During the preclinical drug development phase, compound libraries must be thoroughly tested. For this purpose, cells cultured as monolayers on flat surfaces are the gold-standard model used to evaluate the effectiveness and safety of the therapeutics in development.1 Still, this type of cell culture model has been associated with lack of correspondence between the results obtained in vitro and those obtained in in vivo or in clinical trials.12.–3 Therefore, there has been an increasing need to improve the reliability of the in vitro methodologies in the high-throughput screening of new drugs. To address this issue, three-dimensional (3-D) cell culture models have been developed. In the 1970s, 3-D cellular aggregates—spheroids—started to be produced with the aim of mimicking the features of solid tumors, such as cellular organization, cell–cell and cell–extracellular matrix (ECM) interactions, gene expression profile, and drug resistance mechanisms (reviewed in detail in Refs. 4 and 5).
Despite the advantages of 3-D cell cultures, performing experiments on spheroids introduces new challenges since the methodologies and the equipment currently used for therapeutics screening are only optimized for cells cultured as monolayers. Fluorescence microscopy, such as light-sheet-based fluorescence microscopy, two-photon microscopy, multiphoton microscopy, and single (or selective) plane illumination microscopy have demonstrated an excellent performance in the in vitro analysis of therapeutics on spheroids.4,6,7 Still, most laboratories are equipped with confocal laser scanning microscopy (CLSM) apparatus, which has been used to determine spheroids’ size,8 morphology,9 internal cellular organization,10,11 and expression of proteins (e.g., -tubulin and E-Cadherins),10,12 as well as to evaluate the penetration and efficacy of therapeutics (e.g., drugs and delivery systems) in the spheroid.8,1314.–15 However, taking into account that spheroids are thick opaque samples with diameters between few hundred micrometers and almost 1 mm, the observation of whole spheroids by CLSM is hindered by the low penetration depth of this equipment (usually ).4,6,16 This handicap is usually surpassed by slicing the spheroids into thin sections (3 to ).17,18 However, the sectioning of spheroids is a laborious and time-consuming process that depends on organic solvents and can induce the disruption of the initial morphology of the spheroid (due to the introduction of structural artifacts).4,19,20
Another approach that has been employed to facilitate the imaging of 3-D thick samples is the use of optical clearing methods. These clearing techniques have been applied in the deep imaging of embryos and mouse tissues (e.g., brain, skin, and skeletal muscle)2122.23.24.–25 and more recently started to be applied in the imaging of spheroids.26,27 In general, the optical clearing methods reduce the light scattering induced by the cells and make the samples more transparent, thus enhancing light penetration and, consequently, the images quality.6
The clearing methods developed so far include the 3DISCO, ACT-PRESTO, BABB, , , CLARITY, CUBIC, FocusClear, FRUIT, iDISCO, PACT/PARS, RTF, Scale, Sca/eA2, Sca/eS, Sca/eU2, SeeDB, SeeDB2, TDE, uDISCO, among others (reviewed by Azaripour et al.,25 Tainaka et al.,28 Richardson and Lichtman,6 and Seo et al.29). In 2013, was described for the first time by Kuwajima et al.22 Compared to the other methods, the is quicker, and it does not use detergents or organic solvents. The protocol involves the immersion of the samples in aqueous solutions with increasing concentrations of formamide and polyethylene glycol (PEG).22 These clearing solutions will promote the replacement of the water inside the cells by the aqueous solutions of formamide and PEG until an equilibrium is reached, while maintaining the sample hydrated, reducing the overall refractive index (RI) of the sample and improving samples’ transparency.6,22,29,30 On the other side, PEG is used to maintain the integrity and stability of the fluorescently labeled elements (e.g., proteins) and, thus, avoids the fluorescence quenching prompted by the formamide.6,22,29,30
So far, PEG with a molecular weight (MW) of 8000 Da has been used in the method.22,26 Still, some studies demonstrated that PEGs with other MW can also be used to stabilize proteins.31,32 In this way, disclosing the optimal PEG MW for spheroids clearing by may improve their analysis through CLSM and their potential for high-throughput screening of therapeutics. Herein, the influence of the PEG MW (4000, 8000, and 10,000 Da) used in the method in the imaging of propidium iodide (PI)-stained spheroids was investigated. For this purpose, PI was selected since it is commonly applied for spheroids analysis,10,33,34 and it was also used in other works, where clearing methods were investigated.3536.–37 After imaging whole noncleared and cleared spheroids through optical and CLSM, images were analyzed to characterize the effect of the method variations on spheroids’ morphology, transparency, fluorescence, imaging penetration depth, and cross-section imaging depth.
Materials and Methods
Normal human dermal fibroblasts (HFIB) were bought from PromoCell (Labclinics, S.A., Barcelona, Spain). Cell imaging plates were acquired from Ibidi GmbH (Ibidi, Munich, Germany). Agarose was purchased from Grisp (Porto, Portugal). Cell culture plates, T-flasks, and cell culture consumables were obtained from ThermoFisher Scientific (Porto, Portugal). Dulbecco’s modified Eagle’s medium F12 (DMEM-F12), formamide (), paraformaldehyde (PFA), phosphate-buffered saline (PBS), trypsin, ethylenediaminetetraacetate (EDTA), PEG 4000, 8000, and 10,000 Da were purchased from Sigma-Aldrich (Sintra, Portugal). Fetal bovine serum (FBS) was obtained from Biochrom AG (Berlin, Germany). PI was purchased from Invitrogen (Carlsbad, California).
Cell line maintenance and HFIB spheroids production by micromolding
HFIB were cultured in DMEM-F12, with 10% (v/v) FBS and 1% streptomycin and gentamycin, inside an incubator with a humidified atmosphere at 37°C and 5% .10 After cells attained confluence, they were recovered using 0.25% trypsin (1:250) and EDTA 0.1% (w/v).
Spheroids were fabricated using agarose structures with spherical microwells to guide the cells self-assembly.38 The agarose structures were produced by placing 2% (w/v) agarose solution on 3-D Petri Dish® templates (Microtissues Inc., Providence, Rhode Island).13 Before use, the agarose structures were placed in 12-well cell culture plates and sterilized by UV radiation. HFIB cells were then seeded using per agarose structure, prompting the formation of 81 spheroids. Spheroids were maintained in culture with DMEM-F12 [10% (v/v) FBS and 1% streptomycin and gentamycin] at 37°C in a humidified atmosphere with 5% .
Whole HFIB spheroids fixation and staining with PI
Spheroids used in the experiments grew for 6 days. After this period, whole spheroids were recovered and then subjected to a chemical fixation process that encompasses spheroids incubation with PFA 4% (w/v) during 24 h at 4°C.10 The PFA solution was freshly prepared to minimize its autofluorescence.39 After fixation, the spheroids were washed three times with PBS and then labeled with 1 mL of PI ( in ), as previously described.10,11 After 24 h of spheroids incubation with the fluorescent probe on a plate shaker at 100 rotations per minute (RPM), spheroids were washed three times with PBS to remove the excess of PI.
Whole spheroids clearing using the method
The clearing method variations (, , and that use PEG with an MW of 4000, 8000, and 10,000 Da, respectively) were performed accordingly to previous works (see Fig. 1 for the pipeline overview of the method).22,26
Initially, whole PI-stained spheroids were immersed in a 25% formamide/10% PEG solution for 10 min. Afterward, spheroids were immersed in 50% formamide/20% PEG solution for 5 min and last in another 50% formamide/20% PEG solution for 1 h. PEG 4000, 8000, and 10,000 Da were used to prepare the solutions used in the , , and method, respectively. All methods were performed at room temperature using a plate shaker at 100 RPM.
For comparative purposes, some spheroids were only immersed in PBS instead of the clearing solutions (noncleared PI-stained spheroids). Moreover, nonstained spheroids were subjected to methods (cleared nonstained spheroids) to evaluate the influence of the method in the spheroids’ autofluorescence.
After the clearing of the intact spheroids, samples were transferred to -slide 8-well imaging plates (Ibidi GmbH, Germany) for imaging experiments.
Whole spheroids imaging by optical and CLSM
To observe any possible changes in spheroids size and transparency due to the clearing method, optical microscopy images were acquired using Olympus CX41 inverted optical microscope equipped with an Olympus SP-500 UZ digital camera and a Carl Zeiss Axio Imager A1 inverted microscope equipped with an AxioVision camera.
Images of intact HFIB spheroids were also acquired through CLSM using a Zeiss LSM 710 AxioObserver laser scanning confocal microscope (Carl Zeiss SMT, Inc., Oberkochen, Germany). For comparative purposes, all the samples were analyzed using the same equipment settings. The objective used was a air objective (Plan-Apochromat M27, ). The size of the confocal aperture was 1 Airy disk, and -stacks were collected with intervals. Laser power and master gain were kept constant during image acquisition. PI was visualized with a 514-nm argon laser.
After the acquisition of the CLSM images, an image analysis software—ImageJ, National Institutes of Health,40 was used to determine PI fluorescence levels, imaging penetration depth, and cross-section imaging depth (see Figs. 6Fig. 7–8 in the Appendix for details). The results were compared and normalized to those obtained for noncleared PI-stained spheroids (spheroids that were only treated with PBS).
Results and Discussion
In this study, spheroids composed of HFIB cells were chosen due to their superior capacity to maintain their structure and integrity during the clearing process, as previously demonstrated by our group.10 Such approach bypasses interferences of spheroids’ low cohesiveness on their structure and size during the clearing method.
Spheroids were produced by seeding cells on nonadhesive concave microwells, resulting in the formation of reproducible 3-D spherical cellular aggregates with a diameter of , after 6 days of culture. After spheroids’ fixation, they were stained with PI, a red-fluorescent stain that has been widely used for the evaluation of spheroids cellular viability through fluorescence microscopy10,33,34 and also in other works for optimizing the clearing methods.3536.–37 Then, whole spheroids were cleared by immersing them in formamide/PEG solutions [MW: 4000 Da (), 8000 Da (), or 10,000 Da ()] and then they were observed through optical and CLSM.
Clearings Do Not Influence Spheroids’ Size and Enhance their Transparency
A suitable clearing method must preserve samples’ size and structure in order to allow the study of its initial morphology.35 Additionally, the clearing method should not increase samples’ dimensions, since larger samples require longer imaging times and they may also become unsuited for whole imaging through CLSM.6 Therefore, the effect of methods in the preservation of the spheroids’ morphology was evaluated by comparing the area of the cleared spheroids with that of the noncleared spheroids (Fig. 2).
The mean area of the spheroids cleared by , , and increased by 10.35%, 9.91%, and 10.11%, respectively, when compared to that of the noncleared spheroids [Fig. 2(a)]. The slight increase of spheroids’ size was not statistically significant, which demonstrates that all the methods did not induce any significant changes on the spheroids’ morphology and that PEG MW does not affect spheroids’ size during the clearing procedure (Fig. 2). Similar results were obtained in other studies, which demonstrated that the method performed with PEG 8000 Da did not induce significant volume changes on brain sections 22 or on spheroids.26 In fact, morphological and volumes changes are mostly associated with organic solvent-based clearing methods, which use benzylalcohol/benzylbenzoate (e.g., BABB) or dichloromethane/dibenzylether (e.g., 3DISCO and iDISCO).6
The capacity of the clearing methods to enhance the transparency of spheroids was also observed in the optical microscopy images, as previously reported in the literature.21,22,26,35,36,41 The results demonstrated that, independently of the PEG MW used in the clearing process, all spheroids become gradually more transparent after each immersion step in the clearing solutions [Figs. 3(a1)–3(a4) and (9)]. In fact, the transparency of the spheroids did not seem to differ among the different methods variations. Such is explained by the fact that formamide is the main agent responsible for rendering transparency to the cells,6,42 and that all the samples were immersed in solutions containing the same concentration of formamide.
Clearing Methods Are Reversible
The reversibility of the clearing methods, i.e., if the cleared samples can return to their original morphology and nontransparency, is also important for the application of spheroids in therapeutics screening.22,23 According to Ke et al.,23 the reversibility of the clearing method also enables the analysis of the same samples by other techniques (e.g., immunohistochemistry). Additionally, nonreversible clearing methods may be indicative that the sample is chemically modified during its treatment, leading to unrealistic interpretations.23
Kuwajima et al.22 already demonstrated that the method that uses PEG 8000 Da can be reversed by immersing samples of mice embryos in PBS. Herein, it was evaluated if PEGs with other MW have any influence in the clearing reversal process of the spheroids. For this purpose, the size and transparency of the cleared spheroids were also evaluated after immersion in a PBS solution during 30 min. Independently of the method used, all spheroids maintained their original size [Fig. 2(a)] and were able to return to their nontransparent state (Fig. 9). Therefore, the results indicate that the PEG MW does not influence the reversibility and/or that it does not induce permanent modifications on spheroids’ general structure.
Methods Preserve the PI Fluorescence Intensity
The clearing process should not affect negatively the fluorescence intensity of the samples, i.e., clearing methods must preserve the fluorescence intensity in order to allow their imaging through fluorescence microscopy. In previous studies, it was demonstrated that some clearing methods, such as BABB43 and Scale,41 lack in the preservation of the probes fluorescence intensity due to the fluorescence quenching effect. Such effect was also observed in tissues that were only cleared with formamide ( method).22 This result may be explained by protein denaturation caused by formamide.6 To surpass this drawback, PEG was added to the formamide solutions to stabilize proteins’ structure and maintain their fluorescence. Kuwajima et al.22 reported that the addition of PEG 8000 Da to formamide resulted in the preservation of the fluorescence of proteins, immunohistochemistry labelings, and dye tracers (e.g., 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate, cholera toxin subunit B). Still, in other studies, it was demonstrated that the use of PEG 8000 Da did not contribute to preserve the fluorescence intensity of PI within mice brain sections.35
Therefore, we investigated the influence of the PEG MW on the capacity to preserve the fluorescence intensity of labeled spheroids. For this purpose, the fluorescence of the PI-stained noncleared and , , and cleared spheroids was determined as previously described elsewhere (see Appendix for further details).35 From the analysis of the maximum projection images (thickness equal to 50 and ) [Figs. 3(a5)–3(a12)], it was possible to verify that all the methods under investigation were able to preserve the fluorescence of the PI [Figs. 3(b) and 3(c)]. However, among the methods variations, demonstrated a significantly higher capacity to preserve the PI fluorescence, even at deep penetration depths [Figs. 3(b) and 3(c)]. Importantly, this improved fluorescence intensity was not prompted by the autofluorescence of the clearing solutions constituents, since all the solutions used in the various methods have a transmittance of from 400 to 800 nm (the absorbance peaks of the PEG polymers-formamide solutions occur below 400 nm and according to literature PEG has absorbance below 400 nm44,45) (Fig. 10). In other words, the clearing solutions are permissive to the PI excitation laser light (514 nm), and the emitted fluorescence (550 to 750 nm) is not absorbed by the solutions (Fig. 10). Moreover, nonstained cleared spheroids did not present fluorescence signals when analyzed in the same conditions used for PI-stained spheroids (Fig. 11).
PI Imaging Depth and Cross-Section Imaging Are Significantly Improved Using the Method
The fluorescence signal depth in CLSM is limited by the equipment (e.g., the working distance of the commonly used objectives) and also by the scattering phenomena between the specimen and the excitation and emission photons.4,6 Consequently, CLSM is more prone to be used for the imaging of thin samples [e.g., two-dimensional (2-D) cultures or sliced tissues]. Nevertheless, to improve the imaging of whole thick samples (e.g., spheroids) by CLSM and avoid the handicaps associated with samples sectioning,4,19,20 the reduction of the light scattering using optical clearing methods has been investigated.16 To accomplish that, some clearing methods (e.g., Scale) remove the cellular lipids that are the main source of light scattering.6,21 On the other hand, clearing methods, such as the , reduce the inhomogeneity of the light scatter by equilibrating the RI throughout the sample.6 In the method, the formamide and PEG aqueous solutions have an that will match with the overall RI of the fixed tissue ().6,29,46
To study the possible improvement in the imaging depth after spheroids clearing by methods and the PEG MW influence on this process, the penetration depth of the PI fluorescence signal was measured and presented in Fig. 4 (see Appendix for details). The results show that the improved significantly the penetration depth of the PI signal [Figs. 4(b) and 4(c)]. In fact, the use of PEG 4000 Da allowed to detect PI signal up to , whereas the PEG 8000 and 10,000 Da only enabled the acquisition of a signal up to and , respectively.
The cross-section imaging depth was also analyzed to demonstrate the influence of the clearing method, as well as the PEG MW in the PI-stained spheroids observation by CLSM (see Appendix for details). This evaluation was performed to determine which method is more suitable for acquiring fluorescence from the interior of the spheroid. In comparison to the noncleared spheroids, spheroids cleared with methods allowed the acquisition of images from PI-stained cells that were in the interior of the spheroid (Fig. 5), such is even more evident at the penetration depth of [Fig. 5(b)].
At this penetration depth, the spheroids fluorescence signal in the cross-section improved , , and when cleared with , , and , respectively. To corroborate these observations, plot profiles of the noncleared and cleared spheroids CLSM cross-section images at of penetration depth (-axis) were performed [Fig. 5(c)] (see Appendix for details). These graphs display the color intensity of the pixels throughout the spheroid cross section. As observed in Figs. 5(c1)–5(c4), without the use of a clearing method, the fluorescence observed within spheroids’ core is limited. On the other side, the clearing method allowed a better observation of the spheroid’s interior and also the attainment of a higher PI fluorescence intensity, when compared to the other methods variations [Figs. 5(c5)–5(c8)].
This improved imaging capacity of the method may be linked to the smaller size of the polymer chain of PEG 4000. Such may facilitate the PEG distribution and penetration throughout the spheroid and, consequently, the stabilization of the fluorescence probe in deeper regions, thus allowing a better imaging. In fact, large molecules penetrate slowly into the tissues47 and lower MWs are generally associated with higher diffusion coefficients.48 Moreover, the use of PEG with a low MW may further contribute to a quicker establishment of the water balance between the clearing solution and the spheroids, leading to an improved penetration of the clearing agent through the spheroid caused by the osmotic pressure.4950.51.–52 In future works, it may be interesting to use PEG with smaller MW (e.g., PEG 400 Da) to assess if they further improve the clearing efficacy of the method.
Clearing methods have been used to allow the observation and analysis of thick tissue samples and more recently spheroids by fluorescence microscopy, such as CLSM. In this work, we investigated for the first time the influence of PEG MW on clearing method ability to improve the imaging of PI-stained HFIB spheroids. In general, all the methods variations (, , and ) allowed to obtain transparent spheroids without influencing their original size. Compared to the other methods, the improved the imaging of the spheroids in what concerns (i) fluorescence intensity preservation, (ii) penetration depth, and (iii) cross-section imaging depth. Such improvements allow us to conclude that the use of PEG 4000 Da can be an improved alternative to the conventional methodology for the imaging of thick samples through fluorescence microscopy techniques, namely in the observation of the spheroids’ necrotic core or the cellular death induced by a therapeutic molecule using PI fluorescent staining. Ultimately, this article may contribute to the translation of analytical techniques, commonly used for in 2-D cell cultures, to 3-D cell cultures and, therefore, support the application of these models in the pharmaceutical industry.
Spheroids Optical Microscopy and CLSM Images Analysis
All the analyses described hereafter were performed using an image processing program designed for scientific analysis—ImageJ, National Institutes of Health.40
Measurement of the spheroids area and transparency
Spheroids area before and after the clearing process, and after clearing reversal was determined by analyzing optical microscopy images (Olympus CX41 inverted optical microscope equipped with an Olympus SP-500 UZ digital camera), as previously demonstrated in our work.10 In brief, the area of spheroids was selected in the image using a threshold and then the area of the spheroids was determined by converting the area in pixels to values (Fig. 6).
The transparency of the spheroids before and after clearing, as well as after the clearing reversal, was observed by placing the spheroids on a grill that will serve as a crosshatched background to show relative differences in transparency (as performed previously21,22,26,35,36,41). Then, optical microscopy images were acquired using a Carl Zeiss Axio Imager A1 inverted microscope equipped with an AxioVision camera.
Measurement of PI fluorescence intensity
The determination of the PI fluorescence levels in the spheroids was performed as previously described by Yu et al.35 and Grist and Nasseri.53 Initially, a maximum intensity -projection of the CLSM image with a thickness of 50 and (10 and 20 -stacks, respectively) was performed [Fig. 7(a)]. Then, the spheroid area was selected by applying a threshold [Figs. 7(b) and 7(c)]. Afterward, the integrated density (ID), the selected area (), and the mean fluorescence of the background (MFB) (region without fluorescence) were measured. These values were then used to determine the PI fluorescence intensity through the calculation of the corrected total cell fluorescence (CTCF)54
Measurement of the PI imaging depth and cross-section imaging depth
The imaging depth of PI was determined by multiplying the number of stacks with PI fluorescence signal by the -stacks thickness ().
The determination of the PI cross-section imaging depth was performed by calculating the mean gray value of the spheroid selected area at 25, 50, 75 and of penetration depth (Fig. 8). The mean gray value is a relation between the number of pixels with color in a selected area and the intensity of this color. When the mean gray value is equal to 0, it corresponds to pixels without color (no fluorescence). Mean gray values between 1 and 250 correspond to pixels with color and higher mean gray value corresponds to higher color intensity.
Additionally, similarly to previous works,41,55 the cross-section imaging depth of PI in the spheroids was analyzed by tracing plot profiles, i.e., a graph of the pixels fluorescence intensities along a line traced in the spheroid [Fig. 8(d)]. The plot profiles were obtained at of penetration depth. Lastly, the spheroids transparency (Fig. 9), optical clearing solutions absorbance and transmittance (Fig. 10) and the autofluorescence of cleared nonstained spheroids (Fig. 11) were analyzed.
This work was supported by the FEDER funds through the POCI - COMPETE 2020 - Operational Programme Competitiveness and Internationalisation in Axis I - Strengthening research, technological development, and innovation (Project POCI-01-0145-FEDER-007491) and the National Funds by FCT - Foundation for Science and Technology (Project UID/Multi/00709/2013). André F. Moreira, Duarte de Melo-Diogo, and Elisabete C. Costa acknowledge for their Grants: SFRH/BD/109482/2015, SFRH/BD/103506/2014, and SFRH/BD/103507/2014, respectively. The funders had no role in the decision to publish or in the preparation of the paper.