Worm Breeder's Gazette 12(4): 20 (October 1, 1992)

These abstracts should not be cited in bibliographies. Material contained herein should be treated as personal communication and should be cited as such only with the consent of the author.

Promoter Trapping: Results of a Pilot Screen and Isolation of Two Embryonically Active Fusions

Geraldine Seydoux, Andy Fire

Carnegie Institution, 115 West University Parkway, Baltimore, MD 21210

We have started a screen for expression patterns in C. elegans embryos. This screen is based on the "promoter trapping" method developed by Ian Hope (1), which we modified to allow the screening of a large number of potential gene fusions in a minimum amount of time. In brief, potential gene fusions, created in vitro by fusing genomic DNA fragments to a promoterless lacZ gene, are transformed into E. coli to create a library. Together with rol-6 (d) DNA (2), DNA from the entire library is injected in wild-type hermaphrodites as a single mix to create Roller lines carrying extrachromosomal arrays containing 100 200 potential gene fusions. Established lines are then stained for ß-galactosidase activity. This method eliminates handling of individual bacterial clones prior to the identification of a pattern of interest, and allows a large number of lines to be generated from a single injection mix. Once a pattern of interest is observed, the active fusion construct responsible for this expression pattern is retrieved directly from the ß-galactosidase-expressing line by "plasmid rescue" in E. coli. First, genomic DNA from the ß-galactosidase-expressing line is digested with a restriction enzyme that cuts once in each fusion construct. The digested DNA is then religated and used to transform E. coli to create a mini-library. Clones from this mini-library are then injected in pools into hermaphrodites until the pattern of interest is regenerated and the active construct is identified within the pool.

We have constructed 4 libraries, each based on a different ß-galactosidase-containing vector. These vectors differ from each other in sequences located 5' to lacZ. pGS 3.01 contains a splice acceptor site, followed by an AUG and nuclear localization sequence (AUG-NLS cassette) fused in frame to lacZ (Library 140; 20,000 clones); pPD 21.28 contains a small synthetic intron (with both splice acceptor and splice donor sites) followed by the AUG-NLS cassette (Library 141; 7,500 clones); pPD 16.51 contains the AUG-NLS cassette alone (Library 142; 8,800 clones); and pPD 34.110 contains a transmembrane domain (with no AUG) (Library 143; 7,500 clones). Library 140 should allow B-galactosidase expression from splice fusions, Libraries 141 and 142 from transcriptional and translational fusions, and Library 143 from translational fusions only.

To date we have analyzed a total of 283 lines (corresponding to about 40,000 clones) from the 4 libraries. No significant differences were observed in the frequency of staining among the 4 libraries. 57% of all lines stained in the pharynx. This high incidence of pharyngeal staining was also observed by Ian Hope in his screen and could be due to the presence of a putative pharyngeal enhancer in the vectors used to construct the libraries (1). 53% of all lines stained in tissues other than the pharynx. Examples of staining were observed at all stages from embryos to adults. Staining frequencies in specific tissues were as follows: 33% of all lines stained in neurons (with patterns ranging from a few specific neurons to many neurons), 6% in hypodermal cells, 3% in muscle, 3% in the somatic gonad, 2% in the intestine and/or associated valves, 3% in specific head cells, and 3% in embryos. Three lines (PD9412, PD9419 , PD9431 )gave distinct staining patterns in 200 cell embryos with occasional staining in younger embryos.

Following the method outlined above, we have identified the active fusion constructs in PD9412 and PD9431 .In both cases, when the active construct was purified away from the inactive constructs present in the original array, its expression pattern became stronger and more widely distributed. In particular, lines containing multiple copies of the purified active construct from PD9412 showed a broad pattern of expression in embryos, larvae and adults. Partial sequencing of the active clone revealed that it encodes a splice fusion between the first exon of C. elegans translation factor eIF4 -A(3)and lacZ.

In contrast, the purified active construct from PD9431 shows a pattern of expression restricted to the embryo. Preliminary observations suggest that this fusion is initially expressed in the cytoplasm of most cells in the early embryo (16-28 cell stage), and becomes progressively restricted to the nucleus of a few hypodermal cells by the comma stage. We are currently making lines using spe-26 DNA as a marker to confirm that this expression pattern is not due to rol-6 (d).

Our preliminary results suggest one potential difficulty with the screen: Expression patterns observed in the original lines can change as the active construct is purified. Although the initial screen provides candidates for embryonically expressed constructs, identification of the complete pattern must await purification of the active construct. This problem may be avoided if smaller pools of potential fusions are used to generate the initial lines.

Literature Cited:

1. Hope, I. A. (1991). Development 113, 399408.

2. Mello et al. (1991). EMBO 10, 3959-3970.

3. Roussell D., and Bennett, K. L. (1992) NAR 20, 3783.