Intracellular Photoactivation and Quantification Using Fluorescence Microscopy: Chemical Tools and Imaging Approaches

796 CHIMIA 2016, 70, No. 11 New Faculty iN SwitzerlaNd

doi:10.2533/chimia.2016.796 Chimia 70 (2016) 796–799 © Swiss Chemical Society Intracellular Photoactivation and Quantification Using Fluorescence Microscopy: Chemical Tools and Imaging Approaches

Giovanni Bassolino and Pablo Rivera-Fuentes*

Abstract: Recent advances in optical microscopy enable the visualization and quantification of biological processes within live cells. To a great extent, these imaging techniques remain limited by the physical properties of the chemical probes that are used as fluorescent tags, detectors and actuators. At the same time, the quantification of concentrations in the intracellular medium is not trivial, but a few approaches that employ optical microscopy have been developed. Herein, we highlight a few examples of how a combination of novel chemical probes and microscopy methods could be used to bring a much-needed quantitative dimension to the field of biological imaging. Keywords: Fluorescence microscopy · Fluorophores · Imaging · Photoactivation · Uncaging

Institute of Molecular Medicine), he joined strategies, but rather we showcase a selec- the Laboratory of Organic Chemistry at tion of those that highlight the benefits of ETH Zurich as Assistant Professor (2015). combining novel synthetic probes with Research in his group focuses on develop- quantitative microscopic and bioanalytical ing molecular probes to detect and manip- techniques. ulate small molecules in live cells.

2. Chemical Tools 1. Introduction Photoactivatable molecules, also Optical microscopy has contributed known as ‘caged’ or ‘photocaged’, consti- Pablo Rivera-Fuentes was born in greatly to our understanding of biological tute a class of molecular probes that allow 1984 in Mexico City and received a BSc systems at the cellular level. Recent meth- the activation of a bioactive molecule using (2008) degree in chemical engineering odological advances in this field allow sci- light (Scheme 1a). This activation mode from the National Autonomous University entists to not only visualize but also quan- has the advantage of providing excellent of Mexico. He subsequently received tify biological processes.[1] A judicious spatial and temporal resolution.[3] The gen- MSc (2009) and PhD (2012) degrees in combination of appropriate (bio)chemical eral concept of photoactivatable probes is chemistry from ETH Zurich, working probes and microscopic techniques can that the protecting group masks the bio- under the supervision of Prof. François be used to relate the intensity and spatial logical activity of the molecule of interest Diederich. His PhD work was awarded the distribution recorded in a microscope to until it is cleaved through a photochemical ETH Medal in 2013. Funded by the Swiss structural and chemical quantities such as reaction. This strategy has been applied to National Science Foundation (SNSF), distances, concentrations, interactions, or a very diverse range of biologically active he carried out postdoctoral research at reactions.[2] One branch of our research species, including metal ions,[4,5] diatom- Massachusetts Institute of Technology focuses on the photoactivation of small ic signaling agents (i.e. NO, CO, etc.),[5] with Prof. Stephen J. Lippard (2012– molecules in defined intracellular loca- neurotransmitters,[6–8] lipids,[9] nucleic ac- 2014). After a short postdoctoral stint at tions and quantifiable amounts. For this ids,[8] sugars,[3,8] and proteins.[8] Whereas the University of Oxford with Prof. Harry purpose, our lab is developing new pho- light has been the most popular trigger to L. Anderson (Department of Chemistry) toactivatable groups, intracellular target- activate chemical probes,[3] other stimuli and Prof. Christian Eggeling (Weatherall ing units, switchable dyes, and fluorescent have been successfully applied, including sensors, with the aim of exploring the ef- enzymatic cleavage[10] and bioorthogonal fects that diverse small molecules exert on reactions with small molecules.[11,12] intracellular signaling. In this review, we Many photoremovable groups have provide a brief discussion of representative been applied to the activation of fluoro- chemical tools and imaging methodologies phores.[13,14] One reason is that it is ex- available for the controlled release and perimentally convenient to evaluate the *Correspondence: Prof. Dr. P. Rivera-Fuentes quantification of small molecules in living properties of new photoremovable groups Laboratorium für Organische Chemie cells employing fluorescence microscopy. when the released molecule becomes col- ETH Zürich, HCI G329 We have not attempted to survey exten- ored and/or fluorescent. In addition, pho- Vladimir-Prelog-Weg 3, CH-8093 Zürich E-mail: [email protected] sively all the available probes and imaging toactivatable fluorophores possess other New Faculty iN SwitzerlaNd CHIMIA 2016, 70, No. 11 797

such quantification was included in the report.[25] It is worth highlighting that although HPLC analysis of lysates gives useful information about the population of cells, it does not account for cell-to-cell variability. In particular when studying complex populations of cells, such as tis- sue or whole animals, being able to obtain quantitative information directly from imaging data will greatly increase our ability to understand biological systems. Linker 3, also developed by Shabat and coworkers, operates based on an ortho- benzylic elimination. This linker was com- bined with a hydrogen peroxide-activated trigger to generate probe 5 (Fig. 1), which Scheme 1. Chemical tools and principles useful for controlled release in quantitative fluorescence was used for selective drug delivery of microscopy. a) Photoactivatable compounds and b) self-immolative linkers. A biologically active anticancer agents to mice.[26] Similarly to molecule or a fluorophore (represented as a magenta pentagon) is rendered temporarily inactive the coumarin linker, a correlation between or non-fluorescent (grey pentagon) through covalent binding with a protecting group (blue hexa- fluorescence intensity and amount of drug gon). Interaction with an external stimulus (orange lightning) cleaves the protecting group, resulting in the activation of the protected compound. When self-immolative linkers are used, upon released was measured in vitro. The modu- removal of the protecting group, the linker itself undergoes spontaneous rearrangement (symbol- lar nature of these compounds makes them ized as a change in shape) in order to release the active molecule. attractive candidates for the development of light-activated, self-immolative compounds that could potentially be applied in interesting applications. For instance, tive microscopy because the stoichiomet- quantitative microscopy. emitters that can be turned on with light are ric release of a fluorescent reporter makes The Porter[27] and Jullien groups[28,29] useful for super-resolution microscopy, in it possible to quantify the amount of re- developed a conceptually different fluoro- particular in localization microscopies.[15] leased (bio-)molecule. genic, photoremovable protecting group. Photoactivatable dyes are also useful for Coumarin 1 is based on the elimina- These trans-cinnamic acid derivatives labeling and tracking cells,[16] to evaluate tion of a leaving group from a benzylic- release the molecule of interest triggered exchange rates of intercellular junctions[17] like position promoted by the increased by a photoinduced cis-trans isomerization or to calibrate photolysis experiments.[18] electron-donating ability of the deprotect- and subsequent intramolecular lactoniza- A less explored area is the development ed hydroxide group. This self-immolative tion (Scheme 2b).[27] In particular, com- of protecting groups that become fluores- linker has been incorporated into a series pound 6 (Fig. 1) has the advantage of being cent (fluorogenic) or change their emission of enzyme-activated compounds (i.e. com- efficiently activated under two-photon ex- wavelength after they are cleaved from the pound 4, Fig. 1) and used as reporter in citation conditions,[28,29] which is desirable biologically active molecule. The devel- drug delivery systems by Shabat and co- because light of longer wavelength dis- opment of such protecting groups is chal- workers.[25] The compounds were tested in plays deeper penetration into non-translu- lenging because the functional groups that live cells, and it was observed that more cent media. Exploiting the large contrast in facilitate photoactivation are often incom- intense fluorescence correlated with an in- fluorescence between the protected (trans- patible with high quantum yields of lumi- crease in released drug, as determined by cinnamic acid) and activated (coumarin) nescence (e.g. NO groups).[19] A strategy HPLC analysis of cell lysates. Although product, the photoinduced release of cy- 2 that has overcome these difficulties is to quantitative microscopic approaches could tosolic ethanol from the cinnamic ester in insert a spacer that, upon release from the have been used to determine the intracel- zebrafish embryos was quantified via fluo- protecting group, undergoes a spontane- lular concentration in individual cells, no rescence correlation spectroscopy.[28,29] ous rearrangement resulting in the release of the bioactive molecule of interest and a fluorescent molecule (Scheme 1b).[20,21] These reactive spacers were introduced in the 1980s[21] and are known as ‘self- immolative’ linkers because they trigger their own elimination.[22] A number of self- immolative linkers has been developed, mostly based on benzylic eliminations or spontaneous intramolecular cyclization following the release from the protecting group.[22–24] Among the probes that exploit self- immolative linkers and other rearrange- ments, the three shown in Scheme 2 stand out because, upon release of the molecule of interest, they generate a fluorescent species in a 1:1 stoichiometry. These examples include coumarin derivatives 1 and 2, Scheme 2. Examples of fluorogenic, small-molecule releasing units (1–3) and their mechanisms and cyanine dye 3. This class of protecting of release. The structures of the released fluorophores are shown in green. group is of particular interest for quantita- Hexagon = protecting group; pentagon = activatable molecule. 798 CHIMIA 2016, 70, No. 11 New Faculty iN SwitzerlaNd

Fig. 1. Compounds ence dye and interpolating the measured O used for the fluoro- fluorescence intensity of the sample.[36,37] O O O N H genic release of small N N O Hsu and Dervan further validated their O N N molecules. The fluo- H H quantification by building a calibration O O rogenic unit is colored + curve in a flow cytometer with commer- NH3 red. cially available fluorescent standard beads. 4 O This method also involved the estimation O O O O of the concentration by interpolation into a HN Cl O O N calibration curve. Even though there were B O O O OH N some discrepancies between the results O O N obtained by each method, the measured O O Cl concentration was consistently lower than O [37] Br in the incubating solution used, giving O valuable information about the membrane N+ 5 N+ HO OH permeability of their polyamides. Br Fluorescent standard beads are mi- 6 O S 3 SO3 crospheres with different loading levels of one or multiple dyes.[38,39] The use of such beads for calibration is based on the This study is a good example of how to diffusion time constants and the number of concept of ‘molecules of equivalent solu- photoactivate a bioactive molecule and molecules within the illuminated volume, ble fluorophore’ (MESF).[40] To assign an quantify its concentration directly using for populations of both single and multiple MESF value to a population of fluores- microscopic techniques. emitters.[30,32] In the case of compound 6, cent beads, a calibration curve is built by there is a large contrast between the dark plotting the fluorescence intensity against cinnamic ester and the activated, highly concentration for a series of solutions of 3. Approaches for Quantifying the emissive coumarin, and therefore it is safe the same dye bound to the beads. The fluo- Absolute Concentrations of Dyes to assume that the only emitting species rescence intensity of a suspension with a Using Microscopy is the latter. In this way, the authors could known number of beads is measured, and estimate an initial intracellular concentra- the equivalent number of fluorophores in In general, the concentration of a dye tion of the protected compound, which was solution is assigned by interpolation em- in a biological sample cannot be inferred very close to the one used to incubate the ploying the calibration curve. Assignment directly from the intensity of its emission sample (0.5–10 µM). Another example of of a MESF value to a fluorescent bead does recorded by fluorescence microscopy. The quantification in live cells via FCS is the not require the emission quantum yield of reason is that the emission quantum yields work by Cluzel and coworkers, who quan- the dye in the bead and in solution to be of small-molecule emitters can vary greatly tified the expression of the chemotactic the same or even known. MESF values depending on their concentration, the pH signaling protein Che-Y fused to GFP in are assigned by comparison of the absor- of the medium, the presence of quenchers, order to characterize the behavior of indi- bance-normalized fluorescent intensities the polarity, and even the temperature and vidual flagellar motors in E. coli.[34] that, under the appropriate experimental viscosity of the sample. As a consequence, Alternatively, the construction of cali- conditions, are dependent on the product more information about the photophysical bration curves has been suggested as a of the number of molecules present and properties of the dye and the conditions method to relate the fluorescence intensity the quantum yield of emission, a procedure of the medium are needed for quantitative observed in a confocal microscope to the that does not require knowledge of either microscopy. concentration of the dye of interest in the parameter for the sample to be character- A technique that can provide some sample.[35] This method is based on the as- ized.[40] On the other hand, the dependence of this information is fluorescence cor- sumption that the fluorescence intensity of the emission quantum yield on the en- relation spectroscopy (FCS). In FCS, the measured toward the center of an object vironment matters when using the beads fluctuations in the fluorescence of a small larger than the point spread function gen- to estimate the absolute number of fluoro- ensemble of equilibrated molecules are erated by the microscope optics is directly phores in an unknown sample (e.g. the in- recorded.[30–32] The local concentration of proportional to the fluorophore concentra- tracellular environment),[38,39] because the molecules can be extracted with the aid of tion, provided that the region probed by quantum yield of emission for the dye in computational modeling provided that an the focused beam is entirely contained the sample might well be different from the autocorrelation trace of the fluorescence within the cellular volume.[35] Following one in the standard solutions used to gen- intensity is available.[30–32] An autocorrela- this approach, Loew and coworkers built a erate the original MESF calibration curve. tion function is built by measuring the fluc- calibration curve by measuring the fluores- In fact, the dependence of the dye spectral tuations of the intensity against the mean cence intensity of a series of solutions of features on the environment is a problem value over time and reports on how the different concentrations of the dye (Fura-2, that all these methods with external cali- values correlate as a function of their sepa- a calcium indicator) between coverslips, bration share, and it highlights the impor- ration within the time series (i.e. the time keeping the same instrumental parameters tance of developing new fluorescent dyes interval between signals).[33] In order to used for cell imaging. A similar approach with emission that has minimal sensitivity obtain meaningful autocorrelation curves, was followed by Schultz and coworkers towards the environment. only very few emitters need to be measured to quantify bioactive lipids photoreleased Chiu and coworkers used an alterna- at once, but this difficulty is overcome in from coumarin-protected derivatives[36] tive approach to estimate the number of a confocal setup because the illuminated and by Hsu and Dervan to quantify polyam- emitters per imaged region. In their stud- volume can be decreased to the order of ide-fluorescein conjugates in cellular nu- ies, they fitted the measured intensity dis- femtoliters by tight focusing of the laser clei.[37] In both studies, calibration curves tribution for small ensembles of labeled beam.[31] Models are available to describe were built taking a series of images of stan- proteins with the intensity distribution of the autocorrelation function in terms of the dard solutions of known amounts of refer- a single labeled protein.[41,42] The starting New Faculty iN SwitzerlaNd CHIMIA 2016, 70, No. 11 799

point was to obtain a set of intensities from being able to release small molecules, in [20] P. K. Chakravarty, P. L. Carl, M. J. Weber, J. fluorescent regions where there is only a quantifiable amounts, within selected sub- A. Katzenellenbogen, J. Med. Chem. 1983, 26, 638. single protein and find the distribution that cellular compartments using fluorescence [21] P. L. Carl, P. K. Chakravarty, J. A. best describes the set (usually a lognormal microscopy. A combination of synthetic Katzenellenbogen, J. Med. Chem. 1981, 24, one). That signal was then used as the ba- chemistry, chemical biology, and quantita- 479. sis for fitting the distribution intensities of tive microscopy will be crucial for the de- [22] A. Alouane, R. Labruère, T. Le Saux, F. regions where more proteins are present. velopment and application of these probes. Schmidt, L. Jullien, Angew. Chem. Int. Ed. 2015, 54, 7492. They found that these intensities are not [23] M. E. Roth, O. Green, S. Gnaim, D. Shabat, linearly related (i.e. the intensities of a re- Acknowledgements Chem. Rev. 2016, 116, 1309. gion with 10 proteins is not 10 times the in- Work in our laboratory is funded by a [24] S. Gnaim, D. Shabat, Acc. Chem. Res. 2014, 47, tensity of the individual one). Instead, they start-up grant from ETH Zurich, the ETH 2970. Zurich Research Grant ETH-02 16-1 and the [25] R. Weinstain, E. Segal, R. Satchi-Fainaro, D. estimated the number of emitters by find- Shabat, Chem. Commun. 2010, 46, 553. ing the best fit for the intensity distribution Swiss National Science Foundation grant 200021_165551. [26] O. Redy-Keisar, S. Ferber, R. Satchi-Fainaro, by multiplying the distribution of the indi- D. Shabat, ChemMedChem 2015, 10, 999. Received: August 9, 2016 vidual labeled protein by an integer.[41,42] [27] A. D. Turner, S. V. Pizzo, G. Rozakis, N. A. The lognormal character of the distribution Porter, J. Am. Chem. Soc. 1988, 110, 244. [1] V. C. Coffman, J.-Q. Wu, Trends Biochem. Sci. [28] N. Gagey, P. Neveu, C. Benbrahim, B. Goetz, I. arises from instrumental factors that are 2012, 37, 499. Aujard, J.-B. Baudin, L. Jullien, J. Am. Chem. function only of the position within the im- [2] J. C. Waters, J. Cell Biol. 2009, 185, 1135. Soc. 2007, 129, 9986. age but not the fluorophore concentration; [3] P. Klán, T. Šolomek, C. G. Bochet, [29] N. Gagey, P. Neveu, L. Jullien, Angew. Chem. they also considered factors related to the A. Blanc, R. Givens, M. Rubina, V. Popik, A. Int. Ed. 2007, 46, 2467. Kostikov, J. Wirz, Chem. Rev. 2013, 113, 119. [30] O. Krichevsky, G. Bonnet, Rep. Prog. Phys. labeling efficiency, such as the existence [4] C. Gwizdala, S. C. Burdette, in ‘Inorganic 2002, 65, 251. of a preferential labeling site, quenching of Chemical Biology: Principles, Techniques and [31] P. Schwille, Cell Biochem. Biophys. 2001, 34, the fluorophore by the antibody or differ- Applications’, Ed. G. Gasser, John Wiley & 383. ent extent of labeling between proteins. In Sons, Ltd., Chichester, U.K., 2014, p. 275. [32] S. A. Kim, K. G. Heinze, P. Schwille, Nat. [5] K. L. Haas, K. J. Franz, Chem. Rev. 2009, 109, the system under study, the authors found Methods 2007, 4, 963. 4921. [33] G. E. P. Box, G. M. Jenkins, G. C. Reinsel, in that these factors do not affect the lognor- [6] V. R. Shembekar, Y. Chen, B. K. Carpenter, G. “Time Series Analysis”, John Wiley & Sons, mal character of the distribution within the P. Hess, Biochemistry 2007, 46, 5479. Inc., Hoboken, NJ, 2013, pp. 21. experimental and fit errors.[42] Both stud- [7] N. Senda, A. Momotake, T. Arai, Bull. Chem. [34] P. Cluzel, M. Surette, S. Leibler, Science 2000, ies were performed on proteins that were Soc. Jpn. 2007, 80, 2384. 287, 1652. [8] G. Mayer, A. Heckel, Angew. Chem. Int. Ed. [35] C. Fink, F. Morgan, L. M. Loew, Biophys. J. labeled using organic fluorophores, there- 2006, 45, 4900. 1998, 75, 1648. fore it is possible that this approach could [9] D. Höglinger, A. Nadler, C. Schultz, Biochim. [36] A. Nadler, D. A. Yushchenko, R. Müller, F. be used for small-molecule fluorophores in Biophys. Acta – Mol. Cell Biol. Lipids 2014, Stein, S. Feng, C. Mulle, M. Carta, C. Schultz, solution, but this hypothesis remains to be 1841, 1085. Nat. Commun. 2015, 6, 10056. [10] C. Ritter, N. Nett, C. G. Acevedo-Rocha, tested. [37] C. F. Hsu, P. B. Dervan, Bioorg. Med. Chem. R. Lonsdale, K. Kräling, F. Dempwolff, S. Lett. 2008, 18, 5851. Hoebenreich, P. L. Graumann, M. T. Reetz, [38] Spherotech, Inc., ‘SPHEROTM Calibration E. Meggers, Angew. Chem. Int. Ed. 2015, 54, Particles’. The technical document can be found Closing Remarks 13440. under http://www.spherotech.com/2016 Catalog [11] J. Li, P. R. Chen, Nat. Chem. Biol. 2016, 12, Product Detail Pages/Spherotech Rainbow 129. Calibration Beads for Cytometry Performance With the advent of advanced imaging [12] J. Kim, C. R. Bertozzi, Angew. Chem. Int. Ed. Verification.pdf, accessed July 27, 2016. techniques and improved equipment, the 2015, 54, 15777. [39] Bangs Laboratories, Inc., ‘QuantumTM MESF time is ripe to start using optical microsco- [13] L. M. Wysocki, J. B. Grimm, A. N. Tkachuk, T. Kits’. The technical document can be found py as a widespread quantitative technique. A. Brown, E. Betzig, L. D. Lavis, Angew. Chem. under http://www.bangslabs.com/sites/default/ Int. Ed. 2011, 50, 11206. The development of new fluorogenic files/imce/docs/PDS 821 Web.pdf, accessed July [14] F. M. Raymo, ISRN Phys. Chem. 2012, 619251. 27, 2016. molecules specifically tailored to take ad- [15] F. M. Raymo, J. Phys. Chem. Lett. 2012, 3, [40] A. Schwartz, L. Wang, E. Early, A. Gaigalas,Y.- vantage of this opportunity is pivotal and 2379. z. Zhang, G. E. Marti, R. F. Vogt, J. Res. Natl. demonstrates that chemistry is at the heart [16] D. J. Kozlowski, E. S. Weinberg, Methods Mol. Inst. Stand. Technol. 2002, 107, 83. of modern microscopy and bioimaging. In Biol. 2000, 135, 349. [41] S. A. Mutch, B. S. Fujimoto, C. L. Kuyper, J. [17] W.-h. Li, G. Zheng, Photochem. Photobiol. Sci. S. Kuo, S. M. Bajjalieh, D. T. Chiu, Biophys. J. this review we focused on selected strate- 2012, 11, 460. 2007, 92, 2926. gies that could be applied to the activation [18] G. P. Ahern, S.-F. Hsu, M. B. Jackson, J. [42] S. A. Mutch, P. Kensel-Hammes, J. C. Gadd, B. and quantification of small molecules in Physiol. 1999, 520, 165. S. Fujimoto, R. W. Allen, P. G. Schiro, R. M. the intracellular medium. This is one of the [19] B. Valeur, M. N. Berberan‐Santos, in ‘Molecular Lorenz, C. L. Kuyper, J. S. Kuo, S. M. Bajjalieh, Fluorescence: Principles and Applications’, D. T. Chiu, J. Neurosci. 2011, 31, 1461. goals of our research group at ETH Zurich, Wiley-VCH Verlag GmbH & Co. KGaA, and our current work is directed towards Weinheim, Germany, 2012, p. 75.