Severe osteogenesis imperfecta in cyclophilin B-deficient mice

Abstract

Osteogenesis Imperfecta (OI) is a human syndrome characterized by exquisitely fragile bones due to osteoporosis. The majority of autosomal dominant OI cases result from point or splice site mutations in the type I collagen genes, which are thought to lead to aberrant osteoid within developing bones. OI also occurs in humans with homozygous mutations in Prolyl-3-Hydroxylase-1 (LEPRE1). Although P3H1 is known to hydroxylate a single residue (pro-986) in type I collagen chains, it is unclear how this modification acts to facilitate collagen fibril formation. P3H1 exists in a complex with CRTAP and the peptidyl-prolyl isomerase cyclophilin B (CypB), encoded by the Ppib gene. Mutations in CRTAP cause OI in mice and humans, through an unknown mechanism, while the role of CypB in this complex has been a complete mystery. To study the role of mammalian CypB, we generated mice lacking this protein. Early in life, Ppib-/- mice developed kyphosis and severe osteoporosis. Collagen fibrils in Ppib-/- mice had abnormal morphology, further consistent with an OI phenotype. In vitro studies revealed that in CypB-deficient fibroblasts, procollagen did not localize properly to the golgi. We found that levels of P3H1 were substantially reduced in Ppib-/- cells, while CRTAP was unaffected by loss of CypB. Conversely, knockdown of either P3H1 or CRTAP did not affect cellular levels of CypB, but prevented its interaction with collagen in vitro. Furthermore, knockdown of CRTAP also caused depletion of cellular P3H1. Consistent with these changes, post translational prolyl-3-hydroxylation of type I collagen by P3H1 was essentially absent in CypB-deficient cells and tissues from CypB-knockout mice. These data provide significant new mechanistic insight into the pathophysiology of OI and reveal how the members of the P3H1/CRTAP/CypB complex interact to direct proper formation of collagen and bone.

Figure 1. Reduced body size in CypB–knockout mice.

(A) The third exon of the Ppib gene, encoding CypB, was targeted by loxP sites knocked-in to either side. Arrows indicate SacI sites used for Southern blot analysis of the targeted clones, probed from outside of vector sequences (grey rectangle). Mice bearing this allele were mated to MMTV-Cre transgenic mice to delete the exon in the germ cells of female pups. Their offspring demonstrated complete loss of CypB. (B) Southern blot of genomic DNA from targeted (‘T’) and wild-type (‘WT’) ES clones. (C) Western blot of CypB in thymocytes from wild type (lanes 1, 4) or knockout (lanes 2, 3) mice. (D) Average weights of wild-type, heterozygous, or homozygous CypB–knockout mice. n = 3 WT littermates; n = 10 Ppib+/−; n = 12 Ppib−/− mice. Asterisks indicate points that were statistically significantly different by one-tailed students T-test (p<0.05) from the heterozygote weights. (E, F) Typical appearance of wild type and knockout mice.

Figure 2. Abnormal bone in CypB–knockout mice.

(A) Total body radiographs of 38-week-old mice. Note the enhanced curvature of the spine in the Ppib−/− mouse. Bar = 1 cm. (B) 3D reconstruction of slices from micro CT analysis of femurs from CypB knockout and littermate control mice. (C) Bone volume/total volume, trabecular separation, and trabecular number in femurs analyzed by microCT at the indicated ages.

Figure 3. Altered collagen in CypB–knockout mice.

(A) Skin fibroblasts from wild-type or CypB knockout mice were cultured in serum free medium. Secreted collagen from the supernatant was electropheresed on 5% SDS-PAGE and detected by Western Blotting. Type I collagen from mutant animals demonstrated a subtle but reproducible decrease in migration compared to wild type. (B) Mass-spectroscopic analysis of type I and type II collagen from bone and cartilage. Chromatogram of the parent ions in the mass spectra from wild type and Ppib−/− collagen tryptic-peptides. Wild-type collagen is 4-hydroxylated at pro-981 and pro-987, and 3-hydroxylated at pro-986. Collagen from Ppib−/− mice primarily lacked the pro-986 hydroxylation. Indicated peaks were analyzed by MS2 and MS3 to identify hydroxylated proline residues (‘hP’). (C) Transmission electron micrographs of subcutaneous tissue revealing abnormally thick collagen fibrils in mutant mice in comparison to control mice. (bar = 100 nm).

Figure 4. Abnormal localization of collagen in CypB–deficient cells.

MEFs prepared from wild-type (WT) or CypB knockout mice (KO) were cultured with or without 24hr stimulation with ascorbic acid (+AS). Cells were fixed, permeabilized, and co-stained for intracellular collagen and the Golgi marker GM130 (A) or the ER-resident protein PDI (B). (bar = 40 µm).

Figure 5. Abnormal skin in CypB–deficient mice.

(A) Typical appearance of wild-type (left) and Ppib−/− (right) mice demonstrating the laxity of skin. (B) Sections of skin were fixed and stained with H&E, revealing normal cellular architecture of the dermis. Bar = 200 µm. (C) Sirrius red staining demonstrates reduced collagen in skin sections from CypB-knockout mice. Bar = 250 µm. (D) Quantitation of total type I collagen from skin determined by ELISA. (E) Representative stress vs. strain (SvS) curves of skin from wild type, heterozygote, or CypB knockout mice. (F) Toe lengths of SvS curves indicating the normal range of skin stretchiness. (G) Slopes of SvS curves prior to fracture indicate that CypB−/− collagen fibers are more easily pulled past each other. (H) Force of breakage was much lower in CypB-deficient skin (For all comparison, n = 4; *P<0.05; **P<0.01).

Figure 6. Decreased levels of P3H1 protein in the absence of CypB.

(A) GST-cyclophilin B or GST was mixed and pulled down with gelatin sepharose in the absence or presence of cyclosporine A. Proteins were visualized by blotting with antibody to GST. (B) Lysates from wild type cells were mixed with recombinant GST-cyclophilin B or GST, and bound proteins isolated using GSH-agarose prior to SDS PAGE. Proteins were visualized by western blotting with antibodies to the indicated proteins. (C) Binding of GST-CypB to P3H1 in lysates from wild-type (WT) or CypB-knockout fibroblasts (KO) was detected by chromatography on glutathione-agarose followed by western blotting for P3H1. Although P3H1 is present at much lower levels in the absence of CypB, a greater fraction of it binds to recombinant CypB in vitro. (D) Western blot of the indicated proteins from lysates of wild-type and CypB knockout skin fibroblasts. Blotting for beta-actin was performed as a loading control. (E) Proteins from lysates of wild-type and CypB–knockout skin fibroblasts were isolated on gelatin-sepharose. Bound proteins were eluted and detected by western blotting with antibodies to the indicated proteins.

Figure 7. P3H1 and CRTAP are required for efficient binding of CypB to collagen in vitro.

(A) Mouse fibroblasts were transduced with two different shRNA lentiviruses (#65 and #67) to knockdown P3H1 or a control lentivirus (“C”). Lysates were mixed with gelatin-sepharose to adsorb collagen binding proteins, and then recovered for western blotting, as indicated. Total cellular levels of CypB were not affected by knockdown of P3H1, however binding to gelatin was reduced. (B) As described in (A) in the absence or presence (“AS”) of ascorbic acid in the medium. (C) HeLa cells were transduced with a CypB specific shRNA lentivirus or control and lysates probed for gelatin-binding proteins as described. Knockdown of CypB did not affect CRTAP levels or binding to gelatin. (D) Knockdown of CRTAP in HeLa cells reduced P3H1 cellular levels and blocked the binding of CypB to gelatin in vitro.

Figure 8. A working model to summarize the interaction data posits that CRTAP and CypB both serve to stabilize the accumulation of P3H1 within cells.

P3H1 not only hydroxylates the 3 position of proline 986, but also plays a key role by facilitating the binding of CypB to collagen within the ER. In all 3 conditions that lead to OI, P3H1 hydroxylation is absent, and CypB binding to collagen is reduced or absent.