Models of Müller Glial Cell Disruption and the Consequences on Retinal Health and Visual Function in the Zebrafish

Retina NICOLAS A. YANNUZZI
Department of Molecular and Cellular Biology, Harvard University, Cambridge, Massachusetts 02138

Statement of Research This thesis was completed under the guidance of Dr. Pamela M. Kainz and Dr. John E. Dowling at the Department of Molecular and Cellular Biology at Harvard College. Research was conducted from September, 2005 - May, 2006 and from September, 2006 - April, 2007.

2

ACKNOWLEDGEMENTS I began working at the Dowling Lab with the intention of completing my single semester research requirement. Within a short time, however, I found that the

opportunity to conduct my own research was not only a privilege, but also a chance for me to witness the active pursuit of scientific discovery. The completion of this thesis was the most meaningful, enjoyable, and exciting part of my academic experience at Harvard. It revived my love for science and research, and I am thankful for the freedom I was given to express my thoughts and ideas during the project. I would like to thank Dr. Pamela Kainz for all of her help and support during the process and for teaching me everything I know about scientific research, from experimental design to thoughtful data interpretation. Dr. Kainz is not only a great scientist, but also a great teacher. I would also like to thank the other members of the Dowling Lab for helping me during the process. Finally, I would like to thank my mother for her support and my father for instilling in me a passion for scientific thought and discovery. His achievements in ophthalmology inspired me to pursue research on the retina.

3

ABSTRACT Müller cells are the primary glial cells of the vertebrate retina. They contact and ensheath every retinal cell type and regulate neuronal activity. Recent studies have suggested that Müller cells also provide specific supportive roles for the survival and function of photoreceptor cells, but there are currently few models where this relationship can be explored. The purpose of this study was to evaluate two candidate models of Müller cell disruption and to observe the consequences on retinal health and visual function. In the first chapter I investigated a potential pharmacological model of Müller cell stress. Using the gliotoxin α-aminoadipic acid (α-AAA), I observed only a modest sign of Müller cell stress in adult zebrafish, while the effect on larvae was not glial-specific indicated by the presence of pyknotic nuclei among retinal neurons. Since α-AAA failed to produce clear and reproducible signs of glial cell defects, I decided to discontinue my pursuit of this model and instead investigate a potential genetic model of Müller cell disruption by characterizing the rose mutant, an endothelin receptor B (ET-B) gene knockout. I found that the morphology of the larval rose retina appeared normal. However, when exposed to constant light, wild type larvae were unaffected while the rose larval retina suffered rod outer segment disruption, loss of 10% of the cells in the inner nuclear layer, where the Müller cell bodies lie, and a decrease in visual sensitivity. These defects are consistent with the hypothesis that compromised Müller glial cells lead to a decrease in photoreceptor cell resilience. Questions remain as to what extent Müller cells are involved in the light exposed rose phenotype, but this study provides the groundwork for continued exploration concerning how the absence of ET-B compromises the retina and the ways Müller cells may be involved in the preservation of photoreceptor cells.

4

GENERAL INTRODUCTION Glial cells provide physical support and protection for neurons and guide migrating neurons to their destinations in the brain during development. Glia have also been shown to act as a template for axonal migration in the neural retina (Silver and Rutishauser, 1984). Furthermore, they regulate the formation of synapses that enable neuronal correspondence and promote the survival of some neurons while playing a role in the birth of others (Helmuth, 2001). Once dismissed by neuroscientists as playing a minimal role in the nervous system, glia have recently been suspected to be involved significantly in the pathogenesis of certain diseases. Research has shown that glia are integral to the understanding and causes of neuropathic pain, epilepsy, and neurodegenerative diseases (Miller, 2005). Studies have also suggested that glia may offer a new range of therapeutic targets in a variety of diseases including Multiple Sclerosis and some psychiatric disorders, where post mortem studies have demonstrated that there are abnormal amounts of glia in certain areas of the brain (Miller, 2005). Research on glia has not been limited to their function in the brain, where they outnumber neurons in a ratio of ten to one. Glia have also been studied in the context of the vertebrate retina. Müller glia are the primary support cells in the retina. They contact and ensheath every retinal cell type and regulate neuronal activity by controlling extracellular ion concentration and by recycling excess neurotransmitters used during signaling processes (Newman and Reichenbach, 1996; Sarthy and Ripps, 2001). Specifically, they express gated channels and neurotransmitter receptors which can cause depolarization and intracellular Ca2+ waves. In addition, they transport K+ and glutamate

5

and regulate retinal pH levels via carbonic anhydrase (Newman and Reichenbach, 1996). Although there are more neurons than Müller cells, Müller cell processes contact and ensheath the synaptic and nuclear region of every cell in the retina; their nuclei reside in the inner nuclear layer, which also contains horizontal, amacrine and bipolar cell nuclei, and their endfeet project from the ganglion cell layer (Fig. 1), allowing for extensions that span the entire depth of the neural retina (Peterson et al., 2001). Their apical processes also contact the inner segments of photoreceptors. This framework provides a basis for interaction with every neuron.

6

Apical Processes

Müller Cell Body

Müller Endfeet
Fig. 1. The various layers of the retina are labeled for convenience (Source: The purple and white illustration, except for the illustration of the Müller cell (right) and the photograph of the adult zebrafish retina (left), is courtesy of The Washington University School of Medicine). Müller cells and their processes span the width of the neural retina. Their cell bodies reside in the inner nuclear layer, and their endfeet project out of the ganglion cell layer, while their apical processes reach the outer segments. Rod outer segments are also labeled and are located directly under the pigment epithelium.

7

Both clinical and scientific investigators have suggested that Müller cells play a vital role in several retinal diseases such as X-linked Retinoschisis (Reid et al., 1999), Cystoid Macular Degeneration (Loeffler et al., 1992), Idiopathic Macular Holes, and Foveomacular Schisis (Gass, 1999). There has even been some suspicion that Müller cell disease may be a principal agent in certain forms of Age Related Macular Degeneration, the leading cause of blindness (DiLoreto et al., 1995). Clinical researchers suggest that Müller cells are not necessarily a cause of disease. Instead, they believe that Müller cells become reactive and hypertrophic in response to photoreceptor damage and eventually lead to scarring. Because the activation of Müller cells in response to preexisting

problems with the neural retina is so common, it has been the main focus in the research of retinal glia. Virtually no clinical studies and very few animal studies have explored how Müller cells themselves could be the primary cause of failing retinal health or function and not just a response to preexisting damage. Thus, while some of the functions of Müller cells have been determined, the precise relationship between Müller cells and retinal cell health and maintenance is not yet fully understood. The effect of Müller cell disruption on photoreceptor cells was the specific relationship that I set out to explore in this study. The discovery that stressing or eliminating Müller cells could lead to

photoreceptor degeneration (DiLoreto et al., 1995) provided in part, the inspiration for this thesis. The chosen model organism for studying the relationship between retinal glia and neurons was the zebrafish. Zebrafish are small, approximately one inch long freshwater fish that can be raised inexpensively and in large numbers. Embryos develop rapidly

8

resulting in zebrafish eyes that exhibit light response in just 3 days post fertilization (dpf) (Brokerhoff et al., 1995). The zebrafish retina has the same cell classes and architecture as other vertebrates, including humans. However, unlike the mammalian retina, which contains astrocytes, the zebrafish retina contains only Müller glia and sparse microglia. An advantage of using the larval zebrafish is that it lacks scales; therefore, the absorption of a drug can occur readily through its skin. Finally, the visual behavior of larvae can be tested using the optokinetic response assay (OKR), in which their eyes track the movement of vertical black and white stripes passing through their visual field. This response is not only common to zebrafish but to all vertebrates. Using this assay, their visual threshold can be quantified by determining the lowest level of light at which their eyes consistently track the moving stripes. The forefront of research concerning the role of Müller cells in the zebrafish retina has focused on a genetic mutant called lazy eyes (lze). Larvae homozygous for the lze mutation at 5 dpf respond much less robustly compared to wild type in the OKR assay. Histological observations revealed that the mutation seemed to affect selectively the Müller cells and photoreceptor cells. Some mutant retinas contained fewer Müller cells than wild type retinas, while others contained Müller cells that appeared hypertrophied or unhealthy. Most lze retinas had fewer rods and small cone outer

segments. The combined effects of light stress and genetic manipulation were also studied, and constant light was found to accelerate drastically the Müller cell degeneration and to accentuate the lze functional deficit (Kainz et al., 2003). The lze mutant is thus a striking example of the special relationship between Müller cells and photoreceptor cells. However, larvae homozygous for the lazy eyes mutation for some

9

unknown reason die by 10 dpf. Therefore, there has yet to be a method of studying the effects of Müller glial cell inactivity on the photoreceptor cells of the adult zebrafish retina. The overarching goal of my study was to explore two other potential models of Müller cell disruption, one pharmacological and one genetic, and to determine if either of these models resulted in compromised retinal health or function with special attention to photoreceptors. For each model, photoreceptor stress, in the form of constant light, was introduced and a morphological examination of the photoreceptors and Müller cells was performed as well as an assessment of the function of the retina. In chapter one of this thesis, I investigated a pharmacological model of Müller cell disruption by characterizing the effects of the gliotoxin α-aminoadipic acid (α-AAA) on the wild type adult and larval zebrafish. In chapter two, I examined a second potential model: a zebrafish mutant missing the gene encoding endothelin receptor B (ET-B), which has been shown to be expressed highly on Müller cells and involved in the response to light-induced stress on photoreceptors (Rattner and Nathans, 2005).

10

Chapter 1: The Pharmacological Model INTRODUCTION One of the methods that has been used to study the aftermath of glial cell dysfunction or degeneration is the administration of the DL-isomer of α-aminoadipic acid (α-AAA), a glutamate analogue known to be gliotoxic to the central nervous system, including the retina (Olney et al., 197; Casper and Reif-Lehrer, 1983). The effects of αAAA have been studied in mice through subcutaneous injection. The regularity,

thickness, and number of Müller cell processes were reduced in the inner retina (Reid and Farber, 2004). An electrophysiological and morphological study carried out in the carp retina revealed swollen Müller glial cells after one to two months of intraocular injections with α-AAA at varying doses (Sugawara et al., 1990). Another study found that after treating embryonic Xenopus retinas in culture with α-AAA, photoreceptor outer segments became shortened and dysmorphic, further demonstrating the relationship between Müller cells and photoreceptor health (Jablonski and Iannaccone, 2000). Although these studies demonstrate that compromised Müller cells can result in declining retinal health, they all had pitfalls in their approaches. First, the study on Xenopus retinas was done in culture, where specific aspects of photoreceptor integrity may have been affected already due to the inherent disruption caused by the culture preparation. Second, intraocular injection by itself, causes retinal damage and therefore makes the data difficult to interpret. In addition, in this approach and the others

involving injections of the drug, the concentration of the drug fluctuates dramatically based upon the time of injection. In my pharmacological model, I delivered the toxin to the water where it could be absorbed by the gills of adult zebrafish and through the skin

11

of zebrafish larvae. This delivery method was advantageous because the concentration of the drug and the exposure time could be controlled precisely. I monitored Müller stress both immunologically and histologically. Principally, I followed the expression of antibodies for glial fibrillary acidic protein (GFAP) and glutamine synthetase (GS), known Müller cell markers that have been shown to be upregulated in times of retinal and Müller specific stress (Uhlmann et al., 2003). In addition, I examined the histology of the adult and larval retinas for signs of Müller cell hypertrophy and photoreceptor stress or death since I suspect that the health of these two cell types is linked. MATERIAL AND METHODS Zebrafish maintenance Wild type and lazy eyes heterozygous zebrafish were maintained in accordance to Harvard University and National Institute of Health-approved protocols. The fish were kept on a 14/10-hour light-dark cycle in 28.5ºC fish water containing 2g of Instant Ocean salts per gallon of distilled water supplemented with vitamins. The lze mutant was obtained from a family that had been isolated in a mutagenesis screen in which N-ethylnitrosurea was used to induce DNA point mutations. The lze mutation is homozygous recessive (Kainz et al., 2003). Wild type larvae were obtained by mating several wild type fish together in a basket cross. Lze larvae were obtained by mating adult fish heterozygous for the lze mutation. The resultant progeny of this cross was comprised of wild type larvae, lze mutants, and lze heterozygotes in a ratio of 1:1:2. Mutant lze were identified on 5 dpf based on their failure to respond strongly in the OKR visual behavioral assay. Wild type larvae repeatedly move their eyes with a smooth pursuit motion followed by a saccade in

12

response to moving vertical black and white stripes. Mutant lze larvae respond with weak and infrequent eye movements or fail completely to move their eyes (Kainz et al., 2003). The adult zebrafish used in this experiment were genetically wild type and approximately 30 months old. Alpha-aminoadipic acid treatment of adult zebrafish and zebrafish larvae DL-α-aminoadipic acid was mixed with PBS (pH 7.4) and then adjusted to a pH of 7.3 using 1M NaOH. The resulting solution was mixed with fish water containing 2g of Instant Ocean salts per gallon of distilled water supplemented with vitamins to various concentrations. Treated and untreated adult fish were kept in 400mL volume of liquid. Fish used in experiments lasting longer than one day were fed once daily, and the water was changed shortly after each feeding. Adult fish were exposed to concentrations of 1mM, 10mM, 25mM, and 50mM α-AAA and for an incubation period from one to four days. It has been suggested that adult fish absorb pharmacological agents into the blood stream via the gills. Treated and untreated larvae were kept in an incubator and were not fed throughout the experiment because they still obtain nutrients from the yolk at this time in development. Approximately 15-35 larvae were treated with 80mL of the toxin in a petri dish starting at 3 dpf (after they had hatched from their chorion) and continuing for 48 hours until 5 dpf. The water was not changed after 3 dpf. Larvae were exposed to concentrations of 10mM, 25mM, 50mM, and 100mM α-AAA. Control larvae were reared in normal fish water. The larval exposure period of day 3 to day 5 was chosen for several reasons. On the third day of life, retinal functions first begins in the zebrafish larvae, and on 5 dpf,

13

visual function is readily observable and has been well-characterized (Schmitt and Dowling, 1999; Brokerhoff et al., 1995). Another advantage to using young zebrafish larvae for a pharmacological based experiment is that the larvae have not developed scales yet and their skin is known to absorb chemicals in the fish water. Finally, based on the work of Sugawara et al. on carp, exposure periods as short as 4-8 hours show observable effects on Müller cells. Thus, a 48 hour exposure period was deemed a long enough period of time. Constant light rearing Adult fish treated in constant light were first raised in constant darkness for one week and were then placed in a box lined with several fluorescent bulbs, with a fan used to minimize heat generated by the light for the duration of the treatment. The light level was approximately 15,000 lux, about 50 times brighter than average room light, and the temperature was maintained between 23-25ºC. Preparation of adult eyes for immunohistology Adult zebrafish were euthanized by over anesthetizing them in a 500mg/L Tricaine solution and then decapitated. Adult heads were immediately placed into cold fixative containing 4% paraformaldehyde (PFA) in 0.06M phosphate buffer, 3% sucrose (pH 7.4) with 0.15mM CaCl2. Forceps were then used to loosen connective tissue surrounding the eye, and the optic nerve was cut using a surgical scissor. Each eye was removed gently and the cornea was poked with an insect pin to increase the access of the fixative to the retina. Eyes were then transferred into fresh cold fixative and stored for 48 hours at 4ºC. Eyes were washed for 5 minutes 3X in 0.06M phosphate buffer, 3% sucrose (pH 7.4) with 0.15mM CaCl2 and then placed into 0.06M phosphate buffer, 15% sucrose (pH 7.4) with 0.15mM CaCl2 for one hour at 4ºC. Eyes were then transferred to

14

0.06M phosphate buffer, 30% sucrose (pH 7.4) with 0.15mM CaCl2 at 4ºC overnight. Eyes were removed and were mixed in a 1:1 solution of 30% sucrose and OCT and then transferred to a mixture completely comprised of OCT. Eyes were embedded in OCT and then frozen using dry ice. Eyes were sectioned at 10µm in thickness and placed onto gelatin-coated slides. Immunohistological analysis Sections were removed from the freezer and air-dried for 2 hours. Slides were then washed in PBS (pH 7.4), 5 minutes 3X and blocked in 5% normal goat serum (NGS) in PBS with 0.3% Triton X-100 for 20 minutes at room temperature. Sections were then incubated overnight with primary antibody diluted in blocking solution at 4ºC for 12-18 hours. Sections were then brought to room temperature and washed in PBS 15 minutes 4X. Secondary antibody diluted in blocking solution including the Hoechst nuclear dye was applied, and slides were placed in 37ºC for 30 minutes. Sections were then washed in PBS for 10 minutes 3X. Slides were mounted with Vectashield mounting medium and stored in the freezer. Slides were analyzed using confocal microscopy and images were captured digitally by Pamela Kainz. The following list includes the antibodies used, the working dilution, and the cell types which possess the respective antigens: GFAP, 1:200, Müller glial cells (primarily the cell endfeet); GS, 1:500, Müller cells. antibodies used were AlexaFluor-488 or -555 conjugates. Histology Adult eyes were obtained using the method described above; however, the fixative for this analysis contained 2.5% glutaraldehyde, 1% PFA in 0.06M PBS (pH 7.4), 3% sucrose, 0.15mM CaCl2. Isolated eyes were immediately transferred into fresh cold fixative for 30 minutes. After this period, the eyes were placed in fresh fixative for Secondary

15

2 hours. Eyes were then rinsed for 15 minutes 2X in 0.06M PBS (pH 7.4), 3% sucrose, 0.15mM CaCl2 and were then dehydrated in a graded series of ethanol in 0.06M PBS (pH 7.4), 3% sucrose, 0.15mM CaCl2 and infiltrated with propylene oxide and resin. Transverse sections of 1µm thick were collected and heat-mounted onto a gelatin-coated glass slide and stained with 1% Methylene Blue, 1% Azure in 1% borax. Slides were cover-slipped with DPX. Larvae preparation for histological retinal analysis Five-day old larvae were anesthetized in ice-cold fish water and fixed with 2.5% glutaraldehyde, 1% PFA in 0.06M phosphate buffer (PBS) (pH 7.4), 3% sucrose, 0.15mM CaCl2 for 1.5 hours at 4ºC. Larvae were then rinsed in 0.06M PBS (pH 7.4), 3% sucrose, 0.15mM CaCl2 2X five minutes. Larvae were dehydrated in a graded series of ethanol in 0.06M PBS (pH 7.4), 3% sucrose, 0.15mM CaCl2 and infiltrated with propylene oxide and resin. Transverse sections of 1µm thick were collected and heatmounted onto a gelatin-coated glass slide and stained with 1% Methylene Blue, 1% Azure in 1% borax. Slides were cover-slipped with DPX. Visual behavioral analysis The optokinetic reflex assay (OKR) was used to test the visual sensitivity of larvae at 5 dpf. For testing, 4-5 subjects were transferred into small petri dishes

containing 5% methyl cellulose and placed within a drum lined with vertical black and white stripes, 1cm in width. The drum was illuminated with a tungsten light source, 9.74 * 10-2 W/cm2 and a 2 minute trial was conducted, during which the direction of rotation of the drum was switched 3-4 times. The criterion for a positive response was that each larva either demonstrated a full smooth pursuit-saccade cycle or eye tracking movements

16

in both the clockwise and counterclockwise directions when the drum was rotated accordingly. RESULTS Müller glial cell expression of GFAP and GS is altered by α-aminoadipic acid To analyze the effects of α-AAA on the adult zebrafish Müller cells, I exposed several fish to different concentrations of the gliotoxin: 1mM, 10mM, and 25mM, 50mM for time periods of 24 and 48 hours. A concentration of 50mM compromised severely the overall health of the zebrafish and resulted in violent spasms. These fish were thus sacrificed due to the fact that they appeared distressed and unhealthy. Fish exposed to concentrations less than or equal to 25mM showed no behavioral abnormalities when treated. These adults were fixed for retinal immunological and histological analysis. Retinal sections were obtained, labeled with GS and GFAP, and imaged using confocal microscopy and digital photography. Fish treated with the lowest concentration (1mM) had retinas that appeared highly similar to wild type. Adults treated with the highest, non-lethal concentrations (10mM and 25mM) also had retinas that appeared to be well intact. To examine specifically Müller cells, antibodies to two known cell specific proteins, GS and GFAP, were utilized. Exposure to 10mM α-AAA for 24 hours appeared to affect modestly the expression of GS and GFAP. The retinal images shown in figure 2 were obtained from a specified region in the dorsal part of the retina from treated and untreated animals. The expression of GS in untreated retinas was evenly distributed throughout the Müller cell bodies and therefore, the retina (Fig. 2a), while the expression of GS in treated retinas had a spoke-like pattern (reminiscent of the Müller cell bodies) suggesting that the either the expression of this protein increased in response to the gliotoxin or the size of the Müller cell processes

17

increased (Fig. 2b). In addition, there appeared to be less GS labeling in the inner plexiform synaptic layer in the treated retina compared to controls.

18

a

b

c
PRL INL

wt control

d

wt 10mM

Endfeet

wt control

wt 10mM

Fig. 2. Zebrafish treated with 10mM α-aminoadipic acid for 24 hours exhibited differences in GS (red) and GFAP (green) expression in comparison to controls. The photoreceptor layer (PRL) and inner-nuclear layer (INL) are labeled. a: GS expression in untreated retinas was evenly distributed in each layer of the retina. b: GS expression in treated retinas appeared denser or spoke-like. c: GFAP expression in untreated retinas was confined to the Müller cell endfeet. d: GFAP expression in treated retinas was somewhat elevated in Müller cell bodies and in radial processes reaching the outer-plexiform layer.

19

More substantial differences were found in the expression of GFAP.

In the

controls, immunoreactivity was mostly confined to the Müller cell endfeet (Fig. 2c). However, in the treated retinas, the GFAP immunoreactivity was present throughout the Müller cell bodies in radial processes reaching the outer plexiform layer (Fig. 2d) suggesting that the expression of GFAP increased in response to exposure to the gliotoxin, consistent with what would be predicted. The pattern of Hoechst nuclear dye labeling shown in blue in figure 2c and 2d illustrates how well-preserved the 10mM and 25mM α-AAA treated retinas were compared to untreated controls. No gaps were seen in the nuclear layers indicating that α-AAA did not result in massive cell death, and three distinct nuclear layers were observed implying that retinal organization was not disrupted. In case the mild effect on Müller cells disrupted their ability to support photoreceptors, I examined closely the integrity of the photoreceptors cells in the treated animals. No evidence of photoreceptor cell disruption was observed. Increased dosage and a longer incubation period yielded similar results The 24 hour exposure of 10mM and 25mM Müller appeared to have no major detrimental effect on retinal neurons but did affect mildly Müller glia. Since the goal was to determine whether α-AAA could compromise significantly the Müller glia while having no effect on retinal neurons, I decided to push the system. Since higher

concentrations were lethal, I chose to increase the length of exposure. For this, I exposed adult zebrafish to a concentration of 25mM over a period of 48 hours and found that the severity of the effect had not increased. Again, the expression of GS and GFAP were mildly elevated (data not shown). Greatly increasing the exposure time may have

strengthened the gliotoxin affect; however, we did not feel this was a very practical

20

approach given the large amount of drug we would need to utilize in order to be able to change the 400mL solution of α-AAA daily. Instead I set out to determine the effect αAAA had on larval zebrafish. When combined with constant light, α-aminoadipic acid caused a similar phenotype Knowing that constant and intense light exposure exacerbated the Müller cell phenotype in the larval genetic model of Müller cell disruption, the lze mutant, I tested the impact of combining pharmacological stress with light toxicity in the adult. Although teleost retinas have shown more resistance to light damage than rodents (where the retina is largely rod dominated), light toxicity has been studied in albino adult zebrafish where it has caused rod and cone cell apoptosis (Vihtelic and Hyde, 2000). We chose to avoid the use of albino zebrafish due to their low viability as larvae and extreme susceptibility to light toxicity. Instead, I tested α-aminoadipic acid treated and untreated adult wild type zebrafish in a light regiment of 15,000 lux, nearly twice the intensity necessary to observe photoreceptor cell death in albinos. My initial result suggested that the combination of the toxin (25mM α-AAA) and constant light caused photoreceptor cell death in the adult zebrafish (Fig. 3). This was apparent from observing a Hoechst nuclear dye that indicated that the dorsal portion of the outer-nuclear layer was only one nucleus thick in α-AAA light treated fish but 2-3 nuclei thick in controls (unexposed to light or the drug). When repeated several times, the experiment provided new data to suggest that my light exposure regiment alone could sometimes cause photoreceptor disruption and death (Fig. 4).

21

a

ONL

control

b

ONL

light

c
ONL

light/AAA

light/AAA

Fig. 3. Adult zebrafish treated with 25mM α-aminoadipic acid for three days under constant lighting conditions had photoreceptor cell degeneration. a: The outer nuclear layer (ONL) in untreated retinas appears healthy and is approximately three nuclei thick when viewed under a Hoechst stain. b: The ONL in the retina exposed to constant light appears overall healthy to the control, with the layer spanned by a thickness of 2-3 nuclei. c: The ONL in the retina exposed to constant light and 25mM α-aminoadipic acid is approximately one nucleus thick indicating that many photoreceptors were lost. Pyknotic nuclei are indicated by arrows.

22

a
OS

b

ONL INL
control GCL Fig. 4. Adult zebrafish retinas exposed to constant light exhibit a similar phenotype to those exposed to constant light and 25mM α-aminoadipic acid for three days. a: The untreated retina exhibits healthy outer segments (OS). b: The retina exposed to constant lighting expresses unhealthy and missing outer segments (arrow) and missing photoreceptor nuclei. c: The retina exposed to constant lighting and treated with 25mM α-aminoadipic acid was similar to the retina treated with light alone, having photoreceptor cell loss and outer segment disruption.

light

c

light/AAA

23

Larval health was compromised at concentrations similar to the adult Wild type larvae were treated with various concentrations of the toxin by adding it to their water in petri dishes from 3 dpf to 5 dpf. Survival data is located in Table 1. I found that 81% of larvae survived a two day exposure at concentrations of 10mM and 61% survived a two day exposure at a concentration of 25mM. Survival percentage was 0% at 50mM, although 1/15 treated fish survived a concentration of 100mM (7%). This non-lethal dose pattern was thus similar to adults.

24

α-Aminoadipic Acid Concentration (48hr Exposure) 0mM (control) 10mM 25mM 50mM 100mM

Wild Type Larvae Treated 45 15 36 15 15

Wild Type Larvae Survived 45 14 22 0 1

Survival Percentage 100% 87% 61% 0% 7%

Table 1. Wild type larvae have similar dosage dependent survival as adult zebrafish. Larvae were able to tolerate an exposure of 48 hours at concentrations up to 25mM α-aminoadipic acid before they displayed scant survival. Similarly, the health of adults was severely compromised at concentrations of 50mM or higher.

25

General characteristics of 5 dpf α-aminoadipic acid treated larvae All of the treated larvae at 25mM appeared similar to untreated animals with some exceptions. Treated larvae displayed a constant jaw twitching movement and lacked an inflated swim bladder. Furthermore, they demonstrated little spontaneous activity and had only a moderate response to touch. The yolk of treated larvae was also observed to be partitioned. Heart rate was found to be the same between treated and untreated wild type larvae. The jaw twitching, lethargy, and under-inflated swim bladder were all characteristics in common with the lze Müller glia mutant.
Wild type larvae treated with α-aminoadipic acid demonstrate an OKR similar to controls

To test the vision of wild type larvae treated with the toxin, I utilized the optokinetic reflex assay. I found that larvae treated at a concentration of 25mM

demonstrated an almost identical response to the assay as untreated wild type controls (when the light level in the barrel was not attenuated). Both untreated and treated larvae demonstrated a strong saccade and steady tracking of the black and white stripes. Alpha-aminoadipic acid causes disruptions in the ganglion cell layer, inner nuclear layer, horizontal cell layer, and marginal zone in zebrafish larvae While the visual behavior of treated and untreated wild type larvae was similar, histological findings showed that some retinal neurons were disrupted in larvae exposed to 10mM α-AAA. Pyknotic nuclei, that appear darkened with a halo of empty space surrounding them, were identified within the ganglion cell layer (GCL), indicating that some of these neurons were degenerating (Fig. 5b). In addition, the toxin compromised the marginal zone as indicated by gaps seen in this region where proliferative stem cells are present in controls. At concentrations of 10mM, the toxin also appeared to have an effect on the horizontal cell layer, causing large gaps, indicating the presence of fewer horizontal cells compared to untreated larvae (Fig. 5b).

26

Increasing the dosage of the drug proved to increase inner nuclear layer disruption. At concentrations of 25mM, the toxin caused cell loss within GCL but also caused death in INL (Fig. 5c). Large circular gaps were present in these treated retinas, reminiscent of retinal locations where cell death had just occurred. Gaps in the brain were observed in animals treated with 25mM α-aminoadipic-acid (Fig. 5d).

27

a

b

wt control

wt 10mM

c

d

wt 25mM

wt 25mM

Fig. 5. Wild type zebrafish larvae retinas treated with α-aminoadipic acid at concentrations of 10mM and 25mM and for an incubation period of 48 hours expressed disruptions in the ganglion cell layer, inner nuclear layer, horizontal cell layer, and marginal zone. a: Control retinas showed no signs of cell death. b: Wild type retinas treated with concentrations as low as 10mM expressed pyknotic cells in the ganglion cell layer (orange arrow), missing horizontal cells (red arrow), and deficiencies in the marginal zone (green arrow). c: Increasing the dosage to 25mM resulted in more inner nuclear layer deficiencies and large gaps. d: Concentrations of 25mM also caused larges gaps of missing cells in the brain.

28

Alpha-aminoadipic acid causes cell migration out of the retina through the optic nerve in some wild type retinas In two of the approximately 20 wild type treated retinas that were sectioned (10%), I observed cell migration out of the retina through the optic nerve. I could not ascertain definitively the direction of the migration, although it appears as if cells were funneling out of the retina towards the brain (Fig. 6a). Cells were elongated, which is indicative of migrating cells, and some seemed to be differentiated. A cell associated with an outer segment (a presumed photoreceptor) can be seen within the migratory stream of cells (Fig. 6b). This effect was seen in wild type treated larvae at a

concentration of 10mM and at a concentration of 100mM but never in wild type untreated larvae (data not shown).

29

a

b

wt 10mM

wt 10mM wt 10mM

Fig. 6. In 10% of wild type treated retinas, α-aminoadipic acid caused retinal cell migration out of the retina through the optic nerve. a: Differentiated and elongated cells appeared as though they were migrating out of the retina towards the brain. b: A close up of the optic nerve region. The circle surrounds a cell with an outer segment, indicative of a photoreceptor cell.

30

Alpha-aminoadipic acid does not seem to worsen severely the lze phenotype Since the zebrafish is a genetic model organism, mutants, such as lazy eyes, could be used to investigate the combined effects of genetic and pharmacological manipulation of glia on the neural retina. Knowing that lze mutants have compromised Müller glial

cells, I was curious whether α-AAA would increase the severity of the lze Müller cell phenotype. Thus, the same experiments were carried out on lze larvae. Larvae from the lze clutch that were treated with 10mM α-AAA exhibited subtle histological abnormalities within the spectrum of what was observed in treated wild type larvae and in lze untreated controls (Fig. 7).

31

a

b

lze 10mM

lze control

Fig. 7. Lze larvae treated with α-aminoadipic acid exhibit a phenotype similar to lze untreated controls. a: Lze larvae treated with 10mM α-aminoadipic acid appear to have some inner nuclear layer deficiencies (highlighted by the arrow) but do not appear severely different from lze controls. b: A lze control is pictured.

32

DISCUSSION The goal of the work outlined in this chapter was to explore the gliotoxic effect of α-AAA on the retina using the zebrafish animal model. Previous studies using α-AAA have suggested that retinal glial cells were specifically affected; however, none have been on the zebrafish nor have any been conducted without the use of invasive methods of drug delivery such as intraocular injection. By delivering the drug via the fish water, the concentration and thus dose of the drug at any one time is constant, unlike the unavoidable fluctuation of drug dose with intraocular injection or subcutaneous injection done in all previous in vivo studies. . The investigation of an α-AAA-mediated pharmacological model of Müller cell disruption in the adult zebrafish yielded tepid effects, considerably weaker than those observed in other animal models treated with the toxin. While other studies have

observed Müller cell death, swelling, or hypertrophy, as well as photoreceptor death, I did not. My sole finding was a increase in the expression of GFAP. Although the increase was subtle, upregulation of GFAP is a classic indication of Müller cell stress indicating that the drug was having the desired affect. Furthermore, labeling with the GS antibody revealed that Müller processes appeared thicker in response to treatment suggesting that some degree of hypertrophy may have occurred. Thus, I observed signs of Müller cell stress although these indications were more subtle than predicted. Unfortunately the effects could not be increased in severity by increasing concentration or length of exposure, before lethality became a problem. In the wild type larval retina, I observed evidence of modest neuronal degeneration. In addition, I noticed cell migration of retinal cells out the optic nerve in a

33

few of the retinas. Both of these observations involved retinal neurons and indicated a lack of glial-specificity. It is known that when α-AAA is used at too high of a

concentration, the affects can be neurotoxic (Sugawara et al., 1990) perhaps indicating that my dosage was too high. This is possible but unfortunately neither neurons nor glia were affected by lower doses. In summary, the effect of the gliotoxin varied substantially from fish to fish and was highly dependent on age, as the effects were very subtle in the adult retina and detrimental to retinal neurons in the larval zebrafish. Although more variables in

treatment strategy could be explored, the neurotoxic effects observed in the larvae discouraged me from further pursuing this direction for the present time. As a result, I focused on a different model of Müller cell stress mediated by a genetic mutant.

34

Chapter 2: The Genetic Mode INTRODUCTION The impetus for moving in this direction were the findings of Rattner and Nathans (2006) in their study on the genes related to photoreceptor stress induced either by detachment, genetic photoreceptor mutations, or light toxicity. Using microarray

technology, RNA blots, and in situ hybridization, they quantified the genomic responses to both light damage and inherited photoreceptor degeneration and found that these responses involve a relatively small number of overlapping genes (Rattner and Nathans, 2005). In their research, they discovered that the endothelin pathway is highly linked to photoreceptor and Müller cell stress. Endothelins are vasoactive peptides with various functions throughout vertebrates including cardiovascular systems, pigmentation, and ocular homeostasis (Prasanna et al., 2003). There are three isoforms of the peptide: endothelin 1, 2, and 3. Endothelin receptors come in two subtypes, endothelin receptors A and B, both G-protein coupled receptors (Sakurai et al., 1992). In the retina, endothelin receptor A (ET-A) is mainly localized to the choroid and blood vessels, whereas endothelin receptor B (ET-B) has been found mainly in the neural and glial components of the retina (Maccumber and D’Anna, 1994) although the precise roles endothelins play in the retina are unknown. Using the mouse model, Rattner and Nathans found that endothelin 2 is expressed in photoreceptor cells and highly induced in all of their models of photoreceptor stress. In addition, they found that ET-B localized to Müller cells and its expression was upregulated >10 fold following phototoxic conditions. These data led the authors to the hypothesis that the endothelin pathway plays a critical role in the Müller cell response to

35

stressed or dying photoreceptor cells and may be involved in the neuroprotective support function Müller glia provide for photoreceptors. Other studies have shown that the endothelin pathway is susceptible to pharmacological manipulation. When administered to albino mice under phototoxic

conditions, Tezosentan, a mixed ET-A and ET-B antagonist, lowered the amount of GFAP expression and also resulted in a lower amount of apoptotic cells throughout the retina, as judged by a CC3 cell death assay. These findings imply that inhibition of endothelinergic receptors could play a role in the preservation of vision by sparing photoreceptors (Torbidoni et al., 2005). The investigators hypothesized that the

endothelin pathway triggers the Müller cells to upregulate GFAP expression resulting in a scarring effect and that the prevention of the Müller cell processes could promote neuronal survival and preserve vision. Strong associations between the endothelin pathway, Müller cells, and photoreceptor support mechanisms inspired my investigation of the ET-B knockout zebrafish called rose. Using a genetic mutant allowed me to circumvent some of the drawbacks with a pharmacological model, specifically the variability and fluctuation in drug concentration caused by metabolism. Rose was initially discovered through its abnormal body pigmentation and the initial study concluded that the only defect caused by the absence of the ET-B was the lack of the production of a subset of the adult melanocytes and iridiphores. This phenotype resulted in adults appearing reddish

compared to wild type (Fig. 8) (Parichy et al, 2000). Later studies provided data to support that the ET-B gene is actually expressed in the zebrafish larval retina (Lister et al., 2006).

36

Fig. 8. Wild type and rose mutant adult zebrafish (Source: Parichy et al., 2000). a: Wild type adults demonstrate normal coloring. b: Rose mutants fail to develop the normal amount of melanocytes and iridophores during pigment pattern metamorphosis, accounting for their reddish appearance.

37

In the second chapter of this study, I characterized the homozygous ET-B mutant rose zebrafish larval retina. Research on the links between phototoxicity and

photoreceptor stress in addition to lze’s increased Müller specific susceptibility to light provided the impetus for exposing my genetic model to constant light. Rattner and Nathans’ findings, that the endothelin pathway was involved in the response to phototoxic condition, gave me further reasons to test how a retina missing ET-B would react to constant lighting. I employed histological, behavioral, and quantitative

measurements to characterize the degree to which the health or function of the retina was compromised. The OKR was used to measure the visual threshold of light-treated rose mutants and control retinas, and these results were compared to wild type larvae under the same two conditions: constant light (LL) and a normal light dark cycle (LD). Retinal histology of rose was performed and cell counts on the inner and outer nuclear layers were used to assess the presence or absence of cell death in the regions of the retina containing Müller cells and photoreceptor cells. These experiments continue with the theme of this thesis: exploring potential models where Müller cells are compromised and observing the effects of this stress on photoreceptors and on vision in the zebrafish. MATERIAL AND METHODS Zebrafish maintenance Wild type and rose homozygous recessive zebrafish were maintained as described in Chapter 1. Constant light rearing Adult larvae were maintained in a standard 14/10-hour light-dark cycle until 2 dpf when they were transferred to constant dark conditions because pretreatment with constant darkness intensifies the effect of the light treatment to follow. Dark adaptation

38

from day 0 until day 2 was not necessary since larvae lack developed photoreceptors at this point and opsin does not appear until 48hrs post fertilization (Schmitt et al., 1999). At 4 dpf, larvae treated with constant light were transferred to a box lined with several fluorescent bulbs and a fan used to minimize heat generated by the light. The light level was approximately 15,000 lux, about 50 times brighter than average room light. Controls were kept in standard light dark conditions from 2-6 dpf. Larvae were removed at 6 dpf for visual testing and fixation. Visual Threshold Assay All OKR assays were run on 6 dpf larvae between the hours of 1 PM and 5 PM in a completely darkened room. I used a dim, red head-lamp for visibility when needed. For testing, 4-5 subjects were transferred into small petri dishes containing 5% methyl cellulose and placed within a drum lined with vertical black and white stripes, 1cm in width. The drum was illuminated with a tungsten light source, 9.74 * 10-2 W/cm2, attenuated by 6.5 log units, and the drum was rotated at 10 rpm. A 2 minute trial was conducted, during which the direction of rotation of the drum was switched 3-4 times. The lowest light level that evoked an OKR response for each larva was determined. The criterion for a positive response was that each larva either demonstrated a full smooth pursuit-saccade cycle or eye tracking movements in both the clockwise and counterclockwise directions when the drum was rotated accordingly. Fish that failed to demonstrate a positive response were retested using 0.5 log unit brighter illumination. Histology Six-day old larvae were fixed using the same protocol as Chapter 1. Cell Counts Cell counts were completed in rose and wild type light treated and untreated retinas in the inner and outer nuclear layers. Inner nuclear layer counts did not register

39

horizontal cells or nuclei in the marginal zone (Fig. 9). The marginal zone contains proliferative stem cells, while the inner nuclear layer contains Müller, bipolar, and amacrine nuclei.

40

Fig. 9. The region of the retina in which inner nuclear layer cell counts were performed is enclosed in red. Inner nuclear layer nuclei were counted with careful attention not to include nuclei in the marginal zone (top arrow) or horizontal cell nuclei (bottom arrow) in the count.

41

RESULTS Rose larvae have normal retinal morphology and visual function Morphologically, rose and wild type larvae retinas were nearly indistinguishable when the larvae were raised in a normal light dark cycle (Fig. 10a, 10c). Both had healthy photoreceptors, a continuous span of nuclei in the outer and inner nuclear layers, and healthy, dense, and organized rod outer segments. Furthermore, without dark

adaptation and light attenuation, rose larvae responded equally robustly to wild type in response to the OKR.

42

a

b

wt LD

wt LL

c

d

rose LD

rose LL

Fig. 10. Wild type and rose larvae are strikingly similar when raised in a normal light dark cycle or with constant light. a: Wild type larvae raised in normal light dark cycle. Encircled is the rod dense ventral portion of the retina. Large rod outer segments span the region containing melanin from the pigmented epithelial cells. b: Wild type larvae treated with light. c: Rose mutant larvae kept in normal lighting conditions appear nearly identical to wild type larvae treated in the same conditions. d: Rose mutant larvae treated in constant light show significant loss of rod outer segment material in the ventral retina.

43

When raised in constant light, rose showed higher susceptibility to rod outer segment damage Histological analysis of retinas from rose larvae exposed to constant light revealed that there were limited but visible differences between rose and wild type retinas (Fig. 10b, 10d). Both retinas appeared grossly normal and neither displayed indications that massive cell death had occurred, in the form of pyknotic nuclei or gaps in the nuclear layers. There were also no obvious signs of disruption within Müller cells. Their cell bodies did not appear hypertrophied. There was, however, one clear difference between retinas from constant light reared wild type and rose larvae: the integrity of their rod outer segments. Rose light treated retinas showed several degrees of rod outer segment health. Some appeared slightly swollen in comparison to light dark treated rose larvae (Fig. 11). Most, however, were disorganized, and many retinas were missing rod outer segments. Furthermore, vacuoles within the RPE were very common. In contrast, the rod outer segments of wild type light treated retinas almost always appeared equally healthy to their light-dark counterparts. In some wild type retinas, there were slightly swollen or disorganized rod outer segments; however, in only one of sixteen retinas were there actually fewer rod outer segments.

44

a

rose LL

b

rose LL

c

rose LL
Fig. 11. Rose mutant larvae treated with constant light showed varying degrees of rod outer segment health. a: Rod outer segments appear healthy and organized yet a few small vacuoles appear in the RPE (red arrow). b: Rod outer segments appear substantially disorganized, shortened, and dysmorphic (orange arrow), and the RPE contains vacuoles. c: Significant loss of rod outer segments from the ventral retina has taken place.

45

When treated with constant light, rose larvae had a 10% reduction in inner nuclear layer nuclei, while the amount of inner nuclear layer nuclei in wild type remained constant

The lack of gaps in the nuclear layers and the absence of pyknotic nuclei suggest that cell death in the rose retina did not occur, at least not on day 6 when the larvae were sacrificed. To ascertain whether any cell death had occurred prior to day 6, the number of nuclei were counted in the nuclear layer that contains the photoreceptor nuclei and the nuclear layer that contains Müller cells. Average numbers were compared among lightdark (LD) and light-light (LL) treated rose and wild type animals. Neither rose nor wild type showed a decrease in the number of outer nuclear layer (ONL) nuclei upon the introduction of constant light; however, wild type had on average had more ONL nuclei than rose in both treatment groups. For instance, rose light treated larvae had an average of 156 ONL nuclei, while wild type had an average of 183, 20% higher with a p-value of 0.00022 (Fig. 12, Table 2).

46

200 190

Rose and Wild Type Outer Nuclear Layer

**

Average Number of Nuclei in ONL

180 170 160 150 140 130 120 110 100

rose LD

rose LL

wt LD

wt LL

Fig. 12. Neither rose nor wild type larvae lost photoreceptor nuclei upon treatment with constant light. Although the average number of nuclei in the ONL in rose LD and LL was different than wild type, no difference was observed between rose LD and LL. Error bars represent 95% confidence intervals, and starred bars connect treatment groups for which there was a statistically significant difference in outer nuclear layer.

47

rose LD INL ONL Average Number of Nuclei Standard Deviation Number Observations 95% Confidence Interval 385 50 20 [363, 407] 161 24 8 [144, 178]

rose LL INL ONL 349 36 38 [338, 361] 156 12 17 [151, 162]

wt LD INL ONL 339 34 14 [321, 357] 179 19 17 [170, 188]

wt LL INL ONL 358 31 13 [342, 375] 183 18 13 [174, 193]

Table 2. Average inner and outer nuclear layer nuclei for rose and wild type larvae across the two treatment groups.

48

Untreated rose larvae had an average of 385 nuclei in their INL in contrast to light treated rose which had an average of 349 nuclei, roughly a 10% decrease. This

difference was found to be significant to a p-value of 0.0092 (Fig. 13, Table 2). No such effect was observed in wild type light treated larvae. Another finding was that untreated rose larvae had nearly 14% more INL nuclei than their untreated wild type counterparts, which had an average of 339 nuclei. This difference was significant to a p-value of 0.0037 (Fig. 13, Table 2).

49

410

Average Number of Nuclei in the INL

Rose and Wild Type Inner Nuclear Layer ** **

390 370 350 330 310 290 270 250

rose LD

rose LL

wt LD

wt LL

Fig. 13. When treated with constant light, rose larvae lose 10% of their inner nuclear layer nuclei. Error bars represent 95% confidence intervals, and starred bars connect treatment groups for which there was a statistically significant difference in inner nuclear layer nuclei. Rose untreated larvae had an average of 385 inner nuclear layer nuclei, while rose light treated larvae had an average of 349 inner nuclear layer nuclei (p-value 0.0092). There was also a statistically significant difference between rose untreated larvae and wild type untreated larvae which had an average of 339 nuclei (p-value 0.0037).

50

Untreated rose larvae exhibit a lower visual threshold than rose larvae treated with constant light; wild type shows no loss in visual function after light treatment While I had observed that untreated rose and wild type exhibited equally robust responses to the OKR without dark adaptation, I realized that this was only a qualitative observation. To obtain a quantitative assessment of visual behavior, I chose to measure visual thresholds. Thresholds were found by determining the lowest light level that evoked an OKR response for each larva. Larvae that failed to demonstrate a positive response were retested using 0.5 log unit brighter illumination. In light-dark conditions, rose and wild type exhibited an average threshold of -6.0 log units of light attenuation similar to wild type (Fig. 14, Table 3). When reared in lightlight, rose larvae had a significantly higher threshold than rose larvae raised in a normal light dark cycle and than wild type larvae treated in constant light. Rose light treated larvae had an average threshold of -4.5, while untreated rose larvae had an average threshold of -6.0 (Fig. 14, Table 3). This difference was highly significant to a p-value of 2.8 * 10-7. In addition, this drop in visual sensitivity was not seen in wild type suggesting that the light-light treatment used had no measurable effect on fish having an intact ET-B gene. Light treated wild type larvae had an average threshold of -5.8. The difference between rose and wild type light treated larvae was significant to a p-value of 2.7 * 10-6.

51

Rose and Wild Type Visual Threshold
Average Threshold (Units of Negative Log Attenuation)
6.5

**

**

6.0

5.5

5.0

4.5

4.0

3.5

3.0

rose LD

rose LL

wt LD

wt LL

Fig. 14. When treated with constant light, the visual threshold of rose larvae increases. In contrast, constant light does not affect the visual sensitivity of wild type larvae. Error bars represent 95% confidence intervals, and starred bars connect treatment groups for which there was a statistically significant difference in visual threshold. Rose light treated larvae had an average threshold of -4.5, while untreated rose larvae had an average threshold of -6.0 (p-value of 2.8 * 10-7). There was also a statistically significant difference between rose light treated larvae and wild type light treated larvae, which had an average threshold of -5.8 (p-value of 2.7 * 10-6).

52

rose LD Average Threshold Standard Deviation Number Observations 95% Confidence Interval Number Thresholds -5 or Lower % Thresholds -5 or Lower Number Thresholds Higher Than -5 % Thresholds Higher Than -5 -6 0.5 29

rose LL -4.5 1.5 42

wt LD -6 0.5 30

wt LL -5.8 0.5 33

[-5.8, -6.2] [-4.1, -5.0] [-5.8, -6.2] [-5.7, -6.0] 28 19 30 31 97% 1 3% 45% 23 55% 100% 0 0% 94% 2 6%

Table 3. Average visual threshold for rose and wild type larvae across the two treatment groups.

53

Rose light treated larvae in general varied substantially in their thresholds. Some had thresholds equivalent to wild type, while others had significant drops in their visual sensitivity. To illustrate this phenomenon, I calculated the percent of larvae in each group that had a threshold of -5 or lower. I used -5 as cutoff because this is the highest threshold any light-dark wild type larva ever demonstrated. While rose untreated larvae, wild type untreated larvae, and wild type light treated larvae had thresholds of -5 or lower in 97%, 100%, and 94% of the data points respectively, rose light treated larvae only had a threshold of -5 or lower 45% of the time (Fig. 15, Table 3). The difference between rose light treated and rose untreated larvae in this case was significant to a p-value of 1.4 * 10-7, a highly robust result.

54

Rose and Wild Type Visual Threshold Binomial Data
100.0%

80.0% Percent of Thresholds -5 or Lower

60.0%

40.0%

20.0%

0.0%

rose LD

rose LL

wt LD

wt LL

Fig. 15. The percentage of larvae in each treatment group that had a threshold of -5 or lower is depicted. Error bars represent 95% confidence intervals. While 97% of untreated rose larvae had a threshold of -5 or lower, only 45% of the light treated rose larvae had a threshold of -5 or lower (p-value 1.4* 10-7).

55

Inner nuclear layer count and visual threshold are not correlated in rose light treated larvae

Curious whether there was relationship between the loss of cells in the INL and high visual thresholds, I tested rose light treated larvae in the OKR and separated them into two groups, those with a threshold of -5 or lower and those with a threshold higher than -5. I then proceeded to calculate the average number of INL nuclei in each group. There did not appear to be a correlation between INL count and visual threshold as both groups expressed a nearly identical average number of nuclei (Fig. 16, Table 4).

56

Inner Nuclear Layer Visual Threshold Correlation
390

Average Number of Nuclei in INL

370

350

330

310

290

270

Rose LL Low Threshold

Rose LL High Threshold

250

Fig. 16. When rose light treated larvae were separated into two groups, those with visual thresholds of -5 or lower (Low Visual Threshold) and those with visual thresholds higher than -5 (High Visual Thresholds), there was no significant difference in the average number of inner nuclear layer nuclei across the two groups. Error bars represent 95% confidence intervals.

57

rose LL Low Visual Threshold Average Number of Nuclei in INL Standard Deviation Number Observations 95% Confidence Interval 356 35 7 [330, 382]

rose LL High Visual Threshold 358 31 15 [342, 374]

Table 4. Average number of nuclei in the inner nuclear layer across rose light treated larvae with visual thresholds of -5 or lower (Low Visual Threshold) or with visual thresholds higher than -5 (High Visual Threshold).

58

DISCUSSION In my evaluation of rose, I discovered several interesting characteristics of the mutant. The first was its striking similarity to wild type. ET-B thus is apparently not necessary for early retinal development. However, the absence of ET-B did result in decreased photoreceptor resistance to light damage. Rose had compromised rod outer segments and a higher visual threshold than wild type. The final observation about the rose mutant was its INL vulnerability to constant light. While wild type showed no loss of INL nuclei in response to light treatment, the average number of nuclei decreased by 10% in rose. Thus, the absence of ET-B does indeed compromise the retina’s resilience. At this point however, I have not yet proven that this phenotype is related specifically to Müller cells. The nuclei that reside in the inner nuclear layer belong to bipolar, horizontal, amacrine, and Müller cells. As a final experiment, I applied a Müller cell-specific antibody, carbonic hydrase II, to retinal sections from rose treated and untreated retinas to ascertain whether Müller cells were likely the cell type that was missing or partially missing. While this experiment was attempted twice, I was unable to obtain any labeling of Müller cells in rose retinas using this antibody, for reasons I do not understand. The most logical explanation for a higher visual threshold is fewer photoreceptors. Most research has implicated cones as the predominant contributor to visual sensitivity at 6 dpf since they greatly outnumber rods at this stage in development (Bilotta et al., 2001). However, if the visual sensitivity problems had been cone related, I would have observed a difference in ONL nuclei, which I did not. The only differences I observed were manifested in rods, specifically in the health of their outer segments. This

59

lead us to believe that although there were no pyknotic rod nuclei in the ONL, the disorganization and scarcity of the rod outer segments in rose light treated retinas could imply that the rods were significantly compromised and that rods may play a significant role in visual function at this stage in development. Therefore, we believe that higher visual threshold was likely a result of compromised rod outer segment health. When assessing the reasons behind the rose mutant’s vulnerability to light toxicity, we could not ignore that fact that rose has a deficiency in cells which are pigmented: melanocytes and iridiphores. Thus, perhaps rose is more sensitive to light damage merely because it is missing melanin in the retinal pigmented epithelial cells, much like the albino model. I did not observe evidence in support of this possibility as the melanin density in the retinal pigment epithelium in rose retinas did not appear different from that which was observed in wild type. A second hypothesis which we think holds more promise and that is consistent with the literature is that ET-B is involved in a neuroprotective program, perhaps mediated by Müller cells as implied by the results from mouse models. To follow this idea, I would first need to confirm that Müller cells express ET-B in the zebrafish retina. Then, I would want to explore how and whether this expression level changes as a result of intense light exposure. If the expression of ET-B were specific to Müller cells and increased in response to light exposure, more experiments would be needed to explore the timing of rod photoreceptor cell disruption and the loss of INL nuclei to gain a better understanding of how endothelins and Müller cells are involved in neuroprotection within the retina.

60

FUTURE EXPERIMENTS As mentioned above, in situ hybridization experiments are needed to confirm the expression of ET-B in Müller cells. The findings of Rattner and Nathans on the

association between ET-B and Müller cells, in addition to research on the pig retina, which found ET- B to be expressed by the innermost retinal layers, ganglion cell somata, and by Müller glial cells (Iandiev et al., 2005), have provided compelling circumstantial evidence that ET-B would be expressed by zebrafish Müller cells. However, other studies have demonstrated that ET-B is also expressed in other cells within the mouse retina, including horizontal cells and the retinal pigment epithelium (Torbidoni et al., 2005). Thus, I would need to investigate this possibility in the zebrafish. After obtaining the pattern of ET-B expression in the wild type zebrafish retina, I would repeat the analysis of ET-B expression on rose larvae that had been exposed to constant light to determine whether Müller cells or any other ET-B positive cells had an appreciable change in expression level. This would help to identify the cell type(s) involved in the response to photoreceptor stress. Next, I would like to determine the identity of the missing cells within the INL in light-treated rose larvae. One possibility explaining the absence of these cells is that cell death had occurred during the treatment, prior to day 5 when the animals were sacrificed. To do this, I would use a TUNEL cell death assay on retinas from rose larvae exposed to light for different periods of time and use cell-specific markers to determine the identity of TUNEL positive cells. Once I had more knowledge of the basis for the light-treated rose retinal phenotype, I could combine this model with a morpholino “gene knockdown” approach,

61

targeting candidate genes thought to be involved in the endothelin pathway. Another potentially interesting experiment would be to introduce the rose mutation to the lze mutant to explore the effects of combining models involving Müller cells and lightdependent degeneration. Finally, since many of the genes in the zebrafish have more than one copy, I might learn that there is more than one gene for ET-B. In this case, I would repeat the experiment and analysis using rose larvae treated with a known ET-B inhibitor, called BQ788, and assess the phenotype.

62

GENERAL CONCLUSIONS To examine how Müller cells may be involved in supporting photoreceptors, reliable models where Müller cells are compromised are needed. In this thesis, I

examined two candidate models of Müller cell disruption in the zebrafish retina, one pharmacological and one genetic. In my first model, I introduced a gliotoxin to adult and larval zebrafish through the water and observed modest signs of Müller cell stress in the adults but neuronal deficits in the larval retina. Because my approach did not yield the desired glial-specific effects I had hoped to achieve, I chose to focus my attention on the rose mutant model. Upon the introduction of retinal stress via constant light, photoreceptor cells and cells within the INL were compromised in the rose retina. The appearance of the failing rod photoreceptor outer segments is similar to early stages of photoreceptor degeneration observed in other animal models. Thus, even if I determine that the rose phenotype is not mediated by Müller cell deficiencies, rose still serves as a model of photoreceptor degeneration, where the same issues of neurotrophic support could be investigated. Whether rose will be a useful model to study how Müller cells are involved in the resilience of photoreceptors has yet to be determined; however, my finding that rod photoreceptors were more vulnerable to phototoxic stress coupled with the Rattner and Nathans result that light exposure led to the intense upregulation of ET-B in Müller cells has inspired me to explore more about this mutant.

63

REFERENCES Brokerhoff SE, Hurley JB, Ulrike JB, Neuhauss SCF, Driever W, Dowling JE. 1995. A behavioral screen for isolating zebrafish mutants with visual system defects. Neurobiol 99:10545-10549. Bilotta J, Saszik S, Sutherland SE. 2001. Rod contributions to the electroretinogram of the dark-adapted developing zebrafish. Dev Dyn 222(4): 564-570. Casper DS, Reif-Lehrer L. 1983. Effects of alpha-aminoadipate isomers on the morphology of the isolated chick embryo retina. Invest Opthalmol Vis Sci 24:1480-1488. DiLoreto DA Jr, Martzen MR, delCerro C, Coleman PD, del Cerro M. 1995. Müller cell changes precede photoreceptor cell degeneration in the age-related retinal degeneration of the Fischer 344 rat. Brain Res 698:1-14. Dubois-Dauphin M, Poitry-Yamate C, de Bilbao F, Julliard AK, Jourdan F, Donati G. 2000. Early postnatal Müller cell death leads to retinal but not optic nerve degeneration in NSE-Hu-Bcl-2 transgenic mice. Neurosci 95:9-21. Gass JD. 1999. Müller cell cone, an overlooked part of the anatomy of the fovea centralis: hypotheses concerning its role in the pathogenesis of macular hole and foveomacular retinoschisis. Arch Ophthalmol 117:821-823. Helmuth L. 2001. Science 291(5504): 569. Jablonski MM, Iannaccone A. 2000. Targeted disruption of Müller Cell metabolism induces photoreceptor dysmorphogenesis. Glia 32:192-204. Johns PR. 1982. Formation of photoreceptors in larval and adult goldfish. J Neurosci 2:178-198. Kainz PM, Adolph AR, Wong KY, Dowling JE. 2003. Lazy eyes zebrafish mutation affects Müller glial cells, compromising photoreceptor function and causing partial blindness. J Comp Neurol 463:265-280. Li CM, Yan RT, Wang SZ. 2991. Atrophy of Müller glia and photoreceptor cells in chick retina misexpressing cNSCL2. Invest Ophthalmol Vis Sci 42:3103-3109. Lister JA, Cooper C, Nguyen K, Modrell M, Grant K, Raible DW. 2006. Zebrafish foxd3 is required for development of a subset of neural crest derivatives. Dev Biol 290(1):92104. Loeffler KU, Li ZL, Fishman GA, Tso MO. 1992. Dominantly inherited cystoid macular edema a histopathologic study. Ophthalmol 99:1385-1392.

64

Maccumber MW, d’Anna SA. 1994. Endothelin receptor binding subtypes in the human retina and choroids. Arch Ophthamol 112:1231-1235. Miller G. 2005. The Dark Side of Glia. Science Vol. 308. Newman E, Reichenbach A. 1996. The Müller cell: a functional element of the retina. Trends Neurosci 19:307-312. Olney JW, Ho OL, Rhee V. 1971. Cytotoxic effects of acidic and sulfur containing amino acids on the infant mouse central nervous system. Expl Brain Res 14:61-76. Parichy DM, Mellgren EM, Rawls JF, Lopes SS, Kelsh RN, Johnson SL. 2000. Mutational analysis of endothelin receptor b1 (rose) during neural crest and pigment pattern development in the zebrafish (danio rerio). Dev Biol 227:294-306. Peterson RE, Tu C, Linser PJ. 1997. Isolation and characterization of a carbonic anhydrase homologue from the zebrafish (danio rerio). J Mol Evol 44:432-39. Prasanna G, Narayan S, Krishnamoorthy RR, Yorio T. 2003. Eyeing endothelins: a cellular perspective. Mol Cell Biochem 253:71-88. Rattner A, Nathans J. 2005. The genomic response to retinal disease and injury: evidence for endothelin signaling from photoreceptors to glia. J Neurosci 25(18): 4540-4549. Rattner A, Nathans J. 2006. An evolutionary perspective on the photoreceptor damage response. Am J Ophthalmol 141(3):558-566. Raymond PA, Rivlin PK. 1987. Germinal cells in the goldfish retina that produce rod photoreceptors. Devel Biol 122:120-138. Reid SN, Akhmedov NB, Piriev NI, Kozak CA, Danciger M, Farber DB. 1999. The mouse x-linked juvenile retinoschisis cDNA expression in photoreceptors. Gene 227:257266. Reid SNM, Farber DB. 2005. Glial Transcytosis of a Photoreceptor-Secreted Signaling Protein, Retinoschisin. Glia 49:397-406. Sakurai T, Yanagisawa M, Masaki T. 1992. Molecular characterization of endothelin receptors. Trends Pharmacol Sci 13:103-108. Sarthy V, Ripps H. 2001. The Retinal Müller cell. New York: Kluwer Academic/Plenum Press. Schmitt EA, Dowling JE. 1999. Early retinal development in the zebrafish, danio rerio: light and microscopic analyses. J Comp Neurol 404(4):515-536.

65

Schmitt EA, Hyatt GA, Dowling JE. 1999. Erratum: Temporal and spatial patterns of opsin gene expression in the zebrafish (danio rerio): corrections with additions. Vis Neurosci 16(3):601-605. Silver J and Rutishauser U. 1984. Guidance of optic axons in vivo by a preformed adhesive pathway on neuroepithelial endfeet. Devel Biol 106:485-99. Sugawara K, Torigoe K, Okoyama S, Negishi K, Kato S. 1990. Neurotoxic effects of lalpha-aminoadipic acid on the carp retina: a long term observation. Neurosci 36:155-163. Torbidoni, V, Iribarne, M, Suburo, A.M. 2005. Endothelin Receptors in Light-Induced Retinal Degeneration. Experimental Biology and Medicine 231(6):1095-1101 Uhlmann S, Bringmann A, Uckermann O, Pannicke T, Weick M, Ulbricht E, Goczalik I, Reichenbach A, Wiedmann P, Francke M. 2003. Early glial cell reactivity in experimental retinal detachment: effect of suramin. Invest Ophthalmol Vis Sci 44:41144122. Vihtelic TS, Hyde, DR. 2000. Light induced rod and cone cell death and regeneration in the adult albino zebrafish (danio rerio) retina. J Neurobiol. 44(3):289-307. Yanagisawa M, Inoue A, Ishikawa T, Kasuya Y, Kimura S, Kumagaye S, Nakajima K, Watanabe TX, Sakakibara S, Goto K et al. 1988. Primary structure, synthesis, and biological activity of rat endothelin an endothelium derived vasoconstrictor peptide. Proc Natl Acad Sci USA 85:6964-6967.

66

Master your semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master your semester with Scribd & The New York Times

Cancel anytime.