Biological and biochemical studies on cell surface functions in microorganisms used in brewing and fermentation industry

Abstract

When brewing microorganisms, which include bacteria and fungi, act on solid cereal substrates, the microbial cell surface interacts with the substrate. When microorganisms use sugars and amino acids released by hydrolysis of the substrate, this occurs on the cell surface. Throughout my career, I have focused on functional studies of cell surface molecules such as solute transporters, cell wall components, and bio-surfactants and applied the knowledge obtained to the development of fermentation technologies. In this review, I describe (i) catabolite control by sugar transporters and energy generation coupled with amino acid decarboxylation in lactic acid bacteria; (ii) recruitment of a polyesterase by the fungal bio-surfactant proteins to polyesters and subsequent promotion of polyester hydrolysis; and (iii) hyphal aggregation via cell wall α-1,3-glucan and galactosaminogalactan in aspergilli and the development of a novel liquid culture method with hyphal dispersed mutants lacking these two polysaccharides.

Graphical Abstract

Discovery of new functions of cell surface molecules in fermentation microorganisms and development of new applications of the functions

Open in new tabDownload slide

Brewing of sake, soy sauce, miso (fermented soybean paste), and katsuobushi (fermented dried bonito) are traditional fermentation industries in Japan. To create a flavorful brew, koji mold, lactic acid bacteria, yeast on grain substrates (such as rice, barley, and soybeans), and fish are skillfully combined (Abe and Gomi 2007). Japanese fermentation greatly differs from that used in beer and wine brewing in Europe and the United States. Because of the complexity of Japanese brewing, some microbial reactions are likely to have remained undiscovered. When microorganisms act on solid grain substrates in fermentation, their cell surface layer interacts with the solid substrate, and the cells use low molecular weight nutrients such as sugars and amino acids supplied by grain hydrolysis. On solid substrates, koji molds produce a large amount of hydrolytic enzymes such as amylases, proteases, and lipases (Abe and Gomi 2007). These enzymes hydrolyze solid substrates, and free sugars and amino acids are further fermented by lactic acid bacteria and yeasts. I have focused on the functions of cell surface factors, such as solute transporters, surface-active molecules, and cell wall components, to elucidate their functions and develop applied technologies.

I used to work in the soy sauce brewing industry and was involved in development of the soy sauce lactic acid bacterium Tetragenococcus halophilus (previously named Pediococcus halophilus) that can improve the quality of soy sauce (Abe and Uchida 1989; Abe and Uchida 1990; Abe and Uchida 1991). We found the regulatory function of sugar transporters in catabolite control (Abe and Uchida 1990; Abe and Uchida 1991) and a metabolic energy generation mechanism coupled with amino acid transport and intracellular decarboxylation of an amino acid (Nanatani and Abe 2011). Understanding and manipulating bacterial solute transporters were necessary for creating a new bacterial strain for the soy sauce industry (Nanatani and Abe 2011). This topic is covered in the first section of this review.

I then moved from the fermentation industry to academia, and started to study biology and biochemistry of the cell surface of aspergilli including the koji mold Aspergillus oryzae. Over 3 million species of filamentous fungi are thought to exist (Chambergo and Valencia 2016; Ye and Ai 2022), and they are important in global material circulation because they hydrolyze solid biomaterials (Tanaka et al. 2022). In both pathogenesis and fermentation, filamentous fungi grow on solid bio-materials and hydrolyze them. Since fungal cells grow mainly at the air–solid or liquid–solid interfaces, they also produce the bio-surfactant proteins hydrophobins, which coat the cell surface and enable fungal cells to live at the interfaces (Tanaka et al. 2022). We discovered that hydrophobins not only reduce surface tension but also function as a scaffold to recruit hydrolytic enzymes to the interfaces (Takahashi et al. 2005). I describe this function of hydrophobins in the second section of this review.

Polysaccharides are the major cell surface components of filamentous fungi. Fungal cell walls maintain cell morphology and protect cells from environmental stresses (Yoshimi, Miyazawa and Abe 2016; Miyazawa and Abe 2017; Yoshimi et al. 2022). Fungal cells adapt to environmental stresses by responding appropriately, including rebuilding and repairing their cell walls. The immune systems of plants and animals recognize cell wall polysaccharides of pathogenic fungi (Yoshimi et al. 2022; Latge 2023). We discovered that α-1,3-glucan in the cell wall and galactosaminogalactan (GAG) in the biofilm are adhesive factors for hyphal aggregation in aspergilli grown in liquid culture (Miyazawa et al. 2019). Aspergillus oryzae mutants lacking α-1,3-glucan and GAG showed hyphal dispersion, and the mutants have been used to improve mixing conditions in bioreactors and to increase recombinant enzyme production (Ichikawa et al. 2022). I describe recent developments in this field in the third section of this review.

Solute transporters

In microbial fermentation, productivity has been increased by modifying the intracellular metabolic pathways. To further increase productivity and produce new compounds, substrate uptake into the cells and product secretion out of the cells may be targeted for improvement if these steps are rate limiting. In this section, I describe how transporters of the lactic acid bacterium T. halophilus have been used to improve fermentation.

Catabolite control and sugar transport

Catabolite repression of xylose operon in lactic acid bacterium T. halophilus in fermentation

In soy sauce fermentation, koji-mash is a mixture of steamed soy bean and baked wheat grains fermented by the koji mold A. oryzae or Aspergillus sojae; the koji-mash is mixed with brine and hydrolyzed by amylases and proteases produced by the molds, which is so called moromi-mash (Abe and Gomi 2007). A large amount of glucose is released from wheat starch, and pentoses (such as xylose and arabinose) are released from the cell wall of soy bean and wheat grains, besides galactose. Glucose is used by T. halophilus and yeast Zygosaccharomyces rouxii, but pentoses remain in the fermentation of moromi-mash (Abe and Uchida 1989; Abe and Uchida 1991). Although pentose fermentation in the presence of glucose is necessary to prevent soy sauce browning by amino-carbonyl reaction, pentoses cannot be fermented in the presence of glucose as long as wild-type strains of T. halophilus or Z. rouxii are used (Abe and Uchida 1991).

In bacterial fermentation, uptake of sugars such as glucose and fructose causes catabolite repression, which becomes a limiting factor for production and has to be released in some way. Until the early 1990s, the mechanism of catabolite control in gram-positive bacteria remained unknown (Henkin et al. 1991; Fujita and Miwa 1994; Hueck, Kraus and Hillen 1994), but I succeeded in breeding strains of T. halophilus in which catabolite repression was released by controlling sugar transporters and the glycolytic metabolism. In both gram-positive and gram-negative bacteria, phosphoenolpyruvate-dependent sugar: phosphotransferase system (PTS) is a major sugar transporter family and evokes catabolite control (Abe and Uchida 1989; Abe and Uchida 1990; Abe and Uchida 1991). We found that T. halophilus has mannose: PTS as the main glucose transporter, and mutations in its membrane component EII release catabolite repression of the xylose operon, which consists of xylA for xylose isomerase, xylB for xylulose kinase, and xylT for xylose transporter (Abe and Uchida 1989; Abe and Uchida 1991; Takeda et al. 1998). Besides the single mutation in EII (EII^man-), mutations in the phosphofructokinase (PFK) gene (pfk^-) and glucokinase (GK) gene (gk^-) decrease intracellular levels of fructose-1,6-diphosphate (FDP) by blocking glycolysis and lead to a more effective release of catabolite repression than that caused by the EII^man- mutation (Abe and Higuchi 1998).

In low-GC gram-positive bacteria, including lactic acid bacteria, the molecular mechanism of catabolite repression had been well studied by 2000 and is as follows. The transcription factor CcpA (catabolite control protein A) forms a homodimer, and each subunit associates with the cytoplasmic PTS component HPr, whose Ser46 is phosphorylated (HPr(Ser46-P)) at a high intracellular concentration of FDP during glucose metabolism via mannose: PTS (or glucose: PTS) and glycolysis (Hueck and Hillen 1995; Hueck et al. 1995). The CcpA–HPr(Se46-P) complex attaches to cis-elements upstream of the promoters of target catabolic operons and represses their transcription (Hueck and Hillen 1995; Poolman 2002). We found a CcpA-binding sequence in the xylose operon of T. halophilus, and we think that CcpA–HPr(Ser46-P) evokes catabolite repression of the xylose operon in the presence of both glucose and xylose (Takeda et al. 1998) (Figure 1a). A stronger release of catabolite repression in a triple mutant (EII^man-, pfk^-, and gk^-) than in the single mutant EII^man- is attributable to the lower intracellular levels of FDP which is necessary for the formation of the CcpA–HPr(Ser46-P) complex (Abe and Higuchi 1998) (Figure 1b).

Catabolite inhibition

In addition to the negative regulation of transcription such as catabolite repression, the inhibitory effect of glucose on sugar transporters, known as “inducer exclusion”, is mediated by PTS in gram-positive bacteria, including lactic acid bacteria (Titgemeyer and Hillen 2002). Inducer exclusion is caused by an allosteric inhibition of transporters of substances, which are non-preferentially metabolized, by HPr(Ser46-P). Addition of glucose inhibits maltose transport in ccpA mutants of Lactobacillus casei, suggesting the presence of a catabolite inhibition (inducer exclusion) mechanism independent of CcpA transcription regulation (Viana et al. 2000). In Enterococcus faecalis (Charrier et al. 1997) and Lb. casei (Monedero et al. 2007), glycerol kinase required for glycerol metabolism is activated by HPr(His15-P), whose phosphate is transferred from phosphoenolpyruvate by enzyme I (EI) of the PTS component, and mutations in ptsH or ptsI inhibit glycerol metabolism. When glucose is used in the presence of glycerol, glucose transport by PTS decreases the level of HPr(His15-P) and inhibits glycerol metabolism.

In T. halophilus, maltose, glycerol, xylose, and arabinose are metabolized as non-PTS sugars (Abe, Higuchi and Yamato 1998). We isolated temperature-sensitive EI mutants (EI^ts) that have EI activity at 30°C but not at 42°C; these mutants cannot grow on media containing maltose or glycerol as a sole carbon source at 42°C, similar to EI mutants of E. faecalis (Charrier et al. 1997) and Lb. casei (Monedero et al. 2007). Very interestingly, we found that EI^ts mutants can grow on media containing xylose or arabinose as a sole carbon source at 42°C, which means that pentose transporters for xylose and arabinose are insensitive to inducer exclusion (Abe, Higuchi and Yamato 1998). In other words, sensitivity to inducer exclusion depends on non-PTS sugars (Abe, Higuchi and Yamato 1998). The EI^ts mutants of T. halophilus are used for selective fermentation of non-PTS sugars, including pentoses such as xylose and arabinose, in the presence of glucose (Abe, Higuchi and Yamato 1998).

Amino acid conversion coupled with energy generation

Discovery of l-aspartate decarboxylation in soy sauce fermentation

Nutrient transport by bacteria is usually considered to consume metabolic energy, as this step is typically driven by an ion-motive gradient (e.g. protons or sodium ions) or by hydrolysis of a phosphoester bond (e.g. in ATP or PEP) (Harold 1986; Nanatani and Abe 2011). Since energy generation is often restricted by metabolic control in microbial fermentation, I searched for a metabolic system that can produce substances with high efficiency while generating but not consuming energy in the process of transporting substrates and exporting metabolic products. A new class of nutrient transport reactions was identified (Nanatani and Abe 2011), in which substrate transport generates rather than consumes energy. As a result, we discovered an l-aspartate¹⁻ (l-Asp): l-alanine⁰ (l-Ala) exchange transporter (AspT) in lactic acid bacteria such as Lactobacillus subsp. M3 (Abe, Hayashi and Maloney 1996) and T. halophilus (Abe et al. 2002), analyzed it biochemically, and developed an industrial application.

Lactobacillus subsp. M3 was initially isolated as a contaminant from low-salt (<8% wt/vol) soy sauce in which CO₂ is formed from l-Asp decarboxylation, and we studied the biochemical properties of this decarboxylation (Abe, Hayashi and Maloney 1996). The growth of Lactobacillus subsp. M3 approximately doubles in the presence of l-Asp in MRS-glucose medium. Lactobacillus subsp. M3 catalyzes l-Asp decarboxylation with a nearly stoichiometric release of Ala and CO₂. Because the β-carboxyl group in the side chain of l-Asp is simply eliminated by decarboxylation, it is an intriguing question why the growth of Lactobacillus subsp. M3 is promoted by such a simple reaction. The increased growth implies energy generation coupled with decarboxylation. We analyzed the intracellular level of ATP in intact Lactobacillus subsp. M3 cells catalyzing l-Asp decarboxylation (Abe, Hayashi and Maloney 1996) and detected its increase during l-Asp decarboxylation. ATP can be synthesized either via substrate-level phosphorylation, which produces ATP directly by metabolic processes, or by proton-motive force (pmf)-driven F-type ATPases. We examined the effects of a protonophore CCCP, which collapses pfm across membranes, and an F_oF₁-ATPase inhibitor DCCD on ATP generation coupled with l-Asp decarboxylation. Both compounds inhibited ATP production but not l-Asp decarboxylation in intact cells of Lactobacillus subsp. M3, suggesting that l-Asp: l-Ala conversion generates a pmf that subsequently drives F-ATPase for ATP synthesis. Since l-Asp decarboxylation but not ATP production was observed in cell-free extracts of Lactobacillus subsp. M3, ATP synthesis requires membranes. Furthermore, high extracellular concentration of l-Ala (>300 mm) inhibits both l-Asp: l-Ala conversion and ATP generation coupled with decarboxylation, suggesting that such high levels of extracellular l-Ala inhibit the generation of a pmf that is required for ATP synthesis by F-ATPase.

We also found that some strains of T. halophilus show l-Asp: l-Ala conversion with Asp decarboxylation, and this property depends on some plasmids in which the asp operon consists of the aspD for Asp decarboxylase and aspT for l-Asp: l-Ala antiporter (in this order) (Abe et al. 2002). Because l-Asp is sour and l-Ala is weakly sweet, we industrially applied l-Asp: l-Ala conversion catalyzed by T. halophilus to soy sauce fermentation and made the taste of soy sauce milder (Higuchi, Uchida and Abe 1998).

Biochemical properties of energy generation by l-aspartate decarboxylation

To characterize biochemically the Asp: Ala conversion by lactic acid bacteria such as Lactobacillus subsp. M3 and T. halophilus, we reconstituted the l-Asp: l-Ala antiporter AspT in proteoliposomes (Abe, Hayashi and Maloney 1996; Abe et al. 2002). To determine whether a transporter catalyzes import or export of a solute across the membrane, a high concentration of non-labeled substrate (10-200 mm) is loaded into proteoliposomes and a low concentration of radioactive substrate is added to the outside of the proteoliposomes. The radioactive solute is transported by self-exchange via the transporter, so-called “counter flow,” and this import soon reaches steady state. At this phase, addition of non-radioactive solute (e.g. 10-50 mm) to the outside of the proteoliposomes induces export of the imported radioactive solute, so-called “exchange reaction.” Thus, counter flow and exchange reaction show the presence of an enzyme that catalyzes solute transport. Proteoliposomes containing membrane proteins of Lactobacillus subsp. M3 imported radioactive l-Asp until reaching steady state (counter flow), and subsequent addition of a high concentration of nonradioactive l-Asp induced a rapid efflux of radioactive l-Asp from proteoliposomes (exchange reaction). Liposomes without membrane proteins showed no influx of radioactive l-Asp.

Addition of nonradioactive l-Ala to the outside of proteoliposomes led to a rapid efflux of imported radioactive l-Asp (Abe, Hayashi and Maloney 1996), that is, Asp: Ala exchange occurred in proteoliposomes containing reconstituted membrane fraction from Lactobacillus subsp. M3. In other words, it seemed likely that an AspT carrier protein exchanged Asp and Ala. Since the conversion of Asp to Ala is nearly stoichiometric in Lactobacillus subsp. M3, one would expect the Asp: Ala exchange ratio to be a one-for-one antiport and the exchange to be electrogenic, since at a near-physiological pH (pH 7) the net charge on l-Asp is −1 [pK1(COOH) = 2.09, pK2(NH₃) = 9.82, pK3(COOH) = 3.86], while that on l-Ala is 0 [pK1(COOH) = 2.34, pK2(NH₃) = 9.69]. This prediction was examined in studies of l-Asp transport in l-Ala-loaded proteoliposomes, using valinomycin along with external or internal potassium to generate a membrane potential that should accelerate or retard such an electrogenic reaction (Abe, Hayashi and Maloney 1996) The l-Asp: l-Ala exchange was promoted by the inside positive membrane potential and was strongly inhibited by the inside negative membrane potential, suggesting that Asp: Ala exchange catalyzed by AspT is electrogenic.

The electrogenic nature of l-Asp¹⁻: l-Ala⁰ exchange was also confirmed with purified T. halophilus AspT reconstituted in proteoliposomes (Figure 2a) (Abe, Hayashi and Maloney 1996; Sasahara et al. 2011). A diagram of the mechanism of proton-motive metabolic cycle exemplified by l-Asp: l-Ala exchange is shown in Figure 2b. Continuous electrogenic l-Asp: l-Ala exchange by AspT generates an internally electrically negative membrane potential in cells of Lactobacillus subsp. M3 or T. halophilus. In addition, intracellular decarboxylation of l-Asp to l-Ala consumes cytoplasmic scalar protons and thus generates a pH gradient of physiological polarity across the membrane through l-Asp: l-Ala exchange. Overall, the membrane potential (inside negative) and the pH gradient (inside alkaline) form a pmf across the membrane. If the electrical polarization and pH gradient become larger, the over-polarized membrane with a pmf would retard l-Asp: l-Ala exchange even if the cytoplasmic l-Ala concentration is sufficiently high. However, in bacterial cells, F-type ATPase in the membrane catalyzes ATP synthesis by consuming the generated pmf. Therefore, l-Asp: l-Ala exchange coupled with cytoplasmic l-Asp decarboxylation is continuously maintained until l-Asp in the medium is depleted. When T. halophilus harboring plasmids encoding AspT and AspD is grown in soy sauce moromi mash (see the section “Catabolite repression of xylose operon in lactic acid bacterium T. halophilus in fermentation”, which generally contains 30-100 mml-Asp, most of the l-Asp is converted to l-Ala. Therefore, this conversion generates a pmf required for both ATP biosynthesis and tolerance to low pH caused by lactic acid production. Lactic acid bacteria including T. halophilus obtain metabolic energy mainly by glycolysis, but under unfavorable growth conditions such as low pH, they also use energy generation systems driven by decarboxylation of carboxylic acids, in particular, amino acids(Nanatani and Abe 2011).

Membrane topology of T. halophilus AspT

We biochemically analyzed the trans-membrane (TM) regions of T. halophilus AspT by using a cysteine-scanning method. This method combines site-directed mutagenesis and site-specific chemical modification of the thiols of cysteine residues and has been applied to analyze the membrane topology of solute transporters and receptors (Nanatani et al. 2007). A membrane-impermeable SH-group–modifying reagent such as Oregon Green 488 maleimide (OGM) selectively reacts with SH of cysteine residues in a hydrophilic environment. All cysteine residues in the target protein are replaced with alanine by site-directed mutagenesis, and many mutants with single-cysteine replacements are generated and expressed in Escherichia coli. To determine the periplasmic cysteines, E. coli cells expressing a single-cysteine mutant are exposed to OGM. To determine cytoplasmic cysteines, periplasmic cysteines are blocked with the non-fluorescent SH-reagent 2-(trimethylammonium)ethyl methanethiosulfonate bromide, cells are disrupted and OGM is added. Because each single cysteine mutant has a (6 × His)-affinity tag, the target protein (in our studies, AspT) is solubilized with a detergent, its OGM-modified form is purified by using Ni-affinity columns, electrophoresed by SDS-PAGE, and fluorescence in the gel is detected and used to determine whether the single cysteine residue is located inside or outside the cell (see details in references). Such cysteine scanning revealed that T. halophilus AspT has N- and C-termini on the periplasmic side, 10 TM α-helices, and a hydrophilic loop region of about 180 amino acids between TM5 and TM6. It also revealed that AspT has a unique molecular structure (5 TM—large loop—5 TM). The CDART tool predicted that this is a common structure of AAEx family transporters (Geer et al. 2002).

The TM3 of AspT contains many residues that are conserved in transporters of the AAEx family. A conserved positively charged arginine residue is located in this hydrophobic α-helix. Therefore, we hypothesized that TM3 might be involved in substrate transport and analyzed TM3 using the cysteine-scanning method. In the absence of substrate, Gly62 is exposed to the periplasmic surface, and Pro79 is located at the cytoplasmic end of the hydrophobic core. The substrate binding induces a structural conformation change, and those residues (Gly62 and Pro79) move to the periplasmic end of the hydrophobic core and the cytoplasmic water-filled cavity, respectively (Nanatani, Maloney and Abe 2009).

The characteristic structure of T. halophilus AspT determined by the cysteine-scanning method is a hydrophilic loop of approximately 180 amino acid residues located between TM5 and TM6 (Nanatani et al. 2007). This hydrophilic central loop contains two structures called the TrkA_C domain. The TrkA_C domain was first discovered in TrkA, a protein responsible for regulating K⁺ transport (Anantharaman, Koonin and Aravind 2001), but its function has not been clarified. It is presumed that TrkA_C domain binds an unspecified ligand and controls transport. To obtain detailed structural information, we used a protein structure prediction program, Full Automatic Modeling System (FAMS), to model the three-dimensional structure of this loop. The modeling revealed two α–β complex structures characteristic for TrkA_C domains within the loop (Nanatani et al. 2007).

Substrate recognition by T. halophilus AspT

Detailed kinetic analysis and studies on substrate specificity of membrane transporters have been hampered by the difficulties in the purification of membrane proteins. Transport activity of AspT for l-Asp and l-Ala was demonstrated in proteoliposomes reconstituted from crude membrane fractions (Abe, Hayashi and Maloney 1996; Abe et al. 2002). However, the spectrum of amino and organic acids used by AspT as substrates remained unknown. To clarify this issue, we solubilized T. halophilus AspT in the detergent n-dodecyl-α-d-maltoside and purified it. To examine the inhibitory effect of each amino acid on l-Asp or l-Ala uptake catalyzed by AspT, we reconstituted it into proteoliposomes and used each d- or l- amino acid (15 mm) as a competitor in the external solution. An inhibitory effect of the amino acid used suggested that it is an AspT substrate. We found that an amino group at the α or β position is required for a compound to be an AspT substrate, and the polarity at the β position may be involved in the substrate binding to the Asp-binding site. The uptake of l-Asp was specifically inhibited by l-Cys, and that of l-Ala by l-Ser. l-Asp analogs such as l-cysteine sulfinic acid and l-cysteic acid specifically inhibited l-Asp transport but not l-Ala transport. This indicates that AspT has independent binding sites for l-Ala and l-Asp, and that l-Cys and l-Asp analogs bind to l-Asp-binding sites (Sasahara et al. 2011).

This model of independent binding sites for l-Ala and l-Asp in AspT was modified (Suzuki et al. 2022). In an OGM labeling assay, the labeling efficiency of AspT with single Cys mutations (G62C and P79C) in TM helix 3 depended on the presence of l-Ala or analogs. We observed a concentration-dependent shift of AspT conformation specific to one substrate toward that specific to the subsequently added substrate (l-Ala or l-Asp) (Suzuki et al. 2022). Size-exclusion-chromatography-based thermostability assays indicated different thermal stability of AspT in the presence of l-Ala and l-Asp (Kunii et al. 2024). We concluded that l-Ala binding yields an AspT conformation different from the apo or l-Asp-bound conformations (Figure 3) (Suzuki et al. 2022).

We also discovered a glutamate: GABA transporter with a similar reaction mechanism in Lactobacillus spp. (Higuchi, Hayashi and Abe 1997). We also isolated the oxalate: formate transporter (OxlT) gene and biochemically analyzed OxlT transport (Abe, Ruan and Maloney 1996).

Hydrophobins

Hydrophobins and their biological function

Hydrophobins are small (<20 kDa) amphipathic proteins present in many filamentous fungi. Amino acid sequence similarity among hydrophobins is low, but they have eight conserved Cys residues with a specific number of amino acid residues between them (C-X_5–7C-C-X_19–39-C-X_8–23-C-X₅-C-C-X_6–18-C-X_2–13 or C-X_9–10C-C-X₁₁-C-X₁₆-C-X_8–9-C-C-X₁₀-C-X_6–7) and four disulfide bonds (Kwan et al. 2006). All hydrophobins are similar to each other in that they have β-barrel structures (Kwan et al. 2006; Pille et al. 2015). They are classified mainly into classes I or II according to their hydropathy profiles, primary structures, and self-assembled membrane solubility (Wosten 2001; Tanaka et al. 2022). Some filamentous fungi have at least 10 hydrophobin-encoding genes, for example, 10 in each Aspergillus nidulans and Aspergillus niger, or 16 in Trichoderma atroviride, whereas many filamentous fungi have only a few such genes (Littlejohn, Hooleyb and Cox 2012; Kubicek et al. 2019). The expression profiles of many hydrophobin genes depend on fungal growth stage and culture conditions, and cellular location varies among hydrophobins (Valsecchi et al. 2017; Cai et al. 2020).

Filamentous fungi secrete hydrophobins, which self-assemble and form amphipathic membranes at solid-liquid or air-liquid interfaces (Terauchi et al. 2020; Tanaka et al. 2022). Hydrophobins play a role in the formation of aerial hyphae and conidia because their amphipathic membranes decrease interfacial surface tension (Wosten, Schuren and Wessels 1994; Ball, Kwan and Sunde 2020). Hydrophobins accumulate within aerial hyphae, where they bind to lipid-enriched organelles and may affect their hyphal structure and increase hyphal longevity (Cai et al. 2021). Secretion of hydrophobins is highest at the sporulation phase, when they form a coating that protects conidia. In conidia, hydrophobins are involved in water sensing and germination (Cai et al. 2021). Hydrophobin coating makes the surfaces of aerial structures hydrophobic, which contributes to both conidial dispersal (Cai et al. 2020) and, in the case of pathogenic filamentous fungi, to pathogen adsorption on host insects or plants, as their surfaces are hydrophobic (St Leger, Staples and Roberts 1992; Talbot, Ebbole and Hamer 1993; Tanaka et al. 2017). Because the immune systems of animals (e.g. insects, mammals) and plants fail to recognize hydrophobin-coated hyphae and conidia, hydrophobins are thought to contribute to the ability of pathogenic filamentous fungi to infect their hosts (Aimanianda et al. 2009; Cai et al. 2020; Tanaka et al. 2022). Hydrophobins on solid surfaces can recruit and immobilize various proteins such as bovine serum albumin, IgG, avidin, glucose oxidase, horseradish peroxidase, and cutinases (Corvis et al. 2005; Takahashi et al. 2005; Wang et al. 2010; Pham et al. 2016; Tanaka et al. 2017; Tanaka et al. 2022). The polyphyletic genus Aspergillus in the Ascomycota includes a number of anamorphic fungi (Levetin et al. 2016). Most aspergilli are highly capable of decomposing solid polymers owing to the properties of hydrophobins (Tanaka et al. 2022), and thus have been widely used in the fermentation industry for a long time (Machida, Yamada and Gomi 2008).

Hydrophobins in solid polymer degradation

In 2005, we found that hydrophobin of A. oryzae was specifically induced when the only carbon source was polybutylene succinate co-adipate (PBSA) (Takahashi et al. 2005). The expression of hydrophobin genes is induced in other filamentous fungi such as A. nidulans (Brown et al. 2016), A. niger (Delmas et al. 2012; Pullan et al. 2014), and Trichoderma reesei (Nakari-Setälä et al. 1997) when they are grown on solid polymers of plant origin such as cellulose (Delmas et al. 2012; Pullan et al. 2014) or xylan (Nakari-Setälä et al. 1997), straw (Nakari-Setälä et al. 1997; Delmas et al. 2012), or steam-exploded sugarcane bagasse (Brown et al. 2016), suggesting the involvement of these hydrophobins in the degradation of solid polymers by filamentous fungi. Thus, studying Aspergillus hydrophobins expands our understanding of degradation and utilization of solid polymers, as well as of infection of animals (Thau et al. 1994; Stringer and Timberlake 1995; Aimanianda et al. 2009) or plants (Dolezal et al. 2014; Cohen et al. 2021) by aspergilli. In this section, I focus on the mechanism of enzyme recruitment by hydrophobins of aspergilli attached to solid polymers.

Hydrophobin-cutinase interaction in A. oryzae, A. nidulans, and other fungi

Aspergillus oryzae

In 2005, we reported the first direct evidence for the involvement of hydrophobin in the degradation of solid polymers in A. oryzae (Maeda et al. 2005; Takahashi et al. 2005). On PBSA as a sole carbon source, this fungus coexpresses hydrophobin RolA and cutinase CutL1 and hydrolyzes PBSA (Maeda et al. 2005; Takahashi et al. 2005). The secreted RolA adsorbs to the PBSA surface (Tanaka et al. 2014; Takahashi et al. 2015), and then recruits and condenses CutL1 and a CutL1 homolog, CutC, thereby promoting PBSA hydrolysis (Takahashi et al. 2005; Takahashi et al. 2015) (Figure 4). Cutinases hydrolyze aliphatic esters such as triglycerides, cutin, and PBSA (Maeda et al. 2005). Cutinases are produced by many fungi and bacteria (Tanaka et al. 2022). PBSA is structurally similar to cutin, an insoluble wax polyester in the plant protective cuticle (Fang et al. 2001). PBSA degradation via RolA–CutL1 interaction is thought to mimic infection by pathogenic filamentous fungi (Takahashi et al. 2005; Takahashi et al. 2015) because many of them produce both hydrophobin and cutinase, which promote infection (Talbot et al. 1996; Tanaka et al.  2022).

Several key characteristics of the RolA–CutL1/CutC interaction have been clarified (Takahashi et al. 2015; Terauchi et al. 2017):

(i) RolA adsorbs to the PBSA surface before CutL1 reaches the surface and CutL1 recruitment by adsorbed RolA largely promotes PBSA degradation (Takahashi et al. 2005). The PBSA degradation is only slightly accelerated by simultaneous addition of RolA and CutL1 in comparison with that induced by CutL1 alone. RolA secondary structure changes after its adsorption to a solid surface, and this change is necessary for CutL1 recruitment (Takahashi et al. 2005).

(ii) The adsorbed RolA moves laterally on the PBSA surface but stops in the presence of CutL1 (Takahashi et al. 2005). Therefore, RolA may act as an anchor or scaffold to tether CutL1. The random movement of RolA molecules that do not interact with CutL1 exposes the PBSA surface to CutL1 recruited by other RolA molecules.

(iii) The recruitment of CutL1 by RolA on solid surfaces is driven by ionic interactions between these proteins (Figure 4b) (Takahashi et al. 2015) and is affected by the protonation state of the side chains of amino acid residues in both proteins, which depends on pH; NaCl prevents these ionic interactions.

(iv) Positively charged His32 and Lys34 of RolA and negatively charged Asp30, Glu31, Asp142, and Asp171 on the hydrophilic surface of CutL1 are critical for RolA-dependent CutL1 recruitment (Figures 4b and 5) (Takahashi et al. 2015) (Terauchi et al. 2017). Chemical modification of these residues or their substitution with noncharged residues such as Ser markedly weaken the RolA–CutL1 interaction (Figure 5). Nevertheless, the interactions between the RolA-H32S/K34S and CutL1-E31S/D142S/D171S mutants, and between wild-type RolA and CutL1-D30S/E31S/D142S/D171S are still stronger than the interaction between the wild-type proteins in the presence of 250 mm NaCl (Figure 5) (Terauchi et al. 2017). Therefore, other charged residues (e.g. Lys41, Lys46, and Lys51 of RolA) or complementarity of the three-dimensional structures of RolA and CutL1 may be involved in the interaction. It cannot be excluded that the properties of cutinases such as substrate specificity and thermal stability may change due to their interaction with RolA, but this remains to be studied.

RolA also promotes the degradation of polyethylene terephthalate (PET) by PET-degrading enzyme (Puspitasari, Tsai and Lee 2021), or PETase (Yoshida et al. 2016), from the betaproteobacterium Ideonella sakaiensis; both PETase and cutinases are alpha/beta-hydrolases. PETase (27.6 kDa) is larger than CutL1 (19.7 kDa), and their amino acid sequence identity is very low (19.24%). However, their three-dimensional structures (PETase: protein databank ID 5XJH, Joo et al. 2018; CutL1: protein databank ID 3GBS, Liu et al. 2009) are similar, and some of the negatively charged residues in both proteins are located on the opposite side of the active site. Therefore, the mechanisms of the RolA–PETase and RolA–CutL1 interactions may be similar and RolA may interact with and recruit various cutinases and cutinase-like enzymes and thus enhance the hydrolysis of various aliphatic esters by these enzymes.

Aspergillus nidulans

The model fungus A. nidulans has multiple genes encoding both hydrophobins and cutinases (Grunbacher et al. 2014; Tanaka et al. 2022). Hydrophobin AnRodA interacts with cutinases Cut1 and Cut2, promoting PBSA degradation (Tanaka et al. 2017); they are the orthologs of A. oryzae RolA, CutL1, and CutB, respectively (St Leger, Staples and Roberts 1992; Takahashi et al. 2015). Expression of the cut1 and cut2 genes is induced by lipidic carbon sources such as suberin (Martins et al. 2014), cutin (Bermúdez-García et al. 2019) or olive oil (Castro-Ochoa et al. 2012;). Expression of the AnrodA gene is induced by steam-exploded sugarcane bagasse (Brown et al. 2016), which is composed of cellulose, hemicellulose, lignin, and wax ester (Rezende et al. 2011; Qi et al. 2017). In such a culture, the activity of extracellular polysaccharide-hydrolyzing enzymes (e.g. cellulases or amylases) and the fungal biomass of the ΔAnrodA strain are lower than those of the wild-type strain (Brown et al. 2016). Therefore, AnRodA may promote the degradation of not only aliphatic esters but also polysaccharides. It interacts with Cut1 and Cut2 via ionic interactions in the same way as RolA interacts with CutL1 and CutC (Tanaka et al. 2017). Interestingly, AnRodA also interacts with CutL1 of A. oryzae via ionic interactions, although the interaction is much weaker than that between RolA and CutL1 (Takahashi et al. 2015; Tanaka et al. 2017). Positively charged residues AnRodA (His23, Lys35, and Lys41) are more widely spaced in the primary structure than those of RolA (His32, Lys34, and Lys41) (Takahashi et al. 2015). Thus, charged amino acid residues may be in more suitable positions in the RolA–CutL1/CutC and AnRodA–Cut1/Cut2 interactions (St Leger, Staples and Roberts 1992; Takahashi et al. 2005; Tanaka et al. 2017) than in the AnRodA–CutL1 interaction.

Other fungi

In the rice blast fungus Magnaporthe oryzae, hydrophobins MPG1 (class I) and MHP1 (class II) interact with the cutinase Cut2 (Pham et al. 2016). Using phylogenetic analysis, we detected a variety of potential combinations of hydrophobins and cutinases (Takahashi et al. 2015), such as hydrophobin Pc22g14290 (accession number CAP98717.1)—Cutinase 1 (CAP97019.1) in Penicillium chrysogenum, hydrophobin BCDW1_9126 (EMR82223.1)—cutinase BCDW1_3897 (EMR87444.1) in Botrytis cinerea, and hydrophobin FVG_03 685 (EWG41603.1)—Cutinase 3 (EWG55667.1) in Fusarium verticillioides (all accession numbers are from GenBank).

Some class I hydrophobins are predicted from coding sequences only. In many of them, including those of ascomycetes, multiple positively charged residues are located in the N-terminal regions upstream of the first Cys residue (Takahashi et al. 2015). Most predicted hydrophobins in the clade containing AnRodA and RolA are from Aspergillus and Penicillium species (Takahashi et al. 2015) and have at least three such residues in similar positions (Figure 4). In other clades, multiple positively charged N-terminal residues were detected, for instance, in Hydpt1 of Pisolithus tinctorius (GenBank accession number AAC49307.1; 10 residues), SC6 of Schizophyllum commune (CAA07545.1; 4 residues), and Vmh1 of Pleurotus ostreatus (CAB41405.1; 4 residues) (Tagu, Nasse and Martin 1996). Most ascomycetous and basidiomycetous filamentous fungi that have hydrophobins also have several cutinases, including those predicted from coding sequences only. Some filamentous fungi also have acetylxylan esterases of the carbohydrate esterase 5 family, whose amino acid sequences are highly similar to those of cutinases. Negatively charged residues corresponding to Glu31, Asp142, and Asp171 of CutL1 are conserved in many cutinases of ascomycetes, in some cutinases of basidiomycetes, and in some acetylxylan esterases (Takahashi et al. 2015). Hydrophobin–cutinase ionic interactions may be common in Aspergillus and Penicillium species, and possibly in many ascomycetes and in some basidiomycetes (Takahashi et al. 2015; Tanaka et al. 2017; Terauchi et al. 2017).

Recently, we established an approach to visualize and analyze the function of RolA at interfaces by producing self-assembled monolayers of RolA on hydrophobic and hydrophilic silicon substrates by using the Langmuir-Blodgett method (Terauchi et al. 2020) and transferring these monolayers from droplet surface (Takahashi et al. 2023). We observed the self-assembled monolayer structure on the silicon substrate under an atomic force microscope (Terauchi et al. 2022; Takahashi et al. 2023). Studies on visualization of the RolA membrane–CutL1 interaction are ongoing.

Cell surface and cell wall of filamentous fungi

Response to environmental stress and signaling in Aspergillus fungi

Aspergillus genus contains some of the most important filamentous fungi from the viewpoints of industry, pathogenesis, and mycotoxin production. To understand fungal biology and to make better use of fungi in fermentation and enzyme production, stress adaptation mechanisms in Aspergillus species have been studied intensively (Abe et al. 2010) (Figure 6). A signaling pathway responsible for adaptation to environmental osmotic changes includes an upstream two-component phosphorelay and a downstream HogA/SakA mitogen-activated protein kinase (MAPK) cascade (Hagiwara et al. 2016). In the cell wall integrity (CWI) signaling pathway, the MpkA MAPK cascade regulates α-1,3-glucan synthase genes (ags) and therefore cell wall composition (Fujioka et al. 2007). We focused on these two signaling pathways to elucidate the functions of cell surface sensors and downstream signaling in filamentous fungi and to develop antifungal agents (Figure 6) (Hagiwara et al. 2016; Yoshimi, Miyazawa and Abe 2016; Miyazawa and Abe 2017; Yoshimi et al. 2022). In this section, I describe our findings on the Aspergillus CWI signaling pathway and their applications to industrial fermentation.

Two-component systems and HogA/SakA MAPK cascade

His-Asp phosphorelay (or two-component) systems are implicated in cellular responses to environmental stimuli in prokaryotes, fungi, and plants (Hagiwara et al. 2015; Singh et al. 2021). Most of eukaryotic His-Asp phosphorelay systems consist of a His kinase (HK), a His-containing phosphotransfer intermediate (HPt), and a response regulator (Appleby, Parkinson and Bourret 1996). Phosphorelays among these components mediate signaling that confers adaptation to stimuli. MAPK cascades are essential for adaptation to the environment in eukaryotic cells and are highly evolutionarily conserved; they transduce signals to the nucleus to adjust transcriptional responses (Hohmann 2002; Hagiwara et al. 2016). His-Asp phosphorelay systems and MAPK signaling pathways have been identified and characterized in many eukaryotes (Mizuno 2005; Hagiwara et al. 2016).

In aspergilli, eight HK groups (HK1–HK8) are recognized (Hagiwara et al. 2015; Hagiwara et al. 2016). HK1 is an ortholog of S. cerevisiae Sln1p, a membrane-bound osmotic stress sensor that transmits signals through the high-osmolarity glycerol response 1 (HOG1) MAPK cascade in yeast cells. We cloned and characterized a novel A. nidulans HK gene, tcsB, encoding a membrane-type HK homologous to Sln1p (Furukawa et al. 2002). In an A. nidulans cDNA library, we identified a 3210-bp open reading frame that encoded a putative protein of 1070 amino acids. The predicted TcsB was structurally similar to Sln1p and had two TM regions in its N-terminal half. Overexpression of the tcsB cDNA suppressed the lethality of a temperature-sensitive osmosensing-defective yeast sln1-ts mutant, whereas tcsB cDNAs mutated at His552 (conserved phosphorylation site) or at Asp989 (the phosphorelay site) failed to complement this yeast mutant. Introduction of the tcsB cDNA into a yeast sln1_ sho1_ double mutant, which lacked two osmosensors, suppressed lethality in high-salinity media and activated HOG1 MAPK. These data imply that TcsB functions as an osmosensor HK. We constructed an A. nidulans strain lacking the tcsB gene, but unexpectedly found no detectable phenotype in hyphal development or morphology on standard or stress media. Our data suggest that A. nidulans has more complex and robust osmoregulatory systems than the yeast SLN1-HOG1 MAPK cascade (Furukawa et al. 2005).

Cell wall and CWI signaling

Overview of cell wall structure

Fungal cell walls maintain cell morphology and protect the cells from environmental stresses such as osmotic pressure, temperature, and pH (Yoshimi, Miyazawa and Abe 2016, Miyazawa and Abe 2017; Yoshimi et al. 2022). Cell walls are the first line of defense against environmental stresses, and fungi rebuild and repair them. The cell walls of filamentous fungi are composed mainly of polysaccharides (Figure 7) (Yoshimi, Miyazawa and Abe 2016; Yoshimi et al. 2017). The main cell wall polysaccharides are α-glucans (α-1,3-glucan interconnected by short α-1,4-linked glucose spacers), β-glucans (β-1,3-glucan with β-1,6-branches), galactomannan, and chitin (Yoshimi et al. 2022; Miyazawa et al. 2024). Some fungi also have an extracellular matrix (ECM) composed of polysaccharides synthesized de novo in the outer layers of the cell wall (Gastebois et al. 2009; Beauvais et al. 2014). Galactomannoproteins, some glycosylphosphatidylinositol (GPI)-anchored proteins, and other proteins are also found in the cell wall (Gastebois et al. 2009; Latge 2023). In this section, I review recent findings on the structure, biosynthesis, and biological functions of the polysaccharides of the Aspergillus cell walls.

Cell wall polysaccharides of Aspergillus species can be fractionated by alkali solubility (Yoshimi et al. 2013). The alkali-soluble fraction contains mainly α-1,3-glucan with interconnecting α-1,4-linkages and some galactomannan, and the alkali-insoluble fraction contains chitin, β-1,6-branched β-1,3-glucan, and galactomannan (Yoshimi et al. 2013). Galactomannan is composed of α-core-mannan chains, in which 9 or 10 sets of α-1,2-mannotetraose units are concatenated by α-1,6-linkages, with β-galactofuran side chains containing several β-1,5-galactofuranotetraose units interconnected by β-1,6-linkages. Galactomannans are thought to be synthesized in the lumen of the Golgi apparatus. The biosynthesis and structure of galactomannan are reviewed in Oka et al. (2004). Galactomannan, β-1,6-branched β-1,3-glucan, and chitin are connected via their reducing ends to the non-reducing ends of other polysaccharides organized in a three-dimensional network (Gastebois et al. 2009; Yoshimi et al. 2022; Latge 2023). Some GPI-anchored proteins have transglycosylation activities and are essential for cell wall polysaccharide remodeling (Samalova et al. 2020). A large subset of GPI-anchored proteins is transferred from the membrane to the cell wall in fungi (Vogt et al. 2020). The Aspergillus ECM is primarily composed of secreted proteins and polysaccharides and mediates adherence to inorganic substrates and host cells (Sheppard and Howell 2016). Many secreted proteins are modified with O-glycans that bind to Ser or Thr residues. O-glycans of Aspergillus species are O-linked mannose units containing oligo-β-1,5-galactofuranose (Kudoh, Okawa and Shibata 2015). The ECM contains α-1,3-glucan, galactomannan, and GAG. The latter is composed of linear α-1,4-linked galactopyranose and N-acetylgalactosamine, and is known as the extracellular secreted polysaccharide of Aspergillus species (Fontaine et al. 2011). Partial deacetylation of N-acetylgalactosamine residues of GAG to galactosamine is essential for GAG adhesiveness (Lee et al. 2016). In A. fumigatus, solid-state nuclear magnetic resonance analysis revealed the supramolecular assembly of biopolymers in the cell wall of hyphae (Chakraborty et al. 2021) and the dynamic cell wall remodeling during transition from conidia to hyphae (Lamon et al. 2023). The cell surface structure and biosynthesis of cell wall polysaccharides of Aspergillus species are reviewed in the articles (Yoshimi, Miyazawa and Abe 2016; Lee and Sheppard 2016).

CWI signaling

The CWI signaling pathway senses cell wall abnormalities and regulates gene responses to maintain the cell wall; it has been studied in detail in the budding yeast S. cerevisiae (Levin 2011). The genome sequences of Aspergillus spp. are available (Galagan et al. 2005; Machida et al. 2005; Nierman et al. 2005; Pel et al. 2007), and the genes encoding homologs of the components of the yeast CWI pathway have been identified in Aspergillus spp. (Figure 6) (Fujioka et al. 2007; Kobayashi et al. 2007).

Aspergillus species have homologs of S. cerevisiae Wsc1-3 and mid-type cell wall sensors, upstream components of the CWI signaling pathway; their functions are reviewed in detail in (Yoshimi et al. 2022). On the basis of (i) the comparative genomics of S. cerevisiae and A. nidulans and (ii) functional analyses of mpkA encoding the MpkA MAPK and rlmA encoding the transcription factor RlmA activated by MpkA in A. nidulans, we suggested the following CWI signaling components in A. nidulans (Fujioka et al. 2007). (i) Putative sensor proteins WscA, WscB, and MtlA are important under hypo-osmotic conditions, but WscA and WscB are not essential for MpkA–RlmA signaling. (ii) PkcA is involved in the CWI pathway. PkcA also plays a role in suppressing apoptosis induced via the MpkA pathway, but not in polarity establishment, during hyphal growth independent of the MpkA pathway under heat stress. (iii) Expression of the α-1,3-glucan biogenesis-related genes agsA and agsB depends on MpkA and partly on RlmA. (iv) Other CWI-related genes, such as β-1,3-glucan biogenesis-related genes (fksA, gelA, and gelB) and chitin biogenesis-related genes (chsA, chsB, chsC, chsD, csmA, and csmB) are independent of or partly dependent on the MpkA-RlmA axis. The CWI pathway mainly regulates transcription of α-1,3-glucan biogenesis–related genes. Transcription of β-1,3-glucan- and chitin biogenesis–related genes is mainly regulated by unknown signals that might be activated by cell wall stress (Chen et al. 2024), such as echinocandin (micafungin) treatment. To our knowledge, whereas CWI signaling from the sensor proteins to MpkA-RlmA is well conserved among Aspergillus species, TFs activated by MpkA other than RlmA remain unknown in aspergilli and other filamentous fungi (Fujioka et al. 2007).

Cell wall components and hyphal pellet formation

Hyphal aggregation factors and hyphal dispersion strains in Aspergillus spp.

α-1,3-glucan

Filamentous fungi have dispersed morphology or their hyphae aggregate (Antecka, Bizukojc and Ledakowicz 2016; Veiter, Rajamanickam and Herwig 2018). The mechanisms of hyphal pellet formation are reviewed in detail by (Zhang and Zhang 2016). In industrial production, growing filamentous fungi in liquid culture facilitates the addition and mixing of inducer substrates and nutrients. Aspergillus species form pellets in shake-flask cultures.

As described in the section “CWI signaling”, we compared the CWI pathway of A. nidulans with that of S. cerevisiae and showed that the agsA and agsB genes are its downstream targets in A. nidulans (Fujioka et al. 2007). The cell wall polysaccharide α-1,3-glucan is absent in S. cerevisiae and was considered a potential antifungal target. We generated two ags gene disruptants (ΔagsA and ΔagsB) in A. nidulans and found a non-lethal α-1,3-glucan deficiency in cell walls of the ΔagsB strain (Yoshimi et al. 2013); therefore, α-1,3-glucan biosynthesis is not a potential antifungal target. Interestingly, the mycelia of these strains were dispersed in liquid shake cultures (Yoshimi et al. 2013). The aggregation of A. fumigatus germinating conidia is eliminated by α-1,3-glucanase treatment (Fontaine et al. 2010) and is decreased in a triple α-1,3-glucan synthase gene disruptant, which lacked α-1,3-glucan in the cell wall (Henry, Latge and Beauvais 2012). We generated an α-1,3-glucan-deficient strain of A. oryzae for industrial applications and found that the pellet diameter decreased, but hyphae did not disperse (Miyazawa et al. 2016). As in A. nidulans, agsB primarily functions in α-1,3-glucan synthesis in A. oryzae (Zhang et al. 2017). Mycelial cell diameter decreased in A. luchuensis upon agsE gene disruption (Tokashiki et al. 2019) and in A. fumigatus α-1,3-glucan-deficient strains (Miyazawa et al. 2022). The above observations indicate that α-1,3-glucan functions as the primary hyphal aggregation factor in Aspergillus species.

As mentioned above, we reported that AgsB is important for α-1,3-glucan biosynthesis in A. nidulans (Yoshimi et al. 2013). To clarify its mechanism, we focused on the two α-1,3-glucan synthases in A. nidulans. In wild-type strains, agsB is mainly expressed in hyphae, but agsA expression is low except that it is reportedly expressed during conidiation (He, Li and Kaminskyj 2014). We constructed strains that overexpressed either agsA or agsB under the control of a constitutive promoter in the genetic background of agsB or agsA disruptants, respectively. In liquid culture, the wild-type and agsB-overexpression strains formed tightly aggregated hyphal pellets, whereas hyphae of the agsA-overexpression strain aggregated weakly. Methylation and NMR analyses revealed α-1,3-linked glucose as the main structure of glucan in the alkali-soluble fraction of both overexpression strains. The molecular weight of α-1,3-glucan was 1500  × 10³ in the agsA overexpression strain, but 370  × 10³ in the agsB overexpression strain (Miyazawa et al. 2018). Using α-1,3-glucan-binding domain–fused GFP (Suyotha et al. 2013), we localized α-1,3-glucan in the inner layer of the cell wall in the agsA overexpression strain, but in the outer layer in the agsB overexpression strain. We concluded that the presence of α-1,3-glucan in the outer layer is necessary for adhesion, and proposed a mechanism of cell wall spatial localization due to the molecular weight difference of polysaccharides (Miyazawa et al. 2018).

He, Li and Kaminskyj (2014) found that two α-amylase genes, amyD and amyG, downstream of agsB, are important for α-1,3-glucan biosynthesis. The amount of cell wall α-1,3-glucan was higher in the amyD disruptant than in the parental strain, but smaller in the amyD overexpression strain, indicating that amyD inhibits α-1,3-glucan biosynthesis (He, Li and Kaminskyj 2017). Hyphal pellets were smaller in the amyD overexpression strain than in the parental strain (He, Li and Kaminskyj 2017; Miyazawa et al. 2021). The mechanism of the decrease in α-1,3-glucan content differs between AmyD and α-1,3-glucanase (He, Li and Kaminskyj 2017). The molecular weight of α-1,3-glucan in the amyD overexpression strain was significantly lower than in the non-expression strain (parental or ΔamyD strain) (He, Li and Kaminskyj 2017; Miyazawa et al. 2021; Miyazawa et al. 2022). In a mutant overexpressing amyD with the GPI anchor–coding sequence deleted, the molecular weight of α-1,3-glucan was similar to that of the parental strain. Thus, removal of the GPI anchor abolished the inhibition of α-1,3-glucan biosynthesis by AmyD, but AmyD still hydrolyzed maltooligosaccharides. This suggests that immobilization of AmyD in the cell wall via the GPI anchor is essential for AmyD function. AmyD and its orthologs in aspergilli are classified as GH13 family and have been reported to have high transglycosylation activity (van der Kaaij et al. 2007).

In A. oryzae, AoAgtA is encoded by the agtA gene, an ortholog of A. nidulans amyD (Koizumi et al. 2022). As in A. nidulans, agtA overexpression in A. oryzae grown in submerged cultures decreased both the amount of cell wall α-1,3-glucan and the size of hyphal pellets in comparison with the wild-type strain. We found that recombinant (r)AoAgtA produced in Pichia pastoris degraded soluble starch, but not linear bacterial α-1,3-glucan. It cleaved 3-α-maltotetraosylglucose with a structure similar to the predicted boundary structure between the α-1,3-glucan main chain and a short α-1,4-linked glucose spacer in cell wall α-1,3-glucan (Koizumi et al. 2022). Interestingly, although rAoAgtA specifically cleaved the second α-1,4 bond of maltopentaose from the reducing end, rAoAgtA randomly cleaved only the α-1,4-glycosidic bonds of 3-α-maltotetraosylglucose, indicating that AoAgtA may cleave the spacer in cell wall α-1,3-glucan. Heterologous overexpression of AoagtA in A. nidulans decreased the molecular weight of cell wall α-1,3-glucan (Koizumi et al. 2022). These in vitro and in vivo properties of AoAgtA suggest that GPI-anchored α-amylases degrade the spacer with α-1,4-glycosidic linkages in cell wall α-1,3-glucan before its insolubilization, and this spacer cleavage decreases the molecular weight of cell wall α-1,3-glucan in vivo.

Galactosaminogalactan

In A. oryzae, α-1,3-glucan-deficient strains have a smaller pellet diameter, but their hyphae remain aggregated (Miyazawa et al. 2016), so a factor other than α-1,3-glucan likely contributes to hyphal aggregation in A. oryzae. In A. fumigatus, GAG contributes to immunosuppression of host cells (Fontaine et al. 2011) and to hyphal adhesion to host components (Gravelat et al. 2013). In A. fumigatus, screening of commonly downregulated genes in stuA and medA disruptants with reduced biofilm formation revealed that uge3, which encodes UDP-glucose 4-epimerase, is essential for GAG biosynthesis (Gravelat et al. 2013; Lee et al. 2014). A GAG-deficient strain of A. fumigatus completely lost the ability to form biofilms, as revealed by crystal violet staining (Gravelat et al. 2013; Bamford et al. 2015). Four genes (gtb3, agd3, ega3, and sph3) adjacent to uge3 coordinately contribute to the biosynthesis of GAG in A. fumigatus (Lee et al. 2016; Le Mauff and Sheppard 2023).

We hypothesized that GAG is a second adhesion factor for hyphal aggregation in A. oryzae, and disrupted the uge3 and sph3 orthologs in an α-1,3-glucan-deficient (AGΔ) strain; the disruption led to complete dispersion of the resulting α-1,3-glucan- and GAG-deficient (AGΔ-GAGΔ) strain (Miyazawa et al. 2019; Abe et al. 2021) (Figure 8a). In contrast, a GAG-deficient strain formed larger pellets than the parental strain. During the germination of the wild-type A. oryzae strain, fluorophore-labeled α-1,3-glucan was observed by 3 h after inoculation, and fluorophore-labeled GAG was observed from 6 h (Miyazawa et al. 2019). These data suggest that α-1,3-glucan exposed by conidial swelling is the first to contribute to adhesion, followed by GAG secreted during germ tube formation. We isolated GAG from the culture supernatant by fractional precipitation with ethanol, mixed it with AGΔ-GAGΔ mycelia, and observed mycelial aggregation, suggesting a direct GAG contribution to hyphal aggregation (Miyazawa et al. 2019).

As mentioned in the section “Overview of cell wall structure”, the n-acetylgalactosamine residue in GAG is deacetylated to galactosamine, which is essential for biofilm formation (Lee et al. 2016). In A. oryzae GAG, the degree of deacetylation was determined by colloidal titration using potassium poly (vinyl sulfate) and was found to be about 40% (Miyazawa et al. 2019). GAG-dependent aggregation was abolished by GAG acetylation with acetic anhydride or by inhibition of hydrogen bond formation by adding urea (Miyazawa et al. 2019). These data strongly suggest that the formation of hydrogen bonds via amino groups of galactosamine residues is important for GAG-dependent aggregation. On the contrary, hyphae of the AGΔ-GAGΔ or AGΔ strain are scarcely aggregated by the addition of partially purified α-1,3-glucan (Miyazawa et al. 2019). Because α-1,3-glucan synthesized by α-1,3-glucan synthase seems to be immobilized in the outer layer of the cell wall space (Miyazawa et al. 2018), α-1,3-glucan chains exposed on hyphal surfaces may interact with themselves (probably via hydrogen bonds) but not with β-1,3-glucan or chitin and lead to hyphal aggregation. The α-1,3-glucan- and GAG-dependent mechanisms of pellet formation in Aspergillus species are reviewed in (Miyazawa, Yoshimi and Abe 2020). Most importantly, we demonstrated that the macromorphology of mycelia can be controlled by modification of cell surface polysaccharides.

Improved enzyme production by A. oryzae AGΔ-GAGΔ strain

Regulation of macromorphology of filamentous fungi such as hyphal pellets and pulp form (dispersed hyphae) has been a key issue in fermentation using filamentous fungi; macromorphology can be controlled by adjusting culture conditions such as agitation speed, pH, and medium composition (Antecka, Bizukojc and Ledakowicz 2016; Zhang and Zhang 2016). Addition of microparticles such as titanate and talc to liquid culture media promotes the formation of micropellets that can improve productivity of fermentation of filamentous fungi (Driouch et al. 2012).

Our novel method of improving productivity with cell wall mutants of A. oryzae is shown in Figure 8. Hyphae of the AGΔ-GAGΔ mutant were fully dispersed in submerged culture (Figure 8a), and CutL1 production was significantly higher in AGΔ-GAGΔ than in the wild type in shake-flask culture (Figure 8b) (Miyazawa et al. 2019). The production of secreted Cutl1 was higher in AGΔ-GAGΔ than in the wild-type or mutants lacking α-1,3-glucan (AGΔ) or GAG (GAGΔ) in batch culture in a 5-L lab-scale bioreactor (Ichikawa et al. 2022). The apparent viscosity tended to be lower in AGΔ-GAGΔ than in the wild type at each agitation speed examined (200-600 rpm) (Figure 8c), suggesting improved culture rheology, which increased recombinant protein production (Ichikawa et al. 2022).

The AGΔ-GAGΔ strain produces more recombinant cellulase CBHI than the wild-type strain in a 250-mL bioreactor (Sakuragawa et al. 2021). Glucose consumption, mycelial dry weight, and respiration activity are higher in the AGΔ-GAGΔ than in the wild type. The levels of metabolites of glycolysis and TCA cycle are lower in AGΔ-GAGΔ than in the wild type in liquid culture, suggesting that metabolic flux is higher in AGΔ-GAGΔ than in the wild type (Sakuragawa et al. 2021). The mechanisms underlying the productivity of AGΔ-GAGΔ in the bioreactor are presently being analyzed. Productivity is expected to be further increased by conferring stress tolerance to the AGΔ-GAGΔ mutant and fine tuning culture conditions through the screening for stress factors and multi-omics analyses.

Summary and perspectives

Cell surfaces of microorganisms are involved in various biological phenomena as interfaces between cells and the extracellular environment. In this review, I described microbial cell surface molecules such as an energy-generating solute transporter in lactic acid bacteria; a fungal biosurfactant protein hydrophobin, which adsorbs to polyesters and subsequently recruits polyesterases; and fungal cell wall polysaccharides functioning as mycelial adhesion factors. The discovery of new biological functions of these cell surface molecules has opened the way to new applications in the fermentation industry. Because the biological functions of many cell surface molecules remain unidentified in microorganisms, the discovery of their new functions and physicochemical properties will lead to the development of diverse industrial applications.

Funding

The studies were partly supported by Grants-in-Aid for Scientific Research (JSPS KAKENHI) [20380175, 2629037, 17H03787, 20H02895, 23K26810], the Program for Promotion of Basic and Applied Researches for Innovations in Bio-oriented Industry (BRAIN), A-step programs of Japan Science and Technology Agency (JST), the Institute for Fermentation Osaka Grant [number L-2018–2–014], and projects [JPNP16009, JPNP20011] of New Energy and Industrial Technology Development Organization (NEDO).

Disclosure statement

No potential conflict of interest was reported by the author.

Acknowledgments

The studies described in this review were conducted at the R&D Division of Kikkoman Corporation, and at both the Graduate School of Agricultural Science and New Industry Creation Hatchery Center (NICHe) of Tohoku University. I am sincerely grateful to all of my colleagues and students who worked with me or assisted with my research. Particularly, I thank the late Dr. Fumihiko Yoshida, the former president of the R&D division of Kikkoman Corp., and Dr. Kinji Uchida, the former research leader of the microbiology group of R&D Kikkoman, for providing me with the opportunity to study the soy sauce lactic acid bacterium and industrial fungus A. oryzae. I also thank the late Dr. Peter C. Maloney, the former professor at the Johns Hopkins University School of Medicine for guiding me in the study of bioenergetics and solute transporters. I greatly appreciate Dr. Tasuku Nakajima, the Professor Emeritus of Tohoku University, and Dr. Fumihiko Hasegawa, the President of National Institute of Technology and Evaluation and the former President of NICHe of Tohoku University, for their kind support and encouragement throughout my research in Tohoku University. I also greatly thank Dr. Katsuya Gomi, Professor Emeritus of Tohoku University, for his long collaboration and kind support by his many genetic tools for A. oryzae.

References

Author notes