Dr Rowan Mitchell: Summary
I have a diverse background including crop physiology, plant biochemistry, mathematical modelling and bioinformatics. I have written >30 publications as first or senior author in scientific journals including PNAS, Plant Physiology and Plant Biotechnology Journal and have won research grants with value totalling >£2m. More details here.
In the last 15 years as a project leader at Rothamsted Research, I have been focused on using innovative bioinformatics in gene discovery for plant cell wall research and applications in quality improvement in grass and cereal crops.
Gene Discovery
Genes responsible for synthesis of arabinoxylan
Figure 1. Portion of AX molecule showing 7 Xyl backbone and most common decorations.; a whole molecule has typical length of 100s - 1000s Xyl.
Figure 2. Distribution of ratio of normalised counts of expression sequence tags (ESTs) cereals / dicots. Data recalculated from Mitchell et al. (2007). Groups of genes predicted to synthesise AX are indicated.
Arabinoxylan (AX) is an abundant molecule in grass cell walls, consisting of a long chain of xylose (Xyl) sugars decorated with arabinose (Ara) sugars some of which have a ferulate (FA) attached (Fig. 1). FA is not a sugar but a phenolic capable of oxidative coupling which provides an additional means of cross-linking chains in grass cell walls.
In 2006, the genes responsible for making AX were unknown, so I developed a bioinformatics approach to look for candidate genes. This exploited the differences between grasses and dicots: whilst dicots do have xylan, they have less in primary cell walls than grasses do, they have little Ara decoration of xylan and no FA. I reasoned that genes for AX synthesis should be highly expressed in grasses (because of the high absolute abundance of AX) and much more expressed than the closest matching genes in dicots. To assess this, I derived a measure of expression bias of groups of genes in grasses (actually cereals as this was only grass data available) relative to matching genes in dicots (Fig. 2). I predicted that genes of the right type amongst those showing positive expression bias cereals/dicots were responsible for AX synthesis: clades within glycosyl transfrease families GT43, GT47 and GT61 for making AX and an acyl transferase BAHD gene clade for adding FA (Fig. 2).
Figure 3. Evidence from our papers for role of genes predicted in 2007 to be responsible for synthesis of AX.
In the intervening years, experiments have shown that my predictions of these genes being responsible for synthesis of AX were correct.(Fig. 3). (However, in the 2007 paper I also unwisely speculated on the exact molecular function of each group of genes, which turned out to be much less accurate!).
Independently of my predictions, work from other labs has shown that the equivalent GT43 (IRX9, IRX14) [1, 2] and GT47 (IRX10) genes [3-8] in dicots are responsible for synthesis of the xylan backbone and work from our lab at Rothamsted and others shows they do the same job for AX in grasses [9-11].
We collaborated with Paul Dupree lab in University of Cambridge to show that grass GT61 genes mediate addition of 3-linked Ara to AX . Work from his lab showing that adding the grass genes to Arabidopsis introduced this linkage provided unequivocal evidence for this.
For the BAHD gene clade that I predicted to be responsible for addition of FA to AX, similarly strong evidence has proved difficult to obtain. Some of the genes within the clade have a different function [12] but Laura Bartley and co-workers found evidence that some genes were indeed responsible for acylation of AX and generously named this BAHD clade the "Mitchell Clade" [13]; they found particularly strong evidence for addition of p-coumarate, ( very similar to FA but does not cross-link) . The strongest evidence for a gene in this clade being responsible for addition of FA to AX came from our collaboration with Hugo Molinari group at Embrapa-Agroenergy showing large reductions in FA when one BAHD gene was suppressed in the model grass Setaria viridis.
Other references:
1. Brown, D.M., et al., Plant Journal, 2007. 52(6): p. 1154-1168.
2. Pena, M.J., et al., Plant Cell, 2007. 19(2): p. 549-563.
3. Brown, D.M., et al., Plant Journal, 2009. 57(4): p. 732-746.
4. Wu, A.-M., et al., The Plant Journal, 2009. 57(4): p. 718-731.
5. Chen, X., et al., Molecular Plant, 2013. 6(2): p. 570-3.
6. Jensen, J.K., N.R. Johnson, and C.G. Wilkerson,. Plant Journal, 2014. 80(2): p. 207-215.
7. Urbanowicz, B.R., et al, Plant Journal, 2014. 80(2): p. 197-206.
8. Zeng, W., et al., Plant Physiology, 2016. 171(1): p. 93-109.
9. Zeng, W., et al., Plant Physiology, 2010. 154(1): p. 78-97.
10. Chen, X.W., et al., Molecular Plant, 2013. 6(2): p. 570-573.
11. Chiniquy, D., et al., Frontiers in Plant Science, 2013. 4: p. 83.
12. Withers, S., et al., Journal of Biological Chemistry, 2012. 287(11): p. 8347-8355.
13. Bartley, L.E., et al., Plant Physiology, 2013. 161(4): p. 1615-1633.
AX is an important molecule in two quite different contexts for grass crop quality:
-
in wheat grain it is the predominant non-starch polysaccharide providing dietary fibre in human food, and influencing processing for non-food uses
-
in vegetative tissue (e.g. sugarcane residues or forage grasses) feruloylation (amount of FA) of AX is associated with resistance to digestion, an important trait for production of biofuels and ruminant nutrition.
My discovery of genes responsible for AX synthesis allowed us to try reverse genetic approaches to improve traits in both these areas.
Applications in Crop Improvement
1
Low viscosity wheat
About 25% of AX in wheat grain is extractable in water (WE-AX) giving rise to viscous extracts which are a problem for non-food uses of wheat. For example, during production of Scotch grain whisky, this leads to sticky residues which cause down time in distilleries for cleaning and extra wear on pumps. As proof of principle, we separately down-regulated the wheat GT43, GT47 and GT61 genes using RNAi suppression in transgenic wheat lines; we were able to demonstrate that suppression of GT43 in particular greatly decreased extract viscosity from wheat grain , due to a decrease in amount and chain-length of WE-AX . We then set out to achieve the same low viscosity trait using a non-GM approach called "TILLING" so that it would be acceptable for conversion into a commercial wheat variety suitable for the production of Scotch whisky. We undertook this project in collaboration with the Scotch Whisky Research Institute and the wheat breeders Limagain UK. We successfully achieved a novel non-GM wheat line by stacking 3 loss-of-function alleles for the GT43 gene (now called TaIRX9b) and were able to demonstrate decreased extract viscosity and normal grain size (Fig. 4). Interestingly the grain cell walls seem to compensate for the loss of AX amount and chain length by increasing FA-mediated cross linking.
Figure 4. Low viscosity wheat project. Graphs compare lines on novel wheat with 3 loss-of-function GT43 alleles ("triple") with control line and commercial variety ("Cadenza") showing no effect on grain size but reduced extract viscosity. Photo shows bulking up of backcrossed control and triple lines for testing in distillery (August, 2020).
The next steps are being undertaken by our collaborators at Limagain and SWRI to test the novel wheat at pilot scale in a distillery and to breed the alleles into a modern,suitable genetic background for commercial release. Success will mean that locally grown wheat will be preferred over imported cereals for distillery use.
2
Sugar cane with more digestible residues
Sugar cane production generates huge amounts (100s of millions of tonnes annually in Brazil) of low value residues after the sucrose is removed. These residues are comprised mostly of cell walls in which abundant sugars are locked up ; these sugars can be converted to bioethanol to substitute for fossil fuels and thereby reduce greenhouse gas emissions. The main economic barrier to this is the low digestibility of the residues- the difficulty of releasing the sugars from the polysaccharides in the cell walls. It has long been known that feruloylation (i.e. amount of FA on AX) of grass cell walls is associated with reduced digestibility. We showed in collaboration with the Molinari group at Embrapa Agroenergy that suppression of the BAHD01 gene decreased cell wall feruloylation and increased biomass digestibility in a model grass (Fig. 3 above). The Molinari group, with some contributions from me, has now transferred this work to sugar cane using RNAi to suppress the orthologous BAHD01 gene. Early results show promise and they are now growing in the field with the aim of developing a commercial sugar cane variety with increased efficiency of bioethanol production from residues (Fig. 5).
Figure 5. Sugarcane with enhanced residue digestibility for bioethanol project. Photo shows RNAi BAHD01 and control sugarcane lines growing in the field at Embrapa-Agroenergy, Brasilia (July, 2020).