Structural Diversity and Differential Natural Pairings of MAT1-1-1 and MAT1-2-1 Proteins Essential for Sexual Reproduction in Ophiocordyceps sinensis Strains

1. Introduction

Natural Cordyceps sinensis (DōngChóng XiàCǎo, 冬虫夏草, in Chinese) is found exclusively in alpine meadows (3,000–5,000 m) on the Qinghai‒Tibet Plateau, a restricted habitat that contributes substantially to its scarcity. For centuries, it has been used in traditional Chinese medicine for health maintenance, disease amelioration, post-illness and postoperative recovery, and antiaging therapy [1,2,3,4]. Recent studies have explored the complex biochemical mechanisms that may underlie these clinical applications (e.g., the regulation of glucose and lipid metabolism, immunomodulation, anti-inflammatory and antioxidant activity, hormone-like effects related to libido and fertility, and improvements in perceived energy/vitality). This distinctive therapeutic significance and the high commercial value of this natural substance have contributed to extensive overharvesting, leading to its designation as a LEVEL-II protected natural substance in China [5].

Natural C. sinensis comprises the fruiting body of Ophiocordyceps sinensis and the remains of a mummified Hepialidae moth larva. The larval body retains a thick body wall bearing numerous bristles, together with intact intestinal structures, cephalic tissues, and fragments of other tissues [6,7,8,9,10,11,12,13]. Dual-fluorescence microscopy has revealed multicellular heterokaryotic microstructures within C. sinensis ascospores and hyphae, including mono-, bi-, tri-, and tetra-nucleate cells [14]. Molecular studies have further demonstrated substantial genetic heterogeneity, identifying >90 fungal species across at least 37 genera and 17 genomically independent O. sinensis genotypes that differentially cooccur and vary in natural abundance among distinct compartments of C. sinensis during maturation, which supports their genome-independent natural characteristics [9,10,12,13,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44]. Accordingly, the Chinese Pharmacopoeia defines this medicinal microecosystem as an insect‒fungal complex.

Among the numerous cooccurring fungal species [20,31,32,33,40,41], Hirsutella sinensis has been proposed as the sole anamorph of O. sinensis [45]. However, this hypothesis has not yet been conclusively validated according to all 4 criteria of Koch’s postulates. Among the many attempts to support this hypothesis, Wei et al. [46] reported successful artificial cultivation in an industrial, product-oriented setting. The results of this study revealed a discrepancy between the species identities of the anamorphic inoculants (3 GC-biased Genotype #1 H. sinensis strains) inoculated into Hepialidae larvae and the sole teleomorph (AT-biased Genotype #4 of O. sinensis) identified in the cultivated fruiting body.

Since the 1840s, the Latin designation C. sinensis has been used indiscriminately to refer to both the fungal teleomorph and the wild insect‒fungal complex, despite Edward Doubleday’s original finding in 1842 showing that the insect component of the natural material belonged to Agrotis [47]. This imprecise terminology has obscured the distinction among the fungus, the host insect, and the insect−fungal complex as a unique therapeutic entity. As a result, inconsistencies have emerged across scientific publications, commercial trade in the mass market, and governmental regulations worldwide. In 2007, the fungus was renamed O. sinensis, and H. sinensis strain EFCC7287 (GC-biased Genotype #1 of O. sinensis) was designated the nomenclatural reference [9,10,48,49,50]. Following the “One Fungus=One Name” nomenclature principle established by the International Mycological Association (IMA) [52,53], Zhang et al. [51] interpreted O. sinensis as a single fungal species to satisfy the “One Fungus” prerequisite of the rule and replace the anamorphic name H. sinensis with the teleomorphic name O. sinensis. However, the authors did not fully consider the existence of 17 genomically independent genotypes that differentially coexist in various combinations within different compartments of the C. sinensis insect–fungal complex. In the present study, we retain the anamorphic designation H. sinensis specifically for GC-biased Genotype #1 while referring to the genome-independent Genotypes #2‒17 fungi as O. sinensis in accordance with annotations in public databases such as GenBank and AlphaFold, pending formal taxonomic resolution. For consistency with the longstanding practice in the medicinal literature, we also retain the term C. sinensis to denote both wild and cultivated insect‒fungal complexes, acknowledging that the 2007 taxonomic revision did not resolve the longstanding indiscriminate application of the Latin name to the insect‒fungal complex [9,10,48,51,54]. This provisionally necessary practice may require future refinement, as proprietary and exclusive Latin names are established for individual genome-independent genotypes of O. sinensis and for the insect‒fungal complex as a distinct therapeutic entity.

Sexual reproduction in ascomycetes is regulated by transcription factors encoded at the mating-type (MAT) locus, which mediate mating-type recognition and compatibility, as well as the development of fruiting bodies and sexual structures (ascocarps and ascospores) [55,56,57,58,59,60,61,62]. In O. sinensis, reproductive behavior depends on the coordinated activity of 2 core mating proteins, encoded by the MAT1-1 and MAT1-2 idiomorphs. The MAT1-1-1 protein harbors a mating-type alpha high mobility group box (MATα_HMGbox) domain, whereas the MAT1-2-1 protein contains a high mobility group box ROX1-like (HMG-box_ROX1-like) domain [14,58,63,64,65]. These DNA-binding domains regulate the transcription of genes associated with sexual reproduction. Li et al. [65,66,67,68] reported differential occurrence, alternative splicing, and differential transcription of mating-type and pheromone receptor genes in the genomes of H. sinensis strains, which are inconsistent with self-fertilization but consistent with self-sterility. In addition, Li et al. [65,69] described heteromorphic tertiary structures of the DNA-binding domains of the MAT1-1-1 and MAT1-2-1 proteins in wild-type C. sinensis isolates, which may have originated from either GC-biased Genotype #1 or Genotype #3 of O. sinensis. In aggregation, these observations support the existence of complex reproductive biology in O. sinensis that may involve heterothallic mating or hybrid reproductive mechanisms.

Variations within the DNA-binding domains of the MAT1-1-1 and MAT1-2-1 proteins with diverse amino acid substitutions may alter their tertiary structures and influence interactions between the critical mating proteins. Prior studies characterized mating proteins from numerous impure wild-type C. sinensis isolates characterized by Prof. Zhang Y-J and associates [7,51,65,70,71,72]. In the present study, we investigated the naturally paired MAT1-1-1 and MAT1-2-1 proteins simultaneously coproduced by each of 20 purportedly pure O. sinensis strains reported by Prof. Yao Y-J and collaborators [14,36]. Specifically, we analyzed differential N- and/or C-terminal truncations and diverse amino acid substitutions within the MATα_HMGbox and HMG-box_ROX1-like domains and further assessed how these variants are differentially naturally paired within individual strains. These analyses enable the evaluation of whether the natural pairings and structural diversity of the mating proteins are consistent with their derivation from a single homogeneous fungal source or instead suggest contributions from multiple cooccurring and symbiotic fungal taxa within the purportedly pure, but potentially impure, strains.

2. Results

2.1. Primary Structures of the Truncated MAT1-1-1 Proteins in O. Sinensis Strains

The sequence alignment of 20 MAT1-1-1 proteins (AGW27517−AGW27536 [14]), exhibiting variable N- and C-terminal truncations together with 1–4 amino acid substitutions at distinct sites, and the representative full-length authentic MAT1-1-1 protein AGW27560 derived from H. sinensis strain CS68-2-1229 is presented in Figure 1 [14].

These truncated MAT1-1-1 proteins were encoded by cDNAs amplified from total RNA extracted from 20 O. sinensis strains using the same primer pair, m1F3/m1R3 [14,70]. Relative to the full-length reference protein, these protein variants exhibit truncations of 63–110 residues at the N-termini and 8–18 residues at the C-termini and correspond to 9 distinct AlphaFold-predicted 3D structural morphotypes (U3N9T9, U3N6U0, U3N919, U3N7G5, U3N6U4, U3N6U8, U3NE79, U3N7H7, and U3NE87) (Figure 1; Table 1). In the reference full-length authentic MAT1-1-1 protein AGW27560, the MATα_HMGbox domain spans amino acids 51→225 (highlighted in pink in Figure 1). All 20 variant MAT1-1-1 proteins display variable N-terminal truncations within their MATα_HMGbox domains, ranging from 13–60 amino acid residues (Figure 2). In addition, 1–4 amino acid substitutions were identified in several truncated protein variants and are highlighted in green in Figure 1.

2.2. Primary Structures of the MAT1-2-1 Proteins in O. Sinensis Strains

Figure 3 shows the sequence alignment of 20 MAT1-2-1 proteins (AGW27537− AGW27556 [14]) produced by the 20 O. sinensis strains and the representative authentic MAT1-2-1 protein AEH27625 (AlphaFold code D7F2E9) derived from the H. sinensis strain CS2 [70]. Among these 20 MAT1-2-1 proteins, 15 (AGW27537−AGW27542, AGW27544−AGW27547, AGW27549−AGW27551, AGW27553, and AGW27556) are full-length proteins. The remaining 5 proteins (AGW27543, AGW27548, AGW27552, and AGW27554−AGW27555) exhibit C-terminal truncations of 64–75 amino acids. These truncated proteins were encoded by the MAT1-2-1 cDNAs amplified from total RNA extracted from the O. sinensis strains using the same primer pair, Mat1-2F/Mat1-2R [14,70]. In the full-length reference protein AEH27625 [70], the HMG-box_ROX1-like domain spans amino acid residues 127→197, highlighted in pink in Figure 3. In addition, 1–2 amino acid substitutions were identified in many of the proteins and are marked in green in Figure 3.

Among the 20 MAT1-2-1 proteins, 3 proteins (AGW27538, AGW27539, and AGW27541) were 100% identical to the reference authentic MAT1-2-1 protein AEH27625 and shared the same tertiary structure under the AlphaFold model D7F2E9 (Table 1; Figure 3). The remaining 17 MAT1-2-1 proteins contained tyrosine-to-histidine (Y-to-H) substitutions within the HMG-box_ROX1-like domains, leading to a shift in the hydropathy index from -1.3 to -3.2 (Table S3) [73].

The truncated MAT1-2-1 protein AGW27554 contained an additional glutamine-to-threonine (Q-to-T) substitution within its HMG-box_ROX1-like domain, corresponding to a hydropathy index shift from -3.5 to -0.7 (Table S3; Figure 3) [73].

Two full-length proteins, AGW27537 and AGW27553, exhibited additional amino acid substitutions. Specifically, AGW27537 carried a valine-to-isoleucine (V-to-I) substitution upstream of the HMG-box_ROX1-like domain, corresponding to a hydropathy index shift from 4.2 to 4.5 (Table S3) [73]), whereas AGW27553 contained a lysine-to-unidentified residue (K-to-X) substitution within the HMG-box_ROX1-like domain (Figure 3).

Eleven of the 20 MAT1-2-1 proteins exhibited diverse conformations corresponding to 7 distinct AlphaFold-predicted 3D structural morphotypes (Table 1). The remaining 9 proteins do not have predicted 3D structures available in the AlphaFold database.

In the five C-terminally truncated MAT1-2-1 proteins, the truncated regions (ranging from 12–23 residues) are located within the HMG-box_ROX1-like domains (Figure 3 and Figure 4). All amino acid substitutions, except for the valine-to-isoleucine (V-to-I) substitution, are within the HMG-box_ROX1-like domains. The truncations and amino acid substitutions are associated with differences in hydrophobicity and the predicted secondary/tertiary structure of the HMG-box_ROX1-like domains, corresponding to 4 distinct AlphaFold 3D structural morphotypes (Table 1; Figure 3 and Figure 4), highlighting the structural diversity of this protein family across O. sinensis strains.

2.3. Differential Natural Pairings of the MAT1-1-1 and MAT1-2-1 Proteins Simultaneously Produced by O. Sinensis Strains

Li et al. [36] deposited internal transcribed spacer (ITS) sequences for 15 of the 20 O. sinensis strains in GenBank for phylogenetic positioning. Among these 15 strains, only Group-A (GC-biased Genotype #1 H. sinensis) ITS sequences were reported for 10 strains (highlighted in brown in Table 1). For the remaining 5 strains (highlighted in red in Table 1), ITS sequences corresponding to both GC-biased Genotype #1 and AT-biased Genotype #17 (or #5) were reported (Tables S1−S2). The authors proposed that the AT-biased sequences represented nonfunctional pseudogenic repetitive copies within the genome of GC-biased Genotype #1 H. sinensis and concluded that all the strains represented pure O. sinensis cultures. ITS sequences of the remaining 5 strains (CS26-277, CS36-1294, CS68-5-1216, CS70-1211, and CS70-1212) are unavailable in GenBank (highlighted in black in Table 1).

With respect to the MAT1-1-1 and MAT1-2-1 proteins analyzed here, the assumption of homogeneity for these O. sinensis strains warrants careful interpretation for several reasons. (1) ITS sequences of AT-biased O. sinensis Genotypes #4–6 and #15–17 are absent in the genome of GC-biased Genotype #1; instead, they are apparently genome-independent fungal taxa [9,10,13,35,38]. (2) Li et al. [36] and collaborators [14] reported multicellular heterokaryotic microstructures in C. sinensis ascospores and hyphae that contain multiple mono-, bi-, tri-, and tetra-nucleate cells. Zhang & Zhang [72] discussed the possibility that such polynucleate cells could harbor heterogeneous hereditary substances. (3) Culture-dependent ITS profiling may overlook nonculturable fungal taxa and low-abundance amplicons, particularly when different primer sets and denaturation temperatures differentially affect amplification efficiency among genotypes (cf. Section 3.2) [9,10,12,13,37,90]. (4) Culture-independent studies have reported that C. sinensis ascospores contain psychrophilic GC-biased Genotypes #1 and #13–14 and AT-biased Genotypes #5–6 and #16 of O. sinensis, along with additional mesophilic fungal taxa. Moreover, the stromata and caterpillar bodies of the insect–fungal complexes have been reported to contain additional O. sinensis genotypes and heterospecific fungi [9,10,12,13,15,16,20,21,31,32,34,37,40,41].

Despite the methodological limitations of the original culture-dependent study and its interpretation that the strains were homogeneous, the 20 O. sinensis strains can be categorized into 2 groups according to the observed natural pairing patterns of the 2 mating proteins (Table 2). Group I comprises 5 strains that simultaneously produce paired truncated MAT1-1-1 and MAT1-2-1 proteins. Group II comprises the remaining 15 O. sinensis strains, which simultaneously produce truncated MAT1-1-1 proteins paired with full-length MAT1-2-1 proteins. The differential N- and/or C-terminal truncations and the diverse amino acid substitutions are summarized in Table 2, with proteins sharing identical AlphaFold 3D structural morphotypes denoted by the same suffix symbol (*, ⁋, †, ⁑, ‡, or ♦). The sequence distributions of the differentially naturally paired MAT1-1-1 and MAT1-2-1 proteins co-expressed in each of the 20 O. sinensis strains are visualized in Figure 5. Across strains, these naturally paired mating proteins exhibit variable N- and/or C-terminal truncations and diverse amino acid substitutions (Figure 1, Figure 2, Figure 3 and Figure 4; Table 2).

2.4. O. Sinensis Strains Simultaneously Produce MAT1-1-1 Proteins with Identical Tertiary Structures and MAT1-2-1 Proteins with Various Tertiary Structures

2.4.1. MAT1-1-1 Proteins Sharing the Same AlphaFold-Predicted 3D Structure Under U3NE87 Are Naturally Paired with Structurally Diverse MAT1-2-1 Proteins

Panel (A) of Figure 5 presents the Group-I O. sinensis strain CS70-1219, reported to contain only Group-A ITS sequences, and the Group-II strain CS70-1212, which lacks a deposited ITS record in GenBank (Table 1 and Table 2). These 2 strains produced the MAT1-1-1 proteins AGW27534 and AGW27531, respectively, each carrying 110- and 8-residue truncations at the N- and C-termini, respectively (Figure 1; Table 2). The hydrophobic properties and structural features of the MATα_HMGbox domains of these truncated MAT1-1-1 proteins, assigned to the same AlphaFold-predicted 3D structural morphotype U3NE87 (designated by the suffix symbol “*” in Table 1 and Table 2), were compared with those of the reference protein AGW27560 (AlphaFold code U3N942), as displayed in Figure 6.

Panel (A) of Figure 6 shows that the MATα_HMGbox domains of the proteins AGW27534 and AGW27531 each contain 60-residue truncations at the N-terminus. These truncations within the DNA-binding domains are associated with altered hydrophobicity and predicted secondary-structure features, as reflected by changes in the topological profiles and waveform patterns of the ExPASy ProtScale plots for hydropathy, α-helices, β-sheets, β-turns, and coils (Panels (B)−(F) of Figure 6, respectively). Panel (G) of Figure 6 presents the tertiary structure of reference protein AGW27560 (AlphaFold code U3N942), in which 3 α-helices within the MATα_HMGbox domain assemble into a characteristic core L-shaped fold. Panel (H) illustrates the remodeled conformation of the truncated MAT1-1-1 proteins assigned to the AlphaFold code U3NE87. The truncated segments within the MATα_HMGbox domains overlap with the 3 core α-helices that constitute the hydrophobic core of the MATα_HMGbox domain [91,92].

In addition, the Group-I O. sinensis strain CS71-1219 simultaneously coproduces the MAT1-2-1 protein AGW27554 [14], which contains a 65-residue truncation at the C-terminus (AlphaFold code U3NEA9), compared with the reference MAT1-2-1 protein AEH27625 (AlphaFold code D7F2E9) derived from H. sinensis strain CS2 [70] (Figure 3, Figure 4 and Figure 5; Table 1 and Table 2).

Panel (A) of Figure 7 shows that AGW27554 contains a 12-residue C-terminal truncation within the HMG-box_ROX1-like domain. This protein also has 2 amino acid substitutions: tyrosine-to-histidine (Y-to-H; hydropathy index shift from -1.3 to -3.2; Table 2 and Table S3) and glutamine-to-threonine (Q-to-T; hydropathy index shift from -3.5 to -0.7) [73]. Consistent with these sequence alterations, Panels (B)–(F) of Figure 7 demonstrate corresponding differences in the topological profiles and waveform patterns of the ExPASy ProtScale plots for hydropathy, α-helices, β-sheets, β-turns, and coils, respectively. In the AlphaFold models (Panels (G)−(H) of Figure 7), the truncation and substituted amino acids map to 2 of the 3 core α-helices within the HMG-box_ROX1-like domain, resulting in a remodeled tertiary structure assigned to the AlphaFold code U3NEA9.

On the other hand, the Group-II strain CS70-1212 simultaneously produces the full-length MAT1-2-1 protein AGW27551, which carries a single tyrosine-to-histidine (Y-to-H) substitution within the HMG-box_ROX1-like domain (hydropathy index shift from -1.3 to -3.2 [73]) (Figure 3; Table 2 and Table S3). Panels (A)–(F) of Figure 7 summarize the associated changes in hydrophobicity and primary/secondary structure features in the HMG-box_ROX1-like domain. Because AGW27551 was not assigned an AlphaFold 3D model, it was excluded from the 3D structural comparisons in Figure 7.

2.4.2. MAT1-1-1 Proteins Sharing the Same Tertiary Structure Under the AlphaFold Code U3N9T9 Are Naturally Paired with MAT1-2-1 Proteins Exhibiting Diverse 3D Structures

Among the 8 O. sinensis strains shown in Panel (B) of Figure 5 and Figure 4 strains (CS37-295, CS18-266, CS91-1291, and CS68-2-1229) were reported to contain only Group-A ITS sequences (highlighted in brown in Table 2); 2 strains (CS6-251 and CS34-291) were reported to contain both Group-A and Group-C ITS sequences (highlighted in red); whereas no ITS sequence records were deposited in GenBank for the remaining 2 strains (CS36-1294 and CS26-277; highlighted in black). These strains produce variant MAT1-1-1 proteins (AGW27517−AGW27521, AGW27523, AGW27524, and AGW27528) that each contains 63- and 8-residue truncations at the N- and C-termini, respectively, and share the same AlphaFold-predicted 3D structural morphotype U3N9T9 (designated by the suffix symbol “⁋” in Table 2). The reference MAT1-1-1 protein AGW27560 (AlphaFold code U3N942) originates from the strain CS68-2-1229 [14].

Panel (A) of Figure S1 shows that all 8 MAT1-1-1 proteins contain a common 13-residue N-terminal truncation within their MATα_HMGbox domains. Compared with the reference protein AGW27560 (AlphaFold code U3N942), this truncation is associated with changes in the topological profiles and waveform patterns of the ExPASy ProtScale plots for hydropathy, α-helices, β-sheets, β-turns, and coils (Panels (B)–(F) of Figure S1, respectively). The results of the comparative 3D structural analysis (Panels (G)–(H) of Figure S1) indicate that the truncation is positioned upstream of the 3 core α-helices within the MATα_HMGbox domain.

These 8 strains also simultaneously produced paired MAT1-2-1 proteins corresponding to distinct 3D structural morphotypes. The 2 Group-I strains (CS68-2-1229 and CS91-1291; Table 2; highlighted in blue in Figure 5) coproduced the MAT1-2-1 proteins AGW27548 and AGW27543, which carry 74- and 64-residue C-terminal truncations, respectively (Table 2; Figure 3). Both proteins harbor a tyrosine-to-histidine (Y-to-H) substitution within the HMG-box_ROX1-like domains, corresponding to a hydropathy index shift from -1.3 to -3.2; Table 2 and Table S3; Figure 3 and Figures 3 and S2) [73]. Consistent with these sequence alterations, Figure S2 shows corresponding differences in the predicted secondary and tertiary structures, with AlphaFold models U3N9W5 (AGW27548) and U3N9W0 (AGW27543).

The remaining 6 O. sinensis strains in Panel (B) of Figure 5 belong to Group II (Table 2; highlighted in green). According to the reported ITS records, 2 strains (CS37-295 and CS18-266) contain only Group-A ITS sequences deposited (shown in brown in Table 2); 2 strains (CS6-251 and CS34-291) contain both Group-A and Group-C ITS sequences (shown in red); and 2 strains (CS26-277 and CS36-1294) lack deposited ITS records in GenBank (shown in black). Although all 6 strains produce truncated MAT1-1-1 proteins with the same predicted 3D structural morphotype (AlphaFold code U3N9T9, marked with the suffix symbol “⁋” in Table 2), they simultaneously produce full-length MAT1-2-1 proteins that either correspond to different AlphaFold 3D structural morphs or lack AlphaFold 3D structural models. The 6 full-length MAT1-2-1 proteins can be categorized into 3 subgroups:

• The coproduced full-length MAT1-2-1 proteins AGW27539, AGW27541, and AGW27538 share 100% sequence identity with the reference MAT1-2-1 protein AEH27625 (Figure 3). These proteins were generated by the strain CS37-295 (with Group-A ITS sequences) and by the strains CS26-277 and CS36-1294 (no ITS records in GenBank) (Table 1 and Table 2; Panel (B) of Figure 5). The 3D structures are identical and correspond to the AlphaFold code D7F2E9 (suffix symbol “♦” in Table 2).
• The MAT1-2-1 protein AGW27537 was coproduced by strain CS6-251 (reported to contain both Group-A and Group-C ITS sequences; Table 1 and Table 2; Panel (B) of Figure 5). This protein has 2 amino acid substitutions: a valine-to-isoleucine (V-to-I) substitution upstream of the HMG-box_ROX1-like domain (hydropathy index shift from 4.2 to 4.5; Table S3) and a tyrosine-to-histidine (Y-to-H) substitution (hydropathy index changed from -1.3 to -3.2) within the DNA-binding domain (Figure 3; Table 2 and Table S3; and Figure 12 of Ref. [65]) [73]. These substitutions are associated with an altered tertiary structure (AlphaFold code U3N6V5).
• The MAT1-2-1 proteins AGW27540 and AGW27544 were coproduced by strains CS18-266 (with deposited Group-A ITS sequences) and CS34-291 (contain both Group-A and Group-C ITS sequences), respectively (Table 1 and Table 2; Panel (B) of Figure 5). Both proteins harbor tyrosine-to-histidine (Y-to-H) substitutions within their HMG-box_ROX1-like domains (hydropathy index shift from -1.3 to -3.2; Table S3) and share 99.6% sequence similarity with the authentic MAT1-2-1 protein AEH27625 (Figure 3). No AlphaFold-predicted 3D structural models are available for these 2 proteins.

2.4.3. MAT1-1-1 Proteins with the Same Tertiary Structure Under the AlphaFold 3D Structural Code U3N6U8 Naturally Paired with Diverse MAT1-2-1 Proteins with Different Tertiary Structures

Panel (C) of Figure 5 presents 3 O. sinensis strains: the Group-I strains CS71-1220 and CS68-5-1216 and the Group-II strain CS71-1218. According to the reported ITS records [14,36] (Table 1 and Table 2), only Group-A ITS sequences have been deposited in GenBank for strain CS71-1218, while for strain CS71-1220, both Group-A and Group-C ITS sequences are present, and for strain CS68-5-1216, there are no deposited ITS records in GenBank (Table 1). These strains produced the MAT1-1-1 proteins AGW27535, AGW27532, and AGW27533 (Figure 1 and Figure 5), each carrying 97- and 8-residue truncations at the N- and C-termini, respectively, and sharing the same predicted 3D structural morphotype (AlphaFold code U3N6U8; suffix symbol “†” in Table 2).

Panel (A) of Figure S3 shows that the MATα_HMGbox domains of the MAT1-1-1 proteins AGW27535, AGW27532, and AGW27533 contain a 47-residue N-terminal truncation relative to the reference protein AGW27560. This truncation is associated with changes in the topological profiles and waveform patterns of the ExPASy ProtScale plots for hydropathy, α-helices, β-sheets, β-turns, and coils (Panels (B)−(F) of Figure S3, respectively). Comparisons of the predicted 3D structures shown in Panels (G)–(H) of Figure S3 indicate that the truncated segment overlaps with the first of the 3 core α-helices within the MATα_HMGbox domain, which is consistent with the 3D conformations assigned to the AlphaFold code U3N6U8.

The 2 Group-I strains, CS68-5-1216 and CS71-1220, also produced the MAT1-2-1 proteins AGW27552 and AGW27555, which carry 75- and 74-residue C-terminal truncations, respectively, whereas the Group-II strain CS71-1218 produced the full-length MAT1-2-1 protein AGW27553 (Table 2; Figure 3).

Panel (A) of Figure S4 shows that AGW27552 and AGW27555 contain 23- and 22-residue truncations, respectively, at the C-termini of their HMG-box_ROX1-like domains. All 3 proteins contain tyrosine-to-histidine (Y-to-H) substitutions within the HMG-box_ROX1-like domain, corresponding to a hydropathy index shift from -1.3 to -3.2; Table S3) [73]. In addition, AGW27553 contains an extra lysine-to-unidentified residue (K-to-X) substitution within the DNA-binding domain (Figure 3 and Figure S4). For AGW27552 and AGW27555, the truncations and amino acid substitutions are associated with corresponding differences in hydrophobicity and secondary structural features within the HMG-box_ROX1-like domains, as reflected by altered topological profiles and waveform patterns of the ExPASy ProtScale plots for hydropathy, α-helices, β-sheets, β-turns, and coils (Panels (B)−(F) of Figure S4, respectively). The predicted 3D structures shown in Panels (G)−(H) of Figure S4 indicate that the truncations and replaced residues map to 2 of the 3 core α-helices within the HMG-box_ROX1-like domain, which is consistent with remodeled tertiary conformations assigned to AlphaFold codes U3N6W6 (AGW27552) and U3N9W5 (AGW27555). Because AGW27553 contains an unidentified amino acid substitution, it was excluded from the 2D and 3D structural comparisons.

2.4.4. MAT1-1-1 Proteins with the Same 3D Structure Under the AlphaFold Structural Code U3NE79 Naturally Paired with Diverse MAT1-2-1 Proteins with Different Tertiary Structures

Panel (D) of Figure 5 presents 2 Group-II strains, CS70-1211 and CS70-1208 (Table 1 and Table 2). CS70-1208 contains both Group-A and Group-C ITS sequences, whereas CS70-1211 lacks a deposited ITS record in GenBank. These strains produced the MAT1-1-1 proteins AGW27530 and AGW27529, each of which carried 97- and 8-residue truncations at the N- and C-termini, respectively (Figure 1; Table 2), and shared the same AlphaFold structural morph, U3NE79 (suffix symbol “‡” in Table 1 and Table 2).

Panel (A) of Figure S5 shows that the MATα_HMGbox domains of AGW27530 and AGW27529a contain a 46-residue truncation at the N-termini together with a phenylalanine-to-valine (F-to-V) substitution, corresponding to a hydropathy index shift from 2.8 to 4.2; Table S3), indicating increased hydrophobicity [73]. Panels (B)−(F) of Figure S5 show associated alterations in secondary structural features, as reflected by changes in the topological profiles and waveform patterns of the ExPASy ProtScale plots for hydropathy, α-helices, β-sheets, β-turns, and coils, respectively. Comparison of the predicted 3D structures of these MAT1-1-1 proteins (Panels (G)–(H) of Figure S5) indicates that the truncation and amino acid substitution map to the first of the 3 core α-helices within the MATα_HMGbox domain, which is consistent with the distinct 3D conformations assigned to the AlphaFold code U3NE79.

These 2 O. sinensis strains also coproduce the full-length MAT1-2-1 proteins AGW27550 and AGW27549, each containing a tyrosine-to-histidine (Y-to-H) substitution within the HMG-box_ROX1-like domain (hydropathy index shift from -1.3 to -3.2), indicating reduced hydrophobicity (Table 2 and Table S3; Figure 3) [73]. Although AlphaFold-predicted 3D structural models are not available for AGW27550 and AGW27549, their primary structures are 100% identical to that of the variant MAT1-2-1 protein AGW27542, which has been assigned to AlphaFold code D7F2E3 (Table 2; Figure 12 of Ref. [65]).

2.5. O. sinensis Strains Simultaneously Produce MAT1-2-1 Proteins with the Same Tertiary Structure and MAT1-1-1 Proteins with Different 3D Structures

2.5.1. Truncated MAT1-2-1 Proteins with Identical Tertiary Structures Under the AlphaFold Code U3N9W5 Paired with Diverse Truncated MAT1-1-1 Proteins with Different 3D Structures

As shown in Table 2, the Group-I strains CS68-2-1229 (with deposited Group-A ITS sequences) and CS71-1220 (reported to contain both Group-A and Group-C ITS sequences) produced the MAT1-2-1 proteins AGW27548 and AGW27555, respectively, each of which carried a 74-residue truncation at the C-terminus (Figure 3; Panels (B)−(C) of Figure 5). These proteins share the same predicted 3D structural morph corresponding to the AlphaFold code U3N9W5 (suffix symbol “⁑” in Table 1 and Table 2).

Panel (A) of Figure S6 shows that the HMG-box_ROX1-like domains of AGW27548 and AGW27555 contain a tyrosine-to-histidine (Y-to-H) substitution and a 22-residue C-terminal truncation. The Y-to-H substitution corresponds to a hydropathy index shift from -1.3 to -3.2, indicating reduced hydrophobicity (Table S3) [73]. The structural variations are associated with alterations in the secondary structural features surrounding the sites of variation within the HMG-box_ROX1-like domains, which are reflected by changes in the topological profiles and waveform patterns of the ExPASy ProtScale plots for hydropathy, α-helices, β-sheets, β-turns, and coils in Panels (B)−(F) of Figure S6, respectively. In the predicted 3D structural comparisons (Panels (G)–(H) of Figure S6), the truncation and amino acid replacement map to 2 of the 3 core α-helices, corresponding to the 3D conformation assigned as AlphaFold code U3N9W5.

These 2 Group-I strains also coproduced the MAT1-1-1 proteins AGW27528 (Figure S1) and AGW27535 (Figure S3), which carry 63- and 8-residue truncations and 97- and 8-residue truncations at the N- and C-termini, respectively (Figure 1; Table 2). These proteins correspond to distinct 3D structural morphs assigned to AlphaFold codes U3N9T9 and U3N6U8 (suffix symbols “⁋” and “†” in Table 1 and Table 2) and are shown separately in Panels (B) and (C) of Figure 5, respectively.

2.5.2. Full-Length MAT1-2-1 Proteins with Identical Structures Naturally Paired with Differentially Truncated MAT1-1-1 Proteins Exhibiting Various Diverse Tertiary Structures

Ten full-length MAT1-2-1 proteins (AGW27540, AGW27542, AGW27544− AGW27547, AGW27549−AGW27551, and AGW27556) share identical primary structures (Table 2; Figure 3). Each protein has a tyrosine-to-histidine (Y-to-H) substitution located within the first of the 3 core α-helices of the HMG-box_ROX1-like domain, corresponding to a hydropathy index shift from -1.3 to -3.2 (Figure 3; Table 2 and Table S3) [73]. These proteins were simultaneously produced by the Group-II strains CS18-266, CS560-961, CS34-291, CS76-1284, CS561-964, CS25-273, CS70-1208, CS70-1211, CS70-1212, and CS68-2-1228 shown in Panel (E) of Figure 5. Among these proteins, only AGW27542 was assigned an AlphaFold 3D structural model (D7F2E3) (cf. Figure 12 of Ref. [65]), whereas the remaining 9 full-length MAT1-2-1 proteins do not have available AlphaFold structural models.

These 10 strains also simultaneously produced differentially truncated MAT1-1-1 proteins (AGW27520, AGW27522, AGW27524−AGW27527, AGW27529−AGW27531, and AGW27536) (Figure 1; Table 2). As detailed in Table 2 and illustrated in Figure 6Figures S1, S5, and S7, these MAT1-1-1 proteins exhibit variable N- and C-terminal truncations and diverse amino acid substitutions. The associated structural variations correspond to distinct AlphaFold 3D structural codes (U3N6U0, U3N919, U3N7G5, U3N6U4, U3NE79, U3NE87, and U3N7H7) (Table 2; Figure 2 and Figure 6, S1, S5, and S7). Collectively, these observations indicate that MAT1-2-1 proteins with identical primary structures may differentially naturally pair with structurally diverse MAT1-1-1 proteins across different O. sinensis strains.

2.6. Differences Between the Presence/Absence Patterns of Mating-Type Genes in H. Sinensis Genome Assemblies and Naturally Paired MAT1-1-1 and MAT1-2-1 Proteins Simultaneously Produced by the Analyzed O. Sinensis Strains

Table 3 summarizes reported cooccurrence and differential occurrence patterns of MAT1-1-1 and MAT1-2-1 proteins across C. sinensis insect‒fungal complexes, wild-type C. sinensis isolates, and O. sinensis strains. The MAT1-1-1 and MAT1-2-1 proteins cooccurred in 25.1% of the 183 samples but individually occurred in 52.5% and 22.4% of the samples, respectively.

The MAT1-1-1 and MAT1-2-1 genes coexist in the genome assemblies ANOV00000000 and LKHE00000000 of H. sinensis strains Co18 and 1229, respectively [64,93]. However, the MAT1-1-1 gene is absent from the genome assemblies LWBQ00000000 and NGJJ00000000 of the H. sinensis strains ZJB12195 and CC1406-20395, respectively, whereas the MAT1-2-1 gene is absent from the genome assembly JAAVMX000000000 of the H. sinensis strain IOZ07 [94,95,96]. This differential presence/absence pattern observed at the genomic level in H. sinensis strains, however, was not replicated at the protein level in the dataset analyzed herein. All 20 O. sinensis strains simultaneously produced naturally paired MAT1-1-1 and MAT1-2-1 proteins without omissions (Table 1 and Table 2). Taken together, the contrast between genomic and protein-level occurrence patterns suggest multiple fungal contributors to the mating proteins within these impure O. sinensis strains.

Tyrosine-to-methionine (Y-to-M) substitutions were identified within the MATα_HMGbox domains of MAT1-1-1 proteins encoded by the genome assemblies LKHE00000000 and JAAVMX000000000 of the H. sinensis strains 1229 and IOZ07, respectively [65,93,96]. In addition, serine-to-alanine (S-to-A) substitutions were detected within the HMG-box_ROX1-like domains of MAT1-2-1 proteins encoded by the genome assemblies ANOV00000000, LKHE00000000, LWBQ00000000, and NGJJ00000000 of the H. sinensis strains Co18, 1229, ZJB12195, and CC16-20395, respectively [64,65,93,94,95]. In contrast, neither the Y-to-M nor the S-to-A substitutions were observed in the DNA-binding domains of the MAT1-1-1 and MAT1-2-1 protein variants coproduced by the 20 mycologically impure O. sinensis strains analyzed here.

On the basis of the above observations, the 20 O. sinensis strains simultaneously produced paired MAT1-1-1 and MAT1-2-1 protein variants that exhibited N- and/or C-terminal truncations together with diverse amino acid substitutions at distinct sites. These observations underscore the extensive genetic variation within the MAT loci of distinct fungal genomes present in the impure O. sinensis strains, irrespective of their designation as pure strains previously reported by Li et al. [36]. In contrast to the diverse presence/absence patterns across the genome assemblies of the 5 H. sinensis strains, all 20 O. sinensis strains analyzed here consistently coproduced MAT1-1-1 and MAT1-2-1 proteins as natural pairs with variable domain architectures. Given that no repetitive copies of the MAT1-1-1 or MAT1-2-1 genes exist in the H. sinensis genome [69], these results are compatible with contributions from multiple cooccurring fungal taxa to the observation of differentially naturally paired mating proteins within these strains.

3. Discussion

3.1. The DNA-Binding Domains of Mating Proteins Coproduced in Pairs by O. Sinensis Strains Adopt Altered Tertiary Structures

Li et al. [67,68] reported the differential occurrence and expression of MAT1-1-1, MAT1-2-1, and pheromone receptor genes in H. sinensis. Full-length MAT1-1-1 and MAT1-2-1 proteins and their DNA-binding domains have also been shown to adopt diverse stereostructures and to cluster into several branched Bayesian clusters [65,69]. In aggregation, these genetic, transcriptional, and protein-structural findings are inconsistent with a self-fertilizing reproductive model but consistent with self-sterility for O. sinensis, involving heterothallic mating, hybrid reproduction, or even parasexual processes during the lifecycle of the C. sinensis insect–fungal complex [65,67,68,69].

The present study examined the differential natural pairing patterns of MAT1-1-1 and MAT1-2-1 protein variants simultaneously produced by each of 20 O. sinensis strains. The MATα_HMGbox and HMG-box_ROX1-like domains of these proteins exhibit diverse N- and/or C-terminal truncations and 1–4 amino acid substitutions. The identified variants are associated with alterations in hydrophobicity and secondary/tertiary structure, including changes in the 3D conformations of the hydrophobic cores formed by 3 α-helices characteristic of each domain (Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6 and Figure 7 and S1−S8; Table 1 and Table 2). These results provide protein-level evidence for evaluating how structural diversity among differentially naturally paired mating proteins may influence DNA binding specificity and downstream regulation of mating-type transcriptional programs in O. sinensis.

The HMGbox domains in mating proteins are multifunctional motifs that play essential roles in the transcriptional regulation of mating-related genes, O. sinensis sexual reproduction, and host adaptation within the C. sinensis insect‒fungal complex [58,59,65,67,68,69]. A conservative continuation model has proposed that the MATα-HMGBox domain of the MAT1-1-1 protein derives from a conserved ancestral mating-type HMG lineage, whereas the HMG-box_ROX1-like domain of the MAT1-2-1 protein reflects evolutionary recruitment of a distinct ROX1-like regulatory lineage through functional convergence [62,63,97,98,99,100].

The MATα_HMGbox and HMG-box_ROX1-like domains of the mating proteins each contain a hydrophobic core, consisting of an asymmetric arrangement of 3 tightly packed α-helices, which form a characteristic L-shaped fold that contributes to structural stability and resistance to conformational perturbations that can unfold or distort the domain [73,91,92]. Such conformations facilitate high-affinity, sequence-dependent binding to short AT-rich motifs located within the promoters of AT-rich promoter elements of target mating-related genes [61,101,102,103,104,105,106], which is widely recognized as an initial step in the transcriptional regulation of mating-type programs and controls mating identity and gene expression related to sexual reproduction [62,97,107,108].

Disruption of the hydrophobic core through the structural variations identified in the present study may compromise structural integrity and diminish symbiotic interactions between the MATα_HMGbox and the HMG-box_ROX1-like domains during DNA recognition. Relative to the reference O. sinensis MAT1-1-1 and MAT1-2-1 proteins, these structural variations may alter the precise recognition of target DNA segments and affect the transcriptional regulation of mating-type-specific genes that coordinate sexual reproduction throughout the lifecycle of the C. sinensis insect−fungal complex.

Although both the MATα_HMGbox domain of the MAT1-1-1 protein and the HMG-box_ROX1-like domain of the MAT1-2-1 protein belong to the HMG-box superfamily and share the characteristic L-shaped fold, they are associated with distinct target‒gene regulatory programs in mating-type regulation. The MATα_HMGbox domain primarily activates MAT1-1-associated transcriptional pathways, whereas the HMG-box_ROX1-like domain recognizes motifs within MAT1-2-associated promoters, including those of pheromone-receptor genes. Thus, these DNA-binding domains may function through complementary regulatory pathways that coordinate mating-type specification and downstream reproductive development [108,109,110,111]. The MATα_HMGbox domain is described as a transcriptional activator of cognate mating-type genes, whereas the HMG-box_ROX1-like domain has been implicated both in the repression of genes associated with the opposite mating type (e.g., “a-specific” genes) and in the activation of MAT1-2-specific transcriptional programs. This repressive capacity has been discussed in relation to ROX1 homology, the recruitment of corepressors, and structural interference mechanisms [112]. Therefore, the MATα_HMGbox and HMG-box_ROX1-like domains function specifically to regulate complementary target gene sets that establish mating-type identity during the sexual reproduction of O. sinensis in the lifecycle of the C. sinensis insect–fungal complex.

It has been hypothesized that the MATα_HMGbox domain of the MAT1-1-1 protein and the HMG-box_ROX1-like domain of the MAT1-2-1 protein may form heterodimers that promote outbreeding and initiate the sexual cycle in heterothallic fungi [113,114,115]; however, direct experimental evidence in O. sinensis remains limited. The interplay of the 2 domains may be mediated through complementary electrostatic properties, providing a conceivable basis for cooperative DNA binding. Such in vivo interactions may facilitate the transition from asexual growth to sexual development by modulating DNA binding specificity and downstream transcriptional programs associated with mating and meiosis in O. sinensis [116]. Thus, experimental validation through protein biochemistry and reproductive physiology studies remains necessary to evaluate this hypothesis, clarify species-specific reproductive mechanisms, and determine how structural variations of the mating proteins influence complementary interactions and mating function in O. sinensis [65,67,68,69]. Such investigations are particularly important given that the genome-independent and multi-genotypic nature of O. sinensis is not yet fully understood [9,10,11,12,13,19,24,27,30,35,37,38,54,117]. Protein variations may therefore provide key evidence for evaluating competing hypotheses regarding the anamorph–teleomorph relationship and the reproductive strategy of O. sinensis [14,20,36,45,64,65,67,68,69,72].

3.2. Complex Phylogenetic Heterogeneity of the O. Sinensis Strains

Bushley et al. [14] used a dual-fluorescence microscopy approach and observed multicellular heterokaryotic microstructures within C. sinensis ascospores and hyphae, including mono-, bi-, tri-, and tetra-nucleate cellular configurations. Zhang & Zhang [72] questioned whether such polynucleate cells might harbor heterogeneous hereditary materials in an inconsistent manner. Related observations from culture-dependent and culture-independent studies [12,13,36,37,38] have documented the presence of multiple fungal components within C. sinensis ascospores and ascocarps and in strains described as pure H. sinensis. These findings support the hypothesis proposed by Zhang & Zhang [72] that multicellular heterokaryotic cells may contain heterogeneous hereditary substances.

For phylogenetic assessment, Li et al. [36] deposited ITS sequences for 15 of the 20 O. sinensis strains in GenBank; unfortunately, no ITS sequences were reported for the remaining 5 strains (Table 1). Among the 15 sequenced strains, only Group-A (GC-biased Genotype #1 H. sinensis) ITS sequences were reported in 10 strains, whereas both Group-A and Group-C (AT-biased Genotype #17) ITS sequences were reported for 5 other strains (Table 1 and Tables 1 and S1−S2) [9,10,12,13,36,37]. The culture-dependent methodology employed by Li et al. [36] included (1) 4 primer pairs for ITS sequence amplification (); (2) annealing temperatures of 53 °C for the universal fungal primers ITS5/ITS4 [118], 55 °C for the primers GAF/GAR (specific for Group-A sequences), and 60 °C for the primers GCF/GCR (specific for Group-C sequences); and (3) direct sequencing of mixed amplicon pools. The results of the comparative analysis shown in Table S4 demonstrate various primer similarities/coverages aligned with representative GC- and AT-biased O. sinensis genotypes and suggest variable amplification efficiency among genotypes under the study settings. Consequently, some cooccurring fungal taxa in mixed amplicon samples may have been overlooked. Apart from ignoring nonculturable taxa, Li et al. [36] relied on direct sequencing of dominant amplicons and did not employ extensive amplicon cloning–sequencing with sufficient colony sampling to unbiasedly detect high- and low-abundance sequences [36,37]. Notably, Table S4 also shows identical similarity and coverage values for the ITS5/ITS4 primers when aligned to Genotypes #1 and #13–14, which have been reported to cooccur within C. sinensis ascospores [13]. This observation has prompted discussion regarding the potential co-amplification of low-abundance Genotypes #13–14 sequences from certain ascospore-derived strains.

In contrast, Li et al. [12,13,37] employed culture-independent approaches that included multiple primer pairs, touch-down PCR protocols, cloning–sequencing with sampling of >30 white colonies, and MALDI‒TOF SNP mass spectrum genotyping using multiple extension primers to characterize the fungal composition of C. sinensis ascospores and strains described as pure H. sinensis. These studies revealed the cooccurrence of multiple genomically independent fungal taxa, including psychrophilic GC-biased Genotypes #1 and #13−14 and AT-biased Genotypes #5−6 and #16 of O. sinensis, and mesophilic Samsoniella hepiali, formerly classified as Paecilomyces hepiali [119]. Together with the methodological considerations noted above, these findings suggest that certain lower-abundance amplicons and nonculturable taxa may have been overlooked in the culture-dependent study conducted by Li et al. [36].

Li et al. [36] proposed that all 20 O. sinensis strains represented pure cultures and interpreted the detected AT-biased sequences as nonfunctional repetitive ITS pseudogenic copies within a single GC-biased H. sinensis genome. Li et al. [120,121] further hypothesized that nonfunctional ITS or rRNA pseudogenes may arise through repeat-induced point (RIP) mutation. However, other studies have reported that GC- and AT-biased Genotypes #2–17 of O. sinensis are absent from the genome assemblies of 5 H. sinensis strains and instead represent genome-independent lineages, despite these genotypes may share a common heredity ancestor [9,10,11,12,13,24,38,64,90,93,94,95,96]. Multiple O. sinensis genotypes have also been reported to differentially coexist within different compartments of the C. sinensis insect‒fungal complex, with their relative abundances varying dynamically and reciprocally during C. sinensis maturation, thereby providing additional evidence supporting the genome-independent nature of these genotypes [9,10,11,12,13,27,35,37]. In summary, these findings do not support the pseudogene hypothesis that the AT-biased sequences represent RIP-derived repeats within a single genome of GC-biased Genotype #1 H. sinensis. Instead, they suggest that GC- and AT-biased sequences represent genome-independent taxa.

3.3. Heterogeneous Fungal Sources of MAT1-1-1 and MAT1-2-1 Proteins Simultaneously Produced in Paired form by the Analyzed O. Sinensis Strains

3.3.1. Inconsistency Between Cooccurring Mating Proteins in Reported “Pure” O. Sinensis Strains and the Differential Presence of Mating-Type Genes in Pure H. Sinensis Strains

As summarized in Section 2.6, the MAT1-1-1 and MAT1-2-1 genes exhibit differential presence/absence patterns across the genome assemblies of Genotype #1 H. sinensis strains [64,69,93,94,95,96]. In contrast, the present study reveals the simultaneous production of MAT1-1-1 and MAT1-2-1 proteins in pairs across all 20 O. sinensis strains, despite substantial variation in terminal truncations and amino acid substitutions among these proteins (Table 2; Figure 5) [14,36]. The observed differential natural pairing patterns among highly divergent MAT1-1-1 and MAT1-2-1 variants are not consistent with derivation from a single uniform genetic background. When considered together with the methodological issues discussed in Section 2.3 and Section 3.2, these protein-level observations are compatible with contributions from multiple cooccurring fungal taxa (including taxa potentially overlooked in culture-dependent profiling approaches) to the pool of mating proteins detected in these cultures. Further unbiased, culture-independent characterization may help clarify the composition and cooccurrence patterns of genomically independent O. sinensis genotypes and heterospecific fungal taxa across these strains [12,13,27,35,37].

3.3.2. Potential Heterogeneity Among Reported Pure O. Sinensis Strains

The highly anticipated challenge concerns the strain CS68-2-1229 in Group I (highlighted in dark green and underlined in Figure 2 and Figure 4), which reportedly contain only GC-biased Genotype #1 in this monoascospore-derived culture [36]. Two MAT1-1-1 protein sequences are available in GenBank for this strain: AGW27560 (full-length; 372 amino acids) and AGW27528 (truncated; 301 amino acids) (Figure S1). These 2 sequences are 100% identical throughout their shared region, whereas AGW27528 exhibits 81% query coverage relative to the authentic full-length MAT1-1-1 protein because of 63- and 8-residue truncations at the N- and C-termini, respectively, which is consistent with amplification using different primer sets targeting MAT1-1-1 transcripts. In addition, this strain was reported to simultaneously produce the truncated MAT1-2-1 protein AGW27548 (175 amino acids; Figure 3 and Figure 4 and S2), which shares 99.4% sequence similarity with the authentic full-length MAT1-2-1 protein but displays only 70% sequence coverage because of a 74-residue C-terminal truncation relative to the authentic 249-amino-acid MAT1-2-1 protein. Notably, all 3 mating protein records from CS68-2-1229 were released in GenBank on the same date (28-SEP-2013). The MATα_HMGbox domain (162 amino acids) of AGW27528 is among the 9 longest truncated MAT1-1-1 proteins (Figure 2). In contrast, the HMG-box_ROX1-like domain (49 amino acids) of AGW27548 is 10 residues shorter than the longest truncated HMG-box_ROX1-like domain (59 amino acids) observed in AGW27543 (Figure 4) [65,70]. Moreover, AGW27548 has a tyrosine-to-histidine (Y-to-H) substitution, and exhibits altered secondary and tertiary structural characteristics in the AlphaFold 3D model U3N9W5 (Figure 1 and Figures 1 and S2; Table 1 and Table 2). Because there are no repetitive copies of the MAT1-1-1 or MAT1-2-1 genes across the genome assemblies of 5 H. sinensis strains [69], these observations suggest that AGW27528 may be encoded by the authentic MAT1-1-1 gene in the H. sinensis genome, whereas AGW27548 may have been derived either from a variant MAT1-2-1 gene within the same H. sinensis genome or from a genome-independent taxon cooccurring within the CS68-2-1229 culture. Whether strain CS68-2-1229 represents a single-taxon culture with the significantly variant MAT1-2-1 gene or instead contains multiple fungal taxa warrants re-evaluation.

Among the purportedly homogeneous O. sinensis strains (Table 1), for which only GC-biased Genotype #1 H. sinensis was reported, strain CS561-964 in Group II represents another special example (Table 2) [14,36]. It produced the MAT1-1-1 protein AGW27526, which contains 63- and 13-residue truncations at the N- and C-termini, respectively, together with amino acid substitutions (CDRA-to-SSFT) within the MATα_HMGbox domain (Figure 1 and Figure 2 and S7; Table 2). The authentic peptide segment “CDRA” of AGW27560 (encoded by the DNA segment “TGTGATCGAGCG” in the MAT1-1-1 gene KC437356) is replaced with the variant peptide segment “SSFT” in AGW27526 (encoded by the DNA segment “TCCTCTTTCACG” in the MAT1-1-1 gene KC429528). In combination with the terminal truncations, these sequence variations suggest that AGW27526 was derived from a genome-independent, cooccurring fungal taxon within the CS561-964 culture rather than from a single Genotype #1 H. sinensis; the ITS sequences associated with the contributing taxon may have been overlooked by Li et al. [36], given the methodological limitations discussed in Section 2.3 and Section 3.2.

Reanalysis of the previously reported homogeneous O. sinensis strains raises discussions concerning fungal purification, in vitro culture conditions, and unbiased, comprehensive characterization of cooccurring fungal taxa in such cultures. (1) Do the homogeneous strains listed in Table 1 genuinely harbor only GC-biased Genotype #1, or do they contain additional fungal taxa? (2) Do the strains previously reported to contain Genotypes #1 and #17 (Tables S1−S2) also contain additional cooccurring taxa that were not detected under the culture-dependent settings? One plausible interpretation is that at least some of the differentially naturally paired MAT1-1-1 and MAT1-2-1 proteins exhibiting various truncations and diverse amino acid substitutions are not necessarily produced by GC-biased Genotype #1 H. sinensis but instead originate from additional cooccurring fungal taxa that were overlooked under the culture-dependent and suboptimal experimental conditions employed by Li et al. [36]. Thus, such cooccurring taxa may contribute functionally and symbiotically to sexual reproduction-related processes during the lifecycle of the C. sinensis insect‒fungal complex.

3.4. Cooccurring Fungal Taxa as Potential Sexual Partners for Self-Sterile O. Sinensis During Heterothallic or Hybrid Reproduction

H. sinensis (GC-biased Genotype #1 of O. sinensis) was previously postulated to be the sole anamorph of O. sinensis teleomorph [45]. The anamorphic name was replaced by the teleomorphic name O. sinensis in accordance with the IMA’s “One Fungus=One Name” nomenclature principle [49,51,52,53]. Although the longstanding fungal taxonomic name C. sinensis was revised to O. sinensis in 2007 using H. sinensis strain EFCC7287 as the nomenclature reference [48], the new name O. sinensis has been uniformly applied to all 17 genomically independent genotypes of O. sinensis, including 11 GC-biased and 6 AT-biased genotypes that may share a common ancestral origin [9,10,24,48,49,50].

To support the sole anamorph hypothesis for H. sinensis [45], O. sinensis has been proposed to be self-fertile (homothallic or pseudohomothallic) on the basis of the findings of both the MAT1-1-1 and MAT1-2-1 genes within the genomes of a few H. sinensis strains [14,64]. However, several studies [64,122,123] have reported unsuccessful attempts to artificially cultivate fruiting bodies and ascospores of C. sinensis insect–fungal complexes using single-species fungal cultures. Zhang et al. [51] summarized the multi–decades history of unsuccessful cultivation attempts performed in academic research-oriented settings, and Qin et al. [124] discussed technical obstacles associated with artificial cultivation of C. sinensis fruiting bodies and ascospores. The repeated failures have been attributed, at least partially, to the extremely low and unstable infection rates of H. sinensis [37,125,126,127]. In contrast to the nearly noninfectious characteristics reported for pure H. sinensis cultures, Li et al. [37] advanced an alternative inoculation strategy employing cocultures composed of natural fungal clusters reflecting the native abundance ratios of multiple cooccurring taxa (including several GC- and AT-biased genotypes of O. sinensis together with S. hepiali) isolated directly from the natural C. sinensis insect–fungal complex. This coculture-based approach showed up to 39-fold stronger infectivity to Hepialus armoricanus larvae compared to the conidia or mycelia of pure H. sinensis (P<0.001). These observations suggest that single-species cultivation strategies may be insufficient to reproduce the complete lifecycle of C. sinensis under artificial cultivation conditions and therefore motivate a re-evaluation of the contentious sole-anamorph and self-fertilization hypotheses for H. sinensis in accordance with all 4 criteria of Koch’s postulates.

With respect to reproductive strategies, homothallic or pseudohomothallic reproduction was proposed for O. sinensis [14,64]. However, subsequent investigations revealed extensive genetic diversity within O. sinensis, including at least 17 distinct genome-independent genotypes representing distinct fungal lineages [9,10,12,13,24,27,30,35,54]. Zhang and Zhang [72] and Li et al. [67,68] reported the differential occurrence and expression of MAT1-1-1, MAT1-2-1, and pheromone receptor genes across numerous wild-type C. sinensis isolates and H. sinensis. More recent studies [65,69] have demonstrated heteromorphic tertiary structures among variant MAT1-1-1 and MAT1-2-1 proteins, particularly variations within their DNA-binding domains. The results of the present study further revealed differential natural pairings of the mating proteins with diverse truncations and various amino acid substitutions that were simultaneously coproduced by numerous O. sinensis strains. These observations suggest divergent origins for the mating proteins, involving at least GC-biased Genotypes #1 and #3 and AT-biased Genotype #17 (or #5) of O. sinensis [65, the present study]. Because the MAT1-1-1 and MAT1-2-1 genes each occur as a single copy within the H. sinensis genome with no repetitive copies [69], these structural observations provide protein-level evidence supporting heterogeneous fungal origins of mating-type loci among numerous wild-type C. sinensis isolates and O. sinensis strains. In brief, the available genetic, transcriptional, and protein structural evidence does not support a self-fertilization reproductive model for O. sinensis and instead supports the possibility of self-sterility requiring heterothallic mating, hybridization, or parasexual outcrossing to accomplish sexual reproduction [13,65,67,68,69,128,129,130,131,132,133,134,135,136].

A species discrepancy between the GC-biased H. sinensis strains used as inoculants and an AT-biased Genotype #4 teleomorph in cultivated fruiting bodies was observed in the successful industrial cultivation of C. sinensis insect–fungal complexes [46]. Notably, AT-biased O. sinensis sequences are absent from the genome assemblies ANOV00000000, JAAVMX000000000, LKHE00000000, LWBQ00000000, and NGJJ00000000 of H. sinensis strains Co18, IOZ07, 1229, ZJB12195, and CC1406-20395, respectively [9,10,11,12,13,24,30,35,36,37,38,64,90,93,94,95,96,117]. Moreover, multiple studies have provided genomic, transcriptomic, and protein-structural evidence supporting the genome-independent nature of O. sinensis genotypes [27,30,35,36,37,38,90]. These findings do not support the hypotheses that AT-biased genotypes represent nonfunctional ITS or rRNA pseudogenes generated through RIP mutation within a single GC-biased H. sinensis genome [36,120,121]. Therefore, the industrial cultivation success, together with the inoculant/teleomorph discrepancy, suggests that the sole-anamorph hypothesis for H. sinensis may not satisfy all 4 criteria of Koch’s postulates.

Subsequently, Li et al. [125] reinterpreted the inoculant used in the industrial cultivation project as a conidial suspension derived from cultures of C. sinensis ascospores, thereby revising the previous report that the inoculant was pure H. sinensis strains [46]. The revised report, however, did not provide a detailed description of the fungal components of the inoculant or clarify the teleomorphic fungus in the cultivated insect–fungal complexes. Somewhat confusingly, several independent studies failed to detect AT-biased Genotype #4 in C. sinensis ascospores [12,13,36,37], arguing whether Genotype #4 is indeed involved in the sexual reproduction of O. sinensis. Thus, the inoculant–teleomorph discrepancy reported in cultivated insect–fungal complexes [46] remains unresolved based on the currently available evidence.

The use of multi-taxon inoculants is considered a key factor in the industrial success in cultivating insect–fungal complexes, which is supported by several observations. (1) Li et al. [125] described the use of an inoculant derived from C. sinensis ascospores in an effort to overcome the weak infectivity of pure H. sinensis, and other studies have demonstrated that C. sinensis ascospores harbor multiple genotypic and heterospecific fungal taxa [12,13,36]. (2) AT-biased Genotype #4 of O. sinensis is absent in C. sinensis ascospores [10,12,13,36,38]. Thus, the sole Genotype #4 teleomorph in the cultivated C. sinensis insect–fungal complex [46] may instead originate from the coculture of additional components of the natural complex, such as the stroma and stromal fertile portion that indeed harbor Genotype #4, alongside other fungal taxa. Consistent with this possibility, Li et al. [37] reported that compared with the conidia and mycelia of pure H. sinensis, cocultures containing multiple genotypic and heterospecific fungal taxa derived from wild-type C. sinensis isolates significantly increased the infectivity to H. armoricanus larvae. (3) The industrial cultivation system was supplemented with soil collected from natural C. sinensis production regions on the Qinghai‒Tibet Plateau [46], introducing additional environmental microbial components [137,138,139].

Because reports on the purification of GC- and AT-biased Genotypes #2−17 of O. sinensis are currently limited, the genomic and transcriptomic characterization of diverse MAT1-1-1 and MAT1-2-1 genes across genome-independent O. sinensis genotypes remain unknown, particularly with respect to structural variation within key DNA-binding domains. Naturally paired mating proteins are widely regarded as key determinants of mating compatibility and the initiation of fungal sexual reproduction [56,57,58,59,60]. Evidence demonstrating the differential occurrence and expression of mating-type and pheromone receptor genes in H. sinensis suggests self-sterility rather than self-fertilization for O. sinensis [14,45,63,66,67,68,69]. Furthermore, observations of heteromorphic tertiary structures within variant MATα_HMGbox (MAT1-1-1) and HMG-box_ROX1-like (MAT1-2-1) domains provide additional protein-level support for models favoring heterothallic or hybrid reproduction over strict homothallism, involving divergent symbiotic fungal taxa within heterogeneous cultures. Thus, self-sterility offers a trustworthy explanation for the repeated inability of single-species cultures to reproduce key lifecycle stages under artificial conditions and further underscores the need for continued investigation of multi-taxon synergy across both natural and artificially cultivated C. sinensis insect–fungal complexes.

4. Materials and Methods

4.1. MAT1-1-1 and MAT1-2-1 Protein Sequences from O. Sinensis Strains

AI-predicted 3D structural morphs for 20 truncated MAT1-1-1 proteins (AGW27517− AGW27536) derived from 20 O. sinensis strains were obtained from the AlphaFold database (Table 1) [14]. These strains simultaneously produced MAT1-2-1 proteins (AGW27537−AGW27556). Predicted 3D structural morphs for 5 truncated and 6 full-length MAT1-2-1 proteins, but not for the remaining 9 full-length MAT1-2-1 proteins, were also obtained from the AlphaFold database.

Li et al. [36] reported that 11 of the 20 O. sinensis strains were obtained from fresh caterpillar body samples (marked as “TS” in Table 1) harvested from various C. sinensis production regions on the Qinghai‒Tibet Plateau. The remaining 9 strains were from the culture of monoascospores of C. sinensis insect‒fungal complexes (“SS” in Table 1) harvested from Maqên, Guoluo, Qinghai Province, China.

The MAT1-1-1 protein sequences were encoded by the MAT1-1-1 cDNAs, which were amplified from total RNA extracted from the O. sinensis strains using the primer pair m1F3/m1R3, while the MAT1-2-1 cDNAs were amplified using the primer pair Mat1-2F/Mat1-2R [14]. Li et al. [36] deposited ITS sequences in GenBank for 15 of the 20 O. sinensis strains for phylogenetic positioning and reported the detection of only Group-A ITS sequences (GC-biased Genotype #1 of O. sinensis) for 10 O. sinensis strains (shown in brown in Table 1). For 5 other strains (shown in red in Table 1), they reported the ITS sequences from both Group-A (GC-biased Genotype #1) and Group-C (AT-biased Genotype #5). The remaining 5 O. sinensis strains shown in black in Table 1 lack ITS sequence information in GenBank. Tables S1−S2 show the alignment results between the ITS sequences from the O. sinensis strains reported by Li et al. [36] and the representative ITS sequences of the GC- and AT-biased O. sinensis genotypes.

4.2. Alignment of the DNA-Binding Domain Sequences of the Mating Proteins

The amino acid sequences of the MAT1-1-1 and MAT1-2-1 proteins were aligned using the GenBank Blastp program (https://blast.ncbi.nlm.nih.gov/ (Bethesda, MD, USA), accessed from 18 June 2025 to 3 June 2026).

4.3. Amino Acid Properties and Scale Analysis

The amino acid sequences of the DNA-binding domains of the mating proteins were scaled according to the general physicochemical properties of their side chains using the ExPASy ProtScale tool (https://web.expasy.org/protscale/; Basel, Switzerland; accessed from 18 May 2025 to 14 April 2026) (Table S3) [65,68,69,73,74,75,76,77]. The MATα_HMGbox domain of MAT1-1 proteins (amino acids 51→225 of the reference sequence AGW27560 derived from the H. sinensis strain CS68-2-1229 [14]) and the HMG-box_ROX1-like domain of MAT1-2-1 proteins (amino acids 127→197 of the reference sequence AEH27625 derived from the H. sinensis strain CS2 [70]) were each extended by 9 additional amino acid residues upstream and downstream of the domains to ensure complete domain coverage. These sequences were then plotted sequentially using the ExPASy ProtScale algorithm with a window size of 9 amino acid residues and a linear weight variation model [65,68,69,73,74,75] to generate ExPASy ProtScale plots and predict variations in the hydrophobic properties (hydropathy index) and secondary structural features, including α-helices, β-sheets, β-turns, and random coils. The topological profiles and waveform patterns derived from the ProtScale outputs were compared to characterize variations in hydrophobicity and 2D structural properties within the DNA-binding domains of the mating proteins.

4.4. Tertiary Structures of the Mating Proteins Predicted by AlphaFold

To examine heteromorphic stereostructures in this study, the 3D structures of the MAT1-1-1 and MAT1-2-1 proteins from O. sinensis strains were predicted computationally from their amino acid sequences using AlphaFold, an artificial intelligence (AI)-driven machine learning platform (https://alphafold.com/, Cambridgeshire, UK; accessed from 18 June 2025 to 20 October 2025) [78,79,80,81,82,83,84,85,86,87,88].

The AlphaFold database reports per-residue model confidence using the predicted local distance difference test (pLDDT), which assigns a confidence score ranging from 0 to 100 to each residue [78,79,80,81,83,84,85]. A standardized color-coded scheme is applied to the 3D structural models to visualize confidence levels: residues identified with very high confidence (pLDDT>90) are colored dark blue; those identified with high confidence (90>pLDDT>70) appear light blue; those identified with low confidence (70>pLDDT>50) are shown in yellow; and those identified with very low confidence (pLDDT<50) are shown in orange [65,86,87,89].

4.5. Differential Natural Pairing of Variant MAT1-1-1 and MAT1-2-1 Proteins Simultaneously Produced by O. Sinensis Strains and Correlations Among the Primary, Secondary, and Tertiary Structures of the DNA-Binding Domains

Natural pairs of variant MAT1-1-1 and MAT1-2-1 proteins simultaneously expressed in the O. sinensis strains were compared to evaluate their complex fungal origins. Local magnification of changes in the heteromorphic AlphaFold-predicted 3D structures surrounding the sites of variation in the MAT1-1-1 and MAT1-2-1 proteins to clearly visualize subtle 3D deviations in structure. Truncations and amino acid substitutions occurring within the DNA-binding domains of the mating proteins were further correlated with deviations in topological structural and waveform changes in hydrophobicity and the contents of α-helices, β-sheets, β-turns, and random coils, as illustrated in the ExPASy ProtScale plots, with the locally magnified 3D structures surrounding the sites of variation.

5. Conclusions

This study provides protein structure evidence consistent with self-sterility of O. sinensis and supports reproductive models involving heterothallic mating or hybridization. Across the 20 O. sinensis strains, MAT1-1-1 and MAT1-2-1 protein variants showed diverse N- and/or C-terminal truncations and 1–4 amino acid substitutions within their critical DNA-binding domains. These primary structure variations were associated with corresponding alterations in the hydrophobic properties as well as the secondary and tertiary conformations of the MATα_HMGbox and HMG-box_ROX1-like domains of the differentially naturally paired MAT1-1-1 and MAT1-2-1 proteins, respectively. The observed structural heterogeneity among naturally paired mating proteins suggests heterogeneous fungal sources cooccurring within impure O. sinensis strains. When integrated with previous genetic and transcriptional observations (including differential occurrence, alternative splicing, and differential transcription of the MAT1-1-1, MAT1-2-1, and pheromone receptor genes), the protein-level heterogeneity and differential natural pairing patterns identified in the present study are more consistent with the participation of multiple cooccurring fungal taxa within these cultures than with a strictly homothallic or pseudohomothallic self-fertilization model. In aggregation, these findings refine the current belief of the sexual reproductive biology of O. sinensis and further support the hypothesis that compatible mating partners and multi-taxon synergistic symbiosis may be the key to accomplish sexual reproduction during the lifecycle of the C. sinensis insect–fungal complex on the Qinghai–Tibet Plateau.