In clinical practice, the therapeutic response of patients with ileal Crohn’s disease (iCD) frequently differs from that of patients with extensive CD-associated inflammation of the colon. In 2019, Dulai et al. hypothesized that Crohn’s disease could be reclassified into ileum-dominant or isolated colonic CD subtypes [1
]. For example, clinical outcomes under therapy with the monoclonal tumor necrosis factor alpha (TNFα) antibody, infliximab, are generally superior in patients with colonic CD (cCD) compared with iCD [2
]. Furthermore, studies based on a variety of serological [4
], clinical [5
], microbial [6
] and genetic [7
] markers have described a corresponding heterogeneity in patterns of involvement in CD. Although disease localization has been unequivocally linked to differing clinical outcomes in patients with CD, the mechanisms behind this differential pathology are still unclear. In this context, the identification of a biomarker capable of clearly differentiating the phenotypes of CD would represent a crucial step on the way to an individualized, patient-specific therapy optimized to achieve complete mucosal healing, a central therapeutic goal of CD therapy.
Nikolaus et al. were able to show a decrease in tryptophan (TRP) and an increase in kynurenine (Kyn) concentrations in the serum of patients with inflammatory bowel disease (IBD). In addition, indoleamine 2,3-dioxygenase (IDO1) mRNA expression has been shown to be elevated in inflamed tissue of patients with CD and ulcerative colitis (UC) [10
]. In addition, Gupta et al. observed a correlation between TRP metabolism and the CD activity [11
]. Correspondingly, immigration of innate immune cells into inflamed regions of the gastrointestinal system has been described as the main source of TRP-degrading enzymes and producers of TRP downstream metabolites [12
]. Thus, the TRP metabolization arsenal is carried to the site of inflammation by immune cells. However, it remains unclear (1) whether IDO1 is expressed in an activated state, (2) how high the local mucosal Kyn concentrations are and (3) how further degradation of Kyn into TRP downstream metabolites is regulated.
TRP immune metabolism is strongly involved in keeping the intestinal barrier intact and preventing inflammation by mediating regulatory functions affecting regulatory T cells (Treg) and Th17 adaptive immunity [14
]. It is thought that this immunoregulatory function is performed by IDO and its first downstream TRP metabolite Kyn. In principle, the majority of publications describe Kyn as a regulator of immunosuppression [16
]. Kyn has been shown to interact with the intracellular transcription factor aryl hydrocarbon receptor (AHR), thereby suppressing effector T cells by regulating expression of cytotoxic T-lymphocyte-associated protein 4 and programmed death protein 1 as well as promoting the differentiation of Treg [18
]. It remains unclear why, despite increased expression of IDO1 and the systemic elevation of Kyn, highly active inflammatory processes nevertheless persist in the pathology of IBD.
Whereas both TRP and Kyn, as well as the corresponding TRP-metabolizing enzymes IDO1 and tryptophan 2,3-dioxygenase (TDO) have been examined in some detail, downstream enzymes and metabolites of the TRP metabolism have not been closely studied. In this context, Harden et al. revealed an increase in the expression of the Kyn-degrading enzyme kynureninase (KYNU) in the Th1/Th17 lymphocyte-driven skin disease psoriasis as well as in the general clinical manifestations of chronic inflammation-related diseases [21
]. The authors concluded that KYNU may act as a dichotomous modulator of immune responses, regulating the switch between immunosuppression and inflammation. While in cancer, IDO1 is preferentially upregulated, in chronic inflammatory diseases such as psoriasis and atopic dermatitis, elevation of KYNU was found to be more pronounced than that of IDO1 [21
]. Whether KYNU is increased in IBD, and what function this Kyn-degrading enzyme has in the development and course of chronic inflammatory disorders, is still unclear. This raises the question of whether the inflammation-related pathologies arise due to a decrease in local Kyn concentration, or whether they are caused by an increase in the potentially proinflammatory downstream TRP metabolite 3-hydroxyanthranilic acid (3-HAA) [21
The aim of this study was to investigate the local mucosal TRP immune metabolism with a special focus on the enzyme KYNU. Additionally, we performed a detailed comparison of site-specific inflammation-induced immune cell infiltration in patients with IBD. We found differences in the expression pattern of KYNU in the different CD phenotypes and a decreased level of the immunosuppressive TRP metabolite Kyn in iCD. Furthermore, we identified distinct characteristic immunological profiles for iCD and cCD. In summary, we characterized KYNU as a potential new marker and therapeutic target for iCD as regulator of the ratio of pro- and anti-inflammatory TRP downstream metabolites.
2.1. Ethical Approval
This research was performed in line with the principles of the Declaration of Helsinki. The study (309_19 Bc) was approved by the ethical committee Friedrich-Alexander-Universität Erlangen-Nürnberg.
2.2. Patient and Tissue Sample Characteristics
In order to characterize immune cell markers, enzymes and metabolites of the TRP metabolism through immunohistochemistry (IHC), we included three different cohorts of patients with IBD and defined them as control, CD and UC (Table 1
). Controls were defined as individuals that were scoped within the framework of colorectal cancer screening but had neither symptoms, nor polyps nor inflammatory lesions. The CD cohort was subclassified according to disease localization by differentiating patients with iCD from those with cCD. In both IBD cohorts, CD and UC, biopsies were collected from patients with an active disease state and obtained from the bowel region showing maximally severe inflammation as judged by the endoscopist. In the CD cohort, 14 samples were collected from 12 patients. Nine of these tissue samples showed small bowel inflammation, whereas five revealed colonic inflammation. In the UC cohort of 11 patients, as expected, all samples showed inflammation of the large intestine. In this study setting, one biopsy was taken per individual (patient/control) and location (small intestine/large intestine). Thus, a maximum of two biopsies (small intestine + large intestine) per patient/control were examined. The inflammatory state of the intestinal tissue samples was validated by a pathologist and determined using the Riley score [24
], a four-point scale in which a score of 1 identifies disease remission and 2 or more indicates active disease. An important criterion leading to the choice of the Riley score in preference to other histological indices such as the Nancy score was the possibility of differentiating healthy controls (Riley score = 0) from patients in remission (Riley score = 1), in order to exclude the latter from study participation. Determinant factors of the Riley score are acute or chronic inflammatory cell infiltrates, crypt abscesses, mucin depletion, integrity of surface epithelium and crypt architectural irregularities. The CD cohort showed on average a higher Riley score (3.0) than the UC cohort (2.3). Control biopsies (9 small intestine, 12 large intestine) were collected from 12 healthy individuals. None of the control samples showed any signs of inflammation (Riley score 0; Table 1
Immunohistochemical staining was performed with formalin-fixed and paraffin-embedded sections (2 μm). Xylene deparaffinization and graded alcohol hydration steps were followed by citrate-dependent (pH6) protein antigen retrieval (Dako, Santa Clara, CA, USA). After several washes, endogenous peroxidase was blocked using 0.3% hydrogen peroxide and 0.1% sodium azide diluted in PBS. To block signals from endogenous avidin, biotin and biotin-binding proteins, we used an avidin/biotin blocking kit (Vector Laboratories, Burlingame, CA, USA). Nonspecific binding was blocked with protein block (Dako, Santa Clara, CA, USA) or 2% goat serum in antibody diluent (Dako, Santa Clara, CA, USA) at room temperature. Sections were then incubated overnight at 4 °C with the primary antibody diluted in antibody diluent (Dako, Santa Clara, CA, USA). A negative control section was included, in which only the secondary antibody (anti-mouse/anti-rabbit) was used (Figure S2
). All primary antibodies and corresponding dilutions are listed in Table S1
. After two washing cycles, an anti-mouse or anti-rabbit HRP-envision system (Nichirei Biosciences, Tokyo, Japan) and DAB-chromogen (Vector Laboratories, Burlingame, CA, USA) was applied to visualize the stainings. Tissue sections were counterstained with Meyers hematoxylin (PanReac AppliChem, Darmstadt, Germany) and mounted with Roti-Mount Aqua (CarlRoth, Karlsruhe, Germany). Staining results were documented by Olympus inverted microscopy (Olympus K. K., Tokyo, Japan). Semiquantitative interpretation of positive areas/stained cells was conducted using the BZ Analyzer Software 9000 with its specific IHC-based algorithm (Keyence, Osaka, Japan). Positive staining was defined as clear brown staining. We determined the number of positive cells in each section per biopsy and calculated the percentage of positive area per total area (for CD3, CD68, CD11b, CD11c, MPO, IDO, TDO, KYNU, Kyn and 3-HAA) or positive cells per total area (KMO and FOXP3). A corresponding example is shown in Figure S1
. Quantification was performed by a blinded independent examiner who was unaware of the diagnosis and histopathological scoring. All relevant target structures were stained in each biopsy and included in the statistics. All histology specimens were matched with corresponding samples of each of the other cohorts taken from the same bowel segment.
2.4. Primary Cell Isolation of Antigen-Presenting Cells and CD3+ T Cells
Monocytes, macrophages, dendritic cells, B cells and CD3+ T cells were isolated from Pperipheral blood mononuclear cells (PBMCs) using Ficoll-Histopaque 1077 gradients (Sigma-Aldrich, Steinheim, Germany). For the isolation of monocytes (Catalog#19058), B cells (Catalog#17954) and CD3+ T cells (Catalog#17911), negative selection was performed using EasySep Human Cell Isolation Kits (Stemcell Technologies, Vancouver, BC, Canada). Macrophages and dendritic cells were differentiated as described in [25
2.5. Cultivation and Stimulation of Antigen-Presenting Cells and CD3+ T Cells
Monocytes, macrophages, dendritic cells, B cells or CD3+ T cells were isolated from PBMC as described. Cells were seeded out at a density of 1 × 106/mL and left untreated in RPMI medium (Greiner bio-one, Frickenhausen, Germany), 1% HEPES buffer solution, 100 IU/mL penicillin and 100 IU/mL streptomycin. Cells were cultivated overnight with human AB serum (Sigma-Aldrich, Steinheim, Germany) in a final concentration of 1% (v/v) in RPMI medium. For stimulation, cells were cultivated for 24 h in serum-free RPMI medium, 100 U/mL IFNγ (PeproTech, Hamburg, Germany), 100 ng/mL TNFα (PeproTech, Hamburg, Germany) and 10 ng/mL IL-1β (PeproTech, Hamburg, Germany). All reagents were diluted in PBS/0.1% BSA.
2.6. RNA Isolation, cDNA Synthesis and Real-Time Quantitative PCR
Isolated monocytes, macrophages, dendritic cells, B cells and CD3+ T cells were pelleted, and total RNA was isolated using Isolate II RNA Micro Kit (Bioline, Heidelberg, Germany) according to the manufacturer’s instructions. Equal RNA amounts were transcribed into cDNA by reverse transcriptase with the Precision nanoScript Reverse Transcription Kit (Primerdesign, Southampton, UK), and a standard real-time (RT)-PCR program (65 °C, 5 min; 55 °C, 20 min; 75 °C, 15 min) was used. For real-time quantitative PCR (qRT-PCR) mRNA analysis, 5′FAM labeled testing probes of housekeeping genes GAPDH and RPL13A (Primer Design, Southampton, UK), IDO1 (Hs00984148_m1) and KYNU (Hs01114105_m1) were used in the complete reaction mixtures. All probes were purchased from Applied Biosystems (Foster City, CA, USA) in a final concentration of 250 nM. The qRT-PCR analyses were performed in technical duplicates with 5 µL Precision FAST 2x- qPCR Master-Mix (Primer Design, Southampton, UK), 3.5 µL H2O, 0.5 µL 5′FAM marked testing probe and 1 µL cDNA. The following program was used: 95 °C for 5 min, then 3 s at 95 °C and 30 s at 60 °C (40×), alternately. Relative mRNA expression was calculated based on the normalized ratio of nonregulated GAPDH and RPL13A expression and the 2-ΔCt method.
2.7. Statistical Analysis
Statistical analysis was performed using GraphPad Prism version 6 (GraphPad Software, La Jolla, CA, USA). For the comparison of healthy controls versus inflamed IBD samples and iCD versus cCD samples, Mann–Whitney or Kruskal–Wallis tests were applied for nonparametric distribution. Nonparametric correlation between the Riley score and/or protein/metabolite expression was tested with Spearman’s rho. For the in vitro experiments, the two-tailed Student’s t-test was applied. In all figures, asterisks (*) denote significant comparisons, with * for p ≤ 0.05, ** for p < 0.01, *** for p < 0.001 and **** for p < 0.0001. All data are presented as mean ± SD. The outlier calculator from GraphPad was used for all analyses, and significant outliers were removed.
The aim of this work was to gain a deeper insight into the relationship between mucosal immune TRP metabolism and chronic inflammatory mechanisms in patients with IBD. In particular, we focused on characterizing new biomarkers and revealing molecular mechanisms that may represent novel potential therapeutic targets in IBD. To our knowledge, we are the first group to have identified a distinct mucosal enzyme and metabolite pattern of the local TRP metabolism in patients with IBD. Interestingly, we observed differences in the immunological profile, higher protein expression levels of the Kyn-degrading enzyme KYNU and higher levels of the 3-HAA/Kyn ratio in inflamed ileum versus inflamed colon tissue of individuals with CD. These results can be used to further classify and determine the heterogeneity of CD subtypes.
Why is it important to introduce a further subclassification of CD pathologies? Recent publications support clinical observations of different phenotypes of patients with CD depending on the primary localization of inflammation within the intestine. For example, Ahmad et al. showed specific mutations of the CARD15
genes, which only and specifically occurred in the ileum of patients with CD [7
]. Naftali et al. demonstrated a correlation of enhanced presence of Fusobacterium with the disease activity of iCD by microbiotic profiling [6
]. Differences in therapeutic response have also been noted between patients with CD depending on whether inflammation was predominantly localized in the ileum or in the colon. Specifically, it was proven that infliximab, a monoclonal antibody against TNFα, was associated with a better prognosis when applied in patients with cCD versus iCD [2
Consistent with the identification of active chronic inflammation by means of the Riley score, we found an increased infiltration of immune cells in both CD and UC samples (Figure 1
and Figure S3
). Our data showed that iCD pathophysiology appears to be triggered by T-lymphocyte-driven immunity, whereas cCD tissue followed a more autoinflammatory, innate immune pattern. We determined an increase of CD3+ T cells and a decrease of FoxP3+ regulatory T cells in iCD. In contrast, cCD immunological profile analysis was characterized by a significant increase of CD68+ macrophages and MPO+ neutrophil granulocyte infiltration.
Can we then link TRP metabolism to the immunological profile of patients with CD? Data from our in vitro experiments (Figure 2
and Figure 4
) further substantiate the findings of other research groups [27
], in suggesting that processes of the TRP metabolism can be attributed to specific cells of the immune system. APCs and granulocytes (especially neutrophils and eosinophils) are equipped with TRP-metabolizing enzymes, while T cells and T-cell-specific AHR in general represent targets of these molecules. This is corroborated by the inability of CD3+ T cells to express IDO1, TDO or KYNU, as a result of which these cells cannot produce or degrade Kyn by currently described pathways [28
].We deliberately chose a single-cell analysis of blood cells, since this allows the clear isolation of subpopulations using FACS and the specific stimulation with cytokines, which are also involved in the course of Crohn’s disease. These in vitro data of our mRNA-based screening provided information about KYNU induction and thus the direct influence of inflammatory cytokines. Additionally, a clear assignment of enzymes to cell type was possible. In the overall concept, we were able to show an increased expression of KYNU on the protein level (IHC) as well as on the mRNA level (qPCR), which, we believe, further clarifies the role of KYNU and a possible modulation of this enzyme.
We utilized IHC screening to show that IDO1 expression is significantly increased in inflamed ileal tissue compared with inflamed colon tissue of patients with CD (Figure 2
B). IDO1 is the rate-limiting enzyme of the TRP metabolism and enables the relatively inert TRP to enter the Kyn metabolism with its immunomodulatory effects. In addition, IDO1 has been described as an inflammation-inducible enzyme [30
]. Thus, IDO1 can be understood as a general activity marker for inflammatory processes, as confirmed by the positive correlation between the Riley score and local IDO1 expression in UC samples (Figure S4
). Concurrently, our results revealed very low expression levels of IDO1 in healthy bowel tissue (Figure 2
B). Our data suggest that the TRP metabolism plays a far greater role in ileal than in colonic inflammation in patients with CD. Immune histological analysis also demonstrated that IDO1 and the downstream enzymes of the TRP metabolism were carried into sites of active inflammation by infiltrating immune cells. In support of this hypothesis, the majority of crypts were devoid of IDO1 protein expression. On the other hand, as we detected no significant regulation of TDO in the different intestinal segments (Figure S4
), it appears to play a minor role. This is in line with Badawy et al. [22
], who found that TDO exerted a systemic function by regulating TRP content in plasma, whereas IDO1, which is regulated by inflammatory cytokines produced by immune cells, such as IFNγ, TNFα or IL-1β, in turn modulated local TRP levels [22
We investigated mucosal tissue levels of Kyn using in situ immune histochemical stainings with specific antibodies. Surprisingly, and contrary to our expectations, these measurements failed to show elevated tissue levels of Kyn (Figure 3
B). In both UC and cCD samples, Kyn levels did not differ compared to controls, whereas Kyn levels were clearly reduced in iCD. Kyn has been reported to influence AHR activity in T and epithelial cells and ensure the maintenance of epithelial integrity through production of IL-22 and induction of responsiveness to IL-10 receptor [32
]. Kyn also directs the differentiation of IL-10-producing Treg lymphocytes [20
]. Since our analysis showed Kyn tissue levels to be decreased in patients with iCD, our data implicate that increased IDO1 expression is not automatically associated with elevated Kyn tissue levels. An IDO1/Kyn-dependent immunosuppressive milieu can thus no longer be assumed. In line with this, Treg infiltration was not significantly increased in iCD tissue compared to controls (Figure 1
We additionally analyzed the protein expression by analysis of KMO+ area and KYNU+ area as indirect markers of the further degradation of Kyn. KMO expression levels were generally low, whereas KYNU protein expression was moderate to high in all samples (Figure 4
A). Thus, we concluded that the physiological route of Kyn degradation in the gastrointestinal system might involve accumulation of AA and 3-HAA through the metabolizing enzyme KYNU. Analogously, Giorgini et al. demonstrated in murine brain tissue that the preferred route of Kyn downstream metabolism led to the accumulation of AA and thus to enzymatic catalysis by KYNU [33
]. Herein, for the first time, we determined a significantly higher expression of the Kyn-metabolizing enzyme KYNU in inflamed ileal samples from patients with CD (Figure 4
B). The specific role of KYNU in inflammation pathology has not yet been clarified in sufficient detail. However, one recent study demonstrated that KYNU is strongly expressed in psoriasis [21
]. Harden et al. hypothesized that KYNU regulates the transition between cancer and chronic inflammatory disease by modulating the concentrations of Kyn and other downstream TRP metabolites. Our results are in line with this hypothesis. Data clearly identified different metabolic signatures in iCD and cCD, with a simultaneous prominent presence of IDO1 and decreased levels of Kyn in the ileum, presumably resulting from the strong expression of KYNU in this segment (Figure 4
B). These data are corroborated by the accumulation of 3-HAA in ileal CD biopsies (Figure 5
B). Such an accumulation was observed neither in UC nor in colonic CD tissue samples (Figure 5
B and Figure S7
). The (patho-)physiological effect of 3-HAA is not yet well understood. While various authors have reported immunosuppressive effects of 3-HAA, others showed proinflammatory effects. However, the former were demonstrated by applying nonphysiologically high concentrations (up to 100 µM) of 3-HAA without subsequently assessing cytotoxicity [34
]. In the blood of healthy volunteers, 3-HAA concentrations range from 21 to 383 nM [37
]. In tissue, concentrations are expected to be even lower.
Our investigation and the derived hypothesis may be supported by measuring the ratio between 3-HAA and Kyn in future studies focusing on the differential characterization of ileal and colonic inflammation in CD. In our research, we used the 3-HAA/Kyn ratio as an indirect measurement of KYNU enzyme activity. Since this ratio addresses both the regulatory/immunosuppressive potential of Kyn and the potentially proinflammatory activity of 3-HAA, we expect it to have a greater predictive power as a CD-specific inflammatory marker than individual measurements of Kyn and 3-HAA. We found the 3-HAA/Kyn ratio to be specifically higher in inflamed iCD versus inflamed cCD tissue (Figure 5
C). While a differential diagnosis of iCD and cCD by endoscopic examination is warranted, our results set the stage to establish a systemic biomarker to identify iCD. Further studies to confirm the iCD tissue-specific Kyn/3-HAA ratio by LC-MS/MS in human blood and stool samples are underway. IHC techniques allow an enzyme to be assigned to the immune cell infiltrate parallel to quantitative analysis and other pathological investigations within a single bioptic sample. Common LC-MS/MS- or qPCR-based analyses, on the other hand, do not permit the investigation of single cells and thus fail to take account of the influence of different cell types. In addition, since these methods usually expend an entire biopsy, further analyses necessitate an additional sample. For these reasons, we propose IHC to be the most effective method of investigating mucosal TRP immune metabolism.