TRIP6 and LPP, but not Zyxin, are present at a subset of telomeres in human cells

Abstract

The protection of chromosome ends requires the inhibition of DNA damage responses at telomeres. This inhibition is exerted in great part by the shelterin complex, known to prevent inappropriate ATM and ATR activation. The molecular mechanisms by which shelterin protects telomeres are incompletely understood. Recently, we have implicated for the first time a class of molecules, LIM domain proteins, in telomere protection. This protection occurred through interaction with shelterin, possibly through POT1, and required the pair of LIM proteins TRIP6 and LPP, themselves part of the Zyxin family. The domain similarity between TRIP6, LPP and Zyxin led us to ask whether the latter also interacted with telomeres. Here, we show that there is specificity in the association of LIM proteins with telomeres: Zyxin, despite a high degree of similarity with TRIP6 and LPP, was not detected at telomeres, nor found in a complex with shelterin. TRIP6 and LPP, however, were detected by immunofluorescence at a small subset of telomeres, perhaps those that are critically short. We speculate that specific LIM proteins are part of complex events occurring in the context of the telomere dysfunction response and are possibly at play during the induction of senescence.

Introduction

Most primary human cells are endowed with a finite replicative lifespan, which is controlled by events occurring after significant telomere shortening. Indeed, telomere erosion constitutes one of the triggers for cellular senescence, an irreversible state of cell cycle arrest, known to be an important tumor suppressor mechanism.^1,2 During replicative senescence, short telomeres are able to signal through a pathway akin to the DNA damage response,^3,4 with activation of both ATM and ATR kinases.⁵ These complex pathways are repressed by the activity of a telomere-specific complex, named shelterin,^6,7 known to directly inhibit ATM,^8,9 and sequester the telomeric overhang, preventing ATR activation.^10,11 The disruption of this complex is known to induce telomere abnormalities such as chromosome end-to-end fusions, telomere associations, double-minute chromosomes, and, notably, a significant increase in telomere dysfunction induced foci (TIFs). TIFs result from the accumulation of DNA damage proteins such as p53BP1, γH2AX or MCD1.¹² In tumor cells, a low, endogenous level of DNA damage foci is observed on chromosome arms and telomeres, as measured by the number of γH2AX in nuclei.¹³ Depletion of TRF2 or POT1 by siRNA leads to a great increase in foci located at chromosome ends, indicative of telomere deprotection and leading to apoptosis or premature senescence.^3,14,15 In primary, telomerase-negative, human cells, replicative senescence is also associated with a significant increase in TIFs, resulting from lowered amounts of shelterin complexes on short telomeres.⁴ This pathway of irreversible G₁ arrest has been proposed to result from signaling from a subset of the shortest telomeres and not from overall average telomere size.^1,16

In addition to shelterin, mammalian telomeres contain other proteins that make important contributions to telomere protection and regulation. These accessory factors, while found at telomeres, often possess functions unrelated to telomeres and their presence can be highly regulated or in lower quantities, than those observed for shelterin.³⁰ The functionality of these factors and their presence at telomeres play an important role for telomere integrity. Examples include ATM, ATR and Ku70/80. The presence of accessory factors is thought to come about through recruitment by the shelterin proteins TRF2/Rap1,^17,18 TRF1¹⁹ and POT1.²⁰ The necessity for accessory factors supports the idea that the core six proteins do not handle telomere integrity by themselves. We have argued that LIM-domain proteins TRIP6 and LPP constitute a novel class of telomere-associated factors recruited by POT1.²⁰ The two molecules were found to be in a complex with shelterin, associate with telomeres by ChIP and ensure effective telomere protection by repressing the DNA damage response at telomeres.

LIM domains are found in classes of proteins involved in cell adhesion, differentiation, transcriptional regulation and growth control. They are characterized by a highly conserved motif of CX₂ CX_16–23HX₂CX₂CX₂CX_16–21CX₂(C/H/D) forming two loops of exposed residues, which define protein interaction surfaces.²¹ It has been proposed that, in at least some cases, adjacent LIM domains perform a scaffolding role by bringing together separate functional entities, mediating a variety of functions in cell adhesion, motility or transcription, among others.^21,22 LIM domain proteins Zyxin, TRIP6 and LPP are members of Group 3 LIM proteins, which are characterized by three to four tandem LIM domains located at the C terminus and a Proline-rich pre-LIM domain.²³ The highest degree of similarity is found within the LIM domain region (approximately 60–70%) while the Proline-rich N terminus exhibits a much lower degree of similarity. Group 3 LIM proteins are mostly cytosolic and have been previously demonstrated to interact with cell surface proteins, cytoskeletal proteins or other LIM proteins. However, members of the Zyxin family, and in particular TRIP6 and LPP, are known to shuttle actively between the nucleus and cytoplasm, by virtue of a nuclear export sequence present in the pre-LIM domain.²⁴ Here we report new evidence that a fraction of TRIP6 and LPP is detected at a subset of telomeres, whereas Zyxin does not show telomeric association. We propose, in light of recent advances on telomere and overhang processing, that TRIP6 and LPP participate in the protection of short telomeres and lower the critical point at which telomeres signal into senescence.

Results and Discussion

In order to address specificity of the telomere association observed for TRIP6 and LPP, we chose to investigate whether Zyxin was in a complex with shelterin components and whether it was associated with telomeres. The choice of Zyxin was justified by the fact that it is the next closest related molecule among the LIM domain proteins: the e value between TRIP6 and LPP is at or above 6e⁻¹⁰⁴, and that between Zyxin with TRIP6 or LPP is at or below 3e⁻⁶⁸ (values obtained from the NCBI blast engine). Although the degree of homology with Zyxin is very high and is accounted for by similarities in the LIM domains, it is significantly lower that that observed between TRIP6 and LPP.

A co-immunoprecipitation assay was done in order to determine if Zyxin was in a complex with TRF1 (Fig. 1A). To this end, we cloned the Zyxin cDNA, commercially obtained as an expressed sequence tag, into a retroviral vector, in frame with the MYC epitope. HTC75 cells stably expressing MYC-Zyxin were passaged, along with the MYC-TRIP6 or MYC-LPP-expressing cells. Whole lysates for the three lines, in addition to control cells, were used for immunoprecipitations with TRF1, TRF2, TIN2 or POT1 antibodies. As previously shown, TRIP6 and LPP could be precipitated by TRF1 antibodies (Fig. 1A), as well as the other shelterin antibodies (not shown). In this assay, MYC-Zyxin was not precipitated by TRF1 antibodies (Fig. 1A), or by any other shelterin antibodies used (not shown), arguing for a lack of association between Zyxin and shelterin.

Figure 1.

(A) LIM Proteins TRIP6 and LPP co-immunoprecipitate with shelterin components, but not Zyxin. IP-western blots on lysates made from HTC75 cells stably expressing MYC-TRIP6, MYC-LPP or MYC-Zyxin, as indicated on top. Lysates were used for immunoprecipitations with anti-TRF1 rabbit antibodies, and analyzed for the amounts of each respective protein by western blot with the 9E10 antibody. PI: Preimmune rabbit serum. The Total fraction was ran alongside as indicated. TRIP6: 50 kD; LPP: 70 kD; Zyxin: 60 kD. (B) ChIP analysis of TRIP6, LPP and Zyxin. Lysates prepared from HTC75 cell lines indicated on the left, with the antibodies used listed on top. The total DNA fraction is on the right side of each blot as indicated. Hybridization with a TTAGGG probe shown on top, and with Alu repeats, as a control, at the bottom. Rabbit TRIP6 and LPP (two different sera) antibodies as in reference ²⁰. The Zyxin and PRMT1 antibodies were purchased from Abcam (71842 and 3768). (C) Intranuclear localization of TRIP6, LPP and Zyxin. TRF1 was detected with the 9E10 antibody in the FITC channel and TRIP6, LPP and Zyxin with rabbit antibodies as in (B), in the TRITC channel.

Next, we determined whether Zyxin could associate with telomeres through the ChIP assay. This assay has allowed us to demonstrate telomere association for TRIP6 and LPP.²⁰ Results for Zyxin revealed that it was not found at telomeres in significant levels (Fig. 1B). The co-IP and ChIP results therefore show that not all LIM proteins of group 3 behave the same way and identify a specificity in the telomeric interaction of these proteins. Zyxin would not be expected to play a role at telomeres as TRIP6 and LPP do.

Our published results led to the hypothesis that TRIP6 and LPP recruit a methyl transferase to telomeres, as the related group 3 LIM protein Ajuba does to RARE promoters.²⁵ This association was also suggested by the interesting finding that TRF2 is methylated at specific Arginine residues, by the methyl transferase PRMT1, to mediate telomere protection.²⁶ Therefore, we asked whether PRMT1 could be detected at telomeres by ChIP (Fig. 1B). Our analysis revealed that PRMT1 is present at telomeres, to a degree comparable with that seen for TRIP6 and LPP. Only background counts were detected with the Alu probe, arguing for specific association with telomeres and not along the length of chromosomes. The mode of recruitment of PRMT1 to telomeres, in particular through an interaction with TRIP6 and LPP, is an exciting area to explore.

Immunofluorescence analysis of group 3 LIM domain proteins demonstrates that they shuttle actively between the cytoplasm and nucleus, with predominant staining at steady state in the cytoplasm.²⁷ This dynamic trafficking occurs by virtue of a nuclear export sequence found in the N-terminal pre-LIM domain. Since we have previously shown by ChIP that TRIP6 and LPP are telomere-associated, we asked whether some of the nuclear pool, albeit low, could be detected at telomeres by immunofluorescence. To this end, Hela (1.2.11) cells stably expressing MYC-TRF1 were fixed and subsequently incubated with rabbit-LIM domain proteins TRIP6, LPP and Zyxin in conjunction with the anti-MYC antibody, the last used to detect TRF1 at telomeres. While the presence of these LIM proteins is evidenced by cytoplasmic staining, as reported, there is also a punctate staining seen within the nucleus (Fig. 1C). For TRIP6, we found 32% (±5) of nuclei with 2 or more foci co-localizing with telomeres, with 26% (±4.6) of nuclei without detectable co-localization. For LPP, we found 31% (±11) of nuclei with 2 or more co-localized foci and 45% (±12) of nuclei without telomeric LPP. For Zyxin, although we also observed some degree of punctate nuclear staining, we counted 83% of nuclei without co-localized foci and 19% of nuclei had one focus co-localizing with TRF1 and none with a higher degree of co-localization. We conclude from this data that TRIP6 and LPP were robustly and consistently detected to a small subset of telomeres, whereas Zyxin was not.

One possible interpretation, which requires experimental confirmation, is that TRIP6 and LPP associate with a small subset of telomeres that could be the ones critically short, in a highly regulated manner, for instance specifically during S phase, which could explain the incidence of nuclei with little or no co-localization (1 focus or less). We are currently exploring the localization for TRIP6 and LPP throughout the cell cycle. In contrast, we could find no evidence of telomeric localization for Zyxin, in accordance with our co-IP and ChIP data (Fig. 1A and B).

Overall, our data show a specificity in telomere association for group 3 LIM proteins: TRIP6 and LPP were found at telomeres through various assays, but the next closest member, Zyxin, was not. Moreover, combining our ChIP and IF data, we speculate that a small subset of telomeres are quantitatively bound by TRIP6 and LPP, which would explain why only 2–3 telomeres in a subset of nuclei can generate a significant ChIP signal. We imagine that signaling emanating from critically short telomeres lead to efficient recruitment of TRIP6, LPP and perhaps PRMT1, for protection (Fig. 2). This protection would require the Arginine methylase activity of PRMT1 on TRF2, which was demonstrated to be important in vivo,²⁶ or on other telomeric proteins. Furthermore, the importance of the PRMT1 pathway is underscored by the observation that its depletion by shRNA in primary cells leads to immediate induction of senescence. Given the roles of TRIP6, LPP and PRMT1 in telomere protection, we speculate that they represent a pathway of protection of short telomeres and as such, constitute important factors in determining the senescence point in primary cells. Recent advances have increased our understanding of the interplay between overhang binding factors in regulating telomere replication and preventing telomeres from inducing the DNA damage response in S phase. In particular, it is now demonstrated that RPA, the non-specific single stranded DNA binding protein essential for DNA replication, recombination and repair, is an effective competitor for POT1 on the telomeric overhang.²⁸ In vitro, the affinity of RPA for single stranded TTA GGG sequences is higher than that of POT1. In early S phase, at a time of active telomere replication, RPA is able to compete out POT1 for overhang binding. Later in S phase, RPA is then selectively removed from the overhang by hnRNPA1, allowing POT1 to bind the overhang and inhibit ATR. As telomeres progressively shorten, this interplay between RPA (replication) and POT1 (protection) becomes increasingly biased towards favoring RPA binding, owing to the lower amounts of shelterin, and therefore POT1, in cis. We speculate that it is the ratio of POT1 to RPA that is important at each telomere, for their ability to induce the senescence pathway (Fig. 2). A low ratio, experienced on short telomeres, would result in ATR activation, itself triggered by RPA binding to the telomeric overhang. Although this remains to be established, we find it tempting to propose that the relationship between POT1 and RPA is under the control of pathways able to either delay the senescence point by preventing early ATR activation, or by influencing the binding affinity of POT1 on short telomeres. In this regard, it is intriguing that the association between POT1 and TRIP6 occurs through the POT1 N-terminus,²⁰ which contains the overhang binding domain.²⁹ In vitro studies on the interaction between POT1, TRIP6 or LPP and RPA on the overhang in vitro might be informative in this regard.

Figure 2.

Model for the possible roles of TRIP6 and LPP in telomere protection. See text.

Another interesting question is: why are TRIP6 and LPP recruited only to specific telomeres and not all telomeres? Again, if we retain our hypothesis that they mark particularly short telomeres, one explanation could be that TRIP6 and LPP are recruited through interactions with POT1 or TRF2, which we have been able to detect, but inhibited by other shelterin components, for instance TIN2 or TPP1. As a result, TRIP6 and LPP would not be effectively recruited on long telomeres, but could quantitatively associate with telomeres with less shelterin, the short ones, with lower inhibitory factors. We are currently exploring this possibility by testing the effect of overexpression of individual shelterin components on the association of TRIP6 and LPP with telomeres.

Our current model suggests a role for LIM proteins TRIP6 and LPP at telomeres through the influence in shelterin proteins POT1 and TRF2 (Fig. 2). Both are associated with DNA damage response pathways and have been shown to be in complex with both LIM proteins. The possibility of a direct interaction between these LIM proteins and shelterin warrants further investigation but the idea of a subset of group 3 LIM proteins having a regulatory effect on methyltransferases leads to testable models on the induction of senescence, as well as the repression of the DNA damage response, principally in human primary cells.

Abbreviations

ChIP

chromatin immunoprecipitation

TIF

telomere dysfunction induced foci

co-IP

co-immunoprecipitation

IF

immunofluorescence