|Home | About | Journals | Submit | Contact Us | Français|
Olfactory ensheathing glia (OEG) are distinct from other glia in their developmental origin, presence in both the peripheral and central nervous systems, and highly restricted location. OEG are present only in the olfactory lamina propria, olfactory nerve, and the outer two layers of the olfactory bulb, where they envelop bundles of olfactory sensory neuron axons, in a manner distinct from myelination. Because of their unique properties and their association with the continually generated olfactory sensory neurons, OEG have attracted interest for their potential capacity to support axonal regeneration, for example, after spinal cord injury. However, study of the properties and function of OEG has been hampered by a paucity of neurochemical markers with which to identify and distinguish them definitively from other types of glia. Here we provide evidence through anatomical colocalization studies that OEG express the water channel aquaporin 1 (AQP1), both in vivo and in vitro. We propose that AQP1 expression represents an important distinguishing characteristic of OEG, which may impart unique function to these glia.
The primary sensory neurons of the olfactory system, the olfactory receptor neurons (ORNs), are located in the olfactory epithelium that lines the nasal cavity. The superficial location of the ORNs allows for direct access to odorant molecules, but also exposes these neurons to potential damage from pollutants, microorganisms, and toxic chemicals in the environment. To maintain the integrity of the olfactory sensory apparatus, continual generation of ORNs occurs throughout the life of the animal (Graziadei and Graziadei, 1979). This feature is one of the few examples of ongoing neuronal generation in mature mammals and has been used as a model system to study renewal of neural tissue and reestablishment of functional connections after injury or disease.
From the olfactory epithelium, odorant-related information is carried in the olfactory nerve as it passes through the cribriform plate to the olfactory bulb. The olfactory nerve is composed not only of the axons of ORNs but also of the cell bodies and extensive processes of olfactory ensheathing glia (OEG). This chemically unique type of glia, with characteristics of both astrocytes and Schwann cells (Doucette, 1990), originates from the olfactory placode, not from the ventricular zone or neural crest, which is the origin of other glial cells (Valverde et al., 1992). As demonstrated in ultrastructural morphological studies, OEG envelop or surround large groups of ORN axons along their course within the lamina propria and into the olfactory nerve and glomerular layers of the olfactory bulb (Devon and Doucette, 1992; Franklin et al., 1996; Raisman, 2001).
Because of their unique association, OEG may contribute to the ability of newly generated ORNs to project their axons past the glial limitans and form functional connections within the olfactory bulb. Recent studies have also addressed the use of OEG in studies of regeneration after complete spinal cord transection in adult rats. Indeed, some success in encouraging long-distance regrowth of transected axons and functional recovery following OEG transplantation has been reported (Ramón-Cueto et al. 1998, 2000; Lu et al., 2001, 2002; López-Vales et al., 2006; Kubasak et al., 2008).
The aquaporins (AQPs) comprise an ancient family of integral membrane proteins that function as water-selective channels in animals, plants, and microorganisms. To date, thirteen mammalian aquaporins have been identified (for review, see King et al., 2004). AQP1, the archetypal member of this family, is a constitutively open, bidirectional water channel expressed in red blood cells, renal proximal tubules, capillary endothelium, choroid plexus, and other tissues where water permeability is important to function (Agre et al., 2002). Recently, we also reported abundant expression of AQP1 in unmyelinated primary somatosensory neurons of the mouse (Shields et al., 2007).
In the course of our studies of the nervous system expression of AQP1, we discovered that this protein is also expressed at high levels in the olfactory bulb. Here we detail the expression of AQP1 within the olfactory system, and by comprehensive colocalization experiments demonstrate that the olfactory bulb AQP1 protein does not derive from olfactory receptor neurons, periglomerular cells, or astrocytes. Instead, we show that AQP1 is expressed by OEG, both in peripheral and central olfactory structures, and in olfactory bulb-derived primary cultures. We propose that AQP1 expression represents an important distinguishing characteristic of this unique glial cell.
All animal experiments were approved by the Institutional Animal Care and Use Committees of UCSF and UCLA and were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. AQP1 −/− mice (Ma et al., 1998) were originally provided by Drs. Alan Verkman and Geoffrey Manley (UCSF) and were subsequently produced by interbreeding heterozygotes. Dlx1/Dlx2 and Dlx5 mutant mouse embryos were a gift of Dr. John Rubenstein (UCSF). Genotyping protocols have been described elsewhere (Shields et al., 2007; Qiu et al., 1995; Qiu et al., 1997; Depew et al., 1999).
Adult (20–30 g) male and female mice were deeply anesthetized with 100 mg/kg sodium pentobarbital and perfused transcardially with 0.1M phosphate-buffered saline (PBS), pH 7.4, followed by 10% formalin in PBS or perfused with only phosphate-buffered 4% paraformaldehyde. Brains were collected, post-fixed in the same fixative for 4 hours, cryoprotected overnight in 30% sucrose in 0.1 M PB, embedded in OCT medium (Sakura Finetech USA, Torrance, CA) and frozen before cutting. Mice with Dlx1/Dlx2 and Dlx5 mutations die soon after birth, so tissue was collected from these lines at E18.5 or P0, and immersion-fixed in phosphate-buffered 4% paraformaldehyde. Olfactory bulbs, nerves, and nasal cavities were infiltrated with a series of 5–20% sucrose, incubated in 7.5% gelatin/20% sucrose at 37°C for 1 hr, embedded in plastic molds containing the gelatin mixture, frozen in cold isopentane and stored at −80°C. Tissue was sectioned either on a freezing microtome at 30–40 µm and processed as free-floating sections or at 20–40 µm on a cryostat and processed on slides.
Please see Table 1 for a list of all primary antisera used.
The anti-aquaporin 1 (AQP1) antibody recognized expected bands in Western blot analysis from rat heart, including a 28.8 kDa band and multiple bands of higher molecular weight representing post-translationally modified forms of the protein (manufacturer’s information). Moreover, staining with this antibody was completely absent in AQP1 −/− mice (Shields et al., 2007; this report, Fig 2B).
The anti-neuronal class III β-tubulin antibody recognizes a band of the expected size (50 kDa) on Western blot analysis (manufacturer’s information) and stained a pattern of cellular morphology and distribution in the mouse olfactory nerve that is consistent with previous reports (Akins and Greer, 2005).
OEG were identified with 3 antisera: 1)The anti-p75 nerve growth factor receptor antiserum recognizes a single band at ~75 kDa in Western blots of brain extracts (Huber and Chao,1995), and stained a pattern of cellular morphology and distribution in the mouse olfactory bulb that is consistent with previous reports (Gong et al., 1994; Au et al., 2002). 2) The anti-S100 antiserum recognizes only the expected distinct double peak (S100) in a crossed immunoelectrophoresis assay of human and cow brain extracts (manufacturer’s information), and revealed a pattern of cellular morphology and distribution in the mouse olfactory bulb that is consistent with previous reports (Franceschini and Barnett, 1996; Au et al., 2002). 3) Anti-rat L1 cell adhesion molecule antiserum (gift from Dr. Vance Lemmon, University of Miami) only recognizes proteins with the expected sizes of 220, 180, and 80 kDa immunoprecipitated from mouse brain extracts. This antibody does not bind to Western blots of brain membranes from L1 knockout mice (Nakamura et al., 2010).
The anti-olfactory marker protein (OMP) antibody is widely used as a reliable marker for olfactory receptor neurons and, in our hands, revealed a pattern of cellular morphology and distribution in the mouse olfactory bulb that is identical to that described in a previous report (Monti-Graziadei et al., 1977).
The anti-glial fibrillary acidic protein (GFAP) antibody stains astrocytes and some ependymal cells in the CNS, and staining is absent from skin, connective tissue, lymphatic tissue, muscle, and gastrointestinal tract (manufacturer’s information). In the mouse olfactory bulb, this antibody revealed a pattern of cellular morphology and distribution that is identical to previous reports (Bailey and Shipley, 1993).
The NeuN antibody is widely used as a marker of neuronal nuclei and, in our hands, revealed a pattern of cellular morphology and distribution in the mouse olfactory bulb that is identical to previous reports (Mizrahi et al., 2006).
The anti-tyrosine hydroxylase (TH) antibody recognizes a single band of 60 kDa in Western blot analysis of PC12 rat pheochromocytoma cells (manufacturer’s information) and revealed a pattern of cellular morphology and distribution in the mouse olfactory bulb that is identical to previous reports (Halasz et al., 1977; Mizrahi et al., 2006).
Immunohistochemical experiments were performed as previously described (Shields et al., 2007; Runyan et al., 2005; Runyan and Phelps, 2009). Species-appropriate biotinylated (1:200; Vector, Burlingame, CA) or fluorescent (1:700; Molecular Probes, Eugene, OR) secondary antibodies were used.
To further amplify AQP1 immunoreactivity, we used a Tyramide Signal Amplification kit (NEL704A, PerkinElmer, Boston, MA) following the manufacturer's protocol, but altered reagent concentrations and incubation times. After quenching endogenous peroxidase with 0.3% hydrogen peroxide and 0.1% sodium azide for 30 min, sections were incubated overnight with rabbit anti-AQP1 (1:60,000), washed with TNT buffer (0.1 M Tris-HCl, 0.15 M NaCl, 0.3% Triton X-100, pH 7.5) between steps, incubated in biotinylated goat anti-rabbit IgG (1:1,000; Vector) for one hour, and streptavidin-conjugated horseradish peroxidase (1:150; PerkinElmer) for 30 min. Sections were incubated in fluorophore-labeled tyramide (1:150; PerkinElmer) for 5 min and rinsed thoroughly. Before application of a second rabbit primary antiserum, sections were fixed for 15 min in 4% paraformaldehyde, rinsed and boiled in 10 mM citric acid buffer (pH 6.0) using a 1250W microwave for 3 min on 100% power followed by 2.5 min on 50% power (adapted from Tóth and Mezey, 2007). This step prevents unwanted specific binding of the biotinylated anti-rabbit IgG used to localize the second marker to the rabbit anti-AQP1 primary antibody (Tóth and Mezey, 2007). After the slides cooled for 30 min in citric acid buffer, they were rinsed in TNT buffer before initiating the second immunohistochemical reaction.
To confirm co-localization and exclude the possibility of antiserum cross-reactivity, we conducted control experiments on sections from wild type and AQP1 null mice. Nearby sections were processed for single-labeling to compare the localization pattern and then the double-labeling procedure was conducted, except that the second primary antiserum was eliminated (see Supplementary material).
Primary cultures derived from the outer two layers of mouse olfactory bulbs were prepared using methods adapted from Ramón-Cueto, et al. (2000) and identical to those reported in Runyan and Phelps (2009). Cells were fixed with 4% paraformaldehyde for 15 min and stored at 4°C in phosphate buffer with 0.06% sodium azide until processing.
Digital images were captured using a CCD camera attached to either a Nikon Eclipse or a Zeiss Axioscope 2 microscope. A Zeiss LSM510 confocal microscope was used for co-localization studies. Images were assembled using Adobe Photoshop, adjusting brightness and contrast as appropriate.
We observed AQP1-like immunoreactivity (IR) in sagittal sections through the whole mouse brain. In this view (Fig. 1A), positive immunoreactivity for AQP1 is apparent in the trigeminal nucleus caudalis (the medullary homologue of the spinal cord dorsal horn; see Shields et al., 2007), the choroid plexus (where it has previously been described by Nielsen et al., 1993), the outer surface of the main olfactory bulb (Fig. 1A–E), and within the accessory olfactory bulb (Fig. 1D, inset). We did not detect IR in the brains of mice with a deletion of the AQP1 gene (Fig. 2B and data not shown). Examination of coronal sections at higher magnification shows AQP1 expression in the olfactory nerve layer (ONL) and in the glomerular layer (GL), which comprise the outer two layers of the olfactory bulb (Fig. 1B–C). Often we observed a dense meshwork of tubular-shaped AQP1 immunoreactive profiles within the inner and outer ONL (Fig. 1D–E). AQP1-positive processes in the inner ONL extend into the GL, but never enter individual glomeruli.
We next asked if AQP1 is expressed in the two layers of the olfactory mucosa that line the nasal turbinates, the olfactory epithelium and the lamina propria. The ORN cell bodies and odorant detecting processes are found within the olfactory epithelium, while the ORN axons travel in the lamina propria where they are surrounded by OEG. We examined E18.5 wild type and AQP1 −/− nasal cavities and found that AQP1 is, in fact, robustly expressed throughout the wild type lamina propria (Fig. 2A); the lamina propria of AQP1 null mutants is blank (Fig. 2B). Wild type AQP1-positive cells surround unlabeled circular spaces (Fig. 2C, arrows) within the lamina propria. To determine if these spaces contain ORN axons, we double-labeled nasal cavity sections with AQP1 and the neurofilament marker β-tubulin and found axon bundles closely associated with AQP1-positive structures (Fig. 2D, arrowheads). The AQP1-positive structures that are directly apposed to β-tubulin-positive axons have a morphology consistent with peripheral OEG. Most likely other structures contribute to the heavy AQP1 expression in the lamina propria, including the perineurium, a connective tissue layer made up of epithelial cells joined by tight junctions that forms the blood-nerve barrier (Shantheraveerappa and Bourne, 1962; Thomas, 1963; Matsuzaki et al., 2005; Ablimit et al, 2006). In addition, the isolated AQP1-positive tubular structures in the olfactory epithelium (Fig. 2C, arrowheads) may be the ducts of Bowman’s glands. Together, these results suggest that several populations of peripheral cells express AQP1, but only OEG are also found within the AQP1-positive layers of the olfactory bulb (Fig. 1).
Olfactory ensheathing glia surround axons of ORNs as they course from the lamina propria into the olfactory bulb, and are located throughout the ONL and within the GL (Doucette, 1984, 1993; Pixley, 1992; Ramón-Cueto and Avila, 1997). To determine if the AQP1-expressing processes in the olfactory bulb belong to OEG, we performed colocalization experiments with AQP1 and three OEG markers: the low-affinity nerve growth factor receptor p75, the glial calcium-binding protein S100, and the L1 cell adhesion molecule.
The low-affinity nerve growth factor receptor p75 (p75-NGFR) is an established marker of cultured OEG (reviewed in Ramon-Cueto and Avila, 1998) and is detected early in development (Gong et al., 1994). However, soon after birth p75-NGFR expression is downregulated; only limited levels persist in the outermost ONL and GL of the adult (Gong et al., 1994; Au et al., 2002). We found p75-NGFR immunoreactivity in the outer ONL, where it co-localized with AQP1 immunoreactivity (Fig. 3A1–3, arrows), and also within the glomeruli, a region that never contained AQP1 IR (arrowheads). The limited co-localization of AQP1 and p75-NGFR is consistent with the small number of adult OEG that express p75-NGFR in vivo.
Previous studies demonstrated that OEG within the ONL, GL, and olfactory nerve express S100 (Franceschini and Barnett, 1996; Au et al., 2002), as do astrocytes (Matus and Mugal, 1975; Haan et al., 1982). As shown in Figure 3B1–3, AQP1 colocalizes extensively with S100-labeled processes in the ONL and GL, except for the nuclei (arrowheads), which immunolabel only for S100. To control for antiserum cross-reactivity, we first labeled nearby sections with AQP1 and S100 individually to confirm similar expression patterns (Suppl. Fig 1A–B). We also performed double-labeling in the absence of the second primary antiserum (Suppl. Fig. 1C–D). Results from these experiments confirm that the AQP1 and S100 antisera do not cross react. AQP1 and S100 also colocalize in cross-sections of the olfactory nerve near its entry into the bulb (Fig. 3C1–3). AQP1- and S100-positive processes ensheath the fine olfactory nerve fascicles (between arrows), in the characteristic pattern reported by Field et al. (2003). These findings suggest that OEG express AQP1, but do not differentiate OEG from S100-expressing astrocytes.
L1 CAM is a molecule expressed by both ORN axons (Gong and Shipley, 1996) and OEG (Miragall et al., 1989; Runyan and Phelps, 2009), but not by other CNS glia. Therefore, to confirm that the S100-labeled cells in the ONL are OEG and not astrocytes, we examined the extent of overlap of AQP1 and L1 distribution. AQP1 colocalized with some L1-positive structures in the ONL as well as with processes that surround individual glomeruli (Fig. 3D1–3, arrows). Structures labeled with L1 only are found within the glomeruli, where OEG are excluded, and adjacent to double-labeled structures in the ONL (Fig. 3D2–3, arrowheads). As OEG are the only cell type in the ONL that expresses both L1 and S100, our finding that AQP1 colabels with both of these markers strongly suggests that central OEG are the major source of AQP1 in the olfactory bulb.
OEG can be cultured from the outer two layers of the adult olfactory bulb. Indeed, after culturing, these OEG have been immunopurified and transplanted into the injured spinal cord so as to facilitate functional recovery (Ramón-Cueto et al., 1998, 2000; Lu et al., 2001, 2002; López-Vales et al., 2006; Kubasak et al., 2008). To determine if AQP1 is expressed by OEG in vitro, we prepared primary cultures from adult mouse olfactory bulbs. Figures 4A,B illustrates that a subset of spindle-shaped cells in primary cultures express AQP1. However, many of the cultured cells are marked only by nuclear DAPI labeling. Although cells were immunopositive using both the standard ABC (Fig. 4A) and TSA (Fig. 4B) kits, the TSA protocol increased the intensity of AQP1, even at a 7-fold lower primary dilution. Importantly, spindle-like cells similar to those labeled with AQP1 expressed other OEG markers, including L1 (Fig. 4C) and S100 (Fig. 4D). Based on the morphological similarities in size, shape, and number of labeled cells in wells derived from the same olfactory bulb cultures, we suggest that the AQP1-, S100-, and L1- labeled cells are OEG.
To determine if AQP1 IR in the olfactory bulb derives exclusively from the OEG, we used immunohistochemistry to localize markers of the various neuronal and glial elements that comprise the bulb. We first investigated ORNs, the sensory afferents that transmit olfactory stimuli to the brain. As for AQP1, IR for the ORN marker olfactory marker protein (OMP) is present in a dense band of fibers at the external surface of the bulb (Margolis and Tarnoff, 1973; Monti-Graziadei et al., 1977). However, in contrast to the OMP-positive fibers, which terminate exclusively within the sharply defined borders of glomeruli, AQP1-positive processes course between glomeruli, in a pattern clearly distinct from that of OMP (Fig. 5A–C).
Next, we wanted to exclude definitively olfactory bulb astrocytes as a source of AQP1 IR. As described previously (Bailey and Shipley, 1993), GFAP-positive astrocytes are located in all layers of the olfactory bulb, and are most densely concentrated in the GL. Our immunohistochemical experiments show that AQP1 and GFAP IR co-occur in the GL and ONL, but in patterns that are clearly distinct (Fig. 5D–E). Specifically, AQP1 is located in long, thin processes at the superficial edge of the GL and in the ONL, while intense GFAP immunoreactivity is concentrated in the relatively short processes of stellate-shaped cells located deeper in the bulb. Based on these distinct staining patterns we conclude that AQP1 IR does not derive from olfactory bulb astrocytes.
Periglomerular (PG) cells are a type of local circuit interneuron with small cell bodies that surround the glomeruli. They interact with axons of ORNs as well as with dendrites of mitral/tufted cells (Kosaka et al., 1998). To determine whether PG cells could be a source of the AQP1 IR in the olfactory bulb, we performed immunohistochemical colocalization studies using NeuN (a pan-neuronal marker) and tyrosine hydroxylase (TH), both of which label PG cells (Mullen et al., 1992; Halasz et al., 1977). Figures 6A–F illustrate that neither marker colocalizes with AQP1 IR. We next analyzed tissue from Dlx1/Dlx2 double mutant mice, which have a very specific developmental defect of the olfactory bulb: despite having a normal laminar organization of the bulb, normal axonal projections from the bulb to other regions, and an intact olfactory topographic map, the periglomerular and granule cells of these mice never develop (Anderson et al., 1997; Bulfone et al., 1998). As illustrated in Figures 6G–J, we found AQP1 IR to be intact in Dlx1/Dlx2 mutant mice, compared to their wild type littermates. We conclude that AQP1 is not expressed in PG cells.
In the absence of the Dlx5 homeobox gene, the olfactory bulbs are significantly reduced in size, the nasal cavities are malformed, and ORN axons and OEG cannot cross the cribriform plate to enter the CNS (Long et al., 2003; Levi et al., 2003). Therefore, as a final test to confirm that AQP1-expressing cells in the lamina propria and olfactory bulb are OEG, we compared the distribution of AQP1 immunoreactivity in the olfactory bulbs and nasal cavities of wild type and Dlx5 −/− mice (Long et al., 2003). When we compared AQP1 expression in the E18.5 wild type and Dlx5 mutant nasal cavities, we found that both genotypes contained strong AQP1 IR in the lamina propria (Figs. 7A,C,F,G), despite the significant irregularities of the nasal turbinates of mutants (Fig. 7C). Evidence from β-tubulin/AQP1 double-labeling experiments indicates that ORN axons fasciculate in these mutants and AQP1-positive structures associate with the fascicles (data not shown). In wild type olfactory bulbs, we detected AQP1 IR in the ONL (Figs. 7A,D, arrow), but in Dlx5 mutant bulbs the ONL was absent and only a few isolated immunopositive structures were observed (Fig. 7B,E, arrowhead). The near absence of AQP1 in Dlx5 mutant olfactory bulbs that lack OEG is consistent with our findings that peripheral and central OEG both express AQP1.
In this study, we examined the distribution of the AQP1 water channel in the olfactory bulb and lamina propria of embryonic and adult mice. Based on the colocalization studies, we conclude that AQP1 is expressed in olfactory ensheathing glia and is excluded from olfactory receptor neurons, astrocytes, and periglomerular cells. Expression of AQP1 is, therefore, a characteristic of OEG that distinguishes them from other CNS cell types.
Aquaporins are found throughout the body, in cell populations specialized for rapid transmembrane water fluxes. Here we have demonstrated that AQP1 is concentrated in the outer two layers of the olfactory bulb, an area not previously known to contain cells with enhanced water transport. We also demonstrate that AQP1-expressing cells are OEG, a glial subtype present in both the CNS and PNS and that displays migratory abilities and supports axonal outgrowth. Because AQP1 is not normally found in other CNS-residing glia, these water channels may contribute to the specialized features of OEG.
Of the thirteen mammalian aquaporins known, only AQP4 has been described in CNS glia. Specifically, AQP4 is expressed in astrocytes, predominantly within the perivascular and subpial endfeet (Nielsen et al., 1997), areas that also are enriched for Kir4.1, a potassium channel involved in the clearance of extracellular K+ from active neurons (Nagelhus et al., 1999). Studies indicate that extracellular K+ is taken up by astrocytes and directed into blood vessels or the subarachnoid space through astrocytic endfeet. The resulting osmotic gradient induces water movement that is facilitated by AQP4 (Dietzel et al., 1980; Holthoff and Witte, 2000; Niermann et al., 2001). As a consequence, extracellular K+ is rapidly buffered in active neuropil and ion homeostasis is maintained. Because AQP1 and AQP4 belong to the same subfamily and have similar structural and permeability characteristics, AQP1 may also contribute to the establishment of ion homeostasis in active ORNs. In support of this hypothesis, Gao et al. (2006) showed that AQP1, but not AQP4, is expressed in both myelinating and non-myelinating Schwann cells, which suggested that these two aquaporin channels facilitate similar homeostatic goals, despite their differential expression in the PNS and CNS. That AQP1, but not AQP4, is expressed by OEG may be a reflection of their peripheral embryonic derivation and partial phenotypic similarity to Schwann cells. Therefore, one function of AQP1 in OEG may be to enhance neuronal conduction via ion-water homeostasis.
A second possibility is that AQP1 is related to OEG motility. Interestingly, highly migratory glioma cells express both AQP1 and AQP4 (Endo et al., 1999; Markert et al., 2001; McCoy and Sontheimer, 2007). Results from these studies suggest that the rapid changes in cell volume necessary for tumor cells to invade narrow spaces in tightly packed brain tissue are mediated by aquaporin-dependent water flux. Furthermore, when either AQP1 or AQP4 are reintroduced into isolated, stable glioma cell lines that have lost aquaporin expression, only those lines in which AQP1 expression was reestablished exhibit restored migratory ability (McCoy and Sontheimer, 2007). As OEG exhibit substantial migratory abilities in vitro and during embryonic development, their expression of AQP1 may contribute to this essential function.
OEG migration is guided by lamellipodia at the ends of the leading processes but also by “waves of lamellipodia” along the process shafts (Windus et al., 2007). Time-lapse in vivo and in vitro imaging shows a rapid expansion and retraction of lamellipodia as OEG navigate through their environment and form cell-to-cell contacts (Windus et al., 2007). Other studies report that cultured OEG can rapidly and dramatically change size, shape, and the direction of their migration (van den Pol and Santarelli, 2003). Such activity may involve water channels that facilitate quick changes in cell volume. In support of this idea, McCoy and Sontheimer (2007) found that AQP1 is expressed along lamellipodia at the leading edge of migratory glioma cells. Although more detailed studies are necessary, these observations support the hypothesis that the motility and alterations in size displayed by OEG could be regulated by AQP1-mediated water transport.
Because of the continuous turnover of ORN in the adult, OEG may be required to adjust their volume and the size of their processes. Under normal conditions, OEG form tubular channels of long, thin cytoplasmic processes around large groups of ORN axons. To partially segregate axons into tightly packed clusters, the OEG invest additional thin sheets of cytoplasm inward (Field et al., 2003). After ORN axons undergo retrograde degeneration, they are replaced by new axons. To accommodate these changes, the longitudinal channels formed by OEG are maintained, but the thin cytoplasmic processes dramatically increase in size (Li et al., 2005). This rapid size change is likely related to the active phagocytic response of OEG and may be facilitated by AQP1-dependent water flux. When the phagocytic phase is complete, OEG processes return to their original size (Li et al., 2005), which allows for active recapture and packing of newly generated axons, and again the change in cell volume could be mediated by AQP1.
Transplantation of OEG has been used as a bridge to facilitate regeneration of axons severed by spinal cord injury. Unfortunately, interpreting the results of these studies is limited by the difficulty of locating OEG in vivo, many months after their transplantation. We originally thought that AQP1 might reliably identify OEG following spinal cord injury, but unfortunately AQP1 is also expressed in Schwann cells (Gao et al., 2006), which enter a spinal cord injury site and must be distinguished from OEG. As interest in the reparative potential of OEG increases, it is reasonable to speculate that a better understanding of the function of AQP1 in these cells will provide valuable clues as to how OEG promote axonal repair.
We thank Dr. Steve Runyan for generating the mouse OEG cultures, Tiffany Yap for technical assistance, Drs. John Rubenstein, Jason Long, and Xidao Wang for their generous gifts of tissue, and Dr. John Ngai and members of his laboratory as well as Dr. Almudena Ramón-Cueto for helpful suggestions and advice.
Grant information: This work was supported by NIH grants NS14627, DE08377 and NS54159.