On 2016 Jan 30, Johannes M Dijkstra commented:

Dear Rachel Wolfson,

Thank you for your extensive answer, and my compliments for your work and attitude.

Johannes Dijkstra

On 2016 Jan 28, Rachel Wolfson commented:

My name is Rachel Wolfson and I am one of the co-first authors on this paper. I am submitting these comments on behalf of all the authors.

1) Yes, we have noticed the shift in the Sestrin2 band. The modification on Sestrin2 is phosphorylation. This phosphorylation seems to be induced by binding to GATOR2, as overexpression of GATOR2 is sufficient to drive the phosphorylation of Sestrin2. We have created mutants of Sestrin2 that are either phospho-mimetic or that do not get phosphorylated. In either case, these mutants bind leucine to similar degrees as wild-type Sestrin2, still bind to GATOR2, and have leucine-regulated interactions with GATOR2. Thus, the phosphorylation of Sestrin2 does not seem to be important for any of the conclusions made in our papers. Furthermore, the leucine binding data and crystal structure were obtained from Sestrin2 purified from bacterial cells, which is not phosphorylated. Future work will be needed to understand the function of Sestrin2 phosphorylation.

2) We agree that understanding how leucine can have such a profound effect on the Sestrin2-GATOR2 interaction is a fascinating question, and this was indeed a major motivation for us to solve the crystal structure of leucine-bound Sestrin2. Below are some points to consider:

A) Due to the highly specific and closed nature of the pocket (observed in Fig. 2A of Saxton et. al), leucine is able to make numerous direct contacts with Sestrin2, including 2 salt bridges and 6 hydrogen bonds, in addition to the hydrophobic van der waals contacts, which all together could contribute several kcal/mol of free energy. This is a non-trivial amount in the context of protein-protein interactions (which are also noncovalent). For an intuitive comparison, consider that the mutation of just two acidic residues (DD406-407AA in Fig. 5 of Saxton et. al) likely corresponding the elimination of 2 salt bridges at the Sestrin2-GATOR2 interface, is sufficient to completely abolish the Sestrin2-GATOR2 interaction.

B) In addition, our model actually suggests that the leucine signal is amplified: not by external factors, but rather by its ability to facilitate a conformational change in Sestrin2 that allows for the formation of additional stabilizing contacts. For example, the closing of the lid and formation of the latch contacts, observed in Fig. 3 of Saxton et. al.

With these considerations it becomes easy to imagine how leucine binding could contribute enough free energy to overcome the Sestrin2-GATOR2 interaction. Importantly, the ability of small molecule mediators to have profound effects on protein-protein interactions is not specific to Sestrin, but is actually a common theme found throughout biological signaling systems and forms the basis for many drug discovery efforts.

3) Thank you for catching the mix up in the blots. Both control blots look nearly identical and while our mistake does not affect any of our conclusions we have asked Science to have the correct panel inserted. Here is an image of the published and corrected figure: http://i.imgur.com/aHT8PV0.png

Rachel Wolfson, Lynne Chantranupong, Robert Saxton, and David Sabatini

On 2016 Jan 28, Rachel Wolfson commented:

None

On 2016 Jan 26, Johannes M Dijkstra commented:

The Sabatini group published two related articles, Wolfson et al. and Saxton et al. Saxton RA, 2016, in the January 1^st edition of Science. They convincingly showed that the interaction between Sestrin2 and GATOR2, which affects mTORC1 activity, is leucine-dependent, and that the Sestrin2 molecule has a specific pocket for binding leucine. I think that overall their study is an important and solid contribution to science. However, I have a few issues/questions in regard to their discussion/presentation of the data.

1. The authors did not discuss that their Western blot analyses suggest that Sestrin2 is represented by multiple differently modified forms, the largest of which is increased by leucine starvation and has the highest affinity for GATOR2 (or FLAG-WDR24). I couldn’t find any mentioning in the two articles by the Sabatini group of the fact that Sestrin2 is represented by multiple, probably differently covalently modified, forms. However, Wolfson Fig. 1A cell lysate anti-Sestrin2 shows that absence of leucine in the cell medium enhances the presence of a larger band, and playing with the contrast of the Fig. 1A IP-FLAG anti-Sestrin2 picture suggests that it is this larger band which efficiently binds to FLAG-WDR24 or its associated proteins (the GATOR2 complex). Once one is aware of this, once can find this phenomenon in many figures of the two papers, of which I will only give a few examples. In Wolfson Fig. 3B cell lysate anti-Sestrin2 picture, the larger Sestrin2 band disappears with larger concentrations of leucine, and this larger band seems to represent the form that is most efficiently precipitated with anti-FLAG-WDR. In Wolfson Fig. 4B cell lysate anti-HA-Sestrin2 picture the relative amount of the larger Sestrin2 band increases when the binding site for leucine was mutated, and similar effects might be present in Saxton Figs. 2D, 3D and 4C; admittedly, in those Saxton figures the cell lysates were not directly analyzed prohibiting a distinction between differences in presence and efficiencies of co-precipitation. Actually, this leucine-dependent (probably covalent) modification of Sestrin2 into a form which is more efficiently bound by the GATOR2 complex, fits beautifully with the major conclusion of the authors that Sestrin2 is a leucine-dependent regulator of GATOR2 activity. However, strangely, the authors did not discuss this phenomenon at all, leave alone that they investigated the nature of the modification(s). The only model that they postulated was an effect of leucine on Sestrin2 conformation (and not on covalent modification).

2. I wonder how non-covalent binding of a single leucine can provide enough energy to directly break the specific interaction between two (much larger) proteins. By showing that leucine can interfere with the interaction between Sestrin2 and GATOR2 when added to cell lysates for 2h at 4°C, Wolfson et al. claim to provide evidence that enzymatic activity is not necessary for explanation of the leucine effect. Within the in vitro experimental setting they are probably right in this, and the in vivo effect on apparent Sestrin2 size (see item 1) is not observed. However, I am not a biochemist, but wonder how non-covalent binding of a single amino acid can provide enough energy for disrupting a specific bond between two proteins. Shouldn’t the leucine signal somehow be amplified? I guess that in vivo it may be amplified by stimulation of some covalent change (see item 1), and that in vitro it may be amplified by the binding of Sestrin2 to additional cellular factors or by the immunoprecipitation procedure. I know that the authors do not particularly exclude the possibility of such amplification, and that their Saxton Fig. 5D model may be a simplified depiction of the core principle. Nevertheless, especially since I don’t know enough about these matters and would like to learn, I would appreciate if the authors and others could discuss this matter. Hence, this item 2 is more a question than a criticism.

3. The pictures portraying anti-FLAG-metap2 in Fig. 1B IP:FLAG and cell lysate seem to be identical. From checking the rest of the papers and because of the nature of the duplication, I deem it an unintentional honest mistake.

In summary, the Sabatini group made very important and believable contributions to elucidation of the mTORC1 pathway system in regard to Sestrin2 structure and its leucine sensitivity. However, they failed to note a very important observation, and their explanation model needs discussion. I would very much appreciate if the Sabatini group, but also other experts on the energy of protein interactions, could discuss the above items 1 and 2.

On 2016 Jan 26, Johannes M Dijkstra commented:

The Sabatini group published two related articles, Wolfson et al. and Saxton et al. Saxton RA, 2016, in the January 1^st edition of Science. They convincingly showed that the interaction between Sestrin2 and GATOR2, which affects mTORC1 activity, is leucine-dependent, and that the Sestrin2 molecule has a specific pocket for binding leucine. I think that overall their study is an important and solid contribution to science. However, I have a few issues/questions in regard to their discussion/presentation of the data.

1. The authors did not discuss that their Western blot analyses suggest that Sestrin2 is represented by multiple differently modified forms, the largest of which is increased by leucine starvation and has the highest affinity for GATOR2 (or FLAG-WDR24). I couldn’t find any mentioning in the two articles by the Sabatini group of the fact that Sestrin2 is represented by multiple, probably differently covalently modified, forms. However, Wolfson Fig. 1A cell lysate anti-Sestrin2 shows that absence of leucine in the cell medium enhances the presence of a larger band, and playing with the contrast of the Fig. 1A IP-FLAG anti-Sestrin2 picture suggests that it is this larger band which efficiently binds to FLAG-WDR24 or its associated proteins (the GATOR2 complex). Once one is aware of this, once can find this phenomenon in many figures of the two papers, of which I will only give a few examples. In Wolfson Fig. 3B cell lysate anti-Sestrin2 picture, the larger Sestrin2 band disappears with larger concentrations of leucine, and this larger band seems to represent the form that is most efficiently precipitated with anti-FLAG-WDR. In Wolfson Fig. 4B cell lysate anti-HA-Sestrin2 picture the relative amount of the larger Sestrin2 band increases when the binding site for leucine was mutated, and similar effects might be present in Saxton Figs. 2D, 3D and 4C; admittedly, in those Saxton figures the cell lysates were not directly analyzed prohibiting a distinction between differences in presence and efficiencies of co-precipitation. Actually, this leucine-dependent (probably covalent) modification of Sestrin2 into a form which is more efficiently bound by the GATOR2 complex, fits beautifully with the major conclusion of the authors that Sestrin2 is a leucine-dependent regulator of GATOR2 activity. However, strangely, the authors did not discuss this phenomenon at all, leave alone that they investigated the nature of the modification(s). The only model that they postulated was an effect of leucine on Sestrin2 conformation (and not on covalent modification).

2. I wonder how non-covalent binding of a single leucine can provide enough energy to directly break the specific interaction between two (much larger) proteins. By showing that leucine can interfere with the interaction between Sestrin2 and GATOR2 when added to cell lysates for 2h at 4°C, Wolfson et al. claim to provide evidence that enzymatic activity is not necessary for explanation of the leucine effect. Within the in vitro experimental setting they are probably right in this, and the in vivo effect on apparent Sestrin2 size (see item 1) is not observed. However, I am not a biochemist, but wonder how non-covalent binding of a single amino acid can provide enough energy for disrupting a specific bond between two proteins. Shouldn’t the leucine signal somehow be amplified? I guess that in vivo it may be amplified by stimulation of some covalent change (see item 1), and that in vitro it may be amplified by the binding of Sestrin2 to additional cellular factors or by the immunoprecipitation procedure. I know that the authors do not particularly exclude the possibility of such amplification, and that their Saxton Fig. 5D model may be a simplified depiction of the core principle. Nevertheless, especially since I don’t know enough about these matters and would like to learn, I would appreciate if the authors and others could discuss this matter. Hence, this item 2 is more a question than a criticism.

3. The pictures portraying anti-FLAG-metap2 in Fig. 1B IP:FLAG and cell lysate seem to be identical. From checking the rest of the papers and because of the nature of the duplication, I deem it an unintentional honest mistake.

In summary, the Sabatini group made very important and believable contributions to elucidation of the mTORC1 pathway system in regard to Sestrin2 structure and its leucine sensitivity. However, they failed to note a very important observation, and their explanation model needs discussion. I would very much appreciate if the Sabatini group, but also other experts on the energy of protein interactions, could discuss the above items 1 and 2.