Prolidase PEPD is a cytosolic enzyme that splits dipeptides

Prolidase (PEPD) is a cytosolic enzyme that splits dipeptides with proline or hydroxyproline at the carboxy terminus and is believed to be important for collagen homeostasis, as high levels of the imino acids are present in collagen (Kitchener and Grunden, 2012). Interestingly, we recently found that recombinant human PEPD (rhPEPD) binds as a homodimer, with high affinity (Kd=7.3nM, based on ELISA), to subdomain 3 in the extracellular domain (ECD) of ERBB2, with each PEPD subunit (molecular mass of 54kD) binding to an ECD, thereby cross-linking two ERBB2 monomers (Yang et al., 2014). This was unexpected, as it had long been believed that ERBB2 exists in a closed conformation and cannot be liganded. Indeed, no ERBB2 ligand had been previously identified. The ability of rhPEPD to cross-link two ERBB2 monomers represents a new ERBB2-binding mechanism, as neither trastuzumab nor pertuzumab (another therapeutic anti-ERBB2 monoclonal antibody) cross-links ERBB2 monomers (Cho et al., 2003; Franklin et al., 2004). Also, no ligands of ERBB2 family members cross-link their respective receptors (Hynes and Lane, 2005; Leahy, 2004). In dub inhibitor overexpressing ERBB2, rhPEPD binds to preformed ERBB2 dimers and blocks ERBB2-SRC signaling by causing SRC disassociation from ERBB2, apparently resulting from alteration of ERBB2 conformation (Yang et al., 2014). SRC plays a key role in ERBB2 oncogenesis (Muthuswamy et al., 1994; Sheffield, 1998; Zhang et al., 2011). rhPEPD also binds to ERBB2 monomers, causing ERBB2 dimerization and activation (Yang et al., 2014). However, such ERBB2 activation is transient and may be functionally insignificant, as rhPEPD-bound ERBB2 is efficiently internalized and degraded, causing ERBB2 depletion, and rhPEPD strongly inhibits the malignant phenotype of cells overexpressing ERBB2, while showing little effect on cells without ERBB2 overexpression (Yang et al., 2014). These results not only reveal a new function of human PEPD, but also show that the protein is primarily an inhibitory ligand of ERBB2 and suggest that rhPEPD may be a potential antitumor agent. Among the ERBB2 family members, rhPEPD does not bind to ERBB3 and ERBB4 (Yang et al., 2014), but also binds to ERBB1 ECD (Yang et al., 2013). Interestingly, despite relatively low binding affinity of rhPEPD (Kd=5.3μM, based on ELISA) towards ERBB1, it modulates the receptor at low nM concentrations, causing transient activation and then depletion of the latter (Yang et al., 2013). Notably, intracellular human PEPD does not modulate ERBB1 and ERBB2, and the dipeptidase activity of rhPEPD is not required for modulation of the receptors (Yang et al., 2013, 2014). In this study, we investigate the ability of rhPEPD to inhibit tumor growth in vivo, whether ERBB2 overexpression is a critical indicator of rhPEPD efficacy, and whether rhPEPD is able to silence ERBB2 signaling in the tumor tissues. Notably, trastuzumab, which binds to subdomain 4 in the ERBB2 ECD (Cho et al., 2003), shows similar effect on ERBB2 to rhPEPD in cultured cells (Cuello et al., 2001; Nagata et al., 2004). However, the ability of trastuzumab to downregulate ERBB2 expression or to inhibit ERBB2 tyrosine phosphorylation in tumor tissues in vivo seems to be limited (Gennari et al., 2004; Gijsen et al., 2010), and its Fc domain may play a significant role in tumor inhibition by engaging Fc receptors on immune effector cells and eliciting antibody-dependent cell-mediated cytotoxicity (ADCC) (Barok et al., 2007; Clynes et al., 2000; Spiridon et al., 2004). In contrast, rhPEPD does not have an Fc domain. We also compare the antitumor efficacy of rhPEPD with that of rhPEPDG278D, a mutant which lacks the dipeptidase activity, in order to assess the relevance of the dipeptidase function for rhPEPD to target ERBB2 and to inhibit tumor growth in vivo.
Materials and Methods
Results
Discussion The tumors in the three mouse models used herein either expressed minimal ERBB2 (CHO-K1) or overexpressed it (CHO-K1/ERBB2 and BT-474), and either grew subcutaneously (CHO-K1 and CHO-K1/ERBB2) or orthotopically in the mammary fat pad (BT-474). We show that rhPEPD strongly inhibits the growth of ERBB2-overexpressing tumors, regardless of tumor location, type or size, but it does not inhibit tumors without ERBB2 overexpression. The response of the tumors to rhPEPD in vivo mimics the response of the same cell lines to rhPEPD in vitro (Yang et al., 2014). However, rhPEPDG278D is more attractive than rhPEPD as a potential antitumor agent for several reasons. First, the antitumor efficacy of rhPEPDG278D is superior to that of rhPEPD, as judged by both partial and complete tumor remission. Second, rhPEPD, rather than rhPEPDG278D, stimulates HIF-1α and its downstream targets (VEGF, GLUT-1) in the tumor tissues via its dipeptidase activity. PEPD was also shown to stimulate TGFβ and its receptor via its dipeptidase function (Surazynski et al., 2010). The stimulating effect of rhPEPD on the prosurvival factors mentioned above likely attenuates its antitumor activity. Third, rhPEPDG278D may not interfere with the physiologic function of endogenous PEPD in normal cells and tissues. In the present study, rhPEPDG278D achieved durable CR in 41.7% (5/12) BT-474 tumors, but it is possible that a higher CR rate may be achieved with dose escalation. Other dipeptidase-defective mutants of human PEPD may also possess strong antitumor activity. However, we previously showed that deletion mutants of human PEPD, which are incapable of forming homodimers, have no ability to bind and modulate ERBB2 (Yang et al., 2014).