Using Zebrafish to Learn Statistical Analysis and Mendelian Genetics

ZEBRAFISH Volume 8, Number 2, 2011 Original Articles ª Mary Ann Liebert, Inc. DOI: 10.1089/zeb.2010.0686

Using Zebraﬁsh to Learn Statistical Analysis and Mendelian Genetics

Samantha Lindemann,* Jon Senkler,* Elizabeth Auchter, and Jennifer O. Liang

Abstract

This project was developed to promote understanding of how mathematics and statistical analysis are used as tools in genetic research. It gives students the opportunity to carry out hypothesis-driven experiments in the classroom: students generate hypotheses about Mendelian and non-Mendelian inheritance patterns, gather raw data, and test their hypotheses using chi-square statistical analysis. In the first protocol, students are challenged to analyze inheritance patterns using GloFish, brightly colored, commercially available, transgenic zebrafish that express Green, Yellow, or Red Fluorescent Protein throughout their muscles. In the second protocol, students learn about genetic screens, microscopy, and developmental biology by analyzing the inheritance patterns of mutations that cause developmental defects. The difficulty of the experiments can be adapted for middle school to upper level undergraduate students. Since the GloFish experiments use only fish and materials that can be purchased from pet stores, they should be accessible to many schools. For each protocol, we provide detailed instructions, ideas for how the experiments fit into an undergraduate curriculum, raw data, and example analyses. Our plan is to have these protocols form the basis of a growing and adaptable educational tool available on the Zebrafish in the Classroom Web site.

Introduction Many zebrafish mutant and transgenic strains are available commercially and from the Zebrafish International Resource ndergraduate laboratory courses provide valuable Center, and there is a wealth of information about how to Uopportunities for hands-on learning, the application of successfully raise and maintain them (zebrafish.org/zirc/ concepts learned in lectures and textbooks, and an introduc- home/guide.php).8–10 Our first protocol uses GloFish, a tion to the complexity of science and experimental design. brightly colored and fluorescent strain. GloFish are readily Mendelian genetics has long been taught through laboratory available in pet stores (www.glofish.com/), making it possi- experiments, with protocols for using Drosophila melanogaster ble to carry out challenging genetic experiments in schools in the classroom published as early as 1918.1 Here we present without access to a research zebrafish facility. two protocols that use the zebrafish model system to give Several excellent protocols already exist for using zebrafish students the opportunity to analyze Mendelian and non- to teach genetics (www.bioeyes.org/) (www.insciedout.org/) Mendelian inheritance patterns and rigorously test their hy- (www.glofish.com/classroom.asp).2–7 Our protocols build potheses through statistical analysis. The difficulty of these upon this earlier work by giving students background in how protocols can be easily adapted to students at different levels, genetic screens are used to find new mutants, bringing in the and the material requirements are low, making them acces- analysis of several phenotypes in a single cross, and including sible to a wide range of classrooms. inheritance patterns that do not fall into basic recessive and Zebrafish are being increasingly used for education, and in dominant patterns. Finally, these protocols add richness to the particular have many advantages for genetic experiments.2–7 study of genetics by challenging students to form their own They are easy to raise, and a single pair often produces 100 hypotheses, and then rigorously test these hypotheses through embryos or more when mated, making it possible for each the use of chi-square statistical analysis. Chi-square analysis student to have their own fish. Since adult fish live over 2 requires only the use of basic mathematic techniques, making years, the same fish can be used over many semesters, either it appropriate for middle school and high school students. directly when the adult fish express viable phenotypes or to Knowledge of mathematics is becoming even more important produce clutches containing mutant embryos for analysis. to biologists as they tackle increasingly complex problems and

Department of Biology, University of Minnesota Duluth, Duluth, Minnesota. *These authors contributed equally to this work.

41 42 LINDEMANN ET AL. large data sets.11–13 This has led to a call for greater integration lected into deep (10020 mm) Petri dishes (Cat. #M090501; of mathematics into the undergraduate curriculum.13–15 Our Laboratory Products Sales, Rochester, NY), one dish for each protocols enable the introduction of mathematics into the clutch of eggs. To do this, the adult fish and the insert in the undergraduate curriculum during the freshman or sophomore mating tank were removed and the eggs allowed to settle. The year, providing a foundation for more complex mathematical majority of the water was poured out slowly so that no approaches in subsequent courses. eggs were lost. When about 20 mL of water was left, the re- The two protocols reported here were developed for a maining water and eggs were poured quickly into a Petri dish sophomore level undergraduate genetics laboratory course. (www.zfic.org/common%20techniques/mating.html). The This course met for one 4-hour laboratory session each week fertile eggs were sorted from infertile eggs and waste prod- of the semester, and all of the students (*60 students/ ucts using a dissecting microscope with transmitted light. semester, 10–20 students/laboratory section) either had taken Embryos were moved to the middle of the Petri dish using or were concurrently taking a genetics lecture course. How- the ‘‘embryo swirl’’ (www.zfic.org/common%20techniques/ ever, these experiments will coordinate with virtually every embryo%20swirl.html). The good eggs, which appeared genetics course, as they incorporate the concepts that form the translucent and had normal morphology, were sorted into basis of all genetics studies. In addition, these laboratories are one area of the Petri dish using an embryo loop (www student-driven and problem-based without being extremely .zfic.org/common%20techniques/Embryoloops.html), and labor intensive for the instructors. Thus, they will fit well into waste and infertile eggs removed. Healthy embryos were courses aiming to eliminate ‘‘cook-book’’ laboratories from moved to a new Petri dish containing clean aquatic system their curriculum. Further, the raw data we have included can water using a transfer pipet, and the Petri dish placed in an be used to introduce active learning and problem solving into incubator at 28.58C or in the aquatic fish facility until the fish lecture courses, which has been shown to have a positive were ready to be imaged or raised in a large tank. impact on the ability of students to retain knowledge and their long-term achievement.16 Our goal is to make this an evolving Scoring the phenotype of cyc and sqt mutants and growing collaborative protocol on our Zebrafish in the Between 2 and 3 dpf, developing embryos were counted Classroom Web site (www.zfic.org) that includes variations and scored by their eye phenotype using a dissecting micro- on how to use these ideas in the classrooms for students at scope, with normal embryos having two eyes and mutant diverse levels in their development as scientists. embryos having more closely spaced or cyclopic eyes.

Materials and Methods Imaging

All procedures have been approved by the University of Fish were photographed without anesthesia or with a short Minnesota IACUC Committee. A copy of the approved pro- incubation in 0.017% tricaine methanesulfonate (MS-222) tocol is available on the Zebrafish in the Classroom Web site dissolved in aquatic system water. Larvae were photo- (www.zfic.org/common%20techniques/IACUC-teaching-all graphed using a Leica S6 D stereo light microscope fitted with .pdf). In addition, students complete vertebrate animal safety a Nikon Digital Sight DS-SM camera. For all adult pictures training before starting the laboratories (Supplementary Ma- except the progeny in Figure 5 and all fish in Figure 14, the fish terial 1; Supplementary Data are available online at www were netted onto a flat surface with a small amount of water, .liebertonline.com/zeb). and then photographed using Panasonic DMZ-TZ3 digital camera. Images in Figure 5 were captured with a Canon EOS Rebel XS 18-55IS digital camera. For Figure 14, adult fish Fish stocks were netted onto the glass plate of an Olympus SZX12 Parental fish stocks included the wild-type (WT) strain stereomicroscope connected to a Cannon PowerShot A520 zebrafish Danio rerio (ZDR) (Aquatic Tropicals, Plant City, through one eye piece. Images were taken with white light FL), and strains carrying the following mutations and trans- illumination from above or with fluorescence microscopy. genes: cyclopsm294 (cyc)17; squintcz35 (sqt),18 mylz2:Yellow Fluorescent Protein (GloYFP),19 mylz2:Red Fluorescent Protein Results (GloRFP),19 golden (gol),20,21 and long fin (lof).22,23 Adult fish Laboratory 1: analysis of Mendelian inheritance were maintained using standard protocols.10 Fish to be raised patterns using GloFish were maintained for 8–9 days postfertilization (dpf) in Petri dishes at 28.58C, and then placed in a 10 L tank within a re- This laboratory is designed to be the opening laboratory of circulating, multi-rack aquarium system (Aquatic Habitats, the semester. Students analyze sibling groups of adult zeb- Apopka, FL) in our fish facility. From 9–14 dpf they were fed rafish by counting the number of fish with each phenotype, twice daily with live paramecium10 and a pinch of a powder generate hypotheses about the inheritance of the relevant containing one part dried Spirulina and one part Argent genes based on these data, and test their hypotheses using chi- Chemical Laboratories Hatchfry Encapsulon Grade 0.24 square statistical analysis (Supplementary Materials 1–7). To make this engaging, and to bring in the concept of transgenes, it uses GloFish, a commercially available strain containing Natural matings transgenes that make the fish brightly colored (Table 1) Adult fish were set up in single pair or group matings in the (www.glofish.com).19,25,26 In the most challenging sibling afternoon, and monitored for spawning until late afternoon group, students make hypotheses about the inheritance pat- the following day. At that time, adult fish were placed back terns of three different genes that when combined together into their home tanks, and any embryo produced were col- produce eight different phenotypes in a single clutch of sib- STATISTICAL ANALYSIS AND MENDELIAN GENETICS 43

Table 1. Summary of Mutations and Transgenes Used in These Laboratory Experiments

Type Homozygous Heterozygous Inheritance Gene name Abbreviation of genetic change phenotype phenotype pattern mylz2:YFP GloYFP Transgene Yellow body Yellow body Dominant (or incompletely dominant) mylz2:RFP GloRFP Transgene Red body Red body Dominant (or incompletely dominant) no transgene Glo- None Gray body N/A Recessive golden gol Mutation Attenuated stripes Normal stripes Recessive long fin lof Mutation Long fins Long fins Dominant cyclops cyc Mutation Cyclopia Normal eyes Recessive squint sqt Mutation Cyclopia Normal eyes Recessive, incompletely penetrant

lings (Trihybrid Crosses 1 and 2 below). The experiments in for students to set up their own single pair mating designed to this protocol are accessible to teaching laboratories that do not test one of their hypotheses (Supplementary Materials 5 and have an association with a research zebrafish facility with 6). The last day of the semester, the students have the chance many mutant strains, as all needed materials can be easily to examine the progeny of these crosses and determine if their found in pet stores. For instance, the GloFish available from hypothesis was correct and their experimental design sound pet stores carry transgenes and mutations that enable the (Supplementary Material 2) analysis of recessive, dominant, and incomplete dominance Students are encouraged to work together throughout the inheritance patterns, making it easy to build both simple and laboratory, but the written assignments are to be completed complex crosses for student analysis. individually by each student. The written assignments in- The Glo transgenes, which give these fish their name, use clude the laboratory worksheet, which counts as their labo- the muscle-specific fast skeletal light chain 2 (mylz2) promoter ratory notebook entry, and a homework in which they design to drive expression of Green Fluorescent Protein (GFP), Yellow an experiment to test one of their hypotheses about inheri- Fluorescent Protein (YFP), or Red Fluorescent Protein (RFP/ tance patterns (Supplementary Materials 5–7). Here, we have dsRed), making the body color of the fish green-yellow, yel- provided raw data and analyses for four GloFish crosses that 19,25,26 low, or red, respectively. GloFish purchased from pet could be used for this laboratory. stores are typically homozygous for the recessive gol mutation, which causes them to have very light or non-existent stripes Dihybrid cross 1 (Table 1).20 To create trihybrid crosses, we crossed GloFish to WT zebrafish, which introduces the WT striped allele of The first experiment in this laboratory enables students to the gol gene, and into a line carrying the dominant lof mutation, start with analysis of recessive and dominant genes, the which causes all of the fins to undergo unregulated growth simplest Mendelian inheritance patterns. It analyzes a dihy- (Table 1).22,27–30 WT zebrafish and fish carrying the lof muta- brid cross between a WT male and a female heterozygous for tion are also readily available in pet stores. Because all of these the GloYFP transgene and homozygous for gol mutation genetic changes in our crosses produce viable phenotypes, (Fig. 1). Thus, the male is gray with stripes, whereas the fe- inheritance patterns can be followed over several generations male is yellow and lacks stripes (Fig. 1). Of 14 progeny, 57% and students can be challenged to interpret the inheritance were yellow and 43% were gray, and all were striped (Fig. 1, patterns of several genes in one cross. Table 2, Supplementary Fig. 1). The experimental design and set up for the laboratory are The next step in the experiment is for the students to gen- relatively simple (See GloFish Instructor’s Key: Supplemen- erate a hypothesis about the inheritance patterns in this cross, tary Material 2). Students prepare by completing vertebrate and test their hypothesis with chi-square statistical analysis. animal safety training, and by reading an introduction to the In chi-square analysis, the hypothesis about the inheritance laboratory (Supplementary Materials 1 and 4). At the start of pattern of a gene or genes is termed the null hypothesis. The class, the students are given a short PowerPoint presentation chi-square test produces a probability value ( p-value) that that puts the GloFish experiment into context, lists the goals reports the probability that the observed distribution of val- for the day, and gives examples to illustrate important con- ues is the same as the expected distribution. For instance, a cepts in statistical analysis (Supplementary Material 3). p-value of 0.40 means that there is a 40% chance that the ob- A worksheet guides students through the steps of the lab- served distribution is consistent with the null hypothesis, and oratory and at the same time teaches them how to keep an a 60% chance that it is not. A p-value of 0.05 or less is needed to accurate and complete laboratory notebook (Supplementary reject the null hypothesis. In other words, if there is a 5% or Materials 5 and 7). Briefly, students move in small groups lower probability that the observed distribution is consistent around the laboratory and together sort tanks of sibling fish with the null hypothesis, the null hypothesis can be rejected. It into phenotypic classes and count the number of fish in each is important to note that the converse is not true. A p-value class. After this is complete, the groups together generate null above 0.05 does not prove the null hypothesis. To rigorously hypotheses about the inheritance patterns in each tank, and prove the null hypothesis, one would have to disprove all of test these hypotheses using chi-square analysis. The final step the other possible hypotheses by showing that they produce a in the protocol, which is typically done a week or more later, is p-value less than 0.05. YFP FIG. 1. Dihybrid cross with fish carrying the Glo transgene and gol mutation. The parental (P0) generation consisted of a female fish heterozygous for the GloYFP transgene and homozygous for the gol mutation and a WT male fish that was not carrying the transgene or the gol mutation. Genotypes at the different loci are separated by semicolons with the genotype at the transgene locus listed first. The F1 generation was produced through a single mating of these adult fish, with one progeny of each phenotype shown. Images are lateral views, anterior to the left and dorsal to the top. Images of all of the progeny are included in Supplementary Figure 1. WT, wild type.

A gol gol B Glo- GloYFP

+ +/gol +/gol Glo- Glo-/Glo- Glo-/GloYFP

++/gol +/gol Glo- Glo-/Glo- Glo-/GloYFP

genotype phenotype genotype phenotype +/gol striped (WT) Glo-/Glo- grey body (WT) Glo-/GloYFP yellow body

C gol;Glo- gol;Glo- gol;GloYFP gol;GloYFP +;Glo- +/gol;Glo-/Glo- +/gol;Glo-/Glo- +/gol;Glo-/GloYFP +/gol;Glo-/GloYFP

+;Glo- +/gol;Glo-/Glo- +/gol;Glo-/Glo- +/gol;Glo-/GloYFP +/gol;Glo-/GloYFP

genotype phenotype +/gol;Glo-/Glo- striped, grey body (WT) +/gol;Glo-/GloYFP striped, yellow body

FIG. 2. Hypothesis for dihybrid GloYFP and gol cross. The hypothesis for the inheritance pattern of the dihybrid cross is presented as Punnett squares. The genotypes of the gametes produced by the female parent are shown in the top row and the genotypes of the gametes produced by the male parent are shown in the left column. The genotypes of the progeny were generated by filling in the appropriate inherited genes from each parent, with maternal genes marked in black and paternal genes marked in gray. The hypothesized relationship of ge- FIG. 3. Dihybrid cross of parents heterozygous for GloYFP notype to phenotype is listed below the Punnett squares. (A) and gol. The parental (P0) generation consisted of a female Hypothesis for striped phenotype considered alone. (B) Hy- and a male fish heterozygous for the GloYFP transgene and pothesis for body color phenotype considered alone. (C) Hy- the gol mutation. Genotypes at the different loci are sepa- pothesis for stripe pattern and body color analyzed together in rated by semicolons with the genotype at the transgene locus a dihybrid Punnett square. Because there are two loci that listed first. The F1 generation was produced through a single must be followed at once, there are four possible gamete ge- mating of the P0 pair, with one progeny of each phenotype notypes for each parent. The hypothesis predicts that half of shown. Images are lateral views, anterior to the left and the progeny should be normal colored with normal stripes and dorsal to the top. Images of all of the progeny are included in half of the progeny should be yellow with normal stripes. Supplementary Figure 2.

44 STATISTICAL ANALYSIS AND MENDELIAN GENETICS 45

Table 2. Chi-Square Analysis of a Dihybrid Cross Between a Wild-Type Male and a Female YFP Heterozygous for the Glo Transgene and the GOL Mutation

(1) (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O ED2 ¼ (O E)2 w2 ¼ D2/E

Normal color, 6 7 1 1 0.1 Normal stripes Yellow color, 8 7 1 1 0.1 Normal stripes (7) Total w2 ¼ 0.2 (8) Degrees of freedom (n 1) ¼ 2–1 ¼ 1 (9) 0.6 < p < 0.7

(1) The two phenotypes expected from the hypothesis in Figure 2 were written in the first column. (2) The number of fish observed (O) with each phenotype was recorded in the second column. (3) The expected number of progeny was calculated and recorded in column 3. E (normal color, normal stripes) ¼ 0.51.014 ¼ 7. E (yellow color, normal stripes) ¼ 0.51.014 ¼ 7. (4–6) For each phenotype, D, D2, and w2 were calculated for each phenotype and recorded in the appropriate column. Numbers were rounded to one significant digit, as this is the number of significant digits in column 2. (7) The degrees of freedom were calculated using the number of phenotypes (n) expected by the hypothesis. (8) The w2 value for the experiment was the sum of the w2 values for each phenotype. (9) The range of p-values was calculated by using a w2 table (www.sociology.ohio-state.edu/people/ptv/publications/p%20values/ chi_table.jpg), the w2 value for the experiment, and the degrees of freedom.

To make the experiment more challenging, students are the number of fish increases, but the ratio of fish with each typically not told the genotypes or phenotypes of the parents, phenotype stays the same, the p-value will decrease. Another and have to generate the null hypotheses based only on their possibility is for the students to carry out a chi-square test on observations of the progeny. Even when lacking the parental the alternative (incorrect) hypothesis, providing support for information, generating a correct hypothesis is typically not a their hypothesis by disproving the alternative. Perhaps the challenge for the sophomore level undergraduate students in best approach would be to design an experiment that will give this course. For this cross, two null hypotheses fit the data different outcomes whether the hypothesis is correct or in- equally well. The correct one is that the gol mutation is recessive correct, such as the experiment in Dihybrid cross 2. and the GloYFP transgene is dominant (Fig. 2). Alternatively, This cross could also be used for an even more basic start to the students could hypothesize that the GloYFP transgene is reces- experiment, following only the GloYFP transgene and the body sive, with the yellow parent being homozyogous for the color phenotype as a monohybrid cross. In this case, chi-square transgene, and the gray parent being heterozygous for the analysis for body color generates a p-value between 0.6 and 0.7, transgene. Class discussions can be used to generate a next-step and in fact is exactly the same as the chi-square analysis when experiment to determine which of these two null hypotheses is both phenotypes are considered (Table 3). Note that chi-square correct. Our Dihybrid cross 2 is one such cross (Fig. 3). analysis cannot be used to analyze the striped/nonstriped phe- For simplicity, only the correct hypothesis (Fig. 2) is tested notypes alone because there is only one phenotype in the prog- here. From our null hypothesis, we expect that 50% of the eny, making the degrees of freedom equal to zero. progeny will be yellow and 100% will be striped (Fig. 2, Table 2). The p-value from the chi-square analysis falls between 0.6 Dihybrid cross 2 and 0.7, supporting the null hypothesis (Table 2). The number of progeny for this cross is quite low. This offers a strong Our second cross is designed to distinguish between the two opportunity for discussions on what experiments could be null hypotheses that fit the data for Dihybrid cross 1. In this used to further support the null hypothesis. One possibility is cross, both parents have the same phenotype of a yellow body to repeat the experiment to increase the number of progeny and stripes, and are heterozygous for both GloYFP and gol analyzed. If the null hypothesis is correct, then increased (Fig. 3). The progeny phenotype ratios are approximately 3:1 numbers should cause the p-value to increase. In contrast, if yellow:gray and 1:1 striped:nonstriped (Supplementary Fig. 2,

YFP Table 3. Chi-Square Analysis of a Monohybrid Cross for Fish Carrying the Glo Transgene

(1)a (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O ED2 ¼ (O E)2 w2 ¼ D2/E

Normal 6 7 1 1 0.1 body color Yellow 8 7 1 1 0.1 body color (7) Total w2 ¼ 0.2 (8) Degrees of freedom (n 1) ¼ 2–1 ¼ 1 (9) 0.6 < p < 0.7

aCalculations for each column and row carried out as in Table 2. This analysis was done on the same cross as in Table 2, except the pigment phenotype was ignored. 46 LINDEMANN ET AL.

YFP Table 4. Chi-Square Analysis of a Dihybrid Cross of Fish Heterozygous for Glo and GOL

(1)a (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O ED2 ¼ (O E)2 w2 ¼ D2/E

Normal color, 5 6 1 1 0.2 Normal stripes Normal color, 4 2 2 4 2 No stripes Yellow color, 15 19 4 16 0.8 Normal stripes Yellow color, 10 6 4 16 2.7 No stripes (7) Total w2 ¼ 5.7 (8) Degrees of freedom (n 1) ¼ 4–1 ¼ 3 (9) 0.1 < p < 0.15

aCalculations for each column and row carried out as in Table 2.

Table 4). These ratios should lead students to formulate the mathematical complexity of the chi-square analysis. A cross null hypothesis that the gol mutation is recessive and the between a red, striped female and a yellow, striped male GloYFP transgene is dominant (Fig. 4). Chi-square analysis yields progeny with four different colors: gray, yellow, red, yields a p-value between 0.1 and 0.15 (Table 4). This p-value and orange, and because both parents are carrying one copy supports the null hypothesis, but is close enough to 0.05 to of the gol mutation, two different pigment patterns, striped leave some doubt. This low p-value could be used to prompt and nonstriped (Fig. 5, Supplementary Fig. 3). student discussions on how to decide when a hypothesis is This cross also brings in a new inheritance pattern. The supported. Questions that can be used to stimulate discussion presence of orange fish should lead students to form the null include the following: Does the experiment need to be re- hypothesis that the GloYFP and GloRFP transgenes are in- peated to increase the number of fish analyzed? If Dihybrid completely dominant with each other: blending of yellow and crosses 1 and 2 are considered together, does this increase the red produces orange fish, and orange is dominant over the support for the hypothesis? Are their alternative hypotheses normal gray color. The gol mutation, which causes loss of that could explain these data? Are these alternative hypoth- stripes, is recessive as in the dihybrid crosses. Together, these eses supported or rejected by chi-square analysis? inheritance patterns lead to the expectation that the progeny will be divided equally among each body color, and that 75% Trihybrid cross 1 of the fish will have stripes and 25% will be non-striped (Fig. 6). The chi-square test yields a p-value greater than 0.9, Our next cross, a trihybrid cross, significantly increases the indicating any deviation from expected distributions is likely number of phenotypes found in the progeny, and thus the due to chance and the hypothesis is supported (Table 5).

Trihybrid cross 2 A gol + B Glo- GloYFP The second trihybrid cross enables students to practice gol gol/gol gol/+ Glo- Glo-/Glo- Glo-/GloYFP their analysis skills, introduces a new phenotype and mutation, and explicitly tests whether orange fish are carrying both YFP RFP ++/gol +/+ GloYFP GloYFP/Glo- GloYFP/GloYFP Glo and Glo transgenes. In this cross, an orange male with short fins, one of the progeny from Trihybrid cross 1, was genotype phenotype genotype phenotype mated with a gray female fish with long fins, producing fish +/+ striped Glo-/Glo- grey body that were gray, yellow, red, and orange, with either long or - YFP +/gol striped Glo /Glo yellow body short fins (Fig. 7, Supplementary Fig. 4). gol/gol not striped GloYFP/GloYFP yellow body The presence of red and yellow fish should lead to the YFP C Calculations: hypothesis, as in the first trihybrid cross, that Glo and RFP grey, striped phenotype 1/4 grey X 3/4 striped = 3/16 of progeny Glo are incompletely dominant with each other and indi- grey, not striped phenotype 1/4 grey X 1/4 not striped = 1/16 of progeny yellow, striped phenotype 3/4 yellow X 3/4 striped = 9/16 of progeny vidually dominant over a gray body color (Fig. 8). Presence of yellow, not striped phenotype 3/4 yellow X 1/4 not striped = 3/16 of progeny a long-finned phenotype in approximately half of the progeny should lead to the hypothesis that the lof mutation is dominant (Fig. 8). This hypothesis is supported by the chi-square FIG. 4. Hypothesis for dihybrid cross of heterozygous fish. test, which yields a p-value between 0.4 and 0.5 (Table 6). The hypothesis for the inheritance pattern of the dihybrid cross is presented as Punnett squares generated as in Figure 2. (A) Hypothesis for striped phenotype considered alone. (B) Hy- Laboratory 2: F3 genetic screen pothesis for body color phenotype considered alone. (C) Cal- culations of expected fraction for each phenotype using the Genetic screens are one of the most powerful techniques Punnett squares in (A) and (B). When the loci being analyzed are in biology. They have been used to identify hundreds of not linked, this method of calculating expected fractions of each genes with key roles in development, disease, basic cellular phenotype can be used instead of a dihybrid Punnett square. functions, and a host of other biological processes and have FIG. 5. Trihybrid cross with fish carrying Glo transgenes and the gol mutation. The parental (P0) generation consisted of a female fish heterozygous for the GloRFP transgene and heterozygous for the gol mutation and a male fish heterozygous for the GloYFP transgene and heterozygous for the gol mutation. Genotypes at the different loci are separated by FIG. 7. Trihybrid cross with fish carrying Glo transgenes YFP semicolons with the genotype at the Glo transgene locus and the lof mutation. The parental (P0) generation consisted listed first. The F1 generation was produced through a single of a female not carrying a transgene and heterozygous for the RFP mating of the P0 pair, with one progeny of each phenotype lof mutation and a male fish heterozygous for the Glo and shown. Images are lateral views, anterior to the left and GloYFP transgenes. Genotypes at the different loci are sepa- YFP dorsal to the top. Images of all of the progeny are included in rated by semicolons, the Glo locus listed first. The F1 Supplementary Figure 3. generation was produced through a single mating of the P0 pair, with one progeny of each phenotype shown. Images are lateral views, anterior to the left and dorsal to the top. Images of all of the progeny are included in Supplementary Figure 4.

gol;Glo-;Glo- gol;Glo-;GloRFP gol;Glo-;Glo- gol;Glo-;GloRFP gol;Glo-;Glo- gol;Glo-;GloRFP gol;Glo-;Glo- gol;Glo-;GloRFP + /gol + /gol + /gol + /gol + /gol + /gol + /gol + /gol - - +;Glo ;Glo Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- FIG. 6. Hypothesis for - - - RFP - - - RFP - - - RFP - - - RFP Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo trihybrid GloYFP; GloRFP; + /gol + /gol + /gol + /gol + /gol + /gol + /gol + /gol YFP - +;Glo ;Glo GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- gol cross. The hypothesis Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP for the inheritance pattern + /gol + /gol + /gol + /gol + /gol + /gol + /gol + /gol of the trihybrid cross is +;Glo-;Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- presented as a trihybrid - - - RFP - - - RFP - - - RFP - - - RFP Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Punnett square generated + /gol + /gol + /gol + /gol YFP - + /gol + /gol + /gol + /gol +;Glo ;Glo YFP - YFP - GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- Glo /Glo GloYFP/Glo- Glo /Glo as in Figure 2. Because Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP there are three loci that gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol must be followed in the gol;Glo-;Glo------Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo Glo /Glo cross, there are eight possi- Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP ble gamete genotypes for gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol YFP - gol;Glo ;Glo GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- each parent. The hypothesis Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP predicts that one-eighth of gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol the progeny should have gol;Glo-;Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- Glo-/Glo- - RFP - RFP each of the F phenotypes Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo /Glo Glo-/Glo- Glo /Glo 1 gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol gol/gol shown in Figure 5. gol;GloYFP;Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- GloYFP/Glo- Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP Glo-/Glo- Glo-/GloRFP

47 48 LINDEMANN ET AL.

YFP RFP Table 5. Chi-Square Analysis of Trihybrid Cross Containing the Glo and Glo Transgenes and GOL Mutation

(1)a (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O ED2 ¼ (O E)2 w2 ¼ D2/E

Normal color, 7 7 0 0 0 Normal stripes Normal color, 4 2 2 4 2 No stripes Yellow color, 6 7 1 1 0.1 Normal stripes Yellow color, 2 2 0 0 0 No stripes Orange color, 6 7 1 1 0.1 Normal stripes Orange color, 2 2 0 0 0 No stripes Red color, 7 7 0 0 0 Normal stripes Red color, 3 2 1 1 0.5 No stripes (7) Total w2 ¼ 2.7 (8) Degrees of freedom (n 1) ¼ 8–1 ¼ 7 (9) 0.9 < p < 0.99

aCalculations for each column and row carried out as in Table 2. been the foundations of many Nobel Prizes (for instance, see these mutant lines going was quite difficult, as chromosome http://nobelprize.org/nobel_prizes/medicine/laureates/ segregation during meiosis was often disrupted. In the mid- 1995/ and http://nobelprize.org/nobel_prizes/medicine/ 1990s, zebrafish research began its exponential growth with laureates/2002/). This protocol gives students the opportu- the completion of two very large forward genetic screens, one 32–34 nity to gain experience with forward genetic screens by doing in Tubingen, Germany, and the other in Boston. These a mock screen, biased so that students will successfully find screens used the chemical N-ethyl-N-nitrosourea to induce interesting mutants. Because this F3 screen protocol is longer point mutations, and then kept every mutant line that had a and requires more skills than the GloFish protocol, it has developmental defect (Development Vol. 123, published in typically been done in the middle of the semester. It includes December 1996). Mendelian genetics and the use of chi-square analysis as in the The screening methods developed for these large screens GloFish protocol, but in addition brings in the concepts of are the basis for this protocol (Supplementary Materials 9 and forward genetics and the use of model systems. Further, it 10). The screen is called F3 because it requires three genera- builds skills in embryology, light microscopy, and even fish tions of offspring (F1,F2, and F3) from the parental (P) gen- husbandry (Supplementary Materials 1, 7–13). eration. The students carry out the last step of an F3 screen, Zebrafish is a relatively new model system. When the field which takes three laboratory periods. On the first day of the started, the first mutations were induced with gamma radia- protocol, students set up several single pair matings, called tion. Gamma radiation causes large deletions or transloca- blind intercrosses, between fish in a tank that contains a tions in the genome (for instance, see ref.31). Keeping some of mixture of fish heterozygous for a mutation and fish that are

A gol + B Glo- GloYFP C Glo- GloRFP FIG. 8. Hypothesis for trihybrid GloYFP; GloRFP; lof cross. The hypothesis for the in- + +/gol +/+ Glo- Glo-/Glo- Glo-/GloYFP Glo- Glo-/Glo- Glo-/GloRFP heritance pattern of each phenotype in the trihybrid cross is presented as Punnett - - - - YFP - - - - RFP squares generated as in Figure 2. (A) Hy- ++/gol +/+ Glo Glo /Glo Glo /Glo Glo Glo /Glo Glo /Glo pothesis for striped phenotype considered alone. (B) Hypothesis for yellow body color D Calculations: phenotype considered alone. (C) Hypothesis for red body color considered alone. (D) grey, striped phenotype 1/2 striped X 1/2 not yellow X 1/2 not red = 1/8 of progeny Calculations of expected fraction for each grey, not striped phenotype 1/2 not striped X 1/2 not yellow X 1/2 not red = 1/8 of progeny phenotype using the Punnett squares in (A), yellow, striped phenotype 1/2 striped X 1/2 yellow X 1/2 not red = 1/8 of progeny yellow, not striped phenotype 1/2 not striped X 1/2 yellow X 1/2 not red = 1/8 of progeny (B), and (C). This method of calculating the orange, striped phenotype 1/2 striped X 1/2 yellow X 1/2 red = 1/8 of progeny expected fraction of each progeny can be used orange, not striped phenotype 1/2 not striped X 1/2 yellow X 1/2 red = 1/8 of progeny instead of a trihybrid Punnett square. red, striped phenotype 1/2 striped X 1/2 not yellow X 1/2 red = 1/8 of progeny red, not striped phenotype 1/2 not striped X 1/2 not yellow X 1/2 not red = 1/8 of progeny STATISTICAL ANALYSIS AND MENDELIAN GENETICS 49

Table 6. Chi-Square Analysis of Trihybrid Cross Containing YFP RFP the Glo and Glo Transgenes and LOF Mutation

(1)a (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O ED2 ¼ (O E)2 w2 ¼ D2/E

Normal color, 4 2 2 4 2 Short fins Normal color, 0 2 242 Long fins Yellow color, 1 2 1 1 0.5 Short fins Yellow color, 1 2 1 1 0.5 Long fins Orange color, 2 2 0 0 0 Short fins Orange color, 4 2 2 4 2 Long fins Red color, 2 2 0 0 0 Short fins Red color, 2 2 0 0 0 Long fins (7) Total w2 ¼ 7 (8) Degrees of freedom (n 1) ¼ 8–1 ¼ 7 (9) 0.4 < p < 0.5

aCalculations for each column and row carried out as in Table 2. homozygous WT (Supplementary Materials 8 and 10). These sort out debris and infertile embryos under a dissecting mi- groups of ﬁsh are the mock F2 families. Laboratories main- croscope (Supplementary Material 10). On the third day, taining strains of zebraﬁsh with recessive lethal mutations which can be the next day or several days later, students will typically have mixed tanks such as these available, as screen the F3 embryos for phenotypes, count the ratio of

they are produced in the process of generating the next gen- normal to abnormal embryos, make hypotheses about what eration of ﬁsh (Supplementary Material 8). kind of mutation is present (recessive, dominant, etc.) and the On the second day of the protocol, which ideally would be genotype of the parents, and test their hypothesis using chi- the following day, students learn how to collect embryos and square analysis (Supplementary Materials 10 and 11). The primary difference between this protocol and a real F3 screen is that we know that our mock F2 families are carrying mutations and what phenotypes to expect. The example crosses we present here both use mutants in the Nodal signaling pathway (Figs. 9–12). Zebraﬁsh have three Nodal signaling proteins: Sqt, Cyc, and Southpaw. Mutations in the genes encoding any of these proteins result in

+ cyc

+ +/+ +/cyc

cyccyc/+ cyc/cyc

genotype phenotype FIG. 9. Cross of fish carrying the cyc mutation. The parental +/+ two eyes (WT) (P0) generation consisted of a female and a male fish het- cyc/+ two eyes (WT) erozygous for the cyc mutation. A single natural mating of cyc/cyc cyclopic eye these fish produced progeny, the F1 generation, with two normal eyes or a single cyclopic eye. The fish with cyclopic FIG. 10. Hypothesis for cyc cross. The hypothesis for the eyes also had a severe ventral curvature in the anterior– inheritance pattern of the cyc cross is presented as a Punnett posterior axis. A typical embryo at 3 days postfertilization is square generated as described for Figure 2. The hypothesized shown for each phenotype. All images are lateral views with relationship of genotype to phenotype and the Punnett anterior to the left and dorsal to the top. Images of all square together generate the prediction that three quarters of progeny from this cross are included in Supplementary the progeny should have normal eyes and one quarter Figure 5. should have cyclopic eyes. 50 LINDEMANN ET AL.

FIG. 11. Cross of fish carrying the sqt mutation. The parental (P0) generation consisted of a female and male fish that were heterozygous for the sqt mutation. As with the cyc cross, a single natural mating of sqt heterozygotes fish produced progeny, the F1 generation, with two normal eyes or a single cyclopic eye. Adults are shown in lateral views, anterior to the left, embryos in ventral views with anterior to the left. Pictures of all of the progeny from this cross are included in Supplementary Figure 6.

severe phenotypic defects and are ultimately lethal.35–38 The ing embryo.39–42 Thus, it is not surprising that sqt mutants most obvious phenotype in cyc and sqt mutants is cyclopic share many features in common with cyc embryos, including eyes. At 3 days of development, the eyes are pigmented and cyclopic eyes (Fig. 11, Supplementary Fig. 6). However, sqt very large, making the difference between the two eyes of the mutants have some phenotypes that are distinct from cyc. For WT embryo and the cyclopic eye of the mutants easily dis- instance, cyc mutants are curved ventrally, whereas sqt mu- tinguishable even to students who are observing embryos for tants are straight or curved to the left or right (Figs. 9 and 11, the ﬁrst time (Figs. 9 and 11, Supplementary Figs. 5 and 6). Supplementary Figs. 5 and 6).43 As with the GloFish laboratory, students are encouraged to Also, like cyc,thesqt mutation is homozygous recessive. work together throughout the laboratory, but written as- Therefore, the sqt mutant phenotype is observed only in ﬁsh signments are to be completed individually by each student. with two copies of the mutant allele. However, the sqt in- The written assignments include the completion of a labora- heritance pattern is different from cyc, and thus this cross tory notebook entry for each week of the laboratory (Sup- brings a new aspect of Mendelian genetics to the classroom.

plementary Material 7), and a homework assignment The sqt phenotype is incompletely penetrant. In a cross be- (Supplementary Materials 12 and 13). tween two heterozygous carriers of the sqt mutant allele, 0% to 25% of the offspring of fish heterozygous for the sqt mutation display the mutant phenotype. The expressivity of the Monohybrid cross using the cyclopsm294 (cyc) mutation mutant phenotype is also variable. Some homozygous mu- In this experiment, two adult fish heterozygous for the cyc tants have severe defects, such as complete cyclopia and a mutation are crossed to produce a clutch of sibling progeny curved body axis, some mutants have mild defects such as (Fig. 9). The progeny are then scored according to their eye eyes that are closer together, and other homozygous mutants phenotype using light microscopy and the number of progeny are indistinguishable from their WT siblings.44–46 The pen- with each phenotype is counted. The first cross yields 78% etrance of the different aspects of each phenotype can also progeny with two WT eyes, and 22% progeny with one cy- vary, with an embryo, for example, having a curved body clopic eye (Fig. 9, Table 7, Supplementary Fig. 5). After com- axis but normal eyes.44–46 bining the progeny from several independent cyc crosses, 74% of the progeny have two normal eyes and 26% of the progeny have one cyclopic eye (Table 8). TheratiooffishwithaWTeyephenotypetothosewitha + sqt cyclopic eye phenotype is approximately 3:1 (Tables 7 and 8). This ratio should lead the students to pose the null hypothesis + +/+ +/sqt that the cyc mutation is recessive (Fig. 10). Chi-square analysis of the single cross generates a p-value between 0.5 and 0.6 (Table 7). The combined data from all the cyc crosses, which included sqtsqt/+ sqt/sqt 376 embryos, produces a p-value between 0.6 and 0.7 (Table 8). Thus, both chi-square tests support the hypothesis that cyc is a genotype phenotype recessive mutation. Further, as expected, the p-value becomes +/+ two eyes (WT) higher, and the hypothesis more strongly supported, as the sqt/+ two eyes (WT) number of embryos included in the analysis increases. sqt/sqt cyclopic eyes FIG. 12. Hypothesis for sqt cross. The hypothesis for the Monohybrid cross demonstrating inheritance pattern of the sqt cross is presented as a Punnett incomplete penetrance square generated as described for Figure 2. As with the cyc cross, this hypothesis predicts that three quarters of the The Sqt and Cyc proteins are 55% identical, and their genes progeny should have normal eyes and one quarter should are expressed in many overlapping domains in the develop- have cyclopic eyes. STATISTICAL ANALYSIS AND MENDELIAN GENETICS 51

Table 7. Chi-Square Analysis of a Single CYC Monohybrid Cross

(1)a (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O ED2 ¼ (O E)2 w2 ¼ D2/E

Wildtype 91 88 3.0 9.0 0.10 (two eyes) Cyclopic 26 29 3.0 9.0 0.31 (one eye) (7) Total w2 ¼ 0.41 (8) Degrees of freedom (n 1) ¼ 2–1 ¼ 1 (9) 0.5 < p < 0.6

aCalculations for each column and row carried out as in Table 2.

The percentage of homozygous mutants with an abnormal excellent springboard for discussions aimed at designing ex- phenotype can vary greatly from clutch to clutch. Because the periments to distinguish between this and other possible ex- cyclopic phenotype is the most common and the easiest to planations. identify, the embryos in this cross are scored only for their eye phenotype. In a single clutch generated by a single pair mat- Student feedback and assessment ing of adult fish heterozygous for the sqt mutation, 93% embryos have normal eyes, and 7% embryos have cyclopic eyes One of the key steps in both the GloFish and F3 Screen (Fig. 11, Table 9, Supplementary Fig. 6). From several com- protocols is testing student-generated hypotheses using sta- bined crosses, there are 98% normal embryos and 2% cyclopic tistical analysis. This step in the protocols enabled students to embryos (Table 10). make connections between their statistics courses and on- Students will likely find it challenging to generate a null the-ground experimental biology. This seemed to be much hypothesis for these data, as it does not easily fall into a needed, as informal surveys and our own postlaboratory as- standard Mendelian ratio. In our class, students typically sessment indicate that students find it challenging to match a choose to test the hypothesis that sqt, like cyc, is inherited as a p-value with a conclusion about the experiment (Fig. 13B). The recessive allele with a 3:1 WT:sqt ratio expected in the progeny GloFish and F3 Screen protocols introduce statistical analysis (Fig. 12). Chi-square analysis does not support this hypothe- into a sophomore level course, offering the opportunity to sis. For the single cross, the p-value is between 0.01 and 0.02 build on this initial exposure in subsequent courses. and for the combined crosses, p is less than 0.001. The finding Students reported that the protocols presented here were that the p-value decreases when increased numbers of em- valuable for cementing their learning of Mendelian genetics, bryos are analyzed also indicates that this hypothesis is not as they complemented and built upon the concepts they correct. This analysis of the sqt cross illustrates the ability of learned in their genetics lecture course. Because the experi- statistical analysis to disprove a null hypothesis. ments in our protocols use real clutches of progeny, they bring To prompt students to generate possible explanations for the real complexity of genetic analysis into the classroom. For the phenotype ratios in these crosses, they can be given a instance, in some crosses, more than one hypothesis is sup- follow-up homework that challenges them to generate a ported by statistical analysis (eg., GloFish Dihybrid cross 1). hypothesis that better fits these data (Supplementary Material In other crosses, the ratios of phenotypes in the progeny do 13). When this homework is given, we stress that it is meant to not fit well into any Mendelian ratio, and no hypotheses are be an ill-structured question with many potentially correct supported (eg., F3 Screen sqt crosses). Many students found answers. Students typically do very well with this homework, the lack of a clear outcome frustrating. Instructors could avoid with the most common answer being that this ratio comes this frustration by bringing only sibling groups that fit well from a dihybrid cross, with the parents being heterozygous into Mendelian ratios and support only one hypotheses. Al- for sqt and another recessive gene. Only embryos that are ternatively, this frustration can be a good learning experience homozygous for both mutant alleles express the mutant for the students as it offers the opportunity to think creatively. phenotype. A discussion of this homework can serve as an For instance, students very much enjoyed designing and

Table 8. Chi-Square Analysis for Combined CYC Monohybrid Crosses

(1)a (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O E D2 ¼ (O E)2 w2 ¼ D2/E

Wildtype 278 282 4.00 16.0 0.057 (two eyes) Cyclopic 98 94 4.0 16 0.17 (one eye) (7) Total w2 ¼ 0.23 (8) Degrees of freedom (n 1) ¼ 2–1 ¼ 1 (9) 0.6 < p < 0.7

aCalculations for each column and row carried out as in Table 2. 52 LINDEMANN ET AL.

Table 9. Chi-Square Analysis of a Single SQT Monohybrid Cross

(1)a (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O ED2 ¼ (O E)2 w2 ¼ D2/E

Wildtype 26 21.0 5.0 25 1.2 (two eyes) Cyclopic 2 7 5254 (one eye) (7) Total w2 ¼ 5 (8) Degrees of freedom (n 1) ¼ 2–1 ¼ 1 (9) 0.01 < p < 0.02

aCalculations for each column and row carried out as in Table 2. carrying out their own GloFish cross (Fig. 13C, Supplemen- female or male is determined by multiple genes located on tary Materials 5 and 6). autosomes, although the specific genes are just starting to be A voluntary assessment survey completed by students after isolated.48,49 Environmental factors, such as temperatures out- finishing all but the last step of the GloFish laboratory (they side of the normal range and addition of hormones, can also had not yet observed the progeny from their own crosses) in- influence gender.49 Because of these many contributing factors, dicated a positive experience (Fig. 13). This feedback suggested it is quite common for clutches of sibling fish to be mostly male that students found the analysis and thinking parts of the or mostly female, even when fish are raised in a relatively laboratory most valuable for their understanding of genetics, controlled laboratory setting. Thus, comparing the gender ratios and the work needed to gather the data the least valuable in different clutches can illustrate that the inheritance of even (Fig. 13C). Most students agreed or strongly agreed that the some major phenotypes, like whether a fish is male or female, laboratory increased their confidence with the concepts used, may defy analysis by Mendelian genetics (Supplementary Figs. suggesting that it enhanced student learning (Fig. 13A). 1, 7–9, Supplementary Tables 1 and 2). Uncovering the expla- nation for the variation in gender ratios from clutch to clutch can form a challenging puzzle for upper-level students. Their re- Discussion search into this subject could be an entry into the sophisticated Building on the GloFish laboratory approaches that have been used to uncover traits and diseases

that are controlled by multiple genes. The difficulty of this GloFish protocol can be adapted for Finally, our protocol is aimed at early stage college students, students carrying out their first experiments in genetics to but there are many other possibilities for building GloFish into students beginning their graduate studies. For elementary upper level undergraduate and even graduate courses. For school students, the simplest GloFish crosses (such as Dihy- instance, the GloFish transgenes encode GFP and other related brid cross 1) can be used to illustrate how a single gene can fluorescent proteins (Fig. 14). In 2008, Drs. Osamau Shimomura, dramatically change the phenotype of an organism. For Martin Chalfie, and Roger Tsien won the Nobel Prize in middle and high school students, GloFish can be used to ex- Chemistry for their research on GFP (http://nobelprize.org/ plore how genes interact to produce phenotypes. Further, nobel_prizes/chemistry/laureates/2008/). This protocol on because this protocol includes only pre-algebra mathematics, GloFish could serve as an introduction to this Nobel Prize– the use of Mendelian ratios to calculate expected proportions winning research and how transgenic/genetic engineering of progeny phenotypes and statistical testing of hypotheses approaches are being used to treat human disease. could be easily introduced into pre-college science courses. This protocol could also be expanded by including analysis Building on the F genetic screen laboratory of traits that are not only influenced by genetics. Rather than the 3 simple X/X ¼ female, X/Y ¼ male sex determination in mam- The F3 Genetic Screen protocol can be used to introduce the mals, sex determination in fish is extremely varied.47 In the case concept of model organisms and to teach developmental ge- of zebrafish, there are no sex chromosomes. Whether a fish is netics, embryology, and the genetics of human birth defects and

Table 10. Chi-Square Analysis of Combined SQT Clutches

(1)a (2) (3) (4) (5) (6) Phenotype Observed number (O) Expected number (E) Difference (D) ¼ O ED2 ¼ (O E)2 w2 ¼ D2/E

Wildtype 1740 1330 410.0 1.68105 126 (two eyes) Cyclopic 35 443 408 1.66105 376 (one eye) (7) Total w2 ¼ 502 (8) Degrees of freedom (n 1) ¼ 2–1 ¼ 1 (9) p < 0.001

aCalculations for each column and row carried out as in Table 2. STATISTICAL ANALYSIS AND MENDELIAN GENETICS 53

FIG. 13. Student Assessment of the GloFish Laboratory. Students in the spring 2011 semester of Genetics Laboratory were asked to fill out an assessment of the GloFish laboratory approximately 4 weeks after they had completed all but the last step (observation of the progeny from their own cross, which does not take place until the end of the semester). (A) Assessment of value of the GloFish laboratory for their own learning. (B) Assessment of whether students were able to correctly define key concepts introduced in this laboratory. (C) Student feedback on which aspects of the laboratory were most and least useful for their learning. Al- though students were asked to choose at most two answers under each of the ‘‘most’’ and least’’ categories, many students missed these instructions and checked all of the responses that they thought answered the questions. The design of this survey was based in part on the guidelines at http://tep.uoregon.edu/re- sources/newteach/fifty_cats.pdf

FIG. 14. GloFish fluorescence. (A) Yel- low GloFish under white light. (B) Yellow GloFish under blue light/GFP filter. (C) Head of yellow GloFish under blue light/ GFP filter. (D) Red GloFish under white light. (E) Red GloFish under green light/ RFP filter. (F) Head of red GloFish under green light/RFP filter. (G) Orange Glo- Fish under white light. (H) Orange Glo- Fish under blue light/GFP filter. (I) Orange GloFish under green light/RFP filter. Images are lateral views, anterior to the left and dorsal to the top. GFP, green fluorescent protein; RFP, red fluorescent protein. 54 LINDEMANN ET AL. human disease. For instance, many disease genes, like sqt,have 5. Schmoldt A, Forecki J, Hammond DR, Udvadia AJ. Ex- incomplete penetrance. Mutations in the BRCA1 and BRCA1 ploring differential gene expression in zebrafish to teach genes are associated with increase risk of breast cancer, but are basic molecular biology skills. Zebrafish 2009;6:187–199. in no way indications that 100% of the women carrying these 6. Hutson LD, Liang JO. Making an impact: zebrafish in edu- mutations will develop the disease.50 Environmental influences cation. Zebrafish 2009;6:119–120. as well as genetic background influence whether the women 7. Emran F, Brooks JM, Zimmerman SR, Johnson SL, Lue RA. carrying the mutation will go on to develop breast cancer.50 Zebrafish embryology and cartilage staining protocols for Likewise, studies by Dr. Ben Feldman and colleagues have high school students. Zebrafish 2009;6:139–143. shown that environmental factors and genetic background 8. Nusslein-Volhard C, Dahm R. Zebrafish: A Practical Ap- similarly affect the penetrance of the sqt mutant phenotype.46 proach. Oxford, UK: Oxford University Press, 2002;261. 9. Fukada Y, Okano T. Circadian clock system in the pineal The F Genetic Screen protocol is used as part of a larger set 3 gland. Mol Neurobiol 2002;25:19–30. of laboratories that give students exposure to forward and 10. Westerfield M. The Zebrafish Book. Eugene, OR: University reverse genetics (Supplementary Material 9). To complement of Oregon Press, 2000. these experiments in zebrafish, students also carry forward 11. Gross LJ. Interdisciplinarity and the undergraduate biology genetics experiments on Drosophila. In particular, we have curriculum: finding a balance. Cell Biol Educ 2004;3:85–87. found that the antennapedia mutant flies available from Car- 12. Cohen JE. Mathematics is biology’s next microscope, only olina Biological (www.carolina.com) have characteristics better; biology is mathematics’ next physics, only better. similar to sqt: the phenotype (antennas are replaced by legs) is PLoS Biol 2004;2:e439. incompletely penetrant and has variable expressivity. Since 13. Hoy R. New math for biology is the old new math. Cell Biol Drosophila have the advantage of a very fast generation time, Educ 2004;3:90–92. students are able to follow the penetrance and expressivity 14. Brent R. Intuition and innumeracy. Cell Biol Educ 2004;3: of the phenotype over several generations and generate 88–90. and test hypotheses about the underlying causes of the vari- 15. Bialek W, Botstein D. Introductory science and mathematics ability. Students gain experience in reverse genetics by feed- education for 21st-Century biologists. Science 2004;303:788– ing C. elegans worms with bacteria carrying different 790. constructs that make double-stranded RNA, and by injecting 16. Derting, T. L, Ebert-May D. Learner-centered inquiry in un- antisense morpholinos into zebrafish (www.zfic.org/ dergraduate biology: positive relationships with long-term classroom%20experiments/microinjectionindex.html).51–53 student achievement. CBE Life Sci Educ 2010;9:462–472. 17. Schier AF, Neuhauss SC, Harvey M, Malicki J, Solnica- Acknowledgments Krezel L, Stainier DY, et al. Mutations affecting the development of the embryonic zebrafish brain. Development This work was supported in part through University of 1996;123:165–178. Minnesota Undergraduate Research Opportunity (UROP) 18. Feldman B, Gates MA, Egan ES, Dougan ST, Rennebeck G, Awards to S.L. and J.S. Publication costs were supported by Sirotkin HI, et al. Zebrafish organizer development and the UROP Program, and the University of Minnesota Duluth germ-layer formation require nodal-related signals. Nature Swenson College of Science & Engineering and Department of 1998;395:181–185. Biology. We would like to thank Dr. Paul Bates for his many 19. Gong Z, Wan H, Tay TL, Wang H, Chen M, Yan T. Devel- excellent suggestions for improving the article, Tonya Connor opment of transgenic fish for ornamental and bioreactor by for her expert advice on the student assessment survey for the strong expression of fluorescent proteins in the skeletal GloFish protocol, and Adelle Schumann for her technical muscle. Biochem Biophys Res Commun 2003;308:58–63. support. In addition, we would like to thank the teaching 20. Lamason RL, Mohideen MA, Mest JR, Wong AC, Norton assistants and students of the Genetics Laboratory course for HL, Aros MC, et al. SLC24A5, a putative cation exchanger, piloting this protocol and for all of their valuable ideas for affects pigmentation in zebrafish and humans. Science improvement, and all of the students who have helped with 2005;310:1782–1786. maintenance of the fish and fish facility. 21. Streisinger G, Walker C, Dower N, Knauber D, Singer F. Production of clones of homozygous diploid zebra fish Disclosure Statement (Brachydanio rerio). Nature 1981;291:293–296. 22. Tresnake I. The long-finned zebra Danio. Tropical Fish No competing financial interests exist. Hobby 1981;29:43–56. References 23. Johnson SL, Weston JA. Temperature-sensitive mutations that cause stage-specific defects in Zebrafish fin regenera- 1. Babcock EB, Collins JL. Genetics Laboratory Manual. New tion. Genetics 1995;141:1583–1595. York: McGraw-Hill Book Company, Inc., 1918. 24. Carvalho AP, Araujo L, Santos MM. Rearing zebrafish 2. Fields MC, Adelfio P, Ahmad D, Brown O, Cox B, Davies M, (Danio rerio) larvae without live food: evaluation of a et al. Danio rerio in K-12 classrooms: sparking interest in the commercial, a practical, and a purified starter diet on larval new generation of scientists. Zebrafish 2009;6:145–160. performance. Aquac Res 2006;37:1107–1111. 3. D’Costa A, Shepherd IT. Zebrafish development and ge- 25. Ju B, Chong SW, He J, Wang X, Xu Y, Wan H, et al. Re- netics: introducing undergraduates to developmental bi- capitulation of fast skeletal muscle development in zebrafish ology and genetics in a large introductory laboratory class. by transgenic expression of GFP under the mylz2 promoter. Zebrafish 2009;6:169–177. Dev Dyn 2003;227:14–26. 4. Shuda J, Kearns-Sixsmith D. Outreach: empowering stu- 26. Xu Y, He J, Wang X, Lim TM, Gong Z. Asynchronous acti- dents and teachers to fish outside the box. Zebrafish 2009; vation of 10 muscle-specific protein (MSP) genes during 6:133–138. zebrafish somitogenesis. Dev Dyn 2000;219:201–215. STATISTICAL ANALYSIS AND MENDELIAN GENETICS 55

27. Goldsmith MI, Fisher S, Waterman R, Johnson SL. Saltatory terning and establishing left-right asymmetry. Dev Biol control of isometric growth in the zebrafish caudal fin is 1998;199:261–272. disrupted in long fin and rapunzel mutants. Dev Biol 43. Liang JO, Etheridge A, Hantsoo L, Rubinstein AL, Nowak 2003;259:303–317. SJ, Izpisua Belmonte JC, et al. Asymmetric nodal signaling in 28. Geraudie J, Monnot MJ, Brulfert A, Ferretti P. Caudal fin the zebrafish diencephalon positions the pineal organ. De- regeneration in wild type and long-fin mutant zebrafish is velopment 2000;127:5101–5112. affected by retinoic acid. Int J Dev Biol 1995;39:373–381. 44. Dougan ST, Warga RM, Kane DA, Schier AF, Talbot WS. 29. van Eeden FJ, Granato M, Schach U, Brand M, Furutani- The role of the zebrafish nodal-related genes squint and Seiki M, Haffter P, et al. Genetic analysis of fin formation in cyclops in patterning of mesendoderm. Development the zebrafish, Danio rerio. Development 1996;123:255–262. 2003;130:1837–1851. 30. Iovine MK, Johnson SL. A genetic, deletion, physical, and 45. Aquilina-Beck A, Ilagan K, Liu Q, Liang JO. Nodal signaling human homology map of the long fin region on zebrafish is required for closure of the anterior neural tube in zebra- linkage group 2. Genomics 2002;79:756–759. fish. BMC Dev Biol 2007;7:126. 31. Hatta K, Kimmel CB, Ho RK, Walker C. The cyclops mu- 46. Pei W, Williams PH, Clark MD, Stemple DL, Feldman B. tation blocks specification of the floor plate of the zebrafish Environmental and genetic modifiers of squint penetrance central nervous system. Nature 1991;350:339–341. during zebrafish embryogenesis. Dev Biol 2007;308: 32. Haffter P, Granato M, Brand M, Mullins MC, Hammersch- 368–378. midt M, Kane DA, et al. The identification of genes with 47. Desjardins JK, Fernald RD. Fish sex: why so diverse? Curr unique and essential functions in the development of the Opin Neurobiol 2009;19:648–653. zebrafish, Danio rerio. Development 1996;123:1–36. 48. Tong SK, Hsu HJ, Chung BC. Zebrafish monosex population 33. Driever W, Solnica-Krezel L, Schier AF, Neuhauss SC, reveals female dominance in sex determination and earliest Malicki J, Stemple DL, et al. A genetic screen for mutations events of gonad differentiation. Dev Biol 2010;344:849–856. affecting embryogenesis in zebrafish. Development 49. Orban L, Sreenivasan R, Olsson PE. Long and winding 1996;123:37–46. roads: testis differentiation in zebrafish. Mol Cell Endocrinol 34. Stemple DL, Driever W. Zebrafish: tools for investigating 2009;312:35–41. cellular differentiation. Curr Opin Cell Biol 1996;8:858–864. 50. Beckmann MW, Bani MR, Fasching PA, Strick R, Lux MP. 35. Schier AF, Talbot WS. Nodal signaling and the zebrafish Risk and risk assessment for breast cancer: molecular and organizer. Int J Dev Biol 2001;45:289–297. clinical aspects. Maturitas 2007;57:56–60. 36. Alexander J, Stainier DY. A molecular pathway leading to 51. Timmons L, Court DL, Fire A. Ingestion of bacterially ex- endoderm formation in zebrafish. Curr Biol 1999;9:1147–1157. pressed dsRNAs can produce specific and potent genetic 37. Liang JO, Rubinstein AL. Patterning of the zebrafish embryo interference in Caenorhabditis elegans. Gene 2001;263:103– by nodal signals. Curr Top Dev Biol 2003;55:143–171. 112. 38. Halpern ME, Liang JO, Gamse JT. Leaning to the left: la- 52. Kamath RS, Martinez-Campos M, Zipperlen P, Fraser AG, terality in the zebrafish forebrain. Trends Neurosci Ahringer J. Effectiveness of specific RNA-mediated 2003;26:308–313. interference through ingested double-stranded RNA in 39. Erter CE, Solnica-Krezel L, Wright CV. Zebrafish nodal-re- Caenorhabditis elegans. Genome Biol 2001;2:RESEARCH0002. lated 2 encodes an early mesendodermal inducer signaling 53. Nasevicius A, Ekker SC. Effective targeted gene ‘knock- from the extraembryonic yolk syncytial layer. Dev Biol down’ in zebrafish. Nat Genet 2000;26:216–220. 1998;204:361–372. 40. Sampath K, Rubinstein AL, Cheng AM, Liang JO, Fekany K, Solnica-Krezel L, et al. Induction of the zebrafish ventral Address correspondence to: brain and floorplate requires cyclops/nodal signalling. Jennifer O. Liang, Ph.D. Nature 1998;395:185–189. Department of Biology 41. Rebagliati MR, Toyama R, Haffter P, Dawid IB. cyclops University of Minnesota Duluth encodes a nodal-related factor involved in midline signaling. 1035 Kirby Drive, Rm. 207 Proc Natl Acad Sci U S A 1998;95:9932–9937. Duluth, MN 55812 42. Rebagliati MR, Toyama R, Fricke C, Haffter P, Dawid IB. Zebrafish nodal-related genes are implicated in axial pat- E-mail: [email protected]