Cell Cycle–Specific and Cell Type–Specific Expression of Rb in The

Cell Cycle–Speciﬁc and Cell Type–Speciﬁc Expression of Rb in the Developing Human Retina

Thomas C. Lee,1,2,3 Dena Almeida,1,2 Nidia Claros,1,2 David H. Abramson,2,4 and David Cobrinik1,2

PURPOSE. To define the pattern of Rb expression relative to cell type and developmental stage in which Rb functions have not cycle position and cell type in the developing human retina. been defined.4,5 METHODS. Cryosections of fetal week 11-18 retinas were immu- The human retina begins its development at Fwk 5 and nostained for Rb and cell cycle– or cell type–specific markers. initially consists of proliferating retinal progenitor cells (RPCs) within a neuroblastic layer (NBL).6 RPC nuclei undergo a cell RESULTS. Rb was prominent in retinal progenitor cells (RPCs) cycle–dependent migration within the NBL, in which mitotic expressing the cyclin D1, cyclin A, and cytoplasmic cyclin B nuclei are positioned at the ventricular (outermost) layer, S markers of G1, S, and early to mid G2 phases, but not in RPCs phase nuclei are located in the middle NBL, and G1 and G2 expressing the phosphohistone H3 marker of late G2 and M. Rb nuclei migrate between these positions.7,8 Mitotic RPCs may was not detected in the earliest postmitotic ganglion, ama- give rise to additional RPCs or to postmitotic precursors of crine, horizontal, and bipolar cell precursors migrating away each of the mature retinal cell types, with the fate influenced from the ventricular layer, but was detected as such cells by cell-intrinsic competence states and by extrinsic cues.9 underwent further differentiation. Among photoreceptors, Rb Retinal development proceeds in a central to peripheral direc- was not detected in the earliest RXR␥(ϩ) cone precursors or in ϩ tion, such that proliferation ceases in the central retina by Fwk the earliest Nrl( ) rod precursors, but subsequently rose to 12 and in the far periphery by Fwk 30.10,11 high levels in cones and to low levels in rods. Rb was promi- Histopathological analyses have provided clues to the cell nent at the time when Mu¨ller glia exit the cell cycle and was Kip1 type in which Rb may suppress tumorigenesis. Retinoblasto- generally expressed in a pattern complementary to p27 . mas often contain differentiated cells that express markers of CONCLUSIONS. Rb exhibits cell cycle–specific expression in cones and, to a lesser extent, markers of rods and Mu¨ller glia, 12,13 RPCs, with loss in late G2-M and restoration in G1.Rbis but not markers of other retinal cells. Similarly, retinoblas- re-expressed after postmitotic ganglion, amacrine, horizontal, toma cell lines express proteins that are specific to cones and and bipolar cell precursors migrate away from the ventricular rods but not other cell types.14,15 In tumors, cells that express layer; after the appearance of early cone and rod markers; but glial markers adjoin and may form concurrent with cells that coinciding with Mu¨ller glia cell cycle withdrawal. The results express photoreceptor markers.12,13 These findings suggest suggest that Rb does not mediate the initial proliferative arrest that retinoblastomas may derive from a cell that is committed of retinal neurons, but may indirectly induce arrest in RPCs or to photoreceptor and glial differentiation or from an uncom- maintain an arrest in postmitotic precursors. (Invest Ophthal- mitted cell that is restricted to such differentiation in the tumor mol Vis Sci. 2006;47:5590–5598) DOI:10.1167/iovs.06-0063 environment. Among mammals, retinoblastoma is almost exclusively a etinoblastomas generally arise within the first 4 years of life human disease, as it has been diagnosed in only two individual Rand have been detected prenatally as early as fetal week animals.16,17 Moreover, in mice, loss of Rb does not predispose (Fwk) 21.1,2 The retinoblastoma gene (RB1) and protein (Rb) to retinal tumors, and combined loss of Rb and either of the are crucial to the suppression of retinoblastoma, as RB1 muta- related proteins p107 or p130 results in tumors with amacrine tion, deletion, or silencing occurs in all such tumors, and but not photoreceptor differentiation.18–21 In this study, we germline mutations predispose to numerous retinoblastoma sought to define the spatiotemporal pattern of Rb expression foci.3 Although these observations indicate that Rb suppresses in the developing human retina, as a means of identifying cell tumorigenesis during early retinal development, the retinal cell types and developmental stages in which Rb may suppress tumorigenesis.

1 2 From the Margaret M. Dyson Vision Research Institute, and De- METHODS partment of Ophthalmology, Weill Medical College of Cornell Univer- 4 sity, New York, New York; and the Ophthalmic Oncology Service, Human fetal eyes were obtained under protocols approved by the Department of Surgery, Memorial Sloan-Kettering Cancer Center, New Weill Medical College IRB and were studied in compliance with the York, New York. 3Current address: Division of Ophthalmology, Department of Sur- tenets of the Declaration of Helsinki. Fetal age was determined by gery, Childrens Hospital Los Angeles, Los Angeles, California. femur length. The cornea and lens were removed and the eyes ﬁxed Supported by The C.V. Starr Foundation, The Fred Gluck Founda- overnight at 4°C in 4% paraformaldehyde in PBS (PFA/PBS), incubated tion, The New York Community Trust, and The Fund for Ophthalmic in 30% sucrose/PBS overnight at 4°C, embedded in one part 30% Knowledge. sucrose/PBS and two parts optimal cutting temperature compound Submitted for publication January 20, 2006; revised August 1, (OCT; Miles Laboratories, Elkhart, IN), frozen, and sectioned at 10 to 2006; accepted October 11, 2006. 15 ␮m. Disclosure: T.C. Lee, None; D. Almeida, None; N. Claros, None; For Rb staining, sections were postﬁxed in 4% PFA/PBS for 5 D.H. Abramson, None; D. Cobrinik, None minutes, washed in 0.5 M NaCl/20 mM Tris (pH 8.0; TBS), treated with The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertise- 1 mM EDTA/TBS for 5 minutes, washed with TBS, incubated in 0.1% ment” in accordance with 18 U.S.C. §1734 solely to indicate this fact. H2O2 for 15 minutes, washed in TBS, treated with ABC kit reagent A Corresponding author: David Cobrinik, Margaret M. Dyson Vision (Vector Laboratories, Burlingame, CA) in TBS for 15 minutes, washed Research Institute, Weill Medical College of Cornell University, 1300 in TBS, treated with ABC kit reagent B (Vector Laboratories) in TBS for York Ave, LC303, New York, NY 10021; [email protected]. 15 minutes, washed in TBS, blocked and permeabilized in 5% horse

Downloaded from iovs.arvojournals.org on 09/24/2021 IOVS, December 2006, Vol. 47, No. 12 Expression of Rb in the Developing Human Retina 5591

serumϩ2% human serum in TBS (block 1), with 0.1% Triton-X-100 for line), apart from rare cells at the interface of the outer and 20 minutes, incubated in Rb antibody G3-245 (1:200; BD-PharMingen, middle NBL (Figs. 3A–C, arrows). Notably, cyclin D1(ϩ) cells San Diego, CA) in block 1ϩ0.05% Tween-20 overnight at 4°C, washed often had weak Rb signals relative to the surrounding cells (Figs. in TBS, incubated in biotinylated horse anti-mouse antibody (1:135; 3A–C, compare cells marked by arrowheads with those marked Vector Laboratories) in block 1 for 30 minutes, washed in TBS, washed by asterisks). in 0.1 M sodium bicarbonate and 0.15 M NaCl (balanced saline), Cells expressing cyclin A and cytoplasmic cyclin B were incubated with FITC-conjugated streptavidin (1:100; Vector Laborato- detected in both the middle and outer NBL (Figs. 3E, 3H). In ries) in balanced saline, and washed with TBS. For costaining, sections the middle NBL, nearly all cyclin A(ϩ) and cytoplasmic cyclin were incubated in 5% goat serum and 2% human serum in TBS (block B(ϩ) cells had strong Rb signals (Figs. 3D–I, arrowheads; Sup- 2) for 20 minutes, incubated overnight with primary antibody (Sup- plementary Fig. S2 online). The higher Rb signal in cyclin A(ϩ) plementary Information, online at http://www.iovs.org/cgi/content/ and cytoplasmic cyclin B(ϩ) cells, relative to cyclin D1(ϩ) full/47/12/5590/DC1) in block 2, washed in TBS, incubated for 30 cells, suggests that Rb expression increases as RPCs progress ϩ minutes in block 2 with secondary antibody (Supplementary Informa- from G1 into S and G2. In contrast, many cyclin A( ) and cyclin tion online), and washed in TBS. Sections were stained with (4Ј,6Ј- B(ϩ) cells at the ventricular layer had little or no Rb (Figs. diamino-2-phenylindole (DAPI) and analyzed by indirect immunofluo- 3D–I, arrows), although some cells had punctate Rb signals rescence or confocal microscopy. (Figs. 3D–I, thick arrows) similar to those in Ki67(ϩ) cells in similar positions (Fig. 2). To define further the cell-cycle position in which Rb RESULTS signal declines, we costained for Rb and phosphorylated Distinct Rb Expression Patterns in Peripheral and histone H3 (PH3), which is specific to late G2 and mitotic 27 ϩ Central Retina prophase, metaphase, and anaphase. PH3( ) cells had either a low, barely detectable nuclear Rb signal (not We initially evaluated Rb expression at Fwk 18. At this age, shown), a punctate Rb signal outside the PH-3(ϩ) region there is extensive proliferation in the peripheral retina, but no (Figs. 3J–L, thick arrows), or no detectable Rb in the case of 11 proliferation in the central retina. In the periphery, Rb was anaphase cells (Figs. 3J–L, thin arrow). This observation prominent in nuclei throughout the middle NBL and in occa- suggests that Rb levels decline at or before the time when sional nuclei in the outer NBL (Figs. 1A, 1B). A similar pattern histone H3 is phosphorylated in late G2. was evident at Fwks 12 and 14, although the band of Rb(ϩ) After undergoing mitosis at the ventricular layer, early G1- nuclei was positioned closer to the outer limiting membrane phase RPC nuclei and postmitotic retinal precursor nuclei (OLM), due to the lack of rod precursors in the outer NBL at migrate through the outer NBL in the inward (vitread) direc- these ages (Supplementary Fig. S1; all Supplementary Figures tion. However, all Rb(ϩ) cells in the outer NBL expressed are online at http://www.iovs.org/cgi/content/full/47/12/ cyclin A (Figs. 3D–F and data not shown), which is specific to 5590/DC1). In contrast, in the central retina, Rb was promi- 25 Ϫ S, G2, and early M. The lack of Rb in cyclin A( ) cells implies nent in outer nuclear layer cells having the appearance of that Rb was not appreciably expressed in early G0 or G1 nuclei cones and in inner nuclear layer (INL) positions typical of during their vitread migration. Mu¨ller cell nuclei, but was detected at far lower levels else- In sum, Rb was present in cyclin D1(ϩ) cells in the middle where (Figs. 1C, 1D). Thus, the Rb expression pattern differed NBL, was detected at higher levels in cyclin A(ϩ) and cytoplas- in the highly proliferating peripheral retina compared with the mic cyclin B(ϩ) cells in the middle and outer NBL, and was postmitotic central retina and exhibited cell type–specific dif- diminished in cyclin A(ϩ), cyclin B(ϩ), and PH3(ϩ) cells at the ferences among postmitotic cells. ventricular layer (Fig. 3M). The results imply that Rb expression increases during progression from G into S and G , Rb Expression in RPCs 1 2 declines in late G2 and M coinciding with a punctate distribu- In the peripheral retina, Rb(ϩ) nuclei were in positions that tion pattern, and does not reaccumulate until RPC nuclei re- are characteristic of RPCs (Figs. 1A, 1B). To determine whether turn to the middle NBL. Rb is expressed in such cells, we costained for Rb and the ϩ proliferation marker Ki67. Rb was prominent in all Ki67( ) Rb Expression in Postmitotic Retinal Cells nuclei in the middle and outer NBL (Figs. 2A–D, arrowheads), but was diminished or undetectable in Ki67(ϩ) nuclei at the After undergoing mitosis at the ventricular layer, RPCs give rise ventricular layer (Fig. 2, arrows, and Supplementary Fig. S2 to additional RPCs or to postmitotic retinal precursors. To online). Moreover, Rb had a punctate distribution outside of define the Rb expression pattern during the genesis of postmi- the Ki67(ϩ) region in many ventricular layer cells (Figs. 2B–G; totic cells, we stained for Rb and early markers of the different thick arrows). retinal cell types. The diminished Rb signal in ventricular layer RPCs suggests Rb in Ganglion Cell Precursors. Ganglion cells are pro- 11 that Rb may be decreased in the G2- or M-phase nuclei that are duced in the retinal far periphery at Fwk 12. Islet-1 is an early typically located in this position. To examine Rb in different marker of such cells and is first expressed at the ventricular cell cycle phases, retinas were costained for Rb and a series of layer soon after mitosis.28,29 Islet-1 was highly expressed in

cyclin proteins. Cyclin D1 was used as a marker of cells in G1, precursors migrating through the NBL as well as in the incip- 22 as it is expressed in G1 and degraded in S, and was not ient ganglion cell layer (GCL; Figs. 4A, 4B). Rb was not de- 23,24 ϩ detected in outer NBL positions typical of G2 and M. Cyclin tected in migrating Islet-1( ) cells in the outer NBL (Figs. 25 ϩ A was used as a marker of S, G2, and mitotic prophase, and 4A–C, arrows), but was prominent in Islet-1( ) cells at the cyclin B1 as a marker that accumulates in the cytoplasm in G2, interface of the NBL and GCL (Figs. 4A–C, arrowheads). Rb was enters the nucleus on entry into prophase, and is degraded in also detected at low levels in a subset of cells in the incipient anaphase.26 Consistent with these assignments, cyclins D1 and GCL and in the more mature GCL at later ages (Figs. 1, 4), A were detected in distinct cell populations (Supplementary suggesting that Rb levels may decline during ganglion cell Fig. S3 online). maturation. To deﬁne further the timing of Rb expression, we As in other species, cyclin D1 was detected in the middle costained for Rb and Brn-3b, which is speciﬁc to postmitotic NBL (Fig. 3B). Rb was detected in nearly all cyclin D1(ϩ) ganglion cells.30,31 Rb was prominent in Brn-3b(ϩ) cells at the cells (Figs. 3A–C, arrowheads; Supplementary Fig. S2 on- interface between the NBL and GCL (Figs. 4D–F, arrowheads),

Downloaded from iovs.arvojournals.org on 09/24/2021 5592 Lee et al. IOVS, December 2006, Vol. 47, No. 12

FIGURE 1. Rb expression in the peripheral and central retina. Fwk 18 peripheral (A, B) and central (C, D) retina stained for Rb (green) and DAPI (blue). OLM signal represents background that is independent of primary antibody. OLM, outer limiting membrane; NBL, neuroblastic layer; IPL, inner plexiform layer; GCL, ganglion cell layer; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer.

suggesting that Rb accumulates at or before the time when Brn-3b is ﬁrst expressed. Rb in Horizontal and Amacrine Cell Precursors. Hori- zontal and amacrine cell precursors initially migrate from the ventricular layer to the inner NBL.7,32 However, as Rb was not detected in postmitotic nuclei in the outer NBL, Rb evidently did not accumulate during the early stages of this process. After migrating to the inner NBL, horizontal cell precursors migrate in the outward direction and express the Prox1 and Lim1 proteins.33–35 At Fwk 11, Prox1 was detected at low levels in Ki67(ϩ) RPCs, but at far higher levels in distinctively shaped, Ki67(Ϫ) nuclei in the central and midperipheral retina (data not shown). The time of appearance and position of the strongly Prox1(ϩ) nuclei was consistent with that of horizontal cell precursors.35 At this age, Rb was not detected in the most peripheral, and thus most recently formed, Prox1(ϩ) nuclei

FIGURE 3. Cell cycle–speciﬁc Rb expression in RPCs. (A–L) Fwk 18 peripheral retina costained for Rb (green) and either cyclin D1 (red, FIGURE 2. Rb expression in a subset of RPCs. Fwk 18 peripheral retina A–C), cyclin A (red, D–F), cyclin B1 (red, G–I), or phosphohistone H3 costained for Rb (green), Ki67 (red), and DAPI (blue) and examined by (red, J–L), and examined by confocal microscopy. Arrowheads: cells confocal microscopy. The boxed region in (A) is shown in (B–D). A that costain for Rb and the cell cycle marker. Arrows: cells that separate section is shown at higher magniﬁcation in (E–G). Arrow- strongly stain for the cell cycle marker but not Rb. Thick arrows: cells heads: nuclei that stain for both Rb and Ki67. Arrows: cells that stain with punctate Rb staining. (A–C, ✱) Cells that strongly stain for Rb but for Ki67 but little or no Rb. Thick arrows: punctate Rb staining outside not cyclin D1. (M) Rb expression relative to cell cycle phase and of Ki67(ϩ) region. VL, ventricular layer; O-NBL, outer neuroblastic nuclear position in the NBL. Green shading: subcellular localization layer; M-NBL, middle neuroblastic layer. and intensity of Rb signal.

Downloaded from iovs.arvojournals.org on 09/24/2021 IOVS, December 2006, Vol. 47, No. 12 Expression of Rb in the Developing Human Retina 5593

FIGURE 4. Rb expression in ganglion cell precursors. Fwk 12 peripheral retina costained for Rb (green) and either Islet-1 (red, A–C)or Brn-3b (red, D–F). Arrowheads: nuclei that costain for Rb and either Islet-1 or Brn-3b. Arrows: nuclei that stain for Islet-1 but not Rb. iGCL, incipient ganglion cell layer.

(Figs. 5A–C), but was detected at low levels in occasional and mice is the retinoid X receptor ␥ (RXR␥).36–38 At Fwk 12, Prox1(ϩ) cells in more central positions (Figs. 5D–F, arrow- Rb was detected in only a subset of the RXR␥(ϩ) nuclei in the heads). Similarly, at Fwk 18, Rb was not detected in the most fovea (Figs. 6A–C, arrowheads), and in none of the younger peripheral Prox1(ϩ) horizontal cell precursors (Figs. 5G–I), RXR␥(ϩ) nuclei at more peripheral positions (data not shown). but was highly expressed in the more mature horizontal cells At Fwk 15, Rb was detected in all the RXR␥(ϩ) nuclei in the in the central retina (Figs. 5J–L). Rb expression in horizontal central retina, but not in those in the periphery (Figs. 6D–I). cell precursors was confirmed by costaining with Lim1 (Sup- Similarly, Rb was not detected in peripheral RXR␥(ϩ) nuclei at plementary Fig. S4 online). The expression of Rb in the more Fwk 18 (data not shown). The increased prevalence of Rb in central, but not the peripheral, Prox1(ϩ) nuclei implies that RXR␥(ϩ) cells in more developed retinal regions implies that Rb is expressed after the initial expression of Prox1 during Rb is expressed after RXR␥ appears. horizontal cell maturation. Two other early cone markers are thyroid hormone recep- To date, we have not identified a marker that is specific to tor ␤2 (TR␤2)39,40 and interphotoreceptor retinoid binding postmitotic amacrine precursors and compatible with Rb stain- protein (IRBP), which is specific to cone precursors in the ing (see Supplementary Information, http://www.iovs.org/cgi/ early fovea.10 At Fwk 12, TR␤2 and IRBP were detected only in content/full/47/12/5590/DC1). However, Rb was detected at the fovea, and Rb was detected in all TR␤2(ϩ) and all IRBP(ϩ) low levels in the inner INL where amacrine precursors accu- cells (Supplementary Fig. S5 online). mulate (Fig. 1B), consistent with Rb’s having minimal expres- Rb in Rod Precursors. Rod precursors are initially posi- sion during amacrine cell maturation. tioned below the outer limiting membrane. As additional rod Rb in Cone Precursors. Rb was highly expressed in foveal nuclei accumulate, the earlier born rod nuclei are displaced to cone precursors at Fwk 18 (Fig. 1C). To define the timing of Rb progressively deeper layers.7,32 To evaluate Rb expression dur- expression in this lineage, we stained for Rb and several cone ing rod development, we costained for Rb and Nrl, which is precursor markers. One of the earliest cone markers in chicks the earliest known rod-specific protein.10,41 At Fwk 14, the

FIGURE 5. Rb expression in horizontal cell precursors. Fwk 11 peripheral (A–C) and central (D–F) retina or Fwk 18 peripheral (G–I) and central (J–L) retina costained for Prox1 (red) and Rb (green). Arrows: nuclei that stain for Prox1 but not Rb. Arrow- heads: nuclei that costain for Prox1 and Rb. The positions of strongly Prox1(ϩ) cells are consistent with that of horizontal cell precursors at all ages and with that of a second, uncharacterized population in the Fwk 18 INL.

Downloaded from iovs.arvojournals.org on 09/24/2021 5594 Lee et al. IOVS, December 2006, Vol. 47, No. 12

FIGURE 6. Rb expression in cone precursors. Fwk 12 (A–C) or Fwk 15 (D–I) central (A–F) or peripheral (G–I) retina costained for Rb (green), RXR␥ (red), and DAPI (blue). Ar- rows: cells that stain for RXR␥ but not Rb. Arrowheads: cells that costain for both RXR␥ and Rb.

most peripheral (and hence, least mature) Nrl(ϩ) nuclei were (CRALBP)43 (Figs. 9A–C), confirming their Mu¨ller cell identity. positioned below the outer limiting membrane (Figs. 7A–C, Mu¨ller glia resemble RPCs in their morphology, gene ex- arrows). Rb was not detected in these peripheral Nrl(ϩ) nu- pression profile, and ability to re-enter the cell cycle.44–46 In clei, but was detected in more mature Nrl(ϩ) nuclei at more keeping with this relationship, Mu¨ller glia express cyclin D1 central and basal positions (Figs. 7A–C, arrowhead). This phe- for a limited time while they undergo cell cycle exit.46 We nomenon was more clear in the Fwk 18 periphery, where Rb confirmed that cyclin D1 persisted in the Fwk 18 central retina, was not evident in Nrl(ϩ) nuclei near the outer limiting mem- after proliferation had ceased (Fig. 9E and data not shown). Rb brane, but was detected in more mature Nrl(ϩ) nuclei at more was prominent in all cyclin D1(ϩ) nuclei (Figs. 9D–F), indicat- basal positions (Figs. 7D–G, arrowheads). Similarly, Rb was ing that it is highly expressed as Mu¨ller glia exit the cell cycle. detected in Nrl(ϩ) nuclei in the parafoveal region, albeit at far Ϫ lower levels than in Nrl( ) cones (Supplementary Fig. S6 Complementary Expression of Rb and p27Kip1 online). Notably, Nrl was expressed at higher levels in older, basally positioned Rb(ϩ) nuclei than in the younger, apically The lack of Rb in early postmitotic retinal neurons suggested positioned nuclei that lacked Rb expression (Figs. 7D–G). that other cell cycle regulators mediate the proliferative arrest Rb in Bipolar Cell Precursors. Bipolar cells comprise of such cells. Because p27Kip1 has an important antiprolifera- much of the outer aspect of the INL and can be identified by tive role in murine retina,23,24 we examined its expression their prominent nuclear expression of Chx10.42 Chx10 is also relative to that of Rb. expressed in RPCs, but at far lower levels and without the In the Fwk 18 peripheral retina, p27 and Rb were generally central-to-peripheral distribution of bipolar cell precursors. expressed in a complementary pattern (Figs. 10A–C). A similar At Fwk 15, the most peripheral, and thus youngest, strongly complementarity was evident at Fwk 14 (data not shown). Chx10(ϩ) bipolar cell nuclei were interspersed within the NBL RPCs comprising most of the the NBL prominently expressed and did not have detectable Rb (Figs. 8A–C). However, in the Rb but not p27, whereas rod precursors in the outer NBL and central retina, Chx10(ϩ) nuclei formed a distinct layer, and in amacrine and ganglion cell precursors in the inner retina ex- most cases had moderate, above background Rb expression pressed p27 but little or no Rb. Rare cells expressed both Rb (Figs. 8D–F), indicating that Rb was first detected in bipolar and p27 (Figs. 10A–C, arrowheads), particularly in outer NBL cells after Chx10 accumulation. positions where Rb and Nrl were coexpressed (see Figs. 7D–F), Rb in Mu¨ller Glia. Rb was highly expressed in INL posi- and in middle and inner NBL positions where Rb and Prox1 tions that are characteristic of Mu¨ller glia nuclei at Fwk 18 were coexpressed (see Fig. 5). Thus, Rb may be superimposed (Figs. 1C, 9F). This distribution was also evident at Fwk 15, over p27 expression during the differentiation of these cells. although the pattern was less obvious due to higher Rb expres- Rb and p27 were also expressed in a complementary man- sion in surrounding INL cells (Fig. 9C). Nonetheless, at Fwk 15 ner in the central retina. p27 was highly expressed and Rb had each of the strongly Rb(ϩ) INL nuclei was surrounded by the minimal expression in positions typical of ganglion cells, am- Mu¨ller cell–specific marker, cellular retinal–binding protein acrine cells, horizontal cells, bipolar cells, and rods, whereas

FIGURE 7. Rb expression in rod precursors. Fwk 14 midperipheral retina (A–C) and Fwk 18 peripheral retina (D–G) costained for Rb (green) and Nrl (red). The boxed region in (D)is shown at higher magniﬁcation in (E– G). Arrows: cells that stain for Nrl but not Rb. Arrowheads: cells that costain for Rb and Nrl. The direc- tions toward the retinal periphery or center are indicated for (A–C).

Downloaded from iovs.arvojournals.org on 09/24/2021 IOVS, December 2006, Vol. 47, No. 12 Expression of Rb in the Developing Human Retina 5595

DISCUSSION

This study aimed to deﬁne the pattern of Rb expression in the human fetal retina, from gestational weeks 11 to 18. This interval is characterized by RPC proliferation, production of each of the retinal cell types, and transition from a proliferative to a postmitotic state. The data showed that Rb is expressed in a cell cycle–speciﬁc manner in RPCs and accumulates at vari- ous times and levels in the different types of postmitotic retinal precursors.

Cell Cycle–Speciﬁc Expression of Rb in RPCs Others had previously shown that Rb is expressed in RPCs in the developing mouse and human retina.47–49 By using cyclins and phosphorylated histone H3 (PH3) as markers, we deduced

that Rb is expressed in G1 nuclei in the middle NBL, increases expression in S and early G2 nuclei traversing the NBL, declines to undetectable levels from late G2 to anaphase at the ventricular layer, and does not reaccumulate until G1 nuclei return to the middle NBL (Fig. 3M). This cell cycle-speciﬁc pattern in FIGURE 8. Rb expression in bipolar cell precursors. Fwk 15 peripheral human RPCs is consistent with the general decline in Rb (A–C) or central (D–F) retina costained for Chx10 (red) and Rb 47 (green). Arrows: nuclei that stain for Chx10 but not Rb. Arrowheads: expression in M- and G1-phase RPCs in mice, and further nuclei that costain for Rb and Chx10. indicates that Rb is both dramatically and consistently diminished during RPC mitosis. The cyclical expression of Rb in RPCs contrasts with Rb’s little or no p27 was detected in the strongly Rb(ϩ)Mu¨ller glia constitutive expression throughout the cell cycle in diverse and cones (Figs. 10G–I, arrows). cell types that were previously analyzed with the same anti- To assess the relationship between Rb and p27 expression body.50 In those other cell types, Rb associated with condens- in earlier cones and Mu¨ller cells, we examined a transition zone ing chromatin in prophase, redistributed to the cytoplasm on having the most peripheral (and hence least mature) Rb(ϩ) nuclear envelope breakdown in metaphase, and returned to cone precursors and only rare Ki67(ϩ) cells. In contrast to the daughter nuclei on nuclear envelope restoration in telo- central retina, Rb(ϩ) cone precursors in this region had high phase.50 However, in RPCs, Rb declined before prophase levels of p27 (Figs. 10D–F, arrowheads). This observation im- (when PH3 becomes prominent27) and was not restored to

plies that cone precursors transition from a state in which they telophase, or G1, nuclei until such nuclei had migrated to the express both Rb and p27 to a state in which they express middle NBL.

predominantly Rb. However, we found no evidence of such a The mechanism by which Rb expression declines in late G2 transition in Mu¨ller glia. Rather, the less mature glial cells in the and M is unknown. However, it was notable that some cells transition zone generally had no detectable p27 (Figs. 10D–F), near the ventricular layer had a punctate Rb signal immediately whereas those in the central retina had low but detectable p27 outside of their Ki67(ϩ) or PH3(ϩ) regions (Figs. 2B–G). As Rb (Figs. 10G–I). was not detected in metaphase or later M-phase nuclei, the

FIGURE 9. Rb expression in Mu¨ller glia. Fwk 15 (A–C) or Fwk 18 (D–F) fovea costained for Rb (green), and either CRALBP (red, A–C) or cyclin D1 (red, D–F). Arrowheads: cells that costain for Rb and either CRALBP or cyclin D1.

Downloaded from iovs.arvojournals.org on 09/24/2021 5596 Lee et al. IOVS, December 2006, Vol. 47, No. 12

Kip1 Kip1 FIGURE 10. Complementary expression of Rb and p27 . Fwk 18 retina costained for Rb (green) and p27 (red). Images of peripheral (A–C), transition zone (D–F), and central (G–I) retina are from the same section and were imaged in identical fashion. Arrowheads: nuclei that costain for Rb and p27. Arrows: nuclei that stain strongly for Rb but not p27.

punctate Rb signal may designate sites to which Rb translo- p27 and thereby provides novel insights into the potential role cates before its degradation. Of interest, an earlier study of of Rb in retinal development. osteosarcoma cells showed that high Cdk activity can induce Unexpectedly, Rb was not detected at the time when post- phosphorylation of Rb residue 567 and may thereby induce Rb mitotic retinal neurons were born, but accumulated well after cytoplasmic translocation and degradation.51 Thus, the dramat- the terminal mitosis. For example, Rb was not detected in

ically increased Cdk activity at the G2–M transition may elicit postmitotic ganglion, horizontal, amacrine, and bipolar cell cytoplasmic translocation and degradation of most of the Rb in precursors migrating away from the ventricular layer, but was RPCs, even though it apparently does not do so in other cell detected at the time when ganglion cells first expressed Brn-3b types. and after horizontal cells first expressed Prox1. Similarly, Rb It seems plausible that Rb’s cell cycle–specific expression was not detected in nascent Chx10(ϩ) bipolar precursors, but serves a developmental role. However, the cyclical loss of Rb in was detected in more mature Chx10(ϩ) cells. This evidence the M phase seems unlikely to regulate RPC fate, as Rb was suggests that Rb does not directly mediate cell cycle with- lacking in M-phase cells both in the periphery (where RPCs drawal or early differentiation events in postmitotic retinal mainly produce more RPCs) and in more central positions precursors. (where RPCs mainly produce postmitotic precursors). Alterna- The timing of Rb expression in developing photoreceptors tively, the decline in Rb expression before mitosis may pre- was of particular interest, given that retinoblastoma may derive clude Rb from inhibiting proliferation early in the subsequent from a photoreceptor-directed cell. Surprisingly, Rb was not cell cycle. Our detection of occasional cyclin D1(ϩ), Rb(Ϫ) detected at the time when rod precursors first expressed Nrl in cells in the outer NBL (Figs. 3A–C) suggests that Rb may be the outer retinal layers, but was detected in more mature rod re-expressed only after RPCs acquire cyclin D1-dependent ki- nuclei at more basal positions. This suggests that in rod pre- nase activity that can suppress Rb function. cursors, Rb mediates neither cell cycle withdrawal nor the initial expression of Nrl. That Rb does not mediate a rod precursor arrest is consistent with the lack of rod precursor Delayed Expression of Rb in Postmitotic 49 Retinal Precursors proliferation in Rb-deficient mouse retinas. Nevertheless, the lack of Rb in early Nrl(ϩ) precursors was unexpected, given Our analyses show that Rb is expressed in all cell types in the that Rb-deficient mouse retinas had impaired rod differentia- developing human retina. This result is unsurprising in light of tion and decreased Nrl mRNA.49 Our finding that Nrl was more Rb’s widespread expression and diverse functions in many highly expressed in Rb(ϩ) versus Rb(Ϫ) cells (Figs. 7E–G) developmental contexts.52 Moreover, Rb has been detected in suggests that Rb may promote Nrl expression during late but cell types other than amacrine cells in the early mouse retina.48 not early rod differentiation. However, the present study also defined the timing of Rb As in rods, Rb was not detected in the earliest cones. expression relative to that of cell type–specific markers and Whereas cone precursors are evident at Fwk 9,10 Rb was

Downloaded from iovs.arvojournals.org on 09/24/2021 IOVS, December 2006, Vol. 47, No. 12 Expression of Rb in the Developing Human Retina 5597

detected in only a subset of such precursors in the central Acknowledgments retina at Fwk 12 and was detected in central but not midperipheral cones at Fwks 15 and 18. Although we cannot deter- The authors thank John Saari, Thomas Jessel, Barbara Wiggert, Hemant mine the precise length of the delay, the evidence suggests that Khanna, Anand Swaroop, Lily Ng, and Douglas Forrest for gifts of foveal cones may arrest for 1 to 3 weeks before having detect- antibodies used in this study. able Rb expression. In contrast to rods, Rb rose to high levels 48 in early cones. Spencer et al. similarly detected high Rb References expression in adult cones. We extend their finding by showing that cone precursors prominently express Rb in the fetal pe- 1. Abramson DH, Schefler AC. Update on retinoblastoma. Retina. riod when retinoblastoma tumorigenesis begins. 2004;24:828–848. In contrast to the delayed Rb expression in postmitotic 2. Maat-Kievit JA, Oepkes D, Hartwig NG, Vermeij-Keers C, van Kamp retinal neurons, Rb was highly expressed in Mu¨ller glia at the IL, van de Kamp JJ. A large retinoblastoma detected in a fetus at 21 time of cell cycle exit. As Mu¨ller glia resemble quiescent weeks of gestation. Prenat Diagn. 1993;13:377–384. RPCs,44–46 Rb’s expression may be governed by the same 3. Lohmann DR, Gallie BL. Retinoblastoma: revisiting the model pro- process in the two cell types. Notably, p27 was lacking at the totype of inherited cancer. Am J Med Genet C Semin Med Genet. time when Mu¨ller glia exit the cell cycle. This is consistent 2004;129:23–28. with the lack of p27 in RPCs, but contrasts with the prominent 4. Dyer MA, Bremner R. The search for the retinoblastoma cell of origin. Nat Rev Cancer. 2005;5:91–101. expression of p27 in postmitotic retinal neurons. 5. Cobrinik D. Pocket proteins and cell cycle control. Oncogene. 2005;24:2796–2809. Relationship of Rb Expression 6. Barishak Y. Embryology of the Eye and Its Adnexa. New York: to Tumor Suppression Karger; 2001. 7. Sidman RL. Histogenesis of mouse retina studied with thymidine- In mice, Rb was found to be highly expressed in each of the H3. In: Smelser G, ed. The Structure of the Eye. New York: tissues in which it has crucial developmental roles.53,54 Thus, Academic Press;1961:487–505. although a protein’s expression level does not necessarily re- 8. Donovan SL, Dyer MA. Regulation of proliferation during central flect its function, it is of interest to consider whether Rb might nervous system development. Semin Cell Dev Biol. 2005;16:407– suppress retinoblastoma in the retinal cell types in which it is 421. expressed at high levels. 9. Livesey FJ, Cepko CL. Vertebrate neural cell-fate determination: Rb was most highly expressed in RPCs. However, Rb seems lessons from the retina. Nat Rev Neurosci. 2001;2:109–118. unlikely to directly induce a proliferative arrest in such cells, as 10. O’Brien KM, Schulte D, Hendrickson AE. Expression of photore- RPCs are not known to exit the cell cycle without differenti- ceptor-associated molecules during human fetal eye development. Mol Vis. 2003;9:401–409. ating into one of the postmitotic retinal cell types. Moreover, 11. Provis JM, van Driel D, Billson FA, Russell P. Development of the our data argue that Rb does not accumulate in RPCs in prepa- human retina: patterns of cell distribution and redistribution in the ration for a role after such cells divide, because Rb expression ganglion cell layer. J Comp Neurol. 1985;233:429–451. declined in late G2/M and was not rapidly restored to postmi- 12. Gonzalez-Fernandez F, Lopes MB, Garcia-Fernandez JM, et al. Ex- totic cells. Thus, if Rb were to suppress tumorigenesis through pression of developmentally defined retinal phenotypes in the its expression in RPCs, it would appear to do so by indirectly histogenesis of retinoblastoma. Am J Pathol. 1992;141:363–375. eliciting cell cycle exit and differentiation after a subsequent 13. Nork TM, Schwartz TL, Doshi HM, Millecchia LL. Retinoblastoma: mitosis. Notably, in cortical neurogenesis, progenitor cells un- cell of origin. Arch Ophthalmol. 1995;113:791–802. dergoing neurogenic divisions have longer cell cycles than 14. Bogenmann E, Lochrie MA, Simon MI. Cone cell-specific genes those undergoing proliferative divisions, and artificially length- expressed in retinoblastoma. Science. 1988;240:76–78.

ening G1 suffices to induce neurogenic differentiation after an 15. Di Polo A, Farber DB. Rod photoreceptor-specific gene expression intervening mitosis.55,56 Thus, Rb might indirectly promote in human retinoblastoma cells. Proc Natl Acad Sci USA. 1995;92: 4016–4020. neurogenic differentiation by lengthening G1 or through other 16. Fugaro MN, Kiupel M, Montiani-Ferreira F, Hawkins JF, Janovitz effects that precede Rb’s degradation in late G2 and M. Rb was also highly expressed at the time when Mu¨ller glia EB. Retinoblastoma in the eye of a llama (Llama glama). Vet Oph- thalmol. 2005;8:287–290. exit the cell cycle. Because p27 was not concurrently ex- 17. Syed NA, Nork TM, Poulsen GL, Riis RC, George C, Albert DM. pressed, it is tempting to infer that Rb may be needed to Retinoblastoma in a dog. Arch Ophthalmol. 1997;115:758–763. suppress Mu¨ller cell proliferation. However, arguing against 18. Robanus-Maandag E, Dekker M, van der Valk M, et al. p107 is a this possibility are the lack of Mu¨ller cell markers in prolifer- 12,13 suppressor of retinoblastoma development in pRb-deficient mice. ating retinoblastoma cells and evidence that adult Mu¨ller Genes Dev. 1998;12:1599–1609. 48 glia express high levels of the Rb-related p107 and p130. 19. MacPherson D, Sage J, Kim T, Ho D, McLaughlin ME, Jacks T. Cell In cone precursors, Rb rose to high levels concurrent with type-specific effects of Rb deletion in the murine retina. Genes a decline in p27. This reciprocal increase in Rb and loss of p27 Dev. 2004;18:1681–1694. was unique to cone precursors at the ages that we examined. 20. Zhang J, Schweers B, Dyer MA. The First Knockout Mouse Model Notably, Rb is prominent in cone precursors and mature cones of Retinoblastoma. Cell Cycle. 2004;3:952–959. in the human retina, but in neither developing nor adult cones 21. Chen D, Livne-bar I, Vanderluit JL, Slack RS, Agochiya M, Bremner in mice (this study and Refs. 21,48). Moreover, Rb’s expression R. Cell-specific effects of RB or RB/p107 loss on retinal develop- in human cone precursors correlates with its role as a retino- ment implicate an intrinsically death-resistant cell-of-origin in ret- blastoma suppressor in humans and with the predominant inoblastoma. Cancer Cell. 2004;5:539–551. cone phenotype of differentiated retinoblastoma cells.12,13 22. Guo Y, Yang K, Harwalkar J, et al. Phosphorylation of cyclin D1 at Thr 286 during S phase leads to its proteasomal degradation and These observations suggest that Rb may have a human-specific allows efficient DNA synthesis. Oncogene. 2005;24:2599–2612. function in cone precursors, that coincides with the decline in 23. Dyer MA, Cepko CL. p27Kip1 and p57Kip2 regulate proliferation p27. Nevertheless, additional studies are needed to address in distinct retinal progenitor cell populations. J Neurosci. 2001; whether Rb suppresses retinoblastoma in cone precursors, in 21:4259–4271. other cells that have high Rb levels such as RPCs or Mu¨ller glia, 24. Levine EM, Close J, Fero M, Ostrovsky A, Reh TA. p27(Kip1) or in still other cell types that had less prominent Rb expres- regulates cell cycle withdrawal of late multipotent progenitor cells sion. in the mammalian retina. Dev Biol. 2000;219:299–314.

Downloaded from iovs.arvojournals.org on 09/24/2021 5598 Lee et al. IOVS, December 2006, Vol. 47, No. 12

25. Fung TK, Poon RY. A roller coaster ride with the mitotic cyclins. alpha and N-terminal variant beta receptors. Development. 1992; Semin Cell Dev Biol. 2005;16:335–342. 114:39–47. 26. Pines J, Hunter T. Human cyclins A and B1 are differentially 41. Swain PK, Hicks D, Mears AJ, et al. Multiple phosphorylated located in the cell and undergo cell cycle-dependent nuclear trans- isoforms of NRL are expressed in rod photoreceptors. J Biol Chem. port. J Cell Biol. 1991;115:1–17. 2001;276:36824–36830. 27. Hendzel MJ, Wei Y, Mancini MA, et al. Mitosis-speciﬁc phosphor- 42. Liu IS, Chen JD, Ploder L, et al. Developmental expression of a ylation of histone H3 initiates primarily within pericentromeric novel murine homeobox gene (Chx10): evidence for roles in heterochromatin during G2 and spreads in an ordered fashion determination of the neuroretina and inner nuclear layer. Neuron. coincident with mitotic chromosome condensation. Chromo- 1994;13:377–393. soma. 1997;106:348–360. 43. Bunt-Milam AH, Saari JC. Immunocytochemical localization of two 28. Rachel RA, Dolen G, Hayes NL, et al. Spatiotemporal features of retinoid-binding proteins in vertebrate retina. J Cell Biol. 1983;97: early neuronogenesis differ in wild-type and albino mouse retina. 703–712. J Neurosci. 2002;22:4249–4263. 44. Close JL, Gumuscu B, Reh TA. Retinal neurons regulate prolifera- 29. Galli-Resta L, Resta G, Tan SS, Reese BE. Mosaics of islet-1-express- tion of postnatal progenitors and Mu¨ller glia in the rat retina via ing amacrine cells assembled by short-range cellular interactions. TGF beta signaling. Development. 2005;132:3015–3026. J Neurosci. 1997;17:7831–7838. 45. Walcott JC, Provis JM. Mu¨ller cells express the neuronal progenitor 30. Xiang M. Requirement for Brn-3b in early differentiation of post- cell marker nestin in both differentiated and undifferentiated hu- mitotic retinal ganglion cell precursors. Dev Biol. 1998;197:155– man foetal retina. Clin Exp Ophthalmol. 2003;31:246–249. 169. 46. Blackshaw S, Harpavat S, Trimarchi J, et al. Genomic analysis of 31. Xiang M, Zhou L, Peng YW, Eddy RL, Shows TB, Nathans J. Brn-3b: mouse retinal development. PLoS Biol. 2004;2:E247. a POU domain gene expressed in a subset of retinal ganglion cells. 47. Donovan S, Schweers B, Martins R, Johnson D, Dyer MA. Compen- Neuron. 1993;11:689–701. sation by tumor suppressor genes during retinal development in 32. Hinds JW, Hinds PL. Differentiation of photoreceptors and hori- mice and humans. BMC Biol. 2006;4:14. zontal cells in the embryonic mouse retina: an electron micro- 48. Spencer C, Pajovic S, Devlin H, Dinh QD, Corson TW, Gallie BL. scopic, serial section analysis. J Comp Neurol. 1979;187:495–511. Distinct patterns of expression of the RB gene family in mouse and 33. Liu W, Wang JH, Xiang M. Speciﬁc expression of the LIM/home- human retina. Gene Expr Patterns. 2005;5:687–694. odomain protein Lim-1 in horizontal cells during retinogenesis. Dev Dyn. 2000;217:320–325. 49. Zhang J, Gray J, Wu L, et al. Rb regulates proliferation and rod photoreceptor development in the mouse retina. Nat Genet. 2004; 34. Edqvist PH, Hallbook F. Newborn horizontal cells migrate bi- 36:351–360. directionally across the neuroepithelium during retinal development. Development. 2004;131:1343–1351. 50. Szekely L, Uzvolgyi E, Jiang WQ, et al. Subcellular localization of 35. Dyer MA, Livesey FJ, Cepko CL, Oliver G. Prox1 function controls the retinoblastoma protein. Cell Growth Differ. 1991;2:287–295. progenitor cell proliferation and horizontal cell genesis in the 51. Ma D, Zhou P, Harbour JW. Distinct mechanisms for regulating the mammalian retina. Nat Genet. 2003;34:53–58. tumor suppressor and antiapoptotic functions of Rb. J Biol Chem. 36. Hoover F, Seleiro EA, Kielland A, Brickell PM, Glover JC. Retinoid 2003;278:19358–19366. X receptor gamma gene transcripts are expressed by a subset of 52. Liu H, Dibling B, Spike B, Dirlam A, Macleod K. New roles for the early generated retinal cells and eventually restricted to photore- RB tumor suppressor protein. Curr Opin Genet Dev. 2004;14:55– ceptors. J Comp Neurol. 1998;391:204–213. 64. 37. Mori M, Ghyselinck NB, Chambon P, Mark M. Systematic immu- 53. Jiang Z, Zacksenhaus E, Gallie BL, Phillips RA. The retinoblastoma nolocalization of retinoid receptors in developing and adult mouse gene family is differentially expressed during embryogenesis. On- eyes. Invest Ophthalmol Vis Sci. 2001;42:1312–1318. cogene. 1997;14:1789–1797. 38. Roberts MR, Hendrickson A, McGuire CR, Reh TA. Retinoid X 54. Classon M, Harlow E. The retinoblastoma tumour suppressor in receptor gamma is necessary to establish the S-opsin gradient in development and cancer. Nat Rev Cancer. 2002;2:910–917. cone photoreceptors of the developing mouse retina. Invest Oph- 55. Calegari F, Huttner WB. An inhibition of cyclin-dependent kinases thalmol Vis Sci. 2005;46:2897–2904. that lengthens, but does not arrest, neuroepithelial cell cycle 39. Ng L, Hurley JB, Dierks B, et al. A thyroid hormone receptor that induces premature neurogenesis. J Cell Sci. 2003;116:4947–4955. is required for the development of green cone photoreceptors. 56. Calegari F, Haubensak W, Haffner C, Huttner WB. Selective length- Nat Genet. 2001;27:94–98. ening of the cell cycle in the neurogenic subpopulation of neural 40. Sjoberg M, Vennstrom B, Forrest D. Thyroid hormone receptors in progenitor cells during mouse brain development. J Neurosci. chick retinal development: differential expression of mRNAs for 2005;25:6533–6538.

Downloaded from iovs.arvojournals.org on 09/24/2021