Moreover, we present the crystal structures of cIAP1-BIR3 and XIAP-BIR3 domains in the presence of 9a, describing the molecular details of divalent Smac-mimetic recognition. Taken together, all the experimental evidences here reported suggest that 9a is one of the most powerful divalent Smac-mimetics known to date; the structural analysis of its recognition patterns, here presented, is the basis for further optimization in terms of target affinity and bioavailability.
Results Cellular Cytotoxicity
Preliminary cytotoxicity tests of 9a after 72 hours of treatment versus MDA-MB-231 (a breast cancer cell line that has repeatedly been used to test Smac-mimetics/XIAP inhibitors), HL60 (known to be Smac-mimetic sensitive), and PC-3 cells (as an example of Smac-mimetic refractory cells), were addressed [14]. 9a showed nanomolar cytotoxicity both in MDA-MB-231 and HL60 cell lines, whereas it was inactive against the PC-3 cell line, as expected. The relative IC50 values observed are shown in Table 1.Caspase Activation and cIAP Degradation
To test the capability of inducing caspase activation and apoptosis, MDA-MB-231 cells were treated with 9a, or left untreated. 9a not only inhibited cell growth in the MDA-MB-231 cell line, but Western blot analysis showed activation of caspase-8, -3 and -9, and apoptosis (Fig. 2). Moreover, Western blot analysis shows that 9a induces degradation of cIAP1 and of cIAP2 (Fig. 2), already at 30 min post treatment.
Fluorescence Polarization Assays
Binding and displacement assays based on fluorescent polarization were used to evaluate the affinities of 9a for human cIAP1-, cIAP2-, XIAP-BIR3 and XIAP-BIR2BIR3 domains. Saturation binding experiments were performed to determine the binding affinity of the fluorescent probes to the IAP constructs of interest,Figure 1. Chemical structure of tail-tail dimer 9a. The left inset shows a framed structure of the divalent Smac-mimetics based on the 1-aza-2oxobicyclo[5.3.0]decane scaffold.Table 1. Cytotoxic activity displayed by 9a on MDA-MB-231, HL60 and PC-3 cell lines, determined in three independent experiments (each done in triplicate).as previously reported [11,15,16]. Competitive binding assays revealed that 9a displayed low nanomolar IC50 values for all tested IAP constructs (Table 1).
Analytical Gel Filtration
In order to check whether simultaneous interactions of the divalent inhibitor with two cIAP1-BIR3 or XIAP-BIR3 domains could take place in solution, we performed analytical gel filtration assays mixing the different protein domains (33 mM) with a large excess of 9a (1 mM). The chromatograms obtained for XIAPBIR3 in the presence of 9a exhibited a shift of 1 mL in elution volume (Ve = 10.7 mL) relative to the untreated protein (Ve = 11.7 mL), revealing domain dimerization upon ligand binding (Fig. 3A). In order to exclude that the dimerization of XIAP-BIR3 could be due to the formation of an intermolecular disulfide bridge involving residue Cys351 (induced by the presence of 9a), we performed an analytical gel filtration on the XIAP-BIR3 Cys351Ser mutant, in the absence/presence of the divalent compound, obtaining the same results reported for the wild type protein (Ve = 11.7/10.7 mL, respectively). In contrast, the chromatogram of XIAP-BIR2BIR3 exhibited a slight peak shift (0.3 mL) toward a higher elution volume in the presence of 9a, indicating a more compact conformation of the protein (Fig. 3B). The simultaneous binding of the two heads of divalent 9a to BIR2 and BIR3 domains, resulting in a decrease of their mutual distance, may explain the observed Ve shift (see also the SAXS data below). As expected, analytical gel filtration performed in the presence of a monovalent moiety of 9a did not show any peak shift relative to the apo-proteins. Surprisingly, analytical gel filtration assays performed on cIAP1BIR3 did not show any peak shift in the presence of an excess of 9a. To shed light on such behavior, we performed new gel filtration experiments mixing equimolar amounts (33 mM) of the proteins (XIAP- or cIAP1-BIR3) and of 9a. The resulting chromatograms showed for XIAP-BIR3/9a the same peak shift observed in the presence of an excess of inhibitor, and a peak shift of about 1.3 mL for cIAP1-BIR3/9a (from Ve = 12.9 mL to Ve = 11.6 mL, Fig. 3C). Such apparently contradictory results can be explained taking into account the different affinities of the cIAP1- and XIAP-BIR3 domains for 9a in two different states: 1) when the ligand is free in solution; 2) when the ligand is already bound to one BIR3 domain, as summarized by the dissociation constants K1 and K2: a K1 K2 Bza u BazB u BaB; K1 ~ ; K2 ~ a aB B = BIR3, a = 9a In fact, after mixing together the protein and a large excess of the inhibitor, an expected behavior (K1,K2) will be the saturation of all the available BIR3 domains by one head of 9a, hampering the formation of dimers (no variations in Ve), as observed for cIAP1BIR3.
Figure 2. Western blot of MDA-MB-231 cells untreated or treated with 9a. Upper part: cIAP1 and cIAP2 degradation in the MDA-MB-231 cell line after 30 min and 6h of treatment with 9a. Proteins were revealed by Western blot using polyclonal antibodies specific for cIAP1 and cIAP2. Lower part: activation of caspase-8 (arrows indicate p55, p43/41 and p18 forms), -9 (p37) and -3 (p17 and p19) by 9a (p89). Proteins were revealed by Western blot using rabbit polyclonal antibodies specific for cleaved Parp, and cleaved caspase-8, -9 and -3. Prestained Protein SHARPMASS V (112250 kDa; EuroClone) was used as molecular weight marker. Figure 3. Analytical Gel Filtration Chromatograms. A) XIAP-BIR3 (33 mM) in absence/presence of an excess (1 mM) of 9a. B) XIAP-BIR2BIR3 in absence/presence of an excess (1 mM) of 9a. C) cIAP-BIR3 in absence/presence of an equal amount (33 mM) of 9a.F shifted toward dimer (BaB) formation, as observed for XIAP-BIR3 (variation in Ve). On the other hand, when the amount of inhibitor is comparable with that of the protein, there will be an equilibrium between the monomeric and dimeric adducts even if K1,K2 (Fig. 3C). As a whole, the analytical gel filtration results indicate that 9a is able to bind simultaneously two BIR domains (either BIR2 or BIR3), and bring them to a relatively compact (dimeric for XIAP-BIR3 and cIAP1-BIR3) structure.Crystal Structures of XIAP- and cIAP1-BIR3 Bound to 9a
Overall structure. The binding of 9a to the BIR3 domains of XIAP and cIAPs was investigated through X-ray crystallography. The 3D structures of cIAP1- and XIAP-BIR3 complexes with the ligand were solved through the molecular replacement method, using search models based on the BIR3 structure of cIAP1 (pdb: 3MUP [15]) and of XIAP (pdb: 3CLX [11]) and ??refined at 2.6 A and 3.3 A resolution, respectively (Table 2). In the cIAP1-BIR3 structure, the crystal displays four BIR3 domains and two molecules of 9a in the asymmetric unit, forming a ring-like assembly composed of two dimers (AC and BD; Fig. 4A shows the AC dimer). The crystal packing is the same observed for the structure of cIAP1-BIR3 bound to a monovalent Smacmimetic (pdb: 3MUP; [15]), suggesting that such intermolecular arrangement is independent of the presence of the divalent compound. The four independent BIR3 domains display very ?similar structures, showing r.m.s.d. values in the 0.29 ?0.46 A range (calculated over 101 Ca pairs).
