Plasticity of mouse renal collecting duct in response to potassium depletion. —Renal collecting ducts are the main sites for regulation of whole body potassium balance. Changes in dietary intake of potassium induce pleiotropic adaptations of collecting duct cells, which include alterations of ion and water transport properties along with an hypertrophic response. To study the pleiotropic adaptation of the outer medullary collecting duct (OMCD) to dietary potassium depletion, we combined functional studies of renal function (ion, water, and acid/base handling), analysis of OMCD hypertrophy (electron microscopy) and hyperplasia (PCNA labeling), and large scale analysis of gene expression (transcriptome analysis). The transcriptome of OMCD was compared in mice fed either a normal or a potassium-depleted diet for 3 days using serial analysis of gene expression (SAGE) adapted for downsized extracts. SAGE is based on the generation of transcript-specific tag libraries. Approximately 20,000 tags corresponding to 10,000 different molecular species were sequenced in each library. Among the 186 tags differentially expressed (P < 0.05) between the two libraries, 120 were overexpressed and 66 were downregulated. The SAGE expression profile obtained in the control library was representative of different functional classes of proteins and of the two cell types (principal and α-intercalated cells) constituting the OMCD. Combined with gene expression analysis, results of functional and morphological studies allowed us to identify candidate genes for distinct physiological processes modified by potassium depletion: sodium, potassium, and water handling, hyperplasia and hypertrophy. Finally, comparison of mouse and human OMCD transcriptomes allowed us to address the question of the relevance of the mouse as a model for human physiology and pathophysiology.
- serial analysis of gene expression
- kidney collecting duct
potassium controls many cell functions, the most important of which is membrane potential. This control rests both on the maintenance of whole body potassium content (∼3,600 mmol in a 70-kg human) and the distribution of potassium between intra- and extracellular compartments (∼3,500 mmol and 100 mmol, respectively). Homeostasia of whole body potassium relies on the continuous balance between dietary intake (∼100 mmol/day) and excretion of potassium. Kidneys are not only responsible for the bulk of potassium excretion (∼90%) but also for the adaptive changes in potassium handling in response to alterations of its dietary intake. This renal plasticity toward potassium handling is quantitatively important, as urinary potassium excretion may vary over 20-fold between potassium restriction and potassium loading conditions.
Renal handling of potassium along the renal tubule results from complex reabsorption and secretion processes: ∼90% of the filtered load of potassium is reabsorbed between the glomerulus and the distal convoluted tubule, whereas the downstream collecting duct usually secretes potassium into the tubular fluid. The collecting duct is the main site of homeostasic adjustments of the excreted load of potassium: secretion is enhanced in case of increased dietary intake of potassium, whereas during potassium restriction, the secretion process is abolished and replaced by a reabsorption mechanism (31), which allows recovery of most of the potassium remaining in the tubular fluid beyond the distal convoluted tubule. It is generally assumed that inhibition of potassium secretion during potassium restriction mainly originates in cortical collecting duct (CCD), whereas the potassium reabsorption mechanism prevails in the outer medullary collecting duct (OMCD) where it would be accounted for in part by the induction of a nongastric H-K-ATPase at the apical pole of principal cells (30).
Because the transports of water and of the different solutes in the collecting duct are directly or indirectly coupled at the molecular and/or cellular levels, adaptation to dietary potassium restriction alters not only the transport of potassium but also that of water and other solutes. Furthermore, functional adaptations of the collecting duct not only affect the transport functions of the different cell types but may include also changes in the energy metabolism (to balance ATP availability with active ion transport) (39), cellular hypertrophy and increase in apical and/or basolateral cell membrane surface (which increases the exchange surface available for transport) (26), cell proliferation (thereby increasing the number of transporting cells) (35,50), or even cell transdifferentiation (which modifies the ratio of the different cell types that constitute the collecting duct and thereby the proportion of cells dedicated to the different transport processes) (1). All these adaptive responses are likely mediated through a complex network of induction/repression of gene expression. Thus functional adaptations of the collecting duct appear as pleiotropic phenomena that include changes in the expression pattern of many different genes.
The present study was designed to analyze the pleiotropic adaptation of the collecting duct to dietary potassium depletion. For this purpose, we compared the molecular phenotype of collecting duct from mouse fed either a normal or a potassium-depleted diet. The study was restricted to the OMCD because of its major role in potassium conservation in potassium depletion and of its lesser cellular heterogeneity compared with the CCD. The pattern of gene expression was evaluated using the serial analysis for gene expression (SAGE) microassay compatible with microdissected OMCDs previously developed in the laboratory (59).
Experiments were carried out on two groups of male C57BL/6J mice (8–10 wk old; Charles Rivers Breeding Laboratories). Mice were fed either a potassium-deficient diet (LK) containing 1.5 meq K+/kg (SAFE, Epinay, France) for 3 days (LK-3d) or 14 days (LK-14d), or a similar diet supplemented with KCl (150 meq/kg) (control). All animals had free access to food and were allowed to drink deionized water ad libitum.
For metabolic studies, animals were housed in individual metabolic cages (Phymep, Paris, France) starting 3 days before the beginning of the experiments. Then, 24-h urine samples were collected under mineral oil for 2 days under the K+-supplemented diet followed by 14 days on the LK diet. On the same mice, blood was sampled by retro-ocular puncture at day 2 on the regular diet and by exsanguination at day 14 on the LK diet, for determination of plasma sodium, potassium, calcium, magnesium, and creatinine concentrations. Urine volume and osmolality were gravimetrically and cryometrically determined, respectively. Plasma and urine concentrations of creatinine, sodium, potassium, calcium, and magnesium were determined with an automatic analyzer (Hitachi 911; Boehringer Mannheim). Glomerular filtration rate (GFR) was evaluated by the clearance of creatinine.
In an other group of animals, blood samples were obtained by retro-ocular puncture in control, LK-3d, and LK-14d mice. A single puncture was obtained from each animal. Blood pH and bicarbonate concentration were determined with a blood gas analyzer (ABL30; Radiometer, Copenhagen, Denmark).
In a last series, kidneys were removed from control, LK-3d, and LK-14d mice for determination of fresh and dry weights.
mRNA extraction and generation of SAGE libraries.
Two SAGE libraries were prepared from OMCD of control and LK-3d mice, respectively. After anesthesia (pentobarbital sodium, 140 μg/g body wt), the left kidney was perfused with Hanks’ modified microdissection solution, then with the same solution supplemented with 0.15% collagenase (Serva, Heidelberg, Germany). The kidney was sliced along the corticopapillary axis in small pieces which were incubated for 20 min at 30°C in collagenase-containing (0.25%) microdissection solution. After rinsing, microdissection was performed at 4°C under stereomicroscopic observation.
Approximately 400 and 600 OMCDs dissected from two control and two LK-3d mice were used to generate SAGE libraries. Libraries were generated by using the SAGE adaptation for downsized extracts (SADE) method (59). The main modifications of this protocol consists of the use of oligo(dT)25 covalently bound to magnetic beads (Dynabeads mRNA direct kit; Dynal, Oslo, Norway) to purify poly(A) RNAs, with the use of Sau3AI as the anchoring enzyme.
Sequencing was performed on DNA minipreps by using Big Dye terminator sequencing chemistry (Applera) and an automated sequencer (ABI 377, Applera). Sequence files were analyzed using SAGE2000 software (58). Tags corresponding to linker sequences were discarded, and those originating from duplicate ditags were counted only once. Significant differences between the two libraries were assessed by Monte-Carlo simulation analysis (63), with P < 0.05 being considered as significant.
Identification of tags.
Tags present at statistically different levels in the two libraries were identified through matching to UniGene clusters using the SAGEmap resource (http://www.ncbi.nlm.nih.gov/SAGE) at NCBI. The reliability of tag identification was validated by confirming the correct location and orientation of the tag in relation to the 3′ most Sau3AI site, as well as the presence of a polyadenylation signal and/or poly(A) tail. In some instances, expression of a single gene was detected through two tags contiguous to the 3′ most Sau3AI site and to the penultimate one. This feature was exclusively observed for tags with high occurrence, consistent with incomplete Sau3AI digestion of the cDNA. These tags were referred to as A and B. For tags reliably matching sequences belonging to two unrelated clusters, the two entries were recorded. Tags reliably matching >2 clusters were referred to as multiple matches. For tags without reliable identification, we searched whether they might correspond to neighbor of identified tags. Neighbors, which were defined by the presence of a single nucleotide difference (mutation or shift) with the original tag and thereby likely correspond to sequencing errors or genetic polymorphism, are identified by an asterisk in the list in Supplemental Table S3 (available at the Physiological Genomics web site).1
mRNA extraction and RT-PCR analysis.
The following structures were microdissected from collagenase-treated kidneys according to morphologic and topographic criteria: proximal convoluted and straight tubules (PCT and PST), medullary and cortical thick ascending limb of Henle’s loop (MTAL and CTAL), and CCD and OMCD. Total RNAs were extracted from pools of nephron segments (1–2 cm length) by using a microadaptation (12) of the method of Chomczynski and Sacchi. RT-PCR was carried out on total RNAs corresponding to 1 mm of kidney tubule, which corresponds to ∼400 cells. Reverse transcription and PCR were performed sequentially in the same reaction tube in the presence of [α-32P]dCTP (106 Bq/nmol). The primers (Supplemental Table S1) were selected (Oligo 4.0; MedProbe, Oslo, Norway) from mouse sequences available from GenBank. RT-PCR products were separated by electrophoresis on 2% agarose slab gels and quantitated with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). The number of PCR cycles was adjusted to provide nonsaturating signals detectable by PhosphorImager but not by ethidium bromide staining.
When the expression profile of a transcript along the nephron was analyzed, RNAs from the different nephron segments were processed in the same experiment, and in each experiment, the RT-PCR product quantitated by using PhosphorImager (arbitrary units) in each structure was calculated as percent of the amount in the PCT. Results are expressed as means ± SE from several experiments in different animals.
When comparing expression levels in OMCDs from different groups of mice (control, LK-3d, and LK-14d), RNAs from at least one control and one LK mouse were analyzed in each experiment, and values for the LK animal were expressed as percent of the control. Results are expressed as means ± SE from several experiments in different animals.
Statistical analysis between groups was performed by variance analysis, with P < 0.05 being considered as significant.
Ultrastructure of OMCD.
Kidneys from the three groups of mice were fixed by in situ perfusion with 2% glutaraldehyde in PBS followed by a 1-h incubation in the same solution and washed in 0.1 M cacodylate buffer (pH 6.8). Tissues were postfixed for 1 h in a 1/1 mixture of 2% aqueous osmium tetroxide and 3% aqueous potassium ferrocyanide. Dehydration was performed in graded ethanol baths followed by Epon embedding (Sigma, St Louis, MO). We cut 90-nm sections with a Reichert ultramicrotome (Leica, Wetzlar, Germany), and we counterstained these for 2.5 min with 1% lead-citrate before examination at the electron microscope (model EM400; Philips, Limeil Brévannes, France).
Proliferating cell nuclear antigen (PCNA) labeling.
Kidneys from the three groups of mice were fixed by in situ perfusion with formalin. Double immunohistochemistry with anti-AQP-2 and anti-PCNA antibodies was performed using a three-step streptavidin-biotin method with prior antigen unmasking procedure. Deparaffinized 5-μm thick kidney sections were heated at 98°C for 30 min in TRS buffer (DAKO, Trappes, France). Endogenous peroxidase, avidin, and biotin activities were blocked using 0.3% H2O2 and avidin and biotin blockers (DAKO), respectively. Monoclonal anti-AQP-2 1321 antibody (kindly provided by Dr. A. Blanchard, INSERM U356) was then incubated at 1.25 μg/ml, for 60 min, at room temperature. Secondary biotinylated anti-mouse IgG antibody (Amersham, Les Ulis, France) was used at 1/200 for 30 min. Streptavidin-peroxidase amplification was then performed with the ELITE kit (Vector, Burlingame, CA) according to the manufacturer’s instructions using diaminobenzidine (DAKO) as a chromogen. After rinsing in PBS, the sections were incubated with anti-PCNA monoclonal antibody (PC10 clone, DAKO) at 1/200 for 60 min, at room temperature. Secondary biotinylated anti-mouse IgG antibody (Amersham) was used at 1/200 for 30 min. Streptavidin-phosphatase procedure was performed with the ABC-phosphatase kit (Vector) using Vector blue chromogen (Vector). The slides were then counterstained with hematoxylin. PCNA labeling in OMCD in principal (AQP-2 positive) and intercalated (AQP-2 negative) cells was determined by counting an average of 750 cells per animal in a blinded fashion.
Table 1 summarizes body weights and urinary parameters at day 2 on K+-supplemented diet (control) and days 3 and 14 on LK diet, and plasma parameters on the same mice at day 2 on K+-supplemented diet (control) and day 14 on LK diet. As previously reported in rats (24), body weight decreased in mice fed the LK diet. Plasma creatinine concentration (Pcreat) decreased in parallel with the body weight, suggesting that the latter was mainly due to a loss of muscular mass. Decrease in urinary excretion of creatinine (Ucreat×UV) in LK-14d mice was higher than the decrease in Pcreat, revealing a 50% reduction in the glomerular filtration rate (GFR). Urine flow rate (UV) remained constant during the first week on LK diet and then decreased by ≈30% at day 14. This effect was likely accounted for, at least in part, by the decrease in GFR. Decrease in urinary excretion of potassium (UK×UV) markedly exceeded the decrease in the filtered load of potassium: UK×UV decreased by 53% at day 1, 90% at day 3, and thereafter slowly decreased toward 95% at day 14. Despite this reduction in urinary potassium excretion and in muscle mass, plasma potassium concentration only dropped by 35% after 14 days on LK diet. Urinary excretion of sodium (UNa×UV) decreased in parallel with GFR, and plasma sodium concentration was not modified after 14 days on LK diet. Excretion of calcium (UCa×UV) also decreased in excess of GFR, but this did not alter significantly plasma calcium concentration. In contrast, the marked reduction in magnesium excretion (UMg×UV), which was proportional to that of potassium, induced a significant hypermagnesemia. The parallelism between UK×UV and UMg×UV observed during potassium depletion was previously reported during dietary magnesium restriction (60), suggesting a tight coupling between renal transport of these two ions. However, the cellular and molecular mechanisms of magnesium transport in the distal nephron remain as yet unknown.
Conversely to rats, which develop a metabolic alkalosis in response to potassium depletion (27), mice displayed a slight tendency to develop a moderate acidosis at day 14.
OMCD hypertrophy and hyperplasia.
The increase in the fractional kidney weight observed after 3 days of potassium depletion was totally accounted for by the decrease in body weight, as the fresh and dry kidney weights were not different in control and LK-3d mice (Table 2). In contrast, after 14 days, the fresh and dry kidney weights were increased by 13%, revealing a marked hypertrophy and/or hyperplasia.
Electron microscopic observation revealed that OMCD principal and intercalated cells had similar heights in control and LK-3d mice (Fig. 1, A and B),revealing the absence of hypertrophy. Nonetheless, intercalated cells displayed an amplification at their luminal pole, with the development of numerous microplicae of apical membrane and, to a lesser extent, at their serosal pole with enhanced digitation processes of the basal membrane. Ultrastructure of principal cells was not altered significantly. In contrast, in LK-14d mice both cell types were markedly hypertrophied and had invaded the tubular lumen, which appeared as virtual (Fig. 1, C and D). Also,the membrane alterations of intercalated cells were further developed, especially at the basal membrane, which displayed deep invaginations.
Hyperplasia was evaluated through quantitation of PCNA-positive cells. Principal and intercalated cells were distinguished on the basis of the presence or absence of aquaporin 2 (Fig. 2). In control mice, no PCNA-positive cell was detected, confirming the very slow proliferation rate of tubular epithelial cells in normal kidney. In contrast, OMCDs from LK-3d mice displayed PCNA-positive cells in almost all microscopic fields observed. PCNA-positive cells were more frequent among intercalated than principal cells (Table 3). In OMCDs of LK-14d mice, PCNA-positive cells were still detected consistently, but their frequency was reduced compared with LK-3d mice.
Altogether, these results indicate that adaptation to a LK diet induces hypertrophy and hyperplasia of both principal and intercalated cells of the OMCD and that hyperplasia precedes hypertrophy, although the two phenomena overlap between days 3 and 14.
Transcriptome of OMCD from control mice.
A total of 19,698 tags were sequenced from the control mice library, corresponding to 10,521 different tags [see complete list on Supplemental Table S2 and GEO (accession numbers GSM23474 and GSM23478)]. The gene expression profile in the OMCD of control mice was similar to that previously reported (59). It is characterized by the following general features (Supplemental Table S2): 1) 40% of all tags, representing 75% of the molecular species, were counted only once; 2) the most abundant tag represented ≈2% of all tags; 3) 8 of the 10 most represented tags are encoded by the mitochondrial genome; and 4) the most abundant tag of nuclear origin corresponds to the mRNA encoding the water channel aquaporin 2 (AQP-2).
The OMCD transcriptome was representative of both principal and intercalated cells as it includes specific markers of these two cell types; AQP-2, -3, and -4, and α-, β-, and γ-subunits of amiloride-sensitive epithelial sodium channel (ENaC) for principal cells, and chloride/bicarbonate exchanger kAE1, several subunits of V-type H-ATPase, Rhbg, and Rhcg for intercalated cells (Table 4). In addition to ion transporters, the OMCD library also included tags corresponding to hormone receptors (vasopressin V2 and thyroid hormone receptors, protease-activated receptor 3), proteins involved in hormonal transduction and signalization (adenylyl cyclase 6, GTP binding protein Gs, inositol trisphosphate receptor type 2, calmodulin 1), proteins involved in genetic diseases (11β-hydroxysteroid dehydrogenase 2, carbonic anhydrase 2, 56/58-kDa and 116-kDa subunits of vacuolar H-ATPase), drug targets (cyclophilin and FK506-binding protein), transcription factors (Sp3, Ear2, transcription repressor ring 1A), enzymes of the energy metabolism, and cytoskeleton proteins. In conclusion, the gene expression profile obtained by SAGE is representative of different functional classes of proteins and of the two cell types constituting the OMCD.
Differential gene expression in OMCD from control and LK-3d mice.
Based on the time course of metabolic changes, we analyzed the molecular phenotype of OMCDs after 3 days of potassium depletion, i.e., when 1) adaptation of solute transport was almost complete, 2) cell proliferation was high, and 3) before apparent cellular hypertrophy. A total of 20,200 tags corresponding to 9,833 different tags were sequenced from the LK-3d library (Supplemental Table S2), and the general features of this library were similar to those of the control one.
The number of tags differentially expressed (P < 0.05) in the two libraries reached 186, including 120 tags overrepresented in the LK-3d library (Supplemental Table S3). Among the differentially represented tags, ∼60% were reliably identified as single cDNAs, whereas 13% (14 tags) did not match any GenBank sequence (Table 5). It should be noted that 11/14 tags without GenBank matching sequence corresponded to neighbors of reliably identified tags. Unexpectedly, the proportion of reliably identified tags was much higher for tags overrepresented in the LK-3d library than for tags found at lower level. Over 80% of the cDNAs corresponding to differentially represented tags were functionally characterized. Table 6 and 7 list the tags functionally identified as nuclear transcripts that are present at higher and lower levels in the LK-3d library, respectively. Almost 50% of tags overrepresented in the LK-3d library encode proteins involved in energy or protein metabolism, presumably in relation with hypertrophy and cell proliferation. Next, the most abundant classes of overrepresented tags correspond to proteins related to cell proliferation (17%) and to transporters (11%).
Validation of SAGE data.
To validate the SAGE data, RT-PCR was performed on a selection of five tags over represented in the LK-3d library. Selected tags correspond to reliably and functionally identified genes: growth differentiation factor 15 (Gdf15), urokinase, cofilin, Slc22a4 (the organic cation transporter OCTN1), and non-erythroid Rhesus protein Rhcg. RT-PCR was performed on OMCDs from control mice and mice fed the LK diet for either 3 or 14 days. As shown in Fig. 3, RT-PCR confirmed the stimulation of expression of mRNAs corresponding to four of the selected tags at day 3, whereas for the last one overexpression was revealed only after 14 days of potassium depletion. After 14 days of potassium depletion, the overexpression averaged 2- to 6-fold, except for OCTN1 the expression of which was increased >20-fold.
Expression of OCTN1 in the collecting duct was unexpected, since this transporter is assumed to be specific of the proximal tubule (61). Therefore, we further investigated expression of OCTN1 along the nephron and its regulation during potassium depletion, as well as that of other members of the organic cation transport family, namely OCT1, 2, and 3 and OCTN2 and 3. Results in Fig. 4 confirm that the PCT and PST are the main renal sites of expression of OCTN1. However, a significant level of expression was also observed in the thick ascending limb (both in MTAL and CTAL), whereas expression level was vanishingly low, and at the limit of detection by RT-PCR, along the collecting duct. Potassium depletion did not alter OCTN1 expression in the PCT but increased it in MTAL, CCD, and OMCD (1.7-, 7-, and 20-fold increases, respectively). mRNAs encoding the other members of the organic cation transporters family, except OCT1 and OCT3, were also detected in the OMCDs, but their expression was not or only slightly modified after 14 days of potassium depletion (data not shown).
Other tags of interest.
For tags with low occurrence number in the libraries (0 or 1), a 5- to 7-fold change in abundance was necessary to reach the P < 0.05 significance level, whereas ≈2-fold and 1.2-fold changes were sufficient for tags counted 15 and 100 times, respectively. This means that, at the deepness of analysis reached in this study (20,000 tags sequenced in each library), statistically significant differential expression of transcript of low abundance can be determined only for transcripts displaying a high degree of induction. From a physiological point of view, it is well known that changes of much smaller amplitude are sufficient to induce relevant functional alterations. Furthermore, transcriptome of the human OMCD indicated that some transporters of importance for collecting duct function are expressed at relatively low level (7). Thus we compared the abundance in the two libraries of tags corresponding to transcripts encoding the main transporters expressed in the collecting duct (Table 4). To compare gene expression profiles in mouse and human OMCD, Table 4 also lists the abundance of the cognate tags in human OMCD (7), after normalization of tag counts to 20,000 total tags per library. Comparison between gene expression profiles in mouse and human OMCD was extended to all the transcripts from nuclear genome corresponding to tags representing ≥1‰ of the whole mouse OMCD transcriptome. Note that tag abundance of zero in Table 4 may indicate either 1) that the transcript is not detectable at the depth of analysis attained in this study (20,000 tags) or 2) that mouse OMCDs express a transcript variant with a tag different from that deduced from the cDNA sequence in GenBank.
Renal sodium, potassium, and water handling in LK mice and expression of related genes.
Conversely to the rat, which responds to potassium depletion by a marked polyuria, urine flow rate decreased in LK mice. Along with this difference in urine volume, urine osmolality decreased during potassium depletion in rat, whereas it increased in mice. In the rat, polyuria is associated with downregulation of AQP-2 expression in the kidney cortex and medulla (37). The observed induction of AQP-2 (Table 4) and vasopressin V2 receptor (tag abundance: ncontrol = 1; nLK-3d = 4) mRNAs expression in OMCD of LK mice may account in part for the different behaviors of rat and mouse. Note that expression of constitutive basolateral water channels AQP-3 and AQP-4 was not altered during K depletion (Table 4).
Concerning sodium transporters, transcripts for the three subunits of the epithelial sodium channel ENaC were detected in the mouse OMCD transcriptome. Occurrence of their corresponding tag was not altered consistently during potassium depletion (Table 4), in agreement with the maintenance of a constant fractional excretion of sodium. In contrast, but consistent with previous observations in the rat OMCD (5), expression of mRNAs encoding the three subunits of Na-K-ATPase was increased in OMCD of potassium-depleted mice (note that SAGE allowed us to discriminate three distinct transcripts of Na-K-ATPase β1-subunit corresponding to distinct polyadenylation sites). The physiological significance of increased Na-K-ATPase expression in OMCD of hypokalemic animals remains paradoxical in view of the role of Na-K-ATPase in potassium secretion in the collecting duct. The decreased occurrence of the tag for ROMK (Table 4), although not statistically significant, may explain in part that induction of Na-K-ATPase does not result in a parallel increase in potassium secretion.
The main functional adaptation of collecting duct during potassium depletion is that instead of secreting potassium into the lumen, it reabsorbs it to reduce the renal loss. Potassium reabsorption along the OMCD is thought to be mediated by active uptake of potassium across apical membrane of tubular cells via the nongastric H-K-ATPase (30), the expression of which is induced during potassium depletion both in rat (38) and mouse OMCD (unpublished observation). However, the tag corresponding to the α-subunit of nongastric H-K-ATPase was not detected in the transcriptome of LK-3d mice (Table 4) although 1) the nucleotide sequence of mouse nongastric H-K-ATPase α-subunit available in GenBank displays several restriction sites for the anchoring enzyme (Sau3AI) that specifies the tag, 2) the length of this sequence is consistent with that of the transcript detected by Northern blot in mouse kidney (62), and 3) analysis of AU-rich element (ARE) in the 3′-untranslated region of mouse nongastric H-K-ATPase α-subunit mRNA allocates it to ARE Database group V transcripts (2), i.e., a class of mRNAs with normal degradation rate. In fact, assuming an mRNA abundance of nongastric H-K-ATPase α-subunit similar to that found in potassium-depleted rat OMCD (≈30 mRNAs per cell; Ref. 38) and a total number of 300,000 transcripts per cell, the tag is expected to be counted twice in the SAGE library (20,000 sequences analyzed) but with a probability of only 86% (9). This example illustrates the sensitivity limit of SAGE detection for transcripts of intermediate abundance.
Mucolipin 3 (Mcoln3), the tag abundance of which is 0 and 7 in the control and LK-3d libraries, respectively, might also be involved in changes in transport function of OMCD during potassium depletion. Mcoln3 was discovered by positional cloning in varitint-waddler mice, a strain with hearing loss, vestibular defects, and pigmentation abnormalities (11). Mucolipins constitute a family of cation channels with orthologs in human (MCOLN1), Drosophila melanogaster, and Caenorhabditis elegans (CUP-5). Human MCOLN1 is a high-conductance calcium-sensitive calcium channel that is also permeable to sodium and potassium (29). Mutations of MCOLN1 in human are responsible for mucolipidosis IV (3), a neurodegenerative disease characterized by abnormal endocytosis (OMIM 252650), whereas loss-of-function mutations of C. elegans CUP-5 leads to endocytosis defect and increased apoptosis (13,21). These defects are likely accounted for by the dysfunction of intracellular mucolipin, and indeed, Mcoln3 is located in the cytoplasm of inner ear hair cells (11). However, because Mcoln3 is also located in the plasma membrane of stereocilia (11), it can be speculated that it might contribute to transmembrane cation transport in kidney epithelial cells.
Renal acid/base balance in LK mice and expression of related genes.
Control of acid-base balance by the collecting duct is a function of α- and β-intercalated cells. The high occurrence of the tag for α-intercalated cell-specific basolateral chloride/bicarbonate exchanger kAE1 and the absence of that for pendrin, the apical anion exchanger of β-intercalated cell (Table 4), is consistent with the fact that mouse OMCDs are devoid of β-intercalated cells.
Present results confirm the previous observation (40) that potassium depletion induces no major change in acid-base status in mice, in contrast with the marked alkalosis that prevails in potassium-depleted rats. In agreement with this finding, we found no change in expression of transcripts for carbonic anhydrase II and kAE1, which constitute a transport metabolon (51), the expression of which is controlled by acidosis and alkalosis (49, 55).
Nonetheless, results indicate changes in expression of mRNAs encoding three proteins putatively involved in the maintenance of acid-base balance. The first two correspond to Rhcg, the occurrence of which increased from 26 to 49 (P < 0.008), and Rhbg, whose occurrence decreased, although not statistically significantly, from 7 to 3 during potassium depletion (Table 4). Rhbg and Rhcg are non-erythroid members of the erythrocyte Rhesus protein recently cloned in the kidney (33, 34) that share homologies with Mep/Amt NH3/NH4+ transporters in primitive organisms and plants (23). In rat OMCD, Rhbg and Rhcg are expressed at the basolateral and apical pole of intercalated cells (47), respectively, where they are supposed to mediate NH3/NH4+ secretion. However, the reciprocal changes in expression of Rhcg and Rhbg transcripts and the late time course of Rhcg induction (Fig. 3) suggest that their regulation in potassium depletion might not be related to the regulation of transepithelial NH3/NH4+ secretion. The third transcript corresponds to carbonic anhydrase 15, a putative new member of carbonic anhydrases family for which the only available information concerns the cDNA sequence. In silico analysis of the cognate protein reveals the presence of a signal peptide but no transmembrane or glycosyl-phosphatidyl-inositol anchoring domain, suggesting that carbonic anhydrase 15 might be secreted, as previously reported for carbonic anhydrase 6 (41). If so, overexpression of carbonic anhydrase 15 might not alter the function of OMCD cells but rather some extracellular process away from its synthesis site.
Collecting duct hyperplasia and hypertrophy and expression of related genes.
Kidneys respond to numerous stimuli by increasing the size of nephrons rather than their number. Such increase in size is mainly accounted for by cellular hypertrophy, but cellular proliferation may also occur. In many circumstances, renal cell hypertrophy is linked to increased ammonia production (20, 28), which leads to an alkalinization of the lysosomal compartments and an inhibition of proteolysis and thereby to protein accumulation without progression of the cell cycle (15, 20). Hyperplasia is triggered by growth factors that induce the entry in the cell cycle and secondarily an increase in protein synthesis and a reduction of proteolysis.
In the present study we observed that potassium depletion induces both cellular hypertrophy (Fig. 1) and cellular proliferation (Fig. 2 and Table 3) in both principal and intercalated cells of the OMCD. At the molecular level, the hypertrophy/hyperplasia process is attested by the “overexpression” of transcripts encoding several ribosomal proteins, nucleolin, structural proteins (tubulin, the actin depolymerization factor cofilin, and spectrin β3), and proteins involved in the processing of the extracellular matrix [urokinase, tissue inhibitor of metalloprotease 3 (TIMP3)] and cell-matrix interaction (ADAM9) (36). It is noteworthy, for example, that urokinase is involved in liver regeneration following mass reduction. Also of interest, in relation with the remodeling of cell membrane structure during potassium depletion (Fig. 1), is the “overexpression” of the transcript of Eps8-like 2 (42), a protein of the multimolecular complex that links growth factor-induced activation of PI3-kinase with Rac-induced reorganization of actin (25).
Potassium depletion is also associated with the “overexpression” of transcripts previously found in various epithelial cancers, including kidney carcinomas: urokinase (44), TIMP3 (44), the proto-oncogene kinase Pim3 (10), ferritin (8), tumor protein D52 (6), mucin (19), cofilin (56), tumor differentially expressed protein (45), macrophage migration inhibitory factor, which promotes both tumor growth and tumor associated angiogenesis (48, 53), and the two members of the AP-1 transcription factor, JunD and Fos (57). Also, we observed “down-expression” of the tumor suppressor annexin 6 (18, 54) and of IGF-binding protein 7, also known as mac25, which is a growth-suppressing factor as well as an IGF-binding protein (43). With the exception of IGFBP-7, which may alter the distribution along the nephron of IGF-I, a growth factor involved in potassium depletion-induced renal hypertrophy (14), all these genes may participate to the hypertrophy/hyperplasia response of the OMCD to potassium depletion, but they likely do not play a promoting role in these processes.
Present results also show that hyperplasia preceded hypertrophy, as it was highest after 3 days of potassium depletion, when cell size was normal. This time course is consistent with a switch from hyperplasia toward hypertrophy through blockade of the progression of the cell cycle before S phase. Two mechanisms have been reported to account for this hyperplasia/hypertrophy switch in renal cells. The first one is mediated by signaling molecules, such as transforming growth factor-β (TGF-β) which stops growth factor-induced progression of cell cycle before entry into S phase. The second one is mediated by ammonia, through inhibition of lysosomal function. Both TGF-β and ammonia action on switching EGF-induced hyperplasia to hypertrophy require the activity of proteins of the retinoblastoma family (16, 17, 32). TGF-β inhibits growth factor-induced phosphorylation of retinoblastoma family of proteins via its action on cyclin E/cdk2 kinase (32), whereas the mechanism linking lysosomal inhibition by ammonia with cell cycle remains unknown. Analysis of transcriptome points out several genes possibly involved in this hyperplasia/hypertrophy switch in OMCDs. The first one is Gdf15, the expression of which increased markedly between day 3 and day 14 of potassium depletion (Fig. 3), i.e., at the time of OMCD hyperplasia. Thus, Gdf15, which is a member of TGF-β superfamily (4), may act as an autocrine growth factor in OMCD. Expression of Gdf15 in hepatocytes is dramatically upregulated under circumstances of liver injury and regeneration (22). More recently, Gdf15 was shown to prevent potassium depletion-induced apoptosis in cerebellar neurons (52), suggesting that its expression might be controlled by potassium availability. Other genes possibly involved in the hyperplasia/hypertrophy switch are those encoding Rhcg and Rhbg which likely alter ammonia handling in OMCD intercalated cells (see above). Induction of the expression of OCTN1, which is also a transporter of weak organic bases and therefore may alter the lysosomal pH, may be related to the same process.
Comparison of human and murine OMCD transcriptomes.
Because of the convenience of transgenesis in the mouse, this species has become a must for physiological and pathophysiological studies. However, this raises the question of the relevance of the mouse as a model for human physiology. Comparison of mouse and human transcriptomes may provide a partial answer to this question. From data in Table 4, which compares the abundance of some 40 transcripts in mouse and human OMCD, one can conclude that there is a good correlation between the abundance in the two libraries of a large majority (70%) of tags, including cell-specific and -unspecific tags such as ENaC subunits and ribosomal proteins, respectively. However, the abundance of several markers of collecting duct cells is markedly different in mouse and human libraries. For example, expression of Rhcg and chloride/bicarbonate exchanger kAE1 was >20-fold higher in mouse than in human OMCD. Also, there are marked discrepancies for AQP-2 and AQP-3: first, in human OMCD the abundance of the two water channels transcripts was similar, whereas in mouse OMCD abundance of AQP-2 transcripts was much greater than that of AQP-3. Second, expression of AQP-2 is much higher in mouse than in human OMCD. Whether these differences reflect true species differences is not ascertained, as several epigenetic factors might account for them: 1) present experiments were performed on young adult mice, whereas the study in human was carried in patients aged 59 as a mean (7), and expression of AQP-2 decreases with aging (46); 2) human tissue was obtained from donors undergoing nephrectomy, and therefore the surgical procedure may have altered gene expression, particularly for AQP-2, the expression of which is highly dependent on the hydration status of the organism; and 3) although human mRNAs were extracted from healthy kidney fragments, one cannot exclude that the neighboring tumor for which nephrectomy was performed and/or the presurgical chemotherapy given to the patients might have altered gene expression.
Data in Table 4 also outline that some transcripts are excluded from SAGE analysis. For example, calgizzarin, which is expressed at high level in mouse OMCD, cannot be detected in human libraries because its cDNA lacks Sau3AI site. The same holds for the channel-inducing factor CHIF, one of the most abundant tags in the human OMCD library, which cannot be detected in mouse library.
In summary, this study demonstrates the feasibility of differential analysis of transcriptome at the level of well-defined nephron structures by using SADE. It also shows that, when coupled to functional studies, transcriptome analysis may identify candidate genes for distinct physiological processes. Finally, this work provides a database accessible online for gene expression in the mouse OMCD.
This work was supported by Centre National de la Recherche Scientifique (CNRS) and Commissariat à l’Energie Atomique grants to Unité de Recherche Associée 1859 and CNRS grants to Unité Mixte de Recherche 7134.
We gratefully acknowledge the assistance of Emmanuelle Billon for sequencing SAGE libraries and of Renée Gobin for electronic microscopy.
↵1 The Supplementary Material for this article (Supplemental Tables S1–S3) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00055.2004/DC1.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: A. Doucet, UMR 7134, Institut des Cordeliers, 15 rue de l’Ecole de Médecine, 75270 Paris cedex 6, France (E-mail).
- Copyright © 2004 the American Physiological Society