The CTLA‐4 immune checkpoint protein regulates PD‐L1:PD‐1 interaction via transendocytosis of its ligand CD80

Abstract CTLA‐4 and PD‐1 are key immune checkpoint receptors that are targeted in the treatment of cancer. A recently identified physical interaction between the respective ligands, CD80 and PD‐L1, has been shown to block PD‐L1/PD‐1 binding and to prevent PD‐L1 inhibitory functions. Since CTLA‐4 is known to capture and degrade its ligands via transendocytosis, we investigated the interplay between CD80 transendocytosis and CD80/PD‐L1 interaction. We find that transendocytosis of CD80 results in a time‐dependent recovery of PD‐L1 availability that correlates with CD80 removal. Moreover, CD80 transendocytosis is highly specific in that only CD80 is internalised, while its heterodimeric PD‐L1 partner remains on the plasma membrane of the antigen‐presenting cell (APC). CTLA‐4 interactions with CD80 do not appear to be inhibited by PD‐L1, but efficient removal of CD80 requires an intact CTLA‐4 cytoplasmic domain, distinguishing this process from more general trogocytosis and simple CTLA‐4 binding to CD80/PD‐L1 complexes. These data are consistent with CTLA‐4 acting as modulator of PD‐L1:PD‐1 interactions via control of CD80.

1. Important aspects shown in this manuscript have been reported earlier (e.g. by Sugiura et al). This is a rather small study with a limited set of experiments in cell lines. More experimentsespecially functional experiments should have been performed to gain a more comprehensive insight on the role of CTLA-4 mediated transendocytosis of CD80 in PD-1 function to justify publication in a renowned high-impact journal with a broad readership.
2. As pointed out above the major limitation of the study is that it lacks functional data. It would be important to learn how CTLA4 activity impacts on PD-1 inhibition on T cells or T cell lines (in the presence or absence of CD28 costimulation). Without functional experiments it cannot be claimed that transendocytosis of CD80 is required to effectively control PD-L1-PD-1 interactions. figure 1 there is no information provided how often the experiments were repeated and statistical analysis is lacking 4. No experiments with primary APC have been performed. PDL1-CD80 coexpression mainly occurs on professional APC which frequently also express the second ligand for CTLA4 and in many cases also PD-L2 the second ligand for PD-1. Potentially blockade of PDL1 by CD80 engagement could have less impact if PDL2 is present.

For the experiments summarized in
Minor points: In the binding of therapeutic PDL1 antibodies which bind very strongly to PDL1 also blocked by CD80 binding in cis? Figure 7: the authors report that abatacept does not alter PD-1 binding; it can be expected that this is also the case for belatacept but if would be interesting to address this.
Ligand levels: The authors should show fluorophore expressing and ligand expression (e.g. GFP expression and CD80 staining in Fig. EV4) If CD80 and CD86 are coexpressed is CD80 preferentially removed from the APC via CTLA-4 mediated transendocytosis?
Does CD80 has to be expressed in excess of PD-L1 to prevent PD-1-Ig binding or is this observed when both ligands are expressed at equimolar concentrations?

Referee #2
Kennedy et al. show in in vitro experiments 1) CTLA4 on donor cells specially bind to CD80 in the heterodimer of PD-L1:CD80; 2) CTLA4 transendocytosis selectively remove CD80 but not PD-L1; 3) Effective removal of CD80 via transendocytosis requires CTLA4 cytoplasmic domain; 4) Transendocytosis of CD80 by CTLA4 but not blocking soluble CTLA4 Ig effectively liberates PD-L1. The authors conclude that efficient CD80 depletion by transendocytosis but not simply CTLA-4 binding to CD80 is required to effectively liberate PD-L1 that can engage PD-1. These observations are interesting, and the manuscript is clearly and concisely written.
There is a minor concern need to be addressed. Most of the recent publications on cis PD-L1/CD80 interactions are based on in vitro transfection experiments with PD-L1 and CD80. Those reports and the current report all elegantly show that there is only cis PD-L1/CD80 interactions. However, previous experiments (Park et al: Blood 2010;Yi et al J. Immunol. 2011, Deng et al 2015Ni et al: JCI 2017) clearly indicate the existence of trans PD-L1/CD80 interactions in vivo. In vitro interactions cannot be used to exclude complex in vivo interactions. This notion should be made in the discussion.

Referee #3
Sansom review Adding to several recent publications on the cis CD80/PD-L1 interaction, the authors show that transendocytosis by CTLA4 of a CD80/PD-L1 heterodimer removes CD80 but not PD-L1 from the cell surface, allowing the PD-L1 to interact with PD-1. They propose that CD80 can operate as a switch and by preventing PD-L1 from interacting with PD-1 and by its higher avidity binding to CTLA4 also protect CD86 from transendocytosis by CTLA4, the CD80 can promote enhanced T cell responses by inhibiting both CTLA-4 and PD-1 pathways. The work is well done and deepens our understanding of a wicked complicated set of interactions but is with transfectants and not natural cells. No direct experiments on immune function are performed.
Major: 1. Zhao et al propose that the CD80/PD-L1 interaction, by reducing CD80 to a monomeric state, reduces the avidity for CTLA-4. Do a dose titration of CTLA4-Ig on CHO-CD80 and CHO-CD80/PD-L1 cells and determine the EC50 of CTLA4 binding and see if the avidity is altered. This will have to be done on CHO cell transfectants since endogenous CD80 and CD86 would complicate the analysis in DG-75 cells. 2. Specify concentrations on x-axis of Figure 7 (these are DG-75 cells). Calculate EC50 of CTLA4-Ig binding to CD80 vs CD80/PD-L1 from Fig 7A. 3. What is the molar ratio of PD-L1 to CD80 in the transfectants? Which is in excess? This cannot be determined by FACS but could be done by mass spectroscopy. 4. How much CD80, CD86, PD-L1 does the Burkitts lymphoma cell line DG-75 express (generally Burkitts lines express)? Show in a supplemental figure. Is there binding to GFP-less native CD80 or CD86 which may complicate interpretation?
Minor: 5. In introduction, distinguish characteristics of transendocytosis and trogocytosis. How are they distinguished? 6. "acquisition by the CTLA-4+ recipient population using a well-established assay " describe cells used in assay, particularly CTLA-4 recipient cells. 7. Move Figure EV2 to main text so assay is clear to reader at beginning. 8. The labeling of the figures and descriptions in the figure legends is in many places inadequate to clearly understand what has been done and how. There are many examples but to specify a few: 9. In 1D, specify ab used for IP and for WB 10. The labeling of Figure 2 does not provide enough information. The donor and recipient cells for each panel (and their CTV labeling) should be clearly indicated in the figure.
11. In 2A, are the upper left and right panels duplicate assays? 12. Throughout all figures, where "ligand" is written, specify molecule. Do not use "no ligand" 13. Methods says "ratios of donor: recipient cells and incubation times indicated in the figure legends." But figures often do not provide these numbers. Recipient cells need to be specified in figures. 14. In 3C and similar, the left panel y-axis is labeled "ligand remaining". Specify which ligand in each case; presumably CD80? 15. In methods, give equation of how ligand remaining was calculated in Figure 2D and similar figures. 16. In 2A, choice of overnight incubation is a poor one since the amount of detectable transferred ligand is low (presumably because of degradation). Suggest using a time point where degradation is less dominant. 17. In 2B, the question is whether the PD-L1mCherry is being transendocytosed by the CTLA4 cell. Say the red quadrant is loss of PD-L1. There does not seem to be a shift in the CTV positive population but the MFI intensity of the total cell population of the lower (CTV negative) population is clearly shifted to the right. Why does this shift not mean that PD-L1 is being broadly taken up by the recipient cells, maybe on the cell surface and not internalized? 18. Calculate statistical significance between lines in 2D (CD80 vs CD80+PD-L1) 19. "CTLA-4 cytoplasmic domain was required for effective time-dependent depletion of CD80 and CD86 " Soften this statement (required is too extreme). In general, this data is discussed appropriately but the titles are extreme. There is some trogocytosis/transendocytosis by Cytdel (11% vs control of 2%) 20. "The reason for the differences between these observations currently remain unclear but may relate to differences in how transendocytosis was measured. " Describe differences between your method and Zhao's. 21. In the discussion, also note that in natural cells, PD-L1 is associated with CMTM6 and this is not replicated in transfectants. 22. Specify GFP used in methods and linker length. 23. Specify CTLA-4 del36 in methods. I would therefore like to invite you to submit a revised version addressing the concerns as proposed.
When preparing your letter of response to the referees' comments, please bear in mind that this will form part of the Review Process File, and will therefore be available online to the community. For more details on our Transparent Editorial Process, please visit our website: https://www.embopress.org/page/journal/14602075/authorguide#transparentprocess I thank you for the opportunity to consider your work for publication. I look forward to your revision. We realize that it is difficult to revise to a specific deadline. In the interest of protecting the conceptual advance provided by the work, we recommend a revision within 3 months (19th Sep 2022).
As a matter of policy, competing manuscripts published during this period will not negatively impact on our assessment of the conceptual advance presented by your study. However, we request that you contact the editor as soon as possible upon publication of any related work, to discuss how to proceed.
If you require more time to complete the revisions let me know as as I can grant an extension.
Use the link below to submit your revision: https://emboj.msubmit.net/cgi-bin/main.plex PD1 and CTLA4 are important immune checkpoints that are targeted in cancer therapy. PDL1 and CD80, which are ligands for PD1 and CTLA4 respectively interact with each other. This interaction mainly occurs in cis. Previous work by this group has uncovered a process termed transendocytosis where CTLA4 acquires CD80 and CD86 from adjacent cells. In this study. Kennedy et al have investigated how PDL1-CD80 interaction impact the binding of these ligands to their receptors. In line with some of the earlier studies they observed that upon engagement of PDL1 by CD80 in trans binding of PDL1 to PD-1 is blocked whereas CD80 still can bind CTLA4. In addition, the authors show that CTLA4 mediated transendocytosis of CD80 does not result in depletion of PDL1 bound to CD80. Thus CD80-transendocytosis liberates PD-L1 for engagement of PD-1 in a timedependent manner. Earlier work showed that the cytoplasmic tail of CTLA-4 is required for transendocytosis. In this study the authors provide evidence that simple binding of CTLA4 is insufficient to liberate PDL1 and that trogocytosis which also results in the removal of CD80 and which does not require the cytoplasmic tail of CTLA4 is insufficient to liberate PDL1. This is a wellconducted study and the manuscript is well written and the data are clearly presented. However, the study is limited to binding studies and experiments that monitor the expression levels of CD80 and PDL1 on cell lines in presence of cell lines expressing CTLA4. No functional studies have been performed to address the impact of CTLA4 on reinstalling PDL1 mediated inhibition. Consequently, they have not addressed whether full length CTLA4 is really significantly more efficient in reinstalling PDL1-mediated PD1-inhibition than CTLA4 lacking its cytoplasmic tail. This is an important question in this context since CTLA4 lacking its cytoplasmic tail (termed CTLA4 Del36 in this study) is expressed at much higher levels and might thus strongly engage CD80 (and thereby enhance PD1-inhibition via PDL1) despite the fact that it is less efficient in reducing CD80 on adjacent cells than full length CTLA4. Due to this lack functional data the gives limited insight into how the interaction of PDL1 and CD80 in cis and the interaction of CD80 and CTLA4 in trans impact on the PD1 and CTLA4/CD28 pathways. Major concerns: 1. Important aspects shown in this manuscript have been reported earlier (e.g. by Sugiura et al). This is a rather small study with a limited set of experiments in cell lines. More experiments -especially functional experiments should have been performed to gain a more comprehensive insight on the role of CTLA-4 mediated transendocytosis of CD80 in PD-1 function to justify publication in a renowned high-impact journal with a broad readership. 2. As pointed out above the major limitation of the study is that it lacks functional data. It would be important to learn how CTLA4 activity impacts on PD-1 inhibition on T cells or T cell lines (in the presence or absence of CD28 costimulation). Without functional experiments it cannot be claimed that transendocytosis of CD80 is required to effectively control PD-L1-PD-1 interactions. 3. For the experiments summarized in figure 1 there is no information provided how often the experiments were repeated and statistical analysis is lacking 4. No experiments with primary APC have been performed. PDL1-CD80 coexpression mainly occurs on professional APC which frequently also express the second ligand for CTLA4 and in many cases also PD-L2 the second ligand for PD-1. Potentially blockade of PDL1 by CD80 engagement could have less impact if PDL2 is present. Minor points: In the binding of therapeutic PDL1 antibodies which bind very strongly to PDL1 also blocked by CD80 binding in cis? Figure 7: the authors report that abatacept does not alter PD-1 binding; it can be expected that this is also the case for belatacept but if would be interesting to address this. Ligand levels: The authors should show fluorophore expressing and ligand expression (e.g. GFP expression and CD80 staining in Fig. EV4) If CD80 and CD86 are coexpressed is CD80 preferentially removed from the APC via CTLA-4 mediated transendocytosis? Does CD80 has to be expressed in excess of PD-L1 to prevent PD-1-Ig binding or is this observed when both ligands are expressed at equimolar concentrations?
Referee #2: Kennedy et al. show in in vitro experiments 1) CTLA4 on donor cells specially bind to CD80 in the heterodimer of PD-L1:CD80; 2) CTLA4 transendocytosis selectively remove CD80 but not PD-L1; 3) Effective removal of CD80 via transendocytosis requires CTLA4 cytoplasmic domain; 4) Transendocytosis of CD80 by CTLA4 but not blocking soluble CTLA4 Ig effectively liberates PD-L1. The authors conclude that efficient CD80 depletion by transendocytosis but not simply CTLA-4 binding to CD80 is required to effectively liberate PD-L1 that can engage PD-1. These observations are interesting, and the manuscript is clearly and concisely written.
There is a minor concern need to be addressed. Adding to several recent publications on the cis CD80/PD-L1 interaction, the authors show that transendocytosis by CTLA4 of a CD80/PD-L1 heterodimer removes CD80 but not PD-L1 from the cell surface, allowing the PD-L1 to interact with PD-1. They propose that CD80 can operate as a switch and by preventing PD-L1 from interacting with PD-1 and by its higher avidity binding to CTLA4 also protect CD86 from transendocytosis by CTLA4, the CD80 can promote enhanced T cell responses by inhibiting both CTLA-4 and PD-1 pathways. The work is well done and deepens our understanding of a wicked complicated set of interactions but is with transfectants and not natural cells. No direct experiments on immune function are performed.
Major: 1. Zhao et al propose that the CD80/PD-L1 interaction, by reducing CD80 to a monomeric state, reduces the avidity for CTLA-4. Do a dose titration of CTLA4-Ig on CHO-CD80 and CHO-CD80/PD-L1 cells and determine the EC50 of CTLA4 binding and see if the avidity is altered. This will have to be done on CHO cell transfectants since endogenous CD80 and CD86 would complicate the analysis in DG-75 cells. 2. Specify concentrations on x-axis of Figure 7 (these are DG-75 cells). Calculate EC50 of CTLA4-Ig binding to CD80 vs CD80/PD-L1 from Fig 7A. 3. What is the molar ratio of PD-L1 to CD80 in the transfectants? Which is in excess? This cannot be determined by FACS but could be done by mass spectroscopy. 4. How much CD80, CD86, PD-L1 does the Burkitts lymphoma cell line DG-75 express (generally Burkitts lines express)? Show in a supplemental figure. Is there binding to GFP-less native CD80 or CD86 which may complicate interpretation?
Minor: 5. In introduction, distinguish characteristics of transendocytosis and trogocytosis. How are they distinguished? 6. "acquisition by the CTLA-4+ recipient population using a well-established assay " describe cells used in assay, particularly CTLA-4 recipient cells. 7. Move Figure EV2 to main text so assay is clear to reader at beginning. 8. The labeling of the figures and descriptions in the figure legends is in many places inadequate to clearly understand what has been done and how. There are many examples but to specify a few: 9. In 1D, specify ab used for IP and for WB 10. The labeling of Figure 2 does not provide enough information. The donor and recipient cells for each panel (and their CTV labeling) should be clearly indicated in the figure. 11. In 2A, are the upper left and right panels duplicate assays? 12. Throughout all figures, where "ligand" is written, specify molecule. Do not use "no ligand" 13. Methods says "ratios of donor: recipient cells and incubation times indicated in the figure legends." But figures often do not provide these numbers. Recipient cells need to be specified in figures. 14. In 3C and similar, the left panel y-axis is labeled "ligand remaining". Specify which ligand in each case; presumably CD80? 15. In methods, give equation of how ligand remaining was calculated in Figure 2D and similar figures. 16. In 2A, choice of overnight incubation is a poor one since the amount of detectable transferred ligand is low (presumably because of degradation). Suggest using a time point where degradation is less dominant. 17. In 2B, the question is whether the PD-L1mCherry is being transendocytosed by the CTLA4 cell. Say the red quadrant is loss of PD-L1. There does not seem to be a shift in the CTV positive population but the MFI intensity of the total cell population of the lower (CTV negative) population is clearly shifted to the right. Why does this shift not mean that PD-L1 is being broadly taken up by the recipient cells, maybe on the cell surface and not internalized? 18. Calculate statistical significance between lines in 2D (CD80 vs CD80+PD-L1) 19. "CTLA-4 cytoplasmic domain was required for effective time-dependent depletion of CD80 and CD86 " Soften this statement (required is too extreme). In general, this data is discussed appropriately but the titles are extreme. There is some trogocytosis/transendocytosis by Cytdel (11% vs control of 2%) 20. "The reason for the differences between these observations currently remain unclear but may relate to differences in how transendocytosis was measured. " Describe differences between your method and Zhao's. 21. In the discussion, also note that in natural cells, PD-L1 is associated with CMTM6 and this is not replicated in transfectants. 22. Specify GFP used in methods and linker length. 23. Specify CTLA-4 del36 in methods. PD1 and CTLA4 are important immune checkpoints that are targeted in cancer therapy. PDL1 and CD80, which are ligands for PD1 and CTLA4 respectively interact with each other. This interaction mainly occurs in cis. Previous work by this group has uncovered a process termed transendocytosis where CTLA4 acquires CD80 and CD86 from adjacent cells. In this study. Kennedy et al have investigated how PDL1-CD80 interaction impact the binding of these ligands to their receptors. In line with some of the earlier studies they observed that upon engagement of PDL1 by CD80 in trans binding of PDL1 to PD-1 is blocked whereas CD80 still can bind CTLA4. In addition, the authors show that CTLA4 mediated transendocytosis of CD80 does not result in depletion of PDL1 bound to CD80. Thus CD80-transendocytosis liberates PD-L1 for engagement of PD-1 in a time-dependent manner. Earlier work showed that the cytoplasmic tail of CTLA-4 is required for transendocytosis. In this study the authors provide evidence that simple binding of CTLA4 is insufficient to liberate PDL1 and that trogocytosis which also results in the removal of CD80 and which does not require the cytoplasmic tail of CTLA4 is insufficient to liberate PDL1. This is a well-conducted study and the manuscript is well written and the data are clearly presented. However, the study is limited to binding studies and experiments that monitor the expression levels of CD80 and PDL1 on cell lines in presence of cell lines expressing CTLA4. No functional studies have been performed to address the impact of CTLA4 on reinstalling PDL1 mediated inhibition. Consequently, they have not addressed whether full length CTLA4 is really significantly more efficient in reinstalling PDL1-mediated PD1-inhibition than CTLA4 lacking its cytoplasmic tail. This is an important question in this context since CTLA4 lacking its cytoplasmic tail (termed CTLA4 Del36 in this study) is expressed at much higher levels and might thus strongly engage CD80 (and thereby enhance PD1-inhibition via PDL1) despite the fact that it is less efficient in reducing CD80 on adjacent cells than full length CTLA4. Due to this lack functional data the gives limited insight into how the interaction of PDL1 and CD80 in cis and the interaction of CD80 and CTLA4 in trans impact on the PD1 and CTLA4/CD28 pathways. (Tekguc et al., 2021)

claim that a similar tailless CTLA-4 molecule in mice is effective in regulating PD-L1 expression. However, the Del36 molecule is problematic both because it is highly overexpressed at the plasma membrane and because it does not exist in nature. It is worth reflecting that there are no reported mutations of the highly conserved CTLA-4 cytoplasmic domain in humans, despite a number of mutations being detected in the ligand binding domain. This strongly suggests that such mutations are not tolerated and that an intact CTLA-4 cytoplasmic domain is therefore essential for proper CTLA-4 function.
Thus, while we make comparisons for reference here, it is only the WT CTLA-4 molecule that can be considered to have physiological behaviour. Therefore, the referee's point ".... [Del36] is expressed at much higher levels and might thus strongly engage CD80 (and thereby enhance PD1-inhibition via PDL1) despite the fact that it is less efficient......." is rather moot. Nonetheless, we go on to show this is not the case in our manuscript and that WT CTLA-4 is more efficient in spite of higher levels of surface expression by Del 36.
Major concerns: 1. Important aspects shown in this manuscript have been reported earlier (e.g. by Sugiura et al). This is a rather small study with a limited set of experiments in cell lines. More experiments -especially functional experiments should have been performed to gain a more comprehensive insight on the role of CTLA-4 mediated transendocytosis of CD80 in PD-1 function to justify publication in a renowned high-impact journal with a broad readership. We agree our data support the work of Sugiura et al.,(Sugiura et al., 2019) and we cite their excellent work appropriately in the manuscript. However, our findings represent a significant extension of current understanding. Moreover, we have now provided a number of new experiments which we believe further enhance the importance of our manuscript. 2. As pointed out above the major limitation of the study is that it lacks functional data. It would be important to learn how CTLA4 activity impacts on PD-1 inhibition on T cells or T cell lines (in the presence or absence of CD28 costimulation). Without functional experiments it cannot be claimed that transendocytosis of CD80 is required to effectively control PD-L1-PD-1 interactions. We have now provided data in a new figure 7, which shows that removal of CD80 by transendocytosis (and not trogocytosis) liberates PD-L1 which is capable of binding to PD-1 on Jurkat T cells and inhibiting CD69 expression. This shows that PD-L1 is indeed functional following removal of CD80 by transendocytosis. We also performed these experiments in the absence of CD28 as suggested, which again show that liberated PD-L1 can bind PD-1 and inhibit T cell activation signals propagated via the TCR. 31st Oct 2022 1st Authors' Response to Reviewers 3. For the experiments summarized in figure 1 there is no information provided how often the experiments were repeated and statistical analysis is lacking. We apologise for this oversight; we have added this information to figure 1C. 4. No experiments with primary APC have been performed. PDL1-CD80 co-expression mainly occurs on professional APC which frequently also express the second ligand for CTLA4 and in many cases also PD-L2 the second ligand for PD-1. Potentially blockade of PDL1 by CD80 engagement could have less impact if PDL2 is present. We acknowledge this important point; however it is not clear that all of these additional issues can all be adequately studied within the scope of the present manuscript. The impact of PD-L2 is largely unknown and the cell types and conditions where primary human APCs express CD80, PD-L1, PD-L2 and CD86 are not established. We note that, data from Sugiura et.al., suggest that even in the presence of detectable PD-L2, the major binding of PD-1, was via PD-L1. However, their data used splenic DC and thioglycollate elicited Macrophages to identify PD-L1 expressing cells with differing levels of CD80, neither of which are realistically available for human studies. At present, we believe the best option is to use the B cell lines we describe in the manuscript, which are clearly a relevant APC cell type from an immunological perspective.

Minor points:
In the binding of therapeutic PDL1 antibodies which bind very strongly to PDL1 also blocked by CD80 binding in cis?
We have now provided new data in EV1 to address this point. The data show that even in the presence of high affinity clinical antibodies, the staining of PD-L1 is inhibited by CD80 resulting in an ~4 fold shift in EC 50. It is notable that at the highest concentrations of antibody the stain reaches maximal levels comparable with PD-L1 in isolation. Figure 7: the authors report that abatacept does not alter PD-1 binding; it can be expected that this is also the case for belatacept but if would be interesting to address this. We now show this data in EV8. The referee is indeed correct that belatacept does not enhance PD-1-Ig binding any more than abatacept. Indeed the data are remarkably similar to abatacept. We believe this highlights an important point, which is that Ligand levels: The authors should show fluorophore expressing and ligand expression (e.g. GFP expression and CD80 staining in Fig. EV4) We have now added additional data to EV6, which shows the correlation between antibody stains and GFP or mCherry expression as part of our experiments to more accurately quantify expression levels of these proteins.
If CD80 and CD86 are coexpressed is CD80 preferentially removed from the APC via CTLA-4 mediated transendocytosis? This is the case in our experience, and we are presently putting together a separate manuscript that studies this issue in detail.
Does CD80 has to be expressed in excess of PD-L1 to prevent PD-1-Ig binding or is this observed when both ligands are expressed at equimolar concentrations? We have now provided a new figure (figure 5), which shows that PD-1 binding to PD-L1 can be extinguished at a 1:1 ratio.

Referee #2:
Kennedy et al. show in in vitro experiments 1) CTLA4 on donor cells specially bind to CD80 in the heterodimer of PD-L1:CD80; 2) CTLA4 transendocytosis selectively remove CD80 but not PD-L1; 3) Effective removal of CD80 via transendocytosis requires CTLA4 cytoplasmic domain; 4) Transendocytosis of CD80 by CTLA4 but not blocking soluble CTLA4 Ig effectively liberates PD-L1. The authors conclude that efficient CD80 depletion by transendocytosis but not simply CTLA-4 binding to CD80 is required to effectively liberate PD-L1 that can engage PD-1. These observations are interesting, and the manuscript is clearly and concisely written.
There is a minor concern need to be addressed. We thank the referee for their comments and have now added some further references in the introduction and discussion that provide more balance in this area.

Sansom review
Adding to several recent publications on the cis CD80/PD-L1 interaction, the authors show that transendocytosis by CTLA4 of a CD80/PD-L1 heterodimer removes CD80 but not PD-L1 from the cell surface, allowing the PD-L1 to interact with PD-1. They propose that CD80 can operate as a switch and by preventing PD-L1 from interacting with PD-1 and by its higher avidity binding to CTLA4 also protect CD86 from transendocytosis by CTLA4, the CD80 can promote enhanced T cell responses by inhibiting both CTLA-4 and PD-1 pathways. The work is well done and deepens our understanding of a wicked complicated set of interactions but is with transfectants and not natural cells. No direct experiments on immune function are performed. We thank the referee for their helpful comments and critical reading of our manuscript. We have provided a point-by-point response below.
Major: 1. Zhao et al propose that the CD80/PD-L1 interaction, by reducing CD80 to a monomeric state, reduces the avidity for CTLA-4. Do a dose titration of CTLA4-Ig on CHO-CD80 and CHO-CD80/PD-L1 cells and determine the EC50 of CTLA4 binding and see if the avidity is altered. This will have to be done on CHO cell transfectants since endogenous CD80 and CD86 would complicate the analysis in DG-75 cells.
We have now performed these experiments and provide an EC50 for abatacept (figure 8) and the higher affinity belatacept (EV8). In neither case do we observe significant shifts in EC50, strongly suggesting that 2. Specify concentrations on x-axis of Figure 7 (these are DG-75 cells). Calculate EC50 of CTLA4-Ig binding to CD80 vs CD80/PD-L1 from Fig 7A. We have now shown concentrations of Abatacept in the figures along with the EC50 as above.
3. What is the molar ratio of PD-L1 to CD80 in the transfectants? Which is in excess? This cannot be determined by FACS but could be done by mass spectroscopy. We have now provided a new figure (figure 5), which shows that PD-1 binding to PD-L1 can be extinguished at a CD80-PD-L1 ratio of 1:1. In this figure we show the results from cells expressing different molar ratios quantified using calibration beads to quantify number of molecules. Minor: 5. In introduction, distinguish characteristics of transendocytosis and trogocytosis. How are they distinguished? We have now provided further detail in the introduction. 6. "acquisition by the CTLA-4+ recipient population using a well-established assay " describe cells used in assay, particularly CTLA-4 recipient cells. 7. Move Figure EV2 to main text so assay is clear to reader at beginning. We tried to move EV2 but could not satisfactorily incorporate it into a main figure. We have now tried to alert readers to the assay design further by reference to EV2 in the legend to fig.2. 8. The labeling of the figures and descriptions in the figure legends is in many places inadequate to clearly understand what has been done and how. There are many examples but to specify a few: Apologies we have now been through and corrected these issues in the legends and methods to make this clearer.
9. In 1D, specify ab used for IP and for WB Done 10. The labeling of Figure 2 does not provide enough information. The donor and recipient cells for each panel (and their CTV labeling) should be clearly indicated in the figure. We have tried to improve the labelling, CTV labelling is clearly shown, and we have added recipient and donor labels to 2A to help orient the reader, together with Fig EV2 to  12. Throughout all figures, where "ligand" is written, specify molecule. Do not use "no ligand" We specified ligand throughout and removed the use of "no ligand".
13. Methods says "ratios of donor: recipient cells and incubation times indicated in the figure legends." But figures often do not provide these numbers. Recipient cells need to be specified in figures. We have now specified the recipient and donor cell quadrants in the first TE FACS plot ( Fig. 2A), which together with the cartoon depicting the TE assay in Fig. EV2 should hopefully orient the reader.
14. In 3C and similar, the left panel y-axis is labeled "ligand remaining". Specify which ligand in each case; presumably CD80? We have labelled these more clearly 15. In methods, give equation of how ligand remaining was calculated in Figure 2D and similar figures. We have added this.
16. In 2A, choice of overnight incubation is a poor one since the amount of detectable transferred ligand is low (presumably because of degradation). Suggest using a time point where degradation is less dominant. We agree there are some limitations in showing the longer time point first, however, we wanted to emphasise the extent of removal of ligand from the donor cell. Since a time course showing earlier time points is available in 2C below, we feel the reader should be able to gauge these various nuances from the data.
17. In 2B, the question is whether the PD-L1mCherry is being transendocytosed by the CTLA4 cell. Say the red quadrant is loss of PD-L1. There does not seem to be a shift in the CTV positive population but the MFI intensity of the total cell population of the lower (CTV negative) population is clearly shifted to the right. Why does this shift not mean that PD-L1 is being broadly taken up by the recipient cells, maybe on the cell surface and not internalized? We agree there is a small shift of the recipient cells to the right. However, we don't feel this signifies high levels of uptake. In our experience of transendocytosis assays measuring ligand uptake is the least reliable-as transferred material is often degraded-as is the case for CD80 or CD86. Whenever labelled cells are in contact there is always some transfer by trogocytosis, which occurs at the point at which the cells are separated-resulting in small fluorescent signals. In contrast, removal of ligand from the donor cell in a time dependent manner is a much more rigorous measure and it is clear from this data that the amount of PD-L1 lost from the donor cells must be small. Comparing the graphs in 2E and 2F gives some sense of this.
18. Calculate statistical significance between lines in 2D (CD80 vs CD80+PD-L1) No significant difference was observed between these 2 lines, this has now been stated in the main text and added to fig 2D. 19. "CTLA-4 cytoplasmic domain was required for effective time-dependent depletion of CD80 and CD86 " Soften this statement (required is too extreme). In general, this data is discussed appropriately but the titles are extreme. There is some trogocytosis/transendocytosis by Cytdel (11% vs control of 2%) We have toned down the titles and text where appropriate. 20. "The reason for the differences between these observations currently remain unclear but may relate to differences in how transendocytosis was measured. " Describe differences between your method and Zhao's. We have added this discussion Royal Free & University College Medical School Rowland Hill Street, London NW3 2PF Tel: +44 (0)20 7794 0500 x 22469. Email: d.sansom@ucl.ac.uk http://www.ucl.ac.uk/medicalschool/infection-immunity/ 21. In the discussion, also note that in natural cells, PD-L1 is associated with CMTM6 and this is not replicated in transfectants. We have added some discussion on this point. 22. Specify GFP used in methods and linker length. done 23. Specify CTLA-4 del36 in methods. done 28th Nov 2022 1st Revision -Editorial Decision Dear David, Thank you for submitting your revised manuscript to The EMBO Journal. Your study has now been seen by referees # 1 & 3. As you can see from the comments below, both referees appreciate the introduced changes. Referee #3 has a few remaining points regarding the text that I would like to ask you to address in a final revision.
When you submit your revised version, please also address the following editorial points: -Please upload a "clean" manuscript version with no text markings.
-You are missing a Data Availability section. This is the place to enter accession numbers etc. If no data is generated that needs to be deposited in a databasethen please state: Data Availability: This study includes no data deposited in external repositories.
-You have 8 EV figures but can only have 5. Can you combine some of them for example EV2, 3 & EV7?
-Please make sure that EV figures panels are called out -use EV1a... nomenclature.
-COI needs to be renamed as " Disclosure and competing interests statement"