Unsuspected osteochondroma-like outgrowths in the cranial base of Hereditary Multiple Exostoses patients and modeling and treatment with a BMP antagonist in mice

Abstract

Hereditary Multiple Exostoses (HME) is a rare pediatric disorder caused by loss-of-function mutations in the genes encoding the heparan sulfate (HS)-synthesizing enzymes EXT1 or EXT2. HME is characterized by formation of cartilaginous outgrowths-called osteochondromas- next to the growth plates of many axial and appendicular skeletal elements. Surprisingly, it is not known whether such tumors also form in endochondral elements of the craniofacial skeleton. Here, we carried out a retrospective analysis of cervical spine MRI and CT scans from 50 consecutive HME patients that included cranial skeletal images. Interestingly, nearly half of the patients displayed moderate defects or osteochondroma-like outgrowths in the cranial base and specifically in the clivus. In good correlation, osteochondromas developed in the cranial base of mutant Ext1f/f;Col2-CreER or Ext1f/f;Aggrecan-CreER mouse models of HME along the synchondrosis growth plates. Osteochondroma formation was preceded by phenotypic alteration of cells at the chondro-perichondrial boundary and was accompanied by ectopic expression of major cartilage matrix genes -collagen 2 and collagen X- within the growing ectopic masses. Because chondrogenesis requires bone morphogenetic protein (BMP) signaling, we asked whether osteochondroma formation could be blocked by a BMP signaling antagonist. Systemic administration with LDN-193189 effectively inhibited osteochondroma growth in conditional Ext1-mutant mice. In vitro studies with mouse embryo chondrogenic cells clarified the mechanisms of LDN-193189 action that turned out to include decreases in canonical BMP signaling pSMAD1/5/8 effectors but interestingly, concurrent increases in such anti-chondrogenic mechanisms as pERK1/2 and Chordin, Fgf9 and Fgf18 expression. Our study is the first to reveal that the cranial base can be affected in patients with HME and that osteochondroma formation is amenable to therapeutic drug intervention.

Fig 1. HME patients have cranial base defects.

(A) Sagittal plane T2 MRI sequence showing an example of a normal clivus (delineated by pink arrowheads) and normal posterior foramen magnum (cyan arrowhead) in a HME patient. (B-C) Sagittal plane T1 and T2 MRI sequences showing a very large osteochondroma-like outgrowth (red arrowheads) in a patient that is characterized by continuity with clivus cortical bone and marrow. For orientation, note that the cerebellum is the large tissue mass occupying the upper right half of the images. (D) Sagittal T2 MRI sequence from a patient displaying an irregularity in the occipital lip of the foramen magnum (purple arrowhead). (E) Coronal plane T2 MRI sequence showing an osteochondroma-like outgrowth (orange arrowhead) and a more normal contralateral side (yellow arrowhead).

Fig 2. Osteochondromas form in the cranial base of Ext1 mutant mice over time.

(A-F) Longitudinal sections through the spheno-occipital synchondrosis (sos) in control tamoxifen-injected mice at 2 weeks (A-C), 1 month (D), 3 months (E) and 6 months (F) from injection stained with safranin O and fast green. Note in (B) the characteristic growth plate organization with resting (rt), proliferative (pl), prehypertrophic (ph) and hypertrophic (hp) zones, and note in (C) the distinct inner and outer portions of perichondrium (pc) (arrow and double arrowhead, respectively) along the perichondrium/growth plate (PC/GP) border. (G-R) Longitudinal sections through the spheno-occipital synchondrosis in Ext1^f/f;Col2-CreER mutant mice at 2 weeks (G-I), 3 weeks (M-O), 1 month (J, P), 3 months (K, Q) and 6 months (L, R) from tamoxifen injection. Note that the chondro-perichondrial border is already deranged at 2 weeks as indicated by presence of round, safranin O-positive cells within perichondrium (I, arrowhead). By 3 weeks and thereafter, the border becomes occupied by incipient osteochondromas (O, arrowheads) that enlarge (J) and ossify proximally over time (K and L, asterisks) and maintain a characteristic cartilage cap at their distal end (K and L, arrowheads) with a growth plate-like organization (P and R) with scattered Edu-labeled proliferative cells (Q, arrrowheads). Bar in (A) for A, G and M, 1 mm; bar in (C) for B, C, H, I, N and O, 50 μm; bar in (D) for D, E, F, J-L and R, 3 mm; bar in (P) for P and Q, 300 μm.

Fig 3. Imaging reveals multiple cranial base defects in mutant mice.

(A-B) Bird’s eye μCT images of the skull from control tamoxifen-injected mice at 5 month time point from nasal end (left) to occipital area (right). At higher magnification (B), square area in green shows the smooth and continuous contour of the intrasphenoidal (iss) and spheno-occipital (sos) synchondroses flanked by the sphenoidal (sp) and occipital (oc) bones. (C) Lateral μCT image showing that the control cranial base is flat and relatively thin along and around the synchondroses (vertical arrow) as expected. (D-F) Bird’s eye and lateral μCT images of the skull from companion mutant tamoxifen-injected Ext1^f/f;Col2-CreER mice showing that multiple osteochondromas (E, arrowheads) developed near the iss and sos synchondroses (bracketed areas in E) and that the mutant cranial base was considerably and abnormally thickened (vertical arrow) compared to controls (C). Bar in (A) for A and D, 5 mm; bar in (B) for B and E, 1.3 mm; and bar in (C) for C and F, 1 mm.

Fig 4. Cranial base osteochondromas display growth plate-like gene expression patterns.

(A-E) Longitudinal serial sections through the spheno-occipital synchondrosis and flanking tissues from control mice at one month from tamoxifen injection. Sections were stained with safranin O-fast green to reveal cartilage organization and intact perichondrium (pc) and border (arrows) (A), and were processed for in situ hybridization expression analysis of such typical growth plate genes as: (B) collagen II (Col II); (C) collagen X (Col X); (D) metalloprotease 13 (Mmp13); and (E) collagen I (Col I). Arrowheads in B-D point to characteristic areas/sites of maximal gene expression in growth plates and arrows in E point to Col I expression in bone collar. (F-J) Longitudinal serial sections through osteochondromas present along the nasal aspect of cranial base in companion mutant tamoxifen-injected Ext1^f/f;Col2-CreER mice. Arrows in F point to osteochondromas protruding away from the cranial base into surrounding perichondrium and nasal cavity. Arrowheads in G-I point to areas of maximal and somewhat deranged gene expression patterns within the osteochondromas. Arrows in J point to Col I expression in bone collar. Bar in (F) for all panels, 200 μm.

Fig 5. Stereotypic osteochondromas form in a juvenile HME mouse model.

(A-F) Bright field and fluorescence images of longitudinal sections of spheno-occipital (sos) synchondroses (A-C), tibia (D-E) and ribs (F) from 5 week-old control R26-tdTomato;Agr-CreER mice that had been injected with vehicle 3 days earlier. The fluorescence images (B and E) show absence of reporter activity in growth plates (gp) and perichondrium (pc) at any anatomical site examined and thus, lack of CreER leakage in this model. Note that ribs contain a large resting cartilage (rc) adjacent to the growth plate (F). (G-H) μCT images of cranial base and knee from control Ext1^f/f;Agr-CreER mice sacrificed 6 weeks from vehicle injection and displaying normal and typical anatomical characteristics. (I-N) Bright field and fluorescence images from companion R26-tdTomato;Agr-CreER mice that were administered tamoxifen once and were sacrificed 3 days post-injection. Note that reporter activity is very strong in spheno-occipital synchondrosis (sos), tibia and rib growth plates and rib resting cartilage (J, K, M and N) and is also conspicuous in flanking perichondrial (pc) cells at each anatomical location (arrowheads in K and M). Boxed area in J is shown at higher magnification in K. (O-P) μCT images of cranial base and knee from mutant Ext1^f/f;Agr-CreER mice sacrificed 6 weeks after tamoxifen injection. Note the large multiple osteochondromas (arrowheads) protruding away from the bone surfaces at each anatomical site and better appreciable when contrasted to corresponding images from companion controls (G-H). Bar in (A) for A, B, I and J: 300 μm; bar in (C for C and K, 150 μm; bar in (D) for D, E, L and M, 250 μm; bar in (F) for F and N, 150 μm; and bar in (G) for G and O, 0.8 mm; and in (H) for H and P, 5 mm.

Fig 6. Osteochondroma development in juvenile mice is inhibited by systemic treatment with BMP signaling antagonist LDN-193189.

(A-E) Lateral and bird’s eye μCT images of the cranial base from mutant Ext1^f/f;Agr-CreER mice sacrificed 6 weeks from tamoxifen injection that were administered vehicle daily throughout the treatment period. Note the presence of multiple osteochondromas near the intrasphenoidal (iss) and spheno-occipital (sos) synchondroses highlighted by double arrowheads in A, C and E. Squared areas in B and D are shown at higher magnification in C and E. (F) Representative histochemical image from a serial section throughout the cranial base osteochondromas from above mutant mice. Staining with safranin O and fast green reveals the conspicuous cartilaginous portion of the tumor and presence of a thick perichondrium surrounding its distal end (double arrowheads). (G-K) Lateral and bird’s eye μCT images of the cranial base from mutant Ext1^f/f;Agr-CreER mice sacrificed 6 weeks after tamoxifen injection that were administered LDN-193189 daily throughout that period. Note the significant and obvious reduction in osteochondroma size (arrowheads) that is best appreciable at higher magnification of squared areas in H and J shown in I and K. (L) Representative histochemical image from a serial section throughout the cranial base osteochondromas from above mutant LDN-treated mice. Note the reduction in the cartilaginous tumors (arrowheads). (M and N) Histograms of average bone tumor volume and cartilage tumor volume, respectively, in vehicle-treated and LDN-treated mice. Bar in (G) for A, B, D, G, H and J, 1.2 mm; bar in (C) for C, E, I and K, 0.5 mm; bar in (L) for F and L, 50 μm.

Fig 7. Osteochondroma development in long bones and ribs in juvenile mice is inhibited by systemic LDN-193189 treatment.

(A to D) μCT images (A, B and D) and X-ray image (C) of knee and ribs from mutant Ext1^f/f;Agr-CreER mice receiving vehicle treatment and showing large osteochondromas (double arrowheads). (E to H) Analogous images from companion mutant mice receiving LDN treatment for 6 weeks. Note the appreciable reduction in osteochondorma size in both knee and ribs (arrowhead). Bar in (A) for A, B, E and F, 5 mm; bar in (C) for C and G, 3 mm; and bar in (D) for D and H, 1.5 mm.

Fig 8. LDN-193189 is a strong and direct inhibitor of chondrogenesis.

(A-J) Alcian blue staining and optical quantification of mouse embryo limb bud cell micromass cultures on day 4 and 6 reared in control medium (A and F) or medium containing rhBMP2 (B and G), LDN (C and H) or rhBMP2 plus LDN (D and I). Note the drastic inhibition of chondrogenesis by LDN treatment at both day 4 and 6 (C and H) and partial but still major inhibition in co-treated cultures (D and I). Quantification of alcian blue staining confirms visual data and establish statistical significance (E and J). (** p < 0.005; *** p < 0.001). (K-P) Histograms of qPCR data showing that LDN treatment caused significant decreases in master cartilage genes Sox9 and aggrecan (Acan) on both day 4 and 6 of culture compared to levels in control or rhBMP2-stimulated cultures, while concurrently increasing the gene expression levels of the endogenous BMP antagonist Chordin. (* p < 0.05; ** p < 0.01; *** p < 0.001).

Fig 9. LDN-193189 differentially alters pro- and anti-chondrogenic signaling pathways and factors.

(A-D) Immunoblot images and quantification showing that LDN treatment of micromass cultures significantly reduced levels of pro-chondrogenic pSMAD1/5/8 proteins while concurrently increasing the levels of anti-chondrogenic pERK1/2 proteins relative to controls (** p < 0.01; *** p < 0.001). (E-F) Histograms of qPCR data showing that LDN treatment caused a marked increase in gene expression of Fgf9 and Fgf18, genes normally expressed in perichondrium that may be needed to normally maintain its fibroblastic phenotype.