Relationships between putative G-quadruplex-forming sequences, RecQ helicases, and transcription

Abstract

Background: Putative G-quadruplex-forming sequences (PQS) have long been implicated in regulation of transcription, though the actual mechanisms are not well understood. One proposed mechanism involves the activity of PQS-specific helicases belonging to the RecQ helicase family. However, patterns of PQS that correlate with transcriptional sensitivity to RecQ helicases are not well studied, and no adequate transcriptional model exists to account for PQS effects.

Methods: To better understand PQS transcriptional effects, we analyze PQS motifs in genes differentially-transcribed in Bloom Syndrome (BS) and Werner Syndrome (WS), two disorders resulting in loss of PQS-interacting RecQ helicases. We also correlate PQS genome-wide with transcription in multiple human cells lines while controlling for epigenetic status. Finally, we perform neural network clustering of PQS motifs to assess whether certain motifs are over-represented in genes sensitive to RecQ helicase loss.

Results: By analyzing PQS motifs in promoters of genes differentially-transcribed in BS and WS, we demonstrate that abundance of promoter PQS is generally higher in down-regulated genes and lower in up-regulated genes, and show that these effects are position-dependent. To interpret these correlations we determined genome-wide PQS correlations with transcription while controlling for epigenetic status. Our results identify multiple discrete transcription start site-proximal positions where PQS are correlated with either increased or decreased transcription. Finally, we report neural network clustering analysis of PQS motifs demonstrating that genes differentially-expressed in BS and WS are significantly biased in PQS motif composition.

Conclusions: Our findings unveil unappreciated detail in the relationship between PQS, RecQ helicases, and transcription. We show that promoter PQS are generally correlated with reduced gene expression, and that this effect is relieved by RecQ helicases. We also show that PQS at certain positions on the downstream sense strand are correlated with increased transcription. We therefore propose a new transcriptional model in which promoter PQS have at least two distinct types of transcriptional regulatory effects.

Fig. 1

PQS occurrence in genes differentially expressed in BS and WS. PQS occurrence in sense (S) and antisense (AS) strands was analyzed independently. a Heat map showing PQS occurrence genome-wide and in genes differentially-expressed in BS and WS, represented as raw counts normalized to total number of TSS in each dataset. b Top panels show p-values for comparing PQS abundance per TSS between genes differentially expressed in BS and WS and all other genes. Dotted lines represent p-value cutoffs for determining statistical significance, with less than 1 % of data points from a random gene dataset of the same size as the test dataset having a p-value below this threshold. Bottom panels show PQS enrichment ratio in genes differentially expressed in BS and WS, with values > 1 indicating that PQS are more abundant in the differentially-expressed gene set. Both p-values and enrichment ratios were calculated using a 200 bp bin value repeated at a 10 bp interval. Regions with shaded peaks represent locations of statistically-significant PQS excess or scarcity

Fig. 2

Epigenetic prediction of gene expression to identify PQS positions correlated with altered gene expression. a Correlation of log₂-transformed epigenomic track signals with log₂-transformed gene expression represented as a function of location 1 kbp upstream and downstream of TSS. Bin size is 100 bp. Prior to log₂-transformation, 0.25 % of the maximal bin value was added to each bin feature to avoid the log₂(0) issue. b Representative correlation of CAGE TSS-specific gene expression measurements with epigenetic model predictions generated through Bayesian linear regression. Models were trained on datasets containing half of all PQS-free TSS in the human genome. Gene expression predictions were then generated for the remainder of the PQS-free TSS. c Analysis of model prediction error as a function of PQS position, assigned to sense (S) and antisense (AS) strand effects. Prediction error estimates were sorted into 200-bp bins iterated at a 10-bp interval, based on the position of the PQS nearest to the TSS, and then compared to the prediction error estimates for PQS-free TSS using 1-way ANOVA. The top panels show the p-value from this analysis as a function of position with respect to the TSS. The dashed line represents the p-value threshold used for determining statistical significance in the analysis. The bottom panels show prediction error averages for the binned values. Areas with shaded peaks represent locations where the p-value is statistically significant for rejection of the null hypothesis. Prediction error value > 0 represent positions where PQS presence correlates with higher gene expression. Prediction error values < 0 represent positions where PQS presence correlates with lower gene expression. d Aggregate map of all statistically-significant positions from prediction error analysis based on seven human cell lines

Fig. 3

Analysis of PQS abundance in BS and WS along with position-dependent correlations with transcription. Heat map showing PQS excess or scarcity in genes differentially expressed in BS and WS (yellow = excess, blue = scarcity), as well as the correlation of PQS at these positions with transcription (green = high transcription, red = low transcription)

Fig. 4

PQS motif characteristics in genes differentially expressed in BS and WS analyzed via ANOVA. PQS motif characteristics (total sequence length, loop lengths, base fractions, and number of G-stacks) in genes differentially expressed in BS and WS. P-values and ratios for comparison to the genome-wide distribution are calculated using a 200 bp bin selection, repeated at a 50 bp interval, and in reference to corresponding PQS position on antisense (AS) strand or sense (S) strand for all genes. Red = genes up-regulated in BS and WS. Blue = genes down-regulated in BS and WS. Solid line = AS. Dashed line = S

Fig. 5

Self-organizing map analysis of PQS motifs in genes differentially-expressed in BS and WS. Multidimensional self-organizing map clustering of all human genome PQS within 2 kbp of TSS, based on total sequence length, loop lengths, number of G-stacks, and nucleotide base fractions. a PQS parameters for node centroids. b Counts of PQS binned by distance in multidimensional space to closest node. c Distance between nodes. d PQS node bias in genes differentially expressed in BS and WS. Bias represents number of unique PQS per node, normalized to dataset gene number, compared to the same calculation for all genes genome-wide. Green and red nodes represent positive (excess) or negative (scarcity) bias values that are outside of the 95 % CI for randomly-generated gene datasets of the same size. Green = excess. Red = scarcity. Gray = no statistically-significant difference

Fig. 6

A model to account for PQS transcriptional correlations. Model summarizes PQS position- and strand-dependent correlations with gene transcription determined from the present work and possible molecular mechanisms to explain these observations. An increase in transcription may be facilitated by the formation of intramolecular G-quadruplex structures between 140–270 bp downstream of the TSS on the sense strand (a), resulting in the release of the antisense strand from Watson-Crick base pairing and enabling easier access of the antisense strand to RNAPII, aiding transcription initiation (b). Formation of G-quadruplex structures in this region may also recruit G-quadruplex DNA-binding proteins (c) which may further stabilize G-quadruplex structures and enhance transcription. Resolution of G-quadruplex structures by BLM and WRN helicases (d) attenuates this transcriptional-activating effect. Transcriptional decrease mediated by PQS downstream of the TSS on both DNA strands may be facilitated by intramolecular G-quadruplex formation, resulting in a biophysical roadblock which prevents the passage of RNAPII and hinders transcript elongation (e). Resolution of G-quadruplex structures by BLM and WRN helicases in this region attenuates the transcriptional-repressing effect of the biophysical roadblock (f). Transcriptional decrease mediated by PQS upstream of the TSS may be recruitment of trans-acting factors that attenuate transcription (g). Resolution of G-quadruplex structures in this region attenuates this transcriptional repressing effect (h)