α11β1 Integrin-Dependent Regulation of Periodontal Ligament Function in the Erupting Mouse Incisor

Abstract

The fibroblast integrin α11β1 is a key receptor for fibrillar collagens. To study the potential function of α11 in vivo, we generated a null allele of the α11 gene. Integrin α11^−/− mice are viable and fertile but display dwarfism with increased mortality, most probably due to severely defective incisors. Mutant incisors are characterized by disorganized periodontal ligaments, whereas molar ligaments appear normal. The primary defect in the incisor ligament leads to halted tooth eruption. α11β1-defective embryonic fibroblasts displayed severe defects in vitro, characterized by (i) greatly reduced cell adhesion and spreading on collagen I, (ii) reduced ability to retract collagen lattices, and (iii) reduced cell proliferation. Analysis of matrix metalloproteinase in vitro and in vivo revealed disturbed MMP13 and MMP14 synthesis in α11^−/− cells. We show that α11β1 is the major receptor for collagen I on mouse embryonic fibroblasts and suggest that α11β1 integrin is specifically required on periodontal ligament fibroblasts for cell migration and collagen reorganization to help generate the forces needed for axial tooth movement. Our data show a unique role for α11β1 integrin during tooth eruption.

Integrins constitute a family of multifunctional cell adhesion receptors involved in a variety of biological processes. In higher vertebrates the integrin family is composed of 18 α subunits and 8 β subunits. Out of the 24 integrin heterodimers, α1β1, α2β1, α10β1, and α11β1 act as primary receptors for native collagens. The α3β1 integrin, initially described as a collagen receptor, was later shown to act as a receptor for laminin-5 and does not display any measurable affinity for collagen I (21). The observed role of αvβ3 integrin in collagen lattice remodeling in vitro (13, 33) and in vivo (33) may be related to indirect binding to RGD ligands which are locally deposited in the collagen matrix.

Collagen-binding integrins bind native collagens via their αI domain, recognizing a GFOGER motif (30, 52) or similar sequences with varying specificities and affinities depending on the collagen type and fibrillar status (27). In addition to mediating cell adhesion, cell spreading, and cell migration (22), collagen-binding integrins regulate collagen turnover (32) and take part in assembling (31, 49) and reorganizing three-dimensional collagen matrices (23).

Fibroblasts express characteristic collagen receptor repertoires in a tissue-specific manner, partly reflecting their different embryonic origins. We have previously demonstrated that α11β1 is the only detectable collagen-binding integrin in the incisor periodontal ligament (PDL) fibroblasts of mice (38). The PDL in the mouse incisor is composed of a relatively dense connective tissue on the side facing the tooth and a loose connective tissue region rich in blood vessels on the side facing the bone. The odontogenic ectomesenchyme is the source of the cells populating these two domains in the PDL, which are sometimes referred to as tooth-associated fibroblasts and bone-associated fibroblasts, respectively. Both cell populations are characterized by the expression of Runx2 (50), periostin (29, 43), and collagen XII (41), whereas markers distinguishing between these two cell populations in incisors and molars are scarce.

Studies of tooth eruption have indicated that it is a complex multifactorial process, initially dependent on osteoclast activity to generate an eruptive path through the bone (50). In the continuously erupting rodent incisor, the PDL has been shown to play a central role during the subsequent supraosseous phase of tooth eruption (37), although the exact role of the ligament in this process is controversial.

Two central functions of the PDL, independent of its role in tooth eruption, are to anchor the tooth in its socket and to act as a shock absorber during mastication. For this purpose an elaborate network of collagen fibrils exists. Collagen turnover in the PDL is high, most likely reflecting the constant need for remodeling in order to resist the forces of mastication. Collagen synthesis occurs throughout the mouse incisor PDL, but remodeling has been reported to occur mainly in a middle region called the transitional zone (5). Elaborate control mechanisms exist to maintain a constant width of the PDL and to prevent bone and cementum from encroaching into the PDL space. The transcription factor Msx2 has recently been suggested as a molecular defense mechanism that prevents mineralization in the PDL (51), but the nature of the specific molecular regulators on the surface of the PDL fibroblasts that “sense” the matrix status and control collagen turnover has remained elusive.

We describe here the identification of α11β1 integrin as a central molecule on the surface of PDL fibroblasts. Mice deficient in the α11 integrin chain are dwarfed, show signs of malnutrition, and display increased mortality due to defective incisors. Data from our in vivo and in vitro models suggest that the tooth phenotype is due to a primary defect in incisor PDL and support a model in which α11β1 is needed during the migration/reorganization of fibroblasts throughout the ligament in order to help generate the forces needed for tooth eruption.

MATERIALS AND METHODS

Generation of α11 integrin-deficient mice.

Four bacterial artificial chromosome (BAC) clones were identified from a spotted 129/SvJ genomic library (BAC Mouse Release II; Genome Systems) using a 0.95-kb probe from the 5′ end of human α11 cDNA (nucleotides −37 to 917) (48). BAC clones were obtained from Genome Systems. After cleavage with SacI, a 12-kb fragment reactive with the 0.95-kb probe was identified, subcloned into the pBluescript SK(+) cloning vector, and sequenced (MWG Biotech).

The cell culture, homologous recombination, and microinjection of embryonic stem (ES) cell clones into C57BL/6J blastocysts were carried out as described previously (17). Two mouse strains originating from clones 95 and 215 were obtained by mating the chimeras with C57BL/6J mice. Homozygous mice were obtained by intercrossing of heterozygous mice. The genotyping of the mice was determined by Southern blotting or genomic PCR (38). In order to remove the PGK neo cassette, Itga11^−/− mice were bred with a mouse expressing Cre recombinase in the tyrosine hydroxylase locus (34).

Western blotting and immunoprecipitations.

Western blotting for detection of the integrin α11 protein and immunoprecipitations were performed as described elsewhere (38). Antibodies used for cell cycle analysis were rabbit anti-mouse p16, mouse anti-human p53 (both Santa Cruz Biotechnology, Inc.), mouse anti-human Rb (BD PharMingen), and rabbit anti-human phospho-Rb (Ser795; Cell Signaling).

RT-PCR.

Total RNA isolation from mouse embryonic fibroblasts (MEFs) and PDL tissue as well as cDNA synthesis was is described elsewhere (38). The sequences of the primers amplifying collagen-binding integrins have been described previously (38). Sequences of additional primers used are listed in Table 1. Total RNA from MEFs cultured in gels was harvested by solubilizing the gels directly in RNAzol, and the RNA was extracted according to the manufacturer's instructions (Wak-Chemie Medical GmbH). After a 15-min DNase treatment (Boehringer) and phenol-chloroform extraction, reverse transcription-PCR (RT-PCR) was performed by following the manufacturer's instructions (Gene Amp RNA PCR kit; Perkin Elmer). The cDNA was used to amplify specific transcripts by PCR by following the manufacturer's instructions (REDTaq ReadyMix PCR Mix; Sigma). Amplification of murine S26 was used for normalization.

TABLE 1.

Primer sequences used in the RT-PCR analysis

mRNA	Forward primer	Reverse primer
COL1A2	GACTTCTACAGAGCTGACC	TTCAACATCGTTGGAACCCTGC
Col3a1	TTAATGGACAAATAGAGA	AATGTCATAGGGTGCGAT
Col12a1	TACAAGACAGGAGAAGGAAA	CATCAAACATCCCAAAAACA
MMP2	GTTTATTTGGGGACAGTGA	GGTCAGTGGCTTGGGGTATC
MMP9	GTTTTTGATGCTATTGCTGA	ACCCAACTTATCCAGATCC
MMP13	GATGCCATTACCAGTCTCCG	GCTCAGTCTCTTCACCTCTT
MMP14	CAAGTGATGGATGGATACCC	TCGCTGTCCACTGCCCTGAA
Osteopontin	ACGACGATGATGACGATG	TGCGTTTGTAGGCGGTCTTC
Periostin	AACAATCCGTGCCACTCA	GGAGGGGGCTTCACTGAT
26S	AATGTGCAGCCCATTCGCTG	CTTCCGTCCTTACAAAACGG
β-Actin	GTGTGATGGTGGGAATGGGT	TCTGGGTCATCTTTTCACGGTTGG

Cell culture.

MEFs were obtained from embryonic day 14.5 (E14.5) embryos as described previously (38). The tails of the embryos were cut and used for genotyping. α11 knockout (ko) and wild-type (wt) MEFs were cultured at 37°C in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% fetal calf serum (FCS; PAA Laboratories) and antibiotics.

In order to obtain immortalized MEFs, primary fibroblast cultures from wt and Itga11 ko embryos were infected with recombinant retrovirus-transducing simian virus 40 (SV40) large T at 37°C for 2 h (a gift from S. Johansson, Uppsala University) (26). After infection, cells were cultured in DMEM plus 10% FCS for approximately 14 days. Noninfected cells died during this period, and growing noncloned cells were expanded and used as bulk-immortalized cells.

In order to restore the function of integrin α11, 10 μg of full-length human α11 cDNA was transfected into SV40-immortalized ko cells as described previously (47), and highly expressed clones were isolated. Two independently obtained clones were used in further experiments.

Histology and immunohistochemistry.

For histological analysis, the lower jaws or knee joints were fixed overnight in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS), pH 7.4, and decalcified in 10% EDTA-PBS for 3 to 4 weeks. After the jaw and joints were embedded in paraffin, 6- to 8-μm sections were cut and stained with hematoxylin and eosin (H&E), toluidine blue, or sirius red. Images were visualized under a Leica microscope equipped with a ProgRess LCD camera.

For immunohistochemistry, specimens were collected in polyvinylpyrrolidone-EDTA solution without fixation and subjected to decalcification for 1 to 2 weeks (28). Following decalcification, the specimens were snap-frozen in liquid nitrogen and sectioned on a Leica cryostat. The primary antibodies were hamster anti-α1 integrin (Ha31/8; BD Pharmingen), rat anti-α2 integrin (Emfret Analytics), rabbit anti-α10 integrin (10), rabbit anti-α11 integrin (38), rat anti-PECAM (CD31; BD PharMingen), rabbit anti-collagen I, rabbit anti-collagen III (both obtained from J. Risteli, University of Oulu, Finland), rabbit anti-collagen XII (obtained from M. Koch, University of Cologne, Germany), rat anti-F4/80 antigen (Serotec), rat anti-tenascin-C (46), and mouse anti-human alpha-smooth muscle actin (clone 1A4; Sigma). The secondary antibodies were Cy2, Cy3, or biotin-conjugated goat anti-rat, goat anti-rabbit, and goat anti-mouse immunoglobulin G (Jackson ImmunoResearch Laboratories). An ABC kit (Vector Laboratories) and DAB substrate kit (Zymed Laboratories) were used to detect biotinylated secondary antibodies. Images were visualized under a Zeiss Axioscope microscope equipped with optics for observing fluorescence and were captured using a digital AxioCam MRm camera.

Detection of proliferation and apoptosis.

The proliferation index in vivo was analyzed using bromodeoxyuridine (BrdU) incorporation as described elsewhere (1). At least three animals of each genotype were used per time point. Three sections from each animal were analyzed. The proliferation index in vitro was analyzed using an immunofluorescence assay for the detection of BrdU incorporated according to the manufacturer's instructions (5′-bromo-2′-deoxy-uridine labeling and detection kit I; Roche Diagnostics). The coverslips were mounted in Vectashield medium with 4′,6′-diamidino-2-phenylindole (DAPI; Vector Laboratories). The total number of nuclei (DAPI) and number of BrdU-positive nuclei were counted. Five fields of 150 to 200 cells were counted for each cell type.

A terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling assay using sections of mouse PDL of different ages (2 months, 6 months, and 1 year) was performed to detect apoptotic nuclei. The sections were processed using a TdT in situ apoptosis detection kit (R&D Systems) according to manufacturer's instructions. Three animals of each genotype were analyzed. Nuclei were counted in whole PDL at the level of the molar block, and the apoptotic index was calculated as the ratio of TdT-positive nuclei to the total number of nuclei.

Micro-CT.

Three-dimensional, high-resolution reconstructions of the upper and lower incisors were obtained using a microcomputed tomography (micro-CT) system. The tissues were fixed in buffer containing 4% PFA, 1% glutaraldehyde, and 0.1 mM sodium cacodylate (pH 7.4) and processed as described elsewhere (36).

Electron microscopy.

The mice were anesthetized with ether and sacrificed by cervical dislocation, and the lower jaw bones were removed. The tissues were fixed in 4% PFA and 0.5% glutaraldehyde for 15 min, dehydrated in a graded series of ethanol up to 70%, and embedded in the acrylic resin LR-Gold (London Resin Company). Semithin (1 μm) and ultrathin (0.8 μm) sections were cut according to procedures previously described in detail (19). For orientation purposes, semithin sections were stained with toluidine blue. For electron microscopy, the sections were rinsed with water and stained with uranyl acetate (10 min) and lead citrate (8 min). Sections were examined with an LEO 906E electron microscope. Three wild-type mice and three α11 knockout mice were investigated.

Eruption measurement.

Mice were anesthetized with an intraperitoneal injection of avertin (0.20 mg/ml, 100 μl per 10 g of body weight), and a horizontal mark was made at the gum level on one of the lower incisors using a dental drill (Silfradent) equipped with a diamond saw. The distance between the mark and the gum was measured 1 week later. The measurements were repeated once a week for 5 weeks, making new marks when necessary. The eruption rate was estimated as the average value of the five measurements for each animal. At least five 3-week-old and 6-month-old mice of each genotype were used.

Cell attachment and cell spreading.

Forty-eight-well plates (Nunc Roskilde) were coated with different concentrations of collagen type I or type IV overnight at 4°C and blocked with 2% bovine serum albumin in PBS for 1 h at 37°C, and cell attachment experiments were performed as previously described (47).

To analyze the number of focal contacts formed, cells were left to attach on collagen type I for 1 h and fixed with 2% PFA. Fixed and washed cells were incubated with mouse anti-human vinculin antibodies (clone hVIN-1; Sigma) and fluorescein isothiocyanate-labeled phalloidin (Sigma) and mounted for microscopy. Images were visualized under a Zeiss Axioscope microscope. For cell spreading, the cells were allowed to spread for 45 min, washed with PBS, and stained with 0.1% crystal violet for 30 min. The number of spread cells and total number of attached cells were counted in six fields of each cell type at ×20 magnification.

Collagen gel contraction.

The collagen gel contraction assay has been described previously (23). DMEM with or without blocking antibodies (anti-α1 Ha31/8, anti-α2 Ha 1/29, or both; BD Pharmingen) was ejected into the wells to detach the gels. Free-floating gels were further incubated at 37°C, and gel diameters were measured under a microscope at the time points indicated.

Flow cytometry.

For fluorescence-activated cell sorter analysis, 75-cm² flasks were coated with collagen type I as described above. Confluent α11 ko and wt MEFs in passage 5 were serum starved for 12 h, washed in PBS, and trypsinized. Following trypsinization, 5 × 10⁵ cells were seeded into collagen-coated flasks in DMEM containing 10% FCS. After 2 h, medium was replaced with DMEM containing 2% FCS. Cells were collected after 48 h and resuspended in PBS. MEFs were permeabilized with cold ethanol overnight at 4°C and stained with propidium iodide (500 μg/ml; Sigma) according to a previously described method (45). Cell cycle data were collected using a FACSCalibur flow cytometer (BD Bioscience) and analyzed with CellQuest Pro software (BD Bioscience).

Analysis of matrix metalloproteinases.

MEFs were used at passages 3 and 4. Three-dimensional collagen gels were performed as described earlier (35). Control cultures of MEFs were performed on uncoated (monolayers) tissue culture plates. Twenty-four hours postseeding, cultures were washed three times with PBS and medium was replaced with serum-free DMEM for 24 h.

Serum-free supernatants from monolayers and collagen gel cultures were centrifuged (5 min, 272 × g) to eliminate debris and suspend cells, and then they were analyzed. Equal amounts of supernatants, normalized to the cell number (volume/2,000 cells/well), were used for gelatin zymography. Collagen gels or monolayers were further processed for RNA preparation.

For gelatin zymography, supernatants were separated on nonreducing sodium dodecyl sulfate (SDS) polyacrylamide gels (10%) containing 1 mg/ml bovine gelatin (Sigma). After electrophoresis the gels were washed in 2.5% Triton X-100 for 30 min to remove SDS and then incubated in enzyme substrate buffer (50 mM Tris-HCl, pH 8, 5 mM CaCl₂) overnight at 37°C. Gels were stained with Coomassie blue R250 and destained in water.

RESULTS

Generation of integrin α11-deficient mouse strains.

To introduce a null mutation in the mouse integrin α11 gene (Itga11), we replaced parts of exon 3 and intron 3 with an internal ribosome entry site, lacZ, and a PGK neo cassette by gene targeting (Fig. 1A and B). The targeting construct was introduced into R1 ES cells by electroporation, and resistant colonies were selected with G418. Southern blotting was used to identify four homologously targeted clones out of 325 that were screened (data not shown). The clones 95 and 215 were selected for further work. Southern blotting of these was also used to verify that integration had occurred only once (Fig. 1C). These clones, with the correctly targeted alleles, were injected into blastocysts of C57BL/6J mice. Chimeric males were mated with C57BL/6J mice, and the offspring were screened with Southern blots of tail DNA for germ line transmission of the targeted allele. Both recombinant ES clones transmitted the mutant allele to the germ line. Intercrossing of heterozygous F₁ mice gave rise to live homozygous offspring (Fig. 1D and E) with the expected Mendelian ratios and with no obvious phenotypic defects at birth. The homozygous males and females were fertile.

FIG. 1.

Generation of integrin α11-deficient mouse strains. (A) Schematic presentation of the transgenic construct and probes used. (B) Close-up of reporter and resistance cassettes of the targeting construct. (C) Southern blotting of ES cells using EcoRV digestion and probe 2. Note the single genomic integration of the construct in both clones. (D) Southern blotting using BamHI digestion and probe 1 on ES clones. (E) F₂ generation of mice derived from clone 215. Wild-type (+/+), heterozygous (+/−), and homozygous (−/−) combinations of the targeted allele are shown. (F) RT-PCR was used to verify the lack of α11 mRNA in the α11^−/− cells. Note the lack of a band when using primers corresponding to the exon 2 to 5 sequence (exon 2-5) in the mutant cells. RT-PCR with primers corresponding to the downstream sequences (exon 15-22) detected small amounts of transcript (about 10%) in tissue from the mutant cells. (G) Immunohistochemical analysis of α11 protein in the knee joint perichondrium of +/+ and −/− genotyped littermate E14.5 embryos. The bar represents 50 μm. (H) Lack of α11 protein in −/− tissues. The caudal parts of the vertebrae from 2-month-old +/+ and −/− mice were homogenized, and the solubilized proteins were separated by SDS-polyacrylamide gel electrophoresis, transferred to a nitrocellulose membrane, and incubated with rabbit anti-mouse α11 antibodies and mouse monoclonal β-actin antibodies. Note the absence of the α11-specific band in the −/− tissue.

To show the lack of normal α11 transcripts in the knockout mice, RNA was isolated from E14.5 wild-type (+/+) and null (−/−) embryonic fibroblasts and RT-PCR analysis was performed (Fig. 1F). No PCR product was obtained from the −/− cells using primers located in exons 2 and 5, whereas RT-PCR using primers downstream of the deletion amplified a small amount of α11 RNA from the null cells, most likely representing out-of-frame RNA transcripts resulting from readthrough. Immunohistochemistry and Western blotting of tissues from the wt and −/− mice showed an absence of any specific α11 immunoreactivity in the latter (Fig. 1G and H).

Characteristics of the skeletal system of α11 integrin-deficient mice.

When the tails were cut for genotyping at 3 weeks of age, it was noted that each litter contained mice that were smaller than their littermates (Fig. 2A). Both the female and male homozygous mice revealed a 20 to 30% reduction in weight on average that persisted into adulthood (Fig. 2B). X-rays showed that the overall size of the skeleton was smaller, compatible with proportional dwarfism (data not shown). To rule out any effect of the neo cassette, the cassette which is flanked by loxP sites in the targeting construct was Cre deleted by crossing with Cre deleter mice. The same phenotype was observed in the mice in which the neo cassette had been removed (data not shown), confirming that the weight difference is due to the inactivation of α11 integrin alleles. The neo-deleted mice were used in all further characterizations.

FIG. 2.

Integrin α11-deficient mice show a reduction in body weight. (A) Ten-week-old male mutant mice showed a reduction in size. Hz, heterozygous (α11^+/−); KO, homozygous (α11^−/−). (B) Wild-type (WT), heterozygous, and knockout F₂ offspring were genotyped, and their body weights were measured at different time points. Mice deficient in α11 show a significant reduction in body weight, corresponding to about 70 to 80% of the weight of their wild-type and heterozygous littermates.

Since α11 is heavily expressed in the perichondrium (Fig. 1G) (38, 47), we initially focused our analysis on the growing skeletal system. However, at the light microscopic level we could not observe any structural defects of the forming cartilage or bone at 1 to 8 weeks (data not shown), nor could we detect any significant differences in chondrocyte proliferation rate between the normal and mutant mice (data not shown).

Tooth phenotype in the absence of α11.

The α11-deficient mice showed increased mortality starting from around 1 year of age, and 10 such mice died between 12 and 19 months, whereas only one control mouse died (data not shown). Necropsy of the α11 null mice revealed severe malnutrition, which led our attention to the digestive system, including the teeth.

We have previously established that α11 is coexpressed with α2 in villus cluster fibroblasts (38), raising the question of whether defects in these cells could affect the villus structure, with consequences for nutrient absorption. The analyses performed so far have revealed no such defects, however (S. N. Popova, D. Gullberg, and P. Simon-Assman, unpublished data).

In the older mice a phenotype in the incisors was clearly visible, prompting closer analysis of the forming teeth. The smaller size and tooth phenotype were present in both strains of independently generated Itga11-deleted mice. All future work was performed on the neo-deleted strain derived from ES clone 95. In the mutant mice there was a clear delay in the time of incisor eruption (Fig. 3A), and an altered tooth shape was observed after this delayed eruption (Fig. 3B and C). In the older mice the incisal part of the upper incisors was frequently missing (Fig. 3C), whereas the intraalveolar part proved to be still present upon removal of the soft tissue (Fig. 3D). Monitoring of tooth eruption showed that it was reduced at 3 to 6 weeks and had essentially stopped at 6 to 7 months (Fig. 3E). Micro-CT of 2-month-old incisors showed increased thickness of the dentin layers in the apical region and, as a consequence, pulp closure (Fig. 3F). The enamel layer proved to be mineralized until well into the apical area (data not shown), and the incisors of the older animals showed buckling of their apices (Fig. 3G and H).

FIG. 3.

α11 integrin-deficient mice display abnormal tooth development and structure. (A) Upper and lower incisors from control and mutant mice were analyzed in the oral cavity at 2 weeks (2w) and 4 weeks (4w). Note the irregular tooth surfaces and pitting in the mutant mice. (B) The incisors of 4-month-old control (ctrl) and mutant (mut) mice were excised from their sockets, and the soft tissue was digested away. Note the altered size and shape of the α11^−/− incisors. (C) A lack of the outer portion of the upper incisors was observed in the α11 null mice at 1 year of age. (D) The upper jaws of 1-year-old mice were dissected and treated with 20% KOH in order to remove the surrounding soft tissue. The upper incisors of the mutant mice are significantly reduced in length. Note also their irregular enamel structure. (E) The eruption rate was measured in heterozygous and α11 null mice during the time of active growth and at 6 to 7 months of age. The rate was significantly slower in the mutant mice at 3 to 6 weeks, and eruption was completely blocked at 6 to 7 months. **, P ≤ 0,02; ***, P ≤ 0.005. (F) Micro-CT of the 2-month-old heterozygous and mutant mouse incisors revealed pulp closure (arrow) and an accumulation of dentin in the region occupied by the pulp in the control teeth. (G to J). At 2 months of age the apical side of the incisors displayed buckling and the PDL was increased in width at the level of the molar block. (K and L). Sirius red staining revealed increased collagen content in the PDL tissue of mutant mice. (M and N). H&E staining showed an increased number of rests of Malassez (arrows) in the PDL from 1-year-old α11^−/− mice. T, tooth; d, dentin; brackets, periodontal ligament. The bar corresponds to 100 μm. (O to R). Ultrastructural organization of collagen fibrils in the control and mutant PDL.

Histological examination of incisor PDL showed increased thickness (Fig. 3I and J) due to increased accumulation of collagens, as judged by sirius red staining (Fig. 3K and L). Fibroblast density in the incisal half of the ligament was also reduced by comparison with the even distribution of cells observed in the normal PDL (data not shown). In addition to increased thickness of the PDL, the acellular cementum was increased in thickness, and an increased number of rests of Malassez was a characteristic feature of the PDL in mice older than 6 months (Fig. 3M and N and data not shown). Electron microscopy revealed normal collagen fibrils and a normal collagen network, with occasional apoptotic cells (Fig. 3O to R and data not shown). The PDL of the molars at all ages appeared to be normal. The molars had clearly erupted and did not show any morphological changes with respect to the ligament or with respect to the hard dental tissues and pulps. To examine whether the defective incisor structure was the underlying cause of the proportional dwarfism, we fed weaned mice a soft diet. This regimen was found to rescue the reduced-weight phenotype, although not completely (data not shown). Soft-food supplementation of the regular hard food pellets efficiently prevented malnutrition and was used routinely thereafter.

In vitro phenotype of α11-deficient mouse embryonic fibroblasts.

MEFs isolated from wt and ko embryos were tested for adhesion to collagen I and collagen IV using fibronectin as a control. Attachment to collagen I was reduced to less than 20% of that in the control cells in the absence of α11 (Fig. 4A), whereas cell attachment to collagen IV was only marginally affected (Fig. 4B). Cell attachment to fibronectin was not affected by the absence of integrin α11 (Fig. 4C). The proportion of cells that had spread on collagen I was reduced by 50% in the integrin α11^−/− fibroblasts but was normal in the case of spreading to fibronectin (Fig. 4D). α11^−/− MEFs formed fewer vinculin-positive focal contacts on collagen I (Fig. 4E) and were found to be less efficient in reorganizing a collagen lattice (Fig. 4F). Whereas wt cells incubated with a mixture of α1 and α2 integrin-blocking antibodies efficiently reduced the collagen lattice to 20% of its original size, gels populated with α11^−/− MEFs and incubated with antibodies to α1 and α2 reduced the lattices only to 60% of their original size. To confirm that the changed collagen gel reorganizing properties were α11 dependent, we retransfected SV40-immortalized α11^−/− cells with human α11 cDNA and were able to efficiently correct the collagen gel contraction defect (Fig. 4G). We also compared levels of MMP expression (MMP13 and MMP14) and activity (MMP2 and MMP9) for cells cultured on monolayers in floating and attached collagen gels. Although the absolute levels of MMP RNA varied with cell isolates, a reproducible drop in MMP13 and MMP14 RNA levels was observed in the α11^−/− MEFs (Fig. 5) cultured in floating collagen gels, whereas MMP2 and MMP9 gelatinase activity in MEFs was unaffected by α11 integrin expression.

FIG. 4.

In vitro analysis of α11-deficient cells. (A to C) wt or ko MEFs were seeded on plates coated with collagen type I (A), collagen type IV (B), and fibronectin (C) and allowed to attach. The results shown are from one representative experiment. (D) Cell spreading on fibronectin (FN) and collagen (Col I) of cells containing or lacking α11β1 integrin. (E) The ability of wt and ko MEFs to form focal contacts was analyzed. Note the decreased number of vinculin-positive focal contacts formed by ko MEFs. β-Actin was stained with phalloidin (green). The bar corresponds to 100 μm. (F) Collagen gels were formed, and time-dependent contraction was measured in the presence or absence of anti-integrin α1 antibody (Ab), anti-integrin α2 antibody, or both. The data shown are from one representative experiment. Each point represents 8 to 12 samples. (G) SV40-immortalized mouse embryonic fibroblasts lacking α11 (ko) were stably transfected with human α11 cDNA (clone 14), and collagen gel contraction was compared with SV40 wild-type MEFs (wt). Immunoprecipitation was performed to analyze levels of different collagen-binding integrins in clone 14 (cl. 14) using preimmune (PI) serum and antibodies to α1, α2, and α11. Note the presence of α11 in knock-in (ki) cells.

FIG. 5.

Analysis of metalloproteinases in cultured embryonic fibroblasts. Mouse embryonic fibroblasts (MEFs) from wt and ko mice were cultured in monolayers (m) and inside a floating (f) collagen gel as described in Materials and Methods. RT-PCR was performed to analyze MMP13 and MMP14, whereas MMP2 and MMP9 activity was monitored by zymography.

In vitro cultures of MEFs revealed that integrin α11^−/− cells stopped growing at earlier passages than wt cells. In accordance with this, cells that were homozygous for the α11 deletion showed reduced proliferation and BrdU incorporation on collagen I (Fig. 6A). Fluorescence-activated cell sorter analysis revealed blockage of cells at the G₁/S transition (Fig. 6B), but we failed to detect changed levels of p53, pRb, or p16 in MEFs deficient in α11 (data not shown).

FIG. 6.

Analysis of cell proliferation in α11-deficient cells. (A) Mouse embryonic fibroblasts were cultured with medium containing a BrdU labeling mix for 1 h at 37°C. The total numbers of stained cells were counted and compared. The data shown are from one representative experiment. **, P ≤ 0.02. Note the reduced incorporation of BrdU into ko MEFs. The bar corresponds to 100 μm. (B) Flow cytometry analysis of cell proliferation in wt and ko cells. Note the accumulation of α11 integrin-deficient fibroblasts in the G₀/G₁ phase of the cell cycle.

In situ localization of α11 RNA and immunohistochemical analysis of α11 protein in periodontal ligaments.

To gain an understanding of the differential phenotype in the incisors, we followed the expression of α11 RNA and protein during E14 to E17 and postnatally (Fig. 7A to D). α11 was expressed in the dental follicular mesenchyme that forms the PDL and in the preodontoblasts of the developing molars and incisors. Immunohistochemical staining of the embryonic/postnatal follicular mesenchyme revealed no differences in the levels of expression of integrin α11 or the repertoires of collagen-binding integrins between the incisor and molar PDLs (data not shown). α1 and α2 were present in capillaries, whereas α10 was lacking completely in the PDL. This expression pattern was largely retained in the adult incisor PDL (Fig. 7E to I). One noticeable difference, however, was that the levels of α11 in the molar PDL tissue were low (Fig. 7I) compared to those of the adult incisors (Fig. 7H). We failed to detect any compensatory regulation of collagen-binding integrins in the incisor or molar PDL in the absence of α11 (Fig. 7J to M). This was also confirmed by semiquantitative PCR (Fig. 7N).

FIG. 7.

Analysis of α11 integrin-deficient periodontal ligaments. (A to D) Frontal sections of mouse heads at embryonic day 15 (E15) and postnatal day 2 (P2) were hybridized with antisense riboprobe specific for α11 mRNA. Note α11 expression in incisor and molar odontoblasts (ob). I, incisor; t, tongue; arrow, periodontal ligament. (E to M) Immunohistochemical localization of collagen-binding integrins in PDL of α11 ko and control mice. Sections of lower jaws from 2-month-old mice were incubated with antibodies to integrins α1, α2, α10, and α11. Note the strong signal for integrin α11 in the incisor PDL (H) and weaker signal for α11 in molar PDL (I) Note also the expression of α1 and α2 integrin chains in capillary endothelium on the border between PDL and alveolar bone (arrows in panels E and F) and lack of integrin α10 expression (G) (N) Semiquantitative RT-PCR analysis of integrin expression in incisor PDL. (O to R) Immunohistochemical localization of collagen type III (O and P) and collagen type XII (Q and R) (S and T) Blood vessels in PDL from 2-month-old animals were analyzed using antibodies for PECAM-1 (CD31). The area of the PDL is marked with brackets. ab, alveolar bone; T, tooth; M, molar. The bar corresponds to 100 μm. (U) Semiquantitative RT-PCR of genes expressed in control (c) and mutant (mut) PDL isolated from 4-month-old mice.

The expanded PDL tissue expressed collagen III and collagen XII throughout (Fig. 7O to R).

Since disturbed blood vessel formation during wound healing has been reported in other animal models lacking collagen-binding integrins, we also analyzed CD31-positive cells in the α11^−/− PDL but found no appreciable differences relative to the control mice (Fig. 7S and T). Staining for macrophages failed to reveal any substantial amounts of these cells in either the control or mutant PDL (data not shown). Due to the denaturing conditions in the decalcification protocols, we could not assay MMP activity by in situ methods.

PCR analysis indicated unchanged RNA levels of the PDL marker periostin and of the matrix metalloproteinase MMP2. Interestingly, an increase in the mRNA levels of MMP9 and a decrease in MMP14 mRNA levels were consistently observed in the integrin α11-deficient incisor PDL tissue (Fig. 7U).

DISCUSSION

The α11 integrin chain is most closely related to the α10 integrin chain, and both subunits display restricted expression patterns, being confined embryonically to different interstitial connective tissues. Comparison with the expression patterns of other collagen-binding integrins suggests that the expression of α11 partially overlaps with α2 and is complementary to that of α10 in many locations (38).

Careful analyses of the phenotypes of mice deficient in individual collagen-binding integrins have revealed relatively mild phenotypes in unchallenged mutant mice (8, 11, 18, 25). The unchallenged α11-deficient mice described in this report have an incisor phenotype causing malnutrition and increased mortality.

The relatively mild phenotypes of the individual collagen receptor mutant mice suggest a fine-tuning role for these receptors in cell-collagen adhesive events under normal physiological conditions. Alternatively, these phenotypes may reflect overlapping roles for individual integrin receptors. Circumstantial evidence does indeed suggest that integrins have redundant functions in certain tissues (2, 20, 44). Future studies aimed at inactivating multiple collagen-binding integrins should resolve whether they have overlapping functions. The use of α1 and α2 integrin-deficient mice in disease models has revealed important functions for these integrins in angiogenesis (39), fibrosis (14), inflammation (15), and mast cell activation (16). Little is known about the behavior of mice that are deficient in α10 and α11 integrin chains in disease models.

α11 mice fed on a regular laboratory chow are dwarfed, display increased mortality, and suffer from malnutrition. Despite the abundant expression of α11 in perichondrial cells, α11 expression did not correlate with any cartilage growth defects. Analysis of other tissues in which α11 is normally expressed, such as the cornea, intervertebral disc, intestine, and skin, did not reveal any obvious phenotypes in the corresponding unchallenged α11^−/− tissues (data not shown). Instead, the malnutrition and smaller size observed in α11-deficient mice appear to correlate with the tooth phenotype. Upon supplementing the diet with soft food, the mice no longer suffered from malnutrition and had a normal life span (data not shown). Interestingly, the mutant mice fed on soft food were still smaller, indicating that growth defects must exist in addition to the tooth phenotype. The smaller size of the mice at the time of weaning also supports the notion of a tooth-independent growth defect. Unlike the situation observed recently in periostin-deficient mice (43), the major tooth phenotype in the α11^−/− mice persisted during supplementation of the diet with soft food (data not shown).

The phenotype observed in the α11-deficent mice is highly selective for the incisors, which, unlike the molars, erupt continuously in rodents. Where α11 appears to be expressed in the follicle mesenchyme of both molars and incisors, in the adult dentition it is mainly expressed in the incisors, which most likely explains the restriction of the tooth phenotype to the incisors. Since molar eruption occurs in part by a different mechanism, it is unclear what role integrins and collagen receptors might play in it. Our analysis of the collagen-binding integrin expression patterns suggests that receptors other than collagen-binding integrins may mediate the attachment of molar PDL cells to the collagen matrix.

A number of morphogenic differences exist between molars and incisors, including the types of transcription factors expressed during embryonic development, the growth factors mediating paracrine cellular communication, and the types of MMPs needed at different stages in the tooth eruption process. Epidermal growth factor affects only incisor eruption, whereas CSF-1 preferentially affects the osteoclast-dependent stage of molar eruption (12). FGF-10 supports the stem cell compartment required for growth of the incisors, and incisors, but not molars, are missing in its absence (24). Finally, a lack of MT1-MMP (MMP14) has been reported to affect root development and eruption of molars (3, 7). The list of molecular mechanisms that differ between molars and incisors now also includes a matrix receptor, integrin α11β1.

The PDL has been shown to play a central role during rodent incisor eruption, but the exact nature of this role has been controversial. According to one school, periodontal ligament fibroblasts migrate occlusally through the PDL space and create the tractional force that pulls the tooth towards the surface of the oral mucosa (6), while another school maintains that this eruptive force is provided by the hydrostatic tissue pressure within the vascular tissue of the periodontal ligament (9). Many existing mutations described for the tooth affect the neural crest-derived cells and disturb tooth morphogenesis as a result, but relatively few of these selectively affect tooth eruption.

Until now there has been a lack of good genetic models to study tooth eruption. We believe that the integrin α11-deficient mouse now offers one such model. In vitro analysis of α11^−/− cells is a powerful tool for corroborating the in vivo α11^−/− PDL phenotype. The lack of the α11 integrin chain in fibroblasts was seen to reduce cell attachment to collagen, reduce cell spreading on collagen I, and reduce cell-mediated collagen reorganization in a three-dimensional collagen I matrix in vitro. The major reduction in cell attachment and cell spreading on collagen I of α11^−/− cells agrees well with the observations that α1^−/− (18) or α2^−/− (53) fibroblasts display only slightly reduced interactions with collagen I and shows that α11β1 is the major receptor for collagen I on MEFs. It will be important to determine the relative importance of α11β1 for cell adhesion to fibrillar collagens on other types of fibroblasts. We have previously demonstrated that inactivation of α11β1 in MEFs expressing α1β1, α2β1, and α11β1 increases cell migration on collagen I (38), presumably due to a lower strength of binding to collagen, offering less of a restraint on migration but also leading to less tractional force generation. In the absence of α11β1, PDL cells will bind PDL collagen with lower affinity. We predict that this changed interaction with collagen fibrils disturbs the tractional forces being developed, in turn contributing to the disturbed tissue homeostasis.

The mutant incisor periodontal ligament lacking in α11 is characterized by a distinctly increased thickness, the presence of a wider acellular cementum, and an increased number of rests of Mallasez. The increased width of the cementum layer is most likely related to decreased eruption rate (4). Maintaining the correct width of the PDL is an essential function of the periodontal fibroblast, and this regulation is obviously lost in the absence of periodontal ligament fibroblast α11β1. The present data thus identify α11β1 in periodontal ligament fibroblasts as a central regulator of PDL width. As judged by sirius red staining, collagen accumulates in the mutant PDL. Integrins are known to regulate both collagen and MMP synthesis (40, 42), and we attempted to determine whether increased synthesis was the mechanism behind this collagen accumulation, but quantification by biochemical methods was not possible due to the small amount of tissue. The numbers of PDL fibroblasts that can be isolated from the mouse incisor PDL are too low to allow analysis of collagen in primary cells. Immortalization of this cell population from wt and ko animals could possibly allow such an analysis in the future. Quantitative PCR analysis of RNA from PDL tissue and MEFs cultured under different conditions failed to detect any changes in the RNA levels of collagen I and III (data not shown). However, increased collagen accumulation could be due to a number of mechanisms, including increased RNA stability and posttranslational mechanisms. In the case of the MMPs, we would predict that MMP activity is reduced in the knockout mouse. However, since the phenotype is manifested only postnatally, when ossification has occurred, sectioning of the PDL tissue involves a lengthy decalcification scheme, preventing reliable in situ zymography to detect changes in MMP activity. Interestingly, analysis of MMPs from MEFs cultured inside a collagen gel revealed reduced induction of MMP13 and MMP14 RNA in the α11^−/− cells. Similar analysis of PDL tissue showed increased levels of MMP9 RNA and reduced levels of MMP14 RNA in the mutant PDL tissue. In summary, the exact mechanisms whereby α11 deficiency leads to a change in collagen turnover remain to be clarified, but they appear to involve changes in MMP levels, leading to altered matrix turnover in the absence of integrin α11.

In summary, our data support a role for α11β1 in tooth eruption. We predict that α11β1 is needed on the surface of incisor fibroblasts to generate the tractional force needed for tooth eruption. Consistent with this model, the reduced capacity of PDL fibroblasts to adhere to and remodel the collagen matrix in vitro fits well with the phenotype of reduced tooth eruption observed in vivo.