Site search   
  Home    Publications

Browse Publications
List of Contents

Quantitative Neurogenetics & QTL Mapping

Genetics of Myopia

Control of Neuron Number and Stereology

Growth Cones and Dying Axons

Retina Development and Visual System Mutants

Grant Application

U.S. Patent


Need Help?
Help with Publications
Help with Nervenet
Contact Us

Note to the Reader The full text of a 1995 NSF grant submission that outlines our research objectives and methods. The focus is on mapping quantitative trait loci that affect CNS development.

Print Friendly
Neurogenic Genes and Control of Neuron Number

Robert W. Williams and Dan Goldowitz
Objective. Cell types in the brains of mice and humans are almost indistinguishable in structure and function, but their numbers vary over more than three orders of magnitude. Variation within and between species is at the root of understanding brain evolution. In large part this variation has origins in a set of undefined genes that control rates of cell proliferation in different pools of neural precursor cells. Some genes may have global effects and control the scale of the entire CNS. Other genes must have specific effects on specific populations of cells; for example shifting ratios of rods and cone photoreceptors in species adapting to nocturnal or diurnal niches, or altering the distribution of motor neurons in aquatic and terrestrial species. These genes are not only at the root of brain evolution, but they are very likely to be key genes that normally control development of the brain. It is essential that we learn more about these genes. The tools to do this are now in place.

Background. There has not yet been a concerted effort to isolate genes that control neuron number (neurogenic or proneural genes) in vertebrates. The main problems have been an insufficient density of genetic markers and the amount of work required to get reliable data on cell numbers from hundreds of animals. However, a number of important technical advances now make it practical to map neurogenic genes. These advances include (i) a quadrupling of the density of the genetic map of the mouse, (ii) the ease of genotyping large numbers of marker loci distributed across the entire mouse genome by the polymerase chain reaction, (iii) a far more powerful conceptual framework for mapping quantitative trait loci (QTLs) that underlie complex genetic traits, and (iv) significantly more reliable and economical ways to estimate numbers of neurons.

Methods. This research focuses on four discrete neuron types in the mouse CNS that are amenable to accurate and economical quantitative analysis: retinal ganglion cells (large projection neurons), photoreceptors, horizontal cells (smaller interneurons), and neurons in the dorsal lateral geniculate nucleus. Accurate stereological, electron microscopic, and immunocytochemical methods are used to count these cell populations. The total population of single cell types differ as much as twofold between inbred strains of mice. The chromosomal regions that harbor genes that control this variation are being identified using recombinant inbred strains and intercross-backcross progeny. To map neurogenic candidate genes to within ±5 cM, the DNA from panels of 150-300 progeny is typed by the polymerase chain reaction and analyzed using the maximum likelihood interval method. Developmental studies of inbred and recombinant inbred strains are being carried out in parallel to determine whether these genes control rates of neuron production or rates of survival.

Significance. Genes that modulate absolute numbers of neurons are key determinants of development, individual behavior, species differences in brain structure, and rates of brain evolution. Isolating these genes is a first and crucial step in discovering proteins that normally regulate the proliferation and death of neurons early in development. Understanding how allelic variants of these genes modulate absolute numbers of neurons will ultimately provide much insight into the molecular biology of neuron generation and death.


Aims of Grant

The long-term aim of this proposal is to characterize genetic mechanisms that underlie the remarkable variation in cell number in the mammalian brain. In this proposal we request support for an analysis of neuron number in the mouse visual system. Our immediate goals are to:

1. Measure the strength of genetic and environmental control on variation in neuron number. We will study four well-defined and interconnected cell populations in the mouse primary visual system (photoreceptors, horizontal cells, retinal ganglion cells, and neurons in the dorsal lateral geniculate nucleus). As part of this analysis we will test the hypothesis that a relatively small number of genes have large and specific effects on the size of discrete neuron populations.

2. Map these complex CNS traits. Quantitative trait loci (QTLs) will be mapped using both recombinant inbred strains and the maximum likelihood interval method of Lander and Botstein (1989) on intercross progeny. Mapping genes is a first and crucial step in gaining access to molecular mechanisms that control neuron number and brain size.

3. Determine whether the variation in neuron number is due to differences in cell production or cell death. Large numbers of neurons often die during normal development. It is important to determine what processes neurogenic or proneural genes affect. This will be done by studying the development of the same four cell populations in inbred and recombinant inbred strains of mice.


Significance of goals, and state of knowledge

Variation in brain weight and neuron number. The weight of the vertebrate brain varies enormously, from as little as 7 mg in free-living deep-sea teleosts (Fine et al., 1987) to 7 kg in elephants. This million-fold variation is associated with great differences in neuron numbers. In mammals, the size of homologous populations vary at least 5,000-fold and this variation contributes to the behavioral diversity of this order (Williams & Herrup 1988). There is also wide-ranging variation within single species. Numbers of cells in the trigeminal mesenchephalic nucleus of Xenopus ranges from 185 to 284 (Kollros & Thiesse 1985). Numbers of neurons in the dorsal lateral geniculate nucleus of rhesus monkeys range from 1.0 to 1.6 million (Williams & Rakic 1988a). Variation is less marked but equally pervasive in the nervous systems of invertebrates (Goodman 1979, Macagno 1980, Stewart et al. 1986). The variation in neuron populations among members of a species provides one of the most important foundations for evolutionary change in brain structure (Armstrong & Bergeron 1985, Finlay 1992).

Genetic factors control neuron number. A significant fraction of the variation among members of a single species is caused by environmental and developmental effects. A case in point is the variation in numbers of neurons among isogenic grasshoppers (Goodman 1976). Variation of this type is not heritable and does not contribute to evolutionary change. In contrast, a large, still uncharted component of variation has a genetic basis. Given the complexity of neuron production and death (McConnell 1991, Oppenheim 1991), products of numerous genes—perhaps hundreds—must have specific effects on specific neuronal populations. Products of other genes, for example, hormones and trophic factors, may have global or pleiotropic effects (Berg 1982, Garcia et al. 1992).

Mapping genes that control natural variation. One approach to the problem of the genetic control of neuron number exploits variation among inbred mice. During the process of inbreeding heterozygous loci are forced into homozygosity. Some strains become homozygous for one of the original alleles, other strains for other alleles. The often large phenotypic variation in quantitative traits among inbred strains is principally due to additive genetic factors (Falconer 1981). These inbred strains are therefore superb tools with which to isolate genes that underlie complex phenotypic traits. The work of Wimer and colleagues (1969, 1976) is a good example of how to apply this approach to the genetics of the mammalian CNS. They were able to demonstrate large strain differences in forebrain structure and were also able to estimate the genetic determination of these differences. However, their work was constrained by the problem of estimating cell number in the hippocampus and cortex of large numbers of mice. Perhaps for this reason they did not estimate numbers of effective gene loci controlling variation in CNS structure. Furthermore, until the development of large sets of recombinant inbred strains by Taylor in the 1970s and 1980s (Taylor 1976, 1989) there was also no effective way to map complex quantitative traits (Klein 1978). The situation is now much improved. Quantitative genetics has moved way beyond estimates of heritability, dominance, and epistasis. Complex quantitative traits can now be mapped using recombinant inbred strains (Plomin et al. 1991, Belknap et al. 1992ab) and the maximum likelihood interval method (Lander & Botstein 1989). The genetic dissection of complex traits—of which neuron number is an excellent example—has only begun in the last few years using these new and powerful methods (Lander & Schork 1994).

Number of effective gene loci. If large numbers of genes collectively control variation in many different parts of the nervous system, then allelic differences at any one locus will be associated with minimal differences in phenotype. This is referred to as the infinitesimal model of polygenic action (Lai et al. 1994). In contrast, if a small number of genes control the variation, then individual quantitative trait loci (QTLs) will often have large effects. Such neurogenic and proneural QTLs can in principle be mapped. Recent work on the control of bristle number in Drosophila provides reason for optimism. Mackay and Langley (1990) and Lai and colleagues (1994) have shown that normal allelic variants at the achaete-scute complex and at the scabrous locus each account for 5 to 10% of the natural variation in numbers of these sensory organs. We have begun a comparable analysis of variation in neuron numbers among inbred strains of mice. Our results demonstrate that small numbers of genes control a large fraction of the strain variation in ganglion cell number. Additive genetic factors account for about 40% of the variation, and as much as half of this variation appears to be controlled by one, or possibly two QTLs (see Preliminary Results).

An analysis of four related neuron populations. We propose to analyze four well-defined populations of neurons at different levels of the primary visual system: photoreceptors, horizontal cells, retinal ganglion cells, and neurons in the dorsal lateral geniculate nucleus. We recently finished a system-level comparison of cell populations in the retina and dorsal lateral geniculate nucleus of two species of cat that provides an example of the power of a system-level comparative analysis (Williams et al. 1993). By carrying equally detailed studies of inbred mice and test cross progeny derived from strains with high and low neuron number, we will be able to (i) measure the contributions that genetic factors make to variation in neuron number, (ii) map QTLs that underlie these complex polygenic traits, (iii) measure the phenotypic and genetic correlation between neuron populations, and (iv) determine whether neurogenic QTLs affect the proliferation or death of neurons. In this application we use the term neurogenic quite broadly to describe loci that have an effect on final neuron populations. Such genes can control either the production or survival of neurons. These genes could be expressed in neurons, glia, or even in cells in non-neural organs such as liver or placenta.

Trophic interactions, cell death, and control of neuron number. One important idea in developmental neuroscience is that sizes of neuronal populations are carefully matched by trophic interactions (Berg 1982, Williams & Herrup 1988, Purves 1988). A major role assigned to cell death is to bring interconnected populations into better quantitative alignment. While there is much evidence that favors this hypothesis, there are large gaps in what we know about this process. For example, accurate data on the normal range of variation in ratios of cells are to our knowledge not yet available for a single pair of interconnected neuron populations. This is one of the reasons that we have chosen to study interconnected populations belonging to a single system. Photoreceptors are connected intimately with horizontal cells, and retinal ganglion cells with neurons in the dorsal lateral geniculate nucleus. With this design we can measure correlation between numbers of neurons. We will also be able to test whether common genetic mechanisms control the production of these interconnected neurons, or alternatively whether correlations in numbers emerge only after a period of quantitative adjustment achieved via cell death. Recent work in both the cat (Williams et al. 1993) and several rodent species (Finlay & Pallas 1989) has shown that there can be large variations in ratios and absolute numbers of neurons even within the visual system of closely related species or strains. For example, we have found that strains C57BL/6J and A/J have equal numbers of ganglion cells, but C57BL/6J has twice the number of horizontal cells. Consequently, we can test the hypothesis that the four populations are controlled, at least in part, by unique QTLs.

Project history and long-term strategy. For the past ten years both of us (RW and DG) have been interested in the control of neuron number (Chalupa et al. 1984, Williams et al. 1986, Williams & Rakic 1988a, Hecktroth et al. 1988, Goldowitz 1989, Smeyne & Goldowitz 1990, Williams & Goldowitz 1992ab, Williams et al. 1993, Rice et al. 1995, Scheetz et al. 1995). Much of this work examined effects of experimental manipulations on cell number—increased availability of target tissue, decreased neuronal activity, heterochronic shifts in neuron production, and effects of specific mutations. Several years ago we began to collect data on retinal ganglion cell number from inbred and outcrossed strains of mice as a prelude to making chimeras between strains with large phenotypic differences (Williams & Goldowitz 1992a). Because of the rapid refinement of the genetic map of the mouse, combined with the ease with which polymorphic microsatellites can now be typed, such data can now be put to another and potentially more interesting use—namely, identifying, mapping, and ultimately characterizing neurogenic QTLs.

Mapping neurogenic genes is an important first step in characterizing the genetic basis of neuron proliferation, survival, and death. A daunting second step is to then identify and characterize candidate genes. In this application we do not propose any positional cloning. Our research strengths are in neuroscience, and it therefore makes sense for us to (i) measure strain differences, (ii) estimate numbers of effective gene loci, (iii) map of QTLs to ±5 cM, and (iv) study the development of strain variation. Our rationale is that if we succeed in generating 4-8 QTLs, this will provide a reasonable size pool for a candidate gene approach in subsequent years. It may then also be worthwhile to generate recombinant congenic or consomic strains to isolate effects of single QTLs on nearly isogenic backgrounds (Moen et al 1991, Groot et al 1992). We will also assess the quality of the physical map in the mouse in regions in which QTLs have been mapped and consider whether any of the QTLs are amenable to higher resolution mapping and positional cloning.

In the long-term we may also continue to define and map additional neurogenic QTLs that affect other cell populations in other parts of the nervous system. It may be possible to define classes or systems of neurons influenced by common QTLs. Other long-term directions include cloning and characterizing specific neurogenic genes. This type of analysis might lead to the production of mice in which candidate neurogenic genes have been manipulated by transgenic experiments or eliminated by homologous recombination. At the current pace of progress, it is hard to predict what the best research directions will be in four years. Large segments of the mouse genome will have been physically mapped within this period, and within ten years a great deal of the mouse and human genomes will have been sequenced. We can expect to have a high resolution genetic map of essentially all murine genes within a decade. Some of the studies we are now beginning were not practical even two years ago because of the paucity of PCR-typable marker loci in mouse.


Preliminary Results

The approach that we are using has not yet been exploited to map neurogenic genes in any vertebrate. The findings that are summarize in this section are intended to serve as a proof of concept. We have now studied the ganglion cell population in 40 different strains, including (i) standard inbred mice, (ii) wild inbred mice, (iii) outcrossed and randomly bred mice, (iv) F1, F2, and N2 hybrid mice, and (v) recombinant inbred mice. We have quantified more than 285 cases by quantitative electron microscopy; 155 in the last four months alone. This is the largest systematic study of a neuron population in mouse, and possibly, in any vertebrate. This information is immediately useful in estimating the magnitude of change that might be achieved by means of selection and in estimating the similarity between relatives. In the context of this application, this work also provides a firm basis for the search for, and mapping of, neurogenic QTLs. In the next several pages we present a synopsis of our work on the ganglion cell population as an demonstration of what we expect to accomplish for each of the other three cell populations.

Work in progress under other support. At the earliest stages (1991-1993) a part of this work was supported by a grant from the NIH to study the clonal structure of the mouse retina (RO1-EY08868, now unfunded). At that time we were trying to identify strains of mice with large differences in cell number to exploit in generating aggregation chimeras (Williams & Goldowitz 1992a, Rice et al 1995). This work is now supported by a small intramural grant from the University of Tennessee ($9,600 for one year).

Three papers on the genetic analysis of the ganglion cell population are in preparation or submission. The first deals with the magnitude of natural variation in neuron number and provides estimates of broad- and narrow-sense heritability. The second paper describes work in which we have estimated gene dominance, maternal effect, and the number of effective gene loci controlling ganglion cell number. The final paper will describe mapping data from the BXD recombinant inbred strains.

We are adding new information and cases daily. The most current dataset and manuscripts are accessible for the reviewers of this application on Internet using Mosaic, Netscape, or other World-Wide Web readers at the URL address

The main conclusions of this work can be summarized as follows:


  1. The population of ganglion cells in different strains varies from about 45,000 ±1,800 (SEM) in CAST/Ei to 70,000 ±3,000 in C3H/HeJ. This is a remarkably high level of strain variation. The average number across 15 standard inbred strains is 61,600 ±2,500, the average across 5 randomly bred and outcrossed strains is very close—60,800 ±4,000.


  2. Variation within standard inbred strains is much less than variation between strains (ANOVA F ratio is 11.21 p <.001). This indicates that there is a large additive genetic contribution to the between strain phenotypic differences (Hegmann & Possidente 1991, cf. Roderick et al 1973).
  3. As predicted by standard models of quantitative genetics (Falconer 1981, Crusio 1992), within strain variance in cell number is least in isogenic F1 heterozygotes (13,000), is slightly greater in homozygous inbred strains (18,000), and is greatest in genetically outcrossed and randomly bred strains (68,000). From these data we estimate that 70-75% of the total variance in ganglion cell number is heritable in the broad sense. Between 40-50% of the total variance is due to additive genetic factors. Dominance and epistasis account for the difference.
  4. Dominance and maternal effects controlling ganglion cell number have been examined in six different F1 crosses between high and low stains. In three cases, the F1s had averages very close to the mid-parental value, indicative of additive gene action. However, in three other cases, the F1s had populations that were significantly higher than mid-parental values. Such an effect can be due to heterosis, maternal effect, or overdominance (Wahlsten 1983, Roubertoux et al 1990). Analysis of reciprocal F1s between BALB/cJ and C57BL/6J suggest that maternal effect is a major factor in some crosses in which BALB/cJ is the mother.
  5. Variance in F2 progeny is much greater than in F1 progeny (CAST/Ei by BALB/c). Parental strain phenotypes are easily recovered in the F2. This indicates that the number of effective QTLs controlling ganglion cell number is low—probably under 3 (Wright 1978).
  6. Average ganglion cell numbers in standard inbred strains are distributed roughly bimodally (Figure 1). In contrast, averages of 6 of 11 different F1 hybrids and outcrossed mice fall in this range. The absence of this z score gap in heterozygous animals suggests that a single major-effect QTL has segregated in the homozygous inbred strains (Belknap et al 1992b, Festing 1992).
  7. Average ganglion cell numbers in 9 of 11 of the BXD recombinant inbred strains can be divided into either the low or the high parental phenotype (see Table 2, ANOVA F ratio is 4.59, p < .001). There is a significant level of between strain variation among the high strains (F=2.24, p < .05). In agreement with point 5 above, these data indicate the segregation of a major effect QTL. There is a significant amount of residual genetic variation controlled by one or more secondary QTLs.
  8. By using the BXD strain distribution data, we have excluded all but four chromosomal regions as containing the major effect QTL (see method in Figures 2 and 3). At present two candidate regions stand out—one on Chr 11 near Hoxb, the other on Chr 16 near Son and Sod1. Lod scores approach 3 in both regions.
  9. We have begun a matched developmental study of a high (BALB/c) and a low strain (C57BL/6J). The present data suggests that the difference between these strains is present at birth, before the onset of ganglion cell death.
  10. The population of horizontal cells varies from 7,900 ±250 in A/J to 17,000 ±500 in C57BL/6J. Standard errors of estimate are extremely low due to the high regularity of the horizontal cell mosaic. Both strains have average retinal areas between 14 and 15 mm2. The difference in horizontal cell number is entirely due to a marked difference in cell density.
  11. Strain averages of ganglion cell number and horizontal cell number are not correlated. For example, C57BL/6J has one of the highest horizontal cell populations (17,000), but one of the lowest ganglion cell populations (54,750). The ratio of ganglion cells to horizontal cells varies from 3.2 to 6.7 in six strains for which data are now available.
  12. The correlation between ganglion cell number and brain weight is weak (r=0.35 for n=14 standard inbred strains, r=0.28 for 250 individuals, or r=0.26 for 11 BXD strains). About 12% of the strain variation in neuron number is predictable from information on brain weight. Similarly, 8% of the strain variation in neuron number is predictable from information on body weight.

Plan of the Experiments

Each of the neuron populations will be studied in stages, moving from an analysis of strain variation, through mapping, and concluding with developmental studies.

Stage 1. The first stage involves measuring the variation within and between different strains of inbred and outcrossed mice. The immediate purpose is to determine the degree to which additive genetic factors modulate the size of the target cell population (ganglion cells, horizontal cells, photoreceptors, and neurons in the lateral geniculate nucleus). We have completed this analysis for one population of neurons—retinal ganglion cells. More than 280 individuals belonging to 40 strains of mice have been studied, including standard inbred, recombinant inbred, wild inbred, F1, F2, and N2 hybrids, and several different types of genetically heterogeneous strains. This is the largest quantitative dataset yet produced for any neuron population in any vertebrate.

Comments. Variation among members of the genus Mus. We have discussed variation as a tool that gives us access to neurogenic loci. But variation in neuron number, whether environmental or genetic in cause, is an interesting and important topic in its own right, particularly in exploring the relationship between neuronal populations and behavioral capacity. An obvious question is whether mice with different numbers of ganglion cells have differences in visual acuity. Does the twofold difference in horizontal cell number between A/J and C57BL/6J affect adaptation or the center-surround structure of receptive fields? We hope that our careful quantification of different neuronal populations will catalyze studies of strain differences in visual performance.

We will measure variation within and among 15 or more genetically well characterized and common inbred strains of mice (129/J, A/J, AKR/J, BALB/c, C3H/HeJ, C57BL/6J, C57BL/KsJ, CBA/CaJ, CE/J, DBA/2J, LP/J, NZB/BinJ, NZW/LacJ, PL/J, and SJL/J). All strains have been inbred by successive brother-to-sister matings for more than 80 generations and are consequently homozygous at essentially all loci. These strains were initially selected by us without regard to any known CNS or ocular characteristics, even the presence or absence of known retinal mutations, such as rd (LaVail & Mullen 1976). Provided that the sample of strains is sufficiently large, then differences among inbred strains provides an unbiased estimate of additive genetic control (Hegmann and Possidente, 1981).Of the 15 strains, 7 strains (A/J, AKR/J, BALB/c, C3H/HeJ, C57BL/6J, DBA/2J, LP/J) have been genotyped by Dietrich and colleagues (1994) at more than 5250 polymorphic microsatellite loci.

Heritability estimates. Complementary methods will be used to calculate heritability; that is, the strength of genetic determination of the phenotypic variation. This analysis will allow us to decide how successful a genetic dissection of phenotypic variation is likely to be. We will use Hegmann and Possidente's method (1981) to estimate the additive genetic component from standard inbred mice alone. This method assumes that the variance present among a set of fully inbred mice is approximately twice that in the population that gave rise to the set of inbred strains. We will also use methods described in Falconer (1981) and Crusio (1992) and compare genetically uniform and genetically heterogeneous populations of mice. The genetically uniform mice used for this calculation are either F1 hybrids or standard inbred strains. In estimating heritability of ganglion cell number we correct for scaling problems caused by the use of standard deviations. Heritability is computed using the coefficient of variation.

Evolution, fitness, variation, and why studying the mouse visual system is a good idea. It is often stated that phenotypic traits that are highly variable within a single population contribute less to fitness than those traits that are tightly regulated (Mayr 1970). The rationale is that selective pressures will tend to trim away the extremes thereby minimizing genetic load. Conversely, the retention of high phenotypic variability implies reduced selection of a characteristic. There are obvious exceptions, including sexual dimorphism and within-sex bimodalities associated with niche selection (GC Williams 1992). But in general, an analysis of variation in a natural population can provide insight into the degree of selection that acts on a quantitative trait. Traits that are important in fitness also tend to have high levels of directional dominance and low heritabilities in response to selection (Hahn & Haber 1978). In contrast, traits that are relatively free of selection display more additive genetic variation and comparatively high heritability. Another way of thinking about this is that genetic load and multiple alleles are not as well tolerated or maintained in those system most exposed to selection. The reason this may be worth mentioning in the context of this application is as a counterpoint to the criticism that the mouse is not noted for its high visual acuity and that this mammal is therefore an unwise choice. Mice are nocturnal and rely more heavily on other sensory modalities to get around. However, because high levels of variation and the retention of high levels of polymorphisms are useful, if not essential, for mapping QTLs, the low level of selection on visual system performance in the mouse actually is of considerable advantage in mapping the set of genes that influence neuron populations in this system. Genes that control numbers of neurons in the mouse visual system may display higher than average levels of polymorphism than a more visual competent mammalian species.

Stage 2. Our purpose is to estimate the number of effective QTLs that control variation between strains. This involves counting neuron populations in F1 and F2 progeny generated between high and low cell number strains.

Comments. The ability to map a QTL is critically dependent upon the amount of phenotypic variance that it controls. There is no a piori reason why quantitative traits must be polygenic, and even if traits are controlled by many genes, if one or two of the QTLs account for a disproportionately large share of the phenotypic variance, then these QTLs should be mappable.

Several methods are available to determine the probability that a given complex quantitative trait can be broken into sufficiently large mappable pieces. We have used three methods in our study of the ganglion cell population. First, we have studied the phenotypes of a large and randomly selected group of inbred strains of mice. Discontinuities in the normal probability plots of strain averages indicate single QTL effects (Fentress 1992, Belknap et al. 1992b). A second method involves crosses between high and low strains. A comparison between variance in F1 and F2 progeny can be used to estimate minimum numbers of effective gene loci (Wright 1978, Falconer, 1981). If the variance in the F2 is great relative to that in the F1, then this suggests that the variation is controlled by a few genes with large effects. Wright's equation for the number of effective gene loci is generally biased in a direction that gives higher, rather than lower estimates of the number of effective loci (Wright 1978). The third method involves phenotyping recombinant inbred strains. If the variance of the quantitative parameter is low or quantitative single gene effects are large, then it is in principle possible to detect phenotypes corresponding to a limited number of independent major QTLs.

Stage 3. The third stage is mapping neurogenic QTLs. Two complementary methods will be used. The first method depends on the use of recombinant inbred (RI) strains of mice (Bailey 1981). The second method is molecular linkage analysis. This method involves typing intercross progeny between high and low strains. In many cases we will exploit the large genotypic and phenotypic differences between CAST/Ei (a wild inbred strain) and standard inbred strains, such as BALB/c, DBA/2J, and C57BL/6J. We have begun this analysis for the ganglion cell population.

Comments. Recombinant inbred strains. A major use of recombinant inbred strains is in mapping quantitative trait loci (Klein 1978, Plomin et al. 1991, Nesbitt 1992, Neumann 1992, Belknap et al. 1992ab, 1993, Plomin & McClearn 1993). There are two significant advantages of RI strains for mapping: (i) one can obtain an estimate of average phenotype associated with a fixed recombinant genotype, and (ii) one does not have to genotype animals, because the RI strains have already been typed for many hundreds of loci.

To perform this type of analysis, mice from each of the RI strains is phenotyped for neuron number. We intend to use adults of both sexes. One of the largest set of RI lines was derived from a single cross between a C57BL/6 female and a DBA male, and is referred to as the BXD RI strain set. For our purposes this RI line is ideal because of the large genotypic and phenotypic differences between parental strains (Taylor 1972). Twenty six isogenic recombinant strains from this mating can now be purchased from Jackson Laboratory. A group of animals from each line of BXD mice is typed (i.e., ganglion cell axons are counted), and the average phenotype is compared to that of the parental strains (high, low, or intermediate). Over 1300 gene loci have now been typed and mapped in the BXD lines (Williams 1994). Consequently, the combination of phenotypes with the different RI lines can be read like a code. By matching the neuron number data with pre-existing data for mapped loci, it is usually possible to localize a gene to within 5-10 cM.

While the RI strains are commonly used to map Mendelian traits, they can also be used to map QTLs (Plomin 1994, Manly 1994). The correlation between a set of previously typed genes and the quantitative trait is computed. Each of the parental alleles of previously typed genes is assigned a value of 0 or 1. This is similar to the analysis performed on a standard test cross progeny in mapping a QTL, except that no additional genotyping is required. The calculations are performed by the program Map Manager QTL (Manly 1994), but can also be done directly in a spreadsheet program such as Excel. The result is a map of the genome in which chromosomal segments are assigned correlations, and lod scores are assigned by the probability of there being a linkage between the QTL and allele distribution patterns (Figures 2, 3).

The power of RI analysis depends upon the number of RI strains that are typed. For example we have typed 11 of 26 BXD RI lines and the lod scores for putative ganglion cell QTLs do not yet exceed 2.8. Given the large numbers of comparisons that underlie this analysis, a lod score of more than 3, and preferably of up to 4, is needed to confirm a linkage (Lander & Schork 1994). By completing the analysis of the entire BXD series and by adding additional cases to the BXD strains that we have already begun to type, we expect to be able to define at least one QTL within a short period of time.

Linkage using polymorphic microsatellite loci. A linkage analysis involves finding the set of alleles at polymorphic loci that tend to segregate with a particular phenotype (high or low cell number) in intercross or backcross progeny. To map the ganglion cell number QTLs, inbred BALB/c mice (a high strain) are mated to inbred CAST/Ei mice (a low strain) to produce F1 hybrids. These hybrids are intercrossed to produce F2 in which the CAST/Ei and BALB/c alleles assort and recombine. Cells are counted in the individual F2 progeny when these mice have reached maturity (30-40 days of age). We then look for an association or correlation between CAST/Ei alleles at microsatellite loci and a low cell number phenotype. Conversely, we look for an association between BALB/c alleles and a high cell number phenotype. The stregth of these correlations is assessed by computing lod scores.

Backcross versus intercross progeny. To map recessive Mendelian mutations in the mouse the standard operating procedure is to type backcross progeny. However, in mapping a QTL it is advantageous to generate F2 progeny, because both parents are hybrids and there will be approximately twice as many recombinants between parental alleles. This is an especially important factor when it is expensive or difficult to phenotype animals. A drawback of genotyping F2 progeny is greater complexity of PCR products generated at microsatellite loci. There are three combinations of alleles that need to be distinguished on single lanes of the gels. For example, in a CAST/Ei by BALB/c intercross, these combinations are C/C, B/C, and B/B. In practice, it is sometimes difficult to genotype particular combinations of products—what are called dominant molecular markers—and it is therefore necessary to introduce genotype categories such as "not C/C" and "not B/B". There are currently two QTL programs available—MAPMAKER (Lander et al 1987) and Map Manager QTL (Manly 1994)—both of which handle dominant marker genotype data. Another difference between backcross and intercross progeny concerns the complexity of potential epistatic interactions. In a backcross there are only three combinations of alleles at two unlinked loci (C/C-C/C, C/C-B/C, and C/C-B/B). However, in an F2 intercross there are nine allele combinations, and if one intends to test a particular two locus model then there are advantages in testing the less heterogeneous N2 progeny. Given the complex epistatic interactions between neurogenic loci in Drosophila (Campos-Ortega & Knust 1990) this is a factor that we have to consider seriously. But for the time-being we have decided that the increase in meioses and the associated increase in map resolution in an F2 panel outweigh other factors.

The QTL interval mapping method takes advantage of the large number of microsatellite DNA markers (6,000) that are being generated by Bill Dietrich, Eric Lander, and collegues (see Dietrich et al 1992). Detection of polymorphisms is comparatively easy, involving PCR, gel electrophoresis, and ethidium bromide staining. The Portable Dictionary of the Mouse Genome (Williams 1994, see the Dictionary server on the World-Wide Web at the URL address will be used to select appropriate MIT microsatellite loci to type.

cDNA strategies in relationship to mapping neurogenic QTLs. Substractive hybridization cDNA methods have recently been exploited to define sets of genes expressed in the nervous system preferentially during early stages of brain development. These genes include the 10 Nedd genes, Ndpp1, and Ndn (Kumar et al. 1992, Aizawa et al. 1992, Sazuka et al. 1992). The spatio-temporal expression of several of these genes has been characterized, but their functional signficance is unknown. A cDNA-based approach defines members of the large set of genes expressed during brain develoment. Assessing which of the hundreds or thousands of developmentally expressed neural genes is worth pursuing will be difficult. However, given sufficient map data on neurogenic QTLs known to modulate the size of neuronal populations, it will be possible to rationally select small subsets of candidate cDNAs for more detailed study.

Candidate genes for QTLs. Neurogenic genes in Drosophila, such as Notch, Delta, almondex, vnd, and daughterless, control the relative proportion of neuroectodermal precursors that migrate into the embyro to become neuroblasts (Campos-Ortega & Knust 1990). Members of the achaete-scute complex (T1a, T2-T5) control the relative abundance of different classes of neurons. Homologs of several of these Drosophila neurogenic genes have been cloned in mouse including two unmapped achaete-scute homologs, Mash1 and Mash2 (Franco del Amo et al 1993), and two Notch homologs,Notch1 and Notch2 (Guillemot & Joyner 1993). Both Notch1 and Mash1 are expressed in the ventricular zone early in murine develoment. Mash1 expression begins as early as E10.5 in the brain and E12.5 in the retina. Its expression is restricted to neural tissues, both in the peripheral and central nervous systems. In contrast, Notch1 expression is much more widespread and transcripts can be detected in all three germ layers (Lardelli & Lendahl 1993). The role of these murine homologs is still unknown. As we map QTLs we will review the set of nearby genes, most especially homologs of known neurogenic and cell death genes (Bcl2 and homologs of nematode ced genes; Ellis & Horvitz 1986), helix-loop-helix transcription factors, kinases such as Trkb, etc., that are known to map within ±10 cM.

Stage 4. The final stage involves an analysis of neuron number during development in selected RI and inbred strains. The underlying question is how do QTLs affect neuron number? Do they modulate proliferation or cell death?

Comment. In the case of the ganglion cell population we will count the number of axons in the optic nerves of high and low strains during development. Most counts will initially be done on neonates. This stage of development precedes any appreciable ganglion cell loss (MA Williams et al 1990). If differences between high and low strains are already detected at P0, then this provides evidence that the gene acts on the kinetics of ganglion cell production. However, if no differences are noted at birth, then this supports the hypothesis that cell death is responsible for the strain differences.

Retinal ganglion cells and geniculate neurons are both known to undergo massive cell loss in the mouse and other mammals (MA Williams et al. 1990, Williams & Rakic 1988a). Both populations will be studied at birth using previously described methods (Williams et al. 1986, Williams & Rakic 1988ab). Photoreceptors are not subject to significant cell death (Young 1984, Goldowitz & Williams, in progress) and therefore a developmental analysis is not planned. It is not yet known whether horizontal cells are subject to neuron loss during development (cf. Robinson 1988).


List of specific experiments

  1. Complete mapping of ganglion cell QTLs using BXD recombinant inbred strains. We have finished analysis for 9 strains and will finish typing a total of 20 strains this spring. We will use the correlation methods of Klein (1978) and Plomin et al. (1991) to identify chromosomal regions that harbor major effect QTLs. We will continue to use program Map Manager QTL (Manly 1994) to compute lod scores. This program can map both primary and secondary QTLs. We now are using this program to perform exclusion mapping (Figures 2 and 3). At this point we have excluded most regions of the genome as harboring a QTL and have narrowed our search to several candidate regions. By doubling the number of analyzed RI lines we should be able to assign QTLs to a precision of about ±5-10 cM.
  2. Generate interspecies intercross progeny between CAST/Ei (a low number strain dervied from Mus castaneus) and BALB/c (high number strain standard inbred). We will type a panel of 150-300 F2 progeny. Genotyping will initially involve a low resolution scan for candidate regions using a set of 50-100 MIT loci on pooled DNA samples from the highest and lowest phenotype quartiles (Taylor & Phillips 1994). We will then test the statistical strength of the candidate regions more rigorously by genotyping individual mice (Jacoby et al 1994). Given the differences between strains used for RI mapping and QTL interval mapping we may well identify some unique QTLs using each method.
  3. Begin developmental studies of ganglion cells. We will count ganglion cells (optic axons actually) in newborn mice. Two high (lines 9 and 19) and two low (lines 12 and 18) BXD strains will be studied. We will also analyze the inbred strains CAST/Ei and C57BL/6J (both low), and BALB/c and DBA/2J (both high). If strain differences are of comparable magnitude at birth, before ganglion cell loss (MA Williams et al 1990), then this would support the hypothesis that strain differences are attributable to neuron production or commitment. If the differences are only evident after cell loss then strain differences are attributable to differential survival.
  4. Estimate the strength of genetic control of horizontal cell numbers. We will examine 10-15 strains of inbred mice (5-8 cases per strain). Parental strains of RI lines (A/J, DBA/2J, C57BL/6J, BALB/cBy, etc.) and wild inbreds (CAST/Ei and SPRET/Ei) will be typed.
  5. Generate F1 and F2 hybrids between strains with high (C57BL/6J) and low (A/J) horizontal cell numbers and estimate the number of effective gene loci controlling the variation between A/J and C57BL/6J.
  6. Characterize strain variation in photoreceptor numbers in 10-15 strains of inbred mice (5-8 cases per strains). Wholemount methods will be used (Williams 1991). Parental strains of RI strains and wild inbreds will be typed. This will allow us to estimate the strength of genetic control on photoreceptor numbers. In this study we will avoid strains that carry mutations at any of the retinal degeneration loci.
  7. Determine genetic variation in neuron number in the dorsal lateral geniculate nucleus (LGN). We will use direct three-dimensional counting of 50 microns thick sections from inbred strains. The precise boundaries of the LGN is a critical determinant of count accuracy. Borders will be defined in each strain by anterograde transport of HRP (Williams & Chalupa 1982).
  8. If large variation is found in neuron number in the LGN between either C57BL/6 and DBA/2J or between CAST/Ei and BALB/c, then we will type the brains of cases previously used to estimate retinal ganglion cell numbers. This will include a quantitative analysis of the LGN of the BXD strains previously used to map the ganglion cell QTLs. We will also count the brains of the F2 test cross progeny. In both cases, we will begin by typing cases already known to have extreme ganglion cell phenotypes.
  9. Define candidate QTLs controlling variation in LGN neuron number from the low resolution whole genome screen performed above in Step 2. If new candidate QTL intervals are identified (lod scores of above 3.0), then we will perform higher resolution genotyping of microsatellite markers within regions that harbor the candidate QTLs.
  10. Use direct three-dimensional counting to study the development of strain variation in LGN neuron number at birth in high and low BXD strains and inbred strains. Again the issue is whether strain differences are evident before naturally occurring neuron elimination.


A major advance that makes this research possible is the development of rapid, accurate, and economical methods to count populations of neurons. While estimating numbers of neurons in a progeny panel of 200-500 animals is still a daunting task, we now know that this is practical over a 3-4 year period. In the first part of this Methods section we review four key counting methods: quantitative surveys of the optic nerve, wholemounts analysis of immunohistochemically defined cells, wholemount analysis of photoreceptors, and direct 3-dimensional counting of neurons in the LGN. We have used all of these methods extensively in the past several years.

Synopsis of EM method. To estimate ganglion cell number, 20 to 25 evenly spaced electron micrographs of the optic nerve are taken at X10,000 to X20,000. A low power micrograph (X200) of the entire ultrathin section is taken to determine the total nerve cross-sectional area. Axons are counted directly on each negative. The density of axons in this sample is multiplied by the cross-sectional area of the nerve. Three lines of evidence indicate that there is a 1-to-1 numerical relation between retinal ganglion cells and optic axons: (i) numbers of retrogradely labeled ganglion cells closely match those of optic axons (Rice et al 1995), (ii) reconstructions of single axons indicated that they rarely if ever branch in the optic nerve, even early in development (Williams & Rakic, 1985), and (iii) counts of axons at the proximal and distal ends of the optic nerve do not differ significantly (Lia et al. 1984, Robinson et al. 1987).

Fixation and processing of EM tissue. Mice are anesthetized with an intraperitoneal injection of Avertin and are perfused transcardially with 0.9% phosphate buffered saline and fixative using a small peristaltic pump. The electron microscopy fixative is injected for 2 minutes (5 ml of 1.25% glutaraldehyde and 1.0% paraformaldehyde in 0.125 M phosphate buffer). An additional 5 ml of second double-strength fixative is injected for another 2 minutes. The head is removed and placed in the stronger fixative overnight at 4°C. Optic nerves are subsequently osmicated, stained with uranyl acetate, dehydrated, and embedded in Spurr's resin. Thin sections of the nerve are placed on slot grids and are photographed at a primary magnification of x10,000 to x20,000 on a JEOL 2000 microscope. Brains from these cases will be sectioned on a sledge microtome to analyze cells in the LGN (see below).

Accuracy of EM method. Three main factors influence the accuracy of electron microscopic estimates. The first factor is the quality of the images used for counting. If the tissue is poorly preserved then counts will underestimate the population because small unmyelinated fibers are not resolved. We have studied the correlation between the quality of tissue and the z score of the count (z scores were calculated within strains) for a dataset consisting of 250 cases. Average, good, and excellent cases have z scores of -0.14, -0.03, and +0.05, respectively. The difference in the average z scores is trivial (less than a 750 neurons per estimate). Poorly fixed tissue typically yield low counts (average z = -0.5) because small unmyelinated fibers cannot be counted reliably. Such tissue is discarded.

The second factor is the accuracy with which the image magnification is calibrated. At each sampling session a waffle-type calibration grid is photographed at the same magnification used to sample the nerve. These calibration images are used to compute sample area to within 2% accuracy. The accuracy of the calibration grid has been verified by comparing different calibration grids and by determining axon number using a ratiometric method that requires no information on absolute sample areas.

The third factor is the adequacy of sampling. The greater the number of sites, the lower the sampling error and the more accurate the count. The accuracy of any single estimate has to be weighed against its cost. In practice we have found that an analysis of 20 to 25 negatives gives us a SEM of ±1.5 axons per micrograph. The result is that our estimate from single cases typically have a sampling errors of ±4,000 on a base of 60,000. Replicate counts (n=20 pairs) of the same nerve have a correlation of 0.95. This level of accuracy is sufficient for the analysis of inbred strains, RI strains, and test cross progeny.

Approximately 10-25% of all axons in midorbital parts of the adult mouse optic nerve are unmyelinated. To ensure that they are accurately tallied, the count is performed while wearing 2.5X surgical binocular magnifying glasses. All counts on the negatives are double-checked on the light box also using 2.5X surgical binocular magnifying glasses. Doing the analysis directly on negatives reduces the expense and time required to analyze an optic nerve by more than a factor of two.

Immunohistochemical analysis of wholemounts. Counting calbindin-positive horizontal cells has also proven to give remarkably reliable data in wholemount preparations. For example, analysis of approximately 12 sites in each of four A/J mice gave averages that ranged from 21.44 to 23.67 cells/sample. This very low level of within strain variation, coupled to the very large between strain variation (see Preliminary Results), encourages a QTL analysis using quantitative immunocytochemical procedures. We have not yet demonstrated that we have labeled all horizontal cells in mouse retina [there are 2-3 separate categories in some diurnal species (Kolb et al 1992)] and for this reason we need to be careful to state that we will be counting only calbindin-positive horizontal cells. We suspect that this includes the vast majority of horizontal cells in this species (cf. Hamano et al. 1990). Other antibodies which we will test within the next year, are known to selectively label horizontal cells in mouse, including R4 and R5 and vimentin (Dräger et al. 1984). Animals are perfused with 4% paraformaldehyde in phosphate buffer. Retinas are removed and immersed overnight in 30% sucrose prior to quick freezing at -80°C. Calbindin polyclonal antibodies (Sigma Co. C8666, calbindin-D-28 kDa) are used to label horizontal cells. To improve antibody penetration, the duration has been increased (3-4 days in primary antibody, 2 days in secondary antibody, anti-mouse IgG). In nocturnal species with rod dominated retinas all, or nearly all, horizontal cell bodies and some of their proximal dendrites express high levels of calbinidin (Hamano et al 1990). The calbindin-positive horizontal cells are located in a single plane positioned approximately 40 microns from the retinal surfaces. We have now analyzed 42 retinas from 25 cases belonging to five inbred strains, one F1, and one outcrossed wild species of mouse (M. caroli). Large and very consistent strain differences have been characterized. The phenotype of the five F1 hybrids generated from C57BL/6J (high) and A/J (low) parents was almost precisely at the parental midpoint.

Wholemount analysis of photoreceptors. We have carried out several quantitative studies of photoreceptors in wholemounts of mice and other species (Wikler et al. 1990, Williams 1991, Williams et al. 1993). A major advantage of the wholemount is that minimal tissue processing is involved. Shrinkage and cutting artifacts are usually not a problem. Counting photoreceptors is a straightforward process and it takes approximately 2-3 hours to sample a single mouse retina (10 to 20 evenly distributed sample sites). The mean density of photoreceptor inner segments is multiplied by the area of the retina to obtain a final estimate of the photoreceptor population. We have done this in several strains of mice and have already found in excess of 30% variation in this small sample. Tissue for the wholemount analysis will be perfused as for immunohistochemistry (4% paraformaldehyde). The retina will be prepared as a wholemount in Gelvatol between two coverslips. A microscope equipped with Nomarski optics and video image enhancement equipment is used for this analysis (Williams 1991). The small size of photoreceptor inner segments in mouse requires the use of monochromatic green light to improve resolution.

Sampling Protocol. We will sample sites using a 0.5 to 1.0 mm grid over the entire retina. This is an unbiased method. Typically about 15 to 25 sites will be sampled. The mouse does not have a pronounced central-to-peripheral gradient in cell density, and this number of sample sites is already known to give estimates that have a standard error of the mean of approximately ±5%. Retinas will be processed and counted in batches to ensure uniform application of counting criteria.

Direct 3-dimensional counting. Eight years ago I developed a method of counting neurons called direct 3-D counting that is insensitive that avoids the inaccuracies that surround the Abercrombie correction for split cells (Williams & Rakic 1988b). Both microscopes in my lab are equipped for this method. Provided neuronal populations have a relatively uniform packing density, it can take less than two hours of analysis to obtain estimates that have a sampling error of under 5%. Estimating the volume of a nuclear region, for example the dorsal lateral geniculate nucleus of the mouse takes well under 1 hour. Reliable quantitative estimates of neuron number in the mouse LGN can be obtained in about three hours. Brains of adult mice, previously fixed for EM analysis (see above), will be protected in 30% sucrose and cut frozen at 50 microns on a sledge microtome. The microtome is equipped with a specimen stage encoder to allow accurate determination of mean section thickness. Complete or alternating series of sections through the thalamus will be stained with cresyl violet. Some animals will receive bilateral intraocular injections of 1.0 \0xB5l of 50% HRP. The purpose of this protocol is to allow us to more accurately delimit the boundaries of the LGN. In animals that have had intraocular peroxidase injections 24 h prior to sacrifice, alternate series of sections through the thalamus will also be processed using the tetramethylbenzidine reaction and counterstained with neutral red. Volume of the LGN will be calculated using methods described in Williams and Rakic (1988ab).


QTL Linkage Analysis

Intercross progeny for QTL mapping. We plan an interspecies intercross between CAST/Ei and BALB/c. As part of an experiment now in progress (Williams, Cutler, and Goldowitz, in progress), we have already begun using the MIT primers to detect polymorphism among strains of mice. PCR conditions are uniform for all primer pairs (24 cycles at annealing temperature of 55 °C) and are described in the paper by Dietrich and colleagues (1992). The PCR product is typically run out on a small sequencing gel. Sequencing polyacylamide gels are usually used because the microsatellite products of the PCR are typically short (100 to 300 bp), and because the differences in repeat number between strains are often only a few base pairs. However, in our case we will usually generate interspecies intercrosses (e.g., CAST/Ei and BALB/c) in which the differences in the size of PCR products are greater than 10 bp. The PCR products of these intercross progeny can be run out on a 3% GTG (FMC Inc.) agarose minigel. We have successfully detected the published polymorphisms for each of three MIT microsatellite loci that we have tested on agarose gels using only ethidium bromide for labeling of bands.

Scanning the genome for QTLs with microsatellites. The selection of PCR primers for whole genome QTL scanning involves selecting a set of approximately 50-100 evenly spaced marker loci in which the product differences between CAST/Ei and BALB/c are greater than 10-20 base pairs. Both of these strains have already been typed at all the 5,000 microsatellite loci (Dietrich et al. 1992), so finding a subset of 100 markers that fulfill this condition is simple. We have done this analysis and have generated a list of several hundred candidate markers. The mouse map is about 1,500 cM in length. Consequently, the average distance between any locus and its neighboring microsatellite loci will be about 6-7 cM. If one intended to type each of 50 F2 progeny with each of 100 of these markers, then one would need to carry out 5,000 PCR reactions. There is a much more efficient alternative involving a pooled DNA approach (Taylor & Phillips 1994). DNA from the highest and lowest 20 animals is pooled for PCR reactions with each of about 50 PCR primers (30 cM average spacing, adequate for detecting linkage).This involves setting up only 100 reactions. Every lane will be heterozygous (two bands per lane) since 20 cases are pooled. However, at those microsatellite loci that are linked to QTLs associated with the high and low phenotypes there will be marked quantitative asymmetry in the density of the two bands in each lane. These asymmetric patterns should be reciprocal on adjacent lanes from high and low pools. One can subsequently test whether nearby MIT loci also exhibit the same asymmetries. After this linkage screening process, all progeny are typed for microsatellites within the candidate QTL regions. This allows the informative recombinants to be scored, thus refining the map position of a QTL.

Documentation. UV-excited gels that have been treated with ethidium bromide will be photographed using Polaroid instant film. We have just completed an extensive analysis of genomic DNA from a family of dogs (the achiasmatic mutation), and have experience in documenting and analyzing PCR products (Ezer et al. in submission).

Calculating linkage between loci. The analysis consists of comparing the distribution pattern of the phenotype (high or low neuron or receptor number) with the distribution pattern of parental alleles at the polymorphic microsatellite loci. The first level of analysis is simply to detect a linkage, the second level is to estimate QTL position more precisely by looking at phenotypes of recombinants. Actual calculations will be performed using the program Map Manager QTL from Ken Manly, Roswell Park. This program runs on Macintosh computers. MAPMAKER QTL is also available from the Whitehead Institute. This program runs on UNIX. We have a Sun workstation and we will compare results of the two programs.

Statistical Criteria. Lander and Shork (1994) provide a table of the critical lod scores to use with several different experimental designs. For analysis of RI lines they suggest a value of 3.9 and for backcross or intercrosses, a lod score of 3.3. These values correspond to a genome-wide significance level of .05 and apply to two-sided tests in which alleles from either parental strain may increase neuron number.

A cross between CAST/Ei and BALB/c only tests for genes that are polymorphic between these two strains. Obviously, these two strains may share the same alleles at several loci affecting cell number. However, given the large differences (these strains were derived from different species of mice, Mus castaneus and M. domesticus), this cross will be highly informative. It also has the advantage that the MIT microsatelitte markers will differ substantially in length and we will consequently be able to type most PCR products on agarose gels.


Developmental Analysis of Strain Differences

Developmental analysis of cell number. As in the adult, the size of the ganglion cell population will be estimated by counting axons in the optic nerve. The developmetnal analysis of the optic nerve of the mouse is not difficult and will involve the same methods described in detail in Williams et al (1986) and Williams et al (1992). The key factor is excellent fixation of tissue on the day of birth (P0). We will analyze optic nerves from the following strains as part of this study: parental stocks used in all test cross panels and recombinant inbred lines (CAST/Ei, BALB/c, DBA/2J, and C57BL/6). In addition, we will analyze several other high and low strains. One strain we will study is C57BL/Ks. This Kaliss strain is a high cell number strain closely related to C57BL/6. We have already begun this analysis.

A difference in ganglion cell numbers between strains before ganglion cell death in the mouse (Linden & Pinto 1985, Young 1985) indicates that proliferation or commitment differ between the strains. In this case, we will study progressively younger ages to determine just how early differences can be detected.

The same type of analysis will be performed for LGN neurons. Accurate estimates of the number of neurons in the LGN of neonatal mice will require use of anterograde HRP labeling of retinal afferents to define the medial border of the nucleus. HRP (50%) and DMSO (5%) in water will be injected using a micropipette loaded with 200 \0xB5l of solution. Animals will survive for a 24-hour-period and will then be perfused with mixed aldehydes. The HRP reaction product will be demonstrated using tetramethylbenzidine as a chromogen. I have done this type of experiment in prenatal cats (Williams & Chlaupa 1982). The actual quantitative analysis of neurons in the LGN and criteria to be used in distinguishing neurons from glial cells are discussed in a paper by Williams and Rakic (1988a) in which this cell population was estimated as early as embryonic day 60 in rhesus monkeys. Full text of this and several other papers are available on line via World-Wide Web at the URL address

Horizontal cells are among the first cells to be generated in the mouse retina (Young 1983, 1984). We do not yet know how early these cells can be recognized immunohistochemically, nor do we know whether this population is subject to extensive cell death during normal develoment in mouse. We intend to begin by studying the number and distribution of calbindin-immunoreactive cells in mouse retina at birth. We will also employ antibodies directed against vimentin (Dr\0x8Ager et al, 1984) in this developmental analysis.



Aizawa T, Maruyama K, Kondo H, Yoshikawa Y (1992) Expression of necdin, an embryonal carcinoma-derived nuclear protein, in developing mouse brain. Devel Brain Res 68:265-274.

Anderson L, Haley CS, Ellegren H, Knott SA, Johansson M, Andersson K, Andersson-Eklund L, Edfors-Lilja I, Fredholm M, Hansson I, Hakansson J, Lunstrom K (1994) Genetic mapping of quantitative trait loci for growth and fatness in pigs. Science 263: 1771-1774.

Armstrong E, Bergeron R (1985) Relative brain size and metabolism in birds. Brain Behav Evol 26:141-153.

Atchley WR, Riska B, Kohn LAP, Plummer AA, Rutledge JJA (1984) A quantitative genetic analysis of brain and body size associations, their origin and ontogeny: Data from mice. Evolution 38:1165-1179.

Bailey DW (1981) Recombinant inbred strains and bilineal congenic strains. In: The Mouse in Biomedical Research. Vol I. History, Genetics, and Wild Mice (Foster HL, Small DJ, and Fox JG, Academic Press, New York, pp 223-240.

Belknap, J.K. (1992) Empirical estimates of Bonferroni corrections for use in chromosomal mapping studies with the BXD recombinant inbred strains. Behav. Genet. 22:677-684.

Belknap JK, Philips TJ, O'Toole LA (1992a Quantitative trait loci associated with brain weight in the BXD/Ty recombinant inbred mouse strains. Brain Res Bull 29:337-344. Correlation between brain and body was only 0.28. Not altered by log transformation. Thus body and brain weight are independent systems. One brain size QTL maps close to D7Rp2 on Chr 7.

Belknap JK, Crabbe JC, Plomin R, McClearn GE, Sampson KE, O'Toole LA, Gora-Maslak G (1992b) Single-locus control of saccharin intake in BXD/Ty recombinant inbred (RI) mice: Some methodological implications for RI strain analysis. Behav Genet 22:81-100.

Belknap JK, Metten P, Helms ML, O'Toole LA, Angeli-Gade S, Crabbe JC, Phillips TJ (1993) Quantitive trait loci (QTL) applications to substances of abuse: Physical dependence studies with nitrous oxide and ethanol in BXD mice. Behavior Genetics 23:213-222.

Berg D (1982) Cell death in neuronal development: Regulation by trophic factors. In: Neuronal Development. Spitzer NC (ed). Plenum Press, New York. pp. 297-331.

Bonhomme F (1992) Genetic diversity and evolution in the genus Mus. In Techniques for the Genetic Analysis of Brain and Behavior. Goldowitz D, Wahlsten D, Wimer RE (eds). Elsevier, Amsterdam, pp. 41-56.

Bonhomme F, Gu\0x8Enet JL, Dod B, Moriwaki K, Bulfield G (1987) The polyphyletic origin of laboratory inbred mice and their rate of evolution. Biol J Soc 30:51-58.

Boss BD, Turlejski K, Stanfield BB, Cowan WM (1987) On the numbers of neurons in fields CA1 and CA3 of the hippocampus of Sprague-Dawley and Wistar rats. Brain Res 406:280-287.

Broadhurt PL and Jinks JL (1974) What genetic architecture can tell us about the natural section for behavioral traits. In The Genetics of Behavior, van Abeelen JHF (ed), North-Holland, Amsterdam.

Brown, Pearce, and Van Allen (1926) cited in Yablokov.

Bulman-Fleming B, Wahlsten D (1988) Effects of a hybrid maternal environment on brain growth and corpus callosum defects of inbred BALB/c mice: A study using ovarian grafting. Exp. Neurol. 99:636-646.

Bulman-Fleming B, Wahlsten D, Lassalle JM. (1991) Hybrid vigour and maternal environment in mice. I. Body and brain growth. Behav Process 23:21-33.

Campos-Ortega JA, Knust E (1990) Molecular analysis of a cellular decision during embyronic develoment of Drosohila melanogaster: epidermogenesis or neurgenesis. Eur J Biochem 190:1-10.

Chalupa LM, Williams RW, Henderson Z (1984) Binocular interactions in the fetal cat regulates the size of the ganglion cell population. Neuroscience 12: 1139-1146.

Cowley DE, Pomp D, Atchley WR, Eisen EJ, Hawkins-Brown D. (1989) The impact of maternal uterine genotype on postnatal growth and adult body size in mice. Genetics 122:193-203.

Crusio WE (1992) Quantitative genetics. In. Techniques for the Genetic Analysis of Brain and Behavior. Goldowitz D, Wahlsten D, Wimer, RE (eds) Elsevier, Amsterdam, pp. 231-250.

Crusio WE, Schwegler H, van Abeelen JHF (1989) Behavioral responses to novelty and structural variation of hippocampus in mice. II. Multivariate genetic analysis. Behav Brain Res 32:81-88.

Crusio,WE, Schwegler H, van Abeelen JHF (1989) Behavioral responses to novelty and structural variation of hippocampus in mice. I. Quantitative-genetic analysis of behavior in the open field. Behav Brain Res 32:75-80.

Dietrich W, Katz H, Lincoln SE, Shin HS, Friedman J, Dracopoli NC, Lander ES (1992) A genetic map of the mouse suitable for typing intraspecific crosses. Genetics 131:423-447).

Drager, U.C. (1985) Birth dates of retinal ganglion cells giving rise to the crossed and uncrossed optic projections in the mouse. Proc. R. Soc. Lond. B 224:57-77.

Drager, U.C., and J.F. Olson (1980) Origins of crossed and uncrossed retinal projections in pigmented and albino mice. J. Comp. Neurol. 191: 383-412.

Drager, U.C., and J.F. Olson (1981) Ganglion cell distribution in the retina of the mouse. Invest. Ophthalmol. Vis. Sci. 20:285-293.

Drager UC, Edwards DL, Barnstable CJ (1984) Antobodies against filamentous components in discrete cell types of the mouse retina. J Neurosci 4:2025-2042.

Eisenberg JF (1981) The Mammalian Radiations. An Analysis of Trends in Evolution, Adaptation, and Behavior. Chicago Univ Press.

Elliott RW, Manly KF, and Jacoby RF (1994) Male and Female Map. Mouse Genome Conference 11: 45

Ellis HM, Horvitz H (1986) Genetic control of programmed cell death in the nematode C. elegans. Cell 44:817-829.

Falconer DS (1963) Quantitative inheritance. In Methodology in Mammalian Genetics. Burdette WJ, ed. Holden-Day, Inc. San Fransisco. pp193-216.

Falconer, D.S. (1981) Introduction to Quantitative Genetics, 2nd ed. New York: Ronald Press.

Festing MFW (1993) Origins and characteristics of inbred strains of mice. Mouse Genome 91:393-509.

Festing MFW (1992) From character to gene: some strategies for identifying single genes controlling behavioral characters. In: Genetic Analysis of Brain and Behavior: Focus on the Mouse. D. Goldowitz, R. Wimer and D. Wahlsten, eds. Amsterdam, Elsevier, pp. 17-38.

Fine ML, Horn MH, Cox B (1987) Acanthonus armatus, a deep-sea teleost fish with a minute brain and large ears. Proc R Soc Lond B 230:257-265.

Finlay, B.L. (1992) Cell death and the creation of regional differences in neuronal numbers. J. Neurobiol. 23:1159-1171.

Finlay, B.L. and S. Pallas (1989) Control of cell number in the developing mammalian visual system. Prog. Neruobiol. 32: 207-234.

Franco del Amo F, Gendron-Maguire M, Swiatck PJ, Gridley T (1993) Cloning, sequencing and expression of the mouse mammalian achaete-scute homolog 1 (MASH1). Biochim Biophys Acta 1171:323-327.

Fuller JL, Geils HD (1973) Behavioral development in mice selected for differences in brain weight. Dev Psychobiol 6:469-474.Garcia I, Martinou I, Tsujimoto, Martinou JC (1992) Prevention of programmed cell death of sympathetic neurons by the bcl-2 proto-oncogene. Science 258:302-304.

Johns, P.R., and R.D. Fernald (1981) Genesis of rods in teleost fish retina. Nature 293:141-142. Points out that intercallated production of rods throughout the retina means that some progenitor scells are left behind. Other cell types are diluted in central retinaGoldowitz, D. (1989a) The weaver granuloprival phenotype is due to intrinsic action of the mutant locus in the granule cell: Evidence from homozygous weaver chimeras. Neuron 2:1565-1575.

Goldowitz D (1989) Cell allocation in mammalian CNS formation: Evidence from murine interspecies aggregation chimeras. Neuron 3:705-713.

Goldowitz, D., Agoulnik I., Rice, D.S., and Bishop, C. (1993) A molecular analysis of the region of mouse chromosome 16 that contains the weaver locus. Soc. Neurosci. Abst. 19:76.9.

Goldowitz, D., Seiger, A., Olson, L. (1984) Regulation of axon ingrowth to area dentata as studies by sequential, doublt intraocular brain tissue transplantation. J. Comp. Neurol. 227:50-62.

Goodman CS (1974) Anatomy of locus ocellar interneurons: constancy and variability. J Comp Physiol 95:185-201.

Goodman CS (1976) Constancy and uniqueness in a large population of small interneursons. Science 193:502-504.

Goodman CS (1978) Isogenic grasshoppers: Genetic variability in the morphology of identified neurons. J. Comp. Neurol. 182:681-705.

Goodman CS (1979) Isogenic grasshoppers: Genetic variability and develoment of identified neurons. In: Neurogenetics: Genetic Approaches to the Nervous System. Breakfield XO (ed) Elsevier, New York.

Goodman CS and Williams JLD (1976) Anatomy of the ocellar interneurons of acridid grasshoppers. II. The samll interneurons. Cell Tiss Res 175:203-225.

Gora-Maslak G, McClearn GE, Crabbe JC, Phillips TJ, Belknap JK, Plomin R (1991) Use of recombinant inbred strains to identify quantitative trait loci in psychopharmacology. Psychopharmacology 104:413-424.

Griffith AJF, Miller JH, Suzuki DT, Lewontin RC, and Gelbart, W.M. (1993) An Introduction to Genetic Analysis, 5th edition. Freeman, New York.

Groot PC, Moen, CJA, Dietrich W, JP Stoye, Lander ES, Demant P (1992) The recombinant congenic strains for analysis of multigenic traits: genetic composition. FASEB J 6:2826-2835.

Guillemot F, Joyner AL (1993) Dynamic expression of the murine achaete-scute homologue Mash-1 in the developing nervous system. Mechan Devel 42:171-185.

Hahn ME, Haber SB, and Fuller JL (1973) Differential agonistic behavior in mice selected for brain weight. Physiol Behav 7:21-30.

Hahn ME, Haber SB (1978) A diallel analysis of brain and body weight in male inbred laboratory mice (Mus musculus). Behav Genet 8:251-260.

Hamre, K.M., and Goldowitz, D. (1992) Basis of granule cell loss in the murine cerebellar muation, Meander Tail. Soc. Neurosci. Abst. 18:69.3

Hamano K, Kiyama H, Emson PC, Manabe R, Nakauchi M, Tohyama M (1990) Localization of two calcium binding proteins, calbindin (28 kD) and parvalbumin (12 kD), in the vertebrate retina. J Comp Neurol 302:417-424.

Hecktroth JA, Goldowitz D, Eisenman LM (1988) Purkinje cell reduction in the reeler mutant mouse: A quantitative immunohistochemical study. J Comp Neurol 279:546-555.

Hegmann JP, Possidente B (1981) Estimating genetic correlations from inbred strains. Behav Genet 11:103-114

Hyde JS (1973) Genetic homeostasis and behavior: analysis, data, and theory. Behav Genet 3:233-245.

Jacoby RF, Hohman C, Marshall DJ, Frick TJ, Schlack S, Broda M, Smutko J, Elliott RW (1994) Genetic analysis of colon cancer susceptibility in mice. Genomics 22:381-387.

John D.P, and A.A. Kiessling (1988) Improved pronuclear mouse embryo development over an extended pH range in Ham's F-10 medium without protein. Ferohnson TE, DeFries JC, Markel PD (1992) Mapping quantitative trait loci for behavioral traits in the mouse. Behav Genet. 22:635-653.

Kelley DB, Dennison J (1990) The vocal motor neurons of Xenopus laevis: development of sex differences in axon number. J Neurobiol 21:869-882.

Klein R (1990) Cell 61: 647 (This is a TrkB neurogenic gene paper.)

Klein TW (1978) Analysis of major gene effects using recombinant inbred strains and related congenic lines. Behav Genet 8:261-268.

Kolb H, Linberg KA, Fisher SK (1992) Neurons of the human retina: A Golgi study. J Comp Neurol 318:147-187.

Kollros JJ, Thiesse ML (1985) Growth and death of cells of the nesencephalic fifth nucleus in Xenopus laevis larvae. J Comp Neurol 233:481-489.

Kumar S, Tomooka Y, Noda M (1992) Identification of a set of genes with developmentally down-regulated expression in the mouse brain. Biochem Biophys Res Comm 198:1155-1161.

Lai C, Lyman RF, Long AD, Langley CH, Mackay TFC (1994) Naturally occuring variation in bristle number and DNA polymorphisms at the scabrous locus of Drosophila melanogaster. Science 266:1697-1702.

Lander ES, Botstein D. (1989) Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121:185-199.

Lander ES, Schork NJ (1994) Genetic dissection of complex traits. Science 265:2037-2048.

Lander ES, Green P, Abrahamson J, Barlow A, Daly M, Lincoln S, Newburg L. (1987) MAPMAKER: An interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1:174-181.

Lardelli M, Lendahl U (1993) MotchA and MotchB—two mouse Notch homologues coexpressed in a wide variety of tissues. Exp Cell Res 204:364-372.

Kahn, A.J. (1974) An autoradiographic analysis of the time of appearance of neurons in the developing chick neural retina. Dev. Biol. 38:30-40.

LaVail MM, Mullen RJ (1976) Role of pigment epithelium in inherited retinal degeneration analyzed with experimental mouse chimeras. Exp Eye Res 23:227-245.

LaVail, M.M., J.C. Blanks, and R.J. Mullen (1982) Retinal degeneration in the Lia, B., Williams, R.W., and Chalupa, L.M. (1986) Does axonal branching contribute to the overproduction of optic nerve fibers during early development of the cat\0xD5s visual system. Developmental Brain Research 25:296-301.

Lewis, J. (1973) The theory of clonal mixing during growth. J. Theor. Biol. 39:47-54.

Linden R, Pinto LH (1985) Developmental genetics of the retina: Evidence that the pearl mutation in the mouse affects the time course of natural cell death in the ganglion cell layer. Exp. Brain Res. 60:79-86.

Macagno ER (1980) Number and distribution of neurons in leech segmental ganglia. J. Comp. Neurol. 190:283-302.

Macagno ER, Lopresti,V, Levinthal C (1973) Strucuture and development of neuronal connections in isogenic organisms: Variations and similarities in the optic system of Daphnia magna. Proc Natl Acad Sci USA 70:57-61.

Mackay TFC, Langley CH (1990) Molecular and phenotypic variation in the achaete-scute region of Drosophila melanogaster. Nature 348:64-66.

Manly, KF (1994) New functions, including quantitative trait analysis in MAP MANAGER. Abstract of the Mouse Genome Conference 11:95

Mather K, Jinks JL (1982) Biometrical Genetics. Chapman and Hall, London.

Maurin Y, Berger B, Le Saux F, Gay M, Baumann N (1985) Increased number of locus coeruleus noradrenergic neurons in the convulsive mutant mouse quaking. Neurosci Lett 57:313-318.

Mayr E (1970) Populations, Species, and Evolution. Belknap Press, Cambridge.

McConnell SK (1991) The generation of neuronal diversity in the central nervous system. Ann Rev Neurosci 14:269-300.

McGuire, B.A., Stevens, J.K., and Sterling, P. (1984) Microcircuitry of biopolar cells in cat retina. J. Neurosci. 4:2920-2938.

Miller, M.M. and M. Oberdorfer (1981) Neuronal and neuroglial responses following retinal lesions in the neonatal rats. J. Comp. Neurol. 202:493-504.

McLoon, S.C., and R.B. Barnes (1989) Early differentiation of retinal ganglion cells: An axonal protein expressed by premigratory and migrating retinal ganglion cells. J. Neurosci. 9:1424-1432.

Moen CJA, van der Valk MA, Snoek M, van Zutphen BFM, von Deimling O, Hart, AAM, Demant P (1991) The recombinant congenic strains—a novel genetic tool applied to the study of colon tumor development in the mouse. Mamm Genome 1:217-227.

Moore, W.J., and B. Mintz (1972) Clonal model of vertebral column and skull development derived from genetically mosaic skeletons in allophenic mice. Devel. Biol. 27:55-70.

Mouse New Letter (1984) Mutations that Affect Eye Development. Vol. 70:

Nesbitt MN (1992) The value of recominant inbred strains in genetic analysis of behavior. In Techniques for the Genetic Analysis of Brain and Behavior: Focus on the Mouse. Goldowitz D, Wahlsten D, Wimer RE (eds.) Elsevier, Amsterdam, pp 141-146.

Neumann PE (1992) Inference in linkage analysis of multifactorial traits using recombinant inbred strains of mice. Behav Genet 22:665-676.

Murakami, D., M. Sesma, and M.H. Rowe (1982) Characteristics of nasal and temporal retina in Siamese and normally pigmented cats: Ganglion cell composition, axon trajectory and laterality of projection. Brain Behav. Evol. 21: 67-113.

Oppenheim RW (1991) Cell death during development of the nervous system. Ann Rev Neurosci 14:453-501.

Padeh B, Soller M (1976) Genetic and environmental correlations between brain weight and maze learing in inbre strains of mice and their F1 hybrids. Behavior Genet 6:31-41.

Plomin R (1994) BXD recombinant inbred strains as a first step towards identifying QTL. Mouse Genome Conference 11:116

Plomin R, McClearn GE (1993) Quantitative trait loci (QTL) analyses and alcohol-related behaviors. Behavior Genet. 23:197-11.

Plomin R, McClearn GE, Gora-Maslak G (1991) Use of recombinant inbred strains to detect quantitative trait loci associated with behavior. Behav Genet 21:99-116.

Purves D (1988) Body and Brain. A Tropic Theory of Neural Connections. Harvard Univ Press, Cambridge.

Reeves RH, Crowley MR, Lorenzon N, Pavan WJ, Smeyne RJ, Goldowitz D (1989) The mouse neurological mutant weaver maps within the region of chromosome 16 which is homologous to human chromosome 21. Genomics 5: 522-526.

Reis, DJ, Baker H, Find JS, Joh TH (1981) A genetic control of the number of dopamine neurons in mouse brain: Its relationship to brain morphology, chemistry and behavior. In Genetic Research Strategies in Psychobiology and Psychiatry, ed. E.S. Gershon, S. Matthysse, X.O Breakfield, R.D. Ciarnello, pp. 215-229. Pacific Grove Ca, Boxwood Press.

Rice DS, Williams RW, Goldowitz D (1995) Genetic control of retinal projections in inbred strains of albino mice. J Comp Neurol, in press

Remtulla, S., and P.E. Hallett (1985) A schematic eye for the mouse, and comarison with the rat. Vision Res. 25:21-31. The eye of the mouse is half as big as that of the rat linearlly and 1/8 in terms of volume. Refractive index of lens etc is the same. The mouse retina is about as thick as that of the rat circa 240 microns. This value does not vary substantially with eccentiricty. Radius of curvature of an adult C57 is about -1.6 mm

Robinson SR, Horsburgh GM, Dreher B and McCall MJ (1987) Changes in the numbers of retinal ganglion cells and optic axons inthe developing albino rabbit. Brain Res 432:161-174.

Robinson SR (1988) Cell death in the inner and outer nuclear layers of the developing cat retina. J. Comp. Neurol. 267: 507-515.

Roderick TH, Wimer RE, Wimer CC, Schwartzkroin PA (1973) Genetic and phenotypic variation in weight of brain and spinal cord between inbred strains of mice. Brain Res 64:345-353.

Romer AS (1969) Vertebrate history with special reference to factors related to cerebellar evolution. In: Neurobiology of Cerebellar Evolution and Development. (Llinas R, ed). Chicago, American Medical Association, pp 1-18.

Roubertoux PL, Nosten-Bertrand M, Carlier M (1990) Additive and interactive effects between genotype and maternal environment, concepts and facts. Adv Study Behav 19:205-247.

Sacher GA, Staffeldt EF (1974) Relation of gestation time to brain weight for placental mammals: Implications for the theory of vertebrate growth. Amer Nature 108:593-615.

Sazuka T, Tomooka Y, Kathju S., Ikawa Y, Noda M, Kumar S (1992) Identificiation of a develomentally regulated gene in the mouse central nervous system which encodes a novel proline rich protein. Biochim Biophys Acta 1132:240-248.

Scheetz AJ, Williams RW, Dubin MW (1995) Severity of ganglion cell death during early postnatal development is modulated by neuronal activity and binocular competition. Vis Neurosci: in press.

Smeyne RJ, Goldowitz D (1989) Development and death of external granular layer cells in the weaver mouse cerebellum: A quantitative study. J Neurosci 9:1608-1620.

Silver J (1985) Confidence limits for estimates of gene linkage based on analysis of recombinant inbred strains. J Hered 76:436-440.

Siracusa, L.D., V.M. Chapman, K.L. Bennett, N.D. Hastie, D.F. Pietras, and J. Rossant (1983) Use of repetitive DNA sequences to distinguish Mus musculus and Mus caroli cell by in situ hybridization. J. Embryol. Exp. Morph 73:163-178.

Stent GS (1981) Strength and weakness of the genetic approach to the development of the nervous system. Annu Rev Neurosci 4:163

Sterling P (1983) Microcircuitry of the cat retina. Annu Rev Neurosci 6:149-185.

Stewart RR, Spergel D, Macagno ER (1986) Segmental differentiation in the leech nervous system: The genesis of cell number in the segmental ganglia of Haemopis marmorata. J. Comp. Neurol. 253:253-259.

Sweet HO (1993) Breeding schemes for recessive mutations. JAX Notes 454:6.

Takahashi JB, Hoshimaru M, Kikuchi H, Hatanaka M (1993) Developmental expression of trkB and low-affinity NGF receptor in the rat retina. Neurosci Lett 151:174-177.

Taylor B, Truman JW (1992) Commitment of abdominal neuroblasts in Drosophila to a male or female fate is dependent on genes of the sex-determining hierarchy. Development 114:625-642.

Taylor BA (1972) Genetic relationships between inbred strains of mice. J Hered 63: 83-86.

Taylor BA (1976) Development of recombinant inbred lines of mice. Behav Genet 6:118.

Taylor BA (1989) Recombinant inbred strains. In Genetic Variants and Strains of the Laboratory Mouse, 2nd ed. Lyon MF, Searle AG (eds), Oxford Univ Press, Oxford, pp.773-796.

Taylor BA, Phillips, SJ (1994) Pooled-SSR, A strategy for rapidly mapping new quantitative trait loci (QTLs). Mouse Genome Conference

Thorn RS, Truman JW (1994) Sexual differentiation in the CNS of the mouth, Manduca sexta. I. Sex and segmen-specificity in production, differentiation, and survival of the imaginal midline neurons. J Neurobiol 25: 1039-1053.

Travis, G.H., Brennan, M.B., Danielson, P.E., Kozak, C.A., and Sutcliffe, J.G. (1989) Identification of a photoreceptor specific mRNA encoded by the gene responsible for retinal degeneration slow (rds). Nature 338:70-73

van Abeelen JHF (1980) Direct genetic and maternal influences on behavior and growth in two inbred mouse strains. Behavior Genet 10:545-551.

Waddington CH (1957) The Strategy of the Genes. London: Allen and Unwin.

Wahlsten D (1983) Maternal effects on mouse brain weight. Dev Brain Res 9:215-221.

van Deusen, E. (1973) Devel. Bio. 34:135-158. Axolotl retinal cell transplants into eye results in well defined radial sectors.

Walls, G.L. (1942) The Vertebrate Eye and Its Adaptive Radiation. Hafner, New York (or the Cranbrook Press, Bloomfield Hills, MI)

Watanabe, T. and Raff, M.C. (1990) Neuron 2:461-467.

Wayne RK, Modi WS, O'Brien SJ (1986) Morphological variability and asymmetry in the cheetah (Acinonyx jubatus), a genetically uniform species. Evolution 40:78-85.

Wetts, R., and S.E. Fraser (1987) Multipotent precursors can give rise to all major cell types of the frog retina. Science 239:1142-1145.

Whitney G, McClearn GE, DeFries, JC (1970) Heritability of alchohol preference in laboratory mice and rats. J. Hered. 61:165-169. Wikler KC, Williams RW, Rakic P (1990) Photoreceptor mosaic: Number and distribution of rods and cones in the rhesus monkey retina. J Comp Neurol 297:499-508.

Williams GC (1992) Natural Selection. Domains, Levels, and Challenges. Oxford Univ Press, New York (p. 92).

Williams MA, Pi\0x96on, LGP, Linden R, Pinto LH. (1990) The pearl mutation accelerates the schedule of natural cell death in the early postnatal retina. Exp Brain Res. 82:393-400.

Williams RW (1991) The human retina has a cone-enriched rim. Vis Neurosci 6:403-406.

Williams RW (1994) The portable dictionary of the mouse genome: A personal database for gene mapping and molecular biology. Mamm Genome 5:372-375.

Williams RW, Chalupa LM (1982) Prenatal development of retinocollicular projections in the cat: An anterograde tracer transport study. J Neurosci 2:604-622.

Williams RW, Rakic P (1985) Dispersion of growing axons within the optic nerve of the embryonic monkey. Proc Natl Acad Sci USA 82: 3906-3910.

Williams RW, Bastiani MJ, Lia B, Chalupa LM (1986) Growth cones, dying axons, and developmental fluctuations in the fiber population of the cat's optic nerve. J Comp Neurol 246:32-69.

Williams RW, Herrup K (1988) The control of neuron number. Ann Rev Neurosci 11:423-453.

Williams RW, Rakic P (1988a) Elimination of neurons from the lateral geniculate nucleus of rhesus monkeys during development. J Comp Neurol. 272:424-436.

Williams RW, Rakic P (1988b) Three-dimensional counting: An accurate and direct method to estimate cell numbers in sectioned material. J Comp Neurol: 278:344-352.

Williams RW, Cavada C, Reinoso-Suarez F (1993) Rapid evolution of the visual system: A cellular assay of the retina and dorsal lateral geniculate nucleus of the Spanish wildcat and the domestic cat. J Neurosci 13:208-228.

Wilkinson, D.G., J.A. Bailes, A.P. McMahon (1987) Expression of the proto-oncogene int-1 is restricted to specific neural cells in the developing mouse embryo. Cell. 50:79-88.

Williams RW, Goldowitz D (1992a) Structure of clonal and polyclonal cell arrays in chimeric mouse retina. Proc Natl Acad Sci USA 89:1184-1188.

Williams RW, Goldowitz D (1992b) Lineage versus environment in embryonic retina. A revisionist perspective. Trends Neurosci 15:368-373.

Wimer C and Prater L (1966) Some behavioral differences in mice genetically selected for high and low brain weight. Psychol Rep 19:675\0x1F681.

Wimer RE, Wimer CC, Roderick TH (1969) Genetic variability in forebrain structures between inbred strains of mice. Brain Res 16:257-264.

Wimer RE, Wimer CC, Vaughn JE, Barber RP, Balvanz BZ, Chernow CR (1976) The genetic organization of neuron number in Ammon's horns of house mice. Brain Res 118:219-243.

Wright S (1978) Evolution and the Genetics of Populations. Vol 4. Variability within and among Natural Populations. The University of Chicago Press.

Young RW (1984) Cell death during differentiation of the retina in the mouse. J Comp Neurol 229: 362-373.

Young RW (1985) Anat Rec 212:199-205.Young RW (1983) The life history of retinal cells. Tr. Am Ophth Soc 81: 193-228.


Since 11 August 98


Neurogenetics at University of Tennessee Health Science Center

Print Friendly Top of Page

Home Page  |  Genome DBs  |  Phenome DBs  |  Publications  |  People & Associates
Mouse Brain Library  |  Related Sites  |  |   MBL.ORG

Robert W. Williams | Alex Williams © 2002, modify this page