Redesigning the Drosophila histone gene cluster: an improved genetic platform for spatiotemporal manipulation of histone function

Abstract

Mutating replication-dependent (RD) histone genes is an important tool for understanding chromatin-based epigenetic regulation. Deploying this tool in metazoans is particularly challenging because RD histones in these organisms are typically encoded by many genes, often located at multiple loci. Such gene arrangements make the ability to generate homogenous histone mutant genotypes by site-specific gene editing quite difficult. Drosophila melanogaster provides a solution to this problem because the RD histone genes are organized into a single large tandem array that can be deleted and replaced with transgenes containing mutant histone genes. In the last ∼15 years several different RD histone gene replacement platforms were developed using this simple strategy. However, each platform contains weaknesses that preclude full use of the powerful developmental genetic capabilities available to Drosophila researchers. Here we describe the development of a newly engineered platform that rectifies many of these weaknesses. We used CRISPR to precisely delete the RD histone gene array (HisC), replacing it with a multifunctional cassette that permits site-specific insertion of either one or two synthetic gene arrays using selectable markers. We designed this cassette with the ability to selectively delete each of the integrated gene arrays in specific tissues using site-specific recombinases. We also present a method for rapidly synthesizing histone gene arrays of any genotype using Golden Gate cloning technologies. These improvements facilitate the generation of histone mutant cells in various tissues at different stages of Drosophila development and provide an opportunity to apply forward genetic strategies to interrogate chromatin structure and gene regulation.

Introduction

The organization and regulation of genetic information in eukaryotes is coordinated by the packaging of DNA and core histone proteins (H2A, H2B, H3, H4) into nucleosomes to form chromatin. Nucleosomes are thought to serve as regulatory hubs for signals that direct chromatin folding and gene expression via post-translational modification (PTM) of the highly conserved histone N-terminal tail residues (Strahl and Allis 2000; Kouzarides 2007). A complex pattern of histone PTMs is installed, removed, and interpreted by trans-acting factors that “write”, “erase”, and “read” histone PTMs. The biological functions of histone PTMs are usually inferred by manipulating the activity of readers, writers, and erasers. However, chromatin proteins often have multiple targets in addition to histones, which complicates the determination of histone PTM function using this strategy (Morgan and Shilatifard 2023). For example, the Brahma subunit of BAP and PBAP plays a dual role by reading H3 Lysine 14 acetylation (H3K14ac) and enhancing a writer of H3K27 acetylation, thereby contributing to many downstream effects (Kimura et al. 2003; Tie et al. 2012). Likewise, the H4K20 methyltransferase Set8 has non histone targets in mammals (Shi et al. 2007; Weirich et al. 2015). Last, the TIP60 complex acetylates up to five lysines on the same Histone H4 tail, masking the contribution of any single acetylated H4 lysine via mutation of TIP60 (Jacquet et al. 2016). When manipulating proteins that act on chromatin, we must remain aware of pleiotropic phenotypes that can obfuscate the true function of individual histone PTMs. Consequently, we and others have turned to directly mutating histone tail residues in Drosophila melanogaster to understand histone PTM function more definitively (Günesdogan et al. 2010; McKay et al. 2015; Zhang et al. 2019).

The organization of replication-dependent (RD) histone genes in D. melanogaster makes histone genetics tractable (Corcoran and Jacob 2023). Whereas in most well-studied metazoans (e.g. Homo sapiens, Danio rerio, Xenopus laevis, and Caenorhabditis elegans) the RD histone genes are located in clusters distributed at multiple loci (Fig. 1a), D. melanogaster carry a single RD histone locus on chromosome 2L (HisC) consisting of a 5 kb histone gene unit that is tandemly arrayed about 100 times (Fig. 1b) (Lifton et al. 1978; McKay et al. 2015; Bongartz and Schloissnig 2019). Several groups have taken advantage of this genomic feature to engineer histone genotypes in Drosophila by combining a homozygous HisC deletion with transgenic RD histone gene arrays encoding mutant histone proteins (Fig. 2). Although considerable effort went into developing histone gene replacement platforms in Drosophila, these technologies have limitations. Herzig and colleagues developed the first RD histone replacement platform by deleting the entire HisC locus using the DrosDel system (Ryder et al. 2007), thereby generating Df(HisC^ED1429) which we refer to hereafter as “ΔHisC” (Fig. 2a) (Günesdogan et al. 2010). To rescue the lethal phenotype of a ΔHisC homozygote, four plasmid-based transgenes bearing three tandemly repeated 5 kb histone gene units containing each of the five RD histone genes (i.e. a 3× array of the H1, H2a, H2b, H3, and H4 genes) were inserted into the genome at two different loci using ΦC31 Recombinase (Fig. 2b). Interestingly, the authors found that a total of 12 copies of transgenic, wild-type histone gene units (∼6% of the endogenous number) were sufficient to support viability and fertility of homozygous ΔHisC animals. This platform allowed for the first direct interrogation of histone residue function in metazoan development (Hödl and Basler 2012; Pengelly et al. 2013). Unfortunately, this method was ultimately limited in ease-of-use because of the need to generate and manipulate transgenes at four different loci. Thus, further development by McKay et al. 2015 generated a BAC-based platform that could rescue deletion of the endogenous histone locus with a single 12× histone gene array inserted at one locus (Fig. 2c). These improvements simplified the construction of transgenic genotypes and downstream genetics. Subsequently, a histone replacement platform was developed using CRISPR-Cas9-mediated engineering of the HisC locus (Fig. 2d), replacing the endogenous histone gene array with two transgenic 5× histone gene arrays (Zhang et al. 2019; Fig. 2e). These three gene replacement platforms have been used to greatly expand our understanding of metazoan histone residue function in the context of animal development (Günesdogan et al. 2010; Hödl and Basler 2012; Pengelly et al. 2013, 2015; McKay et al. 2015; Yung et al. 2015; Penke et al. 2016; Graves et al. 2016; Meers et al. 2017, 2018a; 2018b; Armstrong et al. 2018, 2019; Copur et al. 2018; Leatham-Jensen et al. 2019; Zhang et al. 2019; Koreski et al. 2020; Finogenova et al. 2020; Regadas et al. 2021; Lindehell et al. 2021; Crain et al. 2022; Corcoran and Jacob 2023; Salzler et al. 2023).

Fig. 1.

Comparison of replication-dependent (RD) histone genes and proteins among metazoan model organisms. a) Diagram depicting histone gene clusters in H. sapiens, D.rerio, D. melanogaster, X. laevis, and C. elegans. Only chromosomes containing histone gene clusters are shown, and histone gene clusters are represented by orange boxes for all organisms except Drosophila, where the histone gene cluster is represented by a green box. b) D. melanogaster possesses ∼100 tandem-repeats of a nearly identical, 5 kb histone gene repeat unit containing each of the five RD histone genes: H1, H2b, H2a, H4, and H3. c) Multiple sequence protein alignment of the five RD histones from each of the listed organisms. Amino acids are colored by identity (Hydrophobic = dark green, Large Hydrophobic = lime green, Polar = yellow, Small Alcohol = tan, Positive = dark blue green) (Brown et al. 1998) and gaps are represented as gray boxes. Average amino acid similarity and identity were calculated for each species compared to H. sapiens. d) Bar plot of the sizes and identity (determined by nucleotide sequence-based clustering) of each histone repeat unit in the endogenous Drosophila melanogaster HisC locus. Each bar represents a partial-to-full histone repeat, with its position within the array of repeats (start to end) indicated on the X-axis (HisC Coordinate, position of centromere relative to HisC locus indicated by “CEN”). The Y-axis denotes the size of the repeat unit (in kilobases). The colors indicate different clusters of repeats as determined using MeShClust, which revealed five major groups of repeats: 1a and 1b (identified as two sub populations of a larger cluster), 2, 3, and 4. Five additional clusters consisting of only one repeat were also identified. Particular features noted were indicated as: A) cluster 3 with a partial duplication of H2a, B) cluster 4 lacking H4, C) a region with two repeats interrupted by a retrotransposon, D) another region with two repeats interrupted by a retrotransposon, and (E) cluster 2 defined by a lack of an Alu-like retrotransposon sequence element between H1 and H3 (Matsuo and Yamazaki 1989a, 1989b). e) Bar graph of the length of the GA-repeats (Y-axis) located between the divergently transcribed H3 and H4 genes. The X-axis indicates the start and end of the HisC RD histone gene repeat array matching with the above graph.

Fig. 2.

Summary of existing histone replacement strategies in Drosophila melanogaster. a) Diagram depicting the generation of Df(HisC^ED1429) (ΔHisC) by Günesdogan et al. 2010. Wild-type chromosome 2L (top) is shown in purple with the centromere shown in light purple. The replication-dependent (RD) histone cluster is shown as vertical green lines. A zoom-in of the RD histone cluster (middle) depicts the region of chromosome 2L that is deleted in ΔHisC (dotted lines filled with gray), which includes the upstream regulatory region, transcription start site, and 26 bp of the 5′ UTR of the lamp1 gene. Depiction of chromosome 2L after removal of the RD histone cluster in ΔHisC (bottom). b) Diagram depicting histone replacement strategy developed by Günesdogan et al. 2010. Chromosome 3 is shown in orange with the centromere shown in light orange. ΦC31-mediated integration sites (ZH-68E and ZH-86Fb) are shown as green bars at their approximate locations on chromosome 3 (68E1 and 86F8) (Bischof et al. 2007). Four different 3× histone gene unit transgenes were used for rescue of homozygous ΔHisC. c) Diagram depicting the single-site histone replacement strategy developed by McKay et al. 2015. Chromosome 3L is shown in orange with the centromere shown in light orange. The ΦC31-mediated integration site (VK33) is shown as a green bar at the approximate location on chromosome 3 (65B2). d) Diagram depicting the generation of the His^D deletion by Zhang et al. 2019 by independent CRISPR-Cas9 knock-ins flanking the endogenous histone array. His^D was obtained by FLP/FRT recombination between the two CRISPR engineered chromosomes to remove the endogenous histone locus. e) Diagram depicting the histone replacement strategy developed by Zhang et al. 2019, where two 5× Histone array BACs are integrated at the His^D locus.

Each of these RD histone gene replacement strategies have disadvantages that preclude making full use of the arsenal of genetic tools available in D. melanogaster. Therefore, we engineered a novel, designer HisC locus that combines useful features of existing histone replacement platforms with new capabilities, while maintaining a degree of backward compatibility to complement previous work. We have added new functionalities to HisC that expand the utility and afford flexibility in histone genotype engineering. Together this new system provides a more sophisticated toolkit for generating complicated genetic backgrounds, carrying out forward genetics, and performing studies requiring spatiotemporal control of mutant histone gene expression. Here, we describe the features and demonstrate the intended uses of this new system.

Materials and methods

Fly stocks and husbandry

Fly stocks were maintained on standard corn medium provided by Archon Scientific (Durham). The following stocks were obtained from the Bloomington Stock Center:

y[1] sc[*] v[1] sev[21]; P{y[ + t7.7] v[ + t1.8] = nanos-Cas9.R}attP2

w[1118]; P{y[ + t7.7] w[ + mC] = 20XUAS-B2R.PEST}attP2,

w[1118]; P{y[ + t7.7] w[ + mC] = 20XUAS-B3R.PEST}attP2,

y[1] w[*]; P{y[ + t7.7] w[ + mC] = 20XUAS-DSCP-B2R}JK65C/TM6C, Sb[1] Tb[1]

w[*]; P{w[ + m*] = GAL4-ey.H}4–8/CyO

y[1] w[67c23]; sna[Sco]/CyO, P{w[ + mC] = Crew}DH1

Other stocks obtained from colleagues:

y w; ΔHisC/CyO (Günesdogan et al. 2010)

y w; ci-gal4, UAS-GFP (Uyehara et al. 2017; Nystrom et al. 2020)

Databases and online resources

Utilities, tools, and information curated by Flybase (Öztürk-Çolak et al. 2024) and the Alliance of Genome Resources (Kishore et al. 2020) were used throughout the study.

Cloning

ΔHisC^cadillac repair template

Homology arms used for CRISPR engineering at ΔHisC were PCR-amplified from genomic DNA. The 5′ and 3′ ends of each homology arm are listed below:

Distal homology arm (855 bp):

5′genomic sequence—CAACGCGAAAGATTATCATAGATTA

3′genomic sequence—CCCCCTTTTCGGTTGCGCGAGTTGC

Proximal homology arm (1193 bp):

5′genomic sequence—GAGACAAAAACGTATGTGTCAGTGT

3′genomic sequence—CGGAAATAAATTGTCCATCAAGTCC

Homology arms were inserted into pBlueSURF (generously provided by Jeff Sekelsky, UNC) by Gibson assembly (NEB). Sequence containing B2 and B3 recombination sites, attP sites, Act5C-dsRed-Express, and 189 bp of genomic sequence upstream of the lamp1 transcription start site were synthesized by GENEWIZ, Inc. (South Plainfield, NJ) and cloned between the homology arms in the pBlueSURF construct, generating pBlueSURF-ΔHisC^cadillac. PAM sites were abrogated using site-directed mutagenesis (NEB). After experimentally determining that the original guide site in the distal homology arm that we tried did not cut efficiently, 59 bp of genomic sequence were added to the 3′ end of the distal homology arm in pBlueSURF-ΔHisC^cadillac to allow mutations of the new guide site in the repaired chromosome. Plasmid sequence was confirmed using Sanger sequencing and Oxford Nanopore long-read sequencing by Plasmidsaurus, Inc. (Eugene, OR).

Guide RNAs used to target ΔHisC

Guide RNA sequences were cloned into pCFD4-U6:1_U6:3tandemgRNAs was a gift from Simon Bullock (Addgene plasmid # 49411; http://n2t.net/addgene:49411 ; RRID:Addgene_49411) (Port et al. 2014)

Distal gRNA cloning primer sequence:

5′—TATATAGGAAAGATATCCGGGTGAACTTCGCAACTCGCGCAACCGAAAAGTTTTAGAGCTAGAAATAGCAAG—3′

Distal target sequence:

5′—GCAACTCGCGCAACCGAAAA—3′

Proximal gRNA cloning primer sequence:

5′—ATTTTAACTTGCTATTTCTATTTCTAGCTCAAAACATGCGACGCATTTCATTGCTCGACGTTAAATTGAAAATAGGTC—3′

Proximal target sequence:

5′—AGCAATGAAATGCGTCGCAT—3′

CRISPR-Cas9 mutagenesis

Because initial attempts to target ΔHisC in the presence of CyO failed, likely because of Cas9 cutting at the endogenous histone locus on the CyO chromosome, we engineered a stock that expressed Cas9 in the female germline from the nanos promoter (P{y[ + t7.7] v[ + t1.8] = nanos-Cas9.R}attP2) in a ΔHisC homozygous background rescued by a homozygous 12× HWT histone gene array. The pBlueSURF-ΔHisC^cadillac repair plasmid and pCFD4 containing the proximal and distal gRNAs were injected into y w; ΔHisC/ΔHisC; {12×HWT}, nos-Cas9 by Model System Injections (Durham, NC).

We identified transformants that were potential ΔHisC^cadillac repair events by first screening for loss of white expression. Of 11 independent transformants that had white eyes, two were also dsRed+, indicating successful integration of the ΔHisC^cadillac sequences. Full sequence of the locus was confirmed by generating a PCR product with the following primers:

5′—CATGGTATGGGGCGCTATATC—3′

5′—GCCGAGAGTCGGTAAAAATG—3′

followed by TOPO cloning (Invitrogen), and sequencing by Plasmidsaurus, Inc. (Eugene, OR).

PhiC31-mediated integration of 4×, 6×, and 12× histone transgenes

pMultiBAC containing 4×, 6×, and 12× DWT histone gene arrays were prepared as previously described (Koreski et al. 2020), injected into PhiC31; ΔHisC^cadillac/(CyO); 8xHWT^w−/8xHWT^w− and screened for w + transformation by Genetivision (Stafford, TX). The 8xHWT^w− chromosome was generated by Cre-mediated excision between the loxP sites in pMultiBAC and the ZH-86Fb landing site, removing pMultiBAC vector sequences, including mini white and antibiotic resistance components, and the dsRed marker in 86Fb. Progeny were screened for loss of white and dsRed markers.

Distal DWT insertions into ΔHisC^cadillac were validated using the following primers:

5′—GTCGCATTTATAAGTGGCAGTGATATATTTTTTGATTGCCAG—3′

5′—CTCCGAATATGGCCAGTTGGTC—3′

Proximal DWT insertions into ΔHisC^cadillac were validated using the following primers:

5′—GTCGCATTTATAAGTGGCAGTGATATATTTTTTGATTGCCAG—3′

5′—CTAACGAACGTAAGCGACA—3′

Viability assays

Ten yw; ΔHisC^cadillac/CyO; MRKS/TM6B females were crossed with five males of the following genotypes.

1. yw; ΔHisC^{cadillac {D−12xDWT}}; +

2. yw; ΔHisC^{cadillac {D−6xDWT}, {P−6xDWT}}; +

3. yw; ΔHisC; VK33{12xHWT}

CyO+ and CyO− progeny were counted and significance was determined using a Chi-squared test with 1 degree of freedom.

Generation of w- 3×P3-sfGFP and Act5C-sfGFP flies

pattB (generously provided by Jeff Sekelsky) was modified by removing the 5′ region of the white gene (including the promoter) and adding either Act5C-sfGFP or 3xP3-sfGFP sequences. Act5C promoter sequence was PCR-amplified from pBlueSURF-ΔHisC^cadillac plasmid and 3×P3 promoter, sfGFP, and SV40 downstream polyA sequences were PCR-amplified from pScarlessHD-sfGFP-DsRed (Gratz et al. 2015) using the following primer sets.

Act5C

5′—GGGAATTGGGAATTCCATGAATGGCATCAACTCTGAATC—3′

5′—GCCCTTGGACACCATGGTGGCGTCTCTGGATTAG—3′

3×P3

5′—GGGAATTGGGAATTCGGATCTAATTCAATTAGAGACTAATTCAATTAGAGC—3′

5′—GCCCTTGGACACCATGGTGGCGACCGGCTTCG—3′

sfGFP

With 3×P3: 5′—AAGCCGGTCGCCACCATGGTGTCCAAGGGCGAGGAG—3′

With Act5C: 5′—TCCAGAGACGCCACCATGGTGTCCAAGGGCGAGGAG—3′

Shared Rev: 5′—GAGTCGCGGCCGCTACTTGTACAGCTCATCCATGCCC—3′

SV40

5′—GAGCTGTACAAGTAGCGGCCGCGACTCTAGATC—3′

5′—AGAGGTACCCTCGAGTAAGATACATTGATGAGTTTGGACAAACCAC—3′

Promoter (Act5C or 3×P3), sfGFP, and SV40 polyA PCR products were assembled in pattB-w- using Gibson assembly (NEB). Act5C-sfGFP and 3xP3-sfGFP integrations into attP40 were recovered by BestGene, Inc. (Chino Hills, CA) and recombined onto an FRT40A-containing chromosome.

Mosaic eye clones and quantification

eyFLP; Actin5C-sfGFP/CyO; + or eyFLP; Actin5C-sfGFP/CyO; His4r^null females were crossed to yw; ΔHisC^cadillac/CyO; 12xHTG or yw; ΔHisC^cadillac/CyO; 12xHTG, His4r^null males. Twenty adults (10 males and females each) were aged at least 1–2 days after eclosion, placed in a 96-well dish containing molten 1% agarose, and cooled to solidify. Images were obtained on a Leica M205 FCA fluorescent microscope using GFP and RFP band pass filters. Quantification of mutant clone size was determined using FIJI (Schindelin et al. 2012). Area of the eye from stacked RGB eye images was manually cropped to an ellipse using the selection tool. Mutant eye clones were then white pseudocolored using Color Histogram. The image was then converted into a binary image, followed by restoration of the eye area selection. Finally, the area of the eye covered by mutant clones within the eye area selection was determined.

Excision from ΔHisC^cadillac

Genotypes

Figure 7c: y w; ΔHisC^cadillac/ci-Gal4, UAS-GFP; P{y[ + t7.7] w[ + mC] = 20XUAS-B2R.PEST}attP2/+

Fig. 7.

Excision of ΔHisC^cadillac elements and histone gene arrays in a tissue-specific manner. a) Diagram of the predicted 5 kb excision from ΔHisC^cadillac after B3 recombinase expression. b) Diagram of the predicted 5 kb excision from ΔHisC^cadillac after B2 recombinase expression. c) Confocal images of DAPI-stained Drosophila third instar wing imaginal discs in which UAS-B3 is expressed in the anterior compartment by Gal4 under the control of the cubitus interruptus (ci) promoter. UAS-sfGFP is also expressed in the anterior compartment and marks cells with recombinase expression. Loss of dsRed signal indicates excision of the ∼5 kb dsRed expression cassette from the heterozygous ΔHisC^cadillac wing discs. d) As in C except with expression of the B2 recombinase. e) Diagram depicting integration of a 6× wild-type histone gene array into the distal attP site of ΔHisC^cadillac, location of the 41 kb sequence excised upon expression of B2R is noted. f) Confocal images of two DAPI-stained Drosophila eye imaginal discs heterozygous for the 6× wild-type histone gene array in the distal attP site of ΔHisC^cadillac in which UAS-B2 recombinase is expressed in the full eye disc and part of the attached antennal disc by Gal4 under the control of the eyeless (ey) promoter. Loss of dsRed signal indicates excision of the 41 kb dsRed expression cassette plus the 6× wild-type histone array from ΔHisC^cadillac.

Figure 7d: y w; ΔHisC^cadillac/ci-Gal4, UAS-GFP; P{y[ + t7.7] w[ + mC] = 20XUAS-B3R.PEST}attP2/+

Figure 7f: y w; ΔHisC^{cadillac{D−6xHWT}}/ey-Gal4; P{y[ + t7.7] w[ + mC] = 20XUAS-DSCP-B2R}JK65C/+

Procedure

Third instar wandering larvae were placed in 1×PBS and mouth hooks were removed with eye discs attached or larval cuticles were inverted (for wing discs). Mouth hooks and eye discs or inverted larval cuticles were fixed in 4% paraformaldehyde (PFA) in 1×PBS for 10 minutes, then washed 3× with 1×PBST (1×PBS with 0.1%TritonX-100). DAPI was added at a concentration of 0.2 μg/mL for 10 minutes in 1×PBST, followed by a wash in 1xPBST. Fine dissection of eye discs or wing discs was performed and all 1×PBST was removed. Discs were mounted on slides with glass coverslips in Prolong Glass (Thermo Fisher Scientific) and imaged using a Leica confocal microscope with a 20× objective. For display in figures all images were separated into channels using FIJI, pseudocolored appropriately, and saved as flat PNGs. Images were then cropped and sized (within groups) to highlight the tissues while maintaining the relative sizing within each figure panel.

Metazoan histone gene cluster mapping

The approximate locations of the major histone gene clusters in well-annotated chromosomes were determined for H. sapiens, X. laevis, D. rerio, and C. elegans. Briefly, the D. melanogaster H2A (FBpp0085249), H2B (FBpp0085281), H3 (FBpp0085250), and H4 (FBpp0085280) protein sequences were collected from Flybase.org (Gelbart et al. 1997; Jenkins et al. 2022). For each histone protein and each organism, a BLAT search was performed in the UCSC genome browser (hg38, danRer11, xenLae2, ce11; Homo sapiens genome assembly GRCh38.P14—NCBI—NLM, n.d.; Kent et al. 2002; Howe et al. 2013; Session et al. 2016; Davis et al. 2022) to collect tables of conserved histone proteins and their chromosomal locations. A conservative location of major clusters was identified and chosen for display if more than one histone gene was within 2,500 bp of another. The approximate locations were calculated using the coordinates of the located genes relative to the total length of the chromosomes. The measurements and identification are approximate for global illustration. Cartoon representations of adult metazoans were adapted from BioRender.com.

Metazoan histone protein alignment and conservation

Representative sequences for the canonical histone proteins were collected from UniProt (UniProt Consortium 2023). The following accession numbers were used: H. sapiens H1: P07305, H2A: Q6FI13, H2B: P33778, H3: Q71DI3, H4: P62805; X. laevis H1: P22844, H2A: A0A1L8G0Y7, H2B: A0A8J0VAD8, H3: P84233, H4: P62799; D. rerio H1: Q6NYV3, H2A: Q561S9, H2B: Q6PC60, H3: Q4QRF4, H4: A3KPR4; C. elegans H1: P10771, H2A: Q27485, H2B: Q27484, H3: K7ZUH9, H4: P62784; D. melanogaster H1: P02255, H2A: P84051, H2B: P02283, H3: P02299, H4: P84040. Protein sequences were aligned using the ClustalW alignment algorithm in the MegaX program (Thompson et al. 1994; Kumar et al. 2018). Using the Ident and Sim tool of the Sequence Manipulation Suite (bioinformatics.org/sms2; Stothard 2000) we determined the percent residue identity and similarity for each species relative to human. Protein alignments were visualized using Alignmentviewer.org (release 1.0) using the EMBL-EBI Mview color scheme (Kanz et al. 2005).

Endogenous RD histone gene array characterization

The assembled HisC locus sequence was obtained from (Bongartz and Schloissnig 2019). It was noted that the direction of the assembled HisC locus sequence was reversed. Thus, this sequence was flipped to read distal-to-proximal to the centromere. Mapping of full-and-partial histone gene repeats was done using blastn against a 1× wild-type histone repeat sequence. Manual examination and curation were used to properly annotate truncated/expanded arrays. Coordinates were used in BED format to extract individual histone repeats using seqkit subseq –bed arrays.bed histoneconsensus.fasta > single_arrays.fasta. The extracted individual repeats were analyzed using MeShClust with an identity threshold score of 99% (0.99): meshclust -d single_arrays.fasta -o clusters.txt -t 0.99. For display purposes, “C” (center) and “M” (member) cluster members were considered part of the same group. Custom R code was used to plot the location, length, and cluster identity of each histone gene repeat on a ggplot2 bar plot.

The GA-repeat length of the H4 5′ UTR was determined using custom R code to extract the GA-repeat from each identified gene repeat and plot the total number of characters on a ggplot2 bar plot.

High molecular weight Drosophila DNA extraction

Thirty adult flies were collected, frozen, and stored at −80°C. Flies were crushed with a tight-fitting pestle in 270 μL Homogenization Buffer (“HB”, 30 mM Tris-HCl pH 8.0, 10 mM EDTA, 0.1 M NaCl, 0.5% Triton X-100). 250 μL additional HB was used to wash the pestle (total volume ∼520 μL). Homogenized flies were centrifuged at 16,000×g for 5 minutes, before discarding the supernatant. Pellet was resuspended in 520 μL HB by gentle inversion, repeating the centrifugation and resuspension. A total of 60 μL 20% SDS and 20 μL Proteinase K (10 mg/mL) were added followed by incubation at 55°C for 1–4 hours, gently finger flicking every 10 minutes. 1 μL RNase A (100 mg/mL) was added during the final 30 minutes. The mixture was centrifuged at 16,000×g for 5 minutes and the supernatant was carefully transferred into a Phase Lock Gel tube (quantBio 2302830). The supernatant was extracted with 1 volume of phenol:chloroform:isoamyl alcohol twice, and once with equal volume of chloroform:isoamyl alcohol and then the aqueous phase transferred to a 2 mL tube using a wide-bore pipette tip. DNA was precipitated with 0.1 volume of 3 M sodium acetate and 2 volumes of ice cold 100% ethanol. DNA was pelleted by centrifugation at 10,000×g for 5 min. The pellet was washed with 1 mL 70% ethanol twice, before letting it air dry at room temperature for 10 minutes. The pelleted DNA was dissolved in 50 μL nuclease-free water at 4°C overnight, pipetting with a wide-bore 200 μL tip if the pellet remained gelatinous. Qubit and spectrophotometer reading were used to assess DNA quality as outlined in the Oxford Nanopore Native barcoding kit (SQK-NBD114.24).

Oxford nanopore long-read sequencing of plasmids and genomic DNA

Plasmids were grown in E. coli under appropriate conditions and purified using GeneJET plasmid miniprep kit (Thermo Fisher Scientific) using the included instructions. Induction of the BACs was found not to be necessary for obtaining enough intact plasmid for long-read sequencing. Plasmids were verified using Oxford Nanopore long-read sequencing services by Plasmidsaurus, Inc. (Eugene, OR) including preliminary annotation by pLann.

Transgenes were verified using Oxford Nanopore long-read sequencing performed on the GridION platform. High molecular weight genomic DNA was first extracted (as described above) and the Oxford Nanopore kit SQK-NBD114.24 was used to generate libraries using manufacturer instructions. Libraries were sequenced for up to 72 hours, dependent upon the remaining capacity of the flow cell pores.

To assess the quality of the reads the run report was examined, and reads were assessed using the NanoPlot and NanoStat tools of the NanoPack (De Coster and Rademakers 2023). Reads were also examined for contamination by creating short reads from the 100 bases in the center of each read and running FastQ Screen (Wingett and Andrews 2018). We noted a high amount of Wolbachia endosymbiont contamination, with little-to-no contamination from other expected sources (Human, S. cerevisiae, E. coli). These reads were not expected to our reference sequences, however this suggests that our sequencing runs were hampered by these unnecessary DNA contaminants.

Generation of accurate histone gene array plasmid and transgene maps

Alignment visualization and consensus sequence determination required accurate maps of predicted transgenic loci. We provide the sequence of the pCadillBAC vector and ΔHisC^cadillac locus determined by Oxford Nanopore long read sequencing. SnapGene software (www.snapgene.com) was used to assemble histone array plasmids in silico using Actions > Restriction and Insertion Cloning or Actions > Golden Gate Assembly > Insert Multiple Fragments. Careful annotation of genes and features was performed manually and by Snapgene > Features > Detect Common Features. These maps are exported as FASTA to be used as reference sequences for later application. An R script (https://github.com/GreshamLab/labtools/blob/master/R/gff_from_snapgene_features.R) was used to convert Snapgene annotations to GFF annotations for later visualization in IGV (Robinson et al. 2011).

Determination of plasmid consensus sequence

Raw FASTQ reads were obtained from Plasmidsaurus, Inc. (Eugene, OR) and run through a custom pipeline to accurately determine the sequence identity and length of histone multimers (https://doi.org/10.5281/zenodo.11050973). Briefly, seqkit replace was used to arbitrarily rename reads and seqkit subseq was used to select the first 60 bases of a reference sequence (Shen et al. 2016). Reads less than 97% and greater than 103% of the size of the reference sequence were removed using seqkit seq. seqkit fq2fa converted the FASTQ to FASTA files. blat -oneOff = 3 -noHead was used to query the FASTA reads for the beginning of the reference sequence (Kent 2002). Awk selected the names of the reads found to contain the reference sequence. Iterating through each read found to contain a reference sequence, seqkit grep selected the full read, and seqkit restart reorders the read based on the location of the queried reference sequence. Seqtk seq (www.github.com/lh3/seqtk) converted the restarted FASTA reads to FASTQ with arbitrary quality scores. Consensus sequence and alignment were generated using medaka_consensus (www.github.com/nanoporetech/medaka) using the restarted reads. Finally, custom Rscripts generate read length histograms, an alignment of the 5′ end of the restarted reads, and bed files that denote homopolymers and bacterial methylation sites which may be incorrectly ascribed to mutations. Resulting aligned reads were visualized on IGV (Robinson et al. 2011).

Determination of integrated transgene sequence

Raw FASTQ reads were obtained by sequencing on an Oxford Nanopore GridION (described above) and run through a custom pipeline to accurately determine the sequence length of transgenic histone multimers (https://doi.org/10.5281/zenodo.11060915). All relevant FASTQ files are concatenated together. Reads are renamed (seqkit replace), filtered to a minimum length of 5000 bp (seqkit seq –min-len), and converted to FASTA (seqkit fq2fa) (Shen et al. 2016). To ensure BLAT does not suffer a segmentation fault, the renamed FASTA reads are split into 20,000 read files with a split. Iteratively, each read FASTA file is searched for 50 bp sequences that flank the histone repeat array using blat -oneOff = 3 -noHead, appending the results to a table (Kent 2002). Matching read names are taken from the table and filtered such that only one instance of each read is kept (sort names.txt | uniq > blat_names.txt). Iteratively, seqkit grep is used to select each FASTA read that was found to have overlapping sequence. Seqtk seq (www.github.com/lh3/seqtk) was used to convert FASTA to FASTQ with arbitrary quality scores. minimap2 –secondary = no –sam-hit-only -ax -map-ont was used to map the reads to a reference genome of the transgenic locus (Li 2018). samtools sort and samtools view -bq 1 were used to sort and convert to BAM format and remove multiple-mapped reads (Li et al. 2009). Resulting aligned reads were visualized on IGV to find spanning reads (Robinson et al. 2011). The sequence of the ΔHisC^cadillac and VK33 integration sites is available at: https://github.com/DuronioLab/Crain-Nevil-Genetics2024. Since performing the above analysis, a snakemake pipeline was developed to be actively maintained with expanded functionality and ease-of-use: https://github.com/DuronioLab/transgenic_longread_map_snakemake. These improved capabilities include quality screening, 1× locus alignment, and parallel analyses.

Golden gate assembly of histone array multimers

Golden gate assembly was performed largely according to the manufacturer protocol included with the NEBridge Golden gate assembly kit (New England Biolabs) using the alterations noted below (Potapov et al. 2018; Pryor et al. 2020, 2022).

Generation of 1× histone array fragments

Primers were designed against a 1× DWT histone array (pBS-1×DWT) using SnapGene to include 8 bp unique spacer sequences and PaqCI recognition and cut sites compatible with a 4× multimer in the pCadillBAC vector (sequence found at https://github.com/DuronioLab/Crain-Nevil-Genetics2024). The design of these primers follows the suggestions given by Snapgene and (Lee et al. 1996; Pryor et al. 2020, 2022) including eliminating any TTAA overhangs which is readily suggested by Snapgene yet known to be a low-efficiency ligation event in Golden Gate Assembly (Pryor et al. 2020). The primers below were used with a Q5 High Fidelity Polymerase (NEB #M0491S) PCR to amplify individual histone array amplicons in a 50 μL PCR with no modifications to the manufacturer protocol. A small amount of PCR product (5 μL) was run on a 1% agarose gel and visualized with Ethidium Bromide to check for efficient amplification. PCR products were treated with GeneJet PCR clean-up kit (Thermo Scientific) and resuspended in 50 μL of H₂O.

Fragment 1:

5′—TTGGTCCACCTGCTCCTACCGCTAATGCATATGTGGCGAG—3′

5′—TTGGTCCACCTGCTCCTAACAGAGCCGTCTATGTAGTCAAATAAA—3′

Fragment 2:

5′—TTGGTCCACCTGCAAGCTGTTGGCTAATGCATATGTGGCGAG—3′

5′—TTGGTCCACCTGCAAGCACTTTAAGCCGTCTATGTAGTCAAATAAA—3′

Fragment 3:

5′—TTGGTCCACCTGCGGTAAAGTCCTAATGCATATGTGGCGAG—3′

5′—TTGGTCCACCTGCGGTAAGACGTCGGCCGTCTATGTAGTCAAATAAA—3′

Fragment 4:

5′—TTGGTCCACCTGCTGGTGTCTCTAATGCATATGTGGCGAG—3′

5′—TTGGTCCACCTGCATCCCGCCGTCTATGTAGTCAAATAAA—3′

Enzyme pre-digestion

The entire eluted histone repeat PCR amplicons were incubated with PaqCI (NEB #R0745) for 1–3 hours at 37°C according to manufacturer protocol. Digested amplicons were subsequently treated with GeneJet PCR clean up kit (Thermo Scientific) and resuspended in 50 μL H₂O. pCadillBAC vector was incubated with PaqCI for 1–3 hours at 37°C, run out on a 1% agarose gel, and gel extracted with GeneJet PCR clean up kit (Thermo Scientific) and resuspended in 50 μL H₂O. The concentration and volumes of the digested amplicons and vector were noted. The resuspended amplicons and vector were concentrated using an Eppendorf Vacufuge Plus at 60°C using function V-Aq to remove all H₂O without overdrying. Amplicons and vector were resuspended in H₂O to final concentrations of 100 ng/μL.

Golden gate assembly

Equimolar (0.1 pmol) of each digested fragment and vector were mixed and pre-annealed in a thermocycler by heating to 72°C and cooling slowly to 15°C. Upon reaching 15°C, NEBridge Ligase Master Mix (NEB #M1100), NEB PaqCI, and PaqCI activator (NEB #R0745) were added and cycled per manufacturer recommendations for a Golden Gate reaction with four inserts: 37°C for 1 minute, 16°C for 1 minute, cycled 60 times, 37°C for 5 minutes, 60°C for 5 minutes. Assembly mixes were transformed into TransforMax EPI-300 electrocompetent cells (BioSearch Technologies EC300110) and grown on LB + CAM. Correctly assembled vectors were determined by enzyme digest and examination on a 0.4% agarose gel run at <50 V at 4°C. Subsequent verification was done with Oxford Nanopore long-read sequencing as described above.

G418 selection of transformed progeny

Selection of flies harboring hsp40-nptII was performed as described in (Matinyan, Gonzalez, et al. 2021; Matinyan, Gonzalez, et al. 2021) except that fresh G418 was mixed into the food. Briefly, fly food was gently melted in a microwave and allowed to cool in a double boiler with a stir bar. For every 75 mL of melted fly food, an extra 25 mL of distilled water was added. G418 dissolved in distilled water was added to empty fly vials before adding 20 mL of fly food poured in quickly to mix the solution. Vials were allowed to cool and dry overnight, covered by paper towels. The final concentration of G418 used for selection was 150 mg/mL. The reduced number of progeny in G418 vials was visually screened for the absence of dsRed expression, which indicates a complete RMCE event. These progeny were moved from the G418 and kept on normal fly food.

Lamp1 gene expression

Paired-end FASTQ files from four replicates of Oregon-R or y w; ΔHisC/ΔHisC; VK33{12xHWT} (Crain et al. 2024) were aligned to the dm6 release of the D. melanogaster genome with STAR aligner (Dobin et al. 2013), then counts-per-million normalized bigwig coverage files were generated from bam files using deeptools bamCoverage (Ramírez et al. 2016) and visualized using IGV (Robinson et al. 2011).

RT-qPCR

Three replicates of 4 third instar wandering larvae from Oregon-R, yw; ΔHisC/ΔHisC; VK33{12xHWT}, or yw; ΔHisC^cadillac/ΔHisC^cadillac; VK33{12xHWT} were homogenized in TRI Reagent (Zymo) and flash frozen in liquid nitrogen. RNA was isolated using the Direct-zol RNA miniprep kit (Zymo). cDNA was generated from 1 µg of RNA using SuperScript III Reverse transcriptase (Thermo). Real-time PCRs with 2 μL of cDNA, Luna 2× Master mix (NEB), and lamp1 or RpL32 control primers were run on an Applied Biosystems QuantStudio 6 Flex Real-Time PCR System. Relative expression of lamp1 and control was determined for each genotype was by the ΔΔCt method

$2^{-((C{t}_{\mathrm{lamp}1,\exp }-C{t}_{\mathrm{lamp}1,\mathrm{ctrl}})-(C{t}_{\mathrm{RpL}32,\exp }-C{t}_{\mathrm{RpL}32,\mathrm{ctrl}}))}$ (Livak and Schmittgen 2001).

ΔΔCt values across replicates were plotted and significance was determined using t-tests.

lamp1 For:

5′—GGAATCACCATCGACCACCG—3′

lamp1 Rev:

5′—GTGGTAAAGTTTCCCTCCCTAGC—3′

RpL32 For:

5′—ACCAGGAACTTCTTGAATCCG—3′

RpL32 Rev:

5′—CGATATGCTAAGCTGTCGCA—3′

Results

Engineering a multifunctional RD histone gene deletion in D. melanogaster

Drosophila is an excellent experimental organism for manipulating the chromatin environment through direct mutation of histone genes. Importantly, observations made in Drosophila are readily transferable to other organisms because the core histones are highly conserved among metazoans. The average amino acid identity for the core histone tails of H2a, H2b, H3, and H4 among five commonly studied metazoan species are 88.5%, 85.7%, 99.3%, and 99.5%, respectively (Fig. 1c). Recently, long-read sequencing has made it possible to investigate the intraspecies conservation among individual 5 kb histone gene repeats within the RD histone gene array in Drosophila (Bongartz and Schloissnig 2019) (Shukla et al. 2024). We clustered each 1× histone gene repeat via nucleotide sequence similarity and mapped their position along the chromosome (Fig. 1d). We found that most repeats fall within two sub-clusters (1a and 1b in Fig. 1d). The other eight clusters represent structural differences due to partial gene duplication (group a, Fig. 1d), gene deletions (group b, Fig. 1d), retrotransposon insertion (groups c, d, Fig. 1d), and the deletion of an Alu-like retrotransposon element between his1 and his3 (group e, Fig. 1d). Although a small number of SNPs were also observed, a major source of variation among the repeats was the length of a GA-repeat located between the his4 and his3 genes that serves as a binding site for the CLAMP transcriptional regulator (Rieder et al. 2017). Comparing all histone gene repeats from the assembled endogenous array to the cloned wild-type repeat unit we used in building synthetic RD histone gene arrays (McKay et al. 2015) revealed a 99.4% sequence identity with a GA-repeat length in our repeat matching the mode of the endogenous GA-repeat population (20 bp) (Fig. 1e) (Hodkinson and Rieder 2024). There are four SNPs present in coding sequences, one of which (his1^A649G) results in conservative isoleucine to valine difference in H1, with Ile occurring in our cloned repeat unit as well as 28/109 his1 gene copies in the endogenous array and Val occurring in 81 endogenous his1 genes. Therefore, although there is variation in the endogenous HisC locus, the repeat unit we use to engineer RD histone gene arrays represents an approximation of a “typical” RD histone gene repeat.

The endogenous chromatin context has a direct effect on the expression of genes and thus may impact the expression of RD histone genes inserted at ectopic genomic locations (Elgin and Reuter 2013). Therefore, when designing our new platform, we targeted the HisC locus for CRISPR-Cas9 genome editing such that inserted histone genes would be expressed in their endogenous context. We used ΔHisC as our substrate allele for CRISPR engineering to avoid the need to repair across a ∼0.5 megabase region (i.e. on either side of the endogenous 100x histone gene array). We designed a CRISPR-Cas9 guide RNA targeting genomic sequence near the centromere distal 3′ P-element sequence located at ΔHisC, and another guide RNA targeting the 5′ UTR of the centromere proximal lamp1 gene (Fig. 3a). We constructed a repair template containing Actin5C-dsRed between distal and proximal homology arms to facilitate repair of the 5.6 kilobase gap resulting from site-specific cleavage by Cas9 (Fig. 3b). Potential recombinants were obtained by recovering dsRed-positive progeny from G0 animals containing a transformed germ line. Sequencing of HisC PCR products isolated from transformants confirmed a clean repair that extended beyond the homology arms of the repair template into the adjacent, exogenous sequence (Fig. 3c). We also confirmed the intended mutation of the guide RNA sites to abrogate re-cutting of repaired loci (Fig. 3c). A PCR product spanning the locus was analyzed by Oxford Nanopore long read sequencing and confirmed the presence of the entire repair template (Fig. 3d). As this newly engineered locus is designed with many desirable features, we named it the ΔHisC^cadillac locus. As expected, ΔHisC^cadillac is inviable when homozygous and a 12× histone wild-type (HWT) transgene located on the third chromosome rescues this lethality, demonstrating the applicability of ΔHisC^cadillac for a new histone gene replacement platform.

Fig. 3.

Engineering of ΔHisC^cadillac, a novel designer histone deletion locus in Drosophila melanogaster as a platform for histone gene replacement. a) Diagram of the ΔHisC locus with locations of the CRISPR-Cas9 guideRNA target sequences used for gap repair. ΔHisC is marked by mini white (white) with an FRT site integrated at the first intron. Breakpoints between endogenous sequence and P-element sequence are noted using dm6 coordinates (Hoskins et al. 2015). The centromere-proximal breakpoint results in removal of the transcription start site and upstream regulatory region plus 26 bp of the 5′UTR of lamp1. b) A simplified diagram of the CRISPR-Cas9 repair template plasmid, noting the distal and proximal homologies and desired insertion sequence. c) A simplified diagram of the repaired locus with Sanger sequence traces over the mutated guideRNA sequences and junctions between endogenous sequence (distal/proximal homologies) and inserted sequences (B2, and 5′-UTR of lamp1). d) Oxford Nanopore long read sequence validation of the ΔHisC^cadillac locus. Read depth at each position of the locus is represented on the y-axis with the color indicating the percent of reads with the expected sequence. The red positions (A), (B), (C), D), and (E) represent differences between ΔHisC^cadillac and the reference sequence. e) Fully annotated diagram of the ΔHisC^cadillac locus.

ΔHisC^cadillac contains several features that greatly expand the genetic toolkit at the Drosophila RD histone locus (Fig. 3e). First, it precisely deletes HisC, leaving the lamp1 gene intact, in contrast to the widely used ΔHisC allele which deletes the lamp1 promoter. Although lamp1 is not required for viability, it encodes a protein critical for lipid metabolism (Chaudhry et al. 2022). lamp1 expression is substantially reduced in HisC homozygous larvae but is comparable to true wild-type in ΔHisC^cadillac homozygous larvae (Supplementary Fig. 1a, b). Therefore, ΔHisC^cadillac eliminates any potential confounding phenotypes resulting from loss of lamp1. ΔHisC^cadillac contains two co-linear attP sites for integration of either one or two transgenic histone gene arrays. This design also allows for combinatorial integration of two arrays with different genotypes and integration via recombinase-mediated cassette exchange (RMCE). Inclusion of Actin5C-dsRed simplifies scoring of genotypes through positive identification of the ΔHisC^cadillac chromosome (Supplementary Fig. 1c). The attP sequences and Actin5C-dsRed are flanked by B2 and B3 recombination sites, which allow for excision of sequences integrated at the attP sites using UAS/Gal4-driven B2 and B3 site-specific recombinases (Nern et al. 2011). Importantly, the location of these B2 and B3 recombination sites allows for the removal of (1) only sequences integrated into the distal attP site via B3R or (2) sequences integrated into either or both attP sites via B2R. To expand the combinatorial capacity of the locus a loxP site was inserted between the attP sites. This loxP site can be used to remove vector sequences after integration (McKay et al. 2015; Meers et al. 2018a). Last, the ΔHisC^cadillac chromosome also contains the centromere-linked FRT40A to facilitate mitotic recombination. The following sections describe the utility of these and other features of ΔHisC^cadillac.

Generating histone mutant mosaics in the adult Drosophila eye for use in forward genetic screens

A primary draw of histone gene replacement strategies is the ability to assess phenotypes of completely mutant animals. However, due to the importance of many histone residues, finely tuned assessment of effects in certain tissues (e.g. those in adults or pupae) is sometimes precluded by the failure of mutant animals to develop to the necessary stage. Generating genetically mosaic tissues in flies that otherwise remain heterozygous for a lethal histone mutation provides a solution to this problem. We therefore created a strategy for generating mosaic eye tissue composed of clones of cells with different histone genotypes. The Drosophila eye has an easily observable color and structural organization; it is dispensable for survival and can be manipulated using a variety of genetic tools (St Johnston 2002; Baker et al. 2014). Inducing mitotic recombination during development results in a mosaic tissue with clones of homozygous wild-type and clones of homozygous mutant cells that directly compete against one another as the eye grows. For instance, Kanda and colleagues showed that eye clones mutant for the histone methyltransferase trithorax related have a growth advantage over their wild-type neighbors (Kanda et al. 2013). However, such experiments are difficult to perform for histone mutant cells using the existing histone gene replacement platforms.

The primary complication is the white (w) marker gene within ΔHisC (Günesdogan et al. 2010). In most experiments, mutant adult eye clones are also w mutant (and thus white in color) and wild-type eye clones are w+ (and thus red in color). Because the ΔHisC allele carries a white marker gene (Günesdogan et al. 2010; McKay et al. 2015), ΔHisC homozygous cells resulting from mitotic recombination are w+, obscuring the identification of histone mutant clones, especially when they are small. That is, none of the resulting histone mutant clones lack white expression. Furthermore, we found that flies expressing ey-FLP in background with ΔHisC but no other FRT-containing chromosome had red-and-white mosaic eyes, likely due to intrachromosomal excision of sequences between the FRT sites located in ΔHisC and the FRT sites located at 40A (data not shown; FRT sites diagramed in Fig. 3a).

We engineered ΔHisC^cadillac to solve these problems and to allow easy and efficient assessment of clones resulting from mitotic recombination, particularly in the eye. The white gene of ΔHisC is replaced by Actin5C-dsRed in ΔHisC^cadillac, providing a bright marker of the locus under fluorescent light, yet imparting no discernable eye color under white light (Supplementary Fig. 1c). To test whether ΔHisC^cadillac could be used to detect histone mutant mosaic tissue using fluorescent marker genes rather than white, we first created two constructs containing either Act5C-sfGFP or 3xP3-sfGFP and lacking white. We independently inserted these constructs into attP40 on chromosome 2L, followed by recombination to a chromosome containing FRT40A (Supplementary Fig. 1d). Neither sfGFP-containing transgene has discernable eye color under white light but is distinguishable with a fluorescent microscope, where Act5C-sfGFP is expressed in the entire animal and 3xP3-sfGFP is expressed only in the eyes (Supplementary Fig. 1e).

With these new chromosomes, we used ey-FLP to drive mitotic recombination at FRT40A, thereby generating sister clones in the adult that are either genotypically ΔHisC^cadillac/ΔHisC^cadillac or HisC⁺/HisC⁺, both of which contain a 12× histone gene array (12× HWT) on chromosome 3 (Fig. 4a). In this way, we can assess the proliferation of cells that contain only 12× HWT (dsRed+, sfGFP−), cells that contain 212 histone gene copies (dsRed−, sfGFP+) or cells with 112 histone gene copies (dsRed+, sfGFP+) (Fig. 4a). Homozygous ΔHisC^cadillac cells fail to proliferate and do not contribute to the adult eye in the absence of a rescuing transgene (Fig. 4b, c; TM6B). This proliferation defect is rescued by the 12 copies of wild-type histones in the 12× HWT transgene (Fig. 4b, c; compare HWT vs TM6B). Note that because ey-FLP is activated early in development and is expressed continuously in the eye, essentially the entire eye is populated by clones of cells with either 12× histone genes (magenta) or 212 histone genes (green). Quantification of clone size indicates that 12× HWT clones grow equally well next to clones containing 212 copies of wild-type histones in both males and females (Fig. 4c; HWT). This result was anticipated given that 12 copies of wild-type histones are sufficient for whole animal viability and fertility (McKay et al. 2015) (Penke et al. 2016). Notably, the weak white expression from the HWT transgene rescue cassette does not obscure the visualization of dsRed or sfGFP in the adult eye.

Fig. 4.

Fluorescent adult eye mosaic analysis of histone mutant genotypes. a) Left: genotype of fly in which mosaicism was induced using ey-FLP mediated mitotic recombination. Right: diagram of the expected clonal genotypes and fluorescent phenotypes after mitotic recombination during eye development. b) Female and male examples of mosaic adult Drosophila eyes containing ΔHisC homozygous clones rescued with the indicated 12× histone transgenes, with or without deletion of His4r. TM6B indicates flies lacking a 12× histone transgene. For each genotype the mosaic tissue is composed of clonal populations of cells that are homozygous “Wild-type” HisC and homozygous for Actin5C-sfGFP (green; 200 endogenous histone genes and 12 transgenic), homozygous “Mutant” for ΔHisC^cadillac (magenta; 12 transgenic histone genes), or heterozygous (yellow; 100 endogenous histone genes and 12 transgenic). Note that the amount of heterozygous tissue is exceedingly small, due to the expression of FLP recombinase throughout eye development. c) Quantification of the dsRed signal (homozygous ΔHisC^cadillac) as a percentage of the total area of the eye image, displayed as box plots. Data separated into male (♂) and female (♀) flies. Blue boxes are genotypes with endogenous His4r, and orange boxes are genotypes without His4r. Green boxes indicate control tissue. Significance values calculated by Wilcoxon rank sum test: ns = not significant; * = P < 0.05; *** = P < 0.001.

The Drosophila compound eye is a powerful tool to study growth and cell proliferation through forward genetic screens using the FLP/FRT system (St Johnston 2002; Tapon et al. 2002; Pellock et al. 2007; Tseng et al. 2007; Kaufman 2017). We investigated whether we could measure proliferative differences in wild-type vs histone mutant cells using our fluorescent mosaic eye assay. This approach could unmask defects in the growth of histone mutant cells not apparent in whole animal mutants (Penke et al. 2016). During Drosophila imaginal disc development, cells that grow more slowly than their neighbors are outcompeted and actively killed (Morata and Ripoll 1975; Johnston 2009). Consequently, clones containing histone mutant cells that proliferate poorly will either be smaller than wild-type clones or absent from the adult eye altogether. We tested this premise using a previously described H4^K16R substitution mutation (Armstrong et al. 2019). Because acetylated H4^K16 is critical for upregulation of the X chromosome in XY males to balance gene expression with XX females, H4^K16R mutant males are inviable whereas H4^K16R females are viable (Armstrong et al. 2019). We asked to what extent the lethal phenotype of H4^K16R mutant males might be due to defects in cell proliferation. We quantified the fraction of the adult eye composed of histone mutant tissue, with 50% representing no proliferation defect. We found that clones of H4^K16R female tissue were slightly smaller than the control (33.5% for H4^K16R females compared to 45% for HWT females) and that this effect was more pronounced in males (25.5% for H4^K16R males compared to 52% for HWT males) (Fig. 4b, c; compare HWT vs H4^K16R). Interestingly, additional homozygous deletion of the single copy His4r gene, which is located outside of HisC and encodes an identical H4 protein, in H4^K16R cells resulted in a more severe growth defect in male cells but did not affect female cells (12% for H4^K16R, His4r^null males vs 50% for HWT, His4r^null males; 40.5% for H4^K16R, His4r^null females vs 46.5% for HWT, His4r^null females) (Fig. 4b, c; compare His4r vs H4^K16R His4r). These data demonstrate that cells in the developing eye of males proliferate poorly when they can only produce H4^K16R mutant histone. They also demonstrate that we can modulate the growth of histone mutant cells by mutating additional genes, indicating that this method could be utilized in forward genetic screens to identify new genes and pathways that regulate chromatin organization and epigenetic control of gene regulation and cell proliferation.

PhiC31 integration of 4×, 6×, and 12× histone arrays into ΔHisC^cadillac

The ability to insert transgenic histone gene arrays of various copy number and coding potential at the endogenous HisC locus is an essential feature of ΔHisC^cadillac. We designed the locus to contain two co-linear attP sites for integration of attB-containing constructs using established cloning and transgenic injection methods (Fig. 5a) (McKay et al. 2015; Meers et al. 2018a). To test the integration efficacy at these sites, we took advantage of our “Designer Wild-type” (DWT) histone gene repeat, which contains polymorphisms in each of the 5 RD histone genes that allow us to distinguish endogenous wild-type histone genes from transgenic histone genes (Koreski et al. 2020). Using our previously described pMultiBAC vector (McKay et al. 2015), we obtained successful integration of 4×, 6×, and 12× DWT histone gene arrays at ΔHisC^cadillac that were confirmed via PCR and nanopore sequencing (Fig. 5). Using PCR targeting the junctions created during integration, we can determine whether transgenes are integrated at the distal attP, proximal attP, or both attP sites (Fig. 5b, c). We recovered distal, proximal, and double insertions of 4× DWT and 6× DWT transgenes, and a distal insertion of a 12× DWT transgene (Fig. 5c, Supplementary Fig. 2a). Importantly, targeting BACs with single attB sites does not allow for the control of whether distal, proximal, or double insertions occur. A qualitative assessment of the viability of animals with 4×, 8×, 12×, and 24× copies of DWT histones expressed from integrations into ΔHisC^cadillac is similar to previously described histone gene arrays integrated at VK33, with viable adults eclosing only when histone copy number was 8× or greater (Fig. 5c). In addition, the viability of ΔHisC^{cadillac{D-12xDWT}}/ΔHisC^cadillac hemizygous flies (n = 126, 88% observed/expected, P = 0.11), ΔHisC^{cadillac{D-6xDWT},{P-6xDWT}}/ΔHisC^cadillac hemizygous flies (n = 206, 86% observed/expected, P = 0.09), and ΔHisC^cadillac/ΔHisC; VK33{12xHWT} (n = 79, 89% observed/expected, P = 0.31) were equivalent. These data indicate that RD histone gene arrays integrated at HisC and the attP site at VK33 are functionally similar.

Fig. 5.

PhiC31-mediated integration of histone gene arrays into ΔHisC^cadillac. a) Diagram of the predicted integration via ΦC31 recombination between the ΔHisC^cadillac attP sites and the attB site on the BAC vector. b) Diagram of resulting locus after integration at both the distal and proximal attP sites. PCR target sites spanning the junction between the integrated histone gene array/vector sequence and ΔHisC^cadillac sequence are indicated in with pink bars. c) Representative DNA agarose gels with bands indicating amplification across a vector/ΔHisC^cadillac junction. Summary table (right) describes the proximal and/or distal integration of different histone array multimers, the number of independent lines isolated, and (*) whether the integrated transgene supported viability of ΔHisC^{cadillac {Histone Array}} homozygotes. “Yes**” indicates that 1 of 3 independent lines is homozygous viable. d) Nanopore validation of a pMultiBAC-6×DWT (top) and the resulting ΔHisC^{cadillac {D-6xDWT}, {}­^P-6xDWT} (bottom) transformant. Gray lines indicate aligned long read data from plasmid [top; Plasmidsaurus, Inc. (Eugene, OR)] or transformants (bottom; in house). Both sets of data are aligned to the predicted ΔHisC^{cadillac {D-6xDWT}, {}­^P-6xDWT} sequence (middle) including genomic DNA (blue line), pMultiBAC (gray boxes), DWT repeat (green arrows), and dsRed (from ΔHisC^cadillac, red arrowed box). Anchor points for the genomic long read alignment are indicated by asterisks (*) on the cartoon reference genome. Reads that span an entire pMultiBAC-6xDWT transgene and an anchor point to unambiguously assign it to the distal or proximal attP sites are displayed in purple.

We next sought to determine the identity and structure of the histone gene arrays inserted at ΔHisC^cadillac. However, determining the sequence and copy number of these arrays is challenging due to the lack of accessible, cost-effective, and high-throughput technologies for sequencing large, repetitive DNAs. We have used Sanger sequencing of pMultiBAC plasmids prior to injection to confirm the sequence of engineered histone gene arrays, but this short-read method cannot reliably confirm the sequence of each histone gene unit in a large array of repeated units. Further, the only method currently available to determine the number of histone gene units in a transgenic histone gene array is via southern blotting of genomic DNA, which is expensive, low-throughput, and technically challenging (McKay et al. 2015; Zhang et al. 2019).

We, therefore, asked whether leveraging Oxford Nanopore long-read sequencing technology could overcome the hurdles associated with these commonly utilized methods to accurately determine the sequence and repeat number of integrated histone gene arrays (Fig. 6). We sequenced our 12× pMultiBAC injection constructs using a commercial long read sequencing service and attempted to align raw reads to a reference sequence or perform de novo sequence assembly to obtain the sequence of the entire 60 kb 12× array. Unfortunately, neither of these established long-read sequencing alignment methods was successful. We observed that many reads were misaligned or collapsed into smaller repeats, prohibiting the assembly of the entire multi-repeat array (Fig. 6a). Given that repetitive histone gene arrays are flanked by unique DNA sequences in the pMultiBAC backbone, we anchored sequencing reads to these unique sequences to allow for proper alignment (Fig. 6b). Using this strategy, we could generate consensus sequences without sequence collapses and shifts or obfuscation of mutations. Additionally, this method uncovers small differences in base calls due to bacterial methylation of the DNA (Fig. 6c). We have found that this occurs in the absence of a DNA methylation sequencing protocol, likely because the bacterial methylation interferes with normal long-read base calling by changing the charge of the DNA bases. Removal of bacterial DNA methylation alleviates this phenomenon, but our method also provides information to determine whether an apparent mutation is real or the result of methylation.

Fig. 6.

Oxford nanopore validation of BAC-based histone gene arrays before and after genome integration. a) Commercial plasmid sequencing services rely on the random linearization of plasmids prior to Oxford Nanopore sequencing, generally creating a single linear molecule per circular input molecule. De novo assembly of these sequences collapse the repeated histone array units (green) while leaving the BAC vector (purple arrows and black line) well annotated. Sequence alignment with ClustalW (or similar) will result in gaps or masking of histone array repeats. b) Schematic of a strategy to determine the histone gene array length and sequence identity using commercial plasmid sequencing with Oxford Nanopore technology. An accurate, linearized reference sequence is generated with repeat regions (green) flanked by unique vector sequence (purple arrows and black line). The first 50–60 bp are selected as the reference origin sequence. A BLAT search is performed against every read, searching for the reference origin sequence. On every read, sequence 5′ of the identified reference origin sequence is moved to the end of the read. The modified reads are aligned, and an accurate reference sequence is generated. c) Oxford Nanopore sequencing will detect prokaryotic DNA methylation as a mis-called base. Logograms showing the depth of base calls around dcm methylation sites (Top) and EcoKI methylation sites (Bottom) grown in either DH5a (dcm+/EcoKI+) or Stellar (dcm−/EcoKI−) E. coli cell lines. d) Schematic of a strategy to determine the array length and sequence identity of integrated histone gene arrays using Oxford Nanopore sequencing. An accurate reference sequence of the target locus is generated and ∼50 bp of sequence flanking the integrated histone arrays are selected (yellow highlights). A BLAT search is performed against every read, searching for the unique reference sequences. Reads without these sequences are removed. Alignment with minimap2 generates an accurate map of the sequence.

Having validated this approach using DNA purified from E. coli, we next applied it to genomic DNA to determine the structure of the transgenic histone gene arrays integrated into ΔHisC^cadillac. We performed high molecular weight genomic DNA extraction from transformed flies with both distal and proximal 6×DWT integrations in ΔHisC^cadillac (ΔHisC^{cadillac {D-6xDWT}, {P-6xDWT}}) followed by whole genome sequencing on an Oxford Nanopore GridION. Because reads containing only histone gene array sequences result in ambiguous alignments due to the nature of the repeat, we only aligned reads that overlapped with unique, flanking genomic sequences (Figs. 6d, 5d). Using this approach, we identified four reads spanning each 41 kb 6× insertion and >50 reads at each junction between ΔHisC^cadillac and pMultiBAC-6×DWT sequences (Fig. 5d). Despite the de-enrichment of reads over the histone gene arrays, these data confirm the integration of two complete 6× histone gene arrays in the ΔHisC^{cadillac {D-6xDWT}, {P-6xDWT}} transgenic line (Fig. 5d). Thus, this new approach obviates the need for southern blots to confirm the integration, orientation, and length of transgenic histone gene arrays at ΔHisC^cadillac.

Tissue-specific expression of mutant histones without using mitotic recombination

Generating mosaic tissue via the mitotic recombination strategy described above offers a way to assess phenotypes of histone mutants in various tissues like the eye. However, clonal assays based on mitotic recombination also have drawbacks, including competition with neighboring wild-type tissue resulting in small populations of mutant cells that are difficult to study. Additionally, generating sister cells with different genotypes via mitotic recombination requires cell division, restricting the analysis of histone mutant cells to growing tissues at certain developmental stages. Moreover, histone mutant phenotypes currently can only be assessed in post mitotic tissues when the homozygous mutant animal is viable.

To solve these problems, we designed ΔHisC^cadillac to contain recognition sites for the B2 and B3 site-specific recombinases (hereafter B2R and B3R, respectively). These recombinases were originally identified and characterized in yeast and then subsequently engineered for use in Drosophila (Toh-e and Utatsu 1985; Utatsu et al. 1987; Nern et al. 2011; Williams et al. 2019). Importantly, B2R and B3R were shown to have robust activity at their specific recognition sites, B2S and B3S, respectively, with no cross reactivity with other recognition sites, including the FLP recognition site, FRT (Nern et al. 2011). The specificity of these enzymes allows for the utilization of B2S, B3S, and FRT sites in a single genotype. In ΔHisC^cadillac, the distal attP site and Act5C-dsRed expression cassette are flanked by B3S sites, while the entire ΔHisC^cadillac locus, including Act5C-dsRed and both attP sites, is flanked by B2S sites (Fig. 3e). Thus, expression of B3R will excise the distal attP (Fig. 7a), while expression of B2R will result in the excision of both attP sites (Fig. 7b). Each excision event removes Act5C-dsRed resulting in loss of dsRed expression. To test the efficiency of the B2S/B2R and B3S/B3R in excising Act5C-dsRed, we used the UAS/Gal4 system to express either recombinase in third instar wing imaginal discs heterozygous for an empty ΔHisC^cadillac. When expressing B3R in the anterior compartment using ci-Gal4 (marked by GFP + cells) we found efficient loss of dsRed expression (Fig. 7c, Supplementary Fig. 2b). Likewise, when expressing B2R with the same driver (marked by GFP + cells) we also found loss of dsRed (Fig. 7d). Together these results indicate that the B2S/B2R and B3S/B2R pairs work efficiently at the ΔHisC^cadillac locus.

We designed ΔHisC^cadillac with the intention of integrating a wild-type histone gene array at the distal attP site and a mutant histone gene array at the proximal attP site (Supplementary Fig. 3a, b). This arrangement of integrated histone gene arrays would allow for the removal of the wild-type histone gene array via B3 excision, leaving intact the mutant histone gene array at the proximal attP site and the generation of histone mutant tissue (Supplementary Fig. 3c). Note that a similar situation can be generated using mutant histone gene arrays integrated at an ectopic location like VK33 on chromosome 3 (Supplementary Fig. 3d). To test this premise, we asked whether a histone gene array integrated in ΔHisC^cadillac could be efficiently excised. We expressed B2R in the eye disc using ey-Gal4 in an animal with a 6× DWT histone gene array integrated in the distal attP site of ΔHisC^cadillac (Fig. 7e, Supplementary Fig. 2c). Large portions of these eye discs lost dsRed expression, marking the removal of the 6× DWT histone gene array (Fig. 7f). However, we found that excision of 6× DWT was not as efficient as removal of the empty 5 kb cassette, as clones of dsRed-positive cells remained (Fig. 7f). Insertion of a 6× wild-type histone array into ΔHisC^cadillac increases the linear genomic distance between the B2S sites by 8-fold (Fig. 7e). Previous data suggest that efficiency of FLP recombination decreases as the linear genomic distance between FRT sites increases (Golic and Golic 1996). Thus, the reduced efficiency of 6× DWT excision likely results from the 41 kb distance between B2S in this experiment. Although the B2 and B3 recombinase systems are functional in our system, further work is required to fine tune the efficiency of excision of larger insertions to optimize tissue-specific removal of histone gene arrays.

RMCE of histone gene arrays into ΔHisC^cadillac

The ΔHisC^cadillac locus contains two co-linear attP sites, offering distinct integration sites for one or two attB-containing BACs carrying a histone gene array. These attP sites also were designed to be amenable to RMCE, where each genomic attP is paired with one of two BAC attB sites (Fig. 8a) (Schlake and Bode 1994; Bateman et al. 2006). This strategy may be desirable in some contexts because it reduces the amount of exogenous DNA at the engineered locus (Fig. 8b). To test whether ΔHisC^cadillac was competent for RMCE with histone gene arrays we designed a new vector. Due to the positioning of the attP sites relative to Actin5c-dsRed in ΔHisC^cadillac, we chose to take advantage of a drug-selection marker to expedite the isolation of transformants. We re-designed the pMultiBAC vector (McKay et al. 2015) to include an hsp40-nptII gene, a single B3S, PaqCI restriction enzyme sites for Golden Gate cloning (described below), and two co-linear attB sites. We named this new BAC-based vector pCadillBAC (Fig. 8a). The hsp40-nptII confers resistance to food-supplemented G418 sulfate. When a mixed population of larvae consume food containing G418 sulfate, only those expressing nptII will develop to adulthood (Matinyan, Gonzalez, et al. 2021, Matinyan, Karkhanis, et al. 2021). Growing mixed populations of larvae on food with supplemented G418 sulfate without heat-shock was sufficient to select against any flies lacking either a partial (one attP/attB integration, dsRed+) or full RMCE (dsRed−) of a 4×DWT histone gene array. These data demonstrate that the ΔHisC^cadillac is amenable to RMCE and that drug-based selection methods can be used to supplement or replace visible marker-based methods.

Fig. 8.

Recombination-mediated cassette exchange (RMCE) of histone gene arrays into ΔHisC^cadillac and rapid generation of histone gene array BACs. a) Diagram of RMCE between pCadillBAC and ΔHisC^cadillac. pCadillBAC is designed with two co-linear attB sites, an hsp40-nptII selection marker, a B3 site for subsequent selection marker excision, and Golden Gate Assembly-compatible sequences (not shown). b) The resulting locus after RMCE contains little extraneous DNA sequence. c) Diagram of the generation of a 4× histone gene array by Golden Gate assembly through the PCR-generation of 1× histone repeat units (green), reversible cutting of fragments and pCadillBAC, and the final expected product containing the fragments. d) Agarose DNA gel showing generation of each 1× histone gene unit (left) and an agarose DNA gel showing a diagnostic restriction enzyme digest of resultant 3× and 4× histone gene arrays in pCadillBAC.

Rapid generation of BAC-based histone gene arrays

A bottleneck to generating histone mutant genotypes in all existing systems is the synthesis and sequence confirmation of the large plasmids or BACs containing arrays of histone gene units. We asked whether alternative strategies to traditional restriction enzyme and ligase cloning could be used to generate histone gene arrays. Therefore, we turned to the “one-pot” Golden Gate cloning strategy (Engler et al. 2008; Pryor et al. 2022a). This strategy utilizes a Type IIS PaqCI restriction enzyme, which generates ends with overhangs of unique sequence because the cut site is distinct from and adjacent to the recognition site. Fragments are designed with compatible ends that when ligated are refractory to re-cutting by PaqCI allowing the assembly and defined arrangement of multiple fragments into single constructs in one reaction (Fig. 8d). We designed primers with short indexes to individually PCR-amplify 1× histone gene units (Fig. 8e). These PCR reactions generate fragments with compatible ends to either the next array in the sequence or to pCadillBAC. As a result of the PaqCI-ligase reaction, circularized BACs with histone gene arrays are produced (Fig. 8d). Via this method, we efficiently produced 4x histone gene arrays in pCadillBAC for RMCE into the ΔHisC^cadillac locus (Fig. 8e). Using the pCadillBAC-4×DWT generated with this strategy we successfully recovered transformants that had undergone RMCE at the ΔHisC^cadillac locus. Although we have yet to generate larger histone gene arrays by Golden Gate cloning, likely because of compounded ligation inefficiencies as the number of fragments in the reaction increases, these data provide proof of principle that this rapid cloning strategy can be further developed.

Discussion

Delineating direct vs indirect phenotypic effects resulting from mutating histone-modifying enzymes and complexes is challenging due to the multitude of substrates targeted by these factors. The approach of mutating histone residues directly is, therefore, a critical capability for more deeply understanding the effects of histone PTMs. Drosophila melanogaster is the exemplar animal model organism for studying histone biology with many amenable natural features and engineered tools. Histone replacement platforms were developed by several groups and have been widely used to study different histone residues (Günesdogan et al. 2010; McKay et al. 2015; Zhang et al. 2019; Corcoran and Jacob 2023). Each of these platforms has drawbacks that restrict the types of research questions that can be explored using them. Therefore, we developed a new designer histone locus with expanded capabilities and flexibility. We intend for ΔHisC^cadillac to provide a one-stop-shop for researchers interested in chromatin biology with the flexibility to be customized through integration or modification of the elements.

Our novel histone deletion allele provides several improvements over previous strategies. By using fluorescent markers rather than the white gene, we engineered a high-throughput clonal assay in the adult eye to study the effects of histone mutations on cell proliferation and other phenotypes. By modulating H4^K16R clone size via mutation of another gene (i.e. His4r), we demonstrate that ΔHisC^cadillac provides a means to employ forward genetics to study histone biology, a first for any animal experimental system. We demonstrated that the attP sites in ΔHisC^cadillac are amenable to proximal, distal, and double integrations of multiple sizes of histone arrays. To reduce the complexity of generating histone mutant arrays and isolating transformants, we developed a new BAC (pCadillBAC) that is compatible with Golden Gate cloning and includes a drug-selectable marker. This BAC is also compatible with RMCE, eliminating extraneous BAC sequence from being added to ΔHisC^cadillac. We also demonstrate site-specific intragenic recombination at ΔHisC^cadillac using B2 and B3 recombinases. All these functionalities create an opportunity to generate many different histone genotypes in tissue- or temporal-specific ways.

Notably our 6× and 12× DWT integrations represent the first time a transgenic histone array larger than 5× has been integrated into the endogenous HisC locus on chromosome 2L. Integration of the transgenic histone gene arrays into ΔHisC^cadillac is important for mitigating any position effects of ectopic integration sites and to simplify downstream genetic crosses. This system's full utility will provide the ability to study histone PTM function in different developmental contexts, especially when mutation of histone residues would otherwise result in animal lethality. Furthermore, this system also provides the possibility of investigating histone PTM function in the early embryo and post mitotic cells through depletion of wild-type histones.

The ΔHisC^cadillac system is intended to be modular at all levels, allowing for improvements and expansion as new ideas and technologies are developed. The ΔHisC^cadillac locus provides expansion of functionality compared with current histone replacement systems and offers a template for further modulation to fit the needs of individual experiments. We encourage the broader chromatin community to use and modify the ΔHisC^cadillac system as a resource to further our understanding of epigenetics.

Data availability

Strains and plasmids are available upon request or in publicly accessible repositories: Bloomington Drosophila Stock Center (see Fly stocks and husbandry) and Addgene (pCFD4-U6:1_U6:3tandemgRNAs: Addgene plasmid # 49411). The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables. Code used to analyze long-read data is found at https://doi.org/10.5281/zenodo.11060915 and https://doi.org/10.5281/zenodo.11050973. The annotated pCadillBAC, VK33, HisC, ΔHisC, and ΔHisC^cadillac maps used in this study are found at: https://github.com/DuronioLab/Crain-Nevil-Genetics2024. Oxford Nanopore long reads generated by this study can be accessed from NCBI SRA (BioprojectID: PRJNA1133770).

Supplemental material available at GENETICS online.

Funding

This work was supported by National Institutes of Health T32-GM007092 to A.T.C., National Institutes of Health K12-GM000678 to M.N., National Institutes of Health R35-GM145258 to R.J.D., R35-GM136435 to A.G.M., and R35-GM128851 to D.J.M.