In the hopes of identifying members of the nuclear hormone receptor gene family in C. elegans we screened two cosmid libraries (MIT and MRC C-series) with degenerate oligonucleotide probes encoding a highly conserved motif within the zinc finger region of hormone receptors ( and not found in the TFIIIA class of zinc fingers). Two separate sets of oligos were used, one encoding the amino acid sequence CGSCKVFFK ( glucocorticoid receptor class) and the other encoding CE(G/A)CK(G/A)FF (estrogen/thyroid hormone receptor class). This screen identified 63 cosmids which fall into 50 loci on the physical map. Sequence information has been obtained for the hybridizing region of cosmids from nine of these loci; the hybridization signal strengths for the fragments sequenced range from weak to very strong. Only one of these has proved to encode an authentic zinc finger domain. This high percentage of false positives appears to be due at least in part to the high AT content of our oligo probes, especially at the two adjoining phenylalanine codons. Although the probes behaved as predicted on control blots, when used to probe worm genomic libraries the oligos have hybridized to what are most likely intron sequences containing runs of A's and T's, at times bridging two regions of partial homology separated by one or two base pairs. To separate likely candidates for additional authentic hormone receptor genes from other false positives we are further characterizing our collection of cosmids by the following approaches: hybridization with oligos homologous to other regions of the zinc finger domain; low stringency hybridization with known hormone receptor genes; hybridization with hot cDNA prepared from poly-A+ RNA in order to identify transcribed regions; and hybridization with C. briggsae DNA to identify conserved (and we hope exonic) sequences. In addition we are screening cDNA libraries with the oligos, both by hybridization and with PCR. While genes which are expressed at low levels may be missed in cDNA screens, this approach should avoid the problem of spurious hybridization to intron sequences. The one authentic zinc finger gene, which we have called
crf-2 (C. elegans receptor finger) is contained in cosmid AD7. This gene has also been identified by Barry Honda (Simon Fraser University) in a similar screen using a slightly different set of degenerate oligos. The predicted amino acid sequence of the zinc finger domain is shown in the figure aligned with the mouse thyroid hormone receptor, the most closely related known hormone receptor. The two sequences are 50% identical in the region shown. The
crf-2 gene has not been sequenced completely, but so far we have not detected any homology to known hormone receptors outside the zinc finger domain. Developmental Northern blot analysis detects expression of a 1.6 kb
crf-2 RNA in embryos, but not in L1 through L4 and at only very low levels in early adult hermaphrodites. The expression observed in adults could well be due solely to early embryos contained within the gonads. Whether crf- 2 transcription occurs maternally or zygotically or both remains to be determined, but in situ hybridization experiments (see Greenstein and Ruvkun, WBG, this issue) indicate that
crf-2 RNA is present primarily during the first half of embryogenesis. The physical map location of
crf-2 places it on chromosome I, between
dpy-5 and
dpy-14. Efforts are currently underway to map
crf-2 relative to the lethal complementation groups in this region (with the help of generous gifts of strains from the Ann Rose lab), as well as to characterize more completely its gene structure and expression patterns. [See Figure 1]