Different telomere-length dynamics at the inner cell mass versus established embryonic stem (ES) cells

Results

To directly address these possibilities, we first analyzed telomere length at different stages of mouse embryonic and fetal development, including morula, blastocyst, E7.5, E10.5, and E13.5 (Materials and Methods). Embryo sections were hybridized with a telomeric probe and telomere length was measured at a single-cell level by using the telomapping technique (6) (Materials and Methods). We observed that average telomere length significantly increased from morula to the blastocyst stage (Fig. 1A) and that, although average telomere length was shorter at E7.5 compared with the blastocyst stage, it was maintained constant from E7.5 until E13.5, in agreement with the presence of high telomerase activity throughout embryo development (8, 17–21). Strikingly, ES cells processed in parallel showed much longer telomeres than those of blastocyst cells (Fig. 1A). To discard that differences in telomere length are caused by changes in probe accessibility, chromatin status associated to developmental stage, or ploidy, we performed quantitative-FISH (Q-FISH) with a centromeric major satellite probe and found no significant differences in centromeric fluorescence (Materials and Methods and Fig. S1). In this regard, centromeres and telomeres have been reported to share the same heterochromatic marks (22). We next performed a separate analysis of telomere length in trophectoderm (TE) cells versus ICM cells within the same blastocysts by using telomapping. Blastocysts cells were grouped into three categories according to their average telomere fluorescence intensity and a color was associated to each group (Fig. 1B, Top). Most of the cells with the longest telomeres (red color) localize to the ICM, and only a few to the trophectoderm (Fig. 1B), and the mean telomere length for the ICM was significantly higher compared with the TE (Fig. 1B, Bottom). Notably, telomeres of ICM cells were shorter than those of established ES-cell lines, suggesting that ES-cell telomeres undergo a significant lengthening during ES-cell in vitro expansion, in analogy to that previously reported for iPS cells (5). To test this finding, we analyzed in-parallel telomere length in blastocysts and two independent ES-cell lines at both early and late passages by telomapping. Mean telomere length of the ICM was significantly higher than that of the MEFs and trophectoderm cells and of a similar length to early passage ES cells (passage 5) (89 and 83 Kb, respectively) (Fig. 1C). Telomere length further increased from passage 5 (83 kb) to passage 12 (around 125 kb) (Fig. 1C). In addition, the increased recombination rates of ES cells compared with MEFs (23) could account for the increased heterogeneity in telomere length found in increasing passages of ES cells. By performing Q-FISH with a centromeric major satellite probe, we ruled out that these differences in telomere length were because of changes in probe accessibility, chromatin status, or ploidy (Materials and Methods and Fig. S2). Of relevance, this continuous increment of telomere length over passages in pluripotent cells is not observed in established immortal human (Fig. S3) or mouse cell lines (24), which show stable telomeres over passages. To test whether telomere length was further increased after passage 12, we expanded the cells until passage 31 and performed telomapping analysis. We found that telomeres continued to increase their length at these late passages, although the difference between passage 24 and 31 was not statistically significant (Fig. S4). In conclusion, the reset of telomere length during development happens at the blastocyst stage, in accordance with a previous report showing telomere elongation at the transition from morula to blastocyst (25). Importantly, we first demonstrate here that the longest telomeres within blastocysts localize to the ICM, suggesting that telomere elongation specifically occurs in this subset of pluripotent embryonic cells. In addition, ES cells undergo a further increase in telomere length compared with ICM cells of the blastocyst.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Blastocyst ICM bears the longest telomeres which further lengthen upon expansion of ICM-derived ES cells. (A) Quantification of telomere length by telomapping analysis of embryo sections at the indicated stages of development, ES cells (passage 9) and primary MEFs (passage 2). Telomere-length quantification is given in arbitrary units of fluorescence (a.u.). n = number of embryos or independent cell cultures. (B) (Top) Representative image of a telomapping of a blastocyst section. Nuclei are colored according to their telomere length and normalized by the telomere length of ES cells. Because ES cells are derived from the ICM we reasoned that they should have equivalent telomere length. The division of the CY3 intensity value of each blastocyst cell by the mean CY3 intensity value of ES cells should render the blastocyst cells with the longest telomeres (values around or equal to 1). For the blastocyst map we grouped intensity values in three fractions to simplify the identification of the cells with the longest telomeres. Note that the longest telomeres localize to the ICM of the blastocyst. (Scale bar, 10 μm.) (Lower) Quantification of telomere length of blastocyst, ES cells and MEFs, as indicated. n = number of embryos or independent ES and primary MEF cultures. (C) Telomere-length frequency histograms of blastocysts, ES cells at the specified passages, and primary MEFs and mean telomere length for the same samples. Note that the telomere length of ES at early passages is similar to that found in the ICM. n = number of embryos or independent ES or primary MEFs cultures. (D) Scheme of the process of isolation of ES cells from blastocysts. In brief, zona pellucida is removed from blastocysts and they are transferred to a 60-mm dish. After 4 to 6 d the ICM has divided to ∼1,000 cells. Individual ICM colonies are transferred to a 96-well plate. At this step, ES colonies emerge and are transferred to a 24-well plate for expansion. From the 24-well plate, cells are transferred to 25-mm plates and are considered passage 1. Further passages are plate colonies or ES and primary MEF cultures. (E) Mean telomere length and telomere-length frequency histograms in ICM from the blastocyst, in vitro cultured ICM, emerging ES from the 96-well plate, established ES cells (passages 5, 9, and 12), iPS cells (passage 1 and 29), and primary MEFs determined by telomapping. Note that telomeres of the ICM at the blastocyst are longer than those of the cultured ICM.

The telomere length of the blastocyst ICM was comparable to that of ES cells at passage 5, raising the possibility that telomere length of established ES cells was inherited from the blastocyst ICM, and telomere lengthening restricted to in vitro expansion of established ES cell lines. To test this hypothesis, we sought to analyze in-depth telomere dynamics at the earliest steps during establishment of ES cells. The very first step is the in vitro ICM, obtained from 3.5-d blastocysts upon removal of the zona pellucida. After a few days, colonies of about 1,000 cells are formed (scheme in Fig. 1D; images of ICM colonies in Fig. S5; see also Materials and Methods). In vitro ICM colonies are individually trypsinized and transferred to 96-well plates, where ES cells start to emerge. Further expansion leads to the establishment of ES cells (Fig. 1D). We measured telomere length by telomapping in the ICM and trophoblasts in blastocysts, in cultured ICM, in the ICM-derived cells grown in 96-well plates, and in established ES-cell lines at passages 5, 9, and 12 (see scheme in Fig. 1D). We also included iPS cells at both early and late passages. We confirmed that telomeres lengthen during in vitro expansion of ES cells (80 kb at passage 5 compared with 123 kb at passage 12) (Fig. 1E). Similarly, iPS-cell telomeres increased with passages (Fig. 1E) (5). Interestingly, telomeres from the in vitro ICM (55 kb) were shorter than those of the blastocyst ICM (86 kb) but seemed to recover their length at the 96-well plate (89 kb) (Fig. 1E), which showed similar telomeres to early passage (passage 5) ES cells (80 kb). We confirmed these findings by using an independent technique based on Southern blotting (telomere restriction fragment analysis, TRF) (Fig. S6). These results may suggest that the cells from the ICM are susceptible culture-stress–induced telomere-length changes. Indeed, during the establishment of ES-cell lines, the transient ICM of the early blastocyst is forced to artificially exist and divide for several days in vitro. Under culture conditions, most ICM cells differentiate (only 17% and 38.5% of cells express the pluripotency factors Sox2 and Oct3/4, respectively), which in turn may lead to telomere shortening compared with pluripotent stem cells (5, 6).

To better understand the dynamics of telomere lengthening in the cultured ICM, and to avoid contamination with feeder cells (irradiated MEFs), we analyzed telomere length after 4 and 7 d of culture, in the absence of feeders, by using telomapping (Fig. S7). We did not find any statistically significant difference in the telomere length at 4 or 7 d of culture. We also ruled out that mean telomere length of the in vitro cultured ICM was lower than that of the blastocyst ICM because of the contribution of irradiated MEFs.

Next, we set to confirm telomere shortening in the in vitro ICM, as well as telomere lengthening of ES cell over in vitro expansion, by using Q-FISH on metaphase spreads. Metaphase spreads allow analysis of every single telomere at chromosomes of a given metaphase. We confirmed shorter telomeres in the cultivated ICM (50 kb), which increased in length with subsequent passages from a mean telomere length of 112 kb in passage 5 to a mean telomere length of 144 kb in passage 12 (Fig. 2A and Fig. S8A; note that absolute telomere-length values were higher than in the telomapping experiment, most likely because of differences in acquisition and the software used to measure intensity).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Telomere-length dynamics during establishment and expansion of ES cell lines as well as in vivo aggregation of ES cells in morulae. (A) Mean telomere length for ICM cultivated from the 60-mm tissue-culture plate, and successive passages of ES cells. Telomere length was analyzed by metaphase Q-FISH. n = number of ICM colonies or independent ES and primary MEF cultures. (B) Scheme of the aggregation experiments. Established ES cells at passage 16 expressing GFP were microinjected in eight-cell morulae. Blastocyst from injected and noninjected morulae were fixed for the analysis of telomere length by telomapping. (C) Mean telomere length for primary MEFs (passage 2), noninjected and injected blastocysts, as well as GFP-ES cells before injection (passage 16) and ES cells at passage 9. n = number of blastocysts or independent clones of ES cells or primary MEFs. (D) Representative images of an injected blastocyst.

To in vivo test whether established ES cells have longer telomeres than those of the ICM of the blastocyst, we aggregated ES cells with hyper-long telomeres expressing GFP with eight-cell morulae (Fig. 2B and Materials and Methods). At the blastocyst stage, development was stopped and combined telomere FISH/GFP immunofluorescence was performed (Materials and Methods). We found that average telomere length in GFP-expressing ICM cells (derived from aggregated ES cells) was higher than that of non-GFP–expressing ICM cells (derived from the recipient morulae) (Fig. 2 C and D and Fig. S8B). These results demonstrate that established ES cells have longer telomeres than the cells of the blastocyst ICM. In addition, these results rule out possible effects of different developmental stages on telomere-length measurements, as we are comparing the same cell type within the blastocyst ICM. In summary, these findings strongly support the unique finding of active mechanisms, leading to very long telomeres in the process of ES-cell line establishment, which are likely to involve telomere elongation by telomerase (5).

We reasoned that the increase in telomere length observed in established and during the establishment of ES cells could be linked to the structure of chromatin and ultimately to the epigenetic status of telomeres, which is different to that observed in MEFs (5, 22). To test this idea, we first measured the global- and subtelomeric-DNA methylation (SI Materials and Methods). Because pericentric and subtelomeric repeats remain unaltered between ES and differentiated cells (5, 16), we analyzed the interspersed repeats (SINE repeats) and found no substantial difference in DNA-methylation between the passages of ES cells and MEFs (Fig. S9A). We found subtelomeric DNA mostly methylated with small variations between the passages, which were not statistically significant (Fig. S9 B–D). We next analyzed heterochromatic marks at telomeres by performing FISH with a telomere probe combined with immunofluorescence for both trimethylation at lysine 20 of histone H4 (H4k20me3) and at lysine 9 of histone H3 (H3k9me3) (5, 22, 26–29). H4k20me3 average fluorescence was similar in primary MEFs, ICM, and 96-well cells, but very significantly decreased in established ES cells. However, histograms of the frequency of cells with a given H4K20me3 fluorescence already show a population of cells with low H4k20me3 abundance in the cultured ICM and the cells in the 96-well plates. Indeed, the percentage of cells with H4k20me3 fluorescence below 7 arbitrary units increased from MEFs (16%) to the ICM (33.7%) and 96-well plate (36.9%), to reach 83.5% in established ES-cell lines (Fig. 3 A–D). Very similar results were observed for H3k9me3 (Fig. 3 E–H). A lower colocalization of heterochromatic marks with telomeres was also observed during the process of ES cell generation (Fig. 3 C and G) (5). Together, these results indicate a decrease in both H3K9me3 and H4K20me3 heterochromatic marks during the generation of ES cells compared with MEFs, starting in the in vitro ICM. These unprecedented findings suggest that telomere lengthening is concomitant with lower density of heterochromatic marks during the process of ES-cell establishment. Alternatively, only cells with a more open/less-compacted chromatin structure are selected from the blastocyst stage to obtain stable ES-cell cultures.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

The loss of heterochromatic marks accompanies telomere lengthening. (A) Mean H4k20me3 intensity for primary MEFs (passage 2), in vitro cultured ICM, cells from the 96-well plate, and established ES cells at passage 9. (Lower graphs) The H4k20me3 histograms for the same samples. Note that in the ICM as well as in the 96-well plate there are cells with low H4k20me3 signals. (B) Percentage of cells with less than 7 arbitrary units of H4k20me3 fluorescence. Note the portion of cells with low methylation signal in both the cultured ICM and the 96-well plate. (C) Colocalization of the H4k20me3 heterochromatic mark with telomeres in percentage for the samples described in A. (D) Representative images of telomeres and H4k20me3 signals for the samples described in A. (E) Mean H3k9me3 intensity and histograms for the samples described in A. (F) Percentage of cells with less than 7 arbitrary units of H3k9me3 fluorescence. Note the portion of cells with low methylation signal in the cultured ICM and the 96-well plate. (G) Percentage of colocalization of the H3k9me3 heterochromatic mark with telomeres for the samples described in A. (H) Representative images of telomeres and H3k9me3 signals for the samples described in A. n = number of ICM or 96-well plate colonies or independent ES, and primary MEF cultures. Arbitrary units of H4k20me3 fluorescence is plotted. (Scale bars, 10 μm.)

Next, we reasoned that the mechanisms leading to telomere elongation during the establishment of ES-cell lines might be linked to pluripotency (30⇓⇓⇓⇓–35). Indeed, adult stem-cell compartments bear the cells with the longest telomeres in mice (6). As a marker for pluripotency, we first tested Nanog, which is required to maintain pluripotency in the mouse epiblast and ES cells (32, 36). To this end we combined immunofluorescence using a Nanog antibody with FISH for telomeres (Materials and Methods). Again, ICM-cultured cells had shorter telomeres than those from the 96-well plate or established ES cells (Figs. S10 A–C and S11A). Interestingly, Nanog showed very low expression in the cultivated ICM (3% Nanog-positive cells) (Fig. S10B, Lower graph), which was dramatically increased at late-passage ES cells (Fig. S10B). Accordingly, the best positive slope between telomere length and Nanog was found only in established ES cells (Fig. S11B). These results suggest that Nanog expression and telomere length do not correlate during early stages of establishment of ES-cell lines, and this only occurs at later passages.

Several lines of evidence suggest a link between pluripotency and the telomere-binding proteins, known as shelterins (37⇓–39). The shelterin protein TPP1 is essential for telomere elongation by telomerase during reprogramming of MEFs into iPS cells (40). In addition, deletion of TRF1 causes lethality at the blastocyst stage (41), and adult tissues conditionally deleted TRF1, show severe stem-cell defects (40, 42). Thus, we next explored the regulation of TRF1 during establishment of ES cell lines. TRF1 binds and protects telomeres (18, 37, 38) and is proposed to have a role in telomere length regulation (43⇓⇓–46). We observed high TRF1 levels already in the cultured ICM compared with primary MEFs (Figs. S10 D–F and S11D). TRF1 levels were also high in emerging (96-well) and established ES cells, and Nanog showed similar expression to the previous experiment (Fig. S10B). Thus, high levels of TRF1 were associated with high levels of Nanog expression in emerging or established ES cells, but not in the in vitro ICM. We therefore tested whether TRF1 levels in the in vitro ICM associated to other pluripotency markers. Oct3/4 or Sox2 function in the maintenance of pluripotency in early embryos and established ES cells (47⇓–49) and are essential for the reprogramming of differentiated cells into iPS (14⇓–16). To test this possibility, we performed immunofluorescence with TRF1 and Sox2 (Figs. S10 G–I and S12A). Interestingly, the mean intensity value for Sox2 in the cultured ICM was twice higher than in MEFs, and further increased in emerging and established ES cell lines (Figs. S10H and S12A). Similar results were found when TRF1 and Oct3/4 antibodies were used (Fig. 4 A–C and Fig. S12D). Of note, the percentage of positive cells for Sox2 and Oct3/4 in the in vitro ICM (17% and 38.5%, respectively) was higher than that of Nanog (3%). Despite the high levels of TRF1 associated to different pluripotency markers at every stage of establishment of ES cells, correlations were poor (Figs. S11 B and C and S12 B and C). To further study a possible correlation between pluripotency factors and TRF1, we used a mouse antibody against Oct3/4 in combination with our best TRF1 antibody. The mouse cell line L5178Y-R, which bears long telomeres but is not pluripotent, was included in our analysis to discard that association of high levels of TRF1 and pluripotency factors are coincidental. Our results show that the cells from the L5178Y-R line had a higher mean TRF1 intensity than MEFs, bur lower than the cultured ICM (Fig. 4 D, F, and H). Oct3/4 levels were basal in primary MEFs and L5178Y-R (Fig. 4 E, G, and H). Furthermore, we observed a clear correlation between TRF1 and Oct3/4 in established ES cells (Fig. 4I) and in the in vitro ICM in those cells expressing high levels of Oct3/4. Together, these results indicate that high levels of TRF1 occur in the presence of some pluripotency factors (i.e., Oct3/4) from the earliest step of derivation of ES cells. The unprecedented finding of elevated TRF1 levels before telomere elongation (cultured ICM) could represent a previously unnoticed mechanism to enable cells to protect telomeres as they are elongated. In this manner, limiting TRF1 amounts could also limit further telomere elongation by telomerase.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Analysis of TRF1 and Oct4 expression during isolation of ES cells. (A) Mean TRF1 intensity for primary MEFs (passage 2), in vitro cultivated ICM, 96-well plate emerging ES cells, and established ES-cell lines at passage 9 analyzed by telomapping. (B) (Left) Mean Oct3/4 intensity for the same samples described in A. (Right) Percentage of cells with Oct3/4 intensity bigger than 20 a.u. Note that at the cultured ICM stage, a 38% of cells are Oc3/4 positive. (C) Representative images of TRF1 and Oct3/4 expression for the same samples described in A. (Scale bars, 10 μm.) n = number of ICM or 96-well plate colonies or ES and primary MEF cultures. (D) Mean TRF1 intensity values for primary MEFs, the cell line L5178Y-R or R cells, cultured ICM, and ES passage 9. (E) Mean Oct3/4 intensity for the samples described in D. (F) TRF1 expression frequency histograms corresponding for the samples described in D. (G) Oct3/4 expression frequency histograms corresponding to the samples described in D. (H) Representative images of TRF1 and Oct3/4 expression for the samples described in D. (Scale bars, 10 μm.) (I) TRF1 intensity values plotted against Oct3/4 intensity values to analyze correlation. Primary MEFs, cultured ICM, and established ES cells at passage 9 are shown in the Upper panels. (Lower) Cells from the cultivated ICM were divided in high or low TRF1 intensity for the analysis. n = number of ICM colonies or L5178Y-R, ES, primary MEF cultures.