A caged peptide is only as good as the photochemistry that releases it, and the two chromophore families in routine use — o-nitrobenzyl and coumarin-4-ylmethyl — fail in different ways. The choice sets the usable wavelength, how much of the absorbed light does useful work, what reactive species you generate alongside the free peptide, and whether release is fast enough that your kinetics report on the peptide rather than on the cage. None of it is decided by the extinction coefficient on the reagent sheet.
Two chromophores, two mechanisms
The o-nitrobenzyl (oNB) cage does not simply photodissociate. Excitation drives intramolecular hydrogen abstraction by the nitro group to an aci-nitro tautomer, and flash photolysis of 2-nitrobenzyl methyl ether, 1-(2-nitrophenyl)ethyl methyl ether and caged ATP with UV–vis and IR detection identified two further intermediates downstream of it: cyclic 1,3-dihydrobenz[c]isoxazol-1-ol derivatives formed by decay of the aci-tautomers, and 2-nitrosobenzyl hemiacetals [1]. Which step limits release depends on the leaving group and the pH. For the two methyl ethers the hemiacetal is rate-limiting — below pH 8 for the 2-nitrobenzyl compound and below pH 10 for the 1-(2-nitrophenyl)ethyl compound [1]. That study presents itself as a substantial revision of previously proposed release mechanisms, and its hemiacetal result is stated in explicit contrast to an earlier claim for related 2-nitrobenzyl methyl ethers [1], so this is a mechanism that has been argued over rather than a fixed textbook scheme. With a phosphate leaving group the picture differs: time-resolved single-frequency infrared on caged ATP showed release in a single exponential process synchronous with decay of the aci-nitro anion intermediate [2]. A release rate measured for one linkage is therefore not transferable to another, and the linkage you build onto a peptide — carbamate, ether, ester, thioether — is part of the kinetics, not a detail of the synthesis.
Coumarin-4-ylmethyl cages break heterolytically, through a contact ion pair. What the ion pair does next is the efficiency-determining step: across a range of coumarin photocleavable protecting groups, the contact-ion-pair escape mechanism was shown to determine the payload- and solvent-dependent photolysis quantum yield, with an intramolecular type of escape superior to diffusion-dependent escape [3]. The practical reading is unwelcome. A quantum yield measured for a coumarin cage carrying one payload in one solvent is not a property of the chromophore you can carry across to a different payload in a different medium [3].
Quantum yield, and why the published number may not be yours
For a series of oNB derivatives built for red-shifted absorption, uncaging quantum yields after one-photon absorption fell in the 0.1–1% range, and dropped as the chromophore's absorption maximum was pushed to longer wavelength [4]. That inverse trade — redder absorption bought with lower uncaging yield — is the central design problem of the family, and it is why "absorbs further into the visible" is not on its own an improvement.
The 0.1–1% band is not a universal oNB figure. The photolysis quantum yield of 1-(2-nitrophenyl)ethyl phosphate (caged Pi) measured in the near-UV, with a 342 nm peak and 60 nm half-bandwidth, is 0.53 [5]. Two orders of magnitude separate that from the red-shifted ether and ester series [4]; the difference lives in the ring substitution, the benzylic substitution and the leaving group, not in the label "nitrobenzyl".
Peptides make it worse in a way worth knowing before you budget photon dose. When 4,5-dimethoxy-2-nitrobenzyl was attached through histidine and aspartate, the quantum yield of photoconversion was 0.03 for the fluorenylmethoxycarbonyl amino acid derivatives, but approximately 0.01 and 0.005 for the corresponding peptides [6]. Same chromophore, same residues — a three- to six-fold penalty for being in a peptide rather than in a monomeric derivative. Note also what was measured there: photoconversion into the nitroso form, which is a different event from liberation of free peptide.
The efficient end of the nitroaromatic family is nitrodibenzofuran (NDBF). For the NDBF derivative of EGTA, the quantum yield of photolysis is 0.7 and the extinction coefficient 18,400 M−1 cm−1, making its photolysis roughly 16–160 times more efficient than the most widely used caged compounds [7]. Those figures belong to NDBF-EGTA, a caged calcium chelator, and are not a measurement on an NDBF-caged peptide.
On the coumarin side, brominated 7-hydroxycoumarin-4-ylmethyl (Bhc) esters and carbamates release carboxylates and amines with one- and two-photon cross sections up to one or two orders of magnitude better than what was previously available, and conventional UV photolysis of Bhc-glu required less than one-fifth the intensity needed by γ-(α-carboxy-2-nitrobenzyl)-L-glutamate [8]. That comparison is between two caged glutamates under UV, and it is the honest way to state a chromophore advantage: same payload, same excitation, measured side by side.
The nitroso by-product, and what a thiol actually does to it
Every oNB uncaging event produces a nitroso aromatic in stoichiometric amount, and it is not a spectator. Photolysis of 1-(2-nitrophenyl)ethyl phosphates in the presence of dithiothreitol gave 3-methylanthranil as the major final by-product, formed from 2-nitrosoacetophenone, and the by-product species likely to be present on the timescale of most biological experiments using caged compounds is 2-hydroxylaminoacetophenone as a mixture of ring–chain tautomers [2]. That is the correct way to think about scavenging: adding a thiol does not delete the nitroso species, it redirects its fate, and the reduced product then becomes what your assay is exposed to.
The by-product also carries its own spectroscopy into your readout. Photolysis of 4,5-dimethoxy-2-nitrobenzyl esters of acetic, caprylic and arachidonic acids produced large growth of fluorescence emission at around 430 nm, and further irradiation generated a second product fluorescing at longer wavelength (470 nm excitation, roughly 500–600 nm emission), which the authors flag as liable to interfere with widespread dyes such as fluorescein [9]. That study used caged carboxylic acids rather than peptides, but the chromophore chemistry is identical, and it is a direct argument against reading uncaging progress off a fluorescence channel without a photoproduct blank.
Coumarins have a by-product problem too; theirs is a rearrangement of the cage rather than a reactive fragment. Bhc-protected thiols can undergo isomerisation upon photolysis to a dead-end product, a 4-methylcoumarin-3-yl thioether [10], and that undesired photoisomerisation limits the group's uncaging efficiency [11]. Substituting 6-bromo-7-hydroxy-3-methylcoumarin-4-ylmethyl (mBhc) suppressed it: a peptide containing an mBhc-caged thiol underwent clean and efficient photocleavage on irradiation without detectable photoisomer production [11]. Both results come from the same laboratory and the same peptide system, so treat them as a well-characterised comparison within one series rather than a settled ranking of coumarins in general.
Two-photon uncaging is governed by the action cross-section
One-photon extinction predicts nothing useful about two-photon performance, because the quantity that matters is the two-photon uncaging action cross-section — absorption and uncaging efficiency multiplied together. The gap is stark for oNB. Using a fluorescence correlation spectroscopy method after two-photon excitation, a series of oNB-caged fluorescent coumarins showed action uncaging cross-sections in the 0.1–0.01 Goeppert-Mayer (GM) range, in line with the low uncaging quantum yields combined with two-photon absorption cross-sections of 1–50 GM [4]. The authors' own conclusion is the one to carry into a design decision: one- and two-photon absorption cross-sections of oNB groups can readily be improved, but enlarging the corresponding action uncaging cross-sections is difficult given the trend in uncaging quantum yield [4]. Buying absorption at the cost of yield leaves the product where it started.
Numbers that clear the bar come from chromophores with a high uncaging quantum yield to multiply into. Bhc set that relative benchmark [8]; NDBF-EGTA has a two-photon cross-section of approximately 0.6 GM [7]; and a methoxy-substituted NDBF group, used to prepare a protected cysteine suitable for SPPS and then to assemble two peptides, has a two-photon action cross-section of 0.71–1.4 GM [12]. NDBF-caged cysteine peptides deprotect efficiently under either 365 nm one-photon or 800 nm two-photon excitation [10].
Caging the backbone or caging a side chain
Side-chain caging is the default because it is the easiest to build. N-Fmoc-O-(2-nitrobenzyl)-tyrosine can be incorporated at any position by Fmoc solid-phase peptide synthesis, and applied to neuropeptide Y it gave a caged peptide whose binding affinity for the Y1 receptor was one or two orders of magnitude lower than intact NPY, rising back to the intact value on UV irradiation; the photolysis intermediate had a half-life of approximately 7 μs [13]. One to two orders of magnitude of suppression is a real result, and it is also modest — a side-chain cage removes one recognition element and leaves the rest of the epitope intact.
Backbone-amide caging attacks the main chain instead. A backbone-caged speract carrying a 2-nitrobenzyl group on a backbone amide had an IC50 of 950 nM against its receptor; UV photocleavage of the group (τ1/2 = 26 μs) restored an IC50 of 0.67 nM [14]. In that study backbone caging of the biological activity was more efficient than side-chain caging, and the authors present the approach as a general procedure for peptides that have no side chain suitable for introducing a caging group [14]. That is a single head-to-head comparison in one laboratory on one peptide, not a general law — but the structural logic behind it, that a main-chain modification is not contingent on the sequence carrying a derivatisable side chain, is what makes it worth the harder synthesis.
A third option is to cage the post-translational mark rather than the residue. [7-(diethylamino)coumarin-4-yl]methyl (DEACM)-caged phosphoserine, phosphothreonine and phosphotyrosine building blocks were made for Fmoc-based SPPS to allow convenient incorporation into peptides and proteins, so what is masked is the phosphorylation mark itself rather than the residue's other contacts [15].
Getting the cage through the synthesis
Nearly all of this is installed as an Nα-Fmoc building block and carried through standard Fmoc SPPS: N-Fmoc-O-(2-nitrobenzyl)-tyrosine at any sequence position [13]; Fmoc-Cys(NDBF)-OH incorporated by standard solid-phase procedures [10]; Fmoc-Cys(mBhc)-OH for the coumarin thiol cage [11]; DEACM-caged phosphoamino acid building blocks for convenient incorporation into peptides and proteins [15]; and a methoxy-NDBF-protected cysteine used to assemble two peptides [12].
The constraint is that assembly is a long alternation of base and acid. Repeated Fmoc removal exposes the cage to piperidine at every cycle, and final resin cleavage exposes it to strong acid. The coumarins show where this bites: peptides bearing a Bhc-protected cysteine were prepared using Fmoc-Cys(Bhc-MOM)-OH, with the coumarin phenol carried as its methoxymethyl ether through assembly [10]. Anything on the chromophore that is base-labile or acid-labile has to be masked or engineered out before the building block is usable, which is a real part of the cost of a new cage.
The compensating advantage is worth stating plainly, because it is the reason to accept that cost: photoremovable groups are deprotected by illumination without adding acids or bases that can perturb biological activity [12]. And because these chromophores are built to absorb in the near-UV [4], ambient light is itself a deprotection pathway — caged building blocks and caged peptides are light-sensitive stock, not ordinary stock.
Measuring uncaging efficiency
Quantum yield is the number most often quoted and most often incomparable. The 0.53 value for caged Pi rests on results reported in 1978, and the 2016 analysis was written specifically to amplify the methodology in view of different recent estimates of the same quantity [5]. Report the excitation your measurement actually used — in that case a 342 nm peak with 60 nm half-bandwidth [5] — because a figure from a broadband near-UV source and a figure from a laser line are not the same measurement even when both are labelled by a nominal wavelength. For two-photon work the analogous quantity is the action cross-section, and the method has to be stated too; the 0.1–0.01 GM values for the oNB series were obtained by fluorescence correlation spectroscopy after two-photon excitation [4].
Two traps sit between a spectroscopic signal and a released-peptide claim. First, the transient is not the product. Because a hemiacetal can be the rate-limiting intermediate for some leaving groups and pH ranges [1], decay of the aci-nitro absorption bounds an early step, not the appearance of free peptide; only where release was shown to be synchronous with that decay, as for caged ATP [2], does the transient stand in for the payload. This is the same reason the peptide quantum yields of 0.01 and 0.005 in the dimethoxynitrobenzyl series are yields of photoconversion into the nitroso form [6] rather than yields of free peptide. Second, the photoproducts absorb and fluoresce inside the readout band [9], so absorbance and fluorescence changes conflate cage consumption with by-product accumulation. Chromatographic separation of intact caged peptide, free peptide and by-product, run as a function of irradiation dose, is what supports a released-fraction number.
Controls that separate photochemistry from photodamage
The failure mode that destroys an experiment is light doing something other than uncaging. A cellular study of coumarin-caged lipids reported a specific photouncaging artifact triggered by the cage structure coumarin under UV illumination: 340 nm light initiated a membrane conductance specifically in the presence of coumarin at concentrations as low as 30 μmol L−1, and that conductance masked the TRPC3 conductance evoked by photouncaging, while TRPC-mediated cellular Ca2+ responses were largely preserved [16]. The analyte there was a caged lipid and an ion channel, not a peptide — but the effect was attributed to the released caging moiety, so it is a hazard of the chromophore, and it also shows that the artifact can be selective, hitting one readout while sparing another in the same cells [16].
A control set that earns the conclusion has five members. A dark control — caged compound, no illumination — bounds residual activity and gives you dynamic range; the 950 nM IC50 reported for caged speract is precisely this measurement, and it is what makes the 0.67 nM post-photolysis value interpretable [14]. A light-only control applies the identical dose with no caged compound present. A photoproduct control applies the identical dose to the free chromophore or its photolysis products alongside uncaged peptide, which is the control that would have caught the coumarin membrane effect [16]. A pre-photolysed control irradiates first and adds afterwards, separating "the released peptide acts" from "light acts". And if you add a scavenger, treat it as a variable: dithiothreitol changes which by-product your system sees, since photolysis in its presence gives 3-methylanthranil, formed from 2-nitrosoacetophenone, as the major final by-product, while the species likely to be present on the timescale of most biological experiments is 2-hydroxylaminoacetophenone as ring–chain tautomers [2].
What all this licenses is narrow and worth stating. A released-fraction figure from chromatography at a stated dose, plus a dark control, gives you dynamic range for that construct under that lamp. It does not transfer. The same chromophore lost three- to six-fold in quantum yield moving from an amino acid derivative into a peptide [6], and coumarin photolysis quantum yields are payload- and solvent-dependent by mechanism [3]. Measure the cage on the peptide you intend to use, at the wavelength you intend to use, and report both.