Three different α-subunit isoforms are specified by three different genes in mammals

Although in kidney and most other tissues the major expressed isoform is α1, all three isoforms are expressed in mammalian brains. The three isoforms show about 85% sequence similarity, with the most substantial differences occurring in their N-terminal regions. When expressed in HeLa cells, the isoforms differ substantially in their responses to [Na⁺]_i: half-maximal activations, [Na⁺]_0.5V, of α1 < α2 < α3 in one study [3] and in another, [K⁺]_0.5V of α3 < α2 = α1 [4].

Three β-subunit genes have also been identified, the β1 isoform being most generally expressed [5]. In embryonic brain, the β2 isoform was first identified as adhesion molecule on glia (AMOG) that is transiently expressed on the surface of cerebellar Bergman glia during the differentiation of granule cells. As the brain matures, β2 becomes widely expressed on astrocytes and disappears from most neurons. The specific β-subunit isoform that pairs with an α subunit appears to have little effect on pump parameters, and no preferential association of particular α and β isoforms is demonstrable in cerebellum [6]. A major function of β subunits is to target the Na⁺ pump to the plasma membrane [7]. Different β isoforms may determine different cellular and subcellular localizations [8].

A major fraction of cerebral energy production is required for extrusion of intracellular Na⁺ that enters during excitation and secondary transport

Cation flux during action potentials is two to three orders of magnitude greater than in the resting state. For example, Na⁺ entry and K⁺ efflux from a squid giant axon during a single action potential, which has a duration of ~1 msec, is about 3 × 10⁻¹² mol · cm⁻² membrane. The resting membrane flux in this tissue is 12 × 10⁻¹² mol · cm⁻² · sec⁻¹ (Chap. 6). Therefore, it would take the pump about 0.25 sec to regenerate the flux of one spike at the resting membrane-pump rate. Based on these estimates, in order to maintain a steady state at conduction frequencies ranging from 10 to 100 impulses/sec, the Na⁺ pump rate would have to increase by 2.5 to 25 times its resting level.

It is estimated that 25 to 40% of brain energy utilization may be related to Na,K-ATPase activity. The energy expenditure for biosynthetic processes in mature brain, including osmotic work, protein and lipid synthesis and turnover of neurotransmitters, is relatively small, probably less than 10% of total ATP utilization. Other energy-utilizing processes include axoplasmic transport, Ca²⁺ transport, vesicle recycling and aminophospholipid translocation. While all of these are significant, Na⁺ transport accounts for the largest share of energy flux.

Coupled active transport of Na⁺ and K⁺ results from a cycle of conformational transitions of the transport protein

The ATPase activity that is associated with the Na⁺ pump is actually the sum of Na⁺-dependent phosphorylation of an aspartyl residue at the catalytic site of the pump protein and subsequent K⁺-dependent hydrolysis of the enzyme acylphosphate (Fig. 5-3). These molecular events channel metabolic energy into the pumping process. The initial catalytic-site phosphorylation occurs only after three Na⁺ have bound to ionophoric sites from the cytoplasmic side. In this “E1~P” conformation, the phosphorylation by ATP is readily reversible; that is, the energy state of the protein acylphosphate is comparable to that of the ATP phosphate bond. However, phosphorylation initiates a rapid conformational transition to the E2-P state, from which the Na⁺ is discharged extracellularly. K⁺ then binds to E2-P, initiating hydrolysis of the acylphosphate. This causes E2 to spontaneously revert to E1, carrying two K⁺ into the cell. The cycle is completed as the K⁺ dissociates in concert with initiation of the next cycle by binding ATP.

The Na⁺,K⁺ pump and the cell-membrane potential interact with each other

A stoichiometry of three Na⁺ exchanged for two K⁺ as one ATP is hydrolyzed has been observed in nearly all of the numerous cell types and reconstituted systems that thus far have been studied. Because there is a net outward flow of positive charge, the membrane tends to hyperpolarize. Therefore, the Na⁺ pump is termed electrogenic. The membrane electrical potential contributed by Na⁺ pumping is usually less than 10 mV. This can be sufficient to shorten the duration of the action potential and to contribute to negative afterpotentials. In heart muscle, hyperpolarization due to increased Na⁺ pumping can be observed after sustained increases in firing rate and may be a factor in cardiac arrhythmias. Conversely, the membrane potential modifies the pump rate. Hyperpolarization appears to slow Na⁺ dissociation from the pump [9].

Phosphorylation of the catalytic subunit by cAMP-dependent protein kinase reduces pump activity

This occurs at the serine of the consensus sequence RRNSVF in the C-terminal cytoplasmic loop (e in Fig. 5-2A). This sequence occurs in all three isoforms as well as in the gastric H⁺,K⁺-exchange pump and is highly conserved across species. Inhibitory phosphorylation of Na,K-ATPase appears to occur in some neurons. The α1 isoform can also be phosphorylated by protein kinase C (PKC), but the functional consequences of this are uncertain.

The isoforms of the α-subunit genes differ with respect to regulatory DNA base sequences in their 5′-flanking regions

These may enhance or obstruct the assembly of the preinitiation complex of RNA polymerase and relevant transcription factors [10]. The neuron specificity of α3 expression appears to be related to a neural-restrictive silencer element and a positively acting cis element [11].

The “basal promoter,” required for constitutive expression of the α1 gene, is a positive regulatory element termed ARE [12] and contains binding sites for several transcription factors, including the cAMP-responsive element (CRE). CRE occurs within promoters of many cAMP-inducible genes and can interact with CRE-binding (CREB) and -modulatory (CREM) proteins and activating transcription factor 1 (ATF-1), which are a subgroup within the leucine zipper family defined by their amino acid-sequence similarity and ability to heterodimerize with each other. The CRE site is subject to regulation by several pathways, including some involving cAMP, Ca²⁺ and transforming growth factor β (TGFβ). Phosphorylation of ATF-1 and CREB by either cAMP-dependent protein kinase (PKA) or PKC can enhance their binding to the ATF/CRE site and may be required for transcription of the ATPase α1 gene [13].

Regulation by hormones. In the rat hippocampus, the dentate granular cells express the α1 and α3 isoforms (Fig. 5-4). In adrenalectomized rats, aldosterone selectively regulates α3 in these cells. In other cortical neurons which express both mRNAs, however, aldosterone has no effect on either [14]. In the kidney, aldosterone regulates the level of expression of α1. Regulation by aldosterone is evidently determined by local factors, such as cell type and location, in addition to the presence of the corticoid receptor. These local factors in gene regulation, which may be positive or negative, have been termed the “cell context.”

Figure 5-4

A: Immunocytochemical localization of the Na,K pump in cerebellum. Cerebellar cortex has intensely stained basket regions adjacent to Purkinje cells and little staining along the apical dendrites (double arrows) of Purkinje cells. While the reaction product (more...)

During postnatal development days 1 to 20, α3 mRNA is the isoform predominantly expressed in rat brain and is the one most decreased by thyroidectomy [15]. The genes for the β1 and β2 subunits also contain numerous consensus sequences for transcription factors, including, in the case of β1, potential binding sites for thyroid and glucocorticoid receptors [16].

Na⁺-dependent regulation of Na⁺-pump expression. Alterations in the local ionic environment, probably mainly through changes in cytoplasmic Ca²⁺ and Na⁺, can affect expression of the Na⁺ pump and of early-response genes [17]. There is evidence for a Na⁺-responsive element in the 5′-flanking regions of each of the α-subunit genes [18].

Differential localization of isoforms. The activity of rat brain Na,K-ATPase per milligram of protein increases about ten times during the developmental stage just prior to rapid myelination, which occurs 2 to 12 days postnatally and corresponds to the time of glial proliferation, elaboration of neuronal and glial processes and increasing neuronal excitability. In the nervous system, Na⁺ pumps are concentrated in membrane regions associated with high ionic flux. These include neuronal axon terminals, synapses, perikarya and dendritic processes, as well as glial processes that occur in the neuropil of gray matter (see cerebellum, Fig. 5-4A). Na,K-ATPase is demonstrable in neuritic extensions in cell cultures and at the node of Ranvier but not in myelin wrappings of adult rat optic nerve. Within the retina, heavy concentrations are found in photoreceptor inner segments. In retinal pigment epithelium and choroid plexus ependyma, the Na⁺ pump is concentrated on apical or luminal surfaces (Fig. 5-4B). In contrast, almost all epithelial cells adapted for secretion or reabsorption, such as kidney tubules and exocrine glands, express the Na⁺ pump asymmetrically, on the basolateral or antiluminal surfaces but not on the apical surface.

All three α-subunit isoforms are expressed in the CNS but to varying extents in different regions and cell types. The α1 isoform appears in glia, Schwann cells, choroid plexus epithelium and some neurons. The α2 isoform is found diffusely throughout the CNS in glia, although some neuronal localization is not excluded. The α3 isoform is found only in neurons. In the rat nervous system, α3 mRNA is most abundant within the perikarya of retinal ganglion cells, cortical and hippocampal pyramidal cells, certain interneurons, Purkinje cells, brainstem nuclei, subcortical nuclei and anterior and posterior spinal cord gray columns. The α1 mRNA is most abundant in rat hippocampal and cerebellar granule cell layers, sparse in Purkinje cells and diffusely distributed in neuropil and white matter (Fig. 5-4C). In the spinal cord, α1 expression is restricted to a set of laterally situated anterior horn cells and to intermediolateral thoracic cord cells. Different dorsal root ganglion cells express α3 alone or together with α1 but do not express α1 alone. In contrast to the rat, one study of human cerebellum shows that α1 mRNA is about equal in granule and Purkinje cell layers. Aging-related changes in α-isoform gene expression are discussed in Chapter 30.

In human cerebral cortex (Fig. 5-5), as in rat, α3 mRNA is clustered over pyramidal and other neurons while α1 mRNA is distributed diffusely through the neuropil. In nondemented, aged humans, there are slight increases in α1 mRNA, not reaching statistical significance, while there are significant but small reductions in the neuronal perikaryal content of α3 mRNA. In contrast, in dementing Alzheimer's disease (Chap. 46), the neuronal perikaryal and total neuropil α3 mRNA contents diminish markedly prior to neuronal dystrophic changes, reflecting changes in neuronal function early in the neurodegenerative process (Fig. 5-5).

Figure 5-5

In situ hybridization for α1 and α3 mRNAs of Na,K-ATPase in postmortem human superior frontal cortex (layers IV, V). Top row, α1; left to right: 39-year-old normal, 78-year-old normal and 78-year-old Alzheimer's disease (AD). Note (more...)