correspondEnce
signals mediated by the known FLuc and RLuc stabilizers, PTC124 and BTS, respectively (Fig. 1b,c and Supplementary Fig. 2). Using the LOPAC1280 chemical library, we conducted a quantitative HTS experiment, in which full titrations of each compound are tested4 to identify potentiators of the CREB pathway (Supplementary Figs. 3 and 4 and Supplementary Table 1). The screen revealed, for example, coincident FLuc and RLuc signal outputs for 17 adenosine analog agonists of endogenous purinergic 2Y known to signal through G proteins in this cell type5 (Supplementary Table 2a). We observed excellent correlation between the half-maximal effective concentration (EC50) values calculated from the orthogonal reporter outputs (Fig. 1d). Our experiments also illustrated the phenomenon of reporterdependent artifacts: we identified five aryl sulfonamides that showed selective agonist activity for RLuc only (Supplementary Table 2b, compounds 20–24). These compounds share a similar core scaffold with two known RLuc inhibitors and selectively inhibit the enzymatic activity of RLuc over that of FLuc (Supplementary Fig. 5), thus tying these particular artifacts to the phenomenon of reporter stabilization4. Cross-section data analysis of the screen (Supplementary Fig. 6) also demonstrates how coincidence detection enhances the testing of compound libraries in single-concentration format. We conclude that coincidence reporter strategies rapidly discriminate compounds of relevant biological activity from those interfering with reporter function and stability, using a single assay platform. This study outlines an approach to improved use of reporter genes in HTS with numerous coincidence combination types and stoichiometries possible.
To the Editor: Originally developed as sentinels of transcriptional activity to map the regulatory function of genetic elements, reportergene assays have been extensively used in high-throughput screening (HTS) to identify chemical modulators of cellular pathways1. However, HTS chemical libraries consist of structurally diverse small molecules that frequently interact directly with the reporter, thus skewing data interpretation and complicating candidate selection. For example, a recent study indicated that >80% of the apparently active compounds from a 500,000-compound HTS were assay artifacts2. For reporter-gene assays, this poses a formidable challenge. To distinguish compounds that target a biological pathway from those that interfere with a reporter, we designed a coincidence ‘biocircuit’ based on the principle that it is easier to tell signal from noise when the signal is reported by two or more detectors. This concept is implemented here using co-translational expression of nonhomologous reporters: for example, proteins having different catalytic, light-emitting or fluorescence properties (Fig. 1a). We confirmed the function of a preliminary biocircuit design by stoichiometric coexpression of the unrelated bioluminescent reporters firefly luciferase (FLuc) and Renilla luciferase (RLuc), using ‘ribosome skipping’ facilitated by the short P2A peptide3 in a HEK293 cell. (Supplementary Fig. 1 and Supplementary Methods). We demonstrated the accurate discrimination of forskolin-activated adenylyl-cyclase signaling through the cAMP-response element from
a
Input (cpd)
R1
R1 R1
2A
R R2
Ye s Ye s No No
FSK PTC124 BTS
10
0 10
8
6
–Log [cpds] (M)
4
Ye s No Ye s No
Noise reduction p1p2 p1 p2
d 5
–pEC 50 FLuc (M)
20
RLuc activity (RLU ×1,000)
c 30
R2
Output
Response
b
Note: Supplementary information is available at http:// www.nature.com/doifinder/10.1038/nmeth.2170.
Direct reporter modulation
Cell pathway modulation
Biocircuit
FLuc activity (RLU ×1,000)
npg
© 2012 Nature America, Inc. All rights reserved.
A coincidence reporter-gene system for high-throughput screening
4 3 2 1 0
10
8
6
– Log [cpds] (M)
4
ACKNOWLEDGMENTS We thank P. Dranchak and R. MacArthur for assistance with supplementary data and P. Shinn for compound management. This work was supported by the US National Institutes of Health (NIH) Roadmap for Medical Research. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests.
Ken C-C Cheng1 & James Inglese1,2
5
1National Center of Advancing Translational Sciences,
NIH, Bethesda, Maryland, USA. 2National Human Genome Research Institute, NIH, Bethesda, Maryland, USA. e-mail:
[email protected]
6
7
7
6
5
–pEC50 RLuc (M)
Figure 1 | Coincidence reporter biocircuit. (a) General scheme, where R1 and R2 are nonhomologous reporter genes giving rise to proteins R1 and R2. Cpd, compound; p, noise probability for respective reporter. (b,c) Response of FLuc (b) and RLuc (c) to treatment with FSK and inhibitors of FLuc (PTC124) and RLuc (BTS). RLU, relative luminescence units. (d) EC50 correlation plot for compounds activating FLuc and RLuc expression equally; r2 = 0.87. Three classes of compounds were identified: purinergic Y2 receptor agonists (black circles); compound 18, a muscarinic receptor agonist (green); and the adenylyl-cyclase activator FSK (red). EC50 values of compounds that selectively increased RLuc (blue) are plotted along the x axis. Data plotted are averages of replicate (n = 2) determinations; error bars, s.d.
1. Michelini, E., Cevenini, L., Mezzanotte, L., Coppa, A. & Roda. A. Anal. Bioanal. Chem. 398, 227–238 (2010). 2. Lyssiotis, C.A. et al. Proc. Natl. Acad. Sci. USA 106, 8912–8917 (2009). 3. de Felipe, P. et al. Trends Biotechnol. 24, 68–75 (2006). 4. Thorne, N., Inglese, J. & Auld, D.S. Chem. Biol. 17, 646–657 (2010). 5. Mundell, S.J. & Benovic, J.L. J. Biol. Chem. 275, 12900–12908 (2000).
nature methods | VOL.9 NO.10 | OCTOBER 2012 | 937
Supplemental Information
A coincidence reporter-gene system for high throughput screening
Ken C.-C. Cheng1 and James Inglese1,2* 1
National Center of Advancing Translational Sciences, 2National Human Genome Research
Institute, NIH, Bethesda, MD 20892
Mailing address: National Center of Advancing Translational Sciences, 9800 Medical Center Drive, Building B, Room 3005, MSC 3370, Rockville, MD 20850-6386
1
Nature Methods: doi:10.1038/nmeth.2170
Supplementary Figure 1 | Design and characterization of FLuc-P2A-RLuc coincidence reporter biocircuit expression and function.
a
pGL3-Control (Promega)
SV40 Promoter
FLuc
GSGATNFSLLKQAGDVEENPGP
b
Promoter pCI-6.20 pCI-6.24
FLuc
P2A
RLuc
SV40 4xCRE
c
d
Supplementary Figure 1 | Design and characterization of FLuc-P2A-RLuc coincidence reporter
biocircuit expression and function. Arrangement of elements in the (a) SV40-driven FLuc mono-
reporter (pGL3-Control), and (b) the SV40-driven FLuc-P2A-RLuc dual reporter (pCI-6.20) and
4XCRE-driven FLuc-P2A-RLuc dual reporter (pCI-6.24). P2A amino acid sequence (underline) used
in this construct is shown; arrow indicates ribosomal ‘skipping’ site. (c) Western blot analysis
showing the efficient expression of non-tethered reporters, where lane 1 is non-transfection control (transfection reagent only); lane 2, SV40-driven FLuc mono-reporter (pGL3-Control); and lane 3,
FLuc-P2A-RLuc dual reporter (pCI-6.20). Note that co-transfection of 3XFLAG-BAP is to
demonstrate the transfection efficiency was similar. (d) Bioluminescent output from mono FLuc reporter and co-expressed FLuc and RLuc using Dual-Glo reagent; lanes are the same as in (c).
2
Nature Methods: doi:10.1038/nmeth.2170
Supplementary Figure 2| Validation of reporter FLuc-2A-RLuc with SV40 element.
a
b SV40
SV40
Supplementary Figure 2| Validation of reporter FLuc-P2A-RLuc with SV40 element. FLuc and RLuc are both sensitive and commonly used reporters with generally short half-lives and use
different substrates and mechanisms to produce light. Response output from the FLuc (a) or the RLuc (b) reporter from transient transfection of the biocircuit plasmid in HEK293 cell variant,
GripTite MSR in 384-well format. Cells were treated with FSK (●), PTC124 () or BTS (). PTC124
and BTS are inhibitors of FLuc and RLuc, respectively, and act to increase the activity of the
reporters by stabilizing their cellular half-life relative to non-treated control. Note that FSK is
inactive in experiments where reporter expression is driven by the SV40 promoter, only displaying activity when the biocircuit is under control of 4XCRE (see Fig. 1b and 1c). FSK is forskolin.
3
Nature Methods: doi:10.1038/nmeth.2170
Supplementary Figure 3 | qHTS 3-axis plots for output from transiently transfected HEK MSR cells treated with the LOPAC library.
a
b
c
Supplementary Figure 3 | qHTS 3-axis plots for output from transiently transfected GripTite 293
MSR cells treated with the LOPAC library. (a) FLuc, (b) RLuc, and (c) both outputs (red, FLuc and blue, RLuc) plotted as a three-axis graphs for the entire LOPAC library where bold line fits are for compounds displaying qHTS curve classes of 1a, 1b or 2a in either replicate test. The following
classification criteria define curve classes: Class 1 curves display two asymptotes, an inflection point, and r2 ≥0.9; subclasses 1a vs. 1b are differentiated by full (>80%) vs. partial (≤ 80%)
response. Class 2 curves display a single left-hand asymptote and inflection point; subclasses 2a
and 2b are differentiated by a max response and r2, >80% and >0.9 or