Roles of maf family proteins in lens development



Lens provides a good model for studying developmental cues relevant to cellular and molecular interactions. Basic region/leucine zipper (bZIP) transcription factors have been found to play key roles during eye formation in various species, including human, mouse, rat, Xenopus, zebrafish, chick, and quail. Different ocular developmental anomalies associated with MAF mutation in human implicate its active role during eye development. Several members of the maf gene family with this bZIP motif participate directly in lens morphogenesis. One vital Maf protein, L-Maf, is expressed in developing lens cells of chick embryos. Its homolog recently has been detected in lens placode of Xenopus embryos and regulates expression of lens fiber-specific genes in this species. Ectopic expression of L-Maf can induce lens-specific genes in cultured retina cells and embryonic ectoderm. The dominant-negative form of L-Maf causes the suppression of crystallin expression and subsequently inhibits lens formation, indicating that L-Maf plays a central role in chick lens development. Developmental Dynamics 229:440–448, © 2004 Wiley-Liss, Inc.


Development of multicellular organisms requires cell differentiation, which is regulated by spatial and temporal expression of cell-specific genes. Lens is considered an ideal model for studying the process of differentiation because of its simple tissue organization. The classic Xenopus transplant study demonstrated that lens induction occurs through four successive steps, termed bias, competence, specification, and differentiation (for review, see Grainger,1992). After tissue interaction, a signaling event between the optic vesicle and head ectoderm leads to lens induction (for review, see McAvoy,1980). During this process, the ectoderm undergoes thickening and forms the lens placode, which in turn gives rise to the mature lens.

Extensive studies of lens development have explored the involvement of different classes of transcription factors (for reviews, see Kondoh,1999; Ogino and Yasuda,2000; Chow and Lang,2001). Some of these factors are highly conserved among a wide range of species, whereas others are specific to one or a few species. Accumulating evidence indicates that lens-specific crystallins, which constitute the major portion of transparent lens, are regulated by different types of transcription factors. Thus, crystallin genes provide excellent markers for studying lens-specific expression (for review, see Piatigorsky,1981), and some transcription factors that regulate expression of the crystallin genes have been found to be specific for both target genes and cell types.

In previous work from our laboratory, we identified an enhancer element of the αA-crystallin gene, located between nt -119 and nt -99, which plays a pivotal role in determining tissue specificity (Matsuo and Yasuda,1992). Although lens-specific enhancers that activate crystallin genes have been determined (Roth et al.,1991; Matsuo et al.,1992; Sax et al.,1993; Kamachi and Kondoh,1993), we were able to identify L-Maf as a transcription factor that regulates the αA-crystallin expression (Ogino and Yasuda,1998) by screening a chick lens cDNA library with a αCE2 oligonucleotide. L-maf belongs to the large maf gene family, and it activates αA-crystallin expression through an αCE2 enhancer sequence (Ogino and Yasuda,1998). In fact, the crucial role of the maf gene family in lens development was completely unknown before this report. Recent studies have shown that c-Maf is involved in lens fiber differentiation in mouse (Kim et al.,1999, Kawauchi et al.,1999; Ring et al.,2000). Although the zebrafish mutant valentine (i.e., a mutation in the mafB gene) forms a normal lens (Moens et al.,1996), mafB null-mutant mice have been observed to have defective lens formation (Takahashi et al., personal communication). Recently, maf has been found as a disease-causing gene. Pulverulent cataract of the lens along with other ocular abnormalities have been identified due to maf mutation in human (Jamieson et al.,2002,2003; Lyon et al.,2003). Taken together, these reports indicate that Maf proteins are directly involved in lens development in a wide range of species. In this review, we present the structural and functional properties of the maf genes implicated in lens development, and we show that L-Maf is a unique molecule that plays a central role in regulating crystallin expression as well as lens formation in chick.


The maf oncogene family originally emerged with the isolation of v-maf gene from the genome of acute transforming retrovirus AS42, which causes musculoaponeurotic fibrosarcoma in chick (Nishizawa et al.,1989; Kawai et al.,1992). To date, several maf-related genes have been cloned in various species, including human, mouse, rat, frog, quail, chick, and zebrafish. All the proteins in this family share a highly conserved basic region and a leucine zipper domain (bZIP) at their C-terminus, which allows them to bind to the target DNAs and dimerize with the same or other proteins possessing a bZIP domain (for review, see Blank and Andrews,1997). Maf proteins can bind through common recognition elements known as T-MARE or C-MARE (Kataoka et al.,1994a,b). Maf family transcription factors are subdivided into two major groups based on their molecular size. Small Mafs encode proteins that lack a putative activation domain, but can transactivate after forming a heterodimer with CNC, a different class of bZIP containing proteins (for review, see Motohashi et al.,1997). To date, three small Mafs—MafK, MafF, and MafG—have been identified in mammals and chick (Igarashi et al.,1995; Kataoka et al.,1995).

In contrast, large Mafs have a C-terminal DNA-binding bZIP domain, as well as a putative activation domain (Fig. 1A) rich in proline, serine, and threonine at their acidic N-termini (for review, see Blank and Andrews,1997). Large Mafs identified in human and mouse include c-Maf, MafB/Krml, MafA, and NRL (Swaroop et al.,1992, Yang-Feng and Swaroop,1992; Kurschner and Morgan,1995; Chesi et al.,1998; Wang et al.,1999; Kataoka et al.,2002), whereas c-Maf, MafB, and L-Maf have been identified in chicks and Xenopus (Ogino and Yasuda,1998; Kajihara et al.,2001; Ishibashi and Yasuda,2001). We have detected two additional maf genes, Smaf1 and Smaf2, in zebrafish (Kajihara et al.,2001). Based on homology of amino acid residues among the Maf proteins, it appears that Smaf1 in zebrafish, MafA in mouse, human, and quail, and chick L-Maf are homologs, because the acidic, basic, and leucine zipper domains of these Maf proteins possess significant sequence identity (Fig. 1A). In addition to the acidic and bZIP domains, Maf family proteins have a highly conserved amino acid sequence known as extended homology region (EHR), which is located on the amino terminal side of the basic domain (Fig. 1; Kerppola and Curran,1994). The EHR, by participating with the basic region in the conformational change of Maf protein, plays a significant role in the specific recognition of Maf binding site, thus allowing target gene specificity for Maf (Dlakic et al.,2001; Kusunoki et al.,2002). Large Maf proteins have an additional sequence between the activation domain and EHR, termed the hinge region (Fig. 1; exception, NRL), and it is characterized by the presence of glycine and histidine clusters. The role of this polypeptide domain is largely unknown.

Figure 1.

Schematic representation of Mafs. A: Chick L-Maf and its homologs in various species are shown. Acidic, basic, extended homology region (EHR), and leucine zipper (L-ZIP) domains, as well as hinge region are specified. The percent identities of the amino acid residues in the b-ZIP region shown by double-headed arrow are indicated. The total amino acid residues of each Maf protein is also specified. B: Schematic diagram of chick L-Maf, c-Maf, and MafB proteins. Different domains are indicated in the boxes. Indicated percentages specify levels of amino acid identity to L-Maf in EHR and basic domains and also in leucine zipper (L-ZIP) domain.


maf genes are expressed in a highly regulated temperospatial manner. We investigated the expression of the three Mafs during early chick embryogenesis (Yoshida and Yasuda,2002). L-Maf, which has the lowest molecular weight of the three chick Maf proteins (Fig. 1B), is the first of these proteins to be expressed in developing lens cells. Expression begins at Hamburger and Hamilton (HH) stage 11, when the evaginating optic vesicle makes contact with the overlying head ectoderm (Fig. 2, left panel). L-Maf expression may indicate the completion of lens specification, because placode cells positive to L-Maf develop immediately afterward from the head ectoderm at the point of contact. L-Maf is later expressed by both epithelial and fiber cells. In Xenopus, L-Maf is expressed at stage 24, before placode formation, and gradually confined to the lens fiber cells (Ishibashi and Yasuda,2001). A new member of the maf family gene, mafA, recently has been cloned and characterized in humans and mice, which seems to be a homolog of chick L-maf (Kataoka et al.,2002). Preliminary expression of this gene has been detected in mouse eye tissue by reverse transcription-polymerase chain reaction.

Figure 2.

Initiation of Maf proteins in chick lens. Cartoon shows the stages (St.) of lens formation. In situ hybridization (lower panel) depicts the expression of three large Mafs. L-Maf is first expressed in the thickened head ectoderm before placode formation (left) at HH stage 11. c-Maf expression appears a few hours later than L-Maf in the invaginating lens placode (middle) at HH stage12 and MafB expression is detected in both epithelium and fiber cells (right) at much later stage after complete formation of lens vesicle (stage 26). Arrows indicate the respective mRNAs.

By using in situ hybridization, we also assayed the expression of c-Maf and MafB in the developing chick embryo (Fig. 2, middle and right panels). Expression of c-Maf commenced 3–4 hr later than L-Maf expression, corresponding to HH stage 11 for L-Maf (Fig. 2, left panel) and late HH stage 11 or 12 for c-Maf (Fig. 2, middle panel). In the lens, c-Maf is more highly expressed than L-Maf on embryonic day 5 (E5), as shown by Northern blot analysis (Yoshida and Yasuda,2002). In mouse, c-Maf expression is first evident in the presumptive lens ectoderm (PLE) on E9 and it becomes abundant in developing lens later on E10. The expression is preferably higher in primary fiber cells and less in epithelial cells (Kawauchi et al.,1999). In rat, the expression of c-Maf/maf2 is observed at the beginning of the process by which posterior cells differentiate to form primary fiber cells on E13. Expression of c-Maf is eventually restricted to the posterior fiber cells and totally absent from anterior epithelial cells (Sakai et al.,1997; Yoshida et al.,1997). MafB expression appears much later, after the complete formation of lens vesicle (Fig. 2, right panel). Northern blot analysis showed that L-Maf expression gradually decreases from E5 to E14, while c-Maf expression remains consistently high in chick. MafB expression, which is present at intermediate levels between E5 and E8, then declines (Yoshida and Yasuda,2002). Mouse MafB mRNA is apparent in the epithelial cells on E10.5 but not in fiber cells, while NRL is absent in early lens lineage during development (Liu et al.,1996; Kawauchi et al.,1999). In rat, Maf1/mafB is expressed in the differentiating epithelial cells and is confined to the equatorial region (Sakai et al.,1997; Yoshida et al.,1997). MafB expression in Xenopus appears in the PLE at stage 22 and then is restricted to the epithelial cells. MafB expression starts before L-Maf in this species; however, their concurrent expression has also been detected in the initial phase of lens induction (Ishibashi and Yasuda,2001). Both c-Maf and MafB are apparent in zebrafish lens during development, c-Maf is preferentially localized in fiber cells while MafB in all lens cells (Kajihara et al.,2001). In zebrafish, we have identified two additional Mafs, SMaf1 and SMaf2. SMaf1 has high sequence homology with chick L-Maf; however, it is not found in zebrafish lens tissue and, therefore, is the nonfunctional homolog of L-Maf (Kajihara et al.,2001). Similarly, the quail mafA gene, which is developmentally regulated in the neuroretina, is identical to chick L-maf (Benkhelifa et al.,1998).

The spatiotemporal expression of these genes suggests that different Mafs have distinct roles in lens development, probably by regulating the expression of a diverse set of genes. It is also possible that the maf genes exert some overlapping functions when cells require the activity of multiple maf genes to change some intrinsic property favorable to a particular state. For example, the expression of both c-Maf and L-Maf in cells at early stages of development may indicate that enhanced expression of both proteins is necessary to activate downstream genes to an extent sufficient for continued differentiation. In contrast, the short time period during which MafB is expressed suggests that this protein plays a critical role in the subsequent differentiation of fiber cells. Thus, the order and expression pattern of Maf proteins clearly regulates the genetic cascade for lens development. Furthermore, the initiation of lens development is closely associated with the onset of L-Maf expression, making this protein crucial for chick lens formation.

L-Maf, c-Maf, and MafB have varying length of hinge region. Although the homology of the bZIP and acidic domains is highly conserved among these three proteins, their hinge regions differ considerably (Ogino and Yasuda,1998). The differing sequences of the individual hinge regions may provide clues to their differential roles during embryogenesis. In our experiments, a luciferase assay using NIH3T3 cells has shown that deletion of the hinge region of c-Maf reduces transcriptional activity considerably (Yoshida and Yasuda,2002), indicating that the hinge and acidic domains of c-Maf act synergistically to regulate transcription. We have hypothesized that hinge may induce conformational changes in the acidic region, which ultimately enhances the ability of the latter to easily interact with general transcription factors. Consistently, by domain swapping experiments, it was demonstrated that c-Maf requires the activation domain together with the hinge region and bZIP domain for the optimal activity in transactivating γD-crystallin in rat lens cells (Civil et al.,2002). In contrast, deletion of the hinge region of L-Maf did not change its transcriptional activity, although we found that both the acidic and hinge regions of L-Maf are required for full activation of the δ-crystallin gene (Yoshida and Yasuda,2002). These findings suggest that the hinge region of L-Maf interacts physically with other factors that regulate δ-crystallin expression. EMSA using chimeric Maf proteins has shown that the bZIP region of MafB is important for its transactivation potential, although the bZIP region alone had no effect (Yoshida and Yasuda,2002). It has been found that c-Maf and MafB can equally activate rat γD-crystallin promoter when examined in CHO cells. However, in rat lens cells, a lower promoter activity was observed by MafB (Civil et al.,2002). This finding suggests that cellular environment is important so that other transcription factors specific to cell-type can bind to the close proximity to Maf binding site and necessarily determine MARE specificity for Maf proteins to exert their optimal transactivation potential.


Lens induction and development are triggered by several transcription factors. We have identified a transcriptional cascade headed by Pax6, a master regulatory gene in eye development, and we showed that L-Maf expression is dependent on Pax6 (Reza et al.,2002). By using in ovo electroporation, we performed a series of gain-of-function experiments that showed some interesting results. Prox1, a vertebrate homolog of the Drosophila prospero gene (Matsuzaki et al.,1992), is involved in the terminal differentiation and elongation of mouse lens fiber cells (Oliver et al.,1993). We have found that Prox1 can ectopically activate lens markers, including the crystallin gene, in the chick ectoderm (Reza et al.,2002). Ectopic expression of L-Maf in different parts of the head ectoderm in stage 10 embryos induced expression of Prox1, and these cells morphologically resembled normal placode cells. Recently, in situ hybridization with a c-Maf probe after ectopic expression of L-Maf in ectodermal tissue showed a very similar phenotype (H.M.R. and K.Y., unpublished data), in that all of these cells expressed high levels of c-Maf mRNA. In a similar experiment, we also could detect ectopic MafB mRNA expression (Fig. 3). These results suggest that L-Maf directly regulates expression of other maf genes, including c-maf and mafB, during the endogenous evolution of lens. Pax6 and Sox2 are two important upstream genes that have been shown to be indispensable for lens development in various species (Nishiguchi et al.,1998; Ashery-Padan et al.,2000; Kamachi et al.,2001; Reza et al.,2002). When we assayed the expression of Pax6 and Sox2 by overexpressing L-Maf to determine whether there was a feed-back regulatory loop, we found no evidence for one. Taken together, these findings suggest that, together with the other members of the maf gene family, L-Maf plays a critical role in the early differentiation process.

Figure 3.

A model shows the downstream genes regulation by Maf during lens development. The upstream yet unknown signaling factors together with Pax6 and Sox2 induce L-Maf in preplacode, which in turn initiates lens placode. Immediately, c-Maf and Prox1 as well as δ- and αA-crystallins are activated by L-Maf in the lens placode and lens vesicle is formed thereafter. In lens vesicle, L-Maf directly and through c-Maf and Prox1 regulates crystallins expression. Later, L-Maf activates the expression of MafB in differentiating lens fiber and also initiates the expression of β-crystallin and other fiber-specific genes cp95 and cp49. MafB is likely to perform the critical role in regulating the lens-specific genes (dotted line) at later stages, while c-Maf, however, is restricted to the lens epithelium to regulate different set of yet unknown genes.


The multiple roles of the Maf proteins in various developmental processes have been studied in a wide range of species. Some functions are common to many species, while others are divergent. Although it is reasonable to hypothesize that homologous genes may have similar functional properties, there may be species-specific differences. In cultured retina cells, ectopic expression of L-Maf was shown to induce lens differentiation, which was characterized by the expression of several crystallins and other fiber-specific genes. For example, retina cells transfected with a plasmid expressing L-Maf expressed αA- and δ-crystallin, and these crystallin-positive cells were found to be as elongated as normally differentiated lens cells (Ogino and Yasuda,1998). We also observed that L-Maf can induce crystallin expression in primary cultures of retinal pigment epithelium and brain cells. The RCAS-L-Maf was able to induce expression of the terminally differentiated lens fiber-specific genes, βB1-crystallin and filensin, in retina cultures (Ogino and Yasuda,1998). Similar data were also obtained when quail mafA was used in the transfection assay by using chick retina cultures (Benkhelifa et al.,2001). Immunofluorescence analysis using specific antibodies revealed clear expression of δ-, αA-, and βB1-crystallins in the cultures.

When chick embryos were electroporated with L-Maf, ectopic induction of αA- and δ-crystallins was observed. Forced expression of L-Maf also induced the expression of the noncrystallin fiber-specific genes, cp95 and cp49 (Remington,1993; Ireland et al.,2000), in different regions of the head ectoderm of chick embryos (N.S. and K.Y., unpublished data). These findings indicate that L-Maf essentially regulates expression of crystallin and fiber-specific genes during the process of normal lens development (Fig. 3). In contrast, c-Maf in mouse lens has been implicated to have no effect on transcriptional regulation of cp49 and cp115 (DePianto et al.,2003). However, several pieces of evidence indicate that c-Maf function is similarly crucial for lens fiber differentiation and crystallin expression in mouse. For example, while lens development takes place in the c-maf knock-out mouse, fiber differentiation is severely arrested, causing a hollow structure (Kawauchi et al.,1999; Kim et al.,1999; Ring et al.,2000). In these mutant mice, a few primary fiber cells were characterized by the weak expression of αB- and βB-crystallins; however, γ-crystallin was almost absent, suggesting that c-Maf essentially controls β- and γ-crystallins in mouse. These observations indicate that mouse c-Maf and chick L-Maf have similar functions. However, they are not functionally identical because L-Maf plays an important role in placode formation in chick (Reza et al.,2002). c-Maf in mouse does not do so, although c-Maf is found in the PLE before placode formation in this species too (Kawauchi et al.,1999). Ectopic expression of L-Maf in Xenopus leads to activation only of genes expressed in lens fiber while MafB induces both epithelium and fiber-specific genes (Ishibashi and Yasuda,2001). By using rat lens explants, it has been demonstrated that c-Maf is the most potent activator of γD-crystallin promoter in lens fiber cells (Civil et al.,2002).

We also found, however, that ectopic expression of L-Maf alone failed to activate δ-crystallin expression in the entire head ectoderm, although expression of αA-crystallin and CP49 and CP95 were unrestricted, suggesting that the regulation of δ-crystallin occurs through a mechanism unrelated to that regulating the other crystallins.

By a reporter assay using a Maf-responsive element from the αA-crystallin gene, we found that c-Maf and MafB enhanced the transcription of genes more strongly than L-Maf, suggesting that expression of L-Maf requires additional factors, which bind to a cis-element in the regulatory region separate from the Maf-responsive element of αA-crystallin (Yoshida and Yasuda,2002). We previously showed that the lens-specific expression of αA-crystallin is regulated by αCE1 and αCE2 elements in the upstream of the αA-crystallin gene, and we confirmed that CP2 binds to αCE1 element (Murata et al.,1998) Thus, we propose that ubiquitously expressed CP2 functions redundantly with L-Maf in transactivating expression of the αA-crystallin gene. In contrast, functions of c-Maf and MafB are independent of other factors in regulating αA-crystallin gene in chick as judged from reporter assay using different cell types (Yoshida and Yasuda,2002). However, in rat, it has been demonstrated to have a lens-specific partner other than AP1 for c-Maf, which determines the MARE specificity in the process of γD-crystallin regulation (Civil et al.,2002). It has also been shown that the highest promoter activity was detected by c-Maf when in vitro differentiating rat lens cells were transfected with the reporter plasmid containing −73/+10 rat γD-crystallin promoter element. MafB stimulated to a lesser extent and L-Maf completely failed to activate (Civil et al.,2002).

When we compared the ability of the three Mafs to induce expression of δ-crystallin in cultured retina, we found that L-Maf was more efficient than c-Maf or MafB. Northern blot analysis confirmed that L-Maf induced a higher level of δ-crystallin expression than c-Maf or MafB, suggesting that one or more factors cooperate with L-Maf in regulating the δ-crystallin gene (Yoshida and Yasuda,2002). Our recent results from a coexpression study on chick embryos showed that Sox2 cooperates with L-Maf to induce δ-crystallin expression by interacting with an enhancer element located in the third intron of the δ-crystallin gene (Shimada et al.,2003). The ability of Sox2 to bend DNA may promote the stable and sustained docking of L-Maf to the δ-crystallin enhancer. Another study reports that a putative Sox binding site and the −14/−7 positive element rather than MARE in the promoter region of βB2-crystallin mediates the c-Maf activity in rat lens (Doerwald et al.,2001). Sox2 participation can be anticipated in this case also. By following the expression pattern of the three Mafs in chick lens, we may be able to show that regulation of the later expressed crystallins in lens is the likely function of c-Maf and MafB in this tissue.


Unless all the proteins involved in lens initiation are isolated and studied, both separately and together, it is difficult to know which gene initiates lens induction. The pattern of transcription factor expression during embryogenesis suggests, however, that a single transcription factor is unable to trigger the complete formation of an organ. The interdependent activities of several transcription factors probably cause the induction of a preferential, potential, and inevitable molecule, which takes the absolute responsibility for organogenesis of a particular organ. In the case of chick lens formation, we and others have found that many transcription factors are involved in this process, but none of these factors, except for L-Maf, is unique in terms of expression pattern and function (for reviews, see Kondoh,1999; Ogino and Yasuda,2000; Chow and Lang,2001). Many aspects of L-Maf function indicated that this protein is necessary for lens fiber differentiation. However, we observed that targeted loss of L-Maf function in lens primordium inhibits placode formation, suggesting that L-Maf function is required for the initiation of lens placode (Reza et al.,2002). Because the other Mafs are not expressed as early in development as L-Maf, it is likely that this role of L-Maf is very specific. Therefore, we conclude that the initial role of L-Maf is pertaining to the placode formation, while later it becomes vitally involved in the process of fiber differentiation.

In mouse, c-Maf is expressed in the PLE before placode formation; however, it does not play any role in placode formation as evidenced from the c-maf null-mutant mice (Kawauchi et al.,1999; Kim et al.,1999; Ring et al.,2000). Although MafB expression has been exclusively detected in lens epithelium, it begins at the lens vesicle stage (Kajihara et al.,2001), suggesting that this protein again does not participate in placode formation rather, controls the expression of essential ectodermal genes that maintain epithelium in an undifferentiated state. During the initial stages of lens induction in Xenopus, the concurrent expression of MafB and L-Maf suggests that they probably play some redundant roles in placode formation (Ishibashi and Yasuda,2001).

The crucial part played by L-Maf in chick lens formation arises from several observations. First, expression of L-Maf starts in the presumptive lens ectoderm just after contact between the overlying surface ectoderm and the optic vesicle at HH stage 11, after which, the first morphologic structure of the lens, that is, the lens placode, develops. All placode cells extensively express L-Maf. In the developing lens, the early marker genes that encode structural and soluble proteins are δ- and αA-crystallins, which are expressed immediately after L-Maf expression in the lens placode. Furthermore, we have shown that L-Maf can bind directly to the upstream enhancer regions of these genes to promote their expression. L-Maf has been observed to regulate the expression of other Mafs, including c-Maf and MafB, which are endogenously expressed later in lens. In addition, L-Maf directly activates expression of the terminal differentiation factor Prox1, which is also expressed in lens placode later than L-Maf. Ectopic expression of L-Maf can efficiently induce expression of the δ- and αA-crystallins, as well as the fiber-specific genes, βB1-crystallin, cp95, and cp49, in retina cultures and embryos, whereas c-Maf and MafB activate these genes minimally. Loss of L-Maf function essentially suppresses the expression of differentiation factors, fiber-specific genes, and soluble proteins in lens tissue, and L-Maf inactivation in lens primordium completely blocks lens formation. The absence of Pax6 expression in head ectoderm causes no-lens phenotype, which is also observed when L-Maf function was blocked in head ectoderm. Furthermore, L-Maf can rescue the dominant-negative Pax6 effect in growing lens. Pax6 and Sox2 are expressed earlier than L-Maf in head ectoderm and play significant roles in lens development, as well as δ-crystallin expression. We followed the ectopic expression of L-Maf when these two genes were coexpressed in chick embryos. In addition, we have located recently the putative binding sites for both Pax6 and Sox2 in the regulatory region of the L-maf gene (H.M.R. and K.Y., unpublished data), indicating that the significant portion of δ-crystallin is directly regulated by L-Maf, which is turned on by Pax6 and Sox2 during lens formation. Finally, during lens development, L-Maf displays both proliferating and differentiating properties. Taken together, all these observations indicate that L-Maf differs extensively from the other molecules involved in eye development. Thus, L-Maf acts as a central molecule in chick lens development and can directly regulate many if not all of the crystallins expressed in lens.


Growth factors are essential regulators of cell proliferation and differentiation. Maf proteins have defined roles as differentiation factors, thus maintain a regulatory relation with several growth factors. Fibroblast growth factor (FGF) and bone morphogenetic protein (BMP) constitute two important signaling pathways that have been found active throughout the lens development program in mouse and chick (Chow and Lang,2001). Both BMP4 and BMP7 play significant roles in lens induction in mouse (Dudley et al.,1995; Furuta and Hogan,1998; Wawersik et al.,1999). Recent works have shown that BMP4 is involved in the differentiation and elongation of primary lens fiber cells as evidenced by the suppression of lens fiber differentiation when BMP ligand inhibitor, noggin, was overexpressed in mouse lens explants and also in chick embryos (Belecky-Adams et al.,2002; Faber et al.,2002). Similarly, several FGF family members have been implicated in stimulating lens fiber differentiation from epithelial cells and favoring the expression of lens-specific genes (Chamberlain and McAvoy,1987; Schulz et al.,1993; Lovicu and Overbeek,1998; Vogel-Hopker et al.,2000; Le and Musil,2001). Transgenic mice expressing dominant-negative FGF receptor (fgfr1IIIc) in the PLE exhibit defective lens placode formation, indicating that FGF signaling is required in early phase of lens induction (Faber et al.,2001). In chick, a direct relation between FGF8 and L-Maf has been observed as implanted FGF8 beads induced L-Maf–positive lens placode (Vogel-Hopker et al.,2000). We have found that optic vesicle expression of BMP4 and FGF8 are similarly important for the maintenance of L-Maf in differentiating lens fibers after optic cup formation (H.M.R. and K.Y., unpublished data).

Maf function is highly dependent on phosphorylation to be mediated by different signaling pathways. FGF2 has been found to repress the expression of lens-specific genes at the proliferative phase in primary cultured lens cells. The data indicate that FGF2/ERK signaling phosphorylates L-Maf in lens epithelium where the product is degraded (Ochi et al.,2003). Consistently, FGF2 has also been demonstrated to negatively regulate c-Maf in the regulation of γD-crystallin in rat lens (Civil et al.,2002). However, in fiber cells, some mechanisms antagonistic to FGF2 signaling probably enhance L-Maf stabilization; thus, L-Maf continues to exert fiber differentiation activity. After day 3 following treatment of lens cells with FGF2, expression of lens-specific genes was enhanced and persisted (Ochi et al.,2003). Therefore, it is predictable that, once L-Maf is stabilized by some mechanisms, latter regulation by FGF2 may augment transcriptional activity of L-Maf and triggers the expression of lens-specific genes such as crystallins during lens fiber differentiation. A parallel study shows that phosphorylated MafA exerts higher transcriptional activity in retina cultures (Benkhelifa et al.,2001). It is possible that retina cells have intrinsic machinery that may protect phosphorylated MafA from degradation. However, further study is needed to determine how different Mafs are stabilized in the fiber cells. Different growth factors are important candidates to act crucial role in this context.


Maf proteins play important roles in tissue-specific gene regulation and cell differentiation. From extensive studies in a variety of experimental models, it is now clear that the Maf proteins play critical roles in lens differentiation (Ogino and Yasuda,1998; Kim et al.,1999; Kawauchi et al.,1999; Ring et al.,2000; Ishibashi and Yasuda,2001; Reza et al.,2002; Muta et al.,2002; Ochi et al.,2003; Shimada et al.,2003). The spatiotemporal expression patterns of L-Maf, c-Maf, and MafB indicate the importance of each of these genes in lens formation. Although chick L-Maf is known to be involved from induction to maturation of lens, many questions remain unresolved regarding the roles of this protein. We have observed that, at low levels of L-Maf expression, crystallins are abundantly expressed. The latter are associated with a high level of expression of c-Maf in lens cells, indicating that c-Maf regulation is somewhat different from that of L-Maf (H.M.R and K.Y., unpublished data) and that L-Maf and c-Maf regulate different sets of genes at a later stage, if not at an early stage, of differentiation. Diffusible factors that are spatiotemporally regulated may be important in maintaining the stability and functions of Maf proteins. It will be interesting to investigate how the different Maf proteins regulate the expression of crystallin and noncrystallin fiber-specific proteins. Once we determine the downstream genes regulated by Maf, we will be able to study the interactions of each with each Maf protein, thus providing a better picture of the function of each Maf in lens development.

Recent studies demonstrate that developmentally regulated MAF is essential for proper development of eye in human and that its mutation causes different ocular diseases (Jamieson et al.,2002,2003; Lyon et al.,2003). Thus, MAF is now considered as an ocular disease-causing gene together with PAX6, PITX2/PITX3, and FOXE3 (Hanson et al.,1994; Semina et al.,1998,2001). Substitution of a highly conserved arginine with proline at residue 288 (R288P) results in pulverulent cataract, anterior segment dysgenesis, microcornea, and iris coloboma in human. This type of mutation can be closely linked to the mouse mutation R291Q that develops cataract as well (Jamieson et al.,2002; Lyon et al.,2003). R291Q mutation in the basic region of the DNA-binding domain of MAF transcription factor alters the DNA binding affinities, suggesting that the mutant protein is likely to exert differential downstream effect and ultimately perturbs the expression of lens-specific genes. In another case, cataract was associated with translocation, t(5;16)(p15.3;q23.2). The 16q23.2 translocation break point lies in the genomic control region of MAF gene, suggesting an altered regulation of MAF (Jamieson et al.,2002,2003). These findings implicate that MAF genes are directly involved in some genetically controlled eye diseases or abnormalities that can be studied elaborately by dissecting Maf functions in various species.


We thank Drs. K. Kataoka and Y. Kageyama for critical comments and valuable discussions of the manuscript. This work was supported in part by Grants-in-Aid for the 21st Century COE Program from the Ministry of Education, Science, Sports and Culture of Japan. H.M.R. is a recipient of COE fellowship.