Generation of a Yeast Cell Model Potentially Useful to Identify the Mammalian Mitochondrial N-Acetylglutamate Transporter

The human mitochondrial carrier family (MCF) consists of 53 members. Approximately one-fifth of them are still orphans of a function. Most mitochondrial transporters have been functionally characterized by reconstituting the bacterially expressed protein into liposomes and transport assays with radiolabeled compounds. The efficacy of this experimental approach is constrained to the commercial availability of the radiolabeled substrate to be used in the transport assays. A striking example is that of N-acetylglutamate (NAG), an essential regulator of the carbamoyl synthetase I activity and the entire urea cycle. Mammals cannot modulate mitochondrial NAG synthesis but can regulate the levels of NAG in the matrix by exporting it to the cytosol, where it is degraded. The mitochondrial NAG transporter is still unknown. Here, we report the generation of a yeast cell model suitable for identifying the putative mammalian mitochondrial NAG transporter. In yeast, the arginine biosynthesis starts in the mitochondria from NAG which is converted to ornithine that, once transported into cytosol, is metabolized to arginine. The deletion of ARG8 makes yeast cells unable to grow in the absence of arginine since they cannot synthetize ornithine but can still produce NAG. To make yeast cells dependent on a mitochondrial NAG exporter, we moved most of the yeast mitochondrial biosynthetic pathway to the cytosol by expressing four E. coli enzymes, argB-E, able to convert cytosolic NAG to ornithine. Although argB-E rescued the arginine auxotrophy of arg8∆ strain very poorly, the expression of the bacterial NAG synthase (argA), which would mimic the function of a putative NAG transporter increasing the cytosolic levels of NAG, fully rescued the growth defect of arg8∆ strain in the absence of arginine, demonstrating the potential suitability of the model generated.


Introduction
Mitochondrial carriers (MCs) are a family of eukaryotic intracellular transport proteins, primarily localized in the inner mitochondrial membrane, characterized by a tripartite structure consisting of three homologous domains, each composed of approximately 100 amino experimental approach: (i) recombinant transporters, usually expressed as inclusion bodies, need to be refolded in an active form before their reconstitution into liposomes and protein folding may not occur properly; (ii) not all metabolites, known to cross the inner mitochondrial membrane, are available in a radiolabeled form. However, this latter limitation can sometimes be overcome by using chemically related molecules [36][37][38][39][40][41][42][43]. Indeed, the identification of the mitochondrial NAG transporter is negatively affected by the lack of commercially available radiolabeled NAG. In the present study, we set up a yeast cell model that can be used for identifying putative NAG carrier(s). The tool that we have engineered may be used to test the NAG transport activity of the remaining members of the mitochondrial carrier family or other mitochondrial transporter families that are still orphan of a function [44,45]. N-acetyl-glutamate synthase (Arg2p) and N-acetyl-ornithine acetyltransferase (Arg7p) synthesize NAG in the mitochondria. N-acetyl-ornithine acetyltransferase also In E. coli, arginine biosynthesis occurs in the cytosol. argA, NAG synthase; argB, NAG kinase; argC, N-acetylglutamyl-5-phospate reductase; argD, N-acetyl-ornithine amino transferase; argE, N-acetyl ornithine deacetylase; argF/I, ornithine carbamoyl transferase; argG, argininosuccinate synthase; argH, argininosuccinate lyase. (B) In S. cerevisiae, the arginine biosynthesis starts in the mitochondrial matrix from the condensation of glutamate with acetyl-CoA catalyzed by ARG2 and ends with the synthesis of ornithine catalyzed by ARG7. Ornithine is then transported to the cytosol by ARG11, where it is finally converted to arginine. In S. cerevisiae the mitochondrial catabolism of proline may also produce N-acetyl-L-glutamyl-5-semialdehyde bypassing the first two steps of the main biosynthetic pathway. Although mechanistically the first three steps of the two pathways are not identical (see text for more details), the same metabolic intermediates are involved in both species. ARG2, NAG synthase; ARG5,6, N-acetylglutamate kinase and N-acetyl-5-glutamyl-phosphate reductase; ARG8, N-acetyl-ornithine aminotransferase; ARG7, N-acetyl-ornithine acetyltransferase; ARG11, mitochondrial ornithine transporter; ARG3, ornithine carbamoyl transferase; ARG1, argininosuccinate synthetase; ARG4, argininosuccinate lyase; PUT1, proline oxidase; PUT2, (S)-1-pyrroline-5-carboxylate dehydrogenase. MPR1, L-azetidine-2-carboxylic acid acetyltransferase.

Yeast Strains and Growth Conditions
A BY4742 (wild-type, MATα his3∆1 leu2∆0 lys2∆0 ura3∆0) yeast strain was provided by the EUROFAN resource center EUROSCARF (Frankfurt, Germany). The yeast gene deletions (arg2∆/arg7∆ and arg8∆) were achieved as described before [9]. The ARG2 and ARG8 deletion cassettes, carrying the nourseothricin resistance gene, were amplified using the pAG25 plasmid as a template, whereas that of ARG7, carrying the kanamycin resistance gene, was amplified using the pUG6 plasmid as a template. Supplementary Table S1 lists all oligos used. Yeast cells were transformed using the lithium acetate method [46] and transformants were selected on synthetic complete medium (SC) plates lacking uracil (SC-URA) or uracil and histidine (SC-URA/HIS). All yeast strains were phenotyped on (SC-URA) or (SC-URA/HIS) plates in the presence or absence of arginine. Cultures were started from medium log precultures grown on SC selective media in the presence of arginine; then, cells were washed, diluted, and spotted on the selective SC media, with or without arginine, and incubated for 48 h at 30 • C. Ten-fold serial dilutions of wild-type, deleted, and transformed strains were analyzed.

Construction of the Multicopy Expression Vectors
To express up to five genes, we initially cloned various combinations of the genes of interest in the polycistronic yeast expression vector pSP-GM2 [47,48]. The five E. coli genes, argA-E, were amplified by PCR using the DNA of the TOP10 strain as a template [49] argA, argC, and argD were cloned by BamHI/SalI digestion under the control of the TEF1 promoter and carried a Myc-Tag at their C-termini (Supplementary Figure S1). argB and argE, carrying an HA-Tag at their C-termini, were cloned by SpeI/SacI digestion under the control of the PGK1 promoter in the pSP-GM2_argC and pSP-GM2_argD, respectively (Supplementary Figure S1). All plasmids were sequence verified. Supplementary Table S1 lists all oligos and cloning sites used.

Construction of Centromeric Polycistronic Expression Vectors
To place the pSP-GM2 expression cassettes in the yeast centromeric vectors, the CYC1 terminator region of pSP-GM2 was amplified by PCR and cloned by XhoI/SacI digestion into the polylinker of pRS413 or pRS416 vectors ( Figure 2). Adjacent to the XhoI site, the forward primer carried an AscI restriction site which permitted the further subcloning of the pSP-GM2 polycistronic expression cassettes in the two modified centromeric vectors. The whole polycistronic expression modules with part of the CYC1 terminator of pSP-GM2_argC, pSP-GM2_argB-C, and pSP-GM2_argD-E were digested by AscI and BsrGI and then subcloned in the modified pRS413 (argC and argB-C) or pRS416 (argD-E) vectors ( Figure 2). The TEF1-ORF-CYC1 expression cassettes of pSP-GM2_argA was amplified by PCR and cloned AscI and AvrII in the pRS416_argD-E ( Figure 2).
The ARG8-pRS416, ARG2-pRS416 and ARG7-pRS413 plasmids were constructed by cloning DNA fragments of approximately 2200, 2700 and 2200 bp, containing in the centromeric plasmids the open reading frame and approximately 700 bp upstream and 250 bp downstream of the three genes, respectively. All plasmids were sequence verified. All oligos and cloning sites used are listed in Supplementary Table S1.

Figure 2.
Schematic representation of the centromeric plasmid constructs. The pRS413-CYC1 and pRS416-CYC1 entry vectors were produced from the original centromeric vectors by cloning XhoI/SacI the CYC1 terminator in the polylinker region. The presence of AscI restriction site adjacent to XhoI allowed the further cloning of the whole expression cassettes already cloned in the pSP_GM2 plasmids. The pRS413_argC, pRS413_argB-C, and pRS416_argD-E constructs were generated by cloning AscI/BsrGI the whole expression cassettes of the of the pSP-GM2 plasmids reported in Figure  S1. The pRS416_argA-D-E plasmid was originated by cloning AscI/AvrII the PCR-amplified fragment TEF1-argA-CYC1 (pSP-GM2_argA, Supplementary Figure S1) in the vector pRS416_argD-E. See details in the text.
The ARG8-pRS416, ARG2-pRS416 and ARG7-pRS413 plasmids were constructed by cloning DNA fragments of approximately 2200, 2700 and 2200 bp, containing in the centromeric plasmids the open reading frame and approximately 700 bp upstream and 250 bp downstream of the three genes, respectively. All plasmids were sequence verified. All oligos and cloning sites used are listed in Supplementary Table S1.

Western Blot Analysis
Yeast cell extracts were prepared as previously described [14,50]. Briefly, the various yeast strains were grown to 2-2.5 OD600 in liquid SC-URA/HIS and glucose as carbon source. The cell pellet was washed twice with distilled water and incubated in 0.1 M NaOH for 5 min at room temperature. The NaOH-treated cells were centrifuged at 12,000× g for 5 min, resuspended in a suitable volume of Laemmli sample buffer, and boiled for 3 min before loading. Yeast cell lysates (approximately 50 µg of proteins) were separated by SDS-polyacrylamide gel electrophoresis and transferred onto nitrocellulose membrane by a Trans-Blot Turbo Transfer System (Bio-Rad Laboratories, Rome, Italy). Membranes were blocked in Tris-buffered saline (TBS) containing 1% BSA and 0.1% Tween 20 before incubation with an anti-V5-Tag (1:5000 catalog no. 13,202; Cell Signaling Technology, Leiden, The Netherlands) or anti-HA.11 Epitope Tag Antibody clone 16B12 (1:5000 catalogue no. 901501; BioLegend, London, UK) primary antibody overnight at 4 °C. A rabbit antiserum against yeast Aac2p, a kind gift from L. Pelosi, was used for protein normalization [51,52]. Membranes were washed with TBS and Tween 20 and incubated with the appropriate dilution of the horseradish peroxidase (HRP)-conjugated secondary antibody for 1 h at room temperature and washed again [53][54][55]. The SuperSignal West Pico PLUS Chemiluminescent Substrate (no. 34,577; Thermo Fisher Scientific, Monza, Italy) was used for immunodecoration. Images were acquired by GelDoc (Bio-Rad Laboratories, Rome, Italy).

Figure 2.
Schematic representation of the centromeric plasmid constructs. The pRS413-CYC1 and pRS416-CYC1 entry vectors were produced from the original centromeric vectors by cloning XhoI/SacI the CYC1 terminator in the polylinker region. The presence of AscI restriction site adjacent to XhoI allowed the further cloning of the whole expression cassettes already cloned in the pSP_GM2 plasmids. The pRS413_argC, pRS413_argB-C, and pRS416_argD-E constructs were generated by cloning AscI/BsrGI the whole expression cassettes of the of the pSP-GM2 plasmids reported in Figure  S1. The pRS416_argA-D-E plasmid was originated by cloning AscI/AvrII the PCR-amplified fragment TEF1-argA-CYC1 (pSP-GM2_argA, Supplementary Figure S1) in the vector pRS416_argD-E. See details in the text.

Western Blot Analysis
Yeast cell extracts were prepared as previously described [14,50]. Briefly, the various yeast strains were grown to 2-2.5 OD600 in liquid SC-URA/HIS and glucose as carbon source. The cell pellet was washed twice with distilled water and incubated in 0.1 M NaOH for 5 min at room temperature. The NaOH-treated cells were centrifuged at 12,000× g for 5 min, resuspended in a suitable volume of Laemmli sample buffer, and boiled for 3 min before loading. Yeast cell lysates (approximately 50 µg of proteins) were separated by SDS-polyacrylamide gel electrophoresis and transferred onto nitrocellulose membrane by a Trans-Blot Turbo Transfer System (Bio-Rad Laboratories, Rome, Italy). Membranes were blocked in Tris-buffered saline (TBS) containing 1% BSA and 0.1% Tween 20 before incubation with an anti-V5-Tag (1:5000 catalog no. 13,202; Cell Signaling Technology, Leiden, The Netherlands) or anti-HA.11 Epitope Tag Antibody clone 16B12 (1:5000 catalogue no. 901501; BioLegend, London, UK) primary antibody overnight at 4 • C. A rabbit antiserum against yeast Aac2p, a kind gift from L. Pelosi, was used for protein normalization [51,52]. Membranes were washed with TBS and Tween 20 and incubated with the appropriate dilution of the horseradish peroxidase (HRP)-conjugated secondary antibody for 1 h at room temperature and washed again [53][54][55]. The SuperSignal West Pico PLUS Chemiluminescent Substrate (no. 34,577; Thermo Fisher Scientific, Monza, Italy) was used for immunodecoration. Images were acquired by GelDoc (Bio-Rad Laboratories, Rome, Italy).

ARG8 Deletion Induces an Arginine Auxotrophy in S. cerevisiae
A suitable yeast cell model to identify the mammalian mitochondrial NAG transporter could be developed by blocking the arginine biosynthesis at the mitochondrial level and creating, at the same time, a novel NAG-dependent cytosolic pathway. The latter task could be easily achieved by expressing four E. coli enzymes (argB-E) which, starting from NAG, could synthesize ornithine in the cytosol. The former task needed to be investigated in more detail since we needed to block the arginine biosynthesis in the matrix without affecting the NAG synthesis, which, in the final model, had to be exported to the cytosol by the unknown transporter and used by argB-E for its conversion to ornithine. Among the mitochondrial enzymes involved in the arginine biosynthesis, the SGD database (https://www.yeastgenome.org/ (accessed on 21 April 2023)) reports that the single deletion of ARG2 (YJL071w), ARG5,6 (YER069w) and ARG7 (YMR062c) genes leads to an arginine auxotrophy [56][57][58]. The deletion of ARG2 or ARG7 is not helpful for our project since both genes synthesize mitochondrial NAG (Figure 1). The deletion of ARG5,6 is not functional as well because it is known that in yeast Arg5,6p forms a metabolon with Arg2p, and deletions of various length of the ARG5,6 locus affects the enzymatic activity of Arg2p [59].We should also emphasize that performing a blast search of different databases, SGD included, using the human NAG synthase protein sequence, invariably produces the best score with the yeast Arg5,6p and not Arg2p (Supplementary Figure S2), suggesting that in the yeast metabolon, Arg5,6p may be directly involved with Arg2p in the NAG synthesis. For all these reasons, we focused our studies on the ARG8 gene. On the one hand, deleting this gene would impair mitochondrial ornithine and arginine synthesis and increase the mitochondrial NAG levels. Furthermore, ARG8 deletion would also cut the proline-derived N-acetyl-L-glutamyl-5-semialdehyde out. As shown in Figure 3A,F, the arg8∆ strain did not grow on the SC medium without arginine, whereas no growth defect was observed in the rescued strain with the heterologous expression of Arg8p or in the presence of arginine.

arg8∆ Strain Expressing argB-E Shows a Leaky Phenotype When Grown on SC Medium without Arginine
Although the results on the arg8∆ strain were encouraging, it was essential to verify that: (i) yeast does not express any endogenous NAG mitochondrial transporter, which

arg8∆ Strain Expressing argB-E Shows a Leaky Phenotype When Grown on SC Medium without Arginine
Although the results on the arg8∆ strain were encouraging, it was essential to verify that: (i) yeast does not express any endogenous NAG mitochondrial transporter, which would make our final model useless; (ii) the NAG synthesis in yeast occurs only in the matrix. To exclude both possibilities, we expressed four E. coli enzymes, argB-E, in the arg8∆ strain since, in the presence of cytosolic NAG, they could bypass the mitochondrial block generated by the lack of Arg8p.
The four genes were initially cloned in a multicopy plasmid (pSP-GM2) under the control of strong constitutive promoters (TEF1/PGK1) (Supplementary Figure S1). Yeast transformants expressing argB-E showed a different colony morphology (size and edge, not shown), suggesting that the expression of the four bacterial enzymes made the yeast strain unstable. A possible explanation could be that a high expression of the four E. coli proteins overworks the yeast protein synthesis apparatus. This issue was solved by moving all the expression cassettes from the pSP-GM2 to the centromeric pRS vectors (Figure 2). Although we observed poor but visible growth, the successful expression of the four bacterial enzymes ( Figure 3B) did not fully rescue the growth defect of the arg8∆ strain when grown in the absence of arginine ( Figure 3C,F). Since the four bacterial enzymes could partially restore the growth defect by using any intermediates from the NAG to NAG-γ-semialdehyde arginine biosynthetic pathway, we further verified the essential role of NAG in our model by expressing only argC-E in the arg8∆ strain. As shown in Figure 3D, the lack of bacterial N-acetylglutamate kinase (argB) made again the strain fully auxotrophic for arginine. This result clearly demonstrated that the cytosolic synthesis of ornithine in the arg8∆ strain expressing argB-E started from NAG and not from other downstream intermediates.

In Yeast the NAG Synthesis Takes Place Only in Mitochondria
Although the previous experiments suggested that NAG is present in the yeast cytoplasm, they did not test whether NAG originates in the cytosol or the mitochondrion. To answer this question, we constructed a double-deleted yeast strain lacking ARG2 and ARG7, the only known genes able to synthesize NAG in the matrix ( Figure 1B). As expected, the double-deleted strain did not grow on SC lacking arginine. No growth defect was observed in the presence of the amino acid ( Figure 3E,F), suggesting that the lack of Arg2p and Arg7p leads to a shortage of ornithine required to finalize the arginine biosynthesis in the cytosol ( Figure 1B). To test this hypothesis and exclude any cytosolic synthesis of NAG, we expressed the four E. coli genes, argB-E, in this strain because the synthesis of NAG in the mitochondrion would make their enzymatic activity useless. As shown in Figure 3E,F the expression of the four bacterial genes could not rescue the growth defect of the arg2∆/arg7∆ strain on the SC medium without arginine. These results demonstrate that yeast NAG synthesis occurred only in the mitochondrial matrix.

The Enzymatic Activity of argB-E Depends on the Cytosolic Levels of NAG
To prove that our model could be a valuable tool for identifying the putative mammalian mitochondrial NAG transporter, it was essential to understand the causes of the leaky growth showed by the arg8∆ strain expressing argB-E in the absence of arginine. There could have been at least two possible reasons. One is that yeast expressed an endogenous mitochondrial NAG transporter, and the four bacterial enzymes had poor enzymatic activity. The other is that the bacterial enzymes were fully functional and the leaking of NAG from mitochondria was due to its accumulation in the matrix which could induce a poor and non-specific transport by one or more than one mitochondrial transporter. In both cases, our model would be useless. To shed light on this point, we expressed in the arg8∆ and arg2∆/arg7∆ yeast strains the first five enzymes of the bacterial arginine biosynthetic pathway, argA-E (Figure 1), i.e., we also expressed the bacterial NAG synthase. As shown in Figure 4, the expression of the five bacterial enzymes fully rescued to proliferation defect in the absence of arginine in both yeast strains. These results confirmed that: (i) the NAG synthesis in yeast occurs only in the matrix; (ii) the cytosolic bacterial biosynthetic branch was fully functional; (iii) the model we generated was suitable to identify the putative mammalian NAG transporter.

Discussion
The MCF belongs to the sizeable solute carrier protein (SLC) family, which consists of approximately 450 members [60]. The biochemical function of many members of the SLC family still needs to be discovered up to date. The leading cause of this need for more information is the difficulty of obtaining these proteins in a highly purified and active form suitable to be studied in artificial lipid vesicles (liposomes). A significant step forward in the field of MCF was made in 1993 when the first bacterial recombinant mitochondrial transporter was reconstituted in an active state into liposomes [61,62]. Since then, many other mitochondrial carriers have been functionally characterized by the same experimental approach. However, a significant number of them still need to be discovered. The functional identification of this group of proteins is of the utmost importance because their mutation can cause human diseases [32,63]. The biochemical characterization of these transporters facilitates their association with clinical alterations found in patients. To date, mutations in all enzymes and two mitochondrial transporters involved in the UC have been found (https://www.orpha.net/consor/cgibin/OC_Exp.php?lng=en&Expert=79167 (accessed on 21 April 2023)). The only exception is the mitochondrial NAG transporter, whose encoding gene is unknown. Although the NAG transporter is not directly involved in the UC, it has a key role in its regulation. Mammals cannot tightly control the mitochondrial synthesis of NAG and cannot degrade this metabolite in the matrix. When the UC must be downregulated, NAG is exported to the cytosol, where it is de-acetylated [23,24]. Currently, there is no suggested good candidate as mitochondrial NAG transporter among the remaining functionally uncharacterized members of the SLC families. In addition, the radiolabeled NAG, which is needed in transport assays with the recombinant reconstituted putative transporters, is not commercially available. Therefore, we aimed to set up an alternative yeast tool suitable for identifying the mitochondrial NAG carrier ( Figure 5).

Discussion
The MCF belongs to the sizeable solute carrier protein (SLC) family, which consists of approximately 450 members [60]. The biochemical function of many members of the SLC family still needs to be discovered up to date. The leading cause of this need for more information is the difficulty of obtaining these proteins in a highly purified and active form suitable to be studied in artificial lipid vesicles (liposomes). A significant step forward in the field of MCF was made in 1993 when the first bacterial recombinant mitochondrial transporter was reconstituted in an active state into liposomes [61,62]. Since then, many other mitochondrial carriers have been functionally characterized by the same experimental approach. However, a significant number of them still need to be discovered. The functional identification of this group of proteins is of the utmost importance because their mutation can cause human diseases [32,63]. The biochemical characterization of these transporters facilitates their association with clinical alterations found in patients. To date, mutations in all enzymes and two mitochondrial transporters involved in the UC have been found (https://www.orpha.net/consor/cgi-bin/OC_Exp.php?lng=en& Expert=79167 (accessed on 21 April 2023)). The only exception is the mitochondrial NAG transporter, whose encoding gene is unknown. Although the NAG transporter is not directly involved in the UC, it has a key role in its regulation. Mammals cannot tightly control the mitochondrial synthesis of NAG and cannot degrade this metabolite in the matrix. When the UC must be downregulated, NAG is exported to the cytosol, where it is de-acetylated [23,24]. Currently, there is no suggested good candidate as mitochondrial NAG transporter among the remaining functionally uncharacterized members of the SLC families. In addition, the radiolabeled NAG, which is needed in transport assays with the recombinant reconstituted putative transporters, is not commercially available. Therefore, we aimed to set up an alternative yeast tool suitable for identifying the mitochondrial NAG carrier ( Figure 5). Figure 5. The yeast cell model generated to identify the putative mitochondrial NAG transporter/s. Yeast cells lacking the ARG8 gene are auxotrophic for arginine. The expression of the E. coli argB-E in the cytosol makes cells dependent on the exit of NAG from the matrix. In this model, the expression of a functional unknown mitochondrial transporter able to catalyze the efflux of NAG from the matrix will make these cells able to grow in the absence of arginine. The part of the yeast mitochondrial arginine biosynthetic pathway not functional in the model is shaded brown whereas the newly created arginine biosynthetic pathway is shaded blue. argB, argC, argD and argE, E. coli NAG kinase, N-acetyl-glutamyl-5-phospate reductase, N-acetyl-ornithine aminotransferase and Nacetyl ornithine deacetylase, respectively; ARG2, NAG synthase; ARG5,6, N-acetylglutamate kinase and N-acetyl-5-glutamyl-phosphate reductase; ARG8, N-acetyl-ornithine aminotransferase; ARG7, N-acetyl-ornithine acetyltransferase; ARG11, mitochondrial ornithine transporter; ARG3, ornithine carbamoyl transferase; ARG1, argininosuccinate synthetase; ARG4, argininosuccinate lyase; NAGC??, putative mitochondrial NAG transporter.
Taking advantage of the yeast and E. coli twin identical arginine biosynthetic pathways, we placed almost the entire yeast mitochondrial arginine biosynthetic pathway in the cytosol by expressing the bacterial argB-E enzymes. In this new yeast model, the biosynthesis of arginine depends on the efflux of NAG out of the mitochondrial matrix into the cytosol, where NAG is converted by the four bacterial enzymes to ornithine. This arrangement allows the bypassing of the block of mitochondrial synthesis of ornithine caused by the deletion of Arg8p. The expression of a carrier able to export NAG from mitochondria would make this yeast strain able to grow in the absence of arginine, as demonstrated in the yeast model expressing the bacterial enzymes argA-E. Furthermore, the data reported in Figures 3C,F and 4 suggest that the yeast's inner mitochondrial membrane contains endogenous transporters able to transport NAG, although not efficiently, probably when its concentration in the matrix increases due to the downstream block. To find the carrier(s) responsible for the NAG leak, we plan to delete several of yeast mitochondrial transporters, starting from those involved in the dicarboxylates transport [64][65][66]. Knowing the yeast NAG transporter(s) will be instrumental for identifying the putative orthologs in the higher eukaryotic up to humans. Furthermore, similar models could be generated to functionally characterize other mitochondrial carriers involved in the transport of extremely unstable substrates, which may produce artifacts in the transport assays. Figure 5. The yeast cell model generated to identify the putative mitochondrial NAG transporter/s. Yeast cells lacking the ARG8 gene are auxotrophic for arginine. The expression of the E. coli argB-E in the cytosol makes cells dependent on the exit of NAG from the matrix. In this model, the expression of a functional unknown mitochondrial transporter able to catalyze the efflux of NAG from the matrix will make these cells able to grow in the absence of arginine. The part of the yeast mitochondrial arginine biosynthetic pathway not functional in the model is shaded brown whereas the newly created arginine biosynthetic pathway is shaded blue. argB, argC, argD and argE, E. coli NAG kinase, N-acetyl-glutamyl-5-phospate reductase, N-acetyl-ornithine aminotransferase and N-acetyl ornithine deacetylase, respectively; ARG2, NAG synthase; ARG5,6, N-acetylglutamate kinase and N-acetyl-5-glutamyl-phosphate reductase; ARG8, N-acetyl-ornithine aminotransferase; ARG7, Nacetyl-ornithine acetyltransferase; ARG11, mitochondrial ornithine transporter; ARG3, ornithine carbamoyl transferase; ARG1, argininosuccinate synthetase; ARG4, argininosuccinate lyase; NAGC??, putative mitochondrial NAG transporter.
Taking advantage of the yeast and E. coli twin identical arginine biosynthetic pathways, we placed almost the entire yeast mitochondrial arginine biosynthetic pathway in the cytosol by expressing the bacterial argB-E enzymes. In this new yeast model, the biosynthesis of arginine depends on the efflux of NAG out of the mitochondrial matrix into the cytosol, where NAG is converted by the four bacterial enzymes to ornithine. This arrangement allows the bypassing of the block of mitochondrial synthesis of ornithine caused by the deletion of Arg8p. The expression of a carrier able to export NAG from mitochondria would make this yeast strain able to grow in the absence of arginine, as demonstrated in the yeast model expressing the bacterial enzymes argA-E. Furthermore, the data reported in Figure 3C,F and Figure 4 suggest that the yeast's inner mitochondrial membrane contains endogenous transporters able to transport NAG, although not efficiently, probably when its concentration in the matrix increases due to the downstream block. To find the carrier(s) responsible for the NAG leak, we plan to delete several of yeast mitochondrial transporters, starting from those involved in the dicarboxylates transport [64][65][66]. Knowing the yeast NAG transporter(s) will be instrumental for identifying the putative orthologs in the higher eukaryotic up to humans. Furthermore, similar models could be generated to functionally characterize other mitochondrial carriers involved in the transport of extremely unstable substrates, which may produce artifacts in the transport assays.
Funding: This work was supported by a grant from the Italian Ministero dell'Istruzione, dell'Università e della Ricerca (no. 2017PAB8EM to G.F. and E.P.).

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
No new data were produced or analyzed in this study. Data sharing is not applicable to this article.

Conflicts of Interest:
The authors declare no conflict of interest.