N-terminal methionine excision of proteins creates tertiary destabilizing N-degrons of the Arg/N-end rule pathway

All organisms begin protein synthesis with methionine (Met). The resulting initiator Met of nascent proteins is irreversibly processed by Met aminopeptidases (MetAPs). N-terminal (Nt) Met excision (NME) is an evolutionarily conserved and essential process operating on up to two-thirds of proteins. However, the universal function of NME remains largely unknown. MetAPs have a well-known processing preference for Nt-Met with Ala, Ser, Gly, Thr, Cys, Pro, or Val at position 2, but using CHX-chase assays to assess protein degradation in yeast cells, as well as protein-binding and RT-qPCR assays, we demonstrate here that NME also occurs on nascent proteins bearing Met–Asn or Met–Gln at their N termini. We found that the NME at these termini exposes the tertiary destabilizing Nt residues (Asn or Gln) of the Arg/N-end rule pathway, which degrades proteins according to the composition of their Nt residues. We also identified a yeast DNA repair protein, MQ-Rad16, bearing a Met–Gln N terminus, as well as a human tropomyosin–receptor kinase–fused gene (TFG) protein, MN-TFG, bearing a Met–Asn N terminus as physiological, MetAP-processed Arg/N-end rule substrates. Furthermore, we show that the loss of the components of the Arg/N-end rule pathway substantially suppresses the growth defects of naa20Δ yeast cells lacking the catalytic subunit of NatB Nt acetylase at 37 °C. Collectively, the results of our study reveal that NME is a key upstream step for the creation of the Arg/N-end rule substrates bearing tertiary destabilizing residues in vivo.

To understand the impact of Nt-acetylation on the Ubr1mediated degradation of MN-␣2-GST more clearly, N-termi-

MetAPs cleave the Nt-M of MN-␣2-GST and MQ-␣2-GST for degradation
S. cerevisiae contains two MetAPs (Map1 and Map2), which redundantly cut off the Nt-M of nascent proteins, but only if the residue at position 2 is no larger than Val (34). Nonetheless, given the evident involvement of Ubr1, Ate1, and Nta1 in the degradation of MN-␣2-GST or MQ-␣2-GST (Figs. 2 and 3), we presumed that MetAPs might process the Nt-M of the reporters before they are targeted by the Arg/N-end rule pathway (Fig.  4A). To test this possibility, we expressed MN-␣2-GST or MQ-␣2-GST from the P CUP1 promoter on a low-copy-number CEN plasmid in map1⌬ naa20⌬ cells in either the presence or absence of the specific Map2 inhibitor, fumagillin. Tellingly, short-lived MN-␣2-GST and MQ-␣2-GST were strongly stabilized in map1⌬ naa20⌬ cells by fumagillin-mediated Map2 inactivation (Fig. 4, B and C). Altogether, we conclude that MetAPs can noticeably cleave the Nt-M of proteins, even with tertiary destabilizing Asn or Gln present at position 2, thus triggering their degradation by the Arg/N-end rule pathway in vivo (Fig. 4A).

Arg/N-end rule pathway mediates degradation of human MN-TFG
The observation of the Arg/N-end rule-dependent degradation of MN-␣2-GST and MQ-␣2-GST (Fig. 4) post-NME prompted us to search for native MQ/MN-starting proteins that are targeted for degradation by the Arg/N-end rule pathway. Surprisingly (and unexpectedly), our independent yeast two-hybrid (Y2H) screen using a C-terminally truncated human UBR1 1-1031 fragment with the previously defined substrate-binding sites (35) repeatedly isolated a human tropomyosin-receptor kinase-fused gene (TFG) as its binding partner (Fig. 5, A and C). Interestingly, the screened TFG-harboring prey vectors retained a stop codon just upstream of TFG ORF (ORF) following the GAD-coding sequence (Fig. 5A); GAD encodes the Gal4 transcription-activation domain, and native The graphs in C and E show quantitation of data as mean Ϯ S.D. from three independent experiments. F, same as in B but for 0 and 30 min in ubr1⌬ doa10⌬ cells expressing f Ubr1 from the P GAL1 promoter. The yeast cells were grown at 30°C to A 600 ϭ ϳ0.8 in SRaf medium, followed by the addition of either galactose (for UBR1 induction) or glucose (for UBR1 repression). G, same as in F but in naa20⌬ ubr1⌬ doa10⌬ cells expressing vector only, f Ubr1, or catalytically inactive mutant f Ubr1 C1220S . H, extracts of naa20⌬ ubr1⌬ doa10⌬ cells expressing MN-␣2-GST and f Ubr1 or f Ubr1 C1220S were subject to immunoprecipitation with anti-FLAGagarose. The bound proteins were analyzed via SDS-PAGE and immunoblotting with anti-GST or anti-FLAG antibodies. The bottom panel shows the 1% input that was used for immunoprecipitation. I, same as in H but extracts of NAA20 or naa20⌬ S. cerevisiae (ubr1⌬ doa10⌬) cells expressing MN-␣2-GST and f Ubr1 C1220S . J, co-targeting scheme of MN-␣2-GST via the Ubr1-and NatB/Doa10-mediated degradation by the Ub-proteasome system (UPS).

N-terminal Met excision for the Arg/N-end rule pathway
TFG ORF comprises an MN N terminus and an intrinsic transcription-activation domain (36). Hence, we postulated that the resulting MN-TFG would activate the expression of the Y2H reporters through a direct interaction with the Gal4-binding domain-UBR1 1-1031 fusion, even in situations lacking N-terminally positioned GAD (Fig. 5, A and C). To verify this possibility, we further mapped the TFG-binding site of UBR1 by repeating the Y2H assays using the truncated UBR1 derivatives as baits and found that TFG bound to the UBR box (type 1 site)-containing UBR1 1-1031 , UBR1 1-632 , UBR1  , UBR1 93-221 , or UBR1 93-191 , but not to the UBR box lacking UBR1 93-157 and UBR1 167-1031 (Fig. 5, B and C). The identical MN-starting sequences of MN-TFG and MN-␣2-GST suggested that MN-TFG would be subject to NME, followed by Nt-deamidation and subsequent Nt-arginylation before its targeting by UBR1. Indeed, loss of either Nta1 or Ate1 almost completely abrogated the binding of UBR1 to MN-TFG in Y2H assays (Fig. 5D).
Given these results, we sought to examine further whether the Arg/N-end rule pathway mediated the degradation of MN-TFG in mammalian cells by expressing the C-terminal triply FLAG-tagged human MN-TFG (MN-TFG f3 ) in Ate1 ϩ/ϩ WT mouse fibroblast (MEF) cells and Ate1 Ϫ/Ϫ knockout (KO) MEF cells. Upon CHX chases, MN-TFG f3 became short-lived in WT Ate1 ϩ/ϩ MEF cells, but greatly stabilized in Ate1 Ϫ/Ϫ KO MEF cells (Fig. 5, E and F), despite almost no significant changes in the levels of TFG f3 mRNA between Ate1 ϩ/ϩ WT MEF cells and Ate1 Ϫ/Ϫ KO MEF cells (Fig. 5G). Therefore, the observed strong augmentation of the MN-TFG f3 level in Ate1 Ϫ/Ϫ KO

N-terminal Met excision for the Arg/N-end rule pathway
MEF cells most likely arises from alterations in the rate of its initial proteolytic decay (especially in accordance with many previous supporting studies) for the substantial co-translational degradation of nascent proteins (37)(38)(39). Of note, the degradation of MN-TFG f3 also requires the 26S proteasome because the steady-state level of MN-TFG f3 is greatly up-regulated by the presence of the proteasome inhibitor MG-132 (Fig.  5H). Overall, these results suggest that MN-TFG f3 would be degraded by the Arg/N-end rule pathway involving consecutive reactions of NME, Nt-deamidation, as well as subsequent Ntarginylation in mammalian cells.

Arg/N-end rule pathway mediates degradation of S. cerevisiae MQ-Rad16
The positive Y2H interaction of the human UBR box with Nt-arginylated TFG bearing an intrinsic transcription activa-tion domain (without GAD) (Fig. 5, B and C) prompted us to search for Nt-arginylated proteins in S. cerevisiae, in which ϳ9% of DNA-encoded proteins contain an MQ/MN-starting N terminus according to the Saccharomyces Genome Database (https://www.yeastgenome.org/). To this end, we repeated the Y2H screen using a human UBR1 93-151 containing only a UBR box (type-1 binding) site as a bait and a yeast genomic DNA library (as preys) in the ubr1⌬ Y2H strain, in which Nt-arginylated substrates are long-lived because of the ablation of Ubr1, the sole N-recognin of the Arg/N-end rule pathway. The resulting Y2H screen identified Rad16, a 91-kDa nucleotide excision repair protein (denoted MQ-Rad16 hereafter because it starts with an MQ N terminus). MQ-Rad16 also contains its own intrinsic transcriptional activation domain (40). Remarkably, and in agreement with the above observations with MN-TFG (Fig. 5, A-D), the interaction between human UBR1 93-151 and Figure 5. Human MN-TFG is degraded by the Arg/N-end rule pathway. A, schematic representation of human UBR1 1-1031 in Y2H bait vector and human TFG in Y2H prey vectors to be screened. B, UBR box, the N-domain and the RING domain of human UBR1. Fragments of UBR1 used to map its TFG-binding region are depicted below the diagram. C, interaction of TFG with UBR1 fragments upon Y2H assays. D, in vivo detection of UBR1 1-1031 -TFG interactions in ATE1 NTA1, ate1⌬, and nta1⌬ S. cerevisiae using Y2H assay. S. cerevisiae cells co-expressing bait and prey plasmids were serially diluted 5-fold and spotted on SC (ϪLeu/Trp) or SC(ϪLeu/Trp/His/Ade) plates (see "Experimental procedures" for details). E, CHX chases of C-terminal triply FLAG-tagged human TFG (TFG f3 ) in Ate1 ϩ/ϩ WT and Ate1 Ϫ/Ϫ KO MEF cells for 0, 4, 8, and 12 h. F, graph represents quantitation of data in E with mean Ϯ S.D. of three independent experiments. G, relative levels of human TFG f3 mRNAs in WT and Ate1 Ϫ/Ϫ KO MEF cells using RT-qPCR. The data presented are mean Ϯ S.D. in triplicate for each sample. H, steady-state levels of TFG f3 in HeLa cells with or without the proteasome inhibitor MG132.

N-terminal Met excision for the Arg/N-end rule pathway
MQ-Rad16 was abolished by the loss of either Ate1 or Nta1 upon Y2H assays (Fig. 6A).

Ablation of the Arg/N-end rule pathway ameliorates the growth defect of naa20⌬ cells at 37°C
Given that Nt-acetylation competitively inhibits not only the activity of MetAPs, but also the recognition of substrate proteins by Ubr1 (5, 6), we presumed that the loss of NatB (Naa20) Nt-acetylase would trigger the NME-mediated degradation of a vast range of MN/MQ-starting proteins via the Arg/N-end rule pathway, thus increasing the susceptibility of yeast cells to specific stresses. Indeed, the growth of naa20⌬ S. cerevisiae is hypersensitive to pleiotropic stressors, including heat, salts, drugs, ultraviolet light, etc. (41). Strikingly, we observed that the absence of Nta1 substantially rescued the defective growth of naa20⌬ cells at 37°C (Fig. 7A). Moreover, the lack of either Ate1 or Ubr1 more profoundly suppressed the defective growth of naa20⌬ cells at 37°C than that of Nta1 (Fig. 7A), suggesting the possibility that NME may also cause degradation of MD/ME-starting proteins, which bypass Nta1 for degradation via the Arg/N-end rule pathway. Indeed, this is in agreement with our recent observation that degradation of some reporters starting with the MD N terminus requires Ubr1 in naa20⌬ cells (12). Collectively, these results indicate that the previously unexplained slow-growth phenotype of naa20⌬ cells at 37°C (41) may arise, in part, from low levels of Nt-unacetylated MN/MQ-starting proteins because of their vulnerable degradation at higher temperatures via the Arg/N-end rule pathway (Fig. 7B).

Discussion
Because of the strong processing activity of MetAPs toward Nt-M bearing small residues at position 2, NME is assumed to yield only stabilizing Nt-residues (Nt-G, -A, -S, -C, -T, -V, or -P) of the Arg/N-end rule pathway in S. cerevisiae (1,9). As opposed to the typical substrate specificities of MetAPs, this study shows that MetAPs can produce the tertiary destabilizing residues Asn or Gly of the Arg/N-end rule pathway by removing the Nt-M of native proteins (MQ-Rad16 and MN-TFG) and model substrates in vivo (MN-␣2-e K -ha-Ura3, MN-␣2-GST, and MQ-␣2-GST) (Figs. 1-6).
In line with our present findings, global proteomic analyses bear out that the Asn or Gln at the 2nd position in some native proteins is exposed to the N terminus without the retention of the initiator Met (42). Additionally, S. cerevisiae Map1 processes, to an inefficient but significant extent, the Nt-M of a synthetic peptide with a MN N terminus in vitro (43,44). Knop and co-workers (45) also reported, using multiplexed protein stability profiling for the quantitative and systematic mapping of degrons in the yeast Nt-proteome, that 10 MN-starting reporters and an Nt-truncated isoform of the mannosyltransferase MN-Ktr2 (with Met-Asn N terminus) may involve Map1-mediated NME for the Arg/N-end rule pathway.

N-terminal Met excision for the Arg/N-end rule pathway
Pro-Asp-Gly-Thr (MQLSIIDPDGT)-e K -ha-Ura3 (23Q-Ura3) with Gln at position 2, whose degradation requires Nta1, Ate1, and Ubr1. However, the absence of either Map1 or Map2 does not significantly affect the degradation of 23Q-Ura3, suggesting the overlapping action of MetAPs or unknown dedicated aminopeptidases (46). Furthermore, Finley and co-workers (47) isolated an MH-starting PB12 reporter that undergoes NME, thereby exposing the penultimate His (a primary destabilizing residue of the Arg/N-end rule pathway) at its N terminus. Of note, the Ubr1-mediated degradation of the PB12 reporter is also unaffected in map1⌬ cells. Therefore, the possibility cannot be excluded that not only MetAPs but also as yet unknown aminopeptidases can participate in the NME of MN/MQstarting proteins according to their Nt-sequence context or structural conformation.
The present study overtly reveals NME as another critical process of the Arg/N-end rule pathway in the creation of tertiary destabilizing residues by demonstrating that Ubr1-mediated recognition of MN/MQ-starting proteins necessitates consecutive reactions of NME, Nt-deamidation, and subsequent Nt-arginylation (Figs. 2-6 and 7B). Accordingly, the resulting NME enormously increases the number of Arg/N-end rule substrates, considering that MN/MQ-starting proteins encompass about 9% of all nuclear DNA-encoded proteins. Nonetheless, it should be noted that the previously overlooked or underestimated requirement of NME in the degradation of MN/MQ-starting substrates by the Arg/N-end rule pathway most likely stems from two redundant (overlapping) MetAPs (Map1 and Map2) and the alternative proteolytic route by the Ac/N-end rule pathway, because the ablation effects of either single MetAP or the Arg/N-end rule pathway can be efficiently suppressed by the other MetAP or the Ac/N-end rule pathway, respectively.
Notably, Nt-acetylation confers the opposite impacts on protein stability, particularly because it not only creates Ac/N-degrons of the Ac/N-end rule pathway, but also concomitantly precludes the destruction of proteins by the Arg/N-end rule pathway (6,7,9). Hence, many MN/MQ-starting proteins are susceptible to the NME-mediated Arg/N-end rule pathway, and also, alternatively, to the Ac/N-end rule pathway through their Nt-acetylated Met, as demonstrated in the degradation patterns of MN-␣2-e K -ha-Ura3, MN-␣2-GST, MQ-␣2-GST, or MQ-Rad16 ha (Figs. 1-3 and 6).
In addition and analogous to our previous finding that Ntacetylation transforms an M⌽ (a hydrophobic residue)/N-degron into an AcM⌽/N-degron, thus switching the targeting route of an M⌽-starting protein to the Ac/N-end rule pathway (Fig. S1C) (6, 13, 48), the NatB-mediated Nt-acetylation of MN/MQ-starting proteins would not only preclude their NME-mediated degradation by the Arg/N-end rule pathway, but also provoke their alternative proteolytic route via the Ac/N-end rule pathway. Consequently, dual or alternative targeting of MN/MQ-starting proteins by both the Arg/N-end rule pathway and the NatB-mediated Ac/N-end rule pathway would degrade these proteins cooperatively, irrespective of their Nt-acetylation states (Fig. 7B). Indeed and in a similar outcome to that of M⌽/N-degron-containing proteins (13), MN/MQ-starting proteins, such as MN-␣2-GST, MQ-␣2-GST, and MQ-Rad16 ha , were more strongly stabilized in the double mutant ubr1⌬ naa20⌬ cells than in the single mutant ubr1⌬ or naa20⌬ cells (Figs. 2, 3, and 6), indicating the co-targeting of MN/MQ-starting substrates by the Arg/N-end rule The Ac/N-end rule pathway directly targets Nt-acetyl tag of Nt-acetylated AcMN/AcMQ-starting proteins. However, Nt-unmodified but otherwise identical MN/MQ-starting proteins are vulnerable to NME and thereby degraded by the cascades of the Arg/N-end rule pathway.

N-terminal Met excision for the Arg/N-end rule pathway
and the NatB-mediated Ac/N-end rule pathways. However, the actual targeting mechanism of MN/MQ-starting proteins by Ubr1 is substantially different from that of M⌽-starting proteins. In particular, Ubr1 recognizes MN/MQ-starting proteins after their preliminary modifications (consisting of NME, Ntdeamidation, and subsequent Nt-arginylation) in contrast to its direct binding to the nonacetylated Nt-M of M⌽-starting proteins (Fig. 7B) (6, 13, 48).
It is also noteworthy that Nt-acetylation also modulates the activities of E3 Ub ligases, 26S proteasomes, and molecular chaperones (6). Furthermore, the antagonistic (triggering and restraining) impacts of Nt-acetylation on the Arg/N-end rule and the Ac/N-end rule pathways, respectively, increase the complexities of intracellular protein degradation more profoundly than those previously assumed, thereby making it particularly difficult to identify or predict the specific proteolytic pathway of a given protein using specific genetic studies harnessing a single gene deletion or knockdown, etc. (6,7,48).
Deg1 represents the first 67 residues of Mat␣2 (49). In this study, we also demonstrated here that MN-␣2-GST (Deg1-GST) was significantly and substantially stabilized in naa20⌬ cells that lacked a catalytic subunit of NatB Nt-acetylase (Fig.  2B) in agreement with previous observations of MN-␣2-e K -ha-Ura3 (Deg1-e K -ha-Ura3) and MN-␣2-Leu2 M1⌬ (another Deg1 fusion protein) (8,50). Conversely, it is also reported that the loss of Naa20 very weakly affects the degradation of endogenous Mat␣2, as well as other Deg1 fusions such as MN-␣2-FLAG-Ura3 (Deg1-FLAG-Ura3) (49). The cited study interpreted these results as an indication that Nt-acetylation has little effect on the recognition of Doa10 for its substrates (49), in contrast to our previous finding that Doa10 works as an Ac/Nrecognin (8). Although these discrepancies in the Nt-acetylation-dependent degradation of Deg1 fusions remain to be further examined, we infer that the shifting rate or extent of the NatB-mediated Ac/N-end rule pathway to the NME-mediated Arg/N-end rule pathway may cause distinct proteolytic outcomes, particularly according to the sequence context of Deg1 substrates or naa20⌬ strains that are used in our studies and those of others (8,49,50).
Strikingly, and with conceptual similarity to the results with naa20⌬ cells, dfm1⌬ cells (lacking the rhomboid derlin Dfm1) are reported to have very dissimilar proteolytic patterns (51). For instance, some studies do not detect any defects in the degradation of ER-associated protein degradation (ERAD) substrates in dfm1⌬ cells (52,53), whereas others reveal the significant stabilization of a subset of ERAD substrates in the absence of Dfm1 (54,55). More recently, Hampton and co-workers (51) resolved the controversial role of Dfm1 in ERAD by demonstrating that dfm1⌬ cells rapidly assume suppression and thereby compromise ERAD by up-regulating the alternative proteolytic pathway.
Likewise, in naa20⌬ cells, the ablation of the NatB-mediated Ac/N-end rule pathway would rapidly trigger the other NME/ NatB-mediated Arg/N-end rule pathway, thus exhibiting the observed comparable degradation of Deg1 fusions in both WT cells and naa20⌬ cells (49), which remains to be tested. Furthermore, our most recent genetic and biochemical experiments reveal that TEB4 (a mammalian homolog of Doa10) (23) more preferentially binds to Nt-acetylatable native proteins, such as RGS2 and PLIN2, than to their nonacetylatable counterparts, which is in agreement with our previous identification of TEB4 (Doa10) as an Ac/N-recognin (26).
This study also demonstrates that human TFG is a substrate of the NME-mediated Arg/N-end rule pathway. Endogenous TFG is involved in the spatial coordination of the early secretory event from endoplasmic reticulum to Golgi by binding to the coat protein complex II (COPII) (56).
Remarkably, the TFG gene is frequently identified as a chromosomally translocated chimeric gene with many oncogenes, such as NTRK1 (a neurotrophic receptor tyrosine kinase) in thyroid papillary carcinomas, ALK (anaplastic lymphoma kinase), NOR1 (a nuclear orphan receptor), NEMO (NF-B essential modulator), TANK (TRAF-associated NF-B activator), TEC (translocated in extraskeletal genes in some cancer cells), etc. (36,(57)(58)(59). Interestingly, the resulting TFG-fusion oncogenic proteins contain TFG primarily at their N-terminal region. Therefore, the present finding that the MetAPs-mediated Arg/N-end rule pathway degrades MN-TFG suggests that the N-terminal TFG of these oncogenic fusion proteins would act as a portable degron for the NME-mediated Arg/N-end rule pathway. As a result, the TFG-mediated dysregulation of oncogenic proteins most likely promotes tumorigenesis. Elucidating whether Nt-fused TFG can trigger the degradation of oncogenic fusion proteins via the MetAPs-mediated Arg/N-end rule pathway in malignant cancer cells is therefore of great interest.

Plasmid construction
To construct pCH1591 that expressed human UBR1 1-1031 from a Y2H expression bait vector, human UBR1 1-1031 DNA was PCR-amplified from pCH432 carrying human UBR1 cDNA using the primers OCH5001/OCH5003, digested with SfiI/SalI, and then cloned into SfiI/SalI-cut pGBKT7. To create pCH5003 that expressed TFG f3 , the human TFG gene was PCR-amplified from pCH5131, which was derived from the Y2H screen, using the primer pair OCH5017/OCH5018. The resulting PCR fragment was digested with BamHI/XbaI and inserted into BamHI/XbaI-cut pcDNA3.1(ϩ). To construct pCH5054 that expressed MQ-Rad16 ha from P CUP1 promoter on a low-copy-number pRS316 vector, RAD16 was PCR-amplified from yeast genomic DNA using the primer pair OCH5067/ OCH5069, digested with SpeI/SacI, and then ligated into SpeII/ SacI-cut pCH692.
Construction details for other plasmids are available upon request. All constructed plasmids were verified via DNA sequencing.

CHX-chase assays of protein degradation
S. cerevisiae cells were cultured to a 600 nm absorbance value (A 600 ) of Ϸ1.0 in YPD or SC media at 30°C and then treated with CHX (at a final concentration of 0.2 mg/ml). Cell samples (equivalent to 1 ml of cell suspension at an A 600 of 1) were collected at the indicated times via centrifugation for 2 min at  Lab collection pCH1220 p424GAL1, 10 Lab collection pCH1525 Ub R48 -MN-␣2 3-67 -GST in p314CUP1 This study pCH1527 Ub R48 -MN-␣2 3-67 -GST in p314CUP1 This study pCH1531 Human TFG in pACT2 This study pCH1591 Human UBR1

N-terminal protein sequencing by Edman degradation
Approximately 5 g of MN-␣2-GST protein partially purified from naa20⌬ ubr1⌬ doa10⌬ cells was separated on a Tris-SDS-10% polyacrylamide gel at 80 V for 90 min in SDS-PAGE running buffer. After electrophoresis, the gel was equilibrated in CAPS transfer buffer (10% methanol, 10 mM CAPS, pH 11) for 10 min before electroblotting onto a PVDF membrane (Immobilon P SQ , Millipore, Billerica, MA). Electroblotting was performed at 80 mA overnight at 4°C. The PVDF membrane was washed in distilled water for 10 min, stained with Coomassie Blue R-250 (0.1% R-250 in 50% methanol) for 10 min, and then distained twice for 15 min in a destaining buffer (50% methanol, 10% acetic acid). The relevant protein band of MN-␣2-GST was cut and analyzed by Edman degradation using a model 492 cLC procise protein micro-sequencer (Applied Biosystems, GmbH) at the Protein Sequencing Laboratory (Seoul, South Korea).

Total RNA extraction and real-time RT-qPCR
For extraction of total RNA, 1 ϫ 10 5 Ate1 ϩ/ϩ WT MEF cells or Ate1 Ϫ/Ϫ KO MEF cells were seeded onto a 12-well plate in DMEM, 10% FBS plus streptomycin/penicillin on day 1, and then transfected with 0.5 g of PCH5003 and 3 l of PEI. After a 48-h incubation, total RNA was extracted using an RNeasy mini kit (74104, Qiagen, Germantown, MD), according to the manufacturer's protocol. Five hundred ng of total RNAs were converted into cDNA using a TOPscript TM cDNA synthesis kit (Enzynomics, EZ005S, South Korea) in 20-l reactions. Ten ng of cDNA from each sample were used for quantitative real-time RT-qPCR using a StepOnePlus Real-Time System (Thermo-Fisher Scientific), and Power SYBR Green PCR primer pairs OCH6874/OCH6875 for human TFG and OCH8104/ OCH8105 for ACTB (encoding ␤-actin) were designed using NCBI Primer-BLAST. The RT-qPCR cycles were as follows: 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and Table 3 Examples of PCR oligomers used in this study

Statistical analysis
To calculate significant differences (p values), two-tailed paired Student's t tests were used through Microsoft Excel 2016. A p value of Ͻ0.05 was considered statistically significant.
All the values are presented as mean Ϯ S.D.