BMP-induced L-Maf regulates subsequent BMP-independent differentiation of primary lens fibre cells

Authors


Abstract

Bone morphogenetic protein (BMP) signals are essential for lens development. However, the temporal requirement of BMP activity during early events of lens development has remained elusive. To investigate this question, we have used gain- and loss-of-function analyses in chick explant and intact embryo assays. Here, we show that BMP activity is both required and sufficient to induce L-Maf expression, whereas the onset of δ-crystallin and initial elongation of primary lens fibre cells are BMP-independent. Moreover, before lens placode formation and L-Maf onset, but not after, prospective lens placodal cells can switch to an olfactory placodal fate in response to decreased BMP activity. In addition, L-Maf is sufficient to up-regulate δ-crystallin independent of BMP signals. Taken together, these results show that before L-Maf induction BMP activity is required for lens specification, whereas after L-Maf up-regulation, the early differentiation of primary lens fibre cells occurs independent of BMP signals. Developmental Dynamics 240:1917–1928, 2011. © 2011 Wiley-Liss, Inc.

INTRODUCTION

Cell differentiation, including up- and down-regulation of genes and proteins as well as morphological changes, is part of the maturation process which a cell goes through after cell specification. The processes of both cell specification and differentiation are the outcome of interactions of multiple signaling molecules in an integrated temporal sequence. Today, one challenge in developmental and stem cell biology is to understand how individual signals change their roles during a short time window to regulate cell identity. For many decades, the lens has been used as a classical developmental model system to explore how external signalling molecules and transcription factors regulate induction, proliferation, cell fate decisions and differentiation. In this study we have used lens fibre cell differentiation assays to explore the temporal requirement of bone morphogenetic protein (BMP) activity during specification and early differentiation of primary lens fibre cells.

At gastrula and head-fold stages, prospective lens and olfactory placodal cells are intermingled at the rostral neural plate border (Bhattacharyya et al.,2004; Dutta et al.,2005). At the neural fold stage, lens and olfactory progenitor cells have become spatially separated (Couly and Le Douarin,1985; Bhattacharyya et al.,2004), and by early neural tube stages, the presumptive lens ectoderm overlies the optic vesicle. Subsequently, the presumptive lens ectoderm thickens to form the lens placode, which later invaginates into the lens vesicle. The lens vesicle becomes sub-divided into a posterior part, in which cells elongate and differentiate into primary lens fibre cells, and an anterior part, in which epithelial cells continue to proliferate during life, giving rise to secondary lens fibre cells (Lovicu and McAvoy,2005; Robinson,2006).

At gastrula stages, lens and olfactory placodal progenitors, situated at the rostral neural plate border (Bhattacharyya et al.,2004; Dutta et al.,2005), are exposed to Bmp2 and Bmp4, and express the BMP downstream activator, phosphorylated (p) Smad-1/5/8 (Chapman et al.,2002; Faure et al.,2002). At this stage, BMP activity is both required and sufficient to induce lens and olfactory placodal cells, and sustained exposure of placodal progenitors to BMP signals is required for lens induction (Sjodal et al.,2007). Consistently, in mouse, targeted deletions of different components of the BMP pathway lead to disturbed lens formation (Furuta and Hogan,1998; Wawersik et al.,1999; Beebe et al.,2004). At slightly later stages, BMP activity has been suggested to be required for crystallin expression in primary lens fibre cells (Sjodal et al.,2007; French et al.,2009; Rajagopal et al.,2009). However, the temporal requirement of BMP activity during early events of lens development remains to be determined in more detail.

L-Maf belongs to the large Maf proteins, and is a basic leucine zipper (bZip) transcription factor. The bZip motif mediates dimer formation and DNA binding to the Maf-recognition element (MARE; Kataoka et al.,1994; Kerppola and Curran,1994). Around lens placodal stage, lens development involves up-regulation of L-Maf expression in chick and c-Maf expression in mouse (Ogino and Yasuda,1998; Kawauchi et al.,1999). Somewhat later, the first step of primary lens fibre differentiation occurs, which involves up-regulation of crystallin proteins, where δ-crystallin is the first crystallin protein up-regulated in chick (Shinohara and Piatigorsky,1976; Bower et al.,1983; Reza and Yasuda,2004b). The δ-crystallin enhancer constitutes three Sox2, two Pax6, and two Maf binding sites (Muta et al.,2002). Although L-Maf, MafB, and c-Maf have similar DNA binding properties, L-Maf play a distinct role in inducing endogenous δ-crystallin in the lens (Yoshida and Yasuda,2002). Moreover, several in vivo studies have provided evidence that a regulatory complex of L-Maf, Pax6, and Sox2 induces the formation of ectopic lens placodes and δ-crystallin expression in the chick head ectoderm (Ogino and Yasuda,1998; Muta et al.,2002; Reza et al.,2002; Shimada et al.,2003). In addition, a dominant-negative L-Maf inhibits δ-crystallin expression in chick lenses (Reza et al.,2002), and in mouse, double mutations of MafUD binding sites significantly lowered the crystallin enhancer activity (Muta et al.,2002). Whether L-Maf depends on BMP activity to act as an important regulator for lens fibre cell determination has, however, not been defined.

Lens specification and early primary lens fibre cell differentiation occur during a short time-window. To examine the temporal requirement of BMP activity during these processes, we established explant and whole embryo assays of lens fibre cell differentiation in chick, and analyzed a panel of markers that are expressed in lens and olfactory placodal cells. Here, we show that BMP signals are required and sufficient to induce L-Maf expression. Furthermore, our results show that after L-Maf up-regulation, the early differentiation of primary lens fibre cells, indicated by the onset of δ-crystallin expression and initial cell elongation, is BMP-independent. In summary, these results provide evidence that BMP signals are required for the specification, but not for the early events of primary lens fibre cell differentiation, and that BMP-induced L-Maf acts as a regulator of the early differentiation of primary lens fibre cells.

RESULTS

BMP Activity Is Required for the Initial Expression of L-Maf in Lens Cells

As a first step to address whether BMP activity is required for the differentiation of primary lens fibre cells at early neural tube stages, we examined the expression pattern of Bmp4 and its downstream signalling mediator pSmad1/5/8, together with the onset of the lens markers L-Maf and δ-crystallin in Hamburger and Hamilton stage (HH) 11–17 chick embryos (Hamburger,1951). At HH11, Bmp4 and pSmad1/5/8 are expressed in the prospective lens ectoderm, but no L-Maf or δ-crystallin expression is detected (Fig. 1A). At HH13, Bmp4, pSmad1/5/8 and L-Maf, but not δ-crystallin are expressed in the newly formed lens placode as defined by the thickened region of the ectoderm (Fig. 1B). δ-crystallin is up-regulated around HH17, and expressed together with L-Maf in primary lens fibre cells. At this stage Bmp4 is not expressed, and only a few pSmad1/5/8+ cells are detected in the lens (Fig. 1C). Thus, Bmp4 and Smad1/5/8 expression, indicative of BMP activity, precedes the onset of L-Maf and δ-crystallin expression in the developing lens ectoderm.

Figure 1.

Bone morphogenetic protein (BMP) activity precedes the onset of L-Maf and δ-crystallin expression. A: At Hamburger and Hamilton stage (HH) 11, cells in the prospective lens ectoderm express Bmp4 and pSmad1/5/8 staining is seen in prospective lens ectoderm (arrowhead). B: At HH13, cells in the lens placode (arrowhead) express Bmp4 and pSmad1/5/8, and the lens specific marker L-Maf is induced, but not δ-crystallin. C: By HH17, around early lens vesicle stages, L-Maf and δ-crystallin are expressed in lens fibre cells, Bmp4 expression is localized to the ventral periphery of the optic cup (arrow), but not in the lens, and pSmad1/5/8 expression is observed in a few cells in the lens vesicle and in the optic cup.

Both lens and olfactory placodal cells are known to be derived from the rostral neural plate border (Bhattacharyya et al.,2004; Sjodal et al.,2007), and at HH11, Sox2 and Pax6 are co-expressed in the presumptive lens ectoderm as well as in prospective olfactory placodal cells (Supp. Fig. S1A,B, which is available online). However at later stages, these two placodal cell types can be distinguished by the use of a set of different markers. At HH21, in the eye region, L-Maf and δ-crystallin expression are expressed in lens fibre cells, the non-neural marker Keratin is expressed in the surface ectoderm and the postmitotic neural marker HuC/D is expressed in the retina, but no Dlx expression is detected by the pan-Dlx antibody (Panganiban et al.,1995; Fig. 2A). In contrast, at the same stage, cells in the olfactory epithelium express Keratin and Dlx, and a subset of cells express HuC/D, but no L-Maf or δ-crystallin expression are detected (Fig. 2B).

Figure 2.

At Hamburger and Hamilton stage (HH) 11 bone morphogenetic protein (BMP) activity is required for L-Maf and δ-crystallin, but not Pax6 and Sox2 in vitro. A: At HH21, expression of L-Maf and δ-crystallin is observed in lens fibre cells. Keratin is expressed in the surface ectoderm, and the post-mitotic neural marker HuC/D is expressed in a few retinal cells. B: At HH21, cells in the olfactory pit express Keratin and Dlx, and a subset of cells express HuC/D. C: Schematic drawing depicting the explant method. HH11 or HH13 lens/retina (LR) explants were dissected and cultured in vitro to the developmental equivalent time-point of HH21. The explants were then fixated, frozen, and sectioned. D–G: Arrows indicate prospective lens cells, and arrowheads indicate prospective retinal cells. D: HH11 LR explants cultured alone (n = 10) generated a distinct region of L-Maf+ (10/10) and δ-crystallin+ (10/10) lens fibre cells, and a separate region of HuC/D+ (10/10) retinal cells, and a small domain with a few expressing Keratin cells (10/10). No region of Keratin, Dlx, and HuC/D expression (0/10) defining olfactory placodal cells was detected. E: In HH11 LR explants cultured together with Noggin (n = 11), the generation of L-Maf+ (0/11) and δ-crystallin+ (0/11) lens fibre cells was inhibited, and cells switched to an olfactory epithelial fate as characterized by the expression of Keratin, Dlx, and HuC/D in the same region (11/11). F: HH11 LR explants cultured alone (n = 10) generated Pax6+ (10/10) lens epithelial cells, and Sox2+ (10/10) and δ-crystallin+ (10/10) lens fibre cells. Sox2 expression and a few scattered Pax6+ cells were also detected in cells of retinal fate (10/10). G: HH11 LR explants cultured together with Noggin (n = 10) generated Pax6+ and Sox2+ cells in an overlapping Keratin+ region (10/10), and no expression of δ-crystallin (0/10) was detected. Arrows indicate prospective lens cells, and arrowhead indicates prospective retinal cells. Scale bars = 100 μm in A and B.

To examine whether BMP signals are required for the up-regulation of L-Maf and/or δ-crystallin expression, we established an explant assay of lens cell differentiation by culturing HH11 prospective lens/retina (LR) chick explants. Around HH10–HH13, the prospective lens ectoderm consists of only one or a few cell layers (Fig. 1A,B), and the connections between prospective lens and retinal cells are strong. Thus, to avoid cell death caused by the separation of these two ectodermal layers, prospective lens and retina cells were cultured together. The LR explants were cultured alone or in the presence of Noggin, a known BMP inhibitor (Lamb et al.,1993) for around 42 hr, which in intact chick embryos would correspond to approximately HH21/22 (Fig. 2C). Thereafter, the explants were processed, serially sectioned, and analyzed by immunohistochemistry (Fig. 2C).

HH11 LR explants cultured alone (n = 10) generated L-Maf+ and δ-crystallin+ cells, characteristic of primary lens fibre cells, and HuC/D+ retinal cells, and Keratin+ cells were also detected in separate smaller domains of the explants, but no Dlx+ cells were detected (Fig. 2D). In contrast, in the presence of Noggin (n = 11), the expression of L-Maf and δ-crystallin was blocked, and Dlx+ cells and HuC/D+ neurons were generated in the Keratin+ expanded region of the explants; characteristic of olfactory placodal cells (Fig. 2E). In addition, we confirmed that there was no significant change in activated (a) Caspase3+ cells, indicative of cell death, in either the prospective lens or retina domains after 15 hr of culture (Supp. Fig. S2), supporting the conclusion that prospective lens cells switch cell fate after Noggin treatment. Thus, at HH11, BMP activity is required for the induction of L-Maf and δ-crystallin expression in prospective lens cells, and in the absence of BMP activity prospective lens cells acquire olfactory placodal character.

Maintenance of Placodal Sox2 and Pax6 Expression Is BMP Independent

Sox2 and Pax6 are known to regulate δ-crystallin expression (Kamachi et al.,2001; Muta et al.,2002) and are expressed in the prospective lens ectoderm at HH 11 (Supp. Fig. S1A). Thus, another possibility for the loss of δ-crystallin expression in prospective lens ectodermal cells after inhibition of BMP activity is that Sox2 and Pax6 expression are down-regulated. To examine this we analyzed the expression of Sox2 and Pax6 in our explant assays. In HH11 LR explants (n = 10), Sox2+ cells were generated in the δ-crystallin domain, characteristic of primary lens fibre cells, whereas Pax6 was expressed in a separate region, characteristic of lens epithelial cells (Fig. 2F). Sox2 expression and a few scattered Pax6+ cells were also detected in cells of retinal fate, and Pax6+ cells were also detected in the Keratin+ surface ectodermal cells (Fig. 2F). In the presence of Noggin, HH 11 LR explants (n = 10) still generated Pax6+ cells, which co-expressed both Sox2 and Keratin, while the generation of δ-crystallin+ cells was inhibited (Fig. 2G). These results indicate that the loss of δ-crystallin expression in prospective lens cells after BMP inhibition is not due to down-regulation of Sox2 or Pax6 expression. Thus, at HH 11, prospective lens cells specifically require BMP signals for the generation of lens placodal cells, but not for the maintenance of Sox2 and Pax6 expression.

BMP Activity Is Required for the Up-regulation of L-Maf Expression in Intact Chick Embryos

Next, we examined the requirement of BMP signaling for the induction of L-Maf and δ-crystallin expression in primary lens fibre cells in vivo. To address this question, HH11 chick embryos were electroporated in ovo in the lens ectodermal region to transfer a control green fluorescent protein (GFP) vector alone (n = 6) or together with a Noggin-expressing vector (n = 10; Timmer et al.,2002). The electroporated embryos were cultured to approximately HH21, and embryos with GFP staining within the lens region were selected for further analyses. Noggin-electroporated lenses were compared with GFP-electroporated lenses. To verify that BMP signaling was blocked after Noggin electroporation, a pSmad-GFP-reporter construct (BRE-tk-GFP), indicative of activated BMP signalling, and a red fluorescent protein (RFP) construct were electroporated alone or together with a Noggin construct in HH11lens ectoderm and cultured to approximately HH17. During these conditions, pSmad activity was abolished in Noggin-electroporated lenses (n = 4) compared with control-electroporated lenses (n = 4), providing evidence that BMP activity is inhibited in Noggin-electroporated cells (Fig. 3A,B).

Figure 3.

Bone morphogenetic protein (BMP) activity is required for the induction of L-Maf in intact chick embryos. A,B: In ovo electroporation of Hamburger and Hamilton stage (HH) 11 embryos in the prospective lens ectoderm region using a pSmad-GFP reporter construct and a RFP vector alone (n = 4; A) or together with a Noggin vector (n = 4; B) and cultured to HH17. A: In all control embryos, a fraction of the electroporated cells in the lens expressed green fluorescent protein (GFP), indicative of pSmad activity. B: The co-electroporation of Noggin inhibited pSmad activity. C–E: Electroporation as described above using a GFP vector alone (n = 6; C) or together with a Noggin vector (n = 10; D,E) and cultured to HH21. C: All embryos (6/6) electroporated with the control GFP vector exhibited a normal lens morphology and normal expression pattern of L-Maf, δ-crystallin and Keratin. D: In the most severely affected Noggin-electroporated embryos (7/10), the lens failed to develop (arrowhead) and the retina failed to develop beyond the optic vesicle stage. No expression of L-Maf (0/7) or δ-crystallin (0/7) was detected, while Keratin expression was still detected in the remaining surface ectoderm (7/7). E: In the less affected Noggin-electroporated embryos (3/10), a small lens-like structure expressing L-Maf and δ-crystallin was observed (arrowhead). Keratin was detected in the surface ectoderm (3/10). D,E: The development of the retina was also disturbed (10/10; arrows).

After culture to HH21, all control electroporated embryos exhibited normal morphology of the lens and a normal expression pattern of L-Maf and δ-crystallin (Fig. 3C). In contrast, in the Noggin-electroporated embryos, expression of L-Maf and δ-crystallin was reduced or abolished, while Keratin+ cells were generated in the remaining surface ectoderm (Fig. 3D,E). In the most severely affected embryos the lens structure was completely missing, and no expression of L-Maf or δ-crystallin could be detected (Fig. 3D). The slightly less affected embryos exhibited a grossly disturbed lens, and cells within the lens-like structure expressed L-Maf and δ-crystallin (Fig. 3E). In addition, in all Noggin-electroporated embryos on the electroporated side we observed that the retina failed to develop properly beyond optic vesicle stage (Fig. 3D,E), however, we did not observe any change in cell death in the prospective retina in Noggin-electroporated embryos compared with control embryos (data not shown).

To determine whether a change in proliferation and/or cell death accounted for the disrupted lens formation, we analyzed the expression of the cell proliferation marker phosphorylated (p) Histone3 and the cell death marker aCaspase3. The Noggin-electroporated embryos were cultured to approximately HH13/14, HH15, and HH17. pHistone3+/GFP+ and aCaspase3+/GFP+ double-positive cells in the lens ectoderm, defined as the head ectoderm adjacent to the prospective optic vesicle, were quantified. Embryos cultured to HH13/14, before any major morphological differences of the eye region are observed, did not show any significant changes in either pHistone3+ cells or aCaspase3+ cells compared with control GFP-electroporated embryos (Fig. 4A,B and Supp. Fig. S3A,B). In embryos cultured to HH15 and HH17, there was a decrease in pHistone3+ cells (Fig. 4C–F), but no change in aCaspase3+ cells compared with control GFP-electroporated embryos (Supp. Fig. S3C–F). In the Noggin-electroporated embryos cultured to HH15, the lens ectoderm failed to invaginate (Fig. 4C,D). Subsequently, at HH17 the disturbed lens phenotype was more pronounced, in that the lens ectoderm had failed to develop a lens vesicle (Fig. 4E,F). These data indicate that, in the absence of BMP activity, decreased proliferation within the lens ectoderm appears to disrupt lens epithelial invagination and further lens development. Taken together, our results show that before HH11, BMP activity is crucial for the early formation of the lens, and for the up-regulation of L-Maf and δ-crystallin expression in lens cells in intact chick embryos.

Figure 4.

Inhibition of bone morphogenetic protein (BMP) activity reduces cell proliferation in the lens ectoderm. A,B: In Hamburger and Hamilton stage (HH) 11 Noggin-electroporated embryos cultured to approximately HH13/14 (n = 5), the generation of phosphorylated (p) Histone3+ cells was not changed compared with control green fluorescent protein (GFP) -electroporated embryos (n = 5). C,D: In HH11 Noggin-electroporated embryos cultured to approximately HH15 (n = 6), the generation of pHistone3+ cells was decreased compared with control GFP-electroporated embryos (n = 4). E,F: In HH11 Noggin-electroporated embryos cultured to approximately HH17 (n = 5), the generation of pHistone3+ cells was significantly reduced compared with control GFP-electroporated embryos (n = 8). C–F: In Noggin-electroporated embryos, the invagination of the lens ectoderm was disturbed, and the lens ectoderm failed to develop a lens vesicle. Error bars represent mean ± SEM. *P < 0.05.

BMP Activity Is Sufficient to Induce L-Maf Expression

At gastrula stages, Bmp4 is expressed in the ectoderm of the rostral border region, where lens placodal progenitors are situated, but Bmp4 is not expressed in the neural plate region (Fig. 5A; Streit et al.,1998; Chapman et al.,2002). To confirm that BMP signals are sufficient to induce L-Maf expression, we exposed HH4 prospective rostral neural (N) explants to BMP4 (35 ng/ml) in culture for 43–45 hr, which corresponds to approximately HH17. HH4 N explants cultured alone (n = 13) generated L5+ and Sox1+ cells; characteristic of neural cells (Roberts et al.,1991; Pevny et al.,1998), but no L-Maf+ lens cells were detected (Fig. 5B). Addition of BMP4 to HH4 N explants (n = 9) blocked the generation of L5+ and Sox1+ neural cells and induced L-Maf+ lens cells (Fig. 5C). Thus, BMP signals are sufficient to induce L-Maf+ lens cells in prospective rostral neural cells.

Figure 5.

Bone morphogenetic protein (BMP) activity is sufficient to induce L-Maf expression in prospective neural cells. A: At Hamburger and Hamilton stage (HH) 4, Bmp4 expression is up-regulated in cells in the rostral neural plate border region, and at HH6 the Bmp4 expression is increased. No Bmp4 expression is detected in the neural plate (N). B: HH4 rostral neural (N) explants (n = 13) cultured alone generated Sox1+ (13/13) and L5+ (13/13) neural cells, but no L-Maf+ cells (0/13) were detected. C: In HH4 N explants (n = 9), BMP4 (35 ng/ml) blocked the generation of Sox1+ (0/9) and L5+ (0/9) neural cells and induced L-Maf+ lens cells (9/9).

Up-regulation of δ-crystallin Is Indirectly Dependent on BMP Signals

The finding that δ-crystallin expression was blocked in HH11 Noggin-treated lens cells in vitro and in vivo, can either be due to a direct requirement of BMP signals for δ-crystallin up-regulation, or due to an indirect effect caused by the abolished L-Maf expression. To examine this issue, we inhibited BMP activity both in vitro and in vivo at HH13, after the onset of L-Maf expression, but before δ-crystallin up-regulation. HH13 LR explants cultured in the presence of Noggin (n = 10) for around 37 hr, generated L-Maf+ and δ-crystallin+ cells, characteristic of primary lens fibre cells, and Keratin+ cells in a separate region of the explants, (Fig. 6B), in a similar manner as HH13 LR explants cultured alone (n = 10; Fig. 6A). Thus, at HH13, L-Maf positive lens placodal cells differentiate into δ-crystallin expressing lens fibre cells independent of BMP activity in vitro.

Figure 6.

After L-Maf onset, δ-crystallin induction is bone morphogenetic protein (BMP) independent in intact chick embryos. A,B: Hamburger and Hamilton stage (HH) 13 lens/retina (LR) explants cultured alone (n = 10; A), or in the presence of Noggin (n = 10; B) generated a distinct region of L-Maf+ (10/10) and δ-crystallin+ (10/10) lens fibre cells, and separate regions of cells expressing Keratin (10/10). Arrows indicate prospective lens cells, and arrowheads indicate prospective retinal cells. C,D: In ovo electroporation of HH13 embryos in the prospective lens ectoderm region using a green fluorescent protein (GFP) vector alone (C) or together with a Noggin vector (D), and cultured to HH21. Embryos electroporated with GFP alone (n = 5; C) or together with Noggin (n = 5; D) exhibited a normal lens morphology and normal expression pattern of L-Maf, δ-crystallin and Keratin (5/5).

To examine whether L-Maf positive lens placodal cells develop further in the absence of BMP signalling also in vivo, we electroporated HH13 chick embryos in the lens placodal region to transfer a GFP vector alone (n = 5) or together with a Noggin (n = 5) construct, and cultured to approximately HH21. Both the control- and the Noggin-electroporated embryos exhibited normal morphology of the lens, including initial elongation of primary fibre cells, and a normal expression pattern of L-Maf and δ-crystallin (Fig. 6C,D). Taken together, after the onset of L-Maf expression, the up-regulation of δ-crystallin and initial elongation of primary fibre cells is independent of BMP activity.

L-Maf Expression Is Sufficient to Induce Lens Fate Independent of BMP Signals

Our results suggest that after the specification and up-regulation of L-Maf + lens cells, BMP activity is not required for the early differentiation of primary lens fibre cells. To directly test this issue, we electroporated an L-Maf construct (Ogino and Yasuda,1998) alone or together with a Noggin construct in the prospective olfactory placodal region and rostral non-lens head ectoderm of HH9–HH11 chick embryos, and cultured to approximately HH18–HH21. During these conditions, ectopic expression of L-Maf, even in the absence of BMP activity, induced δ-crystallin+ cells in the olfactory placodal region (Fig. 7A,B). Also in the head ectoderm L-Maf induced ectopic δ-crystallin expression (Fig. 7C), which is in agreement with previous results (Ogino and Yasuda,1998; Shimada et al.,2003). Although BMP activity was inhibited in the head ectoderm, L-Maf still induced ectopic δ-crystallin expression (Fig. 7D). Thus, L-Maf expression directs the up-regulation of δ-crystallin independent of BMP activity in intact chick embryos, implicating that L-Maf acts as a regulator of primary lens fibre differentiation. Taken together, our results provide evidence that BMP signals are required for the specification of L-Maf positive lens cells, but not for the subsequent differentiation of primary lens fibre cells.

Figure 7.

L-Maf acts as a regulator of cell fate determination in the lens independent of bone morphogenetic protein (BMP) activity. A–D: Hamburger and Hamilton stage (HH) 9–11 embryos were electroporated with only a green fluorescent protein (GFP) vector and an L-Maf vector, or in combination with a Noggin vector in the surface ectoderm nearby the lens ectoderm region and cultured in ovo to HH18-21. A: Embryos electroporated with GFP and L-Maf in the olfactory placode (n = 6), ectopically expressed L-Maf (6/6) and δ-crystallin (6/6) in the olfactory region (arrowheads). B: Also in embryos electroporated with GFP, L-Maf, and Noggin in the olfactory placode (n = 5), L-Maf (5/5) and δ-crystallin (5/5) were ectopically expressed (arrowheads). C: Embryos electroporated with GFP and L-Maf in the non-lens head ectoderm region (n = 5), ectopically generated L-Maf (5/5) and δ-crystallin (5/5; arrowheads). D: Also in embryos electroporated with GFP, L-Maf, and Noggin (n = 7), L-Maf (7/7) and δ-crystallin (7/7) were ectopically expressed (arrowheads).

DISCUSSION

As the lens forms, cells in the posterior part of the lens differentiate into lens fibre cells. Bmp4 and Bmp7 knockout studies in mice, and gain- and loss- of function experiments in chick have shown that BMP signalling is essential for lens induction (Dudley and Robertson,1997; Furuta and Hogan,1998; Wawersik et al.,1999; Sjodal et al.,2007), without addressing the temporal requirement of BMP signals during primary lens fibre cell differentiation. In this study, we have determined the early requirement of BMP activity during the specification and early differentiation of primary lens fibre cells. Our results provide evidence that BMP signals are required and sufficient to induce L-Maf expression, which acts as an important regulator of lens fibre cell fate. Furthermore, our results show that, after L-Maf up-regulation, the early differentiation of primary lens fibre cells, including up-regulation of δ-crystallin expression and initial elongation of fibre cells, occur independent of BMP signals.

Previous studies in chick, mouse, and zebrafish have suggested that L-Maf and crystallin expression in the lens requires BMP activity (Sjodal et al.,2007; French et al.,2009; Rajagopal et al.,2009). Our results extend this knowledge and provide evidence that BMP activity is both required and sufficient to induce L-Maf expression, whereas the subsequent up-regulation of δ-crystallin is independent of BMP signals. Our results show that, in the presence of L-Maf, δ-crystallin is up-regulated in lens cells even in the absence of BMP activity. In addition, our data show that ectopic L-Maf expression induces δ-crystallin expression in head ectodermal cells independent of BMP activity. Taken together, these results suggest that L-Maf plays an important role in the determination of lens fibre cell identity, and that after L-Maf up-regulation the early differentiation of primary lens fibre cells occurs independent of BMP signals. In agreement with this, previous studies have shown that L-Maf is required for proper δ-crystallin expression (Ogino and Yasuda,1998; Muta et al.,2002; Reza et al.,2002; Shimada et al.,2003), and that L-Maf regulates cell cycle exit of lens cells and fibre cell differentiation by activating p27kip1, a cell cycle inhibitor (Reza et al.,2007). At later stages in chick and mouse, L-Maf expression is restricted to cells at the equator of the lens, the region where epithelial cells differentiate into secondary lens fibre cells (Reza et al.,2007; Takeuchi et al.,2009). Consistently, pSmad expression, indicative of BMP activity, is also localized to the equatorial region of the lens, and inhibition of BMP signals results in reduced sizes of the lens (Belecky-Adams et al.,2002; Faber et al.,2002), suggesting that the differentiation of lens fibre cells are disturbed.

At neural fold stages, BMP signals induce L-Maf and δ-crystallin expression in prospective olfactory placodal cells, and continuous exposure to BMP signals to rostral placodal progenitors promote the generation of lens cells at the expense of olfactory placodal cells (Sjodal et al.,2007). Our results now provide evidence that before the onset of L-Maf expression but not after, prospective lens placodal cells can switch to an olfactory placodal fate in response to decreased BMP activity. Moreover, ectopic L-Maf expression induces δ-crystallin expression in olfactory placodal cells even in the absence of BMP signals. These data provide evidence that BMP-induced L-Maf plays an important role in the decision between lens and olfactory placodal character, and that BMP activity is required for lens specification until the onset of L-Maf expression. This correlates with the fact that pSmad-1/5/8 is preferentially expressed in the prospective lens ectoderm compared with the prospective olfactory placodal region (Sjodal et al.,2007).

Our results show that Pax6 and Sox2 are co-expressed in the presumptive lens ectoderm as well as in prospective olfactory placodal cells, consistent with Pax6 and Sox2 expression patterns in mouse (Donner et al.,2007). Previous findings have described that the expression of Pax6 is still detected in the prospective lens ectoderm in Bmp4 knockout mouse embryos (Furuta and Hogan,1998), and in mice in which the two type I BMP receptors Alk2/Alk3 were inactivated in the surface ectoderm (Rajagopal et al.,2009) using the Le-Cre enhancer (Ashery-Padan et al.,2000). In contrast, the levels of Sox2 expression in the prospective lens ectoderm appear to depend on BMP receptor signaling in a Smad-dependent manner, based on studies using Lens-Cre deletion of Alk2/Alk3, or Smad1/Smad5, or Smad4 (Rajagopal et al.,2009). Consistently, expression of Sox2 is lost in the lens ectoderm of Bmp7 knockout mice (Wawersik et al.,1999), and reduced in Bmp4 knockout mice (Furuta and Hogan,1998). Our results provide evidence that BMP signals are not required for the maintenance of Pax6 or Sox2 expression in prospective lens ectodermal cells. It is important to note, however, that the prospective lens ectodermal cells shift to an olfactory placodal identity, indicating that Pax6 and Sox2 do not regulate the initial differential specification of lens and olfactory placodal cell character.

Previous studies concerning cell differentiation in various contexts have also suggested that L-Maf plays a role in this process. In the pancreas, c-Maf and MafA are differentially expressed in α- and β-cells, and were identified as regulators of glucagon and insulin gene expression, respectively (Kataoka et al.,2002; Olbrot et al.,2002; Artner et al.,2006; Nishimura et al.,2006). In addition, c-Maf promotes different types of interleukin expression in T cells and T helper cells (Ho et al.,1996; Xu et al.,2009; Hiramatsu et al.,2010). Moreover, inhibition of MafA in adipose cells resulted in down-regulation of transcriptional factors essential for adipocyte differentiation (Tsuchiya et al.,2009). Finally, in the hindbrain, MafB has been demonstrated to regulate segmental compartments by hoxa-3 and hoxb-3 expression (Manzanares et al.,1999). Our results now provide additional insights into the molecular mechanisms during the differentiation of lens cells to a lens fibre cell identity, in which BMP activity induces L-Maf expression, which acts as a cell fate determinant. A study regarding dorsal spinal cord patterning has shown that MafB is positively regulated by BMP signalling (Chizhikov and Millen,2004), indicating that the large Maf genes might be regulated by a common mechanism. In summary, we have determined the temporal requirement of BMP activity during early lens development, and provide evidence that BMP signals are required for the specification of L-Maf positive lens fibre cells, but not for the subsequent differentiation of primary lens fibre cells.

EXPERIMENTAL PROCEDURES

Embryos

Fertilized White Leghorn chicken eggs were obtained from Agrisera AB, Umeå, Sweden. Chick embryos were staged according to the protocol of Hamburger and Hamilton (1951). The use of chick embryos in this study was approved by the Ethical Committee on Animal Experiments for Northern Sweden.

Explant Assay

Prospective lens/retina explants were isolated from Hamburger and Hamilton stage (HH) 11 and HH13 chick embryos, and prospective neural explants were isolated from HH4 chick embryos. The explants were cultured in vitro in collagen in serum-free OPTI-MEM (GIBCO) containing N2 supplement (Invitrogen) and fibronectin (Sigma) to desired time points. BMP4 (R&D Systems) was used at 35 ng/ml. Soluble noggin and control-conditioned medium (CM) were obtained from stably transfected or un-transfected Chinese hamster ovary (CHO) cells (Lamb et al.,1993) and cultured in CHO-S-SFM II media (Gibco). Noggin-conditioned media was used at an estimated concentration of 50 ng/ml. Explants cultured in the presence of control CM generated the same combination of cells as explants cultured alone (data not shown).

In Ovo Electroporation

HH11 and HH13 embryos were electroporated in the lens ectoderm with pCaggs-GFP (Andersson et al.,2006; 0.5 μg/μl) and pMiwIII-Noggin (Timmer et al.,2002; 0.5 μg/μl). HH11 embryos were electroporated in the lens ectoderm with BRE-tk-GFP (gift from E. Marti, unpublished construct; 0.8 μg/μl) and pCaggs-RFP (gift from J. Gilthorpe, unpublished construct; 0.6 μg/μl) with or without pMiwIII-Noggin (Timmer et al.,2002; 0.6 μg/μl). HH9–HH11 embryos were electroporated in the prospective olfactory placodal region and head surface ectoderm with a combination of pCaggs-GFP (0.8 μg/μl), pMiwIII-Noggin (0.6 μg/μl) and pCaggs-L-Maf (Ogino and Yasuda,1998; 0.7 μg/μl). DNA was transferred using an Electro Square Porator ECM 830 (BTX, Inc.) by applying five pulses (9–15 V, 25-msec duration) at 1-sec intervals. Embryos were cultured in ovo to the developmental stage of interest.

We confirmed that the development of GFP-electroporated lenses was equal to the development of the contralateral un-electroporated lens. Noggin-electroporated lenses were compared with GFP-electroporated lenses.

In Situ Hybridization and Immunohistochemistry

For the use of in situ RNA hybridization and immunohistochemistry, embryos were fixed in 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS) for 1.5 hr and explants for 25 min at 4°C. In situ RNA hybridization using chick digoxigenin-labeled Bmp4 probe (Francis et al.,1994) was performed essentially as described (Wilkinson and Nieto,1993). Antibodies used were as follows: anti-cytokeratin (Dako), anti-L-Maf (Reza and Yasuda,2004a), anti-Distal-less (pan-Dlx; Panganiban et al.,1995), anti-Sox2 and anti-Sox1 (Patthey et al.,2009), anti-pSmad1/5/8 (Cell Signalling), anti-pHistone3 (Millipore) rabbit antibodies, sheep anti–δ-crystallin (Beebe and Piatigorsky,1981), the monoclonal anti-L5 rat antibody (Roberts et al.,1991), anti-HuC/D (Molecular probes), anti-Pax6 (Developmental Studies Hybridoma Bank), and anti-aCaspase3 (Cell Signalling) mouse monoclonal antibodies. Nuclei were stained using DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; Sigma).

Statistics

To determine the percentage of pHistone3- and aCaspase3-expressing cells in electroporated GFP-positive regions in vivo, the number of antigen-expressing cells was quantified and compared with the total number of cells, determined by DAPI-stained nuclei. The graphs represent the mean number of cells positively stained for pHistone3 and aCaspase3, respectively. Error bars represent ± SEM. P values were obtained comparing GFP-control with GFP and Noggin-construct-electroporated embryos using an un-paired Student's t-test.

Acknowledgements

We thank the following persons for kindly providing antibodies and plasmids: P. Brickell (Bmp4), G. Boekhoff-Falk (Dll), T. Edlund (Sox1, Sox2), J. Ericson (pCaggs-GFP), J. Gilthorpe (pCaggs-RFP), E. Marti (BRE-tk-GFP), L. Niswander (pMiwIII-Noggin), J. Piatigorsky (δ-crystallin), A. Streit (L5), and K. Yasuda (L-Maf, pCaggs-L-Maf). We thank H. Alstermark for technical assistance in the beginning of the project. We thank Michael Wride and Ales Cvekl for comments on the manuscript and members of the Gunhaga laboratory for helpful discussions. L.G. is supported by Umeå University, Sweden and Kronprinsessan Margaretas (KMA) foundation.

Ancillary