A Phenotype-driven Approach to the Molecular and Functional Analysis of the Mouse Genome
Dabney K. Johnson, Donald A. Carpenter, Cymbeline T. Culiat, Karen A. Goss, Mitchell L. Klebig, Edward J. Michaud, Darla R. Miller, Liane B. Russell, Yun You, and Eugene M. Rinchik
Mammalian Genetics and Development Section
Life Sciences Division
Oak Ridge National Laboratory
Oak Ridge, TN
The mouse genetics and mutagenesis program at Oak Ridge National Laboratory (ORNL) is employing chemical mutagenesis and broad-based phenotype screening to recover mutations targeted to specific chromosome regions for ascertaining the whole organism functions of mouse genes. Our strategy of chemical mutagenesis results in pedigrees of mice, each harboring a different DNA mutation for one of the many genes contained in the chromosome region. These mice are then subjected to a broad range of tests to identify a mutant phenotype. This phenotype-driven approach is being applied initially to about 8% of the mouse genome and is adaptable to any genome region as the necessary genetic resources are developed. However, the efficiency with which we can discover genes using this strategy hinges on our proficiency in detecting abnormal phenotypes in the progeny of mutagenized mice. To this end, we have designed broad-based, high-throughput screening assays that are performed on multiple animals from the same pedigree to identify obvious or subtle aberrations in behavior, biochemistry, and/or morphology in mice at young and old ages. Eventually, DNA sequence will be coregistered with functional information for each gene using mouse mutations as the gene-discovery tools and the phenotypes those mutations specify as indicators of gene function.
The mouse genetics program at ORNL began in 1947 under the direction of Dr. William L. Russell. Since then, experimental mutagenesis has been performed mostly to assess genetic risk from exposure to a variety of radiations and
chemicals and has been focused on seven specific loci (Russell 1951) that generated visible phenotypes when mutated. In the mid-1970s, Dr. Russell discovered that the chemical N-ethyl-N-nitrosourea (ENU) is a supermutagen for mouse spermatogonial stem cells (Russell and others 1979), inducing primarily point mutations (single base pair substitutions) (Russell and Montgomery 1982) and thus a variety of types of mutations including nulls and hypomorphic alleles (Bedell and others 1996; Ji and others 1999; Marker and others 1997). In 1986, one of us (E.M.R) launched a pilot ENU-mutagenesis experiment (Rinchik and Carpenter 1999) focused at the albino (c; now called Tyr [tyrosinase]) locus in mouse chromosome (Chr) 7. This region is covered by an extensive series of radiation-induced deletion mutations resulting from the Russell specific locus tests. The mutagenesis strategy for the Tyr region, adapted from a similar approach used by Drosophila geneticists, was to mate ENU-mutagenized males (BALB/cR1) that are homozygous for the c coat-color marker to wild-type females (Rinchik and Carpenter 1999). The F1 mice bearing the BALB/c Chr 7 carrying newly induced point mutations, some of which will be closely linked to c, were then mated to carriers of a large “selector” deletion at c. Of the progeny from this second mating, 25% should be albino and may also express an additional new mutant phenotype if an ENU mutation is so closely linked to c that it maps within the limits of the large c deletion. A set of simple complementation crosses to smaller c deletions localized new mutant phenotypes to intervals suitable for a positional-cloning approach (Rinchik and Carpenter 1993, 1999; Rinchik and others 1993). A similar experiment for the deletion complex surrounding the pink-eyed dilution (p) locus, also on Chr 7, is currently under way (Johnson and others 1995; Rinchik and others 1995; Rinchik, Carpenter, and Johnson, manuscript in preparation). Our successes in designing the genetics and logistics of large mutagenesis experiments have led to the establishment of our current program of inducing new mutations in the proximal two thirds of Chr 7 (which includes the p and c regions), the central one half of Chr 10, the distal half of Chr 15, and a small segment of the X chromosome.
In all of these experiments, broad-based screening for the detection of new mutant phenotypes plays a new and very prominent role. The current program has benefited from our hands-on experience as we increase both the chromosomal region target sizes and our scope and capacity for examining mice for as many different kinds of abnormalities as possible. Our current experiments take advantage of deletion screens, as described above, and more powerful methods utilizing chromosomal inversions as tools to make newly mutagenized chromosomes homozygous —all without molecular genotyping.
EXPERIENCE APPLIED TO NEW EXPERIMENTS
We have learned three important lessons from pilot ENU experiments that have influenced the much broader mutagenesis and phenotype-screening pro
grams now under way. These lessons relate to mutation recovery rate, visual genotyping, and the importance of genetic reagents for targeting mutagenesis.
Mutation Recovery Rate
In the pilot experiment at c, using the 6- to 11-cM deletion c26DVT (Rinchik and Carpenter 1999) as the “selector deletion,” only visible and lethal phenotypes were ascertained. Even so, 31 new mutations were recovered in 4557 gametes (pedigrees) screened for a mutation recovery rate of one in 147 pedigrees tested (Rinchik and Carpenter 1999). We know that mutagenesis and phenotype screening within different regions will result in the recovery of more or fewer mutations due to gene density differences in the regions, and/or to the proportion of genes in the regions that can mutate to a visible or lethal phenotype. Thus, whereas gene density is fixed, we should be able to affect mutation recovery rate by expanding the number and kinds of phenotype screens to increase the proportion of genes for which we can detect a mutation.
Having the progeny class that carries no wild-type copy of the newly induced mutation (i.e., “test class” mice that have one deletion chromosome and one mutagenized chromosome marked by carrying a visible marker like c or that is homozygous for the mutagenized chromosome) provides several important advantages: (1) It eliminates the need for molecular genotyping, an expensive, error-prone, and logistically difficult procedure. (2) It permits 100% ascertainment of lethals, evident when the visibly marked “test” class is absent in progeny or does not survive as long as other genotypes. In our pilot experiments, about half of all new mutations are lethals, which would go undetected if progeny genotypes could not be distinguished by external phenotype. (3) It allows the easy production and testing of multiple test class animals, all carrying the same mutagenized chromosome, for assay in tests with highly variable parameters (such as behavioral tests) or testing in multiple sites. In our program, four test class progeny from each pedigree go through phenotype screening, with the requirement that all four show the variant phenotype before being designated “mutant” and bred for transmissibility of the trait. (4) It likewise allows for the shelving of a set of test class progeny for aging and retesting for late onset recessive mutant phenotypes.
Importance of Genetic Reagents for Targeting Mutagenesis
Although the Tyr- and p-region experiments have been quite successful, it is clear that chromosomal deletions are too small to be used exclusively as selectors for new mutations genome-wide. Deletions are clearly indispensable for comple
mentation crosses to localize new phenotypes within the genome region of interest but generally cannot be large enough (due to the negative effects of large haploinsufficiencies) for efficient genome-wide mutagenesis. Chromosomal inversions, on the other hand, can be larger, and cover a much larger region of chromosome. When suitably marked for visual genotyping (Rinchik 2000), inversions also suppress recombination events that might separate the new mutation from the linked visible “tracking” marker. Accompanied by overlapping sets of nested chromosomal deletions generated in vitro using molecular techniques for the modification of embryonic stem cell chromosomes (Ramirez-Solis and others 1995; Thomas and others 1998; You and others 1997), large inversions become ideal tools for regional mutagenesis. As the mouse and human DNA sequences are acquired and analyzed for these mutagenized regions, we can begin to integrate physical and functional gene maps.
ROBUST AND BROAD-SPECTRUM PHENOTYPE SCREENING
Over the past 2 years, we have employed a basic set of primary phenotypescreening tests to enhance our mutation-recovery rate by detecting mutant phenotypes that are not apparent upon routine observation of test class mice. We have recently expanded and updated our instrumentation to automate our screening as much as possible to accommodate the anticipated load from new and ongoing experiments. Furthermore, all pedigrees will be rescreened at 18 months of age to ascertain later onset abnormalities. Tests in current use have been validated by testing mice that we expected to show a mutant phenotype; we have also used these tests to screen nearly 2000 mice on a high-throughput basis and have identified new mutant phenotypes that would have escaped detection with our previous methods. We are accumulating and testing new equipment fairly constantly and replacing older or less efficient instrumentation as we can. The primary screen is the only opportunity to detect subtle anomalies, so it is crucial that this screen be comprehensive, well-grounded, practical, reliable, and capable of performing as promised.
Because we do have multiple test class animals in every pedigree to screen, we can rely on replication of any abnormality; and with heritability testing, false positives are minimized, even with highly variable traits. We are also certain, by visual genotyping, that we are performing our screens only on test class mice that can be expected to exhibit recessive mutant phenotypes, thus maximizing throughput and minimizing cost. Comprehensive screening, realistically designed, gives us greater power to discover mutant phenotypes of interest that historically have slipped through the cracks. We have planned our screening so that whenever possible, more than one assessment tool will target each category of mutant phenotype of interest to minimize further both false positives and false negatives.
In the establishment of a comprehensive set of phenotype screens, we have chosen tests that are reliable, practical, easily automated, and likely to detect phenotypes of interest to neuroscientists. Individual behavioral tests are rarely definitive measures of a single neurologic process. However, by employing multiple tests, we can develop a pattern of response across tasks to inform us about a particular process. For example, a mouse with heightened anxiety but no memory deficit may perform poorly in the Y-maze spontaneous alternation memory task but show improved performance in the conditioned freezing memory task. However, a mouse with an actual learning deficit should perform poorly on both tests, allowing us to be more confident that the impairment is one of memory and not a nonmnemonic process. In a similar manner, we can dissect sensory and motor components of an aberrant behavior while still employing a test set that can actually be accomplished in an efficient and high-throughput manner.
For each pedigree generated from most mutagenesis experiments, four test class mice (usually two males and two females) first go through the weaning screen at 21 to 25 days of age (P21-25), next the primary screen at P50-60, and then are aged along with two additional males and two additional females for rescreening at 18 months (P548-560). Two pairs of the mice are mated for fertility testing and later separated for storage or sent for further testing if sterility is observed in either sex. Mice are examined at weaning (P21-25) for visible aberrations (overall size and proportionality, external genitalia, limbs, digits and tails, eyes and ears, fur color and quality, posture and gait). The remaining tests are performed over 3 days, in order from the least aversive to the most aversive to avoid intertest impact. Mice rest for at least 1 hour between tests, and all testing is done between 8:00 a.m. and 4:00 p.m. We test approximately 100 mice per week at P50-60 and later retest them at 18 months (P548-560). If recessive visible or lethal phenotypes are apparent, breeding stocks are established without further primary screening. The tests currently used to identify alterations in these traits are outlined in Table 1.
The general flow of mice, when they are 50 days of age and again at 18 months of age, through the behavioral/central nervous system screen is described below. The process is designed to minimize interest impact.
Each mouse is weighed, and simple, gross neurologic observations are performed (reaching reflex, vibrissae response, righting reflex). Next is the 2-minute rotorod test (Accuscan SmartRod, Columbus, Ohio 1) for balance and
aOne paper used both.
Identification of commercial products or manufacturers does not constitute an endorsement by the authors or the National Research Council.
TABLE 1 Primary Behavioral/Central Nervous System Screening Tests
Observation for physical or behavioral abnormalities by trained staff
Head turn in response to stimulation of the whiskers by a pen tip
Lower mice toward flat surface and observe for reaching response
Pupillary reflex in response to light
ICR click box; pinna response noted
Locomotor activity and rearing
Rotorod latency to fall, 4-minute acceleration test
Cued/contextual conditioned freezing, 4-chamber inserts for motor activity system
Startle magnitude, habituation to startle, and prepulse inhibition assessed in a single session.
MicroCT scanner, 1 mouse per pedigree
coordination, designed to measure ability to maintain position on an accelerating rotating dowel. Four mice at a time are then videotaped for 3 minutes (preceded by a 2-minute habituation period) in individual 24”×24” opaque chambers open at the top. Observed behaviors (amount and patterns of movement) are analyzed using software accompanying the Polytrack Video System (San Diego Instruments, San Diego, California) to automatically compute relative dwell time for each user-defined portion of the open-field chamber. This is followed by assessment of acoustic startle response, prepulse inhibition of startle (PPI), habituation to startle in a single session in an automated system (Hamilton-Kinder Co., Poway, California), and return to the home cage.
Open-field activity is measured by counting interruptions of a set of photo-beams over a 20-minute test period to provide information on exploratory and motor behaviors (Hamilton-Kinder). Rearings are monitored, as are total activity count (beams broken) and locations of beam breaks within the enclosure. Next is the tail-flick test, in which latency to react to a heat stimulus is measured by
confining the tail of the mouse in a slot and timing the latency to tail withdrawal after heat is applied; the mouse is removed within 10 seconds if it does not appear to perceive the heat. Last on Day 2 is the conditioned-freezing training trial. The open-field activity system includes 9”×9”×9” opaque white plexiglas inserts that fit into the photobeam frame; the bottom of the insert is an electric grid. Mice are lifted into the insert, confined with an opaque lid, and allowed to acclimate for 2.5 minutes. They are then presented with an 85-dB tone (3000 Hz) for 30 seconds and then a 2-second 0.4-mA footshock, followed by a 2-minute recording of posttone/postshock activity. The mouse is then removed to its home cage.
The first test is the 24-hour conditioned-freezing memory retention trial, in which the mouse is returned to the same chamber that was used for the training trial and allowed to explore for 3 minutes while its movement is monitored to detect freezing behavior. The mouse is removed to the home cage, and the chamber context is changed by insertion of a black liner. The mouse is returned for a 2-minute free exploration period, followed by repetition of the 30-second tone (no shock). Movement is again monitored for 3 minutes to detect freezing. Data are in the form of the total count of photobeam breaks in the first-day free exploration (presound and shock) compared with photobeam breaks the second day for either the 3-minute context test or the 3-minute postsound cue test. By this method of analysis (that is, comparing the mouse with itself in the training vs. retention trials), correction is made for the innate activity level of the individual mouse strain.
Biochemical, Physiologic, and Molecular Tests
Blood, urine, and various tissue samples are harvested from test class mice for a broad array of biochemical tests. Numerous tissues from one male and one female from each pedigree are taken at P60 and frozen for future primary or secondary screens. In this way, tissues can be analyzed retrospectively as additional genome/complex pathway information becomes available. Sperm will also be cryopreserved in the event that later recovery of the pedigree through in vitro fertilization is required. Using fluid or tissue samples from the test class allows efficient use of test class mice inasmuch as samples from any mouse can feed a large number of tests; we also archive a variety of tissues from each pedigree for future primary or secondary screens.
Tests in current use as primary screens for all pedigrees include characterizations of 12 hematologic parameters using a Cell-dyne 3500 Hematology Analyzer (Santa Clara, California) and measurement of six factors in urine using a Bili-Labstix dipstick (Bayer Corp., Elkhart, Indiana). We have recently acquired
the capability to measure blood levels of glucose, cholesterol, and triglycerides and will begin immediately to acquire these values for all test class mice.
ORNL engineers have developed a small animal computed axial tomography X-ray scanner (MicroCT; Oak Ridge National Laboratory, Oak Ridge, Tennes-see) with image reconstruction software and rudimentary organ recognition software for use in phenotype screening. This instrument takes a whole-body scan at less than 1-mm resolution in about 6 minutes and can resolve both soft tissues and skeleton. It is currently equipped with kidney recognition algorithms, and algorithms are under development for lean body mass and whole body fat content determination. Significant skeletal malformations (such as scoliosis) will also be detected from examination of these reconstructed and stored CT images. One test class mouse anesthetized by isoflurane is imaged from each pedigree, with all images stored on CD-ROM.
Tennessee Mouse Genome Consortium (TMGC)
ORNL is a charter member of the newly formed TMGC, a statewide organization designed for enhanced phenotype screening and analysis. Clinical and academic experts from the University of Tennessee Medical Center at Memphis, the University of Tennessee at Knoxville, Vanderbilt University, St. Jude's Children's Research Hospital, the University of Memphis, and Meharry Medical College have joined forces to screen ORNL test class mice comprehensively for a wide variety of phenotypes. Early efforts have concentrated on phenotyping for abnormalities in the nervous system (behavior, drug sensitivity, neuroanatomy, sensory organs, neurochemical pathways, sleep/wake cycles) and are now expanding into heart, blood, and lung phenotypes. Domain experts will also continue phenotype analysis once a mutation is identified. Because ORNL can generate multiple animals per pedigree, live mice or samples from mice can travel statewide (or farther) for primary and secondary screening programs that greatly increase our opportunities for realizing the highest possible mutation recovery rate from our mutagenesis program.
Statistical Analysis and Flagging Mutant Mice
We use the test results from all other test class animals from the same mutagenesis experiment as controls inasmuch as all mice are handled the same and the majority of pedigrees yield normal test class progeny for any given phenotype. This allows us to establish a criterion of two standard deviations from the mean from a very large population, giving us quite adequate statistical power with which to flag a mutant pedigree while testing only four (or even fewer) mice
from that pedigree. If at least three of the four tested show a deviant phenotype, we send that pedigree for transmissibility testing. In our experience, all four will show the variant phenotype, but we take note that we do have a segregating genetic background that could affect the phenotype in any one animal. Potential mutants are first tested for transmissibility of the trait. If the trait proves heritable, then several things can occur:
a breeding stock is established;
the stock can be rederived so that secondary screening of interest can be done in an SPF facility if appropriate, inasmuch as our current colony is conventional;
test class mice are sent to interested researchers for confirmation and more detailed secondary/tertiary screening and analysis of the phenotype and gene-cloning; and
all primary data are currently entered into Excel spreadsheets, and from there will become part of a laboratory information system under development by ORNL's Computational Biosciences/Bioinformatics group.
The goal of this large-scale mouse mutagenesis and phenotype screening program is to annotate DNA sequence with experimentally derived functional information about how individual genes perform in the context of a whole mammalian organism. We have initially undertaken mutagenesis of about 8% of the mouse genome, distributed in different genome regions depending on the current availability of appropriate genetic resources. Resource building continues as a fundamental part of this program to facilitate the application of this strategy to the rest of the mouse genome. Expansion and enhancement of our phenotype screening capabilities are integral components of the entire effort as we continue to extend our capabilities in the detection of all abnormalities that mutant mouse genes can exhibit.
This manuscript was written under contract DE-AC05-96OR22464 with the US government, which retains a nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for US government purposes.
Bedell, M.A., N.G. Copeland, and N.A. Jenkins. 1996. Multiple pathways for Steel regulation suggested by genomic and sequence analysis of the murine Steel gene. Genetics 142:927-934.
Ji, Y., M.J. Walkowicz, K. Buiting, D.K. Johnson, R.E. Tarvin, E.M. Rinchik, B. Horsthemke, L. Stubbs, and R.D. Nicholls. 1999. The ancestral gene for transcribed, low-copy repeats in the Prader-Willi/Angelman region encodes a large protein implicated in protein trafficking, which is deficient in mice with neuromuscular and spermiogenic abnormalities. Hum. Mol. Genet. 8:533-542.
Johnson, D.K., L.J. Stubbs, C.T. Culiat, C.S. Montgomery, L.B. Russell, and E.M. Rinchik. 1995. Molecular analysis of 36 mutations at the mouse pink-eyed dilution (p) locus. Genetics 141:1563-1571.
Marker, P.C., K. Seung, A.E. Bland, L.B. Russell, and D.M. Kingsley. 1997. Spectrum of Bmp5 mutations from germline mutagenesis experiments in mice. Genetics 145:435-443.
Ramirez-Solis, R., P. Liu, and A. Bradley. 1995. Chromosome engineering in mice. Nature 378:720-724.
Rinchik, E.M. 2000. Developing genetic reagents to facilitate recovery, analysis, and maintenance of mouse mutations. Mamm. Genome 7:489-499.
Rinchik, E.M., and D.A. Carpenter. 1993. N-ethyl-N-nitrosourea-induced prenatally lethal mutations define at least two complementation groups within the embryonic ectoderm development (eed) locus in mouse chromosome 7. Mamm. Genome 4:349-353.
Rinchik, E.M., and D.A. Carpenter. 1999. N-Ethyl-N-Nitrosourea mutagenesis of a 6- to 11-cM subregion of the Fah-Hbb interval of mouse chromosome 7: Completed testing of 4557 gametes and deletion mapping and complementation analysis of 31 mutations . Genetics 152:373-383.
Rinchik, E.M., D.A. Carpenter, and M.A. Handel. 1995. Pleiotropy in microdeletion syndromes: Neurologic and spermatogenic abnormalities in mice homozygous for the p6H deletion are likely due to dysfunction of a single gene. Proc. Natl. Acad. Sci. U S A 92:6394-6398.
Rinchik, E.M., D.A. Carpenter, and C.L. Long. 1993. Deletion mapping of four loci defined by N-ethyl-N-nitrosourea-induced postimplantation-lethal mutations within the pid-Hbb region of mouse chromosome 7. Genetics 135:1117-1123.
Russell, W. L. 1951. X-ray induced mutations in mice. Cold Spring Harbor Symp. Quant. Biol. 16:327-336.
Russell, W.L., E.M. Kelly, P.R. Hunsicker, J.W. Bangham, S.C. Maddux, and E.L. Phipps. 1979. Specific-locus test shows ethylnitrosourea to be the most potent mutagen in the mouse. Proc. Natl. Acad. Sci. U S A 76:5818-5819.
Russell, L.B., and C.S. Montgomery. 1982. Supermutagenicity of ethylnitrosourea in the mouse spot test: comparisons with methylnitrosourea and ethylnitrosourethane. Mutat. Res. 92:193-204.
Thomas, J.W., C. LaMantia, and T. Magnuson. 1998. X-ray-induced mutations in mouse embryonic stem cells. Proc. Natl. Acad. Sci. U S A 95:1114-1119.
You, Y., R. Bergstrom, M. Klemm, B. Lederman, H. Nelson, C. Ticknor, R. Jaenisch, and J. Schimenti. 1997. Chromosomal deletion complexes in mice by radiation of embryonic stem cells. Nat. Genet. 15:285-288.
QUESTIONS AND ANSWERS
DR. DELL: Do you perform open field testing during the day or night? I ask because the animals are active 90% of the time at night but only 10% of the time during the day. So your observation period could be shorter under infrared at night.
DR. JOHNSON: Under infrared, I am sure that is true. The logistics of setting up infrared in our colony would simply be prohibitive. I would remind
you that we are measuring all pedigrees against the other thousand pedigrees that have had the same test. We have already found animals that are two standard deviations above and two standard deviations below the population mean. Maybe if we were screening at night we would be finding a whole different set of mutants; however, it is not possible to do everything.
DR. FESTING: Please clarify your statement that you mutagenize the males. How many generations do you have to go through before you get the test animals?
DR. JOHNSON: Our pilot experiments generated data from more than 2000 animals using two-generation hemizygosity screens. In the second generation, the animal carries the mutagenized chromosome opposite a chromosome deletion. That experiment is very simple and effective. It greatly limits the size target in the genome because deletions can be only a certain size before there are haploinsufficiency problems. Some genome regions will not tolerate deletion at all.
We now are doing large chromosomal inversions, which require three generations. In the third generation before you have brought the mutagenized chromosomes to homozygosity, you can make as many animals as you want. Making more animals in that third generation so that you can breed it to make a fourth generation of multiple test class animals will limit the number of overall pedigrees that you can screen, which you do not want to do because only a certain percentage of the pedigrees are mutants in a region, or are mutant to a phenotype.
DR. JACOB: Please describe the average size of the deletion and whether there is a set amount.
DR. JOHNSON: The size of the deletion is very region dependent. We have two deletions that are up to about 6 cM each (such as the ones I described in the p region), and we have some in other regions that are probably 10 or 11 cM. We will not be able to make them much larger than that. For example, there is an imprinted region in the distal end of the p region that has the Prader-Willi/Angelman syndromes in humans, and it is not possible to make a deletion there in males.
DR. MORIWAKI: What kind of mutation is in that region?
DR. JOHNSON: At this time, we are simply mating two of the animals to learn whether they are fertile. If it is male sterile or female sterile, then we know people who are interested in taking that mutant for analysis. Again, this is primary screening. We are only saying that this animal is infertile; we do not know why. Those animals are advertised on our Web site and are available for anyone who is interested. Ours is a conventional facility. We been successful in mailing blastocysts overnight and having people transfer them into clean animals for rederivation of the stock.