The MicroRNA 15a/16–1 Cluster Down-regulates Protein Repair Isoaspartyl Methyltransferase in Hepatoma Cells

Background: Protein carboxyl-O-methyltransferase (PCMT1) repairs isomerized proteins and prevents apoptosis. Results: PCMT1 is down-regulated by microRNAs 15a/16–1. Its silencing results in conspicuous BclxL isomerization and increased susceptibility to apoptosis. Conclusion: MicroRNA 15a/16–1 cluster can modulate PCMT1: a repair methyltransferase with antiapoptotic activity. Significance: PCMT1 may act as an effective modulator of apoptotic cell death, with a potential role in proliferative/degenerative disorders. Asparaginyl deamidation, a spontaneous protein post-biosynthetic modification, determines isoaspartyl formation and structure-function impairment. The isoaspartyl protein carboxyl-O-methyltransferase (PCMT1; EC 2.1.1.77) catalyzes the repair of the isopeptide bonds at isoaspartyl sites, preventing deamidation-related functional impairment. Protein deamidation affects key apoptosis mediators, such as BclxL, thus increasing susceptibility to apoptosis, whereas PCMT1 activity may effectively counteract such alterations. The aim of this work was to establish the role of RNAi as a potential mechanism for regulating PCMT1 expression and its possible implications in apoptosis. We investigated the regulatory properties of the microRNA 15a/16–1 cluster on PCMT1 expression on HepG2 cells. MicroRNA 15a or microRNA 16–1 transfection, as well as their relevant antagonists, showed that PCMT1 is effectively regulated by this microRNA cluster. The direct interaction of these two microRNAs with the seed sequence at the 3′ UTR of PCMT1 transcripts was demonstrated by the luciferase assay system. The role of PCMT1 down-regulation in conditioning the susceptibility to apoptosis was investigated using various specific siRNA or shRNA approaches, to prevent non-PCMT1-specific pleiotropic effects to take place. We found that PCMT1 silencing is associated with an increase of the BclxL isoform reported to be inactivated by deamidation, thus making cells more susceptible to apoptosis induced by cisplatinum. We conclude that PCMT1 is effectively regulated by the microRNA 15a/16–1 cluster and is involved in apoptosis by preserving the structural stability and biological function of BclxL from deamidation. Control of PCMT1 expression by microRNA 15a/16–1 may thus represent a late checkpoint in apoptosis regulation.

This article has been retracted by the publisher. The actin immunoblots in Figs may be also endowed with the ability of regulating partitioning of certain proteins between two possible structures: (Asx)-containing (active) or isoAsp (deamidated and inactive). Co-transfection with a PCMT1 carrying vector prevents apoptosis induced by Bax, in neuronal cell lines (20). Moreover, Bcl xL , an antiapoptotic member of the Bcl2 protein family, contains two labile asparaginyl sites, which are deamidation-prone (i.e. positions 52 and 66). Deamidated Bcl xL is no longer active to block pro-apoptotic proteins, thus leading to cell death (4,21,22,24). In fact, cell damage induces Bcl xL deamidation as a mechanism to commit cells to apoptosis. During cell stress a transient intracellular alkalinization takes place causing Bcl xL deamidation/ isomerization, as alkaline pH conditions increase the tendency of labile asparaginyl residues to deamidate (25). Consistently, reduced levels of Bcl xL deamidation were detected in hepatocellular carcinomas compared with normal liver tissue and this has been linked to the typical apoptosis resistance of such transformed cells (26). In our previous work (27) we demonstrated the antiapoptotic action of PCMT1 and proposed a mechanism of action, based on the ability of this methyltransferase to maintain the structural stability of some crucial antiapoptotic proteins, including Bcl xL , through methylation and repair of their malfunctional L-isoaspartyl-containing derivatives.
Over the last decade, microRNAs have been involved in a number of cell processes, including differentiation and development as well as tissue reshaping and apoptosis. In particular, the microRNA 15a/16 -1 cluster is involved in the regulation of key antiapoptotic mediators, such as Bcl2 (28,29). In consideration of the role of both apoptosis and protein deamidation in tissue reshaping during liver disease, we investigated microRNA-dependent PCMT1 regulation in hepatocarcinoma cells, particularly as related to microRNA 15a/16 -1 (26,30). In particular, we investigated the role of microRNAs on PCMT1 expression and found evidence that this methyltransferase is a target of the microRNA 15a/16 -1 cluster. Selective down-regulation of PCMT1, also induced by various siRNA or shRNA, on HepG2 cell cultures treated with cisplatinum, was utilized to establish the biomolecular basis for the role of PCMT1 as a part of a novel apoptosis regulatory mechanisms.
siRNA and shRNA Transfection-HepG2 cells were transiently transfected with 1 g/ml (final concentration) of shRNA constructs against PCMT1 in pGFP-V-RS vector (OriGene, Rockville, MD) or with 150 ng of target-specific siRNAs designed to knockdown PCMT1 gene expression (Qiagen), using HiPerFect Transfection Reagent (Qiagen) according to the manufacturer's protocol. After 96 h cells were harvested and analyzed.
RNA Extraction-RNA extraction was performed on HepG2 cells using TRIzol reagent (Invitrogen). Briefly, after removing cell culture medium, cells were washed twice with PBS and lysed with 1 ml of TRIzol. Subsequently, RNA was extracted by adding chloroform and subsequent centrifugation and precipitation with isopropyl alcohol. Pellets were washed in 70% ethanol and suspended in H 2 O/diethyl pyrocarbonate. The RNA concentration was measured by NanoDrop UV/Visible microspectrophotometry (ND-1000; NanoDrop Technologies, Wilmington, DE).
cDNA Synthesis-In preparation of real time PCR, RNA retrotranscription into cDNA was accomplished, starting from total RNA extracted from cell cultures. cDNA synthesis was carried out utilizing the QuantiTect reverse transcriptase kit (Qiagen), utilizing 1 g of total RNA.
Analysis of PCMT1 Gene Transcription Levels through Real Time PCR-Amplification reactions were performed with an iCycler thermalcycler (Bio-Rad) using the fluorescence detection system iCycler iQ Real time PCR. The amplification mixture contained 1 l of cDNA, 0.3 M of each primer, 12.5 l of Master mix QuantiTect SYBR Green (Qiagen), and H 2 O/diethyl pyrocarbonate for a final volume of 25 l. GAPDH (as internal reference) primers were: GAPDH sense, 5Ј-TTGG-TATCGTGGAAGGACTCCATG-3Ј, and GAPDH antisense, 5Ј-CAGTAGAGGCAGGATGATGTTC-3Ј.
The QuantiTect Primer Assay mixture (Qiagen), containing specific sense and antisense primer pairs, was utilized for PCMT1 and MAPK. Amplifying conditions were: 95°C for 15 min, followed by 35 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s. Expression levels after treatment were evaluated by comparison of samples treated with siRNAs or shRNAs with control samples, transfected with scrambled sequence siRNA or shRNA. All qPCR results were obtained from the mean of triplicate measurements for the gene of interest in each sample, normalized for the mRNA levels of the housekeeping reference gene (GAPDH, also amplified in triplicate). Relative expression was calculated using the ⌬C t method. A value of 2 Ϫ⌬⌬ Ct Ͼ1 reflects increased expression of the relevant gene, whereas a value of 2 Ϫ⌬⌬ Ct Ͻ1 points to a decrease in gene expression. Cell Extracts-After transfection, cells were trypsinized, washed twice in PBS, and resuspended in a lysis buffer containing 10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 10% glycerol, and 1ϫ protease inhibitor mixture (Roche Diagnostics) for 30 min at 4°C. Lysates were then centrifuged at 13,000 ϫ g for 20 min at 4°C and the supernatants were stored at Ϫ20°C. Protein concentration was determined according to Bradford (31).
Western Blotting-A loading buffer of 2 M Tris-HCl, pH 6.8, 20% SDS, 5% ␤-mercaptoethanol, glycerol, and bromphenol blue was added to transfected HepG2 cell aliquots, containing 20 g of total proteins and incubated for 5 min at 95°C to allow protein denaturation. SDS-PAGE was performed according to Laemmli, utilizing 12% resolving gel and running buffer containing 192 mM glycine, 25 mM Tris, and 0.1% SDS. After electrophoresis, proteins were transferred to nitrocellulose membrane (Hybond-Extra, GE Healthcare), utilizing a TransBlot system apparatus (Bio-Rad) at 100 V for 1 h using as transfer buffer 20% methanol and 5% Tris glycine. Nitrocellulose membrane was then colored with Ponceau Red for 5 min, de-colored, washed with distilled water, and incubated overnight at 4°C with TBS solution containing 0.1% Tween 20, 5% blotting grade blocker Non-fat Dry Milk (Bio-Rad). Membranes were then incubated with a primary monoclonal antibody against the protein of interest: anti-Caspase-3 (C9598 SIGMA), anti-PARP (AF600NA, R&D System), anti-Bcl xL (610211, BD Transduction Laboratories), anti-PCMT1 (610772, BD Transduction Laboratories), and anti-ACTIN as a loading control (sc-1616, Santa Cruz Biotechnology). The membrane was then washed with TBS solution and incubated with secondary antibody conjugated with horseradish peroxidase. Immunocomplex visualization was obtained through chemiluminescence, utilizing ECL Plus kit (GE Healthcare). Western blot images were acquired through Adobe Photoshop software.
MicroRNA Target Prediction-Bioinformatic prediction of microRNA targets was made by using the following algorithms: TargetScan that predicts biological targets of miRNA by searching for the presence of conserved 8-mer and 7-mer sites that match the seed region (nucleotides 2-7 at the 5Ј end) of each microRNA, PicTar that predicts microRNA targets by searching for pairwise alignments that are conserved across species, and miRanda that calculates free energy from microRNA:mRNA heteroduplex formation.
Statistical Analysis-An unpaired Student's t test was performed to compare means from results of silencing experiments. All results are presented as the mean Ϯ S.D. All experiments were done in triplicate.
3Ј Rapid Amplification of cDNA Ends (RACE)-3Ј-RACE was utilized to clone a cDNA molecule containing the PCMT1 seed sequence, and was carried out by the 3Ј RACE System (Invitrogen) according to the manufacturer's protocol.
Briefly, first strand cDNA synthesis was initiated at the poly(A) tail of PCMT1 mRNA using an oligo(dT)-adapter primer and PCMT1 mRNA was converted into cDNA using reverse transcriptase. Specific cDNA was then amplified by PCR, using a PCMT1-specific primer and an adapter primer (AUAP) that targeted the poly(A) tail region. The AUAP is designed to synthesize first strand cDNA from all polyadenylated mRNAs. A certain sequence specificity in the amplification reaction was therefore provided solely from the specific primer. For this reason, a second PCMT1 nested primer was utilized, together with the adapter primer to further amplify the product of the first PCR, in a second amplification reaction, to yield a second highly specific amplification product of the 3Ј RACE procedure. Primer sequences used were: adapter primer (AP), 5Ј-GGCCACGCGTCGACTAGTACTTTTTTTTTTT-TTTTTT-3Ј; AUAP primer, 5Ј-GGCCACGCGTCGACTAG-TAC-3Ј; GSP-PCMT: 5Ј-ACAGTGGATTGCTCATCTCAG-3Ј; GSP-PCMT nested: 5Ј-CTAGTCTAGATGGATAAC-ACCACCATTCAAG-3Ј.
The final PCR product containing the PCMT1 seed sequence was then purified, using the GenElute Gel Extraction kit (Sigma) and sequenced at the S.U.N. DNA sequencing Facility (directed by Prof. Vincenzo Nigro), using an ABI PRISM DNA System (Applied Biosystems).
Dual Luciferase Reporter Assay-For luciferase reporter experiments, a PCMT1 3Ј UTR segment of 480 bp was amplified by PCR 3Ј RACE from human cDNA and inserted, between the XbaI/SalI sites (Takara Bio Inc., Otsu, Shiga, Japan), into a pGL3 control vector with a TK promoter (kindly provided by Dr. Giulio Piluso, Dept. of General Pathology and Oncology, S.U.N.), immediately downstream of the stop luciferase codon. Luciferase activity was measured 24 h after transfection, using a Dual Luciferase Reporter Assay System (Promega, Bergamo, Milan, Italy). Relative luciferase activity was calculated by normalizing the firefly luminescence to the Renilla luminescence. The experiments were performed in triplicate.
Quantitation of PCMT1 Activity in Cells Silenced with siRNA and shRNA-PCMT1 specific activity was quantitated by means of the vapor diffusion radiochemical assay described by Macfarlane (32).
Mass Spectral Analysis of Deamidated Alkali-treated Bcl xL Band-Native Bcl xL (R&D System) was incubated in alkaline buffer (55 mM ammonium bicarbonate, pH 10.0) for various times at 37°C. To optimize trypsin hydrolysis, the pH was then adjusted to 8.5, by addition of 150 l of ammonium bicarbonate (Ambic). Trypsinization was accomplished overnight at 37°C, after addition of trypsin (Roche Diagnostics) in a 1:50 ratio to protein substrate. Samples were then lyophilized and redissolved in 55 mM ammonium acetate, pH 8.0, containing endoproteinase Asp/N (Roche Diagnostics) in a 1:50 ratio with substrate, following overnight incubation at 37°C. The reaction was stopped by lyophilization and samples were resuspended in Milli-Q water and desalted by the ZipTip method (Millipore). 1 l of this mixture was mixed with 1 l of a ␣-cyano-4-hydroxycinnamic acid matrix (Sigma). Matrix-assisted laser desorption mass ionization (MALDI/TOF) was performed by utilizing a MALDI TOF-TOF 4800 instrument (Applied Biosystems, Carlsbad, CA).
In Vitro Enzymatic Methylation of Alkali-treated Bcl xL -Aliquots of alkali-treated Bcl xL were directly lyophilized and resuspended in the PCMT1 assay buffer and assayed for methylation by human recombinant PCMT1 (33). Sample aliquots to be subject to further SDS-PAGE and Western blot analysis were incubated in the presence of unlabeled S-adenosylmethionine as previously described (33) and analyzed by electrophoresis. Parallel sample aliquots were assayed for PCMT1, in the presence of S-adenosyl-[methyl-14 C]methionine to evaluate the presence of the isoAsp residues, using human recombinant PCMT1 and the vapor diffusion radiochemical assay as previously described (33).

PCMT1 Is a Target for MicroRNA 15a/16 -1 Cluster-Mi-
croRNA are one of the most important post-transcriptional regulators of several genes involved in key cell functions including apoptosis. To investigate the potential regulatory properties of microRNAs on PCMT1 expression, we first checked whether PCMT1 could be predicted as a possible microRNA target, by using bioinformatic analysis. To this purpose we employed three of the most widely utilized algorithms (TargetScan, Pic-Tar, and MiRanda), which are based on different prediction criteria and, therefore, used in combination, provide the highest probability of target identification. We identified a perfect complementarity (seed region) between microRNA 15a/16 -1 cluster sequences and the 3Ј UTR of human PCMT1 (Fig. 1A). This seed sequence, representing the potential recognition region for microRNA 15a/16 -1 on PCMT1 3Ј UTR, is also highly conserved in other species (Fig. 1B).
To elucidate the role of microRNA 15a/16 -1 cluster in regulating PCMT1 expression, parallel cell samples were transfected with pre-microRNA 15a/16 -1 or anti-microRNA 15a/ 16 -1 to verify their ability to influence PCMT1 expression. PCMT1 transcript levels were monitored by real time PCR (data not shown) and the relevant protein product by Western blot analyses.
In particular, we transfected HepG2 with pre-microRNA 15a/16 -1 at various concentrations, thus monitoring PCMT1 expression at various times (48, 72, and 96 h). Figs. 2 and 3, respectively, show that pre-microRNA 15a and pre-microRNA 16 -1 are both effective in down-regulating PCMT1, with a comparable efficiency. Treatment with pre-microRNA 15a To demonstrate a specific interaction between the PCMT1 transcript and microRNA 15a/16 -1 cluster, we used a luciferase reporter assay approach. According to this strategy we cloned the putative seed sequence, identified through bioinformatic analysis (Fig. 1A), immediately downstream of the stop codon of the luciferase reporter gene in a pGL3TK vector. HepG2 were co-transfected with this construct (pGL3TKpcmt) and pre-microRNA 15a or 16 -1. In cells co-transfected with pGL3TKpcmt and microRNA 15a/16 -1 the firefly luciferase activity was significantly reduced, whereas in both controls (cells co-transfected with pGL3TKCT and microRNA 15a/ 16 -1 and cells co-transfected with pGL3TKpcmt and FIGURE 1. Bioinformatic prediction of putative miRNA targeting on PCMT1 transcripts. Prediction was made using three algorithms in combination, as described under "Experimental Procedures." Results with the three approaches were highly consistent (not shown). Only the microRNA with the highest scores are shown. A, putative seed sequence for microRNA 15a and 16 -1 at PCMT1 3Ј UTR. B, evolutionary conservation of the above PCMT1 seed sequence. Hsa, human; Ptr, chimpanzee; Mml, rhesus; Rno, rat; Eeu, hedgehog; Cfa, dog; Eca, horse; Bta, cow; Dno, armadillo; Laf, elephant; Ete, tenrec; Mdo, opossum; Oan, platypus; Aca, lizard. DECEMBER 23, 2011 • VOLUME 286 • NUMBER 51 JOURNAL OF BIOLOGICAL CHEMISTRY 43693 microRNA scrambled) firefly luciferase activity was unchanged with respect to control (Fig. 6). Results showed the occurrence of specific, mutual interactions between microRNA 15a/16 -1 and the PCMT1 putative seed sequence, which cause a reduction of firefly luciferase activity expressed by the pGL3TKpcmt recombinant plasmid.

Targeting of Protein Repair Methyltransferase by RNAi
We could conclude that transcripts encoding for PCMT1 are effectively targeted by the microRNA 15a/16 -1 cluster. Therefore the effects we observed in PCMT1 expression upon microRNA 15a/16 -1 transfection of HepG2 occurred at the specific interaction site of such microRNAs with the PCMT1 transcript actually representing the seed sequence we previously identified in silico. PCMT1 Is Effectively Silenced by Various siRNA and shRNA Approaches-Although the potential role of microRNA 15a/ 16 -1 in PCMT1 expression was thus established, to study the biological meaning of PCMT1 silencing in our model, a different approach, based on a more selective way of switching off the expression of the gene, was undertaken. To assess the direct effects of PCMT1 silencing on apoptosis, we designed an approach possibly devoid of pleiotropic effects on other apoptosis mediators, as well as off-target effects on genes structurally unrelated to any PCMT1 substrates. Selective down-regulation of PCMT1 on HepG2 cells treated with cisplatinum, a well known apoptosis inducer, was performed by various siRNA and short hairpin RNA (shRNA).
Experimental conditions to trigger the very initial stages of apoptosis were explored, so that cells would not be irreversibly committed to programmed death, and allowing us to still detect and appreciate even small variations of cell response, possibly due to PCMT1 down-regulation. We then preliminarily tried various conditions including treatments with cisplatinum for various times (data not shown), so that we finally set treatment conditions of HepG2 with cisplatinum at 35 mM for 6 h (condition A), as shown in Fig. 7A. Under condition A, we could observe an initial activation of apoptosis cascade, in that PARP cleavage appeared to be just starting, compared with the pattern detected at the 18-h  Therefore parallel samples of HepG2 cells were treated with cisplatinum, under condition A, and then transfected with either siRNA designed against 3Ј and 5Ј UTRs of the PCMT1 transcript or with four different pGFP-V-RS vectors carrying shRNA constructs against the 5Ј UTR and ORF of PCMT1. Quantitation of PCMT1 transcription levels after transfection was estimated by real time PCR, showing that both siRNAs (siRNApcmt3 and siRNApcmt5; Fig. 7B) and all the shRNAs (shRNAvs5ЈUTRpcmt(a), shRNAvs5ЈUTRpcmt(b), shRNAv-sORFpcmt(a), and shRNAvsORFpcmt(b)) (Fig. 7C), were almost equally highly effective in switching the PCMT1 gene off within 96 h of treatment (p Ͻ 0.01). The high consistency of the results with various RNAi approaches allowed us to rule out the possibility of any off-target effects occurring under such experimental conditions. We also measured PCMT1 enzyme activity in cells upon its gene silencing using a radiochemical methyltransferase assay (32) (Fig. 8). All siRNAs (Fig. 8A) and shRNAs (Fig. 8B) employed were able to cause a significant reduction of PCMT1 activity, in HepG2 cells transfected with the constructs, by at least 80% (p Ͻ 0.05).
Characterization of Isomerized and Methylated Bcl xL Products with Respect to Their Electrophoretic Behavior-Several reports from the literature described the Bcl xL shift on SDS-PAGE as the result of "deamidation" of this protein (21,25,26). However, it should be pointed out that, taking into account the mechanism for asparaginyl deamidation (1, 2), its most abundant product, in most proteins, is in fact the isoaspartyl-containing derivative. Unfortunately this cannot be characterized on the simple basis of band shift, neither previous work done by others has accomplished a thorough characterization particularly regarding the possible isoaspartyl nature of the "deamidated" residue (21,25,26,38,42). To confirm the identity of deamidated/isomerized Bcl xL , a standard of this protein (Bcl xL "n" form) was incubated in alkaline buffer and its product called the Bcl xL "i" form was analyzed by immunoblotting. Results showed that the Bcl xL i form migrated slowly on SDS-PAGE, compared with the Bcl xL n form (Fig. 9, panel A). This retarded migration is completely in agreement as described by previous authors (21,25,26,38,42). Mass spectral analysis of the Bcl xL i form showed a signal profile consistent deamidation/isomerization at positions 52 and 66 of Bcl xL (Fig. 9, inset). The relevant peptide mass fingerprint, obtained from MALDI-TOF analysis of the Bcl xL i form is shown in Table 1. All experimental m/z values are consistent with the masses of expected peptides from Bcl xL , as theoretically predicted by calculation (Table 1). Moreover all the mass values of peptides containing Asp 52 or Asp 66 were 1 mass unit higher than expected, confirming deamidation of the corresponding Asn precursors. In addition, the pattern of peptide fragments in the alkali-treated sample for 18 h is highly comparable with that of the sample deamidated for 6 h, thus allowing us to conclude that all susceptible Asn residues had been extensively deamidated and therefore the Bcl xL i form is stable (Table 1). However, mass spectral analysis could not allow us to distinguish between Asp and isoAsp. Therefore, to evaluate the presence of isoAsp residues in the Bcl xL i form, we used human recombinant PCMT1 as a specific tool. In fact only the isoAsp is a substrate of this enzyme, whereas normal aspartyl derivatives are not modified (1,2). PCMT1 assay showed a significant substrate activity for the Bcl xL i form, which was quantitatively enzymatically transformed in its product: Bcl xL "m" form ( Fig. 9, panel B). This result allowed us to correctly identify the Bcl xL i form as "isoaspartyl-containing" Bcl xL . Therefore a more appropriate denomination for this Bcl xL i form is "isomerized" rather than simply deamidated.
The effects of PCMT-dependent methylation on Bcl xL were monitored by band shift. Isomerized Bcl xL (Bcl xL i form) was incubated in vitro with human recombinant PCMT1, under conditions in which extensive methylesterification of the isoaspartyl sites gives way to the repair reaction sequence (1, 2). The   DECEMBER 23, 2011 • VOLUME 286 • NUMBER 51 product was then analyzed by SDS-PAGE, in parallel with aliquots of the Bcl xL n and Bcl xL i forms, for comparison of the relevant band positions (Fig. 9, panel C).

Targeting of Protein Repair Methyltransferase by RNAi
Results once again confirmed migration delay, on SDS-PAGE for the Bcl xL i form, whereas the Bcl xL m form displayed a pattern of migration absolutely comparable with that of the Bcl xL n form, thus allowing us to conclude that the Bcl xL n and m forms are structurally similar at least for what is related to their electrophoresis behavior.
PCMT1 Silencing Make HepG2 Cells More Susceptible to Apoptosis through Bcl xL Isomerization-The biological effects of PCMT1 silencing on both Bcl xL isomerization and apoptosis induction were investigated. HepG2 cells were treated with cisplatinum under condition A and transfected with siRNA or shRNA against PCMT1. Results (Fig. 10) showed that: 1), the methyltransferase expression is suppressed by all RNAi treatments; 2) in the samples in which PCMT1 is silenced, Bcl xL shows a significant band retardation on SDS-PAGE, compared with cell samples transfected with random siRNA (Fig. 10 panel A) or shRNA (Fig. 10, panel B). This band shift pattern is comparable with what was observed in Fig. 9, thus allowing us the correct assignment of this retarded band as isomerized. This migration behavior is comparable with what previously reported for "native" and "deamidated" Bcl xL (21,25,26,38,42); and 3) apoptosis markers (Caspase 3 and PARP) showed a significant increase upon cisplatinum apoptotic stimulus, in cells where the enzyme was silenced by siRNA (Fig. 10A) or shRNAs (Fig. 10B).
These results, as a whole, allow us to conclude that PCMT1 expression can be effectively down-regulated by the microRNA 15a/16 -1 cluster in HepG2, and that PCMT1 silencing, carried out by an RNAi approach, which allowed us to rule out any pleiotropic or off-target effects, makes the cells more susceptible to apoptosis induced by cisplatinum.

DISCUSSION
In the present paper we show that PCMT1 is effectively silenced, in HepG2 hepatoma cells, by microRNA 15a/16 -1, a cluster involved in down-regulation of various antiapoptosis mediators (i.e. Bcl2 and MCL1) (28,29). Down-regulation of PCMT1 in HepG2 depends on the specific interaction of microRNA 15a/16 -1 with a seed sequence located at the 3Ј UTR of the PCMT1 transcripts. In addition, we showed that PCMT1 is effectively silenced by various specific siRNAs and shRNAs. This effect is also associated with increased Bcl xL isomerization and increased susceptibility to apoptosis induced by cisplatinum. Apoptosis induction was monitored by Caspase 3 activation and PARP cleavage markers. Isomerized Bcl xL was characterized by various approaches also involving identification of isoaspartyl residues by means of human recombinant PCMT1 assay.
PCMT1 has been considered a ubiquitously expressed, housekeeping enzyme, with the ability to prevent accumulation of small amounts of isomerized proteins, spontaneously arising from slow deamidation/isomerization of relatively rare asparaginyl residues in susceptible sequences, which have been selected against during evolution (3). In particular, because most PCMT1 substrates (i.e. proteins containing isoaspartyl residues) occur in largely substoichiometric amounts (for no more than 0.5-1% of the total protein molecules) (2), it has been thought that PCMT1 would never become limiting in its capability to prevent accumulation of such abnormal isomerized proteins, even though evidence actually contradicts such belief. In particular, human erythrocytes accumulate isomerized membrane proteins with aging, although PCMT1 activity does not decay in the same aged red blood cells (16). Modifications of PCMT1 activity have been detected in various mammalian tissues (39) and in mouse testes and ovaries during spermiogenesis (40), where these changes have been regarded as functional adjustments. In pathological conditions, such as uremia/hyperhomocysteinemia, it has been shown that isomerized proteins accumulate in the presence of partial PCMT1 inhibition, due to the increase of S-adenosylhomocysteine, a homocysteine precursor and a powerful methyltransferase inhibitor (18). Isomerized proteins tend to accumulate in various tissues in homozygous knock-out Pcmt1 Ϫ/Ϫ mice, but not in the heterozygous animals. However, there is no report in the literature on the effects of stress conditions (e.g. induced by oxidative treatment) in this model (19). Therefore no specific information on increased susceptibility to apoptosis is available in the PCMT1 knock-out mice. These previous reports in the literature point FIGURE 6. Direct interaction of microRNA 15a and 16 -1 with the PCMT1 3 UTR seed sequence monitored by luciferase reporter activity assay. A construct of the PCMT1 3Ј UTR carrying the putative seed sequence for miRNA 15a/16 -1 interaction was cloned downstream of a luciferase reporter gene in pGL3TK vector. HepG2 were co-transfected with this recombinant plasmid and with either pre-miR 15a or 16 -1. Levels of luciferase activity were normalized against the activities of a Renilla internal standard expressed by a co-transfectant pRLCMV vector in each sample. pGLpcmt/miR15, pGL3TK vector carrying the PCMT1 seed sequence, co-transfected with pre-miR15. pGLpcmt/miR16, pGL3TK vector carrying the PCMT1 seed sequence, co-transfected with pre-miR16. pGLCT/miR15, pGL3TK empty vector, co-transfected with pre-miR15 (control). pGLCT/miR16, pGL3TK empty vector, co-transfected with pre-miR16 (control). pGLpcmt/miRNARD, pGL3TK vector carrying the PCMT1 seed sequence, co-transfected with scrambled pre-miRNA (control).
to the interpretation that, even in the presence of apparently normal PCMT1 activity, various physiological or pathological conditions, mostly associated with human diseases or aging, may overcome the ability of PCMT1 to adequately prevent isoAsp accumulation. In addition, there are situations, such as cancer, in which resistance to apoptosis is a key feature of the pathophysiological mechanism (41). Therefore our findings suggest that the microRNA-dependent regulation mechanism may contribute to "tuning down" PCMT1 activity, under conditions in which cell commitment to apoptosis needs to be reinforced.
A possible question regards the specificity of the microRNA 15a/16 -1 cluster activity on PCMT1 and its possible reflections on apoptosis regulation. These microRNAs down-regulate a number of other genes encoding for antiapoptotic mediators like Bcl2 (28). MicroRNA 15a/16 -1 deregulation, through deletion or mutation, affects tumor resistance to apoptosis, by causing Bcl2 overexpression (37). It has been reported that the locus corresponding to microRNA 15a and microRNA 16 -1 at 13q14 is deleted in more than half of B-cell chronic lymphocytic leukemias (34,35). The undeleted microRNA 15a-microRNA 16 -1 allele, in chronic lymphocytic leukemia patients, may also contain inactivating point mutations (36,37). These observations led to the proposal that microRNA 15a and microRNA 16 -1 may behave as tumor suppressors (37). Therefore, whereas the effects of the microRNA 15a/16 -1 cluster on PCMT1 silencing were truly straightforward, any effect on apoptosis would lack sufficient specificity, due to wide pleiotropic effects of microRNA 15a/16 -1 on various other antiapoptotic mediators (28,29).
To establish the potential biological role of microRNA 15a/ 16 -1-dependent down-regulation of PCMT1 on apoptosis, we used an alternative approach, based on small noncoding RNAs, for selectively silencing only PCMT1 in cell cultures treated with a pro-apoptotic stimulus. To this purpose we employed various siRNAs and shRNAs devoid of off-target or pleiotropic effects and specifically targeted on PCMT1 at different sites in the transcripts, to assess the specific role of PCMT1 in apoptosis, without any interference on other apoptosis mediators. Results showed that selective PCMT1 silencing in cells stressed with cisplatinum makes cells more sensitive to apoptosis, by inducing an increase of Bcl xL deamidation/isomerization, which is devoid of antiapoptotic activity, as previously reported on various models (21,25,26,38,42).
Our results in fact are in line with the reports by previous authors, except that, based on our characterization of deamidated Bcl xL , we propose to more adequately denominate this protein subform as isomerized, also because, as previously shown, the simple replacement of Asn with normal Asp does not cause, per se, any functional loss (21,43). Takara and Takahashi (26) showed decreased deamidation of Bcl xL in human hepatocellular carcinomas, on the basis of its band shift on SDS-PAGE. Therefore these authors hypothesized an association with increased protection against apoptosis in these tumors (26). Later, Zhao and collegues (38) showed that inhibition of the Bcl xL deamidation pathway, detected merely on the basis of band shift, contributed to the development of myeloproliferative disorders. Deamidation of Bcl xL has been related to a mechanism involving up-regulation, under proapoptotic conditions, of a proton pump, which in turn causes alkalinization of the intracellular medium, thus creating conditions for protein deamidation to occur, with special regard to Bcl xL (25). It should be pointed out that no one of these reports was actually concerned with characterization of the shifted Bcl xL band, which in fact was generally referred as deamidated, without mentioning the possibility of isoaspartyl formation. On the  DECEMBER 23, 2011 • VOLUME 286 • NUMBER 51 other hand isoAsp is the major product of protein deamidation under alkaline conditions in various proteins (5)(6)(7)(8)(9). In addition, it has been shown that isoaspartyl containing calmodulin displays an altered molecular radius, which is responsible for its aberrant electrophoretic migration (14).

Targeting of Protein Repair Methyltransferase by RNAi
More recently the isoaspartyl-containing derivative of deamidated Bcl xL has been characterized by proteome analysis, from cells treated with antisense PCMT1 plus oxidative stress. This was accomplished by using a purification proto-col that employed a resin linked to human recombinant PCMT1, which can only recognize isomerized proteins but not the one carrying normal aspartyl (27). These latter experiments provided the first evidence that Bcl xL undergoes deamidation under cell stress conditions, yielding its isomerized derivative, which is devoid of antiapoptotic function.
In the present work we characterized a Bcl xL slower-migrating band on SDS-PAGE showing that under alkaline conditions,  closely resembling those occurring in apoptotic cells (25), Bcl xL isomerizes, because it is methylated in vitro, by human recombinant PCMT1. This isomerized Bcl xL migrates slowly on SDS-PAGE, with a behavior totally comparable with what was observed by previous authors (21,25,26,38,42). In addition, enzymatic methylation of isomerized Bcl xL , by human recombinant PCMT1, reverts its aberrant pattern of migration so that methylated Bcl xL is no longer distinguishable from native Bcl xL by electrophoresis. This result is consistent with the interpretation of the repair role of PCMT1 in agreement as described for many other proteins (1)(2)(3)(4)(5). We then monitored the increase of isomerized Bcl xL under conditions in which PCMT1 had been silenced and apoptosis induced by cisplatinum (see Fig. 10).
Therefore, based on our present result, we can then hypothesize that miRNA-dependent down-regulation of PCMT1 can be envisioned as a way to increase susceptibility to apoptosis, and PCMT1 may represent an additional late checkpoint of apoptosis regulation. We may infer a possible role of PCMT1 up-regulation, induced by the microRNA 15a/16 -1 defect, as a late mechanism for triggering apoptosis resistance in tumors. In this respect it may be interesting to look at PCMT1 activity in cancer, where microRNA 15a and 16 -1 are suppressed and/or Bcl xL repaired by methylation.
The overall scenario may be even more complex, because of the pleiotropic activity of this microRNA cluster on a number of other antiapoptotic genes (23, 28 -30, 35-37, 44), not just PCMT1 and also because other PCMT1 substrates, such as Hsp70, Hsp90, are involved in apoptosis (27), and both of these aspects may further complicate this model and certainly deserve further investigations. These results, as a whole, support a novel interpretation that PCMT1 should not be simply interpreted as a repair methyltransferase of aged/damaged proteins, but it may act as an effective modulator of apoptotic cell death, whose derangement may play a role in proliferative or degenerative disorders.