Frequent Unanticipated Alleles of lethal giant larvae in Drosophila Second Chromosome Stocks

Abstract

Forty years ago, a high frequency of lethal giant larvae (lgl) alleles in wild populations of Drosophila melanogaster was reported. This locus has been intensively studied for its roles in epithelial polarity, asymmetric neural divisions, and restriction of tissue proliferation. Here, we identify a high frequency of lgl alleles in the Bloomington second chromosome deficiency kit and the University of California at Los Angeles Bruinfly FRT40A-lethal P collection. These unrecognized aberrations confound the use of these workhorse collections for phenotypic screening or genetic mapping. In addition, we determined that independent alleles of insensitive, reported to affect asymmetric cell divisions during sensory organ development, carry lgl deletions that are responsible for the observed phenotypes. Taken together, these results encourage the routine testing of second chromosome stocks for second-site alleles of lgl.

STUDIES of lethal giant larvae [l(2)gl or lgl] provided founding genetic evidence for the existence of tumor suppressor genes. Originally isolated by Bridges in 1933 (Bridges and Brehme 1944), lgl mutant cells were later characterized to have potent malignant properties (Gateff and Schneiderman 1967). In the past decade, Lgl has received considerable attention for its varied functions in apico-basal polarity and epithelial maintenance (Manfruelli et al. 1996; Bilder et al. 2000; Bilder 2004), in asymmetric division of embryonic neuroblasts (Ohshiro et al. 2000; Peng et al. 2000), and during multiple asymmetric divisions in adult mechanosensory organ development (Justice et al. 2003; Langevin et al. 2005).

The frequency of mutant lgl alleles in wild populations is extraordinary: ∼3–6% of lethal second chromosomes extracted from wild flies collected in the former Soviet Union (Golubovsky 1978) and in California (Green and Shepherd 1979) failed to complement lgl. This high frequency is due to the fact that lgl is the second protein-coding gene downstream of the subtelomeric region of chromosome 2L, and many spontaneous lgl alleles actually represent terminal deletions of 2L (Mechler et al. 1985).

The high frequency of lgl alleles and terminal deletions of chromosome 2L are not well recognized among Drosophila researchers. We were led to appreciate this during three independent studies using popular public stock collections and previously published mutants, all of which revealed confounding second-site deletions uncovering lgl. Therefore, the possibility of 2L tip aberrations should be regularly considered when analyzing second chromosome stocks of Drosophila melanogaster.

The Bloomington deficiency kit contains frequent alleles of lgl:

The Bloomington deficiency kit is a tiling set of deletions that uncovers ∼95% of D. melanogaster euchromatin (http://flystocks.bio.indiana.edu/Browse/df-dp/dfkit.htm) and is commonly used to map lethal mutants. In addition, it is actively screened for zygotic phenotypes in homozygous deficiency embryos or for the ability of deficiencies to dominantly modify phenotypes of interest (e.g., Guo et al. 2003; Mason et al. 2004; Anderson et al. 2005; Norgate et al. 2007).

While screening the second chromosome deficiency kit for loci required for embryonic neuroblast asymmetry, we found that Df(2R)en-A and Df(2R)CB21 homozygotes failed to localize Miranda asymmetrically in dividing neuroblasts (supporting information, File S1, and data not shown). However, we eventually determined that these stocks, which contain cytologically visible deletions in the right arm of chromosome 2, were also deleted for lgl (Figure 1), whose loss underlies their neuroblast defects.

Figure 1.—

The Bloomington second chromosome deficiency kit contains a high frequency of lgl alleles. The mapped deficiencies that fail to complement the null allele lgl[4] are shown; these include deficiencies on both the left and right arms of chromosome 2. Df(2R)ST1-Df(2R)cn9 and Df(2R)P34-Df(2R)BSC26 are pairs of deficiency stocks that delete partially overlapping sequence.

These findings prompted us to systematically test the second chromosome deficiency kit for failure to complement lgl[4], a characterized null allele. Thirteen of 98 stocks were indeed mutant for lgl, comprising deficiencies of diverse regions of both the left and right arms of the second chromosome (Figure 1 and Table S1). Because the affected stocks were collectively generated in several laboratories using different strategies, we infer that many additional deficiencies not included in the standard kit are contaminated with lgl mutations and/or 2L terminal deletions.

The University of California at Los Angeles Bruinfly FRT40A-lethal P collection contains frequent alleles of lgl:

Collections of lethal stocks bearing randomly inserted transposable elements have been valuable tools for genetic screening (St Johnston 2002; Bellen et al. 2004). Recently, the University of California at Los Angeles Undergraduate Research Consortium in Functional Genetics recombined large collections of lethal P elements onto FRT chromosomes (Chen et al. 2005; Call et al. 2007), thus enabling their use in clonal screens.

We investigated whether these stocks might uncover novel maternally provided genes involved in cuticle patterning. In screening 2L Bruinfly stocks, we identified several that produced cuticular defects in germline clones resembling cobblestones (Figure 2). This phenotype is characteristic of loss-of-function mutations in the neoplastic tumor suppressor genes scribble, discs large, and lgl (Bilder et al. 2000; Tanentzapf and Tepass 2003). Suspicious of the high rate of lgl phenocopy, we tested whether these stocks harbored alleles of lgl and indeed found several lines that failed to complement lgl[4] (Figure 2 and Table S2).

Figure 2.—

The Bruinfly FRT40A collection contains a high frequency of lgl alleles. (Top) Cuticle preparations from wild-type (A), lgl[4] germline clones (B), and Kruppel-Homolog 1 (Kr-H1) (Bruinfly 13097) germline clones (C). Loss of maternal and zygotic lgl disrupts cuticle development, and the mimic of this phenotype in the Kr-H1 line is due to loss of lgl. (Bottom) Five of 24 randomly selected FRT40A-lethal P stocks, containing transposons distributed over the entirety of chromosome 2L, harbored lgl alleles.

The Bruinfly consortium performed excision experiments to ask whether lethality mapped to the P elements (Chen et al. 2005; Call et al. 2007), and as might be expected, many stocks were not revertible. Curiously, some stocks that generated lgl-like embryos were restored to viability upon P excision and complemented lgl[4] in our initial tests. A potential explanation is that such stocks are composed of mixed populations of second chromosomes. To test this, we repeated lgl complementation tests as cohorts of single-fly crosses from each of 24 randomly selected Bruinfly 2L stocks. Five stocks contained lgl alleles, but two of these were indeed mixed with respect to 2L tip status (Table S2). Therefore, conventional complementation/excision tests using multiple fly crosses can falsely attribute phenotypes to the transposon insertion and/or report incorrectly on the integrity of the second chromosome.

Previously described alleles of insensitive are deficient for lgl:

Drosophila external mechanosensory organs are multicellular structures that develop via asymmetric divisions of individual sensory organ precursors (SOPs) (Lai 2004). The Notch-signaling pathway generates asymmetry at each division, yielding four distinct cells in the mature organ. The shaft and socket cells are visible from the exterior and have large nuclei, while the sheath and neuron are strictly internal and have smaller nuclei; a fifth glial cell undergoes apoptosis and is absent from normal organs. The complete loss of Notch activity yields sensory lineages that exclusively generate neurons, while the activation of Notch signaling in all cells of the sensory lineage yields organs composed of multiple sockets.

Insensitive (insv) was originally isolated from expression profiling as a SOP-specific transcript (Reeves and Posakony 2005). Two lethal (and presumably null) deletion alleles were described, insv[23B] and insv[23L], and both of these yielded multiple socket organs in adult thorax clones (Figure 3A). The similarity between insv clones and clones of Notch-inhibitory factors such as numb or lgl suggested that Insv is a novel factor that represses Notch activity in the sensory organ lineage (Justice et al. 2003; Langevin et al. 2005; Reeves and Posakony 2005).

Figure 3.—

The loss of lgl from insv mutant chromosomes is responsible for their reported multiple socket phenotypes. (A) Scanning electron micrograph of insv[23B] notum clones, which differentiate sensory organs with extra socket cells. The dotted circle highlights a four-socket organ, while the dotted square indicates an organ with two sockets and a malformed shaft. (B, B′) Cut staining in a 24-hr after puparium formation (APF) notum bearing insv[23B] clones, marked by the absence of nuclear GFP. Wild-type organs contain two large (arrowheads) and two small (arrows) Cut⁺ nuclei, while insv mutant organs contain four large Cut⁺ nuclei. (C, C′) Suppressor of Hairless [Su(H)] staining in a 24-hr APF notum bearing insv[23B] clones marked by the absence of nuclear GFP. Wild-type organs contain a single Su(H)⁺ cell (arrow), while insv mutant organs contain three to four Su(H)⁺ cells (arrowheads). (D) PCR tests demonstrated that the available insv stocks are actually doubly mutant for lgl (data not shown). Expression of UAS-H2B-RFP in lgl-insv double mutant MARCM clones does not affect the multiple socket phenotype (arrowheads). H2B-RFP is green, and cuticular structures including socket and shaft cells are red. (E) Expression of UAS-lgl-GFP (green) fully rescues the bristle phenotypes of lgl-insv double mutant clones with single sockets (arrowheads) and single shafts (arrows).

We observed that insv mutant sensory organs differentiated four cells with large nuclei (as marked by Cut), suggesting the conversion of inner cell fates into outer cell fates (Figure 3B). Indeed, insv mutant organs failed to differentiate sheath cells (marked by Prospero) or neurons (marked by Elav) and instead contained three to four Su(H)⁺ socket cells (Figure 3C and data not shown). Such cell fate transformations were highly reminiscent of lgl clones. Consistent with this, insv clones were deficient for Lgl protein. However, we were surprised to observe that Lgl was absent both in epithelial cells and in SOPs of the insv clones (data not shown), as insv is specifically expressed by SOPs (Reeves and Posakony 2005).

Because both insv and lgl are located on chromosome 2L, we considered whether insv alleles were deficient for lgl. We observed that insv stocks generated overgrown larvae with tumorous discs and demonstrated their loss of lgl using complementation and PCR tests (data not shown). Thus, the available chromosomes are doubly mutant for lgl and insv. We next used the MARCM system and Ubx-Flp to induce lgl and insv thorax clones for rescue assays. When UAS-Histone2B-RFP was activated in such clones, their sensory organs differentiated with multiple sockets (Figure 3D and File S1). However, expression of UAS-lgl-GFP in these clones completely restored normal external fates with single sockets and shafts in the mutant organs (Figure 3E). Therefore, the deletion of lgl accounts for the sensory organ defects produced by available insv chromosomes.

Conclusion:

We found that commonly used tool-kit stock collections and published second chromosome mutants harbor second-site terminal deletions of chromosome 2L, thereby removing lgl. As this was previously noted to affect a set of deficiency and lethal P stocks (Brumby et al. 2004; Mason et al. 2004; Menut et al. 2007) that are substantially nonoverlapping with the ones that we characterized, the 2L problem is undoubtedly broader than currently recognized.

Unanticipated lgl alleles confounded interpretation of the previously described insv mutant (Reeves and Posakony 2005). Similarly, the neoplastic tumors induced by wingless mutant cells blocked for apoptosis (Perez-Garijo et al. 2005) also appear to be due to the loss of lgl from the wingless mutant chromosome studied (G. Morata, personal communication). We even found that the numb[1] stock (available at the Bloomington Stock Center), which disrupts the prototypical asymmetric cell fate determinant in Drosophila (Uemura et al. 1989; Rhyu et al. 1994), is deficient for lgl.

The observation that lgl mutations exhibit dose-sensitive interactions with a variety of loci (Bilder et al. 2000; Tanentzapf and Tepass 2003; Zarnescu et al. 2005) suggests that unanticipated heterozygosity for lgl can confound dominant modifier screens. Notably, lgl was accidentally discovered to modify a crumbs gain-of-function phenotype via a kekkon P insertion chromosome that fortuitously carried an allele of lgl (Tanentzapf and Tepass 2003). Presumably, at least some other genetic interactions involving second-site lgl alleles have not been appropriately tracked down.

There are further consequences stemming from the fact that most spontaneous lgl alleles represent terminal deletions. Subtelomeric heterochromatic regions in Drosophila generate large numbers of Piwi-interacting RNAs (piRNAs), which mediate post-transcriptional control of transposons (Brennecke et al. 2007) and epigenetic transcriptional activation (Yin and Lin 2007). In addition to Lgl-regulated cuticle patterning, imaginal disc growth, embryonic neuroblast asymmetry, and adult sensory organ development, then, 2L terminal deletions may also affect piRNA-controlled processes, potentially including telomeric silencing (Mason et al. 2004). In conclusion, these observations suggest that Drosophila geneticists should routinely test second chromosome stocks to verify the absence of unanticipated lgl alleles.