Conceived and designed the experiments: BBY. Performed the experiments: DYL TS ZJ WD ZD BBY. Analyzed the data: DYL BBY. Wrote the paper: BBY.
The authors have declared that no competing interests exist.
Mature microRNAs (miRNAs) are single-stranded RNAs of 18–24 nucleotides that repress post-transcriptional gene expression. However, it is unknown whether the functions of mature miRNAs can be regulated. Here we report that expression of versican 3′UTR induces organ adhesion in transgenic mice by modulating miR-199a* activities. The study was initiated by the hypothesis that the non-coding 3′UTR plays a role in the regulation of miRNA function. Transgenic mice expressing a construct harboring the 3′UTR of versican exhibits the adhesion of organs. Computational analysis indicated that a large number of microRNAs could bind to this fragment potentially including miR-199a*. Expression of versican and fibronectin, two targets of miR-199a*, are up-regulated in transgenic mice, suggesting that the 3′UTR binds and modulates miR-199a* activities, freeing mRNAs of versican and fibronectin from being repressed by miR-199a*. Confirmation of the binding was performed by PCR using mature miR-199a* as a primer and the targeting was performed by luciferase assays. Enhanced adhesion by expression of the 3′UTR was confirmed by
Human Genome Project identified approximately 25,000 protein-coding genes, occupying 1.9% of total genomic DNA. The remaining DNA has come to be known as “junk” DNA or cellular detritus and was presumed to serve no particular function because it does not code proteins. Pseudogenes are part of these non-functional DNAs since they are the defective copies of the protein-coding genes. In fact, pseudogenes are nearly as abundant as the protein-coding genes and therefore appear to be an important component of the genome. It has been reported that there are approximately 20,000 putative pseudogenes in the human genome
miRNAs are single-stranded RNA of 18–24 nucleotides in length and are generated by an RNase III-type enzyme from an endogenous transcript that contains a local hairpin structure
To study the effect of versican 3′UTR on modulating miRNA functions, we cloned and placed the 3′UTR under the control of the CMV promoter. The 3′UTR was transcribed as expected (
(a), A fragment of versican 3′UTR (700 bp) immediately after the stop codon containing a small fragment at the 3′ of versican coding sequence was inserted into the pcDNA3.1 plasmid, downstream of the CMV promoter, between XbaI and ApaI sites, producing Ver-UTR construct. Total RNA was prepared from one pooled cell line and two individual clones stably transfected with Ver3′UTR or the empty vector, subjected to RT-PCR, and analyzed on agarose gel electrophoresis. Expression of the 3′UTR was confirmed. (b) Genotyping PCR was performed on tail DNA extracted from the same litter of F1 using two pairs of primers amplifying the promoter region (CMV) and the downstream transcript versican 3′UTR. (c) Expression of the transgene was analyzed by RT-PCR using RNAs isolated from different organs of 3′UTR transgenic mice. (d) The levels of versican 3′UTR were analyzed by real-time PCR in Hek293 cells transfected with the 3′UTR construct or a control vector (detecting endogenous versican). The levels of 3′UTR in the 3′UTR-transfected cells were over 45-fold that of the control. (e) Photograph showing adhesion of the liver to the stomach. (f) Photograph showing organ adhesion occurred between liver and diaphragm. One piece of liver was adhered to the diaphragm. (g) A liver was adhered to the internal body walls of the mouse. (h) Paraffin sections of the adhesion tissues were stained with hematoxylin and eosin (H&E). Both photographs show the linking between the livers and connective tissues.
To study how versican 3′UTR affected organ adhesion, we examined its effects on cell lines stably expressing the 3′UTR. Human astrocytoma cell line U343 was stably transfected with the 3′UTR construct or an empty vector. The cells were cultured in DMEM containing 10% FBS at a cell density of 5×104 cells/ml and examined under a light microscope and photographed. The 3′UTR-transfected cells attached to tissue culture plates slower than the vector-transfected cells (
Vector- or 3′UTR-transfected cells were inoculated in tissue culture dishes overnight. Cell morphology was examined under a light microscope (a). Two days after cell inoculation, the 3′UTR-expressing cells exhibiting island-like morphology (b). The 3′UTR-transfected cells were transiently transfected with siRNA targeting the 3′UTR or a control sequence, followed by real-time PCR analysis of the 3′UTR levels (c, left). Cell adhesion of the two groups of cells was compared with the vector-transfected cells. Down-regulation of the 3′UTR levels increased cell adhesion (c, right). Typical results of cell adhesion are shown (d, left). Cell morphology was also examined. Down-regulation of the 3′UTR levels reversed the morphology (d, right). Luciferase reporter vector harboring the versican 3′UTR was co-transfected with the versican 3′UTR construct at different amount combined with a control vector in U343 cells. Increase rations of versican 3′UTR bound more endogenous miR199a* and thus freeing the translation of luciferase protein, resulting in higher levels of luciferase activities (e).
To confirm the direct effect of the 3′UTR, 4 different siRNAs complementary to the 3′UTR were synthesized. Down regulation of the 3′UTR was confirmed by real-time PCR amplifying a fragment of the 3′UTR (
We developed a number of experiments to analyze the effects of the 3′UTR. It was linked with the construct expressing versican G3 domain
The 3′UTR was also linked with the GFP expression unit (
Furthermore, the 3′UTR was linked to the luciferase report vector producing the lu-VUTR construct. Cells that were transfected with different concentrations of lu-VUTR always produced lower levels of luciferase activity than cells transfected with a control construct (
To test the direct interaction of miRNAs with the 3′UTR, we developed a PCR assay to test the potential binding interactions of the miRNAs with the 3′UTR and to validate the target sites
(a) Scheme for PCR method to test the interaction of miRNAs with versican 3′UTR. An oligonucleotide corresponding to miRNA-X is used as a reverse primer. It binds to the potential targeting sites on the antisense strand of the 3′UTR construct, depending on the extent of complementation. One forward primer docked on a different location of the vector was used to pair with the miRNA primer for PCR. (b) PCR products were obtained showing different size of products corresponding to the forward primer and the miRNA sequences. The expected sizes of PCR products are indicated with arrows. The miRNA sequences used were listed in Supporting
We analyzed the potential miRNAs that bind to the 3′UTR of the Ver-UTR construct. A number of candidates with low binding energy were detected. We tested 17 different miRNAs selected from the potential candidates for the 3′UTR of versican. As shown in
It is expected that one 3′UTR contains many miRNA binding sites and one miRNA might target many 3′UTRs. This would create a balanced network composing of the synergy and counteraction of miRNA-3′UTR interactions. The homeostatsis of miRNAs and miRNA-binding sites might be disrupted through changes in the expression of transcripts or miRNAs. Over-expression of the versican 3′UTR would affect the levels of free miRNAs through binding the miRNAs, which normally regulate versican expression by targeting its 3′UTRs. The formation of miRNA-3′UTR transcript duplex thus decreases the functional miRNA levels. This interaction would affect protein expression, leading to the functional consequences.
Using the computation algorithm FindTar (
To test whether the expression of versican 3′UTR affected versican expression, we prepared protein lysates from the brain, heart, kidneys, lungs, and spleen and analyzed versican expression by western blotting. Our experiments showed that the 3′UTR transgenic mice expressed higher levels of versican compared with the wild-type mice (
(a) Protein lysates were prepared from different organs and subjected to western blot analysis probed with anti-versican antibody. Detection of β-actin on the same membranes served as a loading control. Increased versican expression was detected in the organs harvested from the transgenic mice. (b–f) Paraffin sections of tissues from reproductive system (b), lung (c), liver (d), connective tissues (e), and rib (f) of the 3′UTR-transgenic and wild-type mice were stained with anti-versican antibody. In all sections shown, versican expression levels in the transgenic mice were higher as compared with the wild-type (arrows). (g) The levels of versican expression were higher in the pancreas that adhered to the liver (arrows). (h) The levels of versican expression were higher in the junctions between liver and the surrounding connective tissues (arrows). In the lower panel identified by the circle, some liver tissues (solid arrow) and connective tissues (open arrow), stained with anti-versican antibody, were mixed up, and no border could be identified between the different tissues. (i) Versican 3′UTR (nucleotides 275–299 of the 3′UTR, Upper) was found to be the potential target of
To confirm if versican was a target of miR-199a*, U343 cells were co-transfected with versican 3′UTR-luciferase construct (lu-VUTR) or the mutant lu-Ver-mut, in which the miR-199a* target site was mutated by nucleotide replacement (
To test whether the expression of versican 3′UTR affected fibronectin levels, we prepared protein lysates from the brain, heart, kidneys, lungs, and spleen and analyzed fibronectin expression by western blotting. Our experiments showed that the 3′UTR transgenic mice expressed higher levels of fibronectin compared with the wild-type mice (
(a) Protein lysates were prepared from different organs and subjected to western blot analysis probed with anti-fibronectin antibody. Detection of β-actin on the same membranes served as a loading control. Increased fibronectin expression was detected in the organs harvested from the transgenic mice. (b–f) Paraffin sections of tissues from spleen (b), brain (c), connective tissue (d), liver (e), and rib (f) of the 3′UTR-transgenic and wild-type mice were stained with anti-fibronectin antibody. In the transgenic spleen, the connective tissue structures expressed higher levels of fibronectin than the wild-type tissues did (arrows). In the transgenic brain, fibronectin expression was higher in the blood vessels (arrows). In the connective tissues of the transgenic mice, some areas expressed higher levels of fibronectin as compared with the wild-type (arrows). In the transgenic liver, fibronectin expression was higher along the edges of the liver than the wild-type liver. (g) The levels of fibronectin expression were higher in the 3′UTR pancreas that adhered to the liver (arrows). The connective tissues that adhered to the livers also expressed high levels of fibronectin. (h) In a different mouse, the 3′UTR connective tissues, while expressed high levels of fibronectin, were strongly adhered with the liver. Pulling out the connective tissue severely damaged the liver surface (upper panel, arrow). In the areas identified by the circles, some liver tissues (solid arrow) and connective tissues (open arrow, stained with anti-fibronectin antibody) were completely merged. (i) Adhesion of different organs was detected between liver/liver, liver/muscle, liver/stomach, and pancreas/stomach. The junctions between the organs expressed high levels of fibronectin (arrows). (j) Fibronectin 3′UTR (nucleotides 663–683 of the 3′UTR, Upper) was found to be the potential target of
To confirm that fibronectin was a target of miR-199a*, we cloned the fragment of fibronectin 3′UTR containing the miR-199a* target site. The fragment was inserted into the luciferase reporter vector producing a construct lu-FNUTR. The potential target site of miR-199a* was mutated producing lu-FNmut. U343 cells were co-transfected with fibronectin 3′UTR-luciferase construct (lu-FNUTR) or the mutant lu-FNmut (
Our experiments indicated that the functions of miRNAs can be regulated by a fragment of non-coding transcript. Genomic deletion/truncation leading to translational silencing produces mutant phenotypes as the consequence of protein loss/mutation accompanied by the existence of non-coding/mutation transcripts. This strategy has been extensively used to knock-out genes of interest in studying gene functions. After gene knock-out, the protein is no longer expressed, but it is conceivable that the mutated genes are still able to produce non-coding transcripts. Sometimes, no detectable phenotypes are obtained, and it is said that the mutated/lost proteins may be compensated with others. Although compensation by other proteins is possible, the non-coding transcript may play an important role in compensation and balancing the mutant entity. Our results demonstrate a dramatic functional consequence by expressing a non-coding transcript. Exogenous expression of the 3′UTR construct altered the expression of some proteins functionally associated with the 3′UTR. It is possible that the 3′UTR can play more diverse roles than the protein expressed by the same transcript, although proteins are the executants of biological activities. Therefore, while analyzing the results of gene knock-out, one may need to consider the effects of not only the proteins but also the remaining non-coding transcripts. Our results that exogenous expression of the versican 3′UTR promoted versican expression suggest that each mRNA may exert at east two functional roles: through protein translation and miRNA regulation.
The human genome contains a large number of pseudogenes, which are nearly as abundant as the functional genes and therefore appear to be an important component in the genome. It has been reported that there are approximately 20,000 putative pseudogenes in the human genome
It seemed that expression of the 3′UTR produced a similar functional role as the miRNA inhibitor. However, one 3′UTR has the capacity to modulate multiple miRNAs, while one miRNA inhibitor can only affect one miRNA. In this sense, a 3′UTR may be able to exert diverse biological activities by modulating multiple miRNA functions. As such, an animal gene, with long sequence of the 3′UTR, may have the capacity of exerting complex biological activities. Furthermore, a long fragment of 3′UTR may be more stable, while the miRNA inhibitor may be readily degraded. Thus, expression of a 3′UTR may have great advantage in modulating cell activities. In terms of stability and functionality, the 3′UTR may be better than normal mRNA, in that a mRNA has many tasks to carry out, while a 3′UTR may only exist for miRNA binding. In the former case, binding with multiple factors involved in protein synthesis and formation of secondary structures would decrease the accessibility for miRNAs. In the latter case, a simple 3′UTR fragment would be much more accessible for miRNA binding. This may explain why expression of a non-coding fragment could serve as a vigorous tool and produce potent effects
To study the effect of versican 3′UTR on cell activities, we have cloned the 3′UTR by RT-PCR using two primers Huver-UTRNXbaI and Huver-UTRCApaI. The PCR product was digested with restriction enzymes XbaI and ApaI and inserted into XbaI- and ApaI-opened pcDNA3.1 vector. After transformation, colony selection, DNA mini-preparation, and restriction enzyme digestion, the correct clones were sequenced to ensure identity of the 3′UTR.
A luciferase reporter vector (pMir-Report; Ambion) was used to generate the luciferase constructs. The 3′UTR of versican was cloned using 2 primers, HuverUTR-NSpeI and HuverUTR-CHindIII, by PCR. The PCR products were then digested with SpeI and HindIII and the fragment was inserted into a SpeI- and HindIII-digested pMir-Report Luciferase plasmid (Ambion), to obtain a luciferase construct, lu-VUTR. Primers used in this study are listed in the Supporting
A fragment of the 3′UTR of fibronectin was also cloned using 2 primers, FN1-N3′SacI and FN-199aC3′MluI, by RT-PCR. The PCR products were then digested with SacI and MluI and the fragment was inserted into a SacI- and MluI- digested pMir-Report Luciferase plasmid (Ambion), to obtain a luciferase construct, lu-FNUTR. A mutant construct was generated with two primers FN1-N3′SacI and FN-199aC3′MluI-Mut using similar approach.
To serve as a negative control, a non-related sequence was amplified from the coding sequence of the chicken versican G3 domain using 2 primers, chver10051SpeI and chver10350SacI. It is expected that there is no endogenous miRNA bind to this fragment as it is in the coding region. The PCR product was then inserted into a SpeI- and SacI-digested pMir-Report Luciferase plasmid.
For
The transgene was released from the plasmid by digestion with ApaLI and StuI. The digested product was fractionated by agarose gel electrophoresis and the 3 kb transgene fragment was excised from the gel, purified by Elutip mini-column (Schleicher and Schuell, Keene, NH) and then resuspended in injection buffer (10 mM Tris, pH 8.0 and 0.1 mM EDTA) at a concentration of 1 to 2 ng/µl. The transgene was microinjected into the male pronuclei of C57BL/6×CBA F2 mouse zygotes. Injected embryos were implanted into the oviducts of pseudopregnant recipient females using a standard protocol approved by the Animal Use Subcommittee of the University Council on Animal Care, The University of Western Ontario. Transgenic founder lines were maintained by backcrossing with C57BL/6×CBA F1 mice. Genotyping was performed by PCR, using primers EGFP-347F pairing with EGFP-668R (for CMV promoter and huver10861F pairing with Huversican-UTRCApaI (for versican 3′UTR), and tail snip or ear punch DNA as template. GAPDH served as a control using primers mo-Gapdh1F and mo-Gapdh250R. The transgenic mice were then transferred to Sunnybrook Research Institute (Toronto, Ontario). The methods for tissue harvest and analysis have been approved by the Animal Care Committee of Sunnybrook Research Institute, Ontario, Canada.
Vector- or 3′UTR-transfected cells were plated onto culture dishes at a density of 4×105 cells/ml and incubated for 30 min with DMEM containing 5% FBS. After 30 min, cells were fixed with 3.7% paraformaldehyde. Adhering cells were counted and cell images were captured using a phase-contrast microscope. Ten different fields (100×) were used for cell counting.
Organs were weighted and homogenized with lysis buffer containing protease inhibitors (150 mM NaCl, 25 mM Tris-HCl, pH 8.0, 0.5 M EDTA, 20% Triton X-100, 8 M Urea, and 1× protease inhibitor cocktail). Protein concentration was measured by Bio-Rad Protein Assay kit (#5000-0006). The lysates were subjected to SDS-PAGE and then transferred to nitrocellulose membranes probed with a primary antibody against versican (Lifespan Biosciences, LS-C25140), fibronectin (BD, 610078), or β-actin (Sigma-Aldrich) overnight at 4°C. After incubation with corresponding HRP-conjugated secondary antibodies, the membranes were washed, followed by detection with the ECL kit.
Organs were freshly excised and fixed in formalin overnight, immersed in 70% ethanol, embedded in paraffin, and sectioned by a microtome (Leica RM2255). The sections were de-paraffinized with xylene and ethanol and then boiled in a pressure cooker. After washing with Tris-Buffered-Saline (TBS) containing 0.025% Triton X-100, the sections were blocked with 10% goat serum and incubated with primary antibody against versican, fibronectin, or collagen Iα1 (Santa Cruz, sc-25974) in TBS containing 1% bovine serum albumin (BSA) overnight. The sections were washed and labeled with biotinylated secondary antibody, followed by avidin conjugated horse-radish peroxidase provided by the Vectastain ABC kit (Vector, PK-4000). The staining was developed by DAB kit (Vector, SK-4100). The slides were subsequently stained with Mayer's Hematoxylin for counter staining followed by slide mounting.
U343 cells were seeded onto 24-well tissue culture plates at a density of 3×104 cells/well in DMEM containing 10% FBS and maintained at 37°C for 24 hrs following the methods described by us recently
The results (mean values±SD) of all the experiments were subjected to statistical analysis by
(a) Primers used in this study. (b–c) Photographs showing organ adhesion occurred between liver and stomach (b), between liver and body (c) in a different transgenic line of mice. (d) Vector- or the 3′UTR-transfected cells were inoculated in tissue culture dishes for 2.5 hours. Cell adhesion was examined under a light microscope and photographed.
(8.01 MB TIF)
a, Upper, to test the effect of the versican 3′UTR, the versican G3 domain was linked with or without the 3′UTR producing G3 and G3-UTR constructs. Lower, cell lysates prepared from U343 cells stably transfected with the G3 and G3-UTR constructs were subjected to Western blot analysis probed with anti-G3 and anti-actin antibodies simultaneously. While actin levels were similar, G3 levels were much lower in cells transfected with the G3-UTR construct. Fig S2b, the GFP coding sequence was linked with or without the 3′UTR producing GFP and GFP-UTR constructs (Upper). Cells transfected with the GFP-UTR construct produced lower levels of GFP activities than that transfected with the GFP construct. The levels of fluorescent cells were quantified (Middle). Typical fluorescent levels of U87 and U343 cells transiently transfected with the GFP and GFP-UTR constructs were shown (Lower). Fig S2c, Cells transfected with the GFP and GFP-UTR constructs were also examined under a light and fluorescent microscope. Typical results are shown.
(8.51 MB TIF)
(a) U343 cells were transiently transfected with luciferase reporter vector harboring the versican 3′UTR (lu-VUTR) or a control sequence (ctrl). Luciferase activities were normalized using the control as 100%. The luciferase activities of lu-VUTR never reached the levels of the control, suggesting endogenous miRNAs targeting the versican 3′UTR. Nevertheless, the activities increased with higher does of plasmids, suggesting that increased supplies of versican 3′UTR absorbed some endogenous miRNAs freeing luciferase translation. (b) Luciferase reporter vector harboring the versican 3′UTR was co-transfected with the versican 3′UTR construct at different amount combined with a control vector in U87 cells. Increase amounts of versican 3′UTR bound more endogenous miR199a* and freeing the translation of luciferase protein, resulting in higher levels of luciferase activities. (c) PCR was performed using one forward primer docked on the vector and one of the mature miRNAs as indicated at a different temperature (35°C). PCR products were obtained showing different sizes of products corresponding to the forward primer and the miRNA sequences. (d) Photographs showing organ adhesion occurred between stomach and connective tissues. The sections were immunostained with anti-versican antibody showing that versican was deposited in the adhesion junction areas. (e) The adhesion tissues were sectioned and immunostainined with anti-type I collagen that normally deposits in wound healing areas. Collagen was expressed at high levels in the areas of tissue adhesion.
(9.01 MB TIF)
Paraffin sections of adhesion organs from a different transgenic line of mice were stained with anti-fibronectin antibody. The levels of fibronectin expression were higher in the adhesion junctions between liver and pancreas (a), between liver and connective tissue (b–e), and between liver and liver (e, right). Luciferase reporter vector harboring the fibronectin 3′UTR was co-transfected with the versican 3′UTR construct at different amount combined with a control vector in U343 cells. Increased ratios of versican 3′UTR bound more endogenous miR199a* and thus freeing the translation of luciferase protein, resulting in higher levels of luciferase activities (f).
(9.31 MB TIF)
We thank Dr. Jennifer Ma at the Core Facilities of Sunnybrook Research Institute for her assistance in real-time PCR experiments. We thank Dr. Siu-Pok Yee and Dr. Sara Gatchell for their assistance in the generation of transgenic mice.