Fast neutron mutagenesis in soybean enriches for small indels and creates frameshift mutations

Abstract

The mutagenic effects of ionizing radiation have been used for decades to create novel variants in experimental populations. Fast neutron (FN) bombardment as a mutagen has been especially widespread in plants, with extensive reports describing the induction of large structural variants, i.e., deletions, insertions, inversions, and translocations. However, the full spectrum of FN-induced mutations is poorly understood. We contrast small insertions and deletions (indels) observed in 27 soybean lines subject to FN irradiation with the standing indels identified in 107 diverse soybean lines. We use the same populations to contrast the nature and context (bases flanking a nucleotide change) of single-nucleotide variants. The accumulation of new single-nucleotide changes in FN lines is marginally higher than expected based on spontaneous mutation. In FN-treated lines and in standing variation, C→T transitions and the corresponding reverse complement G→A transitions are the most abundant and occur most frequently in a CpG local context. These data indicate that most SNPs identified in FN lines are likely derived from spontaneous de novo processes in generations following mutagenesis rather than from the FN irradiation mutagen. However, small indels in FN lines differ from standing variants. Short insertions, from 1 to 6 bp, are less abundant than in standing variation. Short deletions are more abundant and prone to induce frameshift mutations that should disrupt the structure and function of encoded proteins. These findings indicate that FN irradiation generates numerous small indels, increasing the abundance of loss-of-function mutations that impact single genes.

Keywords: fast neutron; mutagen; mutation; single-nucleotide variants; soybean.

Figure 1

The detection of de novo mutations in FN treatment lines requires the filtering of naturally occurring variants present in the initial treatment lines, including deletions in the treatment line where reference-based read mapping cannot identify variants. Sequencing issues, including low-quality variant calls and low-coverage regions, are removed to isolate variants unique to treatment lines. Accurate base calls are shown in bold, even if they are filtered out. Erroneous base calls are shown in red.

Figure 2

The number of variants identified in each fast neutron line. The samples were exposed to 8 Gy (dark blue), 16 Gy (light blue), or 32 Gy (orange). Dotted lines indicate the expected number of SNPs and indels based on average mutation rates.

Figure 3

The mutational spectrum of fast neutron variants, rare standing variants (nonreference allele count of two or fewer), and common standing variants (nonreference allele count of three or higher). Each class combines the displayed change and its reverse complement [e.g., the notation C→T* indicates that both cytosine to thymine and the reverse complement guanine to adenine mutations are binned into a single class (see text)].

Figure 4

The nucleotide sequence context in which C→T transitions occur relative to the reference genome for fast neutron variants (A, N = 319), rare standing variants (B, N = 647,614), and common standing variants (C, N = 1,350,269). The relative height of a letter indicates its relative entropy (RE), with overrepresented and underrepresented bases portrayed as right side up and upside down, respectively. The null expectation (zero RE) is based on a nearby nucleotide of the same base that has mutated, chosen at random.

Figure 5

The distribution of insertion and deletion lengths for fast neutron variants, rare standing variants (nonreference allele count of two or fewer), and common standing variants (nonreference allele count of three or higher). Insertions are shown as positive values and deletions as negative values. Variants with lengths greater than 20 bp are not shown.