To investigate the expression level of miR-143 in established cell lines, we profiled miR-143 expression level using Q-PCR in a number of selected cancer cell lines as well as non-tumourigenic cell lines (Additional file 2
: Figure S1). As expected, the expression levels of miR-143 were extremely low or undetectable in all tested cancer cell lines. The highest expression levels were observed in the non-tumorigenic fibroblast cell lines BJ and Tig3. We chose to focus our further studies on the human colon cancer cell line DLD-1 since miR-143 expression was virtually absent from this cell line and thus mimics the situation reported in colon cancer tumors. To confirm previous findings that miR-143 inhibits cell growth we investigated the cell proliferation upon transient transfections with a miR-143 duplex. The effect of miR-143 duplex transfection in DLD-1 cells was confirmed by co-transfection of a luciferase reporter containing a perfect complementary site to the mature miR-143 (Additional file 2
: Figure S2). As demonstrated by cell growth assays, overexpression of miR-143 resulted in a decreased cell proliferation (Figure ).
Figure 1 miR-143 overexpression reduces the proliferative potential of DLD-1 cells. DLD-1 cells transfected with miR-143 duplex exhibit a reduced cell proliferation as measured by crystal violet growth assay. Data are shown as the mean ± S.D. of four replicates. (more ...)
We next sought to identify functionally relevant targets that could explain the underlying role of miR-143 in cancer. To achieve this, DLD-1 cells were transfected with miR-143 duplex or mock transfected. Total RNA was harvested 24
h post-transfection and analyzed on Affymetrix HG-U133 Plus 2.0 human arrays.
To determine whether genes down-regulated by miR-143 were related to specific cellular functions, we performed a search for enriched functional annotations as derived from KEGG and BioCarta pathway databases. In KEGG pathways, down-regulated gene sets were enriched in cell cycle (p-value
), glutamate metabolism (p-value
), N-glycan biosynthesis (p-value
) and glycolysis/gluconeogenesis (p-value
0.01) (Table ). In BioCarta pathways, down-regulated gene set were enriched in the mTOR signalling pathway (p-value
) and the G1/S check point pathway (p-value
) (Table ).
Enriched KEGG pathways among miR-143 down-regulated gene sets
Enriched BioCarta pathways among miR-143 down-regulated gene sets
In addition we also performed a search for enriched transcription factor and miR-143 binding motifs among miR-143 down-regulated genes. The second most significantly enriched motif in the down-regulated gene set was the miR-143 seed site (p-value
), while the most significantly enriched motif was binding site of the transcription factor E2F (Additional file 3
: Table S2).
Seed site enrichment analysis of seed sites present in the 3'UTRs of transcripts showed a very significant enrichment of miR-143 seed sites among the down-regulated transcripts (Figure A). In this analysis we grouped the 3'UTRs into down-regulated (FC
−1.1), no-change (genes with FC centred on 0) and up-regulated (FC
1.1) (Figure A). The p-values for the enrichment of miR-143 seed sites (including 7mer, 7mer-1A and 8mer sites) were 3.4·10−19
when considering the down-regulated transcripts vs. up-regulated transcripts and 5.8·10−28
when considering down-regulated transcript vs. no change transcripts. This was also the case when the seed site enrichment was evaluated by two alternative methods of calculating the seed site enrichment, either as seed site occurrences after correcting the up, down and no-change sets to the same size or as seed site occurrences calculated per kb (Additional file 2
: Figure S3A and S3B).
Figure 2 Microarray based identification of miR-143 targets. A, The percentages of genes in the up, down and no-change sets with seeds sites in their 3′UTRs. Seed sites were mutually exclusive. Mean log fold-changes were 0.193, −0.004 and −0.249 (more ...)
In addition to seed site enrichment analysis, we also performed an unbiased word analysis of words present in 3’UTRs of transcripts ranked according to their FC. The 7mer and the 7mer-1A seed sites of miR-143 were identified as the most significantly enriched 7mer words in the 3'UTRs of transcripts down-regulated after miR-143 overexpression (Figure B). Sequence variations of the miR-143 seed site were also among the highest scoring words. Similarly, a 6mer word analysis indentified the miR-143 6mer seed site as the most enriched 6mer word in 3'UTRs of down-regulated transcripts (Additional file 2
: Figure S4). The overrepresentation of miR-143 seed sites in 3'UTRs of down-regulated transcripts can be visualized by plotting the running sum of the overrepresentation scores of the seed sites in transcripts ranked according to their logFC. As seen in Figure C the overrepresentation scores of the miR-143 7mer seed site are highest among 3'UTRs of down-regulated transcripts (black line). This was not the case, when the same analysis was performed for 100 permutations of the ranked transcript list (grey lines).
As miR-143 possesses a tumor-suppressor function, we would expect a down-regulation of oncogenes and genes promoting cell proliferation upon miR-143 overexpression. Putative miR-143 targets, defined as genes down-regulated upon miR-143 overexpression with a FC
−1.1 and containing either a least one 7mer, 7mer-1A or 8mer seed site in their 3'UTR are listed in Additional file 4
: Table S3. Among the down-regulated genes containing miR-143 seed sites in their 3'UTRs we found a number of genes that have previously been implicated in tumorigenesis. This include the Steroid 5-alpha-reductase SRD5A1, the CCR4-NOT component RQCD1 and the Rab11 effector protein RAB11FIP1 which have all been reported as up-regulated in breast cancers [35
]. Other miR-143 responsive genes with a miR-143 seed site in their 3'UTR were SEMA5A
which have all been shown to be up-regulated in cancers and to promote cell proliferation [38
]. Among the putative miR-143 targets we also found the deubiquitinating enzyme USP22, which have been reported to be associated with a poor prognosis of colorectal cancer [42
] and invasive breast cancer [43
]. In addition we also observed a reduced expression of the glycolytic enzyme hexokinase 2
) upon miR-143 overexpression. HK2 catalyzes the first step of glycolysis by phosphorylation of glucose into glucose-6-phosphate. HK2 is often found upregulated in cancer and facilitates a high rate of glucose metabolism necessary for tumor growth [44
Among genes motioned above, three genes have also been predicted by a target prediction model built on 12 transfection datasets with good prediction posterior probabilities and low FDR (<25%) [45
]. This includes HK2 (posterior probability
0.93; adjusted p-value
0.17), RAB11FIP1 (posterior probability
0.92; adjusted p-value
0.17) and SEMA5A (posterior probability
0.9; adjusted p-value
0.22). This adds supportive evidence that these genes are direct targets of miR-143 beyond a simple seed match search.
As a validation of the microarray data we selected 7 transcripts identified as down-regulated by miR-143 in the microarray analysis for Q-PCR validation. All 7 transcripts including HK2
were found to be down-regulated, confirming the microarray data (Figure D). In accordance with previous reports we also find KRAS
downregulated upon miR-143 overexpression (Figure D) [15
was also found down-regulated in the microarray analysis but because it had a borderline logFC of −0.14, it is not included in our list of potential miR-143 targets as listed in Additional file 4
: Table S3.
Due to HK2’s reported role in promoting tumor growth we wanted to investigate if the tumorsuppressor function of miR-143 can in part be accounted for due to its down-regulation of HK2. To investigate the role of miR-143 mediated regulation of HK2, we firstly wanted to determine whether HK2 is a direct target of miR-143. The 3'UTR of HK2
contain a 8mer seed site for miR-143 (Figure A). To determine if miR-143 directly regulates HK2
through binding to its 3'UTR, 3'UTR luciferase reporter constructs were cloned containing 789 base-pair UTR fragments. Overexpression of miR-143 resulted in a significant decrease of the luciferase activity (p-value
0.002) of a construct holding the wild-type 3'UTR of HK2
(Figure B). This regulation was alleviated when two nucleotides in the seed site had been mutated (Figure B), indicating that the miR-143 directly regulates HK2
. Western blot analysis further confirmed that miR-143 overexpression lead to a down-regulation of HK2 protein levels in both DLD-1 and HCT116 colon cancer cells (Figure C). The effect of miR-143 overexpression in HCT116 cells measured by a miR-143 luciferase reporter was similar to that observed in DLD-1 cells (Additional file 2
: Figure S2B). Notably the endogenous protein level of HK2 in DLD-1 cells is considerably higher than in HCT116 cells, but nevertheless miR-143 overexpression lead to a downregulation of HK2 protein levels in both cell lines.
Figure 3 miR-143 downregulates HK2. A, Sequence alignment of the miR-143 seed region and HK2 3′UTR. (ENSG00000159399) B, Luciferase assay with pGL3+ constructs holding a 789bp 3′UTR fragment of HK2 downstream to the firefly luciferase (more ...)
To further strengthen the connection between miR-143 and HK2 we surveyed the expression levels of both miR-143 and HK2
in data from The Cancer Genome Atlas (TCGA) consortium. TCGA is currently profiling the genomes of a large cohort of colon and rectum adenocarcinomas. We found a significant negative correlation between miR-143 and HK2
−0.22, Pearson correlation) in 184 public TCGA colorectal adenocarcinoma tumor samples with miRNA and mRNA expression data available (Figure D). This observation supports that miR-143 could target and repress HK2
and that HK2
expression could be upregulated in a subset of tumors due to lower levels of miR-143.
To investigate the effect of HK2 on cellular growth, we performed cell proliferation assay upon siRNA-mediated knockdown of HK2. HK2 knockdown was verified on both mRNA and protein level in DLD-1 cells (Figure A and B). We found that, in a similar manner to miR-143 overexpression, HK2 siRNA mediated knockdown also resulted in a reduced cell proliferation (Figure C).
Figure 4 Knockdown of HK2 as well as miR-143 overexpression results in decreased lactate secretion. A, Verification of HK2 knockdown on protein level. Cells were transfected on two consecutive days with HK2 siRNA and protein was harvested 48h after the (more ...)
Next, to determine if downregulation of HK2 mediated by miR-143 resulted in an impairment of glycolysis, lactate production was measured in mock transfected cells and cells transfected with miR-143 duplex or HK2 siRNA. Cells transfected with a HK2 siRNA showed a marked decrease in the rate of lactate secretion over a period of 48
h (Figure D). Importantly, a decrease in the lactate secretion was also observed upon miR-143 overexpression (Figure D), confirming that miR-143 downregulation of HK2 has a functional effect on the glucose metabolism. The observed decrease in lactate secretion caused by miR-143 overexpression is less pronounced than for HK2 siRNA mediated inhibition. However, this might be explained by the more efficient down-regulation of HK2 mediated by the HK2 siRNA than by overexpression of miR-143.