|Home | About | Journals | Submit | Contact Us | Français|
Podosome-type adhesions are actin based membrane protrusions involved in cell-matrix adhesion and extracellular matrix degradation. Despite growing knowledge of many proteins associated with podosome-type adhesions, much remains unknown concerning the function of podosomal proteins at the level of the whole animal. In this study, the spontaneous mouse mutant nee was used to identify a component of podosome-type adhesions that is essential for normal postnatal growth and development. Mice homozygous for the nee allele exhibited runted growth, craniofacial and skeletal abnormalities, ocular anterior segment dysgenesis, and hearing impairment. Adults also exhibited infertility and a form of lipodystrophy. Using genetic mapping and DNA sequencing, the cause of nee phenotypes was identified as a 1 bp deletion within the Sh3pxd2b gene on mouse Chromosome 11. Whereas the wild-type Sh3pxd2b gene is predicted to encode a protein with 1 PX domain and 4 SH3 domains, the nee mutation is predicted to cause a frameshift and a protein truncation altering a portion of the third SH3 domain and deleting all of the fourth SH3 domain. The SH3PXD2B protein is believed to be an important component of podosomes likely to mediate protein-protein interactions with membrane spanning metalloproteinases. Testing this directly, SH3PXD2B localized to podosomes in constitutively active Src transfected fibroblasts and through its last SH3 domain associated with a transmembrane member of a disintegrin and metalloproteinase family of proteins, ADAM15. These results identify SH3PXD2B as a podosomal-adaptor protein required for postnatal growth and development, particularly within physiologic contexts involving extracellular matrix regulation.
Podosome-type adhesions (podosomes and invadopodia) are specialized adhesion structures involved in cell matrix adhesion and extracellular matrix (ECM) remodeling. Structurally, podosome-type adhesions are characterized by an actin dense core surrounded by a ring of additional proteins that drive focal membrane protrusions into surrounding tissues. Functionally, adhesion molecules and matrix metalloproteinases are enriched in podosome-type adhesions, allowing these finger-like protrusions to interact with and modify surrounding ECM. These features are thought to allow podosome-type adhesions to play important physiologic roles in several cell types requiring motility or ECM degradation (Linder 2007; Linder and Aepfelbacher 2003; Spinardi and Marchisio 2006; Weaver 2008).
Many protein components of the podosome-type adhesion machinery have been identified. These include actin and actin modulating proteins that form the core structure, cell adhesion receptors linking cells to ECM, proteinases mediating ECM remodeling, protein kinase transducing signals, and membrane trafficking proteins facilitating protein recruitment (Gimona and Buccione 2006; Gimona et al. 2008; Linder 2007). Amongst these proteins, two key regulators are SRC and SH3PXD2A (commonly referred to as TKS5). The tyrosinase kinase SRC is a key molecule for inducing podosome-type adhesions (Destaing et al. 2008; Gimona et al. 2008). SH3PXD2A is a substrate of SRC (Lock et al. 1998) and has been described as a major organizer of podosome-type adhesions by multiple groups (Courtneidge et al. 2005; Oikawa et al. 2008; Symons 2008; Thompson et al. 2008). Upon SRC activation SH3PXD2A is recruited to the plasma membrane through interactions with PtdIns(3,4)P2, which is enriched in podosomes (Abram et al. 2003; Oikawa et al. 2008). Recruitment of SH3PXD2A subsequently recruits other molecules such as WAS and WASL (commonly referred to as WASp and N-WASp, respectively), which stimulate ARP 2/3 complex-mediated actin nucleation and podosome extension (Oikawa et al. 2008). Recently, a homolog of SH3PXD2A, SH3PXD2B, has been identified and shown to also influence podosomes of cells in culture (Buschman et al. 2009; Courtneidge et al. 2005). Although identification of SH3PXD2A and SH3PXD2B in cells in culture suggests that these adaptor proteins are likely to play determinative roles influencing behavior of podosome-type adhesions, much remains unknown concerning the function of these proteins within the physiological context of the whole animal.
Here, we present identification and mutant phenotypes associated with a unique spontaneously arising allele of Sh3pxd2b in mice. Mice homozygous for the nee mutation exhibit runted growth and striking developmental abnormalities of several mesenchymal derived tissues, including craniofacial structures, the ocular irideocorneal angle, and white adipose tissue. Through genetic crosses designed to identify the basis of the nee mutation, we demonstrate a 1 bp deletion in the Sh3pxd2b gene. Based on sequence homology, the Sh3pxd2b gene is predicted to encode an adaptor protein with homology to SH3PXD2A, a key regulator of podosome-type adhesions. Testing the hypothesis that the SH3PXD2B protein is indeed a podosomal-adaptor protein, we demonstrate here that SH3PXD2B also localizes to podosomes, and as expected for an adaptor protein, is capable of binding transmembrane metalloproteinases such as ADAM15. Combined, these data support a role for SH3PXD2B as a podosomal-adaptor protein and implicate podosomes as important biological regulators of mammalian postnatal growth and development.
All mice were obtained from The Jackson Laboratory, Bar Harbor, Maine. Mice were subsequently housed and bred at the University of Iowa Research Animal Facility. The spontaneous nee mutation arose in strain B10.A-H2h4/(4R)SgDvEg at generation F51 and subsequently was maintained by the Mouse Mutant Resource of The Jackson Laboratory (http://mousemutant.jax.org/index.html). Since both male and female homozygotes are infertile, the strain was maintained by progeny testing or ovarian transplantation. In the past, the mutation has also been referred to as nm2702 (Chang et al. 2005). With advice of The Mouse Genomic Nomenclature Committee, the mutation will here after be referred to as nee to indicate phenotypic defects associated with the mutation (nose, eye, ear). Upon molecular identification of the mutation, mice were genotyped by assaying the absence or presence of an RsaI restriction enzyme site abolished by the nee mutation and the stock maintained in heterozygote X heterozygote crosses. Mice were maintained on a 4% fat NIH 31 diet provided ad libitum and were housed in cages containing dry bedding (Cellu-dri; Shepherd Specialty Papers, Kalamazoo, MI). The environment was kept at 21°C with a 12-hour light: 12-hour dark cycle. All animals were treated in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. All experimental protocols were approved by the Animal Care and Use Committee of The University of Iowa or The Jackson Laboratory.
Craniofacial measurements were made for mutants and controls on skulls prepared by incomplete maceration in potassium hydroxide, stained with alizarin red, and stored in undiluted glycerin (Green 1952). A set of 7 standard skull measurements were collected using digital hand calipers (Stoelting, Wood Dale, IL). Landmarks were adapted from those used by Dr. Joan Richtsmeier to characterize craniofacial morphology in mouse models of Down Syndrome (Richtsmeier et al. 2000) and have been validated for accuracy and precision by the Craniofacial Resource at The Jackson Laboratory (http://craniofacial.jax.org/).
Anterior chamber phenotypes were assessed with a slit-lamp (SL-D7; Topcon, Tokyo, Japan) and photodocumented with a digital camera (D100; Nikon, Tokyo, Japan). Images were taken with identical camera settings and prepared with identical image software processing.
Imaging of adipose tissue by magnetic resonance imaging (MRI) was performed on a 4.7 Tesla Varian Unity/INOVA small-bore MRI system (Varian Inc., Palo Alto, CA). After anesthesia with isoflurane, T1-weighted images were acquired in the coronal plane using a fast spin-echo pulse sequence with TR/TE = 600/12 ms, an echo train length of 2, and 4 averages. A matrix size of 512 × 128 covered a field of view of 60–90 mm × 30 mm with 15 slices of 1.1–1.5 mm thickness separated by a 0.2–0.5 mm gap depending on the size of the animal. Imaging of skeletal morphology by X-ray was performed with euthanized mice and a Faxitron MX20 Specimen X-ray system (Faxitron X-Ray LLC, Lincolnshire, IL) using 25 kV 1 min exposures on Kodak Portal Pack Oncology X-ray film.
Auditory-evoked brain stem response (ABR) threshold analysis was performed using previously described methods (Zheng et al. 1999). Briefly, mice anesthetized with tribromoethanol were presented with broad-band click and pure tone 8-, 16-, and 32-kHz stimuli at intensities varying between 10 and 100 dB SPL (decibels sound pressure level), and the evoked brain stem responses were recorded. Auditory thresholds were obtained for each stimulus by varying the SPL at 5-dB steps to identify the lowest level at which an ABR pattern can be recognized. Inner and middle ears were dissected from the skull following perfusion with Bouin’s fixative. Tissues were decalcified in Bouin’s fixative for 14 days and embedded in paraffin. Cross sections were cut at 7 µm, mounted on glass slides, and stained with hematoxylin/eosin.
Areal bone mineral density of the whole body and percent fat body mass were determined on mice anesthetized by intraperitoneal injection with Avertin (tribromoethanol stabilized in tertiary amyl hydrate), 1 mg per 2 g of body weight. Data were collected using a PIXImus small animal dual-energy X-ray absorptiometry (DEXA) densitometer (GE Healthcare, Waukesha, WI), reconfigured with lower X-ray energy than in human DEXA machines in order to achieve contrast in small specimens, and utilizing manufacturer supplied software (Lunar PIXImus version 1.46). Resolution of the PIXImus is 0.18 × 0.18 mm pixels with a usable scanning area of 80 × 65 mm. Calibration and daily quality assurance was achieved using a phantom with known values.
To determine the chromosomal location of the nee mutation, B10.A-H2h4/(4R)SgDvEg-nee mice were mated with CAST/EiJ mice and the F1 mice, which exhibited no visible abnormalities, were intercrossed to produce F2 mice. Genomic DNA of 45 affected F2 progeny was prepared as previously described (Buffone and Darlington 1985) and genotyped by analyzing polymorphic microsatellite markers by agarose electrophoresis, utilizing standard conditions and protocols. The initial genome scan was carried out on pooled DNA samples. After detection of linkage on Chr 11, microsatellite markers D11Mit295, D11Mit19, D11Mit152, D11Mit215 and D11Mit186 were scored on individual DNA samples. Candidate genes localized to the critical region between markers D11Mit19 and D11Mit186 were identified by scanning the Ensembl genome database (Release 53). Following PCR amplification of exons of interest from genomic DNA isolated from either the mutant B10.A-H2h4/(4R)SgDvEg-nee or wild-type parental B10.A-H2h4/(4R)SgDvEg strain, products were purified (QIAquick PCR Purification Kit; Qiagen, Inc., Valencia, CA), sequenced by automated fluorescence tagged sequencing, and directly compared to each other. Confirmation of the nee mutation was performed by PCR amplification of a 713 bp DNA fragment flanking the deleted adenine residue (using primers: forward: 5' AGGCTCAGTTGCCCTGAATGTA 3'; reverse: 5' TTCTCAGCGGGAACTTGCTCTT 3') from various strains of mice and digestion of the amplicons with RsaI. DNA containing the nee allele is predicted to be resistant to cleavage, while wild-type alleles are predicted to give rise to products of 546 bp and 167 bp. Numeration for the Sh3pxd2b mRNA is based on NCBI RefSeq NM_177364.3.
Constructs generating GST-fused SH3PXD2B were generated by PCR-assisted cloning. To reduce the possibility for cross reaction with the closely related SH3PXD2A protein, the antibody was raised against a region between the third and fourth SH3 domains of SH3PXD2B having low identity with SH3PXD2A. Using oligonucleotides specific to a portion of Sh3pxd2b exon 13 (encoding amino acids 440–751) and genomic DNA of C57BL/6J mice as template, the resulting amplification product was first cloned into a TA cloning vector (pCR2.1-TOPO vector; Invitrogen, Carlsbad, CA) and then subcloned into a GST gene fusion vector (pGEX6p-1; GE Healthcare, Piscataway, NJ). Correct fusion of this construct was confirmed by automated fluorescence tagged sequencing. Following transformation into the BL21 (DE3) strain of Escherichia coli, protein expression was induced and cell lysates collected. GST-fused protein was first purified using a glutathione-Sepharose 4B column followed by excision of GST from the protein (PreScission protease; GE Healthcare, Piscataway, NJ). SH3PXD2B-specific antiserum was generated in rabbits and affinity purified (Sigma-Genosys; The Woodlands, TX).
Plasmids encoding GFP-fused full-length SH3PXD2B (GFP-SH3PXD2B) were constructed by PCR-assisted cloning. DNA fragments containing wild-type Sh3pxd2b were amplified by RT-PCR from total RNA extracted from eyes of C57BL/6J mice. The resulting fragments were first cloned into an entry vector (pCR8/GW/TOPO; Invitrogen, Carlsbad, CA) and then subcloned by LR recombination into an expression vector containing an N-terminus GFP tag (pcDNA-DEST53; Invitrogen, Carlsbad, CA). Plasmids encoding GFP-fused mutant SH3PXD2B with the 1303A deletion (GFP-SH3PXD2B1303delA) were constructed by PCR based site-directed mutagenesis using the wild type construct as a template. Sequences of both plasmid inserts were confirmed by automated fluorescence tagged sequencing.
Cells were maintained at 37°C in a humidified atmosphere containing 5% CO2. HEK 293T and mouse NIH3T3 cells were cultured in DMEM supplemented with 10% FBS, 100 units/ml penicillin, and 100 units/ml streptomycin. Formation of podosomes was induced by transient transfection of a previously described constitutively active Src (Y529F) construct (Sharma et al. 2007). Transfections utilized the Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA) with additional use of the PLUS Reagent enhancing agent (Invitrogen, Carlsbad, CA) in NIH3T3 cells, following the manufacturer’s recommended protocols.
Cells cultured on coverslips were fixed in 4% paraformaldehyde in PBS for 10 min and incubated in permeabilization buffer (0.1M glycine, 0.05% Triton X-100 in PBS) containing 3% BSA for 30 min. Following permeabilization, cells were blocked with permeabilization buffer containing 15% goat serum and 3% BSA for 15 min. Fixed cells were incubated with anti-SH3PXD2B antibody for 1 hr (1:200 dilution in permeabilization buffer with 1% BSA and 1% goat serum), washed with PBS, and incubated with a 1:200 dilution of secondary antibody and 1:100 dilution of phalloidin (Alexa633-conjugated goat anti-rabbit and Alexa546-phalloidin; Invitrogen, Carlsbad, CA). After mounting (ProlongGold; Invitrogen, Carlsbad, CA), cells were imaged with a confocal microscope (Zeiss LCM 510; Carl Zeiss MicroImaging, Inc., Thornwood, NY). Cells transfected with GFP-fusion plasmids were fixed with 4% paraformaldehyde in PBS for 10 min; following washes, cells were mounted and imaged as described above.
GST and a GST-fusion containing the fourth SH3 domain spanning amino acids 842–908 of SH3PXD2B (GST-SH3#4) were expressed in BL21 (DE3) cells and purified using a glutathione-Sepharose 4B column (GE Healthcare, Piscataway, NJ). Cell lysates (50 µg) were incubated with glutathione-Sepharose 4B beads for 2 h at 4°C, washed, and then blocked in EB buffer (10 mM Tris-HCl pH 7.4, 50 mM NaCl, 1% Triton X-100, 50 mM NaF, 0.1% BSA, 2 mM Na3VO4) supplemented with an additional 5% BSA at 4°C overnight. HEK 293T cells expressing MYC-tagged full length ADAM15 were generated using PCR to amplify the Adam15 cDNA followed by cloning with gateway technology into an expression vector (pcDNA-DEST53; Invitrogen, Carlsbad, CA). 293T cells transfected with Adam15-myc plasmid were lysed 48 h after transfection in EB buffer supplemented with 1 mM PMSF and a mix of additional protease inhibitors (Complete Mini Protease Inhibitor Cocktail; Roche Applied Science, Indianapolis, IN). Cell debris was removed by centrifuging at 12,000 rpm for 10 min. GST-fusion coated beads were washed 3 times with EB buffer and incubated with 1 mg total cell lysates for 2 hours at 4°C. Following 6 washes with EB buffer, beads were recovered by centrifugation and dissolved in Laemmli buffer. Bound protein was analyzed by 7.5% SDS-PAGE and immunoblotting. Blots were blocked with 10% BSA in TBST (Tris-buffered saline with 0.1% Tween-20) followed by antibody incubations and washes in TBST using 1:1000 dilution of anti-c-MYC rabbit polyclonal antibody (Novus Biologicals, Littleton, CO) and 1:4000 dilution of horseradish peroxidase conjugated goat anti-rabbit IgG (Invitrogen, Carlsbad, CA). Immunoreactivity was detected by enhanced chemiluminescence methodology (GE Healthcare, Piscataway, NJ).
Protein levels of SH3PXD2B in various tissues were analyzed with western blot. Tissues dissected from a mouse from the nee colony homozygous for wild-type Sh3pxd2b alleles were minced and homogenized with a rotor/stator tissue homogenizer in cell lysis buffer (50mM Tris-HCl pH 7.4, 150mM NaCl, 1mM EDTA, 0.1% Triton X-100, 0.1% SDS) supplemented with 1 mM PMSF and a mix of additional protease inhibitors (Halt Proteinase Inhibitor Cocktail; Thermo Scientific, Rockford, IL). Total tissue lysates were centrifuged at 14,000 rpm for 20 min at 4°C. Protein concentration in the resulting supernatant was determined using the BCA protein assay reagent (Sigma-Aldrich, St Louis, MO). Using 50 µg of lysate per lane, protein from each tissue was electrophoresed on 7.5% SDS-PAGE gels and transferred to membranes (Immobilon-FL membrane; Millipore, Bedford, MA). Membranes were blocked for 1 h at room temperature in Odyssey Blocking Buffer (LI-COR Biosciences, Lincoln, NE) diluted 1:1 in PBS. All antibody incubations were performed using a 1:1 solution of Odyssey Blocking Buffer and PBS supplemented with 0.2% Tween-20. Primary antibody was incubated with membranes at 4°C overnight using a 1:2000 dilution of anti-SH3PXD2B rabbit polyclonal antibody. As a control for protein transfer, membranes were similarly incubated with 1:1000 dilution of anti-β-actin rabbit polyclonal antibody (Sigma-Aldrich, St Louis, MO). After washes with PBS, membranes were incubated for 1 h at room temperature with secondary antibody diluted 1:2000 (IRDye680 goat anti-rabbit IgG; LI-COR Biosciences, Lincoln, NE) and visualized with an infrared scanner (Odyssey Infrared Imaging System; LI-COR Biosciences, Lincoln, NE).
Following a phenotype-driven approach for identifying new mouse models of human disease, the nee mutation is a spontaneously arising mutation that was initially identified by the Mouse Mutant Resource of The Jackson Laboratory. Mice homozygous for the nee mutation were initially recognized because of their small body size, abnormal craniofacial morphology, and ocular defects (Fig. 1).
The most obvious defect of nee homozygotes was their small body size (Fig. 1a, b). At birth, nee homozygotes were visibly indistinguishable from wild-type mice. By 2 months of age, nee homozygotes were dramatically smaller in body weight (10.2 ± 1.5 g; mean ± SD; n = 4) compared to either heterozygote (18.8 ± 0.7 g; n = 9) or wild-type (18.0 ± 1.9 g; n = 5) littermates (P < 0.004 in both comparisons, Student’s t-test). Subsequent to 2 months of age, while heterozygous and wild-type mice both continued to grow in body weight, mice homozygous for the nee mutation did not. As heterozygous and wild-type mice reached normal adult weights by 6–12 months of age (28.3 ± 5.1 g; n = 11 and 28.0 ± 7.9 g; n = 5; respectively), nee homozygotes remained runted at both the 3–5 month (10.9 ± 1.9 g; n = 8) and 6–12 month (11.7 ± 2.7 g; n = 7) age periods. As a consequence, the difference in body weight among adult mice was highly significant (P < 0.0007 comparing homozygotes to heterozygotes or wild-type littermates, 6–12 month old mice, Student’s t-test).
Two additional externally visible phenotypes were initially appreciated in nee homozygotes. All nee homozygotes have craniofacial and ocular abnormalities. Compared to control mice, all nee homozygotes had shortened noses (nose length: skull length = 0.629; heterozygous littermates = 0.687) and domed skulls (skull length: skull width = 1.750; heterozygous littermates = 2.141; n = 12 mice per group; P < 0.0001; Student’s t-test) (Fig. 1c, d). Also, eyes of all nee homozygotes had abnormalities (Fig. 1e – h), typically appearing bulging with white corneal opacities. Examined by ophthalmic slit-lamp, corneal opacities of varying severity were confirmed in all eyes of nee homozygotes (n = 14 mice, 1–8 months in age), whereas corneal opacities were rare in eyes of age-matched controls (a single corneal opacity was observed from all mice examined; n = 12 heterozygous littermate controls, 11 wild-type littermate controls). Where unobstructed views of the anterior chamber were possible in nee homozygotes, peripheral anterior synechia were always observed and the anterior chamber depth was increased. Combined, these clinical observations indicated that nee homozygotes have a developmental anterior segment dysgenesis.
In addition to the externally visible phenotypes of nee homozygotes described above, several other phenotypes were also discovered. First, both male and female homozygotes were infertile. Fertility could be restored by ovarian transplants of ovaries from homozygotes to wild-type mice and mating to heterozygous males, indicating that the reason for female infertility did not reside in the ovary but elsewhere in the female reproductive tract. Second, nee homozygotes had multiple skeletal abnormalities (Fig. 2). Examination of skeletal structures by radiographic analysis indicated that nee homozygotes have a proportionally smaller skeleton (Fig. 2a) and abnormalities of the skull (Fig. 2b, c). Radiography of the skull from nee homozygotes again indicated a reduced length along the anterior-posterior axis. Higher magnification images of the hind leg (Fig. 2d, e) showed reduced dimensions, but otherwise normal gross anatomy. Both the skull and hind leg of nee homozygotes exhibited decreased radio-opacity, suggesting a potential decrease in bone density. To test this quantitatively, DEXA analyses were performed. In comparison to sex-matched heterozygous littermates, 12-week-old nee homozygotes exhibited reduced areal bone mineral density (aBMD). This difference was present in both males (nee homozygotes, 0.035 g/cm2; heterozygous littermates, 0.048 g/cm2) and females (nee homozygotes 0.031 g/cm2; heterozygous littermates, 0.049 g/cm2) and was statistically significant (n = 6 mice per sex per genotype; P < 0.0001; Student’s t-test). Female homozygotes had significantly lower aBMD than male homozygotes (P < 0.001; Student’s t-test); there was no sex difference in heterozygotes. Third, mice homozygous for nee have hearing impairment and middle ear pathology (Fig. 3). ABR measurements revealed that nee mice have elevated thresholds to broad-band click and 8 kHz, 16 kHz, and 32 kHz pure tone stimuli (Fig. 3a). Hearing impairment was neither progressive nor frequency dependent, as elevated ABR thresholds (40–50 dB above normal) were found in all four stimuli at all ages tested (n = 13 nee homozygotes and 3 heterozygous control mice; 34–92 days of age). Histological analysis further revealed that hearing impairment in these mice was likely due to middle ear inflammation (otitis media, Fig. 3b, c). Serous fluid with diffuse neutrophils was found in the middle ear cavities of all nee homozygotes examined, but not in control mice (n = 6 nee homozygotes and 3 heterozygous control mice), and the surrounding epithelium was markedly thickened by fibrous connective tissue and embedded neutrophils. No other structural abnormalities were observed. Similar results have previously been observed; mice with craniofacial abnormalities that include middle ear cavity or Eustachian tube dysmorphologies often develop otitis media (Zheng et al. 2006). Together, these results indicate that nee-mediated hearing deficiency is likely a secondary consequence of the craniofacial abnormalities of nee homozygotes, which presumably render the middle ear sensitive to infection. Finally, nee homozygotes developed a form of lipodystrophy (Fig. 4). Initially noted in dissections (Fig. 4a, b), and subsequently confirmed with MRI (Fig. 4c – h), adult nee homozygotes had severely depleted visceral and subcutaneous white adipose tissue (n = 3 mice examined, 6–11 months in age). In contrast, residual interscapular brown adipose tissue was still present. Adipose defects were less striking in young mice. While adipose defects were not observable by MRI in young mice, DEXA analyses showed a subtle reduction in percent body fat in 12-week-old nee homozygotes compared to heterozygous littermate controls (14% vs 19% respectively, n = 20 mice per group; P < 0.001; Student’s t-test). With the exception of aBMD, no other sex-specific differences were observed in any phenotypes of nee mice. Aside from the above mentioned defects, nee homozygotes were otherwise relatively healthy, with many mice living to at least 12 months of age.
In order to identify the molecular basis of the nee mutation, an intersubspecific intercross between CAST/EiJ and nee mice was performed (Fig. 5). Whereas on the B10.A-H2h4/(4R)SgDvEg genetic background, the nee mutation was fully penetrant and close to the expected 25% of affected progeny were produced in heterozygous matings (60 mutants / 300 total mice born = 20%), heterozygotes from the intercross with CAST/EiJ produced lower than the expected number of mutants in the F2 generation (27 mutants / 224 total mice born = 12%) and of the mice born, 5% of homozygotes died before weaning. Linkage analysis revealed that nee segregates as a simple recessive factor. Haplotype analysis based on 45 F2 affected mice mapped the mutation to a region on Chr 11 (Fig. 5a). The minimal critical region defining the nee interval was delimited proximally by marker D11Mit19 and distally by marker D11Mit186, spanning a 9.7 Mb region.
According to the Ensembl genome database, the nee critical region included a total of 65 protein coding genes encompassed by 718 exons. A sequence-driven approach was utilized to identify the mutation. Based on previously described phenotypes of targeted mutations in mice, several genes in the interval were initially considered reasonable candidates, including Fgf18 and Npm1 (Grisendi et al. 2005; Ohbayashi et al. 2002). However, sequencing of all known exons from these candidates failed to detect any changes. In continuing analysis of the region, a total of 225 different exons divided among multiple genes were analyzed. From these, the nee mutation was identified as a 1 bp deletion in the last exon of Sh3pxd2b at nucleotide position 1303 of the predicted mRNA (Fig. 5b). The 1303delA change was the only change detected and was confirmed by independent amplification and sequencing of two additional affected mice (data not shown).
To stringently confirm the status of the 1303delA alteration as a unique mutation, a restriction enzyme based assay distinguishing this allele was designed. The 1303delA deletion disrupts an RsaI restriction site. Therefore, DNA containing the wild-type allele should be sensitive to RsaI digestion while nee DNA should be resistant. Using this assay, 10 phenotypically normal inbred mouse strains were examined for the potential presence of the nee allele (Fig. 5c). Resistance to RsaI digestion was present only in nee DNA and not in DNA from other phenotypically normal strains, confirming that the 1 bp deletion in Sh3pxd2b was unique and highly likely to be the mutation of nee mice.
The Sh3pxd2b gene is predicted to encode a 908 amino acid protein with an N-terminus PX domain and four SH3 domains (Fig. 5d). The 1303delA deletion is predicted to cause a frameshift altering 37 amino acids before causing a premature stop codon and protein truncation altering a portion of the third SH3 domain and completely deleting the fourth SH3 domain (Fig. 5d). To investigate the potential influence of the1303delA deletion on localization of the SH3PXD2B protein, GFP fusion constructs encoding the wild-type (GFP-SH3PXD2B) and mutant proteins (GFP-SH3PXD2B1303delA) were transfected into HEK 293T cells (Fig. 6). As previously observed by others (Hishida et al. 2008), the absence or presence of SH3 domains in SH3PXD2B influenced localization. Whereas wild-type SH3PXD2B was restricted to the cytoplasm (Fig. 6a), the mutant protein was present throughout the cell, including prominent localization within the nucleus (Fig. 6b). Thus, the Sh3pxd2bnee mutation is predicted to encode a truncated protein, which as a consequence of the mutation, likely has altered subcellular distribution.
The SH3PXD2B protein, and the closely-related SH3PXD2A, have both been demonstrated in cell culture studies to function as adaptor proteins required for formation of podosomes (Abram et al. 2003; Buschman et al. 2009; Courtneidge et al. 2005). The PX domains of both SH3PXD2B and SH3PXD2A bind to the lipids PtdIns(3)P and PtdIns(3,4)P2 while their SH3 domains bind and influence various matrix metalloproteinases and ADAM family proteinases (Abram et al. 2003; Buschman et al. 2009). Through these interactions, SH3PXD2A and SH3PXD2B are thought to tether and regulate the localization of ECM modifying enzymes to precise cellular compartments such as podosomes. To further test how the Sh3pxd2bnee mutation might influence podosomes important to murine postnatal growth and development, we next tested whether SH3PXD2B localizes to podosomes and whether the deleted SH3 domain was capable of mediating a protein-protein interaction with a candidate protein, ADAM15 (Fig. 7).
Properties of podosomes have been well described in Src transformed fibroblasts (Tarone et al. 1985). To test whether SH3PXD2B localizes to podosomes in these cells, NIH3T3 cells were transfected with a constitutively active form of Src (Y529F) and localization of endogenous SH3PXD2B tested with an anti-SH3PXD2B specific antibody (Fig. 7a – c). While SH3PXD2B localized to the cytoplasm in untransfected cells (data not shown), a portion of SH3PXD2B localized to podosome clusters and rosettes in Src (Y529F) transfected fibroblasts. Consistent with other recent findings (Buschman et al. 2009), this result independently confirms that SH3PXD2B localizes to podosomes.
Although direct protein binding partners for SH3PXD2B have not previously been described, several interactions for the closely related SH3PXD2A protein are known. It has previously been demonstrated by in vitro binding assays that SH3PXD2A can form protein-protein interactions with ADAM family proteins, including ADAM12, ADAM15, and ADAM19. These interactions are mediated by multiple PxxP motifs present within the cytoplasmic tail of ADAMs and the fifth SH3 domain of SH3PXD2A (Abram et al. 2003). Because the fifth SH3 domain of SH3PXD2A has homology to the fourth SH3 domain of SH3PXD2B (55% identity), we reasoned that the proteins may have similar binding partners. To test this, a GST pull-down experiment was performed with a candidate of interest, ADAM15 (Fig. 7d). A fusion protein containing the fourth SH3 domain of SH3PXD2B was incubated with lysates of 293T cells transfected with a MYC-tagged full-length ADAM15. While negative controls containing GST alone did not interact with ADAM15, the fusion protein containing the fourth SH3 domain of SH3PXD2B did. This result is consistent with a role of SH3PXD2B as an adaptor protein. Furthermore, because the mutant Sh3pxd2bnee allele is predicted to encode a protein in which all of this fourth SH3 domain is absent, this result suggests that misregulation of ADAM15 likely contributes to mutant phenotypes of nee mice.
In order to further relate functions of SH3PXD2B to the mutant phenotypes of nee mice, western blot analysis was utilized to determine the tissue distribution and levels of SH3PXD2B in wild-type mice (Fig. 7e). A band of 120 kDa was detected in protein lysates isolated from tissues of adult wild-type mice. Presumably a consequence of post-translational modifications, the apparent molecular weight of SH3PXD2B shown on SDS-PAGE gel was slightly larger than predicted (102 kDa). Because the anti-SH3PXD2B antibody was generated against a portion of the SH3PXD2B protein deleted by the nee mutation, it was not possible to test levels or localization of the mutant protein. However, wild-type SH3PXD2B was present in the majority of tissues, including eye, white adipose tissue, brown adipose tissue, lung, heart, brain, spleen, stomach, liver, and skeletal muscle. No protein was detected in the kidney or bone marrow, indicating that the protein is not ubiquitously expressed. With the exception of lung and skeletal muscle tissue, the relative amount of protein in each tissue resembled previously reported Sh3pxd2b mRNA levels in these tissues (Buschman et al. 2009; Hishida et al. 2008).
Phenotype-driven genetics is a powerful approach for studying genetic pathways important to human health and disease. One advantage of this approach is that it often gives rise to animal models with a diversity of mutant alleles and genetic backgrounds. Here, we have utilized the nee mutation, which spontaneously arose on a B10.A-H2h4/(4R)SgDvEg genetic background, to link functions of the Sh3pxd2b gene with multiple defects of postnatal growth and development. Initially recognized because of runted growth, craniofacial abnormalities, and ocular anterior segment dysgenesis recognizable from an early age, we have also found that adult mice homozygous for the nee mutation exhibit infertility, hearing deficiency, reduced areal bone mineral density, and a form of lipodystrophy depleting white adipose tissue. Mapping of the causative mutation led to identification of a protein truncating mutation in Sh3pxd2b, a gene about which relatively little is known (Buschman et al. 2009; Courtneidge et al. 2005; Hishida et al. 2008). Based on previous work (Buschman et al. 2009), and our own direct data presented here, Sh3pxd2b encodes a widely expressed podosomal-adaptor protein. In nearly all of the tissues affected by the Sh3pxd2bnee mutation, podosomes have previously been observed and implicated in functional contributions. Combined, these data add to the mounting evidence linking roles for podosomes in multiple physiological contexts involving ECM regulation.
Podosomes and ECM remodeling are both known to play significant roles in bone biology (Krane and Inada 2008; Ory et al. 2008; Saltel et al. 2008). Podosomes have been particularly well studied in osteoclasts, a specialized cell that digests bone matrix in a process involving formation of podosomes. Growth retardation and craniofacial defects have previously been noted in mice with mutations influencing osteoclast function (Chen et al. 2007; Holmbeck et al. 1999; Kornak et al. 2001; Shi et al. 2008). As would be expected for a podosomal-protein influencing osteoclasts, mutations in Src result in growth retardation, craniofacial defects, and osteopetrosis (Soriano et al. 1991). Interestingly, mutations in other genes linked to podosomes, such as genes encoding various metalloproteinases, can lead to decreases in bone mineral density (Egeblad et al. 2007; Holmbeck et al. 1999; Mosig et al. 2007). Our observation of decreased areal bone density in nee homozygotes suggests that Sh3pxd2b is an important member of this later type of genetic pathway influencing skeletal and craniofacial development.
In the eye, podosome-like structures have been observed in trabecular meshwork cells. When cultured on a layer of collagen, these cells exhibit focal degradations of collagen overlapping spots of podosome-like structures (Aga et al. 2008). The trabecular meshwork is an ECM-rich filter for aqueous humor outflow located in the irideocorneal angle. Trabecular meshwork defects can cause elevated intraocular pressure and likely contribute to human glaucoma (Acott and Kelley 2008). From the constellation of ocular phenotypes we have observed, it is likely that developmental malformation of the irideocorneal angle or inefficient ECM remodeling mediated by a disruption of podosomes results in elevated intraocular pressure (causing increased depth of the anterior chamber) and damaging the corneal endothelium (causing corneal opacities). Given these observations, it is possible that nee homozygotes may also develop glaucoma, a hypothesis we are currently testing directly.
There are also links between podosomes and fertility. In males, the tubulobulbar complex, a type of intercellular junction important in sperm release and spermatocyte translocation, is thought to have properties resembling podosomes (Young et al. 2009a; Young et al. 2009b). Thus, SH3PXD2B may influence male fertility through function of these structures. Another distinct possibility is that SH3PXD2B might contribute to fertility via regulation of ADAMs (Eto et al. 2002; Evans 2001). ADAM15, the protein we have shown may interact with SH3PXD2B through its last SH3 domain, is present in mouse spermatozoa (Pasten-Hidalgo et al. 2008) and binds effectively to oocytes (Eto et al. 2002). Thus, the infertility of nee mice may also be a consequence of ADAM15 mislocalization. In females, potential links to podosomes are less clear, but it is noteworthy that defects in the female reproductive tract have previously been observed in mice with targeted mutations of other genes associated with podosomes (Holmbeck et al. 1999; Shimizu et al. 2005).
Podosomes have not previously been reported in adipose tissue, but may well exist. Aside from our observation that mice homozygous for the nee mutation develop a form of lipodystrophy, a role for SH3PXD2B in adipose tissue has recently been independently reported in differentiation of 3T3-L1 preadipocyte cells (Hishida et al. 2008). Based on the finding that Sh3pxd2b was expressed early in the differentiation process of 3T3-L1 preadipocytes, Hishida et al. performed RNAi-mediated knockdown experiments and found that Sh3pxd2b function is required for preadipocyte differentiation and clonal expansion. Interestingly, adipocyte differentiation is also influenced by multiple proteases, including MMP14 (MT1-MMP), MMP2, and MMP9 (Bourlier et al. 2005; Chun et al. 2006).
In addition to influencing podosomes, an alternative hypothesis is that SH3PXD2B may regulate metalloproteinases with sheddase activity. For example, MMP14 (MT1-MMP) negatively regulates osteoclastogenesis by causing ectodomain shedding of TNFSF11 (RANKL) (Hikita et al. 2006). It is therefore plausible that SH3PXD2B might influence areal bone mineral density by regulating membrane localization and sheddase activity of MMP14. Likewise, SH3PXD2B might influence sheddase activities in other tissues as well. Given the previous demonstration of sheddase regulation by SH3PXD2A (Malinin et al. 2005), a similar function for SH3PXD2B remains possible.
At the molecular level, it will be of particular importance to define additional SH3PXD2B interacting proteins. Our current experiments have been guided in two ways, previous identification of proteins interacting with SH3PXD2A/SH3PXD2B and our molecular identification of the nee mutation; leading to the suggestion that SH3PXD2B may interact with ADAM15. However, many additional interacting proteins likely remain to be identified. Previous studies have demonstrated interactions of SH3PXD2A with ADAM12, ADAM15, ADAM19 (Abram et al. 2003), and WASL (Oikawa et al. 2008). In addition, MMP14 (MT1-MMP) localization has previously been shown to be influenced by SH3PXD2B (Buschman et al. 2009). The possibility of interactions between SH3PXD2B and these candidates, as well as others, remain to be tested.
With a growing list of tissues displaying phenotypes associated with podosome-type adhesions, genes such as SH3PXD2B are attractive candidates for contributing to a number of diseases and complex traits. For example, genetic association studies for human obesity have detected suggestive linkage with SH3PXD2B at 5q35 (Liu et al. 2008; Zhao et al. 2007), although some studies have failed to replicate the association (Vogel et al. 2009). It is also interesting that a quantitative trait locus for growth in mice has previously been mapped near the chromosomal location of Sh3pxd2b (Rocha et al. 2004).
In conclusion, a phenotype-driven search for new mouse models of human disease led to identification of multiple phenotypes caused by the Sh3pxd2bnee mutation in mice. Sh3pxd2b encodes a widely expressed podosomal-adaptor protein influencing postnatal growth and development of several mesenchymal derived tissues, including craniofacial structures, the ocular irideocorneal angle, and white adipose tissue. To our knowledge, this is the first report of mutant in vivo phenotypes associated with either Sh3pxd2b, or the closely related Sh3pxd2a genes. While most studies of podosome-type adhesions to date have been dominated by studies of osteoclasts, macrophages, and tumor cells, the current findings suggest previously unknown roles for podosomes in several additional tissues. In ongoing work, it will be of particular interest to use mice with the Sh3pxd2bnee mutation as a resource to further delineate the genetic pathways regulating function of podosome-type adhesions, to continue mechanistic studies of SH3PXD2B in specific cell-types, and to assess potential contributions of podosomal-adaptor proteins to human health and disease.
We thank Adam Hedberg-Buenz for maintaining mouse colonies at the University of Iowa; Pat Ward-Bailey for help with the nee mapping cross; John Fingert and Val Sheffield for contributions to DNA sequencing; Gloria Lee for kindly providing the Src (Y529F) construct; Michael Henry for assistance with X-ray imaging; Wayne Johnson and Robert Mullins for helpful discussions; and Michelle Curtain, an animal caretaker at The Jackson Laboratory, who observed the founder nee mouse and provided it to The Jackson Laboratory’s Mouse Mutant Resource. We are especially grateful for many helpful discussions with Norm Hawes of The Jackson Laboratory, who recently passed away; his selfless attitude toward science and kindness were a joy and inspiration. This work was supported by EY017673 and a grant from the Knights Templar Eye Foundation, Inc. to MGA. The Mouse Mutant Resource of The Jackson Laboratory is supported by a grant from the National Center for Research Resources (RR01183); contributions to work on the craniofacial and skeletal phenotypes by LRD and BH were supported by EYO15073; contributions to the assessment of the hearing phenotype by QYZ and KRJ were supported by DC005846 and DC004301; contributions to initially recognizing ocular phenotypes of nee mice by BC were supported by the Foundation Fighting Blindness; and work at The Jackson Laboratory was additionally supported by National Cancer Institute Grant CA34196.
This manuscript also appears in Mammalian Genome. The original publication is available at www.springerlink.com