rus
Issue #3-4/2024
D.V.Bagrov, E.R.Pavlova, A.S.Bogdanova, A.M.Moysenovich, T.V.Mitko, A.A.Ramonova, D.V.Klinov
SEM-CLSM CORRELATION MICROSCOPY AND ITS APPLICATION TO ELECTROSPUN GELATIN FIBERS
SEM-CLSM CORRELATION MICROSCOPY AND ITS APPLICATION TO ELECTROSPUN GELATIN FIBERS
DOI: https://doi.org/10.22184/1993-8578.2024.17.3-4.208.218
The most comprehensive information about microstructure of the sample can be obtained by combining different types of high-resolution microscopy. This combination turns out to be especially informative when measurements are carried out not only on the same image, but on the same area of the sample. This approach is called correlation microscopy. Typically, such experiments require careful preparation of the sample and transferring it between the two microscopes. The current work uses correlation microscopy which combines scanning electron microscopy (SEM) and confocal laser scanning microscopy (CLSM). Electrospun gelatin fibers deposited onto metallized glass are studied using these two methods. The possibility of using correlation analysis to combine images obtained by SEM and CLSM is demonstrated.
The most comprehensive information about microstructure of the sample can be obtained by combining different types of high-resolution microscopy. This combination turns out to be especially informative when measurements are carried out not only on the same image, but on the same area of the sample. This approach is called correlation microscopy. Typically, such experiments require careful preparation of the sample and transferring it between the two microscopes. The current work uses correlation microscopy which combines scanning electron microscopy (SEM) and confocal laser scanning microscopy (CLSM). Electrospun gelatin fibers deposited onto metallized glass are studied using these two methods. The possibility of using correlation analysis to combine images obtained by SEM and CLSM is demonstrated.
INTRODUCTION
At the beginning of the 21st century, against the background of rapid development of nanotechnology, high-resolution microscopy methods have become widespread: atomic force microscopy (AFM), transmission and scanning electron microscopy (TEM and SEM), high-resolution optical microscopy and others. These methods allow not only to determine the sample structure with high spatial resolution, but also to measure some local properties. For example, AFM is used to measure local mechanical properties of samples (primarily Young’s modulus and adhesion), and X-ray microspectroscopy technique allows to determine elemental composition in SEM and TEM.
The most complete information about the sample under study can be obtained by using several techniques. In particular, there is an approach called "correlative microscopy", which involves imaging the same area of a sample using several different types of microscopy sequentially. Typically, such experiments involve physically moving the sample between microscopes, or using specialised combined instruments that combine several types of microscopy, e.g. combining AFM and SEM [1] or Raman microspectroscopy and AFM [2].
In this work, SEM and laser scanning confocal microscopy (LSCM) techniques were combined to study local properties of electrospun gelatin fibres. The results of the correlation microscopy experiments were pairs of images obtained by the two methods used. The object of the study was gelatin fibres fabricated by electrospinning. In this process, a charged jet of polymer solution moves in a strong electric field and stretches by repulsive force between close fragments of the jet. After evaporation of the solvent, the jet turns into a polymer fibre, which intensively loops and forms a nonwoven membrane; it is usually formed on a conductive plate (collector). In laboratory jet injection setups, needles with an inner diameter of the order of ~100 µm are typically used, and the fibre diameter lies in the range of ~100 nm to 1 µm. The electrospinning method has traditionally been used for production of air filters [3]; in addition, in recent years it has been used for production of wound coatings, substrates for cell cultivation, and other biomedical products [4].
The most convenient method for imaging electrospun fibres is SEM. It is used to measure fibre diameters, to determine their mutual arrangement [5] and unusual morphologies such as ribbons or beads-on-fibres [6]. In addition, if a fluorescent dye is applied to the fibres or their surface, CLSM can also be used [7]. Although the submicron diameter of fibres makes it difficult to study them with conventional optical methods, and the thinnest fibres can have diameters smaller than resolution of an optical microscope (for example, [8] describes fibres with diameters of ~50 nm), in some cases the CLSM method is extremely useful. For example, it can be used to measure porosity of an electrospun membrane [9].
The aim of this work was to implement the SEM-CLSM correlation microscopy method on a sample of electrofospun fibres and to compare the data on fibre diameters obtained by each of the two types of microscopy used. These experiments will further help to best describe the electrospun membrane structure and, in a broader context to develop correlation microscopy as a promising approach in materials science and biology.
MATERIALS AND METHODS
Sample preparation. Individual thin electrospun gelatin fibres deposited on glasses coated with a thin layer of metal were used as samples. To prepare them, round cover glasses with a diameter of 11 mm were treated were treated in the plasma cleaner Zepto B (Diener electronic, Germany) for 10 minutes at a power of 200 W and a pressure of 600–800 mBar, then a thin layer of metal (~10 nm, gold and palladium alloy) was deposited on them using a Sputter Coater Q150T magnetron sputtering unit (Quorum Technologies, UK). Scratches were made on the metal sputtering, they served as markers for selecting studied area. These coated glasses were used as a collector for electrospinning. Gelatin solution was prepared in 1,1,1,3,3,3-hexafluoroisopropanol with a concentration of 100 mg/ml, with 10% of it being FITC-conjugated gelatin.
Electrospinning was performed on a Nanofiber Electrospinning Unit (China) at an accelerating voltage of 30 kV and a needle-to-collector distance of 30 cm. The sputtering time was short, less than 1 min.
SEM study. The samples were examined using a Merlin microscope (Zeiss, Germany) at an accelerating voltage of 1 kV and an electron current of 70 pA; the working distance was in the range of 3.5–4.5 mm. An Everhart-Thornley secondary electron detector was used. Comparatively small values of accelerating voltage and current allowed us to carry out measurements without metal sputtering, which is usually applied to the surface of samples for SEM studies. The studied areas were chosen so that they were easy to find relative to the marks (scratches in the metal layer on the glass).
Examination by CLSM method. The samples were examined using an Eclipse Ti-E microscope with A1 confocal module (Nikon Corporation, Japan). Imaging was performed through a cover glass placed on top of the fibres. The specimen was rotated so that the scratches in the metal sputtering in the resulting images were approximately orientated relative to the frame boundaries as when the specimen was examined by SEM. Images were acquired using an Apo TIRF Plan Fluor 63×/1.49 objective and a 488 nm laser.
Image Analysis. Image pairs were combined using the bUnwarpJ module [10] of Fiji software [11] and then analysed using Femtoscan Online software [12].
RESULTS AND DISCUSSION
Usually, films with a thickness of ~100 μm are formed by electrospinning, but, in this experiment, it was used to obtain individual fibres on a glass substrate coated with a thin conductive layer. The samples prepared this way were studied by two methods: first by SEM and then by CLSM (Fig.1). A scratch in the form of a cross was made on the sputtering surface, it represented a mark visible in both methods, relative to which the location of the studied area was chosen.
This arrangement of the measurement procedure differs from the usual practice of fibre investigation by SEM. Usually the metal sputtering is on the fibres, while in the described experiments it was, on the contrary, under them. This helped to carry out electrospinning with fibre collection on the collector, and there was no need for a conductive coating of fibres for SEM measurements, since the imaging was carried out at low accelerating voltage (1 kV) and electron current (70 pA).
Fig.2 shows the surface images obtained by SEM and CLSM methods. When two methods are used sequentially in a correlation microscopy experiment, the first one should provide navigation – possibility to find studied area using not only the main label, but also some clearly visible surface features, for example, defects or objects with easily recognisable morphology. The second method is used to find the selected area, and the experiment as a whole results in a pair of images that can be compared and analysed.
Fig.2 shows several images that help to understand the described process course. The image of the large cross in Fig.2a was obtained at a relatively low magnification of 300X, and then a field of view was selected for examination at a higher magnification (Fig.2b). The fragment subsequently examined by the CLSM was localized within this field of view. The measurements performed by the CLSM allowed us to obtain a z-stack, i.e. a set of slices at different depths, which shows the three-dimensional structure of the sample. Once the z-stack was obtained, it was projected onto a plane, i.e. a maximum intensity projection was formed (Fig.2c).
In the CLSM study, the sample appears as a set of green fibres on a black background (FITC was used as a fluorophore); the palette can be chosen arbitrarily, it is important that the fibres are lighter than the background. The SEM image, on the contrary, looks like relatively dark fibres on a light background. For optimal matching, one of these images must be inverted (Fig.2d, inverted SEM image, an alternative will be demonstrated below). Correlation analysis can then be used to compute a pair of transformations (forward g+(x, y) and inverse g-(x, y)) that will combine the images so that their correlation is maximised. The g+ and g- transformations represent two deformation fields; for each image pixel with coordinates (x, y), they define new coordinates. These deformation fields provide shift and rotation of elements relative to the boundaries of each frame, as well as more subtle deformations such as compression or stretching of individual sections. The results of applying transformations to the images are shown in Fig.2e and Fig.2f, they were found using the bUnwarpJ module of the Fiji software.
At a qualitative level, it is visually clear that the SEM and CLSM images correspond to the same surface area, and application of transformations that maximise correlation enhances this similarity. At the same time, these images do not have to be completely identical. Indeed, the CLSM image shows the location of the fluorescent dye inside the fibres, and we are working with a maximum intensity projection, which means that we artificially collect in one frame fibres that lie in different planes. The SEM image, formed mainly by secondary electrons, contains sharp changes in brightness at the fibre edges. The contrast in this image depends on the surface topography (secondary electron emission depends on the angle of inclination of the surface) as well as on the sample material. It is difficult to expect a complete correspondence in every pixel of the two images obtained by these methods. Nevertheless, correlation analysis performed with bUnwarpJ, based on the analysis of intensities, gave a good result – it allowed to combine the images so that visually the location of fibres coincided well.
Let us discuss some features of the studied sample and its images. Firstly, many of the fibres studied were ribbons rather than cylinders, this is noticeable in Fig.2b. Formation of ribbons from various polymers, including gelatin [13], has been described in a large number of papers; the reason for this phenomenon is the loss of stability when the fibre is compressed by the surface tension force [14, 15]. As the charged jet moves from the syringe to the collector, solvent evaporation occurs so that a polymer-rich hard coating can form on the surface of the jet, while the centre of the jet remains relatively soft. When a compressive force (in electrospinning this is the surface tension force) acts on such a "hard coating on a pliable base" type system, the system can lose stability and ribbons or fibres with wrinkled surfaces are formed. According to SEM data, typical thickness of the ribbons was 300–400 nm (it can be measured when the ribbon is turned sideways and "stood on a rib"), and the width was up to 8 µm.
Secondly, the characteristic size of electrospun fibres (characteristic width of the ribbons) lay in the micron range, although it is usually in the submicron range. The diameter of the fibres is determined by the electrospinning conditions (solution concentration, feed rate, accelerating voltage, etc.), and for this work, relatively large fibres were convenient because they were easier to see with the CLSM.
Thirdly, the location of some fibres was different in the images obtained by CLSM and SEM methods. For example, in Fig.2e, Fig.2f, the green triangle on the right side of the frame shows a pair of fibres that are 2.9 µm apart in the SEM data, but directly touch each other in the CLSM data. These situations occurred several times in the images and most likely reflect fibres mobility, which caused their position to change when the specimen was transferred between microscopes.
Fourthly, some ribbons, clearly visible by SEM, turned out to be invisible for CLSM. One of such ribbons is shown with a red arrow in Fig.2e, and Fig.2f. The nature of this "disappearance" may be related to the influence of the metallised glass surface, but a detailed explanation of this phenomenon requires a deeper analysis.
The procedure of correlation analysis of images also needs additional comments. Fig.3a, Fig.3b shows a pair of images prepared for correlation analysis. The red arrows mark the location of the ribbons that are not visible in the image obtained by the CLSM method. In this case, it was inverted, so we see dark fibres on a light background. In general, the result of correlation analysis may depend on how brightness, contrast and gamma were adjusted in the compared images. However, in this paper, for each matched pair, we used minimal or no correction.
The bUnwarpJ module [10] used in this work allows for user-defined landmarks, and these landmarks greatly facilitate the search for optimal image transformations. In the images presented in Fig.3, the anchor points are shown in orange; their numbers and accuracy of their placement also affect the matching result. Usually, 10–15 reference points were sufficient to analyse each pair of images.
Let us compare how the sizes of individual fibres measured by SEM and CLSM methods relate to each other. To answer this question, we measured the ribbons width in each of the images. The results of these measurements are shown in Fig.4; each point on the graph corresponds to a specific point of one of the fibres. It can be seen that most of the measurement results lie above the Y = X line, i.e., the width of the ribbons measured by the CLSM (DCLSM) appears to be systematically larger than by the SEM method (DSEM).
This may be the result of several factors. Firstly, it may be due to the lack of resolution of the CLSM (the fluorescent signal passes through the metallised substrate, which reduces the brightness of the image and worsens the resolution). Secondly, in this case we work with maximum intensity projections, and on them the width of objects is generally not lower than on separate slices. Thirdly, we cannot exclude some mechanical displacement of fibres. It can lead to their sticking together (Fig.2e, Fig.2f, green triangle) and, in general, to more complex deformation, e.g., ribbon rotation. This may explain the rare situation in which DKLSM<DSEM (one point below the line Y = X in Fig.4).
Single descriptions of correlation microscopy experiments combining fluorescence and SEM measurements can be found in the scientific literature. For example, this approach was used to detect artefacts caused by dehydration of eukaryotic cells (human mesenchymal stem cells) during their preparation for SEM examination [16]. Correlation microscopy, which combines fluorescence microscopy and SEM, was used to assist in the manipulation of the sample in the SEM vacuum chamber, mechanically extracting DNA from cell nuclei [17]. The experiments described in this paper are a rare example of combining fluorescence microscopy and SEM in the study of a sample of synthetic rather than biological origin.
CONCLUSIONS
This paper describes the technique of correlation microscopy, which allows imaging the same region of the sample surface by SEM and CLSM methods sequentially. This technique was used to study the structure of individual electrospun gelatin ribbons deposited on metallised glass. It turned out that the typical size of the ribbons measured by maximum intensity projections is usually larger than their size measured by SEM data.
In order to make the sequential application of the two methods possible, SEM measurements were performed without metal sputtering on the sample, at low current. The fluorescence tag required for SEM measurements was incorporated into the electrospun fibres during their manufacture.
Despite the fact that the images obtained by SEM and CLSM methods are radically different in terms of the location of visible elements and the mechanism of contrast generation, correlation analysis can be used to combine them. This is done using the bUnwarpJ module of the Fiji software.
This work reveals a number of unusual effects and opportunities for further development. From the point of data processing, it seems rational to perform a full-fledged analysis of the z-stack, rather than the maximum intensity projection, in order to maximise the use of the data obtained by the CLSM method. This will require development of a non-trivial specialised correlation analysis procedure that compares not a pair of images, but an image with a z-stack. Formation of such procedure will make it possible to extract the maximum information from the described experiments.
Many technical aspects can be modified to provide better resolution of the CLSM. For example, another dye that fluoresces at a shorter wavelength can be used instead of FITC, and the samples can be encapsulated in a casting medium.
A non-trivial effect of fluorescence extinction of gelatin ribbons placed directly on the metal sputtering was found. Usually, the effect of fluorescence quenching near the metal surface appears at a distance of 1–2 nm from it [18], but in our case the thickness of the ribbons is higher and the mechanism of this effect requires further study.
Overall, the described experiments help to develop the high-resolution microscopy and emphasise complementary role of different types of microscopy, in particular SEM and CLSM.
ACKNOWLEDGMENTS
This work was supported by the grant of the Russian Science Foundation, project No. 21-74-10042.
PEER REVIEW INFO
Editorial board thanks the anonymous reviewer(s) for their contribution to the peer review of this work. It is also grateful for their consent to publish papers on the journal’s website and SEL eLibrary eLIBRARY.RU.
Declaration of Competing Interest. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
At the beginning of the 21st century, against the background of rapid development of nanotechnology, high-resolution microscopy methods have become widespread: atomic force microscopy (AFM), transmission and scanning electron microscopy (TEM and SEM), high-resolution optical microscopy and others. These methods allow not only to determine the sample structure with high spatial resolution, but also to measure some local properties. For example, AFM is used to measure local mechanical properties of samples (primarily Young’s modulus and adhesion), and X-ray microspectroscopy technique allows to determine elemental composition in SEM and TEM.
The most complete information about the sample under study can be obtained by using several techniques. In particular, there is an approach called "correlative microscopy", which involves imaging the same area of a sample using several different types of microscopy sequentially. Typically, such experiments involve physically moving the sample between microscopes, or using specialised combined instruments that combine several types of microscopy, e.g. combining AFM and SEM [1] or Raman microspectroscopy and AFM [2].
In this work, SEM and laser scanning confocal microscopy (LSCM) techniques were combined to study local properties of electrospun gelatin fibres. The results of the correlation microscopy experiments were pairs of images obtained by the two methods used. The object of the study was gelatin fibres fabricated by electrospinning. In this process, a charged jet of polymer solution moves in a strong electric field and stretches by repulsive force between close fragments of the jet. After evaporation of the solvent, the jet turns into a polymer fibre, which intensively loops and forms a nonwoven membrane; it is usually formed on a conductive plate (collector). In laboratory jet injection setups, needles with an inner diameter of the order of ~100 µm are typically used, and the fibre diameter lies in the range of ~100 nm to 1 µm. The electrospinning method has traditionally been used for production of air filters [3]; in addition, in recent years it has been used for production of wound coatings, substrates for cell cultivation, and other biomedical products [4].
The most convenient method for imaging electrospun fibres is SEM. It is used to measure fibre diameters, to determine their mutual arrangement [5] and unusual morphologies such as ribbons or beads-on-fibres [6]. In addition, if a fluorescent dye is applied to the fibres or their surface, CLSM can also be used [7]. Although the submicron diameter of fibres makes it difficult to study them with conventional optical methods, and the thinnest fibres can have diameters smaller than resolution of an optical microscope (for example, [8] describes fibres with diameters of ~50 nm), in some cases the CLSM method is extremely useful. For example, it can be used to measure porosity of an electrospun membrane [9].
The aim of this work was to implement the SEM-CLSM correlation microscopy method on a sample of electrofospun fibres and to compare the data on fibre diameters obtained by each of the two types of microscopy used. These experiments will further help to best describe the electrospun membrane structure and, in a broader context to develop correlation microscopy as a promising approach in materials science and biology.
MATERIALS AND METHODS
Sample preparation. Individual thin electrospun gelatin fibres deposited on glasses coated with a thin layer of metal were used as samples. To prepare them, round cover glasses with a diameter of 11 mm were treated were treated in the plasma cleaner Zepto B (Diener electronic, Germany) for 10 minutes at a power of 200 W and a pressure of 600–800 mBar, then a thin layer of metal (~10 nm, gold and palladium alloy) was deposited on them using a Sputter Coater Q150T magnetron sputtering unit (Quorum Technologies, UK). Scratches were made on the metal sputtering, they served as markers for selecting studied area. These coated glasses were used as a collector for electrospinning. Gelatin solution was prepared in 1,1,1,3,3,3-hexafluoroisopropanol with a concentration of 100 mg/ml, with 10% of it being FITC-conjugated gelatin.
Electrospinning was performed on a Nanofiber Electrospinning Unit (China) at an accelerating voltage of 30 kV and a needle-to-collector distance of 30 cm. The sputtering time was short, less than 1 min.
SEM study. The samples were examined using a Merlin microscope (Zeiss, Germany) at an accelerating voltage of 1 kV and an electron current of 70 pA; the working distance was in the range of 3.5–4.5 mm. An Everhart-Thornley secondary electron detector was used. Comparatively small values of accelerating voltage and current allowed us to carry out measurements without metal sputtering, which is usually applied to the surface of samples for SEM studies. The studied areas were chosen so that they were easy to find relative to the marks (scratches in the metal layer on the glass).
Examination by CLSM method. The samples were examined using an Eclipse Ti-E microscope with A1 confocal module (Nikon Corporation, Japan). Imaging was performed through a cover glass placed on top of the fibres. The specimen was rotated so that the scratches in the metal sputtering in the resulting images were approximately orientated relative to the frame boundaries as when the specimen was examined by SEM. Images were acquired using an Apo TIRF Plan Fluor 63×/1.49 objective and a 488 nm laser.
Image Analysis. Image pairs were combined using the bUnwarpJ module [10] of Fiji software [11] and then analysed using Femtoscan Online software [12].
RESULTS AND DISCUSSION
Usually, films with a thickness of ~100 μm are formed by electrospinning, but, in this experiment, it was used to obtain individual fibres on a glass substrate coated with a thin conductive layer. The samples prepared this way were studied by two methods: first by SEM and then by CLSM (Fig.1). A scratch in the form of a cross was made on the sputtering surface, it represented a mark visible in both methods, relative to which the location of the studied area was chosen.
This arrangement of the measurement procedure differs from the usual practice of fibre investigation by SEM. Usually the metal sputtering is on the fibres, while in the described experiments it was, on the contrary, under them. This helped to carry out electrospinning with fibre collection on the collector, and there was no need for a conductive coating of fibres for SEM measurements, since the imaging was carried out at low accelerating voltage (1 kV) and electron current (70 pA).
Fig.2 shows the surface images obtained by SEM and CLSM methods. When two methods are used sequentially in a correlation microscopy experiment, the first one should provide navigation – possibility to find studied area using not only the main label, but also some clearly visible surface features, for example, defects or objects with easily recognisable morphology. The second method is used to find the selected area, and the experiment as a whole results in a pair of images that can be compared and analysed.
Fig.2 shows several images that help to understand the described process course. The image of the large cross in Fig.2a was obtained at a relatively low magnification of 300X, and then a field of view was selected for examination at a higher magnification (Fig.2b). The fragment subsequently examined by the CLSM was localized within this field of view. The measurements performed by the CLSM allowed us to obtain a z-stack, i.e. a set of slices at different depths, which shows the three-dimensional structure of the sample. Once the z-stack was obtained, it was projected onto a plane, i.e. a maximum intensity projection was formed (Fig.2c).
In the CLSM study, the sample appears as a set of green fibres on a black background (FITC was used as a fluorophore); the palette can be chosen arbitrarily, it is important that the fibres are lighter than the background. The SEM image, on the contrary, looks like relatively dark fibres on a light background. For optimal matching, one of these images must be inverted (Fig.2d, inverted SEM image, an alternative will be demonstrated below). Correlation analysis can then be used to compute a pair of transformations (forward g+(x, y) and inverse g-(x, y)) that will combine the images so that their correlation is maximised. The g+ and g- transformations represent two deformation fields; for each image pixel with coordinates (x, y), they define new coordinates. These deformation fields provide shift and rotation of elements relative to the boundaries of each frame, as well as more subtle deformations such as compression or stretching of individual sections. The results of applying transformations to the images are shown in Fig.2e and Fig.2f, they were found using the bUnwarpJ module of the Fiji software.
At a qualitative level, it is visually clear that the SEM and CLSM images correspond to the same surface area, and application of transformations that maximise correlation enhances this similarity. At the same time, these images do not have to be completely identical. Indeed, the CLSM image shows the location of the fluorescent dye inside the fibres, and we are working with a maximum intensity projection, which means that we artificially collect in one frame fibres that lie in different planes. The SEM image, formed mainly by secondary electrons, contains sharp changes in brightness at the fibre edges. The contrast in this image depends on the surface topography (secondary electron emission depends on the angle of inclination of the surface) as well as on the sample material. It is difficult to expect a complete correspondence in every pixel of the two images obtained by these methods. Nevertheless, correlation analysis performed with bUnwarpJ, based on the analysis of intensities, gave a good result – it allowed to combine the images so that visually the location of fibres coincided well.
Let us discuss some features of the studied sample and its images. Firstly, many of the fibres studied were ribbons rather than cylinders, this is noticeable in Fig.2b. Formation of ribbons from various polymers, including gelatin [13], has been described in a large number of papers; the reason for this phenomenon is the loss of stability when the fibre is compressed by the surface tension force [14, 15]. As the charged jet moves from the syringe to the collector, solvent evaporation occurs so that a polymer-rich hard coating can form on the surface of the jet, while the centre of the jet remains relatively soft. When a compressive force (in electrospinning this is the surface tension force) acts on such a "hard coating on a pliable base" type system, the system can lose stability and ribbons or fibres with wrinkled surfaces are formed. According to SEM data, typical thickness of the ribbons was 300–400 nm (it can be measured when the ribbon is turned sideways and "stood on a rib"), and the width was up to 8 µm.
Secondly, the characteristic size of electrospun fibres (characteristic width of the ribbons) lay in the micron range, although it is usually in the submicron range. The diameter of the fibres is determined by the electrospinning conditions (solution concentration, feed rate, accelerating voltage, etc.), and for this work, relatively large fibres were convenient because they were easier to see with the CLSM.
Thirdly, the location of some fibres was different in the images obtained by CLSM and SEM methods. For example, in Fig.2e, Fig.2f, the green triangle on the right side of the frame shows a pair of fibres that are 2.9 µm apart in the SEM data, but directly touch each other in the CLSM data. These situations occurred several times in the images and most likely reflect fibres mobility, which caused their position to change when the specimen was transferred between microscopes.
Fourthly, some ribbons, clearly visible by SEM, turned out to be invisible for CLSM. One of such ribbons is shown with a red arrow in Fig.2e, and Fig.2f. The nature of this "disappearance" may be related to the influence of the metallised glass surface, but a detailed explanation of this phenomenon requires a deeper analysis.
The procedure of correlation analysis of images also needs additional comments. Fig.3a, Fig.3b shows a pair of images prepared for correlation analysis. The red arrows mark the location of the ribbons that are not visible in the image obtained by the CLSM method. In this case, it was inverted, so we see dark fibres on a light background. In general, the result of correlation analysis may depend on how brightness, contrast and gamma were adjusted in the compared images. However, in this paper, for each matched pair, we used minimal or no correction.
The bUnwarpJ module [10] used in this work allows for user-defined landmarks, and these landmarks greatly facilitate the search for optimal image transformations. In the images presented in Fig.3, the anchor points are shown in orange; their numbers and accuracy of their placement also affect the matching result. Usually, 10–15 reference points were sufficient to analyse each pair of images.
Let us compare how the sizes of individual fibres measured by SEM and CLSM methods relate to each other. To answer this question, we measured the ribbons width in each of the images. The results of these measurements are shown in Fig.4; each point on the graph corresponds to a specific point of one of the fibres. It can be seen that most of the measurement results lie above the Y = X line, i.e., the width of the ribbons measured by the CLSM (DCLSM) appears to be systematically larger than by the SEM method (DSEM).
This may be the result of several factors. Firstly, it may be due to the lack of resolution of the CLSM (the fluorescent signal passes through the metallised substrate, which reduces the brightness of the image and worsens the resolution). Secondly, in this case we work with maximum intensity projections, and on them the width of objects is generally not lower than on separate slices. Thirdly, we cannot exclude some mechanical displacement of fibres. It can lead to their sticking together (Fig.2e, Fig.2f, green triangle) and, in general, to more complex deformation, e.g., ribbon rotation. This may explain the rare situation in which DKLSM<DSEM (one point below the line Y = X in Fig.4).
Single descriptions of correlation microscopy experiments combining fluorescence and SEM measurements can be found in the scientific literature. For example, this approach was used to detect artefacts caused by dehydration of eukaryotic cells (human mesenchymal stem cells) during their preparation for SEM examination [16]. Correlation microscopy, which combines fluorescence microscopy and SEM, was used to assist in the manipulation of the sample in the SEM vacuum chamber, mechanically extracting DNA from cell nuclei [17]. The experiments described in this paper are a rare example of combining fluorescence microscopy and SEM in the study of a sample of synthetic rather than biological origin.
CONCLUSIONS
This paper describes the technique of correlation microscopy, which allows imaging the same region of the sample surface by SEM and CLSM methods sequentially. This technique was used to study the structure of individual electrospun gelatin ribbons deposited on metallised glass. It turned out that the typical size of the ribbons measured by maximum intensity projections is usually larger than their size measured by SEM data.
In order to make the sequential application of the two methods possible, SEM measurements were performed without metal sputtering on the sample, at low current. The fluorescence tag required for SEM measurements was incorporated into the electrospun fibres during their manufacture.
Despite the fact that the images obtained by SEM and CLSM methods are radically different in terms of the location of visible elements and the mechanism of contrast generation, correlation analysis can be used to combine them. This is done using the bUnwarpJ module of the Fiji software.
This work reveals a number of unusual effects and opportunities for further development. From the point of data processing, it seems rational to perform a full-fledged analysis of the z-stack, rather than the maximum intensity projection, in order to maximise the use of the data obtained by the CLSM method. This will require development of a non-trivial specialised correlation analysis procedure that compares not a pair of images, but an image with a z-stack. Formation of such procedure will make it possible to extract the maximum information from the described experiments.
Many technical aspects can be modified to provide better resolution of the CLSM. For example, another dye that fluoresces at a shorter wavelength can be used instead of FITC, and the samples can be encapsulated in a casting medium.
A non-trivial effect of fluorescence extinction of gelatin ribbons placed directly on the metal sputtering was found. Usually, the effect of fluorescence quenching near the metal surface appears at a distance of 1–2 nm from it [18], but in our case the thickness of the ribbons is higher and the mechanism of this effect requires further study.
Overall, the described experiments help to develop the high-resolution microscopy and emphasise complementary role of different types of microscopy, in particular SEM and CLSM.
ACKNOWLEDGMENTS
This work was supported by the grant of the Russian Science Foundation, project No. 21-74-10042.
PEER REVIEW INFO
Editorial board thanks the anonymous reviewer(s) for their contribution to the peer review of this work. It is also grateful for their consent to publish papers on the journal’s website and SEL eLibrary eLIBRARY.RU.
Declaration of Competing Interest. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Readers feedback
