Activated kynurenine pathway metabolism by YKL-40 establishes an inhibitory immune microenvironment and drives glioblastoma development

Abstract

Background

Glioblastoma (GB) is the stage IV of glioma and mesenchymal GB represents the most common and malignant subtype characterized with elevated expression of a mesenchymal marker YKL-40 and resistance to immune drug therapy. Here, we determined if YKL-40 regulates kynurenine (Kyn) pathway (KP) metabolism that contributes to establishing an immune suppressive microenvironment in GB.

Methods

Tumor cells expressing YKL-40 from GB patients were isolated and activated cellular metabolisms were identified via gene microarray analysis. KP metabolism was determined by LC/MS/MS system. Indoleamine 2,3-dioxygenase 1 (IDO1), tryptophan 2,3-dioxygenase (TDO2), their regulatory transcription factors AhR and SRF were evaluated using WB. AhR and SRF transactivity was measured by luciferase reporter gene assays with binding motif mutation, while m6A-mediated AhR and SRF mRNA stability was determined in the presence of an METTL3inhibitor. YKL-40 and Kyn-induced tumor cell migration and CD8+ cytotoxic T cell (CTL) apoptosis were measured in cultured cells. Tumors cells expressing YKL-40 were injected to mouse brains to establish orthotpic tumor models. In GB, YKL-40, IDO1 and TDO2 expression was analyzed for correlation with patient survival.

Results

KP metabolism was activated in YKL-40-expressing tumor cells. YKL-40 divergently regulated IDO1 and TDO2 via induction of AhR and SRF, respectively. mRNA levels of AhR and SRF were stabilized by decreased METTL3 and YTHDF2. YKL-40 and Kyn secreted from tumor cells and infiltrating M2 macrophages cooperated to enhance tumor cell migration and inhibit CTL immunity. In xenografts, tumors expressing YKL-40 displayed the elevated KP metabolism and macrophage infiltration, but decreased CTLs. Treatment with an anti-PD-1 antibody Tislelizumab significantly increased YKL-40+ mouse survival. In GB, YKL-40 was positively correlated with IDO1 expression and both were associated with decreased survival, whereas IDO1 was negatively correlated with TDO2.

Conclusion

YKL-40 upregulates IDO1 or TDO2 to activate KP metabolism, and coordinates with Kyn to establish an inhibitory tumor immune microenvironment, leading to tumor immune evasion.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-024-05497-5.

Introduction

Glioblastoma (GB) is the stage IV form of malignant glioma with the median survival rate between 15 and 20 months [1]. The current therapeutic approaches to treat GB mainly include surgery, radiotherapy, chemotherapy, targeted therapy and immunotherapy [2–4]. However, all of these regimens fail to effectively improve patient survival, as the poor prognosis of this disease remains static. It is of note that the mesenchymal glioblastoma, the most common and malignant subtype of GB demonstrates vascularization, invasiveness and resistance to varied therapies including immunotherapy. Hence, revealing molecular mechanisms underlying the malignant transformation and offering novel targets particular for mesenchymal GB are of paramount importance in treatment of this deadly disease.

YKL-40 coded as Chitinase 3 like 1 (CHI3L1), is a secreted glycoprotein with molecular mass of 40 kDa [5]. YKL-40 is highly expressed in multiple types of tumors, including breast cancer, gallbladder cancer and GB [6–8]. Research labs from J. Elias and T. Kessler have reported that elevated serum level of YKL-40mediate tumor angiogenesis, invasion and epithelial-mesenchymal transition (EMT), serving as a mesenchymal marker for the malignancy of mesenchymal GB [9, 10]. Recent research has largely focused on its immune suppressive signature potential for the regulation of tumor immune escape [11, 12]. For example, Tailfour et al. found that YKL-40 stimulated infiltration of neutrophils to develop neutrophil extracellular traps (NETs), thus inhibiting cytotoxic CD8+ T cells (CTLs) in breast cancer [13]. YKL-40 secreted from M2 macrophages cooperated with GDF-15 to inhibit CTLs and promote the progression of gallbladder cancer [14]. In GB, YKL-40 induced the expression of PD-L1 in vascular endothelial cells, which inhibited CTLs [15]. Therefore, these compelling evidence has implicated the novel anti-tumor immunity of YKL-40 in the tumor development.

Tryptophan (Trp) is an essential amino acid which is catalyzed to kynurenine (Kyn)by rate-limiting enzymes indoleamine 2,3-dioxygenase 1 (IDO1) and/or tryptophan 2,3-dioxygenase (TDO2) [16, 17]. Kyn can be metabolized into intermediate products including 3-hydroxykynurine by kynurenine 3-monooxygenase (KMO), anthranillic acid by kynureninase (KYNU) and kynurenic acid by kynurenine aminotransferase (KAT) [18, 19]. 3-hydroxykynurine is catalyzed by KYNU to 3-hydroxyanthranilic acid and subsequently oxidized to 2-amino-3-carboxynuconic-semialdehyde by 3-hydroxyanthranilate 3, 4-dioxygenase (3-HAAO), then yielding quinolinic acid in a non-enzymatic conversion. In the presence of quinolinic acid phosphoribosyl transferase (QPRT), quinolinic acid is further metabolized and finally converted to nicotinamide adenine dinucleotide (NAD+), an essential enzyme co-factor involved in a variety of fundamental physiologic processes including glycolysis, energy metabolism and DNA nucleotide synthesis [20, 21]. A number of clinical evidence has unveiled that increased tumor expression of IDO1 is associated with poor overall survival (OS) in GB [22, 23]. Elevated levels of IDO1 in tumor cells mediate recruitment of regulatory T cells into tumor and/or conversion of naïve T cells into Tregs. Increased Kyn by IDO1 also promotes apoptosis of CTLs [24, 25]. Aryl hydrocarbon receptor (AhR), as the natural ligand of Kyn, displays the ability to induce the conversion of CTLs to exhausted CD8+ cells [26–28]. Given increased Kyn or decreased Trp levels in blood, the Kyn/Trp ratio in clinic studies has been established as a valuable predictor for immunotherapeutic efficacy in GB [29]. However, it remains to be investigated if serum Kyn/Trp ratio represents the bona fide overall KP metabolism of tumor microenvironment (TME), because a number of cell types, not limited to tumor cells, such as tumor-associated macrophages (TAMs) and DCs also contribute to the regulation of the KP metabolism in GB [30]. Furthermore, we still lack knowledge about how the TME establishes an inhibitory immune scenario to assist tumor immune escape via upregulation of YKL-40-driven KP metabolism.

In this study, we first sought to determine if YKL-40 regulates the KP metabolism. In particular, we explored the distinct regulatory mechanisms of YKL-40 for IDO1 and TDO2 expression, in which METTL3 and YTHDF2epigenetically regulate AhR and SRF expression and subsequently controls IDO1 and TDO2 expression, respectively. Then, we found that YKL-40 and Kyn derived from tumor cells and macrophages coordinately inhibited CTL immunity. Understanding of the activated KP metabolism in the TME including tumor cells, TAMs and CTLs will offer mechanistic insights into the YKL-40-mediated immune evasion in GB, ultimately improving the therapeutic strategy with IDO1/TDO2-targeting regimens tailored with immune checkpoint blockers.

Results

YKL-40 was correlated with Kyn pathway metabolism in patients with GB

Given that the mesenchymal subset of GB constitutes the most common and most malignant type of gliomas, we particularly focused on the potentially mechanistic link between the malignant signature and mesenchymal genotype characterized with expression of mesenchymal markers YKL-40 and vimentin. Tumor tissues resected from patients with mesenchymal GB were processed for single cell clone culture in vitro. From nine single cell clones, we selected single clones expressing YKL-40^High and YKL-40^Low for RNAseq microarray analysis (Fig. 1A). KEGG pathway enrichment showed that metabolic pathways were ranked as the top ones in 82 enriched pathways in YKL-40^High relative to YKL-40^Low clones (Fig. 1B left, Table S1). We then sought to define the types of aberrant metabolic pathways and kynurenine pathway (KP) emerged as the activated one (Fig. 1B right, Table S2). InIDO1 and TDO2 levels, the rate-limiting enzymes to control the KP, we found that IDO1 was upregulated together with other enzymes (QPRT, KMO, HAAO) that participate in the regulation of following metabolism of Kyn; unexpectedly, TDO2 was downregulated (Fig. 1C). These protein expression levels were further validated (Fig. 1D) and all other single clones involving YKL-40^High andYKL-40^Low resembled this protein expression signature, suggesting that different expression levels of YKL-40 in GB mediate deregulation of KP.

Fig. 1.

YKL-40 was correlated with Kyn pathway metabolism in patients with GB. (A) Single cell clones with YKL-40^High and YKL-40^Low from patients with GB were grown in culture and YKL-40 in serum-free medium and actin in cell lysates were measured using western blotting. (B) RNAs of these YKL-40^High and YKL-40^Low cell clones were subjected to RNAseq gene microarray analysis. The KEGG pathway enrichment analysis was engaged to define the top 10 ranked pathways (left) from 82 pathways of DEGs that involve 4 enriched metabolism pathways (right). (C) Kynurenine pathway metabolism (left) and differential gene expression of enzymes involved in the KP (right) in the RNAseq analysis. (D) Representatives of all cell clones expressing YKL-40^High and YKL-40^Low for expression of YKL-40, IDO1 and TDO2. (E) GSE51602 from TCGA (https://portal.gdc.cancer.gov/) showed differential gene expression profile of GB in volcano plot, where 3729 DEGs were down-regulated and 2449 DEGs were up-regulated. YKL-40 and genes related to KP were shown. (F) GEPIA 2 (http://gepia2.cancer-pku.cn/) showed different expression of CHI3L1, IDO1 or TDO2 between glioma and adjacent benign tissue in mesenchymal GB. (G)The mRNAseq_325 dataset from CGGA database showed correlation of CHI3L1 with either IDO1 or TDO2, and correlation of IDO1 with TDO2 in primary glioma of all WHO grade, n = 325. (H) The mRNAseq_325 dataset from CGGA database showed patient survival of primary glioma of all WHO grade between high and low expression of CHI3L1, IDO1 and TDO2.n = 116 vs. n = 114. (I) Tissues from 44 cases of GB were used for Immunohistochemistry (IHC) of YKL-40, IDO1 and TDO2. Representatives of positive and negative staining of YKL-40, IDO1 and TDO2. Inserts: amplified cells. Scale bar, 50 μm. (J) Eight different classes of GB cases expressing YKL-40, IDO1 or/and TDO2. Forty-four tissue samples of GB (n = 44) were engaged for IHC study of YKL-40, IDO1 and TDO2. The staining data analyses determined 8 classes with positive and negative staining feature as shown in the bottom cases.(K) Correlation between YKL-40, IDO1 and TDO2 in GB cases. n = 35. The red line represented a mean value*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

We subsequently searched the database GSE51602 of the different expression genes (DEGs) from TCGA database (https://portal.gdc.cancer.gov/) and identified that CHI3L1 and IDO1 were highly expressed in brain tumor, while TDO2 remained unchanged (Fig. 1E). The GEPIA 2 database (http://gepia2.cancer-pku.cn/) showed that CHI3L1, IDO1 and TDO2 in mesenchymal GB were significantly highly elevated relative to those in adjacent benign tissue (Fig. 1F). The mRNAseq_325 dataset from CGGA database (http://www.cgga.org.cn) analysis also unveiled that CHI3L1 was positively correlated with IDO1 or TDO2, and IDO1 was positively correlated with TDO2in primary glioma of all WHO grades (Fig. 1G). In addition, these patients with higher levels of CHI3L1, IDO1 and TDO2 experienced poorer survival than patients with lower levels of these genes (Fig. 1H). These public data analyses supported our findings that expression of YKL-40 was positively correlated with expression of IDO1 or TDO2 in brain tumor, and increased IDO1/TDO2-mediated KP may be associated with malignant transformation. IHC staining from our 44 GB cases showed that most of high YKL-40 expression cases (≥ 3 points, 70% population) were IDO1-positive and TDO2-negative, while most tumors with low YKL-40 levels did not express IDO1 or TDO2 (Tables 1 and 2; Fig. 1I and J), and IDO1-positive cases showed significantly poorer disease free survival. The correlation analyses exhibited that YKL-40 was significantly correlated with IDO1 (Fig. 1K), whereas YKL-40 was not correlated with TDO2. Interestingly, IDO1 was negatively associated with TDO2. Taken together, these results suggest that elevated expression of YKL-40 is positively correlated with expression of either IDO1 or TDO2 that regulates KP during tumor progression.

Table 1.

Clinical pathophysiologic factors of patients with glioblastoma

No.	Age	Sex	Recurrence (yes/no)	Disease-free survival (months)	Overall survival (months)	YKL-40a	IDO1b	TDO2b
1	43	male	yes	NDc	ND	5.5	+	+
2	72	male	yes	2	≥ 4	5	+	+
3	59	male	yes	8	≥ 13	5	+	-
4	52	male	yes	11	≥ 11	5	+	-
5	47	male	yes	10	≥ 20	5	+	+
6	46	female	yes	10	≥ 20	5	+	-
7	48	female	no	ND	≥ 10	5	+	+
8	71	female	yes	3.5	≥ 3.5	5	-	+
9	69	male	yes	5	17	5	+	-
10	51	male	yes	5	≥ 8	5	+	-
11	53	female	yes	9	≥ 24	5	+	-
12	47	male	yes	8	≥ 8	4.5	+	-
13	60	male	ND	1	1	4.5	-	-
14	56	female	yes	4	≥ 4	4	+	+
15	52	male	yes	8	≥ 18	4	+	-
16	44	male	yes	3	≥ 4	4	-	-
17	51	male	yes	2	5	4	+	-
18	46	male	yes	3	ND	4	+	-
19	62	female	yes	3	≥ 4	3.5	+	-
20	62	female	yes	3	≥ 12	3.5	+	-
21	69	female	yes	6	≥ 15	3.5	+	+
22	33	male	yes	41	≥ 41	3.5	-	+
23	62	male	no	ND	≥ 2	3.5	-	-
24	65	male	ND	ND	ND	3.5	-	-
25	40	male	yes	4	≥ 5	3.5	+	-
26	66	male	no	ND	≥ 10	3	+	+
27	56	male	no	ND	ND	3	-	+
28	57	female	yes	7	≥ 7	3	+	-
29	60	male	yes	9	≥ 10	2.5	-	+
30	53	male	yes	32	≥ 41	2.5	-	-
31	35	male	no	≥ 48	≥ 48	2.5	-	-
32	52	male	yes	6	≥ 11	2	+	+
33	40	female	no	≥ 19	≥ 19	2	+	-
34	78	male	yes	1	≥ 4	2	-	+
35	35	female	no	ND	ND	2	-	-
36	51	female	no	ND	ND	1	-	-
37	70	male	ND	ND	ND	1	+	+
38	62	female	ND	ND	ND	1	-	-
39	53	female	yes	ND	ND	1	-	-
40	76	male	yes	5	≥ 5	1	-	-
41	51	male	ND	ND	ND	2.5	+	-
42	69	male	yes	ND	ND	4	-	-
43	67	male	ND	ND	ND	4	+	-
44	66	male	ND	ND	ND	4	+	-

^aYKL-40 value from IHC staining was determined by a semi-quantification method with a range from 0–6 points. ^bIDO1 or TDO2 from IHC was evaluated by positive or negative staining. ^cND=Not determined

Table 2.

Association of IDO1 and TDO2 with clinical outcomes

Characteristics	IDO1		P valuea	TDO2		P valueb
	+	-		+	-
Recurrence			0.316			0.402
YES	20 (71.4%)	9 (31.0%)		10 (35.7%)	19 (65.5%)
NO	3 (37.5%)	5 (62.5%)		3 (37.5%)	5 (62.5%)
Disease-free survival (months)			0.041*			0.91
≤ 14	19 (76.0%)	6 (24.0%)		8 (32.0%)	17 (68.0%)
> 14	1 (25.0%)	3 (75.0%)		1 (25.0%)	3 (75.0%)
Overall survival (months)			0.353			0.732
≤ 14	14 (66.7%)	7 (33.3%)		8 (38.1%)	13 (61.9%)
> 14	7 (70.0%)	3 (30.0%)		3 (30.0%)	7 (70.0%)
YKL-40 score			0.02*			0.752
≥ 3	21 (75.0%)	7 (25.0%)		10 (35.7%)	18 (64.3%)
< 3	6 (37.5%)	10 (62.5%)		4 (25.0%)	12 (75.0%)

^aComparison between IDO1 (+) and IDO1 (-)

^bComparison between TDO2 (+) and TDO2 (-).*p < 0.05

YKL-40 regulated KP metabolism in GB

To determine the role of YKL-40 in regulating the KP, we employed our previously established YKL-40 gene knockdown glioblastoma cell line shYKL-40 GSDC and control cells [31] and a commercial available cell line U87. When CHI3L1 was knocked down, the protein level of IDO1 was decreased, but TDO2 was increased in both cell lines (Fig. 2A), agreed with the cell clone data (Fig. 1D). In GSDCs, Kyn in culture medium was decreased, and the Kyn/Trp ratio which commonly represents the activity of the key metabolic enzymes, was also decreased (Fig. 2B), suggesting that YKL-40 regulates IDO1 to control KP metabolism. Since KP metabolism determines NAD+ synthesis, the core intermediator that participates in the TCA cycle, pentose phosphate pathway and/or nucleotide synthesis, we measured the relevant metabolites in these pathways. In shYKL-40 GSDCs, NAD+ was increased and AMP, GMP, CMP, IMP and Ado in nucleotide synthesis pathway were simultaneously increased; in contrast, succinic acid in TCA cycle was decreased (Fig. 2C), suggesting that the increased TDO2 in shYKL-40 cells may render NAD+ preferential to nucleotide synthesis metabolism. The results indicate that YKL-40 induces IDO1 to control KP and TCA cycle metabolism; whereas suppression of YKL-40 results in induction of TDO2 and consequent nucleotide synthesis metabolism (Fig. 2D).

Fig. 2.

YKL-40 regulated KP metabolism. A Control and shYKL-40 GSDCs and U87 cells were lysed to detect YKL-40, IDO1 and TDO2 expression by immunoblotting. (B) Cultured serum-free medium from control and shYKL-40 GSDCs was collected and measured for Kyn, Trp and Kyn/Trp ratio via LC/MS/MS, n = 3. (C) Cell lysates from control and shYKL-40 GSDCs were tested for relative metabolite levels of NAD+, succinic acid and nucleotides (AMP, GMP, CMP, IMP, Ado) via LC/MS/MS described in Materials and Methods, in which control levels of these metabolites were set as 1 unit. n = 4. (D) A schematic diagram of the regulation of KP metabolism in control GSDCs that predominate Kyn and following TCA metabolism pathway. (E) Control and OE YKL-40 GL261 cells were used to test protein expression level of YKL-40, TDO2 and IDO1 via Western blotting. (F) Cultured serum-free medium from control and OE YKL-40 GL261 cells was collected to test concentration of Kyn, Trp and Kyn/Trp ratio via LC/MS/MS. The cell density was ~ 90% same as GSDCs in 6 wells plates, but GL261 cell number was at least 2 ~ 3 times higher. n = 3. (G) Cell lysates from control and OE YKL-40 GL261 cells were used to measure relative metabolite levels of NAD+, succinic acid and nucleotides (GMP, CMP, Ado) via LC/MS/MS described in Materials and Methods, in which control levels were set as 1 unit. n = 4. (H) A schematic diagram of the regulation of KP metabolism in OE YKL-40 GL261 cells that drive Kyn and downstream both TCA and nucleotide synthesis pathways.(I) GSDCs were treated with IDO1 inhibitor (Epacadostat, 20 µM), TDO2 inhibitor (680C91, 20 µM) or both for 24 h. Serum-free medium was collected to test relative levels of Kyn, Trp and Kyn/Trp ratio via LC/MS/MS, in which control levels were set as 1 unit. n = 3. (J) GSDCs were treated with Epacadostat for 24 h. (K) Control and shIDO1 GSDCs were measured for expression of IDO1. (L) Serum-free medium from control and shIDO1 GSDCs was used to determine relative levels of Kyn, Trp and Kyn/Trp ratio, in which control levels were set as 1 unit. n = 3. Data were presented as mean ± S.E.M. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

To validate YKL-40 activity in the KP metabolism, we utilized murine brain tumor cell line GL261 that doesn’t express YKL-40. After stably enforced to express YKL-40, OE YKL-40 GL261 cells significantly increased to express TDO2, but not IDO1 (Fig. 2E). As expected, Kyn and Kyn/Trp ratio were increased in OE YKL-40 GL261 cells compared with control cells (Fig. 2F). In line with the TDO2-mediated nucleotide synthesis in GSDCs (Fig. 2D), NAD+, Ado, GMP and CMP were all increased in OE YKL-40 GL261 cells in addition to increased succinic acid (Fig. 2G), indicating that YKL-40 stimulates KP and downstream both TCA cycle and nucleotide synthesis metabolism in GL261 cells that only upregulate TDO2 to control the KP metabolism (Fig. 2H).To further confirm this finding, we treated GL261 cells with TDO2 inhibitor 680C91 and found that it decreased Kyn/Trp and succinic acid; but increased NAD+ and CMP. The data suggest that TDO2induces conversion of Trp to Kyn and downstream TCA metabolism. The increased NAD+ and subsequent nucleotide metabolism may be ascribed to other catalytic enzymes (except IDO1) activated to mediate nucleotide synthesis by the inhibitor in GL261 cells (Fig. S1A, B). To evaluate the enzymatic activity of IDO1 or TDO2 in the regulation of KP in GSDC, we introduced both IDO1 inhibitor Epacadostat and 680C91. Treatment with either Epacadostat or 680C91 led to decreases in Kyn and Kyn/Trp ratio (Fig. 2I), and combination displayed the synergistic inhibition of Kyn and Kyn/Trp ratio, suggesting that both IDO1 and TDO2 in GSDCs may contribute to the KP metabolism. Epacadostat inhibited Kyn and Kyn/Trp ratio in a dose-dependent manner(Fig. 2J). Furthermore, when IDO1 was knocked down in the cells (Fig. 2K), Kyn and Kyn/Trp ratio were decreased (Fig. 2L), validating the primary activity of IDO1 inthe KP metabolism. Altogether, these findings demonstrate that elevated expression of YKL-40 upregulates KP metabolism in GB via either IDO1 or TDO2.

YKL-40 differentially regulated IDO1/TDO2 via AhR and SRF in GB

The altered expression of IDO1 and/or TDO2 by YKL-40 had encouraged us to explore potential molecular mechanisms. AhR, the natural ligand of Kyn that acts as a transcription factor to induce multiple targeted genes expression, was first engaged to determine if YKL-40 regulates its expression. When CHI3L1 was knocked down in GSDCs and U87 cells, AhR was abrogated. In contrast, CHI3L1 was overexpressed in GL261 cells, AhR was up-regulated (Fig. 3A), suggesting that AhR may play an essential role in YKL-40-induced IDO1 and TDO2 expression in both cells. In an attempt to interrogate how YKL-40 oppositely regulates IDO1 and TDO2 seen in GSDCs and U87 cells, we used the transcription factor database (http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3) crossing with earlier RNAseq microarray dataset and we identified two candidates SRF and ELF1 in TDO2-regulated and YKL-40-downregulated cells. Indeed, SRF protein level was increased in shYKL-40 GSDCs that express decreased AhR (Fig. 3B). This finding was validated in shYKL-40 U87 cells, but not in OE YKL-40 GL261 cells (Fig. S2). To further confirm that AhR regulates IDO1 in GSDCs and TDO2 in GL261 cells, its localization and function on IDO1 and TDO2 were detected. Nuclear localization of AhR as a transcription factor was diminished in shYKL-40 GSDCs relative to control GSDCs; but was induced in OE YKL-40 GL261 cells compared with control cells (Fig. 3C). In GSDCs, AhR was knocked down and IDO1 not SRF was correspondingly decreased (Fig. 3D). The transcript activity of AhR for IDO1 and TDO2 gene regulation in GSDC and GL261 cells, respectively, was determined using the AhR-binding motif-driven luciferase gene reporter assay (Fig. 3E). Consistent with the protein changes (Fig. 3A), the transactivity of IDO1 driven by AhR, not TDO2, in shYKL-40 GSDCs was decreased by 69.9% relative to control cells, whereas the activity of TDO2, not IDO1, in OE YKL-40 GL261 cells was induced (Fig. 3E). Collectively, the data indicate that YKL-40 regulates AhR to govern expression of IDO1 in GSDCs and TDO2 in GL261 cells, while silence of CHI3L1 gene in GSDCs results in opposite upregulation of SRF that triggers TDO2 expression.

Fig. 3.

YKL-40 differentially regulated IDO1/TDO2 via AhR and SRF. (A) Control and shYKL-40 GSDCs and U87 cells, and control and OE YKL-40 GL261 cells were employed to test AhR expression. (B). Database crossing transcription factor dataset (http://alggen.lsi.upc.es/cgibin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3) were used to seek the common candidates that downregulate TDO2 gene expression. Control and shYKL-40 GSDCs were used to analyze SRF level in Western Blotting.(C) Control and shYKL-40 GSDCs were analyzed for immunofluorescence staining with an anti-AhR antibody. Nuclear immunofluorescence level was quantified in which nucleus was stained with DAPI (blue). n = 3. Scale bar, 100 μm. (D) Control and shAhR GSDCs were measured for AhR, IDO1 and SRF expression. (E) Scheme of AhR-driven transcription activity of IDO1andTDO2 in GSDCs. A dual luciferase report plasmid with AhR binding site of IDO1andTDO2 promoter was transfected to GSDCs or GL261 cells for 48 h and then dual luciferase gene reporter system was tested for the relative luciferase activity of IDO1 and TDO2 in control and shYKL-40 GSDCs, and IDO1 and TDO2 in control and OE YKL-40 GL261 cells, respectively. The luciferase activity result with 124,225 in control GSDCs and 1,418,923 in control GL261 cells were set as 1 unit, respectively. n = 3. Data were presented as mean ± S.E.M. ns = no significance, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

YKL-40 regulated AhR and SRF in an m6A methylation manner

Epigenetic abnormality with m6A methylation modification of RNA is currently appreciated to participate in the development of malignant tumors [32]. To determine whether the m6A methylation modification contributes to AhR and/or SRF regulation, we were particularly interested in the transcript differentiation profile of RNA m6A methyltransferases (METTLs) and m6A readers (YTHDFs) for RNA degradation in the earlier established RNAseq microarray dataset with YKL-40^High and YKL-40^Low clones. As demonstrated in Fig. 4A, most of RNA m6A methyltransferases including METTL3, METTL14, METTL16 etc. and m6A readers includingYTHDF3, YTHDF2 and YTHDF1were down-regulated in YKL-40^High cells relative to YKL-40^Low. As the most important one, METTL3 level was negatively correlated with CHI3L1 and AhR in primary glioma, but positively correlated with SRF (Fig. 4B), analogous with our data (Fig. 3C and D). YTHDF2 resembled METTL3, displaying the similar negative correlation with CHI3L1 and AhR. Accordingly, survival data showed that patients with high expression of CHI3L1 gave rise to decreased disease-free survival compared with low levels of CHI3L1; in contrast, those with high expression of METTL3 were better prognostic than cases with low levels of METTL3 (Fig. 4C). The total m6A methylation was up-regulated in shYKL-40 GSDCs relative to controls, while down-regulated in OE YKL-40 GL261 cells (Fig. 4D& Fig. S3A). Likewise, the expression of METTL3 and YTHDF2, not YTHDF3,was consistent with the m6A expression pattern in GSDC and GL261 cells (Fig. 4E& Fig. S3B, C). These results suggest that YKL-40 may inhibit METTL3 and YTHDF2 to upregulate AhR and downregulate SRF divergently.

Fig. 4.

YKL-40 regulated AhR and SRF in an m6A methylation manner. A Differential gene expression profile for RNA m6A methyltransferases and readers from RNA-seq microarray database of cell clone YKL-40^High and YKL-40^Low. A majority of m6A methyltransferases and YTHDF1-3 were upregulated in YKL-40^Low cell clones relative to YKL-40^High cell clones that upregulated a small populations of the enzymes. (B) Correlation between METTL3/YTHDF2 and CHI3L1, AhR or SRF in GB (GEPIA 2 database) n = 163. (C) Patient survival of GB between high and low expression of CHI3L1 and METTL3 (CGGA database), n = 131 vs. n = 131. (D) Control and shYKL-40 GSDCs and control and OE YKL-40 GL261 cells were used to test total m6A levels via dot blot assay. The membrane was stained with methylene blue (MB). (E) Control and shYKL-40 GSDCs and control and OE YKL-40 GL261 cells were measured for expression level of METTL3 and YTHDF2, 3 via Western blotting. (F) Control GSDCs were treated with STM2457 (40 µM) and actinomycin D (5 µg/mL) and total RNAs were collected at 0, 8, 12, 24 and 36 h. mRNA levels of AhR and SRF were detected by RT-qPCR n = 3. (G) Distribution of m6A motifs in AhR gene (cuilab.cn/sramp). The CDS began at a blue arrow and 3’UTR at a red arrow. (H) Constructs of wild type and mutants of m6A sites in 3’ UTR AhR. The luciferase reporter gene activity was measured in the cells transfected with wild type or mutant constructs in the presence or absence of STM2457 (40 µM), in which control levels were set as 1 unit. n = 3. (I) Control or shYKL-40 GSDCs were treated with or without STM2457 (40 µM) for 24 h. Cell lysates were subjected to WB of AhR, IDO1, SRF, TDO2 and actin. (J) A scheme of METTL3-induced transcription activity of AhR or SRF for IDO1 or TDO2, respectively, in GSDCs. A dual luciferase report plasmid with AhR or SRF binding site of IDO1 or TDO2 promoter (wild type and mutants) was transfected to the cells in the presence of STM2457 (40 µM) for 48 h, then a dual luciferase reporter system was used to test the relative luciferase activity of IDO1 and TDO2, in which AhR-IDO1 activity with 3,204,699 and SRF-TDO2 with 138,736 in wild type control levels were set as 1 unit, respectively. n = 3. (K) Control and OE YKL-40 GL261 cells were treated with or without STM2457 (40 µM) for 24 h and cell lysates were collected for AhR, TDO2 and IDO1 expression via Western blotting. (L) Control GSDCs were treated with STM2457 (40 µM) and actinomycin D (5 µg/mL), and total RNAs were collected at 0, 8, 12 or 24 h, and then IDO1 mRNA level was detected by RT-qPCR. n = 3. Data were presented as mean ± S.E.M. ns = no significance, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

To validate that YKL-40 regulates METTL3 to subsequently control AhR and SRF gene methylation modification and RNA degradation, we exploited an METTL3 inhibitor STM2457 in GSDCs. STM2457 treatment increased mRNA stability of AhR and SRF(Fig. 4F). To further determine if 3’UTR m6A of the gene is accounted for their RNA stability, we sought to locate 3’UTR region rich in m6A methylation motifs RRACH (R = A/G, H = A/C/U). As shown in Fig. 4G, the m6A motifs were enriched in the beginning of 3’UTR in addition to some m6A sequences distributed in the coding sequence of AhR (cuilab.cn/sramp). We then created its consensus motif mutant 1 (5’-aaGct-3’) and mutant 2 (5’-aaGcc-3’) in the 3’UTR (Fig. 4H). In the absence of STM2457, mutant 1 and 2 protected these adenosine sites from methylation, thus stabilizing AhR mRNA by 1.2 and 3.5-fold higher than wild type mRNA. However, in the presence of STM2457, demethylated AhR wild type mRNA was 2.5-fold greater than wild type (methylated) mRNA in the absence of STM2457 (Fig. 4H). When the wild type was replaced with mutant 1 or 2, the elevated m6A-dependent mRNA stability was largely impaired in the presence of STM2457. Agreed with mRNA stability, protein level of AhR was notably induced by STM2457 in a concentration-dependent manner (Fig. S3D). These results suggest that YKL-40 inhibits METTL3 to prevent m6A methylation of AhR 3’UTR and subsequently stabilize AhR mRNA and protein expression.

We then monitored AhR and SRF transactivityforIDO1or TDO2 expression in GSDCs or GL261 cells, Treatment of GSDCs with STM2457 induced AhR, but interestingly inhibited IDO1 compared with controls (Fig. 4I). STM2457 increased SRF and slightly TDO2. In shYKL-40 GSDCs contrary to control GSDCs, the elevated level of METTL3 owe to decreased YKL-40 led to inhibition of AhR and IDO1, and induction of SRF and TDO2 (Fig. 4I). Treatment with STM2457 failed to induce AhR and SRF in shYKL-40 GSDCs relative to cells in the absence of STM2457. However, STM2457 displayed the strong ability to suppress IDO1 and induce TDO2.As transcription factors, AhR and SRF induced by STM2457 promoted IDO1 and TDO2promoter-drivenreporter gene expression by 1.58 and 1.35-fold higher than controls, respectively (Fig. 4J). However, mutation of AhR and SRF binding motifs (5’-cacgca-3’ to 5’-GGGgca-3’ and5’-acataaaaggcacag-3’ to 5’-acaGGGaaggcag-3’, respectively) resulted in failure of gene transactivation, indicating that AhR and SRF drive IDO1 and TDO2 gene expression in GSDCs. In GL261 cells, STM2457 increased AhR and thus induced TDO2, but not IDO1 (Fig. 4K). In OE YKL-40 GL261 cells, like previous findings (Figs. 3A and C and 4E), decreased METTL3 resulted in increased AhR and TDO2(Fig. 4K). The additional treatment with STM2457 substantially enhanced YKL-40 to induce both protein levels. Interestingly, we also found that STM2457 increased the degradation of mRNA of IDO1 (Fig. 4L), which may act as a dominant pathway to directly regulate the expression of IDO1 regardless of increased gene transactivation by AhR, providing the explanation of decreased IDO1 by STM2457 in both control and shYKL-40 cells (Fig. 4I). In sum, all the data deciphered the molecular mechanisms underlying YKL-40-induced IDO1 and/or TDO2 expression that contributes to aberrant KP metabolism in GB.

Role of infiltrating M2 macrophages and CD8 + cytotoxic T cells in KP metabolism

A wealth of research evidence has increasingly reported that TME acts as an indispensable partner to facilitate tumor cell aggressiveness [30]. To evaluate the key immune components including infiltrating M2 macrophages (TAMs) and tumor cytotoxic CD8+ T lymphocytes (CTLs) in the regulation of KP metabolism, we utilized a human macrophage line THP-1 and differentiated them into M2 type macrophages. As shown in Fig. 5A, TDO2 protein expression levels were increased from THP-1, M0, M1 to M2 macrophages, while YKL-40 was only subtly elevated by M2 macrophages and IDO1 was not detectable (see below). In concert with TDO2 increased levels, Kyn level and Kyn/Trp ratio were also elevated from THP-1, M0 to M2 macrophages (Fig. 5B). To further compare the KP metabolism between tumor cells and M2 macrophages, we selected M2 macrophages, GSDCs and a breast cancer line HCC1395, all of which express YKL-40 (Fig. 5C). Comparing with GSDCs that mainly express IDO1 and HCC-1395 that express TDO2, M2 macrophages showed stronger expression of YKL-40 than any one of the other two lines, while TDO2 was higher than that of HCC1395 cells and IDO1 was minimal. Of note, the Kyn level and Kyn/Trp ratio were 5.6-fold and 156.2-fold higher than those in GSDCs (Fig. 5D). When a TDO2 inhibitor 680C91 was added to M2 macrophages, the Kyn level and Kyn/Trp ratio were dramatically suppressed (Fig. 5E). Interestingly, knockdown of YKL-40 gene in M2 macrophages failed to alter TDO2 expression, suggesting that the TDO2-mediated KP metabolism is independent of YKL-40 (Fig. 5F). Although the Kyn and Kyn/Trp ratio in M2 macrophages were notably high, the degree of macrophage infiltration was various in individual tumors from minimal to extensive levels in GB tissues (Fig. 5G), suggestive of varied contributions of TAMs to the KP metabolism in different cases.

Fig. 5.

KP metabolism in M2 macrophages, tumor cells and CD8+ cytotoxic T cells. (A) M0, M1 and M2 macrophages were differentiated from THP-1 in cultures described in Methods. Then, THP-1, M0, M1 and M2 macrophages were tested for YKL-40 and TDO2 expression by immunoblotting. (B) Cultured serum-free media from THP-1, M0 and M2 macrophages were collected to test concentration of Kyn, Trp and Kyn/Trp ratio via LC/MS/MS n = 3. (C) M2 macrophages, GSDCs and breast cancer cells (HCC-1395) were used to determine expression of YKL-40, IDO1 and TDO2 in Western blotting. (D) Cultured serum-free media from M2 macrophages, HCC-1395 cells and GSDCs were used to measure relative concentrations of Kyn, Trp and Kyn/Trp ratio via LC/MS/MS, in which M2 levels were set as 1 unit. n = 3. (E) M2 macrophages were treated with or without TDO2 inhibitor and then relative levels of Kyn, Trp and Kyn/Trp ratio were detected by LC/MS/MS, in which control levels were set as 1 unit. n = 3. (F) M2 macrophages were subjected to YKL-40 gene knockdown and expression of YKL-40 and TDO2 were evaluated using Western blotting. (G) Representative infiltrating M2 macrophages in GB were observed in IHC using an anti-CD206 antibody, where weak, moderate and extensive infiltration of M2 macrophages from the left to the right image were shown. These tissue samples were derived from our recruited 44 GB cancers. Scale bar, 200 μm. (H) Cytotoxic CD8+ T cells TALL-104, M2 macrophages and GSDCs were detected for IDO1 and TDO2 expression by Western blotting. (I) Active or resting TALL-104 cells and GSDCs were grown to test relative levels of Kyn, Trp and Kyn/Trp ratio by LC/MS/MS, in which the levels of GSDCs were set as 1 unit. n = 3.(J) Representatives of IHC staining of YKL-40 and CD8 in GB. Inserts: amplified positive or negative staining. Scale bar, 25 μm. (K)44 cases of GB were engaged to analyze correlation between YKL-40 and CD8 staining value of IHC. The points denoting patients represented multiple individuals within that bin.(L)GEPIA 2 database were used to study relationship between CD8 and CHI3L1, n = 163. Data were presented as mean ± S.E.M. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Next, to investigate the potential KP metabolism of CTLs, we selected CD8+ T line TALL-104 [15, 33]. Given that neither IDO1 nor TDO2 was detectable (Fig. 5H), the KP level and Kyn/Trp ratio were minimal relative to GSDCs (Fig. 5I), implicating that GSDCs and M2 macrophages rather than CTLs mainly contribute to the KP metabolism in TME. To evaluate the interaction between tumor cells and immune cells in TME, we examined association of M2 and CTL infiltration with tumor YKL-40 expression. IHC staining of our 44 cases of GB showed that YKL-40 levels were not correlated with CD206, but were negatively correlated with CD8 expression (Fig. S4; Fig. 5J, K). Their transcript levels from the database of GEPIA 2 also displayed a tendency toward a negative association, but need to be confirmed in future (Fig. 5L).

Coordination of YKL-40 and Kyn in the inhibitory immune microenvironment

To evaluate the anti-immune effect of Kyn secreted from brain tumor cells, we first assessed the effects of Kyn on CTL activity. Incubation of CTLs with Kyn inhibited expression of anti-tumor immune factors including Gramz B, IFN-γ, T-bet, TNF-α, and perforin (Fig. 6A, left). Consistent with this result, culture conditioned media from shIDO1 GSDCs yielded 55.8–89.7% higher immune factors from CTLs than those from control GSDCs (Fig. 6A, right). Cultured conditioned media from shYKL-40 resembled shIDO media to increase these factors (Fig. 6B, left); in contrast, cultured conditioned media from OE YKL-40 GL261 cells inhibited these immune factors(Fig. 6B, right), suggesting that both Kyn and YKL-40 have the ability to inhibit CTL anti-tumor immunity. Like Kyn, YKL-40 blocked these factors expression by 26–59% relative to counterpart controls (Fig. 6C). To validate the inhibitory effects of both factors on CTLs, we examined cell death function. Treatment of CTLs with either Kyn or YKL-40 led to CTL apoptosis by 3.44 ∼ 4.49-fold greater than controls (Fig. 6D). Likewise, CTL apoptosis was increased by 25.3% compared to the control when CM of GL261 OE YKL-40 was introduced to CTLs.

Fig. 6.

YKL-40 and Kyn coordinately promoted tumor cell motility, but inhibited CD8+ T cell immunity. (A) TALL-104 cells were activated and treated with or without Kyn (200 µM) for 48 h (left). Conditioned media from control and shIDO1 GSDCs were added to TALL-104 cells for 48 h (right). Total RNA of TALL-104 cells was used to test mRNA of Gramz B, IFN-γ, T-bet, TNF-α and perforin via RT-qPCR. n = 3. (B)Conditioned media from control and shYKL-40 GSDCs (left) and control and OE YKL-40 GL261 cells (right)were identically used to test TALL-104 cell immune factor expression. n = 3.(C)YKL-40 recombinant protein (200 ng/mL) was identically used to test TALL-104 cell immune factor expression. n = 3.(D) Activated TALL-104 cells treated with or without Kyn (200 µM) or YKL-40 (200 ng/mL) for 48 h (top), and with CM of GL261 control or GL261 OE YKL-40 cells for 48 h (middle). Then cells were collected for apoptosis assay using Annexin V-FITC and PI staining.Q1-UR with purple dash circles showed the apoptosis cells and quantified (bottom). n = 3. (E) Control, shYKL-40 and shIDO1 GSDCs in the absence or presence of Kyn (200 µM) were used to test cell migration using Transwell assay, in which the levels of control cells were set as 1 unit. n = 3. Scale bar, 50 μm. (F) Cell motility of control and OE YKL-40 GL261 cells were detected at 0, 24 and 48 h in a wound healing assay (left) and quantified (right) compared with space at 0 h (100%), n = 3. Scale bar, 100 μm.(G) Control and shYKL-40 GSDCs were treated with M2 macrophage-CM or TDO2 inhibitor-pretreated M2 macrophage CM. Cell motility at 0, 12 and 24 h was analyzed in a wound healing assay (left) and quantified (right) compared with space at 0 h (100%), n = 3. Scale bar, 100 μm. Data were presented as mean ± S.E.M. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Then, to determine if YKL-40 and Kyn possess the ability to promote tumor cell motility, we used a transwell migration assay and found that knockdown of CHI3L1 in GSDCs abrogated cell motility by 70.0% compared with control GSDCs, but exogenous addition of Kyn to the cultured cells, the migration was recovered to 81.3% of the control level (Fig. 6E, top). Further to confirm the activity of endogenous Kyn, we exploited shIDO1 GSDCs and IDO1 knockdown decreased cell migration by 53.0%, and the migration was resumed to 83.5%by adding Kyn(Fig. 6E, bottom). The migration capability of control and OE YKL-40 GL261 cells were also detected in a cell scratch assay and showed that GL261 cells were increased when YKL-40 was overexpressed (Fig. 6F), indicating that both YKL-40 and Kyn promote tumor cell migration. Finally, in order to evaluate the potential effects enhanced by M2 macrophages that secrete a large volume of Kyn on tumor cell migration, we collected conditioned media from M2 macrophages pre-treated with TDO2 blocker 680C91 and observed GSDC motility in the presence of the conditioned media during 24 h in a cell scratch assay (Fig. 6G). In control GSDCs, the conditioned media with 680C91-treatment significantly inhibited GSDC migration relative to counterpart controls in 12 and 24 h. When CHI3L1 gene was knocked in GSDCs, the inhibition of these conditioned media was noticeably enhanced, supporting that YKL-40 and Kyn cooperate to promote tumor cell motility. Overall, these findings highlight the pathological signature of the tumor immune microenvironment in which M2 macrophages collaborate with tumor cells to synergistically inhibit CTL anti-tumor immunity through secretion of YKL-40 and Kyn, thus facilitating tumor invasiveness.

YKL-40 regulated KP metabolism in brain tumor in vivo

Finally, to evaluate the KP metabolism regulated by YKL-40 in brain tumor development in vivo, we established orthotopic murine models. First, an immunocompromised model with nude mice was established using GSDCs. Six weeks following GSDC injection in the brain, TALL-104 cells were supplemented via tail vein injection (Fig. 7A). After ten-week injection with tumor cells, individual mice were sacrificed for tumor sample studies. Consistent with larger tumors in control mice, YKL-40 levels in brain tissue from control mice were higher than those seen in shYKL-40 mice (Fig. 7B and Fig. S5), and Kyn and Kyn/Trp ratio in blood were 1.0 and 1.5-fold higher levels than those in shYKL-40 mice (Fig. 7C). Accordingly, immune fluorescent staining levels of IDO1 and CD206 in control mice were 1.3 and 1.4-fold increase but CD8 was 50% decrease relative to the corresponding levels in shYKL-40 animals (Fig. 7D and E). These results suggest that YKL-40 stimulates KP metabolism to develop an immune suppressive microenvironment with increased TAMs and decreased CTLs in the orthotopic brain tumors.

Fig. 7.

YKL-40 regulated KP metabolism in brain tumor in vivo. (A) Scheme of an established orthotopic murine model of glioma with GSDC transplantation. Control and shYKL-40 GSDCs (15 × 10⁴) were transplanted into the brain of 6 ~ 8-week-old nude mice. At the 7th week, TALL-104 cells (1 × 10⁶) were supplemented via tail vein administration. Sample collection (brain and blood) started from week 10, whenever mice exhibited moribund symptoms (e.g. decreased moving, head bump, breath difficulty, decreased body weight). n = 6.The entire observation time was about 14 weeks. (B) Brain tissue was processed for H& E staining (top), where white dashed lines showed tumor area and green arrows indicated adjacent normal tissue. Levels of YKL-40 in the brain tissue were tested via immunoblotting (bottom). n = 5. Scale bar, 4 mm. (C) Serum samples were tested for concentration of Kyn, Trp, and Kyn/Trp ratio via ELISA. n = 6. (D) Brain tissue samples were subjected for immunofluorescence of IDO1, CD206 and CD8.(E) Quantification of mean fluorensence intensity per field (20X amplification). n = 6.Scale bar, 25 μm. (F) Control and OE YKL-40 GL261 cells (30 × 10⁴) were injected into the brain of 6 ~ 8-week-old C57BL/6 mice with the stereotaxic apparatus. Tumor samples were collected whenever mice showed moribund symptoms. n = 6. The entire observation time was about 6 weeks (G) Mouse survival was recorded and analyzed via Kaplan-Meier survival plot. (H) Brain tissue samples were subjected to immunofluorescence staining of CD206, CD8,IL-6, TNF-α and IFN-γ. Images displayed both tumor center and surrounding staining in which inserts amplified staining. Quantification of mean fluorensence intensity per field (20X) was derived from cancer center (CD206) or surrounding (others) staining analysis. n = 6.Scale bar, 25 μm. Data were presented as mean ± S.E.M. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

To evaluate if immune checkpoint blockers can reinvigorate the immune activity inhibited by YKL-40 and ameliorate survival, we employed a PD-1 antibody Tislelizumab that is currently practiced in clinical trials and preclinical studies in mice [34, 35]. Control and OE YKL-40 GL261 cells were transplanted into the brain of the immune intact C57BL/6 mice and 16 days later, Tislelizumab was injected via intraperitoneal twice a week (Fig. 7F). When mice displayed moribund symptoms, they were individually sacrificed and processed for tumor sample studies. Unexpectedly, the survival of control mice treated with Tislelizumab rapidly died from day 18 to day 21 (Fig. 7G) during which none from any other groups died. This unanticipated event prompted us to determine the likelihood of life-threatening disorder triggered by PD-1 blocker administration. Not surprisingly, it was reported in clinic that treatment of cancer patients with monoclonal antibodies resulted in cytokine release syndrome or cytokine storm [36]. To validate that elevated immune cytokines and immune cell hyperactivation do occur in Tislelizumab-treated control mice, we examined macrophages, CTLs, and cytokines (Fig. 7H). Of note, large populations of infiltrating CD206+ macrophages were accumulated in the tumor while CD8+ CTLs were in tumor surrounding region. Accordingly, inflammatory cytokines IL-6, TNF-α and IFN-γ in Tislelizumab-treated tumors were 2.0 ~ 2.5-fold higher than those in control tumors which were barely detectable. These findings supported our hypothesis that the cytokine storm was stimulated by Tislelizumab in mice. Although the survival of OE YKL-40 mice was moderately decreased relative to control mice, the survival of OE YKL-40 mice treated with Tislelizumab was noticeably longer than the mice with no antibody treatment (Fig. 7G). Immune fluorescent staining revealed that OE YKL-40 tumors harbored decreased CTLs and cytokines (IL-6, TNF-α, IFN-γ) compared with controls, agreed with in vitro immune suppressive activity of YKL-40 (Fig. 6A–G). Additional treatment with Tislelizumab to OE YKL-40 group restored 1.5 ~ 2.3-fold higher levels than cytokines in OE YKL-40 tumors; however, these elevated cytokines were partially recovered as compared with these notably elevated levels seen in Tislelizumab-treated control mice (Fig. 7G), suggesting that a new balance of cytokines due to the opposite activity of YKL-40 and Tislelizumab was established in YKL-40-expressing tumors, the ultimate levels favorable for anti-tumor immunity. Altogether, these in vivo studies demonstrate that YKL-40 upregulates KP metabolism to inhibit tumor immunity, therefore, promoting brain tumor development. Treatment with immune PD-1 antibody in YKL-40⁺ GBs may give rise to promising benefit to disease survival.

Fig. 8.

Scheme of KP metabolism regulated by YKL-40 in the glioblastoma immune microenvironment. Elevated expression of YKL-40 by tumor cells upregulates IDO1/TDO2 via m6A demethylation of AhR/SRF to stimulate KP metabolism, in which increased Kyn cooperates with YKL-40 to block CTL anti-tumor immunity. In addition, infiltrating M2 macrophages also increasingly promote TDO2-dependent KP metabolism, resulting in Kyn accumulation in the TME. Therefore, notably increased levels of Kyn and YKL-40 from tumor cells and M2 macrophages in the TME synergistically inhibit CTL cytotoxicity, leading to tumor immune evasion

Discussion

Highly activated KP metabolism has recently received significant attention in the development of a number of cancers including GB [17, 18, 24]. However, the molecular mechanisms of its rate-limiting enzymes IDO1 and TDO2 underlying the malignant transformation of brain tumor, particular for mesenchymal GB are poorly understood. The current study discovered that YKL-40, the biomarker of mesenchymal GB, displays the rigorous ability to trigger IDO1/TDO2expression that activates the KP metabolism. The elevated expression of IDO1 or TDO2 relied on epigenetic modification (methylation and degradation) of their transcription factors AhR and SRF controlled by METTL3and YTHDF2. The increased Kyn from tumor cells collaborated with Kyn from TAMs to synergistically inhibit CTL cytotoxicity, thus establishing an immune suppressive microenvironment for tumor immune evasion (Fig. 8). Therefore, the current study has highlighted novel pathological insights into the inhibitory immune microenvironment developed by YKL-40-induced KP metabolism that mediates interaction among tumor cells, TAMs and CTLs. The findings may offer an optimal therapeutic devise for the treatment of mesenchymal GB resistant to immunotherapy and others.

With regard to IDO1 and TDO2 expression, we interestingly found that YKL-40 regulated either IDO1 or TDO2 in cultured cells, and when YKL-40 was suppressed, IDO1 was decreased; in sharp contrast, TDO2 was oppositely induced. In the concert with this finding, the staining analysis of IDO1 and TDO2 unveiled their negative correlation in our recruited GB tissue. However, in the public datasets, CHI3L1, IDO1 and TDO2 were positively correlated with each other and patient poor survival. A number of factors may be accounted for the inconsistency between the databases and our studies. First, the public databases utilized gene transcripts for the correlation analyses; however, our studies primarily focused on protein levels which are pathophysiologically functional. Next, it could be ascribed to difference of the recruited populations. In the datasets, patients expressing IDO1 were likely different from patients with TDO2 expression, thus either IDO1 or TDO2 was independently associated with CHI3L1 level or patient survival. In our studies, we enrolled the patients that were evaluated for both expression levels of IDO1 and TDO2. We found the opposite regulations of IDO1 and TDO2 in the same patients, most of which expressed YKL-40 (≥ 3 points, 70.4% GB) and were IDO1-positive but TDO2-negative. In addition, we found in murine glioma cells GL261 that overexpress YKL-40 only elevated TDO2, not IDO1. Indeed, either one of the similarly functional enzymes (IDO1 or TDO2) was sufficient to catalyze the KP metabolism and provide products (NAD+) sufficient for the following metabolic reactions, meeting the principle of cellular energy efficiency. Lastly, different GB subtype cohorts were recruited in different studies, as the dataset employed gliomas, while our GB cohorts included a large population of mesenchymal GB. Collectively, our findings reveal that YKL-40 activates KP metabolism via divergently regulating IDO1 and TDO2.

It is quite of interest to note the difference of Kyn levels and YKL-40-dependent IDO1 or TDO2 regulatory mechanisms between GL261 cells and GSDCs. Although substantially mechanistic insights accounted for these differences remain unknown and are not our primary focus, a few important factors may be taken into account in the divergent scenarios. For example, GL261 cells are derived from mice and classified as a glioma line, whereas GSDCs are from human GB which expresses strong YKL-40, representing different subtypes of brain tumor and genetically distinct species. Unlike GSDCs, GL261 cells do not express YKL-40 and IDO1 regulation is independent of YKL-40. In addition, except YKL-40, a number of factors were identified to regulate KP including inflammatory factors [37], immune mediators [38], hormones [39], thus the varied levels of these factors in these cells remain to be evaluated.

Growing evidence has demonstrated that METTL3 interacts with METTL14 as co-regulators to induce m6A methylation and subsequent degradation by YTHDFs, thus destabilizing targeted genes, although METTL3 has also been reported to increase targeted gene stability [40–42]. For instance, multiple studies have showed that RNA demethylation increased mRNA stability [43–45]. In GB, METTL3 and METTL14 acted as tumor suppressors to inhibit self-renewal and tumorigenesis of GB stem cells [46]. In line with these findings, our study showed that levels of METTL3, METTL14 and YTHDF2 were decreased in YKL-40-overexpressing GB, and bothdecreasedMETTL3 and YTHDF2stabilized mRNA of AhR and IDO1, although the coordination of METTL3 and METTL14 remained to be validated. Nevertheless, our results suggest that METTL3 coupling with YTHDF2functions to differentially regulate AHR and SRF expression, thus controlling IDO1 and TDO2 expression, respectively.

It is noteworthy that when CHI3L1 gene expression was silenced and METTL3 was subsequently induced, resulting in decreased expression of IDO1 but increased expression of TDO2. When IDO1 gene was directly knocked down or its activity was inhibited, TDO2 or YKL-40 expression did not change. Likewise, inhibition of TDO2 activity did not alter IDO1 gene expression. These results indicate that YKL-40 acts as the primary upstream factor to govern individual gene expression of IDO1 and TDO2 via METTL3-mediated epigenetic modification on AhR and SRF in the KP metabolism. This finding may provide explanation for the clinical failure of IDO1-targeting monotherapy in a phase III trial [47, 48], where high expression of YKL-40 can sustain to induce TDO2 gene expression, alternatively activating the KP metabolism. Therefore, combination treatment of IDO1 inhibitor with TDO2 or YKL-40 blocker may be deliberately taken into account in cancer patients that overexpress YKL-40.

Although Kyn/Trp ratio was considerably higher in M2 macrophages than that in tumor cells, varied levels of infiltrating macrophages in GB determined their individual contributions to the overall serum Kyn/Trp ratio, as different cases harbored different levels of TAM infiltration capable of abrogating CTL immunity [30]. Therefore, serum Kyn/Trp ratio reflects the entity of the KP metabolism in the TME orchestrated with tumor cells, TAMs, immune cells and vascular cells. It is of interest to note that YKL-40 did not regulate TDO2 in M2 macrophages in our system; however, YKL-40 was essential for M0 differentiation into M2 type of TAMs [14]. Blockade of YKL-40 in M0 macrophages resulted in M0 differentiation into M1 rather than M2 type macrophages. Given that M2 macrophages primarily contribute to the KP metabolism [49, 50], it is anticipated that attenuated M2 population may ultimately decrease the level of Kyn and Kyn/Trp ratio, if YKL-40 activity is ablated.

Our xenografted animals treated with Tislelizumab exhibited severe cytokine storm syndrome and rapidly died. Indeed, this syndrome was typically documented in the clinical trials with monoclonal antibody therapy with Blincyto (blinatumomab) and Rituxan (rituximab) in leukemia and lymphoma patients [36]. It was pathologically mediated by hyperactive immune cell infiltration and a large volume of cytokine release. Consistent with this clinical syndrome, we found remarkable infiltration of macrophages and CTLs and cytokine accumulation in tumors treated with Tislelizumab, although participation of other immune cell types can be not excluded. In addition, the high levels of Kyn secreted from GL261 cells relative to GSDCs may also contribute to the development of cytokine storm. Therefore, the unexpected results indicate the unfavorable outcomes of PD-1 antibody treatment in mice carrying control tumors; however, the YKL-40-expressing tumor mice treated with Tislelizumab yield significant benefit to the survival due to immune suppressive activity of YKL-40, thus pointing to devising treatment with PD-1 antibody in mesenchymal GB expressing YKL-40.

Like secreted protein YKL-40,Kyn has been reported to induce gene expression of some cytoskeletal proteins, cytokines and growth factors (e.g. IL-10, TGF-β) that are in turn able to promote cell motility [51–53]. In the cells, Kyn can also be further metabolized to other products whereby they induce cell migration. In addition, Kyn binds to AhR and induces PD-1 expression in CD8+ T cells, inhibiting immune activity [26, 28, 54]. Our current research found that both YKL-40 and Kyn promoted tumor cell migration, and inhibited CTL anti-tumor immunity, thus providing the action model of synergistic immune suppression in the TME. Assessment of YKL-40 serum levels combined with serum Kyn or Kyn/Trp ratio may offer diagnostic and prognostic value, and also help devise novel combined therapeutic intervention aimed at YKL-40, IDO1 and/or TDO2.

Conclusion

YKL-40 divergently upregulates IDO1 and TDO2 to activate KP metabolism in tumor cells, and coordinates with Kyn from infiltrating macrophages to inhibit cytotoxic CD8+ T cells, thus establishing an immune suppressive microenvironment for tumor immune evasion.

Materials and methods

Cell lines and cell culture

Human glioblastoma cell line glioblastoma serum-differentiated cells (GSDCs) and shYKL-40 GSDCs [31] and U87 cells, and murine malignant glioma cell line GL261 (Shanghai Fuheng Biotechnology, Shanghai, China) and OE YKL-40 GL261 cells [15] were cultured in DMEM (Gibco, NY, USA) with 10% fetal bovine saline (FBS, Lonsera, Canelones, Uruguay) and 1% Penicillin/Streptomycin (HyClone, UT, USA). Human monocyte leukemia cell line THP-1 cells and shYKL-40 THP-1 cells [14] were cultured in RPMI-1640 (Gibco, USA) with 10% FBS and 1%Penicillin/Streptomycin. For macrophage differentiation, 150 nM Phorbol 12-myristate 13-acetate (PMA, #P8139, Sigma-Aldrich, MO, USA) was added to the culture medium of THP-1 cells for 24 h to induce M0 type macrophages. 10 pg/mL LPS (#L2630, Sigma-Aldrich, USA) and 20 ng/mL IFN-γ (#300-02, Peprotech, NJ, USA) were added to RPMI-1640 with 10% FBS and 1% Penicillin/Streptomycin following the M0 macrophage induction for 24 h to M1-like macrophage differentiation.20 ng/mL IL-4 (#200-04, Peprotech, USA) and 20 ng/mL IL-13 (#200-13, Peprotech, USA) were used in RPMI-1640 with 10% FBS and 1% Penicillin/Streptomycin following the M0 macrophage induction for 72 h to M2-like macrophage differentiation [55]. Human cytotoxic T-cell line TALL-104 (ATCC, VA, USA) was cultured in RPMI-1640 with 10% FBS, 1%Penicillin/Streptomycin and 50 U/ml IL-2 (#200-02, Peprotech, USA).

Human samples

The study on human samples was approved by the Ethics Committee internal review board of Shanghai General Hospital (affiliated to Shanghai Jiao Tong University). Forty-four patients with a primary diagnosis of GB were recruited.

Database analysis

Gene Expression Profiling Interactive Analysis 2 (GEPIA 2, http://gepia2.cancer-pku.cn/) dataset, Chinese Glioma Genome Atlas (CGGA, http://www.cgga.org.cn) dataset and The Cancer Genome Atlas Program (TCGA, https://portal.gdc.cancer.gov/) were used for clinical analysis.

Gene microarray

Total RNA from GSDC^YKL−H and GSDC^YKL−L was isolated using Tri-reagent (Sigma Inc). 1 µg of each sample RNA was used to synthesize Cy-3-labeled cRNA for running Agilent Whole Human Genome Microarray 4 × 44 K (Shanghai Biotechnology Corporation, China), which contained 41,000 human genes or transcripts. The microarray slide was incubated at 65 °C for 17 h in a hybridization oven, model G2545A (Agilent Technologies, USA), washed three times at 25 °C, and then scanned using a Gene Chip Scanner. Feature Extraction software (Agilent Technologies) was used to extract the scanned data of each microarray slide. GeneSpring GX software (Agilent Technologies) was used to analyze gene expression and the quantile normalization method was used to normalize the chip data. Welch’s t-tests and the Significance Analysis of Microarrays (SAM) tests were used to identify genes that were differentially expressed and fold-values > 2.0 were filters for screening genes.

Stable cell line construction

The plasmids of shIDO1 and shAhR were generated by introduction of IDO1 sequence or AhR sequence into a pGIPZ vector (Table S3). For the lentivirus packaging, 293T cells were co-transfected with shRNA plasmid, psPAX2, and pMD2.G vector in the presence of serum free DMEM containing lipofectamine 8000 (#C0533, Beyotime, Shanghai, China) for 6 h as described previously [15]. Stable cells resistant to puromycin after 3 passages were used in the experiments.

LC/MS/MS method and sample detection

For cell supernatant sample preparation, cell cultured medium was collected, mixed with 80% methanol and volatilized for detection. For cell lysate sample preparation, cells were lysed with 80% methanol at -80℃, and centrifuged to get the supernatant extracts for detection.

For the detection of Kyn and Trp, API 4000™ LC/MS/MS system (AB Sciex, CA, USA) was used for detection and ACQUITY UPLC BEH Amide Column (#186004801, 130Å, 1.7 μm, 2.1 mm x 100 mm, Waters, MA, USA) was used for separation. The mobile phase A was acetonitrile containing 0.1% formic acid and the mobile phase B was 10 mM ammonium formate containing 0.1% formic acid. The linear gradient was performed at 3 min with 20% B, 15 min with 50% B, 18 min with 50% B and 20 min with 20% B. Positive ESI mode was used for analysis. 5µL sample was injected for the metabolites detection. Cell number was used for normalization. For the detection of NAD+, succinic acid, AMP, GMP, CMP, IMP, and adenosine (Ado), LC/MS/MS method was specifically described in Table S4.

Western blotting

Cell lysates were subjected to immunoblotting as described previously [15].

RT-qPCR

The qRT-PCR for cellular mRNA was used as described [15]. The primer was listed in Table S6.

Dual luciferase gene reporter assay

pGL3 vector for transcriptional activity detection and pmir-GLO vector for RNA stability detection constructed with specific sequences mentioned above were used (Shanghai Huajin Biotechnology Co., Ltd).AhR-IDO1 promoter:5’-ttacagacacgcaccaccat-3’, AhR-TDO2 promoter: 5’-ctacaggcacgcaccacctt-3’; SRF-TDO2 promoter: 5’- gcaggctacataaaaggcagctgtaga-3’.Dual luciferase gene reporter assays were performed with the instruction of the kit as described previously (#11402ES60, Yeasen, China) [14].

Dot blot

Total RNA was extracted from cells with Trizol. Equal total RNA (1 µg) was spotted on a Hybond-N + membrane, and fixed by UV–cross-linking, and then the membrane was incubated with 5% skim milk at room temperature for 1 h. The membrane blot was incubated with an m6A antibody (#A19136, Abclonal, MA, USA) at 4 °C overnight, followed by incubation with a goat anti-rabbit secondary antibody at room temperature for 2 h. The signal of total m6A level was detected with ECL system. 0.02% methylene blue (MB) was used to stain the membrane as loading controls.

RNA stability

Control GSDCs were treated with STM2457 (40 µM) and actinomycin D (5 µg/mL, #HY-17559, MedChemExpress, NJ, USA), and then cellular total RNAs were extracted from cells with Trizol at 0, 8, 12, 24 or 36 h. AhR, SRF or IDO1 mRNA level was detected with the materials and methods described using RT-qPCR.

Transwell migration assay

Cells were tested for migration as describe previously [15]. The migrated cells were counted under the microscopy (Leica, Wetzlar, Germany).

Wound healing

Cells were seeded into a 6-well culture plate, and scratched with 200 µL tips after treatment with M2-CM or M2-CM pretreated with 680C91. The fresh medium was replaced and the image was captured at 0, 12 or 24 h by microscopy. The migrated space was calculated with Image J software (National Institutes of Health, MD, USA).

Apoptosis

TALL-104 cells were activated by incubation with anti-CD3 antibody (#16-0037-85, Thermo Scientific, MA, USA) and anti-CD28antibody (#16-0288-85, Thermo Scientific, USA) for 24 h and then Kyn (200 µM) or YKL-40 (200 ng/mL, our lab) was added to the culture medium for 48 h. Cells were collected to test apoptosis as described previously [14].

Immunocytochemistry (ICC)

Cells were used for immunocytochemistry staining as described previously [15].

Animal experiments and treatment protocol

The animal experiments were approved by the Institutional Animal Care and Use Committee of Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. For the orthotopic murine model with GL261 cells, the brain of 6 ~ 8-week-old C57BL/6 mice was transplanted with control or OEYKL-40 GL261 cells (30 × 10⁴ cells/10 µL PBS, n = 6) with a stereotaxic apparatus (RWD Life Science, CA, USA).From day 16, Tislelizumab was injected intraperitoneally twice a week (10 mg/kg). Whenever mice showed moribund symptoms like decreased moving, head bump, breath difficulty or decreased body weight, mice were sacrificed and the brain tissue and blood samples were collected. For the orthotopic murine model with GSDCs, the brain of 6 ~ 8-week-old nude mice was transplanted with control or shYKL-40 GSDC cells (15 × 10⁴ cells/10 µL PBS, n = 6) with a stereotaxic apparatus. At the 7th week, TALL-104 cells (1 × 10⁶ cells/100 µL PBS) were supplemented via tail vein administration. Whenever mice showed moribund symptoms, mice were sacrificed and the brain tissue and blood samples were collected.

Enzyme-linked immunosorbent assay (ELISA)

Blood samples were centrifuged to obtain the serum. After addition with dilution solution, Kyn and Trp levels were measured with a tryptophan and kynurenine ELISA kit (#U96-2432E, U96-3009E, YOBIBIO, Shanghai, China) through the microplate reader (Thermo Scientific, USA).

Immune histochemistry staining (IHC) and hematoxylin-eosin staining (H & E)

The fixed brain tissue was dehydrated and embedded with paraffin. The embedded tissue was processed as 3 mm sections on the slides and baked at 60℃for 2 h. IHC was performed with automatic IHC staining instrument Leica BOND RX (Leica, Germany) with primary antibodies described in Table S5. H & E was performed with Leica Autostainer XL (Leica, Germany). The image was captured with microscope.

Immunofluorescence (IF)

The paraffin tissue slides were used for IF as described previously [14].

Quantification and statistical analysis

All experiments were replicated at least three times. Statistical analysis was performed with GraphPad Prism 8 software (GraphPad Software, CA, USA), and data were presented as means ± standard error of mean (S.E.M.). Two-tailed unpaired Student’s t test and one-way analysis of variance (ANOVA) were performed. SPSS Statistics 27 (IBM Company, NY, USA) was used to analyze the p value in clinical characteristics. Results were considered statistically significant when p < 0.05.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

AhR

Aryl hydrocarbon receptor

YKL-40/CHI3L1

Chitinase 3 like 1

CTLs

Cytotoxic CD8+ T cells

GB

Glioblastoma

GDF-15

Growth differentiation factor 15

GSDCs

Glioblastoma serum-differentiated cells

IDO1

Indoleamine 2,3-dioxygenase 1

KP

Kynurenine pathway

Kyn

Kynurenine

METTL3

Methyltransferase 3

PD-L1

Programmed cell death 1 ligand 1

SRF

Serum response factor

Trp

Tryptophan

TDO2

Tryptophan 2,3-dioxygenase 2

Tregs

Regulatory T cells

YTHDF2

YTH domain family, member 2

Author contributions

Hui Chen: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Xuemei Zhang: Resources. Ziyi Wang: Resources. Jing Luo: Resources. Rong Shao and Yingbin Liu: Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – review & editing.

Funding

This work was partially supported by National Natural Science Foundation of China (81772512), the Science and Technology Innovation Plan of Shanghai Science and Technology Commission(20S11901700), and Xinhua Hospital Incubation grant (17DZ2260200).

Data availability

Data will be available on reasonable request.

Declarations

Ethics approval and consent to participate

The study on human samples was approved by the Ethics Committee internal review board of Shanghai General Hospital (affiliated to Shanghai Jiao Tong University). All animal experiments were approved by the Institutional Animal Care and Use Committee of Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine.

Consent for publication

The authors affirm that human research participants provided informed consent for publication of the images in Figs. 1I–K; 5G, J, K; 6 A, B; S4, Tables 1, 2.

Glossary

tumor KP, IDO1/TDO2-mediated KP metabolism, tumor immune microenvironment, cellular collaboration between tumor cells and macrophages, CD8+ T cell immunity.

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.