Bioinformatic and metabolomic analysis reveals miR-155 regulates thiamine level in breast cancer

Abstract

microRNA-155 (miR-155) is one of the well-known oncogenic miRNA implicated in various types of tumors. Thiamine, commonly known as vitamin B1, is one of critical cofactors for energy metabolic enzymes including pyruvate dehydrogenase, alpha ketoglutarate dehydrogenase, and transketolase. Here we report a novel role of miR-155 in cancer metabolism through the up-regulation of thiamine in breast cancer cells. A bioinformatic analysis of miRNA array and metabolite-profiling data from NCI-60 cancer cell panel revealed thiamine as a metabolite positively correlated with the miR-155 expression level. We confirmed it in MCF7, MDA-MB-436 and two human primary breast cancer cells by showing reduced thiamine levels upon a knock-down of miR-155. To understand how the miR-155 controls thiamine level, a set of key molecules for thiamine homeostasis were further analyzed after the knockdown of miR-155. The results showed the expression of two thiamine transporter genes (SLC19A2, SLC25A19) as well as thiamine pyrophosphokinase-1 (TPK1) were decreased in both RNA and protein level in miR-155 dependent manner. Finally, we confirm the finding by showing a positive correlation between miR-155 and thiamine level in 71 triple negative breast tumors. Taken altogether, our study demonstrates a role of miR-155 in thiamine homeostasis and suggests a function of this oncogenic miRNA on breast cancer metabolism.

Introduction

microRNAs (miRNAs) are a type of noncoding RNA 19–23 nucleotide-long in size. They are synthesized by a series of enzymatic processing occurring in the nucleus and cytoplasm [1]. There are ~1400 miRNAs known in human and are predicted to regulate more than 60% of the total repertoire of mRNAs [2]. Particularly, it interacts with 5′ or 3′ untranslated region (UTR) of mRNA, causing degradation or translational repression [3,4]. By extensive studies since last decade, we know that some of the miRNAs are closely related to specific diseases including cancer [5–12]. Among them, miR-155 is a well-known oncogenic miRNA [13–15]. It is transcribed from its own gene called B-cell Insertion Cluster (BIC), which has been known as an insertion site of a virus [16]. When the viral insertion over-expresses the non-coding gene, it causes lymphoma. After miR-155 was shown to be responsible for tumorigenesis, many groups showed that it is also over-expressed in multiple types of solid cancer including breast, pancreatic, colon and lung [17–19]. In addition, we have previously shown its up-regulation in triple negative breast cancer [20]. Molecularly, about 150 validated targets of the miR-155 are known, acting on multiple signaling pathways [21–23]. The functional consequence of the pathways in cancer can be summarized as increased proliferation and migration, decreased apoptosis [24].

Thiamine (vitamin B1) is one of the essential water-soluble vitamins [25]. In contrast to bacteria or plants where thiamine is synthesized, mammalians should uptake it as a diet. It is required for carbohydrate and amino acid metabolism [26]. Specifically, it is known as an essential cofactor for key metabolic enzymes such as transketolase (TKT), alpha-ketoglutarate dehydrogenase (alpha-KGDH) and Pyruvate dehydrogenase (PDH) [27,28]. For cellular uptake of thiamine, thiamine transporters are needed. Two transporters in SLC19A family, SLC19A2 (THTR1 for protein) and SLC19A3 (THTR2 for protein), are major transporters for the task [29]. Another important gene in thiamine metabolism is thiamine pyrophosphokinase-1 (TPK1). TPK1 converts thiamine to thiamine pyrophosphate (TPP), which has activity as a co-enzyme [30,31]. On the other hand, thiamine pyrophosphate carrier (TPC) that is produced from the SLC25A19 gene, transports the TPP through mitochondrial membrane to provide the cofactor to Pyruvate dehydrogenase and alpha-ketoglutarate dehydrogenase [32].

The role of thiamine and thiamine homeostasis genes in cancer has been controversial. Earlier reports demonstrated that SLC19A3 gene was down-regulated in multiple types of cancers [33,34]. In agreement with this, it was reported that SLC19A3 sensitize cells to radiation or doxorubicin. Also the down-regulation of SLC19A3 gene confers the cancer cell more resistant to apoptosis [35].These results suggested SLC19A3 as a tumor suppressor gene [33]. On the other hand, the SLC19A3 was shown to be highly up-regulated by hypoxia, suggesting a role in the hypoxic response [36]. In addition, a recent report showed that three genes namely SLC19A2, SLC25A19, and TPK1 are up-regulated in 43 breast cancer specimens, suggesting their roles in breast cancer [37]. Here, we report a well-known oncogenic miRNA, miR-155 that regulates thiamine metabolism through SLC19A2, SLC25A19, and TPK1 genes. Starting from bioinformatics analysis of NCI-60 cancer cell panel [38], we verified our findings into primary breast cancer cells as well as cell lines. Further, we present a positive correlation between miR-155 and thiamine level in triple negative breast cancer (TNBC).

Materials and methods

Derivation of human primary breast cancer cells

Human primary breast tumor tissues were obtained from breast cancer disease patients diagnosed as triple negative breast cancer (TNBC), determined by IHC (ASAN medical center, Seoul, Korea). Tumor tissues were minced and treated with collagenase type 3 for 18–20 hours. After centrifugation, the cells were seeded onto collagen-coated plates. Primary breast cancer cells, in passages 2~6, were cultured in RPMI1640 medium (Hyclone, Logan, UT, USA) supplemented with 5% fetal bovine serum (FBS) (Hyclone), human epidermal growth factor (hEGF) (200 ng/ml, Gibco, Gaithersburg, MD, USA), Hydrocortisol (10 μg/ml, Sigma-Aldrich, St. Louis, MO, USA), transferrin (10 μg/ml, Sigma) and penicillin/streptomycin (Hyclone). Two human primary breast cancer cells were named as Patient A and Patient B.

Cell culture, transfection and viral infection

Cancer cell lines, MCF7, MDA-MB-436, HeLa (ATCC, Manassas, VA, USA), and 293TN cells (System Biosciences, SBI, Mountain View, CA, USA) were maintained in DMEM Medium (Hyclone) containing 10% FBS and 1% penicillin/streptomycin. For knockdown of miR-155, we used lentiviral constructs containing eitheranti-miR-155 sequence (miRZIP155) or a control sequence (pSIH-H1-siLuc-copGFP) (SBI). The lentiviral production was performed using the SBI instructions. 293TN cells were transfected using lipofectamine 2000 (Invitrogen) with the pPACK plasmids plus control or miRZiP155 lentiviral vectors. To achieve efficient knockdown of the miR-155 by anti-miR-155 lentivirus, first we calculated the MOI of lentiviruses. The lentiviral supernatant were serially diluted and infected into HeLa cells for 3 days. Subsequently, green fluorescent-expressing cells were detected by fluorescence-activated cell sorting (FACS) (Supplementary Fig. S1). After we determine the optimal MOI (10), we applied it to the other cell lines. Control or miRZIP155-containing the lentivirus particles were collected and infected into MCF7, MDA-MB-436 and two human primary breast cancer cells. Twenty-four hours after transduction, these cells were cultured in the presence of Puromycin (1.5 μg/ml; Sigma) for 7 days. The cells with viral infection were photographed with an inverted microscope (Carl Zeiss AG, Oberkochen, Germany).

Quantitative RT-PCR (qRT-PCR)

RNA extraction was performed using TRizol (Invitrogen, Carlsbad, CA, USA) following the instructions from the manufacturer. One microgram of total RNA was used for cDNA synthesis (Superscript First-Strand Synthesis System; Invitrogen) following the manufacturer’s protocol. The expression levels of SLC19A2, SLC19A3, TPK-1 and SLC25A19 were measured by SYBR Green PCR Kit in LightCycler 480 II (Roche Applied Sciences, Indianapolis, IN, USA). Standard PCR conditions are: 15 minutes at 95 °C followed by 40 cycles of 15-second denaturation at 94 °C, 30-second annealing at 60 °C, and 30-second extension at 70 °C. The primer sequences for PCR are shown in Supplementary Table S1. Human ribosomal protein gene RpL13a was used as an internal control gene. To confirm amplification of specific transcripts, melting curve profiles (cooling the sample to 40 °C and heating slowly to 95 °C with continuous measurement of fluorescence) were produced at the end of each PCR. Relative quantification was calculated using the 2-(ΔΔCt) method [39]. For quantitative assessment of miR-155, TRIzol-isolated RNAs were reverse transcribed by miScript II RT Kit (Qiagen, Hilden, Germany) and analyzed by miScript SYBR Green PCR Kit (Qiagen) following the instructions from the manufacturer. Samples were analyzed using the LightCycler 480II for an initial denaturation at 95 °C for 15 minutes followed by 40 PCR cycles with 94 °C for 15 seconds, 55 °C for 30 seconds, and 70 °C for 30 seconds. Fold expression was calculated using the comparative 2-(ΔΔCt) method as mentioned above. Cycle threshold (Ct) values of the analyzed miRNAs were normalized to Ct values obtained for the noncoding, small nuclear RNA molecule U6 (RNU6). Data were presented as fold change versus control.

Western blot analysis

The protein expression of THTR1, TPC, and TPK-1 was measured by Western blot. Preparation of cell lysates and Western blot analysis were performed as previously described [40]. Briefly, cells were cultured in 100-mm culture plates and rinsed with cold PBS and lysed with 0.25 ml of lysis buffer [150 mM NaCl, 1% Triton X-100, 1% Sodium deoxycholate, 50 mM Tris-HCl (pH 7.5), 2 mM EDTA (pH 8.0), and 0.1% SDS]. Concentration of soluble protein from total cell lysate was determined using BCA reagent with BSA as a standard (Pierce Biotechnology, Rockford, IL, USA). Approximately 10–50 μg of protein of total cell lysate per sample was separated on 10% SDS PAGE, transferred to Nitrocellulose membrane, probed with anti-monoclonal THTR1 (1:500, Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-polyclonal TPC (1:1000, Abgent, San Diego, CA, USA) and anti-polyclonal TPK-1 (1:1000, Abcam, Cambridge, UK) antibodies. Specific signal was obtained using an enhanced chemiluminescence (ECL) detection system (Pierce Biotechnology). Membranes were stripped and re-probed with anti-beta-actin monoclonal antibody (1:1000; Santa Cruz Biotechnology) to ensure equal loading. Densitometric analysis was performed using the imageJ 1.47v system (http://imagej.nih.gov/ij).

Reporter construction and luciferase assay

For reporter construction, reverse complement oligonucleotide for matured miR-155 sequence was introduced into the pMIR-REPORT™ Luciferase vector (Applied Biosystems, Inc., Foster City, CA, USA) via SpeI and Hind III sites. MCF, MDA-MB-436 and two human primary breast cancer cells were transfected with 2 μg of plasmid DNA, which included either control vector or the miR-155 3′-UTR Luciferase reporter vector, using Lipofectamine2000 (Invitrogen). Forty-eight hours after transfection, cells were lysed and luciferase activity was measured using Dual-Luciferase Reporter Assay System (Promega Corp, Madison, WI, USA) and a Victor Luminometer (Perkin-Elmer, Waltham, MA, USA) according to the manufacturer’s protocol.

In vitro BrdU/7-AAD incorporation assay

To test whether the excess supply of thiamine can reverse the growth inhibition of cancer cell triggered by the miR-155 knockdown, we supplied excess thiamine (400 mg/ml) into DMEM media. After 24 hours, the cells with control, miRZIP155, and thiamine treated miRZIP155 were incubated with 10μm BrdU for 6 hours. In vitro BrdU kit (BD Pharmingen, San Diego, CA, USA) was used as described in the manufacturer’s manual. Briefly, the cells were fixed with ethanol-containing fixation buffer and washed with Perm/Wash buffer. For BrdU staining, cells were labeled by FITC-conjugated anti-BrdU antibody for 20 minutes at room temperature. After incubation, cells were washed and resuspended with FACS staining buffer. Finally, BrdU-treated cells were analyzed using an AccuriFlow Cytometry (BD Biosciences). The percentage of green fluorescent-positive cells was calculated using the CFlow software.

alamarBlue® cell viability assay

To monitor cell proliferation, the cells in a 96 well plate were treated with control, miRZIP155 and thiamine plus miRZIP155 virus. After 24 hours, 1/10th volume of alamarBlue® reagent (Invitrogen) was added directly to culture media. The cells incubated for an additional 4 hours to assay viability, which was detected by fluorescence measurements using a microplate fluorescence spectrophotometer (GenTeks Biosciences, Inc., San-Chong, Taipei).

Human TNBC sample analysis

We used 71 formalin-fixed, paraffin-embedded (FFPE) tissues of TNBC samples. To check the miR-155 expression level, total miRNAs were extracted from sectioned FFPE tissues (8 μm thickness) by using a miRNeasy FFPE kit (Qiagen). First we checked the miR-155 expression level. That enabled us to classify the samples into two groups; miR-155 high and miR-155 low, that are marked in the Supplementary Table S3 as blue and red, respectively. Among them, we selected 15 samples from each group, based on the frozen tissue availability in our BRC (Bio Resource Center, ASAN Medical Center, Seoul, Korea). These 30 samples were further tested for the thiamine level, as described in the following section.

Sample preparation for metabolite analysis

For metabolite measurement in cells, 1 million human breast cancer cells were lysed by methanol, and thiamine was extracted from aqueous phase by adding CHCl₃ and H2O. For tissues, 50–100 mg of human breast cancer tissue was cut into small pieces using a scissor, and 600 μl of CHCl₃/MeOH (2/1, v/v) was added to the tissue. The sample was homogenized using a Tissue Lyzer (Qiagen) and the supernatant was collected after centrifugation. Three hundred microliter of H2O was added to the supernatant. The sample solution was vortexed vigorously and centrifuged, and aqueous layer was collected. The extracts were dried using vacuum centrifuge, then stored at −20 °C until analysis. The dried matter was reconstituted with 50% Acetonitrile (ACN) prior to LC-MS/MS analysis.

Liquid chromatography-tandem mass spectrometry (LC-MS/MS)

A liquid chromatography-tandem mass spectrometry system equipped with 1290 HPLC (Agilent Technologies, Santa Clara, CA, USA), Qtrap 5500 (AB SCIEX, Framingham, MA, USA) and hydrophilic interaction normal phase column (Zorbax HILIC plus 3.5 μm, 100 × 2.1 mm) was used. The separation gradient for thiamine analysis used mobile phase A (10 mM ammonium formate in ACN/H2O (50/50, v/v) and mobile phase B(10 mM ammonium formate in ACN/H2O (90/10, v/v) and proceeded at 25 °C (400 μ/min). The separation gradient was as follows: hold at 40% of A for 3 minutes, 40% to 90% of A for 0.1 minutes, hold at 90% of A for 12 minutes, 90% to 40% of A for 6 seconds, and then hold at 40% of A for 5 minutes. The selected reaction monitoring (SRM) mode was used in positive ion mode, and the extracted ion chromatogram corresponding to the specific transition (Q1 = m/z 265, Q3 = m/z 122) for thiamine was used for quantification. The calibration range for thiamine was 0.1–1000 nM (r2 ≥ 0.99).

Statistics

All data were presented as means ± standard error of the mean (SEM). Comparison between two groups was performed using t-test. A p value of less than 0.05 was considered to be statistically significant.

Results

Bioinformatic analysis of miRNA and metabolite profiling data in NCI 60 cell lines revealed metabolites correlated with miR-155 level

In order to identify metabolites that are correlated with miR-155 expression, we used two publicly available datasets. The first one is a miRNA expression profile of NCI 60 cell line panel (GSE 26375), and the other one is a metabolite profile of the same cell panel [41]. Out of 60 cell lines, we selected 15, which showed top 25% expression of miR-155. Likewise, another 15 cell lines were selected that showed bottom 25% expression of miR-155 (Fig. 1A and Supplementary Table S2). After that, the consumption and release (CORE) profiles of 219 metabolites from these two groups were compared. The analysis revealed six metabolites that are significantly correlated with miR-155 expression level (Fig. 1B–G and Table 1). Among them, thiamine drew our attention as it was the only metabolite up-regulated in miR-155 high group (positive-correlation). Moreover, thiamine has been implicated in cancer metabolism, especially in breast cancer, based on previous reports [36,42,43]. Therefore, we pursued the role of miR-155 for the regulation of thiamine metabolism in breast cancer.

Fig. 1.

Bioinformatic analysis revealed thiamine has positive correlation with miR-155 expression. (A) Box plot of miR-155 showing various expressions in NCI-60 cell panel. Bottom and top 25% cell lines are selected (indicated in red) and further analyzed. (B~G) Scattered plot of six metabolites (glycodeoxycholate, IMP, fru-1,6-DP, N-carbonyl-beta-alanine, uracil and thiamine), marked as blue (for miR-155 low group) or red (for miR-155 high group) diamonds. miR-155 expression level of each cell line was indicated as green circle. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 1.

List of six metabolites showing significant correlations with miR-155 expression level.

IDa	Metabolite	p-Valued
47	Glycodeoxycholate/glycochenodeoxycholate	0.0131
66	IMPb	0.0147
129	Fru-1,6-DP/fru-2,6-DP/glc-1,6-DPc	0.0191
24	N-carbamoyl-beta-alanine	0.0204
111	Uracil	0.0490
101	Thiamine	0.0495

^aArbitrary serial number for metabolites.

^bIMP, inositol mono-phosphate.

^cFru-1,6-DP, fructose 1,6 diphosphate.

^dCorrelation between miR-155 and metabolite level.

Lentivirus mediated knockdown of miR-155 in breast primary culture and cancer cell lines reduced intracellular thiamine level

To determine whether the miR-155 can alter the thiamine level in breast cancer cells, we introduced two cancer cell lines and two patients derived primary tumor cells (Fig. 2A–D). miR-155 was stably inhibited in these cells by lentivirus mediated antagomiR expression (see Materials and methods for details). The level of miR-155 in these cells was measured by real time PCR, showing efficient knockdown (Fig. 2E). The functional assay for miR-155 using luciferase reporter assay also supported efficient inhibition of miR-155 (Fig. 2F). We determined the level of thiamine in these cells by mass spectrometry (see Materials and methods for details). Even though there was a high variability of thiamine level in four cells tested, we could observe dramatic decrease of the relative cellular thiamine by the knockdown of miR-155 (Table 2 and Fig. 2G). Among the four primary or cancer cell lines tested, we found two (Patient A and MDA-MB-436) cell types that showed significant thiamine level reduction. These results strongly suggest that miR-155 is a regulator of cellular thiamine level.

Fig. 2.

miR-155 dependent thiamine level change in breast cancer cells. (A~D) Images of cells infected by miR-155 knock-down virus (miRZIP155). Patient A (A), Patient B (B), MCF7 (C) and MDA-MB-436 (D) cells were infected with lentivirus overnight, and selected with puromycin for 7 days. These photos were taken on third day after lentiviral infection. Magnification: 200×. (E) qRT-PCR analysis to measure the miR-155 expression level after miR-155 knockdown. Error bar means ± SEM (n = 3). RNU6 was used as control. (F) Measurement of luciferase activity to confirm the miR-155 function was decreased by miRZIP155. A luciferase reporter vector containing a miR-155 binding site on 3′-UTR region or empty vector was transfected into Patient A, Patient B, MCF7 and MDA-MB-436 cells. Error bar means ± SEM (n = 3). **: p < 0.01; ***: p < 0.001. (G) Relative thiamine level in miR-155 knock-down cells, measured by LC-MS/MS. The relative thiamine level was calculated by dividing thiamine in miRZIP-infected cells by thiamine level in control cells. Raw data are provided in Table 2. Error bar means ± SEM (n = 3).

Table 2.

Level of thiamine in human breast cancer cells infected by anti-miR-155 virus (miRZIP).

Sample name	Thiamine (nM)			Ave.	p-Value
	1st	2nd	3rd
Patient A_Control	65	74	87	75	0.0097
Patient A_miRZIP155	28	24	35	29
Patient B_Control	229	100	89	139	0.1127
Patient B_miRZIP155	35	33	15	28
MCF7_Control	5536	1448	1992	2992	0.1488
MCF7_miRZIP155	593	154	288	345
MDA-MB-436_Control	496	205	192	298	0.0179
MDA-MB-436_miRZIP155	342	105	79	175

miR-155 dependent expression of the thiamine homeostasis genes in breast cancer cells

Based on the results above, we next questioned how miR-155 could affect the level of intracellular thiamine. As the miRNA is a gene expression regulator, we analyzed four thiamine metabolism genes including SLC19A2 (THTR1), SLC19A3 (THTR2), SLC25A19 (TPC) and TPK1 in the breast cancer cells used in Fig. 2, by real time PCR analysis. The results revealed that the expression of three out of four genes was greatly reduced by the knockdown of miR-155 (Fig. 3A, C, D), whereas the SLC19A3 gene expression was marginally affected (Fig. 3B). These data are also supported by a western blotting of the THTR1, TPC and TPK1 that showed the reduction of protein expression (Fig. 3E and Supplementary Fig. S2 for quantitation). Thus, these data imply that miR-155 is required to maintain the expression of thiamine homeostasis genes, and thereby keeps the thiamine level in the cells. Of note, miRNA regulation on UTR of a gene usually results in down-regulation of the gene expression as the miRNA can destabilize target mRNA or inhibit translation. Therefore, we reasoned that it is unlikely the miR-155 can directly regulated these genes as the expression was decreased by the knockdown of the miR-155. Indeed, the prediction by TargetScan software showed no putative miR-155 binding sites in the UTRs of the three genes (Supplementary Fig. S3).

Fig. 3.

Thiamine homeostasis genes are regulated by miR-155 in human breast cancer cells. (A~D) qRT-PCR analysis of SLC19A2, SLC19A3, TPK1, and SLC25A19 genes. Data were normalized to human RpL13a level. Error bar means ± SEM (n = 3). *: p < .0.05, **: p < 0.01; ***: p < 0.001 (E) Western blot results of THTR1, TPK1 and TPC proteins. β-actin was used as loading control.

Excess thiamine supply partially rescues cellular defects triggered by the knockdown of miR-155

We next tested the impact of miR-155 knockdown and thiamine regulation on cellular proliferation. As shown in Fig. 4A, knockdown of miR-155 resulted in a defect in cell proliferation, determined by alamarBlue assay. As there are many other targets of miR-155, it is possible that this cellular defect can be caused by mechanisms other than the thiamine regulation. Therefore, we examined the rescuing effect of extra supply of thiamine on the cellular defects. The result in Fig. 5A shows that the extra supply of thiamine can partially, but significantly, rescue the cell proliferation defects caused by miR-155 knockdown. We also measured cell cycle progression upon the miR-155 knock-down by BrdU/7-AAD incorporation assay (Fig. 4B). In support of the alamarBlue data, we observed that miR-155 inhibition reduces BrdU positive S phase cells, which is partially rescued by thiamine supply in culture media (Fig. 4C). Taken together, these data demonstrate that the miR-155 regulates cellular proliferation in part by maintaining thiamine metabolism.

Fig. 4.

Excess thiamine supply partially rescues cellular defects triggered by miR-155 inhibition. (A) alamarBlue assay results after extra thiamine supply. Cell proliferation was detected by a fluorescence spectrophotometer. Control, miRZIP treated and miRZIP plusthiamine supply was marked as black, white and gray bars respectively. ***: p < 0.001 (B) Representative picture of BrdU/7-AAD incorporation assay results. Signals are detected by Flow Cytometry (see Materials and methods), and the BrdU positive cells (in red gate) or BrdU/7-AAD double positive cells (in green gate) were measured. The numbers are % of each population to total cells. (C) Graph of results shown in (B). Error bar means ± SEM (n = 4). *: p < .0.05, **: p < 0.01. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 5.

Positive correlations between miR-155 and thiamine level in human TNBC samples. (A) qRT-PCR analysis results of miR-155 in human 71 TNBC samples. The results of miR-155 high (H1-15) and low (L1-15) samples are shown. The MDA-MB436 cell line was used as a control to divide the TNBC samples into miR-155 high or low groups. (B) Results of thiamine measurement for the 30 samples selected in (A). LC-MS/MS analysis measured the thiamine level in tumor tissues. p = 0.0019 (Students’ t-test). (C) Average of the thiamine level in miR-155 high and low groups shown in (B). Error bar means ± SEM (n = 15). **: p < 0.01. (D) A schematic presentation of the role of miR-155 in the thiamine metabolism. Up-regulated miR-155 can trigger increase in intracellular thiamine by regulating three genes (SLC19A2, TPK1 and SLC25A19), via an unidentified factor X.

Correlation of miR-155 and thiamine level in human TNBC samples

To verify the role of miR-155 on thiamine regulation in human breast cancer samples, we collected 90 triple negative breast cancer (TNBC) samples and measured miR-155 level from FFPE sections. By comparing the 71 reliable miR-155 expression levels with that of the MDA-MB 436 cells, which was shown to be high in miR-155 [40], we selected 24 miR-155 high and 22 low tumors (Supplementary Table S3). Among them, we analyzed thiamine level from 15 specimens for each group (Fig. 5A), based on the availability of fresh tissue. The results shown in Fig. 5B and C indicate that the thiamine level is significantly increased in miR-155 high TNBC samples compared with miR-155 low. These data support that our findings from the human breast cancer primary culture or cell lines are applicable to human breast cancer cases as well.

Discussion

In the present study, we aimed to identify miR-155 related metabolite that can provide a mechanistic insight for the oncogenic nature of this miRNA. Instead of running expensive and time consuming metabolic profiling, we used publicly available data from NCI 60 cancer cell panel [41]. Considering that this panel contains variable types of cancer cells, it would be possible to identify cancer type specific metabolites. Out of six metabolites that showed significant correlation with miR-155, glycodeoxycholate showed the strongest correlation. Considering it can trigger pancreatitis and miR-155 is one of the major miRNA induced by inflammation [44], we speculate that the glycodeoxycholate has a function in cancer inflammation in combination with miR-155. The other four metabolites belong to either energy metabolism (Fru-1,6-DP) or nucleotide/amino acid metabolism (IMP, Uracil, N-carbamoyl-beta-alanine). Intriguingly, however, we found the inverse correlation between these three metabolites with miR-155 level that is difficult to explain considering the oncogenic nature of the miR-155. In contrast, thiamine showed positive correlation with miR-155, even though correlation was not dramatic. Moreover, previous reports showed its function in breast cancer cells that triggered us to further study the thiamine as a miR-155 associated metabolite.

Our study, performed in two primary breast culture and two cancer cell lines, clearly showed that the knockdown of miR-155 reduced the level of thiamine homeostasis genes, thereby the thiamine level was also decreased. As these cells (with the exception of MCF7 cells) express a high level of miR-155, we aimed to inhibit the miR-155 using viral expression of antagomiR to test whether miR-155 is a cause of the thiamine regulation. As the miR-155 controls the expression of multiple transcription factors [13], it is possible that miR-155 controls thiamine homeostasis genes to facilitate its metabolism. A recent report showed that SP1 transcription factor is involved in the regulation of SLC19A3 (THTR2) gene [45] and another report showed that hypoxia condition could trigger its expression [36], whereas no trans-activation signal has been known for the SLC19A2(THTR1) gene. We speculate that the unknown negative regulator for the expression of SLC19A2, SLC25A19 and TPK1 is a target of miR-155 so the up-regulation of miR-155 can result in trans-activation of the gene. This idea is illustrated in Fig. 5D. Finding such regulators will be a next step to fully understand the role of miR-155 in thiamine metabolism. Of note, we observed that MCF7 has very high level of thiamine (Fig. 4A) with the level of low miR-155 (Fig. 2E). These data suggest that there are other ways to increase thiamine metabolism independently to the miR-155, which is in a line with the data in Fig. 3 showing comparable expression of the thiamine homeostasis genes in MCF7 cells.

In Fig. 5, we further tested whether the cell proliferation defects driven by miR-155 knockdown can be rescued, at least partially, by the excess supply of the thiamine. We reasoned that the reduced thiamine availability in cancer cell (Fig. 4A) could be reversed by the excess supply of the thiamine. However, because there are about 150 targets identified for the miR-155 [21], it is becoming clear this oncogenic miRNA exerts its function through multiple aspects of cancer signals. Therefore, the partial rescue of the cellular proliferation would be attributed to other oncogenic impact of the miR-155 that is not reversed by the thiamine supply. Interestingly, a previous report showed that decreased SLC19A3 (THTR2) level can make breast cancer cells less apoptotic [35]. Therefore, it can be speculated that the up-regulated SLC19A3 (THTR2) by miR-155 can sensitize cells to apoptosis. However, as we know the miR-155 can exert its anti-apoptotic effect by regulating multiple targets including Caspase-3, FADD, Apaf-1, we think the high level of miR-155 can override the effect of increased SLC19A3 (THTR2) and drive cells to survival pathway [13].

A previous study has revealed that the miR-155 plays a positive role in glycolysis, through the up-regulation of hexokinase-2 [46]. Our report provides another role of miR-155 in cancer metabolism through the up-regulation of a cofactor, thiamine, that is further verified in TNBC cancer specimens. Future study will be focused on how to prevent such an oncogenic effect by reducing abnormal thiamine metabolism in breast cancer cells. In addition, studying the effect of miR-155 on thiamine in other types of cancer cell will provide a new insight for miR-155 as a general thiamine regulator in cancer.