Background TNF has been shown to contribute to osteoclastogenesis independently and in conjunction with M-CSF or RANKL, two key cytokines involved in osteoclast (OC) development. We have previously demonstrated that TNF enhances the kinetics of RANKL-induced human osteoclastogenesis and that its effects are mitigated more effectively by the anti-TNF biologic adalimumab (ADA) as compared to etanercept (ETN).
Objectives To determine the mechanism responsible for the difference in effectiveness between the two biologics, a label-free quantitative proteomics study was conducted on TNF-activated human OC upon biologic treatment using an EASY Nano LC1000/ QExactive Plus LC/MS system.
Methods Human OC precursors (OCP) were exposed for 5 days to M-CSF, M-CSF+RANKL (RANKL) alone or in combination with 100 ng/mL TNF +/− 5 ug/mL ADA, ETN or human IgG1 (IGG) as a pre-formed complex. OC differentiation was confirmed by measuring tartrate-resistant acid phosphatase 5b (TRAP 5b) activity. Peptides from cell lysates were generated using modified Filter Aid Sample Preparation (FASP). Quality control sample pooling (sham) strategy was used to minimize false discovery by applying a cut off value (ratio >150% or <66.7%) based on reproducibility between sham samples. Data was analyzed with a 3 database search algorithm through Proteome Discoverer (PD) 2.0 and a 4th algorithm using MaxQuant. Protein IDs were aggregated in Scaffold software using the Local FDR data (LFDR) scoring system. Relative quantification was achieved using precursor ion area extraction function within PD and summarized as top 3 precursor ion intensity in Scaffold. Data is expressed as percentage based on ratio of ion counts of each condition to that of RANKL with differences considered significant if ≥150% only for proteins whose percentage between TNF and IGG conditions are similar.
Results Ten OC-related proteins were identified (CA2, ITGAV, ITGB3, OSTF1, NFATC1, ACP5, CTSK, MMP9, TCIRG1, ATP6V0D2), 5 of which (underlined) were induced by TNF (≥161%), decreased by ADA (≤77%), but unaffected by ETN. Overall, TNF upregulated the levels of 110 proteins as compared to that of RANKL. A greater proportion of these proteins had levels reduced following ADA treatment as compared to ETN (69% and 20%, respectively) concordant with a greater reduction in OC maturation in the presence of ADA as to that of ETN (similar to IGG) (TRAP 5b levels of 11, 40, and 38 U/L, respectively). In addition to lowering the levels of TNF-induced proteins concordant with neutralization by adalimumab, we observed indications for OCP development toward an alternative myeloid lineage based on its ≥1.5-fold induction of 25 proteins that are typically expressed at high levels in monocytic cells (e.g., PSMB9, DPP7, IFI30 & BAG3) including 2 proteins reportedly shown to negatively regulate OC development (LGMN & ZFAND5).
Conclusions Our findings demonstrate that ADA treatment, unlike exposure to ETN, dramatically reduces the TNF-induced protein levels in human OC and moreover, that the TNF:ADA complex may potentially alter their proteome to a profile resembling that of a monocytic cell through an increase in negative regulators of OC development.
Acknowledgement Authors thank Drs. Jochen Salfeld, Anna Yarilina and Nathan Brown for their critical review.
Disclosure of Interest B. Harvey Shareholder of: AbbVie Inc., Employee of: AbbVie Inc., C. Hu Shareholder of: AbbVie Inc., Employee of: AbbVie Inc., D. Wang Shareholder of: AbbVie Inc., Employee of: AbbVie Inc., Y. Tian Shareholder of: AbbVie Inc., Employee of: AbbVie Inc., Z. Kaymakcalan Shareholder of: AbbVie Inc., Employee of: AbbVie Inc.