Mechanisms and implications of transcription blockage by guanine-rich DNA sequences

Abstract

Various DNA sequences that interfere with transcription due to their unusual structural properties have been implicated in the regulation of gene expression and with genomic instability. An important example is sequences containing G-rich homopurine-homopyrimidine stretches, for which unusual transcriptional behavior is implicated in regulation of immunogenesis and in other processes such as genomic translocations and telomere function. To elucidate the mechanism of the effect of these sequences on transcription we have studied T7 RNA polymerase transcription of G-rich sequences in vitro. We have shown that these sequences produce significant transcription blockage in an orientation-, length- and supercoiling-dependent manner. Based upon the effects of various sequence modifications, solution conditions, and ribonucleotide substitutions, we conclude that transcription blockage is due to formation of unusually stable RNA/DNA hybrids, which could be further exacerbated by triplex formation. These structures are likely responsible for transcription-dependent replication blockage by G-rich sequences in vivo.

Fig. 1.

Transcription blockage by various G-rich inserts. Lanes designated by vertically written “10” and “100” correspond to denatured 10 and 100 nucleotide DNA size markers, respectively. Above the full gel image, lower exposures for the runoff bands are shown. (A) Linear DNA template. The dashed-lined box shows a higher exposure for the blockage products. At the top of the box, sequences of the inserts (nontemplate strands) are shown above the corresponding lanes, and within the box, block arrows, chevrons, and ovals show repeat-exiting, interruption, and diffuse blockage bands, respectively. (B) The same experiment as in A, but for negatively supercoiled DNA templates. Note that for the supercoiled DNA, we put the term “runoff” within quotation marks, because actually it is likely to be a heterogeneous mixture of long transcription products obtained due to spontaneous transcription termination somewhere within the circular plasmid.

Fig. 2.

Both upstream and downstream G stretches contribute to the blockage signal at the C4 interruption. The block arrow and the chevron show the repeat-exiting and interruption blockage bands, respectively. The interruption blockage band is well-pronounced only for the sequence G8C4G16. Presented results are for scDNA.

Fig. 3.

Effect of monovalent cations and nucleoside substitution on transcription blockage. (A) The blockage is not sensitive to monovalent cations. In the standard transcription buffer (see Materials and Methods), NaCl was replaced by 83 mM of either KCl or LiCl. The ratio of the intensity of the blockage band (normalized to runoff) for K to the intensity of the blockage band (normalized to runoff) for Li was 0.6. Thus, K does not facilitate blockage in comparison with Li. These results are for scDNA. (B) Substitution of ITP, but not 7-deaza-GTP for GTP abolishes blockage. Guanosine (G) was replaced by either inosine (I) or 7-deazaguanosine (D) in transcription reaction with scDNA substrates. It is seen that for inosine there is no detectable difference between blocking insert G32 and control C32. In contrast, in the case of 7-deazaguanosine there is blockage and decrease in runoff product in the case of the G32 insert, though the blockage signal (shown by white rectangle) is somewhat weaker, more smeary, and more shifted downstream from the insert in comparison with guanosine. An additional more contrasted image for lanes 3 and 4 with 7-deazaguanosine is shown at Right.

Fig. 4.

Transcription-dependent replication blockage in E. coli cells. (A) A general scheme for the 2D gel electrophoresis of bubble-like replication intermediates. Replication fork moves unidirectionally from left to right producing replication intermediates with gradually decreasing electrophoretic mobility, which together form an arch on the gel. Fork stalling at an obstacle (gray rectangle) leads to accumulation of corresponding replication product, producing a bulge on the arch (black oval), which intensity reflects the “strength” of the replication blockage. (B) Gel electrophoresis of replication intermediates. The sequences above the figures correspond to the nontemplate (sense) strand for transcription, which is also the lagging strand template for DNA replication. Bulges on the replication arcs (marked by arrows) indicate replication stall sites. The bulges are equally well-pronounced for the G32 insert and its interrupted derivative, G20C4G8, whereas they are barely detectable for the C32 insert and its interrupted derivative, C20G4C8.

Fig. 5.

Suggested model for transcription blockage by G-rich homopurine-homopyrimidine sequences. RNAP is shown as a dashed-line circle, the hPu DNA, hPyDNA, regular DNA, hPuRNA and regular RNA sequences are shown in cyan, orange, gray, magenta, and black, respectively. Watson–Crick and Hoogsteen base pairs are shown as short, thin, black lines, and thicker dark blue lines, respectively. (A) RNAP transcribes into the hPu/hPy sequence without any obstacles, until the hPu transcript reaches the size of the RNA/DNA hybrid inside the transcriptional complex, and the extrastable RNA/DNA duplex has to be unwound to provide nascent RNA extrusion. This might cause a transcription impediment or partial blockage, which could be exacerbated if the RNAP encounters an interruption or end of the G stretch, because in that case it begins to synthesize a less stable RNA/DNA duplex, while still unwinding the extrastable RNA/DNA duplex to extrude nascent RNA. (B) The blockage could be further exacerbated if the nontemplate strand interacts with the downstream duplex forming a triplex. A more realistic drawing of the triplex is shown at Right. (C) During transcription of the pyDNA template, the RNA could rehybridize with the DNA template forming an R loop, which also could exacerbate the blockage (see the text). Triplex and R-loop pathways could be either mutually exclusive, each occurring with a certain probability during transcription, or the triplex formation could be reversible and precede the R-loop formation.