BCL10, a binding partner of MALT1, is an identified MALT1 substrate. BCL10 is a 233 amino acids polypeptide. Cleavage of BCL10 by MALT1 occurs at the C-terminal end of Arg228. Overexpression of BCL10 induces its hyperphosphorylation. The presence of multiple hyperphosphorylated forms and a mere 5 amino acids difference between the substrate and the proteolytic product made it difficult to tell whether the cleavage event took place by using a simple western blot analysis. BCL10 cleavage was once reported to be monitored with a neo-epitope antibody that recognizes the new C-terminus of BCL10. In this study, we offered a way of monitoring cleavage of BCL10 by using BCL10GFP as substrate for MALT1. As shown in Figure 1, we were able to demonstrate the cleavage of BCL10GFP fusion protein as BCL10 in cells with ectopical expression of MALT1. When overexpressed, BCL10GFP formed cytoplasmic filaments as BCL10 did. Mutation or deletion of the major phosphorylation sites (Ser 134, Ser 136, Ser 138, Ser 141, Ser 144, Ser 170, and Ser 171) on BCL10 or BCL10GFP abolished the appearance of hyperphosphorylated forms as well (data not shown). All the data indicated that the fusion of GFP tag on BCL10 did not jeopardize its ability of being phosphorylated or as a substrate for MALT1. Hence, BCL10GFP can be utilized as a convenient tool for studying the determinants for efficient MALT1 cleavage in HEK293T cells.
BCL10 binds directly and constitutively to the DD and the first two Ig-like regions of MALT1[2, 3, 10, 19]. Consistent with others, deletion of the DD of MALT1 (MALT1 127-824) had little influence on its protease activity. Deletion of both DD and Ig-like domains (MALT1 306-824), abolishing the interaction with BCL10, greatly affected its proteolytic activity on BCL10GFP (Figure 2B). However, DD and the first two Ig-like domains of MALT1 are not required for activity in vitro using peptide substrates such as the BCL10-based substrate Ac-LRSR-amc[17, 20]. IAP2-MALT1 fusion oncoprotein induces cleavage of NIK. NIK is recruited to the IAP2 moiety of IAP2-MALT1 fusion protein and is cleaved by the MALT1 paracaspse domain. Cleavage of NIK is not observed in cells expressing activated MALT1. Therefore, for cleavage of natural protein substrates as opposed to peptide substrates, interactions between protease and substrate are critical.
Based on sequence alignment with caspases, metacaspases, miscellaneous eukaryotic and bacterial cysteine proteases, His415 and Cys464 of MALT1 were identified as the possible catalytic diad. Indeed, mutation of these two residues abolished its catalytic activity (Figure 2C). MALT1 1-548 failed to process BCL10GFP. In this context, an extension of catalytic active site into a region C-terminal to the cited caspase-like-domain is suggested. These results coincided with two recent studies indicating that deletion of domains C-terminal to the caspase-like domain jeopardized its stability[16, 17]. In fact, the caspase-like domain and the following Ig domain are shown to appear as a single folding unit.
The expression of MALT1 alone is insufficient to induce proteolytic activation. The activation of MALT1 can be achieved by overexpression of BCL10[11, 12] in cells. Homotypic interactions between the CARD domain results in the oligomerization of BCL10, which in turn mediates the oligomerization and activation of MALT1. Deletion (C’BCL10GFP) or mutation (BCL10L41RGFP) of the CARD domain failed to activate MALT1. Hence, both mutants were not processed in vivo due to the failure of generating a functional MALT1 paracaspase. BCL10L41R was still a substrate for activated MALT1. In vitro, kosmotropic salts mediated the oligomerization of MALT1 and the activation of its catalytic activity, bypassing BCL10-mediated oligomerization. Hence, BCL10L41R was processed by MALT1 in vitro in the presence of kosmotropic salts.
Unlike the related metacaspases, which accept both Lys and Arg as a P1 residue, the analysis of crystal structure of MALT1 paracaspase region suggests a degree of conserved substrate specificity with Arg invariant at the P1 position[16, 17]. The positively charged P1 residue Arg is coordinated by three highly conserved acidic residues in the S1 pocket. The carboxylate side chain of each of the three acidic residues accepts a pair of charge-stabilized hydrogen bonds from the guanidinium group of P1-Arg. Our results that a mutation of Arg228 to Asp, Phe, Gly, Lys, or Ile of BCL10GFP abolished its activity of being processed by MALT1 are consistent with these findings. A mutation of P2-Ser227 to Ala227 on BCL10GFP had no effect on its ability of being a substrate for MALT1, coincided with the finding that P2 position was moderately selective for amino acids with small side chains due to a restricted S2 pocket. A mutation to Gly at the P3 Arg226 (BCL10R226GGFP) retained its ability of being processed by MALT1. The result was in agreement with the notion that there was small contribution of P3 residue for substrate recognition.
The shallow, hydrophobic S4 pocket of MALT1 is open to accommodate hydrophobic residues. Hence, preference for a hydrophobic residue at P4 position is suggested[16, 17]. While the manuscript was in preparation, Hachmann et al. reported the optimal cleavage sequence for MALT1 using positional-scanning peptidyl substrate libraries. Leucine was shown to be strongly favored in the P4 position. The observation that serial deletion mutants BCL10C2GFP, BCL10C3GFP and BCL10C4GFP, though retaining the P1-Arg228, lost their abilities of being MALT1 substrates might simply be due to the loss of a hydrophobic P4 residue. BCL10Δ225LGFP utilizing Pro224 as the cryptic P4 residue lost its ability of being processed by MALT1. A serial of BCL10GFP mutants (L225A, L225G, L225D, L225E, L225Q, L225T, L225R, L225K) with mutation at the P4-Leu225 residue were tested. All but one (BCL10L225RGFP) lost their abilities of being substrates for MALT1. The observation that BCL10L225R could be processed by MALT1 using in vivo and in vitro assay seemed to be conflicted with the notion that polar residues in the P4 position were not tolerated. However, the truncated products from BCL10L225R processed by MALT1 migrated faster than those from wild type BCL10 processed by MALT1 (Figure 3E). Mutation at Phe222 to His, Ser, or Glu on BCL10L225R abolished the cleavage (data not shown). Therefore, a cryptic cleavage site at the carboxyl-end of L225R was highly suspected.
Using BCL10, the natural substrate of MALT1, as a tool, the major determinants for MALT1 cleavage are consistent with what were found in vitro with peptidyl substrates. Since it was not possible to screen for preferences in P1’ positions (C-terminal to the scissile bond) using peptidyl substrate, our system utilizing BCL10GFP can be a good tool to address the issue.