A central role of IKK2 and TPL2 in JNK activation and viral B-cell transformation

IκB kinase 2 (IKK2) is well known for its pivotal role as a mediator of the canonical NF-κB pathway, which has important functions in inflammation and immunity, but also in cancer. Here we identify a novel and critical function of IKK2 and its co-factor NEMO in the activation of oncogenic c-Jun N-terminal kinase (JNK) signaling, induced by the latent membrane protein 1 (LMP1) of Epstein-Barr virus (EBV). Independent of its kinase activity, the TGFβ-activated kinase 1 (TAK1) mediates LMP1 signaling complex formation, NEMO ubiquitination and subsequent IKK2 activation. The tumor progression locus 2 (TPL2) kinase is induced by LMP1 via IKK2 and transmits JNK activation signals downstream of IKK2. The IKK2-TPL2-JNK axis is specific for LMP1 and differs from TNFα, Interleukin−1 and CD40 signaling. This pathway mediates essential LMP1 survival signals in EBV-transformed human B cells and post-transplant lymphoma, and thus qualifies as a target for treatment of EBV-induced cancer.

The experiments in this report provide a comprehensive analysis of this pathway and are well controlled, typically taking different approaches (such as gene knockouts vs shRNA vs inhibitors) to confirm a finding. The data contained here is a mixture of new findings about the effect of LMP1 on this signalling pathway (such as the kinase activity independent role of TAK1 discussed above) together with confirmation that previously established functions (such as IKK2 mediated activation of TPL2) are operating. However, it is known that different mechanisms of activating the IKK complex exist dependent upon the nature of the cell stimulus and receptor complex being used. As such, while these findings provide valuable insights into LMP1 function, they are not in themselves particularly novel from the perspective of NF-kB/IKK signalling.

Specific comments
(1) As the authors themselves comment, the signalling model identified here may play also operate with other viral or cellular oncogenes. Demonstrating whether this is in fact the case (or not) would strengthen the conclusions of the manuscript by placing this data more firmly in the wider context of cancer biology.
(2) The authors make use of a number of gene knockout cell lines. A good experiment, not performed, is to reexpress the gene that has been deleted to demonstrate that this recovers the phenotype seen. Moreover, re-expression of kinase dead versions of these proteins (especially TAK1) would help confirm many conclusions (such as the kinase independent role of TAK1) that currently rely on inhibitors where off target effects are always a concern. Moreover, by not reexpressing mutant proteins, the authors have missed an opportunity to better define the functional domains of proteins that lead to this novel LMP1 dependent signalling module.
(3) There is an over-reliance on MEFs and HEK 293 cells as experimental tools in this manuscript. Figure 6 does look at the BL41 Burkitt's lymphoma B cell line while Figure 7 used EBV transformed cells but the experiments performed are entirely with kinase inhibitors (and so for example the kinase independent role for TAK1 proposed is not confirmed in this more relevant setting for EBV transformation). In Figure 7, only the TPL2 inhibitor is tested (with Supp Fig 4 looking at IKK2 inhibition). An experiment that confirmed that the novel mechanistic aspects of this report were occurring in the context of an appropriate EBV transformed cell would again strongly strengthen the conclusions of this report.
Minor comments (4) In Figure 3d, the TAK1crKO cells appear to have a reduced level of NEMO expression. Is this a general effect and is it also seen when cells are treated with the TAK1 inhibitor? Is this reduction a cause of or consequence of the reduction in NEMO ubiquitination seen in the TAK1crKO cells?
(5) The role of p105 in TPL2 activation is entirely inferred from over-expression in HEK293 cells (Fig. 4d). What is the effect of depleting p105 in other cell systems with active LMP1?
(6) In line 330 the authors discuss previous work using BAY 11-7082, which is described as an IKK/NF-kB inhibitor. While this compound does inhibit the pathway, more recent work has shown that it is not a direct IKK inhibitor but in fact is an inhibitor of ubiquitin E2 ligases (and therefore has very widespread effects).
(7) The authors should provide evidence that the left and right panels in Fig 7a derive from the same gel and from the same western blot exposure time.
Reviewer #2 (Remarks to the Author): Voigt et al. describe robust, interesting findings that the LMP1 oncogene of Epstein-Barr Virus (EBV) activates JNK via a formerly unknown pathway. They provide strong data that LMP1 activates TAK1 which activates IKK2 which activates TPL2 to activate JNK. They show too that the role of TAK1 in this signaling is independent of its kinase activity. They conclude their study with data indicating that this signaling by LMP1 is needed for the survival of EBV-transformed cells. There are two modifications that would strengthen their findings.
The bulk of the data presented is images of western blots. While this data appears to be "black and white", it is not quantified so that there is little/no statistical analysis of it. The authors need to quantify their results and where appropriate provide error analysis to support their conclusions.
The authors use three cell lines to examine the proliferation and survival of LMP1-positve cells in the presence of an inhibitor of TPL2. Two of these are B-cell lines that have been in culture for years; the third is a murine cell that has been propagated in vivo and/or in vitro for many passages. The authors' suggestion that inhibiting TPL2 could be therapeutically useful for EBVassociated cancers would be substantially strengthened if they test recently isolated EBV-positive tumor cells for their survival/proliferation when LMP1 and TPL2 are each independently inhibited.

Reviewer #1 (Remarks to the Author):
This manuscript describes how the EBV LMP1 protein can 'remodel' cell signalling pathways to activate JNK and NF-kB signalling. In response to cytokine signalling, activation of the IKK complex typically requires TAK1 mediated phosphorylation of IKKbeta. However, although LMP1 still requires TAK1 for IKK2 activation, there is apparently no longer any requirement for TAK1 kinase activity in this LMP1 mediated process. Rather, the authors propose that the most likely mechanism is the induction of a conformational change, mediated through the association of the IKK complex with TAK1 and LMP1, that results in trans autophosphorylation by IKK2 (although proof of such a mechanism is not provided).

new Figure 3d
Activated IKK2 results in activation of the TPL2 kinase (through an established mechanism) and this together with TAK1 (in a manner now requiring TAK1 kinase activity) results in activation of JNK. Active JNK in turn is required for B-cell transformation by EBV. The authors propose that their data reveals TPL2 to be a novel target for the treatment of EBV associated cancer.
The experiments in this report provide a comprehensive analysis of this pathway and are well controlled, typically taking different approaches (such as gene knockouts vs shRNA vs inhibitors) to confirm a finding. The data contained here is a mixture of new findings about the effect of LMP1 on this signalling pathway (such as the kinase activity independent role of TAK1 discussed above) together with confirmation that previously established functions (such as IKK2 mediated activation of TPL2) are operating. However, it is known that different mechanisms of activating the IKK complex exist dependent upon the nature of the cell stimulus and receptor complex being used. As such, while these findings provide valuable insights into LMP1 function, they are not in themselves particularly novel from the perspective of NF-kB/IKK signalling.

Specific comments
(1) As the authors themselves comment, the signalling model identified here may play also operate with other viral or cellular oncogenes. Demonstrating whether this is in fact the case (or not) would strengthen the conclusions of the manuscript by placing this data more firmly in the wider context of cancer biology.
(2) The authors make use of a number of gene knockout cell lines. A good experiment, not performed, is to reexpress the gene that has been deleted to demonstrate that this recovers the phenotype seen. Moreover, re-expression of kinase dead versions of these proteins (especially TAK1) would help confirm many conclusions (such as the kinase independent role of TAK1) that currently rely on inhibitors where off target effects are always a concern. Moreover, by not re-expressing mutant proteins, the authors have missed an opportunity to better define the functional domains of proteins that lead to this novel LMP1 dependent signalling module.
no κ new suppl. Figure 3a (3) There is an over-reliance on MEFs and HEK 293 cells as experimental tools in this manuscript. Figure 6 does look at the BL41 Burkitt's lymphoma B cell line while Figure 7 used EBV transformed cells but the experiments performed are entirely with kinase inhibitors (and so for example the kinase independent role for TAK1 proposed is not confirmed in this more relevant setting for EBV transformation). In Figure 7, only the TPL2 inhibitor is tested (with Supp Fig 4 looking at IKK2 inhibition). An experiment that confirmed that the novel mechanistic aspects of this report were occurring in the context of an appropriate EBV transformed cell would again strongly strengthen the conclusions of this report.  The authors have addressed most of the concerns in their revised manuscript but failed to address their lack of validation of their first six figures with statistical support. They argue that "(retrospective) densitometric quantification of immunoblots (films) is not considered an accurate method of linear quantification any more." They do not have to use any method they deem inaccurate; what they cannot do is to ignore the problem. The data they present in the first six figures needs to be validated with biological replicates and statistical analysis of the replicates. Only if they find that the differences they now claim are, in fact, supported by the statistical analysis, can they continue to claim them. The authors, perhaps, are also disingenuous in their response to the reviews in that they do present measurements from their blots in one case, Figure  3C (without detailing how they obtained these numbers nor providing a statistical analysis of them).

Reviewer #1 (Remarks to the Author):
I am happy with the revised manuscript and the authors response to my original review. I have no further concerms.

Reviewer #2 (Remarks to the Author):
The authors have addressed most of the concerns in their revised manuscript but failed to address their lack of validation of their first six figures with statistical support. They argue that "(retrospective) densitometric quantification of immunoblots (films) is not considered an accurate method of linear quantification any more." They do not have to use any method they deem inaccurate; what they cannot do is to ignore the problem. The data they present in the first six figures needs to be validated with biological replicates and statistical analysis of the replicates. Only if they find that the differences they now claim are, in fact, supported by the statistical analysis, can they continue to claim them.
The authors, perhaps, are also disingenuous in their response to the reviews in that they do present measurements from their blots in one case, Figure 3C (without detailing how they obtained these numbers nor providing a statistical analysis of them).