br legend continued on next page br addition of hemin
(legend continued on next page)
addition of hemin, which degrades BACH1, reduced the high migration rates of mTN Midostaurin (PKC412) but did not affect basal migration of mTC cells (Figure 3A). CRISPR/CAS9-mediated knockout of Bach1 did not affect the proliferation of mTC and mTN cells but normalized the high migration rates of mTN cells (Figures 3B and 3C; Figure S4A). Bach1 knockout also prevented the increased migration induced by NRF2 activation in mTC cells (Figure S4B) and by antioxidant treatment in human lung cancer cell lines (Figure S4C–S4F). In NSG mice, Bach1 knockout also prevented mTN cells from metastasizing to the lung after i.v. injection (Figure 3D), but it did not alter the growth of primary tumors when the cells were injected subcutaneously (Fig-ure S4G). We also used the pSECC lentivirus (Sa´nchez-Rivera et al., 2014) for in vivo CRISPR/CAS9-mediated knockout of Bach1 in KP tumors (Figure 3E). In line with the previous exper-iment, BACH1 depletion in vivo prevented vitamin E-induced authentic metastasis in KP mice without affecting the growth of primary tumors (Figures 3F and S4H–S4J).
To determine whether BACH1 is sufficient to stimulate migra-tion and metastasis we used the CRISPR-SAM (synergistic activa-tion mediator) strategy to increase the expression of endogenous BACH1 (Figure 3G). In naive mTC cell lines transduced with SAM-sgBach1 constructs, BACH1 protein levels increased 4- to 6-fold, which increased cell migration and invasion (Figures 3H–3J). The ability of SAM-sgBach1 mTC cells to metastasize to the lung after i.v. injection in NSG mice was 2.5-fold higher than in controls and similar to that of mTN cells (Figure 3K). Moreover, transducing A549 and H1975 human lung cancer cells with SAM-sgBACH1 increased endogenous BACH1 levels, stimulated migration, and increased lung metastases by 3- to 4-fold (Figures S4K–S4N).
Genes Encoding the Glycolytic Enzymes HK2 and GAPDH Are BACH1 Targets
To identify BACH1 target genes responsible for the increased metastasis, we immunoprecipitated chromatin bound to BACH1 in mTC and mTN cells and sequenced the associated DNA (chromatin immunoprecipitation sequencing [ChIP-seq]). The known BACH1 binding motif (from GEO: GSE31477) was present in half of the identified target genes, as was a de novo BACH1-like motif containing the ARE consensus motif (Fig-ure S5A). Known BACH1 target genes were also identified (Fig-ure S5B). BACH1 bound to several regions of the Hmox-1 pro-moter in mTC and mTN cells, but binding levels did not differ significantly (Figures S5B and S5C). Integrating ChIP-seq and RNA-seq data revealed that BACH1 had bound to 240 differen-tially expressed genes (Figure 4A).
Further analysis of these genes in the STRING database of protein-protein interactions identified ‘‘metabolic process’’ as one of three enriched networks (Figure 4A). Among the ‘‘meta-bolic process’’ genes that showed higher BACH1 binding in mTN than mTC cells, Hexokinase 2 (Hk2) was the most signifi-cantly upregulated in the RNA-seq analysis (Figure 4B). We confirmed that expression of Hk2 and also Gapdh, but not Hk1, was 2- to 3-fold higher in mTN than mTC cells (Figures 4C and 4D). Analyses of Hk2 and Gapdh promoters revealed that BACH1-binding sequences were present 300 base pairs (bp) upstream of the annotated transcriptional start site (Fig-ure 4E). ChIP-qPCR analyses confirmed the increased BACH1 occupancy on Hk2 and Gapdh promoters in mTN cells (Figures 4F, 4G, S5D and S5E).
To test whether the new BACH1-binding sequences partici-pate in Hk2 and Gapdh transcriptional activation, we cloned 400- to 500-bp promoter regions containing the wild-type BACH1 binding sequence, or a mutated version that would pre-vent BACH1 binding, into luciferase reporter vectors (Figure 4E). Luciferase activity from the Hk2 and Gapdh promoters was 4- to 5-fold higher in mTN than in mTC cells (Figures 4H and 4I). Knockout of Bach1 essentially abolished the signal in mTN cells, as did expressing the vectors with mutated sequences (Figures 4H and 4I). Thus, BACH1 stimulates transcription of Hk2 and Gapdh, which raises the possibility that BACH1 regulates glycolysis.
Antioxidants Stimulate Glycolysis in a BACH1-Dependent Fashion
Glycolysis rates were 50% higher in mTN than in mTC cells, as judged by analyses of the oxygen consumption rate (OCR) and the extracellular acidification rate (ECAR); the OCR/ECAR ratio was reduced correspondingly (Figures 5A, 5B, and S6A). This was accompanied by increased glucose uptake and lactate secretion (Figure 5C). The involvement of glycolysis prompted further analyses of the RNA-seq data. In addition to Hk2 and Gapdh, mTN cells had increased transcript levels of other glyco-lytic enzymes, including 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase 3 (Pfkfb3) and solute-carrier family 16 member 1 (Slc16a1, encoding the lactate transporter MCT-1) (Figure 5D). Furthermore, analyses of the SEEK co-expression database revealed that in 173 human lung cancer datasets BACH1 expres-sion correlated with HK2 expression (Figure 5E) and tended to correlate with PFKFB3 and SLC16A1 (Figure 5E).