Application of lentivirus-mediated RNAi in studying gene function in mammalian tooth development

Abstract

RNA interference (RNAi) has recently become a powerful tool to silence gene expression in mammalian cells, but its application in assessing gene function in mammalian developing organs remains highly limited. Here we describe several unique developmental properties of the mouse molar germ. Embryonic molar mesenchyme, but not the incisor mesenchyme, once dissociated into single cell suspension and re-aggregated, retains its odontogenic potential, the capability of a tissue to instruct an adjacent tissue to initiate tooth formation. Dissociated molar mesenchymal cells, even after being plated in cell culture, retain odontogenic competence, the capability of a tissue to respond to odontogenic signals and to support tooth formation. Most interestingly, while dissociated epithelial and mesenchymal cells of molar tooth germ are mixed and re-aggregated, the epithelial cells are able to sort out from the mesenchymal cells and organize into a well-defined dental epithelial structure, leading to the formation of a well-differentiated tooth organ after sub-renal culture. These unique molar developmental properties allow us to develop a strategy using a lentivirus-mediated RNAi approach to silence gene expression in dental mesenchymal cells and assess gene function in tooth development. We show that knockdown of Msx1 or Dlx2 expression in the dental mesenchyme faithfully recapitulates the tooth phenotype of their targeted mutant mice. Silencing of Barx1 expression in the dental mesenchyme causes an arrest of tooth development at the bud stage, demonstrating a crucial role for Barx1 in tooth formation. Our studies have established a reliable and rapid assay that would permit large-scale analysis of gene function in mammalian tooth development. Developmental Dynamics 235:1334–1344, 2006. © 2006 Wiley-Liss, Inc.

INTRODUCTION

In mice, the dental mesenchyme derives from cranial neural crest cells migrating from the midbrain and hindbrain regions around embryonic day 8.5 (E8.5) (Imai et al., 1996; Chai et al., 2000). The first branchial arch acquires the tooth-forming capability between E8.75 to E9.0 (Lumsden, 1988; Y. Zhang et al., 2003). Subsequently at E10.5, the determination of the tooth-forming site and tooth type occurs (Neubüser et al., 1997; Tucker et al., 1998; Peters and Balling, 1999). At E11.5, the first morphological sign of molar tooth development appears as a local thickening of the dental epithelium. At the E12.5 and E13.5 bud stage, the thickened dental epithelium proliferates and invaginates into the subjacent mesenchyme to form the epithelial bud. Meanwhile, the mesenchymal cells become condensed around the epithelial bud. At the E14.5 cap stage, the epithelial bud undergoes specific folding, accompanied by the formation of the enamel knot. Cytodifferentiation begins during the bell stage when mesenchymally derived odontoblasts differentiate and secrete pre-dentin and the epithelially derived ameloblasts secrete pre-enamel. These secreted matrices then mineralize into dentin and enamel, respectively.

As is true for vertebrate organ formation in general, mammalian tooth development is largely dependent on sequential and reciprocal epithelial-mesenchymal interactions (Thesleff et al., 1995). These processes involve a series of inductive and permissive interactions that lead to determination, differentiation, growth, and organization of odontogenic tissues. Both dental epithelium and mesenchyme are essential for tooth formation (McCarrol and Dahlberg, 1934; Koch, 1967). Classical tissue recombination experiments have demonstrated that the odontogenic potential, the capability of a tissue to instruct an adjacent tissue to form a tooth, arises in dental epithelium initially, and then at E12.5 shifts to dental mesenchyme where it remains in the dental papilla till birth (Yoshikava and Kollar, 1981; Mina and Kollar, 1987; Lumsden, 1988). Before E11.5, the presumptive molar epithelium is able to induce non-dental mesenchyme to form a tooth, but only the cranial neural crest–derived mesenchyme is odontogenically competent, the capability of a tissue to respond to odontogenic signals and to support tooth formation. After E12.5, the dental mesenchyme acquires odontogenic potential, which enables the dental mesenchyme to direct non-dental epithelium to form a tooth.

Many genes have been identified to be expressed in the mammalian developing tooth. Members of several growth factor families such as Bone Morphogenetic Proteins (BMPs), Fibroblast Growth Factors (FGFs), Wnt, and Hedgehog (Hh) families, may function to regulate epithelial–mesenchymal interactions in tooth development from the very beginning (Jernvall and Thesleff; 2000; Thesleff and Mikkola 2002). The spatial and temporal expression pattern of a number of genes encoding transcription factors have been revealed in the developing tooth. Prominent among them are homeobox-containing genes, including Msx1, Msx2, Pax9, Lef1, Dlx1, Dlx2, Barx1, Lhx6, and Lhx7. The overlapping expression patterns of the transcription factors and growth factors in the developing tooth suggest a relationship between these two classes of gene products in inductive interaction. It was proposed that transcription factors in a tissue layer activate the expression of growth factors in response to the signaling of growth factors produced from the adjacent tissue, forming a signaling network that regulates organogenesis (Chen and Maas, 1998). Functional analysis, using gene targeting and transgenic techniques, have revealed a critical role of many of these genes in the development, differentiation, and patterning of the mammalian tooth.

Gene targeting and transgenic techniques have been widely employed as powerful tools to study both loss- and gain-of-function of a gene of interest in organogenesis and pathogenesis in mice. These techniques have helped to greatly improve our understanding of tooth development and reveal many important genes that regulate odontogenesis. In practice, however, they remain laborious, expensive, and time-consuming, and do not permit large-scale functional genomic studies. RNA interference (RNAi) is an evolutionarily conserved process in a diverse group of organisms including plants, C. elegans, Drosophila, and mammals (Hannon, 2002; Calegari et al., 2002). RNAi has recently emerged as a powerful genetic tool to silence gene expression for the analysis of gene function. A DNA vector-based RNAi technology was recently developed to generate short hairpin RNA (shRNA) that functions as small interfering RNA (siRNA) to stably suppress gene expression in mammalian cells (Paddison et al., 2002; Sui et al., 2002). Various gene transfer techniques have been used to accomplish the introduction of siRNA constructs into cells. Virus-based delivery systems, such as retroviral and lentiviral integration and adenoviral expression, are well-characterized stable expression technologies for siRNA expression in target cells (Barton and Medzhitov, 2002; Robinson et al., 2003; Tiscornia et al., 2003; Shen and Reske, 2004). These virus-based delivery systems have been mainly used in cell culture studies, although lentivirus-mediated RNAi was also shown to silence target gene expression in transgenic animals (Robinson et al., 2003; Tiscornia et al., 2003; 2004; Ventura et al., 2004). However, their application in developing organs of mammalian embryo is highly limited due to the lack of access to specific tissues or organs in developing embryos in uterus and the low efficiency of viral infection in vivo compared to cells in culture. Overcoming these obstacles will permit application of these virus-based systems in studying gene function in organogenesis.

Here we present a method using lentivirus-mediated RNAi to study gene function in mammalian tooth development. We show that mouse embryonic molar tooth germs, once dissociated into single-cell suspension and then re-aggregated, are able to recapitulate normal developmental process and develop into a well-differentiated tooth organ under the mouse kidney capsule. This unique property of mammalian tooth development allows high infection efficiency of siRNA-expressing lentivirus on suspended dental cells. By viral-mediated RNAi knockdown of Msx1 or Dlx2 in the dental mesenchyme, we are able to reproduce the identical tooth phenotype seen in mice deficient for Msx1 and Dlx2, respectively. We also demonstrate that silencing of Barx1 in the dental mesenchyme results in an arrest of tooth development at the bud stage, indicating a critical role for Barx1 in tooth development.

RESULTS

Molar Mesenchyme and Incisor Mesenchyme Differ in Their Odontogenic Potential

Most of our current knowledge on tooth development derives from studies using mouse molar as a model system. Despite the significant difference in the final tooth shape, the basic developmental processes of incisor and molar are thought to be similar, at least in the early stage. However, recent studies have revealed the differential expression of a number of genes in the incisor and molar at the beginning of tooth development. For example, Barx1 expression is restricted to the molar mesenchyme while Islet1 transcripts are only detected in the incisor epithelium before and at the bud stage (Tissier-Seta et al., 1995; Mitsiadis et al., 2003). Thus, incisor seems to differ from molar, at least in terms of the expression of some genes. Previous studies have established that mouse dental mesenchyme after E12.5 acquires the odontogenic potential, capable of inducing non-dental epithelial tissue to form a tooth (Mina and Kollar, 1987; Lumsden, 1988). However, since most of the previous work was carried out using molar mesenchyme, whether the embryonic incisor mesenchyme possesses this odontogenic potential remains elusive. To address this question, we analyzed the odontogenic potential in the incisor mesenchyme by tissue recombination experiments. E13.5 mouse mandibular incisor mesenchyme was recombined with E10.5 2nd branchial arch epithelium of Rosa26 mice. The latter has been used as undifferentiated non-dental epithelium in tissue recombination (Mina and Kollar, 1987). E13.5 molar mesenchyme was used in parallel as positive controls. Histological analyses of tissue recombinants demonstrate that although the incisor mesenchyme was able to induce non-dental epithelium to form an early tooth bud-like structure 24 hr after tissue recombination (Fig. 1A), a similar result seen in that using molar mesenchyme (Fig. 1B), none of 65 tissue recombinants of incisor mesenchyme produced a tooth after 2 weeks in subrenal culture. Instead, keratinized cysts and periodontal bone structures were found in these grafts (Fig. 1C). In contrast, 26 teeth were found in 35 grafts of molar mesenchyme recombined with the 2nd arch epithelium (Fig. 1D). In addition, none of the 11 tissue recombinants of E14.5 incisor mesenchyme and E10.5 2nd arch epithelium gave rise to a tooth after 2 weeks in subrenal culture (data not shown). However, similar to the molar mesenchyme (Fig. 1F,H), the incisor mesenchyme at this stage is also odontogenically competent, as evidenced by tooth formation in the recombinants of the E13.5 incisor mesenchyme and the E10.5 presumptive dental epithelium (from the future molar forming sites) (Fig. 1E,G). We, therefore, conclude that although incisor and molar tooth germs are morphologically similar at E13.5, they begin to exhibit different developmental properties. The incisor mesenchyme is odontogenically competent but does not possess the odontogenic potential at this stage as well as at E14.5.

Figure 1. Embryonic incisor mesenchyme does not possess odontogenic potential but is odontogenically competent. A: An epithelial tooth bud (de) formed in a tissue recombinant of E13.5 incisor mesenchyme and the secondary arch epithelium of E10.5 Rosa26 embryo after 24 hr in organ culture. B: Formation of an epithelial tooth bud (de) is seen in a recombinant of E13.5 molar mesenchyme and E10.5 2nd arch epithelium. C: Fiber cyst instead of tooth and periodontal bone (b) formed in a recombinant of E13.5 incisor mesenchyme and E10.5 2nd arch epithelium. D: A section through a tooth retrieved from a graft after 2 weeks subrenal culture, which consisted of E13.5 molar mesenchyme recombined with a piece of E10.5 2nd arch epithelium. The section was processed with Azan dichromic staining that stained dentin (d) blue and enamel (e) red. E,G: A typical incisor formed in a recombinant of an E13.5 incisor mesenchyme and an E10.5 presumptive dental epithelium (E). Sectioning of the incisor showed formation of dentin (d) and enamel (e) at one side of the tooth, a typical characteristic of an incisor. F,H: A molar tooth retrieved, after 2 weeks in subrenal culture, from a graft of an E13.5 molar mesenchyme recombined with an E10.5 2nd branchial arch epithelium. Arrows point to the characteristic molar cusps (F). A section through a retrieved molar showed formation of dentin (d) and enamel (e). Scale bar = 10 μm in (A,B), 100 μm (C–H).

Suspended Molar Mesenchymal Cells Retain the Odontogenic Potential and Competence

We next asked whether the embryonic molar mesenchyme retains the odontogenic potential following dissociation and re-aggregation. We began with molar mesenchyme from E10.5 to E13.5, and found that molar mesenchymal cells prior to E13.5, after dissociation, do not re-aggregate well. Therefore, we chose to use E13.5 mandibular molar tooth germ in our studies. Molar tooth germs were isolated and the dental epithelia removed. Mesenchymal tissues were dispersed into single-cell suspensions that were plated and cultured for various time points. Cells were harvested and cell number counted. Approximately 1 × 10⁵ cells were re-aggregated into a cell pellet and recombined with the 2nd arch epithelium of a E10.5 embryo. Recombinants were cultured in the Trowell-type organ culture overnight followed by subrenal culture in adult male mice, as described previously (Y. Zhang et al., 2003). Recombinants were harvested at various time points and examined histologically for tooth formation. Strikingly, we found that in 26/40 trials, E13.5 molar mesenchymal cells, once disaggregated into a single cell suspension and re-aggregated immediately, retained their odontogenic potential and induced the formation of morphologically distinct molar when recombined with the 2nd arch epithelium (data not shown). However, when the dissociated molar mesenchymal cells were seeded onto tissue culture dishes for 5 hr before re-aggregation, none of the 20 recombinants formed a tooth, indicating that these cells had lost the odontogenic potential. These results demonstrate that mouse embryonic molar mesenchymal cells retain the odontogenic potential once they are dissociated and re-aggregated immediately, but lose the potential even after a very short period of time in cell culture.

To test if the embryonic molar mesenchymal cells, after cell culture, would retain the odontogenic competence, suspended E13.5 molar mesenchymal cells were plated in cell culture, harvested at various time points, pelleted, and recombined with the E10.5 presumptive dental epithelium, which possesses odontogenic potential. Our results showed that these cells, at day 2, 4, and 6 of culture, retained the odontogenic competence completely and formed teeth in all 33 recombinants (12/12 for day 2, 11/11 for day 4, and 10/10 for day 6; and data not shown). Meanwhile, cell number almost tripled by day 6 of culture, thus providing substantial mesenchymal cells for tissue recombination experiments. Histological analysis demonstrated that tooth development in the recombinants resembles the normal process in vivo, including the bud, the cap, and the bell stages (see Fig. 6A,E,I). However, the odontogenic competence in the cultured mesenchymal cells dropped subsequently as the culture time increased (20/42 at day 7, and 7/51 at day 8 in culture), and was completely lost by day 9 of culture (0/45). In contrast, mesenchymal cells from E13.5 incisor germs lose their odontogenic potential very rapidly after being dissociated and plated in cell culture. Seven teeth were retrieved from 14 recombinants in which incisor mesenchyme was dissociated into single cell suspension and re-aggregated immediately. However, no tooth was found in all 13 recombinants in which incisor mesenchymal cells were plated in cell culture for only 5 hr prior to re-aggregation (data not shown). We, thus, conclude that incisor and molar mesenchyme does not only differ in the odontogenic potential but also in the odontogenic competence, at least at the stage tested.

Figure 6. Tooth development in recombinants infected with lentivirus expressing shRNA. In all cases, E13.5 molar mesenchymal cells were infected with lentivirus indicated, and recombined with E10.5 presumptive dental epithelium from Rosa26 mice. Recombinants were cultured in the Trowell-type organ culture for 2 days and were then harvested for histological analysis, or were grafted for subrenal culture for an additional 2 and 4 days before being retrieved and processed histologically. Rosa26-derived epithelia stained blue. A,E,I: Sections of the recombinants infected lentivirus expressing EGFP alone, following 2, 4, and 6 days of culture, show tooth development at the early bud stage (A), the cap stage (E), and the bell stage (I). B,F,J: Sections of the LV-shDlx2-infected recombinants show the normal tooth development at the bud stage after 2 days in vitro culture (B), the cap stage after 4 days in culture (F), and the bell stage following 6 days in culture (J). C,G,K: Normal tooth development to the bud stage was observed in the recombinants infected by LV-shMsx1 after 2 days in culture (C). However, tooth development remains at the bud stage after 4 days in culture (G), and after 6 days in culture, dental epithelium became disorganized (K). D,H,L: Silencing of Barx1 in dental mesenchyme did not affect tooth development to the bud stage (D), but arrested tooth development at the bud stage (H). The dental epithelium lost organized structures in the recombinants after 6 days in culture (L). Scale bar = 10 μm.

Dissociated Molar Epithelial and Mesenchymal Cells Are Able to Organize into a Tooth Organ Autonomously

Since suspended embryonic molar mesenchymal cells retain their odontogenic potential and competence, we further asked if tooth-forming capability is retained in the suspended molar epithelial cells. To do this, we used molar mesenchyme from E13.5 wild type mice and molar epithelium from the same aged Rosa26 mice. A molar mesenchyme was paired with a piece of dental epithelium, and they were then dispersed into single cell suspension together (Fig. 2A). When the suspended cells were re-aggregated and grafted under the kidney capsule, teeth formed in the re-aggregates (Fig. 2F), similar to the results reported recently (Yamamoto et al., 2003). We further examined the tooth developmental process that occurred in the re-aggregates. We observed that the epithelial cells gradually sorted out from the mesenchymal cells and formed epithelial tissues (Fig. 2B–D). Most interestingly, the sorted epithelial cells organized into a well-defined dental epithelial structure that undergoes typical dental epithelial histogenesis within the grafts (Fig. 2E), and eventually a well-differentiated and mineralized tooth formed (22/22, Fig. 2F). This unique developmental capability, the ability to re-organize and to form a mature tooth after tissue suspension, remains in the developing molar germs until E16.5 (6/18; data not shown), but is lost in the re-aggregates of E17.5 suspended molar epithelial and mesenchymal cells (0/8; data not shown).

Figure 2. Tooth formation in re-aggregates of suspended embryonic molar cells. A: Mixture of suspended E13.5 dental epithelial cells from Rosa26 mice and E13.5 molar mesenchymal cells from wild type mice. B–G: Epithelial cells sorted out at 4 hr (B), 24 hr (D), and 72 hr (F) after re-aggregation and culture in the Trowel-type organ culture, and sections through the corresponding re-aggregates respectively (C,E,G). H–J: Sections through re-aggregates after sub-renal culture for 4 days (H), 6 days (I), and 7 days (J) showed a molar tooth at the bell stage. Rosa26 mouse derived epithelial cells stained blue. K: Teeth formed in the re-aggregates of suspended E13.5 molar tooth germ after 2 weeks in subrenal culture. Scale bar = 100 μm (B,D,F), 1 mm (K), 10 μm (A,C,E,G–J).

To examine if embryonic incisor has similar developmental capability, E13.5 incisor epithelium from an actin-EGFP transgenic mice was paired with incisor mesenchyme isolated from E13.5 Rosa26 mice. After suspension (Fig. 3A), cells were re-aggregated, explanted in the Trowell type organ culture, and then grafted under the kidney capsule. Similar to the re-aggregates of molar germ, dissociated incisor epithelial cells went through the similar early sorting process and formed clumps of epithelial tissue after 3 days in culture (Fig. 3B). However, the incisor epithelial cells subsequently failed to self-organize into tooth structure (Fig. 3C), and eventually formed fiber cysts in the grafts after 2 weeks in subrenal culture (0/29; Fig. 3D).

Figure 3. Failure of tooth formation in re-aggregates of suspended embryonic incisor cells. A: A mixture of suspended incisor mesenchymal cells from E13.5 Rosa26 embryo and incisor epithelial cells from E13.5 actin-EGFP transgenic embryo. B: A re-aggregate of the cell mixture from A after 3 days in the Trowell-type organ culture showed sorting out of fluorescent incisor epithelial cells from the surrounding incisor mesenchymal cells, forming epithelial clumps (arrows). C: A section through a re-aggregate at day 7 in subrenal culture exhibited an organized, but nondetal, epithelial structure (Epi). D: Fiber cyst formed in a re-aggregate after 14 days in subrenal culture. Scale bar = 100 μm (B,D), 10 μm (A,C).

Knockdown of Msx1 and Dlx2 in the Molar Mesenchyme Via Lentivirus-Mediated RNAi Resembles Their Knockout Phenotype

The above studies have established that the normal tooth developmental process can be completely recapitulated in vitro using suspended molar mesenchymal cells. Taking advantage of this unique developmental property of the molar germ, we adopted the lentivirus-mediated siRNA delivery system (Robinson et al., 2003; Tiscornia et al., 2003) in the hope of establishing a novel method to manipulate gene function in tooth development. The HIV-based NL-EGFP/CMV lentiviral vector (Reiser, 2000) was engineered to express EGFP and a specific shRNA sequence under the control of the mouse U6 promoter (Sui et al., 2002). The expression of the EGFP reporter gene in the infected cells was used to determine the infection efficiency by fluorescent microscopy and flow cytometry (Fig. 4A,B). Dispersion of E13.5 molar mesenchyme into a single cell suspension permitted high infection efficiency (>92%) of the pNL lentivirus that co-expressed shRNA and EGFP. Accordingly, we used this viral-mediated RNAi system to manipulate the expression of the homeobox genes Msx1 and Dlx2 that are both expressed in the developing molar mesenchyme (Chen et al., 1996; Qiu et al., 1995; Thomas et al., 1997). The knockout of Msx1 in mice leads to an arrest of tooth development at the bud stage (Satokata and Maas, 1994). Although the Dlx1/Dlx2 double mutant mice have the maxillary molar phenotype, the mice carrying mutated Dlx2 have normal mandibular molars (Qiu et al., 1995; Thomas et al., 1997). Therefore, RNAi knockdown of Msx1 was predicted to halt tooth development while silencing of Dlx2 should have no effect on tooth formation. In these experiments, lentivirus expressing the EGFP gene alone (LV-EGFP) was included as a control. For each target gene, multiple shRNA sequences were tested for their efficacy of repression on mRNA level in the molar mesenchymal cells. To do this, the level of target mRNA expression was determined by real-time PCR in aggregates 62 hr after infection with lentivirus expressing each of the shRNA sequences, and compared to the level of mRNA expression in aggregates infected with LV-EGFP. The shRNA sequence that gave the highest level of repression on the expression of each target gene (Msx1 mRNA by 67% and Dlx2 by 77%) was selected for use in the following experiments (Fig. 4C, see Experimental Procedures section).

Figure 4. Lentiviral infection and gene silencing in embryonic molar mesenchymal cells. A: The efficiency of infection was determined by flow cytometry using EGFP expression as a marker (right histogram) as compared with uninfected control cells (left histogram). We readily detected 92% and higher infection efficiency in our studies. B: Flourescent microscopy shows the highly efficient infection of dental mesenchymal cells by lentivirus expressing EGFP and shRNA against Msx1 (LV-shMsx1). C: Efficacy of lentivirus-mediated RNAi on target genes. The shRNA sequences used in this study suppressed, on average, Msx1 mRNA level to 33%, Dlx2 to 23%, and Barx1 to 13%, as compared to LV-EGFP-infected controls.

Dental mesenchymal cells from E13.5 mandibular molars were infected with lentivirus expressing the shRNA of choice specific for each target gene, pelleted, reaggregated, and recombined with the E10.5 presumptive dental epithelium. After 12 hr in the Trowell type organ culture, the recombinants were grafted for subrenal culture for 2 weeks and then were analyzed histologically for tooth formation. Our data demonstrate that dental mesenchymal cells infected with control LV-EGFP or LV-shDlx2 gave rise to a well-organized tooth organ (Fig. 5A,E,F, and Table 1). These results indicate that infection of dental mesenchymal cells with lentivirus does not interfere with normal tooth development and correlates with the lack of tooth phenotype in the Dlx2 knockout mice. In contrast, grafts of dental mesenchyme infected by LV-shMsx1 produced only keratinized cysts instead of teeth (Fig. 5C; Table 1). This phenotype mimicked additional control assays in which E13.5 Msx1 knockout molar germs were grafted in parallel (Fig. 5D) (Z. Zhang et al., 2003).

Figure 5. Functional silencing of target genes in molar mesenchyme by lentivirus-mediated RNAi. A: A tooth formed with deposition of dentin (d) and enamel (e) in the recombinant of LV-EGFP-infected E13.5 molar mesenchymal cells and E10.5 presumptive dental epithelium after 2 weeks in subrenal culture. B: Keratinized cyst formed in a graft of LV-shBarx1 infected molar mesenchyme and E10.5 presumptive dental epithelium. C,D: Infection of molar mesenchymal cells with LV-shMsx1 produced keratinized cyst (C) in a recombinant after 2 weeks subrenal culture, resembling the phenotype of E13.5 Msx1 mutant molar following 2 weeks of culture underneath the kidney capsule (D). E: Infection of molar mesenchymal cells with LV-shDlx2 did not affect tooth formation. F: Fluorescent image of teeth formed in the grafts infected with LV-shDlx2. Scale bar = 100 μm (A–D), 1 mm (E,F).

Table 1. Tooth Formation by E13.5 Molar Mesenchymal Cells Infected With siRNA-lentivirus

Target gene	Number of implanted recombinants	Number of teeth retrieved
EGfp	24	20
Dlx2	24	17
Msx1	21	0
Barx1	23	0

To determine whether knockdown of Msx1 by LV-shMsx1 in dental mesenchyme exactly resembles the Msx1 knockout tooth phenotype, grafts were harvested at 2-, 4-, and 6-day after tissue recombination and analyzed histologically. For easy visualization, E10.5 presumptive dental epithelia from Rosa26 mouse embryos were used for tissue recombination. The results demonstrated that tooth development in the recombinants infected with LV-EGFP and LV-shDlx2 proceeded through the bud (Fig. 6A,B), the cap (Fig. 6E,F), and the bell (Fig. 6I,J) stages normally, closely resembling the in vivo process. In the recombinants infected by LV-shMsx1, the tooth bud formed normally after two days in culture, as compared with controls (Fig. 6A–C). However, while control teeth had developed to the cap stage by 4 days after recombination (Fig. 6E,F), LV-shMsx1 infected tooth germs remained at the bud stage (Fig. 6G). When assessed 6 days after recombination, the control recombinants had reached the bell stage, while tooth development in the LV-shMsx1 infected recombinants remained at the bud stage (Fig. 6K). The dental epithelium in the LV-shMsx1 infected recombinants became disorganized (Fig. 6K), and eventually these recombinants formed keratinized cysts (Fig. 5C). We conclude that knockdown of Msx1 or Dlx2 mRNAs in the dental mesenchyme via lentivirus-mediated RNAi in this in vitro system faithfully mimics the tooth phenotype observed in the knockout mice, that is arrest at the bud stage and no abnormalities, respectively (Satokata and Maas, 1994; Qiu et al., 1995). Thus, this in vitro knockdown approach provides a rapid approach to assess gene function in molar tooth development.

Barx1 Is Critical for Molar Tooth Development Beyond the Bud Stage

Barx1, a homeobox gene, is expressed specifically in the molar mesenchyme during early tooth development (Tissier-Seta et al., 1995). The restricted Barx1 expression in the presumptive molar mesenchyme is the result of the antagonistic effects of Fgf8 and Bmp4, which leads to the tooth type determination of the molar (Tucker et al., 1998). Although Barx1 knockout mice were recently generated, a tooth phenotype has not yet been reported (Kim et al., 2005). Therefore, we made use of this new RNAi system to investigate the role of Barx1 in tooth development. Real-time PCR quantification indicated 87% down-regulation of the Barx1 mRNA in the molar mesenchymal cells infected by LV-shBarx1 (Fig. 4C). Similar to the Msx1 knock-down, infection of dental mesenchymal cells by LV-shBarx1 produced keratinized cysts in the grafts (Fig. 5B). None of the 23 grafts gave rise to a tooth organ (Table 1). To examine at what stage the tooth development is arrested in the Barx1 knockdown recombinants, tissue recombination was carried out, and grafts were analyzed at 2, 4, and 6 days after recombination, as described above. As shown in Figure 6D, after 2 days in culture, a normal tooth bud formed in the LV-Barx1 infected recombinant. However, after 4 days in culture, the tooth bud remained at the bud stage (Fig. 6H), and the epithelial tissue underwent a degeneration in which the epithelial cells lost their healthy morphology as evidenced by ambiguous cell boundary, hollow space between keratinized cells, and lower reporter-gene expression after 6 days in culture (Fig. 6L). Our results demonstrated for the first time that Barx1, besides its implication in the tooth type determination, plays a critical role in early tooth development, being necessary for progression from the bud stage to the cap stage. These results warrant further investigation of Barx1 in mammalian tooth development.

DISCUSSION

The results presented in this report demonstrate that the mammalian molar tooth germ possesses very unique developmental properties. Molar mesenchyme retains its odontogenic potential after being dissociated into single cell suspension. The molar mesenchymal cells remain odontogenically competent following suspension in cell culture for a relatively longer period. Moreover, the suspended molar cells are able to re-initiate the tooth developmental program and form a morphologically distinct tooth organ. Interestingly, these developmental properties do not reside in the incisor, at least at the stages tested. Our results show that the E13.5 and E14.5 incisor mesenchyme apparently lacks the odontogenic potential, being unable to induce non-dental epithelium to become tooth tissue. A considerable difference apparently exists in the molecular mechanism that regulates odontogenesis of molar and incisor, respectively. Identification of the genetic components that account for the different developmental properties of the molar and incisor germs is currently underway. Certainly, the possibility exists that incisor mesenchyme at a certain stage that was not tested in this study possesses odontogenic potential.

It has been shown previously that well-differentiated teeth form in biodegradable scaffold seeded with single cell suspension dissociated from tooth buds of pig and rat (Young et al., 2002; Duailibi et al., 2004). Our and others' results showed that a scaffold is not necessary for ex vivo development of tooth explants, as demonstrated by the evidence that re-aggregates of dissociated molar tooth germ form teeth underneath the mouse kidney capsule (Yamamoto et al., 2003; Hu et al., 2005a, b; this study). Within the re-aggregates, dispersed dental epithelial cells gradually sort out from mesenchymal cells, form epithelial tissues, and then undergo typical dental epithelial histogenesis. This unique tooth developmental property will provide an excellent model system to study cell-sorting mechanism and molecules involved in the process.

The ease with which it can be accessed, manipulated, cultured in vitro, and transplanted to ectopic sites where it can develop and differentiate into a recognizable tooth organ makes mammalian tooth germ an excellent model organ for studying fundamental questions of developmental biology. Various approaches have been developed to study gene function in tooth development, including bead implantation (Vainio et al., 1993), in vitro organ culture with exogenous proteins or antisense oligonucleotides (Chai et al., 1994; Amano et al., 1999), and DNA electroporation (Angeli et al., 2002). However, these methods are often associated with transient delivery and low penetration of exogenous reagents, or low transduction efficiency of DNA and cytotoxicity or damage to the tissues. RNAi has been widely used to silence gene expression in cell lines as well as in organisms. The lentiviral vector system has been successfully employed to deliver siRNA in various cell lines, tissues, and preimplantation embryos (Robinson, et al., 2003; Tiscornia et al., 2003, 2004; Ventura et al., 2004). However, limited accessibility to specific tissues or organs in developing mouse embryos and less efficiency of viral infection as compared to infection on cells in culture restrict its application in study gene function in a developing organ.

In this study, we utilized the unique molar developmental properties to manipulate the tissue and assess gene function in the developing tooth in vitro. Dissociation of molar primordial tissues into single cell suspension allows all cells to be treated equally and overcomes the lower infection efficiency of non-competent viruses in a solid tissue. The ability of lentivirus to infect both dividing and non-dividing cells and to integrate in the chromosome allows not only the high efficiency of gene transduction but also the long-term effects. The highly successful rate of tooth formation in the tissue recombinants infected by LV-EGFP and LV-shDlx2 indicates that the lentiviral vector alone and the expression of a specific short hairpin RNA do not interfere with the normal function of the molar mesenchymal cells and their capability to form a tooth. In our assay system, we use E10.5 presumptive dental epithelium that possesses the odontogenic potential and relatively old E13.5 molar mesenchyme. The mesenchymal cells remain odontogenically competent after dissociation and in cell culture. When these cells are re-aggregated and recombined with the E10.5 presumptive dental epithelium, tooth development starts from the very beginning, recapitulating the normal tooth developmental process both in morphology and in timing. The knock-down of a target gene via RNAi before recombination would inhibit gene function in the dental mesenchyme from the early stage of molar development. This is supported by the evidence that Msx1 knock-down in dental mesenchyme leads to an identical tooth phenotype that is also seen in the Msx1 knockout mice. Using this lentivirus-mediated RNAi system, we demonstrate that Barx1 is critical for molar development beyond the bud stage, a stage at which tooth development is most frequently arrested in various knockout mice. While we chose to knock down the expression of specific target genes in this study, this system can also be used to study gain-of-gene function in tooth development by lentivirus-mediated expression of a target gene. However, this system has certain limited applications. For example, determined dental epithelium and dental mesenchyme are used for tissue recombination. It is, therefore, impossible to assess genes' function in the commitment of odontogenic fate and the determination of tooth type using this system. In addition, this method can only be applied to study molar development. In summary, we describe the different developmental properties between molar and incisor tooth germs in mice. We also establish a new method that can be used to rapidly assess loss-of-gene function in mammalian molar development.

EXPERIMENTAL PROCEDURES

In Vitro Tooth Germ Manipulation, Tissue Recombination, Organ Culture, and Subrenal Culture

For tissue recombination, mandibular molar or incisor tooth germs were isolated from E13.5 mouse embryo, incubated in 2.25% trypsin, 0.75% pancreatin in PBS on ice for 10 min, and the dental epithelia removed (Zhang et al., 2000). The 1st or the 2nd branchial arches were isolated from E10.5 embryos. The presumptive dental epithelium (the future molar-forming sites) from the 1st or the 2nd arch epithelium (from external site of the arch) was separated from the mesenchyme. Incisor or molar mesenchyme from E13.5 embryo was recombined with the epithelium isolated from either the 1st or 2nd arch. The tissue recombinants were cultured in the Trowell type organ culture for 24 hr prior to being subjected to subrenal culture (Y. Zhang et al., 2003). To make single cell suspension, E13.5 dental mesenchyme isolated from 1 litter of embryos was pooled and incubated in a Ca²⁺/Mg²⁺-free CMF-Tyrode solution at room temperature for 5 min. Tissues were transferred to culture medium and dispersed into a single cell suspension with the aid of a micropipette. Suspended dental mesenchymal cells were either plated on dishes in DMEM supplemented with 20% FCS for various time points, or subjected to lentivirus infection immediately (see below). Plated cells were trypsinized, re-suspended in culture medium, and the number of cells quantified. About 1 × 10⁵ dental mesenchymal cells, either infected with lentivirus or cultured for different periods of time, were added to a 1.5-ml Eppendoff tube, centrifuged at 3,000 rpm for 3 min, and then incubated at 37°C in a 5% CO₂ environment for 2 hr to allow for the formation of a firm cell pellet. Cell pellets were removed from the Eppendorf tubes and recombined with either the presumptive dental epithelium or 2nd branchial arch epithelium of E10.5 mouse embryos. Recombinants were cultured in the Trowell type organ cultures for 12 hr at 37°C in a 5% CO₂ environment, followed by subrenal culture in adult male mice for varying amounts of time. To examine the cell-sorting process and tooth formation in the cultured tooth re-aggregates, a piece of E13.5 mandibular molar epithelium isolated from a Rosa26 embryo was paired with an E13.5 wild type molar mesenchyme in an Eppendorf tube. Tissues were disaggregated into a single cell suspension as described above. After 3 washes with culture medium, cells were spun down to make cell re-aggregates. The re-aggregates were explanted in the Trowell type organ culture for 12 hr and then grafted under the mouse kidney capsule for varying amounts of time. E13.5 Msx1 mutant lower molars were also grafted for subrenal culture following the standard protocol (Y. Zhang et al., 2003; Z. Zhang et al., 2003). Msx1 mutant embryos were determined by a PCR-based genotyping method using genomic DNA from extra-embryonic membranes, as described previously (Chen et al., 1996).

Histology

Samples were fixed in freshly made 4% paraformaldehyde (PFA)/PBS at 4°C. Samples containing tissues from Rosa26 mice were processed for whole-mount staining for β-galactosidase activity, according to the standard protocol (Chai et al., 2000), prior to ethanol dehydration, paraffin embedding, and sectioning. Samples that were harvested after 2 weeks in subrenal culture were de-mineralized in 0.1 M ethylenediaminetetraacetic acid/PBS, followed by dehydration through graded ethanol and paraffin-embedding. Serial sections were made at 10 μm, and were processed with Azan dichromic staining (Presnell and Schreibman, 1997).

Preparation of shRNA Constructs and Lentivirus

The Silencer pre-designed shRNA sequences specific to the selected target genes were obtained from Ambion, Inc (Austin, TX), and were subcloned directionally into the pSilencer vector (Ambion, TX) under the control of the mouse U6 promoter. The DNA cassette containing the U6 promoter and each shRNA sequence was released by XbaI and inserted into a unique XbaI site present 5′ to the CMV promoter driving EGFP of the pNL lentiviral vector (Reiser et al., 2000). The efficacy of each shRNA sequence on target gene mRNA levels was determined by real-time PCR following infection of embryonic molar mesenchymal cells with shRNAi-expressing lentiviruses (see below). Four different shRNA sequences targeting different segments of the Msx1 mRNA, and two of each against Dlx2 and Barx1 mRNAs were tested. The sense siRNA sequences that yielded the highest repression level against each of the target genes were used in this study and are: for Msx1, 5′-AAGGATGCAGAGGCCAAGAGA-3′; for Dlx2, 5′-AAGGAAGACCTTGAGCCTGAA-3′; and for Barx1, 5′-AGTCGCACCGTATTCACTGA-3′.

Lentiviral production was carried out by transfecting human embryonic kidney 293T cells with lentiviral and packaging vectors. The resulting supernatant was collected 48 hr after transfection, filtered and concentrated by ultracentrifugation (Reiser, 2000). Titers were determined by infection of 293T cells with serial dilutions of concentrated viruses using EGFP expression as a marker. We routinely harvested viruses with titers of 1 × 10⁸–10⁹ infectious units/ml.

Infection of Dental Mesenchymal Cells

A single-cell suspension of E13.5 mandibular molar mesenchyme was prepared as described above and the number of cells determined. About 5 × 10⁶ cells were suspended in 200 μl of culture medium in an Eppendorf tube into which 20 μl of concentrated virus was added. The tube was placed with the cap open in an incubator at 37°C for 2 hr. Cells were then plated onto a 35-mm culture dish with 1.5 ml medium (DMEM supplemented with 10% FCS) and cultured for 12 hr in virus-containing medium. To determine the infection efficiency, infected cells were cultured for an additional 2 days in regular medium, and then were harvested and applied to a Beckman-Coulter Cytomics FC 500 Fluorescent Activated Cell Sorter (FACS). Infected cells were harvested 12 hr after plating with virus-containing medium and the cell number was counted. To determine the RNAi efficacy of each shRNA sequence, about 1 × 10⁵ cells were spun down to make a cell pellet, which was then recombined with a piece of E10.5 presumptive molar epithelium. The recombinants were placed in Trowell type organ culture for 48 hr before being harvested for RNA extraction. Real-time PCR was performed to determine the level of the target mRNA expression using pre-designed Taqman Gene Expression Assay kits from Applied Biosystems (Foster City, CA). For tooth formation assays, after 12 hr in a Trowell-type organ culture, the recombinants were subjected to subrenal culture for 2 weeks.

Acknowledgements

We thank Dr. Carol Burdsal for reading and comments on the manuscript. This work was supported by grants from the National Institutes of Health (DE16623, DE15123, DE12329) and the Millennium Trust Health Excellent Fund (HEF-2000-05-04) from the Louisiana Board of Regents to Y.P.C., and National Natural Science Foundation of China (30270652, 30370705) and Fujian Provincial Department of Science and Technology, China (2002I006, C0320002) to Y.Z.