Endothelial LRP1 regulates metabolic responses by acting as a co-activator of PPARγ

Low-density lipoprotein receptor-related protein 1 (LRP1) regulates lipid and glucose metabolism in liver and adipose tissue. It is also involved in central nervous system regulation of food intake and leptin signalling. Here we demonstrate that endothelial Lrp1 regulates systemic energy homeostasis. Mice with endothelial-specific Lrp1 deletion display improved glucose sensitivity and lipid profiles combined with increased oxygen consumption during high-fat-diet-induced obesity. We show that the intracellular domain of Lrp1 interacts with the nuclear receptor Pparγ, a central regulator of lipid and glucose metabolism, acting as its transcriptional co-activator in endothelial cells. Therefore, Lrp1 not only acts as an endocytic receptor but also directly participates in gene transcription. Our findings indicate an underappreciated functional role of endothelium in maintaining systemic energy homeostasis.

1. Fairly well established models are used here in terms of floxed mice and Tie2-Cre excision but given the importance to the hypotheses under study, expression of LRP1 in the liver should be demonstrated before and after Tie2Cre excision. This is relevant given that the liver also has endothelial cells although the structure of these ECs differ from peripheral ECs. Similarly, the reconstitution of LRP1 expression in hematopoietic cells (spleen, leukocytes) but not endothelial cells should be demonstrated, even if shown solely in Supplement.
2. The changes in higher LDL and lower HDL before high fat diet is noted but not the decrease in triglycerides, which is also relevant. Under basal conditions and prior to high fat, mice with deleted LRP1 have lower LDL, HDL and TG but no change in total cholesterol. This warrants some consideration? What were the VLDL levels?
3. This increased physical activity, which is probably due to increased oxidative metabolism of triglyceride-derived fatty acids in heart and skeletal muscle, may explain the decreased triglyceride level in Cre+/BMT mice (Figure 2c) Can the authors provide some data to backup the notion that increased fatty acid oxidation causes increased activity (as suggested in quoted text above)? Usually this is the other way aroundincreased activity results in increased fatty acid oxidation. The increased activity at night may reflect CNS effects. 4. . ...endothelium plays a role in insulin homeostasis, likely indirectly through the regulation of weight gain, adiposity or other mechanisms.
It appears the authors are suggesting the role of the endothelium in insulin homeostasis is indirect and a result of the decreased weight but the writing is not clear and creates an impression of a more direct or important role for endothelial LRP1 on insulin homeostasis when it would have to be considered be a consequence of the improved weight. How does weight gain differ from adiposity? 5. The evidence that LRP1 is increasing PPARγ activity can be bolstered by showing increased expression of canonical PPARγ target genes rather than just relying on PPRE luciferase activity, especially using such high levels of pioglitazone.
6. The finding that LRP1β also regulates the other 2 PPAR isoforms is notable. If this is the case, does LRP1 deletion result in phenotypes that are a consequence of PPARα deletion, which has also been shown to be important in systemic responses? Or PPARδ? 7. The authors may want to revisit the extent to which the discussion (as well as comments made in discussing results) are purely speculative. One example of many: All these data could be resulted from the direct involvement of endothelial cells in lipid and glucose metabolism. Another possibility is that the crosstalk between vascular endothelium and its neighboring lipogenic cells might be crucial for energy homeostasis. Some unknown secreted energy regulators or membrane protein-protein interactions might be involved in this crosstalk, which will become one of our future directions.
How LRP1 deletion in endothelial cells is resulting in increased activity remains unexplained and may need further validation given the importance to understanding the phenotype.
8. The notion of LRP1β as a novel, essential activator of PPARγ is novel and worthy of further exploration/discussion. How does this insight impact the numerous prior LRP1 reports, especially in regards to liver and fat? Do any of those examine action of PPARγ agonists and does the agonist fail to work in that setting?

Reviewer #3 (Remarks to the Author):
Mao H et al study the role of endothelial LPR1 on metabolism using a combination of mouse genetic models and in vitro cellular models. They demonstrate that specific deletion of LRP1 in endothelial cells improves insulin sensitivity, glucose tolerance and energy expenditure while body weight, LDL and HDL levels decrease in mice fed a high fat diet. It is noteworthy that a similar phenotype has been observed on glucose tolerance, energy expenditure and body weight in adipocyte-specific LRP1 knock out mice (Hoffman SM JCI 2017). At the molecular level, Mao et al demonstrate that LRP1 directly interacts with the nuclear PPAR receptors and increases their transcriptional activity.
Although PPARgamma has been shown to enhance LRP1 promoter activity via a PPRE in hepatocytes (Kim HJ Metabolism 2014), the direct interaction between both factors has not been reported so far. Therefore, it seems that part of LRP1's effect is mediated by PPARγ, although a proper demonstration of this model is lacking. In particular, it is unclear what part of the EC LRP1 KO phenotype is due to the lack of PPARg activation, vs activation of the other PPARs or even other mechanisms. The authors should at least show which part of the in vivo phenotype is mediated via the PPARg interaction, for instance by treating the mice with a glitazone (not pioglitazone which may also activate PPARalpha).
In addition, the mechanism regulating the effect of LRP1 in HFD-exposed endothelial cells is incomplete. Indeed, the interaction of PPARγ with LRP1 in endothelial cells does not explain why and how endothelial LRP1 improve insulin sensitivity, glucose tolerance... in mice fed a HFD. The authors should also demonstrate a role for this mechanism in the LRP1-dependent response to a HFD.
Thus, additional experiments should be performed to improve the whole message and strengthened conclusions.
Other comments 1. Figure 2m: As most effects are observed after High Fat Diet feeding, and to be able to compare data obtained in figure 2n-o, insulin tolerance tests should be performed in mice after HFD exposure.

2.
Figure 2a&b: How was LDL and HDL measured? It is assumed that it is rather LDL-cholesterol and HDL-cholesterol? The values are highly unusual for normal mice, which usually carry almost no cholesterol in LDL. The authors should perform lipoprotein cholesterol-distribution analysis by chromatography.
3. Figure 3c: the authors demonstrate that LRP1 does not interact anymore with PPARγ-ΔLBD. The deletion of PPARγ LBD domain may impair proper conformation of the remaining protein. The interaction of the PPARγ LBD with LRP1 (or LRP1 ICD domain) should be properly demonstrated. Figure 4d: do LRP1 deficient endothelial cells still respond to PPARγ agonists in this context?

5.
In addition, figure 1 and figure 2 demonstrate an effect of endothelial LRP1 in HFD-fed mice. What are the underlying mechanisms? What is the effect of long exposure of palmitate, for instance, on PPARγ target genes in both endothelial cell models?

6.
Figure 4g-i: the authors claim that these data may explain why LDL-C levels are higher in figure 2a. However, it does not explain why LDL and HDL levels are lower in figure 2a-b. Similar experiments should be performed in endothelial cells pre-treated with fatty acids and/or on endothelial cells isolated from mice fed a HFD.

7.
Does LRP1-ICD interact with PPARγ-LBD on the promoter of PPARγ target genes? This should be addressed by chromatin immunoprecipitation assays using the different constructs described in this study. Does pioglitazone influence LRP1 binding to PPARγ in this context? What is the effect of fatty acid sensitization on LRP1/PPARγ complex formation? On LRP1 recruitment to PPARγ target genes?

8.
The link between metabolic phenotype and the proposed mechanism is unclear. What is the effect of thiazolidinedione treatment on metabolic parameters, energy expenditure, insulin sensitivity, glucose tolerance in endothelial-specific LRP1 KO mice compared to control littermates?

9.
Mao et al study LRP1 as a coactivator, whose activity is mediated by PPARγ. To investigate such hypothesis, and to decipher the PPARγ-dependent from the PPARγ-independent activity, the authors should at least investigate the effect of LRP1 overexpression on cholesterol uptake (oxLDL loading) and PPARγ target gene expression in PPARγ-deficient or knock down endothelial cells.

10.
The authors should provide additional controls for the in vivo experiments. Since Cre expression may be sometimes non-specific and even spurious, they should check whether the Cre is not active and has knocked-out LRP1 in other metabolic tissues. The authors should also show that the in vivo phenotype is not due to Cre expression in ECs by showing that the Cre+ and Cre-mice (in the background of wt LRP1) display similar phenotypes at baseline and after HFD.
Interestingly, leptin levels decreased in endothelial LRP1 KO mice fed a HFD compared to WT (Figure 2f) mice while the food intake is unchanged (Figure 2g), suggesting an increase of leptin sensitivity in endothelial LRP1 KO mice. It would be interesting to address such possibility.

2.
As VO2 is increased in endothelial LRP1 knock out mice (Figure 2h), it would be interesting to provide the RQ (Respiratory Quotient) to get further insights about the substrate (lipid vs glucose) used by both mouse lines.

4.
Figure legends are confusing, international nomenclature should be used to describe mouse line.

5.
Statistics should be accurately described per panel.
Reviewer #1 (Remarks to the Author): These are novel and highly important observations that provides compelling evidence. Overall, this appears to be a well conducted study with valid approaches and appropriate conclusions drawn from the large amount of in vivo and in vitro data presented. The statistical methods appear appropriate.
1. While many of the phenotypes in endothelial-specific LRP1 deletion mimic that of PPARgamma endothelial specific knockout mice, the endothelial specific PPARgamma deficient mice do not show a difference in weight gain when fed a Western diet. This should be discussed.
We appreciate this great suggestion. The related discussion has been added into the Discussion section. Fig 2a and  We really appreciate the Reviewer's great suggestion about FPLC profiles. We have performed FPLC profiles with serum from Cre-/BMT and Cre+/BMT mice. The data showed that at week 0, even before highfat diet feeding, HDL-cholesterol, triglyceride (TG) levels were lower in Cre+/BMT mice than in Cre-/BMT control mice; however, VLDL-cholesterol and LDL-cholesterol levels were higher in Cre+/BMT mice than Cre-/BMT control mice (Supplementary Fig. 2a-d). Our FPLC data is consistent with the colorimetric assays performed for data shown in Fig. 2a-d. Following high-fat diet feeding for 16 weeks, TG level was significantly lower in Cre+/BMT mice than that in Cre-/BMT control mice, which was correlated to the dramatically decreased VLDL level in Cre+/BMT mice. LDL-cholesterol, HDL-cholesterol and total cholesterol were also significantly decreased in Cre+/BMT mice, compared to Cre-/BMT, following high-fat diet feeding. Taken all together, it suggests that endothelial LRP1 plays distinctive roles in lipid metabolism at the physiological condition and in response to hyperlipidemia stress. (Fig 2c) in Cre+/BMT mice suggests increased lipolysis in these mice. The activity of lipoprotein lipase on the endothelium should be quantified.

The decrease in TG
We really appreciate the great idea that the Reviewer provided to us. We have performed the LPL activity assay with mouse serum samples. The data was shown in Supplementary Fig. 2e. At week 0 and week 16, the LPL activity was indeed increased in Cre+/BMT mice, compared to Cre-/BMT control mice. This elevated LPL activity is inversely correlated with the reduction of serum TG level ( Fig. 2c) in Cre+/BMT mice comparing to Cre-/BMT control mice, suggesting that LRP1 depletion in endothelial cells resulted in increased lipolysis. However, we did not detect dramatic changes in endothelial lipase mRNA levels in endothelial LRP1 knockout mice, compared to wild type endothelial cells (data not shown). The specific roles of different lipases including hepatic lipase, lipoprotein lipase and endothelial lipase in endothelial LRP1-mediated metabolic phenotypes still need more careful studies and will become one of our future research directions. We have incorporated these data in the Results and Supplemental Information sections.
Reviewer #2 (Remarks to the Author): We thank the Reviewer for this great suggestion. We have performed immunostaining for liver sinusoidal endothelial cells of Cre-/BMT and Cre+/BMT mice. The immunostaining intensity of liver sinusoidal endothelial cells for Cre+/BMT is significantly decreased compared to Cre-/BMT control mice, suggesting knockout of LRP1 in liver sinusoidal endothelial cells ( Supplementary Fig. 1e). In addition, we measured LRP1 mRNA levels by using real-time PCR assays for isolated endothelial cells, leukocytes, spleen lymphocytes, adipocytes, liver, kidney and heart of Cre-and Cre+ mice before and after bone marrow transplantation ( Supplementary Fig. 1a, b). The mRNA levels of LRP1 were dramatically decreased in endothelial cells, leukocytes and lymphocytes in Cre+ mice, but not in adipocytes, liver, kidney and heart, compare to Cre-mice before bone marrow transplantation. After bone marrow transplantation, the mRNA levels of LRP1 in leukocytes and lymphocytes of Cre+/BMT mice were recovered back to similar levels as that of Cre-/BMT mice. However, the LRP1 mRNA level in endothelial cells was still significantly lower in Cre+/BMT mice than Cre-/BMT mice, suggesting that BMT resulted in the recovery of LRP1 expression in hematopoietic cells and the Cre+/BMT mice displayed a specific depletion of LRP1 in endothelial cells. We have incorporated these data in the Results and Supplemental Information sections.

The changes in higher LDL and lower HDL before high fat diet is noted but not the decrease in triglycerides, which is also relevant. Under basal conditions and prior to high fat, mice with deleted LRP1 have lower LDL, HDL and TG but no change in total cholesterol. This warrants some consideration? What were the VLDL levels?
The Reviewer has made a great point. As Reviewers suggested, we performed additional FPLC profiling assays to determine the specific lipoprotein particles. Our data based on both lipid colorimetric assays ( Fig.  2a-d) and FPLC analysis (Supplementary Fig. 2a-d) suggest that, under basal conditions and prior to high-fat diet feeding, VLDL-cholesterol and LDL-cholesterol levels were higher, while HDL-cholesterol level is lower in Cre+/BMT mice, compared to Cre-/BMT control mice. This might explain why no significant change in total cholesterol level was observed between Cre+/BMT and Cre-/BMT mice. Our FPLC analysis also confirmed that triglyceride levels were decreased before and after high-fat diet in Cre+/BMT mice, compared to Cre-/BMT mice. Next, we performed the LPL activity assay with mouse serum samples. At week 0 and week 16, the LPL activity was indeed increased in Cre+/BMT mice, compared to Cre-/BMT control mice ( Supplementary Fig. 2e). This elevated LPL activity is inversely correlated with the reduction of serum TG level ( Fig. 2c) in Cre+/BMT mice comparing to Cre-/BMT control mice, suggesting that LRP1 depletion in endothelial cells resulted in increased lipolysis. We have updated the Results and Discussions sections to reflect these findings.
3. This increased physical activity, which is probably due to increased oxidative metabolism of triglyceridederived fatty acids in heart and skeletal muscle, may explain the decreased triglyceride level in Cre+/BMT mice (Fig. 2c). Can the authors provide some data to backup the notion that increased fatty acid oxidation causes increased activity (as suggested in quoted text above)? Usually this is the other way aroundincreased activity results in increased fatty acid oxidation. The increased activity at night may reflect CNS effects.
We really appreciate the Reviewer's thoughtful comments and want to apologize for this misleading discussion. We have performed more thorough literature searches and totally agree with the Reviewer that increased activity usually results in increased fatty acid oxidation. This part has been corrected accordingly and updated in the Results section. Thank you for this great comment as this was not clear in the previous version of the manuscript. Yes, our data suggest that the role of the endothelium in insulin homeostasis is likely indirectly through the regulation of weight gain or other mechanisms. These related sentences have been modified to reflect our thoughts more clearly. Weight gain in response to high-fat feeding is due to the increase of adiposity and liver weight, based on our data in Figure 1. I hope that these changes have addressed your concern.
5. The evidence that LRP1 is increasing PPARγ activity can be bolstered by showing increased expression of canonical PPARγ target genes rather than just relying on PPRE luciferase activity, especially using such high levels of pioglitazone.
We highly appreciate the Reviewer's great comments. We have analyzed the expression of canonical PPARγ target genes in MEFs (Fig. 3g) and endothelial cells ( Fig. 4d-f, Supplementary Fig. 6a-d). In LRP1 knockout MEFs, PDK4 mRNA levels were decreased at basal condition and in response to pioglitazone treatment (Fig.  3g). In LRP1 knockout endothelial cells, there were much lower mRNA and protein levels of PPARγ target genes CD36, PDK4 and C/EBPα, compared to Cre-control cells (Fig. 4d-f). We also performed additional experiments to study how LRP1 activates PPARγ. First, we tested whether LRP1 regulates PPARγ target gene expression by knocking down PPARγ. As expected, treatments of overexpressed LRP1β, pioglitazone or both increased mRNA levels of PPARγ target genes. However, in PPARγ knockdown endothelial cells, these increases were all inhibited ( Supplementary Fig. 6c). In addition, we tested the effects of different agonists including thiazolidinediones (pioglitazone, ciglitazone, rosiglitazone and troglitazone) and palmitic acids on PPARγ target gene expression. As expected, these agonists increased mRNA levels of PPARγ target genes such as CD36, PDK4 and C/EBPα. However, these increases were inhibited in LRP1 depleted endothelial cells ( Supplementary Fig. 6d). Next, we performed chromatin immunoprecipitation (ChIP) assays to determine whether LRP1 is associated with the promoter of PPARγ target gene PDK4. Excitingly, the promoter sequence of PDK4 was detected in LRP1 bound chromatin complex, but not in control-IgG bound complex ( Supplementary Fig. 3a), suggesting that LRP1 is associated with the PDK4 promoter. Our immunoprecipitation experiments also demonstrated that the association of LRP1 and PPARγ was mildly increased by pioglitazone treatment (Supplementary Fig. 3b).
As the Reviewer commented, the in vivo luminescent imaging studies with PPRE-luc+ reporter mice were performed with a high dose of pioglitazone (150 mg/kg). Therefore, we repeated this experiments with much lower dose of pioglitazone (50 mg/kg). As expected, we still observed very strong luminescent signals in CAG-Cre-;PPRE-luc+ mice (revised Fig. 3f). However, the luminescent signals in CAG-Cre+;PPRE-luc+ mice were decreased at basal condition and in response to pioglitazone treatment. Based on our data from in vitro and in vivo reporter assays, real-time PCR and Western blotting of PPARγ target genes, ChIP and coimmunoprecipition assays, we conclude that LRP1 is required for ligand-dependent PPARγ activation and its target gene induction. We have updated these results in the Results section accordingly.
6. The finding that LRP1β also regulates the other 2 PPAR isoforms is notable. If this is the case, does LRP1 deletion result in phenotypes that are a consequence of PPARα deletion, which has also been shown to be important in systemic responses? Or PPARδ?
We appreciate this suggestion to determine whether other 2 PPAR isoforms were mediators for LRP1dependent metabolic responses. Our data demonstrate that LRP1β also binds to PPARα and PPARβ/δ and regulates their transcriptional activities ( Supplementary Fig. 4). It is likely that the changes in PPARα and PPARβ/δ activity in response to LRP1 deletion could also contribute to the improved systemic metabolic responses. We mainly study PPARγ since there is a well-performed report supporting a pivotal role of endothelial PPARγ in metabolic responses 1 . However, we do not and should not exclude the possible roles of other PPARs, which are beyond the scope of our current investigation but will surely become one of our future directions.
Nevertheless, our data suggest that PPARγ activity plays a role in LRP1-dependent metabolic responses. To further confirm the regulatory role of PPARγ, we have performed additional experiments to test whether PPARγ is required for the LRP1-dependent regulation of CD36, PDK4 and C/EBPα induction by using PPARγ specific siRNAs. Our results show that overexpression of LRP1β, treating endothelial cells with PPARγ agonist pioglitozone, or both resulted in increases in mRNA levels of PPARγ target genes CD36, PDK4 and C/EBPα. However, the expression levels of these genes were decreased significantly in PPARγ knockdown cells (Supplementary Fig. 6c). The oxLDL loading assays demonstrate that PPARγ knockdown significantly blocked cholesterol uptake that was induced by overexpression of LRP1β, pioglitozone or both ( Supplementary Fig. 6f). Taken together, our data provide strong evidence suggesting that LRP1-dependent PPARγ activation contributes, at least partially, to the changes of metabolic responses in endothelial LRP1 knockout mice.
7. The authors may want to revisit the extent to which the discussion (as well as comments made in discussing results) are purely speculative.
One example of many: All these data could be resulted from the direct involvement of endothelial cells in lipid and glucose metabolism. Another possibility is that the crosstalk between vascular endothelium and its neighboring lipogenic cells might be crucial for energy homeostasis. Some unknown secreted energy regulators or membrane protein-protein interactions might be involved in this crosstalk, which will become one of our future directions.
How LRP1 deletion in endothelial cells is resulting in increased activity remains unexplained and may need further validation given the importance to understanding the phenotype.
Thank you for these great comments. We have modified these discussion sections accordingly to avoid pure speculations.
In order to further understand how LRP1 deletion in endothelial cells is resulting in increased physical activity, we have performed the LPL activity assay with mouse serum samples. The data was shown in Supplementary Fig. 2e. At week 0 and week 16, the LPL activity was indeed increased in Cre+/BMT mice, compared to Cre-/BMT control mice. This elevated LPL activity is inversely correlated with the reduction of serum TG level (Fig. 2c) in Cre+/BMT mice comparing to Cre-/BMT control mice, suggesting that LRP1 depletion in endothelial cells resulted in increased lipolysis. This increased lipolysis ( Supplementary Fig. 2e) is well correlated with the increased physical activity (Fig. 2h-j) and decreased triglyceride level in Cre+/BMT mice (Fig. 2c). We have incorporated these data in the Results and Supplemental Information sections.
8. The notion of LRP1β as a novel, essential activator of PPARγ is novel and worthy of further exploration/discussion. How does this insight impact the numerous prior LRP1 reports, especially in regards to liver and fat? Do any of those examine action of PPARγ agonists and does the agonist fail to work in that setting?
This suggestion is really great. We haven't found any examinations of PPARγ agonists in prior LRP1 reports. Our observations provide new insights for LRP1's functional roles in metabolism, especially in regards to liver and fat. We have incorporated more detailed discussions in the Discussion section.

Although PPARgamma has been shown to enhance LRP1 promoter activity via a PPRE in hepatocytes (Kim HJ Metabolism 2014), the direct interaction between both factors has not been reported so far. Therefore, it seems that part of LRP1's effect is mediated by PPARγ, although a proper demonstration of this model is lacking. In particular, it is unclear what part of the EC LRP1 KO phenotype is due to the lack of PPARg activation, vs activation of the other PPARs or even other mechanisms. The authors should at least show which part of the in vivo phenotype is mediated via the PPARg interaction, for instance by treating the mice with a glitazone (not pioglitazone which may also activate PPARalpha). In addition, the mechanism regulating the effect of LRP1 in HFD-exposed endothelial cells is incomplete. Indeed, the interaction of PPARγ with LRP1 in endothelial cells does not explain why and how endothelial LRP1 improve insulin sensitivity, glucose tolerance... in mice fed a HFD. The authors should also demonstrate a role for this mechanism in the LRP1-dependent response to a HFD. Thus, additional experiments should be performed to improve the whole message and strengthened conclusions.
We totally agree with the Reviewer that further experiments need to be done to determine which part of the endothelial LRP1 knockout phenotype is due to the lack of PPARγ activation. Therefore, we performed additional in vitro and in vivo experiments to dissect which part of the in vivo phenotype is mediated via the PPARγ activation. We also determined whether LRP1/ PPARγ-dependent signaling plays a role in LRP1dependent responses following high-fat diet feeding, such as oxLDL-loaded cholesterol internalization in endothelial cells.
First, we tested whether LRP1 regulates PPARγ target gene expression by knocking down PPARγ or treating cells with PPARγ agonists. As expected, treatments of overexpressed LRP1β, pioglitazone or both increased mRNA levels of PPARγ target genes CD36, PDK4 and C/EBPα. However, these increases were all inhibited in PPARγ knockdown endothelial cells ( Supplementary Fig. 6c). On the other hand, PPARγ agonists including different thiazolidinediones (pioglitazone, ciglitazone, rosiglitazone and troglitazone) and palmitic acids increased mRNA levels of PPARγ target genes. However, these increases were inhibited in LRP1 depleted endothelial cells (Supplementary Fig. 6d). We also isolated endothelial cells from endothelial LRP1 depleted mice following high-fat diet feeding for 9 weeks. Consistently, mRNA levels of CD36, PDK4 and C/EBPα were increased in Cre-control endothelial cells in response to high-fat diet feeding. However, these increases were abolished in LRP1 depleted Cre+ cells (Supplementary Fig. 6e).
In addition to these in vitro experiments, we also investigated whether PPARγ agonists affect metabolic phenotypes of endothelial LRP1 depleted mice. Due to dual-specific effects of pioglitazone on both PPARγ and PPARα, we decided to test with both pioglitazone and another relatively specific PPARγ agonist-rosiglitazone. Specifically, we analyzed metabolic parameters, energy expenditure, insulin and glucose tolerance responses in endothelial LRP1 knockout or their littermate control mice following the treatment of rosiglitazone or pioglitazone for 3 and 4 weeks, respectively. Our results demonstrate that in response to rosiglitazone and pioglitazone, most of metabolic phenotypes resulted from endothelial LRP1 depletion were still detected in these mice, compared to Cre-/BMT mice ( Supplementary Fig. 7). Taken all together, our additional data provide further mechanistic support for our hypothesis that LRP1 is a coactivator of PPARγ and required for its ligand-dependent target gene induction.
We also performed oxLDL loading assays with PPARγ knockdown endothelial cells. Our results indicate that PPARγ was required for cholesterol internalization induced by overexpression of LRP1, pioglitazone treatment or both ( Supplementary Fig. 6f). In response to PPARγ agonists pioglitazone and palmitic acids, cholesterol uptake was increased. However, these increases were blocked in LRP1-depleted endothelial cells (Fig. 4h, Supplementary Fig. 6g). This suggests that PPARγ activity mediates LRP1-dependent cholesterol internalization. Last, we tested cholesterol uptake with endothelial cells isolated from 9-week-high-fat fed mice. We discovered that cholesterol uptake was significantly decreased in LRP1 knockout endothelial cells in control chow fed mice, compared to Cre-control cells (Supplementary Fig. 6h). However, very surprisingly, cholesterol uptake was increased dramatically in LRP1 knockout endothelial cells isolated from high-fat diet fed mice, compared to Cre-control endothelial cells isolated from high-fat diet fed mice, or LRP1 knockout endothelial cells isolated from control chow fed mice (Supplementary Fig. 6h). This increase, inversely correlated to decreased LDL-cholesterol level following high-fat diet feeding in endothelial LRP1 knockout mice (Fig. 2a, Supplementary Fig. 2a-b), could not be explained by the decreased induction of CD36 in the same cells ( Supplementary Fig. 6e). It suggests that endothelial LRP1 plays an active role in LDL-cholesterol clearance at basal condition and in response to hyperlipidemia. CD36 is required for LRP1-mediated cholesterol internalization at basal condition. However, hyperlipidemia stress activates CD36-independent mediators for cholesterol internalization.
This, together with our other observations such as the potential roles of endothelial LRP1 in HDL and triglyceride homeostasis, insulin sensitivity and glucose tolerance will become our future research directions. Nevertheless, our current results strongly suggest that endothelial LRP1 plays a pivotal role in the regulation of metabolic homeostasis, at least partially, through the regulation of PPARγ transcriptional activation.
Our data demonstrate that LRP1β also binds to PPARα and PPARβ/δ and regulates their transcriptional activities ( Supplementary Fig. 5). It is likely that the changes in PPARα and PPARδ activity in response to LRP1 deletion could also contribute to the improved systemic metabolic responses. We study PPARγ since there is a well-performed report supporting a pivotal role of endothelial PPARγ in metabolic responses 1 . However, we do not and should not exclude the possible roles of other PPARs, which are beyond the scope of our current investigation but will surely become one of our future directions.
1. Figure 2m: As most effects are observed after High Fat Diet feeding, and to be able to compare data obtained in figure 2n-o, insulin tolerance tests should be performed in mice after HFD exposure.
Thank you very much for this great suggestion. We added the data of insulin tolerance tests that we performed in Cre-/BMT and Cre+/BMT mice after high-fat diet feeding (Supplementary Fig. 2h). As mentioned in Fig. 2m, Cre+/BMT mice demonstrated improved insulin sensitivity compared to Cre-/BMT control mice. After high-fat diet feeding for 16 weeks (week 16), Cre+/BMT mice still demonstrated increased insulin sensitivity, compare to Cre-/BMT mice (Supplementary Fig. 2h).

2, Figure 2a&b: How was LDL and HDL measured? It is assumed that it is rather LDL-cholesterol and HDL-cholesterol? The values are highly unusual for normal mice, which usually carry almost no cholesterol in LDL. The authors should perform lipoprotein cholesterol-distribution analysis by chromatography.
Thank you very much for the suggestion. The LDL and HDL measurements were performed with colorimetric assays to quantify the cholesterol contents in LDL or HDL particles. As the Reviewers suggested, we have performed additional LDL, HDL measurements by using lipoprotein cholesterol-distribution analysis with FPLC. Indeed, the numbers of HDL-cholesterol and LDL-cholesterol levels are lower than the readings with colorimetric assays (Supplementary Fig. 2a, b). However, we observed similar trends in LDL and HDL-cholesterol changes in Cre+/BMT mouse serum, compared to Cre-/BMT control before and after high-fat diet feeding. All these data suggest that LRP1 depletion in endothelial cells contributes to the regulation of lipid homeostasis.
3. Figure 3c:  We appreciate the Reviewer's careful evaluation about the experiments for LRP1 and PPARγ interaction. We have demonstrated that the intracellular domain (ICD) of LRP1β bound to PPARγ by using GST pulldown assays with purified recombinant protein of LRP1-ICD (Fig. 3b). Actually we have also performed immunoprecipitation experiments to determine the interaction of purified PPARγ LBD protein (commercially available) and overexpressed Flag-tagged LRP1β protein. As expected, their interaction was observed (Fig. 3h). Together with the results with deletion mutations of PPARγ (Fig. 3c), we conclude that LRP1 is associated with PPARγ through the LBD domain of PPARγ and ICD domain of LRP1. We hope these data have addressed the Reviewer's concerns.

Figure 4d: do LRP1 deficient endothelial cells still respond to PPARγ agonists in this context?
The suggestion to examine the response upon PPARγ's agonists in LRP1 deficient endothelial cells is great. We checked the mRNA levels of PPARγ target genes in LRP1 deficient endothelial cells by using different PPARγ's agonists. Our data demonstrate that PPARγ's agonists, including different thiazolidinediones (pioglitazone, ciglitazone, rosiglitazone and troglitazone) and palmitic acids, increased mRNA levels of PPARγ target genes CD36, PDK4 and C/EBPα. However, these increases were inhibited in LRP1 depleted endothelial cells (Supplementary Fig. 6d). It suggests that LRP1, acting as PPARγ co-activator, is required for PPARγ activity at basal condition and in response to agonists. We have these data incorporated into the Results section accordingly.
5. In addition, figure 1 and figure 2 demonstrate an effect of endothelial LRP1 in HFD-fed mice. What are the underlying mechanisms? What is the effect of long exposure of palmitate, for instance, on PPARγ target genes in both endothelial cell models?
We appreciate this great suggestion and thoughtful questions. In response to the reviewer's questions, we have fed our mice with high-fat diet for 9 weeks and isolated endothelial cells from these mice. Then we analyzed mRNA levels of PPARγ's target genes CD36, PDK4 and C/EBPα in the hyperlipidemia exposed endothelial cells. Our results demonstrate that long term hyperlipidemia resulted in increased expression of PPARγ's target genes CD36 and PDK4, and mildly, C/EBPα. However, these increases were inhibited in LRP1 depleted endothelial cells (Supplementary Fig. 6e). In addition, we treated isolated endothelial cells with palmitic acids for 24 hours and then measured mRNA levels of these PPARγ's target genes. Our results indicated that, similarly to the hyperlipidemia exposure of endothelial cells in vivo, long term exposure of palmitic acids led to increases in mRNA levels of PPARγ's target genes in Cre-control endothelial cells. However, these increases were blocked in LRP1 depleted endothelial cells (Supplementary Fig. 6d). Taken all together, our data suggest that LRP1, acting as a co-activator of PPARγ, is required for expression of PPARγ's target genes at basal condition and in response to high fat challenge.
As I mentioned in the answer to your first comment, to understand the effect of endothelial LRP1 in HFD fed mice, we have investigated whether PPARγ agonists affect metabolic phenotypes of endothelial LRP1 depleted mice. Our results demonstrate that in response to rosiglitazone and pioglitazone, most of metabolic phenotypes resulted from endothelial LRP1 depletion were still detected in these mice, compared to Cre-/BMT mice (Supplementary Fig. 7). We also performed oxLDL loading assays with PPARγ knockdown endothelial cells. We demonstrated that PPARγ was required for cholesterol internalization induced by overexpression of LRP1, pioglitazone treatment or both ( Supplementary Fig. 6f). In response to PPARγ agonists pioglitazone and palmitic acids, cholesterol uptake was increased. However, these increases were blocked in LRP1-depleted endothelial cells (Fig. 4h, Supplementary Fig. 6g). This suggests that PPARγ activity mediates LRP1-dependent cholesterol internalization. Last, we tested cholesterol uptake with endothelial cells isolated from 9-week-high-fat diet fed mice. We discovered that cholesterol uptake was significantly decreased in LRP1 knockout endothelial cells in control chow fed mice, compared to Crecontrol cells (Supplementary Fig. 6h). However, very surprisingly, cholesterol uptake was increased dramatically in LRP1 knockout endothelial cells isolated from high-fat diet fed mice, compared to Crecontrol endothelial cells isolated from high-fat diet fed mice, or LRP1 knockout endothelial cells isolated from control chow fed mice ( Supplementary Fig. 6h). This increase, inversely correlated to decreased LDLcholesterol level following high-fat diet feeding in endothelial LRP1 knockout mice (Fig. 2a, Supplementary  Fig. 2a-b), could not be explained by the decreased induction of CD36 in the same cells ( Supplementary Fig.  6e). It suggests that endothelial LRP1 plays an active role in LDL-cholesterol clearance at basal condition and in response to hyperlipidemia. CD36 is required for LRP1-mediated cholesterol internalization at basal condition. However, hyperlipidemia stress activates CD36-independent mediators for cholesterol internalization.
This, together with our other observations such as the potential roles of endothelial LRP1 in HDL and triglyceride homeostasis, insulin sensitivity and glucose tolerance will become our future research directions. Nevertheless, our current results strongly suggest that endothelial LRP1 plays a pivotal role in the regulation of metabolic homeostasis, at least partially, through the regulation of PPARγ transcriptional activation.
We have included these data in the Results section. We appreciate the Reviewer's great suggestions. Yes, the data in Fig. 4g-i suggest a plausible molecular mechanism for the observed higher LDL-cholesterol level in Fig. 2a. To understand better why the LDLcholesterol level was lower following high-fat diet feeding in Cre+/BMT mice than that in Cre-/BMT control mice, we have performed two additional experiments. First, the isolated LRP1 depleted endothelial cells and Cre-controls cells were treated with palmitic acids for 24 hours and the internalized cholesterol levels were measured following oxLDL loading. Indeed, palmitic acids treatment, which increased CD36 mRNA levels ( Supplementary Fig. 6d), led to an increase in cholesterol internalization in Cre-endothelial cells. However, in LRP1 depleted endothelial cells, palmitic acids failed to increase CD36 mRNA level and cholesterol internalization ( Supplementary Fig. 6d, 6g). We also fed both Cre+ and Cre-mice with high-fat diet for 9 weeks and then isolated their endothelial cells. We performed oxLDL loading assays and discovered that cholesterol internalization was inhibited in Cre-control endothelial cells that have been exposed to high-fat diet for 9 weeks. However, very surprisingly, cholesterol uptake was increased dramatically in LRP1 knockout endothelial cells isolated from high-fat diet fed mice, compared to Cre-control endothelial cells isolated from high-fat diet fed mice, or LRP1 knockout endothelial cells isolated from control chow fed mice ( Supplementary Fig. 6h). This increase, inversely correlated with the decreased LDL cholesterol level in Cre+/BMT mice following high-fat diet feeding for 16 weeks (Fig. 2a), could not be explained by the decreased induction of CD36 in the same cells ( Supplementary Fig. 6e). It suggests that endothelial LRP1 plays an active role in LDL-cholesterol clearance at basal condition and in response to hyperlipidemia. CD36 is required for LRP1-mediated cholesterol internalization at basal condition. However, hyperlipidemia stress activates CD36-independent mediators for cholesterol internalization. As the Reviewer mentioned, we also observed the decrease of HDL-cholesterol level before and after high-fat diet feeding, which is similar to that phenotype with liver-specific LRP1 knockout mice 2,3 . However, whether and how endothelial LRP1 plays a role in HDL metabolism remains mystery and will become one of our future research directions. We have included these data and thoughts in the Results and Discussion sections. We appreciate these thoughtful questions. As the Reviewer suggested, we have performed chromatin immunoprecipitation (ChIP) assays and included the data in our revised manuscript. The stable LRP1β transfected HEK293 cells were used to make soluble chromatin. Following the immunoprecipitation with the chromatin fraction by using LRP1β antibody or rabbit IgG as a control, the promoter of PPARγ target gene PDK4 was detected by PCR with its specific primers. Our ChIP data demonstrate that the PDK4 promoter was detected in LRP1β associated complex, but not in the rabbit IgG associated complex. The association was still observed upon treatments of pioglitazone or palmitic acids ( Supplementary Fig. 3a). We also performed immunoprecipitation experiments to test whether pioglitazone influences LRP1 binding to PPARγ. In mouse endothelial cells, the binding of LRP1 and PPARγ was mildly increased along the treatments of pioglitazone for 5 to15 minutes ( Supplementary Fig. 3b). We have incorporated these data into the Results section.

The link between metabolic phenotype and the proposed mechanism is unclear. What is the effect of thiazolidinedione treatment on metabolic parameters, energy expenditure, insulin sensitivity, glucose tolerance in endothelial-specific LRP1 KO mice compared to control littermates?
We appreciate this great suggestion. As the Reviewer suggested, we have performed further analysis with metabolic parameters, energy expenditure, insulin sensitivity and glucose tolerance following i.p. injection of rosiglitazone or pioglitazone for 3 or 4 weeks, respectively. As shown in Supplementary Fig. 7, following rosiglitazone or pioglitazone injections, endothelial-specific LRP1 knockout mice still demonstrated similar changes in metabolic parameters, increased physical activity, improved insulin sensitivity and glucose tolerance responses. It suggests that LRP1 depletion in endothelial cells leads to improved metabolic responses at basal condition and in response to PPARγ activation by rosiglitazone or pioglitazone. These data further support our hypothesis that LRP1 is a co-activator of PPARγ and is required for its ligand-dependent activation.  Fig. 6c). We also observed that PPARγ was required for cholesterol internalization induced by overexpression of LRP1, pioglitazone treatment or both ( Supplementary Fig. 6f). Taken together, our results further demonstrate that LRP1β is required for the ligand-dependent PPARγ target gene induction and cholesterol uptake. We greatly appreciate the Reviewer's careful evaluation of our in vivo data and totally understand the concerns of the Reviewer. We have performed immunostaining for liver sinusoidal endothelial cells of Cre-/BMT and Cre+/BMT mice. The immunostaining intensity of liver sinusoidal endothelial cells for Cre+/BMT is significantly decreased compared to Cre-/BMT control mice, suggesting knockout of LRP1 in liver sinusoidal endothelial cells (Supplementary Fig. 1e). As the Reviewers suggested, we also measured LRP1 mRNA levels by using real-time PCR assays for isolated endothelial cells, leukocytes, lymphocytes, adipocytes, liver, kidney and heart of Cre-and Cre+ mice before and after bone marrow transplantation ( Supplementary Fig. 1a, b). The mRNA levels of LRP1 were dramatically decreased in endothelial cells, leukocytes and lymphocytes in Cre+ mice, but not in adipocytes, liver, kidney and heart, compare to Cremice ( Supplementary Fig. 1a). After bone marrow transplantation, the mRNA levels of LRP1 in leukocytes and lymphocytes of Cre+/BMT mice were increased to similar levels as that of Cre-/BMT mice. However, the LRP1 mRNA level in endothelial cells was still significantly lower in Cre+/BMT mice ( Supplementary  Fig. 1b). These data suggest that LRP1 expression was recovered in hematopoietic cells and LRP1 in endothelial cells was specifically depleted after bone marrow transplantation.
Regarding to the concern about the non-specific effects of Tie2Cre transgene, before we started these in vivo experiments, we have searched literatures about the Tie2Cre transgenic mice and found from many reports that no obvious changes in metabolic responses were observed in Tie2Cre transgenic mice or Tie2Cremediated knockout mice. For example, studies have been done to compare the differences of Tie2Cre transgenic mice and wild type mice in metabolic responses and no obvious differences in body weight and blood glucose and insulin levels and other metabolic responses were observed 4 . Another study has tracked metabolic responses until mice were one year old and found that the Tie2Cre-mediated insulin receptor knockout mice did not change whole-body glucose tolerance, insulin sensitivity and plasma lipids compared to the intact insulin receptor control mice 5 . Therefore, we hope that these reports have addressed the Reviewer's concerns.

Minor comment
1. Interestingly, leptin levels decreased in endothelial LRP1 KO mice fed a HFD compared to WT ( Figure  2f) mice while the food intake is unchanged (Figure 2g), suggesting an increase of leptin sensitivity in endothelial LRP1 KO mice. It would be interesting to address such possibility.
This is a great suggestion. We have included this point in the Results section.
2. As VO2 is increased in endothelial LRP1 knock out mice (Figure 2h), it would be interesting to provide the RQ (Respiratory Quotient) to get further insights about the substrate (lipid vs glucose) used by both mouse lines.
Thanks for this great comment. We have calculated RER (respiratory exchange ratio), an estimate of RQ, for the data of indirect calorimetry studies. There were no significant differences in RER values of Cre+/BMT and Cre-/BMT mice ( Supplementary Fig. 2g). Both values were at 0.91~0.95 and close to 1.0, suggesting that carbohydrate was the primary fuel source for both endothelial LRP1 knockout and control mice.
Thank the Reviewer for the careful reading. We would like to apologize for our oversight. 'finofibrate' was replaced by 'fenofibrate' in our revised manuscript. Figure legends are confusing, international nomenclature should be used to describe mouse line.

4.
We have corrected the names of the mouse lines in the figure legends.

Statistics should be accurately described per panel.
As the Reviewer suggested, the statistics were updated in the supplemental Methods and the related figure legends.