Science 14Feb2020

predicted to recognize 96.6% ofP. syringae
strains (72.7% excluding AvrE), whereas just
ZAR1 and CAR1 can potentially recognize 94.7%
of strains (43.9% excluding AvrE). Further, 68.0%
ofstrainshavemultipleresistanceNLRsstacked
against them.

A single ETI response can determine
host accessibility

Finally, we tested whether an individual NLR
can limit the ability ofP. syringaestrains to
grow onA. thalianabased on the presence
of their cognate effectors. When analyzing
the distribution of ETI-eliciting effectors, we
observed that the strongA. thalianapathogen
PmaES4326 and the closely related strain (based
on the core genome phylogeny)P. syringae
pv.maculicolaYM7930 (PmaYM7930; also rad-
ish pathogen) differed in their suites of ETI-
eliciting effector profiles by only the presence
of HopAR1 in PmaYM7930; this observation
provided a model for testing whether ETI
can determine host accessibility. Spray inoc-
ulation growth assays showed that PmaYM7930
produces negligible disease symptoms and
reaches a final in planta density approxi-
mately two orders of magnitude lower than
PmaES4326 onA. thalianaCol-0 (Fig. 4). We
hypothesized that the presence of HopAR1 in
PmaYM7930, which is recognized byA. thaliana
NLR RPS5 ( 35 ), was the major cause of this
limited virulence. When we infected PmaYM7930
onA. thalianaCol-0rps5mutant plants, we
found that in planta growth of this strain was
elevated to the level of PmaES4326, with a

corresponding increase in disease symptoms

(Fig. 4).

Although these findings illustrate that the

loss of the HopAR1 ETI-eliciting effector enables

PmaYM7930 to achieve the same high level

ofvirulenceasPmaES4326,thisdoesnotappear

to be the case for strains that are more distantly

related to knownA. thalianapathogens. For

example,P. syringaepv.maculicolaICMP2744

(PmaICMP2744, phylogroup 1 mustard pathogen)

andP. syringaepv.cannabina(PcbICMP2821,

phylogroup 5 hemp pathogen) both carry only

one putative ETI-eliciting effector: PmaICMP2744

carries AvrRpm1, whichis recognized by RPM1;

PcbICMP2821 carries HopA1, which is recognized

by RPS6. Growth assays with PmaICMP2744

onA. thaliana rpm1mutant plants and

PcbICMP2821 onA. thaliana rps6mutant plants

were compared to PtoDC3000 and PmaES4326,

which are in phylogroups 1 and 5, respec-

tively, but divergent from the tested strains.

PmaICMP2744 showed a marginal but signif-

icant increase in bacterial growth when grown

onrpm1mutant plants, whereas no difference

was observed with PcbICMP2821 growth on

wild-type andrps6mutant plants (fig. S15).

Collectively, these results support the idea that

a single NLR-effector interaction has the poten-

tial to change the outcome of a specific host-

pathogen interaction, but this may not be

sufficient to determine host accessibility.

Conclusions

Despite decades of research, we still have only

a limited understanding of the factors that

determine plant pathogen host specificity and

range, particularly from a species-wide per-

spective. The foundation of host resistance

and ETI rests on cultivar- or accession-specific

interactions ( 7 ), which can be difficult to re-

concile with our current understanding of pa-

thogen diversity. For example, our genomic

analysis of 494P. syringaestrains has iden-

tified 4636 unique effector protein sequences,

with nearly every strain carrying a distinct

suite of effectors, whereas the core genome

showed pairwise synonymous substitution

ratesashighas1.0betweenthemostdivergent

strains ( 9 ).

Our PsyTEC library leverages a saturated,

pan-genome effector analysis to reveal the per-

vasive role of ETI mediated by a very small

number of NLRs that counter a highly diverse

and globally important plant pathogen. We

have also revealed the underlying genetic basis

of these interactions in both the host and the

pathogen, yielding numerous new targets for

agricultural engineering of broadly resistant

crops.

TheconvergenceofETIresponsesonasmall

suite of NLRs is surprising. For example, CAR1,

identified in this study, is an NLR that responds

to effectors from theP. syringaeconserved ef-

fector locus. This NLR has the potential to be

particularly potent because it is putatively

capable of recognizing AvrE alleles found in

351 of the 494P. syringaestrains used in this

study (most other strains have AvrE alleles

that do not elicit ETI in this system). Like-

wise, ZAR1 is a powerful antagonist against

Laflammeet al.,Science 367 , 763–768 (2020) 14 February 2020 5of6

Fig. 3. NLR specificity for each ETI-eliciting
effector family.(A) Representative plant images
after bacteria expressing each ETI-eliciting effector
(left) were spray-inoculated onto wild-type (WT) or
mutantA. thalianaCol-0 plants lacking a single NLR
(top). Red boxes indicate loss-of-ETI interactions
(HopF1r was formerly HopF2a). (B) Heat map of the
plant disease scores calculated as proportion of
yellow chlorotic plant tissue in each treatment ( 22 ).
Each effector was sprayed on a total of six
correspondingA. thalianaNLR mutant plants (table
S8) and sixA. thalianaCol-0 wild-type plants. The
proportion of yellow tissue in the NLR mutant plants
corresponding to each cognate ETI-eliciting effector
was significantly greater than the proportion of
yellow tissue in the wild-type plants in all cases.
**P< 0.01, ***P< 0.001 (Studentttests with
Holm-Bonferroni multiple test correction). (C) Empty
vector (EV) levels of in planta bacterial growth were
restored for all novel effector-NLR combinations
when strains harboring ETI-eliciting effectors were
grown in plants lacking their cognate NLR. Error
bars represent SE across eight replicates. Letters
above bars represent significance groups at
P< 0.05 (Studentttests with Holm-Bonferroni
multiple test correction).

RESEARCH | RESEARCH ARTICLE