ABSTRACT
Evolution of cis-regulatory elements (such as enhancers) plays an important role in the production of diverse morphology. However, a mechanistic understanding is often limited by the absence of methods for studying enhancers in species other than established model systems. Here, we sought to establish methods to identify and test enhancer activity in the red flour beetle, Tribolium castaneum. To identify possible enhancer regions, we first obtained genome-wide chromatin profiles from various tissues and stages of Tribolium using FAIRE (formaldehyde-assisted isolation of regulatory elements)-sequencing. Comparison of these profiles revealed a distinct set of open chromatin regions in each tissue and at each stage. In addition, comparison of the FAIRE data with sets of computationally predicted (i.e. supervised cis-regulatory module-predicted) enhancers revealed a very high overlap between the two datasets. Second, using nubbin in the wing and hunchback in the embryo as case studies, we established the first universal reporter assay system that works in various contexts in Tribolium, and in a cross-species context. Together, these advances will facilitate investigation of cis-evolution and morphological diversity in Tribolium and other insects.
INTRODUCTION
Insects display some of the greatest diversity of morphology found among eukaryotic taxa, offering a variety of opportunities to investigate molecular and developmental mechanisms underlying morphological evolution. Decades of studies in evolutionary developmental biology (evo-devo) have revealed that changes in gene regulatory networks (GRNs) have been a major driving force in the production of the diverse morphology seen in insects, as well as in other taxa (Carroll, 2008; Carroll et al., 2005). In general, a GRN can be divided into two components: trans and cis. trans components are transcription factors (TFs) and their upstream regulators (such as signal transduction pathways) that provide instructive cues for patterning and differentiation to the tissues where they are expressed. In contrast, cis components are non-coding DNA elements (i.e. cis-regulatory elements, CREs) that gather and process the upstream trans information, and determine the spatial and temporal expression of the genes downstream in the genetic pathway. Changes in both cis and trans components have been implicated in morphological evolution (Carroll, 2008; Carroll et al., 2005; Halfon, 2017).
By using unparalleled genetic tools, biologists have analyzed both cis and trans components in great detail in the fruit fly Drosophila melanogaster. The accumulated knowledge obtained from Drosophila studies can be used as a reference (i.e. the Drosophila paradigm) when studying other insects and identifying the changes in GRNs responsible for morphological evolution. RNA interference (RNAi)-based gene knockdown techniques have allowed for an investigation of the trans components involved in development and their evolutionary conservation/diversification in various insects (Bellés, 2010). However, the lack of a reliable method for identifying cis components in non-Drosophila insects has made it difficult to study the evolution of cis regulation beyond the Drosophila species, even though this is important in order to gain a comprehensive view of the GRN changes that contribute to morphological evolution.
The major difficulty in identifying CREs, such as enhancers, stems from the labile nature of cis components. The genes that code for trans factors that are important for development are usually evolutionarily well-conserved; thus, it is relatively easy to identify these trans components in various insects based on their homologies (Carroll et al., 2005). In contrast, cis components appear to be more evolutionarily flexible in a variety of aspects. First, the order of TF-binding sites can vary widely within an enhancer region, and the location of enhancers relative to the target gene also appears to be variable. Second, there can be redundancy among multiple enhancers responsible for the same gene (i.e. shadow enhancers) (Hong et al., 2008), allowing the enhancers to evolve more rapidly. In addition, the function of each enhancer tends to exhibit low levels of pleiotropy (Carroll, 2008), resulting in the accumulation of more evolutionary changes in enhancers. These characteristics, along with the faster rate of genome evolution in insects compared with vertebrates (Zdobnov and Bork, 2007), make the identification of insect enhancers a challenging task.
Traditionally, enhancers have been identified using reporter assays, in which the transcriptional activation capability of genomic regions near the gene of interest is assessed via a reporter gene construct (see Suryamohan and Halfon, 2015 for a review of traditional and new methods for identifying enhancers). This is a time-consuming and arduous approach, as an enhancer can sometimes reside hundreds of thousands of base pairs away from the gene that it regulates (Shlyueva et al., 2014). Identification of evolutionarily conserved genomic regions outside of coding regions among several closely related species, such as the Drosophila species group, has been helpful in narrowing down regions to survey for enhancer activity (phylogenetic footprinting) (Frazer et al., 2004; Mayor et al., 2000; Sosinsky et al., 2007; Stark et al., 2007). Enhancer predictions based on the TF-binding motifs have also been helpful in identifying potential enhancer regions, although the prediction appears to work more efficiently for embryonic enhancers because of the clustering tendency of TF-binding motifs within an enhancer that is active during the syncytial blastoderm stage, while enhancers for other stages might be more difficult to identify through current prediction methods (Li et al., 2007). Combinations of these approaches have allowed for successful identification of enhancers that are active in various developmental contexts in Drosophila. More recently, the reporter assay approach has been applied in a genome-wide fashion in Drosophila (as in the FlyLight project), identifying over 10,000 genomic regions capable of activating transcription (Jenett et al., 2012; Jory et al., 2012; Kvon et al., 2014; Pfeiffer et al., 2008). Unfortunately, many of these approaches are technically demanding and resource intensive, and thus are currently only possible in Drosophila (but also see Kazemian et al., 2014 for the successful application of enhancer prediction in non-Drosophila insects).
In parallel to the methods described above, several genomic approaches have been developed for the identification of possible enhancer regions in the Drosophila genome (reviewed by Shlyueva et al., 2014; Suryamohan and Halfon, 2015). One such method is formaldehyde-assisted isolation of regulatory elements (FAIRE) in combination with next-generation sequencing (FAIRE-seq), which identifies open chromatin regions across the genome (Simon et al., 2012). FAIRE-seq has been used in Drosophila, showing that open chromatin regions often correspond to enhancers and other CREs (McKay and Lieb, 2013; Pearson et al., 2016; Uyehara et al., 2017). In addition, FAIRE-seq requires less input material and does not rely on antibodies, thus making it less technically demanding than techniques such as chromatin immunoprecipitation-sequencing (ChIP-seq). These features make FAIRE a promising technique to apply to non-Drosophila insects. However, it is important to note that potential enhancers identified by FAIRE (or other genomic approaches) still require functional validations, such as with a reporter assay. This presents another significant hurdle when studying enhancers in non-Drosophila insects, as the availability of a modern genetic toolkit (such as a versatile reporter construct) is very limited for non-Drosophila species.
In this study, we aimed to establish an enhancer identification and evaluation method in the red flour beetle, Tribolium castaneum. A variety of genetic and genomic tools are available for Tribolium, making this insect a powerful model system for comparative developmental biology and evo-devo studies (Denell, 2008; Schmitt-Engel et al., 2015; Tribolium Genome Sequencing, 2008; Wang et al., 2007). The robust systemic RNAi response of Tribolium has allowed researchers to study trans components in detail (Brown et al., 1999; Bucher et al., 2002; Tomoyasu and Denell, 2004) and to identify changes in GRNs that are responsible for morphological evolution from the trans point of view (see Peel, 2008 for a review of the findings related to the evolution of insect segmentation; Tomoyasu et al., 2009 for insect wing evolution; and Angelini et al., 2012 and Smith et al., 2014 for the evolution of insect appendages). However, studies of cis components in Tribolium are currently limited because of the lack of reliable enhancer identification methods.
For the initial identification of possible enhancer regions, we first implemented FAIRE-seq and obtained genome-wide open chromatin profiles from various tissues and at different developmental stages of Tribolium. The comparison of chromatin profiles between different tissues and stages revealed a distinct set of open chromatin regions in each tissue and stage. In addition, the open chromatin regions detected by FAIRE matched very well with those identified in previous Tribolium enhancer studies (Cande et al., 2009; Eckert et al., 2004; Kazemian et al., 2014; Wolff et al., 1998), as well as with supervised cis-regulatory module (SCRMshaw)-predicted enhancers (Kantorovitz et al., 2009; Kazemian et al., 2011, 2014). Second, we chose the wing expression of nubbin (nub) (Fig. 1) as a case study, and established the first universal reporter assay system that works in Tribolium and also in a cross-species context. To our knowledge, the T. castaneum-nub (Tc-nub) wing enhancer identified here is the first post-embryonic enhancer that has been functionally evaluated in non-Drosophila insects. Furthermore, using hunchback (hb) as another example, we demonstrated that our reporter construct works in other developmental contexts in Tribolium. Together, these advances will facilitate investigation of enhancers in Tribolium and in other insects, which will provide a more comprehensive understanding of the molecular mechanisms underlying the production of the vast morphological diversity seen in insects.
RESULTS
FAIRE-seq revealed a spatially and temporally regulated chromatin profile in the Tribolium genome
To obtain chromatin profiles from diverse tissues and stages of Tribolium, we performed FAIRE-seq with the following six samples: three stages of embryos (0-24, 24-48 and 48-72 h), the second (T2) and third (T3) thoracic epidermal tissues of the last instar larvae that contain the forewing (elytron) and hindwing imaginal tissues, and the brain isolated from the last instar larvae. The sequence reads obtained from FAIRE-seq were then mapped to the Tribolium genome assembly (Tcas_3.0). Each sample displayed a unique set of open chromatin regions (referred to as ‘peaks’; see Fig. 2A for an example), indicating that FAIRE-seq with Tribolium tissues was successful. The overall open chromatin characteristics were similar in Tribolium and Drosophila; however, we also noticed some features that were unique to the Tribolium chromatin profiles. We detected more than 40,000 open chromatin regions in the Tribolium genome across the samples (Table 1). To identify differences in open chromatin profiles between samples, we performed differential peak calling using DiffBind [false discovery rate (FDR)<0.05]. The number of differentially accessible peaks between pairs of samples varied widely. For example, there were over 26,000 differentially accessible peaks between 0-24 h embryos and T3 (Table 1, Fig. S1), reflecting the extensive differences in cis-regulatory control that likely exist between these two samples. By contrast, we found only four differentially accessible peaks between T2 and T3. The similarity in open chromatin profiles in T2 and T3 tissues is remarkable, given the dramatic differences in morphology between forewing and hindwing in Tribolium. However, similar findings were obtained in Drosophila (McKay and Lieb, 2013), suggesting that both species use shared sets of enhancers to shape their appendages. Intriguingly, although the level of nucleosome depletion in the FAIRE-isolated genomic regions is variable between stages and tissues, their positions appear to correlate highly with the guanine-cytosine (GC)-rich regions of the genome (Fig. S2A). Furthermore, these GC-rich and FAIRE-identified regions occur at regular intervals, producing a ‘ruler-like’ pattern of FAIRE peaks throughout the genome (Fig. S2B). This regular periodicity of the GC-rich and FAIRE-identified regions appears to be unique to the Tribolium lineage, as we did not detect a similar periodicity in other coleopteran genomes or in the genome of the lepidopteran Bombyx mori (see Fig. S2C for Drosophila; data not shown for other insects).
Comparison of the FAIRE data with previous enhancer studies in Tribolium
Several previous studies have investigated the activity of Tribolium enhancers. To our knowledge, the only study analyzing enhancer activity in the Tribolium native context is that of Eckert et al., 2004. This study identified enhancers that are important for the stripe expression of the Tribolium hairy gene. Some additional enhancers for Tribolium genes have also been identified, albeit in a cross-species context (i.e. Drosophila). These include enhancers for hunchback (Wolff et al., 1998), single-minded, cactus and short gastrulation (Cande et al., 2009), and labial, Dichaete and wingless (Kazemian et al., 2014). We analyzed the FAIRE profiles at these gene loci and found that our FAIRE peaks matched with many of the enhancer regions identified in these studies (Fig. S3).
More recently, Kazemian et al. applied their enhancer discovery approach, SCRMshaw, to the Tribolium genome and predicted 1214 genomic regions to be potential enhancers (Kantorovitz et al., 2009; Kazemian et al., 2011, 2014). Comparison of our FAIRE data with their SCRMshaw predictions revealed a striking degree of overlap between the two datasets: 78.8% (957/1214) of SCRMshaw predictions overlapped with at least one embryonic FAIRE peak, and 88.1% (1070/1214) of predictions overlapped with at least one larval FAIRE peak (Tables S1, S2; P≈0); overall, 1096 of the 1214 (90.3%) predicted cis-regulatory modules (CRMs) overlapped with at least one FAIRE peak. For certain sets of SCRMshaw predictions, the overlaps were even more extensive: e.g. 98% (97/99) of wing-specific predicted enhancers overlapped with a larval FAIRE peak (Table S1). Taken together, the high degree of overlap between the FAIRE peaks and previously identified enhancer regions, and between the FAIRE-peaks and SCRMshaw-predicted enhancers, verifies that FAIRE-seq is a powerful tool for identifying enhancers in Tribolium.
Identification of the Tribolium nub wing enhancer using a cross-species reporter assay
As mentioned in the Introduction, reporter assays are a time-consuming and laborious task, which makes them difficult to perform in non-Drosophila insects, including Tribolium. However, to fully exploit the benefit of the FAIRE profiling data, it is crucial to have a reliable method to evaluate the function of Tribolium enhancers. Previously, the activity of potential Tribolium enhancers has been successfully evaluated using reporter assays in Drosophila (Cande et al., 2009; Kazemian et al., 2014; Wolff et al., 1998; Zinzen et al., 2006). We reasoned that the enhancer of a gene that has a conserved expression pattern (both temporal and spatial) in Drosophila and Tribolium has the highest chance of being active, even in a cross-species context, and is thus ideal for a case study. The enhancer responsible for the wing expression of nub fits this criterion, as nub is expressed broadly in the tissues that give rise to the wings in both insects (Fig. 1) (Ng et al., 1995; Tomoyasu et al., 2009). In addition, an enhancer trap line for nub is available in Tribolium (pu11; Fig. 1C-E). We have previously determined that this enhancer trap line has a piggyBac construct inserted about 30 kb upstream of the nub transcription start site (Clark-Hachtel et al., 2013) (Fig. 2A), which can be used as a starting point to survey for the wing enhancer.
nub codes for an evolutionarily conserved TF that is important for the proliferation of wing cells (Ng et al., 1995). Drosophila has two nub paralogs (nub and pdm2), whereas Tribolium has one (Tc-nub). FAIRE analysis revealed a number of open chromatin regions located in and near the Tc-nub locus (Fig. 2A). Some of the open chromatin regions were shared across the six samples tested (such as the region corresponding to the promoter), but others were unique to specific tissues and stages. We tested the two open chromatin regions at or near the pu11 insertion site (Tc-nub3 and Tc-nub2) in Drosophila (Fig. 2A,B). In addition, we also tested another major open chromatin region located further upstream of the pu11 insertion site (Tc-nub1). This region is open predominantly in the larval T2 and T3 epidermal tissues (containing the future wing tissues), but not in any of the embryonic samples, suggesting that this region contains enhancers that are specific to the post-embryonic stage (Fig. 2A,B).
The cross-species reporter assay showed that Tc-nub2 and Tc-nub3 do not have enhancer activity in the future wing-related tissues (wing and haltere imaginal discs) in Drosophila (Fig. 2C-F). Tc-nub3 showed activity in a small region near the hinge of the wing and haltere discs, but not in the region that gives rise to the wings (wing and haltere pouches) (Fig. 2C,D). Tc-nub2 drove reporter expression in the leg discs, but did not show any enhancer activity in the wing and haltere discs (Fig. 2E,F). In contrast, Tc-nub1 showed significant enhancer activity in the pouch region of the wing disc (Fig. 2G). Tc-nub1 also drove reporter expression in the leg disc, but was not active in the haltere disc (Fig. 2H). Because Tc-nub1 corresponds to the region that is uniquely open in the larval epidermis in Tribolium, the outcome of our cross-species reporter assay indicates that (1) the open chromatin profiling of various tissues and stages by FAIRE-seq in Tribolium can help predict tissue/stage-specific enhancers from the Tribolium genome, and (2) the cross-species reporter assay can be useful, at least for the enhancers responsible for the post-embryonic expression of nub in Tribolium.
We next sought to minimize the Tc-nub wing enhancer by testing three shorter fragments within the Tc-nub1 region (Fig. 2B). Interestingly, despite covering the main FAIRE peak region of Tc-nub1, Tc-nub1Core did not show any enhancer activity in the wing (Fig. 2K,L). Instead, Tc-nub1L, which corresponds to only a part of the major open chromatin region, drove reporter expression with a pattern and level almost identical to those driven by Tc-nub1 (Fig. 2I,J). Tc-nub1R did not show any enhancer activity (Fig. 2M,N). These results suggest that the important elements for driving wing expression reside within the first 200 bp of Tc-nub1. We tested this idea by making a reporter construct using only the 200 bp region unique to Tc-nub1L (Tc-nub1La, Fig. 2B). This fragment drove reporter expression in the wing and leg discs, albeit with a more-restricted expression domain and/or a lower expression level compared with Tc-nub1L (Fig. 2O,P). We also tested a construct that contained the Tc-nub1L region along with an additional 200 bp sequence outside of Tc-nub1 (Tc-nub1Lb, Fig. 2B), because the location of the functional Tc-nub wing enhancer may be slightly misaligned with respect to the open chromatin region predicted by FAIRE. However, Tc-nub1Lb showed even weaker enhancer activity (Fig. 2Q,R), suggesting that there might be a suppressor element next to the Tc-nub1 region. The constructs we made also drove reporter expression outside the wing and leg imaginal discs. These results are summarized in Table S3.
Identification of the Drosophila nub wing enhancer using a combination of genomic resources, FAIRE profiling and the reporter assay approach in Drosophila
For comparison with the enhancer identified via a cross-species reporter assay described above, we sought to identify the nub wing enhancer that is native to the species used for the reporter assay (i.e. Drosophila). As mentioned earlier, there are two nub paralogs in Drosophila (nub and pdm2), both of which have similar expression in the wing pouch (Ng et al., 1995). We first took advantage of the FlyLight project and surveyed the nub and pdm2 loci for a genomic region that has wing enhancer activity. Among the 33 constructs tested in FlyLight (Fig. 3A), one region (GMR11F02) has a record of enhancer activity in the wing and haltere pouches, along with additional expression in the leg disc (Fig. 3B,C). We then used the previously published FAIRE profile for Drosophila (McKay and Lieb, 2013), and identified three distinct regions within GMR11F02 that are open in the wing and haltere discs (Fig. 3A). We cloned these three regions [Fig. 3B; D. melanogaster-nub (Dm-nub)1, Dm-nub2 and Dm-nub3] and tested their enhancer activity in Drosophila. Among the three regions, Dm-nub2 displayed strong enhancer activity in the wing pouch region (Fig. 3G,H). Dm-nub1 (Fig. 3E,F) and Dm-nub3 (Fig. 3I,J) did not drive reporter expression in the wing and haltere discs. In addition, Dm-nub3 was active in the leg disc, suggesting that the Dm-nub wing/haltere enhancer and leg enhancer are separable (Fig. 3J). To further minimize the Dm-nub wing enhancer, we tested three shorter fragments within Dm-nub2 (Dm-nub2a, Dm-nub2b and Dm-nub2c; Fig. 3D). The wing-related expression is driven by Dm-nub2a, albeit at a weaker level (Fig. 3K,L). This suggests that, although Dm-nub2a contains sufficient components to drive wing expression, a broader genomic region is required for robust wing expression of Dm-nub. In contrast, Dm-nub2b and Dm-nub2c did not drive any expression (Fig. 3M-P). The expression patterns of these constructs in other tissues are summarized in Table S4. Taken together, the Dm-nub2 region that we isolated (1.3 kb) is sufficient to drive a robust wing expression in Drosophila.
Establishing a reporter assay system and evaluating the nub wing enhancers in Tribolium
Although some Tribolium enhancers have been shown to be active in the cross-species context, these enhancers still need to be examined in their native species for functional validation. However, the lack of a reliable reporter construct has been a major obstacle in performing functional evaluation of enhancers in Tribolium. The Gateway system (Katzen, 2007) has been useful in quickly cloning genomic regions into a reporter construct and testing their enhancer activity in Drosophila. We sought to establish a Gateway-compatible reporter construct that is functional in Tribolium.
A key issue in establishing a reporter construct is the choice of promoters. Previous studies have raised concerns about using Drosophila promoters in Tribolium (Schinko et al., 2010). While establishing the Gal4/UAS system in Tribolium, Schinko et al. found that the core promoter isolated from a Tribolium endogenous gene, Tc-hsp68, worked more efficiently than the exogenous promoters that were tested (Schinko et al., 2010). We therefore made a Gateway-compatible piggyBac construct that contained the Tc-hsp68 core promoter driving the dsRed gene [piggyBac Gateway Tc-hsp68 dsRed (piggyGHR), Fig. 4A]. In addition, we added the gypsy element, which is a Drosophila insulator, to either side of the the reporter assay construct to prevent position effects (Fig. 4A). We tested this piggyBac construct with the Tribolium and Drosophila nub wing enhancers (Tc-nub1L and Dm-nub2) in Drosophila. Both Tc-nub1L and Dm-nub2 drove dsRed expression identical to the patterns obtained with the Drosophila reporter construct (compare Fig. 4B,C with Fig. 2I,J, and Fig. 4D,E with Fig. 3G,H), confirming that piggyGHR is functional. However, neither Tc-nub1L nor Dm-nub2 in piggyGHR showed consistent enhancer activity in the wing tissues when transformed into Tribolium (Fig. 4F-M). None of the seven independent transgenic lines obtained for piggyGHR-Tc-nub1L had clear dsRed expression in the wing tissues (Fig. 4F-K). Instead, four lines had dsRed expression in non-wing tissues, with a distinct pattern in each line, likely because of trapping local enhancers (Fig. 4F-K). We obtained only two independent transgenic lines for piggyGHR-Dm-nub2, neither of which had dsRed expression in the wing tissue (Fig. 4L,M). These results indicate that our construct with the Tc-hsp68 core promoter does not work well for reporter assays, at least in the wing-related tissues in Tribolium, although it does work in Drosophila. Alternatively, it is also possible that the Drosophila gypsy insulators that we added to the construct might not be functioning properly in Tribolium.
We next tested a synthetic promoter in Tribolium. Pfeiffer et al. modified the super core promoter 1 (SCP1) (Juven-Gershon et al., 2006) and constructed the Drosophila synthetic core promoter (DSCP), which was used for the FlyLight project as well as in other Drosophila reporter constructs, including pFUGG in this study (McKay and Lieb, 2013). We made a piggyBac construct with the DSCP driving mCherry [piggyBac Gateway universal promoter mCherry (piggyGUM); Fig. 5A]. We removed the Drosophila gypsy insulators from our construct to avoid possible cross-species issues. Similar to piggyGHR, piggyGUM with the Drosophila and Tribolium nub wing enhancers drove reporter expression in the wing disc in Drosophila (Fig. 5B-E), confirming that piggyGUM is functional. In Tribolium, in contrast to the piggyGHR constructs, piggyGUM-Tc-nub1L successfully recaptured the expression pattern of the nub enhancer trap line (pu11) and drove reporter expression in the wing-related tissues (both in T2 and in T3) at both larval and pupal stages (Fig. 5F-I compared with Fig. 1C,D). piggyGUM-Dm-nub2 also showed enhancer activity in the larval wing discs in Tribolium (Fig. 5J-L). The expression driven by Dm-nub2 in Tribolium was mostly in the wing hinge and the margin regions, similar to the pattern observed for this enhancer in the Drosophila imaginal discs (Figs 3G,H and 5D,E). These results indicate that: (1) our Gateway-compatible DSCP piggyBac construct (piggyGUM) can be used for reporter assays both in Tribolium and in Drosophila; (2) the Tribolium nub wing enhancer that was identified through a cross-species reporter assay (Tc-nub1L) is indeed functional as a wing enhancer in Tribolium. In addition, some of the piggyGUM transgenic lines showed mCherry expression in tissues other than wings (data not shown). The expression patterns outside the wing-related tissues were not consistent among the transgenic lines, suggesting that the piggyGUM construct also occasionally traps endogenous enhancers.
We also tested whether the promoter that is endogenous to the enhancer works better for a reporter assay construct in Tribolium. We made a piggyBac construct with the 2 kb sequence upstream of the Tc-nub transcription start site [confirmed by 5′ rapid amplification of cDNA ends (RACE); Clark-Hachtel et al., 2013] as the promoter [piggyBac nub promoter dsRed (piggyNub-proR); Fig. 6A]. We also used the 2 kb sequence downstream of the Tc-nub stop codon (confirmed by 3′ RACE; Clark-Hachtel et al., 2013) as the 3′ untranslated region (UTR) and the polyA signal native to Tc-nub (Fig. 6A). We made a similar construct for Tc-Act5c (with the 1 kb sequence upstream of the transcription start site and the 1 kb sequence downstream of the stop codon as the native promoter and polyA signal, respectively) as a comparison (Fig. 6B). To our surprise, Tc-nub1L in piggyNub-proR did not drive any expression in Tribolium (Fig. 6C-F) or in Drosophila (Fig. 6G,H). Real-time qPCR analysis revealed that there is no transcription of dsRed in these transgenic lines in both species (data not shown), suggesting that the lack of reporter expression is not due to incompatibility of the reporter gene with the Tc-nub UTRs but, rather, to the nub wing enhancer failing to work with the endogenous promoter and/or polyA signal. In contrast to piggyNub-proR-Tc-nub1L, piggyAct5cR showed strong and ubiquitous dsRed expression in Tribolium (Fig. 6I), indicating that our strategy of incorporating the endogenous transcription and translation components is valid. Intriguingly, however, piggyAct5cR did not drive any expression in Drosophila (data not shown), implying a strict species-specific nature of the transcription and/or translation components (such as promoters), even for an evolutionarily highly conserved house-keeping gene that is uniformly expressed in various species, including Drosophila and Tribolium (Chung and Keller, 1990).
Testing the reporter construct in another context in Tribolium
We next tested whether our DSCP reporter system worked in a context other than wings in Tribolium. We chose hb as a case study, and tested the reporter activity during embryogenesis. hb expression in Tribolium starts as a broad posterior domain in the blastoderm, and subsequently clears from the posterior to form an anterior band of expression that covers the pre-gnathal and gnathal segments (Lynch et al., 2012; Marques-Souza et al., 2008). In the early germband stage, the band resolves into a stripe covering the labium (Fig. 7B) (Marques-Souza, 2007; Zhu et al., 2017). Wolff et al. previously identified a genomic region at the Tribolium hb locus that drives blastoderm expression when introduced in Drosophila (Fig. 7A, orange bar) (Wolff et al., 1998). This region corresponds to a SCRMshaw prediction (Fig. 7A, purple bars). Therefore, although the FAIRE signal at this region is weak (likely because of the wide time window of sampling during early embryogenesis), the outcomes of previous studies make this region an excellent candidate enhancer to test with our reporter system in Tribolium. We cloned a 1340 bp fragment containing this genomic region (hb-PE1, Fig. 7A, red bar) and tested its enhancer activity using the piggyGUM construct in Tribolium. In situ hybridization for the mCherry reporter gene revealed that the piggyGUM-hb-PE1 construct recapitulates the hb expression at the early germband stage in Tribolium (Fig. 7C). This result indicates that: (1) our DSCP reporter system works well even during embryogenesis in Tribolium; (2) hb-PE1 contains the hb early germband enhancer.
DISCUSSION
In this study, we demonstrated that FAIRE-based chromatin profiling is a powerful approach for identifying CREs, such as enhancers, in Tribolium. The Tribolium nub wing enhancer that we identified (Tc-nub1L) is over 40 kb away from the nub transcription start site and 10 kb away from the pu11 insertion site, which would be very difficult to identify without the aid of open chromatin profiles. In addition, with the use of the DSCP, we were able to establish a functional reporter assay construct in Tribolium. A combination of FAIRE-based chromatin profiling with this reporter assay system will allow us to assess the function and evolution of enhancers in Tribolium.
FAIRE profiles in Tribolium
Genome-wide FAIRE profiling in Tribolium has identified a significant number of genomic regions whose chromatin status is regulated in a tissue- and stage-specific manner (Table 1, Fig. S1). These regions are promising candidates for future enhancer studies in Tribolium. In addition, our FAIRE analysis has revealed both evolutionarily conserved and diverged aspects of chromatin state regulation in Drosophila and Tribolium. For the conserved aspect, we saw similar chromatin profiles for the T2 and T3 epidermal samples, even though these two tissues differentiate into morphologically distinct structures (the elytron in T2 and hindwing in T3). This outcome echoes the message obtained from the Drosophila FAIRE study, i.e. that chromatin profiles are largely similar among the similar lineages of tissues (such as legs, wings and halteres), with the exception of a handful of ‘master control gene’ loci (McKay and Lieb, 2013). In fact, three of the four differentially open FAIRE peaks between T2 and T3 in our Tribolium FAIRE analysis were within the Ultrabithorax (the T3 selector gene) locus (Fig. S1) (for a review of the function of Ultrabithorax during wing development, see Tomoyasu, 2017). In contrast, the Tribolium FAIRE profiles during embryogenesis showed an interesting difference when compared with those in Drosophila. In Drosophila, the number of genomic regions that are open is fairly consistent throughout embryogenesis, with a distinct set of genomic regions being open in each stage (McKay and Lieb, 2013). In Tribolium, we noticed that a larger number of chromatin regions are open early in embryogenesis, and some of these regions are subsequently closed, resulting in a smaller number of open chromatin regions in later stages. This difference may be a reflection of the different modes of embryogenesis in the two insects (long versus short germband embryogenesis), although the significance of the difference in chromatin profiles has yet to be investigated.
We also noticed a strict overlap between the GC-rich regions and FAIRE-detected open chromatin regions. This raises interesting questions about the evolution of enhancers. Are these regions open because they are functionally important (such as enhancers)? Or have these regions become enhancers, because they were open owing to a bias in their nucleotide content and, thus, were accessible to TFs? There appears to be a similar correlation among the GC-rich regions, enhancers and FAIRE peaks in Drosophila (Li et al., 2007; McKay and Lieb, 2013). It will be interesting to investigate how GC-rich regions overlap with open chromatin regions in other insects. In addition, we found that the GC-rich and FAIRE-positive regions appear at regular intervals throughout the Tribolium genome. The molecular basis and functional implication of this periodicity is currently unknown; however, it is intriguing to speculate that a genome-wide event (such as transposon invasion) might have significantly influenced the chromatin state landscape in the Tribolium lineage.
Overlaps between FAIRE peaks and SCRMshaw enhancer predictions
The high degree of overlap observed between FAIRE peaks and enhancers predicted by the completely different, solely computational SCRMshaw method provides further confirmation that FAIRE is an effective means for enhancer discovery in Tribolium. Overall, the number of FAIRE peaks is well in excess of the number of SCRMshaw predictions. Several factors likely account for this result. First, the SCRMshaw predictions were performed at high stringency in order to minimize potential false-positive results (Kazemian et al., 2014); relaxing the prediction criteria would yield more predicted enhancers. Although this would potentially lead to more false positives, the >90% overlap seen for several specific datasets (Table S1) suggests that stringency could be relaxed in at least some cases. Second, SCRMshaw relies on training data from known Drosophila enhancers; therefore, enhancers with characteristics that deviate significantly from those of Drosophila enhancers will be found only by chromatin profiling, such as FAIRE. Finally, although FAIRE appears to be biased toward enhancers (Song et al., 2011), it also identifies other regions of open chromatin, such as promoters and insulator regions (Giresi et al., 2007), which are not predicted by the enhancer-specific SCRMshaw.
The twin issues of higher SCRMshaw false-positive rates at lower prediction stringencies and the lack of discrimination of FAIRE with respect to enhancers with specific spatial and temporal activity profiles suggest that considerable advantages could be obtained by using the methods in combination. Overlap with FAIRE peaks can be used to filter out false-positive SCRMshaw predictions, allowing predictions to be performed at lower stringency, and thus higher sensitivity. Conversely, SCRMshaw predictions can be used to focus on potentially more relevant FAIRE peaks – helping to avoid selecting enhancer sequences that are active in tissues other than the one of interest, enhancers for a neighboring housekeeping gene, insulators, and cryptic promoters or promoters for unannotated genes. This will be particularly useful in situations such as the one seen here for the larval samples, where cleanly separating the wing from body wall tissue was difficult: a common challenge when attempting to isolate tissues from small organisms such as insect embryos.
Enhancer activity in cross-species contexts and the limitation of non-native reporter assays
Our reporter assays in two insect species showed that both Drosophila and Tribolium nub wing enhancers were at least partially active in the cross-species context. We identified a 20 bp sequence that was shared between the two enhancers. This sequence contained binding sites of some wing-related TFs (such as Brinker and Mad) (Fig. S4), making it a promising candidate for an evolutionarily conserved enhancer motif. However, deletion of this sequence did not influence the activity of these enhancers in Drosophila, indicating that this sequence is dispensable for enhancer function (Fig. S4). We did not recognize any other significant sequence similarity or a conserved TF-binding site architecture between the two enhancers, suggesting that the regulatory landscape in the wings of the two species is evolutionarily maintained (as the nub enhancers can be functional in cross-species contexts), despite the lack of noticeable sequence conservation in the enhancer itself. A thorough examination of trans components that regulate the nub wing enhancers may give us insights into how enhancers evolve in a conserved regulatory landscape.
Although the Tribolium wing enhancer was active in Drosophila, we noticed that the activity of this enhancer was somewhat restricted, as it was active mainly at the dorsal-ventral (DV) compartmental boundary of the T2 wing, and in only a few cells in the haltere. This is in contrast with the expression in Tribolium, which showed a broader activity domain in the entire wing tissue in both the T2 and T3 segments. These differences in the activity domains suggest that some components that regulate the Tribolium nub wing enhancer are missing from the Drosophila T2 wing and are almost entirely absent in the haltere. This highlights the limitation of cross-species analyses and the importance of performing reporter assays in the native species. The reporter assay system we developed allows us to analyze enhancer activities in Tribolium. The successful demonstration of reporter analyses for nub in the wing and hb in the embryo suggest that our reporter construct works in various tissues; however, it is still crucial to evaluate the applicability of this system in diverse contexts.
Choice of core promoters in reporter constructs
Our study showed that the choice of promoters is crucial when assessing enhancer activity. Tc-hsp68 was our first choice because it has successfully been used in the Gal4/UAS system in Tribolium (Schinko et al., 2010). However, although this promoter worked efficiently in Drosophila, in our reporter assay it failed to drive reporter expression even with a functional enhancer in Tribolium (at least in our hands). Interestingly, the transgenic beetles with the Tc-hsp68 reporter construct showed a high occurrence of enhancer trap events (Fig. 4F-M), even though this promoter failed to work with the enhancer that we placed directly upstream of it. One explanation is that this promoter requires a certain distance for optimal interaction with enhancers in Tribolium. The situation might be less strict in Drosophila (for an unknown reason), allowing the Tc-hsp68 promoter to overcome the distance requirement.
We also tried to assess the nub wing enhancer activity with the nub endogenous promoter but, to our surprise, this construct did not drive any expression. There are several possible explanations for this outcome. First, the region we selected might not contain the correct promoter for the nub transcript, although our 5′ RACE results (as well as the published Tribolium genome annotation, Tribolium Genome Sequencing et al., 2008) support our annotation of the nub transcription start site (Clark-Hachtel et al., 2013). Second, the 2 kb region we used as the promoter may contain a suppressor element, interfering with the enhancer to drive reporter expression. Third, the nub promoter might require a long distance to interact properly with the wing enhancer, as the wing enhancer that we identified was 40 kb away from the nub transcription start site. This characteristic might be similar to that of Tc-hsp68, which preferentially interacts with enhancers located at a certain distance. This may further support the idea that Drosophila are more permissive to changes in the enhancer/promoter distance. However, in the case of the nub endogenous promoter, there might be additional issues other than enhancer/promoter distance that prevent this reporter construct from working even in Drosophila.
The reporter construct with the DSCP (piggyGUM) worked efficiently both in Drosophila and in Tribolium. The DCSP is a synthetic core promoter, composed of several common core promoter motifs [i.e. TATA box, initiator element (Inr), motif ten element (MTE) and downstream promoter element (DPE)] isolated from the Drosophila genome. The DSCP has been shown to work efficiently with a diverse array of developmental enhancers in various contexts in Drosophila (Pfeiffer et al., 2008; Zabidi et al., 2015), suggesting that this promoter may also work well with other enhancers in Tribolium. However, it is worth mentioning that a synthetic promoter similar to the DCSP, SCP1 (composed of Drosophila and viral promoter motifs; Juven-Gershon et al., 2006), failed to work when tested in the Gal4/UAS system in Tribolium (Schinko et al., 2010). This again emphasizes the importance of choosing the correct promoter that fits the context of the study, which remains a crucial area for further exploration.
Enhancer studies in evo-devo
The study of enhancers and other CREs is crucial for understanding the molecular basis underlying morphological evolution, as changes in gene regulation, rather than the acquisition of new genes or the modification of protein structures, are often responsible for the evolution of diverse morphology (Carroll, 2008). For example, changes in enhancers can facilitate evolution of novel structures via co-opting pre-existing GRNs into a new context. Acquisition of enhancers de novo may also play a crucial role in morphological novelty. Therefore, studying both evolutionarily conserved and diverged enhancers will help further our understanding of morphological evolution (see Monteiro and Podlaha, 2009 for a comprehensive discussion of how cis studies can help elucidate the molecular basis for the evolution of novel traits). However, it has been a challenge to study enhancers in non-traditional model insects because of the lack of a reliable enhancer identification strategy. In this study, we showed that FAIRE-seq is readily applicable to non-traditional model species. Furthermore, the DSCP can be a useful promoter for establishing a reporter assay system and investigating the evolution of enhancers in non-Drosophila insects. Therefore, FAIRE-based chromatin profiling, along with reporter assay systems applicable to various insects and SCRMshaw enhancer prediction, will make the research on enhancers more accessible, which will provide us with more insights into the evolution of the regulatory mechanisms underlying morphological diversity.
MATERIALS AND METHODS
Fly stocks
The following two Drosophila strains used in this study were obtained from the Bloomington Drosophila Stock Center: P{UAS-Dcr-2.D}1, w1118; P{GawB}nubbin-AC-62 and y1 w*; wgSp−1/CyO, P{Wee-P.ph0}BaccWee-P20; P{20XUAS-6XGFP}attP2.
Beetle cultures
The beetle cultures were reared on whole-wheat flour (+5% yeast) at 30°C in a temperature- and humidity-controlled incubator. The nub enhancer trap line pu11, which has enhanced yellow fluorescent protein (EYFP) expression in the hindwing and elytron discs (Clark-Hachtel et al., 2013; Lorenzen et al., 2003; Tomoyasu et al., 2005), was used to monitor nub expression in Tribolium.
Tissue preparation for FAIRE
For the Tribolium larval T2 and T3 wing tissues, the dorso-lateral region of the epidermal tissues that contain elytron (T2) and hindwing (T3) discs were dissected from the last instar larvae. Although these samples largely consisted of tissues that give rise to wing structures, they also contained body wall tissues as well as larval muscles because of the difficulty of precisely dissecting the wing tissues from larvae. About 50 larvae (100 dissected tissues) were used for each biological replicate, and three replicates were prepared for each wing sample. The brains were dissected from the head of the last instar larvae. About 40 brains were used for each biological replicate, and two replicates were prepared. Embryos were collected in whole-wheat flour (+5% yeast) for 24 h at 30°C. The collected embryos were cultured for 1 and 2 days at 30°C for the 24-48 h and 48-72 h samples, respectively; 0.1 g of embryos was used for each biological replicate, and three replicates were prepared for each sample. These tissues and embryos were crosslinked with 4% formaldehyde for 30 min (larval tissues) or 8% formaldehyde for 30 min (embryos).
FAIRE-seq analysis
FAIRE was performed as previously described (McKay and Lieb, 2013). FAIRE-seq libraries were sequenced on an Illumina HiSeq 2000 at the University of North Carolina High-Throughput Sequencing Facility. 50 bp single-end Illumina reads were obtained for FAIRE-treated samples and two non-FAIRE-treated input samples. Reads were trimmed to remove the index sequence and mapped to the Tribolium reference genome (version 3.0) with bowtie2 (Langmead and Salzberg, 2012). Read alignments were quality filtered (Q<10 dropped), and duplicate reads were removed using SAMtools (http://samtools.sourceforge.net/). For visualization of FAIRE signal, bigwig files were produced by merging tissue/stage-specific replicate bam files with SAMtools and normalizing reads to sequencing depth using deepTools (https://deeptools.readthedocs.io/en/develop/). These files were then visualized with the IGV genome viewer (Robinson et al., 2011; Thorvaldsdóttir et al., 2013). Peaks were called on individual replicates using MACS2 (https://github.com/taoliu/MACS), with the merged input sample bam files as the control. The Drosophila FAIRE profiles used in this study have been previously published (McKay and Lieb, 2013). For differentially open peak analysis, mapped reads (.bam files) for each replicate and the merged input, along with the MACS2 peaks (.narrowPeak files) called for each replicate, were provided as input for DiffBind (http://bioconductor.org/packages/release/bioc/html/DiffBind.html). DiffBind creates a consensus peakset for all replicates provided, requiring a consensus peak to be present in at least two replicates of a sample. An experiment-wide consensus peakset was produced using all samples. Pairwise analysis of differentially open peaks between samples was performed within DiffBind with the DESeq2 method for all consensus peaksets, and plotted using the dba.plotMA() function. The differentially open peaks are listed in Table S6.
Genome-wide GC-contents analysis
Using the experiment-wide consensus peakset described above, 1 kb of sequence upstream and downstream of each peak center was extracted from the genome using BEDTools (Quinlan and Hall, 2010) and custom Python (https://www.python.org/) scripts. For these 2 kb fragments, those free of ‘N’s were subjected to GC analysis. Changes in local GC content (250 bp sliding window, 10 bp step) were plotted against the whole-fragment average of GC content for all fragments. For the GC-rich region distance analysis, first, bedGraphs of GC content fluctuations above and below the genome wide average were computed at 70 and 60 bp resolution for the Tribolium and Drosophila genomes, respectively. The genome of Bombyx mori as well as those of several coleopteran insects (Agrilus planipennis, Dendroctonus ponderosae, Anoplophora glabripennis, Leptinotarsa decemlineata, Nicrophorus vespilloides and Onthophagus taurus) were analyzed at 70 bp resolution. Peaks were then called using the bdgcallpeak command in MACS2. The distance between the edges of adjacent peaks was categorized into 100 bp bins, and the natural logarithm of the number of occurrences was plotted. For the FAIRE peak distance analysis, distances between FAIRE peaks were collected and plotted in the same manner as the GC peaks. A consensus Drosophila FAIRE peakset was obtained from DiffBind with the same setting as those for the Tribolium data applied to the previously published data (McKay and Lieb, 2013).
Comparison between FAIRE and SCRMshaw
Enhancers predicted by SCRMshaw were taken from Kazemian et al. (2014) and converted into BED format. BEDTools (http://bedtools.readthedocs.io/en/latest/) merge was used to combine overlapping and/or redundant (i.e. from more than one SCRMshaw scoring method) predictions, reducing the total number of predicted enhancers to 1214. BEDTools intersect was then used to determine all predicted enhancers with at least 50 bp overlap with a FAIRE peak (-f 0.10). FAIRE peaks that were not assigned to a Tribolium chromosome (i.e. not starting with ‘ChLG’) were omitted. Significance of overlaps was determined using BEDTools fisher; all overlaps were highly significant with –log(P)≥19. Because this method provides only an approximation, a selection of datasets was tested via randomization. BEDTools shuffle was used to generate 1000 random intervals, and the intersections were determined as above. The mean and standard deviation of the randomized intersections were calculated and used with the observed (SCRMshaw) intersection value to determine a z score. P values from all randomization tests were highly significant.
Drosophila reporter assay constructs
pFUGG, a Drosophila Gateway-compatible phiC31 transformation plasmid, was used for reporter assays in Drosophila (McKay and Lieb, 2013). The phiC31 system allows site-specific integration (Bischof et al., 2007), thus preventing position effects due to different insertion sites. An enhancer cloned into pFUGG drives Gal4 as the reporter, whose expression domains can then be visualized by crossing to UAS-EGFP flies.
Gateway-compatible piggyBac reporter constructs
The piggyBac plasmid with the 3×P3-EGFP marker construct and the FseI/AscI cloning site (Horn and Wimmer, 2000) was used to make all piggyBac constructs used in this study. For piggyGHR (piggyBac Gateway Tc-hsp68 dsRed), the gypsy element, the Tc-hsp68 core promoter, dsRed and the SV40 polyA signal were amplified by PCR, assembled through ligation and inserted into the FseI/AscI site of the piggyBac plasmid. The assembled plasmid was then converted to a Gateway-compatible plasmid by Gateway Vector Conversion System (ThermoFisher Science). For piggyGUM (piggyBac Gateway Universal promoter mCherry), the reporter construct including the Gateway cassette was amplified from a Drosophila Gateway-compatible phiC31 transformation vector and inserted into the FseI/AscI site of the piggyBac plasmid. The primers used to make piggyGUM are listed in Table S5. The annotated sequence of the DSCP used in piggyGUM is shown in Fig. S5. The reporter constructs in piggyNub-proR (piggyBac nub promoter dsRed) and piggyAct5cR (piggyBac Act5c promoter dsRed) were de novo synthesized and inserted into the FseI/AscI site of the piggyBac plasmid.
Enhancer cloning
Genomic fragments corresponding to possible enhancer regions were PCR amplified and cloned into pENTR using pENTR-D Directional TOPO Cloning kit (Thermo-Fisher Scientific, K240020). The primers used to clone the enhancer regions from the Drosophila and Tribolium genome are listed in Table S5. Cloned genomic fragments were then inserted into reporter constructs via Gateway Clonase reaction (Thermo-Fisher Scientific, 11791-019).
Drosophila and Tribolium transgenesis
For Drosophila transgenesis, pFUGG constructs were transformed into the attP2 site (68A4) through phiC31 integrase-mediated transgenesis system, and piggyBac constructs were transformed into w1118 with EGFP as a visible marker (BestGene Drosophila transgenic service). For Tribolium transgenesis, piggyBac constructs were transformed into vermilionwhite with EGFP as a visible marker (TriGenES Tribolium Genome Editing Service for the nub and Act5c constructs, Friedrich-Alexander-Universität Erlangen-Nürnberg for the hb construct).
Immunohistochemistry and in situ hybridization
Drosophila imaginal discs were dissected from the third instar larvae and fixed with 4% formaldehyde for 25 min. Tribolium elytron and hindwing discs were dissected from the last instar larvae and fixed with 4% formaldehyde for 25 min. Dissected tissues were then washed and blocked with 10% BSA, and incubated with rabbit anti-mCherry antibody (1:500; Abcam, ab167453) at 4°C overnight. After washing for 1 h, the tissues were incubated with the Alexa 555-conjugated goat anti-rabbit antibody (1:500) for 2 h at room temperature. All the discs were mounted on glass slides with ProLong Gold antifade reagent (Life Technologies) for documentation. in situ hybridization was performed as previously described (Shippy et al., 2009), with digoxigenin (DIG)-labeled riboprobes and alkaline phosphatase-conjugated anti-DIG antibody (Sigma-Aldrich 11093274910). The signal was developed using BM-Purple (Sigma-Aldrich 11442074001). The primers used to amplify the mCherry fragment for riboprobe synthesis are included in Table S5. The hb riboprobe used in this study has been previously described (Wolff et al., 1998).
Image processing and documentation
The images were captured with a Zeiss 710 confocal microscope (mounted discs) and Zeiss AxioCam MRc5 with Zeiss Discovery V12 (Tribolium larvae and pupae). A filter set specific to mCherry (575/50×, 640/50 m) was used to visualize the mCherry expression driven by piggyGUM constructs. Tribolium germband embryos were imaged with a ProgRes CFcool camera on a Zeiss Axio Scope.A1 microscope using ProgRes CapturePro image acquisition software. Some pictures were enhanced only for brightness and contrast with Adobe Photoshop.
Acknowledgements
We thank the Bloomington Stock Center for fly stocks, Johannes Schinko for discussions on Tribolium transgenesis, and the Center for Bioinformatics and Functional Genomics (CBFG) and Center for Advanced Microscopy and Imaging (CAMI) at Miami University for technical support. We also thank Shuxia Yi for technical support, and Courtney Clark-Hachtel, David Linz and other members of Tomoyasu lab for helpful discussions.
Footnotes
Author contributions
Conceptualization: D.J.M., Y.T.; Methodology: Y.-T.L., K.D.D., F.B.-C., N.S., M.S.H., D.J.M., Y.T.; Software: K.D.D., K.S., M.S.H., D.J.M., Y.T.; Validation: K.D.D., F.B.-C., N.S., M.S.H., D.J.M., Y.T.; Formal analysis: Y.-T.L., K.D.D., F.B.-C., M.S.H., D.J.M., Y.T.; Investigation: Y.-T.L., K.D.D., F.B.-C., N.S., H.R., K.S., E.E.-S., M.S.H., D.J.M., Y.T.; Resources: D.J.M., Y.T.; Data curation: Y.-T.L., K.D.D., M.S.H., D.J.M., Y.T.; Writing - original draft: Y.-T.L., M.S.H., D.J.M., Y.T.; Writing - review & editing: Y.-T.L., K.D.D., E.E.-S., M.S.H., D.J.M., Y.T.; Visualization: Y.-T.L., K.D.D., H.R., E.E.-S., M.S.H., D.J.M., Y.T.; Supervision: M.S.H., D.J.M., Y.T.; Project administration: M.S.H., D.J.M., Y.T.; Funding acquisition: M.S.H., Y.T.
Funding
This project was supported by a National Science Foundation (NSF) grant (IOS0950964 and IOS1557936 to Y.T.), a U.S. Department of Agriculture (USDA) grant (2012-67013-19361 to M.S.H.) and The University of North Carolina at Chapel Hill (UNC-CH) start-up funds to D.J.M.
Data availability
FAIRE-seq data have been deposited in Gene Expression Omnibus (GEO) under accession number GSE104495. Sequences of the piggyBac plasmids established in this study are available at Figshare (https://figshare.com/articles/Insect_Transformation_Plasmids/6050486).
References
Competing interests
The authors declare no competing or financial interests.