2. Current Insights
Bone remodeling is mediated by bone-resorbing osteoclasts and bone-forming osteoblasts, two entirely different cell populations. While osteoid, the organic component of the bone matrix, consists mostly of type I collagen that provides the scaffold for the mineral phase consisting of calcium phosphate crystals, there are numerous other proteins that contribute to the composition of the bone matrix and are essential for its proper function 
. In a previous study we have analyzed the transcriptome of differentiating osteoblasts and observed many well-established osteoblastogenesis markers among the genes with the strongest induction during differentiation, thereby demonstrating the validity of our approach. In this screening process, we also identified several genes with a hitherto unknown function in skeletal turnover, despite predominant expression in bone tissue. While the main objective of our previous study was related to Panx3 
, our current investigation is focused on the role of Col22a1
, encoding type XXII collagen (COLXXII). Col22a1
was one of two collagen transcripts identified among the hundred most strongly induced genes during osteoblast differentiation. In the current study we were able to show that Col22a1
, unlike Col13a1
, is predominantly expressed in bone tissue and differentiated osteoblasts, which led us to investigate its putative role in the skeleton.
COLXXII has originally been identified as a novel tissue junction component, as it is mostly localized in the transition zones of tissues, such as the myotendinous junction of heart and skeletal muscle, the articular cartilage/synovial fluid junction, and the junction of hair follicles and the dermis 
. Targeting of Col22a1
in Zebrafish morpholino models has led to the observation that these animals display muscular dystrophy most likely due to a severely disrupted myotendinous junction that could even lead to muscle detachment 
. In a different Zebrafish study, genetic disruption of the Col22a1
gene has led to an increased risk of intracranial hemorrhages due to increased vascular permeability 
. The latter was attributed to a disruption of the proper structure of the basal lamina. These findings were supported by the association of genetic variants of human COL22A1
with a higher risk of aneurysms 
. Importantly, however, since there is no published analysis on the phenotype of a Col22a1
-deficient mouse model, the physiological role of COL XXII is still not well-defined.
Based on the observed Col22a1
expression pattern, our primary focus was to generate such a mouse model in order to study their skeletal phenotype. Importantly, we did not observe any obvious morphological alterations or extraskeletal phenotypes, such as aneurysms. However, we did observe significantly less Col22a1
-deficient mice born than would be expected from the Mendelian ratio. Nonetheless, it remains purely speculative at this point, whether this discrepancy can be attributed to one of the previously proposed functions of COLXXII. With respect to the skeleton, it was also obvious that Col22a1
deficiency did not impair developmental processes, i.e., desmal or endochondral ossification, or skeletal growth. Moreover, we did not detect enrichment of non-mineralized bone, i.e., osteoid, unlike it is the case in mice carrying type I collagen mutations, or in mice lacking Dmp1 
. Finally, there was no indication of impaired osteoblast differentiation and function in association with Col22a1
deficiency. On the other hand, trabecular osteopenia in vertebral bodies, albeit moderate, was observed in all analyzed groups, where wildtype and heterozygous littermates were used as controls.
Of note, we also observed a significant reduction of trabecular bone mass in the distal femoral metaphysis, yet this difference was only found in the 24-week-old mice. Although this may indicate that the axial skeleton is more affected than the appendicular skeleton in Col22a1
-deficient mice, there is also another possible explanation for this apparent discrepancy. In fact, since there are far less trabecular bone structures in long bones when compared to the spine, a reduction of trabecular bone mass is generally more robust, when spine sections are analyzed. On the other hand, assessing bone parameters in femora by µCT is advantageous for quantification of cortical parameters in the midshaft region. Here we observed that cortical bone mass and most cortical bone mechanical properties, as determined by a three point bending test, remained unaffected by the loss of Col22a1
. This indicates that COLXXII is predominantly relevant in the trabecular bone compartment. The only effect on cortical bone that we were able to determine was reduced bone matrix stiffness. It is conceivable to speculate that this phenotype is caused by reduced type I collagen crosslinking via COLXXII, although such a function has not yet been described for COLXXII 
. Indeed, COLXXII belongs to the FACIT subfamily of collagens, which does not form fibrils but copolymerize into superstructures with fibril-forming collagens and mediate interactions with the environment 
When investigating the underlying cause for the reduced trabecular bone mass, we specifically observed increased osteoclast parameters. Since Col22a1
was found barely expressed in osteoclasts, these results were fully unexpected and suggest that they are indirectly mediated. Since osteoblast lineage cells are major producers of Rankl and Opg, which in turn regulate osteoclastogenesis 
, we particularly analyzed the possibility, that Col22a1
-deficient osteoblasts produce more Rankl and/or less Opg. By comparing primary osteoblasts isolated from wildtype and Col22a1
-deficient littermate mice, we did not observe a cell-autonomous osteogenesis defect, and there was no differential expression of the genes encoding Rankl or Opg. Moreover, the serum concentrations of these two factors were not different between wildtype and Col22a1
-deficient littermate mice. Interestingly, it has been shown that Col22a1
colocalizes with integrins such as α2β1 and α11β1 in a pattern that is reminiscent of integrin signaling 
. Moreover, an involvement of Col22a1
in integrin-mediated mechanosensing and negative regulation of chondrocyte hypertrophy has been hypothesized 
. Therefore, it is conceivable that an altered bone matrix composition lacking COLXXII influences osteoclast differentiation or attachment via integrin signaling 
. Further research is required to adequately test these hypotheses and to fully elucidate the underlying mechanism.
Apart from the fact that our current data do not provide a molecular mechanism explaining the increased trabecular osteoclast numbers in Col22a1-deficient mice, there are other limitations of our study. At the age of 24 weeks, we only analyzed male mice. Considering the data obtained in 6-week-old animals, it is likely that female mice might display a similar phenotype. However, this has not been verified. Furthermore, since we only analyzed a global knockout model, we cannot exclude that the observed phenotype might be caused by effects originating from other cell types. Therefore, even if we found Col22a1 to be predominantly expressed by osteoblasts, we truly believe that models with cell-specific deletion of Col22a1 are required to fully investigate the effect of COLXXII on the skeleton.