Xinqidi,Your biological research partner
Xinqidi Biotech Co.,Ltd,Wuhan,China 2008-2021
R&D 13th year

Hexokinase 2-driven glycolysis in pericytes activates their contractility leading to tumor blood ves

Issuing time:2021-10-22 10:36

Abstract

Defective pericyte-endothelial cell interaction in tumors leads to a chaotic, poorly organized and dysfunctional vasculature. However, the underlying mechanism behind this is poorly studied. Herein, we develop a method that combines magnetic beads and flow cytometry cell sorting to isolate pericytes from tumors and normal adjacent tissues from patients with non-small cell lung cancer (NSCLC) and hepatocellular carcinoma (HCC). Pericytes from tumors show defective blood vessel supporting functions when comparing to those obtained from normal tissues. Mechanistically, combined proteomics and metabolic flux analysis reveals elevated hexokinase 2(HK2)-driven glycolysis in tumor pericytes, which up-regulates their ROCK2-MLC2 mediated contractility leading to impaired blood vessel supporting function. Clinically, high percentage of HK2 positive pericytes in blood vessels correlates with poor patient overall survival in NSCLC and HCC. Administration of a HK2 inhibitor induces pericyte-MLC2 driven tumor vasculature remodeling leading to enhanced drug delivery and efficacy against tumor growth. Overall, these data suggest that glycolysis in tumor pericytes regulates their blood vessel supporting role.

Introduction

Pericytes have an important regulatory role in stabilizing blood vessel wall, controlling vessel dilation, and regulating vascular perfusion1,2,3. Unlike other host cells, pericyte expresses specific surface markers/receptors such as platelet derived growth factor receptor β (PDGFRβ) and CD146 (also known as the melanoma cell adhesion molecule (MCAM)), and respond to specific growth factors including PDGF, transforming growth factors and angiopoietins2,4. During angiogenesis, endothelial cells secrete PDGF to recruit pericytes to newly formed vessels via activation of the PDGFRβ signaling1,5. Unlike PDGFRβ, CD146 was originally identified as an endothelial cell marker during angiogenesis6, which has been recently shown to highly express in pericytes too, and important for regulating pericyte-endothelial cell interaction during blood–brain barrier development4. Therefore, these two surface proteins are routinely used as positive markers for pericyte identification.

Just as tumor endothelial cells differ from quiescent endothelial cells in normal tissues, tumor-derived pericytes appear to be abnormal or dysfunctional when compared with pericytes in normal tissues7,8. It is unclear whether cancer cells instruct the pericyte fate changes to facilitate tumor growth. Indeed, defective pericyte-endothelial cell interaction is observed in many cancers, which is characterized by compromised blood vessel integrity9,10. For instance, the loss or detachment of pericytes from tumor blood vessels is frequently observed in some cancers and is believed to promote cancer metastasis11,12. On the other hand, pericytes have been shown to protect endothelial cells from anti-angiogenic drug (i.e. bevacizumab), and therefore combination treatment with PDGFRβ tyrosine kinase inhibitor (i.e. eliminating PDGFRβ positive pericytes) and VEGF inhibition more efficiently block tumor angiogenesis than anti-VEGF therapy alone in several animal models13,14. However, the combination therapy has failed to provide any therapeutic benefit in the clinical trial of renal carcinoma patients and even shows additive toxicity in the patients14,15. Therefore, it is a clinically unmet need to explore the molecular mechanism of pericyte-endothelial cell interaction in preclinical setting. Due to the lack of a cellular model for studying human pericyte biology, the exact cause of abnormal pericyte behavior in tumors is unknown. Previous studies show that the regulation of metabolic pathways determines the cell state and fate decisions in endothelial cells16. For instance, Inhibition of the glycolytic activator PFKFB3 in endothelial cells can improve pericyte coverage and blood vessel function, which therefore enhances the delivery and efficacy of chemotherapeutics drugs in preclinical animal models17. However, the role of metabolic rewiring in human pericyte biology still remains largely unknown.

In this study, we establish the role of pericyte-hexokinase 2 (HK2) driven glycolysis in remodeling tumor vasculature. By developing a method for isolation and culture of pericytes from normal adjacent tissues (NPC) and tumors (TPC) derived from NSCLC (non-small-cell lung cancer) and HCC (hepatocellular carcinoma) patients respectively, we showed that TPC had abnormal blood vessel supporting functions as compared with NPC in two different cancers. Combined proteomics and metabolic flux analysis revealed an increased HK2-driven glycolysis in TPC as compared with NPC, which prohibited its blood vessel supporting role via activation of the ROCK2-MLC2 mediated contractility. Importantly, depletion of HK2/ROCK2 restored the blood vessel supporting function of TPC, while treatment of HK2 inhibitor induces MLC2-driven tumor vasculature remodeling to enhance chemotherapy delivery and efficacy against tumor growth. The role of pericyte-HK2 expression in this process was further illustrated by the fact that administration of chemotherapy reduced tumor growth in mice that were co-injected with tumor cells and HK2-depleted TPC as compared with mice that were co-injected with tumor cells and scramble transfected TPC.

Overall, our work develops a cellular model for studying pericyte biology and uncovers that pericyte-HK2-driven glycolysis induces tumor blood vessel abnormality by activating ROCK2-MLC2 mediated contractility.

Result

Development of the microbead activated vascular cell sorting enrichment method coupled with FACS

Although previous studies have managed to isolate pericytes from mouse brain tissues18,19, there is no effective protocol for isolation and culture of human pericytes from paired normal adjacent tissues and tumors. Due to the fact that both lung and liver tissues contain a rich population of vascular cells, they were therefore chosen as a source of human pericyte isolation in this study (Supplementary fig. 1). Since anti-human CD146 antibody conjugated with FITC and anti-FITC magnetic microbeads are commercially available, we proposed to employ a CD146+-FITC-microbead activated cell sorting (MACS) method to enrich the vascular cell population from tissue samples first, and then used positive and negative selection markers to isolate pericyte population via fluorescence activated cell sorting (FACS) (Supplementary fig. 1). In order to carry out our pericyte purification method, we first examined whether CD146 was expressed on vascular endothelial cells and pericytes in normal adjacent tissue and tumors derived from NSCLC and HCC patients. Based on the fact that CD34 and α-SMA/PDGFRβ is known as a vascular endothelial cell and pericyte marker respectively7,20, they were used for defining the CD146 staining in this study. We then performed triple immunofluorescent staining of these markers with paired normal adjacent tissues and tumors derived from NSCLC and HCC patients respectively, showing that CD146 was expressed in both CD34 positive (+ve) vascular endothelial cells and αSMA/PDGFRβ+ve pericytes within normal adjacent tissues and tumors derived from NSCLC and HCC patients, while ~65–72% percentage of blood vessels were CD146 and α-SMA double positive in both normal adjacent tissues and tumors derived from NSCLC and HCC patients respectively (Fig. 1a–f and Supplementary fig. 2a, b). No significant difference in the percentage of CD146 and α-SMA double positive blood vessels was observed between normal adjacent tissues and tumors (Fig. 1c, f). In addition, we showed that α-SMA was also expressed in PDGFRβ+ve pericytes around CD34 positive blood vessels (Supplementary fig. 2c, d), which was consistent with the previous finding1. Therefore, our results suggested that CD146+-FITC-microbead activated cell sorting method could be used to enrich the vascular cell population from tissue samples derived from NSCLC and HCC patients.

Fig. 1: Purification of pericytes from normal adjacent tissues and tumors derived from NSCLC and HCC patients using CD146+--FITC-MACS enrichment method.
figure1

af Representative triple immunostaining images of CD34 (white), CD146 (green), and α-SMA (red) from 2 different microscopic fields in normal adjacent tissues and tumors derived from NSCLC and HCC patients are given. Bar charts represent the percentage of CD146 and α-SMA-double positive (+ve) blood vessels per field in each group (n = 10 NSCLC/HCC patients). Statistical tests were two-sided. gj Representative FACS plots showing the gating strategy for the purification of pericytes from normal adjacent tissues and tumors derived from NSCLC patients with (+) or without (−) CD146+-FITC-microbead cell soring enrichment (MACS) method. Debris, doublets (TOF), and dead cells (FVD) were excluded, then CD45CD31CD146+PDGFRβ+ pericytes were collected. Percentage refers to the proportion of cells in the previous parent gate. k, l Bar charts show the relative fold change in percentage of pericytes and endothelial cells isolated by the MACS enrichment coupled with FACS protocol compared to that isolated by FACS alone. Statistical tests were two-sided. Percentages of pericyte refer to cell population in the R1 gate. NSCLC non-small cell lung cancer, HCC hepatocellular carcinomas, FSC forward-scatter; SSC side-scatter, TOF time of fly, FVD fixable viability dye, EC endothelial cells, PC pericytes, NEC normal adjacent tissue derived endothelial cells, TEC tumor-derived endothelial cells, NPC normal adjacent tissue derived pericytes, TPC tumor-derived pericytes. FACS plots/bar charts are representative of three individual patient data. Means ± SEM are given. c, f, k, l Student’s t test. Scale bars in a, b, d, e represent 50 μm, a, b, d, e(magnified pictures) represent 25 μm.

To isolate pericytes from normal adjacent tissues and tumors, the dead cells and debris were first removed from mechanically minced and enzymatically digested tissues. The single cell suspension was first labeled with FITC-conjugated CD146 antibody and then enriched by using anti-FITC magnetic microbeads. The enriched vascular cell population was subsequently labeled with antibodies specific for endothelial cells (CD31), immune cells (CD45), and pericyte markers (PDGFRβ). The labeled cells were then collected by using FACS (Supplementary fig. 1). FACS plots showed the gating strategy for the separation of endothelial cells (EC) and pericytes (PC) from normal adjacent tissues (NEC and NPC) and tumor tissues (TEC and TPC) derived from NSCLC and HCC (Fig. 1g–j and Supplementary fig. 2e–h). Importantly, our MACS enrichment method successfully improved the efficacy of isolating human pericytes from paired normal adjacent tissues and tumors derived from NSCLC and HCC patients almost by 7–13 folds as compared to FACS method alone (Fig. 1k, l). Compared with FACS method alone, our protocol enhanced the successful chance of isolating enough pericytes for subsequent culturing and characterization and functional studies (Supplementary fig. 1). Overall, this method enabled us to obtain pure pericytes in culture with high reproducibility, mainly due to MACS enrichment method and temperature/time-controlled incubation with optimized amount of collagenase and DNase I.

Characterization of the isolated pericytes from normal adjacent tissues and tumors derived from NSCLC and HCC patients

To confirm the purity of our isolated pericytes derived from normal adjacent tissues and tumors, we performed flow cytometry and immunostaining experiments with our isolated pericytes, indicating that the TPC and NPC derived from NSCLC or HCC patients expressed PDGFRβ and CD146 as well as newly identified pericyte markers desmin and CD13, but did not express endothelial cell marker CD31 and CD34, immune cell marker CD45, or fibroblast marker FAP (fibroblast associated protein) (Fig. 2a, b and Supplementary fig. 3a, b). Further RT-PCR analysis confirmed that these pericytes either expressed relatively low or undetectable levels of endothelial cell markers CD31 and CD34, immune cell marker CD45, smooth muscle cell markers CNN1 and MYOCD, and pericyte/fibroblast marker ACTA2, fibroblast markers FAP and PDGFRA (platelet derived growth factor α) as compared with human umbilical vein endothelial cells (HUVEC), peripheral blood mononuclear cell (PBMC), and human fibroblasts respectively (Fig. 2c and Supplementary fig. 3c). In contrast, both NSCLC/HCC-derived NPC and TPC strongly expressed pericyte markers PDGFRB and CD146 as compared to fibroblasts and HUVEC respectively (Fig. 2c and Supplementary fig. 3c). In addition, no significant difference in the expression level of pericyte markers between either NPC and TPC or TPC at passage 3 and 12 was observed (Fig. 2a–c and Supplementary fig. 3a–e), suggesting that our culturing method could maintain the pericyte identity of the isolated cells over passages.

Fig. 2: Characterization of normal adjacent tissue derived pericytes and tumor-derived pericytes.
figure2

a Flow cytometry analysis of pericytes isolated from normal adjacent tissue and tumor-derived from NSCLC patients showed that the purity of pericyte preparation was high. Pericyte preparations displayed good expression of pericyte markers PDGFRβ, CD146, and CD13 and undetected or low level of endothelial cell markers CD31 and CD34 as well as immune cell marker CD45. b Immunofluorescent staining analysis of the endothelial cell, fibroblast, pericyte marker expression in NPC, TPC, HUVEC, and fibroblast (n = 3 independent samples). c RT-PCR analysis of the endothelial cell, fibroblast, smooth muscle cells, pericyte marker expression in NPC, TPC, HUVEC, fibroblasts, peripheral blood mononuclear cell (PBMC), smooth muscle cells (SMC) (for CD31, CD34, CD45, CNN1, MYOCD, and ACTA2 expression analysis: n = 4 independent NPC and TPC samples and n = 3 independent SMC, PBMC and HUVEC samples; for FAP, PDGFRA, PDGFRB, and CD146 expression analysis: n = 3 independent samples for each cell type). d Doubling time of NPC and TPC derived from NSCLC/HCC patients (n = 3 independent experiments). e β-galactosidase staining assay showed the percentage of senescent cells at different passages of NSCLC/HCC-derived NPC and TPC (n = 3 independent experiments). Statistical tests were two-sided. f, g Wound healing assay showed no difference in the migration ability of NSCLC/HCC-derived NPC and TPC (n= 3 independent experiments). Statistical tests were two-sided. Means ± SEM are given. c One-way ANOVA. d Two-way ANOVA. eg Student’s t test. Scale bars in b represent 50 μm, eg 100 μm.

Next, we examined their cell morphology, proliferation, doubling time, passage number, senescence, viability, and migration ability. Phase contrast microscope analysis revealed that both NPC and TPC had the morphological characters of pericytes as observed previously in tissue section, which possessed a cell body with a prominent nucleus and a small content of cytoplasm with several long processes (Fig. 2b and Supplementary fig. 3b). Functional studies revealed that the NPC and TPC derived from NSCLC or HCC patients had similar doubling times and could be cultured up to 12 passages with gradual senescence observed during 8–12 passages (Fig. 2d, e), while there was no significant difference in proliferation (Supplementary fig. 3f, g), cell viability (Supplementary fig. 3h), and migration (Fig. 2f, g) between NPC and TPC. To determine whether NPC and TPC showed any difference in endothelial cell dependent recruitment, we plated NPC or TPC in the upper compartment of a transwell, while HUVEC cells were placed into the bottom compartment, indicating that NPC and TPC displayed no difference in response to HUVEC dependent recruitment (Supplementary fig. 3i). Taken together, our work developed a cellular model to study the functional differences between NPC and TPC.

Tumor-derived pericyte possesses defective blood vessel supporting role

Pericytes have an important regulatory role in stabilizing blood vessel wall by regulating endothelial cell properties1,9. To determine the role of NPC and TPC in endothelial cell proliferation and migration, we exposed HUVEC cells to conditioned medium from NPC/TPC derived from NSCLC or HCC patients, indicating that the conditioned media from either NPC or TPC increased the proliferation of HUVEC (Fig. 3a), while neither of them showed any effect on HUVEC migration in a wound healing assay (Fig. 3b). To examine the functional role of NPC and TPC in blood vessel formation, we fluorescently labeled NPC and TPC as well as fibroblasts and smooth muscle cells (SMC), and then co-cultured HUVEC with either of them on matrigel. After co-culturing, we evaluated their effect on total tube length, branch points, and numbers (Fig. 3c–f and Supplementary fig. 4a–d). Although NPC, TPC, fibroblasts, and smooth muscle cells aligned along tubes when added to HUVECs, they had different effects on tube lengths and branching. Strikingly, co-culturing HUVEC with TPC reduced total tube length and branch points and number as compared to HUVEC co-cultured with either NPC, fibroblasts, or HUVEC alone, while normal fibroblasts had no effect on all of these (Fig. 3c–f and Supplementary fig. 4a–d). Unlike NPC, TPC or normal fibroblast, SMCs pulled tubes apart following alignment and caused a rapid regression of the tubular network within 24 h (Fig. 3c–f and Supplementary fig. 4a–d).

Fig. 3: Tumor-derived pericytes display defective blood vessel supporting functions in vitro and in vivo.
figure3

a CCK8 proliferation assay showed that exposure with conditioned medium from NSCLC/HCC-derived NPC or TPC significantly enhanced the proliferation of HUVEC as compared to untreated HUVEC (n = 5 experimental repeats). b Wound healing assay of the HUVEC after exposed with condition medium from NSCLC/HCC-derived NPC or TPC. Representative pictures of wound healing assay in each group are given (n = 3 independent experiments). c HUVEC were either cultured alone or co-cultured with CFSE (green fluorescent dye) labeled NSCLC-derived-NPC/-TPC, fibroblast or smooth muscle cells (SMC) in a tube formation assay. df Bar charts show the relative total tube length, branch points and number in each group (n = 3 independent experiments). gRepresentative images of the serial sections of matrix plugs after immunostained with CD34, PDGFRβ, or collagen IV antibody in each group are given. hk Violin plots show the quantification of the total number of blood vessels, lumen diameter, collagen IV level and percentage of coverage in each group (n = 3 mice per group). Results are given as means ± SEM. a Two-way ANOVA. b, df, hk One-way ANOVA. Scale bars in b represents 100 μm, c 200 μm, g 50 μm.

To evaluate the blood vessel supporting role of NPC and TPC in vivo, we performed in vivo matrix plug assays by embedding NPC/TPC/fibroblasts/SMC in matrigel-fibrin matrix with HUVEC, and then implanted them into nude mice subcutaneously. Matrix gel plug assays indicted that co-transplantation of HUVEC and NSCLC/HCC-derived-NPC and-TPC significantly increased the number of blood vessels as compared with HUVEC transplant alone (Fig. 3g, h). Further histological examination showed that co-transplantation of HUVEC with NPC, but not TPC, supported the formation of functional blood vessels (containing red blood cells) with strong basement membrane marker collagen IV expression (Fig. 3g–j), while smaller and irregular shaped blood vessels, reduced collagen IV staining were observed in HUVEC co-transplanted with TPC (Fig. 3g–j). Previous studies have shown that the expression of basement membrane marker collagen IV has been implicated in blood vessel integrity and perfusion1,7. As expected, both smooth muscle cells and fibroblasts had a very limited effect on tumor blood vessel formation and collagen IV expression (Fig. 3g, h, j), while SMC reduced the lumen diameter of blood vessels most significantly as compared with TPC and NPC (Fig. 3g, i). Finally, neither NPC, TPC, fibroblast, or SMC showed difference in tube coverage (Fig. 3k).

Since pericytes have been implicated in regulating capillary dilation and contraction1,14, we therefore assessed the contractile properties of NPC and TPC by performing time-lapse microscopy studies with them after treated with carbachol. After the treatment, NSCLC/HCC-derived TPC strongly contracted and showed up to a 30% change in surface area, while NSCLC/HCC-derived NPC displayed a modest change in their surface area (Supplementary fig. 4e, f). In contrast, no significant change in the surface area of HUVEC/fibroblast was observed post-treatment (Supplementary fig. 4e, f). Together, these results showed the functional difference between NPC and TPC in vitro and in vivo, indicating that TPC had poor BV supporting functions as compared with NPC, which may be linked to its increased contractility phenotype.

Combined proteomics and metabolic flux analysis reveals an elevated HK2-driven glycolysis in TPC

Metabolic reprogramming determines cell proliferation, migration, and contractility21. Indeed, endothelial cells undergo PFKFB3 dependent glycolytic switch to match their high energy demand for migration and proliferation during angiogenesis. However, the role of metabolic state in determining pericyte’s BV supporting function remains unknown. In addition, it is unclear whether NPC and TPC express different glycolytic gene signature. We therefore performed RT-PCR analysis of the NPC and TPC isolated from NSCLC or HCC patients to compare their glycolytic gene pathway expression, indicating that NPC and TPC had distinct difference in the expression of genes associated with glycolysis (Fig. 4a, b). In both NSCLC and HCC, the expression of glycolytic gene, hexokinase 2 (HK2) was significantly up-regulated in TPC as compared to NPC (Fig. 4a, b). Although previous studies indicated that inhibition of PFKFB3 in human placenta pericytes prohibited their cell migration and proliferation abilities17, we did not observe any difference in the PFKFB3expression between NPC and TPC (Fig. 4a, b). In addition, proteomics analysis indicated that the expression of HK2, but not PFKFB3 as well as other glycolytic enzymes and glucose transporters, was significantly increased in TPC as compared to NPC (Fig. 4c). Western blot analysis confirmed the up-regulation of HK2 protein expression in NSCLC/HCC-derived TPC as compared with NPC (Fig. 4d). Based on these findings, we next examined the glycolytic capacity of NPC and TPC by performing glucose uptake assay, lactate assay, Seahorse analysis as well as metabolic flux analysis. As expected, NPC and TPC showed no difference in the glucose uptake, while the lactate production was significantly increased in TPC as compared with NPC (Fig. 4e, f). Further seahorse assay indicated an increase in extracellular acidification rate (ECAR) but decreased oxygen consumption rate (OCR) in TPC as compared that to NPC (Fig. 4g, h and Supplementary fig. 5a, b), implying that TPC relied on glycolysis for their energy production. To obtain the mechanistic underpinnings of the deregulated TPC metabolism, we carried out 13C-labeled metabolic flux analysis using U-13C6 glucose to estimate metabolic fluxes in TPC and NPC derived from NSCLC or HCC patients. Metabolic tracking revealed the relative abundance of glycolysis metabolite M3 lactate was increased in TPC as compared that to NPC, while no significant difference in the relative abundance of TCA (tricarboxylic acid) cycle metabolites directly derived from U-13C6 glucose, such as M2 citrate, M2 ketoglutarate, M2 succinate, M2 fumarate, and M2 malate, was observed between NPC and TPC (Fig. 4i and Supplementary fig. 5c).

Fig. 4: Combined proteomics and metabolic flux analysis reveals an elevated HK2-driven glycolysis in TPC as compared with NPC.
figure4

a Metabolic network shows the metabolic reactions involved in glycolysis (based on Kyoto Encyclopedia of Genes and Genomes pathway, KEGG), together with main metabolites and 22 selected glycolytic genes. Heatmap analysis of the 22 selected glycolytic gene expression in NSCLC/HCC-derived NPC and TPC (n = 3 individual patients). b Table shows the average fold change and statistical analysis of 22 glycolytic gene expression between NPC and TPC. c Proteomics analysis showed that hexokinase 2 (HK2) was significantly up-regulated in NSCLC-derived TPC as compared with NPC. The x-axis of the volcano plot shows the log2 fold change of each identified proteins and the y-axis represents the −log10 of P-values. Red dots represent significant up-regulated proteins and green dots for significant down-regulated proteins (p < 0.05, fold change more than 2 (log2 = 1)). The horizontal dot line indicates raw P-value = 0.05. The dots having a fold-change less than 2 (log2 = 1) and/or P-value <0.05 are shown in gray. Statistical tests were two-sided. d Western blotting analysis of the HK2 expression in NSCLC/HCC-derived NPC and TPC. Hsc70 was used as loading control (n = 3 independent experiments). e The glucose uptake of NSCLC-derived NPC and TPC was measured by using glucose uptake cell-based assay kit (n = 3 experimental repeats). Statistical tests were two-sided. f The lactate production of NSCLC-derived-NPC and-TPC was measured by glycolysis cell-based assay kit (n = 5 experimental repeats). Statistical tests were two-sided. g, h The extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) of NSCLC-derived-NPC and -TPC were measured by a Seahorse flux analyser (n = 5 independent experiments). Dot lines indicate the time and period of glucose, oligomycin (Oligo), 2-Deoxy-D-glucose (2-DG), carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP), rotenone (ROT), or antimycin (Ant) being added during seahorse experiments. i Schematic diagram represents intracellular fluxes in NSCLC-derived TPC relative to NPC quantified using U-13C6 glucose tracer. Open circle represents 12C and filled black circles for 13C atoms. Bar charts represent the relative abundance of metabolites between NPC and TPC (n = 3 experimental repeats). Red arrow represents increased fluxes in TPC as compared with NPC. Black arrows represent no significant change in fluxes between NPC and TPC. Statistical tests were two-sided. j Representative images of the triple immunostaining of CD34 (magenta), HK2 (green), and α–SMA (red) in normal adjacent tissues and tumors derived from NSCLC patients. Bar charts represent the percentage of HK2/α–SMA double positive blood vessels per field in each group (n = 10 NSCLC patients). Statistical tests were two-sided. k Kaplan–Meier survival study showed that NSCLC patients with high percentage of pericyte-HK2 + ve blood vessels were correlated with poor overall survival (n = 77 NSCLC patients). Statistical tests were two-sided. Results are given as means ± SEM. b, c, ej Student’s ttest. k Log-rank (Mantel-Cox) test. Scale bars in j represents 50 μm.

To prove the clinical significance of our observation, we performed an immunofluorescence study with normal adjacent tissues and tumors derived from NSCLC and HCC patients, indicating that the percentage of pericyte-HK2 positive blood vessels was significantly up-regulated in tumors as compared to normal adjacent tissues (Fig. 4j and Supplementary fig. 5d). We next performed triple immunostaining of PDGFRβ, CD34, and α-SMA with our own NSCLC and HCC patient cohorts, indicating that high percentage of pericyte-HK2 positive blood vessels (i.e. when more than the mean number of tumor blood vessels are pericyte-HK2 positive blood vessels across individual patient samples in our NSCLC and HCC cohorts respectively) was associated with poor overall survival in NSCLC and HCC patients respectively (Fig. 4k and Supplementary fig. 5e). Collectively, our findings establish a major metabolic reprogramming of TPC, evidently through up-regulated HK2-driven glycolysis in TPC as compared to NPC.

Pericyte-HK2-driven glycolysis inhibits the blood vessel supporting function of pericyte via ROCK2-MLC2 mediated contractility

Next, the function relevance of glycolytic switch in TPC was examined. We pre-treated TPC with placebo or HK2 inhibitor for 15 min and then co-cultured them with HUVEC cells in tube formation assays. In both NSCLC and HCC, we showed that pre-treatment with HK2 inhibitor enhanced the total tube length, branch points, and number in HUVEC co-cultured with TPC, when compared with HUVEC co-cultured with placebo pre-treated TPC (Fig. 5a–d). To test its effect in vivo, we embedded HUVEC with TPC pre-treated with placebo or HK2 inhibitor 3-bromopyruvate (3-BP) in matrigel-fibrin matrix, which were then implanted into nude mice and given either placebo or HK2 inhibitor treatment before harvesting. Strikingly, treatment with HK2 inhibitor did not affect the capacity of TPC for promoting blood vessel formation, while it enhanced blood vessel diameter and collagen IV expression in HUVEC co-transplanted with TPC as compared to HUVEC co-transplanted with placebo pre-treated TPC (Fig. 5e–h). No significant difference in percentage of coverage between these two groups (Fig. 5i). Importantly, treatment with HK2 inhibitor did not affect blood vessel density, blood vessel diameter, and collagen IV expression in HUVEC transplant alone as compared with placebo treated group (Fig. 5e–h). Further metabolic study showed that pre-treatment with HK2 inhibitor had no effect on glucose uptake in TPC as compared with placebo treated TPC (Supplementary fig. 6a), while it significantly inhibited lactate production in TPC (Supplementary fig. 6b). In addition, pre-treatment with HK2 inhibitor significantly reduced carbachol-induced cell contraction in TPC as compared to placebo pre-treated TPC (Supplementary fig. 6c). These results demonstrated that the elevated HK2 expression in TPC contributed to its defect in cell contractility and blood vessel supporting function, and that was determined by glycolysis.

Fig. 5: Expression of HK2 regulates the blood vessel supporting function of pericyte via ROCK2-MLC2 mediated contractility.
figure5

a Tube formation assay. HUVEC were co-cultured with CFSE labeled TPC pre-treated with either placebo or HK2 inhibitor (3-bromopyruvate, 3-BP). Representative immunofluorescent/bright field merged pictures of the tube formation after 24-h co-culturing in each group are given. bd Bar charts show the relative total tube length, branch points, and number in each group (n = 3 independent experiments). Statistical tests were two-sided. e Representative IHC pictures of the serial sections of matrix plugs after immunostained with CD34, PDGFRβ or collagen IV antibody. fiMatrix plug assay showed that treatment with HK2 inhibitor significantly enhanced the lumen diameter and collagen IV staining of blood vessels in HUVEC co-transplanted with TPC, compared to that treated with placebo. Violin plots show the relative number of blood vessels, lumen diameter, collagen IV level and percentage of coverage in each group (n = 3 mice per group). jWestern blotting analysis of the ROCK2 and p-MLC2 expression in NSCLC/HCC-derived-NPC and -TPC (n = 3 independent experiments). k Administration of HK2 inhibitor 3-BP reduced the expression of ROCK2 and p-MLC2 in NSCLC/HCC-derived TPC (n = 3 independent experiments). loTube formation assay of the HUVEC co-cultured with CFSE labeled NSCLC/HCC-derived TPC pre-treated with 0.4 μM ROCK inhibitor (GSK429286A) or placebo. Bar charts show the relative total tube length, branch points, and number in each group (n = 3 independent experiments). Statistical tests were two-sided. p Western blotting analysis of the HK2, ROCK2, and p-MLC2 expression in NSCLC/HCC-derived NPC after treated with conditioned medium (CM) from either A549 or HepG2 cells (n = 3 independent experiments). qt Tube formation assay of the HUVEC co-cultured with untreated CFSE labeled NSCLC/HCC-derived NPC or NPC pre-treated with conditioned medium (CM) from A549 or HepG2 cells. Bar charts show the relative total tube length, branch points, and number in each group (n = 3 independent experiments). Statistical tests were two-sided. Results are given as means ± SEM. bd, mo, rt Student’s t test. fi One-way ANOVA. Scale bars in a, l, qrepresents 200 μm, e 50 μm.

Consistent with these findings, Rho-associated coiled-coil containing protein kinase (ROCK) signaling pathway has been implicated in regulating pericyte shape and contractility22,23, while these processes require energy supply mainly from glycolysis24. Non-muscle myosin II is a hexameric actin-binding protein that is formed by two heavy chains and two regulatory light chains, which has an important role in cell contractility25. Interestingly, ROCK can directly phosphorylate MLC2 to increase the activity of myosin II during cell contraction26. As expected, western blot analysis indicated that the expression of ROCK2 and phosphorylation of ROCK2 downstream effector MLC2 (Myosin-light-chain 2) was increased in TPC as compared to that in NPC (Fig. 5j). Importantly, administration of HK2 inhibitor significantly down-regulated the expression of ROCK2 and p-MLC2 in TPC (Fig. 5k). To determine the role of ROCK2 in regulating TPC’s BV supporting function, in vitro tube formation assay was performed by co-culturing HUVEC with NSCLC/HCC-derived TPC pre-treated with either placebo or ROCK inhibitor GSK 429286A in matrigel, indicating that pre-treatment with ROCK inhibitor significantly enhanced total tube length, branch point, and number (Fig. 5l–o). Furthermore, pre-treatment with ROCK inhibitor also prohibited carbachol-induced cell contraction in TPC (Supplementary fig. 6c). Western blot analysis indicated that treatment with ROCK inhibitor down-regulated the phosphorylation of MLC2 in TPC as compared with placebo-treated TPC (Supplementary fig. 6d). Clinically, the pericyte expression of ROCK2 was up-regulated in NSCLC/HCC-derived tumors as compared with normal adjacent tissues (Supplementary fig. 6e), while high ratio of ROCK2 to pericyte marker expression correlated with poor survival in NSCLC and HCC patients respectively (Supplementary fig. 6f, g).

Since human ROCKs consist of two isoforms, ROCK1 and ROCK226, we further examine whether HK2 regulates TPC’s blood vessel supporting role via ROCK1 or ROCK2. We first stably knock-downed HK2, ROCK1, or ROCK2 expression in NSCLC/HCC-derived TPC, and also overexpressed ROCK2 in HK2-depleted TPC, which was confirmed by the RT-PCR and western blot analysis (Supplementary fig. 6h–k). Interestingly, depletion of HK2 down-regulated ROCK2, but not ROCK1 expression, in NSCLC/HCC-derived TPC (Supplementary fig. 6h). Further tube formation assays showed that stable knock-down of pericyte-HK2 expression or depleting pericyte-ROCK2 by shRNA/ROCK inhibitor HA-1077 (i.e. a more potent ROCK2 inhibitor27) increased the total tube length, branch points, and number in HUVEC co-cultured with TPC as compared with HUVEC co-cultured with scramble transfected TPC, while silencing ROCK1 had no significant effect on tube formation (Supplementary fig. 6l–s). Furthermore, overexpression of ROCK2 in HK2-depleted TPC reduced the enhanced total tube length, branch points, and number observed in HUVEC co-cultured with HK2-depleted TPC (Supplementary fig. 6l–s). The above results showed that ROCK2, but not ROCK1, was the downstream effector of HK2 mediated pericyte’s blood vessel supporting function.

Accumulating evidence suggest that tumor cells can secrete paracrine factors to affect their surrounding stromal cell metabolism28. To determine whether the difference in metabolic states between NPC and TPC was driven by tumor cells, we exposed NPC to conditioned medium from either NSCLC cell line A549 or HCC cancer cell line HepG2 and then used them for western blotting, glucose uptake, lactate concentration, and tube formation assays. Strikingly, exposure with the conditioned medium (CM) from cancer cells did not affect the glucose uptake in NSCLC/HCC-derived NPC (Supplementary fig. 6t), while it increased the expression of HK2, ROCK2, and p-MLC2 as well as lactate production in NPC as compared with untreated NPC (Fig. 5p and Supplementary fig. 6u). When these CM treated cells co-cultured with HUVEC in a tube formation assay, it significantly decreased total tube length, branch points, and number as compared to HUVEC co-cultured with untreated NPC (Fig. 5q–t).

Recent studies suggested that tumor-derived hyaluronan (HA) fragments up-regulated glycolytic gene expression (including HK2) to modulate the pro-tumor function of monocytes, which could be rescued by using a HA antagonist (Pep-1)29. Interestingly, our flow cytometry analysis indicated that NSCLC/HCC-derived NPC expressed the HA receptor CD44 on their cell surface (Supplementary fig. 6v). We next examine the role of tumor cell-derived HA fragments in regulating pericyte metabolism by exposing NSCLC/HCC-derived NPC with conditioned medium from A549/HepG2 cells in the presence of HA antagonist (Pep-1) or control peptide (Cpep). Strikingly, our results showed that blockade of HA fragments by a HA antagonist (Pep-1) prohibited the up-regulation of HK2 and ROCK2 in NPC after treated with conditioned medium from A549/HepG2 cells as compared with the group treated with control peptides, while it had no effect on ROCK1 expression (Supplementary fig. 6w). To confirm ROCK2 as the downstream target of HK2 mediated glycolysis in TPC, we therefore pre-treated NSCLC/HCC-derived TPC with ROCK inhibitor, which were then used for lactate concentration measurement and seahorse assays. As expected, inhibition of ROCK did not affect the lactate concentration, ECAR and OCR in TPC as compared to placebo treated group (Supplementary fig. 7a–e), while treatment of ROCK inhibitor had no effect on HK2 expression in TPC (Supplementary fig. 7f, g). Overall, these data suggested that tumor-derived paracrine signal induced pericyte-HK2-driven glycolysis to dysregulate pericyte’s blood vessel supporting role by up-regulating ROCK2-MLC2 mediated contractility.

Administration of HK2 inhibitor enhances blood vessel function, doxorubicin delivery, and efficacy against tumor growth, while reducing hypoxia without affecting blood vessel number and coverage

Previous studies from other laboratories and our own indicated that improved tumor blood vessel functions can enhance chemotherapy delivery and efficacy against tumor growth and progression16,30. However, these works mainly focused on targeting endothelial cells, but not on pericytes. We next tested whether combination treatment with HK2 inhibitor and chemotherapeutic agent doxorubicin enhanced its efficacy against lung and liver tumor growth. Strikingly, our results indicated that co-treatment with HK2 inhibitor and doxorubicin significantly inhibited the growth of subcutaneous lung (i.e. LLC, A549) or liver (i.e. mouse HCC cell line Hepa 1-6) tumors as compared to the group treated with either placebo, doxorubicin, or HK2 inhibitor alone (Fig. 6a–d and Supplementary fig. 8a–h). Notably, treatment with HK2 inhibitor or doxorubicin alone showed limited/modest effect on tumor growth as compared to placebo treated group (Fig. 6a–d and Supplementary fig. 8a–h). Analysis of lung tumor vasculature showed that any treatment including 3-BP had no significant effect on total blood vessel number and pericyte coverage as compared with other treatment groups (Fig. 6e–g and Supplementary fig. 8i–k), while it significantly enhanced tumor blood vessel diameter, basement membrane collagen IV staining, and blood vessel perfusion (Fig. 6h–j and Supplementary fig. 8l, m). Moreover, mice injected intravenously with Hoechst dye showed a significant increase in intratumoral Hoechst dye levels, relative to blood vessel number, after treatment with HK2 inhibitor alone, and with a combination of doxorubicin when compared with placebo or single doxorubicin treated mice (Fig. 6k). No significant difference in the body weight of LLC/A549 tumor-bearing mice during treatment was observed (Supplementary fig. 8n, o). Importantly, co-immunostaining analysis confirmed that LLC tumor pericytes expressed HK2 (Supplementary fig. 8p), while the expression of pericyte-p-MLC2 was reduced in LLC/A549 tumors after treated with HK2 inhibitor as compared with placebo (Supplementary fig. 8q). These results demonstrate that the pro-vascular functional effect of HK2 inhibitor was apparent in vivo, and corresponded with improved doxorubicin efficacy.

Fig. 6: Administration of HK2 inhibitor remodels tumor vasculature to enhance blood flow, tumor perfusion, doxorubicin delivery, and efficacy against tumor growth.

a C57/BL6 mice were subcutaneously injected with 1 × 106 LLC cells and treated with placebo, 3-bromopyruvate (3-BP), doxorubicin (DOX), 3-BP, and DOX combination (n = 5 mice per group). Representative gross image of tumors from each treatment group is given. b Tumor growth curves showed that co-treatment of 3-BP and doxorubicin inhibited the growth of LLC tumors as compared with placebo, doxorubicin or 3-BP treated group. c, d Bart charts indicate the tumor weight and volume in each group. ei Midline sections of LLC subcutaneous tumors were double immunostained for CD34 (green) and α-SMA/collagen IV (red) for blood vessel quantification, percentage of pericyte coverage, blood vessel diameter, and collagen IV intensity analysis. j LLC tumor bearing mice were intravenously injected with a FITC-PECAM antibody (green) to detect perfused blood vessels. k Bar charts show the relative area of Hoechst dye uptake into perivascular tumor cells to area of FITC-PECAM in each treatment group. ln Ex vivo two photon microscopic imaging of FITC-PECAM antibody perfused LLC tumors after treated with DOX alone or 3-BP and DOX combination. Bar charts show the quantification of vessel diameter and doxorubicin in each group. Statistical tests were two-sided. op Doppler ultrasonography in live mice provides relative blood flow (peak enhancement, PE) and tumor perfusion (Wash-in perfusion index, WiPI) in subcutaneous LLC tumors. Representative ultrasound images. Bar charts, quantitation across the whole tumors, including tumor cores. Results are given as means ± SEM. b Two-way ANOVA. c, d, fk, p One-way ANOVA. m, n Student’s t test. Scale bars in a, o represents 1 cm, e, j, k, l 50 μm, i, 20 μm.

Since previous studies have shown a strong correlation between blood vessel perfusion/dilation and chemotherapeutic drug delivery30,31, we next sought to examine whether the increase in blood vessel perfusion and dilation was sufficient to improve doxorubicin delivery in our animal models. Ex vivo two photon imaging showed that LLC tumor blood vessel perfusion, vascular diameter, and doxorubicin delivery were all increased in mice treated combination group as compared with doxorubicin alone treated groups (Fig. 6l–n and Supplementary movie 1, 2), implying that HK2 inhibitor treatment induced MLC2-driven tumor vasculature remodeling. Further confocal microscopy analysis confirmed that the intratumor doxorubicin level was increased in mice treated with HK2 inhibitor and doxorubicin when compared to mice treated with doxorubicin alone (Supplementary fig. 8r). Importantly, Doppler ultrasound analysis of microbubble perfusion revealed that treatment with HK2 inhibitor enhanced LLC tumor blood flow and perfusion in mice as compared with the groups treated with either placebo or doxorubicin alone (Fig. 6o, p). Notably, the increase in these dynamic flow and perfusion readouts was observed across the whole tumor, including the tumor core (Fig. 6o, p). Since tumor hypoxia has been linked with poor prognosis in cancer patients30, we next examined the tumor expression of hypoxia indicator glucose transporter (Glut1) in all treatment groups, indicating that the HK2 inhibitor or HK2 inhibitor and doxorubicin combination treatment significantly reduced tumor hypoxia (Supplementary fig. 8s), suggesting that the improved blood vessel function reduced tumor hypoxia in mice treated with HK2 inhibitor. To further examine whether the observed synergistic inhibitory effect of HK2 inhibitor and doxorubicin on tumor growth in vivo was due to direct inhibition on tumor cells or through modulating TPC function, we treated LLC cells with either placebo, 3-BP, doxorubicin or 3-BP, and doxorubicin combination in vitro, indicating that co-administration of 3-BP and doxorubicin had no synergistic inhibitory effect on LLC tumor cell growth in vitro (Supplementary fig. 8t). To determine whether 3-BP or/and doxorubicin could affect tumor cells mediated pericyte’s glycolytic change, we examined the effect of conditioned medium harvested from cancer cells pre-treated with 3-BP or/and doxorubicin on pericyte-HK2 and ROCK2 expression. Both RT-PCR and western blot analysis indicated that the conditioned medium of 3-BP or/and DOX pre-treated tumor cells could still up-regulate HK2 and ROCK2 expression in NSCLC/HCC-derived NPC as compared to untreated NPC group (Supplementary fig. 9a–e). Altogether, these data suggested that the synergetic inhibitory effect of 3-BP and doxorubicin on tumor growth in vivo was through modulating TPC’s blood vessel supporting function.

To confirm the role of pericyte-HK2 in modulating tumor vasculature and subsequent drug delivery, we subcutaneously co-injected A549/MHCC-LM9 (i.e. human HCC cell line) cells with either stable HK2-depleted TPC, stable ROCK2 overexpressing HK2-depleted TPC or scramble transfected TPC at 10:1 ratio into nude mice and then treated the tumor-bearing mice with placebo or doxorubicin. Strikingly, administration of doxorubicin reduced tumor growth in the mice that were co-injected with A549/MHCC-LM9 cells and HK2-depleted cells as compared to the mice that were co-injected with A549/MHCC-LM9 cells and scramble transfected TPC (Supplementary fig. 10a–h), while co-injection of A549/MHCC-LM9 cells with ROCK2 overexpressing HK2-depleted pericytes reduced the enhanced doxorubicin efficacy against tumor growth observed in mice co-injected with A549/MHCC-LM9 cells and HK2-depleted cells (Supplementary fig. 10a–h). No significance difference in total blood vessel number and pericyte coverage was observed between groups (Supplementary fig. 10i–k). Interestingly, the tumor blood vessel diameter, basement membrane collagen IV staining, blood vessel perfusion, and intratumoral Hoechst were all increased in mice that were co-injected with A549 cells and HK2-depleted TPC as compared to mice co-injected with A549 cells and scramble transfected TPC, while the increased blood vessel functions observed were reduced in mice that were co-injected with A549 cells and ROCK2 overexpressing HK2-depleted TPC (Supplementary fig. 10l–o). Furthermore, the expression of pericyte-p-MLC2 was reduced in tumors arising from the mice that were co-injected with A549 cells and HK2-depleted TPC as compared to A549 cells co-injected with scramble transfected TPC, whilst its expression was increased in the tumors derived from co-injection of A549 cells and ROCK2 overexpressing HK2-depleted TPC (Supplementary fig. 10p). Importantly, ex vivo two photon imaging showed that the tumor blood vessel perfusion, vascular diameter, and doxorubicin delivery were all increased in the mice that were co-injected with A549 cells and HK2-depleted TPC as compared with the mice that were co-injected with A549 cells and scramble transfected TPC, while the increased blood vessel perfusion, vascular diameter, and doxorubicin delivery observed was decreased in mice that were co-injected with A549 cells and ROCK2 overexpressing HK2-depleted TPC (Supplementary fig. 10q and Supplementary movie 35). Doppler ultrasound analysis of microbubble perfusion revealed that depletion of HK2 in TPC enhanced A549 tumor blood flow and perfusion in mice as compared with scramble control group (Supplementary fig. 10r), while overexpression of ROCK2 in HK2-depleted TPC reduced the enhanced tumor blood flow and perfusion observed (Supplementary fig. 10r). These results demonstrated the important role of pericyte-HK2-driven ROCK2-MLC2 mediated contractility in tumor vasculature remodeling and drug delivery.

Overall, our findings show that treatment with HK2 inhibitor or depletion of pericyte-HK2 by shRNA can enhance blood flow and perfusion leading to increased intratumoral drug delivery and efficacy against tumor growth, whilst decreasing tumor hypoxia without affecting blood vessel density and pericyte coverage (Fig. 7). These results indicate that targeting pericyte-HK2 expression/activity could be an effective treatment strategy for cancer patients.

Fig. 7: Schematic diagram represents the role of pericyte-HK2-driven glycolysis in regulating tumor blood vessel abnormality via ROCK2-MLC2 mediated contractility.
figure7

TPC up-regulates HK2-driven glycolysis to induce ROCK2-MLC2 mediated contractility, which in turn negatively regulates its blood vessel supporting role. Administration of HK2 inhibitor 3-BP induces MLC2-driven tumor vasculature remodeling to increase blood flow, tumor perfusion, blood vessel diameter as well as chemotherapeutic drug delivery and efficacy against tumor growth, while reducing tumor hypoxia without affecting blood vessel density and pericyte coverage.

Discussion

Our method successfully purified a population of pericytes from normal adjacent tissues and tumors derived from NSCLC or HCC patients. We showed an elevated HK2-driven glycolysis in TPC as compared with NPC, which contributed to its abnormal blood vessel supporting function by activating ROCK2-MLC2 mediated contractility. Depletion of HK2/ROCK2 by shRNA or inhibitors reduces TPC contractility to remodel vasculature, while combined treatment of HK2 inhibitor and doxorubicin enhanced the efficacy of chemotherapy drug against tumor growth in both lung and liver cancer animal models. This is an unexpected result and call for a reconsideration of our current strategy targeting pericytes for cancer treatment.

Pericytes, also known as contractile cells, play an important role to modulate angiogenesis, blood vessel diameter, blood flow, vascular perfusion, and integrity mainly via their contractile properties1,7,10,32. Interaction between pericyte-endothelial cell in the blood vessel wall is a fundamental process in regulating vascularization, vasoconstriction, vasodilation, and stabilization9,14, while defective interaction between two cell types causes human pathological conditions, including dementia syndrome and tumor angiogenesis1,12. Indeed, genetic knock-out of PDGFB or PDGFRβ both causes vascular dysfunction and perinatal death5. Due to the lack of an effective human pericyte isolation protocol, most of the in vitro pericyte studies were performed by using either bovine/mouse retinal pericytes or mouse brain pericytes4,18,33. Although pericytes can potentially differentiate into other cell types34, the retinal pericytes were neither isolated by FACS, nor fully characterized by pericyte marker expression or cultured in medium supplemented with pericyte growth factors in previous studies32,35. Furthermore, it remains debating about the mesenchymal stem cell property of pericytes in vivo36. Contrary to these studies, we developed a MACS method to enrich vascular cell population from the dissociated tissues and then subjected the cell population for subsequent FACS to purify pericytes, which were then cultured in commercially available medium dedicated for human pericytes. By performing a series of characterization experiments, we confirmed the pericyte identity of our isolated cells derived from NSCLC and HCC, while there was no significant difference in basic function between NPC and TPC. In addition, our culturing method can maintain the pericyte identity of the isolated cells over passages.

We next sought to explore their roles in regulating vasculature in vitro and in vivo. Previously, Mcllroy et al.37, show that exposure of immortalized HUVEC with bovine retina pericyte conditioned medium can inhibit its migration as compared with HUVEC exposed with serum-free medium or endothelial cell conditioned medium, while it also repressed the tube formation ability of HUVEC. On the other hand, Kumar et al.38 found that mesenchymoangioblast-derived pericytes promoted blood vessel formation and stability in vitro and in vivo. The difference amongst these studies may be due to different sources/species of pericytes being used in the tube formation assays. Nevertheless, our data indicated that TPC exerted a stronger repressive effect on the total tube length, branch points, and number as compared with NPC, while in vivo co-transplantation of HUVEC with TPC formed smaller and irregular shaped blood vessels with poor basement membrane structure as compared with HUVEC co-transplanted with NPC. In addition, our cell contraction assays showed that NSCLC/HCC-derived TPC displayed a stronger carbachol-induced cell contraction as compared with NPC, fibroblast, or HUVEC. Overall, our study showed that TPC had defective blood vessel supporting functions as compared with NPC.

Although the contribution of specific metabolic programs to the regulation of cell state/functions have been extensively studied in endothelial cells, the role of metabolic state in regulating pericyte function remains unclear. Transcriptional profiling study showed a consistent increase in glycolytic gene hexokinase 2 expression in NSCLC/HCC-derived TPC as compared with NPC. Since HK2 plays a key role in catalyzing the first step of the glycolytic pathway where glucose is phosphorylated to glucose-6-phosphate39, we demonstrated that the lactate production was up-regulated in TPC as compared with NPC, while seahorse assays showed an increase in ECAR but decreased OCR in TPC as compared that to NPC. Importantly, metabolic flux analysis indicated the relative abundance of lactate, but not the TCA cycle metabolites, was increased in TPC as compared to NPC. These findings were supported by the clinical observation that the pericyte expression of HK2 was up-regulated in tumors as compared with normal adjacent tissues, while high percentage of pericyte-HK2 positive blood vessels correlated with poor overall survival in NSCLC and HCC patients respectively. We next explored the role of pericyte-HK2 in regulating its blood vessel supporting function. Our results showed that pre-treatment with HK2 inhibitor enhanced the total tube length, branch points, and number in HUVEC co-cultured with TPC in a tube formation assay as compared with HUVEC co-cultured with placebo pre-treated TPC, while in vivo co-transplantation of HUVEC with HK2 inhibitor pre-treated TPC enhanced blood vessel diameter and collagen IV expression as compared with HUVEC co-transplanted with placebo-treated TPC. In addition, pre-treatment with HK2 inhibitor reduced carbachol-induced cell contraction in TPC as compared to placebo pre-treated TPC. Notably, treatment with HK2 inhibitor did not affect the tube formation ability of HUVEC on its own as well as the number of blood vessels in subcutaneous LLC and A549 tumors as compared with placebo treated group, suggesting that it has a specific role in pericyte’s blood vessel supporting function. Additionally, our results highlight the potential role of pericyte contractility in regulating blood vessel function. Indeed, previous studies indicate that RhoA-ROCK (Rho-associated protein kinase) signaling pathway has an important role in regulating retinal pericyte contractility22,23, while Kutcher et al. show that pericyte contractility determines endothelial cell cycle progression and sprouting22. He et al.40 has reported that targeting α-SMA+ pericyte contractility could normalize tumor vasculature in a spontaneous pancreatic cancer mouse model. In this study, we observed an up-regulation of ROCK2 expression and its downstream effector MLC2 phosphorylation in TPC as compared with NPC. while administration of HK2 inhibitor reduced ROCK2 and p-MLC2 expression in TPC. In addition, pre-treatment of ROCK inhibitor also decreased a carbachol-induced cell contraction in TPC as compared to placebo pre-treated TPC. Importantly, depleting HK2 or ROCK2, but not ROCK1, enhanced the total tube length, branch point and numbers in HUVEC co-cultured with TPC in tube formation assays as compared with HUVEC co-cultured with scramble transfected TPC, while overexpression of ROCK2 in HK2-depleted TPC reduced the enhanced effect observed. These results suggest that pericyte-HK2-driven glycolysis contributes to vascular dysfunction via activation of the ROCK2-MLC2 mediated contractility.

We next sought to explore the molecular control of HK2 expression in TPC. Since previous work showed that cancer cells can affect surrounding stromal cell metabolism via paracrine signaling28, we therefore exposed NPC with conditioned medium from cancer cells, indicating that exposure of NPC with conditioned medium from cancer cells up-regulated HK2 and ROCK2 expression as compared to untreated NPC. Indeed, recent studies showed that tumor-derived HA fragments can activate glycolytic gene expression, including HK2, in monocytes29. Since NSCLC/HCC-derived NPC was shown to express the HA receptor CD44 in this study, we therefore examined its role in pericyte metabolic reprogramming, indicating that blockade of HA fragments prohibited the up-regulation of HK2 and ROCK2 expression in NPC after exposure with conditioned medium from cancer cells. Overall, our results show that tumor-derived paracrine signal up-regulates pericyte-HK2-driven glycolysis to dysregulate its blood vessel supporting role by up-regulating ROCK2 mediated pathway.

Cancer cells rely on glycolysis to meet their energy demand, which is known as the Warburg effect. Since HK2 is a key mediator of glycolysis, its inhibitor, 3-bromopyruvate, has been extensively tested as a single anticancer agent for treating preclinical tumor models41,42. Unfortunately, it has so far provided modest treatment outcomes in different types of cancers43. Indeed, we showed that treatment with HK2 inhibitor 3-bromopyruvate alone showed modest effect on subcutaneous lung or liver tumor growth, while co-administration of HK2 inhibitor and doxorubicin significantly inhibited the growth of the tumors by inducing MLC2-driven vasculature remodeling, while reducing tumor hypoxia without affecting tumor blood vessel density and pericyte coverage. By performing ex vivo two photon microscopy and Doppler ultrasound imaging, we proved that administration of 3-BP induced tumor vasculature remodeling to enhance blood flow, perfusion, and chemotherapy delivery. Interestingly, treatment with 3-BP and doxorubicin showed no synergistic inhibitory effect on LLC cell growth in vitro, while exposure of NPC with conditioned medium from 3-BP or/and doxorubicin pre-treated tumor cells could still up-regulate HK2 and ROCK2 expression as compared with untreated NPC. Together with the fact that the efficacy of chemotherapy delivery has been shown to be linked with tumor blood vessel perfusion and diameter31, these findings suggest that the synergistic repressive effect of 3-BP inhibitor and doxorubicin on tumor growth in vivo relies on modulating pericyte’s BV supporting role, but not through tumor cell inhibition. Additionally, the role of pericyte-HK2 in regulating MLC-2-driven tumor vasculature remodeling and drug delivery was further illustrated by the co-injection experiments, showing that depletion of pericyte-HK2 expression also enhanced blood vessel perfusion and lumen diameter, which in turn increased chemotherapeutic drug delivery and efficacy against tumor growth, while overexpression of ROCK2 in HK2-depleted TPC reduced the enhanced blood vessel functions and drug delivery and efficacy against tumor growth observed in the mice that were co-injected with HK2-depleted TPC and cancer cells. Furthermore, triple immunostaining experiments indicated that the expression of pericyte-p-MLC2 was decreased in tumors arising from the mice that were co-injected with A549 and HK2-depleted TPC as compared to A549 co-injected with scramble transfected TPC, whilst its expression was enhanced in the tumor-derived from co-injection of A549 cells and ROCK2 overexpressing HK2-depleted TPC. Together, these results indicate that targeting pericyte-HK2-ROCK2-MLC2 mediated contractility improves blood vessel function and drug delivery and efficacy against lung and liver cancer growth.

Overall, our work develops a cellular model for studying human pericyte biology and shows that pericyte-HK2-driven glycolysis induces tumor blood vessel abnormality by activating ROCK2-MLC2 mediated contractility. Importantly, our data demonstrate, using strategy distinct from anti-angiogenesis or targeting pericyte coverage, that a radical approach of modulating the metabolic status of pericytes to remodel tumor vasculature for enhancing chemotherapy efficacy in cancer treatment.


Article classification: Biological abstract
Share to:
Tel:+86-027-87610298
Tel:+86-027-87610297
Add:Room A11-329, 1st Floor, No.1, SBI Venture Street, Optics Valley, East Lake
New Technology Development Zone, Wuhan, China.
Certificate NO.:U18Q28010569R0S