2.3.1 Sources used for field studies.

The buffered formalin solution used in this study was derived from either of two sources: (1) stock formalin (37% v/v formaldehyde) sold commercially in bottled liquid form in Kampala, Uganda and (2) paraformaldehyde powder (100% formaldehyde) obtained from the University of Kisangani, DRC. These sources of formaldehyde were used to freshly prepare 1 L batches of ca. 4% and 10% buffered formalin (100 ml v/v of source (1) or 100 g w/v of source (2) added to 900 ml of water) in the field using 4 g of sodium phosphate monohydrate (NaH₂PO₄H₂O) and 6.5 g of dibasic sodium phosphate anhydrate (Na₂HPO₄) per liter of formalin. The solution prepared from powdered formaldehyde differed from the liquid commercial-grade formalin in a few important ways: (1) it was mixed without heat-mediated depolymerization, due to a lack of electricity and laboratory facilities; (2) it was not filtered; (3) it contained a final concentration of 10% formaldehyde; and (4) it lacked methanol, a common stabilizer which can affect certain immunological reactions. In order to facilitate semi-quantitative comparisons with laboratory-fixed chameleons (fixed using formalin with a final concentration of 4% formaldehyde), the field-caught chameleons listed in Table 1 (Locations 1 and 2) were all fixed using the diluted liquid stock solution (ca. 4% formaldehyde, final concentration). In contrast, the agamid (Location 3) was the only specimen perfused using the solution prepared from powdered fixative (10% formaldehyde, final concentration). Although the pH values for these solutions were not measured in the field, they were likely near neutral pH, based on the buffering ranges of the salts we used.

2.3.2 Source used for laboratory studies.

The fixative used was prepared from freshly depolymerized and cleared granular p-formaldehyde (Electron Microscopy Sciences Inc., Hatfield, PA; Catalog #19210) as a 4% w/v solution in sodium borate buffer (pH 9.5 at 4°C). First validated by Berod and colleagues [27], this high pH solution is used routinely in our laboratory for locating neural antigens with immunohistochemistry [28–31].

2.4.1 Transcardial perfusions.

2.4.1a: Transcardial perfusions under field conditions: Fig 1 shows details of our field procedures, and Table 3 lists the supplies used to perform them. Lizards were deeply sedated by placing them in closed plastic containers containing two cotton balls saturated with liquid isoflurane. When the animals were sedated enough to remain immobile, they were briefly removed from the container to record body weight and snout–vent length before being returned to the container to complete the sedation. Other more detailed morphometric measurements necessary for biodiversity studies, especially of the head, were also recorded at this time (e.g., head length and width, snout length, etc.). Once fully anesthetized (i.e., no Labyrinthine righting reflex), animals were affixed to a silicone mat by a single pin pierced through each appendage (Fig 1B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Images of field-based perfusion technique.

The field laboratory setup (A); pinned lizard on silicone mat prior to opening of the thoracic cavity (B); injections of solution through opening in apex of heart (C, D); partially dissected and exposed formaldehyde-fixed brains (E, F, G).

https://doi.org/10.1371/journal.pone.0155824.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Materials for field perfusion procedures.

https://doi.org/10.1371/journal.pone.0155824.t003

For perfusion, the lizard’s snout was placed inside a 50-ml conical tube that contained an isoflurane-soaked cotton ball at its base. To open the mediastinum (thoracic cavity), scissors, aided with finer incisions from a scalpel, were used to cut anterior–posterior from mid-neck to lower-abdomen. The thoracic cavity was opened without collapsing the pectoral girdle on the brachiocephalic trunks and carotid arteries. The thoracic wall was removed—the sternum was cut free from the ribs, connective tissue was excised, and portions of the ribs and lungs were removed to further expose the heart, right atrium, and carotid arteries (Fig 1C and 1D). The common carotid artery was gently seized with forceps to elevate the heart from the pericardial cavity and better observe the flow of injected solutions toward the head (Fig 1C and 1D). Lizards were exsanguinated from an incision to the right atrium with fine scissors. Two 3-ml syringes, equipped with 18-gauge needles, were used for successive injections into the apex of the heart (Fig 1C and 1D). The needle tip was inserted carefully into the apex and extended through the ventricle to settle visibly just beyond the base of the common carotid. Saline was injected first, followed by buffered formalin solution. In both cases, due to lack of ice or cold storage, the solutions injected were not cold. The small amount of liquid formalin waste (ca. 2 ml) collected after perfusion was diluted with water to a nonhazardous concentration of <0.1% and disposed of down a drain.

2.4.1b: Transcardial perfusions under laboratory conditions: Control-group animals (Table 2) perfused transcardially under laboratory conditions at UTEP underwent identical procedures to those described above for field perfusions with a few notable differences. First, the formulation and source of formaldehyde used were different than the sources used in the field (see Section 2.3.2). Second, when saline and fixative were successively injected into the animal, both solutions were ice cold. Finally, perfusions were performed in a chemical fume hood.

2.4.5 Nissl staining.

Freely floating sections were rinsed twice in an isotonic Tris-buffered saline (TBS; pH 7.6 at room temperature) to wash out cryoprotectant. Sections were mounted on gelatin-coated slides using a fine-tipped paintbrush. Mounted sections were dried overnight (24 h) at room temperature (ca. 20°C) in a vacuum chamber. They were then dehydrated in ascending ethanol concentrations (50–100%; 3 min each), defatted in xylene, stained in 0.5% w/v thionine solution (thionin acetate, Catalog #T7029; Sigma-Aldrich Corporation, St. Louis, MO) [33, 34], and differentiated in 0.4% anhydrous glacial acetic acid. Slides were coverslipped with DPX mounting medium (Catalog # 06522; Sigma-Aldrich) and stored flat within covered slide trays.

2.4.6 Photomicrography and post-acquisition image processing of Nissl-stained tissues.

Nissl-stained tissues were examined under bright field illumination using a Zeiss M2 AxioImager microscope equipped with an X-Y-Z motorized stage (Carl Zeiss Corporation, Thornwood, NY). Wide field mosaic images of stained histological sections were obtained using a cooled EXi Blue camera (QImaging, Inc., Surrey, British Columbia, Canada) driven by Volocity Software (Version 6.1.1; Perkin-Elmer, Inc., Waltham, MA) installed on the Apple Mac Pro computer. Images were exported from Volocity as lossless TIFF formatted files and imported into Adobe Photoshop (Version CS6; Adobe Systems, Inc., San Jose, CA). In Photoshop, images were cropped via the lasso tool, converted and resampled to 300 dpi gray scale, and brightness- and contrast-adjusted via the curves tool. For hollow spaces within the tissue (e.g., third ventricle), any background illumination of the slide remaining after white balancing was not cropped. Care was taken to make all adjustments judiciously across photomicrographs of field- and lab-processed tissue sections and also to track changes in scale during any size conversion.

2.4.7 Semi-quantitative histological evaluation.

To evaluate the relative efficacies of the lab- and field-based perfusion methods for tissues fixed with 4% formaldehyde (final concentration), the condition of the sections obtained by each procedure was evaluated using a semi-quantitative approach. Three members of the UTEP Systems Neuroscience Laboratory (EMW, AM, and KN), who were blind to the treatment conditions of the tissue, independently rated tissue sections from both treatments using a three-point quality scale: poor (1), good (2), and excellent (3). This rating was performed by viewing the tissue sections under bright field illumination. To understand the effect of perfusion treatment (laboratory vs. field) on various aspects of both tissue and stain quality, independent raters were instructed to apply this scale across six criteria: (A) presence of blood in the tissue; (B) evenness of stain; (C) integrity of tissue at the center of the section; (D) integrity of tissue at the edges of the section; (E) clarity of lamination patterns; and (F) visibility and clarity of nuclei and cell clusters. A total of 204 stained tissue sections were evaluated, representing 166 sections prepared under laboratory conditions and 38 sections prepared under field conditions. The sample size discrepancy between treatments was not believed to influence the overall results because the model-based statistical approach we used (see below) is a function of the size of the largest cluster rather than of the number of clusters (i.e., population-averaged estimates).

2.4.8 Statistical analyses.

A heat map of the scores obtained by our independent observers was generated using R function heat map [35]. Because criteria C–F (see Section 2.4.7) are not independent from criteria A and B (i.e., the criteria are interdependent) and because both fixed effects (field vs. lab conditions) and random effects (order of tissue sampling onto slides, variable sample thickness, and subject selected for perfusion fixation) were present, we employed a variation of a general linear mixed model to analyze the results. Specifically, to best compare the scores between the two fixation methods, a generalized estimating equation (GEE) approach [36] was used to account for the dependence structure among clustered ordinal scores, as implemented in geepack package [37] in R. We tested each variable separately. The analyses were conducted based on the raw data and a reduced dataset using the mode score for each slide from each observer. This data reduction did not influence the results.

2.4.9 Immunohistochemistry.

To evaluate the effect of the perfusion method on the immunoreactivity of neural antigens in the tissue samples, we performed a series of indirect immunohistochemistry experiments using distinct primary antibodies (Table 4) across both field- and laboratory-processed tissue samples preserved using solutions containing 4% formaldehyde. In the first set of experiments, we incubated tissue sections from field- (Trioceros johnstoni) and laboratory-perfused (Rieppeleon kerstenii) chameleons with a rabbit polyclonal antibody targeting the catecholamine-synthesizing enzyme tyrosine hydroxylase (TH). Importantly, the specificity of this TH antibody has been validated in Western blots as recognizing a single, 62 kDa protein band from both amphibian and reptile brain homogenates and shown to be immunogenically identical to that of mammalian TH [38, 39]. In a second set of experiments, field- (Rhampholeon boulengeri) and laboratory-perfused (Rieppeleon kerstenii) chameleon tissue sections were incubated with an antibody targeting neuropeptide Y (NPY). In some initial experiments, we also co-incubated the TH antibody with an antibody targeting dopamine beta-hydroxylase (DβH), and the NPY antibody with an antibody targeting calbindin (Table 4). However, since these antibodies produced little to no specific staining, they were not used in subsequent immunohistochemical runs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Summary of Immunohistochemistry Reagent Combinations Used in This Study.

https://doi.org/10.1371/journal.pone.0155824.t004

Tissues were processed for immunohistochemistry as described previously [28–31]. Briefly, all sections were placed for 1.5 hr in a blocking solution consisting of normal donkey serum (2%; EMD Millipore; Catalog #S30-100ML, Lot #2510142), Triton X-100 (0.1%; Sigma-Aldrich; Catalog #T8532-500ML, Lot #MKBH4307V) and TBS (pH 7.4 at room temperature). With the exception of selected tissue sections where the primary antibody step was omitted (w/o 1°), sections were then removed from blocking solution and incubated in a cocktail of primary antibodies according to the reaction parameters summarized in Table 4. All primary and secondary antibodies were prepared in blocking solution. After five washes in TBS, each for five min (5 × 5), all sections (including w/o 1° were then incubated in a cocktail of secondary antibodies (Table 4). All sections were washed again (5 × 5) in TBS, reacted with fluorophore conjugates also prepared in blocking solution, and counterstained (Table 4). Following another 5 × 5 rinse in TBS, sections were mounted onto Superfrost slides and coverslipped with sodium bicarbonate-buffered glycerol (pH 8.6 at room temperature) and sealed with clear nail polish.

2.4.10 Photomicrography and post-acquisition image processing of immunostained tissues.

Immunostained and w/o 1° sections were visualized with the appropriate filters under epifluorescence illumination using the same microscope and software (and were photographed and images exported in the same manner) as described in Section 2.4.6 for the Nissl-stained sections. Files of both individual and merged channels were imported into Adobe Photoshop, linearly adjusted for brightness and contrast across experiments, cropped, and saved as lossless TIFF-formatted images.

2.5.1 Immersion fixation.

In two instances, during collection efforts at Location 2 (Table 1), animals were deeply sedated and their morphological measurements recorded as described in Section 2.3.1, but were then subjected to immersion fixation rather than transcardial perfusion. Specifically, fully anesthetized animals were manually decapitated in the field between the second and third cervical vertebrae using large surgical scissors. The heads were placed immediately in buffered formalin (bottled liquid fixative containing a final concentration of 4% formaldehyde; i.e., “Source 1” in Section 2.3.1).

2.5.2 Diffusible iodine-based contrast-enhanced computed tomography (DiceCT).

The specimens immersion-fixed in the field (Section 2.5.1) were transferred into aqueous solutions of Lugol's iodine (I₂KI) at the Oklahoma State University Center for Health Sciences [25, 26]. Specimen UTEP 21389 (Rhampholeon boulengeri) (trans-quadratic width (TQW) = 8.1 mm) was fully submerged in a 3.75% (i.e., 1.25% w/v of I₂ and 2.5% w/v of KI) for seven days without replacement (of Lugol’s iodine solution). Specimen UTEP 21388 (Trioceros johnstoni) (TQW = 15.5 mm) was fully submerged in a 5.25% (i.e., 1.75% w/v of I₂ and 3.5% w/v of KI) for 21 days without replacement. Upon immersion in Lugol’s iodine, both specimens were agitated for 60 sec using a vortex mixer to remove air bubbles and ensure that the contrast agent accessed the deepest openings within the specimen (e.g., the endocranial space). Stained specimens were removed from the Lugol's solution, rinsed with tap water to remove excess stain, blotted dry, sealed individually in polyethylene bags to prevent dehydration, and then loaded into plastic mounting units for scanning.

Specimens were μCT-scanned at the Microscopy and Imaging Facility at the American Museum of Natural History (New York, NY), with a 2010 GE phoenix v|tome|x s240 high-resolution microfocus computed tomography system (General Electric, Fairfield, CT). A standard X-ray scout image was obtained prior to scanning to confirm specimen orientation and define the scan volume. The specimen UTEP 21389 was μCT-scanned at 130 kV and 150 μA for approximately 45 min and the specimen UTEP 21388 was μCT-scanned at 140 kV and 180 μA for approximately 90 min. Both specimens were imaged using a 0.1 mm copper filter, a tungsten target, air as the background medium, and 400 ms X-ray exposure timing with 3-times multi-frame averaging. Both were scanned at isometric voxel sizes (at 16.185 and 29.476 μm, respectively), and slices were assembled on an HP z800 workstation (Hewlett-Packard, Palo Alto, CA) running VG Studio Max (Volume Graphics GmbH, Heidelberg, Germany).

4.1.1 Cytoarchitectonics.

We used a semi-quantitative approach to evaluate the cytoarchitecture of the tissue sets processed under laboratory and field conditions of perfusion fixation. A recent survey of semi-quantitative methods used to evaluate histology has found that there is no accepted standard for the types of criteria used for such evaluations [56]. Given the many diverse approaches used to rate the quality of brain tissue, the study recommended that at the very least, investigators should provide a rationale for the specific criteria they use [56]. In line with this recommendation, we note here that the criteria we used were selected on the basis of the goals of our larger experimental research program involving these species, which are to examine the gross neuroanatomical relationships among their gray and white matter structures and the general cytoarchitectonic features of their brain tissue such as aggregations of neurons forming nuclei and laminae. These scales of comparative analysis are informed, in part, by seminal comparative neuroanatomical studies published at the turn of the twentieth century by Ramón [50], Edinger [57], and Brodmann [58]. Our approach necessarily constrains the criteria we use to those listed in Section 2.4.7, guided as we are by a rationale of general histological evaluation at the tissue level rather than its examination at the single-cell or subcellular levels. If the focus were on more fine-grained studies of the morphology or intracellular structure of the cells (e.g., the appearance of neurites, the condition of the organelles), then the criteria chosen using Nissl-based methods would likely be different (e.g., see Chapters X–XIV of Barker [59]), and alternatives to the Nissl method would also have been considered (e.g., see Chapter II in Vol. I of Cajal [60]).

Our statistical analyses show that, on the basis of the two major criteria we used to evaluate fixation efficacy (presence of blood and evenness of stain), no differences in tissue quality were observed between field- and laboratory-perfused specimens. This result supports our qualitative observations of the tissue samples. Further, on the basis of the remaining four criteria (integrity of tissue in center, integrity of tissue at edges, visualization of lamination patterns, visualization of cell clustering and nuclei), the field-perfused tissue apparently displayed significantly better tissue quality than lab-perfused tissue. While these results may be somewhat surprising given the more controllable fixation conditions generally available in a laboratory setting, we must interpret these findings with caution for a few reasons. First, from a qualitative standpoint, it is difficult to separate these four criteria from underlying effects that could be due to other confounding factors, such as tissue damage that was incurred during the mounting of the tissue sections onto glass slides, tissue adherence to the slides during their mechanical transfer through separate reagent reservoirs during the thionin staining procedure, and any differences in tissue stability resulting from variations in section thickness or in the pH and salt composition of the fixative solutions we used. Second, our analysis is limited by the variability we observed among the three independent raters, despite the fact that the mixed model we used to analyze our results mitigates this issue to some extent. Finally, these latter four criteria are interdependent to a large degree on the first two criteria; this interdependence of predictors makes it possible that the statistically significant effects we obtained may be more apparent than real, given that the first two major criteria (which are not dependent, or as dependent, on issues such as mounting and mechanical transfer) show no differences between the two groups.

4.1.2 Immunohistochemistry.

In addition to providing validation of our fixation procedures in the field by evaluating cytoarchitectonic criteria within Nissl-stained tissue sections, we sought indications that our fixation was compatible with standard chemoarchitectural localization methods. In particular, given that the field conditions required prolonged post-fixation in formalin-sucrose, we were concerned about the possibilities of over-fixing the tissue. The duration of formaldehyde fixation may lead to absent or weak binding for some epitopes, preventing effective chemoarchitectural studies with immunohistochemistry [61]. At times, poor penetration of the fixative can occur with too short of an exposure time and excessive cross-linkage can occur from prolonged exposure [62]. In addition, prolonged formalin fixation can cause irreversible damage to some epitopes, but this is largely dependent on the antibody used [63].

For our immunohistochemical procedures, we first aimed to identify dopamine-containing neurons in the periventricular hypothalamus, a well-studied neuronal subpopulation that has been documented previously to be present within the lizard brain [64–66]. Our results demonstrating robust TH-immunoreactivity (-ir) in neurons of field-fixed tissues that is comparable to that observed in laboratory-fixed tissues, confirms the findings of others [64–66] that have characterized this cell population as dopaminergic and extends them by demonstrating that the field conditions of prolonged post-fixation did not prevent chemical identification of these neurons. Relative to the field-fixed sample, the lab-fixed sample displayed apparently elevated levels of TH expression in cell bodies and low levels in fibers. Whether this staining difference indicates a difference in species, fixation efficacy, or peptide transport from the cell bodies to distal neurites between animals, is unclear. We also found comparable labeling of NPY-ir, reported to be present in chameleon brain [67], in cell bodies and/or axonal fibers within field- and lab-fixed tissues. These findings demonstrate the extensibility of our field-based fixation methods to antigens of different types (i.e., those that mark the presence of small neurotransmitters or those that mark neuropeptides). Although immunohistochemical labeling and visualization was feasible using our approach, further research is required to understand the degree of immunohistochemical reactivity across a broader range of antigens within field-perfused brains under our protocol conditions.

4.1.3 DiceCT imaging.

In addition to cytoarchitectonic and immunohistochemical validation of our procedures in the field, diceCT scans from immersion-fixed field-collected samples show that our field protocol is compatible with the non-destructive visualization of gross neuroanatomical features in relation to other soft-tissue structures of the head as well as the skull. Our approach to prepare specimens for diceCT scans, immersion fixation and prolonged storage in fixative followed by iodine-enhancement, was demonstrated to have no negative effects on the contrast and visual quality of the scans. Gray and white matter regions were clearly distinguishable in the soft tissue, further demonstrating that our specimen preparation was successful. Importantly, diceCT can be achieved without encephalectomy, allowing for the interrelationships between central and peripheral components of the nervous system to be preserved. Another advantage of diceCT is that 3-D rendering software can be used to rapidly visualize the high-resolution scans as complex, 3-D soft-tissue anatomy [25, 26]. In turn, these datasets can be analyzed to quantify and compare neuroanatomical structures among different body regions and species [68]. Indeed, the possibility with field-fixed samples to visualize neural circuitry from the cellular level to that of entire brain regions sets up the potential for a comprehensive mapping of brain interconnectedness in three dimensions across multiple scales.

4.1.4 Summary remarks about methodology.

In sum, both our qualitative and semi-quantitative comparisons of field- and laboratory-perfused brain samples indicate that none of several variables that differed between laboratory and field experiments (post-fixation duration, formalin pH and buffer composition, environmental exposure) was sufficient to produce major differences in tissue quality, staining or visualization. Moreover, immersion fixation of samples in the field produced tissue that was readily compatible with non-destructive diceCT imaging. Further studies are needed to determine whether laboratory-based immersion fixation with less environmental exposure produces comparable tissue quality for these species that is amenable to diceCT imaging. We anticipate this to be the case given the evidence provided by other lab-based studies [25, 26].