You are on page 1of 16

Article

LDHA-Associated Lactic Acid Production Blunts


Tumor Immunosurveillance by T and NK Cells
Graphical Abstract Authors
Almut Brand, Katrin Singer,
Gudrun E. Koehl, ...,
Wolfgang Mueller-Klieser,
Kathrin Renner, Marina Kreutz

Correspondence
marina.kreutz@ukr.de

In Brief
Brand et al. link altered tumor glucose
metabolism and immune escape and
show that increased lactic acid
production by LDHA in cancer cells
impairs cytokine production, in particular
IFN-g, in tumor-infiltrating T cells and NK
cells, thereby inhibiting tumor
immunosurveillance and promoting
tumor growth.

Highlights
d Human melanoma metastases exhibit a ‘‘Warburg
phenotype’’ with high lactic acid levels

d LDHA-associated lactic acid production and acidification


lead to immune evasion

d Lactic acid and acidification diminish NFAT levels and T and


NK cell activation

d LDHA expression in melanoma patients correlates with


survival and T cell activity

Brand et al., 2016, Cell Metabolism 24, 657–671


November 8, 2016 ª 2016 Elsevier Inc.
http://dx.doi.org/10.1016/j.cmet.2016.08.011
Cell Metabolism

Article

LDHA-Associated Lactic Acid Production


Blunts Tumor Immunosurveillance by T and NK Cells
Almut Brand,1 Katrin Singer,1 Gudrun E. Koehl,2 Marlene Kolitzus,1 Gabriele Schoenhammer,1 Annette Thiel,1
Carina Matos,1 Christina Bruss,1 Sebastian Klobuch,1 Katrin Peter,1,3 Michael Kastenberger,1 Christian Bogdan,4
Ulrike Schleicher,4 Andreas Mackensen,5 Evelyn Ullrich,5,6 Stefan Fichtner-Feigl,2,3 Rebecca Kesselring,2
Matthias Mack,3,7 Uwe Ritter,8 Maximilian Schmid,1,8 Christian Blank,9 Katja Dettmer,10 Peter J. Oefner,10
Petra Hoffmann,1,3 Stefan Walenta,11 Edward K. Geissler,2 Jacques Pouyssegur,12,13 Andreas Villunger,14,15
André Steven,16 Barbara Seliger,16 Stephan Schreml,17 Sebastian Haferkamp,17 Elisabeth Kohl,17 Sigrid Karrer,17
Mark Berneburg,17 Wolfgang Herr,1 Wolfgang Mueller-Klieser,11 Kathrin Renner,1,3 and Marina Kreutz1,3,18,*
1Department of Internal Medicine III, University Hospital Regensburg, 93053 Regensburg, Germany
2Department of Surgery, University Hospital Regensburg, 93053 Regensburg, Germany
3Regensburg Center for Interventional Immunology, University of Regensburg, 93053 Regensburg, Germany
4Mikrobiologisches Institut – Klinische Mikrobiologie, Immunologie und Hygiene, Universitätsklinikum Erlangen,

Friedrich-Alexander-Universität (FAU) Erlangen-Nürnberg, 91054 Erlangen, Germany


5Department of Internal Medicine 5, Hematology/Oncology, University Hospital Erlangen, 91054 Erlangen, Germany
6Cellular Immunology, Pediatric Stem Cell Transplantation and Immunology, Department for Children and Adolescents Medicine of the

University Hospital Frankfurt, Goethe-University, 60590 Frankfurt, Germany


7Department of Internal Medicine II - Nephrology, University Hospital Regensburg, 93053 Regensburg, Germany
8Institute of Immunology, University of Regensburg, 93053 Regensburg, Germany
9Division of Immunology, the Netherlands Cancer Institute, Antoni van Leeuwenhoek Hospital, Amsterdam 1066CX, the Netherlands
10Institute of Functional Genomics, University of Regensburg, 93053 Regensburg, Germany
11Institute of Pathophysiology, University Medical Center of the Johannes Gutenberg University Mainz, 55128 Mainz, Germany
12Institute of Research on Cancer and Aging, University of Nice-Sophia Antipolis, Centre A. Lacassagne, 06189 Nice, France
13Centre Scientifique de Monaco (CSM), 98000 Monaco, Monaco
14Medical University Innsbruck, Biocenter, Division of Developmental Immunology, 6020 Innsbruck, Austria
15Tyrolean Cancer Research Institute, 6020 Innsbruck, Austria
16Martin Luther University Halle-Wittenberg, Institute of Medical Immunology Halle/Saale, 06112 Halle, Germany
17Department of Dermatology, University Hospital Regensburg, 93053 Regensburg, Germany
18Lead Contact

*Correspondence: marina.kreutz@ukr.de
http://dx.doi.org/10.1016/j.cmet.2016.08.011

SUMMARY INTRODUCTION

Elevated lactate dehydrogenase A (LDHA) expres- Accelerated glucose metabolism in tumor cells, the so-called
sion is associated with poor outcome in tumor pa- ‘‘Warburg effect,’’ is based on the upregulation of glucose trans-
tients. Here we show that LDHA-associated lactic porter 1 (GLUT1) and glycolytic enzymes such as lactate dehy-
acid accumulation in melanomas inhibits tumor sur- drogenase A (LDHA), which is essential for the conversion of
veillance by T and NK cells. In immunocompetent pyruvate into lactate. Glycolysis requires continuous export of
lactate from cells by monocarboxylate transporters (MCTs),
C57BL/6 mice, tumors with reduced lactic acid
which co-transport lactate and protons. As a result, lactate
production (Ldhalow) developed significantly slower
and protons (‘‘lactic acid’’) accumulate in the tumor environment.
than control tumors and showed increased infiltra- In primary human tumors, high levels of lactate correlate with
tion with IFN-g-producing T and NK cells. However, incidence of distant metastases (Walenta et al., 2000), and the
in Rag2–/–gc–/– mice, lacking lymphocytes and NK level of LDHA correlates with the size and clinical stage of renal
cells, and in Ifng–/– mice, Ldhalow and control cells cell carcinomas and gastric cancer (Girgis et al., 2014; Sun et al.,
formed tumors at similar rates. Pathophysiological 2014). Based on these data, increased glycolysis by tumor cells
concentrations of lactic acid prevented upregulation appears to promote their growth, progression, and metastasis.
of nuclear factor of activated T cells (NFAT) in T and The relationship among increased lactate production, tumor
NK cells, resulting in diminished IFN-g production. growth, and metastasis has been investigated in different mouse
Database analyses revealed negative correlations models (Fantin et al., 2006; Rizwan et al., 2013; Xie et al., 2014).
Attenuation or disruption of Ldha resulted in reduced tumor
between LDHA expression and T cell activation
growth, and the authors attributed the diminished tumorigenicity
markers in human melanoma patients. Our results
to the compromised ability of tumor cells to grow under hypoxia
demonstrate that lactic acid is a potent inhibitor of or the importance of LDHA for tumor-initiating cells, respectively.
function and survival of T and NK cells leading to Tumor-derived lactic acid inhibits the differentiation and acti-
tumor immune escape. vation of monocytes and T cells in vitro (Dietl et al., 2010; Fischer

Cell Metabolism 24, 657–671, November 8, 2016 ª 2016 Elsevier Inc. 657
(legend on next page)
658 Cell Metabolism 24, 657–671, November 8, 2016
et al., 2007; Puig-Kröger et al., 2003). The tumor-promoting ef- We hypothesized that melanoma is a good model to study
fect of lactic acid might therefore be, in part, related to its immu- the relation between tumor glucose metabolism and immune
nosuppressive effects. Concordant with this hypothesis, Shime cell infiltration. Small hairpin RNAs (shRNAs) complementary
et al. demonstrated that lactic acid regulates expression and to Ldha were used to reduce expression of Ldha in B16.SIY mu-
secretion of the tumor-promoting cytokine interleukin 23 (IL-23) rine melanoma cells (Ldhalow cells). Untransfected cells or cells
(Shime et al., 2008). Furthermore, Husain et al. reported reduced transfected with a nonspecific, scrambled shRNA were used
numbers of myeloid-derived suppressor cells (MDSCs) in the as controls. Levels of Ldha mRNA and LDHA protein were
spleens of mice carrying Ldha-depleted tumors (Husain et al., reduced in Ldhalow1 and Ldhalow2 clones, compared to controls
2013). More recently, Colegio et al. showed that lactic acid (Figure 1D). Since the relative abundance of LDHA versus lactate
promotes development of M2-like macrophages by inducing dehydrogenase B (LDHB) determines the enzymatic activity of
hypoxia-inducible factor 1a (Colegio et al., 2014). These results the tetrameric LDH enzyme complex, we also analyzed LDHB
clearly indicate the importance of glycolysis and tumor-derived expression. LDHB was significantly upregulated at the mRNA,
lactic acid not only for the tumor growth itself, but also for the but not the protein, level in Ldhalow clones in vitro (Figure 1E).
immune cell balance in the tumor environment. In contrast, all clones expressed similar levels of Glut1 mRNA
Here we studied how lactic acid-induced changes modify the (Figure 1F).
anti-tumor immune response. We describe an important inter- To measure the enzymatic activity of LDH, we analyzed lactate
play between tumor-derived lactic acid and immune cells in in cell supernatants. As expected, Ldhalow tumor clones
the tumor environment. secreted significantly less lactate than control cells in vitro (Fig-
ure 1G). To determine whether the clones exhibited a stable
RESULTS phenotype in vivo, 1 3 105 tumor cells were injected subcutane-
ously into immunocompetent C57BL/6 mice. Established tumors
Human and Mouse Melanoma Tumors Exhibit the were dissected and analyzed. Levels of Ldha mRNA, but not
Warburg Phenotype Glut1 mRNA were significantly lower in Ldhalow tumors than
Bogunovic et al. described an immune expression signature in controls; Ldhb mRNA expression was diminished only in
as a predictor of survival in a cohort of patients with meta- Ldhalow2 tumors (Figures 1H–1J). We used quantitative biolumi-
static melanoma (Bogunovic et al., 2009). Using this dataset, nescence imaging to determine the distribution of glucose and
we found a negative correlation between LDHA expres- lactate in frozen sections of isolated tumors; glucose and lactate
sion and survival in these patients (Figure 1A). Furthermore, levels were lower in Ldhalow tumors (data not shown and Fig-
high LDHA expression correlated with low CD3d expression ures 1K and 1L). When we re-cultured tumor cells from Ldhalow
(r value = 0.396, p value = 7.9 3 10 3), indicating a negative and control tumors grown in C57BL/6 mice, we observed that
impact of lactic acid on T cell infiltration. To clarify whether lactate levels remained significantly different between the cell
LDHA expression is associated with increased lactate levels, types (data not shown).
bioluminescence analyses of melanoma of different stages Next, we investigated whether the attenuation of glycolysis
and adjacent healthy skin were performed. Cutaneous metas- would also affect other metabolic pathways that are involved
tases exhibited dramatically elevated lactate levels compared in tumor growth control and immune evasion by measuring
to primary melanomas and healthy skin (Figure 1B). Further- expression of indoleamine 2,3-dioxygenase 1 and 2 (Ido1,
more, glucose levels were significantly reduced in invasive pri- Ido2), cyclooxygenase 1 and 2 (Cox1, Cox2), Arg1, and Nos2.
mary melanomas and metastases. Interestingly, high lactate In tumor lysates we found no significant differences in the
and low glucose levels overlapped in a minority (n = 4) of expression of these genes between Ldhalow and control tumors
patients (Figure 1C and Table S1). (Figures S1A–S1F). Thus, knockdown of Ldha resulted in a stable

Figure 1. The Warburg Phenotype Is Present in Human and Mouse Melanoma Tumors
(A) Overall survival of 44 metastatic melanoma patients with high LDHA (blue) and low LDHA (red) expression levels, calculated with the ‘‘R2: Tumor Melanoma
Metastatic – Bhardwaj – 44 – fRMA – u133p2’’ dataset (http://r2.amc.nl).
(B and C) Quantification of tissue levels of lactate (B) and glucose (C) in human biopsies of melanoma in situ (Mis), primary melanoma (M), cutaneous metastases
of melanoma (M Met), and healthy tissue of the respective patients (Ctrl) by induced bioluminescence imaging. Each dot represents one biopsy (32 patients were
analyzed), and horizontal lines indicate the mean (one-way ANOVA, Tukey’s post test).
(D) qRT-PCR analysis (left panel) of Ldha mRNA in untransfected B16.SIY cells (WT), cells transfected with a scrambled shRNA sequence (Ctrl), and cells
transfected with shRNAs against Ldha (Ldhalow1 and Ldhalow2). The results are presented relative to the level of 18S rRNA. Each dot represents an individual
experiment, and horizontal lines indicate the mean. Immunoblot analysis (right panel) of LDHA and phospho-LDHA (p-LDHA) in whole-cell extracts. Actin was
used as loading control. One representative experiment out of three is shown.
(E) The left panel shows qRT-PCR analysis of Ldhb mRNA, and the right panel shows protein levels of LDHB, analyzed as described in (D).
(F) qRT-PCR analysis of Glut1 mRNA, analyzed as described in (D).
(G) Levels of lactate in the supernatants of cells, cultured for 24 hr (n R 21 independent experiments, mean and SEM). Dashed line represents lactate level in
culture medium.
(H–J) qRT-PCR analysis of Ldha (H), Ldhb (I), and Glut1 mRNA (J) in tumors of C57BL/6 mice after subcutaneous injection of 105 control (Ctrl), Ldhalow1, or
Ldhalow2 cells. The results are presented relative to the level of 18S rRNA. Each dot represents an individual mouse, and horizontal lines indicate the mean.
(K) Induced metabolic bioluminescence imaging of intra-tumor levels of lactate in sections of tumors grown from control or Ldhalow cells. Two representative
images are shown.
(L) Quantification of intra-tumor lactate levels identified as in (K) (n = 3–5 mice per group, mean and SEM, multiple tumor areas per sample analyzed). Statistical
analyses were performed using unpaired Student’s t test for (D–J and L).

Cell Metabolism 24, 657–671, November 8, 2016 659


Figure 2. Growth of B16 Ldhalow and Panc-Ldhanull Tumors Is Controlled in C57BL/6 Mice
(A) Proliferation of control (Ctrl), Ldhalow1, and Ldhalow2 cells, based on 3H-thymidine incorporation, after 24 hr (n R 7 independent experiments, mean
and SEM).
(legend continued on next page)
660 Cell Metabolism 24, 657–671, November 8, 2016
tumor cell phenotype with no effects on the other metabolic Rag2–/–gc–/– mice completely abolished the growth difference
pathways analyzed. between Ldhalow and control tumors (Figures 2E and 2F).
Together, these results show that growth control by T and NK
T Cells and NK Cells Control Tumor Growth in Low cells is possible in Ldhalow but not control B16.SIY tumors.
Lactate Tumors Next, we extended our analysis to another tumor system. Ldha
Little is known about the effects of glycolytic restriction on anti- knockout in the pancreatic adenocarcinoma cell line Panc02-H7
tumor immunity. We therefore studied the growth of Ldhalow and resulted in no detectable LDHA protein expression (Figure 2G),
control B16.SIY melanoma cells in vitro and in vivo. strongly diminished lactate production (Figure 2H), and similar
All clones, with the exception of Ldhalow1, proliferated similarly proliferation in vitro (Figure 2I). In contrast, tumor development
in vitro (Figure 2A), and no differences were noted in cell-cycle in C57BL/6 mice was markedly delayed and reduced when in-
distribution (Figure S2A). Proliferation of Ldhalow cells could be jecting Ldha knockout cells (Panc-Ldhanull) compared to con-
related to their increased mitochondrial content and activity trols (Figure 2J). In vitro analyses revealed a strong dependence
(Figures S2B and S2C). Metabolic analyses with 13C-labeled of Panc-Ldhanull cells on OXPHOS (Figures S2G and S2H).
glucose revealed that Ldha silencing increased the flux into
tricarboxylic acid cycle metabolites (Figure S2D). In line, oxygen Increased Numbers of Cytotoxic Effector Cells in
consumption related to ATP production was elevated in Ldhalow Ldhalow Compared to Control Tumors
clones and proliferation was severely diminished by oligomycin, Flow cytometry was used to analyze the immune cells present in
indicating that proliferation of ‘‘low’’ lactate but not ‘‘high’’ lactate tumors grown from 1 3 105 Ldhalow or control cells, isolated at
clones depends on oxidative phosphorylation (OXPHOS) (Fig- different time points after cells were injected into mice (see Fig-
ures S2E and S2F). Interestingly, both clones seem to depend ure 3A for gating strategy). Using the pan-leukocyte marker
on glutamine, as proliferation decreased by glutamine depriva- CD45, we found that approximately 5% of the cells analyzed
tion (Figure S2F). were hematopoietic cells. We did not observe a relationship
Following injection of 1 3 106 Ldhalow or control tumor cells between tumor size and the number of infiltrating leukocytes
into C57BL/6 mice, tumors grew at similar rates (Figure 2B). In (Figure 3B). However, Ldhalow tumors grown in in C57BL/6 (Fig-
support of our hypothesis, that lower numbers of tumor cells ure 3C), but not in Rag2–/– mice (Figure 3D), had a greater propor-
allow tumor growth to be controlled, 90% of mice injected with tion of CD45+ immune cells than control tumors.
1 3 104 Ldhalow cells were tumor free on day 18, whereas 70% In addition, the composition of the immune cell infiltrate and
of mice injected with control cells developed tumors until this the percentage of myeloid cells, NK cells, and T cells relative
time point (Figure 2C). To allow analysis of tumor immune cell to levels of total CD45+ cells were determined. Myeloid cells
infiltration, mice were injected with 1 3 105 tumor cells. Tumors (CD11b+) and myeloid suppressor cells (MDSCs, CD11b+Gr-1+)
originating from Ldhalow cells grew more slowly than those from were present in high numbers in all tumors; the percentage of
control cells, indicating that tumor control is still possible with myeloid cells did not correlate with tumor size when tumors up
this tumor load. No sex-specific differences in tumor growth to 1,000 mm3 in size were analyzed (Figure 3E). In Rag2–/– and
were observed (Figure 2D). C57BL/6 mice, a higher frequency of CD11b+ myeloid cells
To investigate whether lymphocytes and/or NK cells affect the was detected in control tumors than in Ldhalow tumors, indi-
growth of tumors, we injected 1 3 105 tumor cells into immuno- cating that lactate production promotes myeloid cell infiltration,
deficient Rag2–/– mice (lacking B and T cells) and Rag2–/–gc–/– regardless of the presence of T cells or the strain of mice (Figures
mice (lacking B, T, and NK cells). Growth of control tumors 3F and 3G). A greater proportion of MDSCs was only detected in
was similar in immunocompetent and immunodeficient mice, control tumors from Rag2–/– mice (Figures 3H–3J).
indicating that neither T cells nor NK cells are capable to restrict Myeloid cells accumulate in progressing tumors in humans
growth of tumors with high lactate secretion (Figure 2E). In and mice and limit the anti-tumor immune responses of effector
contrast, in Rag2–/– mice, Ldhalow tumors showed accelerated cells such as NK and T cells. We therefore investigated the
growth (Figure 2F). Lack of lymphocytes and NK cells in numbers of NK1.1+ and CD3+ cells in the tumor types. Again,

(B) Growth of tumors after subcutaneous injection of 106 control (Ctrl) or Ldhalow cells into C57BL/6 mice. Data represent mean tumor volume ± SEM (n = 4–5 mice
per group).
(C) Tumor-free survival of C57BL/6 mice after subcutaneous injection of 104 wild-type (WT), control (Ctrl), Ldhalow1, or Ldhalow2 cells (n = 10 mice per group).
(D) Tumor growth after subcutaneous injection of 105 cells into male (m) and female (f) C57BL/6 mice as described in (B) (n = 4–5 mice per group). In statistical
analyses, growth of tumors was compared between control cells and Ldhalow cells among mice of the same sex.
(E and F) Tumor growth after subcutaneous injection of 105 control (Ctrl) cells (E) or Ldhalow cells (F) into C57BL/6, immunodeficient Rag2–/– mice, and Rag2–/–gc–/–
mice (n = 8–10 mice per group ± SEM). In statistical analyses, growth of tumors in C57BL/6 mice was compared to Rag2–/– mice and Rag2–/–gc–/–, respectively
(black asterisks), and growth of Ldhalow tumors in Rag2–/– mice was compared with growth of Ldhalow tumors in Rag2–/–gc–/– mice (red asterisks).
(G) Immunoblot analysis of LDHA and LDHB in whole-cell extracts of untransfected Panc02-H7 cells (WT) and clones after transfection with Ldha CRISPR/Cas9
plasmids (Panc-Ldhahigh, Panc-Ldhanull). Actin was used as loading control. One representative experiment out of three is shown.
(H) Lactate level in the supernatants of untransfected WT and clones (Panc-Ldhahigh, Panc-Ldhanull) cultured for 24 hr (n R 6 independent experiments, mean and
SEM, unpaired Student’s t test). Dashed line represents lactate level in culture medium.
(I) Proliferation of Panc cells measured as described in (A) (n R 5 independent experiments).
(J) Growth of tumors after subcutaneous injection of 105 Panc tumor cells described in (G) in C57BL/6 mice. Data represent mean tumor volume ± SEM, n = 10
mice per group. Growth of WT tumors was compared to Panc-Ldhahigh tumors (gray asterisks); WT and Panc-Ldhahigh were compared to Panc-Ldhanull (black
asterisks) using unpaired Student’s t test.

Cell Metabolism 24, 657–671, November 8, 2016 661


Figure 3. B16 Ldhalow and Panc-Ldhanull Tumors Contain High
Numbers of Anti-tumor Effector Cells
(A) Gating strategy for flow cytometry analysis of leucocytes (CD45+),
living singular cells (DAPI- FSC-Wlow), myeloid cells (CD11b+), NK cells
(NK1.1+), MDSCs (CD11b+Gr-1+), T cells (CD3ε+), CD4+ T cells
(CD3ε+CD4+), and CD8+ T cells (CD3ε+CD8a+) in tumors from C57BL/6
mice after subcutaneous injection of 105 control (Ctrl) or Ldhalow cells.
Numbers in graphs indicate the percentage of cells. Plots of data from
one representative Ldhalow tumor are shown.
(B–Q) Correlations of CD45+ leukocytes among the entire cell pop-
ulations derived from B16 Ctrl and Ldhalow tumors of C57BL/6 mice with
tumor volume (B) and correlation of myeloid cells (E), MDSCs (H), NK
cells (K), and T cells (N) among living CD45+ leucocytes in tumors of
C57BL/6 mice with tumor volume. Each symbol represents an individual
mouse; correlations were calculated using a Pearson test. Percentage of
CD45+ leukocytes among the entire cell populations derived from tu-
mors of C57BL/6 mice (C) and percentage of immune cell populations
within the CD45+ cell population: myeloid cells (F), MDSCs (I), NK cells
(L), T cells (O), CD8+ T cells (P), and CD4+ T cells (Q). Percentage of
CD45+ leukocytes among the entire cell populations derived from tu-
mors of Rag2–/– mice (D) and percentage of immune cell populations
among the CD45+ cell population: myeloid cells (G), MDSCs (J), and
NK cells (M) (all identified as in A).
(R and S) Percentage of T cells (R) and NK cells (S) among CD45+ cells in
tumors of C57BL/6 mice after subcutaneous injection of 105 Panc-
Ldhahigh and Panc-Ldhanull cells. Each symbol represents an individual
mouse; small horizontal lines indicate the mean (unpaired Student’s
t test).

662 Cell Metabolism 24, 657–671, November 8, 2016


Figure 4. Ldhalow Tumors Contain Higher Amounts of Functionally Active CD8+ T Cells and NK Cells
(A and B) qRT-PCR analysis of Ifng mRNA in control (Ctrl) and Ldhalow tumors grown in C57BL/6 mice (A) and Rag2–/– mice (B). The results are presented relative
to the level of 18S rRNA. Each symbol represents an individual mouse; small horizontal lines indicate the mean.
(C–F) Proportions of IFN-g+ cells among CD8+ T cells (C) and NK cells (E), and granzyme B+ cells among CD8+ T cells (D) and NK cells (F) in tumors from C57BL/6
mice, measured by flow cytometry. Each symbol represents an individual mouse; small horizontal lines indicate the mean (unpaired Student’s t test).

no relation between NK cell infiltration and tumor size was found of mice bearing Ldhalow versus control tumors, indicating that
(Figure 3K). However, the number of NK cells was increased in lactate has a local effect on immune cells in the tumor
Ldhalow tumors grown in C57BL/6 and Rag2–/– mice, compared environment.
with control tumors (Figures 3L and 3M). Furthermore, we
observed no correlation between T cell infiltration and tumor Lactic Acid Impairs Cytokine Production of Tumor-
size, but the number of T cells was increased in Ldhalow tumors Infiltrating T Cells and NK Cells
grown in C57BL/6 mice (Figures 3N and 3O). The frequency of As T cell abundance alone is not informative regarding the anti-
CD3+CD8+, but not CD3+CD4+, T cells was significantly greater tumor potency of T cells, we investigated the functional compe-
in Ldhalow tumors than in control tumors (Figures 3P and 3Q). In tence of tumor-infiltrating T cells. qRT-PCR analysis revealed
Panc-Ldhanull tumors derived from C57BL/6 mice, we also significantly higher mean levels of interferon-g (Ifng) mRNA in ly-
observed an increase in CD3+ T cells by trend compared to sates from Ldhalow tumors as compared to lysates from control
Panc-Ldhahigh tumors (Figure 3R). Furthermore NK1.1+ cells tumors grown in C57BL/6 mice (Figure 4A); a similar trend
were significantly enriched, whereas no change was detected was observed in immunodeficient Rag2–/– mice (Figure 4B). In
regarding myeloid cells (Figure 3S and data not shown). contrast, there was no significant difference in the levels of
Our findings suggest a link between tumor lactate levels and mRNAs of tumor necrosis factor (Tnf) and vascular endothelial
the local composition of tumor-infiltrating immune cells. Since growth factor a (Vegfa) in Ldhalow versus control tumors in
tumors expressing high levels of Ldha modulate immune cell C57BL/6 mice (Figures S4A and S4B). Ldhalow and control tu-
infiltration and differentiation in the spleens of tumor-bearing mors were isolated from C57BL/6 mice, and levels of IFN-g
mice (Husain et al., 2013), we analyzed myeloid and lymphoid and granzyme B were measured by flow cytometry in T cells
cells in the blood and spleen of tumor-bearing C57BL/6 mice. and in NK cells. We found that the proportions of IFN-g and
More myeloid (suppressor) cells and NK cells were found in granzyme B producing CD8+ T cells were significantly higher in
the blood of all tumor-bearing mice than in the blood of healthy Ldhalow tumors than in control tumors (Figures 4C and 4D),
mice, whereas the numbers of CD3+ T cells were significantly whereas no significant differences in the numbers of IFN-
reduced in the blood of tumor-bearing mice (Figures S3A– g+CD4+ and IL-17+CD4+ T cells were observed between tumor
S3D). A similar trend was observed for immune cells in the types (Figures S4C and S4D). NK cells also showed an increased
spleen (Figures S3E–S3H). We detected no significant differ- expression of IFN-g and granzyme B in Ldhalow tumors
ence in the composition of immune cells in the blood and spleen compared to control tumors (Figures 4E and 4F). Thus, Ldhalow

Cell Metabolism 24, 657–671, November 8, 2016 663


Figure 5. Lactic Acid Suppresses Cytokine Production in CD8+ T Cells and NK Cells and Promotes Apoptosis
(A) Flow cytometric analysis of IFN-g+ and IL-2+ cells among stimulated CD8+ T cells (upper panel) and stimulated NK cells (lower panel) after incubation in the
absence (–) or presence (LA) of lactic acid, acidification (pH 6.4), or sodium lactate (NaL) for 24 hr. One representative experiment of three is shown.
(B) qRT-PCR analysis of Ifng mRNA expression in unstimulated or stimulated (PMA/Ionomycin) CD8+ T cells in the absence (–) or presence (LA) of lactic acid after
3 hr. Two independent experiments are shown. The results are presented relative to the level of 18S rRNA.
(C) Quantification of Annexin V+7-AAD+ CD8+ T cells isolated from spleens of C57BL/6 mice (WT) or vav-Bcl-2 mice after incubation in the absence (–) or presence
(LA) of lactic acid, acidification (pH), and sodium lactate (NaL) for 24 hr (n = 3 independent experiments, mean and SEM, paired Student’s t test).
(D and E) Percentage IFN-g+ T cells (upper panel) and IL-2+ T cells (lower panel) after coculture with either wild-type (WT), control (Ctrl), or Ldhalow cells without (D)
or with (E) IFN-g pre-treatment (n = 3–4 independent experiments, mean and SEM, unpaired Student’s t test).

tumors contain a higher proportion of anti-tumor effector cells lular IFN-g and IL-2 by flow cytometry. Lactic acid reduced the
than control tumors, which might impede their growth. expression of IFN-g on protein and mRNA level (Figures 5A
and 5B). NK cells were more vulnerable to the effects of lactic
Lactic Acid, but Not Its Sodium Salt, Inhibits Function acid, as 15 mM lactic acid already completely blocked IFN-g
and Survival of T Cells and NK Cells production (Figure 5A). Acidification alone also diminished cyto-
Knockdown of Ldha in mouse melanoma cells significantly kine production, but to a lesser extent compared to 15 mM lactic
reduced the level of lactate in tumors grown from these cells (Fig- acid. In contrast, sodium lactate had no effect (Figure 5A). Higher
ures 1K and 1L), and lactic acid has been shown to affect expres- levels of lactic acid (20 mM) or the respective acidification (pH
sion of IFN-g, IL-2, and granzyme B in cultured human T cells 5.8) induced apoptosis of T cells irrespective of the expression
(Fischer et al., 2007). We investigated whether lactic acid also in- level of Bcl-2 (Figure 5C). Similar results were obtained with
hibits expression of cytokines in mouse T and NK cells in vitro. NK cells (Figure S5A). Lactic acid did not modulate the expres-
We stimulated mouse CD8+ T cells and NK cells in the absence sion of activation or exhaustion markers on activated CD8+ 2C
or presence of lactic acid, sodium lactate, or acidification corre- T cells but impaired upregulation of CD25 when present during
sponding to 15 mM lactic acid (pH 6.4) and measured intracel- the activation process of naive CD8+ T cells (Figure S5B).

664 Cell Metabolism 24, 657–671, November 8, 2016


Figure 6. Lactic Acid Uptake by CD8+ T Cells Leads to Intracellular Acidification and Inhibition of NFAT Upregulation upon Stimulation
(A) Lactate uptake in CD8+ T cells after incubation with 15 mM 13C1 lactate and HCl corresponding to the pH of 15 mM lactic acid with or without an MCT1 inhibitor
for 4 hr (one experiment performed in duplicates).
(B) Intracellular pH of activated CD8+ T cells after incubation with lactic acid (0–20 mM LA) for 30 min using pH-controlled buffers (pH 4.5, pH 5.5, pH 6.5, and
pH 7.5) for calibration. One representative experiment of three is shown.
(C and D) Intracellular ATP levels of unstimulated and stimulated (PMA/Ionomycin) CD8+ T cells (C) and NK cells (D) in the absence (–) or presence of lactic acid
(LA). One experiment in triplicates (C) and two independent experiments (D) were performed.
(E and F) NFATc1 expression in unstimulated and stimulated (PMA/Ionomycin) CD8+ T cells (E) and NK cells (F) in the absence (–) or presence (LA) of 15 mM lactic
acid, HCl (pH 6.4), and sodium lactate (NaL) for 24 hr by flow cytometry (n R 3 independent experiments for all treatments, except for HCL n = 2). Data calculated
relative to control (PMA/Ionomycin). Data represent mean and SEM; paired Student’s t test.

We also measured expression of IFN-g and IL-2 in SIY- Intracellular Acidification Caused by Lactic Acid
specific CD8+ 2C T cells co-cultured with Ldhalow and control Disturbs NFAT Expression
B16.SIY tumor cells. CD8+ 2C T cells cultured with control tumor Based on our findings with human T cells (Fischer et al., 2007),
cells expressed lower levels of IFN-g and IL-2 (Figure 5D). we hypothesized that uptake of lactic acid by CD8+ T cells leads
In some experiments, tumor cells were pre-incubated with to intracellular acidification. Therefore, mouse CD8+ T cells were
IFN-g to upregulate MHC class I expression, which allows better incubated with 13C1-labeled sodium lactate in the presence of
recognition by SIY-specific T cells. We measured higher HCl to mimic ‘‘lactic acid’’ conditions and measured intracellular
13
levels of cytokines in these cultures, and there was still a C1-lactate levels. Comparable to human T cells, mouse T cells
detectable difference between Ldhalow and control co-cultures showed an uptake of 13C1-lactate, which was poorly inhibited by
(Figure 5E). pre-incubation with a monocarboxylate transporter 1 (MCT1) in-
These data demonstrate that high levels of lactic acid prevent hibitor (Figure 6A). The MCT1 inhibitor did not prevent the inhibi-
expression of IFN-g by mouse CD8+ T and NK cells. Secretion of tion of IFN-g expression or apoptosis induction (data not shown).
lactate and protons by melanoma cells seems to reduce the pro- Uptake of 13C1-lactate was accompanied by an intracellular
portions of effector T and NK cells (Figures 3P and 3L), and their accumulation of unlabeled lactate, which indicates that high
production of anti-tumor cytokines such as IFN-g (Figures 5A extracellular lactate levels disturb efflux of lactate from CD8+
and 5B), which likely promotes tumor immune evasion and T cells (data not shown). As lactate is transported together with
growth. protons, we measured intracellular acidification. Exposure of

Cell Metabolism 24, 657–671, November 8, 2016 665


(legend on next page)

666 Cell Metabolism 24, 657–671, November 8, 2016


CD8+ T cells to lactic acid dose-dependently decreased the Granzyme B expression in tumor-infiltrating CD8+ T cells was
intracellular pH (Figure 6B). Intracellular acidification was paral- reduced in highly glycolytic tumors (Figure 4D); granzyme B and
leled by a drop in ATP levels in T and NK cells, indicating an perforin are required for the cytotoxic activity of T and NK cells.
impaired energy metabolism (Figures 6C and 6D). Therefore, 1 3 105 Ldhalow and control tumor cells were injected
We then analyzed signaling pathways and transcription fac- into mice that lack perforin (Pfp–/– mice), and tumor growth was
tors involved in IFN-g transcription such as nuclear factor of monitored. Unexpectedly, and in contrast to Ifng–/– mice, the
activated T cells (NFAT) and NF-kB proteins (Sica et al., 1997). growth of Ldhalow and control tumors differed significantly in
Besides its activation by calcineurin, NFAT is upregulated during Pfp–/– mice (Figure 7C). This finding indicates that reduced
activation of T cells on mRNA and protein level (Yan et al., 2013). growth of Ldhalow tumors in mice does not result from a per-
In T and NK cells, exposure to lactic acid or HCl inhibited upre- forin-dependent cytotoxic activity of T cells. Accordingly, we
gulation of NFAT during activation (Figures 6E and 6F), whereas did not detect lysis of Ldhalow or control tumor cells by mouse
IkB degradation, a prerequisite for NF-kB activation, was not CD8+ T cells in vitro (data not shown). Based on these results,
altered (Figure S6A). Lactic acid also caused a slight decrease we concluded that the reduced growth of Ldhalow tumors could
in phosphorylated AKT and P38 MAP kinase (Figures S6B and in part be related to the cytotoxic effects of NK cells, because
S6C). These data suggest that inhibition of NFAT is mainly IFN-g has been shown to be important for NK cell activation
responsible for the suppression of IFNg expression in T and and suppression of tumor growth (Takeda et al., 2001).
NK cells. The percentages of NK cells in tumors grown in Ifng–/– and
Ifngr1–/– mice were much smaller than those of C57BL/6 mice
IFN-g Is Required for Immune Control of Ldhalow Tumors (Figure 7D), whereas in spleens of tumor-bearing Ifng–/– mice
Since Ldhalow tumors were infiltrated by greater proportions of NK cell numbers were diminished in control, but not in Ldhalow
IFN-g+ CD8+ T cells and NK cells than control tumors (Figures tumors, compared to healthy C57BL/6 mice (Figure S7A). In
4C and 4E), we speculated that the growth of Ldhalow tumors is contrast, the number of myeloid (suppressor) cells did not differ
regulated via IFN-g production of T and/or NK cells. In this significantly between Ldhalow tumors in C57BL/6, Ifng–/–, and
case, absence of IFN-g expression by immune cells would miti- Ifngr1–/– mice (data not shown). However, a trend for enrichment
gate the growth difference between Ldhalow and control tumors. of myeloid (suppressor) cells was observed in spleens of Ifng–/–
We injected 1 3 105 Ldhalow and control tumor cells into C57BL/6 mice compared to healthy C57BL/6 mice (Figures S7B and
mice and C57BL/6 mice with a disrupted Ifng gene (Ifng–/– mice) S7C). Higher levels of CD8+ T cells were detected in Ldhalow tu-
and compared the growth of tumors from our cell lines. mors of Ifng–/– or Ifngr1–/– mice compared to C57BL/6 mice, but
Again, the growth of tumors from Ldhalow versus control cells the percentage of activated, CD25+CD8+ T cells was signifi-
significantly differed in C57BL/6 mice. This difference was cantly higher in Ldhalow tumors grown in C57BL/6 mice,
completely lost in Ifng–/– mice. In fact, Ldhalow tumors grew compared to knockout mice (Figures 7E and 7F). IFN-g therefore
more rapidly in Ifng–/– mice than in C57BL/6 mice, whereas appears to be important for T cell activation and NK cell recruit-
growth of control tumors was not changed (Figure 7A). These ment in mice.
findings indicate the potential role of IFN-g in immunosurveil- To correlate our findings with patient data, we analyzed im-
lance and growth control of Ldhalow tumors. mune infiltration of cutaneous melanoma metastases, which
IFN-g is a pleiotropic cytokine that has different effects on host exhibit high lactate levels (Figure 1B) in comparison to adjacent
and tumor cells. It directly inhibits tumor cell proliferation but also healthy tissue. Both tissues were heavily infiltrated by T cells;
modulates stroma cell function and activation. To distinguish be- however, only healthy adjacent tissue exhibited activated
tween the direct effects of IFN-g on tumor cells and its effects on CD8+CD25+ T cells, indicating that lactate levels may contribute
the tumor stroma, we injected 1 3 105 Ldhalow or control tumor to immunosuppression in human melanoma as well (Figure 7G).
cells, which express the IFN-g receptor 1 (IFNGR1) into Ifngr1–/– In line, database analyses (TCGA Research Network: http://
mice. Under these conditions, IFN-g secretion by mouse stroma cancergenome.nih.gov/) revealed a negative correlation be-
cells could only act directly on the tumor cells that were injected, tween LDHA expression and T cell activation markers such as
but not on host cells. Just as in Ifng–/– mice, the growth rates of granzyme K (GZMK) and CD25 in human melanoma (Figures
Ldhalow and control tumors were similar in Ifngr1–/– mice. This 7H and 7I).
indicates that IFN-g regulates the growth of tumors indirectly, We conclude that activation of tumor-infiltrating immune cells
by acting on cells in the tumor environment (Figure 7B). is determined by LDHA expression and lactic acid levels in the

Figure 7. Correlation between LDHA, IFN-g, and CD25 Expression in Mouse and Human Melanoma
(A) Growth of control (Ctrl) and Ldhalow tumors in C57BL/6 mice and Ifng–/– mice (mean tumor volume ± SEM, n = 5 mice per group). Growth of control and Ldhalow
tumors was compared in C57BL/6 mice (black asterisks); growth of Ldhalow tumors was compared with growth of Ldhalow tumors in knockout mice (red
asterisks).
(B) Tumor growth in C57BL/6 mice and Ifngr1–/– mice as described in (A).
(C) Tumor growth in Pfp–/– mice as described in (A) (n = 7–8 mice per group).
(D–F) Quantification of NK cells (D) and CD8+ T cells (E) among living leucocytes, and CD25+ cells among CD8+ T cells (F) in control (Ctrl) and Ldhalow tumors
(identified as in Figure 3A). Each symbol represents an individual mouse (small horizontal lines indicate the mean, unpaired Student’s t test).
(G) Flow cytometric analyses of CD8+ cells among CD3+ T cells (upper panel) and CD25+ cells among CD8+ T cells (lower panel) from biopsies of human
cutaneous melanoma metastases and adjacent healthy tissue (control).
(H and I) Correlations of LDHA with GZMK (H) or CD25 (I) gene expression determined in a dataset containing 470 melanoma tumors (‘‘R2: Tumor Skin Cutaneous
Melanoma – TCGA – 470 – rsem – tcgars’’). Patient samples were ordered by LDHA levels (red squares). Pearson’s correlation was calculated.

Cell Metabolism 24, 657–671, November 8, 2016 667


tumor environment. High levels of lactate and concomitant How does lactic acid impair NK and T cell activation and func-
acidification limit immune cell activation and allow immune tion? Activated T cells perform glycolysis and require efficient
escape. secretion of lactate and protons, as intracellular accumulation
of lactate disturbs their metabolism (Fischer et al., 2007). The
DISCUSSION export of lactate by MCTs depends on a gradient from cyto-
plasmic to extracellular lactate concentrations. Accordingly,
A positive correlation between LDHA, high lactate levels, and tu- we show that high concentrations of extracellular lactic acid
mor progression has been documented in various tumor entities lead to lactate and proton uptake and subsequently lowered
including melanoma (Girgis et al., 2014; Walenta et al., 2000). the intracellular pH in T cells. MCT1 inhibition could not block
Immunohistochemical analysis revealed high expression of the acidification, as other MCTs or transporter-independent diffu-
tetrameric LDHA (LDH-5) in human melanoma and metastases, sion may be involved (Parks et al., 2013). In contrast, Pilon-
which was associated with reduced disease-free and overall sur- Thomas and colleagues described that acidification did not
vival of patients (Zhuang et al., 2010). In contrast, high levels of affect cytoplasmic pH in T cells (Pilon-Thomas et al., 2016). Dif-
tumor-infiltrating T cells predict a favorable prognosis in patients ferences in the kinetics could explain these conflicting results.
with melanoma and other types of cancer (Bogunovic et al., Lactate uptake and accumulation disturbed energy metabolism
2009; Galon et al., 2006). A new aspect in this context is the link- as ATP levels were lowered. In line with these data, sodium
age between immune cell activation, tumor cell LDHA expres- lactate or lactic acid have been shown to inhibit T cell glycolysis
sion, and lactate levels in the tumor environment. Our study and function (Dröge et al., 1987; Haas et al., 2015). Impaired
shows that high LDHA expression is associated not only with glycolysis could relieve GAPDH from its glycolytic function, allow
poor prognosis, but also with lower expression of T cell markers binding to IFN-g mRNA, and thereby prevent its efficient transla-
(Bogunovic et al., 2009). Analyses of melanoma biopsies further tion (Chang et al., 2013). We suggest that intracellular acidifica-
revealed that cutaneous melanoma metastases exhibit high tion also interferes with the regulation of NFAT, an important
levels of lactate that could contribute to the immunosuppression transcription factor implicated in the transcriptional control
in melanoma patients. Lactate levels in primary melanomas were of IFN-g. Furthermore, acidification may also disturb NFAT
comparable to the adjacent tissue, indicating that other mecha- translocation to the nucleus as calcineurin, a phosphatase that
nisms lead to immunosuppression in these patients. In this re- dephosphorylates NFAT and thereby regulates its translocation,
gard, studies by Chang et al. and Ho et al. have shown that is suppressed by low pH (Hisamitsu et al., 2012).
accelerated glucose metabolism of tumor cells leads to glucose Moreover, lactic acid concentrations above 20 mM caused T
deprivation and suppresses anti-tumor T cell effector function and NK cell apoptosis, which might explain smaller proportions
(Chang et al., 2015; Ho et al., 2015). Accordingly, we detected of T cells and NK cells in tumors with higher concentrations of
lower glucose levels in 50% of primary melanoma biopsies. lactate. In contrast, myeloid cells seem to resist the apoptosis-
Therefore, highly glycolytic tumors can use different strategies inducing effect of lactic acid, as high numbers were found in tu-
to suppress T cell activation: glucose restriction and/or lactate mors with high lactate levels. Accordingly, we could show that
accumulation. human monocytes do not undergo apoptosis even in the pres-
As in vivo data on the effects of lactic acid are lacking, we ence of high lactic acid concentrations (Dietl et al., 2010). More-
studied its impact in mouse tumor models and observed that over, lactic acid induces polarization of macrophages to a
the level of lactic acid affects immune cell populations and M2-like phenotype via induction of hypoxia-inducible factor 1a
thereby regulates tumor growth. (Colegio et al., 2014). In addition, it stimulates expression of
In our melanoma model we found that tumor-derived lactic IL-23 (Shime et al., 2008) and increases the number of MDSCs
acid reduced the numbers and activity of CD8+ T cells and NK in the spleens of mice carrying LDHA-expressing tumors (Husain
cells in vitro and in vivo. Both immune cell populations were et al., 2013). It is therefore likely that lactic acid promotes tumor
important for the immune control of tumors with low lactate growth by directly suppressing T cell function and generating tu-
secretion as tumor growth was accelerated in immunodeficient mor-promoting myeloid cells. In support of this theory, we also
Rag2–/– mice and Rag2–/–gc–/– mice. Ablation of IFN-g also detected high numbers of MDSCs, especially in control tumors,
blurred the growth difference between Ldhalow and control which could promote tumor growth. Production of IFN-g by
tumors. In these mice, T cells seemed to be non-activated, as T cells and/or NK cells in Ldhalow tumors may counteract
they lack CD25 expression, and NK cells were almost undetect- MDSC development, because IFN-g has been shown to support
able. These findings reveal the importance of IFN-g for T and NK the reprogramming of MDSCs into antigen-presenting cells (Ker-
cell homeostasis in the tumor environment. In line, IFN-g plays an kar et al., 2011).
important role in the development of hepatic NK cells (Wu et al., Although lactic acid inhibits T cell and NK cell activation and
2012). In addition, IFN-g can limit tumor growth by inhibiting leads to MDSC accumulation, we cannot rule out the possibility
angiogenesis (Qin et al., 2003) or directly suppressing tumor that other cells, e.g., B cells, are affected by lactic acid and
cell proliferation (Kakuta et al., 2002). VEGF-regulated angiogen- contribute to the accelerated tumor growth of B16.SIY control
esis did not appear to be affected by lactic acid in our model as cells. B lymphocytes have been shown to foster carcinogenesis
similar levels of Vegf mRNA were detected in Ldhalow and control in squamous cell carcinoma (Affara et al., 2014), whereas plasma
tumors. Nor did it seem that IFN-g was directly inhibiting prolifer- cell signatures are predictors of favorable survival across solid
ation as the difference in growth of Ldhalow versus control tumors tumors in another study (Gentles et al., 2015). Therefore, the
observed in C57BL/6 mice was not seen in Ifngr1–/– mice in impact of lactic acid on B cells and plasma cells remains to be
which tumor cells still express IFNGR1. elucidated.

668 Cell Metabolism 24, 657–671, November 8, 2016


Surprisingly, tumor growth of control and Ldhalow tumors was Induced Metabolic Bioluminescence Imaging of Intra-tissue Lactate
similar in Pfp–/– and C57BL/6 mice, suggesting that the granule and Glucose Levels
Metabolites were measured in serial tumor cryosections by the imBI method,
exocytosis pathway of T and NK cells is not important for the lim-
as described earlier (Walenta et al., 2002).
itation of tumor growth in Ldhalow tumors. In support of our re-
sults, B16.SIY cells are rarely killed by CD8+ T cells because In Vitro Analysis of CD8+ T Cells and NK Cells
PD-L1 limits T cell activity (Blank et al., 2004). CD8+ T cells and NK cells were isolated from spleens by magnetic bead sep-
Checkpoint inhibition by PD-1/PD-L1 or CTLA-4 blockade is a aration (Miltenyi) and cultured under standard conditions. T cells and NK
major advance in melanoma therapy; however, only about half cells were incubated in the absence or presence of L-lactic acid (Sigma-Al-
of the patients benefit from this new treatment. Accumulation drich) or HCl or sodium L-lactate (Sigma-Aldrich). For cytokine and Ifng
mRNA expression analysis, cells were stimulated with 20 ng/ml PMA (Calbio-
of lactic acid in the tumor environment may limit the success of
chem) and 1 mM Ionomycin (Enzo Biochem) for 3 hr. For NFAT analysis,
checkpoint inhibition. Accordingly, blocking acidification prior PMA/Ionomycin stimulation was performed for 30 min. For coculture exper-
to immunotherapy improved the anti-tumor response (Calcinotto iments, irradiated B16.SIY cells were cultured with or without 2 ng/ml murine
et al., 2012; Pilon-Thomas et al., 2016). In line, two studies IFN-g (Pepro Tech). After 24 hr, stimulated 2C CD8+ T cells were added for
showed that LDH is a selection criterion for ipilimumab and another 24 hr.
anti-PD-1 therapy (Diem et al., 2016; Kelderman et al., 2014).
Quantitative Real-Time PCR
Alternatively, genetic manipulation can rescue T cell function in
Total RNA of B16.SIY tumor cells, homogenized mouse tissues, and 2C CD8+
a highly glycolytic tumor environment (Ho et al., 2015). T cells was obtained using the RNeasy Mini Kit (QIAGEN). Complementary
In summary, the (tumor) microenvironment modulates the acti- DNA was synthesized with a M-MLV Reverse Transcriptase kit (Promega)
vation of infiltrating cells and thereby regulates the balance be- and amplified by qPCR with the QuantiFast SYBR Green PCR Kit (QIAGEN)
tween the immune response and tolerance. The finding that using the Mastercyler Ep Realplex (Eppendorf).
tumor-derived lactic acid can suppress the T and NK cells
Quantification of Intracellular pH
adds another piece to the tumor immunosuppression puzzle.
Intracellular pH in 2C CD8+ T cells after incubation with different concentra-
tions of L-lactic acid (Sigma-Aldrich) for 30 min was assessed by flow cy-
EXPERIMENTAL PROCEDURES tometry using the pH-sensitive dye carboxy SNARF-1 AM acetate and
pH-controlled buffers (Life Technologies) for calibration. For analyses, ratios
Human Skin Biopsies of emission wavelength l1, transmitted by a 585/42BP filter and wavelength
Snap-frozen biopsies from tumor and healthy tissue of melanoma patients l2, transmitted by a 670LP filter, were calculated.
were obtained after informed consent. The study was approved by the Institu-
tional Ethics Committee of the University Hospital of Regensburg and de- ATP Measurement
signed and conducted in accordance with the Declaration of Helsinki. Clinical Cellular ATP levels were determined using the ATP determination kit (Life
characteristics are summarized in Table S1. Technologies) according to the manufacturer’s protocol.

Mice and In Vivo Experiments 13


C Tracing Experiments
Animal experiments were performed according to the regulations of the gov- To determine lactate uptake, 2C CD8+ T cells were incubated for 4 hr with
ernment of Upper Palatinate, Regensburg, Germany. Unless otherwise indi- 15 mM 13C1 lactate. Cells were harvested with 80% methanol and analyzed
cated, male C57BL/6 mice (Charles River), Rag2–/– and Rag2–/–gc–/– mice by GC-MS as described in the Supplemental Experimental Procedures.
(Taconic), Ifng–/– and Ifngr1–/– mice (The Jackson Laboratory), and Pfp–/–
mice (The Jackson Laboratory) were used. To generate tumors in mice, 1 3 Flow Cytometry
104, 1 3 105, or 1 3 106 tumor cells were injected subcutaneously in the dorsal Cells were stained with fluorochrome-conjugated antibodies. Intracellular
region of mice. vav-Bcl-2 mice were used for in vitro analysis of T cells. staining for cytokines and NFATc1 was performed using the BD Cytofix/Cy-
toperm Plus Kit (BD Biosciences). Stained cell populations were acquired
Genetic Modulation of Ldha in Tumor Cell Lines by LSR II (Beckton Dickinson) and data analyzed by FlowJo software
B16.SIY mouse melanoma cells (provided by Christian Blank; Blank et al., (Tree Star).
2004) were transfected with shRNA plasmids (GeneClip U1 Hairpin Cloning
System, Promega) to specifically knockdown expression of Ldha (Ldhalow Database Analyses
clones). Untransfected cells (WT) and cells transfected with a non-specific, Kaplan Meier estimation curves for overall survival of individual metastatic
scrambled shRNA (Control) served as controls. melanoma patients were generated with the microarray analysis and visualiza-
In the mouse pancreatic adenocarcinoma cell line Panc02-H7, Ldha tion platform R2 (http://r2.amc.nl) by using the ‘‘R2: Tumor Melanoma
knockout was performed with CRISPR/Cas9 using a pSpCas9(BB)-2A-GFP Metastatic – Bhardwaj – 44 – fRMA – u133p2’’ dataset. Correlation of LDHA
plasmid (Addgene) designed by Jerome Durivault (Centre Scientifique de expression with GZMK and CD25 was determined with the ‘‘R2: Tumor Skin
Monaco). Untransfected cells (WT) served as controls. Tumor cells were Cutaneous Melanoma – TCGA – 470 – rsem – tcgars’’ dataset (http://
cultured under standard conditions. cancergenome.nih.gov).

Proliferation Statistical Analysis


Proliferation of cells was determined by measuring 3H-thymidine incorporation Results represent the mean and SEM. Comparison between groups was
for 24 hr (Hartmann Analytic). performed using paired or unpaired Student’s t test or one-way ANOVA and
Tukey’s post test (*p < 0.05, **p < 0.01, ***p < 0.001). All calculations were
Western Blotting performed using GraphPad Prism 5 Software.
Whole cell lysates were analyzed according to standard procedures.
SUPPLEMENTAL INFORMATION
Lactate Measurements
Lactate concentrations in supernatants of cells cultured for 24 hr were deter- Supplemental Information includes Supplemental Experimental Procedures,
mined enzymatically using an ADVIA 1650 instrument (Bayer) and specific re- seven figures, and one table and can be found with this article online at
agents (Roche). http://dx.doi.org/10.1016/j.cmet.2016.08.011.

Cell Metabolism 24, 657–671, November 8, 2016 669


AUTHOR CONTRIBUTIONS Dröge, W., Roth, S., Altmann, A., and Mihm, S. (1987). Regulation of T-cell
functions by L-lactate. Cell. Immunol. 108, 405–416.
A.B. and K.S. designed experiments, collected data, performed analyses, and Fantin, V.R., St-Pierre, J., and Leder, P. (2006). Attenuation of LDH-A expres-
wrote the paper; W.M.-K. and S.W. performed bioluminescence experiments; sion uncovers a link between glycolysis, mitochondrial physiology, and tumor
G.E.K., M. Kastenberger, G.S., M. Kolitzus, U.S., E.U., R.K., S. Klobuch, S. maintenance. Cancer Cell 9, 425–434.
Karrer, E.K., S.S., K.P., P.J.O., M.S., K.D., A.T., C.M., C. Bruss, S.H., and
Fischer, K., Hoffmann, P., Voelkl, S., Meidenbauer, N., Ammer, J., Edinger, M.,
K.R. were involved in experiments and data collection; S.F.-F., C. Blank,
Gottfried, E., Schwarz, S., Rothe, G., Hoves, S., et al. (2007). Inhibitory effect of
A.M., W.H., J.P., M.B., B.S., E.K.G., and M.M. provided reagents and dis-
tumor cell-derived lactic acid on human T cells. Blood 109, 3812–3819.
cussed data; U.R., C. Bogdan, A.V., and P.H. provided mice and discussed
data; A.S. analyzed data; and M. Kreutz designed the experiments, analyzed Galon, J., Costes, A., Sanchez-Cabo, F., Kirilovsky, A., Mlecnik, B., Lagorce-
data, and wrote the paper. Pagès, C., Tosolini, M., Camus, M., Berger, A., Wind, P., et al. (2006). Type,
density, and location of immune cells within human colorectal tumors predict
clinical outcome. Science 313, 1960–1964.
ACKNOWLEDGMENTS
Gentles, A.J., Newman, A.M., Liu, C.L., Bratman, S.V., Feng, W., Kim, D., Nair,
We acknowledge the excellent technical assistance of Alice Peuker, Stephanie V.S., Xu, Y., Khuong, A., Hoang, C.D., et al. (2015). The prognostic landscape
Faerber, Monika Wehrstein, and Anna Hoehn. We thank Eva Gottfried for help- of genes and infiltrating immune cells across human cancers. Nat. Med. 21,
ful discussions, Fabian Schuler for mouse manipulation, and J. Adams for vav- 938–945.
Bcl-2 mice. This work was supported by DFG KFO 262 and the Mildred Scheel Girgis, H., Masui, O., White, N.M., Scorilas, A., Rotondo, F., Seivwright, A.,
Foundation (Grants MS 110703 and MS 109247). The results shown here are in Gabril, M., Filter, E.R., Girgis, A.H., Bjarnason, G.A., et al. (2014). Lactate
part based upon data generated by The Cancer Genome Atlas Research dehydrogenase A is a potential prognostic marker in clear cell renal cell carci-
Network (http://cancergenome.nih.gov/). noma. Mol. Cancer 13, 101.
Haas, R., Smith, J., Rocher-Ros, V., Nadkarni, S., Montero-Melendez, T.,
Received: October 16, 2015 D’Acquisto, F., Bland, E.J., Bombardieri, M., Pitzalis, C., Perretti, M., et al.
Revised: April 20, 2016 (2015). Lactate Regulates Metabolic and Pro-inflammatory Circuits in
Accepted: August 19, 2016 Control of T Cell Migration and Effector Functions. PLoS Biol. 13, e1002202.
Published: September 15, 2016
Hisamitsu, T., Nakamura, T.Y., and Wakabayashi, S. (2012). Na(+)/H(+)
exchanger 1 directly binds to calcineurin A and activates downstream NFAT
REFERENCES
signaling, leading to cardiomyocyte hypertrophy. Mol. Cell. Biol. 32, 3265–
3280.
Affara, N.I., Ruffell, B., Medler, T.R., Gunderson, A.J., Johansson, M.,
Bornstein, S., Bergsland, E., Steinhoff, M., Li, Y., Gong, Q., et al. (2014). B cells Ho, P.C., Bihuniak, J.D., Macintyre, A.N., Staron, M., Liu, X., Amezquita, R., Tsui,
regulate macrophage phenotype and response to chemotherapy in squamous Y.C., Cui, G., Micevic, G., Perales, J.C., et al. (2015). Phosphoenolpyruvate Is a
carcinomas. Cancer Cell 25, 809–821. Metabolic Checkpoint of Anti-tumor T Cell Responses. Cell 162, 1217–1228.

Blank, C., Brown, I., Peterson, A.C., Spiotto, M., Iwai, Y., Honjo, T., and Husain, Z., Huang, Y., Seth, P., and Sukhatme, V.P. (2013). Tumor-derived
Gajewski, T.F. (2004). PD-L1/B7H-1 inhibits the effector phase of tumor rejec- lactate modifies antitumor immune response: effect on myeloid-derived sup-
tion by T cell receptor (TCR) transgenic CD8+ T cells. Cancer Res. 64, 1140– pressor cells and NK cells. J. Immunol. 191, 1486–1495.
1145. Kakuta, S., Tagawa, Y., Shibata, S., Nanno, M., and Iwakura, Y. (2002).
Bogunovic, D., O’Neill, D.W., Belitskaya-Levy, I., Vacic, V., Yu, Y.L., Adams, Inhibition of B16 melanoma experimental metastasis by interferon-gamma
S., Darvishian, F., Berman, R., Shapiro, R., Pavlick, A.C., et al. (2009). through direct inhibition of cell proliferation and activation of antitumour host
Immune profile and mitotic index of metastatic melanoma lesions enhance mechanisms. Immunology 105, 92–100.
clinical staging in predicting patient survival. Proc. Natl. Acad. Sci. USA 106, Kelderman, S., Heemskerk, B., van Tinteren, H., van den Brom, R.R., Hospers,
20429–20434. G.A., van den Eertwegh, A.J., Kapiteijn, E.W., de Groot, J.W., Soetekouw, P.,
Calcinotto, A., Filipazzi, P., Grioni, M., Iero, M., De Milito, A., Ricupito, A., Cova, Jansen, R.L., et al. (2014). Lactate dehydrogenase as a selection criterion for
A., Canese, R., Jachetti, E., Rossetti, M., et al. (2012). Modulation of microen- ipilimumab treatment in metastatic melanoma. Cancer Immunol. Immunother.
vironment acidity reverses anergy in human and murine tumor-infiltrating 63, 449–458.
T lymphocytes. Cancer Res. 72, 2746–2756. Kerkar, S.P., Goldszmid, R.S., Muranski, P., Chinnasamy, D., Yu, Z., Reger,
Chang, C.H., Curtis, J.D., Maggi, L.B., Jr., Faubert, B., Villarino, A.V., R.N., Leonardi, A.J., Morgan, R.A., Wang, E., Marincola, F.M., et al. (2011).
O’Sullivan, D., Huang, S.C., van der Windt, G.J., Blagih, J., Qiu, J., et al. IL-12 triggers a programmatic change in dysfunctional myeloid-derived cells
(2013). Posttranscriptional control of T cell effector function by aerobic glycol- within mouse tumors. J. Clin. Invest. 121, 4746–4757.
ysis. Cell 153, 1239–1251. Parks, S.K., Chiche, J., and Pouysségur, J. (2013). Disrupting proton dynamics
Chang, C.H., Qiu, J., O’Sullivan, D., Buck, M.D., Noguchi, T., Curtis, J.D., and energy metabolism for cancer therapy. Nat. Rev. Cancer 13, 611–623.
Chen, Q., Gindin, M., Gubin, M.M., van der Windt, G.J., et al. (2015). Pilon-Thomas, S., Kodumudi, K.N., El-Kenawi, A.E., Russell, S., Weber, A.M.,
Metabolic Competition in the Tumor Microenvironment Is a Driver of Cancer Luddy, K., Damaghi, M., Wojtkowiak, J.W., Mulé, J.J., Ibrahim-Hashim, A., and
Progression. Cell 162, 1229–1241. Gillies, R.J. (2016). Neutralization of Tumor Acidity Improves Antitumor
Colegio, O.R., Chu, N.Q., Szabo, A.L., Chu, T., Rhebergen, A.M., Jairam, V., Responses to Immunotherapy. Cancer Res. 76, 1381–1390.
Cyrus, N., Brokowski, C.E., Eisenbarth, S.C., Phillips, G.M., et al. (2014). Puig-Kröger, A., Pello, O.M., Selgas, R., Criado, G., Bajo, M.A., Sánchez-
Functional polarization of tumour-associated macrophages by tumour- Tomero, J.A., Alvarez, V., del Peso, G., Sánchez-Mateos, P., Holmes, C.,
derived lactic acid. Nature 513, 559–563. et al. (2003). Peritoneal dialysis solutions inhibit the differentiation and matura-
Diem, S., Kasenda, B., Spain, L., Martin-Liberal, J., Marconcini, R., Gore, M., tion of human monocyte-derived dendritic cells: effect of lactate and glucose-
and Larkin, J. (2016). Serum lactate dehydrogenase as an early marker for degradation products. J. Leukoc. Biol. 73, 482–492.
outcome in patients treated with anti-PD-1 therapy in metastatic melanoma. Qin, Z., Schwartzkopff, J., Pradera, F., Kammertoens, T., Seliger, B., Pircher,
Br. J. Cancer 114, 256–261. H., and Blankenstein, T. (2003). A critical requirement of interferon gamma-
Dietl, K., Renner, K., Dettmer, K., Timischl, B., Eberhart, K., Dorn, C., mediated angiostasis for tumor rejection by CD8+ T cells. Cancer Res. 63,
Hellerbrand, C., Kastenberger, M., Kunz-Schughart, L.A., Oefner, P.J., et al. 4095–4100.
(2010). Lactic acid and acidification inhibit TNF secretion and glycolysis of Rizwan, A., Serganova, I., Khanin, R., Karabeber, H., Ni, X., Thakur, S., Zakian,
human monocytes. J. Immunol. 184, 1200–1209. K.L., Blasberg, R., and Koutcher, J.A. (2013). Relationships between LDH-A,

670 Cell Metabolism 24, 657–671, November 8, 2016


lactate, and metastases in 4T1 breast tumors. Clin. Cancer Res. 19, 5158– metastases, tumor recurrence, and restricted patient survival in human
5169. cervical cancers. Cancer Res. 60, 916–921.
Shime, H., Yabu, M., Akazawa, T., Kodama, K., Matsumoto, M., Seya, T., and Walenta, S., Schroeder, T., and Mueller-Klieser, W. (2002). Metabolic mapping
Inoue, N. (2008). Tumor-secreted lactic acid promotes IL-23/IL-17 proinflam- with bioluminescence: basic and clinical relevance. Biomol. Eng. 18, 249–262.
matory pathway. J. Immunol. 180, 7175–7183. Wu, X., Chen, Y., Sun, R., Wei, H., and Tian, Z. (2012). Impairment of hepatic
Sica, A., Dorman, L., Viggiano, V., Cippitelli, M., Ghosh, P., Rice, N., and NK cell development in IFN-g deficient mice. Cytokine 60, 616–625.
Young, H.A. (1997). Interaction of NF-kappaB and NFAT with the interferon- Xie, H., Hanai, J., Ren, J.G., Kats, L., Burgess, K., Bhargava, P., Signoretti, S.,
gamma promoter. J. Biol. Chem. 272, 30412–30420. Billiard, J., Duffy, K.J., Grant, A., et al. (2014). Targeting lactate dehydroge-
Sun, X., Sun, Z., Zhu, Z., Guan, H., Zhang, J., Zhang, Y., Xu, H., and Sun, M. nase–a inhibits tumorigenesis and tumor progression in mouse models of
(2014). Clinicopathological significance and prognostic value of lactate dehy- lung cancer and impacts tumor-initiating cells. Cell Metab. 19, 795–809.
drogenase A expression in gastric cancer patients. PLoS ONE 9, e91068. Yan, J., Su, H., Xu, L., and Wang, C. (2013). OX40-OX40L interaction promotes
Takeda, K., Smyth, M.J., Cretney, E., Hayakawa, Y., Yamaguchi, N., Yagita, proliferation and activation of lymphocytes via NFATc1 in ApoE-deficient mice.
H., and Okumura, K. (2001). Involvement of tumor necrosis factor-related PLoS ONE 8, e60854.
apoptosis-inducing ligand in NK cell-mediated and IFN-gamma-dependent Zhuang, L., Scolyer, R.A., Murali, R., McCarthy, S.W., Zhang, X.D., Thompson,
suppression of subcutaneous tumor growth. Cell. Immunol. 214, 194–200. J.F., and Hersey, P. (2010). Lactate dehydrogenase 5 expression in melanoma
Walenta, S., Wetterling, M., Lehrke, M., Schwickert, G., Sundfør, K., Rofstad, increases with disease progression and is associated with expression of
E.K., and Mueller-Klieser, W. (2000). High lactate levels predict likelihood of Bcl-XL and Mcl-1, but not Bcl-2 proteins. Mod. Pathol. 23, 45–53.

Cell Metabolism 24, 657–671, November 8, 2016 671

You might also like