Selective Modification for Red‐Shifted Excitability: A Small Change in Structure, a Huge Change in Photochemistry

Abstract We developed three bathochromic, green‐light activatable, photolabile protecting groups based on a nitrodibenzofuran (NDBF) core with D‐π‐A push–pull structures. Variation of donor substituents (D) at the favored ring position enabled us to observe their impact on the photolysis quantum yields. Comparing our new azetidinyl‐NDBF (Az‐NDBF) photolabile protecting group with our earlier published DMA‐NDBF, we obtained insight into its excitation‐specific photochemistry. While the “two‐photon‐only” cage DMA‐NDBF was inert against one‐photon excitation (1PE) in the visible spectral range, we were able to efficiently release glutamic acid from azetidinyl‐NDBF with irradiation at 420 and 530 nm. Thus, a minimal change (a cyclization adding only one carbon atom) resulted in a drastically changed photochemical behavior, which enables photolysis in the green part of the spectrum.


Introduction
Due to their advantage of spatiotemporal control withoutt he use of additional chemical reagents, photolabile protecting groups (PPGs) or "caging groups" already created an extensive pool of applicationsi nt he fields of biochemistry, [1][2][3] organic synthesis [3,4] and even inorganic materials for coated surfaces [5] or hydrogel formation. [6] Nonetheless, the development and synthesis of PPGs, which can be used by irradiation with visible light-optimally within the "phototherapeutic window" (650-950 nm) [7] and thus in living cells without tissue damage-remains one of the main tasks of modern photochemistry.A lternatively, alsoP PGs which can be activated with visiblel ight within the "green gap" (low absorption of the light-harvesting complexes of plants) from 500-600nma re highly desired for new applications in plants. [8] There are several strategies for achieving the necessary bathochromic absorption shift: psystem extension of the chromophore( maintaining planarity) [9] as well as an attachment of donor (D) and acceptor (A) structures to create ap ush-pull character [10] that enhances electron delocalization and therefore decreases the energy requiredf or excitation,a re few of them. The p-system strategy has often to deal with solubility issues in aqueous media.
For biologically suitable caging groups,a part from the absorbance properties, another important aspecti st he quantum yield of photorelease fr( in competitiont oa lternative relaxation pathways from the excited state) as the uncaging efficiency is determined by the product e·fr. An on-negligible part of the radiation energy is lost for example, by fluorescencee mission or non-radiatived ecay channels like rotation around single bonds and intramolecular vibrational energy redistribution (IVR).
Rivera-Fuentes and co-workersp ublished ac omparison of azetidinyl-coumarin (Az-CM) and the widely used blue-absorbing PPG diethylaminocoumarin (DEACM) regardingt heir photolysis efficiency (Figure 1). [11] At hird derivativec arriedajulolidine [12] substituent, where rotationa round the N-C( donor) bond was prohibited due to the connectiono ft he 6-membered alkyl-rings to the aromatic system. Interestingly,j ulolidine-and azetidinyl-coumarin showed ah ighly similar behavior in all experiments. They investigated the photolysis rates of the derivatives depending on solventp olarity and proticity.I n water,t he derivativew ith the azetidinyl substituent had as ignificantly better frt han DEACM because the small heterocyclic ring appearst oi nhibitp hotochemically unproductive decay channels, whereas the diethylamino derivativel oses the photon energy. One widespread explanation for this phenomenono fs tructural influence is intramolecular charge transfer (ICT) state population,a nother one is the hydrogen bonding (H-bonding) hypothesis. Both effects strongly depend on the surrounding medium. Charge separation can be stabilized in polar solvents and H-bonding induces non-radiative decay in protic solvents. When experiments are performed in water,w hich is both polar and protic, both may play ar ole. ICT states in general are ar elatively common phenomenon in molecules with aD -p-A design.I ft he intramolecular electron transfer from donor to acceptorl eads to at wisting of as ingle bond, this is referred to as T-ICT.I ft he resulting conformational change is notarotation but rather ap lanarization-for example of ap reviously pyramidal amine-this is called P-ICT. [13,14] Ap opulation distribution between locally exited (LE) state and as tabilized charge transfer (CT) state may be detectable by ad ual fluorescence. [15,16] The CT state opens up new relaxation pathways. Knowledge of these pathways and influence of different substituents with torsional angle and bond length restriction possibilities can be exploited for novel PPG design. [17,18] Apart from the photorelease studies of Az-CM, azetidinyl substituents in general have been known in the literature for their positive effects on the photochemistry of rhodamines for al ong time. [19]

Results and Discussion
The nitrodibenzofuran (NDBF)c ore had originally been introduced as aP PG by Ellis-Davies. [20] In ap reviousp ublication guided by theoretical predictions [21] we presented the improvedd imethylamino-NDBF (DMA-NDBF)g roup, [22] which showedasurprising excitation-specific behavior:Aone-photon (1P) irradiation into the red-shifted main absorbance band around4 20 nm did not afford any photolysis any more (f 420 < 0.05 %) in contrast to NDBF (f 420 = 13.6 %), while two-photon (2P) irradiation at 840 nm was very effective (17 times better than NDBF). Our present study contributes to the understanding of the reasonf or this unusual behavior whichm ight not be ar are case but rather ar arely recognized one. [23] Preferably, we wanted to maintain all the positive properties like the redshift and high e,t oo btain ad esirable example of ag reen-light activatable PPG which can be cleaved by 1PE. Azetidinyl N(CH 2 ) 3 should act as ad onor substituent at ring position7 (Az-NDBF), replacing NMe 2 in DMA-NDBF [22] (Figure1).
Additionally,w ei ncludedt wo aryl-NDBF derivatives into this investigation to study the influence of a) stronger electron donors [25] and b) spatially demanding substituents on the photochemistry.B ased on ap revious theoretical study for the optimal substitution pattern of NDBF for new PPGs with bathochromic shift, [21] we chose the tolyl-a nd anisylamino derivatives shown in Figure 2. This previous study concluded that only donor-attachment at the 7-position led to improved performance and that additional substitutions had no substantial furthereffect.
Also for this study,w es tarted out with time-dependent density functional (TDDFT) calculations for initial characterization. Ground state equilibrium structures, vertical excitation energies w ex ,o scillator strengthsf osc as wella s2 Pa bsorption probabilities, which are directly convertible to the absorption cross-section d a in GM were obtained at the CAM-B3LYP/def2-svp level of theory.
Ta ble 1s ummarizes the results for the transition from S 0 to the three energeticallyl owest singlet excited states (S 1 -S 3 )o f DMA-and Az-NDBF-OH and the two phenyl-candidates DTAand DAA-NDBF-OH.W hile DMA-and Az-NDBF-OH show only slight differences in our theoretical calculations, except for their oscillator strength f osc to the S 1 state, the phenylm oieties of compounds DTA-and DAA-NDBF-OH have ap ronounced effect. They lead to remarkably high predicted absorption cross-sectionso f2 59 and 296 GM, respectively.I nc omparison, NDBF-OH,w hich has already been used in living cells, has a calculated value of 0.14 GM (630 nm) with our methoda ndcoupled to EGTAa sC a 2 + releasing agent-an experimentally measured value of 0.6 GM (710 nm). [20] Going from NDBF to DMA-NDBF we had experimentally observedar ed-shift of > 100 nm. [22] Here, af urther aryl substitution to compounds DTA and DAA-NDBF was calculated to afford only as mall additional red-shifto f18-23 nm.
An overview of the synthesis routes is given in Scheme 1. As first synthesis step, all three NDBF derivatives have aB uchwald-Hartwig cross coupling between m-halogenatedp henol  [4,24] azetidinyl-coumarin (Az-CM), [11] dimethylamino-nitrodibenzofuran (DMA-NDBF) [22] and the new azetidinyl-nitrodibenzofuran (Az-NDBF)with LG = leavinggroup. (1, X = iodide or bromide) and as econdary amine (a: azetidine, b: di-p-tolylamine or c: di-p-anisylamine) in common.T his derivatization was followed by iodination in para-position to the corresponding amino function( ortho to the phenolic OH) with NIS. The next synthesis steps were largely similar to the synthesis of DMA-NDBF as published earlier. [22] Couplingt o4 -fluoro-2-nitrobenzaldehyde (4)l ed to the unsymmetrical aryl ethers 5, which were subsequently reduced and methylated with trimethylaluminum and afterward hydrolyzed to obtain alcohols 6.T he closed-ring form is the product of apalladium-catalyzed intramolecular Heck-like reaction. The leaving group -OH can be varied in furthersteps.
1P-Absorption spectrao ft he alcohols were recorded in different solvents (Supporting Information, Figures S4-S6). Figure 3s hows the spectrai nD MSO-along with the one of the unsubstituted NDBF-OH (R = H) for comparison. The longwavelength absorption maximaa re red-shifted from 312 nm (NDBF-OH)t o4 22 nm (Az-and DTA-NDBF-OH)a nd 426 nm (DAA-NDBF-OH). As theoretically predicted, the donor variation does not strongly affect the red-shift, suggesting that we have currently found an optimum of our D-p-A system.T hus, also Az-NDBF-OH showed the expected similar absorption behavior to the one of DMA-NDBF-OH (e max = 424 nm). Not only the wavelength of the maxima turned out to be similarb ut also the respective molar absorbance coefficient e. Within error limits DTA-NDBF-OH showed the highest molar absorbance with e 422 = 17196 Lmol À1 cm À1 in its (second) maximum which is 7% highert han the one of DMA-NDBF-OH (e 424 = 15947 Lmol À1 cm À1 ).
We also recordeds teady-state fluorescencee mission spectra ( Figure 4) to investigate the occurrence of dual fluorescencet o probe for the ICT/-H-bonding hypothesis. We would expect fluorescencef rom the LE and CT states. [26] Experimentally,t he following was observed:a ll derivatives showedw eak fluorescence signals in general.T his corresponds to our TDDFT calculations, which predicted low-energy transitions to the first excited states. Therefore, non-radiative decay is very likely. [27] Ap- Table 1. Calculated values for the vertical excitation energies w ex ,t he one-photon oscillator strengths f osc and the two-photon absorption crosssection d a at the given wavelengthf or the transition to the three energetically lowest exciteds tates S 1 -S 3 of DMA-, Az-, DTA-and DAA-NDBF (LG = OH).    propriately,t he signals are stronger in less polar solvents such as toluenea nd weakeri nm ethanol, whichs tabilizes intramolecular charges. The strongest electron donor,a nd therebyC Tsupporting derivative DAA-NDBF-OH has the weakest toluene fluorescence( light purple vs. light pink and green), whereas the other derivatives are similarly bright in toluene. In MeOH DMA-NDBF is completely dark (dark pink line). This means non-radiative or ultrafast decaysd ominate, in clear difference to Az-NDBF-OH (dark green line). This finding is highly interesting as we see photochemical differencesi nt he fluorescence but hardly in the absorption.
This solvents ensitivity of the two electronically low-lying excited states was further investigated by calculations and is illustrated in Figure5.T he state characters were assigned by means of detachment and attachment densities ( Figure 5A and Be xemplarily for DMA-NDBF-OH), whichd emonstrate a shift of electrond ensity from the amino to the nitro group for the CT state, whereas the LE state is localized on the nitro group. Dipole moments of the excited states further corroborate the respective character (Supporting Information). In vacuum (C left) and non-polar environment such as for example, cyclohexane or toluene, the energeticallyl owest excited state (S 1 ), accessible from the ground state geometry,i saCT state. Due to relaxation to the lowest electronically excited state (Kasha's rule) and ac onical intersection (black triangles) the NDBF derivatives fluorescee ventually from the long-living LE state (bright fluorescence) at around5 50 nm. Them ore polar the environmentb ecomes (e.g.,M eOH or water), the more the CT state is stabilized and energeticallyl owered (C right), whereas the LE state is higheri ne nergy.T he conical in-tersection disappears and the fluorescencea rises from the CT state, which is too fast for steady-state fluorescencem ethods.
Ap ublication by the Sølling group, [28] who examined the dual fluorescenceo fasimilarc ore structure, 2-diethylamino-7nitrofluorene, can be used for comparison. They computationally identifieds everale xcited state minima with CT character, including aP -ICT state, as well as rotation aroundt he amino group (T-ICT) and also the nitro group, which accounts for the majority of the non-radiative decay.T hey observed complex solventd ependence and ultrafast ISC in somec ases (e.g., apolar cyclohexane), which results then in one single fluorescent transition. They stated that the solvent determines not just the fluorescence lifetime, it shapes the potential energy landscape and thereby all relaxation pathways.
Although uncaging mechanisms cannotb ep redicted by TD-DFT,o ne can surely assumet hat the ultrafast decay pathways preventt he photolysis. In our case, this means that uncaging is only possible from the LE state, which depends on the solvent and donor properties. Hence, the solvent polarityd ecides if the NDBF derivative will photolyze and if we assume a planar CT state (PICT hypothesis), in the same solventt he planarization of pyramidal N(CH 3 ) 2 should be easier (= lowered CT) than the one of the small heterocyclic azetidine ring due to required bond length changes and ring strain (= high CT, lower LE).
For furtheri nvestigations, whether Az-NDBF is able to suppress the decay channels competing with uncaging in aqueous medium, we attached aw ater-soluble leaving group. l-glutamic acid (Glu)w as chosen as ap olar and biologically relevant leaving group (Scheme 2). Also, for the phenyld erivatives,r espectivet est compounds were synthesized. Therefore, the alcohols Az-, DTA-and DAA-NDBF-OH were activated with 4-nitrophenyl chloroformate to afford the activeesters 7a-c. l-Glutamic acid then replaced the nitrophenyl moiety,r esulting in Az-, DTA-and DAA-NDBF-Glu. Figure 5. A, B) Electron detachment (lower) andattachment (upper) densities for the energetically lowest excitedstatetransitionso fDMA-NDBF-OH.For the CT state (p!p*) the density shifts from the amino to the nitro group, for the LE state (n!p*) it is localized at the nitro group. C) Franck-Condon diagramsfor the S 0 !S 1 excitation(E exc )a nd the expected relaxationp athwayv ia ac onical intersection (CI, blackt riangles) andthe resulting (radiative)d ecay E fl in vacuum or apolarenvironment, respectively (left), or in polar solvents (right).F or computational methods and data see the Experimental Section and SupportingInformation( section 6). After successful attachment of the amino acid, 1P-photolysis tests were performed in aqueous buffer (1 PBS, pH 7). For reasonso fc omparability,t he quantum yields for all derivatives were determined at 420 nm.
As an example, ap hotolysis curve of Az-NDBF-Glu is shown in Figure 6. After 2h,o nly 27 %o ft he caged compound remained (blue dots). We provedt hat photolysis of the caged glutamic acid is also possible at higher wavelengths (530 nm, green dots). The quantum yields of our "two-photon-only" PPG DMA-NDBF-LG (LG = Glu or DNA) with excitation light above 455 nm are 0%.T he F 420 values for the phenyld erivatives were determined to be 0.3 and 0.4 %. An explanation can be hindrance of planarization caused by the triphenyl structure if the CT state is of the P-ICT type.
The decrease of starting material was monitored by HPLC with the use of an internal standard. To obtain the photon flux we used two different methods:f errioxalate actinometry and our recently published fulgidep hotoswitch actinometer. [29] The quantum yields of our new PPGs are summarized in Ta ble 2.
Ac arbamate linkage has been used before for in vivo 2P-uncaging [30] -itsreleaseisslower than carbonates, but it is signifi-cantly more stable toward hydrolysis at physiological pH. [31] For ah ydrolysis test of Az-NDBF-Glu see the Supporting Information( Figure S3). After 24 ha t3 7 8Co nly 7% decrease of startingm aterial concentration was observed. Another advantage is the possibility of (spectroscopic) CO 2 released etection. Ellis-Davies et al. also studied the releaseo fg lutamate which was attached via the carboxylic acid. [32] For complete characterization, the 2P photochemistry of our derivatives was examined spectroscopically.T wo-photon-induced-fluorescence( TPiF) spectra of Az-, DTA-and DAA-NDBF-OH were recordedi nD MSO,w hich is often chosen in the literature as av iscous polar but non-protic and thus fluorescence-promoting solvent, [33] (Figure 7) and compared with DMA-NDBF-OH [22] and unsubstituted NDBF-OH. [22] The fluorescence intensity in a.u. is related to the ability to absorbt wo photons. [33] As we know,ahighf luorescences ignal is an indicator for few ultrafast decay pathways, which are the main photolysis competitors. In the experiment, all derivatives showed high-intensity signals within the phototherapeutic window, which make them possible candidates for 2P in vivo applications. In comparison with NDBF-OH,w hich has already been used for 800 nm 2P photolysis in living cells, [34] the values of the derivatives are 47 (DMA-and Az-), 32 (DTA)a nd 11 (DAA)t imes higher at 800 nm. Up to 800 nm, the stronger donor DTA,i nc omparison with DMA-NDBF,s howed the expectedh igherf luorescences ignal. In turn, we have previously shownt hat TPiF of DMA-NDBF correlates well with its 2P photolysisr ate in our strand displacement assay. [22] Az-and DMA-NDBF showa gain av ery similarb ehavior,s ot here is no reason to assume that the 2P photolysis for Az-should not be as good as for DMA-NDBF. for HPLC analysis of the amount of starting material.I rradiation was performed with mounted ThorlabsL EDs. The photon flux was determined with af ulgide photoswitch [29] (26.8 nmol s À1 at 420 nm, 93.0 nmol s À1 at 530 nm).

Conclusion
In conclusion, we designed, synthesized and characterized three new representatives for ortho-nitrobenzyl photocages in the present study.T heira bsorption profiles are bathochromically shifteda nd within the visible and less tissue-harmful spectral range maintaining the positive red-shift-effect of alkylamino donors. We want to point out, that the spectral shift is still one of the major goals of photochemists for biological applications. However,i ns trong distinction to our earlier published DMA-NDBF,w hich is interestingly inert against visible light excitation ("two-photon-only" behavior), the best derivative azetidinyl-NDBF is green (1PE) and NIR light (2PE)a ctivatable. Te chnically,t he azetidinea nd dimethylamino derivatives differ only by one carbon atom in the molecular formula, but we successfully tested the release of ab iologically relevantl eaving group (glutamic acid) with 420 and 530 nm irradiation in physiological buffer.T he two phenyld erivatives showeda ni ntermediate photolysis behavior.C ompared with DMA-NDBF they have ah igher 2P absorbance cross section. Most importantly, all these findings allowed us to learn more about excitationspecific photochemistry and optimal PPG design.

Experimental Section Synthesis
In general, all reactions were performed under argon atmosphere and in dry solvents unless otherwise specified. Solvents and reagents were purchased from commercial sources. 3-(Azetidine-1yl)phenol (2a)w as synthesized according to Ref. [19]. DMA-NDBF-OH was synthesized as earlier published. [22] Preparation of new and unpublished compounds and their characterization are provided in the Supporting Information. Microwave reactions were performed in aB iotage Initiator microwave system with matching Biotage vials. Reaction progresses were monitored by TLC analyses (silica gel 60-coated aluminum sheets, UV254 marker,M acherey-Nagel). Reaction product purifications were performed via column chromatography with silica gel 60 by Macherey-Nagel or automated flash chromatography with ap uriFlash XS 420 ULTRA system and associated prep. silica gel columns (15 mmo r3 0 mm) by Interchim. Highly polar compounds were purified by RP-HPLC (MultoKrom columns by CS Chromatographie). NMR spectra of new compounds were recorded on (250 MHz, 400 MHz, 500 MHz, or 600 MHz) Bruker instruments. ESI mass spectra were obtained with aT hermoFisher Surveyor MSQ and high resolution mass spectra (HRMS) were obtained with aM ALDI LTQO rbitrap XL instrument (ThermoScientific).

Photochemical measurements
All absorption and fluorescence emission measurements were performed in standard quartz cuvettes (1.00 cm optical pathlength, Hellma-Analytics) with various maximum volumes. UV/Vis absorption was recorded using ac ommercially available Evolution 300 (ThermoScientific) or our custom-made set up equipped with an Ocean Optics DH-mini light source and USB4000 detector,athermostatic cuvette holder (Thorlabs), all controlled by our in-house programmed PHITS (Photoswitch Irradiator Test Suite) software, which was written in LabVIEW.F or more details see Reinfelds et al. [29] This setup and software were also used for our chemical actinometry.R eference compound was an indolylfulgide photoswitch. Ac oncentrated solution of the fulgide (500-1000 mm)w as irradiated with the respective light source (Thorlabs mounted LED, l max = 420 nm or 530 nm) to convert the photoswitch from its 1Z form to 1C or the other way round. Afterward, the caged glutamic acid of interest could be irradiated with known photon flux. Steady-state fluorescence emission was recorded using aH itachi F-4500 spectrophotometer.T he optical density (OD) was set lower than or equal to 0.1 for fluorescence spectra, otherwise checked for consistency.D etails of the set ups for two-photon induced spectroscopy have been described previously. [22,35,36] See the Supporting Information for additional data.