Aggregation and secondary loop structure of oligonucleotides do not determine their ability to inhibit TLR9.

Toll-like receptor 9 (TLR9) is an endosomal DNA sensor that warns us of the presence of infectious danger and triggers a rapid pro-inflammatory response in dendritic cells, macrophages, and B cells. The consequences of uncontrolled TLR9 activation can be detrimental for the host, contributing to the pathogenesis of bacterial septic shock or autoimmune diseases, such as systemic lupus erythematosus. Therefore, we need to develop TLR9 antagonists. We and others have created inhibitory oligonucleotides (INH-ODN) that are capable of sequence-dependent inhibition of TLR9-induced activation in both human and mouse cells. However, it is not clear whether marked differences in INH-ODN activity related to base sequence derived from polymerization of INH-ODNs or their ability to complex with stimulatory CpG-oligonucleotides (ST-ODN). Furthermore, the 5' end of INH-ODNs may assume a particular loop configuration that may be needed for binding to a critical site on TLR9. Here, we show that 1) G-tetrads required for ODN stacking were compatible with INH-ODN activity but were not necessary; 2) there was no relationship between activity and self-association at endosomal pH; 3) there was no evidence for direct binding between ST-ODNs and INH-ODNs; 4) when a 3G sequence was disrupted, despite a preserved stem-loop formation, INH-ODN activity was abolished. These results support the conclusion that certain features of the primary linear sequence are critical for TLR9 inhibition, but changes in secondary structure or in ODN aggregation are irrelevant.