1. Developmental Biology
Download icon

Enhancer additivity and non-additivity are determined by enhancer strength in the Drosophila embryo

  1. Jacques P Bothma
  2. Hernan G Garcia
  3. Samuel Ng
  4. Michael W Perry
  5. Thomas Gregor  Is a corresponding author
  6. Michael Levine  Is a corresponding author
  1. University of California, Berkeley, United States
  2. Princeton University, United States
Research Article
  • Cited 65
  • Views 5,154
  • Annotations
Cite this article as: eLife 2015;4:e07956 doi: 10.7554/eLife.07956

Abstract

Metazoan genes are embedded in a rich milieu of regulatory information that often includes multiple enhancers possessing overlapping activities. In this study, we employ quantitative live imaging methods to assess the function of pairs of primary and shadow enhancers in the regulation of key patterning genes-knirps, hunchback, and snail-in developing Drosophila embryos. The knirps enhancers exhibit additive, sometimes even super-additive activities, consistent with classical gene fusion studies. In contrast, the hunchback enhancers function sub-additively in anterior regions containing saturating levels of the Bicoid activator, but function additively in regions where there are diminishing levels of the Bicoid gradient. Strikingly sub-additive behavior is also observed for snail, whereby removal of the proximal enhancer causes a significant increase in gene expression. Quantitative modeling of enhancer–promoter interactions suggests that weakly active enhancers function additively while strong enhancers behave sub-additively due to competition with the target promoter.

https://doi.org/10.7554/eLife.07956.001

eLife digest

Only a subset of the genes in a cell is active at any time. Gene activation or ‘transcription’ is controlled by specific DNA sequences called promoters and enhancers. Promoters are found next to genes and recruit the protein machinery needed to transcribe the gene. Enhancers are located further away from genes and interact with the promoter to increase gene transcription.

Some genes have multiple enhancers with overlapping activities, typically including a primary enhancer and secondary or ‘shadow’ enhancer that is even more distant. Multiple enhancers are used to control gene transcription in different types of cells. For example, shortly after fertilization around 1000 enhancers in the embryo of the fruit fly Drosophila regulate several hundred genes that are important for development. These activities establish which cell type any given cell in the embryo will become. Many of these developmental genes have shadow enhancers, which are thought to make gene activity more precise and reliable. It is not known, however, how several enhancers interact with the same promoter at the same time. For example, do the enhancers' individual effects add together? Or can the combined effects of multiple enhancers be more (or even less) than the sum of their parts?

Bothma, Garcia et al. have now examined how multiple enhancers regulate the activity of three developmental genes (called hunchback, knirps, and snail) in early Drosophila embryos. The experiments showed that the individual effects of the knirps enhancers add together as one might expect. On the other hand, the snail enhancers interfere with each other, which means that their combined effect on transcription is less than the sum of the two individual effects. Furthermore, other experiments revealed that the combined effect of the hunchback enhancers depends upon whether another component that is needed for gene activation is in short supply.

To understand these observations, Bothma, Garcia et al. then developed a mathematical model. The model proposes that behavior of enhancers depends upon how strongly they interact with the target promoter. Since ‘weak’ enhancers do not interact very often, their effects can easily add together. However, the effects of ‘strong’ enhancers do not add together because they often compete to interact with the promoter. These findings show how multiple enhancers can work in a complex manner to control gene transcription.

https://doi.org/10.7554/eLife.07956.002

Introduction

There is emerging evidence that metazoan genes occur in a complex regulatory landscape encompassing numerous enhancers (Jeong et al., 2006; Hong et al., 2008; Frankel et al., 2010; Perry et al., 2011; Rada-Iglesias et al., 2011; Buecker and Wysocka, 2012; Lagha et al., 2012; Levine et al., 2014). For example, the mouse Sonic Hedgehog gene is regulated by at least 20 different enhancers scattered over a distance of ∼1 Mb (Jeong et al., 2006). Individual enhancers mediate expression in a variety of different tissues, including the brain, floorplate, and limb buds. Multiple enhancers with overlapping regulatory activities are also used to control gene expression within individual cell types. For example, the transcriptional activation of the gap gene hunchback (hb) in the early Drosophila embryo is mediated by both a proximal enhancer and distal ‘shadow’ enhancer that independently mediate activation in response to high levels of the Bicoid activator gradient (Buecker and Wysocka, 2012). Despite overwhelming evidence for multiple enhancers regulating the same gene it is unknown whether they simultaneously interact with the same promoter in a given cell. Here, we use a combination of quantitative live imaging and theoretical modeling to investigate the function of multiple enhancers for the regulation of a common target gene within a single cell type.

The fate map of the adult fly is established by ∼1000 enhancers that regulate several hundred patterning genes during the 1-hr interval between two and three hrs after fertilization (Levine, 2010; Nien et al., 2011). As many as half of these genes contain ‘shadow’ enhancers with overlapping spatiotemporal activities that are thought to improve the precision and reliability of gene expression (Hong et al., 2008; Frankel et al., 2010; Perry et al., 2010, 2011; Lagha et al., 2012; Miller et al., 2014). For example, the hb shadow enhancer helps produce a sharp boundary of activation by the Bicoid gradient, while its snail (sna) counterpart helps ensure reliable activation under stressful conditions such as high temperatures (Perry et al., 2010, 2011). There is emerging evidence that shadow enhancers are used pervasively in a variety of developmental processes, in both invertebrates and vertebrates (Lagha et al., 2012; Arnold et al., 2013; Miller et al., 2014; Lam et al., 2015).

The underlying mechanisms by which two enhancers with extensively overlapping regulatory activities produce coordinated patterns of gene expression are uncertain. It is possible that they augment the levels of gene expression above the minimal thresholds required to execute appropriate cellular processes (Gregor et al., 2014). However, there is currently only limited experimental evidence for enhancers acting in an additive fashion (Arnold et al., 2013; Lam et al., 2015). An alternative view is that shadow enhancers suppress transcriptional noise and help foster uniform expression among the different cells of a population (Buecker and Wysocka, 2012). To explore these and other potential mechanisms, we examined the timing and levels of gene activity using bacterial artificial chromosomes (BAC) transgenes containing individual enhancers and combinations of primary and shadow enhancers in the early Drosophila embryo.

BAC transgenes containing three key patterning genes, hb, knirps (kni), and sna, were examined in living precellular embryos. Quantitative analyses suggest that shadow enhancers mediate different mechanisms of transcriptional activity. For kni, we observe additive, even super-additive, activities of the primary and shadow enhancer pairs. In contrast, the hb enhancers function sub-additively in anterior regions containing saturating levels of the Bicoid activator but function additively in regions where there are diminishing levels of the Bicoid gradient. Strikingly, sub-additive behavior is also observed for sna, in that removal of the proximal enhancer causes a significant increase in gene expression. These observations suggest that the levels of enhancer activity determine the switch between additive and non-additive behaviors.

Using theoretical modeling, we suggest that these behaviors can be understood in the context of enhancers competing or cooperating for access to the promoter. Weak enhancers work additively due to infrequent interactions with the target promoter, whereas strong enhancers are more likely to impede one another due to frequent associations. Our results highlight the potential of combining quantitative live imaging and modeling in order to dissect the molecular mechanisms responsible for the precision of gene control in development (Gregor et al., 2014) and provide a preview into the complex function of multiple enhancers interacting with the same promoter.

Results

Previous live-imaging studies have relied on simple gene fusions containing a single enhancer attached to a reporter gene with MS2 stem loops inserted in either the 5′ or 3′ UTR (Garcia et al., 2013; Lucas et al., 2013; Bothma et al., 2014). Detection depends on the binding of a maternal mRNA-binding fusion protein (MCP::GFP) expressed throughout the early embryo. In order to examine the interplay between multiple enhancers, we created a series of BAC transgenes containing complete regulatory landscapes (summarized in Figure 1). The BAC transgenes contain an MS2-yellow reporter gene in place of the endogenous transcription units (Figure 1A). For each locus, hb, kni, and sna, we examined a series of three BAC transgenes: containing both primary and shadow enhancers, as well as derivatives lacking individual enhancers (Figure 1B,C). As expected, the BAC transgenes containing both enhancers produce robust expression of the MS2 reporter gene which recapitulate endogenous patterns previously measured using mRNA FISH and immunostaining (Perry et al., 2010, 2011) (Figure 1D–I, Videos 1–3).

Figure 1 with 1 supplement see all
Live-imaging of transcriptional activity of hb and kni loci lacking different enhancers.

(A) General structure of the reporter constructs. A reporter construct with 24 repeats of the MS2 stem loops and the yellow gene was recombined into BACs spanning the hb and kni loci. The 5′ UTR and 3′ UTR of the endogenous genes were left intact. The MCP::GFP protein that binds to the MS2 stem loops is present in the unfertilized egg and in the early embryo. Gene models of (B) the hb and (C) kni loci showing the location of the primary and shadow enhancers (Perry et al., 2011). (D, F, H) Snapshots of Drosophila embryos expressing different versions of the hb BAC>MS2 reporter containing different combinations of the two enhancers 10 min into nuclear cleavage cycle 13 (nc13). The colored bar on the bottom right indicates which enhancer was removed. (E, G, I) Snapshots of Drosophila embryos expressing different versions of the kni BAC>MS2 reporter containing different combinations of the two enhancers in nc14.

https://doi.org/10.7554/eLife.07956.003
Video 1
Dynamics of hunchback expression.

Maximum projection of hb BAC>MS2 transgene from nc10 to gastrulation, MCP::GFP in green and histone in red, anterior to the left and ventral view up. Time elapsed since the start of imaging is indicated in top left. The initial pattern is restricted to the anterior where expression is driven by the primary and shadow enhancers. In late nc13 the central domain enhancer starts to be expressed.

https://doi.org/10.7554/eLife.07956.005
Video 2
Dynamics of knirps expression.

Maximum projection of kni BAC>MS2 transgene from nc10 to gastrulation, MCP::GFP in green and histone in red, anterior to the left and ventral view down. Time elapsed since the start of imaging is indicated in top left. The dynamics of the anterior and central parts of the pattern are evident.

https://doi.org/10.7554/eLife.07956.006
Video 3
Dynamics of snail expression.

Maximum projection of snail BAC>MS2 transgene from nc10 to gastrulation, MCP::GFP in green and histone in red, anterior to the left and ventral view up. Time elapsed since the start of imaging is indicated in top left.

https://doi.org/10.7554/eLife.07956.007

Enhancer ‘deletions’ were created by substituting native sequences with neutral sequences of similar sizes (see ‘Materials and methods’). These substitutions remove most of the critical sequences identified by ChIP-Seq assays (Harrison et al., 2011; Nien et al., 2011; Perry et al., 2011). It is nonetheless possible that critical flanking sequences persist within the transgenes. However, removal of both kni enhancers eliminates detectable transcripts in abdominal regions of early embryos (Figure 1—figure supplement 1), suggesting that any remaining flanking sequences are insufficient to mediate expression.

Qualitative inspection of the hb and kni expression videos suggests that removal of either the primary or shadow enhancer does not cause a dramatic alteration in the overall patterns of gene activity. In order to identify more nuanced changes, we quantified the transcriptional activities of the complete series of BAC transgenes (Figure 2). The fluorescence intensities of active transcription foci were measured during nuclear cleavage cycles (nc) 13 and 14 at different positions across the anterior–posterior (AP) axis. These intensities were converted into an absolute number of elongating Pol II molecules by calibrating with internal standards (see Garcia et al., 2013). Several embryos were analyzed for each time point, and the data were merged to determine the average behavior as a function of AP position and time.

Combined effect of multiple enhancers as a function of AP position.

(A, B) Mean number of Pol II molecules transcribing per nucleus (NPol II) in the hb BAC reporters containing different combinations of enhancers as a function of AP position for two time points in nc13. NPol II is calculated by averaging data from at least three embryos at each AP position. The predicted sum of the individual enhancers is also shown. Note the additivity at the boundary vs the sub-additivity at the core, anterior domain of the pattern. (C, D) Mean number of Pol II molecules transcribing per nucleus (NPol II) in the kni BAC reporters in nc14 as a function of AP position. For kni, we see super-additive behavior in the beginning of nc14 which then becomes additive later in nc14. The absolute number of transcribing Pol II molecules was estimated following a previous calibration (Garcia et al., 2013). Error bars are the standard error of the mean over multiple embryos.

https://doi.org/10.7554/eLife.07956.008

Hb expression was examined during the ∼15 min interphase of nc 13 when both the primary and shadow enhancers are active, but before the onset of later-acting ‘stripe’ enhancers during nc 14 (Perry et al., 2011, 2012). We measured the transcriptional activity of all three hb BAC transgenes (Figure 1D,F,H). Contrary to simple expectations suggested by previous studies (Arnold et al., 2013), we find that the two hb enhancers do not function in an additive fashion in anterior regions (20–40% egg length, EL) of the embryo (e.g., Figure 2A,B). Indeed, the levels produced by the wild-type transgene fall far short of the additive levels predicted by simply summing the levels of expression produced by the transgenes containing either the shadow or primary enhancer alone (Figure 2A,B). Moreover, the removal of the shadow enhancer has no effect on the levels of transcription in anterior regions, which is consistent with the original conception of the shadow enhancer as a ‘back-up’ in the event of stress (Hong et al., 2008).

A very different scenario is observed in central regions of the embryo (40–50% EL) where hb expression switches from ‘on’ to ‘off’ to form a sharp border (Gregor et al., 2007). In this region, the wild-type transgene produces significantly higher levels of expression than either of the transgenes driven by a single enhancer. In fact, these levels correspond to the values predicted by simply adding the activities of the single-enhancer transgenes. Thus, the two enhancers transition from sub-additive to additive behavior in the region of the embryo where there are diminishing levels of the Bicoid activator gradient. We therefore suggest that the hb enhancers function additively only when they are operating below peak capacity (see below).

To further explore the activities of multiple enhancers, we examined the kni gene, which is regulated by an intronic enhancer and a distal 5′ enhancer (Perry et al., 2011). We focus our analysis on central regions of the abdominal expression pattern since previous studies suggest the occurrence of long-range repressive interactions that establish the borders of the ‘stripe’ (Perry et al., 2011). During early periods of nc 14, the two enhancers function super-additively (Figure 2C). That is, the wild-type kni BAC transgene produces higher levels of expression than the predicted sum of the two transgenes containing either enhancer alone. During later stages of development, there is a twofold reduction in the expression levels of the endogenous gene, and at this time the two enhancers work in a simple additive manner (Figure 2D). Note that the maximum number of elongating polymerase (Pol II) complexes falls short of that seen for hb (compare with Figure 2A,B).

Understanding the stark difference in the behaviors of the hb and kni enhancer pairs necessitates measuring the absolute strengths of the different enhancers. Using absolute counts of mRNA molecules (Little et al., 2013), we calibrated our live fluorescence intensity traces to determine the average numbers of actively elongating Pol II transcription complexes (Garcia et al., 2013). The kni transgenes containing single enhancers exhibit as little as fourfold lower levels of expression as compared with the corresponding hb transgenes (Figure 3A,B). At peak activity, the proximal hb enhancer induces ∼50 transcribing Pol II complexes across the yellow reporter gene. By contrast, individual kni enhancers produce an average of only ∼15 elongating Pol II complexes. We propose that the additive and super-additive behaviors of the two kni enhancers reflect their inherently ‘weaker’ activities as compared with the ‘stronger’ proximal hb enhancer (see ‘Discussion’). Note that, despite these differences, the overall output of transcripts and the overall rate of transcript production are essentially identical for all gap genes (Little et al., 2013).

Combined effect of multiple enhancers as a function of time.

(A) Time course of the mean number of Pol II molecules transcribing per nucleus (NPol II) for the different hb BAC transgenes and sum of individual enhancers at 27% EL for the duration of nc13. (B) kni BAC transgenes activities and the sum of individual enhancer activity at 60% EL for the first 50 min of nc14. (C) sna BAC transgenes and the sum of individual enhancer activities averaged over the central mesoderm for the initial 50 min of nc14. Error bars are the standard error of the mean over multiple embryos.

https://doi.org/10.7554/eLife.07956.009

To test the proposed anti-correlation between enhancer strength and additivity, we analyzed the expression of sna, which is essential for delineating the invaginating mesoderm during gastrulation. sna transgenes containing either the proximal or distal enhancer produce peak transcriptional activities of ∼40 actively transcribing Pol II complexes across the yellow reporter gene, similar to the numbers seen for the proximal hb enhancer (Figure 3C). Thus, both sna enhancers are strong and they exhibit strikingly sub-additive behaviors. In particular, the wild-type transgene displays significantly lower levels of expression than the mutant transgene containing only the shadow enhancer. Thus, strong enhancers not only fail to function additively but also interfere with one another, leading to sub-additive expression levels. This observation is also consistent with an earlier study, which suggested that the weaker proximal enhancer attenuates the activities of the stronger distal shadow enhancer (Dunipace et al., 2011).

In an effort to understand how multiple enhancers might function additively or sub-additively, we developed a mathematical model for dynamic enhancer–promoter interactions. In this model, a single enhancer interacts with its promoter via a forward rate kon and a backward rate koff (Figure 4A). The relative values of the forward and reverse rates determine the strength of the enhancer–promoter interaction by controlling what fraction of time the two are bound. When the enhancer and promoter interact, the promoter is in the ON state and initiates transcription at a rate r. This rate can be interpreted as the efficiency of enhancer-mediated transcriptional initiation upon enhancer–promoter interaction. Hence, the observable rate of mRNA production depends on the interaction strength given by the ratio kon/koff, and the efficiency r with which transcription is initiated upon interaction (Figure 4B). This scheme can be generalized to include two enhancers (A and B) interacting with the same promoter (Figure 4C and see ‘Materials and methods’ for details of the mathematical analysis). In this model only one enhancer can interact with the promoter at a given time.

Model of enhancer–promoter interactions and its predictions for mRNA production.

(A) Minimal model of one enhancer engaging a promoter. kon and koff are the rates of promoter engagement and disengagement, respectively, and determine the interaction strength. r is the rate of mRNA production when the promoter is engaged and is a measure of the transcriptional efficiency. The mean number of Pol II molecules transcribing per nucleus (NPol II) is proportional to the rate of mRNA production. (B) As the interaction strength of a single enhancer is increased, the amount of mRNA produced increases up to a maximum value dictated by the transcriptional efficiency. (C) The model in (A) can be generalized to allow for multiple enhancers interacting with the same promoter. (D) In the regime where the interaction strength of both promoters is weak (kon/koff = 0.01), the amount of mRNA produced by having both A and B is simply the sum of the individual contributions of A and B, (r = 1). (E) In the regime where the interaction strength is large, the combined activity of both enhancers can be significantly less than the sum the individual enhancers. A less efficient enhancer A (rA = 0.2 au) can interfere with the more efficient enhancer B (rB = 1 au) such that their combined activity is significantly less than the sum of the activities of individual enhancers.

https://doi.org/10.7554/eLife.07956.010

When the individual enhancers interact infrequently with the promoter (kon/koff << 1) they are unlikely to attempt to engage the promoter simultaneously. In this regime, the enhancers will work additively, and the rate of mRNA production of enhancers A and B will simply be the sum of the production rate of A and B alone as shown in Figure 4D. However, as the strength of promoter–enhancer interactions increases, the combined activity of both enhancers is less than the sum of each individual enhancer. When we model enhancers of different strengths (Figure 4E), the amount of mRNA production is reduced as the enhancer with the weaker transcriptional efficiency interacts more frequently with the promoter. This occurs because the two enhancers compete for access to the promoter, effectively inhibiting one another. Thus, weak enhancers might work additively due to infrequent interactions with the target promoter, whereas strong enhancers interfere with one another due to more frequent interactions (see below).

Discussion

Our quantitative analysis of hb and kni expression provides seemingly opposing results. For kni, we observe additive, sometimes even super-additive, action of the two enhancers within the presumptive abdomen. In contrast, the two hb enhancers do not function in an additive fashion in anterior regions but are additive only in central regions where expression abruptly switches from ‘on’ to ‘off’. We propose that ‘weak’ enhancers function additively or even super-additively, whereas ‘strong’ enhancers can impede one another (Figure 5).

Theoretical expectation and experimental results showing different regimes of combined enhancer action.

(Upper left) Theoretical predictions (yellow) illustrating how the rate of mRNA production from both enhancers, NPol IIPrimary+Shadow, varies with the sum of the activity of the individual enhancers, NPol IIPrimary+NPol IIShadow (yellow). Mean number of Pol II molecules transcribing per nucleus (NPol II) is proportional to the rate of mRNA production. The green line shows perfect additivity for comparison. The model predicts additive behavior (NPol IIPrimary+ShadowNPol IIPrimary+NPol IIShadow) when the rate of production is low and sub-additive behavior (NPol IIPrimary+Shadow<NPol IIPrimary+NPol IIShadow) as the production rate increases. As the interaction strength of individual enhancers increases so does the rate of mRNA production, but the combined activity of both enhancers becomes sub-additive. (Upper right, lower left, lower right) Transcriptional activity of intact loci vs the sum of activities of individual enhancers for hb, kni, and sna at different times. A green line has been drawn in to indicate where NPol IIPrimary+Shadow is equal to NPol IIPrimary+NPol IIShadow. For hb and kni, the plots show data taken at different AP positions at 10 min into nc 13 and 20 min into nc 14, respectively, while for sna the datapoints were at different times. Ellipses indicate standard error of the mean.

https://doi.org/10.7554/eLife.07956.011

Additional support for this view is provided by the analysis of sna. We found that the removal of the proximal enhancer significantly augments expression, consistent with the occurrence of enhancer interference within the native locus. It is also conceivable that a single strong enhancer (e.g., hb proximal or sna distal) already mediates maximum binding and release of Pol II at the promoter, and additional enhancers are therefore unable to increase the levels of expression. However, the increase in the levels of sna expression upon removal of the primary enhancer is inconsistent with this explanation. Perhaps, the proximity of the proximal enhancer to the sna promoter gives it a ‘topological advantage’ in blocking access of the distal enhancer (Dunipace et al., 2011). The proximal enhancer might mediate less efficient transcription than the distal enhancer and thereby reduce the overall levels of expression (see Figure 4E). We do not believe that this proposed difference is due to differential rates of Pol II elongation since published (Garcia et al., 2013) and preliminary studies suggest that different enhancers and promoters lead to similar elongation rates (∼2 kb/min; T Fuyaka and M Levine, unpublished results). A nonexclusive alternative possibility is that deletion of the proximal enhancer removes associated sna repression elements (MacArthur et al., 2009), thereby augmenting the efficiency of the distal enhancer.

A minimal model of enhancer–promoter associations provides insights into potential mechanisms. In the parameter regime where such interactions are infrequent the two enhancers display additive behavior. However, in the regime of frequent interactions, enhancers compete for access to the promoter resulting in sub-additive behavior. Enhancer–promoter interaction parameters are likely to vary not only between different enhancers but also as the input patterns are modulated in time and space during development (Rushlow and Shvartsman, 2012; Kok and Arnosti, 2015).

This simple model explains the switch from sub-additive to additive enhancer activities for hb and sna. However, in order to explain the super-additive behavior of the kni enhancers, it would be necessary to incorporate an additional state in the model, whereby both enhancers form an active complex with the same target promoter. Such a complex would have a more potent ability to initiate transcription than individual enhancer–promoter interactions.

In summary, we propose that enhancers operating at reduced activities (‘weak enhancers’) can function in an additive manner due to relatively infrequent interactions with their target promoters. In contrast, ‘strong’ enhancers might function sub-additively due to competition for the promoter (Figure 4E). For hb, this switch between competitive and additive behavior occurs as the levels of Bicoid activator diminish in central regions where the posterior border of the anterior Hb domain is formed. Similarly, stress might reduce the performance of the sna enhancers to foster additive behavior under unfavorable conditions such as increases in temperature (Perry et al., 2010). Our study highlights the complexity of multiple enhancers in the regulation of gene expression. They need not function in a simple additive manner, and consequently, their value may be revealed only when their activities are compromised.

Materials and methods

Cloning and recombineering

Request a detailed protocol

In brief, BAC clones that map to the region of interest were identified from end-sequenced BAC libraries which can be viewed on a browser at http://pacmanfly.org, and ordered from BacPac Resources (http://bacpac.chori.org/) (Venken et al., 2009). These BACs arrive already cloned into a vector containing an attB sequence for targeted integration, mini-white cassette, chloramphenicol resistance and are in the inducible copy number strain EPI300, Epicentre Biotechnologies (Madison, WI). The following CHORI BACs were used as a starting point: sna BAC (CH322-18I14-1), hb BAC (BAC CH322-55J23), kni BAC (CH322-21P08).

BACs requiring modification were first transformed into the recombineering strain SW102, which was obtained from NCI-Frederick Biological Resources Branch. Cultures containing specific BACs were grown overnight and recombination functions were induced as described (Perry et al., 2010). The induced bacteria were electroporated with targeting constructs that were prepared previously by PCR amplification. Targeting constructs were made using a pair of 90 base pair long oligonucleotides. These contained 25 base pairs specific to the region being amplified that was to be swapped into the BAC, and an additional 65 base pairs of sequence homologous to the target BAC flanking the region to be replaced. The homologous regions, or ‘homology arms’, target the amplified sequence to the region of interest for recombination. After electroporation and a 1 hr recovery period in 2XYT broth, bacteria were plated in a dilution series on LB plates with the appropriate antibiotic for overnight incubation at 30°C. Individual resulting colonies were screened by PCR for appropriate recombination at both homology arm locations. Confirmed positive recombinant colonies were transformed back into EPI300 cells (Epicentre Biotechnologies) and reconfirmed by antibiotic marker selection and PCR; PCR products were sequenced for final confirmation.

Oligonucleotides for amplification to make homology arm constructs (90 base pairs in length) were from Integrated DNA Technologies; shorter primers for colony screening PCR were from ELIM Biopharmaceuticals. Restriction enzymes were from New England Biopharmaceuticals. Qiagen products were used to isolate plasmid DNAs, gel-purify DNA fragments, and purify PCR products. Qiagen taq polymerase was used in colony PCR screening; Invitrogen Platinum pfx was used to amplify targeting constructs.

The first step in the modification was to replace the endogenous coding sequence of sna, hb, and kni genes with that of the yellow-kanamycin reporter gene. The yellow-kanamycin fragment was swapped into the place of the endogenous gene at the ATG start codon at the 5′ end, leaving the 5′ UTR intact. The endogenous 3′ UTR was also left fully intact. In most cases, the different enhancers were replaced with an ampicillin resistance cassette which was PCR amplified from pBluescript. In the case of kni, one of the enhancers is in the intron of the transcribed region and so we replaced enhancer with a fragment of lambda phage DNA using galK positive and negative selection. The next step required the insertion the MS2 stem loop sequences. Copies of the MS2 stem loops were extracted from plasmid pCR4-24XMS2SL-stable (Addgene 31865) and were PCR amplified with primers with appropriate homology sequences.

BAC preparation for microinjection and phiC31-mediated integration

Request a detailed protocol

BACs were induced to high copy number using Epicentre BAC autoinduction solution, according to supplier's instructions, and grown overnight for 16–18 hr at 37°C. DNA was prepared for micro-injection using the Invitrogen PureLink HiPure miniprep kit by following manufacturer instructions with described modifications for BACs and cosmids. DNA was diluted to a final concentration of ∼300–400 ng/µl and 1× injection buffer. At least 200 embryos were injected per construct by BestGene Inc. (Chino Hills, CA). The transgenes were integrated into the following landing sites: BDSC 9723, BDSC 9750, and BDSC 24749. Hb lacking shadow and kni lacking primary were integrated into 9750 and 24,749, respectively, while all other transgenes were integrated into 9723.

Live imaging sample preparation and data acquisition

Request a detailed protocol

Female virgins of line yw; Histone-RFP;MCP-NoNLS-GFP (Garcia et al., 2013) were crossed with males of each reporter line. Collected embryos were dechorinated using bleach and mounted between a semipermeable membrane (Biofolie, In Vitro Systems & Services) and a coverslip (1.5, 18 mm × 18 mm) and embedded in Halocarbon 27 oil (Sigma). The flattening of the embryos makes it possible to image a larger number of nuclei in the same focal plane without causing significant changes in early development processes (Di Talia and Wieschaus, 2012).

Embryos were either imaged using a custom-built two-photon microscope (Liu et al., 2013) and a Zeiss LSM 780 confocal microscope. Imaging conditions on the two-photon microscope were as described in Garcia et al. (2013). The average laser power at the specimen was 10 mW, the pixel size was set to 220 nm and a single image consisted of 512 × 256 pixels. At each time point, a stack of 10 images separated by 1 µm was acquired resulting in a final time resolution of 37 s. Confocal imaging was performed using a Plan-Apochromat 40×/1.4NA oil immersion objective. The MCP-GFP and Histone-RFP were excited with a laser wavelength of 488 nm and 561 nm, respectively. Fluorescence was detected with two separate photomultiplier tubes using the Zeiss QUASAR detection unit (gallium-arsenide-phosphide photomultiplier was used for the GFP signal while the conventional detector was used for the RFP). Pixel size is 198 nm and images were captured at 512 × 512 pixel resolution with the pinhole set to a diameter of 116 µm. At each time point, a stack of 22 images separated by 0.5 µm was captured, spanning the nuclear layer. The final time resolution is 32 s.

Live imaging data analysis

Request a detailed protocol

Analysis was performed as described (Garcia et al., 2013) and full code can be downloaded from https://github.com/PrincetonUniversity/FlyRNAQuant. Histone-RFP slices were maximum projected for each time point. Nuclei were segmented using an object detection approach based on the Laplacian of Gaussian filter kernel. The segmented nuclei were then segmented and tracked over multiple nuclear cycles. Spots are detected in 3D and assigned to their respectively closest nucleus. When multiple spots are detected in the vicinity of a nucleus only the brightest one is kept. Spot intensity determination necessitates an estimate of the local fluorescent background for each particle. A 2D Gaussian fit to the peak plane of each particle column determines an offset, which is used as background estimator. The intensity is calculated by integrating the particle fluorescence over a circle with a radius of 6 pixels and then subtracting the estimated background. The imaging error is dominated by the error made in the fluorescent background estimation (Garcia et al., 2013).

It is possible to measure the average fluorescence per polymerase molecule for the hunchback enhancer > MS2 transgene with 24 MS2 repeats (Garcia et al., 2013). The quantitative imaging for the BAC transgenes was conducted under the exact same imaging conditions on the same microscope. The BAC transgenes also possess 24 MS2 repeats. However, the specific sequence of the stem loops is slightly different as these repeats have been further optimized to facilitate molecular biology work with them (Hocine et al., 2013). Assuming that the MS2 sites are similarly saturated with MCP::GFP protein in both cases we can then use the average fluorescence per polymerase molecule calculated for the hunchback>MS2 transgene to calibrate the BAC fluorescent traces in terms of the absolute number of transcribing polymerases per fluorescent spot.

Mathematical modeling

Request a detailed protocol

We propose a general scheme for enhancer promoter interactions which makes it possible to model the effect of having multiple enhancers activating a single promoter. Here, the enhancer and promoter engage and disengage with one another with characteristic rate constants kon and koff, respectively (see Figure 4A). The ratio between kon and koff determines the strength of the promoter-enhancer interaction. The system can be found in two states. First, the promoter can be unoccupied. We denote the occupancy of this state with [Enhancer]. Second, the promoter and enhancer can be engaged with an occupancy [Enhancer·Promoter]. Following the reaction scheme shown in Figure 4A, the temporal evolution of the occupancy of the state where the enhancer and promoter are engaged is given by

(1) d[Enhancer ·Promoter]dt=kon[Promoter] koff[Enhancer ·Promoter].

While the enhancer is engaged with the promoter it is capable of producing mRNA at a rate r, which we will call the transcriptional efficiency. Hence the rate of mRNA production is given by

(2) dmRNAdt=r[Enhancer ·Promoter].

Since the promoter can only be in two states, engaged or not engaged by the enhancer, the occupancy of these two states is constrained by

(3) [Enhancer ·Promoter]+[Promoter]=1.

Assuming steady state in the temporal evolution of the state occupancies results in a rate of mRNA production of

(4) dmRNAdt=konkon+koff*r.

In Figure 4B we plot this rate as a function of the interaction strength for different values of enhancer efficiency.

When considering two enhancers that can interact independently with the promoter (Figure 4C) we need to calculate the occupancy of the state accounting for the promoter interacting with enhancer A, [EnhancerA·Promoter], and with enhancer B, [EnhancerB·Promoter]. The temporal evolution of these occupancies is given by

(5) d[EnhancerA·Promoter]dt=konA[Promoter]koffA[Enhancer A·Promoter],

and

(6) d[EnhancerB·Promoter]dt=konB[Promoter] koffB[Enhancer B·Promoter].

In this case the rate of mRNA production is given by

(7) dmRNAdt=rA[Enhancer A·Promoter]+rB[Enhancer B·Promoter],

where the different promoter occupancy states are constrained by

(8) [Enhancer A·Promoter]+[Enhancer B·Promoter]+[Promoter]=1.

In analogy to the single-enhancer case we now assume steady state for the temporal evolution of the occupancies described by Equations 5, 6 and use the constraint given by Equation 7 to solve for [EnhancerA·Promoter] and [EnhancerB·Promoter]. Finally we replace these results into Equation 8 leading to

(9) dmRNAdt=rAkonAkoffB+rBkonBkoffAkonBkoffA+konAkoffB+koffAkoffB.

Using this equation we determine the rate of mRNA production as a function of the strength (i.e., r) and the efficiency (i.e., kon/koff) of enhancers A and B. These calculations are used to generate the plots shown in Figure 4D,E.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30

Decision letter

  1. Robb Krumlauf
    Reviewing Editor; Stowers Institute for Medical Research, United States

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

Thank you for submitting your work entitled “Enhancer synergy and interference in living Drosophila embryos” for peer review at eLife. Your submission has been favorably evaluated by Jim Kadonaga (Senior editor), Robb Krumlauf (Reviewing editor), and three reviewers.

The reviewers have discussed the reviews with one another and the Reviewing editor has drafted this decision to help you prepare a revised submission.

The general consensus of the reviewers is that this is a strong paper, detailing high quality work with wide general relevance and import to current questions in the field of gene regulation. The main conclusion, that enhancers can work together in a variety of ways and that these differences are linked to important factors such as enhancer strength, is a significant one that will likely have an impact on the field. The state-of-the-art technique used here to visualize transcription live in vivo, represents a major advance for the quantitative study of enhancer function and gene expression.

No major experimental revisions are required but comments and concerns were raised by all three reviewers primarily in two areas: (1) relevant considerations that are not discussed in the text, and (2) limitations of the very simple theoretical model.

There is real enthusiasm for seeing a revised manuscript that incorporates or rebuts the issues and specific points noted below. The collective view is that this will significantly enhance the general interest of the manuscript.

The points to consider in revision are:

1) The authors' interpretations depend on the assumption that 100% of the (relevant) regulatory activity at each gene locus is contained within the identified enhancers, and that flanking sequences left in the loci after deletion of an enhancer have no impact on gene expression. However, it is possible that sequences that flank enhancer A could interact positively with enhancer B when A is deleted, resulting in a “super-additive” outcome. Minimal versions of enhancers are often weaker than longer versions, so it is possible that not all of the 'primary' or 'shadow' enhancer activity has been removed by the deletions. I think that the authors' assumption is a reasonable one, but it might be safer to acknowledge that when they refer to a transgene containing the primary enhancer “alone” (Results, fifth paragraph), the deletion transgene actually contains much more non-coding sequence than that.

2) The idea that enhancer additivity is a function of enhancer strength is a very interesting and appealing one, and it is central to the paper. Using the existing data, is there a way to plot those two measurements against one another to illustrate this idea more directly?

3) I understand the theoretical model (Figure 4) as it is described in the text, but the authors make a point of saying that it explains enhancer interference, where “the combined activity of both enhancers is less than the sum of both enhancers” (at the end of Results), and I don't see any evidence of enhancer interference in the graph. Shouldn't the red line (2 enhancers) dip below the green line (1 enhancer), and shouldn't the blue line reach an expression ratio lower than 1?

4) On a closely related note, it isn't obvious to me how the model described here could ever drive the expression ratio below 1 by increasing enhancer-promoter engagement time, unless one enhancer were better at engaging the promoter but also worse at activating transcription when engaged. Wouldn't another state, such as non-productive enhancer-promoter interaction, need to be added to the model to achieve interference?

5) In 2011, Perry et al. showed that the knirps intronic enhancer drives an inappropriately wide expression pattern, one that is “normalized” when coupled with the proximal enhancer. This result suggests that the latter element has long-range repression type effects (or acts on the basal promoter in nuclei normally not expressing knirps to silence would-be inputs from the intron element). This data indicates that the authors' modeling of the enhancers as small activating units does not really conform to what we know about these elements. The authors do not discuss this prior finding, but presumably they have seen similar effects in their two photon system? The authors should test how a model involving such negative regulatory inputs would change their predictions. Admittedly this model would be more complex than the simple on/off model presented in this version of the manuscript, but we already know that this simpler presentation is insufficient.

6) In 2009, MacArthur and colleagues measured transcription factor occupancy in the embryo. In that study, these researchers found that the Snail transcriptional repressor directly binds to the Snail regulatory elements that are tested here. The expression of Snail would thus induce some sort of negative feedback, if the binding measured in that study is indeed active. In light of the findings from this manuscript, where the primary enhancer by itself is weaker, and the shadow enhancer by itself is able to drive gene expression at about wild-type levels, it appears that Snail self-inhibition may be compromising promoter activity to reduce the overall output. Previous studies of the related short-range repressor Knirps have shown that this class of repressor may induce local, heterochromatized-like chromatin status (Li and Arnosti, 2011), thus if Snail is doing the same thing, the activity of this proximal element should be considered as a mix of active “silencer”-like activity with “enhancer”-like action in the classic sense. A similar local silencing action by parts of the knirps proximal element would explain the Perry results.

7) The authors should address whether differences in elongation rates (which may be unlinked to promoter firing) would affect their estimations of overall enhancer activity. The genes tested here are very similar, but not identical – the lambda DNA filler for knirps or the unique 5' and 3' UTRs may have differential elongation potential. The overall conclusions about enhancer (non) additivity would still hold in any event.

8) Authors state that hb enhancers do not function in additive fashion in anterior regions of the embryo (are subadditive). It is true, but by looking at the data in Figure 2A–B and 3A, I would say that the transcriptional output of the shadow enhancer deletion overlaps pretty well with the wt, suggesting that hb shadow enhancer does not provide much of a contribution to transcription, at least in the anterior part of the embryo, which perhaps could be stated explicitly.

9) Clearly the theoretical model does not account for the superadditive enhancer behavior, as seen for kni enhancers (which authors acknowledge in the discussion). Nonetheless, existence of the superadditive behavior indicates that a central assumption of the model – that a single enhancer engages the promoter at the time – is probably incorrect at least at a subset of loci. Could authors incorporate an additional state to their model accounting for the possibility of multiple enhancer interactions? How would that affect predictions of their model? What do authors think may determine competitive vs cooperative enhancer engagement with the promoter?

https://doi.org/10.7554/eLife.07956.012

Author response

1) The authors' interpretations depend on the assumption that 100% of the (relevant) regulatory activity at each gene locus is contained within the identified enhancers, and that flanking sequences left in the loci after deletion of an enhancer have no impact on gene expression. However, it is possible that sequences that flank enhancer A could interact positively with enhancer B when A is deleted, resulting in asuper-additiveoutcome. Minimal versions of enhancers are often weaker than longer versions, so it is possible that not all of the 'primary' or 'shadow' enhancer activity has been removed by the deletions. I think that the authors' assumption is a reasonable one, but it might be safer to acknowledge that when they refer to a transgene containing the primary enhanceralone(Results, fifth paragraph), the deletion transgene actually contains much more non-coding sequence than that.

Concerns were raised about the persistence of critical flanking sequences upon deletion of core enhancers. We have added a new paragraph in the Results to address this important issue. The enhancer deletions are actually sequence substitutions, as we now describe in the revised Results. Every effort was made to replace all of the critical binding sites identified in previous ChIP-Seq assays. We nonetheless concede the possibility that critical elements are left intact. However, new data are included to show that any such sequences are not sufficient to mediate activation. For example, “deletions” of both the 5’ and intronic enhancers within the kni locus completely abolish expression of the corresponding transgene in the presumptive abdomen (see Figure 1–figure supplement 1).

2) The idea that enhancer additivity is a function of enhancer strength is a very interesting and appealing one, and it is central to the paper. Using the existing data, is there a way to plot those two measurements against one another to illustrate this idea more directly?

We were asked to revise Figure 5 to provide a clearer representation of the anti-correlation between enhancer strength and additivity. The revised Figure 5 shows how the combined activity of the two enhancers varies as a function of the sum of their individual activities. For hb, when the combined activities are low (left hand side of the x axis) the combined activity is additive. However as the strength of the individual activities increases (right hand side of the x axis) the combined activity becomes sub-additive.

3) I understand the theoretical model (Figure 4) as it is described in the text, but the authors make a point of saying that it explains enhancer interference, wherethe combined activity of both enhancers is less than the sum of both enhancers(at the end of Results), and I don't see any evidence of enhancer interference in the graph. Shouldn't the red line (2 enhancers) dip below the green line (1 enhancer), and shouldn't the blue line reach an expression ratio lower than 1?

We are asked to provide a clearer depiction of enhancer interference. Yes, we agree that the original version of Figure 4 was unclear. We now present a revised version that more clearly illustrates enhancer interference. Moreover, Figure 5 was also revised to show how two enhancers are less than the sum of their individual contributions in the regime of high rates of mRNA production.

4) On a closely related note, it isn't obvious to me how the model described here could ever drive the expression ratio below 1 by increasing enhancer-promoter engagement time, unless one enhancer were better at engaging the promoter but also worse at activating transcription when engaged. Wouldn't another state, such as non-productive enhancer-promoter interaction, need to be added to the model to achieve interference?

We were asked to consider the possibility of non-productive or weakly productive enhancer-promoter interactions to account for the interference phenomenon seen in the snail locus. Yes, this is an excellent point. The text has been revised to include this possibility. Moreover, we have plotted how mRNAs levels change with enhancer-promoter engagement times and different enhancer strengths. When one enhancer is weak and the other strong there is decreased mRNA synthesis as the frequency of engagement by the weak enhancer goes up.

We agree with the referee that this is probably the case for snail, since the shadow enhancer is stronger than the proximal enhancer. We state this clearly in the revised text.

5) In 2011, Perry et al. showed that the knirps intronic enhancer drives an inappropriately wide expression pattern, one that isnormalizedwhen coupled with the proximal enhancer. This result suggests that the latter element has long-range repression type effects (or acts on the basal promoter in nuclei normally not expressing knirps to silence would-be inputs from the intron element). This data indicates that the authors' modeling of the enhancers as small activating units does not really conform to what we know about these elements. The authors do not discuss this prior finding, but presumably they have seen similar effects in their two photon system? The authors should test how a model involving such negative regulatory inputs would change their predictions. Admittedly this model would be more complex than the simple on/off model presented in this version of the manuscript, but we already know that this simpler presentation is insufficient.

We were asked to address the occurrence of long-range repression within the knirps locus. We have revised the text to indicate that this type of repression has been previously documented. We also emphasize that the enhancer analysis is focused on central regions of the knirps expression pattern, outside the limits of the repressive interactions that help establish the sharp borders of the expression “stripe”.

6) In 2009, MacArthur and colleagues measured transcription factor occupancy in the embryo. In that study, these researchers found that the Snail transcriptional repressor directly binds to the Snail regulatory elements that are tested here. The expression of Snail would thus induce some sort of negative feedback, if the binding measured in that study is indeed active. In light of the findings from this manuscript, where the primary enhancer by itself is weaker, and the shadow enhancer by itself is able to drive gene expression at about wild-type levels, it appears that Snail self-inhibition may be compromising promoter activity to reduce the overall output. Previous studies of the related short-range repressor Knirps have shown that this class of repressor may induce local, heterochromatized-like chromatin status (Li and Arnosti, 2011), thus if Snail is doing the same thing, the activity of this proximal element should be considered as a mix of activesilencer-like activity withenhancer-like action in the classic sense. A similar local silencing action by parts of the knirps proximal element would explain the Perry results.

We were asked to consider the possibility that Snail repression elements located within or near the proximal enhancer might attenuate gene expression. This is a valid point and we have revised the text to consider this possibility and now include the MacArthur et al. 2009 citation.

7) The authors should address whether differences in elongation rates (which may be unlinked to promoter firing) would affect their estimations of overall enhancer activity. The genes tested here are very similar, but not identical – the lambda DNA filler for knirps or the unique 5' and 3' UTRs may have differential elongation potential. The overall conclusions about enhancer (non) additivity would still hold in any event.

We were asked whether different Pol II elongation rates might complicate our interpretations. The text has been revised to address this possibility, and we also cite recent and unpublished evidence that many different enhancer-promoter combinations produce the same rate of elongation, approximately 30 nt/s.

8) Authors state that hb enhancers do not function in additive fashion in anterior regions of the embryo (are subadditive). It is true, but by looking at the data in Figure 2A–B and 3A, I would say that the transcriptional output of the shadow enhancer deletion overlaps pretty well with the wt, suggesting that hb shadow enhancer does not provide much of a contribution to transcription, at least in the anterior part of the embryo, which perhaps could be stated explicitly.

We were asked to clarify that removal of the hb shadow enhancer has no effect on the levels of transcription in anterior regions of the embryo. This is now done in the revised manuscript.

9) Clearly the theoretical model does not account for the superadditive enhancer behavior, as seen for kni enhancers (which authors acknowledge in the discussion). Nonetheless, existence of the superadditive behavior indicates that a central assumption of the model – that a single enhancer engages the promoter at the time – is probably incorrect at least at a subset of loci. Could authors incorporate an additional state to their model accounting for the possibility of multiple enhancer interactions? How would that affect predictions of their model? What do authors think may determine competitive vs cooperative enhancer engagement with the promoter?

It was noted that our model fails to explain the super-additive behavior of the knirps enhancers during early stages of nc 14. We really do not have a good explanation for super-additivity. During the later stages of nc 14, the two enhancers function in a simple additive fashion, and this behavior is captured by our model. At earlier stages it would appear that the interaction of one enhancer facilitates the interaction of the other enhancer. This possibility is clearly stated in the Discussion. However, we do not know enough about this type of enhancer cooperativity to formulate a reasonable model. Any such effort would be wildly speculative.

https://doi.org/10.7554/eLife.07956.013

Article and author information

Author details

  1. Jacques P Bothma

    Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States
    Contribution
    JPB, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents
    Contributed equally with
    Hernan G Garcia
    Competing interests
    The authors declare that no competing interests exist.
  2. Hernan G Garcia

    Department of Physics, Princeton University, Princeton, United States
    Present address
    Department of Molecular and Cell Biology, Department of Physics, and Biophysics Graduate Group, University of California, Berkeley, Berkeley, United States
    Contribution
    HGG, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article, Contributed unpublished essential data or reagents
    Contributed equally with
    Jacques P Bothma
    Competing interests
    The authors declare that no competing interests exist.
  3. Samuel Ng

    Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States
    Contribution
    SN, Acquisition of data, Analysis and interpretation of data, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  4. Michael W Perry

    Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States
    Present address
    Department of Biology, New York University, New York, United States
    Contribution
    MWP, Conception and design, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  5. Thomas Gregor

    1. Department of Physics, Princeton University, Princeton, United States
    2. Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, United States
    Contribution
    TG, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    tg2@princeton.edu
    Competing interests
    The authors declare that no competing interests exist.
  6. Michael Levine

    Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, United States
    Present address
    Department of Molecular Biology, and Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, United States
    Contribution
    ML, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    mlevine@berkeley.edu
    Competing interests
    The authors declare that no competing interests exist.

Funding

National Institutes of Health (NIH) (GM34431)

  • Michael Levine

National Institutes of Health (NIH) (GM071508)

  • Thomas Gregor

National Institutes of Health (NIH) (GM097275)

  • Thomas Gregor

Burroughs Wellcome Fund (101880)

  • Hernan G Garcia

Physics Department, Princeton University (Dicke Fellowship)

  • Hernan G Garcia

Searle Scholar Award (10-SSP-274)

  • Thomas Gregor

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Y Yamazaki and E Esposito for help with molecular biology and A Sanchez for discussions on theoretical modeling.

Reviewing Editor

  1. Robb Krumlauf, Stowers Institute for Medical Research, United States

Publication history

  1. Received: April 7, 2015
  2. Accepted: July 15, 2015
  3. Version of Record published: August 12, 2015 (version 1)

Copyright

© 2015, Bothma et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 5,154
    Page views
  • 1,042
    Downloads
  • 65
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Arun Sharma et al.
    Research Article Updated

    Damaging GATA6 variants cause cardiac outflow tract defects, sometimes with pancreatic and diaphragmic malformations. To define molecular mechanisms for these diverse developmental defects, we studied transcriptional and epigenetic responses to GATA6 loss of function (LoF) and missense variants during cardiomyocyte differentiation of isogenic human induced pluripotent stem cells. We show that GATA6 is a pioneer factor in cardiac development, regulating SMYD1 that activates HAND2, and KDR that with HAND2 orchestrates outflow tract formation. LoF variants perturbed cardiac genes and also endoderm lineage genes that direct PDX1 expression and pancreatic development. Remarkably, an exon 4 GATA6 missense variant, highly associated with extra-cardiac malformations, caused ectopic pioneer activities, profoundly diminishing GATA4, FOXA1/2, and PDX1 expression and increasing normal retinoic acid signaling that promotes diaphragm development. These aberrant epigenetic and transcriptional signatures illuminate the molecular mechanisms for cardiovascular malformations, pancreas and diaphragm dysgenesis that arise in patients with distinct GATA6 variants.

    1. Cell Biology
    2. Developmental Biology
    Delia Ricolo, Sofia J Araujo
    Research Article

    Subcellular lumen formation by single-cells involves complex cytoskeletal remodelling. We have previously shown that centrosomes are key players in the initiation of subcellular lumen formation in Drosophila melanogaster, but not much is known on the what leads to the growth of these subcellular luminal branches or makes them progress through a particular trajectory within the cytoplasm. Here, we have identified that the spectraplakin Short-stop (Shot) promotes the crosstalk between MTs and actin, which leads to the extension and guidance of the subcellular lumen within the Terminal Cell (TC) cytoplasm. Shot is enriched in cells undergoing the initial steps of subcellular branching as a direct response to FGF signalling. An excess of Shot induces ectopic acentrosomal luminal branching points in the embryonic and larval tracheal TC leading to cells with extra subcellular lumina. These data provide the first evidence for a role for spectraplakins in single-cell lumen formation and branching.