Conditional deletion of E11/podoplanin in bone protects against load-induced osteoarthritis

Osteoarthritis is a degenerative joint disease and a global health care burden. In osteoarthritis, the articular cartilage undergoes structural deterioration, causing joint pain, loss of joint function and significantly reducing quality of life. However, its underlying molecular mechanisms are not fully understood. As such, there is an ever-growing need for an effective disease-modifying treatment.

Although often considered secondary, subchondral bone (SCB) thickening in osteoarthritic joints is one of the earliest detectable changes and is now considered a potential trigger for subsequent articular cartilage degeneration [1, 2]. Osteoblast-derived osteocytes are the most numerous of all the cells within bone and have a unique morphology with extensive dendritic processes creating bone’s osteocyte-canalicular network. This network is now known to orchestrate bone remodelling [3]. However, in osteoarthritic joints, the osteocytes in the SCB exhibit alterations to their exquisite dendritic morphology, with fewer and more disorganised dendrites [4]. Furthermore, other reports have noted that the expression of sclerostin, the mature osteocyte marker, is disrupted in osteoarthritic SCB [5, 6]. Together, these data suggest that the osteocyte may contribute a central role to pathological SCB sclerosis in osteoarthritis and that an intact osteocytic network is necessary for maintaining healthy SCB architecture.

Numerous genes have been suggested to influence osteocyte formation, one of which encodes the transmembrane glycoprotein E11/podoplanin. We and others have previously shown that E11 is expressed by early embedding osteocytes, thus identifying it as a factor which likely contributes to the vital, early stages of osteocyte differentiation [7‐9]. It is known that mechanical strain in vivo increases E11 mRNA expression [7] and that E11 siRNA abrogates the formation of osteocyte dendrites. [7]. In contrast, over-expression of E11 in ROS 17/2.6 osteoblast-like cells has been found to promote the formation of long dendritic processes [10‐12]. Furthermore, we have recently reported that E11 levels are regulated post-translationally by proteasomal degradation and that their preservation, through the administration of proteasome inhibitors such as Bortezomib, leads to the induction of an osteocyte-like morphology in MLO-A5 pre-osteocytic cells [9]. In accordance with this, we recently showed that the hypomorphic bone-specific ablation of E11 in mice results in disrupted osteocytic dendrite formation, which supports a key role for E11 in regulating the cytoskeletal changes associated with osteocyte process formation and elongation [13].

As the formation of such dendritic processes is a key functional feature of the normal mature osteocyte network, which is perturbed in osteoarthritis [4], we have examined herein whether disruption to the integrity of the osteocyte dendritic processes contributes to the initiation of osteoarthritis. Specifically, we investigated whether proteasome inhibition can stabilise E11 expression in vivo to protect against the osteoarthritis that develops following surgically-induced destabilisation of the medial meniscus (DMM). Moreover, we examined whether the bone-specific conditional deletion of E11 in mice affects early adaptive processes and joint vulnerability to osteoarthritis induction by a mechanically-induced post-traumatic osteoarthritis model.

Here we reveal that the bone-specific conditional deletion of E11 in mice is protective against load induced osteoarthritis pathology. This is evidenced by the restriction of both the load-induced development of articular cartilage lesions and osteophyte formation in our E11 cKO mice. We also reveal that Bortezomib fails to exert any protection against osteoarthritis development in a surgical model of osteoarthritis (DMM). This conclusion was unexpected and opposite to our original hypothesis in which we speculated that disruption to the integrity of the osteocyte network would lead to greater osteoarthritis vulnerability.

Osteocytes are the most numerous bone cell type (> 95% of bone cells) and are essential to bone structure and function. They have a unique morphology with long dendritic processes creating multicellular networks permeating the entire bone matrix [3, 21]. Historically considered passive ‘place-holders’, osteocytes have now emerged as versatile orchestrators of bone remodelling as they regulate both osteoblast (bone-forming cells) and osteoclast (bone-resorbing cells) function [22, 23]. In osteoarthritis, osteocytes are known to have a dysfunctional morphology with shorter and fewer dendritic processes [4]. As E11 is essential for the formation of osteocyte dendrites, we therefore hypothesised that the ablation of E11 from bone cells would lead to SCB thickening and exacerbated osteoarthritis pathology via a decreased osteocyte production of the bone formation inhibitor, sclerostin. However, in contrast to our expectations, we observed the opposite outcome. Our loading model used herein is non-invasive and does not induce anterior cruciate ligament rupture, thus avoiding complications that surgical methods have surrounding risk of disturbances to peri-articular tissues and disease progression. Further, whilst DMM relies on permanent, intransient destabilisation where indirect induction and progression of articular cartilage lesions are inseparable, joint loading in our model is controlled and transient, allowing direct induction of lesions and separation from progression. The data from our loading regime (6 loads in 2 weeks) thus allows examination of load-induced early osteoarthritis induction [14]. Our data indicate that less efficient osteocyte differentiation and dendrite formation, due to the hypomorphic deletion of E11, protects against the induction of osteoarthritic articular cartilage in response to transient loading episodes. Further, our data presented here suggests that a disrupted osteocyte morphology occurs in response to osteoarthritis pathology, rather than being causative. This does not however negate the possibility of E11 deficiency increasing vulnerability to other stratifications of osteoarthritis. It is also important to consider the sample size used for our WT mice as a potential limitation of this study. However, using the in vivo loading model, similar small group sizes have been used to resolve statistically significant load-related differences in articular cartilage lesions, surface strains by digital image correlation and osteocyte protein expression by immunolabelling [14, 24, 25]. These results indicate the very high level of reproducibility and experimental robustness of this loading model and provides us with confidence in our interpretation of our data.

Herein we also show that the bone specific deletion of E11 results in reduced articular cartilage thickness in the lateral femur. This is somewhat surprising as it is thought that a thinner articular cartilage is more susceptible to tensile strains, and therefore load induced trauma [26, 27]. Further, as both sclerostin and E11 are expressed in chondrocytes, and as no effects were seen on osteoclast activity or SCB architecture, the protection to load-induced osteoarthritis afforded by E11 deletion in bone may, at least partially, reflect E11’s currently undefined role in the articular cartilage.

We have recently shown using in vitro osteocyte differentiation that late osteoblast E11 protein levels are regulated post-translationally by proteasome degradation and that their preservation, through use of proteasome inhibitors, such as Bortezomib, leads to the induction of an osteocyte-like morphology [9]. Bortezomib is used in vivo for the treatment of multiple myeloma and it is undergoing clinical trials for epithelial cancer treatment [28]. Moreover, it has been shown that Bortezomib prevents the degradation of collagen type II and the induction of MMP13 in vitro, thereby suggesting that it may have therapeutic effects in the context of osteoarthritis [29]. We therefore speculated that administration of the proteasome inhibitor, Bortezomib, in vivo would exert a protection against osteoarthritis development in an alternative surgically-induced model. We found that the administration of 1 mg/kg Bortezomib, via intraperitoneal injection, to mice undergoing DMM surgery had no effect however on osteoarthritis pathology. This concentration and route of delivery has been shown previously to successively reduce proteasome 20S and mitigate histopathological manifestation of pancreatic injury in mice [18]. This is in contrast to a recent publication which showed that the administration of another proteasome inhibitor, MG132, protects against DMM-induced osteoarthritis [30]. There are many possible explanations as to why we observed these contrasting results, the most likely of which is that MG132 and Bortezomib are different types of proteasome inhibitors – MG132 is a peptide aldehyde which also inhibits certain cysteine proteinases, whereas Bortezomib is a peptide boronate inhibitor [31]. Whilst Bortezomib is currently being developed in the clinic, it would be interesting to examine the effects of other proteasome inhibitors within these subcategories to explore whether they exert modification in osteoarthritis development. It is also pertinent to consider our immunohistochemistry results, which showed that the in vivo administration of Bortezomib was not associated with any modification in E11 expression levels in the SCB osteocyte. The failure of in vivo Bortezomib administration to recapitulate its in vitro effects on E11 expression may indeed offer an explanation for lack of effect on osteoarthritis severity. It is nonetheless intriguing that in vivo Bortezomib treatment instead provoked decreased levels of E11 and sclerostin expression in articular cartilage chondrocytes – thus indicating that our Bortezomib dosing procedure was biologically effective in cartilage. The reasons for these observations require further study. It must also be borne in mind that the proteasome has manifold effects on cellular metabolic and signalling pathways, and its effects will not be limited to those we have analysed here.

Overall, these data suggest that an intact osteocyte network in the SCB contributes to the development of mechanically-driven osteoarthritis. Further, the data presented here indicate that the molecular pathways that preserve the osteocyte network, such as those driven by E11, may be targeted to limit osteoarthritis pathogenesis.