Sirtuins belong to a class of NAD-dependent deacetylases, and include seven distinct isoforms, of which SIRT7 is the least studied member. In the present study, the subcellular expression of SIRT7 in primary fibroblasts undergoing senescence was evaluated by immunocytochemistry and immunoblot assay. Expression of nucleolar SIRT7 in young fibroblast was very prominent, decreased in pre-senescent cells, and became undetectable in the senescent cells. Interestingly, we observed previously unreported staining for cytoplasmic SIRT7 in fibroblasts, and report the existence of a steady-state level of SIRT7 in cytoplasm. Selective localization of the high-molecular-mass (47.5 kDa) SIRT7 in the cytoplasmic fraction and the low-molecular-mass (45 kDa) SIRT7 in the nuclear fraction was observed in the immunoblot analysis of various cell types. The specificity of the N-terminal antibodies for detection of cytoplasmic SIRT7 was confirmed by RNAi and peptide competition assays. The two forms of SIRT7 showed reciprocal expression following serum starvation, nocodazole and okadaic acid treatments, and also during senescence. Using a combination of deletion constructs and site-directed mutagenesis, we defined the role of two distinct SIRT7 sequences in the N-terminal region (amino acids 61–76, LQGRSRRREGLKRRQE) and the C-terminal region (amino acids 392–400, KRTKRKKVT) for nuclear and nucleolar import, respectively. In conclusion, we report for the first time the existence of a cytoplasmic pool of SIRT7 in addition to its well-known nucleolar form, identify distinct localization signals for its nuclear/nucleolar targeting, and describe an association between loss of nucleolar SIRT7 and replicative senescence.
Sirtuins are a family of NAD-dependent protein deacetylases, and their over-expression increases lifespan in lower organisms such as yeast (Saccharomyces cerevisiae) , worms (Caenorhabditis elegans)  and flies (Drosophila melanogaster) . The direct role of sirtuin in mammalian longevity is still unclear. Knockouts and transgenic models in mice (Mus musculus) have suggested roles for sirtuin isoforms in varied responses, including metabolism [4-6], oxidative stress [7, 8] and genomic stability [9-11], all of which may in turn affect ageing. Sirtuin biology in mammals is complicated in view of the presence of multiple isoforms (SIRT1–SIRT7), with different subcellular localizations and a multitude of downstream targets.
Mammalian sirtuin members are mainly restricted to three subcellular compartments, the cytoplasm, nucleus and mitochondria (Table 1), and are dynamic in nature, as they appear to shuttle between them. SIRT1 was previously thought to localize exclusively in the nucleus, but appears to be a nucleo-cytoplasmic shuttling protein as its presence in cytoplasm has been reported in murine pancreatic β-cells , a subset of mouse neurons and ependymal cells . Primary cells show nuclear localization of SIRT1, in contrast to aberrant cytoplasmic localization in cancer cell lines (MCF7, DU145, PC3 and H460) . The nucleo-cytoplasmic shuttling of SIRT1 is mediated by phosphatidylinositol-3-kinase/insulin like growth factor-1 receptor (PI3K/IGF-1R) signalling . SIRT2, which deacetylates tubulin, is predominantly cytoplasmic, but is also present in the nucleus, and has a role in mitosis . SIRT3, which has been implicated in the regulation of lifespan , also resides in nuclei under normal growth conditions in addition to its main mitochondrial niche. The nuclear SIRT3 shuttles to mitochondria under conditions of DNA damage induced by etoposide and UV irradiation . Endogenous as well as over-expressed SIRT4, which has ADP-ribosyltransferase activity, is associated with the mitochondrial matrix . By immunohistochemistry, SIRT4 was found to be predominantly localized in the cytoplasm in a variety of cell types, including vascular smooth muscle cells, striated muscle fibres and islets of Langerhans . In cerebral neurons, SIRT5 was found in the nucleus and cytoplasm and also in mitochondria, where it translocates between the matrix and the inter-membrane space . This appears to be the most dynamic of all sirtuin isoforms. Isoform 1 of SIRT5 shuttles between the nucleus and cytoplasm and is excluded from mitochondria, while isoform 2 is predominantly mitochondrial . Mouse SIRT6 was found predominantly in the nucleus, but shows diffuse cytoplasmic localization when over-expressed . SIRT6 in the nucleus remains associated with heterochromatin, especially the telomeric regions, and has been shown to be involved in DNA repair, telomere maintenance, gene expression and glucose metabolism [4, 21-24].
Table 1. Varied localization pattern of mammalian sirtuin isoforms
Sirtuins control a number of important cellular and metabolic events in cell proliferation, and are generally considered to have a role in longevity. The anti-senescence properties of SIRT1 and SIRT3 had been demonstrated in mammalian cell culture model systems. In addition, studies on the mouse models of SIRT6 and SIRT1 knockouts suggest a role in preventing premature ageing. Mitochondrial sirtuins have been suggested to be metabolic sensors involved in slowing the ageing process [25, 26]. Recently the role of sirtuins in extending the life span of lower organisms was questioned in some of the studies [27, 28].
SIRT7 is the least studied member of the seven mammalian sirtuins. Existing reports on SIRT7 indicated exclusive nucleolar localization [22, 29, 30]. SIRT7 is a positive regulator of rDNA transcription, as it associates with ribosomal RNA polymerase I . SIRT7 interacts with rDNA transcription factor upstream binding factor (UBF), a component of the RNA polymerase I machinery, and was shown to have a role in mitosis . There is some controversy regarding the deacetylase activity of SIRT7. Vakhrusheva et al. () found that SIRT7 deacetylates p53 at lysine 382, but other reports [22, 29, 31] demonstrated no such enzymatic activity. SIRT7 knockout mice have a short lifespan and develop cardiac hypertrophy. Two recent studies suggested a role for SIRT7 in chromatin remodelling, particularly histone H3K18 deacetylation at specific gene loci [31, 33]. SIRT7 is now considered important for maintaining the transformed state of a cell . In general, localization of sirtuins and their shuttling between various subcellular compartments are associated with and possibly dictate their multiple effects in relation to apoptosis, stress, DNA damage, senescence, differentiation, mitosis and metabolism. In the present study, we report strong cytoplasmic staining for SIRT7 in fibroblasts, which led us to undertake a detailed analysis of its subcellular localization in various cell types by both immunocytochemistry and subcellular fractionation using two different antibodies against SIRT7. We demonstrate the presence of endogenous SIRT7 in the cytoplasm, in addition to its previous well-known nucleolar niche. Further, we also identified a stretch of amino acid residues that is rich in lysine/arginine in human SIRT7 that facilitates nucleolar targeting.
SIRT7 levels are affected following replicative senescence
To evaluate the changes in expression levels and localization of SIRT7 during replicative senescence, we used primary TIG3 fibroblasts, which were serially passaged until they attained senescent morphology. Senescent cells were identified by enlarged morphology and were positive for senescence-associated β-galactosidase activity (data not shown). Using the N-terminal antibody H1 (a kind gift from I. Horikawa, National Institutes of Health, Bethesda, MD, USA), we evaluated the expression of SIRT7 in young and senescent cells by immunocytochemistry. Young TIG3 cells (passages 5–10) showed distinct nucleolar and cytoplasmic staining for SIRT7. The presence of SIRT7 in cytoplasm, in addition to its well-known nucleolar niche, appeared unusual, and, to our knowledge, had not been reported previously. Disappearance of nucleolar SIRT7, but not the cytoplasmic form, was observed in the senescent cells when kept in culture for a prolonged duration (Fig. 1A and Fig. S1A). We extended the study to cells subjected to adriamycin-induced premature senescence to determine whether the disappearance of nucleolar SIRT7 is a general feature of senescent cells. We did not find any change in the expression or localization of SIRT7 during stress-induced premature senescence (data not shown). The above results thus indicate a link between reduced levels of nucleolar SIRT7 and replicative senescence.
We then evaluated the expression levels of SIRT7 in young and senescent cells by immunoblot analysis using anti-SIRT7 antibodies that recognize two different N-terminal epitopes: H1 and a commercial antibody S5947. These recognize amino acids 9–25 (SERKAAERVRRLREEQQ) and 35–51 (ILRKAAAERSAEEGRLL), respectively. Interestingly, the immunoblot analysis of SIRT7 protein revealed two signals corresponding to 45 and 47.5 kDa. Serial passaging, which leads to replicative senescence, resulted in a conspicuous decrease in the levels of the 45 kDa signal, with a compensatory increase in 47.5 kDa signals when probed with antibodies H1 (Fig. 1B) or S5947 (Fig. S1B). A detailed experimental analysis, as described below, suggests that the 45 kDa signal corresponds to the nucleolar form of SIRT7, and the 47.5 kDa signal corresponds to the cytoplasmic form.
SIRT7 is localized in the nucleolus and cytoplasm of primary human fibroblasts
The existence of cytoplasmic SIRT7, as revealed by microscopy, and the detection of two variants of SIRT7 (45 and 47.5 kDa) by immunoblot analysis, whose expression correlated with replicative senescence, prompted further in-depth exploration of the subcellular distribution of SIRT7 in a variety of cell types. An exogenously expressed GFP–SIRT7 fusion protein was primarily localized in the nucleolus of HEK 293 cells (Fig. 2A). However, it was detected in both the cytoplasm and nucleolus when over-expressed in young primary fibroblasts (WI38 cells) (Fig. 2A). It is possible that the unusual localization of GFP–SIRT7 in the cytoplasm of WI38 cells is because of its over-expression. Therefore, the distribution of endogenous SIRT7 was assessed in both epithelial and fibroblasts cells by immunocytochemistry using two different SIRT7 antibodies. Immunocytochemistry using H1 antibody showed only nucleolar staining in epithelial cells (U2OS and SiHa), while SIRT7 staining was observed both in the nucleolus and cytoplasm in fibroblasts (WI38) (Fig. 2B). To evaluate whether the cytoplasmic localization of SIRT7 is specific to WI38 cells, we examined its localization pattern in two other primary human fibroblast lines. Both TIG3 and MRC5 cells showed SIRT7 expression in the cytoplasm and nucleolus (Fig. 2C). Immunocytochemistry using antibody S5947 showed the same staining pattern as for H1 in fibroblasts (Fig. 2D) and epithelial cells (Fig. S2A). We then examined whether the localization of SIRT7 varies between immortalized and cancerous epithelial cell lines. We used two pairs of immortalized and cancerous counterpart cell lines: HPLD1 cells (immortalized lung type II epithelium) versus A549 cells (lung carcinoma), and HaCaT cells (immortalized keratinocyte) versus SiHa cells (cervical cancer). SIRT7 showed nucleolar localization in both the immortalized and cancer cell lines (Fig. S1C). Similar results were obtained using the S5947 antibody (data not shown). Both acetone/methanol and formaldehyde/Triton X-100 fixation procedures gave similar results regarding the localization pattern (data not shown), ruling out any possible influence of fixatives. These results suggest cytoplasmic localization of SIRT7, as detected by immunocytochemistry, only in primary fibroblasts.
An intense cytoplasmic presence of SIRT7 was also observed in primary human colon tissues by immunohistochemistry using the two antibodies (Fig. 2E). SIRT7 immunoreactivity was found in the cytoplasm of colonocytes and the goblet cells, but not mucus vacuoles. Interestingly, no nuclear staining for SIRT7 was found in normal colon tissues.
Cell fractionation studies to evaluate the subcellular distribution of SIRT7
To evaluate the differences between the cytoplasmic and nucleolar SIRT7 variants, we subjected cellular homogenates from various cell types to immunoblot analysis with both antibodies. As anticipated, the two antibodies (H1 and S5947) detected a faster-migrating band at 45 kDa, the molecular mass of SIRT7 as predicted by based on sequence information (Fig. 3A). In addition, the antibodies also detected a slower-moving signal corresponding to 47.5 kDa in HEK 293 cells (Fig. 3A and Fig. S2B). The two differently migrating forms of SIRT7 were also detected in other epithelial cells (U2OS, SiHa and HaCaT) as well as primary fibroblasts (WI38, TIG3 and MRC5) (Fig. 3B).
As a second independent method to investigate SIRT7 distribution, we analysed the subcellular fractions of two representative cell lines each of epithelial origin (SiHa and U2OS) and fibroblast origin (TIG3 and WI38), by immunoblotting. The purity of the nuclear and cytoplasmic fraction was ascertained using lamin B and α-tubulin, respectively (Fig. 3C and Fig. S2C). The antibodies detected a faster-migrating (45 kDa) signal in the nucleus and a slower-migrating (47.5 kDa) signal in the cytoplasm of both fibroblast and epithelial cells. The distribution of SIRT7 in various cellular compartments as detected by immunocytochemistry and subcellular fractionation using the two antibodies in fibroblast and epithelial cells is summarized in Table 2.
Table 2. Summary of the endogenous SIRT7 localization pattern in various cell types using various antibodies. +, present; −, absent
Epitope (amino acids)
Fibroblasts and epithelial cells (show same pattern)
Specificity of SIRT7 antibodies as tested by peptide competition assay and RNAi
The specificity of the antibodies was first verified by a peptide competition assay. Pre-incubation of both H1 and S5947 with synthetic peptides corresponding to their respective epitopes resulted in complete loss of both the 47.5 and 45 kDa signals in cell lysates from various cell lines (Fig. 3D). The results were confirmed by immunocytochemistry, which showed a loss of cytoplasmic and nucleolar signals after pre-incubation with the synthetic peptide (Fig. 3E,F). The peptide competition assay therefore confirmed the specificity of both the antibodies in detecting the previously unreported 47.5 kDa signal of SIRT7.
A knockdown assay using siRNA directed against SIRT7 was used as a second method to establish the specificity of the N-terminal antibodies in detecting the cytoplasmic form of SIRT7. This was performed in two ways: (a) transient transfection with siRNA or shRNA in HEK 293 cells, and (b) stable integration of shRNA directed against SIRT7 in U2OS and TIG3 cells.
An siRNA targeting sequence previously validated as specific for SIRT7  was transfected into HEK 293 cells, and cellular homogenates were analysed for SIRT7 expression using H1 and S5947 antibodies. The effective knockdown of the 47.5 and 45 kDa SIRT7 signals (by approximately 50%) confirmed the specificity of the H1 antibody (Fig. 4A and Fig. S2D,E). However, when S5947 was used for probing after siRNA transfection, we observed specific disappearance of only the 45 kDa signal and not the 47.5 kDa signal. Additionally, we performed transient transfection with four shRNAs targeting various regions of SIRT7. This confirmed knockdown of both the SIRT7 signals when probed with H1 antibody, but effective knockdown was observed only for the 45 kDa signal when probed with S5947 (Fig. S3A). These results raise a question regarding the specificity of S5947 for detecting the 47.5 kDa signal. To further confirm our findings, shRNAs targeting various regions of SIRT7 were stably integrated in U2OS and TIG3 cells by retroviral transduction. The transduced cells were selected in puromycin for 10 days, and subsequently analysed for SIRT7 expression. In the SIRT7 knockdown cells (SIRT7 KD), prominent loss of both the 47.5 and 45 kDa signal (approximately 80% in U2OS cells) was observed by both immunoblot (Fig. 4B and Fig. S2F) and immunocytochemistry (Fig. 4C) when probed with H1 antibody. Thus, use of the SIRT7 KD cells confirms not only the specificity of the H1 antibody but also the existence of a cytoplasmic form (47.5 kDa) of SIRT7. In contrast, when probed with S5947, the SIRT7 KD cells showed an appreciable loss (approximately 80%) of the 45 kDa nucleolar form of SIRT7 (Fig. 4B,D and Fig. S2F). Unlike transient transfection with SIRT7 siRNA, where we did not find any decrease in the intensity of 47.5 kDa, the stable SIRT7 KD cells showed a decrease of approximately 25% in expression of the 47.5 kDa protein as seen by immunoblotting and immunocytochemistry using S5947 (Fig. 4B,D). Collectively, these results indicate that the H1 antibody is more specific, while the commercial antibody (S5947) besides detecting the cytoplasmic SIRT7 (47.5 kDa), might also be recognizing an additional non-specific signal.
Expression of SIRT7 variants under various growth conditions
In view of a previous report  suggesting that SIRT7 is dephosphorylated at the exit from mitosis by a phosphatase that is sensitive to okadaic acid, we tested whether expression of the two SIRT7 signals, as detected by the two antibodies, varies with cell cycle or okadaic acid treatment. Three cell lines (U2OS, SiHa and HaCaT) were treated with nocodazole for G2/M arrest, serum-starved for synchronization at G1/G0, or treated with okadaic acid (Fig. 5). After nocodazole and okadaic acid treatments, we observed a significant increase in the intensity of the 47.5 kDa signal, accompanied by a corresponding decrease in the intensity of the 45 kDa signal in homogenates of various cell lines probed with either H1 or S5947antibody (Fig. 5A–D and Fig. S3B,C). Conversely, there was a decrease in the intensity of the 47.5 kDa signal accompanied by an increase in the intensity of the 45 kDa signal when cells were synchronized at the G1/G0 stage by serum starvation (SS) (Fig. 5E,F and Fig. S3D). This reciprocal expression of the two SIRT7 forms strongly indicates the existence of at least two pools of SIRT7 whose relative abundance is dependent on the cell-cycle stage. In accordance with the previous report that SIRT7 is phosphorylated, and that this phosphorylation is sensitive to okadaic acid treatment , we also detected an increase in expression of 47.5 kDa signal upon treatment with the phosphatase inhibitor okadaic acid (Fig. 5C and Fig. S3C). However, there was no decrease in the intensity of the 47.5 kDa signal after treatment with calf intestinal phosphatase (Fig. S3E). The combined results of the okadaic acid and calf intestinal phosphatase treatments are indicative of an additional post-translational modification, rather than simply a phosphorylation event.
Identification of the nuclear and nucleolar localization signals of human SIRT7
Understanding the role of SIRT7 localization signals in nucleolar targeting is of interest because of its prominent presence within nucleolus in most cell types including the young primary fibroblasts. HEK 293 cells were used for localization studies using a full-length GFP–SIRT7 fusion construct, as these cells have high transfection efficiency. When transfected into HEK 293 cells, GFP–SIRT7 localizes predominantly in the nucleolus. The estimated size of SIRT7 is approximately 45 kDa and that of GFP is approximately 27 kDa, producing a fusion construct that is too large to diffuse freely into the nucleolus. We therefore postulated the presence of a nucleolar localization signal (NoLS) in the protein.
In order to identify the amino acid sequences required for nuclear/nucleolar targeting, we evaluated the localization patterns of various SIRT7 deletion constructs fused to GFP (Fig. 6A and Fig. S4A,B) after transient expression in HEK 293 cells. The N-terminal construct (amino acids 1–203) localized primarily to the nucleus and was excluded from the nucleolus, while the C-terminal construct (amino acids 197–400) localized to nucleoli, suggesting a role for the N- and C-terminal regions in targeting to the nuclear and nucleolar compartments, respectively (Fig. 6A). The full-length SIRT7 protein lacked typical monopartite or bipartite nuclear localization sequences (NLSs); instead, we observed stretches of basic amino acid residues, amino acids 66–74 (RRREGLKRR) and amino acids 392–398 (KRTKRKK), in the N- and C-terminal regions, respectively, which we postulated to be putative signals for nuclear/nucleolar localization. This was confirmed by the deletion strategy: the longer N-terminal construct (amino acids 61–203) localized to the nucleus but a shorter construct (amino acids 71–203) lacking the basic amino acid stretch (RRREGLKRR) failed to localize in the nucleus; similarly, the longer C-terminal deletion construct (amino acids 197–400) localized to the nucleolus but a shorter construct lacking amino acids 392–398 (KRTKRKK) remained in the cytoplasm. The importance of the basic amino acids, as established above, in the nuclear and nucleolar targeting signal was confirmed by site-directed mutagenesis. The positively charged lysine/arginine residues in the NLS (amino acids 66–74) and NoLS (amino acids 392–398) were mutated to alanine (Fig. 6B). When these amino acids were mutated simultaneously at both the NLS and NoLS regions, the protein remained in the cytoplasm, as expected. Mutagenesis in either the NLS or NoLS region alone did not alter the nucleolar localization of the full-length protein. However, mutation of these amino acids in the smaller N-terminal construct (amino acids 1–203) and the smaller C-terminal construct (amino acids 197–400) led to loss of their nuclear and nucleolar localizations, respectively. These results indicate that, for full-length SIRT7, basic residues at either the NLS or the NoLS are important and sufficient for nucleolar localization.
Next, we prepared a series of full-length SIRT7 constructs with minimal deletion of amino acids in the N- and C-terminal regions (Fig. 7A) to determine whether there is a minimal stretch of amino acid residues that are essential for the nucleolar targeting. When amino acids 61–76 (LQGRSRRREGLKRRQE) were deleted at the N-terminal region, SIRT7 was localized in the cytoplasm. When amino acids 392–400 (KRTKRKKVT) in the extreme C-terminal region were deleted, SIRT7 was retained in nucleus but failed to localize to the nucleolus. Deletion of both N- and C-terminal signals led to complete cytoplasmic retention of protein, thereby indicating their importance in proper nucleolar localization.
To further confirm the functionality of the candidate NLS and NoLS regions, we tested whether they are sufficient to target heterologous proteins to the nucleus and nucleolus (Fig. 7B). We therefore fused the NLS and NoLS of SIRT7 to enhanced GFP (EGFP) at its C-terminus. Fusion of amino acids 61–76 (LQGRSRRREGLKRRQE) localized the heterologous protein in the nucleus, while fusion of amino acids 392–400 (KRTKRKKVT) to EGFP showed a heterogeneous pattern, with at least 30% of the cells showing nucleolar localization of SIRT7 and the rest showing nuclear localization (Fig. 7B). Nucleolar targeting of EGFP was efficiently achieved when both NLS and NoLS were fused to EGFP. In summary, our study shows that amino acids 61–76 and 392–400 of SIRT7 act as independent nuclear and nucleolar localization signals, respectively. Whereas the NLS alone targets longer as well as shorter peptides to the nucleus, the NoLS targets only shorter peptides to the nucleolus and requires the NLS for efficient targeting of full-length SIRT7.
Using a computational-based approach, we also performed a pattern search for the amino acid sequences of the NLS and NoLS in human SIRT7 in the nuclear and nucleolar signal databases as described by Scott et al. . Our analysis revealed considerable similarity between the identified NLS/NoLS region of SIRT7 and various other known nucleolar proteins (Fig. S5).
A characteristic of most of the sirtuin family members is differential localization and shuttling between various subcellular compartments. By this means, they possibly acquire additional functions. Expression of SIRT7 was observed in both the cytoplasm and nucleoli of young fibroblasts by imaging and subcellular fractionation using antibodies recognizing two N-terminal epitopes. The cytoplasmic localization pattern observed in the present study is unusual and is in contrast to previous reports that described only nucleolar localization [22, 29-31]. A remarkable feature of the senescent cells is the occurrence of SIRT7 at a minimum level in their nucleoli, and this was not due to nucleolar degradation. In addition, the decrease of nucleolar SIRT7 appeared gradual, and complete disappearance was observed only when replicative senescent cells were kept for a prolonged period in culture dishes. Another notable feature was the inverse correlation between expression of SIRT7 signals for the nucleolar (45 kDa) and cytoplasmic (47.5 kDa) forms in senescent cells, which led us to perform a further detailed evaluation regarding SIRT7 subcellular localization. The following significant observations were made in the present study: (a) SIRT7 localizes to both nucleolar and cytoplasmic compartments, (b) the nucleolar and cytoplasmic forms move differentially on immunoblots, indicative of post-translational modification, and (c) expression of the two SIRT7 signals under certain conditions is reciprocal, suggesting a steady-state flux between the two differently localized pools.
Using a number of epithelial and fibroblast cell types, we observed both cytoplasmic and nucleolar localization of SIRT7 by immunocytochemistry in primary fibroblasts, but only nucleolar staining was observed in epithelial cells. However, subcellular fractionation followed by immunoblotting revealed the presence of the faster-migrating nuclear form (45 kDa) and the slower-migrating cytoplasmic form (47.5 kDa) of SIRT7 in both cell types. We expect that the results seen by us may also be extended to other epithelial and fibroblast cell types. Byles et al.  reported a similar difference, and suggested that antibody-specific epitopes play a dominant role in determining the localization pattern in immunocytochemistry assays. Why is there a difference in localization pattern detected by immunocytochemistry versus subcellular fractionation specifically in epithelial cells? A possible explanation may be that the cytoplasmic SIRT7 in epithelial cells assumes a specific conformation on interaction with other proteins or undergoes post-translational modification, thereby masking the epitope and interfering with antibody recognition. However, this epitope is detected once the interaction or modification is reversed during denaturation in SDS/PAGE analysis. Thus combining immunocytochemistry with biochemical assays using subcellular fractionation is the preferred way to confirm localization of a given protein. The existence of SIRT7 forms with various mobility, tested using previously validated antibodies, implies multiple levels of post-translational modifications. Thus discrepancies regarding the localization of proteins in various subcellular compartments may be verified by use of antibodies that recognize different epitopes and hence other forms of the same protein. The intracellular localization of stem cell markers such as OCT4, Nanog and SOX2, which is reported to differ drastically depending on commercial source of antibody and its epitope . The present study therefore suggests that caution in interpretation of results must be exercised when using antibodies that detect different epitopes of the same protein.
We therefore also undertook a comparison between the two antibodies regarding their specificity in two cell lines with stable integration of shRNA directed against SIRT7. The H1 antibody detected loss of both the 45 and 47.5 kDa forms in the SIRT7 stable knockdown cell lines TIG3 and U2OS, indicating not only the specificity of this antibody but also confirming the existence of a cytoplasmic pool of SIRT7. In contrast, when probed with the commercial antibody S5947, the SIRT7 KD cell lines showed prominent loss of 45 kDa but only a partial loss of 47.5 kDa signal. This suggests that in addition to detecting the 47.5 kDa signal of SIRT7, the S5497 antibody may also be detecting a non-specific signal. However, the effective knockdown of 47.5 kDa as detected using H1 argues in favour of the presence of a cytoplasmic pool of SIRT7. Additionally, the reciprocal expression of the 45 and 47.5 kDa form during certain cell-cycle stages and after okadaic treatment, and immunoreactivity in cytoplasm of normal colon tissue, strongly supports the existence of cytoplasmic SIRT7. Previously, Grob et al. suggested that SIRT7 may be subject to okadaic acid-sensitive phosphorylation . Our results confirmed this; however, the intensity of the 47.5 kDa signal remained unaffected following calf intestinal phosphatase treatment of the cellular lysate. This may be attributed to preferential enrichment of mitotic cells from S phase following okadaic acid treatment [36, 37]. We therefore suggest that, in addition to phosphorylation, the 47.5 kDa signal may also have other multiple post-translational modifications, and this warrants further studies.
Furthermore, the present work also supports a prominent nucleolar localization of SIRT7, thereby making it important to understand the mechanism of nucleolar targeting of SIRT7. Nucleolar retention involves complex mechanisms such as target protein binding to nucleolar components such as RNA, DNA or protein and the presence of charged residues in the nucleolar targeting signals that bind the nucleolar components . Nucleolar proteins that are above the size-exclusion limits of nuclear pore complexes require the presence of an additional nuclear localization signal for active transport through the nuclear pore complex . The strategies used for defining a sequence as an NLS/NoLS mainly involve imaging studies to determine the localization patterns using deletion constructs fused to a reporter, site-directed mutagenesis of the putative sites in the full-length protein, and fusing the putative sequence to a heterologous protein. A previous report indicated that deletion of amino acids 1–75 at the N-terminal region of SIRT7 promoted its cytoplasmic localization . However, closer examination of that report indicated that a substantial amount of this deletion construct was retained in the nucleoli. In the present study, we used an extensive mapping strategy that revealed that the N-terminal region is necessary but not sufficient for nucleolar targeting of SIRT7. The N-terminal region containing the NLS (amino acids 61–76, LQGRSRRREGLKRRQE), requires an additional C-terminal NoLS (amino acids 392-400, KRTKRKKVT) for correct nucleolar localization of SIRT7. Site-directed mutagenesis of basic residues in the NLS or NoLS alone had a less severe effect, and SIRT7 was still imported to nucleolus. This indicates that basic amino acids at both sites are important, and the presence of either is sufficient to retain the protein within the nucleolus. However, deletion of either the NLS or NoLS sequences in full-length SIRT7 protein abrogated its nuclear and nucleolar targeting, respectively. This was further supported by the finding that fusion of a minimum stretch of candidate sequence to a heterologous small-molecular-mass protein such as EGFP localized the target protein to either the nucleus or the nucleolus. It should be further noted that fusion of the NoLS alone efficiently localizes smaller peptide sequences (approximately 200 amino acids) to the nucleolus, but an additional NLS is required for efficient nucleolar targeting of larger protein sequences (more than 200 amino acids). These results reiterate the importance of both the NLS and NoLS in correct targeting of SIRT7 to nucleolus.
In summary, the present study indicates the presence of a steady-state cytoplasmic pool of SIRT7 in addition to the well-known nucleolar form, and further confirmation regarding its identity due to either postranslational modification or any other mechanism need to be evaluated in future studies. A possible link between nucleolar SIRT7 and replicative senescence in fibroblasts has emerged. Additionally, nucleolar localization of SIRT7 is dependent on an N-terminal NLS and a C-terminal NoLS. In addition to a prominent role in the nucleolus, SIRT7 may have additional functions in cytoplasm that are yet to be discovered.
Cell culture, treatment and transfections
Cell lines HEK 293, SiHa, HeLa, HaCaT, HPLD, U2OS, TIG3, WI38 and MRC5 were cultured in the appropriate growth medium at 37 °C and 5% CO2. Cells were treated with okadaic acid (120 μm) (Sigma, St Louis, MO, USA) or nocodazole (6 μg·mL−1) (Sigma) for 12 h, and then harvested by trypsinization or mitotic shake off (for nocodazole treatment). For serum starvation, cells were grown in Dulbecco's modified Eagle's medium containing 0.1% serum for 72 h. GFP–SIRT7 was transiently transfected into HEK 293 cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). For treatment with calf intestinal phosphatase (NEB, Ipswich, MA, USA), cells were lysed by the freeze–thaw method, and 1× calf intestinal phosphatase buffer was added. Then calf intestinal phosphatase (1 unit·μg protein−1) was added, and incubated at 37 °C for 15 or 30 minutes.
Full-length SIRT7 was PCR-amplified from a pcDNA-DEST47-SIRT7 construct obtained from I. Horikawa (National Institutes of Health). The PCR-amplified full-length SIRT7 was then cloned into plasmid EGFP-C3 (Clontech, Mountain View, CA, USA). For deletion of the probable localization signals and site-directed mutagenesis, the full-length SIRT7 was cloned into a smaller shuttle vector pOK12 (2.1 kbp) . To delete the candidate localization signals from full-length protein, inverse PCR with primers flanking the putative localization signals was performed as described previously. The cDNA was then re-cloned into the XhoI and EcoRI sites of plasmid EGFP-C3. Site-directed mutagenesis was performed using the SIRT7-pOK12 plasmid, and the mutant SIRT7 was cloned into plasmid EGFP-C3. The chimeric EGFP constructs containing the candidate NLS (amino acids 61–76, LQGRSRRREGLKRRQE) and NoLS (amino acids 392–400, KRTKRKKVT) were made using PCR primers containing these sequences. EGFP-C3 was used as template for PCR. The PCR product was then cloned into the EcoRI and XhoI sites of the pcDNA3.1 vector (Invitrogen, Carlsbad, CA, USA).
Replicative senescence and senescence-associated β-galactosidase assay
For replicative senescence, young primary fibroblasts TIG3 and WI38 were cultured on poly-l-lysine-coated cover slips until they reached confluence, after which they were split and passaged again. Cell morphology was monitored at each passage by light microscopy until the cells lost their typical elongated fibroblast morphology and acquired an enlarged senescent phenotype. Cells with passage numbers between 5 and 10 were considered young, and those with passage numbers between 10 and 20 and 25 and 35 were classified as pre-senescent and senescent cells, respectively. Both the young and growth-arrested senescent cells were fixed in 4% paraformaldehyde and stored at 4 °C for immunocytochemistry at a later stage. The senescent cells were maintained for at least 3 weeks before immunocytochemistry was performed. Senescence was also confirmed by the senescence-associated β-galactosidase assay as described previously .
The following SIRT7 antibodies were used: (a) S5947 (Sigma) which recognizes the N-terminal amino acids 35–51 (ILRKAAAERSAEEGRLL), and (b) H1, which recognizes amino acids 9–25 (SERKAAERVRRLREEQQ), and was a kind gift from I. Horikawa (National Institutes of Health). The nucleolar antibody (anti-nucleolar antigen) was obtained from BioGenex (Fremont, CA, USA). For immunocytochemistry, cells were fixed either with methanol/acetone (1 : 1) or 4% paraformaldehyde/0.1% Triton X-100. Blocking was performed in 2% BSA in NaCl/Pi, and incubation with primary antibody was followed by detection using Alexa Fluor 488- or 594-conjugated secondary (anti-mouse/rabbit) antibodies (Molecular Probes/Invitrogen, Carlsbad, CA, USA) and mounted in Vectashield mounting media containing 4,6-diamidino-2-phenylindole (DAPI) (Molecular Probes). Images were obtained using either a laser scanning confocal LSM510 microscope (Zeiss, Oberkochen, Germany) or a fluorescence inverted microscope (Olympus 1X51, Tokyo, Japan).
Cells were lysed in 20 mm Tris, pH 7.5, 150 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1% v/v NP-40 and protease inhibitor cocktail (Roche, Penzberg, Germany). Equal amounts of proteins were separated by SDS/PAGE and then transferred to poly(vinylidene difluoride) membranes. The blots were blocked with blocking buffer containing 5% non-fat dry milk in NaCl/Pi containing 0.1% Tween 20 for 30 min, followed by incubation overnight at 4 °C with primary antibodies H1 and S5947. Detection was performed using horseradish peroxidase-conjugated secondary antibody (Bangalore Genei, Peenya, India) and developed using ECL Prime reagent (GE Healthcare, Little Chalfont, UK) according to the manufacturer's instructions. All the buffer reagents were from Sigma unless mentioned otherwise.
SIRT7 knockdown and peptide competition assay for checking the specificity of antibodies
SIRT7 siRNA duplexes targeting 5′-GCCUGAAGGUUCUAAAGAA-3′ (sense) and 5′-UUCUUUAGAACCUUCAGGC-3′ (antisense) were obtained from Sigma. siRNA duplexes (15 and 150 nm) were transfected using Lipofectamine 2000 (Invitrogen). As a control, we used the siGENOME non-targeting siRNA control pool (Dharmacon, Lafayette, CO, USA). The cellular homogenates were then subjected to SDS/PAGE immunoblot analysis as described above using the anti-SIRT7 antibodies H1 and S5947. Transient knockdown assays were performed using shRNAs targeting four regions of SIRT7 mRNA, i.e. 807–827, 697–717, 294–309 and 395–405 bp (Mission shRNA; Sigma). Non-targeting shRNAs (Sigma) were used as controls. Stable SIRT7 knockdown U2OS and TIG3 cell lines were generated by transducing lentiviral particles carrying various target SIRT7 shRNAs. The viral particles were packaged by transfecting SIRT7 shRNAs together with lentiviral packaging mix (Sigma) in HEK 293 cells according to the manufacturer's instructions.
SIRT7-specific synthetic peptides corresponding to either amino acids 9–25 (SERKAAERVRRLREEQQ) or amino acids 35–51 (ILRKAAAERSAEEGRLL) were obtained from GenScript (Piscataway, NJ, USA) or SCL Biosciences (Bangalore, India), respectively. The antibody (1 μg) was incubated with or without a 200-fold molar excess of the peptide, and incubated for 1 h at room temperature in blocking buffer containing 5% non-fat dry milk. The pre-incubated antibody was then used for immunoblotting or immunocytochemistry as described above.
Cells were grown to 70% confluency and then collected by trypsinization. Cell fractionation was then performed using a ProteoJET™ cytoplasmic and nuclear protein extraction kit (Fermentas, Vilnius, Lithuania) according to the manufacturer's instructions. Lamin B (Santa Cruz Biotechnology, Dallas, Texas, USA) and α-tubulin (Sigma) were used as markers to examine the purity of the fractionation procedure.
Archived normal colon sample paraffin blocks were sectioned and de-paraffinized, and antigen retrieval was performed using Citra plus antigen retrieval solution (BioGenex, Fremont, CA, USA). Peroxide and power block were performed using a BioGenex kit according to the manufacturer's instructions. Serial sections from the same sample was incubated with the primary antibody S5947 (1 : 100) or the H1 antibody (1 : 150) overnight at 4 °C. Later, super enhancer (BioGenex) was added for 20 min, followed by polymer-horseradish peroxidase secondary antibody for 30 min. Finally, the colour was developed by treatment with diaminobenzidine) as a chromogen. Sections were counter-stained using haematoxylin, and then dehydrated, cleared with xylene and mounted in DPX mountant (HiMedia, Mumbai, India). Negative controls were performed using the secondary antibody only.
We thank I. Horikawa (National Institutes of Health) for providing the H1 SIRT7 antibody and pcDNA-DEST47-SIRT7 construct, and Dr Suresh Thakur (BioGenex, Hyderabad, India) for help with immunohistochemistry. We thank Manjari Sinha (Lab of Computational Biology, CDFD) for helping in bioinformatics analysis. We thank Professor T. Ramasarma (Honorary Scientist, Indian National Science Academy, New Delhi, India) and Dr Lucy M. Anderson (National Cancer Institute, NIH, USA) for critically reading the manuscript. S.K. is supported by a fellowship from the Department of Biotechnology, New Delhi, India, and acknowledges Hyderabad Central University, India, for PhD registration. The study was supported by funds from the Department of Biotechnology, New Delhi, India, and core funding from Center for DNA Fingerprinting and Diagnostics (CDFD). The help received from members of instrumentation facility, imaging facility and sequencing facility of Center for DNA Fingerprinting and Diagnostics (CDFD) is acknowledged. We also thank the anonymous reviewers for their suggestions and comments, which helped us to improve the manuscript.
All the experiments were performed by S.K. N.C. helped with senescent assays and S.S. performed immunohistochemistry. The SIRT7 immunocytochemistry following replicative senescence in primary fibroblasts was performed under the guidance of S.C.K. and R.W. G.R. designed the study and wrote the manuscript with input from S.K., S.C.K. and R.W.