Crystal structures of the sweet-tasting protein honey truffle active component from Mattirolomyces terfezioides

Crystal structures of the sweet-tasting protein honey truffle active component are reported at resolutions ranging from 1.23 to 1.70 Å. The structures correlate well with a previously reported NMR solution-phase protein structure and reveal side-chain configurations that are critical in binding of the protein to the human sweet taste receptor and are also involved in ligand interactions in solution and self-association.

Abstract

Crystal structures of honey truffle active component (HT-AC) originally derived from the fungus Mattirolomyces terfezioides are reported for the first time. HT-AC is a 13 kDa protein and is one of a small class of unrelated proteins that interact with an allosteric site on the human T1R2/R3 sweet taste receptor. These proteins can be used as sweeteners, and result in sweet taste perception at lower concentrations than sugar (the orthosteric receptor target) or many other high-intensity small-molecule sweeteners. Orthorhombic (space group P2₁2₁2₁) and monoclinic (space group P2₁) crystal structures of HT-AC are reported at resolutions ranging from 1.23 to 1.70 Å.

1. Introduction

Honey truffle active component (HT-AC) is one of a small set of proteins that have been found to interact with the human T1R2/R3 sweet taste receptor and is the only known sweet protein originating from a fungal source, Mattirolomyces terfezioides (Vo et al., 2026 ▸). The structure of HT-AC is of interest as an allosteric modulator due to its interaction with the cysteine-rich domain, a region which is found in most class C G-protein coupled receptors (Møller et al., 2017 ▸).

We have previously reported a solution-phase NMR structure of HT-AC and used the structure to model the interaction with the human T1R2/R3 receptor, recently published in its biologically relevant heterodimer structure, via cryoEM (Juen et al., 2025 ▸). This protein–protein binding interaction triggers signal transduction using concentrations many orders of magnitude less than the required concentration of saccharides to trigger a sensory response. HT-AC has a half-maximal effective concentration (EC₅₀) of 0.5 µM, compared with 40 mM for sucrose, the orthosteric ligand.

We sought to determine HT-AC crystal structures to further analyze the structural characteristics that facilitate binding to the receptor. If HT-AC is relatively rigid with little change due to the solution environment, it may act like a lever by providing loci for initial binding contacts. Secondary locations can form additional hydrogen bonds or contacts upon conformational change of the receptor from the inactive to active positions.

2. Materials and methods

2.1. Expression and purification of HT-AC

The gene for HT-AC from M. terfezioides was codon-optimized and expressed in Komagataella phaffii as described previously in McFarland et al. (2024 ▸). HT-AC was expressed intracellularly in K. phaffii and cell disruption, filtration and chromatographic purification were then performed using the parameters described in Vo et al. (2026 ▸) for Escherichia coli and Saccharomyces cerevisiae. A second chromatographic purification was performed via size-exclusion chromatography using Sephacryl 26/60 S-200 HR (Cytiva, Malborough, Massachusetts, USA). HT-AC purity prior to crystallization was determined by BCA assay and RP-HPLC for total protein and HT-AC content, respectively. The final purified protein was then lyophilized.

2.1.1. NMR structure of HT-AC

The NMR structure of HT-AC has previously been reported (PDB entry 9p6w) and was evaluated in a buffer system composed of 100 mM sodium phosphate pH 7.0. The NMR structure contains a methionine at position 1 of the sequence, which is found as an isoform of the protein due to incomplete protein maturation. This isoform was not observed in this work; therefore, the methionine was indexed as position 0 in the sequence and the first residue was identified as Pro1.

2.2. Crystallization

Purified and lyophilized HT-AC was reconstituted in 10 mM sodium acetate pH 4.5 at 19.4, 10.0 and 6.5 mg ml⁻¹ for crystallization screening. All crystallization experiments were set up using an NT8 drop-setting robot (Formulatrix, Dubai, UAE) and UVXPO MRC (Molecular Dimensions, Holland, Ohio, USA) three-drop sitting-drop vapor-diffusion plates at 17°C (290 K). 100 nl protein solution and 100 nl crystallization solution were dispensed and equilibrated against 50 µl crystallization solution in each well. Crystallization was tested using the three protein concentrations in the following commercially available screens: Berkeley (Rigaku Reagents, Bainbridge Island, Washington, USA), Index HT (Hampton Research, Aliso Viejo, California, USA), JCSG+ HT (Molecular Dimensions), Morpheus Fusion HT (Gorrec & Bellini, 2022 ▸; Molecular Dimensions), Morpheus HT (Gorrec, 2009 ▸; Molecular Dimensions) and ProPlex HT (Molecular Dimensions). Crystals displaying a prism/needle morphology appeared overnight from numerous conditions at the highest tested HT-AC concentration. The samples concentrated to 10.0 and 6.5 mg ml⁻¹ produced crystals in a few days to several weeks post-set-up but from a smaller number of conditions. Four conditions that produced crystals were selected for further evaluation and are summarized in Table 1 ▸.

Table 1. Selected crystallization conditions.

Crystal ID	Condition ID	Buffer composition
HT-AC 1	ProPlex HT C5	20%(w/v) PEG 4000, 100 mM Tris pH 8.0
HT-AC 2	JCSG+ C4	10%(w/v) PEG 6000, 100 mM HEPES pH 7.0
HT-AC 3	Morpheus Fusion A12	20%(v/v) glycerol, 10%(w/v) PEG 4000, 0.05 M Tris, 0.05 M Bicine pH 8.5, 0.02 M 2-propanol, 0.02 M 1,4-butanediol, 0.02 M 1,3-propanediol, 0.01 M spermine·4HCl, 0.01 M spermidine·3HCl, 0.01 M 1,4-diaminobutane·2HCl, 0.01 M ornithine·HCl
HT-AC 4	Morpheus Fusion C2	12.5%(w/v) MPD, 12.5%(w/v) PEG 1000, 12.5%(w/v) PEG 3350, 0.05 M sodium HEPES, 0.05 M MOPS pH 7.5, 0.02 MD-glucose, 0.02 MD-mannose, 0.02 MD-galactose, 0.02 ML-fucose, 0.02 MD-xylose, 0.02 MN-acetyl-D-glucosamine, 0.5 mM MnCl2·4H2O, 0.5 mM CoCl2·6H2O, 0.5 mM NiCl2·6H2O, 0.5 mM ZnCl2·2H2O

Crystals from conditions HT-AC 1 and HT-AC 2 were transferred to a cryoprotectant containing 80%(v/v) of their respective crystallization buffers and 20%(v/v) PEG 200 and were stored in liquid nitrogen. The well solution for conditions HT-AC 3 and HT-AC 4 were layered onto the drop and the crystals were harvested directly.

2.3. X-ray diffraction data collection and processing

X-ray diffraction data were collected using a Bruker D8 Venture diffractometer equipped with a Photon III (C14) pixel-array detector. Intensities were integrated using the Bruker PROTEUM 5 software suite (PROTEUM, SAINT and SADABS; Bruker AXS LLC, Madison, Wisconsin, USA) and the Laue class analysis and data scaling were performed with AIMLESS (Evans, 2011 ▸). Structure solution was conducted by molecular replacement with Phaser (McCoy et al., 2007 ▸) using a search model generated using the AlphaFold3 server (Thompson & Petrić Howe, 2024 ▸). Structure refinement and interactive model building were conducted with Phenix (Liebschner et al., 2019 ▸) and Coot (Emsley et al., 2010 ▸) respectively. Structure validation was conducted with MolProbity (Chen et al., 2010 ▸) and figures were prepared using the CCP4MG package (McNicholas et al., 2011 ▸). Crystallo­graphic data are provided in Table 2 ▸. Superpositions were performed using GESAMT (Krissinel & Henrick, 2004 ▸, 2005 ▸) in the CCP4 suite (Agirre et al., 2023 ▸)

Table 2. Crystallographic data for HT-AC structures.

Values in parentheses are for the highest resolution shell.

	HT-AC 1	HT-AC 2	HT-AC 3	HT-AC 4
Data collection
a, b, c (Å)	48.12, 59.02, 84.56	48.11, 58.91, 84.09	48.20, 58.63, 84.77	45.64, 44.47, 63.75
α, β, γ (°)	90, 90, 90	90, 90, 90	90, 90, 90	90, 106.7, 90
Space group	P212121	P212121	P212121	P21
Resolution (Å)	25.43–1.23 (1.25–1.23)	25.31–1.25 (1.27–1.25)	25.45–1.18 (1.20–1.18)	25.17–1.70 (1.73–1.70)
Wavelength (Å)	1.5418	1.5418	1.5418	1.5418
Temperature (K)	100	100	100	100
Observed reflections	1243052	1370475	1077965	370664
Unique reflections	70668	66898	79723	26884
〈I/σ(I)〉	24.0 (1.6)	26.1 (1.6)	18.8 (1.7)	28.3 (1.6)
Completeness (%)	100 (100)	100 (100)	100 (100)	100 (100)
Multiplicity	17.6 (9.0)	20.5 (8.1)	13.5 (8.7)	13.8 (7.6)
Rmerge† (%)	8.7 (163.9)	8.6 (143.3)	9.4 (136.4)	5.3 (156.6)
Rmeas‡ (%)	8.9 (173.8)	8.8 (152.8)	8.7 (145.0)	5.5 (168.0)
Rp.i.m.‡ (%)	2.0 (57.2)	1.9 (50.6)	2.6 (48.7)	1.4 (60.2)
CC1/2§	1.000 (0.735)	1.000 (0.620)	0.999 (0.646)	1.000 (0.767)
Refinement
Resolution (Å)	25.43–1.23	25.31–1.25	25.45–1.18	20.89–1.70
Reflections (working/test)	66679/3608	63481/3327	75643/3984	25391/1299
R factor/Rfree¶ (%)	13.3/16.4	13.2/16.4	12.7/15.2	18.9/24.0
No. of atoms (protein/ligand/water)	1954/17/514	1956/30/476	1988/32/462	1873/12/198
Model quality
R.m.s. deviations
Bond lengths (Å)	0.008	0.008	0.007	0.009
Bond angles (°)	1.001	1.039	1.015	1.019
Mean B factor (Å2)
All atoms	13.6	14.2	13.5	35.8
Protein	9.8	10.1	9.6	35.8
Ligand	45.3	38.1	35.9	43.2
Water	27.0	28.6	27.3	35.3
Coordinate error (maximum likelihood) (Å)	0.13	0.11	0.10	0.25
Ramachandran plot
Most favored (%)	98.3	98.3	98.3	97.5
Additionally allowed (%)	1.7	1.7	1.7	2.5

^†R_merge = , where I_i(hkl) is the intensity measured for the ith reflection and 〈I(hkl)〉 is the average intensity of all reflections with indices hkl.

^‡R_meas is the redundancy-independent (multiplicity-weighted) R_merge (Evans, 2006 ▸, 2011 ▸). R_p.i.m. is the precision-indicating (multiplicity-weighted) R_merge (Diederichs & Karplus, 1997 ▸; Weiss, 2001 ▸).

^§R factor = ; R_free is calculated in an identical manner using 5% of randomly selected reflections that were not included in the refinement.

^¶CC_1/2 is the correlation coefficient of the mean intensities between two random half-sets of data (Karplus & Diederichs, 2012 ▸; Evans, 2012 ▸).

3. Results and discussion

HT-AC adopts a β-sandwich fold containing 11 strands and certain fungal lectins display similar structures with a β-jelly-roll fold (Vajravijayan et al., 2020 ▸). The secondary-structure elements in HT-AC span the following residues: β1, Ser5–Asn12; β2, Phe17–Ser24; β3, Val29–Pro33; β4, Ser38–Ile39; β5, Lys44–Asp51; β6, Gly57–Val66; β7, Trp72–Leu80; β8, Val86–Gly92; β9, Val99–Leu100; β10, Val104–Gly107; β11, Leu111–Thr118. The HT-AC 1, 2 and 3 structures crystallized in an orthorhombic lattice, whereas HT-AC 4 adopted a different monoclinic crystal form. In all cases, the individual monomers of the structures have minimal differences and only one conformation is observed for the protein, regardless of crystal packing or crystallization conditions. All structures are presented in Fig. 1 ▸.

Figure 1.

(a) Dimer structures of HT-AC with A subunits in orange and B subunits in blue. Organic ligands are rendered in cylinder form and chloride ions are identified as yellow spheres. All structures are aligned to the position of the A subunit in HT-AC 1. (b) Secondary structure and identified β-sheets for HT-AC.

3.1. HT-AC structures

The HT-AC 1, 2 and 3 structures contain a twofold-symmetric dimer in the asymmetric unit (Fig. 1 ▸). This dimer has four hydrogen-bonding pairs in this configuration, shown in Fig. 2 ▸ and Supplementary Table S1. HT-AC 1, 2 and 3 are essentially identical except for the presence of variable crystallization-solution ligands and represent a single crystal-packing structure with an r.m.s.d. between 0.085 and 0.196 Å for alignment of C^α atoms and between 0.27 and 0.43 Å for all atoms.

Figure 2.

(a) Inter-subunit hydrogen bonds found in the orthorhombic form for HT-AC 1 (subunit A shown in orange and subunit B shown in blue). (b) Hydrogen-bonding detail of (a). Alternate conformations of Trp44 are shown along with the hydrogen bond to Asp2 of subunit B.

HT-AC 4 adopted a primitive monoclinic form in which each subunit is related by translational symmetry with more complex hydrogen-bonding contacts between subunits (Fig. 3 ▸b), as summarized in Supplementary Table S1. Application of the space-group symmetry operator (−x, y + 1/2, −z − 1/2) to HT-AC 1, 2 or 3 subunit A also produces another potential dimer related by translational symmetry. This second configuration (designated dimer 2 for HT-AC 1, 2 and 3) superimposes well with the monoclinic dimer HT-AC 4, with an r.m.s.d. of 0.42 Å for C^α atoms and 0.98 Å for all atoms, as shown in Fig. 3 ▸(a), and reinforces the prediction of space group and contact surfaces for HT-AC 4, which had a lower quantity of reflections during data collection and a corresponding lower model quality. Structure HT-AC 4 had a large mass of positive electron density near Gln63 and Arg74 of subunit A (Supplementary Fig. S1), which was modeled as a glucose molecule from the crystallization solution.

Figure 3.

Structure of the monoclinic form of HT-AC. (a) Superposition of HT-AC 4 (orange) and HT-AC 1 dimer form 2 (blue). (b) Hydrogen-bonding contacts between subunits in the monoclinic dimer of HT-AC 4. Subunit A is in orange and subunit B is in blue. Dimer form 2 of HT-AC 1, 2 and 3 has a similar hydrogen-bond arrangement that is described in Supplementary Table S1.

3.1.1. Ligands bound to HT-AC from crystallization solutions

In addition to the glucose ligand found in HT-AC 4 (Supplementary Fig. S1), several other ligands from the crystallization solutions were associated with HT-AC dimers. HT-AC 1 contained a PEG molecule and HT-AC 3 had both PEG and a 1,4-butanediol molecule (Supplementary Fig. S2). PEG, acetate and HEPES molecules were bound to HT-AC 2 (Supplementary Fig. S3). While most of the ligands appear on the outer facial regions of the dimer, exceptions include PEG in HT-AC 2 and 1,4-butanediol in HT-AC 3 found between the dimer subunits. Supplementary Table S2 contains the identifiable hydrogen bonds for each of these structures and ligands.

3.1.2. Comparison of crystal-packing structures to solution-phase structures and to protein homologs

Structural analysis using PISA (Krissinel & Henrick, 2007 ▸) predicted a dimeric assembly between subunits A and B of HT-AC 1, with a buried surface area of 534 Ų at the subunit interface and a complexation significance score (CSS) of 0.152, suggesting a potentially relevant but modest interaction. HT-AC 3 had a similar assembly, with a buried surface area 532.2 Ų and a CSS of 0.138. No other assemblies from the other structures had a significant complexation score predicted by PISA. The most probable explanation for these dimer configurations is due to crystal-packing contacts rather than due to biological origin; no evidence of solution-state dimers has been observed in the NMR spectra of this protein.

Minimal differences are observed between any monomer from the four crystal structures and the solution-phase NMR structure of HT-AC (9p6w), with an r.m.s.d. of less than 1.651 Å for all atoms and 1.032 Å for C^α. Possibly relevant differences in side-chain alignment are found for residues Gln30, Gln63, Arg65 and Glu88 (Fig. 4 ▸), with the Gln30, Gln63 and Arg65 side chains having coordinate distances greater than the resolution of the structures as measured by distances of side-chain γ carbons (Gln30 C^γ distance, 3.5 Å; Arg65 C^γ distance, 2.0 Å; Gln63 C^γ distance, 3.0 Å) and a model coordinate error of 0.25 Å. These residues are relevant to binding of HT-AC to the receptor, and while the differences are minor, they could imply some flexibility in the protein side chains in the presence of other ligands. Gln63, which is important for interaction with the receptor but also formed contacts with a HEPES molecule in HT-AC 2 and a glucose molecule in HT-AC 4, could be particularly accessible to ligands in addition to the T1R2/R3 receptor.

Figure 4.

(a) Side-chain alignments of monomeric HT-AC (derived from HT-AC 4, light blue) to side chains in the NMR structure PDB entry 9p6w (cyan) for residues critical for the binding of HT-AC to the sweet taste receptor T1R2/R3. (b) Detail of residues Gln30, Gln63 and Arg65. Gln30 C^γ distance, 3.5 Å. Arg65 C^γ distance, 2.0 Å. Q63 C^γ distance, 3.0 Å.

The DALI (Holm & Rosenström, 2010 ▸; Holm & Laakso, 2016 ▸) and FoldSeek (Van Kempen et al., 2024 ▸) servers were used to search for homologs in the PDB that adopt a similar fold to HT-AC. The structure with the highest homology was cry34Ab1 from Bacillus thuringiensis (PDB entry 4jox; M. S. Kelker, C. Berry, S. L. Evans, R. Pai, D. McCaskill, J. W. Pflugrath, M. Wade, T. J. Wess, M. D. Baker, C. Yang, K. E. Narva, D. G. McCaskill, N. X. Wang & J. C. Russell, unpublished work; Fig. 5 ▸a), which adopts a similar β-sandwich fold but has a slightly extended C-terminus. Additional structures include bryoporin from Physcomitrium patens (PDB entry 7pud; Šolinc et al., 2022 ▸), equinatoxin from Actinia equina (PDB entry 1tzq; Kristan et al., 2004 ▸) and fagaceatoxin C (FraC) from Actinia fragacea (PDB entry 4tsy; Tanaka et al., 2015 ▸) as shown in Figs. 5 ▸(b)–5 ▸(d). The main differences are that PDB entries 7pud and 1tzq contain an α-helix that is positioned on one of the β-sheet groups and PDB entry 4tsy contains an extended N-terminal helix. No fungal proteins were identified as high-similarity protein homologs.

Figure 5.

Comparison of the HT-AC structure (orange) with those of homologs. (a) Cry34Ab1 from Bacillus thuringiensis (PDB entry 4jox). (b) Bryoporin from Physcomitrium patens (PDB entry 7pud). (c) Equinatoxin from Actinia equina (PDB entry 1tzq). (d) Fagaceatoxin C (FraC) from Actinia fragacea (PDB entry 4tsy).

4. Conclusions

The structure of HT-AC does not demonstrate a great deal of flexibility, with little variance observed in the solution-phase structure compared with the X-ray diffraction-derived structures reported in this communication; observed differences within the crystal packings and NMR structure are largely limited to conformations of side chains and very little backbone variance is seen. This supports the hypothesis that during interaction with the receptor, HT-AC does not change in conformation significantly and instead causes a conformational change in the receptor through contacts with the CRD; this conformational change causes receptor activation. Our research group based in silico modeling of the HT-AC–T1R2/R3 interaction on the solution-phase model of HT-AC, and future work can be refined using the higher resolution structures without major concerns about the initial model system.

Gln63 appears to be of particular importance in many binding modes for the protein and, given its importance in the interaction with the T1R2/R3 receptor, could provide avenues for future protein engineering. The small position shift seen in the solution-phase structure for Gln63 brings it nearer to Tyr61, which could add some limited stabilization for this residue and better align with negatively charged surfaces on the T1R2/R3 receptor CRD region. Other side-chain residues (Arg19, Lys25, Tyr61) also interact both in receptor binding and self-association and could be additional valuable targets for manipulation of the protein, as these residues are demonstrably exposed and available for polar interactions with other ligands in the environment.

The protein readily self-associates and crystallizes from solution under milder conditions (such as neutral pH, high ionic strength at 2–8°C) than required for the production of analytical quality crystals, despite limited evidence of biologically relevant quaternary structure. While the protein is likely to be expressed in M. terfezioides as a monomer, in heterologous expression and in bulk manufacturing where concentrations of the protein are above 10 g l⁻¹, crystalline dimeric HT-AC may be present and have characteristics that diverge from the monomeric protein. In silico binding simulations show that the dimer structures have a substantially greater dissociation concentration than the equivalent monomers (Section S1). Future experiments using crystalline HT-AC in receptor-based in vitro assays are likely to confirm the in silico binding model and show lower affinity than the monomer.

5. Related literature

The following references are cited in the supporting information for this article: Bekker et al. (1993 ▸), Dhawanjewar et al. (2020 ▸). Jiménez-García et al. (2015 ▸) and Vangone & Bonvin (2015 ▸).

Supplementary Material

PDB reference: honey truffle active component, orthorhombic, 11os

PDB reference: orthorhombic, with HEPES bound, 11ot

PDB reference: orthorhombic, with 1,4-butanediol bound, 11ow

PDB reference: monoclinic, with glucose bound, 11ox

Conflict of interest

The research presented in this article was financially supported by MycoTechnology, Inc. (Aurora, Colorado, USA). Mycotechnology, Inc. sells products related to this research.

Data availability

Coordinates and structure factors have been deposited in the Worldwide Protein Databank (wwPDB) with accession codes 11os (HT-AC 1), 11ot (HT-AC 2), 11ow (HT-AC 3) and 11ox (HT-AC 4).