Covalent Attachment of Active Enzymes to Upconversion Phosphors Allows Ratiometric Detection of Substrates

Abstract Upconverting phosphors (UCPs) convert multiple low energy photons into higher energy emission via the process of photon upconversion and offer an attractive alternative to organic fluorophores for use as luminescent probes. Here, UCPs were capped with functionalized silica in order to provide a surface to covalently conjugate proteins with surface‐accessible cysteines. Variants of green fluorescent protein (GFP) and the flavoenzyme pentaerythritol tetranitrate reductase (PETNR) were then attached via maleimide‐thiol coupling in order to allow energy transfer from the UCP to the GFP or flavin cofactor of PETNR, respectively. PETNR retains its activity when coupled to the UCPs, which allows reversible detection of enzyme substrates via ratiometric sensing of the enzyme redox state.

and environmental sensing applications than by monitoring enzymeso rs ubstrates directly.
We have previously demonstrated significant diffusion-controlled quenching of UCP upconversion emission by the oxidized flavin cofactor of the enzymes pentaerythritol tetranitrate reductase (PETNR) [8] and glucose oxidase, [9] as well as to vitamin B 12 ,a nd the heme cofactor of cytochrome c. [9] This quenching could be ar esult of an emission-reabsorption (secondary inner filter effect) process and/ord irect energy transfer from UCP to chromophorei ftheir separation (Fçrster distance) is sufficiently short (i.e. quenching via FRET or LRET). As the photophysical mechanism of UCP quenching is not the focus of the present study,w ew ill collectively refer to the UCP quenching process as apparent energyt ransfer (AET). Previous studies have also demonstrated covalent attachment of biomolecules to UCP surfaces, [10] but have not exploited AET from the UCP as as pectroscopicp robe.T he closest example used AET from UCPs to glucose oxidase immobilized on poly(acrylamide) for flow-based applications. [11] Here, we have now created covalentU CP-protein/enzymec onjugates that undergo intra-system AET from the UCP to the protein cofactor while suspended in aqueous solution.W ec hose two exemplar proteins with different intrinsic chromophores:e nhanced green fluorescent protein (GFP) and PETNR.T he methodology should be applicable to any protein that possesses an ative or engineereds urface-exposed cysteiner esidue, so can be adopted by those currently using thiol or maleimide-based organic fluorescent probes. [12] GFP contains the chromophore p-hydroxybenzylidene-2,3-dimethylimidazolidine (HBDI), [13] which has absorptionm aximaa t 395 and 475 nm ( Figure 1). The latter absorption band overlaps with the 475 nm emission band ( 1 G 4 ! 3 H 6 transition) of Tm IIIdoped UCPs, and therefore has the potentialf or efficient AET Scheme1.Simplified representation of overall synthetic scheme;i )Igepal CO-520, NH 4 OH,TEOS,cyclohexane;i i) APTES, cyclohexane;iii)Sulfo-SMCC, GFP,PBS;iv) Sulfo-SMCC, PETNR,PBS;v)KBr,FMN, PBS;see Supporting Information for additionale xperimental details.I nset shows "on-off" apparent energytransfer (AET) concept with PETNR on the surface ofU CPs. (C) Solid-state UV/Vis reflectances pectraofG FP,UCP,and UCP GFP separately drop-cast and dried between two glass slides, with UCP GFP displaying distinct GFP bands. (D) Upconversion emission spectra of UCP and UCP GFP (l ex = 980 nm) in PBS. Bands are normalizedt ot he 800 nm UC emission intensity with full spectra showninF igure S3. (Inset) PhotographofU CP GFP showingd istinct luminousyellow-green coloration from conjugated GFP.U niversallegend for all panels: green = GFP,blue = UCP (PTIR-475), red = UCP GFP . Chem. Eur.J.2020, 26,1 4817-14822 www.chemeurj.org 2020 The Authors. Published by Wiley-VCH GmbH from UCP to GFP ( Figure 1A). Enhanced GFP contains two cysteine residues with one, C48, partially solvent-exposed (Figure S1). Initially,G FP was covalently attachedt ot he surfaceo f maleimide-capped ytterbium(III)-thulium(III) doped gadolinium oxysulfide UCPs (Gd 2 SO 2 :Yb:Tm, PTIR-475) via direct maleimide-thiol chemistry.H owever,a st he UC emissivep rocess is notoriously capricious and easy to quench, we found that any form of direct surface modification with maleimide-containing groups led to loss of UC. Therefore, we first coated the UCPs in as ilica layer.T his layer of silica provides multiple benefits:p rotection of the UCP surface against quenching processes;f airly robust biocompatibility and the ability to functionalize further with ease.
PTIR-475 UCPs were capped with silica using ar eversem icroemulsion synthesis, with IGEPAL CO-520 used to stabilize the procedure during the polymerization of tetraethyl orthosilicate. [14] Then, (3-aminopropyl)triethoxysilane (APTES) was added to the reaction mixture. APTES combines with the silica coating to present an accessible surfacel ayer of primary amines,w hich can be modified relatively easily using N-hydroxysuccinimide (NHS) esters. In order to covalently couple protein cysteines to the amine-coatedU CPs, sulfosuccinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate (sulfo-SMCC) was employeda salinker.T his linker contains both an NHSester and amaleimide and was first allowed to couple( via maleimide-cysteine conjugation) to the protein before the addition of the APTES-coated UCPs.T he reactionm ixture was gently agitated under mild conditions to allow the coupling to progress and after each stage of this multi-step procedure the UCPs were centrifuged and washeds everal times to remove unreactedr eagents;t he overall synthetic Schemei ss ummarized in Scheme1.T he final particles, UCP GFP ,w ere isolateda sa luminous yellow-green powder( Figure 1D,i nset). The average sizes of the unmodified UCP and surface modified UCP APTES UCP GFP particles were determined by dynamic light scattering (DLS) andt ransmission electron microscopy (TEM) measurements and are collatedi nF igures S9 and Ta ble S1 in the Supporting Information. AveragedT EM measurements give particle sizes of 765 nm (UCP), 809 nm (UCP APTES )a nd 878 nm (UCP GFP ).
UV-visible absorption spectra were recordedf rom UCP GFP particles suspended in phosphate buffered saline (PBS). While the particles cause as ignificant amount of scattering,adistinct peak is observed around4 80 nm ( Figure 1B), characteristic of GFP absorption and this band is not observed in the unconjugated UCPs. Likewise, the solid-state UV-vis reflectance spectrum of UCP GFP (Figure1C) also showst he characteristic3 95 and 475 nm bands arising from GFP.T he relative intensity of these bands differs to the those of GFP in solution,l ikely due to scattering from UCP GFP .D irect excitation of the GFP fluorophore at 475 nm gives rise to fluorescence emission at 530 nm, characteristic of GFP ( Figure S2). Excitation of the particles with 980 nm light leads to UCP emission bands at 475, 650, and 800 nm, corresponding to the Tm III transitions of 1 G 4 ! 3 H 6 , 1 G 4 ! 3 F 4 ,a nd 3 H 4 ! 3 H 6 ,r espectively. [15] Comparison of the UC emission from UCP GFP relative to the APTES-coated UCP showeda% 60 %r eduction in emission intensity of the 475 nm band, when normalized to the 800 nm peak ( 3 H 4 ! 3 H 6 transition;F igures 1D and S3). This decrease in emission is consistent with AET from the UCP to GFP,b ut no emission from GFP at 530 nm waso bserved, even at long accumulationt imes, [8] suggesting that fluorescencef rom those GFP moieties acting as AET acceptors is efficiently quenched;t his is likely to be due in part to ac onsiderable reduction of the GFP quantum yield. Unfortunately,a sp reviously observed, we were not able to infer any energy transfer from upconverted emission lifetime measurements. [8,9] Using UV-visible spectroscopy ( Figure 1B), we estimate 3.9 nanomoles of GFP are conjugated to 1mgo f UCP,r esulting in aw orking concentration of 3.9 mm GFP in the 1mgmL À1 UCP solutions (see Supporting Information).
If AET occurs by FRET or LRET,t hen incomplete quenching of UCP emission at 475 nm may be due to inefficient energy transfer from Tm III sites distant from the UCP surfacea nd/or due to emission from as mall population of unreacted UCPs (unobserved by TEM analysis). UCP conjugates were also prepared with synthesized nanoparticles of smaller diametero f % 20 nm ( mal UCP,s ee the Supporting Information), which are small enough for FRET or LRET to occur.S imilar AET behavior was observed ( Figure S4), but the smaller particles required a non-ideal,m uch higherp owerl aser sourcest og enerate comparable UC emission ( % 1W vs. 45 mW). Consequently,t he commercial PTIR-475 microparticles were used for the remainder of the current study.
Following the successful generation of UCP GFP ,asimilar synthetic route was used to conjugate PETNR to the UCPs. PETNR is an NAD(P)H [reduced nicotinamide adenine dinucleotide (phosphate)]-dependent enzyme, [16] whichp ossesses an ative surface-accessible cysteine (Cys222; FigureS1) that is reactive towardst hiol and maleimide derivatives of organic fluorophores. [17] Like GFP,t he 465 nm absorption band of the oxidized FMN cofactor of PETNR has good spectral overlap with the UCP Tm III emission band at 475 nm. [8,9] Upon reduction of PETNR by NAD(P)H, the 465 nm FMN absorption is lost, largely abolishing AET from the UCP,l eading to an increase in UC emission from the UCP at 475 nm. Comparing intensity ratios with the 800 nm emission of the UCPs thereby providesaratiometricd escription of the redox state of the FMN bound to PETNR. [8] During the coupling procedure, much of the relatively weakly-bound (non-covalent) FMN cofactor disassociates from the enzyme. The FMN can be reincorporated into the apoenzyme, [18] however,b ys oaking the apoenzyme-conjugated UCPsystem (UCP apo-PETNR )i nasolutionc ontaining an excess of FMN and 1 m KBr to assist in FMN binding. [18] After 24 hours of gentle agitation at 4 8C, the resulting particles, UCP PETNR ,w ere isolated by centrifugation and repeatedly washed until FMN was no longerp resent in the supernatant ( Figure S5). Throughout this procedure the color of the UCPs progressed from white (UCP APTES ), to straw-yellow (UCP apo-PETNR ), to ac haracteristic deep yellow/orange in UCP PETNR (Figure 2 The UC emissionspectra of UCP APTES ,UCP apo-PETNR and UCP PETNR are shown in Figure 2. Again, they all show the typical Tm III emission at 475, 650, and 800 nm and were normalized to the 800 nm peak for comparison. There is a % 60 %q uenching of the 475 nm UC emission in UCP PETNR ,c onsistentw ithA ET from the UCP to the FMN cofactor in PETNR.S omeq uenching is also observed in UCP apo-PETNR ,l ikely due to low levels of bound FMN in this sample and/ors ome quenching of the UCP by e.g.,v ibrational relaxation due to the presence of the apoprotein. There has been somec ontention as to the exact nature of AET in UCP systems, with some evidence that it is highly dependentont he nature of the size and lattice of the UCP donor and the distance of emitter ions to the acceptor. [19] The data here show the addition of the FMN cofactor to UCP apo-PETNR leads to significant quenching of the UC emission. This is consistent with quenching by AET to an acceptor chromophore with good spectralo verlap with the UCP emission band(s), so this approach should be applicable to other suitable chromophores.
Direct excitation of the FMN in UCP PETNR at 448 nm shows characteristic flavin emission at % 530 nm (Figure 2), further indicatings uccessful functionalization of the UCPs. The companion excitation spectrums hows the expected FMN excitation superimposed with fine structure, which is due to Tm III emission from the UCPs at this wavelength (Figure2,l ower panel inset).
While thesed ata collectivelys uggestw ehave successfully conjugated PETNR to the surfaceo fU CPs, in order to be useful as modelb iosensor,the enzymeneeds to retain its catalytic activity when bound to aU CP.C onsequently,t he steady-state kinetics of UCP PETNR were assessed. The simplified 2-step reaction of PETNR is shown in Equations (1) and (2) (Simplified reaction Scheme for PETNR. k RHR and k OHR describe the reductive and oxidative half reactions, respectively,a nd Si sa no xidative substrate such as ketoisophorone. Note that PENTR ox can also be reduced with sodium dithionite) and kinetic data are shown in Figure 3.
As reduced PETNR will oxidizeu nder ambient aerobic conditions the following experiments were performed under anaerobic conditions (N 2 atmosphere)a tr oom temperature.W e found that PETNR is still active when bound to the UCPs and both PETNR and UCP PETNR show typical "Michaelis-Menten" behaviorw ith the oxidizing substrate ketoisophorone (KI) when NADPH consumption is measured. These data were fitted to the Michaelis-Menten equation [Eq. (3)]: giving K m = 18.1 AE 2.4 mm and k cat = 3.68 AE 0.15 s À1 for PETNRi n solution and K m = 10.9 AE 1.2 mm and V max = 0.042 AE 0.001 mm s À1 per mg mL À1 UCP PETNR (Figure 3). Determination of the rate of turnover, k cat ,r equires knowledge of the exact enzymec oncentration( E 0 ), which is difficult to determine for UCP PETNR .H owever,t he similar K m values (Michaelis constant) for PETNR and UCP PETNR suggest that conjugation of the enzyme to the UCP has not had am ajor effect on the enzyme activity.I fo ne assumest hat k cat is unaffected by UCP conjugation the active enzymec oncentration in the UCP PETNR samples can be estimated to be % 0.37 mga ctive PETNR per mg UCP (see the Supporting Information). Assuming detection is limited to the K d for KI,  Chem. Eur.J.2020, 26,14817-14822 www.chemeurj.org 2020 The Authors. Published by Wiley-VCH GmbH the LOD for this system would be on the order of 10 mm; [8,9] future work is focused on reducing the K d to optimize detection limits. As stated above, the spectral overlap betweent he 475 nm UC emission from PTIR-475 and the absorption of the oxidized flavin PETNR means the emissiono fU CP PETNR is sensitivet ot he oxidation state of the enzyme. [8,9] Reduction of UCP PETNR with NADPH or sodium dithionite leads to as ignificant increase in 475 UC emission and reoxidation with KI or molecular oxygen causes this UC emission to revert to the originalv alue (Figures 4a nd S10). The sample is stable and can be cycled multiple times between oxidized and reduced forms,d emonstrating the potentialo fU CP-enzyme systemsf or ratiometric detection of substrates, coenzymes and/or molecular oxygen. [21] Incorporation of other (flavo)enzymeo xidoreductases would allow detection of awide range of substrates and inhibitors by employing ac ompetition assay approach.
It should be noted that, whilet he scale of the UCPs largely precludes distances involved in classical energy transfer processes, it haspreviously been reported that the uniquei nternal particlee nvironmentm ay enhancet hese distances and luminescent resonance energy transfer (LRET)m ay work much more efficiently than traditional FRET. [22] We also suggest that the majority of emitter ions are excited closer to the surfaceo f the particle, with the largests urface area and lowest penetration depth,a nd the potentialf or energy migration through the lattice from the core to the surface. In this case, energy transfer from the surface would therefore be expected to show significantly more efficiency than calculatedf or the bulk particle.
As the PETNR K m for KI appears to be largely unaffected by conjugation to UCPs, (at least with these UCPs)i ts eems likely that sensors based on UCP-enzymes will benefit from the inherents electivity of native enzymes for their substrates. Sensing of redox state, oxygen levels or of specific molecules may be possible within ac ellular environment using suitable enzymes functionalised to smaller,c ell-permeable UCPs and this af uture directiveo fo ur research. Future incorporation of Nd III into the UCPs may also allow improved sensing through accesst othe more biologically transparent 808 nm excitation. [1f] In summary,w eh ave covalently coupled GFP and PETNR to UCPs,w ith AET from the UCP to protein cofactor observed in both cases.E fficient AET requires spectral overlap of UCP emission and protein cofactor absorption, so ratiometric monitoring is possible by using aU CP emission band with no overlap with protein cofactor (e.g.,t he 800 nm band). PETNR remains catalyticallya ctive when coupled to the UCP and the presenceo f reductant or reducing substrate can be determinedr atiometrically from UCP emission. This approacho ffers ad rop-in alternative to the use of thiol-reactive organic fluorescent probes for use as, for example, "molecularp robes", [17] while benefiting from the inherent advantages of UCP-based detection. [4,5]