Chromosomal Distribution of A. gambiae Orthologs of Genes From Two D. melanogaster Chromosomal Regions

In a first set of experiments aiming at exploring long-range synteny and microsynteny, we identified, among the currently available A. gambiae sequences, putative orthologs of genes in which in D. melanogaster are clustered within two well-studied chromosomal regions, each nearly 3 Mb long. We then determined the genomic locations of these putative orthologs by hybridization to the mosquito polytene chromosomes.

The fruit fly genomic regions that were chosen for these experiments have been completely sequenced and annotated, both in clone-by-clone sequencing projects and as part of whole-genome shotgun sequencing. One of these Drosophila regions is the autosomal Adhregion, covering 2.9 Mb on both sides of the Adh gene, in divisions 34B–35F of chromosomal arm 2L (Ashburner et al. 1999). The other is the tip of chromosome X, encompassing 2.6 Mb in polytene divisions 1–3 (Benos et al. 2000,2001). Both of these regions were also covered by whole-genome shotgun sequencing (Adams et al. 2000).

The 256 genes from the tip of the X and the 219 genes from theAdh region of Drosophila were used to query, byTBLASTN, collections of both STSs and ESTs ofAnopheles: the 17,506 STSs representing end sequences of BAC clones, and the 6012 ESTs that correspond to 2380 potential genes (cDNA clone clusters from a subtracted normalized library; Dimopoulos et al. 2000). To define genes as putative orthologs, hits that satisfied criteria of high score of >40, probability P(N) of <1 e⁻¹⁰, and percentage of identical amino acid residues >30 over a long range were selected in a first round. From them, all spurious hits that were caused by the presence of low complexity segments were eliminated, and the remaining hits were confirmed byBLASTX analysis as best bidirectional hits against a database of 14,080 amino acid sequences of known and predictedDrosophila genes (release 1.0; Adams et al. 2000). Those that passed this test were further verified by direct comparison to the corresponding Drosophila entry, taking into account potential intron-exon boundaries. Henceforth, these validated genes will be referred to as orthologs for convenience (see also Discussion). These procedures (see also Methods) identified 19 mosquito orthologs of unique genes found in the tip region of the Drosophila Xchromosome and 31 orthologs of unique genes found in theDrosophila Adh region. For greater accuracy, we eliminated from consideration additional probable orthologs (18 showing hits toX-tip and nine showing hits to Adh region genes), because they belong to chromosomally dispersed multigene families. This was necessary because the true ortholog can not be chosen among the different members of a given gene family until both genomes are fully sequenced.

The 50 orthologs that were retained for further analysis were present in 33 BAC and 37 cDNA clones (a number of them were detected by both STSs and ESTs). Representative clones were used as probes for in situ hybridization analysis to A. gambiae polytene chromosomes. Tables 1 and 2include the results of this analysis for the X-tip orthologs and Adh region orthologs, respectively. Notably, the Tables show cytogenetic and molecular locations of the 50 Drosophilagenes and the sequence identifiers and cytogenetic locations of the corresponding A. gambiae orthologs. The distribution of theX-Tip and Adh region orthologs among the five polytene chromosome arms of A. gambiae are tabulated in Table3, together with the results of statistical analysis of these distributions using the binomial test of significance, confirmed by the χ² test.

For the statistical analysis, we compared the number of orthologs corresponding to each Drosophila region that were observed in each chromosomal arm of the mosquito to the number expected if the association were random according to chromosomal arm length. To calculate the expected numbers, the lengths of the five mosquito chromosomal arms were estimated according to the number of their lettered subdivisions, as recognized in the map of Coluzzi and associates (22 subdivisions for X, 54 for 2R, 40 for2L, 37 for 3R, and 31 for 3L; or 12.0%, 29.35%, 21.7%, 20.1%, and 16.85% of the total, respectively; the map is accessible at http://www.anodb.gr/AnoDB/Cytomap/). The binomial test is an exact probability test that is used to examine the distribution of a single dichotomy in conditions when only a relatively small sample is available, as is the case here. It provides a one-sample test of the difference between the sampled distribution and a given distribution. In this case, the given distribution is based on the null hypothesis that the genes of each Drosophilachromosomal region are randomly redistributed across all five chromosome arms of A. gambiae according to their lengths. As shown in Table 3 for the gene probes derived from the tip of chromosomeX, all P values are >0.05, and thus, the null hypothesis cannot be rejected. Similarly, the χ² statistic (not shown) is equal to 2.415, lower than the critical value χ² ₄[0.05] = 9.49; therefore, the null hypothesis can not be rejected. By these criteria, none of the five mosquito chromosomal arms is significantly enriched for orthologs of the X-tip genes of Drosophila.

In contrast, the results of the same analysis strongly indicate that the Drosophila arm 2L (at least its Adhregion) corresponds to the chromosome arm 3R of A. gambiae (P = 9.939 e⁻¹²). That mosquito arm includes nearly fourfold as many genes as expected: It contains 24 (77%) of the currently available orthologs of the Drosophila Adh region genes, whereas only 7 (23%) orthologs are scattered over three other mosquito autosomal arms. Furthermore, in three out of four remaining mosquito chromosomal arms (2R, 2L, and3L), the prevalence of orthologs of Drosophila 2Lgenes is statistically significantly lower than expected. Thus, the binomial test clearly rejects the null hypothesis of random redistribution of Adh region genes, in terms of both positive and negative correlations. Rejection is also supported by the χ² analysis, in which the statistic (not shown) is equal to 64.12 with the same critical value as before (χ² ₄[0.05] = 9.49).

It should be noted from Table 3 that the mosquito orthologs of theAdh region genes are not evenly distributed within the mosquito 3R arm: Half of them are located within four chromosomal subdivisions (29C, 31C, 32B, and 33A), whereas the other half are scattered among the other 33 subdivisions of 3R. This apparent clustering may correspond to microsynteny, as will be discussed below.

Distribution of Randomly Selected A. gambiae Sequences and Their D. melanogaster Orthologs

A similar but reverse method was used in a second experiment addressing the question of long-range synteny. In this case, we started by mapping random A. gambiae STSs mapped on the polytene chromosomes and determined their orthologs and the respective cytogenetic locations in D. melanogaster.

Randomly selected BAC clones of A. gambiae from the library that had been used to determine STS end sequences (http://bioweb.pasteur.fr/BBMI; C. Roth and F.H. Collins, pers. comm.) were mapped by in situ hybridization to mosquito polytene chromosomes. A total of 1217 STS were available from 720 cytogenetically mapped clones, and they were used for a BLASTX search of the protein sequences corresponding to the 14,080 known and predictedD. melanogaster genes (release 1.0; Adams et al. 2000). This search led to the identification of 49 mapped STS that were putative orthologs of unique D. melanogaster genes. In addition,A. gambiae genes of known cytogenetic location were used to search the same Drosophila database, yielding 21 additional hits. This number also included cecropin and ADP/ATP, two A. gambiae genes, each of which is homologous to a corresponding small multigene family in Drosophila, clustered at a single cytogenetic location. Table4 lists these 70 mosquito gene sequences by cytogenetic location, together with theirDrosophila orthologs and their locations. Table5 summarizes and correlates the chromosomal locations of corresponding sequences in the two species. As in the previous experiment, the binomial test and the confirming χ² analysis (not shown) used the numbers of orthologs expected on each Drosophila chromosomal arm, in this case according to a random distribution calculated on the basis of the respective known DNA content of the Drosophila arms (Adams et al. 2000).

The data from this second experiment (Table 5) completely confirm and extend the conclusions from the first experiment. They identify statistically significant and unique chromosomal arm homologies with the P values ranging from 0.0193 to 0.0009, as follows:X ^Ag/X ^Dm,2R ^Ag/3R ^Dm,2L ^Ag/3L ^Dm,3R ^Ag/2L ^Dm, and3L ^Ag/2R ^Dm. Except for these, no other pairs even approach statistical significance as homologs. However, the dot chromosome 4 of Drosophila does not exist in the mosquito, and the single known Anopheles homolog of a chromosome 4 gene is found on the mosquito Xchromosome. In this second experiment, as much as in the first, the relative order of orthologous genes within the corresponding chromosomal arms of the two species appeared to be scrambled. Again, however, some residual microsynteny was detected (see below). For an additional statistical analysis of the same data, we took as a starting point the chromosomal distribution of the Drosophila orthologs and compared the observed and expected distributions ofAnopheles genes; this inverse comparison corresponds to that of the first experiment. As shown in Table6, the inverse P values are all significant, convincingly confirming the chromosomal arm homologies established from Table 5.

Local Synteny of Adh Region Orthologs

As noted above, many genes are scrambled within the respective homologous chromosomal arms. However, a careful analysis of gene order between genes of the Adh region in the D. melanogaster 2L and their orthologs in the A. gambiae 3R gave a clear indication that a significant proportion, ∼30%, remain locally clustered with the same neighboring gene. This local synteny may also be called microsynteny, in that it apparently only entails two or three genes at a time. The patterns of both gene scrambling and microsynteny are best displayed graphically, as in Figure 1. It should be noted that because of the availability of the genome sequence, the Drosophila Adhregion genes are placed on both cytogenetic and DNA sequence scales; their orthologs in Anopheles can only be placed on the cytogenetic scale for now.

Figure 1.

Distribution of Drosophila melanogaster genes of theAdh region and their A. gambiae orthologs on chromosome 3R. The left vertical dark blue bar represents theDrosophila Adh region, listing the names of the genes, their cytological location, and their positions on the molecular map (in megabases) according to Ashburner et al. (1999). Underlined are the genes for which corresponding Anopheles orthologs are not localized on chromosome 3R (see Table 2). The dark blue vertical boxes on the right represent the chromosome arm 3R ofAnopheles gambiae, with numbered divisions and lettered subdivisions. Clones that start with 4A3A or 4A3B are cDNAs, and the rest are STSs from chromosomal BACs. The solid lines point to the cytogenetic locations of these clones. The positions of orthologs are indicated by connecting lines between the Drosophila Adh region and Anopheles chromosome 3R. Colored lines connect A. gambiae cytogenetic gene clusters (green, 29C; pink, 31C; red, 32B; and light blue, 33C) and their respective fruit fly orthologs, also shown in the same color. The two chromosomes are not drawn to scale.

Open in new tabDownload PowerPointLink to figure

Of the 31 recognized mosquito orthologs of Adh region genes, 24 map to the Anopheles 3R chromosome, and 13 of these are found clustered in just four subdivisions, forming four cytogenetic clusters that are at least partially microsyntenic. In contrast, the remaining 11 mosquito orthologs are scattered individually amongst the remaining 33 chromosomal subdivisions of the Anopheles 3Rchromosome.

The two distal-most mosquito cytogenetic clusters, on divisions 29C and 31C, are both derived from a tight cluster of 27 Drosophilagenes that are located within ∼150 kb at cytogenetic location 35F6-11 (Ashburner et al. 1999). Of these 27 genes, 10 have known mosquito orthologs, and seven of these map to the mosquito chromosome arm3R; five are microsyntenic. The latter include two adjacent genes (the CG5861 and Sed5 orthologs) that map to the 31C cytogenetic cluster. The 29C cytogenetic cluster includes two adjacent genes (the cact and l(2)35Fe orthologs) plus one outlier (the BG:DS02740.4 ortholog). Each of these clusters additionally encompasses one ortholog of a distantAdh region gene (BG:BACR48E04.2 and lace, respectively).

Similarly, the mosquito 32B cytogenetic cluster includes threeAnopheles orthologs of genes BG:DS00797.7,b, and Sop2 that in Drosophila are part of an 16-gene cluster located within ∼65 kb at 34D1-4 (Ashburner et al. 1999). Two orthologs of other genes from the same cluster,adat and RpII33, are known in the mosquito but do not map at 32B; the orthologs of the 11 remaining genes in the 34D1-4Drosophila cluster are as yet unknown.

Finally, the fourth mosquito cytogenetic cluster at 33C includes the orthologs of adat from the Drosophila 34D1-4 region (see above) plus two genes, beat-B and beat from theDrosophila 35E1-F1 region. In Drosophila, the latter two genes are paralogs with the same exon-intron structure and show 53% identity at the amino acid level. They are separated by ∼100 kb, a region that encompasses three other genes, BG:DS07486.2,beat-C (also a paralog of beat-B and beat), and Bic-C (Ashburner et al. 1999); the orthologs of these three genes are not yet known in the mosquito. Interestingly, the orthologs of beat-B and beat are from the STSs at the two ends of the same mosquito BAC clone (03I12), and thus are also separated by ∼120 kb. It would be interesting to sequence this clone and thus discover whether the mosquito orthologs of theBG:DS07486.2, beat-C, and Bic-C genes are also located in this interval.

Source of Sequence Data

Amino acid sequences of the genes in divisions 1–3 of chromosomeX of D. melanogaster can be obtained by anonymous FTP fromftp://ftp.ebi.ac.uk/pub/databases/edgp/misc/ashburner/EG_genes.991229.pep.fa.gz(Benos et al. 2000, 2001), whereas amino acid sequences of the genes identified in the Adh region are found in http://www.fruitfly.org/sequences/aa_Adh.dros (Ashburner et al. 1999). Amino acid sequences of all genes identified through the whole genome sequence (release 1.0) are available athttp://www.fruitfly.org/sequence/dlMfasta.html (Adams et al. 2000). ForA. gambiae, nucleotide sequences of ESTs from immune-competent cell line cDNA libraries (Dimopoulos et al. 2000) and STSs from the BAC genomic library (C. Roth and F.C. Collins, pers. comm.), as well as other mosquito sequences with known cytological location, can beBLAST-searched at AnoDB, the Anopheles database (http://konops.anodb.gr/cgi-bin/blast2.pl).

Computational Methods and Analysis of Results

For similarity searches, a locally installed WU-BLAST, version 2.0a, suite of programs (Altschul et al. 1990; W. Gish, unpubl.) was used. D. melanogaster amino acid sequences of genes from selected regions were compared to A. gambiae STS and EST databases using TBLASTN with standard default parameters. STS and EST sequences showing similarity with a high score of >40, a probability P(N) of < e⁻¹⁰, and a percentage of identical amino acids >30, were selected and checked as best bidirectional hits after confirming the hit usingBLASTX with standard default parameters against a database of 14,080 amino acid sequences of known and predictedDrosophila genes (release 1.0,http://www.fruitfly.org/sequence/dlMfasta.html#rel1; Adams et al. 2000). Only STSs and ESTs that passed these criteria were selected, and their alignments were further verified using the available exon-intron structure of the corresponding D. melanogaster genes, as shown in the National Center for Biotechnology Information (NCBI) version of the D. melanogaster database (http://www.ncbi.nlm.nih.gov/PMGifs/Genomes/7227.html). The sameBLASTX search criteria were also used in the reciprocal experiment, comparing A. gambiae nucleotide sequences of known cytological location to protein encoding genes in D. melanogaster genes. The names and cytological locations of D. melanogaster genes were taken from FlyBase (http://flybase.bio.indiana.edu; The FlyBase Consortium 1999); additional information and literature references on genes can also be found there.

In Situ Hybridization to A. gambiae Polytene Chromosomes

BAC and cDNA clones were hybridized to preparations of A. gambiae polytene chromosomes essentially as described in Kumar and Collins (1994). The hybridization signals were localized according to the cytological map of M. Coluzzi, A. Sabbatini, M.A. Di Deco, and V. Petrarca (unpubl., accessible at http://www.anodb.gr/AnoDB/Cytomap/).