Reduction of testis growth of pseudaletia separata larvae after parasitization by cotesia kariyai.код для вставкиСкачать
Archives of Insect Biochemistry and Physiology 26:lll-122 (1994) Reduction of Testis Growth of Pseudaletia separata Larvae After Parasitization by Cotesia kariyai Toshiharu Tanaka, Eiko Tagashira, and Sho Sakurai Applied Entomology, School of Agriculfural Sciences, Nagoya University,Ckikusa,Nagoya (T.T., E.T.), and Department of Biology, Faculty of Science, Kanazawa University,Kanazawa (S.S.), Japan The braconid endoparasitoid, Cotesia j=Apante/es) kariyai physiologically influences its host, Pseudaletia separata, through three factors: polydnavirus, venom, and teratocytes. Inhibiting testis development of the host seems to be one factor that is important for successful development of the parasitoid. CkPV (polydnavirus of Cotesia kariyai) plus venom depressed testis development. Testes from unparasitized day 0 last instar transplanted into isolated abdomens increased in volume after stimulation with 20-hydroxyecdysone (20HE). However, day 0 testis preincubated with CkPV plus venom for 6 h and then transplanted into an isolated abdomen did not respond to 20HE. Southern blot analysis indicated CkPV-DNA hybridized to testes-DNA from parasitized hosts, suggesting the possibility that CkPV i s involved in suppression of testes growth. Binding assays using PNA indicated a 2-fold increase in ecdysteroid receptor binding activity during the late stage of parasitism. The increase in receptor activity might be related to the maintenance of a low ecdysteroid titer in parasitized hosts due to a feedback response. 0 1994 Wiley-Liss, Inc. Key words: braconid parasitoid, polydnavirus, venom, ecdysone receptor INTRODUCTION Endoparasitoids often affect the physiological condition and behavior of their hosts [l].Parasitism of Pseudaletia sepavata (Walker) by Cotesia (=Apanteles) kariyai (Watanabe) caused several physiological alterations, including inhibition of host pupation [2,31, changes in fat body metabolism [41, and disruption of the immune system [5,6]. Host behavior is also changed during parasitism. Acknowledgments: We greatly appreciate Dr. M. Strand, Department of Entomology, University of Wisconsin, and Dr. S. Yagi, Tropical Agricultural Research Center, for their invaluable advice. Our grateful thanks also go to Dr. Takashi Tsuge, Laboratoryof Plant Pathology, Nagoya University,for helping in the hybridization experiment.This work is supported in part by ajapan Ministry of Education, Science and Culture Grant-in-Aid for Scientific Research on Priority Areas (#319), Project "Symbiotic Biosphere: An Ecological Interaction Network Promotingthe Coexistence of Many Species." Received February 1, 1993; accepted March 24, 1993. Address reprint requests to Dr. T. Tanaka, Lab. of Applied Entomology, School of Agricultural Sciences, Nagoya University, Chikusa, Nagoya 464, japan. 0 1994 Wiley-Liss, Inc. 11 2 Tanaka et al. For example, rather than moving down from the leaves into the soil in preparation for pupation, parasitized larvae remain on the plant where the parasitoid larva emerges . The testes of several species of Lepidoptera grow during the larval stage; however, testes often do not develop in parasitized larvae IS]. Many parasitoids inject polydnaviruses and/ or venom into their hosts at oviposition. Polydnaviruses enter the cells of host tissues [91 and in some species express several virus-specific transcripts [ l O , l l ] . In the Cuvdiockiles nigriceps-Heliothis vivescens system, Tanaka and Vinson  reported that the PTG" of parasitized hosts is affected directly by polydnavirus plus venom, with the PTG of parasitized larvae producing less ecdysteroid than the PTG of unparasitized larvae when stimulated with PTTH. Here we report that inhibition of testis development occurs shortly after parasitization and is related to the presence of polydnavirus plus venom. MATERIALS AND METHODS Insect Culture The braconid parasitoid C. kariyai was reared in the laboratory on P. sepurutu larvae maintained on an artificial diet (Nihon Nohsan Kohgyo, Kanagawa, Japan) at 25 f 1°Cunder a 16%(1ight:dark)photoregime. Adult wasps were fed a 30% sugar solution in a glass tube as described previously . The day on which each host larva ecdysed to the next instar was designated as day 0. Chemicals PNA, obtained from Takeda Pharmaceutical Industries (Kyoto, Japan), was purified on reverse phase HPLC. The specific activity, determined on the basis of W absorption of the steroid (E243 = 12,300), was 83Ci/mmol. Labeled PNA used for the binding assay was purified on HPLC. 20HE was purchased from Sigma Chemical Co. (St. Louis, MO). Protein Assay Concentrations of protein in cytosol preparations were determined by the colorimetric method of Bradford  using bovine serum albumin as a standard. Ecdysteroid Receptor Binding Assay Specific binding was determined based on the difference between specific and nonspecific PNA binding to total protein. First, all binding protein was bound by cold PNA. Nonspecific binding protein with a weak affinity was *Abbreviations used: Carnoy's fixative solution = absolute ethanol: chloroform: glacial acetic acid, 6:3:1; CkPV = polydnavirus of Cotesia kariyai; DCC = dextran-coated charcoal; Denhardt's solution = 0.02% bovine serum albumin, 0.02% ficoll, 0.02% polyvinylpyrrolidone; Mayer's HE = Mayer's hernatoxylin and 1% eosin solution; P = pupal stage; PNA = ponasterone A; PP = prepupal; PTG = prothoracic gland; PTTH = prothoracicotropic hormone; S E M = scanning electron microscopy; TE = 10 m M Tris-HCI (pH 7.5), 1 m M EDTA; TM buffer = 20 rnM Tris-HCI (pH 8), 5 m M NaCI, 20 rnM 2-mercaptoethanol, 1 m M phenylmethyl-sulfonyl fluoride, 10% glycerol; TNE = 10 m M Tris-HCI (pH 7.5), 0.1 M NaCI, 1 rnM EDTA; TNM = 20 rnM Tris-HCI (pH 7.5), 0.1 M NaCI, 1.5 m M MgC12;20HE =20-hydroxyecdysone;20x SSC = 3.0 M NaCI, 0.3 M sodium citrate; 20 x SSPE x 3.6 M NaCI, 0.2 M sodium phosphate, 0.02 M EDTA (pH 7.7); W = wandering. Reduction of Testis Growth by C. kariyai 11 3 replaced with [3Hl PNA; the binding of cold PNA was completed by i3H1PNA .Radioactivityof these samples where [3H]I"A alone was added represented the total value of specific and nonspecific binding. Specific binding was calculated as the difference between counts in two series samples. Testes from 50-100 male larvae for each stadium were homogenized in 500 ml of modified TM buffer and centrifuged at 50,OOOg for 30 min, according to the method of Deak et al. . Each supernatant was adjusted to 1 pg/p1 after measuring the protein concentration. Samples at each stadium were divided into 4 tubes with the amount of protein per sample standardized. To two tubes was added 5 nM cold PNA and to the other two tubes was added the same volume of ethanol. Tubes were agitated and then held for 15 min at 4°C. i3H1 PNA (25pM) was added to all samples, followed by incubation for an additional 12-16hat4"C. Each sample was incubated for 5 min with 2.1% DCC to remove the excess [3H]PNA.Aftercentrifugingthesarnplesat10,000gfor 15 min, 1 ml of supernatant from each sample was counted by liquid scintillationspectroscopy. Collection of Polydnavirus and Venom Calyx fluid was collected from the lateral oviduct of female wasps as described previously 121. CkPV was purified by centrifugation at 20,6009 for 10 min, rinsed three times with Pringle's saline, and finally adjusted to one female equivalent per pl. A 1/3 female equivalent of polydnavirus and venom was used for injection experiments. Resuspended solution was confirmed to contain polydnavirus alone by S.E.M. (data not shown). Histological Study To characterize histological changes in the testis after parasitization, testes from hosts parasitized at various larval stages were dissected in Pringle's saline, fixed with Carnoy's solution for 12 h, dehydrated in ethanol, and embedded in paraffin. Sections were stained with Mayer's HE. Transplantation To determine whether or not transplanted testes grow with a change in the hormonal milieu, larval testes from day 0 6th instar were transplanted into day 2 6th instar larvae and dissected 6 days after transplantation, coincident with when the recipient pupated. In addition, testes from day 0 6th instar larvae were transplanted into isolated abdomens and stimulated with a high-dose injection of 3 pg 20HE. Isolated abdomens were prepared by ligating larvae behind the thorax on day 2 of the last instar, a testis was transplanted into the isolated abdomen through a proleg, and then the proleg was ligated using fine thread. After transplantation each isolated abdomen was placed in a small Petri dish held in a large plastic container to elevate the humidity. Isolation of Genomic DNA of Host Testes and CkPV-DNA Forty testes from hosts in various stages were homogenized in TNM with a 1 ml glass homogenizer and centrifuged at 2,OOOg for 5 min. The pellet was dissolved in TNE. CkPV from 150 wasps was also dissolved in 300 p.1 TNE SDS and digested for about 12 h buffer. Both samples were added to 20 ~ 1 1 0 % 114 Tanaka et al. at 50°C with 4 ~1proteinase K (10mglml). Samples were extracted withan equal volume of phenol, incubated for 5-6 h at 25"C, the water phase was collected and treated with DNase-free RNase (Boehringer Mannheim, Germany) at 37°C for 1 h following extraction with an equal volume of pheno1:chloroform (1:l). DNAs were precipitated with 100%ethanol and 0.1 M NaCl at -20°C overnight, rinsed with 70% ethanol, and resuspended in TE buffer. Labelling of CkPV-DNA CkPV-DNA was labelled by random priming using a commercial Labelling Kit (UnitedStates Biochemical, Cleveland, OH) and [32P]dCTP(3,000Ci/mmol)(ICN Biomedicals,Irvine,CA)accordingto protocol. Gel Electrophoresis and Southern Hybridization Hind III-digested CkPV-DNA and genomic DNA of host testis in various stages after parasitization were size fractioned on 0.8% agarose gels and transferred to nylon (H bond-N+; Amersham, Arlington Heights, IL) in 20 x SSC. Hybridization to 3&-labelledCkPV-DNAwas carried out at42"C for 12-14 h in 5 x SSPE, 0.5% SDS, 5 x Denhardt's solution. Following hybridization, filters were rinsed and washed twice with 2 x SSPE plus 0.1% SDS at 65°C for 10 min, with 1 x SSPE plus 0.1% SDS at 65°C for 15 min, and 0.1 x SSPE plus 0.1% SDS at 65°C for 10 min according to the manufacturer's recommendation. After wrapping in Saran Wrap, filters were exposed to Kodak X-ray film using an intensifying screen at -80°C for 5 days. RESULTS Parasitization of C. kariyai depresses growth of the host testis. This effect was duplicated experimentally by an injection of CkPV plus venom into day 3 of unparasitized 5th instar hosts (Fig. 1). Inhibition of testis growth by polydnavirus and venom suggested that inhibition of growth occurred immediately after oviposition, because calyx and venom fluids injected into host larvae at this time induces other developmental alterations [3,4]. To test this quantitatively, day 0 last instar larvae were parasitized. The testes of hosts maintained almost the same volume for 10 days until emergence of the parasitoid larvae. In contrast, testes of unparasitized hosts increased 13.4-fold during the same period (Fig. 2). Differences in testicular volume were also noted when larvae were parasitized during the late 2nd, 3rd, and 4th stadia and dissected on day 1 of the last stadium (Fig. 3). In each case testicular volume at the time of dissection was similar to the volume at the time of parasitization. Examination of paraffin sections indicated that in the case of parasitization of late 3rd instar larvae (Fig. 4D), the number of spermatids was greatly reduced in comparison to that of unparasitized controls (Fig. 4A) and were dispersed with no unit formation in the follicles. Many cells also showed signs of degeneration. In the testes from larvae parasitized during late 5th stadium, almost all cells made a unit and appeared normal (Fig. 4E), although some cells contained chromosomes arrested in meiotic division (Fig. 4F). Also, testes from hosts parasitized late in the 4th stadium contained cells with swelled chromosomes Fig. 1. Depression of growth of the host testes by CkPV and venom of Cotesia kariyai. Testicular volume was estimated by the formula 2zr2h/3 (where r = half the width and h = half the length of the testis) . Dissection was performed 6 days after injection. A one-third female equivalent of calyx fluid (calyx) or purified polydnavirus (virus) and venom were injected into day 3 5th instar hosts. One testis per larva was used for measuring. Numbers in parentheses indicate the number of replicates per time point. For each treatment, bars with different letters were significantly different at P i 0.01 (one-way ANOVA). 6 5- - 6 Control * -f- parasitized -5 4- -4 3- -3 2- -2 1- -1 0 0 I I I I I I I I I 1 2 3 4 5 6 7 8 9 Days after I 0 1 D 1 1 parasitization Fig. 2. Testicular volume of hosts parasitized on day 0 of the last stadium and unparasitized larvae on the same day. Asterisks above each bar indicate significant difference at P < 0.01 (Student’s t-test). Tanaka et al. "'" 1 9 0.5 0.4 0.3 0.2 0.1 0.0 Stage at parasitization P Fig. 3. Depression of testes development in hosts parasitized during the late 2nd (2nd-P), 3rd (3rd-P), 4th (4th-P), and 5th instar (5th-P). Testicular volume was measured when hosts were day 1 last (6th) instar larvae. Testicular volume at the late 3rd (3rd-C), 4th (4th-C), 5th (5th-C), and day 1 6th (day16th-C) stadium of unparasitized control larvae are presented in the figure for reference. The letter above each bar indicates the point was significantly different at P<O.Ol (one-way ANOVA). Numbers in parenthesis on each bar indicate the number of replicates. (Fig. 4C). Another notable effect was that the follicular septa of testis from parasitized hosts (Fig. 4D,E) were thicker than that of normal testis (Fig. 4A,B). Transplantation of testes from unparasitized day 0 6th instar hosts into day-2 6th instar hosts resulted in a large induction in testicular volume (Fig. 5). In contrast, transplantation of testes from day 0 larvae injected with CkPV plus venom 6 h before excising into day 2 6th instars did not induce an increase in testicular volume until dissection, suggesting that the testes affected by CkPV plus venom were unresponsive to the hormonal milieu of the host. Testes from unparasitized larvae transplanted into isolated abdomens increased in size with stimulation by 20HE but testes from CkPV plus venom injected larvae did not (Fig. 5). To distinguish whether CkPV plus venom affects the growth of testes directly, testes from day 0 larvae were exposed to CkPV plus venom in vitro for 6 h, and then transplanted into isolated abdomens. Testes exposed to CkPV plus venom remained almost the same size 6 days after 20HE stimulation in isolated abdomen (Fig. 6), again suggesting that testes affected with CkPV plus venom are unresponsive to stimulation of 20HE. Hybridization of CkPV DNA to testes DNA from day 2,4, 6, and 8 larvae indicated the direct infection of polydnavirus-DNA to host testis. Fig. 4. Light micrographs of testis from day 2 (A) and day 3 (B) of 6th stadium unparasitized larvae and day 2 6th stadium larvae parasitized duringthe late 3rd (D),4th (C), and 5th (E) stadium. Spermatids of day3 control (B) are transformed to elongated sperm. In case of parasitization of late 3rd instars (D), spermatids were dispersed, forming no unit. Parasitization of late 4th (C) and 5th (E) instars resulted in abnormal chromosomes (shown by arrowhead). Magnified cells from larvae parasitized as late 5th instars are shown in F. Tanaka et al. Fig. 5. Depression of sensitivity of the host testes to 20HE stimulation by CkPV plus venom of Cotesia kariyai. The testes from day 0 last instars were excised and measured before transplantation (light bar). The volume of testes dissected 6 days after transplantation are shown as a dark bar. Testes from larvae injected with 1/3 female equivalent of CkPV plus venom (CV-day2) or with saline (day0-day2) were transplanted into day 2 6th stadium unparasitized hosts. The testes from day 0 6th instar hosts injected with CkPV plus venom (CV-2OHE) or with saline (dayO-20HE)were stimulated by 20HE after transplantation into isolated abdomens. Dissection was performed 6 days after transplantation. Testicular volume of day 6th instars (dayO) is shown as reference. Different letters above each bar indicate a significant difference at Pc 0.01 (one-way ANOVA). The molting hormone receptor levels of unparasitized control larvae increased gradually with the developmental stage reaching 60 pmol/g protein! individual at pupation. In contrast, receptor levels from parasitized larvae increased to twice this level (Fig. 7). DISCUSSION Our results indicate that CkPV plus venom affects testis development. Baudoin  suggested that castration occurs when the host is investing in resources for reproduction. Thus, depression of testis growth by a larval parasitoid could be advantageous by removing a testis that competes for resources [16-18]. Although testicular volume in parasitized hosts remained unchanged from the time of parasitization (Figs. 2,3), our histological data suggest that growth and differentiation are not completely blocked. This is in contrast to the Reduction of Testis Growth by C. kariyai 119 i n Day0 6D posttrans @7 E E 0.8 v 6h-contr 6h-CV Fig. 6. Inhibition of the response of testis to 20HE in vitro by CkPV plus venom. A day0 testis was incubated for 6 h in Grace's insect medium with (6h- cv) or without (Gh-contr)CkPV plus venom, transplanted into isolated abdomens, and then stimulated by 20HE. Each testis was transplanted into an isolated abdomen of day 2 last stadium larvae and was stimulated by 3 pg 20HE. Testes were dissected 6 days after transplantation and measured. Each testicular volume was significantly different at 0.01 by one-way ANOVA. report of Junnikkala  on parasitism of Pieris brussicae by ApanteIes glomemtus. Parasitization of late 4th and 5th instars caused some spermatids to transform into incompletely elongated sperm [B], a condition that may be related to the inhibition of development of the testimlar septa. The abnormal appearance of chromosomes in some spermatocytes suggests there may exist a critical period when CkPV plus venom affects meiosis. CkPV plus venom may be unable to inhibit spermatogenesis in some spermatocytes if they have developed beyond 200 1 - , control + parasitized 160 200 1160 120 80 40 0 0 1 2 3 4 5 6 7 8 9 10 Days after last larval ecdysis Fig. 7. [3H] PNA receptor titer in unparasitized and parasitized hosts. Measurements were made as described in Materials and Methods. Each value represents the mean of two replicates. Tanaka et al. 120 - M pdv P2 P4 P6 P8 CV C2 CVP2 P4 P6 P8 CV C2 C4 Id3ps 23.1 9.4 - - 6.6'- 4.4 2.3 2.6 - - Fig. 8. Hybridization of testes genomic DNA at various stages post-parasitism. Hind Ill-digested genomic DNA of testis and CkPV were separated in 0.8% agarose gels, transferred to a nylon membrane (Hybond N+) and with 32P-labelledCkPV DNA. The numbers to the left of the figure represent sizes of kilobase pairs (kbps)of Hind Ill-digested h DNA markers (M). Polydnavirus DNA (pdv) hybridizes to testicular genomic DNA from hosts on day 2 (P2), 4 (P4), 6 (P6), 8 (P8), after parasitization but does not hybridize to testicular DNA from unparasitized larvae on day-2 (C2) and 4 (C4) of last stadium. A weak hybridization signal was observed in lane CV, containing testicular DNA incubated in vitro 6 h with CkPV plus venom. a certain critical stage. Pearson and Cheng  suggested that some component of extracts from the parasite Zoogonus lasius acts as a chemical inhibitory factor in the case of certain Ilyanassa obsoleta spermatogonia in G1 phase and that molecule(s) of parasite origin may cause the lysis of immature and mature gametes, inhibit division and differentiationof immature gametes, and antagonize neurosecretions essential for gamete formation and development. The testes from hosts parasitized or injected with CkPV plus venom become insensitive to stimulation by 20HE. The results of transplantation experiments in vivo and in vitro suggest that the effect of CkPV plus venom on the testis is direct. Furthermore, Southern hybridization data (Fig. 8) indicate that CkPV DNA is present in host testes. These facts prompted us to examine whether testes infected with CkPV lose the ability to respond to 20HE stimulation. Receptor binding assays suggest that the receptor titer of parasitized testes increases to higher levels than that of testes from unparasitized larvae. DeJongBrink et al. [21,221 suggested that schistosomin, a peptidergic agent, is present in hemolymph of snails infected by Trichobilharzia ocellata and blocks the effect of female gonadotropic hormone at the level of the receptor complex. Likewise, CkPV may produce a factor affecting the level of an ecdysone receptor or regulation of ecdysone receptor transcription. Reduction of Testis Growth by C. kariyai 121 LITERATURE CITED 1. Vinson SB, Iwantsch GF (1980): Host regulation by insect parasitoids. Q Rev Biol55:143. 2. Tanaka T, Agui N, Hiruma K (1987): The parasitoid Apanteles kariyai inhibits pupation of its host, Pseudaletia separata, via disruption of prothoracicotropic hormone release. Gen Comp Endocrinol67364. 3. Tanaka T (1987): Calyx and venom fluids of Apanteles kariyai (Hymenoptera: Braconidae) as factors which prolong the larval period of the host, Pseudaletia separata (Lepidoptera: Noctuidae). Ann Entomol SOCAm 80:530. 4. Tanaka T (1986): Effects of the calyx and venom fluids of Apanteles kariyai Watanabe (Hymenoptera: Braconidae) on the fat body and hemolymph protein contents of its host Pseudaletia separata (Walker) (Lepidoptera: Noctuidae). Appl Ent Zool 21220. 5. Tanaka T (1987): Morphological changes in haemocytes of the host, Pseudaletia separata, parasitized by Microplitis mediator or Apanteles kariyai. Dev Comp Immunol11:57. 6. Wago H, Tanaka T (1989): Synergistic effects of calyx fluid and venom of Apanteles kariyai Watanabe (Hymenoptera: Braconidae) on the granular cells of Pseudaletia separata Walker (Lepidoptera: Noctuidae). Zool Sci 6:691. 7. Sato Y, Tanaka T, Imafuku M, Hidaka T (1983): How does diurnal Apanteles kariyai parasitize and egress from a nocturnal host larva. Kontyu Tokyo 51:128. 8. Yagi S, Tanaka T (1992): Retardation of testis development in the armyworm, Pseudaletia separata, parasitized by the braconid wasp, Cotesia hriyai. Invert Reprod Develop 22151. 9. Stoltz DB, Vinson SB (1979): Penetration into caterpillar cells of virus-like particles injected during oviposition by parasitoid ichneumonid wasps. Can J Microbiol25:207. 10. Blissard GW, Fleming JGW, Vinson SB (1986): Campoletis sonorensis virus: Expression in Heliothis virescens and identification of expressed sequences. J Insect Physiol215:351. 11. Strand MR, McKenzie DI, Grass1 V, Dover BA, Aiken JM(1992): Persistence and expression of Microplitis demolitor polydnavirus in Pseudoplusia includens. J Gen Virol73:1627. 12. Tanaka T, Vinson SB (1991):Depression of prothoracic gland activity of Heliothis virescens by venom and calyx fluids from the parasitoid, Cardiochiles nigriceps. J Insect Physiol37139. 13. Bradford MM (1976): A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72248. 14. Deak P, Zavorszky P, Maroy P (1988): Moulting hormone regulates its receptor level in Drosophila melanogaster. Insect Biochem 18847. 15. Baudoin M (1975): Host castration as a parasitic strategy. Evolution 29:335. 16. Hurd H (1990): Physiological and behavioural interactions between parasites and invertebrate hosts. Adv Parasitol29:271. 17. Reed-Larsen DA, Brown JJ (1990): Embryonic castration of the codling moth, Cydia pornonella by an endoparasitoid, Ascogator quadridentata. J Insect Physiol36111. 18. Tanaka T, Yagi S, Nakamatu Y (1992):Regulation of parasitoid sex allocation and host growth by Cotesia (=Apanteles) kariyai (Hymenoptera: Braconidae). Ann Entomol Soc Am 85:310. 122 Tanaka et al. 19. Junnikkala E (1985): Testis development in Pieris brassicae parasitized by Apanteles glomeratus. Entomol Exp Appl37283. 20. Pearson EJ, Cheng T (1985): Studies on parasitic castration: Occurrence of a gametogenesis-inhibiting factor in extract of Zougonus lasius (Trematoda).J Invertebr Path01 46239. 21. DeJong-Brink M, Elsaadany MM, Boer HH (1988): Interference with endocrine control of female reproduction of Lymnaeu stagnulis. Exp Parasitol65:109. 22. DeJong-Brink M, Hordijk PL, Schallig HDFH, Bergamin-Sassen MJM, Oosthoek I? (1990): Possible mechanism underlying parasitic castration in tramatode infected snails. In Hoshi M, Yamashita 0 (eds): Advances in Invertebrate Reproduction, vol. 5. Amsterdam: Elsevier Science, pp 141-148.