Circadian rhythm mutations in Drosophila melanogaster affect

Transcription

1 Proc. Nati. Acad. Sci. USA Vol. 77, No. 11, , November 1980 Genetics Circadian rhythm mutations in Drosohila melanogaster affect short-term fluctuations in the male's courtshi song (interulse interval/er mutations/dilo-x transformed males) C. P. KYRIACOU AND JFFRY C. HALL Deartment of Biology, Brandeis University, Waltham, Massachusetts Communicated by Vincent G. Dethier, August 7,1980 ABSTRACT Courtshi song in Drosohila is roduced by the male's wing vibration and consists of ulses of tone roduced at intervals of aroximately 34 msec in D. melanogaster and 48 msec in D. simulans. We have observed that the intervals between these ulses are not constant but fluctuate rhythmically with eriods of aroximately 1 min in D. melanogaster and 0.5 min in D. simulans. In D. melanogaster, three allelic er mutations have been isolated which affect the eriodicity of the circadian oscillators affecting both eclosion and locomotor activity [Konoka, R. & Benzer S. (1971) Proc. Nat]. Acad. Sci. USA 68, ]. ach of the er alleles ers, which shortens the circadian eriod, erl, which lengthens it, and ero, which abolishes it-strikingly affects the 60-sec song rhythm in a arallel fashion. Therefore, both circadian rhythms and a very short, noncircadian oscillation aear to be influenced by the same gene. Courtshi in Drosohila melanogaster consists of a series of highly stereotyed behavioral atterns, the most consicuous of which is the male's wing dislay (1, 2). The male extends and vibrates his wing, roducing a atterned acoustic signal or "song" (3). The song is illustrated in Fig. 1 and has several acoustic comonents, including a train of ulses with an interulse interval (ii) of aroximately 34 msec together with a sinusoidal hum, which has a frequency of 160 Hz (4). The ii is secies secific; males of D. simulans, a closely related secies, sing with an ii of aroximately 48 msec (5). Von Schilcher (6) has reorted that D. melanogaster females mate faster with wingless, deafened consecific males when they are exosed to an artificial song with an ii of 34 msec comared to one having an ii of 48 msec. Therefore, the ii of the courtshi song may lay an imortant role as a recognition factor in reserving the sexual isolation between the two secies. This aer describes some exeriments that demonstrate that the variation in the ii values roduced by males of both D. melanogaster and D. simulans is greater than reviously believed. However, we will show that this variation conforms to a highly structured temoral attern which is dramatically affected in D. melanogaster by three circadian rhythm mutations isolated by behavioral criteria unrelated to courtshi (7). MATRIALS AND MTHODS Drosohila stocks were cultured on cornmeal/agar/ molasses/yeast medium. Unless stated otherwise, all were reared at 250C in a light-dark cycle of 16 hr in the light and 8 hr in the dark. Two wild-tye strains were used, Canton-S and Oregon-K. The courtshi song was recorded by lacing a single 3- to The ublication costs of this article were defrayed in art by age charge ayment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C solely to indicate this fact. a W Pulse song 1m,.1""0---LaidI I,1 ww Sine song ll -i 50 msec FIG. 1. Courtshi song of D. melanogaster males. 5-day-old male with a 3-day-old virgin attached-x female inside a transarent mating cell 2 X 1 X 0.3 cm high. The floor of the chamber was comosed of Nytex (stiff nylon mesh), and the unit was laced 2 mm above the ribbon of a Reslo (Sheffield, ngland) ribbon microhone, which was then encased within a screened aluminum box (12 X 8 X 8 inches) which served as an anechoic chamber. A orthole was cut in the box to enable the exerimenter to observe the flies. The song of the male was amlified and recorded on a Tandberg lox reel-to-reel tae recorder. A visual record of the song was obtained by tracing the recording, together with a time reference, on light-sensitive aer (Duont, Tye 5B) with a multichannel oscillograh (Consolidated lectrodynamics, Newark, NJ, tye 5-124). The aer was exosed, and ii were measured eak to eak with a ruler. The courtshi was divided into 10-sec fractions, and the mean and SM of all ii falling within each successive time eriod was comuted. The minimum number of ii used to obtain a mean was 10, and most means were based on between 25 and 70 individual ii. These means were then used to comute a nonlinear regression of ii against the successive time eriods for each courtshi. The data were fitted to a variety of nonlinear mathematical functions, and the regression line of best fit was estimated by the method of least squares. The BMDP series of comuter rograms was used for this urose (8). The residual variation after fitting a articular function was then used to comute an F ratio for the goodness-of-fit (9). Various arameters of the song were extracted for each courtshi with these analyses, and when comarisons were made between different genotyes, analysis of variance was used together with the Newman-Keuls rocedure for making a osteriori tests between means (10). RSULTS The courtshi songs of wild-tye Canton-S and Oregon-K males were recorded for 5-6 min or until coulation had occurred. Fig. 2 illustrates tyical ii rofiles for males of each genotye. As the courtshi roceeds, the ii roduced by the males follow a sinusoidal ath. The standard errors of the oints are such that the ii roduced at the crest of the sine wave are nearly always significantly longer than those generated at the trough. When these oints were fitted to a sine function, regression analysis Abbreviation: ii, interulse interval(s)

2 6730 Genetics: Kyriacou and Hall D B ; i- + ~ r t M' ' *! c -I t iy, f 35 d4 ll 32 / < FIG. 2. Courtshi song rofiles of Oregon-K (A) and Canton-S (B-) wild-tye males. The means + SM of all ii falling within each consecutive 10-sec eriod of courtshi are lotted together with the best least-squares solution of the regression analysis and the eriod of the ii oscillation. Arrows in C- indicate where a male resumed his courtshi song after interrutions of between 20 and 40 sec. A, = 53 sec; B, = 53 sec; C, = 49 sec; D, = 63 sec;, = 52 sec. revealed highly significant reductions in the total variation. Consequently, the ii roduced by the males can be considered to be oscillating. From the equation describing the best leastsquares fit (see legend in Table 1), we can directly obtain the eriod of the oscillation (), the amlitude (a), and the value around which the ii oscillate (mean ii). The courtshis illustrated in Fig. 2 all show values around 55 sec, a values of aroximately 2-3 msec (or 4-6 msec eak to trough), and mean ii of aroximately 32 msec. Table 1 summarizes the results of the song analyses erformed with the Canton-S and Oregon-K strains, together with similar analyses of yellow males (made nearly isogeneic with the Canton-S strain); males carrying the cuticular markers yellow, singed3, and miniature; dilo-x flies, which were transformed to henotyic males with the third chromosome tra mutation (11); and halo-x males homozygous for tra or heterozygous with tra and the TM6 balancer chromosome. The reasons for studying these mutant males will become clear in the following sections. The results in Table 1 show that even though the males exress a wide variety of mutant genes, they all show similar song atterns with resect to the ii eriod, which is always close to 55 sec, and the amlitude a, which is within the range msec. Interestingly, the most variable character of the song aears to be the value around which the ii oscillate (mean ii), which varies from 30 to 42 msec. Only 1 of the 44 males analyzed failed to show a statistically significant oscillation of ii; in 30 males it was significant beyond the 1 level. There are two striking features of these ii rofiles. First, males do not all initiate their courtshi song at the same hase of the sine wave. This can clearly be seen in the song rofiles of Fig. 2. Second, if a male ceases to sing for one or more of the 10-sec courtshi eriods, as often is the case, when he resumes his song he remains in hase with the ii oscillation reviously t, I Proc. Natl. Acad. Sci. USA 77 (1980) Table 1. Temoral analysis of ii of the courtshi song No. of Mean ii males F SM, a + SM, i SM, Genotye tested sec msec msec Canton-S ± yellow ± Oregon-K ± ysn3m ± ± XX/Y; tra/tra i i ± 0.4 X/Y; tra/tra ± ± 1.2 X/Y; tra/tm i ± 1.2 See text for descrition of genotyes. The raw data consisted of the mean ii for each consecutive 10-sec eriod from the initiation of the courtshi song (see Fig. 1). A least-squares nonlinear regression was comuted by using the sine function f(x) = bi + b2eb3x + (b4 + b5eb6x) sin(b7 + b8x), in which b, is the value of ii around which the sine wave oscillates and which can ascend or descend by incororating the b2eb3x arameter (mean ii = bi + b2eb3x); b4 is the amlitude, which can similarly increase or decrease by using the term b5eb6x (a = b4 + b5eb6x); b7 is the angular dislacement of the sine wave at the first data oint (i.e., when X = 10 sec); and b8 is the eriod of the oscillation. Of the 44 song rofiles examined, 30 showed significant reduction of the total variation on fitting the sine function beyond the 1 level, 12 beyond the 5 level, 1 beyond the 10 level (showed two successive oscillations out of a ossible four), and only 1 was not significant (i.e., considered arrhythmic). initiated. The arrows suerimosed on the three courtshis in Fig 2 C- indicate where a male continued his song after interrutions of sec. These two characteristics of the song rhythm suggest that the ii oscillation is couled to some ongoing temoral rocess. ffects of the er Gene on Song Rhythm. Konoka and Benzer (7) isolated three X-linked mutations in D. melanogaster that alter circadian behavior with resect to two characteristics, eclosion and locomotor activity. These mutations aear to affect one functional gene termed the er locus. One mutation, ers, shortens the normal 24-h rhythm to 19 h, whereas a second, erl, lengthens it to 28 h. The third mutant, er, is arrhythmic. Because oscillations are involved in both circadian behavior and the song rhythm, we wondered whether the er mutations might also affect the exression of the ii attern. We recorded the songs of males carrying each of the er alleles, with each individual mutant gene being tested on two different genetic backgrounds exressing various combinations of the cuticular markers yellow, singed 3, and miniature. Fig. 3 shows the ii rofiles for tyical mutant songs together with a er+ male for comarison. The difference between the er+ and mutant males is clearly illustrated. The eriod of the ii oscillation was reduced to 40 sec in the ers male and lengthened to 76 sec with erl. The er0 male revealed no systematic sinusoidal atterning in his song. Regression analysis of 12 er0 songs failed to find a mathematical function, linear or nonlinear, that could significantly reduce the total variation. We therefore conclude that the ii rofile for er0 is arrhythmic. A summary of these results is resented in Table 2. The two genetic backgrounds on which each of the mutant genes was studied have been ooled because the results with each were essentially identical. The mean eriodicity of the ii oscillation in ers males is 41 sec and that of erl is 82 sec. Analysis of variance revealed that these values are highly significantly different from those of the Canton-S wild-tye strain (P < 0.001) No significant differences were observed in the other song arameters. Therefore, the er mutations have similar effects on the eriodicities of both the 55-sec courtshi song rhythm and the 24-h circadian henotye.

3 Genetics: Kyriacou and Hall Proc. Nati. Acad. Sci. USA 77 (1980) 67 c;._r._ 361 -I t, I"+t' f'a I 4,, f ".."k.f 1~~~. hk-' -;+i st , ' y~~~~~~~~~I 33 t i I ti 4 5 ~~~~~~~~~~~~~ FIG. 3. Courtshi song rofiles of the er mutants and a er+ male (yellow on a Canton-S genetic background). (A) er+, = 56 sec; (B) er8, = 40 sec; (C) er1, = 76 sec; (D) er0, =?. ffects of Light and Temerature Shifts on Song Rhythm. Constant light conditions cause arrhythmia in many circadian rhythms (12). Having uncovered an unexected relationshi between the circadian and courtshi song rhythms, we imagined that if the circadian oscillator "drives" the one mediating the song, then rearing males in constant light might disrut the song attern. We raised wild-tye flies in constant light conditions for five generations, then counted the number eclosing every 8 hr. We observed the characteristic arrhythmic attern when comared to the same stocks reared for 16 hr in the light and 8 hr in the dark. The songs of 10 wild-tye males raised in constant light were then recorded and analyzed. There was no evidence of any change in the normal attern of the song rhythm and, consequently, the oscillation in the courtshi song aears to be insensitive to changes in the light-dark cycle. One of the most rominent features of circadian rhythms is that they are temerature comensated in that changes in temerature do not significantly affect their eriodic characteristic (13). We were therefore interested in seeing whether rearing males and recording their songs at relatively high and low temeratures would affect their ii attern. Canton-S flies were reared at 290C and the male's courtshi song was recorded Table 2. Temoral analysis of the courtshi song of er", er, and ero males ±5+ SM, a ± SM, Mean ii ± SM, Genotye sec msec msec er" k erl er0? and er Twelve males of each genotye were tested. Of the 24 er8 song rofiles, 17 showed significant reductions of the total variation on fitting the sine function beyond the 1 level, 5 beyond the 5 level, and 2 beyond the 101 level, with all er0 males not significant in this resect. See Table 1 for exlanation of symbols W ' 281,( 350C f /#,. / it', It /Ali /" / / t ), y ), f I t f #' / 1,/ /4.fI I/ # / ".1, i I FIG. 4. Courtshi song rofiles of two Canton-S wild-tye males, one reared at 290C and recorded at 350C ( = 51 sec) and the other reared and recorded at 161C ( e 53 sec). in the usual way at 350C after 1 hr of acclimation at this temerature. Males were also reared at and their songs recorded at 16'C. The song rofile of one male at each temerature is illustrated in Fig. 4. Although low and high temeratures have the effect of, resectively, raising and slightly lowering the overall mean ii (see ref. 14), the eriodicity of the song oscillation remains essentially the same. Although significant differences were obtained between the mean ii values at the high and low temeratures (32.2 i 1.0 comared with msec, resectively; P < 0.001) and amlitude a (1.9 ± 0.3 comared with msec, resectively; P < 0.05), no significant differences could be detected in the eriod of ii oscillation between the two grous (high-temerature j = sec; lowtemerature j5 = sec). These results are based on nonlinear regression analyses for 7 songs at 16'C and 6 songs at 350C, of which 8 showed significant reductions in total variation on fitting the sine function beyond the 1 level, 4 beyond the 5 level, and 1 beyond the 10 level. Heterozygosis for the er Locus. Males heterozygous for the various er alleles were constructed by making dilo-x flies that were transformed into henotyic males by the tra mutation. Transformed males are sterile, but they show normal courtshi behavior and song rhythm (see Table 1). All heterozygous combinations of the different er alleles were generated, together with transformed males carrying a er allele and a small deletion of the er locus, Df(l )w (7). ach het- Table 3. Temoral analysis of courtshi songs for transformed males heterozygous for er alleles or Df(1 )w No. of Mean ii males + SM, a I SM, P SM, Genotye tested sec msec msec er'/er eri/er ers/erl I ero/er I ero/ers i 0.8 ero/erl er+/df er'idf i i 0.7 erl/df ± Of the 98 song rofiles examined, 66 showed significant reductions of the total variation of fitting the sine function beyond the 1 level, 23 beyond the 5 level, and 9 beyond the 10 level. See Table 1 for exlanation of symbols. a f,

4 6732 Genetics: Kyriacou and Hall 55 f C;) 52.-r 49.4 FIG. Note h oscillat sec. Of the tote one sh( erozyg;ous combination of er alleles was studied with resect to the ( courtshi song on two different genetic backgrounds, and the res The er+ h from t dominl the er hetero; ers ar 52 sec) tained variati( They o to thevvild-tye allele. However, eri when the allele is made also show that the effect on the song rhythm mas to the er heterozfygous with er+, the eriod length of the ii oscillation locus and is not the result of some other mutation or multigene is signif Sicantly reduced to 40 sec comared with the wild tye combination fortuitously resent in our Drosohila strains. (P 01). Similarly, the era allele significantly reduces the We are currently interested in searching for the secific <0. re- of both the er and oer allelesz(p song eiriod <. )Ts on of the nervous system that is resonsible for the roduction result issnot arallel to the findigs of Konokca and Benzer (7) of of the th mutant song rhythms. Stocks are being reared which muatsn with cihrcadian rhythms in females, in which the ero mutant htm.stck ar ben rrd,hc allele w rhyessiveothms males, w atlh serv mthat will enable us to generate genetic mosaics, which are Drosohila tas recessive to the other alleles. we also observed individuals comosed of both halo and dilo-x tissue. The lacing the deficiency Df(1)wtec with each of the other eri dilo-x tissue will, for examle, exress the heterozygous alleles s imly mimicked the effect of 1er0 heterozygosis. v combination er /er and the halo-x tissue will exress only dently I ferent e Song of D. si songs of five males were recorded, and Fig. 5 illustrates one of behavior. Because the mutant song can be unambiguously the ii Irofiles obtained. The ii of this male also showed an oscillati simulan much sa suggests the ossibility that the er gene may secify a roduct + Ii*,T, that controls a fundamental roerty of temoral regulation I'-", 1 f, '1,'1',lt i~iin this secies. There are several examles of short-eriod osj;,i'+if,' 'a,t' JIB II,' 'I cillating rocesses; e.g., the raid, 0.5-min glycolytic oscillation \.'}.iobserved '4, in yeast (15, 16) and the 5-min oscillation of ATP levels in Acetabularia cells (17). The question of whether such short-eriod oscillations may share common features with the Time, mmn more universal circadian variation has been reviously discussed (18), but until now there has been no evidence to suort this 5. Courtshi song rofile of ad. simulans male ( = 36 sec). notion. iow this male resumes his courtshi song inhasewiththe ii The gl f h hhms f ht t ion he had initiated, even after a comaratively long lull of 110 grlenicana ostrs the five songs examined, four showed significant reduction of is arallel in most resect te to son the results ruts obtained ober by eerozygoe Konoka al variation in fitting the sine function beyond the 1 level and and Benzer (7) with the circadian rhythm of locomotor activity. Owed reduction beyond the 10 level. The ers allele is dominant with resect to the song rhythm and semidominant in circadian locomotor activity variation. Similarly, the er1 allele is recessive in relation to both characters. However, we observed that lacing er0 with any of the other ;ults are resented in Table 3. er alleles always resulted in a significant shortening of the eriod of the song rhythm. This can be interreted as if ero mean eriod length- for the song oscillation in ers/ is a semidominant acting in the same direction as ers. Because eterozygotes is 43 sec, which is not significantly different the same effect is observed when each er allele is made hetthat of ers males. Thus, the ers mutation shows a erozygous with the deficiency Df(1l)w +242, this also suggests ant henotye with resect to this character. In contrast, Proc. Natl. Acad. Sci. USA 77 (1980) '1allele is recessive to the wild-tye allele, with er1/er Beze ( oseredftht occurring. In contrast, Konoka and zygotes having an ii oscillation eriod of 53 sec. When rhyhm ofsheterzy o recessive in fas Howevr cirxadian id er' laced together, a wild-tye henotye (5 = ryhso eeoyosfmls oeearcsiex,detare aed ther, aw e hote linked allele in females could have a different effect in transis Kobt kandṭhesethre ruts aresimilar to those ob- formed males than in regular heterozygous females due to the by Konoka and Benzer (7) withrese tciadian ossible influences of sex er se on X-chromosome dosage n or locomotor activity in heterozygous er remales. comensation in Drosohila (e.g., see refs. 19 and 20). Our observed that erhwas semidominant and er1 recessive results with the er heterozygotes (e.g., ers/ero, ers/df) ero is a null allele (see ref. 7) but aears to have dif- the er1 allele. The mutant and heterozygous tissue will be 1-ffects on circadian and song rhythms. identified by the resence of an internal histological marker of D. simulans Males. We asked if the courtshi song (21). These mosaics will be transformed to henotyic males mulans males also showed oscillation of the ii. The bythe tra mutation so that they will exhibit normal courtshi distinguished from that of the heterozygote (see Tables 2 and on which is suerimosed uon the longer ii of D. 3), we can correlate the internal distribution of halo-x and is. The eriod of the oscillation in this case is 36 sec, dilo-x tissue with the song henotye exressed by the fly. iorter than that for wild-tye D. melanogaster males Mosaic analysis of the er locus has been erformed with (see Table 1). The mean oscillation eriod for this grou of five resect to circadian variation in locomotor activity (22). The males w,as 33.2 i 1.0 sec, and the amlitude of the oscillation results suggest that the mutant focus for the ero allele is in the was msec. Both of these values are significantly dif- brain. Translanting the brain of one er allele into the abdo- the corresonding ones of Canton-S wild-tye males men of another and observing that the reciient may take on ferent from of D. me anogaster (P < 0.01). Therefore, D. simulans males the circadian characteristics of the donor has also raised the also exhiibit an oscillation of the ii of the courtshi song, but ossibility that a neurohumoral factor may be involved (23). it has a ssignificantly shorter eriod and larger amlitude than The song oscillation may also have a focus in the brain. Howhe closely related D. melanogaster males. ever, some genetic mosaic data imlicate the thoracic nervous that of t] DISCUSSION system as an imortant comonent of the song roduction mechanism (24). The effects of circadian rhythm mutations on short-term os- In addition to being studied neurobiologically, the song cillationss in Drosohila courtshi behavior illustrate how mu- rhythm can also be studied from an evolutionary ersective. tant gen es isolated with resect to one system can have ro- The finding that D. simulans males roduce songs that are found influence on another. Indeed, it might have been ex- markedly different from those of D. melanogaster both in the tremely cdifficult to demonstrate a relationshi between the song amlitude and eriod of the ii oscillation suggests that the rate oscillation and circadian rhythm without using a genetic a- of change of ii in the two secies may have some functional roach.' The fact that we have uncovered such a relationshi significance. Perhas it is a factor involved in secies recogni-

5 Genetics: Kyriacou and Hall tion which comlements the basic difference in ii between the two secies. We thank Laurie Tomkins and Kalana White for their comments on the manuscrit. We are grateful to Ronald Konoka for sulying the mutations and Anne Gross for her technical assistance. The work was suorted by a North Atlantic Treaty Organization fellowshi to C.P.K. and U.S. Public Health Service Grant GM lus a Research Career Develoment Award (U.S. Public Health Service GM-00297) to J.C.H. 1. Bastock, M. & Manning, A. (1955) Behaviour 8, Burnet, B. & Connolly, K. (1974) in The Genetics of Behaviour, ed. van Abeleen, J. H. F. (North-Holland, Amsterdam), Bennet-Clark, H. C. & wing, A. (1967) Nature (London) 215, von Schilcher, F. (1976) Anim. Behav. 24, wing, A. & Bennet-Clark, H. C. (1968) Behaviour, von Schilcher, F. (1976) Anim. Behav. 24, Konoka, R. & Benzer, S. (1971) Proc. Nati. Acad. Sci. USA 68, BMDP Biomedical Comuter Programs P-Series (1977) (Univ. Of California, Los Angeles). 9. Daniel, C. & Wood, F. S. (1971) Fitting quations to Data (Wiley, New York). Proc. Natl. Acad. Sci. USA 77 (1980) Winer, B. J. (1971) Statistical Princiles in xerimental Design (McGraw-Hill, London). 11. Sturtevant, A. H. (1945) Genetics 30, Pittendrigh, C. S. (1974) in The Neurosciences Third Study Program, eds. Schmitt, F. 0. & Worden F. G. (MIT Press, Cambridge, MA), Pittendrigh, C. S. (1954) Proc. Natl. Acad. Sci. USA 40, Shorey, H. H. (1962) Science 137, Chance, B., stabrook, R. W. & Ghosh, A. (1964) Proc. Natl. Acad. Sci. USA 51, Pye,. K. (1973) in Biological and Biochemical Oscillators, eds. Chance, B., Pye,. K., Ghosh, A. K. & Hess, B. (Academic, New York), von Klitzing, L. (1969) Protolasma 68, Klein, A. 0. (1979) in Proceedings of the uroean Symosium on Photomorhogenesis, ed. De Greef, J. (Univ. Antwer, Belgium), in ress. 19. Goldschmidt, R. (1954) Science 119, Komma, D. (1966) Genetics 54, Kankel, D. & Hall, J. C. (1976) Dev. Biol. 48, Konoka, R. (1972) Dissertation (California Inst. of Tech., Pasadena, CA). 23. Handler, A. M. & Konoka, R. (1979) Nature (London) 279, von Schilcher, F. & Hall, J. C. (1979) J. Com. Physiol. A 129,