1 Department of Physiology, University Oklahoma Health Sciences Center, Oklahoma City, Oklahoma 73190
2 Department of Internal Medicine, University of Texas Medical Branch, Galveston, Texas 77555
3 Tulane Environmental Astrobiology Center, and Nephrology Section, Tulane University Medical Center, and Veterans Affairs Medical Center, New Orleans, Louisiana 70112
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
gene array; mouse model of disease; urinary bladder inflammation; interstitial cystitis
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Among the possible mechanisms by which LPS induces cystitis, we found that this endotoxin induces upregulation of pro-inflammatory peptide receptors such as bradykinin-1 (B1) (5) and neurokinin-1 (NK-1), the primary substance P receptor (45). During inflammation, B1 receptors are induced by interleukin-1ß (IL-1ß) (26). In addition, both B1 and NK-1 receptor upregulation is dependent on translocation of the transcription factor nuclear factor-B (NF-
B) (26, 45). Additional evidence of NK-1 involvement in response to LPS is the finding that desensitization of terminal sensory nerve endings by reducing the release of substance P decreases bladder hyperreflexia secondary to LPS (22). Moreover, treatment of animals with NK-1 receptor antagonists abrogates tumor necrosis factor-
(TNF-
) transcription and secretion in response to LPS (9). Last, the NK-1 receptor is required in antigen-induced cystitis, where there is participation of these receptors in the chain of events linking mast cell degranulation and inflammation (33).
In addition to upregulation of peptide receptors, LPS is thought to be involved in the recruitment of inflammatory cells, upregulation of cytokine production, adhesion molecules, inducible nitric oxide synthase (iNOS), and cyclooxygenase-2 (COX-2), all of which contribute to the pathology of cystitis. Most of these effects are dependent on NF-B, which plays a central role in bladder responses to inflammation and infection (45). We demonstrated that lactacystin, a proteosome inhibitor, by reducing NF-
B activation, impaired LPS-induced NK-1 receptor expression and inhibited bladder inflammation (45).
However, as our database on the repertoire of inflammatory mediators implicated in bladder inflammation increases, the central mechanisms by which LPS can induce inflammation, cell damage, and bladder injury often become less rather than more clear. For us to make sense of the vast knowledge of the genes and proteins involved in LPS binding, transport, signal transduction, and induced transcription (17, 34, 38, 41, 42, 4749) may require analysis of the patterns of change far more than definition of additional members of the inflammatory cascades.
The Human Genome Project (13) and the associated production of huge libraries of expressed sequence tags provide momentum for the development of new methods to assay large numbers of genes simultaneously (11, 24, 31, 35; see also, http://www.affymetrix.com, http://www.clontech.com, and http://www.incyte.com/reagents/gem/products.shtml). Gene expression analysis by microarrays has provided a rapid, inexpensive, but sophisticated method to meet these needs (1, 10). More importantly, studies of clusters of genes that have similar expression changes over time now allow the definition of functionally meaningful expression patterns.
In this study, we apply the power of time-dependent gene array cluster analysis to define gene expression patterns in the early stages of a model of genitourinary inflammation. Gene array analysis of the inflammatory response allows us to simultaneously observe the time course of LPS-induced regulation of a large number of genes in the mouse urinary bladder. Supporting our hypothesis, the results show that LPS induces a time-dependent upregulation of several different pro-inflammatory gene clusters, as well as changes in the NF-B transcriptional factor pathway.
![]() |
MATERIAL AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Morphological analysis.
The urinary bladder was evaluated for inflammatory cell infiltrates and the presence of interstitial edema. A semiquantitative score using defined criteria of inflammation severity was used to evaluate cystitis (33). A cross section of bladder wall was fixed in formalin, dehydrated in graded alcohol and xylene, embedded in paraffin, and cut serially into four 5-µm sections (8 µm apart) to be stained with hematoxylin and eosin. Histology slides were scanned using a CoolSNAP camera (RS Photometrics, Tucson, AZ) mounted on an Olympus microscope. Image analysis was performed using a MetaMorph Imaging System (Universal Imaging, West Chester, PA). The severity of lesions in the urinary bladder was graded as follows: 1+, mild (infiltration of a low number of neutrophils in the lamina propria, and little or no interstitial edema); 2+, moderate (infiltration of moderate numbers of neutrophils in the lamina propria, and moderate interstitial edema); 3+, severe (diffuse infiltration of moderate to large numbers of neutrophils in the lamina propria and severe interstitial edema).
Sample preparation for cDNA expression arrays.
Three bladders from each group were homogenized together in Ultraspec RNA solution for isolation and purification of total RNA. Mouse bladders were pooled to ensure sufficient RNA for gene array analysis. RNA was treated with DNase according to manufacturers instructions (Clontech Laboratories, Palo Alto, CA), and the quality of 10 µg RNA was evaluated by denaturing formaldehyde/agarose gel electrophoresis.
Mouse cDNA expression arrays.
cDNA probes prepared from DNase-treated RNAs obtained from each of the experimental groups were hybridized simultaneously to four membranes containing Atlas mouse cDNA expression arrays (Clontech). Briefly, 5 µg of DNase-treated RNA was labeled with [-32P]dATP and reverse transcribed to cDNA, according to the manufacturers protocol (Clontech). The radioactively labeled complex cDNA probes were hybridized overnight to cDNA expression arrays (Clontech) using ExpressHyb hybridization solution with continuous agitation at 68°C. After two high-stringency washes, the hybridized membranes were exposed to a phosphor imaging screen overnight (Cyclone storage system; Packard BioScience, Downers Grove, IL). The membranes were also exposed to X-ray film at -80°C with an intensifying screen for various lengths of time, to determine the optimal exposure to generate equally intense hybridization signals for the housekeeping genes. Exposed X-ray film was scanned with Color OneScanner (Apple Computer, Cupertino, CA) to Adobe Photoshop software. The number of experiments (n) represents hybridization of the homogenate of three individual mouse bladders with one cDNA expression array.
Data processing.
The phosphor imaging screen contains phosphor crystals that absorb the energy emitted by the radioactivity of the sample and re-emit that energy as a blue light when excited by a red laser. Results are presented as digital light units (DLU) and were interpreted by using OptiQuant image analysis software (Packard BioScience). Quantification of each detectable band was performed from the DLUs generated by OptiQuant. The background was subtracted, and within each membrane expression was calculated as percentage of ubiquitin, and the average was used for cluster analysis.
Gene clustering.
Cluster analysis was performed using self-organizing maps (SOMs) as described by Tamayo et. al. (39). SOMs are a type of mathematical cluster analysis that is particularly well suited for recognizing and classifying features in complex, multidimensional data (39).
Expression data were analyzed as described (39), including thresholding small and negative expression values to 20. Genes presenting a similar time-dependent peak expression were identified in the mouse bladder experiments based on a Euclidean distance metric, after eliminating genes that failed to vary in expression level within an experiment by a factor of three and an absolute value of 100% of ubiquitin and normalizing within experiments to a mean of 0 and a standard deviation of 1.
Criteria for selecting induced genes.
We set the following arbitrary criteria for determining which genes are induced by LPS. The gene had to be induced at least threefold over baseline (the expression level in unstimulated bladders) by LPS in one replicate of one time point and induced at least twofold over baseline in the second replicate of that time point.
The technique employed here is essentially a different method by which to cluster points in multidimensional space. They can be directly applied to gene expression by regarding the quantitative expression levels of n genes in k samples as defining n points in k-dimensional space (39). The nodes are mapped into k-dimensional space, initially at random, and then iteratively adjusted. Each iteration involves randomly selecting a data point P and moving the nodes in the direction of P. The closest node NP is moved the most, whereas other nodes are moved by smaller amounts depending on their distance from NP in the initial geometry. In this fashion, neighboring points in the initial geometry tend to be mapped to nearby points in k-dimensional space. The process continues for 20,00050,000 iterations. SOMs impose structure on the data, with neighboring nodes tending to define related clusters.
SOMs.
An SOM has a set of nodes with a simple topology (e.g., two-dimensional grid) and a distance function d(N1,N2) on the nodes (39). Nodes are iteratively mapped into k-dimensional "gene expression" space (in which the ith coordinate represents the expression level in the ith sample). The position of node N at iteration i is denoted fi(N). The initial mapping f0 is random. On subsequent iterations, a data point P is selected and the node NP that maps nearest to P is identified. The mapping of nodes is then adjusted by moving points toward P by the formula
![]() |
The method has been implemented in the GeneCluster software, which performs the analytical calculations and provides easy data visualization (http://www.genome.wi.mit.edu). This focuses attention on the "shape" of expression patterns rather than on absolute levels of expression. Using GeneCluster, SOMs were constructed by choosing a 6 x 4 grid that generated 24 clusters. It was also our concern to present the fewest clusters possible that would still give a clear picture of LPS-induced gene expression. Increasing the number of clusters by increasing the grid did not give us any additional correlation between genes.
Partial cloning of mouse ICAM-1, ß-NGF, and TGF-ß genes.
RT-PCR was used to amplify a portion of mouse intercellular adhesion molecule-1 (ICAM-1; GenBank accession no. X52264), ß-NGF (GenBank accession no. K01759), and TGF-ß (GenBank accession no. M13177) from mouse bladder, heart, and brain total RNA. Total RNA was extracted using Ultraspec RNA Reagent (Biotecx Laboratories, Houston, TX). Briefly, 100 mg of fresh tissue was homogenized in 1 ml of Ultraspec reagent, and 0.2 ml of chloroform was added and vortexed for 15 s. The phases were separated by centrifugation (12,000 g, 15 min, 4°C), and isopropyl alcohol was added to the aqueous phase to precipitate total RNA. The resulting RNA was reversed transcribed to single-stranded cDNAs using Superscript II reverse transcriptase enzyme (GIBCO-BRL, Rockville, MD) according to the suppliers protocol.
Three cDNA fragments were generated by RT-PCR using various mouse tissue total RNA (Biotecx Laboratories) and mouse isoform-specific oligonucleotides, as follows: mouse ICAM-1, 5' CGA TCT TCC AGC TAC CAT CC 3' (sense) and 5' CAT CAC GAG GCC CAC AAT GA 3' (antisense); mouse ß-NGF, 5' GCT GTG CCT CAA GCC AGT GA 3' (sense) and 5' GCA AGT CAG CCT CTT CTT GT 3' (antisense); and mouse TGF-ß 5' GCT GTC ATT GCT GGT CCA GT 3' (sense) and 5' GCA AAG CTG TCA GCC TTG CT 3' (antisense).
PCR reactions were carried out in 10 mM Tris hydrochloride (pH 8.3), 50 mM MgCl2, 0.2 mM dNTPs, and 1 mM primers at 95°C for 0.5 min, 58°C for 0.75 min, and 72°C for 1.5 min for 35 cycles in a thermal cycler (model 96; Stratagene, La Jolla, CA). After amplification, an 867-bp ICAM-1 cDNA, a 1,077-bp TGF-ß cDNA, and a 936-bp ß-NGF cDNA were confirmed by PCR using nested primers. After confirmation with nested primers, smaller cDNA fragments were amplified with a T7 sequence. The resulting probes for the RNase protection assay (RPA) were 185-bp ICAM-1, 286-bp TGF-ß, and 339-bp ß-NGF.
RNase protection assay.
Three bladders from each group were homogenized together in Ultraspec RNA solution for isolation and purification of total RNA as described above. RPA was performed as described by Breyer and collaborators (4). Radioactive riboprobes were synthesized in vitro from 1 µg of linearized plasmids containing three different cDNA fragments by using MAXIscript kit (Ambion, Austin, TX) for 1 h at 37°C in a total volume of 20 µl. The reaction buffer contained 10 mM dithiothreitol (DTT), 0.5 mM each of ATP, CTP, and GTP, 2.5 µM of UTP, and 5 µl of 800 Ci/mmol [-32P]UTP at 10 mCi/ml (Amersham Pharmacia, Piscataway, NJ). Hybridization buffer included 80% deionized formamide, 100 mM sodium citrate, pH 6.4, and 1 mM EDTA (RPA II; Ambion, Austin, TX). A quantity of 20 µg of total RNA was incubated at 56°C for 12 h in hybridization buffer with 5 x 104 cpm labeled riboprobes. After hybridization, RNase digestion with Ambion RPA III RNase A (0.250 U/µl) and T1 (10 U/µl) was carried out at 37°C for 30 min. RNA protected fragments were precipitated and separated on 6% polyacrylamide gel at 1,200 V for 20 min. Quantification of each detectable band was calculated based on DLUs generated by OptiQuant and subtracting the background; data are expressed as a ratio with ß-actin gene. Controls included the probe set hybridized to tRNA only.
Statistical analysis.
Results in each cluster are expressed as means ± SE. The statistical analysis of data was performed using SIGMAPLOT (version 5.0; SPSS). Within each cluster, peak expression was determined by ANOVA followed by paired Newman-Keuls multiple comparisons procedure (3). The statistical analysis of morphological data was performed using Wilcoxons rank sum test. Results are expressed as means ± SE. The n values reported refer to the number of animals used for each experiment. In all cases, a value of P < 0.05 was considered indicative of a significant difference (3).
Reagents.
Escherichia coli LPS strain 055:B5 was purchased from Sigma Chemical (St. Louis, MO) and suspended in pyrogen-free saline on the day of the experiment. Ultraspec RNA isolation system was purchased from Biotecx Laboratories. Atlas cDNA expression arrays were purchased from Clontech Laboratories (containing 588 genes as described at the web site http://www.clontech.com/atlas/genelists/search.html)
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
|
|
|
|
|
|
|
|
Early genes were upregulated 0.5 to 1 h after LPS stimulation.
Clusters 35 (Fig. 3) contained genes that had their peak expression at 0.5 h after LPS stimulation. The magnitude of gene expression was high in cluster 3 compared with cluster 4, but both presented the same biphasic pattern with a first peak at 0.5 h and a secondary peak at 24 h. Cluster 5 indicates genes with peak expression at 0.5 h that was maintained up to 24 h. Tables 25 present only the genes that belong to these clusters and had a ratio of expression between LPS- and saline-treated groups above 2.0. Cluster 3 included IL-6 receptor, intracellular kinases, transcription factors, and growth factors. Cluster 4 included growth factors such as - and ß-NGF, oncogenes such as VEGF R1, C-C chemokine receptor, P selectin, cell cycle proteins, stress response elements, ion channels, and transcription factors. However, clusters 3 and 4 are not really distinct from one another but are merely reflections of different promoter strengths. Finally, cluster 5 included oncogenes, intracellular kinases, and transcription factors.
Clusters 610 (Fig. 4) contained genes that had their peak expression 1 h after LPS stimulation. Clusters 6 and 8 had a defined peak at 1 h, followed by a significant reduction in gene expression with time. Cluster 6 differs from cluster 8 in the magnitude of gene expression. Clusters 7, 9, and 10 represent genes with a peak expression at 1 h that was maintained at 4 and declined at 24 h. The magnitude of gene expression was higher in cluster 7 and decreased toward clusters 10 and 9. Tables 2 and 3 indicate genes belonging to these clusters that were upregulated at least twofold by LPS compared with saline-treated tissues. Almost all interleukin genes were present in cluster 9 and therefore upregulated 1 h after stimulation with LPS.
Late genes were upregulated 4 to 24 h after LPS stimulation.
Clusters 1115 had their peak expression between 4 and 24 h after LPS stimulation (Figs. 5 and 6). However, their magnitude of expression in terms of percentage of ubiquitin was different. Therefore, the software clustered the genes of different magnitudes as separate clusters.
With regard to the 4 h group, the magnitude of gene expression was higher in cluster 11 and decreased in clusters 12 and 13 (Fig. 5). Tables 4 and 5 contain the list of genes that belong to those clusters. Cluster 11 included apoptosis, growth factors, heat shock proteins, and proteases and their inhibitors, whereas cluster 12 contained mainly transcription factors of the NF-B family, adhesion molecules (VCAM-1 and ICAM-1), DNA synthesis and repair factors, and proteases (ACE, MMCP-4, and Spi). Cluster 13 contains some of the interleukin receptor genes, cell surface receptors, and adhesion molecules, and apoptosis-related proteins (iNOS and IL-1 receptor).
Clusters 14 and 15 contained genes with peak expression at 24 h (Fig. 6). Cluster 14 included genes with high expression, whereas cluster 15 included genes with low expression. Table 5 presents the genes belonging to those clusters that were upregulated at least twofold by LPS. Cluster 15 included most of the interleukin and chemokine receptors, growth factors, and intracellular kinases.
To access the variation of our experimental conditions, the time course of three well-expressed genes in each cluster is represented in Supplemental Fig. 8 (refer to the Supplementary Material1 to this article, published online at the Physiological Genomics web site). With few exceptions, such as Egr-1 and CD14, the individual gene expression followed the prediction of cluster analysis. A perfect example of such small variability is cluster 6. The three highest expressed genes, c-Jun, c-Fos, and Jun-b, presented the same trend of a time-dependent upregulation. However, the variation associated with a given cluster can be better visualized in Figs. 26.
RNase protection assay.
To confirm some of the gene array results by another method, we developed an RPA. As there is no commercial RPA kit available for suitable mouse genes, we partially cloned ICAM-1, ß-NGF, and TGF-ß genes. Our results indicate that at least within the two time points studied (1 and 4 h), all three genes presented the same trend of upregulation observed by cDNA array analysis of LPS-induced inflammation. Expression of these genes following LPS-induced bladder inflammation can be visualized on Fig. 7A and quantification of the RPA is presented in Fig. 7B. RPA results are in good agreement with those obtained with cDNA array. Peak expression of ß-NGF occurred at 0.5 h (cluster 4, Fig. 3), as determined by cDNA array, which corresponded to 9.1-fold upregulation compared with tissues treated with saline (Table 2). RPA results were obtained at 1 and 4 h after LPS treatment and, therefore, after peak expression. Nevertheless, at 4 h after LPS, both RPA and cDNA array presented similar upregulation of ß-NGF (RPA = 2-fold; cDNA array = 2.9-fold). Moreover, both TGF-ß and ICAM-1 reached peak expression 4 h after LPS stimulation, corresponding to ninefold upregulation as determined either by RPA or cDNA array.
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Our cluster analysis identified gene clusters that presented a constant expression as well as clusters with a peak expression in a particular time interval following LPS stimulation. Although generated without preconceptions, the clusters correspond to patterns of clear biological relevance. For example, this is the case for interleukin genes that had their peak expression at 1 h and were present in the same cluster (cluster 9), whereas most of the interleukin receptor genes had their peak expression 24 h after LPS and were grouped in cluster 15. It remains to be determined whether the time-dependent expression of interleukin and interleukin receptor genes reflects a differential contribution of migrating inflammatory cells. Our results indicate that neutrophils appear as early as 4 h and reached a maximum at 24 h after LPS stimulation.
Most of the known genes upregulated have, in fact, been previously identified in the extensive literature on inflammation. This is the case of the genes responsible for proteins of the NF-B pathway that were grouped in cluster 12. These results support our finding that NK-
B proteins have a mandatory role in LPS-induced bladder inflammation (45). This study, however, identifies the vast majority of these genes in a single experiment and also uncovered additional ones not previously known to be regulated by LPS.
The neurotrophin ß-NGF (K01759) and -NGF (M11434) belong to a group of genes that were upregulated as early as 30 min after LPS stimulation (cluster 4). These results confirm previous observations indicating that NGF is one of the first genes to be upregulated in the bladder during inflammation (23, 27). Bladder neurotrophin mRNA is also increased following spinal cord injury (acute/chronic) or cyclophosphamide-induced cystitis (acute/chronic) (44). In addition, NGF levels in samples from patients with painful bladder conditions are higher than in samples from controls (23), and NGF levels are elevated the urine of patients with cystitis (28).
Some studies suggested that c-Fos may be involved in regulating NGF production because c-Fos translocation to the nucleus precedes NGF production (40). Our results indicate that both - and ß-NGF genes were upregulated by LPS before c-Fos, suggesting a feedback loop in which NGF genes modulate their own transcriptional control.
Some of the genes such as C-C chemokine receptors presented a biphasic pattern of expression. C-C-R2 (U56819) was upregulated as early as 0.5 h, downregulated at 1 and 4 h, and finally expressed at 24 h after LPS. In contrast, CC-R1 (U29678) presented a different pattern of upregulation (cluster 15) with a clear peak at 24 h. Chemokine receptors belong to a family of seven transmembrane spanning proteins, the vast majority of which are receptors that couple to, and signal via, heterotrimeric guanine nucleotide-binding proteins (G proteins) (32). A role of chemokines and their receptors are yet to be defined in cystitis. However, substances known to release IL-8 such as substance P seem to play a major role in this disorder (8, 20, 30, 43). Additional experiments involving gene expression in IL-8 knockout mice or specific antagonists of C-C receptors should elucidate the interdependence of these genes in cystitis.
Interestingly, a proto-oncogene group presented a marked expression only at 1 h after LPS. In the other time points studied, the expression of this group was insignificant. These tightly controlled groups include the immediate early genes c-Fos, Fos-B, Fra-2, Jun-B, and Jun-D. Egr-1, a zinc finger-encoding gene coregulated with c-Fos during growth and differentiation (36) was also expressed at 1 h after LPS stimulation. However, this gene was in cluster 8, which had the highest expression in the array. Transient expression of c-Fos may be an extremely important determinant of gene expression patterns which follow. Combined with c-Jun, c-Fos forms the transcription factor AP-1, which initially undergoes cytoplasmic phosphorylation prior to nuclear translocation, DNA promoter region binding, and modulation of expression of a family of genes. Changes in transcription factors such as c-Fos can therefore potentially participate in subsequent changes in gene cluster expression.
Our results indicate that most of the NF-B genes were clustered together and therefore expressed a concomitant peak of expression 4 h after LPS stimulation as reported by this laboratory (45). The clustering of the NF-
B pathway genes in a single cluster validates the intended examination of functionally significant gene clusters using the SOM approach.
Comparison of our results obtained with whole mouse bladder homogenates with those obtained with mouse macrophages stimulated with LPS (35) revealed a common pathway involved in the response of LPS during inflammation.
Analysis of 588 genes represents only 0.4% of the total mouse genome. It will be instructive to determine whether larger commercially available and custom-made arrays will further clarify the key pathways involved in bladder inflammation or simply add more genes to the clusters identified here. In this type of analysis there is initial evidence that more time points may optimize statistical power for cluster analysis at least as effectively as more genes or more replicates (1). Emphasis needs to be laid on number of time points and replicates in experimental design.
This work determined differences in bladder gene expression secondary to acute LPS stimulation. Our results indicate that significant morphological changes in response to LPS start to occur only after 4 h of stimulation. Therefore, we suggest that early gene expression from 30 min to 4 h most likely represents direct effects of LPS on constitutive cells of the urinary bladder. Gene regulation observed after 4 h of stimulation may indicate both direct and indirect effects of LPS, since after 4 h the contribution of migrating cells needs to be taken in consideration.
Many of the results reported here using gene array techniques confirm previous observations on gene expression using other techniques. For instance, our cDNA array results indicate that NGF upregulation occurs early during acute inflammation, in agreement with reports of an early upregulation of NGF in the bladder following inflammation using Northern blot and RT-PCR techniques (27).
To fairly interpret gene cluster analysis, we must be cognizant of a growing body of evidence that gene and protein changes can be dissociated (14, 16, 44). For instance, increased abundance of urinary bladder NGF mRNA is not always associated with increased total urinary bladder NGF (44). The discrepancy between two measures (mRNA and protein) may reflect retrograde axonal transport of NGF to the dorsal root ganglia (44). Future proteomic correlation must determine how directly mRNA changes reflect translated protein levels and the physiological consequence of these proteins (7).
In conclusion, the cDNA array experimental approach provided a global profile of gene expression changes in bladder tissue after stimulation with LPS. SOM techniques identified functionally significant gene clusters. These responses in gene expression may represent a balance between the cytoprotective and degenerative processes that accompany bladder response to injury. In particular, in future research gene cluster analysis techniques can be applied to begin to understand clinically relevant issues, such as how and why the transition from acute to chronic inflammation occurs only in selected circumstances, and which pathophysiological and therapeutic strategies change the normal balance of apoptosis and necrosis in populations of bladder cells.
![]() |
ACKNOWLEDGMENTS |
---|
This work was supported by National Institutes of Health (NIH) Grant DK-55828-01 (to R. Saban), John Sealy Memorial Endowment Fund Grant 456977 (to R. Saban), and National Institutes of Health Grant DK-51392 (to T. G. Hammond).
![]() |
FOOTNOTES |
---|
Address for reprint requests and other correspondence: R. Saban, Dept. of Physiology, College of Medicine, Oklahoma Univ. Health Science Center, 940 SL Young Blvd., Rm. 605, Oklahoma City, OK 73190 (E-mail: ricardo-saban{at}ouhsc.edu).
1 Supplementary Material (Fig. 8) to this article is available online at http://physiolgenomics.physiology.org/cgi/content/full/5/3/147/DC1.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|