Modulating redox homeostasis and cellular reprogramming through inhibited methylenetetrahydrofolate dehydrogenase 2 enzymatic activities in lung cancer

Expression profile of MTHFD2 in lung cancer (LCa)

Cancers are presumed to proceed successively from less to more advanced preclinical phases. Additionally, their stage-specific correlation with metabolic gene MTHFD2 mediating mitochondrial folate pathway is still unknown. Our data showed a significantly increased expression profile of MTHFD2 mRNA in all stages of LCa tissue compared to normal with increasing tendency (Figure 1A). Similar results were evident in mRNA (Figure 1B) and protein (Figure 1C) levels in three human LCa cell lines A549, H1299 and H441 compared with normal lung epithelial cell line. These outcomes were further corroborated using Oncomine^TM Database in three previous studies [19–21], showing highly significant expression (over 2-fold) of MTHFD2 in LCa compared to normal counterparts (Figure 1D). A similar trend of MTHFD2 expression profiles was also observed in The Cancer Genome Atlas (TCGA) and genotype-tissue expression (GTEx) projects using GEPIA2 online platform (http://gepia2.cancer-pku.cn/#index) (Supplementary Figure 1), which further validates expression levels.

Figure 1. Levels of MTHFD2 in lung cancer tissues and cell lines. (A) Relative MTHFD2 gene expression of normal lung tissues and the different stages of lung cancer tissues. * and ** indicate p<0.05 and p<0.01, respectively using paired t-test, compared to normal group. (B) Relative MTHFD2 gene expression of normal lung cell line BEAS-2B and lung cancer cell lines A549, H1299 and H441. ** and *** indicate p<0.01 and p<0.001, respectively using one-way ANOVA, compared to normal cell line BEAS-2B. (C) Representative MTHFD2 protein expression in normal and lung cancer cell lines, where β-actin was loaded as relative control (D) Online Oncomine database-dependent (https://www.oncomine.org/) in silico analysis of fold changes in MTHFD2 expression in lung cancer tissues. MTHFD2 mRNA expression in normal and malignant lung specimens are presented as box and whisker plots. Sample numbers, fold changes, and p-value for MTHFD2 expression between normal and malignant specimens are indicated. Data are expressed as mean ± SD.

Impact of MTHFD2 knockdown on LCa growth and proliferation

Mitochondrial folate-coupled metabolism is thought to be central in proliferative cancer [22]. Hence, we investigated the role of MTHFD2 in LCa growth and proliferation by establishing vector control and MTHFD2-knockdown cell lines of A549, H1299 and H441 through lentiviral transduction. The efficiency of these cells lines was examined by detecting GFP expression through flow cytometry (Supplementary Figure 2), and the influence of lentiviral transduction to the cell viability was evaluated through MTT assay (Supplementary Figure 3). Our results confirmed gene (Figure 2A) and protein (Figure 2B) levels of expressed MTHFD2, showing clearly established cell lines. Further, MTT assay demonstrated significantly inhibited proliferation ability of MTHFD2-knockdown cell lines (Figure 2C). Eventually, soft agar exhibited significantly decreased number of colonies in all the MTHFD2-knockdown cell lines (Figure 2D), indicating the role of MTHFD2 in malignant transformed cells. However, the cytometric analysis of apoptosis showed that compared to vector control, no significant increase in apoptotic population in early (Q4) as well as late (Q2) phase of MTHFD2-knockdown cell lines was observed (Supplementary Figure 4). This indicate that MTHFD2 knockdown may not directly cause significant cell apoptosis, but might decelerate cell growth ability, which also corresponds to our results of MTT assay and soft agar assay.

Figure 2. MTHFD2 knockdown on growth and proliferation potential of lung cancer. (A) Relative gene expression of MTHFD2 in knockdown group compared to vector control. *** indicate p<0.001 using one-way ANOVA. (B) The MTHFD2 protein expression of vector control group and MTHFD2-knockdown group, where β-actin as loading control. (C) MTT assay-dependent cell viabilities of MTHFD2-knockdown groups of A549, H1299 and H441 compared to vector control. (D) The morphology and quantification of formed colonies of vector control and MTHFD2-knockdown groups of A549, H1299 and H441 cell lines by soft agar assay. * and ** indicate p<0.05 and p<0.01, respectively using one-way ANOVA, compared to shVector group.

MTHFD2-mediated lung cancer stemness

To deduce the relation between MTHFD2 and LCa cell stemness, we conducted sphere formation assay, which is widely used to identify stem cells based on their self-renewal and differentiation potential [23]. Our results showed significantly up-regulated protein expressions of both MTHFD2 as well as HIF-1α in parental A549, H1299 and H441 sphere-forming cells compared to counterparts (Figure 3A), implying that LCa stemness might be attributed to increased MTHFD2 expression, possibly through inducing elevated levels of HIF-1α under low oxygen tension.

Figure 3. MTHFD2 knockdown and assessment of lung cancer stemness through sphere formation assay. (A) Representative protein expression of MTHFD2 and HIF-1α in adherent non-sphere forming cells (Ctrl) and sphere cells (Sphere) of parental cell lines, β-actin was used as loading control. (B) Bright-field images of primary and secondary tumor spheres (at 4X and 10X magnifications) and (C) their relative quantification of vector control and MTHFD2-knockdown cell lines, A549, H1299 and H441. * and ** indicate p<0.05 and p<0.01, respectively using paired t-test.

To further investigate whether MTHFD2 modulate LCa self-renewal ability, vector control and MTHFD2-knockdown group were conducted with primary sphere formation to evaluate cancer stem cell population and secondary sphere formation for detecting maintenance of stemness respectively. We found that using equal concentration of cells seeded for formation of primary and secondary spheres, both the number as well as size of formed spheres were significantly inhibited in MTHFD2-knockdown cell lines (Figure 3B and 3C). This indicate that MTHFD2 is vital for self-renewal and differentiation ability of LCa cells.

Impact of MTHFD2 on cellular metabolic reprogramming-dependent tumor aggressiveness

Besides determining cancer stemness through in vitro sphere formation, an in vivo limiting dilution transplantation assay is highly important to assess the minimal amount of cells capable of initiating LCa, which was done by subcutaneous injection of various dilutions of vector control and MTHFD2-knockdown LCa cells lines in NOD/SCID mice. The results demonstrated that in 5×10⁴ inoculated cells group, more tumors were formed in vector controls (A549 (80%), H1299 (40%) and H441 (80%)), whereas no tumors were formed in MTHFD2-knockdown groups (Figure 4A). Besides, though injected cells at same concentration (5×10⁵) potentially formed tumors in all mice, the tumor size of vector control was significantly larger compared to MTHFD2-knockdown groups (Figure 4B). This indicate the possible role of MTHFD2 knockdown to metabolically reprogram the aggressive phenotypes of LCa cells to a lesser extent.

Figure 4. MTHFD2 knockdown and in vivo tumor formation ability. The cancer initiating ability (A), and tumor volumes (B) after subcutaneous injection of different cell number of respective vector control and MTHFD2-knockdown cell lines of A549, H1299 and H441 into the right flank of 6-week-old NOD/SCID mice (n = 5). *** indicate p<0.001 using one-way ANOVA, compared to vector control group.

Cellular oxygen sensing and MTHFD2-mediated metabolic reprogramming

Available oxygen level and its sensing induce notable and clinically relevant alterations in cancerous cells and tumors by controlling gene expression [24]. Hence, we explored whether mitochondrial folate microenvironment is associated with cellular oxygen sensing and MTHFD2. We treated parental A549, H1299 and H441 cell lines with cobalt chloride (CoCl₂) and digoxin, the low oxygen tension-mimetic agent and hypoxia-inducible factor (HIF)-1α inhibitor respectively, and their combination as well. Our data showed that in the presence of CoCl₂, levels of HIF-1α and MTHFD2 were increased, which were suppressed in presence of digoxin (Figure 5A). Also, HIF-1α and MTHFD2 levels were time-dependently increased following CoCl₂ treatment (Supplementary Figure 5). This infer that CoCl₂ may decrease cellular oxygen sensing potential, leading to elevated levels of HIF-1α and MTHFD2. Further, sphere formation assay was conducted after the cell lines were pre-treated with CoCl₂ for 48 hours. The results demonstrated that only vector control groups showed significantly increased sphere number, whereas MTHFD2-knockdown groups showed no significant increase compared to CoCl₂-treated vector group (Figure 5B and 5C). This was further corroborated by examining protein levels of HIF-1α and MTHFD2 in vector control and MTHFD2-knockdown groups after CoCl₂ treatment, which was significantly increased only in vector control groups, supporting highly formed spheres in this group (Supplementary Figure 6).

Figure 5. Oxygen availability and regulation of MTHFD2. (A) Representative western blots showing hypoxia inducible factor (HIF)-1α and MTHFD2 protein expression in A549, H1299 and H441 cells after treatment with only 100 μM CoCl2 or combined with 100nM digoxin (C+D), where β-actin was used as loading control. (B) Bright-field images of sphere formed by vector control and MTHFD2-knockdown groups of A549, H1299 and H441 cells and (C) their relative quantified number in the presence or absence of 100 μM CoCl2 for 2 days. *, ** and *** indicate p<0.05, p<0.01 and p<0.001, respectively using one-way ANOVA, compared to vector control group.

Influence of MTHFD2 on oxidative stress response and metabolic reprogramming in lung cancer

To investigate the MTHFD2-mediated redox homeostasis under low oxygen tension, we determined the levels of NADPH and ROS after treating vector control and MTHFD2-knockdown groups with CoCl₂. The vector control groups revealed significantly higher expression of NADPH (Figure 6A); whereas ROS level was significantly increased in MTHFD2-knockdown groups (Figure 6B). These results imply that MTHFD2 might contribute in maintaining cellular redox homeostasis under low oxygen tension, knockdown of which led to accumulated ROS levels, resulting into inhibited LCa phenotype.

Figure 6. MTHFD2 and maintenance of redox homeostasis under low oxygen tension. Rate of change (%) of (A) NADPH and (B) ROS production in vector control and MTHFD2-knockdown groups of A549, H1299 and H441 cells after their treatment with 200 μM of CoCl₂ for 24 hours. *, ** and *** indicate p<0.05, p<0.01 and p<0.001, respectively using paired t-test, compared to shVector group.

Cell culture

LCa cell lines A549, H1299 and H441 used in these experiments were purchased from the Bioresource Collection and Research Center, Hsinchu, Taiwan. Cell lines were cultured in RPMI-1640 (Gibco, Grand Island, New York) with 10% (vol/vol) and supplemented with fetal bovine serum (FBS) (Gibco, Mexico) and 1% antibiotic antimycotic solution (Corning, USA). An additional 2.5 μg/ml of puromycin (Sigma-Aldrich, Saint Louis, MO, USA) was supplied in the medium for lentiviral transduced cell lines. The culture medium was maintained at 37°C with 5% CO₂.

In silico analysis of MTHFD2 gene expression

The Oncomine^TM Cancer Microarray Database (https://www.oncomine.org/) was used to perform comparative in silico analysis of MTHFD2 gene expression in cancer versus normal tissue. MTHFD2 mRNA expression was queried in cancer vs. normal in the lung cancer category. The query threshold was set at a p-value <10³, over two-fold difference of expression, top 10% ranking in over-expression genes, and sample sizes are greater than 10. The expression profile of MTHFD2 was further validated in The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) projects using GEPIA2 online platform (http://gepia2.cancer-pku.cn/#index). We chose dataset of lung squamous cell carcinoma (LUSC), which has been represented as box plot, and p-value cutoff is set at 0.05.

Isolation of total RNA and quantitative real-time polymerase chain reaction

The mRNA levels of MTHFD2 between human normal lung and lung cancer tissues were investigated using Lung Cancer cDNA Array V (OriGene, HLRT105), which consisted of 7-normal, 6-IA, 5-IB, 13-IIB, 7-IIIA, 7-IIIB, and 3-IV with a total of 41cancer tissue samples. The sample patient population includes 29 male and 19 females with the age ranging from 44 to 84 years old. For cellular total RNA was extracted with PureLink® RNA Mini Kit (Thermo Fisher Scientific) and was reverse-transcripted (Applied Biosystems) into cDNA using RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Scientific). Quantitative PCR was conducted by using 7300 ABI Real-time PCR system (Applied Biosystems) with SYBR Green Master Mix (Applied Biosystems) by using following primers: β-actin-F: 5’-AGAGCTACGAGCTGCCTGAC-3’; β-actin-R: 5’-AGCACTGTGTTGGCGTACAG-3’; MTHFD2-F: 5’-GATGGCCTCCTTGTTCAGTTG-3’; MTHFD2-R: 5’-ATCCTTGTCTGGAGAAACAGCATT-3’

Gene knockdown by lentiviral transduction

Lentiviral solution was produced by transfecting plasmids pHP, pHEF-VSVG, pGIPZ-shMTHFD2 and pCEP4-tat in 293FT packaging cells (Invitrogen) by using opti-MEM (Gibco) and lipofectamine 2000 (Invitrogen). Infectious viral solution was collected 48h after transfection. Parental cell lines A549, H1299 and H441 were cultured with viral solution with appropriate titers containing 8 μg/ml of polybrene in medium then incubated overnight at 37 °C in 5% CO₂. For selection of lentivirus transduced cells, cells were cultured in medium with additional 2.5 μg/ml puromycin. The efficacy of lentivirus transduction was evaluated by observing GFP expression under fluorescence microscope. Real-time PCR and western blot analysis were utilized to evaluate the level of MTHFD2 expression. The sequences for MTHFD2 shRNA are as following: shMTHFD2#66: 5’-ATTCCCACATCAATGACTG-3’; shMTHFD2#68: 5’-ATTGCATTTCTATTGGCCT-3’.

Cell viability and proliferation

Cell viability and proliferation were evaluated by MTT assay. Thousand cells in 100 μl of medium were seeded per well into 96-wells plate and then incubated for 1, 3, 5 and 7 days, and was replaced with fresh medium after every two days. After incubation, medium was discarded and supplied with 100 μl of MTT solution (Sigma, Saint Louis, MO, USA), then incubated at 37°C for 4 hours. After 4-hours incubation, MTT solution was removed and 50 μl of DMSO was added to dissolve the dark blue formazan crystal, and absorbance was determined using ELISA reader (Multiskan RC Microplate Reader, Thermo) at 595 nm.

Anchorage-independent growth

1 mL of 0.5% agar in complete growth medium was added in each well of six-well plate as a base agar. Top agar was prepared by 1 mL of 0.3% agar in complete growth medium containing 1×10⁴ cells of vector control and MTHFD2-knockdown of A549 and H1299, and 3×10⁵ cells of vector control and MTHFD2-knockdown of H441, top agar was overlaid on the base agar. Growth medium (2 mL) was added on top of the second layer and changed twice a week. After incubation for three weeks, the colonies formed were stained with 0.005% crystal violet in methanol (Fisher Scientific, Hampton, NH, USA), and then enumerated.

Analysis of cell apoptosis

Flow cytometry-based determination of apoptosis was done by following the manufacturers' instructions of commercial Annexin V-PE / 7-AAD Apoptosis Detection Kit (Elabscience, USA) and the representative histogram were obtained through SA3800 spectral cell analyzer (SONY biotechnology, San Jose, CA, USA).

Hypoxia induction

Hypoxia was chemically induced by cobalt chloride (CoCl₂) as described in the previous study [44]. 100 μM of cobalt (II) chloride hexahydrate (Sigma, Saint Louis, MO, USA) was used to treat the target cell lines for 4, 8, 12, 24 and 48 hours. Digoxin (Sigma-Aldrich, D6003) was used as an inhibitor of hypoxia-inducible factor 1-alpha [45]. 100 nM of digoxin was applied to cells with 100 μM of CoCl₂ for 24 hours. For sphere formation assay under hypoxic condition, cell lines were pre-treated with 100 μM of CoCl₂ for 48 hours before seeding for sphere formation.

Sphere formation assay

The medium for sphere formation was DMEM/F12 contained with 1ml of B27 Supplement (50X) (Gibco, NY, USA), EGF 40 ng/ml (Peprotech, USA), bFGF 20ng/ml (Peprotech, USA), Leukemia Inhibitory Factor 20 ng/ml (Sigma, Saint Louis, MO, USA), 100 μl of ITS Supplement (100X) (Sigma, Saint Louis, MO, USA) and 1 % of PSA. For A549, 1 % of extra methylcellulose was added into medium. All the cells were seeded into Ultra-Low Attachment Surface 6 Well Plate (Corning, Kennebunk, ME, USA). An additional 1 ml of medium to the wells was added every 2 days. Cells were incubated at 37°C, 5 % CO₂ incubator for 1-3 weeks until most of the formed spheres had grown to an adequate size (≥ 100 μm).

Western blot

Protein extraction and immunoblotting were performed as previously described [46]. Primary antibodies against β-actin (Genetex, #GTX109639, 1:10000), HIF-1α (Genetex, #GTX127309, 1:600), MTHFD2 (Genetex, #GTX115482, 1:1000) were used. Horseradish peroxidase-conjugated secondary antibodies specific to rabbit IgG were used (Genetex, # GTX213110-01, 1:5000). Lastly, membranes were rinsed with Immobilon Western Chemiluminescent HRP Substrate (Merck Millipore, USA) for chemiluminescent detection, then blots were visualized using MultiGel-21 Western/Gel Image System (Top Bio, Taiwan).

Cellular NADPH determination

To determine cellular NADPH, NADP/NADPH Quantitation Colorimetric Kit (Biovision, #K347) was employed as prescribed by manufacturer’s protocol. For each cell lines, 4×10⁶ cells were seeded and treated with or without 200 μM of CoCl₂ for 24 hours before cellular NADPH determination.

Cellular ROS determination

To determine cellular ROS, In Vitro ROS/RNS Assay Kit (Cell Biolabs, San Diego, CA, USA) was used according to manufacturer’s protocol. For each cell lines, 5×10⁶ cells were seeded and treated with or without 200 μM of CoCl₂ for 24 hours before cellular ROS determination.

In vivo limiting dilution transplantation assay

All the animal studies were approved by Institutional Animal Care and Use Committee (IACUC) of Taipei Medical University Taiwan (Approval No. LAC-2016-0526). In our study, 6-weeks old non-obese diabetic/severe combined immunodeficiency (NOD/SCID) mice were used for in vivo limiting dilution transplantation assay. Mice were anesthetized by mixture of 10X dilution of Zoletil 50 (Virbac, Carros, France) and 10X dilution of Rompun (Bayer, Korea). Both vector control groups and MTHFD2-knockdown groups of A549, H1299 and H441 cells were subcutaneously injected with three different dilutions of cell concentration into the flank of right leg (n=5). The injected cell number of each line were 5×10⁵, 1×10⁵ and 5×10⁴. Tumor size was measured with digital caliper twice a week, and was calculated using following formula:

$Volume=\frac{Lengh\times Widt{h}^2}{2}$

Statistical analysis

The sample size in each experiment was at least in triplicate, unless otherwise indicated. Statistical analyses were conducted using GraphPad Prism 7 (version 7.00, GraphPad Software, San Diego, CA, USA), and Microsoft Excel (Office 2016 Professional Plus, Santa Rosa, California, USA). All data are presented as mean±SD. A p-value of less than 0.05 was considered statistically significant.