N. castellii ALT cells maintain shortened telomeres throughout extensive passaging

N. castellii telomerase knockout cells effectively activate the Alternative Lengthening of Telomeres (ALT) pathway to maintain telomere homeostasis via recombination, which provides the solution for sustained long-term growth (Cohn et al. 2019). To investigate the maintenance of telomeres in ALT cells, we disrupted the telomerase RNA gene (tlc1−) and performed a serial streaking assay (Fig. 1a) (Cohn et al. 2019). Telomere length and structure was analyzed in a terminal restriction fragment (TRF) assay, cleaving the genomic DNA with the restriction enzyme HinfI (Fig. 1c-d). In the N. castellii wild-type (WT) strain, the main part of the terminal DNA fragments exhibits a smear with a mean length of ∼600 bp when hybridized with a telomeric sequence probe (Fig. 1c). During passaging of the telomerase knockout cells, the telomere smear is progressively decreasing in size within the first streaks, until reaching a size of ∼350 bp. The shortened telomeres in the ALT cells show a narrower and more distinct band in the TRF assay, compared to the WT smear. This short and distinct band would generally form at streak 3–4 (S3-S4) corresponding to ca. 60–80 generations from the telomerase disruption (∼20 generations per streak). Remarkably, we observed that the distinct ∼350 bp telomeric band is retained in the continued serial streaking of the ALT colonies, indicating maintenance of a short and stable structure. Although this is a general feature, occasional telomere lengthening events occur, as visualized by the distinct ladder pattern of bands with regular increments (Fig. 1c, S9). These elongations are due to stochastic additions of long arrays of repeats consisting of the subtelomeric TelKO element and telomeric sequences, characteristic of the ALT recombinational telomere elongation events in N. castellii (Cohn et al. 2019).

Intriguingly, the ladder patterns sometimes appear only in a few separate streaks of the assay, leaving no trace in the subsequent passaging. Thus, several of the streaking assays show only the shortened structure, where the telomeres are presumably stabilized without any addition of repeat arrays, seemingly containing just the wild-type subtelomeric sequence and a residual ∼50 bp telomeric sequence (Cohn et al. 2019). This rather contradictory finding piqued our interest to further investigate the ALT telomeres in greater detail to determine the dynamics of their structure.

Fig. 1

Short telomeres are stably maintained in N. castellii ALT cells. (a) Experimental overview of the serial streaking assay of yeast colonies and telomere length analysis. After spore germination on the S1 agar plate, colonies were successively re-streaked, S2-Sn. (b) Schematic overview of the subtelomeric regions in N. castellii containing two variants of the TelKO element; TelKO445 and TelKO220, respectively. In the WT strains the subtelomeric TelKO element is located just internally of the ∼320 bp telomeric sequence in a subset of the 20 chromosome ends. In the ALT strains, the telomeric sequence is shortened to ∼50 bp. The locations of the two subtelomeric hybridization probes are indicated with black bars above the TelKO445 element, the left one is a universal probe and the right one is specific for TelKO445. Subtel-F3, forward primer used in the Telomere-PCR. (c, d) Terminal restriction fragment (TRF) assay of telomerase-deficient strains in serial re-streaking procedures. (c) Haploid wild-type (WT) parental strain (YMC48) and ALT strain (YMC481, tlc1−), streaks S1-S11. (d) Diploid WT parental strain (Y235) and ALT strain (YMC133, tlc1−/ tlc1−), streaks S3-S16. Genomic DNA was digested with HinfI, separated on a 0.8% agarose gel and analyzed by a Southern blot hybridization with a 16-mer telomeric sequence probe

Different TelKO element variants can stabilize and maintain the ALT telomere structure

Interestingly, in some of the serial streaking assays of the telomerase-deficient strains, we discovered a different TRF band pattern, where the telomeric bands were drastically decreased in size to ∼200 bp and showed a much lower hybridization signal (Fig. 1d, S5-S16). Notably, once the new band profile appeared in a streak, it was retained in the following passages, hence indicating that the chromosome end is stabilized by a terminal structure different from the previously characterized TelKO element.

To decipher the terminal DNA sequence present in the different ALT cell lines, we performed Telomere-PCR, followed by gel extraction and sequencing of the amplicon bands. Genomic DNA was C-tailed, and Telomere-PCR performed using a forward primer directed towards the subtelomeric region internally of the TelKO element and a reverse primer annealing to the C-tail. In this way, we recovered two different variants of the TelKO element; a longer 445 bp variant (TelKO445) and a shorter 220 bp variant (TelKO220), which is a truncated version of the longer one (Fig. 1b). The short TelKO220 variant recovered here is identical in sequence to the previously described element (Cohn et al. 2019). The newly discovered TelKO445 element variant contains a 225 bp extension and a HinfI recognition site near its terminus (Suppl. Fig. S1a), which explains the very short telomeric band observed in the TRF assay (Fig. 1d, S5-S16). Consistent with this, re-hybridization of the membrane in Fig. 1d with a subtelomeric probe revealed a band corresponding to the internal HinfI fragment, thus verifying the structure of the TelKO445 element (Suppl. Fig. S1b).

Thus, we have unveiled two variants of the subtelomeric TelKO element, identical in their overlapping regions, which are both able to maintain ALT telomeres. Interestingly, the serial streaking in Fig. 1d displays a marked and clear shift of the TRF band pattern, from the ∼350 bp to the ∼200 bp telomeric band (Fig. 1d, S3 and S5), indicating that the TelKO220 variant is replaced by the TelKO445 variant during the passaging. Notably, the streak in between these samples shows a ladder pattern (Fig. 1d, S4). This makes it tempting to speculate that telomere shortening in the founder cell of this S4 population triggered widespread telomere lengthening activities, eventually leading to the recombinational replacement of TelKO elements in most telomeres in S5.

The TelKO elements are located at the chromosome termini

To verify that the TelKO elements are terminally located, we performed a BAL-31 assay. The BAL-31 exonuclease digests the genomic DNA from the ends and inwards. Wild-type genomic DNA (YMC48) was treated with the exonuclease BAL-31 for increasing amounts of time, and then the time point samples were cleaved with HinfI to release the terminal fragments, followed by separation on an agarose gel and Southern blot transfer. For the hybridization, we firstly used the telomeric probe, which gives the typical smeary telomeric ∼600 bp signal in the sample at time 0 which is undigested by BAL-31 (Fig. 2a). With increasing BAL-31 digestion time, the band is gradually decreasing in size until finally almost totally disappearing at 60 min. This quick and synchronized disappearance of the total signal shows that the telomeric sequences are residing in the termini of all chromosomes, as previously demonstrated for N. castellii (Cohn et al. 2019).

Next, to investigate the subtelomeric sequences, we stripped and re-hybridized the membrane with a probe targeting the common region of the TelKO elements; universal probe (Fig. 2b). In the undigested sample (time 0), the subtelomeric probe produces a narrower signal which is overlapping with the telomeric signal retrieved in the first hybridization. The bands decrease with a similar time course profile as seen with the telomeric sequences, thus confirming the terminal location of the TelKO elements. However, this probe revealed two bands with different behaviors in the time course. While the smeary upper band shows a marked decrease in size already from 30 min, the lower more distinct band (indicated by arrowhead) is not affected until at 75 min. This result presumably depends on the presence of different TelKO variants in the respective bands. The TelKO220 element is fully located within the terminal HinfI fragment and is therefore predicted to show a smeary signal that is immediately affected by the BAL-31 enzyme. In contrast, the additional HinfI site in the TelKO445 element produces an internal ∼470 bp band, which is unaffected until BAL-31 digestion extends beyond this distal HinfI site. To validate this notion, we re-hybridized the membrane with a probe specific for the TelKO445 element (Fig. 2c). As expected, this probe detected only the narrow lower band observed with the universal subtelomere probe, which remained unaffected until the final time point. In conclusion, the BAL-31 results confirm the presence and terminal location of the different TelKO element variants on the chromosomes. To investigate whether the TelKO elements remain terminally located in the ALT cells we performed the BAL-31 assay on a strain having the characteristic short telomere structure (Suppl. Fig. S2). The telomeric sequence hybridization signal showed a similar disappearance in ALT and WT samples (Fig. 2, Suppl. Fig. S2a). Furthermore, re-hybridization of the membrane with the universal subtelomere probe shows that the BAL-31 cleavage affects the TelKO sequences with just a minor delay in the time points (Suppl. Fig. S2b). This result shows that the TelKO elements remain terminally localized on the chromosomes in the ALT cells.

Fig. 2

BAL-31 assays show the presence of the two variants of the TelKO element at the DNA ends. Genomic DNA of the WT strain (YMC48) was cleaved with BAL-31 for increasing amount of time (0–75 min as indicated at the top), followed by the TRF assay procedure using HinfI digestion. The membrane was first hybridized with the telomeric sequence and subsequently re-hybridized twice with different subtelomeric sequences as hybridization probes (probe indicated under the respective image); (a) telomeric probe, (b) universal subtelomere probe; a sequence shared by both TelKO element variants, (c) TelKO445-specific probe; a sequence specific for the TelKO445 variant. See Fig. 1 for the positions of the respective subtelomere probes

Screening populations of colonies reveals a stable and homogeneous telomere structure

To investigate the dynamics of the ALT telomeres, we performed TRF assays on several independent colonies of specific streaks. Analysis of 10 independent colonies from a diploid ALT strain (tlc1−/tlc1−) from streak 3 (S3), showed that they all exhibit a distinct and prominent telomeric band signal at ∼350 bp, characteristic of the presence of the subtelomeric TelKO220 element (Fig. 3a). This result indicates that the terminal structure established in the ALT cells of the parental colony is stably kept in the population of cells that grow into the S3 colonies. The conformity of the bands and the absence of major deviations indicate that an invariant shortened telomere structure is effectively stabilizing the telomeres. Furthermore, for the successive streaks of this strain the same result was obtained, with the predominant telomere signal appearing in the same distinct ∼350 bp band for colonies from S4, S5, S7 and S10, (Fig. 3b-c, Suppl. Fig. S3a). A similar result was obtained for a haploid ALT strain (tlc1−), which gave the same ∼350 bp band in 10 separate colonies of the streaks S3, S7 and S13 (Suppl. Fig. S3b-c). To investigate the subtelomere structure, we stripped and re-hybridized the TRF membranes with the universal subtelomeric TelKO probe, which revealed the expected distinct band, indicating a homogeneous structure (Fig. 3d-f).

The characteristic ALT ladder pattern appears in some of the samples, thus suggesting telomere lengthening with the TelKO element. In both the diploid and haploid strains, only sporadic ladder patterns were observed, exhibiting a similar frequency. Thus, we can conclude that diploid and haploid strains show a conformity regarding the maintenance of ALT telomere structures. Notably, after longer exposures of the TRF membranes, a very faint ladder pattern can sometimes be discerned in some of the other samples (Suppl. Fig. S3d). This implies that array lengthening is occurring at a low frequency, in a small number of cells in the liquid culture.

In the same way, we performed a population screen of 10 colonies from each of two different streaks from which we discovered the new TelKO445 variant. Consistent with the results above, all analyzed colonies within these streaks showed an identical band pattern, hence indicating a stable telomere structure also in telomeres containing the TelKO445 variant (Suppl. Fig. S3e). To verify the subtelomeric structure, we re-hybridized with the universal subtelomeric TelKO probe, which showed a homogeneous structure with the expected internal HinfI band in all samples (Suppl. Fig. S3f). Subsequent rehybridization of the membrane with a probe specific for TelKO445 showed a comparable signal to the universal probe, verifying that the ALT cells in all these colonies indeed contain an abundant amount of this TelKO variant (Suppl. Fig. S3g).

Taken together, our results show the presence of a stable shortened telomere structure in the ALT cells. Notably, the short telomere structure that is established in the early stages of the passaging is stably kept throughout extensive numbers of successive streaks and is the predominant structure found in the population of cells. The stochastic appearance of the array ladder pattern, and the fact that ladder pattern samples also retain the short ∼350 bp band, indicates that the dramatic elongation with the TelKO element is a rather rare event that is not taking place in all cells in a population, and it may not take place in all telomeres within the cell.

Fig. 3

Screening of large populations of colonies indicate a stable and homogeneous telomere structure in the ALT cells. TRF assays were performed on ten separate colonies isolated from each of the indicated streaks (S3, S4, S5, S7) in the serial streaking procedure of an ALT strain (YMC133, tlc1−/ tlc1−). The membranes were first hybridized with a telomere probe (a-c) and then re-hybridized with a subtelomere probe detecting the TelKO element (d-f)

The TelKO elements exhibit a homogeneous sequence in the ALT strains

We further wanted to analyze the terminal DNA sequences of the isolated colonies from the population screen. To this end, we performed Telomere-PCR on the genomic DNA from the 10 separate colonies of the respective streaks. As expected, the colonies within each streak showed a conformity, generating the same size of the PCR amplicon band (Fig. 4, Suppl. Fig. S4). ALT strains containing the TelKO445 element versus the TelKO220 element generate a ∼900 bp and ∼700 bp amplicon band, respectively. Amplicons were extracted and sequenced from three colonies of each streak. Sequence analysis revealed identical TelKO elements among colonies within each respective streak (Cohn et al. 2019). Moreover, both the TelKO445 and TelKO220 elements show identical sequences even when isolated from separate streaks (in total 10 streaks analyzed). Thus, we conclude that the TelKO elements are kept highly homogeneous throughout extensive numbers of generations in the serial streaking.

This result is further underlined by re-hybridization of the TRF assay membranes with a probe specific for the TelKO445 element, detecting the second extended half of the element. Samples containing the TelKO445 element showed very strong and prominent signals in all samples from the streak, indicating an amplification of the TelKO445 elements (Suppl. Fig. S3g, S13). In contrast, colonies from a streak from a different serial line known to contain the TelKO220 element showed a weaker signal than WT, thus indicating a reduction of TelKO445 elements and confirming that the TelKO220 element is predominantly present in these samples (Suppl. Fig. S3g, S7). This remarkable result is indicative of an effective spreading of a specific element in the respective ALT strains, leading to homogenization of the sequences in all subtelomeres.

Fig. 4

ALT cells maintain a homogeneous terminal structure within the respective streak population. Telomere-PCR analysis was performed on genomic DNA from ten colonies of the respective streak. (a-c, f) An amplicon band of ∼700 bp, indicative of the TelKO220 variant, is generated in ALT strains YMC133 (tlc1−/ tlc1−) streaking assay #1 (S3, S7, S10) and YMC481 (S3). (d-e) An amplicon band of ∼900 bp, indicative of the TelKO445 variant, is generated in streaking assay #2 of YMC133

Subtelomeres are undergoing sequence homogenization in the establishment of ALT cells

To get more insight into the subtelomere structure in different chromosome ends, we performed PacBio whole genome sequencing of a WT N. castellii strain (YMC48) and assembled chromosome contigs using TeloClip and Canu. We could locate telomeric sequences and retrieve the adjacent subtelomeric regions in 18 of the 20 chromosome ends. Consistent with the BAL-31 results, we confirmed that TelKO elements are exclusively located in the subtelomeres, present as single copies just internal to the telomeric sequence repeats. Hence, the two variants, TelKO445 and TelKO220, are localized to separate chromosome ends. Moreover, as previously observed in TRF assays, some chromosome ends are lacking TelKO elements (7/18), and some contain partial TelKO445 elements that are extending beyond TelKO220 to various extent (Cohn et al. 2019). Thus, regarding the TelKO elements, the subtelomeres show a quite variable content in the WT strain.

In the same way, we performed PacBio sequencing of an established ALT strain (YMC481, S5). Remarkably, the sequence data shows that all the chromosome ends in this ALT strain have acquired the same TelKO variant; the short TelKO220, thus confirming the subtelomere homogenization observed in the above TRF assays. This means that even telomeres lacking a TelKO element in the WT strain acquired one in the ALT strain. Moreover, it means that a TelKO variant switch has occurred, where the TelKO445 variant present in the WT telomere was replaced with the TelKO220 variant in the ALT strain. Detailed analysis further revealed that the WT strain contains two different alleles of the TelKO220 sequence, differing by a single nucleotide at position 28 (T or C) (Suppl. Fig. S1a). Remarkably, only one of these alleles (C) is present in all telomeres of the ALT strain. Hence, together these observations further substantiate the homogenization of telomeres in the ALT strains.

In the contig assembly of the ALT strain, we observed long elongated arrays of TelKO elements in some of the telomeres, which is correlating to the ladder pattern observed in the TRF assays. Notably, the TelKO220 elements in these arrays are all the same allele (C). A closer analysis of the arrays revealed a 325 bp repeated unit with a composite structure, where the TelKO220 element is flanked by telomeric sequences and additional 13 bp partial TelKO segments that are identical to the end of the TelKO220 element (Fig. 5, Suppl. Fig. S5). Thus, the 325 bp unit forms a higher-order repeat, containing smaller repeated units. The 13 bp terminus of TelKO220 is present in triplicates within this higher-order repeat, separated by identical copies of a 35 bp telomeric sequence stretch (Fig. 5, Suppl. Fig. S5). It is noteworthy, that this 325 bp higher-order repeat unit is completely identical within all the arrays of the different chromosome ends, which further explains the uniform increments in the TRF ladder pattern.

Fig. 5

The subtelomeres of ALT cell chromosomes contain arrays of composite higher-order repeats. Schematic overview of the repeated unit found in arrays at all chromosome ends of the sequenced ALT strain (YMC481). The 325 bp higher-order repeat contains the TelKO220 variant, flanked internally by a short 9 bp interstitial telomeric sequence (ITS, green box) and distally by a 96 bp element containing two identical 35 bp segments of telomeric sequence (green) and two 13 bp segments identical to the end of the TelKO element (yellow). The number of higher-order repeats vary between 2–27 (denoted by n). The sequence at the extreme terminus has the same structure, followed by a terminal telomeric sequence that is shorter than in WT cells

However, intriguingly, we have observed samples exhibiting ladder patterns with different size increments, implying that the arrays in those strains consist of differently sized repeated elements (Fig. 6). Significantly, the band increments show a consistency within each sample, underlining the notion that the same specific element is being added onto the telomeres within the respective separate populations of established ALT cells.

In conclusion, the summarized results from the genome sequencing, TRF assays and Telomere-PCR analyses show a convergent sequence in all the subtelomeres of the ALT cell chromosomes, where one specific TelKO element variant is predominantly spread onto all telomeres in each separate ALT strain establishment. Hence, the TelKO element is a key for the homogenization and stabilization of the shortened telomeres.

Fig. 6

Differences in the ladder pattern increments between strains indicate that differently sized repeat units are used for the elongations. TRF assay of three separate ALT strains show differences in band patterns. WT: YMC48 (left) and Y235 (right). ALT strain #1; YMC482 (S3), #2–3; YMC133 (S4 and S10). The membrane was hybridized successively with a telomere (left) versus a subtelomere probe (right)

The telomere protein Rap1 can bind to TelKO elements, suggesting formation of protective chromatin caps on the short ALT telomeres

Canonical telomeric sequences show a characteristic strand bias, with a TG-rich strand running 5’-3’ towards the end. Interestingly, the sequence of the TelKO element also shows a slight TG-rich strand bias. The TelKO445 has an overall [T + G] content of 75%, with its first part, corresponding to the TelKO220 variant, having an elevated [T + G] content of 82%. This is emphasized by the presence of several short Interstitial Telomeric Sequences (ITS) that are scattered along the element, mostly only partial stretches of the N. castellii telomeric sequence (TCTGGGTG). Since the telomere binding protein Rap1 has previously been demonstrated to show a spatial flexibility and is able to bind to different variant sequences, we reasoned that it might accommodate binding to the TG-rich sequences in the TelKO element (Wahlin and Cohn 2000, 2002). To learn more about the functionality of the TelKO element, we therefore decided to analyze whether Rap1 would be able to bind to the subtelomeric TG-rich stretches in vitro. We designed ds-oligonucleotides distributed along the TelKO region (ST-1 – ST-4) (Fig. 7a). ST-1 is centered around a 9 bp ITS that marks the border of the TelKO element and ST-3 covers an 8 bp ITS close to the end of the TelKO220 variant. Although ST-2 only contains a 5 bp ITS, it is overall highly TG-rich, and ST-4, specific for the TelKO445 variant, incorporates a 6 bp ITS surrounded by a highly TG-rich sequence.

In an Electrophoretic Mobility Shift Assay (EMSA), Rap1 produced the expected shifted DNA-Rap1 complexes, showing that Rap1 can bind to the subtelomeric oligonucleotides, although with a quite low affinity compared to the control oligonucleotide containing a fully telomeric sequence (data not shown). To quantitatively assess the binding capacity to the different sequences, we performed EMSA competition assays (Fig. 7b). Here, the binding of Rap1 to the radiolabeled telomeric oligonucleotide was challenged by adding non-labelled ST oligonucleotide competitors in large molar excess. Addition of the non-labelled telomeric sequence effectively competes away the shifted band, leading to an increase of the free DNA already at 40x molar excess (Fig. 7b, lanes 3–4). The non-labelled ST oligonucleotides also exhibit competition, but a higher molar excess is needed (400x-4000x), indicating a much lower affinity of Rap1 to all these subtelomeric sequences (lanes 5–12). Quantitation of the shifted bands at 400x subtelomeric competitor, shows around 30–90 times higher signal left in the shifted band when compared to the telomeric sequence (Table S1). Notably, ST-4 has the highest affinity among the subtelomeric sequences tested, despite containing only a 6 bp ITS. This result opens for the possibility that other TG-rich stretches in the TelKO element could be possible targets for Rap1 binding.

Intriguingly, the presence of Rap1 binding sites within the TelKO element suggests that the subtelomeric region might be incorporated within the telomeric chromatin in the ALT cells. Although we cannot exclude the possibility that the shortened telomeres may be stable independently in N. castellii, accumulated data in various species show that short telomeres are highly unstable. We therefore hypothesize that the low affinity binding sites may become exposed and accessible for Rap1 when the telomeres shorten, thereby providing a protective cap structure and stabilizing the chromosome ends.

Fig. 7

The telomere protein Rap1 binds with low affinity to the TelKO element. Competition Electrophoretic Mobility Shift Assay (EMSA) of Rap1 incubated with a labeled telomeric probe and increasing amounts of the non-labeled oligonucleotide competitors ST-1-4. (a) Schematic of the subtelomeric region with indicated positions and sequences of the subtelomeric regions tested. Interstitial telomeric sequences are indicated in bold. (b) EMSA gel where the Rap1 binding to the labeled telomeric probe was competed by the addition of non-labeled subtelomeric competitors added in 400 and 4000 times molar excess (lanes 5–12). As a control, the non-labeled telomeric probe was added in 40 and 400 times excess (lanes 3–4). Lanes 1 and 14, no protein added. Lanes 2 and 13, no competitor added. The positions of the free DNA versus the DNA-protein complex are indicated to the right