Mapping and identification of essential gene functions on the X chromosome of Drosophila

Abstract

The Drosophila melanogaster genome consists of four chromosomes that contain 165 Mb of DNA, 120 Mb of which are euchromatic. The two Drosophila Genome Projects, in collaboration with Celera Genomics Systems, have sequenced the genome, complementing the previously established physical and genetic maps. In addition, the Berkeley Drosophila Genome Project has undertaken large-scale functional analysis based on mutagenesis by transposable P element insertions into autosomes. Here, we present a large-scale P element insertion screen for vital gene functions and a BAC tiling map for the X chromosome. A collection of 501 X-chromosomal P element insertion lines was used to map essential genes cytogenetically and to establish short sequence tags (STSs) linking the insertion sites to the genome. The distribution of the P element integration sites, the identified genes and transcription units as well as the expression patterns of the P-element-tagged enhancers is described and discussed.

INTRODUCTION

Drosophila melanogaster has been a model organism for genome research for almost 90 years, starting with Sturtevant’s demonstration that recombination frequencies could be used to map genes in a linear order along the chromosome (Sturtevant, 1913). Physical maps that relate genes to physical sites on the chromosomes (Bridges, 1935), together with the precise mapping of cloned DNA fragments by in situ hybridization to polytene chromosomes, enabled the first chromosomal maps of cloned genomic DNA segments (Wensink et al., 1974). These achievements, the development of procedures for molecular screening (Grunstein and Hogness, 1975) and the assembly of large chromosomal contigs (Bender et al., 1983) have facilitated systematic attempts to construct P1-based physical maps of the Drosophila genome. These maps cover ∼90% of the euchromatic genome and are punctuated by short sequence tag (STS) sites that are, on average, 50 kb apart (Kimmerly et al., 1996). More recently, a whole genome shotgun sequencing project, a collaborative effort between Celera Genomics and the Berkeley Drosophila Genome Project (BDGP), has provided a 95% coverage of the Drosophila genome (Adams et al., 2000; Rubin et al., 2000b). In addition, expressed sequence tag (EST) projects have generated sequence information from >100 000 transcripts (Rubin et al., 2000a).

To assign function to the DNA sequences, the BDGP has undertaken large-scale functional analysis of the genome through gene-disruption projects that are based on mutagenesis by transposable P element insertion. These projects have focused on the isolation of essential genes on autosomes. The current collection of strains disrupt at least 1200 different genes, representing ∼30% of the estimated 3600 genes on the autosomes that are mutable to easily scorable phenotypes, mainly sterility or lethality (Spradling et al., 1995, 1999). Here, we report a large-scale P element insertion screen to identify essential genes on the X chromosome, as well as the construction of an X-chromosomal BAC tiling map, which complements other ongoing studies of the Drosophila genome.

RESULTS AND DISCUSSION

Tiling map of the X chromosome

A physical map of the X chromosome was constructed by hybridizing 521 probes to high-density filters bearing both the EDGP and BDGP BAC libraries. 5771 BAC clones were identified by hybridization. Eleven contigs were assembled using a combination of manual analysis of hybridization data and in situ hybridization. These span the X chromosome from the subtelomeric region of 1A to 20C near the base of the chromosome arm (Figure 1; supplementary information available on the website http://www2.open.ac.uk/biology/research/molecular-genetics/EDGPmap.html). The longest contig, contig 1, spans divisions 1–6 and comprises both cosmid clones (Siden-Kiamos et al., 1990) and BAC clones. Divisions 1–3 were mapped to yield template clones for sequencing (Benos et al., 2000, 2001), and in this region the BAC libraries were screened principally to link cosmid contigs.

Fig. 1. Diagram showing the contig distribution along the X chromosome. Contigs are indicated by the boxes above the schematic representation of the X chromosome. Contig 1 is composed of cosmids and BACs.

Probe sequences were selected from several sources. The EDGP cosmid mapping project determined many sequence tags from the termini of cosmid clone inserts. These sequence tags correspond to cosmid clones of known cytological location, with apparently single-copy inserts that are members of contigs (Madueno et al., 1995). Some additional tags were determined from isolated cosmid clones of known cytological location. Additional probes were derived from sequences reported in databases, generally corresponding to genes of known cytological map position and from sequences of P element insertions, again of known cytological position. Probes used were selected partly on the basis of cytological position, in order to enhance the distribution of probe sequences along the chromosome.

Example for inter-contig gaps

Two probes (ES00262, mapping to 7B1-2, and ES00263, mapping to polytene chromosome bands 7B1-6) fail to hybridize to BAC clones from either library. Both probes are derived from mapped P1 clones. No BAC clones hybridizing to probes on either side of ES00262 and ES00263 were observed. The explanation may be that the sequences from which these probes were derived are incorrect or that there may be polymorphism between the Drosophila strains used in the construction of the BAC and P1 libraries. This gap represents a separation between contig 1 (1A to 7A) and contig 2 (7B to 10A). Other gaps generally correspond to intensely staining polytene chromosome bands (e.g. those in 10B, 11CD, 13A and 17A) and presumably represent long regions with few probes. In a number of cases, clones repeatedly failed to give signals with probes to which, on the basis of cytological location and other hybridization data, they might be expected to hybridize. Possible explanations include clones being absent on certain filters, cloned DNA segments bearing segmental deletion, and other failures in hybridization.

Generation of P element insertions

The euchromatin of the Drosophila X chromosome comprises ∼22 Mb of DNA, containing 2182 predicted protein coding genes (Adams et al., 2000). As an estimated two-thirds of all Drosophila genes show no obvious loss-of-function phenotype (Miklos and Rubin, 1996), ∼800 of the X-chromosomal genes are mutable to a scorable phenotype such as sterility or lethality. In order to assess such genes in a systematic manner, we generated a collection of lethal P element insertions on the X chromosome.

The crossing scheme for generating X-chromosomal P{lacW} insertion lines is depicted in Figure 2. Females with a reinsertion of the P{lacW} element, as shown by the absence of the dominant second chromosome marker Curly and the white⁺ eye colour from P{lacW}, were individually mated to FM7c males. 57 105 such crosses were initially set and 39 900 (69.9%) produced offspring. From these crosses, 501 strains (1.3%) were derived that were either hemizygously lethal or semilethal (defined as <20% viability compared to the sibling balancer males). Taking into account that the X chromosome represents roughly one-seventh of the genomic target size, 1.3% corresponds to a frequency of ∼9% lethal P element insertions, a rate similar to those reported for autosomal insertions with comparable P element vector (Cooley et al., 1988; Bellen, 1999). In addition, 73 lines were isolated that carried a visible mutation or that were sterile in one sex.

Fig. 2. Crossing scheme for isolating lethal X-chromosomal insertion lines. The sex chromosomes (left pair in each genotype) and the two large autosomes are shown schematically with females on the left. Relevant mutations are labelled accordingly. The P{lacW} insertion is indicated by the triangle and its white⁺ selectable marker is represented by the darker chromosome. The transposase source is indicated by the double-headed arrow. Balancer chromosomes are hatched. Absence of the crossed-out males in the F₃ generation indicates a potential lethal P{lacW} insertion. For further details, see Methods.

Distribution of P element insertions

The insertion sites of 401 strains were determined by in situ hybridization on polytene chromosome squashes. In addition, DNA from the insertion sites was isolated either by plasmid rescue or by inverse PCR to generate an STS for all the insertion lines. Figure 3 summarizes the distribution of the P integrations along the X chromosome, showing a higher frequency of P{lacW} insertions in regions closer to the telomere. An observation that remains valid even when considering only non-redundant insertions. This skewed distribution is not mirrored by the relatively even distribution of transcription units along the X chromosome. Hence, it is likely that an inherent insertion site preference of P elements (Liao et al., 2000) is responsible for the observed distribution.

Fig. 3. P{lacW} distribution along the X chromosome. The localization is based on the molecularly defined integration site. Open bars represent all insertions in a given polytene region whereas shaded bars indicate the number of genes affected.

The quality of a P element collection depends largely on the proportion of strains that harbour single P{lacW} insertions. In situ hybridizations to polytene chromosomes were carried out for 401 strains. 89.5% (359 lines) contain single insertions, 9.7% (39 lines) contain two insertions and 0.7% (three lines) contain three P{lacW} insertions. However, for those 14 strains that showed a second hybridization signal in polytene region 3A, we were unable to confirm the second insertion site by molecular means. Conversely, molecular analysis identified eight strains that contain a second integration site and two containing three integrations where only one in situ hybridization signal was observed. In eight such cases, the insertion sites are <50 kb apart, and thus likely to be below the limit of resolution of the in situ hybridization technique. In four of these closely linked strains, the same transcription unit was hit by two P elements, indicative of local hopping during the remobilization of the starting P{lacW}. Collectively, these results indicate that >90% of the lines contain single P element insertions. Hence, this collection of P element insertion strains is comparable with the best P element collections generated to date (Spradling et al., 1999).

P{lacW} insertion and an associated lethal phenotype can be causally linked if excision of the P{lacW} restores wild type function. We tested a total of 206 lines by combining a transposase source P{ry⁺ Δ2-3}(99B) with the the P{lacW} element (Bellen et al., 1989). 134 strains (65.0%) produced fully viable male progeny that lost the white⁺ gene, suggesting that the P element excised precisely. The 65% value should be regarded as a minimum, since the reversion tests were performed with only 10 females bearing both the P{lacW} insertion and the transposase source. Some P elements are difficult to excise and hence exhibit low mobilization frequencies and will almost certainly have been missed in this type of protocol.

Characterization of the lines

As the β-galactosidase gene in P{lacw} can report the expression pattern of adjacent genes, we immunohistochemically stained embryos with anti-β-gal antibody. 398 of the 501 strains (79.4%) expressed the enzyme. This is considerably higher than the 64% reported after random autosomal insertion of the same P element (Bier et al., 1989) and may result from the fact that the X-chromosomal lines are selected for insertions affecting essential genes. Representative patterns for each line were deposited in the FlyView database, http://flyview.uni-muenster.de/ (Janning, 1997), where they can be searched by various criteria.

The lethal phase was determined in 497 of the 501 lines. The most frequent stage where development stops is the larval stage, with 29.2% of the analyzed lines (see Supplementary data). This is not unexpected, since homozygous mutants can often survive up to this stage by relying on maternal contribution of gene products deposited by the heterozygous mother. In addition, the majority of the lines probably do not carry a null allele, but rather hypomorphic mutations.

We have generated an STS flanking the P element insertion site for 496 of the 501 lines by either plasmid rescue or by inverse PCR (the sequences were submitted to the EMBL nucleotide database; see Supplementary data). The integration sites were determined by BLAST search to the published Drosophila genomic sequence (Adams et al., 2000). In cases where the gene annotation is supported by EST data, we can unambiguously identify the affected gene. For predicted genes whose open reading frame is only annotated we assumed that the insertion is integrated in the vicinity of a putative transcription start site upstream of the nearest translational start codon, since there is a strong preference for P integrations in the 5′-end of genes (Spradling et al., 1995).

Since the value of this collection is, in part ,determined by the number of new genes that are affected, we will briefly discuss our mapping data. A total of 513 STS sequences were generated from 496 strains. In five strains, repeated attempts to isolate flanking sequences failed, and in 16 the identification of the integration site was uninformative, since the P element is inserted in repetitive DNA. In 11 of these lines, the insertion is within a yoyo retrotransposon (Whalen and Grigliatti, 1998). The other 497 STSs were generated from the remaining 480 lines and are derived from unique X-chromosomal sequences. Five insertions were >7 kb away from any annotated gene and it is unclear whether the P integration is responsible for the observed lethal phenotype. Three additional P insertions occurred in the vicinity of known EP insertions but were some distance from the next annotated gene, suggesting that the the P elements have inserted into a putative promotor region of an unknown gene. Fourteen P elements, representing five gene pairs, may affect two genes. In each of these cases, one gene is localized within a large intron of another gene.

183 P{lacW} insertions affect 52 different genes that were previously characterized at both the molecular and genetic level, e.g. Notch, pebbled and short gastrulation. Phenotypically well characterized mutations of previously unidentified genes form a second group, consisting of two loci. These are trol, formerly known as zw1, with 12 insertions and stardust with one insertion. The largest group of insertions disrupts genes for which no mutation has previously been reported. In most cases, some molecular information is available or the genes have been predicted by computer algorithms (Adams et al., 2000). In summary, 301 insertions affect 130 genes (including the five nested gene pairs) that remain to be characterized. Among this class are transcription units that code for proteins putatively involved in signaling cascades, e.g. Rala, protein degradation, or Rpt4, intracellular transport, or Ntf-2, as well as many other cellular and developmental processes.

To determine the frequency by which genes were hit, we relied on molecular information. Indeed, allelism cannot easily be determined by genetic criteria, as complementation tests between X chromsome lethals are not possible without first introducing a duplication that rescues the lethality. Genes on the X chromosome that are hit only once represent 96 out of 490 insertions (19.6%). This compares to a rate of 25.8 and 28.8% in the second and third chromosome collection, respectively (Spradling et al., 1999). The 60 so-called warm spots (2–5 insertions per gene; Spradling et al., 1999) correspond to 169 X-chromosomal insertions (34.5%), similar to what was observed for the second and third chromosomes (35.8 and 34.2%, respectively). There are 23 hot spots on the X chromosome that are hit at least six times, resulting in a total of 225 insertions (45.9%). The frequencies for the second and third chromosome hot spots are 39.1 and 38.1%, respectively. The six genes that were hit most frequently are: Notch, 22 insertions; Trf2, 17; inx2, 15; ras, 14; and act5C, ctp and trol 12 each. In total, these six hot spots contained 104 insertions, more than all the transcription units that were hit once. In summary, in this X-chromosomal collection, single hits are under-represented and hot spots are over-represented, when compared to the autosomal P insertion collections. Taken together, of the X-chromosomal genes mutated in our screen each is hit, on average, 2.7 times, whereas the figures for the autosomal collection are 2.3 and 2.2, respectively.

Conclusions

Saturation mutagenesis studies give a ratio of lethal complementation groups to polytene chromosomes bands of 0.81 (Ashburner et al., 1999). This estimate predicts ∼820 essential genes for the X chromosome (1012 bands). Hence, our collection represents mutations in slightly <25% of the X-chromosomal essential genes and is similar to the data obtained for the autosomal collection of lethal P insertions (Spradling et al., 1999). The P element screen for X-chromosomal genes required for adult viability and the BAC tiling map reported here represent a valuable tool for the Drosophila research community. The gene disruption collection provides the first opportunity to link ∼130 X-chromosomal genes with a phenotype. The majority of the lines are available from the Bloomington stock center, together with their in situ hybridization data, and the enhancer trap expression patterns of these strains have been deposited in the FlyView database. Using these resources, the Drosophila research community can utilize the collection for the functional and molecular characterization of affected genes. The BAC tiling map can, on the other hand, serve as a reference point for mapping sequences and, more importantly, it will be essential as a source for a particular genomic fragment, e.g. for germ line transformation. The combined efforts of the community will not only increase our knowledge of the model organism Drosophila but will also, by virtue of the evolutionary conservation of many genes and processes, shed light on human gene function.

METHODS

Drosophila strains and embryos. The following fly strains (Lindsley and Zimm, 1992; FlyBase, 1999) were used in the experiments. The starting element, P{lacW} (Bier et al., 1989) originated from two male sterile insertion strains, 1040B and 1260A, which each contain a single transposon on chromosome 2. The strain used to supply transposase activity was w; CyO/wg^Sp; TM6/Sb P{ry⁺ Δ2-3}(99B) (Robertson et al., 1988). The female fertile FM6 and the female sterile FM7c chromosomes were employed as X chromosome balancers. In addition to being fertile in both sexes, FM6 has the advantage of carrying a white null allele, thereby facilitating the detection of low expression of the white⁺ marker gene.

Crossing scheme (see also Figure 2)

P: w/w; P{lacW}/P{lacW} × w/Y; wg^Sp/CyO; ry Sb P{ry⁺ Δ2-3}(99B)/TM6B

F₁: FM6/FM6 × w/Y; P{lacW}/CyO; ry Sb P{ry⁺ Δ2-3}(99B)/+

F₂: w/FM6; CyO/+ (with P{lacW} somewhere) × FM7c/Y

The absence of non-balancer bearing sons in the F₃ generation indicates a potential lethal insertion in the X chromosome.

Immunohistochemistry. Enhancer trap expression pattern was analyzed by immunohistochemistry. Embryos were collected on apple juice agar plates, fixed and stained with anti-β-galactosidase antibody (Cappel). The secondary peroxidase-coupled antibody was a biotinylated anti-rabbit Ig (Organon Teknika).

Molecular analysis. DNA was isolated from adult flies with QIAGEN DNeasy according to the instructions of the manufacturer. For isolation of flanking sequences by plasmid rescue (Pirrotta, 1987), EcoRI-digested DNA was used, resulting in DNA from the 3¢-end of P{lacW}. Inverse PCR (Silver, 1991) was performed essentially as described in the BDGP webpage (http://www.fruitfly.org/). The construction of the BAC tiling map and their application is described as supplementary information on the website ftp://ftp.ebi.ac.uk/pub/databases/edgp/200111/bac_probes.txt; the P insertion data are stored in the same directory under ftp://ftp.ebi.ac.uk/pub/databases/edgp/200111/PX-lines.txt.

Supplementary data. Supplementary data are available at EMBO reports Online.