C57BL/6J mice were maintained in the University of Illinois’s Institute for Genomic Biology animal facility. Procedures and animal care were in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council) and approved by the Institutional Animal Care and Use Committee. Young adult mice, used for neuroinflammation (intraperitoneal, i.p. LPS or pI:C) experiments, were 12-week-old males at time of treatment. Saline (control), LPS, or pI:C treatments were administered prior to the start of the dark cycle. LPS (serotype 0127:B8, Sigma, St. Louis, MO, USA) was administered at 330 μg/kg body weight; this dose has been shown to elicit neuroinflammation, elevated Ido1, Ido2, and Tdo2 expression and depression-like behaviors (40). pI:C (P0913, Sigma, St. Louis, MO, USA) was administered at 12 mg/kg body weight. Previous work has shown that pI:C also causes cytokine and Ido1 response in the hippocampus and depression-like behaviors (6). Mice were sacrificed 6 h post-treatment.

Heat-inactivated horse serum (SH30074.03), Hank’s balanced salt solution (HBSS, SH30030.03), Eagle’s minimal essential medium (MEM, SH30024.02) were purchased from Hyclone (Logan, UT, USA). d-glucose (15023–021) was purchased from Gibco (Carlsbad, CA, USA). Dex (tested at 1 μM, D4902), pI:C (tested at 10 μg/ml, P0913), and Gey’s balanced salt solution (GBSS, G9779) were from Sigma (St. Louis, MO, USA). Recombinant mouse IFNγ (tested at 10 ng/ml, 315–05) was from Peprotech (Minneapolis, MN, USA). Gal-1 (1245), Gal-3 (1197), and Gal-9 (3535), all tested at 1 μg/ml and TNFα (tested at 10 ng/ml, 410-MT) were purchased from R&D Systems (Minneapolis, MN, USA). IL-1β (tested at 10 ng/ml, IL014) was purchased from Chemicon International (Temecula, CA, USA).

Organotypic hippocampal slice cultures were prepared from hippocampi of C57BL/6J pups, 7–10 days old, as previously described (39). Briefly, hippocampi were cut into 350 μm transverse slices with a McIlwain tissue chopper (Campden Instruments Ltd., UK) and positioned onto membrane inserts (PICM03050, EMD Millipore) with six slices per insert, each slice from a different animal. Inserts were placed into 6-well culture plates with 1.25 ml of medium [50% MEM, 25% horse serum, 1 × penicillin/streptomycin (Pen/Strep), 15 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES buffer, pH 7.4), 25% HBSS, and 0.5% d-glucose]. Every 2–3 days, medium was changed. On the seventh day, OHSCs were rinsed three times with serum-free medium (Dulbecco’s MEM, Pen/Strep, HEPES and 150 μg/ml bovine serum albumin) and incubated for 2 h. Slices were rinsed again with serum-free medium and treatments added to fresh serum-free medium. OHSCs were treated with IFNγ, TNFα, IL-1β, pI:C, or Dex, with or without Gal-9. To determine if effects were specific to Gal-9, independent OHSC preparations were treated with IFNγ and pI:C with or without Gal-1 and Gal-3. Six hours later, tissues and supernatants were collected and kept at −80°C for subsequent analysis.

RNA was extracted using the EZNA. Total RNA Kit II (Omega Bio-Tek, Norcross, GA, USA) and cDNA was prepared with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Grand Island, NY, USA). The cDNA was amplified to quantify steady-state gene expression by qPCR using TaqMan Universal PCR Master Mix and Prism 7900 thermocycler (Applied Biosystems, Foster City, CA, USA). Using the comparative threshold cycle method (GAPDH used to normalize target gene expression), differences in cDNA levels were determined by comparing 2^−ΔΔCts, C_t = cycle threshold (39).

Our previous publication describes validated qPCR assays to quantify the expression of distinct DO transcripts in mouse tissues and OHSCs. Assays for Ido1 (Ido1-FL, Ido1-v1), Ido2 (Ido2-FL, Ido2-v1, Ido2-v6, and Ido2-v3), and Tdo2 (Tdo2-FL, Tdo2-v1, and Tdo2-v2) were described (39). Gene structure for Ido1, Ido2, and Tdo2 is shown in Figure 1. Structures for the previously described transcripts are shown as well as the exon-structure of additional Ido1 (Ido1-v2) and Ido2 variants (Ido2-v2, Ido2-v4, and Ido2-v5) that were quantified for the first time for this manuscript. Assay specifics for DO transcript analyses are shown in Table 1. Probe-based assays were purchased from IDT with custom assays fashioned using the IDT PrimerQuest^® Design Tool. Correct amplicon size was confirmed by gel electrophoresis. Several transcripts (notably Ido1-FL) are not detectable in all naive mouse brain or control OHSC samples (C_t values “undetermined”), thus negating the ability to properly calculate relative gene expression: for mathematical analysis, a C_t value of 40.0 was assigned when this occurred.

Data are reported as mean ± SEM for three to four independent OHSC preparations per treatment combination and six mice/treatment group for the adult animal study. SigmaPlot 13.0 software was used to conduct two-way analysis of variance (two-way ANOVA) using a 2 × 6 (OHSC) or 2 × 2 (in vivo) arrangement of treatments. In the presence of a significant interaction, assessed by Holm–Šídák method, post hoc analyses for multiple comparisons were performed. Comparisons were considered significant at p < 0.05.

We previously detailed the regulation of two Ido1 transcripts and compared their relative expression across multiple areas of the brain. Ido1-v1 was well expressed (39) in the mouse hippocampus and other brain regions, whereas Ido1-FL is low/undetectable across the naive mouse brain (39). Like Ido1-FL, we now show that a third Ido1 transcript, Ido1-v2, is low/undetectable in the mouse hippocampus of naive mice (Figure 2). Not only is hippocampal basal expression of these three Ido1 transcripts different but also the regulation of Ido1 transcripts by inflammatory mediators and corticosteroids is unique as well. Expression of both Ido1-FL and Ido1-v2 were induced by LPS and pI:C in vivo (current work), but not ex vivo (39), suggesting that in vivo hippocampal responses to LPS and pI:C must be mediated by other secondary factors: presumably, cytokines and other immunomodulatory mediators. Using OHSCs, we found that only IFNγ was able to directly induce Ido1-FL and Ido1-v1 expression [Figure 6; (39)]. Since brain/hippocampal IFNγ is inducible by LPS and pI:C, but OHSCs did not express IFNγ (not shown), non-resident cells from the periphery are probably responsible for the elevated IFNγ expression within the brain that occurs in vivo and thus for the increase in Ido1 expression. Indeed, during neuroinflammation, IFNγ⁺ cells infiltrate the brain (42). Hippocampal Ido1-v2 expression is increased by LPS and pI:C in vivo; however, we have not, as yet, identified the mechanism responsible for this response; none of the treatments including IFNγ increased Ido1-v2 expression by OHSCs.

In the naive hippocampus and control OHSCs, only Ido1-v1 is readily detectable, but Ido1-v1 expression is far less sensitive to pro-inflammatory induction compared to Ido1-FL. These data suggest that Ido1-v1 probably has a major role in basal metabolism [supplying Kyn for nicotinamide/NAD⁺ synthesis (43)], whereas the sizeable induction of Ido1-FL by IFNγ plus either Gal-9 (current work) or glucocorticoids (39) is necessary to meet the increased energy/NAD⁺ demand associated with inflammation (44–46) or stress, respectively (Figure 10). Indeed, elevated central nervous system NAD⁺ levels during experimental MS (47) are associated with increased Ido1 (48). An unfortunate consequence of Ido1 induction is elevated production and release of downstream neuroactive Kyn metabolites (QuinA and KynA) that mediate depression-like, anxiogenic, and adverse cognitive behaviors (49).

We recently found that several Ido2 transcripts are differentially expressed across the mouse brain, and again, inflammatory mediators and corticosteroids differentially regulate Ido2 transcripts (39). Obviously, posttranslational processing of Ido2 (and Ido1) is not a random process but is utilized to fine-tune mRNA expression. A switch in posttranslational processing of Ido2 was first addressed using macrophages and B-cells, wherein both cell types can switch from expression of predominantly Ido2-FL to predominantly Ido2-v3 (50). While investigating Ido2 expression in the brain, we had first shown that LPS increased hippocampal Ido2 expression in vivo, using an assay that simultaneously quantified all Ido2 transcripts (40). In the current work, we refined our understanding of central Ido2 expression by quantifying the regulation of its various transcripts. During in vivo neuroinflammation, LPS and pI:C increased hippocampal Ido2 expression in a transcript-specific manner. Both LPS and pI:C increased Ido2-v1 expression; however, only LPS increased Ido2-v3 and Ido2-v6/v2. Since LPS and pI:C were administered intraperitoneally, we used OHSCs to identify direct versus indirect mechanisms responsible for Ido2 induction.

Together with our previous work (39), OHSC expression of Ido2-v1 and Ido2-v3 was increased by IFNγ, LPS, and pI:C, while Ido2-v4, Ido2-v5, and Ido2-v6 were increased slightly but by only IFNγ. These data indicate that LPS and pI:C may act directly or via IFNγ to induce Ido2-v1 and Ido2-v3 within the hippocampus. In contrast, in vivo induction of Ido2-v4, Ido2-v5, and Ido2-v6 are mediated by IFNγ. Uniquely, Ido2-v2 is highly expressed, but unaffected by inflammatory mediators and galectins. Overall, in both the mouse hippocampus and OHSCs, Ido2-v1 and Ido2-v3 are most susceptible to induction by inflammation, whereas the other transcripts are either unchanged in vivo or modestly induced ex vivo. These data suggest that Ido2-v1 and Ido2-v3 induction was necessary to increase flux down the Kynurenine Pathway (Figure 10) to meet the increased energy (NAD⁺) demand associated with inflammation (44–46). A repercussion of their induction may be altered behaviors associated with Kyn production (50) and conversion to QuinA or KynA (20). Other Ido2 transcripts are relatively intractable and involved in basal metabolism.

Excitingly, in the presence of IFNγ, Gal-9 accentuated Ido1-FL and Ido2-v1 expression by OHSCs; neither Gal-1 nor Gal-3 affected DO expression. This statistical interaction suggests that Ido1 and Ido2 are regulated in a transcript-specific manner by additional factors: glucocorticoids (39) and Gal-9 (current work). Although IFNγ initiates Ido1-FL and Ido2-v1 induction [and indeed IFNγ action is necessary for neuroinflammation-induced Ido induction (10)], full induction in vivo requires synergy with glucocorticoids or Gal-9. As Gal-9 is expressed in many cell types within the brain, expression and induction of hippocampal Gal-9 can occur via activation of cells resident to the brain. Gal-9 can bind to many cell surface partners including the T-cell immunoglobulin and mucin-domain containing-3 (TIM-3) receptor, as well as directly associating with several transcription factors (51) within the cell. Further analysis is required to define which of these possible mechanisms are responsible for Gal-9’s ability to enhance IFNγ-induced DO expression.

There are multiple autoimmune disorders associated with elevated levels of both Ido1 and Gal-9 (32, 34–38). As Ido2 function is associated with the development of severe rheumatoid arthritis symptoms (52), Gal-9’s ability to enhance Ido2 expression may promote the development or severity of symptoms of rheumatoid arthritis or other autoimmune conditions. Thus, further characterization of the synergy between IFNγ and Gal-9 during the induction of Ido1 and Ido2 is important to our understanding the pathophysiology of autoimmune disorders such as MS, Hashimoto’s disease, or rheumatoid arthritis. Similarly, Gal-9 is an important regulator of the immune system, promoting the differentiation of T regulatory cells (53) and modulating viral pathogenesis (54). Ido1 plays a similar role in T cell differentiation (55) and also modulates viral pathogenesis (56). These findings open that possibility that one of the mechanisms by which Gal-9 controls T cell differentiation and viral pathogenesis is its ability to direct DO expression. These links warrant further investigation.

Three Tdo2 transcripts are expressed in the mouse brain, only Tdo2-FL is enriched in hippocampus (39, 57). Similar to Ido1 and Ido2, this specific enrichment implies a distinct regulatory mechanism for transcript expression. Although Tdo2-v1 and Tdo2-v2 expression was increased in the hippocampus post-pI:C administration to mice, treatment of OHSCs with pI:C, pro-inflammatory cytokines or LPS (39) did not increase Tdo2 expression. Thus, another factor(s) must mediate increased Tdo2 expression in vivo. Using OHSCs, Tdo2-v1 and Tdo2-v2 expression was increased by Gal-9. Since pI:C was a more potent inducer of hippocampal Gal-9 compared to LPS in vivo, the ability of pI:C administration to increase Tdo2-v1 and Tdo2-v2 expression may be mediated by central Gal-9 induction. In contrast to Tdo2-v1 and Tdo2-v2, Tdo2-FL expression was increased by glucocorticoids (Dex and corticosterone) in OHSCs [Figure 8; (39)], but not cytokines or Gal-9. Thus, like the Ido’s, Tdo2 evolved to permit differential transcript expression in a stress- and inflammation-specific manner within the brain. Tdo2-FL expression is dependent on glucocorticoid activity; whereas Tdo2-v1 and Tdo2-v2 are responsive to Gal-9. These data suggest that different promotors used for Tdo2 induction (58) are necessary to meet the increased metabolic (NAD⁺) demand (Figure 10) associated with stress (Tdo2-FL) and neuroinflammation (Tdo2-v1 and Tdo2-v2), as previously suggested (44–46). A repercussion of aberrant Tdo2 expression may be altered behavior such as those associated with schizophrenia (59, 60) and anxiety (61, 62).