Single-cell, single-mRNA analysis of Ccnb1 promoter regulation

Promoter activation drives gene transcriptional output. Here we report generating site-specifically integrated single-copy promoter transgenes and measuring their expression to indicate promoter activities at single-mRNA level. mRNA counts, Pol II density and Pol II firing rates of the Ccnb1 promoter transgene resembled those of the native Ccnb1 gene both among asynchronous cells and during the cell cycle. We observed distinct activation states of the Ccnb1 promoter among G1 and G2/M cells, suggesting cell cycle-independent origin of cell-to-cell variation in Ccnb1 promoter activation. Expressing a dominant-negative mutant of NF-YA, a key transcriptional activator of the Ccnb1 promoter, increased its “OFF”/“ON” time ratios but did not alter Pol II firing rates during the “ON” period. Furthermore, comparing H3K4me2 and H3K79me2 levels at the Ccnb1 promoter transgene and the native Ccnb1 gene indicated that the enrichment of these two active histone marks did not predispose higher transcriptional activities. In summary, this experimental system enables bridging transcription imaging with molecular analysis to provide novel insights into eukaryotic transcriptional regulation.


Expression of the Ccnb1::Luc-MS2 transgene mimics the native Ccnb1 gene in synchronized cells
To reproduce and validate our findings on Ccnb1 promoter regulation during the cell cycle, we performed cell cycle synchronization experiments to examine cells that were blocked at G1 and progressed into G2 (6 hours after release) and into the next G1 phase (10 hours after release) ( Supplementary Fig. S7). We then examined Ccnb1::Luc-MS2 transgene and native Ccnb1 expression by single molecule RNA FISH ( Supplementary Fig. S7). As expected, G2 cells had higher mRNA counts for both the Ccnb1::Luc-MS2 transgene and the native Ccnb1 gene than G1 cells ( Supplementary Fig. S7). We note that mRNA counts in synchronized G2/M or G1 cells were lower than mRNA counts in G2/M-arrested or G1-arrested cells (Fig. 4b, e), likely due to the prolonged cell cycle arrest induced by nocodazole/mimosine that may increase cell volumes resulting in higher mRNA counts 1,2 . Importantly, we observed reduced transcription site (TS) brightness at both the Ccnb1::Luc-MS2 transgene and the native Ccnb1 gene in G1 than in G2/M ( Supplementary Fig. S7). Under the same assumptions used in this study, we estimated that the average Pol II density at the native Ccnb1 gene was 1.42 kb -1 and 1.74 kb -1 in G1 and G2/M-synchronized cells, respectively and that the average Pol II firing rate at the native Ccnb1 gene was 1.57 ± 0.22 min -1 and 1.93 ± 0.28 min -1 in G1 and G2/Msynchronized cells, respectively. We observed that a higher fraction of cells expressed both the Furthermore, we examined FISH images of cells synchronized in G2 and the next G1 phase. We confirmed the existence of two cell populations that differed in Ccnb1 promoter activity and Luc-MS2 mRNA localization among synchronized G1 or G2/M cells ( Supplementary Fig. S11). We identified that 39% cells and 69% cells exhibited nuclear-  S11). Taken together, our study identified cell-to-cell variations in Ccnb1 promoter activities and Luc-MS2 mRNA localizations among asynchronous cells, G1-or G2/M-arrested cells and among cells synchronized at G1 or G2/M. Therefore, variable activation states of the Ccnb1 promoter likely exist among individual cells during each cell cycle stage.

Differential changes in Luciferase and Ccnb1 expression detected by qPCR and by FISH during the cell cycle
We noticed some differences in Ccnb1::Luc-MS2 transgene expression and native Ccnb1 gene expression detected by FISH or by RT-qPCR. Compared to asynchronous cells, Luciferase mRNA levels decreased more than 2-fold in G1 and increased by ~50% in G2 as detected by RT-qPCR, resulting in an approximately ~4-fold increase of Luciferase mRNA in G2 than in G1 ( Supplementary Fig. S5). In contrast, we observed by FISH that the median Luc-MS2 mRNA counts were 302, 335 and 426 in asynchronous, G1 and G2 cells,respectively (Figs. 2c,4b).
Thus, Luc-MS2 mRNA counts only increased by 30-50% in G2 cells compared to asynchronous cells or G1 cells. Likewise, native Ccnb1 mRNA levels decreased by ~25% in G1 and increased by ~2-fold in G2 (compared to asynchronous cells), resulting in an ~2.5-fold increase of Ccnb1 mRNA in G2 than in G1 ( Supplementary Fig. S5). In contrast, FISH detected about ~50% higher native Ccnb1 mRNA counts in G2 cells than asynchronous cells or G1 cells (Figs. 3c, 4e, median mRNA counts = 391, 406 and 593 in asynchronous, G1 and G2, respectively). Therefore, both RT-qPCR and FISH observed increased Luciferase and Ccnb1 mRNA levels in G2 cells than in asynchronous cells or G1 cells. However, the two methods differed in that decreased Luciferase and Ccnb1 mRNA levels in G1-arrested cells than in asynchronous cells were observed by RT-qPCR ( Supplementary Fig. S5) but were not detected by FISH (Figs. 2c,3c,4b,4e). Because approximately two-thirds of asynchronous cells were in G1 (Supplementary Figs. S5, S12), a 2-3 fold decrease of Luciferase mRNA in G1 cells than asynchronous cells would require a 4-8 fold higher Luciferase mRNA in S/G2 cells than in G1 cells among the asynchronous population.
Although MS2 RNA FISH demonstrated excellent dynamic range in detecting Luc-MS2 transgene expression (Figs. 2a, d), we did not observe such a 4-8 fold higher Luc-MS2 mRNA count in G2 than in G1 either in arrested cells (Fig. 4b) or synchronized cells ( Supplementary   Fig. S7), nor could we distinguish two cell populations differing in mean Luc-MS2 mRNA counts among asynchronous cells (Fig. 2c). Because RT-qPCR can be subject to artifacts during cell lysis and RNA isolation while FISH quantitation is sensitive to enlarged cell volumes 1,2 , we note here that one should be cautious to interpret conflicting FISH and RT-qPCR data on gene expression changes.

Splinkerette PCR
Splinkerette PCR was carried out as described 3 . Briefly, genomic DNA was extracted and digested with BstYI at 60 ºC. The purified digested genomic DNA was ligated to annealed Splinkerette oligonucleotide (Supplementary Table S5) and two rounds of PCR were carried out with Phusion Taq polymerase (NEB) and PCR primers (Supplementary Table S5). The final PCR products were resolved in agarose gel. The PCR products were then treated with Antarctic phosphatase (NEB) and Exonuclease I (NEB) and subjected to sequencing.

Luciferase reporter activity assays
C2C12 cells were plated in a 24-well plate at a density of 4 × 10 4 cells/well and transfected 24 hours later with 180 ng of pGL3 empty vector (Promega) or pGL3 vectors containing four Ccnb1 promoter deletion mutants ( Supplementary Fig. S2). Renilla vector pRL (Promega) (20 ng) was co-transfected to normalize for the transfection efficiency. Luciferase activity was measured 48 hours after transfection using the Dual Luciferase Reporter Assay System according to the manufacturer's instructions (Promega).

RNA extraction and real-time PCR analysis
Total RNA was isolated using the RNeasy Plus Mini Kit (Qiagen). Four hundred nanograms of total RNA was converted to cDNA using SuperScript III Reverse Transcriptase (Invitrogen) in conjunction with oligo(dT) primers (Invitrogen). The resulting cDNA samples were subjected to real-time PCR using gene-specific primers and iTaq Universal SYBR ® Green Supermix (Bio-Rad). Real-time PCR was performed in a CFX-96 Touch Real-Time PCR Detection System (Bio-Rad) and the data were analyzed using the CFX Manager software (Bio-Rad).

Measurement of mRNA decay
mRNA half-lives were calculated by measuring mRNA levels at distinct time points after adding the transcription inhibitor 5,6-Dichlorobenzimidazole 1-β-D-ribofuranoside (DRB) (Sigma) at a final concentration of 30 μg/ml into the culture media. Total RNA was extracted at the time points 0, 0.5 hour, 1 hour, 2 hours and 4 hours after adding DRB. RT-qPCR was carried out using 18S rRNA to normalize the expression. mRNA decay rates were determined as previously described 4 . Briefly, (2 −( − 18 ) ) at distinct time points after DRB treatment were plotted on the y-axis and the corresponding time points were plotted on the x-axis.
was determined by linear regression, and mRNA half-life 1 2 was calculated as 1 2 = 2 .
Three independent experiments were carried out and the average 1 2 was calculated.

Propidium iodide (PI) staining and cell cycle analysis by flow cytometry
Cells were trypsinized and centrifuged at 1,200 rpm for 5 min and were washed with cold 1X PBS three times. imaged in an Olympus IX-81 wide-field microscope (immunofluorescence in the green channel and RNA FISH in the red channel) as described above.
Protein concentration in the soluble fraction was determined by BCA protein assay (ThermoFisher Scientific). Proteins were separated by electrophoresis through 10% Tris-glycine gels (Bio-Rad) and transferred to nitrocellulose membrane. Primary antibodies used for western blot analysis were mouse monoclonal anti-NF-YA (Santa Cruz Biotechnology, sc-10779) at 1:500 dilution and rabbit polyclonal anti-α Tubulin (Abcam, ab18251) at 1:1000 dilution.
Secondary antibodies used were horseradish peroxidase-conjugated goat anti-rabbit IgG (1:5000 dilution). The blot was then developed using SuperSignal West Femto kit (ThermoFisher Scientific) and images were acquired with a ChemiDoc XRS+ Imaging (BioRad) according to the manufacturer's instructions.

Chromatin Immunoprecipitation
Cells were cultured to ~70-80% confluency in five 150 mm plates. Cells were fixed with 1% formaldehyde in serum-free media for 10 min at RT on a shaker. Formaldehyde crosslinking was quenched with a final concentration of 125 mM glycine. Cells on each plate were then washed in cold 1X PBS and scraped in 5 ml of 1X PBS. Cells were then pelleted by spinning at ~5,000 rpm for 5 min and lysed with 4 ml cell lysis buffer (5 mM PIPES, pH 8.0, 85 mM KCl, 0.5% NP-40). After incubating on ice for 5 min, the nuclei were washed with cell lysis buffer and spun down again. The nuclei were then lysed by nuclei lysis buffer (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 1% SDS) and incubated on ice for 10 min. The nuclear extract was transferred into the TPX plastic microtubes (Diagenode) for sonication. The nuclear lysate was then sonicated in a Bioruptor sonicator (Diagenode) at high settings for 20-minute sonication cycle, 30sec ON/30sec OFF at 4 ºC to yield DNA fragments with an average length of ~500 bp.
(g) Fractions of cells detected with Luc-MS2 mRNA FISH signals and native Ccnb1 mRNA FISH signals in the control cell clone pSG5-1 and cell clones stably expressing the NF-YAm29 mutant (m29-5 or m29-3). (h) Fractions of cells displaying uniformly-or nuclear-localized Luc-MS2 mRNA in the control cell clone pSG5-1 and cell clones stably expressing the NF-YAm29 mutant (m29-1 or m29-5). In panels g and h, Fisher's exact test was used to determine the P value. Asterisks (*) indicate statistically significant differences (p < 0.05). FISH images from one experiment were analysed. Asterisks (*) indicate statistically significant differences in mRNA counts or TS brightness between cells with uniformly-or nuclear-localized Luc-MS2 mRNA determined by students' ttest (p < 0.05). FISH images from one experiment were analysed.