Xenopus fibrillin is expressed in the organizer and is the earliest component of matrix at the developing notochord-somite boundary



We identify a Xenopus fibrillin homolog (XF), and show that its earliest developmental expression is in presumptive dorsal mesoderm at gastrulation, and that XF expression is regulated by mesoderm-inducing factors in animal cap assays. XF protein is also first detected in presumptive mesoderm, but is concentrated specifically into extracellular-matrix structures that begin to develop de novo by mid-gastrulation at both of the bilateral presumptive notochord-somite boundaries. Later in embryogenesis, XF protein is localized to the extracellular matrix at tissue boundaries, where it is found surrounding the notochord, the somites, and the neural tube, as well as under the epidermis. This pattern of protein deposition combines to give the appearance of an “embryonic skeleton,” suggesting that one role for XF is to serve as a mechanical element in the embryo prior to bone deposition. Developmental Dynamics 235:1974–1983, 2006. © 2006 Wiley-Liss, Inc.


Fibrillins are a family of large extracellular matrix (ECM) proteins that form the backbone for microfibrils, 10–12-nm-diameter filamentous structures that are widely distributed in ECM either as discrete entities or as a component of elastic fibers (Robinson and Godfrey, 2000). Microfibrils are found in every organ as well as in early embryos, and can perform both structural and signaling roles in tissue homeostasis and development (Ramirez et al., 2004). Humans have three fibrillin genes (Fbns 1–3) that exhibit 61–69% identity at the amino acid level (Corson et al., 2004). This identity is scattered evenly over these proteins, which also exhibit a conserved pattern of distribution of three distinct protein motifs including epidermal growth factor-like (EGF-like) repeats, 8-cysteine based (8-cys) repeats, and a hybrid domain of these two motifs.

Developmental expression of fibrillins has been described in post-gastrulation mouse, bird, and human embryogenesis (Zhang et al., 1994, 1995; Quondamatteo et al., 2002; Visconti et al., 2003; Corson et al., 2004). These experiments reveal distinct patterns of embryonic expression for each of these fibrillin isoforms, but multiple fibrillins are generally expressed in a given tissue including developing heart and vasculature, notochord/axial structures, and developing lung, kidney, and limb bud. The embryonic pattern of deposition of the fibrillin proteins has been determined at the highest resolution in chicken, where Fbn 1 is first found at gastrulation both as fibrils radiating from Hensen's node, the avian equivalent of the frog organizer, as well as along the primitive groove or primary axis prior to notochord formation (Gallagher et al., 1993). Subsequently, it is found around the notochord, dorsal aorta, and in the developing heart, gut, and lung (Gallagher et al., 1993; Burke et al., 2000). Chicken Fbn 2 is also first found in axial structures during somitogenesis, surrounding the notochord and surrounding each somite as it segments from the prechordal plate (Wunsch et al., 1994; Rongish et al., 1998; Czirok et al., 2004). Fbn 3 is also present in embryos, but its distribution is less well characterized (Corson et al., 2004). In the chicken dorsal axis, a mechanical role for the fibrillins in maintaining tissue integrity has been postulated, both prior to notochord development and at later stages as a fibrillar, basket-like network holding axial embryonic elements such as notochord, somites, dorsal aorta, and gut in relative position (Gallagher et al., 1993; Wunsch et al., 1994; Rongish et al., 1998; Czirok et al., 2004). The dynamic behavior of Fbn 2 fibrils during axis formation in the chicken embryo reveals strong cell-matrix interactions, because fibril displacement in the axis is similar to the pattern of cell movement described in the gastrula (Czirok et al., 2004). These fibrils are also observed to undergo a meshwork-to-cable transition, resolving into axially oriented cables. This observation is consistent with fibrillin involvement in the process of transforming global tensional force into axial extension of the chicken embryo. However, to date an experimental challenge to the hypothesis that fibrillin function is biomechanically important in the embryo has been lacking, and potential signaling roles for the multifunctional fibrillins in early development have not been examined.

Embryo shape change driven by physical force generation is an integral feature of early vertebrate development at gastrulation, and the Xenopus laevis embryo develops from a spherically shaped cleaving blastula into an elongated, swimming tadpole in the space of less than a day (Nieuwkoop and Faber, 1967; Oster et al., 1983; Trinkaus, 1984). The bulk of this morphogenesis occurs at gastrulation and is largely driven by convergence and extension of the involuting axial and paraxial mesoderm that gives rise to notochord and somites. This convergence and extension is driven by a patterned cell motility in this tissue termed mediolateral intercalation behavior (MIB) (Keller et al., 2000). This tissue exhibits an anisotropic increase in stiffness and generates directed force to drive normal, asymmetric blastopore closure and archenteron elongation at gastrulation (Moore et al., 1995). Concomitant with both expression of anisotropic biomechanical properties and induction of MIB in involuting, presumptive axial and paraxial mesoderm is the development de novo of a boundary between presumptive notochordal and somitic tissue (Shih and Keller, 1992a, b). This boundary regulates both cell motility, by inducing presumptive notochordal cells that contact the boundary to transit from a bipolar to a monopolar protrusive state, and the expression of cell fate in the presumptive notochord (Domingo and Keller, 1995). However, the molecular basis underlying regulation of this patterned cell motility, the change in stiffness and force production exhibited by this tissue, and the role of the ECM at the developing notochord-somite boundary, remains unclear (Keller et al., 2000).

We identify Xenopus fibrillin (XF) as a member of the fibrillin extracellular matrix protein family. We report its expression pattern in the early embryo, and the localization of XF protein to the developing notochord-somite boundary during notochordal morphogenesis. This localization makes XF a candidate to mediate functions that have been ascribed to this boundary. As the embryo develops, XF protein forms an embryonic skeleton by surrounding and separating the major tissues including the notochord, somites, and neural tube.


Identification of cDNA's Encoding Xenopus Fibrillin (XF)

cDNAs encoding Xenopus Fibrillin (XF) were isolated from an early neurula library in a low-stringency screen utilizing EGF-like repeats (Coffman et al., 1993). The largest XF cDNA represents a partial clone of 5.3 kilobase pairs that encodes an open reading frame of 1,172 amino acids. This open reading frame lacked a start methionine, and is followed by a 3′ untranslated region and poly-A tail. The protein derived from this open reading frame exhibited 68, 77, and 71% identity to the C-terminal portions of the human proteins fibrillin 1, 2, and 3, respectively, suggesting that XF is a homolog of fibrillin 2 (Maslen et al., 1991; Zhang et al., 1994; Corson et al., 2004) (Fig. 1). The C-terminal region of these proteins consists almost entirely of two distinct repeated protein motifs arranged in a stereotypical pattern, and this pattern is conserved between these proteins and XF. One repeated protein motif is the calcium-binding EGF-like repeats (cbEGF), a subset of the EGF repeats that feature six cysteine residues exhibiting stereotypical spacing, and XF exhibits nineteen copies of this motif. These repeats are found widely in extracellular proteins and may mediate protein–protein interactions, or provide a stiff backbone to these extended, linear molecules (Sakai et al., 1991). A second protein motif that is found three times in XF has as its signature eight conserved cysteines including three in a row (termed the 8-cys repeat), and has been identified in both the fibrillin family and in the latent transforming growth factor beta binding protein (LTBP) family of ECM proteins. Some of these domains found in LTBP family members can interact with, and modulate signaling from, TGF-beta family members (Saharinen et al., 1996; Saharinen and Keski-Oja, 2000). XF and the fibrillins also share homology in their carboxy-terminal sequences, which do not correspond to any recognizable modular protein motif. However, this region includes both a conserved furin proprotein convertase processing site at which profibrillins are matured by proteolytic processing, as well as a region that exhibits cell binding activity (Lonnqvist et al., 1998; Ritty et al., 2003). XF also exhibited eight N-linked glycosylation sites of which six are conserved in at least two of the three human fibrillins. This clone encodes the C-terminal portion of a Xenopus member of the fibrillin family, which we term Xenopus fibrillin (XF), because the amino acid identity between the human and Xenopus fibrillins is scattered evenly throughout the multiple domains of these proteins, including all 139 cysteine residues in the appropriate locations (Fig. 1).

Figure 1.

Sequence comparison for XF and the Human Fibrillins. The available deduced amino acid sequence of XF is aligned with Human Fibrillin 1, 2, and 3 sequences. Only human residues that vary from the Xenopus sequence are shown. Introduced gaps to align sequences are represented as an asterisk (*). Introduced gaps across all three proteins, to facilitate display of repeated motifs, are represented as a period (.). Cysteine residues are shaded across all four proteins, revealing the conserved spacing of cysteine residues among these proteins. Nineteen calcium-binding EGF-like repeats are shown, one or two per line, and the consensus repeat (>11/19 identical) is D(I/V)(D/N)EC(X7)C(X4)CXC(X2)N(X2)GS(Y/F)XCXCPXG(Y/F)(X8)C. Three 8-cysteine repeats are shown one per line, and the consensus repeat (3/3 except the first C) is C(X8)C(X10)TK(X2) CCC(X4)G(X2)W(X3)CEXCP(X10)C(X2)G. The nucleotide sequence from which the XF protein sequence is derived has been deposited in GenBank under accession number DQ310728.

Xenopus Fibrillin Is Expressed Specifically in Dorsal Involuting Mesoderm at Gastrulation

RNAse protection assays of staged embryo RNA indicated that Xenopus fibrillin message expression can be detected in the late blastulae (St. 9), and is markedly up-regulated at mid-gastrulae stage (St 10.5) (Fig. 2A). This level of expression of XF is maintained through swimming tadpole stages. Northern blots of staged embryo RNA revealed that there is no maternal component contributing to expression, because XF message is not detected before the onset of zygotic transcription (Fig. 2B). However, XF message is detected at both neurula and tailbud stages, confirming the developmental expression of XF revealed by RNase protection assay. The 11 kilobase (kb) size of this message is consistent with the identification of XF as a fibrillin homolog; the three human fibrillins exhibit message sizes ranging from 9.5 to 11 kb.

Figure 2.

Developmental regulation of XF message. A: RNase Protection assay for XF and elongation factor 1-alpha message levels. One to five staged embryos per lane (note uneven loading, reflected by ef1-alpha signal, required to see early XF expression). Quantitation of XF/control signal is shown. B: Northern blot of staged Xenopus RNA. An 11-kilobase XF message is found at neurula and tailbud stages, but is absent in the egg, indicating that there is no maternal component to the XF message supply in the embryo.

XF message is first detected in the dorsal marginal zone at mid-gastrulation (St 10.5) (Fig. 3A). By late mid-gastrula (St. 11.5), a dorsal close-up of the posterior of an XF stained embryo revealed two posteriorly-directed tongues of expression flanking the midline at the dorsal lip (Fig. 3B). Sagittal sections of similar stage stained whole mounts showed that this XF expression is restricted to recently involuted mesoderm (Fig. 3C). At mid neurula stage (St. 16), a dorsal view revealed intense notochordal expression and weaker expression in somitic mesoderm, as well as anterior expression focused on the eye buds (Fig. 3D). An axial cross-section reveals notochord exhibiting the highest levels of expression, with more lateral somitic mesoderm exhibiting progressively graded XF message from medial to lateral (Fig. 3E). At this stage, message is seen for the first time in the ectodermal lineage with expression in the notoplate, or presumptive floorplate of the neural tube. Additionally, because XF is found in the notochord at this stage in distinct contrast with in situs for another EGF-repeat containing protein at a similar stage, Xenopus notch (fig. 6 in Coffman et al., 1993), this provides good evidence for the specificity of this probe for XF. At stage 25, trunk notochord staining declines relative to expression in the tail to reveal three bands of XF expression running the length of the embryo in lateral view, corresponding to the notochord, floorplate, and endodermal roof of the archenteron (hypochord) (Fig. 3F), and expression in the notochord, but not floorplate and roofplate, gradually disappears from anterior to posterior by stage 28 (see Fig. 5C). At St. 40, all axial expression has faded and expression is limited to the mesenchyme of the head as well as the tail fins (Fig. 3G). The expression of XF in cultured explants of DIMZ (Keller sandwiches) recapitulates the mesodermal expression pattern in whole embryos with XF message found in extending mesoderm, both in the notochord and lateral mesoderm, as well as expression in the anterior neural regions (Fig. 3H). However, in contrast to intact embryos, no XF expression is seen along the extended neural axis of these explants, which lack a floorplate under these culture conditions (Poznanski and Keller, 1997).

Figure 3.

Localization of XF message in the embryo. XF message is detected by whole mount in situ hybridization. Dorsal is to the right in A, and is up in C, E, F, and G. Anterior is to the left in B, C, D, F, and G. A: Vegetal view of a St-10.5 embryo. Black squares are to the left of the blastopore lip. B: Dorsal view of the posterior half of St-11.5 whole mount. Black squares identify the blastopore lip. C: Sagittal section through a St-11.5 whole mount. D: Dorsal view of a St-16 embryo. E: Cross-section through the trunk of a St-16 embryo. F: Lateral view of St-25 embryo. G: Lateral view of St-40 embryo. H: Keller sandwich (St 20) with neural axis running anterior to posterior from top, and mesodermal axis running anterior to posterior from bottom, meeting at middle bend. YP, Yolk plug; N, notochord; SM- Somitic mesoderm; NP, notoplate; RE, roof of endoderm; NT, neural tube; IM, involuting mesoderm; E, endoderm; EB, eye bud; NE, neural extension. Scale bars = 0.25 mm (B–E), 0.5 mm (A,H), 1 mm (F,G).

Figure 5.

Regulation of XF message expression. A: RNase Protection assay for XF and elongation factor 1-alpha message levels. Animal caps cultured with or without the addition of indicated growth factors revealing regulation of XF expression by FGF and Activin. AC, uninduced animal cap; FGF, caps cultured in 50 ng/ml basic FGF; A50 and A500, caps cultured in 50 or 500 pM Activin A; WE, whole embryo. The first lane shows the input probes, for XF on top and the loading control elongation factor 1-alpha (ef-1 alpha) on the bottom. The middle unidentified band is a third input probe for experiments not presented here. Quantitation of XF/ef-1 alpha signal is shown relative to uninduced animal cap. B: Exogastrulated embryos assayed for XF expression by in situ hybridization. C: Posterior of a normal St-31/32 embryo assayed for XF expression by in situ. D: Posterior of a cell division blocked St-31/32 embryo stained for XF expression exhibits little XF expression in the tailbud. E: Lithium-treated dorsalized embryo exhibits strong XF expression. F: UV-treated ventralized embryo exhibits little XF expression. The embryos in E and F were processed for XF in situ together. Scale bars = 0.25 mm (B,E,F) and 0.5 mm (C,D). N, notochord; NE, neuroectodermal ball.

Xenopus Fibrillin Protein Is Enriched at the Notochord–Somite Boundary

Fibrillin protein was first detected by wholemount immunostaining localized to two short parallel stripes in deep dorsal tissue at late midgastrulation (St.11) (Fig. 4A). Staining of the anterior surface of the blastocoel is not specific, because it appears in control reactions processed without primary antibody (data not shown). By late gastrula (St. 12), these stripes have doubled in length as the blastopore closes, consistent with the development of the notochord–somite boundary as more of the presumptive mesoderm has involuted (Fig. 4B). These stripes clearly reveal the anterior-to-posterior progression of convergence of presumptive notochord by virtue of the posterior flaring of these presumptive notochordal-somitic boundaries near the blastopore. In a closer view of these boundaries at early neural plate stages (St. 13), XF protein is detectable in a non-fibrillar, punctate form throughout the dorsal mesoderm that expresses message, but most of the immunoreactivity is concentrated at the notochord-somite boundary (Fig. 4C). After neural tube closure (St. 22), XF is detected in an anterior-to-posterior gradient, outlining the notochord and in the intersomitic clefts (Fig. 4D). An axial cross-section from such an embryo reveals XF protein in most extracellular spaces, including surrounding the notochord, the neural tube, outlining the somites, and under the skin (Fig. 4E). An adjacent section of the same embryo reveals the arrayed intersomitic staining orthogonal to the axis, with fibrils roughly organized to run in the mediolateral direction (Fig. 4F). By stage 39, a dorsal view reveals XF protein is found as a fully formed “axial skeleton” around the notochord and in the tail fins at the midline, presaging the appearance of vertebrae and ribs beginning around St. 50 (Nieuwkoop and Faber, 1967) (Fig. 4G). XF protein is also strongly localized to the developing eye, as well as surrounding the brain. Western blotting of stage 16 embryo extracts reveals that the JB3 antibody identifies both a 350- and 200-kD band (Fig. 4H). The single band running at 350 kD is consistent with the identification by the JB3 antibody of a single fibrillin isoform.

Figure 4.

Localization of XF protein in the embryo. XF protein is detected by immunostaining. Anterior is to the left in A–D and G. Dorsal is up in D–F, and G is a dorsal view from above. A: Dorsal view of St-11 whole mount. White squares identify the location of the blastopore lip. B: Dorsal view of St-12 whole mount. C: Dorsal, mid axial-posterior close-up of St-13 whole mount. D: Lateral view of a St-22 whole mount. E: Cross-section through mid-axis of St-22 embryo. Scale bar is for E and F. F: Adjacent cross-section through mid-axis of St-22 embryo, revealing periodic, orthogonoal intersomitic staining. G: Dorsal view of a St-39 embryo. H: Western blot of St-16 embryo extract probed with the JB3 antibody revealing a single band in the size range of fibrillin isoforms (350 kD), as well as a second, smaller band (200 kD). YP, yolk plug; N, notochord; SM, somitic mesoderm; NT, neural tube; IS, intersomitic space; E, eye; B, brain. Scale bars = 0.125 mm (C, E/F), 0.25 mm (A,B), and 0.5 mm (D,G).

Control of Expression of Xenopus Fibrillin

The onset of XF expression correlates spatially and temporally with dorsal mesodermal differentiation in Xenopus. We examined the regulation of XF expression by experimental induction of both dorsal mesoderm and convergent extension following addition of growth factors that induce mesoderm to cultures of animal caps (Fig. 5A). Basic FGF at a sufficient concentration to induce ventral mesoderm has a small positive effect on baseline expression of XF, consistent with the low-level expression of XF seen in lateral/ventral mesoderm in whole embryos (Howard and Smith, 1993). In contrast, Activin A, an inducer of both dorsal mesodermal fate and convergent extension in animal caps (Green et al., 1992), strongly potentiates XF message expression in a dose-dependent manner (Fig. 5A). This up-regulation is consistent with the normal high expression of XF seen in presumptive dorsal mesodermal tissue. Embryos hyperdorsalized with lithium chloride exhibit a marked upregulation of XF message in this mesodermal ball (Fig. 5E) (Kao and Elinson, 1988; Klein and Melton, 1996). In contrast, UV-mediated ventralization of embryos leads to strongly reduced XF expression levels (Fig. 5F), when assayed by in situ hybridization in the same reaction and developed in the same dish as the dorsalized embryo shown in Figure 5E (Kao and Elinson, 1988). XF expression also correlates with later stages of embryo extension, because experimentally blocking cell division from gastrulation on by treatment of the embryo with the DNA synthesis inhibitors hydroxyurea and aphidicolin does not visibly perturb development of the embryo until tailbud stages when the tailbud fails to elongate, allowing gastrulation and neurulation to proceed normally (Harris and Hartenstein, 1991). This treatment also does not change the pattern of XF expression in the embryo until the tailbud stage, indicating that cell division is not required for the correct localization of expression of XF at gastrula or neurula stages (data not shown). However, at tailbud stage XF expression is reduced in the developing tail of treated embryos (Fig. 5D), relative to untreated controls (Fig. 5C).

Keller explants of dorsal marginal zone exhibited a similar pattern of XF expression in mesoderm as is found in intact embryos, but they fail to express XF along the neuroaxis (Fig. 3H). Culturing embryos in high salt results in exogastrulation, in which the mesoderm fails to internalize but instead extends away from presumptive neural/epidermal tissue, and separates the neuroectodermal ball from mesodermal signals during development (Kintner and Melton, 1987). Exogastrulated embryos expressed XF in extending mesodermal tissue, but do not express XF in the neuroectodermal ball (Fig. 5B).


We show that Xenopus fibrillin (XF) is a member of the fibrillin family of extracellular matrix proteins, with highest homology to fibrillin 2. Expression of XF initiates in the involuting presumptive mesoderm at gastrulation, and becomes graded with high expression in midline notochordal tissue with progressively less expression in more lateral somitic mesoderm, consistent with the observation that both mesodermal inducers Activin A and FGF potentiate XF transcription in animal caps. Both the strong correlation between the XF in situs and JB3 antibody staining, as well as the fibrillar nature and extracellular localization of this epitope, suggest that the smaller protein identified by JB3 in Western blots is only detected under blotting conditions, although we cannot formally rule out a contribution of this antigen to the staining pattern that we see. The XF protein exhibits a more refined localization within the dorsal mesoderm; it is highly enriched specifically in the extracellular matrix deposited at the notochordal-somitic boundaries during gastrulation. XF expression broadens through development, and XF protein is widely distributed in extracellular spaces. By tailbud stage, XF protein is found both under the epidermis as well as surrounding tissues such as notochord, somites, and neural tube; this is similar to the wide distribution of fibrillins in the extracellular spaces surrounding organs in human, mouse, chick, and quail where it subserves numerous functions in development and homeostasis (Robinson and Godfrey, 2000).

The Notochord–Somite Boundary

In Xenopus, the ECM found at the developing notochordal-somitic boundary develops de novo in dorsal mesoderm concomitant with notochordal morphogenesis, and we show that XF is detectable in these boundaries at the time that they first appear morphologically (Shih and Keller, 1992b). Every vertebrate embryo that has been examined exhibits peri-notochordal fibrillin immunoreactivity (Gallagher et al., 1993; Rongish et al., 1998; Quondamatteo et al., 2002; Visconti et al., 2003). High-resolution localization studies of Fbn 2 during avian development using the same antibody (JB3) as employed in this study reveal a difference in the timing of appearance of Fib 2 in the notochord–somite boundary between avian and amphibian embryos. We detect fibrillin immunoreactivity concomitant with boundary formation, whereas in chicken an ECM structure is morphologically apparent at the notochordal-somitic boundary prior to detection of Fib 2. This may reflect differing morpho-mechanical requirements of these two vertebrate embryos at gastrulation, with the holoblastic cleavage strategy employed by Xenopus requiring fibrillin function earlier than the meroblastic strategy used by avian embryos. In Xenopus, the bulk of fibrillin immunoreactivity at gastrulation is detected at the notochord–somite boundary, where it assembles into microfibrils. However, a low level of punctate XF protein staining also appears across the body of the notochord or in somitic mesoderm away from the boundary, in regions that express message for XF. This pattern is consistent with local, regulated assembly of microfibrils, but the mechanism by which XF protein is concentrated in the boundary is not understood and no role for XF in the para-boundary region has been identified.

As the presumptive notochordal-somitic boundary resolves into the matrix-rich mature notochordal sheath, a wide assortment of ECM molecules become localized to or expressed around this maturing structure, including laminin B1, fibronectin, type II collagen, and LTBP (Fey and Hausen, 1990; Su et al., 1991; Altmann et al., 2002; Davidson et al., 2004). In the zebrafish, laminins are required for maintenance of differentiated notochord, but not for its initial morphogenesis (Parsons et al., 2002), and laminins are assembled into the fish notochordal sheath after gastrulation, similar to the pattern of deposition of laminin B1 in Xenopus (Fey and Hausen, 1990). Fibronectin function in Xenopus embryos has been examined by blocking FN-integrin interactions; these embryos close their blastopores during gastrulation although they exhibit reduced convergence and extension, likely due to an observed inhibition of cadherin function that is required for gastrulation (Lee and Gumbiner, 1995; Marsden and DeSimone, 2003). Because we identify XF as the first molecule specifically assembled into this developing matrix structure in Xenopus, XF is a candidate to be involved in assembling or regulating these other matrix elements in the notochordal sheath.

The notochordal-somitic boundary regulates dorsal mesodermal morphogenesis (Shih and Keller, 1992b; Domingo and Keller, 1995), and the regulation of XF message is also consistent with a role in convergent extension of dorsal mesoderm. XF message is markedly up-regulated in animal cap experiments by Activin A, which induces both extension behavior and dorsal mesodermal fates, but is less well up-regulated by basic FGF, a ventral mesodermal inducer (Green et al., 1992; Howard and Smith, 1993). Lithium chloride dorsalized embryos also upregulate XF, while UV ventralized embryos that show no convergent extension exhibit little XF expression, supporting the correlation of dorsal mesodermal convergence and extension behavior with XF expression. This correlation extends into the extending tailbud, which expresses high levels of XF. Perturbing tailbud morphogenesis by blocking cell division also reduces XF expression in the developing tail, suggesting that XF may function in normal axial extension at these stages.

Roles for Fibrillin in Embryos

The fibrillins are multifunctional molecules, allowing them to exert influence in tissue development and homeostasis at several levels simultaneously. One role is to multimerize into microfibrils, which can perform biomechanical roles in the body (Robinson and Godfrey, 2000). Fibrillins 1 and 2 have been shown to be ligands for both integrin and proteoglycan cell surface receptors (Pfaff et al., 1996; Sakamoto et al., 1996; Ritty et al., 2003; Bax et al., 2003; Cain et al., 2005). These interactions likely mechanically link cells to fibrillin containing matrix (Czirok et al., 2004), and signaling functions through these receptors are possible based on analogy with other matrix molecules such as laminins (Hopker et al., 1999). Fibrillins also interact with growth factor signaling in tissue homeostasis; for example, fibrillins directly interact with bone morphogenic protein 7 and can also perturb other TGF-beta signals (Arteaga-Solis et al., 2001; Neptune et al., 2003; Gregory et al., 2005). Microfibrils also serve as scaffolds for construction of ECM aggregates; for example they are an obligate component of elastic fibers. In addition to elastin, other binding partners for microfibrils include MAGP (microfibril associated glycoproteins) -1 and -2, MP 70/78, MFAP (microfibril-associated proteins) -1, -3, and -4, emilin, lysyl oxidase, beta hgIG, and LTBP (reviewed in Robinson and Godfrey, 2000). However, the role of microfibrils in early vertebrate development is largely unexplored, and which activities of fibrillin are relevant to early Xenopus development remains to be determined.

Homologous recombination experiments in mouse shed light on the role of fibrillins in organismal development and homeostasis. However, interpreting these experiments is complicated by functional redundancy exhibited by these molecules, which has its basis in the ability of fibrillins to form either homotypic or heterotypic microfibrils (Lin et al., 2002; Charbonneau et al., 2003). For example, gene targeting of either Fbn 1 or 2 in mouse (and mice lack Fbn 3; Corson et al., 2004) is compatible with normal development and both microfibrils and elastic fibers are assembled in either case, consistent with the observation that both fibrillins are co-expressed through much of development (Pereira et al., 1997; Arteaga-Solis et al., 2001). However, in regions where co-expression is not observed, phenotypes are uncovered in these knockout mice (Arteaga-Solis et al., 2001). Mice heterozygous for mutant alleles of fbn 1 exhibit a range of developmental onset of phenotype, consistent with specific dominant negative alleles perturbing normal fibrillin function to varying degrees and thereby determining the timing of onset of a mutant phenotype and revealing earlier requirements for fibrillin function than revealed by the null mice (Dietz and Mecham, 2000). In combination, this makes it likely that fibrillin function is required for early development in mouse and thus also likely for early frog development, but that either fibrillin-1 or -2 can subserve this function.

Human fibrillopathies also support the hypothesis that fibrillin function is required during development. Microfibril-based diseases include Marfan syndrome (MFS) (associated with Fbn 1) and congenital contractural arachnodactyly (CCA) (associated with Fbn 2) (Dietz et al., 1994; Ramirez, 1996). MFS and CCA patients present overlapping, pleotropic phenotypes with autosomal dominant inheritance (Robinson and Godfrey, 2000). Phenotypes associated with these diseases include defects in multiple organ systems, and can include cadiovascular, ocular, and skeletal involvement. Because life expectancy can range from neonatal to near-normal, normal fibrillin function is likely required throughout human development, making it important to develop vertebrate model systems to assay fibrillin function during development.

Possible Functions for XF in Convergence and Extension

The localization of XF protein to the developing notochord–somite boundaries suggests that XF functions in the process of axial convergence and extension driven by cell intercalation, because these boundaries have been implicated in regulating convergence and extension at several levels (Keller et al., 2000). These matrix-containing structures appear and elongate as straight lines of separation of presumptive notochordal and somitic cells at stage 11–11.5, the midgastrula (Shih and Keller, 1992b), and the pattern of boundary formation is reflected in the pattern of induction of cell motile behaviors that drive intercalation and thus dorsal, axial morphogenesis. Cell intercalation begins with expression of mediolateral intercalation behavior (MIB) in the anterior presumptive somitic and notochordal mesoderm with the formation of the vegetal alignment zone (VAZ). The notochordal-somitic boundaries form within the VAZ and from this origin proceed posteriorly, and the characteristic induction of cells expressing MIB also proceeds posteriorly within a few cell diameters of the boundary, suggesting a role for the boundary in organizing this behavior (Shih and Keller, 1992b). In addition, the bipolar cells expressing MIB exhibit a second boundary-dependent behavior, because they are transformed into monopolar cells on contact with the boundary in a “boundary capture” mechanism that is thought to aid in producing convergence and extension when one protrusive end interacts with the boundary, locally inhibiting motility and causing the cell to spread in the plane of the boundary (Shih and Keller, 1992a). Thus XF is in position to have a role in several processes, including a role in assembling the matrix that constitutes the boundary, in inducing bipolar cell motile behavior near the boundary, or in transiting of cells to a monopolar cell motile behavior upon boundary contact. In addition, XF fibrils in the boundaries are perfectly positioned in the embryo to be responsible for the experimentally observed anisotropic changes in tissue stiffness in the dorsal axis during gastrulation (Moore et al., 1995). These fibrils form at the same time that tissue stiffness is observed to change, and they are oriented axially, which is the same direction in which this tissue develops increased stiffness. These observations make XF a strong candidate to mediate biomechanical changes exhibited by dorsal mesoderm during axial development, as well as to regulate local cell motile behavior, both directly and through the activity of binding partners.

In Xenopus, the first developmental event XF can function in is gastrulation because that is when it is first expressed in the embryo. However, XF may function iteratively in subsequent developmental events as well as in tissue homeostasis, possibly expressing a different range of functionalities in each context. For example, by early tailbud stages the notochord is surrounded by XF and the other matrix molecules that comprise the rounded notochordal sheath. In this context, cell behavior becomes uniform around the periphery of the notochord as cells adopt the “pizza slice” morphology characteristic of differentiated notochord, distinct from MIB (Keller et al., 1989). XF becomes widely expressed at this stage and is also found surrounding the neural tube, somites, and other tissue boundaries, consistent with the idea of it having a role in boundary-dependent signaling events or as a mechanical element in the early embryo.


Characterization of cDNA and Developmental Expression

cDNAs encoding Xenopus Fibrillin were isolated from a neurula library in a low-stringency screen utilizing EGF repeats as probe (Coffman et al., 1993). pXND4 coding for XF was sequenced on both strands with the Sequenase kit (United States Biochemical Corp., Cleveland, OH) utilizing both deletion strategies and oligonucleotides. Northern blotting used P-32 labelled pXND4 and poly-A selected staged RNA blotted onto nitrocellulose membrane with a final wash at 0.2× SSPE and 0.5% SDS at 65°. RNase protections were performed as described (Dixon and Kintner, 1989), using SP6/Dde1 for both pXND4 and p14p (ef1-a), and quantitated using the analyze gels function in ImageJ (http://rsb.info.nih.gov/ij/). Whole mount in situ hybridization followed standard methods (Harland, 1991), using SP6/EcoR1 for pXND4.

Embryo Manipulations and Immunohistochemistry

Both pigmented and albino Xenopus embryos were generated by induction of egg laying and in vitro fertilization (Kintner and Melton, 1987) and staged (Nieuwkoop and Faber, 1967) by conventional techniques. Embryos were dorsalized with lithium and ventralized with UV as described (Kao and Elinson, 1988), and both dorsalized and ventralized embryos were processed for XF in situ in the same reaction vessel to allow comparison of relative expression levels. Animal caps were cut and cultured with growth factors as described (Kintner and Dodd, 1991), and human recombinant activin A was provided by Wylie Vale while basic FGF was provided by David Schubert. Sandwich explants and open-face explants were made and cultured as described (Keller, 1991). Exogastula were produced as detailed (Kintner and Melton, 1987), and inhibition of cell division was as described (Harris and Hartenstein, 1991).

Fluorescence immunocytochemistry was by standard techniques while whole mount immunohistochemistry followed (Hemmati Brivanlou et al., 1991), using an alkaline phosphatase conjugated secondary antibody and amplifying using anti-alkaline phosphatase alkaline phosphatase if required (Boehringer Mannheim, Indianapolis, IN). Fibrillin was detected with the monoclonal antibody [JB3] (Wunsch et al., 1994), which recognizes chick fibrillin 2 (Rongish et al., 1998) and was provided by Charlie Little. Western blotting was performed on whole embryo extracts run on 5% acryamide gels, electroblotted onto nitrocellulose and developed with the JB3 antibody and HRP conjugated secondary antibody and supersignal detection reagents (Pierce, Rockford, IL).