Interdependent Transcription of a Natural Sense/Antisense Transcripts Pair (SLC34A1/PFN3)

Abstract

Natural antisense transcripts (NATs) constitute a significant group of regulatory, long noncoding RNAs. They are prominently expressed in testis but are also detectable in other organs. NATs are transcribed at low levels and co-expressed with related protein coding sense transcripts. Nowadays NATs are generally considered as regulatory, long noncoding RNAs without closer focus on the inevitable interference between sense and antisense expression. This work describes a cellular system where sense and antisense transcription of a specific locus (SLC34A1/PFN3) is induced using epigenetic modifiers and CRISPR-Cas9. The renal cell lines HEK293 and HKC-8 do not express SLC34A1/PFN3 under normal culture conditions. Five-day exposure to dexamethasone significantly stimulates sense transcript (SLC34A1) levels and antisense (PFN3) minimally; the effect is only seen in HEK293 cells. Enhanced expression is paralleled by reduced sense promoter methylation and an increase in activating histone marks. Expression is further modulated by cassettes that stimulate the expression of sense or antisense transcript but disrupt protein coding potential. Constitutive expression of a 5'-truncated SLC34A1 transcript increases sense expression independent of dexamethasone induction but also stimulates antisense expression. Concordant expression is confirmed with the antisense knock-in that also enhances sense expression. The antisense effect acts on transcription in cis since transient transfection with sense or antisense constructs fails to stimulate the expression of the opposite transcript. These results suggest that bi-directional transcription of the SLC34A1/PFN3 locus has a stimulatory influence on the expression of the opposite transcript involving epigenetic changes of the promoters. In perspective of extensive, previous research into bi-directionally transcribed SLC34A loci, the findings underpin a hypothesis where NATs display different biological roles in soma and germ cells. Accordingly, we propose that in somatic cells, NATs act like lncRNAs-with the benefit of close proximity to a potential target gene. In germ cells, however, recent evidence suggests different biological roles for NATs that require RNA complementarity and double-stranded RNA formation.

Keywords: CRISPR-Cas9; DNA methylation; concordant expression; histone modifications; natural antisense transcripts; transcriptional interference.

Figure 1

Architecture and expression of the SLC34A1/PFN3 (sense/antisense) locus. (A) Modified screenshot of the human SLC34A1/PFN3 locus in the UCSC genome browser. The SLC34A1 sense gene and the PFN3 gene in antisense are at the top, published human [7] and murine [23] antisense transcripts are below. Conservation in 100 vertebrates (light blue) and CpG islands (green). (B) Expression of SLC34A1 and PFN3 mRNA and protein in various organs adapted from the human protein atlas (https://www.proteinatlas.org/, accessed on 10 December 2021). (C) Expression of sense (SLC34A1, black bars) and antisense transcripts (PFN3, grey bars) was assessed by RT-qPCR in human testis RNA and RNA extracted from primary kidney cells. The log2 fold change refers to uninduced HKC-8 cells.

Figure 2

Expression of the SLC34A1/PFN3 (sense/antisense) locus in in response to zebularine. The expression of sense (SLC34A1) and antisense transcripts (PFN3) was assessed by RT-qPCR; human testis RNA and RNA extracted from primary kidney cells were used as controls (right panels of (A–C)). (A) Response of the sense transcript to zebularine in HKC-8 cells and (B) HEK293 cells. (C) The most effective dose was used (50 µM) to establish the time course of expression for both the sense (black) and the antisense (grey, only detectable at 24 h) transcripts. Statistical significance was tested by one-way ANOVA followed by Tukey’s test for multiple comparisons, * p < 0.05.

Figure 3

Expression of sense and antisense transcripts in response to dexamethasone in HEK293 and HKC-8 cells. (A) Response of the sense transcript to increasing levels of dexamethasone in HEK293 cells (black) and HKC-8 cells (grey) measured by RT-qPCR. Human testis RNA and human kidney cell RNA were used as positive controls (right panel). (B) Time dependence of the dexamethasone response of sense (black) and antisense (grey) transcripts in HEK293 cells. Control RNAs from testis and kidney on the right. (C) Nuclear and cytoplasmic distribution of sense and antisense transcripts in HEK293 cells. Nuclear and cytoplasmic fractions were enriched followed by RNA extraction and RT-qPCR from control and dexamethasone induced cells (underscored). XIST was measured to estimate nuclear enrichment, GAPDH for cytoplasmic enrichment. One-way ANOVA followed by Tukey’s test for multiple comparisons, * p < 0.05; *** p < 0.001.

Figure 4

DNA methylation and histone H3 modifications of SLC34A1/PFN3 sense and antisense promoters in HEK293 cells. (A) DNA methylation levels of 7 CpGs in the sense promoter (black) and 6 CpGs in the antisense promoter (grey) after exposure to dexamethasone for 1, 5 and 15 days. Fully methylated (100%) and un-methylated fragments (0%) served as controls, methylation levels in cells exposed to dexamethasone for 1, 5 and 15 days were compared to cells without the drug. (B) Global histone H3 acetylation in response to the exposure of dexamethasone for 1, 5 and 15 days. (C) Chromatin immunoprecipitation (ChIP) using antibodies against H3K27Ac and H3K4Me3 followed by qPCR of sense (left panel) and antisense (right panel) promoters. Cells were exposed to dexamethasone for 1, 3 and 5 days or left without the drug as a control. Significance was established using one-way ANOVA and Tukey’s test for multiple comparisons, * p < 0.05; ** p < 0.01; *** p < 0.001; ns = no significance.

Figure 5

Constitutive low-level activation of sense and antisense expression in HEK293 cells. (A) Snapshot of the SLC34A1/PFN3 locus with the insertion sites for the HDR (homology-directed repair) cassettes and primer sites S1–S3 as well as AS1 and AS2. (B) Schematic representation of the HDR cassette containing a CMV promoter, the puromycin resistance, the BGH polyadenylation signal and the two gene-specific flanking regions (homologous arm). The cassette was meant to shut down transcription, but insertion produced low levels of read-through transcripts driven by the CMV promoter. (C) Sense transcript expression in cells with the knock-in construct in sense orientation (middle) and antisense orientation (right). Wildtype HEK293 cells stimulated with dexamethasone (left) were used as positive controls; all expression levels were referred to unstimulated HEK293 wildtype cells. The bars represent specific primer pairs and are color coded, S1 red, S2 blue and S3 black. Of note, primer pair 1 is upstream of the cassette insertion site and does not amplify the read-through transcript. (D) Antisense transcript expression in CRISPR edited HEK293 cell clones, the left panel shows unedited HEK293 cells exposed to dexamethasone as a control. Monoallelic insertion of the cassette placed in sense (middle) and antisense (right) orientation. The primer pair AS1 (blueish) flanks the cassette and only generates a PCR product from clones without an insertion. Data are the mean from three independent biological and three technical replicates. Expression levels are normalized to GAPDH, and the fold change (2^−ΔΔCT) was calculated in comparison to the non-treated HEK293 cells.

Figure 6

Transient expression of sense/antisense cassettes in HEK293 cells. Cells were transfected with two doses (50 and 250 ng/mL) of cDNA encoding, either the sense transcript (dark grey) or the antisense transcript (light grey). Expression levels of sense and antisense transcripts were assayed by RT-qPCR to detect potential post-transcriptional stabilizing effects. All values are related to untransfected HEK293 cells. To estimate PCR background, GFP was amplified in all samples (control and shaded area reflecting non-significant changes).

Figure 7

Functional expression of SLC34A1 in Xenopus oocytes and HEK293 cells. (A) Wild type HEK293 cells were grown with and without dexamethasone for 5 days, followed by uptake measurement using radioactive Pi. Values were scaled to an average control value of 100%, t-test revealed significance, p < 0.01. (B) RNA was extracted from HEK293 cells, untreated (WT) and after 5 days exposure to dexamethasone (WT + dex). The same procedure was carried out with HEK293 cells that had biallelic insertion of the antisense cassette (AS biallelic KO + dex). Total RNA was injected into Xenopus oocytes, and the uptake of radioactive Pi was measured after 3 days. One-way ANOVA followed by Tukey’s test for multiple comparisons, p < 0.01.