Research in the Glaser lab is aimed at understanding the molecular basis of eye development in mammals, using biochemical and genetic approaches, knockout and transgenic mouse models, and mutational analysis of human hereditary eye disease. Projects involve transcription factors and pathways that control patterning, cell fate specification and differentiation, which have been deeply conserved during evolution.
Related projects involve the auditory hindbrain and general aspects of growth control, cell competition, and compartmentation during embryonic development.
1979 - B.S. Stanford University (Biology)
1979 - M.S. Stanford University (Biology)
1988 - Ph.D. Massachusetts Institute of Technology(Biology, Prof. David Housman)
1991 - M.D. Harvard Medical School (Health Sciences and Technology Program)
1991-1994 Research Associate, Harvard Medical School (Prof. Richard L. Maas)
Honors and Awards
1979 - 1984 National Science Foundation Graduate Fellowship
1984 - 1990 Medical Scientist Training Program Fellowship, Harvard Medical School
1987 - 1988 Rockefeller Foundation Fellowship, Harvard Medical School
1989 - Sigma Xi Scientific Society
1991 - Magna Cum Laude, Harvard Medical School (M.D.)
1991 - 1994 Howard Hughes Medical Institute, Postdoctoral Research Fellowship
1994 - 1999 Howard Hughes Medical Institute, Assistant Investigator
2001 - Jerome W. Conn Award for Distinguished Research by a Junior Faculty Member, University of Michigan, Dept of Internal Medicine
2003 - Election to American Society for Clinical Investigation (ASCI)
Brown, NL, Kanekar S, Vetter ML, Tucker PK, Gemza D and Glaser T. Math5 encodes a murine basic helix-loop-helix transcription factor expressed during early stages of neurogenesis. Development 125:4821-4833, 1998.
Lauderdale JD, Wilensky JS, Oliver ER, Walton DS and Glaser T. 3’ deletions cause aniridia by preventing PAX6 expression. Proc. Natl. Acad. Sci. USA 97:13755-13759, 2000.
Brown NL, Patel S, Brzezinski J and Glaser T. Math5 is required for retinal ganglion cell and optic nerve formation. Development 128:2497-2508, 2001.
Tucker P, Laemle L, Munson A, Kanekar S, Oliver ER, Brown NL, Schlecht H, Vetter M and Glaser T. The eyeless mouse mutation (ey1) removes an alternative start codon from the Rx/rax homeobox gene. Genesis 31:43-53, 2001.
Brown NL, Dagenais SL, Chen C-M and Glaser T. Molecular characterization and mapping of ATOH7, a human atonal homolog with a predicted role in retinal ganglion cell development. Mammalian Genome 13:95-101, 2002.
Farjo R, Yu J, Othman MI, Yoshida S, Sheth S, Glaser T, Baehr W and Swaroop A. Mouse eye gene microarrays for investigating ocular development and disease. Vision Res. 42:463-470, 2002.
Oliver ER, Saunders TL, Tarle SA and Glaser T. Ribosomal protein L24 defect in Belly spot and tail (Bst), a mouse Minute. Development 131:3907-3920, 2004.
Brzezinski JA and Glaser T. Math5 establishes retinal ganglion cell competence in a subpopulation of postmitotic progenitor cells, submitted.
1. Math5 mutant mice
2. Cell fate determination in the retina - expression fate mapping
3. Mouse Minutes - riboproteins, tissue growth control and mutant phenotypes
4. Eyeless mice - the Rx transcription factor
5. PAX6 - novel mutant alleles and functions
Additional Lab Links
Nadean Brown, Ph.D.
Kris Coulter, M.D. Ph.D.
Colin Hodgkinson, Ph.D.
Jim Lauderdale, Ph.D.
Gene networks controlling metazoan eye development have been deeply conserved during evolution. An important component in this hierarchy is the Atonal (ato) family of basic helix-loop-helix (bHLH) transcription factors. In the fruit fly Drosophila, the proneural gene ato controls development of the antennae, chordotonal stretch receptors and compound eyes. Within the eye imaginal disc, ato is expressed at the morphogenetic furrow and in ommatidial preclusters, where it is critically required to specify the first photoreceptors (R8). ato mutant flies lack photoreceptors. A few years ago, our lab identified a mammalian homolog, Math5 (Fig. 1), which is highly expressed in the developing eye, in a pattern that coincides with the onset of retinal histogenesis (Brown et al. 1998).The Math5 gene is also designated Atoh7.
Fig. 1 The Math5 bHLH transcription factor (149 amino acids) is predicted to dimerize with E12 proteins (E47, daughterless) and bind to E-box sites in genomic DNA (CANNTG).
We have generated Math5 knockout mice, with a lacZ cassette inserted in the solitary coding exon (Fig. 2). The pattern of lacZ staining reflects endogenous Math5 activity. It begins at E11 in retinal progenitor cells, within the undifferentiated neuroepithelium, and perdures in early-born retinal neurons. Homozygous Math5 ?]/- mice are fully viable, but lack all retinal ganglion cells (RGCs) and optic nerves (Brown et al. 2001).
Fig. 2. Math5 knockout mice. a. targeting strategy. b,c. Math5-lacZ expression in the developing retina. Uncommitted progenitors (arrowhead) and newly formed RGCs (arrow) are indicated. d. model showing progression of hypothetical competence states for different retinal cell types, which develop from multipotent progenitors by a lineage-independent process (after Livesey and Cepko, 2001). The pathway leading to RGC fate is blocked in Math5 mutants. e-j.Math5 phenotypes, including absent optic nerves and chiasm (e,f, ventral brain views) and RGCs (g,h, hematoxylin and eosin; i,j, NF160 antibody stain). RGCs comprise ca. 50% of cells in the GCL.† Neurofilament 160kDa marks horizontal (arrowhead) and ganglion (arrow) cells.
Math5 mutant eyes detect light but are not connected to the brain (Fig. 3). In other respects, the retinas, extraocular muscles and anterior eye structures are remarkably well formed. RGCs are the first class of retinal neurons to exit mitosis and differentiate in all vertebrates. Ganglion cell death is the common pathogenic event in the human disease glaucoma, a leading cause of blindness.† Math5 appears control the RGC fate specification step.
Fig. 3 The optic nerves link the eyes to the brain.
Current projects involve detailed analysis of Math5 mutants, including:
Retinal electrophysiology and circadian behavior.
Abnormal vascular and astrocyte development. During normal eye development, a network of hyaloid blood vessels extends through the vitreous compartment, perfusing the retina and lens. These vessels normally regress after birth, but in Math5 mutants, they proliferate and invade the neuroepithelium. Normal angiogenesis does not occur in Math5 ?]/- retinas compared to wildtype (Fig. 4). The pattern and extent of astrocyte migration from the optic stalk is likewise highly unusual. This may occur because of anatomical defects in the stalk or a deficiency of trophic factors that are normally secreted by RGCs.
Fig. 4. Retinal flatmount preparations. a-d. Defective angiogenesis (Griffonia simplicifolia lectin stain). In wildtype mice, retinal vessels grow radially from the hyaloid artery remnant (arrow) and branch during the first postnatal week. This process fails in Math5 ?]/- mutants. Instead, fetal vessels collapse onto the retinal surface from the vitreous (wavy arrows). e-h. Defective astrocyte migration (glial acidic fibrillary protein antibody). High (e,f) and low (g,h) power views showing an abnormal distribution of astrocytes in the absence of RGCs and vascular arbors. i. In wildtype mice, retinal vessels and astrocytes are coextensive.
Profiling mRNAs from developing and adult retinas of Math5 -/- and wildtype mice, using custom cDNA microarrays (Farjo et al. 2002) and genechips, to identify Math5 targets and RGC-specific transcripts.
Secondary Math5 expression domains in the auditory hindbrain and cerebellum, and potential associated phenotypes.
Mutational analysis of Math5 (ATOH7) in patients with optic nerve aplasia and related defects (Brown et al. 2002).
PROJECT 2: Cell fate determination in the retina ñ expression fate mapping
How does Math5 specify RGCs? In theory, Math5 could behave as an instructive factor, driving previously committed progenitor cells through an RGC differentiation program. Or Math5 could act as a permissive factor, making progenitors competent to develop as RGCs in response to further cues. In each case, Math5 is necessary for RGC fate determination. However, in the first model (instructive), Math5+ progenitors develop exclusively as RGCs, whereas they develop as multiple cell types in the second model (permissive).
To investigate the mechanism of Math5 action, we have tested:
the timing of Math5 expression with respect to the terminal cell division, and
the fate of retinal cells descending from Math5+ progenitors (lineage analysis).
Within the optic cup, proliferating retinoblasts are tethered across the epithelium. Interkinetic nuclear migration occurs during each cell cycle, with M phase nuclei located at the outer (sclerad) surface and S phase nuclei in the middle. Differentiating RGCs accumulate at the inner (vitread) surface. Our results show Math5 is expressed in post?]mitotic cells (Fig. 5). The pattern within the neural epithelium suggests that Math5 is transcribed soon after progenitors complete their terminal M phase.
Fig. 5 Math5-lacZ is not expressed in proliferating cells. Heterozygous E15.5 retinas stained with antibodies to lacZ (green) and BrdU (a) or phosphohistone H3 (b) (red). Embryos were collected one hour after exposure to BrdU, which labels S phase cells. PH3 labels cells in M phase. The staining patterns do not overlap. Similar results were obtained with the proliferation marker Ki67.
Lineage analysis was performed using an expression fate mapping strategy (Fig. 6). We first inserted a Cre recombinase cassette in place of the Math5 coding region, within a bacterial artificial chromosome (BAC). This recombinant BAC contains >100 kb of flanking genomic DNA on each side, to ensure appropriate in vivo transcriptional regulation.† We then generated Math5?]Cre BAC transgenic lines, and crossed these mice to reporter strains carrying lox?]stop?]lox transgenes (Z/AP, R26?]GFP), which are expressed from ubiquitous promoters and activated upon Cre?]mediated excision.
Fig. 6. Math5 lineage analysis. a. Fate?]mapping with the Z/AP reporter. In Math5+ retinal progenitor cells, Cre catalyzes recombination between loxP sites, causing excision of the bgeo?]pA segment. This activates the hPLAP gene (human placental alkaline phosphatase), which is expressed from the CAG promoter in all descendant neurons. The time course is illustrated in panels b-d (E15.5 retinal sections). b.in situ hybridization showing Math5 mRNA in progenitor cells. c.b?]galactosidase stain of heterozygous knockout embryo showing stable Math5?]lacZ activity in progenitors (sclerad) and newly formed RGCs (vitread). d. Lineage?]marked cells in a double transgenic embryo (alkaline phosphatase stain). The steps leading from Math5?]Cre transcription to hPLAP activation occur at the same time progenitors are undergoing nuclear migration and cell fate determination. e. E14.5 eye showing lineage?]marked RGCs and optic nerve (AP stain, purple) and S phase nuclei (BrdU antibody, brown).
In this system, cells descending from Math5+ progenitors are indelibly marked, with hPLAP or GFP, and can be classified in adult retinas according to cell type (Fig. 7). They do not re?]enter the cell cycle.
Fig. 7. Math5+ progenitors are competent to form multiple cell types. Retinal section from an adult Math5?]Cre, Z/AP transgenic mouse, stained with antibodies to hPLAP (green) and BrdU (red), and the nuclear dye DAPI (blue). A cumulative BrdU protocol was used to label cells exiting mitoses between days E10.5 and P0. A variety of cell types are positive for the AP lineage marker, including ganglion cells (g), cones (c, with large pedicles in the outer plexiform layer), horizontal cells (h), amacrines (a) and rods (r). All of the AP+ cells were born before P0. OS, photoreceptor outers segments.
We estimate that 5% of adult retinal neurons are descended from Math5+ progenitors. In addition to RGCs, this cohort contains representatives of all major cell types. The lineage?]marked fraction varies from ca. 30% (cones) to 0.1% (bipolar cells), in accordance with the population birthdates for different cell types and the time course of Math5 expression. Dual?]reporter concordance data suggest that the aggregate level of Cre activity is similar in all Math5+ progenitors, regardless of whether they develop as RGCs or other cell types. Taken together, our results show:
Math5 acts as a permissive factor for RGC development within a subset of postmitotic progenitors. The majority of retinal cells do not transit through a competence state defined by Math5 expression.
Retinal progenitors remain multipotent after they have exited the cell cycle. Fate determination is not strictly coupled to terminal mitosis.
Additional factors - positive or negative, cell intrinsic or environmental ñ are needed to specify RGC fate completely. Current experiments are designed to find them.
PROJECT 3: Mouse Minutes - riboproteins, tissue growth control and mutant phenotypes
Mice with the spontaneous belly spot and tail mutation (Bst/+) have a profound deficiency of retinal ganglion cells, and retinovascular defects that are reminiscent of Math5 ?]/- mutants. The onset of retinal histogenesis is delayed in Bst/+ embryos, and the choroid fissure closes relatively late, causing segmental defects in the optic nerve head (colobomas).
In addition to eye defects, Bst/+ mice have pleiotropic phenotypes including a midventral white spot, vertebral malformations and preaxial polydactyly (Fig. 8). Homozygous embryos die before implantation.
We recently identified Bst as a deletion within Rpl24 (Oliver et al. 2004), a gene encoding ribosomal protein L24, a critical component of the 60S subunit, which is required for protein synthesis in all cells. The mutation impairs mRNA splicing, truncating the L24 polypeptide. Rpl24 BAC and cDNA transgenes rescue these phenotypes. The tissue specificity of Bst phenotypes is surprising, given the universal cellular requirement for ribosomes.
Fig. 8. Bst mutation in the L24 ribosomal protein gene (Oliver et al. 2004).† a.Bst/+ phenotype, with belly spot, white hindfeet and vertebral defects (tail kinks). Bst/+ mice are ~20% smaller than wildtype littermates. b. Major deficiency of RGCs. c. Comparable Minute fruit fly (right) with small thin bristles, blunted wings and an abnormal body shape. d. Bst deletes the exon 1 splicing branchpoint of Rpl24, leading to protein truncation. e. L24 (green) is a component of the large (60S) ribosomal subunit, where it is oriented toward the peptidyl transferase catalytic site. L24 is conserved between eukaryotes and archaebacteria; the structure shown is derived from halophile H. marismortui (Ban et al. 2000). f. Overview of mammalian ribosome biogenesis. The 40S and 60S subunits are assembled from four rRNAs and 79 riboproteins (RPs) in a highly coordinated process involving >100 accessory proteins and snoRNAs. This accounts for up to 80% of metabolism in growing cells.† g. The L24 deficiency in Bst/+ mice slows processing of rRNA precursors (left) but does not affect the steady state abundance of ribosomes (right). Liver RNA was pulse labeled in vivo with 32P and resolved by electrophoresis. h. Cell competition in vivo. Bst/+ and +/+ blastocysts were injected with five ROSA26 lacZ ES cells, and the resulting chimeras evaluated by their agouti coat content and lacZ staining patterns. The ROSA26 cells have an obvious growth advantage in Bst/+ chimeras, similar to that +/+ clones in Minute fly mosaics. i.Bst/+ fibroblasts grow significantly slower than wildtype MEFs in tissue culture.
Bst/+ mice have normal steady state levels of 60S and 40S subunits, but show a significant delay in ribosome biogenesis. Fibroblasts from Bst/+ embryos (MEFs) have decreased overall rates of protein synthesis and grow slower than wildtype, with a prolonged G1 phase. In blastocyst ROSA26 lacZ ES cell chimeras, Bst/+ cells have a significant growth disadvantage (Fig. 8).
Bst is thus equivalent to the large group of Drosophila mutations known as Minutes (M), which were first described in the 1920s. Minutes were instrumental to classical studies that led to concepts of cell autonomy, growth control, cell competition, and developmental compartmentation. M/+ flies are slightly smaller than +/+ flies, and require 3-4 additional days to progress through larval stages. Wildtype somatic clones (mitotic segregants) arising in M/+ flies outgrow surrounding M/+ cells, but do not cross developmental boundaries. The magnitude of cell competition depends on the differential growth rate and is mediated by TGFb signaling. In this way, cell ëfitnessí is assessed and positively selected during organogenesis. Comparable cell scrutiny is likely to occur in vertebrates.
In spite of their abundance and celebrated history in Drosophila research, riboprotein mutations are relatively unexplored in vertebrates. One human genetic disease, Diamond?]Blackfan anemia (pure red blood cell aplasia), is associated with heterozygous mutations in the 40S subunit riboprotein gene RPS19, but the mechanism of pathogenesis is unknown.
Current projects are designed to:
Understand biochemical effects of the Bst mutation and the developmental basis for tissue?]specific phenotypes, including possible extraribosomal functions of L24.
Investigate cell competition effects in chimeras and mosaic mice.
Define the spectrum of mouse riboprotein phenotypes, using targeted mutations in different RP genes, as potential models for human congenital malformation syndromes.
PROJECT 4: Eyeless mice ñ the Rx transcription factor
Early patterning of the optic cup and lens involves a network of inductive signaling and gene regulation. The inbred mouse strain ZRDCT is eyeless because of an interaction between a small number of genetic factors, originally proposed by Herman Chase in the 1940s. In collaboration with Priscilla Tucker (UM Museum of Zoology) and Lois Laemle (UMDNJ Ophthalmology), we identified the major eyeless determinant (ey1) as a hypomorphic mutation in the Rx gene (Tucker et al. 2001). This paired class homeodomain transcription factor is expressed in the anterior neural fold and optic primordia beginning at E8, and is required to form the optic vesicles. Rx is the earliest marker of retinal development. Rx ?]/- knockout mice lack eyes, and die at birth with brain and craniofacial defects. The ey1 mutation (M10L) removes a conserved alternative start codon, which reduces the abundance of Rx protein by 4?]6 fold (Fig. 9).
Fig. 9. Rx (M10L) mutation in an eyeless mouse strain. a. (left) Phenotype of ZRDCT mice. (right) in situ hybridization of wildtype E10.5 embryo showing Rx mRNA in the optic cup. b. PCR microsatellite analysis of DNA pools from eyeless intercross mice (F2 progeny). The skewed distribution of Z alleles indicates linkage between ey1 and D18Mit17. c. Segregation data mapping ey1 near Rx on chromosome 18.† d. Map showing Rx protein structure, genomic organization, and the M10L mutation (Tucker et al. 2001).
ZRDCT mice are viable and fertile. In crosses to other laboratory strains, ey1 acts as one component in a complex trait. Homozygosity for the Rx(M10L) allele is necessary but not sufficient to produce the eyeless phenotype.
Current projects are designed to:
Map ey2 modifier loci on this sensitized genetic background. These are likely to interact, directly or indirectly, with Rx in a common pathway for eye morphogenesis.
Evaluate Rx gene evolution in rodents, testing whether Rx has experienced directional selection or relaxed constraint during eye regression in the evolutionary lineage leading to the blind mole rat Spalax ehrenbergi. This murid rodent has tiny subcutaneous eyes, the smallest of any mammal, and is widely distributed in the Middle East. Its visual system is anatomically similar to ZRDCT mice.
Screen human cases with anophthalmia or severe microphthalmia for mutations in RX and other candidate genes.
PROJECT 5: PAX6 ñ novel mutant alleles and functions
The PAX6 transcription factor (MIM 607108) has been deeply conserved during metazoan evolution. It contains two DNA?]binding domains ñ a paired box and a homeobox - and is critically required for eye morphogenesis in a wide variety of species, including humans, mice, rats, zebrafish and Drosophila (Fig. 10). Vertebrate PAX6 is expressed throughout the developing eye. It is also essential for proper development of the central nervous system, olfactory epithelium and endocrine pancreas.
PAX6 functions near the top of a regulatory hierarchy. When ectopically expressed in Drosophila imaginal discs, PAX6 can initiate a genetic program leading to eye formation. Similar effects have been observed in Xenopus, particularly when PAX6 is coexpressed with other factors.
Fig. 10. PAX6 analysis. a. Map showing the human PAX6 cDNA, protein domains, exon structure, and representative mutations. The dark symbols show mutations identified at UM. Mouse (M) and rat (R) Sey alleles are indicated for comparison. b. Model showing the paired domain in contact with DNA, with six alpha helices and two subdomains (after Xu et al. 1995). Amino acid substitutions identified in human (yellow, blue) and C. elegans (red) PAX6 genes are marked. c.Sey/+ mouse. d. Gene dosage effect in Sey embryos. The homozygous mutant is eyeless. e.PAX6 regulatory allele and human aniridia phenotype. f. Pedigree with a de novo rearrangement 3’of the PAX6 gene. The Southern blot was hybridized with an exon 13 probe The mutation causes aniridia by preventing PAX6 transcription.
Vertebrate eye development is highly sensitive to PAX6 gene dosage. In homozygous mutants, lens induction fails to occur. Eye formation arrests after the optic vesicle stage, producing an eyeless phenotype. In heterozygotes, reduced PAX6 activity causes Small eye (Sey) phenotypes in laboratory mice and rats, and the panocular disease aniridia (MIM #106210) in humans. Aniridia affects 1 in 50,000 births. It occurs sporadically, in children with new PAX6 mutations, or as autosomal dominant trait in pedigrees. Clinical features include severe iris and foveal hypoplasia, nystagmus, and variably progressive glaucoma, cataracts and corneal opacification. Milder phenotypes are associated with partially functional missense alleles. Understanding these mutations is clinically important, and provides molecular information regarding PAX6 structure and function, interacting proteins and transcriptional regulation.
Our lab has characterized a number of PAX6 mutations, working with clinical colleagues. Recent studies have focused on hypomorphic alleles within the paired domain, alternative exon 5a, and regulatory mutations affecting a set of 3’ enhancer elements located >100 kb from the PAX6 transcription unit, within the intron of a neighboring gene (Lauderdale et al. 2000). Current projects include analysis of a dominant?]negative allele associated with severe microphthalmia.