The Na, K-ATPase β-Subunit Isoforms Expression in Glioblastoma Multiforme: Moonlighting Roles

Glioblastoma multiforme (GBM) is the most common form of malignant glioma. Recent studies point out that gliomas exploit ion channels and transporters, including Na, K-ATPase, to sustain their singular growth and invasion as they invade the brain parenchyma. Moreover, the different isoforms of the β-subunit of Na, K-ATPase have been implicated in regulating cellular dynamics, particularly during cancer progression. The aim of this study was to determine the Na, K-ATPase β subunit isoform subcellular expression patterns in all cell types responsible for microenvironment heterogeneity of GBM using immunohistochemical analysis. All three isoforms, β1, β2/AMOG (Adhesion Molecule On Glia) and β3, were found to be expressed in GBM samples. Generally, β1 isoform was not expressed by astrocytes, in both primary and secondary GBM, although other cell types (endothelial cells, pericytes, telocytes, macrophages) did express this isoform. β2/AMOG and β3 positive expression was observed in the cytoplasm, membrane and nuclear envelope of astrocytes and GFAP (Glial Fibrillary Acidic Protein) negative cells. Interestingly, differences in isoforms expression have been observed between primary and secondary GBM: in secondary GBM, β2 isoform expression in astrocytes was lower than that observed in primary GBM, while the expression of the β3 subunit was more intense. These changes in β subunit isoforms expression in GBM could be related to a different ionic handling, to a different relationship between astrocyte and neuron (β2/AMOG) and to changes in the moonlighting roles of Na, K-ATPase β subunits as adaptor proteins and transcription factors.

Specific antisera against cell-and function-specific markers (Table 1) allowed us to study the involvement of β Na, K-ATPase subunit isoforms in the oncobiology and microenvironmental heterogeneity of primary and secondary GBM. Specific immunoreactivity is present in virtually all GBM cell kinds, showing a unique expression phenotype of β isoforms involved in the pathogenesis and progression of GBM. Table 1. Markers used in this study.

Na, K-ATPase β1 Isoform Expression in GBM
In GBM samples, the β1 isoform presented no clearly defined pattern of expression. This isoform was detected in some but not all tumor cells. The subcellular location differed among cells within a given area, while some tumor cells were immunopositive for β1 at the cytoplasmic membrane location ( Figure 1A, white arrow) other cells presented immunostaining in a peri-nuclear localization ( Figure 1G). Immunoperoxidase staining on paraffin-embedded tissue sections was used to localize Na, K-ATPase β1 isoform in samples from GBM patients. The immunoreactivity exhibited two distinct patterns: at the edge of the tumor there was strong positive and fibrillary staining ( Figure 2A); β1 positive cells in the center of the section became less frequent and more globular ( Figure 2D, arrows). The interface between the two zones was easily detectable (Figure 2A, arrows). In areas of blood vessel proliferation of, β1 positive pericyte-like cells were observed surrounding the vessels ( Figure 1M), while endothelial cells varied from β1 negative to slightly positive ( Figure 2C,D and Figure 5A, respectively).
In primary GBM, β1 expression in astrocytes was weak or absent ( Figure 1A-C, yellow arrows and white arrow respectively). GFAP (Glial Fibrillary Acidic Protein) negative cells showed a variable β1-specific staining, mainly in plasma membrane and podosome/invadosome-like structures ( Figure 1A-C,G-I, white arrows).
In secondary GBM, ( Figure 1J-L), β1 immunoreactivity was predominantly located in the nuclear envelope, and sometimes, nucleosol of GFAP negative cells (arrows). Most of GFAP positive astrocytes did not show any β1-specific fluorescence signal. A morphologically heterogeneous pattern (multiform) was noted in different areas of the GBM sections, with areas where β1 isoform positive cells appeared mainly fibrillary (Figure 2), areas with β1-positive staining in the nuclear envelope of spherical cells ( Figure 1J) and pericyte-like cells near blood vessels with a light β1-positive staining in the plasma membrane and the cytoplasm ( Figure 1M).   Immunoperoxidase staining for Na, K-ATPase β1 isoform in secondary GBM. (A) Stronger expression in the edge of the section becoming less intense inwards. Arrows point to the interface between the two zones. (B) Enlargement of the interface line. Na, K-ATPase β1 + cells appear mainly fibrillary (B,C,E), but also as rounded cells (arrow in (D)). (C,E) endothelial cells (arrowheads) are β1 − and vessels appear surrounded by β1 + fibers.

Na, K-ATPase β2/AMOG Isoform Expression in GBM
In primary GBM, β2/AMOG signal was mainly located in the plasma membrane and in the cytoplasm at a lesser intensity in GFAP+ astrocytes; in some instances, positive fluorescence was observed in nuclei ( Figure 3A-C, yellow arrow) or in nuclear envelope ( Figure 3A-C, white arrow). Moreover, GFAP negative and β2/AMOG positive cells were observed ( Figure 3A-C, arrowhead), although β2 expression was less intense than that observed in astrocytes and localized mainly in the cytoplasm and in the nuclear envelope. Definitively, β2/AMOG fluorescence signal in astrocytes was higher than that for β1.
In proliferating blood vessels, positive β2/AMOG staining was observed in endothelial cells and in other cells within the perivascular niche ( Figure 3A).
In secondary GBM Na, K-ATPase β2/AMOG isoform-specific labelling was of lower intensity than that observed in primary GBM and GFAP was more intense ( Figure 3D,E).

Na, K-ATPase β2/AMOG Isoform Expression in GBM
In primary GBM, β2/AMOG signal was mainly located in the plasma membrane and in the cytoplasm at a lesser intensity in GFAP+ astrocytes; in some instances, positive fluorescence was observed in nuclei ( Figure 3A-C, yellow arrow) or in nuclear envelope ( Figure 3A-C, white arrow). Moreover, GFAP negative and β2/AMOG positive cells were observed ( Figure 3A-C, arrowhead), although β2 expression was less intense than that observed in astrocytes and localized mainly in the cytoplasm and in the nuclear envelope. Definitively, β2/AMOG fluorescence signal in astrocytes was higher than that for β1.
In proliferating blood vessels, positive β2/AMOG staining was observed in endothelial cells and in other cells within the perivascular niche ( Figure 3A).
In secondary GBM Na, K-ATPase β2/AMOG isoform-specific labelling was of lower intensity than that observed in primary GBM and GFAP was more intense ( Figure 3D,E).
Co-immunolocalization of β2/AMOG and the telocyte marker CD34 (cluster of differentiation 34) was observed in some cells surrounding proliferating blood vessels ( Figure 2G,H); these cells also expressed β3 (Figure 3, panel I).
Co-immunolocalization of β2/AMOG and the telocyte marker CD34 (cluster of differentiation 34) was observed in some cells surrounding proliferating blood vessels ( Figure 2G,H); these cells also expressed β3 (Figure 3, panel I). isoform (green) and GFAP (red) in GBM. (A-C) primary GBM. β2+ staining is mainly located in plasma membrane and less intense in cytoplasm of GFAP+ astrocytes. Some GFAP+ astrocytes show positive fluorescence in nuclei (yellow arrow) or in nuclear envelope (white arrow). In GFAP negative cells, β2/AMOG positive immunolocalization is present in cytoplasm and nuclear envelope (arrowheads). (D-F) In secondary GBM, Na, K-ATPase β2/AMOG isoform-specific labelling is of lower intensity than the observed in primary GBM and GFAP is more intense. (G,H) Positive immunostaining of β2/AMOG isoform in CD34+ (cluster of differentiation 34) telocyte-like cells (arrows). (I) CD34 + /β2 + telocyte-like cells also express β3 isoform, mainly located in the cytosol (β2 is found in cytosol and more intense in plasma membrane and nuclear envelope).

Na, K-ATPase β3 Isoform Expression in GBM
Most GFAP positive astrocytes exhibited β3 subunit positive immunolabelling ( Figure 4A-C, white arrows) in the cytosol, nucleus and nuclear envelope. β3 positive labelling was also observed in giant, spherical or spindle-shaped cells ( Figure 4A-C, yellow arrow).
PCNA + (Proliferating Cell Nuclear Antigen) cells were β3 positive too, mainly in the nuclear envelope, in cytoplasm and in plasma membrane ( Figure 4G-I, arrows). In determined areas of GBM samples, β3-specific fluorescence was brighter at the tumor front decreasing gradually to the interior ( Figure 4G,I); thus, the ratio β3/PCNA in these areas was higher in the front and lower within the tumor ( Figure 4G-I). In other areas, the β3 signal was homogeneous and very intense, both in the periphery and in the interior ( Figure 5E).  (D-F) In secondary GBM, Na, K-ATPase β2/AMOG isoform-specific labelling is of lower intensity than the observed in primary GBM and GFAP is more intense. (G,H) Positive immunostaining of β2/AMOG isoform in CD34+ (cluster of differentiation 34) telocyte-like cells (arrows). (I) CD34 + /β2 + telocyte-like cells also express β3 isoform, mainly located in the cytosol (β2 is found in cytosol and more intense in plasma membrane and nuclear envelope).

Na, K-ATPase β3 Isoform Expression in GBM
Most GFAP positive astrocytes exhibited β3 subunit positive immunolabelling ( Figure 4A-C, white arrows) in the cytosol, nucleus and nuclear envelope. β3 positive labelling was also observed in giant, spherical or spindle-shaped cells ( Figure 4A-C, yellow arrow).
PCNA + (Proliferating Cell Nuclear Antigen) cells were β3 positive too, mainly in the nuclear envelope, in cytoplasm and in plasma membrane ( Figure 4G-I, arrows). In determined areas of GBM samples, β3-specific fluorescence was brighter at the tumor front decreasing gradually to the interior ( Figure 4G,I); thus, the ratio β3/PCNA in these areas was higher in the front and lower within the tumor ( Figure 4G-I). In other areas, the β3 signal was homogeneous and very intense, both in the periphery and in the interior ( Figure 5E).   Figure 5 shows serial sections from the same GBM patient double immunostained for the 3 βsubunit isoforms and the endothelial/macrophage/telocyte marker CD34 or the endothelial/monocyte-derived macrophage/pericyte marker CD31. β1 and β2 positive staining in endothelial cells was mainly located in the cytoplasm and nuclear envelope of endothelial cells ( Figure 5A,C), while β3 was mainly located in the nucleus of such cells, with higher intensity compared to the other isoforms ( Figure 5E). Moreover, a brighter fluorescence signal for the β3 isoform was observed in the peripheral cells surrounding the blood vessel compared to those of β1 and β2 immune staining ( Figure 5).  Figure 5 shows serial sections from the same GBM patient double immunostained for the 3 β-subunit isoforms and the endothelial/macrophage/telocyte marker CD34 or the endothelial/monocyte-derived macrophage/pericyte marker CD31. β1 and β2 positive staining in endothelial cells was mainly located in the cytoplasm and nuclear envelope of endothelial cells ( Figure 5A,C), while β3 was mainly located in the nucleus of such cells, with higher intensity compared to the other isoforms ( Figure 5E). Moreover, a brighter fluorescence signal for the β3 isoform was observed in the peripheral cells surrounding the blood vessel compared to those of β1 and β2 immune staining ( Figure 5).

Discussion
The physiological role of Na, K-ATPase in astrocytes is to remove the excess of K + from the extracellular space after neuronal depolarization. However, transformed astrocytes in GBM harness ion channels and pumps, including Na, K-ATPase, to sustain their singular growth and invasion instead of regulation [58][59][60]. This study focuses on the Na, K-ATPase β subunit isoforms expression to determine their involvement in GBM oncobiology. Table 2 summarizes the cell-and subcellular-specific Na, K-ATPase β subunit isoforms expression in primary and secondary GBM.
The three Na, K-ATPase β subunit isoforms (β1, β2, β3) were detected in both primary and secondary GBM. β1 expression was observed predominantly in the cell membrane and nucleus of GFAP negative cells, β2 in cytoplasm, plasma membrane and nuclei of astrocytes and β3 in the nuclei of astrocytes. In astrocytes of secondary GBM, β3 was also detected in cytosol and plasma membrane. Regarding expression levels, Na, K-ATPase β2 isoform expressed in astrocytes was lower in

Discussion
The physiological role of Na, K-ATPase in astrocytes is to remove the excess of K + from the extracellular space after neuronal depolarization. However, transformed astrocytes in GBM harness ion channels and pumps, including Na, K-ATPase, to sustain their singular growth and invasion instead of regulation [58][59][60]. This study focuses on the Na, K-ATPase β subunit isoforms expression to determine their involvement in GBM oncobiology. Table 2 summarizes the cell-and subcellular-specific Na, K-ATPase β subunit isoforms expression in primary and secondary GBM.
The three Na, K-ATPase β subunit isoforms (β1, β2, β3) were detected in both primary and secondary GBM. β1 expression was observed predominantly in the cell membrane and nucleus of GFAP negative cells, β2 in cytoplasm, plasma membrane and nuclei of astrocytes and β3 in the nuclei of astrocytes. In astrocytes of secondary GBM, β3 was also detected in cytosol and plasma membrane. Regarding expression levels, Na, K-ATPase β2 isoform expressed in astrocytes was lower in secondary GBM compared to primary GBM and β3 isoform expression was more intense in secondary GBM compared to primary GBM. Na, K-ATPase β2 in a healthy brain is mainly expressed in astrocytes [61]. Conversely, other studies reported that in human gliomas β2/AMOG isoform was downregulated in neoplastic cells astrocytes, and this decrease in expression was correlated with increasing tumor grade and cell migration [56,57]. Moreover, we found that GFAP expression in secondary GBM was more intense than that observed in primary GBM, suggesting that astrocytes in primary GBM are less differentiated than those of secondary GBM. With this in mind, it would be logical to assume that primary GBM astrocytes present an equal or lesser expression of β2 than secondary GBM astrocytes, which has a slower progression. However, we found that in primary GBM the β2 isoform expression was more intense than that observed in secondary GBM ( Table 2).
The Na, K-ATPase β1 isoform studied in mammal brains is predominantly expressed in neurons, and negligible in astrocytes [61]. We found in both primary and secondary GBM, that astrocytes did not express β1 or expressed it weakly; this is consistent with the findings in the literature referred to in the Introduction section. We also observed GFAP negative cells that expressed β1 subunit.
Regarding the oligodendrocyte-specific β3 isoform [40,61], we found positive expression in most GBM cells, either GFAP positive or negative. In secondary GBM there were more cells expressing β3 than in primary GBM, which may be due to a decrease in the β2 expression tied to an increase of β3 expression. The presence of the β3 isoform and the low expression of the β1 and β2 isoforms characteristic of astrocytes are consistent with an oligodendrocyte or oligodendrocyte progenitor phenotype [40].
Another important objective of this study was to correlate the mitotic index related to the expression of isoforms by co-localization of those along with PCNA, the clamp subunit of DNA polymerase δ marker of cell proliferation [62,63], and carry out further analysis by confocal microscopy. No correlation was seen between sodium pump isoforms and PCNA protein expression in GBM cells, that is, high expression of PCNA can be found in cells with either high or low expression levels of the different β isoforms, and vice versa.
Other than in gliomas, abnormal expression of Na, K-ATPase β subunit isoforms has been observed in many carcinomas. β1 and β2/AMOG mRNAs are decreased in renal, lung and hepatocellular carcinomas [64], and expression levels of the corresponding proteins are decreased in human clear cell renal cell carcinoma [51] and bladder carcinoma [65]. Previous work from our laboratory [66] reported opposite patterns of β1 isoform expression in gastric and colon adenocarcinomas in a recent study of subcellular expression of all α and β subunits isoforms in colorectal cancer [67]. The level of expression and the location of the β subunit in epithelial cells are important for maintaining their well-differentiated phenotype, which disappears during cancer progression. Further studies suggest that the transcription factor Snail might be repressing the β1 isoform and E-cadherin expression in carcinomas, associating these events to epithelial-mesenchymal transition (EMT) [65].
Variations in β isoforms expression patterns have been described in the regeneration of dorsal root ganglia and sciatic nerve [68], resembling to a certain extent, changes reported in GBM.
With the purpose of finding an β2/AMOG neuronal receptor-protein, the Matchmaker Two-Hybrid system from Clontech was used to pull-out β2/AMOG interacting factors [69,70]. "The two-hybrid assay is a sensitive in vivo method for identifying proteins that interact with protein of interest and is well-suited for detecting weak or transient interactions". Full-length β2/AMOG prevented cell survival, thus, the protein was split into a carboxy-terminal fragment and an amino-terminal fragment and independently used for screening for interacting proteins in a Human Brain Matchmaker cDNA library (Clontech, Mountain View, CA, USA). 2.7 × 10 6 clones were screened. Both carboxy and amino fragments allowed us to identify interacting proteins, including the cytosolic proteins, endoplasmic reticulum proteins and intra-nuclear proteins [71].
Invadosome formation is a key process in tumor progression including cell growth, angiogenesis, invasion and metastasis.
A previous report from our laboratory noted the presence of podosome/invadopodia-like structures in the progression of GBM [72], showing that, with only one exception in the neurons, all described kinds of cells in GBM present podosome/invadopodia-like structures, including GBM-CSC (Cances Stem Cells) and tumor-associated macrophages (TAMs). Here we show evidence ( Figure 1G-I and Figure 2D) of the presence of Na, K-ATPase β1 and β2 isoforms in podosome/invadopodia-like structures in the tumor invasion front, where GBM cells migrate towards the neighboring normal tissue by extending membrane protrusions (invadopodia) containing metalloproteinases (MMPs) [73]. However, a specific role for β isoforms in the invadosome of any non-α-associated β-isoform needs to be investigated. Figure 6, as a graphical summary, shows the possible functional fate of Na, K-ATPase β subunit isoforms based on the results of this study and compared with current research. However, further studies need to be performed in order to precisely define the moonlighting roles of the Na, K-ATPase β isoforms as transcriptional co-activators or transduction signal adaptors and their potential use as biomarkers of specific GBM staging and progression.  [25]. (3)(4) transcriptional regulator as co-activator of RNA polymerase II, embedded either in the nuclear envelope (3) or intranuclear (4). (5) in the cytosol, ready to be disposed to the plasma membrane or, (6) to act as an adaptor of a transduction signal canonical pathway or, 7, constituting a piece of the podosome/invadosome system.

Patients and Tumor Tissue
The study was approved by the Ethical Committee of Nuestra Señora de Candelaria University Hospital (HUNSC); Santa Cruz de Tenerife, Canary Islands, Spain (no. 198/2008, approved on 16 September 2008) and the Ethics Committee of La Laguna University (La Laguna, Canary Islands, Spain). All patients were treated in the HUNSC between years 2007 and 2017 and provided informed consent for the diagnosis and research of tissue specimens before entering the study. Clinical and pathology data were collected from 41 patients, 33 primary GBM (14 males and 19 females) and 8 secondary GBM (6 males and 2 female). GBM samples were taken after initial surgery before patients received radiation or chemotherapy. Paraffin-embedded tissue samples and corresponding clinical data were used ensuring patient's anonymity.

Patients and Tumor Tissue
The study was approved by the Ethical Committee of Nuestra Señora de Candelaria University Hospital (HUNSC); Santa Cruz de Tenerife, Canary Islands, Spain (no. 198/2008, approved on 16 September 2008) and the Ethics Committee of La Laguna University (La Laguna, Canary Islands, Spain). All patients were treated in the HUNSC between years 2007 and 2017 and provided informed consent for the diagnosis and research of tissue specimens before entering the study. Clinical and pathology data were collected from 41 patients, 33 primary GBM (14 males and 19 females) and 8 secondary GBM (6 males and 2 female). GBM samples were taken after initial surgery before patients received radiation or chemotherapy. Paraffin-embedded tissue samples and corresponding clinical data were used ensuring patient's anonymity.

Image Analysis and Statistical Analysis
Tables were compiled by two independent observers that evaluated the specimens blindly. Staining intensities were graded as strong (+++), moderate (++), weak (+) or absent (−). These cut-offs were established by consensus between each investigator. In the cases where the scores were different by more than one unit, the observers re-evaluated the specimens to reach a consensus. In other cases, the means of the scores were calculated.

Immunohistochemistry
Immunoperoxidase staining of 10% formalin-fixed paraffin-embedded tissue sections was performed using an ordinary avidin-biotin method. Briefly, after deparaffinization in xylene and hydration in a graded series of alcohol baths, tissue sections were heated in sodium citrate buffer (pH 6.0) at 120 • C for 10 min in an autoclave to achieve epitope retrieval. Non-specific sites were blocked with 5% non-fat dry milk in Tris buffered saline (TBS) for 1 h at room temperature. To block endogenous biotin, the Avidin/Biotin Blocking kit (#SP-2001, Vector Laboratories Inc., Burlingame, CA, USA) was used according to the manufacturer instruction. Primary antibodies were incubated over night at 4 • C. Endogenous peroxidase activity was blocked by incubating the slides with 3% hydrogen peroxidase in methanol for 15 min. Biotin-conjugated anti-rabbit secondary antibody was incubated for 2 h at 37 • C, and the specific antibody staining was amplified with the ABC Peroxidase Staining kit (Thermo Fisher Scientific, Inc., Waltham, MA, USA). 3,3 -diaminobenzidine substrate concentrate (#IHC-101F; Bethyl Laboratories Inc., Montgomery, TX, USA) was used to visualize immunohistochemical reactions. Samples incubated without primary antibodies were used as a negative control. Slides were counterstained with Harris hematoxylin solution DC (#253949, Panreac Química SLU, Barcelona, Spain) to visualize cell nuclei and mounted with Eukitt mounting medium (#253681, Panreac Química SLU, Barcelona, Spain). An optical light microscope (BX50; Olympus Corporation, Tokyo, Japan) was used to visualize the results of the immunostaining.

Double Immunofluorescence Simultaneous Staining
Immunofluorescent staining of 10% formalin-fixed paraffin-embedded tissue sections was performed as previously described [51]. Briefly, after deparaffinization in xylene and rehydrated in a graded series of alcohol baths, tissue sections were heated in sodium citrate buffer (pH 6.0) at 120 • C for 10 min in an autoclave to achieve epitope retrieval. Nonspecific sites were blocked with 5% bovine serum albumin or normal donkey serum in Tris-buffered saline (TBS) for 1 h at room temperature, tissue sections were then incubated simultaneously with a mixture of two distinct primary antibodies (i.e., rabbit against human target 1, mouse against human target 2) overnight at 4 • C. Slides were then incubated for 1 h at room temperature in the dark with a mixture of two secondary antibodies raised in different species and conjugated to different fluorochromes. Slides were mounted with ProLong ® Diamond Anti-fade Mountant with DAPI (Molecular Probes ® ; Themo Fisher Scientific, Inc., Waltham, MA USA) to visualize cell nuclei. Slides were analyzed using Olympus FV1000 (Olympus Corporation, Tokyo, Japan) and Leica SP8 (Leica Microsystems, Wetzlar, Germany) confocal microscopes.

Conclusions
Assuming that β2/AMOG is involved in physiological astrocyte-neuron adhesion and the changes reported in this study on the quantitative and topological location of Na, K-ATPase β subunit isoforms, we also propose a role for these proteins in the egoist transformation of the astrocyte from its duty as neuron carer to GBM cancer cell.