Copyright 0 1994 by the Genetics Society of America

Efficient Recovery of ENU-Induced Mutations From the Zebrafish Germline

Lilianna Solnica-Krezel, Alexander F. Schier and Wolfgang Driever Cardiovascular Research Center, Massachusetts General Hospital and Harvard Medical School, Charlestown, Massachusetts 02129 Manuscript received August 27, 1993 Accepted for publication December 29, 1993

ABSTRACT We studied the efficiency with which two chemical mutagens, ethyl methanesulfonate (EMS) and N-ethyl-N-nitrosourea (ENU) can induce mutations at different stages of spermatogenesis in zebrafkh (Brachydanio rerio).Both EMS and ENU induced mutationsat high rates in post-meiotic germ cells,as indicated by the incidence of F, progeny mosaic for the albino mutation. For pre-meiotic germ cells, however, onlyENU was found to be an effective mutagen, as indicated by the frequencies of non-mosaic mutant progenyat four different pigmentation loci.Several mutagenic regimens that varied in either the number of treatments or the concentrationof ENU were studied to achieve an optimal ratio between the mutagenicity and toxicity. For the two most mutagenic regimens: 4 X 1 hr in 3 mM ENU and 6 X 1 hr in 3 mM ENU, the minimum estimate of frequenciesof independent mutations perlocus per gamete was 0.9-1.3 X We demonstrate that embryonic lethal mutations induced with ENU were transmitted to offspring and that they could be recovered in an F, screen. An average frequency of specific-locus mu- tations of 1.1 X lo-’ corresponded to approximately 1.7 embryonic lethal mutations per single mu- tagenized genome. The high ratesof mutations achievable withENU allow for rapid identificationof large numbers of genes involved in a variety of aspects of zebrafish development.

ATURATION mutagenesis screens for embryonic le- Among the vertebrate species in which the mecha- S thal mutations led to the identification of a large nisms of embryogenesis are currently investigated, ze- number of genes that control the development of in- brafish offers several advantages for saturation mutagen- vertebrate animals and plants (e.g., HIRSHand VANDER- esis screens (STREISINGERet al. 1981). Thissmall tropical SLICE 1976; NUSSLEIN-VOLHARD and WIESCHAUS1980; fish has a generation time of 3 months, and maturefe- JCJRGENS et al. 1991; MAYER et al. 1991).Saturation males lay hundreds of eggs at weekly intervals, facilitat- mutagenesis screens are feasible only in experimental ing genetic analysis. A major asset of zebrafish is the systems that fulfill several criteria. First, a mutagenesis accessibility of its embryonic development to visual in- method has to be available that is both very efficient and spection. The embryos develop synchronously outside causes predominantlyintragenic lesions. This mini- the mother and are transparent, such that detailed as- mizes the numberof genomes thathave to be screened pects of embryogenesis and organogenesis can be moni- to reach saturation levels and assures that observed de- tored microscopically (STREISINGERet al. 1981; KIMMEL fects can be linked to lesions in single genes. Second, the 1989). Several interesting mutations have been identi- efficient transmission of mutations to subsequent gen- fied following y-irradiation mutagenesis of pregonial erations should be feasible without seasonal variation cells (KIMMEL et al. 1989; FELSENFELDet al. 1990; WESTER- and in reasonably short time periods. Finally, conve- FIELD et al. 1990; HATTAet al. 1991). nient selection or screening procedures for phenotypes Chemical mutagens are preferentially used for satu- have to be available. The mouse is the only vertebrate in ration mutagenesis as they can efficiently induce lesions which large scale screens forvisible morphological traits that are primarily limited to single genes (SINGERand or metabolic defects have been carried out (JOHNSON GRUNBERGER1983). Among them, two alkylating agents and LEWIS1981; MCDONALDet al. 1990; MOSER et al. are most frequently used for mutagenicpurposes. Ethyl 1990). However, in the mouse, screening forembryonic methanesulfonate (EMS) is the mutagen of choice in lethal mutations, and especially for those affecting the Drosophilamelanogaster and Caenorhabditiselegans earliest stages of development, is difficult due to the in- (ASHBURNER 1989; WOOD 1988). N-Ethyl-N-nitrosourea trauterine mode of embryogenesis. Thus far, the ma- (ENU) is the most potent mutagen of the mouse germ- jority of embryonic lethal mutationshave been foundin line (RUSSELLet al. 1979), anda preliminary report sug- mouse genetic screensby virtue of linkage to visible re- gests that ENU might beeffective in mutagenesis of sper- cessive mutations. This limits saturation mutagenesis to matogonia of the small teleost fish medaka (Oryzias small portions of the murine genome(SHEDLOVSKY et al. latipes) (see discussion in SHIMAand SHIMADA1991). 1986, 1988). Zebrafish spermatozoa can be efficiently mutagenized

Genetics 136: 1401-1420 (April, 1994) 1402 L. Solnica-Krezel,A. F. Schier and W. Driever with ENU in vitro (GRUNWALDand STREISINCER1992). report studies on the mutagenic activity of EMS and Upon fertilization by mutagenized spermatozoa, F, in- ENU on different cellular stages of the male germline. dividuals develop that aremost frequently mosaic for the We demonstrate that ENU but not EMS can induce induced mutations. Several methods forgenetic screen- locus-specific mutations at high rates in pre-meiotic ing have been developed to recover embryonic lethal germ cells ofzebrafish males and that inducedrecessive mutations among progeny of F, females derived from embryonic lethal mutations can be effkiently recovered mutagenized sperm, or of females mutagenized at the in an F, screen. blastoderm stage. Mutations are manifested in haploid gynogenetic progeny of adult females by fertilizing their MATERIALSAND METHODS eggs with sperm rendered genetically inactive by UV ir- Fish tank facilities: Most of the experiments described de- radiation (FELSENFELDet al. 1990). Haploid embryos, pend on the ability to keep large numbers of zebrafish un- however, start to develop abnormally on the secondday der conditions optimal for breeding. Given limited space, we after fertilization (STREISINGERet al. 1981). This hampers had to adopt and develop methods to keep zebrafish at high possibilities for screening forphenotypes affecting pro- population densities. We employed a closed circulating wa- cesses in later embryogenesis. Mutations have alsobeen ter system that guarantees a low water load of compounds deleterious to zebrafish and provides conditioned water of recovered among diploid parthenogenetic embryos consistent chemistry. produced by treatments of activated eggs during thefirst We use 20-liter tanks (acrylic, custom-made: about 15 cm cellcycle with early pressure (EP) (STREISINGERet al. wide, 60 cm long and 25 cmhigh) to keep up to 100 adult fish; 1986) or with heat shock (HS) (STREISINGERet al. 1981) 4liter tanks (acrylic, custom-made: about 12 cm wide, 30 cm (Figure 1, A and B). Both these procedures, however, long and 15 cm high) to keep up to 20 adult fish or upto 100 fish at the age of 1-2 months; and l-liter tanks (Nalgene stor- decrease survival of the eggs, so that only a fraction of age containers from clear plastic 5700-1000; 127 X 73 X 175 diploid embryos develop normally (STREISINGERet al. mm) to keep single adult fish. Four- and 20-liter tanks are 1981; GRUNWALDand STREISINGER1991, 1992). There- equipped with standpipes and/or strainers to control the wa- fore, although these methods are advantageous as they ter level. One-liter boxes have three 2-mm wide holes drilled into the back that serve as drains. One-liter boxes sit in large involve onlylimited breeding (one generation),analysis plastic traysthat are drainedinto a common reservoir. All tanks of subtle or late phenotypes in parthenogenetic embryos are constantly supplied with filtered water through inlet valves would be difficult. at a rate so that water in the tank is exchanged about three A classical three-generation ofcrosses scheme (F, times per hour.Four-liter and 20-liter tanks are aerated atrates screen), similar to the one performedin mice (HALDANE of about 1 liter and 3-5 liters of air per minute, respectively. Pressurized air for the entire fish facility isgenerated by three 1956) (Figure 1, C and D), in which mutations induced 1/2 hp regenerative blowers (Sweetwater Blower $31; Aquatic in the parental generation aredriven to homozygosity in Eco Systems, Apoka, Florida). the F, generation, could circumvent this problem. If The water filtration system serves about 65,000 zebrafish in spermatozoa were mutagenized, however, recovery of tanks with a total volume of about 16,000 liters. Water from mutations would require a large number of crosses be- reservoirs (total volume about 5,400 liters) is recirculated for biological filtration. A 3/4 hp. centrifugal pump (Max-E-Glass, tween F2 siblings, due to the mosaic character of the Sta-Rite Industries, Inc., Delavan,Wisconsin) pumps water germline of F, progeny of mutagenized sperm (Figure from the reservoir through asand pool filter (hydro rate HRV- 1C) (GRUNWALDand STREISINGER1992). The mosaicism 36; Baker-HydroInc.; Augusta, Georgia) and up ontoa wet-dry of the F, germline can becircumvented by mutagenizing trickle filter (ammonia tower with about 1 ms of plastic balls as substratum for denitrifylng bacteria; Aquatec, Brockton, pre-meiotic germ cells such as spermatogonia in adult Massachusetts). Filtered water flows back into the reservoirs. males (RUSSELLet al. 1979). In this case mutations are Five subsystemsdraw waterfrom the main reservoirs to serve fixed via subsequent roundsof DNA replication, before tanks with a total volume of about 3,000-3,500 liters for each differentiation into mature sperm cells. Hence, the F, subsystem. For each subsystem, a 1/2 hp. centrifugal pump progeny of the mutagenized parental animals are non- (Max-E-Glass,Sta-Rite Industries, Inc., Delavan, Wisconsin) drives waterat about10 p.s.i. through particle filter cartridges mosaic heterozygotes each carrying one or several (Jacuzzi CFR 100; Jacuzzi Bros. Div., Little Rock, Arkansas;fil- unique mutations. Subsequently, F, animals become ters are exchanged every 6 months) to remove debris and po- founders of F, families in which 50% of progeny are tential parasites. The filter cartridges are loaded with bagged heterozygous for the mutation(s) inherited from theF, carbon (Darco activated carbon casno 7,440-440; American founder. These mutations manifestedare during screen- Norit Co., Atlanta, Georgia; about 1 kg, exchanged monthly) to remove dissolved organic waste from the water. Water is ing the F, progeny of F, siblings (Figure 1D). An im- sterilized while flowing through a 120 W UV sterilizer (Aqua- portant advantage of this method is that each phenotype netics 120 IL, Aquanetics Systems, Inc, San Diego, California). is revealed on average in everyfourth cross, reducing the Water is distributed to the tanks and drains back through pre- number of crosses needed to detect themajority of mu- filter pads (cleaned weekly) into the main reservoirs. Every day, about 10% of the total volume of water in the tations harbored by a given F, line. system is exchanged with fresh water. Under automatic con- To initiate a saturation mutagenesis screen for em- trol, 28" water (mixed hot and cold tap water, TempControl bryonic lethal mutationsin zebrafishwe searched for an thermostatic water controller, Symmons Industries, Braintree, efficient method to mutagenize the germline. Here we Massachusetts) flows six timesa day for a few minutes through Germline Mutagenesis in Zebrafish 1403

A. F1 screen aftermutagenesis of mature 8. F1 screen after mutagenesisof premeiotic sperm cells germ ceiis

FI founder female: F1 founder female: mosaic rn I + nonmosaic m I +

Heat Shock or Heat Shock or 0000 Early Pressure Early Pressure P‘4 ‘J. F2 diplold F2 diploid gynogenetic gynogenetlc embryos embryos +I+ +I+ +/+ Wrn +I+ nv+ +I+ m/m after EP

F2 diploid gynogenetic embryos +I+ m/rn +I+ m/m after Hs C. F2 scmen after mutagenesis of mature D. F2 screen after mutagenesls of premeiotic sperm cells germ cells 0,

F1 founder: F1 founder: wild type fish +/+ wild type fish +/+ mosaic d+ 4 non-mosaic W+ F2 family: usually less F2 family: than114progeny 112 of progeny carries

F2 screening: the F2 screening: the 4 mutation is found mutation is found a in25Ohofthe d+ 4 d+ siblingcrosses

d+ +I+ n#+ m/m n#+ +I+ nl+ m/rn FIGURE1.-Strategies for induction and recovery of mutations. (A) F, screening strategy after chemical mutagenesis of mature sperm (KIMMEL 1989; GRUNWALDand STREISINGER1992). Upon fertilization with mutagenized sperm, mutations are fixed during early development andF, individuals develop that aremost frequently mosaic for theinduced mutations. Mutations can be revealed in gynogenetic progenyof F, females. First, development of eggs of F, females is initiated by sperm rendered genetically impotent by UV light treatment. Second, diploid embryos are produced by treating haploid activated eggs during the first cell cycle with early pressure (EP) or heat shock (HS) treatments (STREISINCERet al. 1981). Since F, females are mosaic for carried mutations, less than expected 50% and usually lessthan 20% of surviving F, parthenogenetic progeny exhibit mutant phenotypes (GRUNWALD and STREISINCER1992). For the EP treatment proportion of homozygous mutant embryos is also inversly correlated with the locus-centromere distance (STREISINGERet nl. 1986). (B) F, screening strategy after chemical mutagenesis of spermatogonia. Spermatogonial stem cells are mutagenized by treatment of adult males and mutations are fixed before differentiationinto mature sperm cells. F, females derived from such mutant sperm are non-mosaic heterozygotes for the inducedmutations. Consequently, up to 50% (for theEP usually lessthan 50% depending on thelocus-centromere distance) of developing parthenogenetic progeny produced as described in (A) exhibit mutant phenotypes. (C) F, screening strategy after chemical mutagenesis of mature sperm. As in (A) F, progeny of mutagen-treated spermare mosaic heterozygotes for inducedmutation(s). Subsequently, F, animals become founders of F, families. Due tomosaicism of F, founders usually lessthan 25% of F, fish are heterozygous for the inheritedmutations. Consequently, during screeninga large number of crosses between F, siblings is needed to recover the mutation(s). In a cross in which mutation(s) are manifested, 25% of F, embryos exhibit each mutant phenotype. (D) F, screening strategy after chemical mutagenesis of spermatogonia. ,& in (B), after mutagenesis of spermatogonia, sperm cells contain fixed mutations. Hence, the F, progeny are non-mosaic heterozygotes for the mutation(s)carried by the sperm. Due to the non-mosaic character of F, founder fish, 50% of the F, siblings carry the inherited mutation(s).Consequently, the mutation(s)should be manifested in 25%of F, sibling crosses. As in (C), 25% of F, embryos in positive crosses exhibit each mutant phenotype. 1404 L. Solnica-Krezel, A. F. Schier and W. Driever a filtrationunit (CUNO water filtration system: A124and A1 10 2 months, fish are transferred to 20-liter tanks and fedwith a dirt/rust filters and A117 activated carbon filter, CUNO Inc., mixed diet of brine shrimp, freeze dried and X-ray-sterilized Meriden, Connecticut; filters were exchanged every month) krill (Argentum Laboratories), Spirulina algae powder and into themain reservoir. Simultaneously, a metering pumpdis- flake food (TetraMin Flake Food andTetra Conditioning penses a defined amount of a saline solution (3%w/v Instant Flakes, Tetra Werke, Melle, Germany). Using these conditions Ocean Salt mix, Aquarium Systems, Mentor, Ohio) to a salt zebrafish grew to sexual maturity between 3 and 4 months of content of 0.03% w/v in the water. The pH in the system is age. Within afish population, the first breeding could be ob- buffered by introducing cottonbagged crushed coral (Florida served at an age of 10-12 weeks, and at theage of 4 months, Crushed Coral 150, CaribSea Inc., Miami, Florida) into the about 3040% of crosses were successful and yielded mostly particle filter cartridges. more than 50 fertilized eggs. System parameters are: temperature 28.5-29.5” (controlled Strains: Our standardwild-type strain was derived from the by room air temperature);salinity 0.03-0.04% (conductivity: “AB line” from the zebrafish laboratories at the University of 0.9 mS); pHbetween 6.7 and 6.9; ammonia andnitrite below Oregon, Eugene, andhas been describedby CHAKIWBARTIet al. 0.5 ppm; and nitratebetween 30and 60 ppm. Fish are kept on 1983. These fish have been maintained in our laboratory since a 14hr light/l@hr darkcycle (8.30 AM to 10:30 PM). Tanks are spring of 1991 by inbreeding. Every newgeneration is screened kept clean by catfish (Ancistrus and Plecostoma sp.), and de- for the presence of embryonic lethal mutations: at least 10 bris is siphoned off every 2-4 weeks. sibling crosses within one stock are performed, andembryonic Genetic crosses: Crosses between individual male and fe- development is observed until day ’7 after fertilization. If more male zebrafkh are set upin “breeding traps” in order to pre- than 95% of the larvae from each cross develop swim bladders vent the parents form eating thefreshly laid eggs. The design on or before day 7, and none of the embryos shows develop- of the breedingtraps is based on a similar one by C. N~JSSLEIN- mental abnormalities, the stock is considered to be free of VOLHW(Tubingen, Germany)described in CULP et al. zygotic lethal mutations, Tracing the lineage allows us to con- (199 1). The bottom of size A polypropylene mousecages (Nal- clude thatmost of our stocks should also be free of embryonic gene) is cut off and a stainless steel (grade T316) mesh of lethal maternal-effect mutations. 1/8-inch mesh size is melted to the bottom of the breeding For specific locus tests, strains carrying recessive mutations trap. The steel is coated with polyurethane. Such traps are at four different unlinked loci affecting body pigmentation placed inside size A polycarbonate mouse cages (Nalgene), were used. At the 3rd day of development mutant embryos and filled to 2/3volume with water from ourfish facility (“sys- homozygous for the mutationalbino’” (alb; STREISINGERet al. tem water”). A small piece of a green plastic plant placed in 1986) exhibit melanocytes that arevery lightly pigmented com- the breeding trapserves as a “hideout”in case the behavior of pared to the black melanocytes of wild-type embryos. Embryos one of the fish becomes too aggressive. Eggs fall through the homozygous for the mutationgolden” (got WALKERand STRE- mesh, and successful crosses can be identified by visual in- [SINGER 1983) develop light pigmentation of the body and the spection of the clear polycarbonate bottom part of the trap. retina with a distinctbrown tint. Emblyos mutant at brass‘ and Crosses are set up from the late afternoon until evening. brass” are characterized by almost wild-type pigmentation of The fish usually lay eggs during the first hour after the light the retina, but light pigmentation of the body (bra‘ obtained is turned on.At noon, theeggs are collected and washed using from a dealer in Ttlbingen, Germany) does not complement plastic sieves. Fertilized eggs are sorted using a dissecting mi- the locus brah2 described as golden-2 by STREISINGERet al. croscope and transferred into plastic Petri dishes with “egg (1986). A recent decision on gene nomenclaturein zebrafish water” (0.03% Instant Ocean salt mix in Millipore MilliQplus resulted in renaming the golden-2 locus brass (MULLINSet al. deionized water). To inhibitfungal and bacterial growth, the 1993; WESTERFIELD1993). Finally, the embryos homozygous egg water is supplemented with 0.5 mg/liter methylene blue mutant for the mutationsparseb5 (spa;kU3SINGER et aL. 1986) (Argentum Laboratories, Redmont, Washington). exhibit melanocytes of wild-type color, but their distribution Raising and feeding zebrafiih Five days after fertilization, on the body deviates from that of the wild type. In spa/spu larvae are transferred fromPetri dishes to 1-liter beakers and 3day-old embryos melanocytes are missing from the anterior fed with paramecia daily. Paramecia are grown in an axenic portion of the head,accumulate posterior to the ears and are medium (SCHONEFELDet al. 1986) until it becomes clear. Then smaller. paramecia are harvested and washed using filter paper (What- We constructed a quadruply heterozygous stock albh4/f, man No. 1) or centrifugation (Hermle basket centrifuge bra”’/+, golh’/+, spa”5/+by crossing a triply mutant stock SIEVA, setting 50). Every day paramecia from 40 ml of culture ( albb4/albb4,golb’/golb’, spab5/spab5)with brab2/ bra“ mutant are fed tostocks of 50-100 larvae. Ten days after fertilization, fish. We used this heterozygous stock for testcrosses because its the larvae are transferred to “larvae rearing tubes”: acrylic fertility proved superior to that of any multiply homozygous tubes 10 cm in diameter with a 0.5-mm nylon mesh attached stocks we have generated. to thebottom with silicone glue. Themesh allows an easywater Handling andinactivation of EMS and ENU: Both EMS and exchange without loss of the larvae. The larvae rearing tubes ENU are highly mutagenic and carcinogenic and were are placed in 2-liter beakers and aerated at about one to two handled with extreme caution. All work with these substances air bubbles per second. Paramecia from 80 ml of culture are was performed in a chemicalhood with appropriate protective fed per 50-100 larvae daily. Around day 15-25 after fertiliza- clothing. Solutions containing EMS or ENU and any equip- tion, we start tofeed freshly hatched brine shrimpArtemia sp. ment or accessories contaminated with these substances wet-e (small size: Argentemia Platinum, Argentum Laboratories), inactivated by incubation in a 10% solution of sodium thio- and larvae are transferred, within the tubes, into fish tanks. sulfate (inactivating solution) for atleast 24 hr (SHEDLOVSKYet This allows continuous flow of fresh system water through the al. 1986). For EMS the inactivating solution was held at 37”. tubes. At this stage larvae are fed with freshly hatched brine For ENU the inactivating solution was held at roomtempera- shrimps Artemia sp. (San Francisco Bay Brand, Newark, Cali- ture and adjusted to pH -10 with sodium hydroxide. fornia) andSpirulina algae powder (Argentum Laboratories). EMS and ENU were purchased fromSigma in sealed bottles. At the age of 1 month, fish are transferred to 4liter tanks and Bottles containing EMS were opened and theliquid was trans- fed with brine shrimps and flake food (Aquarian Growth Food, ferred using a pipetman. Isopac bottles containing ENU POW- Mardel Laboratories, Glendale Heights, Illinois).At the age of der were never opened. Todissolve ENU, 0.03% Instant Ocean Germline Mutagenesis in Zebrafkh 1405 solution buffered with 10 mM sodium phosphate, pH 6.5, was heterozygous females werescreened for larvae withone of the injected into ENU bottles which were subsequentlyshaken on albino, golden, brass or sparse phenotypes. During the screen- a rocking platform for 1.5-2.0 hr. Theisopac bottles contained ing, each larva was viewed from the dorsal and lateral sides. approximately 1 g of ENU, yielding a solution of final con- Larvae carrying putative allelicmutations at the tester loci centration of approximately 100mM. The pH of the ENU so- were raised and kept individually in 1-liter boxes.To verify the lution has not been determined. The dissolved ENU solution mutations predicted by embryonic phenotypes, singlepair was used within 2.5hr from the moment of introduction of the matings wereperformed between the mature fish carryingthe buffer into the bottles. To achieve the desired final concen- putative new mutant allele and a tester fish homozygousfor a trations of the two mutagens, appropriate volumes of EMS or mutation at the tested locus. The embryonic pigmentation of the 100 mM ENU solution were transferred to I-liter Erlen- phenotype of the resulting progeny was examined as described meyer flasks containing 500 ml of 0.03% Instant Ocean and 1 above. If the embryos exhibited wild-type pigmentation, ad- mM phosphate buffer at pH 7 for EMS, and at pH 6.5 for ENU. ditional crosses with fish carrying the other tester mutations The flasks were immersed in a 21" water bath. were performed. Treatment of zebrafish maleswithEMS and ENU: Wild-type Screening for embryonic lethal mutations:The frequency of zebrafkh males 4-8 months old were transferred to I-liter embryonic lethal mutations induced in germ cells ofmu- Erlenmeyer flasks containing a mutagen solution at 21". Up to tagenized zebrafish males was assessed in a three generation three males were placedin each flask. At the endof a treatment of crosses screen (an F, screen). the mutagen solution from Erlenmeyer flaskswith mu- Production of an F2 screen: To generate F, founders from tagenized males was poured through a fish net held over a EMS mutagenesis, mutagenized males were crossedat weekly container filled with the alkaline sodium thiosulfatesolution. intervals with wild-type females withina period of 4 weeks fol- The recovered males were immediatelytransferred to a tank lowing mutagenesis. F, founders from ENU regimens were filled with system waterat room temperature. Recovery tank(s) generated in weekly crosses between mutagenized males and containing up to 15 males each were placedin a water bath at wild-type females, beginning 3 weeks after mutagenesis. When 21", and the temperature was gradually raised to 28".Any ani- mature, each F,fish was crossed with either wild-type fish mals that died were removedfrom the recovery tankand trans (F, X wild-type line), or alternatively with another F, fish (F, ferred into the alkaline sodium thiosulfate solution. On the X F, line). F, fish that failed toproduce more than 60 embryos following day mutagenized males were usuallyfed once with were returned to tanks and reused for the production of F, live brine shrimps, and their water was exchanged twice. lines. Embryos from crosses in which more than 60 embryos Mutagen-containingwater was transferred to the sodium thio- survived until day 1 of development were raised as described sulfate container. At the endof the day the mutagenized males above. were transferred to a tank in the fish facility. Screening F, lines for embryonic lethal mutations: When F2 Testing efficiency of mutagenesis: The efficiency of mu- fish reached 3-4 months of age, they were screened for em- tagenesis was tested by two approaches. First, we determined bryonic lethal mutations. For each F, line, up to 25 single-pair the frequency of induced mutations at oneor several pigmen- crosses between siblings wereperformed. Eggs from successful tation loci by the specific-locus test(RUSSELL 1951). In the sec- crosses were collected in Petri dishes as described above. Up ond parallel experiment, the effkiency of mutagenesis was es- to 30 fertilized eggs from one cross were transferred to each timated by measurement of the frequency ofzygotic of three wells of six-well tissue culture plates, so that a mini- embryonic lethal mutations in the progeny of mutagenized mum of 20 and a maximum of 90 embryos from any given cross males by an F, screen. These tests are described in detail below. were screened. Embryos were viewed witha dissecting micro- Testcrosses with pigmentation mutant fiih: The specific- scope at day 1,2,3 and 5 of development. At each time dead locus tests were carried out by crossing mutagenized males or degenerated embryos were removed from the dishes, and with tester females carryingone orseveral pigment mutations. their numbers were recorded. A detailed screening protocol During the first 2-4 weeks after mutagenesis the treated males for monitoring various aspectsof embryonic development will were crossed once per week with albb4/aZbb4tester females to be provided elsewhere. We operationally defined an embry- determine the frequency of the newly induced mosaic and onic lethal mutation as leading to death or deviations from non-mosaic mutants at thealb locus. Usually 1 month after the normal development until the end of day 5 of development. completion of a mutagenic regimen females heterozygousfor A phenotype was considered to represent a lethal mutation if each of the four loci alb", golbl, brat and spab5were used in- approximately 1 of 4 (20-30%) of all embryos developed a stead of the albino tester females. Inthese crosses we screened consistent set of defects before death, orexhibited comparable exclusively for non-mosaic mutants at these loci. Specific-locus abnormalities at day 5 of development, and the phenotype was tests werecarried out once or twice per week, up to 4 months found in separate wells. after mutagenesis. Dominant lethal lesions in sperm of mutagenized males: Single pair matings wereperformed between mutagenized The frequencies at which dominant lethal lesions were in- males and tester females. Resulting egglays were individually duced at different stages of male germ celldifferentiation were collected in Petri dishes and fertilized eggs were sorted out as determined by the analysis of the testcross progeny of mu- described above. Gomales from all EMS and oneENU (ENU4) tagenized males. After1 day of development degenerated em- mutagenesis regimens were kept in common tanks with other bryos were counted and removed from the dishes. At day 3 of males from the same regimen. Males from the remaining ENU development embryos wereanesthetized, and the number of regimens, after the first testcross, wereseparated and kept in both normal embryos and embryos with morphological ab- individual 1-liter boxes. normalities was determined. On the 3rd day of development larvae were anesthetized Fertility of mutagenized males: To monitor the fertility of with 0.017% (w/v) 3-aminobenzoic acid ethylester(Sigma) in mutagenized males, we determined the fraction of fertilized 20 mM Tris HCl, pH 7 (WESTERFIELD1993). Using a dissecting eggs in egglays from the testcrosses. The collected egglays were microscope (Wild M5 and Zeiss Technival 2) we screened inspected under a dissecting microscope2-6 hr after mating. among testcross progeny of albino females for larvae with al- At that time an egglay was a mixture of fertilized eggs,unfer- bino eye and body pigmentation, or for larvae with eyes mosaic tilized eggsand degenerated eggs. Degenerated eggs included for the albino phenotype. Progeny of crosses with quadruply abnormal eggs that could not be fertilized, fertilizedeggs that 1406 L. Solnica-Krezel, A. F. Schier and W. Driever iq ments with 1 and 2 mM ENU. As expected, all males were f, alive at thecompletion of each of these treatments, and none died following 1-hr treatments with 1-3 mM ENU 5mMENU (Table 1).However, all males from 2-hr treatments with ."e 4 mM ENU 1 and 2 mM solutions of ENU died within 2 hr after the

....0 .... 2 mM ENU completion of a treatment. After 1-hr treatments with 4 or 5 mM solutions more than 50% of the males died 1 mMENU within 2 hr. This indicated that the latter four treatments 0.5 mM ENU would not be practical for mutagenesis. The 1-hr treat- ments with 3 mM ENU performed at a higher tempera- ture (25") were also found to be too toxic, as less than time [hl 60% of treated males were alive 2 hr after treatment FIGURE2.-Toxic effect of ENU on wild-typezebrafish (data not shown). males. Fraction of males deceased at different times of incu- bation in vaned concentrationsof ENU at 21". Total number Mutagenic regimens: Based on the toxicity studies we of males (N) for each experiment was: 0.5 mM ENU solution chose the following conditions for a single treatment (N= 30); 1 mM (16); 2 mM (14); 4 mM (6); and5 mM (6). with EMS: 3-hr incubation in 10 or 20 mM solutions, and 1-hr incubation in 40 mM solution. For ENU we chose degenerated during the early divisions and degenerated un- 1-hr treatments with 2.0, 2.5, 3.0 and 3.5 mM solutions. fertilized eggs. The various classes of degenerated eggs were In mice, repeated doses ofENU allow adequate not distinguished. However, as the defective eggs are mainly survival and yield higher specific-locus mutation fre- dependent on the females, their levels were constant in our experiments. Any changes in fraction of fertilized eggs in the quencies than can be obtained with single doses at testcrosses should reflect the fertilization ability of the sperm higher concentrations (HITOTSUMACHIet al. 1985). We from mutagenized males. therefore also tested repeated dose regimens for both Statistical analysis: Confidence intervals for thefrequencies mutagens. of specific-locusmutations were calculated assuming that the For ENU, we tested six different mutagenic regimens. distribution of mutations can be approximated by the Poisson distribution, using tables of CROWand GARDNER1959. Allcom- In one series, 1-hr treatment with 3 mM ENU was re- parisons between mean values of different variables studied peated two, four and six times. In the second series, were performed using the t-test, assuming a significance level treatments with 2.0, 2.5, 3.0 and 3.5 mM ENU solutions of 0.05 unless indicated otherwise. All statistical analyses were were repeated four times. It has been demonstrated for done using the computer programs StatView I1 (Abacus Con- mice that applications of ENU spaced seven days apart cepts) and Microsoft Excel (Microsoft Co.) . resulted in recovery of fertility,whereas in regimens with shorter intervals, fewer males remained fertile (HITOT- RESULTS SUMACHI et al. 1985). We therefore chose the time in- Determination of the toxiceffect of EMS and ENU on terval between single treatments of ENU to be 1 week. zebrafiih males:As a methodof in vivodelivery ofEMS For EMS, the treatments were repeated three or six and ENU to zebrafish males, we used incubation of times at 3-4day intervals. males ina mutagen solution (M. MULLINSand C. Survival and fertility of mutagenized males: Practi- N~JSSLEIN-VOLHARD,Tubingen; A. SHIMAand A. SHIMADA, cally all males survived the EMS regimens and most of Tokyo, personal communications). Preliminary toxicity them survived for at least 3 months following the treat- tests indicated that in 1-hr treatments at 21" EMS was ments (data not shown). The survival of ENU-treated lethal at concentrations higher than50 mM. The toxicity males during consecutive mutagenic treatments is of EMS was tested further by exposing zebrafish males shown in Figure 3A. Only 50% of males survived four to 10, 20 and 30 mM solutions of this compound and treatments with 3.5 mM ENU. Survival of males treated monitoring lethality at hourly intervals. This experiment with 3 mM ENU was quite variable between regimens indicated that dying started after 1 hr of incubation in (Figure 3A and data not shown). Greater than 90% of 30 mM solution, after 3 hr in 20 mM and only after 6 hr males survived treatments with 2 and 2.5 mM mutagen in 10 mM solution of this compound. solutions. The survival following the completion of a The toxicityof ENU was studied by incubating regimen, shown inFigure 3B, was poorest for the groups males in a series of solutions from 0.5 to 5 mM at a of males treated six times with 3.0 mM ENU where 70% temperature of 21" and monitoring deaths at hourly of the males that survived the regimens died within the intervals. The results (Figure 2) indicatedthat ze- first 2 months. In contrast,only 15-40% survivors of the brafish males die after 2 hr in 4 and 5 mM solutions of four treatment regimens died during the 3subsequent ENU, and only after longer periods at lower concen- months. trations of this compound. Male mice mutagenized with ENU go through a tran- We then examinedsurvival during andafter 1-hr treat- sient period of infertility (HITOTSUMACHIet al. 1985). ments at concentrations of 1-5 mM ENU, and 2-hr treat- Thus we monitored fertility of zebrafish males exposed Germline Mutagenesis in Zebrafkh 1407

TABLE 1 Survival of males after completion of single ENU treatments

1-Hour treatment 2-Hour treatment

No. of fish No. of survivors No. of fish No. of survivors ENU concentrationat the completion 2completion hrthe after at the 2 hr after the (mM) treatmentof a treatment of a treatmenttreatment

1 9 9 NA NA 2 9 9 6 0 3 6 6 6 0 4 6 3 NA NA 5 10 3 NA NA

a Not assayed.

mens, the fertility during thefirst 2 weeks after comple- tion of a regimenwas significantly lowerthan thefertility in the 3rd month after mutagenesis (ENUS, P < 0.003; + 4X2mM1301 ENU6, P < 0.02; ENU7, P < 0.001; ENU5, P < 0.0001; "0" 4x2.5mM [301 ENU8, P< 0.0001). The reduction of fertility during the ".*" 4x3mM I331 first 2 weeks after the treatmentswas dosage dependent. "-4"-4x3.5mM 1301 That is, males treated twice with 3 mM ENU fertilized --+--6x3mM1321 eggs at significantly higher rates than those treated four *...... m 1211 6x3mM times (PC0.09) ; and males treated fourtimes were more --e--4x3mM1301 fertile then those treated six times (P< 0.06). The fer- 0 2 4 6 tilityof mutagenized males increased gradually over Consecutive treatments time reaching plateau values within 3 weeks, or in case A of higher dosages, within 2 months of completion of treatments. Based on these observations we concluded that ENU treatments lead to a dosage-dependent severe reduction of fertility in the first 2 and upto 4 weeks after completion of mutagenic treatments. Dominant lethal lesions induced by ENU and EMS: 4x3d [191 Chemical mutagens are known to induce dominant le- ----O---6x3mM [SI sions, and their frequencyis assumed to be an indicator ---A" 6x3mM 1291 of the mutagenicity of a given compound (EHLINGet al. - - 4-- - 4x2.5d 1301 1968). We therefore monitored embryonic death and morphological abnormalities among progeny of mu- 0 2 4 6 E 10 12 14 16 tagenized males and tester females. We distinguished B Weeks after mutagenesis three categories of embryos: (i) embryos that died and FIGURE3.Purvival of males during and after mutagenic degenerated during thefirst 24 hr of development and regimens. (A) Survival of males during consecutive mutagenic whose specific phenotype could not be assessed, (ii) treatments as a fraction of the initial population in the ENU regimens: 4 X 2 mM; 4 X 2.5 mM; 4 X 3 mM; 4 X 3 mM; 4 X monstrous embryos that developed body malformations 3.5 mM, 6 X 3 mM, and 6 X 3 mM. All regimens were started until the 3rd day of development and (iii) embryos that with 30-33 wild-type males.(B) Survival of mutagenized males were normal until at least the 3rd day of development. after the completion of a mutagenic regimen, as a fraction of Table 3 shows changes in the frequenciesof these three the survivors of the regimen (N):2 X 3 mM (16); 4 X 2.5 mM classes ofembryos among thetestcross progeny of males (30);4 X 3 mM (19);6 X 3 mM (19), and 6 X 3 mM (29). subjected to different EMS and ENU regimens. Among to ENU by determining the fractionof fertilized eggs in the progeny of males treated with EMS there was only a egglays obtained in testcrosses performed at different small reduction in the frequency of normal embryos times after the completion of mutagenesis (Table 2). during the first weeks after treatment, when compared Treated males were able to fertilize eggs at all times after to crosses performed at latertimes after mutagenesis or mutagenesis. Their fertility, however, was significantly to controlcrosses. A much higher dominantlethality was lower than that of untreated wild-type males of similar observed among progeny of males treated with ENU. age (76.7 2 25.9%) even 3 months after the last ENU Among the testcross progeny obtained during the 1st treatment (ENUS, P < 0.03; ENU8, P < 0.03; ENU6, P < week after treatments, less than 1% of embryos were 0.02; ENU7, P < 0.006). Additionally, for all ENU regi- normal, up to 70% of fertilized eggs died during thefirst 1408 L. Solnica-Krezel, A. F. Schier and W. Driever

TABLE 2 Fertility of males after different ENU regimens

1-2 Weeks 3-4 Weeks 2 Months 3 Months after regimen after regimen after regimen after regimen Percent Percent Percent Percent No. of fertilized NO. of fertilized No. of fertilized No. of fertilized Regimencrosses (mean 2 SD) crosses (mean 2 SD) crosses (mean ir SD) crosses (mean t SD)

ENUG 2 X 3 mM, 1 hr 23 65.5 4238.0 232 32.4 65.1 2 22.4 45ir 24.7 58.6 t 29.3 ENU7 4 X 3 mM, 1 hr 22 24.0 t 21.0 20 38.8 t 29.8 53 56.6 ir55.0 25.224 t 25.0 ENU5 6 X 3 mM, 1 hr 22 12.9 t 17.2 22 29.2 If: 24.6 26 54.5 t 27.5 23 56.3 2 25.6 ENUS 4 X 2 mM, 1 hr 40.738 2 23.5 56 52.8 If: 27.4 111 54.7 2 29.1 33 59.7 ir 28.0 ENU84 X 2.5 mM, 1 hr 38 26.9 2 21.5 39 48.6 ir 20.5 63 48.0 2 22.5 63.2 34 t 19.7 Males were crossed with tester females at weekly intervals. The fraction of fertilized eggs in an egglay was determined as described in MAX-S AND METHODS. A mean and the standard deviation of fraction of fertilized eggs was calculated for all egglaysobtained from one set of mutagenized males during: 1-2 weeks; 3-4 weeks; 2nd month; and 3rd month time periods. The fertility of untreated wild-type males of similar age was 76.7 t 25.9%.

TABLE 3 Changes in frequencies of dominant lethal mutants and normal embryos among progeny of EMS-and ENU-mutagenized males

Weeks after mutagenesis (frequencies given as a mean 2 standard deviation) Phenotypic Mutagenic regimen category 8 1 4 2 3 12

EMS6 6 X 20 mM, 3 hr Dead at 24 hr 0.160.15 ? 0.10 0.15 ir 0.160.27 2 0.33 0.13 ir 0.10 NA Monstrous 0.04 ? 0.05 0.03 2 0.02 t 0.03 0.03 t 0.02 NA Normal 0.810.80 rt 0.810.14 t0.71 0.18 t 0.870.31 2 0.15 NA

EMS7 6 X 40 mM, 1 hr Dead at 24 hr 0.250.07 rt 0.070.08 rt 0.110.27 ir 0.33 0.13t 0.10 NA Monstrous 0.14 0.05 ? 0.02 0.15 t 0.06 0.02 t 0.03 0.03 2 0.02 NA Normal 0.61 0.89 rt 0.09 0.77 ir0.71 0.12 ir 0.32 0.84 ir 0.11 NA

ENUG 2 X 3 mM, 1 hr Dead at 24 hr 0.65 ir 0.220.47 2 0.27 0.23 2 0.18 0.26 2 0.25 0.17 ir 0.18 0.15 rt 0.13 Monstrous 0.34 rt 0.22 0.09 2 0.050.09 rt 0.070.05 t 0.05 0.05 t 0.040.05 t 0.06 Normal 0.01 t 0.010.44 rt 0.24 0.68 t 0.23 0.69 t 0.240.78 2 0.18 0.78 ir 0.18

ENU74 X 3 mM, 1 hr Dead at 24 hr 0.77 t 0.180.32 ? 0.270.12 t 0.090.11 t 0.080.10 rt 0.110.08 t 0.07 Monstrous 0.25 2 0.150.17 ? 0.13 0.05 t 0.050.08 ir 0.05 0.05 t 0.050.05 2 0.08 Normal 0.01 t 0.020.51 ? 0.31 0.84 2 0.070.82 ir 0.090.85 t 0.140.87 rt 0.12

ENU56 X 3 mM, 1 hr Dead at 24 hr 0.83 ir 0.22 0.33 2 0.16 0.16 t 0.120.19 2 0.100.07 t 0.070.10 t 0.10 Monstrous 0.26 t 0.200.11 2 0.080.09 2 0.090.05 5 0.05 0.03 t 0.020.04 2 0.05 Normal 0.01 t 0.010.60 t 0.14 0.75 ir 0.170.80 2 0.090.90 ir 0.97 0.87 t 0.15 Mutagenized males were crossed withtester females at weekly intervals. For each egglay, the fraction of fertilized eggs that (1) died during the first 24 hr of development, (2) exhibited morphological abnormalities at the 3rd day of development and (3) were normal on the 3rd day of development was determined, as described in MATERIALSAND METHODS. Mean and standard deviation reflect variation among egglays from one set of males in a given week. Normal embryos comprised 88.9 2 11.6% of progeny from control crosses between wild-type males and quadruply heterozygous females. NA = not assayed.

24 hr of development, and most of the remaining em- haploid individuals (STREISINGERet al. 1981). Haploid bryos exhibited morphological abnormalities by the 3rd development is known to be induced by spermatozoa day of development. The fraction of normal embryos whose genetic material has been inactivated, by W ir- increased to 40-50% in progeny from testcrosses per- radiation for example (STREISINGERet al. 1981). We ob- formed during the 2nd and 3rd week after treatments. served haploid embryos predominantly in crosses per- Four weeks after mutagenesis normal embryos com- formed during thefirst 2 weeks after completion of ENU prised 80-90% of the testcross progeny, a value not sig- regimens. During this time they were detected in 21% nificantly differentfrom that observed forcontrol of crosses withmales treated with the ENUS (4 X 2 mM; crosses between untreated wild-type males and quadru- 1 hr) regimen, in 38% of crosses after the ENU8 (4 X ply heterozygous females (88.9 2 11.6%). 2.5 mM; 1 hr) regimen, and in 46% of crosses following Among progeny of ENU-, but not of EMS-treated the ENU 7 (4 X 3.0 mM; 1 hr) regimen. In a total of 29 males, phenotypically albino embryos were found that crosses with 10 or morefertilized eggs in which anyhap also exhibited morphological features characteristic of loids were observed, haploid embryos constituted 8 2 Germline Mutagenesis in Zebrafish 1409

€NU or EMS

1 - 6 weeks after 1 - 4 months after mutagenesis mutagenesis

DNA-DS 8lb" a/W+; got+; bW+; spa+ quadruply heterozygous tester stock

alW+: aWlalb alW+ alb"lalb; goU+; bra+: spy+ a/&+; gou+; brY+; spy+ mosaic progeny wild type pigmentation non-mosaic wild type new albmutant pigmentation

A B FIGURE4.-Testing the efficiency of mutagenesis by specific-locus test. Fraction of mosaic and non-mosaic mutants among the progeny of mutagenized males and tester femaleswas determined. (A) During the first3-6 weeks afterthe completionof mutagenic treatments mutagenized males were crossed with albino tester females( albb4/albb4). All the resulting progeny are heterozygotes albb4/+ and phenotypically wild type, unlessa sperm carriesa modification of single strand ofDNA [as in (A)] or a fixed mutation [as in (B)].A modification of a single strandof DNA can causea mutation duringDNA replication in a developing embryo, which eventually is a mosaic ofwild-type tissuealbb4/ + and of mutant tissue albb4/alb* (* indicated a newly induced allele).When a sperm cell carries a fixed mutation the resulting embryois uniformly albb4/alb* and is phenotypically albino. The frequencyof mosaic (non-mosaic) mutants at thealb locus was calculated as:no. of mosaic (non-mosaic) mutant embryos/total no.of normal embryos (equal to the no.of tested alb chromosomes). (B) In the second partof the specific-locus test, the albino females were substituted with quadruply heterozygous females. Every egg produced by such a female carries on average any two of the four tester chro- mosomes. The resulting progeny are wild type unless sperma mutagenized from male carries a mutation that does not complement one of the tester mutations. The mean frequency of mutations per locus per gamete in testcrosses was calculated as: no. of non-mosaic mutant embryos found/no. of normal progeny X 2 (average no. of tester chromosomes in each embryo).

14% of embryos. The peak of occurrence of haploid ENU. For ENU up to 1%, and forEMS up to 4% of sperm embryos coincided in time with the peak in frequency cells from mutagenized males yielded embryos mosaic at of dominant lethal lesions induced by ENU. this locus. For both mutagens the frequency of mosaic Frequenciesof specific-locus mutants induced by mutants decreased as a function of time after comple- EMS and ENU EMS and ENUefficiently induce mosaic tion of mutagenesis. The highest frequencies (3-44 X mutations at the alb locus: Drosophila males treated were observed in the testcrosses performed within with alkylating agents initially produce F, progeny mo- the first 2 weeks after completion of a regimen. The saic forthe induced mutations, and only laternon- frequencies of albino mosaics were lower in the 3rd and mosaic mutant progeny (FAHMYand FAHMY1957). Be- 4th weeks after mutagenesis (0.4-10 X and less cause our objective was to generate non-mosaic mutants, than 0.6 X beyond onemonth aftercompletion of it was important to determinehow long mosaic mutants a regimen. would persist among the progeny after the mutagenic ENUbut not EMS efficientlyinduces non-mosaic treatments of zebrafish males. Frequencies of mutations mutants: The frequencies of locus-specific non-mosaic induced at a specific locus and the mosaic us. non- mutants recovered fromdifferent mutagenesis regi- mosaic character of the mutant progenyof mutagenized mens areshown in Table5. For EMS regimens, only the males were determined by crossing males with tester mu- alb locus was tested. For ENU regimens, both the num- tant females as illustrated in Figure 4. An example of a ber of alb chromosomes, as well asthe numberof chro- mosaic embryo recovered in such a testcross with albb/ mosomes containing all four loci tested for each set of aZbb4 homozygous females isshown in Figure 5. The males are given. In cases when all four tester loci were a1b*/albb4 (db* indicates a newly induced mutation) used, the frequenciesreflect average numbers of muta- patches are visible both in the( albb4/+)eye retina (Fig- tions per locus, although each locus was found to have ure 5A) and on the body (Figure 5B) of the otherwise a different mutability (see below). normally pigmented embryo. The minimal number of independent mutational Table 4 shows the temporal changesin the frequency events shown in Table 5 reflects only the first mutation of mosaic mutants atthe alb locus detected amongprog- at eachtester locus recovered from asingle mutagenized eny of males treated with different regimensof EMS and male. The frequencies given in thelast column of Table 1410 L. Solnica-Krezel, A. F. Schier and W. Driever

FIWRE~.-MOS~~Cand non-mosaic carriers of the ENU- induced alleles at the alb locus. (A and B) Fiveday-old embryo recovered in a cross between an ENU-treated maleand albb4/albb4 homozygous female.The embryo is mosaicfor an ENU-induced mutation at the alb locus and exhibits patches phenotypicallyal- bino in the otherwise pigmentedretina (A) and on the body (B). (C) Group of Sday-old embryos. The rightmost embryois wild- type (alb/alb) and exhibits dark pigmentation of the retina and of the body, while the leftmost embryois a homozygous for the standard albb4 allele. The four embryos in the middle contain each one alb" allele and one of the different, independently in- duced hypomorphic allelesof the alb locus.

5 are based on all mutations, including multiple muta- or three different testerloci; these mutations were tions at the same locus, recovered from a single male. clearly of independent origin. Therefore, the true mu- Multiple mutations recovered from the same male may tation rates were probablyhigher than our minimal fre- have arisen during independentmutational events. Al- quency estimate, but may be lower than the total mu- ternatively they could represent sibling clones derived tation frequencies. from single mutational events that subsequently were The most important finding of the specific-locus tests multiplied during germ cell proliferation. Several ob- was that all ENU regimens resulted in high frequencies servations supported the notion that some of the novel of recovered non-mosaicmutants (0.5 X lo-' and up to mutant alleles recoveredfrom a single maleoriginated 3.9 X io-'). In contrast, only low mutation frequencies from independent mutational events. First,we observed (<0.1-0.2 X lo-') were observedfor EMStreated males. that two sibling mutant progeny carried novel alb mu- Two additional questionswere addressed by the tations of different strengths. Second, of six cases of mu- specific-locus test. First,was the frequency of mutations tant alleles derivedfrom a single malein crosses withthe induced by ENU dependent on the mutagenic dosage? quadruply heterozygous females,five weremutations in Figure 6A shows the mean number of mutations per spa, the most mutable of the tester loci (see below). locus per gamete induced for one series of mutagenic Finally, several malesproduced progeny mutated in two regimens in which l-hr treatments with 3 mM ENU were Germline Mutagenesis in Zebrafish 141 1

TABLE 4

Frequencies of mosaic mutants at the alb locus

1-2 Weeks after regimen 3-4 Weeks after regimen >1 Month after regimen

Mosaic mutants Mosaic mutants Mosaic mutants Chromo- Chromo- Chromc-

Frequency a somes Frequency" somes Frequency a somes Regimen No. x 103 tested No. x 103 tested No. x 103 tested

EMS4 1 X 10 mM, 3 hr 2 X 20 mM, 3 hr 14 (27-67)44 4 315 10 (4-21) 414 1 0.3 (0-2) 2893 EMS5 3 X 20 mM, 3 hr 2130 (15-28) 1408 2 2 (0.4-5) 1267 1 0.6 (0-3) 1589 EMS6 6 X 40 mM, 1 hr 2 (2-25) 8 240 0 <1.0 (<2) 1019 0 ~0.4(

TABLE 5

Frequencies of non-mosaic mutations inducedin four tester loa

Minimum of independent Total no. of mutational events mutational eventS

Frequency (X 10') Frequency (X lo3) No. of chromo- No. of 90% No. of 90% Locus somes mutants Point confidence mutants Point confidence M utagen DosageMutagen tested screened observed estimate observedintervals estimate intervals

EMS4 1 X 10 mM, 2 X 20 mM, 3 hr alb 8,204 - - 0.01-0.55 - 0.12 1 EMS5 3 X 20 mM, 3 hr alb 7,897 - - - 1 0.01-0.57 0.13 EMS6 6 X 20mM, 3 hr alb 2,932 - - - 0 <0.34 <0.83 EMS7 6 X 40 mM, 1 hr alb 3,546 - - - 0 <0.28 ~0.69 ENUG 2 X 3 mM, 1 hr alb 3,486 - - - 3 0.86 0.32-2.15 alb, gol, alb, bra, spa 12,360 5 0.40 0.20-0.796 0.49 0.20-0.92 ENU74 X 3 mM, 1 hr alb 1.402,148 3 0.51-3.502.33 5 1.13-4.52 alb, gol, bra, spa 13,910 17 1.22 0.82-1.76 24 1.73 1.15-2.40 ENU46 X 3 mM, 1 hr alb 1,803 - - - 7" 3.88 1.99-6.95 alb, gol, alb, bra, spa 11,028 13 0.68-1.821.18 15" 0.86-2.02 1.36 E NU 5 6 ENU5 X 3 mM, 1 hr alb 0.46-3.12 1.25 3 2,407 5 1.01-4.04 2.08 alb, gol, alb, bra, spa 2,966 3 1.01 0.37-2.53 6 0.82-3.82 2.02 EN US 4 ENUS X 2 mM, 1 hr alb 3,233 - - - 2 0.16-1.85 0.62 alb, gol, alb, bra, spa0.68 1217,520 0.43-1.05 13 0.74 0.43-1.14 EN U8 4 ENU8 X 2.5 mM, 1 hr alb 2,399 - - 0.04-1.89 - 0.42 1 alb, gol,spa bra, 1927,224 0.70 0.46-0.99 21 0.77 0.50-1.10 ENUlO 4 X 3 mM, 1 hr alb 300 - - - 1 3.33 0.35-15.1 alb, goZ, bra, spa 12,204 16 1.31 0.80-1.95 17 0.93-2.00 1.39 ENUll 4 X 3.5 mM, 1 hr alb, gol, bra, spa 5,744 - - - 8 1.39 0.79-2.36

Mutagenized males were crossed withtester females at weekly intervals, and mutant embryos were screened for as described in MATERIALSAND METHODS. The table presents data obtained from the whole set of mutagenized males. * Males from this set were kept in a common tank. Therefore the minimal number of mutations was calculated as: (total no, of mutants) - (no. of multiple mutants in the same locus from one egglay). This number may be an overestimation of the real minimal number of independent mutants, as single mutants in egglays obtained at different time points could be derived from the same male. repeated two, four and six times. There was more than the numberof treatments was increased to six. Figure 6B a twofold difference in the frequency of mutations be- shows the frequency of specific-locus mutations as a tween the treatments repeatedfour andtwo times. How- function of concentration of ENU for a series of regi- ever, only a small further increase of mutation frequen- mens in which1-hr treatments were repeated fourtimes. cies was observed for one of the two regimens in which There was more than a twofold increase in mutation 1412 L. Solnica-Krezel, A. F. Schier and W. Driever

A cies adjusted for clusters nor for the frequenciesreflect- ing all mutations recovered in the specific-locustest (data not shown). 4.5 1 T Testing the allelism of mutations recovered in the specific-locus test: Mutations detected in the specific- locus test by virtue of not complementing one of the tester loci could represent new allelic mutations at the given locus. Alternatively, they could represent domi- nant mutationsat distinct loci, or non-allelic non- complementing mutations. To distinguish between these possibilities, we tested the fish carrying putative new mutant alleles in crosses with homozygotes at the

1 tester locus in question. For example, a fish carrying a Ol I I I 1 new mutant allele at the alb locus should yield exclu- 0 2 4 6 8 sively albino progeny in crosses with alb/alb tester fish. Number of treatments The presence of wild-type embryos would indicate that B the new mutation is at a locus distinct from alb. 251 A total of 137 mutants were detected in the specific- locus test. Neither of the two fish carrying mutations induced by EMS survived until testing. Frotn the 135 fish carrying the ENU-induced mutations, 52 survived to adulthood, and onewas sterile. The allelism tests were completed for 44 fish and confirmed that all of them carried a mutationat oneof the tester loci. Table 6 shows representative examples of such tests. In most cases (37/ 40) the tests confirmed the allelic classification done at embryonic stages. Additionally, these tests indicated that z some of the newly induced albino and golden mutations, f "1 represented weaker mutant alleles of these loci. When

01 I I 1 I fish carrying such weaker albino alleles were crossed 0 1 2 3 4 with albb4/albb4tester fish, two classes of embryos were ENU [mMI observed among the progeny. 50% of the albino em- FIGURE6.--Specific-locus frequencies of non-mosaic mu- bryos were darker than the remaining alb"4/alb"4ho- tants as the functionof a mutagenic dosage plotted with90% mozygotes and representedheterozygotes of an old albb4 confidence intervaIs. Given frequencies were based on the to- tal number of mutational events (seetext and Table 5 for de- and a new weaker alb* allele (Table 6). Figure 5C shows tails). (A) Frequencies after the 2 X; 4 X; 6 X 3 mM ENU a series of embryos, each carrying a different, indepen- regimens. (B) Frequencies after the 4 X 2.0; 2.5; 3.0 and 3.5 dently induced mutationat the alb locus. The existence mM ENU regimens. of hypomorphic mutations at the alb and go1 loci often made it difficult to identify new mutations unequivocally frequency between regimens utilizing 2.5 mM and 3 mM in embryos derived from the crosses of mutagenized ENU. However, an increase of the ENU concentration males and the quadruply heterozygous females. In sev- from 3 to 3.5 mM has not raised the mutation frequency eral ambiguous cases new mutations could not be clas- further. sified at embryonic stages (M in Table 6). However, Second,did the frequency of non-mosaic mutant when adult fish carrying these M mutations were crossed progeny of mutagenized males change over time after with tester fish, they proved to harbor mutations at ei- the completion of a mutagenic treatment? Thiswas an ther alb or go1 loci. In three of five cases embryos clas- important issue for determining thetime period during sified as carrying new mutations at thebra locus proved which mutagenized males could be used to generate F, upon testing to carry mutations atone of the other three founders for subsequent embryonic lethal screen. To tester loci. answer this question, for each mutagenic regimen, we Based on the above tests, we concluded that most if normalized the mean number of mutations per locus not all animals recovered in the specific-locus test car- per gamete recovered during monthly intervals relative ried a mutation atone of the tester loci, or at a distinct to the 1st month, and subsequently compared average butvery closely linked gene. In addition,these tests dem- monthly mutation rates. No significant differences were onstrated that the ENU-induced mutations at pigmen- observed between mutation frequencies during a period tation loci were transmitted in the germline of the F, of 4 months aftermutagenesis, for neither the frequen- progeny of the mutagenized males. Germline Mutagenesis in Zebrafish 1413

TABLE 6

Testing allelism of the newly induced mutations us. the tester loci

~ Mutation predicted Mutation Resultsof a cross identifying a newly induced tested mutation How closely by an confirmed linked if not embryonic in test the same gene phenotype crosses No. embryos scored No. mutantwild-typeembryos embryos No. (CM)

alb alb alb639 639 0 0.2 alb alb alb 906 906 0 0.1 alb* alb* alb* 164* 310 + 146 0 0.3 go1 342 go1 342 0 0.3 gol* gol* 273 133* + 140 0 0.4 bra260 bra 260 0 0.4 spa spa spa 951 95 1 0 0.1 Spa232 spa 232 0 0.4 Ma 290*alb* 592 + 302 0 0.2 M alb* 623 295* + 328 0 0.2 M gol* 117 117 0 0.9 bra247 go1 247 0 0.4 bra alb* bra 159* 338 + 179 0 0.3 bra spa 483 483 0 0.2

~~ ~ * New allele of a different strength when compared to the canonical allele at this locus in the tester strain. M, in these cases the locus at which the newly induced mutations occurred could not be predicted by the embryonic phenotype.

TABLE 7

Comparison of relative mutability of the tester loci in the specific-locus test

Minimum of independent Total no. of mutational events mutational events

Chromosomes Alleles Frequencies Frequencies Alleles Frequencies Frequencies Gene screened recovered (X 103) 1/x recovered (X 105) 1/x

alb 25,739 19 0.74 1/1355 21 0.82 1/1225 go1 25,739 21 0.82 1/1225 22 0.86 1/1170 bra 25,739 11 0.43 1/2340 12 (6)" 0.47 (0.23) 1/2150 (4290) Spa 25,739 42 1.63 1/610 55 2.14 1/470 Total 102,956 93 0.90 1/1100 110 1.1 1/930

a Five out of 15 putative bra mutants survived and could be tested. Out of these five only two (40%)were confirmed as new bra alleles. Therefore, probably not all of the putative bra mutants carried a mutation in this locus.The numbers in parentheses are adjusted to 40% confirmed surviving mutants.

The distribution among the tester loci of mutations Recovery of induced mutations in an F, screen: The induced by ENU is shown in Table 7. The table includes numbers of embryonic lethal mutations induced in mutations confirmed in the allelism tests aswell as mu- germ cells ofmutagenized males were determined in an tations for which carriers died. It can be seen that mu- F, screen (see MATERIALS AND METHODS). An embryonic tations were induced atall four loci, with the rates at least lethal mutation was operationally defined as leading to 10-fold higher at thespa locus than at theleast mutable death or developmental abnormalities until the end of bra locus. The estimated mean mutation frequencies day 5 of development. This definition deviates from that could be underestimatedif some very weak alleles at the of CHAKRABARTIet al. (1983) who included in the group tester loci were not recognized. Furthermore, several of embryonic lethal phenotypes fish that did not show mutants that exhibited a questionable phenotype sub- an inflated swim bladder at day seven of development. sequently died, and could not be tested, were excluded Our definition is based on two reasons. First, we found from our calculations. On the other hand,the frequen- that individualswith a non-inflated swim bladder eat and cies at the bra locus could be overestimated. Just 5 out swim at day 7. We therefore consider these fish to be in of 15 putative mutants survived to adulthood and only a larval, not embryonic, stage of development. Second, two proved to carry a new bra mutation. This 40% rate several mutations with the noninflated swim bladder of confirmation of surviving putative bra mutants was phenotype that had beenisolated from distinct F2 lines extrapolated toall isolated putative mutants at this locus did not complement one another. This indicated that (Table 7). they probably preexisted in the AB wild-type stocksused 1414 L. Solnica-Krezel, A. F. Schier and W. Driever for mutagenesis. A phenotypewas considered to repre- mutations relative to F, X wild-type lines. On average. sent a lethal mutationif approximately one in four (20- only 2.12 mutations per F, X F, line were recovered 30%) of allembryos developed a consistent set of defects which corresponded to 1.06 mutations per mutagenized before death, or exhibited comparableabnormalities at genome (90% confidence interval: 0.93-1.19). day 5 of development, and the phenotypewas found in separate wells. This strict definition of embryonic lethal DISCUSSION mutation couldresult in omission of potential hypomor- As the first step toward saturation mutagenesis screens phic mutations that lead to a variable phenotype. To for embryonic lethal mutations in zebrafish we com- confirm a specific phenotype, we repeated the cross in pared thecapability oftwo chemical mutagens to induce which a given phenotype was observed and embryos mutations in the germline of males. were rescreened. More than 95% of initiallyisolated mu- Mosaicmutants-mutagenic effect on post-meiotic tations were confirmed in the rescreens. germ cells: Males treated with different EMS and ENU In the case of EMS mutagenesis we screened 71 F, mutagenic regimens produced initially a high incidence lines generated by crossesbetween twoF, founders of embryos mosaic for the albino phenotype in crosses (F, X F,) and thus containing mutations from 142 mu- with females homozygous for a mutationat thealb locus. tagenized genomes. On average 10.5 crosses were per- The frequencies of mosaic mutant progeny of mu- formed during screening of each line. F, founder ani- tagenized males decreased following the 2ndweek after mals used for the generation of F, lines were produced mutagenesis. Mosaic mutants were also reported to be by crossing the mutagenized males during the first 4 induced by EMS and/or ENU exposure of mature sper- weeks after completion of mutagenesis. The high inci- matozoa in Drosophila (ONDR~J1971), mouse oocytes dence of mosaicmutations in the alb locus was observed (LEWISet al. 1985),postgonial germ cells in mice (LEWIS during this time (Table 4). Nevertheless, only 16 em- et al. 1984,1990;FAVOR et al. 1990a), and mouse zygotes bryonic lethal mutations were detected during screen- (RUSSELLet al. 1988). The formation of mosaic mutants ing of these lines. This corresponded to 0.1 1 mutations in zebrafish via action of ENUon post-meiotic germ cells per mutagenized genome. is consistent with earlier reports that fertilization with For the ENU screen, F, lines were produced using F, sperm exposed to ENU results predominantly in mosaic foundersgenerated only laterthan 3 weeks after embryos (GOLIN et al. 1982; GRUNWALDand STREISINCER completion of mutagenic treatments. Therefore, these 1992). In ourstudies, 1-hr treatments of males with 2-3 lines were expected to harbor mutations inducedin pre- mM ENU solutions resulted in frequencies of mosaic mu- meiotic germ cells. Of 116 F, lines screened, 66 were tants at the alb locus (Table 2) very similar to those ob- derived from crosses of F, founders and wild type fish served for this locus in the case ofsperm treatedin vitro (F, X wild type), and corresponded to 66 mutagenized for 10 min with 1 and 2 mM ENU solutions (GRUNWALD genomes from the ENU4 regimen. The remaining 50 F, and STREISINCER1992). lines were produced by crosses between F, founders, and Appearance of mosaic mutant progeny after treat- thus represented mutations from 100 mutagenized ge- ments of postreplicative germ-cell stages with alkylating nomes. Screening a single F, line described here in- agents like ENU or EMS has been explained by the fact volved a minimumof 6 and onaverage 9.4sibling crosses that these mutagens can alkylate baseson onestrand of (90% confidence interval: 9.1-9.7). F, progeny were DNA. In post-meiotic germ cells or a zygote, the modi- then screened forembryonic lethal mutations. A total of fications, if not repaired, become fixed as mutations 214 embryonic lethal mutations was recovered. Each only during divisions followingfertilization. Resulting F, mutation was manifested on average in 22% of crosses embryos would be genetically mosaic (RUSSELLet al. (90% confidence interval: 21-25). This indicated that, 1988). The observed temporal changes in frequency of as expected, most of the mutations were of the non- albino mosaics (Table 4) indicated that the majority of mosaic type. Most mutants in which specific aspects of post-meiotic mutagen-sensitive cells wereutilized within development were affected exhibited unique pheno- a two week period following mutagenesis, only a small types. They most likely represented new mutations in fraction (10%) in the 3rd and 4th week. This is consis- different genes. Complementation tests among mutants tent with the studies of spermatogenesis in the related with similar phenotypes indicated that the majority of teleost, medaka (0.latipes). In medaka, it takes at least mutations affected distinct loci. 12 days for DNA-replicating spermatocytes and 7 days The results of screening F, X wild typelines indicated for spermatids to differentiate into maturespermatozoa that sperm from males mutagenized in the ENU4 and at 25" (EGAMIand HYODCFTAGUCHI1967). The rate of ENU5 regimens (6 X 3 mM ENU, 1 hr), harbored on spermatogenesis is temperature-dependent (EGAMIand average 1.70 (90% confidence interval: 1.47-1.93) em- HYODO-TAGUCHI1967). Hence, thedifferentiation in ze- bryonic lethal mutations per single mutagenized ge- brafish at 28.5" may be slightly faster. It has been sug- nome. Interestingly, F, X F, lines have not yielded the gested that in mouse germ cells, ENU-induced DNA le- expected twofold increase in the average rate of lethal sions may persist through a numberof cleavage divisions G ermline Mutagenesis in Mutagenesis Germline Zebrafish 1415 before being eventually repaired or fixed as a mutation trast, in mice ENUwas found to be less effective in (FAVORet al. 1990b). Therefore, some of the mosaic inducing mutations in post-spermatogonial germ cell albino mutants, obtained at later time points after mu- stages than in spermatogonia (RUSSELLand HUNSICKER tagenic treatments, could be caused by mutagenic le- 1984;VAN ZEELAND et al. 1985).The lower ratio of mosaic sions effected in still dividing germ cells. Alternatively, us. non-mosaic mutants in the mouse compared with they could be derived from mutagenized post-meiotic fish is not likely to be theresult of more efficient repair germ cells utilized at later time points after completion in the mouse, as murine mid-spermatids and mature of mutagenic treatments. spermatozoa are probably incompetent in DNA repair Non-mosaic mutants-mutageniceffect on pre-meiotic (SEGA1974). Rather, during rapid mitoses in fish (and germ cells: DNA lesions formed by chemical mutagens Drosophila) blastoderm, premutationallesions induced in a pre-meiotic germ cell can be fixed as mutations in spermatids and spermatozoa may have a higher fixa- before differentiation into spermatozoa. Embryos tion rate than those in very slowly developing mouse formed upon fertilization with such a sperm would be embryos. non-mosaic for mutations introducedby the sperm. Very Dominant lethal lesions: The peak in ENU-induced low frequencies of non-mosaic mutants were detected dominant lesions was observed during the first 2 weeks after EMS treatments. The high mutagenic potency of after mutagenesis. It corresponded to the peak in fre- EMS in the zebrafish on the later stages of spermato- quencies of mosaic mutants and coincided with occur- genesis and only low activityon the earlierstages ofgerm rence of haploid embryos. There was no obvious cor- cell differentiation is consistent with mutagenicity of relation between the incidenceof dominant lesions and EMS in Drosophila (FmMuand FAHMY1957; VOCEL et al. non-mosaic locus-specific mutants. Although, the cor- 1982) and in the mouse (RUSSELLet at. 1981; VAN respondence between specific-locus anddominant- ZEELANDet al. 1990). lethals patterns is found for several chemical mutagens In contrast, all ENU regimens tested resulted in high in the mouse (RUSSELLet al. 1989), this correspondence frequencies of non-mosaic mutants among theprogeny is not foundfor ENU (GENOROSO1984), consistent with of treated males. The spontaneous mutation rateof 3 X our results in zebrafish. reported for thegol locus in zebrafish (WALKERand EMS treatments lead to very high frequencies of mo- STREISINCER1983), is 50 times lower than the mutation saic mutants; however, neither haploid embryos nor frequency at this locus obtained in thisstudy (e.g., high numbers of dominant lethallesions were observed ENU7: 1.4 X lo-’). The specificlocus test utilizing four among progeny of EMStreated males. Consistent with pigmentation loci indicated thatour most efficient ENU these observations, no significant levels of dominant le- regimens yielded non-mosaic mutants at a rate of 1 per sions were observed after exposure of medaka males to 700-800 mutagenized genomes scored. Such high rates varied doses of EMS (SHIMADAand EGAMI1984). Also in are very similar to the average mutation frequencies at mice EMS proved to be less effective ininducing domi- 7 tester loci in the mouse after the most efficient re- nant lethality than otheralkylating agents (EHLINGet al. peated dose experiments (1 mutation at 1 locus per less 1968). than 700 mutagenized genomes; HITOTSUMACHIet al. Chemical basis for induction of different types of mu- 1985). The comparison of the ENU dosage utilized in tations by ENU and EMS: In post-meiotic germ cell the fish and the mouse is not straightforward. In the stages both EMS and ENU efficiently induced Iocus- zebrafish, ENU was delivered to animals by incubation specific recessive mutations (mosaic mutants). Only of fish in the mutagen solution, whereas mouse males ENU, however, induced a high level of dominant le- received intraperitoneal injections of ENU solution sions. In pre-meiotic germ cells ENU induced locus- (RUSSELLet al. 1979). specific mutations at frequenciesan orderof magnitude Two observations indicated thatapparent whole-body higher than EMS. The basis for differences in effective- mutants carried mutations inducedin pre-meiotic germ ness of ENU and EMS in induction of different types of cells, most probably spermatogonia. First, mutant fish mutations in the two cell types isnot clear. However, the carrying newly induced putative mutations attester loci, distinct spectra of DNA lesions effected by these two produced only mutant progeny in crosses with appro- mutagens are likely to play a role (VANZEELAND et al. priate testers. This indicated that their germlinewas not 1990). The high mutagenic potency of EMS in the late mosaic. Second, non-mosaic mutants were observed at stages ofspermatogenesis is explained by relatively high constant rates among testcross progeny even 4 months levels of N-alkylation and subsequent accumulation of after completion of mutagenic treatments. apurinic sites (VANZEELAND et al. 1990). The DNA le- For all ENUregimens the frequencies of non-mosaic sions induced in postreplicative cell stages leading even- mutants at thealb locus were two to threefold lower than tually to mosaic mutants may be distinct from those lead- the frequenciesof albino mosaics (Table 4). Similar dif- ing to dominant-lethal lesions and genetic inactivation ferences between these two classes of mutants were ob- of the sperm. The latter would be effected rather by served in Drosophila (FAHMYand FAHMY1957). In con- ENU than EMS. 1416 L. Solnica-Krezel, A. F. Schier and W. Driever

The higher mutagenicity of ENUcompared with EMS reduction of fertility during the first two weeks after in spermatogonial cells is explained by the ratioof 0-vs. treatments, the degreeof whichwas dosage dependent. N-alkylations which ishigher for ENU than EMS (SINGER Additionally, mutagenized males have not recovered to and GRUNBERGER1983). It is not clear which of the the fertility of control males in the 3 months following Oalkylations effected by ENU isresponsible for the high treatments. mutagenicity of this compound in spermatogonial cells. Dosage dependence: The frequency of non-mosaic The @-ethylguanine has been suggested to be the rel- mutants generated by ENU was dependent on the dos- evant adduct which eventually should lead to GC -+ AT age. When the concentration of ENU was varied from transitions (LOVELESS1969). Indeed, themajority of mu- 2.0 to 3.0 mM, we observed a significant increase in the tations induced by ENU in E. coli were demonstrated to mutation frequency. However, there was no further in- be GC + AT transitions, and less frequently AT - CG crease in frequencies for the3.5 mM ENU regimen. We transversions (RICHARDSON et al. 1987). In the mouse, were interested in the most effective treatments, and however, most of the ENU induced mutations analyzed thus have not extended ourstudies to lower concentra- at themolecular level havebeen shown to betransitions tions of ENU. Lowerconcentrations ofENU have been and transversions at AT sites, and less frequently GC -+ tested in medaka: a single 2-hr incubation of malesat 27" AT transitions (POPP et al. 1983; LEWISet al. 1985; with up to 1 mM ENU results in specific-locus mutation HARBACHet al. 1992). Thus, in mammalian cells DNA- rates in spermatogonial cells lower than those induced alkylation products other than @-ethylguanine may be by any treatments inthis paper (A. SHIMAand A. SHIMADA, also relevant for ENU mutagenesis. It is not known personal communication). whether the same DNA adducts are responsible for in- As observed with mice (HITOTSUMACHIet al. 1985), we duction of all types ofmutations brought aboutby ENU observed an approximately twofold increase in the fre- in different germ cell stages. quency of locus-specific mutations when the numberof Nature of induced mutations:Three observations in- treatments in a mutagenic regimen was increased from dicated that mutations induced in fish by ENU were two to four. However, onlya small increase in mutagenic probably either point mutations and/or small, intra- frequencies was observed for one of the two regimens in genic lesions. First, induction of mosaic mutations by which the ENU treatment was repeated six times. ENU is consistent with a proposed mode of mutagenic Which of the mutagenic regimens tested would be activity via alkylation of DNA, a process that generally optimal for production of F, founder animals? Besides producespoint mutations (SINGERand GRUNBERGER the dosage-dependence of mutation frequencies dis- 1983). Second,a large proportion of the new mutations cussedabove, for an screen the survivalof mu- induced at the a16 locus were hypomorphic-consistent F2 tagenized males is veryimportant, since they are used for with an intragenic type of lesion. Finally, none of the the generation of F, founders in the period between 3 four tested mutations induced by ENU at the alb locus weeks and 3 months after mutagenic treatments. No- was linked to lethal mutations, supporting the notion ticeably, the X 3.5 mM regimen was very toxic during that the mutationswere limited to single genes (datanot 4 shown). Inother systems, as discussed above, point mu- mutagenic treatments. In both 6 X 3 mM treatments, 50% of the animals that survived the treatments died tations have been demonstrated to constitute the ma- during the first 2 months after the completion of mu- jority of lesions produced by ENU in vivo (RICHARDSON tagenesis. For these reasons, we have chosen the 1-hr et al. 1987; POPPet al. 1983; and see below). Addition- treatment with 3 mM ENU repeated four times as our ally, it has been suggested that the germ cell stage of induction is the major determinant of the nature ofin- routine mutagenesis regimen. However, a measurement duced mutations. Large lesions constitute only a small of frequencies of embryonic lethal mutations in F2 lines proportion of mutations induced by a variety of chemi- produced from males after various mutagenic regimens cals in stem cell or differentiating spermatogonia is needed to decidewhich of them is best for saturation (RUSSELLet al. 1990). mutagenesis screens. Fertility of mutagenized males: Male mice treated Our studies indicated that 3.5 mM is the maximal fea- with ENUgo througha transient periodof infertility, the sible concentration of ENU for 1-hr treatment at 21". length of which positively correlates with the dose of a There are othervariables of mutagenesis that have not treatment, and inversely withthe lengthof intervals be- been explored in this study. For example higher tem- tween treatments (HITOTSUMACHIet al. 1985). This tran- perature or higher ENU concentrations with shorter sient period of sterility observed for mouse males is ex- treatments may be tested for increased frequency of in- plained by ENU-dependent spermatogonial killing duced locus-specific and embryonic lethal mutations. (OAKBERG,as cited in RUSSELLet al. 1979). In contrast, Variable mutabilityof the tester lockAmong the four mutagenized zebrafish males have not exhibited even a pigmentation loci used for specific-locus tests,mutation short period of complete infertility following mutagen- rates in the spa gene were more than two-fold higher esis. We have observed, however, a period of severalfold compared to alb and go1 loci, and more than 10-fold G ermline Mutagenesis Germline in Zebrafkh 1417 higher than in the least mutable bra locus. Similar dif- preexisting pigmentation mutation (DOVE1987). In the ferences of mutability of the tester loci have been ob latter case both therelative mutabilities of single loci, as served in mice (HITOTSUMACHIet al. 1985; RUSSELLet al. well as frequencies of loci with different mutabilites in 1990). Thelow number of mutations isolated at the bra a class ofgenes are important.However, neither ofthese locus might be assigned to two factors. First, the bra lo- two variables between embryonic lethal and pigmenta- cus might be a cold spot forENU mutagenesis, e.g., due tion loci in vertebrates have been rigorously compared. to a small locus size. Second, embryos carrying hypo- Therefore, our extrapolation of mutation frequencies morphic bra alleles might have been scored as wild type from asmall number of pigmentation loci to embryonic at day 3 of embryogenesis. lethal genes could lead to under oroverestimation of the Advantages of mutagenizingspermatogonial stem number of embryonic lethal genes in zebrafish. Third, cells: These have been well recognized and proven in hypomorphic mutations in embryonic lethal genes the mouse (RUSSELLet al. 1979; PETERS 1985;RINCHIK should be detected as efficiently as hypomorphic mu- 1991). Herewe demonstrated that they may also apply tations in the tester pigmentation loci. This might not to zebrafish. A group of males treated with ENU can necessarily be the case, as hypomorphic mutations in continue to produce mutagenized sperm derived from some genes might bemanifested in a range of variable mutagenized spermatogonia until the males cease mat- phenotypes. Such hypomorphic embryonic lethal mu- ing, or die. Because female fish produce large numbers tations could be excluded in our screen due to thestrict of eggs, we were able to generate up to a thousand F, definition of the embryonic lethal phenotype. This strict founder animals from a single mutagenized male. Thus, definition is nevertheless necessary as the nonspecific one group of males subjected to ENU mutagenesis can lethal effects can be exhibited by up to 20% progeny of produce several thousand F, lines. This minimizes the a screen cross, and often of the rescreen cross. Our handling of ENU. fourth assumption was that the number of embryonic The screening of F, lines for embryonic lethal muta- lethal genes per gamete is linearly dependent on the tions has revealed the main advantage of spermatogo- average rate of the specific locus mutations. We have not nial us. sperm mutagenesis: on average a quarter of F, observed, however, the expectedtwo-fold increase in the sibling crosses manifested a given mutation harboredby number of embryonic lethal mutations in F, X F, lines this line. This is very important forseveral reasons. First, us. F, X wild-type lines. The basis for this phenomenon only small numbers of crossesare neededto recover the is not clear. One interpretation is that while the number majority of mutations carried in each line. Second, re- of locus-specific mutations per genome increases lin- covery of a mutation more than once during screening early with the ENU dosage, the number of embryonic of a single line provides rapid confirmation of a mutant lethal mutations permitting asuccessful generation and phenotype. Third, several carriers can be identified at screening of an F, family reaches a plateau. This could once, which enables one to work immediately on the lead to an underestimation of the numberof embryonic mutation, and facilitates its preservation via sperm freez- lethal genes. ing. Finally, the average number of mutations per ge- The estimated number of 1,500 of embryonic lethal nome is probably lower for non-mosaic than mosaic F, genes is threefold smaller than 5,000 of genes essential founders for screens detecting similar numbers of mu- for developmentand viability ofthe zebrafish proposed tants. This makes further geneticanalysis ofmutants de- by GRUNWALDand STREISINCER(1992) based on ENU mu- rived from non-mosaic F, less difficult. tagenesis of sperm cells. This discrepancy results prob An estimate of the number of genes that can mutate ably from different operationaldefinitions of embryonic toa recessive embryonic lethal phenotype: The fre- lethal mutations. The number of essential genes in the quency ofspecific-locus mutations (1.I X 90% con- mouse has been estimated at 5,000-10,000 by CARTER fidence limits: 0.69-1.70 X an average rate of mini- (1957).A similar number was reached after extrapolat- mal independent mutational events for the ENU4 and ing results of saturation mutagenesis of the T/t-H-2 ENU5 sets of mutagenized males used to generate F, region with recessive pre-natal lethal ENU mutations lines, Table 5) was compared with that of embryonic- (SHEDLOVSKY et al. 1986). InD. melanogaster the number lethal mutations per mutagenized genome (1.70; 90% of genes that are requiredzygotically for embryonic de- confidence interval: 1.47-1.93). This yielded an initial velopment has been estimated at 1,500 (NUSSLEIN- estimate of the numberof genes thatcan mutate to em- VOLHARD et al. 1984;JURGENS et al. 1984; WIESCHAUSet al. bryonic lethality in the zebrafish to be around 1,500 1984).A comparison between the fish, fly and the mouse (800-2,800). This estimate was based on four assump- estimates is difficult as we do notknow whether thephe- tions. First, mutations in embryonic lethal genes should notypic class of embryonic lethal genes in the fish or fly be transmitted to subsequent generations with an effi- precisely corresponds to the genes essential for mouse ciency equal to those in tester genes. Second, the aver- development. age target for generating an embryonic lethal allele Toward saturation of zebrafish genome with zygotic should be equal to that for producing an allele for the embryonic lethal mutations: The results reported here 1418 L. Solnica-Krezel, A. F. Schier and W. Driever support the notion thatzebrafish fulfills all the criteria ment obtained from currentembryological and molec- required to perform saturationmutagenesis screens for ular genetic approaches. embryonic lethal mutations (STREISINGERet al. 1981; KIMMEL 1989). First, ENU can induce in the zebrafish germline high rates of mutations affecting predomi- The authorswould liketo thank MONTEWESTERFIELD, in whose labo- ratory at the University of Oregon, Eugene, one of us (W.D.), had nantly single genes. Second, mutations are efficiently started tests of the mutagenic effects of EMS. We are grateful to MARY transmitted to progeny of mutagenized males and seg- MULLINS,MATTHIAS HAMMERSCHMIDT and CHRISTIANENUSSLEIN-VOLHARO regate at Mendelian ratios in the progeny of F, sibling (Tiibingen, Germany) for communicating ideas on EMS mutagenesis crosses. Screening is fast and even subtle defects in or- to us and for discussions. Specialthanks go to AKIHIRO SHIM(Tokyo, ganogenesis can be detected using a dissecting micro- Japan) for discussions and communication of unpublished data on ENU mutagenesis in the medaka. CHARLINEWALKER (Eugene) was help scope. ful withinformation on fish, the mutagenesisprotocols of the zebrafish How many lines would have to be screened to detect laboratory at the University of Oregon, and fish strains (AB wild-type 87% of all genes that can be mutated to a zygotic em- lines and pigment marker stocks). Brass fish were obtained from C. bryonic lethal phenotype?Our initial results allow us to NUSSLEIN-VOLHARD.In Boston, the support of WFISHMANand DIUIER make a rough approximation,based on two assumptions STMNIERduring the establishment of the Fish Core-Facilityat theCar- about the mutability and transmission of embryonic le- diovascularResearch Center was invaluable. IAN CAPERTON (from Aquatech., Brockton, Massachusetts) and THOMASLAWTIENBACH (from thal mutations discussed above. If the distribution of Marine Biotech., Beverly, Massachusetts) helped to plan and build the mutations in the genome can be approximated by the fish facility. Special thanks go to ZEHAVARANGINI and DIDIERSTAINIER Poisson distribution, about 87% of genes would be de- for help in the EMS screen, and to STEPHANNEUHAUSS for help in tected by a screen that obtained on average two muta- screening ENU lines. This study relies to a large degree also on the tions per locus. At an average locus-specific mutation countless efforts andthe excellent technical support by COLLEEN Boccs, LISAVOCELSANG and XIAORONGJI. L.S.K. would like to thank X rate per genomeof 1.27 lo-’ (Table 5, an average for WILLIAMF. DOVE (University of Wisconsin, Madison)and members of the 4 X 3 mM, 1-hr regimens, now used routinely for the the DOVEgroup for encouragement and discussions on what fish can production of F, founders), 1,600 genomes would have learn from mice. Special thanks go to WILLIAMF. DOVE and TIMBUR- to be screened. Since genes are not equally mutable, LAND,and members of our group: DEREK STEMPLE,ZEHAVA RANGINI and however, distribution of mutations in some genes devi- STEPHANNEUHAUSS for constructive criticismof the manuscript. A.F.S. ates from that predictedby Poisson. Forinstance a satu- is a European Molecular Biology Organization fellow. This work was supported by National Institutes of Health grant HDP9761 and by ration screen in Drosophila has revealed that at an av- Bristol-Myers Squibb. erage recovery ofabout fouralleles per locus about 25% of allidentified genes were represented by single alleles (WIESCHAUSet al. 1984) (also see LEFEVREand WATKINS LITERATURE CITED 1986). Consequently, in 1,600 genomes screened we ASHBURNER,M., 1989 Drosophila: A LaboratoryHandbook. Cold would recover more than 95% of loci that are as highly Spring Harbor Laboratory, Cold Spring Harbor, N.Y. CARTER,T. C., 1957 Recessive lethal mutation induced in the mouse mutable as spa (locus-specificmutation rateper genome by chronic y-irradiation. Proc. R. SOC.B 147: 402-411. 2.1 X lo-’) but less than 50% of the loci that have aslow CHAKRABARTI,S., G. STREISINGER, F. SINGER and C. WALKER,1983 Fre- a mutability as bra (locus-specific mutation rate of less quency of yray induced specific locus and recessive lethal mu- tations in mature germ cells of the zebrafish, Bruchydanio rerio. than 0.2-0.4 X lo-’). To isolate mutations in87% of the Genetics 103: 109-123. loci that are as mutable as bra at least 5,000-10,000 ge- CROW,E. L., and R. S. GARDNER, 1959Confidence intervals for the nomes would have to be screened. Currently our labo- expectation of a Poisson variable. Biometrica46 441-453. ratory is screening 160 F, lines per month, making the GULP,P., C.NUSSLEIN-VOLHARD and N. HOPKINS,1991 High-frequency germ-line transmission of plasmid DNA sequences injected into objective of detecting the majority of mutable embry- fertilized zebrafish eggs. Proc. Natl. Acad. USA88 Sci. 7953-7957. onic lethal genes reachable within 2 years. DOVE,W. F., 1987 Molecular genetics of Mus musculus: point mu- Our main goal is the identification of genes that de- tagenesis and millimorgans. Genetics 116 5-8. EGAMI, N.,and Y. HYODO-TAGWCHI, 1967An autoradiographic exami- termine the vertebrate body plan during embryogen- nation of rate of spermatogenesisat different temperatures in the esis. An initial assessment of mutant phenotypes indi- fish, Otyrias latipes. Exp. Cell Res. 47: 665-667. cated that mutations induced by ENU can affect both EHLING,U. H., R. B. CUMMINGand H. V. MALLING, 1968 Induction of early embryogenetic processes: formation of the noto- dominant lethal mutations by alkylating agents in male mice.MU- tat. Res. 5: 417-428. chord, patterning of the neural tube, as well as specific FAHMY,0. G., and M. J. FAHMY,1957 Mutagenic response to the alkyl- aspects of later development and organogenesis: for- methanesulphonates during spermatogenesisin Drosophila mela- mation of the ear, jaw and heart. The availability of nogaster. Nature 180: 31-34. FAVOR,J., A. NEUHAUSER-KLAUSand U. H. EHLING,1990a The fre- mutations in several genes affecting the same process quency of dominant cataract and recessive specific-locusmutation will enable one to use the full power of genetic analy- mosaics in F, mice derived from post-spermatogonial treatment sis in vertebrates and to establish hierarchies in which with ethylnitrosourea. Mutat. Res. 229: 105-114. FAVOR,F., M. SUND,A. NEWHAUSER-KLAUSand U. H. EHLING,1990b A these genes work. The analysis of mutations affecting dose-response analysisof ethylnitrosourea-induced recessive fish embryogenesis will complement and enrichthe specific-locus mutations in treated spermatogonia of the mouse. knowledge about mechanisms of vertebrate develop- Mutat. Res. 231: 47-54. Germ line Mutagenesis in Mutagenesis Germline Zebrafish 1419

FELSENFELD,A. L., C. WALKER,M. WESTEWIELD,C. KIMMEL and G. STRE- affecting pattern of larval cuticle in Drosophila melanogaster. I. ISINGER,1990 Mutations affecting skeletal muscle myofibril struc- Zygotic loci on the second chromosome.Roux’s Arch. Dev. Biol. ture in the zebrafish. Development 108 443-459. 193: 267-282. GENOROSO,W. M., 1984 pp. 347-356 in Genetics: New Frontiers, ed- ONDR~J,M., 1971 The proportion of complete mutations, mosaics ited by W. L. CHAPRAand B. C. JOSHI. Oxford/IBH, New Delhi. and instabilities induced by ethylnitrosourea in Drosophila mela- GOLIN,J., D. J. GRUNWALD,F. SINGER, C. WALKERand G. STREISINGER, nogaster. Mutat. Res. 12: 159-169. 1982 Ethylnitrosourea induced germline mutations inzebrafish PETERS,J., 1985 Ethylnitrosourea as a mouse mutagen. Trends Gen. Brachydanio rerio. Genetics 100: S27. 1: 5-6. GRWNWALD,D.J., and G. STREISINGER,1991 Induction of mutations in POPP,R. A,, E. G. BAILIFF, L.C. SKOW, F.M. JOHNSON and S. E. LEWIS, the zebrafish with ultraviolet light. Genet. Res. 59: 93-101. 1983 Analysis of a mouse olglobin gene mutation induced by GRWNWALD,D. J., and G. STREISINGER,1992 Induction of recessive le- ethylnitrosourea. Genetics 105: 157-167. thal and specific locus mutations in the zebrafish with ethylni- RICHARDSON,K. K, F. C. RICHARDSON,R. M. CROSBY,J. A. SWENBERGand trosourea. Genet. Res. 59 103-116. T. R. SKOPEK,1987 DNA base changes and alkylation following HALDANE, J. B. S., 1956 The detection of autosomal lethals in mice in vivoexposure ofEscherichia coli to N-methyl-N-nitrosourea or induced by mutagenic agents. J. Genet. 54: 327-342. N-ethyl-N-nitrosourea. Proc. Natl. Acad. Sci. USA 84: 344-348. HARBACH,P. R., A. L. FILIPUNAS,Y. WANG and C. S. AARON,1992 DNA RINCHIK,E. M., 1991 Chemical mutagenesis and fine-structure func- sequence analysis of spontaneousand N-ethyl-N-nitrosourea- tional analysis of the mouse genome. Trends Genet. 7: 15-21. induced hart mutations arising in vivo in cynomolgus monkey RUSSELL, B., L. P. B. SELBY, E. VONWLE, W. SHERIDAN and L. VALCOVIC, T-lymphocytes. Environ. Mol. Mutagen. 20 96-105. 1981 The mouse specific-locus test with agents other than ra- HATTA,K., C. B. KIMMEL,R. K. Ho and C. WALKER,1991 The cyclops diations. Interpretation of data and recommendations for future mutation blocks specification of the floor plate of the zebrafkh work. Mutat. Res. 86: 329-354. central nervous system. Nature 350: 339-341. RUSSELL, L.B., J. W. BANGHAM,K. F. STELZNERand P. R. HUNSICKER HIRSH,D., and R. VANDERSLICE,1976 Temperature-sensitive develop 1988 High frequency of mosaic mutants produced by N-ethyl- mentalmutants of Caenorhabditiselegans. Dev.Biol. 49: N-nitrosourea exposure of mouse zygotes. Proc. Natl. Acad. Sci. 220-235. USA 85: 9167-9170. HITOTSWMACHI,S., D.A. CARPENTER and W. L.RUSSELL, 1985 Dose- RUSSELL,L. B., P. R. HUNSICKER,N. L. A. CASHIER, J. W. BANGHAM,W. L. repetition increases themutagenic effectiveness of N-ethyl-N- RUSSELLet al., 1989 Chlorambucil effectively induces deletion nitrosourea in mouse spermatogonia. Proc. Natl. Acad. Sci. USA mutations in mouse germ cells. Proc. Natl. Acad. Sci. 86: 82: 6619-6621. 3704-3708. JOHNSON, F.M., and S. E. LEWIS,1981 Electrophoretically detected RUSSELL,L. B., W. L. RUSSELL,E. M. RINCHIK and P. R. HWNSICKER, germinal mutations induced in the mouse by ethylnitrosourea. 1990 Factors affecting the nature of induced mutations. Ban- Proc. Natl. Acad. Sci. USA 78: 3138-3141. bury Rep. 34: 271-289. JURGENS, G., E. WIESCHAUS,C. NUSSLEIN-VOLHARDand H. KLUDING, RUSSELL,W. L. X-ray induced mutations in mice. Cold Spring 1984 Mutations affecting the pattern of the larval cuticle in Dro- 1951 Harbor Sym. Quant. Biol. 16: 327-336. sophila melanogaster. 11. Zyogtic loci on the third chromosome. RUSSELL,W. L., and P. R. HUNSICKER,1984 Mutageniceffect of Roux’s Arch. Dev. Biol. 193: 283-295. ethylnitrosourea (ENU) on post-stem cell stages in male mice. JURGENS,G., U. MAYER, R. A. TORRESRUIZ, T. BERLETHand S. MISERA, 1991 Genetic analysis of pattern formation in the Arabidopsis Environ. Mutagen. 6: 390. embryo. Development Suppl. 1: 27-38. RUSSELL,W. L., E. M. KELLY,P. R. HUNSICKER,J. W. BANGHAM,S. C. MADDUX et al., 1979 Specific-locus test shows ethylnitrosourea to KIMMEL, C. B., 1989 Genetics and early development of zebrafkh. Trends Genet. 5: 283-288. be the most potent mutagen in themouse. Proc. Natl. Acad. Sci. KIMMEL,C. B., D. A. KANE, C. WALKER, M.R. WARGAand M. B. ROTHMAN, USA 76 5818-5819. 1989 A mutation that changescell movement and cell fate in the SCHONEFELD,U., W. ALFERMANN and J. E. SCHULTZ,1986 Economic zebrafkh embryo. Nature 337: 358-362. Mass Cultivation of Paramecia tetraurelia on a 200 Liter Scale. LEFEVRE,G., and W.WATKINS, 1986 The question of the total gene J. Protozool. 33: 222-225 number in Drosophila melanogaster. Genetics 113: 869-895. SEGA,G. A. 1974 Unscheduled DNA synthesis in the germ cells of LEWIS,S. E., F. M. JOHNSON,L. B. BARNETT, L. C.SKOW and R. A. POPP, male mice exposed in vivo to the chemical mutagen ethyl meth- 1984 Mosaic mutants may be induced in postgonial germ cell ane sulfonate. Proc. Natl. Acad. Sci. USA 71: 4955-4959. stages of the mouse. Genetics 107 (Suppl.): s62. SHEDLOVSKY,A., J.-L. GUENET,L. L. JOHNSON and W. F. DOVE,1986 In- LEWIS,S. E., F. M. JOHNSON,L. C. SKOW, D. POPP,L. B. BARNETTand duction of recessive lethal mutations in theT/t-H-2 region of the R. A. POPP,1985 A mutation in a Pglobin gene detected in the mouse genome by a point mutagen. Genet. Res. 47: 135-142. progeny of a female mouse treated with ethylnitrosourea. Proc. SHEDLOVSKY,T. A., KING and W. F. DOVE, 1988 Saturation germ line Natl. Acad. Sci. 82: 58-29. mutagenesis of the murinet region including a lethalallele at the LEWIS,S. E., L. B. BARNETTand R. A. POPP,1990 Mosaic mutants in- quaking locus. Proc. Natl. Acad. Sci. USA 85: 180-184. duced by ethylnitrosourea in late germ cells of mice. Banbury SHIMA,A.,and A. SHIMADA,1991 Development ofapossible nonmam- Rep. 34: 237-245. malian test system for radiation-induced germcell mutagenesis LOVELESS,A., 1969 Possible relevance of Walkylation of deox- using a fish, the Japanese medaka ( Oryrias latipes). Proc. Natl. yguanosine to mutagenicity and carcinogenicity of nitrosamines Acad. Sci. USA 88: 2545-2549. and nitrosamides. Nature 223: 206-207. SHIMADA,A., and N. EGAMI,1984 Dominant lethal mutations induced MAYER,U., R. A. TORRESRUIZ, T. BERLETH,S. MISERAand G. JURGENS, by MMS and mitomycin C in the fish Oryzias latipes. Mutat. Res. 1991 Mutations affecting body organization in the Arabidopsis 125: 221-227. embryo. Nature 353: 402-407. SINGER,B., and D. GRUNBERGER,1983 Molecular Biology of Mutagens MCDONALD,J. D., V. C. BODE, W.F. DOVEand A. SHEDLOVSKY, and Carcinogens. Plenum Press, New York. 1990 P~h*p”~:A mouse mutant deficient in phenylalanine hy- STREISINGER,G., C. WALKER,N. DOWER,D. JSNAUBER and F. SINGER, droxylase. Proc. Natl. Acad. Sci. USA 87: 1965-1967. 1981 Production of clones of homozygous diploid zebra fish MOSER,A. R., H. C. PITOTand W. F. DOVE,1990 A dominant mutation (Brachydanio rerio). Nature 291: 293-296. that predisposes to multiple intestinal neoplasia in the mouse. STREISINGER,G., F. SINGER,C. WALKER,D. KNAUBER and N. DOWER, Science 247: 322-324. 1986 Segregation analyses and genecentromeredistances in ze- MULLINS,M., 1993 Conventions for namingzebrafkh genes, pp. 7.1- brafish. Genetics 112: 311-319. 7.4 in The Zebrafish Book,edited by M. WESTERFIELD.University of VANZEEIAND, A. A., C. R. MOHN,A. NEUHAWSER-KLAUSand U. H. EHLING, Oregon Press, Eugene. 1985 Quantitative comparison of genetic effects of ethylating NOSSLEIN-VOLHARD,C., and E. WIESCHAUS,1980 Mutationsaffecting agents onthe basis of DNA adduct formation. Use of 06- segmentnumber and polarity in Drosophila. Nature 287: ethylguanine as moleculardosimeter forextrapolation from cells 795-801. in culture to the mouse. Environ. Health Perspect. 62 163-169. NUSSLEIN-VOLHARD,C., E. WIESCHAWSand H.KLUDING, 1984 Mutations VAN ZEELAND,A. A., A. DEGROOT and A. NEWHAUSER-KLAUS,1990 DNA 1420 L. Solnica-Krezel, A. F. Schier and W. Driever

adduct formation in mouse testis by ethylating agents: a com- finding andsynapse formation in a zebrafish mutant lacking func- parison with germ-cell mutagenesis. Mutat. Res. 231: 55-62. tional acetylcholine receptors. Neuron 4: 867-874. VOGEL,E. W., W. G. H. BLIJLEVEN,M. J. H. KORTSELIUSand J. A. ZIJSTRA, WIESCHAUS,E., C.NUSSLEIN-VOLHARD and G. JI~RGENS,1984 Mutations 1982 A search for some common characteristics of the effects of affecting the pasttern of the larval cuticle in Drosophila melano- chemical mutagens in Drosophila. Mutat. Res. 92 69-87. gaster. 111. Zygotic loci on the X-chromosome and fourth chro- WALKER,C., and G. STREISINGER,1983 Induction of mutations by -,-rays mosome. Roux’s Arch Dev. Biol. 193: 296-307. in pregonial germ cells of zebrafish embryos. Genetics 103: WOOD, E. B., 1988 TheNematode Caenorhabditis elegans. Cold 125-136. Spring Harbor Laboratory, Cold Spring Harbor, N.Y. WESTERFIELD,M. (Editor), 1993 TheZebrafish Book. University of Oregon Press, Eugene. WESTERFIELD,M., D. W. LIU,C. B. KIMMEL and C. WALKER,1990 Path- Communicating editor: T. SCH~PBACH