**3. Discussion**

The active site of RTA has been explored extensively as a potential target for antidotes against depurination [32]. However, since SRL is the substrate of RTA, the active site is large and mostly polar and therefore small molecule inhibitor screens have yielded few potential inhibitors with low affinity [32]. Although inhibitors showed activity in enzymatic tests, they failed to protect cells or animals against ricin challenge. Only one small molecule has been shown to have activity in protecting mice against ricin challenge by blocking the retrograde trafficking of ricin [33]. To address this barrier and to establish a starting point for inhibitor discovery, we established the ribosome binding site of RTA as a new target and identified the shortest length of a peptide that can bind to the ribosome binding site of RTA and inhibit its activity.

The C-terminal ends of the ribosomal stalk P proteins interact with a small well-defined hydrophobic pocket on the face of RTA opposite to the active site [27,28]. We show here that peptides derived from the conserved CTD of P proteins can disrupt RTA–ribosome interactions. The longest peptide tested was P11 because this is the smallest peptide reported to inhibit the activity of Stx1 [6]. Since the C-terminal end of this peptide is critical for RTA interaction [27,28], we deleted one amino acid at a time starting from the N-terminal end. We found that the longer peptides had higher affinity and inhibitory activity compared to the shorter peptides. The shortest peptide that could interact with RTA and inhibit its activity corresponded to the last four amino acids of P proteins.

A conserved D106 D107 D108 motif at the N-terminal end of P11 has been shown to be critical for the interaction with trichosanthin (TCS) [34]. In the structure of the TCS–P11 complex, Asp108 of P11 interacted with Lys173 of TCS via salt bridges, while Asp106 of P11 formed hydrogen bonds with Gln169 of TCS [35]. However, in the two structures of this peptide with RTA, the D106 D107 D108 residues were not observed [27,28]. The substitution of Asp108 and Asp106 residues in P2 with alanine abolished the interaction between P2 and TCS [35]. However, these mutations did not affect the interaction of P2 with RTA [28]. Based on these results, Fan et al. concluded that the conserved D106 D107 D108 M109 motif of P2 is not involved in the interaction with RTA and only hydrophobic interactions and hydrogen bonds contribute to the interaction [28]. In contrast, Shi et al. showed that although the D106 D107 D108 motif was not observed in the structure, the binding affinity of RTA measured by isothermal calorimetry (ITC) was lower when this motif was not present on the peptide than when this motif was present [27].

To address the role of the negatively charged motif at the P protein CTD, we deleted these residues one at a time. Our results indicate that individual deletion of Asp107 and Asp108 in the D106 D107 D108 motif had negligible effect on affinity of the peptide for RTA, while deletion of Asp106 had a small effect. This may explain why the GST-tagged P2 variants containing Asp to Ala mutations in this motif interacted with RTA in the pull-down experiments [28]. When affinity was measured directly using Biacore T200, we observed a small decrease in *K*D from 272 μM to about 300 μM when Asp106 was deleted. However, the IC50 increased 2-fold upon deletion of Asp106. The IC50 increased further 2-fold when Asp107 was deleted (Table 2). This data is consistent with our previous study [29] where mutation of positively charged arginines on RTA led to a significant increase in *K* m toward ribosomes without affecting the *K* m or *k*cat towards an RNA mimic of the SRL, indicating that electrostatic contacts contribute to the interaction of RTA with the ribosome [29]. We showed that arginines are critical for maintaining the fast association and dissociation rates of the interaction with the CTD of P proteins [29]. We proposed that these arginines form a positively charged patch on the surface of RTA and interact with the negatively charged D106 D107 D108 motif at the CTD of the P proteins to facilitate the interaction of RTA with the P stalk to allow depurination of the SRL [29]. The results presented here provide direct evidence that the D106 D107 D108 motif is important for ribosome anchoring of RTA for depurination of the SRL.

The qRT-PCR based depurination assay, which directly measures the catalytic activity of RTA on ribosomes showed demonstrable impact on RTAs ability to depurinate ribosomes in a manner which correlated with ribosome binding [36]. Comparison of the binding affinity and the IC50 of RTA for the peptides indicated that the IC50 values were lower than the *K*D values. The P11 with 200 μM *K*D was able to achieve 50% inhibition of RTA activity at 5 μM, which is about 40 times lower than the *K*D. For the shorter peptides, such as P5 and P4, the IC50 (~100 μM) values were about 5 times lower than the *K*D (~500 μM) values. Although peptides do not bind RTA tightly, as indicated by the high *K*D values, they inhibit the activity of RTA, as indicated by the relatively low IC50 values. This difference may be because of allosteric binding sites, where peptides are binding in a location separate from the active site. Thus, peptides do not compete with binding of the active site of RTA to the SRL, but compete with binding of RTA to the P stalk. Our results indicate that inhibition of depurination activity involves both electrostatic and hydrophobic surfaces on the P protein peptide. Electrostatic interactions are critical to maintain the high association and dissociation rates of RTA with the P proteins on the ribosome [28]. Even low affinity binding to the P protein peptide led to a high level of inhibition of depurination by RTA, suggesting that the ribosome binding site is a potentially valuable target distinct from the active site. However, due to the low affinity of the peptides for RTA in combination with the high catalytic efficiency of RTA the therapeutic potential of the peptides used in this study is limited and is not a claim of this paper. They need to be optimized into higher affinity ligands. We establish the ribosome binding site as a potential new target for inhibitor discovery as a proof of concept. Since P protein CTD is the binding site of several RIPs, including the Stxs, inhibitors targeting the ribosome interaction site of RTA could be effective against the Stxs.

The structural analysis showed that the binding between P10 and RTA is mediated by hydrophobic interactions. The Phe111, Leu 113, and Phe114 residues are inserted into a hydrophobic pocket and the Phe114 and Asp115 residues form hydrogen bonds with Arg235 of RTA [27,28]. Consistent with the structural analysis, our results indicated that P5 and P4 showed similar affinity and inhibition activity for RTA (Tables 1 and 2). The binding and depurination inhibition results of P5 over P5b (Figure 4) confirmed that Asp115 plays an important role in the interaction. However, P3 containing both Phe114 and Asp115 only bound very weakly to RTA and could not inhibit RTA activity. Although the conformation of TCS bound to P11 differed from the structure of RTA, both RIPs recognized the Leu113 and Phe114 motif [28]. This LF motif is conserved in both eukaryotic and archaeal ribosomal stalk proteins and has been shown to be necessary for binding of translational GTPases to the stalk proteins [37,38]. Our results sugges<sup>t</sup> that RTA binding to P3 is substantially reduced when Gly112 is deleted possibly because Gly112 accommodates the required backbone to facilitate the insertion of Phe111, Leu113, and Phe114 into the hydrophobic pocket on RTA [28]. The structural analysis showed that the C-terminal sequences of P proteins do not form a stable structure in solution in a ligand-free state [16], but appear as an α-helix upon binding to the hydrophobic pocket of RTA [27,28]. Since 3.6 amino acids are the minimum length needed to form an α-helix, the last three amino acids may not interact well with RTA because they cannot form an α-helix even though they contain all the critical side chains for the interaction. These results indicate that the minimal length of P protein CTD required for binding to RTA and inhibiting its activity is four amino acids.

The C-terminal sequences of the stalk proteins can adapt diverse conformations in order to bind distinct ligands specifically [16]. The P11 appeared as a type II β-turn upon binding to TCS [35]. However, the last six amino acids of P11 formed an α-helix when bound to RTA [27,28]. Similarly, the CTD of archaeal P1 (aP1) formed a β-turn and a 310-helix when bound to eIF5B [20]. In contrast, the CTD of aP1 bound to eIF1A formed a long extended α-helix [37]. The aP1 bound to a hydrophobic pocket on the surface of eIF5B, which is present on the opposite side of the GTP/GDP binding site [20]. It was suggested that the stalk/eIF5B interaction contributes to the recruitment of the GTP binding site of eIF5B to the SRL [20]. Our results indicate that RTA interacts with rapid on and off rates and with low affinity with peptides mimicking the C-terminal sequence of the P proteins. The A1 subunit of Stx1 was also shown to interact with low affinity with P11 [7]. RIPs and translational GTPases may interact with low affinity with the stalk proteins to properly orient their active site and the GTP binding site, respectively, towards the SRL. Thus, our RTA–ribosome interaction model, which proposes that the interaction with the stalk stimulates the catalysis of depurination by orienting the active site of RTA towards the SRL [22], may be applicable to other RIPs and the translational GTPases that interact with the stalk.
