TI - Discussion . AB - The translocation t ( 11 ; 18 ) ( q21 ; q21 ) is a frequent genomic aberration associated with B-cell MALT lymphoma . Nowadays , it is generally accepted that the API2-MALT1 fusion protein expressed from the der ( 11 ) contributes to B-cell transformation and progression of MALT lymphomas via constitutive NF-kappaB activation . In contrast , not much is known about the reciprocal MALT1-API2 fusion generated on the der ( 18 ) chromosome . Although it was initially reported that MALT1-API2 is not consistently transcribed , we show here that the majority of MALT lymphomas with t ( 11 ; 18 ) ( q21 ; q21 ) do express in-frame MALT1-API2 transcripts . Our data furthermore show that the RING domain of API2 provides MALT1-API2 with an E3 ubiquitin ligase activity . In addition , the inclusion of atleast one Ig-like domain of MALT1 in MALT1-API2 results in the binding and ubiquitination of BCL10 , which triggers its degradation by the proteasome in vitro . However , no correlation was found between the presence of a MALT1-API2 transcript and the BCL10 expression level and nuclear staining in MALT lymphoma samples , most likely due to the efficient downregulation of MALT1-API2 protein levels via auto-ubiquitination . Initial reports could not detect expression of MALT1-API2 in the t ( 11 ; 18 ) ( q21 ; q21 ) -positive MALT lymphoma cases examined [32] -[34] . It is possible that the t ( 11 ; 18 ) ( q21 ; q21 ) is accompanied by deletions or other genomic aberrations that obstruct the generation of a MALT1-API2 fusion on der ( 18 ) . One group indeed reported deletions 5' of the breakpoint on der ( 18 ) in all three MALT lymphoma cases examined [32] . Similarly , while only small intronic deletions were detected in the API2 and/or the MALT1 gene in three t ( 11 ; 18 ) ( q21 ; q21 ) -positive MALT lymphoma cases with expression of MALT1-API2 , the failure to amplify a MALT1-API2 transcript in two other cases correlated with a deletion of a substantial part of 3'-API2 on der ( 18 ) ( [35] and Baens , unpublished results ) . Such deletion involving the N-terminal part of the MALT1 gene or the C-terminal part of the API2 gene is however not a general observation in MALT lymphomas with t ( 11 ; 18 ) ( q21 ; q21 ) , as a more recent genomic characterization of 19 t ( 11 ; 18 ) ( q21 ; q21 ) -positive MALT lymphomas demonstrated in 13 cases the generation of a genomic MALT1-API2 fusion , with only small and mostly intronic deletions , duplications and/or insertions . Only in the 6 cases for which the amplification failed , deletions involving the entire C-terminal API2 and/or the entire N-terminal MALT1 gene were expected [36] . To address this apparent discrepancy between the presence of MALT1-API2 fusions at the genomic level and the absence of MALT1-API2 transcripts , we analyzed in this study MALT1-API2 expression in 14 t ( 11 ; 18 ) ( q21 ; q21 ) -positive MALT lymphoma cases , of which 9 cases were previously reported to contain a MALT1-API2 fusion at the genomic level . Surprisingly , RT-PCR and sequencing confirmed the expression of MALT1-API2 in 7 out of these 9 cases ( Table 1 ) , which clearly shows that transcriptional silencing of MALT1-API2 is not a general phenomenon in t ( 11 ; 18 ) ( q21 ; q21 ) -positive MALT lymphomas . Based on the breakpoints in the API2 and MALT1 gene , a number of API2-MALT1 fusion variants have been described [7] . For all these variants , the anticipated reciprocal MALT1-API2 fusion protein always contains the N-terminal DD of MALT1 and the C-terminal RING domain of API2 . The majority of cases however could contain in addition one or two Ig-like domains of MALT1 ( Fig 1A , the M4A8 and M7A8/M8A8 fusion variants ) , as also demonstrated via genomic characterization of the der ( 18 ) chromosome [35] , [36] . Here we show that the inclusion of atleast one Ig-like domain mediates the binding of MALT1-API2 to BCL10 , conform our previous observation that the DD and the Ig-like domains of MALT1 cooperate for BCL10 binding [26] . In contrast , TRAF6 did not interact with any of the fusion variants , despite the presence of the TRAF6 binding site associated with the second Ig-like domain (T6-Ig) of MALT1 in M7A8 and M8A8 . This probably reflects conformational constraints imposed by the C-terminal domains of API2 , as full-length MALT1 efficiently interacted with endogenous TRAF6 [26] , or indicates the necessity of C-terminal sequences of MALT1 for efficient binding of TRAF6 to this site . In favour of the latter is our observation that deletion of the C-terminus of MALT1 in the API2-MALT1 fusion variant A7M3 also abolished TRAF6 binding to the T6-Ig site [26] . RING domains are able to catalyze the transfer of ubiquitin to a SUBstrate . Our data show that the RING domain of API2 also enables MALT1-API2 to function as an E3 ubiquitin ligase capable of auto-ubiquitination , similarly as API2 . Indeed , API2 is a known E3 protein involved in the ubiquitination-induced degradation of itself [9] , [39] and of interacting proteins like RIP1 [40] , TRAF1 [41] and SMAC [10] . Also BCL10 has recently been reported to be a target of API2 -induced ubiquitination and degradation via binding to the BIR1 domain of API2 [29] . In this study , we were however unable to detect an interaction between API2 and endogenous BCL10 , nor did we observe a RING -dependent effect of API2 overexpression on BCL10 stability . Although this confirms our previous observation that endogenous BCL10 does not interact with the BIR domains of API2 in API2-MALT1 [26] , it needs to be determined whether the use of different cell types underlies this discrepancy or whether the interaction of BCL10 with API2 is stimulus -dependent [29] , [43] . Nevertheless , we did detect a RING domain -dependent ubiquitination of BCL10 by M4A8 , resulting in the degradation of BCL10 by the proteasome . Furthermore , while the inclusion of the second Ig-like domain of MALT1 in M7A8 somewhat increased its capacity to bind BCL10 , this fusion variant could only modestly induce BCL10 ubiquitination and degradation , again suggesting conformational constraints . Given the importance of NF-kappaB activation to the oncogenic behaviour of API2-MALT1 [27] , the effect of MALT1-API2 expression on API2-MALT1 -mediated NF-kappaB activation was examined in vitro . Interestingly , all MALT1-API2 fusion variants were found to severely reduce NF-kappaB activation by API2-MALT1 , partly via the presence of an intact RING domain . This can however not be explained by the RING -mediated ubiquitination and degradation of BCL10 , as BCL10 is not required for NF-kappaB activation by API2-MALT1 [26] . Further studies are thus needed to investigate whether the RING E3 ubiquitin ligase activity of MALT1-API2 affects specific proteins involved in API2-MALT1 -mediated signalling to NF-kappaB . Furthermore , these studies could provide insight into the E3 ubiquitin ligase -independent mechanism of MALT1-API2 to reduce NF-kappaB activation by API2-MALT1 . As this could possibly involve the DD of MALT1 and/or the CARD ( caspase recruitment domain ) of API2 , the identification of interaction partners for these domains would form the subject of interesting future research . Moreover , it remains to be investigated whether expression of MALT1-API2 also reduces API2-MALT1 -mediated NF-kappaB activation in t ( 11 ; 18 ) ( q21 ; q21 ) -positive MALT lymphomas , for example by analyzing its effect on the Lys63 -linked polyubiquitination of IKKgamma [23] . Unfortunately we were unable to do so in this study due to insufficient MALT lymphoma material available . In addition to NF-kappaB activation by API2-MALT1 , also the increased BCL10 protein level [29] and the unusual nuclear localization of BCL10 [30] , [31] have been suggested to contribute to the oncogenic character of the t ( 11 ; 18 ) ( q21 ; q21 ) in MALT lymphomas . The destabilizing effect of MALT1-API2 on BCL10 in vitro thus suggests that MALT1-API2 could also interfere in this way with the development of t ( 11 ; 18 ) ( q21 ; q21 ) -positive MALT lymphomas . However , the expression of MALT1-API2 did not affect the nuclear staining of BCL10 in our MALT lymphoma cases analyzed . In addition , we could not confirm decreased levels of BCL10 in MALT lymphoma B-cells expressing M4A8 , this in sharp contrast to our in vitro observations . As MALT1-API2 is very unstable being an efficient target of its own E3 ubiquitin ligase activity , this suggests that its protein level in MALT lymphoma might be too low to affect BCL10 stability and localization . Unfortunately , we were not able to validate this further due to the lack of good quality antibodies able to detect the MALT1-API2 protein . In conclusion , our data show that the MALT1-API2 fusions with one or both Ig-like domains of MALT1 can bind BCL10 and induce its ubiquitination and proteasome-mediated degradation via the RING domain of API2 . However , MALT1-API2 is also an efficient target of its own E3 ubiquitin ligase activity . This property of MALT1-API2 most likely limits its protein level in t ( 11 ; 18 ) ( q21 ; q21 ) -positive B-cells and makes a possible effect on MALT lymphoma development via destabilization of BCL10 unlikely .