To evaluate the utility of the nrd:egfp
transgenic line, we compared the expression of the transgene to that of endogenous neurod
during retinal development, in the adult retina and during photoreceptor regeneration. Previously, RNA in situ
hybridization showed that during early retinogenesis neurod
is first expressed in the ventral nasal patch and then throughout the neuroepithelium. Subsequently, neurod
is transiently expressed in the nascent photoreceptors in the outer nuclear layer and persistently expressed in a subset of amacrine cells in the inner nuclear layer 
. Similarly, we show that the nrd:egfp
transgene is initially expressed adjacent to the ventral nasal patch ( and ), and then throughout the neuroepithelium and nascent photoreceptor layer (). In contrast to the in situ
data, however, EGFP is also present in bipolar cells, in a small fraction of rod photoreceptors at 2 wkpf, and in all rod photoreceptor cell bodies at adulthood.
There are potential explanations for the subtle temporal and cellular disparities in the expression of neurod
, as detected by in situ
hybridizations, and the expression of the nrd:egfp
transgene. One possibility is that the neurod
transgene lacks a required silencer or is influenced by neighboring enhancers near the site of integration. However, it would have to lie far outside the coding region, as the transgene contains 67 kb of sequence upstream and 89 kb of sequence downstream of neurod
open reading frame 
. Another possibility is that mature bipolar cells and rod photoreceptors, not observed following in situ
hybridizations, produce very low levels of endogenous neurod
, and the stability of EGFP more readily allows for the detection of these cells. In support of this interpretation, prior to light treatment we observed weak expression of endogenous neurod
in all rod photoreceptor cells by in situ
hybridization, and strong expression of EGFP in the same cells (, ).
We observed both overlapping and distinct expression profiles for endogenous and transgenic neurod expression during retinal regeneration. In both cases, neurod was not observed in dividing Müller glia or in the early stages of neuronal progenitor amplification. Both endogenous and transgenic neurod were first observed in INL progenitors in later stages of proliferation as these progenitors were migrating to the ONL ( and ). At this point endogenous neurod expression is very strong in these progenitors, whereas EGFP is very weak (cf. and ). By 3 days post light treatment, two distinct bands of expression were observed. At this point, endogenous neurod is downregulated in the first wave of newly regenerated rod photoreceptors that are closest to the INL, whereas EGFP was strongly expressed in these cells. In contrast, endogenous neurod is highly expressed in the next wave of rod photoreceptors located distal to the first band of cells, but EGFP is not yet present. These differences in endogenous and transgene expression may be explained by dynamic changes in endogenous neurod expression compared to the relatively long (~24 hour) half-life of EGFP. In each case, endogenous neurod expression proceeded EGFP expression and EGFP was visualized after the downregulation of endogenous neurod.
Expression of neurod
is often found in tissues with persistent mitotic activity. Although the zebrafish retina continues to grow throughout its life, we did not observe the neurod
transgene in known locations of persistent neurogenesis in the retina. For example, consistent with previously published in situ
transgene expression was not observed during retinogenesis in the progenitors located in the circumferential marginal zone (CMZ), but did overlap with PCNA-positive cells as they exit the CMZ and begin differentiating (). Similarly, during retinal regeneration, endogenous and transgenic neurod
was not observed in Müller glial or their immediate progeny, but in later stage progenitors prior to photoreceptor differentiation (, , ). This is consistent with anti-sense morpholino studies in early zebrafish development which show that in the absence of NeuroD, rod and cone progenitors fail to exit the cell cycle 
. In addition, the developing chick retina requires neurod
for photoreceptor differentiation 
. Together, these data suggest that the major function of NeuroD in the developing retina is in regulating mechanisms that promote cell cycle exit. It has yet to be determined whether NeuroD plays a similar role during retinal regeneration in the adult.
One potential use would be to utilize the line to visualize the reestablishment of the synapses connecting rod photoreceptor and bipolar cells. During intense light damage, rod photoreceptors are lost, but the underlying bipolar cells remain (). Once disconnected from the photoreceptor, the bipolar cell processes hypertrophy and bud out, presumably in an attempt to re-establish the lost connection (data not shown). Once the new photoreceptor is regenerated, this connection is re-established. Since a subset of bipolar cells and newly formed rod photoreceptors are both EGFP-positive, this line could be used for in vivo imaging and genetic manipulation of this dynamic and poorly understood process.
This line also has potential uses for studies on the endocrine pancreas. NeuroD has been shown to be expressed in the endocrine pancreas in a variety of vertebrates 
. Loss of NeuroD in mice results in abnormal pancreatic β-cell maturation and function 
, severe hyperglycemia and neonatal death 
. We show the neurod
transgene is expressed the endocrine pancreas and could be used as a visual marker for β-cell function, particularly in the growing field using zebrafish as a vertebrate model for diabetes 
In summary, given the diverse areas of neurod expression in the developing and adult zebrafish, we anticipate that the Tg(nrd:egfp)/alb line will be a useful tool in multiple disciplines, including future studies on photoreceptor differentiation and retinal progenitor proliferation.