" Vectorology II: Green fluorescent protein (or "It isn't easy being green") Andrew Fire, Joohong Ahnn, Geraldine Seydoux, and SiQun Xu Carnegie Institution of Washington, Baltimore, Md. 21210 Earlier this year, Marty Chalfie introduced the use of Aequorea victoria green fluorescent protein (gfp) as a reporter for following gene expression in live animals. To test the applicability of this reporter in screens for novel expression patterns during embryogenesis, we constructed gfp fusion constructs using several promoters with defined embryonic specificity. Results with lacZ and gfp fusions are compared in the table below: __ _
myo-3,
unc-54, ost-l: lacZ fusions for these genes mimic expression patterns for the endogenous myosin genes, with activity in body muscles beginning around the comma stage of embryogenesis (1,2). Equivalent gfp fusions are also active (i.e., cause uorescence) in body muscles, but expression does not begin until the embryos have ached three-fold elongation. hlh-l: The endogenous gene and lacZ fusions express most strongly in body-wall muscle precursors in cleaving and pre- comma embryos; expression continues in ifferentiated body- wall muscles (3). Equivalent gfp fusions are active only in the differentiated body-wall muscles. ey-l: Somatic expression of the endogenous gene and of lacZ fusions is strongest in arly embryos (where most or all somatic cells express); later expression with lacZ fusions has been seen most intensely in body muscles and pharynx (2). Equivalent gfp fusions are active in differentiated body muscle and in pharynx, but no early mbryonic activity was observed. es-10: The endogenous gene and lacZ fusions express well in somatic lineages in the arly embryo, with occasional low- level postembryonic expression in gut (4). Equivalent gfp fusions show only sporadic activity in differentiated gut. glp-l: expression of the endogenous gene occurs in germ line as well as some somatic issues, most notab~y a set of embryonic cells at the 40-60 cell-stage (5). glp-l :lacZ fusions show strong expression in 40-60 cell embryos and in the spermatheca; glp-l:gfp fusions are active in spermatheca (weakly) but no activity is seen in 40-60 cell embryos. Although the above data is encouraging for experiments that require GFP fluorescence terminally differentiated cells, the results suggest that activity might be blocked in early and mid-stage embryos. The ability to make active GFP in embryos appears to follow differentiation rather than morphogenesis (this was evident from an EMS mutant screen looking for ost- l::gfp activity in comma-stage embryos; the screen yielded many fluorescent embryos in which morphogenesis had arrested while differentiation continued). Activity of gfp constructs in pre-differentiation embryos could in principle be blocked at several mechanistic levels. Analysis of RNA transcripts argues against a transcriptional block: in situ hybridization was used to show that a pes- lO::gfp fusion transcript was produced at high levels in the early embryo, in a pattern indicative of the
pes-10 promoter. We hope to eventually examine the distribution of GF~ protein in these embryos using anti-GFP antibodies. In the interim, we have constructed a hybrid gfp-lacZ fusion gene. This construct produces co-localized B-gal activity and green fluorescence when expressed in differentiated pharyngeal muscle (although the fluorescence is somewhat less than with gfp alone). Expression of the gfp-lacZ fusion gene in early embryos using the
pes-10 promoter results in a product which has B-galactosidase activity but is not fluorescent. As a working hypothesis, we propose that the GFP protein might be made in early embryos but fail to acquire fluorescent properties. In terms of gene expression studies, this suggests some cautions. In particular, GF~ fluorescence could appear or disappear at a given point in development as a result of modulations in the formation of the fluorochrome rather than changes in the level of GFP protein. Our results suggest that this might be less problematic in postembryonic differentiated tissue, although it should be noted that we have not looked at any postembryonic non-differentiated tissues. We have taken several approaches toward producing a gfp vector for use in early embryos. A multiple-intron ~fp gene was constructed (see Fire and Xu article), but did not relieve the inhibition in early embryos. Mutageneses of ~fp fusion containing lines are in progress (in principle it might be possible to find a mutant version of ~fp or a mutant worm strain in which early acquisition of GFP fluorescence was not blocked). For some applications of gfp reporters, labeling of a specific molecule in its normal context is required, while for other applications the critical aim is to label the cells expressing the fusion. We thought that it might be possible to circumvent problems with gfp in the latter cases by targeting the produced protein to different intracellular compartments, potentially allowing a more amenable environment for fluorescence. In lines with gfp expressed with no other protein sequences attached, the fluorescent signal is uniformly distributed through the cytoplasm. We've examined the effects of several different targeting signals on GFP localization and embryonic fluorescence: Nuclear localization signal: Attachment of the SV40 nuclear localization signal (NLS) from our older lacZ vectors (6) results in fluorescence activity which is distinctly stronger in the nucleus than the cytoplasm. Curiously, the degree of nuclear loc~ization is greatly improved in fusions which contain
g7~7, the nuclear localization signal, and an appended mass of protein (lacZ and pieces of myosin and HLH- 1 are variably sufficient for this). The incomplete nuclear localization of the simple NLS-gfp protein may be due to the ability of this relatively small protein to diffuse out of the nucleus. Secretion signal: We have previously used a synthetic secretion signal to drive secretion of several different expressed molecules (e.g., 7). Secreted gfp constructs driven by the
myo-2,
pes-10 and
mec-7 promoters have been tested. The resulting fluorescence accumulation is both intracellular and extracellular. The intracellular staining with the
myo-2 promoter appeared reticular and might represent the internal secretion apparatus of the cell. Interestingly, coeloemocytes appeared to scavenge secreted gfp expressed postembryonically from pes-l O or mec- 7, storing the fluorescent material in internal vacuoles or droplets. Mitochondrial matrix localization signal: An N-terminal mitochondrial matrix localization signal from chicken mitochondrial aspartate aminotransferase (8) was synthesized and incorporated into the ~fp expression cassette. This signal was sufficient to localize gfp to mitochondrial structures in body wall muscle. (This signal might also be useful for other non-GFP experiments;
ced-9 anyone?). Preliminary results in pre-morphogenesis embryos have been most encouraging with mitochondrial-localized ~ffp (M~p). Fluorescence activity from hlh-l :M~p has been seen in lima bean-stage embryos that would not be fluorescent with the cytoplasmic GFP. So far, no fluorescent signals have been seen in premorphogenesis embryos with the nuclear or secreted GFP's. Other GFP notes: a) As reported by Marty, anesthesia with azide or propylene phenoxitol produces rapid fading of fluorescent signal. We have been mounting animals in lmM levamisole, which yields much more stable fluorescence in the immobilized animals . b) We've now made a large number of protein fusions with GFP on the C-terminus. All of these have worked at some level, suggesting that gfp, like lacZ, is relatively tolerant to being appended to the C-terminus of other proteins. One fusion with GFP on the N terminus (the ~fp::lacZ fusion) functioned, but activity may be less than simple GFP. Refs: 1. J. Schwartzbauer, pers comm. 2. V. Plunger, pers. comm. 3. Krause et al., Cell 63, 907. 4. Seydoux & Fire,wbg 13#3, 33. 5. Seydoux & Fire, Development, in press. 6. Fire et al., Gene 93, 189 7. Perry et al., Genes & Dev. 7, 216 8. Jausi et al., J. Biol. Chem. 260, 16060. Thanks to Marty Chalfie (gfp), Verena Plunger (cey-l), Jim Henson (Kerrnit).