Effect of Ibrutinib on the IFN Response of Chronic Lymphocytic Leukemia Cells

The Bruton’s tyrosine kinase (BTK) inhibitor ibrutinib has been a tremendous advance in the treatment of chronic lymphocytic leukemia (CLL) (1). Unfortunately, it does not cure and selects eventually for resistant disease (2). A deeper understanding of the mechanisms that allow CLL cells to persist when BTK-signaling is blocked may suggest new strategies to deepen responses to ibrutinib and improve patient outcomes.

An aberrant cytokine and chemokine network is associated with symptomatic CLL (3, 4). These cytokines include IL-4, IL-6, type I and II IFNs, and others that activate JAKs and promote the growth and survival of CLL cells (3–10). Ibrutinib normalizes the levels of many of these cytokines (11), which may contribute to its therapeutic activity. However, our recent clinical trial suggested JAK signaling remains active in the presence of ibrutinib (3) and may help support the persistence of leukemia cells (5, 6). Specific cytokines that allow CLL cells to survive in the presence of ibrutinib have not yet been identified but might serve as targets to increase the therapeutic efficacy of BTK inhibition.

Type II IFN or IFN-γ is a strong candidate for a cytokine that is active in the presence of ibrutinib. Ibrutinib is known to promote type 1 immunity characterized by IFN-γ production (12), and IFN-γ is a growth and survival factor for CLL cells (7, 8). IFN-γ is made predominantly by T cells and NK cells induced by IL-12 and IL-18. The IFN-γ receptor (IFNGR) is composed of two ligand binding chains (IFNGR1 and IFNGR2) that transmit signals through JAK1 and JAK2 and phosphorylate mainly STAT1 (13). There are hundreds of IFN-γ–regulated genes that include inflammatory mediators along with apoptosis and cell-cycle regulators.

Type I IFNs also promote the growth and survival of CLL cells (9, 10) and could help maintain CLL cells in the presence of ibrutinib. A type III IFN family comprised of IFN-λ1–3 resembles type I IFN in terms of signaling and functional properties but is generally confined to mucosal surfaces (14). Type I IFNs include IFN-β and 13 IFN-α subtypes (15). Almost all cells can make type I IFN with highest amounts from plasmacytoid dendritic cells (16). The IFN-α receptor (IFNAR) is composed of IFNAR1 and IFNAR2 chains associated with JAK1 and TYK2 that phosphorylate STAT proteins, including STAT1, STAT2, and STAT3 (17, 18). pSTAT2 translocates to the nucleus complexed with STAT1 and IRF9 to mediate transcription of IFN-stimulated genes (ISGs), many of which are also induced by IFN-γ (19). ISGs such as the chemokine CXCL10 stimulate immune responses (17), whereas others include immune checkpoint molecules such as LAG3 and death molecules like FAS that can mediate immunosuppression (20, 21). Consequently, type I IFN may have tumor-promoting or tumor-suppressing effects, depending on the magnitude and duration of the response. Concomitant activation of NF-κB also modifies IFN signaling outcomes (18).

The studies in this paper were carried out to determine if there is evidence for IFN activity in patients on ibrutinib, how ibrutinib affects responses of CLL cells to IFNs, and the possibility that IFNs play a role in allowing CLL cells to survive in the presence of ibrutinib.

Fluorescent CD83, LAG3 (CD223), ICAM-1 (CD54), and FAS (CD95) Abs and DAPI were obtained from Pharmingen (San Francisco, CA). Phorbol dibutyrate (PDB), 2-ME, and β-actin Abs were from Sigma-Aldrich (St. Louis, MO). Stock solutions (5 mg/ml) of PDB were made in DMSO. Ibrutinib and ruxolitinib were from SelleckChem (Houston, TX). Human IFN-β-1b (Novartis Pharmaceuticals Canada, Dorval, QC, Canada) was purchased from the Sunnybrook Cancer Centre pharmacy. TNF-α and IFN-γ were from PeproTech (Rocky Hill, NJ). RPMI 1640 was from Invitrogen (Carlsbad, CA), FBS from Wisent Bioproducts (St. Bruno, QC, Canada), and AIM V serum-free media from Thermo Fisher Scientific (Mississauga, ON, Canada). Abs recognizing phospho–(Y705) STAT3 (catalog no. 9134), phospho–(Y690) STAT2 (catalog no. 88410), phospho–(Y701) STAT1 (catalog no. 9171), phospho–(Thr³⁸⁹) p70S6 kinase (p70S6K) (catalog no. 9205), and total STAT1, STAT2, STAT3, p70S6K, and Bcl-2 along with secondary HRP–conjugated anti-rabbit and anti-mouse Abs (catalog nos. 7074 and 7076, respectively) were from Cell Signaling Technology (Beverly, MA). Human IFNGR1 blocking Abs (clone GIR208) were from R&D Systems (Minneapolis, MN). The IFNAR Ab anifrolumab52 was provided by AstraZeneca.

For most in vitro experiments, CLL cells were isolated as before (9) by negative selection from the blood of consenting patients attending a specialized CLL clinic at Sunnybrook. The cells were used directly for experiments, and patients had not been treated for CLL for at least 6 mo prior to blood collection. Protocols were approved by the Sunnybrook Research Ethics Board (Project Identification Number 222–2014).

Frozen plasma stored at −80°C from patients on previously described trials of ruxolitinib in symptomatic CLL patients unfit for conventional therapy (22) and of ruxolitinib combined with ibrutinib (3) were used for cytokine measurements. Plasma was also collected from clinic patients treated with ibrutinib for varying times (3–24 mo), and 10 different samples were pooled to study the effects on signaling and survival of CLL cells. Plasma from patients not on ibrutinib was also pooled to study these effects. Patient characteristics are described in Supplemental Table I.

Purified CLL cells (2 × 106 cells/ml) were uniformly cultured in serum-free AIM V medium plus 2-ME (5 × 10−5 M) in 6- or 24-well plates (BD Labware) at 37°C in 5% CO₂ for the times indicated in the figure legends. To accentuate killing by ibrutinib, CLL cells were cultured in RPMI 1640 with 2% FBS (23–25). To study effects on IFN signaling, pooled plasma from patients on ibrutinib or otherwise untreated were used at final concentrations of 30%, whereas anifrolumab and the IFNGR Abs were used at final concentrations of 9.25 and 2 μg/ml, respectively. Ibrutinib, IFN-β, and IFN-γ were used at final concentrations of 500 nM, 300 U/ml, and 10 ng/ml, respectively, and ruxolitinib at a final concentration of 400 nM.

Viability was measured by washing cells in PBS and staining with DAPI (100 μl) for 10 min. LAG3(CD223), CD83, FAS (CD95), and ICAM-1 (CD54) were measured by staining with the corresponding Abs for 15 min at room temperature, followed by washing with PBS. Lymphocytes were gated using forward and side-scatter parameters and viable cells were gated on the basis of excluding DAPI. Unstained control cells were used to set the thresholds for determining positivity of staining. Ten thousand viable counts were then analyzed with an FACScan flow cytometer using Cellquest software (Becton Dickinson). Standardization of the flow cytometer was performed before each experiment using SpheroParticles (Spherotech, Chicago, IL).

IFN-α2, IFN-γ, IL-18, CXCL10, CCL3, and TNF-α were measured as before by Eve Technologies (Calgary, AB, Canada) using Multiplexing LASER Bead Technology (3, 22). Concentrations were determined from standard curves. Assays were linear between 30 and 1000 pg/ml cytokine.

Protein extraction and immunoblotting were performed as before (9). Proteins were resolved in 10% NaDodSO₄/PAGE and transferred to Immobilon-P transfer membranes (MilliporeSigma, Billerica, MA). Western blot analysis was performed according to the manufacturers’ protocols for each Ab. Chemiluminescent signals were created with SupersignalWest Pico Luminal Enhancer and Stable Peroxide Solution (Pierce, Rockford, IL) and detected with a Syngene InGenius system (Syngene, Cambridge, U.K.). For additional signal, blots were stripped for 60 min at 37°C in Restore Western blot stripping buffer (Pierce), washed twice in TBS plus 0.05% Tween 20 at room temperature, and reprobed as required. Densitometry was performed using ImageJ software. Densitometry values for each sample were normalized to the value for β-actin to obtain the intensities for phosphorylated–STAT1-3, p70S6K, and Bcl-2 reported in the figures.

RNA was prepared with the RNeasy Mini Kit (QIAGEN, Valencia, CA) and cDNA synthesized from 2 μg of RNA using Superscript III reverse transcriptase (Life Technologies, Invitrogen), according to the manufacturer’s instructions. LAG3, CXCL10, and hypoxanthine-guanine phosphoribosyl transferase (HPRT) transcripts were amplified with the following primers: CXCL10, forward 5′-CCAGAATCGAAGGCCATCAA-3′ and reverse 5′-CATTTCCTTGCTAACTGCTTTCAG-3′; LAG3, forward 5′-TCACAGTGACTCCCAAATCCT-3′ and reverse 5′-GCTCCACACAAAGCGTTCTT-3′; and HPRT, forward 5′-GAGGATTTGGAAAGGGTGTT-3′ and reverse 5′-ACAATAGCTCTTCAGTCTGA-3′.

PCRs were carried out in a DNA engine Opticon System (MJ Research, Waltham, MA) and cycled 34 times after initial denaturation (95°C, 15 min) with the following parameters: denaturation at 94°C for 20 s, annealing of primers at 58°C for 20 s, and extension at 72°C for 20 s. Transcript abundance was evaluated by a standard amplification curve relating initial copy number to cycle number. Copy numbers were determined from two independent cDNA preparations for each sample. The final result was expressed as the relative fold change of the target gene to HPRT.

Comparisons between two groups of measurements were tested for significance by Student or paired t tests with a p value <0.05 considered significant. ANOVA with multiple comparisons was conducted to determine the significance of differences between multiple groups.

In a prior phase 1 trial, the JAK inhibitor ruxolitinib was added to 12 patients on ibrutinib for 3 out of 5 wk and repeated seven times in an attempt to deepen clinical responses (3). Ruxolitinib blocks signaling through IFNAR and IFNGR along with many other cytokine receptors (3, 7, 26). Plasma levels of CXCL10, considered a reporter gene for both type I and II IFN responses (19), and IL-18, which is regulated by type I IFN (27), were measured at the beginning of cycle 1 and after ruxolitinib had been used for 3 wk (Fig. 1A) (3). Following a 2-wk break from ruxolitinib, during which ibrutinib was continued, the measurements were repeated for cycles 2 and 3. Results for each patient are shown in Supplemental Fig. 1. Average relative measurements (Fig. 1A) indicate both CXCL10 and IL-18 decreased when patients were treated with ruxolitinib and increased after 2 wk on ibrutinib alone. Similar cycling was reported previously for β-2-microglobulin (β2M), another ISG (3). Although the sources of CXCL10 and IL-18 are unclear and may involve a number of cell types, the findings suggested IFN signaling was active in patients on ibrutinib and could be modulated by ruxolitinib.

Type I and II IFN are elevated in the blood of CLL patients (4, 11) but reported to be lowered by ibrutinib (11). Consistent with these observations, IFN-γ and IFN-α levels in 16 CLL patients on ibrutinib for at least 9 mo (3) were significantly lower than in 15 other symptomatic patients (22) (Fig. 1B) but remained detectable and possibly functional.

To determine if IFN in the blood of patients on ibrutinib was enough to induce a signaling response, CLL cells from otherwise untreated patients were cultured in 30% plasma pooled from 10 patients who had been treated with ibrutinib as a single agent for 3–24 mo (Fig. 1C). After 30 min, phospho–(Y690) STAT1 levels were measured to indicate IFN signaling. Levels of phospho-STAT1 increased and were reduced by IFNAR and IFNGR blocking Abs, suggesting the presence of active type I and type II IFN in the plasma.

Ibrutinib is considered primarily an inhibitor of BTK without a direct role in IFN signaling. However, the drug has off-target effects, inhibits JAK1/3 at high concentrations (>10 μM) (28) and might also affect JAK/STAT signaling in CLL cells by modulating surface expression of cytokine receptors (29). To reveal potential effects of ibrutinib on type I and II IFN signaling, CLL cells were cultured overnight with ibrutinib at 500 nM to approximate therapeutic plasma concentrations in vivo (30). The cells were then stimulated with IFN-β as a representative type I IFN or IFN-γ and levels of pSTAT1, pSTAT2, and pSTAT3 determined over the next 8 h by immunoblotting and densitometry (Fig. 1D). Examples of immunoblots are shown for individual patients along with average values and SEs for time courses of three different patient samples (Fig. 1E). Overall, signaling by type I and II IFN in CLL cells was unchanged by exposure to ibrutinib in vitro.

IFNs can activate additional pathways than JAK/STAT signaling (18) that could be affected by ibrutinib-mediated changes in receptor expression. However, IFNAR1 and IFNGR1 levels in CLL cells treated with ibrutinib for 24 h did not change significantly (Fig. 2A). IFNAR1 and IFNAGR1 levels also did not change significantly in CLL cells from five patients on ibrutinib for at least 3 mo compared with cells obtained before starting treatment (Fig. 2B). IFN-β (Fig. 2C, left panel) and IFN-γ (Fig. 2C, right panel) induced phosphorylation of STAT1 in CLL cells from patients on ibrutinib with a time course similar to CLL cells from patients not on treatment (Fig. 1D). Taken together, the observations in Figs. 1 and 2 suggest that CLL cells that persist in patients on ibrutinib encounter IFN in vivo and are able to respond to it.

Cellular responses to IFN are mediated by transcriptional and translational changes in ISGs. The interferome of CLL cells has not been mapped precisely and may exhibit patient heterogeneity. Candidate CLL ISGs include CCL3, CXCL10, TNF⍺, FAS, and LAG3. These genes are ISGs in many cell types (31–35), and their gene products increased when CLL cells were stimulated with IFN-β (Fig. 3). Despite intact IFN-β-signaling (Fig. 1), expression of these proteins was decreased markedly when CLL cells were treated with IFN-β in the presence of ibrutinib (Fig. 3). Responses to IFN-γ were also inhibited by ibrutinib based on diminished expression of LAG3 and CD54, considered a reporter of IFN-γ (36), compared with controls (Fig. 3D, 3E).

To determine how ibrutinib lowered protein production, CXLC10 gene expression was measured by real-time PCR at 4 and 24 h after stimulating CLL cells with IFN-β (300 U/ml) in the presence and absence of ibrutinib (500 nM) (Fig. 4A). CXCL10 mRNA normally increased after 4 h of IFN stimulation and returned near baseline by 24 h (Fig. 4A). Despite intact canonical IFN signaling (Fig. 1), CXCL10 transcripts were reduced significantly by ibrutinib at 4 h (Fig. 4A).

Like many ISGs, the CXCL10 promoter contains NF-κB response elements, and constitutive NF-κB activity has been shown to be involved in IFN-induced expression of CXCL10 (37). Although ibrutinib does not apparently affect IFN signaling directly, it does inhibit intrinsic BCR and TLR signals that activate NF-κB in CLL cells (38, 39). Accordingly, ibrutinib could negatively affect CXCL10 transcription indirectly by lowering NF-κB activity.

CD83 can be used to report NF-κB activity, as it contains mainly NF-κB–binding sites in its promoter (40, 41). CD83 expression on cultured CLL cells was lowered significantly by ibrutinib (Supplemental Fig. 2A). Consistent with the idea that ibrutinib negatively affects ISGs such as CXCL10 by lowering NF-κB tone, BTK-independent activation of NF-κB, achieved by adding the TLR7-agonist resiquimod along with TNF-α and indicated by restoration of CD83 expression (Supplemental Fig. 2B), increased type I IFN–mediated CXCL10 gene expression in the presence of ibrutinib (Fig. 4B).

The temporal pattern and response to ibrutinib of IFN-mediated LAG3 mRNA expression was different from CXCL10. Peak levels at 4 h were lower and did not decline as rapidly (Fig. 4C). Ibrutinib did not change the pattern of LAG3 mRNA expression significantly (Fig. 4C), suggesting inhibited translation might account for the decrease in protein levels (Fig. 3B).

Ibrutinib has been shown to decrease mRNA translation of some genes (42, 43), and translation is positively regulated by p70S6K (43). Consistent with stimulation of ribosomal activity and induction of protein synthesis, IFN-β caused phosphorylation of p70S6K (PS6K) after 24 h, which was decreased by ibrutinib (Fig. 4D, lanes 1–4; Fig. 4E, first four bars; Supplemental Fig. 3A).

If translational inhibition by ibrutinib was responsible for decreased LAG3 induction by IFN, then BTK-independent activation of translation should restore LAG3 expression. Phorbol esters such as PDB activate protein kinase C in CLL cells (44) and caused p70S6K to be phosphorylated in the presence of ibrutinib (Fig. 4D, lanes 5–7; Fig. 4E, right three bars; Supplemental Fig. 3A). PDB did not affect baseline LAG3 expression. Consistent with restored translation, IFN-induced LAG3 expression did not change significantly in the presence of ibrutinib compared with PDB-activated CLL cells alone (Supplemental Fig. 3B, 3C; Fig. 4F, compare columns 3 and 4 with 6 and 7).

Both type I and II IFN increase survival of CLL cells in the face of cytotoxic stressors (7–10). Survival of CLL cells in serum-containing media (25) was increased significantly by IFN-γ and IFN-β, whereas ibrutinib was cytotoxic in these conditions (Fig. 5A). Despite their negative effects on gene expression (Figs. 3, 4), IFN-β and IFN-γ allowed CLL cells to survive in the presence of ibrutinib. These effects were partially but significantly reversed by ruxolitinib, suggesting they were related to IFN signaling (Fig. 5A).

Ibrutinib-induced death was associated with decreased expression of the antiapoptotic protein Bcl-2 (10, 45, 46) (Fig. 5B, compare lanes 1 and 5). Consistent with the ability to increase survival (Fig. 5A), IFN-β with IFN-γ increased Bcl-2 levels significantly in the presence of ibrutinib (Fig. 5B, compare lanes 5 and 6).

The levels of type I and II IFN in plasma of ibrutinib-treated patients that could induce signaling responses (Fig. 1C) were also able to affect survival of CLL cells in vitro (Fig. 5C). Blocking Abs to IFNAR (anifrolumab) and IFNGR1 significantly increased the death of CLL cells in pooled plasma from patients on ibrutinib at a final concentration of 3% (Fig. 5C, top panel). IFNAR Abs appeared to cause more death than IFNGR1 Abs (Fig. 5C, top panel). Interestingly, IFN Abs did not significantly affect the survival of CLL cells in pooled plasma from untreated patients that contain a richer spectrum of growth-promoting factors (4, 11) (Fig. 5C, bottom panel; Fig. 6).

The results in this paper provide evidence for ongoing but altered IFN signaling in CLL patients on ibrutinib. CLL cells are able to transmit signals from IFNAR and IFNGR in the presence of ibrutinib (Figs. 1, 2), but ISG expression is markedly inhibited (Fig. 3) in several ways. Genes like CXLC10 are inhibited by ibrutinib at the transcriptional level (Fig. 4A, 4B), but others like LAG3 are affected at the level of translation (Fig. 4C–E). Some functions of IFN, such as regulation of cell survival, remain intact in the presence of ibrutinib (Fig. 5A, 5B). Type I and type II IFN levels in the blood of CLL patients are diminished by ibrutinib but still able to signal (Fig. 1) and support survival of CLL cells (Fig. 5C).

The cellular origins of type I and II IFN in patients on ibrutinib are unclear. Major sources of type II IFN are T and NK cells, including large granular lymphocytes. The kinase inhibitor dasatinib causes clonal expansion of large granular lymphocytes in chronic myelogenous leukemia (47), but it is not clear if ibrutinib does the same in CLL. Plasmacytoid dendritic cells are considered the major producers of type I IFN but are reduced in CLL patients (48) and other cells, including leukemia cells themselves (55), may be the source in ibrutinib-treated patients. A schema of potential sources of IFN and IFN signaling responses of CLL cells in patients on ibrutinib is provided in Fig. 6.

Phosphorylation of STAT2 is generally considered to reflect only type I or III IFN signaling (14, 15). IFN-γ apparently phosphorylates STAT2 in CLL cells (Fig. 1), suggesting canonical type II IFN signaling pathways are cell-type dependent. IFN signaling responses in CLL cells also showed patient heterogeneity based on observed variations in IFN-β–mediated induction of chemokines, cytokines, and cell surface proteins (Fig. 3). Patient-specific signaling responses of CLL cells to IFN-α2b have been noted before (9), but there were no obvious relationships between the magnitudes of ISG expression (Fig. 3) and conventional CLL prognostic markers. The finding that ibrutinib decreased production of these gene products in all cases (Fig. 3) suggests BTK-mediated signaling regulates the interferome and may help explain the heterogeneity of IFN responses in CLL cells.

It is unclear if ibrutinib affects IFN responses of normal immune cells in the same manner as CLL cells. Increased numbers of opportunistic infections have been described in CLL patients on ibrutinib (49). Notably, these are mainly fungal infections that require intact Th17 and neutrophil responses and not usually considered highly dependent on IFN (50).

The consequences of ongoing IFN activity in patients on ibrutinib and modulation of IFN signaling responses in CLL cells by ibrutinib are not entirely clear. The effects may be beneficial. For example, by decreasing IFN-mediated LAG3 expression (Fig. 3), ibrutinib might decrease immunosuppression and allow effective antitumor responses against CLL cells (34, 51). Blocking IFN signaling would then diminish the therapeutic efficacy of ibrutinib.

In contrast, the ability of IFN to maintain survival of CLL cells (Fig. 5) might allow CLL cells to persist and ultimately progress in the face of ibrutinib. In that case, blocking IFN signaling should improve the efficacy of ibrutinib. Some insights may be gained from clinical observations of the in vivo effects of ruxolitinib, which blocks signaling from IFNs along with other cytokines. Adding ruxolitinib to patients on ibrutinib modestly improved residual tumor burden in most patients after 7 mo of combination treatment (3). These findings support the idea that persistent IFN signaling in CLL patients on ibrutinib could be detrimental. Perhaps specifically blocking type I and II IFN by combining agents like the IFNAR Ab anifrolumab (52, 53) and IFNGR Ab emapalumab (54) might be an effective way to deepen clinical responses to ibrutinib.

Anifrolumab was provided by AstraZeneca.

The authors have no financial conflicts of interest.