The nonA Gene in Drosophila Conveys Species-Specific Behavioral Characteristics

Corresponding author: Charalambos P. Kyriacou, Room 141, Adrian Bldg., Department of Genetics, University of Leicester, University Rd., Leicester LE1 7RH, United Kingdom., cpk{at}leicester.ac.uk (E-mail)

Communicating editor: J. J. LOROS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The molecular basis of species-specific differences in courtship behavior, a critical factor in preserving species boundaries, is poorly understood. Genetic analysis of all but the most closely related species is usually impossible, given the inviability of hybrids. We have therefore applied interspecific transformation of a single candidate behavioral locus, no-on-transient A (nonA), between Drosophila virilis and D. melanogaster, to investigate whether nonA, like the period gene, might encode species-specific behavioral information. Mutations in nonA can disrupt both visual behavior and the courtship song in D. melanogaster. The lovesong of nonA^diss mutant males superficially resembles that of D. virilis, a species that diverged from D. melanogaster 40–60 mya. Transformation of the cloned D. virilis nonA gene into D. melanogaster hosts carrying a synthetic deletion of the nonA locus restored normal visual function (the phenotype most sensitive to nonA mutation). However, the courtship song of transformant males showed several features characteristic of the corresponding D. virilis signal, indicating that nonA can act as a reservoir for species-specific information. This candidate gene approach, together with interspecific transformation, can therefore provide a direct avenue to explore potential speciation genes in genetically and molecularly tractable organisms such as Drosophila.

THE results of many years of experiments, in which closely related species have been crossed together to detect the genetic architecture of species-specific characters, have usually detected an underlying polygenic system (reviewed in COYNE 1992 ). For obvious reasons, interspecific hybridization is difficult and seldom leads to the production of segregating generations. However, in some cases, a donor species can be transformed with the DNA of another, effectively bypassing the commonly associated sterility and inviability of hybrids. In the behavioral literature, this kind of study has been performed twice, in both cases with the period (per) gene in Drosophila (PETERSEN et al. 1988 ; WHEELER et al. 1991 ). These two studies clearly showed that per genes of the two donor Drosophila species were able to transfer the species-specific characteristics of their rhythmic phenotypes (circadian locomotor activity and ultradian courtship song cycles, respectively) to their Drosophila melanogaster hosts. This switch in the behavior of the transformants was determined by interspecific differences in the coding region of per, as opposed to the regulatory regions. Furthermore, in the case of the ultradian lovesong cycle, the species-specific sequences that were responsible resided in a small fragment that contained a limited number of interspecific amino acid substitutions (WHEELER et al. 1991 ).

A single gene is therefore capable of acting as a reservoir for species-specific information and raises the question of how general this phenomenon might be. Stripped of the cumbersome burden of interspecific genetics, transformation of single genes between species could reveal, as in the case of per, a rather simpler deterministic picture of species-specific behavior than otherwise believed. Consequently, we have decided to extend this approach to another "candidate" locus.

The no-on-transientA (nonA) gene from Drosophila encodes a putative RNA-binding protein, but its function at the biochemical level is unknown (BESSER et al. 1990 ; JONES and RUBIN 1990 ). The nonA locus is sex linked and overlaps partially with the lethal locus l(1)i19e (JONES and RUBIN 1990 ; CAMPESAN et al. 2001 ; SANDRELLI et al. 2001 ). Synthetic deletions of nonA are poorly viable (STANEWSKY et al. 1993 ). Other mutant alleles of nonA produce visual abnormalities (HOTTA and BENZER 1969 , HOTTA and BENZER 1970 ; PAK et al. 1970 ), but the nonA^diss male is, in addition, defective in its courtship song (KULKARNI et al. 1988 ; RENDAHL et al. 1992 ). Interestingly, the latter's mutant song phenotype shows similarities to the wild-type songs of the D. virilis group of species, which has song pulses that are more polycyclic than those of D. melanogaster (HOIKKALA and LUMME 1984 , HOIKKALA and LUMME 1987 and see Fig 1). This situation with nonA^diss mutants is somewhat analogous to that of the per mutants, per^s and TG, whose males have song rhythms that are characteristic of wild-type D. simulans (KYRIACOU and HALL 1980 , KYRIACOU and HALL 1986 , KYRIACOU and HALL 1989 ; YU et al. 1987 ; ALT et al. 1998 ). Thus, like per, this raises the possibility that nonA might also carry species-specific song information.

View larger version (57K): In this window In a new window Download PPT slide   Figure 1. Courtship song bursts from D. melanogaster transformant male hemizygous for the wild-type nonA transgene (top), nonAdiss mutant male (middle), and a D. virilis male (bottom).

The cloning and sequencing of the D. virilis nonA orthologue has been described, as has the transformation of the D. virilis nonA transgene (nonA^vir) into D. melanogaster nonA^- mutants and the associated full rescue of the mutants' poor viability and abnormal electroretinogram (ERG; CAMPESAN et al. 2001 ; SANDRELLI et al. 2001 ). In this study, we investigate the courtship song and another visual phenotype in nonA^vir transformants. Our results suggest that the D. melanogaster transformant hosts may have taken on some of the D. virilis song characteristics.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly strains:
T(1:4)9e2-10/FM7: This strain carries a reciprocal translocation T(1:4), which uncovers the nonA gene and the adjacent essential locus l(1)i19e (STANEWSKY et al. 1993 ) and is balanced with FM7 (LINDSLEY and ZIMM 1992 ). Females crossed to males carrying an autosomal D. melanogaster or D. virilis nonA transgene generate viable males carrying T(1:4)9e2-10 only if the transgene also encodes the neighboring lethal gene l(1)i19e (CAMPESAN et al. 2001 ).

P[(ry) 235R11]: This is a transformant strain carrying an 11-kb EcoRI fragment (homozygous on the third chromosome) that encodes both the D. melanogaster nonA locus and the adjacent lethal gene l(1)i19e (JONES and RUBIN 1990 ). This fragment rescues all nonA mutant phenotypes and the associated lethality due to l(1)i19e (JONES and RUBIN 1990 ; RENDAHL et al. 1992 ; CAMPESAN et al. 2001 ). Hemizygous males that carry a single autosomal copy of this nonA encoding fragment, T(1:4)9e2-10/Y; P[(ry)235R11]/+ (abbreviated to mel/d), were generated and served as the melanogaster control. The particular 235R11 transformant line we used is one of two, which, when crossed to T(1:4)9e2-10 to generate hemizygous nonA⁺ males, gives identical song phenotypes (STANEWSKY et al. 1993 ; RENDAHL et al. 1996 ).

The negative control was provided by nonA^diss/Y males, which were taken from a nonA^diss/FM7a ; ry 506/MKRS strain, which was repeatedly backcrossed to Canton-S, and the nonA^diss mutants re-extracted by selecting for males with the mutant song phenotype. A similar high level of congenicity was maintained between all the genotypes to be compared by prior crossing of all transformant individuals (see below) to a Cantonized w strain.

Transformant lines:
A number of lines were generated that carried a large genomic fragment from the D. virilis nonA region (CAMPESAN et al. 2001 ). This transgene included the corresponding D. virilis sequences of the adjacent lethal gene l(1)i19e, which lies in the promoter region of nonA (CAMPESAN et al. 2001 ; SANDRELLI et al. 2001 ). Lines 113, 67-4, and 168-8 have lost some 3' untranslated material from the transgene, but rescue completely the nonA^- ERG and viability defects (CAMPESAN et al. 2001 ). Males carrying more than one copy of nonA were generated with appropriate crosses.

Courtship song:
Males were recorded for 10 min while in the presence of virgin females at a temperature of 25°–26°, using an electret condenser microphone (RITCHIE and KYRIACOU 1994 ). For D. melanogaster, the wings of virgin females were removed with sharp forceps on collection, and when 1 or 2 days old, they were placed with 2- to 4-day-old males. For D. virilis, a single 15-day-old male and virgin female were placed into the recording chamber. Songs were filtered and analyzed using SPIKE2 software as described previously (RITCHIE and KYRIACOU 1994 ). In addition, the peaks and troughs corresponding to the cycles within each song pulse were detected and marked automatically for each song using one of the SPIKE2 subroutines (see Fig 1). Manual editing was used mainly to remove events caused by extraneous loud noises or to add or remove peaks and troughs that had not been marked properly in the automated analysis. All transformant songs were coded and analyzed "blind" by the experimenter. Further programming allowed the following parameters to be extracted from each song: the intrapulse frequency (PF) for each pulse and the slope of PF (bPF), as calculated by regression of PF against sequential pulse position in a song burst; the peak-to-peak mean interpulse intervals (IPI); the cycles per pulse (CPP) and the slope of CPP (bCPP; measured in the same way as bPF); and the sine song frequency. Only song bursts with 5 or more pulses were analyzed in this way. CPP or PF values at each sequential position in the burst were averaged for each male before performing least-squares regression and ANOVA. A minimum of 200 pulses per male were analyzed in this way, but more vigorous courtships produced >1000 pulses.

Optomotor test:
Three- to 8-day-old flies were dark-adapted for 4 hr and each fly tested individually for its turning behavior in a moving visual field of alternating black and white stripes (BURNET and BECK 1968 ). Ten flies per genotype were tested by placing each in a T-shaped tube in which the arm is painted black, so that the fly is forced to walk out of the opaque tube into a choice point facing the moving stripes. Once outside, it can turn into the right or left arm, and the fly produces the correct response when it turns in the direction of the moving environment. D. virilis flies are much larger than D. melanogaster, so we modified our methods so that flies were placed individually in an inverted empty glass vial positioned at the center of the rotating drum. When the fly reached the top, its movement was scored as the drum rotated. A correct optomotor response was obtained each time a fly performed at least five rotations in the same direction as the moving stripes. Each fly (melanogaster and virilis) was given 20 trials in which the direction of the stripe movement was randomly switched, with either narrow or broad stripes, each subtending an angle of 13.3° or 36°, respectively, from the center of the drum. The drum was rotated at a constant speed of 46 rpm, and a 60-W desk lamp was placed above the drum to uniformly illuminate the center of the cylinder. All tests were performed at room temperature (20°–22°).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Optomotor behavior:
The walking optomotor response is a sensitive test of the fly's capacity to follow movements of the visual environment and reflects more central aspects of the functioning of the visual system than the ERG (HEISENBERG and BUCHNER 1977 ). Hemizygous males carrying one copy of nonA^vir on the double deletion T(1:4)9e2-10 background (these transformants are collectively referred to as vir/d) were compared with their corresponding hemizygous nonAmel male transformant counterparts (mel/d), Canton-S males, partial hybrids carrying the endogenous melanogaster nonA gene plus nonA^vir (collectively termed vir/m), partial hybrids carrying two copies of nonA^vir plus the endogenous melanogaster nonA gene (mel + 2vir), along with males carrying two or three copies of melanogaster nonA (2mel, 3mel), nonAdiss, and D. virilis males.

Fig 2 reveals that the wider (larger) stripes generally gave more turning errors. Analysis of variance revealed a significant genotype effect (F = 5.96, P << 0.0001, d.f. 17,324), stripe width (F = 10.33, P = 0.0014, d.f. 1324), and a marginally significant genotype x stripe interaction (F = 1.71, d.f. 17,324, P = 0.039). Each vir/d transformant line gave a significantly higher number of correct turning responses than the nonAdiss mutants, which turn at random, generating an average of 10 correct responses out of 20 (P << 0.0001, Newman-Keuls a posteriori procedure). The mel/d transformant males rescue the optomotor response to a level similar to that of Canton-S with only lines 191/d and the partial hybrids 168/m having significantly poorer scores than mel/d (P = 0.038, 0.007, respectively). Consequently, the overall conclusion is that the optomotor defect associated with nonA mutants appears to be rescued in the great majority of vir/d lines to a level indistinguishable from that of mel/d.

View larger version (21K): In this window In a new window Download PPT slide   Figure 2. Optomotor responses. The mean number out of 20 of correct turning responses (plus SEM and SD) for broad stripes (open squares) and narrow stripes (solid diamonds) is shown for each genotype/line. #-/d represents a line carrying the relevant species hemizygous transgene in a nonA- background; #-/m represents the transgene in a wild-type melanogaster nonA+ background (see text).

Song analysis:
All mutant nonA transgenes so far examined either affect vision only, or vision and song, but never song only (RENDAHL et al. 1992 , RENDAHL et al. 1996 ; STANEWSKY et al. 1996 ). Both the ERG (CAMPESAN et al. 2001 ) and optomotor responses studied above suggest that nonA^vir gives a full rescue of nonA-mutant phenotypes. It therefore follows that any alterations of the song in a nonA^vir transformant are unlikely to reflect incomplete rescue (because we would expect an accompanying visual defect) and could instead indicate species-specific transfer of song information (see DISCUSSION).

It became clear during song recording that the vigor with which flies sang during the 10-min observation period was very different, one fly producing 15 sec of song and another 171 sec, representing the two extremes. A correlation matrix was generated between each song parameter, CPP, bCPP, PF, bPF, and IPI, and the amount of song generated (pulse plus hum song), but this did not reveal any consistent relationship for any of the 19 genotypes, in that the correlation could be positive or negative and usually not significant (data not shown). However, the correlations between the amount of song produced and PF were always negative, but only in one case (mel + 2vir) was it significant. Nevertheless, PF was corrected for the different amounts of song by analysis of covariance (ANCOVA).

CPP distributions:
One way of examining whether the nonA^vir transformants showed a complete rescue of the nonA mutant song phenotype is to examine overall CPP distributions. Fig 3A shows the results of the CPP distributions for males carrying a single copy of nonA; the nonA^diss males have a long tail in CPP frequency distribution but their modal value is 2, reflecting the mutants' developing polycyclicity as the song burst progresses (see also Fig 1). D. virilis males, on the other hand, have a distribution that is more normally distributed around a modal value of 4.5 CPP. The mel/d males have 95% of their pulses spread between 1, 1.5, and 2 CPP, with 35% having 1 CPP (see Fig 3A). Of the seven vir/d lines examined, six had modal values of 2 or 2.5, with one (97/d) having a modal value of 1.5 (Fig 3A). We examined the skew and kurtosis values for CPP for these genotypes by grouping together individual lines of vir/d after first checking the homogeneity of these lines with a Kruskal-Wallis test. Kolmogorov-Smirnov comparisons revealed that for kurtosis, the vir/d distributions differed significantly from mel/d (P < 0.05) and from nonA^diss (P < 0.001), but not from D. virilis (Fig 3A). In contrast, kurtosis in the partial hybrid vir/m group did not differ from control line me1/d, but was significantly different from D. virilis (P < 0.05) and from nonA^diss (P < 0.001). Therefore, this initial analysis of CPP distributions shows the vir/d lines to be more similar to D. virilis than to D. melanogaster (mel/d) and with no evidence for mutant CPP (>4) in any of the vir/d lines.

View larger version (40K): In this window In a new window Download PPT slide   Figure 3. Results of univariate analyses for courtship songs of all lines/genotypes. (A) Frequency histograms for CPP for males carrying a single nonA gene. diss, nonAdiss; mel/d, hemizygous nonAmel transformants; vir/d, hemizygous nonAvir transformants; D.vir, D. virilis. (B–E) Mean values, SEM and SD, are shown for (B) bCPP (slope of regression of CPP) for all lines except nonAdiss and D. virilis; (C) CPP at the first pulse (CPP1) of a pulse train taken from the regression (D) mean value of bPF (slope of pulse frequency); and (E) mean IPIs.

CPP regression on pulse position:
Another way of describing song characteristics is to examine the CPP regression in song bursts. Mutant nonA^diss songs generally show a steep positive slope (bCPP) in longer song bursts with pulses become increasingly polycyclic (RENDAHL et al. 1992 ; STANEWSKY et al. 1996 ). For each genotype, the number of songs out of 10 (only 6 for line 67-4/m) that had significant regression lines was calculated, and the significant slopes were scored as positive (increasing CPP with pulse position) or negative (decreasing CPP). The slope (bCPP) was calculated for each song and the mean value for bCPP for each transformant line is plotted in Fig 3B and excludes the nonA^diss and D. virilis values, which are off the scale of the y-axis. ANOVA of these data (excluding D. virilis and nonA^diss) gave a significant genotype effect (F = 4.26, d.f. 16,149, P << 0.001). Fig 3B reveals that the vir/d transformants have significantly higher bCPP values than mel/d (vir/d = 0.0158, mel/d = -0.0022; F = 17.3, d.f. 1149, P << 0.001). As expected (RENDAHL et al. 1992 ; STANEWSKY et al. 1996 ), the nonA^diss songs all had highly significant positive slopes (mean bCPP value = 0.257, off scale in Fig 3B). Of the 10 me1/d males, 4 gave significant regressions, 1 positive and 3 negative, with a mean slope of -0.0022. All individual vir/d lines, however, had between 7 and 8 songs with significant slopes, nearly all of which were positive. Partial hybrid lines (vir/m) had fewer significant slopes and this is reflected in their generally lower average bCPP values (Fig 3B). All D. virilis songs had significant positive slopes with a mean bCPP value of 0.186 (off scale in Fig 3B). This aspect of D. virilis songs had not been reported previously. Therefore, as in nonA^diss, D. virilis pulses become increasingly polycyclic during a song burst. Consequently, the positive bCPP values of the vir/d lines means that either rescue of this character is incomplete, or that this component of D. virilis songs has been transferred to the melanogaster hosts.

From the regressions we calculated the value of CPP at the first pulse (CPP1 in Fig 3C. Analysis of the transformant lines only revealed a highly significant genotype effect (F = 15.5, d.f. 16,149, P << 0.001). Planned comparison of mel/d vs. the vir/d values was highly significant (mel/d = 1.607, vir/d = 1.826; F = 16.4, d.f. 1149, P << 0.001). From Fig 3C it is clear that CPP1 values are significantly higher for most of the vir/d lines compared with me1/d (Newman-Keuls tests showed significant differences for lines 67-4/d, 168-8/d, 135/d, and 297-6/d compared with mel/d). Furthermore, partial hybrids vir/m also have significantly higher CPP1 values compared with vir/d lines (vir/m = 1.93; F = 14.87, d.f. 1149, P << 0.001). In general (with the exception of line 135), each individual partial vir/m hybrid line bears a higher CPP1 value than its vir/d corresponding counterpart (Fig 3C), possibly suggesting a dosage effect caused by carrying two nonA genes. However, comparing mel/d scores with 2mel and 3mel does not reveal any linear effect on CPP with increasing nonA dose (Fig 3C). Thus the vir/m values suggest a semidominant effect of adding the virilis copy of nonA to that of D. melanogaster (see DISCUSSION).

The three analyses of CPP described above reveal no convincing evidence that the vir/d transformants show characteristics that may reflect an incomplete rescue of the song phenotype. In contrast, they appear to share features of the D. virilis song.

PF regression on pulse position:
As a possible further indicator of the completeness of the rescue by nonA^vir we examined PF using a least-squares regression as with CPP. Five out of 10 nonA^diss mutant males showed a significant regression, 2 positive but 3 negative, with a mean slope (bPF) of 0.155 (Fig 3D). D. virilis songs were similar to the mutants in this respect, with 1 out of 10 males having a significant negative slope, giving a mean slope for the group of -0.63 (Fig 3D). The vast majority of flies in the other genotypes gave significant negative slopes, revealing that intrapulse frequency is systematically decreased during a song burst (RENDAHL et al. 1992 ). One-way ANOVA revealed a significant difference between the genotypes in bPF even when the nonA^diss and D. virilis values were excluded (F = 4.77, d.f. 16,149, P << 0.001). This difference is caused mainly by each vir/d line having a lower negative slope than its corresponding vir/m partial hybrid counterpart (Fig 3D, planned comparison, F = 41.08, d.f. 1167, P < 0.001). The vir/d lines have similar values to mel/d (F = 2.465, d.f. 1167, P = 0.118), revealing that these subtle changes in PF detected during a song burst are observed in both vir/d and mel/d songs, with no evidence for the mutant nonA^diss , nor the D. virilis-like phenotype.

As a further measure of intrapulse frequency, we calculated PF at the first pulse position in a burst (PF1) using the regression equation, then performed ANCOVA using the amount of song as covariate (see MATERIALS AND METHODS), which revealed a highly significant effect (F = 18.8, d.f. = 18,166, P << 0.001). The adjusted PF1 values revealed that nonA^diss has significantly lower frequencies of 250 Hz compared to the D. virilis value of 410 Hz (P < 0.01 by Newman-Keuls), whereas all the other genotypes gave adjusted PF1 values between 290 and 360 Hz that were significantly higher than nonA^diss (P < 0.05), but significantly lower than D. virilis (P < 0.05). We conclude that the vir/d transformants show no evidence for a lowered PF that might be indicative of incomplete rescue of the song phenotype.

IPI:
Mean IPI values of nonA^diss males are 41 msec and higher than those of mel/d males by 4 msec (Fig 3E). D. virilis males have very short IPIs of 17 msec. The vir/d lines showed significantly lower IPIs than mel/d (vir/d = 35.06 msec, mel/d = 36.67 msec; F = 5.3, d.f. 1167, P = 0.023), although in only two individual lines, 168-8/d and 135/d, was this significant by a posteriori comparison. Each partial hybrid line also showed higher IPIs than its vir/d counterpart (Fig 3E, F = 83.3, d.f. 1167, P << 0.001). There was an apparent dosage effect in that 3mel had significantly higher IPI than mel/d (Fig 3E, P < 0.01), with 2mel males having intermediate scores. These results were unexpected and suggest that the nonA^vir gene may have shortened IPI in the transformants, raising the possibility that nonA might carry species-specific IPI information. IPI values, as measured peak to peak, will be affected by the pulse pattern, which includes the variables that have been examined (CPP, bCPP, PF, and bPF). Consequently, we performed ANCOVA for IPI in which these four characteristics were covariates. The resultant F-ratio was also significant (F = 22.1, d.f. 18,163, P << 0.001) with the only major change from the previous analysis being that the adjusted IPI value for nonA^diss was slightly, but not significantly, lower that that of mel/d. Therefore, if degeneration of pulse pattern of the mutant is taken into account (reflected in its abnormal values of CPP, bCPP, PF, and bPF), the nonA^diss IPI is actually very similar to that of the wild type. However, every vir/d line still maintained a lower IPI than mel/d, which is surprising because vir/d males have higher values of CPP and bCPP, lower PF1 values, and generally more negative bPF. All these factors will serve to increase the duration of a pulse and thereby should indirectly increase IPI because of the way it was measured (peak to peak). Thus the lower IPIs of vir/d compared to mel/d suggest that the virilis transgene is playing an active role in reducing IPIs.

Multivariate analysis:
So far, all the analyses have used a single variable, compared between genotypes. To compare genotypes using all the data simultaneously, a multivariate method, discriminant analysis, was performed with the song characters bCPP, CPP, bPF, PF, and IPI using SYSTAT 7.0 software. All transformant lines were treated individually, and the results for all 19 groups are shown in Fig 4A. An extremely low probability of misclassification of groups is provided by Wilks' of 0.002 (F = 21.7, d.f. 90,790, P << 0.00001). Factor 1 accounts for >80% of the variance, which is primarily generated by the difference between D. virilis songs and the rest. Factor 2 (nearly 10%) is generated largely by the difference between nonA^diss males and the other genotypes. All five song characters contribute significantly (P << 0.0001) to the discrimination between groups.

View larger version (21K): In this window In a new window Download PPT slide   Figure 4. Results of discriminant analysis for song characters. (A) Individual values within D. virilis (squares), nonAdiss (circles), and mel/d (triangles) are represented together with their corresponding ellipses, giving the 95% confidence intervals. For clarity, all the other individuals from each genotype are not shown. (B) Analysis of transformant lines only (see text). Mean values of individual vir/d strains are represented as triangles, partial hybrids (either vir/m or mel + 2vir) as squares, and lines carrying one, two, or three copies of the melanogaster nonA gene, mel/d (1mel), 2mel, and 3mel, respectively, are shown as circles.

Clearly, the difference between D. virilis and the other genotypes is so large that any differences between the transformants are "swamped." We therefore analyzed the transformants only. Lines 113, 67-4, and 168-8, which may have lost some 3' regulatory material (CAMPESAN et al. 2001 ), did not appear any different from the "intact" lines in the univariate analyses. However, to be conservative and for simplicity, we removed these and their corresponding partial hybrids from the subsequent discriminant analysis, which nevertheless proved highly significant (Wilks' = 0.108, F = 5.49, d.f. 50,436, P << 0.0001), with all song characters contributing significantly to the discrimination (P < 0.0001). Fig 4B reveals that the vir/d lines (297/d, 135/d, 191/d, and 97/d) cluster away from the other lines on factor 1, which contributes 62% of the variance. A dosage effect is observed on factor 2 (21.4% of variance) for those flies carrying one, two, or three doses of D. melanogaster nonA, which is not reflected in the vir/m and mel + 2vir comparison. As the vir/m partial hybrids (lines 97/m, 135/m, 112/m, and mel + 2vir) have factor 1 values similar to the melanogaster nonA carriers (mel/d or 1mel, 2mel, and 3mel), the melanogaster nonA gene appears to be dominant with regard to factor 1.

Finally, we took each vir/d transformant line and compared it directly with its corresponding partial vir/m hybrid. In each of the five comparisons, involving lines 113, 135, 168-8, 67-4, and 97, there was almost no overlap between each pair of genotypes on factor 1. Wilks' was significant for all comparisons (P between 0.029 and << 0.0001). All CPP values contributed significantly to the discrimination of each pair, as did IPI in all but one line (line 97, P = 0.06). The other song characters, bCPP, PF, and bPF, contributed sporadically in only one or two pairings. To sum up, the multivariate analyses have buttressed the results of the univariate methods, in that they clearly show a difference between the song characteristics of the vir/d transformants and those flies carrying melanogaster nonA genes.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The courtship song of the nonA^diss mutant male shows some similarities to that of D. virilis, in that pulses are polycyclic. Our study has shown that this superficial similarity runs somewhat deeper, in that the D. virilis song burst also shows the same pattern of progressive polycyclicity as the mutant. The site of the nonA^diss mutation, in which an asparagine is substituted by cysteine (RENDAHL et al. 1996 ), although conserved in D. virilis, nevertheless falls within a seven-residue region in D. melanogaster, KREVDNE (residues 547–553; R is the nonA^diss site) that is altered to KRESDNE in D. virilis. This V to S substitution generates potential cyclic nucleotide-dependent and casein kinase II protein kinase phosphorylation sites in D. virilis (CAMPESAN et al. 2001 ). In addition, the nonA gene from D. virilis shows considerable divergence from its D. melanogaster orthologue, particularly in the N-terminal encoding regions (CAMPESAN et al. 2001 ). Consequently, it is logical to suggest that D. virilis nonA may have the potential to encode species-specific song characteristics.

Mutational analysis of nonA has revealed that the first RNA recognition domain (RRM1) in nonA is absolutely necessary for all the known functions of NONA, whereas mutations such as nonA^diss, which lie in the C-terminal charged region, not only cause severe defects in both visual and song phenotypes, but also invariably reduce the viability of the affected flies (RENDAHL et al. 1996 ; STANEWSKY et al. 1996 ). Further in vitro mutations in the second RRM2 domain left unaltered all the visual, courtship, and viability phenotypes tested or produced defects of visual behavior only (RENDAHL et al. 1996 ; STANEWSKY et al. 1996 ). It was inferred that NONA visual functions are particularly sensitive to alterations in the RRM domains, with courtship song also being affected whenever the charged region is mutated. An alternative interpretation would be that the array of phenotypes observed in nonA mutants results from increasingly severe, nonspecific alterations of the protein. Therefore, relatively mild nonA mutations would disrupt visual behavior only, while more severe protein alterations would also affect courtship song.

This pattern of defects, observed in nonA mutants described above, is thus extremely helpful for interpreting the behavioral results from the nonA^vir transformants. If nonA^vir is indeed generating a slightly mutant song phenotype in melanogaster hosts due to incomplete rescue, rather than a species-specific transfer of behavioral information, then we would predict that vision should be disrupted, as "vision is the first to go" in mutants (RENDAHL et al. 1992 , RENDAHL et al. 1996 ; STANEWSKY et al. 1996 ). In the optomotor responses measured here, broad and narrow stripes were used, the former representing a more challenging stimulus than the latter (HEISENBERG and BUCHNER 1977 ), but the nonA^vir transformants nonetheless appeared completely normal in this aspect of visual behavior. These transformants also have a normal ERG, underscoring the full rescue of the nonA mutant visual phenotype by the nonA^vir transgene (CAMPESAN et al. 2001 ). In contrast, the song characteristics that we have measured consistently show virilis-like features while maintaining sensitive indicators of "normal" melanogaster song ouput such as the subtle change in PF during a song burst (bPF). We also find evidence that nonA^vir appears to confer a virilis-like shorter IPI on the hosts, even though the other characteristics of the transformants' song pulses that have been measured should, indirectly, lead to a longer IPI.

These modest effects of the nonA^vir transgene in heterospecific hosts may occur because nonA is genuinely one of a number of genes that are involved in species-specific behavioral differences. Any nonA^vir-determined adaptive changes that have occurred in the courtship song during the 40–60 million years since the two species have been separated (MORIYAMA 1987 ) may require compensating changes in trans-acting factors for their full effects to be seen in the heterospecific D. melanogaster host transformants. In contrast, the nonA^vir transgene rescues basic visual functioning in transformants perhaps because there is little adaptive species-specific variation in this phenotype, so the relevant vision-specific trans-acting factors have not diverged significantly between the two species. Reciprocal interspecific transformations involving nonA transgenes from two very closely related species with very different song patterns, for example, D. virilis and D. littoralis, could conceivably generate more dramatic song phenotype exchanges because any relevant trans-acting factors would have had less time to diverge (HOIKKALA et al. 2000 ). Recent developments with interspecific transformation technology and gene knockouts could make such transformation experiments feasible (HANDLER et al. 1998 ; MISQUITTA and PATERSON 1999 ). It is further encouraging that the genetic mapping of song differences between D. virilis and D. littoralis has implicated the X chromosomal interval that carries the nonA locus (HOIKKALA et al. 2000 ).

We thus conclude that nonA carries species-specific song information for both pulse pattern and IPIs. This situation differs from that of the per gene, whose coding region controls the species specificity of both locomotor activity patterns and lovesong cycles in an all-or-none manner (PETERSEN et al. 1988 ; WHEELER et al. 1991 ). However, for the lovesong phenotype, only one species-specific component between the two closely related species D. melanogaster and D. simulans is determined by per (KYRIACOU and HALL 1986 ; WHEELER et al. 1991 ), whereas nonA^vir appears to move several song features toward a virilis-like pattern. Thus, nonA could be one of several genes that can evolve to generate novel patterns of courtship song, a species characteristic that shows enormous diversity. The transduced nonA^vir fragment that we employed also encoded the promotor regions of nonA^vir, so that any influences on species behavior could, unlike per, also be due to these regulatory regions, with which the host's transcription factors would interact. Clearly, chimeric nonA genes carrying both virilis and melanogaster material would be helpful in dissecting out whether these species-specific effects are due to coding vs. regulatory sequences.

How might the nonA^vir gene be mediating these effects in the transformants? The NONA protein does have characteristics of a "housekeeping" gene in that it is expressed ubiquitously during development and some nonA mutations can produce an almost lethal phenotype (RENDAHL et al. 1992 ; SANDRELLI et al. 2001 ). It could thus be imagined from its mutational spectra that NONA regulates mRNAs that are involved in the development of the visual system, the central nervous system, or musculature responsible for song production, as well as other tissues that are important for viability. However, the behavioral specificity of the defects in some of the nonA mutations mentioned earlier belie the housekeeping gene scenario painted above. The interspecific results reported here further suggest that nonA may play a more instructive role in the species-specific song pattern than simply providing the permissive conditions for courtship songs to be generated. A heat-shock-inducible nonA⁺ cDNA can rescue normal vision and song patterns in nonA mutants even if activated during the early adult stages (RENDAHL and HALL 1996 ). This would suggest that NONA is not required for directing the development of the relevant neuronal and muscular pathways toward the proper structures required for song production. Alternatively, a more dynamic instructive role for NONA might be reflected in the way the neuromuscular system stimulates song production, perhaps by altering responsiveness via changes in ion channel characteristics.

Such a speculative but seductive scenario would have nonA as a regulator (perhaps via splicing) of a downstream gene, such as the Dmca1A locus, which encodes a voltage-gated calcium channel and displays a complicated pattern of alternative splicing (SMITH et al. 1996 ). The mutant alleles of Dmca1A produce an array of courtship songs and visual and viability defects that are similar but distinguishable from nonA alleles (HEISENBERG and GOTZ 1975 ; VON SCHILCHER 1977 ; PERRIMON et al. 1984 ; KULKARNI and HALL 1987 ). Thus the phenotypic parallels between nonA and Dmca1A are striking and suggest a functional interaction between the two loci. The apparent semidominant effects of the D. virilis nonA gene on CPP suggest that a qualitative and quantitative comparative survey of the transcripts from the Dmca1A locus of D. virilis, as well as those from nonA^diss mutants, could prove worthwhile. This could be extended to other ion channel-encoding genes that, when mutant, generate defective lovesongs in D. melanogaster (PEIXOTO and HALL 1998 ).

In summary, the interspecific transformation of candidate genes is a more direct method of investigating species-specific characteristics than the reliance on hybridization between closely related species, the subsequent genetic analysis (if possible), and the subsequent laborious molecular work for identifying the relevant loci. Focusing on closely related species that are able to hybridize also limits the analysis, because only recent genetic variation will be scanned. In the case of nonA, transformation of the gene from D. virilis to D. melanogaster, two species that cannot hybridize, reveals the subtle behavioral effects of ancient genetic changes in the locus that must have occurred up to 60 mya.

	ACKNOWLEDGMENTS

We thank Martin Couchman for computing assistance. S.C. was supported by a grant from the Human Frontiers Science Program and European Union predoctoral fellowship. Y.D. was supported by the Wellcome Trust.

Manuscript received January 3, 2001; Accepted for publication May 4, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALT, S., J. RINGO, B. TALYN, W. BRAY, and H. DOWSE, 1998  The period gene controls courtship song cycles in D. melanogaster. Anim. Behav. 56:87-97[Medline].

BESSER, H. V., P. SCHNABEL, C. WIELAND, E. FRITZ, and R. STANEWSKY et al., 1990  The puff-specific Drosophila protein Bj6, encoded by the gene no-on-transientA, shows homology to RNA-binding proteins. Chromosoma 100:37-47[Medline].

BURNET, B. and J. BECK, 1968  Phenogenetic studies on visual acuity in Drosophila melanogaster. J. Insect Physiol. 14:855-860.

CAMPESAN, S., D. CHALMERS, A. MEGIGHIAN, F. SANDRELLI, and A. A. PEIXOTO et al., 2001  Comparative analysis of the nonA region in Drosophila identifies a highly diverged 5' gene that may constrain nonA promoter evolution. Genetics 157:751-764[Abstract/Free Full Text].

COYNE, J. A., 1992  Genetics and speciation. Nature 355:511-515.

HANDLER, A. M., S. D. MCCOMBS, M. J. FRASER, and S. H. SAUL, 1998  The lepidopteran transposon vector, piggyBac, mediates germ-line transformation in the Mediterranean fruit fly. Proc. Natl. Acad. Sci. USA 95:7520-7525[Abstract/Free Full Text].

HEISENBERG, M. and E. BUCHNER, 1977  The role of the retinula cell types in visual behavior of Drosophila melanogaster.. J. Comp. Physiol. 117:127-162.

HEISENBERG, M. and K. G. GOTZ, 1975  The use of mutations for the partial degradation of vision in Drosophila melanogaster. J. Comp. Physiol. 98:217-241.

HOIKKALA, A. and J. LUMME, 1984  Genetic control of the difference in male courtship sound between Drosophila virilis and Drosophila lummei.. Behav. Genet. 14:257-268[Medline].

HOIKKALA, A. and J. LUMME, 1987  The genetic basis of evolution of the male courtship sounds in the Drosophila virilis subgroup.. Evolution 41:827-845.

HOIKKALA, A., S. PAALLYSAHO, J. ASPI, and J. LUMME, 2000  Localization of genes affecting species differences in male courtship song between Drosophila virilis and D. littoralis.. Genet. Res. 75:37-45[Medline].

HOTTA, Y. and S. BENZER, 1969  Abnormal electroretinograms in visual mutants of Drosophila.. Nature 222:354-356[Medline].

HOTTA, Y. and S. BENZER, 1970  Genetic dissection of the Drosophila nervous system by means of mosaics. Proc. Natl. Acad. Sci. USA 67:1156-1163[Abstract/Free Full Text].

JONES, K. R. and G. M. RUBIN, 1990  Molecular analysis of no-on-transient A, a gene required for normal vision in Drosophila.. Neuron 4:711-723[Medline].

KULKARNI, S. J. and J. C. HALL, 1987  Behavioral and cytogenetic analysis of the cacophony courtship song mutant and interacting genetic-variants in Drosophila melanogaster.. Genetics 115:461-475[Abstract/Free Full Text].

KULKARNI, S. J., A. F. STEINLAUF, and J. C. HALL, 1988  The dissonance mutant of courtship song in Drosophila melanogaster: isolation, behavior and cytogenetics. Genetics 118:267-285[Abstract/Free Full Text].

KYRIACOU, C. P. and J. C. HALL, 1980  Circadian rhythm mutations in Drosophila melanogaster affect short-term fluctuations in the male's courtship song. Proc. Natl. Acad. Sci. USA 77:6729-6733[Abstract/Free Full Text].

KYRIACOU, C. P. and J. C. HALL, 1986  Interspecific genetic-control of courtship song production and reception in Drosophila. Science 232:494-497[Abstract/Free Full Text].

KYRIACOU, C. P. and J. C. HALL, 1989  Spectral analysis of Drosophila courtship song rhythms. Anim. Behav. 37:850-859.

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

MISQUITTA, L. and B. M. PATERSON, 1999  Targeted disruption of gene function in Drosophila by RNA interference (RNA-i): a role for nautilus in embryonic somatic muscle formation. Proc. Natl. Acad. Sci. USA 96:1451-1456[Abstract/Free Full Text].

MORIYAMA, E. N., 1987  Higher rates of nucleotide substitution in Drosophila than in mammals. Jpn. J. Genet. 62:139-147.

PAK, W. L., J. GROSSFIELD, and K. S. ARNOLD, 1970  Mutants of the visual pathway of Drosophila melanogaster. Nature 227:518-520[Medline].

PEIXOTO, A. A. and J. C. HALL, 1998  Analysis of temperature-sensitive mutants reveals new genes involved in the courtship song of Drosophila. Genetics 148:827-838[Abstract/Free Full Text].

PERRIMON, N., L. ENGSTROM, and A. P. MAHOWALD, 1984  The effects of zygotic lethal mutations on female germ-line functions in Drosophila. Dev. Biol. 105:404-414[Medline].

PETERSEN, G., J. C. HALL, and M. ROSBASH, 1988  The period gene of Drosophila carries species-specific behavioral instructions. EMBO J. 12:3939-3947.

RENDAHL, K. G. and J. C. HALL, 1996  Temporally manipulated rescue of visual and courtship abnormalities caused by a nonA mutation in Drosophila. J. Neurogenet. 10:247-256[Medline].

RENDAHL, K. G., K. R. JONES, S. J. KULKARNI, S. H. BAGULLY, and J. C. HALL, 1992  The dissonance nutation at the no-on-transient A locus of D.melanogaster: genetic control of courtship song and visual behaviors by a protein with putative RNA-binding motifs. J. Neurosci. 12:390-407[Abstract].

RENDAHL, K. G., N. GAUKHSHTEYN, D. A. WHEELER, T. A. FRY, and J. C. HALL, 1996  Defects in courtship and vision caused by amino acid substitutions in a putative RNA-binding protein encoded by the no-on-transient A (nonA) gene of Drosophila. J. Neurosci. 15:1511-1522.

RITCHIE, M. G. and C. P. KYRIACOU, 1994  Genetic variability of courtship song in a population of Drosophila melanogaster. Anim. Behav. 48:425-434.

SANDRELLI, F., S. CAMPESAN, M. G. ROSSETTO, C. BENNA, and E. ZIEGER et al., 2001  Molecular dissection of the 5' region of no-on-transientA of Drosophila melanogaster reveals cis-regulation by adjacent dGpi1 sequences.. Genetics 157:765-775[Abstract/Free Full Text].

SMITH, L. A., X. J. WANG, A. A. PEIXOTO, E. K. NEUMANN, and L. M. HALL et al., 1996  A Drosophila calcium channel 1 subunit gene maps to a genetic locus associated with behavioral and visual defects. J. Neurosci. 16:7868-7879[Abstract/Free Full Text].

STANEWSKY, R., K. G. RENDAHL, M. DILL, and H. SAUMWEBER, 1993  Genetic and molecular analysis of the X chromosomal region 14B17–14C4 in Drosophila melanogaster: loss of function in NON-A, a nuclear protein common to many cell types, results in specific physiological and behavioral defects. Genetics 135:419-442[Abstract].

STANEWSKY, R., T. A. FRY, I. REIM, H. SAUMWEBER, and J. C. HALL, 1996  Bioassaying putative RNA-binding motifs in a protein encoded by a gene that influences courtship and visually mediated behavior in Drosophila: in vitro mutagenesis of nonA.. Genetics 143:259-275[Abstract].

VON SCHILCHER, F., 1977  A mutation which changes courtship song in Drosophila melanogaster. Behav. Genet. 7:251-259[Medline].

WHEELER, D. A., C. P. KYRIACOU, M. L. GREENACRE, Q. YU, and J. E. RUTILA et al., 1991  Molecular transfer of a species-specific behavior from Drosophila simulans to Drosophila melanogaster. Science 251:1082-1085[Abstract/Free Full Text].

YU, Q., H. V. COLOT, C. P. KYRIACOU, J. C. HALL, and M. ROSBASH, 1987  Behaviour modification by in vitro mutagenesis of a variable region within the period gene of Drosophila. Nature 326:765-769[Medline].

This article has been cited by other articles:

				D. Sambandan, M. A. Carbone, R. R. H. Anholt, and T. F. C. Mackay Phenotypic Plasticity and Genotype by Environment Interaction for Olfactory Behavior in Drosophila melanogaster Genetics, June 1, 2008; 179(2): 1079 - 1088. [Abstract] [Full Text] [PDF]

				J. M. Gleason and M. G. Ritchie Do Quantitative Trait Loci (QTL) for a Courtship Song Difference Between Drosophila simulans and D. sechellia Coincide With Candidate Genes and Intraspecific QTL? Genetics, March 1, 2004; 166(3): 1303 - 1311. [Abstract] [Full Text] [PDF]

				M. J. Fitzpatrick and M. B. Sokolowski In Search of Food: Exploring the Evolutionary Link Between cGMP-Dependent Protein Kinase (PKG) and Behaviour Integr. Comp. Biol., February 1, 2004; 44(1): 28 - 36. [Abstract] [Full Text] [PDF]

				B. Chan, A. Villella, P. Funes, and J. C. Hall Courtship and Other Behaviors Affected by a Heat-Sensitive, Molecularly Novel Mutation in the cacophony Calcium-Channel Gene of Drosophila Genetics, September 1, 2002; 162(1): 135 - 153. [Abstract] [Full Text] [PDF]