NOTE: Extended to 1/10/2021 per NSSC information (Ed., 1/21/2020)

NOTE: Extended to 3/10/2020 per NSSC information (Ed., 3/12/19)

The basis for this study involved a comparison between the type of damage caused by HZE radiations compared to that produced following exposure to more common terrestrial radiations, like gamma rays. The type of chromosome damage under investigation involves illegitimate rejoining of genetic material from one chromosome to another in the form of chromosome translocations. We investigated the DNA sequences surrounding the new junctions produced when such translocations are formed. For gamma rays we found that the junctions could be characterized by the resection of 1-6 basepairs of DNA, containing short stretches of sequence homology between the pieces of chromosomes being rejoined, and that the initial breaks in the chromosomes frequently occurred in repetitive DNA, as opposed DNA coding for genes.

Ed. note July 2021: Project continues as Directed Research project, "Molecular Characterization of Transmissible Chromosome Aberrations Produced by Ions of Intermediate and High Atomic Number: grant 80NSSC21K0679," with the same PI, Dr. Michael Cornforth. See that project for subsequent reporting.

NOTE: Extended to 3/10/2020 per NSSC information (Ed., 3/12/19)

Objective 1 of this proposal involves the Isolation and cytogenetic characterization of cell clones to be used in further molecular analysis of chromosomal inversions and translocations. We have collected and cryopreserved several human cell clones that represent the survival and clonal expansion of single cells exposed to gamma rays, 56Fe and 7Li ions. These cell clones harbor a range of transmissible chromosome translocations and inversions. In light of more pressing challenges related to Objective 2, we consider the number of clones representing exposure to the various radiations used to be sufficient. Aside from some incidental inversion analysis by directional genomic hybridization (dGH), Objective 1 has been completed to the point where current efforts were focused on Objective 2.

Objective 2 of this proposal involves the molecular characterization of these clones through the use of Next-Generation Sequencing (NGS), in order to determine the nature of the illegitimate junctions formed at the DNA level.

Sanger sequencing at the base-pair level from the amplified fragment from a paired-end library in clone K1-400C4 showed a 4 bp microhomology at the t(3;4) translocation junction of clone K1-400C4, supporting the idea that such rearrangements are characteristic of microhomology-mediated nonhomologous endjoining (mmNHEJ) misrepair pathways. Significantly, the breakpoint on chromosome 3 mapped to an LTR sequence, while that on chromosome 4 mapped within a LINE element. To our knowledge, this is the very first report involving the sequencing and validation of a known radiation-induced translocation in human cells using modern massively parallel sequencing (Cornforth et al., Radiat Res, 2018. 190(1): p. 88-97).

We still needed to establish mmNHEJ as a consensus mechanism for exchange aberration formation produced by gamma rays, and also whether similar mechanisms underly the fomation of such rearrangements following exposure to densely ionizing (high LET (linear energy transfer)) 56Fe and 7Li ions listed in the proposal. This analysis took the better part of three years to accomplish, which is far too slow to meet the objectives of the grant proposal.

The significant difficulties we experienced in sequencing this rearrangement can now be understood by considering the fact that both breakpoint junctions of this reciprocal translocation occurred in repetitive DNA. For example, the chromosome 3 breakpoint mapped to an LTR sequence and the chromosome 4 sequence mapped within a LINE element. Repeat elements are notorious in causing difficulties for all next-generation sequencing approaches. In hindsight, it is apparent that this would be especially true for the short-read mate-pair approach we were using at the time, which led to thousands of false-positive calls to the reference genome. Moreover, we realized, given the large number of DNA repeats in the human genome, that radiation-induced breakpoints within repeats would be the rule, rather than the exception. Thus, we concluded in last year’s report that strategies making use of longer insert libraries would be necessary, such as sequencing based on mate-pair libraries, for which we provided proof-of-principle efficacy by re-sequencing clone K1-400C4, to show it yielded identical breakpoint assignments compared to paired-end methodologies. We also proposed investigating the use of single molecule/real-time sequencing (SMRT) in the context of its increased cost compared to paired-end and mate-pair strategies.

In this report, we compared whole-genome sequencing via mate-pair sequencing as an alternative approach to identify SVs (structural variations) with a lower false positive rate. Mate-pair sequencing provides short read information for each end of DNA segments that are separated by ~2-4 kilobases in the genome and are therefore likely to be outside the repetitive region where the breakpoint(s) may be located. By comparing mate pair calls with those made from paired-end whole genome sequencing reads, we were able to identify the true-positive breakpoint and its position at single nucleotide resolution for the (3;4) reciprocal translocation of clone K1-400C4 with markedly fewer false-positive SV calls. We also identified the breakpoints of putative translocations in another 11 clones by mate pair sequencing alone, and many of those calls to the reference genome were supported by cytogenetic data (mFISH). It is important to note, however, that we have failed with either approach to call cytogenetically observable inversions as detected by cGH.

More recently, we have successfully employed to SMRT (single molecule/real-time sequencing via PacBio) to characterize SVs. This newer alternative approach is capable of generating very long sequencing reads, which we think will help us to characterize inversions, and improve greatly the throughput of our workflow. This approach was, until recently, far too expensive to be applied for our studies, but with the Sequel System now in place at University of Texas Southwestern (UTSW), these costs have dropped by 80%, now making SMRT sequencing affordable for our project. We used SMRT to sequence the DNA of 6 clones and the parental nonirradiated control clone to comprehensively characterize the SVs.

The DNA samples included one clone that was previously (and definitively) characterized by whole genome paired-end and mate pair sequencing (clone K1-400C4); the remaining 5 clones were previously analyzed by mate pair sequencing. Using a computational algorithm (pbsv) for long read sequencing data, we identified putative breakpoints in the genome of 6 clones. Importantly, we were able to accurately call all translocations that were cytogenetically observed. There were only a few additional/extraneous SVs called, which suggests that SMRT sequencing is plagued by few (if any) false-positive SVs calls. SMRT sequencing appears to be more specific than mate pair sequencing, which, in turn, has much higher specificity than our original short-read paired-end sequencing approach.

We do note, however, that even SMRT sequencing failed to identify an inversion on chromosome 3 of clone K1-400C4. Since we have consistently failed to identify this SV with the three sequencing approaches utilized, we suspect there may a technical issue that prevents the identification of this inversion breakpoint. We speculate that this breakpoint may generate a secondary DNA sequence structure that prevents replication by DNA polymerases, which could prevent either the generation of fragments encompassing this breakpoint during sequencing library preparation or the incorporation of nucleotides during the sequencing reaction. Both of these scenarios would prevent the generation of sequencing reads that span the inversion breakpoint. In either case, we conclude that SMRT sequencing appears to be an efficient and specific approach, which is superior to short-read paired-end sequencing and mate pair sequencing for calling radiation-induced translocations definable with single nucleotide resolution.

This will be the last report prior to the final close-out of this project, which we feel is on the cusp of an important breakthrough. It is expected that these newer two approaches outlined above, either by themselves, or likely in combination with short read technologies, will allow for a more rapid and accurate characterization of breakpoint junctions of radiation-induced large-scale structural variants to human chromosomes.

Objective 1 of this proposal involves the Isolation and cytogenetic characterization of cell clones to be used in further molecular analysis of chromosomal inversions and translocations. We have collected and cryopreserved several human cell clones that represent the survival and clonal expansion of single cells exposed to gamma rays, 56Fe and 7Li ions. These cell clones harbor a range of nonlethal chromosome translocations and inversions.

Over 20 human cell clones have been collected and cryopreserved that represent the survival of single cells exposed to various ionizing radiations. A compilation of data relating to clones in various stages of analysis has been updated with additional data to include mate-pair analysis of clones surviving exposure to gamma rays and 56Fe ions.

As before, the 7Li clones still require further analysis by mFISH and G-banding before they are ready for sequencing and analysis using the newer 3-color/3-chromosome dGH probe sets for inversion analysis. For reasons related to issues with Objective 2, this was temporarily postponed.

Objective 2 of this proposal involves the molecular characterization of these clones through the use of Next-Generation Sequencing (NGS), in order to determine the nature of the illegitimate junctions formed at the DNA level.

Our workflow had been predicated on the following sequence of tasks. Aberration-bearing clones are isolated and characterized on the basis of translocations via mFISH (and inversions via dGH). Next, these clones are subjected to conventional G-band analysis, in order to more succinctly localize the rearrangement breakpoints. Following G-banding, the next step in the workflow was paired-end sequencing, followed by PCR, and eventual Sanger sequencing at the base-pair level from the amplified fragment, an approach that eventually we successfully implemented for one of the clones, and which we recently published. We discovered a 4 bp microhomology at the t(3;4) translocation junction of clone K1-400C4, which substantiated our working hypothesis that such rearrangements are characteristic of mmNHEJ misrepair pathways. It was also discovered that both junctions of the reciprocal translocation occurred in repetitive DNA. The breakpoint on chromosome 3 mapped to an LTR sequence, while that on chromosome 4 mapped within a LINE element. To our knowledge, this is the very first report involving the sequencing and validation of a known radiation-induced translocation in human cells using modern massively parallel sequencing (Cornforth, et al., Radiat Res, 2018. 190(1): p. 88-97).

The analysis of paired-end libraries was deemed far too slow to meet the timely objectives of the grant proposal, and the sequencing of a single translocation is hardly proof of a consensus mechanism, so more work remained. This included analyses of clones that originated from cells exposed to high LET (linear energy transfer) radiation, which we fully recognize is of particular interest to NASA. Consequently paired-end sequencing was scrapped in favor the newer technique of mate-pair analysis, which allows for the assembly of longer-read fragments. Despite encouraging preliminary results using mate-pair described in last year’s progress report, we discovered some glaring inconsistencies between what the cytogenetic data was telling us about certain aberration-bearing clones, compared to the sequencing data derived by mate-paired analysis. Given the preponderance of repeat DNA elements in the human genome, we began to realize the chromosome exchange breakpoints occurring with repetitive DNA is likely to be the rule, rather than the exception. Further, that this fact could explain the aforementioned inconsistencies between mate-pair and cytogenetic endpoints, and the preponderance false-positive calls to the reference genome that plagued our early attempts. We now believe that strategies making use of longer insert libraries should circumvent repetitive DNA problems related to short reads, as is the case for paired-end and, to a lesser extent, mate-pair approaches.

For that reason, we have re-assessed our sequencing strategy to embrace newer long-read technologies. This includes SMRT (single-molecule real-time) sequencing that can generate very long sequencing reads (~20 kb) using a Sequel System (PacBio) that was acquired and implemented in the McDermott NGS core at UTSW (University of Texas Southwestern). While this approach was, until recently, too expensive to be applied for our studies, these costs have dropped substantially, making SMRT sequencing now affordable for our project. We also plan to employ a newer less-expensive approach known as Linked-Read sequencing that utilizes molecular barcodes to tag reads that come from the same long (~50 kb) DNA fragment, thereby providing the long range information missing from standard approaches for a more complete characterization of SVs (structural variations) in the genome. For that approach we will use the Chromium platform from 10x Genomics that was recently acquired and implemented in the McDermott NGS core at UTSW. Linked-read libraries maintain haplotype and other long-range information and are compatible with standard short-read whole genome Illumina sequencing.

It is expected that these two approaches, either by themselves, or likely in combination with short read technologies, will allow for a more rapid and accurate characterization of breakpoint junctions of radiation-induced large-scale SVs, such as the translocations and inversions we have already identified by mFISH and dGH.

Objective 1 of this proposal involves the Isolation and cytogenetic characterization of cell clones to be used in further molecular analysis of chromosomal inversions and translocations. We have collected and cryopreserved several human cell clones that represent the survival and clonal expansion of single cells exposed to gamma rays, 56Fe and 7Li ions. These cell clones harbor a range of nonlethal chromosome translocations and inversions.

Incremental progress on Objective 1 was made. Specific Aim 1a of Objective 1 is largely complete. It is likely that no additional clones will need to be isolated, although we will not know for certain until Specific Aims 1b (G-banding and mFISH analysis) and 1c (inversion analysis) are complete. Progress focused on G-banding (Specific Aim 1b) of this Objective. In previous reports we showed banding analysis for clone K1-400C4. More recently, we partnered with Eli Williams, a clinical cytogeneticist at UVA (University of Virginia), who performed traditional G-banding on 12 additional clones. Data for four clones deriving from exposure to 56Fe ions is now included. G-banding is a vital step prior to sequencing, since it isolates breakpoint junctions to within 5-10 Mb pairs of DNA. This, in turn, is needed to screen out the plethora of false-positive calls that are typical of bioinformatic analysis of DNA sequencing data (Objective 2 below). The 7Li clones collected and cryopreserved still require further analysis by mFISH and G-banding, before they are ready for sequencing. Since KromaTid (the source of our dGH probes) now is capable of producing 3-color/3-chromosome dGH probe sets, we intended to use those for future inversion analysis, as soon as we fit our microscope with the appropriate filter sets.

Objective 2 of this proposal involves the molecular characterization of these clones through the use of Next-Generation Sequencing (NGS), in order to determine the nature of the illegitimate junctions formed at the DNA level.

After initial problems were encountered, last year we decided to concentrate our efforts on clone K1-400C4 in order to determine the best path forward. In the previous progress report we were able to tentatively identify one of the breakpoints in this rearrangement, but had not yet provided validation of its location. This is because a reciprocal translocation, such as the one identified in K1-400C4, actually contains two breakpoint junctions: in this case, both a t(3;4) and a t(4;3) component. Without information about both junctions, it was not possible to fully characterize the rearrangement at the DNA sequence level. In this report we describe below the successful characterization of both breakpoint elements, sequence across both of them, fully characterize the translocation, and suggest a more robust path forward for future studies.

We increased the genomic depth of coverage from 8X to 30X and employed several new and more powerful bioinformatic software algorithms as they became available. We filtered out false positive SVs that occurred from reference mapping artifacts by comparing calls from clone K1-400C4 to those made in a panel of 5 normal DNA samples that included a control clone from this study, and 4 normal samples from the International Cancer Genome Consortium (ICGC). The consensus sequence from MPS for the best-supported candidate translocation indicated a 4bp AAGG overlap between the chromosome 3 and 4, in otherwise alignment between sequences at 3q26.2 and 4q31.1 in the Dec 2013 (GRCh38/hg38) human genome assembly.

A BLAT alignment showed that the chromosome 3 sequence mapped to an LTR sequence (bases 70-133) and that the chromosome 4 sequence mapped within a LINE element (bases 1-73). These repetitive elements were responsible for large number of highly homologous matches within this region. [We believe this led to the plethora of false-positive calls that plagued our initial analysis.]

We used the consensus sequence to design forward and reverse primers to enable PCR amplification across both possible junctions of the reciprocal exchange breakpoints. Sanger sequencing of the extracted and amplified PCR products revealed a translocation event that resulted in a 1 bp deletion of chromosome 4 and a 6 bp deletion of chromosome 3, with the translocation occurring within a 4 bp “AAGG” overlap between chromosome 3 and 4. We take our results as confirmation that t(3;4)(q26.3;q31) translocation has been successfully characterized at the DNA base pair level. These results are consistent with the rearrangement having been produced via microhomology-mediated nonhomologous endjoining (mmNHEJ). The finding that both breakpoints of the translocation in question occurred in repetitive DNA elements is also noteworthy. Although we did not anticipate this result, it is not altogether surprising, given that the majority of sequences in the mammalian genome are composed of DNA repeats of one type or another. Thus we can expect radiation-induced exchange breakpoints within DNA repeat sequences to be commonplace.

We concluded that sequencing longer insert libraries may have a higher specificity for the identification of genomic regions containing repetitive DNA elements. We tested this strategy using Nextera Mate-Pair library preparation (Illumina) followed by Illumina sequencing at ~4X genomic overage. We generated and sequenced 13 mate-pair libraries from 12 additional translocation-bearing clones that derived following exposure to gamma rays, or to 1GeV/amu 56Fe ions. This analysis also included the K1-400C4 clone that was previously analyzed by whole genome paired-end sequencing as described above, and the parental K1 clone.

By this new strategy, we were able to identify a (3;4) translocation breakpoint within K1-400C4, and it was identical to the one found earlier by paired-end analysis (Chr3:170,398,638; Chr4:139,706,161). Furthermore, when we compared the totality of breakpoint calls from the mate-pair sequencing with those of 30X whole genome sequencing for clone K1-400C4, only the (3;4) translocation breakpoint was shared between both sequencing data sets. Importantly also, from the mate-pair sequencing data for an additional 11 clones, we were also able to detect breakpoint regions that are supported by cytogenetic analysis.

These results strongly suggest that the identification of shared SVs predicted from mate-pair and whole-genome sequencing reads provides a robust solution to eliminate false-positive SV calls and to identify true positive genomic rearrangements from sequencing data. This should lead to much higher throughput of SV analysis, compared to paired-end sequencing alone.

To our knowledge, we are the first to actually characterize and validate a known radiation-induced translocation in human cells using modern massively parallel sequencing (MPS). Our working hypothesis is supported by the 4 bp microhomology at the t(3;4) translocation junction of clone K1-400C4, which is characteristic of mmNHEJ repair/misrepair pathways. A paper reporting these results has been submitted to a peer-reviewed journal.

In addition to previously isolated clones surviving exposure to gamma rays and 56Fe ions, we also now have data for clones whose progenitor cells were exposed to 0.2 Gy of 1.5 MeV 7Li ions delivered at NASA Space Radiation Laboratory (NSRL). These were sent to Colorado State University (CSU) for Directional Genomic Hybridization (dGH), a technique that allows us to interrogate the genome for chromosomal inversions. Two of the 7Li clones tested positive for one or more inversions, including clone K1-Li-04. Twelve clones were karyotyped by G-banding at the University of Virginia, which is necessary to localize the breakpoints of exchanges to within 5-10 megabase pairs of DNA.

Working together with KromaTid Inc., we have now have access to a single-color dGH probe set covering chromosomes 1, 2, and 3, 7, and X, which should increase dramatically our ability to screen for radiation induced inversions. We anticipate the construction of 3-color/3-chromosome dGH probe sets in the near future.

Objective 2 of this proposal involves the molecular characterization of these clones through the use of Next-Generation Sequencing (NGS), in order to determine the nature of the illegitimate junctions formed at the DNA level.

We had previously hoped that we had identified the translocation breakpoint in clone K1-400 C4 by NGS. Unfortunately these efforts failed to produce results consistent with the known cytogenetic location of the translocation. The biggest problem we faced was the preponderance of false-positive SV calls to the reference genome. In other words, initial analysis left us with thousands of false-positive results. Since then, we incorporated several changes to our approach that finally produced successful results in one of the clones. This included a newer and improved version of the bioinformatic software used to compare DNA from the irradiated clones, with that of normal DNA. In order to correct for various reference mapping artifacts, we filtered out false positive SVs (structure variants) that occurred from reference mapping artifacts. We did this by comparing potential SVs found using this software to those in a panel of 5 normal DNA samples: that of a control clone from this study and 4 normal samples from the International Cancer Genome Consortium (ICGC). This reduced the total number of potential translocations to a few dozen. We further reduced the number of candidate translocations down to only one position that agreed with the cytogenetic position of the breakpoint at t(3q26.2;4q31.1) by requiring at least two supporting reads for each filtered sequence. PCR primers were made flanking this presumptive breakpoint, and amplified fragments were Sanger-sequenced by its chromatography profile.

The one translocation that we did manage to sequence shared a short four-basepair stretch of DNA sequence homology at the breakpoint junction between chromosome 3 and 4. This suggests that nonhomologous end-joining was responsible for the formation of this translocation, although more clones will need to be sequenced before we reach a consensus mechanism for this type of radiation damage under study. We have determined that most all of the challenges we faced can be overcome by starting with a library with larger inserts. So we have now assembled a mate-pair library, which has been constructed and sequenced at the University of Texas Southwestern. It is currently being interrogated with various bioinformatic algorithms. We expect that these changes will dramatically increase the throughput of our experimental system, and allow for the sequencing of several other translocations and inversion in a timely manner.

Because of issues related to Objective 2, one particular clone (K1-400 C4) was further analyzed by G-banding at Emory University, in order to more accurately assign the cytogenetic location of translocation breakpoints. G-banding localized the translocation breakpoints as occurring between chromosomes 3 and 4. A paracentric inversion involving chromosome 3 that was too small to be seen by either G-banding or mBAND was discovered using directional genomic hybridization (dGH), a cytogenetic approach we developed specifically for the detection of inversions.

Objective 2: Molecular characterization of clones. DNA libraries were made from five clones, each containing a particular radiation induced translocation, and one clone control clone. These were analyzed independently (by whole genome sequencing; WGS) by laboratories at Oregon Health & Science University (OHSU) and UTSW (University of Texas Southwestern Medical Center). Despite experimenting with various established algorithms, neither laboratory was initially able to identify discordant reads that were consistent with the breakpoint locations in any of the cytogenetically verified rearrangements. The decision was made to concentrate sequencing efforts on clone K1-400 C4, which mFISH analysis showed to contain a gamma-ray-induced t(3;4) reciprocal translocation, and which dGH showed to also contain a prominent paracentric inversion in chromosome 3. We were eventually able to identify a putative translocation breakpoint between chromosomes 3 and 4 of this clone, and oligonucleotide primers that were designed that flanked the presumptive breakpoint location yielded a PCR product of anticipated size. Unfortunately sequencing analysis showed this to be a false positive position in the genome. Moreover, all of the presumptive inversions tentatively identified by WGS were inconsistent with the known cytogenetic position of the inversion as well. The biggest problem we faced was the preponderance of false-positive calls to the reference genome.

At OHSU, a deeper-coverage genomic library was made from clone K1-400 C4. It was sequenced using a different bioinformatic algorithm, and calls were stringently filtered against (normal background) structural variants from large panel of normal samples. This allowed us to correct for any reference mapping artifacts. As a result, we were able by WGS to presumptively identify the cytogenetic location of this reciprocal translocation. Although it will require further verification, it is likely that we have successfully identified the genomic location of the reciprocal translocation in clone 400 C4.

Our success with mapping this translocation notwithstanding, and in order to move our research forward at a faster pace, we plan to employ sequencing of longer insert libraries, which has higher specificity for the identification of the genomic regions where the breakpoints are located. We plan to implement this strategy by using mate-pair sequencing, combining this data with that from short read paired-end sequencing.

Objective 3: Cytogenetic Effects of Abrogating CtIP activity. During final negotiations with NASA program management, this objective was deleted from the proposal.