Samples were collected from 10 subjects who suffered from TBI and underwent decompressive craniectomy, with the removal of the contusion at the Neurosurgical Unit of the Hospital Universitario del Valle (Table 1). The tissue corresponded to the contused cortex from the frontal or temporal lobes. The procedures were approved by the Human Ethics Committee of the Faculty of Health, Universidad del Valle, and Hospital Universitario del Valle, in accordance with the Helsinki criteria.

Control tissue samples were obtained from the cerebral cortex of 3 postmortem subjects; they were provided by the Institute of Legal Medicine and Forensic Sciences of Cali and approved by the Human Ethics Committee of the Faculty of Health, Universidad del Valle. The samples came from postmortem male subjects whose cause of death did not compromise the brain; forensic examination did not show macroscopic signs of brain degeneration (Table 2).

All the brain tissue samples were immersed in 9% saline for 5 min and then fixed with 2% paraformaldehyde in phosphate-buffered saline (PBS) for 7–10 days. After immersion in 10, 20, and 30% sucrose (24 h each), the samples were frozen using 2-metylbutane (Product No. 59070, Sigma-Aldrich, Merck) and stored at −75°C until processing. In total, 30 μm thick sections of tissue were obtained using a Leica Jung Frigocut 2800N cryostat microtome, and stored in antifreeze solution (50% PBS, 30% ethylene glycol, and 20% glycerol) prior to selecting consecutive tissue sections to be processed immunohistochemically by free floating with each of the primary antibodies of interest. Sections were immersed in a solution of 30% methanol (Product No. 179337, Sigma-Aldrich, Merck) and 0.3% hydrogen peroxide (Product No. 31642, HoneyWell-Fluka) for 20 min. The adherence to non-specific antigens was blocked with normal goat serum for CRYAB and normal horse serum for NeuN and GFAP (Vectastain Elite ABC-HRP kits: products No. PK-6101 rabbit and PK-6102 mouse, respectively; Vector Labs), for 40 min. Tissue was incubated overnight in 1:100 diluted primary antibody Anti-phospho-α-B Crystallin Serine 59 produced in rabbit (Product No. C7990, Sigma-Aldrich, Merck). 1:400 diluted primary monoclonal anti-GFAP antibody (G-A-5, mouse; Product No. G3893, Sigma-Aldrich, Merck) was used for GFAP labeling and 1:2,500 diluted mouse antineuronal nuclei (NeuN) monoclonal antibody (Clone A60, Product No. MAB377, Millipore-Merck) was used for NeuN. The tissue was subsequently incubated for 40 min in 1:200 diluted biotinylated secondary antibody and Avidin-HRP solution (Vectastain Elite ABC-HRP kits: products No. PK-6101 rabbit and PK-6102 mouse; Vector Labs). A solution containing 4% diaminobenzidine, hydrogen peroxide, and 2% nickel in PBS (Peroxidase Substrate Kit; 3, 3’-diaminobenzidine (DAB) SK-4100, Vector Labs) was then added for 7 min. Three 5 min washes using PBS were undertaken between each reagent incubation. The tissue was mounted on slides that were dried overnight and then covered with mounting medium (Permount Mounting Media, Product No. SP15-500, Fisher Chemical) and coverslips.

Six tissue sections from three contused subjects were chosen for immunofluorescent free-floating double-labeling, due to the scarcity of available tissue. These sections were immersed in 50 mM ammonium chloride for 30 min to increase the exposure of epitopes. Adhesion of non-specific antigens was blocked for 1 h with bovine serum albumin (BSA) (Product No. 0332-100G, Amresco). Overnight incubation was performed using the primary antibodies, 1:400 diluted anti-GFAP, and 1:100 diluted anti-CRYAB (the same antibodies used for DAB-immunohistochemistry). Both antibodies were incubated simultaneously to obtain double-labeling. Each well was incubated the next day for 2 h with secondary antibodies 1:2,000 diluted goat anti-mouse Alexa Fluor 488 for GFAP (goat antimouse IgG1 cross-adsorbed secondary antibody, molecular probes/Invitrogen, product No. A21121) and 1:2,000 diluted goat antirabbit Alexa Fluor 594 for CRYAB (goat antirabbit IgG secondary antibody, Molecular Probes/Invitrogen, product No. A11012). After the incubation of each reagent, the tissue sections were washed three times for five min with PBS, except after the fluorescent secondary antibodies, when tissue sections were washed eight times for five min with PBS, followed by one wash with distilled water for 2 min. These procedures were conducted in a dark room. Mounted tissues were dried overnight, then covered with mounting medium (Vectashield Hard set, product No. H-1400, Vector Labs). Brain tissue sections from all the control and TBI subjects were subjected to the same standard laboratory conditions.

DAB-immunohistochemistry microphotographs were obtained with a digital camera (AxioCam HRC, Zeiss) coupled to a light microscope (Axio Scope A1, Zeiss). For cell counting, sequential photographs were taken with a 10X/0.25 objective lens, to obtained images of sectors from the pia to the deep subcortical white matter for all the three markers. These sections were then fused using Cannon Utilities Photo Stitch. A cortical column was obtained and placed on a two-dimensional 200 μm × 300 μm grid for counting (SigmaScan Pro program 5, SPSS Science, 2000). Quantitative data were obtained only for the GFAP and CRYAB markers, as DAB-immunoreactivity (IR) for NeuN was only used for the general assessment of the cerebral cortex and to identify the limits between cortical layers, given that these limits were not identifiable using the other two markers. Images of counts in homologous sectors of consecutive sections stained for GFAP and CRYAB were obtained. Four sectors were chosen to count positive cells in every section: (a) supragranular or supragranular layers (SUPRA) (between layers II and III), (b) infragranular or infragranular layers (INFRA) (specifically in layer V), (c) transition between layer VI and subcortical white matter or TRANS, and (d) deep white matter or DWM. In total, we obtained DAB-immunopositive cell counts for 20 control tissue sections: 16 marked with anti-CRYAB (n = 64 counts) and 4 marked with anti-GFAP (n = 16 counts), and for 29 contused tissue sections: 17 marked with anti-CRYAB (n = 68 counts) and 12 marked with anti-GFAP (n = 48 counts).

Images of immunofluorescence-labeled sections were taken with a Zeiss LSM700 Laser Scanning confocal microscope. In total, 22 counts of the proportion of cells co-expressing CRYAB and GFAP were obtained for these sections. Images for counts of astrocytes co-expressing CRYAB and GFAP were obtained with a 20X/0.5 objective lens. The number of cells expressing each marker over the full image extent was determined as well as the ratio of CRYAB/GFAP cells.

DAB-immunohistochemical images were acquired with the same filter and light intensity. Immunofluorescence images were acquired under the same conditions of laser intensity, pinhole size, signal emission, channel mode, higher resolution, scanning speed, master gain, digital gain, and digital offset. These conditions were defined by averaging the lowest and highest expression of the two markers in the contused tissue sections. The counts in the DAB-immunohistochemical and immunofluorescent markings were carried out by two independent observers (one was an expert), using the double-blind method to avoid bias and the data obtained were compared to corroborate those differences were below 5%.

The mean and SD of all the quantitative data were obtained for each parameter. Non-parametric Student’s t-tests were used to compare counts in control and TBI samples as well as for comparisons between counts of positive cells and GFAP/CRYAB. Comparisons between cortical regions were analyzed with a Kruskal–Wallis’s test and Dunn’s post hoc multiple comparison test. All differences were considered significant at p < 0.05. The measures of central tendency and dispersion, statistical analysis, and data plotting were obtained using SPSS 10.0 for Windows and GraphPad Prism 4.

Heterogeneous labeling was observed in the cerebral cortex of subjects with TBI using GFAP and CRYAB. Astrocytic forms were seen predominantly in the transition between the gray matter and the white matter, as well as in the subcortical white matter itself (see Figures 1, 2). In most cases, staining excluded the nucleus and was more intense in the cytoplasm and in the process, allowing the identification of astrocyte morphology (see Figures 2D,H, 3). The intensity of labeling was higher in TBI tissue compared with control tissue, where it was very weak or null. The processes were thickened and there was a noticeable increase in cell size (see Figures 1, 2). Both markers allowed the identification of two astrocytic forms: protoplasmic astrocytes with short and intricate processes, predominantly located in the cerebral cortex layers, and larger fibrous astrocytes with longer processes and fewer branches, located predominantly in the subcortical white matter (see Figure 3). Astrocytes associated with numerous blood vessels and end foot extensions were also visible in the subcortical white matter (see Figures 3A,B).

Using consecutive tissue sections marked with Anti-NeuN (the marker that allowed us to visualize the total population of cortical neurons), we identified the boundaries between cortical layers in sections marked with GFAP and CRYAB. In this way, we were able to carry out a more precise qualitative analysis of GFAP and CRYAB marking and select sectors for quantitative analysis. CRYAB-positive cells were observed on surface layers of contused cortical tissue, but they were more frequent in deep sectors of contused cortical tissue and in subcortical white matter. A few of these cells were distinguishable in layer V as neurons, due to their clearly pyramidal morphology and the orientation of their apical dendrites (see Figures 1G,H, 4). The staining of neurons was very thin. Therefore, staining was heterogeneous and always excluded the nucleus; it was seen mainly in the cytoplasm and the proximal processes of all stained cells.

For each analyzed sector (SUPRA, INFRA, TRANS, and DWM), a significantly higher number of GFAP-positive cells were observed in the tissue of patients with trauma than in controls (SUPRA and INFRA: p < 0.05; TRANS and DWM: p < 0.01). There was little staining of control tissues, with no staining of the supragranular layers, and a progressive increase in staining in the deeper sectors of the cortex, with the highest counts in deep white matter. In injured tissues, infragranular layers had the lowest cell counts, with a very slight increase in supragranular layers, and a very large increase in white matter. There were no significant differences in the counts of the analyzed cortical sections of control tissue between individuals (p > 0.05). The comparison of cortical sectors of damaged tissue resulted in significant differences in cell counts between SUPRA vs. TRANS (p < 0.001), SUPRA vs. deep white matter (DWM) (p < 0.001), INFRA vs. TRANS (p < 0.001) and INFRA vs. DWM (p < 0.001). No significant differences were found between SUPRA vs. INFRA (p > 0.05) and TRANS vs. DWM (p > 0.05). The gray matter sector (SUPRA and INFRA) formed a statistical group that was clearly separate from white substance sectors (TRANS and DWM) (see Figure 5A).

For each analyzed sector (SUPRA, INFRA, TRANS, and DWM) a significantly higher number of CRYAB-positive cells was observed in the injured tissue than in the controls (p < 0.01), with a progressive increase in the number of positive cells in deep cortical layers and white matter. The supragranular layers had the lowest number of cell counts, whereas the deep white matter had the highest number. There were no significant differences in the number of CRYAB-positive cells among sectors of the cerebral cortex in control tissue (p > 0.05). The comparison of cortical sectors of injured tissue resulted in significant differences in cell counts between SUPRA vs. TRANS (p < 0.001), SUPRA vs. DWM (p < 0.001), and INFRA vs. DWM (p < 0.001). No significant differences were found between SUPRA vs. INFRA (p > 0.05), INFRA vs. TRANS (p > 0.05), and TRANS vs. DWM (p > 0.05) (see Figure 5B).

Confocal microscopy allowed the observation of double labeling of CRYAB and GFAP proteins at the cellular and subcellular level in tissues of subjects who suffered TBI. There were clear differences in staining: GFAP labeling was almost always of the cytoplasmic content in all its apparent extension and allowed the observation of the proximal and distal processes, whereas CRYAB labeling allowed, in most cases, the identification of intracytoplasmic inclusions in a dotted manner. On the other hand, fluorescence also showed very few cells with pyramidal morphology positively marked for CRYAB in layer V of the cortex, which corroborates the data obtained with DAB-immunohistochemistry (see Figure 4).

The comparison of the GFAP- and CRYAB-positive cell counts for each cortical region of injured tissue showed no significant differences between immune profiles for the two markers (p > 0.05) (see Figure 5C). We did not quantify the proportion of cells with double and single labeling because we did not have a similar number of sections labeled with immunofluorescence and with DAB-immunohistochemistry. However, we were able to qualitatively observe that the double-labeled cells varied according to the sector of the section analyzed, with colocalization being minimal in the SUPRA sectors and maximal in the TRANS sectors. In the latter, between 35 and 100% of cells colocalized for both markers, and all showed astrocytic morphology (see Figure 4).

Crystallin alpha-B fluorescent labeling showed many “extracellular” red dots (see Figure 4), which was the predominant form of labeling in the SUPRA and INFRA sectors of some of the sections. Some of these red dots corresponded to the somatic labeling of cells located on different focal planes from the microphotographs; other points were clearly located outside the astrocytes. The latter could be located in β-amyloid deposits (Aβ-40 and Aβ-42), which were abundant in sections of contused tissue samples (unpublished data) and have also been reported in Alzheimer’s disease (Narayanan et al., 2006).

The similarity in counts of GFAP- and CRYAB-positive cells, as well as the cellular colocalization of both markers found in this study, allow us to propose CRYAB as a possible alternative histological biomarker of reactive astrogliosis in TBI.

Reactive astroglia are characterized by increased cell size and the production of cytoskeletal intermediate filaments; for this reason, GFAP labeling has been widely used to identify this process in a variety of central nervous system disorders, including TBI (Sofroniew, 2009; Sofroniew and Vinters, 2010; Burda and Sofroniew, 2014; Pekny and Pekna, 2014; Liddelow and Barres, 2017). Axotomy as a consequence of trauma is apparently one of the triggers of reactive astrogliosis even in areas distant from the site of injury (Steward and Trimmer, 1997). In this study, GFAP labeling showed that tissue sections from different subjects could be interpreted as mild to moderate astrogliosis in some cases, or severe diffuse astrogliosis in others, according to the classification proposed by Sofroniew and Vinters (2010). None of the samples from subjects who suffered trauma showed severe reactive astrogliosis characteristics with glial scar formation that could be due to trauma evolution, if glial scars were to be considered a phenomenon that can occur over time.

In this study, the cell morphology and distribution of CRYAB-positive cells were qualitatively similar to those reported in other GFAP studies. No significant differences were found between counts of supragranular and infragranular regions with antibodies against CRYAB and GFAP. This may indicate that the astroglial reactivity process tends to be homogenous in the injured section, unlike the selective vulnerability of neurons according to the cortical layer seen in TBI (Riascos et al., 2013). It should be noted that for both GFAP and CRYAB labeling, expression was always higher in subcortical white matter than in cortical gray matter, possibly because the number of astrocytes was higher in subcortical white matter. Axotomy is an integral part of the pathophysiology of TBI, and the astroglial reactivity observed in the white matter might be due to axonal injury of neurons with cell bodies in layers III or V, or to damage to cortical afferents. We can conclude that GFAP and CRYAB labeling does not correlate with the degree of the neuronal lesion between layers of the same cortical sector, but general labeling of both gray and white matter may be physiologically related to damage to the entire sector.

Traumatic brain injury samples used in this study were obtained from patients 48 to 72 h posttrauma (Table 1). Several studies using animal models have established that after direct tissue damage caused by trauma, a complex progression of the injury develops, known as secondary damage, which presents different characteristics depending on the post-trauma time period (Cernak et al., 2004; Hellewell et al., 2010). Hypotension and hypoxia are present during secondary damage; they are responsible for the increase in noxa. It has been reported that neuronal responses directly due to injury tend to be more intense during the first few hours after trauma. However, at later stages, beginning approximately 12 h after injury, there is increased astroglial involvement and its cellular response to injury; this has been shown by using GFAP as a marker. It is important to clarify that some studies have shown that GFAP is not an exclusive marker of astrocytes, although it is a marker of most astrocytes. The definitive categorization of a given cell as an astrocyte could be achieved through the use of joint markers, as is the case when using GFAP staining with AQP4 (Ikeshima-Kataoka, 2016; Kitchen et al., 2020). Astroglial involvement has also been observed in ischemia (Piao et al., 2005; Li et al., 2012), where early CRYAB neuronal immunolabeling in the cerebral cortex of the rats and CRYAB astroglial labeling in the cortex and subcortical white matter beginning 24 h postinjury were identified.

Glia plays a key role 24 h postinjury in the pathophysiology of trauma and ischemia; their biological activity increases to restore environmental conditions. During secondary damage, the glial network could favor the buffering of ions and glutamate uptake, distributing substances through the network. If this phenomenon persists, the connections system could also behave as another aggravating factor in the process, since the substances that are transported between connections, such as glutamate and calcium ions, among others, could induce death cascades in glia and the release of proinflammatory or toxic substances in sectors that may be focal or exofocal, contributing to the spread of damage (Burda et al., 2016). In our group, a study of cerebral ischemia using animal models showed a significant decrease in the DAB-immunohistochemical expression of glial glutamate transporter GLT1 in layer III of the cerebral cortex contralateral to the ischemic core after occlusion of the middle cerebral artery, which could contribute to the hyperexcitability of sectors that are distant to the ischemic core (Medina et al., 2008). These changes were observed just 24 h postinjury and increased up to 72 h after ischemia, corresponding to the same period analyzed for TBI in this study. These changes are probably due to the deafferentation of commissural fibers that originate in sectors of ischemic focus and that can be correlated with axotomy in TBI.

Double labeling of CRYAB and GFAP proteins in contused tissue using confocal microscopy showed the coexistent overexpression of both markers. This co-expression of the two proteins has also been demonstrated for Rosenthal fibers in Alexander’s disease, chronic glial scars, low- and high-grade fibrillary astrocytoma, etc. (Der et al., 2006; Hagemann et al., 2006; Goplen et al., 2010). Likewise, there are reports of GFAP overexpression in mice, indicating that the accumulation of intermediate filaments upregulates the expression of sHSPs. It has been shown that CRYAB controls the monomer–monomer interactions established between the intermediate filament proteins to prevent their pathological aggregation (Der et al., 2006). This would explain the increased expression of both proteins in trauma and cellular co-localization found in this study, as an attempt of CRYAB to stabilize the characteristic increase in intermediate filaments during the gliosis process to prevent its collapse.

However, other studies have found different functions and cytoplasmic locations of CRYAB. For example, CRYAB has been found to inhibit autoproteolytic maturation of caspase 3; it colocalizes in vivo with the intermediate product of this catalysis, peptide p24 (Kamradt et al., 2001; Shin et al., 2009). On the other hand, some data have indicated that CRYAB accompanies the disassembly of the Golgi apparatus during prometaphase, and its reassembly once cytokinesis is complete, as well as its presence in this membrane complex during interphase (Gangalum et al., 2004). Its function during interphase in this organelle is probably related to its role as a chaperone, considering the activity of the Golgi apparatus as the main site of post-translational processing of proteins and as a targeting station for secreted proteins. den Engelsman et al. (2003) described the interaction of CRYAB with FBX4, a component protein of the SKP1/CUL1/F-box (SCF) ubiquitin ligase protein (SKP1/CUL1/F-box); that is, a protein that is responsible for capturing and signaling target proteins so they undergo a degradation process through the proteasomal pathway (den Engelsman et al., 2003). This interaction is strengthened when CRYAB is phosphorylated at serine 19 and 45, which occurs mainly during the mitotic phase of the cell cycle; it would, therefore, be of greater importance in astrocytes than in neurons in the nervous system.

Few CRYAB-positive cells with pyramidal morphology were observed in layer V of the cerebral cortex of injured samples in this study. Layer V of the cerebral cortex contains large pyramidal neurons that project to other cortical regions and to distant areas of the system, such as the striatal nucleus, brainstem nuclei, and spinal cord. Escobar et al. (2008) found that layers III and V of the cerebral cortex were more likely to show alterations of the microvasculature in human TBI. They also reported lower cell counts and greater loss of IR quality when using NeuN and MAP2 labeling in layer V than in other cortical layers. Pyramidal neurons may be compromised by axotomy because of the kinetics of trauma. The vulnerability of the inner pyramidal layer could induce the expression of proteins related to cell survival as part of the response to injury, which would explain the presence of CRYAB IR in the injured tissue. The only difference between the quantitative CRYAB and GFAP profiles found in this study was that there were no significant differences in cell counts between the infragranular layers of the cortex (INFRA) and the transition between layer VI and white matter (TRANS) for CRYAB; and on the contrary, there were significant differences between these two cortical regions for GFAP. This small difference could be due to the number of CRYAB-positive neuronal cells found in the infragranular layers of the cortex.

Although most studies have indicated that many diseases cause CRYAB overexpression in astrocytes, studies of cerebral ischemia in murine models (Piao et al., 2005; Li et al., 2012) showed an early expression of CRYAB in pyramidal neurons of the cerebral cortex during the first hours after occlusion of the middle cerebral artery and reperfusion. Approximately 24 h after occlusion, there was the induction of CRYAB expression in reactive astrocytes, which was maintained for several days. These findings allow us to suggest that CRYAB expression could occur in neuronal cells during early TBI, considering the relationship between CRYAB and ischemic events at the pathophysiological level. If so, the regulatory mechanisms of expression or the activation of the protein may be different in glia and neurons. However, the data presented in this work correspond to the subacute period of trauma, since contusion extraction was carried out between 48 and 72 h after injury, according to medical criteria for each case, which generally coincides with contusion growth and/or bleeding peaks.