LncRNA VEAL2 regulates PRKCB2 to modulate endothelial permeability in diabetic retinopathy

Abstract Long non‐coding RNAs (lncRNAs) are emerging as key regulators of endothelial cell function. Here, we investigated the role of a novel vascular endothelial‐associated lncRNA (VEAL2) in regulating endothelial permeability. Precise editing of veal2 loci in zebrafish (veal2 gib005Δ8/+) induced cranial hemorrhage. In vitro and in vivo studies revealed that veal2 competes with diacylglycerol for interaction with protein kinase C beta‐b (Prkcbb) and regulates its kinase activity. Using PRKCB2 as bait, we identified functional ortholog of veal2 in humans from HUVECs and named it as VEAL2. Overexpression and knockdown of VEAL2 affected tubulogenesis and permeability in HUVECs. VEAL2 was differentially expressed in choroid tissue in eye and blood from patients with diabetic retinopathy, a disease where PRKCB2 is known to be hyperactivated. Further, VEAL2 could rescue the effects of PRKCB2‐mediated turnover of endothelial junctional proteins thus reducing hyperpermeability in hyperglycemic HUVEC model of diabetic retinopathy. Based on evidence from zebrafish and hyperglycemic HUVEC models and diabetic retinopathy patients, we report a hitherto unknown VEAL2 lncRNA‐mediated regulation of PRKCB2, for modulating junctional dynamics and maintenance of endothelial permeability.

non-coding transcript dataset. Finally, the isoforms of all transcripts eliminated at each step of the pipeline were also excluded, to arrive at a final set of 4,897 putative lncRNA. A systematic classification of the predicted lncRNA was achieved by overlapping the putative transcripts with well-annotated RefSeq genes, identifying 925 promoter-associated lncRNA and 3,972 long intergenic RNA (lincRNA). The nearest protein-coding gene for each lncRNA has been listed (Dataset EV2).
In an attempt to provide an independent assessment of the protein-coding capacity of our lncRNA catalog, the translation efficiency scores were calculated on the basis of ribosomal occupancy on the predicted transcripts. For measuring the translational efficiency, the data from the previously reported ribosome profiling study performed across 8 stages of zebrafish development was retrieved. Translation Efficiency Scores (TES) were assigned to 26,829 zebrafish RefSeq genes (coding as well as non coding) along with the 3,866 novel predicted lncRNAs based on their estimated ribosomal occupancy. Our analysis showed that 2247 lncRNAs had a TES equal to zero. Large majority (91%) of the novel lncRNAs showed nil or negligible evidence of translation (TES < 0.001) compared to 82% of known noncoding RNAs and 1.4% of known protein-coding transcripts. The median TES for the novel lncRNAs (0) was found to be significantly lower than that of the protein-coding RefSeq genes (median TES:0.04), and comparable, if not lesser than that of the non coding RefSeq genes (median TES:0.001) ( Figure 1B).

QC of endothelial versus non-endothelial datasets
The differential expression of some of the well studied vascular markers in EC and NEC samples were checked, including 6 upregulated genes (fli1a, flt4, tbx6, kdrl, dlx2a and kdr/flk1) and 2 downregulated ones (ptprn2, dlb) all of which showed significant FPKM fold-change (Appendix table   S6). Apart from these, a few more endothelial modulators like nephrosin (npsn) and hematopoietically expressed homeobox (hhex) were identified, which showed greater than 30fold, and endothelium-specific receptor tyrosine kinase 1 (tie-1) and vascular associated protein (vap) showed more than 4-fold elevated expression in EC. MicroRNA mir130c-1 which is known to be associated with heart development was also found to be EC-enriched (Appendix table S 10).

RACE of veal2 and VEAL2
Total RNA from 24 hpf zebrafish was isolated using TRIzol (Invitrogen, USA) and 2ug of it was used for cDNA preparation of 5'& 3' RACE using manufacturer's protocol. veal2-GSP-1R was used to generate 5'RACE cDNA and adapter primer (dT) were used to generate 3'RACE cDNA.
After tailing of 5'RACE cDNA using Tdt enzyme and dCTP, a gradient PCR with annealing temperature ranging from 60° C to 68° C was setup for 5' RACE using 5'RACE cDNA and veal2-GSP-2R. A very faint band was observed at 62° C which was gel extracted and purified. Two back to back nested PCR using veal2-GSP-3 & 4R were run at 62° C annealing temperature. The nested PCR of 5'RACE showed 4 different bands of size of approx 300bp, 350bp, 470bp and 580bp (Appendix figure S4C). All the bands were gel extracted individually and cloned in pCR 2.1-TOPO vector (Invitrogen, USA) and sanger sequencing was performed on it. The sequence of all the products were aligned with RNAseq annotated veal2. Only band at 580bp showed overlap with previously annotated veal2 from RNAseq at 5'end with no extra nucleotide extension (Appendix figure S4D). The rest of the products were observed to be nonspecific as they did not align to the zebrafish genome.
Similarly, 3'RACE PCR was performed for veal2 and VEAL2 using veal2-GSP-1F and VEAL2-GSP-1F primers on respective cDNA prepared by oligo dT at 56°C annealing temperature. The 3' RACE PCR was followed by nested PCR using veal2-GSP-2F and VEAL2-GSP-2F. We To confirm that the altered vascular development and integrity defects are pertinent to the downregulation of veal2 RNA, we attempted to rescue the phenotype by complementing with mature wild-type veal2 RNA. The delivery of a cocktail of in-vitro synthesized mature wild-type veal2 and anti-sense MO into single-cell zebrafish embryos resulted in 25% of the animals displaying hemorrhage, compared to 50% phenotype in animals injected with morpholino alone.
Thus, about 50% rescue of the vascular integrity defects was achieved by the introduction of spliced veal2 RNA exogenously (p-value<0.5E-3) (Appendix figure S6O).

Ectopic delivery of veal2 induces excessive sprouting of sub-intestinal vessels
To assess the angiogenic activity of the lncRNA, we delivered the in vitro transcribed full-length After the injection of 5' veal2 TALEN, surviving phenotypic animals displaying cranial hemorrhage indicating blood vessel integrity defects (Appendix figure S10) were raised to adulthood and the genotypes were screened. Out of 8 F0 animals, 4 (3 males and one female) displayed editing events at the veal2 locus (Appendix figure S11). Particularly, one pair of the founders were incrossed which gave rise to ~50 percent progeny with cranial hemorrhage (Appendix figure S12A-D) and the genotype was confirmed through capillary sequencing (Appendix figure S12E). The phenotypic progeny were raised to adulthood and genotyped after fin clipping. Of the 26 animals, 20 displayed indels at the targeted loci (Appendix figure S13). Ten animals showed the same 8 bp deletion (veal2Δ8). Possible off-target editing events were checked in 6 of these animals as described in the methods section (Appendix figure S14).