Small interfering RNA production by enzymatic engineering of DNA (SPEED)

Abstract

Small interfering RNAs (siRNAs) potently silence expression of target genes. In principle siRNA libraries can be used to perform effective genome-scale functional genetic screens in mammalian cells, but their development has been hampered by the need to chemically synthesize thousands of oligonucleotides and to incorporate them into expression vectors. We have developed a technology to efficiently convert a double-stranded cDNA library into a retroviral siRNA library in which 21-base siRNAs are produced in infected cells at high levels and efficiently block expression of their target genes. The key steps are the generation of random cDNA fragments that are fused to a hairpin linker, cleavage with the MmeI endonuclease that creates 20- to 21-bp cDNA fragments, conversion to a double-stranded DNA that contains two copies of the cDNA insert in a head-to-head palindrome, and insertion of the construct downstream of a polymerase III promoter. We constructed a siRNA library with 3 x 10(6) clones from a mouse embryo cDNA library; siRNAs were found against many different genes; and multiple siRNAs can be generated from a single mRNA. We further showed that specific siRNAs were efficiently produced in stably infected mammalian cells and resulted in significant and specific reduction of their target mRNAs. Because no prior knowledge about target transcripts is needed, a cDNA-derived siRNA library will generate siRNAs against unknown transcripts and genes. Finally, cDNA-derived siRNA libraries can be readily generated from any cell type or species, enabling genome-wide functional screens in many biological systems.

Fig. 1.

Schematic outline for construction of a siRNA library. Step A, hairpin linker attachment; step B, conversion of extended hairpins into palindromic dsDNAs; step C, cloning into the retrovirus expression vector; step D, creation of a hairpin loop. Details are given in Materials and Methods and Results.

Fig. 2.

Effects of promoter modification. (A) Expression of PU.1 siRNA causes loss of infected cells. UT7 cells were infected with the indicated retroviruses, and the percentage of infected GFP-positive cells, among the total propidium iodide-negative live cells, was measured by FACS at various times after infection. Plotted is the ratio of GFP-positive to -negative cells normalized to the ratio at 50 h postinfection. (B) Specific reduction of PU.1 protein. UT7 cells were infected with the indicated viruses. Two days after infection, GFP-positive cells were sorted and total cell proteins were used for Western blots; the same membrane was blotted with PU.1 and Stat3 antibodies. U6-AAAA-PU718-8 expresses PU718 siRNA with an eight-base palindrome hairpin from a U6 promoter with AAAA as the last bases.

Fig. 3.

Effects of palindrome loop lengths on siRNA function. (A) Expression in NIH 3T3 cells. Cells were infected with the indicated viruses, and after 2 days GFP-positive cells were sorted by FACS and propagated. Northern blots were performed with total cellular RNA (22); the membrane was sequentially hybridized with probes for siRNAs targeting the PU718 site (5′-GAAGCTCACCTACCAGTTC-3′) and for endogenous U6 RNA. (B) Functional expression in UT7 cells. UT7 cells were infected with the indicated viruses, and the relative number of siRNA-infected cells were calculated as described in Fig. 2, except that the ratio of GFP-positive cells to GFP-negative cells was normalized to the ratio at 39 h postinfection.

Fig. 4.

Specific inhibition of CD53 and CD45 expression by siRNAs. Viral vectors expressing a CD45- or CD53-specific siRNA contained a U6 promoter with the AAAA as the last bases and the eight-base palindrome loop (5′-GTCGCGAC-3′). UT7 cells were infected with the indicated viruses, and infected GFP-positive cells were sorted by FACS and propagated. U6-AAAA-CD53-8-infected (A) or U6-AAAA-CD45-8-infected (B) cells and uninfected cells (solid line in each panel) were stained with phycoerythrin (PE)-labeled CD53 (Left) or CD45 (Right) or a PE-labeled rat IgG_2a control antibody (dotted line in all panels).

Fig. 5.

Functional expression of random siRNA clones created from the mouse embryo library. (A) Expression of siRNA clones in NIH 3T3 cells. The indicated clones from the mouse embryo siRNA library were used to infect NIH 3T3 cells. GFP-positive infected cells were purified and cultured. Northern blot analysis was performed on total RNA from purified cells by using probes identical to the 5′ strands of the respective siRNAs (see Table 1 for sequence information). The U6 probe was used as a control. (B) Knockdown of endogenous genes by clones 548 and 550. Real-time PCR was performed on RNA from populations of infected NIH 3T3 cells infected with clones 548 or 550 (see Table 1 for sequence information). The absolute amount of the transcript targeted by the siRNA as well as the GAPDH control were deduced from the standard curve for amplification of the respective genes. The relative abundance of target of siRNA clone 548 (Left) or clone 550 (Right) was normalized to that of GAPDH.