Analysis of the regulation of lin-41 during chick and mouse limb development

Abstract

We have cloned the chicken and mouse orthologues of the Caenorhabditis elegans heterochronic gene lin-41. During limb development, lin-41 is expressed in three phases over developmental time and most notably is associated with the developing autopod. Using chicken and mouse mutants and bead implantations, we report that lin-41 is genetically and biochemically downstream of both the Shh and Fgf signaling pathways. In C. elegans, it is proposed that lin-41 activity is temporally regulated by miRNAs (let-7 and lin-4) that bind to complementary sites in the lin-41 3′-untranslated region (UTR). Taking a bioinformatics approach, we also report the presence of potential miRNA binding sites in the 3′-UTR of chicken lin-41, including sites for the chicken orthologues of both C. elegans let-7 and lin-4. Finally, we show that these miRNAs and others are expressed in the chick limb consistent with the hypothesis that they regulate chicken Lin-41 activity in vivo. Developmental Dynamics 234:948–960, 2005. © 2005 Wiley-Liss, Inc.

INTRODUCTION

In the vertebrate limb, appropriate growth and patterning is dependent on the activities of well-defined signaling centers. The apical ectodermal ridge (AER), a thickened epithelium at the apex of the distal limb, controls outgrowth along the proximal–distal axis through the activity of the fibroblast growth factors (FGFs; Saunders, 1948; Mariani and Martin, 2003). Fgfs from the AER permit continued outgrowth of the limb by preventing apoptosis and ensuring proliferation of responsive cells in the underlying subridge mesoderm (Rowe et al., 1982; Dudley et al., 2002). A group of mesodermal cells located along the posterior border of the developing limb called the zone of polarizing activity (ZPA) controls growth and patterning of the anterior–posterior (AP) axis. ZPA cells synthesize and secrete Sonic hedgehog (SHH), a protein critical for appropriate patterning of the AP axis (Echelard et al., 1993; Riddle et al., 1993; Lopez-Martinez et al., 1995; Marti et al., 1995; Pearse and Tabin, 1998). Additionally, descendants of ZPA cells directly contribute to mouse posterior digits (Harfe et al., 2004).

Limb skeletal determination occurs in a proximal–distal (PD) manner over time (Saunders, 1948; Summerbell, 1974; Rowe and Fallon, 1982). The observation that more proximal skeletal elements differentiate before more distal elements indicates the presence of an inherent timing mechanism in limb development (Searls, 1965; Thorogood and Hinchliffe, 1975). Interestingly, recent work has raised the possibility that a timing mechanism may be essential to understand patterning along the AP axis (Yang et al., 1997; Harfe et al., 2004). Originally, it was hypothesized that SHH secreted by ZPA cells acts as a morphogen to specify cell fate in a concentration dependent manner (Wolpert, 1969; Tickle et al., 1975; Riddle et al., 1993). Recently, a new model emphasizes the importance of developmental time as well as SHH concentration and proposes that the longer a cell expresses SHH, the more posterior its fate (Harfe et al., 2004).

Surprisingly little progress has been made in identifying genes that regulate the timing of developmental events in the vertebrate limb. Although progress has been made in other developing vertebrate systems (e.g., somitogenesis; reviewed by Pourquie, 2003), considerable insight has come from studying the heterochronic pathway in the nematode Caenorhabditis elegans (reviewed in Ambros, 2000; Rougvie, 2001; Pasquinelli and Ruvkun, 2002). Specifically, postembryonic development of C. elegans proceeds through four larval stages (L1–L4) before emergence of the adult. Orderly progression through these larval stages is critical for the realization of the adult phenotype. Each larval stage is characterized by the formation of specific cell types and execution of stage-specific programs. The heterochronic gene pathway controls the timing of these developmental events.

Progression from L4 to the adult developmental program is partially controlled by the heterochronic gene lin-29, which codes for a zinc-finger transcription factor (Rougvie and Ambros, 1995). Worms lacking LIN-29 reiterate the L4 program and do not initiate an adult program (Ambros and Horvitz, 1984). Overexpression of LIN-29 induces precocious initiation of adult programs before completion of the L4 program (Bettinger et al., 1996). Thus, temporal control of LIN-29 activity ensures timely completion of the L4 program and subsequent initiation of the adult program.

One factor known to temporally regulate LIN-29 function is Lin-41 (Slack et al., 2000). C. elegans LIN-41 contains an amino-terminus RING finger domain, two B-boxes, a Coiled-coil and NHL domain. This cluster of conserved motifs, excluding the NHL domain, places LIN-41 in the RING Finger, B-Box Coiled-Coil (RBCC) family of proteins that is now also called TRIpartite Motif (TRIM) proteins (Saurin et al., 1996; Borden, 1998; Reymond et al., 2001). RBCC/TRIM protein family members contain various combinations of the aforementioned domains, whereas the NHL protein domain is only associated with a subset of RBCC/TRIM proteins (Slack and Ruvkun, 1998).

Lin-41 was first isolated and functionally characterized in C. elegans, and loss of Lin-41 function results in precocious LIN-29 activity (Slack et al., 2000). Although the precise mechanism by which Lin-41 regulates Lin-29 is not known, regulation of Lin-41 is better defined. In the 3′-untranslated region (UTR) of lin-41 are complementary binding sites for the miRNAs let-7 and lin-4 (Slack et al., 2000). Recent work has shown that two let-7 complementary sites (LCSs; terminology of Vella et al., 2004a, b) are necessary and sufficient for down-regulation of a reporter construct fused to the 3′-UTR of lin-41 in C. elegans and zebrafish (Kloosterman et al., 2004; Vella et al., 2004a). These results indicate that regulation of lin-41 by let-7 may be conserved across species in vivo.

Two reports demonstrate that orthologues of C. elegans LIN-28 and let-7 are temporally expressed in vertebrate embryos (Moss and Tang, 2003; Pasquinelli et al., 2000). These data combined with the identification of probable vertebrate orthologues of lin-41 strongly suggest conservation of the heterochronic pathway during vertebrate embryogenesis (Slack et al., 2000; Kloosterman et al., 2004). Here, we describe the cloning and expression pattern of the chicken orthologue of C. elegans lin-41 in the developing embryo. We focus our analysis on the developing limb and report that clin-41 is expressed in three distinct phases during development. Furthermore, we perform functional analysis to determine the epistatic relationship between lin-41 and the main limb bud signaling centers in the chick and mouse. Finally, we use a bioinformatics approach to determine potential miRNA binding sites in the 3′-UTR of clin-41 and Northern analysis to determine whether these miRNAs are expressed in the chick limb.

RESULTS

cLin-41 Cloning, Structure, and Homology

A 1.4-kilobase chicken expressed sequence tag (EST) isolated from a subtractive hybridization between limbless mutant limb buds and wild-type limb buds was identified with high similarity to the carboxy-terminus of the C. elegans heterochronic gene LIN-41 (Fig. 1). Using this EST, we screened several chick limb specific cDNA libraries and a chicken genomic library and isolated an additional 1.2 kilobases of 3′-UTR and 1.0 kilobases of putative open reading frame. Based on Northern analysis, the chicken lin-41 (clin-41) transcript is predicted to be approximately 5.0 kilobases (not shown). We were unable to identify a convincing start codon in any EST clone and were unable to deduce a plausible start codon from the published chicken genome.

Figure 1. Comparison of C. elegans LIN-41A with chicken, mouse, human, rat, and zebrafish LIN-41predicted amino acid sequences. Alignments were made with ClustalW and shaded with MacVector. Blackened residues indicate identical residues conserved in over half of the sequences aligned. Overall, C. elegans and chicken LIN-41 proteins are 34% identical and 48% similar. Colored boxes indicate the functional domains of the vertebrate LIN-41 proteins and are coded by the schematic LIN-41 protein in the inset. The C. elegans Ring Finger domain in underlined. Within the domains, functional residues are strongly conserved between C. elegans and vertebrates (e.g., asterisks mark conserved cysteine residues in B-box 1 and 2).

Comparison of chicken LIN-41 to C. elegans LIN-41 reveals an overall similarity of 48%. There is also clear conservation of functional residues within the various protein domains. For example, the characteristic cysteine residues that define the B-boxes are highly conserved between C. elegans and chicken LIN-41 proteins (Fig. 1, asterisks). Searching other published genomes with our chicken sequence allowed us to identify Lin-41 orthologues in human, mouse, rat, and zebrafish. Comparison of LIN-41 among these vertebrates shows high similarity across the entire protein and almost complete identity within the NHL and Coiled-coil domains and B-box 2 (Fig. 1). The predicted zebrafish LIN-41 protein is incomplete but does have these domains (Frank Slack, personal communication; see Schulman et al., 2005).

There is a dramatic drop in similarity among vertebrate proteins at the amino-terminus beginning near B-box 2. All proteins except for the rat orthologue have B-box 1 and only rat and human have a RING finger domain (Frank Slack, personal communication; see Schulman et al., 2005). At this time, the differences in the predicted vertebrate LIN-41 proteins are most likely due to annotation errors. For example, based on our alignments, the start methionine predicted for mouse LIN-41 by NCBI is actually an internal methionine. Also, analysis of the genomic DNA sequence from mouse and chicken revealed sequence gaps within the Lin-41 locus as well as GC-rich sequence. We believe these factors contributed to the annotation errors and that, when corrected, all vertebrate LIN-41 proteins will contain a RING finger domain, two B-boxes, a Coiled-coil domain, and NHL domain.

Chicken lin-41 Expression Domains

We performed whole-mount and section in situ analysis on the chick embryo over a range of stages and report that clin-41 is expressed in many tissues of the developing embryo, including the pharyngeal arches, somites, frontal–nasal mass, mesonephros, liver, and vasculature. In all tissues and organs, clin-41 expression undergoes spatial change over developmental time. This type of dynamic expression is consistent with the possibility that cLin-41 has a specific developmental function. Here, we will focus on the developing limb.

During normal limb development in the chick, limb buds first appear as thickened mesoderm at Hamilton and Hamburger stage 17. Before this time, beginning as early as stage 12, genes critical for limb bud initiation mark the future limb territories in the lateral plate mesoderm (e.g., Wnt and Tbx genes: Gibson-Brown et al., 1996; Gibson-Brown et al., 1998; Isaac et al., 1998; Kawakami et al., 2001). Unlike these early limb genes, clin-41 is not restricted to the future limb territories but instead appears to be expressed at low levels throughout the lateral plate mesoderm (not shown).

Once the limb buds emerge from the body wall, clin-41 is expressed ubiquitously in the mesoderm of both the wing and leg; it is not expressed in the ectoderm (Figs. 2A,A′, 3A). This is the first phase of clin-41 expression where initial ubiquitous expression of clin-41 in the mesoderm remains unchanged through approximately stage 21. The second phase of expression begins at stage 22 where clin-41 expression becomes progressively reduced throughout the limb mesoderm except in the distal, subridge mesoderm (Fig. 2B,B′). Within this distal domain of high clin-41 expression, there is an asymmetry across the AP axis with the highest expression located in the posterior (postaxial) half of the distal expression domain.

Figure 2. Three phases of clin-41 expression in the chick wing and leg. In all figures, the limbs are oriented with the posterior limb on the bottom and distal limb facing right. A,A′: Phase one expression of clin-41 in stage (st.) 20 wings (A) and legs (A′) is ubiquitous throughout the limb mesoderm. B,B′: Initial down-regulation of clin-41 in the proximal limb marks the transition to phase two expression in st. 22 wings (B) and legs (B′). C,C′,D,D′: Phase two expression in st.23 wings (C) and legs (C′) and at st. 24 wings (D) and legs (D′). clin-41 is expressed at high levels in the distal limb with a posterior bias, whereas reduced expression remains detectable in more proximal locations. Note the stronger down-regulation of clin-41 in the zone of polarizing activity (ZPA) region of the leg (arrow). E: Phase three of clin-41 expression in the legs at st. 27 (middle), and st. 29 (far right). E: (Far left) A st. 26 leg showing transition to phase three. clin-41 expression is detectable in muscle cells in the proximal limb. In the st. 27 limb, clin-41 is expressed at the distal end of the developing metacarpals. The st. 29 leg shows differential clin-41 expression in the developing digits of the foot but is absent in the future joint region.

Figure 3. clin-41 Section in situ hybridizations. A,B: Transverse section through a stage (st.) 20 leg buds showing clin-41 phase one expression through out the mesoderm (A) and a neighboring section showing Shh expression restricted to the posterior mesoderm (B). Note the cells that express Shh also express high levels of clin-41 at this stage. D,E: Transverse section through st. 24 leg buds showing clin-41 phase two expression (D) and a neighboring section showing Shh expression (E). At this time, clin-41 is expressed in the distal, subridge mesoderm, and down-regulated in Shh expressing cells (compare arrow in D with Shh expression in E). F: Transverse section of st. 24 legs. clin-41 is expressed in the distal limb and down-regulated in the core mesoderm; expression at the periphery is in the dorsal (D, arrow) and ventral (V, arrow) muscle muscles and vascular cells (arrows with asterisk). G,H: Neighboring transverse section of MyoD expression in both dorsal (D, arrow) and ventral (V, arrow) muscle masses and clin-41 expression in both dorsal (D, arrow) and ventral (V, arrow) muscle masses. clin-41and MyoD are coexpressed in muscle cells.

In phase two, reduced clin-41 expression remains visible in the mesoderm of both the wing and leg. However, section in situ shows this reduced expression domain is restricted to the vasculature (Fig. 3F, arrows with asterisks) and dorsal and ventral muscle masses (Fig. 3F, arrows with D and V, respectively). clin-41 expression is excluded from the central core mesoderm and avascular mesoderm. This reduced expression domain accounts for the apparent “background” seen in both the wings and legs from stage 22–24 and partially obscures the asymmetric expression pattern along the AP axis in the distal limb (Fig. 2B–D,B′–D′). This reduced expression is more pronounced in the wing compared with the leg.

We note that, during development of the in situ hybridizations, the strong distal–posterior expression of clin-41 develops much more rapidly than the expression in the vasculature and muscle masses. The section in situ data demonstrate that what may be interpreted as “background” in the whole-mount in situ hybridizations actually represents the differential mesodermal expression of clin-41 throughout phase two. Transition to phase two expression patterns occurs earlier in the leg bud, becoming obvious at late stage 22, but occurring at stage 23 in the wing, so that, at stage 24, high level clin-41 expression is clearly restricted to the distal limb, displaying a strong posterior bias in both wing and leg (Fig. 2C,C′,D,D′). This asymmetry is always more clearly observed in the leg than the wing.

The distal, posterior expression domain of clin-41 remains prominent through stage 26/27, at which time it begins to be down-regulated, marking the transition to phase three expression patterns (Fig. 2E, left limb, middle limb). At this time, clin-41 expression also becomes pronounced in developing muscle in the proximal limb. To confirm clin-41 expression in muscle, section in situ hybridization was performed on neighboring sections using probes for clin-41 and MyoD, a transcription factor that marks skeletal muscle cells. clin-41 and MyoD are coexpressed in multinucleated muscle cells (Fig. 3G,H).

In the third and final expression phase, clin-41 expression domains are restricted to developing autopod structures, with the exception that there is continued expression in developing muscle (Fig. 2E, right). In the autopod, clin-41 shows differential expression along the PD axis of the developing digit, with highest levels at the most distal region of the developing metatarsals and metacarpals. As the digits continue to grow, clin-41 is expressed in several distinct regions of the digit primordium. It is expressed at low levels in the condensing cartilage, at higher levels in the cells surrounding the developing digit, and at its highest level in the most distal region of the digit. clin-41 is not detectable in the digital joints.

clin-41 Relationship to Limb Signaling Centers

clin-41 expression is dependent upon Fgf signaling from the aER.

Given the dynamic and asymmetric expression patterns of clin-41, we sought to investigate its potential regulation by key signaling molecules associated with the signaling centers of the limb. Because clin-41 is differentially expressed along the PD axis of the limb, we first began by asking whether clin-41 was dependent upon the AER and Fgf signaling. Previous studies from this laboratory (Rowe et al., 1982) and from Dudley and coworkers (2002) have shown that AER removal before stage 23 results in apoptosis of subridge mesodermal cells, whereas removal of the AER at later times results in decreased cell proliferation. Because of these confounding events, it is not possible to simply remove the AER and assay for changes in gene expression in the subridge mesoderm. To circumvent these problems, we performed whole-mount in situ hybridization in chicken and mouse mutants with disrupted AER and Fgf signaling.

The chick limb mutant limbless fails to form an AER and also lacks Shh expression (Grieshammer et al., 1996; Noramly et al., 1996; Ros et al., 1996). However, in limbless, a normal appearing limb bud does form initially but begins to regress at late stage 19. The limbless mutant limb bud is an ideal system to determine whether a gene is dependent upon Fgf signaling from the AER through stage 19. In limbless, clin-41 is briefly expressed but rapidly down-regulated, becoming undetectable during stage 17 (Fig. 4B). This observation is consistent with our identifying clin-41 as being differentially expressed from a subtractive hybridization screen between limbless and wild-type limb buds.

Figure 4. lin-41 expression depends on Fgf signaling from the apical ectodermal ridge (AER). In all figures, the limbs are oriented with the posterior limb on the bottom and distal limb facing right. A,B:clin-41 is expressed throughout the limb mesoderm in stage (st.) 18+ wild-type (WT) leg buds (A) but undetectable in limbless leg buds (B). However, clin-41 is still expressed in the somites of both WT and limbless embryos. C–F:mlin-41 phase two expression in dpc 10.5/10.75 WT forelimb (C) and hindlimbs (E). In Fgf4^−/−; Fgf8^−/− mutant forelimbs (D), mlin-41 expression is altered and only expressed at low levels in Fgf4^−/−; Fgf8^−/− mutant hindlimbs (F).

Because other AER signaling molecules are absent in limbless in addition to Shh, we assayed for lin-41 expression in the developing limbs of Fgf4^−/−; Fgf8^−/− double mouse mutants (Sun et al., 2002) using a mouse specific lin-41 probe (mlin-41). In these mutant limbs, the subridge mesoderm does not undergo apoptosis or reduced cell proliferation that normally accompanies AER removal. mlin-41 expression is greatly reduced in the hindlimb of these mutants but low levels are still detectable. This reduction is less apparent in the forelimb (compare Fig. 4D and F). The observed differences in mlin-41 expression between the two limbs can be explained by the brief burst of FGF8 and SHH activity in the forelimb due to the later gene inactivation of Fgf8 by Cre in the forelimb compared with the hindlimb (Sun et al., 2002).

From these combined results, we propose that, during normal limb development, clin-41 expression is dependent upon the AER and more specifically on the activity of Fgf signaling from the AER. It should be noted that, in both the wing and leg of limbless and the hindlimb of the Fgf4^−/−/8^−/− mouse mutant, absence of Fgf signaling also results in absence of Shh signaling. This finding, combined with the AP asymmetry of clin-41 expression throughout limb development, led us to analyze whether altered or loss of Shh signaling would impact clin-41 expression.

clin-41 expression is maintained by SHH.

clin-41 is coexpressed initially at high levels in the ZPA and Shh-expressing cells through phase one and part of phase two in the wild-type limb. Of interest, by stage 23/24, clin-41 expression is noticeably down-regulated in cells that also express Shh but maintained at higher levels in neighboring non-Shh expressing cells (Fig. 3D,E; see also Fig. 2D′, arrow). To better define the relationship between clin-41 and Shh, we first characterized lin-41 expression in ozd chick mutants and Shh^−/− mice; both lack SHH activity in the developing limb (Chiang et al., 2001; Kraus et al., 2001; Ros et al., 2003). In these mutants, AER development is initially normal and lin-41 phase one expression is also normal, but during the transition to and throughout phase two, lin-41 expression is drastically altered. Instead of expression remaining high distally with a posterior bias, lin-41 becomes restricted to the very posterior border of the limb (Fig. 5B,H). Furthermore, low-level expression normally detected throughout the rest of the limb mesoderm in wild-type limbs is undetectable in ozd and Shh^−/− limbs. By stage 24 in ozd limbs, clin-41 is undetectable throughout the mesoderm (Fig. 5I).

Figure 5. clin-41 Expression is maintained by Shh in the limb. A,D,G:lin-41 expression in wild-type (WT) chick and mouse limbs at the indicated stages. B,C:clin-41expression is greatly down-regulated and confined to the posterior mesoderm in stage (st.) 23 ozd leg buds (B) but up-regulated throughout the distal mesoderm in ta² leg buds (C). E,F,H:mlin-41 expression is undetectable in day post coitum (dpc) 10.5 mouse forelimbs (E) and reduced and confined to the posterior in the mouse hindlimb (H). In the Xt^J mouse forelimb, mlin-41 is up-regulated throughout the anterior mesoderm (arrow). I:clin-41 is undetectable in st. 24 ozd leg buds. J,K: Four hours after implantation of a SHH-soaked bead, clin-41 is up-regulated (arrow in J; the asterisk marks the bead) and maximally up-regulated after 8 hr (arrow in K; the asterisk marks the bead). L,M: Ectopic, anterior Shh expression in the Silkie leg (L, arrow) and ectopic clin-41 (M, arrow) in an age-matched Silkie leg coincident with the ectopic Shh expression.

We next examined clin-41 expression in the ta² chick mutant, a mutant that has constitutive Shh pathway activation throughout the limb mesoderm in the absence of ligand (Caruccio et al., 1999). In these limbs, clin-41 is greatly up-regulated throughout the distal, subridge mesoderm across the entire AP axis (Fig. 5C). Interestingly, clin-41 is undetectable in the ZPA of the ta² mutant, a situation more pronounced in the leg than the wing. These results indicate that clin-41 may be up-regulated in target cells by Shh signaling.

In some cases of polydactyly, Shh is ectopically expressed in the anterior limb mesoderm. We, therefore, determined whether clin-41 expression would be ectopically up-regulated in the legs of the polydactylous Silkie breed chicken and in the limbs of the mouse mutant Extra Toes^J (Xt^J; Johnson, 1967; Hui and Joyner, 1993). In both cases, ectopic anterior clin-41was detected. In the Silkie leg, ectopic, anterior clin-41 expression (Fig. 5M, arrow) was coincident with the ectopic, anterior Shh patch (Fig. 5L, arrow), whereas in Xt^J, mlin-41 was more broadly up-regulated throughout the anterior limb mesoderm (Fig. 5F, arrow).

Finally, to determine how directly the Shh signaling pathway interacts with clin-41, beads soaked in SHH protein were implanted into the anterior limb mesoderm. Up-regulation of clin-41 expression was observed 4 hr after bead implantation (Fig. 5J, arrow) and maximal up-regulation was attained after 8 hr (Fig. 5K, arrow). clin-41 expression remained up-regulated when assayed 24 hr later. These results, combined with the clin-41 expression patterns in the previously mentioned Shh mutant limbs, leads to the hypothesis that SHH is necessary to maintain clin-41 expression during normal limb development and can rapidly up-regulate its expression in ectopic domains.

Potential regulation of vertebrate lin-41 by miRNAs.

By performing genetic screens in C. elegans, mutations in the lin-41 gene were shown to suppress let-7 mutations, placing lin-41 genetically downstream of let-7 (Slack et al., 2000). Six putative let-7 complementary sites (LCSs) were identified in the 3′-UTR of lin-41, and later studies demonstrated that let-7 could directly bind them. Only two LCSs were necessary to down-regulate a reporter in vivo (Vella et al., 2004a).

Because of this regulation in C. elegans and because it is known that let-7 is conserved in the chicken and functionally active in the mouse limb (Pasquinelli et al., 2000), we used a bioinformatics approach to ask whether any miRNA binding sites are present in the clin-41 3′-UTR. We used an algorithm to systematically align and score all Gallus gallus miRNA sequences deposited in the miRNA Registry against the full-length clin-41 3′-UTR (see the Experimental Procedures section). let-7 scored well, as did miR-125b, a homologue of C. elegans lin-4, a second miRNA proposed to regulate C. elegans lin-41 expression (see Table 1; Slack et al., 2000).

Table 1. miRNA Binding Site Prediction Algorithm Output

Duplex score	ΔG	Aln. score	Location	miRNA
106.4	−23.4	83.0	1236–1257	miR-222b
103.1	−19.1	84.0	14–35	miR-125b
102.4	−20.9	81.5	263–284	miR-15a
97.5	−23.5	72.0		C. elegans let-7
94.9	−16.9	78.0	61–92	let-7c, al-3
93.7	−19.7	74.0	1705–1726	let-7k, f
93.6	−16.6	77.0	149–170	let-7f
85.8	−17.8	68.0	2110–2131	let-7d
83.1	−17.1	66.0	221–248	let-7d

For reference, we ran this algorithm using let-7 and lin-41 3′-UTR sequence from C. elegans, and the interaction produced a score of 97.5. In chicken, five potential LCSs were identified, and the interactions were scored as follows: LCS#1- 83.1; LCS#2- 91.1-93.6; LCS#3- 84.5-94.9; LCS#4- 93.7; LCS#5- 85.8. There is a range of scores at sites #2 and #3, because 11 different Gallus gallus homologues of let-7 are present in the miRNA Registry and several were identified as potential regulators at each individual site. Other miRNA scoring exceptionally well were mir-222b (104.6), miR-125b (103.1), and mir-15a (102.4, see Fig. 6). The algorithm also takes into account potential conservation of miRNA binding site by aligning the 3′-UTRs, in this case, from human, mouse, rat, and chicken. LCS#1 and #2 in clin-41 3′-UTR were strongly conserved in human, mouse, and rat. Not only was their general location within the 3′-UTR conserved in each species, but the spacing between each predicted site was also conserved. It is notable that the distance between two LCSs was shown to be functionally important for regulation by let-7 in a reporter assay in C. elegans and is presumed to be true for lin-41 regulation by let-7 in vivo (Vella et al., 2004a, b).

Figure 6. Predicted miRNA-binding sites in clin-41 3′-untranslated region (UTR). Schematic drawing of predicted miRNA interactions with clin-41 3′-UTR miRNA binding sites. Structures are based on m-Fold diagrams (Zucker, 1989). In all structures, the miRNA sequence is on the bottom with the 5′ ribonucleotide to the right. The 3′-UTR sequence is on top and the 5′ ribonucleotide is to the left.

To confirm that a miRNA regulates a specific mRNA during development, it is important to show that both transcripts are expressed in the same tissue at the same developmental time. To do this, we performed Northern analysis on limb tissue for the highest scoring miRNA predicted to have binding sites in the clin-41 3′-UTR (Fig. 7). We chose miR-222b, the highest scoring miRNA and the two potentially conserved regulatory miRNAs miR-125b and let-7. We isolated total RNA from chick limb buds collected from stage 19/20 through stage 26 and divided the limb into proximal and distal halves at each stage. Whereas all the highest scoring miRNAs predicted to regulate clin-41 were expressed in the limb at all times, obvious differential expression of any miRNA was not detected over developmental time or in either region of the limb. Together, our bioinformatic predictions and Northern data are consistent with the hypothesis that miRNAs may regulate Lin-41 in the chick limb.

Figure 7. miRNA expression in the chick limb over developmental time. Northern Blot of miRNAs with predicted complementary binding sites in clin-41 3′-UTR miRNA. In all cases, both the precursor and mature forms of the miRNA can be detected. Several limbs were collected for each stage and dissected into proximal and distal halves. Caenorhabditis elegans RNA was loaded as a positive control for let-7. 5S rRNA was used as an internal loading control.

DISCUSSION

In this study, we report the cloning of the chicken orthologue of C. elegans lin-41. During limb development, clin-41 is expressed in three phases, and we present data consistent with the hypothesis that clin-41 is regulated by FGFs from the AER and SHH from the ZPA. Furthermore, potential miRNA binding sites were identified in the 3′-UTR of chicken lin-41, including five sites for let-7 and one each for miR-125b, miR-225b, and miR-15a. These miRNAs are expressed in the chick limb at the appropriate developmental time consistent with the hypothesis that miRNAs regulate cLin-41 function.

We show that the structure of LIN-41 is conserved in vertebrates from the NHL domain through B-box 2. Because both the chicken and human have B-box 1, it is likely that mouse and rat do as well. Furthermore, it is also likely that chicken and mouse LIN-41 have a RING finger domain. While it is possible that there are species differences in LIN-41 protein structure, we believe that most of the problem in accurately describing vertebrate LIN-41 protein structure stems from this gene being incorrectly annotated in vertebrate genomes. This annotation problem could result from the difficulty in correctly predicting the protein structure of a gene that not only spans almost 100 kilobases of genomic DNA with extensive regions of high GC content but that also contains sequencing gaps in the genomic DNA.

The prevalence of annotation errors became clear to us from three observations. First, the mouse LIN-41 we have aligned is actually composed of two predicted genes that are adjacent to each other on mouse chromosome 9. In the mouse, only one is correctly identified as Lin-41. To construct the putative mouse LIN-41 protein sequence for our alignment, we combined the two protein sequences (accession no. XM357972 and accession no. XM356199). Second, using the nucleic acid sequence coding for the RING finger domain from human lin-41, we were able to identify homologous sequence in the mouse genome very near to the Lin-41 locus on chromosome 9. Third, the predicted chicken LIN-41 proteins from NCBI (GenBank accession no. XM426005) and ENSEMBL (GeneScan ID 00000012620) diverge from each other and with our predicted chicken LIN-41 protein beginning in the Coiled-coil domain. It is notable that our chicken LIN-41 shares greater overall similarity to human, mouse, and rat LIN-41 than do either of the chicken proteins predicted by NCBI and ENSEMBL. Further refinement of vertebrate genomic DNA sequence will help clarify this problem.

Does clin-41 Have a Role in Limb Development?

The Progress Zone model describes patterning events along the PD axis of the vertebrate limb and proposes that specification and determination of the limb occurs in a proximal–distal manner over time (Summerbell et al., 1973). In this model, the longer a cell remains in the Progress Zone (the undifferentiated, subridge mesoderm) under the influence of Fgfs from the AER, the more distal fate it acquires. Therefore, more distal limb structures are specified at progressively later times in development. Recently, new data and reanalysis of original studies related to PD growth challenge the idea that patterning occurs in a proximal–distal manner over time (Dudley et al., 2002; Sun et al., 2002). This work has led to a new model that proposes each region of the limb (stylopod, zeugopod, autopod) is specified very early in development, and expansion of these regions over time results in determination and growth of recognizable skeletal elements.

Whereas the concept of developmental time is explicit in the Progress Zone model, a timing mechanism also underlies the Expansion model. Based on experimental evidence, differentiation of the limb occurs in proximal–distal manner over time, such that more proximal skeletal elements differentiate before more distal elements (Saunders, 1948; Summerbell, 1974; Rowe and Fallon, 1982). With this in mind, we can speculate on a role for cLin-41 in vertebrate limb development.

In the limb, clin-41 is expressed in three phases whose timing correlate with other well-defined developmental and molecular events. Perhaps most significant is the correlation with events related to development of the autopod. According to the Progress Zone model, autopod specification begins between stages 21 and 25 (wing: Saunders, 1948; Summerbell et al., 1973; leg: Rowe and Fallon, 1982), a time that correlates with phase two of clin-41 expression. Also at this time, the Hoxd genes initiate a SHH-dependent expression pattern that also correlates with morphogenesis of the autopod (Nelson et al., 1996; Zákány et al., 2004). Phase changes for both clin-41 and Hoxd genes at similar times is not surprising as both are dependent upon Shh signaling for correct spatial expression. This point is further supported by the observation that the expression patterns of clin-41 and the Hoxd genes are similarly altered in ozd and ta² mutants (compare Fig. 5 in this manuscript with Fig. 5B, F, and J in Ros et al., 2003, and Fig. 1J in Caruccio et al., 1999). For clin-41, the timing of the phase change and distal expression in the future autopod suggests an involvement in autopod morphogenesis.

In the absence of Shh signaling, clin-41 expression is initially normal. It is then down-regulated and becomes undetectable by whole-mount in situ analysis at stage 23/24. Furthermore, loss of Gli3, the primary transcriptional effector of the Shh pathway in the limb, results in significant up-regulation of mlin-41 in the anterior limb mesoderm. Therefore, during normal limb development, we propose that Shh signaling is necessary to maintain appropriate clin-41 expression. We also propose that clin-41 is maintained by Fgf signaling from the AER. The specific removal or reduction of FGF-8 signaling from the AER also results in the absence or reduction of Shh signaling. The interaction between SHH and FGF-8 is the main component of the feedback loop between the AER and ZPA (Laufer et al., 1994; Niswander et al., 1994; Lewandoski et al., 2000; Moon and Capecchi, 2000). In cases where all Fgf signaling is lost and Shh is not expressed, for example in the limbless limb bud, clin-41 expression is rapidly lost. However, when FGF-8 and FGF-4 are removed, Shh signaling is lost but other FGFs remain active in the AER (Sun et al., 2002; Boulet et al., 2004). In this case, low, residual lin-41 expression is maintained. Together, the data indicate that lin-41 is genetically downstream of both Shh and Fgf, and normal lin-41 expression is dependent upon both signaling pathways for normal expression.

Beginning at phase two, clin-41 is down-regulated in Shh-expressing ZPA cells at a time when the autopod is being determined in the leg (see Rowe and Fallon, 1982). Because clin-41 down-regulation occurs in Shh-expressing cells beginning at phase two, we hypothesize that clin-41 down-regulation occurs in cells that have expressed high levels of SHH for long periods of time. It is possible, therefore, that clin-41 is reflecting the temporal gradient of SHH exposure proposed by Harfe and coworkers (2004; see also Yang et al., 1997).

In the limb, Shh signaling influences several aspects of muscle development, including specification, proliferation, and delayed differentiation (reviewed in Christ and Brand-Saberi, 2002). Recent studies have indicated that, in precursor muscle cells, Shh signaling controls muscle growth in the limb by influencing the balance between cell proliferation and differentiation, thus, ensuring the formation of appropriately sized muscle masses (Amthor et al., 1998). Delayed differentiation of precursor muscle cells results in overgrowth of muscle cells, whereas precocious differentiation results in muscle loss, two situations induced by SHH overexpression or loss of Shh signaling, respectively (Amthor et al., 1998; Bren-Mattison and Olwin, 2002). In C. elegans, LIN-41 ensures the completion of L4 larval stage and initiation of the adult program by inhibiting the activity of LIN-29, thus delaying the activation of adult cell specific differentiation (Slack et al., 2000). Based on its role in C. elegans and because clin-41 is likely dependent upon Shh signaling in the limb, we propose that cLIN-41 coordinates the proper timing of muscle cell proliferation and/or differentiation in the developing limb. cLIN-41 would be a Shh signaling effector molecule in the developing muscle precursor cells.

In the developing limb, correct spatial expression of clin-41 is dependent on both Shh and Fgfs. We have identified the presence of miRNA binding sites in the clin-41 3′-UTR and report the presence of miRNAs in the developing chick limb. Using a reporter assay, work from Mansfield and coworkers (2004) showed that let-7c and e are asymmetrically active in the developing mouse limb. Interestingly, the activity domain of let-7c is complementary to the expression pattern we show for mlin-41in the mouse limb at an equivalent stage (compare Fig. 4C or 5D of this manuscript with Fig. 2B in Mansfield et al., 2004). We also used the miRNA prediction algorithm on mouse lin-41 3′-UTR and let-7c scored highly, indicating that it may regulate mlin-41. Similar to C. elegans lin-41, there are several LCSs in vertebrate lin-41 3′-UTRs. Given the potential presence of LCSs in all vertebrate lin-41 3′-UTRs and that let-7 is expressed in the developing limbs of mouse and chick when lin-41 is expressed, our data support the hypothesis that let-7 regulates lin-41 in the developing vertebrate limb. miRNA regulation of vertebrate Lin-28, another C. elegans heterochronic gene orthologue, has also been proposed indicating that specific miRNA-target gene interactions can be highly conserved (Moss and Tang, 2003). A detailed comparison of lin-41 transcript distribution, protein distribution and let-7 activity will be necessary to better characterize this proposed regulatory interaction.

The function of Lin-41 in limb development is unknown. Based on its function in C. elegans, it is reasonable to propose that Lin-41 regulates the timing of cellular proliferation and/or differentiation in the developing limb. Whether removal of LIN-41 function would impact the developmental program of a specific region of the limb, e.g., the autopod, or whether it would affect the entire limb is an intriguing question. In the limb, we hypothesize that clin-41 expression is dependent upon Shh and Fgf signaling and most likely miRNAs. The possible regulation of Lin-41 by miRNAs in the limb suggests the presence of a highly conserved developmental pathway (reviewed in Harfe, 2005; Pasquinelli et al., 2005); this adds another level of complexity to understanding how the output of the Shh and Fgf pathways are ultimately translated into the proper patterning and growth of a developing limb.

EXPERIMENTAL PROCEDURES

Cloning of Chicken and Mouse Lin-41 and Sequence Comparison

A chicken cDNA clone highly homologous to a portion of C. elegans Lin-41 was identified from confirmed, differentially expressed cDNA clones after performing a subtractive hybridization. Differentially expressed genes were obtained by subtracting stage 17–19 limbless mutant limb bud cDNA pools from age-matched wild-type limb bud cDNA pools. The resulting fragment was used to screen a stage 18–24 chick limb bud cDNA library (gift from C. Tabin). Several positive clones were obtained that contained portions of chicken clin-41, but none contained the most 5′ sequence. This sequence was deposited in GenBank under the accession no. DQ117917. Mouse primers used to amplify a mouse clone similar to the original isolated chicken lin-41 clone were as follows: Forward primer: 5′-TTA GAA GAT GAG GAT TCG ATT GTT GC-3′; Reverse primer: 5′-TGA AGG CGA AGT CTC TGT TCC TGC-3′. Protein sequences for C. elegans LIN-41A (AF195610), chicken (DQ117917), human (XM_067369), mouse (XM_357972 and XM_356199), rat (XM_236676), and zebrafish (XM_685160) LIN-41 were aligned using the ClustalW (http://www.ebi.ac.uk/clustalw) alignment program and figures generated with MacVector (accelrys).

Whole-Mount and Section In Situ Hybridization

A digoxigenin-labeled antisense riboprobe specific for chicken lin-41 was generated from a 1.4-kb fragment of clin-41 containing the most 3′ 1,000 bp of the coding region and ∼400 bp of 3′-UTR. A similar region was amplified from a mouse cDNA library and used to generate mouse specific lin-41 ribroprobe. Whole-mount in situ hybridization analysis was performed according to standard procedures (Nieto et al., 1996). Section in situ hybridizations were performed according to standard protocols with minor modifications (Moorman et al., 2001). Sections were cut at 15 μm.

Bead Implants

Heparin acrylic beads (Sigma) were size selected and soaked in a solution of SHH protein (5 mg/ml) for 1 hr at room temperature. Beads were implanted into the subridge mesoderm of the anterior leg bud; eggs were sealed with tape, reincubated, and collected at specific time points (Ros et al., 2000).

Images

Images were captured with a SPOTFIRE camera and software.

Prediction of miRNA Binding Sites in cLin-41 3′-UTR

To identify candidate miRNAs that target clin-41 3′-UTR, we used a miRNA target prediction method (Schageman and Pertsemlidis, unpublished observations). The method is based on modified global sequence alignment and was used to determine complementary base pair interactions between the clin-41 3′-UTR and all chicken miRNA sequences deposited in the microRNA Registry (Griffiths-Jones, 2004). These miRNA:target interactions were then scored and ranked using criteria primarily based on the work of (Doench and Sharp, 2004). First, a score is assigned based on the degree of complementary interactions between the target mRNA and the entire mature miRNA. Next, this score is adjusted to reward perfect complementary interactions between bases 2 and 8 at the 5′ end of the miRNA (allowing for one G:U base pair), a central bulge near base 10 of the miRNA and at least minimal complimentary interactions at the 3′ end of the miRNA. Next, the score is adjusted by the addition of the absolute value of the free energy of hybridization as estimated by mfold (Zuker, 1989). Finally, chicken, mouse, and predicted human and rat lin-41 3′-UTR sequences are aligned to assess conservation of the predicted target. Regions of the 3′-UTR that are highly conserved and are predicted to be targets of homologous miRNAs are inferred to be functionally significant and are given added weight for follow-up Northern analysis.

Northern Analysis for miRNA in Chick Limb

Chicken embryos were staged according to Hamburger and Hamilton (1951, 1992). Similarly staged leg buds were either collected whole and pooled or collected and pooled after dissection along the proximal–distal axis. Total RNA was extracted from homogenized tissue using Tri-Reagent and quantified. Approximately 15 μg of total RNA was loaded per lane on an 11% polyacrylamide gel. Northern analysis was performed as described in Pasquinelli and coworkers (Pasquinelli et al., 2003).

Acknowledgements

We thank Frank Slack and Betsy Schulman for sharing their data and manuscript before publication and Dr. Xin Sun for providing mouse mutant embryos for in situ analysis. We thank members of the Fallon Lab for helpful comments on this manuscript.