Designer Adaptor Proteins for Functional Conversion of Peptides to Small-Molecule Ligands toward In-Cell Catalytic Protein Modification

Peptides are privileged ligands for diverse biomacromolecules, including proteins; however, their utility is often limited due to low membrane permeability and in-cell instability. Here, we report peptide ligand-inserted eDHFR (PLIED) fusion protein as a universal adaptor for targeting proteins of interest (POI) with cell-permeable and stable synthetic functional small molecules (SFSM). PLIED binds to POI through the peptide moiety, properly orienting its eDHFR moiety, which then recruits trimethoprim (TMP)-conjugated SFSM to POI. Using a lysine-acylating BAHA catalyst as SFSM, we demonstrate that POI (MDM2 and chromatin histone) are post-translationally and synthetically acetylated at specific lysine residues. The residue-selectivity is predictable in an atomic resolution from molecular dynamics simulations of the POI/PLIED/TMP-BAHA (MTX was used as a TMP model) ternary complex. This designer adaptor approach universally enables functional conversion of impermeable peptide ligands to permeable small-molecule ligands, thus expanding the in-cell toolbox of chemical biology.

view, the manuscript would actually be improved were this part of the study simply removed.Otherwise, they will need to perform many more genomics type experiments to solidify all this which I think is not the point of this disclosure -even if they did do this, ul�mately this would be of rather low impact given that the H2BK120ub-H3K79me2 axis is already so well understood.Rather, I think the study would be greatly strengthened by further development of the types of experiments in Figure 5 where other acyl units containing clickable handles are installed using the PLIED/CAT system.For example, can they get PLIED/CAT to work on other proteins in cells such as MDM2, or even in the test tube?The authors also men�on that this could be used for "chromosome analysis" or incorpora�ng crosslinkers in proteins such as histones.Perhaps this could be developed a bit more since this seems very promising.
In conclusion, I think from a protein modifica�on perspec�ve the study is quite innovate.However, I feel that the arguments put forward for the u�lity of the system in cells, par�cularly for the H2BK120 acyla�on stuff, are somewhat weak given the data currently provided.

Reviewer: 2
Comments to the Author In this manuscript �tled "Designer adaptor proteins for func�onal conversion of pep�des to smallmolecule ligands", Fujimura et al presents a method (PLIED/CAT system) to site-specifically incorporate acyl group into proteins that can be targeted by known pep�de ligands.The development of the PLIED/CAT system is an extension of the previous BAHA system developed by the authors, which used small molecule ligands to guide the lysine acyla�on on protein of interest.In this manuscript, the authors engineered one target protein of their BAHA system -eDHFR and incorporated different pep�de ligands into the protein, the ligands can subsequently guide the cataly�c lysine acyla�on to happen on their target protein.By predic�ng binding models, the specificity of the acyla�on can be controlled to some extent.As examples, site-specific incorpora�on of acetyla�on on MDM2 and histone H2B were demonstrated using MBP and LANA as their pep�de ligands, respec�vely.Following the development of the PLIED/CAT system, the authors also demonstrated the applica�on of the system in living cells.The authors showed that synthe�c H2BK120 acetyla�on by their method can work as a protec�ng group at H2BK120 and thus inhibits H2BK120 ubiqui�na�on in cells.Finally, the authors demonstrated that the system can be used to incorporate other acyla�ons, such as butyryla�on and unnatural acyla�on that contains azide or alkyne group.Overall, Kanai and coworkers introduce a novel method to incorporate acyla�ons and the manuscript is well-writen.I have some concerns about the results of the experiment, which I listed below:

1.
The catalyst effec�ve region (CER) is ~12Å, within which any lysine residues are modified by the acetyl group.Therefore, there is a poor labelling selec�vity in the residues located within the CER.In this ar�cle, the authors u�lized the PLIED-L52 system to obtain the H2BK120ub both in vitro and in living cell.However, other lysine residues in close proximity to H2B K120, e.g., H2B K116, are significantly acetylated by more than 10%, as shown in Figures 3g, h, and 4c.Therefore, although H2BK120 has the highest acetyla�on level, it is s�ll important not to overlook the biological implica�ons of the other acetyla�on on neighboring sites.In the discussion, the author should comment on the unique biological impact from H2BK120ub decrease in living cells.

2.
In figure 4h, le� panel, the author concludes that the H2BK120ub level in heterochroma�c gene (Oct4 and Nanog) was not significantly affected by PLIED-L52 acetyla�on.But, there is a significant increase of ub level in Nanog.In addi�on, in figure 4h, right panel, a significant decrease in methyla�on level on heterochroma�c gene Oct4 and Nanog was observed.According to the author's conclusion, the methyla�on of Oct4 and Nanog should be unchanged because the ub levels of those genes remain unchanged.The author should repeat the Chip experiment to have high confidence results.If the results s�ll change significantly, the author should discuss possible reasons.

3.
In Supplementary Fig4, why results of the rela�ve enrichment (H2BK120ac/H3) of the genes could support the statement of "equal level of H2BK120 acetyla�on"?Besides, the author should quan�ta�vely compare changes in H2BK120ac level before and a�er PLIED/1 treatment.Theore�cally, it should be possible to observe a significant increase in ac levels in euchroma�n genes and no change in ac levels in heterochroma�n genes a�er PLIED/1 treatment.

4.
For the CHIP experiments, can the authors jus�fy the selec�on of these genes in more detail?In addi�on, the author should further show the data treatment of all the Chip experiments.

5.
The PLIED/CAT system relies on the expression of e.coli DHFR in living cells.Is there any impact on the cells when this protein is overexpressed?This should be demonstrated at two or three common cell lines.
6. Introduc�on of a pep�de ligand into eDHFR protein might either affect the folding of eDHFR (thus affect its binding to TMP) or change the secondary structure of the pep�de ligand (thus affect its binding to protein of interest), both circumstances may affect the applica�on of PLIED/CAT system.Please comment.
7. In the MDM2 example, the authors choose to replace G51-G56 of eDHFR by the pep�de ligands (page 6, line 5).However, in the histone example, the authors insert LANA between eDHFR G51 and R52 (page 7, line45).What are the considera�ons behind this difference?8.The �tle "Designer adaptor proteins for func�onal conversion of pep�des to small-molecule ligands" overstates the scope of the manuscript.The conversion of pep�des to small molecule is limited to the use of BAHA system previously developed by the authors.9. Please provide representa�ve MS/MS spectrum for analysis of acyla�on sites.We are grateful to you and the reviewers for carefully examining our manuscript ("Designer adaptor proteins for functional conversion of peptides to small-molecule ligands toward in-cell catalytic protein modification" by Akiko Fujimura, Hisashi Ishida, Tamiko Nozaki, Shuhei Terada, Yuto Azumaya, Tadashi Ishiguro, Yugo R. Kamimura, Tomoya Kujirai, Hitoshi Kurumizaka, Hidetoshi Kono, Kenzo Yamatsugu, Shigehiro A. Kawashima, and Motomu Kanai, Manuscript ID: oc-2023-009305).We are pleased that both of two reviewers have indicated that our study, if properly revised, is worthy of being published in ACS Central Science.We are now submitting a revised manuscript that has been modified to address all the concerns raised by the reviewers.
A point-by-point response to all the comments by the reviewers is attached.The major changes are: 1) We have synthesized a new acyl donor 6, containing diazirine crosslinker and an alkyne moiety, and showed that our system can be used for incorporation of various acyl units containing clickable handles as well as a diazirine crosslinker on histones and MDM2.This revision has addressed concerns raised by reviewer #1, and has been included in new Figure S5.2) Reviewer #1 suggested that our manuscript would be improved by removing the ChIP data in Figure 4h, since the ChIP data was not the point of our manuscript.We agree with reviewer #1's opinion and have decided to remove previous Figure 4h and related sentences from our manuscript.3) We have changed the title to "Designer adaptor proteins for functional conversion of peptides to smallmolecule ligands toward in-cell catalytic protein modification.".This revision has addressed concerns raised by reviewer #2.
These additional data and changes have improved our manuscript and strengthened our conclusions.We do hope that you find this revised version of our manuscript to be suitable for publication in ACS Central Science.
Thank you for your time and effort in handling our manuscript.

Motomu Kanai
Email: kanai@mol.f.u-tokyo.ac.jpProfessor Graduate School of Pharmaceutical Sciences, The University of Tokyo
We are pleased that you have indicated that our study, if properly revised, is worthy of being published in ACS Central Science.A point-by-point response to your comments is atached below.

Comments:
In this manuscript, the Kanai group build on their previous efforts to develop tools for the chemical modifica�on of protein lysine sidechains using affinity-directed acyla�on catalysts.In the current installment, they have combined their previously reported methotrexate-based catalyst, TMP-BAHA, with engineered DHFR constructs containing appended or embedded pep�de ligands that serve as targe�ng vectors for proteins of interest (POI), in the present study MDM2 and histones.Basically, the idea is that the pep�de directs the DHFR, and by extension TMP-BAHA, to the POI such that proximal lysine in the complex become acylated when a boronate containing acyl donor is added to the system.This is undoubtedly a clever idea, albeit one that has a lot of moving parts.Nonetheless, the system seems to work based on a very nice series of in vitro inves�ga�ons in which structural modeling and MD simula�ons are used to design the DHFR fusions (what they refer to as PLIED/CAT system).Indeed, there is remarkably good agreement between the sites of acetyla�on on the POIs found experimentally and those predicted by the modeling.Overall, the in vitro proof of concept studies are quitethorough and the results fully consistent with the design of the system.The authors then go on to test the PLIED/CAT system in mammalian cells, focusing on chroma�n where they use the well-known LANA pep�de as the direc�ng vector -the Kania group has used this pep�de, which binds to the so-called acidic patch of the nucleosome, in their previous work in the general area.Consistent with the in vitro studies, they find the lysine120 on histone H2B is the major site of acetyla�on when the op�mal PLIED system is employed.The data suppor�ng this conclusion are solid.Less convincing are some follow up studies they perform in which they atempt to show that PLIED/CATmediated acetyla�on of H2BK120 impacts ubiquityla�on at this same site as well as H3K79 methyla�on (which has previously been shown to be dependent on H2BK120ub).In par�cular, the chroma�n immunoprecipita�on studies (ChIP) RT-PCR data in Figure 4h are, in this reviewer's opinion, over interpreted given the errors associated with the reported measurements; the fact that H3K79me2 levels for Oct4 and Nanog are reduced to the same level as GAPDH, RhoB despite that fact that they argue H2BK120ub levels for these silenced genes are not impacted seems inconsistent and probably just reflects data quality.All this makes their argument that PLIED/CAT-mediated acetyla�on of H2BK120 is a beter tool for probing this rela�onship compared to say siRNA methods less than compelling.In my view, the manuscript would actually be improved were this part of the study simply removed.Otherwise, they will need to perform many more genomics type experiments to solidify all this which I think is not the point of this disclosure -even if they did do this, ul�mately this would be of rather low impact given that the H2BK120ubH3K79me2 axis is already so well understood.
Thank you for carefully reading our manuscript.We appreciate for your insigh�ul comments and completely agree with your opinions.As you suggested, we decided to remove ChIP-qPCR data in the previous Figure 4h and related sentences in the text.
Rather, I think the study would be greatly strengthened by further development of the types of experiments in Figure 5 where other acyl units containing clickable handles are installed using the PLIED/CAT system.For example, can they get PLIED/CAT to work on other proteins in cells such as MDM2, or even in the test tube?The authors also men�on that this could be used for "chromosome analysis" or incorpora�ng crosslinkers in proteins such as histones.Perhaps this could be developed a bit more since this seems very promising.
Thanks for your great sugges�on.As you men�oned, we examined the PLIED/CAT system for introduc�on of a clickable azide handle on MDM2.Furthermore, we synthesized new acyl donor 6, containing diazirine crosslinker and an alkyne moiety.As expected, our system can be used for incorpora�on of various acyl units containing clickable handles as well as a diazirine crosslinker on histones and MDM2 (see new Figure S5).
In conclusion, I think from a protein modifica�on perspec�ve the study is quite innovate.However, I feel that the arguments put forward for the u�lity of the system in cells, par�cularly for the H2BK120 acyla�on stuff, are somewhat weak given the data currently provided.
Thanks to your truly valuable comments, our revised manuscript is now much improved.

Comments:
In this manuscript �tled "Designer adaptor proteins for func�onal conversion of pep�des to smallmolecule ligands", Fujimura et al presents a method (PLIED/CAT system) to sitespecifically incorporate acyl group into proteins that can be targeted by known pep�de ligands.The development of the PLIED/CAT system is an extension of the previous BAHA system developed by the authors, which used small molecule ligands to guide the lysine acyla�on on protein of interest.In this manuscript, the authors engineered one target protein of their BAHA system -eDHFR and incorporated different pep�de ligands into the protein, the ligands can subsequently guide the cataly�c lysine acyla�on to happen on their target protein.By predic�ng binding models, the specificity of the acyla�on can be controlled to some extent.As examples, site-specific incorpora�on of acetyla�on on MDM2 and histone H2B were demonstrated using MBP and LANA as their pep�de ligands, respec�vely.Following the development of the PLIED/CAT system, the authors also demonstrated the applica�on of the system in living cells.The authors showed that synthe�c H2BK120 acetyla�on by their method can work as a protec�ng group at H2BK120 and thus inhibits H2BK120 ubiqui�na�on in cells.Finally, the authors demonstrated that the system can be used to incorporate other acyla�ons, such as butyryla�on and unnatural acyla�on that contains azide or alkyne group.Overall, Kanai and coworkers introduce a novel method to incorporate acyla�ons and the manuscript is well-writen.
I have some concerns about the results of the experiment, which I listed below: Thank you for carefully reading our manuscript.We are pleased that you have indicated that our study, if properly revised, is worthy of being published in ACS Central Science.A point-by-point response to your comments is atached below.

1.
The catalyst effec�ve region (CER) is ~12Å, within which any lysine residues are modified by the acetyl group.Therefore, there is a poor labelling selec�vity in the residues located within the CER.In this ar�cle, the authors u�lized the PLIED-L52 system to obtain the H2BK120ub both in vitro and in living cell.However, other lysine residues in close proximity to H2B K120, e.g., H2B K116, are significantly acetylated by more than 10%, as shown in Figures 3g, h, and 4c.Therefore, although H2BK120 has the highest acetyla�on level, it is s�ll important not to overlook the biological implica�ons of the other acetyla�on on neighboring sites.In the discussion, the author should comment on the unique biological impact from H2BK120ub decrease in living cells.
We appreciate your insigh�ul comments.As you suggested, we have men�oned minor but s�ll significant acetyla�on on neighboring sites, in the Discussion part (page 12, lane 2325) as follows; "It should be noted, however, that other lysine residues proximate to H2BK120, such as H2BK116, were also acetylated by the PLIED-L52/1 system, which may contribute to H2BK120ub inhibition.").

2.
In figure 4h, le� panel, the author concludes that the H2BK120ub level in heterochroma�c gene (Oct4 and Nanog) was not significantly affected by PLIED-L52 acetyla�on.But, there is a significant increase of ub level in Nanog.In addi�on, in figure 4h, right panel, a significant decrease in methyla�on level on heterochroma�c gene Oct4 and Nanog was observed.According to the author's conclusion, the methyla�on of Oct4 and Nanog should be unchanged because the ub levels of those genes remain unchanged.The author should repeat the Chip experiment to have high confidence results.If the results s�ll change significantly, the author should discuss possible reasons.In the previous figure 4h, we divided IP (%) of H2BK120ub (or H3K79me2) by that of H3 and showed the average +/-SD value of three independent experiments.However, thanks to your comments, we realized that this calcula�on method caused increased variability of the data.To make the data treatment simpler, we here showed IP (%) in each experiment as well as the average and SD of three experiments in Figure R1 (a-d for H2BK120ub, e-h for H3K79me2) shown below.

Figure. R1 The original data of ChIP analyses for H2BK120ub and H3K79me2
The results were mostly reproduced among three independent experiments, indica�ng high reliability of these data.The level of H2BK120ub or H3K79me2 in the HOXA10, GAPDH, or Rhob region was reduced by PLIED-L52 acetyla�on in all cases, so our conclusion in the original submission [the H2BK120ub or the H3K79me2 level in euchroma�c gene (HOXA10, GAPDH, Rhob) was notably reduced by PLIED-L52 acetyla�on] is valid.However, we found that our previous descrip�on for heterochroma�c regions were not appropriate, as you pointed out.The H2BK120ub level in Oct4 region was comparable in two experiments, and slightly reduced in one experiment.The H2BK120ub level in Nanog region was increased in two experiments, and comparable in one experiment.Therefore, our previous descrip�on "the H2BK120ub level in heterochromatic gene (Oct4 and Nanog) was not significantly affected by PLIED-L52 acetylation" might be incorrect.Rather, we have revised the sentence as follows; "the H2BK120ub level in heterochromatic gene (Oct4 and Nanog) was not significantly reduced by PLIED-L52 acetylation".The level of H3K79me2 in the heterochroma�c genes (Oct4 and Nanog) was significantly lower than euchroma�c regions, which is consistent with the no�on that H3K79me2 is an ac�ve transcrip�on marker.Therefore, it is difficult to precisely evaluate changes in H3K79me2 level in the heterochroma�c genes.Thus, we have revised our descrip�on as follows; "the H3K79me2 level in heterochromatic gene (Oct4 and Nanog) remained low after PLIED-L52 acetylation.".
Meanwhile, reviewer #1 also commented about the ChIP data and suggested that our manuscript would be improved by removing the ChIP data in previous Fig.4h, since the ChIP data was not the point of our manuscript ("In my view, the manuscript would actually be improved were this part (i.e.ChIP data in Fig. 4h) of the study simply removed.Otherwise, they will need to perform many more genomics type experiments to solidify all this which I think is not the point of this disclosureeven if they did do this, ultimately this would be of rather low impact given that the H2BK120ub-H3K79me2 axis is already so well understood.").We agree with reviewer 1's opinion and have decided to remove previous Fig.4h and related sentences from our manuscript.We would appreciate if you kindly agree with this revision.

3.
In Supplementary Fig4, why results of the rela�ve enrichment (H2BK120ac/H3) of the genes could support the statement of "equal level of H2BK120 acetyla�on"?Besides, the author should quan�ta�vely compare changes in H2BK120ac level before and a�er PLIED/1 treatment.Theore�cally, it should be possible to observe a significant increase in ac levels in euchroma�n genes and no change in ac levels in heterochroma�n genes a�er PLIED/1 treatment.
We apologize that the label was not easy to understand, although the graph in the original Supplementary Figure 4 included H2BK120ac levels with and without PLIED/1 treatment.To prevent misunderstanding, we have made the following four changes; 1) we have changed the label in new Figure S4 as "with PLIED/1 treatment" and "without PLIED/1 treatment".2) The result of H2BK120ac and H3 have been separated.3) We have added "n.d." for the data of not detected.4) We have changed the y-axis to log value.Based on the data, we concluded that H2BK120ac level was almost comparable between euchroma�n and heterochroma�n regions.

4.
For the CHIP experiments, can the authors jus�fy the selec�on of these genes in more detail?In addi�on, the author should further show the data treatment of all the Chip experiments.
Thank you for the comment.We selected HOXA10 as a H2BK120ub-posi�ve region according to the previous report by Shema-Yaacoby, E. et al. in Cell Rep 2013.GAPDH/Rhob and Oct4/Nanog were selected as they are well-known representa�ve euchroma�n and condi�onal heterochroma�n regions, respec�vely.For these regions, we used the primer sets that were previously reported by David, Y. et al. in Nat Chem 2015.Regarding the data treatment, we previously divided IP (%) of histone modifica�on (H2BL120ub, H3K79me2, H2BK120ac) by that of H3, which caused increased variability of the data.To make the data treatment simpler, we now showed IP (%) of histone modifica�on and H3 separately (new Figure S4 and Fig. R1).

5.
The PLIED/CAT system relies on the expression of e.coli DHFR in living cells.Is there any impact on the cells when this protein is overexpressed?This should be demonstrated at two or three common cell lines.
We checked if the overexpression of E.coli DHFR affects cell viability in HEK293T and HeLa S3 cells, which are commonly used cell lines.Please see new Figure S3D.The data showed that the overexpression of eDHFR (34 hours a�er transfec�on: the same �mescale as in-cell histone acetyla�on experiments in Figure 4F, 4G) did not affect viability in both cell lines.As controls, we used an empty vector (vehicle) as well as the EGFP plasmid.We have added the following sentence to lane 7-8 in page 10; The overexpression of eDHFR within this timescale did not affect cell viability in HEK293T and HeLa S3 cells (Figure S3D).

6.
Introduc�on of a pep�de ligand into eDHFR protein might either affect the folding of eDHFR (thus affect its binding to TMP) or change the secondary structure of the pep�de ligand (thus affect its binding to protein of interest), both circumstances may affect the applica�on of PLIED/CAT system.

Please comment.
Thank you for the insigh�ul comment.Based on EMSA assay shown in Figure 3D and Figure S2D, LANA-eDHFR, eDHFR-LANA, or PLIED-L23 has weak or no affinity to nucleosomes, sugges�ng that the posi�on of a pep�de ligand introduc�on affects the secondary structure of the pep�de ligand.In addi�on, as shown in Figure S2B, PLIED-L36 was less soluble than other PLIEDs, sugges�ng that LANA inser�on to posi�on 36 affected the folding of eDHFR.PLIED-L51 and PLIED-L52 were soluble and significantly bound to nucleosomes.Therefore, the selec�on of a proper posi�on for pep�de ligand introduc�on is important to construct PLIED/CAT system.We have added the following sentence to lane 16-18 in page 8; Therefore, the insertion of LANA to an appropriate position of eDHFR is important to construct the functional PLIED/CAT system bearing properly folded eDHFR and a peptide ligand.

7.
In the MDM2 example, the authors choose to replace G51-G56 of eDHFR by the pep�de ligands (page 6, line 5).However, in the histone example, the authors insert LANA between eDHFR G51 and R52 (page 7, line45).What are the considera�ons behind this difference?
In the MDM2 example, the binding structures of MDM2 and MBPs (Figure 2D) showed that the binding mo�f of MBPs is α-helix, whose length is approximately 14.4-16.1 Å; 16.1 Å for MBP1, 14.5 Å for MBP2, and 14.4 Å for MBP3.Therefore, we replaced G51-G56 of eDHFR, whose length is approximately 15.4 Å, by the pep�de ligands.In the histone example, LANA was inserted between G51 and R52 of eDHFR, since LANA has a hairpinlike structure (Figure 3A) and the distance between its N and C termini is approximately 5.6 Å.
We have included these considera�ons at lane 3-4 in page 6 for MDM2 and lane 28-29 in page 7 for histones in the main text.

8.
The �tle "Designer adaptor proteins for func�onal conversion of pep�des to smallmolecule ligands" overstates the scope of the manuscript.The conversion of pep�des to small molecule is limited to the use of BAHA system previously developed by the authors.Thank you for the sugges�on.We have changed the �tle to "Designer adaptor proteins for func�onal conversion of pep�des to small-molecule ligands toward in-cell cataly�c protein modifica�on".

9.
Please provide representa�ve MS/MS spectrum for analysis of acyla�on sites.We provided representa�ve MS/MS spectrum for acetyla�on sites analysis in Figure 2C