Engineered indoleamine 2,3-dioxygenase variant: Oxidative activation of prodrug indole-3-acetic acid against cancer cells for targeted therapy

Abstract

Indoleamine 2,3-dioxygenase (IDO) is a rate-limiting enzyme involved in the kynurenine pathway of tryptophan metabolism and has been implicated in immune regulation and tumor immune evasion. Beyond its role in tryptophan metabolism, we found that IDO can catalyze the oxidative activation of indole-3-acetic acid (IAA). This reaction generates reactive oxidative intermediates and triggers potent cytotoxicity in cancer cells. Leveraging human-derived enzymes as “suicide effectors” confers a critical advantage by mitigating the risk of immunogenic rejection. However, the intrinsic kynurenine-pathway activity of wild-type IDO constrains its immediate therapeutic translatability. To surmount this limitation, we further designed and characterized a truncated variant, IDO_268C. This variant retains high catalytic activity toward IAA but lacks the canonical kynurenine-pathway function of IDO. Validation assays demonstrated that both breast cancer MCF-7 and lung cancer A549 cells exhibited pronounced cytotoxic responses to the IDO_268C-IAA pair. Cytotoxicity was achieved through both intracellular overexpression and extracellular protein delivery. Collectively, our findings establish the IDO_268C-IAA pair as an efficient enzyme-prodrug combination, representing a promising therapeutic modality for enzyme-activated prodrug therapy in oncology.

Graphical abstract

Ye and colleagues engineered a human indoleamine 2,3-dioxygenase (IDO) variant, IDO_268C, that selectively activates the prodrug indole-3-acetic acid (IAA) for cancer therapy. By abolishing immunosuppressive kynurenine activity while retaining oxidative potency, IDO_268C induces targeted ROS-mediated cytotoxicity. This human-derived system offers a safer, non-immunogenic alternative for enzyme-activated prodrug therapy in oncology.

Introduction

To improve the selectivity of anticancer agents, enzyme-activated prodrug therapy (EPT) has been developed as a promising strategy. In this approach, non-toxic drugs are delivered to neoplasm tissues and activated specifically by targeted prodrug enzymes.^1,2 Within this context, oxidative stress induced by the intracellular generation of reactive oxygen species (ROSs) has been recognized as an effective mechanism to trigger cancer cell death.^3,4 The combination of the plant growth hormone indole-3-acetic acid (IAA) and horseradish peroxidase (HRP) represents a prominent example.^5,6 This system efficiently induces cell death across a wide range of tumor cell types.^{7,8,9,10,11,12,13,14,15,16,17,18,19,20,21} Importantly, the prodrug IAA naturally exists in human plasma at trace levels and does not have cytotoxic effect alone.²² It only becomes active after oxidative decarboxylation of HRP, forming products with cytotoxicity to mammalian cells.^23,24,25 However, the transition of this promising system from laboratory research to clinical application has been hindered by significant practical obstacles.

First, HRP is a complex plant-derived glycoprotein; the complex structure poses substantial technical challenges for recombinant production.^26,27 Secondly, HRP extracted from plant-derived sources is inherently heterogeneous, consisting of a large family of isoenzymes rather than a single protein species. This leads to high variability in both the composition and abundance of these isoenzymes across different preparations. Due to isoenzyme heterogeneity, achieving efficient delivery and stable expression of the cytotoxic effector is challenging, which in turn compromises the safety and efficacy of suicide gene therapy. To address these challenges, researchers have focused on using biotechnological approach to ensure controlled production processes that obtain single isoenzymes with stable quality, high activity, and reduced surface glycosylation.^12,13 Despite these advances, in vivo studies have been largely unsatisfactory. For example, the expression level of HRP in tumor cells remains insufficient and the resulting cytotoxicity is lower than predictive inference level.^10,13 Beyond these production concerns, the rapid removal of unbound free HRP from circulation remains a formidable hurdle. More importantly, as a non-human protein, HRP can trigger strong immune responses. These drawbacks have hindered the clinical translation of the HRP-IAA system.

To break through the limitations of the HRP-IAA system, attention has shifted toward identifying alternative and clinically feasible enzymes. In this study, we identify indoleamine 2,3-dioxygenase (IDO) as an enzyme capable of oxidizing IAA, demonstrating its potential application in prodrug activation. IDO is an intracellular heme-containing enzyme that is primarily known for catalyzing the kynurenine pathway of tryptophan metabolism.^28,29 However, it also exhibits a lesser-known peroxidase activity and is capable to oxidizing indole.^30,31 This finding is particularly relevant because the prodrug IAA is a structural derivative of indole. Subsequently, Freewan et al. systematically confirmed that human IDO functions as a physiological heme peroxidase.³² Based on these findings, we hypothesized that IDO could be used to activate IAA. Using this human enzyme would minimize the risk of immune reactions compared to plant-derived HRP. However, the therapeutic potential of wild-type IDO is complicated by its native dioxygenase activity. The resulting depletion of tryptophan and accumulation of kynurenine are potent mediators of tumor immune evasion.

Considering the potential risks associated with wild-type IDO, we also explored the use of alternative IDO variants as substitutes within IAA-based prodrug system. The candidate enzyme must fulfill two critical criteria: (1) the ability to oxidize IAA and generate cytotoxic free radicals with sufficient enzymatic activity, and (2) minimal or no involvement in the kynurenine metabolic pathway. In this study, we engineered IDO variants to isolate these two functions. We identified IDO_268C, a rationally designed variant optimized for IAA activation, but it shows near-background levels of immunosuppressive activity. Our findings provide proof-of-concept that IDO_268C can serve as a safer and more effective enzyme in IAA-based EPT systems, thereby offering a promising new avenue for gene-directed cancer therapy.

Results

IDO catalyzes the oxidation

To assess whether IDO possesses oxidative activity, we employed the Amplex Red assay as a fluorometric readout of resorufin generation, with HRP serving as a control. The full-length recombinant wild-type IDO (IDO_WT) exhibited significant oxidation reactions, with its oxidation activity being approximately 77.63% of that of HRP (Figure 1A; Table 1). This result indicates that IDO, beyond its physiological function in tryptophan catabolism, is capable of mediating redox reactions similar to those catalyzed by heme peroxidases.

Figure 1.

Biochemical characterization of engineered IDO variants

(A) Comparative analysis of enzymatic activities. The bar chart summarizes the specific activities for the oxidation pathway across the variants. Data are presented as mean ± SEM (n = 3). (B) Kinetics of oxidative catalytic activity. Real-time monitoring of the fluorogenic reaction was performed to assess the intrinsic oxidative capacity of each variant.

Table 1.

Enzymatic activities of IDO proteins with different truncations

Sample	Residue numbers	Yield (%)	Purification (%)	Specific activity
				Kynurenine pathway (absorbance a.u.)a	Oxidation pathway (fluorescence 107 a.u.)b
IDOWT	M1-G403	100	100	1.6687 ± 0.0910	124.1656 ± 2.3335
IDO120C	L120-G403	18.06	6.32	0.0524 ± 0.0012	22.6316 ± 0.3276
IDO157C	G157-G403	52.78	20.00	0.0518 ± 0.0001	26.8923 ± 0.0942
IDO263C	S263-G403	34.72	13.16	0.0551 ± 0.0007	10.8035 ± 0.4465
IDO268C	S268-G403	65.28	31.05	0.0524 ± 0.0010	56.6936 ± 0.1830
HRP	–	–	–	0.0539 ± 0.0008	159.3829 ± 1.5920
Buffer	–	–	–	0.0571 ± 0.0005	0.0863 ± 0.0032
Heme	–	–	–	0.0600 ± 0.0015	195.4154 ± 1.5920

Values are presented as mean ± SD for specific activity, as mean for yield and purification (n = 3).

^aKynurenine pathway activity was measured with 200 μg of enzyme using L-tryptophan assay.

^bOxidation pathway activity was measured with 30 μg of enzyme using the Amplex Red assay.

However, the intrinsic canonical catalytic activities of wild-type IDO (such as tryptophan depletion and kynurenine biosynthesis) preclude its direct translational application in therapeutic settings. Notably, these inherent metabolic reactions carry the risk of promoting tumor immune evasion or endowing cancer cells with non-immune-mediated survival advantages, which would undermine therapeutic efficacy. For an ideal enzyme-prodrug system, the use of an endogenous protein is preferred to minimize potential immunogenicity, while the ablation of its native physiological catalytic function is essential to avoid off-target biological effects. Guided by these design principles, we next generated a panel of rationally truncated IDO variants, with the dual goal of abolishing its canonical metabolic activity and preserving the structural scaffold required for heme incorporation and specific catalytic activity.

Rational design of IDO truncations

The rational design of IDO truncations hinges on precise selection of truncation sites that align with stringent structural and functional criteria: (1) given that the cofactor heme is central to oxidative catalysis, the integrity of the heme-binding pocket must be predominantly preserved; (2) the hydrophobic core should not be extensively exposed to prevent loss of solubility; (3) truncation within the interior of major secondary structural elements should be avoided; and (4) binding to its natural substrate, tryptophan, must be disrupted.

Following the rational design strategy and utilizing a 1.90-Å-resolution crystal structure of human IDO (PDB: 6E44) to provide detailed structural information, we designed four N-terminally truncated IDO variants (Figure 2). These variants initiate at residues Leu120, Gly157, Ser263, and Ser268, respectively, and share a common C terminus at residue Gly403, identical to that of wild-type IDO. The structure-based rational design yielded a panel of IDO variants suitable for subsequent functional evaluation.

Figure 2.

Schematic overview of the engineering strategy for IDO variant proteins

The engineered variants catalyze reporter reactions, such as the conversion of non-fluorescent Amplex Red to fluorescent resorufin, while lacking the ability to catalyze kynurenine generation. The bottom panel illustrates the four designed truncation sites and their corresponding structural representations.

Biochemical characterization of IDO variants

To evaluate the impact of truncation on IDO’s canonical metabolic activity and oxidative capacity, the four rationally designed IDO truncation variants were expressed and purified for biochemical assessment. We then assessed tryptophan dioxygenase activity and kynurenine production. Only the full-length IDO_WT exhibited robust tryptophan hydrolysis activity with substantial kynurenine accumulation. In contrast, all truncated variants displayed negligible canonical functional activity with levels measured values comparable to those of the buffer control (Table 1).

Notably, all IDO variants retained detectable oxidative activity, albeit with varying catalytic efficiencies (Figures 1A and 1B). Among these engineered constructs, the most extensively truncated variant, IDO_268C (residues number to C terminus) exhibited the most potent oxidative activity while being completely devoid of measurable tryptophan dioxygenase function (Table 1), fulfilling our core design criterion. To further characterize the catalytic efficiency of the engineered variants, we performed assays over 30 min to monitor the real-time oxidation rates of each variant (Figure 1B). As expected, the full-length IDO_WT exhibited the most potent oxidative activity, characterized by a rapid initial burst that reached a plateau within approximately 10 min. Among the engineered truncated variants, IDO_268C showed the highest reaction rate, displaying a significantly steeper slope in the fluorescence-time curve compared to IDO_120C, IDO_157C, and IDO_263C.

To understand why this minimal construct outperformed longer truncations, we analyzed its structural features using AlphaFold3 (Figure S1).³³ Remarkably, the predicted structure of IDO_268C closely matched the corresponding region in the full-length IDO crystal structure (root-mean-square deviation [RMSD]: 0.4–0.6 Å). This structure suggests that removal of the N-terminal 267 residues does not disrupt the core heme-binding architecture or the catalytic pocket. Instead, its removal eliminates domains required for kynurenine-pathway activity while leaving the oxidative machinery intact. In contrast, the shorter delete truncations (IDO_120C, IDO_157C, IDO_263C) likely disrupt structural elements required for stabilizing the heme-proximal core, as supported by their reduced enzymatic activities.

Next, we characterized their secondary structures using circular dichroism (CD) spectroscopy. As shown in Figure S2, the CD spectra of all IDO variants exhibit two characteristic negative minima at approximately 208 and 222 nm, indicating the presence of α-helical structures. However, the signal intensity varied significantly among the variants. IDO_268C showed the most pronounced negative ellipticity, suggesting a better-preserved helical scaffold compared to the other truncated versions. However, IDO_263C exhibited a significant reduction in signal intensity, with its characteristic peaks being notably shallower. This loss of signal in IDO_263C reflects a decrease in helical integrity and structural stability.

These spectroscopic data are highly consistent with our calculated secondary-structure percentages (Table 2). The data revealed that IDO_268C, which showed the highest specific activity among the truncated variants, exhibits the least structural disruption among the variants, retaining 13.62% α-helical content. In contrast, IDO_263C, which differs from IDO_268C by only five additional amino acids, exhibited a notable structural perturbation. The α-helical content of IDO_263C dropped to 10.23%, while the β sheet and β turn content increased to 35.35% and 20.46% (vs. IDO_268C 31.35% and 19.98%). Notably, the unordered coil content remained stable across these two variants (∼35%). This suggests that the inclusion of residues 263–267 in the truncated context does not increase global disorder. Instead, these residues induce a specific conformational drift from α helix to β sheet and β turn.

Table 2.

Secondary-structure composition of full-length IDO and its truncated variants determined by CD spectroscopy (n = 3)

Secondary-structure elements	IDOWT	IDO120C	IDO157C	IDO263C	IDO268C
α helix	15.01%	12.60%	13.58%	10.23%	13.62%
β sheet	31.78%	33.15%	32.08%	35.35%	31.35%
β turn	18.56%	19.35%	19.38%	20.46%	19.98%
Unordered	34.62%	34.84%	35.03%	34.92%	35.04%

Thermal melting CD analysis revealed distinct structural stabilities among the proteins (Figure S3). IDO_WT exhibited a highly robust structure with a sharp and cooperative unfolding transition characterized by a melting temperature (T_m) of 50.05°C (Figure S3A). Although the N-terminal truncation in IDO_268C resulted in a lower T_m of 44.39°C and a broader transition zone, it still maintained a well-defined protein fold, evidenced by its classic cooperative unfolding curve (Figure S3B). In contrast, IDO_263C exhibited significant structural instability, showing a continuous loss of secondary structure starting from 20°C without a distinct melting transition (Figure S3C). Consequently, the T_m for IDO_263C could not be determined using the standard Boltzmann model. These data directly confirm that IDO_268C maintains a more robust structural integrity than the longer IDO_263C variant, although its stability is lower than that of IDO_WT. On the other hand, the relative stability of the helical scaffold in IDO_268C preserves a structural environment better suited for heme-mediated oxidation. Thus, among all constructs, IDO_268C represents the minimal yet structurally self-sufficient catalytic core. It retains the heme-coordinated oxidizing capability while eliminating endogenous metabolic function.

Given the heme-dependent nature of IDO-mediated oxidation, we further assessed whether heme alone could contribute to activation. Indeed, both the Amplex Red and luminol assays revealed a strong oxidation capacity from free heme (Table 1). However, this reaction decayed within ∼4 s, making it unsuitable for any drug delivery or prodrug activation application. While free heme can initiate oxidation, it lacks the structural stability required for sustained catalysis. A protein scaffold, such as the one preserved in IDO_268C, is therefore essential to support the prolonged catalytic activity necessary for effective prodrug activation.

Evaluation of synergistic cytotoxicity induced by IDO and IAA combinations

We have shown that both full-length IDO_WT and IDO_268C variant catalyze the oxidation in vitro. Subsequently, their potential cytotoxic effects on human lung adenocarcinoma A549 cells were further investigated by supplementing the culture medium with exogenous IDO and IAA. Notably, both IDO_WT-IAA (Figure 3A) and IDO_268C-IAA (Figure 3B) combinations induced pronounced cell death when IAA concentrations were adjusted to the 0.5–3.5 mM range. To distinguish the enzyme-mediated effects from the intrinsic toxicity of the substrate, we included a control group treated with IAA alone (dashed line). The results reveal that IAA remains relatively non-toxic (viability > 80%) up to 3.5 mM, confirming that the observed cytotoxicity in the combined treatment is primarily driven by the catalytic activation of IAA by IDO variants. Within this window, a clear enzyme concentration-dependent cytotoxic response was observed particularly within a 0.5–2.0 mM subrange. For instance, at 1 mM IAA, increasing the IDO_WT concentration from 1 to 100 μg/mL resulted in a significant drop in cell viability from 65.67% to 7.36%. A similar dose-dependent trend was observed for IDO_268C, where viability decreased from 85.08% to 16.46% across the same enzyme range. In contrast, when IAA concentrations reached the 2.0–3.5 mM range, even the minimum enzyme loading (1 μg/mL) led to near-complete cell death, with viability dropping below 20%. This observation indicates a high-sensitivity response where trace amounts of enzyme are sufficient to trigger lethal levels of ROSs. Conversely, at IAA concentrations below 0.5 mM, even the maximum enzyme dosage (100 μg/mL) failed to produce potent cytotoxicity.

Figure 3.

Cytotoxicity and ROS generation induced by IDO_WT and IDO_268C in A549 cells

Cell viability of A549 cells treated with IDO_WT (A) or IDO_268C (B) at various enzyme concentrations (1–100 μg/mL) and IAA concentrations (0–4.5 mM). Cell viability was measured after 48 h (means ± SEM, n = 3). The experiment was assessed using a checkerboard method. The dashed line represents the control group treated with IAA alone (without enzyme). (C) Intracellular ROS levels were measured in cells treated with HRP (1.2 μg/mL), IDO_WT (4 μg/mL), or IDO_268C (8 μg/mL) in the presence or absence of 0.5 mM IAA. ROS levels were measured 6 h after treatment using the DCFH-DA fluorescent probe. Fluorescence intensity reflects ROS generation (means ± SEM, n = 3), with statistical significance indicated (p < 0.0001).

Collectively, these results establish two key observations: (1) IAA exerts inherent cytotoxicity only at concentrations exceeding 3.5 mM; and (2) within the moderate IAA range (0.5–2.0 mM), IDO_WT- or IDO_268C-mediated oxidative activation is the primary driver of cell death. Our results thus demonstrate that the catalytic function of IDO_WT or its IDO_268C variant is mandatory for triggering cell death at safe IAA dosages. Potent cytotoxicity is elicited only through their catalytic activity, highlighting their irreplaceable role in realizing the therapeutic potential of the IAA-based prodrug system.

ROS generation and time-dependent cytotoxic responses induced by IDO-IAA pairs

To further delineate the cytotoxic mechanism and optimize dosing for subsequent cell experiments, we compared the capacity of IDO_WT and IDO_268C to induce ROS-mediated cell death in A549 cells, with 1.2 μg/mL HRP serving as a positive control. ROS generation by different IDO_WT-IAA and IDO_268C-IAA combinations was quantified using the DCFH-DA (2′,7′-Dichlorodihydrofluorescein diacetate) fluorescent probe, a well-established indicator of intracellular oxidative stress (Figure 3C). Fluorescence analysis showed that IAA alone does not generate free radicals or ROSs. In contrast, combining 0.5 mM IAA with either 4 μg/mL IDO_WT or 8 μg/mL IDO_268C led to a sharp increase in ROS levels. This confirmed that oxidative stress is triggered specifically by enzymatic activation.

Furthermore, we investigated the effects of IDO_WT-IAA and IDO_268C-IAA prodrug pairs on cell morphology and viability with optical microscopy (Figures 4 and S4). At 0.5 mM, IAA alone did not change the cell appearance or health. The cells remained healthy and firmly attached to the plates. This matches our ROS results, where no radicals were detected for IAA alone. However, adding the enzymes IDO_WT or IDO_268C led to clear toxic effects. At low IAA concentrations (0.5 mM) and short exposure duration (24–72 h), A549 cells exhibited strong resistance to oxidative stress, showing no obvious morphological alterations (Figure S4). Cell viability was maintained above 90% at 24 h and above 60% after 72 h of treatment (Figures 4A and 4C). In contrast, human breast cancer MCF-7 cells treated with exogenous IDO_WT or IDO_268C showed a lower viability at the same IAA concentration, declining from 75% at 24 h to 45% (IDO_WT) and 52% (IDO_268C) at 72 h, respectively (Figures 4B and 4C). As IAA concentration increased to 1–2 mM, prominent cytotoxicity was observed in both cell lines. Following 24 h of treatment with 2 mM IAA, cells underwent shrinkage, morphological distortion, and loss of tight adherence to the culture substrate. By 72 h, all treated cells had rounded up and formed clusters of dead cells, with the survival rates of A549 and MCF-7 cells decreasing to <10% (Figures 4C and S4).

Figure 4.

Effects of IAA treatment on cell morphology and viability in A549 and MCF-7 cells supplemented with exogenous proteins

(A) A549 cells and (B) MCF-7 cells were treated with 1 mM IAA in the absence or presence of HRP (1.2 μg/mL), IDO_WT (4 μg/mL), or IDO_268C (8 μg/mL). Bright-field microscopy images were taken after 24 h, showing cell detachment and morphological changes compared with control group. Scale bar: 100 μm. (C) Cell viability was determined after 24, 48, and 72 h of incubation using the MTT assay (mean ± SEM, n = 3).

It is worth noting that the full-length IDO_WT acts more quickly than IDO_268C during short treatment periods. For instance, after 24 h of incubation with 2 mM IAA, IDO_WT reduced A549 cell survival to 26.58%, while IDO_268C showed 58.12%. This gap disappears as the incubation time increases. By 72 h, both enzymes reached almost identical killing levels at 1 mM IAA, with survival rates around 11.1%. Higher IAA concentrations also help balance the cytotoxic effects between the two enzymes. In MCF-7 cells treated for 48 h, IDO_WT showed slightly higher toxicity at 1 mM IAA (27.22% survival) compared to IDO_268C (37.63% survival). However, increasing the IAA dose to 2 mM brought the survival rates for both enzymes to a nearly equal level of approximately 12.3%. These results show that IDO_268C and IDO_WT share similar overall therapeutic potential. Their performance can be synchronized by simply optimizing the drug concentration or the treatment time.

Cytotoxicity of intracellularly expressed IDO-IAA in transiently transfected cells

To further explore the therapeutic potential of the IDO-IAA prodrug system via nucleic acid-mediated delivery, we next evaluated cytotoxicity in cells transiently transfected with IDO_WT or IDO_268C followed by exogenous IAA administration. Successful transfection and proper expression of IDO_WT or IDO_268C were confirmed by western blotting, which detected the target proteins at their expected molecular weights (∼42 kDa for IDO_WT and ∼16 kDa for IDO_268C) (Figure 5A). Functional validation via kynurenine production assays showed that IDO_WT-transfected A549 and MCF-7 cells exhibited 9.18- and 8.35-fold increases in kynurenine levels, respectively, compared to cells transfected with a fluorescent reporter vector (Table 3). In contrast, IDO_268C-transfected cells showed no significant tryptophan dioxygenase activity relative to the negative control, consistent with the results of our in vitro enzymatic assays. Notably, aligning with the concentration-dependent cytotoxicity observed in the exogenous enzyme system, overexpression of either IDO_WT or IDO_268C had no significant effect on cell viability in the absence of exogenous IAA (Figure S5). This confirms that, under transient transfection conditions, the intracellularly expressed enzymes do not possess detectable inherent cytotoxicity. Likewise, cells transiently transfected with the fluorescent reporter plasmid and treated with 1 mM IAA displayed phenotypes indistinguishable from non-transfected cells, further verifying that the 1 mM IAA or expression vector itself does not exert cytotoxic effects.

Figure 5.

Effects of IAA administration on the viability of cells transiently transfected with intracellularly expressed proteins

(A) Expression of the transfected proteins was confirmed by western blot using a His-tagged antibody in whole-cell lysates from non-transfected, IDO_WT-transfected, or IDO_268C-transfected A549 and MCF-7 cells, respectively. Dose-dependent cytotoxicity of (B) A549 and (C) MCF-7 cells transiently transfected with IDO_WT or IDO_268C following incubation with increasing concentrations of IAA (mean ± SEM, n = 3).

Table 3.

The L-tryptophan assay in different transfected clonal cell lines and controls

Cell lines	Transient gene	Specific activity
A549	IDOWT	0.281 ± 0.003
	IDO268C	0.0331 ± 0.0018
	pcDNA3.1	0.0306 ± 0.0047
MCF-7	IDOWT	0.331 ± 0.0026
	IDO268C	0.0382 ± 0.0012
	pcDNA3.1	0.0396 ± 0.0014

Values shown are mean ± SE, n = 4 and 3, respectively, when A549 and MCF-7 cells were used.

Building on these observations, we next assessed the cytotoxic effects of IAA in combination with transiently expressed IDO_WT or IDO_268C. Consistent with the findings from exogenous protein supplementation, IAA induced dose-dependent cell death in IDO_WT- and IDO_268C-transfected cells (Figure S5). The most crucial point is that, in transiently transfected A549 (Figure 5B) and MCF-7 (Figure 5C) cells, treatment with approximately 1.5 mM IAA already resulted in substantial cell death after 24 h of incubation.

Furthermore, cell line-specific differences in sensitivity to the IDO_WT/IDO_268C-IAA prodrug systems were also observed. For instance, in A549 cells, increasing the IAA concentration from 1 to 2 mM reduced the viability of IDO_WT-transfected cells from 62% to 23% (IC₅₀ = 1.28 mM) and that of IDO_268C-transfected cells from 92% to 51% (IC₅₀ = 2.05 mM) (Figure 5B). In MCF-7 cells, IDO_WT-transfected cells showed a decrease in viability from 45% to 11% (IC₅₀ = 0.98 mM), while IDO_268C-transfected cells exhibited a reduction from 94% to 32% (IC₅₀ = 1.70 mM) (Figure 5C). Nevertheless, according to the results of western blotting, MCF-7 cells displayed remarkably higher expression levels of the transfected proteins compared with A549 cells (Figure 5A). The elevated protein expression in MCF-7 cells appears to be a key factor contributing to their heightened sensitivity to IAA. This is supported by the observed increase in ROS production and the concomitant reduction in cell viability under comparable experimental conditions.

Discussion

Given the inherent limitations of the classical HRP-IAA enzyme-prodrug system, there is an urgent need to develop a more efficient prodrug-activating system to enable successful clinical translation. Ideally, such a system should retain high catalytic efficiency under physiological conditions, particularly maintaining efficient substrates turnover at low concentrations (high k_cat/K_m) without requiring additional enzymatic steps.^34,35 While non-human enzymes such as HRP exhibit robust catalytic activity, their use inevitably elicits strong immune responses and encounters challenges in recombinant expression and batch-to-batch standardization.³⁶ Employing human-derived enzymes could mitigate these immunogenic concerns; however, this strategy presents its own hurdles. The catalytic mechanism of the potency-related oxidation of the enzyme must be distinctly orthogonal to its physiological enzymatic functions to avoid inadvertent prodrug activation in normal tissues, which could otherwise lead to off-target cytotoxicity.³⁷ In this context, we first evaluated human IDO as a potential alternative catalyst for IAA activation, and we confirmed that the IDO-IAA enzyme-prodrug pair exerts cytotoxic potency against cancer cells through either extracellular supplementation or intracellular overexpression. Nevertheless, direct expression or supplementation of full-length IDO is not therapeutically viable, as the enzyme also plays a critical role in immune regulation and may introduce unintended effects.

The rationale for specifically engineering an IDO variant that can bypasses the kynurenine pathway was driven by the necessity to eliminate native IDO functions. Previous research has established that IDO expression inactivates surrounding immune cells through the dual mechanisms of tryptophan depletion and kynurenine metabolite accumulation.^38,39 This metabolic environment is particularly detrimental to T cells, which often undergo cell-cycle arrest in response to tryptophan scarcity.⁴⁰ In the context of oncology, cancer cells often utilize this immunosuppressive ability by upregulating IDO levels, which directly correlates with poor patient prognosis.^41,42 To date, IDO has emerged as a primary therapeutic target, and suppressing its canonical activity is a recognized therapy in modern oncology.^43,44

Therefore, the subsequent research aims to aimed to engineer IDO variants with modified structural features while preserving oxidative activity toward IAA. To achieve this, we designed, screened, and characterized four truncated IDO constructs containing the heme-binding catalytic region. In the subsequent kynurenine decomposition activity assay, none of the four truncates produced physiological enzymatic activity as full-length IDO_WT does. However, under oxidative catalytic conditions, only the truncate IDO_268C exhibited a relatively strong fluorescent signal in the Amplex Red assay. Consistently, overexpression of IDO_268C in transfected A549 and MCF-7 cells resulted in effective elimination of cancer cells following IAA treatment. The catalytic efficiency of IDO_268C is particularly critical, as it allows for prodrug activation at lower IAA levels. This reduces systemic exposure and minimizes off-target cytotoxicity. Furthermore, the efficacy of the IDO_268C and IAA system also depends largely on the drug concentration and exposure time. Our results show that IAA has a favorable dosing window between 0.5 and 2.0 mM (Figures 3A and 3B). At levels above 3.5 mM, IAA causes cell death even without the enzyme. This intrinsic toxicity sets a limit for safe dosing. From a safety perspective, IAA exhibits a favorable dosing window. In this safe range, cell death only occurs when the enzyme activates the prodrug. This synergistic effect allows for selective tumor killing.

The relationship between truncation length and catalytic activity is often unpredictable. In many cases, deleting more residues leads to structural collapse. However, our enzyme activity data and cell experiments show that IDO_268C, the shortest variant, outperforms the longer constructs. Our CD data clearly demonstrate that the extra five residues in IDO_263C (263–267) exert a structural interference effect (Table 2; Figures S2 and S3). In the full-length protein, these residues are stabilized by the complete tertiary structure. However, in the truncated form, they likely introduce steric strain or unfavorable intramolecular interactions that trigger a transition from α helices to β sheets or β turns. Since IDO is strictly dependent on the helical bundle to shield the heme pocket, such a conformational change in IDO_263C likely disrupts the environment for IAA oxidation. However, IDO_268C avoids this structural interference. By removing the 263–267 segment, the protein adopts a more compact helical arrangement. This explains why IDO_268C is more efficient at catalyzing IAA oxidation.

The presence of H₂O₂ during the reaction introduces an additional challenge. It is necessary for the catalytic cycle, but it also triggers suicide inactivation of the heme group.³² Previous studies confirm that H₂O₂ can attack the distal heme pocket and damage the protein scaffold. We found that IDO_268C works effectively by prioritizing catalytic speed. Its high initial reaction velocity allows for a rapid burst of ROSs before the oxidative damage accumulates (Figure 1B). This mode of action is highly effective in a cellular environment. Even if the enzyme eventually degrades, the intracellular oxidative stress generated in the first few hours is sufficient to trigger apoptosis.

We also verified whether the cell culture environment negatively affects with the catalytic function of IDO and IDO_268C. We measured the enzyme’s activity directly in the presence of cells and DMEM medium. Both enzymes remained functional for up to 48 h, despite the natural loss of activity at 37°C (Figure S6). This confirms that the complexity of the cell culture system does not compromise the enzyme’s performance. Therefore, the observed cytotoxicity is a direct result of this sustained enzymatic activity.

Among enzyme-prodrug strategies, gene-directed enzyme-prodrug therapy (GDEPT) delivers therapeutic genes encoding prodrug-activating enzymes to target cells, thereby conferring local catalytic activity.⁴⁵ Notably, several studies on the classical HRP-IAA system have applied a similar principle, using endogenous enzymes to achieve localized prodrug activation.^{7,10,18,46,47} A major challenge of this approach is achieving sufficient gene delivery to tumor cells to enable effective prodrug activation in their vicinity.^48,49 Since not all tumor cells are effectively transfected, the efficacy of GDEPT largely relies on the bystander effect, where activated metabolites diffuse into adjacent non-transfected cells to broaden cytotoxicity.⁵⁰ This effect is particularly valuable in solid tumors, where extreme conditions such as hypoxia may otherwise limit enzyme activity and therapeutic outcomes.^51,52 To model this in vitro, IDO_268C-transfected and non-transfected cells were evenly spread at a 1:1 ratio in a culture dish, and IAA was added after cell adhesion. We observed that supplementation with 1 mM IAA caused a decrease in most transfected cell colonies, while increasing the IAA concentration to 2 mM resulted in cell death across all groups (data not shown). Although this experiment is simple and does not fully replicate in vivo heterogeneity, it confirms that the IDO_268C-IAA system can elicit a robust bystander effect, similar to the conventional HRP-IAA approach. This result emphasizes the potential of IDO_268C-IAA for further optimization and clinical development.

The targeting of the enzyme to the tumor site can be achieved not only via GDEPT but also through antibody-directed enzyme-prodrug therapy (ADEPT) approaches.^45,53,54 The unique therapy of ADEPT delivering an enzyme rather than a toxin to tumor cells, which solves the problem of uncontrolled distribution and accessibility of toxins. In general, ADEPT and GDEPT cannot share the same enzyme-prodrug system, because prodrug activation occurs extracellularly in ADEPT, while it occurs intracellularly in GDEPT.⁵⁵ To explore the potential of ADEPT in our system, we first simulated extracellular prodrug activation by supplementing exogenous IDO_268C protein and IAA to the culture medium. This treatment induced concentration-dependent cell death, consistent with reports that extracellularly generated ROSs can trigger apoptosis and mediate cytotoxicity in various in vitro and in vivo models.^56,57 Taken together, these results suggest that an ADEPT strategy using IDO_268C/IAA is feasible. However, previous studies have indicated that extracellular ROS generated by HRP-IAA in lymphocyte cultures can be neutralized by adding antioxidant enzymes, abolishing the cytotoxic effect.⁵⁸ Therefore, potential extracellular enzyme neutralization should be carefully considered when applying ROS-based ADEPT therapy.

Enzyme-prodrug systems based on IDO_268C-IAA have demonstrated promising effectiveness. At this stage, our evidence is centered on biochemical characterization and in vitro cytotoxicity. The specific impact of this variant for immune-mediated process remains to be experimentally validated in more complex co-culture or animal models. Additionally, the future developments could focus on optimizing enzyme active sites to enhance prodrug activation efficiency. For instance, de novo design of amino acid residues surrounding the heme active site could yield proteins with higher specificity, improved stability, and easier expression. This is particularly relevant because, compared with wild-type full-length IDO_WT, the intracellular expression of IDO_268C is relatively low. In parallel, complementary strategies aimed at increasing cytotoxicity or improving drug release efficiency also hold promise, such as using IAA derivatives. Indeed, several studies have confirmed effective combinations of HRP with IAA derivatives, suggesting potential avenues to optimize IDO-based prodrugs.^{10,59,60,61,62} Beyond molecular optimization, synergistic approaches combining different prodrug systems or other cancer therapies may further enhance efficacy. For example, introducing more than one suicide gene simultaneously allows released cytotoxic drugs to act through different mechanisms, producing synergistic enhancing effects on tumor cells.^63,64 Similarly, combining suicide gene therapy with radiotherapy or immunotherapy has been shown to enhance anti-tumor effects beyond either strategy alone, as demonstrated in HRP-IAA studies.⁶⁵ Combining radiotherapy with hTERTp-HRP-IAA could overcome radio resistance in laryngeal squamous carcinomas cells and amplify the killing effect in targeted cancer cells.⁶⁶ The same research team subsequently explored combining suicide gene therapy with immunotherapy.⁶⁷ Specifically, treatment with AdhTERTHRP-IAA together with AdCMVmIL-12 effectively suppressed tumor cell growth and improved host survival in the Lewis lung carcinoma model without major systemic adverse effects.

In summary, this study has developed and validated a dioxygenase variant, IDO_268C. This enzyme converts the prodrug IAA into cytotoxic metabolites to effectively eliminate tumor cells. It is worth noting that IDO_268C operates without engaging native IDO-mediated immunoregulatory pathways, offering a distinct advantage. This enzyme-prodrug combination demonstrates promising potential for application in both ADEPT and GDEPT strategies and may enable more efficient and safer targeted cancer therapies.

Materials and methods

Protein expression and purification

Protein was expressed and purified using standard methods.^39,68,69 The IDO_WT and truncated proteins (Homo sapiens, GenBank: NM_002164.6), which were cloned into pET-28a vector, were expressed in Escherichia coli BL21 (DE3) cells with codon optimization (Tsingke Biotechnology, Beijing, China). The E. coli cells were grown in LB medium with 100 μg/mL kanamycin at 37°C. When the OD600 nm of the culture reached ∼0.6–0.8, the temperature was cooled to 16°C and incubated overnight, supplemented with 7 μM hemin chloride and 0.5 mM IPTG. The cells were harvested by centrifugation at 4°C (6,000 × g, 20 min), and resuspended in 30 mL of buffer A (20 mM Tris, 300 mM NaCl, pH 8.0) supplemented with 7 μM hemin to ensure protein incorporation. The lysate was clarified by centrifugation at 4°C (12,000 × g, 30 min), and the supernatant was loaded into a Ni-NTA (Nickel-Nitrilotriacetic acid) column (Qiagen, Crawley, UK) equilibrated with buffer A. The red target proteins were eluted with buffer A containing 150 mM imidazole, then concentrated using a 30-kDa molecular-weight cutoff Amicon Ultra centrifugal filter (Millipore, Billerica, USA). The His-tag was removed by loading the digested mixture to Ni-NTA column. The mixture was loaded into anion exchange column with buffer A and further purified by size-exclusion chromatography (SEC) on a Superdex 75 column (GE Healthcare Biosciences, Milwaukee, WI, USA) in SEC buffer (20 mM MES, 25 mM NaCl, pH 6.0). The final pool had an A₄₀₆/A₂₈₀ absorption of ∼80%, indicating high heme occupancy. The purified red proteins were stored at −80°C until usage. In addition, HRP protein were obtained from Solarbio (Beijing, China).

Enzyme activity assay

IDO hydrolysis activity and kynurenine production were determined colorimetrically as described by Takikawa with a minor modification.⁷⁰ Briefly, the mixture was initiated by adding 0.5 mL of catalase suspension to 0.5 mL of incubation medium composed of 500 μM L-tryptophan, 10 mM ascorbic acid, 10 μM methylene blue, and 50 mM potassium phosphate buffer (pH 6.5). The reaction was carried out at 37°C for 40 min and terminated by adding 200 μL of 30% (w/v) trichloroacetic acid. After incubation at 50°C for 30 min, the mixtures were centrifuged at 3,000 × g for 20 min. A total of 800 μL of supernatant was transferred into a 96-well microplate and mixed with equivalent 1% (w/v) p-dimethylaminobenzaldehyde in acetic acid. The absorbance at 480 nm for the yellow pigment derived from kynurenine was determined using a multi-detection microplate reader (Spectra Max i3x, Molecular Devices, USA).

To assess catalytic activity toward hydrogen peroxide, the SignalUp Ultra-Sensitive ELISA Assay Kit with Amplex Red (Beyotime Biotechnology, Shanghai, China) was used. In this assay, the enzyme catalyzes the conversion of hydrogen peroxide into the red fluorophore resorufin in an H₂O₂-dependent reaction. A total of 20 μL of protein samples (∼30 μg) were plated in triplicate in a 96-well plate and mixed gently with 100 μL of Amplex Red working solution (92.75 μL of enhancer solution, 1 μL Amplex Red substrate solution, and 6.25 μL stable peroxide solution). After incubation in the dark for 10 min at 37°C, enzyme activity was quantified in the multi-detection microplate reader based on fluorescence emission at 600 nm upon excitation at 540 nm. The program was set with a pre-detection oscillation of 30 s, a detection interval of 30 s, and a total of 20 cycles.

To further characterize the catalytic oxidation capacity of IDO and its variants through continuous monitoring, a luminol assay was conducted as a complementary approach. The assay used a Promega GloMax 96 plate luminometer (Promega Corporation, Madison, WI, USA) equipped with an automatic injection system to control the injection volume and timing. IDO or variant proteins (7 mg/mL) or heme solution (0.7 μM) were added to the buffer (20 mM MES, 25 mM NaCl, pH 6.0) containing 0.2 mM luminol solution and incubated at room temperature for 30 min. Subsequently, 50 μL of mixture and 50 μL of H₂O₂ (10 mM) were injected, and the chemiluminescence signals were immediately recorded on the luminometer.

CD spectroscopy assay

The secondary structure of the purified proteins was analyzed via CD spectroscopy. Samples were prepared at a concentration of 0.15 mg/mL in PBS buffer and placed in a quartz cuvette with a 1-mm path length. Data collection was performed across a wavelength range of 200–260 nm with a step size of 1 nm. For each sample, three consecutive scans were recorded and averaged to generate the final spectra after subtracting the baseline of the PBS buffer.

Cell culture and transfection

The enzyme-prodrug therapy using IDO variants or HRP in combination with IAA was evaluated in human breast cancer MCF-7 and human non-small cell lung cancer A549 cell lines. MCF-7 and A549 cells (ATCC, Manassas, VA) were cultured in high-glucose DMEM lacking glutamine (Gibco, Life Technologies, USA) supplemented with 10% FBS and incubated at 37°C under 5% CO₂. The mammalian codon-optimized IDO_WT and IDO_268C genes were cloned into the pcDNA3.1-6×His-tag-mScarlet fluorescent protein vector (Tsingke), allowing resultant protein capture with His-tag and monitoring of transfection efficiency by cellular fluorescence. Transient transfections were performed using Lipofectamine 2000 (Invitrogen, USA) according to the manufacturer’s protocol.

Previous studies utilizing HRP-IAA systems have established an optimal experimental window of 1.2 μg/mL HRP combined with 0.1–4 mM IAA, which served as a reference framework to guide our experiments.^7,12,13,71 Accordingly, we employed a checkerboard approach in 96-well plates containing cultured A549 cells to find the best concentration pairs. 3,000 cells per well were seeded in a 96-well plate and cultured for 24 h. In this setup, IAA concentrations ranged from 0.05 to 4.5 mM (0, 0.05, 0.2, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, and 4.5 mM) across the plate (Sigma-Aldrich, St. Louis, MO). Meanwhile, IDO variants (1, 2, 4, 8, 16, 25, 50, and 100 μg/mL) were added along the vertical axis. This grid layout allowed us to see how different combinations affected cell states at a glance. We also used IAA or IDO variants alone as negative controls. The HRP (Sigma-Aldrich) at a concentration of 1.2 μg/mL was used as positive control. The cells were incubated for 24, 48, and 72 h. Finally, cell morphology was captured using a microscope camera (Nikon ECLIPSE Ts2, Tokyo, Japan).

Cell viability assay

Cytotoxicity was assessed using the MTT (3-(4,5)-dimethylthiahiazo(-z-y1)-3,5-di-phenytetrazoliumromide) assay (Beyotime) following drug treatment. After treating with the drug exposure for 48 h, fresh MTT solution (5 mg/mL) was added to each well and incubated at 37°C and 5% CO₂ for 4 h to assess cellular metabolic activity as an indication of cell viability. The supernatant was removed, and formazan crystals were solubilized in 100 μL of solubilization solution on a shaker at low speed for 10 min. Subsequently, the absorbance at 570 nm was measured using a microplate reader, and cell viability was expressed as a percentage relative to the untreated control. In addition, adherent live cells were counted using a hemocytometer. The percentage of cell viability was calculated as number of living cells in the experimental group/number of living cells in the control group × 100%.

Free radical determination

To evaluate intracellular ROS generation, A549 cells were first treated with IAA and IDO variants (or HRP) for 6 h. Following the manufacturer’s instructions, the cell-permeable probe DCFH-DA (2′,7′-dichlorodihydrofluorescein diacetate, MedChemExpress, China) was diluted in serum-free medium (1:1,000) to a final concentration of 10 μM. The treated cells were collected by centrifugation, washed several times with PBS, then resuspended in serum-free medium containing DCFH-DA. Cells were gently inverted and mixed every 5 min during incubation at 37°C for 20 min to ensure the probe contact with cells. Finally, cells were washed with serum-free medium to remove excess DCFH-DA that had not entered the cells. Intracellular ROS levels were quantified by a microplate reader (excitation/emission = 488/525 nm).

Western blot analysis

The transfected cell lysates were prepared and resolved on 4%–20% gradient SDS-PAGE gels (40 μg per lane). Samples were transferred to polyvinylidene fluoride (PVDF) membranes, which were blocked with 5% non-fat milk in TBST (Tris-buffered saline with Tween 20). The membranes were incubated with the primary antibodies against 6 × his-tag at a dilution of 1:10,000 (#66005, Proteintech, China). Protein bands were visualized by Tanon 5200 multi-gel analysis system (Tanon Technology., Shanghai, China).

Data and code availability

All data are available upon request to the corresponding authors.

Acknowledgments

We would like to thank the Imaging Facility (Mr. Jun Chen and Miss. Chunhua Zhang) of the National Center for Protein Sciences Beijing for their assistance. This work was supported by Jing-Jin-Ji Regional Integrated Environmental Improvement-National Science and Technology Major Project (grant No. 2026ZD1218801).

Author contributions

L.L. designed and performed most of the biochemical experiments. W.L. prepared the related figures. S.Y. supervised the project. L.L. and S.Y. wrote the manuscript.

Declaration of interests

The authors declare no competing interests.