Surfactant proteins-A and -D (SP-A and SP-D) are members of the collectin protein family. Mice singly deficient in SP-A and SP-D have distinct phenotypes. Both have altered inflammatory responses to microbial challenges. To further investigate the functions of SP-A and SP-D in vivo, we developed mice deficient in both proteins by sequentially targeting the closely linked genes in embryonic stem cells using graded resistance to G-418. There is a progressive increase in bronchoalveolar lavage phospholipid, protein, and macrophage content through 24 wk of age. The macrophages from doubly deficient mice express high levels of the matrix metalloproteinase MMP-12 and develop intense but patchy lung inflammation. Stereological analysis demonstrates significant air space enlargement and reduction in alveolar septal tissue per unit volume, consistent with emphysema. These changes qualitatively resemble the lung pathology seen in SP-D-deficient mice. These doubly deficient mice will be useful in dissecting the potential overlap in function between SP-A and SP-D in host defense.
- surfactant proteins
- alveolar macrophages
- sequential gene targeting
members of the collagenous subfamily of calcium-dependent lectins (collectins) are structurally related extracellular pattern recognition molecules that participate in the innate immune response to a wide variety of microbial pathogens and particulate allergens (4, 7, 38). In the mouse, the following four collectins have been identified: mannose-binding lectins I and II (Mbl I and Mbl II) and surfactant proteins-A and -D (SP-A and SP-D). The genes encoding Mbl I, SP-A, and SP-D are located within a 60-kb gene cluster on chromosome 14, synthenic with 10q 22–23 in the human (1). Mbl I is predominantly expressed in the liver, and the protein circulates in the serum (6). SP-A and SP-D are predominantly expressed by lung epithelial cells, primarily nonciliated bronchiolar cells and alveolar type II cells (37), and the proteins are found in the alveolar and airway lining fluid. SP-D is also expressed in several other tissues (22).
SP-A is a major structural component of pulmonary surfactant, specifically the extracellular tubular myelin form (36). Despite the close association of SP-A and surfactant lipids, the physiological role of SP-A in surfactant function and metabolism is uncertain. Mice lacking SP-A show no major decrement in surfactant activity or lung function (14, 17). Although it is possible compensatory adaptations to SP-A deficiency may obscure an important role in surfactant function, these mice have been useful in confirming a role for SP-A in pulmonary host defense against multiple microbes. SP-A-deficient mice clear pathogens from the lung slowly and have an exaggerated inflammatory response to microbial challenge (2, 13, 19, 20), consistent with numerous in vitro studies (38). Mice deficient in SP-D display progressive postnatal inflammation without evidence for infection (3, 18). SP-D null mice have increased numbers of activated alveolar macrophages. The reactive oxygen species and matrix metalloproteinases (MMPs) produced by these macrophages contribute to a patchy progressive inflammatory response and air space destruction (35).
In vitro, SP-A and SP-D have qualitatively similar activity in several different assays. Both proteins are able to form tubular myelin-like structures from surfactant phospholipids and enhance surface activity (27). SP-A and SP-D also bind to a common set of microbes and phagocytic cells (5). Although the differences in phenotype between the SP-A- and SP-D-deficient mice support distinct biological functions for SP-A and SP-D, some degree of either synergy or redundancy in function as suggested by in vitro studies has not been excluded. To investigate the functional relationships and overlap between SP-A and SP-D in vivo, we prepared mice deficient in both proteins. Because the two genes are closely linked on chromosome 14, we adopted a sequential targeting strategy in embryonic stem (ES) cells using graded resistance to G-418. We report here the success of this strategy and a preliminary phenotypic characterization of mice deficient in both SP-A and SP-D.
SP-A and SP-D targeting constructs.
The SP-A replacement-type targeting vector containing 1.1- and 8.9-kb homology regions and a Pgk-neo cassette (1.8 kb) for positive selection replaced exons 2, 3, and 4, including the translation start site for murine SP-A and short segments of flanking intronic sequence (F. Poulain and S. Hawgood, unpublished observation). Pgk-tk was inserted 3′ to the regions of homology for negative selection. The SP-D replacement-type targeting vector containing the 1.2- and 4.3-kb homology and a Pgk-neocassette for positive selection replaced all of exon 2, including the translation start site for murine SP-D and short segments flanking the intronic sequence. Pgk-tk was inserted 5′ to the regions of homology for negative selection (3). Both targeting vectors were linearized for transfection using a unique Not I site.
Cell culture, transfection, and selection.
The linearized SP-A construct was electroporated into F1 BL6/129Sv ES cells cultured on mitotically inactivated mouse embryonic fibroblasts as described previously (21). After G-418 (300 μg/ml)/1-(2-deoxy-2-fluoro-d-arabinofuranosyl-S-iodouracil) (0.3 μM) selection for 7 days, surviving colonies were picked and screened for homologous recombination at the SP-A allele by PCR using primers specific for the targeted allele. ES cells heterozygous for SP-A null mutation were electroporated with the linearized SP-D construct and cultured on mitotically inactivated mouse embryonic fibroblasts for 7 days in the presence of a high dose of G-418 (700 μg/ml)/FIAU (0.3 μM). Surviving colonies were picked and screened for homologous recombination at both the SP-A and SP-D allele by PCR using primers specific for the respective targeted alleles.
The presence of the targeted alleles in all clones heterozygous for both the SP-A and SP-D null mutation by PCR was confirmed by Southern blot hybridization. The targeted SP-A allele was confirmed by Southern blot hybridization of EcoR I-digested genomic DNA to a 450-bp Pst I-Hind III probe corresponding to the genomic sequence 3′ to the targeting construct. This probe yielded either the predicted 8.1-kb fragment characteristic of the wild-type allele or the 4.8-kb fragment from the targeted mutant allele. The targeted SP-D allele was confirmed by Southern blot hybridization ofBstX I-digested genomic DNA using a 600-bp BstX I-Xba I probe corresponding to the genomic sequence 5′ to the targeting construct. This probe yielded either the predicted 6.2-kb fragment characteristic of the wild-type allele or the 3.1-kb fragment from the targeted mutant allele.
Generation of chimeric and SP-A- and/or SP-D deficient mice.
Four of the five positive clones were injected separately in day 2.5 postcoital eight-cell to morula stage CD-1 zygotes and transferred to pseudopregnant B6D2 females. Chimeric offspring were bred with albino CD-1 female mice, and their pups were screened by PCR and Southern analysis for germ line transmission of the mutant alleles. Mice heterozygous for both mutations were intercrossed, and subsequent litters were analyzed at both loci by PCR with genomic Southern blot confirmation to distinguish mutations existing in cis ortrans. All experiments described here were conducted on littermate mice of the BL6/129Sv:CD-1 mixed genetic background. Mice were bred and housed under barrier conditions in the transgenic animal facilities of the University of California San Francisco using protocols approved by that facility.
Northern blot analysis.
Total RNA (5 and 2.5 μg) isolated from the lungs and livers of four 8-wk-old mice of each genotype by the guanidine isothiocyanate method was hybridized with [32P]cDNA probes for mouse SP-D, SP-A (lung only), and Mbl I (liver only). Signals were quantified in a PhosphoImager (Molecular Dynamics, Sunnyvale, CA) and normalized to 18S. pTRI RNA 18S was [32P]UTP labeled using the Ambion Maxiscript kit (Ambion, Austin, TX).
Serum from littermate wild-type and SP-A- and SP-D-deficient [SP-AD (−/−)] mice was analyzed by Western blot for Mbl I. The antibody to Mbl I (a gift from Dr. J. R. Wright, Duke University, Durham, NC) was raised in rabbits against the CRD fragment of rat Mbl I.
Bronchoalveolar lavage phospholipid and protein analysis.
Mice 3, 6, 16, and 24 wk old were anesthetized with intraperitoneal pentobarbital sodium. The trachea was canulated, and the lungs were lavaged with 5 × 1-ml aliquots of 10 mM Tris, 100 mM NaCl, and 0.2 mM EGTA, pH 7.4. The bronchoalveolar lavage (BAL) fluid was centrifuged at 250 average gravitational force (g av) for 5 min at 4°C to remove cells. An aliquot of the cell-free BAL supernatant was extracted into chloroform/methanol, and the total phospholipid content was derived from the phosphorus concentration. The total BAL protein concentration was determined using bicinchoninic acid as a substrate. SDS-PAGE and Western blots on cell-free BAL supernatant were performed using standard techniques and developed using the enhanced chemiluminescence detection reagents (Amersham, Arlington Heights, IL). The antibody to mouse SP-D (a gift from Dr. J. R. Wright, Duke University) was raised in rabbits against full-length recombinant mouse SP-D expressed in Chinese hamster ovary cells and purified from the media using maltose chromatography. The antibody against mouse SP-A was raised in rabbits against the CRD fragment of mouse SP-A expressed inEscherichia coli using the pET23 vector system (Novagen, Madison, WI) and was purified by histidine tag-nickel chromatography.
BAL cell analysis.
The pellet from the 250 g av × 5-min BAL centrifugation was gently resuspended in 200 μl of lavage buffer for cell count. Cytospin slides were stained with Dif-Quik (Dade International, Miami, FL) for cell differential. BAL cells were pooled from 4–6 mice/genotype [SP-AD sufficient (+/+), SP-AD (−/−), SP-A deficient (−/−), and SP-D (−/−)] for total RNA isolation. Expression of MMP-12 mRNA was assessed by Northern blot using the32P labeled MMP-12 cDNA probe (a gift from Dr. S. Shapiro, Washington University School of Medicine, St. Louis, MI), and the message level was quantified by ribonuclease protection assay (Pharmingen, San Diego, CA). MMP activity in BAL cells was detected by zymography using precast gelatin gels (Invitrogen, Carlsbad, CA).
The lungs from mice at 3, 6, 16, and 24 wk were fixed at 20 cmH2O by intratracheal instillation of 2% glutaraldehyde and 1% paraformaldehyde in 0.1 M phosphate buffer (2 h, room temperature) and then postfixed overnight in 1.5% osmium tetroxide in veronal acetate buffer at 4°C. They were en bloc stained in 1.5% uranyl acetate in maleate buffer and then quickly dehydrated in cold acetone and propylene oxide. The tissue was finally infiltrated and embedded in LX 112 (Ladd Research Industries, Burlington, VT). Semithin sections were stained with toluidine blue, and ultrathin sections were stained with 5% uranyl acetate and 0.8% lead citrate for electron microscopy.
Lungs from mice at 12 wk (n = 5/genotype) were fixed as above. The lungs were excised with the trachea clamped and stored in the fixative for 2 h at room temperature. The total volume of each lung was determined by fluid displacement (31). Systematic uniform random samples representative of the whole organ were taken from each lung as described in detail previously (8). After osmication, bloc staining in aqueous uranyl acetate, and dehydration, the samples were finally embedded in glycol methacrylate (Technovit, Heraeus Kulzer, Weinheim, Germany). The lungs were analyzed at the light microscope level according to established stereological methods (34), using the computer-assisted stereological toolbox CAST 2.0 (Olympus, Albertslund, Denmark). By means of point and intersection counting, the volume fractions and total volumes of pulmonary parenchyma, alveolar septal tissue, and air space as well as the surface fraction and total surface area of alveolar epithelium were estimated. As a measure for the thickness of the alveolar septum, the volume-to-surface ratio of alveolar septal tissue and alveolar epithelial surface was then determined. As a measure for air space size, the mean linear intercept length and volume-weighted mean volume (ῡv) of terminal air spaces were determined. Theῡv of terminal air spaces was estimated by the point-sampled intercepts method (10), which has been applied successfully to lung alveoli previously (9, 23). Linear intercept length and ῡv measurements were done from “wall to wall,” thereby including alveoli and alveolar ducts (9).
BAL data were analyzed by the two-tailed Student's t-test. Stereological data were analyzed with the nonparametric Mann-WhitneyU-test. Tests were performed using the STATISTICA 6.0 software (StatSoft, Hamburg, Germany). A value of P < 0.05 was considered to be significant.
Generation of collectin-deficient mice.
Because the SP-A and SP-D genes lie within 60 kb in the collectin locus on mouse chromosome 14, separated by the Mbl I gene (1), doubly deficient animals [SP-AD (−/−)] could not be created by mating mice singly deficient in either SP-A or SP-D. We therefore sequentially targeted the two genes in ES cells using the replacement vectors developed for single targeting. Because both vectors contained the neomycin-resistance gene, the single targeted ES cells heterozygous for the SP-A null mutation were selected with high-dose G-418 after transfection with the SP-D replacement vector. Colonies resistant to the high-dose G-418 (700 μg/ml; 463 colonies in all) were picked, and six homologous recombination events were detected by PCR and subsequently confirmed by Southern blot analysis. Chimeric mice were generated from four of these clones, all of which transmitted the mutations to their progeny. By following the segregation of the genes in subsequent generations, we determined two of the four clones were targeted in cis and two in trans. Heterozygous intercross of the cis-targeted mice generated the littermate wild type, doubly heterozygous, and doubly deficient SP-AD mice (Fig. 1). The first 1,075 pups had genotypes in the expected Mendelian proportions with 296 (27%) wild type, 534 (50%) doubly heterozygous, and 245 (23%) doubly deficient. SP-AD (−/−) mice are viable and fertile and have not developed signs of infection or other illness in a barrier facility.
Verification of null alleles.
Lungs from SP-AD (−/−) mice had no full-length SP-D mRNA (1.35 kb), but low levels (<10% wild type) of a truncated message were detected (Fig. 2). We have previously shown that the truncated message lacks exon 2, including the translation start site, and there is no evidence of translation from an alternate start site (3). No SP-A mRNA was detected in the lungs of SP-AD (−/−) mice. Western blot of BAL fluid showed reduced SP-A and SP-D protein levels in the double heterozygous mice and no detectable SP-A or SP-D in SP-AD (−/−) mice (Fig. 3). Thus each mutation is null.
Mbl I levels are decreased in SP-AD (−/−) mice.
Liver Mbl I mRNA and serum Mbl I protein levels were significantly reduced in SP-AD (−/−) mice compared with wild-type littermates. By Northern blot analysis, liver Mbl I mRNA levels were 14% of littermate wild-type levels (data not shown), consistent with a marked reduction in immunoreactive Mbl I in the serum (Fig.4).
Accumulation of lipids and proteins in BAL from SP-AD (−/−) mice.
There was a progressive accumulation of phospholipid and protein in the BAL fluid of SP-AD (−/−) mice (Fig. 5). At 3 wk, both phospholipid and protein trended higher in SP-AD (−/−) mice compared with littermate wild types, but the difference was not significant (n = 5/genotype). By 6 wk, there was significantly more phospholipid (P < 0.002) and total protein (P < 0.05) in the BAL of SP-AD (−/−) mice compared with littermate controls. The accumulation was progressive such that, by 24 wk, phospholipid was increased 10-fold and total protein 4-fold over control. No differences between double heterozygous mice and wild types were seen at any age.
Increased number and size of alveolar macrophages in SP-AD (−/−) mice.
The number of cells recovered by BAL was not significantly different between genotypes at 3 wk, although some enlarged foamy macrophages were seen on cytospins from the SP-AD (−/−) mice but not the controls. By 6 wk, there was a modest but significant increase in the number of BAL cells, with a concomitant increase in the number of large foamy macrophages in the SP-AD (−/−) mice (7.2 ± 0.7 × 105 cells/mouse compared with 5.1 ± 0.9 × 105 cells/mouse in the controls, P < 0.05). At 16 wk, the BAL cell count was increased dramatically in SP-AD (−/−) mice (2.7 ± 0.6 × 106 cells/mouse compared with 4.8 ± 0.8 × 105 cells/mouse in the controls). At all ages, >95% of the BAL cells were macrophages. Large amounts of dense secretions contaminated the BAL cell pellet at 24 wk, making an accurate cell count difficult at this age. The cytospins at 24 wk showed giant foamy macrophages and small particulate material, presumably dense aggregates of surfactant (Fig.6).
MMP expression and activity in BAL from SP-AD (−/−) mice.
A strong signal for MMP-12 mRNA was detected by Northern analysis in BAL cells recovered from 8-wk-old SP-D (−/−) and SP-AD (−/−) mice (Fig. 7). No MMP-12 mRNA was detectable in a comparable amount of total RNA from BAL cells of either wild-type SP-AD (+/+) control or SP-A (−/−) mice (C57/BL6 background). By ribonuclease protection assay, MMP-12 mRNA levels were increased 120-fold in macrophages from SP-AD (−/−) mice and 4-fold in macrophages from SP-D (−/−) mice compared with the levels in macrophages from genotype-matched wild-type controls (n= 3 for each genotype). We also detected strong gelatinase activity at molecular weights consistent with MMP-9 and MMP-2 by zymography in BAL fluid from SP-AD (−/−) relative to both control SP-AD (+/+) BAL or SP-D (−/−) BAL (data not shown). No gelatinase activity was detected in wild-type or SP-A(−/−) mice.
Progressive alteration in pulmonary architecture in SP-AD (−/−) mice.
At 3 wk of age, although large areas of lung appeared quite normal, there were patchy subpleural areas where there was an abnormal accumulation of large foamy macrophages. The alveolar septa in these areas were thickened, and lamellar bodies in type II cells were prominent compared with wild-type mice (Fig.8, A and B). At 6 wk of age, the number of enlarged macrophages and total area of the lung involved were both incrementally increased (data not shown), but, by 16 wk, the area involved, the number and size of the macrophages, the degree of air space enlargement, the type II cell hyperplasia, the septal thickening, and the accumulation of air space secretions had all significantly increased (Fig. 8 C). These changes had further progressed at 24 wk (Fig. 8 D). Two distinct and abnormal cellular accumulations were also seen by 16 wk. These consisted of either a polyp-like grouping of cells arising from the alveolar septa and extending into the alveolar space (Figs. 8 C and9 A), resembling spindle cells of an intra-alveolar organizing pneumonia, surrounded by clumps of giant foamy macrophages, or accumulations of small rounded cells with the typical appearance and location of bronchial-associated lymphoid tissue (Fig. 9 B).
Ultrastructure of surfactant and cells in the alveolar space of SP-AD (−/−) mice.
The intratracheally fixed lungs of 16-wk-old mice were examined in the electron microscope. Alveolar surfactant forms were rarely seen in situ in wild-type lungs, but abundant secretions were present in many but not all alveoli in SP-AD (−/−) null mice. Most of the secretions were in the form of multivesicular spheroids of highly varied sizes. Variably sized vesicles and areas of electron-dense amorphous material, presumably protein, were also present. Notably, there was a complete absence of tubular myelin in the SP-AD (−/−) lungs (Fig.10). As reported in SP-D (−/−) mice, the alveolar interstitium was thickened, and the type II cells were hyperplastic with enlarged lamellar bodies. Also consistent with previous descriptions of SP-D (−/−) lungs, the alveolar macrophages in the lungs of SP-AD (−/−) mice were grossly enlarged and stuffed with membrane-bound phospholipid inclusions, cytoplasmic oil droplets, and electron-lucent crystal-like structures (data not shown).
The stereological data are summarized in Table1. No significant differences in the total volume of air space or septal tissue were noted between the two groups. However, both mean linear intercept length andῡv were significantly increased in the SP-AD (−/−) mice, thereby indicating enlarged air spaces. Becauseῡv is a parameter that contains information on both size and variation in size (10), not only air space enlargement but also a large variation in air space size was present in the SP-AD (−/−) mice. Although the total alveolar surface area in the SP-AD (−/−) mice was not significantly reduced because of a higher parenchymal volume, the alveolar epithelial surface fraction was decreased significantly, thereby indicating a loss of gas exchange surface per unit volume. Furthermore, the volume-to-surface ratio of the alveolar septum was increased significantly, indicating an increase in alveolar septal wall thickness in the SP-AD (−/−) mice.
Mice lacking each of the pulmonary collectins (SP-A and SP-D) have become available over the last several years (3, 17, 18). The phenotypes of these mice and a large number of in vitro studies support the general concept that both proteins participate in the innate immune response to microbial pathogens and organic allergens, modulating the pulmonary inflammatory response (reviewed in Ref.5). The extent to which the functions of these two structurally related and genetically linked proteins overlap or complement each other is poorly understood. There are now several examples in which the generation of mice deficient in two or more members of a protein family by sequential gene targeting has revealed functional redundancy or unsuspected cooperation between the family members (16, 28, 29). The close genetic linkage between SP-A and SP-D on mouse chromosome 14 and the low frequency of recombination in this region (26) precluded simple cross breeding as a method to achieve doubly deficient mice. We therefore sequentially targeted the SP-A and SP-D genes in ES cells using a modification to the reported protocols of incorporating different selection strategies for each round of targeting (16, 28, 29,32). Our two replacement constructs both contained a single copy of the neo r gene, and a high-dose G-418 selection was used for the second round of targeting. The targeting frequency of the second round (1.3%) was comparable to the frequency of homologous recombination we achieved with single targeting of this locus, and there was no interference with the germ line potency of this cell line.
In the mouse the Mbl I gene, encoding the serum collectin Mbl I, lies between the SP-A and SP-D genes disrupted in this study. Mbl I expression is reduced significantly in SP-AD (−/−) mice. In preliminary studies designed to understand this contiguous gene effect, we have determined that the targeted replacement of exons 2, 3, and 4 and some flanking intronic sequence of the SP-A gene, ∼20 kb upstream from exon 1 of the Mbl I gene, is solely responsible for the decreased Mbl I expression (unpublished observation). The sequential targeting of the SP-D gene does not further influence Mbl I expression. Although Mbl I is not present in the lung unless there is significant lung injury and plasma leakage, the reduced Mbl I levels in both the SP-A (−/−) and SP-AD (−/−) mice will need to be factored into the analysis of their host responses to microbial challenges.
The phenotype of SP-AD (−/−) mice (an excessive accumulation of surfactant lipid in the alveolar space, increased numbers of foamy alveolar macrophages with upregulated MMP expression, and emphysema) is qualitatively similar to that of SP-D (−/−) mice (3, 18) and quite distinct from the phenotype of SP-A-deficient animals (Ref.17 and our own unpublished observations). The timing and progression of the lipid and protein accumulation in the alveolar space appear to be different in the SP-AD doubly deficient and SP-D singly deficient mice. In SP-D-deficient mice generated by two separate laboratories in two different outbred strains, BAL phospholipid levels normalized to body weight are elevated maximally by the earliest measurable time point of 3–4 wk, and then levels remain fairly constant or even decline slightly as the mice age (11,15). In contrast, BAL phospholipid levels in doubly deficient SP-AD null mice were not elevated significantly at 3 wk but then rose progressively through 6 mo of age. A similar difference in the pattern of accumulation of total BAL protein was seen between SP-D (−/−) and SP-AD (−/−) mice. At present, we cannot make direct comparisons between perfectly genetically matched single and double knockout animals, but the consistency in the timing of phospholipid accumulation in singly deficient SP-D null mice across strains and in different laboratories suggests the observed differences will not be explained simply by strain-dependent genetic modifiers. The absolute increase in macrophage numbers and the level of MMP upregulation also appear to be greater in the SP-AD (−/−) mice compared with singly deficient SP-D null mice. This finding is consistent with a potential immunomodulatory role for SP-A on macrophage function (24, 25, 30), but further studies will be required with other markers of macrophage activation and in matched genetic backgrounds before firm conclusions can be drawn.
The stereological analysis revealed significant increases in both mean linear intercept length and ῡv of air spaces, clearly demonstrating air space enlargement in the SP-AD (−/−) mice. In addition, there was a decrease in the alveolar epithelial surface fraction, which indicates a loss of gas exchange surface per unit volume (12), in the SP-AD (−/−) mice. Because the lungs of 3-wk-old SP-AD (−/−) mice look qualitatively normal (data not shown), the changes in lung structure in the older mice are more likely the result of progressive alveolar wall destruction than a primary failure of alveolar septation. An increase in alveolar septal wall thickness is also present in SP-AD (−/−) mice. This is consistent with a recent study in human lung emphysema describing a thickening of the interstitium in the remaining alveolar walls because of increases in both elastin and collagen (33). Taken together, the present stereological data suggest emphysema development in SP-AD (−/−) mice. Again, a quantitative comparison with SP-D (−/−) mice waits breeding onto identical backgrounds. Our current studies are directed toward an understanding of whether these morphological changes and the altered surfactant homeostasis result from a dysregulation of normal immune homeostasis secondary to the loss of a regulatory activity of SP-A and SP-D or whether there is an unchecked environmental stimulus driving inflammation.
We thank Dr. J. R. Wright, Duke University, for gift of antibodies against SP-D and Mbl I and S. Freese, A. Gerken, and H. Hühn (Göttingen) for technical assistance.
This work was supported by National Institutes of Health Grants HL-24075, HL-58047, and DK-47766 and Grant 1400490 from the Medical faculty of the University of Göttingen.
Address for reprint requests and other correspondence: S. Hawgood, Box 1944, Univ. of California San Francisco, San Francisco, CA 94118-1944 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2002 the American Physiological Society