Molecular basis of VEGFR1 autoinhibition at the plasma membrane

Ligand-independent activation of VEGFRs is a hallmark of diabetes and several cancers. Like EGFR, VEGFR2 is activated spontaneously at high receptor concentrations. VEGFR1, on the other hand, remains constitutively inactive in the unligated state, making it an exception among VEGFRs. Ligand stimulation transiently phosphorylates VEGFR1 and induces weak kinase activation in endothelial cells. Recent studies, however, suggest that VEGFR1 signaling is indispensable in regulating various physiological or pathological events. The reason why VEGFR1 is regulated differently from other VEGFRs remains unknown. Here, we elucidate a mechanism of juxtamembrane inhibition that shifts the equilibrium of VEGFR1 towards the inactive state, rendering it an inefficient kinase. The juxtamembrane inhibition of VEGFR1 suppresses its basal phosphorylation even at high receptor concentrations and transiently stabilizes tyrosine phosphorylation after ligand stimulation. We conclude that a subtle imbalance in phosphatase activation or removing juxtamembrane inhibition is sufficient to induce ligand-independent activation of VEGFR1 and sustain tyrosine phosphorylation.

Reviewer #1 (Remarks to the Author): The submitted study is aimed at understanding aspects of the structure and function of the vascular endothelial growth factor receptor isoforms 1 and 2 (VEGFR1 and 2), key players in vasculogenesis/angiogenesis, and important drug targets.Data are presented that aim to resolve an outstanding question in the field, specifically why quite closely related VEGF receptor isoforms 1 and 2 (VEGFR1 and VEGFR2) have such different properties.In particular, the R1 form has very high affinity for growth factor but low kinase activity, whereas R2 has low VEGF affinity but high kinase activity.Despite these differences, both play critical roles in blood vessel growth and maintenance.As such, the work presented in the manuscript has the potential to be high impact since it touches on critical aspects of the activation mechanisms of the two receptor isoforms.More specifically, the study tries to explain why VEGFR1, despite being functionally critical, has weak kinase activity compared to its close relative VEGFR2.Like many RTKs, VEGFR2 undergoes ligand-independent activation at (unnaturally) high receptor density in the membrane, whereas VEGFR1 does not.Activation at high receptor density is presumably a stochastic process driven by increasing collisional frequency between monomers in the membrane.Why the two receptor isoforms should be different in this regard remains uncertain, and is addressed in this study.The methodology is fundamentally sound, using a similar approach to that used successfully to look at the mechanism of EGFR by the Kuriyan group (for example, in Endres et al, 2013).Variations in transient expression of different R1 and R2 receptor constructs are correlated with phosphorylation of specific tyrosine residues (detected by immunofluorescence) that in turn reports on receptor kinase activity.Critically, this assay depends on the quality (and specificity) of the antibodies used (see point raised, below).As I see it, the key finding(s) of the study can be summarised: 1.It is known that the 'juxtamembrane region' (JM) controls kinase activity in a range of RTKs, acting as an 'autoinhibitory switch'.In this study, swapping the JM of VEGFR1 for that of VEGFR2 gives the R1 isoform a 'VEGFR2-like' property, high levels of growth factor-independent kinase activity at (unnaturally) high cell surface receptor densities.2. The authors propose that the structure of the VEGFR1 JM is strongly autoinhibitory, making salt bridges with VEGFR1 kinase domain residues.Disruption of this interaction (via 'domain swapping' of the JM or kinase domain mutations) prevents autoinhibition, allowing kinase activation and ligand-independent autophosphorylation. 4. The authors conclude that the low kinase activity of VEGFR1 compared to VEGFR2 occurs as a consequence of a stronger interaction between the JM of VEGFR1 and its kinase domain.In essence, the key finding of this study is interesting and would represent a valuable contribution.Unfortunately, the huge mass of data in the paper compromises the clarity and focus of the work.Some of the data presented either lack statistical significance or are not directly relevant to the main findings.My comments below are essentially aimed at removing a lot of the noise and refocusing the manuscript on the key findings, which do have significance.Specific comments: 1. Figure 2. It would be useful to know what 'Low expressing' and 'High expressing' mean quantitatively -where do these sit on the x-axis of Fig. 2E, for example?In the microscopy images in A, C and D I note that there is no signal at all for phospho-tyrosine-specific antibodies, but none of the corresponding graphs showing a correlation between receptor levels and phosphorylation reach a zero value, hence there is an inconsistency between the microscopy images and the corresponding quantitation.Western blots with pY antibodies also consistently show signal at time zero.The authors should address this, since it somewhat undermines confidence in the assay.2. It could be made clearer how the normalisation in Fig. 2E and F right panels has been performed.3. Fig. 3E.It appears that only the dimerization control has produced any statistically significant effect, so this part of the figure is essentially redundant.I accept that there may be some kind of trend towards a ligand-dependent effect, but if it is not significant then it shouldn't be included.Same argument applies to Fig. 4F and 6E 4. Fig. 5.I don't find the MD simulations compelling -if the JM-S segment is intrinsically flexible and unresolved in xtal structures, isn't the MD simulation little more than a guess?At best, I see this as 'Supplemental' and a speculative discussion point.5. Fig. 7.The authors moved on to include oxidative stress effects and a discussion of differential phosphatase effects -this seems unnecessary to me, causing the story to lose focus.The data seem interesting, but including them (in my opinion) just causes the paper to lose focus and clarity.6.There is far too much 'Supplementary' material, a lot of it just confirmatory and of little value.The manuscript would benefit from having much of this thinned out.
Reviewer #2 (Remarks to the Author): SUMMARY In this study, the authors aimed to understand the mechanisms of ligand-dependent and ligandindependent activation of VEGFR1.They investigated the differences between VEGFR1 and VEGFR2 in terms of auto-phosphorylation and stability, as well as the components responsible for the autoinhibition of VEGFR1.Additionally, they examined the conditions that induce autophosphorylation of VEGFR1 in pathological states.The findings of this study shed light on the regulatory mechanisms of VEGFR1 and have implications for the development of therapeutic strategies targeting VEGFRs.
The authors first compared the phosphorylation levels of VEGFR1 and VEGFR2 in the presence and absence of ligands.They demonstrated that VEGFR1 requires VEGF for phosphorylation, while VEGFR2 can be spontaneously phosphorylated even without ligand stimulation.Furthermore, they recapitulated prior wok showing that the phosphorylation level of VEGFR1 after ligand stimulation was significantly lower than that of VEGFR2.To understand the underlying reasons for these differences, the authors investigated the phosphorylation kinetics and half-life of specific phosphotyrosine residues using immunoblotting.They discovered that the Y1213 residue in VEGFR1 is more transiently stable than the Y1179 residue in VEGFR2, indicating slower activation and a shorter half-life for VEGFR1.This is one major issue with the work, only one VEGFR1 site is examined-while there is a dearth of VEGFR1 antibodies, there are pan antibodies that could be used (non site specific).Therefore, their conclusions are only relevant to this one site, not to VEGFR1 as a whole.
To identify the component responsible for the autoinhibition of VEGFR1, the authors examined different regions of the receptor.They found that mutations or deletions in the extracellular domain (ECD) and transmembrane (TM) segment of VEGFR1 did not affect its autoinhibition.However, mutations in the juxtamembrane (JM) region of VEGFR1 partially or fully restored its autophosphorylation.This led the authors to conclude that the interactions between the TM and kinase domain are the main regulators of VEGFR1 autoinhibition.
Finally, the authors investigated the conditions that induce autophosphorylation of VEGFR1 in pathological states.They demonstrated that under H2O2 or sodium orthovanadate treatment, which mimic pathological conditions, wild-type VEGFR1 is autophosphorylated.This suggests that reactive oxygen species (ROS) and the inhibition of protein tyrosine phosphatase (PTP) play a role in VEGFR1 autophosphorylation.Overall, this study provides valuable insights into the mechanisms underlying the autoinhibition and activation of VEGFR1.VEGFR1 is not widely studied, but it affects vascularization and even immune cell migration-so an improved understanding of its activation is sorely needed.The authors conducted a thorough investigation, using various experimental approaches, to elucidate these mechanisms.The findings have implications for the development of novel therapeutic strategies targeting VEGFRs.However, the title does seem to overstate the impact of the work, given the limited probing of VEGFR1 phosphorylation.Additionally, suggestions and questions regarding specific details and contextualization that could further enhance the manuscript are described below.

++++++++++++++++++++++++++++++ INTRODUCTION
• Page 1|line 7: In the first four sentences, the authors state that VEGFR1 is not autophosphorylated without ligand, unlike VEGFR2, and it shows weak tyrosine kinase activity even after the ligand stimulation.In the next sentence, they talk about VEGFR1 activation under pathological conditions.Since this sentence is after the "ligand stimulation" sentence, the phrase, "which is puzzling", both comes across as awkward and is in conflict with the premise of the paper.It is well established that VEGFR1 transmits angiogenic signals (but to a lesser extent than VEGFR2), so referring to this concept as puzzling is not inline with conventional knowledge of VEGFR1's modulatory behavior The manuscript could be improved if the concept of puzzling is removed.
• The paper mentions 'PDGFR-like JM-in inactive conformation' a few times.Two suggestions on this: (1) spell out PDGFR when introducing it for the first time (Page 3) and ( 2) provide background about this PDGFR JM-in inactive confirmation to contextualize its relevance in this VEGFR study.
• 'Downregulation' is more appropriate than 'deregulation' in this following context: "Overexpression or deregulation of VEGFR1 is linked to several cancers and cancer-associated pain,…": • The paper uses 'basal state' to refer to 'the absence of ligands' (Page 4).However, this usage could cause confusion because 'basal state' typically means 'a pre-activated or a resting condition'.In the absence of ligands, VEGFR2 can be activated, so the absence of ligands is not a basal state for VEGFR2.Additionally, this work examined the diseased condition (ROS) in the absence of ligands, which is not a basal state by conventional definition.+++++++++++++++++++ RESULTS

Page 5|line 1:
The authors raise a very important concept that "receptor density at the plasma membrane is an important determinant of ligand-independent activation of RTKs."This is especially true for VEGFRs, where receptor densities of VEGFR1 and VEGFR2 have been sensitively quantified; however, the citations only point to either generalized RTK papers or papers on FGFR and EGFR, which are not relevant to this work.The manuscript could be improved by citing the relevant VEGFR papers that both cover this concept and quantify these receptor densities.2. Page 5|lines 12-13: Rather than the terminology "peripheral region," terminology like perimembrane or cell perimeter could be used.Additionally, there should be more scientific insight into what specifically this would entail (e.g., cell membrane?How much distance into the cell).Furthermore, the peripheral (perimembrane) region may not be the right location for examining VEGFR1, since a majority of VEGFR1 is intracellular and a significant fraction localizes to the nucleus.3. Page 5|lines 17-19: The manuscript reads "VEGFR2 linearly phosphorylates Y1175 upon ligand stimulation, suggesting the phosphorylation is independent of receptor concentration at the plasma membrane."The phosphorylation level of Y1175 increases as VEGFR2 concentration increases.More insight into why the phosphorylation is independent of receptor concentration would improve the paper.4. Page 8|lines 7-10: In "The observed increase in the Dconfocal for the VEGFR1-GPA-G83I mutant "'&'** I '&'(* KA)E%(# 9CB<?DAE 7 :?A;D%FC%ACBCA;D FD7BE?F?CB CB AGF7F?B= 51 E;=A;BF ?B F>; 6.0/3(%02-9>?A;D7 "'&')( I '&'', KA)E%(# "/?=GD; *.$ 4+0$ 7B: 578@; 4*#&J$ H>7F ?E F>; criterion to determine if VEGFR tends to form a ligand-independent dimer? 5. Page 9|line 5: You mutated the ECD residue in VEGFR1 and saw that it didn't induce the VEGFR1's autophosphorylation.Could you explain more about how you get to the conclusion that the ECD is likely the dominant negative regulator of VEGFR1 activation? 6. Authors devised VEGFR1-only and VEGFR2-only transfected cells models to study VEGFR1 and VEGFR2 phosphorylation; however, these models do not consider the VEGFR1 and VEGFR2 phosphorylation through heterodimerization.This is an important mechanism enlisted by VEGFR1, so some context is required.Do the authors have comments on the heterodimerization mechanism?As a point of reference, it is well established that these receptors dimerize.Please see the two papers below, and there are others: o Autiero, M., Waltenberger, J., Communi, D. et al.Role of PlGF in the intra-and intermolecular cross talk between the VEGF receptors Flt1 and Flk1.Nat Med 9, 936-943 (2003).https://doi.org/10.1038/nm884o Kui Huang, Charlotte Andersson, Godfried M. Roomans, Nobuyuki Ito, Lena Claesson-Welsh, Signaling properties of VEGF receptor-1 and -2 homo-and heterodimers, The International Journal of Biochemistry & Cell Biology, Volume 33, Issue 4, 2001, Pages 315-324, ISSN 1357-2725, https://doi.org/10.1016/S1357-2725(01)00019-X.7.This work focuses on only one phosphorylation site of each VEGFR: Y1213 for VEGFR1 and Y1175 for VEGFR2.While we recognize that there are limited VEGFR1 antibodies available for each of the VEGFR1 sites, it is possible to look at pan-VEGFR1 phosphorylation.What is the authors' rationale for studying only these phosphorylation sites, not others or the total phosphorylation?Is it possible that other VEGFR1 sites can be auto-phosphorylated without the presence of VEGF? 8. Did the authors run a statistical analysis for the bar graphs in Figure 2 G and H? How many replicates (n) are included in these graphs?9. Can the authors give more information about how to interpret the Dconfocal¬¬?Is there a reference value for us to tell whether it's a dimer or monomer state?Is it a standardized measurement that can be compared across different studies?10.Page 10: "The sequence comparison between VEGFR1 and VEGFR2 shows that the residues at the ligand-independent dimer interface are conserved.In contrast, the residues at the liganddependent dimer interface are not conserved."Have the authors considered whether this sequence difference could be responsible for the low ligand-dependent VEGFR1 phosphorylation stability?11.Page 10: There seems to be a logical leap.What is the authors' rationale for the speculation that T763 and C764 make the TM segment incompatible with a ligand-independent dimer?12. How are JM-B, JM-S, and JM-Z defined? 13.I am not sure if I understand how the conclusion was drawn "(VEGFR1) does not require help from a second tyrosine kinase".14.Page 14 "We observed that wild-type VEGFR1 is autophosphorylated upon H2O2 treatment, but in the kinase-dead mutant, Y1213 was marginally phosphorylated (Figure 7A and S8A).Suggesting, under oxidative stress, VEGFR1 spontaneously autophosphorylates the Y1213 and does not require help from a second tyrosine kinase.In human colorectal cancer cells and hyperglycemia, it may be noted that the VEGFR1 and VEGFR2 phosphorylation is mediated by Src tyrosine kinase (80,84)."+++++++++++++++++++++++++++++ CONCLUSION 1.This Conclusion section repeats the Results and did not speak much about the broader impact of these findings.2. Can the authors provide more insights into the basis of this speculation?"We speculate that marginal reduction in PTP activity due to oxidative stress under pathological conditions may be sufficient to stimulate ligand-independent VEGFR1 signaling."VEGFR2.The study has much interest and relevance to VEGFR and receptor tyrosine kinase function.However, there are a number of issues which need to be carefully addressed.These points are listed below: 1) The authors indicate from the very first statement, that the low(er) levels of VEGFR1 tyrosine kinase (TK) activity is unusual; however, VEGFR2 is a notable and powerful TK.The lack of comparison to VEGFR3 (or any other RTK) in this context makes this statement shaky (compared to the 58 members of the human RTK family).
2) The overexpression of VEGFR-mCherry constructs raises serious mechanical issues in the context of RTK activation and possibly trafficking.Can the authors be confident that this is indeed similar to native or endogenous VEGFRs?The attachment of the 27 kDa mCherry to the C-terminal tail will restrict the movement of the flexible ~200 residue C-terminal tail that is likely to have regulatory effects on TK activation; furthermore VEGFR-mCherry trafficking may be modulated or disrupted.What controls have been done to check this?
3) The choice of phosphorylation epitopes in VEGFR1 (Y1213) and VEGFR2 (Y1175) is problematical.VEGFR2-Y1175 (vs.VEGFR1-Y1173) are sites of activation and binding to PLCgamma1 thus showing comparable properties; VEGFR2-Y1214 (vs.VEGFR1-Y1213) are linked to binding c-SRC upon phosphorylation and activation of downstream signalling pathways.Of note, VEGFR2-Y1175 and VEGFR2-Y1213 shows different kinetics upon activation by the same ligand, VEGF-A165 (Fearnley et al. (2016).These experiments should be carried out by checking for total tyrosine phosphorylation using PY20 or PY-100 monoclonal antibodies.That will give a better idea of net activation resulting in the formation of total phosphotyrosine epitopes on the VEGFR1 vs VEGFR2 constructs.
4) The authors seem completely unaware of the multiple number of studies that indicate differential trafficking of VEGFR1 vs. VEGFR2.It is well established that VEGFR2 traffics slowly out of the secretory pathway, accumulating in the Golgi before reaching the plasma membrane and endosomes (Ewan et al., 2006;Manickam et al., 2011;Yamada et al., 2014).In contrast, conflicting reports suggest the majority of VEGFR1 localises to intracellular compartments such as Golgi (Mittar et al., 2009;Yang et al., 2015) and nucleus (Boulton et al., 2008;Zhang et al., 2010) with other reports of intracellular VEGFR1 near the nucleus (Lee at al., 2010).Weddell and Imoukhede (2017) have carried out mathematical modelling which predicts VEGFR1 distribution in endocytic vesicles, endosomes and nucleus but not necessarily at the plasma membrane.If resting VEGFR1 levels are largely (>80%) located within the cell, thus the fraction of VEGFR1 activation by exogenous ligand (e.g.VEGF-A) is relatively small.VEGFR1 is a widely expressed molecule, including epithelial cells with estimated numbers of 500-5000 molecules/cell.This is in contrast to endothelial VEGFR2 which is estimated at 10000-50000 molecules /cell.If VEGFR1 is retained (by virtue) of a targeting signal (e.g.within the JMD) to a different part of the cell, this may explain why VEGFR2 is much more readily available at the plasma membrane.This has not been considered.
5) The authors should remove Fig. 7 with experiments using hydrogen peroxide and phosphatase involvement.This does not add anything to the story and confuses the study as it stands.
6) The authors need to also carefully consider the work from the labs of Kuriyan, Lemmon and Schlessinger on studying EGFR and FGFR signalling and activation.There is much discussion on the roles of the flexible C-terminal tails and JMDs in influencing TK activation.This needs to be better placed within the introduction and discussion.
7) The whole manuscript, especially the figures, needs to be tidied up, data put into supplemental figures if needed, and a more tidy and streamlined article needs to be produced.Currently it feels that the authors have emptied their lab notebooks without any discretion.It does not make for easy reading or digestion by readers in the field.Cb cif ghiXm( kY igYX U acbcaYf UbX U X]aYf Wcbhfc` cZ P?A@L-hc XYhYfa]bY h\Y fY of graph (fig 2) should happen the first time the graph is used.3. KD is conventionally used to describe equilibrium dissociation constant, this is not the appropriate use of this abbreviation.It suggested that a different abbreviation be used.4. Page 6|line 22 (last line): Moving "The ligand bias by …" to the next subsection might make the paragraphs more readable.5. Page 7|lines 5-6: Could you match the illustration style showing VEGFR1 and VEGFR2 in Figure S3A so that the readers can compare their structures more easily?6. Page 7|line 11: You might need to include a brief definition of "C482R".7. Page 9|line 2: It is not clear if you mean "Figure S5A-B", or "Figure 5A-B".Reviewer #3 (Remarks to the Author): This study by Charaborty et al. provides a new finding, namely that the VEGFR1 juxtamembrane domain (JMD) has an autoinhibitory function that is not present in another related molecule,