The Automation of Science

Ross D. King, et al.

Science 324, 85 (2009);
DOI: 10.1126/science.1165620

The following resources related to this article are available online at

www.sciencemag.org (this information is current as of August 20, 2009 ):

Updated information and services, including high-resolution figures, can be found in the online
version of this article at:
http://www.sciencemag.org/cgi/content/full/324/5923/85

Supporting Online Material can be found at:

http://www.sciencemag.org/cgi/content/full/324/5923/85/DC1
A list of selected additional articles on the Science Web sites related to this article can be
found at:
http://www.sciencemag.org/cgi/content/full/324/5923/85#related-content
This article cites 16 articles, 5 of which can be accessed for free:

Downloaded from www.sciencemag.org on August 20, 2009

http://www.sciencemag.org/cgi/content/full/324/5923/85#otherarticles

This article has been cited by 1 articles hosted by HighWire Press; see:
http://www.sciencemag.org/cgi/content/full/324/5923/85#otherarticles

This article appears in the following subject collections:

Computers, Mathematics
http://www.sciencemag.org/cgi/collection/comp_math

Information about obtaining reprints of this article or about obtaining permission to reproduce
this article in whole or in part can be found at:
http://www.sciencemag.org/about/permissions.dtl

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the
American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright
2009 by the American Association for the Advancement of Science; all rights reserved. The title Science is a
registered trademark of AAAS.
 REPORTS
We have demonstrated the discovery of 3. J. Hanc, S. Tuleja, M. Hancova, Am. J. Phys. 72, 428 19. Y. Liang, B. Feeny, Nonlinear Dyn. 52, 181 (2008).
physical laws, from scratch, directly from ex- (2004). 20. M. Mor, A. Wolf, O. Gottlieb, in Proceedings of the 21st
4. D. Clery, D. Voss, Science 308, 809 (2005). ASME Biennial Conference on Mechanical Vibration and
perimentally captured data with the use of a 5. A. Szalay, J. Gray, Nature 440, 413 (2006). Noise (ASME Press, Las Vegas, NV, 2007), pp. 1–8.
computational search. We used the presented 6. R. E. Valdés-Pérez, Commun. Assoc. Comput. Mach. 42, 21. P. Gregory, R. Denis, F. Cyril, in Evolution Artificielle,
approach to detect nonlinear energy conservation 37 (1999). 6th International Conference, vol. 2936, L. Pierre,
laws, Newtonian force laws, geometric invari- 7. R. D. King et al., Nature 427, 247 (2004). C. Pierre, F. Cyril, L. Evelyne, S. Marc, Eds. (Springer,
8. P. Langley, Cogn. Sci. 5, 31 (1981). Marseilles, France, 2003), pp. 267–277.
ants, and system manifolds in various synthetic 9. R. M. Jones, P. Langley, Comput. Intell. 21, 480 22. E. D. De Jong, J. B. Pollack, in Genetic Programming and
and physically implemented systems without (2005). Evolvable Machines, vol. 4 (Springer, Berlin, 2003),
prior knowledge about physics, kinematics, or 10. J. R. Koza, Genetic Programming: On the Programming of pp. 211–233.
geometry. The concise analytical expressions that Computers by Means of Natural Selection. (MIT Press, 23. S. H. Strogatz, Nature 410, 268 (2001).
Cambridge, MA, 1992). 24. P. A. Marquet, Nature 418, 723 (2002).
we found are amenable to human interpretation 11. S. Forrest, Science 261, 872 (1993). 25. This research was supported in part by Integrative Graduate
and help to reveal the physics underlying the 12. J. Duffy, J. Engle-Warnick, Evolutionary Computation in Education and Research Traineeship program in nonlinear
observed phenomenon. Many applications exist Economics and Finance 100, 61 (2002). systems, a U.S. NSF graduate research fellowship, and NSF
for this approach, in fields ranging from systems 13. F. Cyril, B. Alberto, in 2007 IEEE Congress on Creative-IT grant 0757478 and CAREER grant 0547376.
Evolutionary Computation, S. Dipti, W. Lipo, Eds. We thank M. Kurman for editorial consultation and
biology to cosmology, where theoretical gaps
(IEEE Press, Singapore, 2007), pp. 23–30. substantive editing of the manuscript.
exist despite abundance in data. 14. B. Elena, B. Andrei, L. Henri, in Seventh International
Might this process diminish the role of future Symposium on Symbolic and Numeric Algorithms for Supporting Online Material
Scientific Computing (SYNASC '05) (IEEE Press, 2005), www.sciencemag.org/cgi/content/full/324/5923/81/DC1
scientists? Quite the contrary: Scientists may use Materials and Methods
pp. 321–324.

Downloaded from www.sciencemag.org on August 20, 2009

processes such as this to help focus on interesting SOM Text
15. J. Bongard, H. Lipson, Proc. Natl. Acad. Sci. U.S.A. 104,
phenomena more rapidly and to interpret their 9943 (2007). Figs. S1 to S7
meaning. 16. S. Nee, N. Colegrave, S. A. West, A. Grafen, Science 309, Tables S1 to S3
1236 (2005). References
Movie S1
References and Notes 17. P. Jäckel, T. Mullin, Proc. R. Soc. London Ser. A 454,
3257 (1998). Data Sets S1 to S15
1. P. W. Anderson, Science 177, 393 (1972).
2. E. Noether, Nachr. d. König Gesellsch. d. Wiss. zu 18. T. Shinbrot, C. Grebogi, J. Wisdom, J. A. Yorke, 15 September 2008; accepted 19 February 2009
Göttingen, Math-Phys. Klasse 235 (1918). Am. J. Phys. 60, 491 (1992). 10.1126/science.1165893

The Automation of Science experiments can be executed, each individual ex-

periment cannot be designed to test a hypothesis
about a model. Robot scientists have the potential
Ross D. King,1* Jem Rowland,1 Stephen G. Oliver,2 Michael Young,3 Wayne Aubrey,1 to overcome this fundamental limitation.
Emma Byrne,1 Maria Liakata,1 Magdalena Markham,1 Pınar Pir,2 Larisa N. Soldatova,1 The complexity of biological systems neces-
Andrew Sparkes,1 Kenneth E. Whelan,1 Amanda Clare1 sitates the recording of experimental metadata in
as much detail as possible. Acquiring these meta-
The basis of science is the hypothetico-deductive method and the recording of experiments in data has often proved problematic. With robot
sufficient detail to enable reproducibility. We report the development of Robot Scientist “Adam,” scientists, comprehensive metadata are produced
which advances the automation of both. Adam has autonomously generated functional genomics as a natural by-product of the way they work.
hypotheses about the yeast Saccharomyces cerevisiae and experimentally tested these hypotheses Because the experiments are conceived and ex-
by using laboratory automation. We have confirmed Adam’s conclusions through manual ecuted automatically by computer, it is possible
experiments. To describe Adam’s research, we have developed an ontology and logical language. to completely capture and digitally curate all as-
The resulting formalization involves over 10,000 different research units in a nested treelike pects of the scientific process (11, 12).
structure, 10 levels deep, that relates the 6.6 million biomass measurements to their logical To demonstrate that the robot scientist meth-
description. This formalization describes how a machine contributed to scientific knowledge. odology can be both automated and be made
effective enough to contribute to scientific knowl-
omputers are playing an ever-greater role advantages of logic, most scientific knowledge is edge, we have developed Robot Scientist “Adam”

C in the scientific process (1). Their use to

control the execution of experiments con-
tributes to a vast expansion in the production of
expressed only in natural languages. This is now
changing through developments such as the
Semantic Web (4) and ontologies (5).
(13) (Fig. 1). Adam’s hardware is fully automated
such that it only requires a technician to period-
ically add laboratory consumables and to remove
scientific data (2). This growth in scientific data, A natural extension of the trend to ever-greater waste. It is designed to automate the high-
in turn, requires the increased use of computers computer involvement in science is the concept of throughput execution of individually designed
for analysis and modeling. The use of computers a robot scientist (6). This is a physically imple- microbial batch growth experiments in micro-
is also changing the way that science is described mented laboratory automation system that exploits titer plates (14). Adam measures growth curves
and reported. Scientific knowledge is best ex- techniques from the field of artificial intelligence (phenotypes) of selected microbial strains (geno-
pressed in formal logical languages (3). Only (7–9) to execute cycles of scientific experimenta- types) growing in defined media (environments).
formal languages provide sufficient semantic tion. A robot scientist automatically originates Growth of cell cultures can be easily measured in
clarity to ensure reproducibility and the free hypotheses to explain observations, devises exper- high-throughput, and growth curves are sensitive
exchange of scientific knowledge. Despite the iments to test these hypotheses, physically runs the to changes in genotype and environment.
experiments by using laboratory robotics, inter- We applied Adam to the identification of
1
Department of Computer Science, Aberystwyth University, prets the results, and then repeats the cycle. genes encoding orphan enzymes in Saccharomy-
SY23 3DB, UK. 2Cambridge Systems Biology Centre, Depart- High-throughput laboratory automation is trans- ces cerevisiae: enzymes catalyzing biochemical
ment of Biochemistry, University of Cambridge, Sanger forming biology and revealing vast amounts of reactions thought to occur in yeast, but for which
Building, 80 Tennis Court Road, Cambridge CB2 1GA, UK.
3
Institute of Biological, Environmental and Rural Sciences, new scientific knowledge (10). Nevertheless, ex- the encoding gene(s) are not known (15). To set
Aberystwyth University, SY23 3DD, UK. isting high-throughput methods are currently in- up Adam for this application required (i) a
*To whom correspondence should be addressed. E-mail: adequate for areas such as systems biology. This comprehensive logical model encoding knowl-
[email protected] is because, even though very large numbers of edge of S. cerevisiae metabolism [~1200 open

www.sciencemag.org SCIENCE VOL 324 3 APRIL 2009 85

REPORTS
reading frames (ORFs), ~800 metabolites] (15), Adam formulated and tested 20 hypotheses The reason that Adam considered them to be
expressed in the logic programming language concerning genes encoding 13 orphan enzymes orphans was due to the use of an incomplete bio-
Prolog; (ii) a general bioinformatic database of (16) (Table 1). The weight of the experimental informatic database. These six genes therefore
genes and proteins involved in metabolism; (iii) evidence for the hypotheses varied (based on ob- constitute a positive control for Adam's meth-
software to abduce hypotheses about the genes servations of differential growth), but 12 hypothe- odology. A possible error was also revealed
encoding the orphan enzymes, done by using a ses with no previous evidence were confirmed (Table 1) (SOM).
combination of standard bioinformatic software with P < 0.05 for the null hypothesis. To better understand the reasons why the
and databases; (iv) software to deduce experi- Because Adam’s experimental evidence for its identity of the genes encoding these enzymes has
ments that test the observational consequences of conclusions is indirect, we tested Adam’s conclu- remained obscure for so long, we investigated
hypotheses (based on the model); (v) software to sions with more direct experimental methods. The their comparative genomics in detail (16). The
plan and design the experiments, which are based enzyme 2-aminoadipate:2-oxoglutarate amino- likely explanation is a combination of three com-
on the use of deletion mutants and the addition of transferase (2A2OA) catalyzes a reaction in the plicating factors: gene duplications with retention
selected metabolites to a defined growth medium; lysine biosynthetic pathways of fungi. Adam hy- of overlapping function, enzymes that catalyze
(vi) laboratory automation software to physically pothesized that three genes (YER152C, YJL060W, more than one related reaction, and existing func-
execute the experimental plan and to record the and YGL202W) encode this enzyme and ob- tional annotations. Adam’s systematic bioinformatic
data and metadata in a relational database; (vii) served results consistent with all three hypotheses and quantitative phenotypic analyzes were required
software to analyze the data and metadata (gen- (Table 1). To test Adam’s conclusions, we pu- to unravel this web of functionality.
erate growth curves and extract parameters); and rified the protein products of these genes and Use of a robot scientist enables all aspects of a

Downloaded from www.sciencemag.org on August 20, 2009

(viii) software to relate the analyzed data to the used them in in vitro enzyme assays, which scientific investigation to be formalized in logic.
hypotheses; for example, statistical methods are confirmed Adam’s conclusions [supporting on- For the core organization of this formalization,
required to decide on significance. Once this in- line material (SOM)] (Fig. 2). we used the ontology of scientific experiments:
frastructure is in place, no human intellectual inter- To further test Adam's conclusions, we ex- EXPO (11, 12). This ontology formalizes generic
vention is necessary to execute cycles of simple amined the scientific literature on the 20 genes knowledge about experiments. For Adam, we
hypothesis-led experimentation. [For more details investigated (Table 1) (16). This revealed the ex- developed LABORS, a customized version of
of the software, and its application to a related istence of strong empirical evidence for the cor- EXPO, expressed in the description logic lan-
functional genomics problem, see (16) and figs. rectness of six of the hypotheses; that is, the guage OWL-DL (17). Application of LABORS
S1 and S2]. enzymes were not actually orphans (Table 1). produces experimental descriptions in the logic-

Fig. 1. The Robot Scien-

tist Adam. The advances
that distinguish Adam from
other complex laboratory
systems are the individual
design of the experiments
to test hypotheses and the
utilization of complex in-
ternal cycles. Adam’s basic
operations are selection of
specified yeast strains from
a library held in a freezer,
inoculation of these strains
into microtiter plate wells
containing rich medium,
measurement of growth
curves on rich medium,
harvesting of a defined
quantity of cells from each
well, inoculation of these
cells into wells containing
defined media (minimal syn-
thetic dextrose medium plus
up to four added metab-
olites from a choice of six),
and measurement of growth
curves on the specified me-
dia. To achieve this func-
tionality, Adam has the
following components: a,
an automated –20°C freezer;
b, three liquid handlers (one
of which can separately control 96 fluid channels simultaneously); c, three strain and defined-growth-medium experiments each day (from a selection of
automated +30°C incubators; d, two automated plate readers; e, three robot thousands of yeast strains), with each experiment lasting up to 5 days. The
arms; f, two automated plate slides; g, an automated plate centrifuge; h, an design enables measurement of OD595nm for each experiment at least once
automated plate washer; i, two high-efficiency particulate air filters; and j, a every 30 min (more often if running at less than full capacity), allowing ac-
rigid transparent plastic enclosure. There are also two bar code readers, seven curate growth curves to be recorded (typically we take over a hundred mea-
cameras, 20 environment sensors, and four personal computers, as well as the surements a day per well), plus associated metadata. See the supporting
software. Adam is capable of designing and initiating over a thousand new online material for pictures and a video of Adam in action.

86 3 APRIL 2009 VOL 324 SCIENCE www.sciencemag.org

 REPORTS
programming language Datalog (18). In the course mentation involving over 10,000 different research and takes up 366 Mbytes (16). Making such
of its investigations, Adam observed 6,657,024 units (segments of experimental research). This experimental structures explicit renders scien-
optical density (OD595nm) measurements (form- has a nested treelike structure, 10 levels deep, that tific research more comprehensible, reproduc-
ing 26,495 growth curves). These data are held in logically connects the experimental observations ible, and reusable. This paper may be considered
a MySQL relational database. Use of LABORS to the experimental metadata. (Fig. 3). This struc- as simply the human-friendly summary of the
resulted in a formalization of the scientific argu- ture resembles the trace of a computer program formalization.

Table 1. The orphan enzymes and Adam’s hypotheses. The hypothesized of metabolites tested. Existing annotation is the summary from the
genes are those which Adam abduced encoded an orphan enzyme. Prob. Saccharomyces Genome Database of the annotation of the ORF. Dry is the
is Adam’s Monte Carlo estimate of the probability of obtaining the summary of whether the annotated function is the same as predicted by
observed discrimination accuracy or better with a random labeling of Adam. If a gene already has an associated function, we do not consider
replicates. The discrimination is between the differences in growth curves this to be contradictory to Adam’s conclusions unless this function is
observed with the addition of specified metabolites to the wild type and capable of explaining the observed growth phenotype, for example, BCY1.
the deletant. Acc. is the highest accuracy for a metabolite species in ida indicates inferred from direct assay and iss, inferred from sequence or
discriminating between the growth curves observed with the addition of structural similarity (5). Wet is the result of our manual enzyme assays.
specified metabolites to the wild type and the deletant. No. is the number See (16) for details.

Hypothesized
Orphan enzyme Prob. Acc. No. Existing annotation Dry Wet

Downloaded from www.sciencemag.org on August 20, 2009

gene
Glucosamine-6-phosphate YHR163W <10−4 97 8 6-Phosphogluconolactonase, ida – –
deaminase (3.5.99.6) (SOL3)
Glutaminase (3.5.1.2) YIL033C <10−4 92 11 Cyclic adenosine 3´,5´- ✗ –
(BCY1) monophosphate (cAMP)–
dependent protein kinase
inhibitor, ida
L-Threonine 3- YDL168W <10−4 83 6 Alcohol dehydrogenase, ida – –
dehydrogenase (SFA1)
(1.1.1.103)
Purine-nucleoside YLR209C <10−4 82 11 Purine-nucleoside ✓ –
phosphorylase (2.4.2.1) (PNP1) phosphorylase, ida
2-Aminoadipate YGL202W <10−4 80 3 Aromatic–amino acid ✓ ✓
transaminase (2.6.1.39) (ARO8) transaminase, ida
5,10-Methenyltetrahydrofolate YER183C <10−4 80 4 5,10 Formyltetrahydrofolate ✓ –
synthetase (6.3.3.2) (FAU1) cyclo-ligase, ida
Glucosamine-6-phosphate YNR034W <10−4 79 2 Possible role in tRNA export – –
deaminase (3.5.99.6) (SOL1)
Pyridoxal kinase (2.7.1.35) YPR121W <10−4 78 1 Phosphomethylpyrimidine – –
(THI22) kinase, iss
Mannitol-1-phosphate YNR073C <10−4 78 6 Putative mannitol – –
5-dehydrogenase (1.1.1.17) dehydrogenase, iss
1-Acylglycerol-3-phosphate YDL052C 0.0001 80 6 1-Acylglycerol-3-phosphate ✓ –
O-acyltransferase (SLC1) O-acyltransferase ida
(2.3.1.51)
Glucosamine-6-phosphate YGR248W 0.0002 78 2 6-Phosphogluconolactonase, ida – –
deaminase (3.5.99.6) (SOL4)
Maleylacetoacetate YLL060C 0.0003 76 3 Glutathione S-transferase, ida – –
isomerase (5.2.1.2) (GTT2)
Serine O-acetyltransferase YJL218W 0.0005 78 2 Unknown function – –
(2.3.1.30)
L-Threonine YLR070C 0.0052 75 6 Xylitol dehydrogenase, ida – –
3-dehydrogenase (XYL2)
(1.1.1.103)
2-Aminoadipate YJL060W 0.0084 73 3 Kynurenine – ✓
transaminase (2.6.1.39) (BNA3) aminotransferase, ida
Pyridoxal kinase (2.7.1.35) YNR027W 0.0259 76 2 Involved in bud-site – –
selection, iss
Polyamine oxidase YMR020W 0.0289 78 4 Polyamine oxidase, ida ✓ –
(1.5.3.11) (FMS1)
2-Aminoadipate YER152C 0.0332 74 3 Uncharacterized – ✓
transaminase (2.6.1.39)
L-Aspartate oxidase YJL045W 0.1300 72 1 Succinate dehydrogenase – –
(1.4.3.16) isozyme, iss
Purine-nucleoside YLR017W 0.1421 72 6 Methylthioadenosine ✓ –
phosphorylase (2.4.2.1) (MEU1) phosphorylase, ida

www.sciencemag.org SCIENCE VOL 324 3 APRIL 2009 87

REPORTS
Fig. 2. Assay results for 2A2OA activ- 1.2
A major motivation for the formalization of
control YDL168W
ity. The proteins encoded by YGL202W, experimental knowledge is the expectation that
no protein added
YJL060W, YER152C, and YDL168W were 1 such knowledge is more easily reused to answer
YGL202W
expressed from OpenBiosystems (www. other scientific questions. To test this, we investi-

Absorbance at 340nm
0.8
openbiosystems.com) yeast ORF clones YER152C
gated whether we could reuse Adam’s functional
and purified. Activity was tested in an 0.6 YJL060W
genomic research (16). An example question
assay of NADPH (reduced form of nico- investigated was the relative growth rates (mmax)
0.4
tinamide adenine dinucleotide phosphate) in rich and defined media of the deletion strains
production based on (22). L-a-aminoadipic 0.2
compared with those of the wild type. What
acid and 2-oxoglutarate were provided was observed, in both media, was a skewed dis-
0
as substrates and pyridoxal phosphate
tribution, with a few deletants having a much
as cofactor. Glutamate production was -0.2
lower mmax than that of the wild type, but most
assayed by using commercially available 0 5 10 15 20 25

yeast glutamate dehydrogenase, which Time (min)

having a slightly higher mmax. These observations
uses NADP as cofactor and deaminates question the common assumption that wild-type
glutamate, producing ammonia and NADPH and regenerating 2-oxoglutarate (16). Also consistent with S. cerevisiae is optimized for mmax and provide
2A2OA activity is experimental evidence indicating a higher activity with L-a-aminoadipic acid over either quantitative test data for yeast systems biology
alanine or aspartate (16). models (19).
It could be argued that the scientific knowl-
edge “discovered” by Adam is implicit in the

Downloaded from www.sciencemag.org on August 20, 2009

formulation of the problem and is therefore not
novel. This argument that computers cannot
originate anything is known as Lady Lovelace’s
objection (20): “The Analytical Engine has no
pretensions to originate anything. It can do
whatever we know how to order it to perform”
(her italics). We accept that the knowledge
automatically generated by Adam is of a modest
kind. However, this knowledge is not trivial, and
in the case of the genes encoding 2A2OA, it
sheds light on, and perhaps solves, a 50-year-old
puzzle (21).
Adam is a prototype and could be greatly
improved. Its hardware and software are “brittle,”
so although Adam is capable of running for a few
days without human intervention, it is advisable
to have a technician nearby in case of problems.
The integration of Adam’s artificial intelligence
(AI) software also needs to be enhanced so that it
works seamlessly. To extend Adam, we have de-
veloped software to enable external users to pro-
pose hypotheses and experiments, and we plan to
automatically publish the logical descriptions of
automated experiments. The idea is to develop a
way of enabling teams of human and robot sci-
entists to work together. The greatest research
challenge will be to improve the scientific in-
telligence of the software. We have shown that a
simple form of hypothesis-led discovery can be
automated. What remain to be determined are the
limits of automation.

References and Notes

1. Nature 440, 383 (2006).
2. T. Hey, A. Trefethen, in Grid Computing: Making the Global
Infrastructure a Reality, 36 809 (Wiley, New York, 2003).
3. S. Toulmin, in Encyclopaedia Britannica Deluxe Edition
2004 CD (Encyclopaedia Britannica UK, London, 2003).
4. T. Berners-Lee, J. Hendler, O. Lassila, Sci. Am. 284, 34
(May 2001).
5. M. Ashburner et al., Nat. Genet. 25, 25 (2000).
Fig. 3. Structure of the Robot Scientist investigation (a fragment). It consists of two main parts: an 6. R. D. King et al., Nature 427, 247 (2004).
investigation into the automation of science and an investigation into the reuse of formalized 7. B. G. Buchanan, G. L. Sutherland, E. A. Feigenbaum, in
experiment information. The top levels involve AI research (red), which requires research in functional Machine Intelligence, B. Meltzer, D. Michie, Eds.
(Edinburgh Univ. Press, Edinburgh, UK, 1969), vol. 5,
genomics (blue) and systems biology (yellow). Each level of research unit (studies, cycles, trials, tests, and pp. 209–254.
replicates) is characterized by a specific set of properties (fig. S3) (16). Such a nested structure is typical of 8. P. Langley, H. A. Simon, G. L. Bradshaw, J. M. Zytkow,
many scientific experiments, where the testing of a top-level hypothesis requires the planning of many Scientific Discovery: Computational Explorations of the
levels of supporting work. What is atypical in Adam’s work is the scale and depth of the nesting. Creative Process (MIT Press, Cambridge, MA, 1987).

88 3 APRIL 2009 VOL 324 SCIENCE www.sciencemag.org

 REPORTS
9. J. M. Zytkow, J. Zhu, A. Hussam, in Proceedings of the 16. Materials and methods are available as supporting Exhibition of 1851, the Engineering and Physical
Eighth National Conference on Artificial Intelligence material on Science Online. Sciences Research Council, and the Royal Academy of
(AAAI-90), 889 (AAAI Press, Menlo Park, CA, 1990). 17. I. Horrocks, P. F. Patel-Schneider, F. van Harmelen, Engineering to A.C.; and by a RC-UK Fellowship to
10. L. Hood, J. R. Heath, M. E. Phelps, B. Lin, Science 306, Web Semantics 1, 7 (2003). L.N.S. We thank D. Struttman for help with yeast and
640 (2004). 18. J. D. Ullman, Principles of Database and Knowledge-Base M. Benway for help with Adam.
11. L. N. Soldatova, R. D. King, J. R. Soc. Interface 3, 795 Systems, Vol. I: Classical Database Systems (Computer
(2006). Science, New York, 1988). Supporting Online Material
12. L. N. Soldatova, A. Clare, A. Sparkes, R. D. King, 19. M. J. Herrgard et al., Nat. Biotechnol. 26, 1155 (2008). www.sciencemag.org/cgi/content/full/324/5923/85/DC1
Bioinformatics 22, e464 (2006). 20. A. Turing, Mind 236, 433 (1950). Materials and Methods
13. R. D. King et al., The Robot Scientist Project, www.aber. 21. T. M. Zabriskie, M. D. Jackson, Nat. Prod. Rep. 17, 85 (2000). Figs. S1 to S3
ac.uk/compsci/Research/bio/robotsci/ (2008). 22. M. Young, J. Exp. Bot. 24, 1172 (1973). Table S1
14. J. Warringer, E. Ericson, L. Fernandez, O. Nerman, 23. This work was funded by grants from the Biotechnology References
A. Blomberg, Proc. Natl. Acad. Sci. U.S.A. 100, 15724 and Biological Sciences Research Council to R.D.K.,
(2003). S.G.O., and M.Y.; by a SRIF 2 award to R.D.K.; by 8 September 2008; accepted 10 February 2009
15. K. E. Whelan, R. D. King, BMC Bioinformatics 9, 97 (2008). fellowships from the Royal Commission for the Great 10.1126/science.1165620

Priming in Systemic Plant Immunity

(JA) accumulates to a high level in petiole exu-
dates from leaves infected with SAR-inducing
bacteria, JA does not seem to be the critical sig-

Downloaded from www.sciencemag.org on August 20, 2009

Ho Won Jung,1 Timothy J. Tschaplinski,2 Lin Wang,3* Jane Glazebrook,3 Jean T. Greenberg1† nal for SAR (5, 6). Instead, SAR and the pro-
duction of active exudates require the DIR1
Plants possess inducible systemic defense responses when locally infected by pathogens. protein, a predicted secreted protein and puta-
Bacterial infection results in the increased accumulation of the mobile metabolite azelaic tive signal carrier in the lipid transfer protein fam-
acid, a nine-carbon dicarboxylic acid, in the vascular sap of Arabidopsis that confers local and ily, and other proteins involved in glycerolipid
systemic resistance against the pathogen Pseudomonas syringae. Azelaic acid primes plants to biosynthesis (2, 3, 7). Additionally, SAR and
accumulate salicylic acid (SA), a known defense signal, upon infection. Mutation of the AZELAIC exudate-induced resistance appears to require the
ACID INDUCED 1 (AZI1) gene, which is induced by azelaic acid, results in the specific loss of phenolic metabolite salicylic acid (SA) (3, 8) and
systemic immunity triggered by pathogen or azelaic acid and of the priming of SA induction in possibly methylsalicylate (MeSA) and its methyl
plants. Furthermore, the predicted secreted protein AZI1 is also important for generating vascular
sap that confers disease resistance. Thus, azelaic acid and AZI1 are components of plant systemic 1
Department of Molecular Genetics and Cell Biology, The Uni-
immunity involved in priming defenses. versity of Chicago, 1103 East 57th Street EBC410, Chicago, IL
60637, USA. 2Oak Ridge National Laboratory, Environmental
Sciences Division, Oak Ridge, TN 37831–6341, USA. 3Depart-
hole plant immunity, called systemic ondary infection (1). Leaves infected with SAR- ment of Plant Biology, Microbial and Plant Genomics Institute,

W acquired resistance (SAR), often de-

velops after localized foliar infections
by diverse pathogens. In this process, leaves dis-
inducing bacteria produce vascular sap, called
petiole exudate, which confers disease resistance
to previously unexposed (naïve) plants (2, 3). This
University of Minnesota, 1500 Gortner Avenue, St. Paul, MN
55108, USA.
*Present address: Boyce Thompson Institute for Plant Research,
Tower Road, Ithaca, NY 14853–1801, USA.
tal to the localized infection become primed to indicates that a mobile systemic signal(s) is involved †To whom correspondence should be addressed. E-mail:
activate a stronger defense response upon sec- in SAR (4). Although the hormone jasmonic acid [email protected]

A B C D
5 mM MES
107 1 mM AzA 107
1 2 3 4
108 106
106
cfu/leaf disc

108
cfu/leaf disc

cfu/leaf disc

cfu/leaf disc

* * * **
107 107 105 *
105 ** ** **
106
106 104
104 105
105 1 2 3 4 M C8 C9 C10
0 1 10 100 1000 6 12 24 48 Local Systemic No. of carbon
Azelaic acid (µM) Exposure period (h) leaf leaf
Fig. 1. Azelaic acid specifically confers resistance to Pseudomonas syringae. (A)
Azelaic acid–induced resistance is concentration-dependent. Plants were sprayed with 1, E 18 35
10, 100, and 1000 mM azelaic acid in 5 mM MES (pH 5.6) or 5 mM MES (pH 5.6) alone 2 16 30
(µg/g fresh weight)

days before infection with P. syringae pv. maculicola strain PmaDG3 (OD600 = 0.0001). 14
(µg/ml exudate)

acid
acid

25
(B) Induced resistance is time-dependent. Plants sprayed with 5 mM MES or 1 mM azelaic 12
10 20
2H-azelaic
2H-azelaic

acid for the time periods indicated were subsequently inoculated with PmaDG3. (C)
5 mM MES or 1 mM azelaic acid was injected into local leaves. Two days later, either 8 15
local or systemic leaves were infected with PmaDG3. (D) Dicarboxylic acids (1 mM) of 6 10
different carbon-chain lengths were applied to Arabidopsis. M, 5 mM Mes; C8, suberic 4 5
acid; C9, azelaic acid; C10, sebacic acid. (E) Mobility of deuterium-labeled azelaic acid 2
0
[HOOC(CD2)7COOH] injected into WT leaves. Azelaic acid amounts were determined 0
18 24 48 72 6 24 48
in petiole exudates (left) and distal leaves (right) after local injection with 1 mM Hours of Hours after
azelaic acid. *P < 0.05; **P < 0. 01; t test. Error bars indicate SE. collection (h) injection (h)

www.sciencemag.org SCIENCE VOL 324 3 APRIL 2009 89