Escherichia coli strain Mach1 T1 was used for plasmid construction and cultivated in Luria–Bertani (LB) broth. A. niger CBS513.88 (△kusA△pyrG) was used as the host strain for gene manipulation (Dong et al., 2021). All Aspergillus strains were generally cultivated on Czapek-Dox (CD) medium (Ozeki et al., 1994). High osmotic pressure Czapek–Dox (HCD) medium with sucrose (1 M) was used for protoplast transformation. Dextrose peptone yeast extract (DPY) liquid medium was used for DNA extraction and RNA isolation (Imanaka et al., 2010). Complete medium (CM) was used for the observation of mycelial growth, which consisted of starch (1% w/v), peptone (1% w/v), beef extract (0.3% w/v), yeast extract (0.2% w/v), and sodium chloride (0.2% w/v). For fermentation, all Aspergillus strains were cultivated in the shake flask medium as described previously reported (Lin et al., 2020). Saccharomyces cerevisiae strain NMY51 was used as the parental strain for the Y2H screening. SD medium with different DO supplements used for the screening and cultivation of hybrid yeast strains.

We pre-constructed a universal expression vector based on the pMD18T plasmid. The promoter and terminator are Ptef and Ttef, left with the enzyme cut site BglII in the middle, followed by ANpyrG as the screening marker, and the upstream and downstream 1,000 bp fragment of the pyrG as the homology arm at both ends. All DNA fragments were amplified with the high-fidelity polymerase PrimeSTAR (Takara, Japan) following the manual standardized procedure, and ligated to the pMD18T plasmid using the NEBuilder HiFi DNA Assembly Cloning Kit (NEB, Ipswich, United States). To construct receptor overexpression plasmids, the gene for each receptor was amplified from the genome and inserted into the universal vector linearized by BglII. To construct fluorescently labeled mstE and snf3 plasmids, mstE and snf3 cDNAs and GFP were fused and inserted into the middle of Ptef-Ttef. The pyrithiamine resistance gene ptrA followed the expression cassette and was flanked by a 1,500 bp homology arm at the outermost edge for integration into the A. niger genomic AM270982:171718-174718 locus. To construct the Y2H bait plasmid, the cDNA of Erv14 was amplified by PCR, and the PCR product was constructed into the linearized plasmid pBT-STE, which was digested with SfiI. The prey plasmid library was constructed by Shanghai Biogene Biotech company. Here, the whole cDNA was integrated with the linearized plasmid pPR3-NubG, which was digested with SfiI.

The sgRNAs for target genes were searched by the online tool Cas-Designer (CRISPR RGEN Tools)¹ and screened for base preference at each position of the 20 nt protospacer sequence (Doench et al., 2014). All sgRNA sequences targeting each receptor were listed in Supplementary Table S1. The ribonucleoprotein (RNP)-based CRISPR-Cas9 method was used for the genetic manipulation of A. niger as recently reported (Yang et al., 2020). The RNP complex was pre-assembled from 6 μg of SpCas protein and 10 μg of in vitro synthesized sgRNA, which was subsequently transformed into protoplasts along with 100ug of donor DNA fragments. The donor DNA was inserted into the genome in the microhomology-mediated end joining (MMEJ) manner (Zhang et al., 2016), which consisted of the middle selection marker ANpyrG and a 39 bp homologous arm at both ends.

To observe spore germination, 10⁴ spores were inoculated into 500 μl of CD liquid medium on coverslips and incubated for 40 h at 30°C. Carefully removed the medium, and gently washed with PBS buffer. 50 μl of PBS buffer was added to the glass slide and the coverslip with mycelium was pressed over it. To observe the localization of monosaccharide transporters in the control strain and the DEL-ERV14 strain, 10⁴ spores were inoculated into 500 μl of DPY liquid medium on coverslips and incubated for 12 h at 30°C. To observe conidiophore, 10⁴ spores were inoculated around a small block of CD medium on a slide, and then covered with a coverslip and incubated on a plate for 2 days. All images were observed with a Zeiss Axio Imager A2 microscope equipped with a 100× oil objective lens.

All Aspergillus strains were cultivated in the shake flask medium at 30°C for 6 days. 2 ml sample was taken and centrifuged at 12,000 g at 4°C for 5 min, and the supernatant was used for subsequent experiments. The total protein in the culture supernatant was determined via BCA Protein Assay Kit (GBCBIO, Guangzhou, China). Glucoamylase enzymatic activity was measured as described previously reported (Zhu et al., 2017). The dry weight was measured as previously reported (Fiedler et al., 2018). Each experiment was performed with two biological replicates. Statistical analysis was performed by one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, p values were obtained using Tukey’s multiple comparison test.

We chose the mating-based split-ubiquitin system (mbSUS) alternative to the classical yeast two-hybrid (Y2H) system in particular (Grefen et al., 2009). The workflow of the high-throughput Y2H screening is the following: RNA was extracted from A. niger cultured in three different media, then cDNA was synthesized and ligated to the prey vector after the addition of an adaptor. The prey library was transformed into the bait strain and then painted in amino acid-deficient media. Ten transformants were picked in one tube within 50 μL ddH₂O, which was used directly as a template for PCR. Then the target genes were amplified. Finally, all the PCR products were mixed for NGS sequencing.

Y2H Data analysis was divided into three parts: quality control reads matching back to the reference genome, and gene expression quantification. (1) Quality control: FastQC (v0.11.9) was used to check whether the junctions of the reads obtained from sequencing were removed cleanly; (2) matching: Hisat2 (version 2.2.1) was used to match the reads back to the reference genome to obtain sam files, and Samtools was used to convert the sam files into bam files and filter out the unmatched and low-quality reads. Use Samtools to convert the sam files into bam files and filter out the unmatched and low-quality reads with the following parameters: view -F4 -q 30 -bS, and use the sort function in Samtools to sort the bam files; (3) the number of reads on the match is set with the parameters: -T 10 -p -t gene -g gene_id.

Three biological replicates of DEL-Erv14 and control strains were incubated in shake flasks with fermentation medium for 5 days. Mycelium was collected by vacuum pump, washed twice with 0.8 M NaCl and water, and put at −80°C for lyophilization. The lyophilized blocks were ground into the powder with liquid nitrogen and then placed in the tissue grinding tube with cell lysis solution and PMSF for further extraction. The lysate was transferred to a centrifuge tube and centrifuged at 12,000 g for 10 min at 4°C. The supernatant was considered as soluble protein, while the sediment was used to extract membrane proteins. The subsequent steps followed the protocol of the filamentous fungal membrane protein extraction kit (HR9086, Biorab, Beijing, China). The extracted membrane proteins were sent to GeneCreate company (Wuhan, China) for mass spectrometry detection and analysis.

To determine ER cargo receptors of A. niger, we searched for ER receptor homologous genes in the genomes of Homo sapiens, Mus musculus, Arabidopsis thaliana, Saccharomyces cerevisiae, Neurospora crassa, Aspergillus nidulans, and A. niger. Nine putative receptors in A. niger were identified. Phylogenetic analysis revealed that these receptors were mainly divided into three clusters representing three families (Figure 1A). Emp47 (An04g08830) and Vip36 (An02g04250) belonged to the ERGIC53 family; while Erv29 (An08g03960), Svp26 (An18g06740), and Erv14 (An07g09160) belonged to the ERV family. The p24 family consisted of numerous members which are divided into four main subfamilies - α, β, γ, and σ. Except for that of Arabidopsis thaliana, each p24 protein in each subfamily showed relatively good homology in all species. Only Erv25 (An01g08870), Emp24 (An08g03590), Erp2 (An09g05490), and Erp1 (An04g01780) belonged to each subfamily, respectively, in A. niger, which is much less than those of the mammals and yeast (Supplementary Table S2).

To explore the characteristics of the receptors in each family, we predicted the functional domains of nine receptors in A. niger. Both Emp47 and Vip36 of the ERGIC53 family contained a signal peptide, a Lectin-like domain, and a transmembrane region. The p24 family receptors, in addition to the conserved GOLD domain, also contained a signal peptide and a transmembrane region (Figure 1A). The ERV family receptors were characterized by the presence of multiple transmembrane regions, although the length and position of each transmembrane region were variable. Furthermore, we used deep learning tools AlphaFold2 to predict the 3D structural models of individual receptors. Each receptor, especially its conserved regions, was well-matched with those of its homologous proteins in S. cerevisiae (Supplementary Figure S1).

Overall, A. niger possessed three major families of receptors similar to those of other species such as S. cerevisiae, but differences in receptor structure and number might indicate differences in function.

To investigate the function of each putative receptor in A. niger, we constructed knockout and overexpression strains of each receptor. These strains and the control strain were inoculated on the plate with the sporulating medium (PDA), minimal medium (CD), and completed medium (CM) to observe mycelial growth (Figure 1B). Most of the overexpression strains showed no significant phenotypic changes compared to the control strains, except for the OE-Erp2 strain which appeared to grow weakly on the complete medium. However, the deletion of individual receptors from different families resulted in different effects on mycelial growth. In the ERGIC-53 family, the colony size of the DEL-Emp47 mutant was significantly smaller compared to that of the DEL-Vip36 mutant and the control strain on the PDA medium and complete medium, and the mycelium of DEL-Emp47 was not as dispersed as they were on minimal medium. In the ERV family, the DEL-Erv29 mutant showed weak spore production on the PDA medium and complete medium. The colony of the DEL-Svp26 strain showed sectors on the PDA medium, while it did not differ significantly from the control strain on the other media. The DEL-Erv14 mutant exhibited particularly pronounced phenotypic alterations in all three mediums. It had a more aggregated colony morphology and a significantly reduced number of spores. In the p24 family, the DEL-Erp1 mutant showed a marked reduction in growth in both mediums. However, the DEL-Erv25, the DEL-Emp24, and the DEL-Erp2 mutants did not show significant differences, and even the DEL-Erp2 strain grew better on the minimal medium compared to the control strain. To sum up, the deletion of Vip36, Erv25, and Emp24 had no effect on mycelial growth, and the deletion of Emp47, Erv29, Svp26, and Erp1 slightly slowed down strain growth, while the deletion of Erp2 promoted strain growth on minimal medium. The growth of the DEL-Erv14 mutant was severely suppressed and was significantly inferior to the other mutant and control strains in terms of colony size and sporulation capacity.

A. niger is considered an industrially natural host for protein production due to its exceptional secretion capacity, and how the receptor affects its extracellular protein secretion caused our concern. All the receptor overexpression strains and the deletion mutants as well as the control strain were fermented in shake flasks and sampled on the sixth day. After measuring the total protein in the supernatant of the fermentation broth, we found that overexpression of Emp47, Erv29, Svp26, and Erv14 had almost no impact, while overexpression of Vip36, Erv25, Erp2, and Erp1 showed a 10–20% decrease. The OE-Emp24 strain was the only one with an elevated total extracellular protein yield of 15% compared to the control strain. Among the receptor deletion strains, the DEL-Erv14 strain declined the most significantly, and the DEL-Emp24 decreased by 20%, while the other strains did not show much difference (Figure 1C). Furthermore, glucoamylase activity was also measured as a representative of the endogenous secreted proteins of A. niger. Overexpression of individual receptors did not seem to have much effect on promoting the secretion of glucoamylase. The highest elevation was only 15% in the OE-Erv14 strain, while it was even reduced by 30% in the OE-Emp24 strain. However, the glucoamylase activity varied among the deletion mutants, the DEL-Vip36 had higher enzyme activity than the wild strain, while the DEL-Svp26, the DEL-Erv25, the DEL-Emp24, and the DEL-Erp2 were almost unchanged. Knockout of Emp47, Erv29, Erv14, and Erp1 inhibited the secretion of glucoamylase in varying degrees, with the enzyme activity of the DEL-Erv14 strain reduced by about 70% (Figure 1D).

On the whole, overexpression of receptors had limited effects on mycelial morphology and protein secretion, while deletion of some receptors such as Erv14 and Erp1 suppress hyphae growth and glucoamylase secretion.

Since Erv14 appeared to be the most significant influence from the above characterization of the entire ER receptor, we decided to reveal its function in mycelial growth and protein secretion in detail. First, we observed the conidial peduncles of the OE-Erv14 and the DEL-Erv14 strains on the solid medium. Compared with the control strain, the OE-Erv14 strain showed no significant difference, while the DEL-Erv14 strain showed a reduced number of conidial peduncles and some conidial heads were defective or even naked (Figure 2A). In the liquid medium, spore germination of DEL-Erv14 strain was also significantly delayed compared to the control strain and overexpression strain. After 20 h of inoculation, the spores of the control strain had germinated into mycelium, and the spores of the OE-Erv14 strain had germinated into mycelium with multiple branches, while the DEL-Erv14 strain was still spore-like. After 40 h of incubation, different from the germling tube of the control and the overexpression strains that maintained polarized extension, the spores of the deletion strain sprouted into short but expanded mycelium (Figure 2B). Furthermore, we tested the tolerance of the overexpression and the deletion strains under the high salt, reduction, and cell wall stresses. The mycelium of all strains aggregated in the presence of high Na⁺ and K⁺ salt ion concentrations, although the DEL-Erv14 strain showed more significant changes than the control strain. In the medium with 50 mM CaCl₂ or 2 mM DTT, the OE-Erv14 strain grew better than the control strain, while the DEL-Erv14 strain grew weaker. Under cell wall stress, the Congo Red completely inhibited the growth of the DEL-Erv14 mutant and had a few effects on the OE-Erv14 strain (Figure 2C). On the other hand, we speculated the growth of the OE-Erv14 and the DEL-Erv14 strains in liquid fermentation by continuous measuring changes in dry weight, total extracellular protein and glucoamylase. The biomass of DEL-Erv14 strain was significantly less than the other two strains during the whole fermentation process. The growth trend of OE-Erv14 strain was similar to that of the control strain, except that the growth was slower in the early stage (Figure 2D). To prevent interference from differences in biomass, we normalized total secreted protein and glucoamylase enzyme activity by dry weight and then compared them. In terms of total proteinper unit dry weight, the amount of the OE-Erv14 strain was highest in the early growth phase and then decreased. In contrast, the DEL-Erv14 strain instead had a relatively high amount of protein on the sixth day (Figure 2E). In terms of glucoamylase, the enzyme activity of both the control and the OE-Erv14 strain reached a peak on the fourth day and then decreased. The enzyme activity of strain DEL-Erv14 maintained at a low level but elevated a little on the sixth day (Figure 2F). In a word, the DEL-Erv14 strain showed significant phenotypic changes - delayed germination, deformed mycelium, poor tolerance to environmental stress. Moreover, although the protein secretion level in the DEL-Erv14 strain was decreased, it might be due to the slow growth of the mycelium, rather than directly affecting the transport of secreted proteins.

The transport of cargo proteins depends on receptor recognition and vesicle assembly. To fully characterize the proteins associated with Erv14 transport in A. niger, we need to construct a high-throughput screening method. Y2H is a well-established genetic tool for identifying protein–protein interactions, but it has limitations (Rao et al., 2014). To improve the throughput of the assay, we combined Y2H and NGS techniques. The workflow was showed in Figure 3. First, cDNAs from different culture conditions were extracted and used to construct a prey library. We then examined the quality of this prey library. The concentration of the library was found to be 2.3 × 10⁶ CFU/ml by colony counting. The PCR results of randomly selected 96-well colonies showed a 97% positive rate with an average insert size of 1,200 bp. To test the library coverage, the library plasmids were sequenced by NGS. The results showed that the library contained a total of 6,057 protein-coding genes located at various locations within the cells (Supplementary Figure S2). Any gene of interest could be cloned on the bait plasmid. Following the transfer of the bait plasmid into the host and testing for not self-activation, the library plasmids were transferred into the host. Transformants containing both the bait plasmid and the prey plasmid were able to grow on SD-Leu/Trp/His/Ade medium. Each strain from the transformation plate was picked onto the amino acid-deficient medium and then every 10 colonies were mixed into one well for PCR. Finally, the resulting mixed PCR samples were sequenced by NGS.

Following the above approach, we used Erv14 as bait to screen for proteins interacting with it in the library. As the negative control, we also co-transformed the empty bait vector with the prey library. Unexpectedly, transformants also grew on the screening plate even in absence of any receptor genes. We washed down all these transformants and sent them for sequencing as well, which would be considered as background “Empty” (Supplementary Figure S3). Mapping the sequencing reads to the genome, we identified 1,528 potential Erv14-interacting proteins, 323 of which were also present in the “Empty” (Figure 4A). To eliminate the background, we calculated the FPKM value from the mapped reads and length of each gene. There were many genes with high FPKM values in the background - often those genes that are highly expressed by A. niger itself - probably transferred into the same host along with other genes. Then we used the FPKM (Fragments Per Kilobase of exon model per Million mapped fragments) values of each gene in the background as the baseline and calculated the fold change and value of p. Only those genes of logFC_Erv14/EMPTY greater than 1.5 and p-value less than 0.05 were accepted. We finally obtained 1,276 proteins that interacted with Erv14 after filtering (Supplementary Table S3). Of these proteins, 239 were predicted to have at least one transmembrane region, while the remaining 1,037 proteins did not. Comparing the proportion of transmembrane and non-transmembrane proteins in each FPKM value range, we found that transmembrane proteins had higher FPKM values (Figure 4B). It is generally believed that the higher the FPKM value of which gene is, the higher his confidence level is. Therefore, we thought that Erv14 is more likely to interact with these transmembrane proteins. To further resolve the function of these proteins, the GO distribution of annotated genes was generated using g: Profiler (Supplementary Table S4). The molecular function of the genes that interacted with Erv14 was enriched in various transporter activities (Figure 4C). To verify this result, we selected different types of transporters to construct prey vectors for point-to-point hybridization, including monosaccharide transporter, ion transmembrane transporter, and amino acid or oligopeptide transporter. The COPII complex subunit Sec24 (An08g10650) was reported to act as a cargo adapter to interact directly with transmembrane proteins, so it has also been constructed on bait vectors to detect interactions with transporters. We found that both Erv14 and Sec24 can interact with all six transporters (Figure 4D). To further verify that the transport of these transporters is associated with Erv14 in A. niger, we observed the subcellular localization of two polysaccharide transporters. The mstE and snf3 fused with GFP were overexpressed in the control and DEL-ERV14 strains, respectively. As shown in Figure 4E, although mstE and snf3 were diffusely distributed throughout the mycelium in both strains, they were significantly aggregated at the edge of the mycelium in the control strain, but not in the DEL-ERV14 strain. In a word, we applied a high-throughput Y2H assay to screen 1276 proteins that may interact with Erv14, and demonstrated high confidence in the interaction between Erv14 and multiple transporters by point-to-point Y2H validation and fluorescence localization.

Since Erv14 could interact with many transporters in vitro, membrane proteins of the DEL-Erv14 strain were analyzed by mass spectrometry to reveal whether the membrane proteins were reduced with the absence of Erv14 in A. niger. Three independent cultures of the deletion and the control strains were extracted (Supplementary Figure S4). And the quantitative membrane proteome study was performed by Sequential Window Acquisition of all Theoretical fragment ions(SWATH)technique. A total of 728 membrane proteins were determined in all strains. Among them, most proteins were not significantly different, but the abundance of 78 proteins was significantly up-regulated and 200 proteins were significantly down-regulated (Figure 5A). The up-regulated proteins were mainly enriched in endomembrane systems and were associated with NADH dehydrogenation. And among the down-regulated proteins, the majority were located in the cell membrane and endoplasmic reticulum, exercising transporter function, which was consistent with the functions of the proteins that we identified in the Y2H screening (Figure 5B; Supplementary Figure S5). In addition, Erv29, another receptor of the Erv family, and Erp2, a receptor of the p24 family, were found to have decreased protein amounts. The expression levels of vesicle transport-associated SNARE proteins SsoA and SncA were also impaired (Supplementary Table S5). Comparing the proteins down-regulated in the membrane proteome to those screened by Y2H, we found 37 proteins identified by both methods, of which nearly one-third were transporters (Figure 5C). These transporters belonged to different families. Among them, the transporters from the Major Facilitator family (MFS) were mainly responsible for the transport of monosaccharides, while different types of ATPase-coupled transporters were associated with the transport of metal ions. Moreover, there were also permeases, drug transporter and nucleoside transporter, etc. (Table 1). Furthermore, we tried to draw the associated network between Erv14 and the down-regulated proteins. 57 of 200 proteins were predicted to be capable of forming protein–protein interaction networks by STRING. As shown in Figure 6, Erv14 was closely related to other proteins located on the ER such as other receptors, and molecular chaperones, which then dispersed to vesicular transport-related proteins and eventually associated with functional proteins. Among these proteins, chsA (An09g02290) and chsD (An09g04890) were known as chitin synthases, and fksA (An06g01550) and gelA (An10g00400) encoded glycoside synthases, which were associated with fungal-type cell wall biogenesis process. Additionally, several putative peptidases (An14g03420, An02g06330, An18g03980) and alkaline phosphatase (An07g07520) were thought to be associated with the organic substrate metabolic process. Another cluster of processes related to lipid metabolism includes acyltransferase (An13g00040), diacylglycerol pyrophosphate phosphatase (dppA, An02g01180), Cytochrome b5 (An11g04370), phosphatidate cytidylyltransferase (An07g09570). In conclusion, we identified many proteins with reduced abundance due to Erv14 deletion by quantitative proteomics, which was associated with vesicle transport, cell wall synthesis, and lipid and organic substrate metabolism.