Transvection Arising from Transgene Interactions in Zebrafish

Rapid advances in the technologies for genetic manipulation in zebrafish have led to an extensive array of transgenic lines. Recently we encountered a novel phenomenon that demonstrates a precautionary principle regarding use of these approaches.

We generated a Tol2-based transgenic line expressing the human inwardly rectifying potassium channel gene kir2.1 (KCNJ2), Tg(UAS:hkir2.1-mCherry). kir2.1 has been used in prior studies including in zebrafish to hyperpolarize and silence neurons. Interestingly, when we crossed this line to a double transgenic line we have used previously, the triple-transgenic Tg(foxP2.A.2:Gal4), Tg(UAS:GFP), and Tg(UAS:kir2.1-mCherry) embryos had GFP expression throughout the brain and not just in the expected subset of foxP2 neurons (Fig. 1A–D) (“green brain” phenotype). There was no expansion of endogenous foxP2 expression (in situ data, not shown). Intriguingly, UAS:GFP and UAS:kir2.1-mCherry expression did not match (Fig. 1B–D); neither transgene's expression matched the normal expression pattern of foxP2.A.2. ¹ This was unexpected, given that both UAS:GFP and UAS:kir2.1-mCherry transgenes share the same UAS enhancer and minimal core promoter (MCP) sequences. Using a GCaMP5 reporter, inhibition of electrical activity matched the altered expression pattern of UAS:hkir2.1-mCherry labeling. For experiments we used the widely used Gal4-VP16_413–470 and 10x UAS constructs. Although we did not formally test whether any of the transgenes had more than one insertion, we have used these lines over multiple generations and not noted any variegation of expression, suggesting that they are either single copy; or if multiple copies, are closely linked.

OPEN IN VIEWER

FIG. 1.

(A–E) Maximum intensity confocal z-stack projections of whole-mount embryo immunofluorescence; (ob, olfactory bulb, tel, telencephalon, ey, eye, ot, optic tectum). (A) Neurons labeled by foxP2A.2:Gal4; UAS:GFP (green) show the expected expression pattern for foxP2.A.2 enhancer. (B–D) Trans interaction between foxP2A.2:Gal4 and UAS:hkir2.1-mCherry expands UAS:GFP labeling to produce the “green brain” phenotype. Although both UAS:hkir2.1-mCherry (red) and UAS:GFP (green) transgenes have the same UAS and MCP, they have different expression patterns. (E) A “red brain” phenotype occurs in foxP2.A.2:Gal4; UAS:kir2.1-mCherry embryos when crossed to UAS:RFP. (F, G) Live whole embryos at 24 hpf. foxP2A.2:Gal4; UAS:hkir2.1-mCherry, UAS:GFP embryos show green brain phenotype (closed arrowheads) in uninjected sibling controls. In Gal4 morpholino-injected triple transgenic animals, the green brain phenotype is absent (open arrowheads). (H) Somatic pairing of transgenes causes transvection, where the regulatory region of gene A influences transcription in trans of gene B. The ability of an enhancer to bind a transcriptional activator (MCP) is sufficient to drive transcription in trans. Many zebrafish transgenes contain the same MCP sequence. (I) In our model, the trans interaction of the UAS enhancer from UAS:hkir2.1-mCherry transgene, with the MCP of the foxP2A.2:Gal4 transgene, expands the expected Gal4 expression pattern because of genomic pairing of these transgenes. We hypothesize that endogenous cis-acting elements near the transgene insertion for UAS:hkir2.1-mCherry constrains the UAS enhancer, leading to an unexpected expression pattern for UAS:hkir2.1-mCherry. MCP, minimal core promoter.

We were unable to recapitulate the green brain using chemical or genetic mechanisms to disrupt neuronal activity, including glutamate receptor antagonists (AP5, CNQX), a sodium channel inhibitor (MS-222), increased extracellular K⁺, or crossing to tetanus toxin Tg(UAS:TeTx-CFP) fish. The potassium channel blocker chloroquine that specifically inhibits kir2.1 did not suppress the green brain phenotype. We did not observe a similar expansion in GFP labeling using other Gal4 driver lines. However, a Gal4 morpholino prevented the green brain phenotype (Fig. 1F, G). The expansion of GFP segregated with only the joint presence of the foxP2.A.2:Gal4 and UAS:kir2.1-mCherry transgenes, and we could create a “red brain” phenotype in foxP2.A.2:Gal4, UAS:RFP, and UAS:kir2.1-mCherry embryos (Fig. 1E). Outcrosses demonstrated that the transgenes were not at the same or a nearby genomic locus. Thus, the green brain is Gal4/UAS dependent, is dependent on two specific transgenes (foxP2.A.2:Gal4 and UAS:hkir2.1:mCherry), but is distinct from the normal expression pattern of foxP2.A.2.

This phenomenon is consistent with transvection, the trans interaction of genetic elements. ² Transvection can involve cis and/or trans interactions of genetic elements, often between paired (allelic) insertions (Fig. 1H). ³ Transvection can occur in trans when an enhancer drives expression by binding and activating expression at the paired allele. For example, in Drosophila, inserting two transgenes with different enhancers but a shared MCP into the same genomic location, including for UAS-driven transgenes, is sufficient to elicit transvection. ⁴

Many transgenes in the zebrafish community use the same MCP containing the viral E1b TATA box fused to the carp β-actin 5′ UTR. We hypothesize that the green brain phenotype is because of transvection between the UAS:hKir2.1:mCherry and foxP2.A2:Gal4 transgenes, containing the same MCP, which by chance have inserted into interacting positions in the genome. The UAS:hKir2.1:mCherry may have inserted into a genomic region where an endogenous cis-regulatory element constrains its UAS expression, but leads to trans-driven expression of foxP2.A.2:Gal4, and the subsequent expansion of GFP (or any other UAS-dependent gene) (Fig. 1I). Our findings are an important reminder to include controls for expected transgene expression patterns because of the potential for transvection interactions when using multiple transgenes in zebrafish.