Optical Nanosensors for in vivo Physiological Chloride Detection for Monitoring Cystic Fibrosis Treatment

Abstract

Personalized approaches for continuous monitoring of chloride levels are potentially valuable for evaluating the efficacy of new treatments of genetic disorders such as cystic fibrosis. In this report, we validated optode-based nanosensors for real-time chloride monitoring in the interstitial fluid of living animals. These nanosensors take advantage of a ratiometric sensing scheme which demonstrates reversible and selective chloride detection in the physiological range. We further investigate how skin pigmentation affects the sensor performance during in vivo fluorescence imaging. We successfully monitored endogenous chloride changes using nanosensors during pharmacological treatment in a cystic fibrosis mouse model. We believe this platform is a valuable tool for chloride detection which could assess the efficacy of new treatments for cystic fibrosis.

Graphical Abstract

Endogenous chloride fluctuations were successfully monitored in real-time using optode-based nanosensors during pharmacological treatment in a cystic fibrosis mouse model.

Personalized medicine (PM) targeting customized treatment regimens and preventive care has been getting increasing attention over the years, especially for cancer and genetic diseases^{1, 2}. For example, PM has been considered as one of the most promising approaches for cystic fibrosis due to the disease complexity and variations^{3, 4}. Cystic fibrosis (CF) is a monogenetic disease, caused by a genetic mutation in the cystic fibrosis transmembrane conductance regulator (CFTR) gene that produces the CFTR protein⁵ with impaired sodium-chloride homeostasis and disruption of chloride transport⁶. To date, over 2,000 types of CFTR mutations have been reported^{7, 8}; these range from common mutations encompassing six main classifications based on defect to rarer mutations with no known source⁹.

The concept of PM is to offer a tailored treatment based on an individual’s phenotype, environment, and genetic traits instead of “all-in-one” approach^{10, 11}. Importantly, PM requires careful and continuous monitoring of therapeutic drug levels or therapy-related indicators and biomarkers as key parameters^12-14. The (quasi) real-time monitoring of these substances is crucial to help improve the efficacy and reduce side effects during treatment. This approach could be particularly important during CF treatment, where combination therapies are needed for different variations in mutations. Continuous monitoring of efficacy indicators for each individual could provide real-time feedback to a specific therapy strategy.

Currently, the gold standard for preliminary CF diagnosis is the sweat test, however, this historical tool has its limitations including off-line results and insufficient amount of sweat for robust assessment. An important treatment evaluation method is the nasal potential difference (NPD) measurement^{15, 16} which can validate CFTR function and provide in vivo evidence of abnormal chloride transport^{17, 18}. By placing an electrode on the lining of the nose of the patient, the potential across the nasal epithelium changes correspondingly to mucosal perfusion with varying salt solutions^{19, 20}. NPD testing is uncomfortable to patients and is technically difficult to perform; thus, it is not suitable for long-term monitoring and assessment²¹. In addition, the interpretation of PD values can be challenging in animal studies since the relative distribution of respiratory and olfactory epithelium differs with distance from the nose and it is unclear where in the mouse nose the PD should be measured. Therefore, NPD measurements are highly variable in rodents and leads to limited within-subject repeatability in rodent research²². As such, the design of better preclinical assays that can help monitor chloride transport on an individual basis are highly needed.

One potential solution is continuous monitoring of chloride level fluctuations in the interstitial fluid. The CFTR protein is a chloride channel expressed on epithelial cell membranes in organs such as sweat glands as well as tissues in the central, peripheral and enteric nervous systems^23-25. In sweat glands, defective CFTR chloride channels lead to a high chloride concentration in the ductal lumen and the interstitial chloride concentration becomes lower than normal physiological levels²⁶. Thus, measurement of interstitial fluid chloride levels could provide information of chloride fluctuations caused by CFTR channels²⁷.

Over the decades many types of chloride sensors based on an array of detection principles have been described^28-31. The optical counterparts of ion-selective electrodes, called optodes, do not require an electrical connection between sensing and detection units. Thus, they are well-suited to monitoring a wide range of analytes with various shapes and sizes for biological sensing applications ^{32, 33}. These nanosensors have major advantages, such as tunable response in the desired dynamic range, reversible performance to detect analyte fluctuations³⁴, maximal image penetration depth by incorporating near-infrared (NIR) fluorophores³⁵. The first nanometer scale anion sensing fluorescent spherical nanosensors, or PEBBLEs (probes encapsulated by biologically localized embedding) were demonstrated for the intracellular monitoring of chloride by the Kopelman Group³⁶. The PEBBLES were developed to show robust sensitivity for intracellular chloride detection in the range of 5-15 mM in vitro. However, this formulation has not have been used for in vivo applications since its dynamic range and stability are better suited to intracellular measurements.

In the present study, we report the development of optode-based chloride-selective nanosensors capable of detecting chloride dynamics in vivo. We evaluated the analytical performance of the nanosensors, and further demonstrated the capability of subcutaneously injected nanosensors to monitor chloride changes in the interstitial fluid (ISF) in a cystic fibrosis mouse model after treatment of phosphodiesterase (PDE) inhibitors³⁷. The chloride level changes could give insight into therapeutic effectiveness of treatment based on chloride transport, which is vital for evaluation of personalized medicine efficacy and the development of CF therapy. Moreover, we also explored the effects of skin pigment on sensor performance which indicated nominal interference in darker skin that could be calibrated due to the ratiometric signal of the sensor. Taken together, we believe this work not only provides a successful advance of nanosensors for in vivo chloride monitoring, but also validate a platform for sensing other anions.

Material and methods:

Materials

Poly (vinyl chloride), high molecular weight (PVC), bis(2-ethylhexyl) sebacate (DOS), chromoionophore IV (CHIV), chloride ionophore IV (Cl IV), Tridodecyl methyl ammonium chloride (TTDMAC), tetrahydrofuran (THF), 4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid (HEPES), dichloromethane (DCM), sodium chloride, Pluronic F127 was purchased from Sigma-Aldrich (St. Louis, MO, USA). 1,2-Disteroyl-sn-glycero-3-phosphoethanolamine-N- [methoxy (polyethylene glycol)-550] ammonium salt in chloroform (DSPE-PEG) was purchased from Avanti Polar Lipids (Alabaster, AL, USA). (Dil) was purchased from Life Technologies (Grand Island, NY, USA). CellBrite™ NIR750 was purchased from Biotium (Fremont, CA). 2-Amino-2-hydroxymethylpropane-1,3-diol, 2 M solution (TRIS, 2M), was purchased from Fisher Scientific (Waltham, MA, USA). Phosphate-buffered saline (PBS with Ca²⁺ and Mg²⁺, pH = 7.4) was purchased from Boston Bioproducts (Ashland, MA, USA).

Nanosensor Fabrication

The protocol for nanosensor fabrication is described in previous papers published by our group^{38, 39}. The optode cocktail comprised PVC, DOS, and all other sensing components in 500 μL of THF. The formulation of the optode cocktail in this paper was 30 mg of PVC, 66 μL (60mg) of DOS, 0.7 mg of CH IV, 1.4 mg of Chloride ionophore IV, 0.7 mg of TTDMAC, 1 mg of DiI and 2 mg of NIR750.

4 mg of Pluronic F127 in THF was dried in a 4-dram scintillation vial and then resuspended in 4 mL of 10 mM HEPES solution (pH = 7.4, adjusted with 1M Tris Base) with a probe tip sonicator for 30s with an altitude of 20% (Branson, Danbury, CT, USA). Later, 75 μL of DCM was added in a PCR tube and mixed with 50 μL of optode cocktail solution by pipetting. The mixture was added into the Pluronic F127 solution under probe tip sonication (3 minutes with 20% altitude). Finally, to remove any larger pieces of polymer, the solution was filtered with a 0.45 μm syringe filter (Pall Corporation, Port Washington, NY, USA).

The size and zeta-potential of nanosensors were characterized with a Dynamic lightening scattering (DLS) Brookhaven 90Plus (Brookhaven Instruments, Holtsville, NY, USA) with the intensity of scattered 640 nm light at a fixed angle of 90°. The size was measured in triplicates with diluted sensor solution at a detector count rate of 150-450 kcps. The zeta-potential measurements were taken in PALS mode with 10 runs per cycle and total 25 cycles.

Nanosensor characterization

Nanosensors were calibrated at different concentrations of sodium chloride solutions in a 96-well plate. The fluorescence intensities for DiI (λ_EX: 540 nm, λ_EM: 575 nm) and NIR 750 (λ_EX: 748 nm, λ_EM: 780 nm) were measured with plate reader (SpectraMax M3, Molecular Device, San Jose, CA). The ratio, R, of the emission fluorescence intensities of two fluorophores (Dil/NIR750) was calculated as:

$$R=\frac{575nmEmission}{780nmEmission}$$ (1)

The ratio for 575 nm/780 nm were then normalized by calculating the protonation state degree (1-α) as:

$$1-\alpha =\frac{Ri-Rd}{Rp-Rd}$$ (2)

Where Ri is the ratio of sensor fluorescence at a particular concentration of NaCl. Rp and Rd represent the fully protonation and deprotonation states, respectively. The center value of dynamic range, which is the Cl⁻ concentration at half-maximal response was calculated based on the calibration curve fitted according to a dose-response (Hill) equation from different chloride concentrations and their corresponding values of protonation degree.

Selectivity

Here we used mixed solution method which is suitable for complex systems to evaluate the selectivity of the chloride-selective nanosensors. The mixed interference solution mimics the ISF environment by incorporating all the primary anions with their physiological concentrations including 0.02 mM Nal, 0.07 mM NaBr, 1.3 mM NaHPO₃, 0.4 mM Na₂SO₄, 28.0 mM NaHCO₃, and 0.05mM NaNO₃ ⁴⁰. We prepared different concentrations of NaCl in two buffer systems: one is 10 mM HEPES buffer (pH 7.4) without interfering ions; the other is the mixed interfering solution in 10 mM HEPES buffer (pH 7.4). Thus, we were able to compare the chloride nanosensor performances with and without interferences and evaluate the specificity of chloride nanosoensors.

In vivo studies

All animal procedures were approved by the Northeastern University Institutional Animal Care and Use Committee (IACUC) and were in accordance with the National Institutes of Health guidelines. Fluorescence imaging experiments were performed by using a Lumina II in vivo imaging system (IVIS) (Perkin Elmer Instruments, Shelton, CT, USA). Animals were anesthetized with 2% isoflurane in oxygen and then placed on the warm stage of the IVIS with a 1.5% isoflurane maintenance rate. The fluorescence signals were taken with two channels. Dil: excitation filter centered at 535 nm (bandpass: 30 nm) and emission filter from 575 to 650 nm and 1 second exposure. NIR750: excitation filter centered at 750 nm (bandpass: 30 nm), emission filter from 810 nm to 875 nm and 1 second exposure time. All nanosensors to be injected in vivo were concentrated 30-fold with Amicon Ultra centrifugal filters (100kDa, 0.5 mL, MWCO, Millipore Corporation, Billerica, MA, USA).

In order to assess the effects of skin pigmentation on nanosensor response, we used Sentinel rats (Charles River Laboratories, Wilmington, MA), which have both pigmented and non-pigmented areas of the skin after depilation. Nanosensors suspended in solution with varying levels of sodium chloride were injected intradermally into the rats on the back: 50 mM, 100 mM, and 150 mM in both areas, respectively. After injection, the intensities at both wavelengths were measured immediately in both non-pigmented and pigmented areas every 2 minutes in the duration of 1-hour. Regions of interest were chosen based on where nanosensors injected and the responses were plotted as a fluorescence ratio of Dil/NIR750 over time. We normalized the ratio to the 50 mM of NaCl on each animal and averaged these data among the six animals. Error bars represented the value of the standard deviation in each group. P value was calculated to determine the significant difference between the mean in two groups using the t-test from GraphPad.

We used CFTR^tm1Unc Tg(FABPCFTR)1Jaw/J (The Jackson Laboratory, Bar Harbor, ME)) mice as the model for cystic fibrosis and CD-1 mice as the control group. 5 μL concentrated chloride-selective nanosensors were injected to the two footpads of the mice where sweat glands are located. Baseline fluorescence measurement was acquired 10 minutes before Vardenafil administration. The chloride concentration in ISF for the baseline in CD-1 control mouse is assumed as standard 100 mM based on literature⁴¹. An intraperitoneal injection of Vardenafil at 0.14 mg/kg was administered and image was acquired in the interval of 10 minutes for another 80 minutes. Chloride concentrations was then calculated based on the relation between raw fluorescence ratio (Dil/NIR750) and their corresponding chloride level in the linear dynamic range. The linear range was determined by calibration curve with linear deviation of 5%. Error bars represented the values of the standard deviation in each group (n=12). The curve fit was generated based on a two-compartment model with first-order input using MATLAB/SimBiology. Statistical analysis was calculated using student’s t-test with GraphPad and significance was set at 95% confidence (a=0.05). When the p < 0.05, the difference was considered as significant. P value was calculated to determine the significant differences among different time points for within group as well as in both CD-1 and cystic fibrosis groups.

Results and discussion

The mechanism for anion recognition and signaling is based on the co-extraction between the lipophilic sensor particle and the surrounding aqueous phase (Figure 1), and is based on extensive previous work³⁶. In brief, the optode is hydrophobic since it is comprised of a highly plasticized polymer, and retains lipophilic sensing components that are free to diffuse throughout the sensing particle. Namely, an ion-selective ionophore, a pH-responsive chromoionophore, and an ionic additive are used to tune the response to the appropriate dynamic range. The chloride-selective ionophore extracts chloride ions into the sensing phase, while chromoionophore co-extracts protons at the same time to maintain the electroneutrality in the sensor. This changes the protonation state and consequently the fluorescent signature of the chromoionophore, resulting in an indirect metric of the chloride concentration. The optode-based chloride-selective nanosensors are sensitive in the physiological range, reversible for multiple cycles, and selective to chloride in the presence of other anions in the ISF.

Fig. 1:

Co-extraction sensing mechanism for chloride nanosensors. Chromoionophore (C), Chloride Ionophore (I), NIR 750 (Ref), DiI (D), and Ionic Additive (R) are encapsulated in the polymeric optode matrix. Chloride ion and proton are extracted simultaneously to the sensor core which causes protonation of pH-sensitive absorber chromoionophore until it reaches equilibrium. This will lead to fluorescence intensity change of Dil through the inner filter effect. The intensity ratio between Dil and reference dye NIR 750 therefore can be used to quantify the chloride concentration in the physiological surrounding. Ionic additive is used to maintain electroneutrality.

The dynamic range for chloride monitoring in vivo is from 60 mM to 140 mM, which corresponds to the reportable range of clinical monitoring devices. Earlier studies have shown that the ratio of the components, especially the charged ionic additives, will shift the dynamic range of the sensor⁴². Thus, the sensor response can be tuned to a dynamic range that facilitates further in vivo study. Although the materials, including the polymer, plasticizer, and ionic additive were fairly standard, we made some adjustments in the optical components in order to meet the design requirements for in vivo analysis (REF Tim’s review paper). We chose CHIV as the chromoionophore for two reasons: (1) the pKa of CHIV (ETH 2412) is over 17.00±0.04 ⁴³, which shifts the dynamic range into that appropriate for the interstitial chloride ; and (2) chloride ions do not interfere with the protonated form of ETH 2412 in PVC–DOS system⁴³. However, CHIV does not have ideal fluorescent properties. Specifically, while it absorbs well at 535 nm when deprotonated, the fluorescence emission is too weak for in vivo use. Thus, we introduced another hydrophobic fluorophore, Dil, whose excitation overlaps with the deprotonation absorbance peak (Figure S1) and leads to energy transfer⁴⁴. Additionally, NIR750 is used in the optode formulation as an internal reference standard to avoid error caused by differences in nanosensor concentration or depth of injection.

We used a nanoemulsion fabrication process that had been successfully developed for other nanosensors to detect ions and small molecules^{42, 45-47}. We compared different coating materials for the chloride nanosensor formulation (Figure S2). Since chloride-selective nanosensors need to extract surrounding negatively charged chloride ions, the surface charge of nanosensors also plays a key role in the responsiveness. Traditional PEGylated coatings lead to a negative surface charge and repel chloride ions. Thus, Pluronic F127, which provides a slightly positive surface charge, was used in this formulation. The size and zeta-potential of chloride-selective nanosensors coated with Pluronic F127 were measured as 177.7 ± 18.5 nm and 14.0 ± 2.5 mV with dynamic light scattering (DLS) compared to 186.4 ± 6.3 nm and - 15.4 ± 1.5 mV of the counterparts coated with PEG-Lipid. The calibration curve for the Pluronic F127 formulation is shown in Figure 2 is a function of sodium chloride concentration verses ratiometric measurement. The center of the dynamic range of the Pluronic F127 chloride nanosensor is 170 ± 31 mM and the total change in the physiological range (60-140mM) is 25 ± 3 % and the sensitivity is 0.35% change per mM. Although the ratio of all the sensing components impacts the sensitivity of the measurement, the ionic additive has the largest effect. Figure 3a shows the effect of changing the amount of additive on the sensitivity of the chloride measurement. The trend shows that a smaller relative amount of additive leads to a steeper slope of the linear portion of the calibration curve, indicating a higher sensitivity of the response.

Fig. 2:

Sensor response to different chloride concentrations. The sensor performance can be characterized by the effective chloride concentration at half-maximal sensor response (defined here as EC₅₀). The EC₅₀ value shown in the] calibration curve is 170 ± 31 mM. and has linear response in the reportable range of interests, which is 60 to 140mM.

Fig.3.

(a) Changes in sensor formulation affect nanosensor performance. Nanosensor is less sensitive with an increasing amount of ion additive, for instance, as demonstrated by reducing slope at the linear range of the calibration curve; (b) The nanosensor is selective to chloride against other interfering physiological anions. Red: the calibration curve of chloride nanosensors in the presence the interfering anions in the 10 mM HEPES buffer; Black: the calibration curve without any interfering anions. The calibration curves almost overlapped, which indicates that the nanosensor response is specific to chloride.

To study the specificity of the chloride-selective nanosensors, we used a mixed solution method with interfering solutions that contain the primary anions at their physiological concentrations. The mixed solution method is used here since ISF is a complex system and this method can take the interfering effects as a whole into account. The optode-based nanosensor selectivity is usually determined by the ionophore in the optode formulation. According to the literature, the chloride ionophore IV shows a good selectivity for chloride ions^{48, 49}. Our results, as seen in Figure 3b shows that our chloride-selective nanosensors responded as anticipated in the presence of interfering anions in the 10 mM HEPES buffer (pH 7.4) and the calibration curves shifted from 195 ± 13 mM to 236 ± 18 mM at the center of the dynamic range, showing no significant difference (p= 0.1949). This selectivity study showed that our nanosensor responses are specific to chloride with minimal interference from other primary anions in the physiological ISF enviroment⁴⁰.

Stability was also demonstrated by the chloride nanosensor response stability to show that they maintain their response characteristics. Figure S3 shows that the dynamic range of response shifted less than 5 mM over eight days in solution, which is similar stability compared to previous polymer-based sensors developed by our lab⁵⁰. Collectively, we have fabricated the chloride-selective nanosensors that have the characteristics for reporting chloride change in vivo by showing millimolar dynamic range without interference from other primary anions in ISF.

There are always challenges in in vivo fluorescence imaging. Namely, tissue absorption and scattering notably impact fluorescence emission and limits its use for quantitative measurements. Moreover, darker skin pigmentation may further attenuate fluorescent signal and make in vivo optical measurements vary among individuals. Thus, we evaluated the effect of skin pigmentation of animals on the response of our optode-based nanosensors. We used sentinel rats which have areas of both white and black coat colors and thus non-pigmented and pigmented skin after depilation, as shown in Figure 4. We prepared nanosensors in NaCl solutions at 50mM, 100 mM, and 150 mM, which cover the whole pathological range and used these to produce an in vivo calibration curve by injected each solution separately. Figure S4 shows the chloride-selective nanosensors’ response in non-pigmented and pigmented skin regions at time zero minutes. The normalized values of these concentrations indicate a sensitivity based on the slope of the fit of 4.5 ± 0.8 % and 4.7 ± 0.8 % fluorescence change per 10 mM chloride change, respectively. The p value of the sensitivity was 0.9148, indicating there is no significant difference between non-pigmented and pigmented skin.

Fig. 4.

Effect of skin pigments to the performance of the chloride nanosensors. (a,b) Sensors with different concentrations of NaCl solutions were injected and imaged at the pigmented skin region in one Sentinel rat. a: DiI channel; b: NIR 750 channel, (c) Changes in sensor response with 50 mM, 100 mM and 150 mM NaCl solutions over time at pigmented area (n=3). (d,e) Sensors with different concentrations of NaCl solutions were injected and imaged at the non-pigmented area in one Sentinel rat. d: DiI channel; e: NIR 750 channel, (f) Changes in sensor response with 50 mM, 100 mM and 150 mM NaCl solutions over time at non-pigmented area (n=3). Trendlines are included for visual guidance.

Subsequently, fluorescence measurements were taken every 2 minutes for duration of 1 hour. The result also shows that the NaCl solution used as a carrier diffused quickly after injection; the nanosensors with 50 mM NaCl solution equilibrated with physiological concentration within minutes which resulted in an increased fluorescent intensity (Figure 4c and 4f). In contrast, the fluorescence signal decreased in the spot where nanosensors with 150 mM NaCl solution were injected as the solution equilibrated to physiological levels. Therefore, this time-series figure validated that the chloride-selective nanosensors are responsive to different chloride levels in both non-pigmented and pigmented skin regions and after several minutes, nanosensors equilibrate to physiological chloride concentration. Notably, the intensity ratio values in these two groups are different as well as the raw fluorescence intensities. We assume that the melanin attenuates the emission of NIR750 less than the that of the shorter wavelength Dil^{51, 52}. Thus, the ratio in the pigmented group decreases compared to the non-pigmented group. This demonstrates that variation in skin tone lead to attenuated response despite taking a ratiometric measurement. Taken together, we predict that these sensors could be used in varying skin types, but that the response would need to be calibrated to an individual and normalized as above. Thus, ratiometric measurements can cannot be seen as a panacea to skin variability, but rather need to be normalized to an individual.

Previous studies have shown that phosphodiesterase type 5(PDE5) inhibitors can improve defective cAMP-dependent CFTR chloride transport across the CF mouse by increasing cAMP concentration^37,53,54 CFTR-dependent chloride channels not only are located in the nasal mucosa, but also are highly expressed in the sweat gland^{55, 56}. To evaluate the nanosensors, we used a murine CF model, treated with the PDE5 inhibitor Vardenafil, and determined the chloride concentration changes in the sweat glands by injecting the nanosensors in the footpads of mice where the sweat glands are located⁵⁷. CD-1 mice were used as control, since the fully-functional chloride channels should not be affected by the presence of drug. The changes in fluorescence intensity of both the CF model and the control during drug administration are shown in Figure S5. In the CF group, the nanosensor fluorescence intensity initially increased 22 % after 30 min and declined gradually after 60 min, while in the CD-1 the nanosensors do not have significant fluctuations in fluorescence. Since the fluorescence intensity changes are proportional to the chloride level in the ISF, we could retrieve valuable information of underlying chloride dynamics based on sensor response during pharmacological stimulation.

Furthermore, we estimated the chloride concentrations in mice based on the sensor responses obtained at the non-pigmented area (Figure.S5). We assumed that the basal chloride level in the CD-1 ISF is 100 mM which we correlate to the corresponding nanosensor response 10 minutes before drug administration^{41, 58, 59}. Based on a linear relationship of sensor responses at physiological chloride levels, the sensor readings can be converted to estimated chloride concentrations, as shown in Figure 5. Specifically, the CF group chloride levels reach a peak of 112 ± 11 mM at 30 minutes, which is a significant increase from the initial level of 96 ± 6 mM measured before the drug administration (p =0.003). The chloride concentration plateaued at 30 minutes then declined to 106 ± 15 mM. Moreover, the chloride level was treated as a response to drug administration, so we fit the chloride dynamics after Vardenafil treatment using a two-compartment model with first-order input. We found this time-course is consistent to the pharmacokinetic profile of Vardenafil³⁷ where the elimination half-life of PDE5 inhibitors in rodents is approximately 0.4h to 1.3h⁶⁰. We further observed the initial chloride concentrations between the two groups are slightly different with a lower concentration for the CF model, possibly due to chloride deficiency in the CF model⁶¹. The errors bars in the CF model are larger than those in the CD-1 group, and we postulate that diseased animals may have greater variability in chloride levels than healthy CD-1 mice, although this is difficult to confirm since there is currently no gold standard for independent measurement of chloride levels in ISF in mice.

Fig 5.

Chloride nanosensor response to endogenous chloride dynamics in both CD-1 and Cystic fibrosis (CF) mice under the treatment of Vardenafil. Time 0 is defined as the time when Vardenafil was administrated via intraperitoneal method. Chloride concentrations were calculated based on in vivo calibration curve from non-pigmented area (n = 12 for each group). Red line: The fitted curve of chloride response in CF model is based on a two-compartment model with first-order input.

Conclusions

In this study, we developed optode-based chloride-selective nanosensors based on a well-understood co-extraction mechanism. Our work used a nanoemulsion process to fabricate the chloride-sensitive sensors for monitoring chloride levels continuously in real-time in the ISF. Moreover, the specificity to chloride shows these nanosensors can accurately monitor the chloride levels in the presence of other interfering anions in the physiological environment. Finally, we validated the sensor performance in vivo by comparing the sensor response both in non-pigmentated and pigmentated skin areas. Also, we used nanosensors to continuously monitor the endogenous chloride level changes for over 1 hour in CF mice model during administration of PDE inhibitors. The chloride-selective nanosensors have potential to be a tool for preclinical drug screening for cystic fibrosis, or possibly for monitoring drug efficacy on an individual basis.