vibrational spectroscopy-based chemometrics to …vibrational spectroscopy-based chemometrics to map...

36
Vibrational Spectroscopy-Based Chemometrics to Map Host Resistance to Sudden Oak Death Pierluigi (Enrico) Bonello Department of Plant Pathology 6 th SOD Symposium, San Francisco

Upload: others

Post on 27-Jun-2020

9 views

Category:

Documents


0 download

TRANSCRIPT

Vibrational Spectroscopy-Based

Chemometrics to Map Host

Resistance to Sudden Oak Death

Pierluigi (Enrico) Bonello Department of Plant Pathology

6th SOD Symposium, San Francisco

163

Brice McPherson

UC Berkeley

Coauthors

Luis Rodriguez-Saona

Ohio State Univ.

Anna Conrad

Ohio State Univ.

(Currently U. Kentucky)

Dave Wood

UC Berkeley

164

Jason Smith

Univ. Florida

Coauthors on Paper Presented at the North American

Forest Insect Work Conference, DC, June 3, 2016

David Showalter

Ohio State Univ.

Kenneth Raffa Univ. Wisconsin

Daniel Herms Ohio State Univ.

Richard Sniezko

USDA-FS DGRC Sandy Liebhold

USDA-FS NRS

165

• Identification and utilization of host

resistance is essential

– for effective, feasible, long-term

management of

– select tree-killing pests

– for which top-down control cannot work

• Non destructive, quick tools for

screening are necessary

• Early and sustained support is required

Main points

166

Tree-killing Phytophagous Insects and

Phytopathogens (PIPs)

• intimately and cryptically associated with hosts

• damage high fitness value host tissue

• kill a large proportion of naïve host trees

• e.g. canker, wilt and rust fungi, bark and wood

borers

Tree-Killing PIPs

Laurel wilt Emerald ash borer White pine blister rust

167 Adapted from Blackburn et al 2011. Trends in Ecol and Evol.

26:7

Invasion Progression

Stages

PIP

Source

Barriers

Geo

gra

phy A

2

Establishment

Dis

pe

rsa

l

C

3

Spread

4

Outbreak Host

Tre

e D

efe

nses

PIP

Natu

ral E

nem

ies

Abio

tic E

nvironm

ent

D

Surv

ival and

Re

pro

ductio

n

B

1 Introduction

to Naïve

Ecosystem

Management

Prevention

Eradication

Containment

Mitigation

Our focus is on established PIPs

168

Disease/Pest Triangle

Host Pathogen/P

est

Environment

Amount

of

Disease

/Pest

Damage

169

170

Available approaches

I. Short-term ecosystem maintenance

Millar and Stephenson 2015. Science. 349:6250

II. Long-term ecosystem transition

171

…with tree-killing PIPs, those that: • are cryptically associated with their hosts (extremely difficult to detect and eradicate)

• are intimately associated with their hosts (facilitates exchange of molecular signals)

• damage high fitness value host tissue (low damage tolerance, short acceptable lag for PIP control)

• kill a large proportion of naïve host trees (or coevolved trees with compromised defenses)

Host resistance is effective...

172

173

Modern Host Resistance Programs

• Trait Discovery

– Selection and screening of available germplasm

• Trait Development

– Breeding or genetic engineering to combine traits

– Screening/ verification of continued selections

– Mechanistic basis, interactions

• Trait Deployment

– Incorporating heterogeneity for durability/ resilience

174

Feasibility of Modern Trait Discovery and Development

• Marker-assisted selection/ molecular

breeding – reduce time and labor cost of phenotyping continued selections

– genetic, genomic, transcriptomic, chemical markers

– enables non-destructive screening of naïve populations,

informing management

Harper et al 2016. Sci. Rep. 6:19335

175

Feasibility of Modern Trait Discovery and Development

• Mechanistic understanding of resistance traits though manipulative studies – facilitates development and deployment

• Cisgenesis – rapid and controlled trait incorporation

– potentially more widely acceptable than transgenesis

• Transgenesis – rapid and controlled trait incorporation

– dramatically expands germplasm from which resistance traits can be drawn

– See chestnut blight resistance example provided by Bill Powell’s group at SUNY ESF

176

(Emerging) Effectiveness of Modern Resistance Deployment

• Understanding of resistance durability – Combining diverse quantitative and qualitative

mechanisms across time and/or space

– Guided by assessments of PIP evolutionary potential

• Associational/ landscape resistance concepts – May allow for deployment of genetically

diverse resilient populations vs. only resistant individuals

– Includes other forms of heterogeneity • Stand structure/age, species composition

177

Deployment of host resistance is feasible and essential for:

• effective

• long-term management (forest transitions)

of select tree-killing pests, such as Phytophthora ramorum

Conclusion

178

COAST LIVE OAK SUSCEPTIBILITY VARIES

External canker length measured 10 months following coast

live oak inoculation with P. ramorum (N = 154).

Resistant

Susceptible

Google

The Sudden Oak Death Case

179

CANKER LENGTH PREDICTS SURVIVAL

External canker length measured 9 months following inoculation can be used to

predict coast live oak survival 7 years following inoculation (McPherson et al., 2014).

Meanwhile, in Marin County…

180

• Ellagic acid and a tyrosol derivative are associated with

resistant CLO (Nagle et al., 2011).

• Concentrations of 4 putative phenolic biomarkers of

resistance were identified from asymptomatic tissue of

already infected CLO (McPherson et al., 2014).

• Ellagic acid and crude methanol extract from CLO

phloem tissue both inhibit the growth of P. ramorum in

vitro (McPherson et al., 2014).

PHYTOCHEMICALS AND DEFENSE

181

PHLOEM PHENOLICS PREDICT RESISTANCE

Relationship between resistance and selected putative phenolic biomarkers of resistance. The plot shows the estimated probability of resistance and logit values. The probability of resistance is greater than 80% when logit values are greater than 1.39 (dashed line).

From: McPherson et al., 2014

182

• Test the feasibility and efficacy of Fourier Transform-

Infrared (FT-IR) spectroscopy for discriminating

between resistant and susceptible coast live oaks.

OBJECTIVE

183

• Measures light absorbance for a range of wavelengths.

• Functional groups, like those found in phytochemicals, have

characteristic FT-IR spectra shape and position.

• Variation in intensity and presence or absence of certain

spectral bands can be used to distinguish between samples.

• Rapid, reproducible, and non-destructive.

• Predictive models are “easily” developed using commercially

available chemometric software.

WHY FT-IR IS USEFUL

184

DISEASE PHENOTYPES REVISITED

Only trees classified as

resistant or susceptible in

2012 were used to build

the FT-IR model

Resistant Susceptible

Resistant CLO (n = 22) have significantly smaller

canker lengths than susceptible CLO (n = 24)

(independent t-test, P < 0.001).

Back to Briones…

185

Fourier-Transform IR spectroscopy Vibrational spectroscopy-based technique exploits asymmetric

molecular stretching, vibration, and rotation of chemical bonds

as they are exposed to IR radiation

186

Fourier-Transform IR spectroscopy

Chemical fingerprint data can be

analyzed using various chemometric

methods, such as PCA, SIMCA or PLSR

187

IMPORTANT SPECTRAL REGIONS IDENTIFED

Soft independent modeling of class analogy (SIMCA) was used to identify regions of spectrum

that differed between resistant and susceptible CLO and for developing a model for predicting

tree resistance. Only extracts from trees classified as resistant or susceptible in 2012 (N = 46)

were used for this analysis.

188

SIMCA DISCRIMINATES BETWEEN RESISTANT AND SUSCEPTIBLE CLO

LEFT SIMCA 3-D class projection plot. Dashed lines indicate 95% CI for each group.

RIGHT Coomans plot from 4 factor SIMCA analysis. Dashed lines indicate critical sample residual threshold.

Interclass distance = 2.4

Resistant

Susceptible

189

• FT-IR identified two regions, corresponding primarily to carbonyl group vibrations, that were important for identifying resistant trees.

• Spectral differences may be associated with phenolic compounds, e.g. quercetin and ellagic acid.

• 100% of extracts from resistant trees (n = 24) and 100% of extracts from susceptible trees (n = 36) were correctly classified, with an interclass distance of 2.4

(the larger the interclass distance, the less likely samples will be classified as both resistant and susceptible by the SIMCA model)

• The SIMCA model can be used in the future to predict resistance of naïve trees.

FT-IR, PHYTOCHEMICALS, AND

ESTIMATES OF RESISTANCE

190

• Resistant CLOs constituted 16% of the naïve Briones

population (14% based on disease expression after

inoculation, i.e. canker lengths)

• In a prior study in Marin County (McPherson et al. 2014),

and based on phenolic biomarkers, we estimated that

25-30% of RESIDUAL trees (i.e. after much of the

epidemic in the 90s/early 2000s) were resistant

FT-IR, PHYTOCHEMICALS, AND

ESTIMATES OF RESISTANCE

191

0

20

40

60

80

100

2000 2005 2010 2015

Pe

rcen

tag

e K

ille

d

Coast Live Oak Mortality for 18 Marin County Plots

192

• FT-IR spectroscopy coupled with chemometric analysis can identify

resistant CLO.

• Implementation of handheld FT-IR or Raman devices may make in-

field identification of resistant CLO a reality in the future.

NEW APPROACHES TO ASSESS RESISTANCE

193

Next Steps

194

• Using these approaches, naïve CLO could be screened for

resistance to P. ramorum relatively quickly and without a need for

inoculation.

• Resistance could then be mapped on the landscape to improve

epidemiological understanding and lead to the formulation of

rational management plans

NEW APPROACHES TO ASSESS RESISTANCE

195

RedwoodPark,bleeding+killedcoastliveoaks,2011 Peakheightandredcolorrepresentdiseaseintensity

Joshua O’Neill, MS Thesis, UC Berkeley

196

Funding supporting our work

• USDA FS FHP - Conducting Activities Related to

Monitoring, Extension, Management and Mitigation of the

Sudden Oak Death Disease Caused by Phytophthora

ramorum

• OARDC SEEDS Program – A new tool for the rapid

identification of pest-resistant trees

197

THANK YOU!