Annotated Bibliography

Levels of naturally occurring DNA polymorphism correlate with recombination rates in D. melanogaster David J. Begun & Charles F. Aquadro

Secti on of Genetics and Development, Biotechnol ogy Buildi ng, Cornell University, Ithaca, New York 14853 - 270 3 . USA

Two genomic regions with unusally low recombination rates in Drosophila melanogaster have normal levels of divergence but greatly reduced nucleotide diversity'· 2, apparently resulting from the fixation of advantageous mutations and the associated hitch­ hiking effect 3.4. Here we show that for 20 gene regions from across the genome, the amount of nucleotide diversity in natural popula­ tions of D . melanogaster is positively correlated with the regional rate of recombination. This cannot be explained by va riation in mutation rates and/or functional constraint, because we observe no correlation between recombination rates and DNA sequence divergence between D. melanogast er and its sibling species, D. simulans. We suggest that the correlation may result from genetic hitch-hiking associated with the fixation of advantageous mutants . Hitch-hiking thus seems to occur over a large fraction of the Drosophila genome and may constitute a major constraint on levels of genetic variation in nature.

T a ble I summ a riz es le v el s o f DNA v a ri ati o n a nd inte rs pe cifi c di ve rg enc e ( b et wee n D. m ela nogaste r a nd D. simulan s) wh ere av ail a bl e. T he se estimate s of D NA pol y morphi sm are d eri ve d fr o m re stri ction site survey s w ith o ne e x cepti o n ( cu bitu s int errup­ rus) a nd th erefore a r e estimate s o f average l eve ls of va ri ati o n ov er 13 to 65 kil o ba ses ( kb ) from eac h ge n e re g ion. T o explor e the rel ati o n ship b et we en levels o f D N A sequen ce v a ri a tion a nd r eco mbin ati o n ra tes we co mp are d estim at es o f nu cle otid e di ve r ­

5 sit y ( w ) and th e coe ffi c i e nt of ex ch an ge\ a mea sur e of r ec ombi ­ n ation ra t e p er p h ys i ca l d ista n ce .

E 0.014

.?:­ ·;;; ID 0 .010 > i5 (I) 1::> 0.006 -~ (I)

u ::, z 0.002

•

O l!H~~~~~~~~-~~~-~~~~~~~~--

0 0.02 o.o• 0.06 o.oa 0 .1 Coefficient of exchange

FIG. 1 Scatte rplot of nucleoti de divers ity ( 7T) versus coef ficient of exchange in D. melanogaster. Autoso mal and X-linked genes are represented by hatched and closed circles. respect ively. To make autosoma l and X-linked genes direc t ly compa rable we made the simplifying assu mpti on of equal numbers of males and fema les. Then, under neutrality , nucleot ide heterozy­ gos it y is an estimate of 3N µ. fo r X-linked genes and 4Nµ. fo r autosomal genes (N is the e ff ective populat ion size and µ. is the neut ral mut ation rate) . Therefo re, befo re doing regress ion or co rrelat ion analyses. we mult i­ plied est imates of 7T from X-linked gene regions by fou r-th irds . The recombi­ nat ion rates est imated by the coefficie nt of exchange are for fema les. An X-linked gene reg ion spends two -th irds of t he t ime in fema les (where it can recombine) and only one-thi rd of the time in males (where it canno t recom­ bine ). wherea s an autosome spends half it s t ime in fema les and half in males. Therefore , we multi plied the coeff icient of exchange fo r autos omal genes and X-linked regio ns by one-half and two- t hirds. respective ly. Regressio n line is indicated by a solid line.

NATURE · VOL 356 · 9 APRIL 1992

LETTERS TO NATURE

TABLE 1 Coefficient s of exchange and nucleotide heterozygosities in D. me/anogaster and divergence with D. simulans

Coefficient Gene region of exchange " Divergence Reference

Chromosome I (X) yello w.achaete (y, ac) 0.0045 0.001 0.054 1 phosphogluconate

dehydrogenase gene (Pgd) 0.0154 0.003 0.029 1

zest e- tko (z, tko) 0.0222 0.004 12 period (per) 0.0520 0.001 0.050 1 white (w) 0.1400 0.009 13 Notch (N) 0.1212 0.005 14 vermilion (v) 0.0590 0.001 0.047 (D.J.B. and C.F.A.,

unpublished results)

forked ( f) 0.0455 0.002 15 glucose-6.phosphate

dehydrogenase gene (Z w) 0.0485 0.001 16

suppre ssor of fork ed (sul f)) 000 50 0.000 15

Chromosome II sn-glycerol 3-phosphate

dehydogenase gene (Gpdh) 0.0800 0.008 17

alcohol dehydrogenase gene (Adh) 0.0647 0.006 0.045 18,19

DOPA decarboxylase IC.FA et al .. gene (Ode) 0.0184 0.005 unpublished

results) amylase gene (Amy) 0.0435 0.008 20 Punch (Pu) 0.0718 0.004 17

Chromosome Ill esterase-6 gene (Est-6) 0.0604 0.005 21 metallothionein-A gene

(MCnA) 0.0083 0.001 0.072 22 heat.shock protein. 70A

gene (Hsp70A) 0.0069 0.002 0.023 23, 24 rosy (ry) 0.0471 0.003 0.050 25

Chromosome IV cubitus in terruptus

Dominant (ci°) o• 0.000 0.050 2

Nucleotide diversity (,r) is the average pairwise difference for all pairs of sequences drawn at random from a population, and can be thought of as heterozygosity per nucleotide5 . The coefficient of exchange for a gene region was calculated by selecting two genetically defined loci26 that f lank the region of interest and dividing the distance in map units between the flanking loci by the number of polytene bands between the loci6 . The number of polytene bands between loci was determined from Bridge's maps27_ An important assumpt ion underlying the use of thi s me t ric as an index of recombination rate is that over large stretches of the genome (for example. 20 to 40 polytene bands), the average amount of DNA per polytene band is roughly similar between regions. Available data suggest that this is a reasonable assumption for at least much of the Drosophila genome 6 28 2 " .

• The recombination rate on the fourth chromosome is effectively zero30

F i gu re I is a sca tt erpl o t o f w ve r su s th e coeffici ent of e xc h a n ge fo r 20 ge ne s in D. m ela nogaste r. It i s a p pare n t t h at l e v els o f nucl eo tid e d ive r si ty i n crease as r ates of r eco mbin a t io n in c r ease. Varia ti o n in reco mb ina ti on r at es ex p la i ns a l arge fr act ion o f th e va ri at i on i n nucl eo tid e di v er sity a n d t he null h y p o th es i s th at th e sl ope i s ze ro i s rej ec t ed with hi g h p ro b a b i l i t y (F , = 16.8, P = 0.0007). Th e n o n -pa r a m etri c Spea rm a n a nd K e nd all r egress i o n t es t s are a l so si g nifi ca ntl y diff ere nt from ze ro ( Spea r­ m a n 's D = 544, P < 0.0 1; K end a ll 's r = 0.437, P < 0 .0 1) . T h e sa m e co n cl u si o n i s reac h ed w h e n d ata from th e white reg i o n ( w hi ch h as a pa rt i cul a rl y hi g h l eve l o f va ri a t io n in D. me lanogas ter) is ex clu d ed fr o m the an a l ys i s ( F, = 5.8, P = 0 .03; Spe ar ma n 's D = 544, 0.02 < P < 0.05; K end a ll 's r = 0.374, 0.02 < P < 0.05).

O n e h y p o th es i s to expl a in thi s tr end is tha t gene reg i o n s in areas of reduced reco mbi na t io n h ave l owe r n eu tra l m ut a tion rates. Per h aps recom bin a tio n itse l f i s mut age n ic. If th is w er e t ru e, th en und e r a n e u tra l m o d el th ese gene reg i o n s sho uld a l so be l ess di verged be t wee n spec i es th an gene reg i ons i n areas of

519 © 1992 Nature Publishing Group

LETTERS TO NATURE

0.08

0.06 QJ 0 C

~ 0.04 Q) > o 0.02

e

•

• •

•

0 .1-,-~~~-.--~~ ....... ~---,.--~,-~ ....... ~-.--~--.-

0 0 .010 0.020 0 .030 0.040

Coefficient of exchange

greater recombination rates 7. In Fig. 2 we show a plot of diver­ gence between D. melanogaster and D. simulans versus the coefficient of exchange, including a ll gene region s for which we have estimate s of divergence. Clearly, estimate s of divergence from a larger number of gene regions are desirable. Nevertheless, the lack of a significant positive regression coefficient (F , = 0.001, P == 0.983) with the available data argues against the hypothesis that gene regions in areas of low recombination rates have, on average, lower substitution rates.

Theoretical results show that at the time offixation of a neutral variant , the amount of linked neutral variation is reduced, and that the magnitude of the reduction depends on th e recombina­ tion rate 8 • But at a random time (which is an y time a genomic region in a population is sampled), the average amount of neutral nucl eotide polymorphism is unaffected by the recombi­ nation rate 8 9• . Therefore, we are unable to arrive at a sa tisfactory neutral explanation for the patterns seen in Figs 1 and 2.

We propose that the positive correlation between DNA vari­ ation and recombination rate results from th e selective fixation of advantageous mutants over a significant portion of the genome. This correlation suggests that levels of neutral va riation in man y of the gene regions for which variation has been measured have been reduced by one or more hitch -hiking events. Provided that a new selectively favoured mutation goes to fixation before another advantageous mutation arises close to it, each fixation will be surrounded by a ' window ' of redu ced polymorphism, the relative size of which is proportional to the rate of recombination for that region of the genome 4 • Thus, where recombin ation rates are very low, ea ch fixation will cause a wide window of reduced polymorphism, wherea s in reg ion s of higher recombination, the window will be proportionatel y smaller . Mo ving along a chromosome tow ards regions of pro ­ gressively lower recombin ation , the windows becom e clos er , and ma y begin to overl ap sub stanti ally. Thu s, re gions of low

Received 19 December 1991; accepted 7 Febru.Jry 1992 . 18 . Langley, C. H., Montgomery. E_ & Quat tlebaum, W. F. Proc. natn. Acad. Sci. US.A. 79 , 5631-5635 (1982) .

19 . Aquadro, C. F., Deese, S. F .. Bland, M. M., Langley, C. H. l aurie~Ahlberg, C. C. Genetics U4 , 1 . & Begun, D J. & Aquadr o . C. F. Genetics 129 , 1147 - 1158 (1991 ). 1165-1190 (1986) 2. Berry, A . J .. Ajioka. J. w. & Kr ei tman . M. Genetics 129 . 1111 -11 17 (199 1).

20. Langley. C. H. et al. Genetics 119 , 619 - 629 (1988 ). 3 . Maynar d Smit h. J. & Ha igh, J. Genet . Res . 23, 23 - 35 (197 4). 21. Game. A . Y. & Oakes ho lt. J. G. Genetics 126 , 1021-1031 (1990) . 4 . Kaplan . N. L.. Hudson , R. R. & Langley. C. H. Genetics 123. 887 - 899 (1 989) 22. Lange, B. W .. Langley . C.H . & Step han . W . Genelics 126, 92 1-93 2 (1990) 5. Nei, M . Molecular £vo futionary Genetics (Columbia Univ. Press . 1987) 23 . J. 80. 5350-5354 (1983). 6 . Linds le y , D. L. & Sand ler , L. Phil. Trans. R. Soc B. 277, 295 - 312 (19 77). Leigh Brown. A. Proc. natn. Acad . Sc( US.A. 24 . J. D. 290 , 677 - 682 (198 1). 7. Kimura, M. The Neutral Theory of Molecular Evolution & (Cambridge Univ. Press. 1983 ). Leigh Brown, A. !sh -Horow itz . Natu re 25 . Aq uadro . C. F .. 125, Lado . K. M. & Noon . W A. Genetics 119, 87 5 -888 (19881. 8 . Taji ma, F. Geneti cs 447 -4 54 (1990) .

R. 26 . Ashbumer, M. OrosophHa Genetic Maps (Drosophila Information Service 69. 1991) . 9 . Hudson . R. Theor. Popular . Biol . 23. 183 - 20 1 (1983) . 27. Lindsley. D. L. & Grell, E. H. Gene tic Variations of Drosophila melanogaster (Carnegie Inst itute, 10 Birky. C. W. Jr & Wa lsh . J. B. Proc. natn A cad Sci . US .A 85 , 6414-64 18 (1988)

Washington DC. 196 7) 11. McDonald, J, H. & Kre itman, M. Na ture 351, 65 2- 65 4 (1991) . 28. Sousa. V. Chromosome Maps of Drosophila (CRC, Boca Raton. Florida. 1988) . 12 . Aguade. M ., Miyashita , N. & Langley. C.H. Molec. Biol. E:vol, 6, 123 - 130 (19881 . 29 . J .. & 254 . 221 - 225 (1991) . 13 . Miya sllita . N. & L angl ey. C. H. Genetics 120. Merriam. Ashburner . M.. Hartl , D . L. Ka fatos. F. C. Science 199 - 212 (19881.

w., c. 5. 30 . Hochman. B. in Genetics and Biology of Drosophila Vol. 1b (eds Ashburner, M. & Novitski, E.) 14 . Schaeffer. S. A qua dro , C. F. & Langley, H. Mol ec. Biol. Evol. 30 - 40 (1988) . 903 - 928 (Academ ic. New York, 1976) . 1 5 . Langley, C. H. in Population Biology of Genes and Molecule s (eds Takahata, N. & Crow. J F.)

75 - 91 (Baifuk an. Japan). 16 . Eanes . W. F., Aj ioka, J. W .. Hey, J. & Wes ley, C. Molec. Bio/. £vol. 6, 38 4 - 397 (1989). ACKNOWLEDGEMENTS. We thank M. Nachman f or comments and all members of our laboratory for 17 . Takano . T. S .. Ku sakabe . 5 . & Mukai. T. Genet ics 129, 753 - 761 (1991) . discussion. This work was supported by the NIH and NSF

520

FIG. 2. Scatterp lot of sequence divergence between D. me/anogaster and D. simulans versus coefficient of exchange in 0. melanogaster. Autosomal and X-linked genes are represented by hatched and closed circles. respec­ tively . Coefficients of exchange for X-linked and aut osomal regions are modified as described in Fig. 1 legend. Regression line is indicated by a solid line.

recombination are 'hit' by selective sweep s more often, keeping polymorphism at a lower average level. Mutations driven to fixation by meiot ic drive or biased gene conversion would ha ve similar evo lutionary consequences. Hitch-hiking does not affect interspecific divergence'° , con sistent with the observed lack of a corre lation between DNA sequence divergence and recombi­ nation rate .

McDonald and Kreitm a n 11 propo sed that patterns of synony ­ mous and non-synonymous va riation at the Adh locu s in and between three Drosophila species are incompatible with neutralit y. They sugg ested that selective fixation of amino -acid polymorphisms at this locus is the best expl a nation for their data. Furthermore, they speculate th at selective fixations occur at a large number of loci. Our results are consistent with thi s view . However, the number of sele ctively favoured nucleotides relative to the size of the genome could still be quite small 4 •

C learly, hitch-hiking and reco mbination rate s do not explain all of the heterogeneity in levels of var iation across the D. melanogaster genome. Variation in mutation rate and functional constraint , as well as several different forms of selection, have roles in shaping local leve ls of DNA sequence variation . But the analy sis pre sented here pre se nts the first eviden ce th at hitch ­ hiking driv en b y selective fixation of new mutations may con­ stra in le vels of nucleotide pol ymorphi sm over large portions of the D. melanoga ster genome. Inferenc e of effective population size from levels of DNA variation may be co mpromis ed by this phenomenon.

Much effort is being expended to assemble physical and genetic maps in sever al species. An unexpected benefit of these genome mapping projec ts is that it will be possible to examine wheth e r correla tion s between recombination rates and levels of DNA variation a re a general phenomenon in natural populations o f other taxa , including human s. D

NATURE · VOL 356 · 9 APRIL 1992

© 1992 Nature Publishing Group