Engineered microRNA scaffolds for potent gene silencing in vivo

Abstract

RNA interference (RNAi) is emerging as a powerful strategy for therapeutic targeting of "undruggable" targets. However, efficacy of currently used siRNA-based therapies is often hindered by transient effects and limited modeling possibilities. Artificial microRNAs (amiRNAs or miRNA scaffolds) present a durable and precise approach to gene silencing, opening new avenues for developing long lasting targeted therapies. In this study, we engineered highly expressed primary miRNAs (pri-miRNAs) with sequence determinants known to enhance processing efficacy and precision. The resulting amiRNAs were extensively tested both in vitro and in vivo and proved to efficiently silence a target gene when virally delivered via adeno-associated virus (AAV) into mice brains. This study provides a set of novel amiRNAs with potential therapeutic application as well as a pipeline to generate and validate novel amiRNAs from endogenous pri-miRNAs.

Fig. 1

Design of novel amiRNAs and screening with reporter assay. (a) Secondary structure of the endogenous pri-Let7a1 (bottom) compared to the secondary structure of the engineered amiRNA derived from pre-Let7a1 with all the modifications indicated (top); sequence determinants are shown in red; darker blue indicates greater stability of the local structure. (b) A schematic of the reporter assay for the silencing efficiency evaluation of the novel amiRNAs; a reporter cell line stably expressing an mCherry cassette responsive to the screening guide (RNA responsive elements) was infected with a single copy of lentiviral particles encoding the amiRNAs under the control of the weak promoter EF-1α short. Single infection rate was confirmed by cytometric analysis three days post transduction (GFP + cells < 15% of total population). (c) Cytometry analysis six days post-transduction showing mCherry signal reduction was used as an indicator for silencing efficiency of the novel amiRNAs; an empty vector encodes for GFP only was used as a negative control. The mCherry signal from infected cells (GFP+) was normalized over the mCherry signal from non-infected cells (GFP−) within the same sample. Brown bars represent the amiRNAs selected for the rest of the studies described in (d). (d) Silencing efficiency of selected amiRNAs loaded with a weak guide sequence. (e) Reporter assay for novel amiRNAs loaded with guide sequences previously reported to have different efficiency ranging from weak, moderate to strong silencing; timepoints and readout used were the same as described for (b,c). Statistical analysis carried with one-way ANOVA; P values: *<0.05 - ** <0.01 - ***<0.001 - ****<0.0001.

Fig. 2

Validation of the novel amiRNAs for the silencing of endogenous genes and delivery with rAAV9 vectors. (a) Schematic of single-copy lentiviral infection for silencing Cd9 in 3T3 cells. The single infection rate was confirmed by cytometric analysis three days post-transduction, with GFP + cells comprising less than 15% of the total population. (b) Six days post-transduction, a fraction of the cells was stained for CD9 and analyzed by cytometry to quantify the knockdown efficiency. (c) The remaining fraction was selected with puromycin to obtain a homogeneous pool of cells expressing the amiRNAs from a single locus, and RNAs were isolated to quantify the residual amount of GFP as a proxy for processing efficiency. (d) Schematic of rAAV9 infection in the reporter line. (e) Five days post-transduction, cytometry was employed to assess mCherry silencing in GFP-high (high viral cargo expression) and GFP-low (low viral cargo expression) populations, with mCherry expression normalized to that in GFP- (non-infected) cells within the same sample. “Negative control” indicates Let7f2 WT, a non-functional amiRNA from the screening shown in Fig. 1b,c. Statistical analysis was performed using one-way ANOVA, with P-values indicated as *<0.05, **<0.01, ***<0.001, and ****<0.0001.

Fig. 3

AmiRNAs processing studies in hIPSC-derived neurons. (a) Schematic of transduction of hIPSC-derived NGN2 neurons with rAAV9 encoding the novel amiRNAs loaded with a guide against PTEN. (b) qPCR was used to quantify the residual expression of PTEN; “no amiRNA” refers to the virus encoding GFP only. “Negative control” indicates Let7f2 WT, a non-functional amiRNA from the screening in Fig. 1b,c; “Mock” refers to non-transduced cells. (c,d) Analysis of mature amiRNA processing precision: (c) The percentage of reads starting and ending with the corresponding nucleotide. miR26bAll is used as an example; the guide strand sequence is in uppercase, and the adjacent amiRNA nucleotides are in lowercase. (d) The percentage of reads starting with the correct 5′ nucleotide for each amiRNA; dots represent replicates. (e) Normalized quantification of guide and passenger strands. (f) Correlation [Pearson r value: 0.6141, P-value (two-tailed): 0.0195] between guide strand amount (determined from small RNA sequencing) and silencing efficiency (determined from qPCR; “1” = 100% silencing). Statistical analysis was carried out using one-way ANOVA and Pearson’s Correlation test. P-values: *<0.05, **<0.01, ***<0.001, ****<0.0001.

Fig. 4

Delivery of novel amiRNAs in vivo via Intracerebroventricular injection of rAAV9. (a) Schematic of in vivo intracerebroventricular injection of rAAV9 encoding the novel amiRNAs loaded with a guide targeting Ataxin2. (b) qPCR analysis to measure Ataxin2 expression; “Vehicle” refers to PBS treatment. (c) Representative immunofluorescence images showing ATAXIN2 (red), GFP (green), and nuclei (blue). (d) Quantification of ATAXIN2 expression in GFP-positive (transduced) versus GFP-negative (non-transduced) cells (n = 4). (e) Normalized quantification of guide (G) and passenger (P) strands. (f) Analysis of mature amiRNA processing precision: the percentage of reads starting and ending with the correct nucleotide; the guide strand sequence is in uppercase, and adjacent amiRNA nucleotides are in lowercase. Statistical analysis was performed using one-way ANOVA. P-values: *<0.05, **<0.01, ***<0.001, ****<0.0001.