Drosophila Mitotic Domain Boundaries as Cell Fate Boundaries

See allHide authors and affiliations

Science  08 Aug 1997:
Vol. 277, Issue 5327, pp. 825-828
DOI: 10.1126/science.277.5327.825


Fate determination in Drosophila embryos is evidenced by the appearance of mitotic domains. To identify fate or fates of cells, individual cells in mitotic domains 2, 8, and 15 were marked and monitored through development. Comparison of the different fates indicated that domain boundaries are cell fate boundaries. Cells were marked by expression of GAL4-dependent transgenes after photoactivation of a caged GAL4VP16 analog that had its DNA binding activity inhibited with a photolabile blocking reagent. Caged GAL4VP16 was also used to induce gene expression in Xenopus embryos. Thus, photoactivated gene expression is a versatile tool for spatiotemporal control of gene expression.

To control the temporal and spatial expression of selected genes at the single-cell level for the purpose of fate mapping and genetic manipulation, we devised a method for “caging” the DNA binding activity of GAL4VP16, a potent transcriptional activator. Caging is a form of photo-reversible chemical modification that has been used in the light-mediated activation of molecules such as adenosine 5′-triphosphate, Ca2+-chelators, and actin (1). Caged GAL4VP16 was produced by modifying lysine residues of purified GAL4VP16 (2) with the amine-reactive compound 6-nitroveratrylchloroformate (NVOC-Cl) (3). GAL4VP16 DNA binding activity was abolished after a 30-min incubation with 2 mM NVOC-Cl under mildly basic conditions (Fig.1A). More than 50% of the initial binding activity was recovered by irradiating the caged GAL4VP16 with a low-intensity, long-wavelength (365 nm) ultraviolet (UV) lamp.

Figure 1

(A) Caging of lysines inhibits DNA binding activity of GAL4VP16, but irradiation partially reverses this inhibition. GAL4VP16 was modified with different concentrations of NVOC-Cl, incubated for 15 min with a radiolabeled, double-stranded (ds), 19-nucleotide DNA containing the GAL4 consensus sequence (25), and run on a 16% polyacrylamide gel. Irradiation of the 2 mM samples occurred at 4°C. The percentage of dsDNA bound byGAL4VP16 is shown. Quantitation was done by fluorimaging (Ambis). (B) In vivo assay for transcriptional activity of GAL4VP16. Caged or unmodified GAL4VP16 was coinjected with a fluorogenic β-Gal substrate (RGPEG) into embryos of aUASGlacZ strain, and the changes in whole-embryo fluorescence were recorded by time-lapse recordings with a cooled charge-coupled device (CCD) camera mounted on a fluorescence microscope. The fluorescence intensity units (IU) represent the CCD output values scaled according to the neutral density filter used in recording the image (26). Error bars indicate the standard deviation. Within 15 min after injection, embryos injected with caged GAL4VP16 were irradiated with light of a mercury lamp passing through a DAPI (4′,6′-diamidino-2-phenylindole) filter. (□) Average (n = 6) of embryos injected with unmodified GAL4VP16, (•) caged GAL4VP16 irradiated (n = 6), (○) caged GAL4VP16 not irradiated (n = 4), and (▴) RGPEG only (n = 4). (C) Immunostain after photoactivation of UASGlacZ in a single cell of the amnioserosa. Arrowhead points to the labeled cell. Because the level of lacZ expression in the amnioserosa was generally weaker than in other tissues, the contrast of the image was enhanced to highlight the stained cell. (Inset) Enlargement of the stained cell. (D) Single-cell photoactivation ofUASGlacZ in the dorsal epidermis. The embryo was fixed at stage 13, which occurs after the normal onset of apoptosis. (Inset) Enlargement of the eight-cell cluster with an outline of the individual cells.

Caging of GAL4VP16 with 0.5 mM NVOC-Cl, which modified about 8 of the 14 GAL4VP16 lysines (4), completely inhibited in vivo transcriptional activation in Drosophila embryos (5). This level of caging did not affect GAL4VP16 DNA binding activity in vitro (Fig. 1A). It is not known why the lower level of caging inhibited in vivo activity (6). Inhibition of the transcriptional activity of caged GAL4VP16 could be reversed in vivo with 365-nm light from a 100-W mercury lamp shone through a microscope objective via the epi-fluorescence light path of a standard inverted microscope. Experiments with Drosophila embryos required 3 to 4 s of irradiation (7) for maximal photoactivation.

We determined the efficiency of GAL4VP16-mediated photoactivated gene expression by quantitating the fluorescence of coinjected RGPEG (8), a fluorogenic β-galactosidase (β-Gal) substrate, in embryos that contained a GAL4-dependent lacZ construct (UASGlacZ) (Fig. 1B). GAL4VP16 was usually injected at a concentration of 0.2 mg/ml or less (9). Concentrations of unmodified or caged GAL4VP16 greater than 0.4 mg/ml caused developmental defects. This may have resulted from squelching, where general transcription factors bound to the acidic domain of unbound GAL4VP16 (10). Injection of RGPEG alone or with caged GAL4VP16, but not followed by irradiation, gave curves that initially decreased to a constant, low level of fluorescence. This decrease may have been due to quenching of the fluorescence of trace-free resorufin in the RGPEG preparation by free thiols in the embryo such as glutathione (11). Irradiation of control embryos injected with RGPEG alone did not cause any increase in fluorescence (12). Injection of unmodified GAL4VP16 showed a marked increase in RGPEG fluorescence, indicating the induction of lacZ expression. Irradiation of embryos after injection of caged GAL4VP16 gave a response that was 74% of the unmodified GAL4VP16. Caged GAL4VP16 was photoactivated as early as nuclear cycle 12 and up to 4 hours after gastrulation at 25°C (12). Most of the irradiated embryos developed into normal stage 17 embryos. Whole-embryo irradiation did not alter cell death patterns as seen by time-lapse fluorescence microscopy of embryos coinjected with acridine orange to monitor apoptosis (13) and either GAL4VP16 or buffer (12).

To photoactivate gene expression in single cells, we narrowed the UV beam to about 5 to 10 μm by inserting a pinhole aperture at the field-stop position of the epi-fluorescence light path of the microscope (14). To show that only single cells were activated, we irradiated amnioserosa cells, which do not divide after gastrulation, at the start of gastrulation. Irradiation was restricted to single cells in more than 90% of the attempts; the remainder had two activated cells (Fig. 1C). Photoactivated gene expression did not perturb cellular division or induce cell death. Dorsal epidermis cells, which undergo three rounds of postblastoderm divisions, were irradiated at the start of gastrulation. By stage 13, which is after the normal onset of programmed cell death at stage 12, eight labeled cells could be observed (Fig. 1D).

Mitotic domains are believed to be indicators of cell fate (15). DiI labeling has been used to show that cells from mitotic domain 14 give rise to a limited set of neurons (16). To test the mitotic domain hypothesis, one must determine the fates adopted by cells of adjacent mitotic domains and show that different mitotic domains produce discrete sets of fates. We compared the fates of cells derived from mitotic domains 2 (∂2), 8 (∂8), and 15 (∂15) (17), which are adjacent domains in the anterior ventral region of the embryo (Fig.2A).

Figure 2

Cell lineage tracing of cells in domains 2, 8, and 15 after photoactivation ofUASGlacZ. (A) Schematic diagram of the ventral mitotic domains in the head primordium. (B) ∂2: Single-cell activation labeled cells of the hypopharyngeal epithelium (arrowhead) and associated muscle cells. (C) ∂2: Single-cell activation marked the hypopharyngeal epithelium (arrow-head) and macrophages. (D) Single-cell activation in ∂8 labeled a cluster of cells in the anterior midgut. (E) Activa tion of a single ∂15 cell lateral to ∂8 stained endodermal tissue of the esophagus and proventriculus. The line highlights the lumen of the proventriculus and of the esophagus.

Both single-cell and multiple-cell photoactivation in the head region of ∂2 primarily gave rise to the hypopharyngeal epithelium (floor of the pharynx), macrophages (or hemocytes), and a few cells associated with the pharynx that were provisionally identified as muscle cells. Single-cell activation produced subsets of the multiple-cell activation experiments. The pharyngeal cells often appeared in discrete patterns with associated muscle cells or macrophages (Fig. 2, B and C).

Single-cell irradiation of ∂8 primarily gave rise to endodermal cells that formed the anterior midgut epithelium (Fig. 2D). ∂15 was discerned from ∂8 in that ∂15 cells divide late and abut ∂2. Cells of ∂15 that resided posterior to ∂8 also gave rise to endodermal clones of the anterior midgut epithelium and appeared morphologically similar to ∂8 clones (12). It is not clear if cells from both domains intermingle as they migrate from their original location to reside in the endoderm. Irradiation of ∂15 cells that were located lateral to ∂8 stained endodermal cells of the esophagus and proventriculus (Fig. 2E). We have not examined the fates of ∂15 cells that are anterior to ∂8. Clones from both domains often formed clusters and appeared spatially restricted, suggesting a possible targeting mechanism. Whole ∂8 irradiation gave rise to two additional structures, dorsal pharyngeal muscles and visceral muscles of the esophagus and proventriculus. Irradiation of the entire domain labeled mesodermal cells underneath ∂8 because photoactivation of larger areas could not be focused solely to superficial cells and allowed penetration of irradiation to deeper cell layers. We confirmed this result by irradiating ∂10 mesodermal precursors before invagination. Those cells gave rise to the pharyngeal muscles and visceral muscles (12).

Fate mapping ∂2, ∂8, and ∂15 showed that there was no fate overlap between ∂2 and ∂15 and limited overlap between ∂15 and ∂8. Clones from ∂8 and ∂15 that contributed to the anterior midgut epithelium could not be discerned by morphology; however, there may be a physiological difference. Also, ∂15 gave rise to clones in the foregut, which were not seen in ∂8. It has been suggested that ∂15 is distinct from ∂8, rather than being a late-dividing component of ∂8, because these cells divide in the plane of the epithelium, not perpendicularly like ∂8. Our results support this conclusion.

It is expected that cells from different regions of the embryo give rise to different tissues and structures. However, the spatial arrangement of domains ∂2, ∂8, and ∂15 allowed fate mapping of cells that were immediate neighbors on the embryonic surface. It is therefore unlikely that the consistent fate differences from one cell to the next are merely a reflection of the overall location. Thus, the mitotic domain hypothesis is correct: Cells within a particular domain are destined to assume specific fates. Mitotic domain boundaries can therefore be viewed as cell fate boundaries, at least for the mitotic domains described here.

Photoactivated gene expression was also used to induce ectopic expression of a GAL4-dependent UbxIa transgene (UASGUBX). Compared to unirradiated embryos (Fig. 3A), the irradiated embryos showed ectopic UBX expression at the site of irradiation (Fig. 3B). Overexpression of UBX in 4- to 7-hour embryos repressesAntennapedia (Antp) transcription (18). Unilateral photoactivation of UBX repressed ANTP expression on one side of the embryo, indicating that the photo-induced UBX was functional (Fig. 3C).

Figure 3

Immunostains for UBX (A andB) or ANTP (C) after photoactivation ofUASGUbxIa at the onset of gastrulation. (A) Control. Injection of caged GAL4VP16, no irradiation. (B) Injection of caged GAL4VP16 irradiated anteriorly. (C)UASGUbxIa embryo stained for ANTP; unilateral photoactivation of UBX repressed Antptranscription only in one-half of the embryo. The approximate area of irradiation is indicated by the dotted line.

To ectopically activate GAL4-dependent transgenes inXenopus, we injected caged or unmodified GAL4VP16 intoXenopus embryos at the four-cell stage together with a reporter plasmid that carried a GAL4-dependent green fluorescent protein (GFP) construct (19). Injection of unmodified GAL4VP16 produced high levels of GFP expression (Fig.4A) that was still visible at stage 31/32. Embryos that were injected with caged GAL4VP16 but not irradiated did not exhibit GFP fluorescence (Fig. 4B); this result was similar to that for embryos that were injected with reporter plasmid alone (12). Injection of caged GAL4VP16 followed by irradiation at stage 9 strongly activated expression of GFP, with expression levels comparable to that for unmodified GAL4VP16 (Fig. 4C).

Figure 4

Photoactivated gene expression inXenopus . (A) Coinjection of unmodified GAL4VP16 and reporter plasmid caused high levels of GFP fluorescence. (B) Coinjection of caged GAL4VP16 and reporter plasmid, no irradiation. (C) Coinjection of caged GAL4VP16 and reporter plasmid followed by irradiation.

Genetic dissection of development requires both removal and ectopic addition of gene activity. For ectopic gene expression inDrosophila, researchers commonly use transgenic heat-shock constructs (20) or the GAL4 enhancer trap system (21). Also, localized heating of cells with a laser was used to activate the expression of heatshock–responsive transgenes in single cells in Drosophila and Caenorhabditis elegans (22). Photoactivated gene expression complements those two methods in that it extends the window of inducible gene expression to earlier developmental stages and provides a reliable method for spatiotemporal control of ectopic gene expression. The caged GAL4VP16 expression system has the added benefit that localized transient activation of plasmid-borne genes, as was shown in Xenopus embryos, may be used in other organisms that are less amenable to genetic manipulation and may increase the throughput of testing different DNA constructs. The main impetus for developing the caged GAL4VP16 photoactivated gene expression system was to provide a method for single-cell fate mapping and ectopic gene expression. Using commercially available reagents and standard biochemical methods as well as standard embryo-preparation techniques, we established a system for light-induced gene expression inDrosophila and Xenopus embryos. Caged GAL4VP16 can be prepared in a few hours from the purified protein stock and is stable for more than 1 year at −80°C. Typically, we processed about 150 embryos per day.

We have demonstrated that the mitotic domain hypothesis is correct in that mitotic domains are indicators of cell fate. The fates of the cells arising from different mitotic domains had been based on existing fate maps that were generated by visual inspection of unmarked living or fixed embryos. These predictions were confirmed for the domains analyzed thus far. However, direct marking of these cells revealed much more detail about the actual array of fates.

  • * To whom correspondence should be addressed. E-mail: minden{at}


View Abstract

Stay Connected to Science

Navigate This Article