ARTICLE

Histology-Based Screen for Zebrafish Mutants withAbnormal Cell DifferentiationManzoor-Ali P.K. Mohideen,1 Lee G. Beckwith,1 Gladys S. Tsao-Wu,2 Jessica L. Moore,1 Andrew C.C. Wong,3

Mala R. Chinoy,2 and Keith C. Cheng1,4*

The power of histology to define states of cell differentiation was used as the basis of a mutagenesis screen inzebrafish. In this screen, 7-day-old parthenogenetic half-tetrad larvae from potential carrier females were screenedfor mutations affecting cell differentiation in hematoxylin and eosin-stained tissue sections. Seven,noncomplementing, recessive mutations were found. Two mutations affect only the retina: segmentedphotoreceptors (spr) show a discontinuous photoreceptor cell layer; vestigial outer segments (vos) has fewerphotoreceptor cells and degenerated outer segments within this cell layer. Three mutants have gut-specific defects:the epithelial cells of kirby (kby) are replaced by ballooned cells; the intestines of stuffy (sfy) and stuffed (sfd) containincreased luminal mucus. Two mutations affect multiple organs: disordered neural retina (dnr) has disrupted retinallayering and mild nuclear abnormalities in the gut and liver; and in huli hutu (hht), the retinal cell layers aredisorganized and multiple organs have mild to severe nuclear abnormalities that are reminiscent of the atypia ofhuman neoplasia. Each mutation appears to be homozygous lethal. This screen is proof of principle for the feasibilityof histologic screens to yield novel mutations, including potential models of human disease. The throughput for thistype of screen may be enhanced by automation. Developmental Dynamics 228:414–423, 2003.© 2003 Wiley-Liss, Inc.

Key words: zebrafish mutant; larval arrays; histology; cell differentiation; retina; cancer

Received 14 May 2003; Accepted 4 August 2003

INTRODUCTIONInfinite combinations of architec-tural and cytologic features of tissuecells are detectable by the light mi-croscopic study of stained tissuesections. Histologic studies distin-guish a multitude of cell types, cellprocesses, and abnormalities. Thispower is routinely exploited in thefield of pathology, where histologicfeatures are used to assess biologi-

cal behavior, identify abnormalitiesin cell differentiation, and evaluatecell injury in human diseases, includ-ing cancer (Demay, 1996; Cotran etal., 1999). We reasoned that thesmall size of zebrafish larvae contain-ing primordia of most adult tissues,combined with the high resolutionand sensitivity associated with lightmicroscopy of tissue sections, couldbe harnessed to detect subtle

changes in cellular structure. Clon-ing of such mutations would allowthe dissection of their genetic basis.The subtlety of many lesions routinelyassessed in pathology indicate thatmany abnormalities in organogene-sis and/or cytology would be difficultor impossible to detect by micros-copy of live larvae, even whentransparent. Pigmentation of the lar-vae further impairs detection of cel-

1The Jake Gittlen Cancer Research Institute, Department of Pathology, Pennsylvania State University College of Medicine, Hershey,Pennsylvania2The Jake Gittlen Cancer Research Institute, Department of Surgery, Pennsylvania State University College of Medicine, Hershey, Pennsyl-vania3Department of Human Genetics, University of Chicago, Chicago, Illinois4The Jake Gittlen Cancer Research Institute, Department of Biochemistry and Molecular Biology, Pennsylvania State University College ofMedicine, Hershey, PennsylvaniaGrant sponsor: National Institutes of Health; Grant numbers: RO1-CA73935; RO1-HD040179; NRSA IF32GM19794; Grant sponsor: NSF; Grantnumber: MCB-93198174.*Correspondence to: Keith C. Cheng, Jake Gittlen Cancer Research Institute H059, Pennsylvania State University College of Medicine,500 University Drive, Hershey, PA 17033. E-mail: kcheng@psu.edu

DOI 10.1002/dvdy.10407

DEVELOPMENTAL DYNAMICS 228:414–423, 2003

© 2003 Wiley-Liss, Inc.

lular changes. We expected a histol-ogy-based screen to find previouslyundiscovered mutants for two rea-sons. First, stained tissue sections pro-vide a higher resolution than unsec-tioned, unstained, whole embryos,as used in two large publishedscreens (Haffter et al., 1996; Drieveret al., 1996). Second, organogenesisis more advanced in 7-day larvaethan in embryos or younger larvae,which permits identification of muta-tions affecting later stages of cell dif-ferentiation. The ability to embed ze-brafish larval arrays in agaroseconsisting of 64 larvae per block, de-veloped in our laboratory (Tsao-Wuet al., 1998) made the screen possi-ble. The first histology-based geneticscreen, performed in Drosphila, wasreported by Heisenberg and Bohl(1978). Here, we report the first histol-ogy-based genetic screen in a ver-tebrate.

Heterozygous mutations in carrierfemales were rendered homozy-gous in their progeny by gynogene-sis (Streisinger et al., 1981; Chengand Moore, 1997). Seven recessivemutations with novel phenotypeswere obtained from this initial screen,demonstrating that histology-basedscreens can detect interesting mu-tants that do not otherwise exhibitany easily identifiable nonhistologicphenotypes. The mutants reportedhere include two with defective ret-inal photoreceptor differentiation,three with abnormal differentiation

of the intestinal epithelium, and twowith multiorgan defects. The lattertwo show disruption of retinal layersin addition to cytologic abnormali-ties in multiple organs. The mutantsfound include models of retinal dys-plasia and of cytologic dysplasiaseen in human cancer. The muta-tions identified in this screen are dif-ferent from those reported in earlierzebrafish screens, further validatingthe utility of histology-based screens.

RESULTS AND DISCUSSION

Half-tetrad progeny were producedfrom F1 females carrying N-ethyl-N-nitrosourea (ethylnitrosourea, ENU)-induced mutations and werescreened for histologic defects at 7days of development (Fig. 1). Thesmall size of zebrafish larvae (�5-mm-long at 7 days postfertilization)made it possible to fit a section froman array of 64 larvae on one micro-scope slide (Tsao-Wu et al., 1998).The screen was performed at 7 days,because most organs have distinctcell patterns and appearances bythat age (Kimmel et al., 1995). Lar-vae with gross abnormalities notedunder a dissection microscope werediscarded, based on the asumptionthat mutations causing such defectswould have been found in past ge-netic screens. We chose not to useolder larvae because gut mutationsaffecting absorption would be ex-pected to be lost beginning on day

8, when yolk is normally depleted. Atotal of 3,281 larvae from 72 histo-logic arrays of up to 7-day-old par-thenogenetic half-tetrad larvae perF1 female were initially screened at�25–�400 magnification, and asnecessary, examined at �1,000magnification for additional cyto-logic detail. Among the embeddedlarvae, 2,038 (62%) larvae werescreened for eye morphology and1,670 (51%) were screened for intes-tinal morphology in at least one ofthree stained sections. Seven puta-tive mutants were obtained, all ofwhich showed transmission to atleast one succeeding generationand exhibited Mendelian inheri-tance. Five organ-specific and twopleiotropic mutations were found(Table 1). Of these, vos and hht aredescribed in greater detail.

Eye-Specific Mutants

The two eye mutants vestigial outersegments (vos) and segmentedphotoreceptors (spr) appear normalunder the dissecting microscope.Histologic examination of these twomutants revealed defects specific tothe photoreceptor cell layer (PCL) ofthe retina. The photoreceptor cellsof the wild-type 7-day zebrafish eyeare elongated and are oriented per-pendicular to the outer circumfer-ence of the eye (Fig. 2A,B). The PCLconsists of the outer nuclear layer,and inner and outer segments. Theouter nuclear layer stains blue with

Fig. 1. Genetic screen for histologic mutants in zebrafish. N-Ethyl-N-nitrosourea (ethylnitrosourea, ENU) was used to mutagenize adultmales, which were outcrossed with homozygous goldenb1 (gol) females to generate F1 progeny for screening. Early pressure partheno-genesis (EP) was used to generate half-tetrads from F1 females. At 7 days of age, half-tetrad larvae were fixed in 10% neutral bufferedformalin, and up to 64 sibs from individual EP families were arrayed in individual agarose blocks. After embedding in paraffin, 6-�m-thicksections were cut and stained with hematoxylin and eosin for histologic screening. F1 carriers that generated half-tetrad progeny withinteresting phenotypes were then outcrossed for further study, including confirmation of Mendelian inheritance. m, mutant.Fig. 2. vos specifically affects the photoreceptor cell layer of the developing zebrafish retina. Figure depicts coronal sections through theeye. A–D: Wild-type retina. E–H: vos mutant retina. A,B,E,F: hematoxylin and eosin–stained sections; C,D,G,H are scanning electronphotomicrographs. B,F: Enlarged views of the photoreceptor cell layer (indicated between arrowheads in A and E, respectively). C,D,G,H:Outer segments of the photoreceptor cells; D and H are the magnified images of the boxed area in C and G, respectively. rpe/c, retinalpigment epithelium/cells; is/os, inner segment/outer segment of the photoreceptor cells; onl, outer nuclear layer; inl, inner nuclear layer;ipl, inner plexiform layer; gcl, ganglion cell layer; opl, outer plexiform layer; the space between the white arrowheads indicates thethickness of the photoreceptor cell layer; asterisks, photoreceptor outer segment; red arrows, vacuolar spaces; black arrowheads,mitochondria. Scale bars � 10 �m in A,B,E,F, 1 �m in C,D,G,H.Fig. 3. The loss of outer segments of the photoreceptor cells in vos mutants occurs at 5 days postfertilization (dpf). Embryos (from 3 to 6dpf) from pair-wise crosses of previously identified vos heterozygotes were fixed, embedded, sectioned, stained with hematoxylin andeosin, and were examined under the light microscope for any abnormalities in the developing retina. At 3 and 4 dpf, the developingretina looked normal in all embryos (data not shown). At 5 dpf, in approximately 25% of the embryos (vos), the outer segments are notapparent (A,B, wild-type; C,D, vos). In addition, the retinal pigment epithelium layer in vos appears to be thicker than the wild-type. At 6dpf, approximately 25% of the embryos (vos) continue to exhibit the loss of outer segments (data not shown). A,C are low-power viewsshowing all the cell layers of the developing retina; B,D, are enlarged views of the photoreceptor cell layer (between red arrowheads),Scale bars � 10 �m.

ZEBRAFISH HISTOLOGIC MUTANT SCREEN 415

Fig. 1.

Fig. 2.

Fig. 3.

hematoxylin and corresponds ultra-structurally to the nucleus. The innerand outer segments stain pink witheosin (Fig. 2A,B). Ultrastructurally, theinner segment includes mitochon-dria and Golgi apparatus, whereasthe outer segment contains longstacks of membranes (Fig. 2C,D)containing the light-sensitive opsinsresponsible for light detection (Adlerand Hewitt, 1994; Schmitt and Dowl-ing, 1999). The PCL in wild-type is ap-proximately 20 �m thick, includingnuclei and outer segments (Fig.2A,B). In contrast, the PCL in vos isonly 7–10 �m thick, consisting pri-marily of a nuclear layer (Fig. 2E,F).Furthermore, the number of photo-receptor cells in vos is decreased byapproximately 50% (Fig. 2E), and theouter segments of the remainingphotoreceptor are essentially unde-tectable by light microscopy (Fig.2F). Vacuolar spaces suggestive ofcellular degeneration are also scat-tered near the nuclei of the PCL invos (Fig. 2E,F). The photoreceptor

cells that remain have outer seg-ments that are misoriented and di-minished in size (Fig. 2G,H). Low-

power electron photomicrographsof vos retina were used to determinethat the volume of outer segments

TABLE 1. Summary of Mutants Found in Zebrafish Histologic Screena

Mutant (abbreviation) Linkage group Eye defects Gut and other defects

segmentedphotoreceptors (spr)

* Photoreceptor layer discontinuous,normal to reduced thickness

None

vestigial outersegments (vos)

LG23**(Z23370;0.6 cM)

Reduced number ofphotoreceptor cells;photoreceptor outer segmentsfragmented and parallel tocircumference

None

kirby (kby) LG18**(Z13692;2.3 cM)

None Ballooned cells, mild nuclearpleomorphism, nuclearfragments

stuffy (sfy) LG23**(Z20039;1.1 cM)

None Increased luminal mucus;clusters of mucous cells

stuffed (sfd) LG6**(Z9738;4.6 cM)

None Increased luminal mucus

disordered neuralretina (dnr)

Not determined Small eyes; poorly formed lens;disorganized retina layers

Mild nuclear pleomorphism,hyperchromasia, enlargement,and irregularity in intestine andliver

huli hutu (hht) LG12**(Z22103;2.3 cM)

Severe retinal disorganization,decreased cell number, nuclearfragments

Severe atypia of esophagus,intestine, liver, exocrinepancreas, pancreatic duct,and swim bladder; enlarged,dark, irregular nuclei, nuclearpleomorphism, multinucleation;rare tripolar mitosis (seen inbrain and pancreas)

aLinkage group data presented as nearest tested linked marker and distance from its centromere.*, spr is not linked to LG05, 06, 12, 18, and 23; **, vos mapped to the south arm, while kby, sfy, sfd, and hht mapped to thenorth arm of their respective linkage groups.

Fig. 4. The photoreceptor cell layer in spr mutants is formed in patches. At 7 days postfer-tilization, spr mutants do not have a continuous photoreceptor cell layer (PCL). The PCL ispresent as patches (indicated by white arrowheads). There is also no distinct outer plexi-form layer in spr mutants (compare with wild-type in Fig. 2A). Scale bar � 10 �m.

Fig. 5. Histologic mutations affecting the intestine. Low-power (upper row) and high-power(lower row) photographs of wild-type (A,B), kby (C,D), and sfy (E,F). Intestinal epithelialcells are ballooned in kby (C) and lack brush border (D); the lumen is indicated byasterisks. Mucus accumulates in sfy (E); abnormal clumps of mucous cells are occasion-ally found in sfy mutants (E, F, black arrowheads). wt, wild-type. Scale bars � 10 �m in A–F.

Fig. 6. Mutants with multiorgan defects. The mutations dnr (A,B,D) and hht (E–H) showarchitectural and cytologic effects in multiple organs. At low power, disruption of theneural retina is evident in both dnr (A) and hht (E). The dnr mutation shows cellular atypiain the intestine (white arrowheads in B; compare nuclear size with normal intestine in Fig.3A) and liver (white arrowhead in D; compare with normal liver shown in C). Atypical cellsof hht (F,G) include giant cells in the intestine (white arrowheads in F; a single giant cell isshown in G), pancreas (F), central nervous system, and other endodermally derivedepithelia. An atypical mitosis in hht pancreas is shown in F (downward arrow, lower left)and H (arrowhead). Cells with nuclear atypia were present in proximal intestine andpancreas (F, black arrows). Scale bars � 10 �m in A–H.

Fig. 7. The distributions of nuclear areas in wild-type vs. hht intestine. The area of 100 hhtand 100 wild-type intestinal nuclei were obtained as described in Experimental Proceduressection and plotted by size, wild-type in black and hht in red.

ZEBRAFISH HISTOLOGIC MUTANT SCREEN 417

Fig. 4.

Fig. 5.

Fig. 6.

Fig. 7.

was less than 10% of that seen inwild-type retina (not shown).

However, the electron photomi-crographs show that vestigial outersegments are present in vos (Fig.2G,H); in the outer segments in vosare oriented parallel, rather thanperpendicular, to the circumfer-ence of the eye (Fig. 2H). To deter-mine the time of onset of the vosphenotype, we examined larvaeobtained from a pair of vos het-erozygous parents from 2 to 7 dayspostfertilization (dpf). Histologically,photoreceptors can be distin-guished at 2 dpf with the outer seg-ments beginning to appear by 2.5dpf (Branchek and Bremiller, 1984).The first defects were detected inthe photoreceptors of vos larvae atday 5, when 25% of the larvae fromcrosses between heterozygoteslacked definitive outer segments(Fig. 3D). The retinal pigmented epi-thelium (RPE) appears to be thickerin vos (Figs. 2E–G, 3D) than wild-type(Figs. 2A–C, 3C). We do not know atthis time whether the changes in theRPE have any bearing on the PCLdefects of vos. In Xenopus laevis, ret-inal pigmented epithelium has beenreported to participate in the orga-nization of photoreceptor outer seg-ments (Stiemke et al., 1994), and inmouse, it is required for the develop-ment and maintenance of the neu-ral retina (Raymond and Jackson,1995). In zebrafish, analyses of themosaic eyes mutant has providedthe first genetic evidence that sig-nalling from the RPE to the retina isrequired for the induction of properretinal organization (Jensen et al.,2001). Furthermore, the gantenbeinzebrafish mutant shows both conedystrophy and RPE degeneration(Biehlmaier et al., 2003). Adults fromcrosses between vos heterozygoteswere genotyped for SSR markerstightly linked to vos; no homozygotesfor the linked allele were identified,indicating that the vos mutation islethal (data not shown).

In wild-type larvae, photorecep-tors form a continuous, distinct layerjust inside the retinal pigmented ep-ithelium (Fig. 2A,B). In contrast, sprhas a discontinuous photoreceptorlayer, with patches of variable sizeand thickness (Fig. 4). The spr pheno-types varied in the size and number

of patches, and the thickness of thediscontinuous PCL. It also lacks aclear outer plexiform layer. At 7 dpf,the RPE also appears to be thickerthan those of wild-types. We do notknow whether spr mutants first de-velop an intact PCL that is then lostin patches or whether the PCL formsonly in patches to begin with. Thismutant did not exhibit any other de-fects identifiable under the dissec-tion microscope. This defect wasfully penetrant and followed Men-delian inheritance. Unfortunately,the spr mutation was lost before wecould perform further characteriza-tion.

Gut-Specific Mutants

Three mutations, kirby (kby), stuffy(sfy), and stuffed (sfd) affect only theintestine. kby is characterized by bal-looned gut epithelial cells, which alsodisplay perinuclear clearing (Fig.5C,D; compare with the correspond-ing wild-type sections in 5A,B). In ad-dition, the epithelial cells are deficientin brush borders (Fig. 5D). The bal-looned cells suggest degenerativechanges, while brush border defi-ciency suggests a defect in cell differ-entiation. The nuclei in kby intestinalepithelium (Fig. 5D) are larger andmore variable in size and shape thanwild-type (Fig. 5B). Occasional nu-clear fragments and pyknotic nucleiare also found in kby (not shown). Fur-ther study of this mutant may provideinsights into mechanisms of cell differ-entiation, nuclear atypia, and cell de-generation in the intestine.

The proximal intestine of wild-type7-day larvae contains little to no mu-cus (Fig. 5A,B). In the distal intestine,small amounts of mucus are associ-ated with scattered, solitary mucuscells surrounded by absorptive epi-thelium. However, in the two intesti-nal mutations, sfy and sfd, there isincreased luminal mucus in the prox-imal intestine (Fig. 5E, only sfy isshown). In addition, in sfy, clusters ofmucus cells appear in the proximalintestinal epithelium (Fig. 5E,F, indi-cated by arrowheads). The mucuspresent in these mutants stains withthe same weak intensity as in wild-type with both mucicarmine andcolloidal iron mucus stains (Luna,1960; not shown). All three gut mu-

tants are fully penetrant and showMendelian inheritance in outcrosses.None of these phenotypes were de-scribed among the previously de-scribed zebrafish intestinal mutants(Pack et al., 1996; Chen et al., 1996).

Pleiotropic Mutations

Two mutations showed multiorganand cytologic defects (Fig. 6). Thefirst, disordered neural retina (dnr),exhibits defects in the eye, intestine,and the liver (Fig. 6). The eyes of dnrmutants show severely disrupted ar-chitecture of the neural retina (Fig.6A). Histologically, the lens and reti-nal layers of dnr are poorly formed.The different layers of the retina arenot distinct and have a disorderedappearance. The dnr retina con-tains scattered nuclear fragmentsand vacuolar spaces. This eye phe-notype is detectable as early as day3 (not shown). The dnr gut pheno-type is characterized by a consistentnuclear pleomorphism of the intestine(Fig. 6B) and nuclear enlargement inthe liver (Fig. 6D; compare with nucleiof normal liver in 6C). The nuclearatypia in the intestine and liver of dnrmutants cosegregated consistentlywith the eye phenotype, suggestingthat a mutation in the dnr gene is re-sponsible for these defects. The dnrmutants could be identified by theirsmaller eye phenotype with the helpof a dissection microscope. However,their gut defects could not be de-tected without histologic examina-tion. The mutant larvae do not surviveto adulthood. Unfortunately, this mu-tant has also been lost before furthercharacterization was possible.

The second pleiotropic mutation,huli hutu (hht), causes striking archi-tectural and cytologic changes inseveral organs (Fig. 6E–H). The eyesof a variable fraction of mutants arereduced in size. The cell types of dif-ferent retinal layers are mixed to-gether to varying degrees. In themost extreme cases, no discernibleretinal layering is evident, similar towhat is seen in dnr. The eye pheno-type is detectable at 2 dpf (notshown) and can be mild to severe at7 dpf. The retinal disorganization isalso associated with scattered nu-clear fragments suggestive of apo-ptosis (Fig. 6E). Accordingly, the

ZEBRAFISH HISTOLOGIC MUTANT SCREEN 419

number of nuclei per cross-sectionof the eye was 25% to 75% of normal.The decrease in cell number and se-verity of retinal disorganization wereroughly proportional to the degreeof disruption of layers and to eyesize, which ranged from normal toapproximately 75% of normal. Thenuclear defect is not evident in mus-cle, blood, gills, or blood vessels.

The most interesting of the hhtphenotypes is severe cytologicatypia in the intestine, pancreas,liver, swim bladder, and pneumaticduct. This atypia includes a greatvariability in nuclear size and shape,frequent multinucleation, and archi-tectural disorganization (Fig. 6F,G).There were occasional giant cells inthe intestine and pancreas (Fig.6F,G). Despite architectural disorga-nization of the intestine of 7-day-oldhht larvae, formation of microvilliwas preserved (not shown). In 100randomly selected cells in hht, 19had multiple nuclei (up to 3 per cell).An intriguing but rare finding was tri-polar mitosis (Fig. 6H); the signifi-cance and potential relationship ofthis phenomenon to the other phe-notypes of hht remains to be deter-mined.

To quantify the degree of nuclearenlargement in hht intestine, cross-sectional nuclear areas were mea-sured digitally for the intestinal epi-thelial nuclei of 100 cells of hhtlarvae and wild-type siblings. Thearea of hht nuclei vary in size andare significantly larger than wild-type (Fig. 7); mean nuclear areaswere 34.8 �m2 for hht and 12.1 �m2

for wild-type. Because the densityof stain is at least as dark as wild-type, we expect this difference toreflect differences in DNA content.The degree of nuclear irregularity,which correlates with malignantpotential in human cancer (De-may, 1996), was measured as theratio of circumference to nucleararea (“circularity”); greater valuescorrelate with increased irregular-ity. The mean circularity was 16.9for hht and 14.3 for wild-type. Thus,the nuclei of intestinal epithelialcells in hht are larger, more vari-able in size, and are irregular in out-line compared with wild-type.

Complementation Analyses

To determine the number of genescorresponding to the seven reces-sive mutations found, we performedcomplementation crosses and half-tetrad mapping. Histologic analysisof reciprocal complementationcrosses between known carriers ofthe four mutations with eye pheno-types (spr, vos, dnr, and hht) showedonly wild-type progeny, consistentwith these mutations occurring infour different genes. Half-tetrad ge-netic mapping (Streisinger et al., 1986;Mohideen et al., 2000) of the five mu-tants with intestinal phenotypes (kby,sfy, sfd, dnr, and hht) showed thatthey reside on different chromo-somes. Each of the seven mutantswere obtained from a different ENU-treated G0 male. Taken together, thedata suggest that we have identifiedseven different single locus mutations.

Genetic Map Positions

Five mutations (vos, kby, sfy, sfd, andhht) were mapped to specific chro-mosome arms on the SSR geneticmap by centromere-linkage map-ping (Table 1); fine mapping for vosand hht are in progress. Mapping ofspr and dnr is incomplete due to lossof these mutant lines.

Viability

The mutations identified in thisscreen are lethal after 8 dpf. Up tothis time, the eye-specific mutantsspr and vos, as well as the gut mu-tants kby, sfy, and sfd and the multi-organ mutant dnr look normal, withsome hht mutants exhibiting smalleyes. However, they all die soon af-ter 8 dpf. We speculate that the his-tologic defects in spr and vos mightrender them blind, which inhibitstheir feeding ability. After 8 dpf, theyolk becomes depleted and the lar-vae depend on food from their en-vironment for survival. We also be-lieve that, in kby, sfy, and sfd, theirhistologic gut defects may causethem to die because of insufficientfood absorption. The combined de-fects in both the eye and gut in dnrand hht confer both disadvantages.Progeny from crosses between het-erozygotes of these mutations havebeen grown to adulthood. Homozy-

gotes have not been recovered. Inthe case of vos and hht, we havegenotyped adults with tightly linkedSSR markers and have not detectedany homozygotes among progenyderived from known heterozygotes.

Frequency of Mutants

In this screen, half-tetrad progenyfrom heterozygous mothers werescreened for homozygous mutationsaffecting their histologic appear-ance. Because centromeric muta-tions are more frequently homozy-gous than are telomeric mutations inhalf-tetrads (Streisinger et al., 1986),half-tetrad screens have a bias to-ward detecting mutations that tendto be close to centromeres (Grun-wald and Streisinger, 1992; Henion etal., 1996). Thus, the method used in thisreport will underestimate the totalnumber of mutations that were in-duced and that could be detected inthe histologic screen. Three levels ofhistologic sections were examinedper larva in this screen. Examining ad-ditional levels per block, while increas-ing the amount of work per block,might be expected to yield additionalmutants. Under our conditions, ap-proximately 1 in 10 mutagenized F1individuals had a histologically de-tectable mutation, whereas 1 in 1,000F1s had a mutation at a single pre-defined locus, golden. We estimatethat approximately 100 genes existthat, when mutated, will yield defectsnot seen in the dissecting scope butseen with hematoxylin and eosin.

Relevance to Human Disease

The mechanisms responsible for de-fective photoreceptor layers in sprand vos may relate to one or morehuman genetic retinal diseases (He-witt and Adler, 1994). In particular, thehistologic phenotypes of photorecep-tor cell layer and pigment cells of vosare similar to human retinal degener-ative diseases such as retinitis pigmen-tosa. For example, a similar deficiencyof outer segments is seen in mice withmutations in the otx-related ho-meobox gene crx (Furukawa et al.,1997, 1999). Mutant alleles of crx inman are associated with retinal de-generation syndromes, includingcone–rod dystrophy, Leber congeni-

420 MOHIDEEN ET AL.

tal amaurosis, and retinitis pigmentosa(Furukawa et al., 1997, 1999; Freund etal., 1997; Swain et al., 1997; Sohocki etal., 1998; Swaroop et al., 1999). The ze-brafish crx homolog has been clonedand subsequently mapped to LG05 onthe zebrafish genetic linkage map (Liuet al., 2001). However, we have con-firmed that vos and spr are not al-leles of crx, because the vos locus ison LG23 and spr is not linked toLG05.

Histologic sections allow us to de-tect changes otherwise hidden be-neath normal, opaque choroidalpigments that appear after day 3 inzebrafish. Both mutant phenotypesare clearly different from those withphotoreceptor defects found in ear-lier screens (Malicki et al., 1996; Fa-dool et al., 1997; Neuhauss et al.,1999). The photoreceptor mutantsidentified in previous screens havesmall eyes and are usually associ-ated with additional phenotypessuch as poor touch response, braindefects, and abnormal pigmenta-tion. The vos and spr mutants havenormal-sized eyes and do not haveany other scoreable defects. Onlyupon histologic examination aretheir photoreceptor cell defects ap-parent. It is unclear whether the pri-mary defect in either mutant lies inthe photoreceptor cells themselvesor surrounding cells, such as the pig-mented epithelium, upon which dif-ferentiation or maintenance of thephotoreceptor outer segment de-pends (Hewitt and Adler, 1994).

The common histologic pheno-types of human cancer include disor-ganization of tissue architecture, inva-sion, and a variety of cytologicfeatures, including nuclear hyper-chromasia, nuclear pleomorphism, ir-regular nuclear outlines and chroma-tin patterns, multinucleation, andmultipolar mitoses (Frost, 1986; De-may, 1996; Cotran et al., 1999). De-spite the critical importance of thesehistopathologic phenotypes in clinicalmedicine, their genetic origins re-mains largely unknown. To our knowl-edge, the hht mutation is the first ver-tebrate mutant to be identified by itsatypical histologic phenotype. Be-cause the hht phenotype includesseveral of the cytologic features ofhigh grade dysplasia, which fre-quently progresses to cancer, further

studies may contribute to our under-standing of cytologic cancer pheno-types. It will be interesting to deter-mine whether hht heterozygotes aresusceptible to cancer.

Rudolf Virchow not only played akey role in defining the cellular basisof disease, largely as revealed byhistologic analysis, but also pointedout the importance of genetics ingaining understanding of normaland abnormal cell physiology (Vir-chow, 1855). This screen reflects thefirst application of a classic geneticapproach to these fundamentalideas.

Potential Throughput ofHistologic Screens

A histology screen such as the onereported here involves several steps,some of which can be automatedto improve the throughput of thescreen. First, the embedding step in-volves careful positioning of individ-ual larvae into each agarose well.The maximum speed we haveachieved for embedding is approx-imately 10 min per block of 64 lar-vae, or approximately 10 sec perlarva. We imagine that a mold thatclosely matches the shape of thelarva may allow more rapid position-ing than presently possible. There ispotential for robotic positioning ofthe embryos (Schneider, 2003) andautomated removal of excess waterfrom the well that otherwise causespoor sections. The next step thatcould be automated is that of scor-ing sectioned and stained embryos.While screening of slides of larval ar-rays is most rapidly done by an ex-perienced human operator, writtenand image documentation of mu-tant phenotypes can be greatly fa-cilitated by combining databasesoftware with virtual slide scanningsoftware such as that offered by Mi-croBrightField (Williston, VT) or AperioTechnologies (Vista, CA). The virtualimages may be digitized, individualembryos recognized and labeled,and linkage to appropriate data-base fields be automated. Our lab-oratory is currently developing suchautomations to facilitate histologicscreening of zebrafish larvae. In-creases in the throughput of histo-logic screens can be applied not

only for genetic screens but also fortoxicologic, forward genetic (Nas-evicius and Ekker, 2000), and chem-ical screens (Peterson et al., 2000).

EXPERIMENTAL PROCEDURES

Zebrafish Strains and Fish Care

Wild-type male zebrafish used formutagenesis were obtained fromEkkwill (Gibsonton, FL) and Liles(Ruskin, FL); microsatellite alleles inthe two strains were similar at severalloci (unpublished data). Homozy-gous goldenb1 females used forcrosses with mutagenized maleswere descendants of goldenb1 fishobtained from the Zebrafish Re-source Center, University of Oregon.A strain obtained from India was ini-tially used to introduce polymor-phisms by crosses to identified fe-male carriers; these were discardedafter poor breeding of the progenyled to the loss of mutant lines. The fishwere raised in a recirculating systemas described (Beckwith et al., 2000).Fish were maintained and bred forgenerating larvae as previously de-scribed (Westerfield, 1995). Beforefixation for histology, larvae wereanesthetized using tricaine methanesulphonate (Westerfield, 1995).

Mutagenesis

Males, 7- to 9-month-old, were usedfor mutagenesis. The males were se-lected for high fertility by previouspair-wise matings. Germline mutationswere induced in adult male zebrafishwith 2.5 mM ENU (Sigma), as de-scribed (Mullins et al., 1994; Beckwithet al., 2000). Mutagenized males werefirst crossed 1 week after mutagenesisto enable expulsion of sperm with sin-gle-strand DNA damage. Beginning 2weeks postmutagenesis, males wereoutcrossed to homozygous goldenb1

females to determine germline muta-tion rates and to generate F1 forscreening.

Early Pressure Parthenogenesis

Mutations induced by ENU in G0males and carried in �6-month-old F1females were made homozygous by“early pressure” gynogenesis (Fig. 1;Streisinger et al., 1981) using condi-tions described previously (Gestl et al.,

ZEBRAFISH HISTOLOGIC MUTANT SCREEN 421

1997). The resulting half-tetrads wereexamined for mutations in histologicsections. The fraction of homozygoushalf-tetrads from maternal heterozy-gous mutations ranges from 5% to50%, depending upon distance fromthe centromere (Streisinger et al.,1986; Kauffman et al., 1995).

Histology

Seven-day-old larvae without any vis-ible defects were fixed in 10% neutralbuffered formalin for 16 to 64 hr at4°C. Arrays of 32 to 64 larvae perclutch were embedded in individualagarose blocks, processed into paraf-fin, and cut into 6-�m-thick sections,three sets at a time as described pre-viously (Tsao-Wu et al., 1998). Oneslide per set was stained with hema-toxylin and eosin by using standardprotocols (Luna, 1960), and the re-maining sections were left unstainedfor further study or for extracting DNAfor genetic mapping. F1 females thatproduced half-tetrad progeny with in-teresting histologic phenotypes wereoutcrossed to the India or AB strainsfor further study.

Genetic Mapping

Mutations were first mapped to achromosome arm by centromerelinkage analysis as used by Johnsonet al. (1995), using markers de-scribed previously (Mohideen et al.,2000). For genetic mapping, DNAfrom mutant half-tetrad larvae aswell as from wild-type sibs were ex-tracted from corresponding un-stained, deparaffinized sections asdescribed previously (Tsao-Wu et al.,1998). Centromere-linkage analysiswas performed using SSR markersthat are linked to the centromeres ofthe 25 linkage groups (Mohideen etal., 2000). Upon establishing linkageto a specific linkage group, markersflanking the centromere were usedto determine the specific arm onwhich the mutation is located.

Electron Microscopy

Larvae were fixed for electron mi-croscopy in 2.5% glutaraldehyde(Electron Microscopy Sciences, FortWashington, PA) in 0.1 M sodium ca-codylate buffer (Sigma) at 4°C for 2

hr, or in 1% glutaraldehyde 4% Ultra-pure formalin (EM Sciences) in 0.1 Mcacodylate buffer at 4°C for 2 hr.Fixed larvae were washed in 0.1 Mcacodylate buffer on ice, post-fixedwith 2% osmium tetroxide in 0.1 Mcacodylate buffer for 1 hr, dehy-drated in prechilled graded alcoholsolutions, and embedded in EMbed812 (EM Sciences). Light microscopyof methylene blue–stained 1-�mtransverse sections were used toscreen sections for electron micros-copy. Ultra-thin, silver–grey sections(60 nm) were cut from preselectedblocks, stained with uranyl acetateand lead citrate, and examined ona Philips 400 transmission electron mi-croscope.

Photography

Sections were photographed withKodak Ektachrome T160 slide film on aZeiss Axiophot with a Zeiss MC spotcamera, using a variety of objectives.The 35-mm slides were scanned usinga Polaroid SprintScan 35 Plus scanner(Cambridge, MA), and labeling andadjustments were done with AdobePhotoshop.

Morphometry

For cell counts in hht, nuclei werecounted manually on 8- � 10-inchenlargements obtained by using acolor photocopier of full coronalsections of normal and hht eyes. Nu-clear size and irregularity were de-termined by digital photography ofimages from 6-�m thick, hematoxy-lin and eosin–stained sections of7-day-old wild-type and hht siblingsderived from early pressure parthe-nogenesis or crosses. Digitized im-ages were acquired from the slidesby using an Olympus BH-2 micro-scope with a Zeiss �40 dry objective,a Sony DKC-5000 digital camera,and Adobe Photoshop. Outlines ofdigital images of 100 nuclei, eachfrom several wild-type and mutantlarvae, were traced by using Opti-mas 4.2 software (Silver Spring, MD)to calculate nuclear areas and cir-cumference/area ratios.

ACKNOWLEDGMENTSWe thank Lynn Budgeon and Xiao-Hong Wang for histologic work, Dr.

Ian Zagon for help with determiningnuclear areas, Roland Meyers forelectron microscopy, Michele Arosfor technical assistance, and PeggyHubley for fish facility management.This work was supported by the JakeGittlen Memorial Golf Tournamentand by grants from the NIH (toK.C.C. and J.L.M.) and the NSF (toK.C.C.).

REFERENCES

Adler R, Hewitt AT. 1994. The outer retina:structural and functional consider-ations. In: Ryan SJ, editor. Retina. St.Louis: Mosby. p 1857–1868.

Beckwith LG, Moore JL, Tsao-Wu GS, Harsh-barger JC, Cheng KC. 2000. Ethylnitro-sourea induces neoplasia in zebrafish(Danio rerio). Lab Invest 80:379–385.

Biehlmaier O, Neuhauss SC, Kohler K.2003. Double cone dystrophy and RPEdegeneration in the retina of the ze-brafish gnn mutant. Invest OphthalmolVis Sci 44:1287–1298.

Branchek T, Bremiller R. 1984. The devel-opment of photoreceptors in the ze-brafish, Brachydanio rerio. I. Structure.J Comp Neurol 224:107–115.

Chen JN, Haffter P, Odenthal J, Vogel-sang E, Brand M, van Eeden FJ, Furu-tani-Seiki M, Granato M, Hammer-schmidt M, Heisenberg CP, Jiang YJ,Kane DA, Kelsh RN, Mullins MC, Nus-slein-Volhard C. 1996. Mutations affect-ing the cardiovascular system andother internal organs in zebrafish. De-velopment 123:293–302.

Cheng KC, Moore JL. 1997. Genetic dis-section of vertebrate processes in thezebrafish: a comparison of uniparentaland two-generation screens. BiochemCell Biol 75:525–533.

Cotran RS, Kumar V, Collins T. 1999.Pathologic basis of disease. Philadel-phia: WB Saunders. 1425 p.

Demay RM. 1996. The cell, the art andscience of cytopathology. Chicago:American Society of Clinical Patholo-gists’ Press. 40–44 p.

Driever W, Solnica-Krezel L, Schier AF,Neuhauss SC, Malicki J, Stemple DL,Stainier DY, Zwartkruis F, Abdelilah S,Rangini Z, Belak J, Boggs C. 1996. Agenetic screen for mutations affectingembryogenesis in zebrafish. Develop-ment 123:37–46.

Fadool JM, Brockerhoff SE, Hyatt GA, Dowl-ing JE. 1997. Mutations affecting eyemorphology in the developing zebrafish(Danio rerio). Dev Genet 20:288–295.

Freund CL, Gregory-Evans CY, FurukawaT, Papaioannou M, Looser J, Ploder L,Bellingham J, Ng D, Herbrick JA, Dun-can A, Scherer SW, Tsui LC, Loutradis-Anagnostou A, Jacobson SG, CepkoCL, Bhattacharya SS, McInnes RR. 1997.Cone-rod dystrophy due to mutationsin a novel photoreceptor-specific ho-meobox gene (CRX) essential for main-

422 MOHIDEEN ET AL.

tenance of the photoreceptor. Cell 91:543–553.

Frost JK. 1986. In: Bibbo M, editor. Com-prehensive cytopathology. Philadel-phia: WB Saunders. pp 69–89.

Furukawa T, Morrow EM, Cepko CL. 1997.Crx, a novel otx-like homeobox gene,shows photoreceptor-specific expres-sion and regulates photoreceptor dif-ferentiation. Cell 91:531–541.

Furukawa T, Morrow EM, Li T, Davis FC,Cepko CL. 1999. Retinopathy and atten-uated circadian entrainment in Crx-de-ficient mice. Nat Genet 23:466–470.

Gestl EE, Kauffman EJ, Moore JL, ChengKC. 1997. New conditions for genera-tion of gynogenetic half-tetrad em-bryos in the zebrafish (Danio rerio).J Hered 88:76–79.

Grunwald DJ, Streisinger G. 1992. Induc-tion of recessive lethal and specific lo-cus mutations in the zebrafish with ethylnitrosourea. Genet Res 59:103–116.

Haffter P, Granato M, Brand M, MullinsMC, Hammerschmidt M, Kane DA,Odenthal J, van Eeden FJ, Jiang YJ,Heisenberg CP, Kelsh RN, Furutani-SeikiM, Vogelsang E, Beuchle D, Schach U,Fabian C, Nusslein-Volhard C. 1996. Theidentification of genes with unique andessential functions in the developmentof the zebrafish, Danio rerio. Develop-ment 123:1–36.

Heisenberg M, Bohl K. 1978. Isolation ofanatomical brain mutants of Drosoph-ila by histological means. Z Naturforsch34:143–147.

Henion PD, Raible DW, Beattie CE,Stoesser KL, Weston JA, Eisen JS. 1996.Screen for mutations affecting devel-opment of zebrafish neural crest. DevGenet 18:11–17.

Hewitt AT, Adler R. 1994. The retinal pig-ment epithelium and interphotorecep-tor matrix: structure and specializedfunctions. In: Ryan SJ, editor. Retina. St.Louis: Mosby. p 58–71.

Jensen AM, Walker C, Westerfield M. 2001.mosaic eyes: a zebrafish gene requiredin pigmented epithelium for apical local-ization of retinal cell division and lamina-tion. Development 128:95–105.

Johnson SL, Africa D, Horne S, Postle-thwait JH. 1995. Half-tetrad analysis inzebrafish: mapping the ros mutationand the centromere of linkage group I.Genetics 139:1727–1737.

Kauffman EJ, Gestl EE, Kim DJ, Walker C,Hite JM, Yan G, Rogan PK, Johnson SL,Cheng KC. 1995. Microsatellite-centro-mere mapping in the zebrafish (Daniorerio). Genomics 30:337–341.

Kimmel CB, Ballard WW, Kimmel SR, Ull-mann B, Schilling TF. 1995. Stages ofembryonic development of the ze-brafish. Dev Dyn 203:253–310.

Liu Y, Shen Y, Rest JS, Raymond PA, ZackDJ. 2001. Isolation and characteriza-tion of a zebrafish homologue of thecone rod homeobox gene. Invest Oph-thalmol Vis Sci 42:481–487.

Luna G, editor. 1960. Manual of histologicstaining methods of the armed forcesinstitute of pathology. New York:McGraw Hill.

Malicki J, Neuhauss SC, Schier AF, Solnica-Krezel L, Stemple DL, Stainier DY, Abdeli-lah S, Zwartkruis F, Rangini Z, Driever W.1996. Mutations affecting developmentof the zebrafish retina. Development 123:263–273.

Mohideen MA, Moore JL, Cheng KC.2000. Centromere-linked microsatellitemarkers for linkage groups 3, 4, 6, 7, 13,and 20 of zebrafish (Danio rerio).Genomics 67:102–106.

Mullins MC, Hammerschmidt M, Haffter P,Nusslein-Volhard C. 1994. Large-scalemutagenesis in the zebrafish: in searchof genes controlling development in avertebrate. Curr Biol 4:189–202.

Nasevicius A, Ekker SC. 2000. Effectivetargeted gene ’knockdown’ in ze-brafish. Nat Genet 26:216–220.

Neuhauss SC, Biehlmaier O, Seeliger MW,Das T, Kohler K, Harris WA, Baier H. 1999.Genetic disorders of vision revealed by abehavioral screen of 400 essential loci inzebrafish. J Neurosci 19:8603–8615.

Pack M, Solnica-Krezel L, Malicki J, Neu-hauss SC, Schier AF, Stemple DL, DrieverW, Fishman MC. 1996. Mutations affect-ing development of zebrafish digestiveorgans. Development 123:321–328.

Peterson RT, Link BA, Dowling JE, SchreiberSL. 2000. Small molecule developmentalscreens reveal the logic and timing ofvertebrate development. Proc NatlAcad Sci U S A 97:12965–12969.

Raymond SM, Jackson IJ. 1995. The reti-nal pigmented epithelium is requiredfor development and maintenance ofthe mouse neural retina. Curr Biol 5:1286–1295.

Schmitt EA, Dowling JE. 1999. Early retinaldevelopment in the zebrafish, Daniorerio: light and electron microscopicanalyses. J Comp Neurol 404:515–536.

Schneider I. 2003. Novel high-throughputscreening technologies. Genet EngNews 23:10–13.

Sohocki MM, Sullivan LS, Mintz-Hittner HA,Birch D, Heckenlively JR, Freund CL,McInnes RR, Daiger SP. 1998. A rangeof clinical phenotypes associated withmutations in CRX, a photoreceptortranscription-factor gene. Am J HumGenet 63:1307–1315.

Stiemke MM, Landers RA, al-Ubaidi MR,Rayborn ME, Hollyfield JG. 1994. Photo-receptor outer segment developmentin Xenopus laevis: influence of the pig-ment epithelium. Dev Biol 162:169–180.

Streisinger G, Walker C, Dower N, KnauberD, Singer F. 1981. Production of clones ofhomozygous diploid zebra fish (Brachy-danio rerio). Nature 291:293–296.

Streisinger G, Singer F, Walker C, KnauberD, Dower N. 1986. Segregation analy-ses and gene-centromere distances inzebrafish. Genetics 112:311–319.

Swain PK, Chen S, Wang QL, AffatigatoLM, Coats CL, Brady KD, Fishman GA,Jacobson SG, Swaroop A, Stone E,Sieving PA, Zack DJ. 1997. Mutations inthe cone-rod homeobox gene are as-sociated with the cone-rod dystrophyphotoreceptor degeneration. Neuron19:1329–1336.

Swaroop A, Wang QL, Wu W, Cook J,Coats C, Xu S, Chen S, Zack DJ, SievingPA. 1999. Leber congenital amaurosiscaused by a homozygous mutation(R90W) in the homeodomain of the ret-inal transcription factor CRX: direct ev-idence for the involvement of CRX inthe development of photoreceptorfunction. Hum Mol Genet 8:299–305.

Tsao-Wu GS, Weber CH, Budgeon LR,Cheng KC. 1998. Agarose-embeddedtissue arrays for histologic and geneticanalysis. Biotechniques 25:614–618.

Virchow R. 1855. Cellular-pathologie. Vir-chows Arch 8:1.

Westerfield M. 1995. The zebrafish book: aguide to the laboratory use of zebrafish(Danio rerio). Eugene, OR: University ofOregon Press.

ZEBRAFISH HISTOLOGIC MUTANT SCREEN 423