Inducible nitric oxide synthase–nitric oxide signaling mediates the mitogenic activity of Rac1 during endochondral bone growth

Growth of endochondral bones is regulated by the coordinated proliferation and hypertrophic differentiation of growth plate chondrocytes (Kronenberg, 2003; Olsen et al., 2000). Cell volume increase during hypertrophic differentiation provides one driving force for bone growth, whereas ordered proliferation is required for the generation of sufficient numbers of cells to undergo subsequent hypertrophy (Beier, 2005). Thus, inhibition of either proliferation or hypertrophy can result in dwarfism. Disruption of these processes by, for example gene mutations, endocrine imbalances and mechanical or nutritional influences, causes numerous diseases. Many growth factors and hormones have been shown to regulate chondrocyte proliferation, including fibroblast growth factors, parathyroid hormone-related peptide and Indian hedgehog (Ballock, 2003; van der Eerden, 2003). However, the intracellular mechanisms controlling chondrocyte proliferation and hypertrophy are much less understood.

Studies by our group and others have shown that Rho GTPases are important regulators of chondrocyte proliferation and differentiation (Kerr et al., 2008; Kumar and Lassar, 2009; Wang and Beier, 2005; Wang et al., 2004; Woods and Beier, 2006; Woods et al., 2009b; Woods et al., 2005; Woods et al., 2007a; Woods et al., 2007b). Rho family GTPases act as molecular switches that become activated in response to specific extracellular stimuli, including growth factors and integrin ligands, through the activity of guanine nucleotide exchange factors that exchange GTP for GDP (Heasman and Ridley, 2008; Vega and Ridley, 2007; Vega and Ridley, 2008). Activated, GTP-bound Rho proteins then interact with a large variety of downstream effectors, including cytoskeletal regulators and various kinases, to control a number of cellular events such as cell migration, cell cycle progression and gene expression. During the early steps of chondrogenesis, regulation of actin dynamics and cellular morphology appear to mediate many of the effects of Rho GTPases (Beier and Loeser, 2010; Woods and Beier, 2006; Woods et al., 2005; Woods et al., 2007a). However, additional mediators are probably involved in these activities of Rho GTPAses, in particular after cells have differentiated into chondrocytes.

We have previously reported that chondrocyte-specific inactivation of the gene Rac1, encoding a Rho family GTPase with a central role in cellular signaling pathways, causes dwarfism and numerous skeletal abnormalities (Wang et al., 2007). A similar phenotype of reduced long bone growth was observed upon limb-specific inactivation of Rac1 (Suzuki et al., 2009). The defects we observed in cartilage-specific Rac1 knockout (KO) mice appear to be due, at least to a large extent, to reduced chondrocyte proliferation (Wang et al., 2007), but the downstream mechanisms connecting Rac1 to chondrocyte proliferation are unknown. Rac1 has been shown to control many cellular processes, including actin organization and transcription (Bustelo et al., 2007; Lee and Dominguez, 2010; Pai et al., 2010). In addition, Rac1 activation leads to the generation of reactive oxygen species (ROS) that mediate certain aspects of Rac1 signaling in various tissues (Hordijk, 2006; Sulciner et al., 1996; Sundaresan et al., 1996; Zhang et al., 2011). ROS have also been implicated in the promotion of chondrocyte differentiation (Kim et al., 2010; Morita et al., 2007), prompting us to investigate whether altered ROS levels contribute to the effects of Rac1 deficiency in chondrocytes. There are many interactions between the metabolism of ROS and nitric oxide (NO) (Thomas et al., 2008); however, NO or the enzymes involved in its synthesis have not been identified as major mediators of Rac1 signaling.

Three different nitric oxide synthase (NOS) proteins have been identified, eNOS, iNOS and nNOS (endothelial, inducible and neuronal NOS, respectively) (Bryan et al., 2009). Although pharmacological manipulation of NO levels has been shown to affect chondrocyte differentiation (Teixeira et al., 2005; Teixeira et al., 2001), the roles and regulation of individual NOS genes in cartilage development are largely unknown. iNOS is capable of producing large amounts of NO, often in response to extracellular signals, such as pro-inflammatory cytokines (Teixeira et al., 2008). It has been shown that Nos2 (iNOS) KO mice show accelerated cartilage degeneration in surgical models of osteoarthritis (Clements et al., 2003), but growth plates of these mice have not been described. Here we identify iNOS as a novel target of Rac1 that mediates the mitogenic activities of this pathway during endochondral bone growth.

We had previously shown hypocellularity and reduced proliferation in Rac1-deficient growth plates (Wang et al., 2007). We first examined whether we could recapitulate this phenotype in vitro to facilitate biochemical studies on the underlying mechanisms. Primary chondrocytes were isolated from Rac1^fl/fl mice that carried two alleles of a conditional (floxed) Rac1 allele. Cells in monolayer culture were infected with adenoviruses expressing green fluorescent protein (GFP; control) or Cre recombinase (Ad-Cre). Infection with Ad-Cre resulted in a marked reduction of Rac1 protein (Fig. 1A); densitometry determined that Rac1 proteins levels were reduced by 70% in Ad-Cre-infected cultures (data not shown). Control cells increased their numbers steadily over the 3-day time course of the experiment, whereas this increase was much slower in cells infected with Ad-Cre (Fig. 1B). This was accompanied by reduced expression of cyclin D1 (Fig. 1C), as we had observed in vivo. The similarity of these effects to the phenotypes observed in vivo suggested that our in vitro culture system provides an appropriate model to study the mechanisms employed by Rac1 to regulate chondrocyte proliferation.

Because Rac1 is able to stimulate ROS generation, we measured levels of ROS and NO. Rac1 deletion caused an approximately 50% reduction in ROS levels (Fig. 1D), but NO production was inhibited to a much larger degree (Fig. 1E). We therefore examined the molecular cause of reduced NO production and showed that expression of iNOS protein was markedly reduced in Rac1-deficient chondrocytes, whereas the expression of eNOS and nNOS was similar to that in control cells (Fig. 1F). Densitometry showed a significant 51% reduction in iNOS protein levels in Cre-expressing chondrocytes, whereas no significant changes were seen for eNOS and nNOS proteins (data not shown). To analyze the effects of Rac1 deficiency on iNOS expression in vivo, we performed immunohistochemistry on paraffin sections from 7-day old (P7) control and cartilage-specific Rac1 KO mice. In control mice, iNOS protein was found mostly in late proliferating and prehypertrophic chondrocytes (Fig. 1G). iNOS staining was notably reduced in growth plates from Rac1 mutant mice that also showed disorganization of the proliferative zone, as we described previously (Wang et al., 2007). A similar decrease in staining in mutant mice was observed using an antibody for nitrotyrosine, an indicator of cellular NO levels (Fig. 1H). Collectively, these data suggest that iNOS expression and NO generation are reduced in Rac1-deficient chondrocytes.

We next examined whether addition of exogenous NO could overcome the effects of Rac1 deficiency. We treated cells with the phorbolester, phorbol 12-myristate 13-acetate (PMA; an inducer of ROS that did not change NO levels; data not shown) and observed no effects on numbers of control or Rac1-deficient chondrocytes, suggesting that the mitogenic defects in Rac1-deficient chondrocytes are not due to lower ROS production (Fig. 2A). By contrast, addition of the compound 3-morpholinosydnonimine (SIN-1), which has been show to induce NO synthesis in chondrocytes (Del Carlo and Loeser, 2002; Fay et al., 2006; Mathy-Hartert et al., 2003) and increased both ROS and NO generation in our experiments (supplementary material Fig. S1), restored the number of Rac1-deficient chondrocytes in monolayer culture (Fig. 2A). Similarly, western blotting and subsequent densitometry analysis demonstrated that cyclin D1 expression in Rac1-deficient chondrocytes is restored by SIN-1 treatment, but not by PMA (Fig. 2B).

These data suggested that NO production mediates the mitogenic effects of Rac1 in chondrocytes and that inhibition of NO synthesis by other means would result in similar phenotypes to those resulting from deletion of Rac1, whereas NO donors should promote chondrocyte proliferation. To test this prediction, we incubated chondrocytes with the NOS inhibitor N(G)-nitro-L-arginine methyl ester (L-NAME) and the NO donors S-nitroso-N-acetylpenicillamine (SNAP) and 2,2′-(hydroxynitrosohydrazino) bis-ethanamine (NOC18). L-NAME reduced the number of chondrocytes, and both donors increased cell numbers (although only SNAP produced a statistically significant difference in these studies; Fig. 2C,D). In additional experiments, statistically significant increases in the number of chondrocytes was produced by NOC18 and sodium nitroprusside (SNP; supplementary material Fig. S1B). Collectively, these data suggest a model in which Rac1 promotes chondrocyte proliferation through iNOS–NO-mediated induction of cyclin D1 expression.

We next examined whether NO regulates chondrocyte proliferation and tissue architecture in intact growth plates, using an organ culture system of embryonic mouse tibiae (Agoston et al., 2007). Tibiae were grown with DMSO (control) and L-NAME for 6 days. Histochemical analyses of Haematoxylin and Eosin-stained paraffin sections from cultured tibiae followed by measurement of the growth plate zones demonstrated that L-NAME treatment resulted in a shortened proliferative zone, whereas the resting and hypertrophic zones of the growth plate were not affected (Fig. 3A,B; supplementary material Fig. S2). This was accompanied by reduced numbers of BrdU-incorporating cells (Fig. 3C,D), providing additional evidence for an important role of NO production in chondrocyte proliferation.

To investigate whether NO donors can rescue the effects of Rac1 inhibition in intact growth plates, tibiae were incubated with the pharmacological Rac1 inhibitor NSC23766 (Gao et al., 2004). Our earlier micromass culture studies had demonstrated that this compound has similar effects on early chondrogenesis as genetic inactivation of Rac1, providing evidence for the specificity of the inhibitor (Woods et al., 2007b). NSC23766 treatment for 6 days resulted in drastic reduction of tibial growth in our organ culture system (Fig. 4). The NO donor SNP did not affect tibial growth on its own, but completely rescued the effects of Rac inhibition on overall tibial growth (Fig. 4). These data demonstrate that restoration of NO signaling overcomes the effects of Rac1 deficiency in the authentic three-dimensional context of the growth plate.

Because our data suggested a role for endogenous Rac1–iNOS signaling in growth plate chondrocyte proliferation, we next analyzed the cartilage phenotype of iNOS-deficient mice (Laubach et al., 1995). Alcian Blue and Alizarin Red staining showed that the skeletons of these mice were morphologically normal, but slightly smaller than those of control littermates (supplementary material Fig. S3). Growth plates from these mice showed normal zonal and columnar architecture, but appeared slightly hypocellular in their centers (Fig. 5A; supplementary material Fig. S3), similar to but more subtle than those of mice with cartilage-specific deletion of the Rac1 gene (Wang et al., 2007). Cell counts confirmed hypocellularity in the proliferative zone in mutant mice (Fig. 5B). Immunohistochemistry and counts of positive cells demonstrated lower numbers of chondrocytes incorporating BrdU (Fig. 5C,D) and staining positively for cyclin D1 protein (Fig. 5E,F) in iNOS-deficient growth plates. Both of these features had also been observed in cartilage-specific Rac1 KO mice (Wang et al., 2007).

We had shown previously that cyclin D1 expression in chondrocytes is positively regulated by the transcription factors ATF2 and CREB and repressed by ATF3 (Beier et al., 2001; Beier et al., 1999; Beier and LuValle, 2002; James et al., 2006). Immunohistochemistry demonstrated limited expression of ATF3 protein in chondrocytes of control mice, with strongest staining at the transition from the proliferating to the prehypertrophic zone and weaker staining in hypertrophic chondrocytes (Fig. 6A). Loss of either Rac1 or iNOS resulted in a marked increase in the number of stained cells, as well as increased staining intensity (Fig. 6A). This suggests increased expression of ATF3 that in turn can repress cyclin D1 transcription and result in earlier cell cycle withdrawal. In agreement with this model, Atf3 mRNA levels were increased in Rac1-deficient cartilage in vivo (Fig. 6B) and in primary Rac1^fl/fl chondrocytes infected with Ad-Cre (Fig. 6C). Western blotting revealed that Rac1 deletion also increased ATF3 protein levels, whereas SIN-1 treatment slightly reduced ATF3 protein levels in both control and Rac1-deleted chondrocytes, and PMA did not affect ATF3 expression in either case (Fig. 6D). These data suggest that the Rac1–iNOS–NO pathway acts to suppress ATF3 expression in order to maintain cyclin D1 transcription and chondrocyte proliferation.

Chondrocyte proliferation is one of the main determinants of endochondral bone growth and thus final height. Although many extracellular signals that control these processes are known, the intracellular signaling pathways connecting cell surface receptors to specific cell cycle genes are less well understood. We had shown that the small GTPase Rac1 is essential for normal chondrocyte proliferation and cyclin D1 expression in mice (Wang et al., 2007), but the molecular mechanisms involved were not known. Here we demonstrate that Rac1 controls chondrocyte proliferation by stimulation of iNOS expression and NO production, which in turn suppresses ATF3 expression and thus allows increased transcription of the cyclin D1 gene (Fig. 7).

Although the regulation of ROS generation by Rac1 is well established, there is much less known about the interactions between Rac proteins and NO signaling. The requirement for Rac1 for normal iNOS expression in our study is consisted with an earlier report showing reduced levels of iNOS (Nos2) mRNA upon adenoviral delivery of dominant-negative Rac to rat livers (Harada et al., 2003). However, Rac1 activation or inhibition did not alter iNOS induction by interleukin 1β in smooth muscle cells (Finder et al., 2001), suggesting that the regulation of iNOS expression by Rac1 is stimulus and/or cell-type specific. Moreover, Rac1 deficiency in endothelial cells leads to reduced expression and activity of eNOS (Sawada et al., 2008), whereas eNOS protein levels were not altered in our system. Together, these data indicate that the connections between Rac1 (and possibly other Rho GTPases) and components of the NO signaling pathway act in a cell-type-specific manner.

Similar to the links between Rho GTPases and iNOS, studies from other tissues suggest that the role of iNOS itself in cellular proliferation is cell-type and/or context specific. For example, iNOS is required for hepatocyte proliferation during liver regeneration (Rai et al., 1998) but suppresses proliferation of smooth muscle cells (Kibbe et al., 2000). Our studies clearly show a requirement for iNOS in chondrocyte proliferation. These in vivo data are further supported by use of the NOS inhibitor L-NAME in vitro, which also resulted in a reduction in the number of chondrocytes in monolayer culture and reduced proliferation in tibia organ culture.

NO and NOS genes have been studied extensively in the context of bone remodeling and arthritis (Scher et al., 2007; van't Hof and Ralston, 2001), but much less is known about the roles of this pathway in endochondral bone development (Teixeira et al., 2008). Teixeira and colleagues showed that excess NO promotes hypertrophic differentiation and apoptosis of chondrocytes in chicken (Teixeira et al., 2005; Teixeira et al., 2001), whereas our data suggest that basal levels of NO are required for chondrocyte proliferation. Interestingly, we recently demonstrated a similar phenotype as described here (e.g. reduced proliferation and cyclin D1 expression, increased levels of ATF3 and premature cell cycle exit) in eNOS-deficient mice (Yan et al., 2010). Together, these studies suggest a dose-dependent effect of NO in the growth plate where low levels are required for proliferation, but high levels promote terminal differentiation and ultimately apoptosis. This model is in agreement with suggested dose-dependent effects of NO (Thomas et al., 2008). It should be noted in this context that in vitro data from our and other laboratories suggest Rac1 functions during chondrocyte hypertrophy (Kerr et al., 2008; Wang and Beier, 2005); therefore, it is possible that iNOS or other NOS enzymes also contribute to this role of Rac1. However, it appears probable that the Rac1–iNOS interaction demonstrated by us is not exclusive; for example, Rac1 probably acts through additional effectors, including ROS and the actin cytoskeleton, and iNOS activity and expression in chondrocytes is probably controlled by additional upstream regulators.

The similar phenotypes found in this and our earlier study (Yan et al., 2010) also suggest partial overlap of iNOS and eNOS function in chondrocytes. NO generated from either source inhibits the onset of ATF3 expression, allowing continued expression of cyclin D1 and thus delaying cell cycle exit. Because the effects of SIN-1 on ATF3 expression are less pronounced than its rescue of cyclin D1 levels, it is probable that additional transcriptional or post-transcriptional events contribute to the effects of iNOS–NO signaling on cyclin D1 expression. However, eNOS expression was not affected by loss of Rac1, suggesting that the two genes are regulated by different upstream pathways and mediate mitogenic responses to different external stimuli.

In conclusion, this study identified the iNOS/NO pathway as a central mediator of the mitogenic activity of Rac1 in growing endochondral bones. Further research into this will provide us with novel approaches towards the treatment of various skeletal diseases such as chondrodysplasias.

Timed-pregnant CD1 mice were purchased from Charles River Laboratories at embryonic day (E) 15.5. S-nitroso-N-acetylpenicillamine (SNAP), 2,2′-(hydroxynitrosohydrazino) bis-ethanamine (NOC18) and the Rac1 inhibitor (NSC23766) were obtained from Calbiochem. N(G)-nitro-L-arginine methyl ester (L-NAME), 3-morpholinosydnonimine (SIN-1) and sodium nitroprusside (SNP) were obtained from Sigma and reconstituted in dimethyl sulfoxide (DMSO, vehicle) according to the manufacturer's instructions. Cell culture materials and general chemicals were obtained from Invitrogen, Sigma or VWR unless otherwise stated.

The following antibodies were employed in this study: iNOS ab15323-500, nNOS 1376 and eNOS ab3868-100 (Abcam); Rac1 sc-217, nitrotyrosine sc-32757, ATF3 sc-188, goat anti-mouse sc-2005 and goat anti-rabbit sc-2004 (all from Santa Cruz Biotechnology); cyclin D1 (SP4) 9104-S1 (NeoMarkers Inc.); BrdU 03-3900 (Zymed Laboratories).

Mice with a deletion of the iNOS gene (Nos2^tm1Lau; stock no. 002609) (Laubach et al., 1995) and control mice with C57/BL6 background were obtained from Jackson Laboratory (Bar Harbor, ME), and exposed to a 12-hour light–dark cycle and given tap water and regular chow ad libitum. All procedures involving animals were approved by the University of Western Ontario Animal Care and Use Committee. For PCR genotyping, tail snips were used to prepare DNA for PCR analysis. PCR genotyping was performed by simultaneous amplification of the wild-type and null Nos2 alleles as described previously (Tranguch and Huet-Hudson, 2003). PCR fragments were analyzed by agarose gel electrophoresis.

Primary chondrocytes were isolated essentially as described (James et al., 2006; Woods et al., 2009a). Our previous studies had established that these cells maintain their chondrogenic phenotype for several days, based on their gene expression profile (Bursell et al., 2007; Stanton and Beier, 2007). Long bones from limbs were isolated from E15.5 mouse embryos and placed in α-minimum essential medium (α-MEM; Invitrogen) containing 0.2% bovine serum albumin (BSA), 1 mM β-glycerophosphate, 0.05 mg/ml ascorbic acid and penicillin–streptomycin. Then the medium was removed and the bones placed in 4 ml of 0.25% trypsin–EDTA (Invitrogen) for 15 minutes at 37°C. Trypsin was subsequently replaced with 1 mg/ml collagenase P (Roche) in Dulbecco's modified Eagle's medium (DMEM)–10% fetal bovine serum (Invitrogen), and cells were incubated at 37°C with rotation at 100 rpm for 90 minutes. Following digestion, the cell suspension was centrifuged for 5 minutes at 1000 rpm, and the collagenase containing supernatant was decanted. Chondrocytes were resuspended in medium containing 2:3 DMEM:F12, 10% fetal bovine serum, 0.5 mM L-glutamine and penicillin–streptomycin (25 units/ml). Cells were seeded in six-well NUNC plates at a density of 5×10⁴ cells per ml and incubated overnight. Primary monolayer chondrocytes were treated with 10⁻³ M L-NAME, 10⁻⁵ M SNAP or NOC-18 or the DMSO control (vehicle) diluted in fresh medium supplemented with 0.25 mM ascorbic acid (Sigma) and 1 mM β-glycerophosphate (Sigma) and incubated for up to 5 days with replacement of the medium and treatments every other day.

Adenovirus expressing Cre recombinase or GFP (control) was purified using the Adeno-XTM virus mini-purification kit (Clontech) and incubated with primary chondrocytes from Rac1^fl/fl mice at a multiplicity of infection (MOI) of 50. Medium was changed 4 hours after culturing and cells were then cultured as described above.

Intracellular ROS and NO were measured fluorometrically. The nonfluorescent dye 2′,7′-dichlorofluorescein diacetate (DCF-DA) permeated cells easily and hydrolyzed to fluorescent 2′,7′-dichlorofluorescein (DCF) upon interaction with intracellular ROS. Cultured cells were washed twice with PBS. Cells were then incubated with 20 μM of DCF-DA for 30 minutes and washed with PBS. The DCF fluorescence was analyzed using a fluorometer (SAFIRE microplate reader, TECAN, Australia) at an excitation of 485 nm and an emission of 530 nm. Nitric oxide production in chondrocytes was analyzed using DAF-FM DA (4-amino-5-methylamino-2′,7′-difluorofluorescein diacetate), a fluorescence probe specific for NO, that was diluted in culture medium and added to cells (10 μM). Cells were incubated with DAF-FM for 30 minutes at 37°C and then washed with PBS. Measurements were made using the SAFIRE microplate reader (TECAN, Australia; excitation, 495 nm; emission, 515 nm).

Cell proliferation assays were performed using CyQUANT NF Cell Proliferation Assay Kit (Invitrogen) according to manufacturer's protocol, as described previously (Yan et al., 2010). Chondrocytes were isolated as described above and seeded in 96-well NUNC plates at a density of 1000 cells/100 μl/well. After treatment, the medium was discarded and replaced with CyQUANT NF dye reagent diluted in Hanks' balanced salt solution (HBSS) and incubated for 30 minutes in 37°C incubator. Measurement of the fluorescence intensity of each sample used SAFIRE microplate reader (TECAN, Australia) with excitation at 480 nm and emission detection at 520 nm with a minimum of six individual wells per treatment in duplicate and three independent cell isolations. Images of primary chondrocyte were taken after treatments of 1 day in culture using a Nikon SMZ1500 microscope with Photometric CoolSNAP color digital camera (Nikon Canada) and PTI Image Master 5 program.

Tibiae were isolated from E15.5 embryos from CD1 timed-pregnant mice (Charles River Canada) using the Stemi DV4 Stereomicroscope (Zeiss) as described (Agoston et al., 2007; Ulici et al., 2008). Dissection day was considered to be day 0 and tibiae were allowed to recover from dissection overnight in serum-free α-MEM containing 0.2% BSA, 0.5 mM L-glutamine, 40 units penicillin/ml and 40 μg streptomycin/ml as described (Agoston et al., 2007). The following morning, bones in 24-well Falcon plates were measured using an eyepiece in the Stemi DV4 Stereomicroscope and treated with 10⁻³ M L-NAME or 5×10⁻⁵ M Rac1 inhibitor (NSC23766) with or without 10⁻⁵ M SNP, and DMSO as the vehicle. Medium was changed every 48 hours beginning on day 1, and bones were measured on days 1 and 6. Results are expressed as change in length relative to day 1. Experiments were repeated at least three times, with four to six bones per treatment for each trial.

After experiment completion, tibia samples from organ culture or newborn mice were rinsed with PBS and fixed in 4% paraformaldehyde overnight. Bones were then placed in 10% formalin solution and sent for embedding and sectioning at the Molecular Pathology Core Facility at the Robarts Research Institute (London, Ontario, Canada). Following sectioning, bones were stained with Hematoxylin and Eosin (H&E) or Safranin O and Fast Green using standard protocols (Gillespie et al., 2011; Solomon et al., 2009; Ulici et al., 2009; Wang et al., 2007; Yan et al., 2010). For immunohistochemistry, sections were incubated with primary anti-cyclin D1 antibody (1:50 dilution), anti-ATF3 antibody (1:200 dilution), anti-nitrotyrosine antibody (1:200 dilution) or anti-iNOS antibody (1:1000 dilution) overnight at 4°C. For BrdU labeling, newborn mice were injected intraperiteally 1 hour before being killed with BrdU (Roche) at a dose of 0.01 ml/g. BrdU was detected in paraffin sections using anti-BrdU antibody (1:100 dilution). Bound antibody was visualized using the UltraVision LPValue detection system (Lab Vision) with AEC chromogen substrate (Lab Vision).

Total RNA was isolated from epiphyseal cartilage of long bone from newborn mice or primary cells using TRIzol reagent (Invitrogen), according to the manufacturer's recommendations. Taqman real-time PCR was performed as described previously (Woods and Beier, 2006; Woods et al., 2005) with primers and probe sets from Applied Biosystems (Atf3 Mm00476032_m1; Gapdh Mm99999915_g1). Data were normalized to Gapdh mRNA levels and given as means ± s.e.m. from direct comparison of mutant and control littermates from three different crosses.

Alcian Blue and Alizarin Red staining of newborn Nos2 KO and control mice was done as described (Solomon et al., 2009; Ulici et al., 2009; Wang et al., 2007). Briefly, mice were skinned, eviscerated and dehydrated in 95% ethanol and acetone, then stained with 0.015% Alcian Blue, 0.05% Alizarin Red and 5% acetic acid in 70% ethanol for several days. After clearing in 1% KOH and passing through a decreasing KOH series, skeletons were stored in a 1:1 mixture of glycerol and ethanol.

All studies were done in at least three independent cell isolations or mice. Statistical significance was determined by Student's t-test or by one-way ANOVA with Bonferroni post test, using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego CA; www.graphpad.com). In figures where values were shown as percentages, data were first log-transformed before applying ANOVA.