A method for estimating intracellular ion concentration using optical nanosensors and ratiometric imaging

Optical nanoparticle (NP)-based sensors have been widely implemented as tools for detection of targeted ions and biomolecules. The NP sensing platform offer a modular design that can incorporate different sensing components for greater target specificity and the ability to tune the dynamic range, as well as encapsulation of multiple dyes to generate a ratiometric signal with varying spectra. Despite these advantages, demonstrating quantitative ion imaging for intracellular measurement still possess a major challenge. Here, we describe fundamentals that enable intracellular validation of this approach using ion-selective nanosensors for investigating calcium (Ca2+) as a model ion. While conventional indicators can improve individual aspects of indicator performance such as Kd, wavelength, and ratiometric measurements, the use of NP sensors can achieve combined benefits of addressing these issues simultaneously. The nanosensor incorporates highly calcium-selective ionophores and two fluorescence indicators that act as signal transducers to facilitate quantitative ratiometric imaging. For intracellular Ca2+ application, the sensors are fine-tuned to physiological sensing range, and live-cell imaging and quantification are demonstrated in HeLa cells loaded with nanosensors and their responsiveness to carbachol-evoked store release (~400 nM). The current nanosensor design thus provides a promising sensing platform for real-time detection and optical determination of intracellular ions.

Nanoparticle (NP)-based sensors [16][17][18][19][20][21][22] provides an efficient and promising approach for developing robust imaging probes for quantitative ion detection. For example, while molecular probes can improve individual aspects of indicator performance such as Kd, wavelength, and ratiometric measurements, the use of NP-based sensors can achieve combined benefits of addressing these issues simultaneously. By incorporating multiple agents in the NP matrix, the advantages of these classes of sensors include: (i) brighter and photostable probes by integrating a high number of dyes; (ii) ability to fine-tune the sensor dynamic range; (iii) greater selectivity and range of analyte detection; (iv) adjustable emission spectra and ratiometric response; (v) reduced effects from light-induced cell damage 23 by integrating longer wavelength dyes; and (vi) low cytoplasmic mobility [24][25][26][27] to prevent dye diffusion, and protection from intracellular organelle sequestration 28,29 . Indeed, the recent advances in NP-based sensors from various nanomaterials have been widely implemented in cellular tracking and imaging applications [16][17][18][19][20][21][22] . For example, there are several nano-and micro-particle sensors for detection of ions reported for sodium [30][31][32] , potassium [33][34][35][36][37] , calcium 38,39 , magnesium 40,41 , zinc 42 , silver 43 , and chloride 44 . Other efforts include nanosensors termed PEBBLEs (probes encapsulated by biologically localized embedding) developed by Kopelman and co-workers 28,45 . These sensors incorporate molecular indicators such as rhod-2 into a polymeric matrix, however, the flexibility to tune the dynamic range remains challenging with this mechanism 46 . While the development of NP-based sensors has been an increasingly growing field of research, the application for measurement of intracellular ions, and detailed procedure for quantitative ratiometric ion imaging has not been fully established in these sensors.
To address this issue, we present a ratiometric sensor termed calcium-optode nanosensor (opCaNS) for quantitative fluorescence imaging of ions in living cells, and establish a procedure for estimating ionic concentration using ratiometric fluorescent signal. Calcium is selected as a model target ion to reliably assess our sensor characteristics to the extensive work established on the local control of Ca 2+ signaling, as widely exemplified by cellular processes ranging from gene expression, synaptic transmission, hormone secretion to cell survival and death [47][48][49][50] .
In the nanosensor design, the ion-selective optodes (ISOs) 21,46,51 , an optical counterpart to ion-selective electrodes (ISEs), were applied and miniaturized to nanoscale for intracellular investigation. In the ISO sensing scheme, three reagents are embedded within the NP matrix: a highly selective, optically silent ion carrier (ionophore) 21, 46, 51 that is capable of reversibly binding ions; a fluorescent pH indicator dye (chromoionophore) which acts as a reporter to yield a fluorescence readout based on its protonation degree; and a lipophilic charge-carrying molecule (ion exchanger) to maintain electroneutrality within the NP. Hence, ion-specific optical change is facilitated by binding of calcium ions by the ionophores, which causes displacement of protons of the pH-indicator with a quantifiable fluorescence signal. Furthermore, we have incorporated a second pH-indicator [octadecyl rhodamine (R18)] and term it as reference dye, whereby fluorescence intensity moves in opposite direction to pH changes in comparison to the chromoionophore to achieve a highly sensitive sensor for ratiometric measurement of intracellular Ca 2+ . Most importantly, the sensing range of the probe is easily modified and tuned by changing the ratio of components, which provides flexibility to the user for specific imaging applications.
Overall, the opCaNS provides a sensing platform suitable for intracellular Ca 2+ measurement using ratiometric approach with sensor characteristics tuned to exhibit physiological sensing range, high selectivity, fast response time, and a reversible profile. Herein, we demonstrate the performance of this class of ion-selective nanosensors for cellular applications, achieving real-time detection and quantitative ion measurements by live-cell imaging for the first time.

Results
Principle and characterization of ratiometric nanosensor. To achieve nanosensors for real-time detection and measurement of intracellular free calcium ions ([Ca 2+ ] i ), we applied ISOs as the principle sensing scheme, and further developed into miniaturized nanoscale sensors for ratiometric fluorescence imaging. In this sensing platform, the nanosensors are fabricated by co-immobilizing hydrophobic optode components comprised of calcium ionophore (I), a fluorescent chromoionophore (C), ion exchanger (R), and reference fluorophore (Ref) within a biocompatible plasticizer nano-emulsion matrix, and stabilized by a biocompatible polyethylene glycol (PEG)-lipid as surfactant (Fig. 1). The hydrodynamic diameter of nanosensors measured 67 nm (polydispersity index 0.16) (Fig. 2a) with zeta-potential of −29.9 ± 1.4 mV, indicating nanoparticles that are stable in aqueous solution. Mechanistically, the highly selective calcium ionophore acts as a recognition element that extracts Ca 2+ into the sensor, which causes the pH-sensitive fluorophore (chromoionophore) to deprotonate to maintain charge neutrality within the sensor with an accompanying change in fluorescence. In this scheme, the opCaNS respond to Ca 2+ in accordance with optode function (Eqs 1-3) 21, 46, 52 . The ion-exchange optode mechanism in a system at equilibrium is described by: where, M, I, and C are the ion of interest, ionophore, and chromoionophore, respectively, and aq and org denote an ion, M, in aqueous or organic phase of the optode component, respectively. The exchange constant to describe the partitioning between the two phases in the ion-exchange system is then given by:  46 . Under these conditions, the relation between the deprotonation ratio (α) of the nanosensor and the ion activity is: For sensor calibration, the emission intensity ratios of two fluorophores (CHIII:R18, Eqs 4-6, see Methods) in response to Ca 2+ was used to plot a calibration curve with a Hill fit to determine effective Ca 2+ concentration at half-maximal sensor response (defined here as EC 50 ). The emission spectra for individual fluorophores are shown in Supplementary Figure S1. The sensing range can be fine-tuned (Fig. 2b) by altering the amount of the individual sensing components (CaI-II, CHIII, NaTFPB, see Methods) as previously demonstrated 53 .
For intracellular applications, the ratiometric response of nanosensor can be tuned to exhibit EC 50 at 105 ± 19 nM in solution (Fig. 2c), which is close to 100 nM intracellular calcium basal level 48 . There were batch-to-batch differences in sensor response and calibrations were routinely performed to ensure the sensors maintain optimal responsiveness for subsequent cellular studies. An index of nanosensor sensitivity, as determined by the slope of linear portion of the normalized response at EC 50 (Fig. 2c, inset), measured a decrease of 80% between 38 to 600 nM.
For sensor selectivity, the sensor responded to Ca 2+ with high selectivity over the primary intracellular interfering ion, magnesium (Mg 2+ ) (Fig. 2d). The opCaNS demonstrated EC 50 for magnesium >1 mM, which produces a log selectivity coefficient of 4 (Eq. 7, zero background ion). The physiological range of ionized Mg 2+ concentration is known to be 0.5-1 mM 54 , suggesting there is minimal interference of sensor response by Mg 2+ .
It is critical to determine the sensor response time to assess the ability of nanosensors to capture ionic transients which occur on the millisecond-to-second time scale [47][48][49][50] . Response time of the nanosensor was assessed by a two-photon (2P) Ca 2+ uncaging experiment (Fig. 2e). We first demonstrated that the opCaNS is 2P-excitable with maximum fluorescence observed at excitation wavelengths centered at 875 nm (Supplementary Figure S2). Subsequent uncaging data shows that nanosensors responded instantaneously to elevations of local Ca 2+ levels when imaged at a frame-rate of 6 Hz, as indicated by an increase of fluorescence from baseline. This suggests the capability of nanosensors to capture intracellular Ca 2+ signals by store-release mechanisms (see next section). It is worth noting that the measured response time of the nanosensors here is limited by the imaging frame-rate, and the temporal resolution can be significantly improved by other imaging methods 31 .
Reversibility is also key to monitoring Ca 2+ release from the intracellular store due to its rise and decay characteristics [47][48][49][50] . The ratiometric response of the opCaNS exhibited a clear reversibility profile (Fig. 2f) over repeated cycles of alternating Ca 2+ concentrations (100 and 1350 nM), which is indicative of excellent cycling capability of Collectively, the results show applicable sensor characteristics for reporting local cytoplasmic Ca 2+ concentration by exhibiting nanomolar response range, high selectivity, fast response time, and reversible response.
Sensing intracellular calcium signals. To demonstrate the capabilities of our nanosensors for intracellular applications, we examined carbachol-induced Ca 2+ store release in HeLa cells (n = 3) which acts on inositol 1,4,5-triphosphate receptors (IP 3 Rs) on the endoplasmic reticulum (ER) membrane [55][56][57] . First, nanosensors were loaded into HeLa cells using pressure controlled microinjection (Fig. 3a, inset). Although nanosensors have previously been introduced to intracellular space through endocytosis 39,58 , the microinjection technique ensures nanoparticle delivery to the cell cytosol and prevents encapsulation by endosomes 31,[59][60][61] . Each injection was about one tenth of total volume correlating to ~10 2 −10 3 nanosensors (picomolar) per cell (Supplementary Figure S4), which is an amount that provided robust signal-to-noise ratio for imaging without inflicting damage to the cells.
Time-lapse confocal imaging (Fig. 3a) revealed an initial short-lived decrease followed by a slower increase in sensor intensity ratio. This indicates an elevation of Ca 2+ levels followed by a slow decay of cytoplasmic Ca 2+ to its basal level. When the intensity ratios are converted to Ca 2+ concentration, the cytosolic elevations reached up to ~400 nM (Fig. 3b). The empirical conversion involved performing in situ cell calibration (Fig. 3c) and curve fitted by the Hill equation (Eq. 8) to determine the EC 50 after the carbachol experiment on the same cell. Briefly, in situ cell calibrations were performed by superfusing nanosensor-loaded HeLa cells with varying concentrations of Ca 2+ using calibration buffer in presence of 5 µM ionomycin, which serve as ion carriers to facilitate the equilibrium of Ca 2+ levels across the cell membrane. Although measurement of EC 50 in solution can serve as an index of sensor performance, the behavior can change when loaded in living cells 62,63 . As shown in Fig. 3c, the EC 50 of in situ cell calibration was determined to be 110 ± 42 nM. In comparison, the EC 50 of the same batch of nanosensor in solution was 54 ± 14 nM. We note that the differences between in situ and in solution calibrations in our sensor response (EC 50 ) is less or comparable to those observed in previous studies using dye  (Table S1), suggesting the validity of nanosensors for intracellular studies. The in situ calibration curve defined by optode function, α, is also provided in Supplementary Figure S5. The recovery of Ca 2+ signal to baseline in our sensor carbachol experiment suggests that reversible characteristics of the nanosensors remained intact. Also, comparable signal amplitudes were observed from Ca 2+ indicator, Fluo-4 64 (Supplementary Figure S6), suggesting that taken together, we demonstrate the use of optode-based nanosensors as a sensing platform for monitoring and quantitative imaging of intracellular ion dynamics.

Discussion
The main advantage of our sensor platform is the flexibility to tune the sensor to a given response range and selectivity for various applications by altering the sensing components. While molecular probes can improve individual aspects of indicator performance such as Kd, wavelength, and ratiometric measurements, NP-based sensors benefit from addressing these issues simultaneously. For example, sensor sensitivity can be tuned to a dynamic range of interest which can mitigate issues with indicator buffering and affinities that are either lower or higher relative to the magnitude of the Ca 2+ signal -which often leads to indicator saturation and causing erroneous estimation of the magnitude of the signal 65,66 . Also, the current sensing scheme utilizes a recognition agent that is separate from an optical signal transducer. As a result, a wide range of highly selective ionophores 22, 31, 32 applied for potentiometric biosensing 51, 67-72 can be explored for optical detection of other physiologically important ions that lack such specific probes. In addition, the current nanosensor platform permits selection of fluorophores in the visible and infrared range to permit ratiometric imaging where possible photodamaging effects induced by UV light excitation of common ratiometric indicators such as Fura-2 2, 13 and Indo-1 14 , can be minimized. While the damage imposed on cells was reduced during microinjection procedure where possible, microinjection is an invasive method which may perturb the normal physiology and delay restoration of elevated Ca 2+ levels. This can be improved by further sensor miniaturization through bottom-up fabrication approaches 22,73 and sensor delivery through facilitated uptake methods such as functionalizing nanosensors to cell penetrating peptides (CPPs) [73][74][75] in order to minimize physical perturbation of living cells. Lastly, we find that the nanosensors are dynamically fluorescent under 2P excitation, demonstrating their potential use in vivo.
Similar to the pH dependent nature of conventional dye indicators 76,77 , our nanosensor response is also subject to pH changes due to the encapsulation of pH-sensitive dyes. This is illustrated in Supplementary Figure S7, based on theoretical optode function (Eq. 3). However, the intracellular environment is governed and protected by a buffered system to maintain pH homeostasis near pH 7.2 78 , and the effect of pH is minimized. Also, the CHIII (ETH 5350) used in this work has a pKa of >10 21 , which is highly protonated under standard physiological conditions, and as such, the predominant sensor response is dictated by the changes in Ca 2+ concentrations. Currently, there are developments of other efforts to circumvent such pH dependency 52,79 .
This work expands on the continual development of optode-based nanosensors and highlights its importance as tools for monitoring physiological analytes. The current work introduces a ratiometric method for demonstrating quantitative determination of intracellular Ca 2+ dynamics using these sensors which has not been described previously. The current sensor platform combines the ease of synthesis and the flexibility to allow incorporation of different sensing components. As such, the sensor response can be tuned for specific dynamic range, selectivity, and wavelengths. In future work, detection of other physiologically important ions such as sodium, potassium, and chloride will be explored for intracellular signaling applications. Instrumentation. Fluorescent spectra for routine nanosensor calibration in solution and selectivity experiments were measured with Spectramax M3 plate reader (Molecular Devices). Sensor reversibility test and cell imaging were performed with Zeiss LSM700 inverted confocal laser scanning microscope (Carl Zeiss). Hydrodynamic diameter and zeta-potential measurements were determined using a 90 Plus particle size analyzer (Brookhaven Instruments Corporation). Ca 2+ uncaging/imaging experiments were carried out on a two-photon microscope (Bruker) equipped with two Ti:Sa lasers (SpectraPhysics) and separate beam paths/galvanometer mirrors for independent imaging and uncaging. Laser power was controlled independently for each beam path using two Pockels cells (Conoptics).
Particle Size and Zeta-Potential Measurement. Hydrodynamic diameters were measured by dynamic light scattering (DLS) based on the intensity of scattered 640 nm light at a fixed angle of 90°. All particle size measurements were taken in triplicates with sample solutions diluted in 10 mM HEPES buffer, pH 7.2 adjusted to a detector count rate of 150-450 kcps. Zeta-potentials were measured in PALS mode collected over 25 cycles each 10 runs per cycle. Size and zeta-potential measurements were analyzed from three separate batches of nanosensors each from three discrete optode stock preparations. Fluorescent Response to Calcium. Nanosensors were calibrated in solution for fluorescent response to a series of Ca 2+ EGTA buffers with controlled concentrations of free calcium ion (Ca 2+ calibration buffer). In order to determine the fraction protonated during measurements in Ca 2+ , acid and base standards were used as endpoints for fully protonated or deprotonated chromoionophore conditions, respectively.
The fluorescence intensities for CHIII (λ EX : 639 nm, λ EM : 670 nm) and R18 (λ EX : 555 nm, λ EM : 575 nm) were measured with plate reader in bottom read mode through clear-bottom 96-well plates (emission spectra of the fluorophores shown in Figure S1). The emission fluorescence intensity ratio, R, of two fluorophores (CHIII:R18) was calculated as: Emission at nm Emission at nm 670 575 (4) Traditionally, optode data is converted to a normalized value termed α, by ratios of 670 nm/575 nm defined as: where, R is the fluorescence ratio measured experimentally, and R P and R D is the ratio of fully protonated and deprotonated state of the chromoionophore, respectively. The effective Ca 2+ concentration at half-maximal sensor response (defined here as, EC 50 ), for the nanosensors was calculated from Ca 2+ concentrations and the corresponding values of alpha, α, was fitted according to the dose-response (Hill) equation: LogC p min max m in ( ) 50 where R(α) is intensity ratio of measured response, C, denotes Ca 2+ concentration at which a ratio was measured, and R min , R max , and EC 50 are parameters obtained from the fitted sigmoidal curve defined by the Hill equation.
Selectivity for Calcium over Magnesium. Nanosensor selectivity was determined by measuring the ratiometric fluorescent response (CHIII:R18) to Ca 2+ and Mg 2+ as described above, and fitting the calibration data to the equation comparable to the Nicolskii-Eisenman model with a fixed interfering ion 51,80 . The selectivity coefficient for an optical sensor is defined as: ij opt i j 50 50 where, EC i 50 and EC j 50 are the constants for the interfering ion (Mg 2+ ) and primary ion (Ca 2+ ), respectively.

Confocal Microscope Settings.
For reversibility and cell imaging experiments (see below), confocal laser scanning microscope (Zeiss LSM 700, Carl Zeiss) was used to measure fluorescence intensity ratios (CHIII:R18), as described above. Briefly, simultaneous images were acquired from dual excitation (λ EX : 555 and 639 nm) and peak intensities were recorded at two emission channels (R18 λ EM : 575 nm and CHII λ EM : 670 nm) using diffraction grating set to split the emission at 620 nm. The 40× oil-immersion objective lens (NA = 1.3) was used with the pinhole aperture setting at 2 µm which remained constant throughout the entire experiment.
Sensor Reversibility. For reversibility experiments, nanosensors were firstly bound to the surface of a neutravidin functionalized rectangular glass flowchamber (0.1× 2 mm ID) as described by Reinhard et al. 81 , and imaged on the confocal microscope. Simultaneous images from both color channels were acquired every 10 s intervals with alternating solutions of 100 or 1350 nM calcium buffers exchanged every 2 min for five exchange cycles. Images were analyzed using Matlab with identical regions of interest defined on both channels. Delivery and in situ Cell Calibration. Nanosensors were delivered to cytosolic compartment of HeLa cells using a microinjection technique (Eppendorf InjectMen 4) with 1.0 mm/0.5 mm (OD/ID) borosilicate glass tubes pulled to a tip size 0.1-0.5 µm using P-97 micropipette puller (Sutter Instrument). For in situ cell calibration, nanosensor loaded cells were initially incubated with Ca 2+ -free imaging buffer containing 5 µM ionomycin for 15 min to equilibrate Ca 2+ levels within the cytosolic environment to the extracellular space. Ca 2+ calibration buffer with incremental levels of Ca 2+ were subsequently added and fluorescence images from the confocal microscope were acquired after 10 min incubation at each calcium concentration. The values of alpha, α, and the EC 50 was determined based on the same methods used to calibrate sensors in solution. There were batch-to-batch differences in sensor response and consequently, in solution and in situ cell calibrations were routinely performed to ensure the sensors maintain responsiveness to the intracellular Ca 2+ levels.

Two-Photon
Carbachol-Induced Intracellular Calcium Measurement. Single cell determination of the intracellular calcium concentration ([Ca 2+ ] i ) was carried out in opCaNS-loaded HeLa cells. Carbachol (1 mM) was used to invoke changes in [Ca 2+ ] i by acting on IP 3 R on the endoplasmic reticulum (ER) membrane. Fluorescence confocal images were acquired at 1 s interval for 10 min after addition of carbachol to the 50 mm culture dish. Fluorescence intensities in the regions of interest (ROI) on each channel (575 nm and 670 nm) were first background subtracted by the intensities of nearby area that were void of cells. For individual regions of interest (ROI) in the cytosolic cellular region we subsequently used the corrected fluorescence intensities on both channels to calculate intensity ratio R (F 670 /F 575 ) normalized by the photobleaching profile. Changes in nanosensor fluorescence ratio were expressed as changes in Ca 2+ concentration based on in situ calibrations as described above. Conversion of fluorescence ratio to Ca 2+ concentration, C, involved empirically deriving from acquired fluorescence ratio using the dose-response (Hill) equation (Eq. 6) and plotted against time. Image processing and data analysis were performed using Matlab and Origin.