1. Biochemistry and Chemical Biology
  2. Cell Biology
Download icon

Identification of human TERT elements necessary for telomerase recruitment to telomeres

  1. Jens C Schmidt
  2. Andrew B Dalby
  3. Thomas R Cech  Is a corresponding author
  1. Howard Hughes Medical Institute, University of Colorado Boulder, United States
Research Article
  • Cited 45
  • Views 3,286
  • Annotations
Cite this article as: eLife 2014;3:e03563 doi: 10.7554/eLife.03563

Abstract

Human chromosomes terminate in telomeres, repetitive DNA sequences bound by the shelterin complex. Shelterin protects chromosome ends, prevents recognition by the DNA damage machinery, and recruits telomerase. A patch of amino acids, termed the TEL-patch, on the OB-fold domain of the shelterin component TPP1 is essential to recruit telomerase to telomeres. In contrast, the site on telomerase that interacts with the TPP1 OB-fold is not well defined. In this study, we identify separation-of-function mutations in the TEN-domain of human telomerase reverse transcriptase (hTERT) that disrupt the interaction of telomerase with TPP1 in vivo and in vitro but have very little effect on the catalytic activity of telomerase. Suppression of a TEN-domain mutation with a compensatory charge-swap mutation in the TEL-patch indicates that their association is direct. Our findings define the interaction interface required for telomerase recruitment to telomeres, an important step towards developing modulators of this interaction as therapeutics for human disease.

https://doi.org/10.7554/eLife.03563.001

eLife digest

In the nucleus of a cell, the DNA that contains the cell's genetic information is packaged into long structures called chromosomes. Every time a cell divides, its chromosomes are duplicated. However, the proteins that are responsible for copying the DNA cannot reach the very end of the DNA strand, causing the chromosomes to progressively shorten. To ensure that this does not cause genetic information to be lost, each chromosome ends in a repetitive stretch of DNA called a telomere. Though the end of the telomere is lost whenever the DNA is copied, an enzyme called telomerase replaces the sequence that has been lost and counteracts the shortening of the telomeres.

Shelterin is a protein complex that binds to telomeres to protect them and also helps telomerase to work correctly. Shelterin contains a specific site that attaches to telomerase, but exactly how the human versions of these two molecules bind to each other is unknown. A possible interaction site had been identified on the telomerase, which, when mutated, stops the telomerase working properly. However, as this region is also involved in lengthening the telomeres after the chromosomes have duplicated, it is not certain that these problems result from the telomerase failing to bind to shelterin.

The enzyme telomerase is unusual; it has both RNA and protein components. Like all other proteins, the telomerase protein is made up of strings of amino acids. Schmidt et al. discovered that replacing two specific amino acids in human telomerase prevents its binding to shelterin. Cells that produced the modified form of the telomerase had chromosomes with shortened telomeres. However, if the cells also produced modified versions of the shelterin complex that were designed to bind to the modified telomerase, telomere length was normal. This indicates that the telomerase interacts directly with shelterin, rather than through a ‘bridging’ molecule.

Mutations in the genes coding for both shelterin and the telomerase enzyme cause a number of human diseases, and cancers rely on the activity of telomerases to grow. Knowing how shelterin and telomerase interact could therefore help to design drugs that may either restore or disrupt the interaction and therefore can be used to treat these diseases.

https://doi.org/10.7554/eLife.03563.002

Introduction

Linear human chromosomes end in G-rich telomeric repeat DNA with a single-stranded overhang (Blackburn, 2005). Chromosome ends pose two challenges to cells: they resemble DNA damage sites, and they shorten progressively each cell cycle due to the end-replication problem (Levy et al., 1992). The six-protein shelterin complex specifically binds telomeric DNA and prevents the DNA damage machinery from recognizing telomeres (Palm & de Lange, 2008). Additionally, shelterin recruits telomerase, an RNA-containing reverse transcriptase that adds telomeric repeats to the 3ʹ-end of the single-stranded overhang to maintain telomere length (Nandakumar and Cech, 2013). Loss-of-function mutations in telomerase and shelterin components cause human diseases like dyskeratosis congenita, aplastic anemia, and pulmonary fibrosis (Armanios and Blackburn, 2012). In addition, like stem cells and germ cells, 90% of cancers depend on telomerase activity to maintain the potential to proliferate indefinitely (Stewart and Weinberg, 2006; Shay and Wright, 2011).

Telomerase is a ribonucleoprotein (RNP) complex. Its catalytic core is comprised of hTERT and the telomerase RNA component (hTR), which contains the template used for telomeric DNA synthesis (Cech, 2004). In the holoenzyme, hTR also associates with dyskerin for stability and with TCAB1 for telomerase maturation and trafficking to Cajal bodies (Mitchell et al., 1999; Venteicher et al., 2009). The hTERT protein contains four domains, the telomerase N-terminal domain (TEN-domain), the telomerase RNA-binding domain (TRBD), the reverse transcriptase domain (RT-domain), and the C-terminal extension (CTE) or putative thumb domain of the reverse transcriptase. The TRBD- and RT-domains of hTERT bind hTR and position the template of hTR in the RT-domain active site to synthesize telomeric repeats (Cech, 2004). The TEN-domain participates in the catalysis of telomeric repeat addition (Jacobs et al., 2006; Jurczyluk et al., 2011; Robart and Collins, 2011; Wu and Collins, 2014). Additionally, both the TEN-domain and CTE contain DAT (dissociates activities of telomerase) regions, N-DAT and C-DAT, respectively (Armbruster et al., 2001; Banik et al., 2002). Mutations in either DAT-region disrupt the ability of telomerase to immortalize cells, but they retain catalytic activity (Armbruster et al., 2001; Banik et al., 2002). Based on experiments with N-DAT mutants, the TEN-domain has been proposed to mediate telomerase recruitment to telomeres (Armbruster et al., 2003, 2004).

Telomerase is recruited to telomeres in S-phase, while it remains in Cajal bodies for the remainder of the cell cycle (Jady et al., 2006; Tomlinson et al., 2006). Telomerase localization to Cajal bodies requires the interaction of hTR with TCAB1 (Venteicher et al., 2009). TCAB1 is not required for enzymatic activity but for function in vivo (Venteicher et al., 2009), indicating that recruitment to Cajal bodies is a key step in telomerase trafficking or maturation. Recruitment of telomerase to the telomere requires the shelterin component TPP1 (Xin et al., 2007; Abreu et al., 2010). TPP1 is recruited to the shelterin complex via an interaction with TIN2 (Abreu et al., 2010). TPP1 also tightly binds to POT1 (Wang et al., 2007), which specifically binds the telomeric single-stranded DNA overhang (Baumann and Cech, 2001; Lei et al., 2004). Biochemical studies have demonstrated that the interaction between the POT1-TPP1 complex and telomerase stimulates its repeat addition processivity (RAP), that is, increases the number of consecutive repeats that telomerase synthesizes in a single round of telomeric DNA addition (Wang et al., 2007). Recently, the TEL-patch, a group of mostly acidic amino acids on the oligosaccharide binding (OB)-fold domain of TPP1 was shown to be necessary for telomerase recruitment in vivo and RAP stimulation in vitro (Nandakumar et al., 2012; Sexton et al., 2012; Zhong et al., 2012). Furthermore, cell biological assays demonstrated that the OB-fold domain of TPP1 is sufficient to recruit telomerase to a non-telomeric chromosomal locus (Zhong et al., 2012). Although TPP1 is critical for recruitment of telomerase to the telomere, it has been unclear whether TPP1 makes a direct interaction with telomerase or if their interaction is bridged by another component.

Conversely, the site on telomerase that associates with the telomere is not well defined. The TEN-domain of hTERT is a potential candidate to contain key residues necessary for the interaction with TPP1. A mutation in the TEN-domain of hTERT, G100V, is defective in stimulation of telomerase processivity in vitro (Zaug et al., 2010) and fails to localize to telomeres in vivo (Zhong et al., 2012). However, G100V telomerase has substantially reduced enzymatic activity (Zaug et al., 2010); thus, it is difficult to assess whether its failure to function in vivo is due to an activity or recruitment defect. Furthermore, the interaction between TPP1 and telomerase may require a bridging factor. In Schizosaccharomyces pombe, Ccq1 bridges the interaction between Tpz1, the TPP1 homologue, and telomerase (Miyoshi et al., 2008; Moser et al., 2011). A human homologue of Ccq1 has not been identified (Nandakumar and Cech, 2013). Thus, it is unclear if telomerase directly interacts with TPP1 in human cells.

In this study, we identify separation-of-function mutations in the TEN-domain of hTERT. The mutants retain high catalytic activity, but their interaction with POT1-TPP1 is compromised in vitro. Furthermore, the TEN-domain mutants fail to localize to telomeres and cannot maintain telomere length in vivo. Importantly, a compensatory mutation in the TEL-patch of TPP1 rescues a mutation in the TEN-domain of hTERT both in vitro and in vivo, indicating that the interaction between telomerase and the OB-fold domain of TPP1 is direct. These observations open the door to more directed approaches for targeting the TPP1-telomerase interface as a therapeutic strategy for human diseases.

Results

TEN-domain hTERT mutants defective for POT1-TPP1-mediated processivity enhancement

Conserved acidic amino acids on the surface of the TPP1 OB-domain (TEL-patch) are necessary for the recruitment of telomerase to telomeres (Nandakumar et al., 2012; Sexton et al., 2012; Zhong et al., 2012). We hypothesized that a corresponding region of basic amino acids would exist on the surface of hTERT, which directly associates with the TEL-patch via charge–charge and other interactions to recruit telomerase to telomeres (Figure 1A).

Figure 1 with 2 supplements see all
Identification of mutants in the TEN-domain of hTERT that affect the interaction with POT1-TPP1 in vitro but not enzymatic activity.

(A) Domain structures of hTERT and TPP1 proteins. Question mark, proposed interaction between the TEL-patch on TPP1 and the N-DAT region of the TEN-domain of hTERT. In TPP1, the OB-domain (OB) is followed by the POT1-binding domain (PBD), and finally the TIN2-binding domain (TIN2 BD). For hTERT, the Telomerase Essential N-terminal (TEN) domain is adjacent to the RNA binding domain (RBD), which precedes the reverse transcriptase domain (RT) and finally the C-terminal extension (CTE). (B) An alignment of the N-DAT region of the TEN-domain of TERT proteins from selected species. Basic residues conserved in greater than five of the seven species are highlighted blue. The conserved residue G100 is highlighted in orange above the alignment. (C) Western blot, probed for hTERT, showing the immuno-purification of telomerase over-expressed in HEK 293T cells. To monitor relative quantities of hTERT, equal fractions of lysate, flow through (FT), IgG bead capture (capture), and cleaved eluate (elution), were analyzed by SDS-PAGE. (D) Western blot of the relative quantities of wild-type and mutant hTERTs after immuno-purification of the telomerases. (E) Direct telomerase activity assays in the absence and presence of the POT1-TPP1 heterodimer (PT) for wild-type (WT) and mutant telomerases. LC, loading control. +4, oligonucleotide marker corresponding to the addition of the first four nucleotides to primer. Numbers on left, telomeric repeats added. (FH) Bar graphs representing the quantification of activity, RAP, and RAP stimulation (decay method) by wild-type POT1-TPP1. Values are normalized to WT telomerase, and to WT telomerase with WT POT1-TPP1 for RAP stimulation (n = 3, Mean ± SD).

https://doi.org/10.7554/eLife.03563.003

To identify candidate residues on the TEN-domain of hTERT that might interact with the TPP1 TEL-patch, we performed a multiple sequence alignment of TERT from higher eukaryotes (Figure 1B). The alignment revealed a number of conserved basic amino acids; we tested these, as well as some non-conserved basic amino acids in close proximity in the primary sequence. We generated a panel of hTERT mutants, replacing individual or combinations of basic amino acids on hTERT with the acidic amino acid aspartate. Mutant hTERTs and hTR were over-expressed in HEK293T cells and the telomerases were immuno-purified (Figure 1C). The hTERT variants were expressed at levels similar to those of wild-type telomerase (Figure 1D).

To identify separation-of-function hTERT mutants that do not interact with POT1-TPP1 but retain near wild-type telomerase activity in vitro, we carried out direct telomerase extension assays. (In these assays, dATP, dTTP, and 32P-dGTP are added to the 3ʹ end of a DNA oligonucleotide by immuno-purified telomerase and products are separated on a polyacrylamide sequencing gel and visualized by autoradiography. The enzymatic activity of each telomerase is determined by quantifying the total amount of 32P-dGTP incorporated into reaction products, and RAP is measured by analyzing the distribution of product lengths.) To measure the physical interaction between TPP1 and telomerase, we carried out direct telomerase assays with and without a previously described minimal POT1-TPP1 complex bound to substrate oligonucleotides (Wang et al., 2007) and calculated ‘RAP stimulation by PT’ (equation given in ‘Materials and methods’). RAP stimulation by PT depends on the association of telomerase with POT1-TPP1 and therefore provides a biochemical readout for this interaction.

In the hTERT mutant screen, some telomerases had defects in either activity (R72E, R143E, R142E;R143E), RAP (R87E;R91E;K94E, R120E) or had no effect (R142E) (Figure 1—figure supplement 1, Table 1). However, hTERT K78E retained wild-type activity and processivity (92% and 100% respectively), but RAP stimulation by POT1-TPP1 was reduced to 68% that of wild-type telomerase (Figure 1E–H; Figure 1—figure supplement 2). Additionally, the R132D mutant was reported to have wild-type activity but failed to localize to telomeres in vivo (Stern et al., 2012). We generated a R132E mutant hTERT, which retained moderate activity (73%) but was defective in RAP stimulation by POT1-TPP1 (24% of wild-type hTERT, Figure 1E–H; Figure 1—figure supplement 2). Furthermore, the K78E;R132E double mutant retained 63% activity but further reduced RAP stimulation by POT1-TPP1 to 18% of wild-type hTERT (Figure 1E–H). Two different methods of quantitating RAP gave comparable results (Figure 1—figure supplement 2).

Table 1

Telomerase TEN domain mutant activity, processivity, and RAP stimulation by wild-type TPP1

https://doi.org/10.7554/eLife.03563.006
TEN domain mutantActivity % of WT*Processivity % of WTRAP stimulation by PT % of WT
R72E48 ± 399 ± 191 ± 2
K78A19 ± 2101 ± 788 ± 2
K78E92 ± 1100 ± 168 ± 5
R87E;R91E;K94E74 ± 675 ± 0.361 ± 4
R120E999655
K78E;R120E82 ± 1887 ± 338 ± 2
R132E73 ± 292 ± 324 ± 2
R132E;K78E63 ± 1094 ± 818 ± 1
R142E97 ± 7111 ± 10103 ± 4
R143E16 ± 1281 ± 3N.D.
R142E;R143E4 ± 1N.D.N.D.
V144M48 ± 695 ± 544 ± 7
  1. *

    Percentage of wild-type telomerase activity, processivity, or RAP stimulation. Activity values normalized to hTERT levels, loading control, ± standard deviation for 2 or more replicates.

  2. Repeat addition processivity (RAP) stimulation upon addition of WT POT1-TPP1, values relative to WT telomerase with WT POT1-TPP1 (i.e. RAP stimulation by PT % of WT = ((WT PT RAP stimulation of telomerase mutant)/(WT PT RAP stimulation WT telomerase))*100).

  3. Not determined (N.D.) due to low telomerase activity.

An alignment of the human TEN-domain with the Tetrahymena thermophila TEN-domain in combination with a secondary structure prediction shows that K78 and R132 are in close proximity to each other, on a surface distinct from the proposed DNA-binding region of the TEN-domain (Figure 2, Figure 2—figure supplement 1).

Figure 2 with 1 supplement see all
Conserved basic residues K78 and R132 are found in close proximity on the surface of the TEN-domain.

Cartoon representation of the TEN-domain crystal structure from Tetrahymena thermophila (PDB: 2B2A). Amino acid numbers are from Tetrahymena (human counterparts in parentheses). Anchor site (DNA binding) residues are highlighted in blue. The positions of the residues corresponding to human K78 and R132 are highlighted in red, and F143 corresponding to human V144 is highlighted in green.

https://doi.org/10.7554/eLife.03563.007

These results demonstrate that K78 and R132 within the TEN-domain of hTERT make a much larger contribution to RAP stimulation by POT1-TPP1 than to enzymatic activity in vitro. RAP stimulation defects in K78E and R132E mutants are most likely due to the failure of telomerase to interact with TPP1. Thus, this surface of hTERT is a candidate TPP1-interacting element.

TEN-domain mutants fail to localize to telomeres in vivo

If mutations of K78 and R132 in hTERT affect the interaction of telomerase with TPP1, they should disrupt telomerase localization to telomeres in vivo. To test whether the mutant telomerases are recruited to telomeres, we transiently over-expressed mCherry-tagged hTERT variants and hTR in HeLa cells and determined the subcellular localization of telomerase by immuno-fluorescence (IF). Under all conditions hTERT and hTR were expressed at similar levels, indicating that mutations in the TEN-domain did not affect hTERT or hTR stability (Figure 3A–B). As previously described (Zhong et al., 2012), wild-type telomerase localized to telomeres and promoted the formation of neo-Cajal bodies at most telomeres, as shown by the co-localization of TRF2, hTERT, and coilin foci (Figure 3C). In contrast, all three hTERT mutants tested (K78E, R132E, K78E;R132E) localized to bona fide Cajal bodies, forming ∼1–3 large foci per cell that co-localized with coilin but did not localize to telomeres marked by TRF2 (Figure 3C). These results demonstrate that TEN-domain mutants that do not interact with POT1-TPP1 in vitro also fail to localize to telomeres in vivo. Furthermore, because Cajal body localization of hTERT requires its association with hTR, our observations indicate that telomerase assembly is unaffected by mutations in the TEN-domain.

TEN-domain mutations disrupt telomere localization of telomerase.

(A) Western blots of lysates of HeLa cells transfected with expression plasmids for various hTERT alleles and hTR, probed with an antibody against hTERT. Actin was used as a loading control. (B) Northern blots of RNA isolated from HeLa cells transfected with expression plasmids for various hTERT alleles and hTR, using probes for hTR. In vitro transcribed hTR (500 pg) was used as positive control. Blots were probed for RNase P RNA as loading control. (C) Immuno-fluorescence (IF) analysis of HeLa cells transiently transfected with mCherry-hTERT and hTR plasmids. Cells were fixed and probed with antibodies against mCherry, coilin, and TRF2 to visualize telomerase, Cajal bodies and telomeres, respectively. Images were deconvolved. Numbers indicate the fraction of cells analyzed showing the displayed phenotype (scale bar = 5 μm).

https://doi.org/10.7554/eLife.03563.009

TEN-domain mutants are defective in telomere maintenance in vivo

To determine the effects of the mutant telomerases on telomere maintenance, we generated cell lines stably expressing mCherry-tagged hTERT variants by retroviral transduction. Overexpression of hTERT alone leads to a moderate increase in telomerase activity per cell, more closely resembling endogenous telomerase levels than those obtained by overexpression of both hTERT and hTR (Cristofari et al., 2007; Xi and Cech, 2014). Thus, while the absolute expression level of mCherry-hTERT varied between the stable cell lines (Figure 4A), telomerase immuno-purified from the cell lines had comparable activities per cell, and these activity levels were threefold to fourfold higher than levels in untransfected control cells (Figure 4B). This result confirmed that mutations in the TEN-domain did not strongly reduce telomerase activity. The exogenous hTERT was overexpressed relative to endogenous hTERT, which was not detectable by western blot (Figure 4A). Therefore, the majority of endogenous hTR should be assembled into telomerase RNPs containing the mutant hTERT protein (Figure 4A).

TEN-domain mutants that do not localize to telomeres fail to elongate telomeres in vivo.

(A) Western blot probed for hTERT and for Actin as a loading control, showing hTERT expression in lysates of parental HeLa cells and cell lines stably expressing mCherry-hTERT variants. hTR was not ectopically expressed. (B) Direct enzyme assay of telomerase immuno-purified from lysates, generated using equal number of cells, from parental HeLa or cell lines stably expressing mCherry-hTERT variants. Activity per cell relative to WT hTERT overexpressing cells (n = 4, Mean ± SD, p < 0.05). (C) Telomeric restriction fragment Southern blot of cell lines stably expressing mCherry-hTERT variants over the time course of 8 weeks.

https://doi.org/10.7554/eLife.03563.010

To analyze the effect of mutant hTERT expression on telomere length, DNA isolated from stable cell lines over the course of 8 weeks after viral transduction was subjected to telomeric restriction fragment (TRF) analysis by Southern blot. The telomere length of the parental HeLa cell line was stable over the course of the experiment at ∼6.6 kb (Figure 4C). Expression of wild-type hTERT caused a progressive increase in telomere length to ∼14.8 kb (Figure 4C). In contrast, TEN domain mutants K78E and R132E failed to elongate telomeres, despite the fact that telomerase enzymatic activity per cell was higher than in the parental HeLa cell line (Figure 4B,C). In addition the TEN domain mutant K78E;R132E also failed to elongate telomeres, but due to its reduced activity (Figure 1F), we cannot conclude that this is exclusively due to a localization defect. These observations reinforce the conclusion that telomerase with K78E or R132E mutations in the TEN-domain fails to elongate telomeres due to an inability to localize to telomeres, not an activity defect.

A compensatory mutation in TPP1 rescues the mutant hTERT—TPP1 interaction in vitro

Taken together, our experiments demonstrate that mutations in the TEN-domain of hTERT recapitulate the effects of mutations in the TEL-patch of TPP1 (i.e., loss of telomerase recruitment to telomeres in vivo, and reduced POT1-TPP1 RAP stimulation in vitro). If the stimulation of telomerase RAP and telomerase recruitment to telomeres are the manifestation of a direct interaction between charged residues in the TEN-domain of hTERT and the OB-fold of TPP1, a charge-swap mutation in the OB-fold of TPP1 could rescue the TEN-domain mutation (Figure 5A). On the other hand, if a bridging factor were necessary for the interaction of TPP1 with hTERT, individually deleterious mutations in both TPP1 and the TEN-domain should have an additive negative impact on the TEN-TPP1 interaction when combined (Figure 5A).

Figure 5 with 3 supplements see all
A compensatory mutation in the TEL-patch of TPP1 rescues RAP stimulation of TEN-domain mutant telomerase in vitro.

(A) Schematic of the charge-swap experiment to test the interaction between specific amino acids on the TEL-patch on TPP1 and the N-DAT region of the TEN-domain of hTERT. Predicted experimental outcomes are illustrated for two competing models: a direct TPP1 telomerase interaction and an interaction bridged by a yet-unidentified factor (BF?). (B) Coomassie-stained gel of WT, E169K, and E215K TPP1 co-purified with wild-type POT1. (C) Overlays of the Superdex 200 16/60 sizing column chromatograms for wild-type and mutant TPP1-POT1 complexes. (D) Direct telomerase activity assay to test the rescue of various telomerase containing WT, K78E, R132E, and K78E;R132E mutant hTERTs in the absence of POT1-TPP1 (No PT) or WT, E169K, and E215K TPP1 in complex with wild-type POT1. LC and +4 marker as in Figure 1E. (E) Quantification (decay method) of the RAP stimulation by POT1-TPP1 (PT) relative to the stimulation of wild-type telomerase with wild-type POT1-TPP1 (n = 3, Mean ± SD, *p < 0.05, **p < 0.01, Student's t test).

https://doi.org/10.7554/eLife.03563.011

To test this hypothesis, we generated a number of mutant TPP1N proteins, changing individual conserved negatively charged residues in the TEL-patch to the basic amino acid lysine. Mutant TPP1 proteins were co-purified with POT1 to apparent homogeneity (Figure 5B). All mutant POT1-TPP1 complexes had identical elution profiles from the sizing column (Figure 5C). The presence of a single elution peak indicated that all TEL-patch mutant TPP1 proteins were not globally misfolded and formed heterodimers with POT1 (Figure 5C).

To test the ability of the mutant TPP1 proteins to bind TEN-domain mutants, we used direct telomerase enzyme assays to measure RAP stimulation by POT1-TPP1. Strikingly, TPP1 E215K fully rescued the processivity defect of hTERT K78E; processivity increased from 68% to 98% relative to wild-type telomerase and wild-type TPP1 (Figure 5D,E). Importantly, the non-cognate combinations of (i) wild-type telomerase and E215K TPP1 and (ii) K78E telomerase with wild-type TPP1 showed a similar reduction of RAP stimulation, to 78% and 68% of the full level, respectively (Figure 5D,E). TPP1 E169K gave dramatic reductions in RAP and did not stimulate any of the tested telomerases above 24% of wild-type (Figure 5D,E). hTERT mutant R132E and the double mutant K78E;R132E were stimulated at levels less than 23% of wild-type by all TPP1 proteins tested (Figure 5D,E). An alternative method for quantification of RAP stimulation by POT1-TPP1 gave similar results (Figure 5—figure supplement 1). Furthermore, to rule out the possibility of a bridging factor that co-purifies with telomerase from HEK293 T cells, we carried out the charge-swap experiment with telomerase purified from rabbit reticulocyte lysates (RRLs). The charge-swap rescue was statistically significant when using telomerase from RRLs, although the absolute PT stimulation of RRL telomerase was diminished (Figure 5—figure supplement 2).

To further illustrate the specificity of the charge-rescue, a disease allele V144M found in some cases of idiopathic pulmonary fibrosis (IPF) (Tsakiri et al., 2007) was tested for interaction with POT1-TPP1. V144M telomerase retained moderate telomerase activity and high processivity (48% and 95% of wild-type telomerase, respectively) but POT1-TPP1 RAP stimulation was significantly decreased to 61% when compared to wild-type telomerase (Table 1, Figure 5—figure supplement 3). Furthermore, TEL patch mutants E169K and E215K each resulted in an additive loss of RAP stimulation when tested with disease mutant V144M, to 28% and 37% respectively (Figure 5—figure supplement 3). Thus, the specific and compensatory nature of the hTERT K78E and TPP1 E215K mutant combination strongly suggests that K78 on the TEN-domain of hTERT directly interacts with E215 on the TEL-patch of TPP1 to stimulate RAP in vitro.

A compensatory mutation in the TEL-patch of TPP1 restores the telomerase TPP1-OB interaction at a non-telomeric locus in vivo

To analyze mutant hTERT-TPP1-OB-fold interactions in vivo, we utilized a Lac-repressor (LacI) assay previously described (Zhong et al., 2012). The TPP1-OB-fold domain was inserted between GFP and the LacI protein (Figure 6A). GFP-TPP1-OB-LacI was expressed alongside mCherry-tagged hTERT and hTR in the U2OS 2-6-3 cell line, which contains a single Lac-operator DNA array on chromosome 1 (Janicki et al., 2004), thereby tethering the OB-fold domain of TPP1 to a non-telomeric chromosomal locus (Figure 6A). The interaction between telomerase and the TPP1-OB-fold domain was assessed by co-localization of GFP- and mCherry foci in cell nuclei. Wild-type telomerase co-localized with the TPP1-OB-fold domain in ∼90% of nuclei. In contrast, in cells expressing K78E, R132E, or K78E;R132E telomerase, the fraction of TPP1-OB-fold domain foci showing telomerase signal was reduced to ∼55%, ∼30%, and 0%, respectively (Figure 6B,C, additional examples in Figure 6—figure supplement 1). The reduction in co-localization demonstrated that mutation of the hTERT TEN-domain interferes with the interaction between telomerase and the TPP1-OB-fold domain in vivo.

Figure 6 with 1 supplement see all
A compensatory mutation in the TEL-patch of TPP1 rescues TEN-domain mutant telomerase binding to the TPP1 OB-domain in vivo.

(A) Model showing the experimental design. Fusion of the OB-domain of TPP1 to the lac repressor (LacI) recruits the OB-domain to a single non-telomeric chromosomal locus (LacO array), allowing the interaction between telomerase and the TPP1 OB-domain to be assessed by co-localization of GFP (TPP1 OB) and mCherry (hTERT) in cell nuclei. (B) Fluorescence images showing the localization of GFP-TPP1-OB-LacI and mCherry-hTERT fusion proteins in cell nuclei stained by DAPI. Cells were fixed, permeabilized, and stained with DAPI. The intrinsic GFP- and mCherry-fluorescence was used to detect GFP-TPP1-OB-LacI and mCherry-hTERT (scale bar = 5 μm). (C) Quantification of the experiments shown in (B), showing the fraction of nuclei with co-localization of GFP- and mCherry-foci (n = 3, 117–176 nuclei total, Mean ± SD, *p < 0.05, **p < 0.01, Student's t test).

https://doi.org/10.7554/eLife.03563.015

To determine whether mutations introduced in the OB-fold domain of TPP1 can compensate for the mutations introduced in hTERT in vivo, we carried out the LacI-assay with E215K TPP1-OB. As shown in Figure 6B,C, the co-localization of wild-type telomerase with the E215K TPP1-OB-fold domain was greatly reduced; instead, wild-type telomerase localized to many foci distinct from the LacI site, presumably telomeres. In contrast, introduction of the E215K mutation significantly increased the co-localization of K78E telomerase with the TPP1-OB-fold domain (p < 0.05). Thus the E215K mutation in the TPP1-OB-fold compensates for the presence of the K78E mutation in hTERT in vivo. Importantly, E215K TPP1 had no effect on the localization of R132E and K78E;R132E telomerases, demonstrating that E215K specifically compensates for the presence of the K78E mutation in hTERT (Figure 6B,C). If another protein were bridging telomerase and TPP1, any combination of two defective mutations should be even more defective. Thus, finding a compensatory double mutant demonstrates that telomerase is recruited to telomeres by a direct interaction between the TEN-domain of telomerase and the TPP1-OB-fold domain in part by a contact formed between K78 in the TEN-domain and E215 in the TPP1-OB-fold domain.

TPP1 E215K rescues telomere maintenance in cells expressing K78E hTERT

To address whether TPP1 E215K could rescue recruitment of hTERT K78E to telomeres, we transduced cell lines stably expressing WT and K78E mCherry-hTERT with retrovirus expressing wild-type (WT), E169K, and E215K TPP1-FLAG alleles. The resulting cell lines expressed both TPP1-FLAG and mCherry-hTERT proteins (Figure 7A–B). Importantly, retroviral transduction was carried out after substantial telomere erosion had occurred due to the expression of K78E hTERT (Figure 4C, transduction carried out after 8 weeks).

TPP1 E215K rescues telomere maintenance in cells expressing hTERT K78E.

(A) Western blot probed for TPP1-FLAG and for Actin as a loading control, showing TPP1 expression in lysates of parental HeLa cells and cell lines stably expressing TPP1-FLAG and mCherry-hTERT variants. (B) Western blot probed for hTERT and for Actin as a loading control, showing hTERT expression in lysates of cell lines stably expressing TPP1-FLAG and mCherry-hTERT variants. (C) Telomeric restriction fragment Southern blot of cell lines stably expressing K78E mCherry-hTERT and TPP1-FLAG variants over the time course of 6 weeks.

https://doi.org/10.7554/eLife.03563.017

To test whether the expression of E215K TPP1 rescued telomere maintenance in cells expressing K78E hTERT, we carried out TRF analysis by Southern blot (Figure 7C). Telomere length in the cell line expressing only K78E hTERT was constant at ∼3.4 kb during the 6-week experiment. Expression of WT TPP1 led to an increase of telomere length to ∼6.2 kb. Importantly, expression of E215K TPP1 triggered a more pronounced increase in telomere length to ∼7.7 kb over the same time period, while E169K gave lead to a subtle increase to ∼4.9 kb. These observations demonstrate that expression of E215K TPP1 in cells expressing K78E hTERT rescues telomere length maintenance, most likely by driving telomerase recruitment to telomeres.

Discussion

Telomerase recruitment to telomeres is a crucial step in telomere maintenance. Here, we demonstrate that K78 and R132 within the TEN-domain of hTERT are critical residues that mediate the direct interaction of telomerase with the TEL-patch of TPP1, recruiting telomerase to telomeres (Figure 8). Mutation of either interaction partner results in sequestration of telomerase in Cajal bodies and failure to elongate telomeres in vivo. Our observations provide a molecular mechanism explaining the failure of previously described N-DAT mutants to function in vivo. Additionally, the identification of the interface between telomerase and the telomere will allow more targeted approaches to alter telomerase recruitment as a potential therapeutic approach for human diseases.

A model for the recruitment of telomerase to telomeres through the direct interaction between the TEL-patch of TPP1 and the hTERT TEN-domain.

Throughout the cell cycle, telomerase associates with TCAB1 and localizes to Cajal bodies. During late S/G2, telomerase is recruited to telomeres through a direct interaction between K78 on the TEN-domain of hTERT and E215 within the TEL-patch of TPP1. R132 in the TEN-domain also participates in the interaction, but the corresponding residue on the TEL-patch and the nature of the interaction remains unknown (?). TPP1 is recruited to telomeres through interaction with TIN2 and further stabilized by the single-stranded DNA-binding protein POT1. TRF1 and TRF2 comprise the double-stranded DNA binding proteins of shelterin. Mutations in the OB-domain of TPP1 or TEN-domain of hTERT that disrupt the interaction are sufficient to prevent recruitment to telomeres, instead resulting in the sequestration of telomerase in Cajal bodies.

https://doi.org/10.7554/eLife.03563.018

Specific hTERT TEN-domain amino acids are necessary for telomerase recruitment to telomeres

During the S-phase of the cell cycle, telomerase moves from Cajal bodies to telomeres to counteract the progressive shortening of the telomere that occurs during DNA replication (Jady et al., 2006; Tomlinson et al., 2006; Cristofari et al., 2007; Stern et al., 2012). The regulatory mechanisms underlying this process are not well understood, in part because the necessary molecular interfaces are not fully defined. On the telomere side, TPP1 is well established as being necessary for the recruitment of telomerase (Abreu et al., 2010; Nandakumar et al., 2012; Sexton et al., 2012; Zhong et al., 2012). Localization of telomerase to an artificial non-telomeric focus by the TPP1 OB-domain (Zhong et al., 2012) suggests that the TPP1 OB-domain is the minimal sufficient telomerase recruitment module of the shelterin complex. The TEL-patch provides a molecular surface that mediates telomerase binding, as single mutations in the TEL-patch result in loss of telomere recruitment and of RAP stimulation of telomerase by TPP1 (Nandakumar et al., 2012; Sexton et al., 2012; Zhong et al., 2012).

While the telomeric half of the interface required for telomerase recruitment to telomeres is well understood, the molecular determinants on telomerase required for telomere localization were thus far ill defined. Several lines of evidence, including naturally occurring disease-associated alleles of hTERT (V144M) (Tsakiri et al., 2007) and synthetic mutations (G100V, R132D, N-DAT), have implicated the TEN-domain of hTERT in telomerase recruitment to telomeres (Armbruster et al., 2001; Zaug et al., 2010; Stern et al., 2012). Unfortunately, due to the involvement of the TEN-domain in catalysis and RAP, many of these mutations also alter the enzymatic activity of telomerase. Thus, TEN-domain mutants can have pleotropic effects and may not specifically disrupt the telomere–telomerase interface (Zaug et al., 2010, 2013). Our results demonstrate that a region in the TEN-domain including K78, R132, and V144 is required for telomerase recruitment in vivo and POT1-TPP1-mediated RAP stimulation in vitro, albeit to varying degrees. The differences in RAP stimulation may reflect the importance of the respective residue in an interaction with TPP1. The recruitment defects and failure to maintain telomeres in vivo occur in the context of wild-type levels of activity, suggesting that the mutants described in this study are separation-of-function alleles that dissociate enzymatic activity from recruitment. Importantly, the recruitment defects of K78E and R132E telomerases recapitulate those of TEL-patch mutant TPP1 proteins (Zhong et al., 2012; Nandakumar et al., 2012). The interaction between the TEL-patch of TPP1 and the TEN-domain of hTERT provides a molecular mechanism to explain the failure of N-DAT mutations to immortalize cells.

Telomerase recruitment to telomeres in human and other organisms

Our observation that charge-swap mutations on the TEN-domain of hTERT and the TEL-patch of TPP1 restore telomerase recruitment indicates that these elements interact directly; for other examples see Jucovic and Hartley (1996); Tansey and Herr (1997); Pennock et al. (2001). If instead there were an ‘adapter’ protein bridging the two elements, the combination of two individually deleterious mutations would lead to an additive loss of function (Figure 5A). We note that hTERT K78E and TPP1 E215K lead to similar reductions in RAP stimulation when combined with wild-type TPP1 and wild-type hTERT, respectively. This observation adds additional support to the model that hTERT K78E and TPP1 E215K eliminate reciprocal residues of the same molecular interaction. In contrast, the V144M mutation in hTERT and the E215K mutation in the TEL-patch of TPP1 display an additive reduction in RAP stimulation, indicating that they affect separate components of the telomerase–TPP1 interface.

In other organisms, however, the association of telomerase with the telomere occurs through different mechanisms. For example, a bridging factor Ccq1 appears to be necessary for recruitment of telomerase to the telomeric protein Tpz1 (TPP1 homolog) in S. pombe (Miyoshi et al., 2008; Moser et al., 2011). A homologue of Ccq1 has not been identified in humans (Nandakumar and Cech, 2013). Another variation occurs in Saccharomyces cerevisiae, where Est3 shares structural conservation with TPP1 (Yu et al., 2008; Rao et al., 2014). Est3 is part of the telomerase holoenzyme instead of the telomeric cap (Hughes et al., 2000). Est3 is thought to make direct contacts with the TEN-domain of Est2, the yeast TERT (Friedman et al., 2003; Talley et al., 2011; Yen et al., 2011). However, Est3 appears to stimulate telomerase activity rather than act as a recruitment factor (Talley et al., 2011). Recruitment is instead mediated by the interaction of the Est1 subunit of the telomerase holoenzyme and Cdc13 at the telomere (Evans and Lundblad, 1999), and charge-swap mutations indicate that this interaction is direct (Pennock et al., 2001). Thus, in budding yeast the interaction between a TPP1-like OB-domain protein (Est3) and the TEN-domain of TERT (Est2) is conserved, but the recruitment mechanism differs significantly from that in humans.

In addition to the interaction between TPP1 and hTERT in humans, other factors contribute to the productive recruitment of telomerase to the telomere. Depletion of TIN2 from shelterin (Abreu et al., 2010) and perturbation of the TCAB1-hTR interaction (Stern et al., 2012) both result in a reduction of telomerase association with the telomere. In spite of the aforementioned defects, these proteins likely play important but indirect roles in telomeric recruitment. TCAB1 is necessary for telomerase maturation and trafficking to Cajal bodies (Venteicher et al., 2009), while TIN2 is required for TPP1 localization to telomeres (Abreu et al., 2010).

Previous work also suggested that the C-DAT region of the CTE of hTERT contributed to telomerase recruitment (Banik et al., 2002; Zhong et al., 2012). However, synthetic C-DAT mutants have severe activity defects (Huard et al., 2003; Jurczyluk et al., 2011). In addition, disease-associated mutations in and near C-DAT have varying effects on activity and cellular localization of telomerase. E1117X has ∼10% of wild-type activity and appears to be sequestered to Cajal bodies in vivo (Tsakiri et al., 2007; Zhong et al., 2012); in contrast, F1127L has reasonable activity (∼70%) and localizes to telomeres, and it is stimulated by POT1-TPP1 (Zhong et al., 2012; Zaug et al., 2013). Thus, CTE mutants do not have clear recruitment defects dissociated from activity loss, despite their importance based on disease association. In contrast, the observation that a mutant TPP1-OB-domain increases the localization of a mutant hTERT to a non-telomeric locus in vivo, while it strongly decreases the association with wild-type hTERT, demonstrates that a direct TEN-domain-TEL-patch interaction is necessary and may be sufficient for telomerase recruitment to telomeres in humans.

The interface of telomerase recruitment as a therapeutic target

The identification of a direct protein–protein interaction surface between telomerase and the telomere is key to identifying modulators of telomerase recruitment as potential therapeutic agents. Multiple disease mutations associated with IPF—P33S, L55Q, Pro112ProfsX16, and V144M—cluster in TEN-domain of hTERT. Mutation carriers have significantly shorter telomeres than non-carrier relatives (Armanios et al., 2007; Tsakiri et al., 2007). These mutations have variable impacts on telomerase enzymatic activity (Armanios et al., 2007; Tsakiri et al., 2007; Zaug et al., 2013). Importantly, V144M is defective in TPP1-mediated RAP stimulation in vitro (this study) and fails to localize to telomeres in vivo (Zhong et al., 2012). Given the propensity of both synthetic and IPF-associated mutations in the TEN-domain of hTERT to disrupt telomere localization, it is likely that some cases of IPF are directly caused by a decrease in telomerase recruitment to telomeres. Agonists of the TEL-patch-TEN-domain interaction that compensated for these mutations and attenuated telomere shortening associated with the disease might be pharmaceutically important.

In contrast to the loss of telomerase function associated with IPF, reactivation of telomerase is a hallmark of most human cancers. Inhibitors that disrupt telomerase recruitment to telomeres would provide an additional approach to target telomerase activity in cancer cells. In support of this idea, disruption of the TEL-patch diminishes cell growth and triggers apoptosis in HeLa cells, and this effect is exacerbated in combination with a small molecule inhibitor of telomerase activity (Nakashima et al., 2013). Our finding that the TEN-domain of hTERT interacts directly with the TEL-patch of TPP1 to bring telomerase to the telomere defines a key interface and provides a direct target for the design of novel therapeutic inhibitors of telomerase action.

Materials and methods

Telomerase purification

Request a detailed protocol

Telomerase was overexpressed in HEK293T cells and purified as previously described (Sauerwald et al., 2013). Briefly, 1 ml whole cell lysates of 50–60 × 106 HEK293T cells overexpressing ZZ-TEV-3xFlag-hTERT variants and hTR were incubated with ∼350 µl IgG-Sepharose at 4°C for 3–4 hr. Following a wash with 50 ml wash buffer (20 mM HEPES-KOH pH 7.9, 300 mM KCl, 2 mM MgCl2, 1 mM EDTA, 1 mM DTT, 1 mM PMSF, 0.1% Triton X-100, 10% glycerol), telomerase was eluted from the beads by cleavage with TEV-protease. The telomerase bound to IgG-Sepharose was incubated with ∼500 µl of elution buffer (20 mM HEPES-KOH pH 7.9, 150 mM KCl, 2 mM MgCl2, 1 mM EDTA, 1 mM DTT, 1 mM PMSF, 0.1% Triton X-100, 10% glycerol) supplemented with 5 µl of Act-TEV (Life technologies, Carlsbad, CA) and 5 µl RNAsin+ (Promega, Madison, WI) overnight at 4°C. Telomerase-containing eluate was aliquoted, snap frozen in liquid nitrogen, and stored at −80°C until use.

To purify endogenous telomerase from HeLa cells and HeLa cell lines overexpressing mCherry-hTERT variants, telomerase was immuno-purified using a sheep polyclonal antibody (a kind gift from S Cohen) as previously described (Cohen and Reddel, 2008). Briefly, cell lysates of 1 × 106 cells were incubated with 40 µg of anti-hTERT antibody, which was captured using protein-G agarose. Telomerase was eluted using the peptide antigen used to raise the antibody. Eluates were used for direct telomerase extension assays.

Telomerase was produced in RRLs as previously described (Zaug et al., 2013) using ProA-tagged hTERT and purified using the same protocol as telomerase from HEK293T cells described above.

Western blot

Request a detailed protocol

Western blots were carried out using 4–12% polyacrylamide Bis Tris Glycine gels (Life Technolgies) and antibodies against hTERT (Ab32020; 1:1000; Abcam, UK), beta-Actin (A5441; 1:5000; Sigma), and an HRP-conjugated FLAG-antibody (A8592; 1:1000; Sigma). Secondary antibodies (Jackson ImmunoResearch, West Grove, PA) were used at 1:5000. Detection was carried out using SuperSignal Western Pico Chemiluminescence substrate (ThermoFisher Scientific, Waltham, MA).

POT1-TPP1N purification

Request a detailed protocol

Cell pellets from insect cells overexpressing GST-POT1 and Escherichia coli cells overexpressing 6xHis-SUMO-TPP1N were lysed by sonication in lysis buffer (PBS supplemented with 250 mM KCl, 1 protease inhibitor tablet [ThermoFisher Scientific], 1 mM PMSF). Lysates were cleared by centrifugation at 40,000×g for 35 min at 4°C. Equal volumes of the insect-cell lysates expressing GST-POT1 and bacterial lysates expressing 6xHis-SUMO-TPP1N were combined and incubated with 1 ml of glutathione–sepharose resin for 1 hr at 4°C. After washing three times with 50 ml of GST-wash buffer (PBS supplemented with 250 mM KCl, 1 mM DTT, and 1 mM PMSF), the POT1-TPP1N complex was eluted with 3 ml of GST-elution buffer (50 mM Tris pH 8.1, 75 mM KCl, 10 mM glutathione). The elutions were supplemented with 0.5% w/w PreScission protease and SUMO-protease and incubated on ice for 0.5–1 hr, followed by size-exclusion chromatography on a Superdex 200 16/60 column (GE Healthcare, UK) in gel-filtration buffer (50 mM Tris pH 7.0, 150 mM NaCl, 1 mM DTT). Size-exclusion fractions were pooled, concentrated, supplemented with 10% glycerol vol/vol, snap-frozen in liquid nitrogen and stored at −80°C until use.

Direct telomerase activity assay

Request a detailed protocol

Direct telomerase activity assays were carried out as previously described (Zaug et al., 2013), with slight modifications. The final 20 µl reactions contained 2 µl eluted telomerase extract, 500 nM primer, 500 nM POT1-TPP1 complex, and 150 mM KCl. Direct telomerase assays were quantified as previously described (Zaug et al., 2013), except the summed counts incorporated into extension products were normalized to both the hTERT levels (determined by western blot) and the loading control and then normalized to WT telomerase. Processivity was measured using the decay method previously described (Latrick and Cech, 2010). Briefly, counts for repeats 1–20 were corrected for the number of dGTP nucleotides incorporated. The natural log of the counts left behind in each repeat was graphed vs repeat number and fit by linear regression. The slopes of the linear regressions were compared and normalized to WT telomerase. RAP stimulation by PT, for direct extension assays with POT-TPP1, was measured in the same manner as processivity except values were normalized to WT telomerase with WT POT1-TPP1 (i.e., RAP stimulation by PT = processivity of mutant telomerase with TPP1/processivity of WT telomerase with WT POT1-TPP1). The processivity in the absence of POT1-TPP1 was not included the calculation, because (1) it did not vary greatly among the mutants (Table 1) and (2) it was so small relative to the processivity in the presence of POT1-TTP1 that dividing by this distorted the calculations and led to irreproducibility. An alternative fraction method of quantitation was also used to determine RAP stimulation by PT, calculated as the fraction of counts in bands 9 and above divided by the total sum of counts incorporated, normalized as described for the decay method. Telomerase from stable cell lines was immuno-purified from identical number of cells and was subjected to direct telomerase assays as previously described (Cohen and Reddel, 2008). Activity was determined by dividing the total counts by the loading control and normalized to the activity level of cells overexpressing WT hTERT. Telomerase purified from RRLs was assayed using 10 µl of eluate under identical conditions as telomerase from HEK293T cells.

Northern blot

Request a detailed protocol

Northern blots were carried out as previously described, using 5 separate probes for hTR and 2 probes for RNase P (Xi and Cech, 2014).

Molecular cloning

Request a detailed protocol

The mCherry-hTERT vector was generated by restriction cloning of the hTERT gene into a modified pBABE-puro vector (a kind gift from Iain Cheeseman) containing the N-terminal mCherry-tag using SalI/EcoRI. GFP-TPP1-OB-LacI was cloned by restriction cloning of the TPP1-OB-fold domain into a modified pBABE-blast vector (a kind gift from Iain Cheeseman) containing the GFP and LacI sequences using XhoI/EcoRI. All point mutations in hTERT and TPP1 were introduced using the Quickchange II mutagenesis kit (Agilent, Santa Clara, CA). The presence of the mutations was verified by Sanger sequencing.

Cell culture

Request a detailed protocol

All human cell lines were cultured at 37°C, 5% CO2 in growth medium (Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum, 2 mM glutamax [Life Technologies]), 100 units/ml penicillin, and 100 µg/ml streptomycin. The growth medium for the U2OS 2-6-3 cell line (a kind gift from David Spector) was additionally supplemented with 100 µg/ml hygromycin B to maintain the LacO array.

Transient transfection of human cells

Request a detailed protocol

All transient transfections of human cells were carried out using Lipofectamine 2000 (Life Technologies) according to the instructions of the manufacturer.

Generation of stable cell lines

Request a detailed protocol

Retroviruses were generated by transfection of pBABE vectors encoding mCherry-hTERT variants and a puromycin resistance gene along with a vector encoding the coat protein VSV-G into the packaging cell line 293-GP. Virus-containing supernatants were harvested 72 hr post-transfection, filtered through a 0.22-µm filter, and supplemented with 8 µg/ml polybrene. HeLa cells were incubated with virus-containing supernatant for 16 hr. After infection, the growth medium was replaced with fresh medium containing 1 µg/ml puromycin for mCherry-hTERT constructs and 1 µg/ml blasticidin for TPP-FLAG constructs to select for transduced cells. To ensure maintenance of the transgenes, cell lines were kept under selection with 1 µg/ml puromycin and 1 µg/ml blasticidin (TPP1 rescue only) for the duration of the experiments.

Immunofluorescence and fluorescence

Request a detailed protocol

IF experiments were carried out as previously described (Nandakumar et al., 2012), using the following antibodies: mouse monoclonal anti-TRF2 (IMG-124A; 1:500; Imgenex), rabbit polyclonal anti-coilin (sc-32860; 1:100; Santa Cruz, Dallas, TX), rat monoclonal anti-mCherry (M11217; 1:1000; Life Technologies), and rabbit polyclonal anti-RAP1 (NB-100-292; 1:500; Novus Biologicals, Littleton, CO). Secondary antibodies (Life Technologies, Abcam) were pre-absorbed to prevent cross-reactivity between rat and mouse antibodies.

Telomere length analysis

Request a detailed protocol

Telomeric restriction fragment length analysis by Southern blot was carried out as previously described (Nandakumar et al., 2012). Briefly, 1.5 µg of genomic DNA extracted from human cell lines was digested with RsaI/HinfI and separated by gel electrophoresis using a 0.8% 1xTBE agarose gel for a total of 1100 V-hours. DNA was transferred onto a Hybond N+ membrane (GE Healthcare) and telomeric restriction fragments were detected using a radiolabeled (TTAGGG)4 probe. To determine mean telomere length, lane intensity profiles were extracted using ImageQuant TL and fit to a Gaussian using Kaleidagraph. The mean of the Gaussian distribution was used as telomere length of the respective sample.

LacI recruitment assay

Request a detailed protocol

Plasmids encoding mCherry-hTERT, hTR, and GFP-TPP1-OB-LacI were co-transfected into U2OS 2-6-3 cells grown on coverslips as described above. Cells were fixed for 5 min in PBS supplemented with 0.4% formaldehyde followed by permeabilization with PBS supplemented with 0.2% (vol/vol) Triton X-100 for 5 min. Finally, coverslips were embedded in DAPI containing mounting media (Vector Laboratories, Burlingame, CA). Imaging was carried out visualizing the intrinsic fluorescence of the mCherry- and GFP-fusion proteins without additional staining.

Microscopy

Request a detailed protocol

All images were acquired on a Deltavision Core deconvolution microscope (Applied Precision, GE Healthcare) using a 60× 1.42NA PlanApo N (Olympus, Japan) or 100× UPLanSApo 1.4NA (Olympus) objective and a sCMOS camera. 20 Z-sections with 0.2 µm spacing were acquired for each image with identical exposure conditions within each experiment. For analysis, maximum intensity projections were generated and scaled identically. For presentation in figures, representative images were deconvolved (where indicated), followed by generation of maximum intensity projections of 10 Z-sections, which were scaled identically for all experimental conditions.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54

Decision letter

  1. Carol Greider
    Reviewing Editor; Johns Hopkins University, United States

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

Thank you for sending your work entitled “Identification of human TERT elements necessary for telomerase recruitment to telomeres” for consideration at eLife. Your article has been evaluated by James Manley (Senior editor) and 3 reviewers, one of whom is a member of our Board of Reviewing Editors.

The Reviewing editor and the other reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.

In this manuscript the authors carry out a number of experiments to examine the interaction of the TEN domain of TERT with the OB fold domain of TPP1. They present in vivo experiments to document the interaction and specificity. They identify two residues in the TEN domain, K78 and R132 that affect telomerase RAP processivity as well as some effects on catalytic activity and telomere length. They then go on to identify one mutation in TPP1, E215 that compensates for the K78 mutant in the TERT TEN domain. The authors conclude that unlike in S. pombe, where TERT and the TPP1 homologue are bridged by a protein, CCQ, in human cells, these two proteins interact directly. Unfortunately all of the evidence presented here is in vivo and thus no clear conclusion about any possible bridging protein can be made. Rather than directly addressing the potential interaction of the TEN domain and TPP1 by using biochemical assays, the authors used a series of in vivo surrogates that suffer from a number of caveats. For example, high-level protein overexpression is used to examine cellular localization, when over expression is well known to distort protein localization. Without any direct demonstration of an interaction of TPP1 and the TEN domain that is mediated by the charge swap mutants that are described, the conclusions here are not supported by the data.

Major concerns:

If the authors would like to conclude that there is a direct interaction of hTERT TEN domain with TPP1 OB fold evidence for direct protein interacts needs to be presented. The hypothesis they are trying to test is whether these two proteins interact and if that interaction is mediated by the charged residues on the TEL patch of TPP1 and the opposite charges on the TERT- TEN domain. There are a number of biochemical assays that they could use. They could purify the specific proteins or the domains and test interaction and affinities on Biacore chips. They could use GST pull down experiments, affinity columns or similar techniques. This assays presented here including the Cajal body localization and LacO array recruitment are very indirect and are in vivo. Thus they do not exclude a bridging protein, as the authors try to conclude. For the authors to conclude that the interaction of TERT and TPP1 is a direct interaction, and mediated by the charged residues they identify, evidence of direct protein interaction in vitro needs to be presented.

Specific comments:

1) Figure 1b: the diagram has blue highlighted text indicating conserved residues; however additional mutants that are not in blue are shown in the Figure 1 supplement. There needs to be an explanation for how specific residues were chosen for mutation and need to state in text, all of the mutants that were made and why. How the mutant combinations in in Table 1 were chosen should be described.

2) The results shown in Figure 1 and Table 1 are essential for understanding the rest of the paper. There are a number of issues in these two figures that need to be clarified. In Figure 1 F, G and H the authors set up the activity assay and the mutants that are the basis for all of their experiments going forward. It is important to explain in the text what these assays are, and what is concluded from them. I am assuming that the intrinsic changes in processivity are being ignored (Figure 1G) and that the PT stimulation is being used as a measure of the interaction of TEN 1 and TPP1. However this is never stated directly. The authors need to explain how they define processivity and PT stimulation differently and how it is quantitated.

This data is quantitated differently from a previous report by the Cech lab in Latrick et. al 2010. The authors should provide an explanation for changing the processivity analysis from previous work, since choosing to sum all counts above band #9 seems a bit arbitrary and less rigorous than previous studies. In Figure 1G, there is a clear increased processivity of R132E and K78E+ R132E. This 2-fold increase is not evident from the gel in Figure 1E. And also it is not commented on in the paper. If this is true and not due to a normalization artifact, how does this change in processivity affect telomere length? This increased processivity is also reported in table 1 as 186%, a large effect.

Currently there is very little description on how the authors determine the values they report on Table 1 for telomerase activity, RAP, and RAP stimulation by POT1-TPP1. Since IP-efficiencies and/or expression of hTERT vary greatly among different hTERT mutants, it was often hard to see the correlation between what is shown on gels vs. values in Table 1. Thus, it would be very important for the authors to better describe their quantification methods, and provide raw data before normalization. Were these measurements done on serial dilutions of the extract to increase the accuracy? How reproducible are these assays? How many experiments were done to generate are in the error bars shown? Figure1 and Table 1 describe some of the same mutants. Figure 1H axis label says “normalized PT stimulation” while Table 1 column 4 says “% RAP stimulation”. I assume these are the same thing? Need to use consistent terminology if the reader is to follow.

The authors say that K78E and R132E are required for RAP stimulation but not essential for enzyme activity, however the double mutant has only 63 % activity. This is overstating the conclusion that these residues are not needed for activity and they need to make conclusions that represent the data presented.

3) The overexpression studies need to be cautiously interpreted. To determine if the TERT TEN domain mutants localize to telomeres, the authors overexpress the mutants and examined localization to telomeres or Cajal bodies. This experiment is not sufficient to conclude the proteins fail to localize to telomeres. Overexpression of proteins is known to cause spurious cellular localization. The fact that overexpression of the wt gives different pattern than overexpression of the mutant only indicates that the mutants may aggregate differently in the cell. If they want do make conclusion about protein localization, a second method without overexpression is needed. The authors state they have an antibody to TERT, so ChIP and Co-IP might be an alternative. Finally the authors do not show the level of over expression or look at the stoichiometry of the various holoenzyme components. The authors conclusion from this experiment that TEN domain and and TPP interact is not supported by these experiments. It is not clear this experiment is helpful to the conclusions given the caveats of over expression.

4) It is not clear how the authors are measuring telomerase activity, as it does not seem to be consistent. In Figure 1 the K78E+R132E mutant is reported to have 60 % telomerase activity but here Figure 4b the activity is more than wildtype. I wonder how these numbers are normalized to give these values. It would be much better to show a titration activity with amounts normalized by hTERT levels. The conclusion in the text is that there is no effect of the mutants on activity, but an effect on telomere length. This conclusion is not supported by the data given the results of reduced activity in Figure 1.

5) The method used to show localization of telomerase does not have any controls. In Figure 4C the authors show a robust signal for endogenous hTR localization in cells. Given that there is a copy number of about 100 molecules of hTR per cell, it is not clear what could be generating such a robust signal. To know that this technique can indeed localize endogenous hTR, the authors need a control experiment. They should use a cell line that does not express hTR such as VA13 and show there is no signal with this technique, or use mouse cell lines that are deleted for the endogenous mTR and show there is no signal. Unless a control for specificity is shown this experiment should be omitted. The authors conclude from this experiment that mutant hTERT exerts a dominant effect on endogenous hTR localization. This conclusion is not justified by the data.

6) Figure 4D. In all of the previous experiments the authors overexpressed both hTERT and hTR, in this experiment they only overexpressed hTERT; it is not clear why this was done. Given that the mutants K78E and R132E showed decreased telomerase enzyme activity in the experiments in Figure 1, it is not surprising that there is less telomere elongation seen with these overexpressed mutants. Over expression of these mutant enzymes that have decreased activity will sequester all of the hTR and result in overall reduced telomerase activity and thus explain reduced telomere elongation. The authors' conclusion that this data shows that the TEN domain mutants fail to localize but are active is not warranted by the data presented.

7) The authors describe a charge swap experiment in which they mutate residues in TPP1 that they think might interact with the TEN domain acidic residues in Figure 5. The authors state that they mutated “a number of residues in TPP1”, however they then just show E169K and E215K. It is important for the authors to list all the mutants tested and show us what the phenotypes are for all of them, otherwise the specificity of these chosen mutants that will be followed going forward is not clear.

8) The authors use an assay developed by Zhong et al to examine ectopic localization of Tert to a chromosome locus where the OB fold domain of TPP1 has been tethered in Figure 6. They use this assay to demonstrate that mutants in the TEN domains disrupt localization but the double mutant of TEN K78E with TPP1 E215K has an increased number of foci. The figure shown in Figure 6 B is not helpful. This figure only shows single cell nuclei and the point of the assay is to examine the number of nuclei with foci. So it would only make sense to show a field of cells, not single cells. So either do not show Figure 6 B or show a field that would represent the data they are trying to relate. More importantly since this is an in vivo assay, if there were a bridging protein it would still be present, so the main conclusion about a charged interaction is being the main interaction cannot be concluded from this experiment.

9) The authors add analysis of V144 at the end of the paper in Figure 7. This figure does not contribute in any way to the point of this paper. The V-144 mutation may be of interest in telomerase function in IPF, but the addition of this mutant onto the end of this paper is distracting and does not belong here. This data is best published in another study and should be removed from this manuscript.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled “Identification of human TERT elements necessary for telomerase recruitment to telomeres” for further consideration at eLife. Your revised article has been favorably evaluated by James Manley (Senior editor) and a member of the Board of Reviewing Editors. The manuscript has been improved but there are some issues that need to be addressed with further revisions, as outlined in the following review.

The rebuttal letter by the authors is persuasive in many areas, and a number of helpful changes have been added to the manuscript. However some of the changes and additions are confusing and perhaps some figures are mislabeled. If the authors can clarify and revise the writing of the manuscript, it is in principal acceptable for publication.

Here are some of the concerns, listed sequentially as they appear in the manuscript, not in order of importance (some are minor suggestions).

1) In the rebuttal letter the authors argue that it is not possible to look at direct protein interaction with hTERT and TPP1 because of solubility issues that many in the field acknowledge. I would argue that interactions of soluble proteins made in RRL using specific antibodies or GST fusions could be used to look at pull downs and protein or domain interactions. However, I will not push this point too far. The authors are persuasive in the rebuttal that the PT stimulation is an excellent proxy for direct protein-protein interaction' They state in the letter “we consider the charge-swap mutants to provide such a strong argument for a direct protein-protein interaction”

However, in the text of the manuscript the authors did not make a strong case for how the PT stimulation assay is being used as a proxy for protein-protein interactions. Instead the text was altered in the opposite direction. They have removed the wording that stated they were assaying RAP stimulation and instead now say they are measuring protein interactions. I think it is important for the reader to understand the actual experiments that are done measure RAP stimulation, and then at the end of the paragraph the authors can say that they interpret this as a direct interaction and they will go on to test this.

Further on in the text, the authors are again mixing the descriptions of the results with their interpretation. The previous text that is marked as deleted gives a more straightforward description of that was measured and is interpretation from that result. A reversion to this kind of language would help the manuscript significantly.

2) The yellow color in Figure 2 is very difficult to see, I suggest using a different color.

3) The authors use the jargon “Super telomerase” in some cases in the text but not constantly. They are referring to a term used in a paper by Lingner's group when hTERT and hTR are co-overexpressed. The authors use “super telomerase” without explaining what it means. In the methods section the authors state they use the direct telomerase assay described by Zaug 2013. In that paper both hTERT and hTR are co-overexpressed but the term “Super telomerase” is not used. Since the Zaug 2013 assay is used in Figure 1, technically this could be called super telomerase. I suggest that the term be eliminated and the experiment just described as they are. (Note Figure 5–figure supplement 3 also uses the term super telomerase.)

4) In Figure 4, it will be important to show that it is the lack of interaction of TERT with TPP1 that is responsible for the short telomeres and not the decreased enzyme activity intrinsic in the hTERT mutants. I made this point before; however, the authors disagreed in their rebuttal. Figure 4B shows activity assays from cell lines with different levels of telomerase expression. The authors argue that the activity per cell for each of the mutants is the same. And the quantitation below the lanes suggests that the over expressed hTERT mutants have slightly more activity than overexpressed while type. This is discordant with the careful quantitation shown in Figure 1 E and F that shows a decreased catalytic activity of the 78 and 123 mutants.

In their rebuttal the authors state:

“The results in Figure 4B demonstrate that all cell lines over-expressing the hTERT alleles have a 3-4-fold higher telomerase activity than the parental HeLa cell line. This observation is crucial since it rules out lower telomerase activity as a possible explanation for the reduced telomere length in the cell lines expressing hTERT with mutations in the TEN-domain. To the contrary, telomeres shrink in cell-lines expressing TEN-domain mutant alleles despite 3-4-fold higher telomerase activity per cell.”

I am not sure I follow the logic of comparison of the level of activity in untransfected HeLa cells. As I understand it, the comparison here is not with the untransfected HeLa cell telomere length, but between the cell lines that are overexpressing either Wt or various hTERT mutants. The telomere length of untransfected HeLa cells over the 8-week period shown in lanes 1 and 2 are not changed, as expected. When Wt hTERT is overexpressed telomeres clearly lengthen. When the mutants are expressed telomeres fail to lengthen. Given the intrinsic heterogeneity of the telomeres, and following just one clone, is not possible to make accurate comparisons of the various lengths in cell lines where there is a failure to elongate. But it is clear that each of the three mutant analyzed fail to elongate the telomeres while overexpression of wt hTERT does show elongation. In Figure 1 the intrinsic telomerase activity of the mutants were all shown to be reduces and in fact the 78;132 mutant is 60% that of wt. (Why this is not reflected in their quantitation in Figure 4B is not clear, however this in vivo assay may not be as sensitive as the direct assay shown in Figure 1, so I am assuming the discrepancy is due to the inherent experimental error in this in vivo analysis from clones of cells). My question is simply what would happen if an hTERT mutant that is WT in the TEN domains but has 60% catalytic activity were assayed in this manner? Would there be elongation?

The experiments shown in the new Figure 7 go a long way toward solving the problem in Figure 4D. Given that there are a number of other issues with Figure 7 (see below), I would suggest that to put this issue to rest the, the authors might find a way to combine Figure 4D with Figure 7 D in one figure and not have to resort to arguments over the intrinsic 60% activity of the mutant telomerase.

The experiments to show localization of hTR to telomeres in Figure 4C are problematic and probably not necessary for the conclusions of this paper. I suggest eliminating Figure 4C. The authors show a control in Figure 4–figure supplement 1 that suggests this method to examine hTR might have serious technical issues. The HeLa cells shown in Figure 4–figure supplement 1 show that almost all of the green dots do not co-localize with TRF2 or with Colin. The authors statement: 'Although our FISH method readily detected hTR foci that co-localized with telomeres in telomerase-positive HeLa cells, hTR signals were completely absent in telomerase negative VA13 and U2OS cell lines, confirming that FISH is a valid approach to determine the subcellular localization of hTR' is not supported by the data shown in the figure. In the HeLa cells there are 6-8 large green dots and many small green dots. In the Va13 and U2OS the large green dots are not seen but just as many small green dots appear. Given that the there is only one hTR per active telomerase, and the signal would be expected to be quite faint, it is not clear what the large green dots that do not co-localize that are seen in the HeLa cells represent.

Eliminating this experiment from the paper would not alter the conclusion and could strengthen the paper.

5) Figure 5 shows the charge swap experiments. The direct assay data for the hTERT K78E TPP1 E215K combination is compelling. However the authors acknowledge that the R132E mutant does not show much compensation. It is not clear why they this is still shown in the model at the end in Figure 8 as being involved with a charge swap, when it might be some other kind of interaction.

6) In Figure 5–figure supplement 1 there seems to be is a problem with labeling of the gel, perhaps? Why is the “no PT” in this experiment more processive than the WT PT? This higher processivity of no PT is not seen in Figure 5D, Figure 5–figure supplement 3 or Figure 5–figure supplement 3.

7) (As a side note, I do not know how supplemental figures should be referred to; it seems confusing to have multiple supplemental Figure 5's. Also some of the figures could be combined or eliminated; not sure why you need Figure 5–figure supplement 1 when that same data shown later; the authors should tighten up their use of supplemental figures.)

8) Figure 6 is now out of order, the new experiments in Figure 7 relate to what is shown in Figure 4 and 5, while Figure 6 is something different.

The variability of the number mCh-hTERT dots in the nucleus of both of Figure 6B and Figure 6–figure supplementl 1 is very confusing. In the top panel of Figure 6b only one dot is shown for mCh-hTERT. This would suggest hTERT is not at normal telomeres but rather only at the lacO array. However in the lower panel with the E215TPP1 mutant, not the Wildtype hTERT is at many loci. One way to interpret that is that in this mutant now the hTERT goes to telomeres. But of course this is the opposite of the authors' conclusion. What do all those other dots represent? Do they co localize to telomeres? These experiments are very difficult to interpret. Was this quantitation shown in Figure 6C done in in a blinded fashion? Best practice for this kind of experiment is to have the person scoring the localization blind to the genotype of the cells being examined. Given the noise in this experiment, it would be a good idea to go back to the images and blind them and redo the statistics with the blinked data. The figure legend would then reflect that this experiment was carried out using best practices for such image analysis.

9) Figure 7C is equally difficult to interpret; I suggest eliminating it and merging Figure 7 A, B and D with the Southern blot in Figure 4D.

In the images in this figure most of the hTR signal does not co-localize with the Rap1. (Why was Rap1 used here when the previous figure used TRF2?) Was this quantitation carries out in a blinded fashion? If I understand the numbers in white on the final column, cells expressing K78E hTERT and WT TPP1 show 49/52 or 94% co-localization, whilst cells expressing K78E hTERT and E215K TPP1 show 20/51 or 40% co-localization. Isn't this the opposite of what you would expect? This experiment needs to be redone in a blinded fashion, or simply eliminate these co-localization experiments that are problematic.

10) The model of TPP1 interaction with TERT is shown in Figure 8 and also in Figure 5A. It is not clear that it needs to be in both places. The way that TPP1 is drawn here however might confuse people. The three domains should not have black outlines around them; it makes it look like this is three Different proteins. The color difference between TPP1 and TIN2 is subtle, so hard to know which protein the reddish colors circle belongs to. Eliminating the black line will help this. Similarly if you eliminate the black outline around the TEN domains where it touches the TERT, it will look much more like a domain and not a separate protein.

[Editors' note: further clarifications were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled “Identification of human TERT elements necessary for telomerase recruitment to telomeres” for further consideration at eLife. Your revised article has been favorably evaluated by James Manley (Senior editor) and a member of the Board of Reviewing Editors. The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

In the revised manuscript the authors address a number of concerns. However the authors have not addressed two major points that were raised.

First the evidence shown in Figure 4D is not sufficient to say that “telomeres shrink” when the TEN domain mutants are overexpressed. The heterogeneity in the telomere lengths is just too variable. It is clear that telomere do not elongate as they do when the Wt TERT is overexpressed. To convince the readers that there is actually significant shortening of telomeres, with these heterogeneous lengths, it would be necessary to show several independent clones and show a decrease occurs reproducibly in several independent clones. It is well established that there is significant telomere heterogeneity in culture, so one sample that appears marginally shorter in one lane, without repetition is not convincing. I understand redoing this experiment is a lot of work. For that reason it would be best to simply narrow the conclusion to what can clearly be established from this data: that the TEN domain mutants fail to elongate telomeres like the Wt TERT does. My earlier comment that the double mutant has less activity than the WT TERT is still a concern. This decreased activity is not shown in this figure but is shown in Figure 1 where more quantitative activity assays are done.

The second point is that the hTR localization to telomeres is expected to be a very difficult thing to measure. It is clear that hTR localization to cajal bodies can be established, and it is expected that many hTR molecules would be localized in this processing center, however it has not been clearly established that a single telomerase at a telomere can be detected by this technique.

The authors state, “Note that the FISH method we are using in this study is well established in the literature and has been used by other labs, including the Artandi and Terns labs.” Just because others use an assay, does not make it robust. For example many labs do TRAP assays to quantitate telomerase activity using 40 or more rounds of PCR; I am sure the Cech lab appreciates that while widely used, that TRAP assay is not quantitative. None of the prior publications on hTR FISH that the authors refer to did controls to show their hybridization detected single telomerase molecules at telomeres. Were mutant oligos that are mismatched still giving a signal in this assay? What to the foci that do not co localize with telomeres represent? What do the few foci of hTR hybridization represent given that that most telomeres show no signal? This lack of signal at most telomeres was even more apparent in the previous Figure 7c that the authors decided to remove.

Do the authors conclude that a signal at a telomere mean they are visualizing ONE telomerase RNA molecule localized to a telomere? Is it possible to detect just three the fluorescent probe oligos under these conditions? The TRF2 signal clearly comes from many hundreds of TRF2 proteins bound to the telomere, and yet the foci of hTR signal are bright. Have the authors considered that this may be an aggregate of hTR? It would be helpful for these authors to state directly if they think they are measuring a cluster of hTR molecules or if that they believe they are seeing one telomerase molecule on one telomere, with their microscopy. If they authors conclude there may be multiple hTR molecules in these foci, it would be helpful to explain why there would be multiple telomerase molecules and how that affects the charge swap interaction they are proposing as shown in Figure 8.

The uncertainties in these experiments weaken the manuscript. The conclusion on processivity can be made without this FISH data.

One final note it was surprising to see the authors state in the rebuttal that under some salt conditions TERT alone is more processive than TERT + Pot1/Tpp1. They say “Figure 5–figure supplement 1 is labeled correctly. The gel was a screen carried out under non-physiological salt conditions, and under these conditions telomerase alone is more processive.” This reduced my confidence in the biological result of the Pot1/Tpp1 stimulation. Why is the effect so sensitive to salt conditions? It is not clear what “non-physiological salt conditions” is and why they would have used these conditions in the experiment shown in Figure 5–figure supplement 1. Presenting this figure will be very confusing to the readers and it is not clear why it is needed. If the salt conditions are this sensitive such that the exact opposite of the conclusion the authors are making can easily be obtained, it is important to state this in the methods, and not just refer to previous papers for the details. For others to be able to reproduce the results, understanding the salt issues is important.

https://doi.org/10.7554/eLife.03563.019

Author response

We thank you for your detailed comments, which we’ve taken very seriously in our attached revision. In several cases we present new experiments: new Figure 3A and B in response to request to quantify expression levels, a Figure 4–figure supplement 1 showing the requested controls for hTR FISH using telomerase-negative VA13 and U2OS cell lines, the requested test of “direct binding” by expressing telomerase in RRL’s in a new Figure 5–figure supplement 3, and a new Figure 7 showing dramatic rescue of telomerase recruitment and telomere length extension in vivo with the charge-swap mutant hTERT and TPP1 combination. The requested re-analysis of enzyme processivity using a second method is presented in Figure 1–figure supplement 2. Finally, the writing has been clarified throughout, and a revised cartoon in Figure 5A should make it easier for the reader to understand why we consider the charge-swap mutants to provide such a strong argument for a direct protein-protein interaction, especially in light of other mutants at non-interacting sites which show the opposite effect.

Major concerns:

If the authors would like to conclude that there is a direct interaction of hTERT TEN domain with TPP1 OB fold evidence for direct protein interacts needs to be presented. The hypothesis they are trying to test is whether these two proteins interact and if that interaction is mediated by the charged residues on the TEL patch of TPP1 and the opposite charges on the TERT- TEN domain. There are a number of biochemical assays that they could use. They could purify the specific proteins or the domains and test interaction and affinities on Biacore chips. They could use GST pull down experiments, affinity columns or similar techniques. This assays presented here including the Cajal body localization and LacO array recruitment are very indirect and are in vivo. Thus they do not exclude a bridging protein, as the authors try to conclude. For the authors to conclude that the interaction of TERT and TPP1 is a direct interaction, and mediated by the charged residues they identify, evidence of direct protein interaction in vitro needs to be presented.

Purification of telomerase in quantities large enough for biochemical binding assays, such as the ones suggested by the reviewers, has not been accomplished in any lab. The data presented in this study include biochemical assays testing the interaction between TPP1 and telomerase in vitro. It is well established in the literature that the stimulation of telomerase processivity in vitro by POT1-TPP1 is due to an interaction between the TEL-patch of TPP1 and telomerase (Nandakumar et al, 2012; Sexton et al, 2012; Zaug et al, 2010). We carry out these assays with recombinant POT1-TPP1 complex, purified from insect cells and bacteria respectively, and telomerase immuno-purified from HEK293T using an affinity tag and protease mediated elution after stringent washing steps. Therefore, the only way a potential bridging factor could still be present is if it were to be tightly associated with telomerase. To address the reviewer’s concerns about a co-purifying bridging factor, we have repeated the in vitro charge swap experiments with telomerase produced and purified from rabbit reticulocyte lysates (RRL). The results with obtained using RRL telomerase are consistent with those observed with immuno-purified super-telomerase, with statistically significant differences. In addition we have also included new data, demonstrating that E215K TPP1 restores telomeric localization and telomere maintenance in cells. In total we now demonstrate the charge rescue using multiple independent experiments in vivo and telomerase from two sources in vitro. In addition we would like to point out that a compensatory effect of combining two individually deleterious mutations is inconsistent with the presence of a bridging factor. We have now illustrated this concept in more detail in Figure 5A and provided several additional examples from the literature in the Discussion, which use charge-swap experiments to assess direct protein-protein interactions.

Specific comments:

1) Figure 1b: the diagram has blue highlighted text indicating conserved residues; however additional mutants that are not in blue are shown in the Figure 1 supplement. There needs to be an explanation for how specific residues were chosen for mutation and need to state in text, all of the mutants that were made and why. How the mutant combinations in in Table 1 were chosen should be described.

We have modified the text to further clarify our rationale for choosing basic residues in the TEN domain . Without a priori knowledge of which hTERT residues interact with TEL-patch, we chose to mutate as many basic residues as we reasonably could. Mutants with multiple mutations were made to screen TEN domain residues more efficiently, or to examine the compound effects of single mutations.

2) The results shown in Figure 1 and Table 1 are essential for understanding the rest of the paper. There are a number of issues in these two figures that need to be clarified. In Figure 1 F, G and H the authors set up the activity assay and the mutants that are the basis for all of their experiments going forward. It is important to explain in the text what these assays are, and what is concluded from them.

We thank the reviewers for their comments; these issues have been corrected in the text of the manuscript.

I am assuming that the intrinsic changes in processivity are being ignored (Figure 1G)

Correct. In addition, the intrinsic processivity differences were influenced by a normalization artefact. We have re-calculated processivity as per the reviewer suggestion, and the differences are now minimal.

…and that the PT stimulation is being used as a measure of the interaction of TEN 1 and TPP1. However this is never stated directly.

Indeed PT stimulation is used as a measure of the interaction between TEN and TPP1, we have further clarified this in the text.

The authors need to explain how they define processivity and PT stimulation differently and how it is quantitated.

We have also clarified this within the text and Materials and methods.

This data is quantitated differently from a previous report by the Cech lab in Latrick et. al 2010. The authors should provide an explanation for changing the processivity analysis from previous work, since choosing to sum all counts above band #9 seems a bit arbitrary and less rigorous than previous studies.

We now present processivity and RAP stimulation by PT values made using the decay method described by Latrick and Cech, so Table 1, Figure 1, and Figure 5 are updated. In addition we compare the +9 (fractional) and decay methods directly Figure 1–figure supplement 2, Figure 5–figure supplement 2. Importantly, both methods of quantitation provide similar numbers as observed in previous comparisons (Nandakumar et al, 2012; Zaug et al, 2013).

In Figure 1G, there is a clear increased processivity of R132E and K78E+ R132E. This 2-fold increase is not evident from the gel in Figure 1E. And also it is not commented on in the paper. If this is true and not due to a normalization artifact, how does this change in processivity affect telomere length? This increased processivity is also reported in table 1 as 186%, a large effect.

We thank the reviewers for the suggestion; the large processivity effects were influenced by normalization and have now been corrected.

Currently there is very little description on how the authors determine the values they report on Table 1 for telomerase activity, RAP, and RAP stimulation by POT1-TPP1.

We have further clarified our description in materials and methods, as well as in the table and in the text itself (and Materials and methods).

Since IP-efficiencies and/or expression of hTERT vary greatly among different hTERT mutants, it was often hard to see the correlation between what is shown on gels vs. values in Table 1. Thus, it would be very important for the authors to better describe their quantification methods, and provide raw data before normalization. Were these measurements done on serial dilutions of the extract to increase the accuracy?

We have further clarified the quantification of telomerase enzyme assays in the Materials and methods and Table 1. The purpose of supplement to

Figure 1 is to show the raw data for a variety of hTERT mutants that were screened in order to identify separation-of-function mutants. The mutagenesis and production of super-telomerase is technically challenging, expensive, and time consuming. For this reason the mutants were produced in small groups at different times, which can result in expression variability. We controlled for this by simultaneously producing wild-type telomerase with each group of mutants. As the purpose of the screen was to identify separation-of-function mutants, we did not further characterize many of the mutants from the screen.

The data in supplement to Figure 1 and in Table 1 show that we screened a variety of mutants. In contrast, Figure 1 presents data that is integral to the main conclusions of the paper. For Figure 1, the production of wild-type and mutant super-telomerases were carried out simultaneously, and in this case the hTERT expression levels were quite similar, see Figure 1D. Although we do not typically do serial dilutions, the telomerase assays were done under standard Cech Lab conditions, and the linearity of our measurements with respect to concentration and time have been fully documented in of a previous report (Zaug et al, 2013).

How reproducible are these assays? How many experiments were done to generate are in the error bars shown?

The number of experimental replicates is listed for all in vitro telomerase assays as well as in vivo cell-based experiments, in each figure legend or detailed in the table. The error bars on the graphs or standard deviations noted in Table 1 give a measure of the reproducibility of the assays.

Figure1 and Table 1 describe some of the same mutants. Figure 1H axis label says “normalized PT stimulation” while Table 1 column 4 says “% RAP stimulation”. I assume these are the same thing? Need to use consistent terminology if the reader is to follow.

Yes. This oversight on our part has been corrected.

The authors say that K78E and R132E are required for RAP stimulation but not essential for enzyme activity, however the double mutant has only 63 % activity. This is overstating the conclusion that these residues are not needed for activity and they need to make conclusions that represent the data presented.

The text has been modified to more accurately emphasize the point we intended to make. K78E hTERT has more than 90% of wild-type activity and intrinsic processivity, and all experiments testing the direct nature of the hTERT-TPP1 are done using this allele.

3) The overexpression studies need to be cautiously interpreted. To determine if the TERT TEN domain mutants localize to telomeres, the authors overexpress the mutants and examined localization to telomeres or Cajal bodies. This experiment is not sufficient to conclude the proteins fail to localize to telomeres. Overexpression of proteins is known to cause spurious cellular localization. The fact that overexpression of the wt gives different pattern than overexpression of the mutant only indicates that the mutants may aggregate differently in the cell.

In terms of overexpression (OE) causing aggregation, the catalytic activity of telomerase is very sensitive to the integrity of this RNP complex; our careful measurements of the activity (Vmax and Km) of the overexpressed and endogenous telomerases show that they have the same activity per molecule (Xi & Cech, 2014), arguing against aggregation upon OE. Furthermore, the measurements herein show that key mutants retain WT-like activity, arguing against differential aggregation. And while we agree with the reviewers that protein OE can in some cases lead to abnormal sub-cellular localization of proteins, our mutants do not form inclusion bodies or localize to non-physiological structures. Instead, our TEN-domain mutants fail to localize to telomeres despite OE, and remain sequestered in Cajal bodies. Since TEN-domain mutants localize to Cajal bodies and not some other structure, there is no reason to believe that there is an assembly or maturation defect caused by overexpression. We believe the fact that TEN-domain mutants can’t be forced to localize to telomeres even when over-expressed strengthens the conclusion that they fail to interact with TPP1. We would also like to point out that over-expression of telomerase has been previously used to analyse its subcellular localization, including the relevance of hTERT residues on this process (Cristofari et al, 2007; Stern et al, 2012; Zhong et al, 2012).

If they want do make conclusion about protein localization, a second method without overexpression is needed. The authors state they have an antibody to TERT, so ChIP and Co-IP might be an alternative.

We appreciate the reviewers’ comment, but the only approach to avoid over-expression is genome editing, which is far beyond the scope of this study. We would like to point out that we also tested telomerase localization by hTR FISH under conditions where only hTERT is over-expressed, which only raised telomerase activity per cell about 3-4-fold, compared to the ∼200-fold in super-telomerase cells (Cristofari et al, 2007). Furthermore, our LacO experiments in Figure 6 demonstrate that K78E telomerase is fully capable of associating with the OB-fold domain of TPP1 when the E215K mutation is introduced. Finally, we have now included experiments showing that introduction of E215K TPP1, but not WT or E169K TPP1, rescues telomere localization of K78E telomerase, adding further support for our model (Figure 7).

Finally the authors do not show the level of over expression or look at the stoichiometry of the various holoenzyme components. The authors conclusion from this experiment that TEN domain and and TPP interact is not supported by these experiments. It is not clear this experiment is helpful to the conclusions given the caveats of over expression.

We have now included western and northern blots that demonstrate that hTERT and hTR are over-expressed to similar levels under all conditions, ruling out the possibility that the phenotypes observed are due to variable expression levels of the holoenzyme components (Figure 3A-B).

4) It is not clear how the authors are measuring telomerase activity, as it does not seem to be consistent. In Figure 1 the K78E+R132E mutant is reported to have 60 % telomerase activity but here Figure 4b the activity is more than wildtype. I wonder how these numbers are normalized to give these values. It would be much better to show a titration activity with amounts normalized by hTERT levels. The conclusion in the text is that there is no effect of the mutants on activity, but an effect on telomere length. This conclusion is not supported by the data given the results of reduced activity in Figure 1.

Thank you for this comment. We were not clear in the text about the differences between the experiments in Figures 1 and 4, and have proceeded to point out the specific differences and conclusion in the revised text. In Figure 1 we analyse the activity of telomerase over-expressed in HEK293T cells followed by immuno-purification using the ProA-tag on the N-terminus of hTERT. The activity levels in the experiments in figure 1 are normalized to both the loading control and the intensity of the western blot band shown. The activity shown in Figure 4 is a measure of telomerase activity per cell and is normalized to both loading control and the number of cells used to purify telomerase. Since endogenous telomerase lacks an affinity tag we had to use an alternative approach to purify telomerase from the cell lines stably expressing the different hTERT alleles and the parental HeLa cell line. We chose to use a well-established method using a polyclonal antibody against hTERT to immuno-purify telomerase from these cell lines (Cohen & Reddel, 2008). The results in Figure 4B demonstrate that all cell lines over-expressing the hTERT alleles have a 3-4-fold higher telomerase activity than the parental HeLa cell line. This observation is crucial since it rules out lower telomerase activity as a possible explanation for the reduced telomere length in the cell lines expressing hTERT with mutations in the TEN-domain. To the contrary, telomeres shrink in cell-lines expressing TEN-domain mutant alleles despite 3-4-fold higher telomerase activity per cell. This observation in combination with the lack of telomerase present at telomeres in these cells, lead us to the conclusion that telomeres shrink due to the failure of telomerase association with telomeres. The differences in activity observed could be due to the experimental set up or purification procedure, but they don’t compromise the main conclusions of either figure: K78 and R132 are involved in the interaction between telomerase and TPP1 (and to a lesser degree impact catalytic activity) and thus fail to maintain telomere length due to the failure to localize to telomeres.

5) The method used to show localization of telomerase does not have any controls. In Figure 4C the authors show a robust signal for endogenous hTR localization in cells. Given that there is a copy number of about 100 molecules of hTR per cell, it is not clear what could be generating such a robust signal. To know that this technique can indeed localize endogenous hTR, the authors need a control experiment. They should use a cell line that does not express hTR such as VA13 and show there is no signal with this technique, or use mouse cell lines that are deleted for the endogenous mTR and show there is no signal. Unless a control for specificity is shown this experiment should be omitted. The authors conclude from this experiment that mutant hTERT exerts a dominant effect on endogenous hTR localization. This conclusion is not justified by the data.

We have now included VA13 and U2OS ALT cell lines as controls for the FISH experiment (Figure 4–figure supplement 1) and neither cell line has detectable hTR foci at telomeres. Thus the conclusion that over-expression of an hTERT allele, that fails to localize to telomeres, displaces endogenous hTR from telomeres is fully supported by the data presented.

6) Figure 4D. In all of the previous experiments the authors overexpressed both hTERT and hTR, in this experiment they only overexpressed hTERT; it is not clear why this was done.

To assess telomere length over an extended period of time it was necessary to stably express all hTERT variants in HeLa cells. Importantly, all hTERT variants are overexpressed relative to endogenous telomerase (see western, Figure 4A). Therefore, the vast majority of hTR, which is the limiting component for telomerase assembly under these conditions, should be assembled with exogenous hTERT molecules. Additionally, overexpression of only hTERT leads to a 3-4-fold increase in telomerase activity since endogenous hTR is not limiting the level of telomerase, while overexpression of both components leads to a 200-fold increase (Cristofari et al, 2007). Therefore, by choosing to only overexpress hTERT, telomerase activity is much more similar to endogenous telomerase activity than if both components were overexpressed. We have now clarified the rationale for the experimental set up in the text.

Given that the mutants K78E and R132E showed decreased telomerase enzyme activity in the experiments in Figure 1, it is not surprising that there is less telomere elongation seen with these overexpressed mutants. Over expression of these mutant enzymes that have decreased activity will sequester all of the hTR and result in overall reduced telomerase activity and thus explain reduced telomere elongation. The authors' conclusion that this data shows that the TEN domain mutants fail to localize but are active is not warranted by the data presented.

This concern is not true, and we have clarified the situation in the revision. In fact, all cell lines have at least 3-fold higher telomerase activity per cell, relative to the parental HeLa cell line, due to hTERT overexpression (Figure 4B). As described (Xi & Cech, 2014), overexpression of hTERT increases the fraction of hTR assembled in telomerase RNPs. We have now clarified this reasoning in the text. The parental HeLa cell line maintains a constant telomere length. In contrast, telomeres shrink in cell lines expressing K78E, R132E, or K78E;R132E hTERT despite ∼3-4-fold higher telomerase activity per cell than the parental HeLa cell line (Figure 4D). Additionally, the cell line expressing wild-type hTERT displays significant telomere lengthening with similar telomerase activity per cell compared to all mutant cell lines. Thus, decreased telomerase activity can’t explain telomere shrinking in TEN domain mutant cell lines. Since hTR is sequestered in Cajal bodies in all mutant cell lines, the data fully support the model that telomeres shrink because mutations in the TEN-domain compromise telomerase localization to telomeres.

7) The authors describe a charge swap experiment in which they mutate residues in TPP1 that they think might interact with the TEN domain acidic residues in Figure 5. The authors state that they mutated “a number of residues in TPP1”, however they then just show E169K and E215K. It is important for the authors to list all the mutants tested and show us what the phenotypes are for all of them, otherwise the specificity of these chosen mutants that will be followed going forward is not clear.

TPP1 mutant E169K was included as a control to demonstrate specificity of E215K for telomerase K78E. In addition we have added Figure 5–figure supplement 1, including all other residues mutated in TPP1, demonstrating that only E215K compensates for the presence of the K78E mutation in telomerase.

8) The authors use an assay developed by Zhong et al to examine ectopic localization of Tert to a chromosome locus where the OB fold domain of TPP1 has been tethered in Figure 6. They use this assay to demonstrate that mutants in the TEN domains disrupt localization but the double mutant of TEN K78E with TPP1 E215K has an increased number of foci. The figure shown in Figure 6 B is not helpful. This figure only shows single cell nuclei and the point of the assay is to examine the number of nuclei with foci. So it would only make sense to show a field of cells, not single cells. So either do not show Figure 6 B or show a field that would represent the data they are trying to relate.

Thank you for the suggestion. We fully agree and have added additional examples of cells for all panels relevant for the charge-swap as Figure 6–figure supplement 1. Due to the transfection efficiency being less than 100% it is not possible to capture fields with high densities of relevant cells. The images presented in the main figure are representative images of the 117-176 nuclei evaluated per condition in three independent experiments.

More importantly since this is an in vivo assay, if there were a bridging protein it would still be present, so the main conclusion about a charged interaction is being the main interaction cannot be concluded from this experiment.

We believe that the LacO-assay presented in Figure 6 fully supports our model that the OB-fold of TPP1 and the TEN-domain must interact directly, via the following argument. Mutation of the TEN-domain or the TEL-patch independently reduces the association between telomerase and the OB-fold domain of TPP1, which is manifested in a reduced fraction of nuclei in which telomerase and the GFP-TPP1 OB-LacO foci co-localize. If each independent mutation reduces the affinity for one or more potential bridging factors, simultaneous mutation of the TEN-domain and the TEL-patch should lead to an additive reduction of the fraction of nuclei that show co-localization of telomerase and OB-fold foci. We observe the opposite result: mutation of the TEL-patch rescues the presence of a mutation in the TEN-domain. While WT-telomerase co-localizes with E215K OB-fold in only 20% of the nuclei, K78E-telomerase co-localizes with E215K OB-fold in over 75% of the nuclei. The only plausible explanation is that mutation of the TEN-domain or TEL-patch independently eliminates a direct interaction between these charged residues, which is re-established by swapping the charged residues on both surfaces. We have now added an illustration to figure 5A outlining not only how a charge swap can support a direct interaction between the TEN-domain and the TEL-patch, but also how it can rule out the presence of a bridging factor. (See also response to #9 below.)

9) The authors add analysis of V144 at the end of the paper in Figure 7. This figure does not contribute in any way to the point of this paper. The V-144 mutation may be of interest in telomerase function in IPF, but the addition of this mutant onto the end of this paper is distracting and does not belong here. This data is best published in another study and should be removed from this manuscript.

We agree with the reviewers that the data on the V144M mutation is out of place at this point in the manuscript, yet it is of interest to explain the molecular basis of this allele causing IPF. We have instead included it as supplement to Figure 5 since it illustrates how mutations in the TEL-patch and the TEN-domain display additive losses in RAP stimulation if they affect residues that do not directly interact, solidifying our in vitro analysis of the interaction between telomerase and TPP1.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

1) [… In] the text of the manuscript the authors did not make a strong case for how the PT stimulation assay is being used as a proxy for protein-protein interactions. Instead the text was altered in the opposite direction. They have removed the wording that stated they were assaying RAP stimulation and instead now say they are measuring protein interactions.

This sentence is in the introduction summarizing our conclusions.

I think it is important for the reader to understand the actual experiments that are done measure RAP stimulation, and then at the end of the paragraph the authors can say that they interpret this as a direct interaction and they will go on to test this.

Further on in the text, the authors are again mixing the descriptions of the results with their interpretation. The previous text that is marked as deleted gives a more straightforward description of that was measured and is interpretation from that result. A reversion to this kind of language would help the manuscript significantly.

We have taken the reviewer’s suggestion and reverted to the original language used to discuss the results at the end of section 1.

2) The yellow color in Figure 2 is very difficult to see, I suggest using a different color.

We have revised Figure 2 for better visibility.

3) The authors use the jargon “Super telomerase” in some cases in the text but not constantly. They are referring to a term used in a paper by Lingner's group when hTERT and hTR are co-overexpressed. The authors use “super telomerase” without explaining what it means. In the methods section the authors state they use the direct telomerase assay described by Zaug 2013. In that paper both hTERT and hTR are co-overexpressed but the term “Super telomerase” is not used. Since the Zaug 2013 assay is used in Figure 1, technically this could be called super telomerase. I suggest that the term be eliminated and the experiment just described as they are. (Note Figure 5–figure supplement 3 also uses the term super telomerase.)

We have removed the term “super telomerase” from text and figures.

4) […] In their rebuttal the authors state:

“The results in Figure 4B demonstrate that all cell lines over-expressing the hTERT alleles have a 3-4-fold higher telomerase activity than the parental HeLa cell line. This observation is crucial since it rules out lower telomerase activity as a possible explanation for the reduced telomere length in the cell lines expressing hTERT with mutations in the TEN-domain. To the contrary, telomeres shrink in cell-lines expressing TEN-domain mutant alleles despite 3-4-fold higher telomerase activity per cell.

I am not sure I follow the logic of comparison of the level of activity in untransfected HeLa cells. […] My question is simply what would happen if an hTERT mutant that is WT in the TEN domains but has 60% catalytic activity were assayed in this manner? Would there be elongation?

We have sought to clarify this further within the text. We understand the reviewer’s concerns about the differences between the activity levels in Figures 1 and 4. The purification methods are different in these experiments and might be the reason for the differences. There is a higher degree of variability in the data shown in Figure 4, as is reflected in the error bars. The conclusion we are trying to draw from these experiments is that telomeres shrink due to a defect in telomere recruitment of telomerase, not an activity defect. The untransfected control maintains the same telomere length, while overexpression of WT hTERT leads to a significant growth in telomere length, due to a 4-fold increase in telomerase activity per cell compared to the untransfected control. In contrast, overexpression of TEN-domain mutants leads to telomere shortening, even though the telomerase activity per cell is increased to a similar degree as with WT hTERT. Thus, cells expressing TEN-domain mutants have much higher telomerase activity than untransfected cells, yet their telomeres are not maintained at the same level as untransfected cells, therefore an activity defect cannot be the cause of telomere erosion. If the telomere shortening of R132E and K78E;R132E were merely due to an activity defect, overexpression of hTERT and thereby raising the activity level 3-4 fold should compensate for the activity defect, which it does not.

The experiments shown in the new Figure 7 go a long way toward solving the problem in Figure 4D. Given that there are a number of other issues with Figure 7 (see below), I would suggest that to put this issue to rest the, the authors might find a way to combine Figure 4D with Figure 7 D in one figure and not have to resort to arguments over the intrinsic 60% activity of the mutant telomerase.

All conclusions in Figure 7 as well as the charge-swap supporting the direct interaction of telomerase with TPP1 are based on the K78E mutant, which has greater than 90% activity in all assays. Combining Figures 4 and 7 would disrupt the logic of the paper, so we have not incorporated the reviewer’s suggestion in this case.

The experiments to show localization of hTR to telomeres in Figure 4C are problematic and probably not necessary for the conclusions of this paper. I suggest eliminating Figure 4C. The authors show a control in Figure 4–figure supplement 1 that suggests this method to examine hTR might have serious technical issues. The HeLa cells shown in Figure 4–figure supplement 1 show that almost all of the green dots do not co-localize with TRF2 or with Colin. The authors statement: 'Although our FISH method readily detected hTR foci that co-localized with telomeres in telomerase-positive HeLa cells, hTR signals were completely absent in telomerase negative VA13 and U2OS cell lines, confirming that FISH is a valid approach to determine the subcellular localization of hTR' is not supported by the data shown in the figure. In the HeLa cells there are 6-8 large green dots and many small green dots. In the Va13 and U2OS the large green dots are not seen but just as many small green dots appear. Given that the there is only one hTR per active telomerase, and the signal would be expected to be quite faint, it is not clear what the large green dots that do not co-localize that are seen in the HeLa cells represent.

Eliminating this experiment from the paper would not alter the conclusion and could strengthen the paper.

In the FISH control HeLa cells show 10 bright signals that are not present in the VA13 or U2OS cells in addition the background signals that are present in all samples. 9 out of these 10 signals co-localize with either the telomeric or the Cajal body marker, which is the expected localization of hTR. We have clarified the text to accommodate the reviewer’s concern. (Note that the FISH method we are using in this study is well established in the literature and has been used by other labs, including the Artandi and Terns labs. We scaled the images to include some background signal, which is the appropriate practice with microscopy data.)

5) Figure 5 shows the charge swap experiments. The direct assay data for the hTERT K78E TPP1 E215K combination is compelling. However the authors acknowledge that the R132E mutant does not show much compensation. It is not clear why they this is still shown in the model at the end in Figure 8 as being involved with a charge swap, when it might be some other kind of interaction.

As noted in the text, mutation to R132 compromises RAP stimulation by PT, suggesting that it interacts with the TPP1 TEL-patch. Although we cannot rescue the defect of R132E with a charge-swapped TPP1, the stimulation defect suggests that it is interacting. We have clarified the text in figure legend 8 to make it clear that the question mark does not represent a salt-bridge.

6) In Figure 5–figure supplement 1 there seems to be is a problem with labeling of the gel, perhaps? Why is the “no PT” in this experiment more processive than the WT PT? This higher processivity of no PT is not seen in Figure 5D, Figure 5–figure supplement 3 or Figure 5–figure supplement 3.

Figure 5–figure supplement 1 is labeled correctly. The gel was a screen carried out under non-physiological salt conditions, and under these conditions telomerase alone is more processive. We have clarified this in the figure legend, and note that the charge-swap works under a variety of in vitro conditions.

7) (As a side note, I do not know how supplemental figures should be referred to; it seems confusing to have multiple supplemental Figure 5's. Also some of the figures could be combined or eliminated; not sure why you need Figure 5–figure supplement 1 when that same data shown later; the authors should tighten up their use of supplemental figures.)

We are following the eLife guidelines for supplemental figures. Figure 5–figure supplement 1, as well as the majority of other figure supplements, were included to address reviewer concerns, or requests for additional data.

8) Figure 6 is now out of order, the new experiments in Figure 7 relate to what is shown in Figure 4 and 5, while Figure 6 is something different.

The logic of the paper is as follows. We first demonstrate the interaction between the TEN-domain in vitro and in vivo and then go on to demonstrate that the interaction is direct in Figures 5-7. Figure 5 shows the in vitro rescue and Figure 6 the in vivo rescue. We therefore wish to maintain the figure order as is.

The variability of the number mCh-hTERT dots in the nucleus of both of Figure 6B and Figure 6–figure supplement 1 is very confusing. In the top panel of Figure 6b only one dot is shown for mCh-hTERT. This would suggest hTERT is not at normal telomeres but rather only at the lacO array. However in the lower panel with the E215TPP1 mutant, not the Wildtype hTERT is at many loci. One way to interpret that is that in this mutant now the hTERT goes to telomeres. But of course this is the opposite of the authors' conclusion. What do all those other dots represent? Do they co localize to telomeres? These experiments are very difficult to interpret. Was this quantitation shown in Figure 6C done in in a blinded fashion? Best practice for this kind of experiment is to have the person scoring the localization blind to the genotype of the cells being examined. Given the noise in this experiment, it would be a good idea to go back to the images and blind them and redo the statistics with the blinked data. The figure legend would then reflect that this experiment was carried out using best practices for such image analysis.

We have added text to further clarify the lacO experiment. When expressing WT mCh-hTERT alongside E215K TPP1, telomerase cannot interact with the TPP1-OB-LacI fusion, but since it is WT it can localize to endogenous telomeres, as visualized by the large number of foci present under these conditions. When it is expressed alongside WT TPP1 it localizes to the LacI focus since the amount of TPP1 present at this site is far greater than any individual telomere. When telomerase can interact with neither the telomere nor the TPP1-OB fusion, as is the case with R132 and R132/K78E mCh-hTERT, it is diffusely localized in the nucleoplasm. We agree with the reviewer that best practice is to blindly quantify experiments if they involve the judgment of the observer. In this case, however, the results are very clear: there is either a mCherry signal present at the GFP focus or there is not. The experiment was carried out three independent times quantifying a large number of nuclei and is therefore highly reliable.

9) Figure 7C is equally difficult to interpret; I suggest eliminating it and merging Figure 7 A, B and D with the Southern blot in Figure 4D.

In the images in this figure most of the hTR signal does not co-localize with the Rap1. (Why was Rap1 used here when the previous figure used TRF2?) Was this quantitation carries out in a blinded fashion? If I understand the numbers in white on the final column, cells expressing K78E hTERT and WT TPP1 show 49/52 or 94% co-localization, whilst cells expressing K78E hTERT and E215K TPP1 show 20/51 or 40% co-localization. Isn't this the opposite of what you would expect? This experiment needs to be redone in a blinded fashion, or simply eliminate these co-localization experiments that are problematic.

We acknowledge the reviewer’s concerns regarding the FISH experiment in Figure 7. Since it is not essential to the conclusions of the paper, we have therefore eliminated it.

10) The model of TPP1 interaction with TERT is shown in Figure 8 and also in Figure 5A. It is not clear that it needs to be in both places. The way that TPP1 is drawn here however might confuse people. The three domains should not have black outlines around them; it makes it look like this is three Different proteins. The color difference between TPP1 and TIN2 is subtle, so hard to know which protein the reddish colors circle belongs to. Eliminating the black line will help this. Similarly if you eliminate the black outline around the TEN domains where it touches the TERT, it will look much more like a domain and not a separate protein.

The schematic in Figure 5A explains the expected results and logic for a charge-swap experiment. In contrast, Figure 8 depicts a model for telomerase recruitment in cells, and the figures are not redundant. We have modified the colors and black outlines in Figure 8 to enhance the clarity of our model.

[Editors' note: further clarifications were requested prior to acceptance, as described below.]

[… It] would be best to simply narrow the conclusion to what can clearly be established from this data: that the TEN domain mutants fail to elongate telomeres like the Wt TERT does.

Using precise language is always good, so we thank the reviewer for this suggestion. We now state that telomerase with mutations in the TEN-domain fail to elongate telomeres.

My earlier comment that the double mutant has less activity than the WT TERT is still a concern. This decreased activity is not shown in this figure but is shown in Figure 1 where more quantitative activity assays are done.

We have added a clause in the text acknowledging that the double mutant fails to elongate telomeres, but we cannot rule out the impact of reduced activity for this mutant. Thus, our conclusion rests on the two single mutants.

The second point is that the hTR localization to telomeres is expected to be a very difficult thing to measure. It is clear that hTR localization to cajal bodies can be established, and it is expected that many hTR molecules would be localized in this processing center, however it has not been clearly established that a single telomerase at a telomere can be detected by this technique. […] If they authors conclude there may be multiple hTR molecules in these foci, it would be helpful to explain why there would be multiple telomerase molecules and how that affects the charge swap interaction they are proposing as shown in Figure 8.

The uncertainties in these experiments weaken the manuscript. The conclusion on processivity can be made without this FISH data.

We have now removed the FISH experiments as suggested.

One final note it was surprising to see the authors state in the rebuttal that under some salt conditions TERT alone is more processive than TERT + Pot1/Tpp1.

Not true – this is a misunderstanding (see below).

They say “Figure 5–figure supplement 1 is labeled correctly. The gel was a screen carried out under non-physiological salt conditions, and under these conditions telomerase alone is more processive.” This reduced my confidence in the biological result of the Pot1/Tpp1 stimulation. Why is the effect so sensitive to salt conditions?

The effect of POT1-TPP1 is not so sensitive to salt conditions; it is the intrinsic processivity that is sensitive. The charge swap effect occurs in a variety of salt concentrations, as we stated in the last response. Interestingly, the effect of salt on processivity (higher salt encourages product dissociation and therefore low processivity) has been seen for many different polymerases (e.g., P. Von Hippel et al. (1994) “On the Processivity of Polymerases,” Annals of the N.Y. Academy of Science); that paper also defines physiological salt.

It is not clear what “non-physiological salt conditions” is and why they would have used these conditions in the experiment shown in Figure 5–figure supplement 1. Presenting this figure will be very confusing to the readers and it is not clear why it is needed.

The resubmitted manuscript now has this supplemental figure removed, as the reviewer suggests.

https://doi.org/10.7554/eLife.03563.020

Article and author information

Author details

  1. Jens C Schmidt

    Department of Chemistry and Biochemistry, BioFrontiers Institute, Howard Hughes Medical Institute, University of Colorado Boulder, Boulder, United States
    Contribution
    JCS, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Andrew B Dalby
    Competing interests
    The authors declare that no competing interests exist.
  2. Andrew B Dalby

    Department of Chemistry and Biochemistry, BioFrontiers Institute, Howard Hughes Medical Institute, University of Colorado Boulder, Boulder, United States
    Contribution
    ABD, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Contributed equally with
    Jens C Schmidt
    Competing interests
    The authors declare that no competing interests exist.
  3. Thomas R Cech

    Department of Chemistry and Biochemistry, BioFrontiers Institute, Howard Hughes Medical Institute, University of Colorado Boulder, Boulder, United States
    Contribution
    TRC, Conception and design, Drafting or revising the article
    For correspondence
    thomas.cech@Colorado.EDU
    Competing interests
    The authors declare that no competing interests exist.

Funding

National Institutes of Health (RO1-GM099705)

  • Andrew B Dalby
  • Thomas R Cech

Howard Hughes Medical Institute

  • Thomas R Cech

Damon Runyon Cancer Research Foundation (DRG-2169-13)

  • Jens C Schmidt

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Iain Cheeseman, Scott Cohen, Joachim Lingner, and David Spector for generously providing experimental reagents, and members of the Cech lab for helpful discussions. J C S is a Merck fellow of the Damon Runyon Cancer Research Foundation (DRG-2169-13). T R C is an investigator of the HHMI. This work was supported by NIH grant R01 GM099705 to T R C.

Reviewing Editor

  1. Carol Greider, Johns Hopkins University, United States

Publication history

  1. Received: June 3, 2014
  2. Accepted: October 1, 2014
  3. Accepted Manuscript published: October 1, 2014 (version 1)
  4. Version of Record published: November 3, 2014 (version 2)

Copyright

© 2014, Schmidt et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,286
    Page views
  • 376
    Downloads
  • 45
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Leslie A McCauliff et al.
    Research Article Updated
    1. Biochemistry and Chemical Biology
    2. Microbiology and Infectious Disease
    James Lee et al.
    Research Article