The PI3Kδ-Selective Inhibitor Idelalisib Minimally Interferes with Immune Effector Function Mediated by Rituximab or Obinutuzumab and Significantly Augments B Cell Depletion In Vivo

Idelalisib is a highly selective oral inhibitor of PI3Kδ indicated for the treatment of patients with relapsed chronic lymphocytic leukemia in combination with rituximab. Despite additive clinical effects, previous studies have paradoxically demonstrated that targeted therapies potentially negatively affect anti-CD20 mAb effector mechanisms. To address these potential effects, we investigated the impact of PI3Kδ inhibition by idelalisib on the effector mechanisms of rituximab and obinutuzumab. At clinically relevant concentrations, idelalisib minimally influenced rituximab- and obinutuzumab-mediated Ab-dependent cellular cytotoxicity and phagocytosis on human lymphoma cell lines, while maintaining the superiority of obinutuzumab-mediated Ab-dependent cellular cytotoxicity. Consistent with this, idelalisib did not influence obinutuzumab-mediated B cell depletion in whole-blood B cell–depletion assays. Further, idelalisib significantly enhanced obinutuzumab-mediated direct cell death of chronic lymphocytic leukemia cells. In murine systems, in vivo inhibition of PI3Kδ minimally interfered with maximal rituximab- or obinutuzumab-mediated depletion of leukemic targets. In addition, the duration of rituximab- and obinutuzumab-mediated depletion of leukemia cells was extended by combination with PI3Kδ inhibition. Collectively, these data demonstrate that PI3Kδ inhibition does not significantly affect the effector mechanisms induced by rituximab or obinutuzumab and provides an effective in vivo therapeutic combination. Therefore, combinations of obinutuzumab and idelalisib are currently being assessed in clinical studies.

affect binding properties and effector functions that control the therapeutic effect of anti-CD20 mAbs (6,7). In contrast, type II anti-CD20 mAbs (such as obinutuzumab) do not induce significant CD20 redistribution and, as such, impart enhanced therapeutic effects, including direct killing of cellular targets by homotypic adhesion (7)(8)(9). In addition to its type II properties, obinutuzumab is glycoengineered and consequently offers enhanced affinity for FcgRIII and increased Ab-dependent cellular cytotoxicity (ADCC) and Ab-dependent cellular phagocytosis (ADCP) in comparison with rituximab (10,11).
Obinutuzumab has been approved for first-line treatment of CLL patients in combination with chlorambucil in the United States and Europe and for first-line treatment of FL in Europe, based on headto-head trials comparing obinutuzumab regimens with the respective rituximab regimen using a flat dose of 1000 mg for obinutuzumab and 375 mg/m 2 for rituximab, as well as for the treatment of rituximab-refractory FL patients (12)(13)(14)(15). In first-line diffuse large B cell lymphoma, obinutuzumab did not show superior outcomes (16,17).
Because anti-CD20 mAbs are the standard of care, it is important to understand whether new targeted agents affect their function. Previous work has shown that the covalent Bruton's tyrosine kinase inhibitor, ibrutinib, can interfere with immune effector function and, ultimately, with in vivo efficacy of rituximab in preclinical models (18). Because PI3K isoforms also play a role in immune effector cells and FcgR signaling (19), we investigated the effect of PI3Kd inhibition by idelalisib on the immune effector functions of rituximab and obinutuzumab and the efficacy of in vivo anti-CD20 mAb therapy in a murine model of CLL.

Reagents and chemicals
Idelalisib was synthesized at Gilead Sciences, dissolved in DMSO at 10 mM, and stored at 220˚C. Rituximab and obinutuzumab were provided by Hoffmann-La Roche (Basel, Switzerland). Palivizumab was used as a negative control and was produced at Gilead Sciences.
Cell culture WIL2-S cells were obtained from the American Type Culture Collection (Manassas, VA) and maintained in IMDM supplemented with 10% ultralow Ig FBS and 1% penicillin-streptomycin (all from Life Technologies [Thermo Fisher Scientific], Grand Island, NY).
For macrophage polarization, frozen CD14 + monocytes enriched by negative selection were thawed and cultured in T75 tissue flasks in AIM-V medium (Life Technologies) with 60 ng/ml M-CSF (PeproTech, Rocky Hill, NJ). On day 7, monocyte-derived macrophages (MDMs) were washed and plated in AIM-V with polarizing cytokines. For differentiation to M1 macrophages, cells were plated for 24 h in 100 ng/ml IFN-g (R&D Systems, Minneapolis, MN) and 100 ng/ml LPS (derived from Escherichia coli strain 055:B5; Sigma-Aldrich); for differentiation to M2c macrophages, cells were plated for 48 h in 10 ng/ml IL-10 (R&D Systems).

ADCC assay with PBMC effectors
PBMCs were prepared by Histopaque (Sigma-Aldrich) density centrifugation of fresh blood obtained from healthy human donors. WIL2-S target cells (2.5 3 10 4 cells per well) were incubated with isolated human PBMCs (6.25 3 10 5 cells per well), as well as with titrations of obinutuzumab or rituximab (0.01-1000 ng/ml) in the presence or absence of 256 nM idelalisib in AIM-V medium for 4 h in a humidified incubator at 37˚C, 5% CO 2 . Ten microliters per well of a 1:10 predilution of anti-CD107a-PE Ab (catalog number 328608; BioLegend, San Diego, CA) was added into the culture to determine NK cell degranulation. Tumor cell lysis was assessed after 4 h of incubation at 37˚C, 5% CO 2 by quantification of lactate dehydrogenase (LDH) released into cell culture supernatants (LDH detection kit, catalog number 11644793001; Roche Applied Science, Indianapolis, IN). Maximal lysis of the target cells (= 100%) was achieved by incubation of target cells with 2% Triton X-100. Spontaneous release (= 0%) refers to target cells coincubated with effector cells without Abs. ADCC was calculated using the following formula:

ADCC assay with CLL targets
Frozen CLL PBMCs were thawed and washed in medium before resuspending in PBS (Severn Biotech) at 1 3 10 7 cells per milliliter. Ten microliters of Calcein, AM (1 mg/ml; Life Technologies) was added per milliliter, and cells were incubated at 37˚C in 5% CO 2 for 30 min with periodic mixing. Cells were washed three times (300 3 g, 5 min) with PBS (10% FCS) and resuspended in medium at 9 3 10 5 cells per milliliter. Effector cells (PBMCs) from healthy donors were isolated from anonymized leukocyte cones (University of Southampton Tissue Bank, Southampton, U.K.) using Lymphoprep (AXIS-SHIELD) density gradient centrifugation and resuspended in medium at 20 3 10 6 cells per milliliter. Target and effector cells were preincubated separately without treatment or with vehicle (DMSO) or with idelalisib (1 mM-1 nM) for 40 min at 37˚C in 5% CO 2 . A total of 90 ml of target cells was transferred to a roundbottom 96-well plate (Thermo Fisher Scientific), and Ab was added to a final concentration of 10 mg/ml. Cells were incubated for an additional 20 min, after which 100 ml of effector cells with matching pretreatment was added (E:T ratio, 25:1), and the plate was briefly pulse centrifuged and incubated for 4 h at 37˚C, 5% CO 2 . After incubation, the plate was centrifuged at 200 3 g for 5 min, and 90 ml of supernatant was transferred to a white-walled 96-well plate. Calcein release was measured using a Varioskan plate reader (excitation 494 nM, emission 515 nM; Thermo Fisher Scientific), and lysis was calculated as a percentage of maximum lysis (4% Triton X-100) after subtraction of background (target cells plus effector cells without Ab).

ADCP assay
Idelalisib was added to plated macrophages in AIM-V media and incubated at 37˚C for 1 h, followed by the addition of rituximab or obinutuzumab. WIL2-S target cells were labeled with CellTracker Red (CTR; Molecular Probes; Thermo Fisher Scientific), as per the manufacturer's protocol, and combined with the macrophages at an E:T ratio of 3:1. The cocultures were incubated for 2 h at 37˚C. Cells were then stained with

B cell depletion in human whole blood
Normal B cell depletion was also assessed using fresh heparinized human blood from healthy volunteers. Briefly, fresh blood was collected in heparin-containing syringes. Blood aliquots (180 ml per well) were placed into 96-well deep-well plates, supplemented with 10 mg/ml obinutuzumab or rituximab (10 ml per well) in the absence or presence of idelalisib, and incubated for 1 d at 37˚C in 5% CO 2 in a humidified cell incubator. After incubation, blood was mixed by pipetting up and down before blood aliquots (35 ml per well) were transferred into 96-well round-bottom plates and incubated with fluorescent anti-CD45 (anti-human CD45 allophycocyanin, catalog number 304037), anti-CD19 (anti-human CD19 PE, catalog number 302208), and anti-CD3 (anti-human CD3 PE-Cy7, catalog number 300420; all from BioLegend). After a 15-min incubation at room temperature, BD FACS Lysing Solution (BD Biosciences) was added to lyse erythrocytes and to fix cells prior to flow cytometry. B cell depletion was evaluated using samples not treated with Ab as a 100% control and the following formula:  Enriched NK cells were used as effectors with the WIL2-S cell line as target over various E:T ratios. Anti-CD20 mAb potency was assessed at a saturating concentration (10 mg/ml) in the presence or absence of idelalisib (5-500 nM) by LDH release after 4 h (n = 3, in triplicate, mean 6 SD). Palivizumab was used as a negative control Ab. NK cells and WIL2-S cells were incubated without Ab as an Ab-independent cellular cytotoxicity (AICC) control.
1002 100 B=T cell ratio in control 3ð½B=T cell ratio in sample containing AbÞ: The average B cell depletion and SDs of the triplicates of each experiment were calculated.  annexin V-allophycocyanin, washed, fixed with 4% paraformaldehyde, and acquired on a FACSCanto II flow cytometer (BD Biosciences). CD5 + CD19 + cells were identified and gated for analysis. The CD5 + CD19 + population was plotted as annexin V-allophycocyanin versus live/dead. Quadrant gates were drawn: Q1 is annexin V 2 and live/dead + , Q2 is annexin V + and live/ dead + , Q3 is annexin V + and live/dead 2 , and Q4 is double negative. Results are reported as the sum of Q1, Q2, and Q3 as a percentage of the total cells.  (20), and monitored for disease presentation. Terminal animals were sacrificed, and splenocytes were harvested and cryopreserved. Sex-matched C57BL/6 or SCID syndrome mice were inoculated with 1 3 10 7 splenocytes derived from Em-TCL1-Tg or CD20-Tg Em-TCL1-Tg mice, respectively, and monitored for disease presentation by weekly blood sampling and flow cytometry. Following disease presentation, mice were treated with 250 mg of anti-mouse CD20 (clone 18B12 mouse IgG2a), 250 mg of rituximab (human IgG1

Animals
[hIgG1]), or 250 mg of obinutuzumab (hIgG1) by i.p. injection, together with 10 mg/kg GS-9820 by mouth twice a day or an appropriate vehicle control, as described previously (21). Disease progression/therapeutic responses were subsequently monitored by blood sampling and flow cytometry.

Impact of idelalisib on immune effector functions of rituximab and obinutuzumab
The potential impact of idelalisib on the immune effector functions of rituximab and obinutuzumab was first evaluated by measuring ADCC against the B cell lymphoma line WIL2-S in primary PBMC cultures with titrated amounts of Ab, in the presence or absence of a single concentration of idelalisib, representative of the cell culture medium protein-binding-adjusted maximum peak concentration (C max ) of 256 nM. As expected, the magnitude of ADCC mediated by obinutuzumab was greater than that by rituximab, and the addition of idelalisib showed little to no effect on ADCC with either Ab (Fig. 1A). Effector NK cell surface expression analysis by flow cytometry showed a concentration-dependent increase in the frequency of CD107a + NK cells with rituximab and obinutuzumab, with obinutuzumab inducing more profound effects (Fig. 1B). Addition of idelalisib had little effect on the frequency of CD107a + cells or the density of surface CD107a (data not shown) with obinutuzumab or rituximab. When ADCC was evaluated over various E:T ratios using isolated NK cells (effectors) and WIL2-S cells (targets) and at saturating concentrations of rituximab and obinutuzumab (10 mg/ml), idelalisib had no effect on ADCC when tested over a dose range from 5 to 500 nM (Fig. 2). The effect of idelalisib over a range of concentrations (1-1000 nM) was evaluated in ADCC assays using a fixed E:T ratio and saturating amounts of obinutuzumab or rituximab. Idelalisib did not affect rituximab-or obinutuzumab-mediated ADCC in the context of the maximal ADCC activity of each anti-CD20 Ab (Fig. 3A). The impact of the FcgRIIIa genotype was assessed with NK effectors derived from donors with FcgRIIIa 158 phenylalanine homozygous (F/F), phenylalanine/valine (F/V), or valine homozygous (V/V) genotypes. Idelalisib at concentrations up to 1 mM had no significant effect on rituximab-or obinutuzumab-mediated ADCC in FcgRIIIa 158 F/F and F/V genotypes (Fig. 3B). ADCC with the FcgRIIIa 158 V/V effectors was affected slightly by treatment with idelalisib (∼15% inhibition) at the highest concentration tested.

Impact of idelalisib on ADCC with primary CLL specimens
Next, obinutuzumab-or rituximab-mediated ADCC was assessed in the presence of idelalisib using primary CLL PBMCs as targets. Obinutuzumab and rituximab were effective at inducing ADCC at a saturating concentration of Ab (10 mg/ml), with obinutuzumab inducing 47% ADCC and rituximab inducing 20% ADCC (Fig. 3C, 3D). The amount of ADCC remained largely unchanged for both Abs in combination with idelalisib over a clinically relevant concentration range.

Whole-blood Ab-mediated B cell depletion in the presence of idelalisib
The effects of idelalisib in an autologous B cell-depletion assay using a dose titration of obinutuzumab were assessed in the presence of clinically relevant concentrations of idelalisib (760 and 4200 nM, minimal concentration and C max in whole blood). In both donors, B cells were depleted in a dose-dependent manner in the presence of increasing concentrations of obinutuzumab (Fig. 5A). Although there was a slight trend toward inhibition of B cell depletion in one donor, there were no statistically significant effects with idelalisib across the range of obinutuzumab concentrations. Autologous B cell depletion was also tested in the presence of titrated idelalisib (54-6810 nM), in combination with saturating concentrations of obinutuzumab or rituximab, in three donors. B cell depletion by obinutuzumab was largely unaffected by idelalisib (Fig. 5B). In contrast, B cell depletion by rituximab, in combination with idelalisib, varied across donors, with the most inhibition seen at the highest concentration. In all donors tested, idelalisib alone showed no B cell depletion at any concentration.

Direct induction of cell death in primary CLL specimens
Next, we examined the ability of idelalisib, obinutuzumab, or the combination to induce direct cell death in primary CLL cultures stimulated with anti-IgM/IgG/CD40 to mimic prosurvival signaling within the tissues. Idelalisib (30-480 nM) showed a small, but significant, increase in cell death (8-10%) across all doses, whereas obinutuzumab (0.5-5 mg/ml) showed a dose-dependent induction of cell death of 2-21%, which was most prominent at 5 mg/ml (Fig. 6). However, more striking was the ability of the combination of idelalisib (480 nM) and obinutuzumab (5 mg/ml) to elicit a statistically significant increase in cell death (39%) above either agent alone.

In vivo therapy of Em-TCL1-Tg leukemia-bearing mice
We next assessed the impact of PI3Kd inhibition on the extent and duration of anti-CD20-mediated B cell depletion in a murine model of CLL: the Em-TCL1-Tg mouse (22). Because idelalisib is not suitable for use in a murine setting as a result of unfavorable pharmacokinetic properties (21) (S. Tannheimer, Gilead Sciences, unpublished observations), we used the structurally related surrogate PI3Kd inhibitor, GS-9820. Treatment of leukemia-bearing animals with a suboptimal dose (250 mg per mouse) of anti-mouse CD20 (18B12 mouse IgG2a) or anti-human CD20 mAb (rituximab or obinutuzumab [250 mg of each hIgG 1 per mouse]) imparted a rapid and significant reduction in the leukemic burden of treated animals 48 h posttreatment (Fig. 7A). Consistent with previous studies, initial depletion of target cells with obinutuzumab was greater than that achieved with rituximab (p , 0.005). In this model system, concomitant PI3Kd inhibition (via GS-9820) induced a statistically significant reduction in the extent of obinutuzumab-mediated depletion and a trend toward reduction in the rituximab and 18B12 treatment settings. In a monotherapy setting, GS-9820 provided a modest therapeutic benefit and effectively reduced the extent of leukemia deposits within secondary lymphoid organs (21) (Fig. 7B, 7C, Supplemental  Fig. 1). Therefore, the apparent reduction in anti-CD20-mediated depletion 48 h posttreatment in the presence of PI3Kd inhibition may reflect a redistribution of target cells into the periphery, as is observed in clinical trials (23), rather than inhibition of Ab effector mechanisms. In support of this, concomitant administration of a PI3Kd inhibitor with a single dose of anti-mouse or antihuman CD20 mAb provided a more durable depletion of leukemic cells over the long term and significantly enhanced overall survival in comparison with Ab therapy alone (Fig. 7B, 7C). Although obinutuzumab appeared to outperform rituximab in a short-term depletion setting, long-term superiority of one agent over the other was observed between different Em-TCL1-Tg tumors (Supplemental Fig. 1). This observation is most likely attributable to the suboptimal nature of the Ab-dosing strategy. These data show that, despite the marginal impact of PI3Kd inhibition on both in vitro Ab effector mechanism assays and in vivo depletion in the first 48 h with obinutuzumab, these combinations provide additional benefit to the duration of leukemic depletion in a therapeutic model.

Discussion
The recent regulatory approvals of multiple new targeted therapies, including the anti-CD20 mAb, obinutuzumab, have provided many treatment options for patients with CLL. Multiple small molecule inhibitors that target the BCR pathway have been approved for use in CLL, with idelalisib approved for use with rituximab in relapsed and refractory CLL (4,5). A key step in the clinical integration of these therapies is a clear understanding of how these agents might be combined. Our studies evaluated the effect of idelalisib (or its surrogate GS-9820) at physiologically relevant concentrations on direct killing, ADCC, ADCP, and in vivo therapy in combination with rituximab or obinutuzumab. Rituximab is a type I Ab, whereas obinutuzumab is a type II glycoengineered Ab that has enhanced affinity for FcgRIIIa on effector cells, which leads to enhanced ADCC (11,24). Differences between the two classes of Abs have also been shown for direct cell killing (9,25), complement-dependent cytotoxicity (6), and binding-induced CD20 internalization (26), but they seem to have comparable effects on ADCP (10,27,28).
Our results using obinutuzumab and rituximab in ADCC assays corroborate previous studies showing that obinutuzumab results in better outcomes, regardless of the CD20 levels of the target, Ab concentration, or E:T ratio used in the assay (10). Our studies showed that idelalisib had no effect on anti-CD20 ADCC with obinutuzumab or rituximab, in contrast to other studies that used higher concentrations than those in the clinic (29,30). Previous reports have shown that the F/V polymorphism of FcgRIIIa can impact NK cell binding to Ab (31), as well as progression-free survival, when evaluated in the context of rituximab treatment in genotyped patients (32).
When the impact of idelalisib was assessed in vitro on NK cells with known FcgRIIIa polymorphisms, it showed no inhibition of ADCC, regardless of genotype. Our findings also demonstrated that F/F and F/V polymorphisms were no less effective in ADCC cell assays with obinutuzumab. Characterization of NK cells from PI3Kd kinase-dead or -knockout mice shows altered NK cell migration, extravasation, receptor activation, and cytokine production, although alterations in cytotoxicity toward tumor cells in vitro or in vivo was not observed (33)(34)(35). In our in vitro studies investigating the effects of idelalisib on NK cell-mediated ADCC, we saw a lack of inhibition of NK cell function, as well as observed that CD20 levels on target cells are not altered by PI3Kd inhibition (data not shown). Additionally, unlike ibrutinib, which has been shown to suppress ADCC and ADCP function due to strong inhibition of IL-2-inducible tyrosine kinase, idelalisib has no activity on Bruton's tyrosine kinase or IL-2-inducible tyrosine kinase (16,36,37).
MDMs, polarized to M2c or M1, showed robust phagocytosis of target cells, with M2c demonstrating superior activity. Idelalisib had no effect on macrophage ADCP function on either subset up to the protein-adjusted C max ∼ 300 nM. The robust phagocytosis of myeloid cells in this in vitro assay suggests that idelalisib treatment has little to no impact on macrophage function. More significantly, there is no effect of idelalisib on anti-CD20-mediated B cell depletion from whole blood in healthy donors, an assay that may best model the clinical experience as a result of the integration of multiple mechanisms for depletion (i.e., ADCC, complement-dependent cytotoxicity, induction of cell death). In agreement with these findings, administration of a surrogate PI3Kd inhibitor (GS-9820) alongside suboptimal doses of antimouse CD20, rituximab, or obinutuzumab only minimally influenced maximal depletion of leukemic targets in a murine model of CLL in vivo. These observations indicate that key Ab effector mechanisms are not significantly affected in vivo by PI3Kd inhibition, and any apparent reduction in anti-CD20-mediated depletion in the presence of PI3Kd inhibition may reflect a redistribution of target cells into the periphery, as is observed in clinical trials (23). More significantly, in this model, administration of PI3Kd inhibition alongside anti-CD20 mAbs imparted a more durable depletion of leukemic targets and enhanced therapeutic effects. These observations are reminiscent of clinical trial results demonstrating the therapeutic enhancement of rituximab therapy offered by coadministration of idelalisib in relapsed CLL patients (23).
When the combination of idelalisib and obinutuzumab was evaluated in terms of cell death induction of primary donor CLL cells, a strong additive antitumor activity was confirmed. This unexpected enhanced activity, in conjunction with the lack of idelalisib inhibition of obinutuzumab-mediated ADCC and ADCP activity, suggests that a glycoengineered type II anti-CD20 Ab may function as effectively as a type I Ab in combination with idelalisib.
In conclusion, at clinically relevant doses, idelalisib does not have significant effects on rituximab-or obinutuzumab-driven ADCC, ADCP, or B cell-depletion activity. Additionally, the combination of obinutuzumab and idelalisib appears to have additive activities on direct killing of CLL cells and control of leukemic burden in vivo compared with either agent alone. These studies provide preclinical rationale that the addition of idelalisib to type II Abs, such as obinutuzumab, is likely to be effective clinically without negatively impacting any Ab-mediated immune functions. Based on these findings, obinutuzumab is being tested in combination with idelalisib in clinical trials (NCT02962401, NCT02968563, NCT02445131). Anti-CD20-mediated depletion of leukemic targets in vivo. (A) Animals bearing Em-TCL1-Tg (left panel) or human CD20-Tg Em-TCL1-Tg leukemia (right panel) were treated with 250 mg of anti-CD20 mAbs (anti-mouse CD20 clone 18B12 mouse IgG2a, rituximab hIgG1 [RTX], or obinutuzumab hIgG1 [OBZ]) alongside 10 mg/kg GS-9820 administered by mouth twice a day or an appropriate vehicle control and were monitored for peripheral leukemia levels 48 h later by flow cytometry. Data are normalized to pretreatment values. (B) Em-TCL1-Tg tumor (Em-TCL1 Tg 020)-bearing animals from (A) were maintained on PI3Kd inhibition treatment or an appropriate vehicle control for the duration of the experiment and were monitored for leukemia levels by weekly blood sampling and flow cytometry. A representative example of the peripheral leukemic fraction 4 wk posttreatment, expressed as the percentage of total lymphocytes is shown (middle panel). (C) Human CD20-Tg Em-TCL1-Tg tumor (human CD20-Tg Em-TCL1-Tg FU2)-bearing animals from (A) were maintained on PI3Kd inhibition treatment or an appropriate vehicle control for the duration of the experiment and were monitored for leukemia levels by weekly blood sampling and flow cytometry. Data are represented as the percentage of CD5 + B220 + cells present in the lymphocyte gate. Error bars represent SEM. Statistical analyses were performed using a paired/unpaired Student t test (A) and two-way ANOVA or logrank survival analysis (B and C). *p , 0.05, **p , 0.005, ***p , 0.0005, ****p , 0.000005. GS, GS-9820; n.s, not significant.