Upregulating endogenous genes by an RNA-programmable artificial transactivator

Abstract

To promote expression of endogenous genes ad libitum, we developed a novel, programmable transcription factor prototype. Kept together via an MS2 coat protein/RNA interface, it includes a fixed, polypeptidic transactivating domain and a variable RNA domain that recognizes the desired gene. Thanks to this device, we specifically upregulated five genes, in cell lines and primary cultures of murine pallial precursors. Gene upregulation was small, however sufficient to robustly inhibit neuronal differentiation. The transactivator interacted with target gene chromatin via its RNA cofactor. Its activity was restricted to cells in which the target gene is normally transcribed. Our device might be useful for specific applications. However for this purpose, it will require an improvement of its transactivation power as well as a better characterization of its target specificity and mechanism of action.

Figure 1.

Design of the ribo-transactivator and its proof-of-principle validation. (A) Structure of the NMHV apo-activator and its RNA cofactor. NMHV includes: NLS2, nuclear localization signal 2x; RBD, MS2 RNA-binding domain; HA, hemagglutinin epitope and TAD, VP16-related transactivator domain, 3x. The RNA cofactor includes: MF6, MS2-high affinity, stem-and-loop finger, 6x; and ‘bait’, short, target gene specific, RNA tag. GOI is the GOI. (B) Subcellular distribution of NMHV in HEK293T cells, in the presence or in the absence of its MF6-NAP22 RNA cofactor, as revealed by anti-HA immunofluorescence. Beta-actin and H3K9me3 distributions, cytoplasmic and nuclear, respectively, are also shown, as references. (C) Schematics of human FMR1, NAP22 and NRGN loci, with natural transcripts, artificial baits and diagnostic amplicons used in this study. Nucleotide numbering refers to UCSC-hg19. Color code: blu, DNA; green, sense-oriented RNA; red, antisense-oriented RNA. (D–F) Upregulation of FMR1, NAP22 and NRGN mRNAs in HEK293T cells cotransfected with NMHV- and MF6-bait-encoding plasmids, as evaluated 72 h post-transfection. Pgkp1-EGFP (‘EGFP’) and MF6-øplasmids were used as controls. Results were normalized against GAPDH and further normalized against the EGFP/MF6-øcombination. (G, H) Upregulation of NAP22 protein by NMHV/MF6-NAP22. (G) Western blotting of NAP22 in HEK293T cells, four days after transfection by NMHV/MF6-NAP22 or EGFP/MF6-ø, via the ‘Lipofectamine 3000 protocol’. (H) Quantification of NAP22 protein detectable in (G). Results were normalized against βACT and further normalized against the EGFP/MF6-øcombination. Numbers of biological replicates, n, are displayed under the graphs. Bars represent s.e.m.'s.

Figure 1.

Design of the ribo-transactivator and its proof-of-principle validation. (A) Structure of the NMHV apo-activator and its RNA cofactor. NMHV includes: NLS2, nuclear localization signal 2x; RBD, MS2 RNA-binding domain; HA, hemagglutinin epitope and TAD, VP16-related transactivator domain, 3x. The RNA cofactor includes: MF6, MS2-high affinity, stem-and-loop finger, 6x; and ‘bait’, short, target gene specific, RNA tag. GOI is the GOI. (B) Subcellular distribution of NMHV in HEK293T cells, in the presence or in the absence of its MF6-NAP22 RNA cofactor, as revealed by anti-HA immunofluorescence. Beta-actin and H3K9me3 distributions, cytoplasmic and nuclear, respectively, are also shown, as references. (C) Schematics of human FMR1, NAP22 and NRGN loci, with natural transcripts, artificial baits and diagnostic amplicons used in this study. Nucleotide numbering refers to UCSC-hg19. Color code: blu, DNA; green, sense-oriented RNA; red, antisense-oriented RNA. (D–F) Upregulation of FMR1, NAP22 and NRGN mRNAs in HEK293T cells cotransfected with NMHV- and MF6-bait-encoding plasmids, as evaluated 72 h post-transfection. Pgkp1-EGFP (‘EGFP’) and MF6-øplasmids were used as controls. Results were normalized against GAPDH and further normalized against the EGFP/MF6-øcombination. (G, H) Upregulation of NAP22 protein by NMHV/MF6-NAP22. (G) Western blotting of NAP22 in HEK293T cells, four days after transfection by NMHV/MF6-NAP22 or EGFP/MF6-ø, via the ‘Lipofectamine 3000 protocol’. (H) Quantification of NAP22 protein detectable in (G). Results were normalized against βACT and further normalized against the EGFP/MF6-øcombination. Numbers of biological replicates, n, are displayed under the graphs. Bars represent s.e.m.'s.

Figure 2.

Mechanisms underlying NAP22 transactivation in HEK293T cells. (A) Schematics of the NAP22 locus with NAP22-bait and diagnostic amplicons employed for the analysis. (B) Modulation of NAP22-pre-mRNA in NMHV/MF6-NAP22-expressing cells, evaluated by intronic qRTPCR. Results were normalized against GAPDH and further normalized against the EGFP/MF6-øcombination. (C) Recruitment of NMHV at NAP22 promoter in NMHV/MF6-NAP22-expressing cells, evaluated by anti-HA-ChIP/qPCR. The potential NAP22-offtarget SLC4A2 was used as a specificity control. (D) RNApolII binding at different sites of the NAP22 locus in NMHV/MF6-NAP22-expressing cells, evaluated by anti-RNApolII-ChIP/qPCR. In (C, D) results were normalized against input chromatin and further normalized against the NMHV/MF6-øcontrol. Noticeably, in (B, D) cells were transfected by the ‘Lipofectamine 3000 protocol’. In (B-D) cell culture timing was as described in Figure 1. Numbers of biological replicates, n, are displayed under the graphs. Bars represent s.e.m.'s.

Figure 3.

NMHV-mediated transactivation of Emx2 in murine, embryonic cortico-cerebral precursors and its molecular and biological correlates. (A) Schematics of the murine Emx2 locus, with natural transcripts, artificial baits and diagnostic amplicons used in this study. Nucleotide numbering refers to UCSC-mm10. Color code: blu, DNA; green, sense-oriented RNA; red, antisense-oriented RNA. (B)Emx2-mRNA upregulation in precursors infected by NMHV- and MF1-Emx2-encoding lentivectors. EGFP and MF1-øviruses were used as controls. Results were normalized against Gapdh and further normalized against the EGFP/MF1-øcombination. (C) Reduced frequency of cells immunopositive for the neuron-specific Tubb3 marker, in cultures of NMHV/MF1-Emx2-overexpressing precursors. Results were normalized against EGFP/MF1-øcontrols. (D) Examples of aTubb3 immunofluorescences referred to in (C). (E) Enhancing Emx2 transactivation by improving the RNA cofactor structure. The assay was run similar to Figure 2A, replacing the monomeric-finger ‘MF1-bait’ cofactor by its dimeric-finger ‘MF2-bait’ derivative. (F) Enhancing Emx2 transactivation by increasing the m.o.i. ratio of NMHV- and MF2-Emx2-encoding lentiviruses. The best results were obtained by an NMHV/MF2-Emx2 ratio of 3:1. (G)Emx2 promoter enrichment in chromatin of NMHV/MF1-Emx2-overexpressing precursors, immunoprecipitated by an anti-HA antibody. Results were normalized against input chromatin and further normalized against the NMHV/MF1-øcontrol. Promoters of potential Emx2-offtargets Ptpn12, CachD1 and Dvl3 were used as specificity controls. In all cases (B-G), cortico-cerebral cells were dissociated from E12.5 embryos, acutely infected, cultured for 96 h and finally analyzed. Throughout the figure, lentiviral multiplicities of infection (moi's) and numbers of biological replicates, n, are displayed under the graphs. Bars represent s.e.m.'s.

Figure 4.

NMHV-mediated transactivation of Foxg1 in murine, embryonic cortico-cerebral precursors and its biological correlate. (A) Schematics of the murine Foxg1 locus, with natural transcripts, artificial baits and diagnostic amplicons used in this study. Nucleotide numbering refers to UCSC-mm10. Color code: blu, DNA; green, sense-oriented RNA. (B)Foxg1-mRNA upregulation in precursors infected by NMHV- and MF1-Foxg1-encoding lentivectors. EGFP and MF1-øviruses were used as controls. Results were normalized against Gapdh and further normalized against the EGFP/MF1-øcombination. (C) Reduced frequency of cells immunopositive for the neuron-specific Tubb3 marker, in cultures of NMHV/MF1-Foxg1-overexpressing precursors. (D) Examples of aTubb3 immuno-fluorescences referred to in (C). In all cases (B–D) time-frame of the experiments as well as representation of moi's and statistical parameters are as in Figure 3.

Figure 5.

Lack of responsivity to Emx2- and FoxG1-specific ribo-transactivators by neural precursors of non-cortico-cerebral origin. (A) Unchanged Emx2-mRNA levels in E10.5 rhombo-spinal precursors, infected by NMHV- and MF2-Emx2-encoding lentiviruses or EGFP- and MF2-øcontrols. (B) Unchanged FoxG1-mRNA levels in E10.5 mesencephalic precursors, infected by NMHV- and MF1-FoxG1-encoding lentiviruses or EGFP and MF1-øcontrols. In both (A) and (B), transactivation of the two genes in E12.5 cortico-cerebral precursors is shown, as a positive control. Moreover, in both cases, results were normalized against Gapdh and further normalized against E12.5 cortico-cerebral precursors treated by EGFP/MF2-øand EGFP/MF1-øcombinations, respectively. Time frame of the experiments as well as representation of moi's and statistical parameters is as in Figure 3. (C) Idealized representation of the murine E10.5 neural tube. cx, cerebral cortex; m, mesencephalon; rs, rhombo-spinal tract.

Figure 6.

Consequences of RNA-bait shortening and mutagenesis. (A-C) Comparison of NMHV-dependent gene activation driven by select ‘primary’ baits, NAP22 (A), FMR1.1 (B) and Emx2-S (C), and ‘secondary’ baits obtained by heminested shortening of the former ones. (D, E) Full or partial suppression of NMHV-dependent NAP22 transactivation driven by the NAP22.60L bait, upon replacement of 30% (D) or 15% (E) of its original bases by mutant ones. Mutant bases were distributed in up to four (D) and up to 10 (E) equispaced mismatching modules, as shown in top panels (sequences of mutant baits are reported in Supplementary Table S7). Results were normalized against GAPDH (A, B, D, E) and Gapdh (C) and further normalized against EGFP/MF6-ø(A, B, D, E) and EGFP/MF2-ø(C) control samples. Time frame of the experiments as well as representation of moi's and statistical parameters was as in Figures 1 (A, B, D, E) and 3 (C).