Lactate dehydrogenase B is required for the growth of KRAS-dependent lung adenocarcinomas.

PURPOSE
This study is aimed to identify genes within the KRAS genomic amplicon that are both coupregulated and essential for cell proliferation when KRAS is amplified in lung cancer.


EXPERIMENTAL DESIGN
We used an integrated genomic approach to identify genes that are coamplified with KRAS in lung adenocarcinomas and subsequently preformed an RNA interference (RNAi) screen to uncover functionally relevant genes. The role of lactate dehydrogenase B (LDHB) was subsequently investigated both in vitro and in vivo by siRNA and short hairpin RNA (shRNA)-mediated knockdown in a panel of lung adenocarcinoma cells lines. LDHB expression was also investigated in patient tumors using microarray and immunohistochemistry analyses.


RESULTS
RNAi-mediated depletion of LDHB abrogated cell proliferation both in vitro and in xenografted tumors in vivo. We find that LDHB expression correlates to both KRAS genomic copy number gain and KRAS mutation in lung cancer cell lines and adenocarcinomas. This correlation between LDHB expression and KRAS status is specific for lung cancers and not other tumor types that harbor KRAS mutations. Consistent with a role for LDHB in glycolysis and tumor metabolism, KRAS-mutant lung tumors exhibit elevated expression of a glycolysis gene signature and are more dependent on glycolysis for proliferation compared with KRAS wild-type lung tumors. Finally, high LDHB expression was a significant predictor of shorter survival in patients with lung adenocarcinomas.


CONCLUSION
This study identifies LDHB as a regulator of cell proliferation in a subset of lung adenocarcinoma and may provide a novel therapeutic approach for treating lung cancer.


INTRODUCTION
Lung cancer is one of the most prevalent cancer forms, responsible for over one million annual deaths worldwide. In the clinical setting, lung cancer is classified according to two main histological types, small-cell lung cancer (SCLC) and non-small cell lung carcinoma (NSCLC) (1). Eighty-five percent of all lung cancers are attributable to non-small-cell lung cancer (NSCLC), of which lung adenocarcinoma is the most frequent histological subtype (1). On a molecular level, lung adenocarcinomas frequently harbor mutations in the KRAS oncogene (25%) and the tumor suppressor protein p53 (33%) (1). The importance of these mutations is highlighted by the development of genetically engineered mouse models that harbor KRAS and p53 mutations and are able to drive lung adenocarcinoma initiation and progression (2). In recent years, molecular and cancer genomic approaches have increased our understanding of the pathogenesis of NSCLC and have led to therapies that directly target the precise genetic alterations that drive tumor growth.
The molecular genotyping of NSCLC into discrete molecular classes has redefined therapeutic approaches. Oncogenic alterations in the kinases Epidermal Growth Factor Receptor (EGFR) and Anaplastic Lymphoma Kinase (ALK) are found in 15-25% of NSCLC (3,4).
Patients with tumors harboring the activated EGFR and ALK kinase respond to the tyrosine kinase inhibitors gefitinib and erlotinib (5,6) and crizotinib (7), respectively. Importantly, ALK and EGFR activation is mutually exclusive of KRAS mutation, underscoring the need for identifying and characterizing therapeutic options for KRAS driven lung cancer. Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
RAS genes are members of the small GTPase super-family (11) and function to propagate growth factor signaling through activation of c-RAF and PI 3-kinase pathways. In spite of concerted efforts, KRAS mutation defines a genetic subtype of lung cancer that is currently not amenable to therapeutic intervention (12). Recent studies have utilized high throughput chemical and genetic screens to identify synthetic lethal phenotype in the context of KRAS mutations (13)(14)(15)(16). These screens have yielded promising kinases and other cellular machinery. Significant work is required to determine whether these interactions will translate into therapies for KRAS driven tumors.
While previous studies have mostly focused on identifying synthetic interactions in KRAS mutant cell lines, here we set out to identify genes within the KRAS genomic amplicon that are both co-upregulated and critical for cell proliferation when KRAS is amplified. We characterized the 12p11/12 amplicon where KRAS resides, and report the presence of 18 additional genes that are co-amplified and overexpressed in lung adenocarcinomas. Using a loss of function RNAi screen on these 18 genes, we identified Lactate Dehydrogenase B (LDHB) as being specifically required in KRAS amplified lung cancer for cell proliferation. We find that LDHB expression significantly correlates to both KRAS copy number gain and KRAS mutation in lung adenocarcinoma cell lines and tumors. LDHB knockdown reduced cell growth in KRAS mutant lung cancer both in vitro and in vivo. Lastly, LDHB was a prognostic indicator of poor outcome in lung adenocarcinomas. Our work identifies LDHB and more broadly, lactate metabolism, as a potential therapeutic target for KRAS driven lung cancer.

siRNA screen
An RNAi screen for genes that regulate lung cancer growth was carried out in four lung adenocarcinoma cell lines, two with chromosome 12p KRAS amplification (NCI-H838, NCI-H322T) and two without KRAS amplification (RERF-LC-KJ, NCI-H1568). Genes were selected Dharmacon siGENOME siRNAs were individually reverse transfected using DharmaFECT 4 (Dharmacon) in 96-well format in triplicate. Four independent siRNAs were utilized for each gene. Cell number was determined with CellTiter-Glo (Promega) six days after transfection. Values were normalized to a non-targeting control siRNA, and then Z scores were calculated using the following formula: (gene value -plate average) / plate standard deviation.
To reduce the Z scores for each gene to a single comparable value, a ∆Z score was calculated: average of the Z scores from the two 12p amplified cell lines minus average Z scores of the nonamplified cell lines. Genes with a ∆Z score <-1 were considered hits in the 12p amplified cell lines. Only genes with two or more siRNA oligos with a ∆Z score <-1 were considered further.

Gene expression analysis
For cell lines, RNA was harvested from 96 well plates three or four days after siRNA transfection using the TaqMan Gene Expression Cells-to-CT kit (Applied Biosystems). For mouse tissues/tumors, RNA was harvested using Qiagen RNeasy kit. Quantitative RT-PCR was carried out with the Taqman One-Step RT-PCR Master Mix Reagents kit using Taqman Gene Expression Assays according to manufacturer protocol (Applied Biosystems). All samples were normalized to a GAPDH control.
For each shRNA cell line, 5 x 10 6 cells were injected subcutaneously into the backs of female NCr nude mice (Taconic) to initiate tumor growth. After tumors reached 200-300 mm 3 in size, the animals from each cell line were split into two groups and fed either 5% sucrose or 5% sucrose + 1 mg/ml doxycycline to induce hairpin expression. Tumors with a starting volume <200 mm 3 at the time of treatment were excluded from further analysis. After seven days, three mice from each group were euthanized and the tumors were harvested for LDHB knockdown analysis. Tumor measurements were carried out on the remaining mice every 3-4 days until Day GSE11969; (22)). LDHB expression was mean-centered across all tumors then separated by above or below the mean. All patients were censored from the study after >5 years of follow-up, thus the 5-year survival is shown. P-value was calculated by a log-rank test.  Table 1; (9)). To further characterize the KRAS locus in lung adenocarcinoma, we performed a GISTIC analysis on copy number data derived from 735 NSCLC tumors and cell lines. While KRAS is the most significantly amplified gene (Q-value = 10 -22 ), the region of copy number gain is relatively broad and harbors 18 additional genes that are both amplified and overexpressed ( Figure 1A, B).

Immunohistochemistry
To functionally interrogate the candidate genes residing in the 12p12 region of copy number gain, we performed a loss of function RNAi screen in lung adenocarcinoma cell lines that harbored either 12p amplification (NCI-H838, NCI-H322T) or were disomic at this region (RERF-LC-KJ and NCI-H1568). Gene expression and knockdown for all genes in the screen was confirmed by quantitative RT-PCR in NCI-H838 cells (Supplementary Figure 1). By comparing the Z scores between the amplified and disomic cell lines, genes with a differential cell proliferation effect were identified ( Figure 1C). In addition to KRAS, two additional genes, LDHB and MED21, met our significance criteria of at least two independent siRNAs scoring at a ∆Z score of -1 or below (Z score chromosome 12p amplified lines minus Z score disomic lines). MED21 (hSrb7) is a member of the Mediator complex and is a general regulator of transcription (23). Lactate dehydrogenase B (LDHB) is a metabolic enzyme that catalyzes the interconversion of lactate and pyruvate (24,25).
Since LDHB and MED21 are frequently co-amplified with KRAS, we first examined whether expression of these genes also correlate to KRAS mutation status. We assessed LDHB To ensure that these effects were not due to technical differences in knockdown levels, loss of LDHB was confirmed by immunoblot in each cell line tested. Moreover, no changes in LDHA protein levels were observed upon LDHB knockdown ( Figure 3B). Since KRAS-mutation does not always confer dependence on KRAS for growth (27) To determine whether LDHB itself is a downstream KRAS target, we examined LDHB gene expression after KRAS knockdown. We observed that LDHB gene expression did not directly change after KRAS knockdown (Supplementary Figure 3B), implying that LDHB is required in KRAS mutant cells, but is not directly regulated by KRAS activity.
Since LDHB expression does not appear to be directly regulated by KRAS, we considered whether LDHB is upregulated in lung cancers that contain mutations in other oncogenic drivers (ie. EGFR mutation, c-Met amplification, and ALK fusion). Therefore, we immunoblotted a panel of cell lines that were annotated for mutation status (Supplementary Figure 4A). LDHB was also upregulated in other lung cancer subtypes, in particular those driven by c-MET (2/2 cell lines) and EGFR (3/8 cell lines). The level of overexpression was similar to that observed in KRAS mutant cancers. To functionally test whether LDHB overexpression reflected a cellular dependence on LDHB, we knocked down LDHB with two independent siRNA oligos in a subset of these cell lines (Supplemental Figure 4B, C). We find that cell lines with high levels of LDHB are statistically more sensitive to loss of LDHB (P = 0.00005) than LDHB low expressing lines ( Figure 3C). Taken together these data suggest that targeting LDHB may provide a broad therapeutic option for lung cancer patients that specifically overexpress LDHB.
The metabolic and growth requirements of a three-dimensional tumor can be starkly different from cancer cells grown in vitro. To test whether LDHB knockdown affects tumor growth in vivo, we utilized an inducible short hairpin (shRNA) lentiviral system to acutely deplete endogenous LDHB in fully formed tumors. Two independent shRNAs to LDHB (shLDHB) and a non-targeting control (shNTC) were introduced into a human KRAS mutant lung cancer cell line (NCI-H2122) and subsequently grown as xenografted tumors in mice.
Doxycycline-induced LDHB knockdown in fully formed tumors led to significant tumor growth inhibition when compared to the uninduced shLDHB tumors or shNTC controls ( Figure 4A).
LDHB knockdown in the tumors was confirmed by immunohistochemistry and immunoblot at day 7 after doxycycline administration ( Figure 4B, C). These observations indicate that LDHB is required for growth of KRAS mutant lung tumors in vivo.

KRAS mutant lung cell lines and adenocarcinomas are dependent on glycolysis.
The 'Warburg effect' is named after the observation that many cancer cells generate energy using a high rate of glycolysis, instead of oxidative phosphorylation (28). The high rate of glycolysis occurs regardless of oxygen supply and produces an excess of pyruvate, which gets converted to lactate by LDHA or LDHB and subsequently exported from the cell. We hypothesized that LDHB upregulation may be associated with a more general shift in cellular metabolism toward glycolytic dependence. Since LDHB is overexpressed in KRAS mutant tumors, we analyzed and compared the expression of a defined glycolytic gene signature (see

Methods) between KRAS wild-type and KRAS mutant lung tumors. Consistent with our
Research.
on August 13, 2021. © 2012 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
hypothesis, KRAS mutant tumors exhibited an elevated expression of the glycolytic gene signature compared to KRAS wild-type tumors ( Figure 5A). Moreover, lung tumors that have increased expression of the RAS gene signature also show increased expression of the same glycolysis genes ( Figure 5A). This effect appears to be specific to lung cancer, as a correlation between RAS signature expression and glycolytic signature expression was not observed in colon tumors (Supplementary Figure 2C). This suggests that lung tumors with KRAS mutation and RAS pathway activation are associated with a dependence on glycolysis.
To explore the functional significance of these gene expression changes, we tested the sensitivity of seven KRAS wild-type (low LDHB expressing) and six KRAS mutant (high LDHB expressing) cell lines to the glycolysis inhibitor 2-deoxyglucose (2DG). KRAS mutant cell lines were significantly more sensitive to 2DG inhibition than KRAS wild-type cells ( Figure   5B), suggesting that KRAS mutant cells are more addicted to glycolysis. Taken together these data imply that KRAS mutant tumors are associated with a shift in tumor metabolism that is characterized by increased dependence on glycolysis for energy demands.
The dependence of KRAS-mutated lung tumors on glycolysis and LDHB implies that these tumors are sensitive to perturbations in the end-stage of glycolysis, namely lactate production. In addition to LDH isoforms, which generate lactate, the monocarboxylate transporters (MCT1-4) are key transmembrane proteins that regulate the import/export of lactate (29). While MCT1-4 all have similar structure and function, MCT1 and MCT4 are differentially upregulated in various tumor types and have garnered the most attention in cancer (30). Therefore, we examined the expression of MCT1 and MCT4 in lung adenocarcinomas. IHC analysis revealed that MCT4, but not MCT1, was overexpressed in lung adenocarcinomas ( Figure 5C). Similar to LDHB, MCT4 expression was associated with RAS pathway activation  Figure 5D). We suggest that KRAS mutant adenocarcinomas have upregulated components of the lactate machinery to accommodate an increased dependence on glycolysis.

LDHB expression in lung adenocarcinomas correlates with poor clinical outcome
We next examined whether the expression of LDHB in lung cancer correlated with clinical outcome. We scored and analyzed LDHB protein levels by IHC on a cohort of 383 lung adenocarcinomas ( Figure 6A, B), of which nearly half (n = 170) had associated survival data (Supplementary Table 3). In Kaplan-Meyer analysis, five-year patient survival was significantly lower in LDHB high (2 or 3 staining) as compared to LDHB low (0 or 1) tumors (33% versus 48%, log-rank p = 0.005). When we assessed the independent prognostic effect of LDHB using a multivariate Cox model that adjusted for additional predictors of survival, such as sex, age and tumor grade, high LDHB expression was associated with reduced five-year survival (Hazard ratio = 1.67, 95% confidence interval (1.06, 2.62), p = 0.027) ( Figure 6C, D and Supplementary Figure 5). Since we observed that LDHB is upregulated in a subset of cancers containing other oncogenic drivers, we also examined if LDHB expression is predictive of lifespan in KRAS wild-type tumors using a microarray dataset that contained associated outcome data.
Interestingly, KRAS wild-type lung tumors with high expression of LDHB mRNA trended with a poorer clinical outcome (Supplementary Figure 5C). These data indicate that LDHB is upregulated in a significant fraction of lung adenocarcinomas and is associated with poor patient outcome. Research.
on August 13, 2021. © 2012 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

DISCUSSION
In this study, we characterized the KRAS amplicon at 12p12 and found 18 genes that are co-amplified with KRAS in NSCLC tumors. Using a loss of function RNAi screen, we subsequently knocked down all 18 genes and identified LDHB as an essential regulator of lung cancer. LDHB was upregulated and required for the growth of KRAS mutant and amplified lung cancers both in vitro and in vivo. Gene expression and functional analyses indicate that KRAS mutant lung tumors, compared to KRAS wild-type tumors, have a greater predilection for using glycolysis for their energy demands. Consistent with such a role, we find that in addition to LDHB, the lactate transporter MCT4 is also upregulated in KRAS mutant lung cancer, suggesting a more global shift in metabolic requirements upon KRAS mutation. Finally, we show that LDHB overexpression in patients with lung adenocarcinoma have a poorer prognosis.
Carcinogenesis is a complicated, multi-step process that involves combinatorial alterations in oncogenes, tumor suppressors, and metabolic pathways. It is important to note that these alterations in intracellular signaling pathways are a type of evolution that conforms to the changing tumor micro-environment and bioenergetic requirements that allows uncontrolled cell growth. Since Warburg's observation over 50 years ago, in which tumor cells shift from oxidative phosphorylation to aerobic glycolysis, we are only now beginning to understand the molecular details of this transformation.
In the last decade there has been accumulating evidence that oncogenes, such as PI 3- and stimulating phosphofructokinase activity (32). Likewise, genes involved in the metabolism of glutamine, which tumor cells need for the production of various metabolic intermediates, are downstream targets of the MYC oncogene (33,34). Direct regulatory inputs on metabolism are not restricted to oncogenes, as deletion of tumor suppressors, such as p53 and PTEN, also contribute to the metabolic shift (31). Taken together, the concerted upregulation of oncogenes and downregulation of tumor suppressors directly alters the transcriptional controls of the metabolic circuitry.
The direct role of KRAS in regulating tumor metabolism is less well defined. Tumors with oncogenic RAS correlate with numerous metabolic aberrations, including increased consumption of glucose and glutamine, increased production of lactic acid, altered expression of mitochondrial genes, and reduced mitochondrial activity (35)(36)(37)(38). Our findings that a glycolysis gene signature is specifically upregulated in KRAS mutant lung tumors and that KRAS mutant cells are more sensitive to the glycolysis inhibitor 2DG, are consistent with such changes in tumor metabolism. Moreover, a recent report demonstrated that KRAS can also regulate glycolysis and glucose metabolism in a pancreatic cancer mouse model (39). Our observations, however, show that in other cancer types, such as colon cancer, KRAS mutant tumors do not display elevated dependence on glycolysis. Cumulatively, these studies suggest that examining the effect of KRAS on cellular metabolism should be conducted in a tumor-specific context. Lactate dehydrogenase enzymes have important roles in aerobic glycolysis, as they are required to convert the surplus of pyruvate generated by a high rate of glycolysis into lactate (25).
The LDHA isoform has received the most attention because it is ubiquitously expressed in tumors and is a downstream target of HIF1. While an essential function of LDHA in tumor metabolism has been described (40), the role of LDHB is less clear. One report has shown that Research.
on August 13, 2021. © 2012 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
the mTOR pathway controls LDHB expression, while another report has shown that LDHB transcription is shut off by promoter methylation in bladder, colon, and prostate cancers (41-44).
In the context of lung adenocarcinoma, we observed that LDHB is upregulated in KRAS aberrant tumors and in a subset of cell lines containing other oncogenic drivers. Importantly, LDHB is required to maintain cell growth in this context. While our data does not support that LDHB transcription is downstream of KRAS signaling, we hypothesize that a pathway functioning in parallel with KRAS activity might drive LDHB expression. In normal tissues, LDHB is expressed in liver, red blood cells, kidney, and heart raising concern that its inhibition may have deleterious consequences (45,46). Intriguingly, however, patients with complete loss of LDHB expression due to a hereditary recessive trait have no significant phenotypic impairment (47,48). This suggests that an LDHB small molecule inhibitor may have minimal off-target affects in normal tissue. Future studies are necessary to identify signaling pathways that act upstream of LDHB.
Our data showing that LDHB inhibition reduced cell growth, even in the presence of LDHA, raises an important question regarding the biochemical activity of the LDHB isoform.
Since the LDH enzymes function as both homo-and hetero-tetramers (46), it is possible that reducing LDHB somehow affects LDHA activity and the forward reaction. However, an alternative explanation is that LDHB has an exclusive function from LDHA. While biochemical data indicates that LDHA and LDHB can run in both the forward and reverse direction, LDHB has a biochemical predilection for the reverse direction (45). Perhaps LDHB functions in certain contexts to oppose LDHA or act as a governor to LDHA function. In this scenario LDHB would at the very least slow down the conversion of pyruvate to lactate. Future work is necessary to test these hypotheses, but is beyond the scope of this work.   lung adenocarcimoma tumors with low (n = 97) or high (n = 129) LDHB expression (below or above mean LDHB expression across all tumors, respectively). P-value is a Student's t-test.    with high LDHB expression (n = 99; IHC score 2 or 3) had a worse overall survival compared to patients with low LDHB expression (n = 71; IHC score 0 or 1) for a 5-year survival (C) or overall survival, as measured by death or being censored from the study (D). Research.