Convergent extension movements in growth plate chondrocytes require gpi-anchored cell surface proteins

Abstract

Proteins that are localized to the cell surface via glycosylphosphatidylinositol (gpi) anchors have been proposed to regulate cell signaling and cell adhesion events involved in tissue patterning. Conditional deletion of Piga, which encodes the catalytic subunit of an essential enzyme in the gpi-biosynthetic pathway, in the lateral plate mesoderm results in normally patterned limbs that display chondrodysplasia. Analysis of mutant and mosaic Piga cartilage revealed two independent cell autonomous defects. First, loss of Piga function interferes with signal reception by chondrocytes as evidenced by delayed maturation. Second, the proliferative chondrocytes, although present, fail to flatten and arrange into columns. We present evidence that the abnormal organization of mutant proliferative chondrocytes results from errors in cell intercalation. Collectively, our data suggest that the distinct morphological features of the proliferative chondrocytes result from a convergent extension-like process that is regulated independently of chondrocyte maturation.

Fig. 1.

Chondrodysplasia in Piga mutant limbs. (A) Gpi-anchored proteins are composed of a lipid tail, a sugar-phosphate linkage, and a protein. (B) Alcian Blue and Alizarin Red staining show a delay in cartilage formation beginning at 12.5 dpc (black arrow) and a delay in mineralization at 14.5 dpc and 16.5 dpc in mutant samples (black arrow). At P0 and P12, mutant bones are mineralized but shorter than wild type. (C) At P0, mutant bones are significantly shorter (P<0.0001) and wider (P<0.0001) than wild type. Heterozygous females (mosaic) have an intermediate length (wild type versus mosaic, P=0.0004; mutant versus mosaic, P<0.0001) and width (wild type versus mosaic, P=0.003; mutant versus mosaic, P=0.0001). (D) Col2a1 expression in immature cartilage is found in both wild-type and mutant limbs. Hypertrophic chondrocytes expressing Col10a1 persist in the diaphysis of mutant bones at 15.5 dpc. Scale bars: in B, 50 μm for 12.5 dpc and 14.5 dpc, 100 μm for 16.5 dpc and P0, 125 μm for P12; in D, 100 μm.

Fig. 2.

Piga mutant cartilage lacks gpi-anchored proteins. (A) Alkaline phosphatase (Alp) activity is analyzed in wild-type (Wt), mosaic, and mutant (Mut) growth plates. The black arrows highlight the chondrocytes and the white arrows point to the perichondrium. The black dashed line in the mosaic sample demarcates the boundary between mutant (left) and wild-type (right) chondrocytes. Alp-positive cells only derive from columnar proliferative chondrocytes. (B) ISH for Akp2 reveals normal expression profiles in wild-type and Piga mutant growth plates. No signal is observed with the sense probe. (C) Flow cytometry shows that Piga mutant (black) chondrocytes lack Gpc3 and have comparable fluorescence to IgG controls (red), in contrast to wild-type chondrocytes that have increased fluorescence (green). (D) Transfected wild-type chondrocytes display gpi-Gfp on the cell surface but Piga mutant chondrocytes do not. Localization of myristolated-Gfp is similar in wild-type and mutant transfected cells. Scale bars: in A, 200 μm for 10× images, 50 μm for 40× images; in B, 200 μm.

Fig. 3.

Chondrocyte organization is altered in Piga mutant limbs. (A) The growth plate contains distinct zones, including resting (RZ), proliferative (PZ), prehypertrophic, and hypertrophic chondrocyte (HZ) zones. (B) Masson's trichrome-stained sections of P0 cartilage reveals that the transition from round disorganized cells to discoid columns is perturbed in Piga mutant chondrocytes. BrdU staining shows that the proliferation rate in Piga mutant RZ is increased compared with in wild type (9.8±0.2% versus 5.7±0.7%); the proliferation rate is increased in the RZ of Piga mutants compared with wild type (14.7±2.2% versus 12.3±1.6%). Cleaved caspase 3 antibody staining shows that apoptosis is absent in the growth plate (box) and normal in mineralized regions (arrow). (C) Trichrome staining of mosaic growth plates demonstrates that stacked discoid wild-type chondrocytes (black arrow) are found adjacent to round disorganized mutant chondrocytes (red arrow). Scale bars in B: 50 μm for trichrome staining; 100 μm for immunostaining.

Fig. 4.

Piga mutant chondrocytes undergo normal stages of maturation. (A) Each zone is present in Piga mutant growth plates as evidenced by ISH using the following markers: fibroblast growth factor receptor 1 (Fgfr1), bagpipe (Bapx1), fibroblast growth factor receptor 3 (Fgfr3), proline/arginine-rich end leucine-rich repeat (Prelp), Indian hedgehog (Ihh) and Col10a1. (B) Fractional length is expressed as a ratio of the measurement of a particular zone over the length of the entire growth plate (articular surface to mineralized region). Fgfr1 expression is similar between wild-type and mutant limbs (P=0.033), but Prelp and Ihh are significantly decreased (P<0.001, P<0.005). Col10a1 is not significantly reduced (P=0.173). Although Fgfr1 expression is similar, the region between the epiphysis and the distal end of Prelp expression is significantly increased (red bracket, P<0.0001).

Fig. 5.

Cell autonomous delay in Piga chondrocyte maturation. (A) Mutant proliferative chondrocytes express Prelp later in maturation than do wild-type chondrocytes in mosaic sections. This is also the case for Ihh. (B) Measurement of fractional length in wild type, mosaic and mutant growth plates. Piga mutant chondrocytes are not deficient in receiving all signals because the downstream target of hedgehog signaling (Ptc1) is expressed similarly in mutant and wild-type chondrocytes in mosaic sections (P=0.389). Scale bars: 200 μm for 10× images; 50 μm for 40× insets.

Fig. 6.

Piga mutant chondrocytes fail to intercalate into columns. (A) Daughter cells were identified by phalloidin labeling (red) of the cleavage furrow (white arrow); nuclei are stained with DAPI. (B) RZ chondrocytes divide at arbitrary angles, whereas the PZ chondrocytes for both wild-type and mutant growth plates divide nearly perpendicular to the long axis of the bone (θ∼90°; distinct from resting cells P<0.0001). (C) Canonical Wnt signaling is unaltered in Piga mutant chondrocytes, as measured by β-galactosidase (β-gal) activity from the TOPGAL reporter. The patchy β-gal staining corresponds to clonally related chondrocytes. (D) In the proliferative zone, the total number of cells in a clone and the clone width were counted, revealing that Piga mutant clones are significantly wider (P<0.0001) and shorter (P<0.0001) than wild-type clones. (E) The midbody (acetylated tubulin, green) connects the dividing daughter cells in both wild-type and mutant chondrocytes. Scale bars: in A, 50 μm; in C, 200 μm; in E, 100 μm.

Fig. 7.

Gpi-anchored proteins regulate polarity in hair cells. (A) Stereocilia labeled with phalloidin (red) and kinocilia labeled with acetylated tubulin (green) are organized with the apex at an ∼90° angle to the basal surface of the cochlea. Hair cell orientation was calculated by measuring the angle that the kinocilia deviate from 90°. (B) Piga mutant inner hair cells have normal orientation (P=0.092). Piga mutant outer hair cells are significantly perturbed (OHC1, P=0; OHC2, P=0; OHC3, P<0.0001). (C) Phalloidin (red) and Vangl2 (green) show that Piga mutant outer hair cells have aberrant architecture and reduced localization of Vangl2 compared with wild type.