Phylogenetic analysis of fungal Ecm14
We first took a bioinformatics approach to understand the role of Ecm14. All fungal proteins within the NCBI database were searched (June 2019) using BLASTp for homology to bovine CPA1. This resulted in a dataset of 2068 sequences, after the deletion of sequences shorter than 150 amino acids or those lacking the majority of the carboxypeptidase domain. Alignment of these sequences using Clustal omega resulted in a phylogenetic tree which, when midpoint rooted, revealed M. daphnia CPB (XP_013238090.1) to be an ideal outgroup (Fig. 2). Mitosporidium daphniae is a microsporidium parasite, a group previously considered protozoan, but recently placed on a branch ancestral to the fungi [34, 35]. This was used as the root of our phylogenetic tree.
Protein sequences generally fell into three large groups, CPA-like, CPB-like, and Ecm14-like proteins (Fig. 2). CPA-like enzymes are predicted to cleave aliphatic or aromatic C-terminal amino acids due to a hydrophobic amino acid at the bottom of their specificity pocket (residue 255 in bovine CPA1), while CPB-like enzymes are predicted to cleave basic C-terminal amino acids with an acidic amino acid, often aspartic acid, at their residue 255 equivalent [18]. Fungal proteins identified within the CPB clade often had an Asp at position 255; however, a lot of variation was seen at this position, as nearly every amino acid was represented. Similar results were observed in the CPA clade; while most proteins had a hydrophobic amino acid at position 255, many other amino acids were found to occupy this position also. Notably, while a few representatives from the basal fungi (Chytridiomycota and Zygomycota) were found in each of these three major groups, no CPB sequences were found from Basidiomycota, and only 28 of 737 sequences in the CPA group were from Basidiomycota, a small group suggestive of a horizontal gene transfer event.
The Ecm14 clade was the largest of the three, similar to the CPA clade in number of sequences but far exceeding it in terms of species representation (481 species for Ecm14 versus 226 species for CPA; Fig. 2). All Ecm14 sequences were derived from ascomycete fungi and were strictly defined by the presence of Asp at position 144 and Lys at position 270. Active metallocarboxypeptidases have a conserved Asn at position 144, which is involved in substrate binding, and a critical Glu at position 270 which serves as a proton donor and acceptor in general acid-base catalysis [14].
A large group of CP-like proteins from Basidiomycota were identified as a sister clade to Ecm14, in which the active MCP Asn144 was conserved. However, in about half of these sequences, Arg145 was replaced with a Ser, likely resulting in the inability to bind the carboxyl group of a substrate. In the other half of these sequences, three active site substitutions were found: Arg127Tyr, His196Arg, and Glu270His (Fig. 2). Arg127 typically functions in binding of the scissile carboxyl group during cleavage, while His196 coordinates the zinc ion cofactor of active enzymes and, of course, Glu270 is necessary for acid-base catalysis. The presence of substitutions at these important residues suggests that these proteins are also inactive enzymes. Altogether, these results suggest that Ecm14 is a widely-distributed protein within the ascomycete branch of the fungal kingdom with unique active site residues that are conserved and thus likely to be functional.
Ecm14 purification and biochemical analysis
In order to determine whether Ecm14 was functional as a carboxypeptidase enzyme or as a pseudoenzyme, we attempted purification by metal-affinity chromatography. Histidine-tagged Ecm14 was expressed in its native environment, the yeast S. cerevisiae, under the control of a galactose-inducible promoter from the plasmid pEMBLyEx4. This resulted in abundant Ecm14-His6 expression, typically observed as a doublet approximately 45 kDa in size (pro-Ecm14) and another doublet at 35 kDa (mature Ecm14), as detected by anti-His6 western blotting (Fig. 3A). The two doublets were due to N-glycosylation of Ecm14, confirmed by a reduction of doublets to singlets upon incubation with EndoH (Fig. 3A). N-glycosylation was predicted at two sites by NetNGlyc, both sites found on the surface of Ecm14 (Fig. 3B,C).
Ecm14-His6 protein overexpressed in yeast was mostly insoluble and was not able to be purified even following denaturation in urea or guanidine. In contrast, when Ecm14-His6 was expressed in the Sf9 insect cell system using a baculovirus expression vector, about 50% of Ecm14-His6 protein was secreted into the media in soluble form by 3 days post-infection, typically as a doublet at 45 kDa (see Additional file 1, Figure S1A). A small amount of this 45 kDa protein was processed to produce a 35 kDa protein in Sf9 cells (see Additional file 1, Fig S1A), suggesting the removal of the prodomain. Addition of trypsin or chymotrypsin to conditioned media converted the large majority of 45 kDa protein to the 35 kDa form, and also greatly enhanced the purification of Ecm14-His6 by metal-affinity chromatography (see Additional file 1, Figure S1B-D).
The observation that Ecm14 could be processed into a smaller form by endopeptidase digestion both in vivo and in vitro suggested possible prodomain removal, and that this cleavage may be necessary for full activity of Ecm14. To investigate this maturation process further, Ecm14-His6 within conditioned Sf9 media was digested with trypsin and chymotrypsin. Digestion with 5 µg/ml chymotrypsin showed the reaction to be complete after 40 minutes at room temperature, with at least two products observed at about 35 kDa. Optimal cleavage by 5 µg/ml trypsin was observed after only 1–5 minutes, with one distinct band observed at 35 kDa (Fig. 4A). To identify the sites of cleavage, Edman degradation was performed on purified Ecm14-His6 following cleavage by either chymotrypsin or trypsin. It was found that chymotrypsin cleaved at 3 locations within the linker region between the prodomain and the carboxypeptidase domain, while trypsin cleaved only once, C-terminal to an arginine in this same region (Fig. 4B, C, see Additional file 1, Figure S2). The much cleaner and more rapid digestion by trypsin suggested that a trypsin-like enzyme might be important for full Ecm14 activity in vivo.
Purified Ecm14-His6 was incubated with standard carboxypeptidase substrates (3-(2-Furyl)acryloyl-Phe-Phe, -Phe-Ala or -Glu-Glu) at 0.5 mM and at a range of pH values from 5.5 to 7.5. No activity was detected under any of these conditions, even though the prodomain was no longer present.
Biological function of Ecm14 through targeted genetic analyses
A number of large-scale screens in yeast have suggested possible functions for Ecm14. Ecm14 was first identified in a large screen searching for mutations leading to sensitivity to calcofluor white, suggesting a cell wall defect [33]. We have not been able to clearly replicate this phenotype. While a slight sensitivity to calcofluor white was observed in the BY4741 mat a strain (see Additional file 1, Figure S3A), deletion of the ECM14 gene (ecm14Δ) in the same strain stored independently in another lab did not result in the same sensitivity (see Additional file 1, Figure S3B).
Hillenmeyer et al. attempted to tease out the functions of nonessential yeast genes through a large-scale screen in which they measured the fitness of yeast mutants upon incubation with a variety of chemicals or in various environmental conditions [36]. Decreased fitness was observed for ecm14Δ cells under a number of conditions. We incubated wild-type and ecm14Δ yeast in YPD media supplemented with many chemicals that were shown by Hillenmeyer et al. to effect fitness, including paraquat (5 mM), miconazole nitrate (25, 50, and 100 uM), alverine citrate (500 uM), hydrogen peroxide (2 and 5 mM), D-sorbitol (0.75 and 1.5 M), mercury (II) chloride (15.6 and 62.5 uM), and lithium chloride (150 and 300 mM). Of these, a small and reproducible effect was observed for lithium chloride only, that being a small increase in ecm14Δ cell growth rate compared to wild-type cells (see Additional file 1, Figure S3C).
Other reports have described a number of genetic synthetic lethal phenotypes. For example, yeast deficient for arp1 were shown to be inviable together with an ecm14 mutation [37]. Our attempts to replicate this result also came up empty-handed. Our BY4741a yeast doubly deficient for ecm14 and arp1 remained viable.
Biological function of Ecm14 through synthetic lethal genetic analysis.
In lieu of any clearly identifiable phenotype for ecm14Δ cells, we sought to identify the function of ECM14 through a synthetic lethal approach, in which the lethality of two gene mutations is rescued by expression of one wild-type gene from a plasmid. While many genes putatively synthetic lethal with the ecm14 gene have been listed in the yeast database (yeastgenome.org) from large genome-wide screens in recent years, many of these putative genetic interactions do not exceed default Pearson correlation coefficient thresholds (see thecellmap.org), and so we reasoned that a synthetic lethal experiment targeted specifically at ecm14 would provide better-validated data. Therefore, ecm14Δ cell suspensions, transformed with pSLS1-ECM14, were treated with ethyl methanesulfonate (EMS) and plated on rich media containing galactose to destabilize the pSLS1-ECM14 plasmid (see Additional file 1, Figure S4 for a map of this plasmid). Those yeast acquiring secondary mutations requiring the ECM14 gene for viability would maintain the plasmid and would be identified due to red pigmentation produced by the ADE3 gene on the plasmid (Fig. 5A). Yeast not requiring the plasmid for viability would rapidly lose it, resulting in white or sectored colonies. YPD plates were used as controls to determine the effectiveness of EMS mutagenesis and to tune EMS exposure time. An EMS exposure time of 45 minutes was determined to be effective at generating the ideal cell death (~ 40%; (Winston, 2008)) in the yeast strains used in this study (see Additional file 1, Figure S5).
Because the pSLS1-ECM14 plasmid contains two additional yeast genes (ADE3 and URA3), this approach could also identity false-positive synthetic lethals caused by a dependency on either the plasmid-borne ADE3 or URA3 genes. Effectively silencing either the ADE3 or URA3 gene could potentially improve the sensitivity of the synthetic lethal screen and decrease the number of false-positives to be subsequently evaluated. Because the URA3 gene was not involved in the synthetic lethal screen, a modification was made in later experiments in which YPGal media containing extra uracil (0.001%) was used to decrease undesirable URA3 false-positives. In addition, a modified YPGal medium containing 20 µg/ml calcofluor white was also used for screening EMS-treated cells. Exposure to calcofluor white (CFW) amplifies the effect of mutations in genes involved in cell wall architecture and remodeling [33], and therefore it was thought that CFW might increase the sensitivity of the synthetic lethal screen to novel mutations in genes functionally related to ECM14. An active, plasmid-borne ADE3 gene was required for the ADE3/ADE2 color assay essential to the synthetic lethal screen. Therefore, modifications to decrease ADE3 false-positives were not made.
Approximately 27,000 EMS-treated ecm14Δ [pSLS1-ECM14] yeast colonies were screened for novel plasmid dependencies on galactose media (either YPGal, YPGal + CFW, or YPGal + Ura + CFW). Three rounds of EMS mutagenesis were performed, with each yielding approximately 9,000 screenable EMS-treated yeast colonies. Colonies of interest were those that produced the desired non-sectored red phenotype on these media. These non-sectored red colonies of interest were re-streaked for single colonies on fresh media to further evaluate the development and the maintenance of the desired non-sectored red phenotype (Fig. 5B). The first round of EMS mutagenesis yielded five putative synthetic lethal mutants (SL1-5) derived from YPGal. The second round yielded two putative synthetic lethal mutants derived from YPGal + Ura + CFW (SL6-7) and seven from YPGal (SL8-14). The final round yielded eight putative synthetic lethal mutants derived from YPGal + Ura and YPGal + Ura + CFW (SL15-22). SL10, 11, and 13 were unique in that they did not retain their red color when plated onto YPGal containing calcofluor white. Some colonies of interest, particularly SL 9, 10, and 13, that initially produced the desired non-sectored red phenotype on these media following EMS treatment produced a mosaic phenotype when re-streaked on galactose media, suggesting that they may not be true synthetic lethals.
Additional rounds of growth on both SC-Ura and SC + 5FOA were attempted to test for true synthetic lethal mutants caused by novel secondary mutations or for potential false-positives caused by reversions of marker genes. Counterselection against the plasmid-borne URA3 gene using 5FOA should result in plasmid loss and, thus, cell death for true synthetic lethal mutants. This test was performed on the first five putative synthetic lethal mutants (SL1-5; see Additional file 1, Figure S6). Although these five putative synthetic lethal mutants seemed to retain the desired phenotype when subsequently re-plated/spotted on YPGal, all five were unable to grow appreciably on any SC-based media. However, because the EMS-treated white non-synthetic lethal colony (ecm14Δ) and untreated control strain (ecm14Δ [pSLS1-ECM14]) were able to grow on SC-based media, the apparent lack of growth of these synthetic lethal mutants seemed to be the result of secondary mutations conferring significant fitness deficiencies on SC-based media.
Synthetic lethal mutants were further analyzed to confirm synthetic lethality with ecm14 (Fig. 6). Mutants were crossed with wild-type yeast of an isogenic background, containing a plasmid (pRS316-NatMX) allowing growth on nourseothricin, and heterozygous diploids were selected by growth on YPD + G418 + nourseothricin. Diploid colonies were grown on SC + 5FOA to eliminate plasmids, and then transformed with pSLS1 or pSLS1-ECM14. Sporulation of diploids was performed and random spores germinated on YPGal plates. The percentage of unsectored red colonies was determined, and compared to expected segregation of the phenotype. If only two mutations were required for the red synthetic lethal phenotype (ie. the ecm14Δ mutation and mutation of one additional gene by EMS), then we would expect approximately ¼ of spores derived from a heterozygous diploid to be red. If two EMS mutations were required for synthetic lethality, then approximately 1/8 of spores would be red, and so on.
An average of 204 germinated spores were analyzed per synthetic lethal clone. Nine of 16 synthetic lethal clones were rescued by the pSLS(-) plasmid itself, indicating synthetic lethality with another element on the plasmid (Table 1). In these cases, the fraction of total colonies being red suggested that 1–3 EMS mutations were required for the synthetic lethality. Very few, if any, red colonies were observed for the remaining seven clones, suggesting that these were either not true synthetic lethals, or that four or more EMS mutations were required to generate the synthetic lethality, and, therefore, that useful genetic information would be difficult to identify (Table 1).
Table 1
Analysis of isolated synthetic lethals.
Name | pSLS1(-) rescue (% red colonies) | pSLS-ECM14 rescue (% red colonies) | # of causative mutations | Possible causative mutation1 |
WT | 0 | 0 | -- | -- |
SL1 | 3.9 | 2.7 | 3 | ND |
SL2 | 0 | 0 | ≥ 4 | ND |
SL4 | 0 | 0 | ≥ 4 | ND |
SL6 | 16.8 (1/6) | 20.6 (1/5) | 1 | SHM2 D219N (ADE3 SL) |
SL7 | 21.0 (1/5) | 18.5 (1/5) | 1 | SHM2 D219N (ADE3 SL) |
SL8 | 7.7 | 0 | 3 | |
SL9 | 0 | 1.5 | ≥ 4 | ND |
SL10 | 0 | 0 | ≥ 4 | ND |
SL11 | 10.5 (1/10) | 10.6 (1/10) | 2 | SHM2 G403R (ADE3 SL) |
SL12 | 0 | 0 | ≥ 4 | |
SL13 | 0 | 0.8 | ≥ 4 | SHM1 P252S (ADE3 SL) |
SL14 | 6.9 | 8.9 | 2 | SHM2 G252D (ADE3 SL) |
SL15 | 15.0 | 9.1 | 2 | ND |
SL18 | 0 | 0 | ≥ 4 | ND |
SL20 | 5.0 | 0 | 3 | ND |
SL21 | 15.1 | 7.6 | 2 | ND |
1 ND, not determined |
Identification and analysis of ECM14 interacting genes.
ECM14 interacting genes were analyzed in two ways. First, all genes reported within the Saccharomyces genome database (SGD) to interact either physically or genetically with ECM14 were submitted for Gene Ontology (GO) analysis. The top reported GO function and process terms, found to be enriched in this gene list compared with all yeast genes (p < 0.05) included terms such as vesicle-mediated transport, lipid binding, and organelle organization (Fig. 7). These terms are consistent with a secreted or secretory role for ECM14 in cell wall maintenance. Genes reported in the SGD to physically or genetically interact with ADE3 or URA3 were also assessed by GO analysis. This GO analysis was performed in order to compare ADE3 and URA3 interacting genes with those mutated in our synthetic lethal assay, to further determine likely relevance.
Seven putative synthetic lethal mutants (SL6, SL7, SL8, SL11, SL12, SL13, and SL14) were selected for whole-genome sequencing using the Illumina MiniSeq platform. Clones were chosen for a diversity of red color phenotypes prior to the above complementation analysis. Sequencing resulted in 30-50x coverage (Table S1) and, after elimination of noncoding and silent variants when compared to the WT sequence, 46–123 variants were identified for each synthetic lethal clone. Because this was too many to be useful, further analysis was performed in PROVEAN to identify those variants likely to be deleterious, using a stringent cutoff of -4.0. GO analysis was performed on these gene lists. For all clones except SL7 and SL13, some enrichment (although not always reaching significance, p < 0.05) was found for GO terms associated with ADE3 interacting genes. Very few of the GO terms identified for remaining synthetic lethal gene mutations reached a p < 0.05 level of significance. In some cases, missense mutations were identified in genes known to be synthetic lethal with ade3. For example, both SL6 and SL7 presented very consistently red synthetic lethal colonies; a mutation in the SHM2 gene resulting in a D219N amino acid change was identified in both (Table 1). Likewise, SL11 and SL14 contained mutations in the SHM2 gene, while SL13 contained a mutation in the SHM1 gene. Both the SHM1 and SHM2 genes are reported in the SGD to be synthetic lethals of the ade3 gene.