Peripheral blood and bone marrow aspirates were obtained from CLL patients attending the CLL clinic at CancerCare Manitoba. Informed consent of patients and control subjects was obtained under a protocol approved the Research Ethics Board at the University of Manitoba. Rai staging was determined using standard clinical criteria. Clinical biomarkers including CD38, IgM, IgG, IgA and lymphocyte count were determined using standard protocols and obtained from the Manitoba Tumor Bank and CAISIS database. Mononuclear cells were isolated using Ficoll-Paque density gradient and cryopreserved in 10% DMSO medium prior to analysis. Lymph node biopsies were formalin-fixed and paraffin-embedded prior to sectioning.

For assessment of T_FH subpopulations, peripheral blood or bone marrow mononuclear cells were stained for the markers CD3, CD4, CD14, CD19, CXCR5, PD-1, ICOS, CD45RA, CCR7, CXCR3, CCR6, TIGIT (antibody details in Supplementary Table 1 ) and LIVE/DEAD™ Fixable Aqua viability dye (Invitrogen™) at room temperature for 30 minutes. Stained cells were run on a Beckman Coulter Cytoflex instrument. T_FH populations were quantified as percent of the singlet, CD4+, Dump (CD19/CD14/LiveDead) negative population as illustrated in Supplementary Figure 1 .

For assessment of cytokine production, cryopreserved peripheral blood mononuclear cells (PBMC) were cultured overnight and then stimulated for 6 hours with 50ng/ml PMA and 1μg/ml ionomycin (Selleck Chemicals), with 10μg/ml Brefeldin A (Selleck) added for the last 4 hours. Cells were then cell surface stained for CD3, CD14, CD19, CXCR5, CCR6, CXCR3 as described above, fixed and permeabilized using eBioscience™ Fixation/Permeabilization buffer and stained intracellularly for CD4, IFNγ, IL-21 and CD40L at room temperature for 45 minutes. Intracellular CD4 staining was utilized to improve discrimination of CD4+ cells, as cell surface CD4 is downmodulated upon treatment with PMA+ionomycin. Antibody details are provided in Supplementary Table 1 .

Lymph node tissue sections were deparaffinized, rehydrated and boiled for 20 minutes in Target Retrieval Solution, Citrate pH 6.1 (Agilent). After washing, serial tissue sections were blocked with 1% BSA and 2% FBS in PBS followed by staining with unconjugated primary antibodies at 4°C overnight. After washing, sections were incubated with secondary antibodies for 4 hours at room temperature with shaking, washed and then stained with directly conjugated primary antibodies at 4°C overnight. Following a final wash, sections were air dried and mounted with ProLong™ Gold antifade reagent (Invitrogen) and kept at -20°C until analysis. Antibody details are provided in Supplementary Table 2 . Images were captured with a CSU-X1M5000 spinning disc confocal microscope (Carl Zeiss) equipped with 405/488/561/635nm lasers.

Autologous CD4 T cells and CLL B cells were purified from PBMC using EasySep™ Human CD4+ T Cell Isolation Kit and EasySep™ Human B Cell Enrichment Kit II without CD43 Depletion (both STEMCELL Technologies), respectively. Purified CD4 T cells were suspended at 1x10^6 cells/mL in RPMI 1640 media (GE Healthcare) with 10% FBS (Life Technologies). T cells were cultured overnight in 24-well plates with/without addition of ImmunoCult™ Human CD3/CD28/CD2 T Cell Activator (STEMCELL Technologies) at 25 μL cocktail/mL of cells. B cells were cultured overnight in U-bottom 96-well plates at 1x10^6 cells/well, in the presence of sCD40L+IL-4 (both 50ng/mL; R&D Systems). After 14-16 hours incubation, cells were washed to remove stimuli. B cells were stained with carboxyfluorescein succinimidyl ester (CFSE) (Sigma-Aldrich) at 0.3 μM in PBS for 5min at room temperature, then washed with culture media. T cells (with or without pre-activation) were co-cultured together with CFSE-labelled autologous CLL-B cells in U-bottom 96-well plates, using 2x10^5 T and 1x10^6 B cells per well. T and B cells alone were included as controls. Starting at day 2 of co-culture, 100 μL of culture medium was gently taken out and fresh medium added to each well daily. At indicated times, wells were harvested and flow cytometry analyses were carried out using the panel detailed in Supplementary Table 1 . Briefly, cells were stained for CD4, CD19, CXCR5, CXCR3, CCR6, CCR7, CD69, CD25, CD134/OX40, PD-1, CD38 and LIVE/DEAD™ Fixable Aqua viability dye (Invitrogen™). Following wash, cells were fixed and permeabilized as above and stained for Ki-67 at room temperature for 45 min.

Flow data were analyzed by FlowJo^® V10 (FlowJo, LLC). Statistical analysis was performed with GraphPad Prism (GraphPad Software Inc). Confocal images were processed by ImageJ (V1.47). Box and whisker plots illustrate the median, interquartile range and 10-90% percentile values. Statistical tests used are indicated in figure legends and differences were considered to be statistically significant at values of *(p<0.05), **(p<0.01), ***(p<0.001) and ****(p<0.0001).

Peripheral blood mononuclear cells collected from CLL patients, MBL patients or age-matched controls were analyzed by multicolor flow cytometry to assess T follicular helper (T_FH) cell populations. Gating on CD4+CXCR5+CD19-CD14- live lymphocytes ( Figure S1 ) revealed a significant elevation in both T_FH frequencies and overall T_FH numbers in CLL but not MBL patients ( Figure 1A ). CLL T_FH express higher levels of T_FH-associated activation markers PD-1 and ICOS than corresponding non-T_FH CD4 T cells and more PD-1 than control T_FH ( Figure 1B ). Compared to control T_FH, CLL T_FH populations contain a higher proportion of CD45RA-/CCR7- effector memory cells and fewer CD45RA-CCR7+ central memory cells ( Figure S2 ). Within the T_FH population we further examined T_FH1, T_FH2 and T_FH17 subset composition based on expression of chemokine receptors CXCR3 and CCR6 (24) and found that CLL patients demonstrate significant skewing towards the CXCR3+CCR6- T_FH1 population ( Figure 1C ). The increased T_FH1 skewing in CLL patients was accompanied by significantly reduced frequencies of the T_FH2 population, whereas no significant change in T_FH17 cells was observed ( Figure 1C ). Together these results indicate that T_FH cells are expanded and phenotypically altered in CLL.

Within the spectrum of CLL patients in the cohort analyzed, we examined whether T_FH frequencies were associated with clinical parameters or established biomarkers of disease. Strikingly, we found that both frequency of T_FH among CD4+ T cells and T_FH1 skewing were positively correlated with blood lymphocyte counts ( Figure 2A ). T_FH expression of PD-1 or ICOS were also positively correlated with lymphocyte count ( Figure S3 ). Frequencies of T_FH and skewing to T_FH1 were significantly elevated in high risk CLL (Rai 3-4) relative to low risk CLL (Rai 0) and CD38+ CLL patients showed more T_FH1 skewing than CD38- patients ( Figure 2B ). The latter is consistent with a report that CD38 expression is driven by the T_H1 cytokine IFNγ²⁵. Interestingly, T_FH frequency was inversely correlated with plasma IgM and IgG levels ( Figure 2C ). Together these results suggest that T_FH accumulation and selective skewing to a T_FH1 phenotype occurs in parallel with expansion of the leukemic B cell clone and decline in normal B cell function.

To determine how CLL treatment impacts T_FH populations, we examined patients during the first year of ibrutinib treatment. Ibrutinib treatment led to a gradual decline in the frequency of T_FH cells over time, in parallel with a decline in total lymphocyte count ( Figure 3A ). In one patient who exhibited prolonged lymphocytosis after treatment, there was a concurrent transient increase in the T_FH population prior to normalization. Notably, T_FH composition changed post-ibrutinib treatment, with patients exhibiting a gradual re-balancing of T_FH1, T_FH2, and T_FH17 subsets ( Figure 3A ). Significant reductions in both T_FH frequencies and T_FH1 skewing were observed after 40 weeks of treatment ( Figure 3B ). This was accompanied by increased frequencies of CXCR3-CCR6- T_FH2-like cells, while T_FH17 frequencies were not altered ( Figure 3C ). These results suggest that ibrutinib treatment can normalize T_FH subsets concomitant with reduction in disease burden.

We further examined expression of costimulatory molecules and cytokines by CLL T_FH cells. We found that T_FH express higher levels of CD40L than non-T_FH in both CLL patients and controls, however CLL T_FH exhibit a strikingly elevated expression of CD40L (approximately 5-fold on average) relative to control T_FH ( Figure 4A ). In addition, we found that CLL T_FH express high levels of TIGIT ( Figure 4A ), an inhibitory immunoreceptor previously thought to function in T:B cell interactions (25, 26). We examined the ability of CLL T_FH to produce the canonical Type 1 cytokine IFNγ and the T_FH-associated cytokine IL-21 ( Figure 4B ). Remarkably, we observed a substantial increase in both the frequency IL-21 producing and IL-21/IFNγ double-producing cells in CLL patients, with the CLL T_FH population containing significantly more IL-21 and double-producing cells than control T_FH cells ( Figure 4B ). T_FH1 cells produced significantly more IL-21 and IFNγ than other T_FH subsets, but interestingly produced slightly less CD40L ( Figure S4 ). We further assessed whether levels of CD40L, TIGIT or IL-21 expression by T_FH are associated with disease burden or stage. Interestingly, while CD40L and TIGIT expression did not show strong associations, IL-21 expression by T_FH was significantly associated with lymphocyte count and Rai stage ( Figure 4C ). These results indicate that CLL T_FH cells produce abnormally high levels of costimulatory molecules and cytokines known to stimulate CLL survival and proliferation and the expression of IL-21 by these cells is associated with adverse biomarkers and disease burden.

CLL cells are present at varying levels in the bone marrow and lymph nodes, and signals present in these microenvironments are thought to drive CLL proliferation. To investigate whether circulating T_FH cells may represent the counterpart of T_FH populations present within lymphoid tissues, paired blood and bone marrow samples from CLL patients were analyzed. We found a significant correlation between blood and marrow T_FH and T_FH1 cell frequencies from the same patients ( Figure 5A ). Interestingly, T_FH populations in bone marrow showed increased activation status relative to those in peripheral blood of the same patients, expressing significantly more IL-21, IFNγ and IL-21/IFNγ double-producing cells ( Figure 5B ). While levels of PD-1 was also elevated, CD40L and TIGIT were similar in bone marrow and peripheral blood T_FH ( Figure 5C ). We further examined lymph node tissue sections from CLL patients using immunofluorescent staining and could identify T cells present within proliferation centers expressing the T_FH1 markers CD3, CD4, CXCR5, CXCR3 and PD-1 ( Figure 6 ). These results suggest that T_FH cells are present in both circulation and within lymphoid tissues, with the latter showing similar skewing to T_FH1 and a highly activated phenotype.

We developed an in vitro system to study CLL B cell interaction with autologous CD4+ T cells. CD4 T cells were isolated from patient PBMC, pre-cultured overnight with a T cell-activation cocktail or medium alone, then washed and co-cultured with B cells isolated from the same patient that were labeled with cell division tracking dye CFSE. Resulting CLL-B cell division, and expression of activation markers by B and T cells, were assessed after 2 and 6 days of co-culture. At day 2, a significant increase in CLL B cell expression of CD69 and nuclear proliferation antigen Ki-67 was observed in the presence of activated CD4 T cells, compared to co-cultures with non-activated T cells or B cells alone ( Figure 7A ). Activated CD4 T:B cell co-culture also promoted an increase in CD25 activation marker expression among B cells and increased CLL B cell survival at this time point, whereas increased CD38 expression or cell division (as assessed by CFSE dilution) were not observed at day 2 ( Figure S5A ). After 6 days of co-culture, CLL-B cell division was observed ( Figure 7B ) and most divided cells also expressed Ki-67 ( Figure S5B ). At day 6, CLL B cells also exhibited significant upregulation of the activation marker CD25 ( Figure 7B ) as well as CD38 and CD69 ( Figure S5C ) when co-cultured with activated CD4 T cells. Notably, there was a positive correlation between patient T_FH frequency and the frequency of divided, Ki-67+ and CD25+ CLL B cells observed in co-cultures ( Figure 7C ), consistent with a potential role for T_FH cells in driving CLL-B cell activation and proliferation.

During CLL:T cell interactions, CLL-B cells can also impact CD4+ T cell activation (27), but it is unclear whether particular cell subsets are preferentially targeted. In order to assess CLL B cell-dependent activation of CD4+ T cell subsets in co-culture, CD4 T cells untreated with activation cocktail were assessed for their expression of various activation markers in the presence or absence of autologous CLL cells. It was found that the presence of CLL B cells led to an increase in expression of activation markers in a small fraction (less than 15%) of autologous CD4 T cells ( Figure 8A ). This upregulation was significant for CD25, CD69 and PD-1 within 2 days of co-culture ( Figure 8A ). The frequency of CD4 T cells co-expressing both CD25 and OX40, associated with antigen-specific T cell recall responses (28), was also increased in the presence of the B cells at both timepoints ( Figure 8A ). To determine whether T_FH cells were preferentially activated by CLL-B cells, we assessed activation marker expression on T_FH and non-T_FH fractions. The presence of B cells provoked an increase in CD69+ cells in both populations, however the fold increase in expression was significantly higher for T_FH ( Figure 8B ). In addition to CD69, T_FH cells showed greater fold increases in expression of CD25, CD25/OX40 and PD-1 than non-T_FH cells when co-cultured with CLL ( Figure 8B ). T_FH cell expression of activation markers CD25 and OX40 increased between day 2 and day 6 of culture, while no further increase in CD69 or PD-1 were observed after day 2 ( Figure 8C ). We noted that while the patient-specific frequencies of T_FH cells were very stable over the first two days of culture, they began to decline by day 6 in cultures without B cells but were well maintained in the presence of CLL B cells ( Figure 8D ). Taken together, these results indicate that in this co-culture system, CLL-B cells can promote T_FH cell activation and maintenance.