INTRODUCTION

TOP ABSTRACT INTRODUCTION INBRED STRAIN CHARACTERIZATION:... MUTAGENESIS PHENOTYPIC SCREENS PHENOTYPIC DEVIANTS IDENTIFIED DISCUSSION REFERENCES

CARDIOVASCULAR DISEASE (CVD) is controlled both genetically and environmentally. There is good evidence that CVD is driven by aberrant transcription of not just one but many genes that collectively participate to describe the various disorders contributing to heart disease (6, 10, 19, 30). CVD is often coincident with hypertension, obesity, and diabetes, and these disorders share common risk factors (27). Often, certain of these risk factors, including hypertension, hypertriglyceridemia, reduced high-density lipoprotein (HDL)-cholesterol (HDL-C), visceral obesity, hyperglycemia, and insulin resistance, are clustered in some individuals (20). On the basis of widespread observation of these concurrent disorders among individuals, this phenomenon has come to be recognized by clinicians as metabolic syndrome (17, 20), metabolic cardiovascular syndrome (2), insulin-resistance syndrome (21), or syndrome X (31). As attempts are made to describe the molecular mechanisms that lead to these disorders, it appears that additional risk factors are still being uncovered that significantly impact the incidence of disease. Recently, bone mass was shown to be inversely correlated with severity of atherosclerosis in mice (15), a finding that supports previous clinical observations in women (1, 3). In addition, increasing evidence shows that long-term effects of untreated obstructive sleep apnea may also be detrimental to cardiovascular structure and function. Patients with sleep apnea experience repeated episodes of airway occlusion during sleep, resulting in conditions of hypoxemia, hypercapnia, and significant changes in intrathoracic pressure, which, in turn, can produce acute changes in cardiovascular performance, as well as sustained effects such as an overall increased blood pressure (29).

As dissection of the complex nature of CVD proceeds, some important rodent models have been developed that either overexpress or underexpress a particular key gene (transgenics and knockouts, respectively) known to contribute to the pathophysiology of this disease process. The laboratory mouse, representing hundreds of distinct strains, is a rich genetic resource made increasingly more valuable by available and emerging technologies used to manipulate the mouse genome and to distinguish gene expression and phenotype in a quantitative manner. Some of these models [LDLR knockout (26) and ApoE knockout (37)] are now commonly used as a background for additional genetic manipulation, allowing further identification of pertinent underlying pathways. From spontaneous genetic mutations such as ob/ob (leptin deficient) and db/db (leptin receptor deficient), we have learned the importance of leptin in obesity (13, 22, 39, 40), and these models have been used extensively to study other disorders in which obesity is present. Leptin levels are elevated in obese individuals, who are often predisposed to atherosclerosis. Recently, ob/ob and db/db mice were used to investigate the role of leptin in platelet aggregation after vascular injury. It was shown in this study that leptin contributes to arterial thrombosis in vivo and that this effect appears to be mediated by the platelet leptin receptor (7). These and other genetically engineered models will continue to be of utility, although their development depends on a priori identification of target genes.

One approach to finding genes underlying a complex disease is quantitative trait loci (QTL) analysis, which relies on measurable phenotypic differences between inbred mouse or rat strains (31a). With the use of this strategy, the segregation of one or more traits of interest can be followed by crossing inbred strains that differ significantly in the measured phenotype(s). Backcrossed or intercrossed progeny are then phenotyped for the trait(s) and genotyped for a series of genetic markers spanning the genome that are polymorphic between the parental strains. Results from QTL studies have added significantly to our understanding and appreciation of the complexity of cardiovascular and other diseases. Furthermore, QTL studies have allowed investigation of gene interactions that may not have been revealed by other approaches and have provided a basis for defining major components of disease through analysis of concordance of QTL across mammalian species (42, 43).

Even with the multitude of genetic manipulations and known mutations currently available in rodents for studying cardiovascular and related disorders, we are still at a relatively early stage with respect to our full understanding of these disease processes and how they interact. There is still a need for additional robust models to use as tools to promote discovery of the genetic mutations underlying disease. Without knowing all of the genes involved in the progression of CVD, how can we develop promising new research models? We can do this by gaining a better understanding of existing genetic variation and by random whole-genome intervention (e.g., mutagenesis) followed by phenotypic assessment. The goal of our laboratory is to develop new models through a phenotype-driven approach, discussed in this review, utilizing two important mouse resources. One source for these models lies in the natural variation among the many existing inbred mouse strains. Inbred strains are a powerful, enduring resource because they provide unique opportunities to repeatedly access a genomically fixed population, enabling reliable comparison of data across laboratories. Because their genotypes are fixed, inbred strains are invaluable for deciphering the influence of environment on phenotype. Although inbred strains exhibit an enormous range of phenotypic diversity and provide an extraordinary resource for functional genomics, a conspicuous lack of universally available phenotypic data for many commonly used strains leaves much of their potential for improving human health untapped. The Mouse Phenome Project, in which our program participates, is an international collaborative effort to promote the quantitative phenotypic characterization of a defined set of mouse strains under standardized conditions for a wide range of phenotypes of biomedical relevance. A second source for new models is in genome-wide mutagenesis, followed by phenotypic characterization. Because the rate of spontaneous mutation in rodents is relatively low, we cannot depend on those rare models to rapidly advance our investigation; therefore, much effort has been made to introduce new mutations experimentally with the use of chemical mutagenesis. These protocols include generation of both subtle and gross changes in DNA sequence from point mutations to inversions, insertions, and deletions and have been used in many organisms.

Through the Programs for Genomic Applications (PGA) mechanism recently developed by the National Heart, Lung, and Blood Institute, we have established The Center for New Mouse Models of Heart, Lung, Blood, and Sleep Disorders at The Jackson Laboratory (http://pga.jax.org). Drawing from both natural inbred strain variation, as well as from new chemically induced variation, our goal is to link genetic variation to biological function and dysfunction. Using a phenotype-driven approach, we are conducting high-throughput, noninvasive screens of more than 40 existing inbred mouse strains and of chemically mutagenized mice to establish new mouse models to further our understanding of the genetic origins of disease. This review will highlight the phenotyping screens we perform, with an emphasis on CVD, and will feature results from both the strain characterization and the mutagenesis screen.

	INBRED STRAIN CHARACTERIZATION: THE MOUSE PHENOME PROJECT

TOP ABSTRACT INTRODUCTION INBRED STRAIN CHARACTERIZATION:... MUTAGENESIS PHENOTYPIC SCREENS PHENOTYPIC DEVIANTS IDENTIFIED DISCUSSION REFERENCES

Inbred strains of mice have been used for decades to study normal mammalian biology and disease (4, 16). Representing the natural genetic variation observed in human populations, inbred mouse strains potentially contain a mélange of genomic variants with multiple compensatory mechanisms responsible for a particular phenotype. At the request of the research community, and with the guidance of an International Steering Committee, The Mouse Phenome Project has coordinated efforts to choose an optimal set of strains, collect phenotypic and genotypic data, and build a public database for collecting and disbursing data generated from these efforts (36). Data are currently available for selected physiological and behavioral parameters in the Mouse Phenome Database (MPD; http://www.jax.org/phenome). These data promote a comparative approach that is used routinely with panels of inbred strains for choosing optimal strains for QTL analysis and are springboards for detecting and mapping both key regulatory loci and modifier genes.

We are participating in this effort by characterizing inbred strains using our screens for heart, lung, blood, and sleep phenotypes. We followed Mouse Phenome Project recommendations (http://aretha.jax.org/pub-cgi/phenome/mpdcgi?rtn=docs/recommendations) by using 10 animals of each sex per strain to maximize the value of our survey for the research community. Specific guidelines are accessed from the MPD homepage and include a list of priority strains and recommendations for standardization. An example of the data we have collected is presented in Fig. 1. This scatter analysis of our serum chemistry data emphasizes strain- and sex-specific differences with respect to plasma lipid profiles in response to consumption of a high-fat diet (32). Using MPD analysis tools, we found that, in mice that consume standard laboratory chow, HDL-C and total cholesterol (TC) levels are highly correlated among the strains, as shown in Fig. 1, left. Figure 1, right, demonstrates a dramatically different response among strains consuming a high-fat diet, as seen in the disproportionate increase in TC levels compared with HDL-C levels. Although values for both measurements tend to increase after consumption of the high-fat diet, the correlation of TC with HDL-C is diminished, mostly due to non-HDL-C but also because of the differential response to diet both among sexes and between strains. Some strains, such as DBA/1J and DBA/2J, are affected minimally by the high-fat diet, whereas other strains show an increase in both parameters (e.g., NOD/LtJ) or in only one parameter (e.g., LP/J has a greater change in HDL-C and CAST/Ei has a greater change in TC). Likewise, there are striking sex-specific responses exhibited by some strains, such as BUB/BnJ, where females are more susceptible than males to a significant increase in TC on the high-fat diet. Other strains show no statistical difference in response between sexes (e.g., BALB/cByJ and SM/J). Changes in TC in response to the high-fat diet are further demonstrated by the MPD analysis presented in Fig. 2. Here, measurements of increase are stratified by strain and by sex. Raw and summary data used for these analyses have been contributed and are Web accessible in the MPD (35).

View larger version (46K): [in this window] [in a new window]   Fig. 1.   Scatter analyses of selected inbred strains showing a diminished correlation of total plasma cholesterol (x-axes) and high-density lipoprotein (HDL)-cholesterol levels (y-axes) in response to the atherogenic diet. Plasma was collected after a 4-h fast for animals that consumed standard laboratory chow at 8 wk of age (left) and again at 25 wk of age after animals consumed the high-fat diet for 17 wk (right). Top: females. Bottom: males. Pearson correlation coefficients are shown in the top left corner of each plot.

View larger version (39K): [in this window] [in a new window]   Fig. 2.   Evaluation of plasma total cholesterol among inbred strains in response to consumption of a high-fat diet. Total plasma cholesterol levels are shown before (black bars) and after (gray bars with SE) consumption of the high-fat diet for 17 wk. Left: females. Right: males. Strain names are shown once between panels. Strains are listed in order according to female values for 17-wk cholesterol levels. Males are listed in the same order to draw attention to sex differences in some strains. The diet-induced increase in cholesterol level was normalized for each strain and is shown in black bars above the value panels (fold-increase = 17 wk level  8 wk level).

These dramatic differences in both TC and HDL-C among a broad sampling of inbred mouse strains challenged with the high-fat diet suggest that genetic determinants are responsible for these phenotypes. This large-scale survey helped identify numerous strains with a variety of distinct lipid profiles for further study to determine the genetic bases of the effects of diet on the development of CVD. It is hoped that this and other data generated by our program will enable the choice of optimal strains for use in mapping chemically induced mutations, modeling disease processes, and analyzing complex traits by QTL mapping. The strength of this approach lies in the broad diversity of phenotypes that can be identified by surveying such a large set of strains.

	MUTAGENESIS

TOP ABSTRACT INTRODUCTION INBRED STRAIN CHARACTERIZATION:... MUTAGENESIS PHENOTYPIC SCREENS PHENOTYPIC DEVIANTS IDENTIFIED DISCUSSION REFERENCES

Chemical mutagenesis in mice offers a promising resource for increased frequency and variety of mutations for functional genetic studies. Subjecting mice to chemical mutagens creates genetic variants that are identical to the parental background strain with the exception of the induced mutations that may be responsible for a shift in phenotype compared with the parental strain. Beginning with a single mouse strain, this approach creates a set of mutants that, depending on the mutagen employed, differ minimally in genomic sequence from the parental strain but differ robustly and reproducibly in phenotype. We are using N-ethyl-N-nitrosourea (ENU), an alkylating agent that induces point mutations in the DNA of spermatogonial stem cells by virtue of single-base mismatching to the unrepaired alkylated base. Mutations are induced randomly throughout the genome, although ENU appears to preferentially target A:T pairs, resulting in both transversions to T:A and transitions to G:C (33). ENU mutagenesis will induce genetic variation at an estimated 1,000-fold higher rate than spontaneous mutations occur.

To capture recessive as well as dominant mutations, we are performing a third-generation screen of ENU-derived mice. This scheme is outlined in Fig. 3. Briefly, males of inbred mouse strain C57BL/6J are injected with three, once weekly doses of 85 mg ENU/kg body wt. These males, referred to as the G₀ generation, undergo a sterility period of ~10 wk. After these mice return to fertility, they are mated to nonmutagenized females of the same inbred strain to produce the G₁ generation. Each G₁ animal represents one unique mutagenized genome. Based on analysis by the specific locus test (14) at this dose of ENU, we expect that the G₁ mice will carry 150 mutations, on average. For each G₀ male, we generate 10 G₁ males and mate them each to a nonmutagenized female to produce the G₂ generation. Four female G₂ animals are then mated back to their G₁ fathers to produce the G₃ generation. Statistically, one in eight G₃ progeny will be recessive for any given locus mutated in the G₀ animal. Phenotyping is performed on the G₃ animals. It takes approximately 7 mo from the initial injection of the G₀ male to the birth of G₃ mice. Our goal is to survey ~200 ENU-mutagenized genomes per year or an average of 4,000 G₃ mice annually. To achieve this, we perform multiple rounds of injections into new G₀ males throughout a 12-mo period. Our program started 2 yr ago, and, in addition to performing the inbred strain characterization in this initial period, we have screened 250 mutagenized genomes.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   N-ethyl-N-nitrosourea (ENU)-mutagenesis scheme for generating recessive mutants. See text for more detailed description. Open box follows the segregation of a hypothetical ENU-derived mutation.

	PHENOTYPIC SCREENS

TOP ABSTRACT INTRODUCTION INBRED STRAIN CHARACTERIZATION:... MUTAGENESIS PHENOTYPIC SCREENS PHENOTYPIC DEVIANTS IDENTIFIED DISCUSSION REFERENCES

Success in capturing disruptions in genes that normally drive cardiovascular fitness is directly dependent on the breadth of the phenotyping protocol. Our goal is to perform all tests on every G₃ animal. Outliers, or phenotypic deviants, are identified on the basis of their presenting significant deviation in a measured phenotype from threshold values established in a multistep process that considers control and G₃ data and evaluation of biological relevance. Animals are first observed at birth for anemias or platelet defects that may be indicated by pallor, petechiae, or bruising. Observations are made again at wean, when some obesities can be noted by visually comparing littermates. Mice are then aged to 6 wk before they are entered into a comprehensive, 9-wk protocol designed to identify phenotypic deviants in a broad range of screens for heart, lung, blood, and sleep disorders. For the purposes of this review, we will focus on those phenotypes that will provide primary utility in studies related to CVD, but the full range of phenotypes can be viewed from our web site (http://pga.jax.org/protocols. html).

Because mice are, in general, naturally resistant to the development of some features of CVD, especially to fatty streak progression toward atherosclerosis, we feed them a synthetic, dairy fat-derived, high-fat and high-cholesterol diet (32). This enables observation of conditions that more closely resemble those seen in human disease. In our protocol, mice begin the high-fat diet at 9 wk of age and continue consuming it for many of the screens we perform, through 14 wk of age. Plasma TC, HDL-C, triglycerides, and glucose are measured after a 4-h fast both before and after a 6-wk consumption of this diet. As a follow-up to animals presenting deviations in glucose levels, either while consuming chow or after consuming the atherogenic diet, we perform an UltraSensitive ELISA test for insulin (ALPCO Diagnostics, Windham, NH).

Systolic blood pressure and pulse are measured by a tail-cuff method [Visitech Systems, Apex, NC (28)]. Mice are first trained to the apparatus for 3 days before measurements are made during the following 2 days. At the end of this test, mice have consumed the high-fat diet for 2 wk. Detection equipment includes a warmed platform on which four unanesthetized mice are held individually with a magnetic restrainer, which is open at the nose end. A computer drives the inflation and deflation of the tail cuff, as a sensor positioned at the base of the tail detects when blood flow starts and stops. Training and measurements are performed in the morning over a period of ~20 min per platform of four mice. At the end of the screen, we obtain an average systolic blood pressure based on 60 measurements.

Doppler, M-mode, and two-dimensional echocardiography are used to assess hemodynamic, diastolic, and systolic heart function in adult animals. We use a Philips Sonos 5500 ultrasound machine with a 15-MHz probe. Mice are lightly anesthetized with pentobarbital sodium to achieve a heart rate between 450 and 550 beats/min. This important noninvasive cardiac evaluation, including time for the anesthetic to take effect, takes ~30 min per mouse. An ECG is obtained from unanesthetized animals with an apparatus developed by Mouse Specifics (Boston, MA; Ref. 11); this is especially useful for providing additional evidence in support of a prior assessment of cardiac dysfunction from blood pressure or ultrasound screens. The ECG equipment consists of three pediatric conductance leads set on the platform of a tower on which the mouse is placed. The mouse must make contact with each of its hind feet and one of its front feet for the record to be made. Mice are allowed to acclimatize to the tower for 5-10 min, and readings are then obtained over an additional 3-5 min. Recordings are analyzed by software that includes Fourier analysis and interpretation of P, Q, R, S, and T waves for each beat. Data for heart rate, heart rate variability, QRS complex, and specific wave intervals are obtained.

To track the development of obesity, weight is followed in response to consumption of the atherogenic diet. In addition to weight monitoring, we use dual-energy X-ray absorptiometry (DXA) with anesthesia to measure lean tissue and body fat percentages, as well as bone mineral density and content. This analysis is done under tribromoethanol anesthesia by using a Lunar Piximus DXA machine and takes ~5 min to perform once the animal is properly anesthetized. Animals identified by the DXA scan as obese or lean are further screened with an ELISA test to measure nonfasted plasma leptin levels (CrystalChem, Chicago, IL).

Using a modified metabolic cage monitoring system developed at Columbus Instruments (Columbus, OH) in collaboration with Dr. Kevin Seburn at The Jackson Laboratory, we collect data on food and water intake, rest and activity cycles, and respiratory exchange ratios during multiple successive light and dark photoperiods. Mice are in these cages for a 3-day period, enabling screening of approximately one-third of the G₃ animals generated. Our primary interest in the data obtained from the metabolic cages is to develop a noninvasive method to identify disordered sleep patterns. When a mouse interrupts infrared beams positioned horizontally and vertically in these cages, this information is recorded, allowing rest and activity assessment. By analyzing these data, especially during 3-h epochs around photoperiod transition times, we have been able to determine the value of activity counts giving the best estimation of sleep and wakefulness. Our estimates of sleep-wake in C57BL/6J control mice are reasonable compared with published data (A. Pack, personal communication; Refs. 18, 25). We use this system as an initial estimate of disordered sleep and follow this observation with more invasive EEG and electromyograph monitoring. However, the metabolic data may also enable us to predict obesity before its manifestation as weight or body fat gain. As a preliminary test of the effectiveness of this system, the obese mouse model tubby (tub) was used. This mutant becomes overtly obese between 12 and 16 wk of age but was analyzed in the metabolic screen at 7-8 wk of age. The statistical algorithm developed to accompany this test showed aberrant data for tubby, most notably by a reduced respiratory exchange ratio, which was most pronounced during periods of high activity. Furthermore, the expected metabolic shift toward increased carbohydrate use did not occur (J. Naggert, personal communication).

	PHENOTYPIC DEVIANTS IDENTIFIED

TOP ABSTRACT INTRODUCTION INBRED STRAIN CHARACTERIZATION:... MUTAGENESIS PHENOTYPIC SCREENS PHENOTYPIC DEVIANTS IDENTIFIED DISCUSSION REFERENCES

Identification of phenotypic deviants is achieved by drawing on three sources of control data. First, nonmutagenized C57BL/6J animals are phenotyped seasonally, following the 9-wk protocol as if they were G₃ animals. Second, we look at the data generated by the G₃ animals themselves to understand population trends. Third, having determined thresholds for deviant values using the first two approaches, we consult experts in the appropriate research fields to confirm the biological relevance of our thresholds. The boundaries that we set must allow identification of abnormal phenotypes that are of interest in the biological context of the measurement and that define a truly deviant animal. Outlying values cannot be set globally for every trait by simply using a predetermined statistical limit such as 2.5 standard deviation units from the mean. Rather, data for each phenotype are interrogated separately by using this three-step approach. Once an ENU-derived G₃ animal is identified as exhibiting a phenotypic measurement that is significantly different from normal, the measurement is repeated. If the original observation is confirmed in the second test, the animal is mated appropriately to a C57BL/6J animal for heritability testing of the observed deviant trait. Intercrossing of resultant offspring will allow recovery of a recessive mutation, and we typically test 20 progeny from the intercrossed individuals. Affected progeny are then mated to each other to establish a colony. Although observation of more than one affected G₃ animal from a given G₁ founder is promising initial evidence that the trait is heritable, all confirmed affected G₃ animals are progeny tested. Depending on how many screens an animal has undergone in the protocol by the time it presents a deviant phenotype, we may have multiple phenotypes to draw on for a fuller assessment of the deviation. For instance, if a mouse presents diet-induced hypertriglyceridemia, observed at the end of the protocol, we are able to review measurements for blood pressure and body fat content for indications of the metabolic syndrome. Otherwise, additional phenotype information is gathered from the line established from the mutant.

Table 1 lists some representative mutants for hypertension, obesity, and cardiac function identified to date in our program and described below. Other phenotypic deviants identified thus far with potential use in cardiovascular research include those with altered lipid profiles (hypercholesterolemia, low HDL-C, high TC with low HDL-C), which have only recently been identified and therefore are currently undergoing the heritability testing process. Of 23 hypertensive phenotypic deviants thus far identified, we have evidence for heritability in 14; 4 of these are shown in Table 1. Retesting to confirm this phenotype is done by taking 5 additional days of measurements, at 30 measurements/day, resulting in an average systolic blood pressure based on 150 measurements. Of 32 obese animals identified, we have evidence for heritability in 7; 3 of these are shown in Table 1. Obesity is often initially described in terms of accelerated weight gain over time or absolute weight differences from controls. The advantage of the DXA screen is that we can determine whether the extra weight is due to an increased lean or fat tissue mass. Model HLB147 was observed at wean (4 wk of age) to be significantly larger than his littermates and by 6 wk of age presented a value for percent body fat, as assessed by DXA, that exceeded that seen in 12-wk-old obese phenotypic deviants. Features of this model are further presented in Fig. 4. Elevated plasma leptin in HLB147 rules out the possibility that this animal is a remutation of ob/ob. However, we are currently testing whether the defect is due to a remutation of db/db. This model exhibits elevated plasma TC and high HDL-C and elevated glucose and insulin and is hypotensive (data not shown).

View larger version (56K): [in this window] [in a new window]   Fig. 4.   Obesity model HLB147 ("Thor"). A: photograph of the subject (right) with an age-matched control (left). B and C: output from dual-energy X-ray absorptiometry (DXA) scans revealing body composition (%fat), bone mineral composition (BMC), and bone mineral density (BMD) data for control and affected subject, respectively. For DXA scanning, the head region has been removed from the analysis, as indicated by the shaded oval.

In addition, the DXA measurements allow us to identify animals presenting apparently normal weight but with increased body fat compared with control animals (HLB93, HLB46), which would have been missed by monitoring weight alone. We have, however, also noticed animals with a weight significantly exceeding that of normal animals at the same age, and the obesity is confirmed by an increased body fat content (HLB179). Growth anomalies can also be detected in a heavy animal with a normal body fat measurement (HLB166), although this phenotype is not of central interest to our program.

While surveying strain A/J for ultrasound parameters, as part of the inbred strain characterization, we noticed an anomalous left ventricular pattern from the M-mode display in one female, A/J 05. The pattern reflected no structural defects, and we determined that it was not an effect of the anesthetic because no other animals from this strain presented the anomaly. We suspected arrhythmia and were able to follow up the evaluation with the unanesthetized ECG screen. From the MPD (23), we found that the normal ECG pattern of strain A/J is not significantly different from that of strain C57BL6/J, shown for comparison in Fig. 5B. As shown in Fig. 5C, the A/J mutant animal has an arrhythmia. Hence, this mouse represents a new mutation on the A/J inbred strain.

View larger version (82K): [in this window] [in a new window]   Fig. 5.   Identification of an arrhythmic subject from a survey of inbred strains. A: initial observation by echocardiography in the M-mode display from the short-axis view of the left ventricle, taken using a 15-MHz probe. Note the irregularity in the pattern from one cardiac cycle to the next, where an occasional pause causes the left ventricle to remain open longer than in the previous cycle (noted with arrows). A normal pattern appears in Fig. 6A. B and C: ECG patterns of unanesthetized control and subject, respectively, confirming the arrhythmia initially identified in the subject by echocardiography (noted with arrows). In the ECG recordings, millivolts are on the y-axis and time in seconds is on the x-axis.

In the case of HLB133, presenting dilated cardiomyopathy, we saw the same phenotype in a littermate, supporting the possibility that the trait is heritable. Details of the ultrasound measurements and the M-mode display for this animal are presented in Fig. 6.

View larger version (63K): [in this window] [in a new window]   Fig. 6.   Dilated cardiomyopathy in model HLB133. M-mode views were obtained from a 15-MHz probe oriented to acquire the short-axis view of the left ventricle. A: typical left ventricular pattern seen in unaffected animals. Regular cardiac cycles are depicted in the M mode of a normal subject as a pattern of adjacent circles (arrow). Diastole is represented by the maximum diameter of each circle, and systole is represented by the closure between the circles. B: pattern from HLB133, who had a likewise affected littermate, suggesting heritability. Note the failure of the left ventricle to close properly in systole, abolishing the regularly repeated closed circle pattern seen in unaffected animals such as is shown in A. C: echocardiographic measurements obtained from 12 control C57BL/6J female animals and the subject. * Significant (>2 SD) deviation from control values. PWd/s, posterior wall thickness in diastole or systole (mm); IVSd/s, interventricular septum thickness, diastole or systole (mm); LVDd/s, left ventricular diameter, diastole or systole (mm); EnFs, endocardial fractional shortening; EF, ejection fraction; LVM/BW, left ventricular mass divided by body weight.

The hyperglycemic model, HLB62, presented with hyperglycemia before it began the atherogenic diet. In cases presenting with a severe phenotype, the subjects are removed from the protocol and not challenged with the high-fat diet but are entered into heritability testing as soon as possible in an effort to maintain the animal's health and allow it to produce offspring. Model HLB62 is a case for which we have little or no other information gathered on the subject that would allow further evaluation of the disorder and for which we must generate a heritable line to investigate a potentially more complex phenotype.

We are genetically mapping up to 50 heritable phenotypic deviants per year by generating F2 crosses with an inbred strain that shows a significant difference in the featured phenotype or has been previously used in a QTL study with C57BL/6J, providing an indication of where other influential loci are. It is hoped that the data we collect from the inbred strain characterization will enable us to make well-informed choices about the strains we use for mapping. In addition, we are performing microarray analyses on proven phenotypic deviants in collaboration with another PGA initiative hosted by The Institute for Genomic Research (http://pga.tigr.org).

	DISCUSSION

TOP ABSTRACT INTRODUCTION INBRED STRAIN CHARACTERIZATION:... MUTAGENESIS PHENOTYPIC SCREENS PHENOTYPIC DEVIANTS IDENTIFIED DISCUSSION REFERENCES

Physiological pathways may be genetically dissected with a variety of complementary approaches. We support a dual approach for identifying new functionally pertinent mammalian resources for the study of heart, lung, blood, and sleep disorders. Without a thorough inventory of the genes involved in disease, comprehensive phenotype-driven strategies are an important step toward understanding disease. A thorough analysis of existing inbred strains and subsequent QTL analyses with the use of strains differing in multiple phenotypes will help us identify genetic loci associated with disease. To complement the QTL approach, we are also conducting large-scale, high-throughput analysis of chemically induced mutant mice. Although the random mutagenesis approach will allow us to identify and confirm single genes responsible for a phenotype of interest, inbred strains will point the way to modifier genes and will assist in dissecting genomic components and environmental factors responsible for complex diseases. By using these approaches in parallel, our goal is to gain access to key entry points from which an understanding of pathways and proteins describing these complex traits can be accelerated. Without favoring one approach over the other, we have combined these methods to produce new models and thereby complement any disadvantages of using only one approach (5). Identification of the ENU mutation leading to these observed heritable phenotypes will be left to the end user of these models and admittedly requires significant additional effort, although the availability of the mouse sequence (34) has lessened this burden. By making the mutagenized models available to the scientific community, our program enables other laboratories to capitalize on this large-scale effort. New models are posted on our web site (http://pga.jax.org/models.html) and are available to academic investigators for the cost of shipping only, as part of the PGA initiative to provide the research community with new tools to further ongoing study into the etiology of complex disorders. Animals showing an overt phenotype not of central interest to our program, and therefore that we will not pursue by heritability testing, are also available from the Web site. We do not expect that all of the ENU mutations recovered in our program will be novel, but remutations of known genes can extend allelic series and thereby enhance functional genetic characterizations (9, 38).

With careful measurement of the physiological effects of functionally relevant genetic mutations, we can begin to link these mutations to biological processes. A variety of mutagenesis protocols with the mouse are being implemented worldwide (8, 12, 24, 41) and, combined with an increased effort to characterize extant inbred mouse strains, will collectively serve to interrogate the genome for functional elucidation of human disease.