Light‐Induced Uncaging for Time‐Resolved Observations of Biochemical Reactions by MAS NMR Spectroscopy

Abstract Light‐induced activation of biomolecules by uncaging of photolabile protection groups has found many applications for triggering biochemical reactions with minimal perturbations directly within cells. Such an approach might also offer unique advantages for solid‐state NMR experiments on membrane proteins for initiating reactions within or at the membrane directly within the closed MAS rotor. Herein, we demonstrate that the integral membrane protein E. coli diacylglycerol kinase (DgkA), which catalyzes the phosphorylation of diacylglycerol, can be controlled by light under MAS‐NMR conditions. Uncaging of NPE‐ATP or of lipid substrate NPE‐DOG by in situ illumination triggers its enzymatic activity, which can be monitored by real‐time 31P‐MAS NMR. This proof‐of‐concept illustrates that combining MAS‐NMR with uncaging strategies and illumination methods offers new possibilities for controlling biochemical reactions at or within lipid bilayers.

The development of light-induced uncagings trategies forb iochemicals ubstrates, which have been inactivated with ap hotolabile protection group, enables ar ange of experiments with high spatial and temporalr esolution especially in the cellular context.Abroad set of tools for the caging of biologically relevant compoundsh as been developed enablingt oa ddress a wide range of biological applications. [1] The high versatility of the uncaging approach with respectt ot emporal control and wavelength selectivity also provides av ariety of opportunities for NMR spectroscopy or other biophysical methods for in situ triggering of enzymatic reactions, folding eventso ro ligomeri-zation/complex formation.I namolecular biophysical context, caged compounds have been utilized for example in NMR spectroscopy of soluble samples to induce folding of proteins, DNA and RNA or for enzymes tudies. [2] Also lipids have been an important target for developing uncaging strategies. [1f, 3] Solid-state NMR is extensively used for the investigation of lipids andm embrane proteins within intact bilayers, but uncagingh as not been exploredy et for these applications. It could be advantageous because solid-stateN MR relies on fast sample rotation at the magic angle using sealed rotors, which makest he addition of substrates during the experiment almosti mpossible. Pre-mixing before the NMR experiment followed by fast sample transfer into the magnet is in principle possible but requires tailoring of the experimental conditions towardss low kinetics and relies on ag ood distribution of the substrate within ah eterogeneous proteoliposome sample. This becomes especially challenging for example for hydrophobic compounds such as lipid substrates or for targeting binding sites within the lumen of sealed liposomes.
To test whether these limitations could indeed be addressed by uncaging, we have chosen the E. coli membrane protein diacylglycerol kinase (DgkA), which phosphorylates diacylglycerol under ATPc onsumption ( Figure 1). Its homotrimerics tructure was determined by X-ray crystallography in lipidic cubic phases. [4] Its interfacial enzymatic reaction has been observed with time-resolved 31 P-MASN MR [5] and its secondary structure, [6] trimer symmetry and protomer interactionsw ithin the lipid bilayer were resolved by 3D-and DNP-enhanced MAS-NMR. [7] In mammalian cells, DAGs act as second messenger and get phosphorylated by lipid kinases, which are structurally rather distinct from the E. coli variant.A ltered functions of individual DgkA isoforms have been implicated in ar ange of diseases, which requires ab etter understanding of their function. [8] Therefore, developing tools by which such reactions could be studied directly within the membrane interface could have wide implications.
Here, the DgkA activity has been controlled by either uncaging NPE-ATP ( Figure 1a)o rb yr eleasing NPE-DOG, aD AG variant ( Figure 1b). Therefore, arobustand cost-effective illumination setup for efficient in situ illumination under MAS at ah igh magnetic field was established. Similart ot he illumination setup previously described for photo-CIDNP, [9] af iber bundle with am acor ferrule was inserted into the MAS statorf rom beneath through ahole in the coil pedestal ( Figure 1c). Astretched radiofrequency coil geometry was used to enablee fficient illumination of the sample volume, which was restricted to 15 mLi nt he center of the MAS rotor by insertion of rubber disks ( Figure S1).
Using ah igh radiance UV LED as light source with ap eak wavelength of 365 nm, 20 mW of UV light were availablea t the end of the fiber bundlef or sample illumination. This setup was first tested on as ample containing DOPC liposomes and NPE-ATP.W ithin minutes, NPE-ATP could be successfully uncaged under MAS-NMR conditions and the reactionc ould be monitored by 31 P-NMR direct detection( Figure S2). As the availablesignal-to-noise ratio limits the time resolution, the observed uncaging rate should be sufficientf or ar ange of applications.H owever,l iposomes as well as the uncaged NPE-group absorb light and therefore reduce the uncaginge fficiency at these concentrations to approx.80%.
Uncaging of NPE-ATP was then carried out in the presence of DgkA ( Figure 2) to induce its enzymatic activity.T he lipid substrate DAG or other long-chain variants were omitted initially,s ot hat only basal ATPh ydrolysis occurs. [5,10] Indeed, as shown in Figure 2a,u ncaging of NPE-ATP by light is followed by ATPc onsumption and build-up of ADP and Pi. Build-up rates are comparable to those observed by us before under similar conditions butw ithout uncaging of NPE-ATP. [5] Within 5min an uncaging efficiency of approx. 65 %w as achieved (see Figure S2). The amount of DgkA within the liposomes was chosen so that the bulk turnover of uncaged ATPb yD gkA is significantly slower than the uncaging reaction. As ac ontrol, the same experiment was performed in the absence of Mg 2 + for which uncaging but no ATPh ydrolysis could be observed ( Figure S3).
The data in Figure 2d emonstrate the feasibility of triggering an enzymatic reaction by light-inducedu ncaging of NPE-ATP in the presence of DgkA proteoliposomes. In order to general-ize this approach, it would be desirable to bring also al ongchain diacyclglycerol lipid substrate such as dioleoylglycerol (DOG) under light control. It is highly hydrophobic and cannot be added by simple mixingbut would have to be incorporated already at the stage of liposomef ormation. [11] DOG was therefore protected with an NPE group at the hydroxyl moiety, which prevents phosphorylation by DgkA without uncaging. [3] The NPE group was connected via an oxycarbonyl linker,initially used for caging nucleoside 5'-hydroxyls, in ordert oe nhance the uncaging efficiency as hydroxyls are poor leaving groups compared to phosphates (see SI for further details on the synthesis). [12] Unlike NPE-ATP,N PE-DOG has no 31 Ps pectroscopicm arker as direct NMR-readout for successful uncaging via solid-state NMR. However,s uccessful uncaging can be shown indirectly: In the presenceo fl ipid substrate, the basal ATPase activity of DgkA gets stimulated and turns into ap hosphoryl transfer reaction. [5] As ar esult,a ni ncrease in ATPc onsumption but ad ecrease in Pi productioni so bserved as the g-phosphoryl group is transferred to DOG.This is indeed the case upon illumination of as ample containingD gkA within DOPC and NPE-DOG bilayers ( Figure S4): Upon addition of Mg.ATP,A TPase activity is observed by the build-up of Pi. Irradiation with UV light uncages NPE-DOG, which leads to an increasei nA TP turnover but not Pi.
Chem.E ur.J.2020, 26,6789 -6792 www.chemeurj.org 2020 The Authors. Published by Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim nance. We have therefore repeated the experiment but replaced Mg.ATP by Mg.ATPgSa sn ucleotide substrate. Here, the thiophosphoryl group of ATPgSi st ransferred to uncaged DOG and the resulting lipid product thiophosphatidic acid (ThioPA) is significantly downfield shiftedfrom the main lipid resonance. Indeed,a fter illumination and uncaging of NPE-DOG, an increasings ignal at 44 ppm is detected confirming successful uncagingo fN PE-DOG and subsequentp hosphorylation of the released DOG within the membrane (Figure 3a nd Figure S5). The small ThioPAs ignal observable before illumination causes ab aselineo ffset and can be attributed to slightly incomplete caging of DOG, which has been thiophoshorylated upon addition of ATPgSd uring the dead time of the experiment. Additional purifications teps might furtherd ecreaset he fractiono f lipid educt if required. [3] Hydrolysis of ATPgSb yD gkA is knownt ob ea no rder of magnitude faster compared to ATP. [5] Therefore, stimulated hydrolysis of ATPgSu pon uncaging could also be seen in these experiments as build-up of thiophosphate increased after illumination. However,t he amounto fl ipid product formed after illumination reveals that only as mall fraction of DOG (4 %w ith respectt ot he amount of incorporated NPE-DOG, calculated by comparison with the DOPC integral) was thiophosphorylated. One reason is probably limited accessibility of the DgkA binding sites for uncaged DOG. An additional factor could be insufficient uncaging of NPE-DOG. It can be assumed that uncaging of NPE-DOG is less efficient compared to NPE-ATP as primary alcohols caged with carbonate derivatives of highly efficient Coumarin based photocages exhibit ap oor photolysis efficiency comparedt ogPc aged ATPd erivatives. [13] Despite efficient illumination of the small active volumeo f the MAS rotors used by high-performanceL EDs, uncagingi s relativelys low and incomplete. One reasoni sp robably the relativelyh igh concentration (> 20 mm with respect to total sample volume) of the caged compoundsw ithin the liposome sample in the MAS rotor and subsequentc ompeting light absorptionb yt he leaving group. It has been demonstrated under solution-state NMR conditions, that submillimolar concentrationso fc aged compounds can be releasedw ithin seconds by laser illumination setups capable of delivering several watts of radiantf lux. [2a, 14] Ah igherr adiant flux could therefore also for MAS NMR experiments be beneficial to achieve higher uncaging efficiencies at reasonable concentrations and illumination times.
The performed experiments demonstrate that biochemical reactionss tudied by solid-state NMR can be brought under light control using caged compounds. The main advantage of light triggered reactions as demonstrated by uncaging al ipid substrate will thereby lie on initiation of reactions that cannot be started by mixingorwhen components have to be prevented from reacting durings ample preparation.T his proof-of-concept illustrates that combining MAS-NMR with uncaging strategies andi llumination methods offersanew possibility for controlling biochemical reactions in situ.

Experimental Section
General The E. coli dgkA wild-type gene carrying aN -terminal hexa-His tag sequence was cloned from pSD005 into Novagen) by changing the HindIII recognition sequence to NdeI and inserting the gene between the NcoI and NdeI recognition sequence. Production and purification of DgkA were performed as described previously [15] with minor changes mentioned below (for SDS-PAGE see Figure S6). For solubilization and purification Empigen BB was replaced by OG in the same weight amounts. The HEPES concentration in all buffers was 50 mm and LiCl was replaced by 30 mm NaCl. The protein was eluted from the Ni-NTAr esin with 0.4 m imidazole, 0.1 %( w/v) DDM, 50 mm HEPES pH 7.5, 30 mm NaCl and 1mm BHT.T he yield for DgkA was typically 50 mg per liter of culture. For reconstitution into liposomes, DOPC and NPE-DOG were dissolved in chloroform/methanol 3:1( v/v), dried and dissolved in buffer (50 mm HEPES pH 7.5 and 30 mm NaCl). Liposomes were destabilized with 3mm DDM before performing freeze-thaw cycles. The NPE-DOG containing liposomes (100 mm HEPES pH 7.5 and 30 mm NaCl) were first sonicated for 10 min in an ultrasonic bath before addition of DDM and freeze-thaw cycling. Purified DgkA was reconstituted into the liposomes at the desired lipid to protein molar ratio under slow stirring at room temperature. Detergent removal and imidazole removal was performed as described [7] before. Proteoliposome samples were pelleted and resuspended at the desired concentration by vortexing before being transferred into a3.2 mm sapphire MAS rotor. Solid-state NMR experiments were performed on aB ruker Avance III 850 WB spectrometer operating at 850.31 MHz 1 Hf requency using 10 kHz MAS rate and 30 8C. The sample temperature was ref- eal-timeM AS NMRe xperiments on DgkA in DOPC liposomes containing NPE-DOG andA TPgS. Basal ATPase activityi so bserved. Upon illuminationf or 5min performedduring acquisitionoft he spectrumm arked in red, NPE-DOGgets uncaged and an increasing signal of ThioPAiso bserved at 44 ppm. The asterisk denotes at hiophosphate (ThioPi)s ide-product (see Figure S5b). b) Time traces of the 31 Preal-timeN MR experiment depict basal ATPase activity before uncaging of NPE-DOG andk inase activity in conjunction with enhanced ATPase activity after uncaging as seen by formation of ThioPA and enhanced built-upo fT hioPi (mainly observed in form of its sideproduct (*)). The sample contained75nmol ATPgSa nd 13.3 mgD gkA reconstituted into DOPC liposomes with 20 mol %N PE-DOG (L:P 2000:1, 50 mm HEPES, pH 7,5,30mm NaCl and2:1 molarr atio MgCl 2 :ATP). The sample volumew as 15 mL. Spectra were recorded at 30 8CataM AS rate of 10 kHz. erenced via KBr T 1 relaxation measurements at the same spinning speed. To limit sample heating no decoupling was used. 31 Ps olidstate NMR spectra are referenced to H 3 PO 4 using ac hemical shift of 58.62 ppm for triethylphosphine as external reference. For 31 P real-time NMR measurements pH of ATPa nd ATPgSs olutions was adjusted to 7.5. NPE-ATP was dissolved in 100 mm HEPES pH7.5. MgCl 2 was added to the nucleotide solutions in two-fold molar excess as lower ratios lead to severe 31 Pl ine broadening of ATP and ADP signals. Nucleotide was added to the rotor containing the desired proteoliposome amount directly before measurements. Insitu illumination was achieved using am odified 3.2 mm DVT MAS NMR probe containing ac ustom-built 2mmd iameter LUV 70 mm fiber bundle with macor ferrules (Leoni Fibertech). AU VL ED with 500 mW output power connected to ac omputer-controlled LED driver (LCS-0365-11-22, SLC-AV02-US, Mightex Systems) in combination with al ightguide adapter (LCS-LGA22-0515, Mightex Systems) was used for illumination at ap eak wavelength of 365 nm (see Figure S1). To initiate photocleavage and induce the reaction, samples were typically illuminated for 5min. The output power at the end of the fiber guide was determined with al aser thermal power sensor (P/N 1Z02146, Ophir) connected to aN ova Display power meter (1Z01500, Ophir). Time traces were generated by integration of the respective peak area and scaled to the amount of added nucleotide.