RESEARCH | REPORTS • −1 −1 • Rout TdðTout − Tin Þ,whereRout is the outgoing 10. O. Pauluis, J. Atmos. Sci. 68,91–102 (2011). 30. O. M. Pauluis, A. A. Mrowiec, J. Atmos. Sci. 70, 3673–3688 – long-wave radiation and Tin, Tout ,andTd are the 11. O. Pauluis, V. Balaji, I. M. Held, J. Atmos. Sci. 57, 989 994 (2000). (2013). – mean temperature of atmospheric heat input, 12. O. Pauluis, J. Dias, Science 335, 953 956 (2012). 31. V. Lucarini, K. Fraedrich, F. Ragone, J. Atmos. Sci. 68, 13. S. Sherwood, R. Roca, T. Weckwerth, N. Andronova, 2438–2458 (2011). output, and dissipation, respectively. Observations Rev. Geophys. 48, RG2001 (2010). 32. D. L. Hartmann et al., Climate Change 2013: The Physical of recent tropospheric warming [figures 2.26 and 14. I. Held, B. Soden, J. Clim. 19, 5686–5699 (2006). Science Basis. Contribution of Working Group I to the Fifth 2.27 in (32)] show that temperature trends are 15. R. Caballero, P. L. Langen, Geophys. Res. Lett. 32, L02705 (2005). Assessment Report of the Intergovernmental Panel on Climate ’ – somewhat uniform in the vertical, which sug- 16. P. O Gorman, T. Schneider, J. Clim. 21, 5797 5806 (2008). Change, T. F. Stocker et al., Eds. (Cambridge Univ. Press, 17. T. Schneider, P. A. O’Gorman, X. Levine, Rev. Geophys. 48, Cambridge and New York, NY, USA, 2013). T −1 − T −1 gests that the difference out in might increase RG3001 (2010). more slowly than either Tin or Tout.Thisslower 18. G. Vecchi, B. Soden, J. Clim. 20, 4316–4340 (2007). ACKNOWLEDGMENTS • 19. R. Feistel et al., Ocean Science 6,91–141 (2010). increase may explain why d Qtotal does not follow a We acknowledge the Global Modeling and Assimilation Office surface Clausius-Clapeyron scaling and why one 20. Materials and methods are available as supplementary (GMAO) and the Goddard Earth Sciences Data and Information material on Science Online. would expect moist processes to limit the work Services Center (GES DISC) for the dissemination of MERRA data. 21. K. Emanuel, Atmospheric Convection (Oxford Univ. Press, MIT, This work was supported by the G8 Research Initiative grant output in simulations with anthropogenic forcing. Oxford, 1994). “ExArch: Climate analytics on distributed exascale data archives” Simulations over a wider range of climates would 22. D. R. Johnson, International Geophysics, D. A. Randall, Ed. made available through the Natural Sciences and Engineering – help verify this hypothesis. (Academic Press, Waltham, 2001), pp. 659 720. Research Council (NSERC). 23. P. R. Gent et al., J. Clim. 24, 4973–4991 (2011). Our comparison of thermodynamic cycles in 24. R. H. Moss et al., Nature 463, 747–756 (2010). – CESM and MERRA show many similarities; how- 25. M. M. Rienecker et al., J. Clim. 24, 3624 3648 (2011). SUPPLEMENTARY MATERIALS 26. J. D. Zika, M. H. England, W. P. Sijp, J. Phys. Oceanogr. 42, ever, we find that CESM requires less power to www.sciencemag.org/content/347/6221/540/suppl/DC1 708–724 (2012). maintain its hydrological cycle than the reana- Materials and Methods 27. K. Döös, J. Nilsson, J. Nycander, L. Brodeau, M. Ballarotta, Supplementary Text lysis, due to the smaller amplitude of its moist- J. Phys. Oceanogr. 42, 1445–1460 (2012). Figs. S1 to S5 ening inefficiencies. We suggest that this difference 28. J. Kjellsson, K. Döös, F. B. Laliberté, J. D. Zika, J. Atmos. Sci. References (33–36) might be a consequence of the idealized nature 71, 916–928 (2014). of parameterized convection schemes, and it is 29. S. Groeskamp, J. D. Zika, T. J. McDougall, B. M. Sloyan, 9 June 2014; accepted 3 December 2014 F. Laliberté, J. Phys. Oceanogr. 44, 1735–1750 (2014). 10.1126/science.1257103 likely that it might also influence the response of CESM to anthropogenic forcing. Typically, con- vection schemes artificially transport moisture along a moist adiabat without accounting for the OPTICAL IMAGING work needed to lift this moisture, but in the real on January 30, 2015 world, this work is necessary to sustain precip- itation. Any increase in global precipitation there- Expansion microscopy fore requires an increase in work output; otherwise, precipitation would have to become more effi- Fei Chen,1 Paul W. Tillberg,2 Edward S. Boyden1,3,4,5,6† cient, for example, by reducing the frictional dis- * * sipation of falling hydrometeors (11, 12). This is one reason we should interpret the constraint in In optical microscopy, fine structural details are resolved by using refraction to magnify work output in CESM as a constraint on the large- images of a specimen. We discovered that by synthesizing a swellable polymer network within a specimen, it can be physically expanded, resulting in physical magnification. scale motions and not on the unresolved subgrid- www.sciencemag.org scaleconvectiveevents. By covalently anchoring specific labels located within the specimen directly to the polymer Our work illustrates a major constraint on the network, labels spaced closer than the optical diffraction limit can be isotropically large-scale global atmospheric engine: As the cli- separated and optically resolved, a process we call expansion microscopy (ExM). Thus, mate warms, the system may be unable to in- this process can be used to perform scalable superresolution microscopy with crease its total entropy production enough to diffraction-limited microscopes. We demonstrate ExM with apparent ~70-nanometer lateral resolution in both cultured cells and brain tissue, performing three-color offset the moistening inefficiencies associated 7 with phase transitions. This suggests that in a superresolution imaging of ~10 cubic micrometers of the mouse hippocampus with a conventional confocal microscope. future climate, the global atmospheric circulation Downloaded from might comprise highly energetic storms due to explosive latent heat release, but in such a case, icroscopy has facilitated the discovery fixed and permeabilized brain tissue (Fig. 1B) the constraint on work output identified here of many biological insights by optically sodium acrylate, a monomer used to produce will result in fewer numbers of such events. magnifying images of structures in fixed superabsorbent materials (2, 3), along with the Earth’s atmospheric circulation thus suffers from M cells and tissues. We here report that comonomer acrylamide and the cross-linker the “water in gas problem” observed in simu- physical magnification of the specimen N-N′-methylenebisacrylamide. After triggering lations of tropical convection (6), where its ability itself is also possible. free radical polymerization with ammonium to produce work is constrained by the need to We first set out to see whether a well-known persulfate (APS) initiator and tetramethylethy- convert liquid water into water vapor and back property of polyelectrolyte gels—namely, that lenediamine (TEMED) accelerator, we treated again to tap its fuel. dialyzing them in water causes expansion of the tissue-polymer composite with protease to the polymer network into extended conforma- homogenize its mechanical characteristics. After 1 — REFERENCES AND NOTES tions (Fig. 1A) ( ) could be performed in a bi- proteolysis, dialysis in water resulted in a 4.5-fold 1. L. Barry, G. C. Craig, J. Thuburn, Nature 415, 774–777 (2002). ological sample. We infused into chemically linear expansion, without distortion at the level 2. M. H. P. Ambaum, Thermal Physics of the Atmosphere (Wiley, of gross anatomy (Fig. 1C). Digestion was uniform Hoboken, 2010), pp. 203–220. 1Department of Biological Engineering, Massachussetts throughout the slice (fig. S1). Expanded speci- – 3. O. R. Wulf, L. Davis Jr., J. Meteorol. 9,80 82 (1952). Institute of Technology (MIT), Cambridge, MA, USA. mens were transparent (fig. S2) because they 4. J. P. Peixoto, A. H. Oort, M. De Almeida, A. Tomé, J. Geophys. 2Department of Electrical Engineering and Computer Res. 96 (D6), 10981 (1991). Science, MIT, Cambridge, MA, USA. 3Media Lab, MIT, consist largely of water. Thus, polyelectrolyte gel 5. R. Goody, Q. J. R. Meteorol. Soc. 126, 1953–1970 (2000). Cambridge, MA, USA. 4McGovern Institute, MIT, Cambridge, expansion is possible when the polymer is em- 6. O. Pauluis, I. M. Held, J. Atmos. Sci. 59, 125–139 (2002). MA, USA. 5Department of Brain and Cognitive Sciences, bedded throughout a biological sample. – 6 7. D. M. Romps, J. Atmos. Sci. 65, 3779 3799 (2008). MIT, Cambridge, MA, USA. Center for Neurobiological We developed a fluorescent labeling strategy 8. S. Pascale, J. Gregory, M. Ambaum, R. Tailleux, Clim. Dyn. 36, Engineering, MIT, Cambridge, MA, USA. 1189–1206 (2011). *These authors contributed equally to this work. †Corresponding compatible with the proteolytic treatment and 9. O. Pauluis, I. M. Held, J. Atmos. Sci. 59,140–149 (2002). author. E-mail: [email protected] subsequent tissue expansion described above, SCIENCE sciencemag.org 30 JANUARY 2015 • VOL 347 ISSUE 6221 543 RESEARCH | REPORTS to see whether fluorescence nanoscopy would errors in length were small (<1% of distance, FWHM by the known immunostained micro- be possible. We designed a custom fluorescent for errors larger than the imaging system point tubule width [55 nm (6)], conservatively ignoring label (Fig. 1D) that can be incorporated directly spread function size; n = 4 samples) (Fig. 2C). the width of the trifunctional label, and ob- into the polymer network and thus survives Throughout the paper, all distances measured tained an effective resolution for ExM of ~60 nm. the proteolytic digestion of endogenous bio- in the post-expansion specimen are reported di- This conservative estimate is comparable with the molecules. This label is trifunctional, comprising vided by the expansion factor (supplementary diffraction-limited confocal resolution [~250-nm a methacryloyl group capable of participating materials, materials and methods). lateral resolution (8)] divided by the expansion in free radical polymerization, a chemical fluo- We next compared pre-ExM conventional factor (~4.5). rophore for visualization, and an oligonucleotide superresolution images to post-ExM confocal Clathrin-coated pits were also well resolved that can hybridize to a complementary sequence images.