The setup for a fluorescence optical sensor for saxitoxin is analogous to commonly used ion sensors. A fluorophore is linked to a recognition unit (receptor/ionophore) resulting in a fluoroionophore . Typically, the receptor unit bears a tertiary amine group which is responsible for the emission enhancement in the presence of ions due to the reduced PET effect.
To date, receptors for saxitoxin detection were based on aliphatic aza-crown ethers. Those receptors are highly pH sensitive at physiological conditions because the amine can be easily protonated, which would result in a fluorescence enhancement similar to analyte binding. Moreover, most of the fluorophores which were used for the optical detection of saxitoxin were excitable in the UV region (330–390 nm), which can cause fluorescence background from biological samples (e.g. shellfish extract). Additionally, for measurements in the required low concentration ranges, the complex stability of crown ethers with analytes may be too weak in aqueous solutions. Complex stabilities in organic solvents are typically better and may be sufficient; however, usually aqueous conditions are required for the measurement of environmental samples.
To improve the commonly used setup, we introduced a lariat ether at the ortho position with respect to the nitrogen atom of the crown. This increases the binding efficiency of the analyte, since the two additional oxygen atoms also participate in the complexation . We also decided to use an aromatic crown ether (substituted aniline), which is not sensitive to pH in the relevant range, since the pKA value of the tertiary amine is ~ 5.5. As indicator, we used a commonly known BODIPY fluorophore which is excitable at > 400 nm, possesses a high photostability and molar absorption coefficient, and shows a high quantum yield.
Using this new indicator, the response to saxitoxin in solution was tested under similar conditions as in previously published work (H2O/EtOH/THF mixture, phosphate buffer at pH 7.2). A high fluorescence enhancement is obtained upon protonation of the amine group of the aza-crown ether indicating that the PET effect is suppressed. However, treating with saxitoxin did not show any fluorescence enhancement (Fig. S1, ESI). This negative result raises two fundamental questions: (1) is the complex stability (KD) of the complexation of saxitoxin in the crown ether sufficient to detect saxitoxin in the micromolar range? (2) If saxitoxin is complexed, does it suppress the PET effect or have any other influence on the photophysical properties?
Since the concentration of saxitoxin is limited by the certified reference material (6.63 × 10−5 M stock solution) and it is not possible to obtain saxitoxin in higher concentrations, we investigated if a fluorescence enhancement can be obtained using structurally similar compounds at higher concentration (200× higher). Figure 1b summarizes the surrogate compounds 2–4 used to simulate saxitoxin as they are all subunits of saxitoxin itself and cover the whole molecule. Above all, guanidinium (3) is known to have a high binding affinity to 27-crown-9 and was proposed in previous work to be the structural compound of saxitoxin to inhibit PET [11, 17]. Additionally, we evaluated K+ and NH4+, and ethylenediamine because they are known for their binding affinity to the 18-aza-6-crown.
A high fluorescence increase can be observed in the presence of 10 mM K+ or NH4+, whereas a less pronounced response is caused by ethylenediamine (Fig. 1a, Fig. S2 ESI). However, we did not obtain any significant increase in fluorescence upon adding the surrogates. The same experiment was conducted in a DMSO/H2O (4 + 1) mixture as it is known, that the PET effect is more pronounced in more polar solvents. Again no significant increase of fluorescence using these surrogates was observable (Fig. S3, ESI).
The complexation behaviour is highly depending on two factors—the solvent and the size of the cavity and the analyte. Generally, crown ethers show the highest binding constants in methanol and the lowest in aqueous solution as a higher ratio of organic solvents are beneficial for the complexation . The low complexation in water is due to a too-strong tendency to undergo hydration of the ion instead of getting complexed as the hydration shell around the ion needs to be stripped off . Methanol or other organic solvents are much weaker solvating mediums and therefore hydration competes less with complexation yielding stability constants around three to four decades more than in water.
Another important parameter besides the solvent is the size of the crown cavity and the guest ion. As size of the 18-crown-6 is between 2.6 and 3.2 Å, it shows optimal interaction with K+ ion (2.66 Å) and NH4+ (2.86 Å) . The corresponding stability constants of these in H2O are lgK = 2.05 for K+ and lgK = 1.44 for NH4+ [20, 22]. As the ammonium ion is substituted higher, the stability constants lower since the ion gets sterically hindered to fit into the crown ether . This trend is observable in our data for K+, NH4+, ethylenediamine, and the surrogates. The amine group of the latter is highly substituted which consequently prevents the complexation.
However, reported saxitoxin-sensitive fluoroionophores which were used in aqueous solution show a binding constant 1000 × higher than for K+ [13, 14, 16]. Additionally, it was reported that complex stabilities of saxitoxin are higher in pure H2O than in an EtOH/H2O mixture, which is in contrast to the trend of measured binding constants of all crown–ion interactions in different solvents .
The published utilized fluoroionophore for saxitoxin measurements in water is based on a methoxycoumaryl-aza-crown dye, with which it was possible to measure the concentrations of saxitoxin in the micromolar range with 137 mM NaCl and 2.7 mM KCl as background . Na+ and K+ did not “turn on” the sensor even though an aza-18-crown-6 was used as the recognition unit. In this work, saxitoxin binds to the receptor and inhibits the PET in a K+ background that is 27 times higher, whereas K+ does not turn on the sensor.
We synthesized this saxitoxin-sensitive coumarin indicator dye as described in literature, and response to saxitoxin was tested under conditions similar to those reported (Fig. 2a) . However, we could not observe any immediate increase in fluorescence with saxitoxin, but observed an increase in fluorescence intensity and a slight blue shift over the course of 20 min, similar to the published work. However, as a blank sample without any indicator was measured, we detected that saxitoxin itself starts to fluorescence upon illumination at 330 nm. This emission at 390 nm is shifted compared to the coumarin emission at 401 nm and superimposition with the coumarin fluorescence could explain the blue shift of the emission of the probe which is untypical for PET-indicators (Fig. 2a). Excitation spectra and emission spectra of both the coumarin dye and the saxitoxin illumination product are very similar and overlap over a broad range (Fig. S4, ESI). The saxitoxin decomposition product shows excitation and emission peaks of 334 and 390 nm, respectively. This corresponds to the fluorescent decomposition product that is usually obtained during the pre- or post-column oxidation of the HPLC–fluorescence detection method, where saxitoxin is chemically oxidized to the fluorescent purine derivate 5 (Fig. 2b) [6, 7, 23]. From this experiment, we concluded that the increase of fluorescence is not caused by inhibition of the PET effect by saxitoxin. Instead, we were able to determine that this increase in fluorescence can be attributed to a photooxidation product of saxitoxin itself.
To investigate the formation of this fluorescent saxitoxin product, we recorded the emission spectra of a buffered solution (pH 7.2) of saxitoxin alone (Fig. S5a, ESI). When saxitoxin is stored without exposing to light, no increase in fluorescence is observed. In contrast, strong fluorescence is detectable after illumination with UV light. The fluorescence intensity of the oxidation product after the illumination of saxitoxin with different intensities in the fluorimeter clearly shows that the formation of this fluorescent saxitoxin product is highly dependent on the intensity of the applied UV light and that saxitoxin does not form this product by simple exposure to ambient air (Fig. S5b, ESI).
It should be stated that Gawley et al. also detected this background fluorescence of saxitoxin, but interpreted it as a trace impurity of the saxitoxin solution and did not observe its increase during their measurements . In our opinion, attributing the fluorescence increase of saxitoxin using the coumarin indicator to PET inhibition is a misinterpretation of data. The saxitoxin product shows very similar excitation and emission spectra to the used coumarin indicator and an addition of both fluorescence spectra explains the observed fluorescence enhancement by saxitoxin. Moreover, fluorescence enhancement due to saxitoxin oxidation can explain the observed blue shift of the emission in our measurement and in literature which is untypical for PET-indicators [14, 15]. With this in mind and the comparison of published stability constants of saxitoxin–crown interaction with well-known ion–crown complexations, we believe that the measurement of saxitoxin using this method in water is not achievable in the environmental necessary concentration range.
However, crown ether sensors for saxitoxin based on other fluorophores have been developed and work at different excitation/emission wavelengths. For these probes, the fluorescence increase is not influenced by this background fluorescence . It is also important to note, that Gawley et al. in their earlier contributions were using non-aqueous solutions or a very high percentage of organic solvents which would be beneficial for the complexation of saxitoxin and much higher saxitoxin concentrations were used for the measurements .