| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Biomedical Engineering, Duke University, Durham, North Carolina; 2United States Department of Agriculture, Agriculture Research Service, Livestock Issues Research Unit, Lubbock, Texas
Submitted 5 October 2006 ; accepted in final form 21 February 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
3,000 genes. One-way ANOVAs showed that the number of frequency-dependent genes increased as the mean shear stress was reduced. At 1 Hz, several inflammatory transcripts were repressed relative to steady flow, including VCAM and IL-8, whereas several atheroprotective transcripts were induced. The anti-inflammatory response at 1 Hz was reversed at 2 Hz. The proinflammatory response evoked by the higher frequency was most pronounced under reversing and oscillatory shear. This study suggests that arterial regions subject to both shear reversal and dominant frequencies that exceed the normal heart rate are at greater risk for atherosclerotic lesion development.
atherosclerosis; microarray; gene expression; heart rate; harmonic analysis
Whereas these studies suggest that the shape of the shear stress waveform is an important factor in disease progression, they do not delineate which features of the waveform are most important. A time-dependent waveform can be described to a large degree by three parameters: level of mean shear stress, frequency content, and amplitude of shear oscillation. It is generally accepted that the level of mean shear stress has a strong influence on endothelial cell gene expression. Recent computational flow dynamic simulations performed in our laboratory (11) have demonstrated that there are regions in the vasculature where the shear waveform is dominated by frequency components higher than the heart rate, and that the mean shear stress tends to be lower in these regions; this suggests that high harmonic content, possibly more than low mean shear, is the distinguishing characteristic of atherogenic "disturbed flow." The effect on endothelial biology of the individual components of the shear waveform has not been thoroughly examined. Thus the current study was designed to examine how the vascular endothelium responds to time-varying shear forces of different mean values and frequencies, for a fixed, physiologically realistic, shear amplitude.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
The experimental protocol was designed to test independently the effects of shear stress frequency and mean shear stress on endothelial cell morphometry and gene expression. Steady shear and four periodic shear waveforms were used. Waveforms at 1, 2, and 3 Hz consisted of a single frequency and had a sinusoidal form. An additional waveform, referred to as the physiological waveform, had a frequency composition similar to a porcine iliac artery flow waveform. Each of the five waveforms (including steady flow) was applied at three levels of mean shear stress: 0, 7.5, and 15 dyn/cm2. The amplitude of oscillation for all of the pulsatile experiments was 15 dyn/cm2. Thus the flow for the three levels of mean shear was, respectively, oscillatory, reversing, and nonreversing. Four biological replicates of each condition were performed.
Experiments were performed in an in vitro flow system. Porcine aortic endothelial cells (PAECs) were grown to confluence on a standard glass microscope slide and placed in a parallel plate flow chamber. The flow region was defined by a silicone gasket, which fit over the assembly holding the microscope slide. The flow channel was 2.15 cm wide and 0.0254 cm high. The total length of the flow region was 11.2 cm, with inlet and outlet regions 1.8 cm long bordering the glass slide. Steady fluid flow through the flow chamber was achieved by means of a constant hydrostatic pressure difference. Pulsatile flow was superimposed on the steady flow by a programmable bellows-type pump inserted upstream of the flow chamber and driven by a stepper motor. The stepper motor (IMS Mdrive 23, East Hartford, CT) has a computer interface that enables motion programming using proprietary software supplied by the manufacturer.
The volumetric flow rate at the inlet of the flow chamber was continuously monitored using a transit-time tubing flow probe (Transonic Systems, Ithaca, NY). A LabVIEW software program (National Instruments, Austin, TX) recorded the flow and adjusted an upstream valve as needed to maintain the desired flow. The flow waveform was recorded at the chamber inlet; Fourier analysis confirmed that only the desired frequency was present. Fourier analysis was also performed on the physiological flow waveform to verify that its harmonic content was similar to that of the original iliac artery flow waveform. The porcine waveform had strong first and second harmonics and a weak third harmonic component, as did the physiological waveform generated by the stepper motor (Fig. 1).
|
Cell Culture
Aortic endothelial cells were harvested from recently euthanized healthy female pigs weighing 60 kg; procedures preceding and including euthanasia were carried out under a Duke IACUC-approved protocol. For all flow experiments, cells between passages 2 and 5 were seeded at a density of 100,000 cells/cm2 onto sterilized uncoated glass slides. Cells were grown to confluence in Medium 199 supplemented with 10% fetal bovine serum, 2% porcine serum, and 1% antibiotic-antimycotic. The cells typically reached confluence within 2–3 days. When the cells reached confluence, the medium was changed to a low-serum medium (Medium 199 with 2% porcine serum and antibiotic-antimycotic) to prevent overgrowth. The cells were inserted into the flow chamber after 24 h in the low-serum medium, and the low-serum medium was used as the perfusion fluid for all experiments. The flow system was inside a humidified incubator maintained at 37°C and 5% CO2. Gas exchange between the flow medium and the incubator environment was permitted via filters on the medium reservoir. The temperature and pH of the cell culture medium at the inlet of the flow chamber were measured initially and at the end of each experiment to verify that they were within the physiological range. After shear exposure, the cells were observed at x10 magnification under phase contrast to confirm that they remained adherent. Twenty fields evenly distributed along the length of the slide were photographed using a Q-Capture CCD camera (Qimaging, Burnaby, Canada) for cell shape analysis. The slide was then rinsed using PBS with Ca2+ and Mg2+, and the cells were harvested for microarray and PCR analysis using the Ambion Protein and RNA Isolation System (PARIS) kit. The total RNA was divided into two portions: one for microarray analysis and one for real-time PCR.
Cell Shape Analysis
A MATLAB (Mathworks, Natick, MA) program was used to identify the orientation of cells with respect to the flow direction. Approximately 140 cells were identified in each x10 field. Each cell identified by the program was fit by an ellipse. The angle between the major axis of the ellipse and the direction of flow was determined for each cell. If this angle was <10°, the cell was considered to be aligned with the direction of flow. The percentage of cells aligned in each experiment was calculated. A two-way ANOVA was then performed using the statistical software MINITAB to determine whether frequency or mean shear affected cell alignment.
Microarray Analysis
The integrity of each RNA sample was validated using an Agilent Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA) to measure the ratio of 28s/18s ribosomal RNA. RNA was prepared for microarray analysis as follows. Total RNA (200 ng) from each sample was amplified using the Ambion MessageAmpII kit. Two rounds of amplification were performed on each sample. In the second round of amplification, aminoallyl-UTP was incorporated into the amplified product.
The samples were then sent to the Duke University Microarray Facility (DUMF), where Cy dyes were incorporated into the RNA strand by a coupling reaction with the aminoallyl-UTP. The samples were hybridized to Operon porcine oligonucleotide arrays (Version 1.0, Operon Biotechnologies, Huntsville, AL) printed at the DUMF. Each array assays for the expression of 10,655 optimized 70 mers, specific to Sus scrofa gene sequences, plus 24 control genes. A standard reference RNA, derived from nonsheared cells grown under normal culture conditions and labeled with a different Cy dye, was also hybridized to each array. The arrays were scanned, and the sample intensity, relative to the reference intensity was determined for each gene. In compliance with the MIAME standard for microarray data, the raw data are available in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO www.ncbi.nlm.nih.gov/projects/geo/) under the accession number GSE4281. The raw data were imported into GeneSpring (Agilent Technologies, Palo Alto, CA), and Lowess normalization was performed. The Cross Gene Error model was employed to determine a control signal threshold, which was used to select genes whose fluorescence intensity was large enough to provide a reliable result (20).
A two-way, parametric ANOVA was performed for each gene to determine whether frequency or mean shear affected gene expression. In addition, three one-way ANOVAs were performed to determine the effect of frequency at each mean shear level. A P value cutoff of 0.05 was used to screen for differentially expressed genes. The Benjamini and Hochberg False Discovery Rate multiple testing correction test (3) indicated a potential false discovery rate of 5%.
Quantitative RT-PCR
The dependence of the expression of VCAM, E-selectin, cyclooxygenase-1 (COX-1), monocyte chemoattractant protein (MCP-1), and c-jun on mean shear at 1 Hz was confirmed by using quantitative real-time PCR, as was the expression of VCAM as a function of frequency and mean shear. A two-step RT-PCR reaction was performed using a Bio-Rad MyIQ system (Bio-Rad Laboratories, Hercules, CA) to quantify the mRNA levels of these genes relative to the housekeeping gene GAPDH. Primers were designed for the PCR reaction using Primer 3 software (MIT Whitehead and Howard Hughes Medical Institute, http://frodo.wi.mit.edu/primer3/primer3_code.html). All primers were made by Integrated DNA Technologies (Coralville, IA). The amplification efficiency of each of the primer sets was tested by performing serial dilutions on a single cDNA sample and determining the number of cycles required to reach threshold for each dilution. The primer sequences and efficiencies are shown in Table 1. The 2–
CT method (18) was used to quantify the change in expression of the target genes relative to a standard control sample. Three replicates of each PCR measurement were performed.
|
An unreleased version of HTGOFAT (http://liru.ars.usda.gov/mainbioinformatics.html) was utilized to assign Gene Ontology (GO) (2), Enzyme Commission numbers (26), and mappings to the Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways (21). Statistical analyses related to over representation of GO or KEGG categories were performed using a Fisher Exact methodology similar to that described by Al-Shahrour et al. (1). Statistical P values that are equal or smaller than 0.05 indicate GO and KEGG categories that are strongly represented. Data mining for PubMed IDs was performed using a beta version module within HTGOFAT that searches PubMed abstracts using experimental conditions or terms (e.g., atherosclerosis) that co-occur along with gene names and symbols that are represented within a given dataset.
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
In all experiments, the endothelium remained adherent and the monolayer was confluent. The percentage of cells aligned in the flow direction for each experimental condition is depicted in Fig. 2. The means and standard deviations of four replicate experiments are shown. A two-way ANOVA was performed on the complete data set to determine whether frequency or mean shear stress had an effect on cell alignment. Cell alignment was significantly affected by mean shear level (P < 0.001) but not by frequency (P = 0.229).
|
The following statistical analyses are summarized in Table 2. The global effect of frequency and shear level on gene expression was evaluated using a two-way ANOVA, which identified 232 genes affected by frequency and about three thousand genes affected by mean shear. These results indicate a modest effect of pulsatile frequency, and a much larger effect of mean shear, on endothelial gene expression. Of the 232 genes sensitive to differences in frequency, 185 were also sensitive to mean shear.
|
Frequency effects. To investigate more comprehensively the effect of pulsatile frequency on gene expression, the behavior of the larger group of 232 frequency-sensitive genes was examined in greater detail. A complete list of these genes can be found in the online supplementary data. The frequency-sensitive genes were clustered using the GeneSpring Gene Tree function to compare the effect of the five waveforms on gene expression. The results of the clustering process are shown in Fig. 3. The expression of a single gene across all frequencies (including the pulsatile waveform) is represented as a horizontal row of color-coded blocks. Each column describes, for a given frequency and for all of the frequency-sensitive genes, the mean expression level of each gene at that frequency relative to the reference RNA, averaged over the 12 experiments performed at that frequency; i.e., 4 replicates x 3 mean shear levels.
|
The frequency-sensitive genes were also clustered using the GeneSpring Quality Threshold (QT) clustering algorithm. In this approach, genes are grouped into clusters based on a specified distance metric. A minimum cluster size of 10 genes was specified, and the distance metric chosen was the standard correlation, with a cutoff value of 0.9 for inclusion in the cluster. Since this methodology specifies a cutoff criterion for inclusion in a cluster, not every gene will be included in a cluster. With the use of these criteria, three clusters were identified among the frequency-sensitive genes. Roughly half of the 232 genes were included in one of the clusters. The expression patterns identified in these clusters are depicted in Fig. 4. The most common expression profile, exhibited by the 76 genes in cluster 1, is shown in Fig. 4A. These genes demonstrated decreased expression when exposed to the 1- and 3-Hz waveforms relative to their response to the other frequencies. The 25 genes in cluster 2 exhibited increased expression at 1 Hz (Fig. 4B). The expression pattern in cluster 3, which included 14 genes (Fig. 4C), is similar in appearance to cluster 1. The complete list of genes included in each cluster can be found in the online supplementary data.
|
(TGF-) and tight junction pathways were overrepresented in clusters 1 and 3. Genes with a potential role in the development or inhibition of atherosclerosis are listed in Table 4 by cluster. The genes populating clusters 1 and 3 were primarily atherogenic, whereas most of the genes in cluster 2 are thought to be atheroprotective. Among the genes present in cluster A are the heat shock protein Hsp60, and the adhesion molecule VCAM. Expression of heat shock proteins is known to be increased in atherosclerotic plaques, and heat shock proteins have also been shown to increase cytokine and adhesion molecule levels in cultured endothelial cells (19). Three antioxidant enzymes and an antioxidant protein (natural killer cell enhancing factor A, NKEF-A) were present in cluster B, the genes that showed a pattern of upregulation at 1 Hz. Cluster C, of which its pattern of downregulation at 1 Hz and 3 Hz was similar to that observed in cluster A, included the cytokine IL-8.
|
|
|
B, a molecule that sequesters NF-B in the cytoplasm and thus prevents NF-B from inducing transcription of inflammatory genes, was also repressed. However, some potentially pro-atherogenic genes were downregulated under oscillatory flow. Among these genes were calcyclin, a marker of oxidative stress in the cell (17), the proliferative transcription factor GATA-6, and the oxidative enzyme monoamine oxidase. A pro-apoptotic protein (Bcl-2 binding component 6) transcript was also repressed.
A large number of structural and remodeling genes were expressed differently under these two flow regimes. Eight transcripts associated with the extracellular matrix, including those encoding fibronectin and collagen type V, were downregulated by oscillatory flow. Expression of the transcripts for the cell junction protein 
-catenin and three cytoskeletal proteins also decreased under oscillatory flow. This may indicate that, while the cell can undergo adaptation in response to nonreversing flow, it is inhibited to some extent from doing so by low mean shear or oscillatory flow.
Fewer gene transcripts were differentially expressed under reversing flow relative to nonreversing flow. The inflammatory mediators E-selectin, MCP-1, and VCAM, which were induced by oscillatory flow, were either modestly induced by reversing flow or not differentially regulated by reversing flow, viv-a-vis nonreversing flow. Moreover, only 14 gene transcripts were induced by both oscillatory and reversing flow. Functionally similar transcripts repressed by both reversing and oscillatory flow, relative to nonreversing flow, included those encoding cathepsin S and COX-1. Several known antioxidant transcripts were repressed under reversing flow, including those encoding catalase and superoxide dismutase (SOD). The potentially pro-atherosclerotic transcripts for annexin I, connective tissue growth factor (CTGF), and TNF- were also downregulated.
Frequency-shear interactions. Figure 6 demonstrates the interactive effect of frequency and mean shear on the quantity of mRNA encoding three genes of interest: VCAM, IL-8, and c-jun. Low mean shear and the physiological and 2-Hz flows interact synergistically to promote the expression of these genes. Their differential response to frequency is more pronounced under oscillatory and reversing flow than under nonreversing flow.
|
The mRNA expression levels of MCP-1, E-selectin, COX-1, and c-jun, following 24 h of exposure to 1 Hz flow at the three mean shear levels, as measured by quantitative real time PCR, are shown in Fig. 7. These genes were chosen because the microarray experiments showed that their expression was sensitive to mean shear. A one-way ANOVA was used to test for an effect of shear level on the mRNA expression of each gene. Significant differences in expression among the shear levels were found for E-selectin, c-jun, and COX-1 (P = 0.025, P = 0.063, and P = 0.034, respectively). Consistent with the expression profiles observed in the microarray experiments, MCP-1, E-selectin, and c-jun expression decreased as the mean shear level increased. Also in agreement with the microarray findings, the expression of COX-1 increased with the level of mean shear.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
The objective of this work was to determine whether the endothelium is sensitive to the frequency of time-varying shear stress. There was no apparent effect of frequency on cell alignment, which suggests that structural adaptation to flow, at least during the first 24 h after onset, may be insensitive to pulsatile frequency. However, the array data indicate that, at the molecular level, the endothelium can sense and respond to frequencies in the 0- to 3-Hz range. This sensitivity is much more pronounced as the level of mean shear is reduced. Indeed, cells exposed to the highest mean shear did not show any statistically significant expression differences among frequencies. This finding could indicate that endothelial gene expression is most directly influenced by the frequency of flow reversal. Clustering of the frequency-sensitive genes revealed that cells exposed to any of the applied periodic flows exhibit an expression profile distinctly different from that of cells exposed to steady flow, lending further support to the notion that endothelial gene expression under steady flow does not accurately represent the response of the endothelium to in vivo fluid shear stress. The responses to the 2-Hz and physiological waveforms were similar, since 2 Hz was the strongest frequency component in the latter waveform.
Transcripts for several proinflammatory molecules (CTGF, IL-8, and VCAM) demonstrated a pattern of repression at 1 and 3 Hz and induction at 2 Hz and under the physiological waveform. In the swine model, we have shown using computational fluid mechanics that certain regions of the iliac arteries experience a shear stress profile that is dominated by the second harmonic of the flow waveform (i.e., twice the heart rate) (11). This is a consequence of the nonuniform flow field created by the branching geometry at the aortic trifurcation. Although there are some geometric differences between human and swine iliac arteries, there are enough similarities to suggest that human iliac arteries experience similar oscillations in shear stress. The average resting heart rate in humans is about 72 beats/min or 1.2 Hz, so the second harmonic is about 2.4 Hz in an average individual. Thus in vivo locations dominated by the second harmonic are likely to exhibit a phenotype similar to that of the cells in this study that were exposed to 2 Hz and may accordingly be more susceptible to atherosclerosis. Furthermore, these regions are also characteristically exposed to lower levels of mean shear stress. Since the expression of numerous inflammatory molecules was increased under oscillatory flow, the negative impact of the 2-Hz component of the shear waveform may be exacerbated in these regions; indeed, some of the apparent atherogenic effects of low shear stress may reflect the induction of inflammatory genes by a range of higher frequency components in the shear profile. This notion is supported by the PCR measurements of VCAM gene expression, in which VCAM was upregulated at 2 Hz, for all mean shear levels.
The majority of genes differentially regulated by frequency in this study do not have a known role in the progression of atherosclerosis. Many of these molecules may be involved in signal transmission within the cell. KEGG pathway analysis (provided in the supplementary data) demonstrated that a significant number of gene products are involved in the mitogen-ativated protein kinase, Wnt, TGF-, Toll-like receptor, and JaK-STAT signaling pathways. Gene ontology analysis of the gene clusters that showed higher expression under the 2-Hz and physiological waveforms indicated that the TGF- pathway was also frequency sensitive. Upstream effectors of these pathways may be part of the shear-sensing mechanism of the cell. These pathways suggest a role for cell junction proteins, receptor tyrosine kinases, and G-coupled protein receptors in the shear transduction process.
The results of this study also raise the possibility that individuals whose arteries are chronically subjected to a particular range of shear frequencies, such as those with high resting heart rates, may be at greater risk for lesion development. Indeed, this hypothesis is supported by several epidemiological studies, which found heart rate to be independently associated with a greater risk of heart disease in adult men (8). However, the small sampling of shear frequencies presented here prohibits an unambiguous specification of the harmonic environment that might induce an atherogenic response. It is unclear why the endothelium exhibited an inflammatory response at 2 Hz but not at the higher frequency of 3 Hz. The response of the endothelium to shear exposure at 3 Hz was similar to that at 1 Hz. It is possible that the response of the mechanosensor responsible for transmitting the force experienced by the cell may be too slow to transmit shear fluctuations >2 Hz into the cell. A more thorough investigation into the response of the endothelium to frequencies between 1 and 3 Hz may provide insight into how endothelial cells sense and respond to shear, as well as the role of heart rate in disease susceptibility.
|
GRANTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
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.
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
CT) Method. Methods 25: 402–408, 2001.[CrossRef][Web of Science][Medline]This article has been cited by other articles:
|
|
 |
|
  D. Nam, C.-W. Ni, A. Rezvan, J. Suo, K. Budzyn, A. Llanos, D. Harrison, D. Giddens, and H. Jo Partial carotid ligation is a model of acutely induced disturbed flow, leading to rapid endothelial dysfunction and atherosclerosis Am J Physiol Heart Circ Physiol, October 1, 2009; 297(4): H1535 - H1543. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. M. Tinken, D. H.J. Thijssen, N. Hopkins, M. A. Black, E. A. Dawson, C. T. Minson, S. C. Newcomer, M. H. Laughlin, N. T. Cable, and D. J. Green Impact of Shear Rate Modulation on Vascular Function in Humans Hypertension, August 1, 2009; 54(2): 278 - 285. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Coppola and C. Caro Arterial geometry, flow pattern, wall shear and mass transport: potential physiological significance J R Soc Interface, June 6, 2009; 6(35): 519 - 528. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. H.J. Thijssen, E. A. Dawson, T. M. Tinken, N. T. Cable, and D. J. Green Retrograde Flow and Shear Rate Acutely Impair Endothelial Function in Humans Hypertension, June 1, 2009; 53(6): 986 - 992. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Torii, N. B. Wood, N. Hadjiloizou, A. W. Dowsey, A. R. Wright, A. D. Hughes, J. Davies, D. P. Francis, J. Mayet, G.-Z. Yang, et al. Stress phase angle depicts differences in coronary artery hemodynamics due to changes in flow and geometry after percutaneous coronary intervention Am J Physiol Heart Circ Physiol, March 1, 2009; 296(3): H765 - H776. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Zhang, K. A. Burridge, and M. H. Friedman In vivo differences between endothelial transcriptional profiles of coronary and iliac arteries revealed by microarray analysis Am J Physiol Heart Circ Physiol, October 1, 2008; 295(4): H1556 - H1561. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. H. Laughlin, S. C. Newcomer, and S. B. Bender Importance of hemodynamic forces as signals for exercise-induced changes in endothelial cell phenotype J Appl Physiol, March 1, 2008; 104(3): 588 - 600. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
