Development 115, 421-426 (1992)Printed in Great Britain © The Company of Biologists Limited 1992

421

A transgene containing lacZ is expressed in primary sensory neurons in

zebrafish

THOMAS A. BAYER and JOSE A. CAMPOS-ORTEGA

Institut fUr Entwicklungsbiologie, Universitht zu Ktiln, Gyrhofstr. 17, 5000 K6ln 41, Federal Republic of Germany

Summary

In order to screen for developmentally active chromo-somal domains during zebrafish embryogenesis, wegenerated transgenic fish by microinjecting two differentlacZ reporter constructs into fertilized eggs. Transgenicfish were screened among the progeny of injected fish(Fo) crossed to non-injected fish. Groups of 15 to 20progeny of each cross were tested for lacZ expressionand/or transmission of injected sequences using PCRand Southern hybrizations. Progeny from 2 of 102 fishinjected with supercoiled constructs containing Ronssarcoma virus promoter sequences showed apparentlyspatially regulated /J-galactosidase (/J-Gal) activity.However, we were not able to detect this reporterconstruct in DNA from fins of Ft fish. Injections of alinear reporter construct containing mouse heat-shockpromoter sequences revealed transmission of injectedsequences to Fx progeny in about 6% of cases (8 of 129fish, tested with PCR). We found one /acZ-expressingline that showed a spatially and temporally restrictedexpression of lacZ and, therefore, features typical

characteristics of "enhancer trap" lines. In this line,lacZ expression starts at 16 hours post-fertilization intrigeminal ganglion cells. At about 24 hours lacZexpression can be detected in trigeminal ganglionneurons and Rohon-Beard neurons, indicating that thedevelopment of these two cell types shows commonfeatures. The reporter gene has integrated as a singlecopy. The founder fish was mosaic: 19% of its offspring(3 of 16 tested animals) carried the reporter construct intheir fins; about 51% (13 of 27 tested animals) of theprogeny of Ft fish were /J-Gal positive indicating fullhemizygosity. We traced the heritability up to the 4thgeneration and showed that the reporter construct isstably integrated and inherited in a Mendelian manner.These results demonstrate that it is possible to generate"enhancer trap" lines in zebrafish, albeit with lowefficiency.

Key words: zebrafish, enhancer trap, primary sensoryneurons.

Introduction

The zebrafish Brachydanio rerio is particularly usefulfor studying developmental processes in higher eucar-yotes, since the embryos develop outside the motherand are optically transparent, allowing direct obser-vation of their embryonic development. Screening forrecessive mutations is facilitated by methods for theproduction of haploid and gynogenetic offspring(Streisinger et al., 1981; Walker and Streisinger, 1983;Chakrabarti et al., 1983; Kimmel, 1989). Furthermore,production, maintenance and analysis of transgenic ormutant lines is aided by a relatively short generationtime of 2-3 months. Recently, it has been shown that itis possible to produce transgenic zebrafish via micro-injection of foreign DNA into the cytoplasm offertilized eggs (Stuart et al., 1988, 1990; Culp et al.,1991).

In some organisms, transgene expression may beinfluenced by endogenous elements near the position of

integration (Hazelrigg et al., 1984; Palmiter andBrinster, 1986). Gene constructs containing weakpromoters have been used in different transgenicorganisms to detect endogenous genomic regulatoryelements (O'Kane and Gehring, 1987; Allen et al.,1988; Kothary et al., 1988; Gossler et al., 1989). InDrosophila, P-element-mediated enhancer detection iswidely used to screen for developmentally regulatedgenes on a large scale, using protocols that permitdevelopmental, genetic and molecular analysis of thesegenes (Bier et al., 1989; Bellen et al., 1989; Wilson etal., 1989; Ghysen and O'Kane, 1989). The main goal ofour study was to determine whether this "enhancertrap" technique can also be applied to zebrafish. Weused two different constructs, with the Rous sarcomapromoter (pRSV-fiGal), and with a truncated mouseheat-shock promoter (phs6-lacZA). Following injec-tions of this latter construct in linear form, we isolatedone stable transgenic line of zebrafish expressing lacZwith enhancer trap features. The reporter gene lacZ is

422 T. A. Bayer and J. A. Campos-Ortega

expressed exclusively in primary sensory neurons,including Rohon-Beard neurons of the spinal cord andtrigeminal ganglion neurons in the head.

Materials and methods

Preparation of plasmid DNA for injectionThe fusion construct pRSV-fiGal (a gift of Manfred Schartl,Wiirzburg; Edlund et al., 1985) contains Rous sarcoma viruspromoter sequences linked to lacZ coding for /3-gal. The otherfusion construct used for our studies is phs6-lacZA (a gift ofAchim Gossler, Koln; personal communication). It contains atruncated mouse heat-shock promoter linked to lacZ. In bothconstructs, lacZ is linked to SV40 polyadenylation signalsequences which were extracted with phenol/chloroform priorto injection. In one set of injections, phs6-lacZA was digestedwith Pstl, which liberates the reporter as a 4.8 kb fragment.This DNA fragment was isolated and recovered by precipi-tation with an equal volume of isopropanol in the presence of0.2 M NaCl. Transgenic lines r200 and r215 were produced bymicroinjection of circular pRSV-fiGal; h32 was generated bymicroinjecting the linear 4.8 kb Pstl fragment oiphs6-lacZA.

MicroinjectionReporter construct DNA was injected into fertilized zebraflsheggs prior to first cleavage, essentially as described by Stuartet al. (1988), with the following modifications. Soon afterspawning, fertilized eggs were washed in embryo medium (amodified Hanks saline containing 1.3 mM CaC^, 1 mMMgCl2 and 4 mM NaHCO3) and placed in plastic Petri dishes(5 cm diameter). Injections were carried out under sightcontrol using a dissecting microscope and with the aid of amicromanipulator. In contrast to previous studies (Stuart etal., 1988,1990;Culpetal., 1991), we did not dechorionate thefertilized eggs prior to injection. Chorions were removedabout 24 hours later, at which time the zebrafish embryos arewell developed and crucial steps in embryogenesis have beencompleted. Micropipettes were made on a horizontal pullerusing 1 mm Hilgenberg glass capillary tubing with an innerfilament for backfilling the pipettes with DNA solution.Injection solutions contained 0.5% phenol red to allowestimation of the injected volume.

DNA extraction and blot analysisDNA manipulations were carried out essentially according toSambrook et al. (1989). DNA for PCR and Southern analysiswas extracted from whole fish at 1-3 weeks of age, from fins ofadult fish or from whole adult fish by dissolving the tissue inextraction buffer (10 mM Tris-HCl pH 8.0, 0.1 M EDTApH8.0 and 0.5% SDS). The samples were digested with 100 /ig/mlproteinase K overnight at 50°C. After removing undissolvedmaterial by centrifugation, DNA samples were subsequentlyextracted in phenol/chloroform/ethyl ether and precipitatedwith isopropanol as described above for plasmid DNA. DNAsamples were resuspended in TE (10 mM Tris pH 7.5, 0.25mM EDTA). For Southern analysis, DNA samples weredigested with restriction enzymes as specified by the manufac-turer, followed by gel electrophoresis and capillary transferonto nitrocellulose filters (Sambrook et al., 1989). Blots wereprehybridized overnight at 42°C with a solution containing 5x SSPE, 5 x Denhardt's, 0.2% SDS, 50% formamide and 0.1mg/ml sheared calf thymus DNA. The blots were thenhybridized overnight at 42°C in this solution containingradioactive probes for the plasmids pRSV-^Gal or phs6-lacZA, labelled to a specific activity of 108 cts/min/^g by

random priming (Feinberg and Vogelstein, 1983). The blotwas then washed twice in 2 x SSPE, 0.2 SDS (5 minutes),twice in 0.5 x SSPE, 0.2 SDS (15 minutes) and twice 0.1 xSSPE, 0.2 SDS (15 minutes) at RT. Exposure to film wascarried out at — 80°C with an enhancing screen.

Screening of transgenic fish by the polymerase chainreaction (PCR)DNA from either 15-20 fish 1-3 weeks old or from fins ofsingle adults was used. Approximately 1 fig template DNAwas used for PCR following a modification of the protocol ofSambrook et al. (1989): 67 mM Tris-HCl (pH 8.8); 16 mM(NH4)2SO4; 6.7 mM MgCl2; 0.67 /*M EDTA; 1 mM 2-mercaptoethanol; 200 /JM dNTP; primers 1 ^M; Taq DNApolymerase 2-5 units. Sense primer was derived from thepromoter sequence of the injected construct, torphs6-lacZA:5'CCCGATCCTCGGCCAGGACC 3'; for pRSV-^Gal:5'CCATGGCCACCCACTTCTGGTC3\ The anti-senseprimer was derived from the lacZ sequence: 5'CGCGTAAAAATGCGCTCAGG 3'. These primers gener-ated a PCR product of 614 bp for phs6-lacZA and about 1.1kb for pRSV-fiGal. The reaction was carried out as follows:melting temperature 95°C 3 cycles for 2 minutes, 35 cycles for10 seconds, annealing temperature 62°C 3 cycles for 1 minute,35 cycles for 1 minute and primer extension 72°C 3 cycles for 2minutes, 35 cycles for 1 minute, and a 15 minutes final step at72°C elongation. Amplified fragments were verified byhybridization analysis.

Detection of ft-Gal activityZebrafish embryos at various stages were fixed in 12.5%glutaraldehyde for 10 minutes (in 0.1 M PBS, pH 7.2) andwashed several times in PBT (0.3% Triton-X in 0.1 M PBS,pH 7.2). f3-Gal activity was made visible by incubation in X-Gal staining solution: [0.2% 5-bromo-4-chloro-3-indolyl-/J-D-galactoside (Biomol) in dimethylformamide; 5 mM K3 (Fe HI(CN)6); 5 mM K4 (Fe H (CN)6) in 0.1 M citrate buffer pH 8.0]overnight at 37°C. Stained embryos were dehydrated inincreasing concentrations of ethanol, incubated for 5 minutesin xylene and mounted in Durcupan (Fluka).

Double-labelingDouble-labeling of embryos from the line h32 was carried outusing X-Gal staining followed by incubation with zn-12 (amonoclonal antibody that recognizes the L2/HNK-1 epitope,a gift of Charles Kimmel, Eugene, Oregon). Fixation andstaining for X-Gal was as above, followed by several 5 minutewashes in 0.1 M PBS pH 7.2 and then distilled water, all atroom temperature. The staining protocol for antibodylabeling was performed essentially as described (Westerfield,1989). The embryos were permeabilized with acetone for 7minutes at —20°C, followed by 5 minute washes in distilledwater and PBS. The samples were then incubated in 1% BSA,1% DMSO in PBS for 5, 15, 30 and then 60 minutes. Theembryos were next treated overnight at 4°C with thesupernatant containing the monoclonal zn-12 antibody. Afterwashing the samples again 4 times in 1% BSA, 1% DMSO inPBS (5, 15, 30 and 60 minutes), the secondary antibody(HRP-coupled goat anti-mouse) diluted 1:500 with theBSA/DMSO/PBS solution was applied for 1 hour at 37°C. Theembryos were then washed 4 times (5, 15, 30 and 60 minutes)in PBS. Immunoreactivity was made visible with a 1 mg/mlsolution of diamino benzidine (Sigma) in PBS pH 7.2 to which1 jA of a 3% H2O2 solution was added. Stained embryos weredehydrated and mounted as above.

Enhancer trap in zebrafish 423

HistologyEmbryos to be sectioned for histology were treated as above.Prior to mounting in Durcupan, the samples were incubatedin a mixture of xylene and Durcupan (1:1) overnight at 4°C.After evaporation of xylene at room temperature, theembryos were embedded in fresh Durcupan, then incubatedfor 24 hours at 45°C and then for 30 hours at 60°C. 2 to 3 /ansections were cut with a microtome (Reichert), followed bycounter staining with methylene blue-borax and mounted inDePeX (Sigma).

Results

Transient expression of lacZ in early embryosWe injected several hundred fertilized zebrafish eggsprior to first cleavage with either pRSV-fiGal or phs6-lacZA. The recombinant lacZ genes in these constructswere expected to be functional in zebrafish due to thepresence of eucaryotic transcription initiation (RSV-LTR, heat-shock promoter from mouse) and termin-ation signals (SV-40 polyadenylation signal). Theresults of Stuart et al. (1990) had already indicated thatthe RSV-LTR functions in transgene zebrafish. Inmouse, it has been shown that the truncated mouseheat-shock promoter can be used as a minimalpromoter in enhancer trap experiments (Gossler et al.,1989). To verify the ability of these constructs toproduce functional /S-Gal, injected embryos were fixedand stained for /3-Gal activity about 24 hours afterinjection. Embryos injected with circular pRSV-^Galexhibited staining mainly in cells of mesodermal origin(data not shown; Overbeek et al., 1986; Swain et al.,1987). In contrast, the 4.8 kb Pstl fragment of phs6-lacZA gave rise to expression of lacZ in cells of most ifnot all tissues (data not shown).

Functional lacZ genes are inherited by the progeny ofinjected fishPresence or absence of foreign DNA in the germline ofinjected fish was determined by crossing injected fishwith non-injected fish and examining their progeny for^-Gal activity and/or by PCR or Southern hybridiz-ation. Since transgenic founder fish have been reportedto be mosaic (Stuart et al., 1988, 1990; Culp et al.,1991), groups of 15 to 20 progeny of each cross wereexamined. We performed three sets of injections: (1) Fxprogeny from 102 fish injected with pRSV-fiGal (assupercoiled construct, 30 to 100 ̂ g/ml, Table 1) wereexamined for lacZ expression by using the X-Galreaction. The results indicated that two of the injectedparental fish had /S-Gal activity in their germ-lines(Table 1). (2) V1 progeny from 129 fish were tested fortransmission of the 4.8 kb Pstl fragment of phs6-lacZA(40 to 80 /ig/ml, Table 2) using PCR. We detected theexpected PCR product in offspring of eight of thesefish. Progeny of one of these eight lines (b.32) exhibitedjS-Gal activity in identifiable cells. (3) Injections ofphs6-lacZA as a supercoiled fusion construct (50 /xg/ml)did not result in germ-line transmission (n=46), astested by PCR (Table 2).

Founder fish r200 and r215 injected with supercoiled

Table 1. Transgenic founder fish after injection ofsupercoiled pRSV-/3Gal DNA

Injected eggs

Number of survivorsafter 10 d

Number screened*(P)

Southern positive(Fi)

PCR positive(Fi)

X-GAL positive(Fi)

*: Total number of

I15 /ig/ml

199

69

-

—

—

—

P founder

Groups (concentration)

n30 /ig/ml

706

120

32

—

0(20)

1(32)

fish screened

in100 /ig/tnl

3465

208

97

1(27)

2(56)

1(70)

IV330/ig/ml

196

0

-

-

-

—

by various means.

Table 2. Transgenic animals after injections of phs6-lacZA DNA

Injected eggs

Number of survivorsafter 10 d

Number screened*(P)

PCR positive(Fi)

X-GAL positive(Fi)

•: Total number of P

Groups (concentration)

I(linear)

40 /ig/ml

907

145

96

6

0

founder fish

II(linear)

80/ig/ml

785

39

33

2

1

screened by

III(supercoiled)

50 /ig/ml

770

92

46

0

0

various means.

pRSV-fiGal DNA produced Fx progeny expressing lacZin a few cells of mesodermal origin (not shown).However, DNA was prepared from fins of 20 Fi r200and 15 Fx r215 siblings and found negative afterscreening by PCR. One line with /3-Gal activity wasestablished (h32) from fish injected with the 4.8 kb Pstlfragment of phs6-lacZA. The founder fish was indeedmosaic, since we detected the presence of the PCRproduct in the fins of three of a total of 16 offspringtested (19%; Table 3). Using the X-Gal reaction, theoffspring of 18 of 35 Fx fish (51%) were positive,whereas using PCR, 7 of 15 (47%) were positive,indicating that Fx fish were fully hemizygous. We wereable to demonstrate heritability of /J-Gal activitythrough the 4th generation (Table 3).

The reporter construct has integrated as a single copyin the genome of line h32Stable transmission through the germline over several

424 T. A. Bayer and J. A. Campos-Ortega

Table 3. Inheritance of lacZ expressivity in transgenicline h32

A 1 2 3 4 5 6 7

Generation

PF,F2»F2tF3

/S-Gal

—18+/3535+/4916+/2912+/25

(%)

-(51)(71)(55)(48)

PCR

3+/167+/15

-—-

(%)

(19)(47)

-—-

genotype

mosaichemizygoushemizygoushemizygoushemizygous

Generation: Generation of founder fish.•Hemizygous F2 fish were mated with their siblings.tHaploid embryos generated from hemizygous F2 female /3-Gal:

embryos were tested for lacZ expression.PCR: Offspring were grown and fin DNA was tested for

presence of the construct using PCR.

generations, as shown above, provides strong evidencefor genomic integration. Since extrachromosomalinheritance may also occur in other species (Stinchcombet al., 1985; Rassoulzadegan et al., 1986; Mello et al.,1991, see above), the presence of junction fragments ofreproducible size on Southern blots can provide furthersupport for this conclusion. As mentioned above, wefound no evidence for integration in offspring offounder fish r200, and r215, suggesting extrachromo-somal inheritance of the pRSV-f3Gal construct.Southern blot analysis of DNA from the line h32,however, showed that the 4.8 kb Pstl fragment hasintegrated as a single copy in the genome. Whereasdigestion with EcoRl, which has one cleavage site in theinsert, revealed two bands, digestion with BamHI,which has two sites in the insert, generated three bands.Digestion with Hindlll reveals only one hybridizingband at about 5.5 kb, indicating cleavage within specificflanking regions, since Hindlll does not cut within theinsert. The hypothesis of genomic integration is furthersupported by the observation that inherited foreignDNA migrated with high molecular weight (genomic)DNA, if left undigested (Fig. 1).

lacZ is expressed exclusively in primary sensoryneurons in line h32lacZ expression was found in a few cells of thetrigeminal ganglion and in individual neurons within thespinal cord (Fig. 2). Using double labeling with X-Galand the monoclonal antibody zn-12, we identified the/acZ-expressing cells as Rohon-Beard neurons in thespinal cord and trigeminal neurons in the head. Byanalyzing the /acZ-expression pattern at various stages,we detected the onset of lacZ expression at about 16hours post-fertilization in the trigeminal ganglion (datanot shown) and at about 24 hours in the spinal cord;expression in the trigeminal ganglion persists until 72hours and until 48 hours in the Rohon-Beard cells.Onset of lacZ expression in trigeminal cells coincideswith the onset of immunoreactivity to the monoclonalantibody zn-12 in these cells; developing axons begin togrow at about the same stage (Metcalfe et al., 1990).This temporal coincidence suggests that lacZ expressionmight be related to a neuronal cytodifferentiationprocess. lacZ expression is fully penetrant with respect

7.1-6.1-

RI RI RI

1 kb

Fig. 1. Southern analysis of line h32 DNA. (A) Lanes 1, 6and 7 contained about 10 ng samples of BamHI (lane 1),Hindlll (lane 6) and EcoKl (lane 7) digests. Lane 5contained about 3 /ig undigested DNA. Lanes 2, 3 and 4contained Pstl (lane 2, 500 ng) and Hindlll (lane 3, 500ng; lane 4, 50 ng) digests of phs6-lacZA as positivecontrols. The position of length standards (1 kb ladder) areshown on the left. (B) Diagram of the derived restrictionmap of the insert and flanking regions (see text).Abbreviations: B, BamHI; RI, EcoRI.

to the trigeminal ganglion cells, but not with respect tothe Rohon-Beard cells; that is to say, all embryos thatcarried the construct showed one or more lacZ-expressing trigeminal cells, but not all embryos showed/acZ-expressing Rohon-Beard cells. In addition to thisvariable penetrance, expressivity was found to vary aswell, for the number of /3-Gal positive cells differedbetween embryos, from one to about 20 /3-Gal-positivecells per trigeminal ganglion per embryo. We could notcorrelate /acZ-expressing Rohon-Beard neurons withpositions along the anteroposterior axis. Hence, it isunlikely that lacZ expression occurs in a distinct subsetof these cells. However, since Rohon-Beard neuronsare primary sensory cells mediating tactile sensitivity(Metcalfe et al., 1990), it might well be that the subsetof trigeminal neurons expressing lacZ are also specificfor tactile stimuli (Fig. 2).

B

Enhancer trap in zebrafish 425

Fig. 2. lacZ expression pattern in embryos of the line h32.The embryo in C is 30 hours old, all others are 24 hoursold. Photographs A and B were taken of the same embryo(for orientation, the arrow in both photographs points tothe same /3-Gal positive cell in the spinal cord). (A)Neurons show /3-Gal activity in both trigeminal ganglia; (B)several Rohon-Beard neurons show /3-Gal activity.Photographs C-G show embryos double-stained for lacZexpression with X-Gal and for expression of the L2/HNK-1carbohydrate with the zn-12 antibody. (C) Note /3-Galactivity in the trigeminal ganglion (arrow) and the networkof sensory axons stained with the zn-12 antibody. (D)shows lacZ expression in a neuron (overstained)neighbouring several zn-12 positive (Rohon-Beard) cells.The arrow in E points to the membrane, which is brownbecause of L2/HNK-1 expression, whereas the nucleus isstained in blue (lacZ expression). (F and G) show crosssections through the mesencephalon (F) and the spinalcord (G). The arrows point to /3-Gal positive (blue) cells.The membranes of the trigeminal ganglion cells (F) and ofthe Rohon-Beard cells (G) (brown) are labelled with thezn-12 antibody. Whole-mounts are oriented with anteriorto the left. Abbreviations: notochord (n), canalis centralis(cc). Magnification bar in A (same in B) is 50 /an; C is 30/an; D (same in G) is 9 /an; E (same in F) is 14 /an.

Discussion

We present evidence that the 4.8 kb Pstl fragment ofphs6-lacZA can integrate functionally into the zebrafishgenome. In addition, we find that the truncated mouseheat-shock promoter can be used as a minimalpromoter in zebrafish, which may come under theinfluence of genomic regulatory elements, conferringspatially and temporally restricted expression of lacZon a specific tissue. In the case reported in this paper,expression of lacZ is restricted exclusively to primarysensory neurons during the period of approximately 16to 72 hours embryonic development. This may be dueto cw-acting effects from nearby enhancers or othertissue-specific regulatory sequences. As in other species(O'Kane and Gehring, 1987; Allen et al., 1988; Gossleret al., 1989), lacZ expression in the zebrafish maytherefore serve as a valuable morphological marker tostudy patterns of gene expression at the cellular level.

In three previous studies, rates of germline trans-mission and stable integration in the zebrafish genome,following injection of foreign DNA, were found to varygreatly (Stuart et al., 1988,1990; Cu!p et al., 1991). Thisvariation may be due to a number of factors, such as theconditions of injection and/or the method used toscreen for germline positives. Injection of supercoiledplasmids was reported to result in a higher frequency ofgermline transformation (5% in Stuart et al., 1990; orup to 25% in Culp et al., 1991) than injection of linearconstructs [no transgenic was found by Culp et al.(1991) and one was found among 20 P fish by Stuart etal. (1988) with linear constructs]. Since our mainconcern was the production of /acZ-expressing trans-genics, rather than the production of transgenics assuch, our material was screened in various ways;consequently, our data cannot be directly compared to

previously published data. Since our main interest wasfunctional lacZ expression, we did not verify bySouthern blotting whether all our PCR-positive fishwere germline transformants.

Indeed, we believe that all available numbers, thosepreviously published as well as our own data, are toolow to permit any meaningful generalizations and thatthere are still many imponderables which are beyondour control. Founder fish r200 and r215 providedevidence compatible with transmission in an extra-chromosomal state (Table 2; see below). Hence, it maybe that some of the PCR-positives may still have carriedthe constructs as plasmids rather than integrated intheir genome. Alternatively, rearrangements or sup-pression of expression after passage through thegermline may also account for the failure to detect lacZexpression in the other seven PCR-positive lines.However, one fish of the eight PCR-positive founderfish injected with linear DNA (h32) carried the reporterconstruct as a single copy in its genome. This obser-vation is novel, since all other transgenic zebrafishstudied so far (Stuart et al., 1988; 1990; Culp et al.,1991) carried the foreign DNA in a rearranged state oras multiple copies; this has also been observed fortransgenic mice (Palmiter and Brinster, 1986).

Following injections of supercoiled pRSV-^Gal,Culpe et al. (1991) tested five of their 19 transgenic linesfor lacZ expression. Although all were transgenic fish(verified by Southern analysis), none showed /3-Galactivity in their offspring. Following injections of pRSV-PGal (also supercoiled), we observed lacZ expression inthe Fi offspring of two injected animals; however, westudied Fx lines established from these animals andcould not find evidence for stable germline transform-ation (see above). We cannot decide to what extent thefailure to detect stable lacZ expression is due to thepRSV-flGal construct. In C. elegans and mouse, it ispossible for injected sequences to be maintainedextrachromosomally, even when passed through thegermline (Stinchcomb et al., 1985; Rassoulzadegan etal., 1986; Mello et al., 1991). Therefore, we assumethat, in the lines r200 and r215, sequences were passedthrough the germline in an extrachromosomal state.

One of the main goals of our study was to applyenhancer trap techniques to the zebrafish. Transgenicanimals carrying a reporter gene can serve as a tool tosearch for developmentally regulated genes, for gener-ation of specific deletion mutations using the expressionof the reporter gene as a dominant marker, and forfurther molecular work on the corresponding genes.The line h32 examined in this study exhibits a preciselyregulated expression pattern in time and space.Double-labeling allowed us to identify the lacZ-expressing cells as primary sensory neurons. Althoughwe followed the inheritance through four generations,in which the pattern of /3-Gal activity is reproducible,lacZ expression is not fully penetrant. We have nocompelling explanation for this latter observation; avariety of possibilities, such as the position of theinsertion in the genome, may be considered. Neverthe-less, this technique can in principle be applied to the

426 T. A. Bayer and J. A. Campos-Ortega

zebrafish, as it has been applied in similar fashion toseveral other animal species (O'Kane and Gehring,1987; Allen et al., 1988; Gossler et al., 1989). However,the frequency of enhancer detection is apparently verylow in the zebrafish, at least with the constructs used inour study. We screened by various means the progenyof a total of 304 P animals injected with one of twodifferent constructs and found only one case in whichlacZ expression was transmitted stably over severalgenerations.

The pattern of expression in this case (line h32) isremarkable. Metcalfe et al. (1990) characterized withthe antibody zn-12 the temporal and spatial pattern ofexpression of the L2/HNK-1 tetrasaccharide in zebra-fish embryos, focusing in particular on the primarysensory neurons. Primary neurons are a distinct set oflarge neurons that arise early, rapidly develop longaxons and interconnect to form a simple neural networkthat mediates the early behaviour of the embryo(Grunwald et al., 1988; Kimmel and Westerfield, 1990).It has been suggested that the primary sensory neurons(Rohon-Beard neurons in the trunk and trigeminalneurons in the head) follow a common developmentalprogram. The developmental profile of the transgeneline h32 seems to support this observation, since lacZexpression can be found exclusively in these two celltypes. We detect /3-Gal activity at about 16 hours, whentrigeminal cells are being established. Considering thatlacZ transcription starts earlier, the endogenous zebra-fish gene could have an early function in the develop-ment of the primary sensory neurons.

We would like to thank Eva Varus for expert technicalassistance, Paul Hardy for constructive criticisms on themanuscript, Achim Gossler for advice and discussions, andBeate Schmitz, Christiane Bierkamp for discussions. Theresearch reported here was supported by a doctoral Fellow-ship of the Fritz-Thyssen Stiftung to T. A.B. and by grants ofthe Deutsche Forschungsgemeinschaft (DFG, SFB 243).

References

Allen, N. D., Cran, D. G., Barton, S. C , Hettle, S., Reik, W. andSuranl, M. A. (1988). Transgenes as probes for active chromosomaldomains in mouse development. Nature 333, 852-855.

Bellen, H. J., O'Kane, C. J., Wilson, C , Grossnlklans, U., Pearson,R. K. and G«hring, W. J. (1989). P element mediated enhancerdetection: a versatile method to study development in Drosophila.Genes Dev 3, 1288-1300.

Bier, E., Vaessln, H., Sheperd, S., Lee, K., McCall, K., Barbel, S.,Ackerman, L., Carretto, R., Uemura, T., Grell, E., Jan, L. Y. andJan, Y. N. (1989). Searching for pattern and mutation in theDrosophila genome with a P-lacZ vector. Genes Dev 3, 1273-1287.

Chakrabarti, S., Strelsinger, G., Singer, F. and Walker, C. (1983).Frequency of X-ray induced specific locus and recessivelethalmutations in mature germ cells of the zebrafish, Brachydanio rerio.Genetics 103, 109-123.

Culp, P., NQsslein-Volhard, C. and Hopkins, N. (1991). High-frequency germ-line transmission of plasmid DNA sequencesinjected into fertilized zebrafish eggs. Proc. Natl. Acad. Sci. USA88, 7953-7957.

Edlund, T., Walker, M. D., Barr, P. J. and Rutter, W. J. (1985). Cell-specific expression of the rat insulin gene: evidence for a role of twodistinct 5' flanking elements. Science 230, 912-916.

Feinberg, A. P. and Vogelstein, B. (1983). A technique forradiolabeling DNA restriction endonuclease fragments to highspecific activity. Anal. Biochem. 137, 266-267.

Ghysen, A. and O'Kane, C. (1989). Neural enhancer-like elements ascell specific markers in Drosophila. Development 105, 35-52.

Gossler, A., Joyner, A. L., Rossant, J. and Skarnes, W. C. (1989).Mouse embryonic stem cells and reporter constructs to detectdevelopmentally regulated genes. Science 244, 463-465.

Grunwald, D. J., Kimmel, C. B., Westerfield, M., Walker, C. andStreisinger, G. (1988). A neural degeneration mutation that sparesprimary neurons in the zebrafish. Dev. Biol. 126, 115-129.

Hazelrigg, T., Levis, R. and Rubin, G. M. (1984). Transformation of. white locus DNA in Drosophila: dosage compensation, zeste

interaction, and position effects. Cell 36, 469^481.Kimmel, C. B. (1989). Genetics and early development of zebrafish.

TIG 5, no.8, 283-288.Kimmel, C. B. and Westerfleld, M. (1990). Primary neurons of the

zebrafish. In Signals and Sense (ed. G.M. Edelman and- M.W.Cowan) New York: Wiley Interscience.

Kothary, R., Clapoff, S., Brown, A., Campbell, R., Peterson, A. andRossant, J. (1988). A transgene containing lacZ inserted into thedystonia locus is expressed in neural tube. Nature 335, 435-437.

Mello, C. C , Kramer, J. M., Stinchcomb, D. and Ambros, V. (1991).Efficient gene transfer in C. elegans: extrachromosomalmaintenance and integration of transforming sequences. EMBO J.10, no. 12, 3959-3970.

Metcalfe, W. K., Myers, P. Z., Trevarrow, B., Bass, M. B. andKimmel, C. B. (1990). Primary neurons that express the L2/HNK-1carbohydrate during early development of the zebrafish.Development 110, 491-504.

O'Kane, C. J . and Gehring, W. J. (1987). Detection of regulatoryelements in Drosophila. Proc. Natl. Acad. Sci. USA 84, 9123-9127.

Overbreek, P. A., Lai, S., Van QuU, K. R. and Westphal, H. (1986).Tissue-specific expression in transgenic mice of a fused genecontaining RSV terminal sequences. Science 231, 1574-1577.

Palmlter, R. D. and Brinster, R. L. (1986). Germ-line transformationof mice. Ann. Rev. Genet. 20, 465-499.

Rassoulzadegan, M., Leopold, P., Vallly, J. and Cuzin, F. (1986).Germ-line transmission of autonomous genetic elements intransgenic mouse strains. Cell 46, 513-519.

Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). MolecularCloning: A Laboratory Manual (Cold Spring Harbor Lab., ColdSpring Harbor, NY), 2nd Ed.

Stinchcomb, D. T., Shaw, J. E., Can-, S. H. and Hirsh, D. (1985).Extrachromosomal DNA transformation of Caenorhabditiselegans. Molec. cell. Biol. 5, 3484-3496.

Streisinger, G., Walker, C , Dower, N., Knauber, D. and Singer, F.(1981). Production of clones of homozygous diploid zebrafish(Brachydanio rerio). Nature 291, 293-2%.

Stuart, G. W., McMurray, J. V. and Westerfield, M. (1988).Replication, integration, and stable germ-line transmission offoreign sequences injected into early zebrafish embryos.Development 103, 403-412.

Stuart, G. W., Vlelklnd, J. R., McMurray, J. V. and Westerfield, M.(1990). Stable lines of transgenic zebrafish exhibit reproduciblepatterns of transgene expression. Development 109, 577-584.

Swain, J. L., Stewart, T. A. and Leder, P. (1987). Parental legacydetermines methylation and expression of an autosomal transgene:A molecular mechanism for parental imprinting. Cell 50, 719-727.

Walker, C. and Streisinger, G. (1983). Induction of mutations by g-rays in pregonial germ cells of zebrafish embryos. Genetics 103,125-136.

Westerfield, M., ed. (1989). The Zebrafish Book. Eugene: Univ. ofOregon Press.

Wilson, C , Pearson, R. K., Bellen, H. J., O'Kane, C. J.,Grossniklaus, U. and Gehring, W. J. (1989). P element-mediatedenhancer detection: an efficient method for isolating andcharacterizing developmentally regulated genes in Drosophila.Genes Dev. 3, 1301-1313.

(Accepted 24 February 1992)