Essential Zebrafish Methods: Genetics and Genomics

Methods.

Haploid Screens and Gamma-Ray Mutagenesis

Charline Walker, Institute of Neuroscience, University of Oregon, Eugene, Oregon 97403-1254

Update

I. Introduction

II. Production of Haploid Embryos

A. In Vitro Fertilization

B. UV Irradiation to Make Haploid Embryos

III. Development of Haploid Embryos

IV. Genetic Background

A. Clonal Lines

B. AB Wild Type

C. *AB-Derived Wild Type

V. The Haploid Screen

A. Morphology Screen

B. Behavioral Screen

C. In Situ Screen

D. PCR Screen

E. Haploid Versus Diploid Phenotype

VI. Limitations of Haploids

VII. Mosaicism in F1 Screens

VIII. Haploid Screens and Gamma-Ray Mutagenesis

A. Gamma-Irradiated Blastula Stage Embryos

B. Gamma-Irradiated Sperm

C. Gamma-Irradiated Spermatogonia

IX. Nature of Gamma-Ray-Induced Mutations

A. Characterized Mutations

B. Translocations

X. Conclusion

References

Update

Producing haploid and gynogenetic diploid embryos continue to be effective ways to identify mutation-bearing females and to find new mutations in a forward genetic screen with efficient use of space and time.

I Introduction

Mutational analysis has been a productive way to dissect, at the cellular and molecular levels, pathways and key events of development. A number of large-scale mutagenesis screens have been conducted to identify developmental mutations in zebrafish (Driever et al., 1996; Haffter et al., 1996). In addition, smaller screens are ongoing in many other laboratories (Kimmel, 1989; Riley and Grunwald, 1995). Regardless of which type of mutagen is used, it is important to screen as efficiently as possible in terms of time, space, and the number of fish used. In this chapter, I describe the use of haploids to screen for early developmental mutations in mutagenized zebrafish, with an emphasis on the use of gamma rays as a mutagen. Screening haploids allow rapid identification of mutation-bearing females in a parental P or an F1 screen, avoiding the necessity of raising several generations of fish stocks prior to screening (Fig. 1).

Fig. 1 Haploid screens and gamma-ray mutagenesis strategies. (A) The primordial germ cells are targeted in the blastula stage embryo, mutagenized fish are raised to adulthood, and the haploid progeny of mutagenized females are screened for mutations. This parental screen is contrasted to the F1 screens shown in (B) and (C). (B) Mature sperm is collected in a small vial, mutagenized, and used to fertilize normal eggs. Haploid progeny of the resulting F1 females are screened for mutations. (C) Adult males are mutagenized, sperm from mutagenized spermatogonia are used to produce an F1 generation, and haploid progeny from female F1 adults are screened for mutations. The mutagen used in each case is gamma-ray irradiation, but any mutagen can be used in these screening strategies.

Gynogenetic haploid embryos are produced when eggs are fertilized by UV-irradiated sperm; the resulting haploid embryos develop solely from maternal genetic information. Although the UV-irradiated sperm provide no male genetic contribution to the embryo, they are necessary to activate embryonic development. Because recessive mutations are no longer masked by diploidy, the use of haploids is a very powerful tool to identify newly induced mutations in a mutagenesis screen. A relatively lethal-free genetic background is particularly important for a haploid screen.

Haploid embryos live about 4 days postfertilization (dpf), a stage at which their diploid counterparts are free swimming and feeding larvae; this allows screening for mutations that affect a large number of developmental processes. However, because haploid development is not completely normal, there are limitations to their use in certain types of screens.

Developmental mutations uncovered in a haploid screen are then characterized in diploids. It is reassuring that, in the majority of cases, the homozygous diploid phenotype is very similar to the original haploid phenotype.

Gamma rays are an effective mutagen in producing early developmental mutations (e.g., Fritz et al., 1996; Grunwald et al., 1988; Halpern et al., 1993; Talbot et al., 1998). They generate a whole range of lesions from just a few base pair changes to chromosomal rearrangements and large deletions (Chakrabarti et al., 1983; Fritz et al., 1996; Kubota et al., 1992; Russell and Russel, 1959; Talbot et al., 1998). Here I present results from mutagenesis studies comparing gamma irradiation at different stages: the 1000-cell blastula, mature sperm, and spermatogonia, with the aim of producing recoverable mutations in the germline of zebrafish.

II Production of Haploid Embryos

Zebrafish eggs can be fertilized in vitro, which makes it possible to manipulate the ploidy of the resulting embryos. One can produce gynogenetic haploid, gynogenetic diploid, triploid, and tetraploid embryos using techniques developed by Streisinger et al. (1981). Ploidy can be verified by preparing chromosome spreads from 24-h whole embryos (Westerfield, 1995) as shown in Fig. 2, which compares a haploid to a diploid chromosome spread. Androgenetic haploids can also be produced by using normal sperm to fertilize eggs irradiated with large doses of gamma rays, X rays, or UV light. Gynogenetic and androgenetic haploids look virtually the same (Corley-Smith et al., 1996, 1999).

Fig. 2 Chromosome spreads from 24-h whole embryos. (A) Haploid spread, 1n = 25 chromosomes; (B) diploid spread, 2n = 50 chromosomes.

A In Vitro Fertilization

1. Females are separated from males late in the afternoon before the day of the experiment so that they will not lay their eggs early.

2. On the morning of the experiment, sperm is collected from males. Males are anesthetized in tricaine (0.2 mg/ml 3-aminobenzoic acid ethyl ester, methane-sulfonate salt, brought to pH 7 with 0.4 mg/ml Na2HPO4) until rapid gill movements slow down, rinsed, and placed upside down on a damp sponge. The genital region is blotted dry with a tissue. After parting the anal fins, sperm is extruded by gentle pressure and stroking on the sides of the abdomen with smooth forceps and collected in a 20-μl microcapillary. The amount of sperm collected per male varies (on average 1–5 μl); sperm is collected and pooled in Hank’s saline (Westerfield, 1995) until the solution is cloudy. Sperm held in Hank’s on ice (4 °C) maintains the capacity to fertilize eggs efficiently for at least 90 min. Usually there is a great excess of sperm. At this point, sperm can be inactivated by UV irradiation as described later.

3. Starting at dawn in the fish light cycle, each female is similarly anesthetized, rinsed, blotted on a paper towel, and laid in a 35-mm plastic dish. Eggs are manually extruded or squeezed from the female by using one finger to brace the back, and gently pressing the belly with a finger of the other hand. Eggs should come out easily, and after using a spatula to move the eggs away from the female, she is placed into a recovery container. Care should be taken to avoid excessive wetness, because when water contacts an egg, the chorion swells and the egg can no longer be fertilized.

4. The eggs are immediately fertilized by adding about 25 μl sperm in Hank’s and 0.75 ml water. After about 1 min, more water is added to fill the dish three-quarters full. Sperm in Hank's saline remain quiescent, but as soon as the saline is diluted with water, the sperm are activated and able to fertilize eggs, remaining active for only about 1 min. Thus, all the eggs in a clutch are fertilized essentially synchronously. Sperm in Hank’s can be gently mixed with eggs before adding water. In this case, the time of fertilization is when water is added. Premature contact of sperm or eggs with water should be avoided.

Besides the successful females which give good eggs, other females may not be prepared to release eggs, and they should not be pressed too hard. Still other females may give eggs which have already broken down (bad eggs), and these eggs are not worth fertilizing.

Both male and female fish are significantly stressed when handled in an in vitro fertilization experiment. This can lead to illness or poor production of gametes. We find that by allowing males a minimum of 3 weeks and females a minimum of 4 weeks to recover before using them again, the fish are more likely to continue producing eggs and sperm. They can be set up for natural matings during this waiting period.

B UV Irradiation to Make Haploid Embryos

UV-irradiation cross-links DNA. When sperm is UV irradiated, the DNA is destroyed although the sperm retains the ability to activate the egg. The haploid embryo, resulting from an egg fertilized by UV-irradiated sperm, has no genetic contribution from the male. A 0.5-ml aliquot of freshly collected sperm in Hank’s saline is spread in a watch glass that sits on top of ice (ice should never touch the sperm) and is exposed for 2 min to UV light from an 18 in. (43 cm) Sylvania germicidal lamp (254 nm) at a distance of 15 in. (38 cm) from the watch glass. After UV irradiation, the sperm is put into a clean vial and kept on ice. UV sperm is used just like normal sperm to fertilize freshly squeezed eggs.

Care must be taken to shield oneself from UV by wearing safety glasses and latex gloves. Sperm should not be too concentrated, and solid material such as feces should be removed, to prevent shielding the sperm from the UV source.

• Many labs have a Stratalinker which can be used as an alternative to a germicidal lamp to UV-irradiate sperm. An aliquot of sperm (up to 0.5 ml) suspended in Hank’s saline is spread on a watch glass sitting on ice as before. A dose of 900–1200 J gives results comparable to a germicidal lamp.

III Development of Haploid Embryos

Haploid embryos live for about 4 dpf. The schedule of early development is approximately the same for haploids and diploids. Interestingly, whereas early cleavage cycles are the same, haploids enter midblastula transition (when cells slow their division rates and zygotic transcription begins) one cleavage cell cycle later than diploids (Kane and Kimmel, 1993). Thereafter, haploids contain more numerous and smaller cells than diploids (especially evident for pigment cells).

The smaller cells, in addition to gene haploinsufficiency, probably contribute significantly to the inviable haploid syndrome (Fig. 3) which is fairly consistent from embryo to embryo: the haploid body is short and stocky, yet has the appropriate number of muscle segments. Morphogenesis of the brain is abnormal, the brain looks less clearly defined and has more cell death than in diploids. From dorsal view, the brain folds or kinks from side to side; possibly this defect is related to the general body shortening. The eye is incompletely formed at the choroid fissure; the otic vesicles are always present but may be duplicated on one or both sides of the body. Blood cells seem too large for the blood vessels, and most haploid embryos have circulation problems. As development proceeds, certain parts of the body such as the pericardial cavity start swelling. Eventually, there is generalized edema, and the embryos die. However, as you can see in Fig. 3, all the major body parts can be identified in haploids, including features such as the notochord, floor plate, heart, and blood.

Fig. 3 Comparison of day 1 and day 2 haploid and diploid embryos, lateral views. Anterior is to the left, and dorsal is to the top. (A) Day 1 haploid; (B) day 1 diploid. Details of trunk region in (C) day 1 haploid and (D) day 1 diploid. (E) Day 2 haploid; (F) day 2 diploid. Head detail of (G) day 2 haploid and (H) day 2 diploid. cc, central canal; fp, floor plate; n, notochord; y, yolk extension; e, ear; m, melanocyte; hc, heart cavity; hgc, hatching gland cells.

Another feature of normal haploid clutches is that in each batch there are a variable number of defective embryos, easily distinguished from the well-developed haploids (Fig. 4). Because deleterious genes are unmasked in haploids, the genetic background of the fish influences the frequency of defective embryos. Egg quality has an effect as well, and if the mother is unhealthy, the haploids she produces do not develop well. Such defective embryos are easily recognizable, and when screening a batch of haploids, it is useful to sort embryos into groups based on how well formed they are. We sort embryos into three categories, although the severity of defects occurs along a continuum. Table I shows, in various backgrounds, the proportion of embryos belonging in each category: A or normal-looking embryos; B embryos, which have a twisted, short axis but still have a recognizable head and tail; and C embryos, which are bits of tissue on a live yolk. Figure 4 shows examples of each category.

Fig. 4 Examples of day 1 haploid embryos from a single clutch showing the categories of good-looking and defective embryos typical of haploids. The proportion of A embryos in the entire clutch is higher than shown. (A) Unsorted haploids; (B) sorted A embryos; (C) sorted B embryos; (D) sorted C embryos.

Table I Clonal and Selected Wild-Type Strains Have More Than 50% A Haploid Embryos

a The clonal strains are homozygous and embryonic lethal free (see Section IV).

b The AB strain is the semiselected Oregon wild-type line (see text).

c The *AB strain is the gynogenetically derived wild-type line currently being used in the Oregon mutagenesis screens.

During a morphological screen, it is sometimes difficult to decide whether a few similarly defective embryos are mutant or not, but the variability of the haploid defective embryos helps to distinguish them from real mutant embryos, which usually have a characteristic phenotype.

Within their limitations, haploids are consistent enough to detect mutations and have been very useful in finding mutations which affect early development. Haploids may not be appropriate for some screens (see Section VI); for example, a screen for phenotypes later than day 3 would have to be done in diploids. However, haploid analysis has the advantage that mutant progeny are represented in the same proportion as is the mutation in the gametes of the mother.

IV Genetic Background

A consistent background free of lethal mutations is advantageous for genetic screens; there is less chance of interactions with preexisting mutations, and one can be confident that a new mutation has been induced by the mutagen. A good genetic background is particularly important for a haploid mutagenesis screen because it is beneficial to have a high proportion of A embryos to screen.

A Clonal Lines

In the 1970s, George Streisinger produced lethal-free clonal lines of fish with the idea that the ideal genetic background would be one in which all the fish were identical to each other prior to mutagenesis. These lines were gynogenetically derived first by making fish homozygous diploid with heat shock (HS) which inhibits the first mitotic division, then by making identical clonal progeny from a single homozygous female with early pressure (EP) which inhibits the second meiotic division of the eggs (Streisinger et al., 1981).

To produce homozygous fish, eggs are fertilized with UV-irradiated sperm to eliminate the male genetic contribution. During the first mitosis, the maternal chromosomes replicate. Heat shock prevents the separation of the replicated chromosomes and prevents the first cell division, establishing diploidy to the embryo. Subsequent replications and cell divisions are allowed to proceed (method described in Corley-Smith et al., 1999).

Fertilization triggers the start of the second meiotic division of the egg. A diploid number of condensed chromosomes align on a spindle in preparation for the meiotic reduction division which normally results in a haploid pronucleus and a polar body which is extruded. High-hydrostatic pressure applied at this time breaks down the spindle and keeps the full diploid complement of chromosomes together. With no further intrusion, a high percentage of embryos (60–90%) go through normal development. Specifically, eggs are fertilized with UV-irradiated sperm then immediately put in a small glass vial capped with a rubber sheet. At 1.4 min postfertilization (mpf), 8000 lb/in.² is applied by a French press. At 6 mpf, pressure is relieved, and the eggs are allowed to continue developing without disturbance for 3–4 h.

The first batches of homozygous fish produced from the wild-type strain were of very poor quality, but by crossing the best homozygous males and females and performing several successive generations of homozygosing, clonal lines were produced which were nearly as good as wild-type strains in terms of viability and fecundity.

Gamma-ray mutagenesis screens were conducted with clonal lines for a number of years, but as the scope of the screen increased in size, the lack of vigor of these fish proved to be a problem; the clutch sizes were too small (<100 eggs), and the quality of the eggs was often poor, resulting in too many defective embryos.

B AB Wild Type

Turning to the nonhomozygous Oregon wild-type strain, AB, from which the clones were derived, we noticed that some females produced beautiful haploid embryos. This line had already gone through a preselection process to enrich the stock with females that produced more than 50% A haploid embryos by discarding females that produced poor-quality haploids. Males were unselected. Once established, the AB strain was kept closed for over 50 generations; however, at each generation, gametes from many individuals (20–30 females and 60 males) were mixed in an overt attempt to maintain heterozygosity within the closed line. Clutch sizes from AB females are over double that of clonal females (Table I).

C *AB-Derived Wild Type

Unlike the homozygotes, the AB strain was not lethal free, and the haploid screen revealed a number of background mutations with visible embryonic phenotypes which interfered with the screen. In 1991–1992, by a single round of stringent selection we produced a new line, called *AB, from gynogenetic diploid progeny of 21 AB founder females. Haploid embryos from 106 AB females were put through the entire 3-day morphological mutagenesis screen (see Section V). Out of those, the selected group of 21 females produced haploid embryos which were normal at all screening points and which had 50% or more A embryos at 24 h. Gynogenetic diploid progeny were made for each of these females by the EP technique described earlier. The EP progeny of the 21 selected females were male and female. Crossing all possible combinations of males and females together established the *AB line. Although this line has a limited gene pool, care is taken to keep it as diverse as possible, so like the AB line, 20–30 females are crossed to about 60 males each generation to conserve the hybrid vigor within the population. Haploids from the *AB line have an improved genetic background as seen by the higher percentage of A embryos in Table I.

The *AB strain is not homozygous nor is it necessarily lethal free; it has been selected to produce good haploids for the first 3 days of development.

V The Haploid Screen

Our haploid screen covers the first 3 days of development at 28.5 °C. We are interested in mutations affecting early development, especially the patterning of embryonic mesoderm and nervous system.

Eggs are squeezed from P or F1 potentially mutation-bearing females and fertilized in vitro with UV-irradiated sperm to make haploids (Fig. 1). Each female is kept in a 1-l individual container until the screen is completed. After a few hours, fertile eggs are sorted into embryo medium (Westerfield, 1995) containing 100,000 units penicillin and 100 mg streptomycin/l, at about 100 embryos per 100-mm plastic Petri dish. (We do not use antibiotics to raise stocks of fish, but they help the haploid embryos survive the entire screening procedure.). At 24 hours postfertilization, dead embryos are counted and removed. The remaining embryos are taken through a screening regimen which is described in detail later. Females carrying potential mutations of interest are then outcrossed with wild-type *AB males for recovery of the mutation.

Our approach is to combine several screening methods to increase the probability of finding an interesting mutation. The morphological screen and the early qualitative behavioral screen pick up de novo mutations; the RNA in situ hybridization screen and the PCR-based screen find mutations affecting known genes.

A Morphology Screen

The morphological screen is conducted using a dissecting microscope. Haploid embryos at four developmental stages of the morphological screen, tailbud, day 1, day 2, and day 3, are shown in Fig. 5.

1. At the tailbud stage (10–12 h), the cells in the embryo are just completing gastrulation movements (Kimmel et al., 1995). Mutations affecting epiboly, axis, and cell cycle can be distinguished at this time.

2. On day 1 (24–30 h, or the pharyngula stage), the embryo has formed many of its primary organs. At this time, we screen for morphological features including: body shape, eyes, ears, hatching gland, notochord, floor plate, somite shape and approximate number, and cell death.

Fig. 5 Haploid embryos at key stages of the morphological screen; anterior is to the left, dorsal is to the top. (A) Tailbud stage; (B) day 1; (C) day 2; (D) day 3.

Screening is made easier by sorting the A, B, and C embryos into separate groups so that your eye can pick out patterns. Mutants usually have a characteristic phenotype, and this distinct look helps to distinguish mutants from variably defective haploids (Fig. 4).

3. At day 2 (48–54 h), the embryos are screened for heartbeat, blood, pigment cells (melanophores and xanthophores), pectoral fins, new cell death, and body shape.

4. By day 3 (72–78 h), some of the haploid embryos have begun to swell, but we can screen the good-looking haploids for jaw defects, the presence of Mauthner cell axons, and the presence of neuromasts of the lateral line; the two latter screens are conducted with a compound microscope.

B Behavioral Screen

Although haploids never perform normal swimming movements, we can screen at earlier times for motility and sensory defects. The following early behavioral screens test motility in 24-h embryos and sensory reflexes in 30-h embryos, the tests are noninvasive and are conducted using a dissecting microscope.

Haploid embryos, like diploids, produce spontaneous flexing movements at 24 h of development, which later cease (Grunwald et al., 1988). It is easy to look systematically through a batch of 24-h A embryos for ones that do not flex. This lack of behavior could be caused by problems with neurons, muscles, or neuromuscular junctions. An example of a mutant which failed this test is the recessive lethal mutation fibrils unbundled (fub1b45) (Felsenfeld et al., 1990). Embryos homozygous for the fub1 mutation have unstriated skeletal muscles. The myofibrils in somitic myotubes are severely disorganized; the major contractile proteins are present, but not properly aligned. These fish barely move during the wiggle test.

A test of the early sensory system can be performed at about 30 h. Haploids, like diploids, respond reflexively with a sharp flex of the body to light touches on the head or body with a tiny probe of nylon monofilament. This screen detects mutants that block or alter development of the early functional sensory-motor pathway. We did not identify any mutants using this screen on several thousand embryos. Nevertheless, such a screen holds promise because the hyperactive mutants techno trousers (tnt) and roller coaster (roc) were identified in another touch test in the Germany diploid mutagenesis screen (Granato et al., 1996).

C In Situ Screen

RNA in situ hybridization can be used efficiently to screen for mutations that affect the normal expression pattern of known genes; these mutations may be in the gene itself or in an upstream gene. Twenty embryos are fixed in 4% paraformaldehyde at each of two developmental stages, 10 and 22 h. The embryos are processed essentially following the in situ hybridization protocol of Oxtoby and Jowett (1993).

For the tailbud screen, embryos are fixed at late gastrula, about 10 h, and are incubated with a mix containing riboprobes for a number of genes whose expression patterns do not overlap or interfere with each other. At present we examine the expression pattern of the following genes which mark the following structures: hatching gland1 (hgg1) in the prechordal plate mesoderm, floating head (flh) in the epiphysis and notochord, valentino (val) in hindbrain rhombomeres 5 and 6, forkhead6 (fkh6) in presumptive neural crest, hairy-enhancer split-related1 (her1) in trunk and tailbud, and paxb in the presumptive midbrain-hindbrain junction (Fig. 6, hggl and paxb expression not shown).

Fig. 6 In the haploid in situ hybridization screen, a cocktail of riboprobes reveals patterns of gene expression at late gastrula stage (A), and at 22 h (B)

(photos courtesy of Sharon Amacher).

A second set of embryos, fixed at 22 h, are screened with the following probes: sonic hedgehog (shh) in floor plate and forebrain, myoD in the somites, distalless2 (dlx2) in developing pharyngeal arches 1, 2, and 3, krox20 in hindbrain rhombomeres 3 and 5, engrailed3 (eng3) at the midbrain-hindbrain border, and lim5 in the diencephalon (Fig. 6).

The mutation valentinob361 was identified in the 22-h screen because it alters the krox20 expression pattern. Normally krox20 is expressed in two very sharp bands of cells in the position of rhombomeres 3 and 5 in the hindbrain. In valb361, the krox20 expression in r5 is reduced whereas expression in r3 is normal (Moens et al., 1996). In this case, a mutation in an upstream gene, valentino, modified the expression of the known gene, krox20. The valentino gene was cloned molecularly by homology to mouse kreisler (Moens et al., 1998), and a probe for that gene is now used in the early screen to mark the r5 and r6 territory.

D PCR Screen

A polymerase chain reaction-based (multiplex PCR) assay is used to screen individual haploid embryos for deletions of known gene sequences (Fritz et al., 1996).

DNA is isolated from approximately 10 haploid embryos per clutch at 24 h. Combinations of the following PCR primers are used to amplify specific DNA sequences from each embryo: axial, dlx2, dlx3, elrC, eng1, eng2, goosecoid, hoxb4, hoxb5, hoxb6, msxA, msxB, msxD, noggin, tenascinC, shh, hoxc3. An F1 female which produces a clutch of embryos where one or more embryos are missing a specific PCR amplification product is likely to bear a mutation in that gene. The mutation T(msxB)b220 was identified because two of ten haploid progeny from the original F1 female were missing the msxB PCR product (Fritz et al., 1996). Further analysis showed that the mutation was the result of a translocation and that other genes in addition to msxB were missing. Figure 7 shows another example where three of nine haploid progeny were missing the goosecoid PCR product.

Fig. 7 The PCR-based screen uses a specific mixture of compatible primers to amplify gDNA sequences of randomly selected individual haploid embryos of potential mutation-bearing females. Here, three out of nine haploid embryos from a single female are missing the bottom goosecoid band. PCR products used in this reaction, from the top band down, are krox20, msxD, hoxb4, msxA, eng2, hoxc3, dlx3, dlx2, and goosecoid (gsc)

(photo courtesy of Andreas Fritz).

E Haploid Versus Diploid Phenotype

Usually the phenotype of a mutant in a haploid is strikingly similar to the homozygous diploid phenotype. The tail is always shorter, and the phenotype is more severe in a haploid; two such mutants are shown in Fig. 8C-F.

Fig. 8 Three mutants recovered from gamma-ray mutagenesis screens, as a haploid on the left and a diploid on the right. The mutation b225 is likely to be a translocation, and there are two phenotypes in a clutch of embryos from a mutation-bearing female. (A, B) b225 phenotype 1 would be considered a neural degeneration (ned) phenotype (also see Table II). (C, D) b225 phenotype 2 would be classified in the tail category although there is also considerable cell death. (E, F) cerebumb305 would also be put in the tail category

(cerebumb305 photos courtesy of Marnie Halpern).

Exceptions to this rule are the mutations producing neural degeneration (ned) haploid phenotypes. Sometimes these mutants can look very different as diploids, usually being less severe, and they may not be necrotic at all. An example of a ned mutation as a haploid and a diploid is b225 phenotype 1 (Fig. 8A and B).

VI Limitations of Haploids

Because of irregularities caused by the haploid syndrome discussed earlier, it is not possible to screen for all features in haploids. Because, for example, haploids may have more than one otic vesicle on each side, and otoliths may be absent, they can be screened for mutations that block ear development completely, but not for mutations that perturb subtleties of how it develops.

In diploids, a single pair of Mauthner neurons develops in the hindbrain, extending their axons into the spinal cord where they can be recognized uniquely in live embryos by their large diameters (Kimmel et al., 1978). With a background of less than 1%, well-formed haploids always have Mauthner neurons that project into the spinal cord, as we determined from Nomarski screens involving 4600 embryos. However, the axon diameters are smaller than normal and axons run along an incorrect spinal pathway in more than 10% of well-formed haploids, and in about the same fraction there are one or more extra Mauthner cells. Hence, one can use haploids to identify a mutation that deletes the Mauthner neuron or that changes the course of the Mauthner axon so that it does not enter the spinal cord, but not for mutations that duplicate the cells or change the axonal pathway within the spinal cord.

In haploids, morphogenesis of the brain is abnormal as previously described. Identifying details of brain structure are problematic in the morphology screen; however, the expression domains of markers such as krox20 or engrailed3 in the in situ screen are quite consistent in haploids and allow for screening developing regions of the haploid brain. In fact, the valentino mutation, which alters krox20 expression in rhombomere 5, also has a morphological defect which a trained eye can pick out in diploid embryos, but it is unlikely that this mutation would have been found in the haploid morphology screen. This is an example of how a subtle phenotype can be found in a haploid, using the appropriate