Genetic and biochemical dissection of the ADP-heptose-ALPK1-TIFA innate immune signalling pathway

  • Tom Snelling

Student thesis: Doctoral ThesisDoctor of Philosophy


At the outset of my studies an innate immune signaling pathway had recently been discovered in which the bacterial metabolite ADP-heptose activated the atypical mammalian protein kinase ALPK1, leading to the phosphorylation of TIFA at Thr9, the polymerisation of TIFA, the recruitment of the E3 ligase TRAF6, the activation of the transcription factor NF-κB and the secretion of the chemokine IL-8. The research that I have carried out over the past three years has greatly advanced understanding of this signaling network in the following ways:-

1. I showed that the stimulation of HEK293 cells with ADP-heptose triggered the formation of K63-linked and M1-linked ubiquitin chains within minutes.

2. I found that the expression of TRAF6 but not its E3 ligase activity was required for the activation of canonical IKKs and the MAPKs p38α, p38γ, JNK1 and JNK2.

3. I established that the kinase activity of TAK1 was essential for the ADP-heptosestimulated activation of the canonical IKKs and MAPKs.

4. I demonstrated that ADP-heptose induced the formation of a complex in HEK293 cells that included TRAF6, the components of the TAK1 and IKK complexes, the E3 ligases c-IAP1 and LUBAC and the inactive pseudo-E3 ligase TRAF2.

5. I found that the E3 ligase activity of TRAF6 was not essential because it operates redundantly with c-IAP1 to produce the K63-linked ubiquitin required to activate TAK1. The E3 ligase c-IAP1 is recruited into the signaling complex because it interacts with TRAF2, which is bound to TIFA.

6. I showed that the ADP-heptose-stimulated activation of TAK1 does not occur in HEK293 cells lacking TAB2 and TAB3, which are the regulatory subunits of the TAK1 complex that bind to K63-linked ubiquitin chains. The findings in 1, 5 and 6 are consistent with a model in which K63-linked ubiquitin chains bind to TAB2 and/or TAB3, inducing a conformational change that activates TAK1.

7. Further experiments revealed that activation of the canonical IKKs induced by ADP-heptose was reduced in MEFs in which the M1-linked ubiquitin-generating E3 ligase HOIP, a component of LUBAC, was replaced by an E3 ligase-inactive mutant. The activation of IKKs was abolished in MEFs in which the NEMO regulatory subunit of the canonical IKK complex, which preferentially binds to M1-linked ubiquitin, was replaced by a ubiquitin-binding-deficient mutant. These observations support a model where the binding of M1-linked ubiquitin to NEMO induces a conformational change that facilitates activation of the canonical IKKs.

8. I showed that ADP-heptose stimulated the ubiquitylation of TRAF6, TRAF2 and  c-IAP1 and that in HEK293 cells the ubiquitin chains attached to TRAF6 and  c-IAP1 are largely K63-linked. The ubiquitylation of TRAF6, TRAF2 and c-IAP1 was still observed in HEK293 cells expressing an E3 ligase-inactive mutant of TRAF6 indicating that the ubiquitylation of TRAF6 and c-IAP1 was catalysed by both TRAF6 and c-IAP1 rather than solely by autoubiquitylation.

9. Unexpectedly, I did not detect any ubiquitylation of TRAF6 in IL-1β-stimulated HEK293 cells or in BMDM stimulated with R848, a synthetic activator of the TLR7/8 heterodimer, even though at one time TRAF6 ubiquitylation was thought to be the key driver of these MyD88-dependent signaling pathways.

10. I developed an assay for ALPK1 in which the incorporation of 32P-radioactivity into TIFA from [γ32P]-ATP could be measured in vitro. I demonstrated that ALPK1 phosphorylated TIFA directly at Thr9, and that the mutation of this residue to Ala, Ser, Asp or Glu prevented the binding of TIFA to TRAF6, TRAF2 and c-IAP1. These results showed that ALPK1 phosphorylates TIFA directly and not by activating another protein kinase that then phosphorylates TIFA at Thr9.

11. My studies also revealed that ALPK1 directly phosphorylated TIFA at Thr177 in the TRAF6-binding motif and that the phospho-mimetic mutant TIFA[T177D] impaired the interaction of TIFA with TRAF6 and suppressed the activation of the TAK1 and canonical IKK complexes. This suggested that the phosphorylation of Thr177 may restrict the strength and/or duration of ADP-heptose signaling.

12. The T177D mutant of TIFA did not impair its interaction with TRAF2 or c-IAP1, indicating that the sites that mediate interactions with these proteins are distinct from the TRAF6-binding motif.

13. I found that A20 and ABIN1 were components of the signaling complex formed in response to ADP-heptose. A20 and ABIN1 interact with one another and are known to be ubiquitin-binding proteins that negative regulate MyD88-dependent signaling pathways. I also found that prolonged ADP-heptose signaling induced the disappearance of TIFA from the extracts of HEK293 cells. These results suggest that the recruitment of A20/ABIN1 and the disappearance of TIFA are mechanisms to prevent hyperactivation of the ADP-heptose signaling pathway. 

14. I found that ADP-heptose stimulated the phosphorylation of the IKK-related kinase TBK1 at Ser172, the site required for activation. The phosphorylation of TBK1 was reduced in TRAF6 KO HEK293 cells and abolished in cells lacking both TRAF2 and TRAF6. TBK1 is known to restrict the activation of the canonical IKKs by phosphorylating them at sites that inhibit their activity. This may be a further mechanism to prevent the hyperactivation of ADP-heptose signaling.

15. I found that the ADP-heptose-induced phosphorylation of TBK1 still occurred in TAK1 KO HEK293 cells but was suppressed in these cells by the inclusion of the TBK1 inhibitor GSK8612 in the culture medium. These and other results indicate that the ADP-heptose stimulated phosphorylation of TBK1 was catalysed by the canonical IKK complex and by autophosphorylation.

16. Mutations in ALPK1 cause three diseases. The ALPK1[T237M] mutation causes ROSAH syndrome and the ALPK1[S924P] mutation a case of PFAPA, which are inherited in an autosomal dominant manner. The ALPK1[V1092A] mutation is the cause of 40-50% of all spiradenomas and spiradenocarcinomas. I found that all three mutants were inactive in vitro in the absence of ADP-heptose. In the presence of ADP-heptose, ALPK1[T237M] displayed 32% of the activity of WT ALPK1, but the activities of ALPK1[S924P] and ALPK1[V1092A] were 37% and 47% higher than WT ALPK1, respectively.

17. When overexpressed in HEK293 cells, the ALPK1[V1092A] mutant induced robust NF-κB-dependent gene transcription in the absence of ADP-heptose, but ALPK1[T237M] and ALPK1[S924P] did not. The molecular mechanism of this requires further investigation.
Date of Award2023
Original languageEnglish
SupervisorPhilip Cohen (Supervisor)

Cite this