The Nucleolar Localization Domain of the Catalytic Subunit of Human Telomerase*

Telomerase is the enzyme essential to complete the replication of the terminal DNA of most eukaryotic chro-mosomes. In humans, this enzyme is composed of the telomerase reverse transcriptase (hTERT) and telomerase RNA (hTR) subunits. hTR has been found in the nucleolus, a site of assembly of ribosomes as well as other ribonucleoproteins (RNPs). We therefore tested whether the hTERT component is also found in the nucleolus, where it could complex with the hTR RNA to form a functional enzyme. We report here that hTERT does indeed localize to the nucleolus, and we mapped the domain responsible for this localization to the hTR-binding region of the protein by deletion analysis. Substitution mutations in two of the three conserved hTR-binding domains in this nucleolar localization domain (NoLD) abolished nucleolar localization. However, another mutation that impeded hTR binding did not alter this subcellular localization. Additionally, wild type hTERT was detected in the nucleolus of cells that failed to express hTR. Taken together, we propose that the nucleolar localization of hTERT involves more than just the association with the hTR subunit. Furthermore, the coincidental targeting of both the hTR and hTERT subunits to the nucleolus supports the premise that the assembly of telomerase occurs in the nucleolus.

Telomerase is a reverse transcriptase ribonucleoprotein (RNP) 1 complex composed of a reverse transcriptase catalytic protein subunit (TERT) that copies a template region of an accompanying RNA subunit (TR) onto telomeres as DNA (1). In humans, this enzyme is of great medical importance because of its pivotal role in unlimited cellular proliferation, a hallmark of cancer cells (2). The union of the RNA and protein subunits to form an RNP is essential for telomerase activity (3).
Based on a comparison of the amino acid sequence of TERT from organisms of many different kingdoms and on mutational analysis, it has been possible to identify a number of discrete domains in TERT. The central region of the catalytic subunit contains seven motifs found in reverse transcriptases, which define the catalytic core (4 -17). The C terminus of TERT is highly divergent, both at the sequence and the functional level (18,19). The N-terminal region is more conserved, containing domain I and the DAT (dissociates activities of telomerase) domain (18,20) followed by domains II and III (18,20,21) and the T motif (1,5,6), which are essential for telomere elongation. Substitution mutations in domains II and III or the T motif decrease TERT binding to the telomerase RNA in yeast (18), ciliate (22,23), or human cells (20,24) and correspondingly result in a dysfunctional enzyme. Deletion analysis has also defined the region extending from amino acids 326 to 613, which harbors all three of the aforementioned domains, as the minimum region required for hTR binding (22), although mutations in domain I can also have some effect on hTR binding (24).
The site of assembly of the telomerase RNP has not been determined; however, there is growing evidence supporting a connection to the nucleolus in vertebrate systems. The nucleolus is well known as the site of ribosome assembly and has been speculated to be a site for the assembly of other RNPs (25)(26)(27). The hTR RNA component of human telomerase contains a sequence/structure motif characteristic of Box H/ACA small nucleolar RNAs, which guide rRNA processing and modification within the nucleolus (28 -30). Moreover, hTR co-immunoprecipitates the small nucleolar RNA-binding proteins dyskerin (31), GAR1 (32), NHP2, and NOP10 (33,34), indicating that hTR can exist in a complex with these nucleolar proteins. Lastly, hTR has been found to localize to the nucleolus by virtue of its Box H/ACA motif (28 -30). However, the same may not be true in lower eukaryotes. For example, yeast telomerase RNA lacks a Box H/ACA motif (28) but instead shares features of spliceosomal small nucleolar RNAs (35,36). Therefore, unlike other functions of telomerase, understanding the biogenesis of this enzyme in humans may be achieved only by studying this process in higher eukaryotes. This is of particular importance because improper telomerase RNP accumulation has been linked to the human disease dyskeratosis congenita (31,37).
Given that hTR is found in the nucleolus, a site for the assembly of ribosomes and possibly other RNPs (25)(26)(27), we tested whether the hTERT catalytic subunit is also found in this subnuclear compartment by monitoring the subcellular localization of hTERT when fused to the yellow fluorescent protein (YFP) in human cells. We found that hTERT localized to the nucleolus, and we mapped the corresponding nucleolar localization domain to the region encompassing the N-terminal domains II and III and the T motif. Although point or substitution mutations in any of these three domains decreased hTR binding in vitro, not all of these mutations disrupted hTERT nucleolar localization, suggesting that localization can occur independent of hTR binding. Indeed, hTERT was found in the nucleoli of human cells that do not express hTR. These results demonstrate that hTERT contains a discrete nucleolar localization domain (NoLD) that targets the molecule to the nucleolus and that this subcellular distribution can be mediated through interactions with factors other than hTR.

EXPERIMENTAL PROCEDURES
Cell Culture-The human osteosarcoma cell line U2OS, the human cervical adenocarcinoma cancer cell line HeLa, and the SV40 transformed human fibroblast cell lines WI38 VA13/2RA (American Tissue Type Collection) and LM217 (38) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum.
Plasmids-pYFP-hTERT, hTERT-II 386 , or TERT-III 512 were made by subcloning the EcoRI/SalI FLAG-hTERT cDNA either in the wild type format or with a six-amino acid (NAAIRS) substitution of the sequence 386 YWGMRP in domain II or 512 MSVRGC in domain III, respectively (20), into the same sites of plasmid pEYFP-C1 (CLONTECH). FLAGtagged NoLD was made by PCR-amplifying hTERT cDNA with the primers 5Ј-GGAATTCGCCACCATGGACTACAAAGACGATGACGAC-AAGCTGAGGCCCAGCCTG-3Ј and 5Ј-TGCGGTCGACTCATCTGGAC-GTCAGCAG, digesting the product with EcoRI/SalI, and cloning the fragment into the same sites in plasmid pEYFP-C1, which had a modified polylinker to accommodate this reading frame or the same sites of pCI-neo (Promega) for in vitro synthesis. The same six amino acids described above beginning at positions 386 and/or 512 were substituted with the sequence NAAIRS, and/or amino acid Phe 561 in the T motif was mutated to Ala in the NoLD fragment by site-directed mutagenesis as described previously (20), and the resultant products were confirmed to be correct by direct sequencing. These products were again subcloned into the EcoRI/SalI sites of the modified pEYFP-C1 plasmid or pCI-neo. hTERT deletion fragments were generated by PCR with the primers 5Ј-TCCCCGCGGAAGCTTGCACCATGCCGCGCGCTCCCCGC-3Ј and 5Ј-ATTGGATCCATGGCCTGAGTGGCAGCGCC-3Ј (1-183), 5Ј-CGCG-GAAGCTTGCTACCAGCTCGGCGCTGCC-3Ј and 5Ј-ATTGGATCCAT-CCAGTGCAGGAACTTGGC-3Ј (170 -546), 5Ј-CGCGGAAGCTTGCCC-AGGGGTTGGCTGTGTT-3Ј and 5Ј-CCGCTCGAGTCGACTAGTGGGC-CGGCATCTGAAC-3Ј (523-924), and 5Ј-CGCGGAAGCTTGCGATGAT-TTCTTGTTGGTG-3Ј and 5Ј-ATTGGATCCATGTCGACTCAGTCCAG-GATGGTC-3Ј (867-1132), confirmed to be correct by direct sequencing, digested with HindIII/SalI or HindIII/BamHI, and cloned into the same sites in pEYFP-C1.
hTR-hTERT Co-immunoprecipitation-As described previously (20), hTR was transcribed and 32 P-labeled with the T7-coupled Maxiscript kit (Ambion) using 1 g of linearized pBluescriptSK-hTR plasmid (20), after which unincorporated nucleotides were removed by using a G-25 minispin column (Amersham Biosciences). FLAG-tagged NoLD, either wild type or with the described mutations, was expressed from the T7 promoter of pCI-neo and 35 S-labeled in the presence of 1 l of hTR using the T7 quick coupled TNT system (Promega). As a negative control, FLAG-HDAC1 was expressed from the T7 promoter of pCMV-HDAC1. 10 l of anti-FLAG M2-agarose (Sigma) was used for immunoprecipitation after preblocking with 100 ng/ml bovine serum albumin, 100 ng/ml casein, 100 ng/ml tRNA, 250 ng/ml yeast total RNA, and 100 ng/ml glycogen in PBS supplemented with 1.5 mM dithiothreitol, 0.5% CHAPS, 1 mM benzamidine, and 0.1 mM phenylmethylsulfonyl fluoride. TNT reactions were diluted in supplemented PBS with nonspecific blockers and 200 units of RNasin (Promega) and immunoprecipitated at room temperature for 1 h with M2-agarose. The agarose beads were then washed three times with prechilled supplemented PBS, heated in SDS buffer, and resolved by SDS-PAGE.
Detection of hTR RNA-Water or 100 ng of total RNA isolated with the RNazol reagent according to the manufacturer's instructions (Tel-Test) from WI38 VA13/2RA or telomerase-positive PC3 (39) cells was reverse transcribed and PCR amplified to detect either total hTR or porphobilinogen deaminase (PBGD) mRNA using the LightCycler TeloTAGGG hTR quantification kit and the LightCycler system in accordance with the manufacturer's instructions (Roche Molecular Biochemicals).
Visualization of YFP-tagged Proteins and Fibrillarin-Localization of YFP fusion proteins was visualized in U2OS, LM217, HeLa, or WI38 VA13/2RA cells grown on coverslips coated with 100 mg/ml poly-Dlysine, M r Ͼ 300,000 (Sigma). Cells were transiently transfected with the above described plasmids encoding the appropriate YFP fusion protein using either the FuGENE 6 (Roche Molecular Biochemicals) or LipofectAMINE 2000 (Invitrogen) reagents according to the manufacturers' protocols. After 48 h, live U2OS or LM217 cells were observed under PBS at ϫ400 magnification on a Zeiss Axioskop fluorescence microscope. VA13 and HeLa cells were fixed in 4% formaldehyde in PBS for 10 min at room temperature, washed twice with PBS, mounted, and observed at ϫ630 magnification on a Zeiss Axiovert S100 inverted fluorescence microscope. To visualize fibrillarin, U2OS cells transfected with hTERT expression constructs were similarly fixed and then per-FIG. 1. hTERT is found in the nucleolus. A, an example of U2OS cells transiently expressing YFP-hTERT and stained with an antifibrillarin antibody to detect the nucleolar protein fibrillarin (left) or viewed as a fluorescence image to detect YFP-hTERT (middle) or a merge (right) of both images. B, an example of a U2OS, LM217, or HeLa cell transiently expressing YFP-hTERT is shown as a differential interference or phase contrast image (left) to visualize nucleoli (arrows) or as a fluorescent image (right) to visualize the YFP-tagged protein. a Only cells with nuclear YFP were scored (full-length hTERT was found in the nucleus of ϳ80% of transfected cells). Wild type full-length hTERT was found in the nucleolus and the nucleolusϩnucloplasm; hence, only cells with such staining were scored as positive for being in the nucleolus. meabilized with 0.5% Nonidet P-40 in 1ϫ PBS and incubated with the human polyclonal anti-fibrillarin antibody (40) at 1:500 dilution. The primary antibody was detected with the rhodamine (TRITC)-conjugated donkey anti-human IgG antibody (Jackson ImmunoResearch Laboratories) and visualized as above.

RESULTS AND DISCUSSION
hTERT Is Found in the Nucleolus-To address whether hTERT is found in the nucleolar compartment of the cell, YFP was fused in-frame with the N terminus of full-length hTERT. We chose to monitor hTERT localization by detecting YFP because the fusion of large polypeptides at the N terminus has no measurable effect on telomerase activity (Ref. 41 and data not shown), YFP-tagged proteins can be visualized in live cells (42), and there is an absence of antibodies readily capable of detecting endogenous hTERT at the subcellular level. The YFP-hTERT construct was transiently expressed in the human osteosarcoma cell line U2OS because these cells lack telomerase activity (43), which eliminates possible interference by multimerization with the endogenous hTERT protein (20, 44 -46). The subcellular distribution of YFP-hTERT was then assayed by fluorescence light microscopy. We found that YFP-hTERT is located predominantly in the nucleolus or both the nucleolus and nucleoplasm as assessed by co-localization with   Fig. 3A for a description of the fusion constructs. b Wild type NoLD was found primarily in the nucleolus; hence, only cells with predominant nucleolar staining were scored as positive for being in the nucleolus. the nucleolar protein fibrillarin detected by indirect immunofluorescence (40) or by co-localization with nucleoli as identified using differential interference or phase contrast optics (Fig. 1, A and B, and Table I).
To rule out the possibility that this observation was unique to these cells, we introduced YFP-hTERT into another telomerase negative human cell line, LM217 (38). Consistent with our observations using U2OS cells, YFP-hTERT was found predominantly in the nucleolus. YFP-hTERT was even detected in the nucleolus of the telomerase-positive line HeLa (Fig. 1B and Table  I). We therefore conclude that hTERT is found in the nucleolus and that this event is independent of cell type.
The Nucleolar Localization Domain of hTERT Encompasses Domains II and III and the T Motif-To map the region of hTERT required for nucleolar localization, we fused YFP to a series of fragments that represent key regions of hTERT and collectively span the entire length of the protein (Fig. 2A). Each of these constructs was introduced into U2OS cells, and Ͼ100 transfected cells were scored for subcellular localization. Fragment 1-183, which encompasses domain I and the DAT domain, did not localize to the nucleolus. Similarly, fragment 523-924, encoding the T motif and all the reverse transcriptase domains, or fragment 867-1132, encompassing the entire C terminus, also failed to localize to the nucleolus (Fig. 2B). Since all of these fragments were detected in the nucleus, we discount the possibility that the peptides were excluded from the nucleolus due to a failure to enter the nucleus. We therefore surmise that neither the DAT domain, the catalytic core, nor the C terminus of hTERT contains a nucleolar localization domain. However, fragment 326 -620, which encodes the hTR-binding domains of hTERT (domains II and III and the T motif), was always found in the nucleolus (Fig. 2B and Table II), indicating that this fragment contains sequences sufficient for nucleolar targeting. We found that the overlapping fragment 170 -546,

FIG. 3. Mutations in the nucleolar localization domain that disrupt hTR binding.
A, top, a scale representation of the NoLD, the minimum fragment (326 -620) found to be required for nucleolar localization. Black boxes denote the most conserved regions, domains II and III and the T motif, whereas shading denotes bordering sequences that are less conserved, and white shading identifies regions known to be dispensable for enzyme activity. The positions and sequence of specific amino acids mutated are shown below. Bottom, diagram depicting the size and position of mutations of the NoLD fragments fused to YFP. On the right are listed the names of the mutants. aa, amino acids. B, wild type (WT) and the described mutants of the FLAG-tagged NoLD domain of hTERT were 35 S-labeled and made in the presence of 32 P-labeled hTR. The irrelevant protein HDAC-1 was used to assess the level of non-specific RNA binding. RNA-protein complexes were immunoprecipitated by an anti-FLAG antibody, and products were resolved on an SDS-PAGE. RNA input is represented as 1/100 of the post-immunoprecipitation lysate. which contains domains II and III, or the fragment encompassing the catalytic region (523-924), which also contains part of domain III and the T motif, was not localized to the nucleolus. We therefore reasoned that at least two elements are required for nucleolar targeting of the 326 -620 region encompassed by domain II to the T motif. We term this portion of hTERT the nucleolar localization domain (NoLD). We next confirmed that the NoLD was localized to the nucleolus in other cell types. Specifically, we expressed the YFP-NoLD polypeptide in LM217 and HeLa cells and again found the YFP-tagged protein in the nucleolus of 99 -100% of the cells (Table II). Thus, the NoLD encodes a potent nucleolar localization sequence that functions regardless of cell type.

Mutations in Domain II and the T Motif Affect the Nucleolar Targeting of hTERT-The
NoLD encompasses conserved sequences denoted as domains II, III, and the T motif. These domains also map to regions determined by mutational analysis to be essential for telomerase activity in humans, yeast, or ciliates (18, 20 -23, 45, 47). We have previously shown that a six-amino acid substitution (with the sequence NAAIRS) beginning at position 386 in domain II or 512 in domain III of hTERT reduces the association of this protein with hTR (20). Similarly, an alanine substitution of the highly conserved phenylalanine residue in the T motif is known to decrease telomerase RNA binding in ciliates (23). Collectively, these data argue that these three domains are responsible for most telomerase RNA binding. Because nucleolar localization was detected with a fragment encompassing the sequences encoding the hTR RNAbinding activities of hTERT (but not with fragments outside this region) and hTR is known to localize to the nucleolus (28 -30), we queried whether the RNA-binding activity of hTERT was essential for the observed nucleolar localization of hTERT. We introduced the aforementioned mutations in domains II, III, and the T motif one at a time or in combination into the NoLD fragment of hTERT (Fig. 3A). We then confirmed that the NoLD fragment harboring mutations in domains II and III and the T motif abolished hTR binding in vitro. Specifically, 35 S-labeled FLAG-tagged NoLD fragments, either in the wild type format or with the 386, 512, or the 561 mutations, were incubated with 32 P-labeled hTR and immunoprecipitated with an anti-FLAG antibody. The FLAG-tagged NoLD polypeptide, but not an irrelevant FLAG-tagged protein (HDAC1), co-immunoprecipitated hTR. However, mutations in any of the three hTR-binding domains abolished this association with no further loss observed, even when all three mutations were introduced into the same polypeptide (Fig. 3B). Thus, we confirm that mutations 386, 512, and 561 abolish detectable hTR association with the NoLD in vitro.
Having confirmed that the mutations abolish detectable hTR binding, we next monitored the subcellular localization of the YFP-tagged proteins in U2OS cells by fluorescence microscopy. We have found that a substitution mutation in domain II (mutant 386), which eliminated hTR binding and is known to abolish the catalytic function of hTERT ( Fig. 3B and Ref. 20), greatly reduced the nucleolar localization of the NoLD fragment. In Ͼ400 cells scored, the mutated YPF-NoLD fragment was found in the nucleolus only one-third of the time, a 3-fold drop compared with the wild type NoLD fragment. Similarly, a mutation in the T motif (561) crippled the ability of the NoLD fragment to localize to the nucleolus by 4-fold. Interestingly, mutating domain III (512) had little effect on the nucleolar localization of the NoLD fragment ( Fig. 4 and Table II). We discount the possibility that the localization of these three mutant NoLD proteins was dependent upon cell type, because all of these proteins were similarly localized when expressed in LM217 or HeLa cells (Table II). Thus, disrupting two of the three hTR-binding domains altered the nucleolar targeting by the NoLD, and these effects were independent of cell type.
Mutations in More than One hTR-binding Domain Do Not Further Disrupt Nucleolar Localization-We next addressed whether mutations in more than one hTR-binding region fur- ther decreased nucleolar localization. YFP was fused to a series of NoLD fragments containing different combinations of domains II and III and T motif mutations (Fig. 3A). The resultant constructs were then expressed in U2OS cells and assayed for nucleolar localization. We found that mutating either two or all three hTR-binding domains did not have an additive effect. A mutation in domain II disrupted the nucleolar localization of the NoLD fragment to approximately the same degree whether domain III or the T motif or both were also mutated ( Table II). The same was true with the T motif; additional mutations did not further diminish nucleolar localization (Table II). These results are consistent with a model whereby mutations in domains II or the T motif are alone capable of disrupting the accumulation of hTERT in the nucleolus.
The NoLD Is Required for Nucleolar Localization of Fulllength hTERT-Mutation in domain II greatly reduced the nucleolar accumulation of the NoLD fragment when expressed in human cells. To directly test whether this domain was essential to target full-length hTERT to the nucleolus, we created a fusion of the YFP with hTERT containing the aforementioned NAAIRS substitution mutation in domain II. The fusion protein was transiently expressed in U2OS cells and assayed for subcellular localization. Although the full-length hTERT molecule was found in the nucleolus, the introduction of a NAAIRS substitution mutation at position 386 in domain II reduced this localization by almost one-half ( Fig. 5 and Table I). Thus, we conclude that the NoLD identified by deletion analysis mediates the localization of target full-length hTERT to the nucleolus. hTERT Nucleolar Localization Is Independent of hTR Binding-We have shown that the nucleolar localization domain of hTERT maps to the region of the protein linked to hTR binding (Fig. 2) and that mutations in two known hTR-binding domains, domain II and the T motif, crippled the ability of the protein to accumulate in this compartment of the nucleus. Such observations are consistent with the hTR recruiting hTERT to the nucleolus via the Box H/ACA motif. However, one mutation that abolished all measurable association of hTERT with hTR ( Fig. 3B) had very little effect on the accumulation of the NoLD fragment (Fig. 4) or full-length hTERT (Table I) in the nucleolus. We interpret these data in one of two ways. Interaction with hTR may not be required for targeting hTERT to the nucleolus. Alternatively, hTR binding may be essential for hTERT localization, but the mutation in domain III does not disrupt hTR binding to the same extent in vivo as observed in vitro. To differentiate between these two models, YFP-tagged full-length hTERT and the NoLD fragment were introduced into WI38 VA13/2RA human cells. These cells reportedly lack both hTERT and hTR transcripts (48,49) and require the ectopic expression of these two subunits to restore telomerase activity (49). Indeed, we confirm by the exquisitely sensitive real-time quantitative reverse transcription-PCR that the hTR transcript is not present in these cells despite being readily detected in telomerase-positive cells (Fig. 6A). If hTR binding is indispensable for the nucleolar localization of hTERT, then the hTERT or NoLD proteins should not be detected in the nucleolus when expressed in these cells. However, we find that ectopically expressed YFP-hTERT as well as the YFP-NoLD itself was found almost exclusively in the nucleolus, supporting the premise that hTERT can be targeted to this subcellular structure independent of hTR binding. Moreover, a mutation in the T motif (561) that drastically reduced nucleolar accumulation of YFP-NoLD protein in the U2OS and LM217 cells was similarly effective in the cells lacking hTR (Fig. 6B and Table  II). These observations indicate that hTR is not essential for the localization of hTERT to nucleoli.
Conclusions-We now show that hTERT is localized to the nucleolus when transiently expressed in human cells. Because both the hTR and hTERT subunits localize to this structure, we suggest that this localization may reflect a part of telomerase biogenesis such as the assembly of the hTR and hTERT subunits into an RNP. However, it is possible that the targeting of hTERT to the nucleolus may have other functions such as the sequestration of telomerase from its telomeric target (30). The minimal fragment of hTERT defined by deletion analysis that mediates localization extends from amino acids 326 to 620. The most striking feature of this NoLD is that it encompasses all of the known hTR binding elements of hTERT. Furthermore, mutating two of the three RNA-binding domains of hTERT reduced the localization of either the NoLD or the full-length hTERT to the nucleolus. Thus, it initially appeared that the hTR served as a bridge, tethering hTERT protein to the Box H/ACA motif for nucleolar localization. However, we also found that a mutation in hTERT that disrupts hTR binding did not grossly affect the ability of the protein to accumulate in the nucleolus and that hTERT is found in the nucleolus of cells lacking hTR. The simplest interpretation of these data is that the NoLD can target hTERT to the nucleolus in a manner that is independent of hTR binding. Thus, the targeting or retention of hTERT within the nucleolus may involve other factors that interact specifically with domain II and the T motif within the NoLD of hTERT.