Rapid dynamics of signal-dependent transcriptional repression by Capicua

SUMMARY

Optogenetic perturbations, live imaging, and time-resolved ChIP-seq assays in Drosophila embryos were used to dissect the ERK-dependent control of the HMG-box repressor Capicua (Cic), which plays critical roles in development and is deregulated in human spinocerebellar ataxia and cancers. We established that Cic target genes are activated before significant downregulation of nuclear localization of Cic and demonstrate that their activation is preceded by fast dissociation of Cic from the regulatory DNA. Furthermore, we discovered that both Cic-DNA binding and repression are rapidly reinstated in the absence of ERK activation, revealing that inductive signaling must be sufficiently sustained to ensure robust transcriptional response. Our work provides a quantitative framework for the mechanistic analysis of dynamics and control of transcriptional repression in development.

eTOC Blurb

Using optogenetics, the MS2-MCP system, and ChIP-seq techniques, Keenan et al. investigate the dynamic transcriptional response of ERK signaling in the early Drosophila embryo. They show that ERK signaling relieves repression of genes on a short timescale and that repression can return when signal is removed.

INTRODUCTION

ERK signaling plays essential roles in development, most commonly by controlling gene expression during tissue patterning and morphogenesis (Patel and Shvartsman, 2018). Active ERK can control gene expression in multiple ways, including changes in DNA binding, stability, and subcellular localization of transcription factors and components of basal transcriptional machinery (Carlson et al., 2013; Foulds et al., 2004; Kedage et al., 2017; Tootle et al., 2003; Yang et al., 1999, 2013). While the list of the ERK substrates involved in transcription is already extensive and still growing, the relative contributions of different regulatory mechanisms in any given developmental context remain unknown. Here, we address this question by studying the first of multiple ERK-dependent events in Drosophila embryogenesis, which patterns the head and tail structures of the future larva.

This ERK signaling event is induced by Torso, a uniformly expressed receptor tyrosine kinase, which is presented with active ligand at the poles of the embryo (Casali and Casanova, 2001; Goyal et al., 2018). The patterning functions of ERK are mediated by Capicua (Cic), the HMG-box transcriptional repressor that controls its target genes through the highly conserved binding sites in their regulatory DNA (Ajuria et al., 2011; Jimenez et al., 2012). For example, tailless (tll), a gene essential for terminal patterning, is repressed by Cic in the middle of the embryo, where ERK is inactive, but is expressed in the head and tail regions where ERK is activated by Torso (Casanova, 1990; Jimenez et al., 2000; Liaw et al., 1995). ERK signaling at the poles has a dual function: inducing the formation of terminal structures and repressing the segmentation gene expression program. Both of these effects rely on antagonism of transcriptional repression by Cic.

Cic is loaded into the embryo as a spatially uniform maternal RNA and translated after egg fertilization. Active ERK phosphorylates Cic, triggering its nuclear export and cytoplasmic degradation (Astigarraga et al., 2007; Futran et al., 2015). As a consequence of this regulation, nuclear levels of Cic are reduced at the poles, where ERK is activated by Torso. This correlation suggests that the ERK-dependent expression of tll and other targets of Cic depends on its nuclear export and degradation (Ajuria et al., 2011; Dissanayake et al., 2011; Grimm et al., 2012; Jimenez et al., 2000; Lim et al., 2013). Indeed, a previous study has shown that ERK activation reduces the rate of Cic nuclear import and increases its export rate (Grimm et al., 2012). However, other evidence indicates that ERK-target genes repressed by Cic may be induced even before Cic leaves the nucleus. In particular, studies with time-ordered fixed embryos showed that during a second wave of ERK signaling in the embryo, a gene called intermediate neuroblast defective (ind) is induced within minutes of ERK activation, while Cic remained localized in the nucleus (Lim et al., 2013). Based on these observations, it was proposed that the ERK-dependent antagonism of Cic repression is a two-tiered process, with fast relief of repressor function, followed by slower nuclear export and cytoplasmic degradation (Lim et al., 2013). The nature of the fast step in this process remained unknown.

Here we present the results of our direct tests of this model using optogenetic perturbations and live imaging experiments. While past studies relied on genetic perturbations that alter the endogenous ERK signals, we use optogenetic tools to deliver precisely timed ERK activation pulses to probe how the system responds. Using this approach, we established that short pulses of ERK activation lead to rapid gene expression, preceded by dissociation of Cic from regulatory regions of its target genes. We also demonstrate that sustained signaling is needed to keep Cic off DNA, indicating that repression is both rapidly relieved and rapidly reestablished.

RESULTS

Initial evidence for fast transcriptional response to ERK activation

We used optogenetic control of ERK activation by the OptoSOS system to investigate how precisely staged live embryos respond to short pulses of ERK activation (Figure 1A), (Johnson et al., 2017; Toettcher et al., 2013). For all presented experiments, blue light, which turns on ERK at high levels throughout the embryos within a minute (Johnson et al., 2017), was applied for varying amounts of time, and embryos were selected at early nuclear cycle 14, the peak time for endogenous ERK signaling during terminal patterning (Figure 1A, Goyal et al., 2017). In embryos with the Tor^D4021 mutation, which causes sustained and spatially uniform Torso activation, the segmented pattern of the future larva is disrupted, and the embryo does not hatch. We found that similar defects could be induced by 30 minutes of optogenetic stimulation (Figures 1B and 1C). This is expected based on previous findings that showed active ERK reduces the Cic protein half-life from 30 to 15 minutes (Grimm et al., 2012). Therefore, 30 minutes of light should reduce levels of Cic in the nucleus, and expression of Cic targets should lead to morphological defects (Figures 1B and 1C). Surprisingly, embryos exposed to 5 minutes of light also displayed patterning defects. While most of these embryos hatch, about 50% had segmentation defects, most commonly characterized by missing or fused denticle belts in the middle of the embryo (Figures 1B and 1D). The small defects that are seen in larvae from embryos exposed to 5-minutes of light may be the result of low levels of ectopic gene expression, suggesting that transcription of Cic target genes occurs quickly in response to ERK signaling prior to Cic export and degradation.

Figure 1. ERK activation induces a fast biological response and gain-of-function phenotypes.

(A) Mothers with optoSOS and Cic-sfGFP produce embryos used in this study. Embryos were collected for 2–4 hours and stimulated with blue light to activate ERK (dually phosphorylated ERK, dpERK). Following activation, embryos that were early NC 14 were selected from the collection.

(B) Cuticle preparations of representative gain-of-function phenotypes after light stimulation. Longer light stimulation induces stronger defects in abdominal segmentation. Scale bar: 100 μm.

(C) Embryonic lethality following different durations of light activation at NC 14, calculated as percent of unhatched embryos (30 – 100 total embryos counted per condition).

(D) Percentages of unhatched embryos and hatched larvae with normal phenotypes and abdominal defects following varying durations of light stimulation (30 – 100 total embryos counted per condition).

To confirm that Cic indeed leaves the nucleus on a timescale longer than 5 minutes, we examined nuclear localization of Cic in response to varying durations of light exposure. We used CRISPR-directed gene editing to generate Drosophila in which Cic is tagged with superfolder GFP (sfGFP) at the endogenous gene locus (Pédelacq et al., 2006). The Cic-sfGFP flies are homozygous viable and reproduce the wild type pattern of Cic localization (Figures 2A and S1). Levels of Cic in the middle of the embryos, where ERK signaling is normally absent, were compared to levels at the anterior, where ERK signaling is normally high. Consistent with previous measurements of Cic degradation time (Grimm et al., 2012), Cic-sfGFP levels were strongly attenuated in embryos exposed to 30 minutes of blue light (Figures 2B and S2).

Figure 2. Cic is exported from the nucleus after 30 minutes.

(A) Quantified nuclear Cic profile in a wild-type Cic-sfGFP embryo (fluorescence intensity measured as A.U.), and representative images of Cic localization under different activating conditions. Scale bar: 100 μm.

(B) Representative images of Cic at the poles vs. nuclei in the center of a fixed embryo. Quantification of nuclear Cic signal, represented as average ± standard deviation of intensities (A.U.) measured from three images for each condition. Cic is nuclear after 5 minutes of ERK activation, but is degraded after 30 minutes of ERK activation. Scale bar represents 15 μm. See also Figures S1 and S2.

A similar reduction in nuclear Cic is seen in the abdominal regions of mutant embryos with the Tor^D4021 mutation, which causes ligand-independent ERK activation starting from the second hour of development. Five-minute pulses of light cause no detectable changes in Cic levels or in its nuclear localization, regardless of whether it was scored immediately or 25 min later (Figures 2B, S1, and S2). Furthermore, intermediate durations (5–15 minutes) of light did not cause measurable changes in nuclear Cic (Figure S1). These results provide initial evidence that short pulses of ERK signaling can induce functionally significant transcriptional response while Cic is still in the nucleus.

ERK activation induces rapid and reversible relief of tailless repression

As a more direct test of the possibility that Cic target genes can be expressed on timescales that are shorter than those needed for Cic nuclear export, we focused on tll. This gene is an established target of Cic; furthermore, segmentation defects caused by genetic removal of Cic are fully suppressed in embryos that lack tll (Figure S3). We used the MS2:MCP system (Chubb et al., 2006; Garcia et al., 2013) to design a live reporter of tll transcription, using a regulatory region that recapitulates the wild type pattern of tll expression (Ajuria et al., 2011; Liaw et al., 1995). While the wild type expression pattern of the tll-MS2 reporter showed no activity in the central region of the embryo, consistent with direct repression of tll by Cic, we could detect the first signs of ectopic reporter activity ~6 minutes after the onset of optogenetic stimulation. More robust transcriptional activity is observed after 10 minutes of optogenetic stimulation (Figure 3A, Movies S1 and S2). Taken together with our analysis of Cic-sfFP nuclear levels, this fast response demonstrates that repression can be relieved before any significant reduction of the nuclear Cic concentration.

Figure 3. Transcriptional repression of the tll gene in response to ERK signal is rapid and reversible.

(A) tll transcription is shown using tll-MS2:MCP-mCherry following blue light stimulation for 15 minutes. Images depict max intensity projections of nuclei in the middle of the embryo where there is no endogenous ERK signal. Images have been segmented in ImageJ to separate nuclei from the OptoSOS signal. Nuclei with an MS2 dot, indicating active transcription, have been falsely colored purple. Quantification shows average number of nuclei with active transcription at each time for 7 movies. Error bars represent standard deviation. Scale bar: 15 μm.

(B) Same as (A), but tll transcription is shown after stimulating with blue light for 5 minutes, followed by 15 minutes of darkness. Quantification done for 6 movies. Scale bar: 15 μm. See also Figure S3 and Movies S1 – S4.

We next tested whether the ERK-dependent derepression of tll is irreversible or, alternatively, whether it requires sustained ERK signaling. In these experiments, embryos were stimulated by activating light for only 5 minutes. One minute after turning off the blue light, we see that the tll reporter begins to be activated, following the same kinetics as before. However, the reporter activity is no longer detected 15 minutes after blue light has been removed (Figure 3B, Movies S3 and S4). Thus, short activation pulses lead to only transient tll expression, demonstrating that repression is reestablished in the absence of sustained ERK signaling. This would explain the differential gain-of-function phenotypes that are observed with varying durations of blue light (Figures 1B and 1C). Indeed, with only five minutes of light, tll is expressed for a very short amount of time, resulting in low levels of accumulated tll and very small abdominal defects. On the other hand, embryos exposed to 30 minutes of light encounter higher cumulative levels of tll expression for longer amounts of time, inducing much stronger phenotypes.

Gene derepression is preceded by loss of Cic/DNA binding

Rapid relief of repression in response to ERK activation could reflect fast dissociation of Cic from the regulatory regions of its target genes. As a first step in evaluating this possibility, we used ChIP-qPCR to analyze how Cic binding to one of the tll enhancers (Ajuria et al., 2011; Lohr et al., 2009) is affected by ERK activation pulses of varying duration. The extent of Cic/DNA binding was assessed relative to binding in embryos with normal levels of ERK activation, where a majority of DNA comes from nuclei located in regions of the embryo where ERK is inactive. Consistent with the strongly attenuated levels of nuclear Cic, we found that binding was abolished in embryos exposed to either 30 minutes of optogenetic stimulation and in embryos with constitutively active Torso. Strikingly, Cic/DNA binding was also strongly reduced in embryos subject to only 5 minutes of ERK activation (Figure 4A). Thus, ERK activation can disrupt Cic/DNA binding in about 5 minutes. Furthermore, we performed ChIP-qPCR on embryos exposed to 5 minutes of light followed by differing durations of dark. Our results are consistent with the tll-MS2 data: once ERK signal is removed, Cic returns to DNA within 15 minutes, which explains why a 5-minute ERK activation pulse leads to only transient tll-MS2 expression (Figure 4B).

Figure 4. Short pulses of ERK signaling cause loss of Cic-DNA binding.

(A) ChIP-qPCR results for binding of Cic at the tll enhancer. ChIP signal is quantified as percent of DNA input that was pulled down by the GFP antibody. Error bars represent standard deviation of three replicates made up of 100 embryos each.

(B) ChIP-qPCR at the tll locus for embryos stimulated with 5 minutes of light, followed by a varying dark periods. Error bars represent standard deviation of three replicates.

(C) Heat maps depicting Cic-DNA binding (z-score calculated from read counts per million, see methods). Each row represents one of 103 regions that Cic binds to, and the read counts 1kb upstream and downstream of each peak. Read counts are merged from 3 replicates.

(D) Cic-DNA binding profiles for 4 different genes, at varying times of light stimulation, depicted as read counts per million from 3 merged replicates. Schematics of gene expression profiles in the embryo are also shown.

To test whether rapid dissociation from DNA is limited to tll or holds true for other targets of Cic, we combined optogenetic perturbations with time-resolved ChIP-seq assays. First, we found where Cic binds in the genome using Cic-sfGFP embryos without OptoSOS (Figure 4C). We focused on 103 binding regions that correspond to 89 unique gene targets (see Methods). We found that within these binding regions, there are a total of 92 optimal Cic binding consensus sites, ‘TC/GAATGAA’. If we also consider the number of weak Cic consensus sites, defined as having one nucleotide difference from the optimal binding site sequence, then there are a total of 255 sites. According to this count, only a small number of Cic molecules are needed to effectively repress gene targets throughout the embryo. The top Cic targets in our list include previously established Cic targets, such as tll, hkb, wntD, ind, and pnt. Our list closely matched previously published ChIP data performed with Cic antibody (Papagianni et al., 2018).

As a first step in analyzing which of these targets respond to ERK activation, we carried out ChIP-seq on embryos expressing a Torso GOF mutation. In the Tor^D4021 condition, we found that 72.8% of binding regions had reduced Cic interaction, confirming that Cic behaves similarly in response to ERK signaling for most gene targets. Using the OptoSOS system, we looked closer at the dynamics of Cic/DNA binding. As expected, we saw a similar loss of Cic/DNA binding after 30 minutes of light stimulation, with 80.6% of binding regions having reduced Cic. Consistent with what was observed at the tll enhancer, 5 minutes of activation also led to a significant reduction in Cic occupancy in over half of binding regions (Figures 4C and 4D). Thus, rapid loss of DNA binding is a general feature of the ERK-dependent relief of gene repression by Cic.

DISCUSSION

Several lines of evidence suggest that reducing nuclear localization and/or the total nuclear concentration of Cic is needed to relieve repression of target genes. This includes the negative correlation between nuclear Cic and tll expression in the early Drosophila embryo (Grimm et al., 2012; de las Heras and Casanova, 2006), as well as upregulation of Cic targets in cancers caused by loss of functional Cic (Bunda et al., 2019; Okimoto et al., 2017; Yang et al., 2017). Nevertheless, induction of several Cic targets was reported to commence before appreciable downregulation of nuclear Cic levels (Lim et al., 2013), suggesting the existence of a fast step in the ERK-dependent relief of transcriptional repression by Cic.

We presented several new results that corroborate this previously hypothesized fast step. First, patterning defects can be induced by signals of short duration that do not cause detectable attenuation of nuclear Cic localization or concentration (Figures 1 and 2). Second, live imaging of tll-MS2 activity provides direct proof that transcription can be robustly detected within minutes of ERK activation (Figure 3). Third, ERK activation causes loss of Cic binding to regulatory DNA on the timescale that is clearly shorter than either nuclear export or degradation of Cic (Figure 4). Thus, our study produced clear support for the two-tiered model of the ERK-dependent gene derepression and identified loss of DNA binding as the fast step in this model.

In addition to demonstrating that relief of repression is indeed fast, we found that gene derepression by ERK is not irreversible and that Cic targets can be rapidly re-repressed in the absence of ERK signaling (Figures 3 and 4). Once light stimulus is removed, we expect ERK activity to gradually decrease due to persistence of SOS binding to the membrane and dephorphoshorylation of ERK (Johnson and Toettcher, 2019; Toettcher et al., 2013). Thus, the process of gene activation by ERK can be likened to ringing a doorbell, which starts ringing as soon as the pressure on the doorbell is applied but gradually stops once pressure is relieved. How is repression reinstated? ChIP-seq data show that Cic binds to ~100 regulatory regions (Figure 4). Based on the known Cic consensus motifs and our ChIP-seq data, each of the 103 regulatory regions possess 2–3 binding sites on average. On the other hand, the total number of Cic molecules in the nucleus is estimated ~1000 (unpublished data from Liu Yang). This implies that most Cic molecules are not actively engaged in repression in the absence of signaling. It follows that one potential explanation of re-repression is that a short ERK activation pulse would target only DNA-bound Cic molecules, leaving the rest of Cic ready for reinstatement of repression once the pulse terminates. Although a recent study in mammalian cells proposed this idea (Bunda et al., 2019), this mechanism is ruled out by studies that demonstrate that in regions of high ERK signaling, nuclear exclusion is readily detected even for Cic variants with strongly attenuated DNA binding (Astigarraga et al., 2007). Based on this, we hypothesize that fast derepression requires phosphorylation of most of nuclear Cic molecules and that fast re-repression reflects rapid dephosphorylation of Cic and restoration of its function as a repressor. A further possibility could be that new Cic molecules are synthesized in the cytoplasm and imported into the nucleus. However, this scenario would most likely only be relevant on longer timescales.

All of the identified attributes of the ERK-dependent antagonism of gene repression by Cic – fast derepression, rapid re-repression, and slower nuclear export/degradation – appear to be essential for proper patterning of the embryo. Fast derepression is important because the syncytial embryo develops under tight temporal constraints. As an illustration, some Cic targets, such as wntD, must be expressed during nuclear cleavage cycles 11 and 12 (Ganguly et al., 2005). The interphases of these cycles are shorter than the timescale of Cic nuclear export and since transcription is limited to interphases, repression must be relieved on the timescale of minutes. On the other hand, rapid re-repression is important in the central regions of the embryo, where the segmentation processes must be guarded from inadvertent short pulses ERK activation. Specifically, pulses of ERK activation that last 1 to 3 minutes are observed throughout the embryo during mitotic phases of nuclear cleavages (unpublished data from Liu Yang). Since even such short pulses can relieve repression, it must be rapidly reestablished during the following interphases. This also means that ERK signaling must be sufficiently long-lasting to generate robust transcriptional responses for the induction of terminal cell fates and antagonism of cell fates that are normally limited to the central regions of the embryo. Our results suggest that such signals must last at least 30–45 minutes, which is comparable to the duration of endogenous ERK signaling at the poles and is sufficient to cause significant nuclear export and degradation of Cic.

In conclusion, our results provide new insights into dynamics of one of the best-studied examples of developmental ERK signaling and pose several questions for future studies. For instance, it will be very interesting to determine the biochemical origins of the fast dissociation DNA-bound Cic. In addition, it will be important to investigate how other ERK substrates, such as transcriptional corepressor Groucho, contribute to the timescales and mechanism of terminal patterning. Our study lays the foundations for studying dynamic control of transcriptional repression in a wide range of developmental contexts that depend on ERK activation.

STAR ★ METHODS

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Stas Shvartsman (stas@princeton.edu). Fly lines generated in this study are listed in the Key Resources Table and are available at the laboratory upon request.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Sheep polyclonal anti-GFP	Bio-Rad	Cat#4745-1051; RRID: AB_619712
Donkey Anti-Sheep IgG H&L (Alexa Fluor® 488)	Invitrogen	Cat#A-11015; RRID: AB_141362
Chemicals, Peptides, and Recombinant Proteins
DAPI fluorescent DNA stain	ThermoFisher Scientific	Cat#D1306 RRID: AB_2629482
Critical Commercial Assays
Protein G Dynabeads for immunoprecipitation	Invitrogen	Cat#10003D
Proteinase K	NEB	Cat# P8107S
NEBNext Ultra II DNA Library Prep Kit for Illumina	NEB	Cat#E7645S
NEBNext Multiplex Oligos for Illumina Kit	NEB	Cat#E7335S
Zymogen ChIP DNA Clean and Concentrator Kit	Zymogen	Cat#D5205
PowerUp SYBR Green Master Mix	Applied Biosystems	Cat#A25741
Deposited Data
Raw ChIP-seq data and list of Cic targets	This paper	GEO: GSE130584
Analysis scripts	This paper	https://github.com/keenansh/SEK_Cic_ChIP_Analysis.
Experimental Models: Organisms/Strains
D. melanogaster:–_OregonR	Bloomington Drosophila Stock Center	Stock# 5 RRID:BDSC_5
D. melanogaster:–_UAS-OptoSOS	(Johnson et al., 2017)	N/A
D. melanogaster:–_CicsfGFP	This paper	N/A
D. melanogaster:–_orD4021	(Schüpbach and Wieschaus, 1986)	FlyBase: FBal0016921
D. melanogaster:–_MTD-Gal4 (Maternal Triple Driver)	Bloomington Drosophila Stock Center	Stock# 31777 RRID:BDSC_31777
D. melanogaster: tll-MS2	This paper	N/A
D. melanogaster: MCP-mCherry	Levine Lab	N/A
Oligonucleotides
Forward primer for qPCR at tll enhancer, 5′ GGGTGGTTACCTGGCTTAGG 3′	This paper	N/A
Reverse primer for qPCR at tll enhancer 5′ ACGCCGTTCAGTCGCTAATC 3′	This paper	N/A
Software and Algorithms
Fiji	(Schindelin et al., 2009)	RRID: SCR_002285
SAMtools	(Li et al., 2009)	RRID:SCR_002105
MACS2	(Zhang et al., 2008)	RRID:SCR_013291
PicardMarkDuplicates	(http://broadinstitute.github.io/picard/)	RRID: SCR_006525
RStudio	https://RStudio.com	RRID:SCR_000432
R package: GenomicRanges	(Lawrence et al., 2013)	RRID:SCR_000025
R package: ChIPSeeker	(Yu et al., 2015)	https://www.bioconductor.org/packages/release/bioc/html/ChIPseeker.html
Galaxy	https://usegalaxy.org	RRID:SCR_006281

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Drosophila Strains and Fly Husbandry

Oregon-R (OreR), Tor^D4021/+, Cic-sfGFP/+, UAS-optoSOS (Johnson et al., 2017), tll-MS2, MCP-mCherry (provided by the Levine lab), and MTD-gal4 (Bloomington #31777) stocks were used in this study. CicsfGFP represents Cic endogenously tagged with superfolder GFP (Pédelacq et al., 2006). Flies were kept at room temperature in vials containing a standard mixture of agar, cornmeal, and yeast, provided by the Drosophila Media Core Facility within the Princeton Molecular Biology Department. To collect embryos, flies were placed in cages with an agar plate made with apple juice and supplemented with a yeast paste.

METHOD DETAILS

Generation of Crispr-tagged Cic with sfGFP

We generated a germline driven form of PBac transposase by cloning the PBac ORF into pNos-cas9 (Port et al., 2014) by isothermal assembly with NEB HiFi assembly master mix. This was accomplished by PCR amplifying the vector with primers GCGAATCCAGCTCTGGAG & GGCGAAAATCCGGGTCGAAAG. The PBac ORF was amplified from phsp-PBac (Handler and Harrell, 1999) with the primers gtaactttcgacccggattttcgccATGGGTAGTTCTTTAGACG & gcctctgctccagagctggattcgcTCAGAAACAACTTTGGCAC. The final vector pNos-PBac was inserted into the fly genome at landing sites attP40 and attP2 on the second and third chromosomes accordingly (Bischof et al., 2007a; Groth et al., 2004). Transgenesis was conducted by BestGene Inc. We utilized the scarless gene editing strategy developed by S. Gratz and K. M. O’Connor-Giles (flycrispr.molbio.wisc.edu,Bier et al., 2018). In short, the plasmid pU6-BbsI-chiRNA (Gratz et al., 2013) was digested with BbsI and the annealed primers cttcGTTCTGCGTTTGTTACTTAT & aaacATAAGTAACAAACGCAGAAC were cloned into the cut vector via T4 ligation. The homology directed repair vector pHD-sfGFP-ScarlessDsRed was PCR amplified in two separate pieces with primers AATTCGTTTAAACCTGCAGGAC & CGCGGCCGCTAAATTCAATTC in addition to primers GGTGGATCTGGAGGTTCC & GGAACCTCCTGAACCACC. The left homology arm was PCR amplified with primers gggcgaattgaatttagcggccgcgGCCCTAAAAAAGAAGAACG & agccgccggaacctccagatccaccACTGGTGGTCTGCATATTC and the right homology arm was amplified with primers aggttctggtggttcaggaggttccGCGGCAGGTAATTAGAAG & ctagtcctgcaggtttaaacgaattGACTGAAAATAGTGGAAGAAAG. The four PCR products with circularized with NEB HiFi assembly master mix and transformed into DH5α. This vector subsequently had the CRISPR PAM site mutated by the NEB Q5 Site-Directed Mutagenesis Kit with primers TTACTTATTGtTTTTTAGGTTATAAAGATC & CAAACGCAGAAAATGTTTAG. Transgenesis was conducted by BestGene Inc. The 3xP3::dsRed cassette was subsequently removed by crossing to a line containing the attP40 inserted Nos-PBac.

tll-MS2 reporter generation

The tll^0.15 enhancer sequence was amplified from pCaSpeR-hs43-lacZ (described in (Ajuria et al., 2011)) via PCR. The resulting fragment was then inserted in the pBphi-evePromoter-24X MS2-yellow vector (Bothma et al., 2014) via Gibson assembly. Transgenic lines carrying the tll^0.15-MS2 construct were generated by phiC31 integration in the 65B2 landing site (VK00033) conducted by BestGene Inc. (Bischof et al., 2007b; Groth et al., 2004).

Cuticle Preparations and Lethality

Embryos were collected on an agar plate for 2–4 hours in the dark. The plate of embryos was placed in a box illuminated by blue LED lights for varying durations of time. Embryos were then staged in halocarbon 27 oil on a stereomicroscope under red light. Embryos at early NC 14 were selected and placed on a separate agar plate. Selected embryos were allowed to develop in the dark for another 24–36 hours. Lethality was calculated by counting unhatched embryos and empty eggshells on the plate. Unhatched embryos were dechorionated in bleach. Dechorionated embryos and larvae were mounted onto a slide in media containing lactic acid and Hoyer’s medium (1:1) at 65°C overnight. Dark-field imaging was done with a Nikon Eclipse Ni at 10X objective.

Immunostaining

Embryos were collected for 2–4 hours, dechorionated under red light (to suppress activation of optoproteins) and exposed to blue light for varying lengths of time. From there, immunostaining protocols were performed as described elsewhere (Lim et al., 2015). Antibodies: sheep anti-GFP (1:500, Bio-Rad, 4745–1051), Alexa Flour conjugate 488 (1:1000, Invitrogen), and DAPI (1:5000, Molecular Probes, D1306). Embryos were fixed at the beginning of NC 14. In Figures 2A and S1, images of embryos have been cropped from the original image and rotated so that the orientation is anterior on the left and posterior on the right.

Microscopy

Fixed imaging for Cic-sfGFP antibody staining and live imaging of tll transcription with the MS2:MCP system were done using confocal microscopy, performed on a Nikon A1 RS (Princeton Microscopy Core). The 20x air and 60x oil objectives were used respectively. For Cic-sfGFP visualization, embryos were imaged at the midsagittal plane. For imaging of tll-MS2:MCP-mCherry, z-stacks were taken starting from the surface of the embryo, into a depth of 7 m with a step size of 0.7 m. Embryos were imaged every 30 seconds. A digital micro-mirror device was used to stimulate embryos with blue light while imaging in the red channel for tll-MS2 transcription.

Embryo Collection for ChIP

Embryos with Cic-sfGFP and OptoSOS constructs were collected for 2–4 hours in the dark and dechorionated in bleach for 2 minutes under red light, which keeps optogenetic constructs turned off. Dechorionated embryos were placed under blue light for varying amounts of time. Embryos were fixed for 15 minutes in a mixture of 20% paraformaldehyde, PBS, and TritonX-100. Crosslinking was quenched with glycine in PBS + TritonX-100. Fixed embryos were staged on a dissecting microscope to select embryos at early NC14 and stored at −80°C. For more details, see Hannon et al., 2017. Each ChIP sample was composed of ~100 embryos at early NC14.

Chromatin Immunoprecipitation

A solution of RIPA buffer, protease inhibitor cocktail, and DTT were added to frozen embryos. Embryos were homogenized with a pellet pestle, followed by 7 rounds of sonication at 20% output for 20 seconds at 4°C. 1% of sonicated chromatin solution was save d as “input” DNA, which is effectively a sample of all chromatin that exists in the embryo. The remaining solution was incubated with Anti-GFP antibody (AB3080P) overnight at 4°C was used to pull down Ci c-sfGFP. Protein G DynaBeads (10003D) were also blocked with a BSA solution at this time. Antibody/chromatin solution was transferred to the tubes with the magnetic beads and incubated at 4°C for 1 hour on a nutator. The beads were washed a total of 5 times. Purified protein/DNA complexed were eluted by shaking the beads at 65°C for 10 minutes. DNA was uncrosslinked from the protein using Proteinase K (P8107S), and it purified using Zymogen ChIP DNA Clean and Concentrator Kit (D5201). For more details, please see Blythe and Wieschaus, 2015.

ChIP-qPCR at the tll enhancer

ChIP samples were analyzed by qPCR using Sybr Green detection in triplicate. Primers were designed to amplify a region of the tailless enhancer near Cic octameric binding sites: forward (5’ GGGTGGTTACCTGGCTTAGG 3’), reverse (5’ ACGCCGTTCAGTCGCTAATC 3’). RT-qPCR was performed with the ViiA7 Real-Time PCR System. ChIP signal was quantified as the percentage of the input pulled down by the antibody (Haring et al., 2007)

ChIP Library Preparation and Sequencing

Purified ChIP and input DNA were prepared for Illumina sequencing using the NEBNext Ultra II DNA Library Prep Kit for Illumina (E7645S). The primers used for multiplexing the library came from the NEBNext Multiplex Oligos for Illumina Kit (E7335S). Single end sequencing of libraries using Illumina HiSeq 2500 with read length of 75 bp. Three biological replicates were run for each condition.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical details are included in the figure legend for the corresponding experiment.

Quantification of Cic-sfGFP Nuclear Signal

Composite images of lateral cross-sections stained for GFP and DAPI were loaded into imageJ. Images were cropped to include 10–15 nuclei from the anterior portion of the embryo or the middle of the embryo. A mask was created using the Dapi signal to identify nuclei. Average fluorsence intensity (A.U.) of the Dapi and Cic-sfGFP signals were calculated across nuclei. The final calculation to compare nuclei in the middle and the anterior of the embryo was as follows:

$$\frac{(\text{Intensity}\text{of}\text{GFP}\text{in}\text{Middle}\text{Nuclei})}{(\text{Intensity}\text{of}\text{DAPI}\text{in}\text{Middle}\text{Nuclei})}÷\frac{\text{(Intensity}\text{of}\text{GFP}\text{in}\text{Anterior}\text{Nuclei)}}{\text{(Intensity}\text{of}\text{DAPI}\text{in}\text{Anterior}\text{Nuclei)}}$$

For Cic anterior- to - posterior profiles provided in Figure 2A and Figure S2, images of NC14 embryos taken at the midsagittal plane were analyzed in MATLAB to measure Cic-sfGFP intensity. A MATLAB script was generated in the past (Coppey et al., 2008) to find the perimeter of the embryo and average the staining intensity across a line normal to the boundary. This is done for 1000 points along the entire perimeter. These are averaged for multiple images and plotted as arbitrary fluorescence units along the AP axis.

Quantification of tll-MS2 Dots

Z-stacks of tll-MS2:MCP-mCherry embryos were loaded into ImageJ. Images were smoothened and max intensity projections were done for each time frame. It is important to note that the OptoSOS construct is labeled in red in addition to MCP-mCherry. OptoSOS can be seen being recruited to the membrane surrounding nuclei within a minute of blue light being applied. In this way, OptoSOS outlines the nuclei. Local thresholding in ImageJ using the Otsu algorithm was used to create a mask to outline nuclei and determine the number of nuclei within a given frame. To remove the OptoSOS signal, Gaussian blur was used to smooth the OptoSOS signal and this was subtracted from the original image. This was done twice. A threshold was applied to MS2 dots remaining in the image using Otsu algorithm. Total number of dots were counted based on this threshold intensity and a pixel size of 3 or greater for each frame.

ChIP-Sequencing Data Analysis

Raw sequencing data for each of the three replicates for each condition were uploaded to galaxy.princeton.edu. Within Galaxy, sequences were trimmed and mapped to the Drosophila genome (dm6). SAMtools (Li et al., 2009) was used to remove sequences with a quality score less than 20, and PicardMarkDuplicates (http://broadinstitute.github.io/picard/) were used to remove the duplicate reads from each replicate. Processed BAM files were downloaded and each replicate was merged into one BAM file for each condition using SAMtools. Regions of enrichment were determined for Cic-sfGFP embryos using MACS2 (Zhang et al., 2008). Because the GFP antibody non-specifically pulls down certain regions of DNA, our control for enrichment was OreR embryos ChIPped with the same GFP antibody. MACS2 peak calling parameters were: -q 0.01 --keep-dup all –nomodel –extsize 145. The parameter ‘extsize’ was set as the predicted fixed size of fragments calculated for the merged replicates from the MACS2 ‘predictd’ function. Other analysis was done using several R packages including: GenomicRanges and ChIPSeeker (Lawrence et al., 2013; Yu et al., 2015). MACS2 originally called 468 enriched peaks; We only considered peaks with a log₂(fold change) of 5 or greater. Additionally, we only kept peaks in chromosomes: 2R, 2L, 3R, 3L, 4, and X. Lastly, we removed peaks found in the dsor1 and rl loci because they were suspected to be contamination. We were left with 103 consensus peak regions for Cic binding. Coverage of read counts per million were determined across 10 base-pair windows of the genome. Data was normalized with z-scores, which were calculated using mean CPM and standard deviation of CPM in noise peaks, generated based on a procedure described elsewhere (Blythe and Wieschaus, 2016). Loss of DNA binding in each peak was calculated as the log₂ fold change in read counts for each peak compared to wild type. Reads counts in the summits of light-stimulated peaks that were less than 50% of the read counts in wild type peaks were considered to be ‘reduced’.

Supplementary Material

1. Movie S1, related to Figure 3. Embryo 1 showing induction of tll.

Active transcription of tll following light stimulation. Several minutes into nuclear cycle 14, blue light was applied to the embryo while imaging in the red channel. Signal represents OptoSOS tagged with RFP, which is recruited to the membrane surrounding nuclei upon light stimulation. Dots present within nuclei are tll-MS2 transcripts binding to MCP-mCherry protein. Blue light was applied continuously for 15 minutes. Z-stacks 7 μm into the embryo were acquired every 30 seconds. Scale bar: 15 μm.

2. Movie S2, related to Figure 3. Embryo 2 showing induction of tll.

Active transcription of tll following light stimulation. Several minutes into nuclear cycle 14, blue light was applied to the embryo while imaging in the red channel. Signal represents OptoSOS tagged with RFP, which is recruited to the membrane surrounding nuclei upon light stimulation. Dots present within nuclei are tll-MS2 transcripts binding to MCP-mCherry protein. Blue light was applied continuously for 15 minutes. Z-stacks 7 μm into the embryo were acquired every 30 seconds. Scale bar: 15 μm.

3. Movie S3, related to Figure 3. Embryo 1 exhibiting return of tll repression.

Active transcription of tll following light stimulation. Several minutes into nuclear cycle 14, blue light was applied to the embryo while imaging in the red channel for 5 minutes. Blue light was turned off and imaging continued for an additional 15 minutes. Signal represents OptoSOS tagged with RFP, which is recruited to the membrane surrounding nuclei upon light stimulation. Dots present within nuclei are tll- MS2 transcripts binding to MCP-mCherry protein. Z-stacks 7 μm into the embryo were acquired every 30 seconds. Scale bar: 15 μm.

4. Movie S4, related to Figure 3. Embryo 2 exhibiting return of tll repression.

Active transcription of tll following light stimulation. Several minutes into nuclear cycle 14, blue light was applied to the embryo while imaging in the red channel for 5 minutes. Blue light was turned off and imaging continued for an additional 15 minutes. Signal represents OptoSOS tagged with RFP, which is recruited to the membrane surrounding nuclei upon light stimulation. Dots present within nuclei are tll- MS2 transcripts binding to MCP-mCherry protein. Z-stacks 7 μm into the embryo were acquired every 30 seconds. Scale bar: 15 μm.

Highlights

• ERK signaling in the Drosophila embryo antagonizes repression in a two-step process

• ERK activation leads to rapid loss of a transcriptional repressor from the DNA

• Transcriptional repression is reestablished once signal is removed

• Signal must persist for proper gene expression and pattern formation