Author response: GPIHBP1 expression in gliomas promotes utilization of lipoprotein-derived nutrients Article Swipe

Article Figures and data Abstract Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract GPIHBP1, a GPI-anchored protein of capillary endothelial cells, binds lipoprotein lipase (LPL) within the subendothelial spaces and shuttles it to the capillary lumen. GPIHBP1-bound LPL is essential for the margination of triglyceride-rich lipoproteins (TRLs) along capillaries, allowing the lipolytic processing of TRLs to proceed. In peripheral tissues, the intravascular processing of TRLs by the GPIHBP1–LPL complex is crucial for the generation of lipid nutrients for adjacent parenchymal cells. GPIHBP1 is absent from the capillaries of the brain, which uses glucose for fuel; however, GPIHBP1 is expressed in the capillaries of mouse and human gliomas. Importantly, the GPIHBP1 in glioma capillaries captures locally produced LPL. We use NanoSIMS imaging to show that TRLs marginate along glioma capillaries and that there is uptake of TRL-derived lipid nutrients by surrounding glioma cells. Thus, GPIHBP1 expression in gliomas facilitates TRL processing and provides a source of lipid nutrients for glioma cells. https://doi.org/10.7554/eLife.47178.001 Introduction GPIHBP1, a GPI-anchored protein of capillary endothelial cells, is required for lipoprotein lipase (LPL)–mediated processing of triglyceride-rich lipoproteins (TRLs) (Beigneux et al., 2007). The principal function of GPIHBP1 is to capture LPL within the interstitial spaces, where it is secreted by parenchymal cells, and then to shuttle this enzyme to the luminal surface of capillary endothelial cells (Davies et al., 2010). GPIHBP1 is a long-lived protein (Young et al., 2011; Olafsen et al., 2010) that moves bidirectionally across endothelial cells, with each trip to the abluminal plasma membrane representing an opportunity to capture LPL and bring it to the capillary lumen (Davies et al., 2012). When GPIHBP1 is absent or defective, LPL is stranded within the interstitial spaces, where it remains bound to sulfated proteoglycans near the surface of cells (Young et al., 2011; Davies et al., 2010; Allan et al., 2017a; Fong et al., 2016). The inability of LPL to reach the capillary lumen in the absence of GPIHBP1 expression profoundly impairs TRL processing, resulting in severe hypertriglyceridemia (chylomicronemia) (Beigneux et al., 2007; Davies et al., 2010; Goulbourne et al., 2014). GPIHBP1 is expressed in the capillary endothelial cells of peripheral tissues, with particularly high levels of expression in heart and brown adipose tissue (Beigneux et al., 2007; Davies et al., 2010; Fong et al., 2016). Most of the LPL within those tissues is bound to GPIHBP1 on capillaries (Beigneux et al., 2007; Davies et al., 2010; Davies et al., 2012; Allan et al., 2017a; Fong et al., 2016; Allan et al., 2017b; Allan et al., 2016), and the processing of TRLs in these tissues is robust, generating fatty acid nutrients for nearby parenchymal cells (Fong et al., 2016; Jiang et al., 2014a; He et al., 2018a). By contrast, GPIHBP1 is absent from capillaries of the brain (Young et al., 2011; Davies et al., 2010; Olafsen et al., 2010), a tissue that depends on glucose for fuel (Mergenthaler et al., 2013). When wild-type mice are injected intravenously with a GPIHBP1-specific antibody, the antibody rapidly binds to GPIHBP1-expressing capillaries in peripheral tissues and disappears from the plasma (Davies et al., 2010; Olafsen et al., 2010). By contrast, there is no antibody binding to the capillaries of the brain (Davies et al., 2010; Olafsen et al., 2010). For the lipolytic processing of TRLs to proceed, lipoproteins in the bloodstream must marginate along the luminal surface of capillaries (Goulbourne et al., 2014). TRL margination along capillaries depends on GPIHBP1, more specifically on GPIHBP1-bound LPL (Goulbourne et al., 2014). In GPIHBP1-deficient mice, TRLs never stop along heart capillaries and instead simply 'flow on by' in the bloodstream (Goulbourne et al., 2014). In wild-type mice, TRLs marginate along heart capillaries, but TRL margination is absent along capillaries of the brain (Goulbourne et al., 2014). Even though GPIHBP1 is not found in brain capillaries, there is ample evidence for LPL expression within the brain (Ben-Zeev et al., 1990; Bessesen et al., 1993; Goldberg et al., 1989; Vilaró et al., 1990; Yacoub et al., 1990; Eckel and Robbins, 1984). Several groups have found LPL in the rat brain, specifically in neurons of the dentate gyrus and hippocampus, in pyramidal cells of the cortex, and in Purkinje cells of the cerebellum (Ben-Zeev et al., 1990; Bessesen et al., 1993; Goldberg et al., 1989; Vilaró et al., 1990; Eckel and Robbins, 1984). Using single-cell RNA sequencing, Zhang et al. (2014) found Lpl transcripts in the resident macrophages of the brain (microglia), with lower levels in astrocytes, neurons, and oligodendrocytes. Using the same approach, Vanlandewijck et al. (2018) found LPL expression in brain smooth muscle cells and in perivascular fibroblasts (at even higher levels than in microglial cells). Given the absence of GPIHBP1 expression in brain capillaries and the absence of TRL margination along brain capillaries, we have proposed that the LPL in the brain probably has an extravascular function, presumably to hydrolyze glycerolipids within the extracellular spaces (Young et al., 2011; Adeyo et al., 2012). Despite the absence of GPIHBP1 expression in brain capillaries, we were curious about whether GPIHBP1 might be expressed in the capillaries of gliomas. Glioma capillaries are morphologically distinct from normal brain capillaries (Yuan et al., 1994; Hobbs et al., 1998; Monsky et al., 1999; Bullitt et al., 2005), and the blood–brain barrier is often defective (Zhang et al., 1992). Electron microscopy has suggested that glioblastoma capillaries resemble capillaries in peripheral tissues (Vaz et al., 1996). If GPIHBP1 were to be expressed in glioma capillaries, it could be relevant to glioma metabolism. The GPIHBP1 might capture locally produced LPL, allowing for TRL margination and TRL processing, and thereby providing a source of lipid nutrients for glioma cells. Interestingly, Dong et al. (2017) documented LPL expression in gliomas. Also, several studies have raised the possibility that glioma cells use fatty acids for fuel (Lin et al., 2017; Guo et al., 2011; Guo et al., 2009a; Guo et al., 2009b; Guo et al., 2013) and that levels of free fatty acids are higher in gliomas than in normal brain tissue (Guo et al., 2013; Gopal et al., 1963). In the current study, we sought to determine whether glioma capillaries express GPIHBP1 and, if so, whether it binds LPL and facilitates TRL margination and the lipolytic processing of TRLs. In our study, we took advantage of NanoSIMS imaging, a high-resolution mass spectrometry–based imaging modality that makes it possible to visualize TRL margination and TRL processing in tissue sections (He et al., 2018a; Jiang et al., 2014a; Jiang et al., 2014b; He et al., 2017a; He et al., 2017b; He et al., 2018b; He et al., 2018c). This imaging modality allowed us to visualize TRL margination in glioma capillaries as well as the entry of TRL-derived nutrients into tumor cells. Results GPIHBP1 is expressed in the endothelial cells of human gliomas We sectioned 20 human gliomas (Table 1) and screened them for GPIHBP1 expression by confocal microscopy with three GPIHBP1-specific monoclonal antibodies (mAbs) — RF4, which binds to residues 27–44 downstream from GPIHBP1's acidic domain (Kristensen et al., 2018); and RE3 and RG3, which both bind to GPIHBP1's LU (Ly6/uPAR) domain (Hu et al., 2017). GPIHBP1 in capillary endothelial cells was detected in 14 of 20 gliomas (Table 1) and colocalized with von Willebrand factor, an endothelial cell marker (Figure 1). GPIHBP1 expression in glioma capillaries did not appear to correlate with glioma grade, 1p/19q co-deletions, or IDH1 mutations (Table 1). GPIHBP1 was not detectable in the capillaries of human brain specimens (Figure 1). The GPIHBP1 in glioma capillaries could be detected with all three GPIHBP1-specific mAbs (Figure 2A). To be confident in the specificity of the antibodies, we performed studies in which recombinant human GPIHBP1 was added to the GPIHBP1-specific mAbs before incubating the solution with the glioma sections. As expected, the presence of recombinant GPIHBP1 eliminated binding of the GPIHBP1-specific mAbs to glioma capillaries (Figure 2B). GPIHBP1 expression in glioma capillaries could also be detected by immunoperoxidase staining (Figure 1—figure supplement 1). Figure 1 with 1 supplement see all Download asset Open asset GPIHBP1 expression in the endothelial cells of several human gliomas. Immunohistochemical studies on surgically resected gliomas (Gliomas 1, 5, 9; Table 1) and non-diseased human frontal lobe (n = 3), revealing GPIHBP1 expression in capillaries of gliomas but not in frontal lobe specimens. GPIHBP1 (detected with a combination of the mAbs RE3 and RF4, 10 μg/ml each [red]) colocalized with von Willebrand factor (vWF, a marker for endothelial cells [green]), but not with glial fibrillary acidic protein (GFAP, a marker for astroglial cells [magenta]). DNA was stained with DAPI (blue). Three sections of each tumor and normal brain were evaluated and representative images are shown. Scale bar, 50 μm. https://doi.org/10.7554/eLife.47178.002 Figure 2 Download asset Open asset Detecting GPIHBP1 in capillaries of human glioma specimens with three different monoclonal antibodies (mAbs) against GPIHBP1. (A) Confocal fluorescence microscopy studies on sections from glioma sample 1 (Table 1), demonstrating the detection of GPIHBP1 with three different human GPIHBP1–specific monoclonal antibodies (mAbs). Tissue sections were fixed with 3% PFA and then stained with mAbs against human GPIHBP1 (RF4, RE3, or RG3, 10 μg/ml [red]), an antibody against von Willebrand factor (vWF[green]), and an antibody against glial fibrillary acidic protein (GFAP [magenta]). All three GPIHBP1-specific mAbs detected GPIHBP1 in capillaries, colocalizing with von Willebrand factor. DNA was stained with DAPI (blue). Scale bar, 50 μm. (B) Immunofluorescence confocal microscopy studies on human glioma sample 5, performed with mAbs RF4 and RE3 (10 μg/ml) in the presence or absence of 50 μg of recombinant soluble human GPIHBP1 (hGPIHBP1). Adding recombinant hGPIHBP1 to the antibody incubation abolished binding of the GPIHBP1-specific mAbs to GPIHBP1 on glioma capillaries. Images show GPIHBP1 (red), vWF (green), GFAP (magenta), and DAPI (blue). Three sections of tumors were evaluated; representative images are shown. Scale bar, 50 μm. https://doi.org/10.7554/eLife.47178.004 Table 1 Human glioma tumor specimens. Expression of GPIHBP1 was assessed by immunohistochemistry with mAbs against human GPIHBP1 (RF4, RE3, RG3). Those conducting the studies were blinded to diagnoses. This table details the tumor diagnosis, location, 1p/19q co-deletion, and IDH1 mutation status, as well as the presence of GPIHBP1. https://doi.org/10.7554/eLife.47178.005 Sample IDTissue diagnosisLocation1p/19q co-deletionIDH1 mutationGPIHBP11Glioblastoma (GBM)Right frontal, parietalNoNegativeYes2GBMLeft temporalNoNegativeYes3GBMRight occipitalNoNegativeYes4GBMLeft frontalNoNegativeYes5Oligodendroglioma Grade IILeft anterior temporal, left posterior temporalYesNegativeYes6Oligoastrocytoma Grade IIIRight temporalNoNegativeYes7GBM + oligodendroglial componentLeft frontalYesNegativeYes8GBM + extensive oligodendroglial componentRight frontalNoNegativeYes9Oligodendroglioma Grade IIILeft frontalYes+R132HYes10Oligodendroglioma Grade IIILeft frontalYes+R132HYes11OligoastrocytomaRight parietalNoNegativeYes12Oligodendroglioma Grade IIIRight parietalYes+R132HYes13Oligodendroglioma Grade IIIRight parietalYesNegativeYes14Oligoastrocytoma Grade IIILeft temporalNo+R132HYes15Oligoastrocytoma Grade IIIRight temporalNo+R132GNo16Oligoastrocytoma Grade IIIRight frontalNo+R132HNo17Oligodendroglioma Grade IIILeft frontalYesNegativeNo18Oligodendroglioma Grade IIILeft frontalYes+R132HNo19Oligodendroglioma Grade IIILeft temporalYesNegativeNo20Oligodendroglioma Grade IIIRight temporalYes+R132HNo GPIHBP1 is present in the capillary endothelial cells of mouse gliomas To determine whether GPIHBP1 is expressed in a mouse model of glioblastoma, spheroids of syngeneic C57BL/6 mouse CT-2A glioma cells (Seyfried et al., 1992; Oh et al., 2014), modified to express a blue fluorescent protein (BFP) (Mathivet et al., 2017), were engrafted into the brains of mice harboring an endothelial cell–specific Pdgfb-iCreERT2 transgene (Claxton et al., 2008) and a ROSAmT/mG reporter allele (Muzumdar et al., 2007). ROSAmT/mG is a two-color fluorescent, membrane-targeted Cre-dependent reporter allele. In the absence of Cre, all cells express a membrane-localized tdTomato and fluoresce red. In the setting of Cre expression, cells express membrane-localized EGFP (rather than tdTomato) and fluoresce green. Before tumor implantation, mice were injected with tamoxifen to induce Pdgfb-driven Cre expression in endothelial cells; thus, the endothelial cells of the mice expressed EGFP and fluoresced green. Mice harboring gliomas (after three weeks of growth) were injected intravenously with an Alexa Fluor 647–conjugated antibody against mouse GPIHBP1 (11A12) (Beigneux et al., 2009). Mice were perfused with PBS and then perfusion-fixed with 2% PFA, and tumor sections were processed for confocal immunofluorescence microscopy. GPIHBP1 was detected in endothelial cells of the gliomas, colocalizing with EGFP (brain endothelial cells), but GPIHBP1 was absent from capillaries in the adjacent normal brain (Figure 3, Figure 3—figure supplement 1). Using transmission electron microscopy, we observed large and irregularly shaped capillaries in gliomas, with numerous villus-like structures on the luminal surface of endothelial cells (Figure 3—figure supplement 2), similar to findings reported for capillaries in human gliomas (Vaz et al., 1996; Coomber et al., 1987; Weller et al., 1977). Figure 3 with 3 supplements see all Download asset Open asset GPIHBP1 is expressed by capillary endothelial cells in mouse gliomas. Confocal microscopy images of a BFP-tagged CT-2A glioma implanted in a ROSAmT/mG::Pdgfb-iCreERT2 mouse, revealing the expression of GPIHBP1 in capillary endothelial cells of the glioma but not those of normal brain. Tamoxifen was administered prior to implantation of the glioma spheroid to activate membrane-targeted EGFP in endothelial cells (green). After three weeks of glioma growth, mice were anesthetized and injected via the tail vein with an Alexa Fluor 647–labeled antibody against mouse GPIHBP1 (11A12; red). The mice were then perfused with PBS and perfusion-fixed with 2% PFA in PBS. Glioma and adjacent normal brain were harvested, and 200-μm-thick sections were imaged by confocal microscopy. GPIHBP1 was present on endothelial cells of the glioma (blue) but was absent from normal brain. High-magnification images of the boxed area are shown on the right. Three mice were evaluated; representative images are shown. Scale bar, 50 μm. https://doi.org/10.7554/eLife.47178.006 The factors that regulate Gpihbp1 expression in the capillary endothelial cells of peripheral tissues and gliomas are incompletely understood. However, a recent study found that Gpihbp1 transcript levels in rat aortic endothelial cells are upregulated by vascular endothelial growth factor (VEGF) (Chiu et al., 2016), an angiogenic factor known to be expressed at high levels by glioma cells (Plate et al., 1994; Pietsch et al., 1997; Christov et al., 1998). We found that Gpihbp1 expression in the mouse brain endothelial cell line bEnd.3 is upregulated by recombinant VEGF (Figure 3—figure supplement 3). GLUT1 is expressed in the capillaries of gliomas and normal brain We used immunofluorescence microscopy to examine the expression of GPIHBP1 and GLUT1 (the main glucose transporter in brain capillaries [Maher et al., 1994; Pardridge et al., 1990]) in mouse gliomas and adjacent normal brain. GPIHBP1 expression was detected in gliomas but was absent in the normal brain. The signal for GLUT1 was strong in the endothelial cells of the normal brain and was easily detectable in the capillaries of gliomas (Figure 4, Figure 4—figure supplements 1–2). Consistent findings were observed in single-cell RNA-seq studies on vascular cells of gliomas (Ken Matsumoto, manuscript in preparation) and normal brain vascular cells (Vanlandewijck et al., 2018; He et al., 2018d). Endothelial cells of gliomas (identifed by high von Willebrand factor [vWF] expression) exhibit high expression of Gpihbp1 and somewhat lower levels of Glut1 expression (e.g., Endothelial cell cluster 5 in Figure 4—figure supplement 3). In normal brain, Glut1 was highly expressed in endothelial cells, whereas Gpihbp1 expression was absent (Figure 4—figure supplement 3). In Gpihbp1-deficient mice, GLUT1 expression was detectable in the capillaries of gliomas and normal brain (Figure 4—figure supplement 4). Figure 4 with 4 supplements see all Download asset Open asset Expression of GPIHBP1 and GLUT1 in the endothelial cells of mouse gliomas. Immunohistochemical studies of a BFP-expressing CT-2A glioma (after three weeks of growth). Mice were injected via the tail vein with an Alexa Fluor 647–labeled antibody against mouse GPIHBP1 (11A12; green), then perfused with PBS and perfusion-fixed with 2% PFA. Glioma and adjacent normal brain tissue were harvested, then 200-μm thick sections cut, fixed with 4% PFA, and stained with an antibody against GLUT1 (red). GPIHBP1 was present in the capillaries of mouse gliomas (blue) but absent from the capillaries of the normal brain. High-magnification images in the boxed region are shown below. Three mice were evaluated; representative images are shown. Scale bar, 50 μm. https://doi.org/10.7554/eLife.47178.010 LPL is present on GPIHBP1-expressing capillaries of mouse gliomas Most of the LPL in peripheral tissues (e.g., heart or brown adipose tissue) is bound to GPIHBP1 on capillaries; consequently, LPL and GPIHBP1 colocalize in tissue sections (Young et al., 2011; Davies et al., 2010; Davies et al., 2012; Allan et al., 2017a; Fong et al., 2016; Allan et al., 2017b; Allan et al., 2016). We hypothesized that GPIHBP1-expressing endothelial cells of gliomas could capture LPL. Several observations prompted us to consider this hypothesis. First, as noted earlier, there is ample evidence for LPL expression in the brain (Ben-Zeev et al., 1990; Bessesen et al., 1993; Goldberg et al., 1989; Vilaró et al., 1990; Yacoub et al., 1990; Zhang et al., 2014), and it seemed reasonable that some of that LPL would reach high-affinity GPIHBP1-binding sites on endothelial cells. Second, gliomas contain large numbers of macrophages (F4/80-expressing cells; Figure 5—figure supplement 1), and macrophages are known to express LPL (Mahoney et al., 1982). We found that LPL could be detected in peritoneal macrophages from wild-type mice but not in macrophages harvested from Lpl–/– mice carrying a skeletal muscle–specific human LPL transgene (Lpl–/–MCK-hLPL) (Levak-Frank et al., 1995) (Figure 5—figure supplement 2). We also found that LPL could be detected in some of the macrophages in mouse gliomas and in normal brain of wild-type mice, but not in the brain of Lpl–/–MCK-hLPL mice (Figure 5—figure supplement 3). These findings were consistent with single-cell RNA-seq data from glioma and normal brain, in which Lpl transcripts were found in the macrophages of gliomas and microglia of normal brain (Figure 4—figure supplement 3). Lpl transcripts are not present in capillary endothelial cells. Third, the most highly upregulated fatty acid metabolism gene in human gliomas, compared to normal brain tissue, is LPL (Figure 5—figure supplement 4). The second most perturbed gene in gliomas is CD36, which encodes a putative fatty acid transporter (Figure 5—figure supplement 4). To determine whether LPL is bound to GPIHBP1-expressing capillaries of gliomas, we performed immunohistochemical studies, taking advantage of an affinity-purified goat antibody against mouse LPL (Page et al., 2006). These studies revealed colocalization of GPIHBP1 and LPL in glioma capillaries (Figure 5, Figure 5—figure supplement 5). LPL was not present in the capillaries of the normal brain or in the capillaries of gliomas from Gpihbp1–/– mice (Figure 5, Figure 5—figure supplement 5). As expected, the binding of the goat LPL antibody to tissues of Lpl–/–MCK-hLPL mice was low (Figure 5, Figure 5—figure supplement 5), whereas mouse LPL was easily detectable in the heart capillaries of wild-type mice (colocalizing with GPIHBP1) (Figure 5—figure supplement 6). Consistent with earlier publications (Ben-Zeev et al., 1990; Vilaró et al., 1990), we observed a strong mouse LPL signal in the hippocampal neurons of wild-type mice but not of Lpl–/–MCK-hLPL mice (Figure 5—figure supplement 7). Of note, LPL was undetectable in 'secondary antibody–only' experiments (i.e., when the incubation of the primary antibody with tissue sections was omitted) (Figure 5, Figure 5—figure supplement 5–7). Figure 5 with 8 supplements see all Download asset Open asset Lipoprotein lipase (LPL) colocalizes with GPIHBP1 in glioma capillaries. Confocal immunofluorescence microscopy studies on glioma and normal brain from wild-type and Gpihbp1–/– mice, along with the brain from an Lpl–/– mouse carrying a skeletal muscle–specific human LPL transgene (MCK). Glioma and brain sections (10-μm-thick) were fixed with 3% PFA and then stained with a mAb against mouse GPIHBP1 (11A12; green), a goat antibody against mouse LPL (red), and a rabbit antibody against CD31 (white). LPL colocalizes with GPIHBP1 and CD31 in the capillaries of gliomas; GPIHBP1 and LPL were absent from normal brain capillaries and from glioma capillaries in Gpihbp1–/– mice. DNA was stained with DAPI (blue). No LPL was detected in the capillaries of Lpl-deficient mice (MCK) or when the incubation with primary antibodies was omitted (Secondary Only). Staining of all tissue sections was performed simultaneously, and all images were recorded with identical microscopy settings. Three mice per genotype were evaluated; representative images are shown. Scale bar, 50 μm. https://doi.org/10.7554/eLife.47178.015 There is little reason to suspect that the expression of LPL influences the expression of GPIHBP1 in capillaries. The overexpression of human LPL in the skeletal muscle of Lpl–/–MCK-hLPL mice did not alter levels of Gpihbp1 expression (Figure 5—figure supplement 8). Margination of TRLs along glioma capillaries and uptake of TRL-derived nutrients in glioma cells Given the presence of GPIHBP1-bound LPL on glioma capillaries, we suspected that we might find evidence of TRL margination and processing in gliomas. To test this idea, TRLs that were heavily labeled with deuterated lipids ([2H]TRLs) (He et al., 2018a) were injected intravenously into mice harboring CT-2A gliomas (after three weeks of glioma growth). After allowing the [2H]TRLs to circulate for either 1 min or 30 min, the mice were euthanized, extensively perfused with PBS, and perfusion-fixed with carbodiimide/glutaraldehyde. Heart, brain, and glioma specimens were harvested and processed for NanoSIMS imaging. 12C14N– or 1H– images were used to visualize tissue morphology, and 2H/1H images were used to identify regions of 2H enrichment. The scale in the 2H/1H images of brain and glioma specimens ranges from 0.00018 to 0.0003 (i.e., from levels slightly above 2H natural abundance to levels twice as high as 2H natural abundance). The scale in the heart 2H/1H images ranges from 0.00018 to 0.0006. In mice euthanized 1 min after the [2H]TRLs injection, [2H]TRL margination was visualized along the luminal surface of glioma and heart capillaries, but not along the capillaries of normal brain (Figure 6A–B). After 1 min, deuterated lipids from the [2H]TRLs had already entered glioma cells and were even found in cytosolic neutral lipid droplets of those cells (Figure By contrast, 2H was absent in normal brain. As (He et al., we observed of lipids in in cytosolic lipid In gliomas harvested 30 min after the of we observed similar TRL margination along capillaries of gliomas and heart and the uptake of TRL-derived nutrients by glioma cells and (Figure 7). [2H]TRL margination was absent in capillaries of the normal brain at the and we did not find 2H in the parenchymal cells of the normal brain. We however, low levels of 2H in capillary endothelial cells of normal brain. Given the absence of TRL margination in normal brain capillaries, we that the low of 2H in brain capillary endothelial cells to [2H]TRL processing in the by the uptake of acids by endothelial cells of the brain. Figure Download asset Open asset NanoSIMS imaging margination of [2H]TRLs along glioma capillaries and 2H in adjacent glioma cells. C57BL/6 mice harboring CT-2A gliomas were for 4 and then injected intravenously with of After 1 min, mice were euthanized and perfusion-fixed with carbodiimide/glutaraldehyde. Tissue sections were processed for NanoSIMS imaging. (A) NanoSIMS images margination of [2H]TRLs in glioma capillaries. 1H– images were to visualize tissue 2H/1H and 1H– (blue) images [2H]TRLs in glioma and heart capillaries and lower The lower are images of the regions in the 2H/1H were to show TRLs. Scale 4 μm. (B) NanoSIMS images 2H in glioma 12C14N– images were to visualize tissue 2H/1H images margination of [2H]TRLs within the capillary lumen and lipid droplets in gliomas and There was no 2H in normal brain Scale 4 μm. The the in the 2H/1H above natural The was performed in mice with a of images for each were assessed a with Figure with 2 supplements see all Download asset Open asset NanoSIMS imaging 2H in gliomas 30 min after an of C57BL/6 mice harboring CT-2A gliomas were for 4 and then injected intravenously with of After 30 min, mice were euthanized and perfusion-fixed with carbodiimide/glutaraldehyde. of brain, and heart were processed for NanoSIMS imaging. 12C14N– images were to visualize tissue 2H/1H images margination of [2H]TRLs along the capillary lumen and 2H in glioma and in cytosolic lipid Images of normal brain revealed 2H in capillary endothelial cells. Scale 4 μm. The the in the 2H/1H above natural The was performed in mice, with a of images for each were assessed with a with both the and we observed in 2H in glioma cells, with perivascular cells 2H enrichment. We not the of the highly perivascular cells (i.e., whether are tumor cells, or we some cells within the glioma took more lipids than cells. As an we injected a mouse with PBS than with As expected, there was no 2H in the tissues of that mouse (Figure supplement 1). We performed an study in which [2H]TRLs were injected intravenously into a wild-type mouse and a Gpihbp1–/– After min, the and brains from these mice were harvested and processed for NanoSIMS imaging. The 2H/1H images revealed 2H in the heart of the wild-type mouse but 2H in the heart of the Gpihbp1–/– mouse in lipid droplets was than natural (Figure supplement 2). In the of 2H in the heart of the Gpihbp1–/– mouse was probably not the large of in the bloodstream of Gpihbp1–/– mice higher than that in wild-type the we were to 2H in the brain of either the wild-type mouse or the Gpihbp1–/– mouse (Figure supplement 2).