Simple and Rapid Endotoxin Recognition Using a Dipicolylamine-Modified Fluorescent Probe with Picomolar-Order Sensitivity.

Endotoxin is a lipopolysaccharide (LPS) that is found in the outer membrane of the cell wall of Gram-negative bacteria. Due to its high toxicity, the allowable endotoxin limit for water for injection is set at a very low value. Conventional methods for endotoxin detection are time-consuming and expensive and have low reproducibility. A previous study has shown that dipicolylamine (dpa)-modified pyrene-based probes exhibit fluorescence enhancement in response to LPS; however, the application of such probes to the sensing of LPS is not discussed. Against this backdrop, we have developed a simple and rapid endotoxin detection method using a dpa-modified pyrenyl probe having a zinc(II) center (Zn-dpa-C4Py). When LPS was added into Zn-dpa-C4Py solution, excimer emission of the pyrene moiety emerged at 470 nm. This probe can detect picomolar concentrations of LPS (limit of detection = 41 pM). The high sensitivity of the probe is ascribed to the electrostatic and hydrophobic interactions between the probe and LPS, which result in the dimer formation of the pyrene moieties. We also found that Zn-dpa-C4Py has the highest selectivity for LPS compared with other phosphate derivatives, which is probably caused by the co-aggregation of the probe with LPS. We propose that Zn-dpa-C4Py is a promising chemical sensor for the detection of endotoxin in medical and pharmaceutical applications.

Introduction

Endotoxin is a lipopolysaccharide (LPS) and the main component of the outer membrane of the cell wall of Gram-negative bacteria. LPS has an amphiphilic structure that is made up of three components: lipid A consisting of hydrophilic phosphorylated glucosamines and hydrophobic fatty acid chains, a core region, and a repeating unit (O-antigen) (Figure ).[1,2] This molecule is highly negatively charged due to the phosphate groups and carboxyl groups.[3] A tiny amount of endotoxin adversely affects the human immune system, causing fever, leukocytosis, tachycardia, and fatal multi-organ failure termed sepsis.[4−6] In 2017, 11.0 million sepsis-related deaths were reported worldwide, which account for 19.7% of all global deaths.[7] The European Pharmacopeia, the United States Pharmacopeia, and the Japanese Pharmacopeia have established a strict endotoxin limit of 0.25 EU/mL (EU = unit of measurement for endotoxin activity) for water for injection (WFI), which is highly purified, in order not to adversely affect a patient’s safety.

Figure 1

Schematic diagram of LPS (left) and structure of lipid A in Escherichia coli (right).

Currently, endotoxin is detected either by the rabbit pyrogen assay, immune assays, and the limulus amebocyte lysate (LAL) assay, which uses an aqueous extract of blood cells from the horseshoe crab, namely, LAL reagent or lysate reagent. The LAL assay is most widely used and has high sensitivity, thanks to a cascade system that amplifies the enzymatic reaction with endotoxin.[8] On the other hand, this method requires a long measurement time,[9] normally more than 60 min, and is expensive.[10] Moreover, there are unfavorable differences in reactivity among LAL reagent lots. Hence, an alternative technique to detect endotoxin is eagerly desired. Many attempts to establish new endotoxin detection methods have been made from various perspectives, e.g., electrochemistry, biochemistry, physical chemistry, and photochemistry, using aptamer, peptide, protein, graphene oxide, gold nanoparticle, quartz crystal unit, and fluorescent chemical sensors.[11−13] In particular, fluorescent chemical sensors are appealing in terms of sensitivity, selectivity, and real-time detection.[11,14] For example, Liu et al. designed and synthesized a series of pyridinium-functionalized dibenzo[a,c] phenazine fluorescent sensors for the selective detection of LPS and clarified the influence of the alkyl chain length in the probe molecule on their optical properties.[3] Lim et al. reported peptide-assembled graphene oxide as a fluorescent Turn-ON chemical sensor for LPS, which has excellent sensitivity and the detection limit of 130 pM, one of the lowest values for a synthetic fluorescent chemical sensor to date.[11,15] However, even this method is inferior to the LAL assay in terms of sensitivity. Moreover, many fluorescent chemical sensors for LPS require either time-consuming sample preparation or expensive materials.

We have reported several dipicolylamine (dpa)-modified fluorescent probes for the recognition of phosphate derivatives.[16−19] The dpa moiety forms a chelate complex with a metal ion that has unoccupied coordination sites to which the oxygen atoms of the phosphate groups coordinate in water.[19] The probes require neither cumbersome sample preparation nor costly substances. Thus, we expect that the dpa-modified probes will be able to detect LPS because LPS possesses several phosphate units in its structure. Meanwhile, Cabral et al. reported that the fluorescence intensity of several dpa-modified pyrene-based probes is enhanced by recognizing broad-spectrum bacteria, and for Gram-negative bacteria, they identified negatively charged membrane components including LPS as binding points between the probes and bacteria.[20] Moreover, they ascribed this fluorescence change to the excimer signal produced by the proximity-based stacking of multiple pyrene units through the chelated Zn2+ by dpa for phosphate binding. Although they were able to develop excellent chemical sensors, detailed experiments characterizing them as a practical LPS sensor were not fully carried out because their detection target was bacteria.

Herein, we report a simple and rapid detection method for LPS using the dpa-modified fluorescent probe reported by Cabral et al. (hereinafter denoted as dpa-C4Py). We synthesized two dpa-modified probes, dpa-C1Py and dpa-C4Py (Chart ), to investigate the effect of the length of the spacer connecting the recognition site (dpa moiety) and the reporter site (pyrene moiety). The optical responses of these probes with various centered metal ions (M-dpa-C, M = Co2+, Ni2+, Cu2+, Zn2+, and Cd2+) toward LPS were investigated by ultraviolet–visible (UV–vis) absorption and fluorescence measurements to determine the optimal sensor. Particle size and zeta potential were also determined by dynamic light scattering (DLS) measurements to examine the morphology of the probe with LPS. Moreover, selectivity and interference assays were carried out using phosphate derivatives.

Chart 1

dpa-C (n = 1, 4)

Experimental Section

Please refer to the Supporting Information for the reagents and the synthesis of dpa-C (Schemes S1 and S2).

Apparatus

1H NMR spectra were measured using an Avance III HD 400 (Bruker Japan K. K., Kanagawa, Japan) at 400 MHz and JNM-ECA500 (JEOL Ltd., Tokyo, Japan) at 500 MHz at 298 K. All pH values were recorded using a Horiba F-52 pH meter (HORIBA, Ltd., Kyoto, Japan). UV–vis absorption spectra were measured using a 10 mm quartz cell and a Jasco V-760 UV–vis spectrophotometer (JASCO Corporation, Tokyo, Japan) equipped with a Peltier thermocontroller. Fluorescence spectra were measured using a 10 mm quartz cell and a HITACHI F-7000 fluorescence spectrophotometer (Hitachi High-Technologies, Co., Tokyo, Japan) equipped with a Peltier thermocontroller. DLS measurements were carried out at 25 °C using a Zetasizer Nano ZS (Malvern Instruments Ltd., Malvern, Worcestershire, UK).

Preparation of LPS stock solution

LPS from Escherichia coli O55: B5 (purified by phenol extraction) was dissolved in 5 mM HEPES buffer (pH 7.4). The LPS stock solution was vortexed for 3 min followed by sonication for 5 min.

Evaluation of the LPS Recognition Function of M-dpa-C

The UV–vis and fluorescence spectra of dpa-C with various metal ions were recorded in the absence and presence of 1.0 μM LPS. For titration tests, LPS was successively added into 10 μM M-dpa-C solution until LPS concentration reached 1.0 μM. Each sample was measured within 1 min after the sample preparation. Fluorescence spectra were obtained at the excitation wavelength of 350 nm at 25 °C. The slit widths were set at 5 nm for both excitation and emission. The scan rate was set at 240 nm/min.

Results and Discussion

LPS Recognition by M-dpa-C

The LPS recognition functions of dpa-C1Py and dpa-C4Py were evaluated in the absence and presence of the following metal ions: Co2+, Ni2+, Cu2+, Zn2+, and Cd2+. In the absence of the metal ions, dpa-C4Py showed strong excimer emission centered at 470 nm (Figure S1). This is probably caused by the lower water solubility of dpa-C4Py, which possesses a longer alkyl chain, because such a result was not observed for dpa-C1Py. The excimer emission of dpa-C4Py was quenched by the addition of the metal ions except for Ni2+ (Figure and Figure S1). The metal ions improved the solubility of the probes by complexation with the dpa moiety, which possibly brought about electrostatic repulsion among the probe molecules, leading to the monomerization of the pyrene dimers. On the other hand, we found that after the addition of 1 μM LPS into M-dpa-C4Py solution, the excimer emission of Zn-dpa-C4Py and Cd-dpa-C4Py was enhanced (Turn-ON), whereas that of the Ni2+ complex was diminished (Turn-OFF) (Figures and 4). LPS seemingly coordinated to the centered metal ions of Zn2+ and Cd2+, resulting in the dimer formation of the pyrene moiety. The existence of intermolecular π–π stacking of two pyrene moieties was also evidenced by the band broadening and the small redshifts in the UV–vis spectra (Figure S2).[21,22] The removal of the metal ions from dpa by LPS, which can enhance excimer emission, did not occur because the UV–vis spectrum of Zn-dpa-C4Py with LPS was obviously different from that of dpa-C4Py; the absorption peaks of the probe clearly shifted to the shorter wavelength region by the addition of Zn2+, and the absorbance was decreased with a slight redshift by the addition of LPS (Figure S2). The Turn-OFF response toward LPS was unique to Ni-dpa-C4Py (Figure S3). The complexation rate of Ni2+ is known to be lower than those of the other divalent metals due to its electron configuration.[23] In fact, the intensity of the excimer emission decreased with time after the addition of Ni2+ into dpa-C4Py solution (Figure S4). Because Ni2+ was slowly coordinated by dpa-C4Py, the probe gradually dissolved with producing weak excimer emission (Figure ). We found that the decrease rate of the excimer emission intensity was high when LPS co-existed (Figure S4). LPS may work as a surfactant to improve probe solubility in water. On the other hand, excimer emission was not enhanced by the addition of Co2+ and Cu2+. Ligand-to-metal charge transfer (LMCT) plausibly occurs because Co2+ and Cu2+ possess unoccupied d-orbitals, whereas Zn2+ and Cd2+ are homologous elements of group 12 in the periodic table, and their d-orbitals are occupied by electrons (Figure ).[16]

Figure 2

Fluorescence spectra of M-dpa-C4Py (M = Cu2+, Ni2+, and Zn2+) without (solid line) and with (dashed line) 1 μM LPS in 1% DMSO/99% water (v/v) (λex = 350 nm). [dpa-C4Py] = 0.01 mM, [HEPES] = 5 mM, [M(NO3)2] = 0.01 mM, pH 7.4, and 25 °C.

Figure 3

Changes in fluorescence intensity of dpa-C4Py and M-dpa-C4Py (M = Co2+, Ni2+, Cu2+, Zn2+, and Cd2+) by the addition of 1 μM LPS in 1% DMSO/99% water (v/v) (λex = 350 nm). [dpa-C4Py] = 0.01 mM, [HEPES] = 5 mM, [M(NO3)2] = 0.01 mM, pH 7.4, and 25 °C.

Figure 4

Proposed sensing mechanisms of M-dpa-C4Py (M = Co2+, Ni2+, Cu2+, Zn2+, and Cd2+) for LPS.

In contrast, little change in the fluorescence spectra of dpa-C1Py was observed by the addition of the metal ions and LPS (Figure S1). Hence, the length of the spacer drastically influenced the recognition abilities of the probes. The methylene spacer of M-dpa-C1Py would be too short or too rigid to form the pyrene dimer upon binding to LPS.

From the perspective of practical sensors, the Turn-ON response is generally preferable because it can be clearly distinguished from the background emission.[24] In addition, because Cd is highly toxic, we selected Zn-dpa-C4Py as the best candidate for practical usage, and the feasibility of this probe was further investigated, as described below.

Titration of LPS by Zn-dpa-C4Py

Figure shows the fluorescence spectra of Zn-dpa-C4Py with different LPS concentrations. The excimer emission at 470 nm was dramatically enhanced by increasing the concentration of LPS. Furthermore, the change in emission color was clearly observed by the naked eye under UV irradiation in the presence of LPS (Figure A and Figure S5). The fluorescence intensity was increased linearly with increasing LPS from 0.1 to 1 nM. The limit of detection (LOD) was determined to be 41 pM. To the best of our knowledge, this LOD is the lowest among those of reported synthetic fluorescent chemical sensors for LPS (Table ).

Figure 5

Fluorescence titration spectra of Zn-dpa-C4Py upon addition of LPS (A) and calibration curve (B) in 1% DMSO/99% water (v/v) (λex = 350 nm). The inset in (A) is a photograph of the corresponding color change in the absence (left) and presence (right) of 1 μM LPS under UV irradiation in the dark. [dpa-C4Py] = 0.01 mM, [HEPES] = 5 mM, [Zn(NO3)2] = 0.01 mM, pH 7.4, and 25 °C. Each data point is the average of five measurements under the same conditions.

Table 1

Comparison of Reported Synthetic Fluorescent Chemical Sensors for LPS with Zn-dpa-C4Pya

fluorescent probe	LOD (pM)	reaction time (min)	ref
GO	130	N/A	(15)
QD-Apt-GO	870	30	(26)
ROX-LBA/GO	1570	N/A	(27)
MTA-Au	N/A	20	(28)
HDT-AuNPs	650	N/A	(29)
[Pt(N∧N∧N)Cl]+	5700	N/A	(30)
CPT1	270	N/A	(31)
peptide-diacetylene amphiphiles	N/A	N/A	(32)
BD2C	2600	N/A	(3)
TPEPyE	370	N/A	(33)
BT-5	120	0.5	(34)
CTPY-P16	6970	N/A	(35)
DMQA	100,000	N/A	(22)
BPTG	5000	N/A	(36)
Sp-Py	1000–10,000	N/A	(37)
Zn-dpa-C4Py	41*	<1	this work

The molecular weight of LPS was assumed to be 10 kDa to calculate LOD in pM units.[15,25] *41 pM is equivalent to 4.1 EU/mL, given that 100 pg/mL of LPS corresponds to 1 EU/mL.[43]

Sensing Mechanism

From the spectrum change obtained by the titration experiment, we calculated the Py value (Figure ), which is the ratio of the fluorescence intensities of pyrene band I (375 nm)/band III (385 nm) at various LPS concentrations. This value describes the polarity of the microenvironment around the probe; a higher Py value indicates a more polar environment.[38−40] Judging from Figure , the probe existed in a more hydrophobic environment with increasing concentration of LPS. It is likely that the probe self-assembles with LPS because LPS forms aggregates like micelles/vesicles in water (Figure ).[41−43] Amphiphilic Zn-dpa-C4Py possessing hydrophobic n-butyl pyrene and cationic Zn2+ coordinated by dpa will show high affinity toward the fatty acid chains and the phosphate groups of LPS. The decline of Py value caused by LPS addition was also reported using pyrene to investigate the aggregation behavior of LPS in the micromolar order.[41] However, Zn-dpa-C4Py showed the decline of Py value at lower concentrations of LPS in the nanomolar order. The positive divalent charge of Zn2+ probably offsets the negative charge of the phosphate groups in LPS to facilitate the aggregation of LPS at lower concentration because the counter ions of ionic amphiphiles generally reduce the charge repulsion between the amphiphiles and lower the critical micelle concentration.

Figure 6

Py values of Zn-dpa-C4Py at various LPS concentrations shown in Figure . Each plot is the average of five measurements under the same conditions.

Figure 7

Schematic diagram of the co-aggregation of Zn-dpa-C4Py with LPS.

To elucidate the co-aggregation behavior, we measured the particle size distribution and the zeta potential by the DLS technique (Figure ). The results showed that LPS itself formed particles measuring 39 ± 4 nm in diameter, which is consistent with reported data (10–50 nm range).[41,44,45] The average particle size of Zn-dpa-C4Py was 31 ± 13 and 36 ± 4 nm before and after the addition of LPS, respectively, suggesting that Zn-dpa-C4Py formed micelles/vesicles even without LPS because of its amphiphilic structure. Although the particle sizes were almost identical, each peak showed a near-Gaussian distribution, indicating that each compound formed almost homogeneous aggregates. The zeta potential of LPS particles was – 8.6 ± 0.8 mV, which is consistent with reported values (−14 to −6 mV).[45−47] This negative charge is derived from the phosphate groups and carboxyl groups. In contrast, Zn-dpa-C4Py showed the positive value of 1.0 ± 0.4 mV because of the charge of Zn2+. We found that Zn-dpa-C4Py showed the zeta potential of −8.3 ± 1.5 mV in the presence of LPS, implying the formation of a supramolecular complex with LPS.

Figure 8

Size distribution (A) and zeta potential (B) of micelles/vesicles formed by LPS, Zn-dpa-C4Py, and the complex of Zn-dpa-C4Py with LPS, respectively, in 5 mM HEPES (pH 7.4). Each result is the average of three measurements under the same conditions.

We conclude that surfactant-like Zn-dpa-C4Py recognized amphiphilic LPS through multiple points including negative phosphate groups and hydrophobic fatty acid chains by forming co-aggregates like micelles/vesicles. Then, the pyrene moieties of the sensor got close to each other, resulting in the excimer emission at 470 nm. It is possible that the aggregate formation enables sensitive detection for the following reasons: (1) it increases the local concentration of LPS from bulk water, and (2) it boosts the fluorescence response to LPS such as several sensors exploiting micelle formation with enhanced quantum yield.[14,48,49]

Selectivity of Zn-dpa-C4Py

The selectivity of Zn-dpa-C4Py was evaluated by monitoring its fluorescence response to other biologically important phosphate derivatives (Pi: phosphate, PPi: pyrophosphate, Tri: triphosphate, AMP: adenosine monophosphate, ADP: adenosine diphosphate, and ATP: adenosine triphosphate) as possible interferents. Figure shows that LPS displayed the strongest excimer fluorescence enhancement compared with the other phosphate interferents (black columns). Cho et al. reported that Zn-dpa-C4Py is a pyrophosphate sensor but did not evaluate LPS as an analyte.[50] We demonstrated that this sensor recognized LPS more selectively than the other pyrophosphate derivatives. Moreover, small effects on LPS sensing by the other phosphate derivatives were found in the interference assay (gray columns). The large amount of negatively charged groups on the LPS molecules including two phosphate groups in lipid A make LPS highly negatively charged.[47] The difference in the number of phosphates per unit molecule and the highly negative charge of LPS probably contributed to the excellent selectivity of the probe for LPS because of the more recognition targets for dpa moieties and the strong electrostatic attraction between LPS and the positively charged sensor. A similar mechanism was reported for the positively charged tetraphenylethylene-based sensor for LPS.[33] In addition, the two-point sensing mechanism and the co-aggregation described above seemed to produce a stronger binding affinity and signal toward LPS.

Figure 9

Changes in fluorescence intensity of Zn-dpa-C4Py in the presence of phosphate derivatives in 1% DMSO/99% water (v/v) (λex = 350 nm). [dpa-C4Py] = 0.01 mM, [HEPES] = 5 mM, [Zn(NO3)2] = 0.01 mM, pH 7.4, and 25 °C.

Conclusions

Zn-dpa-C4Py showed pyrene-moiety-derived excimer emission after the addition of LPS without cumbersome sample preparation. Fluorescence measurements demonstrated that the probe formed co-aggregates with LPS by multi-point recognition of LPS through electrostatic and hydrophobic interactions, leading to the formation of pyrene dimers. Zn-dpa-C4Py showed excellent sensitivity down to the picomolar order (LOD = 41 pM), which is the best among the reported synthetic fluorescent chemical sensors for LPS. Furthermore, selectivity and interference assays using a series of phosphate derivatives revealed that the selectivity of this probe was the highest for LPS and interference was limited even in the presence of the other phosphate derivatives. This fluorescent Turn-ON sensor offers great potential for practical endotoxin/LPS detection to control the quality of pure water used in pharmaceuticals and dialysis.