1. Developmental Biology
  2. Neuroscience
Download icon

Neural retina-specific Aldh1a1 controls dorsal choroidal vascular development via Sox9 expression in retinal pigment epithelial cells

  1. So Goto
  2. Akishi Onishi  Is a corresponding author
  3. Kazuyo Misaki
  4. Shigenobu Yonemura
  5. Sunao Sugita
  6. Hiromi Ito
  7. Yoko Ohigashi
  8. Masatsugu Ema
  9. Hirokazu Sakaguchi
  10. Kohji Nishida
  11. Masayo Takahashi
  1. RIKEN Center for Developmental Biology, Japan
  2. Osaka University Graduate School of Medicine, Japan
  3. Kobe City Eye Hospital Research Center, Japan
  4. RIKEN Center for Life Science Technologies, Japan
  5. Shiga University of Medical Science, Japan
Research Article
  • Cited 4
  • Views 1,306
  • Annotations
Cite this article as: eLife 2018;7:e32358 doi: 10.7554/eLife.32358

Abstract

VEGF secreted from retinal pigment epithelial (RPE) cells is responsible for the choroidal vascular development; however, the molecular regulatory mechanism is unclear. We found that Aldh1a1–/– mice showed choroidal hypoplasia with insufficient vascularization in the dorsal region, although Aldh1a1, an enzyme that synthesizes retinoic acids (RAs), is expressed in the dorsal neural retina, not in the RPE/choroid complex. The level of VEGF in the RPE/choroid was significantly decreased in Aldh1a1–/– mice, and RA-dependent enhancement of VEGF was observed in primary RPE cells. An RA-deficient diet resulted in dorsal choroidal hypoplasia, and simple RA treatment of Aldh1a1–/– pregnant females suppressed choroid hypoplasia in their offspring. We also found downregulation of Sox9 in the dorsal neural retina and RPE of Aldh1a1–/– mice and RPE-specific disruption of Sox9 phenocopied Aldh1a1–/– choroidal development. These results suggest that RAs produced by Aldh1a1 in the neural retina directs dorsal choroidal vascular development via Sox9 upregulation in the dorsal RPE cells to enhance RPE-derived VEGF secretion.

https://doi.org/10.7554/eLife.32358.001

eLife digest

The retina is the part at the back of our eyes that detects light and sends this information to our brain. Within the retina is a layered structure containing the light-sensitive cells, known as the neural retina, and another protective layer of cells called the retinal pigment epithelium. A surrounding network of blood vessels, the choroid, keeps the retina healthy by supplying oxygen and nutrients. When the choroid does not work properly, eye disease can result. A common example is age-related macular degeneration, where blood vessels in the choroid either break down or start growing uncontrollably in the wrong places. In both cases, light-sensitive cells are damaged and eventually die. This causes vision loss that worsens over time.

The choroid forms early in life, within the developing embryo. The retinal pigment epithelium helps the choroid to develop properly by producing a protein, VEGF, which supports the growth of blood vessels. However, it was not clear what signals tell this tissue to start making VEGF in the first place and then to keep making it.

To address this, Goto et al. looked at eye development in mutant mice that lack an enzyme called Aldh1a1. This enzyme’s role is to make a molecule called retinoic acid, which is known to be vital for many biological processes including the growth and development of embryos. Aldh1a1 is not made in the choroid of normal mice, just in the neural retina. Yet the choroid in the mutant mice without Aldh1a1 still grew fewer blood vessels than normal. Their retinal pigment epithelium also produced less VEGF and had lower levels of a protein called Sox9, which is known to make the gene for VEGF more active.

Goto et al. went on to show that simply removing retinoic acid from the diet of normal mice produced the same choroid defect as in the mutant mice with no Aldh1a1. Genetically manipulating otherwise normal mice to decrease the levels of Sox9 in the retinal pigment epithelium had a similar effect. In contrast, giving Aldh1a1-deficient mice extra retinoic acid or artificially increasing their levels Sox9 was enough to make the choroid develop normally. These experiments showed that retinoic acid produced in the neural retina directs choroid development by making Sox9 more active, which in turn encourages the retinal pigment epithelium to produce VEGF.

These findings bring new insights into the molecular signals that control choroid development. In the future, they may also help scientists to better understand why blood vessels in the choroid become abnormal in eye diseases like age-related macular degeneration.

https://doi.org/10.7554/eLife.32358.002

Introduction

The retina is the light-sensitive tissue at the back of the eye that consists of photoreceptor cells, retinal pigment epithelium (RPE), and its basement (Bruch’s) membrane. Disorder of the retina causes vision loss. The primary nutrient source for the retina is the choroid (Bill et al., 1980), which is a highly vascularized tissue layer surrounding the retina. The choroid consists of three layers: Haller’s layer (large blood vessel layer), Sattler’s layer (medium size blood vessels), and the choriocapillaris (Hogan et al., 1971; Nickla and Wallman, 2010). The choriocapillaris is a unique anastomosed vascular structure with an extraretinal fenestrated capillary bed that lies in a single plane below Bruch’s membrane.

In the mouse retina, choroidal development begins from embryonic day (E)13.5 (Marneros et al., 2005; Saint-Geniez et al., 2006). Vascular endothelial growth factor (VEGF) is known to be a central regulator of vascular development during embryogenesis (Coultas et al., 2005), and VEGF secreted from RPE cells is indispensable for the vascular development and maintenance of the choroid (Saint-Geniez et al., 2009; Le et al., 2010; Kurihara et al., 2012). VEGF expression is enhanced by a number of transcription factors such as hypoxia-induced factor-1α (HIF-1α), estrogen-related receptor-α (ERRα), and peroxisome proliferator-activated receptor gamma coreceptor 1α (PGC-1α) (Ziello et al., 2007; Ueta et al., 2012). However, the molecular mechanism of choroidal vascular development and the regulatory mechanism of VEGF secreted from RPE have not been clarified.

In the present study, we found that aldehyde dehydrogenase one family, member A1 (Aldh1a1) knockout (Aldh1a1–/–) mice have hypoplasia in the dorsal region of the choroid. Aldh1a1 is an enzyme that synthesizes retinoic acids (RAs) and is expressed in the dorsal neural retina from the embryonic stage, not in the RPE and the choroid (Matt et al., 2005; Molotkov et al., 2006; Luo et al., 2006; Kumar et al., 2012). RAs are essential for biological activities such as reproduction, development, growth, and maintenance (Kam et al., 2012; Rhinn and Dollé, 2012; Cunningham and Duester, 2015). We analyzed choroidal vascular development in Aldh1a1–/– mice, and demonstrated how Aldh1a1 expressed in the neural retinas in trans enhances VEGF secretion from the RPE to the choroid. We also found that, mechanistically, Sox9 expression in RPE is downstream of the signaling pathway mediated by Aldh1a1 in the neural retina.

Results

Aldh1a1 is preferentially expressed in the dorsal region of the mouse neural retinas, not in the RPE/choroid complex

It was previously reported that Aldh1a1 is expressed in the dorsal region of embryonic and adult retinas (McCaffery et al., 1991; Fan et al., 2003; Matt et al., 2005). However, the retinal cell types that express Aldh1a1 have not been precisely identified. Therefore, we performed immunohistochemistry and in situ hybridization studies in embryonic and adult mouse retinas.

At E12.5, Aldh1a1 was localized in the progenitor cells at the dorsal neural retina, and no Aldh1a1 expression was observed in the RPE and choroid complex (Figure 1A). Aldh1a3, an Aldh1 family enzyme that generates RA, was mainly localized in the ventral region of the retina and RPE. At E17.5, Aldh1a1 expression was observed in the dorsal region and the ventral edge, but no expression was detected in the RPE/choroid complex (Figure 1B). In adult tissue, in situ hybridization showed that Aldh1a1-positive cells were preferentially localized in the middle of the inner nuclear layer (INL) cells, and had spread to the dorsal region of the retina (Figure 1—figure supplement 1A and B). We also observed Aldh1a1-positive cells at the ventral edge of the adult retina (Figure 1C). Immunohistochemistry of sectioned tissues showed Aldh1a1 colocalized with glutamine synthetase (GS), a marker of Müller glia. These results indicate that Aldh1a1 is preferentially expressed in the retinal progenitor cells of the dorsal region from the embryonic stage, and in Müller cells in the dorsal region and the ventral edge of the retina in adulthood.

Figure 1 with 1 supplement see all
Aldh1a1 is predominantly expressed in the dorsal neural retina during embryonic and adult stages, and Aldh1a1–/– mice have less pigmentation and a thin choroid in the dorsal area.

(A–C) Section immunohistochemistry of the mouse retinas labeled with Aldh1a1, Aldh1a3 and glutamine synthetase (GS) antibodies. (A) At E12.5, Aldh1a1 (green) and Aldh1a3 (red) are expressed in dorsal and ventral neural retina (NR), respectively. At the dorsal edge (white box), Aldh1a1 is expressed only in NR, not in RPE (right upper panel), while some RPE cells are Aldh1a3-positive (right lower panel). (B) At E17.5, Aldh1a1 (green) is expressed in the dorsal half and the ventral edge of the retina, but no Aldh1a1-positive RPE cells were detected in the dorsal (right upper panel) and ventral (right lower panel) regions. (C) At 8 weeks, Aldh1a1 (green)-positive cells were double-labeled with GS (red) at the dorsal, boundary, and ventral edge regions (see Figure 1—figure supplement 1A, B; for the ‘boundary’ region). ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. (D) Dissected eyeballs and choroidal flat-mounts of adult (8-week-old) wild-type (WT) and Aldh1a1–/– mice. Loss of pigmentation was observed in the dorsonasal region (red arrowhead) of Aldh1a1–/– eyes. (E) Hematoxylin and eosin (H and E) staining of WT and Aldh1a1–/– retinal sections. The pigmented layer (choroid) of the dorsal Aldh1a1–/– retina is thinner than those on the other sides. ONL, outer nuclear layer; IS/OS, inner segment/outer segment; RPE, retinal pigment epithelium; Chd, choroid. (F–H) Quantitative evaluation of choroidal thickness (F), retinal thickness (G), and the length of outer segments (H) of the H and E-stained retinal sections. Data represent the average ±SD; n = 8–9 from 4 to 5 mice per group. *p<0.05. N.S., not significant. [Scale bars, 50 μm (A and C), 200 μm (B), 1 mm (D), and 20 μm (E).].

https://doi.org/10.7554/eLife.32358.003

Next, we analyzed the morphological differences between wild-type (WT) and Aldh1a1–/– eyes. In adult Aldh1a1–/– mice, the eyeballs displayed a loss of pigmentation in the dorsonasal area (Figure 1D). Hematoxylin and eosin (H and E) staining of the paraffin sections revealed that pigmentation was lost from the choroidal region, not from RPE cells (Figure 1E). The choroidal thickness was significantly less than that of the dorsal and ventral sides of WT eyes and the ventral side of Aldh1a1–/– eyes (Figure 1F). There were no obvious morphological differences in the thickness of the neural retina or in the length of the outer segments of photoreceptor cells (Figure 1G and H), as reported previously (Fan et al., 2003). These findings suggest that the choroidal vessels of Aldh1a1–/– mice are hypoplastic.

Aldh1a1–/– mice exhibited choroidal hypoplasia in the dorsal region

To characterize choroidal vascularization in the Aldh1a1–/– mice, we first immunostained choroidal flat-mounts for endomucin antibody, a specific marker for choriocapillaris, and isolectin B4, which mainly visualizes choroidal medium-sized/large blood vessels. Surprisingly, the Aldh1a1–/– mice exhibited choroidal hypoplasia in the dorsal region of the eyes (Figure 2A), although Aldh1a1 is normally expressed in the neural retina. In the dorsal and ventral regions of WT and the ventral region of Aldh1a1–/– eyes, the choriocapillaris was normal, with a dense mesh structure, whereas fewer choroidal vessels with more avascular areas were detected in the dorsal region of Aldh1a1–/– eyes (Figure 2B and C). We also performed whole-mount immunohistochemistry for ZO-1, a tight junction marker of RPE cells (Figure 2D). In the dorsal region of Aldh1a1–/– eyes, the vascular density was significantly lower than that in the other regions (Figure 2E), whereas there were no significant morphological differences in the localization of ZO-1 signals and RPE size between WT and Aldh1a1–/– eyes (Figure 2F). Taken together, these results demonstrate that Aldh1a1–/– eyes show choroidal hypoplasia without degeneration of the RPE in the dorsal region.

Hypoplasia at the dorsal side of choroid in Aldh1a1–/– mice.

(A) Representative choroidal flat-mount immunohistochemistry of adult (8-week-old) WT (left panel) and Aldh1a1–/– (right panel) posterior eyes stained for endomucin (Emcn, red) and isolectin B4 (IB4, green). (B) High-magnification Z-stack images of the choroidal flat-mounts collected from the dorsal and ventral areas of WT and Aldh1a1–/– mice (four white boxes shown in (A)) stained with Emcn (red) and IB4 (green). The dorsal image of Aldh1a1–/– mice indicates poor vascularization. (C) Orthogonal images of the Z-stacks (broken lines in (B)) showing breaks/holes in the dorsal region of Aldh1a1–/– eyes stained with Emcn antibody (red) and IB4 (green). (D) RPE flat-mount immunohistochemistry of WT and Aldh1a1–/– stained with ZO-1 (green). (E and F) Quantitative evaluation of the vascular density and size of RPE cells of adult WT and Aldh1a1–/– mice. Data represent the average ± SD; n = 8 per group. *p<0.05. N.S., not significant. [Scale bars, 500 μm (A), 50 μm (B–D).].

https://doi.org/10.7554/eLife.32358.006

To investigate further the morphological features of the choroidal vasculature and the neural retina/RPE/Bruch’s membrane complex, sections of WT and Aldh1a1–/– eyes were examined by transmission electron microscopy (TEM). The TEM images demonstrated that Aldh1a1–/– eyes exhibited apparently normal morphology of photoreceptor and RPE cells, despite the complete absence of choroidal pigmentation and the presence of a thin dorsal choroid layer (Figure 3A). Bruch’s membrane did not differ significantly between WT and Aldh1a1–/– eyes (Figure 3B). Remarkably, the choroidal vessels in the hypoplastic dorsal section of Aldh1a1–/– eyes showed fenestrations, a characteristic of capillaries, as did those in the dorsal and ventral regions of WT and the ventral region of Aldh1a1–/– eyes (Figure 3C). Taken together, these results suggest that the hypoplastic blood vessels of the Aldh1a1–/– eyes maintain the characteristics of the choriocapillaris, including an intact RPE and Bruch’s membrane.

Figure 3 with 1 supplement see all
Choroidal vessels of Aldh1a1–/– mice have features of capillaries.

(A) Electron micrographs of the dorsal and ventral areas from adult (8-week-old) WT and Aldh1a1–/– mice. The absence of choroidal pigmentation and a thin choroid layer is observed in the dorsal section of Aldh1a1–/– mice (yellow arrowhead). (B) Higher magnification of electron micrographs of retinal pigment epithelium (RPE) and Bruch’s membrane (BrM). There are no differences between WT and Aldh1a1–/– mice in the morphology of RPE and BrM. M, melanocyte; Mt, mitochondria; N, nucleus. (C) The higher magnification of electron micrography allows visualization of the fenestrations in the choriocapillaris. Representative fenestrated structures are indicated by red arrowheads. In the dorsal section of Aldh1a1–/– mice, the choriocapillaris shows fenestrations similar to those in the dorsal and ventral sides of WT and the ventral side of Aldh1a1–/– mice. [Scale bars, 10 μm (A), 0.5 μm (B), and 0.1 μm (C).].

https://doi.org/10.7554/eLife.32358.008

Aldh1a1−/− choroidal vessels exhibited an abnormal distribution of VEGF receptor subtypes

Although several reports have suggested that FMS-like tyrosine kinase 1 (Flt1, also called VEGF receptor 1) and kinase insert domain protein receptor (Kdr, also called VEGF receptor 2) are expressed in choroidal vessels (Zhao and Overbeek, 2001; Witmer et al., 2003; Cross et al., 2003; Saint-Geniez et al., 2006), the blood vessel type-specific receptor expression remains ambiguous. Therefore, we examined the regional distribution of Flt1-DsRed and Kdr-EGFP in choroidal vessels in WT and Aldh1a1−/− eyes by using Flt1-BAC-DsRed;Kdr-BAC-EGFP mice (Matsumoto et al., 2012). In the choroidal flat-mounts from adult WT mice, the choriocapillaris coexpressed both Flt1-DsRed and Kdr-EGFP, while the medium-sized/large choroidal vessels expressed only Flt1-DsRed (Figure 3—figure supplement 1A–D). In addition, more Flt1-DsRed appeared to be expressed in the medium-sized/large choroidal vessels than in the choriocapillaris (Figure 3—figure supplement 1D). In contrast, in the dorsal region of Aldh1a1–/– eyes, all the blood vessels associated with the hypoplastic choroid coexpressed both Flt1-DsRed and Kdr-EGFP (Figure 3—figure supplement 1B–D). Also, the level of expression of Flt1-tdsRed in the Aldh1a1–/– blood vessels appeared lower than that in the medium-sized/large choroidal vessels.

Aldh1a1-mediated RA production regulates dorsal choroidal vascular development by enhancing VEGF secretion from RPE cells

We next explored the developmental time point at which choroidal hypoplasia appears in Aldh1a1–/– mice. Since the formation of the choroidal vascular network starts at approximately E13.5 (Zhao and Overbeek, 2001; Marneros et al., 2005; Saint-Geniez et al., 2006), we compared the vascular development in the choroid of the eyes of WT and Aldh1a1–/– mice from E16.5 to postnatal (P)28 using a choroidal flat-mounts immunostained with antibodies against endomucin and ETS-related gene (ERG), which are enriched in endothelial cells (Figure 4A; Figure 4—figure supplement 1). In the dorsal region of Aldh1a1–/– eyes, endomucin immunostaining was patchy, indicating that choroidal hypoplasia was already detectable (Figure 4A and B). There were significantly fewer endothelial cells in the area than in WT eyes (Figure 4C).

Figure 4 with 2 supplements see all
Retinoic acids modulate VEGF secretion by RPE cells.

(A) Choroidal flat-mount immunohistochemistry of E16.5 WT and Aldh1a1–/– embryos stained for endomucin (Emcn, red) and ETS-related gene (ERG, green). Note that hypovascularization and fewer vascular endothelial cells were observed in the dorsal region of Aldh1a1–/– embryonic eyes. (B and C) Quantitative evaluation of the density of Emcn-positive vessels and the number of ERG-positive cells in E16.5 WT and Aldh1a1–/– embryos. Data represent the average ±SD; n = 4 per group. *p<0.05. (D) ELISA analysis of VEGF secreted from WT and Aldh1a1–/– RPE-choroid complex at E17.5 and P24. Data represent the average ±SD; n = 4 independent samples per group. *p<0.05. (E) In situ hybridization on E16.5 WT and Aldh1a1–/– eyes with the Vegfa probe (DIG-labeled, purple). Vegfa expression was reduced in the dorsal RPE cells of Aldh1a1–/– eyes (upper RPEs from red arrowhead). Right panels show the higher magnification images of left panels (four white boxes) and red square brackets indicate RPE layer. (F) Retinoic acid (RA)-dependent enhancement of VEGF secretion of human primary RPE cells evaluated by ELISA. Data are means three times ELISA determinations. ***p<0.001. (G) Choroidal flat-mount immunohistochemistry of P3 WT and Vitamin-A-deficient (VAD) mice stained for endomucin (Emcn, red). Dorsal choroidal hypoplasia was observed in VAD mice. (H) Quantitative evaluation of the vascular density of P3 WT and VAD mice. Data represent the average ±SD; n = 5–6 per group. *p<0.05. (I) Choroidal flat-mount of P3 Aldh1a1–/– and Aldh1a1–/– mice from a mother treated with RA by oral gavage between E10 and E16, immunostained with anti-endomucin antibody (Emcn, red) This RA treatment of the mother restored the choroidal vascularization in P3 Aldh1a1–/– pups. (J) Quantitative evaluation of the vascular density in Aldh1a1–/– mice and Aldh1a1–/– mice from a mother treated with RA. Data represent the average ± SD; n = 4 per group. *p<0.05. N.S., not significant. [Scale bars, 50 μm (A, right panels in E, (G and I), 200 μm (left panels in E)].

https://doi.org/10.7554/eLife.32358.010

Based on studies showing that VEGF secreted from RPE cells is necessary for choroidal vascular development (Sakamoto et al., 1995; Saint-Geniez et al., 2009; Le et al., 2010), we hypothesized that RAs synthesized by Aldh1a1 in the neural retina stimulate the RPE to enhance VEGF secretion to the basal side of the choroid. To test this hypothesis, we first evaluated the VEGF level in E17.5 and P24 RPE-choroid complexes by ELISA. VEGF levels in the Aldh1a1–/– RPE-choroid complex were significantly lower than those in WT eyes at both time points (Figure 4D). In addition, in situ hybridization revealed that Vegfa mRNA expression of the E16.5 Aldh1a1−/− RPE cells was reduced at the dorsal half where Aldh1a1 is expressed in the WT neural retina (Figure 4E, shown as arrowheads). Next, we tested RA-dependent VEGF secretion by the primary human RPE cells. We measured VEGF in the culture medium after RA treatment. As a result, RAs significantly enhanced VEGF secretion in a dose-dependent manner (Figure 4F).

Because RA is the active metabolite of vitamin A (Shams et al., 1993; Amengual et al., 2012), we generated vitamin A-deficient (VAD) mice by feeding a vitamin A-deficient diet (Chihara et al., 2013). At P3, VAD mice showed dorsal choroidal hypoplasia in the flat-mount analysis (Figure 4G). In the dorsal region of VAD eyes, the vascular density was significantly lower than that in the other regions such as the dorsal and ventral regions of WT and the ventral region of VAD eyes (Figure 4H). Also, RA administration to pregnant Aldh1a1–/– mice by oral gavage from E10 to E16 significantly suppressed the dorsal choroidal hypoplasia in their offspring (Figure 4I and J). These results indicate that RA controls dorsal choroidal vascular development and that dorsal choroidal hypoplasia in Aldh1a1–/– mice is causally related to a deficiency in RA synthesis.

These observations raised the question of whether Aldh1a3, another RA-producing enzyme in the mouse retina, also affects choroidal vascularization, because Aldh1a3 begins to be expressed in the entire RPE cell layer and in the ventral region of the neural retina from E10.5 (Matt et al., 2005). To test this possibility, we conditionally disrupted Aldh1a3 in the neural retina (floxed Aldh1a3 mice crossed with Pax6-α-Cre), however, in these mice, choriocapillaris development was found to be normal (Figure 4—figure supplement 2).

In conclusion, RAs derived from Aldh1a1 but not Aldh1a3 control choroidal vascular development by enhancing VEGF secretion from RPE cells.

Sox9 is downregulated in both the dorsal RPE and neural retina in the Aldh1a1−/− mice

We next investigated the transcription factors that directly enhance VEGF expression in dorsal RPE cells. Recently, it was reported that conditional disruption of Pax6 in the RPE at the early embryonic stages resulted in the loss of pigmentation (Raviv et al., 2014), and Pax6 conditional knockout mice in which the Vegfa promoter is synergistically transactivated by Pax6 and Sox9 exhibit choroidal hypoplasia (Cohen et al., 2016). Therefore, we performed immunohistochemistry to detect Pax6 and Sox9 in sections of embryonic WT and Aldh1a1–/– retinas. The intensity of Pax6 immunofluorescence in the dorsal Aldh1a1–/– RPE was slightly lower than WT at E12.5 and E14.5, but did not show a significant difference (Figure 5—figure supplement 1A–C). Next, we measured the developmental expression of Sox9 (Figure 5A and B). In the E12.5 WT neural retina, Sox9 was predominantly expressed in the dorsal region rather than in the ventral region, and the expression level at the dorsal region became comparable to the ventral region at E14.5. In Aldh1a1–/– neural retinas, Sox9 immunofluorescence in the dorsal region was reduced as much as that of the ventral region (Figure 5A and C). In the E12.5 WT RPE cells, there was no difference in Sox9 immunofluorescence between dorsal and ventral region, and the intensity increased at E14.5. In the E12.5 Aldh1a1–/– RPE cells, the immunofluorescence was comparable to WT, but was significantly lower than that of E14.5 WT (Figure 5B,D and E). These densitometry results suggest that Aldh1a1 enhances Sox9 expression in the dorsal neural retina and RPE cells during development.

Figure 5 with 1 supplement see all
Sox9 expression is downregulated in RPE cells of Aldh1a1–/– mice.

(A) Sox9 (green) and Aldh1a1 (red) staining of WT and Aldh1a1–/– eyes at E12.5. (B) Sox9 (white) expression in neural retina and RPE (white arrowheads) of WT and Aldh1a1–/– eyes at E12.5 (upper panels) and E14.5 (lower panels). Sox9 was strongly expressed in the dorsal neural retinas of E12.5 WT eyes, and downregulated in RPE cells of E14.5 Aldh1a1–/– eyes. [Scale bars, 50 μm (A and B).]. (C–E) Quantitative evaluation of the Sox9 immunofluorescence intensity in embryonic WT and Aldh1a1–/– eyes. Sox9 intensity was quantified in the E12.5 neural retina (C), E12.5 RPE cells (D), and E14.5 RPE cells (E). Data represent the average ±SD; n = 4 per group. *p<0.05. N.S., not significant. (F–K) Sox9 and Vegfa mRNA expression in primary RPE cells in response to RA exposure (F and G), Sox9 overexpression (H and I), and Sox9 knockdown (J and K). Relative expression of Sox9 mRNA (F, H, and J) and Vegfa mRNA (G, I, and K) normalized to β-actin mRNA are shown. Data are representative of three experiments. *p<0.05, **p<0.01, ***p<0.001. N.S., not significant.

https://doi.org/10.7554/eLife.32358.014

To determine whether Sox9 enhances VEGF in RPE cells in an RA-dependent manner, we measured Sox9 and Vegfa mRNA expression in primary RPE cells in response to RA exposure. The results showed that both Sox9 and Vegfa mRNAs (Figure 5F and G) were enhanced in an RA-dependent manner. To examine whether Sox9 regulates Vegfa in RPE cells, we performed overexpression and knockdown experiments. Overexpression of Sox9 by transient transfection of a pCAGIG-Sox9 vector resulted in upregulation of Vegfa mRNA (Figure 5H and I). In contrast, knockdown by transient transfection of Sox9 siRNA resulted in downregulation of Vegfa mRNA (Figure 5J and K). Taken together, these results strongly suggest that Sox9 enhanced by Aldh1a1-mediated RA upregulates Vegfa expression in RPE cells.

Conditional disruption of Sox9 in RPE cells phenocopies choroidal hypoplasia in the Aldh1a1–/– mice

We next explored further whether the Aldh1a1-driven Sox9 expression in the dorsal neural retina and RPE is involved in choroidal vascular development. To generate mice with selective deletion of Sox9 in the developing RPE or neural retina, mice with a conditional deletion of Sox9 (Sox9flox/flox; Kist et al., 2002) were mated with either Tyr-Cre (RPE-cKO of Sox9; Delmas et al., 2003) or Pax6-α-Cre (Retina-cKO of Sox9; Marquardt et al., 2001), respectively. In the Cre-reporter assay with R26R-H2B-mCherry mice (Abe et al., 2011) on an albino background, mCherry expression from E16.5 Tyr-Cre mice was observed in all RPE, although a few mCherry-positive cells were found in the neural retina (Figure 6—figure supplement 1A). Also, mCherry expression in Pax6-α-Cre mice was restricted to the dorsal and ventral portions of the neural retina as reported previously (Marquardt et al., 2001), but no mCherry-positive cells were found in the choroid (Figure 6—figure supplement 1B). In RPE-cKO of Sox9, we found less pigmentation in the dorsal region, and significantly poorer vasculature in the dorsal choroidal area than the other areas (Figure 6A and B), which phenocopied Aldh1a1–/– eyes. Interestingly, although Sox9 was disrupted in all RPE cells, the poor vasculature phenotype was restricted to the dorsal region, and did not appear in the ventral region (Figure 6B). Conversely, Retina-cKO of Sox9 showed no hypoplasia of the choroidal vasculature or lack of pigmentation (Figure 6A and B), indicating that Sox9 in the neural retina is not responsible for choroidal vascular development.

Figure 6 with 1 supplement see all
RPE-derived Sox9 controls choroidal vasculature development.

(A) Endomucin expression (Emcn, red) in 5-week-old choroidal flat-mounts. RPE-specific conditional knockout of Sox9 (Sox9 RPE-cKO) mimicked the choroidal hypoplasia seen in Aldh1a1–/– mice (middle panel), but retina-specific conditional knockout of Sox9 (Sox9 Retina-cKO) did not (right panel). Insets represent each choroidal flat-mount. (B) Quantitative evaluation of the vascular density of 5-week-old WT, Sox9 RPE-cKO, and Sox9 Retina-cKO. Data represent the average ±SD; n = 4–5 per group. *p<0.05. N.S., not significant. (C) Endomucin (Emcn, white) expression in P7 choroidal flat-mounts. The choroid was hypovascularized in Tg-CAG-mRFPfloxed-SOX9-ires-EGFP;Aldh1a1–/– eyes (middle panel). Choroidal hypoplasia in Aldh1a1–/– mice was rescued in Aldh1a1–/–;RPE-specific conditional overexpression of Sox9 (Aldh1a1–/–;Tyr-Cre, right panel). Insets represent reporter (mRFP and EGFP) expression. (D) Quantitative evaluation of the vascular density of P7 control, Aldh1a1–/–, and Aldh1a1–/–;Sox9 RPE-cOE. Data represent the average ±SD; n = 5 per group. *p<0.05. N.S., not significant. [Scale bars, 1 mm (insets in A), 50 μm (A and C), 500 μm (insets in C).]. (E) Model summarizing that neural retina-specific Aldh1a1 controls choroidal vascularization in the dorsal region. Retinoic acids (RA) synthesized by Aldh1a1 regulate Sox9 expression, and then Sox9 enhances VEGF secretion from the dorsal RPE cells.

https://doi.org/10.7554/eLife.32358.018

Next, we attempted to rescue the choroidal hypoplasia phenotypes imposed by Aldh1a1 deletion by restoring Sox9 signaling using Cre-inducible Sox9-overexpressing mice (RPE-cOE of Sox9, Kim et al., 2011). Tyr-Cre-induced overexpression of Sox9 in these mice significantly recovered the choroidal hypoplasia phenotypes in the dorsal region (Figure 6C and D). These results indicated that Sox9 expression enhanced by Aldh1a1 in developing RPE cells is critical for normal choroidal vasculature development.

Discussion

The molecular mechanism underlying the development and function of the neural retina has been well studied, but that responsible for the development of the choroidal vessels has not been thoroughly investigated, despite the pathophysiological importance of these structures in human eye disease. Here, we could immunohistochemically discriminate choriocapillaris from medium-sized/large blood vessels in Haller’s/Sattler’s layer in mouse eyes and then demonstrated that disruption of Aldh1a1, which is predominantly localized in the dorsal neural retina, resulted in choroidal hypoplasia in the dorsal portion of the eyes.

Previous developmental and electrophysiological analyses of the Aldh1a1–/– mouse has shown that the neural retina is apparently normal (Fan et al., 2003). By broadening the phenotyping scope, we identified vascular hypoplasia in the dorsal choroid of Aldh1a1–/– eyes. Aldh1a1 is expressed in dorsal retinal progenitor cells from E10.5 (Matt et al., 2005), and we observed no Aldh1a1 expression in either RPE cells or choroid in E12.5 and E17.5 mouse eyes (Figure 1A and B). CD31-positive endothelial cells and VEGF expression in mouse eyes can be observed in choroid and primitive RPE cells at E10.5 (Saint-Geniez et al., 2006), and choroidal vascular development is initiated at E11.5 when periocular vessels emerge following formation of the vascular network at around E13.5 (Zhao and Overbeek, 2001; Marneros et al., 2005; Saint-Geniez et al., 2006). The requirement for RPE-derived VEGF during embryonic development was reported using RPE-specific Vegfa-knockout mice (Vegfflox/flox;VMD2-Cre mice), in which Cre expression can be conditionally induced in RPE cells by doxycycline administration (Le et al., 2008). Although conditional disruption of Vegfa in RPE cells at E10 or E13 resulted in decreased choroidal vascular density, loss of the RPE-produced VEGF after E15 caused no significant defects in the choroidal vasculature (Le et al., 2010). Together with our demonstration that choroidal VEGF production is downregulated in Aldh1a1–/– embryos, these data suggest that Aldh1a1 expression during E10 to E13 in trans potentiates VEGF secretion from RPE cells to allow the development of normal choroidal vascularization. In addition, it should be noted that choroidal VEGF level was still reduced at P24 in the Aldh1a1–/– mice. Considering that severe choriocapillaris vasoconstriction occurs in adult tamoxifen-induced Vegfflox/flox;VMD2-Cre mice (Kurihara et al., 2012), it is possible that Aldh1a1 is also responsible for maintenance of the choriocapillaris.

We also demonstrated that RAs enhanced VEGF secretion from primary RPE cells and that VAD mice showed choroidal hypoplasia in the dorsal region. Also, RA administration suppressed the choroidal hypoplasia phenotype in Aldh1a1–/– mice. These results strongly suggest that Aldh1a1-mediated RA is responsible for normal choroidal vascular formation by controlling VEGF secretion from RPE cells. Moreover, we observed downregulation of Sox9 in the dorsal Aldh1a1–/– eyes. RPE-specific disruption of Sox9 replicated the phenotype of Aldh1a1–/– eyes, and Sox9 overexpression in the Aldh1a1–/– RPE cells rescued dorsal choroidal hypoplasia. Considering that in primary RPE cells RA exposure enhances both Sox9 and Vegfa expression and that overexpression and knockdown of Sox9 influences Vegfa expression, it is more likely that Aldh1a1-mediated RA production stimulates Sox9 expression in dorsal RPE cells and Sox9 then transactivates the Vegfa promoter. Retinoic acid receptor α (RARα) and retinoid X receptor α (RXRα) are expressed in RPE cells at an early embryonic stage (Mori et al., 2001). It is plausible that RARα and RXRα enhance Sox9 expression in dorsal RPE cells, although the precise molecular mechanism remains to be investigated.

We observed interactions between Aldh1a1, Sox9, and Vegfa genes; however, the results raised the question of why VAD mice and RPE-cKO of Sox9 show choroidal hypoplasia only in the dorsal region because the reduction of vitamin A and Sox9 also occurs in the ventral region. We do not yet have the answer, but one possibility is that ventral choroidal vascular development is governed by a different molecular pathway. For example, Aldh1a3, another RA-producing enzyme in the mouse retina, is expressed in RPE cells from E10.5 to E12.5 and in the ventral neural retina during embryogenesis (Matt et al., 2005), Figure 1B). However, conditional disruption of Aldh1a3 in the neural retina (floxed Aldh1a3 mice crossed with Pax6-α-Cre) did not result in either choroidal hypoplasia or loss of pigmentation in the ventral region (Figure 4—figure supplement 2), suggesting that Aldh1a3-mediated RAs from the ventral neural retina are unlikely to affect ventral choroidal development. In addition, we did not observe upregulation of Sox9 in the ventral RPE cells from E12.5 to E14.5 (Figure 5D and E), suggesting that Sox9 in the ventral RPE cells is unnecessary for ventral choroidal development. Given that mRFP and GFP fluorescent intensity seems different between dorsal and ventral RPEs in Tg-CAG-mRFPfloxed-SOX9-ires-EGFP mice (Figure 6C), the difference might be the result of different developmental or physiological characteristics.

In summary, we demonstrate a novel role of Aldh1a1 in dorsal choroidal vascular development. The results of the present study strongly suggest that RA production resulting from Aldh1a1 expression in the dorsal neural retina upregulates Sox9 expression in the dorsal RPE cells to enhance RPE-derived VEGF secretion (Figure 6E). In addition, our results suggest that embryonic RA exposure may regulate future dorsal choroidal vessels in the adult. Because vascular hypoplasia could result in hypoxia and impaired nutrient supply, which could affect adjacent RPE and photoreceptor cells, future studies should investigate the degeneration of RPE and photoreceptor cells. For example, age-related macular degeneration (AMD) is the leading cause of severe vision loss in humans (de Jong, 2006) and is caused by abnormalities of the subfoveal choriocapillaris. Aldh1a1−/− and VAD mice would be useful not only to study regional differences in choroidal development and maintenance but also to explore both a risk factor and a potential therapeutic target for AMD, because the RA concentration in the neural retina is affected by various environmental factors such as dietary intake of Vitamin A and light irradiation.

Materials and methods

Animals

All animal experiments were conducted with the approval of the RIKEN Center for Developmental Biology Ethics Committee (No. AH18-05-23). Timed pregnant CD1 and C57BL/6 mice were purchased from the Laboratory for Animal Resources and Genetic Engineering, RIKEN Center for Developmental Biology. Aldh1a1–/– mice (Fan et al., 2003) were purchased from Jackson Laboratory and mated with CD-1 mice to allow visualization of the choroidal vessels. Flt1-BAC-DsRed;Kdr-BAC-EGFP mice (Matsumoto et al., 2012) were mated with CD-1 or albinized Aldh1a1–/– mice. To conditionally disrupt Aldh1a3 in neural retinas, Aldh1a3-flox mice (Dupé et al., 2003) were mated with Pax6-α-Cre mice (Marquardt et al., 2001). To conditionally disrupt Sox9 in RPE cells and neural retinas, Sox9-flox mice (Kist et al., 2002) were mated with Tyr-Cre (Delmas et al., 2003) and Pax6-α-Cre mice, respectively. To conditionally overexpress Sox9 in RPE cells, CAG-mRFP1floxed-Sox9-ires-EGFP transgenic mice (Kim et al., 2011) were mated with Tyr-Cre mice. Albinized R26R-H2B-mCherry mice (Abe et al., 2011) were used for the Cre reporter assay of Tyr-Cre and Pax6-α-Cre mice (Figure 6—figure supplement 1A and B). PCR primers used for genotyping are listed in Table 1. VAD diet feeding and RA treatments were performed as described previously (Chihara et al., 2013; Fan et al., 2003). Eight-week-old CD-1 mice were fed a vitamin A-deficient (VAD) diet (AIN93G-D13110GC; Research Diets, New Brunswick, USA). After 16 weeks of feeding this diet, the animals were used as breeding pairs. Pregnant mice received the VAD diet until 3 days postpartum. All-trans-RA (Sigma) was suspended in ethanol (5 mg/ml) and then either diluted in sunflower oil (125 μg/ml) and administered by oral gavage to pregnant females (2 mg/kg of body weight) every 12 hr from E10 to E16.

Table 1
Oligonucleotides used in this study
https://doi.org/10.7554/eLife.32358.021
Mouse primersOligonucleotide sequences (5’−3’)
Aldh1a1-/-ForwardTGAGCAAATCCTCCACAGCCCTGTTC
ReverseCTGCTAAAGCGCATGCTCCAGACTG
Aldh1a3 floxForwardTCTCTGACCAGCTTTCCAACCTTCAG
ReverseCTCAAACCAGCACCACCTCCATATTG
Sox9 floxForwardTCAGCAAGACTCTGGGCAAGCTCT
ReverseCTCAAAATCTGAGCCACTCCCTC
Tyr-Cre, Pax6-α-CreForwardCCTGGAAAATGCTTCTGTCCGT
ReverseGTGTCCACATAGTCATTGGCAGAGTG
Sox9-Tg, Kdr-BAC-EGFPForwardAGCTGACCCTGAAGTTCATCTG
ReverseGTCGTCCTTGAAGAAGATGGTG
Flt1-BAC-tdsRedForwardGCTGCAGGCGCGGAGAAGGGCTCTC
ReverseCTTCACGTACACCTTGGAGC
IL2 (internal control)ForwardGCCTAGAAGATGAACTTGGACCTCTG
ReverseGTGGAAGGATTCACTTGCACAGTGAC
Human RPE primers
Sox9ForwardCGTACCCGCACTTGCACAAC
ReverseTCTCGCTCTCGTTCAGAAGTC
VegfaForwardTGCCCGCTGCTGTCTAAT
ReverseTCTCCGCTCTGAGCAAGG
β-actinForwardCCAACCGCGAGAAGATGA
ReverseCCAGAGGCGTACAGGGATAG

Immunohistochemistry

Request a detailed protocol

For cryosections, embryos were fixed in 4% paraformaldehyde (PFA), cryoprotected in 30% sucrose in phosphate-buffered saline (PBS) overnight at 4°C, embedded in OCT compound (Tissue Tec; Sakura Fine Technical, Japan), and sectioned at 10 μm using a cryostat (HM560; Thermo Fisher Scientific, Waltham, MA). Specimens were blocked with horse serum for 1 hr at room temperature and incubated with primary antibodies overnight at 4°C, followed by incubation with secondary antibodies for 1 hr at room temperature.

For flat-mount immunostaining of the RPE/choroid, the tissues were fixed with 4% PFA at room temperature for 30 min, and washed three times with PBS containing 0.5% Triton X-100 (PBST, Nacalai Tesque, Japan), incubated with primary antibodies overnight at 4°C, followed by incubation with secondary antibodies for 1 hr at room temperature (Zhu et al., 2012). The primary antibodies and dilutions used were as follows: goat anti-Aldh1a1 (1:1,000; Abcam), rabbit anti-Aldh1a3 (1:1,000; Sigma), rat anti-endomucin (1:400; Millipore), mouse anti-GS (1:1,000; Millipore), FITC-isolectin B4 (1:100; Vector Laboratories), rabbit anti-ZO-1 (1:100; Invitrogen), rabbit anti-GFP (1:1,000; Abcam), rabbit anti-ERG (1:400; Abcam), rabbit anti-Pax6 (1:200; BioLegend), and rabbit anti-Sox9 (1:200; Millipore).

Electron microscopic analysis

Request a detailed protocol

Mice were euthanized and perfused with ice-cold 4% PFA. Eyes were fixed with 2% glutaraldehyde in 4% PFA overnight. After washing with PBS, the eyes were postfixed with ice-cold 1% OsO4 in 0.1 M sodium cacodylate buffer, pH 7.3, for 2 hr. The samples were then rinsed with distilled water, stained with 0.5% aqueous uranyl acetate for 2 hr or overnight at room temperature, dehydrated with ethanol and propylene oxide, and embedded in Poly/Bed 812 (Polyscience). Ultrathin sections were cut, double-stained with uranyl acetate and Reynolds’ lead citrate, and viewed with a JEM 1010 or JEM 1400 transmission electron microscope (JEOL) at an accelerating voltage of 100 kV.

Analysis of images

Request a detailed protocol

Labeled cells were imaged using an LSM 780 confocal microscope (Carl Zeiss). Images were processed using Photoshop CS2 software (Adobe Systems). Choroidal vascular density was analyzed using ImageJ software (NIH) as described previously (Le et al., 2010). Fluorescence intensity was quantified using Zen Black software (version 2012; Carl Zeiss) according to the manufacturer’s instructions. All images shown are representative of three to eight independent experiments.

In situ hybridization

Request a detailed protocol

In situ hybridizations were performed as described previously (Acloque et al., 2008).

Enzyme-linked immunosorbent assay for VEGFA

Request a detailed protocol

The RPE/choroid was separated from the mouse eye and was homogenized and resolved in 100 μl RIPA buffer containing a protease inhibitor. The total VEGFA protein (pg) per mg of the extracted RPE/choroid tissue was calculated using ELISA development kits (R and D Systems, Minneapolis, MN) (Ueta et al., 2012).

Cell culture, RA treatment, and transfection

Request a detailed protocol

Primary human RPE cells (Lonza) were maintained in Dulbecco’s minimal essential medium (DMEM) supplemented with 5% fetal bovine serum (FBS) without antibiotics. Before RA treatment or transfection experiments, the RPE cells were suspended at 2 × 105 cells per well of a 6-well plate and cultured in DMEM supplemented with 5% FBS without antibiotics. After overnight culture, the medium was changed to RA-supplemented DMEM without FBS, and the cells were harvested after 24 hr incubation. For overexpression, 1 μg of pCAGIG and pCAGIG-Sox9 expression vectors (Addgene #11159) were transfected using Lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions. For knockdown, control and human Sox9 siRNAs (Santa Cruz) were transfected using siRNA Transfection Reagent (Santa Cruz) according to the manufacturer’s instructions. After 7 hr incubation, the medium was changed to DMEM without FBS (for overexpression) or DMEM plus 10% FBS (knockdown), and cells were harvested after another 24 hr incubation to quantify Sox9 and Vegfa mRNA by reverse transcription–quantitative polymerase chain reaction (RT-qPCR).

Reverse transcriptase–quantitative polymerase chain reaction

Request a detailed protocol

RT–qPCR was performed as described in a previous report (Sugita et al., 2015). Briefly, Sox9, Vegfa, and β-actin expression was analyzed in triplicate samples using a LightCycler model 480 (Roche Diagnostics), qPCR MasterMix (Roche Diagnostics), and highly specific Universal ProbeLibrary assays (Roche Diagnostics). The tested primers are described in Table 1, and the Universal Probes used were Probe#61 (Sox9 and Vegfa), and Probe#64 (β-actin). Relative mRNA expression was normalized to ΔΔCt of β-actin using relative quantification software (Roche Diagnostics). Results were reported as the relative expression of each molecule (ΔΔCt: control cells = 1).

Statistical analysis

Request a detailed protocol

All data are presented as the means ± SD. JMP Pro version 10.0.2 (SAS Institute Inc.) was used for statistical analysis, and data were analyzed using analysis of variance (ANOVA) followed by the Tukey–Kramer multiple-comparison test. When only two groups were compared, a two-sided Student’s t-test was used.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
    Permeability of the intraocular blood vessels
    1. A Bill
    2. P Törnquist
    3. A Alm
    (1980)
    Transactions of the ophthalmological societies of the United Kingdom 100:332–336.
  6. 6
  7. 7
    Pax6 role in the regulation of retinal pigmented epithelium maturation
    1. Y Cohen
    2. P Blinder
    3. M Idelson
    4. B Reubinoff
    5. S Itzkovitz
    6. R Ashery-Padan
    (2016)
    Investigative Ophthalmology & Visual Science 57:6055.
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
    Histology of the Human Eye
    1. MJ Hogan
    2. JA Alvarado
    3. JE Weddell
    (1971)
    M. J Hogan, J. A Alvarado, J. E Weddell, editors. WB Saunders Company.
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
    Aldehyde dehydrogenase is a positional marker in the retina
    1. P McCaffery
    2. P Tempst
    3. G Lara
    4. UC Dräger
    (1991)
    Development 112:693–702.
  29. 29
  30. 30
    Systematic immunolocalization of retinoid receptors in developing and adult mouse eyes.
    1. M Mori
    2. NB Ghyselinck
    3. P Chambon
    4. M Mark
    (2001)
    Investigative Ophthalmology & Visual Science 42:1312–1320.
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
    Effect of vitamin A deficiency on the adhesion of rat corneal epithelium and the basement membrane complex
    1. NB Shams
    2. LA Hanninen
    3. HV Chaves
    4. G Frangieh
    5. CV Reddy
    6. DT Azar
    7. KR Kenyon
    (1993)
    Investigative Ophthalmology & Visual Science 34:2646–2654.
  38. 38
  39. 39
  40. 40
  41. 41
    Regulation of choroid development by the retinal pigment epithelium
    1. S Zhao
    2. PA Overbeek
    (2001)
    Molecular Vision 7:277–282.
  42. 42
  43. 43
    Hypoxia-Inducible Factor (HIF)-1 regulatory pathway and its potential for therapeutic intervention in malignancy and ischemia.
    1. JE Ziello
    2. IS Jovin
    3. Y Huang
    (2007)
    The Yale Journal of Biology and Medicine 80:51–60.

Decision letter

  1. Jeremy Nathans
    Reviewing Editor; Johns Hopkins University School of Medicine, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Neural retina-specific Aldh1a1 controls dorsal choroidal vascular development via Sox9 expression in RPE cells" for consideration by eLife. Your article has been favorably evaluated by Marianne Bronner as the Senior Editor, Jeremy Nathans as the Reviewing Editor and three reviewers. The following individual involved in review of your submission has agreed to reveal her identity: Jing Chen (Reviewer #3).

As you will see, all of the reviewers were impressed with the importance of your work, but they also had a number of specific comments.

I am including the three reviews at the end of this letter, as there are a variety of specific and useful suggestions in them. Most of the comments are focused on greater documentation of the ocular phenotypes. One point that may be more challenging to address in a reasonable time frame is the specificity of the Tyr-Cre and Pax6-α-Cre lines. This can be done by crossing these Cre lines to a reporter (a nuclear localized reporter might be best). This point is summarized as item 4 in the comments of reviewer #2, which I am reproducing below.

"One concern relates to the foundations for the mechanistic conclusions of the paper. The model and main interpretations leading to the model are based on genetic manipulations using Cre-lines that the authors claim to be specific, i.e., Tyr-Cre, specific for RPE, and Pax6-α-Cre, specific for neural retina. However, since the specificities of these are critical for the conclusions of the paper, the authors need to better characterize them. Certainly, these lines have been published before but they were not described in the retina/choroid context. For instance, how specific is the Tyr-Cre line to the RPE? Since the choriocapillaris has melanin, it is possible that the manipulations of Sox9, either the KO or the overexpression rescue experiment, are not totally RPE-specific. This might affect the authors' interpretations in terms of the mechanism. Note that Sox9 is also expressed in the choroid (Figure 4). Is Pax6-α-Cre expressed in all neural retina?"

The challenge is made greater because the best design of this experiment would be to cross the Cre lines to reporters in an albino background (to avoid the difficulty of visualizing fluorescent reporters or immunofluorescent staining in heavily pigmented cells, i.e. RPE and choroid).

Reviewer #1:

In this study, the authors link neuroretinal Aldh1a1 to dorsal choroidal vascular development via regulation of RPE synthesis of VEGF. The authors found a compelling phenotype of hypoplasia choroidal with insufficient vascularization in the dorsal region in Aldh1a1 knockout mice. The authors found that Aldh1a1, an enzyme used for retinoic acid synthesis, is localized in the dorsal neuroretina from the very early embryonic stage to adult stage. By using in situ techniques and conditional knockout mice, the authors found that Aldh1a1-induced RA in neural retinal directs dorsal choroidal vascularization via RPE-derived VEGF production by increasing Sox9 expression in RPE. They provide a well-organized and logical set of experiments that provide a viable and interesting mechanism involving retinoic acid and Sox9 expression in the RPE. The data looks solid, with good quality images. The mechanisms governing choroidal vascular development remain poorly understood, so the current study provides a novel and important advance to the field.

Although the study is overall very interesting and insightful, one slight weakness is that the manuscript only provides correlation of Aldh1a1, Sox9, and VEGF expression, and lacks the mechanistic evidence on how Aldh1a1 regulates Sox9 and therefore regulates VEGF expression in RPE cells.

Specific comments:

Although Sox9 has been previously been reported to regulate VEGF expression, experiments showing the expression of VEGF in Sox9 overexpressing mice and knockout mice would strengthen the conclusion of the manuscript.

The authors suggest that Aldh1a1-derived RA from dorsal neural retina upregulate Sox9 and then VEGF in RPE. Can vitamin A diet rescue choroidal hyperplasia in Aldh1a1 knockout mice?

The vascularized area (%) from both dorsal and ventral choroid should be plotted in Figure 5A and 5B.

Reviewer #2:

This paper reveals an interesting observation of choroid development in the mouse. By analyzing the Aldh1a1-/- mice the authors found a peculiar phenotype of choroidal hypoplasia only in the dorsal retina. Interestingly the ventral choroid is not dependent on RA signaling, as Aldh1a3 conditional KO does not have an effect on choroid development. The authors suggest a model of the mechanism, where RA synthesized by Aldh1a1 in the dorsal domain of the neural retina induces Sox9 in the dorsal RPE that ultimately leads to increased secretion of VEGF from RPE, and subsequent regulation of choroid development. These findings contribute to our understanding of choroid development, suggesting that there are differential mechanisms for regulation of choroid development along the dorso-ventral (D/V) axis of the mouse retina.

1) The phenotype should be thoroughly investigated using comprehensive characterization of the choroid along thDV axis. EM analyses along the D/V axis in WT and Aldh1a1-/- at "critical" domains, such as dorsal, central and ventral, would improve the understanding of these processes.

2) The authors should do a better job explaining the phenotype in Aldh1a1-/-. How penetrant is the loss of pigmentation phenotype? Why is it patchy? And does it always appear in the same location or does it vary within the dorsal domain? In the retina shown in Figure 1D the unpigmented area seems to expand towards the central domain, outside of the Aldh1a1 expression region. Is this correct? Are the other phenotypes, i.e., choroid thickness reduction and decrease in vascular area, also patchy or are they observed throughout the whole dorsal region, correlating with the domain of Aldh1a1 expression? It would be helpful to correlate the appearance of phenotypes with the expression pattern of Aldh1a1. Along these lines, it is not clear what the authors think is the most affected layer of the choroid in the Aldh1a1-/-? The Haller's (large), Sattler's (medium) or choriocapillaris?

3) Due to the homeostatic regulation of RA signaling, the authors should check if the expression of Aldh1a3 and Cyp26s were altered in the Aldh1a1-/- and vice-versa.

4) One concern relates to the foundations for the mechanistic conclusions of the paper. The model and main interpretations leading to the model are based on genetic manipulations using Cre-lines that the authors claim to be specific, i.e., Tyr-Cre, specific for RPE, and Pax6-α-Cre, specific for neural retina. However, since the specificities of these are critical for the conclusions of the paper, the authors need to better characterize them. Certainly, these lines have been published before but they were not described in the retina/choroid context. For instance, how specific is the Tyr-Cre line to the RPE? Since the choriocapillaris has melanin, it is possible that the manipulations of Sox9, either the KO or the overexpression rescue experiment, are not totally RPE-specific. This might affect the authors' interpretations in terms of the mechanism. Note that Sox9 is also expressed in the choroid (Figure 4). Is Pax6-α-Cre expressed in all neural retina?

5) The phenotype in the Sox9 RPE-conditional KO should be quantified, as in the Aldh1a1-/-, by measuring the% of vascular area. Likewise, RPE cell integrity should be assessed as well as the reduced levels of VEGF produced in the RPE-choroid complexes of the Sox9 RPE-KO. Similarly, the phenotype in the rescue experiment of overexpression of RPE-Sox9 in the Aldh1a1-/- should be quantified in the same way. It seems that it is only rescued to some extent. And are the secretion levels of VEGF also rescued?

- The authors claim that endomucin is a specific marker of choriocapillaris (Results subsection “Aldh1a1-/- mice exhibited choroidal hypoplasia in the dorsal region”) but a previous paper shows that expression of endomucin extends to other layers of the choroid (Saint-Geniez et al., 2006). CD31 seems to be more specific to the choriocapillaris than endomucin (Saint-Geniez et al., 2006).

- At the end of the second paragraph in the Discussion, the authors mentioned that it is "possible that Adh1a1 is also responsible for maintenance of the choriocapillaris". Their data on the VAD-diet experiment strongly suggests that indeed this is the case, as the VAD-diet was used starting from P0 and not applied in embryonic stages by feeding the pregnant mice.

- In the Figure 2 legend (C) mentions "orthogonal images of the Z-stacks (…) showing thin choroid…". It is actually hard to see the thinning of the choroid on these images. There is certainly more "breaks/holes" suggesting more avascular areas.

- Also, it is not clear if the RPE flat mount immunohistochemistry of the Aldh1a1-/- shown in Figure 2D is underlying a region of reduced choroid vascular area. Again, the authors need to describe better the phenotype patterns of Aldh1a1-/-.

- Could not find reference "Cohen Y. et al., 2016".

- In the Discussion, can the authors comment on the fact that there is a fairly large distinct domain in the mouse retina that lacks expression of Aldh1a1 and Aldh1a3 in the central domain, clearly visible in the RARE-lacZ transgenic mouse line. How is the choroid structure in this central domain? Is the choroid development still dependent on RA signaling at this region?

Reviewer #3:

This manuscript by Goto and coworkers evaluated the role of retinal ALDH1a1, an enzyme that synthesizes retinoic acid, in mediating dorsal choroidal vascular development in mice. The authors showed that mice with retinal neuronal deficiency of ALDH1a1 exhibited decreased VEGF and Sox9 production in RPE (retinal pigment epithelium), and dorsal choroidal hypoplasia, which was phenocopied in mice with RA-deficient diet and also in mutant mice with RPE specific loss of Sox9. Based on these findings the authors concluded that neuronal ALDH1a1 affected dorsal choroidal vascular development via influencing retinoic acid-dependent, Sox9-mediated VEGF production in RPE. Overall this is a beautiful study that convincingly demonstrated the molecular interaction between retina and RPE via ALDH1a1/RA/Sox9/VEGF axis to control dorsal choroidal vascular development, which is of high interest for research in both basic development biology and also eye diseases with choroidal abnormalities such as age-related macular degeneration. This reviewer has just a few concerns as following.

1) There are patches of depigmentation in ALDH1a1 KO choroid/RPE (Figure 1D). Do ALDH1a1 KO eyes have any defects in melanin production in RPE melanosomes and choroidal melanocytes, in a potentially RA-dependent manner?

2) Reduced levels of Sox9 in ALDH1a1 KO eyes were demonstrated with immunohistochemistry. It would be helpful to confirm this finding with Western blot and/or RT-PCR in isolated retinas/RPE. In addition, does RA treatment increase Sox9 levels in RPE cell culture?

3) Figure 3G WT images also showed a small patch of choroidal thinning in top left part. Is this normal? Can the authors explain?

4) Figure 1CAldh1a1 staining does not completely overlap with GS staining, suggesting cells other than muller cells, e.g. photoreceptors, may also express Aldh1a1.

https://doi.org/10.7554/eLife.32358.029

Author response

As you will see, all of the reviewers were impressed with the importance of your work, but they also had a number of specific comments.

I am including the three reviews at the end of this letter, as there are a variety of specific and useful suggestions in them. Most of the comments are focused on greater documentation of the ocular phenotypes. One point that may be more challenging to address in a reasonable time frame is the specificity of the Tyr-Cre and Pax6-α-Cre lines. This can be done by crossing these Cre lines to a reporter (a nuclear localized reporter might be best). This point is summarized as item 4 in the comments of reviewer #2, which I am reproducing below.

"One concern relates to the foundations for the mechanistic conclusions of the paper. The model and main interpretations leading to the model are based on genetic manipulations using Cre-lines that the authors claim to be specific, i.e., Tyr-Cre, specific for RPE, and Pax6-α-Cre, specific for neural retina. However, since the specificities of these are critical for the conclusions of the paper, the authors need to better characterize them. Certainly, these lines have been published before but they were not described in the retina/choroid context. For instance, how specific is the Tyr-Cre line to the RPE? Since the choriocapillaris has melanin, it is possible that the manipulations of Sox9, either the KO or the overexpression rescue experiment, are not totally RPE-specific. This might affect the authors' interpretations in terms of the mechanism. Note that Sox9 is also expressed in the choroid (Figure 4). Is Pax6-α-Cre expressed in all neural retina?"

Thank you for your comments. We managed to obtain albinized Tyr-Cre and Pax6-α-Cre mice carrying the R26R-H2B-mCherry conditional reporter, which express nuclear mCherry in a Cre-dependent manner (Abe et al. 2011). We characterized mCherry expression at E16.5 because choroidal hypoplasia in Aldh1a1–/– mice was already observed at that time point, and we did not have enough time before the revision deadline to raise the pups.

In the Tyr-Cre; R26R-H2B-mCherry mice, mCherry expression was observed in all RPE cells. No mCherry expression was found in choroidal endothelial cells that were not co-immunostained with ERG and Sox9 antibodies. We found some mCherry expression in the neural retinas, but few cells were double-labeled with Sox9. These results indicate that Sox9 was conditionally disrupted in RPE cells in Sox9RPE-KO mouse eyes. However, in Pax6-α-Cre and R26R-H2B-mCherry mice, mCherry expression was observed in the dorsal and ventral portions of retinal progenitor cells as described previously (Marquardt et al., 2001), and no mCherry expression was found in RPE cells.

Taken together, these results indicate that Sox9 in RPE cells is responsible for dorsal choroidal vascularization. We added these data as Figure 6—figure supplement 1 and revised our manuscript as follows.

“To generate mice with selective deletion of Sox9 in the developing RPE or neural retina, mice with a conditional deletion of Sox9 (Sox9flox/flox; Kist et al., 2002) were mated with either Tyr-Cre (Sox9RPE-KO) mice (Delmas et al., 2003) or Pax6-α-Cre (Sox9Retina-KO) mice (Marquardt et al., 2001), respectively[…] Also, mCherry expression in Pax6-α-Cre mice was restricted to the dorsal and ventral portions of the neural retina as reported previously (Marquardt et al., 2001), but no mCherry-positive cells were found in the choroid (Figure 6—figure supplement 1B).”

“Albinized R26R-H2B-mCherry mice (Abe et al., 2011) were used for the Cre reporter assay of Tyr-Cre and Pax6-α-Cre mice (Figure 6—figure supplement 1A and B).”

The challenge is made greater because the best design of this experiment would be to cross the Cre lines to reporters in an albino background (to avoid the difficulty of visualizing fluorescent reporters or immunofluorescent staining in heavily pigmented cells, i.e. RPE and choroid).

Reviewer #1:

[…] Although the study is overall very interesting and insightful, one slight weakness is that the manuscript only provides correlation of Aldh1a1, Sox9, and VEGF expression, and lacks the mechanistic evidence on how Aldh1a1 regulates Sox9 and therefore regulates VEGF expression in RPE cells.

We thank the reviewer for pointing this out. To address this question, we used cultured primary human RPE.

To demonstrate mechanistic evidence for the link between Aldh1a1 and Sox9, we maintained the RPE cells with or without retinoic acids (RA), because RAs are synthesized by Aldh1a1 and vitamin A deficient diet and administration of RA regulate choroidal vascular formation (Figure 4G–J). First, by ELISA, we confirmed a significant increase in the level of VEGF protein in the culture medium in response to RA exposure. We also quantified Sox9 and Vegfa mRNAs, showing a significant RA-dependent increase. Subsequently, to examine whether Sox9 regulates Vegfa expression in RPE cells, we performed overexpression and knockdown experiments. Overexpression of Sox9 by transient transfection of pCAGIG-SOX9 vector resulted in upregulation of Vegfa mRNA. In contrast, knockdown by SOX9 siRNA transfection resulted in downregulation of Vegfa mRNA.

Based on these results, we concluded that Sox9 enhanced by Aldh1a1-derived RA upregulates Vegfa expression in RPE cells. We added this result as Figure 5F–K and revised our manuscript as follows. We also replaced the data in Figure 4F with that obtained from the primary RPE experiments.

“To determine whether Sox9 enhances VEGF in RPE cells in an RA-dependent manner, we measured Sox9 and Vegfa mRNA expression in primary RPE cells in response to RA exposure. […] Taken together, these results strongly suggest that Sox9 enhanced by Aldh1a1-mediated RA upregulates Vegfa expression in RPE cells.”

“Considering that in primary RPE cells RA exposure enhances both Sox9 and Vegfa expression and that overexpression and knockdown of Sox9 influences Vegfa expression, it is more likely that Aldh1a1-mediated RA production stimulates Sox9 expression in dorsal RPE cells and Sox9 then transactivates the Vegfa promoter.”

Specific comments:

Although Sox9 has been previously been reported to regulate VEGF expression, experiments showing the expression of VEGF in Sox9 overexpressing mice and knockout mice would strengthen the conclusion of the manuscript.

The authors suggest that Aldh1a1-derived RA from dorsal neural retina upregulate Sox9 and then VEGF in RPE. Can vitamin A diet rescue choroidal hyperplasia in Aldh1a1 knockout mice?

We thank the reviewer for these comments. We performed oral administration of RA to Aldh1a1-/- mice from E10 to E16 and found that the choroidal hypoplasia was rescued in P3 pups. We added the data in Figure 4I and J and revised our manuscript as follows.

“The level of VEGF in the RPE/choroid was significantly decreased in Aldh1a1–/– mice, and RA-dependent enhancement of VEGF was observed in in vitro-cultured primary RPE cells. An RA-deficient diet resulted in dorsal choroidal hypoplasia, and simple RA treatment of Aldh1a1–/– pregnant females suppressed choroid hypoplasia in their offspring.”

“Also, RA administration to Aldh1a1−/− mice by oral gavage of pregnant mothers from E10 to E16 significantly suppressed the dorsal choroidal hypoplasia (Figure 4I and J). These results indicate that RA controls dorsal choroidal vascular development and that dorsal choroidal hypoplasia in Aldh1a1−/− mice is causally related to a RA synthesis deficiency.”

The vascularized area (%) from both dorsal and ventral choroid should be plotted in Figure 5A and 5B.

We thank the reviewer for these comments. We added the appropriate graphs to Figure 6B and D and revised our manuscript as follows.

“In Sox9RPE-KO mice, we found less pigmentation in the dorsal region, and significantly poorer vasculature in the dorsal choroidal area than the other areas (Figure 6A and B),”.

“Tyr-Cre-induced overexpression of Sox9 in these mice significantly recovered the choroidal hypoplasia phenotypes in the dorsal region (Figure 6C and D).”

Reviewer #2:

[…] 1) The phenotype should be thoroughly investigated using comprehensive characterization of the choroid along thDV axis. EM analyses along the D/V axis in WT and Aldh1a1-/- at "critical" domains, such as dorsal, central and ventral, would improve the understanding of these processes.

Thank you for your suggestion. We performed EM analysis in 8-week-old WT and Aldh1a1–/– mouse eyes. As shown by the H&E staining (Figure 1E), EM sections of the dorsal Aldh1a1–/– eyes also showed a thinner choroidal layer and less pigmentation than those of the other experimental samples (dorsal and ventral WT eyes and ventral Aldh1a1–/– eyes). Despite the observation that the choroidal blood vessels in Aldh1a1–/– mice exhibited morphological and immunohistochemical abnormalities, they have fenestrations, which is a characteristic structure of choriocapillaris. We added these data as Figure 3A–C and revised our manuscript as follows.

“To investigate further the morphological features of the choroidal vasculature and the neural retina/RPE/Bruch’s membrane complex, sections of WT and Aldh1a1–/– eyes were examined by transmission electron microscopy (TEM). […] Taken together, these results suggest that the hypoplastic blood vessels of the Aldh1a1–/– eyes maintain the characteristics of the choriocapillaris, including an intact RPE and Bruch’s membrane.”

The central domain of choroidal flat-mount showed the same density of vascular area in both dorsal and ventral areas of WT and the ventral area of Aldh1a1–/– choroid. We refer the reviewer to Author response image 1 below, which shows the central part of the choroid in WT and Aldh1a1–/–mice.

Author response image 1
Representative choroidal flat-mount IHC of the central region of adult (8-week-old) WT (left panel) and Aldh1a1–/– (right panel) eyes immunostained with FITC-labeled isolectin B4 (IB4, green) and anti-endomucin antibody (Emcn, red).

There was no difference in choroidal vascular density between WT and Aldh1a1–/– mice.

https://doi.org/10.7554/eLife.32358.024

2) The authors should do a better job explaining the phenotype in Aldh1a1-/-. How penetrant is the loss of pigmentation phenotype? Why is it patchy? And does it always appear in the same location or does it vary within the dorsal domain? In the retina shown in Figure 1D the unpigmented area seems to expand towards the central domain, outside of the Aldh1a1 expression region. Is this correct?

We thank the reviewer for these comments. We first became interested in the phenotype. However, we could not identify an appropriate genotype–phenotype relationship. In the mating record, as shown in Author response image 2, not all Aldh1a1–/– mice had a loss-of-pigmentation phenotype. Also, some WT mice had the same phenotype as observed in Aldh1a1–/– mice. This observation suggests that multiple factors besides Aldh1a1 cause the loss-of-pigmentation phenotype. However, all Aldh1a1–/– mice have dorsal choroidal hypoplasia. For this reason, we just described the loss-of-pigmentation phenotype and did not proceed with further analyses.

Author response image 2
The frequency of appearance of pigmentation loss did not correspond with the principles of Mendelian inheritance.
https://doi.org/10.7554/eLife.32358.025

Are the other phenotypes, i.e., choroid thickness reduction and decrease in vascular area, also patchy or are they observed throughout the whole dorsal region, correlating with the domain of Aldh1a1 expression? It would be helpful to correlate the appearance of phenotypes with the expression pattern of Aldh1a1.

As the reviewer pointed out, choroidal hypoplasia in Aldh1a1–/– mice is observed in a dorsal portion of the choroid, where Aldh1a1 is to be expressed in the WT neural retina (see Figure 2A and Figure 1—figure supplement 1B). This observation indicates that neural retina-specific expression of Aldh1a1 controls dorsal choroidal vascular development.

Along these lines, it is not clear what the authors think is the most affected layer of the choroid in the Aldh1a1-/-? The Haller's (large), Sattler's (medium) or choriocapillaris?

Our understanding is that all types of choroidal blood vessels in the dorsal region are affected in Aldh1a1–/– mice. In fact, Figure 2B indicates that the hypoplasia phenotype is observed not only in choriocapillaris (endomucin-positive), but also in Haller/Sattler vessels (IB4-positive). Also, as shown in Figure 3—figure supplement 1, differential expression of VEGFR1 and VEGFR2 in choroidal vessels disappeared, suggesting that all types of dorsal choroidal vessels are immature in Aldh1a1–/– mice.

It might be considered that choriocapillaris is most affected because the more vascular region of choriocapillaris decreased more than the Haller/Sattler vessels. However, TEM images indicated that the choriocapillaris in Aldh1a1–/– dorsal choroid still had fenestrations, suggesting that the choriocapillaris retains its physiological characteristics.

3) Due to the homeostatic regulation of RA signaling, the authors should check if the expression of Aldh1a3 and Cyp26s were altered in the Aldh1a1-/- and vice-versa.

We thank the reviewer for these comments. Spatiotemporal analysis of RA production in Aldh1a1–/– mouse eyes has been performed previously using RARE-LacZ mice, which showed that the RA-free zone was expanded to include the dorsal area (Fan et al., 2003; Matt et al., 2005). These observations indicate that RA is not synthesized in the dorsal region in Aldh1a1–/– mice. Also, we performed immunohistochemistry (IHC) to check the compensatory effect of Aldh1a3 in E17.5 Aldh1a1–/– eyes. The results showed that the spatial distribution of expression of Aldh1a3 was not altered (Author response image 3), indicating no compensatory effect of Aldh1a3.

Author response image 3
Section IHCs of E17.5 mouse eyes labeled with anti-Aldh1a3 (green) and anti-Aldh1a1 (red) antibodies.

The region of Aldh1a3 expression did not expand.

https://doi.org/10.7554/eLife.32358.026

4) One concern relates to the foundations for the mechanistic conclusions of the paper. The model and main interpretations leading to the model are based on genetic manipulations using Cre-lines that the authors claim to be specific, i.e., Tyr-Cre, specific for RPE, and Pax6-α-Cre, specific for neural retina. However, since the specificities of these are critical for the conclusions of the paper, the authors need to better characterize them. Certainly, these lines have been published before but they were not described in the retina/choroid context. For instance, how specific is the Tyr-Cre line to the RPE? Since the choriocapillaris has melanin, it is possible that the manipulations of Sox9, either the KO or the overexpression rescue experiment, are not totally RPE-specific. This might affect the authors' interpretations in terms of the mechanism. Note that Sox9 is also expressed in the choroid (Figure 4). Is Pax6-α-Cre expressed in all neural retina?

We thank the reviewer for these comments. Please see the answer to the editor comments.

5) The phenotype in the Sox9 RPE-conditional KO should be quantified, as in the Aldh1a1-/-, by measuring the% of vascular area. Likewise, RPE cell integrity should be assessed as well as the reduced levels of VEGF produced in the RPE-choroid complexes of the Sox9 RPE-KO. Similarly, the phenotype in the rescue experiment of overexpression of RPE-Sox9 in the Aldh1a1-/- should be quantified in the same way. It seems that it is only rescued to some extent. And are the secretion levels of VEGF also rescued?

As you suggested, we added quantification data to Figure 6B and D and revised our manuscript (please see the last part of the answer for specific comments from reviewer #1). Since we were not able to obtain sufficient Sox9RPE-KO and Aldh1a1–/–;Sox9RPE-OE mice to compare the VEGF secretion levels, we instead quantified VEGF expression in cultured primary RPE cells in response to overexpression and knockdown of SOX9. Please see the answer to reviewer #1’s main comments.

- The authors claim that endomucin is a specific marker of choriocapillaris (Results subsection “Aldh1a1-/- mice exhibited choroidal hypoplasia in the dorsal region”) but a previous paper shows that expression of endomucin extends to other layers of the choroid (Saint-Geniez et al., 2006). CD31 seems to be more specific to the choriocapillaris than endomucin (Saint-Geniez et al., 2006).

Besides the rat anti-endomucin antibody and FITC-labeled isolectin B4 used in this study, we had checked the specificity of rat anti-CD-31 antibody before starting this study. Because these markers had been tested previously only in section immunohistochemistry, we performed flat-mount immunohistochemistry to figure out which marker explicitly discriminated choriocapillaris from choroidal medium-sized/large vessels. As shown in Author response image 4, in adult WT mice (8 weeks old), anti-CD-31 antibody mostly visualized medium-sized/large vessels of the choroid that were double-labeled with Isolectin B4, but hardly visualized choriocapillaris. In contrast, anti-endomucin antibody specifically labeled choriocapillaris that was not double-labeled with Isolectin B4. Based on these preliminary experiments, we chose the anti-endomucin antibody to visualize and quantify choriocapillaris vascularization.

Author response image 4
The representative dorsal region of choroidal flat-mounts of 8-week-old WT and Aldh1a1–/– mice stained with FITC-isolectin B4 (IB4, green), anti-endomucin (Emcn, red; upper panels), and anti-CD-31 (red; lower panels).

Emcn specifically visualized choriocapillaris, whereas CD-31 visualized medium-sized/large vessels that were IB4-positive.

https://doi.org/10.7554/eLife.32358.027

- At the end of the second paragraph in the Discussion, the authors mentioned that it is "possible that Adh1a1 is also responsible for maintenance of the choriocapillaris". Their data on the VAD-diet experiment strongly suggests that indeed this is the case, as the VAD-diet was used starting from P0 and not applied in embryonic stages by feeding the pregnant mice.

We assessed VAD pups using a procedure described previously (Chihara et al., 2013) and quantified the vascularization from choroidal flat-mounts immunostained with anti-endomucin antibody. The results showed that P3 VAD pups phenocopied the dorsal choroidal hypoplasia observed in Aldh1a1–/– mice, indicating that RA is responsible for dorsal choroidal vascularization. We added the results to Figure 4G and H and revised our manuscript as follows.

“At P3, VAD mice showed dorsal choroidal hypoplasia in the flat-mount analysis (Figure 4G).”

- In the Figure 2 legend (C) mentions "orthogonal images of the Z-stacks (…) showing thin choroid…". It is actually hard to see the thinning of the choroid on these images. There is certainly more "breaks/holes" suggesting more avascular areas.

Thank you for your suggestion, we revised the Figure 2C legend as follows.

“Orthogonal images of the Z-stacks (broken lines in (B)) showing breaks/holes in the dorsal region of Aldh1a1–/– eyes stained with Emcn antibody (red) and IB4 (green).”

- Also, it is not clear if the RPE flat mount immunohistochemistry of the Aldh1a1-/- shown in Figure 2D is underlying a region of reduced choroid vascular area. Again, the authors need to describe better the phenotype patterns of Aldh1a1-/-.

Author response image 5 shows X-Y images of the dorsal region of WT and Aldh1a1–/– choroidal flat mounts immunostained with anti-endomucin and anti-ZO-1 antibodies. Thinner choriocapillaris formation was observed in Aldh1a1–/– mouse eyes, but RPE morphology visualized by ZO-1 appeared normal.

Author response image 5
Representative dorsal choroidal flat-mounts of adult (8-week-old) WT (left panel) and Aldh1a1–/– (right panel) eyes immunostained with anti-endomucin (Emcn, red) and anti-ZO-1 (green) antibodies.

There is no difference in the size of RPE between WT and Aldh1a1–/– mice.

https://doi.org/10.7554/eLife.32358.028

- Could not find reference "Cohen Y. et al., 2016".

Thank you for the comment. We mentioned URL of ARVO abstract as follows.

“Cohen Y, Blinder P, Idelson M, Reubinoff B, Itzkovitz S, Ashery-Padan R. 2016. Pax6 role in the regulation of retinal pigmented epithelium maturation. Invest Ophthalmol Vis Sci 57: 6055. http://iovs.arvojournals.org/article.aspx?articleid=2563982.”

- In the Discussion, can the authors comment on the fact that there is a fairly large distinct domain in the mouse retina that lacks expression of Aldh1a1 and Aldh1a3 in the central domain, clearly visible in the RARE-lacZ transgenic mouse line. How is the choroid structure in this central domain? Is the choroid development still dependent on RA signaling at this region?

We thank the reviewer for these comments. Please see the response to the major comment by reviewer #2.

Reviewer #3:

[…] This reviewer has just a few concerns as following.

1) There are patches of depigmentation in ALDH1a1 KO choroid/RPE (Figure 1D). Do ALDH1a1 KO eyes have any defects in melanin production in RPE melanosomes and choroidal melanocytes, in a potentially RA-dependent manner?

Thank you for the comment. Please read the answer to reviewer #2’s major comment 2.

2) Reduced levels of Sox9 in ALDH1a1 KO eyes were demonstrated with immunohistochemistry. It would be helpful to confirm this finding with Western blot and/or RT-PCR in isolated retinas/RPE. In addition, does RA treatment increase Sox9 levels in RPE cell culture?

Please see the answer to reviewer #1’s main comment and reviewer #2’s major comment 5.

3) Figure 3G WT images also showed a small patch of choroidal thinning in top left part. Is this normal? Can the authors explain?

A small patch of the choroid is normal, which is why the choroidal vascular densities of WT are around 90% (Figures 2A and 6B).

4) Figure 1C Aldh1a1 staining does not completely overlap with GS staining, suggesting cells other than muller cells, e.g. photoreceptors, may also express Aldh1a1.

We repeated and revised the IHC shown in Figure 1C. We also added high magnification images of section ISH for Aldh1a1 mRNA as Figure 1—figure supplement 1A, showing that the ISH signals were restricted to the middle of the INL, where some Muller glia were localized. No ISH signals were found in the ONL.

https://doi.org/10.7554/eLife.32358.030

Article and author information

Author details

  1. So Goto

    1. Laboratory for Retinal Regeneration, RIKEN Center for Developmental Biology, Kobe, Japan
    2. Department of Ophthalmology, Osaka University Graduate School of Medicine, Suita, Japan
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4171-0781
  2. Akishi Onishi

    1. Laboratory for Retinal Regeneration, RIKEN Center for Developmental Biology, Kobe, Japan
    2. Kobe City Eye Hospital Research Center, Kobe, Japan
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    aonishi@cdb.riken.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2775-6567
  3. Kazuyo Misaki

    Ultrastructural Research Team, RIKEN Center for Life Science Technologies, Kobe, Japan
    Contribution
    Data curation, Formal analysis, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
  4. Shigenobu Yonemura

    Ultrastructural Research Team, RIKEN Center for Life Science Technologies, Kobe, Japan
    Contribution
    Data curation, Supervision, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
  5. Sunao Sugita

    1. Laboratory for Retinal Regeneration, RIKEN Center for Developmental Biology, Kobe, Japan
    2. Kobe City Eye Hospital Research Center, Kobe, Japan
    Contribution
    Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
  6. Hiromi Ito

    Laboratory for Retinal Regeneration, RIKEN Center for Developmental Biology, Kobe, Japan
    Contribution
    Resources, Data curation, Formal analysis, Validation, Investigation
    Competing interests
    No competing interests declared
  7. Yoko Ohigashi

    Laboratory for Retinal Regeneration, RIKEN Center for Developmental Biology, Kobe, Japan
    Contribution
    Resources, Data curation, Formal analysis, Validation, Investigation
    Competing interests
    No competing interests declared
  8. Masatsugu Ema

    Department of Stem Cells and Human Disease Models, Research Center for Animal Life Science, Shiga University of Medical Science, Otsu, Japan
    Contribution
    Resources
    Competing interests
    No competing interests declared
  9. Hirokazu Sakaguchi

    Department of Advanced Device Medicine, Osaka University Graduate School of Medicine, Suita, Japan
    Contribution
    Supervision
    Competing interests
    No competing interests declared
  10. Kohji Nishida

    Department of Ophthalmology, Osaka University Graduate School of Medicine, Suita, Japan
    Contribution
    Supervision, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
  11. Masayo Takahashi

    1. Laboratory for Retinal Regeneration, RIKEN Center for Developmental Biology, Kobe, Japan
    2. Kobe City Eye Hospital Research Center, Kobe, Japan
    Contribution
    Conceptualization, Supervision, Funding acquisition, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared

Funding

Japan Agency for Medical Research and Development (17bm0204002h0005)

  • Masayo Takahashi

Japan Society for the Promotion of Science (24687010)

  • Akishi Onishi

Japan Society for the Promotion of Science (17K11471)

  • Akishi Onishi

The Kato Memorial trust for NAMBYO Research

  • Akishi Onishi

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We received generous support from all members of the Takahashi Laboratory. We thank Laboratory for Animal Resources and Genetic Engineering, RIKEN Center for Developmental Biology for providing and maintaining mice, and RIKEN BioResource Center, Drs. R Morita (RIKEN CDB), H Fujiwara (RIKEN CDB), Y Kubota (Keio University), R Ashery-Padan (Tel Aviv University), H Akiyama (Gifu University), P Chambon (IGBMC), and S Mader (Université de Montréal) for providing mice. This work was supported in part by grants from Japan Agency for Medical Research and Development (grant number 17bm0204002h0005), JSPS KAKENHI (grant number 24687010 and 17K11471), and grants-in-aid of The Kato Memorial trust for NAMBYO Research. SG was financially supported from RIKEN by a Junior Research Associate (JRA) program for graduate students.

Ethics

Animal experimentation: All animal experiments were conducted with the approval of the RIKEN Center for Developmental Biology Ethics Committee (No. AH18-05-23)

Reviewing Editor

  1. Jeremy Nathans, Johns Hopkins University School of Medicine, United States

Publication history

  1. Received: September 28, 2017
  2. Accepted: March 5, 2018
  3. Version of Record published: April 3, 2018 (version 1)
  4. Version of Record updated: April 3, 2018 (version 2)

Copyright

© 2018, Goto et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,306
    Page views
  • 158
    Downloads
  • 4
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Developmental Biology
    2. Evolutionary Biology
    Kate A Rawlinson et al.
    Research Article
    1. Developmental Biology
    Ismael Lamas-Toranzo et al.
    Research Article