3,3 ′ -thiodipropanol as a versatile refractive index-matching mounting medium for fluorescence microscopy

: High resolution fluorescence microscopy requires optimization of the protocols for biological sample preparation. The optical and chemical characteristics of mounting media are among the things that could be modified to achieve optimal image formation. In our search for chemical substances that could perform as mounting media, 3,3 ′ -thiodipropanol (TDP) emerged as a sulfide with potentially interesting characteristics. In this work, several tests of its performance as a mounting medium for fluorescence microscopy of biological samples were performed, including the labeling of filamentous actin with fluorescent phalloidins. The refractive index dispersion curve of pH-adjusted TDP was experimentally obtained in the visible range and compared to the dispersion curves of commercial and lab-made mounting media. The effects on the fluorescence of commonly used dyes were tested by using TDP as a solvent and measuring the relative fluorescence quantum yield of the dyes. By being able to mix TDP in any concentration with water and 2,2 ′ -thiodiethanol (TDE), it was possible not only to fine-tune the refractive index of the resulting solution, but also to preserve the compatibility of TDP with the most popular and efficient fluorescent actin staining used in biological microscopy.

In a search among different simple compounds, TDP has emerged as an interesting candidate for mounting medium: it is liquid, colorless, possesses a high n when undiluted, it is miscible in water, non-toxic and its potential use as a mounting medium would represent a cheaper alternative compared to other commercial products. In this work, a thorough optical characterization of TDP was performed and the chemical was tested for its use as mounting medium for biological samples stained with common used fluorophores. The fluorescent slides were subsequently imaged with confocal microscopy, confirming its suitability for this application.
The major improvement that comes with the use of TDP as mounting medium is its compatibility with the use of toxin-based actin markers, in particular phalloidin. Differently from what happens with the related molecule TDE, phalloidin staining of microfilaments maintains its morphology and fluorescent properties after mounting in TDP. Additionally, TDP and TDE can be mixed, to finely adjust the n of the medium for high resolution fluorescent microscopy without losing the capability of staining actin with phalloidin. This type of actin decoration is by far the most popular in biomedical microscopy experiments, both for staining and counterstaining, due to the simplicity of the protocol and the quality of the outcome. For this reason, widening the application range of sulfides mounting to actin staining will add an important tool to high resolution biological fluorescent microscopy protocols.

Mounting media and immersion oil
2,2′-thiodiethanol (TDE) and 3,3′-thiodipropanol (TDP) (Sigma Aldrich No. 166782 and 205346 respectively) pH were adjusted to 7 ± 1 using sodium hydroxide. 1 mL of TDE is adjusted to the desired pH value using approx. 2 µL of 0.05M NaOH. In the case of TDP, 1 mL is adjusted using 12.5 µL of 0.25M NaOH, although the volumes could vary depending on the lot number. The terms aTDE and aTDP refer to the media prepared as above. TDE and TDP are viscous compounds and provide inconsistent readings with the electrodes normally used for pH measurement, therefore pH indicator strips (Macherey Nageland) and the pH indicator phenol red (No. P4758, Sigma Aldrich) were also used.

Refractive index
The refractive index of mounting media was measured using an Abbe refractometer Type-WY1A (Edmund Optics, Barrington, U.S.A.). The n D value refers to refractive index measured at 589 nm obtained using a white light source directed to the illumination window of the refractometer and using the compensating prism to deflect other wavelengths.

Dispersion measurements
Two different types of light sources were used to cover the visible range from 410 to 680 nm: a SCT500 supercontinuum laser (FYLA, Spain) whose spectrum ranges from 500 to 2150 nm and 2 LED sources with spectra centered in 420 nm (thereafter called LED 420 , FWHM: 18 nm) and 440 nm (LED 440 , FWHM: 19 nm). LED 420 was used for the 410 nm reading and LED 440 for the 440 and 470 nm. Seven data points in the range of 500-680 nm were measured with the laser in wavelength steps of 30 nm. The 590 nm point was replaced with 589 nm since it approximates to the sodium line conventionally used for reporting refractive indices.
Through a double-convex lens (Edmund Optics, U.S.A), the light beam from the LEDs was focused to a DMC1-03 monochromator (Optometrics, USA). Data from the manufacturer indicates a resolution of 2.16 nm for a slit width of 300 μm. For the laser, no lens was required since an optic collimator is coupled to the fiber. Another double-convex lens was placed to expand the beam so that it could fill the illumination window of the Abbe refractometer. The refractometer is provided with a compensator dial to filter out the sodium D line and deflect other wavelengths. During the measurements, the compensator was set to position 30 to avoid wavelength-dependent beam deviation. A CCD cell phone camera focusing on the eyepiece of the refractometer was mounted on a mechanical support, allowing an instantaneous view of the eyepiece image to be seen on the cell phone screen. The whole system was placed in a temperature-controlled room at 21°+/− 0.5° C.

Curve fitting for dispersion data
The dispersion curve for the refractive index in the visible range was approximated using the least squares fitting method for a four terms Cauchy's dispersion equation: where A i (i = 0,1,2,3) are the Cauchy's dispersion parameters and λ is the wavelength in nm. The statistical parameter used to define how well the measured data fit the Cauchy's dispersion equation is the adjusted coefficient of determination, 2 . adj R

Abbe number
To further compare dispersions of the liquids analyzed, Abbe numbers were calculated for D, F and C spectral lines, according to the formula [2]: Where n D, n F and n C are the refractive indices of the material at 589.3, 486.1 and 656.3 nm respectively.

Intensity along the optical axis
A 1 mM solution of the dye Rhodamine b was dissolved 10% v/v into each of the mounting media tested in this work, and mounted in slides. 70 µm Z-stacks were captured with a 63X 1.4 NA oil immersion objective, with a 1 µm separation between optical sections.

Spectra
The absorption and emission spectra were measured with a Cytation 5 spectrophotometer (Biotek, Vermont, U.S.A.): suspensions of fluorophores in TDP, TDE, VS, Dako and PBS were measured within single wells of a black-walled, clear-bottom 96-well microplate (Corning Incorporated, NY, U.S.A.). Alexa Fluor phalloidin probes were used at a concentration of 2.6 μM and DAPI at 11.4 mM. The volume used in each measurement is 100 μL.
For the imaging experiments, cells were seeded on glass coverslips. When grown to a confluency of about 80%, they were fixed in 4% formaldehyde. The cells were subsequently permeabilized with 0.1% Triton-X 100, washed in PBS, then incubated in blocking buffer composed of 10% bovine serum, 0.05% Triton X-100 in PBS for 30 minutes and incubated with the primary antibody (polyclonal anti pan-tubulin, ATN02, Cytoskeleton). After overnight incubation at 4°C, the cells were washed 3 times with PBS for 10 min and incubated with the secondary antibody (Alexa Fluor 488, A-11015, Thermo Fisher Scientific) for 1.5 h. Nucleus staining was performed by incubating cells in the presence of 3 µM DAPI for 10 min (62248, Thermo Fisher Scientific). Actin cytoskeleton was stained by incubating for 20 minutes in 130 nM Alexa Fluor phalloidin or 100 nM SiR-actin.
For fluorescent protein expression, HeLa cells were seeded on glass coverslips and transfected after 24 hours using lipofectamine 2000 (Thermo Fisher Scientific, No 11668019) with either pAC γactin EGFP or pAC ßactin mCherry. Transfection was performed according to the protocol of the manufacturer.

Mounting procedure and imaging
Fixed samples prepared for microscopy were incubated in different dilutions of aTDP to facilitate the exchange of water and mounting medium as described in [3]. The three dilutions were prepared with 10, 25 or 50% of aTDP, 25% PBS and water. Incubation time for each solution was 10 minutes. Two final incubations were performed in aTDP before the final mounting in aTDP. Mounting in VectaShield and Prolong Diamond was performed according to the respective technical data sheet.

Cell size analysis
Twenty micrographs of HEK293 cells stained with fluorescent phalloidins and DAPI were included in the analysis, 4 mounted in PLD, 7 mounted in aTDP and 9 mounted in PBS. Independent binarization of the 2 channels was performed estimating thresholds by Otsu's algorithm [4]. Once the binarized images were obtained, a morphological closure operation using monostructural elements was applied [4]. The number of cells in a micrograph was determined by the number of nuclei; when apparently joint nuclei were detected, a k-means clustering analysis was used to solve the nuclei division. The regional average size of the cells was calculated dividing the surface filled with actin by the number of nuclei counted in the region. The whole image analysis was performed using Matlab (MathWorks).

Mixtures of TDP-TDE
Three mixtures containing aTDE and aTDP at different concentrations were also tested as mounting media: solution M1 contains 25% aTDP, solution M2 contains 50% and solution M3 is made of 75% aTDP. For the mounting process, the same procedure as in the case of aTDP was followed: cells were incubated in solutions of 10, 25 and 50% concentration of mixture in PBS before mounting.

Refractive index
The refractive indices of TDP and TDE at increasing concentrations in water were measured. The experimental set of data obtained corresponds to n D and is shown in Fig. 1, where the linear relation found between refractive index and concentration in water is reported. The refrac 1.508 for TDP

Dispersio
For the meas refractometer used (LED or measurements prisms dispers The avera described by results when other hand, t comparable o  For immersion oil, five n reported in the technical datasheet of the product (Immersionsoele, 2004 [8]) were compared to the values obtained with our system, at corresponding wavelengths: the average difference of the two series of n is 0.00020, with the maximum variation being 0.00044, which shows good agreement between the two sets of data (see Data File 1).
For the commercial medium VectaShield, the only refractive index available from the literature was the one corresponding to 590 nm [9]: the value that can be interpolated from our model (1.449) corresponds to the published value (1.45).
TDE was presented as mounting medium by Staudt et al. and some points of its dispersion curve were reported graphically in the original paper [3]. In our work, we have analyzed the dispersion behavior in a larger spectral region and, as for the other media, we have calculated the mathematical approximation that allows to interpolate the refractive index in any given point of the visible spectrum.
In the case of the new medium TDP, the only n information previously available, to our knowledge, was the value given by the producer, n D 20 = 1.51; the value we measured is 1.5056, again in good agreement.

Abbe number
The constringence or Abbe number (Eq. (2)) of the liquids analyzed, which measures the variation of the refractive index with the wavelength, shows that the substance with the highest V D among those analyzed in this work is VectaShield (57.35), while aTDP (46.17) and aTDE (45.88) show dispersion similar to that of immersion oil (46.07) ( Table 2).

Effects of n mismatch on confocal image stacks
We have measured the fluorescent intensity along the optical axis in slides where the dye Rhodamine b was diluted and mounted in the different mounting media used. As expected, the mounting media with higher n are those that better perform in this test (Fig. 3). TDP, whi more pronoun not comparab most rapid lo fluorescent in the n mismatc

Spectra
For the three solvents. The in Fig. 4    as mounting m rinking could h der to estimate to cells mount labeled with flu nted in PLD re nking is thus ob medium doesn tograms of the ctin. Two ng protein uorescent mounting . media for happen if e whether ted in the uorescent esulted in bservable n't induce e regional

Mixtures
The three solu medium for phalloidin. Im of labeled act observed and higher fluores in M1 (Fig. 1 consistent wit dense stress f M1 and M2 (F  since they differ in cell size and morphology, allowing the imaging and test for performance of our mounting medium with different cytoskeletal architectures. The analysis of samples confirmed that TDP could be used as mounting medium, since the conventional distribution of DAPI, tubulin and filamentous actin staining in epithelial, cervix cancer and neuroblastoma cells was observed. The lack of morphological distortions was also confirmed by our cell size analysis. Additionally, transfected cells mounted in TDP were also imaged, confirming that TDP is compatible also with fluorescent proteins.
Knowing the solvatochromism of probes dissolved in the mounting media is important to determine excitation and detection wavelengths of the fluorescence microscope set up and establish if the excitation sources and filters, as well as the detection filters, are appropriate and efficient for the fluorescence of each experiment. The excitation wavelengths used in this work were those recommended by the fluorophore providers and each fluorophore was imaged using the same excitation and emission parameters in all experimental conditions, i.e. when mounted in PLD, aTDE or aTDP. The bathochromism observed in the spectra of the dyes when TDP is the solvent is subtle, therefore no changes in filters or detectors were necessary. Fluorophore-solvent interactions also have an effect over the fluorophore quantum yield of the dyes used: for Alexa Fluor 647 and DAPI, the relative quantum yield indicates stronger fluorescence brightness when embedded in TDP relative to PBS. In case of Alexa Fluor 488 the brightness is lower, similarly to what happens with Oregon Green 488 as reported in [3]. This might suggest that similar non-covalent molecular interactions take place between "fluorescein-based dyes" and both sulfides media.
A key feature of TDP compared to most commercial media is its refractive index, which by being close to that of immersion oil reduces the refraction caused by the n mismatch and the consequent aberrations in Z. It has been demonstrated that the loss of intensity when imaging deeper into a fluorescent medium, in a confocal system, can be solely explained by aberrations [1]. Spherical aberrations induced by a n mismatch provoke a larger spread of the focus; the spreading of the illumination and detection PSF caused by aberrations affects resolution and intensity [13]. When imaging, the object space, made of the immersion liquid, glass coverslip and the mounting medium-embedded sample, is necessarily heterogeneous. However, differences in the dispersion behavior of the components could led to additional aberrations (i.e. chromatic). Therefore, the selection of the mounting and immersion fluids could be based on the assumption that, for equal conditions of n D , similar dispersion on the two sides of the coverslip is most desirable, especially when detecting several fluorophores in a Z-stack, as is the case when studying colocalization in a volume [14]. As the homogeneity of n improves the axial resolution, which is critical when imaging deep into thick samples, optical sectioning techniques like confocal, lightsheet and multiphoton microscopy are already benefitting from the use of refractive index matching liquids that enable even deeper imaging [15]. Noteworthy, some index matching liquids were also shown not to interfere with the generation of second harmonics for the stain-free imaging of collagen fibers in the forward direction [16].
Another factor that affects imaging inside 3D samples is unwanted lateral scattering caused by heterogeneities in biological tissues. Clarification, a group of techniques that achieves the downsizing of light deviation by means of chemical reduction of tissue scattering [17], makes extensive use of media with high refractive index, including TDE [18]. It is safe to predict that TDP, due to the chemical similarity to TDE and high n, could be useful for volume fluorescence microscopy, whether this implies clarification or not.
The physico-chemical behavior of mixtures is often unpredictable; for instance, the refractive index of binary mixtures is not always accurately described by mixing rules [19]. For this reason, the effect of mixing aTDE with aTDP on the refractive index of the solution was tested, as well as the performance of the mixtures as mounting media in biological slides. We found that the refractive indices of the mixtures linearly increased with the addition of aTDE making it possible to tune the n from 1.505 to 1.5123. Also, by mixing aTDP and aTDE in three different proportions we observed dramatic differences on the confocal images, which were tightly dependent on the aTDP:aTDE ratio employed for mounting the slides: on the one hand, the destabilizing effect of TDE on the labeling of actin with fluorescent phalloidins was confirmed; on the other hand, staining loss was hindered by increasing the content of TDP. In other words, by playing with the amount of each medium, we were able to increase the n of the mounting medium based on TDP, besides diminishing the destabilizing effect of TDE on the toxin-based actin staining.
Index matching liquids are necessary in microscopy either for imaging into tissue or to achieve the highest possible resolution when required; TDP represents a new, flexible option for these microscopy applications. Additional optical uses of TDP may include optofluidics [20], which combines liquids and photonics, or the related area of tunable optics [21]. Tunable n liquids can be also advantageously used in particle image velocimetry, where tunability associated to miscibility in water make TDP a potentially good solvent for both dynamic light scattering, where differences in n must be minimized, and diffusing wave spectroscopy, where large differences in n are pursued [22,23].