Fig 1.
Quantitative parameters, stereological test systems, sampling, section orientation and stereological probes in quantitative stereological analyses.
Volume densities (VV(structure of interest/reference space)), length densities (LV(structure of interest/reference space)) and surface area densities (SV(structure of interest/reference space)) are estimated in representative, systematically uniform random (SUR) sampled 2-D sections of the reference space. Volume densities are deduced from the fractional areas of the structure of interest and the reference space, determined e.g., by point counting. Length densities are estimated on isotropic uniform random (IUR) sections from the number of intersections of the structure of interest with the section area. (Note that the present guidelines do not cover quantitative stereological estimation of length parameters of gill structures. Determination of the true harmonic mean of the diffusion barrier thickness in the secondary lamellae (SL) is described in Section 15.) For estimation of surface area densities, the number of interactions of the examined surface area with appropriate stereological probes is counted in vertical uniform random (VUR) sections. Estimation of numerical volume densities (NV(structure of interest/reference space)) requires 3-D test systems, such as the physical disector, to sample and count particles. A physical disector is a stereological probe used for unbiased counting and sampling of particles. It consists of two parallel histological sections (a reference section and a look-up section) with a defined distance, thus defining a known tissue volume. Particles that are sectioned in the reference section, but not in the look-up section are counted (Q-), using the unbiased counting frame. Estimation of NV(structure of interest/reference space) using the physical disector is described in detail in Section 14. Absolute quantities of volumes, lengths, surfaces and numbers are obtained from the respective densities and the total reference space volume. Mean particle volumes are calculated from their volume densities and their numerical volume densities in the reference space, as described in Section 14. *The section plane orientation illustrated for the corresponding parameters is highly recommended, but there are several options for most morphological parameters regarding the orientation of the section plane.
Table 1.
Adequate sample section plane orientation, embedding medium and tissue shrinkage correction factor (fs) for different quantitative morphological parameters.
Fig 2.
Schematic illustration of the 3-D histo-architecture of the gills and corresponding 2-D histological sections.
Fig 3.
Appearance of histological gill section profiles in different section plane orientations (schematically indicated).
A. Transverse section. B. Frontal section. C. Sagittal section. Important morphological structures are indicated: GS: Gill arch support skeleton, CR: Cartilage rod, PL: Primary gill lamellae, SL: Secondary gill lamellae. Haematoxylin and Eosin (HE)-stained sections of formalin-fixed and paraffin-embedded (FFPE) gills. Bars = 200 μm.
Fig 4.
Gill histomorphology and ultrastructure.
Important morphological structures are indicated: PL: Primary gill lamellae, SL: Secondary gill lamellae, PC: Pillar cell, EC: Epithelial cell, ERY: Nucleated erythrocyte inside a SL-capillary, BM: Basement membrane of the SL-capillary. A, B. Light-microscopic histomorphology of gill filaments (sagittal section). A. The anatomical border between the SL and the PL is indicated by arrows and a dashed green line. FFPE. HE. B. Semithin section of Epon-embedded gill filaments. Toluidine blue (TB) staining. Bars = 50 μm. C. Ultrastructure of the secondary gill lamella. The oxygen diffusion barrier between the epithelial surface of the SL and the capillary lumen is indicated by a double arrow. Transmission electron micrograph. Bar = 5 μm.
Table 2.
Relevant quantitative stereological gill parameters.
Fig 5.
General experimental design for quantitative stereological analyses of trout gill morphology in ecotoxicological studies.
The paper sections containing the respective sample processing and analysis steps are indicated (5–15). At the end point of the study, fish are euthanized, using a gill-preserving killing method (Section 5). Optional perfusion fixation, excision-, fixation- and volumetry of gills after removal of the gill arch are described in Sections 6–8. Section 9 presents the systematic uniform random (SUR) sampling of representative gill filament specimens for stereological analyses. aAfter SUR sampling of the specimens, it is strongly recommended to store the remaining fixed gills and not discard them until final completion of the study. The subsequent processing steps and analysis methods depend on the individual quantitative stereological parameters of interest (Sections 10–15).
Fig 6.
Using scissors, the peritoneal cavity is opened, starting with a transverse incision ventro-caudal to the base of the pectoral fins (B). This incision is elongated caudally along the ventral midline, up to a few millimeters cranial of the anogenital papilla (A). Just behind the pectoral fins and the cleithrum, the incision is continued in dorso-cranial direction, ending in the opercular chamber by severing the cleithrum (A). Then, the operculum is removed and the ventral part of the gill basket is disconnected from the viscerocranium (in D, the orientation of the incision line is indicated by a green dotted line). The dorso-cranial connection between the gill basket and the skull is disconnected by severing the rostral pharyngobranchial bones (indicated by a blue dotted line in D). Subsequently, the esophagus is cut through (indicated by a black dotted line in C), the dorsal connection between the gills and the skull is transected (indicated by a red dotted line in D) and the gill basket is removed from the body by gently pulling the gill arches in ventral direction (arrow in C). E. Dissected gill basket. Top: dorsal aspect. Gills I-IV are indicated. Bottom: ventral aspect. Rostral (ro). F. Dissected gill arches of the left (l) and right (r) side, prior to immersion fixation. Bars = 1 cm.
Fig 7.
Determination of gill filament density and sample volume, using the submersion method (Archimedes`principle).
A. Fixed gill after excision of the gill arch, briefly blotted dry on lab-paper towel to remove adhering liquid. B. Scale tared to the weight of the sample holder. C. Measurement of the GF sample weight (m(GF)). D. Scale tared to the weight of a container filled with physiological saline or fixative of known density and the submerged sample holder. Note that the sample holder is not placed on the scale sensor and does not have contact to the bottom or the walls of the container. The arrow indicates the position up to which the sample holder is submerged into the liquid. E. The sample is attached to the sample holder and completely submerged into the liquid up to the marked position on the sample holder. Note that the sample does not have contact to the bottom or the walls of the container. The weight of the fluid displaced by the GF sample (m(F)) is recorded. The sample volume is calculated from the sample weight (m(GF)) and the GF density (ρ(GF)), using Eq 1.
Fig 8.
Systematic uniform random (SUR) sampling of representative gill filament samples.
A. The 4 formalin-fixed gills of one side before the removal of the gill arches. B. After removal of the gill arches, the gills are placed on their opercular (i.e., lateral) sides. Optionally, the GF are briefly dipped in liquid agar for stabilization. C. The GF are randomly superimposed with an appropriately sized cross-grid printed on a plastic transparency. Here, a 6 mm grid is used. To randomize the position of the grid relative to the gills, the upper left cross of the grid is placed over a random point outside of the GF (indicated by a red spot). D. All crosses hitting the GF are marked (red crosses). In the present example, 27 crosses hit the GF. E. The sampling interval (i) is defined by the number of crosses hitting the sampled reference compartment (n) and the number of SUR samples to be generated (s) for analysis of a specific quantitative stereological gill parameter (i = n/s). In the present example, ≥ 5 samples are to be generated. Accordingly, every 5th position (27/5 = 5.4) where a cross hits the GF is sampled. The position of the first location to be sampled is determined randomly within the sampling interval (1-i; here position N°3), using a random number table/generator. Thus, in the present example 5 GF locations are SUR sampled (N°3, N°8, N°13, N°18, N°23), as indicated by red circles. (Counting proceeds from the left to the right and from top to bottom.) F. Detail enlargement of the second gill with the sampling location N°8, indicated by the red circle. G. SUR sampled GF sites are excised, using a 6 mm biopsy punch. H. Detail enlargement of an excised SUR sampled GF specimen. Bars = 1 cm in A-G and = 0.25 cm in H.
Table 3.
Recommended sampling design and sample number for quantitative stereological analysis of different morphological gill parameters.
Fig 9.
Generation of IUR sections of a SUR sampled gill filament specimen with the Isector method.
A. A SUR sampled specimen of fixed gill filaments is carefully cut to a size of approximately 1 mm x 1 mm x 1 mm (suitable for Epon-embedding and electron microscopy), preserving the secondary lamellae as structure of interest. B. The specimen is embedded in a sphere of epoxy resin, using a spherical casting mould. C, D. After polymerization of the embedding medium, the sphere is rolled across a flat surface and stopped at a random position. F, G. The sphere is sectioned at this random position, resulting in an IUR section plane.
Fig 10.
Generation of VUR sections of a SUR sampled gill filament sample.
A. SUR sampling positions on the GF are marked by confetti paper. The vertical axis (VA) is indicated. B. SURS sample excised with a biopsy punch. The orientation of the sample relative to the gill is marked by a black line on the confetti paper (0°-180°-line). C. The sample is placed on an equiangular circle, corresponding to the 0°-180°-mark on the confetti paper. D-G. The first SURS specimen is randomly rotated around the VA by an angle between 0° and 36°, determined using a random number generator (here: 2°). The following four samples are systematically rotated around their VA in a predefined rotation interval of 36° (here: 38°, 74°, 110°, and 146°). The samples are sectioned at the corresponding positions (parallel to the VA). H. The samples are embedded in plastic medium (e.g., GMA/MMA), maintaining the orientation of their VUR section surfaces, the VA is still identifiable in light microscopy. I. The resulting histological sections are VUR sections, used for analysis of the surface area densities of the secondary gill lamellae in the gill filaments (SV(SL/GF)), as described in Section 13.
Fig 11.
Determination of volume shrinkage of gill filaments due to embedding in plastic embedding media.
A-D. Volume determination of SUR sampled GF samples prior to embedding. After dissection of the gill arch (A), the density of the (fixed) GF samples is determined (B). The GF samples are excised by biopsy punch of 0.2 cm diameter (C) and the samples weight is recorded (D). GF sample volume is calculated from GF density and sample weight (submersion technique, refer to Section 8 and Eq 1). E-H. Embedding of GF samples in plastic medium and exhaustive serial sectioning of the embedded samples. For stabilization and visual contrast, the GF samples are embedded in ink-dyed black agar (E) and subsequently routinely processed and embedded in plastic embedding medium (F) (here: GMA/MMA). The embedded samples are then exhaustively sectioned over the entire sample height (h) with a defined section thickness (G). From the section series (H), sections are taken in a defined interval (e.g., every 40th section) and mounted on a glass slide. The factual thicknesses of the individual sections are determined by spectral reflectance measurement (not shown) [80]. I-J. Determination of GF sample section areas in equidistant serial sections. The section profile areas of all samples in all examined sections are determined (here: point counting). Digital microscopic section images are randomly overlaid with a grid of equally spaced test points (crosses) of known distance at the given magnification (i.e., every point (P) is associated with a defined area (A)). The number of points hitting GF section profiles are counted. Since the entire height (h) of the GF samples was sectioned into equidistant, parallel sections, the volume of the embedded GF samples can be calculated according to the principle of Cavalieri [24,59,63], from the total section profile area of all samples in all sections (A(GF sample section profiles)) and the mean distance between two examined sections (i.e., section thickness x section interval). A(GF sample section profiles) is calculated from the total number of counted points (∑P) and the area associated with each point (A/P) (refer to Eqs 2 and 4). The proportional volume shrinkage of GF samples associated with the embedding in the histological plastic embedding medium is calculated from the quotient of the sample volume prior to and after embedding. The linear tissue shrinkage factor (fs) for gill filaments embedded in plastic medium is calculated as shown in Eq 3. Bar = 1 cm in I (left image side) and = 500 μm in I (right image side) and J.
Fig 12.
Estimation of the volume density of the secondary lamellae in the gill filaments.
A. SUR sampled GF specimens (please compare to Section 9). B, C. SUR sampling of sections from samples embedded in paraffin in non-random orientation. B. Paraffin block with sagittally embedded GF samples. C. SUR sampling of sections. The entire block is exhaustively sectioned. For subsequent analyses, sections are taken in a defined sampling interval (i). The first section is randomly taken within the sampling interval (>0≤i). D-G. Estimation of VV(SL/GF) by point counting. D. SUR sampled section of GF specimens. E. SUR sampling of test fields within the section, performed at a factor of magnification allowing for a reliable differentiation of PL and SL (e.g., 40x-100x microscopic magnification). All sampled sections per case are entirely screened, following a defined meander pattern and test fields are SUR sampled in a defined interval i (i.e., every ith field of view containing GF section profiles), with the first field of view being randomly selected within the sampling interval. F. SUR sampled test field overlaid with an appropriately sized cross grid (here: 8x8 points at 100x microscopic magnification). The number of points hitting GF section profiles (P(GF), indicated by bold crosses) in all sections per case are counted, as well as the number of points hitting SL section profiles (P(SL), indicated by green crosses). 34 points hit the entire GF section profile, 23 points hit the SL section profile. VV(SL/GF) is calculated as the point density of P(SL) and P(GF). The total SL volume (V(SL,GF)) is calculated as the product of VV(SL/GF) and the GF volume V(GF).
Fig 13.
Estimation of the surface area density of the secondary lamellae in the gill filaments.
A SUR sampled microscopic test field in a VUR section of a GMA/MMA-embedded (SUR sampled) GF sample is superimposed with a stereological test system combining 35 cycloids and 70 points. The short side of the rectangular frame of the system and therewith the minor axis of the cycloids is aligned parallel to the orientation of the vertical axis of the VUR GF section (VA, indicated by the arrow on the left). All points hitting GF section profiles (P(GF), indicated in green) are counted, as well as all intersections of cycloids with the epithelial surface of SL section profiles (I(SL), encircled in red). The SL surface area density in the GF is calculated from the sum of intersections (∑I(SL)) and points (∑P(GF)), counted in all examined test fields of all sections of all samples per case, using Eq 6. In the presented example, the length of one cycloid (l) = 1/10 of the frame width [62], the test curve length in general is calculated from cycloid arch height h as: l = 2 x h [104]. GMA/MMA. HE. Bar = 50 μm.
Fig 14.
Estimation of the numerical volume density of epithelial cells in the secondary gill lamellae.
The number of epithelial cells (EC) per volume unit of secondary lamellae (SL) is estimated, using the physical disector as a 3-D stereological test system for unbiased counting of particles. A physical disector consists of two parallel histological sections (a reference section and a look-up section) with a defined distance (disector height, h). The reference section is SUR sampled from a series of technically impeccable, parallel, consecutive sections with a defined nominal section thickness. The factual physical thickness of the sections (d) defines the disector height. The present example shows corresponding fields of view in a series of 5 consecutive, toluidine blue stained, semithin IUR GF sections with a nominal thickness of 0.5 μm (the examined fields of view in the section are SUR sampled at the given factor of magnification). From the series of five sections, the second section is SUR sampled as reference section. The fourth section (i.e., with a distance of h = 2 x d = 2 x 0.5 μm = 1 μm) is defined as look-up section. Corresponding section areas in the reference- and the look-up section are overlaid with an unbiased counting frame [110] of known area and a cross grid of equally spaced test points. The volume of the reference compartment defined by the disector probe, i.e., the volume of SL within the 3-D space defined by both sections of the disector, is given by the disector height and the mean area of the SL section profile(s) (i.e., the reference compartment) present in the reference- and the look-up section. The section area of SL within the area of the unbiased counting frame is determined by point counting: the number of crosses hitting SL section profiles is counted and multiplied by the area associated with a single point/cross of the grid (i.e., the quotient of the number of crosses in the counting frame and the area of the counting frame). SL-EC nuclei that are sectioned in the reference section, but not in the look-up section are counted (Q-), using the unbiased counting frame (particle sections are only counted if they are completely located within the unbiased counting frame or if they touch one of the “acceptance” (green) border lines. Any particle section profiles touching an “exclusion” (red) line are not counted) [25]. In the presented example, a SL-EC nucleus section profile that is present in the reference section, but absent in the look-up section, is indicated by red arrows. The numerical volume density of the SL-EC is then calculated from the EC number counted in all sections of all samples per case in all analyzed disectors and the cumulative reference compartment (SL) volume in all analyzed disectors (Eq 8).
Fig 15.
Determination of the true harmonic mean of the diffusion barrier thickness in secondary gill lamellae.
A printed transmission electron microscopic (TEM) image of a SUR sampled field of view of an IUR section of a secondary lamella (SL) is superimposed with a test grid of lines (red). The transections of the gridlines with the epithelial SL surface are marked by red circles. At these locations, the shortest distance (t, double arrow) between the epithelial surface of the SL and the inner surface of the blood space (green arrows) is determined (dashed circles). Along these lines, the diffusion barrier thickness is measured, using a superimposed logarithmic ruler. In the shown example, the measured distance falls in class 9. Th(DB) is calculated from the number of measurements and the corresponding classes (Eq 13), detailed illustration is given in Hirose et al. [115]. Epon. TEM. Bar = 2 μm.
Fig 16.
Optical clearing of rainbow trout gills.
A. Formalin-fixed (non-transparent) rainbow trout gills, placed on a mm ruler. B. The identical gills after optical tissue clearing with the 3DISCO protocol. Note the transparency and the shrinkage of the cleared gills. Note that the images in A and B show intact gills (with gill arches). For LSFM-based quantitative morphological analyses, the gill arches are removed from the gills and the gill filaments are cleared after determination of their weight or volume.
Fig 17.
Laser light sheet fluorescence microscopy (LSFM) of an optically cleared gill filament sample.
The cleared GF sample in the sample chamber is stepwise illuminated by thin laser light sheets of defined wavelength ranges. Here, autofluorescence signals emitted by GF structures excited by the laser light energy are detected by a digital charge-coupled device (CCD) camera, resulting in a 2-D fluorescent image of the illuminated focus plane in the sample. The sample is gradually moved through the laser light sheet(s) along its vertical axis (red arrow), resulting in the acquisition of a z-stack of serial fluorescence images, used to compute a 3-D reconstruction of the sample. In the present example, autofluorescence images of a 3DISCO-cleared GF sample acquired at 520/40 nm (excitation range) (ex) and 585/40 nm (emission range) (em) are shown. Bar = 200 μm.
Fig 18.
Generation of virtual optical VUR sections of cleared gill filament (GF) samples by LSFM.
A. Left image side: top view of a sample holder with a rotatable axis equipped with a needle for sample attachment. Right image side: 8 stripe-shaped samples of optically cleared GF containing the SUR sampled locations (SURS-Loc., indicated by red circles). B-E. For generation of (virtual, optical) VUR sections, the GF sample is pinned to the rotatable axis (B) of the sample holder, so that the opercular side of the sample is oriented perpendicular to the (user-defined, virtual) vertical axis (VA, indicated by a red arrow). The sample is then rotated in a defined rotation interval (compare to Fig 10). In D, a rotation angle of ~40° is shown. A green schematic plane indicates the orientation of a corresponding vertical section plane (VSP), relative to the VA and the sample. The sample holder with the attached sample is then transferred into the sample chamber of the LSF-microscope (E), maintaining the orientation of the sample to the (horizontal) plane of the laser light sheet. F, G. A z-stack series of digital autofluorescence images (i.e., virtual vertical section planes, parallel to VA) of the SURS-Loc. in the sample is acquired at an appropriate magnification (G). In F, the 3-D reconstruction of the region of the GF sample that contains the SURS-Loc. is shown. G, H. Depending on the applied magnification factor and the examined parameter, one to three images are (systematically) randomly sampled from the z-stack series of digital virtual optical VSP images of the SURS-Loc. (e.g., N°10) of each GF sample for subsequent analysis of SV(SL/GF) (and VV(SL/GF)). For estimation of SV(SL/GF), inclusion of a calibrated size ruler and indication of VA in the VUR image are mandatory. Note that the orientation of the VA of the sample is always recognizable (in the cleared sample, the 3-D reconstruction, the (virtual) VUR section images, and in the SUR sampled fields of view within these VUR sections). Bar = 1 cm in A and = 200 μm in H.
Fig 19.
Estimation of VV(SL/GF) in VUR autofluorescence images acquired by LSFM of optically cleared GF samples.
For practical reasons, VV(SL/GF) and SV(SL/GF) can be estimated in the same images, i.e., using SUR sampled fields of view from (virtual) optical VUR sections. In the presented example of a SUR sampled field of view of a (virtual) GF VUR section, the vertical axis is not indicated. A (virtual) grid of equally spaced test points (crosses) is superimposed on the SUR sampled test field. All points hitting GF section profiles (P(GF)) (including PL section profiles (P(PL), yellow crosses) and SL section profiles (P(SL), red crosses)) are counted [in the presented example: 21 P(PL) and 10 P(SL), i.e., 31 P(GF)]. VV(SL/GF) is calculated from P(SL) and P(GF), using Eq 4. LSFM-autofluorescence image acquired at 520/40 nm (ex) and 585/40 nm (em). Bar = 100 μm.
Fig 20.
Estimation of SV(SL/GF) in VUR autofluorescence images acquired by LSFM of optically cleared GF samples.
Here, a SUR sampled field of view from a (virtual) optical VUR section is shown, generated as described in Section 16.2 and Fig 18. The orientation of the vertical axis (VA, red arrow) and the size ruler are indicated. The VUR images are overlaid with a stereological test system combining cycloids and test points. The short side of the rectangular frame of the system is aligned parallel to the orientation of the VA. All points hitting GF section profiles (P(GF), including PL section profiles (P(PL), yellow crosses) and SL section profiles (P(SL), red crosses)) are counted, as well as all intersections of cycloid arches with the epithelial surface of SL section profiles (I(SL), encircled in red). [In the present example, a test system combining 35 cycloid arches and 70 points is used, refer to Fig 13. 21 P(PL) and 16 P(SL) (i.e., 37 P(GF)), and 41 I(SL) are counted]. SV(SL/GF) is calculated from the sum of intersections (∑I(SL)) and points (∑P(GF)), counted in all examined test fields of all sections of all samples per case, using Eq 6. LSFM-autofluorescence image acquired at 520/40 nm (ex) and 585/40 nm (em). Bar = 100 μm.