Renal response to metabolic acidosis: Role of mRNA stabilization

Abstract

The renal response to metabolic acidosis is mediated, in part, by increased expression of the genes encoding key enzymes of glutamine catabolism and various ion transporters that contribute to the increased synthesis and excretion of ammonium ions and the net production and release of bicarbonate ions. The resulting adaptations facilitate the excretion of acid and partially restore systemic acid-base balance. Much of this response may be mediated by selective stabilization of the mRNAs that encode the responsive proteins. For example, the glutaminase mRNA contains a direct repeat of 8-nt AU-sequences that function as a pH-response element (pH-RE). This element is both necessary and sufficient to impart a pH-responsive stabilization to chimeric mRNAs. The pH-RE also binds multiple RNA binding proteins, including ζ-crystallin, AUF1 and HuR. The onset of acidosis initiates an ER-stress response that leads to the formation of cytoplasmic stress granules. ζ-crystallin is transiently recruited to the stress granules and concurrently, HuR is translocated from the nucleus to the cytoplasm. Based upon the cumulative data, a mechanism for the stabilization of selective mRNAs is proposed. This hypothesis suggests multiple experiments that should better define how cells in the kidney sense very slight changes in intracellular pH and mediate this essential adaptive response.

Renal response to metabolic acidosis

Metabolic acidosis is a common clinical condition that is characterized by a decrease in blood pH and bicarbonate concentration and is caused by overproduction of an acid or excessive loss of base (1). To restore acid-base balance, the kidneys initiate a complex set of responses that include increased ammoniagenesis and gluconeogenesis, enhanced acid and ammonium ion excretion, and a net production and release of bicarbonate ions. All of these changes are mediated by a pronounced increase in renal extraction and catabolism of plasma glutamine. During normal acid-base balance, the kidneys extract and metabolize very little of the plasma glutamine. For example, the measured rat renal arterial-venous difference is normally less than 3 percent of the arterial concentration of glutamine (2), whereas approximately 7 percent of the plasma glutamine is extracted by the human kidneys following an overnight fast (3). However, during chronic acidosis, more than one-third of the total plasma glutamine (2) is extracted in a single pass through this organ. The extracted glutamine is catabolized predominantly within the proximal tubule. Since the amount extracted exceeds the fraction filtered by the glomeruli, both apical and basolateral transport must contribute to the cellular uptake of glutamine during acidosis. Utilization of the extracted glutamine requires its additional transport into the mitochondrial matrix (4) via a transporter that is specific for glutamine and asparagine and is inhibited by various thiol reagents (5). Within the mitochondria, glutamine is deamidated by a phosphate-activated glutaminase (GA) and then oxidatively deaminated by glutamate dehydrogenase (GDH) to yield two ammonium ions and α-ketoglutarate. In humans, the resulting α-ketoglutarate is primarily converted to glucose (6). This process utilizes the cataplerotic activity of phosphoenolpyruvate carboxykinase (PEPCK) to convert intermediates of the tricarboxylic acid cycle to phosphoenolpyruvate. This pathway generates 2 H⁺ and 2 HCO₃⁻ ions per mole of α-ketoglutarate. However, the 2 H⁺ ions are subsequently consumed during the conversion of phosphoenolpyruvate to glucose or oxidation to CO₂ and H₂O.

Previous micropuncture studies (7) and assays using microdissected nephron segments (8) established that the preponderance of renal ammoniagenesis occurs within the proximal convoluted tubule. During acidosis, the levels of key enzymes of glutamine metabolism are increased solely within this segment of the nephron (9). Onset of acidosis causes a rapid induction of the PCK1 gene that encodes the cytosolic PEPCK (10). More gradual increases in the levels of mitochondrial GA (11, 12) and GDH (13) also occur within the proximal convoluted tubule. The adaptations in GA and PEPCK levels result from increased rates of synthesis of the proteins (14, 15) that correlate with comparable increases in the levels of their respective mRNAs (16, 17). However, the increase in GA results from the selective stabilization of the GA mRNA (18-20), whereas the initial increase in PEPCK activity results from enhanced transcription of the PCK1 gene (21). The adaptive increases that occur within the proximal convoluted tubule during acidosis may be initiated by a rapid and sustained decrease in intracellular pH (22, 23).

The activities of a number of key transporters are also increased in the proximal convoluted tubule during acidosis. For example, chronic acidosis causes a pronounced increase in SN1 (24), a reversible transporter that couples the Na⁺-dependent uptake of glutamine to the efflux of a H⁺ ion (25, 26). Under normal acid-base conditions, rat renal SN1 is localized primarily to the basolateral membrane of the proximal straight tubule (27) where it may promote a pH-dependent release of glutamine. However, during chronic acidosis, increased expression of the SN1 transporter occurs primarily in the basolateral membrane of the proximal convoluted tubule (24). Given the sustained increase in H⁺ ion concentration within these cells, the increase in the SN1 transporter could contribute to the basolateral uptake of glutamine. The activities of the mitochondrial glutamine transporter (28); NHE3, the apical Na⁺/H⁺ exchanger (29); and NBC1, the basolateral Na⁺/3HCO₃⁻ co-transporter (30), are also increased during acidosis. The increase in NHE3 contributes to the acidification of the fluid in the tubular lumen and the active transport of ammonium ions (31). As a result, increased renal ammoniagenesis provides an expendable cation that facilitates the excretion of titratable acids while conserving sodium and potassium ions. The increased Na⁺/H⁺ exchanger activity also promotes the tubular reabsorption of HCO₃⁻ ions. Activation of NBC1 facilitates the translocation of reabsorbed and of de novo-synthesized HCO₃⁻ ions into the renal venous blood. Thus, the combined adaptations also create a net addition of HCO₃⁻ ions to the blood that contribute to the ability of the kidney to partially restore acid-base balance.

Early micropuncture studies established a strong correlation between the level of ammonium ions in the luminal fluid of the late proximal convoluted tubule and the level of renal ammonium ion excretion (32, 33). However, nearly 80% of the secreted ammonium ions are reabsorbed from the luminal fluid between the proximal tubule and the accessible portion of the distal convoluted tubule. The primary site of ammonium ion reabsorption within the loop of Henle is the thick ascending limb (34). This segment is a site of further acidification of the luminal fluid and of HCO₃⁻ reabsorption (35). Since the developing pH gradient would favor the further trapping of ammonia in the luminal fluid, ammonium ion reabsorption must be an active process. This process is mediated primarily by BSC1/NKCC2, an apical Na⁺-K⁺-2Cl⁻ cotransporter (36). During metabolic acidosis, increased ammonium ion reabsorption (37) results from increased delivery from the proximal tubule and an adaptive increase in the level of the BSC1/NKCC2 transporter (38). This adaptation is not mediated by increased transcription, but is due to selective stabilization of the BSC1/NKCC2 mRNA (39). Chronic metabolic acidosis also causes an adaptive increase in NHE3 mRNA and protein levels in the rat thick ascending limb (40) that may contribute to the increased HCO₃⁻ reabsorption. The combined adaptations in the thick ascending limb generate a corticomedullary concentration gradient that facilitates the final transport of ammonium ions into the lumen of the collecting duct and their excretion in the urine.

Approximately 80% of the excreted ammonium ions are translocated across the cells of the collecting duct (41). During chronic acidosis, there is a marked increase in acid secretion in the collecting duct that is paralleled by a corresponding increase in ammonium ion secretion (7). A second Na⁺-K⁺-2Cl⁻ cotransporter, BSC2/NKCC1, is localized to the basolateral membranes of the α-intercalated cells of the outer and initial inner medullary collecting ducts of the rat (42) and the terminal portion of the inner medulary collecting ducts of the mouse (43). Increased expression of BSC2/NKCC1 during chronic acidosis may contribute to the basolateral uptake of ammonium ions (44). In rat kidney, the ammonia transporter, RhBG, is also expressed in the basolateral membrane of the distal convoluted tubule, the connecting segment and the collecting duct (45). RhCG, a second ammonia transporter, has an identical distribution along the rat nephron except that it is localized to the apical membrane (46). Within the collecting duct, expression of RhBG and RhCG is greater in the α-intercalated cells than in the principal cells (46, 47). Because the former cells are the primary sites of H⁺ ion secretion within the collecting duct, the disequilibrium decrease in luminal pH would be greatest in the region adjacent to the apical membrane of the α-intercalated cells. Therefore, the high levels of the two ammonia transporters in these cells would facilitate the rapid transport of ammonia and the subsequent luminal trapping of ammonium ions (48).

The cellular content and localization of RhBG are unaltered during chronic acidosis (49). Furthermore, the selective knockout of the RhBG gene had no apparent effect on systemic ammonium ion metabolism or acid-base balance (50). In addition, the knockout mice responded to a chronic acid challenge with an appropriate increase in ammonium ion excretion. Thus, it was concluded that RhBG is not essential for the physiologic transport of ammonia (51). Recent immunostaining (52) and immunogold-labeling (49) experiments indicate that the levels of RhCG are selectively increased in the apical membrane during chronic acidosis. The observed increases occur without a corresponding increase in RhCG mRNA and may result from increased apical targeting and the selective expansion of the apical membrane in the intercalated cells of the medullary collecting duct. Thus, additional knockout experiments are necessary to determine if the two Rh proteins are functionally redundant or if only RhCG is essential for the physiological transport of ammonia. A more detailed description of the renal adaptations in ammonium ion synthesis and excretion can be found in recent review articles (9. 41).

Proteomic analysis

A proteomic approach was used to identify additional proteins that exhibit altered expression in rat renal proximal tubules during metabolic acidosis and to assess the role of increased mRNA stability (53). Proximal tubules were highly purified from control and acidotic rats by collagenase digestion and Percoll density gradient centrifugation. Difference gel electrophoresis and MALDI-TOF/TOF mass spectrometry were used to quantify the changes in proteins that exhibit enhanced or reduced expression. This analysis confirmed the well-characterized adaptive responses in GA, GDH, and PEPCK and identified 17 previously unrecognized proteins that are increased with ratios of 1.5 to 5.6 and 16 proteins that are decreased with ratios of 0.67 to 0.03 when comparing tubules from 7-d acidotic versus control rats. Temporal studies using proximal tubules isolated from normal and 1-d, 3-d, and 7-d acidotic rats confirmed that PEPCK is maximally induced (7-fold) within 1 d. In contrast, GA and GDH increased gradually and reached an 8-fold and 3-fold maximal induction, respectively, after 7 d. The temporal studies also identified 6 additional proteins that exhibit induction profiles similar to GA and GDH. All of the mRNAs that encode the latter proteins and the PEPCK mRNA contain an AU-sequence that is highly homologous to the pH-response element (pHRE) found in the GA mRNA (19, 20). Recent studies have established that a pH-responsive stabilization contributes to the sustained induction of the PEPCK mRNA during chronic acidosis (unpublished data of S. Hajarnis, L. Taylor and N. P. Curthoys). Thus, selective mRNA stabilization may be the predominant mechanism by which protein expression is increased in response to acidosis.

Stabilization of GA mRNA

The rapid degradation of mammalian mRNAs is initiated by the binding of specific protein(s) to unique element(s) that are usually located within the 3′-untranslated region (3′-UTR) of the mRNA (54, 55). Various mRNA binding proteins, such as AU-factor 1 (AUF1) (56), recruit PARN, a poly(A)-specific ribonuclease (57), and the exosome, a complex of 3′→ 5′ exonucleases (58), to remove the poly(A) tail and to accomplish the rapid degradation of the deadenylated mRNA. Alternatively, the deadenylated mRNA may undergo decapping and 5′→ 3′ exonuclealytic degradation within a cytoplasmic loci, termed processing bodies (59). This subcellular structure contains Dcp1/Dcp2, the decapping enzyme, and Xrn1, a 5′→ 3′ exonuclease and therefore constitutes the primary site of 5′→ 3′ degradation. A third mechanism of mRNA turnover is initiated by sequence-specific endonucleolytic cleavage that generates sites for rapid exonucleolytic degradation (60). Selective stabilization of the latter class of mRNAs is mediated by sequence-specific binding of a protein that inhibits the endonucleolytic cleavage.

Characterization of the turnover of GA mRNA has become the paradigm for determining the mechanism by which mRNAs are stabilized in response to metabolic acidosis. The presence of a sequence element that regulates the turnover of the GA mRNA was initially demonstrated by stable expression of various β-globin (βG) reporter mRNAs (18) in LLC-PK₁-F⁺ cells, a pH-responsive line of porcine proximal tubule-like cells (61). The parent βG construct produced a stable mRNA (t_½ > 30 h) that was not affected by transfer of the cells to an acidic medium (pH 6.9, 10 mM HCO₃⁻). By contrast, the βG-GA mRNA, that includes a 956-bp segment from the 3′-UTR of rat GA mRNA exhibited a more rapid turnover (t_½ = 4.6 h). Transfer of cells expressing this construct to acidic medium resulted in a pronounced stabilization of the βG-GA mRNA. Therefore, this segment of the GA mRNA contains a pHRE.

RNA electrophoretic mobility shift assays indicated that cytosolic extracts of rat renal cortex contain a protein that binds to the 3′-UTR of the GA mRNA (19). The observed binding was mapped by deletion analysis to a 29-nt fragment that contains a direct repeat of two 8-nt AU sequences. The binding interaction was reduced significantly by mutating either 8-nt element and was completely lost by mutating both elements. The binding was effectively competed by an excess of the same RNA, but not by adjacent or unrelated RNAs. The functional significance of the observed binding site was further characterized by measuring the effect of acidic medium on the half-lives of various chimeric βG-GA constructs (20). Insertion of short segments of the GA mRNA containing the direct repeat or a single 8-nt AU-sequence was sufficient to impart a 5-fold pH-responsive stabilization to the chimeric mRNA. Therefore, a single 8-nt AU-sequence can function as a pHRE.

A tetracycline-responsive promoter system was developed in LLC-PK₁-F⁺ cells to perform a more accurate pulse-chase analysis of the half-life of the βG-GA mRNA and to characterize the potential role of deadenylation in its degradation (62). This approach accomplishes the rapid induction and shut-off of the synthesis of a single mRNA and avoids indirect effects that are caused by the use of a general transcription inhibitor (63). With this approach, the measured half-life of the βG-GA mRNA was 2.9 h in LLC-PK₁-F⁺ cells maintained in normal medium, whereas transfer of the cells to an acidic medium caused a pronounced stabilization of the chimeric mRNA. RNase H cleavage and Northern analysis of the 3′-ends established that rapid deadenylation occurred concomitant with the rapid decay of the βG-GA mRNA in cells grown in normal medium. Stabilization of the βG-GA mRNA in cells treated with acidic medium was associated with a pronounced decrease in the rate and the extent of deadenylation. The data indicate that deadenylation is the initial and potentially rate-limiting step in the turnover of the GA mRNA. Mutation of the pHRE within the βG-GA mRNA blocked the pH-responsive stabilization observed when the cells are transferred to acidic medium, but not the rapid degradation when cells are maintained in normal medium. Therefore, the 3′-UTR of the GA mRNA must contain additional instability elements. Insertion of only a 29-bp segment containing the pHRE into the βG mRNA was sufficient to produce both rapid degradation and pH-responsive stabilization (62). Therefore, the identified pHRE contributes to the rapid turnover of the GA mRNA and is both necessary and sufficient to mediate its pH-responsive stabilization.

A biotinylated oligoribonucleotide containing the pHRE was used as an affinity ligand to purify potential pHRE binding proteins from a cytosolic extract of rat renal cortex (64). The purified binding activity retained the same specific binding properties as observed with crude extracts and correlated with the elution of a 36-kDa protein. Microsequencing of the purified 36-kDa protein by mass spectroscopy identified 11 peptide sequences (~ 50% sequence coverage) that are identical to tryptic peptides derived from rat ζ-crystallin/NADPH:quinone reductase (ζ-cryst). The identity of this protein as ζ-cryst was confirmed by western blot analysis. A second protein purified by this protocol was identified as the T-cell restricted intracellular antigen-related protein (TIAR). However, the purified TIAR neither bound nor affected the binding of ζ-cryst to the pHRE. Furthermore, specific antibodies to ζ-cryst, but not TIAR, blocked the formation of the complex between the pHRE and either the crude cytosolic extract or the purified protein. Thus, ζ-cryst is the primary pHRE binding protein in rat kidney cortex.

During chronic metabolic acidosis, an adaptive increase in rat renal GDH also contributes to the sustained increase in ammoniagenesis (9, 13). RNA electrophoretic mobility shift assays established that purified ζ-cryst also binds to the full-length 3′-UTR of the GDH mRNA (65). This region contains four 8-nt sequences that are 88% identical to one of the two pHREs present in the GA mRNA. Direct binding assays indicated that the four individual GDH sequences bind ζ-cryst with different affinities. Insertion of the 3′-UTR of the GDH cDNA into the βG expression vector produced a chimeric mRNA that was stabilized when LLC-PK₁-F⁺ cells were transferred to acidic medium. A pH-responsive stabilization was also observed using a βG construct that contained only a single GDH element. Therefore during acidosis, a pH-responsive stabilization of the GDH and NKCC2 mRNAs may be accomplished by the same mechanism that affects an increase in the GA mRNA.

To test the functional significance of ζ-cryst binding, adenoviruses were produced to over express mouse ζ-cryst or a siRNA that is specific for the porcine ζ-cryst. Experiments using the adenovirus to knockdown ζ-cryst expression in LLC-PK₁-F⁺ cells failed to produce a significant effect on the normal turnover or the pH-responsive stabilization of the βG-GA mRNA (Fig. 1). The level of virus used in these experiments produced an 85% knockdown of ζ-cryst protein. These data suggest that the binding of ζ-cryst may not be the rate determining event that mediates the pH-responsive stabilization of the GA mRNA. Alternatively, the endogenous ζ-cryst is present in such large excess that covalent modification of a small fraction of the normal level may be sufficient to mediate the turnover of the GA mRNA. Thus, higher levels of recombinant adenovirus should be tested to determine if a greater knockdown of ζ-cryst can be achieved and what effect this has on the turnover of the GA mRNA.

Fig. 1. siRNA knockdown of ζ-cryst does not affect the pH-responsive stabilization of βG-GA mRNA.

LLC-PK₁-F⁺ cells that express the βG-GA mRNA were infected with a control adenovirus that encodes green fluorescence protein (GFP) or an adenovirus that encodes an siRNA specific for porcine ζ-cryst. After 72 h, the cells were transferred to normal (pH 7.4) (Panel A) or acidic (pH 6.9) medium (Panel B) containing 50 ng/ml Dox for 10 h. The cells were washed twice with PBS and incubated in medium minus Dox for 3 h to create a transcriptional pulse. Fresh medium containing 1 μg/ml Dox was added and RNA was isolated at 0, 3, 6, and 9 h after Dox addition. Samples from uninfected cells were also harvested for half-life analysis. The level of βG-GA mRNA was quantified by Northern analysis and divided by the corresponding level of 18S rRNA to correct for errors in sample loading. The log of the normalized data was plotted versus time of plus Dox treatment. The reported values are the mean ± SE of data obtained from 3 separate determinations. The lines represent a least squares fit of the data points as determined by KaleidaGraph.

RNA gel shift assays demonstrated that the recombinant p40 AUF1 also binds to the pHRE of the GA mRNA with high affinity and specificity (62). AUF1 is a known RNA binding protein that interacts with the 3′-UTRs of various unstable mRNAs including c-myc (66), TNFα (67), GM-CSF (68), and COX-2 (69). AUF1 was also identified as hnRNP D (70), a protein that shuttles between the nucleus and the cytoplasm (71). Four isoforms of AUF1 (p37, p40, p42 and p45) are produced by alternative splicing of the initial AUF1 transcript (70). Western blot analysis demonstrated that rat kidney cortex and various kidney cell lines express all 4 isoforms of AUF1 (62). Additional experiments have established that recombinant HuR, a protein that participates in mRNA stabilization (72), also binds to the pHRE (unpublished data of Y. J. Lee and N. P. Curthoys). Concentrations of 0.1-0.4 μM of the two recombinant proteins are required to shift 50 percent of the labeled RNA. Therefore, the two proteins may bind to the pHRE with similar affinities. Thus, AUF1 and HuR may augment or counteract the effect of ζ-cryst and contribute to the regulation of the turnover of the GA mRNA.

ER-stress response

Eukaryotic cells sense a variety of stress conditions including nutrient deprivation; heat shock; changes in redox balance, osmolarity, or calcium concentration; decreased ATP levels; and increased synthesis or altered glycosylation of secretory proteins through the accumulation of unfolded proteins in the endoplasmic reticulum (ER) (73, 74). The resulting ER-stress response leads to increased expression of specific genes and the formation of cytosolic stress granules where translation of mRNAs that encode constitutively expressed proteins is inhibited and other mRNAs are targeted for rapid degradation or selective stabilization. This response is initiated by BiP, an abundant chaperone protein that participates in protein folding in the ER. BiP is localized to the ER through a cleaved N-terminal signal peptide and retained there by a C-terminal KDEL-retention signal. Normally, a large fraction of the BiP protein is sequestered by binding to transmembrane receptors on the luminal surface of the ER membrane. When unfolded proteins accumulate within the lumen of the ER, they interact with BiP. The release of BiP from the individual transmembrane receptors activates different downstream signaling events. For example, release of BiP from the IRE1 receptor results in receptor dimerization and activation of a cytosolic RNase domain that appropriately splices XBP1 mRNA leading to increased synthesis of the encoded transcription factor (75). In addition, the release of BiP from the transmembrane p90-ATF6 protein allows its translocation to the Golgi network where it undergoes proteolytic processing and release into the cytoplasm as p50-ATF6, a second transcription factor (76). The XBP1 and p50-ATF6 proteins are subsequently translocated to the nucleus where they enhance transcription of genes that encode ER chaperones and enzymes that participate in the degradation of unfolded proteins. As a result, the levels of BiP protein are significantly increased (77). This response facilitates the folding of proteins in the ER and gradually inhibits the activated signaling pathways.

In addition to transcriptional regulation, the ER-stress response also affects the rate of translation of specific mRNAs. The release of BiP from a third transmembrane receptor, PERK (PKR-like endoplasmic reticulum kinase) results in receptor dimerization and activation of its cytosolic kinase activity (78). Activation of PERK correlates with auto-phosphorylation of residue Thr980 (79). The activated PERK subsequently phosphorylates eIF2α on Ser51. Phosphorylation of eIF2α blocks formation of the active eIF2/GTP/Met-tRNA_i^Met ternary complex that participates in initiation of translation (80). This results in accumulation of 48S preinitiation complexes that contain the 40S ribosomal subunit, the inactive eIF2/GDP complex, and an mRNA. The 48S RNA/protein complexes then associate with TIA-1 and TIAR proteins to form cytoplasmic stress granules (81). TIA-1 and TIAR are RNA binding-proteins that contain a glutamine-rich prion-like domain that is essential for stress granule formation (82). Under normal conditions, TIA-1 and TIAR function primarily in the nucleus, but they are translocated to the cytoplasm in response to ER-stress (83). The mRNAs that accumulate in stress granules lack the 60S ribosomal subunit and are translationally inactive. This process results in the transient inhibition of translation of most mRNAs that encode housekeeping proteins. However, mRNAs, encoding the proteins that are induced in response to ER-stress, are excluded from stress granules and are actively translated on polysomes. Recent studies also suggest that stress granules function to triage mRNAs (84). Certain mRNAs are transferred from stress granules to adjacent processing bodies (84, 85) where they undergo rapid 5′→ 3′ degradation. Other mRNAs are released from stress granules and reassociate with polysomes where they are actively translated.

Previous data suggest the possibility that stabilization of rat renal GA mRNA during onset of acidosis may involve the transient association of the GA mRNA with stress granules. These observations include the initial finding that following acute onset of acidosis the increase in GA mRNA is initiated after an 8-10 h lag, whereas PEPCK mRNA levels are fully induced by this time (17, 86). Mass spectroscopic analysis of proteins in rat renal cortical homogenates that bind to a biotinylated RNA containing the pHRE identified TIAR as a potential pHRE-binding protein (64). Proteomic analysis, using ICAT-labeling, also established that the levels of BiP and calreticulin, an additional ER-chaperone protein, are increased in renal proximal convoluted tubules following onset of acidosis (unpublished data, L. Taylor and N.P. Curthoys). These observations suggest the hypothesis that changes in intracellular pH may affect protein folding in the ER of the renal proximal convoluted tubule and initiate an ER-stress response. To test this hypothesis, western blot analyses were performed using protein extracts prepared from WKPT cells, an established cell line derived from isolated rat renal proximal tubules (87). The transfer of WKPT cells to acidic medium results in a rapid (6 h) and pronounced (5-fold) increase in BiP protein (unpublished data of C. Cooper, Y. J. Lee and N. P. Curthoys). In contrast, the level of eIF2α protein is increased slightly (2-fold) and only after 48 h of treatment with acidic medium. However, there is a significant increase in phosphorylation of eIF2α within 24 h (2-fold) and a more pronounced increase after 48 h (5-fold). These data strongly support the hypothesis that kidney cells activate an ER-stress signaling pathway in response to onset of metabolic acidosis.

Immunostaining experiments indicate that in normal medium, ζ-cryst is largely distributed throughout the cytosol of WKPT cells (unpublished data of Y. J. Lee and N. P. Curthoys). However, treatment with acidic medium for 1 d results in the initial, but transient, association of ζ-cryst with granules that are localized near the endoplasmic reticulum (ER). After 1d, ζ-cryst co-localizes with eIF2α, indicating that the ζ-cryst may enter stress granules. The formation of stress granules that contain eIF2α and ζ-cryst occurs with the same time course as the increase in phosphorylation of eIF2α. However, after 5 d in acidic medium, ζ-cryst was redistributed through out the cytosol, whereas the phosphorylated eIF2α was retained in stress granules. The transient recruitment of ζ-cryst to the stress granules may prevent it from acting as a destabilizing factor of the GA mRNA or alternatively it may transiently recruit the GA mRNA to the stress granules where it may bind the stabilizing factor, HuR.

HuR is a member of the embryonic lethal abnormal vision (ELAV)-like family of RNA binding proteins (88). Through identification of its target transcripts, HuR has been implicated in the control of cell division, carcinogenesis, immune responsiveness, and the response to cellular stress (89). HuR has also been shown to stabilize ARE-containing mRNAs during hypoxia (90), UV irradiation (91), heat shock (92), nutritional deprivation (93) and energy depletion (94). While HuR is normally localized to the nucleus (95), the stabilization of the ARE-containing mRNAs in response to stress conditions correlates with the translocation of HuR to the cytoplasm. Additional immunostaining experiments (unpublished data of J. Mufti, Y. J. Lee and N.P. Curthoys) confirmed that HuR is predominantly localized in the nucleus of WKPT cells that are maintained in normal medium. However, when the cells are treated with acidic medium for 1 d, HuR is largely redistributed to the cytoplasm. This effect is sustained if the cells are maintained in acidic medium for 3 d, but is reversed when cells are returned to normal medium. These data and the observation that addition of recombinant HuR greatly stabilized the GemGAA₆₀ mRNA in an in vivo decay assay (unpublished data, Y. J. Lee and N.P. Curthoys) indicate that HuR may contribute to the selective stabilization of the GA mRNA in response to onset of acidosis.

Model for mRNA stabilization

Based upon the completed experiments, the following model is proposed to explain the normal degradation and the pH-responsive stabilization of the GA mRNA (Fig. 2). Additional experiments are required to test this model and to establish the role of the various pHRE binding proteins and the importance of the cellular localization of various RNA:protein complexes as possible regulatory factors in the pH-responsive stabilization of the GA mRNA. It will also be important to follow the movement of the GA mRNA within the cell in response to onset of acidosis and to determine in vivo the temporal interaction of the ARE-binding proteins with the pHRE. The resulting data should substantially increase knowledge of how this essential adaptive response is mediated and regulated.

Fig. 2. pH-responsive Stabilization of GA mRNA.

Panel A. At pH 7.4, the GA mRNA is constitutively translated to produce the basal level of GA protein. The binding of ζ-cryst and/or AUF1 to the pHRE within the 3′-UTR of the GA mRNA promotes the recruitment of a deadenylase and subsequent degradation of GA mRNA by the exosome or P-bodies. This process produces the rapid turnover (t_½ = 3 h) of the GA mRNA. Panel B. At pH 6.9, the translocation of ζ-cryst to stress granules and the release of HuR from the nucleus would favor the binding of HuR, a stabilizing factor, to the GA mRNA. The resulting stabilization of the GA mRNA leads to an increase in the level of GA mRNA and the more gradual increase in the level of the GA protein.