Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2016 May 10;15(6):1144-60.
doi: 10.1016/j.celrep.2016.04.029. Epub 2016 Apr 28.

Metabolic Symbiosis Enables Adaptive Resistance to Anti-angiogenic Therapy that Is Dependent on mTOR Signaling

Elizabeth Allen  1 , Pascal Miéville  2 , Carmen M Warren  3 , Sadegh Saghafinia  1 , Leanne Li  1 , Mei-Wen Peng  1 , Douglas Hanahan  4
Affiliations

Metabolic Symbiosis Enables Adaptive Resistance to Anti-angiogenic Therapy that Is Dependent on mTOR Signaling

Elizabeth Allen et al. Cell Rep. .

Abstract

Therapeutic _targeting of tumor angiogenesis with VEGF inhibitors results in demonstrable, but transitory efficacy in certain human tumors and mouse models of cancer, limited by unconventional forms of adaptive/evasive resistance. In one such mouse model, potent angiogenesis inhibitors elicit compartmental reorganization of cancer cells around remaining blood vessels. The glucose and lactate transporters GLUT1 and MCT4 are induced in distal hypoxic cells in a HIF1α-dependent fashion, indicative of glycolysis. Tumor cells proximal to blood vessels instead express the lactate transporter MCT1, and p-S6, the latter reflecting mTOR signaling. Normoxic cancer cells import and metabolize lactate, resulting in upregulation of mTOR signaling via glutamine metabolism enhanced by lactate catabolism. Thus, metabolic symbiosis is established in the face of angiogenesis inhibition, whereby hypoxic cancer cells import glucose and export lactate, while normoxic cells import and catabolize lactate. mTOR signaling inhibition disrupts this metabolic symbiosis, associated with upregulation of the glucose transporter GLUT2.

PubMed Disclaimer

Figures

None
Graphical abstract
Figure 1
Figure 1
Effects of Potent Angiogenesis Inhibition on mTOR Signaling; Upregulation and Relocalization into Focal Clusters (A) Quantitation of p-S6 intensity in western blots (WB) of tumors treated with sunitinib versus sham control (top left). All samples were first normalized using a Bradford assay and a WB for actin. The blots were probed with p-S6 antibodies, and the relative values of Mono-S- (n = 8, red) treated tumors were compared to controls (blue, n = 7) by quantitation using the Fusion FX7 imaging system (see Supplemental Information). A similar trend was seen for tumors treated with axitinib (data not shown). The protein lysates were prepared from PanNET tumors collected from RIP1Tag2 mice, following a 1-week trial from 14–15 weeks of age. The mice were treated daily (see Supplemental Information) with vehicle control (Con), rapamycin monotherapy (Mono-R), sunitinib monotherapy (Mono-S), or a combination of the two (R+S) (top right). The lysates were normalized to actin by WB, blotted, and probed with anti-p-S6 as a readout for mTOR signaling and reprobed for actin normalization. (B) The βTC4 cell line was cultured in normoxic (20% O2/5% CO2) or hypoxic (1% O2/5% CO2) conditions, and the lysates were prepared and analyzed by WB as above. (C) Tissue sections from tumors treated for 4 weeks in an intervention trial (see Figure 4A for a description of trial formats) were used for IHC using anti-p-S6. The representative images are shown. The scale bars represent 100 μm. (D) Representative images from tissue sections of tumors treated with sunitinib (top) or axitinib (bottom). Pimonidazole (pimo, green) staining indicates tumor hypoxia (right), p-S6 reactivity (red, arrowheads) indicates that mTOR/p-S6 signaling is mostly excluded from the hypoxic regions (middle), and the merged images are shown in the left image. The scale bars represent 50 μm. (E) A highly regressed sunitinib-treated tumor after 8 weeks of Mono-S (13–21 weeks). The islands of highly proliferative cells (Ki-67+) are embedded in fibrotic tissue, which are organized around the few remaining blood vessels (CD31, blue, arrows). The scale bar represents 700 μm in the leftmost image and 200 μm in all others.
Figure 2
Figure 2
Gene Specific Expression Analysis Reveals a Strongly Glycolytic Signature in Mono-S-Treated Tumors (A) Gene set overlapping analysis on the top 2,500 upregulated genes in sunitinib treatment compared to vehicle treated controls, revealing that glycolysis and hypoxia were among the most enriched pathway signatures (left). The glycolysis enrichment plot for sunitinib-treated tumors, including the profile of running ES Score are shown (right). (B) Top five ranked genes from the hallmark glycolysis gene set, adopted from the Molecular Signatures Database of the Broad Institute, for the sunitinib-treated cohort. (C) MCT4 expression was assessed using IHC on 15 week control tumors (Con, left) and tumors treated with rapamycin monotherapy (Mono-R, middle left), axitinib (Mono-A, middle right), or sunitinib (Mono-S, right). The MCT4 expression is high in 4/4 larger and 5/6 smaller Mono-A- and 4/4 larger Mono-S-treated tumors and absent in Con and Mono-R-treated tumors. The scale bars represent 2 mm in top image and 200 μm in the bottom image. (D) Pimonidazole (pimo, green) staining was performed to assess tumor hypoxia (center left); the glucose transporter, GLUT1, is shown in red (center right), and the leftmost image depicts the merged images. A merged image of R+A-treated tumors for MCT4 and pimonidazole is depicted in the rightmost image. The GLUT1 and MCT4 staining is highest in the most pimonidazole+/hypoxic regions, but can also be found in the peri-hypoxic areas. The scale bars represent 100 μm in the three left images and 50 μm in the right image. (E) IHC using anti-GLUT1 (middle) and anti-MCT4 (right) indicates that their expression is highly reduced/absent in tumors containing a cell-type-specific (β-cell) gene knock out of HIF1α that were treated with sunitinib. The top row shows a representative Rip1Tag2_Rip1Cre_Hif1αWT tumor, while the bottom row shows a Rip1Tag2_Rip1Cre_Hif1αfl/fl littermate whose tumors do not express Hif1α; this result is representative of all tumors from 4/4 wild-type (WT) versus 6/6 HIF1α KO mice, all similarly treated with sunitinib. The scale bar represents 100 μm. (F) Monotherapy with sunitinib (middle row) or axitinib (bottom row) elicits upregulation of MCT4 (green, first, second, and fourth) versus control untreated tumors (top row); in addition, MCT1 (in red) is upregulated in both AI-treated arms (middle and bottom rows, first, third, and fourth) versus controls (top row, first, third, and fourth). The scale bar represents 25 μm.
Figure 3
Figure 3
Mouse PNET Cell Lines and Tumors Consume and Catabolize Lactate to Establish Metabolic Symbiosis (A) βTC3 cancer cells cultured in normoxic conditions show low levels of GLUT1 (green) and MCT4 (green) expression, whereas both proteins are upregulated in hypoxic conditions. The scale bars represent 25 μm. (B) NMR analysis of βTC3 cells, cultured in low glucose (6.25 mM glucose + 2 mM Gln + FCS), plated in the same media at 7 × 106 for 2 days, switched to 0% glucose/0% Gln + FCS ON, then plated in 20 mM 1–13C-glucose + 2 mM Gln +FCS (top) or 20 mM 3–13C-lactate + 2 mM Gln+ FCS, cultured 16 hr in hypoxic (1% O2/5% CO2) or normoxic conditions (20% O2/5% CO2), and harvested for NMR. The net production of lactate in CM from time 0 in 1–13C-glucose conditions in hypoxia was +15.6 mM and normoxia +4.6 mM, while 3–13C-lactate levels were either increased in hypoxia by +0.4 mM or reduced in normoxia by −2.7 mM. The gray arrow highlights 4–13C-glutamine, which was found as a product of 3–13C-lactate catabolism in βTC3 cells when these results were replicated (data not shown). (C) Tumor-bearing mice were treated for 10 days with Mono-S or R+S, or vehicle control, and then infused with 3–13-C-lactate for 90 min prior to sacrifice. The tumors were excised, quick frozen in liquid nitrogen, and prepared for NMR (see Experimental Procedures).
Figure 4
Figure 4
Metabolic Regulation of mTOR Signaling (A) Pimonidazole (green) is merged with anti-GLS2 (red) in left images, and right images depict tumors anti-GLS2 only (red) in control untreated tumors (top) and sunitinib-treated tumors (bottom). The arrow indicates GLS2-negative staining, while the arrowhead indicates GLS2-positive staining in hypoxic cells. The scale bars represent 25 μm. (B) βTC3 cells, acclimated to low glucose as in Figure 3, were cultured in 2 mM glutamine (Gln), or 2 mM Gln + 20 mM lactate ± selective metabolic inhibitors, and assessed for p-S6 levels. Both βTC3s and lactate avid SiHa cells (Figure S2C) markedly upregulated p-S6 when cultured in lactate + Gln versus Gln alone; this upregulation could be reversed by 100 nM rapamycin treatment, or partially reversed using 40 μM DON, a competitive inhibitor of glutaminase. In addition, blocking lactate uptake with the MCT1 inhibitors 1 mM CHC or 10 μM 7ACC2 also reversed this upregulation, as did 200 μM AOA, which blocks the transamination of pyruvate + glutamate to alanine + α-ketoglutarate. The bottom images depict experiments performed with the selective GLS1 inhibitor, 50 μM BPTES, which failed to reverse the lactate induced pS6 upregulation, in contrast to DON, CHC, or 7ACC2. Below the blot is a graphic depicting the net consumption or production of lactate from time = 0. (C) Graphic depicting the relative production of α-ketoglutarate in Gln + lactate versus Gln-only conditions. The samples were normalized by protein concentration.
Figure 5
Figure 5
Therapeutic _targeting of AI-Induced, p-S6+ Clusters with Rapamycin (A) Trial designs are indicated in this schematic: trials were initiated and terminated at discrete time points, as established by previous studies in the Rip1Tag2 model (Bergers et al., 1999). The intervention trials were initiated at 11 weeks when tumors were small (Figure 5C), while fixed endpoint regression trials (Figure 5B) and open endpoint survival trials (Figure 5D) were initiated at 13 weeks when tumors were already large; molecular efficacy trials commenced at 13.5–14 weeks and proceeded for 7–10 days. (B) Fixed endpoint regression trials were performed from 13–15 weeks, as described in the Supplemental Information. Mean values ± SEM are indicated, and a two-tailed Mann-Whitney test was used to assess statistical significance; 13 week timepoint (TP) control versus R+S St and R+A St, p = 0.003∗∗ and versus R+S Sim, p = 0.0003∗∗∗; 15 week TP control versus Mono-R, p = 0.02∗, R+S St, p = 0.0003∗∗∗, R+S Sim, p = 0.0004∗∗∗, Mono-A, p = 0.006∗∗, and R+A St, p = 0.0005∗∗∗; Mono-R versus R+S Sim, p = 0.0008∗∗∗; Mono-S versus R+S St, p = 0.044∗ and R+S Sim, p = 0.0008∗∗∗; and Mono-A versus R+A, p = 0.004∗∗ (∗p = 0.05–0.01, ∗∗p = 0.009–0.001, and ∗∗∗p < 0.001). (C) Fixed endpoint intervention trials were performed for 4 weeks, commencing at 11 weeks, and described in Supplemental Information. Mean values ± SEM are indicated. Two-tailed Mann-Whitney test for statistical significance; 11–15 week vehicle control versus Mono-R and Mono-S, p = 0.0004∗∗∗, R≥S and R+S St, p = 0.0001∗∗∗, R+S Sim, p = 0∗∗∗, and 1/2R+1/2S and R+1/2S, p = 8 × 10−5; Mono-R versus R+S St, p = 0.036∗ and R+S Sim, p = 0.0006∗∗∗; Mono-S versus R+S St, p = 0.0009∗∗∗ and R+S Sim, p = 0.0001∗∗∗; R->S versus R+S St, p = 0.0032∗∗ and R+S Sim, p = 8 × 10−5∗∗∗ and R+1/2S (p = 0.0023∗∗); R+S St versus R+S Sim, p = 0.002∗∗; and 1/2R+1/2S versus R+S Sim, p = 0.0009∗∗∗ (∗p = 0.05–0.01, ∗∗p = 0.009–0.001, and ∗∗∗p < 0.001). (D) Open endpoint survival trials were performed in Rip1Tag2 mice from 13 weeks of age until animals became moribund and were sacrificed. Cohorts were composed of 7–12 mice/arm, and p values were derived using the log rank test (LR). Survival was assessed for each cohort as mean survival in weeks ± SD: control vehicle-treated mice (n = 10; 16.9 ± 0.8 weeks); Mono-S (n = 10; 21.8 ± 1.6 weeks); Mono-R (n = 10; 22.9 ± 1.7 weeks); Mono-A (n = 12; 21.7 ± 2.2 weeks); and R+S St (n = 9; 24.51 ± 3.3 weeks), R+S Sim (n = 7; 25.4 ± 3.8 weeks), and R+A St (n = 9; 25.8 ± 4.4 weeks). The R+S Sim arm was comprised of fewer animals than the other arms, signified by a stippled line, and was not statistically different than each cognate monotherapy. Long surviving, healthy mice from different arms were sacrificed for evaluation of TB and metastasis at the following time points: R+S St, 30.4 weeks; R+S Sim, 30.1 weeks; and two mice from R+A St, 28.4 weeks. All of the treated arms show significantly higher survival than the vehicle-treated mice of p < 0.0001∗∗∗, except for Mono-A, which has a p < 0.0002∗∗∗. R+S St versus Mono-S, p = 0.015∗; R+A St versus Mono-A, p = 0.002∗∗; and Mono-R, p = 0.004∗∗ (∗p = 0.05–0.01, ∗∗p = 0.009–0.001, and ∗∗∗p < 0.001). See also Figures S3A and S3B.
Figure 6
Figure 6
mTOR Inhibition Disrupts Metabolic Symbiosis (A) Evaluation by tissue immunostaining of MCT1 (red) and MCT4 (green) expression in representative tumors treated with Mono-R or Mono-S (top row) or with the combination (bottom row). The Mono-R-treated tumors (top left) appear similar to control vehicle-treated tumors (Figures 2F and S7) in having relatively weak expression of the MCT1/4 transporters. In the R+S-treated tumors (bottom), MCT4-expressing regions (green) are reduced versus MCT1-expressing regions, which have expanded (red), in contrast, note relatively more expanded MCT4-positive regions in Mono-S-treated tumors (top: center and right: green). The scale bar represents 100 μm. See also Figure S7. (B) In contrast to the overlapping distribution between GLUT1 (red) and hypoxia (pimo, green) in both Mono-S and R+S-treated tumors (top row), GLUT2 (red) is selectively upregulated in the non-hypoxic compartment of R+S-treated tumors (bottom row). The scale bars represent 25 μm. See also Figure S5B.
Figure 7
Figure 7
Schematic Conceptualizations of AI-Induced Metabolic Symbiosis (A) Untreated PanNET tumors are highly vascularized and express low levels of MCT1, MCT4, and GLUT1, but appreciable levels of GLUT2. Potent AIs _targeting the VEGFR and PDGFR pathways elicit regression of the tumor vasculature with consequent regional hypoxia and tumor compartmentalization, marked by upregulation of MCT4 and GLUT1 in a hypoxic compartment and elevated levels of MCT1 and pS6 along with reduced levels of GLUT2 in a normoxic compartment. Combined inhibition of VEGFR/PDGFR and mTOR produces necrotic cell death in the hypoxic compartment, associated with GLUT2 upregulation and altered lactate metabolism in the normoxic compartment, followed by eventual necrosis. (B) Extensive vascular collapse resulting from AI treatment results in hypoxia that induces HIF1α, which in turn upregulates the glycolytic _targets GLUT1 and MCT4 in the hypoxic cancer cells, leading to high levels of lactate secretion. Accumulating extracellular lactate induces expression of the MCT1 at the cell surface and lactate import into the normoxic compartment. In concert with serum-derived glutamine, lactate is catabolized in normoxic cancer cells with consequent induction of mTOR signaling to promote tumor metabolism. The data imply that normoxic cancer cells spare glucose for the hypoxic cells and fuel themselves by importing the lactate byproduct of glycolysis operative in the hypoxic cells in conjunction with glutamine. The bottom image suggests that metabolic symbiosis is disrupted by inhibition of mTOR through (indirect) upregulation of GLUT2 in the normoxic cells within the putative symbiotic clusters. Additionally, NMR studies indicate that mTOR inhibition disrupts the conversion of lactate-derived pyruvate/glutamate to alanine/α-ketoglutarate (anaplerosis) in dual-treated tumors, thereby disrupting the production of TCA cycle intermediates and potentially further enhancing toxic lactate accumulation.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Bergers G., Hanahan D. Modes of resistance to anti-angiogenic therapy. Nat. Rev. Cancer. 2008;8:592–603. - PMC - PubMed
    1. Bergers G., Javaherian K., Lo K.M., Folkman J., Hanahan D. Effects of angiogenesis inhibitors on multistage carcinogenesis in mice. Science. 1999;284:808–812. - PubMed
    1. Bergers G., Song S., Meyer-Morse N., Bergsland E., Hanahan D. Benefits of _targeting both pericytes and endothelial cells in the tumor vasculature with kinase inhibitors. J. Clin. Invest. 2003;111:1287–1295. - PMC - PubMed
    1. Brugarolas J., Lei K., Hurley R.L., Manning B.D., Reiling J.H., Hafen E., Witters L.A., Ellisen L.W., Kaelin W.G., Jr. Regulation of mTOR function in response to hypoxia by REDD1 and the TSC1/TSC2 tumor suppressor complex. Genes Dev. 2004;18:2893–2904. - PMC - PubMed
    1. Casanovas O., Hicklin D.J., Bergers G., Hanahan D. Drug resistance by evasion of antiangiogenic _targeting of VEGF signaling in late-stage pancreatic islet tumors. Cancer Cell. 2005;8:299–309. - PubMed

Publication types

MeSH terms

Supplementary concepts

  NODES
INTERN 1
Note 1
twitter 2