| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Published online before print October 4, 2007
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the Unit of Pharmacology and Therapeutics (UCL-FATH 5349),* and the Biomedical Magnetic Resonance Unit, Université catholique de Louvain, Brussels, Belgium
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
-smooth muscle actin-stained mural cells. The increase in blood plasma volume in Cav–/– tumors was confirmed by dynamic contrast enhanced-magnetic resonance imaging and found to be associated with a more rapid tumor growth. Finally, an in vitro wound test and the aorta ring assay revealed that silencing caveolin expression could directly impair the migration and the outgrowth of smooth muscle cells/pericytes, particularly in response to platelet-derived growth factor. In conclusion, a decrease in caveolin abundance, by promoting angiogenesis and preventing its termination by mural cell recruitment, appears as an important control point for the formation of new tumor blood vessels. Caveolin-1 therefore has the potential to be a marker of tumor vasculature maturity that may help adjusting anticancer therapies.
The reversibility of the biological pathways supporting the tumor vessel leakiness led us to examine the role of caveolin in this process. Caveolin is the structural protein of vesicle-like plasmalemmal microdomains termed caveolae that are involved in transcytosis, ie, the transcellular movement of macromolecules from the luminal side of capillary endothelial cells to the interstitial space.10 Through the regulation of intercellular interactions, caveolin is also a modulator of endothelial barrier integrity.11 Whether this capacity to regulate physiological vascular permeability is also involved in tumor vessel leakiness is primarily unexplored.
In the cardiovascular context, we and others have previously documented that the recombinant expression of caveolin12 or the administration of caveolin-derived peptides13 exert anti-angiogenic effects through the inhibition of the endothelial nitric-oxide synthase (eNOS). Caveolin has also been shown to co-purify with the VEGF-R2 receptor in caveolae and thereby to modulate VEGF signaling.14,15 Interestingly, endogenous (native) caveolin has been reported to be down-regulated in the tumor vasculature16 and in endothelial cells exposed to angiogenic growth factors including VEGF and basic fibroblast growth factor17,18 or when co-cultured with tumor cells.16 Angiogenic mediators such as nitric oxide19 or proangiogenic stresses such as ionizing radiation20 and hypoxia16 have also been associated with a decrease in endothelial caveolin abundance. Conversely, angiogenesis inhibitors including angiostatin, fumagillin, and thalidomide were reported to block VEGF-induced down-regulation of caveolin-1.17
In this study, we used caveolin-deficient mice bearing B16 melanoma to evaluate the hypothesis of an active role of the caveolin down-regulation in the modulation of tumor vessel permeability and angiogenesis. By means of a variety of in vitro and in vivo techniques including immunohistochemistry and dynamic contrast enhanced-magnetic resonance imaging (DCE-MRI) to evaluate the structural/functional characteristics of the tumor vasculature, we found that angiogenesis was orchestrated in tumors grown in Cav–/– mice via the extravasation of plasma proteins leading to the constitution of a provisional matrix required for endothelial tube formation and the deficit in smooth muscle cell (SMC)/pericyte recruitment, thereby preventing the tumor vessel maturation and the termination of the neovascularization process.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Cav-1-deficient (Cav–/–) mice (originally obtained from Drs. M. Drab and T.V. Kurzchalia, Max Planck Institute for Molecular Cell Biology and Genetics, Dresden, Germany) and their control littermates (Cav+/+) were generated through heterozygous mating and housed in our local facility. A minimum of six backcrossings was performed before using the animals described in this study. We used 8- to 10-week-old animals that were subcutaneously injected into the left flank with 106 B16 melanoma cells; tumor diameters were determined with an electronic caliper. In some experiments, luciferase-encoding B16 melanoma cells (Xenogen, Hopkinton, MA) were used to track early appearance of tumor mass; a single dose of luciferin (150 mg/kg) was injected intraperitoneally into anesthetized mice and in vivo bioluminescence was detected 15 minutes later on an IVIS50 imaging system (Xenogen). These procedures were approved by the local authorities according to national animal care regulations.
Melanoma (B16-F10) cells and 10T1/2 cells (SMC/pericyte progenitors) were routinely cultured in serum containing Dulbecco’s modified Eagle’s medium and Eagle’s basal medium, respectively. 10T1/2 cells were always maintained at a confluence level <80 to 90% and used between passages 5 and 15. Both cell types were from the American Type Culture Collection (Manassas, VA). In some experiments, SMCs/myofibroblasts were also derived from aorta strips, as previously reported21
; more than 90% of the cells isolated through this procedure were found to be positive for -SMA staining, and less than 1% were CD31-positive.
Immunostaining
Collected tumors were cryosliced and probed with antibodies against fibrinogen/fibrinogen degradation products or -SMA (both from DAKO, Heverlee, Belgium) and against CD31 (BD Pharmingen, Lexington, KY). 10T1/2 cells were stained with caveolin antibodies (BD Pharmingen) and fluorescein isothiocyanate (FITC)-labeled phalloidin (Sigma). 4,6-Diamidino-2-phenylindole (DAPI)-containing mounting medium (Vector Laboratories Inc., Burlingame, CA) was used to stain the nuclei. Secondary antibodies were coupled to tetramethyl-rhodamine isothiocyanate (TRITC) or FITC fluorophores (Jackson, De Pinte, Belgium). Tumor slices or cells were examined with an Axioskop microscope (Zeiss, Wetzlar, Germany) equipped for fluorescence.
In Vivo Albumin Clearance
125I-Albumin was prepared by classical radioiodination. Briefly, serum albumin (20%; Croix Rouge, Brussels, Belgium) was incubated in the presence of carrier-free 125I (2 mCi; MDS Nordion, Fleurus, Belgium), using chloramine as oxidizing agent and Na2S2O5 to stop the reaction. The labeled protein was purified by permeation gel chromatography (Sephadex G-25, NAP 25 column; Amersham Biosciences, Uppsala, Sweden) and then sterilized by filtration on a ministart filter (0.2 µm; Sartorius, Goettingen, Germany). The radiochemical purity monitored by paper chromatography was higher than 99%. For animal experiments, 106 dpm of the 125I-albumin solution was injected in the tail vein. Sixty minutes after injection, mice were sacrificed, and tumors were microdissected and weighted. The amount of 125I-albumin in each tissue sample was then determined using a 1480 Wizard counter (Wallac, Turku, Finland).
IFP Measurements
IFP was measured using a wick-in-needle apparatus, as previously described.22 Measurements were done by inserting the needle into the center of the tumor and injecting 50 µl of 0.9% sodium chloride to ensure fluid communication between the tumor and the pressure monitor system. Large-size tumors are needed for these measurements, and in this study, only 9-mm-diameter tumors were used to evaluate IFP.
DCE-MRI
This technique was used to compare tumor perfusion and vascular permeability in tumor-bearing mice, as previously described.23-26 In brief, MRI was obtained with a 4.7-T (200 MHz, 1H) 40-cm inner diameter bore system (Bruker Biospec, Ettlingen, Germany) using the 6.5-kDa blood pool contrast agent P792 (Vistarem, Guerbet, France). High-resolution multislice T2-weighted spin echo anatomical imaging was performed just before dynamic contrast-enhanced imaging. The tracer concentration changes were fitted to a two-compartment pharmacokinetics model as previously described.23,27,28 An operator-defined region of interest encompassing the tumor was analyzed on a voxel-by-voxel basis to obtain parametric maps. The Ktrans and Kep values (min–1) reflect the influx volume transfer constant from the plasma to the interstitial space and the fractional rate of efflux from the interstitial space back to the blood, respectively. Vp is the blood plasma volume per unit volume of tissue. The fraction of perfused tumor was defined as the number of voxels with statistical significance for Ktrans divided by the total number of voxels in the whole tumor region of interest.
siRNA Transfection and Immunoblotting
10T1/2 cells were transfected with duplex short interferring (si)RNA targeting caveolin-1 (corresponding to coding sequences 206 to 226 and 286 to 306) using Lipofectin (Invitrogen, Carlsbad, CA), according to the manufacturer’s protocol. A scramble siRNA was used as control and specific reduction of caveolin-1 expression level was confirmed by immunoblotting of 10T1/2 cell lysates with caveolin antibodies (BD Pharmingen). Anti-Akt and anti-phospho-Akt (Ser473) antibodies (Cell Signaling, Beverly, MA) were also used to immunoblot SMC/myofibroblast lysates.
To evaluate possible changes in caveolin expression within the vasculature of growing tumors, we also used lysates of microdissected tumor blood vessels, as previously described.29,30 In brief, B16 melanoma cells were injected at the vicinity of the saphenous arteriole in one of the rear legs. The arteriole and the downstream branches (diameters between 100 and 200 µm) were progressively trapped in the growing tumor mass. After 2 weeks, tumor-co-opted vessels and size-matched healthy arterioles from the contralateral leg were microdissected, lysed, and immunoblotted with caveolin-1 antibodies. The microvessels from five mice were pooled to get enough material for immunodetection.
In Vitro Migration and ex Vivo Angiogenesis Assay
To evaluate 10T1/2 cell migration, a 0.5-mm-wide line was scraped across confluent serum-starved cells. For the quantitative analysis, a migration index was defined as the ratio (expressed as percentage) of the density of migrating cells (observed 24 hours after the wound) in the center of the wounded area versus the density of cells in size-matched area of the unwounded region.20 To evaluate ex vivo angiogenesis, aortic rings (obtained from Cav–/– and Cav+/+ mice) were embedded in rat tail collagen gel and cultured for 12 days in MCDB131 medium supplemented with 5% autologous serum, as previously described.31 Cell migration and endothelial tube formation were observed through an inverted phase contrast microscope and quantified by two blinded investigators using images randomly captured by a video camera system.
Statistical Analyses
Statistical analyses were made using Student’s t-tests except for the fibrinogen accumulation scoring (Wilcoxon two-sample test) and for the tumor growth (two-way analysis of variance test).
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
We first used radioiodinated albumin (125I-albumin) to have an instant picture of the acute permeability status of B16 melanoma established in caveolin-1-deficient (Cav–/–) mice or in the wild-type littermates (Cav+/+). One hour after 125I-albumin intravenous administration to B16 melanoma-bearing mice, blood, skeletal muscle, and tumor samples were collected. As shown in Figure 1
, blood clearance of 125I-albumin appeared significantly greater (P < 0.01) in Cav–/– mice. This indication of a global increase in vascular permeability in Cav–/– mice was further confirmed by the higher accumulation of radioactivity in peripheral tissues such as skeletal muscle (P < 0.01) (versus Cav+/+ mice). By contrast, accumulation in size-matched (6-mm diameter) tumors reached similar levels in both mouse genotypes (Figure 1)
.
|
Tumor vessel permeability and perfusion were also monitored in both mouse genotypes via dynamic contrast-enhanced MRI using intravenous infusion of the macromolecular blood-pool agent P792 (Figure 2A)
. Both influx and efflux parameters, Ktrans and Kep, did not significantly differ between tumor-bearing Cav–/– and Cav+/+ mice, indicating that the overall tumor vessel permeability was similar in the two mouse genotypes (Figure 2B)
. By contrast, the plasmatic volume fraction (Vp) was significantly higher in Cav–/– tumors than in Cav+/+ tumors (Figure 2, A and B)
.
|
To understand the apparent resistance to increased permeability in Cav–/– tumors (versus Cav–/– healthy tissues), we next searched for a possible difference in tumor IFP. Using the wick-in-needle technique, we found a net increase in IFP in Cav–/– tumors (versus size-matched Cav+/+ tumors) (Figure 3A)
; this invasive measurement was only possible in large tumors (9-mm diameter). To examine whether an excess extracellular matrix could account for this difference, we evaluated the presence of fibrinogen and its degradation products in sections of tumors collected at different sizes (3-, 6-, and 9-mm diameters). The scoring of the amounts of immunolabeled fibrinogen/fibrinogen degradation products in tumors (see examples on the top of Figure 3B
) revealed that fibrinogen accumulation was larger in Cav–/– than in Cav+/+ mice (Figure 3B
, bottom). It is important to note that these differences were already detectable and eventually more prominent in small tumors (3-mm diameter) (Figure 3B
, bottom left), indicating that in the early stages of tumor growth in Cav–/– mice, vessel leakiness was more elevated than in Cav+/+ tumors. Higher scores of fibrinogen accumulation were also found in 6-mm Cav–/– tumors (see Figure 3B
, bottom right), thereby supporting the elevated IFP observed in large tumors (Figure 3A)
. Of note, the increasing extent of necrosis rendered difficult the interpretation of the fibrinogen staining in the largest 9-mm-diameter tumors (not shown).
|
We then aimed to identify alterations in the Cav–/– tumor vasculature that could account for the increased tumor Vp and IFP values. The combined use of antibodies directed against -smooth muscle actin (-SMA) and CD31 unraveled a dramatic difference both in the general pattern of the tumor vasculature and in the density of mural cells between Cav–/– and Cav+/+ mice (Figure 4A)
. First, the distribution and the size of tumor CD31-positive structures were homogenous in Cav+/+ mice but very dependent on the tumor topography in Cav–/– mice. In the latter, large and tortuous vascular structures were observed at the tumor periphery whereas a dense pattern of very small vessels was found in the core of the tumors (see red staining in the single- and dual-color panels in Figure 4A
). Regardless of the tumor regions, the CD31-positive staining area in Cav–/– tumors was 2.6-fold larger than in Cav+/+ tumors (P < 0.01, n = 5, 10 sections per mouse). Second, -SMA-positive staining was associated with a large proportion of CD31-labeled tumor vascular structures in Cav+/+ mice but was generally lacking around tumor endothelial structures in Cav–/– mice, independently of the central or peripheral location within the tumor (see green staining in Figure 4A
). A scoring procedure led two blinded investigators to estimate that at least 50% of the tumor CD31-positive structures were in close contact with -SMA-positive mural cells in any tumor-bearing Cav+/+ mice (five of five mice; 10 sections analyzed per mouse) whereas only one of five Cav–/– mice presented >50% mural cell coverage.
|
shows that a positive difference in the tumor diameter progressively installed between the two mouse genotypes. Accordingly, 18 days after tumor cell injection, tumor growth tended to plateau in Cav+/+ mice whereas it continued to progress in Cav–/– mice. In another series of experiments, we also used luciferase-encoding B16 melanoma cells to detect early tumor mass formation by in vivo bioluminescence. Figure 4C
shows that the tumors grown in Cav–/– mice were more rapidly detectable on luciferin injection (than those grown in Cav+/+ mice), confirming the direct positive impact of caveolin deficiency on tumor growth. Silencing Caveolin-1 Prevents the Migration of SMCs/Pericytes and Naturally Occurs in Vivo in the Tumor Vasculature
To explore further the role of caveolin in the recruitment of mural cells, we used cultures of SMC/pericyte precursors (10T1/2 cells) and evaluated their migration capacity in an in vitro wound assay. We first examined the changes in caveolin subcellular localization during 10T1/2 cell migration, a previously described hallmark of the caveolin implication in endothelial cell migration. Figure 5A
shows that from a diffuse location in quiescent, confluent cells, caveolin-1 presented a polarized distribution during migration. Caveolin-1 could be found in the cell nuclei and at the rear of migrating 10T1/2 cells, just at the opposite of the migratory front. We then used small interfering RNA to silence caveolin-1 expression in 10T1/2 cells and examined the migratory capacity of these cells in the wound assay. As shown in Figure 5B
, a 75% inhibition of caveolin expression led to a significant reduction in the migration of 10T1/2 cells; this effect was confirmed using a second siRNA and persisted for at least 72 hours (not shown).
|
shows that a significant reduction in caveolin expression could be authenticated in the microvessels influenced by the tumor microenvironment versus size-matched control microvessels isolated from the contralateral leg. The Lack of Caveolin Inhibits the Basal and Platelet-Derived Growth Factor (PDGF)-Driven Mobilization of SMCs/Myofibroblasts
To explore further a possible lack of mural cell migration/recruitment during the angiogenic process in Cav–/– mice, we also used an ex vivo assay based on the three-dimensional culture of aorta rings in collagen gels. This assay allows simultaneous evaluation of the organization of proliferating EC in tubes (see rectilinear structures in Figure 6, A and C
) and the outgrowth of myofibroblasts/SMCs (see scattered cells in Figure 6, A and C
). Figure 6A
shows that a significantly higher number of endothelial tubes formed from Cav–/– aorta rings, reflecting a higher intrinsic angiogenic potential (Figure 6B
, left). Conversely, the amounts of scattered myofibroblasts/SMCs were approximately twofold less around the Cav–/– aorta rings than the Cav+/+ aorta (Figure 6B
, right).
|
show that although PDGF stimulated a dramatic outgrowth of SMCs/myofibroblasts from the Cav+/+ aorta ring, a limited amount of these scattered cells could be identified around the Cav–/– aorta rings (see also bar graph in Figure 6C
). Note also that the endothelial tubes nicely formed from the PDGF-exposed Cav–/– aorta rings but not from the Cav+/+ aorta explants (Figure 6C
, top). Finally, to validate the alterations in PDGF signaling, we examined the phosphorylation status of Akt, a well-known kinase located downstream PDGF stimulation. Figure 6D
shows that the extent of Akt activation (ie, phosphorylation) was dramatically reduced in SMCs outgrown from Cav–/– aorta rings. |
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
. The absence of caveolae in lung capillaries of caveolin-deficient mice was previously reported to be associated with an increase in vascular permeability.32
Our study now documents that the phenomenon is exacerbated and accelerated in tumors (see detection of excess fibrinogen as early as in 3-mm tumors) but is then progressively buffered by the consecutive increase in IFP (Figure 3A)
. Indeed, at the time an elevated vascular permeability is still detectable in healthy tissues (Figure 1A)
, extravasation of radioiodinated albumin in tumors did not differ between Cav–/– and Cav+/+ mice (Figure 1A)
, neither were the permeability constants Ktrans and Kep different between the two mouse genotypes, as determined by DCE-MRI (Figure 2)
. Moreover, we found that in Cav+/+ mice, the tumor growth was associated with a decrease in the abundance of caveolin in tumor-perfusing microvessels (Figure 5C)
, confirming previous observations of a down-regulation of caveolin in angiogenic endothelial cells.16-20
Altogether, these data support a pathogenic link between the caveolin down-regulation that occurs naturally in the tumor microvasculature and the dynamic increase in vessel permeability to plasma proteins known to lead to the production of a scaffold provisional matrix for cell migration.
A second key step in the development of tumor vascularization that seemed to be regulated by the caveolin down-regulation is the dynamic process of tumor vessel maturation. During embryogenesis and in the rare circumstances of the adult life in which angiogenesis is switched on, neoformed endothelial tubes are rapidly covered by mural cells to give rise to mature blood vessels gifted with vasomotion.5
In tumors, the imbalance between the levels of pro- and anti-angiogenic molecules tends to promote a continuous remodeling of the tumor vasculature.4,5,33,34
In the current study, the lack of caveolin was shown to prevent the recruitment of -SMA-positive mural cells in growing tumors, a phenomenon that we found associated with a higher density of endothelial structures in the center of Cav–/– tumors (Figure 4A)
. The lack of vessel maturation in Cav–/– tumors also translated at the tumor periphery in the development of very large, tortuous vessels, supporting the concept of a continuous endothelial growth in absence of a minimal recruitment of mural cells. DCE-MRI confirmed that the blood plasma volume (Vp) was significantly increased in Cav–/– tumors in good correlation with the higher density and/or size of neovessels in these mice (Figure 4)
. Of note, a small but significant increase in B16 melanoma growth rate was observed in Cav–/– mice (versus Cav+/+ mice) (Figure 4, B and C)
, likely to reflect the higher angiogenic status of Cav–/– tumors.
To establish further the link between caveolin down-regulation and tumor vessel immaturity, we also used in vitro migration and vascular morphogenesis assays. We showed that SMC/pericyte precursor 10T1/2 cells treated with a silencing caveolin siRNA presented a defect in their capacity to migrate and that more endothelial tubes and lesser outgrowth of myofibroblasts arose from Cav–/– aorta rings. These assays also led us to identify clues in the defective Cav–/– vessel morphogenesis. We found that in nontransduced SMC/pericyte precursors, caveolin was polarized at the rear of the cell during migration, strongly suggesting a dynamic role in cell migration, such as the prevention of lamellipod protrusion as previously reported in endothelial cells.35
Among the possible effectors that master vessel maturation,36
we also showed that the PDGF signaling pathway was altered in Cav–/– SMCs. PDGF is the prototypical growth factor released by proliferating endothelial cells to recruit mural cells to stop ongoing angiogenesis.37
In our study, we found that the capacity of PDGF-B to stimulate SMCs and myofibroblasts outgrowth in the aorta ring assay (Figure 6C)
as well as the stimulation of the downstream kinase Akt (Figure 6D)
were lost in the absence of caveolin. Altogether and although alterations in other pathways than PDGF are likely to be involved in the observed Cav–/– vascular phenotype, our data indicate that caveolin is a master regulator of the maturation of tumor blood vessels.
Hassan and colleagues38 recently reported that caveolin-deficient SMCs presented cell autonomous abnormalities in proliferation and migration. In addition, in another study by the same authors, caveolin down-regulation was shown to favor activation of signal transduction pathways usually associated with SMC proliferation such as extracellular signal-regulated kinase phosphorylation.39 Although obvious differences in the biological context of the down-regulation of caveolin in these studies may account for the discrepancies with our study, another explanation may also be found in the double role of caveolin, ie, blocking the basal activity of enzymes residing in caveolae but facilitating their activation on agonist stimulation.40,41 In other words, the deficiency in caveolin/caveolae may lead either to a tonic activation of these signaling enzymes in the absence of stimuli (or in response to stimulatory pathways not concentrated in caveolae), or inversely to an incapacity to transmit the signal because of the deficit in the caveolar compartmentation of the different actors. This so-called caveolae paradox may explain why Cav–/– SMC proliferation was found to be associated with an increased extracellular signal-regulated kinase activation in basal conditions39 or in response to the noncaveolar ETB receptor stimulation,38 whereas in other studies42 (including ours) stimulation of the caveolar PDGFR-ß receptor led to a lesser activation of extracellular signal-regulated kinase or Akt in SMCs exposed to caveolin siRNA.
Finally, Woodman and colleagues43
have reported a defect in tumor angiogenesis in caveolin-deficient mice using the same syngeneic B16 melanoma model. In this study, however, they evaluated angiogenesis by vessel counting from hematoxylin and eosin-stained paraffin-embedded sections. The mural cell-free endothelial structures and immature vascular morphology may have led these authors to underestimate the extent of tumor angiogenesis in caveolin-deficient mice. In our hands, the stimulated angiogenic process was only detectable through the use of specific immunomarkers, including von Willebrand factor, MECA32, and CD31 antibodies, to probe endothelial structures within the tumor mass (as reported in Figure 4A
).
In conclusion, our study emphasizes the close relationship between mural cell recruitment and termination of angiogenesis. One generally admitted view is that tumor vessel maturation is incomplete because the exponential growth of endothelial structures beats the capacity of pericytes/SMCs to be recruited in time and in numbers to sufficiently cover neoformed endothelial tubes.33 Here, our study supports another (although nonexclusive) reading of this phenomenon in which the defect in vessel maturation would, in fact, be actively supported by the dynamic down-regulation of caveolin. Furthermore, our data underscore how caveolin deficiency contributes through an early increase in tumor vessel permeability to build up an extravascular matrix network that promotes angiogenesis. Thus, the previously reported angio-genic growth factor-driven decrease in caveolin abundance16,17 supports the advantages of an increased endothelium permeability, the stimulation of EC proliferation/migration and the prevention of the angiogenic process termination by mural cell recruitment. Caveolin-1 has therefore the potential to be an important prognostic indicator in delineating the degree of tumor vessel maturity and thereby the therapeutic efficacy of anti-angiogenic drugs.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Supported by the Fonds Spécial de la Recherche, the Fonds de la Recherche Scientifique Médicale, the Fonds National de la Recherche Scientifique, the Télévie, the Belgian Federation Against Cancer, the J. Maisin Foundation, and an Action de Recherche Concertée from the Communauté Française de Belgique (no. 04/09-317).
O.F. and C.D. are Fonds National de la Recherche Scientifique Senior Research Associates.
Accepted for publication August 3, 2007.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
