to BioTechniques free email alert service to receive content updates.
A simple high-throughput technology enables gain-of-function screening of human microRNAs
Wen-Chih Cheng1,2, 3, 4, Tami J. Kingsbury1,3, Sarah J. Wheelan5, and Curt I. Civin1,2, 3, 4
Full Text (PDF)
Supplementary Material

Figure 1.  Design of nested PCR strategy and miR-HTS. (Click to enlarge)

Cell growth assay

One day prior to transfection, IMR90 cells were seeded into 96-well plates (3000 cells per well). MiR mimics (5 or 10 nM, Dharmacon, Lafayette, CO, USA) were transfected using 0.15 μL DharmaFECT2 (Dharmacon) per well in triplicates. On day 5 post-transfection, 10 μL alamarBlue (Life Technologies) was added to each well and incubated for 4 h. The alamarBlue fluorescence signal was read on a VictorX3 plate reader (PerkinElmer, Waltham, MA, USA) using 530/578 nm excitation/emission filters. Caspase activity

Cells were transfected using the same protocol as for the cell growth assay. Caspase activity was measured 4 days post-transfection using the Apo-ONE homogenous caspase 3/7 assay (Promega, Madison, WI, USA) according to the manufacturer's instructions. Results and discussion

To interrogate the roles of the human miRNome in regulating cellular processes, we developed a functional high-throughput screening (HTS) method that combines a pooled miR overexpression lentiviral (lenti-miR) library with a novel panel of qPCR assays to quantitate each lenti-miR in the library. Lentiviral vectors allow screening in either dividing or quiescent cell types (e.g., stem cells). Infections with the lenti-miR library as a pool make it a scalable screening procedure. In addition, qPCR is widely practiced with standard data-analysis protocols, in contrast to more complex data analyses necessary for custom microarray or high-throughput sequencing. The lenti-miR library used in this study contains 578 lentiviruses encoding 539 individual miRs and 39 miR clusters. Each lentivector contains a miR cassette consisting of the native human miR hairpin, plus a stretchof 200–400 bp of genomic flanking sequences to allow for endogenous miR processing. The expression of each miR cassette is controlled by a CMV promoter; an EF1α promoter-driven GFP is located downstream of the miR cassette (Figure 1A). Transduced cells co-express the miR and GFP, allowing flow cytometric quantification of transduced cells.

As a proof-of-concept evaluation of the miR-HTS strategy, we screened for miRs capable of regulating cell “growth” (i.e., proliferation and/or survival) of the IMR90 primary human lung fibroblast cell line (Figure 1B). We chose IMR90 for this proof-of-principle study for three reasons. First, miR-34a was already known to potently inhibit IMR90 growth (28), providing a positive control. Second, performing the functional screen in this nonimmortalized cell line might generate a more comprehensive set of human miRs involved in growth than would be obtained using a transformed cell line with deregulated growth control. Third, defective miR processing has been observed in many cancer cell lines (29-31). Approximately 3 million IMR90 cells were transduced with the pooled lenti-miR library at ~20% efficiency (i.e., MOI = 0.2) and then cultured for 1 month. This MOI should result in <5% of infected cells containing >1 lentiviral insertion (32), thus avoiding combinatorial effects due to multiple transduced miRs in a single cell (33, 34). Infecting 20% of ~3 million cells predicts that each miR lentivirus will integrate into ~1000 cells. Each cell transduced should have an independent genomic integration site (35), reducing the chance of detecting a lentiviral insertion site–specific effect (e.g., host gene disruption). In addition, the many different integration sites for each miR lentivirus should average positional effects on miR expression (e.g., clonal silencing of miR expression) (36).

Identifying the subset of miRs whose overexpression enhanced or reduced cell growth via the miR-HTS requires reliable detection of the changes in representation (i.e., number, abundance or relative frequency) of the cells containing each lenti-miR transgene within the entire transduced population over time in culture. Therefore, we developed the GRE-qPCR assay to quantify the abundance of cells infected with each lenti-miR by using the unique miR cassette sequence as a barcode. Each GRE-qPCR assay has three components (Figure 1A): (i) an internal miR-specific forward primer to identify each miR cassette, (ii) an internal common reverse primer, and (iii) an internal common probe. The internal common reverse primer and probe detect every lenti-miR transgene in the library, but not the endogenous miR locus of the host genome. For high-throughput analysis, each of the 578 different internal miR-specific forward primers for the GRE-qPCR assays was plated in a well of a 384-well plate.

Before conducting the actual miR-HTS, we determined the dynamic range and sensitivity of GRE-qPCR assays, and the amount of genomic DNA required for the GRE-qPCR assay to quantitate cell number accurately. First, three cultures of IMR90 cells were separately infected with miR-155, miR-222, or vector-only lentivirus at an MOI = 0.2, to mimic the miR-HTS conditions. Specific numbers of miR-155- and miR-222-infected cells were mixed at desired ratios to provide a 4-log range of different cell numbers infected with either of these lenti-miRs (i.e., 10–105 infected cells). As there were 578 different miR cassettes in the library, the expected relative frequency of each specific miR cassette was 1 per 578 transduced cells. To mimic this expected frequency of each miR cassette/barcode in the library and the potential changes in frequency (under- or over-representation) during the 1 month miR-HTS culture, we added vector only–transduced cells at numbers ~20- to 1800-fold higher than that of lenti-miR virus infected cells (see Supplementary Table S2 for details of the different cell mixtures prepared). Genomic DNA was extracted from each mixture and subjected to GRE-qPCR assays for miR-155 and miR-222, as well as qPCR assay for GFP (loading control present in every lentiviral integrant). The corrected cycle of threshold (Ct, corrected for template amounts) values of qPCR assays were plotted against the numbers of cells infected with each lenti-miR, in mixtures containing 10–105 cells transduced with the given miR cassette (Figure 2A, B). As expected, we saw a log-linear inverse correlation between the number of infected cells and the correct Ct values. The fitted regression line of the GRE-qPCR assay for miR-155 lost linearity below 50 cells, indicating a detection limit of 50 cells (Figure 2A; regression lines fitted to values with 50–105 cells or 10–105 cells had r2 = 0.94 or 0.84, respectively). The fitted regression line of miR-222 maintained linearity to as low as 10 cells (Figure 2B; regression line fitted to values with 10–105 cells had r2 = 0.98). We determined that 1 or 1.5 μg genomic DNA template per GRE-qPCR assay was required to accurately quantify the numbers of lenti-miR infected cells over a dynamic range of 50–105 infected cells, when 20% of the cell population was transduced in these model experiments.

  1    2    3    4    5