Published online before print April 14, 2006
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
* Department of Experimental Dermatology and Interdisciplinary Clinical Research Center and
Integrated Functional Genomics, University of Münster, Germany;
Department of Pediatrics, University Hospital Münster, Germany; and
Department of Dermatology, University Hospital Mannheim, University of Heidelberg, Germany
1 Correspondence: Department of Experimental Dermatology, University of Münster, Röntgenstr. 21, 48149 Münster, Germany. E-mail: viemannd{at}uni-muenster.de
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: gene regulation • inflammation • microarray
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The diversity of the vascular bed is crucially determined by the type of endothelial cells (ECs) lining the inner vessel surface. In many studies, EC cultures are the basis for experimental designs investigating inflammatory reactions of endothelia to derive conclusions for vascular responses regarding inflammation. The most frequently used models to study micro- and macrovascular ECs in vitro are human microvascular ECs (HMEC) and macrovascular human umbilical vein ECs (HUVEC). However, results of studies based on a given EC type are sometimes transferred uncritically to other types of ECs [1 2 3 ]. Merely, E-selectin is stated frequently to be expressed differentially in HUVEC and HMEC-1, which suggests potential functional differences [4 ]. Nevertheless, it is tacitly assumed that the responses of both EC types to distinct inflammatory stimuli are basically comparable. Recent studies provided first evidence for differences in basal gene expression profiles of microvascular and macrovascular ECs [5 , 6 ]. So far, only one study exists, which compares the gene expression response of two different ECs (human blood/brain-barrier EC vs. HUVEC) upon stimulation with tumor necrosis factor (TNF) [7 ]. However, the validity of such data for individual genes is limited as a result of the small number of independent experiments and a limited statistical analysis. Such deficits generally result in a low rate of confirmation of microarray data by independent methods. It remains unclear whether differences in basal gene expression lead to an altered cellular response to distinct inflammatory stimuli in microvascular and macrovascular ECs or whether a potent inflammatory stimulus such as TNF indeed generates roughly comparable responses in both types of EC despite differences in basal gene expression. Characterizing and understanding the diversity of ECs therefore appear to be prerequisites for the interpretation of data obtained from in vitro models with regard to their relevance for inflammatory diseases in man.
We have recently determined the gene expression profile of HUVEC induced by TNF [8 ]. As such information is not available for HMEC-1, for the first time, we here characterize their gene expression response upon stimulation with TNF using oligonucleotide microarray technique. A high number of independent experiments, strict quality criteria, and an extensive statistical evaluation of 14 independent microarray analyses allowed us to reliably compare microvascular and macrovascular ECs. Besides many genes comparably induced in HMEC-1 and HUVEC, we identified a considerable number of others, which are regulated differentially by TNF, depending on the specific type of EC. It is surprising that cell type-dependent regulation showed only minor differences regarding genes involved in cell differentiation and cell cycle control but rather striking differences with respect to genes encoding chemokines, cytokines, and cell surface molecules.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
was obtained from R&D Systems (Wiesbaden, Germany). All other reagents were purchased from Sigma-Aldrich (Deisenhofen, Germany) unless otherwise specified.
Cells and cell culture
HMEC-1 were kindly provided by Dr. Francisco Candal (Centers for Disease Control, Atlanta, GA). They were cultured at 37°C in 3% CO2 in MCDB131 (Biochrom, Berlin, Germany) supplemented with 10% fetal calf serum (FCS) gold (PAA Laboratories, Cölbe, Germany), 20 mM L-glutamine, 50 µg/ml (
30 units/ml) gentamicin (Cytogen, Berlin, Germany), 10 ng/ml epidermal growth factor (EGF; Boehringer, Mannheim, Germany), and 1 µg/ml hydrocortisone (HC; Sigma-Aldrich). HUVEC were obtained from Clonetics (via Cell Systems, St. Katharinen, Germany) and cultured as described elsewhere [8
, 9
]. HMEC-1 (after thawing) and HUVEC were used for exposure to TNF between Passages 3 and 4.
DNA microarray hybridization
HMEC-1 were exposed to medium as control or 2 ng/ml TNF for 5 h in three independent experiments. Total cellular RNA was isolated, processed, and hybridized to Affymetrix Human Genome 133 A Gene Chip arrays (Affymetrix, Santa Clara, CA) according to the manufacturer’s instructions. Fluorescence data were processed by the MicroArray Suite software 5.0 (Affymetrix). With respect to HUVEC, four independent experiments had been performed as described [8
]. Fluorescence raw data of the hybridization were computed and statistically analyzed using the Expressionist Suite software package (GeneData, Basel, Switzerland) as described elsewhere [8
]. The complete datasets are deposited in the Gene Expression Omnibus database [Accession Numbers GSE2638 (HMEC) and GSE2639 (HUVEC)].
Comparing TNF-stimulated HUVEC or HMEC-1 with their unstimulated controls, we considered genes with a fold-change of >2.5 or <–2.5 and a P value of <0.05 as expressed differentially. Regarding genes that were called present (on) or absent (off), we defined genes with "on:off" ratios of 3:0, 3:1, 2:0, 0:2, 1:3, and 0:3 between unstimulated to TNF-stimulated HMEC-1 as significantly regulated. On/off events were categorized in grades of expression over background as described before [8 ], and + = n-fold range 5–10; ++ = n-fold range 10–25; +++ = n-fold range 25–50; and ++++ = n-fold >50. Indicated gene accession numbers were derived from the GeneBank database.
Real-time reverse transcriptase-polymerase chain reaction (RT-PCR)
The procedures and conditions for real-time RT-PCR using the QuantiTect SYBR Green PCR kit (Qiagen, Valencia, CA) were the same as described elsewhere [8
]. Primers (Supplemental Data, Table 1
) were designed using the Primer Express software package (Applied Biosystems, Foster City, CA) and obtained from MWG Biotech (Ebersberg, Germany). Gene expression was normalized with respect to the endogenous housekeeping control gene glyceraldehyde 3-phosphate dehydrogenase. Relative expression differences of respective genes were calculated by using the comparative threshold cycle (CT) method as described [10
].
|
View this table: [in a new window] |
Table 1. HMEC-1-Characteristic, TNF-Mediated Gene Expression
|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-induced protein 6 (TNFAIP6), and matrix metalloproteinase 12 (MMP12; Supplemental Data, Table 2A
and 2B
). |
View this table: [in a new window] |
Table 2. HUVEC-Characteristic, TNF-Mediated Gene Expression
|
 View larger version (24K): [in a new window] |
Figure 1. Differential regulation of gene expression in HUVEC and HMEC-1 after stimulation with TNF. The diagram shows the distribution of genes specifically induced by TNF in HUVEC (solid bars, n=46) and in HMEC-1 (shaded bars, n=33) analyzed by microarray analysis. Individual genes were assigned to functional groups as described in Materials and Methods.
|
Confirmation of microarray data by real-time RT-PCR
To confirm the expression pattern of genes, which were inducible in both HMEC-1 and HUVEC, we focused on the chemokine and cytokine repertoire selecting nine chemokines/cytokines and the IL-7 receptor as well as two transcription factors newly identified in HMEC-1 and HUVEC as TNF-inducible for quantitative real-time PCR (Table 3
). Microarray results of these genes ranged from n-fold regulations of 2.5 (IL-15) to 17.3 (IL-8; Supplemental Data, Table 2A
, and ref. [8
]). RT-PCR was performed in all experiments of each EC type in seven independent approaches altogether. For all 10 genes, the inducibility by TNF could be confirmed in both EC types, thereby validating microarray data (Table 3)
.
|
View this table: [in a new window] |
Table 3. TNF-Induced Gene Expression in HUVEC and HMEC-1, As Confirmed by Real-Time RT-PCR
|
|
View this table: [in a new window] |
Table 4. Differential Regulation of Gene Expression by TNF in HUVEC and HMEC-1, As Confirmed by Real-Time RT-PCR and Microarray Analysis
|
Correlation of gene expression results with protein expression
To verify gene expression at the protein level, intracellular and surface flow cytometry analyses were performed in at least three independent experiments. First, we selected ICAM-1, MCP-1, and IL-8, which are known to be TNF-inducible in ECs. For these genes, inducibility by TNF was confirmed by microarray analysis and verified at the protein level (Fig. 2
). According to microarray data, IL-1ß is one of the genes that is especially induced in HMEC-1. Intracellular flow cytometry revealed that IL-1ß is up-regulated in HMEC-1 but not in HUVEC, thus confirming microarray data (Fig. 2)
.
 View larger version (23K): [in a new window] |
Figure 2. Validation of gene expression programs in HUVEC and HMEC-1 by flow cytometry. Five genes were selected to confirm their expression at protein level. HMEC-1 and HUVEC were stimulated for 16 h with 2 ng/ml TNF and processed for surface (ICAM-1, CD69) or intracellular flow cytometry (MCP-1, IL-8, IL-1ß). Specific antibody profiles are shown as bold lines and isotype control profiles, as thin lines. Histograms show one of three representative experiments.
|
The kinetic progression of TNF-regulated gene expression in HUVEC and HMEC-1
To study whether the differences in the TNF-mediated gene expression profiles of HMEC-1 and HUVEC are a result of a different time-course of gene expression, we stimulated both EC types for 1, 3, 5, and 8 h with 2 ng/ml TNF in three independent experiments. We performed quantitative RT-PCRs for a selection of genes, which were designated before to be regulated primarily in HUVEC ("HUVEC -specific genes", Fig. 3A
) or HMEC-1 ("HMEC-1-specific genes", Fig. 3B
) or to be induced significantly by TNF in both EC types after 5 h of TNF stimulation ("HUVEC- and HMEC-1-shared genes", Fig. 3C
). CXCL6 (GCP-2), CXCL2 (Gro-ß), and LT-ß were confirmed to be induced primarily in HUVEC (Fig. 3A)
. However, for LT-ß, a significantly higher induction in HUVEC compared with HMEC-1 was not visible before 5 h of TNF stimulation (Fig. 3A)
. It might be a result of this dynamics that these experiments now confirmed the HUVEC specificity of LT-ß detected by microarray analysis, which we failed to confirm with the former independent experiments (Table 4)
. The time-course of the expression of TL1 shows that a predominant induction in HMEC-1 compared with HUVEC is only true after 5 h of TNF stimulation, whereas the peak of TL1 expression in HUVEC is reached already within 3 h (Fig. 3B)
. However, TL1 is the only one of nine genes where a different time-course of gene expression is responsible for our assignments as HMEC-1-specific, HUVEC-specific, or "shared" genes, according to microarray results. With respect to the kinetics of IGFBP5 and MMP12 expression, RT-PCR nicely demonstrates their persistent and predominant induction in HMEC-1 (Fig. 3B)
. Furthermore, the assignments of MCP-1, CXCL1, and HIVEP2 to be induced significantly in both EC types hold true over the whole time-dependent course of 8 h (Fig. 3C)
.
 View larger version (33K): [in a new window] |
Figure 3. Kinetic studies of gene expression in HUVEC and HMEC-1 after stimulation with TNF for different time periods. HMEC-1 (open bars) and HUVEC (solid bars) were stimulated for the indicated time periods with 2 ng/ml TNF or left untreated. mRNA expression of selected genes was quantified by real-time RT-PCR. The n-fold value represents the mean of the amount of TNF-induced mRNA expression compared with unstimulated control cells (n=3).
|
 View larger version (29K): [in a new window] |
Figure 4. Comparison of the influence of different culture conditions on gene expression in HUVEC and HMEC-1, which were cultured under standard conditions (open bars) as described in Materials and Methods or in the absence (HMEC-1, solid bars) or presence of EGF and HC (HUVEC, solid bars). In three independent experiments, ECs were stimulated with 2 ng/ml TNF or left untreated and processed for quantitative RT-PCR. The mean of n-fold induction of gene expression by TNF compared with the control cells is shown.
|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
With our work, we strive for the detailed comparison of the gene response patterns of HMEC-1 and HUVEC to TNF. In a recent study [8 ], we generated the TNF-mediated gene expression profile of HUVEC. In the present study, we also use oligonucleotide microarray technology for a comprehensive analysis of TNF-regulated gene expression in HMEC-1. To not alter the gene responses of the ECs by suboptimal culture conditions, we consciously used the most appropriate culture condition for each cell type to guarantee optimal cell viability and growth. A potential bias in gene expression patterns as a result of slightly different culture conditions was corrected by relation of all TNF-induced gene expression data to the relevant basal expression levels of each individual gene in both EC types. Thus, we focused our analysis exclusively on TNF-induced genes and excluded differences of basal gene expression patterns, which may be a result of differences in cell cycle or metabolism of these two EC types. Compared with other studies [5 , 7 ], we carefully applied strict quality criteria for a reliable data analysis, which included a sufficient number of independent experiments and the a priori definition of statistical inclusion and exclusion criteria. Consequently, a high reproducibility of statistically validated microarray data by RT-PCR is achieved, confirming 85% of 26 examined genes in HMEC-1. Microarray results, which could not be confirmed by RT-PCR, were mostly related to distinct sensitivities of both methods at low expression levels. Our examples indicated that the inclusion/exclusion criteria used for microarray analysis might miss genes with basal expression levels below the probe set-specific noise. RT-PCR might still detect differences in the expression of such genes as a result of a higher methodological sensitivity. The comprehensive analysis by RT-PCR and direct comparison with microarray data excellently demonstrate the importance of a careful definition of inclusion criteria, which adequately consider gene expression levels. Our study added 35 previously unknown genes to the list of TNF-regulated genes in HMEC-1. These include genes encoding functionally well-characterized proteins such as CD70 [12 ], IGFBP5 [13 14 15 ], TNFAIP6 [16 ], MMP12 [17 ], or human elongation factor-1 [18 , 19 ].
It is surprising that comparing the gene expression profiles of HUVEC and HMEC-1 after TNF stimulation, only less than half of the genes were found to be regulated by TNF in a similar manner in both cell types. Even more striking is the fact that the relative distribution of TNF-regulated genes onto functional groups differs considerably between HMEC-1 and HUVEC ECs. Especially when focusing on chemokines and cytokines, our results are in contrast to previous statements that the expression of cytokines in ECs from various tissues does not differ significantly after stimulation with potent, proinflammatory molecules [1
]. In HUVEC, TNF stimulation leads to a predominant EC-type, specific expression of chemokines, cytokines, and cell surface molecules. The remarkable, specific induction of genes such as CXCL2 (Gro-) [20
], CXCL6 (GCP-2) [21
], CSF-1 (GM-CSF) [22
, 23
], or fractalkine [24
] in HUVEC points to a higher chemoattractant activity generated after TNF stimulation, especially to neutrophils, monocytes, and activated T lymphocytes. Furthermore, the higher transcription of immunomodulatory genes by HUVEC, such as BMP-2 [25
] or LT-ß [26
, 27
], may provide an immunologically different microenvironment for recruited leukocytes. In contrast, the major part of genes induced exclusively by TNF in HMEC-1 is related to apoptosis, cell proliferation, and cell structure genes, which implies that changes in cell structure and EC survival might be especially relevant in these inflammatory microvascular ECs modulating endothelial integrity and barrier function. The most prominent representative of this functional group is VEGF C. The only proinflammatory gene, which is transcribed significantly stronger in HMEC-1 than in HUVEC, is IL-1ß, one of the major acute-phase cytokines responsible for systemic inflammatory responses [1
].
In the present study, we demonstrate for five selected genes that the expression at the mRNA level correlates well with the expression at the protein level. For strongly induced genes such as MCP-1 and IL-8 microarray analysis, RT-PCR and flow cytometry generated consistent results. In contrast to microarray data, results obtained by RT-PCR suggested an induction of the IL1-ß gene in HMEC-1 as well as in HUVEC; however, the expression level in the latter was low. It is interesting that analysis at the protein level confirmed microarray data of a relevant induction of IL-1ß only in HMEC-1, indicating that the high sensitivity of RT-PCR sometimes amplifies effects at the RNA level, which finally turns out not to be of biological relevance when studying expression at the protein level. Furthermore, we used quantitative RT-PCR analyses to address possible expression differences that might be a result of different kinetic responses in HMEC-1 and HUVEC and to different culture conditions (absence or presence of HC and EGF in the culture media). When selecting three HUVEC-specific genes, three HMEC-1-specific genes, and three HUVEC- and HMEC-1 common genes (Fig. 3) , the recording of RNA expression data over a time period of 8 h validates with only one exception (TL1): the microarray-based assignment to EC type-specific groups regarding expression regulation by TNF. Likewise, RT-PCR analyses of the same selection of genes also demonstrate, with only one exception (LT-ß), that the addition of HC and EGF to the culture medium of HUVEC or the omission of HC and EGF in the HMEC-1 medium does not significantly change the TNF-induced regulation of genes in both EC types as detected by microarray analysis (Fig. 4) .
These results, which support that microarray analyses are a reasonable tool to classify expression profiles in a superior, principal manner and focus on single genes in ECs, always demand the consideration of every condition such as kinetic course of expression, cell type, culture conditions, passage time, and others. Our data clearly indicate that by no means should it be assumed that the response to a stimulus, even to such a strong stimulus such as TNF, is roughly the same in distinct subtypes of ECs. The response to TNF stimulation is highly specific for HUVEC and HMEC-1. It has therefore to be considered that there is a much higher diversity of ECs in vivo, which may result in even greater differences of distinct tissue-specific ECs to inflammatory stimuli in vivo.
In summary, our study generates a comprehensive, TNF-mediated gene expression profile in HMEC-1 and characterized in detail the considerably different responses of HMEC-1 and HUVEC to TNF. Our results sensitize to a more cautious interpretation of in vitro data, which should result in a stronger consideration of EC type-specific effects. Conversely, the basis for a selective manipulation of EC subtypes in different inflammatory diseases may be provided by defining EC type-specific expression patterns to distinct, inflammatory stimuli.
Received September 26, 2005; revised January 17, 2006; accepted February 22, 2006.
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Brain Res. Mol. Brain Res. 115,130-146[Medline]
B- and p38 MAP kinase-dependent gene expression in TNF--stimulated primary human endothelial cells Blood 103,3365-3373 B via the I B kinase complex is both essential and sufficient for proinflammatory gene expression in primary endothelial cells J. Biol. Chem. 276,28451-28458B-dependent transcription of the monocyte chemoattractant protein-1 gene in primary endothelial cells Blood 97,46-55, GRO ß, and GRO 
activate human neutrophil and basophil leukocytes J. Biol. Chem. 268,15419-15424 and LTß Adv. Exp. Med. Biol. 512,43-48[Medline]This article has been cited by other articles:
 |
|
 |
|
  H. Chaudhury, M. Zakkar, J. Boyle, S. Cuhlmann, K. van der Heiden, L. A. Luong, J. Davis, A. Platt, J. C. Mason, R. Krams, et al. c-Jun N-Terminal Kinase Primes Endothelial Cells at Atheroprone Sites for Apoptosis Arterioscler Thromb Vasc Biol, March 1, 2010; 30(3): 546 - 553. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Mostertz, M. Stevenson, C. Acharya, I. Chan, K. Walters, W. Lamlertthon, W. Barry, J. Crawford, J. Nevins, and A. Potti Age- and Sex-Specific Genomic Profiles in Non-Small Cell Lung Cancer JAMA, February 10, 2010; 303(6): 535 - 543. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Bruegmann, R. Gruemmer, J. Neulen, and K. Motejlek Regulation of soluble vascular endothelial growth factor receptor 1 secretion from human endothelial cells by tissue inhibitor of metalloproteinase 1 Mol. Hum. Reprod., November 1, 2009; 15(11): 749 - 756. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Anguiano, S. A. Tuchman, C. Acharya, K. Salter, C. Gasparetto, F. Zhan, M. Dhodapkar, J. Nevins, B. Barlogie, J. D. Shaughnessy Jr, et al. Gene Expression Profiles of Tumor Biology Provide a Novel Approach to Prognosis and May Guide the Selection of Therapeutic Targets in Multiple Myeloma J. Clin. Oncol., September 1, 2009; 27(25): 4197 - 4203. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Johnson TNF-induced activation of pulmonary microvessel endothelial cells: a role for GSK3{beta} Am J Physiol Lung Cell Mol Physiol, April 1, 2009; 296(4): L700 - L709. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Sakakibara, C. A. Pise-Masison, J. N. Brady, and G. Tosato Gene Regulation and Functional Alterations Induced by Kaposi's Sarcoma-Associated Herpesvirus-Encoded ORFK13/vFLIP in Endothelial Cells J. Virol., March 1, 2009; 83(5): 2140 - 2153. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Marmon, J. Hinchey, P. Oh, M. Cammer, C. J. de Almeida, L. Gunther, C. S. Raine, and M. P. Lisanti Caveolin-1 Expression Determines the Route of Neutrophil Extravasation through Skin Microvasculature Am. J. Pathol., February 1, 2009; 174(2): 684 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Mori, R. E. Rempel, J. T. Chang, G. Yao, A. S. Lagoo, A. Potti, A. Bild, and J. R. Nevins Utilization of Pathway Signatures to Reveal Distinct Types of B Lymphoma in the E{micro}-myc Model and Human Diffuse Large B-Cell Lymphoma Cancer Res., October 15, 2008; 68(20): 8525 - 8534. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. R. Andrechek, S. Mori, R. E. Rempel, J. T. Chang, and J. R. Nevins Patterns of cell signaling pathway activation that characterize mammary development Development, July 15, 2008; 135(14): 2403 - 2413. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. R. Acharya, D. S. Hsu, C. K. Anders, A. Anguiano, K. H. Salter, K. S. Walters, R. C. Redman, S. A. Tuchman, C. A. Moylan, S. Mukherjee, et al. Gene Expression Signatures, Clinicopathological Features, and Individualized Therapy in Breast Cancer JAMA, April 2, 2008; 299(13): 1574 - 1587. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. A. O'Brien, M. A. Richardson, D. T. Berg, B. Gerlitz, A. Gupta, and B. W. Grinnell The Puzzling Role of TRAIL in Endothelial Cell Biology Arterioscler Thromb Vasc Biol, February 1, 2008; 28(2): e5 - e6. [Full Text] [PDF]
|
|
|
|
 |
|
 D. Viemann, M. Schmidt, K. Tenbrock, S. Schmid, V. Muller, K. Klimmek, S. Ludwig, J. Roth, and M. Goebeler The Contact Allergen Nickel Triggers a Unique Inflammatory and Proangiogenic Gene Expression Pattern via Activation of NF-{kappa}B and Hypoxia-Inducible Factor-1{alpha} J. Immunol., March 1, 2007; 178(5): 3198 - 3207. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
