Galactic Cosmic Rays (GCR) and Solar Cosmic Radiation (SCR) constitute two major risks for life forms of all types in the spaceflight environment. The biological effects of GCR and SCR at various dosage levels have been studied extensively to elucidate the mechanisms involved in repairing the DNA Double Strand Breaks (DSBs) typical of these exposures (Arena et al., 2014; Kavanagh et al., 2013). Interactions between these two types of Ionizing Radiation (IR) and microgravity-induced effects on the DNA damage repair response have been studied in both microgravity simulations and true spaceflight experiments, but no consistent synergistic effects have been identified (Moreno-Villanueva et al., 2017). Further research and application of this knowledge will contribute to efforts to effectively safeguard the genomic integrity of astronauts and other organisms beyond the protection of Earth's magnetic field (Durante, 2014; Kennedy, 2014). Research into the effects of radiation remains a high priority for the further advancement of crewed space exploration, both in terms of reducing risk to the mental and physical health of humans, as well as the viability of plants and other attendant biology vital to the support of long duration missions and exploration (NASA, 2018).
Ions of high-Z (charge) and energy (HZE) constitute 1% of GCR; the remainder is a combination of protons (85–90%), helium (11%), and electrons (1%) (Badhwar and O’Neill, 1992). HZE include the electron-stripped nuclei of Lithium (Z = 3) and all heavier elements, and pose significant danger to biological samples due to their capability for high levels of Linear Energy Transfer (LET), which causes damage along their paths. These impacts can be detected either in real-time or post-experimentally, and an array of hardware capable of measuring IR are available (Benton and Benton, 2001). Columbia Resin 39 (CR-39) is one such detector, allowing for post-exposure characterization of IR impacts. CR-39 is a polyallyl diglycol carbonate (PADC) polymer, which passively records HZE passage via bond breakage within its superstructure (Cartwright et al., 1978).
Efforts to replicate the spaceflight radiation environment in an accurate manner for purposes of advancing the field of radiation biology continue to be made in the NASA Space Radiation Laboratory (NSRL) at Brookhaven National Laboratory (BNL) (Miller and Zeitlin, 2016). Bombardment of biological samples with radiation of known ion composition and dosage rate allows for tightly-controlled experimental designs, the results of which can be extrapolated to support downstream studies of DNA damage resulting from cosmic radiation in more complex environments (Kiffer et al., 2017; Suman et al., 2017).
The upper polar stratosphere of Antarctica provides a useful environment for longer-term studies of IR and its effects, without requiring flight on a rocket beyond Low Earth Orbit. Access to this radiation environment is accomplished using various types of scientific balloons (Smith and Sowa, 2017). An average dose rate of 5.47 ± 0.31 μGy/hr was calculated from data acquired in the 2003 balloon flight of the TRACER (Transition Radiation Array for Cosmic Energetic Radiation) instrument hardware in this environment (Ave et al., 2011; Benton, 2012). These low dosages thus require a longer exposure time to reach the same absorbed dose as an artificial radiation source of increased intensity.
Seeds, including those of the model organism
The Cosmic Ray Exposure Sequencing Science (CRESS) payload system was developed as an end-to-end proof of concept experimental payload aimed at examining large numbers of seeds exposed to space radiation environments. The hardware was developed to fly on high-altitude balloon flights launched from Antarctica. The design of the hardware was based on components that could also be adapted for other experimental needs. The post-flight procedures were designed to assay the effort and components necessary for the screening and processing of large quantities of seeds. Long-read DNA sequencing technologies were applied to tissues arising from somatic mutants in order to assess the ability of the exposure and screening procedures to provide novel insights into the genomic impacts of space radiation exposure.
Hardware design for the CRESS payload was conducted with several critical specifications in mind. The containment vessel was designed to maintain an internal pressure of approximately 1 atm, record the impacts of IR, meet the 1U CubeSat standards, and carry a well-organized seed container to facilitate highly controlled post-flight processing and analysis. The payload layout can be seen in Figure 1. The basic unit of the payload is the seed cassette, which is composed of a seed tray flanked by sheets of CR-39 Solid-State Nuclear Track Detector (SSNTD) material [Track Analysis Systems Ltd] (Figure 1A). The seed trays are modified Nunc™ 1536-bin plates [ThermoFisher Scientific]. Each seed tray was hand-cut from the original 1536-bin plate and contains 576 bins (24 x 24 bin orientation). Each bin holds an average of 250 vertically stacked seeds (Figure 1A bottom). To secure the seeds in the tray, a polyolefin sealant film [USA Scientific] was placed over each loaded seed tray. The ANT biological payload of approximately 500,000
Figure 1D shows the components of the ANT payload containment system. The biological payload was bolted into a 3D printed containment vessel made from VeroClear™ resin [Stratasys], which was constructed using an Objet Eden 3D printer [Stratasys] utilizing PolyJet™ technology. This design was necessary to ensure that the containment vessel could hold 1 atm of internal pressure. To allow ease of integration onto the BACCUS hardware, the containment vessel was set into a 3D-printed support structure and then bolted onto the integration plate; both structures were produced using a CraftBot Plus 3D printer [CraftUnique], which utilizes Filament Deposition Modeling (FDM) technology. The assembled payload was bolted onto a corner of the BACCUS support structure (Figure 1E). Figure 1F shows the balloon being filled with helium on the ice, and Figure 1G shows the path of the balloon over Antarctica.
For this proof of concept experiment and comparison, the full CRESS payload that flew in Antarctica is referred to as ANT. The payload elements that were exposed at the NSRL at the Brookhaven National Laboratory are referred to as BNL. In the figures and tables ANT will refer to data from the Antarctica exposure while BNL will refer to data from the NSRL exposure.
A single seed cassette, which contained approximately 145,000 seeds, was encased in three CR-39 sheets on each side, and this assembly was placed inside a modified T-25 flask (Figure 1C). Particle accelerator time was reserved at NSRL and a specific ion dosage designed to simulate GCR was requested, as outlined in Table 1. The holder was oriented perpendicularly to the radiation source, and particles were projected in unidirectional paths so that ions passed in a straight line through the assembly. The beam had an area of effect of 20 cm x 20 cm, such that only a portion of the total dosage would impact the biological payload. Seeds were exposed to H, He, O, and Ti ions via the particle accelerator and further details of the entire BNL mixed-beam can be found in Table 1.
Profile of the mixed-beam for cosmic radiation simulation used in the Brookhaven National Laboratory (BNL) seed treatment.
The assembled CRESS payload was en route to Antarctica via the NASA Wallops Flight Facility and New Zealand from November 2, 2016 to November 16, 2016. The payload was removed from cold stowage and integrated onto the BACCUS experimental hardware six hours before launch (Figure 1F). The high-altitude balloon supporting the experiments was then launched from the McMurdo polar station in Antarctica on November 28, 2016 and maintained an altitude between 36 and 40 km above sea level for most of the 30-day float period (Kim et al., 2017) (Figure 1E, F, G). After this exposure to the upper-stratospheric radiation environment, the experiment landed and was recovered. The payload was stored at −20°C until it could be transported. The payload was returned to Gainesville, Florida on March 29, 2017; all seeds remained stored in the hardware at −20°C during shipment and after arrival.
From the ANT experiment, the top seed cassette (#1) was removed from the hardware and stored at 4°C for downstream assessment, while the other three seed cassettes remained in −20°C storage. Seed tray #1 was chosen due to its position at the top of the biological payload, where it was likely to receive a greater amount of radiation than the trays below. The top three CR-39 SSNTDs placed above this tray (#8–10) were also removed and stored at 4°C for further analysis. Eighteen bins were randomly selected from both the BNL and ANT seed trays, with 36 total bins being used for phenotypic screening.
The CR-39 sheets were etched with 6M NaOH for 8 hours at 82°C. Etching in a heated NaOH solution increases the diameter of the aperture created by the penetration, and this remains proportional to the size of the incident particle (Cartwright et al., 1978). Post-etching images of CR-39 track detectors were taken with an Olympus SZX12 stereomicroscope. The etched CR-39 sheets in Figure 2A–B show sections of CR-39 from the BNL and ANT exposures, respectively. The images in Figure 2C were taken under an Olympus SZBX12 compound microscope and illustrate the post-etching track shape expected from particles impacting the plate in a non-perpendicular manner. In addition, the etched BNL CR-39 sheet showed a uniform angle of particle tracks, and these tracks covered almost all the seed tray bins. This was due to the unidirectional source of radiation from the particle accelerator. However, on the etched ANT CR-39, the angle of tracks was less uniform and X-ray imaging allowed better visualization of radiation impacting the payload in the upper stratosphere.
An etched track detector from the ANT payload (#8) was scanned via X-ray Computed Tomography with the Phoenix v|tome|x m scanner [General Electric] and Datos Acquisition Software v. 2.4.0 [General Electric]. A 100 kV X-ray beam was directed at the CR-39 detector at a focus-object-distance (FOD) of 7.32 mm, and a 109x magnified image was transmitted onto a detecting plate opposite the X-ray source at a focus-detector-distance of 799.92 mm. The sheet was rotated approximately 360° and approximately 2500 images were taken with a pixel size of 200 μm2. Images were then imported into the Volume Graphics (v. 3.2) software and compiled into a three-dimensional digital reconstruction of the CR-39 sheet, with a pixel-to-voxel ratio of 1:1 (Figure 2D)
Irradiated seeds were removed in a single mixed batch from a randomly selected bin of the seed tray, hereinafter referred to as the M0 generation of seed as per convention. This process is illustrated in Figure 3A. M0 seed batches were then transferred into individual 1.5 mL microcentrifuge tubes for wet sterilization. Briefly, seeds were soaked in 70% ethanol for 5 minutes, followed by 10 minutes in 50% bleach with 2 drops of Tween-20 added per 5 mL of solution, and rinsed with sterile water 6 times to remove residual chemicals (Figure 3B). After sterilization, M0 seeds were stored in sterile water and vernalized for 4–5 days at 4°C before planting.
A total of 122 M0 seeds were planted on each Petri plate (150 mm x 15 mm Petri dishes [Fisher Scientific]) under sterile conditions (Figure 3C). A 0.5% Phytagel™-based nutrient media supplemented with 0.5x Murashige and Skoog (MS) salts [Caisson Laboratories], 0.5% (w/v) sucrose [Sigma], and 1x Gamborg's Vitamin Mixture [Caisson Laboratories] was used. Each bin of seeds required approximately two plates. A gridded label was adhered to the bottom of each plate, and one seed was planted per grid-square (1 cm2) to ensure equidistant spacing and consistent seed counts. After planting, plates were sealed with porous tape [MicroporeTM] to prevent contamination, labelled, and grown in growth chambers at 19 ± 2°C, under 24-h fluorescent light at approximately 80 μmol m−2 s−1 PAR.
Plates were scored for germination 14 days after planting and storage in the growth chamber (Figure 3D). Both the total number of seeds planted on the plate and the number that germinated were recorded. Plates were also generally screened at the 14-day time point to identify signs of phenotypic differences. Examples of phenotypic aberrations that were scored for include: microgrowth (crown diameter < 8 mm), presence of a single cotyledon only, absence of plant structures, and alterations of pigmentation. Atypical plants demonstrating sufficient health and size after scoring were transferred to soil for the generation of M1 seeds via self-pollination, while plants with stunted growth and/or absent organs were transferred to individual Phytagel™ plates for further growth and observation. Both of these groups were grown in the same conditions described above. Eventually, plants that displayed severely inhibited growth and development were flash-frozen in liquid nitrogen and stored at −80°C for subsequent DNA extraction. Any plates displaying contamination were discarded and no data were recorded.
To allow study of the effects of radiation damage on subsequent generations, a subset of bins from the BNL seed tray were dedicated to producing M1 generation seeds. Seeds harvested from the BNL seed tray were chosen for this analysis as the etched BNL CR-39 detectors showed a dense and uniform distribution of impacts that covered almost every bin in the seed tray. M0 seeds allotted for M1 analyses were de-integrated, wet-sterilized, and planted directly onto soil. M0 seeds remained separated by bin throughout seed processing and growth, and seeds harvested from M0 plants were stored according to the bin from which their parental plant originated. A total of 13 M1 seed stocks were collected, and 244 seeds were planted from each stock. M1 plants were screened using the same parameters as M0 plants.
Survey images of all plates were taken at 4-, 8-, and 14-day time points using an Axiocam 506 Color camera (Carl Zeiss Microscopy). Plants that displayed atypical phenotypes at the time of scoring were also imaged at later time points under an Olympus SZX12 stereomicroscope (Figure 3E) and tracked until maturity or until preservation (Figure 3F). Anomalous phenotypes that were discernible prior to the 14-day old time point were tracked and imaged at earlier time points.
Immediately following seed planting, plates were labelled by generation (e.g., M0, M1), test group, (e.g., ANT, BNL), bin (i.e., the location of the bin on the seed tray), and the numerical ID of the Petri plate on which the plant was grown. Individual plants were also denoted by their location within the Petri plate grid, which was labelled using the Cartesian coordinate system with alphabetical labels along the y-axis and numerical labels along the x-axis. This system of nomenclature was developed to organize image files over the course of the screening process and to serve as an effective tracking method from seed de-integration to DNA sequencing. All image filenames included the generation and origin of the seedling, as well as the date of imaging and the age of the seedling at the time of imaging. A sample file name of 180812_M0_ANT_F3-2_L12_8D would indicate that the image was taken August 12, 2018 of plant L12 at 8 days of age that originated from the F3 bin of the ANT seed tray.
Genomic DNA was extracted from selected mutants, which are displayed in Figure 4, for sequencing. Using the aforementioned system, these mutants were named: Mutant1-M0ANT W3-2 D7 (Figure 4A), Mutant2-M0ANT H16-2 K9 (Figure 4B), Mutant4-M1BNL H13-1 H10 (Figure 4C), Mutant5-M1BNL P18-3 F3 (Figure 4D), and 6 pooled mutants collectively referred to as Mutant7-M0ANT-Pooled-Mutants (Figure 4E–J). An ethanol precipitation-based method adapted from a previous publication was used (LeFrois et al., 2016). Briefly, either fresh or previously frozen mutant plants were ground directly in liquid nitrogen without rinsing in wash buffer. The aforementioned protocol, beginning with the addition of the lysis buffer, was followed accordingly until the phenol/chloroform/isoamyl alcohol extraction step, which was eliminated. After the pellets were suspended in Tris/EDTA/RNase A (TER) buffer and incubated at room temperature for 20 minutes, 7.5M ammomium acetate (AmOAc) and 95% ethyl alcohol (EtOH) were added to the tubes, inverted to mix, and placed at −20°C overnight. As described in the protocol, the tubes were then spun at 4°C for 10 minutes, the supernatant decanted, and the pellets thrice rinsed with 70% EtOH. The samples were dried on the benchtop for 15 minutes and suspended in 0.25x TE buffer. The concentrations of the DNA samples were quantified using a Qubit fluorometer [Invitrogen] and these values are displayed in Table 2.
Quantification of gDNA extractions via Qubit.
DIN: DNA Integrity Number
Barcoded, large-insert (10 kb) libraries were constructed using 600 ng of pure, high MW DNA (>40 kb) from the five different
The demultiplexed subreads from this PacBio SEQUEL run were aligned to the TAIR10 release of the
For each experimental group, 18 bins were de-integrated from the seed trays and used for determining rates of germination. Each bin contained an average of 250 seeds. As can be seen in Figure 5A, M0 BNL and ANT seeds displayed mean germination rates of 76.4% and 82.5%, respectively. In the non-irradiated controls, the mean germination rate was 98%, which is consistent with reports from other studies (Kazama et al., 2008). In both M0 BNL and ANT, the germination rate was significantly lower than the non-irradiated controls (Student's
To determine the effects of radiation on the next generation of seeds, the germination rates for M1 BNL seeds were also measured. The box-and-whisker plot in Figure 5B shows that germination rates for the M1 generation of sampled BNL bins were similar to the control group (at 98%) and significantly higher than the M0 BNL seeds (Student's
Mutation rate was calculated based on the number of phenotypic aberrations observed per 100 germinated seeds to avoid data skewing due to plants that displayed multiple aberrations. As was mentioned above, the mutation rate was calculated from screening 18 randomly selected bins, each containing roughly 250 seeds. Phenotypic aberrations were defined as observable traits that deviated in a significant manner from that of normally developing seedlings. These included color differences, developmental anomalies, and substantial size differences. Absence of true leaves, an ablated primary root or a substantial delay in root growth, and lack of cotyledons were the developmental aberrations most prominently observed in the plants grown from both ANT and BNL M0 seeds.
The box-and-whisker plot in Figure 5C shows the mean numbers of mutations observed in the M0 plants per 100 germinated seeds, as well as the overall distribution of mutation rates among bins. The non-irradiated controls showed a mean mutation rate of 1 event per 100 germinated seeds, whereas in the irradiated BNL and ANT M0 samples, the mean mutation rates were significantly higher than the control group at 12 and 4 aberrations per 100 seeds, respectively (Student's
To determine radiation-induced effects on the next generation of seeds, mutation rates were also determined for the M1 generation of BNL plants. The box-and-whisker plot in Figure 5D shows that BNL M1 sample had a significantly higher mutation rate than the non-irradiated control plants (Student's
As a proof of concept for the application of long read DNA sequencing technologies for detecting genomic structural variations in irradiated tissues, genomic DNA from the following plant samples was isolated and sequenced (see Figure 4):
Structural variants were detected in the genomic DNA from all five samples, with sizes that ranged from less than 100 bp to greater than 1 Mbp in length. Due to DNA repair of DSBs via non-homologous end joining (NHEJ) having potentially resulted in translocations and intrachromosomal SVs, these genomic rearrangements were of primary interest in identifying the genomic effects of IR in the sequenced samples. In Figure 6, Circos plots graphically represent the translocations detected in the genome (Krzywinski et al., 2009). Breakpoints of translocations that were bridged during sequencing are indicated by lines joining one chromosome to another. In the individual ANT mutants (Mutant1-M0ANT W3-2 D7 and Mutant2-M0ANT H16-2 K9) that were sequenced, 15 translocations (TLs) indicated by blue lines and 11 TLs indicated by red lines were detected in each mutant, respectively (Figure 6A). The six ANT M0 mutants (M0ANT P19-1 H2, M0ANT R2-1 E12, M0ANT P19-1 H3, M0ANT R2-1 F7, M0ANT P19-1 G12, and M0ANTC18-2 C10) from which DNA was pooled and sequenced (M0ANT Pooled Mutants), exhibited 12 TLs whose breakpoints are indicated by green lines (Figure 6B). In the individually sequenced M1 BNL mutants (Mutant4-M1BNL H13 1H10 and Mutant5-M1BNL P18-3 F3), 15 TLs indicated by orange lines and 9 TLs indicated by black lines were detected in the respective mutants (Figure 6C). As expected, the SV detection rate in the simulated Col-0 control datasets was much lower, with only 1 detected SV remaining after the aforementioned filtering steps (Figure 6D).
The total numbers of detected translocations, as well as their distribution, were largely similar between the ANT and BNL mutants. In some genomic regions of both the BNL and ANT mutants, the density of detected translocation breakpoints appeared to be elevated when compared to the remainder of the genome. These regions include the beginning of chromosome 5 (chr5:1–3 Mb) and two additional regions in chromosome 1 (chr1:9-15 Mb, 23–25 Mb). However, while the set of partner breakpoints associated with these potential regions of DSB enrichment demonstrated no fully conserved pattern in their distribution, there did seem to be a tendency for certain regions to be preferred as matches for repair. For example, the breakpoints in the (chr1:23-25 Mb) region connected primarily to corresponding breakpoints in the (chr3:5-17 Mb) region in both individually barcoded ANT mutants, while another connection is shared from this region to the (chr2:3-5 Mb) region in both an ANT mutant (Mutant1-M0ANT W3-2 D7) and a BNL mutant (Mutant5-M1BNL P18-3 F3). Chromosome 2 showed a larger number of identified breakpoints in the BNL mutants than the individual ANT mutants, but not the pooled ANT mutants.
Structural variants other than translocations were also detected. In Table 3, it is apparent that for most SV types the raw number of SVs that were detected was similar between mutants. A clear enrichment of inverted duplications (INVDUPs) was seen in the ANT Pooled Mutants sample, 96 INVDUPs having been detected in this sample whereas no more than 4 were detected in any other samples. M0ANT H16-2 K9 is also distinguished by its overall reduction in the number of detected SVs compared to the other samples. M1BNL H13-1 H10 in contrast demonstrates a slight elevation in the number of insertions detected.
Structural variants detected in the nuclear genomes of sequenced mutants.
DEL: Deletion, DUP: Duplication, INS: Insertion, INV: Inversion, INVDUP: Inverted Duplication
The methods described here provide a framework in which the effects of radiation can be examined at both the phenotypic and genomic levels in large numbers of seeds. The advantages of technologies such as PacBio sequencing, X-ray computed tomography, and access to deep space radiation analog environments facilitated the process of data collection, from seed irradiation through to post-sequencing bioinformatics.
As BNL was designed to serve in part as a positive control, it was expected that M0 BNL could have a high mutation rate relative to M0 ANT, and that the quantity of particle radiation events/mm2 of CR-39 sheets would also be elevated in the BNL experiment compared to that of the ANT exposure. A high dosage (Table 1) was also used in the BNL exposure due to the total area of the beam (20 cm x 20 cm) being larger than that of the seed tray used (5.4 cm x 5.4 cm). ANT seed tray #1 was selected for these analyses as it was expected to receive the most radiation of the payload, minimizing the dosage gap between the two seed populations being studied. The BNL exposure produced a visibly higher number of events/mm2 as visualized through CR-39 SSNTD, as well as more mutations in the BNL M0 than were seen in the ANT seedlings.
These results indicated that there may be significant variation in the observed biological effects resulting from differences in both the composition and dosage of the radiation presented to each group of seeds. A more thorough understanding of the radiation dosage for each individual bin could allow better explanation of the difference in mutation rates between the two samples, and further research into mapping the CR-39 SSNTDs with X-ray Micro-CT or other scans is necessary to accomplish a complete mapping of radiation events to seed bins. However, a lack of tracking of seed position and orientation within bins during the de-integration process prohibits conclusions being drawn on any significant protective effects dependent on sub-bin positioning.
Joint United States Air Force/National Oceanic and Atmospheric Administration (USAF/NOAA) Report and Forecast of Solar and Geophysical Activity (RSGA) documents from the time period of the ANT exposure (November 28, 2016 – December 28, 2016), indicate either low or very-low levels of solar activity for all but 2 days with moderate activity levels (November 29–30) in this period (NOAA, 2016). This indicates that the GCR was likely the primary source of IR to which the ANT payload was exposed. Previous studies have also shown that a large range of heavy ions (Z = 6–28) contribute to high-altitude HZE (NCRP, 1995). This array of HZE ions contrasts with the radiation environment experienced by the BNL seed cassette, which was composed of H, He, O, and Ti ions (Z = 1, 2, 8, and 22, respectively). The LET expected of the average HZE event in Antarctica would therefore be higher than that of the average HZE event in the BNL simulation. However, due to the fact that the BNL M0 exhibited a higher rate of mutations, it appears that the high dosage rate of the exposure relative to that of the ANT exposure contributed more to mutagenesis than differences in HZE ion composition. Deeper and detailed analyses of SSNTDs collected from the ANT payload will also provide information on which ions impacted individual bins.
Genomic damage, regardless of its source, has been shown to cause inhibition of the germination process; if the genome sustains an irreparable degree of damage the germination process can fail altogether (Waterworth et al., 2015; Waterworth et al., 2016). Germination may not fail if the genomic damage results in a mutator phenotype with increased tolerance of radiation-induced damage (Walker et al., 1997; Zhang et al., 2008). Study of these mechanisms is thus vital to guaranteeing significant rates of seed viability for long-term space exploration missions. The concern of how best to store seeds during these missions arises, as seeds irradiated in monolayers have been shown to exhibit lower germination rates (Kazama et al., 2008). However, little effort has been directed toward better understanding the extent of shielding provided by packaging many seeds into the same container. Study of other seed trays from the ANT payload could reveal some of these effects in future work, where seed trays further isolated within the payload may demonstrate different germination or mutation rates from peripheral trays. In the present study, the effects of upper stratosphere exposure and simulated GCR exposure on seed germination were assessed using only one tray from each condition.
Analysis of irradiated M0 ANT and BNL seeds compared to non-irradiated seed samples showed that radiation damage significantly reduced the germination rates of the experimental groups, which is in line with previous studies despite the array of radiation conditions under which they were conducted (De Micco et al., 2011; Kranz, 1986; Tepfer and Leach, 2017; Tepfer et al., 2012). This also served to validate the ability of the BNL mixed-beam to faithfully reproduce this basic GCR effect, to the level that there was no significant difference found between the BNL and ANT M0 germination rates. Interestingly, germination for the BNL M1 seed recovered to nearly that of the control group (Figure 5B). It is logical that the recovered germination rate in the M1 generation would be a result of the decreased viability of M0 seed due to lethal mutations, when the connections between germination regulation, growth regulation, and cell cycle checkpoints for genomic damage are taken into account (Waterworth et al., 2016; Zhang et al., 2008). Further study using lines deficient in these genes could enable the discovery of new radiation-induced phenotypes.
Germinated seeds with genomic damage of a high enough level to produce distinct mutant phenotypes were classified and counted to quantify the rates of mutation in each exposure sample. Compared to the non-irradiated control, both the BNL and ANT M0 plants exhibited significantly higher mutation rates. In contrast to the similarities seen in the germination rate data, observations of the mutation rates in M0 BNL and M0 ANT showed that significant differences existed between them, with the BNL rate being higher than was observed in ANT. This serves as an indicator that differences in downstream biological effects do exist between these radiation sources, and that there is still room for improvement of GCR simulations. Of particular interest is the significant difference in mutation rate between the BNL M1 samples and the control samples, which clearly demonstrates the presence of heritable mutations that were the result of radiation damage and repair in the genomes of the M0 BNL seeds.
In the M0 BNL and ANT test groups, the type of mutations and their frequencies differed. The discolorations observed in the BNL M0 were most prevalent near the shoot apex, and the phenotype was often accompanied by microgrowth. Deficiencies of cotyledon development were also seen to be enriched in the BNL M0 seedlings. However, the ANT M0 also contained a subset of abnormally-colored and stunted mutants, classified based on their cotyledons, which exhibited albinism. Dwarf architecture has been prevalent in past studies, and was statistically significant in its enrichment within the M0 BNL test group (De Micco et al., 2011; Lake et al., 2009). These findings suggest that there may be links between the levels of growth and discolorations observed. Radiation exposure has previously been connected with abnormal coloration, but this was also shown to be an LET-dependent effect (Kazama et al., 2008). While this could be a product of overall genomic damage, as stress-induced pigmentation changes are well-known to occur (Gangappa and Botto, 2016), it would be expected that the discoloration events would be more similar between samples if this were the case, unless the two types of radiation are significantly dissimilar in terms of LET values. Further study into the compositions of GCR and SCR as well as sampling of IR alongside biological experiments may allow determination of the true nature of these phenotypes.
Analysis of sequencing data for the purpose of detecting structural variants revealed that irradiation of
The similarities in observed translocations also indicated the potential for non-random factors having influenced the probability of certain chromosomal rearrangements occurring, such as spatial proximity and chromatin structure of genomic loci when IR-induced DSBs occur. IR can, in addition to impacting the genome directly, lead to the generation of reactive oxygen species that have the potential to create DSBs and other genomic lesions. Seeds have some protections against these threats, such as endogenous antioxidants (Roldán-Arjona and Ariza, 2009). While our approach is unable to distinguish between these direct and indirect types of damage, both are biologically relevant effects of IR and are of interest as a result. Prior investigations into the off-target repair frequencies of DSBs resulting from radiation have found trends similar to those observed in this study (Hada et al., 2011; Lavelle and Foray, 2014). The (chr1:23–25 Mb) region of chromosome 1 serves as an example of this concept, where it shows preferential off-target repair to chromosome 3 in only the ANT mutants. Previous studies have also used high-throughput sequencing methods to map genomic domains with high frequencies of interaction in
Mutant phenotypes in the ANT and BNL M0 were likely to have occurred due to somatic mutations in the seed embryo, which resulted in the death or delayed growth of certain plant organs, such as in the case of the ANT “rootless” mutants (Figure 4A and B). The resultant mosaicism offers the unique opportunity to sequence wild-type and mutant genomes within the same plant, but would create issues with tissue amplification measures such as tissue culture, where wild-type cells would be expected to outcompete mutant cells exhibiting deleterious phenotypes. However, the presence of heritable mutations in the BNL M1 samples, as evidenced by their observed mutation rate (Figure 5D), and the detection of structural variants in the sequenced BNL M1 individuals (Figure 6C), support the notion that intergenerational screening would be sufficient to affix these IR-induced SVs and allow better study of their biological effects.
Pooling multiple mutants (Figure 4E–J) into a single DNA sample before library preparation showed some potential as a strategy for the detection of SVs in samples that cannot provide enough DNA for individual sequencing libraries. As the mutants that were of the highest interest due to their severe phenotypes fell into this category, it was important to explore methods that could enable their study. It is noted, however, that in all respects other than the heavy enrichment of INVDUPs (Table 3), the SVs detected in the pooled ANT sample (Figure 6B) are similar to those in the individually barcoded samples (Figure 6A, C). Further study can shed light on both the possible causes and biological effects of these INVDUPs, and whether pooling of mutants is a viable option for SV analysis on a larger scale.
The issue of tissue availability was the primary driving force for the development of a yield-optimized DNA extraction protocol. DNA of sufficient quality and quantity for library preparations was successfully isolated from low-biomass samples despite the removal of cleaning steps that would have resulted in significant losses of DNA. In the event of low yields, DNA from multiple mutants could be pooled before barcoding and SVs were detected (Figure 6B), but at the cost of an inability to determine the sample of origin for each SV. The presence of organelle genomes in extracted DNA led to false SV calls due to high coverage of the organellar genomes leading to spurious alignments, but this was considered acceptable given that isolation of nuclei would lead to further losses of material. In future studies, organelle genomic sequences would be excluded from read alignment steps unless they are of particular interest.
Of the detected SVs, there was potential for the blunt-end ligation of SMRTbell adaptors to generate fusion artifacts between DNA fragments during library preparation (Griffith et al., 2018), but no DNA was able to be retained for validation of SVs due to the small amount available. However, it was notable that in the simulated datasets only one SV was detected (Figure 6D). This demonstrated that fusion artifacts were rarely called as SVs in the sequencing output of these other studies that also utilized blunt-end ligation during library preparation (Chin et al., 2016; Liang et al., 2018). In future work, the A/T overhang adaptor ligation method will be used to ameliorate this effect, and it is likely that a small but significant number of SVs detected in this study were artefactual.
The methods described here were developed for the purpose of defining an experimental pipeline that would allow workable experimentation on large numbers of seeds exposed to space radiation, with the goal of using seeds to characterize genomic damage by space radiation. The extended goal will be to create large datasets of space-irradiated genomes in order to deeply characterize genomic damage tendencies that may accumulate in long duration exposures to deep space environments. We found that these test exposures, screening procedures, and sequencing methods provided a tractable experimental path that resulted in clearly different phenotypes during the early development of
These data show that long duration Antarctic balloon flights provide exposure to enough IR to realistically emulate, not just simulate, deep space exposure to IR for the purposes of gathering data on genomic impacts. The balloon flight platform and operations provided enough exposure to make searching for mutations both tractable and productive. The outcomes include the genomic characterization of actual space IR mixes, rather than shorter, high fluence exposures meant to simulate deep space IR. These data clearly demonstrate the potential for long duration and low dosage rate experiments to provide unique insights on the biological effects of IR on eukaryotes relevant to space exploration. The further development of biological payloads and hardware to facilitate balloon-based sampling of naturally occurring IR can thus allow advances in scientific knowledge as well as protective technologies – without requiring travel beyond the Earth's magnetic field.
Quantification of gDNA extractions via Qubit.
Profile of the mixed-beam for cosmic radiation simulation used in the Brookhaven National Laboratory (BNL) seed treatment.
Structural variants detected in the nuclear genomes of sequenced mutants.