A fast and specific fluorescent probe for thioredoxin reductase that works via disulphide bond cleavage

Small molecule probes are indispensable tools to explore diverse cellular events. However, finding a specific probe of a target remains a high challenge. Here we report the discovery of Fast-TRFS, a specific and superfast fluorogenic probe of mammalian thioredoxin reductase, a ubiquitous enzyme involved in regulation of diverse cellular redox signaling pathways. By systematically examining the processes of fluorophore release and reduction of cyclic disulfides/diselenides by the enzyme, structural factors that determine the response rate and specificity of the probe are disclosed. Mechanistic studies reveal that the fluorescence signal is switched on by a simple reduction of the disulfide bond within the probe, which is in stark contrast to the sensing mechanism of published probes. The favorable properties of Fast-TRFS enable development of a high-throughput screening assay to discover inhibitors of thioredoxin reductase by using crude tissue extracts as a source of the enzyme.


Contents
The solution was then acidified to pH = 1 using 1 M HCl and concentrated under reduced pressure.
The residue was dissolved in 100 mL of ethyl acetate and washed with brine (2×50 mL), dried over

Supplementary Notes
Detailed Mechanism of TRFS-green Activation. One major unfavorable property of TRFS-green is its slow response to TrxR 4 . As shown in Figure 1A, Taken together, these results strongly support that the cyclization of the intermediate is a rate-limiting step for the activation of TRFS-green.
Screening of TRFS probes. TCEP is a strong reducing agent, and may readily reduce disulfide and diselenide bonds. To simplify the experimental procedure, we employed TCEP in place of TrxR/NADPH for the initial screening of the probes (TRFS3-8). Incubation of TRFS3 with TCEP for 3 h gave no apparent ANA signal when the mixture excited at 438 nm (Supplementary Figure 8A).
However, when the mixture was excited at 363 nm (the maximal absorbance of TRFS3), a fast increase of the emission signal centering at ~495 nm was observed, and the fluorescence intensity reaches a plateau (~10-fold increase) within 2 min (Supplementary Figure 1A). When TRFS4 was incubated with TCEP, a weak but repeatable ANA fluorescence signal appeared, and the fluorescence intensity centering at 538 nm increased constantly within 4 h giving a ~4-fold elevation of the emission signal (Supplementary Figure 1A). Compared to the TRFS-green that shows nearly an 18-fold increase of fluorescence intensity within 3 h ( Figure 2C), this observation led us to conclude that the release of the fluorophore is unfavorable when a six-membered cyclization product (referring to the second step of TRFS-green in Figure 1A) was formed. TRFS5, a selenium analogue of TRFS-green, responded to TCEP to release the ANA fluorophore (Supplementary Figure 1C). It gave a >80-fold increase of the ANA emission signal within 10 min, which is much faster than the probe TRFS-green ( Figure 2C). This observation indicated that replacement of the five-membered cyclic disulfide moiety with the corresponding diselenide could improve the response rate drastically. It is known that selenolates have stronger nucleophilicity than do thiolates 17, 18 , which may account for the observed fast response of TRFS5. Similar to its sulfur analogue TRFS3, TRFS6 displayed a negligible ANA fluorescence signal when it was incubated with TCEP for 3 h (Supplementary Figure   8B). Also, a great increase (>100-fold) of the emission signal centering at ~495 nm was observed, and the fluorescence intensity reached a plateau within 5 min (Supplementary Figure 1D).  Figure 1F). For clarity, the response of the probes TRFS3-8 was summarized in Table 2 (the second and third columns).

Replacement of the recognition part in TRFS
Based on these results, a preliminary structure-activity relationship (SAR) of these probes ( Figure 1B & Figure 3) in responding to TCEP could be drawn. First, the linker atom directly connecting to the fluorophore (X in Figure 1B) determines the stability of the probes, and the nitrogen atom is expected to improve the stability while probes contain the oxygen atom are not stable (TRFS1 and TRFS2). Second, both the Y atom and the number of carbons between the recognition part and the Y atom (Y and n in Figure 1B) Figure 2A), suggesting that the 5-membered cyclic diselenide could be reduced by GSH. Other TRFS probes showed negative response to GSH either excited at 438 nm or 363 nm (Supplementary Figure 9). We next determined whether TrxR could turn on the fluorescence of these probes. Incubation of TRFS3 with TrxR and NADPH, no significant ANA fluorescence signal appeared when excited at 438 nm, which is similar to its reduction by TCEP (Supplementary Figure   8A). However, when the reaction mixture was excited at 363 nm, a fast increase (~6-fold) of the fluorescence signal centering at ~495 nm was observed (Supplementary Figure 2B). TRFS4 gave an emission signal centering at ~475 nm with the excitation at 382 nm (Supplementary Figure 2C), but there was no apparent signal of the ANA fluorophore. This observation indicated that TrxR could reduce the disulfide bond but the following CDR process did not happen due to the unfavorable formation of a six-membered cyclization product. The fluorescence signal of TRFS5 was also switched on by TrxR, and a rough 6-fold increase of the ANA signal (λ ex =438 nm) was observed within 3 h (Supplementary Figure 2D). There was no significant fluorescence increment when incubation of TRFS6, TRFS7 or TRFS8 with TrxR and NADPH, either excited at 363 nm or 438 nm (Supplementary Figure 10), indicating that the six-membered cyclic disulfides and five-membered cyclic diselenides are likely not reduced by TrxR. Taken together, after we examined the response of the probes with GSH and TrxR, we could conclude that the recognition part containing five-membered cyclic disulfides showed selectivity for TrxR over GSH. For clarity, the response of the probes to GSH and TrxR was summarized in Table 2 (the last two columns).
Sensing Mechanism of TRFS3. An interesting observation from the reduction of TRFS3-8 by TCEP is that there are two different types of fluorescence spectra. One is the blue emission centering at  Figure 1A) also supported a direct reduction of TRFS3 by TCEP. It was reported that carbonylation of the amino group of ANA causes a blue shift of its absorbance from ~440 nm to ~370 nm. In addition, this modification also changes the green emission of the ANA (~540 nm) to a blue emission of the amide form of ANA (~470 nm) 32,33,34 .
Taken together, these results demonstrated that a direct reduction of TRFS3 without the following cyclization ( Figure 4B) occurred in the response of the probe to TCEP. Furthermore, the off-on fluorescence signal of TRFS3 (and other probes, such as TRFS6 and TRFS8) in response to TCEP also suggested that the disulfide/diselenide bond could quench the emission of certain fluorophores, and may serve as a trigger in designing fluorescent probes.  Table 3), and studied their interactions with TrxR and GSH. The detailed synthetic procedures and characterization of these molecules were described in the Supplementary Information. It is not convenient to determine the direct reduction of the molecules by GSH. Thus, we adopted a coupled enzymatic assay with GR and NADPH 44 . The decay of absorbance at 340 nm (A 340 ), due to the oxidation of NADPH, was monitored for 10 min, and the rates of A 340 decay within the initial 3 min were calculated and summarized in Table 3. Without the test compounds, the background decay of A 340 in the TrxR assay is 0.54x10 -4 s -1 . In the presence of compounds 1, 2, 3, 5, and 9, the decay of A 340 was significantly higher than the background, indicating these compounds are substrates of TrxR. Without the test compounds, the background decay of A 340 in the GR/GSH assay is 0.70x10 -4 s -1 . In the presence of compounds 4, 5, 8 and 9, the oxidation of NADPH was faster than the background, indicating that GSH could reduce these compounds. Based on these data, the SAR could be drawn: 1) The 1, 2-dithianes (6-membered cyclic disulfides, compounds 6 & 7)

Reduction of Cyclic
cannot be reduced by either TrxR or GSH; 2) The 1, 2-dithiolanes (5-membered cyclic disulfides, compounds 1, 2 & 3 are substrates of TrxR but cannot be reduced by GSH. The reduction of the cyclic diselenides is a little bit complicated. Compounds 5 and 9 are substrates of both TrxR and GSH, while compound 8 is resistant to TrxR but appears a weak substrate of GSH. Interestingly, compound 4 seems to be selectively reduced by GSH but not by TrxR. In the comparison of 1, 2-dithiolanes reduction, the reduction of aminodithiolane (3) is much faster than the hydroxydithiolane (2) and carboxylicdithiolane (1). Under our assay conditions, the amino group in 3 and the carboxylic group in 1 are positively and negatively charged, respectively. The positively charged amino group would have a favorable electrostatic interaction with the negatively charged C-terminal active site of TrxR, while the interaction between the negatively charged carboxylic group and the latter is unfavorable. This might account for the observed different reduction rates of the three dithiolanes by TrxR. Taken together, although more data are needed to obtain a clear picture of reduction of cyclic diselenides, it is evident that 1, 2-dithiolanes display promising selectivity to TrxR over GSH, which strongly supports the selective activation of TRFS3 and TRFS4 by TrxR.
This discovery demonstrated that the 1, 2-dithiolane moiety may serve a general ligand in designing chemical tools to target TrxR selectively.  Table 2, for molecule R-Fast-TRFS, the absorbance of S 1 excited state makes major contribution to the absorption peak near 300 nm (corresponding to the absorption peak at 346 nm in experiment).
Therefore, we consider that Fast-TRFS will be excited from ground state to S 2 excited state, while R-Fast-TRFS will be excited from ground state to S 1 excited state.
Then we optimized the S 2 excited state structure of Fast-TRFS, and the S 1 excited state structure of R-Fast-TRFS, their fluorescence emission spectra were calculated based on the optimized excited state structures.
From Supplementary Figure 6 and Supplementary Table 3 it can be seen that: when the molecule Fast-TRFS is in S 2 excited state structure, its S 1 emission peak has low oscillator strength (0.0011), while S 2 emission peak has high oscillator strength (0.6753). However the strong S 2 emission peak is not observed in experiment, we thought that internal conversion might occur from S 2 to S 1 excited state. When the molecule Fast-TRFS is in S 2 excited state structure, the energy difference between S 2 and S 1 excited states are small (~0.22 eV, Supplementary Table 3), making the internal conversion possible. After internal conversion of Fast-TRFS, S 1 emission took place, leading to low intensity of fluorescence.
From Supplementary Figures 5 and 7, Supplementary Tables 2 and 4 we can draw conclusion that: for molecule R-Fast-TRFS, the UV-Vis absorption and fluorescence emission can be mainly attributed to HOMO-LUMO transition type, therefore its absorption peak and emission peak have similar oscillator strengths.