Engineering an indoleamine 2,3-dioxygenase immunotherapy via selective cysteine-to-serine mutations

Abstract

Indoleamine 2,3-dioxygenase is an immunomodulatory enzyme that shows great promise when delivered exogenously as a protein therapeutic. However, IDO activity is under complex redox control, mediated in part by multiple cysteine residues within its primary sequence. We have characterized three IDO mutants in which solvent-accessible cysteine residues were mutated to chemically-similar serine residues, “IDO_C4S4” with C112S, C159S, C206S, and C308S mutations and “IDO_C5S3” with C112S, C159S, and C308S mutations based on prior reports that C206 is necessary for catalytic function, and IDO_C0S8, in which all cysteine residues were mutated to serines. IDO_C0S8 was expressed in poor yield and demonstrated less than 1% activity when compared to wild-type IDO. In contrast, IDO_C4S4 and IDO_C5S3 demonstrated robust enzymatic activity, though IDO_C5S3 had a slower V_max than wild-type and IDO_C4S4. Computational predictions and experimental measurements suggested a high degree of structural similarity between the wild-type IDO and IDO_C4S4, with subtle perturbation of α-helical content for IDO_C5S3. The structure of IDO_C0S8 was predicted to be significantly different than that of wild-type IDO. IDO_C4S4 and IDO_C5S3 were more stable than wild-type IDO over time at physiological, ambient, and reduced temperatures, likely due to diminished oxidation of the mutant IDO forms. Based on the increased V_max and robust thermal stability of IDO_C4S4, we fused it to the anchoring moiety galectin 3, to evaluate its effectiveness in a mouse model of psoriasis. The IDO_C4S4-galectin-3 fusion blunted the rate and severity of disease as compared to wild-type IDO-galectin-3 fusion. When compared to historical data with Cys-Ala IDO mutants, this study highlights the importance of employing amino acid substitution according to similarity in isosteric and isostructural shape to advance IDO as an immunomodulatory therapeutic.

---

Design, System, Application

Enzymatic stability poses a considerable challenge to translation of biotherapeutics, as changes in conformational state can disrupt catalytic activity and functionality. If protein misfolding is driven by key amino acids, mutation into a more stable residue can reduce protein instability. We present a selective cysteine to serine mutation strategy for the immunomodulatory enzyme indoleamine 2,3-dixoygenase (IDO). By targeting only the most solvent accessible cysteines of IDO, the IDO mutants shared a high degree of secondary structure similarity, which in turn preserved enzymatic activity. These mutants also improved thermal stability during warm and cold storage, while also improving its therapeutic efficacy in a mouse model of psoriasis as compared to wild-type IDO. Computational predictions allowed for pre-screening of candidate mutants, where several key metrics predicted low enzymatic function in non-selective Cys-Ser mutations, which was seen after microbial expression. Not only does our work demonstrate improvements to a promising immunotherapeutic, but it also highlights the importance of employing amino acid substitution judiciously, and choosing replacement amino acids according to similarity in structural shape.

Introduction

Indoleamine 2,3-dioxygenase (IDO), an enzyme catabolizing tryptophan,¹ is an attractive immunoregulatory therapy. For example, IDO provided via overexpression in transferred cells can ameliorate type 1 diabetes^2,3 and psoriasis.⁴ mRNA delivery of IDO has shown efficacy in the treatment of T cell mediated diseases.⁵ Exogenously supplied IDO protein confers dendritic cell resistance to LPS stimulation⁶ and is a robust anti-inflammatory.⁷ Yet, whether delivered as a gene or as protein, IDO enzymatic activity is under complex redox control.^1,8–13 Immune cell activation in diseased tissue results in increased redox factor production, which can cause oxidative and nitrosative stress, redox signaling disruption, and molecular damage.^14,15 IDO cysteine residues, of which there are eight, have been reported to contribute in part to enzymatic inactivation in high oxidant conditions.^8,16,17

Cysteine residues often pose problems in protein therapeutic development. Cysteines can undergo oxidation into disulfides,¹⁸ which can lead to protein misfolding and aggregation.¹⁹ The sulfhydryls in cysteines can also oxidize into sulfinic, sulfenic, and sulfonic acids,²⁰ which can disrupt co-factor binding (e.g., metal ion coordination)²¹ and hinder substrate binding.²² Cysteine oxidation can furthermore change protein conformation, resulting in altered if not inactivated function.^23,24 Given the complex character of cysteines, it can be advantageous to remove such residues, if possible to do so without significantly perturbing protein form and function.

In the case of IDO, the role of each of its eight cysteine residues is poorly defined. IDO does not contain any disulfide bridges, and cysteines in proximity to the heme active site are not thought to be directly necessary for enzymatic activity.^17,25 Chemically modifying the cysteines of IDO can lead to inhibition. Selenyl–sulfide bonds formed between cysteine and the organoselenium drug ebselen inhibit IDO by changing the conformational state and disrupting enzymatic activity.²⁶ Later work also showed multimerization and inactivation of IDO after oxidation through the formation of protein-centered radicals, though the specific amino acid residues involved have not yet been identified.⁸ Mutating cysteines to alanines can also reduce enzymatic activity. In particular, IDO activity decreased by more than 50% when C85, C206, C272, and C335 were individually mutated into alanines (Table S1).²⁷

Mutating cysteines to alanines can result in unexpected changes to protein form and function due to the differences in their side-chain chemistry.^28,29 In contrast, cysteine and serine are similar in isosteric and isostructural shape, differing only by atomic substitution of the sulfur in cysteine with oxygen in serine.^30,31 Cys-Ser mutations have shown successes in engineering more stable protein therapeutics. In one instance, surface Cys-Ser mutants of galectin-1 resisted oxidative inactivation by eliminating susceptibility to covalent cross-linking, thus retaining activity in oxidative environments.³² Furthermore, Cys-Ser substitutions on the surface of IL-18 improved solubility and reduced aggregation.³³ In contrast, in the copper binding proteins human ATP7B and E. coli CueR metalloregulator, Cys-Ser mutations were not tolerated at sites where cysteine coordinated copper ions and diminished enzymatic activity.³⁴ Thus, judicious investigation of candidate mutation sites is necessary to balance protein biophysical properties for improved stability and function.

We investigated Cys-Ser IDO mutants to determine the impact of those mutations on enzymatic function, secondary structure, stability under different storage conditions, and therapeutic effectiveness. We limited mutations to C112, C159, C206, and C308, which are the cysteine residues predicted to be most solvent accessible (i.e., on the “surface” of the protein) based on crystal structure and in silico predictions. We generated two mutants – IDO_C4S4, where all surface cysteines were mutated into serines, and IDO_C5S3, where C112, C159, and C308 were mutated into serines but C206 was maintained, based on a prior report suggesting its role in enzymatic function.²⁷ We found that both Cys-Ser mutants generally retained IDO's enzymatic function and structure. IDO_C4S4 was observed to be the most stable variant at reduced (4 °C) and body (37 °C) temperatures and outperformed wild-type IDO domain in the fusion protein IDO-galectin3 (IDO-Gal3) in vivo in an imiquimod (IMQ)-induced psoriasis mouse model.

Experimental

IDO cysteine mutants design and production

Selective mutations of cysteines into serines were formed at sites 112, 159, 206, and 308 for IDO_C4S4 and at 112, 159, and 308 for IDO_C5S3. Total mutation of cysteine to serine was formed at all sites for IDO_C0S8. The genes encoding these mutant IDOs were inserted into pET-28a (+) vectors between NdeI and Xhol restriction sites, and were produced by and purchased from Genscript. The plasmids were then transformed into Origami B(DE3) E. coli competent cells (Sigma-Aldrich) for protein expression, with successful transformation confirmed by Sanger sequencing (Genewiz). Protein expression was performed using the autoinduction method.³⁵ Bacteria were pelleted and lysed, and IDO was purified from the lysis supernatant through a HiTrap TALON cobalt column (Cytiva) and size-exclusion chromatography (GE Life Sciences). Purity was ensured by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Coomassie staining. Yield for IDO_C5S3 and IDO_C4S4 mutants was ∼2.8 mg and ∼2.5 mg of protein per liter of growth volume, respectively. Concentration was determined via the Pierce Bradford Protein Assay Kit (ThermoFisher), prior to use. Absorbance spectra were captured by NanoDrop ND-1000 Spectrophotometer. Wild-type IDO production was performed following the same method as above. IDO-galectin3 (Gal3) fusion mutant were engineered through a combination of the IDO_C4S4 mutant and our previously established production of IDO-Gal3.⁷ Yield for IDO_C4S4-Gal3 mutant was ∼3.5 mg of protein per liter of growth volume.

Enzymatic activity

Wild-type IDO_C8S0, IDO_C5S3, and IDO_C4S4 were stored at 4 °C, room temperature, and 37 °C, and were sampled periodically to test for changes in enzymatic activity, with no additional steps taken to reduce oxidizing conditions. 17.88 pmoles of IDO was combined with tryptophan, catalase, and the electron donors methylene blue and ascorbic acid. The absorbance of N-formyl-kynurenine (NFK), a precursor to kynurenine, was read at 321 nm for ten minutes, and specific activity (unit of pmol NFK per min per pmol IDO) was determined from the slope of absorbance at 321 nm and the extinction coefficient of NFK, with a pathlength of 0.32 cm. Enzymatic activity for all timepoints were normalized against t = 0 IDO activity, and activity half-life was calculated via best fit. The kinetic parameters of wild-type IDO_C8S0, IDO_C5S3, and IDO_C4S4 were determined by the Michaelis–Menten equation after varying tryptophan concentrations. Activity half-life and IDO kinetic parameters were compared using a one-way ANOVA and Tukey's multiple comparison test (Graphpad Prism).

In silico analysis

Predicted secondary structures of wild-type IDO_C8S0, IDO_C5S3, and IDO_C4S4 were generated using AlphaFold v3.³⁶ Structural similarity was assessed by rigid FATCAT pairwise structural alignment^37,38 (PDB) comparison of root mean square deviation of atomic positions (RMSD) and template modeling (TM) score. Predicted secondary structure composition for each protein was calculated using STRIDE,³⁹ and then percentage of each structural component was compared between wild-type IDO_C8S0 vs. IDO_C5S3 and IDO_C4S4. Analysis was also performed for one variant where all cysteines were mutated into serines (IDO_C0S8) and another variant where all cysteines were mutated into alanines (IDO_C0A8). Residue distance between r85 and r129 was measured using Mol* Viewer (RCSB PDB).⁴⁰

Galectin 3 binding affinity

Affinity of IDO_C4S4-Gal3 for lactose was determined using affinity chromatography in an AKTA Pure chromatography system (GE Life Sciences) equipped with consumer-packable glass column (GE Life Sciences) packed with α-lactose agarose affinity resin (Sigma-Aldrich). Proteins were eluted with a linear gradient of β-lactose (Sigma-Aldrich) in phosphate buffer.

Secondary structure

Protein secondary structure characteristics were measured using Fourier-transform infrared (FT-IR) spectroscopy (PerkinElmer, USA) and circular dichroism (CD). CD was recorded on a Chirascan™-plus qCD (Applied Phosophysics) with a quartz cuvette of path length 1 mm from wavelengths between 190 and 260 nm with 2 nm intervals. Each sample contained purified protein at ∼0.1 mg mL⁻¹ and both sample and baseline contained 10 mM phosphate, pH 7.4, and 140 mM KF. Samples were scanned 5 times with the average spectra reported. For FT-IR, wild-type IDO_C8S0, IDO_C5S3, and IDO_C4S4 in distilled water were lyophilized from frozen. Samples were transferred onto the attenuated total reflectance (ATR) stage and then powder samples were compressed by pressure device to give uniform contact on the ATR crystal. The absorbance spectrum was measured over the range of 900–4000 cm⁻¹. The baseline of each spectrum was corrected using the Baseline Correction function within the PerkinElmer Spectrum software. Absorbance at 1650 cm⁻¹ represents amide I band, while absorbance at 1550 cm⁻¹ corresponds to the amide II band. Secondary structure comparisons were made from the peak shape and intensity of amide I and II bands.

Imiquimod-induced psoriasis

All studies were conducted with the approval of the University of Florida Internal Animal Care and Use Committee, in compliance with the United States Public Health Service policy on Humane Care and Use of Laboratory Animals. An inflammatory mouse model, imiquimod-induced psoriasis,^7,41 was used to confirm therapeutic efficacy of IDO-Gal3 mutants. After shaving and removal of remaining hair with depilatory cream, 62.5 μg of 5% imiquimod cream (Patterson Veterinary Supply, cat. num. 07-893-7787), a potent inflammatory agent, was applied to the backs of female C57/BL6J mice (8 weeks old, Jackson labs, PBS: n = 8, all others: n = 4). Mice were clinically observed and scored daily using a modified psoriasis area and severity index (PASI) score to determine disease severity. Erythema (redness), scaling, and thickening were scored independently and assigned a score between 0 to 4: 0, none, 1: slight, 2: moderate, 3: marked, 4: very marked. The cumulative score was reported as a measure of the severity of inflammation (scale 0–12). On the first day of clinical score emergence (day 3), five 10 μg doses of IDO_C4S4-Gal3, IDO-Gal3, or an equivalent volume of PBS were injected subcutaneously and evenly distributed across the back. Application of imiquimod cream and clinical scoring was performed every day for a duration of 10 d. Mean clinical scores were compared using a one-way ANOVA and Dunnett's multiple comparisons test. Area under the curve was calculated for each animal, and compared using an unpaired t-test. Initial rate of change of scores was measured between the baseline score to plateau for each animal, and compared using an unpaired t-test. All statistics were performed using Graphpad Prism.

Results and discussion

Cysteine to serine IDO mutants are enzymatically active

We designed two different IDO mutants, one where the cysteines located at residues 112, 159, 206, and 308 were substituted for serines, “IDO_C4S4”, and a second at residues 112, 159, and 308, “IDO_C5S3” (Fig. 1A). Proteins corresponding to the predicted molecular weights of IDO_C4S4 and IDO_C5S3 were recovered in good yield and purity from microbial hosts (Fig. 1B). Soret bands (λ = 405–425 nm), which are used to identify the heme unit of IDO,⁴² had similar maxima for WT IDO, IDO_C4S4 and IDO_C5S3 (Fig. 1C). Wild-type IDO (“WT IDO_C8S0”) had a higher Soret band/protein (λ = 405/280 nm) peak ratio than that of IDO_C4S4 and IDO_C5S3, suggesting that there may be more heme bound to WT IDO. Both IDO mutant forms were enzymatically activity (Fig. 1D). There were no differences in K_M for either mutant relative to WT IDO; however, IDO_C5S3 had significantly slower V_max value than WT IDO_C8S0 and IDO_C4S4. This slower V_max and weaker Soret band of IDO_C5S3 are correlative, as heme binding to oxygen is required for the dioxygenase reaction of tryptophan.^17,25

Fig. 1 IDO Cys-Ser mutants. (A) Predicted IDO crystal structures highlighting the location of Cys residues in each IDO variant. Cysteines are highlighted in pink while cysteine to serine point mutations are highlighted in green. (B) SDS-PAGE gel of IDO_C5S3 and IDO_C4S4 showing a pure product after recombinant expression. (C) Absorbance spectra of WT IDO_C8S0, IDO_C5S3, and IDO_C4S4, with Soret peak around 405 nm. (D) Enzymatic parameters of WT and mutant IDO, compared between mutant and WT. *: P < 0.05, ns: no significance.

In silico analysis and in vitro secondary structure characterization display a high degree of similarity between WT IDO and cysteine-based mutants

Computational models predicted that the Cys-Ser mutations in IDO_C5S3 and IDO_C4S4 would not significantly alter secondary structures compared to WT IDO. Composition of predicted secondary structure components showed no significant changes (Fig. 2A), and three-dimensional structural predictions between WT IDO and IDO mutants had a high degree of similarity (Fig. 2B). Likewise, CD and FT-IR measurements showed that the secondary structures of IDO_C5S3 and IDO_C4S4 were highly similar to that of WT IDO. The CD spectra of all three proteins were characterized by minima at 208 and 222 nm, along with a maxima at 192 nm (Fig. 2C), consistent with a protein that is predominantly folded into alpha-helices. Additionally, all three proteins demonstrated a prominent peak at 1645–1655 cm⁻¹ in FT-IR spectra, indicating a structure predominated by alpha-helical coiled-coils (Fig. 2D), matching characterizations of IDO in the literature.^17,43 Of note, IDO_C5S3 displayed a slight blue shift in the maxima of the primary amide band to 1652.5 cm⁻¹ (Fig. 2E), which suggests a subtle shift in secondary structure,⁴⁴ and which correlates to the slower V_max seen in this mutant.

Fig. 2 IDO mutants have similar in silico predicted and in vitro secondary structures. (A) Secondary structure composition comparison and (B) pairwise structural alignment comparison between AlphaFold predictions of WT IDO and IDO Cys-Ser mutants. (C) Circular dichroism spectra of IDO proteins over the range λ = 180 to λ = 280 nm at 20 °C. (D) Fourier transform infrared spectroscopy (FT-IR) absorbance spectra of IDO and Cys-Ser mutants. Dashed lines at indicate primary and secondary amines. (E) Inset of FT-IR trace, showing primary amine.

Cysteine to serine mutants improve IDO thermal stability at physiological and reduced temperatures

To assess the effect of the Cys-Ser mutants on thermal stability of IDO, we measured the enzymatic function of WT IDO and the two mutants after warm (37 °C), room temperature (25 °C), and cold (4 °C) storage. Cys-Ser mutants demonstrated higher enzymatic activity over time when compared to WT IDO (Fig. 3A). At physiological temperatures, IDO_C4S4 extended enzymatic half-life by ∼50% as compared to WT IDO_C8S0, whereas IDO_C5S3 only extended stability by ∼20% (Fig. 3B). At room temperature, IDO_C4S4 increased half-life by ∼200% compared to WT IDO_C8S0, while IDO_C5S3 only improved the half-life on the order of hours rather than days (Fig. 3B). Cold storage showed the best improvement in enzymatic stability, as both mutant forms increased the enzymatic half-life by 200% (∼1 week increase) as compared to WT (Fig. 3B). Consistent with this result, cysteine point mutations have been previously reported to improve cold storage and thermostability,^45,46 suggested to be due to slowed or absent cysteine oxidation.⁴⁷ Such mutations, however, would not be expected to affect the susceptibility of the heme to oxidative degradation, as has been shown for IDO and other proteins.^8,48

Fig. 3 Cys-Ser IDO mutants increase stability of IDO. (A) WT vs. IDO_C5S3 vs. IDO_C4S4 normalized enzymatic activity over time after storage at 37 °C, 25 °C, and 4 °C. (B) Mutant IDO activity half-life also increases for all temperatures, as compared to WT IDO_C8S0. *: P < 0.05, ****: P < 0.0001.

Cysteine to serine IDO mutations improve therapeutic efficacy in imiquimod-induced psoriasis

We created a fusion of IDO_C4S4 and galectin 3 (Gal3) (Fig. 4A) based on our prior work showing that IDO-Gal3 is effective for treating local inflammation in various diseases.^7,49 IDO_C5S3 was excluded from these studies based on its diminished V_max and weaker thermal stability when compared to IDO_C4S4. IDO_C4S4-Gal3 was expressed and recovered in high yield and purity (Fig. 4B). Tryptophan catabolism was comparable between IDO_C4S4 and IDO_C4S4-Gal3 (Fig. 4C). The Gal3 domain endowed IDO_C4S4-Gal3 with binding affinity for immobilized lactose that IDO_C4S4 lacked, as expected (Fig. 4D).

Fig. 4 Anchored Cys-Ser IDO mutant improves therapeutic efficacy. (A) Schematic of IDO_C4S4-Gal3, with the IDO arm shown in yellow and the Gal3 arm shown in blue. Cysteines are highlighted in pink while cysteine to serine point mutations are highlighted in green. (B) SDS-PAGE gel of IDO_C4S4-Gal3, showing a pure product after recovery. (C) Tryptophan catabolism of IDO_C4S4 and IDO_C4S4-Gal3, as shown by absorbance at 321 nm over time. (D) Lactose binding affinity of IDO_C4S4 and IDO_C4S4-Gal3, showing preserved galectin 3 functionality in the fusion mutant. (E) Timeline of IMQ-induced psoriasis mouse model. Subcutaneous injection sites are depicted as purple stars. (F) Mutant IDO_C4S4-Gal3 reduces clinical scores in psoriatic mouse model as compared to PBS treatment, while IDO-Gal3 does not achieve statistical significance of cumulative clinical scores. (G) Area under the curve of clinical scoring is significantly reduced in IDO_C4S4-Gal3 as compared to PBS. (H) IDO_C4S4-Gal3 blunts peak of disease and slows rate of inflammation as compared to IDO-Gal3. *: P < 0.05, **: P < 0.005, ns: no significance.

To test the in vivo immunomodulatory capabilities of our IDO Cys-Ser mutant, we used the IMQ-induced psoriasis mouse model, as previously described.⁷ Mice were treated one time with 50 μg subcutaneous injections of IDO-Gal3 or IDO_C4S4-Gal3, or with an equivalent volume of PBS vehicle control at the first sign of clinical symptoms (erythema, skin thickening, and skin scaling) (Fig. 4E). IDO_C4S4-Gal3 blunted disease progression of psoriasis (Fig. 4F). Treatment with the mutant led to significantly lower area under the curve for clinical scores relative to vehicle control (Fig. 4G). Treatment with the mutant also delayed peak of disease and slowed rate of inflammation compared to the existing IDO-Gal3 fusion (Fig. 4H). This improved effectiveness of the IDO_C4S4 mutant corresponded with its increased enzymatic stability and faster V_max.

Models predict the influence of cysteine mutations to IDO structure

Modeling of crystal structures predicted that Cys-Ala mutations would change the inter-residue distance between residue 85 and residue 129, which surrounds the enzymatic pocket.²⁷ This was suggested to change substrate binding through altered hydrophobic interactions, resulting in a loss of enzymatic activity. Modeling of IDO using AlphaFold v3 predicted that total protein architecture, as measured with RMSD and TM, was not altered in the IDO_C4S4 and IDO_C5S3 mutants (Fig. 5B), whereas it was altered in the Cys-Ala mutants reported previously (Fig. 5A). Likewise, AlphaFold v3 models predicted that that inter-residue distances would increase when all Cys were mutated to Ala, IDO_C0A8, while the selective mutations of IDO_C5S3 and IDO_C4S4 were not predicted to increase inter-residue distances (Fig. 5B). Notably, mutating all IDO Cys residues to Ser (IDO_C0S8) was predicted to significantly increase inter-residue distances, which we expected would lead to reduced activity. Indeed, the IDO_C0S8 mutant was recovered in poor yield and showed less than 1% activity when compared to IDO_C8S0, IDO_C4S4, and IDO_C5S3 (Fig. 5B). However, even when inter-residue distances were predicted to be similar, as with IDO_C5S3 and IDO_C4S4, inserting Ser residues into IDO could have some effect on enzyme activity, as suggested by the slower V_max for IDO_C5S3 than for wild-type and IDO_C4S4, despite IDO_C5S3 sharing C206 with wild-type, whereas IDO_C4S4 did not (Fig. 1D). Collectively, these data suggest that while computational models can be used to predict the tolerability of IDO structure to mutations, the impact on enzymatic activity is complex, depending on both the type and placement of the residues, and so must be arrived at empirically.

Fig. 5 Mutation of cysteines to serines preserves residue distance surrounding enzymatic site. (A) Predicted crystal structures of all Cys-Ser (left) and all Cys-Ala (right) mutations of IDO. Cysteine to serine point mutations are highlighted in green and cysteine to alanine point mutations are highlighted in orange. (B) Activity, predicted distance between site 85 and site 129, and pairwise structural alignment comparison for all mutations.

Conclusions

The IDO_C4S4 mutant demonstrated marked improvement in thermal stability and reaction velocity, which provided for a more effective protein therapeutic in the context of psoriasis. Improving reaction velocity and thermal stability of enzymes is often difficult, as protein properties in vitro versus in vivo can vary due to many different factors.^50,51 These improvements were made by mutating surface accessible cysteines to chemically-similar serines, while maintaining cysteines thought to be buried within the fold of IDO. In contrast, prior work using Cys-Ala mutations led to significant decreases in IDO activity.²⁷ Collectively, these data reinforce the idea that amino acid choice and site selection are critical for engineering improved mutant enzymes.

Author contributions

J. A. S., A. W., B. G. K., and G. A. H. conceived of this study. J. A. S., A. W., and A. S. conducted protein production and enzymatic activity assessment. R. A. C. and J. P. conducted secondary structure characterization. J. A. S. performed computational modeling and animal experiments. J. A. S. performed data analysis and wrote the draft of the manuscript. J. A. S., B. G. K., and G. A. H. finalized the manuscript. All authors approved the submitted version of this paper. Artificial intelligence, such as ChatGPT, did not write the manuscript or generate figures outside of AlphaFold v3 predictions, which have been disclosed.

Conflicts of interest

There are no conflicts to declare.

Data availability

Supplementary information (SI): amino acid sequences. See DOI: https://doi.org/10.1039/d5me00106d.

Amino acid sequences can be found in the supplement.

Acknowledgements

We would like to thank Dr. Brandon D. Wiles for his assistance with mathematical modeling. B. G. K. and G. A. H. disclose support for the research described in this study from the National Institutes of Health (R01 DE027301). B. G. K. discloses support from the National Institutes of Health (R01 DK091658, R01 DK098589, and R01 AI133623). G. A. H. discloses support from the National Institutes of Health (R03 EB019684 and R35 GM133697) and the National Science Foundation (DMR-1455201).

References

1. A. W. S. Yeung, A. C. Terentis, N. J. C. King and S. R. Thomas, Role of indoleamine 2,3-dioxygenase in health and disease, Clin. Sci., 2015, 129(7), 601–672.
2. Y. Zhang, R. B. Jalili, R. T. Kilani, S. S. Elizei, A. Farrokhi and M. Khosravi-Maharlooei, et al., IDO-Expressing Fibroblasts Protect Islet Beta Cells From Immunological Attack and Reverse Hyperglycemia in Non-Obese Diabetic Mice, J. Cell. Physiol., 2016, 231(9), 1964–1973.
3. R. B. Jalili, Y. Zhang, A. Hosseini-Tabatabaei, R. T. Kilani, M. Khosravi Maharlooei and Y. Li, et al., Fibroblast Cell-Based Therapy for Experimental Autoimmune Diabetes, PLoS One, 2016, 11(1), e0146970.
4. S. S. Elizei, M. Pakyari, M. Ghoreishi, R. Kilani, S. Mahmoudi and A. Ghahary, IDO-expressing Fibroblasts Suppress the Development of Imiquimod-induced Psoriasis-like Dermatitis, Cell Transplant., 2018, 27(3), 557–570.
5. L. L. Kenney, R. S. Y. Chiu, M. N. Dutra, A. Wactor, C. Honan and L. Shelerud, et al., mRNA-delivery of IDO1 suppresses T cell-mediated autoimmunity, Cell Rep. Med., 2024, 5(9), 101717.
6. E. Bracho-Sanchez, A. Hassanzadeh, M. A. Brusko, M. A. Wallet and B. G. Keselowsky, Dendritic Cells Treated with Exogenous Indoleamine 2,3-Dioxygenase Maintain an Immature Phenotype and Suppress Antigen-specific T cell Proliferation, J. Immunol. Regener. Med., 2019, 5, 100015.
7. E. Bracho-Sanchez, F. G. Rocha, S. K. Bedingfield, B. D. Partain, S. L. Macias and M. A. Brusko, et al., Suppression of local inflammation via galectin-anchored indoleamine 2,3-dioxygenase, Nat. Biomed. Eng., 2023, 7(9), 1156–1169.
8. M. Freewan, M. D. Rees, T. S. S. Plaza, E. Glaros, Y. J. Lim and X. S. Wang, et al., Human indoleamine 2,3-dioxygenase is a catalyst of physiological heme peroxidase reactions: implications for the inhibition of dioxygenase activity by hydrogen peroxide, J. Biol. Chem., 2013, 288(3), 1548–1567.
9. H. Ohtaki, H. Ito, K. Ando, T. Ishikawa, M. Hoshi and R. Tanaka, et al., Interaction between LPS-induced NO production and IDO activity in mouse peritoneal cells in the presence of activated Valpha14 NKT cells, Biochem. Biophys. Res. Commun., 2009, 389(2), 229–234.
10. D. Alberati-Giani, P. Malherbe, P. Ricciardi-Castagnoli, C. Köhler, S. Denis-Donini and A. M. Cesura, Differential regulation of indoleamine 2,3-dioxygenase expression by nitric oxide and inflammatory mediators in IFN-gamma-activated murine macrophages and microglial cells, J. Immunol., 1997, 159(1), 419–426.
11. C. Hucke, C. R. MacKenzie, K. D. Z. Adjogble, O. Takikawa and W. Däubener, Nitric oxide-mediated regulation of gamma interferon-induced bacteriostasis: inhibition and degradation of human indoleamine 2,3-dioxygenase, Infect. Immun., 2004, 72(5), 2723–2730.
12. M. Hill, R. Zagani, C. Voisine, C. Usal and I. Anegon, Nitric oxide and indoleamine 2,3-dioxygenase mediate CTLA4Ig-induced survival in heart allografts in rats, Transplantation, 2007, 84(8), 1060–1063.
13. H. Fujigaki, K. Saito, F. Lin, S. Fujigaki, K. Takahashi and B. M. Martin, et al., Nitration and inactivation of IDO by peroxynitrite, J. Immunol., 2006, 176(1), 372–379.
14. J. Lugrin, N. Rosenblatt-Velin, R. Parapanov and L. Liaudet, The role of oxidative stress during inflammatory processes, Biol. Chem., 2014, 395(2), 203–230.
15. R. R. Manoharan, A. Prasad, P. Pospíšil and J. Kzhyshkowska, ROS signaling in innate immunity via oxidative protein modifications, Front. Immunol., 2024, 15, 1359600.
16. A. Poljak, R. Grant, C. J. D. Austin, J. F. Jamie, R. D. Willows and O. Takikawa, et al., Inhibition of indoleamine 2,3 dioxygenase activity by H2O2, Arch. Biochem. Biophys., 2006, 450(1), 9–19.
17. H. Sugimoto, S. i. Oda, T. Otsuki, T. Hino, T. Yoshida and Y. Shiro, Crystal structure of human indoleamine 2,3-dioxygenase: catalytic mechanism of O2 incorporation by a heme-containing dioxygenase, Proc. Natl. Acad. Sci. U. S. A., 2006, 103(8), 2611–2616.
18. J. N. Cobley, 50 shades of oxidative stress: A state-specific cysteine redox pattern hypothesis, Redox Biol., 2023, 67, 102936.
19. L. J. Perry and R. Wetzel, The role of cysteine oxidation in the thermal inactivation of T4 lysozyme, Protein Eng., 1987, 1(2), 101–105.
20. D. Garrido Ruiz, A. Sandoval-Perez, A. V. Rangarajan, E. L. Gunderson and M. P. Jacobson, Cysteine oxidation in proteins: structure, biophysics, and simulation, Biochemistry, 2022, 61(20), 2165–2176.
21. D. E. Fomenko, S. M. Marino and V. N. Gladyshev, Functional diversity of cysteine residues in proteins and unique features of catalytic redox-active cysteines in thiol oxidoreductases, Mol. Cells, 2008, 26(3), 228–235.
22. C. Klomsiri, P. A. Karplus and L. B. Poole, Cysteine-based redox switches in enzymes, Antioxid. Redox Signaling, 2011, 14(6), 1065–1077.
23. K. H. Moon, B. L. Hood, P. Mukhopadhyay, M. Rajesh, M. A. Abdelmegeed and Y. I. Kwon, et al., Oxidative inactivation of key mitochondrial proteins leads to dysfunction and injury in hepatic ischemia reperfusion, Gastroenterology, 2008, 135(4), 1344–1357.
24. S. K. Suh, B. L. Hood, B. J. Kim, T. P. Conrads, T. D. Veenstra and B. J. Song, Identification of oxidized mitochondrial proteins in alcohol-exposed human hepatoma cells and mouse liver, Proteomics, 2004, 4(11), 3401–3412.
25. S. Klaessens, V. Stroobant, E. De Plaen and B. J. Van den Eynde, Systemic tryptophan homeostasis, Front. Mol. Biosci., 2022, 9, 897929.
26. A. C. Terentis, M. Freewan, T. S. Sempértegui Plaza, M. J. Raftery, R. Stocker and S. R. Thomas, The selenazal drug ebselen potently inhibits indoleamine 2,3-dioxygenase by targeting enzyme cysteine residues, Biochemistry, 2010, 49(3), 591–600.
27. C. J. D. Austin, P. Kosim-Satyaputra, J. R. Smith, R. D. Willows and J. F. Jamie, Mutation of cysteine residues alters the heme-binding pocket of indoleamine 2,3-dioxygenase-1, Biochem. Biophys. Res. Commun., 2013, 436, 595–600.
28. R. P. Bywater, Why twenty amino acid residue types suffice(d) to support all living systems, PLoS One, 2018, 13(10), e0204883.
29. M. H. Ahmed, C. Catalano, S. C. Portillo, M. K. Safo, J. Neel Scarsdale and G. E. Kellogg, 3D interaction homology: The hydropathic interaction environments of even alanine are diverse and provide novel structural insight, J. Struct. Biol., 2019, 207(2), 183–198.
30. F. J. Gisdon, E. Bombarda and G. M. Ullmann, Serine and Cysteine Peptidases: So Similar, Yet Different. How the Active-Site Electrostatics Facilitates Different Reaction Mechanisms, J. Phys. Chem. B, 2022, 126(22), 4035–4048.
31. C. Catalano, M. H. Al Mughram, Y. Guo and G. E. Kellogg, 3D interaction homology: Hydropathic interaction environments of serine and cysteine are strikingly different and their roles adapt in membrane proteins, Curr. Res. Struct. Biol., 2021, 3, 239–356.
32. M. M. Fettis and G. A. Hudalla, Engineering Reactive Oxygen Species-Resistant Galectin-1 Dimers with Enhanced Lectin Activity, Bioconjugate Chem., 2018, 29(7), 2489–2496.
33. J. Saetang, N. Roongsawang, S. Sangkhathat, S. P. Voravuthikunchai, N. Sangkaew and N. Prompat, et al., Surface cysteine to serine substitutions in IL-18 reduce aggregation and enhance activity, PeerJ, 2022, 10, e13626.
34. M. Pavlin, Z. Qasem, H. Sameach, L. Gevorkyan-Airapetov, I. Ritacco and S. Ruthstein, et al., Unraveling the Impact of Cysteine-to-Serine Mutations on the Structural and Functional Properties of Cu(I)-Binding Proteins, Int. J. Mol. Sci., 2019, 20(14), 3462.
35. eLabProtocols - ZYM-5052 medium for auto-induction [Internet], [cited 2024 Mar 7], Available from: https://www.elabprotocols.com/protocols/#!protocol=38033.
36. J. Abramson, J. Adler, J. Dunger, R. Evans, T. Green and A. Pritzel, et al., Accurate structure prediction of biomolecular interactions with AlphaFold 3, Nature, 2024, 630(8016), 493–500.
37. Y. Ye and A. Godzik, Flexible structure alignment by chaining aligned fragment pairs allowing twists, Bioinformatics, 2003, 19(Suppl 2), ii246–ii255.
38. Z. Li, L. Jaroszewski, M. Iyer, M. Sedova and A. Godzik, FATCAT 2.0: towards a better understanding of the structural diversity of proteins, Nucleic Acids Res., 2020, 48(W1), W60–W64.
39. M. Heinig and D. Frishman, STRIDE: a web server for secondary structure assignment from known atomic coordinates of proteins, Nucleic Acids Res., 2004, 32(Web Server issue), W500–W502.
40. D. Sehnal, S. Bittrich, M. Deshpande, R. Svobodová, K. Berka and V. Bazgier, et al., Mol* Viewer: modern web app for 3D visualization and analysis of large biomolecular structures, Nucleic Acids Res., 2021, 49(W1), W431–W437.
41. L. van der Fits, S. Mourits, J. S. A. Voerman, M. Kant, L. Boon and J. D. Laman, et al., Imiquimod-induced psoriasis-like skin inflammation in mice is mediated via the IL-23/IL-17 axis, J. Immunol., 2009, 182(9), 5836–5845.
42. W. J. Huber and W. L. Backes, Quantitation of heme oxygenase 1: heme titration increases yield of purified protein, Anal. Biochem., 2008, 373(1), 167–169.
43. U. F. Röhrig, O. Michielin and V. Zoete, Structure and Plasticity of Indoleamine 2,3-Dioxygenase 1 (IDO1), J. Med. Chem., 2021, 64(24), 17690–17705.
44. K. Y. Chiang, F. Matsumura, C. C. Yu, D. Qi, Y. Nagata and M. Bonn, et al., True Origin of Amide I Shifts Observed in Protein Spectra Obtained with Sum Frequency Generation Spectroscopy, J. Phys. Chem. Lett., 2023, 14(21), 4949–4954.
45. C. A. Seid, K. M. Jones, J. Pollet, B. Keegan, E. Hudspeth and M. Hammond, et al., Cysteine mutagenesis improves the production without abrogating antigenicity of a recombinant protein vaccine candidate for human chagas disease, Hum. Vaccines Immunother., 2017, 13(3), 621–633.
46. J. Lee and M. Blaber, The interaction between thermodynamic stability and buried free cysteines in regulating the functional half-life of fibroblast growth factor-1, J. Mol. Biol., 2009, 393(1), 113–127.
47. T. Nakaniwa, H. Fukada, T. Inoue, M. Gouda, R. Nakai and Y. Kirii, et al., Seven cysteine-deficient mutants depict the interplay between thermal and chemical stabilities of individual cysteine residues in mitogen-activated protein kinase c-Jun N-terminal kinase 1, Biochemistry, 2012, 51(42), 8410–8421.
48. G. Di Rocco, F. Bernini, G. Battistuzzi, A. Ranieri, C. A. Bortolotti and M. Borsari, et al., Hydrogen peroxide induces heme degradation and protein aggregation in human neuroglobin: roles of the disulfide bridge and hydrogen-bonding in the distal heme cavity, FEBS J., 2023, 290(1), 148–161.
49. S. A. Farhadi, E. Bracho-Sanchez, M. M. Fettis, D. T. Seroski, S. L. Freeman and A. Restuccia, et al., Locally anchoring enzymes to tissues via extracellular glycan recognition, Nat. Commun., 2018, 9(1), 4943.
50. K. S. Hingorani and L. M. Gierasch, Comparing protein folding in vitro and in vivo: foldability meets the fitness challenge, Curr. Opin. Struct. Biol., 2014, 24, 81–90.
51. R. Feng, M. Gruebele and C. M. Davis, Quantifying protein dynamics and stability in a living organism, Nat. Commun., 2019, 10(1), 1179.

---