Establishing Developmental Genetics in a Self-fertilizing Fish (Krytolebias marmoratus)

Abstract

Kryptolebias marmoratus is a synchronous hermaphroditic vertebrate that utilizes an ovotestis for reproduction. This fish develops externally, is easy to maintain, and has about a 100-day life cycle, making it a desirable developmental genetic model organism. Here, we present a pilot zygotic mutant screen utilizing the common chemical mutagen, N-ethyl-N-nitrosourea (ENU) to establish genetics in this model species. Selection of clonal stocks and optimal conditions for mutagenizing this fish are presented and the types and frequencies of zygotic mutants are documented in comparison to other fish models. Kryptolebias marmoratus is an exemplar model organism that will complement future developmental genetic screens in vertebrates.

Introduction

Forward genetic screens are fundamental for determining the functions of genes, uncovering cellular and developmental pathways, investigating evolutionary-developmental patterning, and constructing models of human disease. Although recent developments in targeted mutagenesis have advanced in some model organisms such as transcription activator-like approaches (Bogdanove and Voytas 2011), chemical mutagens remain the mainstay for large-scale genetic screens due to loci-random distribution and high mutation rate.

In contrast, chemical mutagens or transposon insertion-tagging may also be used selectively by the reverse genetic technique to identify putative mutant alleles in a gene or genes of interest (at the genotype level) prior to genetic observations of mutant phenotypes. However, reverse genetic techniques require knowledge of the gene(s) or genome under study and are limited in their genetic utility. For example, loss-of-function mutations (amorphic alleles) may be identified through reverse genetic approaches, but the ability to screen for functional allelic variants affecting particular phenotypes among multiple genes using the forward genetic technique is immensely more powerful. Indeed, missense mutations such as hypo-, hyper-, or neo-morphic alleles have classically been identified through forward genetic or genetic modifier screens prior to their characterization at the molecular level.

The chemical mutagen, N-ethyl-N-nitrosourea (ENU), is an established and potent agent of choice for forward or reverse genetic screens in many organisms. For example, in the detection of genes required for early development and patterning of the vertebrate embryo, ENU has successfully been employed in genetic screens for zebrafish (Danio rerio) and medaka (Orzyias latipes) (Haffter et al. 1996; Loosli et al. 2000).

The arrival of new model organisms for exploring different biological sub-disciplines has surged due to the unique biological features of each model and the accessibility of applicable cellular and molecular techniques such as genomics (Crotty and Gann 2009). Kryptolebias marmoratus, hereafter referred to as K. mar., is a unique oviparous vertebrate that primarily reproduces by internal self-fertilization via a mixed gonad, the ovotestis (Harrington 1961; Cole and Noakes 1997). K. mar. has an extensive history as a model organism, first described >50 years ago by Harrington and Rivas (1958). For example, the ecology and remarkable life history of this organism and its population genetic structures have been widely investigated over that last several decades (see Taylor, Tatarenkov et al.; both in this issue). In the laboratory, K. mar. has been developed as a model for cancer and ecotoxicology (reviewed by Lee et al. 2008); physiology, toxicology, early vertebrate development, and behavior (see Wright, Bielmyer et al., in this issue; Mourabit and Kudoh, in this issue; Chang et al., in this issue).

Harrington and colleagues also made the earliest observations on K. mar.’s egg-laying patterns and developmental stage upon oviposition and demonstrated through tissue grafts that inbreeding produced isogenic offspring (Harrington 1963; Kallman and Harrington 1964; Harrington and Kallman 1968). These isogenic lines were referred to by Harrington as “clones,” a term that at present remains the general descriptive for isogenic K. mar. stocks.

Recently, 21 clonal-isogenic laboratory stocks were established within the scientific community through microsatellite and mitochondrial DNA analysis (Tatarenkov et al. 2010). This is a valuable genetic resource similar to common laboratory strains of the hermaphroditic nematode Caenorhabditis elegans because the variability in the genetic background of these organisms at the DNA level is reduced. Genomic isogeny potentially makes the identification of causative mutations easier to identify because the level of background polymorphisms is reduced. For example, contemporary whole-genome sequencing techniques are readily used for identifying mutations among multiple C. elegans strains by filtering out the known genetic polymorphisms (Flibotte et al. 2010). Furthermore, starting single nucleotide polymorphisms unique among strains prior to mutagenesis can be used to map new mutations by genetic linkage studies.

Therefore, K. mar. is poised as an alternative vertebrate model for genetic study of multiple fields of comparative biology and its genomic isogeny is a unique and important feature. Moreover, a considerable reduction in resources and time are realized when working with a vertebrate hermaphrodite; the equivalent of an entire family of individuals for maintenance of a genetic stock can be maintained in one individual of K. mar. K. mar. biology is also exceptional among vertebrates to include (to name a few) its mode of sex determination, highly adaptive physiology, phenotypic plasticity in both behavior and development, and ability to undergo embryonic diapause similar to other organisms. Yet, despite its use as a model for multiple fields of study, there have been no reports on genetic investigations and most researchers in the scientific community are not aware K. mar. even exists as a potential comparative vertebrate model.

Here, we describe the first forward genetic screen in K. mar. and demonstrate the efficacy of ENU mutagenesis to identify zygotic mutations affecting early development. We aimed to optimize breeding conditions and identify appropriate laboratory clonal stocks for mutagenesis and compare our genetic results to those of previous screens in other fish models to establish developmental genetics in K. mar.

Materials and methods

All experiments presented here were approved by the Valdosta State University Institutional Animal Care and Use Committee (AUP-00023-2009) under the National Institutes of Health’s Office of Laboratory Animal Welfare (Assurance Number A4578-01).

Fish husbandry

All developmental stages of K. mar. were maintained in an Aquatic laboratory maintained at 28°C on a 14 h light: 10 h dark photoperiod. Larvae and adult fish were housed in either 1 L plastic food containers or 2 L breeder tanks (14-955-118A, Fisher Scientific; LBTANK, Aquatic Habitats) in a brackish water solution (17 ppt Instant Ocean) with biweekly water changes. Embryos were collected from the breeder tanks by first removing the adult hermaphroditic parents using the basket inserts followed by retrieval with 3-mL disposable pipettes. Embryos were grown and observed with student-style dissecting microscopes in plastic petri dishes (100 × 15 mm) containing brackish water until hatching and then transferred to 1 L plastic food containers. Juvenile fish were fed freshly hatched Artemia nauplii (Artemac USA) harvested from a commercial grade hatchery funnel daily (BS6, Aquatic Ecosystems).

ENU mutagenesis

A total of 34 fish of similar size and age were treated (30–35 mm from the mouth to the posterior margin of the peduncle muscle of the tail fin; >1 year old). Selected stock fish were mutagenized on two separate weeks with varying concentrations of ENU (Sigma N3385). The first week of treatments (round one) included the control 0.0 mM versus 3.3 mM ENU treatment group which was monitored carefully for a month post treatment so that modifications to protocol might be made prior to the second week of treatments (round two) at different concentrations of ENU (2.8, 3.0, 3.8 mM). We followed established zebrafish protocols for ENU mutagenesis with minor modifications (Mullins et al. 1994; Westerfield 2007). Since K. mar. are maintained in brackish water and mutagenesis was performed in zebrafish water (60 mg/L Instant Ocean), fish were first acclimated in 500 ml of zebrafish water for 1 h followed by a 1 h exposure to the appropriate ENU concentration (of 0.0 control, 2.8, 3.0, 3.3, or 3.8 mM) made by diluting the ENU stock in 300 mL of fresh pH 6.6 mutagenesis buffer (10 mM Phosphate Buffer in zebrafish water). All ENU stock solutions (range of 85–106 mM) were quantified prior to dilution at an OD₂₃₈ using a Beckman DU 600. To remove the ENU from the fish, two subsequent washes were performed for 1 h each in 500 mL zebrafish water. Fish were then transferred back to individual 2 L breeder tanks containing brackish water and transported to the aquatic laboratory for normal feeding for 48 h between ENU treatments. The above ENU protocol was repeated three times for 1 week, every other day; constituting a round of mutagenesis. All four transfers of fish during the above protocol were performed in 1 L breeder tanks (SBTANK, Aquatic Habitats) by placing four to seven fish in the first acclimation tank and simply transferring fish with the lid and basket across the remaining three treatment tanks. In addition, all four transfers contained 2 mg/L Tricaine (MS-222) to help calm and reduce mortality in a quiet, dedicated hood at room temperature. An equal volume of ENU neutralization solution (20% NaSO₄, 1% NaOH) was added to all solutions or materials contacted by ENU for 24 h prior to disposal.

Photography of embryos

The chorion of K. mar. is rather difficult to dissect away from developing embryos. Chorions were removed from embryos by the hatching-enzyme method adapted from medaka (Mourabit et al. 2011). Digital images were taken with an Olympus DP72 camera mounted to an Olympus SZX16 stereomicroscope.

Results

A pilot genetic screen was designed to identify early embryonic developmental defects (zygotic mutants) in K. mar. as proof of the principle that ENU is an effective mutagen for this species. Prior to mutagenizing K. mar., we optimized fecundity for rearing and growing fish and selected stocks based on certain criteria. During the mutagenesis procedure, the effects of ENU treatments on overall production of embryos and parental fertility across all ENU treatment groups over a 10-week period were monitored. Finally, the zygotic mutants identified in the pilot screen were rescreened to confirm and establish stock lines as part of a larger ongoing genetic screen.

Optimizing fecundity

K. mar. display considerable variability in the number of eggs it lays (oviposits) by self-fertilization in a week or even through cyclical periods of time (Harrington 1963; Grageda et al. 2005). Juvenile fish prefer fresh rotifers or brine shrimp, Artemia nauplii (Pandey et al. 2008), but it is convenient to feed both juveniles and adults brine shrimp ad libitum. We sought to increase egg production consistently from our adult fish to maximize numbers of offspring in our genetic screen and reduce subsequent generation times. We hatched brine shrimp (to 350 ± 10 nauplii/mL) and fed adult fish 1 mL versus 10 mL daily and recorded overall egg production (n = 10 fish). We found that by increasing the food 10-fold we could increase egg production per week by two-fold (Fig. 1A). Based on these results, we fed a high-food regimen of either approximately 350 or 1000 brine shrimp daily, to juvenile or adult fish, respectively (i.e., 1 mL or 3 mL of ∼350 nauplii/mL).

Fig. 1.

Optimizing fecundity and selection of stock fish for mutagenesis. (A) Ten K. mar. fish of 32.8 ± 4.3 mm standard length were fed low food, then high food (see ‘Materials and methods’ section). The mean daily number of embryos produced by all fish was normalized by first dividing the total number of eggs produced by the standard length of the parent fish and then again by the number of collections (standard error included). Asterisk indicates Paired t-test indicates significant difference between low versus high feed (P = 0.004). (B) Example of stock selection based on average daily embryos produced by various stock fish under high food conditions (normalized as in A above). Asterisk indicates Hon9 stock produced a significantly greater number of eggs than did the other clones examined by multiple pair-wise comparisons except for 50.91 stock (Tukey-Kramer multiple comparisons test, P < 0.005). Both Hon9 and 50.91 stocks were used for mutagenesis.

Parental stock selection for mutagenesis

Isogenic lineages or stocks were first selected on the basis of the genetic composition of animals surveyed from multiple laboratories around the world (Tatarenkov et al. 2010). Briefly, the Hon9, Hon11, and 50.91 stocks at Valdosta State University were selected based on (1) complete homozygosity across all 32 microsatellite markers tested, (2) slight genotypic differences among the three stocks for future molecular genetic discrimination of stocks, (3) commonality of stock fish among laboratories, and (4) sociability (low levels of aggression in group tanks). Stocks were further selected on the basis of high fecundity. For example, 50.91 and Hon9 were selected because of a higher average number of embryos laid per week versus other stock fish under the same high-food regimen (Fig. 1B). More importantly, the final set of selected stocks for mutagenesis did not exhibit embryonic lethal background mutations.

Pre versus post ENU treatment effects on parental fish

In order to optimize and establish a consistent protocol for ENU mutagenesis in K. mar., it was critical to measure the effects of ENU treatment throughout the screen. First, we measured the overall daily egg production of both Hon9 and Hon11 parental fish (n = 22) for 10 weeks prior to 3.3 mM ENU treatment (round one of mutagenesis, see Table 1 and ‘Materials and methods’ section). Next, we measured the overall daily egg production by these fish for 10 weeks after ENU treatment. We observed that production did not change significantly pre versus post treatment for the Hon9 stock, while it actually increased for the Hon 11 stock (Fig. 2). Therefore, the ENU treatment did not decrease the overall egg production in our hermaphroditic parental fish, which allowed us to set up a second round of ENU mutagenesis using fewer parental fish (under varying concentrations, see Table 1 and ‘Materials and methods’ section).

Table 1.

Effects of ENU treatments on fertility of parental fish

N	P Stock	mM ENU	F1 Embryos	Percent fertile
6	Hon9	0.0a	986	92
4b	Hon9	2.8c	790	70
6b	50.91	3.0c	1219	67
6	Hon9	3.3a	935	64
16	Hon11	3.3a	3062	54
4	Hon9	3.8c	1312	66

N = total number of fish treated.

^aRound one of ENU treatments (see ‘Materials and methods’).

^bOne P fish died in each of these treatment groups.

^cRound two of ENU treatments (see ‘Materials and methods’).

Fig. 2.

ENU effects on weekly egg production by parental fish, Hon9 and Hon11 (round one). Standard error shown for the mean number of eggs collected weekly (replicated collections in parenthesis). Asterisk indicates Hon11 produced significantly more eggs after treatment (P = 0.008).

Next, we analyzed the fertility rate of both rounds of mutagenesis across 10 weeks post ENU treatment and compared them to untreated controls (Table 1). Fertility is defined as the percentage of overall embryos fertilized and is detectable by the presence (fertilized) or absence (nonfertilized) of the perivitelline membrane space easily identified under a dissecting microscope upon collection of embryos. This is different from our initial observations of overall egg production in round one treated fish described above and is important because it is the fertilized embryos that are capable of survival that form the next generation. The overall average fertility across all treatments groups averaged ∼30% less than the control, untreated group (Table 1). We conclude that ENU reduces overall production of viable sperm in our hermaphroditic fish leading to lower rates of fertility, as predicted from previous initial screens in zebrafish where male fertility is reduced during the first several weeks after ENU treatment (Mullins et al. 1994).

To further assess the effects of ENU on fertility of the parental fish, we compared fertility by treatment-group versus week of egg deposition across the 10-week post ENU treatment period (Fig. 3). Beginning in Week 1 post treatment, we collected all embryos from treated fish and recorded fertility as described above. We only found zero fertilized F1 fish from Week 1 post treatment and of the few fertilized embryos from Weeks 2 to 3 of collections, only 24 fertilized embryos survived to hatching. Among the fertilized embryos collected up to Week 4, we observed severely deformed monster embryos resulting from chimaeras of nonheritable hetero-duplexed mutations. These embryos rarely survived and if hatched were not carried forward in the screen. In summary, all ENU treated parental fish exhibited reduced fertility up to Week 5 compared to the untreated control. The variance across the mean values of the treatment groups was not significant as compared to the control across all 10 weeks observed. However, by Week 5 all treated parental fish began to recover normal levels of fertility. Indeed the majority of fertilized embryos collected from these treated parental fish were from Weeks 5 to 10 in our genetic screens described below.

Fig. 3.

Post ENU treatment effects on parental fish fertility across 10 weeks. Eggs were collected and scored for fertility from parental fish treated with 0.0, 2.8, 3.0, 3.3 and 3.8 mM ENU across 10 weeks. See Table 1 for clonal stocks and numbers of parental fish treated and aggregate totals of embryos for each treatment group. Note: fertility was not recorded for Week 1 of the 3.3 mM treatment group.

Pilot genetic screen

A total of 36 parental fish (P) were treated with varying doses of ENU and allowed to self-cross (Fig. 4). Only two fishes died during these treatments and most P fish survived ∼1 year post treatment (356 days ± 114). In the next familial generation (F1), 7350 embryos were collected over a 10-week period. The F1 embryos were maintained in 149 unique families labeled by parent and week of collection (e.g., Hon9A-5 is a family collected from Hon9 stock, P fish A during Week 5 post ENU treatment). Eighteen percent of these F1 embryos (1334) hatched and survived to adulthood. Two hundred eighty four F1 fish were selected for the pilot screen by setting up at least two F1 fish from each of the 149 unique families in order to maximize the diversity and number of different starting haploid genomes potentially containing mutations (+*, Fig. 4). These 284 F1 fish were allowed to self-cross to produce the second familial generation (F2) and 2436 F2 embryos were scored for mutant zygotic defects during early development (Table 2). Close to 90% of these F2 embryos were fertilized and >60% were viable with an average of nine viable F2 embryos screened per F1 parent. Close to 40% of the F2 embryos died within the first day of development (nonviable), probably due to early embryonic lethal mutations. These nonviable embryos were simply removed 24 h after collection (i.e., end of Day 1).

Fig. 4.

K. mar. genetic screen outline. Parental hermaphroditic K. mar. (P) are mutagenized with different concentrations of ENU and self-crossed. Germ-line homozygous recessive mutations are predicted to occur across the diploid genome (+*/+*) and passed from the first familial generation (F1; m/+) to the second as zygotic mutants (F2; m/m).

Table 2.

Summary of F2 embryos scored in the pilot genetic screen

	N	Fertilized	Nonfertilized	Viable	Nonviable
Sum	2436	2150	286	1335	815
Mean	8a	15	2	9	5
Range	–	1–66	0–16	0–50	0–27
%	–	88	12	62b	38b

N = total F2 embryos collected from 284 F1 parents.

^aMean number of F2 fish per F1 parent.

^bBoth viable and nonviable embryos are derived as a percent of total F2 fertilized embryos scored (column 2).

Since K. mar. undergoes internal fertilization (oviposition), the stage at which they are laid is variable (Harrington 1963). Therefore, we scored the viable F2 embryos from 2 days after being laid through 14 days of development in broods collected weekly from their single F1 parents. This developmental period corresponds to >36 h post fertilization (stage 17, head-to-tail stage embryo, after gastrulation) through hatching stage (for embryonic stage series, see Mourabit et al. 2011). Initially, a total of 73 F1 fish produced discernable zygotic defects in their F2 progeny during this developmental period (m/m; Fig. 4).

Since each of the 284 F1 fish potentially inherited two mutagenized haploid genomes transmitted through the sperm and egg of their mutagenized hermaphroditic parent (+*/+*, Fig. 4), it is logical to conclude we screened 568 haploid genomes with an estimated 0.13 zygotic mutants identified per genome screened. However, this is a crude underestimate because (1) many mutations were probably missed in our simple microscopy screen, (2) we did not characterize the early embryonic lethal (nonviable) mutations as part of the total, and (3) our screen was not saturating for all chromosomal combinations (haploid n = 24; Solo et al. 1997) due to the low average number of F2 viable embryos we screened per F1 parent. As for point three, we probably reached saturation on some of our broods because we simply scored more viable embryos within the range of 0–50 (Table 2). To address some of these issues, we performed a rescreen on the 73 F1 fish producing zygotic mutants to confirm our initial observations and provide a larger range and average number of viable F2 offspring scored.

Since we treated parental fish with different concentrations of ENU, we also looked at the effect of dose on mutation rate across the 10 weeks of collection. The distribution and frequency of zygotic mutants was random across all treatment groups and weeks indicating no observable differences among ENU treatment groups. The most likely explanation is that our doses did not span a wide enough range (2.8–3.8 mM ENU) to identify differences in mutation frequency in a dose dependent manner.

Zygotic mutant rescreen

From a careful rescreen of >2000 F2 embryos produced by 73 self-crossing F1 fish, ∼60% were viable versus 30% nonviable (Table 3); a similar result to our previous pilot screen (Table 2). However, we were able to screen more viable F2 embryos on average per F1 fish (22, range of 4–84) as compared to our pilot screen because our overall mean number of F2 fish scored per F1 fish increased from 8 to 35.

Table 3.

Summary of F2 embryos scored from zygotic mutant rescreen

	N	Viable	Nonviable	Mutant
Sum	2539	1636	903	495
Mean	35a	22	12	7
Range	12–97	4–84	1–37	0–23
%	–	64	36	30b

N = total F2 embryos collected from the 73 F1 parents.

^aMean number of F2 fish per F1 parent.

^bMutant percentage is expressed as percent of total viable embryos scored (column 2).

Two of the 73 F1 fish failed to reproduce the previous mutant phenotype in their F2 offspring (3%). For the remainder of F1 fish screened, the types of mutants and their observed frequencies can be distributed among nine phenotypic categories (Table 4 and Fig. 5). The most obvious observable zygotic defects were those affecting morphology of the tail, body axis, and head (72%). For example, shortened, reduced, or curled tail fins were common (Fig. 5B and C), as well as those affecting skull/eye or jaw/mouth formation (Fig. 5D and E). We observed a reduced or dwarfed embryo as compared to 14-day sibling controls that usually contained a second phenotype such as curled tail (Fig. 5A and F). The appearance of fluid pouches (edema) were observed in ∼10% of the mutants (Fig. 5G). We also observed broods of F2 containing consistent unresolved phenotypes shortly after 2 days of development, presumably defects in gastrulation (Fig. 5H). Furthermore, ∼11% of our F1 fish produced lethal phenotypes resulting in developmental arrest displayed in ∼25% of F2 broods within the second day of development so are presumed to be pregastrulation lethal mutants (data not shown).

Table 4.

Summary of zygotic mutant rescreen

Typea	Phenotypic Class	N	Same P Founderb
A	Wild-type	–	–
B	Tail fin	15	4
C	Curly tail/body	16	10
D	Skull/eye	14	8
E	Jaw/mouth	2	–
F	Dwarf body	4	2
G	Fluid Pouch (edema)	7	2
H	Unresolved (postgastrula)	5	–
–	Embryonic lethal (pre-gastrula)	8	–
Total	9	71	26

N = total number of F1 fish producing phenotypic class of mutants in their F2 offspring.

^aType corresponds with Fig. 5.

^bNumber of F1 fish born from the same common P fish founder producing a similar mutant phenotype in their F2 offspring.

Fig. 5.

Zygotic mutants of K. mar. Examples of seven phenotypic categories of mutants were observed (Table 4). (A) Wild-type 14-day-old larval fish (hatching stage 32, see Mourabit et al. 2011). (B) Short Tail fin defect. (C) Curly tail defect. (D) Skull/eye defect. (E) Jaw/mouth defect. (F) Dwarf/retarded body development defect. (G) Fluid pouch defect. (H) Unresolved gastrulation defects.

The second day, pre-gastrula lethal embryos are important to distinguish from our nonviable F2 progeny that we defined as lethal within one day (early lethal). The overall percent of nonviable embryos were not scored within individual broods but simply recorded as such across all F2 progeny scored—as in our pilot screen. Therefore, these nonviable progeny probably represent early lethal mutations segregating throughout our zygotic screen, but were not characterized further in this study.

As for the phenotypic variance or penetrance level of the zygotic mutants we observed in our screen, we did not measure this because we chose to focus on obvious and most likely, fully-penetrable phenotypes. We note that multiple mutant phenotypes were segregating within broods indicating multiple mutant alleles could contribute to the simple zygotic mutant phenotypes. However, we focused on strong, fully penetrant phenotypes occurring in the predicted Mendelian ratios in our rescreen in order to simply establish mutagenesis in K. mar. Ultimately, breeding each zygotic mutant true into subsequent generations among multiple clonal lineages will allow us to measure multi-allele or phenotypic variance and penetrance levels.

It is common in forward genetic screens for the same mutant allele to be identified repeatedly that is derived from a single mutagenized founder, referred to as a “clonal” pattern of inheritance. We monitored the inheritance patterns of our zygotic mutants to determine if any shared a common mutated parent indicative of a clonal pattern. Nearly 40% of our F1 fish producing similar phenotypic classes of zygotic mutants are likely clonal (Table 4). However this would have to be confirmed by genetic complementation crosses followed by molecular characterization. Genetic complementation crosses are difficult at present in this species, yet we have begun molecular characterization of the tail fin mutants by the candidate gene approach (Type B, Table 4).

Conclusions and discussion

Here, we demonstrate that K. mar. is amenable to forward genetic screening through ENU chemical mutagenesis. A simple feeding and stock selection protocol was presented that is useful for setting up mutagenesis screens in this fish species. ENU treatment results in an initial loss of sperm function (fertility) in this hermaphroditic fish in the dosage range we tested (2.8–3.8 mM). However, fertility is quickly recovered and zygotic mutants are identifiable in the F2 generation. From a pilot screen of >500 genomes, we demonstrated this potential to identify nine different common phenotypic classes of zygotic mutants.

The zygotic mutants observed in our screen are similar to those observed in the zebrafish and medaka initial screens. For example, general retardation of growth; patterning defects in whole body, tail fin, and head structures; and embryonic lethality were common to our screen as compared to zebrafish (Mullins et al. 1994; Solnica-Krezel et al. 1994). A comparison of zygotic mutants to medaka is also similar; early defects of the eye and body axis (Loosli et al. 2000). We also identified a dominant body axis defect from three different F1 founders in our screen, analogous to the medaka headfish mutant and are currently following up on various viable mutants producing adult phenotypes (data not shown).

Clearly K. mar. genetics has its advantages as a vertebrate model of development, such as (1) the shorter cross time of one less generation due to mutagenesis in the hermaphroditic parent and (2) the ability to self-cross F1 individuals and maintain mutant alleles within a few single heterozygous carrier-fish. For example, the >500 genomes we screened were from <300 single F1 fish each housed in a small 1.5–2 L tank of stagnant, brackish water. A similar screen in zebrafish or medaka would require thousands of F2 fish grown in circulating tanks of families that are systematically pair-mated to each other for detection of zygotic mutants into the F3 generation. As for point two, we plan to confirm the zygotic mutants detected here in our F2 screen into the F3 generation. This will establish the rate of heritability of our zygotic mutant alleles and demonstrate maintenance of stocks into further generations.

Some current shortcomings of K. mar. as a genetic vertebrate model include (1) small weekly broods of 10–20 embryos weekly per fish (Fig. 2), (2) lack of sufficient outcrossing protocols for genetic complementation and meiotic mapping of mutant alleles, (3) germ-plasm archival of mutant alleles as sperm freezer stock for recovery or reverse genetic techniques, (4) a reference genome for mapping and molecular characterization of mutated genes, and a (5) lack of recessive viable markers leading to phenotypes that can be used to mark outcrossing or directly measure mutation rate by the locus specific complementation as demonstrated in zebrafish (Mullins et al. 1994). As for the last point, we were surprised we did not find many useful phenotypic markers of this sort. A simple explanation is our screen focused mainly on obvious patterning zygotic defects or our pilot screen may not have been large enough to detect such mutants. Regardless, we continue searching for these types of markers. As for the other current disadvantages noted above, these resources were not common to other established genetic model organisms, but rather were developed over years of the scientific community’s contributions. For example, it is anticipated that a first draft of the K. mar. genome will be available soon (Kelley et al., this issue).

Although inbreeding by self-crossing is novel to K. mar. as a vertebrate model and makes generation time quicker and simplifies animal husbandry, the mutation load (accumulation of many mutations simultaneously), is a critical issue as indicated by the 30% of nonviable, embryonic lethality we consistently identified in our pilot and rescreen. In similar zebrafish or medaka ENU mutagenesis screens, mutagenized males are outcrossed to females and in subsequent generations continued crosses, either through outcross or inbreeding within a family, reduces the overall mutation load of embryonic lethal mutations. An ideal scenario in K. mar. would be to mutagenize males and cross them to hermaphrodites (androdiecy), followed by hermaphroditic inbreeding into clonal lineages thereafter to identify mutations while reducing mutation load. Although outcrossing has been demonstrated in the laboratory by both natural and in vitro fertilization methods, these techniques need to be further optimized (Mackiewicz et al. 2006; Nakamura et al. 2008).

Other types of mutant screens could be performed in K. mar. such as those in search of late-developmental defects, behavior, or maternal effects (an F3 generation screen). We have already begun screening for dominant behavioral defects in our F1 fish and currently raising F2 fish to adulthood to screen for genes affecting fertility and maternal/paternal effect mutants. Supporting such expansion of this screen to late acting genes, late-developmental defects have been observed from a subset of the F2 fish generated from this screen by Matthew Harris and Helena Boldt at Harvard Medical School and Children’s Hospital in Boston, MA (personal communication). Further screens will extend our initial findings and generate important markers such as viable body color mutants useful for outcrossing and genetic mapping. We anticipate K. mar. will be utilized by more researchers in the future given the simplicity and hidden potential of this model organism.

Funding

Research was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development by the Academic Research Enhancement Award mechanism (award number R15HD060017). The authors also acknowledge support for the presentation of this work by NIH Conference Grant (award number R13HD070622) from the Eunice Kennedy Shriver National Institute of Child Health & Human Development; SICB through the DCE, DCPB, DAB and the C. Ladd Prosser Fund; Student & Faculty Development Funds from Center for Applied Research in the College of Arts & Science and Academic Affairs Office of the Valdosta State University; and the College of Agriculture and Natural Resources, University of Maryland.