Brain, Behavior, and Immunity xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Brain, Behavior, and Immunity

journal homepage: www.elsevier .com/locate /ybrbi

Short Communication

Non-tumor cell IDO1 predominantly contributes to enzyme activity andresponse to CTLA-4/PD-L1 inhibition in mouse glioblastoma

http://dx.doi.org/10.1016/j.bbi.2017.01.0220889-1591/� 2017 Elsevier Inc. All rights reserved.

⇑ Corresponding author at: 300 E Superior Street-Tarry Bldg 2-703, Chicago, IL60611, USA.

E-mail address: derekwainwright@northwestern.edu (D.A. Wainwright).

Please cite this article in press as: Zhai, L., et al. Non-tumor cell IDO1 predominantly contributes to enzyme activity and response to CTLA-4/PD-Lbition in mouse glioblastoma. Brain Behav. Immun. (2017), http://dx.doi.org/10.1016/j.bbi.2017.01.022

Lijie Zhai a, Erik Ladomersky a, Carlos R. Dostal b, Kristen L. Lauing a, Kathleen Swoap a, Leah K. Billingham a,Galina Gritsina a, Meijing Wu a, Robert H. McCusker c, David C. Binder d, Derek A. Wainwright a,e,f,g,⇑aDepartment of Neurological Surgery, Northwestern University Feinberg School of Medicine, Chicago, IL 60611, USAbNeuroscience Program, The University of Illinois at Urbana-Champaign, Urbana, IL, USAcDepartment of Animal Sciences, The University of Illinois at Urbana-Champaign, Urbana, IL, USAdDepartment of Radiation Oncology, University of Colorado School of Medicine, Aurora, CO 80045, USAeDepartment of Microbiology-Immunology, Northwestern University Feinberg School of Medicine, Chicago, IL 60611, USAfDepartment of Medicine-Hematology/Oncology, Northwestern University Feinberg School of Medicine, Chicago, IL 60611, USAgRobert H. Lurie Comprehensive Cancer Center of Northwestern University, Chicago, IL 60611, USA

a r t i c l e i n f o

Article history:Received 17 October 2016Received in revised form 28 December 2016Accepted 28 January 2017Available online xxxx

Keywords:IDO1GliomaTryptophanKynurenineImmunosuppressionIDO2TDO2

a b s t r a c t

Glioblastoma (GBM) is the most common malignant brain tumor in adults with a median survival of14.6 months. A contributing factor to GBM aggressiveness is the intratumoral expression of the potentlyimmunosuppressive enzyme, indoleamine 2,3 dioxygenase 1 (IDO1). The enzymatic activity of IDO1 isassociated with the conversion of tryptophan into downstream kynurenine (Kyn), which has previouslybeen hypothesized to contribute toward the suppression of tumor immunity. Utilizing the syngeneic,immunocompetent, intracranial GL261 cell GBM model, we previously demonstrated that tumor cell,but not non-tumor cell IDO1, suppresses T cell-mediated brain tumor regression in mice. Paradoxically,we also showed that the survival advantage mediated by immune checkpoint blockade is abrogated bynon-tumor cell IDO1 deficiency. Here, we have built on our past observations and confirm the maladap-tive role of tumor cell IDO1 in a novel mouse GBM model. We also demonstrate that, non-tumor cells,rather than mouse GBM cells, are the dominant contributor to IDO1-mediated enzyme activity. Finally,we show the novel associations between maximally-effective immune-checkpoint blockade-mediatedsurvival, non-tumor cell IDO1 and intra-GBM Kyn levels. These data suggest for the first time that,GBM cell-mediated immunosuppression is IDO1 enzyme independent, while the survival benefits ofimmune checkpoint blockade require non-tumor cell IDO1 enzyme activity. Given that current clinicalinhibitors vary in their mechanism of action, in terms of targeting IDO1 enzyme activity versusenzyme-independent effects, this work suggests that choosing an appropriate IDO1 pharmacologic willmaximize the effectiveness of future immune checkpoint blockade approaches.

� 2017 Elsevier Inc. All rights reserved.

1. Introduction (Reardon et al., 2015; Binder et al., 2015), based on the recent

Glioblastoma (GBM) is a primary malignant brain cancer inadults with a median overall survival of 14.6 months (Stuppet al., 2005). Patients diagnosed with GBM typically undergo con-ventional therapy including surgical resection of the tumor, whenpossible, followed by irradiation and chemotherapy. Even withthese aggressive treatments, the overall prognosis remains grim;highlighting the need for more effective and less toxic approaches.To address this clinical necessity, immunotherapy has been pro-posed as a potential beneficial strategy for GBM patients

demonstration of improved survival in terminal cancer patientsthat are otherwise refractory to treatment (Wolchok et al., 2013;Larkin et al., 2015). However, applying immunotherapy to GBMis challenging given the unique immunologic specialization ofthe central nervous system (CNS) that includes: anatomical barri-ers limiting lymphatic drainage, low MHC expression levels byparenchymal glial cells and immunosuppressive gene expressionthat increases with age (Ladomersky et al., 2016).

Indoleamine 2,3-dioxygense 1 (IDO1) is an interferon-induciblemediator and rate-limiting enzyme associated with the conversionof tryptophan (Trp) into kynurenine (Kyn). This pathway is well-established to play an important role in driving cancer-inducedimmunosuppression and is a high-profile therapeutic target(Munn et al., 2012; Zhai et al., 2015b; Uyttenhove et al., 2003). In

1 inhi-

2 L. Zhai et al. / Brain, Behavior, and Immunity xxx (2017) xxx–xxx

vitro, IDO1-mediated Trp depletion and/or Kyn accumulation, leadsto effector T cell apoptosis (Fallarino et al., 2002) and/or naïve CD4+

T cell conversion into inducible regulatory T cells (Tregs; CD4+-CD25+FoxP3+) (Mezrich et al., 2010). In support of its therapeutictargeting potential, multiple IDO1 enzyme or pathway inhibitorsare being tested in clinical trials including epacadostat (Incyte),GDC-0919 (Genentech), PF-06840003 (Pfizer) and indoximod(D1-MT; New Link Genetics), with additional agents undergoingevaluation in the preclinical pipeline.

Our group previously demonstrated that, GBM cell- but notnon-tumor cell IDO1, facilitates increased Treg accumulation andnegatively impacts overall survival in a syngeneic, immunocompe-tent, intracranial mouse GBM model (Wainwright et al., 2012).Wealso demonstrated the paradoxical observation showing that, max-imal responsiveness to immune checkpoint blockade requires non-tumor cell IDO1 (Wainwright et al., 2014), providing evidence ofnon-overlapping immunomodulatory roles for IDO1 in GBM versusnon-GBM cells. Based on the historical function of IDO1 as a canon-ical Trp catabolic enzyme, we next hypothesized that, IDO1enzyme activity is primarily regulated by GBM cells - accountingfor the tumor cell IDO1-induced immunosuppression that we pre-viously discovered. Here, we unexpectedly found that, non-tumorcell-, rather than GBM cell IDO1, predominantly contributes tothe tryptophan catabolism in mouse GBM. We then confirmedthe relevance of these findings by verifying that GBM-cell IDO1negatively impacts overall survival in a novel orthotopic mouseglioma model. Finally, we confirm and expand upon the surprisingrequirement for non-GBM cell IDO1 to achieve maximal survivalbenefit after treatment with CTLA-4/PD-L1 blockade in mice withexperimental brain tumors.

2. Materials and methods

2.1. Mice

Wild-type (WT) C57BL/6 (B6) mice (Cat# 000664) and systemicIdo1 knockout (Ido1�/�, B6 background; Cat# 005897) mice werepurchased from Jackson Laboratories, maintained in the North-western University Center for Comparative Medicine Facility andintracranially-engrafted between the ages of 6 and 8 weeks.Tamoxifen-inducible transgenic mice that spontaneously developglioma (tGBM; GFAP(ERT2)? Cre+/�; Ptenfl/fl; Rbfl/fl; p53fl/fl] weregenerously provided by Dr. Suzanne J. Baker, PhD (St. Jude Chil-dren’s Research Hospital, Memphis, TN) on a mixed FVB/B6 back-ground and maintained as previously described (Chow et al.,2011). To obtain Ido1 deficient tGBM mice, Ido1�/� mice weremated with the tGBM founder line, followed by backcrossing tothe founder line for 6 generations. Genotyping to confirm the sta-tus of all transgenes (Cre+/�, Ptenfl/fl, Rbfl/fl, p53fl/fl and Ido1�/�) wereperformed in accordance with established protocols. All mousestrains used in these studies were provided autoclaved food pelletsand water ad libitum.

2.2. Primary GBM cell culture, cell proliferation, viability, mAbdelivery, RNA isolation, reverse transcription and quantitative PCR(RT-PCR)

Procedures are described in the Supplementary Materials andMethods section.

2.3. Trp and Kyn analysis by high performance liquid chromatography(HPLC)

Procedures for HPLC sample processing and analysis have previ-ously been described (Zhai et al., 2015a).

Please cite this article in press as: Zhai, L., et al. Non-tumor cell IDO1 predomibition in mouse glioblastoma. Brain Behav. Immun. (2017), http://dx.doi.org/1

2.4. Statistical analysis

For all quantitative data, the normality of residuals and homo-geneity of variances were analyzed with the Shapiro-Wilk testand Fligner-Killeen test, respectively. Two-way ANOVA withTukey’s multiple comparison test were used for analysis whenboth assumptions were met, otherwise non-parametric analysiswere used instead. Tryptophan (ng/mL) levels were log-transformed and then tested for the normality. Survival curveswere generated using the Kaplan-Meier method and comparedby Log-rank test. Differences were considered to be statisticallysignificant when P < 0.05. Means ± SEM are presented throughoutthe manuscript. Statistical analyses was performed using GraphPadPrism5.0 (Graphpad Software Inc, La Jolla, CA) and R version 3.3.2(R Foundation for Statistical Computing, Vienna, Austria.). Package‘‘car” was used for ANOVA analysis.

3. Results

3.1. Non-tumor cell IDO1 predominantly contributes to kynurenine(Kyn) accumulation in mouse GBM

Our previous work demonstrated that IDO1 expression bymouse GL261 GBM cells, but not non-tumor cells, enhances theaccumulation of tumor-infiltrating Treg and decreases the survivalof brain tumor-bearing mice (Wainwright et al., 2012). This led tothe hypothesis that, the majority of IDO1-mediated enzyme activ-ity in brain tumors is attributable to GBM cells. To evaluate thishypothesis, Trp and Kyn levels were assessed in naïve mice thatdid not receive an intracranial injection, or in mice engraftedGL261 cells at 2 weeks post-intracranial injection (2wp-ic.).Universally, wild-type (WT) mice possess a higher Kyn level andKyn/Trp ratio, when compared to Ido1�/� mice in the serum, naïvebrain and brain tumor (Fig. 1A; P < 0.001). In contrast, GBM cellIDO1 neither affects Kyn levels nor the Kyn/Trp ratio in mousebrain tumors. Taken together, these data suggest that non-tumorcell, rather than GBM cell IDO1, primarily contributes to IDO1enzyme activity and Kyn accumulation in mouse brain tumors.

To address the IDO1 expression potential in mouseGBM, unmodified- or GL261-cells transduced with scrambledshRNA (GL261Ido1-WT) or IDO1-targeted shRNA (GL261Ido1-KD)(Wainwright et al., 2012) were cultured with or without mouseIFNc; a potent inducer of IDO1 expression (Fig. 1B) (Carlin et al.,1987). Additionally, novel mouse GBM cell lines isolated from atamoxifen-inducible mouse model of GBM [GFAP(ERT2)Cre+/�;p53fl/fl; pTENfl/fl; Rbfl/fl (Chow et al., 2011)] that was WT(tGBMIdo1-WT) or backcrossed with Ido�/� mice (tGBMIdo1-KO), werealso analyzed for IDO1 expression potential. Mouse IDO1 mRNAwas significantly increased in all WT mouse GBM cell lines whenstimulated with IFNc. Confirming the capability to express IDO1(P < 0.001). In contrast, all glioma lines targeted for IDO1 geneknockdown/deletion showed a P90% decrease in IDO1 mRNAexpression (P < 0.001).

To confirm the negative impact of IDO1 in GBM cells, which wepreviously evaluated in the GL261 brain tumor model (Wainwrightet al., 2012), FVB/B6 mice were intracranially-engrafted noveltGBMIdo1-WT and tGBMIdo1-KO mouse glioma cells (Fig. 1C). Themedian survival of mice with IDO1-WT glioma is 27 days, whichis significantly decreased when compared to mice with IDO1-deficient glioma that possessed a median survival of 46 days(P < 0.01). Importantly, the rate of proliferation (MFI; mean fluo-rescence intensity) (Fig. 1D) and apoptosis (% cell viability) (Fig. 1E)was not different between the tGBMIdo1-WT and tGBMIdo1-KO mouseglioma cells, in vitro. These data confirm the maladaptive nature ofIDO1 expression in an independent, immunocompetent mouse

nantly contributes to enzyme activity and response to CTLA-4/PD-L1 inhi-0.1016/j.bbi.2017.01.022

Fig. 1. The effects of glioblastoma (GBM) cell and non-tumor cell IDO1 on enzyme activity, survival and inflammatory gene expression in a mouse brain tumor model. (A) WT(red bars) or Ido1�/� (blue bars) mice were not injected (untreated and naïve; solid bars) or intracranially-engrafted 2 � 105 GL261 cells transduced with lentiviral particlesencoding scrambled shRNA (GL261Ido1-WT; hatch-filled bars) or -shRNA specific to IDO1 (GL261Ido1-KD; horizontal line-filled bars). Tryptophan (Trp) and kynurenine (Kyn) wasquantified in the serum (left), uninjected normal brain or GBM (middle) and draining cervical lymph nodes (cLN; right) by HPLC in naïve (n = 5/group) or tumor-bearing mice(n = 4–7/group) at 2 weeks post-intracranial injection. (B) RT-PCR analysis of Ido1 mRNA expression in unmodified GL261 (GL261), scrambled-shRNA transduced GL261(GL261Ido1-WT), IDO1-shRNA transduced GL261 (GL261Ido1-KD), as well as tGBMIdo1-WT and tGBMIdo1-KO cells isolated from symptomatic mice with tamoxifen-inducible wild-type [GFAP(ERT2)? Cre+/�; Ptenfl/fl; Rbfl/fl; p53 fl/fl] or IDO1-deficient [GFAP(ERT2)? Cre+/�; Ptenfl/fl; Rbfl/fl; p53 fl/fl; Ido1�/�] tumors and treated without (blue bars) or with100 ng/mL mouse IFNc (red bars) for 48 h. All mRNA levels were normalized to unmodified GL261 cells at 48 h post-stimulation with 100 ng/mL IFNc. Data compiled fromfive independent experiments is shown. For each experiment, all groups were run in triplicate. (C) Kaplan-Meier survival analysis of WT FVB/B6 mice intracranially-injected2 � 105 tGBMIdo1-WT (n = 5) or tGBMIdo1-KO cells (n = 9). One representative dataset from two experiments is shown. In vitro characterization of WT (red bars) and Ido1�/� (bluebars) tGBMIdo1-WT and tGBMIdo1-KO cells for proliferation (D) and viability (E). For viability studies, staurosporine (Staurospor.) was utilized to induce cell death. Cells negativefor the Zombie and Annexin V dyes were gated as living cells. For both proliferation and viability assays, the quantitative data compiled from five independent experiments isshown. For each experiment, all groups were run in triplicate. (F) RT-PCR mRNA analysis of intratumoral gene expression for markers associated with T cells, proinflammatorycytokines and Ido1 in brain tumors isolated fromWT FVB/B6 mice intracranially-injected 2 � 105 tGBMIdo1-WT (n = 8) or tGBMIdo1-KO cells (n = 8) at 3 weeks post-engraftment.Gene expression levels were compared to samples iso lated from mice engrafted tGBMIdo1-WT cells. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

L. Zhai et al. / Brain, Behavior, and Immunity xxx (2017) xxx–xxx 3

glioma model and suggest that the differences in survival are unre-lated to tumor intrinsic cell growth or death properties and rather,likely represent an immune-mediated anti-tumor effect.

To characterize the immune regulation that might account forthe differences in survival between mice engrafted tGBMIdo1-WT

and tGBMIdo1-KO glioma cells, tumors were resected at 3 weekspost-intracranial engraftment and analyzed for mRNA expressionof T cell-associated markers, Cd8a and Cd3e, as well as the proin-flammatory cytokines, Il1b, Ifng and Il6 (Fig. 1F). All immune indi-cators were significantly decreased in the IDO1 WT-, whencompared to the IDO1-deficient-brain tumor (P < 0.05, respec-tively). Notably, the intratumoral IDO1 mRNA expressionlevel was unchanged between mice engrafted tGBMIdo1-WT andtGBMIdo1-KO glioma cells, likely reflecting the compensatory IDO1

Please cite this article in press as: Zhai, L., et al. Non-tumor cell IDO1 predomibition in mouse glioblastoma. Brain Behav. Immun. (2017), http://dx.doi.org/1

mRNA expression by non-tumor cells in the glioma microenviron-ment. Collectively, these data show a correlation betweenincreased inflammatory gene expression and a benefit to survivalin immunocompetent mice with intracranial GBM cell IDO1deficiency.

3.2. Immune checkpoint blockade-mediated survival and intracranialmouse GBM kynurenine (Kyn) levels

Previous work in a peripheral (outside of the CNS) B16 mousemelanoma model demonstrated that the therapeutic effects ofCTLA-4- or PD-1-mediated immune checkpoint blockade, syner-gize with germline Ido1 deletion in mice (Holmgaard et al.,2013). In contrast, we previously showed that the dual inhibition

nantly contributes to enzyme activity and response to CTLA-4/PD-L1 inhi-0.1016/j.bbi.2017.01.022

4 L. Zhai et al. / Brain, Behavior, and Immunity xxx (2017) xxx–xxx

of CTLA-4/PD-L1 is less effective at achieving an overall survivalbenefit in Ido1�/�, when compared to WT mice with GL261 braintumors (Wainwright et al., 2014). To further explore the require-ment for germline IDO1 to achieve a maximal survival benefit fromimmune checkpoint therapy against brain tumors, WT and Ido1�/�

mice engrafted unmodified GL261 were administered Syrian ham-ster IgG and rat IgG2B or CTLA-4 and PD-L1 mAbs at 2 weeks post-

Fig. 2. The effects of CTLA-4/PD-L1 blockade on survival, enzyme activity and inflammamouse glioblastoma (GBM). (A) A timeline indicating a dosing schedule of control IgG anmice intracranially-engrafted 2 � 105 unmodified GL261 cells and treated with control IgIdo1�/� mice were engrafted 2 � 105 unmodified GL261 cells and treated with CTLAindependent experiments is shown. (C) WT (red bars) or Ido1�/� (blue bars) mice were iIgG (solid bars) or CTLA-4/PD-L1 mAbs (hatch-filled bars). Tryptophan (Trp) and kynurserum, contralateral brain without tumor from a tumor-bearing mouse (Contral. brain), G(n = 10); WT mice + CTLA-4/PD-L1 mAbs (n = 11); Ido1�/� mice + IgG (n = 5); Ido1�/� micfor markers associated with T lymphocytes, proinflammatory cytokines or Ido1 at 3 weeintracranially-engrafted 2 � 105 unmodified GL261 and treated as shown in (A). Sample

Please cite this article in press as: Zhai, L., et al. Non-tumor cell IDO1 predomibition in mouse glioblastoma. Brain Behav. Immun. (2017), http://dx.doi.org/1

intracranial injection (Fig. 2A). While WT mice treated with CTLA-4/PD-L1 mAbs have a median survival of 55 days, Ido1�/� miceundergoing similar treatment possess a decreased median survivalof 38 days (P < 0.0001), similar to WTmice treated with IgG controlantibodies (Fig. 2B). Notably, the CTLA-4/PD-L1-mediated survivalbenefit was durable in a subset of WT mice, while no Ido1�/� micesurvived >45 days post-GBM cell engraftment (Fig. 2B). These data

tory gene expression in wild-type (WT) and IDO1-deficient (Ido�/�) mice engraftedtibodies (Abs) or CTLA-4 and PD-L1 mAbs. (B) Kaplan-Meier survival analysis of WTG (green triangle, n = 5) or CTLA-4 and PD-L1 mAbs (red circle, n = 10). Additionally,-4 and PD-L1 mAbs (blue square, n = 10). One representative dataset from twontracranially-engrafted 2 � 105 GL261 cells followed by the treatment with controlenine (Kyn) levels were measured at 3 weeks post-intracranial engraftment in theL261 GBM lysate and draining cervical lymph nodes (cLN) by HPLC. WT mice + IgG

e + CTLA-4/PD-L1 mAbs (n = 6). (D) RT-PCR analysis of intratumoral gene expressionks post-intracranial engraftment of tumor cells isolated from WT and Ido1�/� micesize is the same as in (C). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

nantly contributes to enzyme activity and response to CTLA-4/PD-L1 inhi-0.1016/j.bbi.2017.01.022

L. Zhai et al. / Brain, Behavior, and Immunity xxx (2017) xxx–xxx 5

confirm the unexpected requirement for non-tumor cell IDO1 toachieve maximal survival in response to immune checkpointblockade in mice with experimental GBM.

Based on the paradoxical requirement for non-tumor cell IDO1to achieve maximal therapeutic effectiveness with immune check-point blockade, we next explored the role of IDO1 enzyme activityand inflammatory gene expression in the experimental mouseGBM model. These findings, in combination with our otherobservations demonstrating the role of non-tumor cell IDO1 as aprimary contributor to enzyme activity (Fig. 1A), led us to hypoth-esize that, non-tumor cell IDO1 contributes an essential pool ofKyn that facilitates responsiveness to immune checkpoint block-ade. To pursue this question, Trp and Kyn levels were quantifiedin the sera, contralateral hemisphere without brain tumor (Con-tral. brain), GBM and the draining cervical lymph nodes in miceat 3 weeks post-GL261 cell engraftment (Fig. 2C). Notably, themouse treatment schedule for administration of control IgG Absand CTLA-4/PD-L1 mAbs were in accordance with the time linepreviously described (Fig. 2A). Trp levels are not affected byimmune checkpoint inhibition regardless of the germline IDO1status in mice. However, Kyn levels and the Kyn/Trp ratio are sig-nificantly increased in the sera and GBM of WT when compared toIdo1�/� mice (P < 0.05, respectively). In contrast, there is no changein the Kyn levels and Kyn/Trp ratio of contralateral brain isolatedfrom GBM-bearing WT and Ido1�/� mice, suggesting that the tryp-tophan catabolic effects of immune checkpoint blockade selec-tively affect different anatomical compartments in subjects withbrain tumors. Unexpectedly, immune checkpoint blockade hadno significant effect on intratumoral inflammatory gene expres-sion associated with markers expressed by T cells, Cd8a andCd3e, as well as the proinflammatory cytokines, Il1b and Ifng(Fig. 2D). Interestingly, while we observed a trend for increasedintratumoral Ido1 expression in WT mice treated with immunecheckpoint blockade, there was no trend in Ido1�/� mice, suggest-ing the possibility that CTLA-4/PD-L1 inhibition marginallyincreases IDO1 expression in non-tumor cells. Collectively, thesedata suggest the hypothesis whereby immune checkpoint inhibi-tion requires increased Kyn levels, provided by non-tumor cellIDO1, to achieve maximal immune-mediated antitumor effectsagainst GBM.

4. Discussion

The canonical function of IDO1 is associated with its ability tocatalyze the conversion of Trp into Kyn, leading to the inhibitionof T cell effector functions (Uyttenhove et al., 2003; Munn et al.,1998, 1999). However, not all immunosuppressive aspects ofIDO1 can be explained through the Trp depletion/Kyn accumula-tion theory. Indoximod (D1-MT) is a non-enzymatic IDO1 pathwayinhibitor (Zhai et al., 2015b; Lob et al., 2009) that is currently inclinical trials, having previously shown antitumor activity in pre-clinical cancer models when combined with standard of careapproaches (Hou et al., 2007). Additionally, mouse plasmacytoidDCs (pDCs) treated with transforming growth factor-b (TGF-b)cause IDO1 to mediate tolerogenic effects in T cells that are mech-anistically independent from its enzymatic activity (Pallotta et al.,2011). The results from our current study provide further supporttoward the hypothesis that, IDO1 possesses a non-enzymatic,immunosuppressive function in mouse GBM cells. This is basedon GBM cell IDO1 facilitating Treg accumulation and limitingmouse survival (Wainwright et al., 2012) through anIDO1 enzyme-independent mechanism (Fig. 1A), combined withnon-GBM cell IDO1 possessing no effect on mouse survival(Wainwright et al., 2012) despite its primary contribution towardincreasing Kyn levels in mouse GBM (Fig. 1A).

Please cite this article in press as: Zhai, L., et al. Non-tumor cell IDO1 predomibition in mouse glioblastoma. Brain Behav. Immun. (2017), http://dx.doi.org/1

Given the current enthusiasm of combinatorial immunotherapyfor the treatment of cancer, the results from this study demon-strate the importance of understanding the effects of differentIDO1 targeting approaches that may vary in effectiveness, depend-ing on the type of cancer being investigated. Our findings highlightthe importance of developing future strategies that inhibit thenoncanonical IDO1 activity in GBM cells, without disrupting thenon-tumor cell IDO1-mediated enzyme activity associated withresponsiveness to immune checkpoint blockade (Fig. 2B and C).We do recognize that our study is limited by the evaluation of sin-gle time points after tumor engraftment, rather than a rigoroustime course analysis of Trp/Kyn and inflammatory gene expressionlevels. In the future, we also hope to better understand the non-canonical IDO1 effects in mouse GBM cells. To address this, whilekeeping in-mind the spontaneous tumor rejection that occurs inimmunocompetent mice engrafted GL261Ido1-KD cells, our grouphas developed Ido1�/� Rag1�/� double knockout mice. This modelwill support the future evaluation of comparing IDO1 activity inthe absence of GBM-infiltrating T cells, which is critical for com-paring to the previous findings of Grohmann et al. reporting that,CTLA-4-induced immunotolerance is contingent upon effectiveTrp catabolism in dendritic cells (Grohmann et al., 2002). Giventhat DCs infiltrate intracranial GBM (Curtin et al., 2009), it becomesmore likely that non-tumor cells including microglia, macrophagesand/or DCs, require IDO1 to mediate enzyme-dependent respon-siveness to immune checkpoint inhibition.

It was surprising to find increased intratumoral mRNA expres-sion for T cell markers and proinflammatory cytokines in mice thatsurvive longer when engrafted IDO1-deficient GBM cells(Fig. 1C and F), but not in WT mice treated with CTLA-4/PD-L1mAbs that also possess a substantial survival advantage(Fig. 2B and D). It is therefore important to keep in-mind our noveldata showing that mouse GBM cells do not significantly contributeto IDO1-mediated Kyn accumulation in brain tumors (Fig. 1A),while the non-tumor cell IDO1 is required for maximal therapeuticresponse to CTLA-4/PD-L1 blockade (Fig. 2B) and associated withhigher Kyn levels in WT-, when compared to Ido1�/�-miceengrafted GBM (Fig. 2C). Based on these collective observations,we hypothesize that the increased survival in mice intracranially-engrafted IDO1-deficient GBM cells reflects the inability of tumorcells to suppress T cell-mediated cytolytic effects. We furtherhypothesize that the diminished therapeutic response to CTLA-4/PD-L1 mAb treatment in Ido1�/� mice reflects the unexpectedrequirement for tryptophan catabolism to fuel T cell-mediatedtumor rejection in brain tumors; independent of T cell infiltrationand proinflammatory cytokine expression.

One surprising observation that was unexpected is the signifi-cantly increased Kyn/Trp ratio in draining cervical lymph nodes(cLN) of WT mice engrafted GL261Ido1-WT- when compared toGL261Ido1-KD glioma (Fig. 1A). Given that mice with GL261Ido1-WT

brain tumors possess significantly increased Treg levels when com-pared to GL261Ido1-KD (Wainwright et al., 2012), it is possible thatthe increased Kyn/Trp ratio is somehow associated with enhancedTreg generation in the cLN. However, we find this possibility unli-kely, given that there is no increase in the Kyn/Trp ratio in Ido1�/�

mice engrafted IDO1-competent GL261 glioma, which also resultsin tumor-infiltrating Treg accumulation (Wainwright et al., 2012).Another interesting finding was the increased Kyn/Trp ratio in theGBM of WT when compared to Ido1�/� mice (Fig. 2C), which didnot occur in the contralateral brain without tumor. This suggeststhat under normal conditions, IDO1 activity is negligible in theCNS, but is significantly increased by traumatic and/or proinflam-matory events such as GBM development.

Taken together, our data demonstrate for the first time thatGBM cell IDO1 suppresses tumor immunity through a mechanismindependent of intratumoral Trp? Kyn catabolism in an

nantly contributes to enzyme activity and response to CTLA-4/PD-L1 inhi-0.1016/j.bbi.2017.01.022

6 L. Zhai et al. / Brain, Behavior, and Immunity xxx (2017) xxx–xxx

immunocompetent, experimental mouse model of malignantglioma. In contrast, non-tumor cell IDO1 is required for the maxi-mal therapeutic effect of immune checkpoint blockade and is asso-ciated with increased intratumoral Kyn levels in WT, whencompared to Ido1�/� mice. Since the GL261 cell lines, techniquesand time points utilized in the current study are identical to thosein previous studies (Wainwright et al., 2012), these findings are inalignment with the intriguing hypothesis that, IDO1 regulatesbrain tumor immunity through distinctly different mechanismsdepending on the CNS cell type under consideration. Given thatmost IDO1 targeting strategies focus on inhibiting enzyme activity(Röhrig et al., 2015), our findings also provide novel considerationsfor designing second generation inhibitors that target the ill-defined immunosuppressive aspects of IDO1 in tumor cells. Ulti-mately, these data provide strong rationale for the continued pur-suit of determining how IDO1 suppresses the immune response inmalignant brain tumors and an extension and expansion of theseanalyses into human glioblastoma.

Conflict of interest disclosure

None.

Acknowledgments

This work was supported NIH grants K99 NS082381 (D.A.W.),R00 NS082381 (D.A.W.), R01 MH101145 (RHM) and R01NS097851-01 (D.A.W.), Robert H. Lurie Comprehensive CancerCenter – Zell Scholar Program of the Zell Family Foundation Gift(D.A.W.) and Cancer Research Institute – Clinic and LaboratoryIntegration Program.

We thank Dr. Suzanne J. Baker, Ph.D. (St. Jude Children’sResearch Hospital) for generously donating her tamoxifen-inducible mouse glioma model for these studies.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.bbi.2017.01.022.

References

Binder, D.C., Davis, A.A., Wainwright, D.A., 2015. Immunotherapy for cancer in thecentral nervous system: current and future directions. Oncoimmunol. 5,e1082027.

Carlin, J.M., Borden, E.C., Sondel, P.M., Byrne, G.I., 1987. Biologic-response-modifier-induced indoleamine 2,3-dioxygenase activity in human peripheral bloodmononuclear cell cultures. J Immunol. 139, 2414–2418.

Chow, L.M., Endersby, R., Zhu, X., Rankin, S., Qu, C., Zhang, J., et al., 2011.Cooperativity within and among Pten, p53, and Rb pathways induces high-grade astrocytoma in adult brain. Cancer Cell 19, 305–316.

Curtin, J.F., Liu, N., Candolfi, M., Xiong, W., Assi, H., Yagiz, K., et al., 2009. HMGB1mediates endogenous TLR2 activation and brain tumor regression. PLoS Med. 6,e10.

Please cite this article in press as: Zhai, L., et al. Non-tumor cell IDO1 predomibition in mouse glioblastoma. Brain Behav. Immun. (2017), http://dx.doi.org/1

Fallarino, F., Grohmann, U., Vacca, C., Bianchi, R., Orabona, C., Spreca, A., et al., 2002.T cell apoptosis by tryptophan catabolism. Cell Death Differ. 9, 1069–1077.

Grohmann, U., Orabona, C., Fallarino, F., Vacca, C., Calcinaro, F., Falorni, A., et al.,2002. CTLA-4-Ig regulates tryptophan catabolism in vivo. Nat. Immunol. 3,1097–1101.

Holmgaard, R.B., Zamarin, D., Munn, D.H., Wolchok, J.D., Allison, J.P., 2013.Indoleamine 2,3-dioxygenase is a critical resistance mechanism in antitumorT cell immunotherapy targeting CTLA-4. J. Exp. Med. 210, 1389–1402.

Hou, D.Y., Muller, A.J., Sharma, M.D., DuHadaway, J., Banerjee, T., Johnson, M., et al.,2007. Inhibition of indoleamine 2,3-dioxygenase in dendritic cells bystereoisomers of 1-methyl-tryptophan correlates with antitumor responses.Cancer Res. 67, 792–801.

Ladomersky, E., Genet, M., Zhai, L., Gritsina, G., Lauing, K.L., Fangusaro, J., et al., 2016.Improving Vaccine Efficacy Against Malignant Glioma. Oncoimmunol. 5,e1196311.

Larkin, J., Chiarion-Sileni, V., Gonzalez, R., Grob, J.J., Cowey, C.L., Lao, C.D., et al.,2015. Combined nivolumab and ipilimumab or monotherapy in untreatedmelanoma. N. Engl. J. Med. 373, 23–34.

Lob, S., Konigsrainer, A., Zieker, D., Brucher, B.L., Rammensee, H.G., Opelz, G., et al.,2009. IDO1 and IDO2 are expressed in human tumors: levo- but not dextro-1-methyl tryptophan inhibits tryptophan catabolism. Cancer Immunol.Immunother. 58, 153–157.

Mezrich, J.D., Fechner, J.H., Zhang, X., Johnson, B.P., Burlingham, W.J., Bradfield, C.A.,2010. An interaction between kynurenine and the aryl hydrocarbon receptorcan generate regulatory T cells. J Immunol. 185, 3190–3198.

Munn, D.H., Zhou, M., Attwood, J.T., Bondarev, I., Conway, S.J., Marshall, B., et al.,1998. Prevention of allogeneic fetal rejection by tryptophan catabolism. Science(New York, NY) 281, 1191–1193.

Munn, D.H., Shafizadeh, E., Attwood, J.T., Bondarev, I., Pashine, A., Mellor, A.L., 1999.Inhibition of T cell proliferation by macrophage tryptophan catabolism. J. Exp.Med. 189, 1363–1372.

Munn, D.H., 2012. Blocking IDO activity to enhance anti-tumor immunity. Front.Biosci. (Elite ed.). 4, 734–745.

Pallotta, M.T., Orabona, C., Volpi, C., Vacca, C., Belladonna, M.L., Bianchi, R., et al.,2011. Indoleamine 2,3-dioxygenase is a signaling protein in long-termtolerance by dendritic cells. Nat. Immunol. 12, 870–878.

Reardon, D.A., Gilbert, M.R., Wick, W., Liau, L., 2015. Immunotherapy for neuro-oncology: the critical rationale for combinatorial therapy. Neuro Oncol. 17(Suppl 7), vii32–vii40.

Röhrig, U.F., Majjigapu, S.R., Vogel, P., Zoete, V., Michielin, O., 2015. Challenges in theDiscovery of Indoleamine 2,3-Dioxygenase 1 (IDO1) Inhibitors. J. Med. Chem.58, 9421–9437.

Stupp, R., Mason, W.P., van den Bent, M.J., Weller, M., Fisher, B., Taphoorn, M.J., et al.,2005. Radiotherapy plus concomitant and adjuvant temozolomide forglioblastoma. N. Engl. J. Med. 352, 987–996.

Uyttenhove, C., Pilotte, L., Theate, I., Stroobant, V., Colau, D., Parmentier, N., et al.,2003. Evidence for a tumoral immune resistance mechanism based ontryptophan degradation by indoleamine 2,3-dioxygenase. Nat. Med. 9, 1269–1274.

Wainwright, D.A., Balyasnikova, I.V., Chang, A.L., Ahmed, A.U., Moon, K.-S.,Auffinger, B., et al., 2012. IDO Expression in Brain Tumors Increases theRecruitment of Regulatory T Cells and Negatively Impacts Survival. Clin. CancerRes. 18, 6110–6121.

Wainwright, D.A., Chang, A.L., Dey, M., Balyasnikova, I.V., Kim, C., Tobias, A.L., et al.,2014. Durable therapeutic efficacy utilizing combinatorial blockade againstIDO, CTLA-4 and PD-L1 in mice with brain tumors. Clin. Cancer Res. 20, 5290–5301.

Wolchok, J.D., Kluger, H., Callahan, M.K., Postow, M.A., Rizvi, N.A., Lesokhin, A.M.,et al., 2013. Nivolumab plus ipilimumab in advanced melanoma. N. Engl. J. Med.369, 122–133.

Zhai, L., Dey, M., Lauing, K.L., Gritsina, G., Kaur, R., Lukas, R.V., et al., 2015a. Thekynurenine to tryptophan ratio as a prognostic tool for glioblastoma patientsenrolling in immunotherapy. J. Clin. Neurosci. 22, 1964–1968.

Zhai, L., Spranger, S., Binder, D.C., Gritsina, G., Lauing, K.L., Giles, F.J., et al., 2015b.Molecular pathways: targeting IDO1 and other tryptophan dioxygenases forcancer immunotherapy. Clin. Cancer Res. 21, 5427–5433.

nantly contributes to enzyme activity and response to CTLA-4/PD-L1 inhi-0.1016/j.bbi.2017.01.022