Using full-length transcript sequencing to reveal the fate of mRNA in aging seeds

Dr. Margaret Fleming from Michigan state university spoke about using Oxford Nanopore cDNA sequencing to detect and quantify mRNA degradation in seeds over time. Dr. Fleming started by stating “Seeds… are not rocks, they may look like rocks, and they would be much easier if they were, but they are alive”. Dr. Fleming explained that most plant seeds are dry, therefore metabolism is halted and molecules such as degraded mRNAs cannot be regenerated. This is particularly important in relation to germination, as these mRNAs, which typically have very short half-lives, must persist for potentially decades in order for this process to occur. Dr. Fleming then showed examples of seeds belonging to Verbascum blattaria and Phoenix dactylifera which were successfully germinated after 120 years and ~2000 years respectively, suggesting that there are mechanisms in place which allow these unstable molecules to persist. Seeds are exposed to similar environmental forces as those which cause human aging, for example free radical oxidation, but as they are quiescent, i.e. their metabolism has halted, no response to this damage can occur. This is a likely reason as to why seeds and seed stocks eventually lose their ability to germinate after long-term storage in some cases. Germination rates can be measured as a bulk assay on the whole seed stock as a decrease in germination proportion over time, however, it is very difficult to predict when a seed stock will start to show the effects of this accumulated damage, or more specifically, determine which seeds will not germinate at all. Information linking RNA degradation to reduced germination rates would be highly beneficial to seed banks in order to preserve important plant species for later propagation.

Using soybeans harvested in two different years (2015 and 1994) and stored at 4°C, Dr. Fleming extracted RNA and used this material to determine if there was a difference in RNA degradation profiles between the two time points. Analysed 2 years after the 2015 harvest, the most recent harvest had 100% germination success, whilst the 1994 harvest had 2% success. The main questions for this study were: Does RNA degrade during dry storage, if so, what does this degradation look like, and can it be quantified?

Examining Ribosomal integrity numbers (RIN), a measure of total RNA degradation, the 1994 harvest exhibited more degradation than that of the 2015 harvest (average RIN = 6.7 vs 7.9 respectively). Furthermore, polyA-selected mRNA from the 1994 harvest was shorter when examined on a bioanalyser. At this point, Dr. Fleming stated “I’d just like to say, I was really pleased I could do this myself in my lab; I got my MinION, I prepared my library and it was very easy”. Using the Oxford Nanopore cDNA by strand-switching protocol with barcoding, multiple samples (n = 5 per timepoint) with external RNA controls consortium (ERCC) spike-ins were run on a single flow cell. Reference transcripts (n = 2,211) were selected in order to compare degradation patterns between the two harvests time points. They were chosen on the basis that: they were shared amongst all 10 replicates (five from each time point), the sequences covered >75% of the transcript in the 2015 harvest samples, and each had a read depth of greater than 20. In order to determine patterns of mRNA degradation, read depth at each base pair location was normalised as a proportion of the maximum count per transcript. Dr. Fleming then showed plots of mRNAs that either exhibited no degradation between each time point, degradation in the 1995 harvest or, in one case, significant degradation in both harvests. In order to quantify these observations, the difference in normalised sequence depth was used. Here, Dr. Fleming gave examples of a 60S ribosomal protein transcript which showed little to no difference between the two harvests, while a DC8-like embryonic protein transcript was drastically degraded in the 1994 harvest when compared with the 2015 harvest. In terms of actually calculating this relative degradation (RD), Dr. Fleming showed that the area under the coverage plot curve was a good starting point however only the middle 50 % of the transcript should be considered to avoid bias from reduced coverage at the ends. Using this figure, a delta RD can be calculated, which is a measure of the relative degradation between two conditions. If this value is high, it suggests high degradation rates in one condition vs the control.

Summarising the findings, Dr. Fleming suggested that when comparing the 1994 harvest with the 2015 harvest, the main determinant of mRNA degradation was transcript length. Using an ordination plot, the group showing the least relative degradation were transcripts less than 1200 bp in length while those showing the most degradation in comparison with the control were >2500 bp. Other than this, as yet no particular gene function, sequence motif or GC content was associated with degradation but on the basis of these results, long transcripts are currently being tested as markers for seed health.