Altered A-to-I RNA editing in human embryogenesis

Abstract

Post-transcriptional events play an important role in human development. The question arises as to whether Adenosine to Inosine RNA editing, catalyzed by the ADAR (Adenosine Deaminase acting on RNA) enzymes, differs in human embryogenesis and in adulthood. We tested the editing of various target genes in coding (FLNA, BLCAP, CYFIP2) and non-coding sequences at their Alu elements (BRCA1, CARD11, RBBP9, MDM4, FNACC), as well as the transcriptional levels of the ADAR1 enzymes. This analysis was performed on five fetal and adult human tissues: brain, heart, liver, kidney, and spleen, as well as on human embryonic stem cells (hESCs), which represent the blastocyst stage in early human development. Our results show substantially greater editing activity for most adult tissue samples relative to fetal ones, in six of the eight genes tested. To test the effect of reduced A-to-I RNA editing activity in early human development we used human embryonic stem cells (hESCs) as a model and tried to generate hESC clones that overexpress the ADAR1-p110 isoform. We were unable to achieve overexpression of ADAR1-p110 by either transfection or lentiviral infection, though we easily generated hESC clones that expressed the GFP transgene and overexpressed ADAR1-p110 in 293T cells and in primary human foreskin fibroblast (HFF) cells. Moreover, in contrast to the expected overexpression of ADAR1-p110 protein following its introduction into hESCs, the expression levels of this protein decreased dramatically 24-48 hr post infection. Similar results were obtained when we tried to overexpress ADAR1-p110 in pluripotent embryonal carcinoma cells. This suggests that ADAR1 protein is substantially regulated in undifferentiated pluripotent hESCs. Overall, our data suggest that A-to-I RNA editing plays a critical role during early human development.

Figure 1. RNA editing activity of Alu elements is reduced in human fetal tissue samples relative to adult tissue samples. RNA

editing levels were determined for samples of fetal tissues, adult tissues, and hESC lines H9.2 and I6 at passages 50–60. Mean levels of two independent measurements are presented, as assessed using the Sequenom Mass ARRAY compact analyzer. Editing level was significantly reduced for most fetal tissues in four genes tested: BRACA1 (A), CARD11 (B), RBBP9 (C), and MDM4 (D) (Mann-Whitney, P<0.05). FANCC (E) exhibited no difference between adult and fetal editing activity.

Figure 2. RNA editing activity of coding regions is reduced in human fetal tissue samples relative to adult tissue samples.

RNA editing levels were determined for samples of fetal tissues, adult tissues, and hESC lines H9.2 and I6 at passages 50–60. Mean levels of two independent measurements are presented, as assessed using the Sequenom Mass ARRAY compact analyzer. Fetal editing of FLNA (A), BLCAP spleen (B), and CYFIP2 (C) was found to be significantly reduced (Mann-Whitney, P<0.05). No difference was found between fetal BLCAP (B) editing level of brain, liver, heart, and kidney compared to respective adult tissues.

Figure 3. ADAR1 isoforms and ADAR2 mRNA expression levels in human fetal and adult tissue samples.

1 µg RNA, from samples obtained from fetal and adult tissues was reverse transcribed to generate cDNA. The mRNA expression level was determined using Quantitative Real Time PCR (QRT-PCR) analysis. The Y axis represents the relative ADAR expression levels of (A) ADAR1-p110 isoform, (B) ADAR1-p150 isoform, (C) ADAR2, and (D) ADAR3. The data plotted are representative of three independent experiments. It appears that ADAR1 transcript levels are mostly reduced, while ADAR2 transcript levels are mostly increased in human fetal tissues relative to respective adult tissues.

Figure 4. Transfection of ADAR1-p110 gene into hESC line H9.2.

Attempts to generate stable hESC lines H9.2 overexpressing ADAR1-p110 failed: The pcDNA3.1/HisC plasmid, expressing ADAR1-p110 conjugated to His-tag and Xpress epitope at the amino end of the protein (which adds approximately 6 kDa to the protein) under the CMV promoter, was transfected into H9.2 hESCs. A similar plasmid was transfected into 293T cells for validation of plasmid functionality. H9.2 transfected cells were cultured on Neomycine (Neo) resistant Mouse Embryonic Fibroblast (MEF) feeder cells, and selection was performed with 50 µg/ml Neo for 10 days. Neo-resistant hESC clones were isolated and expanded. RNA and protein extracts were generated from 40 selected hESC clones and subjected to (A) QRT-PCR analysis and (B) Western Blot analysis. Representative data for 6 Neo-resistant hESC clones (C1–C6) are shown here relative to untransfected hESC H9.2 samples. For the Western Blot a positive control of 293T transfected cells, overexpressing ADAR1-p110 was added (left lane). Arrow head indicates the exogenous His-tag conjugated ADAR1-p110. No hESC Neo-resistant clone showed significant overexpression of ADAR1-p110 isoform following transfection and Neo-selection. 293T cells were harvested 4 days post transfection. RNA and protein were extracted from untransfected (UT) and transfected cells. ADAR1-p110 expression was analyzed by (C) QRT-PCR and (D) Western Blot analysis, lane 1– control untransfected cells, lane 2– transfected cells. Results show that ADAR1-p110 overexpression could be induced in 293T cells but not in selected hESC clones.

Figure 5. Lentivirus infection of ADAR1-p110 into the hESC line H9.2. ADAR1 overexpression could not be generated in H9.2 although the virus did integrate into the cells.

H9.2 hESCs were infected with the lentivirus PTK vector harboring ADAR1-p110 gene under the CMV promoter. As a control, a similar plasmid harboring GFP under the CMV promoter or PTK with no insert (empty vector; emp) were used to infect H9.2 hESCs. Simultaneously, Human Foreskin Fibroblasts (HFF) and 293T cells were also infected with similar lentiviral vectors. (A) GFP-expressing cells demonstrated infection efficiency of ∼90% for 293T and HFF, and ∼50% for H9.2 hESCs. (B) mRNA expression levels of ADAR1-p110 transcript, following infection of H9.2, HFF, and 293T cells with lentivirus. Analysis was performed by QRT-PCR using specific primers, at time periods of two days post infection (2dpi) and seven days post infection (7dpi), relative to non-infected cells. (C) Western Blot analysis of ADAR1-p110 isoform protein level following infection of H9.2, HFF, and 293T cells. Infection was performed with the lentiviral vectors PTK-ADAR1-p110 and PTK-GFP. Protein was collected at 2dpi and 7dpi. Total protein levels were verified by β-Actin. Lanes: 1-uninfected cells, 2- PTK-ADAR1-p110 2dpi, 3- PTK-ADAR1-p110 7dpi, 4- PTK-GFP 7dpi, 5- PTK-emp 2dpi, 6- PTK-emp7dpi. (D) Analysis of lentivirus integration into H9.2, HFF, and 293T infected cell genome by PCR for genomic DNA. Primers were designed so that the left primer corresponds to the CMV promotor and the right to the ADAR1-p110 sequence. PCR for a genomic segment of OCT4 and the RRE element of the viral construct was performed as control. Lanes: 1–293T uninfected cells, 2- 293T, PTK-ADAR1-p110 7dpi, 3- H9.2 uninfected, 4- H9.2, PTK-ADAR1-p110 13dpi 5- H9.2, PTK-empty 7dpi, 6- HFF PTK-GFP 7 dpi, 7-HFF PTK-ADAR1-P110 7dpi 8- No template DNA - negative control.

Figure 6. Lentivirus infection of ADAR1-p110 into the EC lines NTERA2 and 2102Ep. ADAR1-p110 overexpression could not be generated.

THE EC lines NTERA2 and 2102Ep were infected with the lentivirus PTK vector harboring ADAR1-p110 gene under the CMV promoter. As a control, a similar plasmid harboring GFP under the CMV promoter was used. (A) mRNA expression levels of ADAR1-p110 transcript following infection. Analysis was done by QRT-PCR using specific primers, at 2dpi and 7dpi, relative to non-infected cells and normalized to GAPDH. (B) Western blot protein analysis. Lanes: 1-4- NETRA2, 5-8- 2102Ep; 1,5- uninfected, 2,6- ADAR1-p110 2dpi, 3,7- ADAR1-p110 7dpi, 4,8- 7dpi GFP infected. (C) Lentivirus integration into H9.2, HFF, and 293T infected cell genome verified by PCR for genomic DNA. Primers were designed so that the left primer corresponds to the CMV promotor and the right to the ADAR1-p110 sequence. PCR for a genomic segment of OCT4 and the RRE element of the viral construct was performed as control. Lanes: 1-4 NETRA2, 5-8 2102Ep; 1,5- uninfected 2,6- 15dpi, 3,7- 7dpi, 4,8- GFP infected 9- no template control.