A Disintegrin and Metalloproteinase10 (ADAM10) Regulates NOTCH Signaling during Early Retinal Development

ADAM10 and ADAM17 are two closely related members of the ADAM (a disintegrin and metalloprotease) family of membrane-bound sheddases, which proteolytically cleave surface membrane proteins. Both ADAM10 and ADAM17 have been implicated in the proteolytic cleavage of NOTCH receptors and as such regulators of NOTCH signaling. During retinal development, NOTCH signaling facilitates retinal neurogenesis by maintaining progenitor cells in a proliferative state and by mediating retinal cell fates. However, the roles of ADAM10 and ADAM17 in the retina are not well defined. In this study, we set out to clarify the roles of ADAM10 and ADAM17 during early retinal development. The retinal phenotype of conditionally abated Adam17 retinae (Adam17 CKO) did not differ from the controls whereas conditionally ablated Adam10 retinae (Adam10 CKO) exhibited abnormal morphogenesis characterized by the formation of rosettes and a loss of retinal laminae phenotypically similar to morphological abnormalities identified in mice with retinal NOTCH signaling deficiency. Additionally, Adam10 CKO retinae exhibited abnormal neurogenesis characterized by fewer proliferating progenitor cells and greater differentiation of early photoreceptors and retinal ganglion cells. Moreover, constitutive activation of the NOTCH1-intracellular domain (N1-ICD) rescued Adam10 CKO abnormal neurogenesis, as well as abnormal retinal morphology by maintaining retinal cells in the progenitor state. Collectively these findings provide in vivo genetic evidence that ADAM10, and not ADAM17, is indispensable for proper retinal development as a regulator of NOTCH signaling.


Introduction
During retinal development, all retinal cell types are derived from a single population of pluripotent retinal progenitor cells (RPCs). The birth order of retinal cells is unidirectional and highly conserved although at any given developmental time point there is an overlap in the generation The goal of this study was to determine the roles of ADAM10 and ADAM17 in the retina and establish whether ADAM10, ADAM17, or both regulate retinal NOTCH1 signaling. Our results demonstrate that ADAM10, and not ADAM17, is indispensable for early retinal development as a key regulator of NOTCH1 signaling.

Mice and clinical eye evaluation
All animal procedures were performed in accordance with NIH guidelines and approved by the Medical College of Wisconsin Institutional Animal Care and Use Committee (IACUC). Mice carrying a floxed Adam10 allele (Adam10 tm1.1Zhu ) were generously provided by Tian Xu. Mice carrying a floxed Adam17 allele (Adam17 tm1.2Bbl ), were generously provided by Carl Blobel. Six3-Cre mice carrying a Cre allele driven by the promoter of the Six3 gene were generously provided by Yasuhide Furuta. Gt(ROSA)26Sor tm1(NOTCH1)Dam /J. Each mouse strain was genotyped as previously described [24,32,36,37] using primers in S1 Table. Mice with Six3-Cre; Adam10 tm1Zh/tm1Zhu or Six3-Cre; Adam17 tm1Bbl/tm1Bbl genotypes were referred to here as Adam10 CKO mice and Adam17 CKO mice respectively. Littermates with genotypes Adam10 tm1Zhu/+ and Adam10 tm1Zhu/ tm1Zhu as well as Adam17 tm1Bbl/+ and Adam17 tm1Bbl/tm1Bbl are referred to as controls.
Mouse eyes from the eight week old control, Adam10 CKO, and Adam17 CKO mice were examined with a Topcon SL-D8Z slit lamp biomicroscope with a Nikon SLR-based Photo Slit Lamp imaging system following mydriasis with 1% atropine sulfate (Bausch & Lomb) as previously described [38]. Control (n = 3) and Adam17 CKO (n = 3) animals at 8 weeks of age were dark adapted for 2 hr and ERGs were recoded after a single flash of intensity sufficient to evoke the maximum retinal response as previously described [39].

RT-PCR
Eyes from newborn mice (n = 3) of each genotype were collected and the cornea, iris, and lens were removed. RNA was isolated and reverse transcribed as previously described [40]. Adam10 or Adam17 transcript levels were analyzed via semi-quantitative RT-PCR while in the exponential phase of the PCR amplification using Gapdh as an internal control as previously described [40] using primers in S1 Table. PCR band intensities were quantified using ImageJ software (http://rsbweb.nih.gov/ij/) and are expressed relative to Gapdh. The results represent at least three independent experiments performed in triplicates and the data is expressed as mean ± SEM. Comparison between the genotype groups was analyzed by Student's t-test and P<0.05 was considered as statistically significant.

Histology and immunohistochemistry
Tissues were fixed in 4% PFA, or Davidson's solution, embedded in paraffin, sectioned to 4 μm thickness and Hematoxylin and Eosin (H&E) stained as previously described [38]. For immunohistochemistry, following deparaffinization and rehydration, antigen retrieval was performed using 1x citrate buffer pH 6.0 (Life Technologies) as previously described [38] and then sections were blocked in 10% normal serum with 1% BSA in 1xPBS for 1-2 hr and incubated with primary antibody in 1xPBS with 1% BSA O/N at 4°C. 2x 5 min washes in 1xPBS, 0.025% Triton X-100 were followed by secondary antibody incubation in 1xPBS with 1% BSA for 1 hr at RT. Primary antibodies used in this study include: ADAM10 (R&D Systems), BRN3B, ADAM17 OTX2, and N-cadherin (Abcam), ISL1 (Hybridoma Bank), β-catenin, NOTCH1 and HES1 (Cell Signaling). Secondary antibodies used were Alexa Fluor 546 goat-anti rabbit and Alexa Fluor 488 goat-anti mouse (Life Technologies). All slides were counterstained with DAPI (Life Technologies), mounted using Prolong gold anti-fade reagent (Life Technologies) mounting media, and photographed with a Nikon DS-Fi1 camera on a Nikon eclipse 80i microscope. The percentage of OTX2 and ISL1positive cells were determined by counting the numbers of OTX2 and ISL1 positive cells and total DAPI positive cells from a minimum of 3 separate genotypes with at least 10 sections per genotype. Significance was calculated via Students t-test (GraphPad Prism), and P<0.05 was considered significant.
EdU, BrdU, and TUNEL assays EdU (Life Technologies) or BrdU was injected intraperitoneally at a dose of 125 mg/kg of body weight into pregnant dams, 13.5 days after a vaginal plug was found and sacrificed 2 hr later. Embryos were fixed in 4% PFA for 2 hours and dehydrated through a graded series of ethanol washes, were paraffin embedded, and cut into 5 μm sections. EdU detection was performed with Click-iT EdU Alexa Fluor 488 Imaging Kit (Life Technologies) according to the manufacturer's protocol and BrdU was detected using a mouse monoclonal anti-BrdU antibody conjugated to Alexa Flour 488 (Life Technologies). TUNEL staining was performed utilizing the ApopTag Plus In Situ Apoptosis Fluorescein Detection Kit (Millipore) according to the manufacturer's recommendations. DAPI (Life Technologies) was used to stain the DNA, slides were mounted using Prolong gold anti-fade reagent (Life Technologies) and imaged using a Nikon DS-Fi1 camera on a Nikon Eclipse 80i microscope. The percentage of EdU, BrdU, and TUNEL positive cells after counting the numbers of EdU, BrdU or TUNEL positive cells and total DAPI positive cells from a minimum of 3 separate genotypes with at least 10 sections per genotype. Significance was calculated via a t-test (GraphPad Prism) and P<0.05 was considered significant.

ADAM10 and ADAM17 are ubiquitously expressed retinal proteins
As the initial step, we evaluated the expression of ADAM10 and ADAM17 during retinal development in mice. In control embryonic E13.5, E16.5 and newborn P0.5 retinae both ADAM10 ( Fig 1A, 1C and 1E) and ADAM17 (Fig 1B, 1D and 1F) exhibit ubiquitous expression in ganglion cell and neuroblastic layers. By P7, as the retina starts to mature, ADAM10 and ADAM17 are expressed in the ganglion cell layer, inner and outer plexiform layers, and in the photoreceptors (Fig 1G and 1H) and this pattern of expression is maintained in the adult retina for both ADAM10 ( Fig 1I) and ADAM17 ( Fig 1J).

ADAM10, but not ADAM17, is required for early retinal development
To determine the roles of ADAM10 and ADAM17 during early retinal development, we generated mice with retina-specific ablation of Adam10 (Adam10 CKO), as well as retina-specific ablation of Adam17 (Adam17 CKO) (Fig 2A and 2B). In newborn Adam10 CKO and Adam17 CKO retinae, only residual levels of Adam10 transcript containing exon 3 ( Fig 2C) and Adam17 transcript containing exon 2 ( Fig 2D) were identified when compared to the controls. Therefore, Adam10 and Adam17 were successfully disrupted in most retinal cells in Adam10 CKO and Adam17 CKO mice respectively.
The evaluation of adult (8 weeks of age) Adam10 CKO eyes by observation revealed a smaller eye size ( Fig 2F) when compared to age-matched control littermates (Fig 2E). Adam17 CKO mice did not exhibit any obvious eye size differences ( Fig 2H) when compared to agematched control littermates (Fig 2G). Histological analysis of adult Adam10 CKO eyes identified profoundly disrupted retinal laminae (Fig 2J and  To exclude the possibility that Adam17 CKO retinae may exhibit functional, but not morphological abnormalities we performed electroretinograms (ERGs). Our analysis did not identify any differences in ERGs  A similar approach was taken to generate Adam17 CKO mice after homozygote Adam17 tm1.2Bbl mice were crossed to Six3-Cre mice. (C) Semi-quantitative analysis of the relative expression of Adam10 transcript containing exon 3 normalized to Gapdh expression from RNA isolated from newborn control and Adam10 CKO littermates revealed significantly reduced (P<0.0001; n = 3) levels in Adam10 CKO retinae when compared to the controls. (D) Semi-quantitative analysis of the relative expression of exon 2 containing Adam17 mRNA normalized to Gapdh expression from RNA isolated from newborn control and Adam17 CKO retinae also revealed significantly reduced (P<0.0002; n = 3) levels in Adam17 CKO retinae when compared to the controls. In both (C) and (D), the bars represent mean values ± SEM. Significance was established following Student's t-test analysis and P<0.05 was considered significant. (F) Adam10 CKO eyes appeared smaller when compared to the eyes in the control mice (E) whereas Adam17 between the control and Adam17 CKO mice (not shown). Collectively these findings provided evidence that ADAM10, and not ADAM17, is required for early retinal development.
The onset and progression of retinal phenotypes in Adam10 CKO mice The first morphological abnormalities in Adam10 CKO retinae were evident at E13.5 characterized by disrupted cellular architecture ( Fig 3B). Along the apical-basal axis nuclei from Adam10 CKO retinae appeared rounded and disorganized resulting in misshapen apical and basal surfaces (Fig 3B'). By contrast, the E13.5 control retinae exhibited elongated and organized nuclei with distinct apical and basal surfaces ( Fig 3A and 3A'). While morphological abnormalities were identified within the Adam10 CKO central retinae some regions of peripheral retinae retained undisrupted retinal cell organization ( Fig 3B). It was reported previously that Six3-Cre is expressed across the entire retina although the highest expression is within the central retina [36]. To explore this further, we immunostained the control and the Adam10 CKO retinae for ADAM10. Within the central Adam10 CKO retinae, the majority of cells did not express ADAM10 (Fig 3D and S2B Fig). However, at the retinal periphery, we identified residual ADAM10 immunostaining (Fig 3D and S2B Fig). This finding suggested that milder retinal abnormalities identified at the retinal periphery of E13.5 Adam10 CKO may have been caused by the lower efficiency of CRE-mediated Adam10 ablation relative to the central retina. By E16.5 the Adam10 CKO retinae started to exhibit rosettes within the central retinae ( Fig  3F); the identified rosettes were of different sizes (Fig 3F'). Immunostaining of E16.5 Adam10 CKO retinae for ADAM10 did not identify any residual staining (Fig 3H). We also morphologically evaluated the postnatal retinal development of Adam10 CKO mice. By P0.5 in Adam10

ADAM10 inactivation disrupts retinal neurogenesis
Next, we wanted to determine if Adam10 CKO retinae exhibited disrupted neurogenesis. At E13.5 in control retinae, EdU-positive cells were highly organized and restricted to the neuroblastic layer (Fig 4A) whereas in E13.5 Adam10 CKO retinae EdU-positive cells were highly disorganized ( Fig 4B). Quantification analysis revealed a significantly lower (P = 0.019; n = 3) percentage of EdU-positive cells in the Adam10 CKO retinae when compared to the controls ( Fig 4C). Next, we hypothesized that the fewer proliferating cells identified in Adam10 CKO retinae (Fig 4A-4C) may be caused by precocious retinal cell differentiation. Immunostaining of the control retinae for orthodenticle homeobox 2 (OTX2), which is a marker for early photoreceptors [41], revealed sparse OTX2-positive staining in cells along the retinal apical surface consistent with the development of early photoreceptors (Fig 4D). By contrast, Adam10 CKO retinae exhibited highly disorganized OTX2-positive cells (Fig 4E). Quantification analysis revealed a significantly greater (P = 0.003; n = 3) percentage of OTX2-positive cells in the Adam10 CKO retinae when compared to the controls (Fig 4F). We also evaluated the ISL1 expression as a marker for early ganglion cells [42]. In control retinae, ISL1 immunostaining CKO eyes (H) that did not appear to differ from age-matched control littermates (G). H&E staining of Adam10 CKO eyes at 8 weeks of age (J) revealed severe retinal abnormalities characterized by the formation of rosettes (J', arrow) in contrast to the retinal laminae in the age-matched controls (I-I'). Histological analysis of Adam17 CKO eyes (L-L') did not identify any morphological differences when compared to the age-matched controls (K-K').  was restricted to the cells at the retinal basal layer consistent with the developing ganglion cells (Fig 4G). In the Adam10 CKO retinae, ISL1-positive cells were highly disorganized (Fig 4H). Quantification analysis revealed a significantly greater (P = 0.004; n = 3) percentage of ISL1-positive cells in E13.5 Adam10 CKO when compared to the control retinae (Fig 4I). Similar findings were observed following immunostaining with BRN3B (S3A-S3C Fig), a marker for early ganglion cells [42]. In both the control (Fig 4J) and Adam10 CKO retinae (Fig 4K), the OTX2-positive and ISL1-positive cells did not co-stain for EdU, indicating that OTX2-positive and ISL1-positive cells had exited the cell cycle.
Characterization of rosettes in E16.5 Adam10 CKO retinae Morphological evaluation of Adam10 CKO retinae revealed the formation of rosettes at E16.5 ( Fig 3F and 3F'). Next we set to characterize which cells formed the rosettes in E16.5 Adam10 CKO retinae. In the control, the retinae BrdU-positive cells are positioned between the OTX2positive photoreceptors on the apical side and ISL1-positive ganglion cells at the basal side ( Fig  5A and 5C). Immunostaining of Adam10 CKO retinae revealed that BrdU-positive cells were forming the rosette outer cellular layers (Fig 5B) whereas OTX2 positive photoreceptors were forming the rosette inner cellular layers (Fig 5D). However, in the Adam10 CKO retinae, OTX2-positive photoreceptors were also identified as highly disorganized cells present at the apical and basal surface as well as cells randomly scattered throughout the retina (Fig 5D). The ISL1-positive cells did not associate with rosettes and appeared to be randomly distributed throughout the Adam10 CKO (Fig 5D). It was shown previously that both N-cadherin and βcatenin are indispensable for the formation of retinal laminae [43,44], thus we wanted to determine if N-cadherin and β-catenin expression had been altered in the Adam10 CKO retinae. In the control retinae both N-cadherin and β-catenin (Fig 5E-5G) were expressed throughout the retina especially in the cells forming apical and basal surfaces consistent with previous reports [44]. In the Adam10 CKO retinae both N-cadherin ( Fig 5F) and β-catenin ( Fig 5H) were highly expressed in cells forming the rosette lumens whereas cells forming apical and basal surfaces intermittently expressed both N-cadherin and β-catenin. Next, we wanted to determine if cell death was contributing to the formation of rosettes. In E16.5 control retinae, we identified a very few TUNEL-positive cells (Fig 5I) whereas in Adam10 CKO the TUNEL-positive cells were scattered throughout all layers of the retinae (Fig 5J). Although a significantly greater (P = 0.003; n = 3) percentage of TUNEL-positive cells were identified in Adam10 CKO retinae (Fig 5H) when compared to controls, the overall percentage of TUNEL-positive cells was very small (<5%) in both the WT and Adam10 CKO retinae (Fig 5K). Consistent with this, the overall number of DAPI positive cells did not significantly differ (P = 0.19; n = 3) between the two genotypes ( Fig 5L).

ADAM10 regulates NOTCH signaling
Next, we set out to determine if disrupted NOTCH1 signaling contributed to the retinal phenotypes identified in Adam10 CKO retinae. Immunostaining of E13.5 control retinae revealed that NOTCH1 positive immunostaining was restricted to the neuroblastic layer (Fig 6A). By contrast in E13.5 Adam10 CKO retinae very few cells within the central retina stained positive for NOTCH1, although towards the retinal periphery highly disorganized NOTCH1 positive cells were evident (Fig 6B). Quantification analysis confirmed a significantly (P = 0.001; n = 3) laminae in the age-matched control retinae (E-E'). Immunostaining of E16.5 Adam10 CKO retinae (H) did not identify any ADAM10-positive cells in contrast to the age-matched controls (G). BA = basal surface, AP = apical surface. Scale bars = 50 μm.    (Fig 6C). Immunostaining for HES1, which is a NOTCH1 downstream target [15], revealed a very few HES1-immunopositive cells within the central E13.5 Adam10 CKO retinae that appeared highly disorganized (Fig 6D). However, at the retinal periphery HES1 positive cells were identified restricted to the neuroblastic layer (not shown). Quantification analysis confirmed a significantly (P = 0.0004; n = 3) lower percentage of HES1-positive in E13.5 Adam10 CKO retinae when compared to the age-matched controls.
In order to determine if disrupted NOTCH1 signaling is directly caused by a functional loss of ADAM10, we utilized a genetic approach. We focused on establishing if overexpression of the NOTCH1-intracellular domain (N1-ICD) can rescue aberrant morphogenesis and/or neurogenesis identified in Adam10 CKO retinae. We crossed Adam10 CKO mice to Gt(ROSA) 26Sor tm1(NOTCH1)Dam /J mice and generated mice with genotypes Six3-Cre;Adam10 CKO; Rosa N1-IC/+ (Adam10 CKO;N1-ICD) wherein the same retinal cells Six3-driven CRE recombinase activity deleted Adam10 exon 3 and overexpressed N1-ICD; mice with Six3-Cre; Rosa N1-IC/+ (N1-ICD) genotypes, where Six3-CRE-mediated constitutive retinal activation of N1ICD, were used as controls (Fig 6G). At E13.5 cells, both N1-ICD retinae ( Fig 6H) and Adam10 CKO;N1-ICD retinae (Fig 6I), were highly organized along the apical-basal axis with well-defined apical and basal surfaces. EdU incorporation revealed that in both N1-ICD retinae ( Fig 6J) and Adam10 CKO; N1-ICD retinae (Fig 6K) proliferative cells were present in all retinal layers suggesting that all retinal cells in these mice were maintained in the progenitor state. The percentage of EdU positive cells did not significantly differ between the genotypes ( Fig  6L). To explore this further, we immunostained E13.5 N1-ICD retinae and Adam10 CKO; N1-ICD retinae for OTX2 and ISL1 as early photoreceptor and ganglion cell markers respec-

Discussion
In this study, we show that ADAM10 is indispensable for retinal neurogenesis as well as the formation of retinal laminae. The rescue of aberrant retinal phenotypes in ADAM10-deficient retinae by overexpression of the N1-ICD provides in vivo genetic evidence that ADAM10 is a key regulator of NOTCH1 signaling. In mice during early retinal development NOTCH1 maintains retinal cells in the undifferentiated progenitor state, but it also has a role in repressing photoreceptor cell fate [10,11]. As such the role of NOTCH1 signaling has been proposed catenin (G) were expressed in all retinal cells although the highest expression was in the cells forming the apical surface (white arrows in E and G). In Adam10 CKO retinae cells forming the rosette lumens were highly expressing N-cadherin and β-catenin (white arrows F and H); cells at the apical and basal surface were intermittently expressing N-cadherin (green arrows in F) and β-catenin (red arrows in H). Very few TUNELpositive cells were identified in the control (I) and Adam10 CKO (J) retinae. Quantification analysis (K) identified in the controls 0.4 ± 0.01% (gray bar) and in Adam10  to have a key role in assuring the sufficient number of progenitor cells necessary for the development of all retinal cell types [10,11]. Results from this study showed that ADAM10 deficiency decreased the pool of retinal progenitor cells and increased the production of early photoreceptors further supporting the idea that ADAM10 is a regulator of NOTCH1 signaling. However, our results also show that ADAM10 retinal deficiency increased the production of ganglion cells, a phenotype not identified in mice with NOTCH1-deficient retinae [10,11]. ADAM10 ablation in the developing retina phenocopies aberrant retinal neurogenesis in mice with retinae deficient for both NOTCH1 and NOTCH3 [45] as well as in mice with retinae deficient for RPBJ [12]. While it is well established that NOTCH1 is a regulator of the progenitor cell pool and represses photoreceptor fate, recent studies have shown both NOTCH3 and NOTCH1 contribute to the repression of ganglion cell fate [45]. Furthermore both NOTCH1 and NOTCH3 signal through RBPJ as a common downstream target [12]. Therefore, it was hypothesized that NOTCH1+NOTCH3>RBPJ repress retinal ganglion cell fate. We propose that ADAM10 is a common NOTCH signaling regulator by facilitating the cleavage of both NOTCH1 and NOTCH3 which in turn facilitates maintenance of the progenitor cell pool and repression of both photoreceptors and ganglion cells. In support of this hypothesis are recent reports showing that in cell culture ADAM10 cleaves NOTCH3 via a similar mechanism as NOTCH1 [46]. Additionally, during retinal development expression of NOTCH1 and NOTCH3 spatially and temporally coincide [47]. However, further studies are needed to unequivocally establish if ADAM10 mediates the cleavage of both NOTCH1 and NOTCH3 during early retinal neurogenesis.
In addition to aberrant neurogenesis, ADAM10 retinal ablation also results in severely disrupted formation of retinal laminae. Prior to differentiation, RPCs are positioned along the apical-basal axis and are attached to both the apical and basal surfaces [48,49]. As RPCs start to differentiate, they undergo detachment allowing the cells to migrate to the appropriate retinal layer thereby establishing the retinal laminae and neuronal circuitry required for vision [48,49]. Our results show that ADAM10 retinal deficiency results in the formation of rosettes similar to the rosettes reported for mice with disrupted retinal NOTCH1 signaling [9][10][11][12][13][14][15]. It was shown that RBPJ retinal deficiency causes discontinuous expression of β-catenin resulting in the formation of rosettes; ectopic overexpression of β-catenin in RBPJ-deficient retinae rescued the rosette formation without rescuing neurogenesis defects [13]. How disrupted retinal NOTCH1 signaling alters β-catenin expression still remains unclear. Active β-catenin is a downstream effector of Wnt signaling [50], but active β-catenin is also bound to N-cadherin at the cell membranes [51,52]. A functional deficiency of either N-cadherin or β-catenin disrupts the formation of retinal laminae without disrupted neurogenesis [43,44,53,54]. However, it was shown previously that ADAM10 mediates the cleavage of N-cadherin and as such regulates the release of active βcatenin into the cytoplasmic pool [51]. At this point, it is unclear if the rosetting phenotypes observed in ADAM10-deficient retinae are solely caused by the disrupted NOTCH1 signaling or if disrupted ADAM10-mediated cleavage of N-cadherin also contributes to the aberrant retinal morphogenesis. The rescue of aberrant retinal morphology in ADAM10-deficeint retinae by overexpression of N1-ICD supports the idea that the failure in the formation of laminae in the ADAM10-deficeint retinae is solely caused by disrupted NOTCH1 signaling. However, we cannot exclude the possibility that ADAM10 may be regulating the cleavage of N-cadherin as a downstream event following cleavage of the NOTCH1 receptor.
While this study establishes ADAM10 as indispensable for the early retinal development, the role of ADAM17 still remains unclear. Our results show that ADAM10 and ADAM17 exhibit an overlapping spatiotemporal expression pattern in the developing mouse retinae consistent with the expression of ADAM10 and ADAM17 previously reported for the developing chick retinae [34]. However, it should be pointed out that another study using a different antibody identified that in mice, the retinal expression of ADAM17 starts at P7 [55]. Furthermore, the same study revealed that in the adult (P175) mouse retinae ADAM17 expression is restricted to the ganglion cell layer, the inner and outer edge of the inner nuclear layer [55]. This discrepancy may have been caused by the variable specificity of the antibodies used as well as different time points selected for the analysis. Nonetheless, our findings unequivocally indicate that ADAM17 is dispensable for retinal development. In the developing chick retina, in addition to ADAM10, ADAM17 exhibits similar spatiotemporal expression pattern with ADAM9, ADAM12, ADAM13, ADAM22, and ADAM23 [34]. Except for ADAM9, that has been shown to have a role in the development of retinal vasculature and homeostasis [56,57], the roles of other ADAM proteins during retinal development and homeostasis are unknown providing no information if ADAM17 may have functional overlaps with other ADAM proteins. Studies focusing on the role of ADAM17 as a regulator of retinal homeostasis are currently in progress.  Table. A list of primers used in the study. (DOCX)