https://www.cell.com/molecular-cell/fulltext/S1097-2765(26)00198-X
chatGPT:
Here’s a concise review of the uploaded paper, Structural basis for TORC2 activation
Summary
This paper asks how TORC2 is turned on at the plasma membrane. The authors purify endogenous yeast TORC2 and solve a very high-resolution cryo-EM structure, with focused maps reaching about 2.2 Å. That lets them place essentially all subunits and several previously unresolved or flexible regions more confidently than earlier work.
The central mechanistic finding is that the Avo1 PH domain can sit directly in the Tor2 catalytic cleft. On page 8, the structural model shows this PH domain contacting both kinase lobes through two loops positioned so that they would interfere with substrate access. In other words, the PH domain is acting like a gate near the active site rather than just a passive lipid-binding appendage.
But the story is more subtle than “PH domain equals inhibition.” When the authors delete the Avo1 PH domain, TORC2 does not become constitutively active. Instead, basal signaling drops and TORC2 can no longer mount proper activation responses to hypoosmotic shock or myriocin. Their structural comparison suggests why: when the PH domain is absent from the cleft, the active site narrows and the Tor2 negative regulatory domain, NRD, can intrude into the active site. So the PH domain seems to both restrain access in one state and prevent collapse into a more deeply inhibited state in another.
They then connect this to lipids. The Avo1 PH domain has a positively charged PI(4,5)P2-binding surface, but in the “PH-in” conformation that pocket is buried in the active site and inaccessible to membrane lipid. Mutating that pocket reduces PI(4,5)P2 binding in vitro and blocks TORC2 activation in cells, supporting a model in which PI(4,5)P2 pulls the PH domain out of the cleft and thereby promotes activation.
The paper also identifies a positively charged membrane-facing surface on TORC2, centered on Avo3 and Tor2, as a likely plasma-membrane interaction patch. Within that region, a positively charged Avo3 HD pocket is required for full activation, even though mutating it does not obviously abolish membrane localization. That argues the pocket is more about activation/allostery than mere recruitment.
A further contribution is a model for substrate delivery. The Avo1 CRIM domain is positioned flexibly near the active site and, with AlphaFold-guided modeling, is proposed to rotate inward while carrying substrate such as Ypk1 into the catalytic cleft. The paper therefore proposes a multi-step activation scheme: PH-in resting state, PI(4,5)P2-induced primed state, then CRIM-assisted substrate delivery into an active state. Figure 7 on page 11 summarizes this clearly.
What is novel
The biggest novelty is the direct structural observation of the Avo1 PH domain in the Tor2 active-site cleft. Earlier models had suggested something like this for mSin1/mTORC2, but this paper provides direct structural evidence in yeast TORC2.
A second novelty is the reinterpretation of the PH domain as a bidirectional regulator. Rather than being simply inhibitory, it appears to stabilize an activation-competent architecture by preventing active-site closure and NRD intrusion, while still needing to move away for full substrate entry. That is a more interesting mechanism than a simple on/off plug.
A third novelty is the structural definition of a likely membrane-binding PM patch and the identification of the Avo3 HD pocket as an activation-relevant feature. This gives the field a concrete candidate site for lipid-mediated allostery in TORC2.
The paper also adds important architectural insights beyond the activation model: it clarifies how Tor2 specificity for TORC2 is distributed across multiple interfaces rather than a single determinant, and it strengthens the case that Bit61 and Protor1 are orthologs despite low sequence similarity.
Critique
The paper is strong overall: the structural work is excellent, the figure logic is good, and the authors tie structure to genetics and signaling phenotypes rather than stopping at description. Still, there are some limits.
The main weakness is that part of the activation model depends on inference from structural classes and AlphaFold modeling, especially for the flexible Avo1 CRIM domain and the proposed substrate-delivery motion. The authors acknowledge this explicitly in their limitations section. So the mechanistic model is plausible and well motivated, but not fully captured in one experimental structure.
Related to that, the putative membrane-bound activated state is not directly visualized for yeast TORC2 in this study. The model of PI(4,5)P2-driven activation is supported by mutagenesis and lipid-binding data, but the key transition—PH withdrawal upon membrane engagement—is still inferred rather than directly seen.
A second limitation is that the work is done on purified endogenous yeast TORC2 outside its native plasma-membrane environment. That is unavoidable for this kind of structure, but important, because membrane tension, local lipid composition, and accessory factors may reshape the conformational landscape in vivo. The authors note this themselves.
Third, the evidence for the Avo3 HD pocket is good functionally but still incomplete mechanistically. Mutation impairs activation without clearly disrupting membrane localization, which is suggestive of allostery, but the paper does not directly show what binds that pocket. PI(4,5)P2 is a reasonable candidate, yet that remains a hypothesis rather than a demonstrated ligand interaction in this system.
Fourth, the claim that activation and inactivation are mechanistically uncoupled is interesting and probably right, but it leaves a major gap. The authors propose that NRD insertion may participate in inhibited states, yet they could not directly test the importance of the Tor2 NRD by deletion. So the inhibitory arm of the model is less securely established than the activation arm.
Finally, the evolutionary extrapolation to mammalian mTORC2 is sensible but should still be treated carefully. Some features look conserved, but others differ, including subunit composition and flexible regions. The paper supports a conserved principle more than a fully demonstrated one-to-one mammalian mechanism.
Bottom line
This is a strong mechanistic structural paper. Its main advance is showing that TORC2 regulation involves a dynamic PH-domain gate coupled to phosphoinositide sensing, membrane engagement, and substrate delivery, rather than simple recruitment of substrate to a constitutively open kinase. The model is convincing in broad outline, though some of the most interesting moving parts still await direct structural capture in the membrane-bound active state.
If you’d like, I can turn this into a more detailed claim-by-claim critique table.