A red-shifted two-photon-only caging group for three-dimensional photorelease

With a new photolabile protecting group – exclusively cleavable by two-photon-excitation – complex light scenarios for three-dimensional uncaging are possible.


Introduction
Over the last decades photolabile protecting groups (PPGs) became a frequently used tool to regulate bioactive molecules 1 such as neurotransmitters, 2-5 hormones 6,7 and even macromolecules like proteins 8,9 and oligonucleotides. [10][11][12] One crucial long-term goal is a red-shiing 13 of the lightinduced photorelease (uncaging 14 ) into the therapeutic window ($650-950 nm (ref. 15)). It is less harmless for living cells and deeper tissue penetration becomes possible in biological applications due to less absorption and scattering of e.g. blood. 16 In the 1990s a promising (un)caging strategy for higher wavelengths (>650 nm) has emerged, based on two-photon (2P) sensitive photolabile groups. [17][18][19][20] PPGs with 2P-absorption character are cleavable with femtosecond pulsed lasers. This non-linear optical process can be seen as a simultaneous absorption of two photons. In many casesbut not all 21the resulting electronically excited state is the same when photons of half the energy are used. This process was rst described by Maria Göppert-Mayer. 22 It can be used to realize photochemistry with 3D spatial resolution since the excitation depends on the squared intensity (p $ I 2 ). Excitation volumes can be as small as a femtoliter. 16 The 2P-uncaging efficiency d u can be described (analogous to the "quantum product" in 1P-uncaging) by the product of absorption cross section d a and the uncaging quantum yield F u (d u ¼ d a $F u ). 16 The d a of chromophores depends on the length and planarity of the p-electron system and substituent effects (i.e. push-pull-systems). 16,23,24 The positions of the substituents are crucial: a dipolar character as well as quadrupolar or octupolar enhance the 2P-absorption. 23 In 2006, Ellis-Davies et al. introduced the new chromophore 3-nitrodibenzofuran (NDBF) for 2P-photolysis of NDBF-EGTA:Ca 2+ with d u ¼ 0.6 GM at 720 nm. 25 In this paper, based on the NDBF core (see compound 1, Fig. 1), we rationally designed and synthesized the dimethylamino derivate DMA-NDBF-OH (2, Fig. 1). Due to calculations with the DFT/B3LYP method and a 6-31*G basis set for the ground state equilibrium structures and TDDFT/BHLYP for exited states the dipolar structure should be red-shied and have an increased d a . 26 Despite the availability of sophisticated computational methods the optimisation of 2P-chromophors remains a formidable challenge.

Experimental and results
The simulation of various NDBF derivatives predicted a preferred substitution of ring-position 7 with a dimethylamino (DMA) functionality as donor (Fig. 1). The expected Fig. 1 The caging group precursor NDBF-OH (1) and its new dimethylamino derivative DMA-NDBF-OH (2). uncaging efficiency of DMA-NDBF based on computed 2P-absorption spectra should be more than 20 times higher than the one of NDBF at its respective red-shied maximum. 26
For 1PE characterisation of DMA-NDBF-OH (2) an absorption spectrum was recorded ( Fig. 2) in DMSO and compared to the one of unsubstituted NDBF-OH (1). The absorption maximum of 2 is shied bathochromically from 312 to 424 nm. With 15 947 L mol À1 cm À1 at 424 nm, the molar absorption is 98 times higher than the one of compound 1 (at 420 nm it is 79 times higher).
Then, 2P-uorescence excitation (TPE) spectra of 1 and 2 were determined. Fig. 3 shows the relative uorescence intensities observed in the visible spectral range ($400-700 nm) aer 2PE of 1 (cyan) and 2 (magenta) in DMSO using a wavelength range between 770 and 1060 nm. We observed a generally higher responsiveness to 2PE for 2 than for 1. The resulting uorescence intensity for 2 at 840 nm e.g. is 40 times higher and should be proportional to the absorption cross section d a of compound 2. Of course, these 2P-uorescence excitation spectra do not directly reect the 2P-uncaging efficiency d u as there is generally no direct relationship between a molecules uorescence quantum yield F f and the quantum yield for the uncaging photochemistry F u . However, the observation of high uorescence intensities aer two-photon excitation is a strong indication that compound 2 can indeed be two-photon excited quite readily, which is obviously a decisive prerequisite for any subsequent uncaging photochemistry (for more details see the discussion below and the ESI †).   protected with tert-butyldimethylsilyl (TBDMS) before it was reacted with 9, 4-DMAP and DIPEA in DMF. Aer a nearly quantitative TBAF-deprotection of the alcohols in THF the 5 0 -OH was protected with 4,4 0 -dimethoxytrityl (DMTr). Phosphoramidite 11 was obtained in a reaction with DIPEA and 2-cyanoethyl-N,N 0 -diisopropylchlorophosphoramidite in CH 2 Cl 2 .
Test-sequences DNA1 and DNA2, 15-mers with the cage at position 8, were used for rst investigations of 1P-absorption behaviour aer irradiation (Fig. 4) and 1P-photolysis (Fig. 5). All irradiation-tests were performed with a concentration of 20 mM in PBS. The 1P-photolysis of DNA1 was found to be signicantly faster at every tested wavelength compared to DNA2. For example, aer 1200 s irradiation at 385 nm DNA1 was quantitatively uncaged, whereas 77% starting material remained in case of DNA2. Aer the same time of irradiation at 420 nm we observed 87% remaining starting material for DNA1 and 97% of DNA2, even though DMA-NDBF has its absorption maximum at 424 nm. Further experiments revealed that with wavelengths >420 nm (i.e. 455 nm) no 1P-photolysis could be detected at all for DNA2regardless of the power we used! The 1P-quantum    yield F 0 420 for DNA1 with a dA NDBF residue was found to be 13.6% and for DNA2 with dA DMA-NDBF 0.05%. 27 The respective quantum yields F 0 340 were 24.05% and 1.10%, showing that irradiation into the transition of DMA-NDBF at around 340 nm results in some extent of uncaging whereas the lower-energy transition does not (see also ESI † for a more detailed analysis).
Compound 2 shows some similarity to an amino-substituted ortho-nitrobenzyl caging group which has been investigated by Bochet et al. 28 In their case only prior protonation of the amino group led to a state with productive photolysis while irradiation into the unprotonated state yielded a charge transfer transition instead of uncaging. Investigations addressing the charge transfer character in the photochemistry of compound 2 can be found in the ESI. † However, even at pH 2 there was no conversion of DNA2 upon irradiation at 455 nm. 29 For the 2-photon-characterisation of the photocages in DNA strands we used our recently published hydrogel-uorescenceassay on a confocal microscope and laser setup (ESI †). 30,31 DNA3 and DNA4 with either dA NDBF or dA DMA-NDBF residues, respectively, were immobilised via thiol-linkers in a hydrogel (PVA-PEG), as illustrated in Scheme 3. Then the gel was soaked with a buffered solution containing a duplex of a 15-mer-strand with a 5 0 -terminally attached uorophore (ATTO565) and a 3 0 -quencher (BHQ2) 11-mer strand. The quencher strand displacement aer uncaging of DNA3 or DNA4 led to increased uorescence in the focal plane. The uncaging-wavelengths between 720 and 980 nm were generated with a pulsed Ti:sapphire laser for 2P-excitation. For each of the wavelengths investigated one line was written into the hydrogel (Fig. 6b) and the resulting uorescence was quantied (Fig. 6a). The power was kept constant for every line. The form of the pink spectrum in Fig. 6a (with a maximum at 840 nm) resembles very much the one of the pink spectrum in Fig. 2 (with a maximum at 424 nm). However, 1P-irradiation at 420 nm results in a poor conversion whereas 2P-irradiation at 2 Â 420 nm efficiently produces the desired uncaged product strand.
Apparently, DMA-NDBF is one of the few cases where 1PE and 2PE with twice the wavelength do not result in the same photochemical behaviour. In our case, the excited state aer 1PE appears to have a low uncaging quantum yield, i.e. this caging group can be considered as "one-photon-stable" (at least in the visible range) whereas aer 2PE the intended uncaging reaction readily occurs with irradiation conditions that have been shown to be compatible with living cells. 30 Because of its low 1P-photolysis rates, we decided to use DMA-NDBF as a "2P-only-cage" which should be applicable for complex orthogonal uncaging together with various 1P-cagesespecially also red-shied ones that are currently the focus of much attention. We used DNA4 with dA DMA-NDBF residues and DNA5 with a different sequence and dA NDBF residues for individual addressing together in the same hydrogel. For a twocolour read-out ATTO565 and ATTORho14 were used as uorophores F1 and F2 and BHQ2 (Q1) and BBQ-650III (Q2) as matching quencher pairs (for an overview see Scheme S2 in the ESI †). Fig. 7 provides an overview of optimised irradiationconditions, tested for the triply caged strands. With 420 nm (1PE, 780 nW, 2 scans) it was possible to only uncage the NDBF, with 730 nm (2PE, 15 mW, 5 scans) bothhowever DMA-NDBF much more efficiently than NDBFand 840 nm (2PE, 15 mW, 5 scans) only DMA-NDBF, while leaving the other cage intact in the same hydrogel. Based on these results, seven rectangular shapes (steps) were written into the hydrogel with 840 nm (Fig. 8), as well as a circular shape with 420 nm. The laser beam Scheme 3 Hydrogel-fluorescence-assay to monitor uncaging. After irradiation with a defined wavelength the triply caged antisense strand becomes uncaged and is able to bind the 5 0 -fluorophore-labelled strand. By competition, the counter strand with the quencher is removed, the fluorescence increases.  direction followed the z-axis. The 2P-steps with spatial resolution in z had a distance of 25 mm between them. The circular 1Pirradiation resulted in the cylindrical staircase core shown in Fig. 8. A z-stack (Fig. S4 †) with two detection channels (Ch. 1: uorescence excitation 543 nm, detection at 557-612 nm; Ch. 2: excitation 633 nm, detection 671-721 nm) was imaged at laser setup 2 (ESI †). Fig. 8a shows the magenta-channel 1, cyanchannel 2 and the overlaid combination in the x/y-plane. For Fig. 8b and c the staircase was rotated in the z-/y-plane. The colours in Fig. 8c demonstrate the height in the hydrogel. The laser powers used were in a range compatible with living cells. For instance, 18-24 mW were found to be tolerated by HEK cells, dorsal root ganglia and liver cells for 10-20 scans. 32

Conclusion
In summary we designed, synthesised, and characterised a new dimethylamino derivative of the NDBF photolabile protecting group (PPG). It shows a red-shied one-and two-photon absorption compared to the NDBF groupwhich is important for biological applications. As predicted by theoretical calculations the DMA-NDBF PPG shows a better two-photon photolysis behaviour compared to NDBF. However, to our surprise it turned out that the one-photon photolysis efficiency of DMA-NDBF is rather poor, especially for wavelengths beyond 400 nm. Based on our calculations we propose that this is a rare case of excitation-specic photochemistry. Both 1PE and 2PE at twice the wavelength populate the same excited state, since the S 1 exhibits substantial one-photon oscillator strength as well as two-photon absorption cross section. 26 However, different photochemical behaviour is induced, because 1PE and 2PE electronic excitations couple to different molecular vibrations. 33 This unusual "two-photon-only" behaviour offers interesting applications for light-regulation scenarios with increased complexity. We had recently presented orthogonal two-colour, two-photon uncaging 30 where we had to carefully control the two-photon irradiation conditions (especially the power). With DMA-NDBF, we can now perform efficient two-photon uncaging with red light leaving a broad spectral window open for orthogonal 1P-uncaging in the red part of the spectrum. In addition, 1PE can now be performed before 2PE with DMA-DNBF, which is surprisingly "one-photon-stable". With previous two-photon caging groups, the 2PE had to be performed as rst photochemical operation. This adds another degree of freedom to ever more complex scenarios of complex light control. 34,35 Its one-photon-stability and yet easy and selective photolysis under 2PE, its red-shied 2P absorption which lies perfectly in the therapeutic window and its perfect stability to regular ambient light make the DMA-NDBF a very interesting caging group for future biological applications.

Conflicts of interest
There are no conicts of interest to declare.