Chop (Ddit3) Is Essential for D469del-COMP Retention and Cell Death in Chondrocytes in an Inducible Transgenic Mouse Model of Pseudoachondroplasia

Abstract

Cartilage oligomeric matrix protein (COMP), a secreted glycoprotein synthesized by chondrocytes, regulates proliferation and type II collagen assembly. Mutations in the COMP gene cause pseudoachondroplasia and multiple epiphyseal dysplasia. Previously, we have shown that expression of D469del-COMP in transgenic mice causes intracellular retention of D469del-COMP, thereby recapitulating pseudoachondroplasia chondrocyte pathology. This inducible transgenic D469del-COMP mouse is the only in vivo model to replicate the critical cellular and clinical features of pseudoachondroplasia. Here, we report developmental studies of D469del-COMP-induced chondrocyte pathology from the prenatal period to adolescence. D469del-COMP retention was limited prenatally and did not negatively affect the growth plate until 3 weeks after birth. Results of immunostaining, transcriptome analysis, and qRT-PCR suggest a molecular model in which D469del-COMP triggers apoptosis during the first postnatal week. By 3 weeks (when most chondrocytes are retaining D469del-COMP), inflammation, oxidative stress, and DNA damage contribute to chondrocyte cell death by necroptosis. Importantly, by crossing the D469del-COMP mouse onto a Chop null background (Ddit3 null), thereby eliminating Chop, the unfolded protein response was disrupted, thus alleviating both D469del-COMP intracellular retention and premature chondrocyte cell death. Chop therefore plays a significant role in processes that mediate D469del-COMP retention. Taken together, these results suggest that there may be an optimal window before the induction of significant D469del-COMP retention during which endoplasmic reticulum stress could be targeted.

Cartilage oligomeric matrix protein (COMP) is an extracellular matrix (ECM) protein found primarily in cartilage and other musculoskeletal tissues.^1–10 The COMP gene (previously PSACH; also known as TSP5) is the fifth member of the thrombospondin family and is a homopentamer.^1–7 It is composed of a N-terminal pentamerization domain, four EGF-like repeats, seven type II calcium-binding repeats, and a C-terminal globular domain.⁷ COMP has several potential functions including interacting with other ECM molecules in the matrix, regulating type II collagen fibril assembly, and enhancing chondrocyte attachment and proliferation.^11–17

Mutations in COMP, which result in protein misfolding and intracellular retention, cause two autosomal dominant skeletal dysplasias: pseudoachondroplasia, a severe dwarfing condition, and multiple epiphyseal dysplasia (MED/EDM1), a milder disorder.^18–23 Children with pseudoachondroplasia have normal birth parameters, but develop a waddling gait and have decelerating linear growth by 2 to 3 years of age.^24,25 Bone abnormalities develop over time and include shortened limbs, brachydactyly, and pectus carinatum.^24,25 Early-onset osteoarthritis, affecting all of the joints, causes chronic pain, often necessitating hip replacement in early adulthood.²⁵ Although natural history studies have defined the clinical progression of the pseudoachondroplasia phenotype, little is known about the mechanism or mechanisms underlying the chondrocyte cellular phenotype that causes the diminished bone growth. Most information about the in vivo state is derived from bone biopsies of the pseudoachondroplasia growth plate or from in vitro tissue culture studies, and thus provides little mechanistic information.

Growth-plate chondrocytes from both pseudoachondroplasia and MED/EDM1 patients show giant rough endoplasmic reticulum (rER) cisternae containing COMP and other ECM proteins [collagen types II and IX and matrilin 3 (MATN3)] that form an ordered matrix network.^2,13,19–40 The inability of mutant COMP to fold properly traps the protein in the ER, thereby allowing intracellular matrix assembly.^23,35,36,41 These matrix-filled cisternae occupy most of the chondrocyte cellular volume, ultimately compromising chondrocyte cellular function and causing premature chondrocyte cell death.^{13,18,19,27,32,35,36,39,41}

Earlier pseudoachondroplasia mouse models have not faithfully recapitulated the chondrocyte phenotype and thus have not allowed for analysis of growth plate pathology during skeletal development.^42,43 To circumvent this problem, we generated an inducible transgenic D469del-COMP mouse that we have shown reproduces the characteristic pseudoachondroplasia chondrocyte pathology by 1 month of age, including intracellular retention of COMP, types II and IX collagens, and MATN3, as well as premature intracellular ECM assembly, growth plate abnormalities, and increased numbers of apoptotic chondrocytes.⁴⁴ Here, we define the mechanism of chondrocyte loss and ontogeny of D469del-COMP retention and cell death during prenatal and postnatal skeletal development. Most importantly, we show that, in the absence of Chop, D469del-COMP is not retained and D469del-COMP-induced cell death is eliminated.

Materials and Methods

Generation of Constructs

Two plasmids, pTRE-COMP and pTET-On-ColII, were generated as described previously⁴⁴ and were used to produce bigenic mutant D469del-COMP mice. Briefly, the pTRE-COMP construct contains the coding sequence of the human COMP (+FLAG tag) gene driven by the tetracycline responsive element (TRE) promoter. The pTET-On-ColII construct contains the rtTA coding sequence driven by a type II collagen promoter.⁴⁵

Generation of Bigenic Mice

Standard breeding was used to generate bigenic animals. Genotypes of the transgenic offspring were verified using human COMP and rtTA-specific primers. Mice were pre- and postnatally administered doxycycline (500 ng/mL) through drinking water (supplemented with 5% sucrose). All animal studies were approved by the Animal Welfare Committee of the University of Texas Medical School at Houston.

Generation of Chop Null Bigenic Mice

Chop null mice were procured from Jackson Laboratories and mated with D469del-COMP bigenic mice to obtain a strain expressing D469del-COMP in a Chop null background.⁴⁶ Genotypes of the Chop null mice were verified using Chop-specific primers.⁴⁶

Growth Plate Analysis

Upper tibial growth plates were obtained from at least 10 different mice of each strain. Chondrocytes were counted by marking each chondrocyte using Image-Pro software version 4 (Media Cybernetics, Silver Spring, MD) at ×4 magnification to visualize the entire growth plate. Five measurements along the width of the growth plate were obtained from different limbs, and the measurements were then averaged as previously described.⁴⁷

Immunohistochemistry

Hind limbs of D469del-COMP and C57BL/6 control mice were collected and tibial growth plates were analyzed as previously described.⁴⁴ Briefly, the limbs were fixed in 95% ethanol for immunostaining for COMP [rabbit polyclonal antibody, 1:200 (Kamiya, Seattle, WA) or ab11056-rat antibody, 1:100 (Abcam Cambridge, MA)], FLAG (Abcam, Cambridge, MA), PCNA [93–1143 (Invitrogen, Carlsbad, CA)], AIF [4642, 1:50 (Cell Signaling Technology, Danvers, MA)], and H2AX [9718, 1:100 (Cell Signaling)] or in 10% formalin for TUNEL staining. The COMP rabbit antibody cross-reacts with both endogenous mouse and transgenic human COMP, whereas the COMP rat antibody is specific for human COMP. FLAG specifies the recombinant human D469del-tagged COMP.

Retention Scoring

A scoring strategy was devised to measure the relative amount of D469del-COMP retention in the growth-plate chondrocytes of transgenic mice. Growth plates were processed as described above and as described previously.⁴⁷ Each chondrocyte in the visual field of images at ×400 magnification was scored for retention in 10 separate images. The scorer visually estimated the size of D469del-COMP accumulation area, compared with the area of the nucleus for each chondrocyte. An average score was calculated from these images. Chondrocytes were scored as follows: 1 = no retention, 2 = retention area is less than that of the nucleus, 3 = retention area is the same size as the nucleus, and 4 = retention area is larger than the nucleus in each chondrocyte.

RNA Extraction and Real-Time qRT-PCR

Mice of various ages were collected and soft tissue was removed from the skeleton at age embryonic day 15.5 (E15.5), or knee and elbow joints at postnatal day 1 (P1), or knee joint at all other ages. Tissue was immediately homogenized in TRIzol reagent (Sigma-Aldrich, St. Louis, MO) and RNA was extracted according to the manufacturer's protocol. Quantitative real-time RT-PCR (qRT-PCR) was performed using an ABI 7700 sequence detector (Applied Biosystems, Foster City, CA). DNA was removed using amplification grade DNase (Invitrogen) according to the manufacturer's instructions. Each assay was replicated three times, and each sample was measured in triplicate, including a control without reverse transcriptase. The final data were normalized to Hprt1 × 100 (percentage of the normalizer transcript). qRT-PCR was performed using standard methods and the following primers: Grp78 5′-CATGGTTCTCACTAAAATGAAGG-3′ and 5′-GTAAGCTGGTACAGTAACAACTG-3′; Xbp-1 5′-GTCTGCTGAGTCCGCAGCAGGT-3′ 5′-CAGACTCAGAATCTGAAGAGGC-3′; Edem1 5′-AGTCAAATGTGGATATGCTACGC-3′ 5′-ACAGATATGATATGGCCCTCAGT-3′; Chop 5′-TGCCTTTCACCTTGGAGAC-3′ and 5′-CGTTTCCTGGGGATGAGATA-3′; Ero1B 5′-CGTCCTTTAAATCCTTTGGCG-3′ and 5′-ACACAAACCTTCTAGCCACG-3′; GADD34 5′-AGAGAAGCCAGAATCACCTTG-3′ and 5′-ACTGTGACTTTCTCAGCGAAG-3′; GADD45A 5′-CCGAAAGGATGGACACGGTG-3′ and 5′-TTATCGGGGTCTACGTTGAGC-3′; and Hprt1 (normalizer) 5′-CCTCATGGACTGATTATGGACAG-3′ and 5′-TCAGCAAAGAACTTATAGCCCC-3′.

Microarray Analysis

Total RNA (300 ng) from both hind-limb knee joints of each sample was purified using TRIzol reagent and RNeasy columns (Qiagen, Valencia, CA). RNA was amplified using a total prep RNA amplification kit (Illumina, San Diego, CA). RNA was assessed by RNA integrity number (RIN number), and type II collagen was assessed to verify that both the C57BL/6 and D469del-COMP RNA preparations were primarily from chondrocytes. For each time point, three control and three D469del-COMP animals were examined, to control for variation in expression. In vitro transcription was performed and biotinylated cRNA was synthesized by 14-hour amplification with dNTP mix containing biotin-dUTP and T7 RNA polymerase. Amplified cRNA was subsequently purified and concentration was measured. An aliquot of 1.5 μg of amplified products was loaded onto Illumina Sentrix BeadChip mouse WG-6.v2 arrays, hybridized at 58°C in an Illumina hybridization oven for 17 hours, washed, and incubated with streptavidin-Cy3 to detect biotin-labeled cRNA on the arrays. Arrays were dried and scanned with an Illumina BeadArray reader.

Data were analyzed using Illumina GenomeStudio software. Clustering and pathway analysis were performed with GenomeStudio and Ingenuity Pathway Analysis software (Ingenuity Systems, Redwood City, CA), respectively. A P value of <0.05 was considered significant.

The mRNAs were classified into functional categories.

Limb Length Measurements

Hind limbs were obtained from 12 control and 12 D469del-COMP mice and the soft tissue was removed. The skeleton was stored in 70% ethanol and then subjected to radiographical examination. The limbs were attached to a grid to ensure consistency and accuracy of measurements. Measurements were made from end to end on both long bones in the hind limb, using Image-Pro software, as previously described.⁴⁷ A t-test was used to compare tibial measurements from control and D469del-COMP mice limbs.

Skeletal Preparations

Mice were collected and the soft tissue removed and then fixed overnight in 95% ethanol. Skeletons were stained for 48 hours with Alcian Blue (0.015% solution in 20% acetic acid and 80% methanol), then transferred to 95% ethanol for 2 hours, and then to 2% KOH for 3 hours. The mineralized bone was then stained with Alizarin Red (0.005% in 1% KOH) for 3 hours. The skeleton was cleared in 1% KOH in 20% glycerol for 48 hours and then stored in a glycerol ethanol mixture (1:1).

Results

D469del-COMP Growth Plates Become Disorganized and Thinner during Juvenile Development

We first asked whether the presence of D469del-COMP in our transgenic mouse causes growth plate abnormalities over time. The growth plate organization was similar in D469del-COMP and control mice until 2 weeks (P14; Figure 1, A–C and G–I). At 3 weeks (P21; Figure 1, D and J), more acellular gaps were seen and the growth plate was thinner than control (Figure 1, M and N). By 4 weeks, the growth plate continued to be thinner, compared with control, and loss of chondrocytes was evident (Figure 1, E, K, M, and N). At 2 months, when skeletal growth is complete, the D469del-COMP growth plate had very few chondrocytes, unlike the control (Figure 1, F and L). Changes in the D469del-COMP growth plate are attributed to expression of D469del-COMP rather than to doxycycline exposure, because all mice were subjected to the same doxycycline treatment. These results indicate that D469del-COMP expression causes loss of growth-plate chondrocytes, which is evident by 3 weeks of age. There is little change in the number of chondrocytes and growth plate height before this developmental time point.

Figure 1.

Histomorphometric analysis of D469del-COMP growth plates from E15 to 2 months of age. H&E staining of growth plates at various ages in control C57BL/6 (A–F) and D469del-COMP (G–L) mice. D469del-COMP and control growth plates are similar until P14 (D and J), when the growth plates of D469del-COMP mice are more disorganized. Original magnification, ×20 (A–L). M: Quantification of growth plate width. Growth plates from D469del-COMP mice are narrower than those of controls at P21, 1 month, and 2 months of age. N: Quantification of number of chondrocytes. Fewer chondrocytes are present in the D469del-COMP growth plate, compared with controls, at 1 and 2 months of age. *P ≤ 0.05, **P ≤ 0.01.

D469del-COMP Retention Is First Evident in Prenatal Growth-Plate Chondrocytes

We next asked when D469del-COMP becomes retained in the chondrocytes of the developing growth plate. Previously, we showed that D469del-COMP is retained in the rER and forms an ordered intracellular matrix composed of collagens types II and IX and MATN3 in growth plates from 1-month-old mice, identical to what is observed in pseudoachondroplasia patient chondrocytes.^21,44 In these studies, a FLAG antibody recognizes the transgenic D469del-COMP-FLAG protein and a COMP antibody recognizes both the endogenous murine and transgenic human COMP-FLAG. Few chondrocytes showed intracellular retention at E15, compared with control (Figure 2, A and G). Both the D469del-COMP and control chondrocytes show a patchy distribution of COMP (Figure 2, A and G). Intracellular retention was measured using the scoring methodology described under Materials and Methods and presented schematically in Figure 2M (an example of D469del-COMP growth-plate chondrocytes with a retention score of 4 is presented in Figure 2N). The intracellular retention increases, with many chondrocytes showing retention by P3; almost all chondrocytes are retaining D469del-COMP by P21 (Figure 2, B–D and O); In comparison, in the nontransgenic controls, all endogenous COMP is extracellular (Figure 2, G–L). This indicates that intracellular retention is a progressive process with the maturing of the growth plate and eventually encompasses all chondrocytes in the growth plate.

Figure 2.

Intracellular retention of COMP in growth plates of D469del-COMP mice from E15 to 2 months of age. Immunostaining of D469del-COMP (A–F) and control C57BL/6 (G–L) growth plates with FLAG (green) and COMP (red) antibodies. Patchy intracellular retention (yellow indicates merged image) begins before birth at E15; progressive intracellular D469del-COMP accumulation is observed with increasing age. Control C57BL/6 growth plates show no intracellular retention of COMP. Original magnification, ×20 (A–L). M: Scoring methodology. A numerical score was assigned by visually comparing the cumulative size of D469del-COMP-positive rER cisternae with the area of the nucleus. C, retention; N, nucleus. N: Two examples of D469del-COMP growth-plate chondrocytes with a retention score of 4. O: Quantitation of progressive intracellular D469del-COMP in growth plates from E15 to 2 months of age. *P ≤ 0.05.

To quantitate the progression of D469del-COMP retention, we developed a scoring system based on the area of the COMP-positive intracellular retention relative to the area of the nucleus as defined by DAPI staining (Figure 2B). This system was designed to compare the relative amount of D469del-COMP retained in growth-plate chondrocytes at different developmental ages. There was a significant increase in D469del-COMP retention starting shortly after birth (P3), compared with that seen at E15.5 (Figure 2B). D469del-COMP retention continued to increase significantly up to P14, at which point the levels remained relatively constant, suggesting that a steady state level had been attained.

Cell Death Is Detected Only at P21 When Significant Intracellular D469del-COMP Retention Is Present

Previously we showed that expression of D469del-COMP in the mouse growth plate induces premature cell death.⁴⁴ Here, we define the onset and distribution of cell death using TUNEL staining (Figure 3, G–L and S–X). DAPI was used to localize the nucleus of each chondrocyte (Figure 3, A–F and M–R). Little or no cell death was seen in the D469del-COMP and control growth plates until P14, when it was restricted to the hypertrophic zone (Figure 3). A significant increase in cell death affecting chondrocytes throughout the D469del-COMP growth plate was seen at P21 and older ages (Figure 3, J–L). In contrast, in the control growth plates cell death was limited and restricted to the hypertrophic zone (Figure 3, V–X). Cell death correlated with the presence of significant accumulation of D469del-COMP in the growth-plate chondrocytes (Figure 3, J–L, and Figure 2, D–F). Levels of cell death remained high at 1 and 2 months of age (Figure 3, K, L, and Y), corresponding to fewer chondrocytes in the growth plate (Figure 1M). These results suggest that the level of intracellular retention of D469del-COMP induces cell death, which translates into significant loss of growth-plate chondrocytes.

Figure 3.

Chondrocyte cell death in D469del-COMP and control growth plates from mice at ages E15 to 2 months. TUNEL and DAPI staining of D469del-COMP (A–L) and control C57BL/6 (M–X) growth plates. Cell death in D469del-COMP and control growth plates is equivalent until P21, when D469del-COMP cell death is increased, compared with control (J–L and V–X). Original magnification, ×20 (A–X). Arrowhead indicates increase relative to control. Y: Quantitation of TUNEL-positive chondrocytes in growth plates from P21 to 2 months of age. Cell death in D469del-COMP growth plates is threefold higher at P21, fourfold higher at 1 month, and twofold higher at 2 months of age, compared with controls. **P ≤ 0.01.

Long Bones Are Significantly Reduced in D469del-COMP Mice

Finally, we assessed how loss of growth-plate chondrocytes affects the mouse skeleton. Radiographical measurements showed that the D469del-COMP tibia was wider and 12% shorter than the control tibia (Figure 4). The skeleton of D469del-COMP mice was smaller than that of the controls (Figure 5, A and B). The skull, snout, fore limbs, hind limbs, and phalanges of the D469del-COMP mouse were reduced, compared with controls (Figure 5, C, E, G, I, and K). The rib cage was shortened, and there was a prominent sternum. These results indicate significant reduction in all of the bones derived from endochondral bone formation; the reductions were similar to those observed in pseudoachondroplasia patients.^24,40

Figure 4.

D469del-COMP limbs are shorter than C57BL/6 controls. A and B: Radiographs of hind limbs from 1-month-old C57BL/6 (A) and D469del-COMP (B) mice. C: Scatter plot of individual measurements. Horizontal bars indicate means, and error bars indicate ± SD. D469del-COMP mice have a 12% decrease in tibial length (P < 0.000001). Tibial measurements were taken from ≥12 limbs and the average length was calculated. A t-test was used to compare D469del-COMP tibial length to control.

Figure 5.

Skeletal elements of D469del-COMP and C57BL/6 control mice. A and B: Whole skeletal preparations of C57BL/6 control (A) and D469del-COMP (B) mice showing cartilage (Alcian Blue) and bone (Alizarin Red). C–F: Lateral and superior images of the skull of C57BL/6 (C and E) and D469del-COMP (D and F) mice. G–J: Fore and hind limbs of C57BL/6 (G and I) and D469del-COMP (H and J) mice. K and L: Front paws of C57BL/6 (K) and D469del-COMP (L) mice. Red and blue bars reflect skull measurements of a C57BL/6 mouse, for comparison with D469del-COMP mouse.

D469del-COMP Expression Induces a Proapoptotic Environment before D469del-COMP Retention and Cell Death Is Detected in the Growth Plate

To define the molecular mechanisms that contribute to intracellular retention and chondrocyte cell death, we analyzed the transcriptome from P1 and from 1-, 2-, 3-, and 4-week-old hind-limb knee joints from transgenic and control mice, using a microarray. Proapoptotic transcripts were modestly elevated, and antiapoptotic transcripts were slightly down-regulated at P1 (Table 1). Specifically, the transcripts from four prosurvival genes were expressed at lower levels in D469del-COMP P1 mice: Bag3 (0.8×), Bcl2 (0.6×), Bcl2l1 (0.5×), and Bcl2l2 (0.7×). In contrast, proapoptotic transcripts from five genes were up-regulated: Casp8 (1.3×), Bad (1.6×), Bax (1.3×), and Chop (1.4 to 1.6×). Generally, expression of genes encoding proteins involved in generation of antioxidants and ROS were increased from 2 to 4 weeks (Table 1). Inflammatory gene expression spiked at 3 weeks of age in the D469del-COMP mice, which correlated with high levels of cell death in the growth plate (Figure 3 and Table 1).

Table 1.

Time-Dependent Altered mRNAs Involved in Cell Death, Oxidative Stress, and Inflammation in MT-COMP Mice

Genes	P1	1 week	2 weeks	3 weeks	4 weeks
Apoptosis: antiapoptotic
Apip				2	1.6
Bag3	0.8				0.8
Bcl2	0.6		0.8
Bcl2l1	0.5
Bcl2l2	0.7	1.2			0.8
Bnip3l	1.2				1.2
Bcl2a1b		1.7	0.8		0.6
Bcl2a1d		2	0.8		0.7
Cfl1	1.6	1.3		0.3
Apoptosis: caspase recruitment domain
Casp1				1.5
Casp6					0.8
Casp8	1.3			1.4	0.8
Nol3	1.5
Apoptosis: proapoptotic
Bax		1.3	0.6		0.8
Bad	1.6		0.8
Apip
Chop	1.4	1.6
Antioxidants
Epx				7	2
Ttn	0.5		1.7	0.6
Mt3		1.9
ROS metabolism
Prg3				5	2
Prg2				4	1.7
Angptl7	0.2	3	0.4		0.6
Atox1	1.5
Dusp1	0.6				1.6
Foxm1	1.5
Glrx2				0.5
Sepp1	0.7	0.8		1.6	1.8
Stk25	0.8				0.6
Inflammatory response
Cd40		1.6		1.6
Ly96		1.5
Hdc	0.3			1.5	1.2
Pou2af1				4.2
Cd79b				3.5
Ear2	4			6.2	2
Ear3				5.2	2.6
Ear4				10.4
Ear6				14.6
Irf9	0.5			2.7

Includes articular and growth-plate cartilage.

Other general trends in mRNA expression were found for genes involved in the UPR activation, ubiquitination, proteasome activation, DNA damage repair, and cell cycle regulation (Table 2). Expression of genes involved in protein folding and binding of unfolded proteins was elevated at P1 and depressed at 4 weeks. Ubiquitin ligase (E1) expression was elevated at P1; PSMB8, a component of the proteasome, was expressed at a higher level in 2- and 4-week-old D469del-COMP mice, compared with controls. mRNAs for proteins that bind to damaged DNA were elevated from P1 to 4 weeks, and DNA repair gene mRNAs were increased from P1 to 2 weeks (Chaf1a, Fen1, Polh, and Ung). Transcripts of genes that regulate the cell cycle in response to DNA damage were modestly increased at 2 and 4 weeks (Gadd45a, Smc3, and Rbbp4). Changes in mRNA expression of selected genes of the UPR, including Chop, Gadd34, and Ero1B, were confirmed using qRT-PCR analysis (data not shown). Taken together, these trends indicate that a proapoptotic environment is established early and that oxidative stress, DNA damage, and inflammation collectively promote chondrocyte cellular death. The observed changes in DNA damage response, UPR, and ubiquitin-proteasome mRNAs support this model.

Table 2.

Time-Dependent Expression Trends of Genes Involved in UPR, Ubiquitination, Proteasome, and DNA Damage in MT-COMP Mice

Gene	P1	P7	P14	P21	1 month
Protein folding, binding of unfolded protein	↑				↓
Ubiquitin ligase (E1)	↑
Component of the proteasome PSMB8			↑		↑
Bind to damaged DNA	↑	↑	↑	↑	↑
DNA repair (Chaf1a, Fen1, Polh, Ung)	↑		↑
Cell cycle control in response to DNA damage (Gadd45a, Smc3, Rbbp4)			↑		↑

Includes articular and growth-plate cartilage. ↑ indicates increased expression; ↓ indicates decreased expression.

D469del-COMP Is Not Retained in the Absence of Chop

Microarray transcriptome analysis showed a modest increase in Chop mRNA at P1 and P7 that was validated by qRT-PCR. Because Chop plays a significant role in UPR pathway and cell death, we generated a mouse line that expresses D469del-COMP on a Chop^−/− background (Jackson Laboratory). The growth plates from these mice were analyzed at 1 month for intracellular retention of D469del-COMP, using a human COMP-specific antibody. In the absence of Chop, D469del-COMP was not retained in the chondrocyte, but was observed in a pericellular halo around the chondrocyte (Figure 6, G–I). However, as expected, D469del-COMP was distinctly evident within each chondrocyte in the D469del-COMP growth plate in mice expressing Chop (Figure 6, D–F), in contrast to controls (Figure 6, A and B).⁴⁴ Because the loss of Chop permits D469del-COMP to be exported, we asked whether the export of D469del-COMP alleviates other aspects of the pseudoachondroplasia chondrocyte pathology. We assessed proliferation and cell death by PCNA and TUNEL staining in the Chop^−/− D469del-COMP growth plates at 1 month of age. We found that PCNA and TUNEL staining was significantly reduced in the Chop^−/− D469del-COMP growth plate, compared with levels in the D469del-COMP mouse, but similar to levels in the control (Figure 6, J–O). Several inflammatory-related transcripts that were up-regulated in the D469del-COMP mice were down-regulated by twofold (Ear2, Ear3, Ear4, Ear6, and Irf9) in the Chop^−/− D469del-COMP mice. These data indicate that these inflammatory-related transcripts are involved in the pseudoachondroplasia chondrocyte pathological process, consistent with the proposed model.

Figure 6.

COMP immunostaining of C57BL/6, D469del-COMP, and Chop null D469del-COMP growth plates. DAPI staining marks the nucleus. A human specific COMP antibody (Abcam) was used. A–I: COMP and DAPI staining of C57BL/6, D469del-COMP, and Chop^−/− D469del-COMP growth plates of 1-month-old mice. D469del-COMP was retained in all D469del-COMP chondrocytes (D–F) but had a pericellular distribution in the absence of Chop (G–I). J–L: PCNA immunostaining of growth plates of 1-month-old C57BL/6 (J), D469del-COMP (K), and Chop^−/− D469del-COMP (L) mice. Decreased PCNA immunostaining is seen in D469del-COMP, compared with control and Chop^−/− D469del-COMP mice. M–O: TUNEL immunostaining of growth plates of 1-month-old C57BL/6 (M), D469del-COMP (N), and Chop^−/− D469del-COMP (O) mice. Increased number of apoptotic chondrocytes are found in D469del-COMP, compared with control and Chop^−/− D469del-COMP growth plates. Original magnification, ×40 (A–I), ×20 (J–O).

To further investigate the effects of the absence of Chop on ER stress genes, the expression of Chop, Xbp1, Gadd45a, Gadd34, Ero1b, Grp78, and Edem1 was measured. D469del-COMP expression up-regulated the transcript level of Chop, Xbp1, Gadd45a, Ero1b, and Grp78 in mice at 1 month of age, which was consistent with the microarray findings (Table 1 and Figure 7). In Chop^−/− D469del-COMP mice, the expression of Xbp1, Gadd45a, and Gadd34 was reduced, compared with D469del-COMP mice (Figure 7). These findings show that Chop plays a significant role in regulating the UPR genes that are up-regulated in D469del-COMP chondrocytes.

Figure 7.

Loss of Chop reduces D469del-COMP-induced cellular stress. Quantitation of Chop, Xbp1, Gadd45a, Gadd34, Ero1b, and Grp78 mRNAs from hind-limb joints of 1-month-old C57BL/6 (control), Chop^−/−, D469del-COMP, and Chop^−/− D469del-COMP mice. D469del-COMP increases expression of specific ER stress mRNAs. The absence of Chop in the presence of D469del-COMP lowers the expression of these stress genes. *P < 0.05.

D469del-COMP Increases Cell Death-Inducing Factor (AIF) and Phosphorylated Histone H2AX (γH2AX) Associated with Necroptosis

Our in vitro analysis of the UPR, stimulated by D469del-COMP, showed up-regulation of markers of necroptosis.⁴⁸ Necroptosis, a form of programmed death mediated through caspase-independent mechanisms, involves rupture of the plasma membrane.^49–52 To determine whether the same mechanisms occur in the D469del-COMP growth plate, we assessed the level of AIF and γH2AX, which are associated with necroptosis. AIF is increased in the D469del-COMP growth plate at 1 month, compared with control (Figure 8, A–L). Additionally, cleaved AIF necessary for necroptosis execution was observed in the nucleus of D469del-COMP chondrocytes but not in controls. Moreover, γH2AX (the binding partner of AIF) was up-regulated in the D469del-COMP growth plate (Figure 8, M–O). Cleaved caspase 3 was not detected in either D469del-COMP or control growth plates (data not shown). Taken together, these findings indicate that D469del-COMP induces cell death in the growth-plate chondrocytes by necroptosis.

Figure 8.

AIF and γH2AX immunostaining of C57BL/6 and D469del-COMP growth-plate chondrocytes. AIF and γH2AX antibodies were used to localize these necroptotic proteins; DAPI staining marks the nucleus. Dashed outlines mark the nucleus, regardless of stain. A–L: AIF and DAPI staining of C57BL/6 (A–F) and D469del-COMP (G–L) growth plates of 1-month-old mice. AIF was observed in the nucleus of D469del-COMP chondrocytes (J–L), but only in the cytoplasm of the controls (D–F). Asterisks mark AIF localization. M–O: γH2AX staining of C57BL/6 (M) and D469del-COMP (N–O) growth plates of 1-month-old mice. γH2AX was elevated in D469del-COMP growth plate, compared with control. Scale bar = 50 μm (A–C and G–I). Boxed area locates the image of a single nucleus, shown at higher magnification in the next row. Original magnification, ×20. Inset cells enlarged digitally.

Discussion

COMP plays an important role in homeostasis of the growth plate and articular cartilage, but the exact function or functions still remain to be defined.^{11–13,16,17,53,54} Absence of COMP causes mild perturbations in the growth plate, whereas D469del-COMP causes catastrophic effects, such as premature loss of chondrocytes, which translates into pseudoachondroplasia, a well-characterized form of dwarfism.^22,23,55 Despite cross-sectional information from iliac crest biopsies about growth plate disturbance in humans and transgenic mice expressing mutant COMP, the pathological processes in growth-plate chondrocytes resulting from D469del-COMP retention have not been characterized during development and growth. In the present study, we used doxycycline-inducible transgenic mice that express D469del-COMP in growth-plate chondrocytes to study the effects of D469del-COMP on growth-plate development. We show that, although D469del-COMP retention begins very early, before birth, it does not become pervasive until after birth, at which time it establishes a proapoptotic environment. Later in skeletal maturation, the pervasive D469del-COMP retention was associated with the onset of cell death and expression of mRNAs involved in ROS metabolism, antioxidant function, and inflammation. Importantly, by reducing the UPR Chop-mediated response, D469del-COMP was secreted from the chondrocytes, and the growth plate had a more normal appearance. These results have important implications for therapy.

Babies with pseudoachondroplasia have normal birth parameters, but show decreased linear growth by 2 years of age, and then very little long-bone growth after the age of 6 years. Mouse long-bone growth is essentially complete by 8 weeks, when the growth plate becomes quiescent. Similar to pseudoachondroplasia babies, the D469del-COMP mice appear normal at birth. However, the D469del-COMP mouse at 1 month had a dwarfed appearance, with an abnormal protuberant rib cage, similar to the human pectus carinatum seen in pseudoachondroplasia (Figure 5). There was significant reduction in long bones of the fore and hind limbs, with a 12% reduction in tibial length (Figure 4). Moreover, the digits in the fore limb were short (Figure 5). The face in pseudoachondroplasia is very angular, and the head circumference is normal. Similarly, the D469del-COMP mouse shows a reduction in snout length and a smaller cranial vault (Figure 5). Thus, we show that the D469del-COMP mice manifest most of the skeletal abnormalities of pseudoachondroplasia and truly mimic the human dwarfing condition.

Analysis of the growth plate showed that intracellular retention of D469del-COMP was limited to a few chondrocytes at E15.5. At birth (P1), intracellular retention of D469del-COMP had a patchy distribution in the growth plate that became extensive by P21 (Figure 2). Of note, only at P21 did gross growth plate changes become obvious, with fewer chondrocytes and reduced growth-plate width (Figures 1 and 3). The timing of cell death in the D469del-COMP growth plate followed the temporal pattern of intracellular retention, suggesting that significant retention in most growth-plate chondrocytes was necessary to induce cell death (Figure 3).

To evaluate the processes contributing to chondrocyte cell death, transcriptome analysis was performed on hind-limb joints from mice at ages P1 to 4 weeks. Increases in proapoptotic genes (eg, Chop) and repressions of antiapoptotic factors (eg, Bcl-2 family) were found at P1 and P7, even though there was relatively little D469del-COMP retention. Despite the early apoptotic stimulus, D469del-COMP chondrocytes survived until 3 weeks, when D469del-COMP retention became widespread. Of note, select inflammatory markers showed a dramatic increase at that time point, but subsequently returned to normal values. In the Chop^−/− D469del-COMP mouse, these same inflammatory transcripts were reduced two-fold, indicating that retained D469del-COMP elevates these inflammatory mRNAs but that D469del-COMP in the extracellular matrix does not. An increase in ROS metabolism genes and antioxidant genes coincided with the widespread cell death. ER stress is known to stimulate ROS production and oxidative stress, thereby causing inflammation.^56,57 Excessive and prolonged oxidative stress mediates cell death through caspase-independent mechanisms, and specifically through necroptosis.⁵⁸ Mild oxidative stress leads to caspase cleavage and apoptosis.⁵⁸ AIF, a necroptosis executioner, and its binding partner, γH2AX, were both elevated in D469del-COMP growth-plate chondrocytes (Figure 8). Our results show that at P1 to P7 apoptosis was stimulated when D469del-COMP retention and oxidative stress were minimal. However, necroptosis took over at 3 to 4 weeks, when extensive D469del-COMP retention generated excessive oxidative stress.

Studies in a mouse model of MED/EDM1 caused by a T585M COMP mutation suggested that Chop plays a role in mutant COMP-induced cell death.⁴² This mouse showed little intracellular retention, but had increased apoptosis and decreased chondrocyte proliferation. Similarly, Chop mRNA was increased at P1 and P7 in our D469del-COMP mice, leading us to examine the effect of loss of Chop on the retention of D469del-COMP. Absence of Chop resulted in less retention of D469del-COMP, with export to the ECM (Figure 6). Importantly, the loss of Chop in the Chop^−/− D469del-COMP mice results in an increase in proliferation and decrease in cell death, as assessed by PCNA and TUNEL staining, respectively (Figure 6). Additionally, expression of several stress-related genes were depressed in the Chop^−/− D469del-COMP mouse, compared with the control mice (Figures 6 and 7). These results indicate that Chop-mediated cellular stress is a consequence of D469del-COMP retention, rather than of the presence of D469del-COMP in the extracellular matrix, as was previously suggested.^32,42 This also suggests that the ER quality control systems, particularly the Chop arm, prevent D469del-COMP export, and that by interrupting these sensing mechanisms D469del-COMP can be secreted.

Our unique transgenic mouse, which expresses D469del-COMP in a tissue-specific manner, is a substantial improvement over other models, because it recapitulates both the chondrocyte pathology and the skeletal abnormalities associated with pseudoachondroplasia.^42–44,59 We show the progressive nature of D469del-COMP accumulation and chondrocyte loss, which now explains the deceleration of human long-bone growth in pseudoachondroplasia. Although COMP accumulation at birth is minimal, it steadily increases in all of the growth-plate chondrocytes. Chondrocytes are able to maintain normal function initially, but the D469del-COMP intracellular load subsequently appears to overwhelm the cellular coping systems, leading to cell death. These novel findings suggest a molecular model in which D469del-COMP triggers apoptosis signaling soon after birth (Figure 9). Later, when most chondrocytes are retaining D469del-COMP, oxidative stress leads to inflammation and DNA damage, and finally chondrocyte cell death occurs through necroptosis. Also shown is that Chop plays a significant role in the chondrocyte intracellular pathology. Chop and other proapoptotic mRNAs were elevated from P1 to P7, but antiapoptotic mRNAs were decreased. Despite unfavorable changes in apoptotic and survival mRNAs, growth-plate chondrocytes survived until P21, when all showed significant D469del-COMP intracellular retention and inflammation and ROS stress was triggered. Our results suggest that adverse molecular events occur before clinical diagnosis (at approximately 2 to 3 years of age in humans and by 3 weeks of age in mice). Therapeutics would be most effective if administered very early, before oxidative stress and inflammation responses are mounted. The tolerance of growth-plate chondrocytes to intracellular D469del-COMP may define a therapeutic window in which treatments would be the most effective before the initiation of inflammation and reactive oxygen stress.

Figure 9.

Schematic of D469del-COMP growth plate pathology during maturation of the mouse skeletal. Shortly after birth, there is little D469del-COMP intracellular retention and very little cell death. As retention increases, there is induction of oxidative stress and inflammation markers and abundant cell death occurs throughout the growth plate, rather than being limited to the hypertrophic zone.