Transposon-mediated telomere destabilization: a driver of genome evolution in the blast fungus

Abstract The fungus Magnaporthe oryzae causes devastating diseases of crops, including rice and wheat, and in various grasses. Strains from ryegrasses have highly unstable chromosome ends that undergo frequent rearrangements, and this has been associated with the presence of retrotransposons (Magnaporthe oryzae Telomeric Retrotransposons—MoTeRs) inserted in the telomeres. The objective of the present study was to determine the mechanisms by which MoTeRs promote telomere instability. Targeted cloning, mapping, and sequencing of parental and novel telomeric restriction fragments (TRFs), along with MinION sequencing of genomic DNA allowed us to document the precise molecular alterations underlying 109 newly-formed TRFs. These included truncations of subterminal rDNA sequences; acquisition of MoTeR insertions by ‘plain’ telomeres; insertion of the MAGGY retrotransposons into MoTeR arrays; MoTeR-independent expansion and contraction of subtelomeric tandem repeats; and a variety of rearrangements initiated through breaks in interstitial telomere tracts that are generated during MoTeR integration. Overall, we estimate that alterations occurred in approximately sixty percent of chromosomes (one in three telomeres) analyzed. Most importantly, we describe an entirely new mechanism by which transposons can promote genomic alterations at exceptionally high frequencies, and in a manner that can promote genome evolution while minimizing collateral damage to overall chromosome architecture and function.

. E 2G4SS4-20 had 79 rDNA-telomere reads; SS6-1 had 10 and SS15-11 had 17. F Novel rDNA-telomere junctions were only "called" if: i) an extended telomere repeat was present at the start (CCCTAA strand) or the end (TTAGGG strand) of the read; and ii) the rDNA blast alignment extended to within 10 bp of the telomere junction. G Telomere position approximated due to unclear telomere boundary in MinION read(s). H Sequence of telomere seed undetermined due to poor sequence quality at boundary.    telomere (TEL1), while the filled arrowhead marks a stable telomere that rarely exhibited rearrangement (TEL5). Novel telomeric restriction fragments that were cloned characterized and described here are labeled "a", "f", and "n." Black dots highlight rearranged rDNA telomere fragments. The asterisk marks a novel TRF that is shared by culture 5, 6, 8 and 9 and which arose through a MoTeR2 insertion in TEL-C -presumably via a transposition event (see Fig. S8 variant ii).  hybridized using the telomere and TLP7 probes. TEL7 variants are labeled i, ii, iii and iv. The asterisk is to highlight a TRF that is believed to correspond to variant iv but a definitive link has not yet been established; B) Structures of the four variants as determined from MinION reads.
All structures were inferred from individual MinION reads that spanned the entire distance from the chromosome unique sequence to the telomere. In most SS cultures, the TLP7 probe hybridized to fragments with a molecular size of ~10.5 kb (bands i and ii). Isolates SSs 12, 13, 17 and 18 had signals at a position corresponding to ~ 16 kb (band iv), SS4 had one slightly larger than 16 kb (band iii), and the 2G1SS1 parent culture had a signal at ~13 kb.
The MinION reads for 2G4 SSs 6-1 and 15-11 revealed that the 10.5 kb TRF -which was also present in the original LpKY97 culture (see Fig. 1) but not 2G1SS1 (this figure, panel A)had a single, full-length MoTeR1 insertion in the telomere with a short interstitial telomere repeat (TTAGGG)2 (i & ii). In contrast, most reads from SS4-20 identified a TEL7 variant with four truncated MoTeR copies distal to the intact element (see panel B.iii). These four elements were duplicates of truncated variants found at the end of TEL-D (Fig. S5). Considering that the parental telomere lacks obvious break-inducing features, we hypothesize that the first three elements were acquired through a multi-step process.
Step 1 likely involved break-induced replication of a composite MoTeR cassette from TEL-D; in step 2, a terminal duplicate of these last three elements was then generated by D-loop formation & extension; step 3 involved terminal resection that eliminated the two distal tMoTeR copies; and, finally, step 4 involved healing of the TTAGGG seed at the tMoTeR2:tMoTeR1 junction. Under this scenario, the slightly shorter variant with just two truncated elements distal to the intact MoTeR1 could be the product of more extensive resection prior to healing (iv). The low abundance of variant (iv) in SS4-20 (and its similar size) is probably why a corresponding band is not visible in the SS4-20 lane but it likely corresponds to band "iv" seen in SS 12, 13, 17 and 18. We were unable to characterize the ~13 kb TRF in LpKY97 but given that it is slightly smaller than variant "iv," it possibly comprises an array with a telomere-healed break at the (TTAGGG)3 repeat.
Summary of inferred rearrangement mechanisms: i -> iii) acquisition of a composite MoTeR cassette comprising the three distal elements in TEL-D (see Fig. S5), followed by acquisition of a truncated MoTeR2 into the newly acquired telomere; iii -> iv) as above, but with resection/attrition and telomere healing as an alternate pathway to the second MoTeR2 acquisition.

Fig. S5. A single MoTeR array with a MAGGY insertion in the LpKY97 parental strain.
The combined MinION assembly identified a single MAGGY insertion in a complex MoTeR array in TEL-D on minichromosome 2. Note: the entire array was captured in a single MinION read thereby confirming the array's structure.         MoTeR1 copy 2 Fig. S15C). Variant 6.iv almost certainly arose from 6.iii via interstitial telomere breakage and healing; and it seems likely that 6.vi was generated from 6.iii, 6.iv or 6.v by the same mechanism. On the other hand, 6.v was probably derived from 6.iii through an intrachromatid or unequal sister chromatid exchange between MoTeR1 copies 3 and 4.  PstI fragment of between ~46 kb and 48 kb and the subterminal sequences comprise another tandem repeat array. Note that this array is disqualified from being classified as a "subtelomeric" tandem repeat due to its presence at a single chromosome end (see "Terminology"). In SS4-20 and SS6-1, the telomere contained a truncated MoTeR2 and an intact MoTeR1, with short interstitial repeats (≤3 TTAGGGs) (i & ii). Differences in the sequence of variant repeats subtending the telomere suggests that the TEL2.ii form was derived from TEL2.i via replication slippage. In SS15-11, a truncated MoTeR1 was found inserted between the tMoTeR2 and MoTeR1 copies and two different terminal organizations were identified (iii & iv). The newly acquired tMoTeR1 was truncated at the same position as two copies found in TEL4 (Fig. 6), suggesting that it was acquired via gene conversion, or that TEL2 and TEL4 underwent an exchange of sequence. These scenarios also account for the acquisition of the drastically truncated MoTeR1 that is found in a terminal position in TEL4. Variant TEL2.iv presumably arose from TEL2.iii through an interstitial telomere break, followed by resection and, eventually,  four truncated MoTeR1s inserted tandemly in the telomere, two of which were missing the same amount of 5' sequence (264 nt). The TRF was unusual in that it also contained a second, drastically truncated MoTeR1 (MoTeR1 relic) at an "internal" location approximately 2 kb away from the telomere but whose orientation was inverted with respect to the telomeric copies. Also present was an inverted, interstitial telomere just 300 bp away from the start of the TTAGGG repeats at the MoTeR array border. It is clear that the internal relic was once telomeric because it had 1.5 telomere repeat units at its 3' end. In SS SS4-20, TRF10-1 had an identical organization, except that it possessed just one copy of the larger truncated MoTeR1 (variant ii). As was seen with TEL14 (Fig. 5), SS cultures possessing TRF10.i often exhibited signals at the same position as TRF10.ii (panel A), which suggests that 10.i also undergoes spontaneous truncations at a high frequency, that gives rise to structure 10.ii in some nuclei. However, this conclusion was not supported by the sequence data because, while two variants were identified among reads for SS15-11, their sizes were ~15 kb and > 20 kb. Furthermore, because the MoTeR array contained no long interstitial telomeres (panel B), there was no reason to suspect it would undergo recurrent breakage. Instead, the identification (and structure) of the longer (20 kb+) variant, and its presence -albeit as another faint band -in several SS cultures that possessed the short form,

Summaries of inferred rearrangement mechanisms
presented an alternative mechanism that fully explained the data. Specifically, the large and short variants are the expected products of an unequal sister chromatid exchange between the two larger tMoTeR1 insertions. It is not clear what would promote such a high level of recombination in this array.