Transvection between nonallelic genomic positions in Drosophila

Abstract

In Drosophila, pairing of maternal and paternal homologous chromosomes can permit trans-interactions between enhancers on one homolog and promoters on another, an example of transvection. Although trans-interactions have been observed at many loci in the Drosophila genome and in other organisms, the parameters that govern enhancer action in trans remain poorly understood. Using a transgenic reporter system, we asked whether enhancers and promoters at nonallelic, but nearby, genomic positions can communication in trans. Using one transgenic insertion carrying the synthetic enhancer GMR and another nearby insertion carrying the hsp70 promoter driving a fluorescent reporter, we show that transgenes separated by 2.6 kb of linear distance can support enhancer action in trans at the 53F8 locus. Furthermore, transvection between the nonallelic insertions can be augmented by a small deletion flanking one insert, likely via changes to the paired configuration of the homologs. Subsequent analyses of other insertions in 53F8 that carry different transgenic sequences demonstrate that the capacity to support transvection between nonallelic sites varies greatly, suggesting that factors beyond the linear distance between insertion sites play an important role. Finally, analysis of transvection between nearby nonallelic sites at other genomic locations shows evidence of position effects, where one locus supported GMR action in trans over a linear distance of over 10 kb, whereas another locus showed no evidence of transvection over a span <200 bp. Overall, our data demonstrate that transvection between nonallelic sites represents a complex interplay between genomic context, interallelic distance, and promoter identity.

Introduction

In Drosophila and other dipteran organisms, homologous chromosomes are closely paired from end to end in virtually all somatic cells (McKee 2004; Joyce et al. 2016). This chromosomal configuration can permit genetic regulatory regions to communicate between the 2 homologous chromosomes, a phenomenon known as transvection (Lewis 1954). Although examples of transvection and related trans-sensing phenomena continue to be uncovered and characterized in Drosophila and other organisms, the parameters that govern how genetic regulatory elements interact between 2 chromosomes in trans remain poorly understood.

In one form of transvection, an enhancer on one homolog can act in trans on a promoter on its corresponding homolog, stimulating transcription when the chromosomes are paired. Enhancer action in trans was first uncovered as a mechanism underlying intragenic complementation between specific types of mutant alleles. For example, if one allele lacks an enhancer (“Enhancerless”) and the other lacks a functional promoter (“Promoterless”), neither can support gene expression alone, but when paired, the remaining functional enhancer and promoter of the 2 alleles can interact in trans to achieve transcription (Fig. 1a; Lewis 1954; Geyer et al. 1990; Duncan 2002; Kennison and Southworth 2002). In most cases, alleles that carry a fully intact promoter in cis to a functional enhancer show weak enhancer action in trans (Martínez-Laborda et al. 1992; Casares et al. 1997; Gohl et al. 2008; Tian et al. 2019), or none at all (Geyer et al. 1990; Morris et al. 1998), suggesting that a promoter in cis to the functional enhancer is a preferred competitive target relative to a promoter in trans.

Fig. 1.

Transvection between nearby nonallelic positions. a) Schematic showing enhancer action in trans between paired homologous chromosomes, where one allele lacks an enhancer and the other allele lacks a promoter, at allelic positions (top), or nonallelic positions (bottom). Circle, enhancer; bent arrow, promoter; X, mutated or deleted promoter; rectangle, transcribed gene. b) Transgenic insertions used for prior transvection studies in polytene band 53F8. P{attP.w+.attP}JB53F, just upstream of the GstS1-RB TSS, is an RMCE landing site (Bateman and Wu 2008; Bateman et al. 2012), whereas P{TV2-GMR}53F, 2.6 kb distal to P{attP.w+.attP}JB53F, and just upstream of the CG46491-RB TSS, is based on the pWFL vector for transgene coplacement (Siegal and Hartl 1996; King et al. 2019). Both GstS1 and CG46491 are endogenously expressed in wild-type eye-antennal discs (Potier et al. 2014; Bateman and Johnson 2022). c) Third instar larval eye disc showing GMR acting in trans on the hsp70 promoter driving GFP expression. Transgenes are inserted on homologous chromosomes at the P{attP.w+.attP}JB53F landing site. A and P, anterior and posterior. Dashed box indicates general field of view for higher-resolution images in Fig. 2.

More recently, the study of enhancer action in trans has been bolstered by transgenic approaches, with site-specific integrases or recombinases allowing designed “alleles” of reporter genes to be inserted at identical positions in the genome (Chen et al. 2002; Bateman et al. 2012; Mellert and Truman 2012). In one approach, a P-element carrying recognition sequences for the integrase phiC31 is first randomly inserted in the genome; subsequently, the integrase attachment sites are used to precisely integrate Enhancerless and Promoterless fluorescent reporters that mimic alleles of endogenous genes where transvection has been observed (Bateman et al. 2012). Flies can then be generated where the Enhancerless transgene is carried on one homolog and the Promoterless transgene is carried at the precise allelic position on the other homolog, allowing the transgenes to pair and thereby permitting transvection between them (Fig. 1a). Transgenic systems have thus far been used to demonstrate that the Drosophila genome is generally permissive to transvection (Chen et al. 2002; Kravchuk et al. 2016; King et al. 2019), that the capacity to act in trans is common among enhancers (Mellert and Truman 2012; Blick et al. 2016), and that the incorporation of insulator sequences into transgenes can augment enhancer action in trans (Kravchenko et al. 2005; Schoborg et al. 2013; Lim et al. 2018; Piwko et al. 2019).

Since cytological evidence suggests that transvection in Drosophila ultimately relies on somatic homolog pairing, it may be reasonably assumed that productive trans-interactions between enhancers and promoters will be restricted to those that are encoded in allelic positions on homologous chromosomes. Indeed, disruption of homolog pairing by chromosomal rearrangement typically leads to loss of transvection and transvection-disrupting rearrangements are considered a form of experimental proof that bona fide pairing-dependent interactions are taking place between 2 alleles (Lewis 1954). Similarly, Enhancerless and Promoterless transgenes that are inserted into different locations in the genome will typically fail to support enhancer action in trans, presumably because they fail to pair with each other (Chen et al. 2002; Bateman et al. 2012; Mellert and Truman 2012), although the incorporation of sequences such as insulators and Polycomb Response Elements (PREs) into transgenes can sometimes cause insertions at nonhomologous positions to come together (Bantignies et al. 2003; Kravchenko et al. 2005). Notably, past experiments that tested interactions between nonallelic positions made use of transgenic insertions that are relatively far away from one another along the chromosome, or on different chromosomes altogether. However, it is as yet unclear whether Enhancerless and Promoterless transgenes that are inserted relatively close to one another on homologous chromosomes have the potential to productively interact (Fig. 1a). Exploring this question may reveal insight into how strictly homologous chromosomes align with one another.

Here we assess the capacity of Enhancerless and Promoterless transgenes that are inserted at nearby nonallelic positions to support enhancer action in trans using existing transgenic insertions from prior analyses of transvection (Bateman et al. 2012; King et al. 2019) and from large-scale screens for gene disruption (Bier et al. 1989; Huet et al. 2002; Bellen et al. 2004, 2011; Thibault et al. 2004). Our data show that transvection can be observed for transgenic insertions separated by at least 10 kb; however, in some cases, insertions separated by mere tens or hundreds of base pairs fail to support transvection, suggesting that factors such as genomic position and transgene identity have a major influence on productive trans-interactions between enhancers and promoters.

Materials and methods

Stocks and fly husbandry

Transgenic insertions used to assess transvection are described in Table 1. H{Pdelta2-3}HoP8 is an X-chromosomal hobo insertion that constitutively expresses the P-element transposase. A multiply marked second chromosome carrying mutations in aristaeless (al), dumpy (dp), brown (bw), and speck (sp) was used to uncover male recombinants. All flies were cultured at 25°C on standard Drosophila cornmeal, yeast, sugar, and agar medium with p-hydroxybenzoic acid methyl ester as a mold inhibitor (Bateman et al. 2012).

Table 1.

Insertions used in this study.

Insert	Polytene position	Genomic positiona	Strand	References
P{attP.w + .attP}JB53Fb	53F8	2R:17,097,510	+	Bateman and Wu (2008)
P{TV2-GMR}53Fc	53F8	2R:17,100,154	−	King et al. (2019)
P{TV2-GMR}42Ac	42A13	2R:6,214,221	+	King et al. (2019)
P{TV2-GMR}37Cc	37C5	2L:19,158,447	−	King et al. (2019)
P{lacW}GstS1[k11405]	53F8	2R:17,097,378	+	Bier et al. (1989); Bellen et al. (2004)
P{lacW}GstS1[k08805]	53F8	2R:17,097,399	+	Bier et al. (1989); Bellen et al. (2004)
P{lacW}GstS1[k09854]	53F8	2R:17,097,454	+	Bier et al. (1989); Bellen et al. (2004)
P{lacW}GstS1[k11301]	53F8	2R:17,097,505	+	Bier et al. (1989); Bellen et al. (2004)
P{lacW}GstS1[k10815]	53F8	2R:17,097,631	+	Bier et al. (1989); Bellen et al. (2004)
PBac{RB}CG30431[e01618]	42A13	2R:6,203,346	+	Thibault et al. (2004)
P{lacW}Ttc7[k15603]	42A13	2R:6,214,072	+	Bier et al. (1989); Bellen et al. (2004)
P{Epgy2}EY20090	42A13	2R:6,214,143	−	Bellen et al. (2011)
P{wHy}Ttc7[DG04810]	42A13	2R:6,214,285	+	Huet et al. (2002)
P{Epgy2}bin3[EY11308]	42A13	2R:6,214,574	+	Bellen et al. (2011)
P{Epgy2}bin3[EY09582]	42A13	2R:6,214,695	−	Bellen et al. (2004)
P{XP}brat[d11404]	37C5	2L:19,158,258	−	Thibault et al. (2004)
P{Epgy2}brat[EY01093]	37C5	2L:19,158,440	+	Bellen et al. (2004)
P{XP}brat[d08172]	37C5	2L:19,161,727	+	Thibault et al. (2004)

^a Insertion positions according to Release 6 of the Drosophila genome (dm6).

^b Landing site for phiC31-mediated cassette exchange. The original insertion is marked with mini-white; Enhancerless and Promoterless GMR constructs were subsequently inserted using phiC31 integrase, which removes mini-white from the genome (Bateman et al. 2012).

^c Insertion based on the pWFL vector for transgene coplacement (Siegal and Hartl 1996); Enhancerless and Promoterless GMR constructs are derived using Cre and FLP, respectively (King et al. 2019).

Screen for male recombinants and flanking deletions

A screen to generate flanking deletions around the Promoterless GMR P-element insertion near GstS1 is outlined in Fig. 3b. Male flies carrying P{GMR(P-)lacZ}JB53F were crossed to virgin females of genotype H{Pdelta2-3}HoP8 y w; al dp bw sp/CyO, dp, sp to generate F1 male progeny of genotype H{Pdelta2-3}HoP8 y w/Y; P{GMR(P-)lacZ}JB53F/al dp bw sp. In the germlines of these flies, aberrant mobilization of the P-element at 53F can create Chromosome II recombination events at the site of the P-element that may also generate a small flanking deletion or duplication (Preston and Engels 1996); since the transgene is located between dp and bw on the second chromosome map, proximal deletions would be associated with recombinant chromosomes of genotype al dp P{GMR(P-)lacZ}JB53F bw⁺sp⁺, whereas distal deletions (in the direction of the Enhancerless GFP construct near CG46491) would be associated with recombinant chromosomes of genotype al⁺ dp⁺ P{GMR(P-)lacZ}JB53F bw sp. To screen for recombinant chromosomes, F1 males were crossed to virgin females carrying a CyO balancer marked with dp and sp mutations, allowing F2 progeny to be scored for at least one marker to the left and to the right of the P-element insertion. Only males were scored in the F2 to ensure loss of the P-element transposase that was carried on the X chromosomes of F1 males.

Fig. 3.

Deletions flanking a transgenic insertion can augment transvection between nonallelic positions. a) Schematic showing theoretical changes to paired chromosomes when one homolog carries a deletion located between 2 nonallelic positions. Subsequent “looping out” of unpaired DNA on the intact homolog results in a decreased linear distance of paired DNA between the 2 sites. b) Genetic scheme for generating deletions flanking P{GMR(P-)lacZ}JB53F. See Materials and Methods for a detailed description of the screen. c–e) Transvection between P{GMR(P-)lacZ}JB53F with flanking deletions and P{hsp70GFP}TV2-53F on an intact chromosome. Columns of images show merged or individual channels for staining with anti-GFP or anti-Elav; schematics above each image set show configurations of transgenes. c) Control P{GMR(P-)lacZ}JB53F^7L carrying no deletion. Arrows indicate GFP-positive cells. d) P{GMR(P-)lacZ}JB53F^2R with an 813 bp distal deletion. e) P{GMR(P-)lacZ}JB53F^4L with an approximately 1 kb deletion internal to the transgene that includes the entire 5′ P-element end. f) Quantification of GFP-positive cells per disc for each of the genotypes presented in c–e). Each genotype carrying a deletion (2R, 4L) is significantly different from the control (P < 0.05, Mann–Whitney U test).

Approximately 2050 F2 males were scored, and 20 candidate recombinants were identified. Fifteen of these were successfully placed into stock by crossing to virgin females carrying a CyO balancer; of these, 9 were candidates for distal deletions, and 6 were candidates for proximal deletions based on flanking markers. Several candidates were tested for the presence of the Promoterless GMR P-element and for flanking deletions using primers GstS1_1, GstS1_2, GstS1_2_Ex1, Pend, RNXG9, Pry4, and Sp1 (see Supplementary Fig. 1 and Table 1 for primer locations and sequences). For P{GMR(P-)lacZ}JB53F^4L, a proximal deletion candidate, sequencing of a GstS1_1/RNXG9 PCR fragment showed that it carries a 1,017 bp deletion that extends from a position 15 bp into the 5′ P end at the proximal end to a position 110 bp downstream of the transgenic AAUAAA polyadenylation signal at the distal end, thereby removing all but the terminal-most 15 bp of the 5′ P end in addition to the proximal attR recombination sequences. For P{GMR(P-)lacZ}JB53F^7L, also a proximal deletion candidate, sequencing of a GstS1_1/Sp1 PCR product showed that it carries a small 30 bp tandem duplication of the 5′ P end inverse terminal repeat, but no other structural changes. For P{GMR(P-)lacZ}JB53F^2R, a distal candidate, sequencing of a Pry4/GstS1_2_EX1 PCR product shows a deletion beginning 19 bp into the 3′ P end at the proximal end to a position 813 bp distal to the P insertion.

Staining, microscopy, imaging, and scoring

Third instar eye-antennal discs were dissected, fixed, and stained with antibodies as previously described (Blick et al. 2016) using polyclonal rabbit anti-GFP (Invitrogen) diluted 1:2,000, mouse monoclonal anti-Elav [9F8A9, Developmental Studies Hybridoma Bank (DSHB)] diluted 1:400, mouse monoclonal antibeta-galactosidase (40-1a, DSHB) diluted 1:110, goat antirabbit AlexaFluor-488 secondary antibodies (Invitrogen) diluted 1:2,000, and/or goat anti-mouse Cy3 secondary antibodies (Jackson Immunoresearch) diluted 1:250. Stained discs were mounted in Fluoromount G with DAPI (Affymetrix) and visualized using a Zeiss Axio Imager.A2 epifluorescence microscope and AxioCam Mrm camera with Zen Lite software, with the exception of high magnification views of discs shown in Fig. 2, which were imaged using a Zeiss Axioplan 2 microscope with a 510 Meta confocal laser scanning system, with optical sections collected at 0.7-mm increments. Confocal images were processed in FIJI image analysis software (Schindelin et al. 2012) to combine z-stacks into a single flat image (using max intensity) for publication.

Fig. 2.

GMR acts in trans between nonallelic positions separated by 2.6 kb. Columns of images show merged or individual channels for staining with anti-GFP (indicative of transvection) or anti-Elav, a neural marker for all photoreceptors. Schematics above each image set show configurations of transgenes located at the 2 insertion sites shown in Fig. 1b. a and b) Enhancerless and Promoterless transgenes at allelic positions near GstS1-RB a) or near CG46491-RB b). c and d) Enhancerless and promoterless transgenes at nonallelic positions, with c and d carrying reverse configurations between the 2 insertion sites. Arrow in D, weak GFP signal in a single photoreceptor.

Counts of GFP-positive cells presented in Fig. 3 were determined from images of whole eye-antennal discs. To assess mini-white eye pigmentation, adult male flies of each genotype were collected within 8 h of eclosion and aged 5 days prior to scoring. Fly eyes were imaged under consistent light and exposure levels using a Canon EOS Rebel Tli digital camera mounted on a Leica MZ7.5 stereomicroscope using EOS Utility Imaging Software (Canon Inc.). At least 10 flies of each genotype were examined and scored separately by at least 2 scorers. Note that scoring was performed at the microscope and not from digital images.

Results

We previously incorporated transgenes carrying fluorescent reporters that mimic Enhancerless and Promoterless alleles of classical transvection systems into several locations in the Drosophila genome (Bateman et al. 2012; King et al. 2019). Specifically, these reporters carry a coding region for a reporter gene (lacZ or GFP), and either the eye-specific synthetic enhancer GMR (Moses and Rubin 1991) or the minimal core promoter of the hsp70 gene. When Enhancerless and Promoterless constructs are placed at allelic positions on homologous chromosomes within the same organism, enhancer action in trans can be observed via GFP fluorescence within differentiated photoreceptors of third instar larval eye discs (Fig. 1a–c).

By chance, 2 of our insertions are separated by only 2.6 kb between the GstS1 and CG46491 genes within polytene band 53F8 on chromosome arm 2R (Fig. 1b). Hi-C analysis of third instar eye-antennal discs supports that the 2 insertions are within the same Topologically Associated Domain (TAD) (Viets et al. 2019). To address whether transvection can be supported between transgenes at these nearby but nonallelic positions, we crossed flies carrying an Enhancerless construct at one location and a Promoterless construct at the other location and assessed fluorescence in third instar larval eye discs in the resulting progeny. In control crosses where Enhancerless and Promoterless transgenes are at allelic positions, GFP positive cells are apparent in the majority of mature ommatidial clusters, consistent with previous observations (Fig. 2a and b; Bateman et al. 2012; King et al. 2019). When the Enhancerless construct near CG46491 is paired with the Promoterless construct near GstS1, GFP positive retinal cells are observed in all discs examined (n = 7), with fewer positive cells relative to the control discs carrying constructs in allelic positions (Fig. 2c). In the reverse configuration with an Enhancerless construct near GstS1 and a Promoterless construct near CG46491, very few GFP positive cells are observed (Fig. 2d); however, at least 1 GFP positive cell was observed in each disc examined (n = 7), whereas negative control discs carrying Enhancerless or Promoterless constructs alone show no evidence of GFP expression (Bateman et al. 2012; King et al. 2019). Thus, enhancer action in trans is supported between transgenic insertions separated by 2.6 kb, with differences in the levels of activation depending on the positions of the enhancer and promoter.

A deletion on one homolog augments transvection between nonallelic sites

We next asked whether transvection between nonallelic sites could be improved by decreasing the distance between the positions of the transgenic insertions on one of the homologous chromosomes, which we predicted would alter the pairing configuration of the locus to “loop out” the unpaired portion of the unaffected chromosome and thereby bring the enhancer and promoter into closer proximity (Fig. 3a). To accomplish this, we used P-element-induced male recombination to generate chromosomal deletions flanking the Promoterless construct near the GstS1 gene (Preston and Engels 1996). In this method, a P-element is given the opportunity to remobilize by introducing the P transposase via genetic cross. In some instances, an aberrant remobilization event can create a recombinant chromosome carrying a small deletion adjacent to the P-element that otherwise remains in its original position; these events can be easily detected by screening for recombination events in the male germline where conventional meiotic crossing over does not take place (Preston and Engels 1996; Duttaroy 2002).

We screened 2,050 male progeny according to the cross scheme in Fig. 3b and identified 20 putative recombinants around the Promoterless construct near GstS1, a rate of 0.1% that is consistent with previous observations (Preston and Engels 1996; see Materials and Methods). Subsequent analysis showed that one line, P{GMR(P-)lacZ}JB53F^2R, carried an approximately 800 bp deletion of genomic DNA immediately distal to the P-element, which meets our criteria of reducing the distance between the transgenic insertion sites on one homolog. A second recombinant line, P{GMR(P-)lacZ}JB53F^4L, carried an approximately 1 kb deletion that removes the 5′ P-element end, with no changes to flanking genomic DNA. Finally, P{GMR(P-)lacZ}JB53F^7L showed a small 30 bp duplication of the 5′ P-element terminal repeat but no major flanking deletions, and was used as a control.

We crossed each of the 3 recombinant lines to the Enhancerless construct near CG46491 and assessed enhancer action in trans in larval eye discs. The recombinant control line with no deletions, P{GMR(P-)lacZ}JB53F^7L (Fig. 3c), appeared qualitatively indistinguishable from the original P{GMR(P-)lacZ}JB53F construct that lacks structural changes (Fig. 2d), with very few GFP positive cells in each disc. In contrast, the recombinant line with the 800 bp distal deletion, P{GMR(P-)lacZ}JB53F^2R (Fig. 3d), showed more GFP positive cells in each disc, indicating that the presence of the deletion augmented the capacity of GMR to act in trans from a nonallelic site. Surprisingly, the recombinant carrying a deletion of the 5′ P-element end, P{GMR(P-)lacZ}JB53F^4L (Fig. 3e), showed a more dramatic increase in GFP expression, even though the deletion was on the opposite side of the Promoterless transgene relative to the Enhancerless transgene. To quantify these differences, we scored the number of GFP positive cells per disc for each genotype (Fig. 3f), which showed a significant increase in the number of GFP-positive cells per disc for P{GMR(P-)lacZ}JB53F^2R (28.8 ±17.4 (mean ±standard deviation), n = 7 discs; P = 0.001, Mann–Whitney U test) and for P{GMR(P-)lacZ}JB53F^4L (113.0 ±31.3, n = 6 discs; P = 0.002) relative to control P{GMR(P-)lacZ}JB53F^7L discs (3.6 ±1.4, n = 6 discs). In sum, our data support that enhancer action in trans between nearby nonalleleic positions can be improved by decreasing the distance between the 2 locations on one of the homologous chromosomes, and that sequences within the 5′ P-element end may be detrimental to enhancer action in trans.

Parameters additional to linear distance impact transvection between nonallelic sites

To further explore the potential for transvection between nearby nonallelic positions, we sought other transgenes carrying Enhancerless reporters that could be assessed for activation by GMR. To this end, several large-scale screens have generated thousands of insertions of P-elements and other transposons marked with mini-white (Bier et al. 1989; Huet et al. 2002; Bellen et al. 2004, 2011; Thibault et al. 2004). The mini-white reporter is essentially an Enhancerless transgene that relies on the local chromatin environment to permit transcription and subsequent eye color pigmentation, ranging from yellow (low transcription) to red (high transcription) when assessed in an otherwise white⁻ background (Pirrotta 1988). We first determined whether mini-white could be activated by GMR from an allelic position in trans by crossing the Promoterless GMR transgene near GstS1 to its parent landing site P{attP.w⁺.attP}JB53F, which is marked with mini-white (Bateman and Wu 2008). In the absence of GMR in trans, mini-white expression from P{attP.w⁺.attP}JB53F results in an orange eye color, reflecting an intermediate level of transcription. When GMR is placed in trans at the allelic position, the resulting eye color is a robust red, supporting that mini-white transcription can be augmented by GMR enhancer action in trans (Supplementary Fig. 2).

We then searched the Drosophila genome annotation for transgenes carrying mini-white that had been inserted near GstS1 in polytene band 53F. Fortuitously, the GstS1 gene is a P-element hotspot, and many insertions have mapped near the promoter of the GstS1-RB transcript (Wangler et al. 2015). To more effectively compare transgenes inserted in different locations, we identified 5 nearby insertions of a single transgene type, P{lacW} (Table 1; Bier et al. 1989), for which stocks were available. All 5 insertions were mapped within 250 bp of one another surrounding the GstS1-B transcriptional start site, and all were inserted in the same orientation on the chromosome (Bellen et al. 2004; Fig. 4a).

Fig. 4.

GMR action in trans on nearby nonallelic mini-white insertions at the 53F locus. Each schematic shows the position of a transgene carrying GMR on homolog 1 and the positions of 5 different P{lacW} insertions near GstS1-RB on homolog 2. Images show adult eye pigmentation for flies where GMR is carried on homolog 1 (bottom row) and for control flies where homolog 1 does not carry a transgene (top row). The linear distance in base pairs from the P{lacW} insertion site to that of the GMR transgene [either proximal (p) or distal (d)] is indicated above each image set. Qualitative assessment of enhancement of pigmentation in the presence of GMR relative to control is indicated below the images for each P{lacW} insertion. a) GMR transgene P{GMR(P-)lacZ}JB53F inserted near GstS1-RB; b) GMR transgene P{GMR(P-)GFP}TV2-53F inserted near CG46491-RB.

We first crossed each P{lacW} line to the Promoterless GMR construct near GstS1 (Fig. 4a). Surprisingly, only 3 of the 5 P{lacW} insertions showed elevated mini-white expression relative to control flies without GMR in trans, even though the largest distance between P{lacW} insertions on one chromosome and the Promoterless GMR insertion on the homolog was just 132 bp. Notably, the 2 P{lacW} insertions that failed to show evidence of transvection were both inserted in the 5′ UTR of the GstS1-B transcript, whereas the 3 insertions with clear enhancement by GMR were inserted in upstream noncoding DNA (Fig. 4a). We then crossed the 5 P{lacW} insertions to the Promoterless GMR construct inserted near CG46491 approximately 2.5 kb away (Fig. 4b). We observed an identical pattern of enhancement as in the previous experiment, with the 3 P{lacW} insertions in noncoding DNA showing enhancement of mini-white relative to controls lacking GMR, and the 2 insertions in the GstS1-RB UTR showing no evidence of enhancement. Our data therefore show that the linear distance between transgenic insertions is not the sole predictor of whether transvection can be supported between them.

Each P{lacW} insertion carries a second transgene encoding Beta-galactosidase (from the Escherichia coli lacZ gene) and driven from the endogenous P-element promoter, which was originally intended as an “enhancer trap” to report the activity of endogenous enhancers near the point of insertion (Bier et al. 1989). As a complementary test of GMR's capacity to act in trans on nearby nonallelic positions, we crossed each of the 5 P{lacW} lines to the Promoterless GMR transgene near GstS1 and stained third instar larval eye discs using antibodies against Beta-galactosidase (Fig. 5a and b). In control discs lacking the GMR enhancer in trans, each P{lacW} insertion line showed similar levels of weak to undetectable Beta-galactosidase signal. Surprisingly, in the presence of GMR in trans, Beta-galactosidase staining showed a trend almost completely opposite to that shown by mini-white derived from the identical insertions; specifically, the 2 P{lacW} insertions in the 5′ UTR of GstS1-B now showed robust Beta-galactosidase staining, whereas the 2 insertions furthest upstream of GstS1-B showed little to no detectable change in Beta-galactosidase levels. The remaining P{lacW} insertion, positioned just upstream of GstS1-B, showed modest activation of lacZ in larval eye discs in addition to activation of mini-white in adult flies, thereby behaving differently from insertions of identical constructs located only 50 bp on either side. In sum, our observations indicate that the capacity to support transvection between nearby nonallelic positions does not simply scale with linear distance along the chromosome; rather, other parameters unique to each insertion site and transgene must also play a role in permitting transvection at detectable levels.

Fig. 5.

GMR action in trans on nearby nonallelic lacZ insertions differs from mini-white activation at the same positions. a) Schematic showing GMR transgene P{GMR(P-)lacZ}JB53F inserted near GstS1-RB on homolog 1 and 5 P{lacW} insertions near GstS1-RB on homolog 2. b) Representative third instar eye discs stained with antibeta-galactosidase antibodies where GMR is carried on homolog 1 (right column) and for control flies where homolog 1 does not carry a transgene (left column). The linear distance in base pairs from the P{lacW} insertion site to that of the GMR transgene [proximal (p) or distal (d)] is indicated to the left of each image set.

Position effects impact transvection between nonallelic sites

Finally, we asked whether transvection between nearby nonallelic sites can be supported at other genomic locations. We previously characterized GMR enhancer action in trans between allelic transgenic insertions at additional locations, including one upstream of the Ttc7-RA transcript in polytene band 42A13 and another upstream of the brat-RB transcript in polytene band 37C5 (King et al. 2019). For the Ttc7 gene region, we found existing stocks for 6 transgenic insertions that were marked with mini-white and inserted relatively close to the Ttc7-RA transcription start site (Fig. 6a, Table 1; Bier et al. 1989; Huet et al. 2002; Bellen et al. 2004, 2011; Thibault et al. 2004). When flies carrying each mini-white insertion were crossed to the Promoterless GMR construct upstream of Ttc7, we observed an increase in the pigmentation of the adult eye relative to control flies lacking the GMR enhancer in trans; although the change in eye color was subtle in some cases, it was consistently observed over many age- and sex-controlled adults, demonstrating that transvection between the nonallelic sites occurred for each combination (Fig. 6a). Notably, the largest distance from the insertion site of the mini-white transgene to that of the Promoterless GMR transgene was 10.8 kb. In contrast, for the brat gene region, we found just 3 suitable stocks carrying mini-white insertions near the brat-RB transcription start site (Fig. 6b, Table 1). At this locus, only the mini-white transgene nearest (7 bp distal) to the insertion site of the Promoterless GMR construct showed a mild but consistent increase in eye pigmentation relative to control flies, whereas inserts that were 189 bp distal or 3.3 kb proximal to the GMR insertion site showed no evidence of transvection (Fig. 6b). In sum, our data show that enhancer action in trans between transgenes at nonallelic positions is possible for distances of at least 10.8 kb, but factors including position effects and transgene identity have a major influence on whether transvection will be supported.

Fig. 6.

Position effects impact GMR action in trans on nonallelic mini-white insertions. Each schematic shows the position of a transgene carrying GMR on homolog 1 and the relative positions of several insertions carrying mini-white on homolog 2. a) Insertions near Ttc-RA and bin3 at polytene position 47A13. b) Insertions near brat-RB at polytene position 37C5. RNA-seq shows that each of Ttc7, bin3, and brat is endogenously expressed in wild-type eye-antennal discs (Potier et al. 2014). Images show adult eyes for flies where GMR is carried on homolog 1 (bottom row) and for control flies where homolog 1 does not carry a transgene (top row), with the linear distance in base pairs from the mini-white insertion to that of the GMR transgene indicated above each image set. Qualitative assessment of enhancement of pigmentation in the presence of GMR relative to control is indicated below the images. Eye pigmentation was scored at the microscope by at least 2 different scorers for each mini-white insertion.

Discussion

Transvection in Drosophila relies on intimate pairing of homologous chromosomes so that regulatory elements can communicate in trans. In this study, we sought to better understand the pairing requirements for transvection by testing the capacity of enhancers and promoters to communicate in trans between nearby but nonallelic transgenic insertions. Our observations show that enhancer action in trans can indeed occur between some nonallelic insertions that are separated by a linear distance of at least 10 kb. However, our data also demonstrate a lack of transvection between some transgenes that are inserted at positions separated by only tens of base pairs, indicating that factors beyond the linear distance separating insertion sites likely play an important role in determining whether transvection between nonalleleic positions can be supported.

At the 53F8 locus, our data show that transgenic insertions separated by 2.6 kb can support transvection. Curiously, our data showed more robust enhancer action in trans by GMR when the enhancer was located near the GstS1 gene and the hsp70 promoter and GFP coding region were located near the CG46491 relative to the reversed configuration. While this was not anticipated, it is possible that the chromatin environment upstream of GstS1 is more favorable for transcription, or that subtle differences between the 2 types of transgenes used have an unknown influence on transcriptional efficiency. Note that our method of examining GFP fluorescence reflects the total output of transcription and translation resulting from transvection, and therefore does not directly inform our understanding of transcriptional dynamics of transvection as has been described elsewhere (Lim et al. 2018).

We predicted that a deletion of genomic DNA adjacent to one of the transgenic insertions would increase the efficiency of transvection between nonallelic sites by changing the local chromatin topology (Fig. 3a). In support of this prediction, an approximately 800 bp deletion flanking the promoterless GMR insert near GstS1 increased the number of GFP-positive cells relative to a transgene lacking a deletion when each was placed in trans to the hsp70-GFP transgene near the CG46491 gene. According to our model, the increased enhancer action in trans in the presence of the deletion is due to the unpaired DNA of the intact chromosome being “looped out” in this configuration, leading to an effectively reduced linear distance of paired chromatin between the insertions that could increase the probability of productive interactions between enhancer and promoter. While this model is supported by the data, we cannot exclude the possibility that some other aspect of the altered pairing topology is relevant to the observed change in transvection, or that the deletion in our experiment removed a DNA sequence that interferes with transvection independently of local chromatin folding. In our screen for flanking deletions, we also uncovered a deletion of sequences within the 5′ P-element end of the Promoterless transgene near GstS1, which dramatically increased the capacity of GMR to act in trans relative to the transgene without a deletion. Interestingly, a recent quantitative assessment of enhancer action in trans by GMR at multiple positions in the genome showed that an outlier line with an extraordinarily high capacity for transvection also carried a deletion in the 5′ P-element end (King et al. 2019). While this may be purely coincidental, Fujioka et al. (2021) recently showed that a promoter in the 5′ P-element end can be activated by transgenic enhancers in cis, which can result in reduced transcriptional output from the transgene depending on the orientation and configuration of transgenic elements. It is therefore possible that the presence of a nearby 5′ P-element end is detrimental to enhancer action in trans, and studies based on P-element transgenes may underestimate the strength of transvection.

We were further surprised by our analysis of GMR action in trans using P{lacW} insertions near GstS1, where differences in the activation of mini-white in trans differed dramatically among 5 lines with insertions that were separated by only 250 bp. It is unlikely that these differences are due simply to changes in linear distance between the different P{lacW} insertions and the position of the GMR transgene since the same pattern was seen when using either the Promoterless GMR construct near GstS1 (within hundreds of bp from the P{lacW} insertions), or the Promoterless GMR construct near CG46491 (approximately 2.5 kb away). Moreover, analysis of beta-galactosidase production resulting from lacZ transcription from the same P{lacW} insertions showed a remarkably different pattern, with robust staining observed for the lines that showed little to no evidence of mini-white activation. Although puzzling, these data may result from competition and/or interference among the promoters at the 53F8 locus. For example, for each of the P{lacW} insertions, transcription of mini-white occurs on the opposite strand relative to GstS1; with this configuration, transcriptional interference between the mini-white and GstS1 promoters could prevent robust activation of mini-white for the P{lacW} insertions in the GstS1-RB 5′ UTR, whereas interference would not be predicted for insertions upstream of the GstS1-RB TSS. Furthermore, the lacZ transgene carried by P{lacW} insertions in the GstS1-RB 5′ UTR is positioned downstream of the mini-white transcription unit relative to the GstS1-RB promoter (Fig. 5). As such, any transcription of mini-white would be predicted to protect lacZ from transcriptional interference with GstS1, permitting lacZ transcription and subsequent beta-galactosidase production. In contrast, for the P{lacW} inserts upstream of the GstS1-RB TSS where transcriptional interference would not be predicted to occur, differences between the transgenic promoters that drive lacZ and mini-white may lead to favorable expression of the latter relative to the former. Specifically, lacZ is driven by the endogenous P-element promoter (Bier et al. 1989), which encodes only an Initiator (Inr) core promoter element and has previously been demonstrated to be a poor competitive target for enhancers acting in trans (Hodgetts and O’Keefe 2001; Mellert and Truman 2012). Conversely, the white promoter driving mini-white has both Downstream Promoter Element (DPE) and Inr core promoter elements, and prior analysis of transvection at the yellow locus has shown that it is a favorable target in trans (Kutach and Kadonaga 2000; Morris et al. 2004; Lee and Wu 2006). Ultimately, our observations highlight the potential complexity of outcomes when enhancers choose among promoter targets in cis and in trans.

Our data further show that transvection between nonallelic transgenic insertions is influenced by position effects, where an insertion near Ttc7 supported GMR activation of mini-white in trans from more than 10 kb away, whereas an identical transgene inserted near brat showed no evidence of trans-activation of a mini-white insertion <200 bp away. Several potential mechanisms could account for these differences. For example, the chromatin environment near Ttc7 could simply be permissive for higher levels of transcription; in support of this, King et al. (2019) showed that GMR activation of hsp70-GFP in cis results in a roughly 2-fold increase in fluorescence for the insertion near Ttc7 relative to the insertion near brat. Alternatively, Hi-C strategies have recently shown that the “tightness” of pairing varies across the genome, with some loci holding homologous chromosomes in tight register, and others allowing looser “sloppy” pairing (AlHaj Abed et al. 2019; Erceg et al. 2019). It could therefore be that the Ttc7 locus reflects a looser pairing state that permits higher levels of contact between nearby nonallelic regions of homologous chromosomes, whereas the brat locus may be held in a tighter paired state where trans-interactions are more restricted. Other analyses of transvection using transgenic insertions have shown that sequences such as insulators and PREs can facilitate interactions between insertions at nonallelic positions, possibly due to clustering of these elements into nuclear bodies (Sigrist and Pirrotta 1997; Muller et al. 1999; Kassis 2002; Bantignies et al. 2003; Pirrotta and Li 2012). Thus, position effects on transvection could also reflect the presence or absence of nearby insulator and PRE sequences. Ultimately, it is likely that a complex interplay of chromatin structure and topology influenced by local genetic elements will determine the capacity for productive trans-interactions between nonallelic sites over varying linear distances.

Data availability

The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables. Stocks are available upon request.

Supplemental material available at G3 online.

Funding

This work was supported by grants from the National Institutes of Health (P20 GM0103423), a Faculty Early Development (CAREER) award from the National Science Foundation to J.R.B. (1349779), and Bowdoin College.