but due to anatomical constraints much 16. R. C. Hardison, J. Biol. Chem. 258, 8739 (1983). Genoways, Ed. (Plenum, New York, 1990), vol. 2, 17. B. F. Koop, D. A. Tagle, M. Goodman, J. L. chap. 12. homoplasy resulted. Pettigrew argues that Slightom, Mol. Biol. Evol. 6 (no. 6), 580 (1989). 26. D. P. Mindell, C. W. Dick, R. J. Baker, Proc. Nati. "functionally obscure" neurological fea- 18. S. Harris, J. R. Thackeray, A. J. Jeifreys, M. L. Acad. Sci. U.S.A. 88, 10322 (1991); R. M. Adkins tures are less apt to be under selective Weiss, Mol. Biol. Evol. 3, 465 (1986). and R. L. Honeycutt, ibid., p. 10317. 19. S. G. Shapiro, E. A. Schon, T. M. Townes, J. B. 27. M. J. Stanhope, J. Czelusniak, J. Nickerson, M. constraints and similarities between pri- Ungrel, J. Mol. Biol. 169, 31 (1983). Goodman, unpublished data. mates and megabats are due to recent 20. F. Smith and M. S. Waterman, ibid. 147, 195 28. T. H. Jukes and C. R. Cantor, in Mammalian common ancestry rather than convergence (1981). Protein Metabolism, H. N. Munro, Ed. (Academic DNA 21. K. C. Beard, in Origines de la Bip6die chez les Press, New York, 1969), pp. 21-123. (7). Our E-globin noncoding data Hominid6s, CNRS, Ed. (Paris, France, 1991), pp. 29. E. M. Prager and A. C. Wilson, J. Mol. Evol. 27, strongly oppose the "flying primate" view 91-99. The tree placing flying lemur as the sister 326 (1988). of wing convergence during the descent of group of Primates adds only 2 to the lowest 30. We thank the Carnegie Museum for supplying two separate bat lineages, in support of the parsimony score of 3614 in agreement with the voucher specimen numbers and R. Baker at the weak strength of grouping result in Fig. 2 for the Texas Tech University Museum of Natural Histo- classical hypothesis of a monophyletic putative primate-tree shrew-rabbit clade. ry for supplying frozen tissue samples of C. Chiroptera and a common origin of mam- 22. P. H. A. Sneath and R. R. Sokal, Numerical variegatus, C. sphinx, M. lyra, and T. glis; the malian flight. Taxonomy: The Principles and Practice of Numer- Duke Primate Center for frozen tissues of T. ical Classification (Freeman, San Francisco, CA, syrichta; and the Yerkes Primate Center for 1973). frozen tissues of P. troglodytes, P. paniscus, G. REFERENCES AND NOTES 23. N. Saitou and M. Nei, Mol. Biol. Evol. 4, 406 gorilla, and M. mulatta. We also thank J. (1987). Czelusniak for providing us with his computer 1. C. Unnaeus, Systema Naturae per Regna Tria 24. The PTRFC parsimony program, written by J. programs, K. Hayasaka for providing the capu- Naturae, Secundum Classes, Ordines, Genera, Czelusniak to accommodate extensive sequence chin monkey subclone, C. Skinner for assistance Species, cum Characteribus, Differentuis, Syn- data sets, swaps branches by all possible bisec- in sequencing, B. Koop, M. Stanhope, D. Fitch, onymis, Locis, I. Tomus, Ed. (Holmiae, Impensis tions and pairwise reconnections as illustrated by and M. Miyamoto for helpful suggestions and Direct. Laurentii Salvii, Stockholm, 1758), pp. 18- D. L. Swofford and G. J. Olsen [in Molecular discussion. This study received support from 47. Systematics, D. M. Hillis and C. Moritz, Eds. NSF grant BSR-9007056 and NIH grant 2. J. R. Wible and M. J. Novacek, Am. Mus. Novit. (Sinauer, Sunderland, MA, 1990), p. 489, figure HL33940. 2911, 1 (1988). 22]. 3. R. J. Baker, M. J. Novacek, N. B. Simmons, Syst. 25. J. Czelusniak et al., in Current Mammalogy, H. H. 28 October 1991; accepted 30 January 1992 Zool. 40 (no. 2), 216 (1991). 4. M. J. Novacek, in Current Mammalogy, H. H. Genoways, Ed. (Plenum, New York, 1990), vol. 2, chap. 1 1. 5. J. D. Pettigrew, Science 231, 1304 (1986). 6. _ , Syst. Zool. 40 (no. 2), 199 (1991). Maternal-Effect Selfish Genes in Flour Beetles 7. J. D. Pettigrew et al., Philos. Trans. R. Soc. Lon- don Ser. B 325, 489 (1989). 8. J. D. Smith and G. Madkour, in Proceedings, Fifth R. W. Beeman,* K. S. Friesen, R. E. Denell International Bat Research Conference, D. E. Wil- son and A. L. Garder, Eds. (Texas Tech Press, A previously unknown class of dominant, maternal-effect lethal Mfactors was found to be Lubbock, TX, 1980), pp. 347-365. widespread in natural populations of the flour beetle, Tribolium castaneum, collected on 9. R. J. Baker, L. Honeycutt, R. A. Van Den Bussche, several continents. Such factors are integrated into the host chromosomes at variable in Contributions to Mammalogy in Honor ofKarl F. Koopman, T. A. Griffiths and D. Klingener, Eds. locations and show the remarkable property of self-selection by maternal-effect lethality to [Bull. Am. Mus. Nat. Hist., NY206, 42 (1991)]. all hatchlings that do not inherit a copy of the factor itself. Offspring are rescued by either 10. J. Czelusniak, M. Goodman, N. D. Moncrief, S. M. paternally or maternally inherited copies. The M-bearing chromosome is thereby perpet- Kehoe, in Methods in Enzymology, R. F. Doolittle, Ed. (Academic Press, San Diego, CA, 1990), vol. uated at the expense of its non-M homolog. M factors that map to different regions of the 183, chap. 37. genome do not rescue one another's maternal-effect lethality. Factors expressing these 11. M. Goodman, B. F. Koop, J. Czelusniak, M. L. properties are predicted to spread in a population, even in the absence of any additional Weiss, J. L. Slightom, J. Mol. Biol. 180, 803 factors also occur in the related T. confusum. (1984). selective advantage. Similar species 12. Scientific names, abbreviations, and GenBank accession numbers are as follows: human (Homo sapiens, Hsa), common chimpanzee (Pan troglo- dytes, Ptr, M81361), pygmy chimpanzee (Pan paniscus, Ppa, M81362), gorilla (Gorilla gorilla, Much of an animal's genome may be com- maternal-effect lethality to all progeny that Ggo, M81363), orangutan (Pongo pygmaeus, posed of parasitic or "selfish" DNA that do not inherit a copy of the factor itself. To Ppy), gibbon (Hylobates lar, Hla, M81377), rhesus serves no immediate function for the host our knowledge similar mechanisms are un- monkey (Macaca mulatta, Mmu, M81364), capu- chin monkey (Cebus albifrons, Cal, M81409), but rather exploits the host for its own known in the animal kingdom. tarsier (Tarsius syrichta, Tsy), galago (Galago propagation (1). Parasitic DNA may thrive In a screen for hybrid dysgenesis (3) crassicaudatus, Gcr), lemur [Lemur macaco (ful- and spread by replicative transposition, by between geographically diverse strains of T. vus) mayattensis, Lfu], flying lemur (Cynoceph- mechanisms in- castaneum, we found numerous cases of alus variegatus, Cva, M81365), tree shrew (Tu- segregation distortion, by paia glis, Tgl, M81366), megabat (Cynopterus volving supernumerary chromosomes (2) or reciprocal hybrid female semisterility. One sphinx, Csp, M81367), microbat (Megaderma by other non-Mendelian mechanisms. We example of such bidirectional, female-spe- lyra, Mly, M81368), rabbit (Oryctolagus cuniculus, report a novel mechanism by which a selfish cific semisterility, observed in hybrids be- Ocu), goat (Capra hircus, Chi). Sequences for PCR primers are as follows: A, 5'CACTGCTGAC- gene (or gene complex) may facilitate its tween strains collected in Singapore (SP) CCTCTCCTGACCTG-3'; B, 5'TCTCCATCCCT- own propagation at the expense of its un- and the United States (US) (4), is docu- CAGCCCATGAAC-3'; C, 5'GGAAGAGGCTG- selfish homolog. We show evidence for the mented in Fig. 1A. Crosses within each GAGGTGAAGCCTTGGG-3'; and D, 5'A-ATAAT- CACCATCACGTTACCCAGG-3'. The nucleotide widespread distribution in nature of a chro- strain and reciprocal crosses between strains sequence alignment can be obtained through mosomally integrated factor that confers are fertile. F1 hybrid males from either EMBL file server file name (ALIGN:DS8841.DAT). interstrain cross are fertile when back- Voucher specimen numbers are as follows: Me- to from either gaderma lyra, M81366; Cynopterus sphinx, R. W. Beeman, U.S. Grain Marketing Research Labo- crossed females parental M81365; Tupaia glis, M81367; and Cynocephalus ratory, Agricultural Research Service, U.S. Depart- strain. F1 hybrid females from either inter- variegatus, M81368. ment of Agriculture, 1515 College Avenue, Manhattan, strain cross are fertile when crossed to SP 13. F. S. Collins and S. M. Weissman, Prog. Nucleic KS 66502. K. S. Friesen and R. E. Denell, Division of Biology, males, but semistenile when crossed to US Acid Res. Mol. Biol. 31, 315 (1984). Kansas State KS 66506. 14. B. F. Koop et al., J. Mol. Evol. 24, 94 (1986). University, Manhattan, males. This semisterility is due to a preadult 15. D. A. Tagle et al., J. Mol. Biol. 203, 439 (1988). *To whom correspondence should be addressed. mortality rate of 50 to 80% among the SCIENCE * VOL. 256 * 3 APRIL 1992 89 IN progeny. Finally, F1 hybrid females crossed ever, when M/+ females are mated to +/+ (whether telomeric or otherwise) is un- to hybrid males show a more modest reduc- or M/+ males, a subset of progeny dies known. tion in fertility, manifested by a progeny during a period ranging from just prior to The observation that M au/+ + females mortality of about 25 to 50%. larval hatching through the second larval crossed to + au/+ au males yield only The semisterility exhibited by hybrid instar (5). The doomed larvae appear nor- aureate progeny (excepting rare recombi- females crossed to US males is also ex- mal with respect to external morphology nants whose genotypes were confirmed by pressed by all of their daughters when sim- but become sluggish, uncoordinated, or progeny testing) provides direct evidence ilarly mated (Fig. 1A). This faithful mater- paralyzed before death. for the hypothesis that the ensuing semi- nal transmission persists for many genera- We mapped the M factor derived from sterility is due to the death of zygotes tions of such backcrosses. However, the the SP strain by outcrossing M/+ females to inheriting the non-Medea maternal locus, observations that hybrid females derived males homozygous for various recessive and also demonstrates the complete pene- from either reciprocal cross are semisterile variants, and then backcrossing F1 daugh- trance of lethality. We also crossed M au/+ in crosses to US males and that hybrid ters to males of the paternal genotype. The + females to M au/M au males, and found males transmit the basis of semisterility to expectation is that linkage of a visible the 1:1 ratio of aureate and wild-type prog- some of their daughters show that this trait mutant marker to the M locus will result in eny expected when no semisterility occurs is not strictly maternally inherited. a decrease in the predicted 1:1 ratio of (Table 3). When these same females were These observations are consistent with mutant to wild-type progeny because of recrossed to + au/+ au males, the expected the hypothesis that the SP strain is fixed for preferential maternal-effect lethality to mortality of the non-aureate progeny class an autosomal dominant factor that acts progeny inheriting the marked (= non-M) was observed (Table 3). maternally to cause the death of any prog- homolog. (This strategy assumes that all Although these data suggest strongly eny which do not inherit it (Table 1 and laboratory stocks lack Medea factors, an that the zygote's maternally derived M Fig. 1B). Such a mechanism for genetic assumption consistent with results obtained chromosome can mediate zygotic rescue, it self-selection appears to be unprecedented. to date.) M assorted independently of all is not clear from the data in Table 3 We designate this factor Medea (M), an linkage groups tested except the third. whether the paternally derived zygotic res- acronym for maternal-effect dominant em- Mapping experiments (5) (Table 2) involv- cue factor is associated with paternal chro- bryonic arrest. We designate chromosomes ing four visible markers on linkage group mosomes, sperm cytoplasm, seminal fluid, not showing M behavior with a "+" sym- (LG) 3 showed that M is located about 1 or possibly other sources. To provide direct bol, although we do not know whether such unit to the right of aureate (au), making it evidence that zygotes are rescued from ma- chromosomes carry an alternative allele or the rightmost known marker for this link- ternal-effect lethality by paternally derived entirely lack a homologous locus. age group. LG 3 appears to reside on the M chromosomes as well as maternally de- M/+ females have no apparent abnor- longest autosome, a metacentric (6). The rived ones, we crossed M au +/+ au Bamp27 malities in ovarian morphology or oogene- au region of LG 3 is strongly disfavored as a males with M + +/+ au + females, then sis. They show normal fecundity, and fer- site for radiation-induced, semisterility-as- analyzed progeny phenotypes. If rescue tilization and embryonic development sociated rearrangements (7), but the posi- were not mediated by the paternally derived appear normal on gross examination. How- tion of this region on the chromosome M chromosome, then the Bamp, aureate phenotypic class should be equal in size to all other classes. The results (Table 4) are A B consistent with the hypothesis that the Usd'XUSqu9 sdXSP9 SPad X SP9 *4dX14X 44*dXfI9 MCPdX f paternally derived M chromosome itself confers rescue, but are inconsistent with the possibility that other chromosomes or extrachromosomal factors mediate such res- Fl PdFl9W* c],IrMg XUSd' XUe d XUSe XUS9 XUSd X4A4 X44e v4,4W'CX *d X44d cue. The observed frequency of Bamp, au- reate progeny (Table 4) matches that pre- 627) F297 2(6 F2CP w 2e) W* or * Ehiud e Whl el W We morbm Xusc XUse XusC XUsq Xus X *ec X *e& X *dO X *V, X *e0 dicted from normal meiotic recombination 5a27}* 2Ho8 up 4Age) -l. m bd between M and Bamp in the male with the assumption that only those sperm contain- $pdXUSg ~C X 1*/ ing recombinant M au Bamp chromosomes, and not sperm containing + au Bamp non- F10 recombinant confer rescue F16&(13) luhb19*0 4 + chromosomes, to zygotes that do not happen to inherit an XU8 X 14X) d XP9 Xi W9 X X *e xd 9U(6) ~711(169 0)~ xswid' gb xm.9~ "x IM M factor from their mother. Progeny re- P29 P29t P2 P \11)Att ceiving a single, maternally derived M fac- F29 F,F2,wewe CFe W*eWan( BOWS9 W+, W+ Wwor~ Xus & XUse Xus c], XUa9 XU& X 4*e X &*X *e X X 4 Set 4If 3316) 81)* Mr E m 1 Im a h . Table 1. Interactions of Medea-encoded cyto- Fig. 1. (A) Fertility (expressed as percentage of survival of progeny) for various crosses from type and genotype. M cytotype implies M/+ or parental strains derived from the United States (US) and Singapore (SP). Each datum is the mean M/M maternal genotype. The + cytotype im- deviation in of 50 crosses. determinations were based plies +/+ maternal genotype. Zygotic geno- (standard parentheses) single-pair Fertility of MIM is with + on pooled data, usually from two consecutive 3-day oviposition periods (total, 6 days oviposition type impossible cytotype. yielding about 50 to 60 eggs per female) or, in some cases, on a single 3-day oviposition period, with sexually mature young adults, 2 to 3 weeks old. Asterisk indicates progeny groups in which Cytotype Genotype Phenotype both fertile and semisterile daughters appear to be segregating. (B) Behavior of the Medea (M) M M/M Viable factor in crosses between MIM and +/+ strains. The flow diagram is redrawn from (A) and reflects M M/+ Viable the assumption that M is fixed in the SP strain and is absent from the US strain. Designation of M +/+ Lethal crosses as "fertile," "semisterile," or "mixed" suggests our interpretation of the data in (A), and is + M/+ Viable consistent with our proposed self-selection mechanism. Note that crossing M/+ males to +/+ + +/+ Viable females produces a mixture of semisterile (M/+) and fertile (+/+) daughters. 90 SCIENCE * VOL. 256 * 3 APRIL 1992 "Ir-.;","mi". tor (phenotypically Bamp, excepting re- from a strain collected in Pakistan, neither nally controlled self-selection mechanism combinants), as well as progeny receiving rescues nor is rescued by M' (5), although just described is that a Medea factor intro- M factors from both parents (phenotypical- the two are closely linked. M2 maps 4 duced into a wild-type population, even at ly wild type, excepting recombinants), were units to the left of au (toward the Bamp a low frequency, should gradually invade also rescued, as expected (Table 4). locus), or about 5 units from M'. In the population and become fixed, even in Although the death of +/+ progeny of addition, many MW-bearing strains harbor the absence of any selective advantage M/+ mothers occurs with complete pene- another M factor different from M' or M', beyond the immediate survival conferred trance, in some cases the rescue of M/+ or again by the criterion that they do not by resistance to its own maternal lethal M/M progeny does not. For example, in provide rescue from one another's mater- effect (11). This prediction is based on the Fig. 1A the mean percentages of progeny nal-effect lethality. Such factors have assumption that no loss of fitness is asso- survival in crosses involving M/+ (= F1) or been found in association with M' in ciated with the M factor other than the M/M (= SP) females were somewhat lower strains collected on all three continents reduced number of offspring that directly than the model predicts. This could have where M' has been found. Conversely, reflects the death of +/+ progeny of M/+ occurred because a portion of the M-bear- not all M-bearing strains carry M' (for mothers. Although the invasive nature of ing females deposited a higher than average example, the M2-bearing strain does not). M factors has not yet been demonstrated dose of lethal product in their eggs, result- It is not yet known how much heteroge- in laboratory populations, we have ob- ing in the death of a significant portion of neity exists among these additional factors served that most M-bearing individuals are their M/+ or M/M progeny. However, in in genomic location or interaction class, equal to wild-type in rate of development, other genetic backgrounds rescue was near- but they assort independently of LG 3. fecundity, longevity, and mating success ly complete (see Table 3). Maternal effects have previously been (5). Medea factors in strains sampled from implicated in interspecific hybrid sterility M-like factors are not restricted to T. dispersed natural populations show diver- or inviability within the genus Drosophila castaneum. We tested four strains of T. sity in both their genomic location and (8). Hutter et al. (9) described mutations confusum from four Asian countries and interactions. The one carried by the SP in Drosophila melanogaster that rescued the found evidence of an M-like factor in a strain (designated M') represents the most embryonic lethality of certain such hy- strain from Thailand. This factor con- commonly found type worldwide. So far brids. Such mechanisms have been inter- ferred dominant, female-specific semiste- we have found M1 in several strains col- preted as "complementary lethal" systems rility that was faithfully transmitted to all lected within the last 5 years in Asia, in which certain alleles at one locus are surviving daughters by heterozygous moth- Africa, and South America. In all cases lethal in the presence of certain alleles at ers but that segregated in heterozygous the factor maps to the same position, other (normally complementary) loci. males (5). Horizontal transmission of M approximately 1 unit to the right ofaureate Build-up of such complementary lethal factors between these two species may on LG 3. However, not all factors that systems is considered to be an important have occurred through an infectious map to this region are M1. M2, derived mechanism for the evolution of reproduc- agent, but could not have occurred by tive isolation (expressed as hybrid invia- normal gene flow in recent evolutionary Table 2. Mapping of an Asian Medea (M) factor bility) during speciation in both animals time, because crosses between the two on the third linkage group (LG). Percentage of and plants (10). Cases of reproductive species are entirely sterile. The existence crossover (in cis or trans) between M and the incompatibility in animals that combine of M-like factors in two congeneric species indicated visible LG 3 marker was tested by maternal and zygotic influences appear to that cannot interbreed raises intriguing backcrosses to appropriate homozygous re- involve multilocus interactions and occur questions about the distribution of such cessive (non-Medea) testers. In crosses involv- ing lod, both parents were homozygous pearl. in hybrids between separate species or elements more broadly within the genus Data indicate percentage of visible mutant clearly distinct subspecies. In contrast, the and at higher taxonomic levels. The (trans tests) or wild (cis tests) phenotype in Medea system appears to be entirely con- Medea system could be viewed as an un- progeny. Male data merely reflect percentage trolled by any one of several autonomous of segregation of visible mutant gamete, and not and independently functioning loci in a Table 4. Zygotic rescue by paternal or maternal Medea factors. Test cross was M au au recombination, because the M phenotype being species that shows no other evidence of +/+ scored is maternal effect lethality to non-M prog- Bamp male x M + +/+ au + female (five of into races. eny. Recombination occurs equally in the two subdivision genetic each sex). The Bamp au phenotypic class (in sexes in Tribolium. ND, not determined. A predicted consequence of the mater- boldface) stems from meiotic recombination between M and Bamp in the male. All other Male Female Table 3. Meiotic segregation of Medea in fe- classes derive from noncrossover chromo- males. Fifteen M aul+ + females (dams) were somes. Data represent two complete oviposi- tested individually. Data are expressed as tions. Predicted number of progeny is based on Genotype of Appar- Prog- Appar- Prog- means ± 1 standard deviation. Each female the assumption of tight linkage between au and test parent cross eny cross eny was crossed first to a pair of M au/M au sires, M, 13% recombination between au and Bamp crovss scored cross scored then to a pair of + aul+ au sires. Fertility is (the published value), and 297 total progeny. (%)r (no.) (%)r (no.) defined as percentage of survival of progeny and is based on all eggs laid during a 3- to Prog- Ob- Pre- Mater- Pater- 6-day oviposition period. Progeny phenotypes served dicted M +/+ au 46 273 1 8926 ey nal nal are based on counts from all progeny pro- Mau/+ + ND 0 241 prog- prog- chromo- chromo- M +/+ Bamp 46 383 9* 509 duced in the first 3 days of oviposition after a pheno- interval. type eny eny some some MBamp/+ + 52 702 9 1455 3-day premating (no.) (no.) M+/+ lod 46 384 15 387 Mlod/+ + 50 606 21 802 Progeny of given + 105 99 M++ Mau+ Bamp 94 99 M+ + + au M +I+ b 50 141 43 295 Fertility phenotype (no.) Mb/+ + 50 307 41 241 Sire genotype Bamp + au au 85 86 +au+ Mau+ *Confirmed by test crosses. All Bamp recombinants Bamp au 13 13 + au + M au appeared mosaic or aneuploid. Several apparent so- 93 3 22 + 4 21 5 Bamp matic mosaic Bamp progeny had wild-type (non- Mau/Mau Bamp) germlines. + au/+ au 46 + 12 0 _ 0 16 5 Total 297 297

SCIENCE * VOL. 256 * 3 APRIL 1992 91 ...... 1 1 usual strategy for the self-propagation of 5. R. W. Beeman, K. S. Friesen, R. E. Denell, unpub- is an immunoglobulin G1/K, known from lished data. selfish DNA that can also function as a 6. R. W. Beeman and J. J. Stuart, J. Econ. Entomol. immunochemical data to strongly recog- postzygotic isolating mechanism to facili- 83, 1745 (1990). nize residues 2 through 5 and 11 and tate an ongoing process of speciation. 7. R. W. Beeman, T. R. Johnson, S. M. Nanis, J. weakly recognize residues 1 and 9 of CsA Hered. 77, 451 (1986). (14). Messenger RNA sequencing of its REFERENCES AND NOTES 8. H. A. Orr, Evolution 43, 180 (1989); Genetics 1 16, 555 (1987). variable region revealed an unusually long 1. L. E. Orgel and F. H. C. Crick, Nature 284, 604 9. P. Hutter, J. Roote, M. Ashburner, Genetics 124, H3 loop (15). The structure of CsA in the (1980); W. F. Doolittle and C. Sapienza, ibid., p. 909 (1990). Fab R4545 1 1-CsA complex, derived here 601. 10. P. Christie and M. R. Macnair, J. Hered. 75, 510 from our crystallographic analysis, is 2. U. Nur, J. H. Werren, D. G. Eickbush, W. D. Burke, (1984); Evolution 41, 571 (1987). T. H. Eickbush, Science 240, 512 (1988). 11. M. J. Wade, personal communication. shown in Fig. 1A. The difference electron 3. M. G. Kidwell, J. F. Kidwell, J. A. Sved, Genetics 12. We thank J. F. Crow, D. L. Harti, M. S. Thomson, density map clearly shows the shape of the 86, 813 (1977); G. Picard, Biol. Cell. 31, 235 and M. J. Wade for critical reading of the manu- CsA cycle as well as the location of side (1978). script, and S. U. Khattak and S. H. Ho for provid- chains. The conformation is different from 4. The SP strain was derived from beetles collected ing beetle strains. Supported by ARS and by in 1989 in a rice warehouse in Senoko, Singapore; USDA research grant 8901004. that of the isolated form (Fig. 1B), also the US strain was collected in a farmer's corn bin obtained by x-ray diffraction (8). Howev- in Georgia in 1980. 28 October 1991; accepted 1 1 February 1992 er, the conformation of CsA in interac- tion with the antibody closely resembles that recently observed, with NMR analy- ses, in a complex with CYP (10, 11). The conformation of isolated CsA (Fig. iB) A Conformation of Cyclosporin A in Aqueous contains two antiparallel 1 strands formed Environment Revealed by the X-ray Structure of a by residues 11 to 3 and 4 to 7. All four nonmethylated main-chain nitrogens are Cyclosporin-Fab Complex involved in internal hydrogen bonds with carbonyl groups. In the structure of CsA Daniele Altschuh,* Olivier Vix, Bernard Rees, shown in Fig. 1A, no internal hydrogen Jean-Claude Thierry bonds are formed. Instead, main-chain nitrogens and carbonyls are available for The conformation of the immunosuppressive drug cyclosporin A (CsA) in a complex with hydrogen bonding to the protein or to the a Fab molecule has been established by crystallographic analysis to 2.65 angstrom res- solvent. The internal structure of the cy- olution. This conformation of CsA is similar to that recently observed in the complex with cle is mainly hydrophobic because most the rotamase cyclophilin, its binding protein in vivo, and totally different from its confor- main-chain N-methyls, as well as the side mation in an isolated form as determined from x-ray and nuclear magnetic resonance chain of MeVal5, point to the inside of the analysis. Because the surfaces of CsA interacting with cyclophilin or with the Fab are not cycle. identical, these results suggest that the conformation of CsA observed in the bound form The connection between residues Me- preexists in aqueous solution and is not produced by interaction with the proteins. Leu9 and MeLeu10 is best fit by a trans peptide bond, as in the conformation of CsA in interaction with CYP (10, 1 1), in contrast to the cis peptide bond found in Cyclosporin A (CsA), an immunosuppres- chain conformation. The conformation of isolated CsA (Fig. 1B). Although topolog- sive drug that acts as an inhibitor of T cell CsA in aqueous solution has never been ically similar overall, the conformations of activation (1), forms a complex in vivo studied because of its poor solubility. Two CsA bound to the Fab or to CYP (10, 1 1) with cyclophilin (CYP) (2), a cis-trans NMR studies ofCsA in a complex with CYP are not identical. For example, when CsA isomerase, and can inhibit its activity in yielded, however, a conformation of CsA is bound to the Fab, MeBmt' does not fold vitro (3, 4). The molecular mechanisms of with a totally rearranged backbone confor- back onto the molecule as when CsA is immunosuppressive activity are not under- mation (10, 11). The cis peptide bond found bound to CYP (MeBmt, 3-hydroxy-4- stood. Furthermore, the relation between between residues MeLeu9 and MeLeul0 (Me, methyl-2-methylamino-6-octenoic acid). immunosuppressive activity and inhibition N-methyl) in free CsA was replaced by a Differences, however, are small compared of rotamase activity has recently been ques- trans peptide bond in the conformation of to those of the totally unrelated conforma- tioned [for review, see (5, 6)1. Knowledge of CsA in interaction with CYP. tions of CsA in complex with the Fab the conformation of the drug that is recog- Is the radical conformational rearrange- (Fig. 1A) and isolated CsA (Fig. 1B). nized by CYP is central to an understanding ment of the whole CsA molecule a result The antibody is unlikely to distort CsA of its mechanism of action. Two crystallo- of its binding to CYP (12, 13)? Although through a rotamase activity that would graphic (7, 8) and two nuclear magnetic some conformational adaptation may oc- mimic CYP. Indeed, the binding modes of resonance (NMR) analyses in different apo- cur upon interaction of CsA with an the Fab and of CYP are different: CYP is lar solvents (9) of isolated CsA yielded antibody combining site, such drastic known from NMR experiments to recog- superposable structures, suggesting that CsA changes are not expected from its free nize residues 1 through 3 and 9 through 11 is a rigid molecule with a distinctive main- conformation in solution. We have estab- of CsA (10, 1 1). Crystallographic analysis lished by crystallographic analysis to 2.65 of the Fab-CsA complex clearly shows a side chain D. Altschuh, Laboratoire d'immunochimie, Institut de A resolution the structure of CsA in that residues 9 and 10 and the Biologie Mol6culaire et Cellulaire du CNRS, 15 rue complex with the Fab of an antibody and of residue 1 are not involved, whereas Descartes, 67084 Strasbourg, France. analyzed its interaction with the antibody residues 11 and 2 to 5 and the main chain 0. Vix, B. Rees, J.-C. Thierry, Laboratoire de Cristal- of residue 1 are critical for binding (Fig. lographie, Institut de Biologie Mol6culaire et Cellulaire combining site. du CNRS, 15 rue Descartes, 67084 Strasbourg, The monoclonal antibody to CsA 2). These latter residues are also those France. R454511, which was chosen for crystallo- found from immunochemical data to be *To whom correspondence should be addressed. graphic analysis of an Fab-CsA complex, important for recognition (14). 92 SCIENCE * VOL. 256 * 3 APRIL 1992