Endothelial deletion of SHP2 suppresses tumor angiogenesis and promotes vascular normalization

SHP2 mediates the activities of multiple receptor tyrosine kinase signaling and its function in endothelial processes has been explored extensively. However, genetic studies on the role of SHP2 in tumor angiogenesis have not been conducted. Here, we show that SHP2 is activated in tumor endothelia. Shp2 deletion and pharmacological inhibition reduce tumor growth and microvascular density in multiple mouse tumor models. Shp2 deletion also leads to tumor vascular normalization, indicated by increased pericyte coverage and vessel perfusion. SHP2 inefficiency impairs endothelial cell proliferation, migration, and tubulogenesis through downregulating the expression of proangiogenic SRY-Box transcription factor 7 (SOX7), whose re-expression restores endothelial function in SHP2-knockdown cells and tumor growth, angiogenesis, and vascular abnormalization in Shp2-deleted mice. SHP2 stabilizes apoptosis signal-regulating kinase 1 (ASK1), which regulates SOX7 expression mediated by c-Jun. Our studies suggest SHP2 in tumor associated endothelial cells is a promising anti-angiogenic target for cancer therapy.

The manuscript by Xu et al describes a novel and very unexpected anti cancer effect of SHP2/PTPN11 inhibitors. These drugs are considered to have anti cancer effects by blocking signals from receptor tyrosine kinases (RTKs) to downstream effectors. As such, most trials with these drugs are focussing on RTK mutant cancers. The present manuscript demonstrates that SHP2 also has a critical role in the endothelium of the cancer, as they demonstrate that deletion of SHP2 in only endothelial cells has a major effect on tumor growth. As such, these findings shed a new light on how SHP2 inhibitors can inhibit cancer growth. This finding will without doubt be of interest to a wide readership. The biological data are supported by mechanistic studies that point towards a major role for SOX7 downstream of SHP2 in endothelial cell function.
There are a few points that the authors should address: 1. The manuscript should be corrected by a person that is fluent in English. 2. Figure 1h needs a legend for the staining 3. Figure 3a: why focusing only on downregulated genes? The upregulation of VEGFR is also quite strong and interesting, but it is not even discussed. 4. Figure 4: How was the AAV vector delivered? I can guess intranasal or intratracheal injection, but it is not mentioned 5. The rationale behind looking at cJun as SOX7 regulator is unclear.
Reviewer #2 (Remarks to the Author): In this manuscript, Zhang et al. investigated the role of SHP2 in tumor-associated angiogenesis. The authors concluded that tumor-associated endothelial cells have increased levels of total and activated SHP2 that is required to de-phosphorylate and stabilize ASK1, activate c-Jun and consequently increase the SOX7 transcription. SHP2 role in endothelial cell functions has been extensively studied and SHP2 has been already implicated in regulation of vessel stability and permeability via different molecular mechanisms. Also, it was already shown that inhibition of SHP-2 suppresses angiogenesis in vitro and in vivo. The noteworthy result from the current manuscript is the new molecular mechanism implicating SHP2 in increasing the expression of SOX7 in endothelial cells. Even though the mechanism is somewhat experimentally defined, mostly in vitro, additional evidence needed and specified below. Also, it is unclear how the new mechanism is related to the previously shown role of SHP2: the authors and others have previously shown that SHP2 depletion caused endothelial cell junction disruption and lung edema. Clinical relevance of the conclusions is weak and needs to be clarified: is it relevant to the specific types of cancer or ubiquitous to all cancers. Major points are outlined below.
Major concerns: 1. Since endothelial cells in different organs are phenotypically and functionally different, the authors need to use orthotopic tumor models to assess tumor angiogenesis, which is also tissue specific. They need to inoculate LLC tumor cells into the lung tissue and B16 tumor cells intradermally to make the clinically relevant models of these two types of cancer. 2. In their previous publication, the authors demonstrate that conditional deletion of Shp2 in endothelial cells (EC) caused embryonic lethality associated with disruption of endothelial cell junctions and massive hemorrhage, and the inducible deletion of Shp2 in EC (Shp2iECKO) in adult mice resulted in lung vessel permeability. It is essential to measure lung vessel permeability in Shp2iECKO mice that are used for the experiments in this manuscript and their survival. Structural and functional changes in lung endothelial cells should be assessed at different time points after tamoxifen treatment, including at the time of tumor harvest, and whether those changes are similar or different in lung vs subcutaneous endothelial cells. 3. The conclusion that SHP2 is increased and activated in tumor-associated endothelial cells in vivo is not based on strong evidence. The immunofluorescent images of mouse, and especially human lung cancer tissues (Fig. 1b-c) are of low quality and must be replaced. High and low magnifications of IF staining should be added. It should be also clarified how the relative p-SHP2 or SHP2 were quantified using these images. In addition, in it is unclear why the authors use lung EC to show protein levels of SHP2 (Fig. 1b-c), but mammary EC to show mRNA levels of SHP2 (Fig. 1d). 4. The statement that there are more α-SMA-positive endothelial cells in Shp2iECKO vessels does not make any sense since there are no such cells. 5. The statement that Shp2iECKO tumors have decreased macrophage infiltration is incorrect, since CD11b is not a specific marker of macrophages. 6. Since Sox7 is a transcription factor, it is not clear why the IF staining for SOX7 is cytoplasmic in Fig. 3i. 7. The authors conclusion that SOX7 is regulated by SHP2 has to be verified in human samples by at least showing the SOX7 levels in normal vs tumor-associated endothelial cells using human lung cancers and melanoma. 8. The authors should show the levels of SOX7 in normal and tumor-associated endothelial cells after using AAV vector to over-express SOX7. 9. Most conclusions about c-Jun and ASK1 roles in tumor angiogenesis are based on in vitro experiments or Matrigel plugs. The authors need to verify them in orthotopic mouse models of cancer and in human tissue samples.
Reviewer #3 (Remarks to the Author): In this manuscript the authors investigated the tumor extrinsic function of SHP2 in promoting tumor growth by enhancing tumor angiogenesis and vascular abnormalization in endothelial cells. The authors used a GEMM, previously generated in their lab, in which SHP2 is selectively deleted in the adult endothelium upon tamoxifen induction. The authors found that SHP2 was a key regulator in tumor angiogenesis since its deletion was correlated with reduced tumor growth and microvascular density in mice where the tumor formation was previously induced by subcutaneous injection of two different syngeneic cancer cell lines. Similar results were also obtained in vitro using one human endothelial cell line. Mechanistically, they proposed that SHP2 promotes tumor angiogenesis not through the canonical activation of the MAPK pathway but rather the formation of a new molecular proangiogenic axis. Specifically, the authors concluded that ASK1 protein stability is enhanced by the tyrosine phosphatases function of SHP2 that in turn promoted c-Jun activation and increase the expression of the proangiogenic factor SOX7 in tumor associated endothelial cells. While SHP2 has been identified as a regulator of tumor angiogenesis (Fedele et al. Cancer Discovery 2018 andJEM 2021), this manuscript is the first to report that specific genetic deletion in adult endothelial cells promotes endothelial cell death and the involution of tumor vessels resulting in reduced tumor growth. However, more study are necessary to understand long-term effect of SHP2 deletion/inhibition in the resting vasculature of normal tissues. The current identification of SHP2 as a previously unrecognized regulator of the tumor vasculature may be a finding of exceptional relevance since the development of several SHP2 inhibitors currently in phase 1/2 trials. The manuscript is of potential interest; however, main concerns are present since some experiments are not well controlled, using few in vitro and in vivo models and sometimes the conclusions are not fully supported by the presented results.
MAJOR points: 1) The authors used only HUVEC cell line as in vitro model. Keys results should be validated in other endothelial cell lines as HDMEC and/or BMEM 2) The effect of SHP2 deletion in TME need more characterization. Only two syngeneic cell lines were used to generate tumors in the iECKO mouse model. Keys results in fig 1 and 2 should be validated using other models with a bona fide tumor microenvironment (e.g. orthotopic implantation of mammary 4T1, pancreas/lung KPC, melanoma YUMM and or D4M3A) 3) In fig. S1a the author showed a not complete KO of SHP2 upon tamoxifen induction in the iECKO model. Is a partial downregulation of SHP2 enough to drive the reduction in tumor growth and promote vascular normalization? Further characterization using allosteric inhibitor in vivo is necessary. Parallel evaluation of tumor sizes and vascular normalization should be performed in immunocompromised mice (e.g NSG or nude) implanted with tumor cells whose growth is SHP2independent (e.g. ectopic expression of drug-resistant PTPN11 mutant SHP2T253M/Q257L) and treated with different doses of the SHP2 inhibitor (e.g. SHP099 75, 37.5 and 18.75 mg/kg). 4) How the standard anti VEGF treatment (e.g. bevacizumab) in mouse models compare with SHP2 deletion/inhibition in reducing tumor growth and promoting vascular normalization? 5) In fig 3, 5 and 6, in vitro rescue experiments of the shSHP2 effect should be performed to exclude any off-target effect of the shRNA. Does the co-delivery of shSHP2 and a shRNA-resistant SHP2 cDNA rescues the SOX-7 downregulation ( fig. 3), cJun and ASK reduced phosphorylation ( fig. 5 and 6)? Rescue validation should be conducted in experiments in which shSOX7 and shASK1 were used as well ( fig S2 and Fig 6) 6) The authors claimed that the phosphatases activity of SHP2 is important to promote ASK1 stability. Not enough evidences are provided in fig.6 to support this conclusion. In vitro rescue experiments in which co-delivery of shSHP2 and a shRNA-resistant phosphatases-dead SHP2 cDNA (e.g. C459S mutant) should be performed. IP of ASK1 followed by pTyr WB should be performed as well using the SHP2 inhibitor in cells expressing either WT or the inhibitor-resistant mutant SHP2T253M/Q257L. 7) Based on the results provided in fig 5 a-b, the authors claimed that in endothelial cells, SHP2 promotes tumor angiogenesis with a MAPK-independent mechanism. However, not enough evidences are provided to support this conclusion. Contrasting data are also provided in fig. S3 j in which HUVEC cells treated with the SHP2i showed decreased level of pERK as well as phospho and total cJun. A better characterization is needed. Experiment using different MAPKi (e.g. SHP2, MEKi or ERKi) should be performed in vitro. Proliferation, migration assay should be performed. WB analysis of SOX7, phospho cJun, ASK1, pERK etc. should be performed as well.
Minor points: 1) Concentration and duration of experiments conducted with RMC-4550 inhibitors were not reported either in the text or in figure legends. 2) Quality of many WB should be improved (e.g. Fig 3f; Fig 5 a

Reviewer #1 (Remarks to the Author)
The manuscript by Xu  Reply. We appreciate the positive remarking. The anti-cancer function of SHP2 inhibitors, especially combining with other anti-tumor drugs, has been well demonstrated recently. Here we show dramatic anti-cancer function of SHP2 deletion in tumor endothelial cells. Together with two recent studies revealing anti-angiogenic function of SHP2 inhibitors, our results using genetic modified mice and multiple mouse tumor models support to target SHP2 for anti-angiogenic therapy in cancers. In addition, we present a new signaling axis, ASK1-c-Jun-SOX7, as a major downstream pathway for SHP2 in regulating tumor angiogenesis. Also, our results suggest SHP2-independent tumors could also benefit from SHP2 inhibitors. SHP2 deletion improves blood perfusion in tumors which highlights the necessity in combination with other anti-tumor drugs.

The manuscript should be corrected by a person that is fluent in English.
Reply. Thanks for the comment. We have read through the manuscript. In addition, the writing is improved by the Editage (www.editage.cn).

Figure 1h needs a legend for the staining.
Reply. Thanks for pointing out this question. Figure 1h is H&E staining and necrosis areas are draw out by pink staining due to decreased or vanished nuclei. In figure legends of revised version, we added the staining and measuring methods.
3. Figure 3a: why focusing only on downregulated genes? The upregulation of VEGFR is also quite strong and interesting, but it is not even discussed.
Reply. Thanks for the comments. To identify mechanisms for SHP2 in regulating tumor angiogenesis, angiogenesis-related genes were checked in SHP2-knockdown HUVECs ( Figure 3a for the old version, and Figure 4a for the revised version). Multiple pro-angiogenic factors including VEGFA, VEGFR1, and VEGFR2 were upregulated upon SHP2-knockdown, which was not in line with decreased angiogenesis in SHP2 deleted mice and SHP2 knockdown endothelial cells. It is possible that the upregulation of these factors is due to the compensation to reduced angiogenesis.
In the other hand, proangiogenic factor, SOX7, was significantly reduced. SOX7-knockdown in endothelial cells impaired cell proliferation, migration, and tube formation, suggesting its proangiogenic function. Importantly, re-expression of SOX7 in SHP2-knockdown and SHP2 deleted endothelial cells restored endothelial function and tumor angiogenesis respectively, emphasizing the critical importance of SOX7 for SHP2 in regulating tumor angiogenesis. In the Results section of the revised manuscript, we added the description that we focused on the downregulated genes because of the reduced angiogenesis.

Figure 4: How was the AAV vector delivered? I can guess intranasal or intratracheal injection, but it is not mentioned
Reply. Thanks for bringing out the question. The AAV expressing SOX7 was intra-tumor injected once for the viral content of 5X 10^10 TU/injection. The information was added in figure legend ( Supplementary Fig. 6i) in the revised version.

The rationale behind looking at cJun as SOX7 regulator is unclear.
Reply. Thanks for bringing out the question. First, SOX7 is among the four transcriptional factors for SOX7 predicted in three different databases. Secondary, SHP2 is well demonstrated in regulating MAPK pathways, which including RAS-ERK, p38, and JNK-c-Jun. Subsequently, we found that in endothelial cells, SHP2 knockdown did not affect the activation of ERK and p38. On the contrary, c-Jun was dramatically deactivated upon SHP2 knockdown or inhibition in endothelial cells. Finally, we identify SHP2-ASK1-c-Jun-SOX7, a novel pro-angiogenic signaling axis in tumor associated endothelial cells. Therefore, c-Jun mediates the regulation of SOX7 by SHP2.

Reviewer #2 (Remarks to the Author)
In this manuscript, Zhang   The importance of SHP2 in regulating tumor angiogenesis is demonstrated by two syngenetic mouse tumor models using LLC and B16 cancer cells and one orthotopic breast cancer model using E0771 cells. In addition, SHP2 inhibitor also reduces tumor angiogenesis and growth. Comparable with anti-angiogenic cediranib, SHP2 inhibitor suppressed tumor growth, especially in SHP2independent tumors. All these results, together with other group's excellent studies, demonstrate that SHP2 inhibitors are excellent anti-angiogenic drugs in cancer therapy.

Since endothelial cells in different organs are phenotypically and functionally different, the authors need to use orthotopic tumor models to assess tumor angiogenesis, which is also tissue specific. They need to inoculate LLC tumor cells into the lung tissue and B16 tumor cells
intradermally to make the clinically relevant models of these two types of cancer.
Reply. Thanks for the good suggestion. To increase clinical relevance of our studies, an orthotopic mouse model using E0771 mouse mammary tumor cells was introduced. As shown in Supplementary Fig. 1 and 4, SHP2 deletion in an orthotopic mouse tumor model significantly reduced tumor growth and angiogenesis and promoted vascular normalization. Therefore, SHP2 deletion displayed great anti-angiogenic function in multiple preclinical mouse tumor models. Shp2iECKO mice including lungs, spleens, kidneys, livers, and hearts ( Supplementary Fig. 3). Shp2 deletion also did not affect microvessel density in lungs, spleens, kidneys, and livers ( Supplementary   Fig. 3). With or without tumor burden, Shp2 deletion didn't alter vessel permeability in lungs, livers, and kidneys ( Supplementary Fig. 3). These data showed that SHP2 inhibition is somehow safe in cancer therapy. It is worth mentioning that Shp2 deletion disrupted endothelial barrier in mouse embryos, while increasing endothelial barrier in tumor vessels. These results suggest different roles of SHP2 in regulating physiological and pathological angiogenesis.

The conclusion that SHP2 is increased and activated in tumor-associated endothelial cells in vivo
is not based on strong evidence. The immunofluorescent images of mouse, and especially human lung cancer tissues (Fig. 1b-c) are of low quality and must be replaced. High and low magnifications of IF staining should be added. It should be also clarified how the relative p-SHP2 or SHP2 were quantified using these images. In addition, in it is unclear why the authors use lung EC to show protein levels of SHP2 (Fig. 1b-c), but mammary EC to show mRNA levels of SHP2 (Fig. 1d).
Reply. Thanks for bringing out the questions. We replaced the representative images for SHP2 and p-SHP2 immunofluoscence staining in Fig. 1b, c with high-quality ones. Quantitative data were measured by using CD31 to label endothelial cells and the Image J software for measuring. In Fig.   1d we showed increased SHP2 mRNA in tumor associated endothelial cells. The result was extracted from the GEO dataset GSE118904, which used endothelial cells in mouse normal mammary glands and orthotopic mouse E0771 mammary tumors for single-cell RNA sequencing 2 .
The dataset GSE118904 was used here because we couldn't to find similar data for lung cancer or other cancers. NSCLC tissues and conditioned media were used in our studies; however, SHP2 regulates tumor angiogenesis and vascular abnormalization shouldn't be unique to lung cancers. We didn't answer how SHP2 is upregulated and activated in tumor associated endothelial cells. This is a limitation in our studies.

The statement that there are more α-SMA-positive endothelial cells in Shp2iECKO vessels does not make any sense since there are no such cells.
Reply. Thanks for bringing out the incorrect description. α-SMA is expressed in pericytes, not in endothelial cells. We revise the description. There were more endothelial cells surrounded with αsmooth muscle actin (αSMA) positive cells in tumor vessels of Shp2 iECKO mice, indicating increased pericyte coverage.

The statement that Shp2iECKO tumors have decreased macrophage infiltration is incorrect, since
CD11b is not a specific marker of macrophages.
Reply. Thanks for pointing out the question. "The infiltration of CD11b positive myeloid cells" was used to replace with the old one. Fig. 3i. Reply. Thanks for bringing out the question. We did immunofluorescence staining for SOX7 in LLC tumors ( Fig. 4i and supplementary Fig. 6j), NSCLC tissues (Fig. 4j), HUVECs (supplementary Fig.   5e) and Matrigel plugs (supplementary Fig. 5i). SOX7 is nuclear localization in HUVECs, in endothelial cells in LLC tumors and NSCLC tissues. SOX7 staining in the plugs was not exactly nuclear and non-specific staining is possible due to Matrigel. Same SOX7 antibody was used for all staining.

The authors conclusion that SOX7 is regulated by SHP2 has to be verified in human samples by at least showing the SOX7 levels in normal vs tumor-associated endothelial cells using human lung cancers and melanoma.
Reply. Thanks for the suggestion. As suggested, we did immunofluoscence staining for SOX7 in human NSCLC tissues and paired adjacent normal tissues (Fig. 4j). SOX7 was highly expressed in tumor vessels. Compared with normal endothelial cells, SOX7 was increased in tumor associated endothelial cells.

The authors should show the levels of SOX7 in normal and tumor-associated endothelial cells after using AAV vector to over-express SOX7.
Reply. Thanks for the suggestion. As suggested, we did immunofluoscence staining for SOX7 in tumors in the rescue experiments ( Supplementary Fig. 6j). SOX7 was decreased in tumors in Shp2iECKO mice. AAV-SOX7 injection increased SOX7 in Shp2iECKO mice. Notably, no virusmediated SOX7 was expressed in Shp2f/f mice without Cre recombinase. The ZsGreen signal also supported the effectively expression of AAV in tumors.

Most conclusions about c-Jun and ASK1 roles in tumor angiogenesis are based on in vitro
experiments or Matrigel plugs. The authors need to verify them in orthotopic mouse models of cancer and in human tissue samples.

Reply. Thanks for the suggestions. ASK1 expression was increased in endothelial cells in LLC
tumors in Shp2iECKO mice (Fig. 7g) and in NSCLC tissues (Fig. 7i). Phospho-c-Jun was increased in endothelial cells in LLC tumors in Shp2iECKO mice (Fig. 6d). We tried to do immunofluorescence staining for phosphor-c-Jun in NSCLC tissues and failed to have high-quality results. Thus, with the new data, all components in the pro-angiogenic signaling axis SHP2-ASK1c-Jun-SOX7 are upregulated or activated in tumor associated endothelial cells. Reply. Thanks for the suggestion. As suggested, we used hCMEC (human cerebral microvessel endothelial cell) as another endothelial cells line (Supplementary Fig. 2). Similar to those in HUVECs, SHP2 knockdown in hCMEC impaired cell proliferation, migration and tube formation.

In this manuscript the authors investigated the tumor extrinsic function of SHP2 in
Together with the results we gained in primary lung endothelial cells isolated from Shp2 iECKO mice, SHP2 is essential for endothelial function.
2) The effect of SHP2 deletion in TME need more characterization. Only two syngeneic cell lines were used to generate tumors in the iECKO mouse model. Keys results in fig 1 and 2  Reply. Thanks for the suggestions. As suggested, we introduced orthotopic mouse mammary tumor model using E0771 mammary cancer cells ( Supplementary Fig. 1, 4). Similar to the results in other two mouse tumor models, SHP2 deletion reduced tumor angiogenesis and growth, as well as tumor vascular abnormalization. Thus, the anti-angiogenic function of SHP2 inhibition is not specific to certain tumors, while being specific to basic regulating mechanisms in tumor vessels. fig. S1a the author showed a not

4) How the standard anti VEGF treatment (e.g. bevacizumab) in mouse models compare with SHP2 deletion/inhibition in reducing tumor growth and promoting vascular normalization?
Reply. Thanks for the suggestions. In the figure 3, cediranib, a VEGFR inhibitor was used in a mouse tumor model. Our results showed that cediranib (1.5 mg/kg) exhibited anti-tumor effect.
SHP099 was comparable to cediranib in reducing tumor growth, which supports the anti-angiogenic function of SHP2 inhibition.
VEGFA-VEGFR signaling pathway is the major pro-angiogenic signal in endothelial cells. In clinician, anti-angiogenic therapy is only effective in a few tumors. Drug resistance is a major limitation. SHP2 mediates multiple receptor tyrosine kinase (RTK) signals in both cancer cells and endothelial cells. Targeting SHP2 will be more effective in anti-angiogenesis, compared with separately targeting VEGFA-VEGFR, or FGF-FGFR. Moreover, synergistic anti-tumor effect of SHP2 inhibitors and inhibitors for MAPK pathway such as MEK inhibitors has been well demonstrated in cancer cells in preclinical tumor models. It is rationale that SHP2 inhibitor, combining with existing drugs targeting VEGFR pathway will exhibit excellent anti-tumor effect and eliminate drug resistance. These are worth further investigations.  S2 and Fig 6).
Reply. Thanks very much for the suggestions. As suggested, in both HUVECs and hCMECs, SHP2 was re-expressed in SHP2-knockdown cells (Supplementary Fig. 2). All defects caused by SHP2 knockdown including cell proliferation, migration, and tube formation in endothelial cells were restored by SHP2 re-expression. To knockdown SHP2, targeting sequence was designed in the untranslated region (utr) in SHP2 mRNA. SOX7 and phosphorylated c-Jun was restored by the reexpression in SHP2 knockdown endothelial cells ( Supplementary Fig. 5a, 7ba). Similarly, SOX7 was re-expressed and defects caused by SOX7 knockdown were restored ( Supplementary Fig. 6ad). ASK1 was re-expressed to restore SOX7 expression in ASK1-knockdown endothelial cells ( Supplementary Fig. 8e). These new data reduced the possibility of off-target effect of shRNA and therefore enhanced our conclusion.

6) The authors claimed that the phosphatases activity of SHP2 is important to promote ASK1
stability.
Not enough evidences are provided in fig.6 9 . No commercial antibody for phosphor-ASK1 (Tyr718) is available.
In addition, SHP2 inhibitor increased ASK1 in cells expressing wildtype SHP2, but not in cells expressing inhibitor-resistant SHP2 (SHP2T253M/Q257L). The results again support the conclusion that phosphatase activity of SHP2 is essential to stabilize ASK1.

7) Based on the results provided in fig 5 a-b, the authors claimed that in endothelial cells, SHP2
promotes tumor angiogenesis with a MAPK-independent mechanism. However, not enough evidences are provided to support this conclusion. Contrasting data are also provided in fig. S3  Reply. Thanks a lot for bringing out these questions. It is cell-type dependent for SHP2 in regulating Ras-Erk signaling pathway. Our previous studies showed that phosphor-Erk was decreased upon  11 . In current manuscript, phosphor-Erk didn't change upon SHP2 inefficiency in both SHP2-knockdown HUVECs and Shp2-knockout mouse lung endothelial cells (Fig. 6a-b). Therefore, the role of SHP2 in regulating Ras-Erk signaling pathway relies on cell types.
The change in phosphor-Erk in HUVECs treated with SHP2 inhibitor didn't agree with that in SHP2-knockdown HUVECs. Phospho-Erk was decreased in HUVECs treated with RMC4550 ( Supplementary Fig 6c), but unchanged in SHP2-knockdown HUVECs (Fig 6a). Our results about RMC4550 were consistent with recent published data (Wang Y, et al. EMBO Mol Med 2021) 5 . We noticed the difference and therefore these experiments were repeated many times. Same issue was happened in our ongoing project. The off-target effect of SHP2 knockdown was excluded by SHP2 re-expression. We think it is possible that SHP2 allosteric inhibitors (SHP099 and RMC4550) have other targets than SHP2. Unfortunately, we don't have evidence for this yet.
SOX7 and phosphor-c-Jun were decreased in both SHP2 inhibition and knockdown HUVECs. Our data showed MEK inhibitor AZD6244 decreased SOX7 expression without affecting c-Jun signaling in endothelial cells (Figure 1). Moreover, SHP2 knockdown in HUVECs didn't change Erk activation induced by various growth factors ( Figure 2). Therefore, in endothelial cells, SOX7 expression is controlled by SHP2 via c-Jun signaling pathway. Western blot for SOX7, c-Jun, ERK, and p-38 and associated phosphorylated forms in HUVECs treated with MEK inhibitor (AZD6244, 24 h). β-Actin was used as a loading control. Quantitative data were expressed as mean ± SEM for three independent experiments. **, p < 0.01, ***, p < 0.001, by one-way ANOVA with multi-comparisons. Reply. Thanks for pointing out this incorrect. Various concentrations of RMC4550 were used to study endothelial function (supplementary Fig. 2a-c), SOX7 expression (supplementary Fig. 5d) and ASK1-c-Jun signaling (supplementary Fig. 7c, 8b). In a concentration dependent manner, RMC4550 decreased endothelial function and ASK1-c-Jun-SOX7 signaling.
Reply. Thanks for the suggestions. We have redone the experiments and new results were shown in the new version (Fig 3f to Fig 4f; Fig 5a to Fig 6a; Fig 5b to Fig 6b; Fig 5g to Fig 6h; Fig 5h to   Fig 6c; Fig S1f to Fig S1k; fig S3 g to Fig S5h).
Reply. Thanks for pointing out this mistake. We make a revision in the new version.

4) Graphs in fig 4d-e and j-m miss the legends.
Reply. Sorry for the careless errors. These errors are corrected in the new manuscript. Reply. Thanks for these nice suggestions. All references are added as suggested in our new manuscript.