Monitoring correlates of SARS-CoV-2 infection in cell culture using two-photon microscopy and a novel fluorescent calcium-sensitive dye

The organism-wide effects of viral infection SARS-CoV-2 are well studied, but little is known about the dynamics of how the infection spreads in time among or within cells due to the scarcity of suitable high-resolution experimental systems. Two-photon (2P) imaging combined with a proper subcellular staining technique has been an effective tool for studying mechanisms at such resolutions and organelle levels. Herein, we report the development of a novel calcium sensor molecule along with a 2P-technique for identifying imaging patterns associated with cellular correlates of infection damage within the cells. The method works as a cell viability assay and also provides valuable information on how the calcium level and intracellular distribution are perturbed by the virus. Moreover, it allows the quantitative analysis of infection dynamics. This novel approach facilitates the study of the infection progression and the quantification of the effects caused by viral variants and viral load.


126
The calcium-sensitive fluorescent dye readily enters the cell The spectroscopic characterization of BEEF-CP involved one-photon (1P) and 2P fluorescent measurements and the 137 quantum yield (QY) determination in the presence and absence of Ca 2+ ions. The measured fluorescence intensity at 138 pH 7.2 showed a single-step enhancement upon Ca 2+ addition in accordance with the formation of a single 139 monometallic complex. The fluorescence enhancement was found to be around threefold in the biologically relevant 140 Ca 2+ concentration range, from 10 nM to 1 M ( Supplementary Fig. S15). The dissociation constant of the Ca 2+ 141 complex (KD = 175.6 nM) is comparable with previously synthesized molecular probes for in vivo calcium level 142 determination. 35,36 The background fluorescence of BEEF-CP is not zero in the absence of Ca 2+ , which provides 143 appropriate contrast to visualize even the healthy cells with low level of Ca 2+ ion during the cell viability assay. 144 Furthermore, the high-slope region of the Ca 2+ -dependent fluorescent curve is adequately positioned to distinguish 145 between healthy and malfunctioning cells. 146 5 The further applicability of BEEF-CP in cellular systems was tested by flow cytometry measurements on HEK-293 147 cell line. Stained cells showed an increased mean fluorescent intensity by around one order of magnitude compared 148 to the control level ( Supplementary Fig. S12). In a parallel experiment, the stained HEK cells were treated with 149 ethylene glycol-bis ( -aminoethyl ether)-N,N,N N -tetraacetic acid (EGTA) prior to the analysis, which resulted in a 150 somewhat lower mean fluorescent intensity. In contrast, cells stained in the presence of ionomycin using otherwise 151 identical procedure exhibit substantially higher mean fluorescent intensity. The ionomycin releases exclusively the 152 intracellular Ca 2+ storage, which proves that the dye is internalized by the cells as it responds to the cytosolic Ca 2+ 153 levels. 154 Subcellular dye localization is characteristic and reproducibly different in uninfected and infected 155 cell cultures.

156
To perform high-resolution quantitative analysis of the level of SARS-CoV-2 infection we used 2P imaging. This 157 imaging technique has a high, subcellular spatial resolution (~350 nm), and virtually no out-of-focus signal from non-158 studied cells or compartments, above or below of the imaging plane, which ensures that the detected signal originates 159 only from the cells of interest. Samples were prepared in a 96-well bio-assay plate with a clear, flat bottom for good 160 optical access. One to three high-resolution, 2P, full-161 FemtoViruScope (Femtonics Ltd, Budapest) at 700 nm wavelength (see Supplementary Fig. S2 for wavelength 162 selection) from one or more representative area within each sample, see Methods for details. 163 Acquired images were analyzed with two complementary methods (manual and automatic). During manual 164 evaluation, the cells were sorted according to their morphology ( Fig. 1b-d), while using the automatic analysis, based 165 on statistical methods, consisted of the classification of the full images based on a set of image parameters (Fig. 2). 166 Both methods take the different features of the acquired raw images into account for the analysis, and for both, we 167 investigated the fine details of the images (at cellular or single-pixel level) for designing a quantitative measure 168 correlating with the initial challenge infection level. Two variants of concern of SARS-CoV-2, namely D614G and 169 B.1.1.7 were studied. These caused pathophysiologically and clinically distinguishable disease patterns in the clinic, 170 so the putative infection-quantitation and infection-assay properties of BEEF-CP would be tested as the distinguishing 171 features of cellular Ca 2+ related effects of these two variants. Different cellular features such as the state of infection 172 and the apoptotic-like state were visible and distinguishable on the images of Vero E6 cells infected with the two 173 SARS-CoV-2 variants (Fig. 1b). The investigations were done in a phase when still living cells (both healthy and 174 infected) adhered to the plate surface. Cell compartments, such as cell membrane, nucleus, and nucleolus were visible 175 and distinguishable in the samples. With increasing virus titer, the mean fluorescence of infected cells increased 176 significantly as this caused more Ca 2+ to enter the cytoplasm. When the infecting viral titer increased even further, the 177 morphology of cells changed, and dead cells detached from the wells plate; thus, the number of these cells could be 178 estimated Due to the subcellular resolution of this cell viability 179 assay, we could detect the formation of several syncytial formations, as giant multinucleated cells in the most infected 180 cultures, which agrees with the earlier observation that SARS-CoV-2 virus causes this type of cellular fusion 37  Manual evaluation 208 During the manual evaluation cells where manually localized on the recorded images, then the individual cells were 209 visually categorized into five groups, representing different stages of the infection, see Fig. 1c and Methods for more 210 details on cell categories. The individual images from different wells were subjected to a single-blind evaluation by 211 human operator. The cells were classified into four categories (Fig. 1c), while the fifth category, cell death, was 212 determined by the absence of living cells, estimated from the surface area that was not covered with cells, as shown 213 in Fig. 1c. The ratio of the cells belonging to each category was calculated for the different virus variants, virus titer 214 and dye concentration showing an increasing number of cells in the healthy or initial infection categories with the 215 virus titer decreasing (Fig. 1d). The 50th percentile values were calculated for each virus titer by averaging the result 216 from all wells infected with the given titer. For this calculation, cells were sorted by the severity of their infection 217 from healthy to dead cells, and the position of the median was calculated as the ratio of all the cells. The 50th percentile 218 values also show a clear trend with increasing viral load. Furthermore, there is a clear jump in the severity of the 219 infection in the cell culture between TCID50 virus titer 10 -4 and 10 -3 mL -1 . 220 Automatic analysis 221 An ImageJ 38 macro was created to identify particle objects based on an intensity threshold (Fig. 2a), and to extract 222 image data from the fluorescence microscopy images in the form of a few numerical parameters. The ImageJ macro 223 is available in the Supplementary Information. Seven parameters characterizing the detected particles were defined in 224 each image, more specifically i: Relative signal area, ii: Image mean intensity, iii: Mean of the threshold area, iv: 225 Maximum particle intensity, v: Average particle size, vi: Particle percentage area, and vii: Particle mean intensity. 226 8 The input images correspond to experimental points that can be described by two infection-related variables, namely 227 virus titer and virus variant and by two measurement-related variables, namely BEEF-CP sensor concentration and 228 photomultiplier tube (PMT) voltage. Analysis of the image parameters as a function of the infection-related and 229 measurement-related variables can yield information about the efficiency and robustness of the automatic image 230 analysis, respectively. Fig. 2b c, 2e f show that the image parameters vary according to different acquisition settings, 231 i.e., virus concentration, virus type, dye concentration and PMT voltage. Importantly, certain image parameters such 232 as Relative (Fig. 2d) can be a good indicator of the viral infection, as it shows increasing parameter value 233 with increasing virus titer. However, images of the two variants did not show clear separation by any parameter values 234 without eliminating the variation caused by the measurement-related variables. 235 Other prevalent Ca 2+ sensors, such as OGB-1 AM (Oregon Green 488 BAPTA-AM), are used at a high concentration, 236 around 1 mM (ranging 0.9-1.4 mM 39 42 ). We used BEEF-CP at lower concentrations, at 0.1 mM and 0.01 mM. Both 237 concentrations were suitable to differentiate between little or no infection and a high level of infection using automatic 238 image analysis. Moreover, differentiation of the variants responsible for the infection became possible when the 239 images obtained with higher dye concentration (0.1 mM) were analyzed in those cases when the well was highly 240 infected (TCID50 > 10 -3 mL -1 ) (Fig. 2g). Therefore, for further analysis, only those images were considered that were 241 acquired from samples stained with 0.1 mM dye concentration. 242 A cluster analysis that considers all seven parameters was run on all the images acquired using 0.1 mM dye 243 concentration. Two-dimensional visualization of the seven-dimensional clustering (Fig. 2h) (Fig. 2g). The results suggest that the fluorescent intensity of infected cells related 271 to the image mean intensity is delimited by the dye concentration at 0.01 mM and by the Ca 2+ concentration at 272 0.1 mM dye load. Second, the cluster analysis of the data could distinguish three separate groups. Among these was 273 the non-infected group, the images falling into this cluster showed similar characteristics to the control measurements. 274 However, more interestingly, the remaining images were almost perfectly separated by the virus variant used for the 275 infection (Fig. 2h). These results demonstrate that this method is capable of quantifying the infection level and can 276 also differentiate the two applied variants, D614G and B.1.1.7. 277 Discussion 278 This study presents a general concept of quantitative calcium-based viral cellular harm assay, here applied on SARS-279 CoV-2 viral variants. For quantitative analysis of SARS-CoV-2 infection, we chose to monitor the Ca 2+ -dependent 280 cellular mechanism previously described in the literature as it strongly correlates to the infection rate. For monitoring 281 the intrinsic concentration of Ca 2+ , a novel cell membrane permeable internalized, two-photon active Ca 2+ selective 282 sensor molecule was designed and synthesized, which allows the measurement of both extra and intracellular Ca 2+ 283 concentration, without chemical compromises. BEEF-CP as a Ca 2+ -selective sensor exhibits several advantages for 284 practical applicability, such as i) appropriate cell internalization 44 ; ii) the effective concentration of the dye is 10-100 285 times lower (0.01 0.1 mM) compared to other widely used compounds (>1 mM); and iii) is free from toxic leaving 286 groups hence infection processes are not affected by the chelation of Ca 2+ using the novel dye, unlike in the case of 287 working concentrations for BAPTA-AM; iv) appropriate spectrophysical characteristics and sufficient pH 288 insensitivity around the physiological pH. 289 ViruScope has provided high resolution and extreme sensitivity by fast 2P fluorescence imaging. The selection of the 290 2P imaging method was indicated by its fine z-sectioning ability compared to single-photon imaging, as shown in 291 Supplementary Fig. S1. As we used the fluorescent intensity profile of the images acquired for quantitative analysis, 292 it was imperative to measure only a single cellular layer without contamination from the out-of-focus cells or 293 compartments. This was achieved by the means of 2P imaging. Furthermore, it should be noted that the proposed 294 imaging method can be directly applied to tissue slices or organoids, opening new ways to test the effects of SARS-295 CoV-2 infections in more complex systems as well. The high-resolution detection of the Ca 2+ level within the infected 296 cells requires a sensor molecule with appropriate 2P sensitivity and adequate cell internalization. The developed 297 sensor, BEEF-CP meets these criteria owing to its diethylfluorescein fluorophore. The fluorescent dye concentration 298 influences the information content of the images; 0.1 mM unlike 0.01 mM proved to be sufficient to determine 299 not only the level of infection but also the virus variant. Importantly, neither applied dye concentrations induced cell 300 death in the control wells. The incubation time (60 minutes) was sufficiently short for rapid imaging of the fast 301 dynamics of the infection. 302 Two different methods, with comparable results, have been used to evaluate imaging data. The first method allowed, In conclusion, we propose a new calcium-sensitive dye for correlates of SARS-CoV-2 infection determination, 312 quantification and monitoring in two-and three-dimensional cell cultures using 2P microscopy. This molecule is at 313 least one order of magnitude more sensitive than previously applied calcium dyes. Our BSL-3 2P microscopy 314 measurements have also established that the applied concentrations of the dye do not interfere with viral replication 315 and viral fusion events. We postulate that this method thus ensures proper monitoring of the full viral entry spectrum 316 of events. At the same time, it also enabled us to distinguish intracellular details of cell damage, such as vacuole and 317 apoptotic body formation. Using clustering analysis, we could use the 2P microscopy calcium fluorescence images 318 for the distinction of two different viral variants in cell cultures. Using this 2P microscopy method, we could also 319 establish correlates of infection related to the initial viral multiplicity of infection numbers. Further research may 320 utilize this technique to study and compare the effects of antiviral drug candidates on the infection in cell cultures. 321

322
Two-photon (2P) imaging 323 All experiments were performed on a FemtoViruScope (Femtonics Ltd., Budapest, Hungary). Laser pulses were 324 generated by a Mai Tai HP laser (SpectraPhysics, Santa Clara, CA), the laser wavelength was tuned between 750 and 325 850 nm, depending on the experiment. Laser intensity was controlled using a Pockels-cell (PC) electro-optical 326 modulator (model 350 80 LA; Conoptics) after beam expansion (1:2, Thorlabs). For excitation and signal collection, 327 a XLUMPLFLN20XW lens (Olympus, 20 , NA 1.0) was used, separated using a dichroic mirror (700dcxru, Chroma 328 Technology) before the two-channel detector unit, which was sitting on the objective arm (travelling detector system) 329 as described in detail elsewhere 16,45 . Emitted light entering to the detectors were filtered with emission filters, 330 ET520/60 for the green and ET605/70 for the red channel (Chroma Technology, Bellow Falls,VT). Fluorescence 331 signal was collected to GaAsP photomultiplier tubes (PMT) fixed on the objective arm (H7422P-40-MOD, 332 Hamamatsu). A set galvanometric mirrors was used to acquire full field images. All acquired images were recorded 333 at 1000 1000 pixels resolution, spatial resolution was 0.36 . Single pixel integration time 334 resulting a 6.7 s image acquisition time. The PMT maximal amplification voltage (~300 V) and the imaging laser 335 power (~50 mW) were kept constant through all measurement and for all wavelength for a more reliable comparison 336 between the acquired fluorescent images. Dynamic range of the PMT is 16bit, w 337 most of the range but have a maximum of 50% of the pixels dark on the dimmest and no saturated pixel on the brightest 338 image. 339 We acquired a series of images using different stimulation wavelengths to calibrate the optimal laser wavelength. To 340 compensate for the change of the output laser power experienced when tuning the output of the pulse laser, the PC 341 amplification was aligned individual at every different wavelength tested, to keep to laser intensity reaching the sample 342 constant at ~50mW power. The amplification levels were calibrated prior to the acquisition with an intensity meter 343 placed under the objective. Note, that the transmission of the objective is also dependent on the laser intensity; i.e. 344 measuring the intensity under the objective also eliminates this error, besides the laser output modulation. Furthermore, 345 not just the transmission but the material dispersion of the objective is wavelength dependent. Temporarily elongated 346 laser impulses can also affect the 2P efficacy. To minimize this effect, we used glass stand chamber well plate, where 347 the effect of the material dispersion is minimal. The maximum calculated elongation of the pulse, caused by the 348 material dispersion, when tuning the laser from 750 to 850 nm is 19 fs which. Due to the quadratic nature of the 2P 349 power on the excitation, and because the full pulse length is still below 200 fs at 850 nm, pulse elongation has less 350 than 1% effect on the stimulation quantum efficacy. This is much lower than the effects we demonstrated and falls 351 within its noise level, therefore we took no count for this effect during the analysis of the wavelength optimization. 352 One to three high-resolution image is acquired from every well. The stage was moved between the plates automatically, 353 with the know horizontal and vertical distances between the wells; however, the field of view and the focus were 354 selected manually. Images were saved as multichannel tiffs (for the dual wavelength detection). Image acquisition 355 was performed with a custom Matlab code. 356 Virus categories during manual analysis 357 Imaged cells were manually localized on the recorded images, and then individual cells were visually categorized into 358 five groups according to the following criteria: 359 a) Healthy: Healthy cells display a pale, low level fluorescence, due to low Ca 2+ level and their shape is regular. 360 The distribution of cells is homogenous, their arrangement on the plate is normal.

Automatic analysis 376
An ImageJ 38 macro was created to load PMT normalized images sequentially, run a despeckle filter to remove noise, 377 run a threshold command to select the signal, measure the mean, max and relative area of the selection, measure the 378 mean intensity of the whole image (including empty space) and finally create a selection based on the threshold. To 379 quantitate particle texture and roughness, a copy was made, and a gaussian blur was applied. The converted image set 380 was then subtracted from the original to enhance any variations in the image. Threshold areas in these images were 381 then selected as before to create a second mask which is again analyzed. Seven parameters characterizing the detected 382 particles on the images were defined for each image, more specifically i: Relative signal area, ii: Image mean intensity, 383 iii: Mean of the threshold area, iv: Maximum particle intensity, v: Average particle size, vi: Particle percentage Area, 384 and vii: Particle mean intensity. T-distributed stochastic neighbor embedding (t-SNE) 2D plot was obtained in 385 MATLAB using the built-in tsne function with default random number generation, a perplexity value of 10 and 386 exaggeration value of 50 on all seven parameters of images acquired with 0.1 mM dye concentration. Gaussian mixture 387 model and k-means clustering analyses were performed in the seven-dimensional space determined by the seven image 388 parameters. The number of clusters was set to 3. For the gaussian mixture fitting, the regularization parameter value 389 was set to 1. 390 Data Availability

391
The authors declare that all the data supporting the findings of this study are available within the paper and 392    control triplicate using 10 -1 TCID50 and a negative uninfected control triplicate were also plated.
2 for forty-eight hours to allow for the cytopathogenic effects of viral replication to complete.
For imaging, the fluorescent dye solution was added to each well in 10 microliters of volume in a solution of 10 or 100 times diluted original stock (1 mg of dye dissolved in 1 mL of ethanol, which was diluted with 9 mL of distilled water). The supernatant replaced after 60 minutes incubation prior to the imaging.

Synthetic Notes
General synthetic methodologies and characterization Coupling constants are given with an accuracy of 0.1 Hz. Splitting patterns are abbreviated as follows: singlet (s), doublet (d), triplet (t), quartet (q), multiplet (m), broad singlet (bs) and combinations thereof.
Assignment of spectra was aided by 2D NMR spectroscopy ( 1 H-13 C HSQC and HMBC).
HRMS and MS-MS analyses were performed on a Thermo Velos Pro Orbitrap Elite (Thermo Fisher Scientific) system. The ionization method was ESI operated in positive (or negative) ion mode. The protonated (or deprotonated) molecular ion peaks were fragmented by collision-induced dissociation (CID) at a normalized collision energy of 40-65%. For the CID experiment helium was used as the collision gas. The samples were dissolved in methanol. Data acquisition and analysis were accomplished with Xcalibur software version 4.0 (Thermo Fisher Scientific).

S23
The preparation of the calcium dye was carried out in a five-step synthetic route using 4-ethyl resorcinol, 1,2,4-benzenetricarboxylic anhydride and tetraethyl BAPTA23 in high overall yield.

Synthesis of ketone 5
4-Ethylresorcinol (2, 1.38 g, 10.0 mmol, 1.0 equiv.) and 1,2,4-benzenetricarboxylic anhydride (1, 2.11 g, 11.0 mmol, 1.1 equiv.) were dissolved in 1,2-dichloroethane (55.0 mL). AlCl3 (7.99 g, 60 mmol, 6.0 equiv.) was added, the mixture was stirred for 72 hours at room temperature. The solvent was evaporated, 100 mL of EtOAc, then 50 mL of aqueous HCl (4M) were added then the layers were separated. The aqueous phase was washed with EtOAc (3 100 mL), the combined organic layers was washed with aqueous HCl (1M, 50 mL), brine (50 mL), dried over MgSO4. The solvent was evaporated under reduced pressure. The residue (3.51 g) was purified by preparative HPLC (water acetonitrile 0.1% TFA, using the gradient method). After purification, the fractions were lyophilized. The target isomer (5, 1.49 g, yield 45%) and its regioisomer (5_co 1.02 g, yield 31%) were isolated as a yellow powder. Spectroscopic data of 5_co   0.8 mg of BEEF-CP was dissolved in DMSO to obtain a stock solution. pH Buffers for the range of 3-9 based on NaOAc/AcOH (pH [3][4][5] or Na-HEPES/HEPES (pH 6-9) were prepared in the concentration of 5 mM. The ionic strength was set to 50 mM by the addition of KCl and either 1.5 mM EGTA or 1.5 mM CaCl2 was dissolved in the solution. The fluorescent spectra of BEEF-CP has been recorded using a Shimadzu 1900 spectrophotometer in each Ca 2+ -containing and Ca 2+ -free pH buffer. The used excitation wavelength was 504 nm, both the excitation and emission slits were set to 5 nm. The spectra were recorded at a scan speed of 2000 nm/min by recording datapoints every in 0.5. nm. The instrument was used in low sensitivity mode. The emission intensity values were plotted against for pH in both the Ca 2+ -containing and Ca 2+ -free solutions shown in Fig. S16.

Two-photon cross-section measurements
The two-photon absorption cross section (TPCS) was determined with the two-photon excited fluorescence (TPEF) method 5 . The measurements were performed using an inverted two-photon microscope (FemtoSmart2D, Femtonics), equipped with a XLUMPFLN20XW Olympus objective (numerical aperture; NA = 1.0) and a tunable high-power Ti:Sapphire laser (Coherent Chameleon Discovery COM5, wavelength of the excitation light is between 700 nm and 1040 nm). The incident light source was focused into at capillary filled with either the sample or the reference solution (Rhodamine 6G in MeOH 6 ) and integrated fluorescence emission was detected in a wavelength window from 475 to 575 nm (green channel of the microscope). The power of the laser source was kept constant at 15 mW. TPCS at each excitation wavelength was calculated according to the following equation: where A is the mean TPEF emission intensity, c is the concentration of the dye, n is the refractive index of the solvent measured at the sodium D-line; a is a ratio derived from one-photon emission measurements calculated as the integral of one-photon emission spectrum from 475 to 575 nm divided by the total one-photon emission spectrum integral, ref is reference, sam is sample.