Introduction

CCN5/WISP2 is part of the CCN family of matricellular proteins, which is named for its founding members: CCN1/CYR1, CCN2/CTGF, CCN3/NOV, CCN4/WISP1, CCN5/WISP3, and CCN6/WISP3. Members of the CCN family of matricellular proteins share similar modular protein structures, containing an N-terminal secretory signal peptide, an insulin-like growth factor binding protein (IGFBP), a von Willebrand factor type C repeat (vWC), a thrombospondin type 1 repeat (TSR), and a cysteine knot motif within the C-terminal (CT). CCN5 is unique among members of this family by lacking the CT domain (Fig. 1). The biological role of CCN5 is unclear, but as a matricellular protein, CCN5 resides in the extracellular matrix, and likely serves regulatory rather than structural roles. Exogenous CCN5 has been shown to regulate an array of processes, including proliferation, migration, angiogenesis, tumorigenesis, differentiation, adhesion and ECM synthesis. Whether CCN5 is required in vivo is unknown.

Fig. 1
figure 1

Modular organization of members of the CCN family of matricellular proteins. All CCN family members contain conserved domains of N-terminal secretory signal peptide (SP), insulin-like growth factor binding protein (IGFBP), von Willebrand factor type C repeat (vWC), thrombospondin type 1 repeat (TSR), and a cysteine knot motif within the C-terminal (CT). CCN5 is the only member of the family that does not contain the C-terminal cysteine knot (CT). The hinge region is highly variable among the family members

Full size image

CCN5 was originally cloned in the 1990s by several groups. The first publication by Delmolino et al. (Delmolino and Castellot 1997) found that CCN5 was up-regulated in human vascular smooth muscle cells after treatment with heparin (Delmolino and Castellot 1997). This group named it Heparin-Induced CCN-like Protein (HICP). About the same time, another group found that Ccn5 was upregulated in the mouse mammary epithelial cell line C57MG after transformation by Wnt-1(Pennica et al. 1998). This group named it Wnt-Inducible Secreted Protein-2 (WISP-2). Several other names were assigned to Ccn5 by other groups around this time.

All six members of the CCN family are expressed in bone (Chen et al. 2014, Parisi et al. 2006). To date, functions in bone have been described for CCN1/Cyr61, CCN2/Ctgf, CCN3/Nov, and CCN4/Wisp1. Conditional knockout of Ccn1/Cyr61 using Osteocalcin-Cre led to a low bone mass phenotype that included thinning of the cortical bone (Zhao et al. 2018). Ablation of Ccn2/Ctgf in osteoblasts using Osteocalcin-Cre also led to a mild low bone mass phenotype, but was only seen in males and not in females; cortical bone was unaffected (Canalis et al. 2010a, b). While Ccn3/Nov is expressed in mature osteoblasts (Matsushita et al. 2013), Ccn3/Nov−/− mice showed no skeletal abnormality (Canalis et al. 2010a, b, Matsushita et al. 2013). However, they exhibit accelerated bone regeneration (Matsushita et al. 2013), consistent with studies showing that Ccn3/Nov inhibits osteoblast differentiation (Kawaki et al. 2011, Rydziel et al. 2007). In contrast, in Ccn4/Wisp1−/− mice, cortical bone thickness, cross-sectional area, and endocortical mineral apposition rate are significantly reduced (Maeda et al. 2015). Hence, some CCN family members (CCN1/Cyr61, CCN2/Ctgf, CCN4/Wisp1) have anabolic functions in bone, while CCN3/Nov has opposing functions.

Several studies have investigated the potential role of CCN5 in bone. Kumar et al. first identified Ccn5 mRNA in primary cultures of human osteoblasts (Kumar et al. 1999). In situ hybridization showed CCN5 as being highly expressed in bone-forming osteoblasts and in alkaline phosphatase positive bone marrow cells (Kumar et al. 1999). A later study by Kawaki et al. showed with immunohistochemistry that CCN5 protein co-localized with osteocalcin positive regions in mouse calvaria (Kawaki et al. 2011). In vitro functional studies showed that CCN5 protein promoted the adhesion of osteoblasts, inhibited the binding of fibrinogen to purified integrin receptors, and inhibited the production of osteocalcin by rat osteoblast-like Ros 17/2.8 cells (Kumar et al. 1999). Additionally, CCN5-treated primary murine calvaria osteoblasts showed increased mineralization with upregulation of the osteogenic genes Osterix, Alp, and Bsp (Kawaki et al. 2011). While these studies provide good support for anabolic function of CCN5 in bone, direct in vivo evidence for CCN5 is not available. The goal of this study is to generate Ccn5−/− mice to enable characterization of CCN5 function in bone and other tissues.

Methods

Vertebrate animals

Ccn5 knockout mice (Ccn5LacZ) were generated from targeted ES cells with the vector map shown in Fig. 2, obtained from the Knockout Mouse Project (KOMP) repository at UC Davis (Wisp2tm1(KOMP)Vlcg). ES cells were injected into C57Bl6J blastocysts, and germline transmission was confirmed at UC Davis. All animals were treated in accordance with the National Institutes of Health guidelines for care and use of animals, and approved by the UCLA Institutional Animal Care and Use Committee.

Fig. 2
figure 2

a Schematic of reporter-tagged deletion allele for CCN5 using ES cells obtained from IMPC (Wisp2tm1(KOMP)Vlcg), LacZ cassette is inserted directly behind ATG starting codon located in Exon2. b Ccn5 gene expression in mouse tibia showing no Ccn5 mRNA in Ccn5LacZ/LacZ mice

Full size image

LacZ staining

Whole-mount LacZ staining was performed on Ccn5 heterozygous (Ccn5LacZ/WT) mice to examine the expression pattern of Ccn5 in bone. A standard X-gal staining protocol was utilized as previously described (Jiang et al. 2017). Briefly, mice were euthanized and hind limbs were dissected and fixed in 0.2% glutaraldehyde LacZ fixative solution. After fixation, the tissue was washed and stained with X-gal overnight at 37 °C. After X-gal staining, the tissue was washed and fixed in 4% paraformaldehyde and decalcified in 19% EDTA. After decalcification, the tissue was embedded in paraffin and sectioned. Sectioned slides were counterstained with Eosin and visualized with on a microscope (Model BX60F; Olympus Optical Co., Japan) equipped with a digital camera (Model 01-RET-OEM-F-CLR-12; QImaging, Surrey, Canada). Photomicrographs were taken with a Nikon Ti-DH Microscope. Images were processed in Photoshop (Adobe).

Micro-computed tomography (μ-CT) analysis

Bone parameters were quantified on femurs from mutant and wild type (Ccn5+/+) mice by μ-CT (Skyscan1172; Bruker MicroCT, Kontich, Belgium) using CTAn (v.1.14.4) and CTVol (v.2.2) software. The microradiography unit was set to an energy level of 55 kVp, intensity of 181 μA, and 900 projections; specimens were scanned at a 10μm3 voxel resolution. A three-dimensional reconstruction was generated with NRecon software (Bruker MicroCT, Kontich, Belgium) from the set of scans. The regions of interest (ROI) of trabecular bone were defined as the areas between 1 mm and 3 mm from the growth plate in the metaphyseal region of distal femurs. The ROI of cortical bone was defined as 0.75 mm segments of the femoral middle-diaphysis. All ROIs were drawn automatically and trabecular regions were assessed for bone mineral density (BMD) and bone volume fraction (bone volume/total volume, BV/TV). The mid-shaft average cortical bone thickness (Ct.Th) values were analyzed. All abbreviations and nomenclature are standardized according to previously published guidelines (Bouxsein et al. 2010).

In vivo gene expression

For in vivo samples, femurs and tibias were dissected and soft tissue was removed. Bones were then cut to remove both 3 mm termini, and the bone marrow was flushed out with cold PBS until the cortical bone became white. The resulting cortical bone was flash frozen with liquid nitrogen, homogenized with a grinder, and re-suspended in 1 mL Trizol (Thermo Scientific, MA). Total RNA was isolated by the phenol-chloroform method and converted to cDNA using SuperScript III (Thermo Scientific). cDNA was amplified and quantified using Maxima SYBR Green qPCR master mix (Thermo Scientific). Analysis of Ccn5 gene expression was done using following primers: 5’-TGTGTGACCAGGCAGTGATG-3′ and 5’-GGATACTCGGGTGGCTATGC-3′. Ccn5 expression is then normalized to the housekeeping gene Gapdh: 5’-CTTTGGCATTGTGGAAGGGC-3′ and 5’-CAGGGATGATGTTCTGGGCA-3′.

Statistical analysis

The data represent results from at least three independent experiments, and are presented as the mean ± standard deviation (SD). Statistical significance between experimental and control groups was compared by Student t test. For experiments with more than two parameters, one-way analysis of variance (ANOVA) followed by Tukey’s post hoc analysis was used. A value for p < 0.05 was considered significant.

Results

Ccn5 expression in bone in vivo

The Ccn5LacZ mouse used for this study is a knockout/knock-in where exon 2 to exon 5 of the Ccn5 gene is knocked-out and is replaced with a LacZ cassette (Fig. 2a). We confirmed the absence of Ccn5 mRNA in the limbs of Ccn5LacZ/LacZ mice using qPCR (Fig. 2b). Ccn5Lacz/WT heterozygous mice did not present any obvious phenotypes. LacZ is expected to be expressed instead of Ccn5 where and whenever Ccn5 is being expressed. Thus we examined the presence of LacZ in Ccn5LacZ/WT heterozygous mice as a reporter for Ccn5 expression. Given previous studies showing Ccn5 expression in bone (Kawaki et al. 2011), we examined LacZ expression in this tissue. In trabecular bone, we did not observe any significant staining in any of the compartments in Ccn5LacZ/WT heterozygous mice (Fig. 3a). However, LacZ staining was very strong in bone periosteal cells (Fig. 3b). We did not see any staining in the bone marrow stromal cells, osteocytes, osteoblasts or endosteal cells.

Fig. 3
figure 3

LacZ staining of 3 month old Ccn5LacZ/WT heterozygous mice: a trabecular bone; b cortical bone. Strong LacZ staining on the periosteal surface of cortical bone. No staining was observed in any other compartment

Full size image

Ccn5 LacZ/LacZ bone phenotype

We examined the bone phenotype of Ccn5LacZ/LacZ mice using μCT. When compared to WT littermates, we found that loss of Ccn5 had no apparent effect on trabecular bone mineral density (BMD) or bone volume fraction (BV/TV). Additionally, loss of Ccn5 did not have an effect on cortical bone thickness (Fig. 4). These observations are consistent with subsequent skeletal analyses performed by International Mouse Phenotyping Consortium (IMPC).

Fig. 4
figure 4

Micro-CT analysis of 3 month old Ccn5LacZ/LacZ mice. a Representative μCT 3D reconstruction of 3 month old mice. Quantitative analysis of bone morphology with μCT (b-d). No differences were seen between WT and Ccn5LaZc/LacZ mice in b BMD, c BV/TV and d Cortical bone thickness

Full size image

Discussion

In this study, we found LacZ under the control of the endogenous Ccn5 regulatory sequences to be highly expressed in periosteal cells in adult bone, but we did not observe any LacZ staining in any other compartment of bone. These data are somewhat contradictory with previous studies showing Ccn5 to be expressed in bone-forming osteoblasts and alkaline phosphatase positive bone marrow cells (Kumar et al. 1999). Several factors can contribute to this difference, as the previous study was done in human fetal femoral growth plate. Similarly, Kawaki et al. found co-localization of CCN5 protein with osteocalcin in neonatal mouse calvaria. It is conceivable that the pattern of CCN5 expression differs in neonatal and adult mice. Furthermore, we did not specifically examine calvarial expression. In this study we found that Ccn5/LacZ expression is localized on the periosteal surface of adult bone, which has not been previously reported.

Currently there remains some debate as to the expression profile of Ccn5 during development. Jones et al., using immunohistochemistry and qRT-PCR, showed that Ccn5 is expressed in all the organs examined and is specifically high in skeletal muscle and epidermis of the skin (Jones et al. 2007). Data from Eurexpress/Genepaint showed a general hazy background staining for Ccn5 with in situ hybridization. It is very possible that Ccn5 is expressed at low levels in many tissues during development, which can be difficult to detect with in situ hybridization, but can be easily detected with qRT-PCR. Additionally, since CCN5 is secreted, the protein may accumulate in the extracellular matrix and be detected with immunohistochemistry. In the future we plan to use our LacZ reporter mice to examine in detail the pattern of Ccn5 expression during development.

IMPC analysis showed that Ccn5LacZ/LacZ mice had a normal phenotype in most organs. The only abnormal phenotype exhibited by Ccn5LacZ/LacZ mice was a change in auditory brainstem response in the IMPC analysis. These data are a significant departure from a previous observation indicating that both Ccn5-null (not published) and overexpressing mice are embryonic lethal (Myers et al. 2012, Russo and Castellot 2010). Since the Ccn5-null mouse strain described previously is not available, it is difficult to explain the drastically different phenotypes. The Ccn5LacZ strain we analyzed represents a loss of function allele since all of the conserved protein coding exons are replaced by LacZ. There have been other examples of genetic modifications within the CCN family that showed major differences in phenotypes. These are typically the result of incomplete knockout of the gene and the production of a mutant protein that can have antagonistic effects. CCNs are unique in that all members share a similar modular structure, and within the genome, each exon usually encodes one of the modular domains of the protein. Most of the exons are in frame, so deletion of any one or multiple exons often does not result in a frame shift in the remaining mRNA (Jiang et al. 2017). For example, Ccn3/Nov knockout mice where exons 1-3(Matsushita et al. 2013) and exons 1-5 (Canalis et al. 2010a, b) were deleted showed no apparent skeletal phenotype (Canalis et al. 2010a, b, Matsushita et al. 2013), while Ccn3/Nov mutant mice in which exon 3 alone was deleted (Novdel3) produced multiple skeletal changes characterized by overgrowth of multiple skeletal elements (Heath et al. 2008). In this latter strain, in-frame exon skipping generates a form of CCN3 lacking the vWC domain.

Previous papers have demonstrated that CCN5 enhances osteogenic differentiation in vitro through both Wnt (Grünberg et al. 2014, Robinson et al. 2006) and integrin (Kumar et al. 1999) signaling. Ccn5 expression is directly upregulated by canonical Wnt signaling and CCN5 has been shown to attenuate Wnt signaling in vitro (Grünberg et al. 2014). In this study we examined the effect of Ccn5 on skeletal tissues using Ccn5LacZ/LacZ mice. We found that Ccn5LacZ/LacZ mice are viable with a normal skeletal phenotype; this is consistent with IMPC skeletal analysis. This lack of phenotype suggests that Ccn5 is dispensable for bone homeostasis, in spite of its strong expression in the periosteum (this study), and strong expression in different bone compartments reported by other groups during neonatal development. The lack of a discernable phenotype does not signify that Ccn5 is unimportant in bone biology. Ccn3/Nov knockout mice did not exhibit an overt skeletal phenotype but were shown to have accelerated bone healing after injury (Canalis et al. 2010a, b, Matsushita et al. 2013). This also could be the case for Ccn5. Thus, we plan to conduct future experiments that would induce stress, via fracture, mechanical loading and ovariectomy, to further examine the role of Ccn5 in bone biology in vivo.