Immunolocalization of Water Channel Proteins AQP1 and AQP4 in Rat Spinal Cord

Experimental Animals

Pathogen-free male Sprague-Dawley rats (270–310 g) were obtained from Charles River Laboratories (Orient Bio, Seongnam, Korea). The animal protocols were approved by the Animal Care and Use Committee of Kyungpook National University, and all animal experiments were conducted according to the guidelines of Kyungpook National University. Rats were housed in controlled temperature rooms (21 ± 1C) with a 12 hr light/dark schedule. Food and water intake were supplied ad libitum.

Perfusion Fixation, Spinal Cord Isolation and Sample Collection

The rats were anesthetized with isoflurane inhalation and perfusion fixation was performed according to the previous study (Gage et al. 2012). The diaphragm was incised along the entire length of the rib cage and a cut on the both sides through the rib cage up to the collarbone was made to allow for lifting the sternum away and providing a clear view of the heart and major vessels. A small incision was made on the posterior end of the left ventricle to pass a perfusion needle and place the tip of it into the ascending aorta. Then, an outlet incision was made in the right atrium and blood was flushed with warm normal saline (37C) for 30 sec, before switching to 4% paraformaldehyde in 0.01 M PBS (pH 7.4) for 3 min. Immediately after perfusion, the spinal cords were collected by hydraulic extraction method proposed by Meikle and Martin (1981). Briefly, a tip of the blunt needle attached to a 10-ml syringe filled with normal saline was placed in the hub of the lumbar canal. Then, after giving a firm and rapid push, the spinal cord popped out onto the working surface with very few histological artifacts. The spinal cord was divided on the basis of anatomic differences into lumbar enlargement representing lumbar part (L1–L5), cervical enlargement representing cervical part (C5–C1) and the thoracic part, which was in turn divided into thoracic lower (T12/13–T6) and thoracic upper (T6–T1) parts.

Immunohistochemistry

To have the desired orientation of the tissue sections, spinal cord segments were put in plastic straws with a direction marker indicating caudal to cephalic direction before placing them in the histological cassettes. The samples were subsequently immersion fixed in 4% paraformaldehyde in 0.01 M PBS (pH 7.4) for 1 hr followed by dehydration in ethanol (70, 96 and 99% each for 2 hr), clearing in xylene overnight, and, finally, embedding in paraffin (Lee et al. 2011). The paraffin-embedded tissues were cut (2-μm thickness) on a rotary microtome (Leica Microsystems, Wetzlar, Germany). The sections were dewaxed and rehydrated. For immunoperoxidase labeling, endogenous peroxidases were blocked by immersion in a solution prepared by dissolving 2.4 ml of 30% H₂O₂ in 240 ml of absolute methanol for 10 min at room temperature. To reveal antigens, sections were put in 1 mmol/l Tris solution (pH 9.0) supplemented with 0.5 mM EGTA and heated using a microwave oven for 10 min. Thereafter, the sections were cooled for 30 min in the same buffer. Nonspecific binding of IgG was precluded by incubating the sections in 50 mM NH₄Cl for 30 min, followed by blocking in PBS supplemented with 1% BSA, 0.05% saponin, and 0.2% gelatin. The sections were incubated overnight at 4C with primary antibodies diluted in PBS supplemented with 0.1% BSA and 0.3% Triton X-100. The following primary antibodies were used: rabbit polyclonal AQP1 (1:1000–1:4000; Terris al. 1996; Kwon et al. 2005), rabbit polyclonal AQP4 (1:1000–1:3000; Terris et al. 1995; Christensen et al. 2006), rabbit polyclonal AQP4 (AB2218, 1:35,000–1:40,0000, EMD Millipore; Billerica, MA;), mouse monoclonal AQP4 (ab9512, 1:100, Abcam; Cambridge, MA), mouse monoclonal rat endothelial cell antigen-1 (RECA-1, ab9774, 1:50, Abcam), rabbit polyclonal transient receptor potential cation channel subfamily V member 4 (TRPV4, AB39260, 1:1000, Abcam), mouse monoclonal glutamine synthase (GS, MAB302, 1:1,000, EMD Millipore), mouse monoclonal glial fibrillary acidic protein (GFAP, MAB3402, 1:1,000, EMD Millipore), mouse monoclonal neurofilament protein 200 kDa (NF200, MAB5262, 1:2000, EMD Millipore), and mouse monoclonal peripherin (MA3-16724, 1:50, Thermo Scientific; Wilmington, DE). After rinsing with PBS supplemented with 0.1% BSA, 0.05% saponin, and 0.2% gelatin for 3×10 min, the sections were incubated in horseradish peroxidase-conjugated secondary antibodies (Dako; Glostrup, Denmark) diluted 1:200 in PBS supplemented with 0.1% BSA and 0.3% Triton X-100, followed by incubation with diaminobenzidine. The microscopy was carried out using a Leica DMRE light microscope. Immunofluorescence labeling was prepared using Alexa fluor secondary antibodies diluted 1:200 (anti-rabbit Alexa fluor 488, anti-mouse Alexa fluor 594) and DAPI (in 1:10,000 solution prepared from 5 mg/ml stock). After a final wash with PBS, the coverslips were mounted with anti-fade agent (#P36930, Life Technologies; Carlsbad, CA). Laser scanning confocal microscopy was carried out (Zeiss LSM 5 EXCITER, Jena, Germany). The specificity of AQP1 and AQP4 antibodies was validated by peptide pre-absorption control experiments in the rat kidney, brain, and spinal cord sections using the peptide against which the antisera was raised (Terris et al. 1996; Terris et al. 1995).

AQP1 Expression

Immunolabeling of AQP1 was of a similar intensity from the cervical region to the lumbar levels of the rat spinal cord. The highest intensity of AQP1 immunolabeling was seen in the dorsal horns as a scattered, spotted pattern, especially in laminae I and II (Fig. 1A and 1B). In laminae III and IV, AQP1 was mostly expressed at the lateral and medial boundaries of the dorsal horns (Fig. 1A). In lamina V, the immunolabeling showed a tangled, filamentous pattern (Fig. 1A and 1C). In other parts of the gray matter, AQP1 expression was seen as dispersed dots along the medial boundary of the dorsal horn up to lamina X and throughout the immediate surroundings of the central canal (Fig. 2D). In white matter, AQP1 immunolabeling was sporadic but present along all levels of the spinal cord and was mostly found close to the glia limitans (Fig. 1D and Fig. 2CI–2CIII).

OPEN IN VIEWER

Figure 1.

Immunoperoxidase microscopy of AQP1 at the thoracic level of rat spinal cord tissue. Expression of AQP1 in the dorsal horn at thoracic level of the spinal cord (A) was strong in laminae I and II (B). (C) Immunolabeling of AQP1 was also observed at the lateral and medial boundaries of the dorsal horns up to lamina V, where the immunolabeling pattern appeared as a tangled filamentous arrangement. (D) In white matter, AQP1 labeling was sparse and found mostly close to the glia limitans. The dorsal orientation of the spinal cord is indicated by encircled arrows.

OPEN IN VIEWER

Figure 2.

Colocalization of AQP1 with neuronal, endothelial and astrocytic markers at the thoracic level of the rat spinal cord. AQP1 was colocalized with peripherin in laminae I and II of the dorsal horn (AI–AIII), with weaker signal intensity up to lamina III. Colocalization between AQP1 and RECA-1 was found likely in the small arterioles and capillaries branching from the arterial vasocorona, which penetrated into the spinal cord core in the dorsal horn of gray matter (BI–BIII) and white matter (CI–CIII). No co-labeling was noticed between AQP1 and GFAP in AQP1-expressed structures around the central canal (D), dorsal horn (E), or white matter (F). The dorsal orientation of the spinal cord is indicated by encircled arrows; colocalization by arrows. The blue signal in the images represents DAPI staining of the nuclei. AQP1, aquaporin-1; GFAP, glial fibrillary acidic protein; PERIPH, peripherin; and RECA, rat endothelial cell antigen-1. Scale bars, 10 µm.

Colocalization analysis demonstrated that AQP1 was associated with peripherin, a marker for unmyelinated neuronal fibers (Fornaro et al. 2008; Goldstein et al. 1991). The intensity of the colocalization was strong in laminae I and II (arrows in Fig. 2AIII). The immunolabeling intensity decreased and was restricted to linear structures in lamina III (data not shown). Although colocalization between AQP1 and peripherin was obvious in laminae I and II, a large fraction of AQP1 labeling did not colocalize with peripherin (Fig. 1AI–1AIII). In white matter, AQP1 was colocalized with RECA-1, an endothelial cell marker (Duijvestijn et al. 1992), mainly in the immediate vicinity of the glia limitans in the white matter (Fig. 2CI–2CIII). Some co-labeled signals were also observed in lamina I and II of the dorsal horn in the gray matter (Fig. 1BI–1BIII). In contrast, fiber-like or filamentous structures labeled by AQP1 in lamina V (Fig. 1C) were not co-labeled with any of the examined antibodies (peripherin, RECA-1, GFAP, and NF200).

AQP4 Expression

Like AQP1 expression, the immunolabeling intensity of AQP4 was similar and unchanged from the cervical to lumbar levels. AQP4 was highly expressed in astrocytes in the white and gray matter. In white mater, AQP4 was expressed in a radial pattern protruding from the gray matter to the glia limitans along astrocytes processes (Fig. 3C, 3E and 3F). In gray matter, AQP4 was expressed in a laminar pattern, where the strongest immunolabeling was seen at the laminae I and II in the dorsal horns and glia limitans (Fig. 3A and 3B). AQP4 labeling was also abundant in lamina IX in the ventral horns, where it was localized to the astrocytes adjacent to capillaries and around motor neurons (Fig. 3C and 3D); it was not, however, observed in the neuronal somata (Fig. 3D). Other areas demonstrating AQP4 expression were in the surroundings of the central canal in lamina X (Fig. 4DI). Blood vessels were encircled by AQP4 labeling (Fig. 3D).

OPEN IN VIEWER

Figure 3.

Immunoperoxidase microscopy of AQP4 at the thoracic level of rat spinal cord tissue. AQP4 was highly expressed along the whole spinal cord. It was most abundant in laminae I and II, and in the glia limitans of the dorsal horn (A and B). In the ventral horn, AQP4 labeling intensity was higher in lamina IX (C and D). Lamina IX consisted of large motor neurons, wherein the AQP4 signal was attributed to circular structures of capillaries and astrocytes processes, but was not found in the neuronal somata (D). In white matter, AQP4 was labeled in astrocytic fiber bundles protruding from the gray matter up to the glia limitans (E and F). The dorsal orientation of the spinal cord is indicated by encircled arrows.

OPEN IN VIEWER

Figure 4.

Colocalization of AQP4 and GFAP at the thoracic level of rat spinal cord tissue. In the dorsal horns, the highest co-expression between AQP4 and GFAP was observed in the glia limitans and laminae I and II (AI–AIII). Weaker colocalization of AQP4 and GFAP was present in the ventral horns, as demonstrated in lamina IX, which consists of large motor neurons (BI–BIII). The colocalization was also found at the circular structures of capillaries highlighted by AQP4 with red spots of GFAP, possibly astrocytic processes (BIII). In white matter, AQP4 was co-labeled with GFAP, especially along the astocytic processes (CI– CIII). In lamina X around the central canal, co-labeling of AQP4 and GFAP was also observed (DI–DIII). The dorsal orientation of the spinal cord is indicated by encircled arrows; colocalization by arrows. The blue signal in the images represents DAPI staining of nuclei. AQP4, aquaporin-4; GFAP, glial fibrillary acidic protein. Scale bars, 10 µm.

Colocalization analysis demonstrated that AQP4 was co-expressed with GFAP at the astrocytes in the laminae I and II in the dorsal horns and glia limitans of the gray matter (Fig. 4AI–4AIII), and in astrocytes in the white matter (Fig. 4CI–4CIII). Less-intense colocalization of AQP4 with GFAP was noticed in laminae IV–VIII in the gray matter, where AQP4 labeling was mostly expressed along the outline of capillaries, co-labeled with GFAP, presumably astrocytic end-feet (data not shown). Moreover, co-labeling of AQP4 and GFAP was also noticed as a linear structure of astrocytic processes, especially in the ventral horn of lamina IX close to large motor neurons (Fig. 4BI–4DIII). Around the central canal in lamina X, the degree of colocalization between AQP4 and GFAP was slightly higher than that in other parts of the gray matter, excluding laminae I and II, and the glia limitans (Fig. 4DI–4DIII).

In contrast, colocalization between AQP4 and glutamine synthase (GS) was rarely seen in the spinal cord (Fig. 5A, 5C, 5E and 5G), indicating that AQP4 was mostly present in astrocytes around the capillaries (i.e., end-feet) and astrocytic processes; GS is predominantly found in the astrocytic cell body (Fig. 5A, 5C and 5E).

OPEN IN VIEWER

Figure 5.

Colocalization of AQP4 and GS or AQP4 and RECA-1 at the thoracic level of rat spinal cord tissue. Colocalization was rarely seen between AQP4 and GS along the whole spinal cord in the dorsal horns (A), ventral horns (C), white matter (E) and around the central canal (G). GS was expressed in the cell body of astrocytes, whereas AQP4 was restricted to the astrocytic plasma membrane around the capillaries or astrocytic processes. No co-labeling was encountered between AQP4 and RECA-1 (B, D, F, and H). RECA-1 immunolabeling was encircled by AQP4 in vascular units in the ventral horns (D), or their labeling patterns were adjacent in laminae I and II (B), as well as in lamina X with the central canal (H) or in the white matter (F). The dorsal orientation of the spinal cord is indicated by encircled arrows. The blue signal in the images represents DAPI staining of nuclei. AQP4, aquaporin-4; GS, glutamine synthase; and RECA, rat endothelial cell antigen-1. Scale bars, 10 µm.

No co-labeling was encountered between AQP4 and RECA-1. RECA-1 immunolabeling was closely encircled by AQP4 in the neurovascular units (Fig. 5D). In white matter, RECA-1 immunolabeling was rarely found and mostly presented close to the glia limitans near the expression of AQP4 (Fig. 5F). In other laminae, such as I and II (Fig. 5B) as well as lamina X with the central canal(Fig. 5H), the immunolabeling signal was more or less adjacent to that of AQP4.

The expression of transient receptor potential cation channel subfamily V member 4 (TRPV4) has been reported previously in the kidney, liver and rat brain (Liedtke et al. 2000; Strotmann et al. 2000; Delany et al. 2001), with remarkable analogous expression to that of AQP4 in the superficial layers of the neocortex and the astrocytic end-feet (Benfenati et al. 2007). Thus, we examined whether immunolocalization of TRPV4 is associated with AQP4 in the spinal cord. The results demonstrated that TRPV4 immunolabeling was similar to the expression pattern of AQP4, but colocalization between the two proteins was not clearly seen. In the gray matter of spinal cord, TRPV4 expression represented a similar distribution to that of AQP4 (TRPV4 in Fig. 6A and 6B vs AQP4 in Fig. 3A and 3B), but it was not present in the glia limitans (Fig. 6A and 6B). In the ventral horns, TRPV4 labeling was found in the immediate vicinity of the plasma membrane of neuronal cells and in the fiber-like structures projecting toward the white matter (Fig. 6C and 6D). Moreover, TRPV4 immunolabeling was seen in the neuronal cell bodies (arrowheads in Fig. 7B). Importantly, no colocalization between TRPV4 and GFAP was found in AQP4-rich structures of laminae I and II in the dorsal horns (Fig. 7A). Furthermore, no clear overlapping between TRPV4 and AQP4 was observed in the laminae of the dorsal horn (Fig. 7C) or ventral horn (Fig. 7E) in the gray matter, white matter (Fig. 7F), at the border between the gray and white matter close to the sulcal artery (Fig. 7GI-GIII) or in other gray matter regions, as presented in lamina VIII (Fig. 7HI–7HIII).

OPEN IN VIEWER

Figure 6.

Immunoperoxidase microscopy of TRPV4 at the thoracic level of rat spinal cord tissue. The labeling of TRPV4 was seen in the dorsal horn of the gray matter, with slightly higher intensity in laminae I and II (A). No labeling of TRPV4 was seen in the glia limitans or in the dorsolateral fasciculus contiguous to lamina I (B). TRPV4 signal was found in the immediate vicinity of the neuronal plasma membrane in the ventral horn and in the fiber-like structures projecting toward the white matter (C and D). The dorsal orientation of the spinal cord is indicated by encircled arrows.

OPEN IN VIEWER

Figure 7.

Colocalization of AQP4 and TRPV4 or TRPV4 and GFAP at the thoracic level of rat spinal cord tissue. No colocalization was noticed between TRPV4 and GFAP in the dorsal horn (A) or in the ventral horn (B). AQP4 was not colocalized with TRPV4 in AQP4-enriched structures in the dorsal horn (C), surroundings of the central canal (D), ventral horn (E) or in the white matter (F). The lack of colocalization between TRPV4 and AQP4 was also demonstrated at the border between the gray matter and white matter close to sulcal artery (GI–GIII) and in lamina VII of the gray matter (HI–HIII). The dorsal orientation of the spinal cord is indicated by encircled arrows; arrowheads in panel B indicate TRPV4 labeling in neuronal cell body. The blue signal in the images represents DAPI staining of nuclei. AQP4, aquaporin-4; GFAP, glial fibrillary acidic protein; and TRPV4, transient receptor potential cation channel subfamily V member 4. Scale bars, 10 µm.

AQP1 and AQP4 Expression

In the spinal cord, the immunolabeling intensity of AQP1 was overwhelmed by that of AQP4. The characteristic patterns of AQP1 and AQP4 labeling were compared. The main structure where AQP1 was abundantly expressed was the dorsal horn near the glia limitans (Fig. 1A and 1B and Fig. 8B), whereas AQP4 was abundantly expressed along the whole glia limitans in the rat spinal cord (Fig. 3A and 3E). AQP1-labeled structures co-labeled with RECA-1 in white matter were presumably penetrating arterioles from pial arteries (Fig. 2CI–2CIII). These arterioles were surrounded by GFAP-expressing astrocytic end-feet (Fig. 2F). In contrast, no colocalization was found between the structures labeled by RECA-1 and AQP4 (Fig. 5F), where AQP4 labeled astrocytes and RECA-1 labeled endothelial cells in arterioles, respectively. In contrary, AQP1 and RECA-1 co-expression was not found in the neurovascular units of the spinal cord (Fig. 8C), whereas the neurovascular units were strongly encircled by AQP4 (Fig. 5D).

OPEN IN VIEWER

Figure 8.

Comparison of AQP1 and AQP4 labeling and peptide pre-absorption control study. AQP1 was immunolabeled in dorsal horn in a scattered spot pattern, especially in laminae I and II, but it was not expressed in the glia limitans of the dorsal horn (GL in panel A, indicated by a dotted line) and was not co-labeled with NF200 (A) or AQP4 (B). No AQP1 labeling was detected along the vascular units in the gray matter labeled with RECA-1 (C), which were tightly encircled by AQP4 (Figure 5D). AQP1 immunolabeling was not detected in vascular structures in the gray matter of the spinal cord (D) or in the neuronal cells of the brain (E), whereas the epithelium of the choroid plexus was abundantly labeled by AQP1 (F). (G–P) Peptide pre-absorption control study. AQP1 and AQP4 antibodies labeled the proximal tubule of the kidney cortex (I) and collecting ducts (K), respectively, whereas no labeling was seen with peptide-pre-absorbed antibodies (J and L). Moreover, AQP1 and AQP4 antibodies labeled the rat spinal cord (H and M) and AQP4 antibodies labeled the rat brain (O), respectively, whereas no labeling was seen with peptide-pre-absorbed antibodies (G, N, and P). AQP1 control staining in the brain was made at approximately the position of Bregma -4.52 mm in (F), whereas (E) represents the parietal association cortex. The dorsal orientation of the spinal cord is indicated by encircled arrows; GL and dotted lines in panel A indicate the glial limitans. The blue signal in the images represents DAPI staining of nuclei. AQP1, aquaporin-1; AQP4, aquaporin-4; NF200, neurofilament protein 200 kDa; and RECA-1, rat endothelial cell antigen-1.

Importantly, no colocalization was seen between AQP1 and NF200, a marker for myelinated neuronal fibers (Groeneweg et al. 1993), in any parts of spinal cord (Fig. 8A), and no AQP1 immunolabeling was seen in the neuronal somata (Fig. 8D). Moreover, AQP1 was not colocalized with GFAP, an astrocyte marker (Groeneweg et al. 1993), at any level of the rat spinal cord (Figure 2D-F).

Specificity of AQP1 and AQP4 Antibodies

Specificity of AQP1 and AQP4 antibodies was confirmed by immunoperoxidase microscopic examination of the rat brain cortex, choroid plexus, kidney, and spinal cord (Fig. 8E–8P). AQP1 labeling was not seen in the rat brain cortex (Fig. 8E), whereas AQP1 was abundantly expressed in the choroid plexus epithelium (Fig. 8F). Moreover, AQP1 was highly expressed in the rat kidney proximal tubules (Fig. 8I) and spinal cord (Fig. 8H), whereas peptide pre-absorbed AQP1 antibodies did not label the proximal tubules (Fig. 8J) or spinal cord (Fig. 8G). AQP4 expression was seen in the kidney collecting duct principal cells (Fig. 8K), whereas the labeling was not seen in the sections labeled by peptide pre-absorbed AQP4 antibodies (Fig. 8L). Moreover, AQP4 labeling was seen in the rat spinal cord (Fig. 8M) and in the rat brain (Fig. 8O), whereas the labeling was abolished by peptide pre-absorbed AQP4 antibodies (Fig. 8N and 8P), indicating the specificity of the AQP1 and AQP4 antibodies.