1005 - Toll-like receptor 9 signalling in CLL: a resistance mechanism to B-cell receptor-targeted treatments, and a potential tool for therapeutic stratification.

Background:
CLL cell trafficking to secondary lymphoid tissues is fundamental to disease progression, since within these protective niches, CLL cells encounter an abundance of activating, pro-proliferative and pro-survival signals. The primary mechanism of CLL cell activation is via the B-cell receptor (BCR)-signalling pathway, and BCR-targeted treatments are extremely effective at releasing tissue-resident cells into the peripheral blood and inducing apoptosis. Despite these successes, currently available agents fail to achieve complete tumour clearance, and CLL remains incurable.

Toll-like receptor 9 (TLR9) is an intracellular pattern recognition receptor, which responds to unmethylated CpG motifs in bacterial/viral/mitochondrial DNA. We recently showed unmethylated DNA levels are 28-fold higher in CLL patient plasma relative to healthy controls, and that TLR9-ligation induces an NF-κB and STAT3-driven activation/migratory phenotype in primary CLL cells[1]. We hypothesise TLR9 signalling is a BCR-independent contributor to CLL homing and potential resistance mechanism to BCR-targeted agents.

Results:
In a cohort of 74 primary CLL samples, a dichotomous migratory response was observed in response to the TLR9 agonist ODN 2006. 53% showed an increase in CLL cell migration (categorised as ‘Responders’[R]), whilst 47% showed either no change or a decrease (categorised as ‘Non/Reverse Responders’[NR/RR]). There was a significant (but not exclusive), correlation with mutational status, with 71% vs. 33% of M-CLL vs. U-CLL samples being categorised as R, respectively (P=0.03, n=48), and no correlation was found between CD49d+ve vs. CD49d-ve or CD38+ve vs. CD38-ve samples.

Surprisingly, there was no significant difference in total TLR9 expression between R and NR/RR samples (P=0.23, n=45), and ODN 2006 stimulation induced TLR9 signalling in both subgroups. Post-stimulation, IRAK1 and IκBα were degraded in R (n=4) and NR/RR (n=3) and the activation marker CD69 was equally upregulated in each subgroup (P < 0.0001, n=34 and P < 0.0001, n=28, respectively). This indicates the observed dichotomous migratory response is due to downstream signalling divergence, as opposed to a complete lack of response to stimulation.

Previous studies have shown constitutive BCR activation to be heterogeneous in CLL, and generally elevated in U-CLL relative to M-CLL[2]. We hypothesised that the most basally activated CLL cells may reach their maximal activation/migratory capacity through BCR-signalling-alone, rendering them non-responsive to further external stimuli. This is supported by our finding that NR/RR have significantly lower basal CD5 (P=0.02, n=54), and significantly higher NF-κB p-p65 (P <0.01, n=12) and MCL-1 (P=0.03, n=16).

Using in-silico modelling, we simulated TLR9 activation in states of low/high basal BCR activity, using canonical NF-κB activity as a functional read-out. Our model showed cells with low basal activation responded strongly to TLR9 stimulation, whilst cells with high basal activation were unresponsive. These results were verified in-vitro as basal p-BTK and p-p65 (representative of constitutive BCR activation) showed negative correlation with the migratory response to ODN 2006 in the NR/RR subgroup (P < 0.01, R=-0.91, n=8 and P=0.02, R=-0.93, n=5, respectively); these data indicate constitutive signalling impairs the ability of NR/RR samples to respond to alternative stimuli. Interestingly, we found responsiveness to ODN 2006 (migration) correlated strongly with responsiveness to IgM (p-BTK expression) (P=0.03, R2=0.33, n=14), signifying that this trend is not TLR9-specific, but extends to other microenvironmental interactions.

Importantly, when simulating NR/RR treatment with a BTK inhibitor in-silico, we observed a renewed responsiveness to TLR9 activation, which was later verified in-vitro. In the presence of the BTK-inhibitor ibrutinib, we found a subset of previously non-responsive NR/RR samples became ‘sensitised’ to TLR9 activation, showing an increase in CLL cell migration in response to stimulation with ODN 2006. This suggests a switch from BCR to TLR9 signalling in some NR/RR samples and implicates TLR9 signalling as a tumour escape mechanism following BTKi therapy. We hypothesise that TLR9 ‘responder’ and ‘sensitised’ subgroups may benefit from BCR/TLR dual targeted therapy.

We are currently investigating NF-κB as a potential therapeutic target to simultaneously inhibit BCR/TLR9 signalling. Whilst pan NF-κB inhibitors have proven extremely toxic (in-vivo), we are opting for a more streamlined subunit-specific approach. We are working to create patient-specific NF-κB ‘fingerprints’ (i.e., subunit expression profiles), to ascertain whether different subgroups show targetable differences in NF-κB subunit expression. Early analyses suggest fingerprints to be distinct between R and NR/RR samples, rendering this technique a potential tool for personalising future NF-κB-targeted therapies.