Osteochondroma formation is independent of heparanase expression as revealed in a mouse model of Hereditary Multiple Exostoses

Abstract

Hereditary Multiple Exostoses (HME) is a rare, pediatric disorder characterized by osteochondromas that form along growth plates and provoke significant musculoskeletal problems. HME is caused by mutations in heparan sulfate (HS)-synthesizing enzymes EXT1 or EXT2. Seemingly paradoxically, osteochondromas were found to contain excessive extracellular heparanase (Hpse) that could further reduce HS levels and exacerbate pathogenesis. To test Hpse roles, we asked whether its ablation would protect against osteochondroma formation in a conditional HME model consisting of mice bearing floxed Ext1 alleles in Agr-CreER background (Ext1^f/f;Agr-CreER mice). Mice were crossed with a new global Hpse-null (Hpse^−/−) mice to produce compound Hpse^−/−;Ext1^f/f;Agr-CreER mice. Tamoxifen injection of standard juvenile Ext1^f/f;Agr-CreER mice elicited stochastic Ext1 ablation in growth plate and perichondrium followed by osteochondroma formation, as revealed by micro-computed tomography and histochemistry. When we examined companion conditional Ext1-deficient mice lacking Hpse also, we detected no major decreases in osteochondroma number, skeletal distribution and overall structure by the analytical criteria above. The Ext1 mutants used here closely mimic human HME pathogenesis but have not been previously tested for responsiveness to treatments. To exclude some innate therapeutic resistance in this stochastic model, tamoxifen-injected Ext1^f/f;Agr-CreER mice were administered daily doses of the retinoid Palovarotene, previously shown to prevent ectopic cartilage and bone formation in other mouse disease models. This treatment did inhibit osteochondroma formation compared to vehicle-treated mice. Our data indicate that heparanase is not a major factor in osteochondroma initiation and accumulation in mice. Possible roles of heparanase upregulation in disease severity in patients are discussed.

1. INTRODUCTION

Hereditary Multiple Exostoses (HME) is a rare, autosomal dominant, musculoskeletal pediatric disorder with an estimated incidence of 1:50,000 individuals worldwide.¹ Known also as Multiple Osteochondroma (MO), HME is characterized by benign tumors (referred to as exostoses or osteochondromas) that develop within perichondrium flanking the growth plate in long bones, ribs and other skeletal elements in children and adolescents. ^2–4 The osteochondromas initiate with formation of ectopic cartilage, acquire a growth plate-like structure themselves and elongate outwardly, and undergo endochondral ossification in their proximal region, thus remaining connected to the neighboring osseous elements from which they originate. Because of their large number, size and multiple locations, the osteochondromas can cause a variety of health problems that include skeletal deformities, growth retardation, chronic pain, and impingement of nerves, muscles and vessels.^{5; 6} Surgery is used to resect the most symptomatic osteochondromas, but because of their number and challenging anatomical locations, many osteochondromas are left in place, causing life-long problems and increasing the frequency of malignant transformation into osteosarcomas.⁷ At present, there is no approved drug treatments to prevent, reduce or reverse the development of osteochondromas.^{3; 8}

The majority of HME cases are linked to a loss-of-function heterozygous mutation in the tumor suppressor genes EXT1 or EXT2. ^9–12 These genes encode Golgi glycosyltransferases responsible for heparan sulfate (HS) synthesis and assembly onto the core proteins of cell surface and matrix proteoglycans such as syndecans and glypicans.^13–16 Because both EXT1 and EXT2 proteins are needed for efficient HS polymerization, the heterozygous EXT mutation carried by HME patients causes a systemic decrease of HS levels of about 50%.¹⁷ Such partial decrease is found to elicit appreciable changes in certain HS-dependent processes such as lipid metabolism in liver,¹⁸ but is thought to be insufficient to cause osteochondroma development on its own.³ In accordance with general traits and rules of tumorigenesis,^{19; 20} osteochondroma development requires a stochastic somatic second hit that would directly or indirectly lead to a further drop in HS levels, rendering local cells virtually devoid of HS.^{2; 3} In line with these theses, patient genetic studies did detect loss-of-heterozygosity (LOH), aneuploidy or other chromosomal changes in human osteochondroma surgical retrieval specimens, rendering some local cells EXT1- or EXT2-null.^21–24 Other studies detected additional genetic changes, including compound heterozygous EXT1 plus EXT2 mutations in osteochondromas in the same patient that would be expected to reduce HS levels by about 75%.^25–29 Findings in Ext mouse mutants are in agreement with human studies. Whereas global heterozygous loss of Ext1 or Ext2 causes minimal phenotypic changes in young mice,^{30; 31} conditional homozygous loss of both Ext1 or Ext2 alleles as well as compound heterozygous loss of Ext1 plus Ext2 causes widespread osteochondroma development, mimicking human HME.^31–33 The overall notion emerging from human and mouse studies is that osteochondroma formation requires a steep loss of HS in affected local cells, switching their phenotype to tumorigenic.

Intriguingly, previous studies in other cancer types showed that decreases or loss of EXT expression and HS production are accompanied by an increase in gene expression of heparanase (Hpse), the only vertebrate secreted enzyme that acts on and metabolizes extracellular HS.^34–38 These seemingly paradoxical observations indicate that when HS synthesis is reduced, cancer cells actually up-regulate production of extracellular heparanase to further lower their HS levels and do so in proportion to the degree of their malignancy.³⁸ The higher heparanase activity is widely thought to facilitate cancer cell proliferation and survival by liberating extracellular HS-bound growth factors³⁹ and making them bio-available to the cells.^{35; 40–42} Interestingly, excessive Hpse expression has also been found to occur in osteochondromas from HME patients.⁴³ We confirmed these trends by immunohistological analysis of human osteochondromas compared to normal surgical discard growth plates, and we found also that experimental reduction of HS levels in cultured cells boosted Hpse expression.⁴⁴ Together, these previous studies raised the possibility that heparanase could be a culprit in HME pathogenesis. In the present study we tested this important thesis by delineating and comparing osteochondroma formation in conditional stochastic Ext1-deficient HME mice³² bearing or not global Hpse ablation.

2. METHODS

2.1. Ethics statement regarding mouse studies

All experiments involving wild type and transgenic mice were reviewed by the IACUC at The Children’s Hospital of Philadelphia (protocol no. IAC 14–000952, principal investigator MP). Animals were handled, treated and cared for according to the approved protocols and all procedures fully complied with the ARRIVE guidelines.

2.2. Generation and verification of Hpse-null mice

Mice harboring a 905 bp genomic deletion spanning the proximal promoter and exon 1 of the heparanase (Hpse) gene were obtained by co-injection of Cas9 mRNA and single guide (sg) RNAs (Table 1) into C57BL/6J fertilized embryos via non-homologous end joining (NHEJ) – mediated repair, using core services provided by Jackson’s Laboratories. Hpse gene deletions were screened by PCR using primers spanning the target sites (Table 2). Sanger sequencing analysis identified and confirmed mosaic founder mice. These founders were bred to wildtype (WT) C57BL/6J mice (Jackson Labs #664, Bar Harbor, ME) to generate F1 heterozygous progeny for subsequent intercrossing. To confirm the generation of Hpse^+/− and Hpse^−/− mice, DNA was extracted from tail biopsies, amplified with primers targeting the deleted region and ran under standard PCR electrophoretic conditions. To confirm the absence of Hpse gene expression, total RNA was isolated from the liver of WT, Hpse^+/− and Hpse^−/− mice using TRIzol reagent (Life Technologies, Carlsbad, CA) according to the manufacturer’s protocol. Whole RNA was quantified by Nanodrop., and 1 μg aliquots were reversed transcribed using the Verso cDNA synthesis kit (Thermo Fisher Scientific, Waltham, MA). cDNA was amplified using primers and PCR conditions as described.⁴⁵ The amplified products were separated by electrophoresis on a 1% agarose gel and visualized by SYBR Safe DNA gel stain (Invitrogen, Carlsbad, CA).

TABLE 1.

Primers for single guide RNAs used to delete the targeted Hpse region

Location	Sequences
Upstream Guides	HpsePro_sgRNA1:AGAAATCGGTCCTTGCACGG HpsePro_sgRNA2:ATCACCACCACCCCCGTGCA HpsePro_sgRNA3:TAGAAATCGGTCCTTGCACCG
Intron 1 Guides	HpseInt1_sgRNA1:CCTGTCCAATCACACACTCGCG HpseInt1_sgRNA2:CTGTCCAATCACACTCGCGA

The indicated single guide (sg) RNAs were co-injected along with Cas9 mRNA into C57BL/6J fertilized embryos to elicit non-homologous end joining-mediated repair and deletion of 905 bp genomic segment spanning Hpse proximal promoter and exon 1.

TABLE 2.

Primer pairs to genotype Hpse mouse mutants

Primer*	Sense (5’−3’)	Antisense (5’−3’)	Length (bp)
Hpse	GGGTTGCTCAGGATTCCAAG	GGCTCCAGACAAAGTGCTAAG	WT:1257 KO: 352
Deleted region	CCAATGCTCGGATCA	CCAGGACTCAGTACC	485

^*Primers in top row flank the targeted deleted Hpse region and generate a 1257 bp product in WT and a 352 bp product in knock-out mice by PCR. Primers in lower row are from within the deleted region itself and generate a PCR product of 485 bp in WT and heterozygous mice.

2.3. Transgenic mouse lines, husbandry and drug treatment

Mice bearing an inducible loss-of-function allele of Ext1 in which exon 2 is flanked by head-to-head loxP sites (heretofore referred to as Ext1^f/f mice) were created and described previously³². All mice were housed in conventional housing (5 mice per cage) and given unlimited access to food and water. Shepard shacks were added to the cages for environmental enrichment. The mice were mated with Agr-CreER^T2 transgenic mice expressing CreER recombinase under the control of aggrecan enhancer sequences⁴⁶ to eventually generate compound Ext1^f/f; Agr-CreER^T2 mice. Hpse^−/− mice were bred to Ext1^f/f; Agr-CreER^T2 to generate Ext^+/f; Agr-CreER^T2;Hpse^+/− mice. Hpse^−/− mice were also bred to Ext1^f/f mice to generate Ext^+/f;Hpse^+/− mice that were mated with Ext^+/f;Agr-CreER^T2;Hpse^+/− mice to generate experimental Ext^f/f;Agr-CreER^T2;Hpse^−/− mice as well as control littermates. At 5 weeks of age, experimental and control mice were given a single intraperitoneal injection of tamoxifen (1 mg/13.3grs per body weight); stock solution was 20 mg/ml in ethanol: corn oil mixture at 1:4 ratio. Mice were euthanized at 6 weeks post tamoxifen injection to assess osteochondroma development. A total number of 3–5 mice were used per group to evaluate phenotype. ⁴⁷

For Palovarotene (Atomax Chemicals, ShenZhen, China) treatments, the drug was dissolved in DMSO at 10 mg/ml stock solution. Aliquots were prepared and stored at −20°C. Fresh aliquots were used every two weeks. On the day of treatment, an aliquot was thawed and mixed with the appropriate amounts of DMSO and corn oil to generate a working solution; this was used for two weeks and then replaced with a new one. Sufficient numbers of mice per experiment were placed in separate cages and randomized into control and experimental groups to minimize bias. Ext1^f/f;Agr-CreER^T2 mice were given Palovarotene at 3 mg/kg by oral gavage 5 days per week for 6 weeks. Companion controls were given the vehicle mixture of DMSO and corn oil. Each subgroup contained a minimum of 5 mice per experiment. Treatment started one day after tamoxifen injection. Before gavage, mice were anesthetized by inhalation of 1.5% isoflurane in 98.5% oxygen to minimize discomfort as prescribed by IACUC. Oral gavage probes were 22 gauge and 25 mm long (Instech Laboratories, Plymouth Meeting, PA). After drug administration, animals were regularly monitored for signs of overt toxicity, including lethargy and dystonia, and Institutional LAS personnel provided additional daily animal monitoring and care.

2.4. Micro computed tomography (μCT) and histochemical analyses

Ribs and long bones were fixed for 72 hours in 4% paraformaldehyde, washed with 1X PBS and stored in ethanol for scanning. Ribs and long bones were scanned using a μCT45 scanner (Scanco Medical AG, Switzerland) and analyzed using manufacturer’s issued software. Serial 21 μm 2D and 3D images were acquired at 55 kVp energy, 145 μA intensity and integration time of 200 msec. Raw μCT data were compiled into 2D gray scale images. Ribs and long bones were contoured, and binary images were generated using thresholds of 200 and 260, respectively. After imaging procedures, samples were decalcified and embedded in paraffin. Serial 5 μm sections were stained with safranin-O and fast green and processed for histochemical osteochondroma analyses as described.⁴⁷

2.5. Osteochondroma quantification

Histological images of ribs were loaded into ImageJ software and pixel dimensions were converted to millimeters (mm). Regions of interest were drawn around the bony and cartilaginous areas of the osteochondromas from multiple ribs per animal to measure the areas. Surface areas were analyzed using Prism 9 (GraphPad Software Inc.). Student’s t-test and one-way analysis of variance (ANOVA) were used to establish statistical significance. Threshold for significance for all tests was set as p ≤ 0.05.

3. RESULTS

3.1. Creation and verification of Hpse-null mice

As indicated above, heparanase is the only vertebrate secreted enzyme that is responsible for metabolism and cleavage of HS chains present in extracellular and cell surface proteoglycans such as syndecans and perlecan.³⁵ Because HS cleavage has major functional consequences including liberation of HS-bound growth factors and cytokines and strong cellular responses, Hpse is normally expressed by relatively few tissues and is under strict transcriptional regulation.^{35; 40–42} To directly test its possible roles in osteochondroma formation during HME, we first set out to create a line of Hpse-null mice, expecting it to be viable based on previous studies⁴⁵ and thus amenable to be genetically introduced into conditional stochastic HME model mice. The mouse Hpse gene comprises 12 exons with the translation start site within exon 1 (MGI: 1343124). Thus, we used a CRISPR/Cas9 approach to specifically target and ablate a 905 base pair segment that comprises the proximal promoter and exon 1 (Figure S1A,B). We obtained mosaic founders that were intercrossed to propagate the mutation. Using PCR primer pairs spanning the entire targeted region (Table 2), we monitored progenies and found that they displayed Mendelian inheritance. Thus, the expected WT DNA amplification product of 1257 bp was appreciable in homozygous WT and heterozygous mutants but not in homozygous mutants (Figure S1C, left panel lanes), while the expected knock-out amplification band of 352 bp was observed in heterozygous and homozygous mutants only (Figure S1C, left panel lanes). To double-check the results, we used a primer pair from within the deleted region itself. The expected amplification product of 485 bp was seen in WT and heterozygous mice only, but not homozygous mutants (Figure S1C, right panel lanes).

Ablation of the 905 bp region spanning the upstream promoter and exon 1 should result not only in lack of gene expression but also absence of alternative transcripts. To make sure this was the case, we isolated total RNA from the liver of WT and homozygous Hpse mutant mice above and processed them for cDNA synthesis and RT-PCR, using primer pairs specific for the 5’, middle and 3’ portions of Hpse transcript. Amplification products of expected sizes (332, 330 and 331 bp, respectively) were detected in WT sample but not samples from two knock-out mutants (Figure S1D, WT vs KO 1 and KO 2, left and central panels). As internal control for RNA quality, we used primers for housekeeping ribosomal protein L19, and an RT-PCR amplification product of 419 bp was elicited by all samples (Figure S1D, far right panel).

3.2. Osteochondromas form in ribs and long bones in both Ext1- and Ext1;Hpse-null mice

Having obtained an effective Hpse-null mouse line, we moved on to test whether this gene has roles in the development and accumulation of osteochondromas in mice. To this end, we relied on an effective transgenic HME mouse model previously created by Jones et al.³² in which head-to-head LoxP sites introduced into the Ext1 gene lead to conditional Cre-induced stochastic loss of normal alleles, thus mimicking stochastic events such as loss-of-heterozygosity (LOH) seen during pathogenesis of osteochondromas in HME patients.^{22; 23} Floxed Ext1 mice were mated with aggrecan-CreER^T2 (Agr-CreER^T2) mice that elicit tamoxifen-dependent expression of CreER^T2 in growth plate chondrocytes and perichondrial cells normally involved in osteochondroma development.² These mice were then mated to Ext1^+/f;Hpse^−/− mice to ultimately generate compound homozygous Ext1^f/f;Agr-CreER;Hpse^−/− mice as well as control littermates including Ext1^+/f;Agr-CreER, Ext1^f/f;Agr-CreER, and Ext1^f/f;Agr-CreER;Hpse^+/− mice. All mice received a single tamoxifen injection (1 mg/13.3 gr/bw) at 5 weeks and were then monitored over time. Mice were sacrificed after 6 weeks post tamoxifen injection to assess osteochondroma development and growth and to establish similarities and differences amongst the various genetic backgrounds. Given that osteochondromas are most prominent and identifiable in ribs and long bones in mouse models based on previous studies, we focused on these skeletal elements to analyze the phenotypes using μCT. As to be expected, no osteochondromas had developed in conditional heterozygous Ext1-deficient mice in ribs (Figure 1A) or long bones (Figure 2A), with their skeletal elements being essentially identical to those in wild type mice (not shown).⁴⁷ Prominent osteochondromas had instead formed and accumulated in conditional homozygous Ext1-deficient littermates in which the tumors were appreciable at the costochondral junction in ribs (Figure 1B) and along the epiphyseal portions of femur and tibia (Figure 2B, arrowheads), confirming the disease severity in this model. The corresponding portions in control unaffected heterozygous Ext1-deficient mice were smooth and even (Figure 2A, arrowheads). When we examined homozygous Ext1-deficient mice bearing also a monoallelic and biallelic deletion of Hpse, we did not observe any major effect on osteochondroma formation as revealed by μCT (Figure 1C,D; Figure 2C,D), with the osteochondromas appearing to be similar in overall appearance and frequency to those seen in ribs and long bones in companion Ext1-deficient mice (Figure 1B and Figure 2B). Because the ablation of Hpse had no obvious effects, it was difficult to compare μCT data. Thus, we quantified tumor surface area in Ext1-deficient versus double Ext1;Hpse-deficient mice by histomorphometry (see below).

FIGURE 1.

Analysis of osteochondroma formation in ribs of control and mutant mice. (A) Representative μCT image of rib cage from control heterozygous Ext1^+/f;Agr-CreER mice that were sacrificed 6 weeks after tamoxifen injection. White box area on left panel identifies the osteochondral junction in neighboring ribs shown at higher magnification on the right panel. Note the characteristic asymmetric bone tissue flanking the growth plate that is translucent by μCT (arrowheads). (B-C) Rib cage images from standard Ext1-deficient mice (B, Ext1^f/f;Agr-CreER) and Ext1-deficient mice in heterozygous Hpse-null background (C, Ext1^f/f;Agr-CreER;Hpse^+/− ). Note the conspicuous osteochondromas around the osteochondral junction in all mice, best appreciable in the higher magnification panels on the right (B,C, arrowheads). (D) Ribs from mutant Ext1^f/f;Agr-CreER;Hpse^−/− mice displaying prominent osteochondromas despite homozygous genetic ablation of Hpse. Scale bar: 1 mm. n = 3 Ext1^+/f;Agr-CreER , n = 3 Ext1^f/f;Agr-CreER, n = 3 Ext1^f/f;Agr-CreER;Hpse^+/− and n = 5 Ext1^f/f;Agr-CreER;Hpse^−/−.

FIGURE 2.

Osteochondroma formation in the knees in control and mutant mice. (A) Representative μCT image of the knee from control heterozygous Ext1^+/f;Agr-CreER mice that were sacrificed 6 weeks after tamoxifen injection. Red arrowheads point to normal anatomical characteristics of femur and tibia. (B-C) Representative images of knees from conditional Ext1-deficient mice without and with heterozygous Hpse ablation (Ext1^f/f;Agr-CreER and Ext1^f/f;Agr-CreER;Hpse^+/− mice) 6 weeks after tamoxifen injection, respectively. Note prominent osteochondromas protruding from femur and tibia depicted by red arrowheads in all mice. (D) Representative images of knees from 2 compound Ext1-deficient mice with homozygous Hpse ablation (Ext1^f/f;Agr-CreER;Hpse^−/− mice) 6 weeks after tamoxifen injection. Note the large osteochondromas along femur and tibia (red arrowheads) that appear quite similar to those in mutant knees shown in B-C. Scale bar: 1 mm. n = 3 Ext1^+/f;Agr-CreER, n = 3 Ext1^f/f;Agr-CreER, n = 3 Ext1^f/f;Agr-CreER;Hpse^+/− and n = 5 Ext1^f/f;Agr-CreER;Hpse^−/−.

3.3. Osteochondromas display prominent cartilaginous outgrowth in single and double mutant mice

As indicated above, osteochondroma development starts with formation of ectopic cartilage flanking the growth plate, followed by endochondral ossification. To complement the above μCT analysis of mineralized tissues in mutant mice, ribs and long bones harvested from mouse groups at the 6-week time points as above were decalcified and processed for histochemical staining with Safranin O to delineate presence, structure and organization of cartilaginous tissues in relation to conditional Ext1 ablation and Hpse co-ablation. As shown in Figure 3, ribs and long bones collected from unaffected conditional heterozygous Ext1-deficient mice showed a clear, normal and well defined border between growth plate and adjacent perichondrial and periosteal tissues, with the latter containing typical flat-shaped fibroblastic cells best appreciable at higher magnification (Figure 3E,M). In comparison, the rib border in companion homozygous Ext1-deficient mice was occupied by numerous ectopic cartilage cells that were arranged in growth plate-like structures and protruded away into surrounding tissues (Figure 3B,F). The femurs of these mutant mice displayed similar large osteochondromas some of which occupying, and infringing into, the joint space (Figure 3J,N). When we analyzed homozygous Ext1-deficient mice with monoallelic and biallelic deletion of Hpse, we found that ectopic cartilage formation was as advanced and severe as in companion Ext1-deficient mice and was so in both ribs and femurs (Figure 3C,G,K,O and Figure 3D,H,L,P). The data also indicated that there were no obvious differences in chondrocyte organization into growth plate-like structures amongst the different mutants. The surface area of the rib osteochondromas from homozygous Ext1-deficient mice and homozygous Ext1-deficient mice with monoallelic and biallelic deletion of Hpse were quantified by ImageJ (Figure S2).

FIGURE 3.

Histochemical examination of osteochondromas in ribs and long bones. (A-D) and (E-H) Longitudinal sections of osteochondral rib junctions in control and mutant mice at 6 weeks post injection stained with safranin O and fast green. Box areas in A-D are shown at higher mag in E-H. In control heterozygous Ext1^+/f;Agr-CreER mice, note that the border along growth plate and perichondrium is even and normal (A,E). However, in homozygous Ext1-deficient mice bearing heterozygous or homozygous Hpse loss (B-D and F-H), the border displays prominent osteochondromas in all genotypes. (I-L) and (M-P) Longitudinal sections of femurs in control and mutant mice at 6 weeks post injection stained with safranin O and fast green. In controls (I,M), the growth plate and perichondrium are typical and smooth. However, in all mutants (J-L and N-P), there are prominent osteochondromas protruding out of growth plate. Boxed areas in I-L are shown at higher mag in M-P. Scale bar: 100 μm. n = 3 Ext1^+/f;Agr-CreER, n = 3 Ext1^f/f;Agr-CreER, n = 3 Ext1^f/f;Agr-CreER;Hpse^+/− and n = 3 Ext1^f/f;Agr-CreER;Hpse^−/−.

3.4. Stochastic model HME mice are responsive to drug treatment

The lack of major effects by Hpse ablation on osteochondroma development was quite unexpected based on heparanase’s reported roles in mechanisms of tumor formation in several other systems.³⁵ It should be noted that previous studies on potential therapeutic strategies in HME utilized traditional floxed Ext1 HME mice bearing standard head-to-tail loxP sites, eliciting effective Ext1 ablation in all cells targeted by Cre action.^{47; 48} However, the HME model mice used here are stochastic, resemble human HME pathogenesis more closely, but have not been used for testing possible therapeutic strategies. Thus, it became important to clarify whether these mice are actually amenable and responsive to treatment and do not have some intrinsic resistance to it because of their lower degree of Ext1 ablation after Cre action. Accordingly, we tested whether osteochondroma formation in these stochastic model HME mice could be inhibited by systemic administration of Palovarotene, a drug we previously identified for prevention of heterotopic ossification (HO) in mouse models of the severe pediatric disorder Fibrodysplasia Ossificans Progressiva.⁴⁹ HO and osteochondromas share the basic fact that both pathologies start with ectopic cartilage formation followed by endochondral ossification.⁵⁰ Stochastic Ext1^f/f;Agr-CreER mice at 5 weeks of age were given a single tamoxifen injection as above and were then treated with Palovarotene (3 mg/kg)^{49; 51} or vehicle by oral gavage (5 days/week) for a total of 6 weeks. To assess effects on osteochondroma formation, we collected ribs and long bones from drug- and vehicle-treated mutant mice and subjected them to μCT. Both ribs and long bones in vehicle-treated Ext1-deficient mutant mice displayed the expected significant number of osteochondromas (Figure 4A,E; red arrowheads). Instead, there was a clear decrease in osteochondroma formation in Palovarotene-treated mutant mice in both ribs (Figure 4B–D, red arrowheads, triplicate samples) and long bones (Figure 5F–H, red arrowheads, triplicate samples). Following μCT analysis, the samples were decalcified and processed for sectioning and staining with Safranin O. Palovarotene treatment had reduced to the development of the cartilaginous portion of osteochondromas in ribs, tibia and femur (Figure 5B,C; 5E,F; and 5H,I, respectively) compared to vehicle-treated mutant mice (Figure 5A,D,G) albeit to varying degrees from mouse to mouse but without causing apparent changes or closure of adjacent endogenous growth plates (Figure 5). Measurements of surface area by imaging in specimens from companion untreated and treated mice indicated that the decrease was over 50% and statistically significant (Figure S3).

FIGURE 4.

Analysis of osteochondroma formation after Palovarotene administration. (A-D) Representative μCT images of rib cages from vehicle- and Palovarotene-treated Ext1-deficient (Ext1^f/f;Agr-CreER) mice at 6 weeks after tamoxifen injection. In vehicle-treated mouse (A), large typical osteochodromas are present at the costochondral junction, better visible in white boxed area shown at higher magnification in lower panel (red arrowheads). In companion Palovarotene-treated mice (B-D), there is a clear decrease in osteochondroma formation best appreciable in higher magnification images (lower panels, red arrowheads). Data are from 3 mice to demonstrate consistency of observations. (E-H) Representative μCT images of knees from vehicle- and Palovarotene-treated Ext1-deficient (Ext1^f/f;Agr-CreER) mice at 6 weeks after tamoxifen injection. Prominent osteochondromas in femur and tibia in vehicle-treated mice (E, red arrowheads) are clearly decreased after Palovarotene treatment (F-H, red arrowheads). Data are from 3 mice to demonstrate consistency of observations. Scale bar: 100 μm. n = 4 vehicle-treated and n = 5 palovarotene-treated mice per group.

FIGURE 5.

Histochemical analysis of osteochondroma formation after Palovarotene administration. (A,D,G) Sections stained with safranin O and fast green shown at higher mag in lower panels. Note the sizable osteochondromas present at the rib osteochondral junction (A, arrowheads) and along tibia and femur growth plates (D,G, arrowheads) in vehicle treated Ext1-deficient mice at 6 weeks after tamoxifen injection. (B-C, E-F and H-I) Sections stained with safranin O and fast green shown at higher mag in lower panels. Note the reduced sizes of cartilaginous osteochondroma areas after Palovarotene treatment in companion Ext1-deficient mice. Data from two mice are shown to demonstrate overall consistency of response, albeit with some variability from mouse to mouse. Scale bar: 100 μm.

n = 4 vehicle-treated and n = 5 Palovarotene-treated mice per group.

4. DISCUSSION

Given the essential roles HS and HS-rich proteoglycans play in physiology and pathology, the synthesis, structure, distribution and roles of these macromolecules remain at the center of much research activity.^{13; 15} This is reflected also by the number of basic and translational studies on enzymes that metabolize and functionally modify extracellular HS including heparanase and SULF1 and SULF2 and on means by which these activities could be targeted therapeutically.^{13; 35; 42} In this vein, the original report of excessive heparanase in human osteochondromas from HME patients at both protein and RNA levels was thus similarly indicative of potential pathogenic role by this enzyme.⁴³ We confirmed the data subsequently and found also that heparanase was particularly abundant along growth plate perichondrium where the osteochondromas form and grow.^{44; 52} In addition, we showed that exogenous recombinant heparanase greatly stimulated pathways such as the bone morphogenetic protein (BMP) signaling likely by releasing HS-bound BMPs. This in turn led to a marked stimulation of chondrogenesis and cartilage formation that are the first differentiation steps in osteochondroma development.^{44; 52} Together, these and other observations did suggest that heparanase could be an important culprit in osteochondroma pathogenesis in HME but the data reported here do not lend support to that thesis, at least in the aggressive mouse model used and under the experimental regimens followed here. Thus, the current data do not explain what the excess heparanase levels observed in patient osteochondromas may signify and imply, but the following possibilities can be considered in relation to aspects of HME pathology and severity.

It has long been appreciated that HME disease severity varies greatly from patient to patient, causing minor if any symptoms in some patients to causing severe, debilitating skeletal malformations, joint ankyloses, chronic pain and additional limitations in other patients.^{5; 53–55} Perplexingly, this range of severity can occur even within siblings bearing the same germline heterozygous mutation in EXT1 or EXT2. Indeed, over 700 independent mutations in either EXT gene have been identified so far in HME patients worldwide; while they invariably are all loss-of-function missense, nonsense, frame shift or splice-site mutations, they have proven to be not useful in establishing definitive genotype-phenotype correlations and predicting disease course and severity.^{12; 29; 56} To this day then, the basis for HME pathological variability remains a mystery, but many hypotheses have been put forward, including patient genetic background, disease modifiers and interactions amongst unrelated genes.⁵⁷ While the data here indicate that Hpse is not a key player in osteochondroma formation in the aggressive mouse model used and its ablation cannot protect against disease progression, the data do not exclude the possibility that heparanase could still be a contributor to disease variability and severity in patients. Presence of excess heparanase was found in the original study by Trebicz-Geffen et al. and confirmed in ours,^{43; 52} and together these studies examined about 11 patients. However, there was no attempt to relate heparanase levels and/or activity to disease severity given the low number of patients. Hence, a dedicated and much larger study could be carried out to address the possibility of a direct relationship between enzyme levels and disease severity. It should be pointed out that osteochondromas in patients are not composed of tumor cells only that have all undergone an additional somatic mutation such as LOH, but are a mixture of tumor cells and normal cells, the latter bearing only the original germline heterozygous EXT mutation and recruited into tumor growth.^{11; 22} Thus, enzymes such as heparanase could have roles in facilitating recruitment of these more normal cells into tumor growth by reducing their HS levels, exacerbating the disease phenotype in some patients. This mechanism may be less important in mouse models where the Cre drivers simultaneously ablate both floxed Ext alleles in most/all cells engaged in osteochondroma formation, offering one explanation for the observed lack of effects of Hpse ablation reported here.

The basic pathogenic defect in HME is the deficiency of HS, and the ability of such deficiency to cause so many skeletal health problems in most patients reflects the multifaceted and essential roles this glycosaminoglycan has in cell and tissue physiology.¹³ It is widely assumed that the HS deficiency renders HS-binding growth factors and cytokines more broadly distributed and available for biological action, undermining normal cellular and tissue mechanisms and eliciting tumor growth.^{2; 3} Factors normally interacting with -and bound to- HS include BMPs, activins, hedgehogs and fibroblast growth factors, thus representing the most fundamental and powerful growth factors in the body. These factors all possess a specific HS-binding domain containing basic amino acid residues arranged in Cardin-Weintraub motifs.^{39; 58} In recent studies, we showed that these domains are not only distinct in different proteins, but are located in specific locations, including the N- or C-terminal ends of the mature protein or in different regions within the pro-protein.^{59; 60} Impressively, these differences are conserved through evolution and for over 300 million years, based on our analysis of genomic databases.³⁹ Reciprocally, HS is not a mere simple polysaccharide made of repeating disaccharide units, but undergoes multiple and sophisticated modifications during its assembly in the Golgi brought about by a large number of enzymes and producing distinct patterns of sulfation and sugar epimerization. The result of this elaborate synthetic process are HS chains containing successive 8–12 saccharide-long segments each of which with distinct modification patterns and diverse protein binding abilities.^{13; 61} It is not clear the extent to which HS chains produced by EXT heterozygous mutant cells within and around osteochondromas are structurally different from those made by wild type cells in terms of average length and sugar modifications, possibly resulting in modified protein binding properties.^{17; 31} However, what is well established is that heparanase is unique in its ability to cleave HS chains into sizable fragments which in turn can exert distinct biological activities including: ferrying growth factors within the surrounding environment; maintaining factor-factor dimeric interactions and activity; facilitating factor binding to respective cell surface receptors; and modulating factor turnover.^{13; 62} It is such HS fragmentation and ensuing functions that could be facilitated by excessive heparanase levels in osteochondromas, influencing tumor growth and pathogenic severity in certain patients. Future studies could tackle these and related questions that when answered, will have important basic and translational medicine significance and relevance.

Supplementary Material

fS1

Figure S1 Strategies for construction and verification of Hpse-null mice. (A) Schematic of the exon-intron structure of mouse Hpse gene. The red box indicates the about 0.9 kb region containing the proximal promoter and exon 1 that was deleted by CRISPR/Cas9. (B) Nucleotide sequence of the 0.9 kb region (in blue) and amino acids encoded by exon 1 that were successfully deleted in founder mouse. The 4 nucleotides highlighted in purple are the 5’ protospacer adjacent motif (PAM) downstream of the target DNA sequence (a portion of which is in green) cleaved by Cas. (C) Electrophoretic analysis of genomic DNA amplification products. Primers flanking the targeted region elicited a 1257 bp product in wild type (+/+) and heterozygous (+/−) mice and a 352 bp product in heterozygous and homozygous mutant (−/−) mice (left 3 lanes after DNA ladder). The 3 lanes on the right are amplification products obtained with primers from within the targeted region and eliciting a 485 bp product in wild type and heterozygous mice only. (D) Electrophoretic analysis of amplification products using whole RNA extracted from liver of wild type (WT) and 2 knock out (KO) mice. Three different primer pairs were used to assess different regions (5’, middle, and 3’) of Hpse transcript that elicited the expected 332, 330 and 331 bp products, respectively, in WT sample only. Ribosomal protein, L19, was used as an internal control, and its transcript generated a 418 bp product with RNA samples from both WT and KO mice.

fS2

Figure S2 Ablation of Hpse has no effect on osteochondroma development. Scatterplot showing the surface area of multiple osteochondromas from ribs. Each point represents a rib osteochondroma from a total of 3 mice per group. Data were analyzed by ANOVA. p = 0.755 for Ext1^f/f;Agr-CreER vs. Ext1^f/f;Agr-CreER;Hpse^-/−. p = 0.6750 for Ext1^f/f;Agr-CreER vs. Ext1^f/f;Agr-CreER;Hpse^+/−. p = 0.9706 for Ext1^f/f;Agr-CreER;Hpse^−/− vs. Ext1^f/f;Agr-CreER;Hpse^+/−.

fS3

Figure S3 Palovarotene reduces osteochondroma formation. Scatterplot showing the surface area of multiple osteochondromas from ribs. Each point represents a rib osteochondroma from a total of 4 vehicle-treated mice and 5 palovarotene-treated mice per group. Data were analyzed by ANOVA. p = 0.0001.