Comparison of siRNA-induced off-target RNA and protein effects

Abstract

The downregulation of many mRNAs has been observed through bioinformatic analysis of microarray results following transfection of short interfering RNAs (siRNAs). Many of these mRNA changes are due to the interaction of the siRNA guide strand with partially complementary sites and thus are considered "off-target" effects. To examine the mRNA:siRNA interactions important for off-target effects, we generated a panel of mRNA:siRNA combinations containing single and double mismatches, bulges, and noncanonical base-pairing interactions in the 9th, 10th, and 11th positions of two siRNA binding sites located in the 3' UTR of an integrated reporter gene. Approximately half of the mRNA:siRNA combinations containing mismatches in positions 9-11 result in a twofold or more mRNA decrease with varying degrees of protein knockdown. However, mRNA and protein analysis of the various mRNA:siRNA combinations reveals instances in which mRNA and protein levels do not correlate. Analysis of the resulting degradation products recovered from an imperfectly complementary siRNA interaction with an endogenous gene reveals a small fraction of products that map to the canonical siRNA cleavage site. Furthermore, downregulation of ARGONAUTE 2 (AGO2), the only AGO family protein known to catalyze canonical siRNA-mediated cleavage, did not significantly affect the degree of mRNA knockdown observed for one of the stably expressed reporters after transfection of an imperfectly complementary siRNA. These results indicate that although some degree of canonical siRNA cleavage can take place between a siRNA and an off-target transcript, most off-target mRNA reductions are likely attributable to AGO2-independent degradation processes.

FIGURE 1.

Effects of mismatches in positions 9–11 on RNA and protein levels. (A) Average normalized RNA and protein reductions, expressed as fold change, of the 30 mRNA:siRNA combinations (Table 1). Five CXCR4 siRNAs (Table 1), each containing a single nucleotide change within positions 9–11, were transfected into six different stable cell lines expressing Renilla luciferase reporters, which also contained a single nucleotide change within positions 9–11 in both CXCR4 binding sites. Cells were collected for RNA and protein analysis 48 h after transfection. RNA analysis was performed by quantitative real-time PCR. TBP was used as an internal control. Protein analysis was performed by Renilla luciferase assays. For both RNA and protein analysis, samples were normalized to an unrelated siRNA (targeting GFP) transfection. For each mRNA:siRNA combination the average normalized RNA and protein reductions (fold change) were calculated from at least three independent experiments. Each bar graph represents the average normalized RNA reduction (black) overlapped with the average normalized protein reduction (gray), both expressed as fold change, for a particular mRNA:siRNA combination. Each combination is numbered and represented in Table 1. (B) Ratio of normalized protein knockdown to normalized RNA knockdown (fold change) for each of the 30 mRNA:siRNA combinations. The normalized protein reduction (expressed as fold change) of each mRNA:siRNA combination was divided by its corresponding normalized RNA reduction (fold change) to calculate the ratio between these two measurements. The protein to RNA ratio was then plotted versus normalized RNA knockdown (fold change) for each combination. The combinations were grouped into four distinct subgroups (labeled I–IV) according to their distribution in this plot.

FIGURE 2.

Imperfectly complementary siRNAs result in degradation products that correspond to canonical siRNA-mediated cleavage events. (A) Schematic of the interaction between two CXCR4 siRNAs that contain nucleotide changes in position 10 (CXCR4-A) and position 11 (CXCR4-B) and the endogenous CXCR4 transcript. The perfectly complementary CXCR4 siRNA (antisense strand) is also shown binding to the endogenous CXCR4 mRNA for reference. (B) 5′ RACE analysis performed on the endogenous CXCR4 mRNA after transfection with imperfectly complementary siRNAs. HeLa cells were transfected with 100 nM of Xrn1 siRNAs and 10 nM of either control siRNA (GFP), perfectly complementary CXCR4 siRNA, or imperfectly complementary siRNA (CXCR4-A or CXCR4-B). Cells were collected 48 h after siRNA transfection and subjected to RNA ligation followed by reverse transcription, nested PCR, and TOPO cloning. Each number corresponds to a distinct degradation product. The canonical siRNA cleavage product is highlighted with an asterisk. The CXCR4 siRNAs target sequence and the CXCR4 inner reverse PCR primer are indicated. (C) Summary of the 5′ RACE analysis performed on the endogenous CXCR4 mRNA after transfection with perfectly and imperfectly complementary siRNAs. The total number of clones sequenced and the number of clones that mapped to each degradation site are indicated for all the transfected siRNAs. Degradation products that mapped to the siRNA binding site have been highlighted.

FIGURE 3.

AGO2 is not required for off-target mRNA downregulation. (A) RNA analysis of a stably expressed Renilla luciferase reporter after transfection of 10 nM of a perfectly complementary siRNA, CXCR4-D siRNA (combination 27—Perf), or 10 nM of an imperfectly complementary siRNA, CXCR4-B (combination 15, mismatches at position 10 and 11—Imperf). These cells were transfected with either 100 nM of an unrelated siRNA (targeting GFP) or 100 nM of Ago2 siRNAs (Ago2-1 and Ago2-2) 16 h prior to the CXCR4 siRNAs transfections. (B) RNA analysis of Ago2 levels for experiment in A. RNA of analysis for A and B was performed by quantitative real-time PCR. TBP was used as an internal control. Samples were normalized to an unrelated siRNA (targeting GFP) transfection. Normalized average mRNA knockdown (fold change ± SD) was calculated from two independent experiments.