De novo assembly and annotation of the genome of Clogmia albipunctata reveals multiple lineage-specific duplications of early segmentation gene homologs

We generated a high quality de novo genome assembly of Clogmia albipunctata with 6 chromosome-size scaffolds (Fig 1, Table 1, S1 Appendix), consistent with the known chromosome number in this species [87,88]. The total length of the chromosome-size scaffolds is 304.3 Mb, close to genome size estimates based on flow cytometry data (316.6 Mb) [89]. An additional 268 small scaffolds (<1 MB) contained mostly repetitive sequences. We annotated this new reference genome using RNA-seq transcripts from embryonic, larval, pupal, and adult stages (S1 Table), protein sequences, and ab initio gene prediction software, and identified 15,047 protein coding genes. The annotated genome recovered 88%−94% of Complete Benchmarking Universal Single-Copy Orthologs (BUSCO) (Table 2), in line with recently published dipteran genomes [90–92]. While synteny between the chromosomes within the Cyclorrhapha clade can be highly conserved (time of divergence ~145 million years ago) [91], synteny between the chromosomes of D. melanogaster and C. albipunctata (divergence time ~250 million years ago) was essentially lost (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Clogmia albipunctata reference assembly statistics and quality metrics.

https://doi.org/10.1371/journal.pbio.3003896.t001

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Clogmia albipunctata annotation statistics and BUSCO scores.

https://doi.org/10.1371/journal.pbio.3003896.t002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Synteny analysis of Drosophila melanogaster chromosomes and Clogmia albipunctata scaffolds.

Synteny analysis between Clogmia albipunctata chromosomes-sized scaffolds and Drosophila melanogaster chromosomes. Groups of collinear genes are represented by contact lines connecting the positions shared between D. melanogaster chromosomes and C. albipunctata scaffolds. Colors correspond to D. melanogaster chromosomes. Positions of developmental genes of interest are indicated by solid purple bars. Position of the HOX gene cluster indicated by a solid brown bar.

https://doi.org/10.1371/journal.pbio.3003896.g001

Next, we interrogated the genome for the presence of predicted early segmentation genes (S2 Appendix). As expected, no bicoid ortholog was found but we recovered two putative Hox class 3 genes to which bicoid belongs [95], including the previously reported homolog of zerknüllt (Cal-zen) [96] on scaffold 7 and a more diverged potential Cal-zen paralog in position 3 of the Hox gene complex on scaffold 3 (evm.TU.scaffold_3.2943). Our annotation also included homologs of Drosophila melanogaster’s early segmentation genes, including gap genes (tailless, huckebein, orthodenticle/ocelliless, empty spiracles, buttonhead, hunchback, Krüppel, giant, and knirps), pair-rule genes (even-skipped, fushi tarazu, odd-skipped, runt, sloppy-paired, paired, hairy, and odd-paired), and other important regulators of the segmentation network (caudal, Dichaete) [20,22]. Some of these genes have more than one copy in the Clogmia genome, including 2 copies of hunchback (Cal-hb1, Cal-hb2), knirps/knirps-like (Cal-kni/knrl1, Cal-kni/knrl2), orthodenticle (Cal-otd1, Cal-otd2), even-skipped (Cal-eve1, Cal-eve2), tailless (Cal-tll1, Cal-tll2) and caudal (Cal-cad 1, Cal-cad2), 3 copies of sloppy-paired1/2/fd19B (Cal-slp1, Cal-slp2, and Cal-slp3), and 4 copies of empty spiracles (Cal-ems1, Cal-ems2, Cal-ems3, and Cal-ems4) (Fig 1, S2 Table). Such gene duplications are a potential source of redundancies in Clogmia’s segmentation gene network (see below).

Patterns of chromatin opening and closing during nuclear cleavage cycles in Clogmia suggest coordinated regulation of groups of genes

To identify accessible chromatin before and during the main wave of ZGA, we performed single-embryo ATAC-seq at NC11, NC12, NC13, and NC14 (before cellularization), and Late NC14 ~20 min before gastrulation. The embryos were staged in vivo using developmental time after egg activation at 25 °C and morphological criteria as previously described [83] and validated under our laboratory conditions. At least three replicates were performed for NC11 though NC14, and two replicates were performed for Late NC14. We pooled the ATAC-seq data from all stages to create a master peak list of 32,157 open chromatin regions (peaks), using MACS2 [97]. We also computed the number of peaks for each developmental stage excluding stage-specific peaks that were not recovered as significant in the master peak list. In this analysis, we found that the overall number of peaks increases with each NC, suggesting that the chromatin landscape is continuously changing (Table 3, S3 Appendix). Since promoter accessibility does not necessarily indicate gene expression [98], we divided the accessible chromatin regions into transcription start site (TSS) proximal regions (within 500 bp of TSS) and TSS distal regions (>500 bp) to facilitate the distinction of open promoters and other open cis-regulatory elements. ATAC-seq peaks that only mapped to intergenic or intronic sequence were treated as candidate enhancers or silencers (henceforth collectively referred to as enhancers). The relative proportion of candidate enhancers increased by about 10% from NC11 (46.8%) to NC14 (58%) and then decreased from NC14 to Late NC14 (52.7%) (S1 Fig). These observations suggest similar dynamic changes in chromatin accessibility during ZGA in Clogmia and Drosophila [99,100].

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Comparison of chromatin accessibility peaks, nuclear cycle length in minutes, and haploid genome size of Clogmia albipunctata and Drosophila melanogaster.

https://doi.org/10.1371/journal.pbio.3003896.t003

To identify peaks with dynamic patterns of chromatin accessibility, we compared the degree of accessibility of each peak between successive timepoints (Fig 2A) using DESeq2 [101]. This analysis identified a total of 6,930 dynamically regulated peaks (adjusted p-value ≤ 0.05 and |log₂ fold change| ≥ 1; 21.6% of all 32,157 peaks) and enabled us to quantitatively determine the behavior of each peak (i.e., maintaining, gaining, or losing accessibility) between consecutive developmental stages. We then clustered dynamic peaks into groups based on shared changes in accessibility from stage to stage and delineated 21 dynamic groups (S2 Fig). Nearly 80% (5,403/ 6,930) of the dynamic peaks fell into one of 8 groups, which reveal the major accessibility trends we observe in the data (Fig 2B). Many of the dynamic peaks closed (14.6%; Groups 5, 13) or opened (20.4%; Groups 8, 12) gradually over time. The remaining dynamic peak groups exhibited more complex dynamics between NC11 and Late NC14, commonly with an inflection at NC13 (e.g., Groups 3, 1, 15, 14; see also S2 Fig). These observations suggest that remodeling of chromatin accessibility (both opening and closing) between NC11 and Late NC14 at individual peaks often follows patterns that are shared by many peaks and may reflect co-regulation of the associated genes.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Changes in chromatin accessibility during NC11 to Late NC14.

A) Volcano plots of differentially accessible ATAC-seq peaks between consecutive developmental stages. Significant change of adjusted p-value ≤ 0.05 and |log2 fold change| ≥ 1are highlighted in red (log2 fold change ≤ −1) or blue (log2 fold change ≥ 1). B) Dynamic peak groups were identified using DEGreport (Pantano, 2017) (R package version 1.38.5). Y-axis, Z-score of abundance. X-axis, developmental stage. Underlying data is available in S1 Data and NCBI (bioproject: PRJNA1198932).

https://doi.org/10.1371/journal.pbio.3003896.g002

Cal-zld dependent gene expression and chromatin accessibility

Previous studies in Drosophila have shown that temporal changes in chromatin accessibility are regulated by the interplay of TFs that can interact with nucleosome-bound DNA (pioneer factors) and TFs that have limited or no ability to initiate chromatin accessibility [47,54]. The earliest changes in chromatin accessibility are primarily driven by the pioneer factor Zld [44,52,56,99,102]. As ZGA accelerates, additional pioneer factors, such as GAF [61], CLAMP [59,60], and Opa [76,86] are required for maintaining, modulating, or gaining chromatin accessibility. In Clogmia, a motif enrichment analysis of peaks with the MEME software package [103,104] revealed an enrichment of a Zelda-binding motif (5′-CAGGTA), GA-rich motifs recognized by GAF and CLAMP [61,105–107], in addition to (CA)_n motifs recognized by Combgap (Cg) [108], and others (S3 Fig). Opa-binding motifs were not recovered in this analysis. These results are comparable to findings in Drosophila [76,79,86] and suggest that motif enrichment within chromatin accessible at ZGA is very similar between Drosophila and Clogmia.

In early Drosophila embryos, loss of Zld causes widespread reduction of chromatin accessibility and failure to activate a large cohort of early zygotic genes, including genes required for pattern formation and cellularization [44,52]. Both motif affinity and occupancy positively correlate with Zld’s pioneering activity [44,52,54,109]. Homologs of zelda have been identified in many insects and some crustaceans [110], and several transcriptomic studies suggest that Zld’s function as pioneer TF during ZGA is conserved across insects [111–115]. However, these comparative studies did not directly test for Zld-dependent effects on chromatin accessibility. We wanted to know how zelda might affect chromatin accessibility in regions that open or close as the Clogmia embryo reaches NC14. We identified a single ortholog of zelda in the Clogmia genome (Cal-zld; S4 Fig) and observed high levels of Cal-zld expression at preblastoderm and blastoderm stages (S5 Fig). To examine phenotypic effects of Cal-zld depletion, we injected batches of ~30 embryos with Cal-zld dsRNA (double-stranded RNA) within the first hour of development (NC2/3) and monitored their development in vivo until the end of NC14 (roughly 7–7.5 hours after injection). Preblastoderm mortality of injected embryos (33/59; 56%) was higher than in the uninjected controls (6/22; 27%), but this difference may have been caused by unspecific side effects of the injection procedure. Across three independent replicates, a subset of the injected embryos reached NC14 but failed to cellularize and developed abnormally (S1 Movie). This phenotype is comparable to the phenotype of zld mutant Drosophila embryos (S2 Movie) [56] and suggests that Zld performs similar developmental roles in both species.

To assess Cal-zld’s role in shaping chromatin accessibility and zygotic transcription, we adapted our ATAC-seq protocol to isolate RNA from single-embryo ATAC-seq library preparations and used the RNA-seq data to quantify Cal-zld expression levels in Late NC14 Cal-zld RNAi embryos and uninjected control embryos (S6 Fig and see Materials and methods). We then compared gene expression (RNA-seq data) and chromatin accessibility (ATAC-seq data) between RNAi embryos with knockdown efficiencies greater than 75% (n = 4; zldKD group), RNAi embryos in which no knockdown of Cal-zld was detected (n = 3; zldnoKD group), and uninjected embryos that were otherwise treated like the RNAi embryos (n = 5; alignment control group, AC) using DESeq2. To better discern effects of Cal-zld knockdown on the initiation of zygotic transcription, we distinguished genes with maternal expression (227), genes with maternal and zygotic expression (6658), and genes with only zygotic expression (691) (Fig 3, S4 Appendix). As expected, only very few maternal genes were affected by knockdown of Cal-zld. In contrast, genes with maternal and zygotic expression were predominantly upregulated (1.7-fold in zldKD-vs-AC; 2.1-fold in zldKD-vs-zldnoKD) while genes with strictly zygotic expression were predominantly downregulated (3.3-fold in zldKD-vs-AC; 13.7-fold in zldKD-vs-zldnoKD). While the predominant upregulation of many genes with maternal and zygotic expression could reflect indirect posttranscriptional roles of Cal-zld in their regulation, the predominant downregulation of strictly zygotic genes in response to reduced Cal-zld levels is consistent with an important role of Cal-Zld in their transcriptional activation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Cal-Zld promotes zygotic transcription.

Individual panels show the number of downregulated and upregulated genes when comparing Cal-zld RNAi embryos with efficient Cal-zld knockdown to alignment controls (zldKD vs. AC) or to Cal-zld RNAi embryos without Cal-zld knockdown (zldKD vs. zldnoKD). The comparison of Cal-zld RNAi embryos without Cal-zld knockdown to alignment controls (zldnoKD vs. AC) serves as a negative control. On the y-axis, genes are grouped based on maternal, maternal and zygotic, or zygotic expression (see Materials and methods). Log₂ fold change is shown on the x-axis. Only genes with an adjusted p-value ≤ 0.05 and |log₂ fold change| ≥ 1 are shown. Venn diagrams show overlap of downregulated zygotic genes between zldKD vs. zldnoKD (pink) and zldKD vs. AC (blue) or zldnoKD vs. AC (green). Underlying data is available in S2 Data and NCBI (bioproject: PRJNA1198932).

https://doi.org/10.1371/journal.pbio.3003896.g003

At the level of chromatin accessibility, (Figs 4, S7, S5 Appendix), 4,102 regions (12.8% of all 32,157 ATAC-seq peaks) were affected in the zldKD-vs-AC comparison, and 2,145 peaks (8.6% of all peaks) were affected in the zldKD-vs-zldnoKD comparison. In both cases, the majority of the peaks lost accessibility (1.4-fold difference in zldKD-vs-AC; 3.2-fold difference in zldKD-vs-zldnoKD; 5.7-fold differences between consistently downregulated and upregulated peaks). These observations should be compared to our negative control (zldnoKD-vs-AC), in which only 2.3% of all peaks exhibited differential accessibility and no enrichment of downregulated peaks was observed (Fig 4). Depletion of Cal-zld mRNA affected both dynamic peaks and nondynamic peaks. Dynamic peaks that lost accessibility in response to Cal-zld knockdown mostly belonged to groups that opened before late NC14 (e.g., groups 8, 15, 14, 12, 6, 10; S8 Fig). Taken together, these findings suggest that Cal-Zld plays an important role in opening chromatin while a limited additional role in closing select chromatin regions cannot be ruled out.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Cal-Zld promotes chromatin accessibility.

Volcano plots of differentially accessible chromatin regions (ATAC-seq peaks) between Cal-zld RNAi embryos with efficient Cal-zld knockdown and alignment controls (zldKD vs. AC) or Cal-zld RNAi embryos without Cal-zld knockdown (zldKD vs. zldnoKD). The comparison of Cal-zld RNAi embryos without Cal-zld knockdown to alignment controls (zldnoKD vs. AC) serves as a negative control. Significant change of adjusted p-value ≤ 0.05 and |log₂ fold change| ≥ 1 are highlighted in red (log₂ fold change ≤ −1) or blue (log₂ fold change ≥ 1). The inserted pie charts show the proportion of peaks with at least one Zld motif (black) as reported in FlyFactorSurvey or inferred from the discriminative MEME analysis on NC11 peaks in wild-type Clogmia embryos (cf. position weight matrix logos). Underlying data is available in S3 Data and NCBI (bioproject: PRJNA1198932).

https://doi.org/10.1371/journal.pbio.3003896.g004

To assess if the changes in accessibility were a direct or indirect consequence of Cal-zld knockdown, we performed two types of motif analyses. First, we performed a differential motif enrichment analysis with MEME on Cal-zld sensitive peaks to identify motifs enriched centrally within 50 bp of the peak summit (S9 Fig). Relative to all nonsensitive peaks, the Zld motif was enriched in peaks that lost accessibility following Cal-zld knockdown (zldKD versus AC: 5-CAGGTA, E = 1.9e⁻⁷⁶; zldKD versus zldnoKD: 5-CAGGTA, E = 1.9e⁻⁸⁹), consistent with a direct role of Cal-Zld in nucleosome depletion and chromatin opening. Conversely, the Zld motif was not enriched in peaks that gained accessibility. We then extended the search for Cal-Zld motifs to the full width of each peak, using Zld motifs derived from our MEME analysis in Clogmia and experimental data in Drosophila melanogaster [116]. 81.4% of the 2,399 downregulated peaks in zldKD-vs-AC and 92.1% of the 1,632 downregulated peaks in zldKD-vs-zldnoKD contained at least one Zld motif compared to ~60% of the respective upregulated peaks or the Cal-Zld-insensitive peaks. This difference was slightly more pronounced when restricting the comparison to the 1,091 consistently downregulated peaks (93.3% with at least one Zld motif; S7A Fig) and the 191 consistently upregulated peaks (47.1% with at least one Zld motif; S7B Fig). These findings suggest that Cal-Zld plays a direct role in opening chromatin. In support of this hypothesis, we recall that both motif affinity and occupancy positively correlate with Zld’s pioneering activity in Drosophila [54,109]. Conversely, the closure of many chromatin regions that we observed during the final cleavage cycles in wild-type Clogmia embryos may generally occur without direct Cal-Zld input, although a direct role in closing some of these regions cannot be ruled out since 90 of the 191 consistently upregulated peaks in the zldKD-vs-AC and zldKD-vs-zldnoKD comparisons contain at least one Zld motif.

Maternal Cal-Opa drives chromatin accessibility and expression at homeobrain and sloppy-paired loci in complementary anterior domains

The early zygotic segmentation gene network of dipterans is dominated by TF-encoding genes, such as the gap genes and pair-rule genes [20,22,30,80,117–120]. Bcd breaks axial symmetry by targeting these genes in a concentration-dependent manner throughout the blastoderm stage, including NC14. In contrast, maternal Cal-opa exerts its symmetry-breaking function in axis specification prior to the onset of zygotic Cal-opa expression during blastoderm cellularization at mid NC14 [3]. Therefore, we focused our search of Cal-Opa targets on TF genes that are expressed before NC14 and asked whether maternal Cal-opa activity affects their expression and chromatin accessibility during NC12 and NC13.

To assess the impact of maternal Cal-Opa activity on gene expression and chromatin accessibility, we performed RNA-seq and ATAC-seq on single Cal-opa RNAi embryos (6 replicates for NC12, 8 replicates for NC13) and selected four NC12 embryos and five NC13 embryos with knockdown efficiencies greater than 75% (NC12_opaKD, NC13_opaKD; S10 Fig and Materials and methods). At NC13, we also identified two RNAi embryos without Cal-opa knockdown (NC13_opanoKD) and used them as a negative control group. In parallel, we processed stage-matched embryos that were prepared for injection under oil but not injected (alignment controls, AC; 4 replicates for each developmental stage), and embryos that were injected with dsRNA of the extraneous gene DsRed (injection controls, IC; 3 replicates for each stage). We then used DESeq2 to compare gene expression (RNA-seq peaks; Fig 5, S6 Appendix) and chromatin accessibility (ATAC-seq peaks; Fig 6, S7 Appendix) at NC12 and NC13 between embryos of the opaKD groups and their respective control groups (AC and IC for NC12; AC, IC, and opanoKD for NC13). At NC12, opaKD-vs-AC identified 79 downregulated genes (including 11 TF genes) and 5 upregulated genes (no TF genes) while opaKD-vs-IC identified only 22 downregulated genes (including 5 TF genes) and no upregulated genes. An opanoKD group was not available for this stage. Cal-opa (positive control), Clogmia’s three sloppy-paired homologs (Cal-slp1, Cal-slp 2, Cal-slp3), and homeobrain (Cal-hbn) were the only consistently downregulated TF genes at NC12 (Fig 5). At NC13, the opaKD-vs-AC comparison yielded 561 downregulated genes (including 67 TF genes) and 486 upregulated genes (including 49 TF genes). However, far fewer down- and upregulated genes were identified when the opaKD group was compared to the injected control groups (Fig 5). In opaKD-vs-IC, 27 genes were downregulated (including 5 TF genes: Cal-hbn, Cal-slp 1, Cal-slp2, Cal-hb1, Cal-hb2) and 17 were upregulated (including 1 TF gene). Downregulation of Cal-opa in the IC control group was not statistically significant (log₁₀ adjusted p-value of 0.1169), falling slightly above our significance cutoff: log₁₀ adjusted p-value of < 0,05. This observation likely reflects the low detection of Cal-opa transcript in one of the three embryos in the injection control group (S10 Fig). In opaKD-vs-opanoKD, only 13 genes were downregulated (including 6 TF genes: Cal-hbn, Cal-slp1, Cal-slp2, Cal-hb1, Cal-hb2, and Cal-opa) and 3 genes were upregulated (0 TF genes). These results should be compared to our negative control for NC13 (opanoKD-vs-AC), which only yielded a random set of 3 slightly downregulated genes (0 TF genes) and no upregulated genes (S11A Fig). Taken together, our results suggest that maternal Cal-opa promotes the activation of Cal-slp1, Cal-slp 2, Cal-slp3, and Cal-hbn at NC12 and of Cal-slp1, Cal-slp2, Cal-hbn, Cal-hb1, and Cal-hb2 at NC13. Of these genes, only the loci of Cal-slp1, Cal-slp 2, and Cal-hbn were associated with ATAC-seq peaks that spanned Opa motifs (S7 Appendix).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Maternal Cal-opa activates a small set of transcription factor genes at NC12 and NC13.

Volcano plots of differentially expressed genes between embryo groups with reduced Cal-opa expression (opaKD) and control groups (AC and IC for NC12; AC, IC, and opanoKD for NC13). See S11A Fig for a negative control (opanoKD vs. AC at NC13). Significant changes of adjusted p-value ≤ 0.05 and |log₂ fold change| ≥ 1 are highlighted in red (log₂ fold change ≤ −1) and blue (log₂ fold change ≥ 1). Consistently affected TF genes are labeled. Underlying data is available in S4 Data and NCBI (bioproject: PRJNA1198932).

https://doi.org/10.1371/journal.pbio.3003896.g005

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Cal-Opa is required for establishing a selective set of chromatin accessibility peaks.

Volcano plots of differentially accessible chromatin regions (ATAC-seq peaks) between Cal-opa RNAi embryos with reduced Cal-opa expression (opaKD) and control groups (AC and IC for NC12; AC, IC, and opanoKD for NC13). See S11B Fig for negative control (opanoKD vs. AC). Significant changes of adjusted p-value ≤ 0.05 and |log₂ fold change| ≥ 1 are highlighted in red (log₂ fold change ≤ −1) and blue (log₂ fold change ≥ 1). Underlying data is available in S5 Data and NCBI (bioproject: PRJNA1198932).

https://doi.org/10.1371/journal.pbio.3003896.g006

To assess the effects of Cal-opa knockdown on chromatin accessibility we performed DESeq2 analyses on the ATAC-seq peaks (Fig 6, S7 Appendix). At NC12, the opaKD-vs-AC comparison identified 79 downregulated and 7 upregulated peaks. Sixty-three of the 79 peaks downregulated peaks but none of the 7 upregulated peaks contained at least one Opa motif. The opaKD-vs-IC comparison only yielded 35 downregulated peaks of which 30 contained at least one Opa motif. The 32 consistently downregulated peaks were associated with the loci of Cal-slp1, Cal-hbn, Cal-opa, and various non-TF genes (Figs 7, 8, S7 Appendix). One additional ATAC-seq peak associated with Cal-slp2 narrowly missed our significance threshold in the opaKD-vs-IC comparison (adjusted p-value 0.062).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Expression and chromatin accessibility at the Cal-hbn locus at NC12 and NC13.

RNA-seq and ATAC-seq tracks (CPM) are shown for NC12 and NC13 with alignment control in blue, injection control in light blue, Cal-opa RNAi without KD in green, and Cal-opa RNAi with KD in red. Gene models are shown in pale yellow. ATAC-seq peaks with significant changes in accessibility are highlighted in pink and are marked with a plus sign if they contain at least one Opa motif. The scale of the y-axis is adjusted for each data type but kept constant between stages and is indicated with numbers in square brackets: [0-25] for RNA-seq data; [0-10] for ATAC-seq data. Underlying data is available in NCBI (bioproject: PRJNA1198932).

https://doi.org/10.1371/journal.pbio.3003896.g007

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Expression and chromatin accessibility at the Cal-slp1 and Cal-slp2 locus at NC12 and NC13.

RNA-seq and ATAC-seq tracks are shown for NC12 and NC13 with alignment control in blue, injection control in light blue, Cal-opa RNAi without KD in green, and Cal-opa RNAi with KD in red. Gene models are shown in pale yellow. ATAC-seq peaks with significant changes in accessibility are highlighted in pink and are marked with a plus sign if they contain at least one Opa motif. The scale of the y-axis (CPM) is adjusted for each data type but kept constant between stages and is indicated with numbers in square brackets: [0-100] for RNA-seq data; [0-25] for ATAC-seq data. Underlying data is available in NCBI (bioproject: PRJNA1198932).

https://doi.org/10.1371/journal.pbio.3003896.g008

At NC13, opaKD-vs-AC yielded 250 downregulated peaks and 42 upregulated peaks, OpaKD-vs-IC identified 11 downregulated peaks and 5 upregulated peaks, and opaKD-vs-opanoKD recovered only 5 downregulated peaks and 5 upregulated peaks without Opa motifs (Fig 6, S7 Appendix). Of the consistently downregulated peaks, two were associated with Cal-hbn (Fig 7), and two were associated with Cal-slp1 (Fig 8), all covering Opa motifs. The fifth consistently downregulated peak was associated with an uncharacterized gene without homology to known genes (evm.TU.scaffold_7.1426) and did not cover an Opa motif. Significant downregulation of peaks associated with Cal-hb1 and Cal-hb2 was only observed in comparisons with AC and IC controls at NC13, and no Opa motifs were detected in these peaks (S12 Fig). In our negative control (opanoKD-vs-AC), none of these ATAC-seq peaks were recovered (S11B Fig, S7 Appendix).

Given that maternal Cal-opa is required for zygotic transcription homeobrain and sloppy-paired homologs, we examined the expression of these genes in more detail. Maximum-likelihood trees with dipteran and nondipteran sloppy-paired homologs generated with IQ-TREE v.2.3.6. [121] and ModelFinder [122] suggest that Cal-slp1, Cal-slp2, and Cal-slp3 resulted from recent gene duplications in the Clogmia lineage (S13, S14 Figs). Zygotic transcription of Clogmia’s sloppy-paired and homeobrain homologs was detected during and after NC11 and was confined to the anterior half of the blastoderm (Fig 9). Cal-hbn was expressed in the anterior-most portion of the blastoderm, complementary to the overlapping expression domains of Cal-slp1, Cal-slp2, and Cal-slp3 (which was only expressed at low levels prior to NC14). In the case of Cal-slp2, we also detected high levels of maternal transcript, previously found to be distributed in a shallow anterior-to-posterior gradient [3]. The early timing and spatial expression of Clogmia’s homeobrain and sloppy-paired homologs suggest that these genes likely play an important zygotic role in the initiation of anterior pattern formation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 9. Expression of homeobrain and sloppy-paired homologs in Clogmia.

A) Expression levels (CPM) of Cal-hbn, Cal-slp1, Cal-slp2, and Cal-slp3 at 6 consecutive stages, including NC2/3 (maternal, ~45-min old embryos), NC11, NC12, NC13, NC14 and Late NC14. Gene models are shown in pale yellow. Synteny of the three sloppy-paired loci is indicated by three dots. The y-axis is the same for all tracks at each locus. Cal-hbn RNA-seq: 0-30 CPM. Cal-slp1-3 RNA-seq: 0-120 CPM. B) Whole-mount fluorescent in situ hybridizations at consecutive blastoderm stages. Embryos are shown with anterior left and dorsal up. Underlying data is available in NCBI (bioproject: PRJNA1198932).

https://doi.org/10.1371/journal.pbio.3003896.g009

Dynamic chromatin states in the embryo of Clogmia albipunctata

In this study, we have established the resources for dissecting developmental gene networks of early Clogmia embryos by providing an annotated chromosome-level genome sequence for our inbred laboratory strain and by describing chromatin accessibility of precisely staged consecutive blastoderm stages. We have used these resources to examine how the activities of zelda and maternal odd-paired shape chromatin accessibility and gene expression in Clogmia. We found that, as in Drosophila, the period spanning ZGA is accompanied by large-scale changes in chromatin accessibility. However, there are notable differences between Clogmia and Drosophila in the early patterns of establishment and maintenance of accessible states before NC14. In Clogmia, the chromatin landscape is highly dynamic in that gains as well as losses occur during the nuclear cleavage cycles of the blastoderm stage. In Drosophila, only gains in accessibility are observed until widespread ZGA and lengthening of the cell cycle at NC14 [74,76,77,79,123]. This difference could be a consequence of cell cycle timing. In Clogmia, NC11 through NC13 last ~3x longer than in Drosophila [83], in which NC11 only lasts ~10 min, NC12 ~12 min, and NC13 ~21 min [82]. The much shorter cell cycle times of Drosophila present significant limitations to establish and maintain chromatin states necessary for transcriptional activity [79]. Assuming that kinetic rates for nuclear import, DNA replication, and DNA binding of TFs is similar between Clogmia and Drosophila, the significantly longer syncytial interphase times in Clogmia could alleviate constraints on the initiation of zygotic transcription. However, the timing of AP patterning between the two species appears comparable [84,117], suggesting that the timing of ZGA might also be conserved. Alternatively, the longer syncytial interphase times in Clogmia could alleviate constraints on maintaining or silencing active chromatin states. In Drosophila, Polycomb group (PcG) proteins begin to establish heritable, transcriptionally silenced chromatin states during NC14 [124], but in Clogmia, the longer nuclear cleavage cycles may provide time for the maturation of repressive histone marks at accessible chromatin regions before NC14.

Role of Zld in regulating chromatin accessibility during Clogmia’s nuclear cleavage cycles

We found that reduced Cal-zld expression results in widespread losses of chromatin accessibility, consistent with a conserved role for Zld to pioneer chromatin accessibility within the context of ZGA. In Drosophila, Zld almost exclusively drives gains in chromatin accessibility during nuclear cleavage cycles [71,76,102,125,126]. However, two recent studies found that during ZGA, Zld activity directs how repressive chromatin marks are deposited. Loss of zld results in context-specific losses and gains of H3K27me3 within specific PcG domains [124]. Zld also enables histone acetylation at promoters and enhancers through its interaction with Chip, Drosophila’s homolog of LIM-domain-binding protein 1 (Ldb1), which in turn recruits the histone acetyltransferase CBP [127]. The longer nuclear cleavage cycles in Clogmia might enable similar roles of Cal-Zld before NC14. In this case, Clogmia and Drosophila may differ in the timing of when specific histone marks are deposited and affected by Zld depletion. Differences in the activities of Zld and Cal-Zld could also reflect differences in their protein structures or isoforms. For example, Cal-Zld contains an additional zinc finger that has been eroded in Drosophila (S4 Fig) [110]. Additionally, stage-specific isoforms of Zld have been described in Drosophila, including a variant lacking the C-terminal three zinc fingers [128–131]. This isoform functions as a dominant negative form in cell culture although flies in which the expression of this isoform is suppressed are viable and fertile [130]. Therefore, the molecular underpinnings of the functional readouts of Zld and Cal-Zld could differ and involve both direct and indirect mechanisms.

Functional comparison of the anterior determinants in Clogmia and Drosophila

Bcd contributes to the activation of many early segmentation genes; however, its role in the posterior embryo is much more subtle than in the anterior embryo, where all expression domains directly or indirectly depend on Bcd. Bcd target enhancers have been classified based on how their accessibility depends on Bcd and Zld [71]. The accessibility of Bcd target enhancers that drive expression in the posterior of the embryo is established independent of Bcd (e.g., Krüppel CD1, knirps posterior, giant posterior). However, enhancers that open in a Bcd-dependent manner are associated with genes expressed in the anterior of the embryo (e.g., orthodenticle, empty spiracles, buttonhead, giant, knirps, hunchback, sloppy-paired 1, fd19B, even-skipped, paired, cap’n’collar, huckebein, and homeobrain). Moreover, many correspond to enhancers that drive reporter gene expression in the anterior embryo including giant anterior (−6), giant anterior (−10), knirps anterior, orthodenticle, buttonhead, even-skipped stripe 1/5, hunchback P2, hunchback shadow enhancers, eve-skipped stripe 2, an early paired enhancer, cap’n’collar, and huckebein (reviewed in [71]). The majority of Bcd-dependent anterior expression domains first appear during NC13, with the noteworthy exception the P2 transcript of hunchback. The promoter of this transcript is in an open conformation as early NC10 [79] and drives Bcd-dependent expression throughout the anterior half starting at NC11 [45,66,132], thereby boosting the maternal Hb gradient which forms independent of Bcd [133].

In Clogmia, neither the expression nor chromatin accessibility at the Cal-hb1 and Cal-hb2 loci was reduced in NC12 opaKD embryos (S12 Fig) and maternal hunchback activity does not contribute to embryo polarity in this species, given that knockdown of maternal Cal-opa transcripts results in symmetrical double abdomens [3]. Additionally, homologs of several Bcd target genes were not expressed at NC12 and NC13 (e.g., Cal-gt, Cal-kni/knrl1-2, Cal-ems1-4, and Cal-hkb) or their expression levels at these stages were not affected by the knockdown of maternal Cal-opa activity (e.g., Cal-cnc, Cal-prd, and Cal-btd). However, maternal Cal-Opa was required for the zygotic transcription of sloppy-paired and homeobrain homologs in complementary, evolutionarily conserved domains across the anterior half of the embryo. In Drosophila, these genes are regulated by Bcd-bound enhancers whose accessibility is sensitive to the concentration of Bcd [134,135], even though neither of them has received much attention as Bcd target genes. Taken together, these findings suggest that the anterior determinant of Clogmia targets a small subset of the genes known as direct targets of Bcd in Drosophila.

Mechanistically, Cal-Opa could prime accessibility at target loci through a canonical pioneering mechanism which requires binding of nucleosome occupied DNA and promoting nucleosome instability [47,48,56,60,136,137]. A vertebrate homolog of Opa, Zic3, was found to bind nucleosomal DNA in vivo, suggesting that Cal-Opa may pioneer chromatin directly [138]. Alternatively, Cal-Opa binding may compete with nucleosomes to bind unoccupied DNA during each nuclear cleavage cycle [139], similar to Bcd [54,71,78].

Role of homeobrain and sloppy-paired in breaking axial symmetry in insects

It is conceivable that homeobrain nor sloppy-paired homologs serve a more broadly conserved role in the initiation of anterior pattern formation. In Drosophila, homeobrain mutant embryos lack the labrum and anterior portions of the brain [140–143], indicating that homeobrain is essential for specifying the most anterior portion of the embryo. A more extreme homeobrain loss-of-function phenotype has been observed in the beetle Tribolium castaneum, where this gene plays an important zygotic role in establishing the embryo’s AP polarity [144]. In Tribolium, knockdown of homeobrain resulted in loss of the entire head and—in extreme cases—duplications of the tail end. The duplicated abdomen was defective, and thoracic structures persisted in these embryos, but complete double abdomens could be obtained by simultaneously suppressing the expression Tc-zen1 (like bicoid a Hox3 derivative [35,95], thereby preventing the anterior repression of caudal, an important gene for posterior patterning [145]. Additionally, the Tribolium ortholog of odd-paired, which is expressed maternally and required for head development (including the labrum, antennae, and head capsule) [146], could potentially complement the strictly zygotic roles of Tc-zen and Tc-hbn in breaking symmetry along the AP axis of the Tribolium embryo.

The sloppy-paired gene, a member of the forkhead domain gene family [147], is best-known for its role as a pair-rule segmentation gene [22,81], but it is initially expressed in a head domain of the syncytial Drosophila embryo, along with its paralogs sloppy-paired 2 and fd19B [135,148–153]. The sloppy-paired paralogs, which are common across insects (S13, S14 Figs), may have obscured a fundamental role of sloppy-paired activity in initiating anterior pattern formation in Drosophila. This also applies to the two sloppy-paired homologs of Tribolium, in which only the gnathocephalon (mandibular, maxillary, and labial segment) and even-numbered trunk segments are lost when Tca-slp1 is knocked down [154,155]. We therefore propose that along with homeobrain, sloppy-paired may play a more widely conserved role in establishing the AP axis in insect embryos.

Limitations of the study

It is conceivable that homeobrain and sloppy-paired activities are sufficient for breaking symmetry along the AP axis of the Clogmia embryo. To test this hypothesis, it might be necessary to completely suppress the activities of four genes, including three with early zygotic expression (Cal-hbn, Cal-slp1, and to a lesser degree Cal-slp3) and one with anterior-enriched maternal transcript and zygotic expression (Cal-slp2). This is particularly challenging in Clogmia because this species did not tolerate more than one microinjection per embryo in our hands. Ongoing research in our laboratory seeks to overcome this technical issue by developing more effective knockout and knockdown reagents. Alternatively, maternal Cal-Opa could control the zygotic expression of additional genes. We did not observe other TF genes that were consistently downregulated in embryos with reduced maternal Cal-opa transcript, but this could be due to incomplete knockdown of the maternal Cal-opa transcript or limited statistical power of our opaKD-vs-IC and/or opaKD-vs-opanoKD comparisons or a combination of both. To further explore this possibility and to formally prove a direct interaction of maternal Cal-Opa with the identified target genes, we are developing protocols and reagents for Cal-Opa chromatin immunoprecipitation experiments. In the present study, the identification of Clogmia’s anterior determinant target genes rests on Opa-dependent chromatin accessibility along with the presence of Opa-binding motifs, Opa-dependent gene expression in the anterior blastoderm, and the early timing of transcriptional activation.

Clogmia culture and embryo collection

The Clogmia culture was maintained in 100 mm × 15 mm Petri Dishes (Fisher Scientific FB0875712) on moist, compressed cotton (Jorvet SKU: J0197) sprinkled with parsley leaf powder (Starwest Botanicals SKU: 209895-5). For embryo collections, ~2-day old females were transferred to round 250 mL glass bottles with a ~1-inch bottom layer of moistened and compressed cotton and mated at room temperature for 2–3 days. Female ovaries were dissected with a pair of fine forceps (Dumont 5) and eggs released into water for egg activation and embryogenesis. For staged egg collections, embryos were released into flat bottom glass 100 mm × 15 mm Petri Dishes (Corning CM: 101425,55) containing 15 ml of water. Care was taken to disperse the eggs on the bottom of the dish as crowding can delay development. Petri Dishes were covered with a lid and moved to an incubator set at 25 °C until the desired stage. Egg harvesting post activation was on average 47 min for NC2/3, 261 min for NC11, 292 min for NC12, 332 min for NC13, 401 min for NC14, and 465 min for Late NC14.

Embryo fixation

Embryos were dechorionated using a 50% dilution of commercial bleach (8.25% sodium hypochloride) for 60 s. Dechorionated embryos were fixed in a 50 mL falcon tube, using 20 mL of boiling salt/detergent-solution (100 μL 10% triton-X; 500 μL 28% NaCl; up to 20 mL of water). The mixture was gently swirled for 10 s. To stop the heat fixation, 30 mL of deionized water was added. Embryos were transferred to 1.5 mL reaction vials and devitellinized in a 1:1 mixture of n-heptane and methanol by vigorous shaking. This step was followed by 3 washes with cold methanol (stored at −80 °C). Embryos were inspected under a stereomicroscope and manual devitellinizations were performed as needed using sharp tungsten needles in a 3% agar plate covered with methanol. Devitellinized embryos were stored in 100% methanol at −20 °C.

Microinjections of embryos and RNAi efficiency

Embryo injection was done as previously described [3]. We note here that RNAi efficiency in Clogmia embryos varies [3]. This variability could be due to technical variables (e.g., injection needle, amount injected) or biological variables (e.g., differences in the RNAi response between individual embryos, or the general health or genetic background of the embryos or their mothers) or a combination of such effects. Therefore, we took care to control the age of the females and selected all experimental embryos from females with healthy looking ovaries and eggs. Embryos were injected within the first hour of development (NC2/3). To test Cal-zld dsRNA, we quantified knockdown of the Cal-zld transcript in individual embryos by RT-qPCR and performed fold change calculations of Cal-zld mRNA levels with the average level in uninjected control embryos as reference (S3 Table). In one quarter of these Cal-zld RNAi embryos (5/20), Cal-zld mRNA levels were reduced by ~30%–90%. For Cal-opa RNAi, we used dsRNA of the isoform-specific first exon of the maternal Cal-opa transcript and confirmed that it can induce the double abdomen phenotype. We also quantified knockdown efficiencies of the Cal-opa transcript in individual embryos by RT-qPCR and performed fold change calculations of Cal-opa mRNA levels with the average level in uninjected control embryos as reference (S4 Table). This analysis suggests that Cal-opa RNAi can nearly abolish the Cal-opa transcript.

dsRNA synthesis

A PCR template was made using primers specific for the target gene with T7 overhangs and either cDNA, gDNA, or a plasmid depending on the target sequence. The PCR product was run on a gel and purified using the Zymo Gel DNA Recovery Kit (D4007/D4008) following the manufactures instructions. The PCR template was used as the input for the transcription reaction producing dsRNA. dsRNA synthesis and purification were done using the Invitrogen MEGAscript RNAi Kit (AM1626) following the manufacturer’s instructions. Purified dsRNA was eluted in injection buffer (0.1 mM NaH₂PO₄ Ph 6.8; 5 mM KCl).

Forward and reverse primer sequences for dsRNA (lengths of dsRNAs in brackets; gene specific sequence of primers underlined):

Cal-opa, maternal transcript (201 bp):

5′-CAGAGATGCATAATACGACTCACTATAGGGAGAAAACAATTGTGAAGTGCGACA

5′-CAGAGATGCATAATACGACTCACTATAGGGAGACAAATTTCCAAACGATGACAGA

Cal-zld (a combination of two dsRNAs were used 579 bp and 668 bp):

5′-TAATACGACTCACTATAGGGAGAAGTCCCGCAATTGATACAGC

5′-TAATACGACTCACTATAGGGAGAAGGATGTTGGTGGACACTCC

5′-TAATACGACTCACTATAGGGAGACGGAATGGTGTGTGAAACAG

5′-TAATACGACTCACTATAGGGAGATTCGAGGGGTTATTGTCCTG

dsRed (273 bp):

5′-CAGAGATGCATAATACGACTCACTATAGATGCAGAAGAAGACTATGG

5′- CAGAGATGCATAATACGACTCACTATAGCTACAGGAACAGGTGGTG

RT-qPCR

RT-qPCR reactions were performed using the Luna Universal One-Step RT-qPCR Kit (New England Biolabs Ipswich, MA).

Forward and reverse primer sequences for targets (lengths of amplicon in brackets):

Cal-opa, maternal transcript (112 bp):

5′- TTGCGCTTCATTCTGTCATCG

5′- CACTGAGGGCTGATTCTCGG

Cal-zld (138 bp):

5′-AAAGATGAACCACCGTGCGA

5′-CATTGATGGCAGTGGACCCT

Cal-rpl35 (119 bp):

5′- CTGTCCAAAATCCGAGTCGT

5′- AGGTCGAGGGGCTTGTATTT

Fluorescent HCR in situ hybridizations

Gene specific probes and amplifier-fluorophore sets were designed and synthesized by Molecular Instruments Inc (Los Angeles, CA). The amplifier-fluorophores were used as follows: B1-647 for Cal-slp1, B2-546 for Cal-hbn and Cal-slp3, and B3-488 for Cal-slp2, and Cal-otd1. Buffers (probe hybridization, probe wash, and amplification buffers) throughout the protocol were purchased from Molecular Instruments. The HCR in situ hybridizations were performed as described [156,157]. Briefly, fixed embryos were rehydrated and washed 3 times in PBS-Tween20 (1x PBS, 0.1% Tween-20). Embryos were permeabilized for 30 min at room temperature (100 ml Permeabilization buffer: 100 µL 100% Triton X-100, 50 µL Igepal CA-630, 50 mg Sodium Deoxycholate (Sigma D6750-25G), 50 mg Saponin (Sigma 47036-50G-F), and 200 mg BSA Fraction V (Sigma A5611-5G). Embryos were then pre-hybridized in probe hybridization buffer for 30 min in a 37 °C water bath. Once pre-hybridization was complete, fresh hybridization solution containing gene specific probes was added, and the embryos were left to incubate overnight in a 37 °C water bath. On the second day the hybridization solution was removed, and the embryos were washed 4 times with pre-warmed probe wash buffer followed by two washes in 5x SSCT (5x SSC, 0.1% Tween-20). An amplifier hairpin solution was prepared following the manufacturer’s instructions. Embryos were incubated with pre-warmed amplification buffer for 30 min at room temperature before adding amplifier hairpin solution. Embryos were left to incubate overnight at room temperature in the dark. The following day the embryos were washed 5 times in 5x SSCT at room temperature. The embryos were then counter stained with DAPI and mounted in Aqua-Polymount for imaging.

Image acquisition and processing

Confocal imaging was performed on the Zeiss LSM 900 using a Plan-Apochromat 20x/0.8 M27 objective. Pixel-density was 1,024 x 1,024 and physical pixel size was X = 0.46980808984799 µm and Y = 0.46980808984799 µm. Excitation wavelengths were 561 nm for Alexa Fluor 546 with B2 amplifier (Cal-hbn/Cal-slp3; collection window: 550–645 nm), 640 nm for Alexa Fluor 647 with B1 amplifier (Cal-slp1; collection window: 645–700 nm), 488 nm for Alexa Fluor 488 with B3 amplifier (Cal-slp2; collection window: 410–550), and 405 nm for DAPI (collection window: 410–550 nm). Z-slice spacing was 1 µm. For each image a Z-stack of images was captured from the embryo’s surface to approximately halfway through the embryo. Then a maximal projection was created from this Z-stack using FIJI/Image J. To minimize background fluorescence, the display range minimum of each channel was slightly increased in all images (0–255 scale): 18 for 541 nm, 25 for 640 nm, and 40 for 488 nm. No further adjustments to the images were made.

ATAC-seq library preparation

Single-embryo ATAC-seq library preparations were done essentially as described previously in Drosophila melanogaster [76,79]. Embryos developed in an incubator at 25 °C until the desired NC. NC was confirmed by time after egg activation and the morphology was checked under water using a transmission light microscope (Zeiss Stemi 2000-C). Once at the desired NC individual embryos were placed in the lid of a 1.5 ml low-retention Eppendorf tube and homogenized in 10 μl of cold lysis buffer (10 mM Tris-HCl pH 7.4; 10 mM NaCl; 3 mM MgCl2; 1% Igepal CA-630) [158], using a fire-sealed microcapillary tube (Drummond Microcap 25). Once homogenized, 40 µl of lysis buffer was added to the lid, and the body of the tube was carefully closed over the lid. Samples were placed on ice until all embryos were processed. The nuclei were pelleted by centrifugation at 800 RCF at 4 °C for 10 min. Supernatant removal was observed under a stereo microscope to ensure that the nuclei pellet was not dislodged. The nuclear pellet was resuspended in 7.5 µl of Tagmentation buffer (Illumina Tagment DNA TDE1 Enzyme and Buffer Kit) and placed on ice until all samples were resuspended. Then 2.5 µl of Tn5 Transposase was added and mixed by pipetting up and down. Tagmentation and amplification of ATAC-seq libraries were performed as described previously [159]. The tagmentation reaction was performed at 37 °C and 800 RPM for 30 min using an Eppendorf Thermomixer. Immediately after this step, the reactions were purified using the Qiagen Minelute kit following the manufacturer’s instructions. Purified tagmented DNA was eluted in 10 µl of buffer EB. Library amplification was performed as described previously [159]. Typically, 12–14 total PCR were performed. Amplified libraries were purified using 1.8x Ampure SpRI beads following the manufacturer’s instructions. Purified ATAC-seq libraries were eluted in 15 µl of Qiagen elution buffer (10 mM Tris-Cl, pH 8.5). The ATAC-seq library profile was evaluated using Agilent high sensitivity bio-analyzer. Concentrations were estimated using a Qubit fluorometer. Libraries from the same experiment were pooled and sequenced together. Sequencing was done at the University of Chicago Genomics Core (Chicago, IL, USA) using the Illumina NextSeq (PE 75 bp) and at Admera Health (South Plainfield, NJ, USA) using the Illumina HiSeq (PE 150 bp) and the Illumina NovaSeq X Plus (PE 150 bp).

ATAC-seq libraries were amplified using a set of modified Buenrostro primers that introduced Unique Dual Indexes (UDI) [76]. The generalized UDI ATAC primer sequences are:

ATAC UDI Index Read 2 (i5): 5′- AATGATACGGCGACCACCGAGATCTACACnnnnnnnnTCGTCGGCAGCGTCAGATGT*G -3′

ATAC UDI Index Read 1 (i7): 5′- CAAGCAGAAGACGGCATACGAGATnnnnnnnnGTCTCGTGGGCTCGGAGATG*T -3′

Primers were synthesized with a terminal phosphorothioate bond (*) by IDT (Integrated DNA Technologies, Coralville, Iowa).

RNA isolation for RT-qPCR and RNA-seq experiments

Embryos developed in an incubator at 25 °C until the desired NC. NC was confirmed by time after egg activation and the morphology was checked under water using a transmission light microscope (Zeiss Stemi 2000-C). RNA isolation was performed using the Zymo Quick-RNA Tissue/Insect Microprep Kit following the manufacturer’s instructions except for the tissue homogenization step. Instead, homogenization was done as follows. Once at the desired NC individual embryos were placed in the lid of a 1.5 ml low-retention Eppendorf tube and homogenized in 10 μl of cold lysis buffer (10 mM Tris-HCl pH 7.4; 10 mM NaCl; 3 mM MgCl2; and 1% Igepal CA-630) [158], using a fire-sealed microcapillary tube (Drummond Microcap 25). Once homogenized the lysate was added to 450 µl of Zymo RNA Lysis Buffer and thoroughly mixed. RNA isolation resumed following the manufacturer’s instructions, including DNAse treatment. Purified RNA was eluted in DNase/RNase-Free Water.

RNA-seq library preparation

RNA quality and quantity were assessed using the Agilent bio-analyzer. NonStrand-specific RNA-seq libraries were prepared using the Smarter v4 Ultra-Low Input protocol from Takara for cDNA synthesis and Nextera XT from Illumina for Library prep (protocols provided by Takara and Illumina). Library quality and quantity were assessed using the Agilent bio-analyzer. Libraries from the same experiment were pooled and sequenced together. Sequencing was done at the University of Chicago Genomics Core (Chicago, IL, USA) using the Illumina NovaSeq X (PE 150 bp) and at Admera Health (South Plainfield, NJ, USA) using the Illumina NovaSeq X Plus (PE 150 bp).

Experimental setup of ATAC/RNA-seq experiments in Cal-zld and Cal-opa RNAi embryos

RNAi embryos were processed in parallel with stage-matched untreated embryos that were allowed to develop under water in the dish used for egg activation (wild-type controls), embryos that were prepared for injection under oil but not injected (alignment controls, AC), and embryos that were injected with double-stranded RNA of the extraneous gene DsRed (injection controls, IC). Cal-zld and Cal-opa expression levels were quantified using the RNA-seq data (see Fold change calculations).

ATAC-seq library preparation and RNA isolation

The first steps were the same as in the standard ATAC-seq protocol above. Individual embryos were homogenized in cold lysis buffer, and a centrifugation step was performed to pellet the nuclei. The supernatant (~47 µl) was then removed and mixed thoroughly with 450 µl Zymo RNA lysis buffer in a fresh low-bind reaction vial. The nuclear pellet was resuspended in 7.5 µl Tagmentation buffer and the ATAC-seq protocol was continued as described above. RNA isolation proceeded immediately, or the RNA lysate was stored at −20 °C for up to ~1 week. RNA isolation was performed using the Zymo Quick-RNA Tissue/Insect Microprep Kit following the manufacturer’s instructions. Isolated RNA was for RT-qPCR and RNA-seq experiments.

Preparation of injected embryos for ATAC-seq/ATAC-seq and RNA isolation

Following the injection, embryos were left to develop on the injection slide at 25 °C. Once the embryos had reached the desired stage, any excess oil was removed from the slide using a pipette. The embryos were then washed 6 times using deinonized water. Embryos were removed from the injection slide using a blunted tungsten needle and transferred to the lid of a low bind reaction vial. 10 μl of cold lysis buffer was added and the embryos were homogenized using a fire-sealed microcapillary tube. The ATAC-seq protocol was conducted as described above.

ATAC-seq data processing

Demultiplexed reads were trimmed of adapters using TrimGalore! [160] (version 0.6.10, cutadapt version 4.4) and mapped to the Clogmia albipunctata assembly (calbi-uni4000) using Bowtie2 [161] (version 2.4.1) with option -X 2000. Suspected optical and PCR duplicates were marked by Picard MarkDuplicates (version 2.21.4) using default parameters (https://broadinstitute.github.io/picard/). Mapped, trimmed, duplicate marked reads were imported into R using the GenomicAlignments [162] (version 1.38.2) and Rsamtools [163] (version 2.18.0). Reads that mapped, were properly paired, nonsecondary and had map quality scores ≥ 10 were imported. Only reads mapping to scaffolds 1–5 and scaffold 7 were used for downstream analysis. The start and end coordinates of the reads were adjusted to account for the Tn5 interaction site. Watson strand start sites had four base pairs subtracted, and Crick-strand start sites had five base pairs subtracted [158]. Reads with mapped length ≤ 120 bp were considered to have originated from ‘open’ chromatin.

RNA-seq data processing

Demultiplexed reads were trimmed of adapters using TrimGalore [160] (version 0.6.10, cutadapt version 4.4; https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/) and mapped to the Clogmia albipunctata assembly (calbi-uni4000) using Bowtie2 [161] (version 2.4.1) with default parameters and maximum fragment length (-X) of 2000. Suspected optical and PCR duplicates were marked by Picard MarkDuplicates (version 2.21.4) using default parameters. Mapped, trimmed, duplicate marked reads were imported into R using the GenomicAlignments [162] (version 1.38.2) and Rsamtools [163] (version 2.18.0). Reads that mapped, were properly paired, nonsecondary and had a map quality scores ≥ 10 were imported.

Genome assembly

We used Cantata Bio’s genome assembly services to generate a de novo reference genome for Clogmia albipunctata from ~70 freshly eclosed males of a 12-generation inbred line of an established laboratory culture [169]. High-fidelity (HiFi) PacBio reads were used to generate an initial draft assembly, subsequently refined with Omni-C reads (S1 Appendix). Details on sequencing and software used to assemble the genome have been described elsewhere [91]. We estimated genome heterozygosity and sequencing error rates using jellyfish to count 21-mers and GenomeScope2.0 to analyze the k-mer frequencies.

Genome annotation

We used RepeatModeler [170] (version v2.0.4) to identify repeat elements in the Clogmia albipunctata genome and RepeatMasker’s [171] (version v4.1.5) fambd.py script to extract Arthropoda records from the Dfam database. These sequences were combined to generate a custom repeat library, which was then used to soft-mask the genome. To predict gene structures, we used RNA-seq and protein evidence. mRNA data from 26 individual embryos of various stages, pooled samples of first and fourth-instar larvae, one female and one male pupa generated in this study were mapped to the genome (S1 Table). Additionally, 11 RNA samples (9 embryonic and two adult) from other studies were downloaded from NCBI (S1 Table). For protein evidence, we downloaded all available Dipteran sequences from NCBI’s RefSeq (January 2024) and UniProt’s complete protein sequences (Release 2023_04). A detailed description of the annotation pipeline is available elsewhere [91]. Briefly, we assembled and mapped the RNA transcripts and protein sequences to the reference and used these data as evidence of gene structures and to predict gene models. Evidence Modeler [172] (EVM, version 2.1.0) was used to create a consensus gene set, which was further refined with Program to Assemble Spliced Alignments’ (PASA version 2.5.3) to add UTRs and identify alternative transcripts. Gene structures overlapping with repeats and transposable elements (TEs) were filtered out. Functional annotation was performed with EggNOG-mapper [173,174] (version 2.1.12), assigning gene names based on orthology. We assessed the final annotation quality using BUSCO (Benchmarking Universal Single-Copy Orthologs) [175,176] on the predicted protein sequences. Finally, we analyzed synteny between the C. albipunctata and D. melanogaster genomes using MCScanX [177] (primary release) with parameters: match_score = 50 (default), match size = 20 (minimum genes per syntenic block), gap_penalty = −1 (default), and max_gaps = 100. Results were visualized using the circlize R package [178] (version 4.1.2). Annotations of Cal-eve1, Cal-eve2, Cal-tll1, and Cal-tll2 were manually corrected in the final output file (.gff3) based on RNA-seq data and transcript data available on NCBI.

TF identification

Any gene in our annotation of the Clogmia genome that contained the Cluster-of-Orthologous-Gene (COG) symbol “K” for transcription or the Gene Ontology (GO) term “0003700” for DNA-binding TF activity was treated as a putative TF gene.

Genome browser

The genomic data produced in this study can be further interrogated using a genome browser ecosystem centered on JBrowse2 [179] and delivered as a cloud image designed for individual use, currently available on NSF’s Jetstream2 cloud platform, with a portable Docker image with documentation and BSGenome package created for R. Jetstream2’s user-friendly interface (image: Clogmia albipunctata Genome Resources) [180,181] integrates tools for genomic analysis, including BLAST search (SequenceServer2.0) [182], CRISPR guide RNA design (modified crisprDesigner) [183], R shiny-driven differential gene expression (DGE) analysis (freecount) [184], and synteny mapping for comparative genomics (ShinySyn) [185]. Each tool is connected back to the browser, leveraging JBrowse2’s ability to dynamically visualize and contextualize the genomic data. BLAST results link directly to genomic regions, enabling users to explore expression tracks, variants, and synteny relationships. CRISPR tools provide sgRNA design with metrics for selection and visualization of target regions alongside expression data, off-target effects, and variants. DGE analysis visualizations are linked back to the genome browser for further exploration. All gene model annotations in the browser connect to common external databases, such as NCBI, Uniprot, FlyBase, and Google Scholar, for further exploring and interpreting the data. Future enhancements will include expanded sgRNA profiling, and further optimization of workflows, ensuring this ecosystem remains a powerful and accessible solution for the use of our genomic data presented here and elsewhere [91] for research and education. Installation instructions for accessing the Clogmia albipunctata genome browser and related tools are available at https://github.com/kallistaconsulting/genomic_resources_clogmia. The container requires a Linux web server on any cloud platform. The docker was developed on Jetstream2, which can be accessed via instructions at https://jetstream-cloud.org/get-started/index.html. This on-demand container-based model reduces resource costs and security concerns compared to hosting a centralized site. Administrative tools (e.g., automated reboot scripts, system service management, load balancing for multi-user groups or workshops) enhance usability while minimizing operational overhead.