Our demonstration is based on a series of complementary observations. First, excavations without collagen left-over, thus where collagen degradation was as fast as demineralization, had the shape of continuous trenches reflecting long-lasting resorption events. In contrast, excavations with collagen left-over, thus where collagen degradation was slower than demineralization, had the shape of discrete

round pits reflecting intermittent short-lasting resorption events. This relation between collagen and duration of resorption was already suggested by SEM pictures [17], and is now further supported by our quantitative analysis. Second, if specifically decreasing the collagen degradation rate by using a CatK inhibitor, collagen accumulated faster, resorption stopped at smaller depths and generated clusters of discrete pits, at the expense Sotrastaurin solubility dmso of deep continuous resorption trenches, as also recently reported by Leung et al.

[19]. Furthermore, we show that this response to pharmacological inhibition is not artefactual and results directly from CatK inactivation, since the prevalence of pits and trenches varied similarly with the natural variation of CatK levels amongst different OC preparations. Third, conversely, if decreasing specifically and slightly the rate of demineralization in order to allow collagen degradation to proceed as fast as demineralization, collagen did not accumulate in the excavations and resorption continued over longer distances thereby generating continuous HSP inhibitor resorption trenches instead of discontinuous resorption pits. Thus, paradoxically, a resorption inhibitor may favor continuous bone resorption. The same result was obtained if the OCs were offered bone

slices where collagen had been damaged by a NaOCl pretreatment, which is an alternative way to facilitate removal of collagen and to render it as fast as demineralization. Observations in line with this were obtained Rolziracetam by others after damage induced by NaOCl- or heat-treatment of bone [19], [25] and [26], or by culturing OCs on pure mineral [27] and [28]. Together these observations lead to a model (Fig. 7) where the OC starts resorbing along a perpendicular axis to the bone, down to a certain depth, and thereafter continues resorbing parallel to the bone surface. However, since collagenolysis on average is slower than demineralization in cultures of control OCs, most OCs already stop resorbing while still along the perpendicular axis thereby generating a round pit, and not a trench. When collagenolysis is further slowed down compared to demineralization, the resorption stops even sooner resulting in shallower pits. In contrast, when collagenolysis is as fast as demineralization, resorption continues parallel to the surface resulting in continuous resorption trenches.