and Mishra 2021). The three genes of the BX-C: Ubx, abd-A and Abd-B are regulated by nine cis-regulatory modules abx/bx and bxd/pbx regulating Ubx in posterior T2 through A1, iab2 through iab4 regulating abd-A in A2 through A4, and, iab5 through iab8 regulating Abd-B in A5 through A8 (Maeda and Karch 2015). The iabs are arranged in a spatially collinear manner with respect to the segments that are affected upon mutating them. Also, the numbers of the iabs corroborate with the number of abdominal segments formed in the fly. In the present study, we generated a set of FRT transgenes to alter the cis-regulatory landscape of the posterior-most Hox gene Abd-B. These lines can be used independently to insert test elements in the fly or can be repurposed to alter the regulatory landscape in Drosophila.
The role of many of the cis-regulatory elements or modules is known in their native structure (Zhou and Levine 1999; Cléard et al. 2006; Iampietro et al. 2008; Iampietro et al. 2010). Notedly, many of the existing studies regarding genome manipulation of a similar kind are limited by the status-quo of the targeted manipulations, that is, the alterations lack the flexibility to incorporate novel features or re-shuffle the existing modules (Guo et al. 2015; Li et al. 2015; Fabre et al. 2017).
In the current study, the use of FLP-FRT at modular junctions makes the system more dynamic and robust. The alterations in the genome are quickly interchangeable. Furthermore, each of the lines can be followed individually in different directions. For example, it is interesting to probe the chromatin landscape and chromatin interaction in the fly bearing an inversion of iab5 through iab7. Previous studies have shown sequential de-repression of the BX-C cis-regulatory modules from anterior to posterior segments; a feature called open for business model of the BX-C regulation (Bowman et al. 2014; Maeda and Karch 2015). However, one key aspect to probe in the sequential opening model is the ability of these modules to sense the spatial cue of accessibility (Kyrchanova et al. 2015). Therefore, in the case of inversion, segment-specific methylation marks would reveal novel aspects of sequential opening. Moreover, understanding segment-specific interactions of enhancers and promoters via techniques like ORCA and Hi-C will shed light upon the functioning of these modules in the altered scenario (Mateo et al. 2019).
In cases such as Dp_iab6-Fab8, where the modularity is intact with respect to the positioning of CRMs and boundaries but the number of modules has been altered; the variability in phenotype is interesting to probe. For instance, the variability can arise from the change in dosages of the non-coding RNAs (ncRNA) generated from the cis-regulatory locus of Abd-B (Gummalla et al. 2014; Garaulet and Lai 2015). Or, the variability could be caused by ectopic transvection happening in different cell types in a stochastic manner (Hendrickson and S. Sakonju 1995; Vazquez et al. 2006). One way to probe the latter is to investigate the role of epigenetic modifiers such as PcG and trxG proteins to modulate the phenotypic variability (Singh and Mishra 2015). Genes like Zeste are known players of transvection (Sipos et al. 1998; Birve et al. 2001). The effects of mutations in such genes can be probed to understand the nature of variability present in the system. The other possibility pertaining to the differential dosages of ncRNAs can be probed by in-situ hybridization (Arib et al. 2015). However, the definitive changes in the ncRNAs corresponding with the adult phenotypes require further investigation. With the advancement of genomics and gene-editing techniques, many such mutations as presented in the current study can be generated to understand the evolutionary significance of the CRMs and Hox arrangement in different organisms. Additionally, comprehending the epigenomic landscape and DNA looping becomes important to grasp the depths of CRM functioning in such conditions. Thus, the actual complexity of the BX-C regulation cannot be explained by simplistic models such as the open for business but requires deeper attention to the functional evolution of such complexes across animal kingdom.
In summary, we generated a comprehensive map of the Abd-B cis-regulatory landscape that can be utilized to design downstream experiments and predict functionality of different modules. Such principles can be coupled with high-throughput single cell sequencing databases to invent tools of functional relevance (Calderon et al. 2022). Next, we also generated a set of transgenic flies that can alter the modularity of cis-regulatory domains in the BX-C of Drosophila melanogaster and open avenues to explore the fundamental basis of body axis formation. The novel and unanticipated phenotypes obtained in the study clearly demonstrate that our current understanding of the mechanisms of regulation of bithorax complex is substantially inadequate. Further, evolutionary constraint arrangement of Hox genes could be a function cis-regulatory module and not necessarily Hox genes alone. In fact, even in vertebrates, large inversions of Hox complex altered the temporal collinearity due to the juxta positioning of the genes beside anterior regulators (Kmita et al. 2000; Rodríguez-Carballo et al. 2017).
Our study strengthens the potential of targeted re-arrangement of the modules to elucidate the role of the genomic landscape of the bithorax complex. Further, the resources generated in this study are dynamic and provide flexibility for re-arrangement of various cis-regulatory elements. These resources will shed light on important aspects of Hox regulation including but not limited to finer details of the order of the CRMs, chromatin conformations, enhancer-promoter interactions, modus operandi of regulatory boundaries, or causal relation between histone modifications and expression of associated Hox. Genetic interactions of the novel modules will provide crucial insights underlying their concerted functioning and open a new paradigm for the business of the bithorax complex.
Materials and Methods
Comprehensive mapping of Abd-B cis-regulatory landscape
Region spanning distal portion of iab4 through Abd-B encompassing Chr3R: 16,865,213..16,974,406 was annotated in SnapGene software (V6.1, https://www.snapgene.com/; Supp data S1_Cis-regulatory landscape of Abd-B).
DNA sequences for Abd-B gene and transcripts were downloaded from https://flybase.org/. Transcription factors binding sites (TFBS) were annotated from UCSC genome browser as curated by https://flybase.org/. The transcription factors (TFs) were broadly classified into different categories based on their primary roles during embryogenesis (Russell et al. 1996; Epstein et al. 1997; Decotto and Ferguson 2001; Mishra et al. 2003; Négre et al. 2011; Kimelman and Martin 2012; Stein and Stevens 2014; Rossi et al. 2021) and annotated in color scheme as provided in supplemental data (SD1). Chromatin domain boundaries were annotated based on their function, transgenic assays or deletion allele data (Karch et al. 1994; Barges et al. 2000; Schweinsberg and Schedl 2004; Kyrchanova et al. 2011b; Postika et al. 2021a). Subsequently, PREs adjacent to boundaries were mapped using co-ordinates from previous studies ((Barges et al. 2000; Busturia et al. 2001; Mishra et al. 2001; Singh and Mishra 2015; Postika et al. 2021b). The regions between the boundary-PREs were marked as broad domains of CRMs namely, infraabdominal-5 (iab5), iab6, iab7, iab8/9. Region proximal to MCP-boundary was annotated as iab4 CRM. Experimentally validated regions such as initiators (Iampietro et al. 2010), promoter targeting sequences (Zhou and Levine 1999), promoter tethering element (Akbari et al. 2008), repressors (Pankratz et al. 1990; Ho et al. 2009) and enhancers (Hendrickson and Shigeru Sakonju 1995; Estrada et al. 2002; Hogga and Karch 2002; Mihaly et al. 2006; Starr et al. 2011; Postika et al. 2021c)were annotated from respective studies.
Regions spanning multiple A-P or D-V axes TFs were annotated as single region of A-P or D-V TFBS respectively. The experimentally validated data was independently annotated as copies of individual elements and represented together as “experimentally validated data”. Regions spanning A-P and D-V axes TFBS were annotated as common regions. The regions with maximum binding of multiple TFs (minimum threshold of 5 different factors) were annotated as regulatory hotspots.
The detailed annotation of the Abd-B locus can be found in supplemental data SD1 and SnapGene file, SF1.
Analysis of TF binding on the Abd-B cis-regulatory binding
The total nucleotides spanning the binding site for a particular class of TF such A-P TF was fractionated with respect to the total nucleotides bound by any TF. The fraction was normalized to total nucleotides bound by any TF to obtain percentage occupancy of a particular TF class (Supplemental data SD1.2) and plotted as a doughnut chart (Fig 1B).
For the percentage overlap of A-P and D-V axes TFs (Supplemental Data SD1.3), the common regions for A-P and D-V axes TFBS were subtracted individually from A-P or D-V TFBS. The obtained value of total unique A-P and D-V TFBS was summed along with the common regions to get the total nucleotides bound by A-P or D-V axes TFs. Finally, values for unique A-P, unique D-V and common regions occupied by A-P and D-V were normalized against the total nucleotides bound by A-P or D-V axes TFs. Data was plotted as pie chart (Fig 1C).
The regulatory hotspots were normalized for the total nucleotides bound by the TFs and plotted as a pie chart (Supplemental data SD1.4, Fig 1D).
Primer designing for FRT insertion
The primers for targeting Cas9 were obtained after submitting query sequences for different regions of the BX-C at http://crispor.tefor.net/. Drosophila melanogaster BDGP release 6 (R6.0) was selected as the reference genome, and 20bp crRNA with S. pyogenes Cas9 5’-NGG3’ PAM was curated for obtaining targets. From the list of suitable gRNAs, sequences with an MIT specificity score of 97 or more (preferably 100), CFD specificity score of 97 or more (preferably 100) and a Lindel score of 80 or more were selected as preferred guides. Due care was taken to select primers with the least number of off-targets as indicated in the relevant column.
Upon successful selection of guides, 1000-1200 bp upstream and downstream region of the cut site was selected as a query to design primers for amplification and cloning of homology arm for donor constructs in https://www.ncbi.nlm.nih.gov/tools/primer-blast/. For all the FRT insertions, the centromeric proximal homology arm to the cut site was denoted as the left homology arm (LH), while the distal homology arm was denoted as the right homology arm (RH). The reverse primer for LH was selected as 20-24 bp immediately centromeric proximal to the Cas9 cut site (3 bases upstream of PAM). Similarly, the forward primer for RH was selected as 20-24 bp immediately distal to the cut site. Keeping the LH reverse primer as constant (constraint), a forward primer was curated on the online tool.
Similarly, keeping forward primer of RH the constraint, a reverse primer for amplifying RH was curated with the same parameters as stated earlier. Further, to clone the arms with FRTs in the desired orientation and, in pBSKS via Takara® In-fusion (Cat# 638915), the forward primer was added with an overhang complementary to 15 bp immediately upstream of EcoRV digested pBSKS. The reverse primer was also incorporated with a 15 bp o/h homologous to 15 bp immediately downstream of EcoRV digested pBSKS. The reverse primer of LH was provided with an overhang for 2/3rd the size of the 34 bp minimal FRT sequence:
5′-GAAGTTCCTATTCtctagaaaGtATAGGAACTTC-3′
Note that the nucleotides indicated by uppercase alphabets represent binding sites for subunits of the flippase enzyme (FLP). The lowercase alphabets indicate the asymmetric core of the FRT site, which gives the site a particular orientation. A similar o/h was also provided on the forward primer of RH, ensuring that it also has a 15 bps homology to the FRT overhang of LH reverse primer. The complete list of primers is provided in Supplemental Table S4-S6.
Fly stocks and culture
Flies were grown in standard cornmeal yeast extract medium at 25°C unless otherwise specified.
Generation of transgenic lines with FRT insertions at various regions within the Abd-B cis-regulatory landscape
All the donor constructs to target FRT were cloned in pBSKS vector with ~1 Kb homology arms from the region flanking the site of FRT insertion. The arms were amplified with overhangs for FRT sequences with 15-20 bp homology. The homology arms with FRT sequences were cloned in EcoRV digested pBSKS vector using standard protocol for Takara® in-fusion assembly (Cat#102518). All the constructs to express the guideRNA were cloned in pCFD3 vectors as described earlier.
The donor and guide expressing constructs were co-injected in nos-Cas9 Drosophila embryos (BDSC#54591) as previously described or outsourced to Centre for Cellular and Molecular Platforms (C-CAMP), Bangalore, India. The injected G0s and G1 flies were crossed with third chromosome balancer as single fly crosses. The G1s were screened for the desired transgenes by using primers that eccentrically flanked the FRT site. The amplified product was digested using Xba-I to validate the presence of FRT prior to sequencing based confirmation. A detailed schematics for obtaining the FRT transgenes is shown in Supplemental Table S2-S3 and S7-S8.
Generation of recombinants by Flp mediated recombination
The transgenic lines with desired FRTs were crossed and obtained with a cassette containing Flippase recombinase coding region under the influence of heat-shock inducible promoter (hsFLP; BDSC#7). The FLP activity was validated using a fly expressing Gal4 under the influence of Act5C promoter ubiquitously upon FRT excision (BDSC#4779) combined with UAS-GFP and hsFlp expressing flies (see supplementary information, Supplemental Fig. S2).
To obtain deletion and duplication, the flies with desired FRTs were crossed to obtain trans-heterozygous fly with respect to the position of FRT. The trans-heterozygous late second instar larvae are given heat shock at 37°C for 90 mins followed by a daily heat-shock periodically after every 24 hrs till eclosion. The detailed schematics of the crosses for obtaining deletion and duplication is shown in Supplemental Table S9-S10. For obtaining inversion, a fly with two FRTs in -cis is given heat shock as standardized. The schematics of genetic crosses are provided in Supplemental Table S11-S12.
Population assay of variable phenotype for Dp_i6-F8
25 females and 15 males of either Dp_i6-F8 or 25 females of Dp_i6-F8 and 15 males of CS lines were mated and the number of progenies eclosing counted with different phenotypes. The distribution of heterozygous and homozygous progenies was recorded. The flies with affected abdominal segments A1-A8 were counted as category A. The flies with a distortion in A5-A8 were counted in category B. Category C included flies with aberrations in A6-A7. The flies having a defect in only one segment (often A6 or A7) were grouped as Category D. Category E mutants had minor defects in segments without perturbing the segmental boundaries, often characterized as wavy segmental borders. The category F flies appeared normal when compared to a wild-type fly. The distribution was normalized as the percentage of the total population for a given category. The detailed counting is presented in Supplementary Data in form of a workbook.
Cuticle preparations of adult abdominal segments
Adult abdominal cuticles were prepared as described earlier. Briefly, 2-3 days old adult flies were dehydrated in 70% ethanol for 24 hrs. The flies are then boiled at 70°C in 10% potassium hydroxide for 90 mins followed by three washes of autoclaved distilled water. Next, the flies are washed in autoclaved water for 90 mins at 70°C and transferred to room temperature 70% ethanol.
Mounting and imaging of cuticle preparations
The cuticles are transferred to a watch glass containing 70% ethanol and head-thorax region is separated from the abdominal segments by a dissection needle. The abdominal segments are given a sharp cut along the dorsal midline and transferred to a clean glass slide containing P700 halocarbon oil (Sigma #H8898). The cuticle is spread in desired orientation using dissection needles and covered with cover slip. The cuticles were imaged at 1.0X objective, 35X zoom on Zeiss AxioZoom V16 stereo microscope.