Regeneration and Developmental Enhancers Are Differentially Compatible with Minimal Promoters

Abstract

Enhancers and promoters are cis-regulatory elements that control gene expression. Enhancers are activated in a cell type-, tissue-, and condition-specific manner to stimulate promoter function and transcription. Zebrafish have emerged as a powerful animal model for examining the activities of enhancers derived from various species through transgenic enhancer assays, in which an enhancer is coupled with a minimal promoter. However, the efficiency of minimal promoters and their compatibility with multiple developmental and regeneration enhancers have not been systematically tested in zebrafish. Thus, we assessed the efficiency of six minimal promoters and comprehensively interrogated the compatibility of the promoters with developmental and regeneration enhancers. We found that the fos minimal promoter and Drosophila synthetic core promoter (DSCP) yielded high rates of leaky expression that may complicate the interpretation of enhancer assays. Notably, the adenovirus E1b promoter, the zebrafish lepb 0.8-kb (P0.8) and lepb 2-kb (P2) promoters, and a new zebrafish synthetic promoter (ZSP) that combines elements of the E1b and P0.8 promoters drove little or no ectopic expression, making them suitable for transgenic assays. We also found significant differences in compatibility among specific combinations of promoters and enhancers, indicating the importance of promoters as key regulatory elements determining the specificity of gene expression. Our study provides guidelines for transgenic enhancer assays in zebrafish to aid in the discovery of functional enhancers regulating development and regeneration.

Introduction

Enhancers are noncoding cis-regulatory elements that serve as vital regulators of spatiotemporal gene expression (Panigrahi and O’Malley, 2021; Shlyueva et al., 2014; Tippens et al., 2018). Enhancers selectively activate gene expression in a highly regulated manner in various conditions and developmental stages (Jindal and Farley, 2021; Rickels and Shilatifard, 2018). Thus, identifying and characterizing enhancers in particular tissues and contexts will expand our understanding of the regulatory mechanisms governing essential biological phenomena, such as development and regeneration.

Enhancer candidates can be identified through genome-wide assays of accessible chromatin regions via assay for transposase-accessible chromatin using sequencing (ATAC-seq) or through assaying specific enhancer-associated histone modifications, such as histone H3 lysine 27 acetylation (H3K27ac), via chromatin immunoprecipitation followed by sequencing (ChIP-seq) (Barski et al., 2007; Bonn et al., 2012; Buenrostro et al., 2013; Shlyueva et al., 2014). While the presence of open chromatin and specific histone marks is suggestive of active enhancers, the in vivo activity of enhancer candidates must be verified empirically (Cao et al., 2022; Goldman et al., 2017). A widely used in vivo method is the transgenic reporter assay, in which an enhancer is paired with a minimal promoter and a reporter gene and then inserted into the genome of a model organism, with reporter gene expression providing a readout of the enhancer’s activity (Begeman et al., 2020; Dickel et al., 2016; Dobrzycki et al., 2020; Goldman et al., 2017; Hewitt et al., 2017; Kang et al., 2016; Kvon, 2015; Osterwalder et al., 2018; Wong et al., 2020). Because promoters are also important regulators of gene expression, a minimal promoter with little or no background expression is required to obtain conclusive results from transgenic enhancer assays.

Zebrafish (Danio rerio) have emerged as a powerful system for examining enhancers, owing to their genetic tractability, transparency during early developmental stages, and high fecundity (Kang et al., 2016; Yuan et al., 2018). Enhancers derived from other species, such as humans, mice, and sponges, can also direct conserved expression patterns in zebrafish, highlighting the usefulness of zebrafish for cross-species enhancer investigation (Heller et al., 2022; Perez-Rico et al., 2017; Wong et al., 2020; Yuan et al., 2018). Recent genome-wide analyses of zebrafish have rapidly expanded the list of enhancer candidates that may contribute to development and regeneration (Cao et al., 2022; Goldman et al., 2017; Kang et al., 2016; Quillien et al., 2017). However, published enhancer studies in zebrafish have employed a variety of minimal promoters (Supplementary Table S1), and no promoter has yet been accepted as a common standard by the zebrafish community. Moreover, the leakiness of these minimal promoters and their compatibility with distinct classes of enhancers have not been systematically determined.

In this study, we assessed the efficiency of six minimal promoters and comprehensively interrogated their compatibility with developmental and regeneration enhancers. We found that two minimal promoters yielded high rates of leaky expression that may complicate the interpretation of enhancer assays. Notably, four promoters drove little or no ectopic expression, making them suitable for transgenic assays. We also demonstrated significant differences in compatibility among specific combinations of promoters and enhancers. Our study provides guidelines for transgenic enhancer assays in zebrafish to aid in the discovery of functional enhancers regulating development and regeneration.

MATERIALS AND METHODS

Zebrafish maintenance and transgenesis

Embryos were collected by mating wild-type or transgenic male and female zebrafish from the Ekkwill (EK) strain. Water temperature of adult fish was maintained at 26 °C. F₀ transgenic larvae were generated using Tol2 transgenesis (Kawakami et al., 2004). One-cell-stage wild-type or transgenic cmlc2:mCherry-NTR embryos were injected with 1 pL of a mixture of 30 ng/μL plasmid DNA, 40 ng/μL Tol2 mRNA, and 0.1% phenol red. Injections were performed using a WPI PV820 microinjector. Embryos and larvae were maintained at 28 °C in egg water containing 300 mg/L sea salt, 75 mg/L calcium sulfate, 37.5 mg/L sodium bicarbonate, and 0.0001% methylene blue. Larvae were screened for mCherry-positive lenses at 3 dpf.

Plasmid generation

Plasmids were generated through Gibson assembly or restriction enzyme subcloning. Enhancer, promoter, EGFP, and SV40 polyadenylation signal sequences were assembled on the forward strand of each plasmid. To enable identification of larvae carrying transgenic constructs, a reporter cassette containing mCherry driven by the eye lens-specific alpha crystallin (cryaa) promoter was placed downstream in the reverse complement direction. The construct is flanked by I-SceI meganuclease and Tol2 transposon sites, enabling transgenesis through either method (Kawakami and Shima, 1999; Thermes et al., 2002). E1b, P0.8, P2, LEN, and EGFP sequences were subcloned from previously generated plasmids (Kang et al., 2016). DSCP was subcloned from the pBPGUw plasmid (Addgene plasmid #17575; http://n2t.net/addgene:17575; RRID:Addgene_17575) (Pfeiffer et al., 2008). 103runx1EN and gata2a-i4 enhancer sequences were amplified from EK zebrafish genomic DNA. Primers used for subcloning are listed in Supplementary Table S2.

Injury models and imaging

To injure hearts with larvae, cmlc2:mCherry-NTR^pd71Tg (Chen et al., 2013) larvae were incubated in 10 mM metronidazole (Mtz) for 24 hours from 3 to 4 dpf to induce CM ablation. To prevent light inactivation of Mtz, larvae were kept in the dark during incubation. Larvae were transferred to fresh egg water following Mtz treatment for recovery. For tail fin fold injury, 4-dpf larvae were first anesthetized in 0.02% tricaine methanesulfonate. The tail was amputated using a scalpel at the ventral gap in melanophores at the distal end of the tail. Larvae were transferred to egg water following amputation for recovery. To injure eyes, 4-dpf larvae were anesthetized in 0.02% tricaine methanesulfonate and placed on a 1% agarose pad with the right eye facing upwards. The right eye of each larvae was fully penetrated by a glass capillary needle. The left eye was left uninjured as a control. Following injury, larvae were returned to egg water for recovery. For imaging, anesthetized larvae placed on a 1% agarose pad. Wholemount fluorescent and brightfield images were captured using a Zeiss Axiozoom V16 stereomicroscope. Work with zebrafish was performed in accordance with University of Wisconsin–Madison guidelines.

Quantification and statistical analysis

Statistical significance of transgenic construct expression rates was determined via Pearson’s chi-squared test with post hoc pairwise analysis. Significance levels were set to 0.05. Multiple-comparison correction was performed using the Bonferroni correction. n represents the number of larvae observed. The listed numbers of larvae include only those that expressed the cryaa:mCherry transgenesis marker in their eyes. Developmentally abnormal larvae were excluded. All larvae counted as having positive EGFP expression exhibited expression patterns similar to those shown in the representative images, though variability in the strength and extensiveness of EGFP expression was observed in some larvae due to the mosaic nature of randomly inserted F₀ transgene expression.

Results

Minimal promoters drive different rates of leaky expression

We performed a literature search of enhancer assays and selected five minimal promoters to test for efficiency in transgenic enhancer assays in zebrafish: E1b, leptin b (lepb) 0.8 kb (P0.8), lepb 2 kb (P2), fos, and the Drosophila synthetic core promoter (DSCP) (Supplementary Fig. S1, S2; Supplementary Table S3). The E1b promoter, referred to as E1b hereafter, originates from the adenovirus early region 1b gene and has been widely used for validating developmental enhancers in zebrafish (Supplementary Table S1) (Argenton et al., 1996; Bar Yaacov et al., 2019; Birnbaum et al., 2012; Booker et al., 2013; Hirsch et al., 2018; Kim et al., 2014; Laarman et al., 2019; Li et al., 2010; Oksenberg et al., 2014; White, 2001; Yanovsky-Dagan et al., 2015). P0.8 and P2 are the 793-bp and 2045-bp sequences, respectively, immediately upstream of the lepb transcription start site (Kang et al., 2016). lepb shows no detectable developmental expression in the heart or fins but is strongly upregulated in those tissues during regeneration. A lepb-linked regeneration enhancer known as LEN is located 5.2 kb upstream of lepb and is required for regeneration-dependent lepb induction (Kang et al., 2016; Thompson et al., 2020). P0.8 and P2 have been used as basal promoters for identifying and testing regeneration enhancers, including cardiac LEN (cLEN), which is the enhancer fragment of LEN that directs expression in injured hearts (Begeman et al., 2020; Geng et al., 2021; Kang et al., 2016; Karra et al., 2018; Lee et al., 2020). The 100-bp fos promoter, referred to as fos hereafter, is derived from the mouse fos gene and has been used in zebrafish for studying both developmental and regeneration enhancers (Cao et al., 2022; Goldman et al., 2017; Scott and Baier, 2009; Thompson et al., 2020). The 155-bp DSCP is an optimized synthetic core promoter that has been utilized for examining neuronal enhancers and for performing large-scale developmental enhancer screens in Drosophila (Kvon et al., 2014; Pfeiffer et al., 2008).

An important characteristic of a minimal promoter is having limited intrinsic leaky activity, as such activity can hinder the interpretation of the activity of the enhancer being tested. To determine the leakiness of the selected minimal promoters, we generated constructs containing E1b, P0.8, P2, fos, and DSCP upstream of EGFP in the absence of an enhancer. Each construct also included an eye lens-specific mCherry reporter gene that enables the selection of transgenic larvae. We assessed the basal activity of these constructs by microinjecting them into one-cell-stage zebrafish embryos, selecting larvae with mCherry-positive lenses, and examining EGFP expression during larval development (Fig. 1A).

Fig. 1. E1b, P0.8, and P2 drive minimal leaky expression in uninjured tissues.

(A) Experimental design. Transgenic constructs were injected into one-cell-stage wild-type or Tg(cmlc2:mCherry-NTR) embryos, and EGFP expression was observed at 3 dpf. (B) 3-dpf F₀ uninjured hearts, eyes, and fin folds. Arrows indicate EGFP expression. Eyes are demarcated by dashed lines. (C, D) Tables of larvae displaying EGFP expression in each tissue for each promoter alone (C) or when paired with LEN (D). Scale bars: 100 μm.

We found that E1b, P0.8, and P2 yielded low rates of background expression with no detectable EGFP expression in the heart, eye, or fin fold of larvae (n=196, 57, and 88, respectively) (Fig. 1B, C; Supplementary Table S4). In contrast, fos and DSCP yielded high rates of leaky expression (Fig. 1B; Supplementary Table S4). fos drove expression in the heart, eye, and tail fin fold in 10%, 2%, and 78% of larvae, respectively (n=134). Among larvae carrying the DSCP construct, 43%, 3%, and 100% directed EGFP expression in the heart, eye, and tail fin fold, respectively (n=63). These results suggest that E1b, P0.8, and P2 may be preferable to fos and DSCP due to their lower basal activity.

We next determined whether placing an enhancer fragment upstream of the minimal promoters influences their activity in larvae (Fig. 1A). We selected LEN, a well-characterized regeneration enhancer that exhibits no activity during development or in uninjured tissues but drives injury-responsive expression in the heart and fin (Kang et al., 2016). Because LEN activity is regeneration-specific, we expected to observe no detectable expression in uninjured larvae if the promoter exhibited no leaky expression. We observed limited basal activity for LEN-E1b, LEN-P0.8, and LEN-P2, as no larvae expressed EGFP in the heart, eye, or fin fold (n=227, 234, and 402, respectively) (Fig. 1B, D; Supplementary Table S4). In contrast, a substantial number of LEN-fos and LEN-DSCP larvae displayed leaky expression throughout the body (Fig. 1B, D; Supplementary Table S4). A total of 2%, 2%, and 61% of LEN-fos larvae expressed EGFP in the heart, eye, and tail fin fold, respectively (n=85), while 51%, 0%, and 96% of LEN-DSCP larvae exhibited expression in the heart, eye, and tail fin fold, respectively (n=47). The low levels of background expression of E1b, P0.8, and P2 indicate that they can serve as adequate minimal promoters for F₀ transgenic enhancer assays in zebrafish. In contrast, the high rates of leaky activity of fos and DSCP may mask the activity of paired enhancers and thus complicate conclusions drawn from enhancer assays using F₀ mosaic embryos.

Developmental enhancers display preferences for specific promoters

Another important feature of minimal promoters is the ability to drive expression at the locations and timepoints specified by the paired enhancer. Thus, we sought to determine how efficiently E1b, P0.8, and P2 direct expression with developmental enhancers. We placed each promoter downstream of two validated developmental enhancers: 103runx1EN and gata2a-i4 (Fig. 2A). 103runx1EN is located 103 kb upstream of the zebrafish runx1 gene and displays multiple activities depending on the tissues involved and injury status (Goldman et al., 2017). 103runx1EN directs expression in the notochord and CNS of uninjured developing larvae and in the heart and fin in response to injury, indicating that it can function as either a developmental or regeneration enhancer. gata2a-i4 is an evolutionarily conserved enhancer that regulates the developmental expression of gata2a in endothelial cells (Dobrzycki et al., 2020; Gao et al., 2013).

Fig. 2. Developmental enhancer activity is preferentially compatible with E1b.

(A) Experimental design. Transgenic constructs were injected into one-cell-stage wild-type or Tg(cmlc2:mCherry-NTR) embryos, and EGFP expression was observed at 3 dpf. (B) 3-dpf F₀ hearts, eyes, fin folds, notochords, and vasculature in uninjured larvae. Arrows indicate EGFP expression. Eyes are demarcated by dashed lines. (C, D) Tables of larvae displaying EGFP expression in each tissue for each promoter when paired with 103runx1EN (C) or gata2a-i4 (D). Scale bars: 100 μm.

In agreement with previous reports (Dobrzycki et al., 2020; Gao et al., 2013; Goldman et al., 2017), our F₀ transgenic enhancer assays revealed that 103runx1EN drove EGFP expression in the notochord and CNS but not the heart, eye, or fin fold (Fig. 2B, C; Supplementary Table S4). However, the expression rate of each 103runx1EN-promoter construct varied significantly, with notochord expression being detectable in 15% of 103runx1EN-E1b, 1% of 103runx1EN-P0.8, and 6% of 103runx1EN-P2 larvae (n=414, 246, and 166, respectively). The expression rates driven by gata2a-i4 also varied significantly among promoters (Fig. 2B, D; Supplementary Table S4). Expression was observed in the hearts and vasculature of 77% and 95% of gata2a-i4-E1b larvae, respectively (n=168). In contrast, only 2% of gata2a-i4-P0.8 larvae expressed EGFP in hearts or vasculature (n=61), while no detectable cardiac or vasculature expression was observed in gata2a-i4-P2 larvae (n=73). These data suggest that minimal promoters exhibit different degrees of compatibility with developmental enhancers in F₀ transgenic enhancer assays. Among the tested minimal promoters, our results indicate that E1b exhibits the most efficient compatibility with developmental enhancers.

Differential injury-induced expression among enhancer-promoter pairs

Regeneration enhancers are characterized as being quiescent in uninjured tissues but robustly activated upon injury (Begeman and Kang, 2018; Kang et al., 2016). Important features of the minimal promoters used for investigating regeneration enhancers include a lack of intrinsic injury-responsive activity and the ability to mediate injury-inducible activation of paired enhancers. To determine how tightly E1b, P0.8, and P2 are regulated in injured contexts, we examined the activity of each promoter both alone and when paired with the regeneration enhancers LEN and 103runx1EN. Previous studies validated the activities of these two regeneration enhancers in injured hearts and fins (Goldman et al., 2017; Kang et al., 2016). To selectively injure the hearts of larvae, we employed the cardiac myosin light chain 2 (cmlc2):mCherry-nitroreductase (NTR) system, which enables the genetic ablation of cardiomyocytes (CMs) (Fig. 3A). Under the control of the cmlc2 promoter, the bacterial NTR gene is expressed specifically in CMs. When larvae are treated with the innocuous prodrug metronidazole (Mtz), NTR converts Mtz into a cytotoxic molecule, resulting in the ablation of a substantial number of CMs (Curado et al., 2007; Curado et al., 2008; Dickover et al., 2013).

Fig. 3. Regeneration enhancer activity in hearts is preferentially compatible with P0.8 and P2.

(A) Experimental design. Transgenic constructs were injected into one-cell-stage Tg(cmlc2:mCherry-NTR) embryos, larvae were treated with Mtz for 24 hours starting at 3 dpf to ablate cardiomyocytes, and EGFP expression was observed at 6 dpf. (B) 6-dpf F₀ regenerating hearts following Mtz-induced cardiomyocyte ablation. Arrows indicate EGFP expression. (C) Table of larvae displaying cardiac EGFP expression for each promoter alone or when paired with LEN or 103runx1EN. Scale bar: 50 μm.

We observed little or no cardiac EGFP expression in E1b, P0.8, and P2 larvae at 3 days post-treatment (dpt) (2%, n=154; 0%, n=96; and 0%, n=109, respectively), indicating minimal leaky induction of these promoters following injury (Fig. 3B, C; Supplementary Table S4). When paired with cardiac regeneration enhancers, E1b, P0.8, and P2 displayed different degrees of compatibility. LEN drove cardiac expression in 18%, 61%, and 42% of larvae when paired with E1b, P0.8, and P2, respectively (n=74, 61, and 139) (Fig. 3B, C; Supplementary Table S4). 103runx1EN displayed a similar but more biased pattern, driving cardiac EGFP expression in 5%, 48%, and 45% of larvae when paired with E1b, P0.8, and P2, respectively (n=270, 148, and 102) (Fig. 3B, C; Supplementary Table S4). These data suggest preferential compatibility of cardiac regeneration enhancer activity with P0.8 and P2 over E1b.

We next utilized larval tail regeneration assays, which are a model for studying appendage regeneration (Fig. 4A). Amputating the distal tip of the tail removes multiple tissues, including the notochord, blood vessels, epithelia, and mesenchyme (Mateus et al., 2012; Yoshinari et al., 2009). Of note, we previously found that P2, but not P0.8, drove injury-induced expression in the larval tail, although this intrinsic injury-inducible ability of P2 was not observed in amputated adult fins (Kang et al., 2016). Consistent with our previous reports, no P0.8 larvae directed expression in the amputated larval tail (n=55), while 38% of P2 larvae drove expression (n=56) (Fig. 4B, C; Supplementary Table S4). Meanwhile, 4% of E1b larvae exhibited EGFP expression in the injured tail (n=124) (Fig. 4B, C; Supplementary Table S4), suggesting that E1b possesses limited but noticeable injury-responsive activity. This observation suggests that E1b and P2 may contain injury-responsive elements, which agrees with the findings of a previous study that many promoters possess intrinsic enhancer activity (Nguyen et al., 2016). Given that E1b and P2 do not direct injury-dependent expression in adult fins (Supplementary Fig. S3), the injury-responsive elements are likely repressed during maturation.

Fig. 4. Injury-induced activity of enhancer-promoter constructs in amputated fin folds.

(A) Experimental design. Transgenic constructs were injected into one-cell-stage wild-type embryos, fin folds were amputated at 4 dpf, and EGFP expression was observed at 6 dpf. (B) 6-dpf F₀ regenerating fin folds following injury. Arrows indicate EGFP expression. (C) Table of larvae displaying injury-induced fin fold EGFP expression for each promoter alone or when paired with LEN or 103runx1EN. Scale bar: 100 μm.

We next assessed minimal promoter activity in injured fins when paired with regeneration enhancers. For all tested minimal promoters, LEN and 103runx1EN robustly increased the rate of EGFP expression in injured fins, though the efficiency differed among promoters. Injury-induced expression in the tail was observed in 24%, 44%, and 60% of LEN-E1b, LEN-P0.8, and LEN-P2 larvae, respectively (n=46, 45, and 67) (Fig. 4B, C; Supplementary Table S4). 103runx1EN drove injury-induced expression in 31%, 71%, and 86% of larvae when paired with E1b, P0.8, and P2, respectively (n=39, 34, and 118) (Fig. 4B, C; Supplementary Table S4). Synergy between the injury-responsive activities of P2 and the regeneration enhancers LEN and 103runx1EN likely contributes to the extremely high rates of injury-dependent expression observed in injured fins. In both hearts and fins, the efficiency of E1b when paired with the regeneration enhancers was lower than that of P0.8 or P2. Given that P0.8 and P2 originate from a regeneration-specific gene, our studies imply that regeneration enhancers can more effectively direct injury-responsive expression when paired with a minimal promoter derived from a regeneration-inducible gene.

Enhancer-promoter pairs display differential eye injury-induced activation

Our F₀ transgenic assays of LEN and 103runx1EN recapitulated their injury-responsive activities in hearts and fins (Goldman et al., 2017; Kang et al., 2016), indicating the feasibility of our approach for revealing novel enhancer activity. We next applied our transgenic enhancer assays to a new injury model. As the activities of LEN and 103runx1EN in other injured contexts remain unknown, we tested their injury-dependent activities in eyes. To induce eye injury, we passed the tip of a glass needle fully through the right eye of each larva, damaging multiple tissues, including the sclera, retina, and lens (Fig. 5A). We first examined the leakiness of the minimal promoters and then assessed the injury-dependent activity of the enhancers. Our assays performed with E1b, P0.8, and P2 alone demonstrated no directed expression in uninjured or injured eyes (n=75, 55, and 57, respectively), verifying that all three promoters can be used for enhancer testing (Fig. 5B, C; Supplementary Table S4). Next, we examined the activities of LEN and 103runx1EN in injured eyes, as lepb and runx1 are both induced upon eye injury (Kramer et al., 2021; Zhao et al., 2014). LEN drove injury-induced expression in the eye when paired with each tested promoter (21%, n=43 for E1b; 49%, n=41 for P0.8; 40%, n=35 for P2) (Fig. 5B, C; Supplementary Table S4), indicating that LEN functions as a regeneration enhancer in eyes. In contrast, 103runx1EN drove expression only when paired with P0.8 and P2 (0%, n=38 for E1b; 9%, n=33 for P0.8; 20%, n=51 for P2) (Fig. 5B, C; Supplementary Table S4). The reduction or lack of injury-induced expression when regeneration enhancers are paired with E1b suggests that specific promoter-enhancer interactions are necessary for achieving full enhancer activity. Overall, these results demonstrate that our F₀ transgenic assays can be used to identify regeneration enhancer activity in a wide range of regenerative contexts.

Fig. 5. Activity of minimal promoters and regeneration enhancers in injured eyes.

(A) Experimental design. Transgenic constructs were injected into one-cell-stage wild-type embryos, eyes were injured at 4 dpf, and EGFP expression was observed at 6 dpf. (B) 6-dpf F₀ regenerating eyes following injury. Arrows indicate EGFP expression. (C) Table of larvae displaying eye EGFP expression for each promoter alone or when paired with LEN or 103runx1EN. Scale bar: 100 μm.

Downstream elements influence enhancer-promoter compatibility

Our assays revealed differences in the efficiency of E1b and P0.8/P2 depending on enhancer type. E1b favors developmental enhancer activity, while P0.8 and P2 favor regeneration/injury enhancer activity. To identify potential factors contributing to the higher efficiency of E1b when paired with developmental enhancers, we analyzed the sequences within the E1b promoter construct. We found that the E1b promoter construct used in this and previous studies (Bar Yaacov et al., 2019; Birnbaum et al., 2012; Booker et al., 2013; Hirsch et al., 2018; Kim et al., 2014; Laarman et al., 2019; Li et al., 2010; Oksenberg et al., 2014; Yanovsky-Dagan et al., 2015) is composed of the 42-bp E1b minimal promoter and two downstream elements—the 5′ untranslated region (UTR) of the carp beta-actin gene and an intron from the rabbit beta-globin gene (Supplementary Fig. S1A, S2A). These downstream elements provide a transcriptional start site and splicing sites and have been reported to improve transgene activity (Chatterjee et al., 2010; Distel et al., 2009; Horstick et al., 2015; Scheer and Campos-Ortega, 1999), with promoter-proximal splicing sites being known to influence gene transcription (Choi et al., 1991; Furger et al., 2002). Thus, we hypothesized that adding a 5′ UTR and an intron to P0.8 would improve its efficiency when paired with developmental enhancers without reducing the efficiency of regeneration enhancers. To this end, we generated a zebrafish synthetic promoter (ZSP) in which the carp beta-actin 5′ UTR and the rabbit beta-globin intron were added immediately downstream to a fragment of P0.8 (Fig 6A). We first assessed the leakiness of ZSP activity by examining ZSP either alone or when paired with LEN in uninjured developing larvae. ZSP displayed low leakiness, driving expression in the uninjured heart, eye, and fin fold in no more than 1.5% of larvae (n=206 and 134 for ZSP and LEN-ZSP, respectively) (Fig. 6B, C; Supplementary Table S4). These results suggest that ZSP activity is tightly regulated in uninjured contexts, indicating its usefulness for enhancer assays.

Fig. 6. Adding downstream elements to the P0.8 promoter increases developmental enhancer activity.

(A) Structure of the ZSP synthetic promoter. (B-E) 3-dpf F₀ uninjured heart, eye, fin fold, notochord, and vasculature of ZSP (B), LEN-ZSP (C), 103runx1EN-ZSP (D), and gata2a-i4-ZSP (E) larvae. Arrows indicate EGFP expression. Eyes are demarcated by dashed lines. The percentage of EGFP⁺ larvae were compared between larvae harboring each ZSP construct and each corresponding P0.8 construct. (F-H) 6-dpf F₀ regenerating hearts, eyes, and fin folds following injury in ZSP (F), LEN-ZSP (G), and 103runx1EN-ZSP (H) larvae. (I) Table of larvae displaying EGFP expression in each tissue for ZSP alone or when paired with LEN or 103runx1EN. Scale bars: 100 μm. ****, p < 0.0001.

We next measured the efficiency of ZSP when paired with the developmental enhancers 103runx1EN and gata2a-i4. EGFP was expressed in the notochord in 22% of 103runx1EN-ZSP larvae (n=116), a significantly higher rate than that of 103runx1EN-P0.8 (Fig. 6D; Supplementary Table S4). Nonspecific expression of 103runx1EN-ZSP was undetectable in the uninjured heart and eye and occurred in less than 1% of fin folds. EGFP expression was also significantly higher in the hearts and vasculature of gata2a-i4-ZSP larvae (35% and 48%, respectively; n=82) than in gata2a-i4-P0.8 larvae (Fig. 6E; Supplementary Table S4). These data suggest that adding the 5′ UTR and intron sequences to P0.8 significantly improves its compatibility with developmental enhancer activity.

To determine the efficiency of ZSP when paired with regeneration enhancers and its intrinsic activity in injured contexts, we examined the expression of ZSP, LEN-ZSP and 103runx1EN-ZSP in hearts, eyes, and fin folds upon injury. While ZSP alone directed expression in 19% of injured fin folds (half the rate recorded for P2; n=32), it did not drive EGFP in injured hearts and eyes (n=82 and 39, respectively) (Fig. 6F, I; Supplementary Table S4). LEN-ZSP drove injury-induced expression in 54% of hearts, 59% of eyes, and 67% of fin folds (n=61, 32, and 43, respectively), showing a significant increase in fin fold expression relative to LEN-P0.8 (Fig. 6G, I; Supplementary Table S4). The 103runx1EN-ZSP construct also drove significantly higher rates of EGFP induction in injured hearts and fin folds than 103runx1EN-P0.8 (81%, n=43 for hearts; 90%, n=31 for fin folds), and its activity in injured eyes was similar to that of 103runx1EN-P0.8 (13%, n=32) (Fig. 6H, I; Supplementary Table S4). These data suggest that the inclusion of downstream promoter elements can improve the efficiency of promoters when paired with developmental enhancers without decreasing their compatibility with regeneration enhancer activity.

Overall, our study demonstrates that both enhancers and promoters play important roles in governing gene expression and that they interact with each other in complex ways. Thus, the characteristics of each enhancer and promoter should be carefully considered when designing constructs for transgenic assays in zebrafish.

Discussion

Transgenic enhancer assays in zebrafish have been performed using many different minimal promoters (Supplementary Table S1), and it remains unclear how compatibly each promoter performs with various developmental and regeneration enhancers. Here, we assessed the performance of the E1b, P2, P0.8, fos, DSCP, and ZSP promoters in transgenic enhancer assays in zebrafish. Our study showed that the mouse-derived fos promoter and synthetic Drosophila-derived DSCP promoter exhibited high levels of leaky expression, displaying ectopic expression in numerous tissues. E1b, P2, P0.8, and ZSP drove minimal or no ectopic expression, making them more suitable for use in transgenic screens because enhancer activity is not masked by leaky promoter activity. Developmental enhancer activity appears to exhibit better performance when paired with E1b, while injury-responsive enhancer activity works more efficiently with P0.8 and P2, indicating differential enhancer-promoter preference. The synthetic promoter ZSP, consisting of a fragment of P0.8 combined with additional downstream elements, displays a higher efficiency with developmental enhancers than P0.8 while maintaining its compatibility with injury-responsive enhancer activity. These results imply that elements downstream of the promoter, such as UTRs, splicing sites, and introns, are also important cis-regulatory elements affecting transcription.

Based on our findings, we provide guidelines for designing transgenic enhancer assays in zebrafish. First, minimal promoters are crucial regulatory elements that influence gene expression and cannot be considered universal. A previous study in zebrafish investigated combinations of the enhancers and promoters of genes that are active in early development, revealing that the intensity of transgene expression and the level of tissue specificity varied dramatically among different enhancer-promoter combinations (Gehrig et al., 2009). A study in Drosophila also reported enhancer-promoter specificity (Zabidi et al., 2015). In the fly, the promoters of housekeeping and developmental genes markedly differ in their characteristics, including their enrichment for specific transcription factor motifs, compositions of core promoter elements, and bound trans-acting factors. A recent high-throughput study in cultured human cells also found distinct categories of promoters (Bergman et al., 2022). Promoters of ubiquitously expressed genes exhibit strong intrinsic activity and low responsiveness to enhancers, while promoters of genes that are variably expressed across cell-types have weak intrinsic activity and high compatibility with enhancers. Because transcriptional specificity is encoded by both enhancers and promoters, appropriate minimal promoters need to be selected when performing transgenic enhancer assays.

Second, a minimal promoter needs to be tailored to the enhancers being investigated. Our study and previous studies have provided evidence that the most appropriate minimal promoter is the cognate promoter from the flanking gene of the enhancer being tested (Quillien et al., 2017). A zebrafish study revealed that endothelial enhancers showed increased expression levels in endothelial cells when paired with their cognate promoters versus a basal promoter (Quillien et al., 2017). In agreement with this finding, LEN drove higher rates of injury-responsive expression when paired with its cognate promoter (P2, P0.8, and the P0.8-derived ZSP) than with E1b. Given these data, in can be inferred that a cognate promoter of an enhancer’s target gene can provide a more robust and accurate readout of the enhancer’s activity in its native context. However, this is unlikely to be feasible in a high-throughput assay. In such a case, we recommend selecting a promoter from a gene that is a well-characterized marker in contexts in which the tested enhancers are predicted to be active. However, examining whether the promoter’s intrinsic activity directs ectopic expression is a crucial prerequisite. As shown in a high-throughput study, most promoters appear to possess intrinsic enhancer activity (Nguyen et al., 2016). In our studies, we found that P2 and, to a lesser extent, E1b and ZSP could drive injury-induced expression in the absence of an enhancer during the larval stage. Thus, care should be taken to ensure that the promoter does not intrinsically drive expression in the cell types, tissues, or contexts where the enhancer is expected to drive expression.

In summary, our work demonstrates that enhancer-promoter combinations are subject to complex interactions that can drive dramatically different expression patterns during development or regeneration. The proper selection of minimal promoters is important to ensure the accuracy of transgenic assays, and the characteristics and potential compatibilities of each enhancer and promoter should be carefully considered when designing transgenic constructs. We propose that it may be helpful to test multiple minimal promoters or use each enhancer’s cognate promoter to conclusively determine enhancer activity in F₀ transgenic assays.

Limitations of the study

Our transgenic approach utilizes random insertions, so the activity of integrated constructs may be influenced by surrounding regions, known as the position effect. To avoid this, researchers studying mice and Drosophila recently established targeted insertion methods to increase reproducibility and decrease chromosomal effects on transgene expression patterns and intensities (Kvon et al., 2014; Kvon et al., 2020). Developing an efficient method for site-directed transgenesis in zebrafish will improve the effectiveness and reproducibility of transgenic enhancer assays.

Supplementary Material

Supplementary Material

Supplementary Fig. S1. Promoter features of promoter constructs.

(A-F) Promoter features found in the E1b (A), P0.8 (B), P2 (C), fos (D), DSCP (E), and ZSP (F) promoters. DPE, downstream promoter element; MTE, motif ten element.

Supplementary Fig. S2. Annotated promoter sequences.

Detailed annotation of promoter features found in the E1b (A), P2 (B), fos (C), DSCP (D), and ZSP (E) promoters. P0.8 is annotated as part of P2. DPE, downstream promoter element; MTE, motif ten element.

Supplementary Fig. S3. E1b and P2 do not drive injury-responsive expression in adult fins.

Uninjured and 2 days post-amputation (dpa) adult F0 E1b:EGFP, LEN-E1b:EGFP, P2:EGFP, and LEN-P2:EGFP caudal fins. White arrowheads indicate amputation plane. White arrows indicate injury-induced EGFP expression. Numbers indicate number of fins expressing EGFP. Scale bar: 100 μm.

Supplementary Table S1. Minimal promoters used in zebrafish and Drosophila.

Supplementary Table S2. Sequences of primers used in this study.

Supplementary Table S3. Promoter sequences.

Supplementary Table S4. Animal numbers used in this study.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
Metronidazole	Sigma-Aldrich	M3761–100G
Tricaine methanesulfonate	Pentair	TRS1
Experimental models: Organisms/strains
Zebrafish: Ekkwill: Wild type	Ekkwill Breeders	EK (EKW), RRID:ZFIN_ZDB-GENO-990520–2
Zebrafish: Ekkwill: Tg(cmlc2:mCherry-NTR)	(Chen et al., 2013)	pd71Tg/+, RRID:ZFIN_ZDB-GENO-140310–4
Oligonucleotides
See Table S3 for subcloning primers	This paper	N/A
Recombinant DNA
Plasmid: Tol2-ISceI-E1b:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-P0.8:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-P2:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-fos:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-DSCP:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-ZSP:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-LEN-E1b:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-LEN-P0.8:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-LEN-P2:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-LEN-fos:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-LEN-DSCP:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-LEN-ZSP:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-103runx1EN-E1b:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-103runx1EN-P0.8:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-103runx1EN-P2:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-103runx1EN-ZSP:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-gata2ai4-E1b:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-gata2ai4-P0.8:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-gata2ai4-P2:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Plasmid: Tol2-ISceI-gata2a-i4-ZSP:EGFP;cryaa:mCherry-ISceI-Tol2	This paper	N/A
Software and algorithms
R	R Core Team	https://www.r-project.org/
RStudio	RStudio, PBC	https://www.rstudio.com/
Zeiss ZEN	Carl Zeiss AG	https://www.zeiss.com/microscopy/us/products/microscope-software/zen.html

Highlights.

• fos and DSCP minimal promoters drive high rates of leaky expression

• Developmental enhancers are preferentially compatible with E1b promoter

• Regeneration enhancers are preferentially compatible with P0.8 and P2 promoters

• ZSP promoter is highly compatible with developmental and regeneration enhancers