The present article argues that the Abl ligand imatinib binds to the ATP, not allosteric, site and that this leads to disassembly of the regulatory core of Abl. The authors show NMR titration data that indicate tight binding of the ligand at a protein concentration of 79 μM. The data are presented in response to a paragraph in a recent article by Xie et al. (2022), who report not having found any evidence of a shift in conformational equilibrium when imatinib binds to the catalytic pocket. There appears to be a clear discrepancy in the interpretation of data regarding the Abl-imatinib complex, which the authors of the present article attribute to different protein constructs used. Imatinib is an approved anti-cancer drug (Gleevec), i.e. the questions raised are important.

The binding curves of Figure 1B are characteristic of slow exchange on the chemical shift time scale, indicative of a dissociation constant at least 100-fold smaller than the protein concentration (reported to be 79 μM). Still, a more detailed argument would help convince the reader that an exchange rate < 200 s^-1 is expected for nanomolar affinities and not compatible with micromolar affinities (line 176).

Arguably, the spectral data shown in the present article do not, by themselves, identify the site of binding, as ligand binding can cause changes in chemical shifts all over the place, if the target protein is an allosterically active, flexible protein. Can the argument for binding to the ATP site be strengthened? Previous research seems to have clearly identified the catalytic site as the site of tight binding and, to my understanding, the publication by Xie et al. (2022) does not claim anything else.

Granted that binding is in the ATP site, it is still difficult to follow the authors’ conclusion that ‘the imatinib-induced opening of Abl’s regulatory core is caused by imatinib binding to the ATP site’. What exactly is the evidence of opening of the structure? This deserves better explanation.

Section 3.3 reads like a referee report, which would be unusual in a regular article. I am certain that the technical queries can easily be answered by the authors of the article by Xie et al. (2022). In this case, I suggest to remove the entire section prior to publication of the article in Magn. Reson. If the queries remain unanswered, the section should still be condensed to a few sentences to simply and concisely state the open questions.

One may wonder why the communication between the authors of the articles is so difficult?

In general, it is a regrettable trend that many high-profile articles present some of the data with incomplete descriptions, which prevents others from reproducing the results with reasonable effort. Referees and authors should do better.

In this vein, the authors of the present article could be encouraged to show the complete TROSY spectra as supporting information, for the benefit of readers who do not wish to access the data deposited in Zenodo but are still interested to get an impression of their quality, as the construct used was very big (452 residues) and the protein, which was apparently only labelled with ¹⁵N, relatively low in concentration (79 μM).

Minor suggestions:

Line 14: remove ‘some’

Line 15: should it be ‘Although’ instead of ‘Albeit’?

Lines 11 and 19: conventional style would be to abbreviate journal names to Proc. Natl. Acad. Sci. and J. Am. Chem. Soc. and omit issue numbers.

Line 35: P1 for the article by Xie et al. is an unfortunate abbreviation, as P1 is later used to refer to a protein state (Fig. 1D). The different articles may be better referred to as A1, A2 etc.

Line 37: what is 1b numbering -  a reference might help.

Line 54: why ‘now’, if no new information has been provided by the authors of P1 since publication of the P1 article?

Line 58: The way the sentence starting with ‘However’ reads, it sounds as if the reference to a past article is sold as an experimental result. It may be less confusing for the reader to start with ‘We regret that the authors of P1 …’

Fig. 1: use uniform font sizes for the figure titles, axis labels and residue numberings in the structure.

Line 80: no deuteration?

Thoughout the text, legibility would improve if variables like K_DA were written in italics style (as in Fig. 1D).

Line 116: it appears to me that the argumentation regarding binding to the allosteric site at concentrations above 100 μM is correct. Would it be correct to say ‘While high concentrations can lead to binding of imatinib to the allosteric site, in the following we discuss and provide evidence that binding of imatinib to the catalytic site is sufficient to disassemble the regulatory core.’

Line 119: instead of starting a sentence with ‘However’, it often is better style to start with ‘In contrast’.

Line 131: explain ‘DFG out’

Line 158: ‘when’ instead of ‘since’ would be clearer.

Line 160: ‘reported’ instead of ‘have observed’

Line 162: ‘broadened beyond detection’ instead of ‘bleached out’

Line 174: for all these residues

A recent PNAS article has come to the attention of this reviewer that may deserve consideration by the authors. The article reports the results of nanoBRET experiments in HEK293T cells to measure the dissociation of imatinib from wild-type Abl fused with NanoLuc luciferase in vivo. The half-life time of the complex was determined to be about 20 min. (Lyczek et al., 2021), in agreement with the slow exchange regime indicated by the NMR experiments of the present article. Interestingly, some of the imatinib resistance mutants showed practically unchanged IC₅₀ values, but their dissociation rates were increased, suggesting that the more rapid decrease of inhibition can lead to lesser efficacy in patients despite nominally unchanged dosage.

Lyczek, A., Berger, B.-T., Rangwala, A. M., Paung, Y., Tom, J., Philipose, H., Guo, J., Albanese, S. K., Robers, M. B., Knapp, S., Chodera, J. D., and Seeliger, M., Mutation in Abl kinase with altered drug-binding kinetics indicates a novel mechanism of imatinib resistance, Proc. Natl. Acad. Sci., 118, e2111451118, https://doi.org/10.1073/pnas.2111451118j1of10, 2021.

In this manuscript, Grzesiek and co-workers describe new experimental data and use previously published data to critically evaluate a manuscript by the Kalodimos group in JMB. Correcting flawed data in the literature is extremely important for the scientific community and an integral part of our work, and, hence, this manuscript deserves much attention (as well as a critical reading).

It is a pity that the JMB editor(s) have decided not to consider the manuscript on their own pages, but it is also my experience that journals are hesitant to publish critical views of manuscripts earlier reported on their pages.

The matter of debate here is whether imatinib, a small molecule that acts as an ATP-site inhibitor of the Abl kinase, binds to the myristoyl binding site, in addition to the ATP binding site. There are two conflicting models: the Grzesiek group has claimed that binding to the ATP site (nM affinity) leads to opening of the regulatory core. In contrast, the Kalodimos group has proposed that the binding may occur at an allosteric site instead of the orthosteric site, and that this binding may lead to disassembly of the core and thus activation.

These two models (nM-affinity binding to the ATP-site or microM-affinity binding to an allosteric site) shall have very different consequences for NMR titrations. And the authors of the present manuscript go and do exactly this experiment: a titration experiment. The data, shown in Figure 1 A, B, are extremely convincing to me, and they show that binding occurs at the orthosteric site, rather than the allosteric site.

In itself, this experiment clarifies the debate: binding is in agreement with nM affinity binding, but not with uM allosteric binding. This new data is, thus, important to the field.

However, the data do not directly show that the binding occurs at the ATP site. The groups seem to agree that nM binding is necessarily at the ATP site. But could there be any direct NMR evidence for this? I wonder if the NMR chemical shifts during the titration provides hints to the binding site. Has this been done in the previous papers? This should be clarified.

I also agree with the additional problematic points that the authors listed (lines 190 – 225; see some more specific comments below). It would be nice if these points could be clarified with Kalodimos’ group before the present manuscript gets out.

My conclusion is, thus, that this manuscript should be published. The ideal place would be JMB, and Kalodimos and co-workers may want to have a chance to reply to the criticism (although the case seems really clear to me). But if JMB editors feel like they are not ethically bound to clarifying/correcting what is on their pages, then Magn. Reson. is a possible place.

The question is whether the current manuscript can be made more easy to understand and read. It would be tremendously helpful if the structure was displayed and the important structural features labeled somewhere. For example, the “A-loop”, and “pushing forces” onto that loop, as well as the helix alpha-I are described in the text; likewise, the authors write about the “DFG out” state. But without seeing where these things are located, it is difficult. It would already help to label Figure 1C accordingly, and a scheme like the one of Figure 1A and 4A of reference P1 (the Kalodimos paper); some scheme to show the binding sites and the changes (disassembly, loops being pushed, helix alpha-I kinked etc).

And then the critical question is which observations have lead Kalodimos’ group into conclusions which the present data show as incorrect. Now the authors provide some bits of information why the Kalodimos group may have gotten it wrong (e.g. ITC not done on the relevant constructs).

I would find it very helpful if the text was structured as follows:

• Show and discuss the structures/sketch/models that highlight the divergent views of the two groups
• Highlight what the two models imply, e.g. for a titration experiment
• Discuss the new titration experiment
• Discuss the possibly reasons why the Kalodimos group may have come to wrong conclusions (ITC, short constructs,….)
• Discuss the further problematic points, which in my opinion are all good
• “Reply” to the criticism that Kalodimos raised against Grzesiek’s work (which is now the first part of 3.1.)

Personally, it would find this much nicer to read and understand. Currently, one needs to read the JMB paper, and possibly read the present manuscript more than once to get the point.

Another important point is that the authors make the new data totally open – something that lacks in Kalodimos’ paper. It is nice to see that the titration spectra are all deposited on zenodo. It would be nice to have them additionally also as figures. I would like to see the full spectra, and/or multiple zooms, rather than only four examples.

Below are further points that are comments or suggestions for improvements.

Point 3. on lines 209-211 is valid: the concentrations are an important information. In addition, having looked at the original paper P1, and the Supp Info of the P1 paper by Kalodimos, I could not really identify that peak in the full spectrum, and I would have liked seeing the full spectra of all the states of Figure 4B of P1. It is a poor practice to show only a zoom onto one peak; it does not allow getting a comprehensive picture. The authors may want to make this point.

The same is true for the point 4, which refers to Figure 5. Again, the authors of P1 have decided not to show the full spectra, which is really a pity. It is also noteworthy that the identity of the spectrum shown in orange color in Figure 5 of P1 is not specified.

Line 165: “It also needs to be indicated that the obtained KD solution structures (Xie et al., 2020), on which the argument is based, are of low definition having at most 3 NOE and 2 dihedral (the origin of which is not documented) constraints per residue. Hence, a well-defined A-loop conformation cannot be postulated without additional evidence.» This is indeed an important and compelling argument by the authors. It would be useful to state here explicitly where they have obtained the information about the number of constraints. I assume they retrieved the constraints in the BMRB/PDB.

On lines 171-173, the titration data are described, and four residues for which a new peak appears are named. I would find it useful to state explicitly which residues, in addition to the named ones, show a second peak. Moreover, please show the full spectra in the Supplementary Information.

On lines 197-200, it is stated that «often» the description in P1 is insufficient. I would recommend to be as specific as possible, because a direct criticisms calls for being very specific. I propose that the authors explicitly name occurrences where information is lacking.

Methods: The protein concentration, specified as “79 uM” appears to be very precise, but I wonder if this precision is real. “Ca. 80 uM” (with some error estimate?) is most likely more realistic.

Figure 6 of P1 contains an interesting y-label in a plot, “Lorem ipsum”. Not a sign of very careful work; but probably nothing that can be addressed here.

The manuscript by Grzesiek and coworkers is a response to a recent publication by the Kalodimos lab (Xie et al JMB 2022) who concluded that imatinib, a well-established nanomolar ATP-site inhibitor of Abl kinase, can also bind to the allosteric myristoyl site, although with 10 uM affinity. Xie et al concluded that binding to the myristoyl and not the ATP site may be responsible for the imatinib-induced opening of the Abl regulatory core, which would suggest an alternative mechanisms and might rationalize the effect of imatinib on ATP-site resistance mutations.

The response by Stephan Grzesiek and coworkers, presents a number of clear and convincing argumentsm why the conclusions put forward by Xie et al are not supported by data presented and likely are linked to the use of truncated protein constructs. The manuscript presented here provides additional support for their previous work, which had shown that imabinib binding to the ATP-site indeed induces opening of the regulatory core.

There are key arguments and data presented that suggest that several aspects of the work by Xie et al are flawed, and that reconfirm the previous work by Grzesiek and Jahnke groups. Key points made are:

• Xie et al do not show key results using the full regulatory core kinase construct. Rather, evidence is based on a crystal structure that shows binding of imatinib to the allosteric site, however, using a truncated fragment of the protein, which in fact deletes parts of helix alpha1 that regulates the myristoyl site. This is suggests a crystallization artifact.
• No evidence was shown using solution techniques with the full regulatory core by Xie et al. Grzesiek et al now show NMR titrations that are fully consistent with binding of imatinib to the ATP-site and inconsistent with significant binding occupancy of the myristoyl site in solution.
• This is also consistent with mass-action analysis of binding site occupancies of imatinib with known nM Kds to the ATP-site and the 10 uM Kd reported by Xie et al, which shows that the higher affinity binding site for the ATP sites is dominant.
• Previous competition experiments using a 19F labeled reporter (Skora/Jahnke,2017) with the full regulatory core has established that imatinib does not compete with a 43 uM binding reporter to the allosteric site.

Additional arguments and data presented and missing experimental details and incorrect labeling in the paper published by Xie et al, indicate that the conclusions made by Xie et al likely are linked to the observation of a binding site in a crystal structure with a truncated kinase fragment and lack of rigor and consistency in analyzing the experimental data.

It is very unfortunate and dissappointing, that apparently neither JMB editors nor the corresponding author of the manuscript published in JMB, have responded to the arguments presented here. An editorial response and an offer to present the arguments in a public letter to JMB would be the correct way to address this situation.

In any case, the arguments presented are fully convincing and should be published as soon as possible to make the scientific community aware of the issues with the conclusions presented by Xie et al.