Contrasting Molecular Population Genetics of Four Hexokinases in Drosophila melanogaster, D. simulans and D. yakuba

Corresponding author: Walter F. Eanes, Department of Ecology and Evolution, State University of New York, Stony Brook, NY 11794., walter{at}life.bio.sunysb.edu (E-mail)

Communicating editor: C.-I WU

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

As part of a larger study contrasting patterns of variation in regulatory and nonregulatory enzymes of the central metabolic pathways we have examined the molecular variation in four uncharacterized hexokinase genes unique to muscle, fat body, and testis in Drosophila melanogaster, D. simulans, and D. yakuba. Earlier isoenzyme studies had designated these genes as Hex-A, Hex-C, and Hex-t. There are two tightly linked testes-specific genes designated here as Hex-t1 and Hex-t2. Substantial and concordant differences across species are seen in levels of both amino acid and silent polymorphism. The flight muscle form Hex-A is the most conserved followed by the fat body hexokinase Hex-C and testis-specific hexokinases Hex-t1 and Hex-t2. While constraints acting at the amino acid level are expected, the silent polymorphisms follow this pattern as well. All genes are in regions of normal recombination, therefore hitchhiking and background selection are not likely causes of interlocus differences. In D. melanogaster latitudinal clines are seen for amino acid polymorphisms at the Hex-C and Hex-t2 loci. There is evidence for accelerated amino acid substitution in Hex-t1 that has lost residues known to be associated with glucose and glucose-6-phosphate binding. D. simulans shows substantial linkage phase structuring that suggests historical population subdivision.

THE generation of complex phenotypes involves many steps, irrespective of whether the phenotype is a simple metabolic pool or a morphology whose expression involves a developmental pathway. Implicit in the understanding of genetic architecture is the expectation that all steps in a pathway may not be equal in their contribution to the phenotypic variation. A major challenge in understanding both local adaptation and long-term evolutionary divergence is to establish an understanding of which steps, because of their inherent properties, will be the most responsive to selection. In studies of metabolic regulation, both physiological biochemistry (HOCHACHKA and SOMERO 1984 ) and metabolic control theory (KACSER and BURNS 1973 ) have focused on the issue of pathway regulation. Physiological biochemistry has emphasized the role of key regulatory enzymes that display allosteric control and nonequilibrium reaction states. In contrast, metabolic control has emphasized the distribution of pathway regulation across many steps. In the 1970s the relationship between selective responses and pathway position first emerged in arguments about whether regulatory and nonregulatory enzymes should show more or less genetic polymorphism (JOHNSON 1971 ), then defined by allozymes. Amid much early speculation these questions remained unanswered. However, the introduction of DNA sequencing adds enormously more power and insight into the study of the impact of selection across steps in a pathway.

The concentration of pathway control in a few critical steps requires specialization in those enzymes. The central metabolic pathway, represented by glycolysis and the Krebs cycle, is active in all tissues; yet, for a given pathway step, selective pressures may be expected to be tissue or stage specific. The duplication and functional divergence of enzymes to meet the differing metabolic demands and modulator environments of different tissues is a pervasive theme in metazoan evolution, especially in the vertebrates. Divergence of function is typically achieved through gene duplication or the appearance of isozymes, which may arise from post-translational modification or alternative splicing. It appears in Drosophila melanogaster that most of the enzymes of the central metabolism show single forms derived from single-copy genes, although several, such as Gpdh and Ald, show multiple forms derived from alternative splicing (BEWLEY et al. 1989 ; KALM et al. 1989 ; SHAW-LEE et al. 1992 ). The notable exception is the hexokinases that are routinely encoded by multiple loci in Drosophila (MURRAY and BALL 1967 ; BECKENBACH and PRAKASH 1977 ; CAVENER 1980 ). The hexokinases represent the initial step in the oxidative phosphorylation of hexoses and are critically viewed as textbook "regulatory" or nonequilibrium enzymes that control flux into the pathway in response to several effectors, in particular the adenylate charge (HOCHACHKA and SOMERO 1984 ). In D. melanogaster four different prominent hexokinase isozymes have been reported from studies (MURRAY and BALL 1967 ). The so-called HEX-A and HEX-B isozymes are products of a single X-linked locus, Hex-A, while the HEX-C form is the product of a locus on the second chromosome, Hex-C, and segregates a common allozyme polymorphism in natural populations (MUKAI and VOELKER 1977 ). The HEX-A and HEX-B forms are largely expressed in adult thorax and larval muscle tissues, respectively, while the HEX-C enzyme appears confined to the adult and larval fat body, an organ similar in function to the vertebrate liver. In addition, a fourth form, HEX-T, which is expressed in the testis, has been reported (MURRAY and BALL 1967 ; CAVENER 1980 ). Given that these three hexokinases may regulate glycolysis in tissues with very different metabolic demands, we might expect very different evolutionary constraints acting on each gene. This report focuses on the molecular population genetics of the hexokinase loci of D. melanogaster and its relatives, D. simulans and D. yakuba. It extends our study of the population genetic and molecular evolution of genes involved in central metabolism and raises the possibility that genes dedicated to control of pathways that are operational in different tissues can have very different patterns of molecular variation and evolution that are perhaps predictable by their metabolic biology (EANES 1999 ).

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Origin of wild lines:
More than 1500 isofemale lines of D. melanogaster representing 10 local populations distributed from Florida to Vermont were collected in the fall of 1997. They are prefaced as HFL (Homestead, FL), MFL (Merritt Island, FL), JFL (Jacksonville, FL), SC (Eutawville, SC), NC (Smithfield, NC), VA (Richmond, VA), MD (Churchville, MD), CT (Middlefield, CT), MA (Concord, MA), and VT (Whiting, VT). Lethal-free second and third chromosomes were genetically extracted to establish isochromosomal lines using the CyO and TM6 balancers. A total of 14 D. simulans lines collected in 1996 and 1997 from Davis Peach Farm (DPF96s), Middleton, CT (CT96s), Homestead, FL (HFL97s), and Virginia (VAs96) were subjected to 14 generations of full-sib mating. Samples of flies from east Africa (ZimH and ZimS) were kindly donated by the lab of Chung-I Wu. The line of D. yakuba was obtained from the Bowling Green Stock Center (BG1013).

Identification and isolation of hexokinase sequences:
Short 353- and 640-bp segments of the Hex-A and Hex-t2 genes, respectively, were recovered from a third instar cDNA library using degenerate PCR primers designed to amino acid motifs conserved among hexokinase genes from yeast and mammals. These degenerate PCR products were sequenced, confirmed to have high similarity to hexokinases, and the 353-bp fragment was subsequently used to isolate the complete transcript sequence of Hex-A from the cDNA library by the method of VERRELLI and EANES 2000 . The complete genomic sequences for Hex-C, Hex-t1, and Hex-t2 emerged from BLAST searches of the Drosophila Genome Project database (accession nos. AA567642 and AC005813) using the degenerate PCR-amplified sequences of both Hex-A and Hex-t2 as the query sequences.

Northern blot analysis:
Probes for each hexokinase gene included an 1.2- to 1.4-kb fragment of coding sequence and were prepared by direct incorporation of digoxigenin-11-dUTP (Boehringer Mannheim, Mannheim, Germany) during PCR. Total RNA was prepared according to standard protocols (SAMBROOK et al. 1989 ) and 30 µg of each sample was electrophoresed through a 1% agarose-formaldehyde gel. RNA samples were transferred by capillary blotting onto positively charged nylon membranes (Boehringer Mannheim). Hybridization and detection were performed using the nonradioactive digoxigenin system (Boehringer Mannheim) following the protocols of ENGLER-BLUM et al. 1993 .

PCR amplification and DNA sequencing:
DNA sequences representing the complete coding regions for each of the hexokinase genes were amplified according to the following conditions: 10 ng of genomic DNA was amplified in 10 mM Tris (pH 8.3), 50 mM KCl, 0.01% gelatin, 1.5 mM MgCl₂, 2 units of platinum Taq polymerase (Life Technologies, Rockville, MD), and 100 nM of each primer. The amplification products were purified (Prep-A-Gene DNA purification kit; Bio-Rad, Hercules, CA), and double-stranded templates (500 ng per reaction) were sequenced manually (Sequenase, v.2.0; United States Biochemical, Cleveland) using primers spaced at 300-bp intervals. The primary sequences for each hexokinase gene were determined for both strands. In the population level studies, all ambiguous sites were verified on both strands.

Single nucleotide polymorphism screening:
Geographic data on allozyme polymorphisms were collected by screening a single female from each isofemale line using standard staining techniques (JELNES 1971 ). For the two third chromosome loci, 553 lethal-free third chromosomes were extracted with the TM3 balancer. Targeted single nucleotide polymorphisms (SNPs) were screened by performing single-base sequencing reactions in regions spanning the target polymorphisms.

Sampling strategy and statistical methods:
Sample size varied among loci because of different questions associated with specific amino acid polymorphisms. For estimation of population genetic parameters, the subsampled eastern population was considered the sampling unit and gene copies were sampled across the range from Florida to Vermont. A stratified sampling strategy was used. For mutations of unique interest, such as allozymes and chromosome inversions, sampling was expanded to examine within-lineage variation. For statistical tests and parameter estimation, a constructed random sample (HUDSON et al. 1994 ) was generated for each gene from the stratified data sets by random subsampling conditioned on independent population estimates of targeted allozyme and inversion frequencies. Parameters and statistics were determined using DnaSP (ROZAS and ROZAS 1999 ) and MEGA (SUDHIR et al. 1993 ).

To assess whether balancing or directional selection was operating on each gene, a series of tests were carried out to evaluate departures in the data from expectations for neutral variation evolving in a Wright-Fisher population. To evaluate the expected correlation between intra- and interspecific polymorphism a four-locus HKA test (HUDSON et al. 1987 ) was carried out. Failure to reject this model would imply that differences among locus polymorphisms arise from differences in neutral mutation rates. The two-locus version of this test is most commonly seen, but as originally proposed by HUDSON et al. 1987 the test can be extended to L loci with 2L-2 d.f. TAJIMA 1989 and FU and LI 1993 tests, which examine departures from the expected neutral allele frequency spectrum, were also carried out to detect historical balancing selection or, conversely, vestiges of hitchhiking events (CHARLESWORTH et al. 1995 ; FU 1997 ). The sequencing of D. yakuba as an out-group permitted a designation of all changes as derived or ancestral and allowed us to use the D-statistic of FU and LI's (1993) test (the F-statistic gives a similar result). Monte Carlo simulations were also carried out to evaluate the distribution of silent polymorphism with respect to specific mutation events that are defined by amino acid polymorphisms (HUDSON 1990 , HUDSON 1993 ). These follow the logic and design described in EANES et al. 1996 . To assess if any of the hexokinase genes possess elevated rates of amino acid substitution between species (see AKASHI 1999A , AKASHI 1999B ) or, conversely, elevated levels of amino acid polymorphism, we applied MCDONALD-KREITMAN (1991) tests to each of the four genes using polymorphism and fixed differences recovered from the D. melanogaster and D. simulans data sets.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Identification and characterization of paralogous hexokinase genes:
We isolated the complete coding sequences and intervening introns for four paralogous hexokinase genes in D. melanogaster, D. simulans, and D. yakuba. The chromosomal locations and tissue-specific expression patterns of these genes are consistent with the three previously described D. melanogaster hexokinase genes inferred from isoenzyme studies (MURRAY and BALL 1967 ; CAVENER 1980 ). Although not available at the time of recovery of these genes, a recent search of the complete D. melanogaster genome sequence (http://www.fruitfly.org/annot) has failed to find any other genes with hexokinase homology.

The putative Hex-A locus exhibits an uninterrupted coding region of 1347 bp and has subsequently been localized to polytene bands 8E10 in the Drosophila Genome Project (Gadfly gene CG3001). Its cytogenetic position had been mapped previously to band 8D using deletion mapping (LINDSLEY and ZIMM 1992 ). A Northern blot analysis confirmed that Hex-A is strongly expressed in the thorax of male and female adult flies, with no detectable transcript in adult abdomens (results not shown). This gene was reported to be expressed primarily in adult flight muscle (MOSER et al. 1980 ). The Hex-C locus exhibits an uninterrupted coding region of 1365 bp and was localized to band 51E9 by the Drosophila Genome Project (Gadfly gene CG8094). The Hex-C locus had been previously cytologically mapped to bands 51B–52E (BURKHART et al. 1984 ). A Northern blot analysis demonstrated that Hex-C is expressed only in the abdomens of male and female adult flies, consistent with its reported localization to fat body tissues (MOSER et al. 1980 ). Finally, the identity of the Hex-C locus was unequivocally confirmed by the identification, at the nucleotide level, of two inferred charge-changing amino acid polymorphisms that are correlated with known electrophoretic allozyme polymorphisms at this locus (described below).

The third hexokinase gene recovered in our PCR using degenerate primers was localized to band 97B2 of the third chromosome by Drosophila Genome Project (Gadfly gene CG5443). The 3086-bp region revealed two tandemly repeated hexokinase genes separated by a short variable-length intergenic sequence of 185–230 bp. The first gene exhibits an uninterrupted coding sequence of 1392 bp while the second gene contains a coding sequence of 1359 bp that is interrupted by a single variable-length intron of 88 bp in D. melanogaster and 71 bp in D. simulans. This intron is missing in D. yakuba. The Northern blot hybridization patterns for both genes were identical and localized to the abdominal segment in adult male flies. Although the specific function for these latter hexokinase genes is not known, the Northern hybridization patterns are consistent with the expression pattern reported previously for Hex-t, which was shown by isoenzyme study (MURRAY and BALL 1967 ; CAVENER 1980 ) to be male specific and localized to the testes. We have designated the two loci as Hex-t1 and Hex-t2. The predicted isoelectric points of the four proteins are also consistent with the relative electrophoretic mobilities of HEX-A, HEX-C, and HEX-T.

Inferred amino acid sequences show high similarity to other hexokinases in regions of established conservation (Fig 1). The amino acid sequences from the four hexokinase genes indicate that Hex-t1 and Hex-t2 are most recently derived and because of their close linkage perhaps have originated from a tandem duplication event. Levels of amino acid divergence among the four genes are quite high, ranging from 38 to 58% (ignoring gaps). This level of divergence suggests that the duplication events that gave rise to this gene family are quite ancient, allowing ample time for divergent expression patterns and different functional constraints to have developed.

View larger version (105K): In this window In a new window Download PPT slide   Figure 1. Amino acid sequence comparisons of the four candidate D. melanogaster hexokinases, Schistosoma mansoni hexokinase (Q26609), Saccharomyces cerevisiae hexokinase A (6321168), and human glucokinase (P35557). Glucose- and glucose-6-phosphate-binding residues are shown with dark backgrounds.

Intraspecific polymorphism and interspecific divergence:
Tables of the distribution of polymorphic sites across lines for each of the four genes for D. melanogaster and D. simulans samples can be viewed on the Eanes lab web site (http://www.life2.sunysb.edu/ee/eaneslab/hexdata) and all 130 sequences (including D. yakuba) are deposited in GenBank (accession nos. AF257522–AF257652). Table 1 provides the associated summary statistics and parameter estimates for the silent sites, as well as the Hex-t2 intron and Hex-t1-Hex-t2 intergenic region. Hex-A did not exhibit any amino acid polymorphisms in either species. At the Hex-C locus, four amino acid replacement polymorphisms were detected in 50 gene copies of D. melanogaster, while none were observed in 14 copies from D. simulans. Two of the replacement polymorphisms (nucleotides 113 and 1277) in D. melanogaster appear as singletons in our sample, while the other two (1049 and 1231) occur at intermediate frequencies. The polymorphism at position 1231 results in a Lys-to-Glu replacement at residue 411 and is responsible for the Hex-C⁴ allele (Hex-C^I, LINDSLEY and ZIMM 1992 ), a widely distributed allozyme polymorphism at the Hex-C locus (JELNES 1971 ; MUKAI and VOELKER 1977 ). The polymorphism at position 1277 results in an Ala-to-Glu replacement at residue 426 and the combined charge change with residue 411 results in the widespread but rarer Hex-C⁶ (Hex-C^F, LINDSLEY and ZIMM 1992 ) allozyme allele. The second common amino acid polymorphism (Lys to Arg at residue 350) is seen at position 1049 and results in no charge change. We also sequenced 12 copies found on In(2R)NS-bearing chromosomes. This inversion lies distally to the Hex-C locus, and since both standard and inversion-associated copies of Hex-C show the same haplotypes and harbor both common amino acid polymorphisms, recombination between arrangements in this region of the chromosome is apparently common.

Both D. melanogaster and D. simulans possess intraspecific amino acid polymorphisms in Hex-t1 and Hex-t2. In D. melanogaster three of the amino acid polymorphisms (positions 97 in Hex-t1, 2117 and 2764 in Hex-t2) appear as singletons and the other three (positions 728 in Hex-t1, 3076 and 3081 in Hex-t2) are at intermediate frequencies. In D. simulans one of the sites (position 4 in Hex-t1) appears as a singleton, while three other sites, all in Hex-t2 (positions 1895, 2149, and 2964), are at intermediate frequencies.

While numbers of amino acid polymorphisms vary across genes, there are also marked differences in levels of silent polymorphism, varying nearly 13-fold across loci in D. melanogaster ( ranges from 0.0020 to 0.0258). In D. simulans the contrast is even more striking: the Hex-A locus shows no variation at silent sites (Table 1 and Table 2). The levels of intraspecific polymorphism are also correlated between species. Using TAVARE 1984 equation 9.5 we have put approximate upper and lower confidence limits on our silent site-based estimates for all four loci. The 95% confidence limits of for Hex-A and the Hex-t1 and Hex-t2 loci do not overlap in either species (Table 1).

In contrast to the large disparities in levels of intraspecific polymorphism across genes, interspecific divergence between D. melanogaster and D. simulans is more modest (pairwise differences vary from 0.0151 to 0.0494, Table 1). The number of fixed amino acid differences per replacement site differs substantially across loci (Table 2). There were no amino acid replacements found at Hex-A, 2 fixed replacement differences were found at Hex-C, and 8 and 13 fixed amino acid replacements were observed in Hex-t1 and Hex-t2, respectively. The inclusion of sequence data for D. yakuba further extends this pattern (Table 2). Notably, there are still no amino acid differences at Hex-A.

A four-locus HKA test (HUDSON et al. 1987 ) was carried out to test the null hypothesis of concordance of silent polymorphism levels with silent site divergence in D. melanogaster and D. simulans. We derived joint parameter estimates of ₁ = 6.8 x 10^-3, ₂ = 12.2 x 10^-3, ₃ = 21.5 x 10^-3, ₄ = 21.0 x 10^-3, = 0.86, and = 6.77 and used these to generate expectations and variances for levels of polymorphism and divergence. For the combined species the result was not statistically significant (X² = 9.33, d.f. = 6), and we cannot reject the explanation that the varying silent polymorphism levels are due to among-locus variation in the neutral mutation rate. When the HKA test is partitioned into separate species tests, the D. simulans test is statistically significant (X² = 8.512, d.f. = 3, P < 0.05), and the largest deviation is associated with the absence of polymorphism associated at the Hex-A locus. Although not significant, the same deviation is apparent in the D. melanogaster data, suggesting that there is possibly more to the lower polymorphism levels at Hex-A than simply decreased mutation. To increase the power the four-locus HKA analysis was also carried out separately for D. melanogaster and D. simulans using D. yakuba to estimate divergence. Deviations from neutral expectations were nonsignificant and smaller than in the joint melanogaster-simulans contrast.

There is clear evidence for decreasing constraint in amino acid substitution from the flight- to the sperm-specific enzymes. A G-test of the homogeneity across genes for ratios of numbers of fixed synonymous and replacement mutations between D. melanogaster and D. simulans (Table 2) is highly significant (G = 13.78, d.f. = 3, P < 0.003). As a measure of relative constraint (assuming synonymous sites are neutral), K_a/K_s ratios for all three interspecific comparisons range from 0 for Hex-A to 0.128 in Hex-t2 (for D. simulans/D. yakuba) with a rank order of Hex-A < Hex-C < Hex-t1 < Hex-t2 (Table 2). The results for the four MCDONALD-KREITMAN (1991) tests were statistically significant (P < 0.006) for only the Hex-t1 gene, and it appears to deviate from the expectation either because there are too many amino acid fixations or too few silent fixations.

Samples of amino acid polymorphisms were surveyed for large-scale patterns of geographic variation. The Hex-C², Hex-C⁴, and Hex-C⁶ alleles were screened in isofemale lines in the 10 D. melanogaster populations ( = 150). The Hex-C⁶ allele is universally rare, while the ancestral Hex-C² allele shows statistically significant frequency variation (F_ST = 0.006, P < 0.05; F_ST adjusted for sampling variance), decreasing in more northern populations and forming a subtle but statistically significant cline against latitude (r = 0.74, P < 0.01). We also examined the pattern of variation for the Hex-C⁶ and Hex-C⁴ alleles combined, which represents the change in frequency of the derived Glu mutation. That cline is also highly significant (r = 0.69, P < 0.02) and is shown in Fig 2. The Hex-t1^T728A (Tyr/Phe polymorphism) allele was surveyed in 533 third chromosomes sampled across the same populations using an SNP screen. There was no significant allele frequency variation and the mean frequencies are not clinal (Fig 2). In contrast, although located downstream only 2348 bases from the Hex-t1 polymorphism, the derived Hex-t2^AAA allele (the Asn-Ile haplotype) shows significant geographic variation (F_ST = 0.061, P < 0.0001) and a highly significant latitudinal cline (Fig 2, r = 0.80, P < 0.01).

View larger version (15K): In this window In a new window Download PPT slide   Figure 2. Plot of derived (using D. simulans as an outgroup) allele frequency vs. latitude for 10 D. melanogaster population samples from the eastern United States. The Hex-CK411E amino acid polymorphism, the SNP sites at the Hex-t1 (Tyr/Phe polymorphism at nucleotide 728), and Hex-t2 (sites at nucleotides 3076, 3080, and 3081). Two additional points for Hex-CK411E were added (Raleigh, NC, and Stony Brook, NY).

Although there is evidence for selection from the latitudinal clines for Hex-C and Hex-t2, there is no evidence from the Monte Carlo simulations for significant genealogical departures from an expected Wright-Fisher model. Significant departures might be expected if there were adaptive polymorphisms that had recently increased in frequency or were under persistent balancing selection (HUDSON 1990 ; HUDSON et al. 1994 ; EANES et al. 1996 ). It should be emphasized, however, that such tests are conservative since recombination will decouple individual site genealogies and make inferences about specific sites under selection more difficult. There are a number of examples where all four gametes segregate in two-site combinations and this provides ample evidence for recombination. Using the method of HUDSON and KAPLAN 1985 there are estimated to be a minimum number of four recombination events in the sample of D. melanogaster Hex-C sequences from North America, while the Hex-t1 and Hex-t2 sequences show eight and seven recombination events, respectively, in the same sample.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Most enzymes of the glycolytic pathway in Drosophila are present as single forms, irrespective of the cellular environment in which they are expressed, but a few notable exceptions occur. Among these are the hexokinases. The hexokinases are believed to tightly regulate the entry of glucose into the glycolytic pathway in response to changes in reducing equivalents or metabolic charge. Consequently, features of the regulatory control function of hexokinases will be expected to vary along with the diverse requirements of various tissue environments. The four hexokinase genes that occur in the Drosophila genome show dramatically distinctive patterns of polymorphism and divergence, and this may reflect the intensities and patterns of selection imposed in order to regulate glycolytic flux in each of the unique tissue-specific environments where these genes are expressed. We begin with a discussion of general patterns of variation across all four hexokinase loci and then follow with a discussion of the specific patterns of polymorphism present at each of the individual loci.

The most notable differences exhibited among the four hexokinase paralogs were in the levels of both silent and replacement polymorphism and divergence. Differences in levels of replacement polymorphism and divergence among the four genes are consistent with intuition about the level of functional constraint imposed by metabolic demands in the various tissue environments. We expected a priori that the flight muscle- and the testis-specific forms would be the most and least constrained, respectively. This prediction was supported by the rank order of K_a/K_s ratios and by the relative levels of amino acid polymorphism among each of the four genes. However, levels of silent polymorphism differed significantly among paralogs as well (Hex-A significantly different from Hex-t1 and Hex-t2, Table 1). The failure to reject the neutral model in the four-locus HKA test argues that levels of silent polymorphism and divergence are correlated and consistent with locus-specific differences in mutation rate. Furthermore, the concordance of polymorphism and divergence at both silent and replacement sites suggests that the constraints operating on amino acid changes may transfer to synonymous site variation.

The substantial differences observed in silent polymorphism among the four hexokinases could be explained by extrinsic genomic factors if the recombinational landscape differed among loci. In D. melanogaster, low variation as a consequence of hitchhiking is correlated with regionally low levels of recombination, as seen in telomeric and centromeric regions (BEGUN and AQUADRO 1992 ; HUDSON and KAPLAN 1995 ), or with chromosomes lacking recombination entirely (BERRY et al. 1991 ; ZUROVCOVA and EANES 1999 ). However, since all four genes occur in regions of normal recombination, it is unlikely that hitchhiking due to selective sweeps or background selection has substantially reduced levels of variation as seen, for example, with Hex-A.

Another explanation for the concordance of polymorphism and divergence at silent and replacement sites could be if intensity of selection for codon bias varied among the four genes according to levels of functional constraint. However, levels of codon bias were unexpectedly low for all four hexokinase loci, Hex-A and Hex-C in particular. Most enzymes of central metabolism possess very high codon bias. For example, enolase (Eno), aldolase (Ald), and triose phosphate isomerase (Tpi), all farther down the glycolytic corridor, rank at or near the top of 428 genes surveyed and most others rank in the top 10% (Fig 3, R. KLIMAN, personal communication). This index is typically correlated with genes that undergo high rates of translation and are typically abundant (see MORIYAMA and POWELL 1997 ). An explanation for the relatively low bias might be that, unlike the other steps in the pathway, hexokinases operate as nonequilibrium enzymes that have not evolved to an upper neutral limit (see HARTL et al. 1985 ). Enzyme levels for such regulatory steps must be fine tuned to effectors that may occur at low concentrations. As a result, extremely low interspecific divergence and a near absence of polymorphism without high codon bias, such as seen for Hex-A, could be the consequence of stabilizing selection for tightly regulated protein levels. In contrast, many equilibrium enzymes, which are not duplicated and must function in many different metabolic contexts, have been selected to operate well below their catalytic capacity and do so by being abundant in many tissues with regulation occurring by high protein turnover. This will lead to these enzymes possessing high codon bias because of the need for high translational efficiency. It will be interesting to examine the pattern of polymorphism and interspecific divergence in phosphofructokinase (Pfk), which is also a textbook regulatory enzyme and has only average codon bias (Fig 3) even though it is flanked in the glycolytic pathway by high-bias genes such as Pgi and Ald. Whether patterns of codon bias can be used to make this type of inference is an interesting question.

View larger version (30K): In this window In a new window Download PPT slide   Figure 3. Codon bias of glycolytic enzymes representing the glycolytic pathway plotted against the distribution of codon adaptation indices (CAI) for 428 D. melanogaster genes (R. KLIMAN, personal communication).

The HEX-A isoenzyme, which shows very little or no intra- and interspecific variation and is specialized to the adult flight muscle, exhibits the highest affinity for glucose (MOSER et al. 1980 ), which, along with trehalose (which is converted to glucose), is used as the primary source of energy for flight in insects (SACKTOR and WORMSER-SHAVIT 1966 ). The importance of hexokinase as a regulator in high-metabolic-demand tissues, such as in muscle and brain, is evidenced by its being bound to the porins of the outer membrane of the mitochondria, where it can be the most sensitive to the immediate ATP demands of the cell (WILSON 1995 ). In contrast to other glycolytic corridor enzymes [e.g., phosphoglucose isomerase (PGI)], it is shown that in the high-flux flight muscle, hexokinase is operating at its maximum catalytic capacity or near saturation (SUAREZ et al. 1996 , SUAREZ et al. 1997 ). It would appear that these stringent demands have imposed very tight functional constraints. Furthermore, the Hex-A gene is highly conserved not only at the amino acid level, but also at synonymous positions and possesses the lowest K_a/K_s ratios (zero) in all interspecific comparisons. The lack of silent and replacement polymorphism in both D. melanogaster and D. simulans is probably not the result of a recent hitchhiking event since it would require that such an event occur independently in each species. Moreover, for Hex-A the nearest linked coding regions are over 9 kb upstream and 12 kb downstream. An argument against internal hitchhiking events is also supported by the complete conservation of the amino acid sequence among all three species surveyed in this study.

Two isozymes are associated with this locus (MOSER et al. 1980 ), but it is not known if these arise from alternate splicing or post-translational modification. We are not able to address this issue since our screen recovered only one cDNA, but an exhaustive search for potential exons in the 5' and 3' sequences flanking the gene (GenBank accession no. AC012813) failed to detect any potential exons.

The HEX-C enzyme is expressed in the adult and larval fat body and should be involved in the partitioning of energy resources and perhaps gluconeogenesis. While it exhibits activity on glucose and mannose substrates, HEX-C has its highest affinity for fructose and is likely primarily involved in metabolism of the latter (MOSER et al. 1980 ). There are multiple potential fates for fructose. It can be kinased to fructose-6-phosphate and enter the glycolytic pathway at phosphoglucose isomerase, or to fructose-1-phosphate and then be lysed into dihydroacetone phosphate and glyceraldehyde. In this sense HEX-C serves as a branch point enzyme, not unlike G6PD and phosphoglucomutase (PGM), and correspondingly is less conserved than Hex-A at both nonsynonymous and synonymous sites. Furthermore, consistent with amino acid polymorphisms at other branch point enzymes in glycolysis (Gpdh, OAKESHOTT et al. 1982 ; G6pd, OAKESHOTT et al. 1983 ; Pgm, B. C. VERRELLI and W. F. EANES, unpublished results) we see clines in allele frequencies with the derived Hex-C⁴ allozyme systematically increasing with latitude in our samples (Fig 2). The Lys-to-Glu amino acid polymorphism at residue 411, responsible for this common allozyme in D. melanogaster, is in a functionally important structural position. The three-dimensional structure of hexokinase from Schistosoma reveals that the position homologous to HEX-C site 411 (also a lysine residue) is located at a bend in an -helix that forms part of the ATP-binding domain (see Hex-C residue 411, Fig 1; MULICHAK et al. 1998 ). While the Lys is not strictly conserved, it is never occupied by a negatively charged amino acid and is flanked by a number of conserved residues in nearly all other hexokinases.

The pair of testes-specific genes represent an old duplication event in which new roles appear to have emerged. Because the HEX-T1 enzyme does not retain crucial hexose-binding residues (discussed below), we predict that the HEX-T2 enzyme is the testis-specific form identified by MURRAY and BALL 1967 . At Hex-t2 there are 13 amino acid replacements fixed between D. melanogaster and D. simulans (Table 2), and in D. melanogaster we see a pair of amino acid polymorphisms at the carboxy terminus of the protein. These two amino acid polymorphisms may be under latitudinal variable selection since the Asn-Ile haplotype varies over 45% in frequency in a steep cline between Vermont and Florida. This is comparable in magnitude to the well-characterized clines in Adh and Gpdh (OAKESHOTT et al. 1982 ; BERRY and KREITMAN 1993 ). There is evidence that selection is targeted to this site and not a larger chromosome region since a silent polymorphism at nucleotide site 3041, only 35 nucleotides away and also detected in our SNP screen, is only weakly associated with this pair of sites and shows no significant geographic variation. Furthermore, the common Try-Phe polymorphism in Hex-t1, which is 2395 bp away, is only in weak disequilibrium with these polymorphisms (n = 533, R² = 0.017, P < 0.0016) and also shows no cline. How selection might operate on these residues is unknown, although a look at the three-dimensional structure of hexokinase in Schistosoma reveals that the carboxy terminus, while not highly conserved, lies in very close proximity to the glucose-binding domain.

We hypothesize that the Hex-t1 gene has acquired a new sperm- or testis-related function that has co-opted its hexokinase properties. The overall amino acid alignment shows Hex-t1 is unmistakably related to other hexokinases, yet it shows changes in 6 of the 13 residues considered important for glucose and glucose-6-phosphate binding (see Fig 1; WILSON 1995 ; MULICHAK et al. 1998 ), and these residues are all conserved in the HEX-A, HEX-C, and HEX-T2 proteins, with a single exception in HEX-T2. The importance of these critical residues is well established and mutations at these sites are associated with maturity onset type diabetes, in the human 50-kD pancreatic glucokinase homologue (see GIDHJAIN et al. 1993 ; VELHO et al. 1997 ).

Nonetheless, even with this apparent loss of hexose-binding ability, Hex-t1 is still clearly under selective constraint. Most of the variable sites in Hex-t1 are synonymous and the average K_a/K_s ratio (Table 2) is smaller than that for Hex-t2, implying greater constraint. When the physiochemical differences (see GRANTHAM 1974 ) for each amino acid substitution among all three species are examined, the mean differences for Hex-t1 and Hex-t2 are not statistically different when compared by Wilcoxon two-sample test (SOKAL and ROHLF 1981 ). This indicates that on the average replacement substitutions at Hex-t1 are no less physiochemically conservative than changes at Hex-t2. Most importantly, by the criteria of the McDonald-Kreitman test Hex-t1 shows a significant excess of interspecific replacement substitutions and this argues that it has been subjected to directional selection.

If Hex-t1 and Hex-t2, which are relatively weakly expressed only in the whole abdomens of males, are functioning exclusively in the sperm (or are more broadly expressed in the testes) then they could be influenced by sexual selection. In mammalian sperm there are multiple isoforms of hexokinase-1, and a germ line isoform, HK-sc, is believed to act in both energy generation and as a sperm surface receptor for a ligand in the zona pellucida (TRAVIS et al. 1998 ). This suggests that the glucose-binding ability of hexokinases can be co-opted for carbohydrate binding in the extracellular matrix of the egg. With respect to the Drosophila hexokinases this is only speculation, but the studies in mammal sperm emphasize the multiple roles that hexokinases can play. Given the recurring observation of new function in genes associated with testis-specific specialization (NURMINSKY et al. 1998 ) and the associated observation that male-specific genes are under accelerated evolution (TING et al. 1998 ; WYCKOFF et al. 2000 ), the pattern at Hex-t1 further extends this class of observations to an enzyme whose function was probably limited to glycolytic energy production in sperm before duplication.

Patterns of polymorphism are more complex in D. simulans at Hex-C, Hex-t1, and Hex-t2. The Hex-C locus has no amino acid polymorphisms, and there is very strong linkage disequilibrium (resulting in a significant positive Fu and Li D-statistic), where only three haplotypes emerge from 10 segregating polymorphisms. Likewise, there is very strong linkage disequilibrium in the Hex-t1, -t2 region. The Hex-t1 locus shows only a singleton amino acid variant, while Hex-t2 shows a complex set of replacement polymorphisms: a common ancestral amino acid sequence segregating most of the silent-site polymorphisms at the locus, and two derived sequences, one bearing two amino acid differences from the ancestral sequence and a second bearing a single difference from the ancestral sequence. Both derived sequences show no within-lineage variability and this extends across the entire Hex-t1, -t2 region. Such complex haplotype structure is a recurring theme in D. simulans (HASSON et al. 1998 ; HAMBLIN and VEUILLE 1999 ; LABATE et al. 1999 ) and could be the result of historical population subdivision. D. simulans simply may not be a good candidate for a species in genetic equilibrium, and this makes contrasts of polymorphism with D. melanogaster less informative.

Finally, the polymorphism difference between species also differs with other studies. D. simulans possesses lower amino acid polymorphism than D. melanogaster, consistent with other studies, but it also possesses lower silent polymorphism (f = 0.86 from the HKA test), unlike the general trend seen for the loci in MORIYAMA and POWELL's (1996) survey, where silent polymorphism is on average two- to threefold higher. Reduced levels of replacement polymorphism in D. simulans are generally assumed to reflect greater efficacy of purifying selection in a species with larger effective population size (AQUADRO et al. 1988 ; AKASHI 1995 ). Therefore, reduced levels of silent polymorphism in D. simulans may provide an additional indication that silent sites are subject to purifying selection at hexokinase loci. Finally, while in D. melanogaster amino acid polymorphisms in several metabolic genes show evidence of age and geographic patterning favoring selection (EANES 1999 ), D. simulans continues to show no such patterns.

	FOOTNOTES

¹ Present address: Department of Biological Sciences, Southern Illinois University, Edwardsville, IL 62026.

	ACKNOWLEDGMENTS

We thank Brian Verrelli, Luciano Matzkin, Efe Sezgin, and Yihao Duan for their contributions to numerous genetic constructions and other demands of this project. This project was supported in part by U.S. Public Health Service grant GM-45247 and National Science Foundation grant DEB-9318381 to W. F. Eanes. This is contribution number 1057 from the Graduate Program in Ecology and Evolution at Stony Brook.

Manuscript received January 13, 2000; Accepted for publication July 14, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AKASHI, H., 1995  Inferring weak selection from patterns of polymorphism and divergence at "silent" sites in Drosophila DNA. Genetics 139:1067-1076[Abstract].

AKASHI, H., 1999a  Inferring the fitness effects of DNA mutations from polymorphism and divergence data: statistical power to detect directional selection under stationarity and free recombination. Genetics 151:221-238[Abstract/Free Full Text].

AKASHI, H., 1999b  Within- and between-species DNA sequence variation and the `footprint' of natural selection. Gene 238:39-51[Medline].

AQUADRO, C. F., K. M. LADO, and W. A. NOON, 1988  The rosy region of Drosophila melanogaster and Drosophila simulans. I. Contrasting levels of naturally occurring DNA restriction map variation and divergence. Genetics 119:875-888[Abstract/Free Full Text].

BECKENBACH, A. T. and S. PRAKASH, 1977  Examination of allelic variation at the hexokinase loci of Drosophila pseudoobscura and D. persimilis by different methods. Genetics 87:743-761[Abstract/Free Full Text].

BEGUN, D. J. and C. F. AQUADRO, 1992  Levels of naturally occurring DNA polymorphism correlate with recombination rates in D. melanogaster.. Nature 356:519-520[Medline].

BERRY, A. J., J. W. AJIOKA, and M. KREITMAN, 1991  Lack of polymorphism on the Drosophila fourth chromosome resulting from selection. Genetics 129:1111-1117[Abstract].

BERRY, A. and M. KREITMAN, 1993  Molecular analysis of an allozyme cline: alcohol dehydrogenase in Drosophila melanogaster on the east coast of North America. Genetics 134:869-893[Abstract].

BEWLEY, G. C., J. L. COOK, S. KUSAKABE, T. MUKAI, and D. L. RIGBY et al., 1989  Sequence, structure and evolution of the gene coding for sn-glycerol-3-phosphate dehydrogenase in Drosophila melanogaster.. Nucleic Acids Res. 17:8553-8567[Abstract/Free Full Text].

BURKHART, B. D., E. MONTGOMERY, C. H. LANGLEY, and R. A. VOELKER, 1984  Characterization of allozyme null and low activity alleles from two natural populations of Drosophila melanogaster.. Genetics 107:295-306[Abstract/Free Full Text].

CAVENER, D. R., 1980  Genetics of male-specific glucose oxidase and the identification of other unusual hexose enzymes in Drosophila melanogaster.. Biochem. Genet. 18:929-937[Medline].

CHARLESWORTH, D., B. CHARLESWORTH, and M. T. MORGAN, 1995  The pattern of neutral molecular variation under the background selection model. Genetics 141:1619-1632[Abstract].

EANES, W. F., 1999  Analysis of selection on enzyme polymorphisms. Annu. Rev. Ecol. Syst. 30:301-326.

EANES, W. F., M. KIRCHNER, J. YOON, C. H. BIERMANN, and I. N. WANG et al., 1996  Historical selection, amino acid polymorphism and lineage-specific divergence at the G6pd locus in Drosophila melanogaster and D. simulans.. Genetics 144:1027-1041[Abstract].

ENGLER-BLUM, G., M. MEIER, J. FRANK, and G. A. MULLER, 1993  Reduction of background problems in non-radioactive Northern and Southern blot analyses enables higher sensitivity than ³²P-based hybridizations. Anal. Biochem. 210:235-244[Medline].

FU, Y. X., 1997  Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 147:915-925[Abstract].

FU, Y. X. and W. H. LI, 1993  Statistical tests of neutrality of mutations. Genetics 133:693-709[Abstract].

GIDHJAIN, M., J. TAKEDA, L. Z. XU, A. J. LANGE, and N. VIONNET et al., 1993  Glucokinase mutations associated with non-insulin-dependent (type-2) diabetes-mellitus have decreased enzymatic activity-implications for structure/function relationships. Proc. Natl. Acad. Sci. USA 90:1932-1936[Abstract/Free Full Text].

GRANTHAM, R., 1974  Amino acid difference formula to help explain protein evolution. Science 185:862-864[Abstract/Free Full Text].

HAMBLIN, M. T. and M. VEUILLE, 1999  Population structure among African and derived populations of Drosophila simulans: evidence for ancient subdivision and recent admixture. Genetics 153:305-317[Abstract/Free Full Text].

HARTL, D. L., D. E. DYKHUIZEN, and A. M. DEAN, 1985  Limits of adaptation: the evolution of selective neutrality. Genetics 111:655-674[Abstract/Free Full Text].

HASSON, E., I. N. WANG, L. W. ZENG, M. KREITMAN, and W. F. EANES, 1998  Nucleotide variation in the triosephosphate isomerase (Tpi) locus of Drosophila melanogaster and Drosophila simulans.. Mol. Biol. Evol. 15:756-769[Abstract].

HOCHACHKA, P. W., and G. N. SOMERO, 1984 Biochemical Adaptation. Princeton University Press, Princeton, NJ.

HUDSON, R. R., 1990 Gene genealogies and the coalescent process, pp. 1–44 in Oxford Surveys in Evolutionary Biology, edited by D. J. FUTUYAMA and J. ANTONOVICS. Oxford University Press, New York.

HUDSON, R. R., 1993 The how and why of generating gene genealogies, pp. 23–36 in Mechanisms of Molecular Evolution, edited by N. TAKAHATA and A. G. CLARK. Japan Scientific Societies Press and Sinauer Associates, Tokyo and Sunderland, MA.

HUDSON, R. R. and N. L. KAPLAN, 1985  Statistical properties of the number of recombination events in the history of a sample of DNA sequences. Genetics 111:147-164[Abstract/Free Full Text].

HUDSON, R. R. and N. L. KAPLAN, 1995  The coalescent process and background selection. Philos. Trans. R. Soc. Lond. [Biol.] 349:19-23[Medline].

HUDSON, R. R., M. KREITMAN, and M. AGUADE, 1987  A test of neutral molecular evolution based on nucleotide data. Genetics 116:153-159[Abstract/Free Full Text].

HUDSON, R. R., K. BAILEY, D. SKARECKY, J. KWIATOWSKI, and F. J. AYALA, 1994  Evidence for positive selection in the superoxide dismutase (Sod) region of Drosophila melanogaster.. Genetics 136:1329-1340[Abstract].

JELNES, J. E., 1971  Identification of hexokinases and localisation of a fructokinase and a tetrazolium oxidase locus in Drosophila melanogaster.. Hereditas 67:291-293[Medline].

JOHNSON, G. B., 1971  Metabolic implications of polymorphism as an adaptive strategy. Nature 32:347-349.

KACSER, H. and J. A. BURNS, 1973  The control of flux. Symp. Soc. Exp. Biol. 27:65-104[Medline].

KALM, VON L., J. WEAVER, J. DEMARCO, R. J. MACINTYRE, and D. T. SULLIVAN, 1989  Structural characterization of the alpha-glycerol-3-phosphate dehydrogenase-encoding gene of Drosophila melanogaster.. Proc. Natl. Acad. Sci. USA 86:5020-5024[Abstract/Free Full Text].

LABATE, J. A., C. H. BIERMANN, and W. F. EANES, 1999  Nucleotide variation at the runt locus in Drosophila melanogaster and D. simulans.. Mol. Biol. Evol. 16:724-731[Abstract].

LINDSLEY, D. L., and G. G. ZIMM, 1992 The Genome of Drosophila melanogaster. Academic Press, New York.

MCDONALD, J. H. and M. K. KREITMAN, 1991  Adaptive evolution at the Adh locus in Drosophila. Nature 351:652-654[Medline].

MORIYAMA, E. N. and J. R. POWELL, 1996  Intraspecific nuclear DNA variation in Drosophila. Mol. Biol. Evol. 13:261-277[Abstract].

MORIYAMA, E. N. and J. R. POWELL, 1997  Codon usage bias and tRNA abundance in Drosophila. J. Mol. Evol. 45:514-523[Medline].

MOSER, D., L. JOHNSON, and C.-Y. LEE, 1980  Multiple forms of Drosophila hexokinases. Purification, biochemical and immunological characterization. J. Biol. Chem. 255:4673-4679[Abstract/Free Full Text].

MUKAI, T. and R. A. VOELKER, 1977  The genetic structure of natural populations of Drosophila melanogaster XIII. Further studies on linkage disequilibrium. Genetics 86:175-185[Abstract/Free Full Text].

MULICHAK, A. M., J. E. WILSON, K. PADMANABHAN, and R. M. GARAVITO, 1998  The structure of mammalian hexokinase-1. Nat. Struct. Biol. 5:555-560[Medline].

MURRAY, R. F. and J. A. BALL, 1967  Testis-specific and sex-associated hexokinases in Drosophila melanogaster.. Science 156:81-82[Abstract/Free Full Text].

NURMINSKY, D. I., M. V. NURMINSKAYA, D. DE AGUIAR, and D. L. HARTL, 1998  Selective sweep of a newly evolved sperm-specific gene in Drosophila. Nature 396:572-575[Medline].

OAKESHOTT, J. G., J. B. GIBSON, P. R. ANDERSON, W. R. KNIBB, and G. K. CHAMBERS, 1982  Alcohol dehydrogenase and glycerol-3-phosphate dehydrogenase clines in Drosophila melanogaster on different continents. Evolution 36:86-96.

OAKESHOTT, J. G., G. K. CHAMBERS, J. B. GIBSON, W. F. EANES, and D. A. WILLCOCKS, 1983  Geographic variation in G6pd and Pgd allele frequencies in Drosophila melanogaster.. Heredity 50:67-72.

ROZAS, J. and R. ROZAS, 1999  DnaSP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformation 15:174-175.

SACKTOR, B. and E. WORMSER-SHAVIT, 1966  Regulation of metabolism in working muscle in vivo. I. Concentrations of some glycolytic, tricarboxylic acid cycle and amino acid intermediates in insect flight muscle during flight. J. Biol. Chem. 241:624-631[Abstract/Free Full Text].

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual, Ed. 2. Cold Spring Harbor Laboratory Press, Plainview, NY.

SHAW-LEE, R., J. L. LISSEMORE, D. T. SULLIVAN, and D. R. TOLAN, 1992  Alternative splicing of fructose 1,6-bisphosphate aldolase transcripts in Drosophila melanogaster predicts three isozymes. Biol. Chem. 267:3959-3967.

SOKAL, R. R., and F. J. ROHLF, 1981 Biometry. W. H. Freeman, New York.

SUAREZ, R. K., J. R. B. LIGHTON, B. JOOS, S. P. ROBERTS, and J. F. HARRISON, 1996  Energy metabolism, enzymatic flux capacities and metabolic flux rates in flying honeybees. Proc. Natl. Acad. Sci. USA 93:12616-12620[Abstract/Free Full Text].

SUAREZ, R. K., J. F. STAPLES, J. R. B. LIGHTON, and T. G. WEST, 1997  Relationships between enzymatic flux capacities and metabolic flux rates: nonequilibrium reactions in muscle glycolysis. Proc. Natl. Acad. Sci. USA 94:7065-7069[Abstract/Free Full Text].

SUDHIR, K., K. TAMURA and M. NEI, 1993 MEGA: Molecular Evolutionary Genetics Analysis, version 1.0. The Pennsylvania State University, University Park, PA.

TAJIMA, F., 1989  Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123:585-595[Abstract/Free Full Text].

TAVARE, S., 1984  Line-of-descent and genealogical processes, and their applications in population genetics models. Theor. Popul. Biol. 26:119-164[Medline].

TING, C.-T., S.-C. TSAUR, M.-L. WU, and C.-I. WU, 1998  A rapidly evolving homeobox at the site of a hybrid sterility gene. Science 282:1501-1503[Abstract/Free Full Text].

TRAVIS, A. J., J. A. FOSTER, N. A. ROSENBAUM, P. E. VISCONTI, and G. L. GERTON et al., 1998  Targeting of germ cell-specific type 1 hexokinase lacking a porin-binding domain to the mitochondria as well as to the head and fibrous sheath of murine spermatozoa. Mol. Biol. Cell 9:263-276[Abstract/Free Full Text].

VELHO, G., H. BLANCHE, M. VAXILLAIRE, C. BELLANNE-CHANTELOT, and V. C. PARDINI et al., 1997  Identification of 14 new glucokinase mutations and description of the clinical profile of 42 MODY-2 families. Diabetologia 40:217-224[Medline].

VERRELLI, B. C., and W. F. EANES, 2000 Extensive amino acid polymorphism at the Pgm locus is consistent with adaptive protein evolution in Drosophila melanogaster. Genetics 156(4) (in press).

WILSON, J. E., 1995  Hexokinases. Rev. Physiol. Biochem. Pharmacol. 126:65-198[Medline].

WYCKOFF, G. J., W. WANG, and C. I. WU, 2000  Rapid evolution of male reproductive genes in the descent of man. Nature 403:304-309[Medline].

ZUROVCOVA, M. and W. F. EANES, 1999  Lack of nucleotide polymorphism in the Y-linked sperm flagellar Dynein gene Dhc-Yh3 of Drosophila melanogaster and D. simulans.. Genetics 153:1709-1715[Abstract/Free Full Text].

This article has been cited by other articles:

				J. Flowers, E Sezgin, S Kumagai, D. Duvernell, L. Matzkin, P. Schmidt, and W. Eanes Adaptive Evolution of Metabolic Pathways in Drosophila Mol. Biol. Evol., June 1, 2007; 24(6): 1347 - 1354. [Abstract] [Full Text] [PDF]

				A. Sanchez-Gracia and J. Rozas Unusual Pattern of Nucleotide Sequence Variation at the OS-E and OS-F Genomic Regions of Drosophila simulans Genetics, April 1, 2007; 175(4): 1923 - 1935. [Abstract] [Full Text] [PDF]

				W. F. Eanes, T. J. S. Merritt, J. M. Flowers, S. Kumagai, E. Sezgin, and C.-T. Zhu Flux control and excess capacity in the enzymes of glycolysis and their relationship to flight metabolism in Drosophila melanogaster PNAS, December 19, 2006; 103(51): 19413 - 19418. [Abstract] [Full Text] [PDF]

T. J. S. Merritt, D. Duvernell, and W. F. Eanes Natural and Synthetic Alleles Provide Complementary Insights Into the Nature of Selection Acting on the Men Polymorphism of Drosophila melanogaster Genetics, December 1, 2005; 171(4): 1707 - 1718.