Described here is a protocol for characterizing modules of biologically synergistic miRNAs and their assembly into short transgenes, which allows simultaneous overexpression for gene therapy applications.
The biological relevance of microRNAs (miRNAs) in health and disease significantly relies on specific combinations of many simultaneously deregulated miRNAs rather than the action of a single miRNA. The characterization of these specific miRNAs modules is a fundamental step in maximizing their use in therapy. This is extremely relevant because their combinatorial attributes can be practically exploited. Described here is a method to define a specific miRNA signature relevant to the control of oncogenic chromatin repressors in glioblastoma. The approach first defines a general group of miRNAs that are deregulated in tumors in comparison to normal tissue. The analysis is further refined by differential culture conditions, underscoring a subgroup of miRNAs that are co-expressed simultaneously during specific cellular states. Finally, the miRNAs that satisfy these filters are combined into an artificial polycistronic transgenes, which is based on a scaffold of naturally existing miRNA clusters genes, then used for overexpression of these miRNA modules into receiving cells.
miRNAs offer an unmatched opportunity for the development of a broad gene therapy approach to many diseases1,2,3, including cancer4,5. This is based on several unique features of these biological molecules, including their small size6, simple biogenesis7, and natural tendency to function in association8. Many diseases are characterized by specific miRNA expression patterns, which often converge on the regulation of complex biological functions9. The purpose of this method is first to define a strategy to identify groups of miRNAs that are synergistically relevant for specific cellular functions. Consequently, it provides a strategy for the re-establishment of such miRNA combinations in downstream studies and applications.
This method allows for functional analysis of multiple miRNAs at once, leveraging on their simultaneous targeting of a large number of mRNAs, thus recapitulating the complex landscapes of diseases. This approach has been recently employed to define a group of three miRNAs that 1) are simultaneously downregulated in brain cancer and 2) show a strong co-expression pattern during neural differentiation as well as in response to genotoxic stress by radiation or a DNA alkylating agent. The combinatorial re-expression of this module of three miRNAs by the clustering method described below results in profound interference with the biology of cancer cells and can be easily used as a gene therapy strategy for preclinical studies10. This protocol may be of particular interest to those involved in miRNA research and its translational applications.
1. Characterization of Functionally Associated miRNAs in Glioblastoma
2. Assembly of miRNA Modules into a Polycistronic Transgenic Cluster
3. Obtaining Transgenes by DNA Synthesis
This method allowed characterization of a module of three miRNAs that are consistently downregulated in brain tumors, which are co-expressed specifically during neuronal differentiation (Figure 1) and involved in the tumor survival response after therapy (Figure 2). This is accomplished by regulating a complex oncogenic chromatin repressive pathway. This co-expression pattern suggested a strong synergistic activity among these three miRNAs (Figure 3). Consequently, taking advantage of the small size and simple biogenesis of miRNAs, the second part of this protocol was used to design a transgene (Figure 4) that could simultaneously recapitulate the expression of the three miRNAs in glioblastoma cells, both in vitro and in vivo, with significant interference in tumor biology and promising translational applicability (Figure 5A,B,C)10. Additionally, it was demonstrated that the transgenic cluster is also functional in a breast cancer model (Figure 5D,E).
Figure 1: Characterization of a functional miRNA module in glioblastoma. (A) Volcano plot representing the differentially expressed miRNAs in human glioblastoma samples (n = 516) vs. normal brain (n = 10) obtained from TCGA. In green are miRNAs with >4-fold difference. The red circle represents the 10 most significantly downregulated miRNAs in glioblastoma. (B) Relative expression of the 10 miRNAs selected in (A) during induction of different differentiation pathways in neural stem cells, showing clear upregulation of the miR-124, miR-128, and miR-137 modules during induction of neural differentiation. Mean ± SD from three biological replicates (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; Student's t-test, two-tailed). This figure has been modified with permission from reference10. Please click here to view a larger version of this figure.
Figure 2: Confirmation of co-expression patterns of miRNA modules during genotoxic stress. Relative expression of the three miRNAs defined in Figure 1 in multiple glioblastoma cells and cell lines (G62-mesenchymal; MGG4-proneural; U251, U87 pro-neural-like, and T98G mesenchymal-like glioblastoma cell lines) after induction of resistance to temozolomide (TMZ, pink bars) or ionizing radiation (RT, green bars). Reported are means with SD from two independent replicates. This figure has been modified with permission from reference10. Please click here to view a larger version of this figure.
Figure 3: Analysis of functional convergence of the targetomes of co-expressed miRNAs. Venn diagram output from the Bioinformatics and Evolutionary Genomics website, simultaneously crossing the targetome of labelled miRNAs with the mRNAs constituting a GO category of interest (in this case, chromatin repressors). mRNAs uniquely targeted by single miRNAs were chosen for further downstream functional studies. Please click here to view a larger version of this figure.
Figure 4: 2D structure of an engineered miRNA sequence encoding the three miRNAs cluster. Graphical output from the RNAweb Fold program. Note the presence of three well defined stem-loop structures which represent the hairpins of each respective miRNA (miR-124, miR-128, miR-137) encoded by the transgene. Please click here to view a larger version of this figure.
Figure 5: Evidence of transgene processing and its downstream biological effect. (A) Relative quantification of miRNA expression after lentiviral-mediated transduction of G34 glioblastoma cells with the transgenic miRNA cluster (Cluster 3) or negative control (ctrl). Reported are means from three independent experiments ± SD. (B) Fluorescence microscope pictures of G34 glioblastoma spheres expressing negative control transgene vs. Cluster 3 transgene. Scale bar = 100 µm. (C) In vivo growth of intracranial human G34 cell xenografts expressing either control or Cluster 3 transgenes. Scale bar = 1 mm. This figure has been reproduced with permission10. (D) Relative quantification of miRNA expression after lentiviral-mediated transduction of MDA-MB-231 breast cancer cells with the transgenic miRNA cluster (Cluster 3) or negative control (ctrl). (E) Fluorescence microscope images of MDA-MB-231 breast cancer cells expressing negative control transgene vs. Cluster 3 transgene. Scale bar = 100 µm. All experiments were performed in triplicates (*p < 0.05; **p < 0.01; Student's t-test, two-tailed, multiple comparisons). Please click here to view a larger version of this figure.
Supplementary Figure 1: TargetScan workflow. (A) Home page screenshot, showing selection options for miRNA search. (B) Representative search results for miR-137. List of target genes is in the left column. Red box denotes genes with conserved targeting sites (suggesting higher confidence of real targeting). Please click here to view this figure. (Right-click to download.)
Supplementary Figure 2: ToppGene Suite workflow. (A) Home page screenshot showing the search box where the list of genes to be analyzed is inserted. (B) Representative search result for the miR-137 targetome, showing the most statistically significant Gene Ontology (GO) categories. Please click here to view this figure. (Right-click to download.)
This protocol is based on the notion that rather than functioning in isolation, miRNAs are biologically relevant by working in groups, and these groups are transcriptionally determined by specific cellular contexts26. To justify this approach from a translational perspective, a follow-up protocol that allows recreation of this multi-miRNA pattern in cells/tissues is introduced. This is possible by taking advantage of the relatively simple biogenesis of miRNAs, whereby the recognition of the characteristic miRNA hairpin by microprocessor is necessary and sufficient for correct miRNA processing27. At the same time, this minimal requirement allows use of the genetic scaffold of naturally occurring miRNA clusters as a backbone for the expression of desired miRNA modules, which are contained within short DNA sequences that can be fitted into any delivery vectors of choice. The major requirement of this protocol is maintaining the hairpin structure and sufficient length of the stem component to allow appropriate cleavage.
There are two major critical considerations regarding the execution of this protocol. The first point is the accurate determination of the optimal miRNA combinations. This is determined by careful analysis of not only the miRNA expression signature of cells or tissues compared to controls, but also of any simultaneous expression changes observed as the cells are experimentally manipulated. Once a set of miRNAs is defined, it is also fundamental to ascertain that the artificial modulation of each singularly does not modify expression of the others.
The second critical aspect concerns the construction of chimeric sequences to artificially recapitulate this functional miRNA clustering. It is fundamental to abide by the structural requirements of the miRNA processing machinery, which is the presence of a long enough stem sequence (measuring at least 11 nucleotides) at the origin of each miRNA hairpin18, as well as maintenance of original spacing sequences of the native miRNA cluster scaffold. In our experience, fulfilling these two requirements has consistently yielded appropriate RNA folding (Figure 4) and resulted in successful multi-miRNA expression.
The major limitation of this technique is the finite number of miRNAs that can be clustered together into a functional transgene. We have successfully engineered sequences overexpressing up to six different miRNAs but observed some decrease in processing efficiency as the number of hairpins increases. So far the genetic structure of the miR-17-92 cluster has been used, because it is the one encoding the highest number of hairpin (six) within the shortest DNA sequence (~800 base pairs)8. As there are other natural miRNA clusters, it is anticipated that they could also be used for this purpose28. Finally, it has been observed that modification of spacing sequences among the hairpin decreases their processing, so there are some constraints regarding to what extent the native structure can be modified.
The most significant and advantageous aspect of this proposed method is that it allows the engineering of transgenes that are able to simultaneously overexpress multiple desired miRNAs mainly using an in silico approach, limiting the need for tedious and time-consuming molecular cloning steps, as previously described29,30,31. The protocol described is easy to execute and does not require specialized equipment or skills.
In consideration of the recognized importance of miRNAs in molecular biology and their potential in therapeutic applications, this protocol is of interest to a large audience of researchers. This approach is expected to encourage future studies that focus on the combinatorial properties of miRNAs and will serve as a simple and robust tool for their execution. More importantly, these clustered transgenes represent ideal cargos for gene therapy vectors. Evidence of successful in vivo delivery of a 3-miRNA cluster has already been obtained via direct intratumoral intracranial injection of AAV vectors (unpublished data). It is thus anticipated that this technique will significantly boost the translational aspects of miRNA research.
The authors have nothing to disclose.
The authors wish to thank the members of the Harvey Cushing Neuro Oncology Laboratory for support and constructive criticism. This work was supported by NINDS grants K12NS80223 and K08NS101091 to P. P.
0.4% low melting temperature agarose | IBI Scientific | IB70058 | |
0.45 µM sterile filter unit | Merck Millipore | SLH033RS | |
1.5-mL Microcentrifuge tube | Eppendorf | 22431081 | |
6-Well plates | Greiner Bio-One | 657160 | |
Athymic mice (FoxN1 nu/nu) | Envigo | 069(nu)/070(nu/+) | |
B-27 Supplement | Thermo Fisher Scientific | 12587010 | |
Cell culture flask | Greiner Bio-One | 660175 | |
Cell Scraper, 16cm | Sarstedt | 83.1832 | |
Cesium 137 irradiator | JL Sheperd and Associates | Core Facility (Harvard Medical School) | |
Chloroform | Sigma-Aldrich | 439142-4L | |
DMEM, high glucose, pyruvate | Thermo Fisher Scientific | 11995040 | |
Dulbecco’s phosphate-buffered saline | Gibco | 14190144 | |
Eosin Y solution | Sigma-Aldrich | E4009 | |
Fetal Bovine Serum | Sigma-Aldrich | F9665 | |
Formalin solution | Sigma-Aldrich | HT501128 | |
GlutaMAX Supplement | Thermo Fisher Scientific | 35050061 | |
HEK-293 | American Type Culture Collecti | ATCC CRL-1573 | |
Hematoxylin solution | Sigma-Aldrich | 1051750500 | |
Human primary glioma stem-like cells (GBM62) | Provided by Dr. E. A. Chiocca (Brigham and Women’s Hospital, Boston, MA) | ||
Human primary glioma stem-like cells (MGG4) | Provided by Dr. Hiroaki Wakimoto (Massachusetts General Hospital, Boston, MA) | ||
Lentiviral vector pCDH-CMV-MCS-EF1-copGFP | System Biosciences | CD511B-1 | |
Lipofectamine 2000 | Thermo Fisher Scientific | 11668019 | |
Microcentrifuge refrigerated | Eppendorf | model no. 5424 R, cat. no.5404000138 | |
Mounting medium | Thermo Fisher Scientific | 4112APG | |
Nalgene High-Speed Polycarbonate Round Bottom Centrifuge Tubes | Thermo Fisher Scientific | 3117-0380PK | |
NanoDrop | Thermo Fisher Scientific | 2000c | |
Neural Progenitor cells (NPC) | Provided by Dr. Jakub Godlewski (Brigham and Women’s Hospital, Boston, MA) | ||
Neurobasal Medium | Thermo Fisher Scientific | 21103049 | |
Nikon eclipse Ti motorized fluorescent microscope system | Nikon, Japan | 14314 | |
Opti-MEM | Thermo Fisher Scientific | 31985088 | |
PCR tubes | Sigma-Aldrich | CLS6571-960EA | |
Penicillin-Streptomycin | Thermo Fisher Scientific | 15140122 | |
Petri-Dishes 94/16 | Greiner Bio-One | 632180 | |
Poly-D-Lysine | Sigma- Aldrich | P4707 | |
Recombinant Human EGF | PeproTech | AF-100-15 | |
Recombinant Human FGF-basic | PeproTech | AF-100-18B | |
Retinoic acid | Gibco | 12587-010 | |
RNA Miniprep Kit | Direct-zol | R2050 | |
S1000 Thermal Cycler | Bio-Rad | 1852196 | |
Small Animal Image-Guided Micro Irradiator | Xstrahal Life Sciences, UK | Core facility (Dana-Farber Cancer Institute, Boston, MA) | |
Sorvall WX+ Ultracentrifuge | Thermo Fisher Scientific | 75000100 | |
StemPro Accutase | Thermo Fisher Scientific | A1110501 | |
StepOne Real-Time PCR System | Applied Biosystems | 4376357 | |
SterilGARD biosafety cabinet | The Baker Company | SG403A-HE | |
Sucrose | Sigma-Aldrich | S9378 | |
T98-G | American Type Culture Collecti | ATCC CRL-1690 | |
TaqMan MicroRNA Reverse Transcription Kit | Thermo Fisher Scientific | 4366596 | |
TaqMan Universal PCR Master Mix | Thermo Fisher Scientific | 4324018 | |
Temozolomide | Tocris Bioscience | 2706 | |
Tissue-Tek optimum cutting temperature | Fisher Scientific | NC9636948 | |
TRIzol Reagent | Thermo Fisher Scientific | 15596026 | Lysis reagent |
U251-MG | American Type Culture Collecti | ATCC HTB-17 | |
U87-MG | American Type Culture Collecti | ATCC HTB-14 | |
ViraPower Lentivector Expression system | Thermo Fisher Scientific | K4970-00 | |
Water, HPLC grade | Fisher | W54 | |
Xylene | Sigma-Aldrich | 534056 |