Metabolic enzyme LDHA activates Rac1 GTPase as a noncanonical mechanism to promote cancer

Abstract

The glycolytic enzyme lactate dehydrogenase A (LDHA) is frequently overexpressed in cancer, which promotes glycolysis and cancer. The oncogenic effect of LDHA has been attributed to its glycolytic enzyme activity. Here, we report an unexpected noncanonical oncogenic mechanism of LDHA; LDHA activates small GTPase Rac1 to promote cancer independently of its glycolytic enzyme activity. Mechanistically, LDHA interacts with the active form of Rac1, Rac1-GTP, to inhibit Rac1-GTP interaction with its negative regulator, GTPase-activating proteins, leading to Rac1 activation in cancer cells and mouse tissues. In clinical breast cancer specimens, LDHA overexpression is associated with higher Rac1 activity. Rac1 inhibition significantly suppresses the oncogenic effect of LDHA. Combination inhibition of LDHA enzyme activity and Rac1 activity by small-molecule inhibitors displays a synergistic inhibitory effect on breast cancers with LDHA overexpression. These results reveal a critical oncogenic mechanism of LDHA and suggest a promising therapeutic strategy for breast cancers with LDHA overexpression.

Editor summary:

Liu et al. identify a non-metabolic mechanism through which lactate dehydrogenase A (LDHA) promotes cancer progression. This study shows that LDHA, independently of its enzymatic activity, directly interacts with and activates Rac1

Introduction

Metabolic reprogramming is a hallmark of cancer cells and a key contributor to cancer progression ^1–3. The Warburg effect is the most well-known metabolic change in cancer; most cancer cells are reprogrammed to highly depend on glycolysis for increased demands of energy and biomolecules to support their rapid growth and proliferation ^1–3. The enhanced glycolysis plays a critical role in promoting cancer progression, which is a promising target for cancer therapies ^1–3. As a key glycolytic enzyme, LDHA converts pyruvate to lactate in the final step of glycolysis ^{2, 3}. LDHA is frequently overexpressed in cancers, including breast, lung, prostate, pancreatic, brain, and liver cancer ^4–11. LDHA overexpression plays a critical role in promoting cancer progression, and inhibition of LDHA suppresses cancer progression ^{6–8, 11–18}. LDHA expression can be transcriptionally induced by different oncoproteins, including HIF-1α and c-Myc ^{11, 19}, and can also be upregulated through post-transcriptional and post-translational regulations in cancer ^{8, 20}. Clinical studies have shown that high LDHA expression in different types of cancers, including breast cancer, is associated with enhanced metastasis and poor clinical outcomes in cancer patients ^{4, 5, 9, 10, 21, 22}. Therefore, LDHA has been regarded as a biomarker and is being actively tested as an attractive therapeutic target for many cancers.

As a glycolytic enzyme, the oncogenic function of LDHA has been attributed to its function in promoting glycolysis ^{6, 7, 14–17, 22–24}. Different small-molecule LDHA inhibitors have been developed to target the lactate dehydrogenase (LDH) enzyme activity of LDHA to inhibit glycolysis and suppress tumor progression ^{21, 22, 24}. However, existing LDHA inhibitors have shown limited clinical efficacy, and currently, there are no LDHA inhibitors available for clinical cancer therapies ^{21, 22, 24}. Interestingly, recent studies have shown that some metabolic enzymes display noncanonical functions to promote cancer progression in addition to their well-known roles in metabolism ²⁵. For instance, PFKFB4 phosphorylates and activates steroid receptor coactivator-3 (SRC3) to promote breast cancer ²⁶. PFK1 binds to TEADs and activates YAP/TAZ transcription to promote breast cancer ²⁷. Understanding these noncanonical mechanisms of metabolic enzymes can deepen our understanding of cancer metabolic reprograming and lead to the identification of novel therapeutic vulnerabilities in cancer.

Rac1 is a small GTPase, which exists in two conformational states in cells, the active GTP-bound (Rac1-GTP) and inactive GDP-bound states (Rac1-GDP) ^{28, 29}. Guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs) play critical roles in regulation of Rac1 transition between these two states ^{28, 29}. GEFs promote Rac1 activation by facilitating the exchange of GDP for GTP. Once activated, Rac1-GTP regulates various cellular events including cytoskeleton organization, cell cycle and cell motility. In contrast, GAPs inhibit Rac1 activity by promoting GTP hydrolysis to convert Rac1-GTP to Rac1-GDP ^{28, 29}. In addition, many other proteins and mechanisms have been reported to regulate Rac1 activity ^28–31. Deregulation of these mechanisms frequently leads to the aberrant Rac1 activation in cancer, which in turn promotes proliferation, metastasis and therapeutic resistance of cancer cells ^28–30.

In this study, we found that LDHA binds to Rac1-GTP and inhibits its interaction with Rac1 GAPs, resulting in Rac1 activation independently of LDHA glycolytic enzyme activity to promote tumorigenesis. Importantly, blocking LDHA enzyme activity and Rac1 activity simultaneously displayed a synergistic inhibitory effect on breast tumors with LDHA overexpression, suggesting a promising therapeutic strategy for breast cancers with LDHA overexpression.

Results

Rac1 is a LDHA-interacting protein.

LDHA overexpression is frequently observed in breast cancer and is associated with poor clinical outcomes of breast cancer patients ^{5, 10, 22, 24}. The meta-analysis of different cohorts of breast cancer specimens (total n=3057) and normal breast tissues (total n=287) from Oncomine database (http://www.oncomine.org) ³² showed that breast cancer tissues had significantly higher LDHA mRNA expression in 19 out of 35 studies (Supplementary Fig. 1). Analysis of the breast cancer dataset from the Cancer Genome Atlas (TCGA) showed that LDHA was frequently overexpressed in breast cancers; 50% of breast cancer specimens showed increased LDHA mRNA expression (by >1.5-fold) compared with their paired adjacent non-tumor tissues (n=57 out of 114; Fig. 1a). Furthermore, LDHA overexpression was observed in different breast tumor subtypes in terms of the estrogen receptor (ER), progesterone receptor (PR) or HER2 status, and not linked to any specific subtypes in these specimens (Extended data Fig. 1a–c). Similarly, LDHA protein levels were significantly higher in breast cancer samples (n=120) compared with non-tumor breast tissues (n=48) in a tissue microarray (TMA) as examined by immunohistochemistry (IHC) staining (Fig. 1b). Data from Kaplan-Meier plotter, an online cancer survival analysis tool (http://kmplot.com) ³³, showed that high LDHA mRNA expression in breast cancer was significantly associated with poor prognosis of cancer patients (n=4929; Fig. 1c), which was observed in different subtypes of breast tumors (Extended data Fig. 1d–f).

Figure 1. LDHA directly interacts with Rac1.

a, Higher LDHA mRNA levels in human breast cancer specimens compared with matched adjacent non-tumor breast tissues (n=114). The data were obtained from TCGA. P<0.0001, two-tailed paired Student’s t test. b, Higher LDHA protein expression in breast cancer specimens (n=120) compared with non-tumor breast tissues (n=48) analyzed by IHC. Left panels: Representative IHC staining images of LDHA in a human breast TMA (TMA-BR2082a; US Biomax). Scale bar: 20μm. Data represent mean ± SE. P<0.0001, two-tailed unpaired Student’s t test. c, High LDHA mRNA expression is associated with poor relapse-free survival in breast cancer patients. The data were obtained from Kaplan-Meier plotter. The P value was analyzed by the log-rank (Mantel-Cox) test. d, LDHA-Flag interacted with Myc-Rac1 in cells. Hs578T cells expressing LDHA-Flag and Myc-Rac1 were employed for co-IP assays using the anti-Myc and anti-Flag antibodies, respectively, followed by western-blot analysis. IP: immunoprecipitation; IB: Immunoblotting. e, Co-IP analysis of the interaction between endogenous LDHA and Rac1 in wild-type (WT) Hs578T and SK-BR3 cells as well as corresponding cells with LDHA knockout (KO) by CRISPR/Cas9. Two different LDHA KO clonal lines were used. f, Co-IP analysis of the interaction between endogenous LDHA and Rac1 in Hs578T cells with transduction of a control (Con) or two different lentiviral shRNA vectors against LDHA. g, The L4 region is required and sufficient for LDHA to interact with Rac1. Hs578T cells expressing Myc-Rac1 and WT or different deletion mutants of LDHA-Flag were used for co-IP assays. Upper panels: schematic representation of vectors expressing serial deletion mutants of LDHA-Flag. h, Co-IP analysis of the interaction between L4 LDHA-Flag and endogenous Rac1 in Hs578T and SK-BR3 cells expressing L4 LDHA-Flag. i, The direct interaction between LDHA and Rac1 proteins analyzed by in vitro GST pull-down assays using purified recombinant His-Rac1 and WT or mutant GST-LDHA proteins. j, The in situ interaction between endogenous LDHA and Rac1 proteins in WT and LDHA KO Hs578T and SK-BR3 cells analyzed by the proximity ligation assay (PLA). Red: PLA signals of the LDHA-Rac1 interaction in cells. Scale bar: 10 μm. In d–j, data represent three repeats with similar results.

To deepen our understanding of oncogenic mechanisms of LDHA, we screened for LDHA-interacting proteins in human breast cancer Hs578T cells with or without expression of LDHA-Flag-HA by co-IP assays followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. Through this strategy, Rac1 was identified as a potential LDHA-interacting protein (Supplementary Table 1). The interaction between ectopic LDHA-Flag and Myc-Rac1 was confirmed by co-IP followed by western-blot analysis in Hs578T cells co-expressing LDHA-Flag and Myc-Rac1 (Fig. 1d). The interaction between endogenous LDHA and Rac1 was observed in control LDHA wild-type (WT) human breast cancer cells, including Hs578T and SK-BR3 cells, but not in cells with LDHA knockout (KO) by the CRISPR/Cas9 system (Fig. 1e). Similarly, the interaction between endogenous LDHA and Rac1 was observed in Hs578T cells but was largely abolished in Hs578T cells with LDHA knockdown by shRNA vectors (Fig. 1f).

To determine the domain of LDHA required for its interaction with Rac1, a series of Flag-tagged deletion mutants of LDHA were constructed (Fig. 1g). Co-IP assays showed that while L1, L2 and L4 (amino acids 21–163 that contains the coenzyme binding domain) mutant LDHA interacted with Myc-Rac1, the L3 LDHA, which lacks the L4 region, did not interact with Myc-Rac1, indicating that the L4 region is required and sufficient for LDHA to interact with Rac1 (Fig. 1g). The interaction between L4 LDHA-Flag and endogenous Rac1 was confirmed in Hs578T and SK-BR3 cells expressing L4-LDHA-Flag (Fig. 1h).

We further examined whether LDHA directly interacts with Rac1 by in vitro glutathione S-transferase (GST) pull-down assays using purified GST-LDHA and His-Rac1 recombinant proteins. WT GST-LDHA directly interacted with His-Rac1, and furthermore, L4 but not L3 GST-LDHA protein directly interacted with His-Rac1 in vitro (Fig. 1i). The interaction between endogenous LDHA and Rac1 in cells was further confirmed in Hs578T and SK-BR3 cells by the proximity ligation assay (PLA), which permits the detection of protein-protein interactions in situ at endogenous protein levels in cells ³⁴; the interaction signal was observed in WT but not LDHA KO cells (Fig. 1j). Collectively, these results indicate that LDHA directly interacts with Rac1 in cells.

LDHA activates Rac1 in human breast cancer cells.

Given that LDHA interacts with Rac1, we tested whether LDHA regulates Rac1 activity or levels in cells. The Rac1 activity was determined by analysis of the ratio of the active Rac1-GTP compared with the total Rac1 in cells ^{31, 35, 36}. The Rac1 binding domain of PAK1, a major downstream effector protein of Rac1 that specifically binds to Rac1-GTP but not Rac1-GDP, was used to pull down Rac1-GTP in cells ^{31, 35, 36}. Breast cancer cell lines of different subtypes were employed for assays, including Hs578T (ER−/PR−/HER2−), BT-549 (ER−/PR−/HER2−), SK-BR3 (ER−/PR−/HER2+), MCF7 (ER+/PR+/HER2−) and ZR-75–1 (ER+/PR+/HER2+), all of which displayed much higher levels of LDHA protein compared with normal breast tissues (Fig. 2a). Notably, expression of LDHA-Flag significantly increased levels of Rac1-GTP but not total Rac1 protein in these cells (Fig. 2b), suggesting that LDHA enhances Rac1 activity. Furthermore, knockdown of LDHA by shRNA significantly reduced levels of Rac1-GTP but not total Rac1 protein in these cells (Fig. 2c). Rac1 activation by LDHA was confirmed in Hs578T and SK-BR3 cells with LDHA KO; compared with control WT cells, LDHA KO cells displayed significantly lower levels of Rac1-GTP (Fig. 2d). It has been reported that the binding of Rac1-GTP to PAK1 results in PAK1 auto-phosphorylation at multiple sites, including Ser199/204, which activates PAK1 to mediate Rac1-GTP functions ^{28, 37}. LDHA-Flag expression in breast cancer cells enhanced PAK1 Ser199/204 phosphorylation (Fig. 2b), whereas LDHA knockdown or KO reduced PAK1 Ser199/204 phosphorylation (Fig. 2c & d), further indicating that LDHA enhances Rac1 activity.

Figure 2. LDHA activates Rac1 in breast cancer cells and mouse mammary tissues.

a, LDHA expression in normal breast tissues and different human breast cancer cell lines examined by western-blot assays. b, Ectopic expression of WT LDHA-Flag enhanced Rac1 activities represented by increased levels of Rac1-GTP and p-PAK1 (Ser199/204) in different breast cancer cell lines. c, Knockdown of endogenous LDHA by two different shRNA vectors reduced Rac1 activities represented by decreased Rac1-GTP and p-PAK1 (Ser199/204) levels in breast cancer cell lines. d, Knockout (KO) of LDHA by CRISPR/Cas9 decreased the levels of Rac1-GTP and p-PAK1 (Ser199/204) in Hs578T and SK-BR3 cells. e, Expression of WT and L4 LDHA-Flag but not L3 LDHA-Flag enhanced the Rac1 activity in Hs578T and SK-BR3 cells. b-e: Left panels: represented results of Rac1 activity. Right panels: relative Rac1-GTP/total Rac1/Actin levels. f, g, L4 LDHA lacked the lactate dehydrogenase (LDH) enzyme activity. In f, Hs578T and SK-BR3 cells transduced with control, WT or L4 LDHA-Flag vectors were used for LDH activity assays. In g, purified WT and L4 GST-LDHA recombinant proteins as well as GST only protein were used for LDH activity assays. h, The interaction between endogenous LDHA and Rac1 in mammary tissues of R26-Cre-ERT2, LDHA^flox/flox mice detected by co-IP assays. LDHA was deleted by Tamoxifen treatment in mice. i, LDHA deletion reduced levels of Rac1-GTP and p-PAK1 (Ser198/203) in different tissues in R26-Cre-ERT2, LDHA^flox/flox mice. Left panels: represented results of the Rac1 activity assays in mammary tissues. Right panel: relative Rac1-GTP/total Rac1/Actin levels in mammary, liver and spleen tissues. In a, h, data represent three repeats with similar results. In b-g, i, n = 3 independent experiments (b–e, and g) or n=6 independent experiments (f, i), two-tailed unpaired Student’s t-test (b, i) or one-way ANOVA followed by Dunnett’s test (c-e) or Tukey’s test (f, g). Data represent mean ± SD. *: P<0.0001.

We also examined whether L4 LDHA activates Rac1. Consistent with the effect of WT LDHA, L4 but not L3 LDHA expression enhanced Rac1 activity (but not total Rac1 protein levels) and PAK1 Ser199/204 phosphorylation levels in cells (Fig. 2e). As expected, expression of WT but not L4 LDHA-Flag in cells significantly enhanced LDH enzyme activity analyzed by using a LDH activity assay kit (Fig. 2f). This result was confirmed by using recombinant WT and L4 GST-LDHA proteins for assays; WT but not L4 GST-LDHA protein displayed LDH enzyme activity (Fig. 2g). These results together demonstrate that LDHA activates Rac1 independently of LDHA glycolytic enzyme activity.

LDHA deletion reduces Rac1 activity in mouse tissues.

Both mouse LDHA and Rac1 proteins share high (≥97%) amino acid homology with human LDHA and Rac1 proteins, respectively (Supplementary Fig. 2). To investigate whether LDHA regulates Rac1 activity under physiological conditions, the conditional LDHA knockout mice (R26-Cre-ERT2, LDHA^flox/flox) were treated with Tamoxifen to delete LDHA. Co-IP assays showed the interaction between endogenous LDHA and Rac1 in mammary tissues of WT but not LDHA-deficient mice (Fig. 2h). Compared with WT mice, mammary tissues from LDHA-deficient mice displayed significantly lower levels of Rac1-GTP (but not total Rac1) and reduced PAK1 Ser198/203 phosphorylation (equivalent to human PAK1 Ser199/204) (Fig. 2i). Similar results were observed in other tissues, including the liver and spleen (Fig. 2i). These results indicate that the regulation of Rac1 activity by LDHA is a conserved function in both mice and human beings.

LDHA inhibits Rac1-GAPs interaction to activate Rac1.

Rac1 activity is tightly regulated by GEFs and GAPs; GEFs specifically bind to Rac1-GDP to activate Rac1, whereas GAPs bind to Rac1-GTP to inactivate Rac1(Fig. 3a) ^{28, 37}. The Switch I & II region of Rac1 is required for Rac1 to interact with GEFs and GAPs ^{28, 37}. To investigate which domain is required for Rac1 to interact with LDHA, different Myc-tagged deletion mutants of Rac1 were constructed, including ΔC33, ΔN29, and ΔSwitch (Fig. 3b). Co-IP assays showed that ΔC33 and ΔN29 but not ΔSwitch (deletion of Switch I & II) Rac1 interacted with LDHA-Flag (Fig. 3b), indicating that the Switch region is required for Rac1-LDHA interaction.

Figure 3. LDHA binds to Rac1-GTP and inhibits its binding with Rac1 GAPs to activate Rac1.

a, The regulation of Rac1 activity by GAPs and GEFs. b, The Switch I & II region is required for Rac1 to bind to LDHA. Hs578T cells expressing LDHA-Flag and WT or different deletion mutants of Myc-Rac1 were used for co-IP assays. Left panels: schematic representation of vectors expressing serial deletion mutants of Myc-Rac1. c, LDHA preferentially bound to the active Rac1-GTP form but not the inactive Rac1-GDP form in cell lysates. Lysates of Hs578T cells expressing Myc-Rac1 and LDHA-Flag were pretreated with GDP and GTPγS to convert Rac1 into Rac1-GDP and Rac1-GTP form, respectively, before co-IP assays. d, LDHA preferentially bound to the constitutively active Rac1-G12V but not the inactive dominant negative Rac1-T17N in cells. e, LDHA inhibited the interaction between Myc-Rac1 and HA-RACGAP1 or HA-FilGAP in cells. Hs578T cells were transfected with Myc-Rac1 vectors and HA-RACGAP1 (left) or HA-FilGAP vectors (right), together with different amounts of LDHA-Flag vectors (0, 1 and 3 μg) for co-IP assays. f, LDHA inhibited the interaction between endogenous Rac1 with RACGAP1 or FilGAP in cells. Hs578T cells were transfected with LDHA-Flag vectors (0, 1, 2, and 4 μg) for co-IP assays. g, LDHA KO promoted the interaction between endogenous Rac1 and RACGAP1 or FilGAP in Hs578T and SK-BR3 cells as analyzed by co-IP assays. h, L4 LDHA inhibited the interaction between endogenous Rac1 and RACGAP1 or FilGAP in cells. Hs578T cells were transfected with different amounts of L4 LDHA-Flag vectors (0, 1, 2, and 4 μg) for co-IP assays. i, A schematic model depicting that LDHA interacts with Rac1-GTP and inhibits the binding of Rac1 GAPs with Rac1-GTP to activate Rac1. In b–h, data represent three repeats with similar results.

To determine the mechanism underlying Rac1 activation by LDHA, we compared the binding of LDHA with Rac1-GTP and Rac1-GDP. To this end, lysates from Hs578T cells expressing LDHA-Flag and Myc-Rac1 were pretreated with GDP or GTPγS (a non-hydrolyzable GTP analog) to convert Rac1 into Rac1-GDP or Rac1-GTP form in cell lysates before co-IP assays ^{35, 38}. Interestingly, LDHA-Flag preferentially bound to Myc-Rac1 in cell lysates pretreated with GTPγS but not GDP, suggesting that LDHA preferentially interacts with Rac1-GTP (Fig. 3c). This result was confirmed by co-IP analysis of the interaction between LDHA-Flag and Rac1-G12V or Rac1-T17N. Rac1-G12V is a constitutively active mutant existing constitutively in Rac1-GTP form, whereas Rac1-T17N is a dominant negative mutant existing constitutively in the Rac1-GDP form ^{28, 39}. Consistently, LDHA-Flag preferentially bound to Myc-Rac1-G12V in cells (Fig. 3d). We further tested the interaction of LDHA with Rac1-T35S mutant, which contains a mutation in the Switch I domain and has a reduced affinity to Rac1 effector proteins ^{40, 41}, and found that LDHA displayed a much weaker interaction with Rac1-T35S compared with WT Rac1 (Extended Data Fig. 2).

The finding that LDHA specifically binds to Rac1-GTP raised a possibility that LDHA activates Rac1 through competing with Rac1 negative regulator GAPs for Rac1-GTP binding. To test this possibility, Hs578T cells were transfected with Myc-Rac1 vectors and different amounts of LDHA-Flag vectors together with vectors expressing two HA-tagged Rac1 GAPs, RACGAP1 and ARHGAP24 (also known as FilGAP), which bind to Rac1 Switch I & II region to inactivate Rac1 ^{28, 39}. Co-IP assays showed that LDHA-Flag inhibited the binding of HA-RACGAP1 and HA-FilGAP with Myc-Rac1 (Fig. 3e), and also the interactions of endogenous RACGAP1 and FilGAP with Rac1 in cells (Fig. 3f). In contrast, LDHA KO promoted the interactions of endogenous RACGAP1 and FilGAP with Rac1 in cells (Fig. 3g). Like WT LDHA-Flag, L4 LDHA-Flag inhibited the binding of RACGAP1 and FilGAP with Rac1 in cells (Fig. 3h). These results demonstrate that LDHA interacts with Rac1 to block Rac1-GAPs interaction, leading to Rac1 activation (Fig. 3i).

Rac1 contributes to LDHA oncogenic effect in cancer cells.

Rac1 plays a critical role in promoting proliferation, migration and invasion of cancer cells ^{28, 37}. We investigated whether LDHA overexpression promotes proliferation, migration and invasion of breast cancer cells through Rac1 activation in addition to promoting glycolysis. Different breast cancer cells expressing WT or L4 LDHA-Flag were employed for colony formation and transwell assays. Expression of WT or L4 LDHA significantly promoted colony formation (Fig. 4a), migration (Fig. 4b) and invasion (Fig. 4c) of these cells. Further, WT LDHA displayed a more pronounced promoting effect than L4 LDHA (Fig. 4a–c). Notably, Rac1 knockdown by shRNA (the knockdown was shown in Supplementary Fig. 3a, b) clearly reduced the promoting effect of WT LDHA on cells and largely abolished the promoting effect of L4 LDHA (Fig. 4a–c), suggesting that L4 LDHA enhances cell proliferation, migration and invasion mainly through Rac1 activation. Further, expression of the dominant negative Rac1-T17N significantly reduced the promoting effect of WT LDHA and largely abolished the promoting effect of L4 LDHA on cells (Extended Data Fig. 3a–c). Similar results were observed in LDHA KO Hs578T and SK-BR3 cells that were transduced with retroviral vectors to stably express WT or L4 LDHA-Flag to a comparable level with that of endogenous LDHA in control cells (Fig. 4d). Compared with WT cells, LDHA KO significantly inhibited colony formation, migration and invasion of cells, which were fully rescued by WT LDHA-Flag and partially but significantly rescued by L4 LDHA-Flag (Fig. 4d–f). These results suggest that Rac1 activation is an important mechanism contributing to the oncogenic effect of LDHA.

Figure 4. LDHA activates Rac1 to promote colony formation, migration and invasion of breast cancer cells.

a, Knockdown of endogenous Rac1 significantly attenuated the promoting effect of WT LDHA on colony formation and largely abolished the promoting effect of L4 LDHA on colony formation in different breast cancer cells. Left panel: representative images of colony formation of Hs578T cells. Right panel: quantification of colonies formed by different breast cancer cells. Cells with or without Rac1 knockdown by two different shRNA vectors were transduced with WT or L4 LDHA-Flag retroviral vectors for colony formation assays. b, c, Knockdown of Rac1 significantly attenuated the promoting effect of WT LDHA-Flag on cell migration (b) and invasion (c), and largely abolished the promoting effect of L4 LDHA-Flag on cell migration (b) and invasion (c) as analyzed by transwell assays. Left panels: representative images of migrating (b) and invading (c) Hs578T cells. Scale bar: 200μm. d, The impaired ability of colony formation of Hs578T and SK-BR3 cells resulted from LDHA KO was fully rescued by stable ectopic expression of WT LDHA-Flag and partially rescued by expression of L4 LDHA-Flag in cells. Left panels: The level of endogenous LDHA protein in control cells and the level of WT or L4 LDHA-Flag protein in LDHA KO cells analyzed by western-blot assays. Endo LDHA: endogenous LDHA. e, f, The impaired abilities of migration (e) and invasion (f) of Hs578T and SK-BR3 cells resulted from LDHA KO were fully rescued by ectopic expression of WT LDHA-Flag and partially rescued by L4 LDHA-Flag. In a-f: data represent mean ± SD (n=6 independent experiments), two-way ANOVA followed by Tukey’s test (a–c) or one-way ANOVA followed by Tukey’s test (d–f). *: P<0.0001.

PAK1 is an important downstream effector protein of Rac1 ^{28, 37}. While PAK1 knockdown by shRNA significantly reduced the promoting effect of WT and L4 LDHA on colony formation, migration and invasion of Hs578T and SK-BR3 cells, PAK1 knockdown displayed a less pronounced effect on cells than Rac1 knockdown (Supplementary Fig. 4a–d). Notably, Rac1 knockdown but not PAK1 knockdown largely abolished the promoting effect of L4 LDHA on cells (Supplementary Fig. 4b–d). These results suggest that while PAK1 plays a major role in mediating the LDHA/Rac1 signaling, other Rac1 effector proteins may partially mediate the LDHA/Rac1 signaling, which deserves future studies.

Invadopodia are actin-based protrusions of the plasma membrane that penetrate into and degrade the extracellular matrix (ECM), which are important for invasion and metastasis of cancer cells ^{42, 43}. Rac1 was reported to be involved in invadopodia regulation but not physically localized in invadopodia in some cell lines ^{44, 45}. We examined whether LDHA and Rac1 complex is localized in invadopodia in breast cancer cells. Co-localization of invadopodia markers Tks5 and F-actin in a punctate manner in cells is an indication of invadopodia formation ^{42, 43, 46}. In Hs578T and MDA-MB231 cells, two breast cancer cell lines widely used for invadopodia assays ^{47, 48}, we found that LDHA but not Rac1 was co-localized with Tks5 and F-actin in invadopodia, indicating that LDHA but not Rac1 is localized in invadopodia in these cells (Extended Data Fig. 4), which is consistent with previous reports using other cell lines ^{44, 45, 49}.

Lactate, the metabolite of LDHA in cells, plays an important role in tumorigenesis ^{3, 14}. Our results showed that expression of WT but not L4 LDHA significantly promoted lactate production in breast cancer cells (Supplementary Fig. 5a). Further, expression of WT or T17N Rac1 did not clearly affect lactate production in cells, and neither did Rac1 knockdown by shRNA or treating cells with small-molecule Rac1 inhibitor NSC23766 ^{29, 50} (Supplementary Fig. 5b–d). To further examine whether glycolysis contributes to the oncogenic effect of L4 LDHA, Hs578T and SK-BR3 cells were treated with 2-Deoxy-d-glucose (2-DG), a glucose analog that blocks glycolysis ⁵¹. 2-DG clearly affected the promoting effect of WT but not L4 LDHA on colony formation, migration and invasion of cells (by comparing the fold change of WT and L4 LDHA in the control and 2-DG treatment groups) (Supplementary Fig. 6a). Galactose enters glycolysis through the Leloir pathway, which occurs at a much lower rate than glucose entry into glycolysis, and therefore, galactose inhibits glycolysis ^52–54. To inhibit glycolysis, cells were cultured in the medium containing galactose instead of glucose. Galactose clearly reduced the promoting effect of WT but not L4 LDHA on colony formation, migration and invasion of cells (Supplementary Fig. 6b). These results suggest that promoting glycolysis is an important oncogenic mechanism for WT LDHA but not L4 LDHA.

Given that LDHA binds to Rac1 at its Switch domain, it raises the possibility that LDHA binds to and activates some other Rho family proteins, including RhoA, Cdc42 and Rac3, whose Switch domains share high amino acid homology with Rac1 ^{28, 37}. Co-IP assays showed that LDHA-Flag interacted with Myc-Cdc42 and Myc-Rac3 but not Myc-RhoA (Extended Data Fig. 5a). Compared with Rac1, LDHA displayed much weaker interactions with Cdc42 and Rac3, and activated them at a much less extent (Extended Data Fig. 5a, b). Knockdown of Cdc42 or Rac3 by shRNA reduced colony formation, migration and invasion of breast cancer cells, but at a much less extent compared with Rac1 knockdown. Notably, knockdown of Cdc42 or Rac3 did not clearly affect the promoting effect of L4 LDHA on cells (Extended Data Fig. 5c, d). Analysis of the RNA-Seq data from Cancer Cell Line Encyclopedia (CCLE; https://sites.broadinstitute.org/ccle/) showed that Rac1 expression levels were much higher than the levels of Cdc42 and Rac3 in majority of breast cancer cell lines (Extended Data Fig. 6). These results suggest that Rac1 plays a major role in mediating the oncogenic effect of LDHA in breast cancer cells.

Rac1 contributes to LDHA oncogenic effect in vivo.

To investigate whether Rac1 activation contributes to LDHA function in promoting breast tumorigenesis, orthotopic breast tumor models in immunocompromised nude mice and syngeneic breast tumor models in immunocompetent BALB/c mice were employed. Expression of WT or L4 LDHA-Flag in Hs578T cells significantly promoted the growth of orthotopic breast tumors formed by Hs578T cells (Fig. 5a). Compared with L4 LDHA, WT LDHA showed a more pronounced effect on tumor growth (Fig. 5a). Expression of WT or L4 LDHA enhanced Rac1 activity in tumors as shown by increased levels of Rac1-GTP and p-PAK1 (ser199/204) in tumors (Fig. 5b & c). Expression of WT and L4 LDHA also significantly enhanced the percentage of cells with positive IHC staining of cell proliferation marker Ki-67 in tumors (Fig. 5c). Further, LDHA KO significantly inhibited growth of orthotopic tumors, which was almost fully rescued by WT LDHA-Flag and partially rescued by L4 LDHA-Flag (Fig. 5d). Notably, Rac1 knockdown in Hs578T cells greatly reduced the promoting effect of WT LDHA-Flag on tumor growth and largely abolished the promoting effect of L4 LDHA-Flag (Fig. 5e).

Figure 5. LDHA activates Rac1 to promote growth and metastasis of breast tumors.

a, WT or L4 LDHA-Flag promoted growth of orthotopic breast tumors formed by Hs578T cells in nude mice. Left panel: representative images of tumors at day 30 after the mammary fat pad implantation. Right panel: Tumor weights at day 30. b, The levels of Rac1-GTP and p-PAK1 (Ser199/204) in Hs578T tumors. c, IHC staining of p-PAK1 (Ser199/204) and Ki-67 in Hs578T tumors. Left panels: representative images. Scale bar: 20 μm. Right panels: percentage of Ki-67 positive cells in tumors. d, The impaired growth of Hs578T orthotopic tumors resulted from LDHA KO was fully rescued by expression of WT LDHA-Flag and partially rescued by L4 LDHA-Flag. Stable cell lines presented in Fig. 4d were used. e, f, Rac1 knockdown attenuated the promoting effect of WT LDHA-Flag on growth of Hs578T orthotopic tumors (e) and 4T1 syngeneic orthotopic tumors (f) and largely abolished the promoting effect of L4 LDHA-Flag on tumor growth. Right panel in f: the levels of Rac1-GTP in 4T1 tumors. g, h, Rac1 knockdown attenuated the promoting effect of WT LDHA-Flag and largely abolished the promoting effect of L4 LDHA-Flag on lung metastasis of Hs578T cells. Mice were sacrificed at 12 weeks after the tail vein injection. Presented are representative bioluminescent images (left panels in g), normalized photon flux of lung metastases (right panel in g), and representative H&E staining of lung tissues (h) at 12 weeks. In h, arrows indicate metastatic nodules. Scale bar: 200μm. i, The impaired lung metastasis of Hs578T cells resulted from LDHA KO was fully rescued by WT LDHA-Flag and significantly rescued by L4 LDHA-Flag. j, Rac1 knockdown attenuated the promoting effect of WT LDHA-Flag and largely abolished the promoting effect of L4 LDHA-Flag on lung metastasis of syngeneic orthotopic 4T1 tumors. Data represent mean ± SD. In a, d, e, left panel of f, g, i & j, n = 8 mice/group, and in b, c & right panel of f, n=6 tumors/group. Statistical differences were determined by one-way ANOVA followed by Tukey’s test. *: P<0.0001.

Similar results were observed in syngeneic breast tumors in immunocompetent mice established by mammary fat pad implantation of murine breast carcinoma 4T1 cells ⁵⁵. Expression of WT or L4 LDHA-Flag in 4T1 cells, which activates Rac1 (Supplementary Fig. 7a, b), significantly enhanced tumor growth and Rac1 activity in tumors (Fig. 5f). Further, WT LDHA-Flag showed a more pronounced promoting effect on tumor growth than L4 LDHA-Flag (Fig. 5f). Rac1 knockdown by shRNA in 4T1 cells (Supplementary Fig. 7a, c) clearly reduced the promoting effect of WT LDHA-Flag on tumor growth and largely abolished the promoting effect of L4 LDHA-Flag (Fig. 5f).

We further examined the effect of LDHA on metastasis of breast cancer cells to the lung, one of most common metastatic sites for human breast cancer ⁵⁶. Hs578T cells expressing luciferase and WT or L4 LDHA-Flag were injected into nude mice via the tail vein, and lung metastasis was examined by in vivo bioluminescence imaging and histopathological examination of lungs after mice were sacrificed. WT or L4 LDHA-Flag expression significantly promoted lung metastasis while WT LDHA-Flag displayed a more pronounced effect on metastasis than L4 LHDA-Flag (Fig. 5g, h). Notably, Rac1 knockdown in Hs578T cells clearly reduced the promoting effect of WT LDHA on lung metastasis, and largely abolished the effect of L4 LDHA (Fig. 5g, h). Similar results were observed in tail vein injection-induced lung metastasis models using LDHA KO Hs578T cells; LDHA KO significantly inhibited lung metastasis, which was fully rescued by WT LDHA-Flag and partially but significantly rescued by L4 LDHA-Flag (Fig. 5i). These results were confirmed in the spontaneous metastasis model by implanting 4T1 cells into the mammary fat pad; expression of WT or L4 LDHA-Flag in cells significantly promoted lung metastasis, and Rac1 knockdown in cells clearly attenuated the effect of WT LDHA-Flag on lung metastasis and largely abolished the promoting effect of L4 LDHA-Flag (Fig. 5j). Collectively, these results indicate that Rac1 activation plays an important role in mediating the function of LDHA in promoting breast tumor growth and metastasis.

The synergistic effect of LDHA and Rac1 inhibitors on cells.

We next explored whether simultaneous inhibition of LDHA enzyme activity and Rac1 activity by the combination treatment with small-molecule inhibitors displays a better therapeutic effect than the single inhibitor treatment in breast cancer cells with LDHA overexpression. The LDHA inhibitor FX11 is a NADH competitive and selective small-molecule that binds to LDHA to inhibit LDHA enzyme activity, which suppresses glycolysis to exert anti-tumor effects ^{12, 14, 22}. The small-molecule Rac1 inhibitor NSC23766 blocks Rac1 interaction with GEFs to inactivate Rac1 ^{29, 50}. While FX11 or NSC23766 alone significantly inhibited the colony formation of different breast cancer cell lines, the combination treatment with FX11 and NSC23766 displayed a much more pronounced inhibitory effect on colony formation than the single inhibitor treatment (Fig. 6a, b, and Extended Data Fig. 7a). The combination treatment showed a synergistic inhibitory effect on colony formation of these cells (Fig. 6a, b, and Extended Data Fig. 7a). In contrast, LDHA KO in Hs578T and SK-BR3 cells greatly reduced the inhibitory effect of FX11 on colony formation, and abolished the synergistic effect of the combination treatment (Fig. 6a,b).

Figure 6. LDHA and Rac1 small-molecule inhibitors display a synergistic inhibitory effect on colony formation, migration and invasion of breast cancer cells.

a, b, The combination treatment with the LDHA inhibitor FX11 and the Rac1 inhibitor NSC23766 displayed a synergistic inhibitory effect on colony formation of Hs578T (a) and SK-BR3 (b) cells. For colony formation assays, cells with or without LDHA KO were treated with the indicated concentrations of FX11 and/or NSC23766 for 4 days. Combo: FX11 + NSC23766. The combination index (CI) value: <1: synergism; =1: additive effect; >1: antagonism. Data represent mean ± SD (n=6 independent experiments). c, d, The combination treatment displayed a much more pronounced inhibitory effect on migration of Hs578T (c) and SK-BR3 (d) cells than the single inhibitor treatment as analyzed by transwell assays. e, f, The combination treatment displayed a much more pronounced inhibitory effect on invasion of Hs578T (e) and SK-BR3 (f) cells than the single treatment as analyzed by transwell assays. In c–f, Hs578T and SK-BR3 cells with or without LDHA KO were treated with the indicated concentrations of FX11 and/or NSC23766 for 24 h. Data represent mean ± SD (n=6 independent experiments), one-way ANOVA followed by Tukey’s test. *: P<0.0001. NSC: NSC23766.

Transwell assays were performed to assess the inhibitory effect of the combination treatment with FX11 and NSC23766 on migration and invasion of breast cancer cells. While FX11 or NSC23766 alone significantly inhibited the migration and invasion of these cells, the combination treatment displayed a much more pronounced inhibitory effect than the single inhibitor treatment (Fig. 6c–f; Extended Data Fig. 7b,c). LDHA KO significantly reduced the inhibitory effect of FX11 on migration and invasion of cells, and the combination treatment did not display a more significant inhibitory effect on the migration and invasion of LDHA KO cells than the single inhibitor treatment (Fig. 6c–f).

FX11 has been reported to have off-target effects ¹². Here, LDHA KO Hs578T and SK-BR3 cells with stable expression of WT or L4 LDHA-Flag were treated with FX11 and then used for colony formation, migration and invasion assays. FX11 treatment (for 4 days) displayed a significantly less pronounced inhibitory effect on colony formation in cells expressing L4 LDHA than cells expressing WT LDHA (Supplementary Fig. 8a). Similarly, FX11 displayed a much less pronounced inhibitory effect on migration and invasion of breast cancer cells in cells expressing L4 LDHA than cells expressing WT LDHA (Supplementary Fig. 8b,c). Notably, FX11 treatment (for 24 h) for migration and invasion assays did not result in obvious cell death (Supplementary Fig. 8d). These results suggest that the inhibitory effects of FX11 on colony formation, migration and invasion of breast cancer cells are largely on-target effects.

Synergistic effects of LDHA and Rac1 inhibitors on tumors.

We further examined the effect of combination treatment with FX11 and NSC23766 on tumor growth using orthotopic and syngeneic breast tumor models. Mice were treated with inhibitors when the tumor volume reached ~100 mm³. FX11 or NSC23766 alone significantly inhibited the growth of orthotopic breast tumors formed by WT Hs578T cells (Fig. 7a). In contrast, FX11 did not significantly inhibited growth of tumors formed by LDHA KO Hs578T cells (Fig. 7a). Compared with the single inhibitor treatment, the combination treatment displayed a more pronounced inhibitory effect on the growth of tumors formed by WT but not LDHA KO Hs578T cells (Fig. 7a). These results were confirmed in syngeneic orthotopic breast tumors formed by 4T1 cells with or without LDHA knockdown (Fig. 7b; the knockdown was shown in Supplementary Fig. 7d, e). Further, the combination treatment did not clearly affect the body weight of mice (Extended Data Fig. 8).

Figure 7. LDHA and Rac1 small-molecule inhibitors display a synergistic inhibitory effect on primary and metastatic breast tumors.

a, b, FX11 and NSC23766 (NSC) combination treatment displayed a much more pronounced inhibitory effect on the growth of Hs578T orthotopic breast tumors (a) and 4T1syngeneic orthotopic breast tumors (b) than the single inhibitor treatment. Hs578T with LDHA KO and 4T1 cells with LDHA knockdown and their control cells were used for tumorigenesis. When tumor volumes reached ~100 mm³, mice were treated for 3 and 2 weeks in a & b, respectively. Scale bars: 10 mm. c, d, The combination treatment displayed a much more pronounced inhibitory effect on lung metastasis of Hs578T cells (c) and 4T1 cells (d) than the single inhibitor treatment. In c, Hs578T cells with or without LDHA KO were injected via the tail vein for lung metastasis. Mice were treated for 3 weeks starting one day after cell injection, and sacrificed at 10 weeks after the completion of treatments. In d, mice were treated for 3 weeks when luminescent photon intensities of lung metastases reach ~1×10⁶ ph/sec, and sacrificed at a week after the completion of treatments. e, f, The combination treatment displayed a much more pronounced inhibitory effect on the growth (e) and lung metastasis (f) of mammary tumors of MMTV-PyMT mice than the single inhibitor treatment. Seven-week-old mice that developed early-stage mammary tumors were treated for 3 weeks. Left panels in f, representative H&E staining of lung tissues. Arrows indicate metastatic nodules. Scale bar: 200μm. In a-f, data represent mean ± SD. In a-d, n=8 mice/group two-way (a, b) or one-way (c, d) ANOVA followed by Tukey’s test; in e, f, n=6 mice/group, one-way ANOVA followed by Tukey’s test. *: P<0.0001. Con: vehicle; NSC: NSC23766; Combo: FX11 + NSC. g, LDHA expression is positively associated with p-PAK1 (Ser199/204) levels in three cohorts of human breast cancer specimens. LDHA and p-PAK1 levels were analyzed by IHC staining. Scale bars: 20 μm. p<0.0001; χ² test. h, The proposed model depicting combination inhibition of LDHA and Rac1 activities to treat breast cancers.

We then investigated the effect of the combination treatment on lung metastasis by employing the tail vein injection of Hs578T cells and mammary fat pad implantation of 4T1 cells. While FX11 or NSC23766 treatment alone significantly inhibited lung metastasis of WT Hs578T and 4T1 cells, FX11 did not significantly inhibited lung metastasis of LDHA KO Hs578T or 4T1 cells with LDHA knockdown (Fig. 7c,d). Compared with the single inhibitor treatment, the combination treatment displayed a more pronounced inhibitory effect on lung metastasis of WT Hs578T and 4T1 cells but not LDHA KO Hs578T cells or 4T1 cells with LDHA knockdown (Fig. 7c,d).

The effect of the combination treatment on breast tumor growth and metastasis was further examined in MMTV-PyMT mice that develop spontaneous mammary tumors which metastasize to the lung ⁵⁷. The 7- and 10-week-old female MMTV-PyMT mice that developed early- and late-stage tumors, respectively, were treated with FX11 and/or NSC23766 for 3 weeks. In 7-week-old mice, the combination treatment displayed a more significant inhibitory effect on tumor growth (both tumor weights and numbers) and metastasis compared with the single inhibitor treatment (Fig. 7e & f). In 10-week-old mice, the combination treatment significantly reduced tumor weights and lung metastasis although the inhibitory effect appeared to be much less pronounced compared with that in 7-week-old mice (Extended Data Fig. 9a). The LDHA-Rac1 interaction in mammary tissues and tumors of MMTV-PyMT mice was confirmed by co-IP assays (Extended Data Fig. 9b). Collectively, these results suggest that blocking LDHA enzyme activity and Rac1 simultaneously is a promising strategy to treat breast cancers with LDHA overexpression.

We further investigated whether LDHA activates Rac1 in human breast cancer specimens by analyzing the association between LDHA expression levels and p-PAK1 (Ser199/204) levels in three different breast cancer TMAs (n=200, 203, and 120, respectively) determined by IHC staining. A very significant positive correlation between the levels of LDHA and p-PAK1 (Ser199/204) were found in these three cohorts of breast cancers (Fig. 7g; p<0.0001), suggesting that LDHA overexpression is an important mechanism for Rac1 activation in human cancers. Again, LDHA expression was not linked to any specific breast tumor subtypes in terms of ER, PR or HER2 status (Extended Data Fig. 10).

Discussion

LDHA overexpression in cancer plays a critical role in promoting glycolysis and cancer progression ^{4–9, 12, 13, 18}. Although some earlier studies suggested a potential role of LDHA in transcriptional regulation of gene expression independently of its enzyme activity ⁵⁸, most of the studies on LDHA in cancer have focused on its role as a glycolytic enzyme ^{6, 7, 11, 14–17, 22–24}. In this study, we found that LDHA directly interacted with Rac1 to inhibit the interaction of Rac1 GAPs with Rac1, leading to Rac1 activation to promote breast tumor progression. Blocking the Rac1 signaling genetically or pharmacologically significantly compromised the oncogenic function of LDHA (Fig 7h). While different small-molecule inhibitors have been developed to specifically inhibit LDHA enzyme activity, currently, none of these LDHA inhibitors has progressed to the point of being a clinically viable treatment ^{21, 22, 24}. Our results showed that blocking LDHA and Rac1 activities simultaneously displayed a much more pronounced inhibitory effect on breast cancer cells with LDHA overexpression than the single inhibitor treatment, suggesting a promising strategy to treat breast cancers with LDHA overexpression. Although many studies have been done on Rac1, the mechanism of aberrant Rac1 activation in cancer is incompletely understood ^28–30. Furthermore, currently, there is no clinically viable treatment developed from Rac1 inhibitors ^28–30. Our results revealed that LDHA is an important activator for Rac1. Given that LDHA is frequently overexpressed in cancer, LDHA overexpression constitutes an important mechanism contributing to aberrant Rac1 activation in cancer. Further studies to investigate novel Rac1 regulators, such as LDHA, will lead to identification of new key players in the Rac1 signaling, which can be developed as potential pharmacological targets to antagonize Rac1 activation in cancer. Given that our current study is focused on breast cancer, it will be important to extend our study to different types of cancers in future. In addition to Rac1, LDHA may also regulate other signaling pathways independently of its enzyme activity under both physiological and pathological conditions, which deserves further studies.

In summary, our results reveal that LDHA binds to and activates Rac1 as a critical mechanism for Rac1 activation, and Rac1 activation by LDHA is an unrecognized noncanonical mechanism for LDHA to promote tumor progression independently of its glycolytic enzyme activity. These findings also suggest that pharmacologically targeting LDHA enzyme activity and Rac1 activity simultaneously is a promising strategy to treat breast cancers with LDHA overexpression (Fig 7h).

Methods

Cell lines, vectors and reagents.

Hs578T (HTB-126), SK-BR3 (HTB-30), MCF7 (HTB-22), ZR-75–1 (CRL-1500), BT-549 (HTB-122), MDA-MB231 (HTB-26) and 4T1 (CRL-2539) cells were obtained from American Type Culture Collection (ATCC). Cells were authenticated by short tandem repeat profiling. Cells were regularly tested for mycoplasma using Lookout Mycoplasma PCR detection kit (MP0035, Sigma-Aldrich) and only used when negative. The retroviral vectors expressing LDHA (pLPCX-LDHA-Flag and pLPCX-LDHA-Flag-HA), Rac1 (pLPCX-Myc-Rac1) and their deletion mutants, as well as Rac3 (pLPCX-Myc-Rac3), Cdc42 (pLPCX-Myc-Cdc42), and RhoA (pLPCX-Myc-RhoA) were constructed by PCR amplification ^{31, 35}. The pLPCX-Myc-Rac1-G12V, T17N and T35S vectors were constructed by using a Quikchange II XL Site-Directed Mutagenesis Kit (#200522, Stratagene/Agilent Technologies). The pCMV-HA-RACGAP1 and pCMV-HA-FilGAP vectors were constructed using PCR amplification. Two lentiviral shRNA vectors against LDHA (#1: V3LHS_388269 and #2: V3LHS_388270), Rac1 (#1: V3LHS_374856, and #2: V3LHS_374854) and control shRNA vectors were obtained from Open Biosystems (Huntsville, AL). The lentiviral shRNA vectors against PAK1, Rac3, Cdc42, mouse LDHA (mLDHA) or mouse Rac1 (mRac1) were constructed by inserting the shRNA sequences into the GIPZ lentiviral shRNA vector, and their target sequences are listed in Supplementary Table 2. FX11 (HY-16214) and NSC23766 (HY-15723) were purchased from MedChemExpress (Monmouth Junction, NJ).

Generation of LDHA KO lines using CRISPR/Cas9 system.

The sgRNAs were designed by the CRISPR sgRNA design web tool ⁵⁹. The sgRNA sequences are as follows: sgRNA-a: 5′- CTTTTCTCTAGACTATAATG -3′; sgRNA-b: 5′- GCCGTGATAATGACCAGCTT -3′. The annealed oligonucleotides were ligated into the pSpCas9n(BB)-2A-GFP plasmid (Addgene 48140). GFP-positive single cells transfected with sgRNAs were sorted by flow cytometry, and single-cell colonies were selected by sequencing PCR products of the edited regions ⁶⁰. LDHA deletion was validated by western-blot assays

Western-blot assays.

Standard western-blot assays were employed. Cellular and mouse tissue proteins were extracted using RIPA lysis buffer (FNN0011, Thermo Fisher Scientific) containing the protease inhibitor cocktail (524625, Sigma-Aldrich). Protein concentration in extracts was measured using Bradford reagent. Protein samples were resolved by SDS-PAGE, transferred onto PVDF membranes, and blocked with TBST containing 5% nonfat dry milk. The following antibodies were used for assays: anti-Flag-M2 (F1804, Sigma-Aldrich; 1:20,000 dilution), anti-β-Actin (A5441, Sigma-Aldrich; 1:10,000 dilution), anti-Myc (9E10, Roche; 1:1,000 dilution), anti-HA (3F10, Roche; 1:1,000 dilution), anti-Rac1 (23A8, Millipore; 1:5,000 dilution), anti-LDHA (3582, Cell Signaling; 1:2,000), anti-p-PAK1 (Ser199/204) (09–258, Millipore; 1:1,000 dilution), anti-PAK1 (07–1451, Millipore; 1:1,000 dilution), anti-RACGAP1 (sc-166477, Santa Cruz; 1:1,000 dilution), anti-FilGAP (SAB1401870, Sigma-Aldrich; 1:1,000 dilution), goat anti-mouse (31430, Thermo Fisher Scientific; 1:5,000 dilution) and goat anti-rabbit (31460, Thermo Fisher Scientific, 1:5,000 dilution) antibodies.

Co-IP and LC-MS/MS assays.

To screen for LDHA-binding proteins, LDHA-Flag-HA protein in Hs578T cells transduced with the pLPCX-LDHA-Flag-HA vector was pulled down by two sequential rounds of co-IP ^{60, 61}. In brief, LDHA-Flag-HA was pulled down by co-IP using anti-Flag beads (A2220, Sigma) and eluted with the Flag peptide, which were then subjected for a second round of co-IP using anti-HA beads (A2095, sigma) and eluted with the HA peptide. Cells transduced with control vectors were used as a control. The eluted proteins were subjected to LC-MS/MS analysis performed at the Biological Mass Spectrometry facility of Rutgers University.

To determine the interactions of LDHA with Rac1 and other proteins, including RhoA, Cdc42, and Rac3, as well as the domains for the LDHA-Rac1 interaction, co-IP was performed with 300 to 500 μg of cell lysates prepared by using NP40 buffer (J60766.AP, Thermo Fisher Scientific) ^{60, 61}. Exogenous Flag-tagged proteins or Myc-tagged proteins were pre-cleared with protein A/G-agarose beads prior to immunoprecipitation with the anti-Flag (A2220, Sigma) or anti-Myc (A7470, Roche) agarose beads, respectively. The eluted proteins were then subjected to western-blot assays.

In vitro GST pull-down assays.

In vitro GST pull-down assays were performed using recombinant proteins purified from E. coli ^{60, 61}. In brief, E. coli (BL21 DE3 strain) transformed with pGEX-5X-1-GST-LDHA and pET-32a-His-trx-Rac1 vectors were induced with 0.4 mM IPTG for 16 h at 16°C to express GST-LDHA or His-Rac1 protein. The purified WT or mutant GST-LDHA recombinant protein (200 ng) was immobilized on Glutathione-Sepharose beads, which were then incubated with purified 200 ng of His-Rac1 protein. GST protein alone was used as a negative control. After washing, proteins bound to the beads were eluted and analyzed by western-blot assays.

The proximity ligation assay (PLA).

The PLA, an assay widely used for detecting in situ protein-protein interactions (at distances < 40 nm) at endogenous protein levels in cells and tissues ³⁴, was used to detect the in situ interaction of endogenous Rac1 and LDHA proteins in cells by employing Duolink^™ secondary antibodies and a detection kit (DUO92101, Sigma) according to the manufacturer’s instructions ⁶². In brief, cells were fixed and incubated with Rac1 (24072-1-AP, Proteintech; 1:100 dilution) and LDHA (Sc-137243, Santa Cruz; 1:100 dilution) antibodies together with Duolink secondary antibodies. Ligation steps were performed to ligate the secondary antibodies that are in close proximity. 4’, 6-diamidino-2-phenylindole (DAPI) was used for counterstaining the nucleus.

Analysis of Rac1, Rac3 and Cdc42 activities.

The Rac1, Rac3 and Cdc42 activities were determined by analysis of the ratio of their active GTP-bound forms compared with their total protein levels in cells ^{31, 63}. Briefly, cells were lysed in lysis buffer, and lysates were then incubated at 4°C with PAK-PBD Protein GST Beads (#PAK02, Cytoskeleton Inc.). The bead pellet was washed once with buffer containing 1% NP-40, twice without NP-40, and suspended in 20 μl Laemmli sample buffer (S3401, Sigma). The levels of precipitated GTP-bound Rac1, Rac3 and Cdc42 were measured by western-blot assays using antibodies against Rac1 (23A8, Millipore; 1:1,000 dilution), Rac3 (#ab129062, Abcam; 1:1,000 dilution) and Cdc42 (05–542, Millipore; 1:1,000 dilution), respectively, and then normalized to the levels of their total proteins and Actin in total cell lysates measured by western-blot assays.

Lactate dehydrogenase activity assays.

Hs578T and SK-BR3 cells transduced with vectors expressing WT or L4 LDHA-Flag or control empty vectors were used for LDH enzyme activity assays using a LDH activity assay kit (# MAK066, Sigma-Aldrich). This assay is based on that LDH in samples reduces NAD to NADH, which can be specifically detected by colorimetric (450 nm) assays. Cells were collected and homogenized on ice in LDH assay buffer. The samples were then incubated with LDH substrate mix in 96-well plates and detected at 450 nm with an Infinite 200 PRO microplate reader (Tecan Group Ltd.).

In addition, WT and L4 GST-LDHA recombinant proteins and GST protein were purified as described above and used for LDH enzyme activity assays ^{64, 65}. Standard assay conditions for the reaction (NAD to NADH) were 50 mM phosphate buffer pH 8.0, 0.1 mM NAD+, and 20 mM L-lactate in a 200 μL total volume. Assays were started by the addition of 0.5 μg of purified proteins, and the LDH enzyme activity was analyzed with a spectrophotometer at 340 nm. Protein concentrations were determined using the Bradford protein assay.

Quantitative real-time PCR assays.

Total RNA in cells was prepared with the RNeasy Kit (#74136, QIAGEN). The cDNA was prepared using a TaqMan Reverse Transcription Kit (N8080234, Applied Biosystems), and real-time PCR was performed with the TaqMan PCR Mixture (#4369016, Applied Biosystems) according to the manufacturer’s instructions ⁶¹. The expression of genes in cells was normalized to the expression of the Actin gene.

Colony formation assays.

Cells were seeded in 6-well plates in triplicates and subjected to different treatments for 4 days. After incubation for additional 14 days, plates were washed with PBS and stained with 0.1% of crystal violet. Colonies were counted using Image J software (NIH). The doses of inhibitors used for the combination treatment were based on the IC50 of the single treatment, which include the IC50, and doses higher or lower than the IC50 as previously described ⁶⁶. The combination effect was evaluated with Compusyn software (ComboSyn Inc) using the Combination index (CI) analysis ⁶⁷. CI<1, = 1, and >1 indicate synergistic, additive, and antagonistic effects, respectively. To study the effect of 2-DG or galactose on colony formation, cells were cultured in the medium containing 2-DG (1 mM; D6134, Sigma) or galactose (G0750, Sigma; 25 mM galactose instead of 25 mM glucose) for colony formation assays.

Transwell analysis of cell migration and invasion.

The transwell system (8 μm pore size; #35309, BD Biosciences) was employed for cell migration and invasion assays ^{31, 61}. Cells in the serum-free medium were seeded into upper chambers for assays. For invasion assays, the upper chambers were coated with matrigel. For small-molecule inhibitor treatments, different doses of inhibitors were added to the lower compartment. For 2-DG or galactose treatments, the medium containing 2-DG (4 mM) or galactose (25 mM galactose instead of 25 mM glucose) was added to the lower compartment. Cells on the lower surface were fixed, stained and counted at 24 h after seeding.

Invadopodia formation assays.

For invadopodia formation assays, coverslips were coated with 0.1mg/ml Poly L-lysine, 0.5% glutaraldehyde, 2% gelatin B, and 5 mg/ml sodium borohydride, and cells were seeded on coverslips and incubated overnight before fixation and staining ⁶⁸. Invadopodia were identified by co-staining with phalloidin that selectively stains F-actin (#A12380, ThermoFisher Scientific; 1:1,000 dilution) and the Tks5 antibody (#MABT336, Sigma; 1:100 dilution). Antibodies against LDHA (1998F-1-AP, Proteintech; 1:100 dilution) and Rac1 (24072-1-AP, Proteintech; 1:100 dilution) were used to stain LDHA and Rac1, respectively.

Lactate production assays.

Lactate levels were measured using the L-Lactate Assay Kit (ab65330, Abcam) following the manufacturer’s instructions ⁵⁴. Briefly, the diluted culture medium was mixed with L-lactate assay solution provided in the kit in a 96-well plate. The plate was then incubated at room temperature for 30 minutes. The absorbance was measured at a wavelength of 570 nm with an Infinite 200 PRO microplate reader (Tecan Group Ltd.).

Animal studies.

All mouse experiments were approved by the Rutgers University Institutional Animal Care and Use Committee (IACUC). Mice were maintained at an ambient temperature of 22±1°C and relative humidity 40–60% under a 12h:12h light: dark cycle. LDHA^flox/flox mice (# 030112, Jackson Laboratory) ⁶⁹ were bred with R26-Cre-ERT2 mice (# 008463, Jackson Laboratory) to generate R26-Cre-ERT2, LDHA^flox/flox mice containing Cre-ERT2. To delete LDHA, 8-week-old female mice were injected (i.p.) with 2 mg of Tamoxifen or vehicle each day for 4 days. Different tissues were collected at 3 days after the last injection of Tamoxifen for western-blot and Rac1 activity assays.

For orthotopic xenograft breast tumor models, human breast cancer Hs578T cells (5×10⁶ in a 50:50 mix of DMEM: Matrigel) were injected into the mammary fat pad of 8-week-old female immunocompromised BALB/c athymic nude mice (Taconic; n = 8 mice/group). For syngeneic mouse models, mouse breast cancer 4T1 cells (5×10⁴ in a 50:50 mix of DMEM: Matrigel) were injected into the mammary fat pad of 8-week-old immunocompetent female BALB/c mice as we described (Taconic; n = 8 mice per group) ⁶⁰. For treatments with small-molecule inhibitors, when the tumor volume reached ~100 mm³, mice were injected (i.p.) with FX11 (1 mg/kg/day) and/or NSC23766 (1.5 mg/kg/day) once/day for 2–3 weeks. Control mice were injected with vehicle. Tumor volume = 1/2 (length × width²).

Lung metastasis assays were performed by the tail vein injection of Hs578T cells and mammary fat pad implantation of 4T1 cells in mice, respectively ^{31, 61}. For tail vein injection, Hs578T cells (1×10⁶) stably transduced with lentiviral vectors expressing luciferase were injected into 8-week-old female BALB/c athymic nude mice via the tail vein (n=8 mice/group). Lung metastasis was monitored once/week by using IVIS Spectrum in vivo imaging system (PerkinElmer). Mice were sacrificed at 12 weeks after cell injection, and the lung metastasis was confirmed by routine histopathological analysis. For lung metastasis from syngeneic orthotopic 4T1 breast tumors, 4T1 cells stably transduced with lentiviral vectors expressing luciferase (1×10⁵ in a 50:50 mix of DMEM: Matrigel) were implanted into mammary fat pads of 8-week-old female BALB/c mice (n=8 mice/group). Primary 4T1 tumors were surgically removed when they reached a volume of ~200mm³ to eliminate the effect of primary tumor size on metastasis ^{61, 70}, and mice were sacrificed at 4 weeks after primary tumor removal. Lung metastasis was monitored by using IVIS Spectrum in vivo imaging system and confirmed by routine histopathological analysis after mice were sacrificed. For treatments with small-molecule inhibitors, mice with tail vein injection of cells were treated (i.p.) with FX11 (1 mg/kg/day) and/or NSC23766 (1.5 mg/kg/day) once/day for 3 weeks at one day after cell injection, and mice were sacrificed at 10 weeks after the completion of treatments. For lung metastasis of 4T1 tumors, when luminescent photon intensities of lung metastases reach ~1×10⁶ ph/sec, mice were treated as described above for 3 weeks and sacrificed at a week after the completion of treatments. Control mice were treated with vehicle.

MMTV-PyMT mice (FVB-Tg(MMTV-PyVT)634Mul/LellJ; Strain #:002374) were obtained from The Jackson Laboratory ⁵⁷. Female mice of 7- and 10-week-old that developed early- or late-staged mammary tumors, respectively, were treated (i.p.) with FX11 (1 mg/kg/day) and/or NSC23766 (1.5 mg/kg/day) once/day for 3 weeks. Control mice were treated with vehicle. After mice were sacrificed at the end of treatments, the number and weight of primary mammary tumors per mice were determined, and lung metastasis was examined by routine histopathological analysis. All mice were scheduled for euthanasia once the tumor volume reached 1,700mm³ as indicated in the approved IACUC protocol. The maximal tumor size of all mice used in the present study did not exceed 1,700mm³.

IHC assays of human breast cancer TMAs.

The use of the TMA-RCINJ (obtained from the Rutgers Cancer Institute of New Jersey/RCINJ) containing 200 primary breast tumor tissues were approved by Rutgers institutional review board. The TMA-BR2161 and TMA-BR2082a containing 203 and 120 different human breast tumor tissues, respectively, were obtained from US Biomax (five duplicated tumor tissues in these two TMAs were excluded from the TMA-BR2161 during analysis). All tumor specimens were deidentified samples. IHC staining and scoring were performed to determine the expression levels of LDHA and p-PAK1 (Ser199/204) in tumor samples ^{60, 61}. IHC staining was performed using anti-LDHA (sc-137243; Santa Cruz; 1:100 dilution), anti-p-PAK1 (Ser199/204) (09–258, Millipore; 1:100 dilution), and Ki-67 (ab16667, Abcam; 1:200 dilution) antibodies, respectively. Signals in tumor cells were visually quantified using a scoring system from 0 to 9. The scores were obtained by multiplying the intensity of signals with the percentage of positive cells (signal: 0=no signal, 1=weak signal, 2=intermediate signal, and 3=strong signal; percentage: 0=0%, 1≤25%, 2=25–50%, and 3≥50%). Low and high expression were defined as scores of<6 and≥6, respectively ^{60, 61}.

Statistical analysis.

All data in this study were presented as the mean±SD. Two-tailed Student’s t-test was applied for statistical analysis between two groups. Analyses of variance (ANOVAs) followed by Tukey’s, Dunnett’s or Bonferroni’s multiple comparison tests were applied for multiple group comparisons. The differences between two Kaplan-Meier survival curves were analyzed by the log-rank (Mantel-Cox) test. The association of LDHA and p-PAK1 expression in TMAs was analyzed by χ² test. The statistical analysis was performed using GraphPad Prism 9.

Extended Data

Extended Data Fig. 1. LDHA overexpression and its association with clinical outcomes in different subtypes of breast cancers classified by the status of ER, PR or HER2.

a-c, The increased LDHA mRNA levels in different subtypes of breast cancers compared with matched adjacent non-tumor breast tissues. The data were obtained from TCGA, and the p-value was analyzed by two-tailed paired Student’s t-test. d-f, High LDHA mRNA expression is associated with poor relapse-free survival in patients with different subtypes of breast cancers. The data were obtained from Kaplan-Meier plotter (http://kmplot.com) and analyzed by the log-rank (Mantel-Cox) test. ER: estrogen receptor; PR: progesterone receptor.

Extended Data Fig. 2. LDHA displayed a much weaker interaction with Rac1-T35S mutant compared with WT Rac1 in cells.

Hs578T cells with ectopic expression of LDHA-Flag and WT Myc-Rac1, Myc-Rac1-T17N, or Myc-Rac1-T35S were employed for co-IP assays followed by western-blot assays. Data represent three repeats with similar results.

Extended Data Fig. 3. The dominant negative Rac1-T17N mutant reduces the promoting effect of L4 LDHA on colony formation, migration and invasion of breast cancer cells.

a-c, Hs578T and SK-BR3 cells with ectopic expression of WT or L4 LDHA were transduced with control or Rac1-T17N expression vectors for colony formation (a), migration (b) and invasion (c) assays. Data represent mean ± SD (n=6 independent experiments), two-way ANOVA followed by Tukey’s or Bonferroni’s test. *: P<0.0001.

Extended Data Fig. 4. LDHA but not Rac1 is localized in invadopodia in breast cancer cell lines.

HS578T and MDA-MB231 cells seeded on gelatin-coated coverslips were labeled with anti-LDHA or Rac1 in far-red, anti-Tks5 in green, and phalloidin (to stain F-actin) in red. The co-localization of Tks5 and F-actin in a punctate manner in cells was used as an indication of invadopodia formation. Arrows indicate invadopodia in cells. Scale bars: 20 μm. Data represent three repeats with similar results.

Extended Data Fig. 5. The potential role of different Rho family proteins in mediating the oncogenic effect of LDHA in breast cancer cells.

a, The interaction between LDHA with different Rho family proteins analyzed by co-IP assays. Hs578T cells expressing LDHA-Flag and Myc-Rac1, Myc-Cdc42, Myc-Rac3, or Myc-RhoA were used for co-IP and western-blot assays. b, LDHA activated Rac3 and Cdc42, but at a much less extent compared with its effect on Rac1. Hs578T and SK-BR3 cells expressing LDHA-Flag were used for PAK-PBD pull-down assays to measure the levels of GTP-bound Rac1, Rac3 and Cdc42. Left panels: represented results. Right panels: relative Rac1, Rac3 and Cdc42 activities analyzed by comparing the levels of GTP-bound Rac1, Rac3 and Cdc42 to the levels of total Rac1, Rac3 and Cdc42 proteins, respectively, in cells. Data represent n = 3 independent experiments, two-tailed unpaired Student’s t-test. c, Knockdown of Rac3 or Cdc42 displayed a much less pronounced effect on colony formation, migration and invasion of breast cancer cells with WT or L4 LDHA expression compared with Rac1 knockdown. Data represent mean ± SD (n=6 independent experiments), two-way ANOVA followed by Tukey’s test or Dunnett’s test. d, Knockdown of Rac3 and Cdc42 in cells was confirmed by quantitative Taqman real-time PCR assays. Their mRNA levels were normalized with Actin. Data represent n = 3 independent experiments, one-way ANOVA followed by Dunnett’s test. *: P<0.0001.

Extended Data Fig. 6. The relative expression levels of Rac1, Rac3 and Cdc42 in breast cancer cells.

Rac1 has a much higher expression level than Rac3 and Cdc42 in majority of breast cancer cell lines, including cell lines used in this study (labeled in red) as shown by the RNA-Seq data from Cancer Cell Line Encyclopedia (CCLE; https://sites.broadinstitute.org/ccle/).

Extended Data Fig. 7. LDHA and Rac1 small-molecule inhibitors display a much more pronounced inhibitory effect on colony formation, migration and invasion of breast cancer cells.

a, The combination treatment displayed a much more pronounced inhibitory effect on colony formation of breast cancer cells. BT-549, MCF7 and ZR-75-1 cells were treated with indicated concentrations of FX11 and/or NSC23766 for 4 days before colony formation assays. Combo: FX11 + NSC23766. b, c, The combination treatment displayed a much more pronounced inhibitory effect on migration (b) and invasion (c) of breast cancer cells than the single inhibitor treatment as analyzed by transwell assays. In b, c, BT-549, MCF7 and ZR-75-1 cells were treated with the indicated concentrations of FX11 and/or NSC23766 for 24 h. NSC: NSC23766. Data represent mean ± SD (n=6 independent experiments), one-way ANOVA followed by Tukey’s test. *: P<0.0001.

Extended Data Fig. 8. The effect of small-molecule inhibitor treatments on the body weights of tumor-bearing mice.

a, Small-molecule inhibitor treatments did not significantly affect the body weights of female nude mice bearing orthotopic tumors formed by WT Hs578T cells. b, Small-molecule inhibitor treatments did not significantly affect the body weights of female BALB/c mice bearing orthotopic tumors formed by 4T1 cells transduced with the control lentiviral shRNA vector. Mice were treated with FX11 (1 mg/kg/day; i.p.; once/day) and/or NSC23766 (1.5 mg/kg/day, i.p.; once/day), or vehicle (−). Mice with Hs578T tumors were treated for 3 weeks, and mice with 4T1 tumors were treated for 2 weeks. The body weights of mice were measured and recorded at the days indicated. Data represent mean ± SD. n=8 mice/group. Statistical differences were determined by two-way ANOVA followed by Tukey’s test.

Extended Data Fig. 9. The effect of FX11 and/or NSC23766 on the growth and lung metastasis of mammary tumors in MMTV-PyMT mice.

a, The effects of FX11 and/or NSC23766 on the growth and metastasis of mammary tumors in 10-week-old MMTV-PyMT mice. The 10-week-old female MMTV-PyMT mice that developed late-stage tumors were treated with FX11 (1 mg/kg/day; i.p.; once/day) and/or NSC23766 (1.5 mg/kg/day, i.p.; once/day) for 3 weeks before they were sacrificed for analysis. Data represent mean ± SD; n=6 mice/group. One-way ANOVA followed by Student’s t-test. Con: vehicle; NSC: NSC23766; combo: FX11 + NSC. b, The interaction between endogenous LDHA and Rac1 in both non-tumor and tumor mammary tissues of MMTV-PyMT mice detected by co-IP and western-blot assays. Non-tumor mammary tissues were obtained from 5-week-old MMTV-PyMT mice, and mammary tumor tissues were obtained from 10-week-old MMTV-PyMT mice. Normal mammary tissues from 5-week-old LDHA-deficient mice (the R26-Cre-ERT2, LDHA^flox/flox mice treated with Tamoxifen to delete LDHA as presented in Fig. 2h) were used as negative controls for assays. NC: negative control.

Extended Data Fig. 10. LDHA expression is not linked to any specific subtypes of breast cancer.

LDHA protein levels in two breast cancer TMAs, including TMA-RCINJ (a-c; n=200) and TMA-BR2082a (d-f; n=120), were analyzed by IHC staining, and were compared in different cancer subtypes classified by the status of ER (a, d), PR (b, e), or HER2 (c, f). Statistical studies were performed by using χ² test. The TMA-BR2161 does not have information on the status of ER, PR or HER2.

Data availability:

Publicly available datasets used in this study are TGCA (https://portal.gdc.cancer.gov/), Oncomine (https://www.oncomine.com/), and CCLE (https://sites.broadinstitute.org/ccle/). All data supporting the present study are available within the article and supplementary information files. Source data are provided with this paper.