The Capacity to Act in Trans Varies Among Drosophila Enhancers

Abstract

The interphase nucleus is organized such that genomic segments interact in cis, on the same chromosome, and in trans, between different chromosomes. In Drosophila and other Dipterans, extensive interactions are observed between homologous chromosomes, which can permit enhancers and promoters to communicate in trans. Enhancer action in trans has been observed for a handful of genes in Drosophila, but it is as yet unclear whether this is a general property of all enhancers or specific to a few. Here, we test a collection of well-characterized enhancers for the capacity to act in trans. Specifically, we tested 18 enhancers that are active in either the eye or wing disc of third instar Drosophila larvae and, using two different assays, found evidence that each enhancer can act in trans. However, the degree to which trans-action was supported varied greatly between enhancers. Quantitative analysis of enhancer activity supports a model wherein an enhancer’s strength of transcriptional activation is a major determinant of its ability to act in trans, but that additional factors may also contribute to an enhancer’s trans-activity. In sum, our data suggest that a capacity to activate a promoter on a paired chromosome is common among Drosophila enhancers.

THE spatial organization of the interphase eukaryotic genome is characterized by extensive long-distance interactions between distal chromosome regions (Sanyal et al. 2012). Interactions have been identified between sequences on the same chromosome (in cis) or on different chromosomes (in trans) (Lieberman-Aiden et al. 2009; Duan et al. 2010; Sexton et al. 2012; van de Werken et al. 2012; Nagano et al. 2013; Zhang et al. 2013). Many long-distance interactions in cis underlie the activation of specific genes, and in some cases, sequences have been identified that facilitate interactions between a distal enhancer and a specific promoter target (Zhou and Levine 1999; Calhoun et al. 2002; Calhoun and Levine 2003; Lin 2003; Akbari et al. 2008; Fujioka et al. 2009; Majumder et al. 2015). In contrast, the genetic impacts of trans-interactions between chromosomes are less clearly understood. Examples of gene regulation involving interchromosomal associations have been described (Spilianakis et al. 2005; Bacher et al. 2006; Xu et al. 2006; Apostolou and Thanos 2008; Sandhu et al. 2009; Markenscoff-Papadimitriou et al. 2014; Patel et al. 2014), but it remains unclear whether it is common for sequences that regulate gene expression to communicate between different chromosomes when they are physically juxtaposed.

In Drosophila melanogaster, extensive trans-interactions are observed between homologous chromosomes in virtually all somatic tissues, a phenomenon known as somatic homolog pairing (reviewed by McKee 2004; Bosco 2012). The close proximity of homologous chromosomes in Drosophila can permit an enhancer to act in trans on a promoter on the paired homolog, a form of pairing-dependent gene regulation called transvection (Lewis 1954). Evidence for enhancer action in trans has been uncovered at a handful of genes in the Drosophila genome, often providing an explanation for unexpected intragenic complementation of loss-of-function alleles (Lewis 1954; Gelbart 1982; Geyer et al. 1990; Leiserson et al. 1994; Hendrickson and Sakonju 1995; Casares et al. 1997; Morris et al. 1998; Southworth and Kennison 2002; Marin et al. 2004; Coulthard et al. 2005; Gohl et al. 2008; Juni and Yamamoto 2009), or the nonadditive activity of paired mutant and wild-type alleles (Gibson et al. 2000; Lum and Merritt 2011; Bing et al. 2014).

More recently, transgenic approaches based on site-specific recombination have been developed for Drosophila to query specific enhancer fragments and genomic positions for support of transvection (Chen et al. 2002; Kravchenko et al. 2005; Bateman et al. 2012a; Mellert and Truman 2012; also see Kassis et al. 1991). Thus far, transvection has been observed at all genomic insertion sites tested, suggesting that the Drosophila genome is generally permissive to enhancer action in trans. Furthermore, transgenic experiments have demonstrated that promoters at allelic positions in cis and trans to an enhancer will compete for the enhancer’s activity, consistent with earlier classical observations (Geyer et al. 1990; Martinez-Laborda et al. 1992; Casares et al. 1997; Morris et al. 1999; Gohl et al. 2008). However, enhancer activation of promoters in cis and trans are not equivalent, with enhancers showing a strong preference for a promoter in cis, and frequently showing cell-to-cell variability in the activation of a promoter in trans (Bateman et al. 2012a; Mellert and Truman 2012).

Recently, Mellert and Truman (2012) used a transgenic approach to test 21 Drosophila enhancers for the capacity to support transvection. Their assay provided evidence that some, but not all, tested enhancers could activate a promoter on a paired homolog, raising the question of whether the capacity to act in trans is special to some enhancers. It does not appear that specific “tethering” sequences are required for an enhancer to act in trans since minimal enhancers consisting of multimerized binding sites for a single transcription factor can function in trans, even when that transcription factor is not native to Drosophila (Bateman et al. 2012a; Mellert and Truman 2012). Notably, in the system employed by Mellert and Truman, enhancers were generally juxtaposed to a strong promoter in cis, which may have decreased the likelihood of observing enhancer action in trans. Thus, their analysis may represent an underestimate of the proportion of enhancers that can support transvection.

Here we assess the generality of enhancer action in trans by testing a collection of known Drosophila enhancers for the capacity to support transvection. We employed a transgenic system that placed a weak, or absent, promoter in cis and a fluorescent reporter in trans to a collection of enhancers that act in third instar larval discs, tissues known to support transvection. Our data show evidence for enhancer action in trans for all enhancers tested, implying that the capacity to support transvection is a common property shared by Drosophila enhancers.

Materials and Methods

Stocks and fly husbandry

Flies carrying an enhancerless GFP construct downstream of an hsp70 minimal promoter at recombinase-mediated cassette exchange (RMCE) site 53F were described previously (Bateman et al. 2012a). Flies of genotype w[*]; P[w[+mC]=Ubi-p63E(FRT.STOP)Stinger]15F2 (Evans et al. 2009), carrying an FRT-flanked stop cassette between a ubiquitous promoter and a nuclear EGFP coding region on chromosome 3 (“G-TRACE cassette”), were obtained from the Bloomington Drosophila Stock Center (stock no. 32251). All flies were maintained at 25° on standard Drosophila cornmeal, yeast, sugar, and agar medium with p-hydroxybenzoic acid methyl ester as a mold inhibitor.

Plasmid construction and transgenics

Construction of piB-LTL-lacZ, an RMCE donor vector with loxP sites flanking an hsp70 minimal promoter upstream of a lacZ coding region, and GMR-LTL-lacZ, was described previously (Bateman et al. 2012a). To generate piB-LTL-lacZ derivatives carrying different enhancers, potential target enhancers were identified using the REDFly database (Gallo et al. 2011), and primers for amplifying enhancer sequences were either identified from a previous publication (Aerts et al. 2010) or designed using Primer3 software (Untergasser et al. 2012). Enhancer sequences were amplified via PCR from Drosophila genomic DNA and cloned into either pcr2.1 using a TOPO-TA cloning kit (Invitrogen) or pSC-A using a Strataclone PCR cloning kit (Agilent Technologies). Subcloned fragments were digested from these vectors and cloned into piB-LTL-lacZ using BamHI. Primer sequences for amplifying enhancers are listed in Supplemental Material, Table S1.

An RMCE donor vector carrying the hsp70 minimal promoter upstream of a FLP coding region was generated via a PCR-based strategy (Bateman and Wu 2008) using the template pCaSpeR-DEST5 (obtained from the Drosophila Genomics Resource Center) and primers hsp70FLP_5_attB40 and hsp70FLP_3_attB40 (additional primer sequences are provided in Table S2). The final donor vector was created by cloning the resulting PCR product into pSC-A using a Strataclone PCR cloning kit.

Paxpaxpax-LTL-lacZ and salsalsal-LTL-lacZ, carrying three tandem copies of the pax and sal enhancers, respectively, were created by serial cloning of PCR-amplified enhancer fragments with primer-encoded restriction sites into piB-LTL-lacZ. Briefly, a PCR fragment carrying the enhancer, a single XhoI site, and flanked by BamHI sites, was cloned into a BamHI site upstream of the promoter of piB-LTL-lacZ. Next, a second fragment carrying the same enhancer with a single StuI site and flanked by XhoI sites was cloned into the XhoI site created in the previous step. Finally, a third fragment carrying the enhancer flanked by StuI sites was cloned into the new StuI site.

All donor constructs were integrated into an RMCE target site at position 53F via RMCE as previously described (Bateman et al. 2006, 2012a; Bateman and Wu 2008). Following integration, the orientation of each insert was assessed using primers lac4 or 3′-Pend1, which are complementary to the 5′ and 3′ P-element ends flanking the RMCE cassette, respectively, and RNXG9, which is complementary to the SV40 3′-UTR downstream of the lacZ and FLP coding regions. All constructs were analyzed using insertions oriented with the 3′-UTR closest to the 5′ P-element end, with the exception of CG1625-LTL-lacZ, for which only one insertion in the opposite orientation was obtained. Promoterless derivatives of pax-LTL-lacZ and klu-LTL-lacZ insertions were generated via Cre/loxP recombination as previously described (Siegal and Hartl 1996; Bateman et al. 2012a).

Staining and microscopy

Imaginal discs were dissected from wandering third instar larvae in phosphate-buffered saline (PBS) and fixed in 4% formaldehyde (Electron Microscopy Sciences) in PBS for 20 min. Discs were then rinsed three times in PBS plus 0.1% Triton X-100 (PBST) and blocked in 4% normal goat serum (NGS) in PBST for 1 hr, then incubated with primary antibodies overnight at 4°. Antibodies used were polyclonal rabbit anti-GFP (1:2000; Invitrogen), mouse monoclonal antibodies elav-9F8A9 (Elav), 2B10 (Cut), and 40-1a (β-galactosidase, β-gal) (1:400, 1:100, 1:110, respectively; Developmental Studies Hybridoma Bank, DSHB), and rat monoclonal antibody Rat-Elav-7E8A10 (1:250; DSHB). Next, discs were rinsed and then washed two times for 20 min each in PBST, blocked in 4% NGS/PBST for 1 hr, and then incubated for 1 hr with secondary antibodies antirabbit Alexa Fluor-488 (1:2000; Invitrogen), antimouse-Cy3 (1:250; Jackson Immunoresearch), antirat-Cy3 (1:250; Jackson), and/or antimouse-Cy5 (1:250; Jackson) as required. Discs were then washed four times for 20 min each in PBST and mounted in Fluoromount G with DAPI (Affymetrix eBioscience). Discs were imaged using a Zeiss Axio Imager.A2 fluorescence microscope with an AxioCam MRm camera and Zen software. Confocal imaging was performed using a Zeiss Axioplan 2 microscope with a 510 Meta confocal laser scanning system. Statistical analyses were performed using Prism (GraphPad Software) or R Statistical Software packages. A D’Agostino and Pearson omnibus normality test was used to determine whether to test data via parametric or nonparametric methods.

To stain for β-galactosidase activity, imaginal discs were dissected from wandering third instar larvae in PBS, fixed for 15 min in 1% glutaraldehyde/PBS, rinsed three times in PBST, and then incubated in X-gal staining solution [10 mM NaH₂PO₄ pH 7.2, 150 mM NaCl, 3.3 mM K₃[Fe(CN)₆], 3.3 mM K₄[Fe(CN)₆], 0.1% Tween, 0.2% X-gal] at room temperature. Following staining, discs were rinsed three times in PBST and then mounted in 80% glycerol. Discs were imaged using an Olympus BX51 compound microscope equipped with a Media Cybernetics Evolution VF color camera and QCapture Pro Software.

Quantitative RT-PCR

Assessment of mRNA levels via quantitative RT-PCR was performed as previously described (Bateman et al. 2012a). Briefly, for each sample, 20 imaginal discs were dissected from wandering third instar larvae and frozen at −80°. Tissue homogenization, genomic DNA elimination, and RNA purification were carried out using an RNeasy plus kit (Qiagen) according to the manufacturer’s protocol. PCR was performed on a StepOne Real-Time PCR system (Applied Biosystems) using cDNA diluted 1:5 into SYBR green PCR Mastermix (Applied Biosystems). Primers SV40_3′-UTR_2F and SV40_3′-UTR_2R were used to amplify lacZ cDNA, and RP49-58F and RP49-175R were used to amplify the housekeeping rp49 cDNA as an internal reference. Relative levels of transcript were calculated via the ΔΔCt method using StepOne software.

Data availability

The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article.

Results

To test whether the capacity to act in trans is a common property of Drosophila enhancers, we designed a transgene-based strategy using lacZ and GFP reporters. Since some enhancers were previously shown to have decreased action in trans in the presence of a promoter in cis, the lacZ reporter was designed to permit removal of its hsp70 promoter via Cre/loxP recombination [designated the “loxP-TATA-loxP” (LTL) promoter] (Figure 1). The reporter construct also carried attB sequences suitable for RMCE, permitting targeted transgene insertion in the absence of phenotypic markers (Bateman et al. 2006). Our strategy was to clone various enhancers into the reporter, then target each construct to a common location in the Drosophila genome where an enhancerless hsp70-GFP reporter had already been placed. The activity of each enhancer in cis can then be assessed by β-gal activity, while trans-action can be determined by GFP fluorescence (Figure 1).

Figure 1.

Study strategy. (A) Each enhancer is cloned into a common reporter carrying a minimal hsp70 promoter flanked by loxP sites and upstream of a lacZ coding region. Enhancer activation of the promoter in cis can be assessed by staining for β-gal activity. Enhancer action in trans is assessed by placing each lacZ construct in trans to an enhancerless hsp70-GFP construct at the same genomic location and assessing GFP fluorescence. Note that the promoter of the lacZ construct can also be removed by Cre/loxP-mediated recombination. (B) Cartoon representation of a third instar eye disc. The morphogenetic furrow (MF) represents a wave of mitotic divisions that moves from the posterior (down) to the anterior (up) of the disc. Cells anterior to the MF are precursor cells that have not yet differentiated, while those posterior to the MF initiate a pattern of differentiation to form ommatidial clusters (OC). Each mature cluster consists of eight photoreceptor cells, R1–R8, surrounded by four cone cells (c), all of which develop via a stereotyped developmental program following the passage of the MF. OC schematic is adapted from Mavromatakis and Tomlinson (2013).

We first tested the eye-specific enhancer GMR, which was previously shown to support transvection in the developing larval eye disc (Bateman et al. 2012a). In comparison to a GMR-hsp70-lacZ transgene lacking loxP sites, β-galactosidase activity from the GMR-LTL-lacZ reporter was substantially weaker, with an ∼40-fold difference in lacZ transcript levels as assessed by quantitative RT-PCR (Figure S1). We did not further pursue the nature of the decreased activity, possibly caused by interference of the loxP sites with the function of the promoter, but we reasoned that the weak action of the LTL promoter should be advantageous to our analysis of transvection since a weak promoter in cis is more likely to release an enhancer’s activity to a strong promoter in trans (Morris et al. 1999; Mellert and Truman 2012). Importantly, GMR-LTL-lacZ activation of GFP in trans was observed in the expected pattern in cells posterior to the morphogenetic furrow of third instar eye discs (Figure 1B and Figure 2), demonstrating that this system supports enhancer action in trans.

Figure 2.

Enhancers active in the third instar eye disc act in trans with varying strengths. Each panel shows a representative eye disc carrying the indicated enhancer in trans to hsp70-GFP, with the fraction of scored discs that show GFP-positive cells indicated. The wing-specific enhancer sal was used as a negative control. Each enhancer was qualitatively assigned to a strong, moderate, or weak transvection class based on the number of GFP-positive cells observed (see main text). Arrowheads indicate weak GFP-positive cells.

We then selected an additional 14 enhancers from the REDfly database of known Drosophila cis regulatory elements (Gallo et al. 2011) (Table 1). We focused on enhancers that are active in stereotyped patterns across the developing ommatidial clusters of the third instar larval eye disc, reasoning that this tissue is generally permissive to transvection based on prior analyses (Leiserson et al. 1994; Wu and Howe 1995; Bateman et al. 2012a). Each enhancer was cloned upstream of the LTL-lacZ construct, targeted to a common RMCE site and crossed to flies carrying the enhancerless hsp70-GFP at the same genomic location. As a negative control, we also assayed a wing disc-specific enhancer fragment from the gene sal (Guss et al. 2001).

Table 1. Enhancers analyzed in this study.

Enhancer name	Size (bp)	Cell-type specificity	Distance to promoter (kb)	Reference
aos	292	Cone cells	Intron, 6	Wildonger et al. 2005
CG1625	801	Photoreceptor precursors	Upstream, 0	Aerts et al. 2010
CG30492	601	Photoreceptor precursors	Upstream, 0	Aerts et al. 2010
dscam	1703	Photoreceptor precursors	Intron, 10	Aerts et al. 2010
ey	212	Eye primordial cells	Intron, 12	Hauck et al. 1999
fas2	539	Photoreceptor precursors	Intron, 13	Aerts et al. 2010
GMR	190	Photoreceptors	Upstream, 0	Moses and Rubin 1991
klu	364	R1, R6, R7, cone cells	Intron, 10	Wildonger et al. 2005
neur	809	Photoreceptor precursors	Intron, 8–14	Aerts et al. 2010
nmo	754	Photoreceptor precursors	Intron, 50	Aerts et al. 2010
pax	361	Cone cells	Intron, 7	Flores et al. 2000
pros	1220	R3, R4, R7, cone cells	Intron, 32	Xu et al. 2000
Rapgap1	802	Photoreceptor precursors	Intron, 1	Aerts et al. 2010
Rh4	159	Photoreceptors	Upstream, 1	Fortini and Rubin 1990
sev	476	R3, R4, R7, cone cellsa	Intron, 6	Bowtell et al. 1991
Ser	812	Wing margin	Downstream, 22	Yan et al. 2004
sal	328	Wing pouch quadrants	Upstream, 9	Guss et al. 2001
ct(2.7)	2.7 kb	Wing margin, Air Sac Primordium (ASP)	Upstream, ∼80	Jack et al. 1991
ct(668)	668	Wing margin	Upstream, ∼80	Guss et al. 2001

^aHighest early expression in R3 and R4, with later high expression in R7 and cone cells. Reports conflict regarding weak expression in R1, R6, and mystery cells (Tomlinson et al. 1987; Bowtell et al. 1991; Domingos et al. 2004).

In third instar eye discs carrying the sal wing enhancer, we observed only a low level of background autofluorescence distributed evenly across the tissue (Figure 2). In contrast, for all 14 eye-specific enhancers, we observed cells with green fluorescence above background levels, consistent with each enhancer having some capacity to activate a promoter in trans. In all cases, the pattern of fluorescence was unevenly distributed across the regularly patterned ommatidial clusters, consistent with the variegated pattern of transvection previously observed for GMR in this tissue (Bateman et al. 2012a) (Figure 2). The levels of GFP fluorescence and numbers of GFP-positive cells varied greatly for the different enhancers, with some enhancers showing broad GFP fluorescence across the tissue, while others showed just a few fluorescent cells in some, but not all, discs. We qualitatively assigned each enhancer to one of three general classes of transvection strength: strong, with many GFP-positive cells present in all discs (four enhancers, including GMR), moderate, with fewer scattered GFP-positive cells present in all discs (three enhancers), and weak, with some discs showing a few GFP-positive cells and others showing none (eight enhancers). For two enhancers of the weak class, pax and klu, we used Cre/loxP-mediated recombination to remove the promoter in cis to the enhancer and then repeated the analysis of transvection. In the case of the klu enhancer, the proportion of discs showing GFP-positive cells increased significantly (3/9 vs. 16/20, P = 0.03, Fisher’s exact test), whereas no significant change was observed for the pax enhancer (12/20 vs. 7/11, P = 1.0), suggesting that enhancers may differ in sensitivity to a weak promoter in cis. Given the weak nature of the LTL promoter and the widespread enhancer action in trans observed in its presence, we did not pursue removal of the promoter for other enhancer constructs. In sum, our analysis suggests that the capacity to act in trans is a general property of enhancers acting in the eye disc, but the strength with which they do so varies greatly.

A FLP-based assay further supports transvection by weak class enhancers

Although the enhancers of the weak class each showed evidence of transvection when placed in trans to hsp70-GFP, the GFP fluorescence was frequently at our limit of detection using fluorescence microscopy. We therefore wished to provide further confirmation that these enhancers indeed activate promoters in trans. To this end, we designed a complementary system to detect transvection based on the G-TRACE tools for cell lineage analysis (Evans et al. 2009). In this system, the recombinase FLP acts on a G-TRACE cassette to remove a transcriptional stop signal between a ubiquitous enhancer and a GFP transgene, resulting in robust GFP fluorescence in a cell’s lineage (Figure 3A). To adapt G-TRACE to the study of transvection, we targeted an hsp70-FLP construct to RMCE site 53F so that it could be placed in trans to enhancers of the weak transvection class. We reasoned that this scheme would result in a highly sensitive detection method; indeed, in negative control discs carrying the G-TRACE cassette in which an enhancerless hsp70-lacZ was placed in trans to hsp70-FLP, we observed an occasional GFP-positive cell posterior to the morphogenetic furrow, suggesting that the level of background activity of the hsp70-FLP construct in the absence of an enhancer is very close to the threshold level required for removal of the transcriptional stop signal of the G-TRACE cassette.

Figure 3.

G-TRACE-based strategy further supports enhancer action in trans by weak class enhancers. (A) Schematic for an alternative detection scheme, where hsp70-GFP (Figure 1A) is replaced with hsp70-FLP. Trans-activation of FLP at suitable levels removes a transcriptional stop signal on a G-TRACE cassette located at a different genomic location, resulting in strong GFP fluorescence in that cell. Schematic adapted from Evans et al. (2009). (B) Scoring of GFP-positive cells using the G-TRACE system. Each symbol represents the number of GFP-positive cells in a single disc; lines overlaying each cluster of symbols represent mean ± SD. (C) Representative eye discs from flies carrying the indicated enhancer activating the G-TRACE system (nmo and aos not shown).

We crossed flies carrying each enhancer-LTL-lacZ construct of the weak transvection class to flies carrying hsp70-FLP and the G-TRACE cassette and scored the number of GFP-positive cells in each eye disc for at least 10 discs of each genotype (with the exception of ey; see below). For each enhancer, the mean number of GFP-positive cells was significantly higher than that of negative control discs lacking an enhancer (adjusted P < 0.05 for each enhancer, Kruskal–Wallis test with Dunn’s multiple comparisons test) (Figure 3, B and C), further supporting that each enhancer has the capacity to activate a promoter in trans. Notably, this analysis confirmed that the enhancer Rh4 supports transvection in the larval eye disc, an unexpected result given that the Rh4 gene is normally active later in development (see Discussion).

The weak class enhancers analyzed above are active in postmitotic cells, which are not expected to undergo cell divisions once GFP has been activated. In contrast, the weak class enhancer ey becomes active in eye-specific precursor cells during embryogenesis (Halder et al. 1998; Hauck et al. 1999), allowing us to use the G-TRACE system to determine the stage(s) at which this enhancer can act in trans; if transvection by the ey enhancer is restricted to postmitotic cells, we should only observe single GFP-positive cells posterior to the morphogenetic furrow, whereas if the enhancer acts in trans before this time, we should see clonal patches of GFP-positive cells. Analysis of the ey enhancer showed evidence of both of these predictions: among differentiated cells, some single GFP-positive cells were surrounded by nonfluorescent neighboring cells, whereas in other areas of the disc, larger clusters of GFP-positive cells were observed (Figure 4). Note that clonal patches of GFP-positive cells were also observed anterior to the furrow, consistent with the early action of the ey enhancer (Figure S2). Importantly, clusters of fluorescent cells were never observed in negative control eye discs lacking an enhancer in trans to hsp70-FLP (n = 26) nor in discs carrying other weak class enhancers that are not active in embryonic precursor cells (n = 141). Among the discs analyzed, two large clonal patches created by the enhancer ey, likely derived from individual GFP-positive precursors, showed ∼180–200 GFP-positive cells, suggesting that at least seven to eight cell cycles had passed since the activation of FLP led to removal of the stop signal in the G-TRACE cassette (Figure 4C). Thus, the enhancer ey has the capacity to act in trans early in the development of the eye.

Figure 4.

The enhancer ey acts in trans early in development. (A) Eye disc with clonal regions of GFP-positive cells resulting from trans-activation of hsp70-FLP by the ey enhancer and subsequent activation of the G-TRACE cassette. Dashed line, morphogenetic furrow. (B) Z-projected confocal image of inset area from A. Clones (dashed circles) were defined as overlapping/touching groups of GFP-positive cells (>2) with no evidence of intervening GFP-negative cells. Single GFP-positive cells contacting only GFP-negative cells (arrows) were not scored. The region shown was scored as six clonal patches, with between 6 and 24 GFP-positive cells per clone. (C) The number of GFP-positive cells per clone was counted or estimated for 30 clonal patches. Overlaid lines represent mean ± SD.

Cell-type specificity of transvection correlates with strength and timing of cis-activity

Among the enhancers analyzed, GMR and sev were the most consistent and robust activators of GFP in trans. In both cases, the majority of mature ommatidia appeared to have at least one GFP-positive cell (Figure 2). However, it was rare for more than one or two cells per ommatidium to be GFP positive, in contrast to the expected patterns of expression in cis, where each enhancer can act in multiple cell types within each ommatidial cluster (Table 1). To determine which cell types showed GFP expression in these discs, we used anti-Elav, a marker of neuronal photoreceptor cells. Specifically, anti-Elav highlights eight cells per mature cluster, R1–R8, each of which occupies a stereotyped position relative to the axes of the disc (Tomlinson et al. 1987; Bowtell et al. 1989) (Figure 1B). In discs where GFP was activated in trans by GMR, high-resolution confocal microscopy confirmed that most ommatidial clusters contained just one or two GFP-positive cells (Figure 5A). Based on their positions within the photoreceptor clusters, GFP-positive cells were most commonly either R3 (57% of ommatidia scored) or R4 (84%), with other cell types rarely showing GFP fluorescence (9%) (Table 2). We confirmed the strong bias of GFP-positive cells to R3 and R4 using the marker mδ0.5-lacZ, which shows highest expression levels in R4 (Cooper and Bray 1999; Domingos et al. 2004) and occasional weak labeling of R3 in later ommatidia (Figure S3). This bias was not due to a position effect at 53F, as identical constructs inserted at a different genomic location showed a similar bias for R3 and R4 (data not shown). Importantly, all GFP-positive cells resulting from GMR transvection at 53F were also positive for Elav, indicating that trans-activation by the GMR enhancer was restricted to the expected cell types where GMR is known to act.

Figure 5.

Cell-type specificity of transvection correlates with strength and timing of cis-activity. (A) Eye discs showing trans-activation of hsp70-GFP by the enhancer GMR. Costaining with anti-Elav shows that the majority of GFP-positive cells are either R3 or R4. (B) Image shows a single ommatidial cluster with GFP driven in cis by GMR. The positions of R2–R4 are highlighted. Relative fluorescent intensities of R4 vs. R2 (blue) or R3 (red) show higher R4 expression relative to R2 (2.17 ± 0.55), but equivalent expression in R4 and R3 (1.06 ± 0.26). (C and D) Eye discs showing trans-activation of hsp70-GFP by sev (C) or fas2 (D) enhancers with costaining for the neuronal marker Elav.

Table 2. Cell-type specificity of hsp70-GFP trans-activation.

Enhancer	Ommatidial clusters scored	R3 positive (%)	R4 positive (%)	Other R cell positive (%)
GMR	145	57	84	9
sev	203	27	84	3a
fas2	194	5	11	9

^aOne Elav-negative cell, scored a cone cell, was also positive for GFP.

To further investigate the bias in transvection by GMR, we analyzed the relative strength of the enhancer in cis across different photoreceptor identities. At high resolution, expression of lacZ driven by the weak LTL promoter was patchy and inconsistent across the tissue, and therefore not suitable for quantitative imaging of single cells (data not shown). We therefore analyzed expression of an existing construct wherein GFP is driven in cis by GMR and an hsp70 promoter lacking loxP sites (Bateman et al. 2012a). GFP expression was noticeably high in R3 and R4 in virtually all ommatidial clusters (Figure 5B; Figure S4); to quantify this pattern, we compared the fluorescence intensity of the R4 cell in each cluster to that of R2, representive of a cell type with low fluorescence, or R3, with fluorescence similar to that of R4. Averaged across 86 ommatidia, R4 was roughly twice as bright as R2 (mean relative intensity = 2.17 ± 0.55) but equivalent in intensity with R3 (mean relative intensity = 1.06 ± 0.26). Thus, a plausible explanation to account for cell-type bias in GMR-mediated transvection is that R3 and R4 show the highest level of GMR activity in cis, and therefore may have a higher probability of showing activation of a promoter in trans.

Similar to the case with GMR, sev expression in cis is highest in R3 and R4 in early ommatidial clusters, followed by later increases in expression in R7 and in cone cells (Tomlinson et al. 1987; Bowtell et al. 1989, 1991; Domingos et al. 2004). Notably, discs where GFP was activated in trans by sev once again showed a strong bias for R3 and R4 vs. other cells (Figure 5C; Table 2). Specifically, out of 203 ommatidia scored for transvection by sev, we observed only six cases of GFP-positive cells that were not R3 or R4; of those six, five were also positive for Elav, though we did not precisely determine their identities, and one was negative for Elav, with positioning and appearance consistent with a cone cell (data not shown). In contrast to GMR and sev, the strong class enhancer fas2 has not been reported to show biased expression in subsets of photoreceptors (Aerts et al. 2010). In discs where fas2 activates GFP in trans, GFP-positive cells were restricted to neuronal fates as expected, and the bias toward R3 and R4 was strongly reduced (Figure 5D; Table 2), consistent with a model wherein the cell-type-specific strength of enhancer action in cis is predictive of biased cell-type specificity of transvection. However, for all three strong class enhancers analyzed, trans-activation of GFP was always observed most often in R4, with the strongest bias observed for the sev enhancer and the mildest for fas2. Thus, while the strength and timing of an enhancer’s activity may have a strong effect on the likelihood of transvection in a particular cell identity, other factors may also favor transvection in particular cell types.

We also attempted to determine cell-type specificity for GFP-positive cells resulting from weak class enhancers acting on hsp70-GFP in trans, but the GFP fluorescence produced by these genotypes was typically at the limit of detection and not sufficiently robust to confidently assign cell identities at high resolution. We therefore turned to our FLP-based assay, where GFP fluorescence produced by the G-TRACE cassette is strong and easily detected. However, analysis of this system was complicated by the low frequency spontaneous FLP activity in the absence of transvection (Figure 3). To better understand the spontaneous activity, we used either anti-Elav, a marker of neuronal photoreceptors, or anti-Cut, a marker of cone cells (Blochlinger et al. 1993), to costain control discs carrying the G-TRACE cassette wherein hsp70-FLP was placed in trans to an enhancerless hsp70-lacZ cassette. Our data indicate that the spontaneous activity is restricted to cells of neuronal fate, with complete overlap between GFP and Elav staining (23/23 cells from eight discs) and no overlap between GFP and Cut staining (0/19 cells from eight discs). We then analyzed discs carrying the G-TRACE cassette wherein hsp70-FLP was placed in trans to a cone-cell-specific enhancer, either pax or aos (Flores et al. 2000; Wildonger et al. 2005) and costained with anti-Cut, reasoning that any overlap between GFP and Cut staining must result from the activity of the enhancer and not from spontaneous FLP activity on the G-TRACE cassette. In both cases, we saw colocalization between GFP and Cut staining that differed significantly from control discs (Figure S5). Specifically, for aos-mediated transvection using the G-TRACE system, 32/34 GFP-positive cells were also positive for Cut, whereas pax transvection led to fewer total numbers of GFP-positive cells, 9/45 of which were positive for Cut. Although we cannot exclude the possibility that weak class enhancers activate GFP outside their expected cell types (see Discussion), our data are consistent with transvection by pax and aos enhancers being primarily restricted to cone cells.

Enhancer transvection class correlates with enhancer strength

Although each of the 15 enhancers tested is active broadly across the eye disc when acting on a promoter in cis, the enhancers vary greatly in the number of cells where transvection is observed in each disc. While several parameters may influence variation in transvection by the different enhancers, a simple possibility that may contribute to their difference is the overall strength of transcriptional activation supported by each enhancer. Specifically, those enhancers that are generally strong transcriptional activators in cis may have a greater chance of acting on a promoter in trans, whereas those enhancers that act weakly in cis may be much less likely to activate a promoter in trans. To test this possibility, we first stained imaginal discs representing a subset of strong and weak class enhancers for β-gal activity, which reflects the strength of cis-activation. Using identical staining conditions for discs carrying each enhancer tested, we found that enhancers of the strong class tended to produce darker β-gal staining than those of the weak class (Figure 6A; data not shown). We then used quantitative RT-PCR to assess lacZ transcript levels in third instar eye-antennal discs from five lines each of strong and weak class enhancers (Figure 6B). Overall, we observed a >100-fold difference in lacZ transcript levels between the strongest and weakest enhancers. Consistent with the β-gal activity analysis, the enhancers with the highest activation of lacZ in cis tended to be of the strong class supporting transvection, whereas those with the lowest levels of cis-activity tended to be of the weak transvection class. Comparison of lacZ transcript levels grouped by enhancer class (strong vs. weak) showed that this difference is significant (P = 0.036, Mann–Whitney U-test), supporting that an enhancer’s strength in cis is a major determinant of its capacity to act in trans.

Figure 6.

Enhancer transvection class correlates with enhancer strength in cis. (A) β-Gal staining of eye discs from representative strong and weak class enhancers. All discs were stained under identical conditions. (B) Relative levels of lacZ mRNA generated by different enhancers as assessed by quantitative RT-PCR from eye-antennal discs. Levels are normalized to the enhancer ey (= 1.0). Enhancers of the strong transvection class were more likely to drive higher levels of lacZ mRNA in cis.

To further support a link between enhancer strength and transvection class, we created a construct carrying three tandem copies of the weak class enhancer pax. Multimerization of enhancer and/or transcription factor binding site sequences has been previously employed in transgenic organisms to boost an enhancer’s strength of activation, likely by recruiting higher local concentrations of specific transcriptional activators (e.g., Moses and Rubin 1991; Scott et al. 1999; Horn et al. 2000; Zimmerman et al. 2000). Quantitative RT-PCR showed that lacZ transcript levels in paxpaxpax-LTL-lacZ eye-antennal discs are approximately sevenfold higher than those produced by pax-LTL-lacZ, indicating an increased level of cis-activity by the multimerized enhancer (data not shown). Interestingly, staining of paxpaxpax-LTL-lacZ eye-antennal discs with antibodies to β-gal showed a pattern of cis-expression noticeably broader than that expected for cone cells, the established specificity of the pax enhancer (Figure S6). Higher resolution imaging confirmed that, in discs carrying paxpaxpax-LTL-lacZ, β-gal staining overlapped that of the neuronal marker anti-Elav (Figure 7A), demonstrating that tandem multimerization of the pax enhancer expands its typical cell-type specificity to include neurons in addition to cone cells. When placed in trans to hsp70-GFP, paxpaxpax-LTL-lacZ produced robust GFP fluorescence in many cells posterior to the morphogenetic furrow in all discs analyzed (n = 11), a pattern consistent with the strong class of enhancers and in stark contrast to the weak trans-activation of hsp70-GFP produced by a single copy of the pax enhancer (Figure 7B). Consistent with the broad cell-type specificity of the multimerized enhancer, costaining of discs carrying paxpaxpax-LTL-lacZ in trans to hsp70-GFP with either anti-Elav or anti-Cut showed that GFP fluorescence overlapped both neurons and cone cells (Figure 7, C and D). Thus, increasing enhancer strength via multimerization, which in this case increases transcriptional activity and expands the number of cells in which the enhancer is active, can convert an enhancer with barely detectable trans-activity to one with robust trans-activity.

Figure 7.

Multimerization of the pax enhancer increases its capacity to support transvection. (A) Staining of eye discs carrying paxpaxpax-LTL-lacZ with antibodies against β-gal and Elav. Arrowheads show β-gal-positive cells that overlap Elav-positive neurons, whereas arrows show cells with positions consistent with cone cells. Note that the signal for β-gal appears weak at high resolution due to the weak nature of the LTL promoter. (B) Representative eye discs carrying pax or paxpaxpax showing trans-activation of hsp70-GFP. Multimerization of the pax enhancer results in a change from weak to strong transvection class. (C and D) Eye discs showing paxpaxpax activation of hsp70-GFP in trans, with staining for GFP and either the cone cell marker Cut (C) or the neuronal marker Elav (D). Cells showing overlap (arrowheads) and no overlap (arrows) of signals are visible for both markers.

Analysis of wing enhancers further supports a link between enhancer strength and transvection

To determine whether a link between enhancer strength in cis and transvection may be common to enhancers of other tissues, we created LTL-lacZ constructs carrying three different enhancers that are active in the third instar larval wing disc: Ser, sal, and ct(2.7) (Table 1). Qualitative β-gal staining shows that Ser is the strongest of the enhancers, producing dark staining at the wing margin, while sal and ct(2.7) act more weakly in cis (Figure 8A). When placed in trans to hsp70-GFP, 14/14 discs carrying the Ser enhancer showed robust GFP fluorescence. In contrast, trans-activation of hsp70-GFP by the ct(2.7) enhancer was comparatively weaker and variable, and the sal enhancer produced no detectable GFP in trans, even when the promoter in cis was removed via Cre/loxP-mediated recombination (Figure 8A and data not shown). Further testing of the three enhancers using hsp70-FLP and the G-TRACE cassette demonstrated strong transvection by Ser in all discs, but weak transvection by ct(2.7) and sal, with just a few GFP-positive cells in the activation domains of each enhancer (Figure 8A). An exception was the air sac primordium (ASP) in discs carrying the ct(2.7) enhancer, which showed strong β-gal staining and more consistent GFP fluorescence in the G-TRACE system relative to the wing margin, consistent with a link between enhancer strength and transvection.

Figure 8.

Analysis of wing enhancers further supports a link between enhancer strength and transvection class. (A) β-Gal staining and trans-activation of hsp70-GFP and the hsp70-FLP/G-TRACE system is shown for the wing disc enhancers Ser, ct(2.7), and sal. Ser, the strongest enhancer in cis, shows more consistent and robust trans-action than the weaker ct and sal enhancers. Arrowheads highlight β-gal-positive or GFP-positive cells in the expected domains of the wing pouch for ct and sal; arrows marks the air sac primordium (ASP). The fraction of scored discs that showed GFP-positive cells is given for each test for trans-activation. The ASP was not scored in discs activating hsp70-GFP due to high levels of background fluorescence in that region. (B) Multimerization of sal to create salsalsal increases cis-expression ∼45-fold as assessed by quantitative RT-PCR on wing discs. Data represent two independent experiments. Inset shows β-gal staining of a wing disc carrying salsalsal-LTL-lacZ. (C) salsalsal activates hsp70-GFP in trans at high levels relative to a single copy of the sal enhancer.

As a final test, we multimerized three tandem copies of the weak sal enhancer to create salsalsal-LTL-lacZ. We predicted that the multimerized version would boost cis-activity and that it would subsequently show robust fluorescence when placed in trans to hsp70-GFP. Indeed, quantitative RT-PCR showed that lacZ transcript levels were >45 times higher in wing discs from flies carrying the salsalsal enhancer relative to those without multimerization, and β-gal activity assays showed intense staining that was restricted to the pattern expected for a single copy of the sal enhancer (Figure 8B; Figure S1). Notably, trans-activation of hsp70-GFP, which was not detectable for the single copy of sal, was robust and easily detected in all discs carrying salsalsal-LTL-lacZ (n = 9) (Figure 8C). Thus, enhancers of the wing disc further support a relationship between enhancer strength and transvection.

Factors beyond enhancer strength also influence transvection

The ct(2.7) enhancer described above is a large 2.7-kb fragment derived from a region ∼80 kb upstream of the ct gene promoter (Jack et al. 1991). We also tested a smaller 668-bp ct enhancer fragment, ct(668), that is contained within the larger 2.7-kb enhancer and is also capable of driving expression at the wing margin of third instar wing discs (Guss et al. 2001). Quantitative RT-PCR on wing disc tissue showed that lacZ transcript levels from the smaller fragment are roughly twofold higher than that of the larger fragment (Figure 9A), although β-gal staining suggests that both large and small ct enhancers are weaker than the Ser enhancer described above (Figure 8A and Figure 9B). Notably, although cis-activation by the smaller ct(668) enhancer appears to be stronger than that of the larger ct(2.7) enhancer, trans-activation of hsp70-GFP by the small enhancer was not observed (0/8 discs; Figure 9C) as it was for the large enhancer (10/10 discs; Figure 8A). Analysis of the smaller ct(668) enhancer using the G-TRACE system was complicated by the widespread appearance of large patches of GFP-positive cells, presumably due to nonspecific trans-activation of the hsp70-FLP transgene in precursor cells, although GFP-positive cells could also be seen at the wing margin in some discs (Figure S7). In sum, although there is an overall correlation between the strength of an enhancer acting in cis and its capacity to support transvection, additional factors likely also contribute to the ability of some enhancers to act in trans.

Figure 9.

Factors beyond enhancer strength may also contribute to transvection. (A) Truncated fragment of the 2.7-kb ct enhancer produces higher levels of lacZ expression in cis as assessed by quantitative RT-PCR. Data represent three independent experiments. (B) β-Gal staining shows the ct(668) enhancer fragment is active at the wing margin. (C) No trans-activation of hsp70-GFP is detected at the wing margin for the shorter ct enhancer fragment.

Discussion

The generality of gene regulation via transvection in Drosophila has been a long-standing question. Prior classical data provided evidence for enhancer action in trans at a handful of loci, demonstrating that several Drosophila enhancers can activate a promoter on a separate chromosome. However, in the majority of these cases, transvection was uncovered serendipitously through the chance isolation of specific types of mutations, and it remained unclear whether enhancers that can act in trans represent a distinct class of regulatory elements. In this study, we systematically analyzed known enhancers for a capacity to act in trans. We focused our analysis on tissues known to support transvection and created a system based on a weak promoter in cis to each enhancer to optimize the likelihood of observing enhancer action in trans. Our data showed evidence of transvection for all enhancers tested, albeit at greatly varying levels. In sum, our data support the hypothesis that enhancers that act in trans do not represent a special functional class; rather, it is plausible that any enhancer has the capacity to act on a promoter of a closely paired chromosome.

A previous study by Mellert and Truman (2012) used a related transgenic approach to test 21 neural-specific enhancers for an ability to support transvection and found no evidence for enhancer action in trans for roughly half of their sample, contrasting the results of our study. The two studies differ in several fundamental respects; perhaps most importantly, we used a weak (or absent) promoter in cis and a strong promoter in trans, whereas the prior test of 21 neural enhancers relied on transgenes carrying identical strong promoters in cis and in trans. Thus, it is conceivable that enhancers that were scored negatively for transvection in the previous study were, in fact, restricted from acting in trans at detectable levels due to competition from the promoter in cis, whereas other enhancers in the study may have been less sensitive to the promoter in cis. Consistent with this hypothesis, the enhancers of the yellow gene do not appear to act in trans in the presence of a strong promoter in cis (Geyer et al. 1990; Morris et al. 1999), whereas enhancers from other genes will show low levels of transvection when juxtaposed to a strong cis-promoter and higher levels of transvection in the absence of a cis-promoter (Martinez-Laborda et al. 1992; Casares et al. 1997; Gohl et al. 2008; Bateman et al. 2012a; Mellert and Truman 2012). We also found that two enhancers from our study differed in sensitivity to a promoter in cis, as Cre/loxP removal of the LTL promoter increased transvection by the klu enhancer but not by the pax enhancer. In sum, the differences between the present study and that of Mellert and Truman (2012) likely reflect the use of different strategies, with one study focusing on conditions favorable for observing transvection and the other representing a more stringent test.

Enhancer strength is a major determinant of transvection

Our data indicate that the varying levels of transvection displayed by different enhancers are largely due to their differing strengths, as broadly defined by the activation of a promoter in cis. This conclusion was supported in part by quantitative RT-PCR data from diverse enhancers that act in the eye disc and by multimerization of two weak class enhancers that led to increased cis-activity and nonadditive conversion to strong class transvection phenotypes. Formally, an enhancer’s output in cis could impact transvection in at least two nonmutually exclusive ways; in one model, the transcriptional output of an enhancer in an individual cell would influence whether or not transvection could occur in that cell, or, in another model, the proportion of cells of a given tissue in which the enhancer is active could create more or fewer opportunities for transvection to occur in that tissue. Our data are consistent with both of these models for the influence of enhancer strength on transvection. Specifically, multimerization of the sal enhancer dramatically increases transcriptional output in cis, but does not perceptibly alter the spatial pattern of the enhancer’s activity, implying that the impact of multimerization on transvection reflects increased transcription on a per-cell basis. Our data regarding the strong bias for sev and GMR transvection for R3 and R4, where activity in cis is strongest in early developing ommatidia, are also consistent with this model. In contrast, multimerization of the pax enhancer appears to increase both the transcriptional output of individual cells and the range of cell types in which the enhancer is active. Note that prior analysis of this enhancer also demonstrated a change in cell-type specificity to include neurons when the sequence was rearranged rather than multimerized (Swanson et al. 2010), and, more generally, strengthening the efficiency of reporter construct output has previously been shown to expand the detectable cell types in which an enhancer is active (Pfeiffer et al. 2010). In the case of the multimerized pax enhancer, it is likely that its increased ability to act in trans relative to a single pax enhancer reflects both higher rates of transcription per cell and increased opportunites for transvection via activity in an expanded number of cells.

In sum, our data support a model wherein, when acting in cis, an enhancer of the weak transvection class activates fewer rounds of transcription relative to a strong class enhancer, either due to lower levels of transcription in each cell, fewer cells in which the enhancer is active, or both. When acting in trans, a weak class enhancer is less likely to activate detectable levels of transcription in any given cell within the tissue, leading to fewer GFP-positive cells relative to a strong enhancer.

Factors other than enhancer strength influence transvection

Despite a general relationship between enhancer strength and trans-action, exceptions within our data indicate that other factors likely play a role in transvection for some enhancers and further highlight potential differences between cis- and trans-activity. For example, although the bias for transvection by GMR and sev in photoreceptors R3 and R4 may be explained in part by the strength of the enhancers’ cis-activity in those cells, we also found that transvection was observed more frequently in R4 than in any other cell, including R3. However, in the case of the GMR enhancer, we found no significant different between cis-activity in R3 and R4 as assessed by levels of GFP fluorescence. Thus, it is possible that factors and/or parameters that influence transvection may exist at different levels in different cell types and at different times of development. This possibility is consistent with prior observations; for example, analysis of transvection in the larval nervous system implied that certain neural cell types were more conducive to transvection than others (Mellert and Truman 2012). Furthermore, factors that promote or hinder somatic homolog pairing, a requirement for transvection, have been identified, and may function differently in different cell types (Hartl et al. 2008; Bateman et al. 2012b; Joyce et al. 2012). Finally, data based on DNA-FISH show high levels of somatic homolog pairing in third instar larvae, but lower levels during embryogenesis, which would likely negatively impact transvection in early tissues (Fung et al. 1998). Consistent with this, very few examples of transvection have been described in Drosophila embryos (Hendrickson and Sakonju 1995; Ronshaugen and Levine 2004; Fujioka et al. 2016) relative to later developmental time points. A broad study of enhancer action in trans across different tissues and developmental time points would be highly informative of different parameters that affect transvection.

In addition to biases in cell-type specificity of transvection, our data support the existence of DNA sequences that can modulate transvection. In the case of the ct enhancer, a large fragment with moderate expression at the wing margin showed a greater degree of transvection than did a smaller enhancer fragment with more robust expression at the wing margin. It is possible that the differences in trans-activity observed for these enhancer fragments are simply caused by their different sizes; for example, a larger fragment may alter local pairing between the two homologous chromosomes in a manner that facilitates access to the promoter in trans. However, the other enhancers in our study show no correlation between fragment size and strength of transvection (Table 1; data not shown), indicating that a size effect is not a general phenomenon. An intriguing possibility is that the larger ct enhancer fragment could possess sequences that aid in targeting distal promoters, including promoters in trans, that are not present in the smaller fragment. Indeed, the ct enhancer is normally 80 kb away from its promoter target, and sequences that promote long-range enhancer–promoter interactions in cis and in trans have been characterized for other genes (Hopmann et al. 1995; Zhou and Levine 1999; Calhoun et al. 2002; Calhoun and Levine 2003; Lin 2003; Akbari et al. 2008; Fujioka et al. 2009; Majumder et al. 2015). The notion that variation in transvection could be partially dependent on factors beyond enhancer strength is supported by previous studies. For example, pairing the same enhancer fragment with transgenes carrying different genomic fragments upstream of a trans-promoter can result in varying levels of transvection, suggesting that “compatibility” of paired sequences is important for robust enhancer action in trans (Mellert and Truman 2012). Furthermore, insulator sequences have previously been shown to have a positive influence on transvection and other long-distance enhancer–promoter interactions (Kravchenko et al. 2005; Schoborg et al. 2013; Fujioka et al. 2016). In addition, some examples of transvection at the abdominal-B locus require a 10-kb sequence in addition to the enhancer (Hopmann et al. 1995), and transvection of the Men locus varies greatly across genetic backgrounds (Lum and Merritt 2011). Thus, while our data support that an enhancer’s strength in cis is also reflected in its action in trans, it is likely that other sequences can further modulate transvection.

Finally, we cannot exclude the possibility that some enhancers act with different specificity in trans relative to their activities in cis. We were unable to reliably determine the cell-type specificity of transvection by many weak class enhancers due to the weak GFP staining produced by trans-activation of hsp70-GFP and complications of spontaneous FLP activity using the G-TRACE system. However, we were surprised to find that the Rh4 enhancer supported weak class transvection in the larval eye disc, given that the Rh4 gene is normally activated in this tissue at a later period in development (Pollock and Benzer 1988; Fortini and Rubin 1990). It is possible that the enhancer fragment used in our analysis also shows activity in the larval eye disc in cis that is not representative of the wild-type expression pattern, although we were unable to confirm this using β-gal staining and the weak LTL promoter (data not shown). Alternatively, it may be that low-level activation of a promoter in trans can occur outside the normal spatiotemporal domain of a given enhancer’s cis-expression. Consistent with this, prior analysis has shown that altering the identity of the promoter downstream of an enhancer in cis can also alter its cell-type specificity (Mellert and Truman 2012), implying an inherent fluidity of enhancer function in different cell identities that depends on the nature of enhancer–promoter communication.

Does transvection play a role in the wild?

Studies of transvection in Drosophila have traditionally relied on mutant and/or transgenic organisms where evidence of gene regulation in trans can be easily observed. While these analyses have provided great insight into mechanisms of enhancer action in trans, they leave outstanding questions of a potential role for transvection in wild-type organisms. Does somatic homolog pairing and/or transvection provide direct benefit to Drosophila and/or other organisms? Recent analysis has shown that programmed stochastic expression of Rh3 and Rh4 opsins in Drosophila is controlled in part by trans-interactions between homologs of the Spineless gene (Johnston and Desplan 2014), and other interallelic interactions that help to pattern gene expression may also exist. Furthermore, for the promoters of some genes, it may be that the combined input of enhancers in cis and trans can aid in ensuring tight regulation of patterned gene expression, as has been postulated for seemingly redundant enhancers found at some developmental genes (so-called “shadow” enhancers) (reviewed by Lagha et al. 2012). Given the wide range of transvection strength observed in our study, our data suggest that such combined action of enhancers in cis and trans may be more relevant to some genes than others. Consistent with this, enhancer action in trans observed using classical alleles of different genes can produce widely varying levels of transcription, from ∼1% or less of wild-type levels using certain alleles of yellow (Morris et al. 2004), to near 100% of wild-type activity for some alleles of the Men and sn-Glycerol-3-Phosphate Dehydrogenase genes (Gibson et al. 1999; Lum and Merritt 2011). In addition, many other types of regulatory sequences, including insulators and Polycomb response elements, may function more efficiently when paired or clustered (reviewed by Bantignies and Cavalli 2011; Herold et al. 2012). In non-Dipteran organisms, where somatic homolog pairing is the exception rather than the rule, interchromosomal interactions between nonhomologous sequences are routinely uncovered in whole-genome interaction assays (Lieberman-Aiden et al. 2009; Duan et al. 2010; Sexton et al. 2012; van de Werken et al. 2012; Nagano et al. 2013; Zhang et al. 2013), and specific trans-interactions have been linked to several examples of gene regulation (Spilianakis et al. 2005; Bacher et al. 2006; Xu et al. 2006; Apostolou and Thanos 2008; Sandhu et al. 2009; Markenscoff-Papadimitriou et al. 2014; Patel et al. 2014). Thus, it appears that interchromosomal communication is adaptable to the different global principles for genome organization present in Dipteran and non-Dipteran organisms and likely plays a role in maintaining correct patterns of gene expression across diverse species.