点击显示 收起
【摘要】 Hypoxia has been suspected to trigger transdifferentiation of renal tubular cells into myofibroblasts in an epithelial-to-mesenchymal transition (EMT) process. To determine the functional networks potentially altered by hypoxia, rat renal tubule suspensions were incubated under three conditions of oxygenation ranging from normoxia (lactate uptake) to severe hypoxia (lactate production). Transcriptome changes after 4 h were analyzed on a high scale by restriction fragment differential display. Among 1,533 transcripts found, 42% were maximally expressed under severe hypoxia and 8% under mild hypoxia (P O 2 = 48 mmHg), suggesting two different levels of oxygen sensing. Normoxia was required for full expression of the proximal tubule-specific transcripts 25-hydroxyvitamin D 1-hydroxylase ( Cyp27b1 ) and L -pyruvate kinase ( Pklr ), transcripts involved in tissue cohesion such as fibronectin ( Fn1 ) and N-cadherin ( Cdh2 ), and non-muscle-type myosin transcripts. Mild hypoxia increased myogenin transcript level. Conversely, severe hypoxia increased transcripts involved in extracellular matrix remodeling, those of muscle-type myosins, and others involved in creatine phosphate synthesis and lactate transport ( Slc16a7 ). Accordingly, microscopy showed loss of tubule aggregation under hypoxia, without tubular disruption. Hypoxia also increased the levels of kidney-specific transcripts normally restricted to the less oxygenated medullary zone and others specific for the distal part of the nephron. We conclude that extensive oxygen supply to the kidney tubule favors expression of its differentiated functions specifically in the proximal tubule, whose embryonic origin is mesenchymal. The phenotype changes could potentially permit transient adaptation to hypoxia but also favor pathological processes such as tissue invasion.
【关键词】 gene expression profiling hypoxia cell adhesion metabolism aging
HYPOXIA HAS BEEN SUSPECTED to trigger transdifferentiation of renal tubular cells into myofibroblasts in an epithelial-to-mesenchymal transition (EMT) process. In fact, the kidney is able to withstand transient periods of hypoxia, during which blood flow is redirected to other organs. For example, during strenuous exercise, renal blood flow can be divided by five ( 29 ), and at 5,800 m above sea level, kidney blood flow is divided by a factor of two and arterial P O 2 was reduced by 40% ( 34 ). On the other hand, an extended period of reduced tubule capillary perfusion during glomerular sclerosis in aging has been suspected to cause harmful local hypoxia. From studies made in cultured cells as well as in animal models ( 19 ), hypoxia was shown to trigger a transdifferentiation process to a mesenchymal cell type, myofibroblasts, because smooth muscle proteins were induced in renal cells by hypoxia ( 19, 23, 45 ). EMT could be one of the first events leading to renal fibrosis ( 12, 45 ), during which there is a thickening of extracellular matrix (ECM). Conversely, hypoxia has been suspected to favor tumor invasion through disruption of ECM, although it can decrease tumor growth ( 37 - 39 ). Therefore, exhaustive data were needed to clarify and delineate the changes induced by hypoxia in renal proximal tubules.
The aim of this study was to investigate early changes in transcriptome in the normally well-oxygenated renal proximal cortical tubule. Rat tubule suspensions were differentially oxygenated under three conditions that reproduce normoxia with lactate uptake and two conditions of controlled hypoxia with a progressive shift toward a glycolytic metabolism. Transcriptome changes were analyzed on a high scale by restriction fragment differential display (RFDD) to determine the different functional networks altered by hypoxia and potentially involved in changes of tubular phenotype.
METHODS
Experimental Hypoxia
Renal cortical tubule suspensions were prepared by collagenase digestion as previously described ( 33 ) from male Sprague-Dawley rats weighing 300 g (Charles River, Les Oncins, France). In each experiment, kidney cortexes from four animals were pooled to prepare a purified tubule suspension that was distributed into separate flasks. The DMEM/F12 medium used for preparation and incubation of the suspensions was equilibrated with 75% N 2 -20% O 2 -5% CO 2 and contained 25 mM bicarbonate, 10 mM HEPES at pH 7.4, 5 mM glucose, 0.5 mM lactate, 4 mM glutamine, and 20 mg/l egg white lysozyme as an antibacterial agent, but neither serum nor albumin except during collagenase digestion. The suspension was incubated at 37°C in 25-ml flasks, the atmosphere of which was equilibrated for 2 min through the stopper with air-CO 2 and shaken for 4 h at 120 strokes/min under the following conditions: condition 1 : 2.5-ml suspension, 1.75 mg protein/ml; condition 2 : 2.5-ml suspension, 3.5 mg protein/ml; and condition 3 : 5-ml suspension, 3.5 mg protein/ml. Therefore, differential oxygenation was achieved by modulating the balance between oxygen supply (diffusion across a variable distance) and oxygen consumption (variable concentrations of cells). At the end of the incubation time, P O 2, P CO 2, and pH of the liquid phase were immediately measured with a gas analyzer (Radiometer ABL 300, or ABL 5), and photomicrographs were taken. Meanwhile, the remaining suspension was centrifuged in the cold for 5 min at 50 g. The supernatant was saved at -20°C in 2% perchloric acid for metabolite determination, and the pellet was resuspended in RNA later (Ambion) for further RNA extraction.
Glucose and lactate were determined in neutralized perchloric extracts as previously described ( 33 ).
Transcriptome Analysis by RFDD
RNA extraction and double-stranded cDNA synthesis. Total RNA was extracted with the TRIzol reagent (Invitrogen, Cergy Pontoise, France), treated with RNase-free DNase (Ambion), and purified with phenol/chloroform/isoamylic alcohol (Invitrogen).
The first cDNA strand was prepared in 20 µl containing 1 µg RNA, 80 pmol oligoVdT 25, 20 pmol random octamers, 20 nmol dNTP mix, 2 µl Qthermo RT reverse transcriptase, and 1 µl RNAsin. Incubation was performed for 30 min at 37°C, 30 s at 70°C, 30 min at 37°C, 30 s at 70°C, and 5 min at 37°C. After denaturation, the RNA strand was hydrolyzed with 0.8 µl RNase H for 20 min at 37°C. The second DNA strand was synthesized in the same tube (final volume 50 µl) by addition of 5 nmol dNTP mix, 0.5 µl Taq DNA polymerase, 3 mM MgCl 2, and incubation for 6 min at 95°C, 30 s at 55°C, and 5 min at 72°C. All reagents were obtained from MP Biomedicals (Illkirch, France). Double-stranded cDNAs purified with phenol-chloroform extraction and alcohol precipitation were resuspended in 20 µl H 2 O and stored at +4°C.
Preparation of adapted fragments and discriminating PCR amplification. RFDD-PCR (Display PROfile expression profile products, MP Biomedicals) was performed as recommended by the manufacturer. Briefly, cDNA samples were fragmented with an endonuclease ( Taq1 or Hha1 ) and ligated to one of the adaptor couples specifically designed for each type of endonuclease. A couple was made of adaptor A, designed to be recognized during the RFDD-PCR by a blue fluorescent PCR primer, and adaptor B, designed to be recognized by the common part of the discriminating primers. The discriminating primers will hybridize to adaptor B and to the three nucleotides flanking the restriction site on the fragment by their discriminating sequence. Because there are 64 combinations of 3 nucleotides, there are 64 discriminating primers per endonuclease kit. RFDD-PCR touch-down amplification of the ligated fragments was made on an Eppendorf MasterCycler following the manufacturer's instructions. The PCR products were checked on a 1.7% agarose gel and diluted two- to sixfold. Five microliters were separated on a ABI 3100 capillary sequencer (Applied Biosystems) in the presence of formamide and a red fluorescent size marker (ROX 2500B, Applied Biosystems).
Analysis of fragments and data organization. After separation of the amplicons by electrophoresis, it was possible to identify each of them from its size and its discriminating triplet owing to the fact that a database of predicted fragments had been previously developed in silico. The sizes and fluorescence intensities of the separated PCR products (amplicons) were listed on a text file by GeneScan software (Applied Biosystems). The results were organized, and candidate transcripts were assigned to each amplicon using the program available on the website http://pbil.univ-lyon1.fr/software/RFDD/. The threshold for peak alignment under different conditions was 0.25 bp. The threshold for assignment of a candidate transcript was 0.6 bp. Fluorescence intensities were normalized by the total fluorescence in each electrophoresis, and in a given PCR type, the three fluorescences from three matched peaks were normalized by division by the maximal fluorescence. Subsequently, a variation in fluorescence between two of the three conditions was taken into account when 0.45.
Another program in Javascript in free access on http://pbil.univ-lyon1.fr/software/CCOV was used to crosscheck the variations of several amplicons putatively originating from the same transcript. The variations in fluorescence in response to hypoxia were sorted into five types of response: increased, or decreased under severe hypoxia, no change, decreased maximally by mild hypoxia, or increased maximally by mild hypoxia. After this sorting, a response to hypoxia was accepted if it was found in a majority of PCRs amplifying different fragments from the same transcript, or if two ends of the same fragment were amplified in the two related PCRs with the same type of response (de Laplanche E et al., unpublished observations).
Quantitative PCR
RFDD results were checked on double-stranded cDNAs from separate but similar experiments. Real-time PCR was made using a Light Cycler (Roche Diagnostics) with SYBR Green 1 (Roche Diagnostics) as a fluorescent probe. Semiquantitative PCR was performed in several tubes arrested after an increasing number of cycles, and the products were analyzed by agarose gel electrophoresis. 2 -Microglobulin and ribosomal L32 protein were used as control transcripts. The primer sequences are given in supplementary data (the online version of this article contains supplemental data).
RESULTS
Experimental Hypoxia and Normoxia
Figure 1 shows a clear-cut difference between condition 1 and condition 3, with excellent reproducibility of the difference, as attested by the Student's analysis of paired data. Condition 1 resembles normoxia (95 mmHg in human arterial blood); however, P O 2 was not sufficient to characterize oxygenation status, because P O 2 may be high in the venous blood of organs that take up oxygen at a low rate and in the medium surrounding cells that produce ATP from anaerobic glycolysis ( 3 ). As a consequence, the Pasteur effect, i.e., the shift of lactate uptake ( condition 1 ) toward lactate production and glucose uptake ( condition 3 ) was probably a better index of the oxygenation status in tubular proximal cells. Incubation was performed in 5 mM glucose and 0.5 mM lactate, i.e., the normal plasma content of glucose and lactate.
Fig. 1. Metabolic parameters in experimental hypoxia. P O 2, P CO 2, glucose consumption, and lactate production were measured after 4 h of incubation. Results are expressed as means ± SD of 4 experiments. Open bars, normoxia; gray bars, mild hypoxia; filled bars, severe hypoxia. Data were analyzed by ANOVA and Student's t -test. * P < 0.05, ** P < 0.01.
Under condition 3, despite the very low P O 2 value in the medium, one molecule of glucose was not totally metabolized into two molecules of lactate like in cultured cells ( 3 ), because, on average, 0.18 µmol glucose was taken up and only 0.05 µmol lactate was produced in the same time period ( Fig. 1 ). In addition, CO 2 content increased in parallel with that of bicarbonate, reflecting cell respiration. Therefore, some undetectable amount of oxygen must cross the liquid phase and condition 3 was not total anoxia but rather resembled ischemic hypoxia. Under condition 2, P O 2 presented an intermediate value between those of conditions 1 and 3 and the metabolite contents resembled that of condition 1. Therefore, we considered condition 2 as mild, physiological hypoxia, close to that encountered at high altitude and to that of fetal life, because, in fetal arterial blood of lambs just before birth, P O 2 is as low as 18-19 mmHg ( 26 ).
Tubule suspensions could be incubated for up to 6 h without significant changes in P O 2, in glucose consumption rate ( 28 ), and their morphology after 6 h was essentially maintained ( Fig. 2 ).
Fig. 2. Loss of tubule adhesion under hypoxia. A : tubule suspension after collagenase digestion in the presence of albumin. B : tubules incubated as described in METHODS during 6 h of severe hypoxia. C : tubules after 2 h of normoxia. D : tubules after 6 h of normoxia. E : after 2 h of mild hypoxia, tubule agglutination is already detectable.
Reproducible Synthesis of Long-Size Double-Stranded cDNA
RFDD, as well as other differential cDNA analyses, requires a good quality of RT (i.e., reproducibility in yield and in range of cDNA sizes). Moreover, the fact that RFDD is prone to display several fragments per transcript was used to confirm their variation by crosschecking the results, and therefore it was important to get cDNAs as long as possible. Good quality was obtained by using a thermo-resistant reverse transcriptase with pulses at 70°C during the synthesis of the first cDNA strand, and Taq DNA polymerase at 72°C for the synthesis of the second strand, preventing the formation of loops and of consecutive deletions in the process of cDNA synthesis. The cDNA lengths were increased by adding a minor proportion of random primers to oligoVdT primers. As a result, agarose gel analysis showed five similar smears, each with a wide range of sizes stretching to 2,652 bp and more ( Fig. 3 ). The quantitative reproducibility of the method was checked in purified cDNAs by real-time PCR of 2 -microglobulin reverse transcribed from transcripts naturally present in tissue RNA ( Fig. 3 ). Moreover, a synthetic and nonhomologous poly(A) standard mRNA (SmRNA; patent no. 03290958.2, April 17, 2003) was added to the RNA sample before five different RT reactions, and the resulting cDNAs were checked in the five purified reaction products by real-time PCR. The good reproducibility of the cDNA synthesis was assessed by crossing point values varying by no more than ±0.56% (SD; n = 5; Fig. 3 ).
Fig. 3. Reproducible method of RT yielding long-size cDNAs. Total RNAs were extracted from a tubule suspension and treated to eliminate DNA, as described in METHODS. From 1 RNA sample, 5 separate 1-µg aliquots into which were added 80 pg SmRNA as a nonhomologous external standard were reverse transcribed, duplicated in double-stranded cDNAs, and purified with phenol-chloroform, as described in METHODS. A : purified cDNA separated in 1.5% agarose gel. M, size marker. Lanes 1-5, purified cDNAs from 5 separate reactions. B : purified cDNAs were tested by real-time PCR amplifying the external standard (reverse transcribed from SmRNA) or 2 -microglobulin (m) reverse transcribed from endogenous RNAs. In each PCR, 4 µl of DNA were diluted 40-fold.
High-Scale Analysis of Transcript Changes After 4 h of Hypoxia
Double-stranded cDNAs were prepared from tubule suspensions incubated for 4 h under the three different conditions as described above. cDNAs from each condition were fragmented with two different endonuclease kits ( Taq 1 and Hha1, respectively) and processed as indicated in METHODS. Eighty-three discriminating RFDD-PCRs of 128 provided satisfactory results under the 3 experimental 50 peaks/electrophoregram, similar number of peaks ( n = ±15%), and a similar range of sizes. Among the 4,582 transcripts present in RFDD-Base (version 1) for Rattus norvegicus, 3,853 were proposed as candidates, and, after crosschecking of the variations, the response to hypoxia of 1,533 transcripts was retained as valid (de Laplanche E et al., unpublished observations). The file of whole results can be accessed at http://pbil.univ-lyon1.fr/datasets/delaplanche2005.
The response to hypoxia of 33 transcripts was checked by quantitative PCR (qPCR), and 29 were confirmed (this study and de Laplanche E, unpublished observations). The tables show the variations of 85 transcripts selected for their relevance to EMT, energy production, and cell-cell interactions, among which 26 have been validated by qPCR ( Fig. 4 ).
Fig. 4. Confirmation of restriction fragment differential display results by quantitative PCRs. Values are means ± SD of 4 determinations in 3 separate experiments ( a, b, and c, respectively) analyzed by real-time PCR (histograms), and representative gels of semiquantitative PCRs, with the no. of cycles indicated above each lane. Open bars, normoxia; gray bars, mild hypoxia; filled bars, severe hypoxia; N, normoxia; M, mild hypoxia; S, severe hypoxia. A : kidney-specific transcripts [ Table 1; SGLT2 ( Slc5a2 ); heparanase A ( Hpse ); CYP1 ( Cyp27b1 ); UT3 ( Slc14a1 ); LPK ( Pklr ); KCC2 ( Slc12a5 )]. B : muscle-related transcripts [ Table 2; myogenin ( Myog ); myodulin ( Tnmd )]. C : tissue cohesion [ Table 3; fibronectin ( Fn1 ); N-cadherin ( Cdh2 ); leprecan ( Lepre1 ); ankyrin 3 ( Ank3 ); gel A ( Mmp2 ); gel B ( Mmp9 )]. D : energy production, substrate supply [ Table 4; aldolase A ( Aldoa ); PGM ( Pgm1 ); MALO COA: Mlycd; oxo glut: 2-oxoglutarate carrier; RGD:708476 ); MCT2 ( Slc16a7 )]. AHR, aryl hydrocarbon receptor; SDHA, succinate dehydrogenase A.
Five possible types of variations were retrieved, including biphasic responses. The chosen stringent threshold of variation permitted the discrimination between the transcripts that did not change and those that varied under hypoxia. The transcripts whose level was not changed by hypoxia were frequently those encoding abundant and ubiquitous proteins such as matrix metalloprotease (MMP-1; Pmpca gene, see Table 4 ), one of the protease-maturing proteins imported into the mitochondrion, succinate dehydrogenase A (SDHA; Sdha gene), and cytochrome c oxidase 2, subunits of respiratory chain complexes II and IV, and the 1 -subunit of the plasma membrane Na-K-ATPase ( Atp1a1 ) (see Table 4 ). Among the transcripts whose level was not changed by hypoxia was also found a molecule encoding a kidney-specific transporter, the Na-driven glucose transporter SGLT2 ( Slc5a2 ), and nonmuscle myosin heavy chain-B ( Myh10, see Table 2 ). As shown by qPCR ( Fig. 4 ), the relative abundances of these unchanged mRNAs were generally high because their levels were close to that of the transcript encoding ribosomal protein L32.
Table 1. Response to hypoxia: kidney-specific or heterogen-in-nephron transcripts
Table 2. Response to hypoxia: transcripts whose products are relevant to epithelial-mesenchymal transition
Table 3. Response to hypoxia: extracellular matrix, cell adhesion, and cytoskeleton
Table 4. Response to hypoxia: energy production and storage and related products
Among 1,533 transcripts, 128 transcripts were maximally increased by mild hypoxia, i.e., their levels were the lowest both in severe hypoxia and in normoxia. Symmetrically, 20 transcripts were decreased by mild hypoxia. qPCR confirmed the variation of the three tested transcripts whose level was increased by mild hypoxia ( Fig. 4 ). Therefore, there seem to be two different levels of oxygen sensing. Intermediate hypoxia, close to fetal values, favors expression of myogenin (see Table 2 and Fig. 4 ), a transcription factor that had been shown to be transiently expressed during development and to be absent from fully differentiated muscle cells ( 5 ). The 2-oxoglutarate carrier was also maximally increased by an intermediate condition of oxygenation (see Table 4 and Fig. 4 ).
Proximal tubule-specific transcripts but not distal-specific transcripts were maximally expressed under normoxia. Table 1 shows that several mRNAs normally restricted to proximal tubules were decreased by hypoxia. The transcript of Cyp27b1, which encodes the final maturating enzyme for active vitamin D, 25-hydroxyvitamin D 1-hydroxylase, was decreased by hypoxia, and the result was confirmed by qPCR ( Table 1 and Fig. 4 ). So were the transcripts of Pklr, encoding the liver-type pyruvate kinase that is normally expressed in proximal tubule ( Table 1; confirmed by qPCR as shown in Fig. 4 ), and that of Atp60a1, which encodes a lysosomal proton-transporting ATPase subunit, involved in the very active endocytic degradation of proteins in proximal tubule ( Table 1 ). Three transcripts involved in solute transport in this part of the nephron were found by RFDD; however, at variance with the preceding transcripts, they were not affected by hypoxia ( Table 1 ). Table 1 also shows that the transcripts activated by hypoxia were not always related to EMT because, typically epithelial transcripts, kidney specific, were maximally expressed under hypoxia. The furosemide-sensitive Na-K-Cl transporter, restricted to the thick ascending limb of Henle, and UT3, a urea transporter isoform, whose transcripts were increased by hypoxia, are normally present in the medullary zone and absent from the cortex for UT3. The response to hypoxia of UT3 was confirmed by qPCR ( Fig. 4 ). Hypoxia also increased Scnn1g, an isoform of the amiloride-sensitive sodium channel of the distal nephron, cortical and medulla segments ( Table 1 ).
Myosins and muscle differentiation. Eight transcripts of myosin isoforms were found by RFDD. The transcript abundance of nonmuscle myosin heavy chain A ( Myh9 gene), which is normally expressed in the kidney, and heavy chain MYR 1 ( Myo1b gene) were decreased by hypoxia. Conversely, those of muscle-type myosin II heavy chain ( Myh1 gene) and light chain ( Mlc3 gene) from fast-twitch skeletal muscle were increased under severe hypoxia ( Table 2 ). This represented a general shift of myosin isoform transcripts toward muscle-type expression, with the exception of Myh10, for which the transcript level did not change.
The transcript of myodulin, a muscle-specific angiogenic protein, was increased by hypoxia, contributing to the general scheme of a transcriptome closer to a muscle type.
ECM, related products, and transdifferentiation factors. After only 4 h of hypoxia, there were strikingly consistent modifications in the abundance of transcripts encoding the ECM, its processing enzymes, and proteins involved in cell adhesion ( Table 3 ). Gelatinases A and B (MMP-2, Mmp2 gene and MMP-9, Mmp9 gene) and heparanase ( Hpse gene) were overexpressed under hypoxia, whereas fibronectin ( Fn1 gene) was less expressed, suggesting remodeling of the ECM. N-cadherin mRNA ( Cdh2 gene), encoding a cell adhesion protein that is normally present in the kidney, was lowered by hypoxia, whereas hypoxia increased transcripts involved in inhibition of cell-cell interaction such as those encoding ADAMTS-1 ( Adamts1 gene), a disintegrin/metalloprotease expressed during kidney development and localized to the medullary collecting duct in adults, and CASPR ( Cntnap1 gene), a contactin-associated protein. The cell junction-related desmin mRNA ( Des gene) remained unchanged, along with barmotin ( Bartomin gene), a tight junction component. Taken together, these results show that under severe hypoxia, there is a general increase in the transcripts involved in ECM disruption or remodeling, and a general decrease in the transcripts involved in ECM tightness and tissue cohesion.
Transcripts of TGF- 2 and -3 were overexpressed under hypoxia, in agreement with the observations made in other models of hypoxia ( 32, 45 ). It should be mentioned that variations in TGF- 1 could not be detected by RFDD because its cDNA does not harbor restriction sites for Taq1 or for Hha1. In addition to TGF, the product of Fgfr1, the fibroblast growth factor receptor 1 ( -isoform) transcript, was overexpressed under hypoxia.
Energy production and storage, substrate and oxygen supply. The transcripts involved in energy production and intermediary metabolism, mostly enzymes ( Table 4 ), were less sensitive to hypoxia than those involved in tissue structure cited above. However, the mRNAs encoding the angiogenic factor VEGF-C ( Vegfc gene), the glycolytic enzyme aldolase A ( Aldoa gene), and the lactate transporter MCT2 ( Slc16a7 gene) were increased by hypoxia. Aldolase A is a well-known hypoxia-induced transcript ( 32 ). The increase in MCT2 mRNA during hypoxia could be related to a transition to a collecting duct tissue type, because this latter part of the nephron is able to express MCT2 ( 10 ). Less expected were the changes in two transcripts involved in muscle-type energy storage, encoding, respectively, creatine kinase B and L -arginine: glycine amidinotransferase ( Gatm gene), which is involved in creatine synthesis ( Table 4 ). The two latter transcripts were overexpressed under hypoxia, favoring the hypothesis of EMT with a new type of marker. Creatine kinase M transcript ( Ckm gene), which is preferentially expressed in oxidative-type muscles, was decreased under hypoxia. CKM is thought to be involved in fine regulation of ATP levels more than in storage of energy under the form of creatine phosphate ( 40 ).
Loss of Tubule Adhesion During Hypoxia
Microscopy showed tubule agglutination in normally oxygenated tubule suspensions, whereas hypoxic tubules remained well dissociated ( Fig. 2 ). Under all conditions, tubules remained birefringent. Agglutination was not due to a lower concentration of total proteins under normoxia because the protein concentration was identical under mild hypoxia or severe hypoxia, and the tubule suspension nevertheless agglutinated much more under mild hypoxia, as soon as 2 h after the beginning of the incubation period ( Fig. 2 ). Tubule or cell suspensions are generally incubated in the presence of fatty acid free albumin to prevent the cells from clustering together, and to bind by-products of degradation. However, in our experiments, incubation of tubules was done in the absence of albumin because it had been shown to induce tubular defects ( 4 ), and this disclosed the dissociating effect of hypoxia. In this latter condition, however, cell junctions were not disrupted and tubular structures were conserved ( Fig. 2 ).
DISCUSSION
This is the first high-scale report of hypoxia's effects on freshly isolated tubules. We chose to use this ex vivo model rather than cultured cells ( 14 ) to prevent the well-known adaptations of tubular cell phenotype to nonstirred medium supplemented with growth factors ( 1, 3 ). Another frequently used model for transcriptome studies is in vivo ischemia and reperfusion ( 17 ), but its drawback is that the level of oxygenation is not controlled. The shortcoming of tubule suspensions is that they can be maintained for but a few hours and therefore cannot model the whole cascade of events putatively leading to fibrosis. However, the first step in the cascade is crucial, and, in this respect, it was mandatory to begin by describing early changes under physiological hypoxia. The control of oxygenation status in this study resulted in the finding that even moderate hypoxia may induce profound changes in mRNA levels and that ex vivo models of kidney cells should be provided with high amounts of oxygen.
Owing to the good reproducibility of our model, we could test the soundness of the results observed by RFDD because, if RFDD analysis can be obtained with only 1-2 µg RNA, confirmation of the results by qPCR requires much higher amounts of material. An abundant supply of oxygen to tubule suspensions resulted in metabolic features normally observed in freshly isolated kidney cortex and in in vivo measurements ( 18 ), and the RFDD data from the normoxic condition reflected the expected tubule transcriptome, similar in several points to that described by serial analysis of gene expression in freshly dissected nephrons from human kidney ( 6 ).
The transcriptome method we chose was not as quantitative as serial analysis of gene expression; however, it presented the combined advantages of a relatively low cost, good capacity to find unexpected and rare transcripts, and its low level of false positive and of false negative because the unchanged transcripts could be reported and confirmed.
One of the main findings of the present study is that the very moderate hypoxia of 40-60 mmHg is able to favor the expression of a subset of transcripts less expressed both in normoxia and in severe hypoxia. These include enzymes of the intermediary metabolism and a transcription factor, myogenin, known to be involved in fetal development of muscle cells but absent in adult muscle. The mechanism underlying a biphasic response to oxygen could be that two or several sensors are recruited at different levels of oxygen. Previous studies have already reported different levels of responses to hypoxia. The Chandel group ( 8 ) has shown in cultured fibroblasts that the hypoxia-inducible factor-1 system is activated through different mechanisms according to the level of oxygen deprivation: 1.5 or 0% O 2. In vivo, angiogenesis during placental development has been shown to be controlled by different factors, probably in relation to the levels of hypoxia: during early placental development, branching angiogenesis occurs under hypoxic conditions, recruiting the VEGF system, whereas during late pregnancy, nonbranching angiogenesis takes place under less hypoxic conditions under the control of PlGF ( 26 ).
The second important finding is that physiological oxygenation status is required for fully differentiated expression of kidney-specific transcripts, this specifically in the proximal tubule. New oxygen-regulated transcripts were identified such as those encoding vitamin D final maturing enzyme and Pklr. 25-Hydroxyvitamin D 1-hydroxylase, the product of Cyp27b, is a monooxygenase and could be regulated by oxygen not only at the substrate level, but also at the transcript level. The so-called liver-type pyruvate kinase ( Pklr ) is in fact also expressed in other glucose-synthesizing organs such as the kidney proximal tubule and small intestine ( 43 ). In the liver, Pklr is modulated by the relative levels of glucose and oxygen ( 13 ), but in the kidney it is sensitive to glycerol rather than to glucose ( 43 ), and data on the effect of oxygen were lacking.
The fact that transcripts normally restricted to the medullary zone, such as those encoding UT3 and hepatocyte nuclear factor-3, were increased by hypoxia in our cortical preparation was unexpected because, during dissection, the medulla was largely removed from the cortex. However, it could have a biological meaning if we consider that this zone is normally less oxygenated than the cortex and that some transcription factors such as hypoxia-inducible factor-1 ( 46 ) are restricted to this zone. The suggested hypothesis is that the highly specialized functions of the different segments of the nephron are modulated, in part, by tissue oxygenation and that an epithelial cell can undergo a transition toward another epithelial phenotype. More unexpected was the fact that transcripts of the amiloride-sensitive sodium transporter and furosemide-sensitive K-Cl transporter, which are normally present in the cortical distal tubule, were increased by hypoxia. The transcript of heparanase also belongs to this relationship. This suggested that the proximal tubule's functions are more sensitive to hypoxia than those of the distal nephron.
In parallel, hypoxia increased the level of other transcripts that are related to an EMT transition toward myofibroblasts, in agreement with previous in vitro and in vivo studies ( 19 ). Nonmuscle myosin mRNAs were decreased under hypoxia, whereas skeletal muscle-type myosin mRNAs and troponin mRNA were induced by hypoxia. Myogenin transcript expression was increased by mild hypoxia. This was strikingly similar to the observations of other authors showing, in myofibroblasts, increased sarcomeric myosin expression, together with that of MyoD, myogenin, and troponin mRNAs ( 20 ). However, in this study, at this stage of hypoxia, there was not a true transdifferentiation because tubule structure and morphology remained well maintained and the changes were just initiated. Besides, if there was a transition toward a glycolytic muscle type of energy storage, creatine phosphate, it could represent a positive response to facing transient periods of hypoxia, together with the increase in transcripts involved in glycolysis.
EMT also involves changes in tissue cohesion, which is essential for maintaining the highly organized morphology of nephrons with their surrounding vasculature, and in cell-cell interactions, necessary for epithelial barrier organization. It has been shown that an increase in gelatinase A/MMP-2 was sufficient to induce EMT ( 7 ); however, data from the literature are still needed to clarify the effect of hypoxia on MMP-2 levels: increase or decrease ( 9, 15 )? Hypoxia's effect could depend on its duration and level ( 9, 15 ). This study clearly shows that early hypoxia decreased transcripts involved in tissue cohesion, such as fibronectin, and increased those of ECM remodeling, such as gelatinase A, whereas transcripts involved in cell-cell junctions were not changed. In agreement, tubule suspensions that naturally aggregated in clusters under normoxia were well dissociated in separate tubules under hypoxia. Heparanase, an ECM-degrading enzyme, is normally found in the inner medulla and in distal parts of the nephron ( 16 ), and an increased level of its transcript could be involved both in ECM remodeling and invasion ( 24 ), and in release of growth factors trapped in ECM ( 30 ), therefore favoring tissue regeneration and tumor invasion. Fibronectin, that plays a pivotal role in cell adhesion to ECM ( 41 ), is required for tumor suppression by von Hippel-Lindau protein ( 36 ) to which it directly binds, and it is absent from von Hippel-Lindau protein-deficient renal tumors ( 11 ). Its mRNA level had been previously reported to be insensitive to extreme hypoxia in nonstirred cultured cell lines ( 2, 42 ), but in the present study its transcript was downregulated by early physiological hypoxia. TGF- has been proven to be a major factor of EMT in several tissues either during normal development or during the fibrotic process, especially in kidney disease ( 35, 45 ); however, TGF- protein only increased after 24 h of hypoxia ( 22 ). This suggests that the early effects observed in the present study could hardly be relevant to TGF- protein signaling, even if hypoxia increased the transcript levels of TGF- 2 and -3.
Disruption of the epithelial barrier by hypoxia has been reported in other models of very severe ischemia with ATP depletion ( 21, 25, 44 ) or unstirred cultured cells ( 14 ), whereas in our model with no ATP depletion ( 33 ) cell junctions were essentially maintained. Therefore, ECM changes in mRNA levels appear early during physiological hypoxia but remain modest: they could trigger or favor invasion of the tissue by other cell types such as macrophages, fibroblasts or tumor cells, without altering the tubular barrier, whereas normoxia favors ECM formation and could be required to maintain the well-organized morphology of the kidney cortex.
In conclusion, an abundant supply of oxygen seems an important factor for a fully differentiated phenotype of the proximal cortical tubule in the kidney, and a global approach can provide clues for a better understanding. The proximal part of the nephron is formed during embryogenesis from mesenchymal cells that undergo mesenchymal-to-epithelial transition ( 27, 31 ), whereas the collecting duct originates from the epithelial-type ureteric bud. It is therefore tempting to hypothesize that proximal tubular cells easily revert to a mesenchymal, fetal type of tissue under hypoxia. Proximal phenotype shift toward glycolysis and storage of energy as creatine phosphate could favor adaptation to transient periods of hypoxia, but loss of tissue adhesion in the adult quiescent kidney could permit a deleterious tissue invasion by fibroblasts, tumor cells, and new vessels. In addition, the differential hypoxia sensitivity of kidney functions described in this study provides insight into how nephron heterogeneity may be generated and, sometimes, impaired. Investigations are now in progress to understand the mechanisms involved in some of the changes disclosed in this study.
GRANTS
This work was supported by funds from the Centre National de Recherche Scientifique, the UCBL (University of Lyon 1) and its subdivision, the Institut Fédératif de Recherche 41; the Région Rhône-Alpes (project Emergence); French League against Cancer (Rhône and Loire committees); and the Institut National pour la Santé et la Recherche Médicale (project Vieillissement). E. de Laplanche was the recipient of a grant from the League against Cancer (Limousin).
DISCLOSURES
We acknowledge financial support from Qbiogene, Inc. (now MP Biomedicals, Irvine, CA), essentially in the form of free reagents and mutually helpful friendly technical discussions. However, our work was undertaken independently, and publication of the presented results was by no means submitted for their approval.
ACKNOWLEDGMENTS
The authors are greatly indebted to Nancy Uhrhammer (Laboratoire d'Oncologie Moléculaire, Centre Jean Perrin, Clermont-Ferrand, France) for skillful analysis of fragments and acknowledge Olivier Levillain for expert reading of the RFDD results.
【参考文献】
Aleo MD and Schnellmann RG. Regulation of glycolytic metabolism during long-term primary culture of renal proximal tubule cells. Am J Physiol Renal Fluid Electrolyte Physiol 262: F77-F85, 1992.
Bluyssen HA, Lolkema MP, van Beest M, Boone M, Snijckers CM, Los M, Gebbink MF, Braam B, Holstege FC, Giles RH, and Voest EE. Fibronectin is a hypoxia-independent target of the tumor suppressor VHL. FEBS Lett 556: 137-142, 2004.
Bolon C, Gauthier C, and Simonnet H. Glycolysis inhibition by palmitate in renal cells cultured in a two-chamber system. Am J Physiol Cell Physiol 273: C1732-C1738, 1997.
Bruzzi I, Benigni A, and Remuzzi G. Role of increased glomerular protein traffic in the progression of renal failure. Kidney Int Suppl 62: S29-S31, 1997.
Buckingham M, Bajard L, Chang T, Daubas P, Hadchouel J, Meilhac S, Montarras D, Rocancourt D, and Relaix F. The formation of skeletal muscle: from somite to limb. J Anat 202: 59-68, 2003.
Chabardes-Garonne D, Mejean A, Aude JC, Cheval L, Di Stefano A, Gaillard MC, Imbert-Teboul M, Wittner M, Balian C, Anthouard V, Robert C, Segurens B, Wincker P, Weissenbach J, Doucet A, and Elalouf JM. A panoramic view of gene expression in the human kidney. Proc Natl Acad Sci USA 100: 13710-13715, 2003.
Cheng S and Lovett DH. Gelatinase A (MMP-2) is necessary and sufficient for renal tubular cell epithelial-mesenchymal transformation. Am J Pathol 162: 1937-1949, 2003.
Emerling BM, Platanias LC, Black E, Nebreda AR, Davis RJ, and Chandel NS. Mitochondrial reactive oxygen species activation of p38 mitogen-activated protein kinase is required for hypoxia signaling. Mol Cell Biol 25: 4853-4862, 2005.
Fine LG, Orphanides C, and Norman JT. Progressive renal disease: the chronic hypoxia hypothesis. Kidney Int 65: S74-S78, 1998.
Garcia CK, Brown MS, Pathak RK, and Goldstein. JL. cDNA cloning of MCT2, a second monocarboxylate transporter expressed in different cells than MCT1. J Biol Chem 270: 1843-1849, 1995.
He Z, Liu S, Guo M, Mao J, and Hughson MD. Expression of fibronectin and HIF-1alpha in renal cell carcinomas: relationship to von Hippel-Lindau gene inactivation. Cancer Genet Cytogenet 152: 89-94, 2004.
Iwano M, Plieth D, Danoff TM, Xue C, Okada H, and Neilson EG. Evidence that fibroblasts derive from epithelium during tissue fibrosis. J Clin Invest 110: 341-350, 2002.
Kietzmann T, Krones-Herzig A, and Jungermann K. Signaling cross-talk between hypoxia and glucose via hypoxia-inducible factor 1 and glucose response elements. Biochem Pharmacol 64: 903-911, 2002.
Leonard MO, Cottell DC, Godson C, Brady HR, and Taylor CT. The role of HIF-1 alpha in transcriptional regulation of the proximal tubular epithelial cell response to hypoxia. J Biol Chem 278: 40296-40304, 2003.
Leufgen H, Bihl MP, Rudiger JJ, Gambazzi J, Perruchoud AP, Tamm M, and Roth M. Collagenase expression and activity is modulated by the interaction of collagen types, hypoxia, and nutrition in human lung cells. J Cell Physiol 204: 146-154, 2005.
Levidiotis V, Kanellis J, Ierino FL, and Power DA. Increased expression of heparanase in puromycin aminonucleoside nephrosis. Kidney Int 60: 1287-1296, 2001.
Liang M, Cowley AW, and Greene AS. High throughput gene expression profiling: a molecular approach to integrative physiology. J Physiol 554: 22-30, 2004.
Mandel LJ. Metabolic substrates, cellular energy production, and the regulation of proximal tubular transport. Annu Rev Physiol 47: 85-101, 1985.
Manotham K, Tanaka T, Matsumoto M, Ohse T, Inagi R, Miyata T, Kurokawa K, Fujita T, Ingelfinger JR, and Nangaku M. Transdifferentiation of cultured tubular cells induced by hypoxia. Kidney Int 65: 871-880, 2004.
Mayer DC and Leinwand LA. Sarcomeric gene expression and contractility in myofibroblasts. J Cell Biol 139: 1477-1484, 1997.
Molitoris BA, Dahl R, and Hosford M. Cellular ATP depletion induces disruption of the spectrin cytoskeletal network. Am J Physiol Renal Fluid Electrolyte Physiol 271: F790-F798, 1996.
Nakagawa T, Lan HY, Zhu HJ, Kang DH, Schreiner GF, and Johnson RJ. Differential regulation of VEGF by TGF- and hypoxia in rat proximal tubular cells. Am J Physiol Renal Physiol 287: F658-F664, 2004.
Ng YY, Huang TP, Yang WC, Chen ZP, Yang AH, Mu W, Nikolic-Paterson DJ, Atkins RC, and Lan HY. Tubular epithelial-myofibroblast transdifferentiation in progressive tubulointerstitial fibrosis in 5/6 nephrectomized rats. Kidney Int 54: 864-876, 1998.
Parish CR, Freeman C, and Hulett MD. Heparanase: a key enzyme involved in cell invasion. Biochim Biophys Acta 1471: M99-M108, 2001.
Racusen LC. Alterations in human proximal tubule cell attachment in response to hypoxia: role of microfilaments. J Lab Clin Med 123: 357-364, 1994.
Regnault TR, de Vrijer B, Galan HL, Davidsen ML, Trembler KA, Battaglia FC, Wilkening RB, and Anthony RV. The relationship between transplacental O 2 diffusion and placental expression of PlGF, VEGF and their receptors in a placental insufficiency model of fetal growth restriction. J Physiol 550: 641-656 2003.
Ribes D, Fischer E, Calmont A, and Rossert J. Transcriptional control of epithelial differentiation during kidney development. J Am Soc Nephrol 14, Suppl 1: S9-S15, 2003.
Rossignol F, de Laplanche E, Mounier R, Bonnefont J, Cayre A, Godinot C, Simonnet H, and Clottes E. Natural antisense transcripts of HIF-1alpha are conserved in rodents. Gene 339: 121-130, 2004.
Rowell LB. Human cardiovascular adjustments to exercise and thermal stress. Physiol Rev 54: 75-159, 1974.
Sanderson RD, Yang Y, Kelly T, Macleod V, Dai Y, and Theus A. Enzymatic remodeling of heparan sulfate proteoglycans within the tumor microenvironment: growth regulation and the prospect of new cancer therapies. J Cell Biochem 96: 897-905, 2005.
Saxen L. What is needed for kidney differentiation and how do we find it? Int J Dev Biol 43: 377-380, 1999.
Semenza G. Signal transduction to hypoxia-inducible factor 1. Biochem Pharmacol 64: 993-998, 2002.
Simonnet H. Glucose-6-phosphatase mRNA levels in kidney isolated tubule suspensions are increased by dexamethasone and decreased by insulin. Metabolism 48: 1052-1056, 1999.
Singh MV, Salhan AK, Rawal SB, Tyagi AK, Kumar N, Verma SS, and Selvamurthy W. Blood gases, hematology, and renal blood flow during prolonged mountain sojourns at 3500 and 5800 m. Aviat Space Environ Med 74: 533-536, 2003.
Stahl PJ and Felsen D. Transforming growth factor-beta, basement membrane, and epithelial-mesenchymal transdifferentiation: implications for fibrosis in kidney disease. Am J Pathol 159: 1187-1192, 2001.
Stickle NH, Chung J, Klco JM, Hill RP, Kaelin WG Jr, and Ohh M. pVHL modification by NEDD8 is required for fibronectin matrix assembly and suppression of tumor development. Mol Cell Biol 24: 3251-3261, 2004.
Thews O, Wolloscheck T, Dillenburg W, Kraus S, Kelleher DK, Konerding MA, and Vaupel P. Microenvironmental adaptation of experimental tumours to chronic vs. acute hypoxia. Br J Cancer 91: 1181-1189, 2004.
Vaupel P. Is there a critical tissue oxygen tension for bioenergetic status and cellular pH regulation in solid tumors? Experientia 52: 464-468, 1996.
Vaupel P and Mayer A. Hypoxia and anemia: effects on tumor biology and treatment resistance. Transfus Clin Biol 12: 5-10, 2005.
Ventura-Clapier R, Kuznetsov A, Veksler V, Boehm E, and Anflous K. Functional coupling of creatine kinases in muscles: species and tissue specificity. Mol Cell Biochem 184: 231-247, 1998.
Wierzbicka-Patynowski I and Schwarzbauer JE. The ins and outs of fibronectin matrix assembly. J Cell Sci 116: 3269-3276, 2003.
Wykoff CC, Sotiriou C, Cockman ME, Ratcliffe PJ, Maxwell P, Liu E, and Harris AL. Gene array of VHL mutation and hypoxia shows novel hypoxia-induced genes and that cyclin D1 is a VHL target gene. Br J Cancer 90: 1235-1243, 2004.
Yamada K and Noguchi T. Nutrient and hormonal regulation of pyruvate kinase gene expression. Biochem J 337: 1-11, 1999.
Ye J, Tsukamoto T, Sun A, and Nigam SK. A role for intracellular calcium in tight junction reassembly after ATP depletion-repletion. Am J Physiol Renal Physiol 277: F524-F532, 1999.
Yu HT. Progression of chronic renal failure. Arch Intern Med 163: 1417-1429, 2003.
Zou AP, Yang ZZ, Li PL, and Cowley AWJR. Oxygen-dependent expression of hypoxia-inducible factor-1 in renal medullary cells of rats. Physiol Genomics 6: 159-168, 2001.
作者单位：1 Centre de Génétique Moléculaire et Cellulaire, Unit 553 2 Laboratoire de Biométrie et Biologie Evolutive, Unit 555 and 3 Laboratoire de Physiologie Intégrative Cellulaire et Moléculaire, Unit 512 Centre National de la Recherche Scientifique and University