Biochemical and genetic analysis of ANK in arthritis and bone disease

Abstract

Mutations in the progressive ankylosis gene (Ank/ANKH) cause surprisingly different skeletal phenotypes in mice and humans. In mice, recessive loss-of-function mutations cause arthritis, ectopic crystal formation, and joint fusion throughout the body. In humans, some dominant mutations cause chondrocalcinosis, an adult-onset disease characterized by the deposition of ectopic joint crystals. Other dominant mutations cause craniometaphyseal dysplasia, a childhood disease characterized by sclerosis of the skull and abnormal modeling of the long bones, with little or no joint pathology. Ank encodes a multiple-pass transmembrane protein that regulates pyrophosphate levels inside and outside tissue culture cells in vitro, but its mechanism of action is not yet clear, and conflicting models have been proposed to explain the effects of the human mutations. Here, we test wild-type and mutant forms of ANK for radiolabeled pyrophosphate-transport activity in frog oocytes. We also reconstruct two human mutations in a bacterial artificial chromosome and test them in transgenic mice for rescue of the Ank null phenotype and for induction of new skeletal phenotypes. Wild-type ANK stimulates saturable transport of pyrophosphate ions across the plasma membrane, with half maximal rates attained at physiological levels of pyrophosphate. Chondrocalcinosis mutations retain apparently wild-type transport activity and can rescue the joint-fusion phenotype of Ank null mice. Craniometaphyseal dysplasia mutations do not transport pyrophosphate and cannot rescue the defects of Ank null mice. Furthermore, microcomputed tomography revealed previously unappreciated phenotypes in Ank null mice that are reminiscent of craniometaphyseal dysplasia. The combination of biochemical and genetic analyses presented here provides insight into how mutations in ANKH cause human skeletal disease.

Figure 1.

ANK stimulation of PPi translocation across the plasma membrane. A, Agarose gel of wild-type (WT) Ank and mutant ank cRNAs, with western blot of membrane-enriched protein lysates from Xenopus oocytes 3 d after cRNA injection. The cRNA appears intact, and the ANK protein was detected. The mutant ank cRNA encodes a truncated protein (E440X) that does not contain the C-terminal epitope recognized by the ANK antibody and therefore served as a control for antibody specificity. B, Time course of PPi uptake by oocytes in transport buffer with 1 μM ³³PPi. After subtracting uptake by water-injected controls, the calculated rate of ANK-dependent transport is 1.54±0.04 fmol per oocyte per min (n=3). C, Rate of ANK-stimulated PPi transport as a function of substrate concentration. The transport rate was saturable, and the apparent Km was 1.33±0.59 μM after subtracting uptake by water-injected controls (n=8). A single asterisk (*) indicates P<.05; a double asterisk (**) indicates P<.01; a triple asterisk (***) indicates P<.001.

Figure 2.

PPi-transport assay with disease-causing Ank alleles. A, Oocytes expressing mutant Ank alleles were assayed for PPi uptake over a 20-min period at near-saturating conditions (7 μM). Note that oocytes expressing either of the CCAL2 proteins (P5T or M48T) exhibited wild-type uptake, whereas oocytes expressing the truncated ank protein (E440X) or either of the CMD proteins (C331R or G389R) exhibited little to no uptake. An asterisk (*) indicates P⩽.0005 versus wild-type control; n⩾3). B, Apparent Km measurements for wild-type (WT) and CCAL2 proteins. Neither mutant was significantly different from wild-type controls tested in parallel (n=4 for each mutant allele; n=8 for wild type). ND = not determined. C, Oocytes were co-injected with a 1:1 ratio of wild-type and mutant Ank cRNA. Both CMD proteins (C331R and G389R) had a statistically significant dominant negative effect, although the effect from C331R was stronger. Numbers on the X-axis are ng of cRNA injected. Values from water-injected oocytes were subtracted from experimental values. A single asterisk (*) indicates P<.05; a double asterisk (**) indicates P<.003 versus wild-type control; n=4). D, Surface biotinylation assays for plasma membrane localization of wild-type and mutant ANK proteins when expressed in Xenopus oocytes. All of the tested variants were detected at the surface, with the exception of the truncated ank protein, which served as a control for antibody specificity. The C331R mutant protein showed decreased surface expression compared with wild type. When the dominant negative CMD mutant proteins (C331R and G389R) were coexpressed with wild-type ANK, they did not grossly affect its surface expression. A smaller band is seen in some lanes, but its presence was inconsistent in multiple trials, and we were unable to determine its nature. PDI localizes to the endoplasmic reticulum and served as a negative control for surface expression.

Figure 3.

Transgenic expression of Ank disease alleles in mice. A, Breeding strategy. BACs were modified by homologous recombination in bacteria and were used to generate transgenic founders. Founders were crossed for 2 generations with +/Ank^null mice. B–C, Analysis of RT-PCR products from +/Ank^null;BAC+ and control +/Ank^null;BAC− mice. In panel B, direct sequencing confirmed the expression of the Ank^M48T allele, which is derived from a T→C bp change (arrow). In panel C, restriction digests confirmed the expression of the Ank^G389R allele (arrow), which does not contain an FokI site. The wild-type allele was detected in all cases. D, μCT-derived volumetric reconstructions of digits 2–4 of the right forepaw at age 6 wk. Whereas the Ank^M48T allele rescued the Ank null phenotype in two of three transgenic lines, the Ank^G389R allele did not rescue in any of the four lines. As a control, the wild-type ANK protein expressed from an unmodified BAC rescued the Ank mutant phenotypes in four of four transgenic lines (and data not shown).

Figure 4.

Skulls from Ank^null/Ank^null mice, which exhibit CMD-like features. A, Volumetric reconstructions of the foramen magnum from skulls of wild-type and Ank null mice aged 23 wk. The entryway for the spinal chord was significantly smaller in mutant mice (n=6 mice for each genotype). Blackened bars indicate wild type; unblackened bars indicate Ank null; 1, 2, and 3 are the distances represented by red lines in the wild-type image. B, Ventral view of posterior region of skulls scanned by μCT. The skull images are color-coded for thickness, and the histogram plots the average number of pixels per thickness value for +/+, +/Ank^null, and Ank^null/Ank^null mice. The skulls of mice homozygous for the null allele are significantly thicker than those of controls. Note that the increase in thickness is not global but is instead restricted to discrete regions of the skull (n=6 mice for each genotype). C and D, Two dimensional μCT “slice” images depicting the fusion of middle-ear bones in Ank null mice (m = malleus, i = incus, and s = stapes). A single asterisk (*) indicates P<.05; a double asterisk (**) indicates P<.01; a triple asterisk (***) indicates P<.001 versus wild type.

Figure 5.

μCT analysis of the femur. A, Volumetric reconstructions of cortical and cancellous bone from the metaphysis of wild-type and mutant distal femurs. The cortex from Ank null mice was slightly thinner but similar in shape, whereas the cancellous bone was sparse compared with wild type. B, Quantified values for images in panel A for mice euthanized at age 6 wk and 6 mo. Ank null mice exhibited decreased metaphyseal cortical thickness at age 6 wk but not at age 6 mo. In contrast, these mice exhibited a decreased ratio of bone:tissue volume (BV:TV) relative to wild type in the metaphyseal cancellous bone at both age 6 wk and age 6 mo. Note that the presence of BAC-Ank^G389R had no affect on wild-type or Ank null mice. However, this BAC had a significant dominant negative effect on the cancellous region of +/Ank^null mice at age 6 mo. A single asterisk (*) indicates P<.05; a double asterisk (**) indicates P<.01; a triple asterisk (***) indicates P<.001 versus wild-type control; a pound sign (#) indicates P<.01 for BAC+ versus BAC−; the number of mice analyzed is indicated at the base of each bar.

Figure 6.

Range of mutations and phenotypes at the ANKH locus. The nature of mutations is suggested on the basis of biochemical assays and functional tests in transgenic mice. HA = hydroxyapatite crystal deposits; CPPD = CPPD crystal deposits; Dom neg = dominant negative; neomorph = neomorphic allele; eye dist = distance between the eyes.