Abstract
Transcriptional profiling of specific elements of vasculature from animal models of vascular toxicity is an approach to gain insight into molecular mechanisms of vascular injury. Feasibility of using laser capture microdissection (LCM) to evaluate differential gene expression in selected elements of mesenteric arteries (MA) from untreated rats and rats given a single vasotoxic dose of 100 mg/kg Fenoldopam and euthanized 1 or 4 hours postdose was assessed. Regions of MA (endothelial cells [EC] and vascular smooth muscle cells [VSMC]) were selectively microdissected from optimal-cutting-temperature (O.C.T.)-embedded-frozen tissue sections. RNA was isolated, linearly amplified (LA), and hybridized to Affymetrix GeneChips®. Enrichment for specific vascular elements was evident by unique gene-expression profiles. Statistical analysis indicated that Fenoldopam treatment resulted in differential expression of 333 versus 458 genes in EC and 371 versus 618 genes in VSMC at the 1-hour or 4-hour time point, respectively. Analysis of regulated EC and VSMC genes common to both time points identified several gene functions or pathways affected by treatment. Several genes were identified in EC and/or VSMC that have not been previously linked to vascular structure or function. These data indicate that tissue–element-enrichment by LCM in conjunction with LA and GeneChip analysis offers a refined approach for assessment of injury-mediated transcriptome changes in distinct elements of the vasculature.
Keywords
Introduction
Numerous drugs are known to produce vascular damage in animal species used in nonclinical drug safety assessment (Kerns et al., 2005). Assessing clinical relevance of these findings to humans has been challenging because of the lack of selective and predictive biomarkers of vascular damage and a limited understanding of molecular and biochemical mechanisms of drug-induced vascular injury (Kerns et al., 2005). Model systems of species-specific vascular toxicity can be used as tools to investigate pathogenic mechanisms including transcriptional changes induced by vascular toxicants (Enerson et al., 2006; Kerns et al., 1989).
Fenoldopam (SKF-82526), a selective dopaminergic 1 (DA1) receptor agonist, is useful for investigating drug-induced vascular toxicity, as it reproducibly induces mesenteric artery damage in rats following a single subcutaneous vasotoxic dose ≥ 60 mg/kg (Kerns et al., 1989; Newsholme et al., 2000). Although comparison of transcriptional activity may provide insight into the mechanism of vascular toxicity and guide identification of putative biomarkers, data from expression profiling of intact vascular beds is confounded by the fact that signal is derived from whole vessels (intima, media, and adventitia) and associated tissues (fat and/or muscle). For greatest sensitivity and selectivity, transcriptional profiles need to be vascular-element–specific, that is, generated from enriched cell populations. Isolation of enriched cell populations for transcriptional profiling has been addressed through the use of laser capture microdissection (LCM; Stagliano et al., 2001).
Expression profiling of microdissected samples enriched for selected regions or cellular components of the vasculature has been demonstrated by several laboratories that have applied focused quantitative real-time reverse transcriptase polymerase chain reaction (TaqMan®) analysis to RNA extracted from micro-dissected vascular elements (Simmons et al., 2004; Stagliano et al., 2001). Although a powerful tool, expression analysis using TaqMan is restrictive in that it is limited to a set of predefined genes. In contrast, gene-expression profiling by microarray analysis can be used to explore changes in the entire transcriptome. Challenges in applying this approach to microdissected samples are the limited amount of starting cellular RNA that can be reasonably collected from tissue sections by LCM and the requirement for high-fidelity RNA amplification.
Given these obstacles, the primary objectives of this study were to assess the feasibility of using LCM, RNA linear amplification (LA), and Affymetrix GeneChip®analysis to obtain transcription profiles from LCM-enriched vascular tissue elements (e.g., endothelial cells [EC] and vascular smooth muscle cells [VSMC]) from rats given a single vasotoxic dose of Fenoldopam. Affymetrix GeneChip microarray data obtained from amplified LCM-derived samples confirmed the feasibility of using LCM to collect highly enriched populations of EC and VSMC from rat mesenteric vasculature. This method further provided evidence that Fenoldopam administration effected changes in gene functions and pathways common to both EC and VSMC and resulted in regulation of element-specific genes before onset of morphologic lesions in mesenteric arteries.
Materials and Methods
Animals, In Vivo Treatment, and Tissue Collection
Twelve groups (6/group) of 12-week-old male Sprague-Dawley rats were obtained from Charles River Laboratories (Raleigh, North Carolina). Rats were individually housed in a controlled environment (64 to 84°F; 30% to 70% relative humidity) with a 12-hour light-dark cycle and were given Certified Rodent Diet (PMI Feeds, Inc., St. Louis, Missouri) and water ad libitum. All animal care and experimental procedures were approved and carried out at GlaxoSmithKline under standard operating procedures and in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 1996).
Each rat was given a single subcutaneous injection of 0 (vehicle) or 100 mg/kg Fenoldopam (SKF-82526; GlaxoSmithKline, King of Prussia, Pennsylvania) in 0.9% sodium chloride (10 mL/kg). All rats given vehicle or 100 mg/kg Fenoldopam were euthanized by CO2 asphyxiation and exsanguination and necropsied 1, 2, 4, 12, 24, or 48 hours postdose. A separate vehicle (control) group was used for each time point. Dose administration was staggered so that terminal procedures for each rat, at each time point, were performed at the same time of day and completed within a 1-hour period. Following exsanguination, the entire mesentery from each rat was excised, mesenteric lymph nodes were removed, mesentery was cut in half, and each half was rolled on a 3-mM diameter plastic cylinder. One half of each mesentery was placed in the base of a cryomold (Tissue-Tek, Sakura Finetek, Torrance, California) covered with Tissue-Tek O.C.T. (Optimal Cutting Temperature) compound (Sakura Finetek), snap frozen immediately in liquid nitrogen, placed on dry ice, and stored at −70°C until sectioning. The mesenteric roll was kept flat in the bottom of the cryomold to maximize the area of tissue that could be sectioned. The second half of each rolled mesentery was placed in a Tissue-Tek processing capsule (Sakura Finetek), fixed in 10% neutral buffered formalin, processed routinely into paraffin blocks, sectioned at 5 microns, placed on glass slides, and stained with hematoxylin and eosin (H&E). Mesenteric arteries were evaluated under light microscopy for presence of morphological changes.
LCM and RNA Isolation
Seven-micron–thick cryosections from mesenteric rolls of rats necropsied at the 1-hour (6 control/6 treated) or 4-hour (6 control/6 treated) postdose time point were cut using a Leica cryotome equipped with a cryosectioning aid maintained at −20°C, mounted on chilled, precleaned, 4× adhesive-coated slides (Instrumedics, Inc., Hackensack, New Jersey) that had been treated to eliminate RNAses, and stored at −70°C. A minimum of 10 sections were prepared from each rat for LCM. Immediately following removal from −70°C, slides containing frozen sections were fixed in 70% ethanol for 30 seconds, dehydrated in graded alcohols (70%, 95%, and 100%; 2 times, 30 seconds each) and xylene (2 times, 3 minutes each), air-dried completely, and dessicated for 5 minutes. Each slide was then prepared for LCM by removing any loose or nonspecific attached components by using an LCM CapSure Pad (Arcturus Engineering, Mountain View, California).
Areas containing selected regions of rat mesenteric vasculature (EC or VSMC) were microdissected onto Macro CapSure Transfer Film Carriers (Arcturus) using a modified version of previously described methods (Bonner et al., 1997). Briefly, sections of mesentery to be microdissected were visualized on the Arcturus PixCell IIe System microscope at a magnification of 4×. Mesenteric arteries were located, and the laser was first pulsed over the area just inside the vessel wall where endothelium is known to be located. Following collection of EC-enriched samples from all suitable arteries, VSMC were microdissected. Typically, each tissue section contained approximately 10 to 12 mesenteric arteries suitable for LCM. A total of 10 slides were microdissected from each rat, corresponding to collection of EC or VSMC from ~100 to 120 mesenteric arteries/rat. A 7.5-μm laser spot size at a power range of 30 to 45 mW (milliwatts) and pulse duration of 500 μsec to 1.0 msec (milliseconds) was used to capture identified cells. The total time required to process slides and perform microdissection from endothelium-and VSMC-enriched regions of mesenteric arteries from up to ~12 slides for any given rat was ~1.5 hours following removal from −70°C. Several LCM caps containing regions of EC or VSMC from each individual rat were pooled into a single tube containing RNA lysis buffer (Stratagene, La Jolla, California) and vortexed (upside down) so that a single EC- and VSMC-enriched sample was generated for each rat. Samples from each rat were then frozen at −70°C until isolation.
Total RNA was isolated from each enriched sample, and each sample was DNAsed on an RNA isolation column using a modified version of the Absolutely RNA Microprep Kit (Stratagene). The RNA from each individual rat sample was isolated separately, and then RNA from each element of specified rats at each time point was pooled so that at each time point, 2 samples/LCM-enriched region were generated. At the 1-hour time point, each EC and VSMC sample represented a combination of RNA isolated from 3 rats/element (~300 to 360 arteries microdissected), and each LCM-enriched EC or VSMC sample collected at the 4-hour time point represented a combination of RNA isolated from 6 rats/element (~600 to 720 arteries microdissected). Following combining of RNA samples, each pooled, enriched sample was concentrated using the RNA Clean-up Kit-5 (Zymo Research, Orange, California) into a total volume of 12 μL using a modified version of the manufacturer protocol, which included 2 successive elutions (10 and 6 μL, respectively) with RNAse-free water heated to 65°C. The concentration of each pooled sample was quantified using the RiboGreen RNA Quantification Low Range Assay (Molecular Probes, Eugene, Oregon) in a total assay volume of 60 μL compared with the manufacturer-recommended assay volume of 200 μL. A 60 μL assay volume was used to increase the concentration of the samples so that concentrations would fall within the quantitative range of the assay.
Total RNA was qualitatively assessed for the presence of 18S and 28S ribosomal fragments using an Agilent 2100 Bioanalyzer and RNA Pico LabChip Kit (Agilent Technologies, Palo Alto, California). All RNA samples were diluted to approximately 3 ng/μL before qualitative assessment as it was previously determined that concentrations within a range of 1.5 to 5 ng/μL were optimal for reliable qualitative assessments of LCM-derived material (data not shown).
Affymetrix GeneChip Expression Analysis
Methods used for Affymetrix GeneChip sample preparation, hybridization, data analysis, and quantitation have been previously described (Raghavendra Rao et al., 2003). Briefly, biotinylated cRNA probes were generated from 15 ng total RNA from each sample according to the Affymetrix GeneChip Eukaryotic Small Sample Target Labeling Assay (Version II) as per the manufacturer protocol (Affymetrix, Santa Clara, California). Following each round of amplification, amplified RNA (aRNA) was quantified using the RiboGreen RNA Quantification Kit Low Range Assay (Molecular Probes, Eugene, Oregon) in a total assay volume of 60 μL and qualitatively assessed for size of transcripts using an Agilent 2100 Bioanalyzer and RNA Nano LabChip Kit (Agilent Technologies). Biotinylation of each amplified sample was performed during the second round of amplification using the BioArray™ HighYield™ T7 transcription Kit (Enzo Life Sciences Inc., Farmingdale, New York) according to the manufacturer protocol. Labeled cRNA was fragmented, size of fragments were determined using an Agilent 2100 Bioanalyzer and RNA Nano LabChip Kit (Agilent Technologies), and 10 μg of each sample was hybridized to a separate Rat Genome 230A GeneChip containing 15,841 probe sets representing 11,220 rat genes and expressed sequence tag (ESTs; unknown genes) clusters using standard Affymetrix protocols (www.affymetrix.com). Following hybridization, Affymetrix GeneChips were washed, stained, and scanned according to standard Affymetrix protocols (www.affymetrix.com).
GeneChips were analyzed using the Affymetrix Microarray Analysis Suite (MAS) 5.0 software according to standard protocols (Affymetrix, 2001, 2002a, 2002b) to assess chip quality control metrics and to evaluate and compare gene-expression data. Following MAS 5.0 analysis, a list of EC and VSMC genes that were up-regulated and/or down-regulated following Fenoldopam treatment were identified using a moderated t-test (p ≤ .00001) and a log-fold change of ≥ 1.5× compared with control. Lists of genes were further restricted to include only genes regulated at both 1-hour and 4-hour time points in respective tissue elements. Gene functions, particularly those related to vascular structure or function, were annotated by searching PubMed (http://www.ncbi.nlm.nih.gov) by gene name. Genes were subsequently categorized by reported function.
Confirmatory TaqMan Analysis
Analysis of mRNA levels from 29 selected genes shown to be regulated by Affymetrix GeneChip analysis following Fenoldopam treatment at the 1-hour and/or 4-hour time points were confirmed by TaqMan analysis. To generate sufficient quantities of cDNA from EC- or VSMC-enriched samples for TaqMan analysis and to assess possible RNA amplification bias, RNA was amplified using 2 different methods. Fifteen ng total RNA from each EC- or VSMC-enriched sample was linearly amplified, using 2 rounds of amplification, with the RiboAmp RNA Amplification System (Arcturus, Mountain View, California) followed by generation of single-stranded cDNA starting with 500 ng aRNA using SuperScript III Reverse Transcriptase (RT; Qiagen, Valencia, California). cDNA was generated for each enriched sample using 40 ng aRNA, SensiScript RT (Qiagen), and random primers (Invitrogen, Carlsbad, California). Another 15 ng total RNA from each EC- and VSMC-enriched sample was linearly amplified, and cDNA was generated using the Ovation Aminoallyl RNA Amplification System (Nugen Technologies, Inc., San Carlos, California) according to the manufacturer protocol.
Primer Express software (PE Applied Biosystems, Foster City, California) was used to design primers and fluorogenic probe sets for select genes shown to be up-regulated or down-regulated following Fenoldopam treatment. A primer and probe set for 18S ribosomal RNA (18S) was included and used as a normalizing gene. Sequences used to generate primer/probe sets for each gene of interest were chosen based on published cDNA sequences located in GenBank (www.NCBI.com). Gene-specific primers (Invitrogen, Carlsbad, California) and fluorogenic probe (BioSource International, Camarillo, California) sets used for EC- or VSMC-enriched samples are listed in Table 1. TaqMan analysis was conducted using the TaqMan PCR Core Reagent Kit (PE Applied Biosystems) for real-time RT-PCR as previously described (Dalmas et al., 2005). Data were quantitatively analyzed on an ABI Prism 7900 Sequence Detection System (PE Applied Biosystems) according to the methods of Scicchitano et al. (2003). TaqMan analysis was performed for each gene using 100 ng single-stranded cDNA. Negative controls included no template reactions (data not shown). For each transcript, TaqMan analysis was conducted in duplicate. Normal rat tissues known to express genes of interest were used to generate standard curves for each profiled gene. Standard curves were constructed for each gene to verify primer/probe efficiency. Data were evaluated using the ΔΔCT method as described in Applied Biosystems User Bulletin 2: ABI Prism 7700 Sequence Detection System (Foster City, California) and reported as fold change relative to control.
Nonisotopic In Situ Hybridization (ISH)
In situ expression of selected Fenoldopam-regulated genes was confirmed at the 1-hour time point by ISH. Digoxiginin (DIG)-labeled riboprobes (RP) were generated to genes identified to be up-regulated in EC- and/or VSMC-enriched samples following Fenoldopam treatment and included heme oxygenase 1 (HO-1), metallothionine (MT-1), p21, syndecan-4, elastase II, and colipase. Riboprobes were generated by in vitro transcription (IVT) using reverse transcriptase polymerase chain reaction (RT-PCR)–derived cDNA templates amplified with rat-gene–specific PCR primers. Primers used for generation of RP were designed using OLIGO 5.0 software (National Biosciences, Inc., Plymouth, Minnesota) from published cDNA sequences obtained from GenBank (http://www.ncbi.com) for genes of interest (Table 2). Templates were amplified from cDNA derived from rat liver using 2 rounds of RT-PCR. RT-PCR products were purified using the QIAQuick PCR Purification Kit (Qiagen, Valencia, California). In vitro transcription was performed using an RNA DIG labeling system (Roche, Indianapolis, Indiana) according to the manufacturer instructions. Serial dilutions of column purified DIG-labeled RNA products were run on ethidium-bromide–stained agarose gels to verify product size and compare relative sense and antisense probe concentrations. DIG incorporation was verified using anti-DIG antibody test strips (Boehringer Mannheim, Indianapolis, Indiana). Gels were assessed using an ImageMaster VDS-CL imaging system and Total Lab Software (Amersham Pharmacia, Uppsala, Sweden). Probes were diluted in water as appropriate to match sense and antisense concentrations and stored at −70°C as previously described (Dalmas et al., 2005). ISH was performed on cryosections of mesentery using riboprobes generated with T7 or T3 polymerase binding sequence (sense and antisense probe, respectively) and a Ventana Discovery System® and DAKO autostainer as previously described (Dalmas et al., 2005; Zimmerman et al., 2003).
Evaluation of Apoptosis by Caspase-3 Immunohistochemistry (IHC) and TUNEL Assay
Evaluation of mesenteric EC and VSMC for apoptosis was assessed using Caspase-3 IHC and TUNEL assay for each rat necropsied at the 1-, 2-, 4-, 12-, 24- and 48-hour time points. For Caspase-3 IHC, 5-micron–thick formalin-fixed paraffin-embedded (FFPE) sections of mesentery from each rat were placed on the Ventana Discovery System, and all reagents, with the exceptions noted below, were obtained from Ventana (Ventana Medical Systems, Inc., Tucson, Arizona). Briefly, sections were deparaffinized, rehydrated, and subjected to heat antigen retrieval (CC1 EDTA buffer; Ventana Medical Systems, Inc., Tucson, Arizona). Nonspecific staining was blocked with 3% hydrogen peroxide, an avidin/biotin-blocking system, and protein block (Dakocytomation, Carpinteria, California) with 5% normal goat serum. Polyclonal rabbit IgG anticleaved Caspase-3 antibody (Cell Signaling, Beverly, Massachusetts) was applied manually at a dilution of 1:250 for approximately 1 hour. Negative control reactions included omitting primary antibody and nonimmune rabbit IgG (SouthernBiotech, Birmingham, Alabama) matched for concentration. Anticleaved Caspase-3 antibody was labeled with biotinylated goat antirabbit IgG (Vector Labs, Burlingame, California) followed by streptavidin horse radish peroxidase (HRP) and 3, 3-diaminobenzidine (DAB). Sections were counterstained with hematoxylin, dehydrated in graded alcohols, cleared in xylene, coverslipped, and examined microscopically.
For assessment of apoptosis using the TUNEL assay, adjacent 5-micron–thick FFPE sections of mesentery from each rat necropsied at the 1-, 2-, 4-, 12-, 24-, and 48-hour time points were placed on the Ventana Discovery System, and all reagents, with the exceptions noted below, were obtained from the ApopTag® peroxidase In Situ Apoptosis Kit (Intergen Company, Purchase, New York). Briefly, sections were deparaffinized, rehydrated, and incubated in protease 1 (Ventana Medical Systems, Tucson, Arizona) for 4 minutes. Sections were blocked with 3% hydrogen peroxide followed by equilibration buffer and incubated for 32 minutes in working strength TdT (terminal deoxynu-cleotidyl transferase) enzyme followed by incubation with prediluted mouse antidigoxigenin biotinylated secondary antibody (Innogenex, San Ramone, California). Detection of apoptosis was accomplished by applying streptavidin HRP and DAB (Ventana Medical Systems). Negative controls included omitting TdT enzyme. Sections were counterstained with hematoxylin dehydrated in graded alcohols, cleared in xylene, coverslipped, and examined microscopically.
Results
Histopathology
A representative H&E section of mesentery from each rat was examined microscopically, and the incidence of mesenteric arterial lesions and number of susceptible arteries were tabulated (Table 3). Rat mesenteric arteries ranging from 100 to 800 μM in diameter have been shown to be susceptible to Fenoldopam-mediated vascular injury (Kerns et al., 1989). The evaluations indicated no microscopic evidence of vascular injury in mesenteric arteries of control rats at any time point or in Fenoldopam-treated rats at time points ≤4 hours. Images taken from a representative section of a control and Fenoldopam-treated rat at the 4-hour time point are shown in Figures 1A and 1B, respectively. In contrast, mesenteric arterial vascular medial necrosis was observed in mesenteric arterial smooth muscle of all Fenoldopam-treated rats necropsied at the 12-hour (Figure 1C), 24-hour (Figure 1D), and 48-hour time points. Vascular lesions observed at the 12-hour time point were characterized as focal to segmental areas with loss of medial smooth muscle cells, which were replaced by accumulation of red blood cells. Mesenteric lesions observed at the 24-hour time point were greater in severity when compared with lesions observed at the 12-hour time point and were characterized by segmental to circumferential loss of medial smooth muscle cells and replacement by accumulation of red blood cells and perivascular acute inflammatory cell infiltrates. At the 48-hour time point, vascular medial necrosis noted in Fenoldopam-treated rats was less severe, but perivascular inflammatory cell infiltrates were of greater severity than in mesenteric arteries at the 24-hour time point. Morphologic changes were consistent with those previously described (Kerns et al., 1989; Newsholme et al., 2000).
LCM, RNA Isolation and Amplification
LCM and transcriptional profiling was restricted to mesenteric arteries from rats necropsied at the 1-hour and 4-hour time points, which were before the onset of morphological changes evident by light microscopy. Figure 2A–2E illustrates microdis-section of EC- and VSMC-enriched samples from a mesenteric artery with sequential capture of samples enriched for EC or VSMC on separate caps using the Arcturus PixCell IIe LCM instrument. EC or VSMC total RNA isolated from enriched regions of ~300 to 320 mesenteric arteries from rats necropsied at the 1-hour time point yielded, on average, ~88 ng total RNA (Table 4), and total RNA isolated from enriched regions of ~600 to 720 mesenteric arteries from rats killed at the 4-hour time point yielded, on average, ~206 ng total RNA (Table 4).
Qualitative assessment of total RNA isolated from LCM samples using the Agilent Nano Lab-on-a-Chip and Biosizing assay showed intact 18S and 28S ribosomal bands (Figure 3A). The presence of the 18S and 28S ribosomal bands indicated RNA degradation was minimal. Fifteen nanograms of starting cellular RNA yielded, on average, ~0.221 μg and 90 μg aRNA following first- and second-round amplification, respectively (Table 5). aRNA transcript size for second-round amplification products ranged from approximately 200 to 2,000 bp (base-pairs; Figure 3B). aRNA quality of EC- and VSMC-enriched samples from both the 1-hour and 4-hour time points assessed by evaluation of standard Affymetrix array quality metrics including Raw Q, scaling factor, percent calls, and GAPDH/β-Actin 3’/Middle (M) signal intensity ratios (Table 6) indicated that array performances for LCM-enriched EC and VSMC samples were optimal and in accordance with the Affymetrix Data Analysis Fundamentals Manual (Affymetrix, 2002b) and comparable to non-LCM material (Luzzi et al., 2003). Similar results were obtained for all duplicate samples.
Assessment of Selective Enrichment for EC and VSMC by LCM
To assess the purity and reproducibility of EC- and VSMC-enriched samples collected by LCM, control (vehicle-treated) EC- and VSMC-enriched samples were evaluated to identify genes known to be selectively expressed in each respective cell type. A number of genes identified in literature to be selectively expressed in normal (untreated) EC or VSMC were identified by Affymetrix GeneChip Analysis to be present in control EC- or VSMC-enriched samples at both the 1-hour and 4-hour time points. In EC-enriched samples only, endothelin I, von Willebrand Factor, vascular endothelial growth factor (VEGF), and vascular cell adhesion molecule-1 (VCAM) were expressed, whereas in VSMC-enriched samples, smooth muscle alpha actin was selectively expressed. Similar expression patterns for these genes were observed for each respective independent control EC- or VSMC-enriched sample at each time point.
In addition, EC or VSMC gene lists generated from EC and VSMC-enriched samples collected from control and Fenoldopam-treated rats necropsied 1 or 4 hours postdose were compared to assess regulation of element-specific genes. This comparison was also performed to further assess success of enrichment and reproducibility of the experimental approach used to collect each enriched sample type. Evaluation of regulated genes identified in EC- and VSMC-enriched samples following Fenoldopam treatment also indicated significant enrichment for each specific vascular element (Table 7). Concordance between VSMC and EC genes are shown in Venn Diagrams (Figure 4A to 4D). Only those genes that were considered to be significantly regulated (p ≤ .00001, fold change ± 1.5-fold) in both independent EC- or VSMC-enriched samples were considered to be selectively expressed and considered for inclusion in Venn Diagrams. At the 1-hour time point, 212 of 333 regulated genes were unique to EC-enriched samples, 250 of 371 were unique to VSMC-enriched samples, and 121 were common to both (Figure 4A). At the 4-hour time point, 241 of 458 and 401 of 618 regulated genes were unique to EC- and VSMC-enriched elements, respectively, and 217 were common to both (Figure 4B).
To assess the number of regulated genes at the 1-hour and 4-hour time point for EC- and VSMC-enriched samples, additional Venn Diagrams were constructed (Figure 4C to 4D). In EC-enriched samples, 333 regulated genes were noted at the 1-hour time point versus 458 genes at the 4-hour time point, and 110 regulated genes were shared between both time points. In VSMC-enriched samples, 371 regulated genes were seen at the 1-hour time point versus 618 genes at the 4-hour time point, and 141 regulated genes were shared between both time points. Data generated for each independent EC- or VSMC-enriched sample showed similar expression patterns at each respective time point. Taken together, these data show that enrichment for the specific vascular elements were both selective and reproducible, thereby validating the methodology used in this study for selective enrichment of each vascular element.
Genes Regulated at Both 1-hour and 4-hour Time Points
Up-regulated and/or down-regulated genes and respective average fold changes and p-values for EC- and VSMC-enriched samples, common to both 1-hour and 4-hour time points, are given in Tables 8 and 9, respectively. Regulated genes identified in Tables 8 and 9 showed similar gene-expression patterns in respective independent samples at both the 1-hour and 4-hour time points. Genes showing regulation in only 1 of the 2 independent samples for each enriched sample type or at only 1 of the 2 time points were excluded from analysis. Genes are listed in categories of reported function based on annotation derived from PubMed searches. EC and VSMC gene lists and annotations are provided as supplemental information located at the publisher’s Web site (http://tpx.sagepub.com/supplemental). As shown in Tables 8 and 9, EC and VSMC gene functions regulated by Fenoldopam treatment and common to both time points include genes relating to cellular transport and proliferation, cell-cycle regulation, cytoskeletal, nuclear and plasma membrane migration, immunomodulation, dopamine receptor signaling, oxidative and cellular stress, inflammation, coagulation, extracellular matrix (ECM) signaling and remodeling, and vasomotor, vascular, and thyroid function. A number of other regulated genes, including various growth factors, hormones, transcription/nuclear factors, and genes related to cellular signaling, were common between time points (Tables 8 and 9). In addition, Fenoldopam-regulated genes, including those involved in calcium regulation/homeostasis and angiogenesis, were shown to be uniquely regulated in VSMC-enriched samples at both time points.
Confirmatory TaqMan Analysis
Nineteen genes unique to EC-enriched samples, 3 unique to VSMC-enriched samples, and 7 common to both elements that were identified as Fenoldopam-regulated by microarray analysis were confirmed by TaqMan analysis. Several genes chosen for analysis were confirmed using two different amplification methods including the Arcturus RiboAmp RNA Amplification Kit (Arcturus) and Ovation Aminoallyl RNA Amplification System (Nugen Technologies) to assess possible bias of amplification methodologies (Table 10). TaqMan analysis confirmed regulation of the selected genes following Fenoldopam treatment, and independent EC- and VSMC-enriched samples at each time point showed similar gene-expression patterns (individual replicate data not shown). Although fold changes between TaqMan and microarray results were not identical due to the increased sensitivity and dynamic range of the Taqman PCR platform, the resultant data were in similar ranges (data not shown), and therefore, the trends of regulation patterns for each gene were confirmed.
Nonisotopic In Situ Hybridization
Localization and up-regulation of genes associated with oxidative stress (MT-1 [metallothionine 1] and HO-1 [heme oxygenase 1]), cell-cycle regulation (p21), cell-extracellular matrix signaling (syndecan-4), angiotensinogen processing (elastase II), and pancreatic triglyceride hydrolysis (colipase) in mesenteric arteries of rats given 100 mg/kg Fenoldopam at the 1-hour time point were confirmed by nonisotopic ISH in mesenteric arteries of rats given 100 mg/kg Fenoldopam at the 1-hour time point. MT-1 and HO-1 expression was not observed in endothelium, VSMC, or adventitia of control rats (Figure 5A and 5C, respectively), but expression was evident in EC, VSMC, and perivascular stromal cells following treatment (Figure 5B, 5D). Minimal expression of p21 was observed in some EC, VSMC, and adventitia of control rats (Figure 5E), but expression was markedly up-regulated in all 3 compartments following treatment with Fenoldopam (Figure 5F). Basal expression of syndecan-4 was detected in EC, VSMC, and perivascular stromal cells in mesent (Figure 6A) with apparent up-regulation in all elements following treatment (Figure 6B). Elastase II and colipase expression were not observed in endothelium, VSMC, or adventitia of control rats (Figure 6C and 6E, respectively), but expression was up-regulated in EC, VSMC, and perivascular stromal cells following treatment (Figure 6D, 6F). Expression of MT-1, HO-1, p21, syndecan-4, elastase II, and colipase was not observed in slides containing mesenteric arteries hybridized with respective sense riboboprobes (data not shown). Localization and up-regulation of genes assessed by ISH showed similar expression patterns for each rat evaluated at the 1-hour time point.
Evaluation of Apoptosis by Caspase-3 Immunohistochemistry (IHC) and TUNEL
Evaluation of mesenteric EC and VSMC for apoptosis was assessed using Caspase-3 IHC and TUNEL analysis for each rat necropsied at the 1-, 2-, 4-, 12-, 24-, or 48-hour time point. Each mesenteric artery section was evaluated for the presence of apoptotic EC and VSMC, and the incidence of affected rats at each time point was tabulated (Tables 11 and 12). Apoptotic endothelial cells were rarely present by either method (Table 11), and in no case was a single animal positive for endothelial apoptosis by both TUNEL and activated Caspase-3 methods. When present, there were single or a few positive ECs in a single arterial vessel in the mesenteric arcade of an affected animal. The incidence of animals with EC apoptosis was similar between control animals and animals given 100 mg/kg Fenoldopam. When endothelial apoptosis was present in animals given Fenoldopam, it only occurred at the 24-hour and 48-hour time points (Table 11). Apoptosis of VSMC was apparent by both methods of detection in rats given 100 mg/kg Fenoldopam but was not detected in control animals (Table 12). Incidence of animals with smooth muscle apoptosis first occurred in a single rat at the 4-hour time point and increased in incidence with time postdose from 12 hours through 48 hours (Table 12). Examples of mesenteric arterial sections evaluated for Caspase-3 IHC and TUNEL analysis are illustrated in Figure 7.
EC and VSMC apoptosis was not observed by Caspase-3 IHC or TUNEL analysis for slides containing mesenteric arteries when primary antibody was omitted, when non immune rabbit IgG was used, or when TdT enzyme was omitted for each respective assay.
Discussion
This study demonstrates the feasibility of obtaining gene-expression profiles from LCM samples enriched for selected vascular elements. Sufficient amounts of high-quality total RNA can be isolated from microdissected O.C.T.-embedded samples to generate sufficient quantities of biotinylated aRNA for Affymetrix GeneChip analysis. Transcriptional data were confirmed by TaqMan analysis and ISH of selected regulated genes.
Various technical considerations were critical when collecting and performing microdissection of mesenteric vasculature for enrichment and isolation of RNA from EC and VSMC for subsequent transcript amplification and Affymetrix GeneChip profiling. To obtain a sufficient number of mesenteric arteries for subsequent microdissection, it was important to roll the mesentery on a 3-mM diameter plastic cylinder before being placed in a cryomold and frozen in liquid nitrogen. Compared with number of arteries present in unrolled mesentery cryosections, the use of rolled mesenteries increased the surface area and number of suitable arteries available for microdissection and subsequent RNA isolation.
Another critical technical aspect included the use of 4× adhesive-coated glass slides, which were necessary to provide appropriate tissue adhesion for selective microdissection. Treatment of the dehydrated slides with LCM CapSure Pads (Arcturus) was also important to limit inadvertent collection of any loose, nonadherent tissue debris. The order in which tissue elements were microdissected was also important. To effectively enrich for EC, endothelium was microdissected before VSMC. If EC were not collected first, EC would have been artifactually removed during microdissection. With the use of these critical steps, successful enrichment for each vascular element was evident by the presence of differentially regulated genes unique to each respective element (Table 6). Parameters used during LCM, including a spot size of 7.5 μm, pulse duration between 500 μsec to 1.0 msec, and power range between 30 to 45 mW, were essential for successful isolation of EC- or VSMC-enriched samples from mesenteric arteries, as the mesentery is composed mostly of fat and connective tissue. The use of larger spot sizes, longer pulse durations, and higher power ranges reduced the ability to isolate specific cellular elements.
Under the LCM operating conditions employed, tissue-lifting efficiency was selective and consistently >90%, ensuring a sufficient and near-constant amount of isolated RNA. Additionally, the time period beginning from the removal of the cryosections from −70°C to placement of the microdissected sample into lysis buffer was kept constant to ~1.5 to 2 hours to prevent RNA degradation. Preliminary studies revealed that LCM sessions greater than 2 hours led to reduced recovery of intact RNA (data not shown).
At the time this evaluation was performed, 15 ng of total RNA was the minimum amount of RNA required for amplification in our laboratory. Because of this limitation, pooling of samples from enriched elements was essential to make GeneChip microarray analysis feasible. Each pooled EC- or VSMC-enriched sample (n = 2 unique samples/enriched element/time point) contained RNA from 3 rats/sample type. The first 3 numerical rats/group/time point were combined and designated as replicate 1, and the second 3 rats/group/time point were combined and designated as replicate 2. Samples for each respective element (EC or VSMC) contained respective RNA from approximately 300 arteries per rat (~900 arteries total), which reduced the likelihood of animal-selective phenomena that may have skewed the data set. aRNA quality of EC- and VSMC-enriched samples from both the 1-hour and 4-hour time points assessed by evaluation of standard Affymetrix array quality metrics including Raw Q, scaling factor, percent calls, and GAPDH/β-Actin 3’/M signal intensity ratios indicated array performances for LCM-enriched EC and VSMC samples were optimal and in accordance with the Affymetrix Data Analysis Fundamentals Manual (Affymetrix, 2002b) and comparable to non-LCM material (Luzzi et al., 2003). The success of this evaluation provided data that indicated the resultant data were suitable for further evaluation.
Gene lists were constructed using stringent criteria (p ≤ .00001) and included only genes regulated at both 1-hour and 4-hour time points. Fenoldopam treatment effected changes in numerous genes involved in diverse cellular functions in both EC- and VSMC-enriched samples. Expression patterns of those genes shown to be regulated at both 1-hour and 4-hour time points were similar for independent EC- and VSMC-enriched samples. Several genes related to the dopamine pathway and G-protein coupled receptor (GPCR) signaling were regulated in both elements. In EC-enriched samples, DEXRAS1, a negative regulator of PKC delta and inhibitor of adenyl cyclase (Nguyen and Watts, 2006), was initially up-regulated at the 1-hour time point then down-regulated. Key regulators of G-protein signaling (Feldman and Gros, 2006; Wieland and Mittmann, 2003), were interestingly down-regulated in EC-enriched samples at the 4-hour time point but up-regulated in VSMC at both time points. Another key negative regulator of dopamine receptor signaling, G protein-coupled receptor kinase 6 (Willets et al., 2003), was surprisingly down-regulated in VSMC at both time points. In VSMC-enriched samples, expression of dopamine-1A receptor, the pharmacologic target of Fenoldopam, was also markedly up-regulated in a time-dependent manner at both time points. Up-regulation of dopamine receptor transcripts has been reported as an acute response to agonist stimulation (Chen et al., 2003) maximal at 2 hours with progressive reduction in expression. However, the increase was sustained and progressive up to 4 hours in this study. Resistance to receptor desensitization has been described for dopamine receptors that depend on de novo synthesis (Ryman-Rasmussen et al., 2005), and lack of desensitization and chronic stimulation of dopamine signaling has been reported to induce oxidative stress and cell death via induction of nitric-oxide synthase (NOS) enzymes (Chen et al., 2003). Dopamine administration has also been reported to induce oxidative stress, independent of receptor signaling, with profound effects on endothelial chemokine production (Beck et al., 2001).
Transcriptional signals suggestive of oxidative and cellular stress were regulated in both EC and VSMC. Up-regulated genes linked to oxidative stress not only included MT-1 (Cai et al., 2000) and HO-1 (Maines, 1997) but mitogen-inducible gene 6 (MIG6), which is an immediate early gene induced by mechanical strain and hypoxia (Saarikoski et al., 2002) and stellate specific protease 2, a peroxide scavenger identified in stellate liver cells (Kawada et al., 2001). Genes up-regulated and linked to cell stress included DNA-J protein (cardioprotective chaperone; Depre et al., 2003), proteasome subunit (Chuang and Madura, 2005), preproenkephalin-related sequence (cardioprotective endogenous opiate; Weil et al., 2006), ubiquitin-specific protease 2 (deubiquitinating enzyme; Ovaa et al., 2004) and Wolfram syndrome 1 (Fonseca et al., 2005), a component of unfolded protein response that counteracts ER stress. Oxidative stress has been suggested to be a key component in the mechanism of vascular injury following administration of a selective type III phoshodiesterase (PDE) inhibitor (Zhang et al., 2006).
A pattern of transcriptional changes was observed that was suggestive of cell-cycle arrest with antiproliferative and prosurvival signals in both vascular elements. P21 and NGFI-B, mediators of cell-cycle arrest (Jacobs et al., 2004; Martinez-Gonzalez and Badimon, 2005; Weber et al., 2004), were markedly up-regulated in both elements. Predominant directional changes in genes linked to cell proliferation supported an antiproliferative response suggested by down-regulation of protein tyrosine phosphatase type IV A (Werner et al., 2003) and protein kinase C theta (Altman and Villalba, 2003) and up-regulation of RNase A family 1 (Benner and Alleman, 1989) and RNB6 (Bear et al., 2000; Ohta et al., 1997) in EC and up-regulation of prostacyclin synthase in VSMC (Hara et al., 1995). In addition, there was prominent up-regulation of nephroblastoma overexpressed gene (NOV) in EC and NR4A2 and osteopontin in VSMC, genes linked to cell survival (Lin et al., 2003; Li et al., 2006; Weintraub et al., 2000). There were mixed transcriptional signals for apoptosis. Pro-apoptotic signal was evident in both elements by up-regulation of p22/PACAP response gene 1 (PRG1), an early response gene implicated in p53 mediated apoptosis (Schafer et al., 1998). Coincident anti-apoptotic signals included down-regulation of serine threonine kinase 17b, a member of death-associated protein 3 kinases (Mao et al., 2006) in EC and up-regulation of osteoprotegerin, a decoy receptor of RANKL that also binds/neutralizes TRAIL (Collin-Osdoby, 2004). Interestingly, osteoprotegrin has also been demonstrated to protect against vascular calcification (Bucay et al., 2006).
Although the gene patterns do not clearly indicate an anti- or pro-apoptotic outcome, apoptosis does not appear to play a major role in the early pathogenesis of Fenoldopam-induced vascular medial necrosis. TUNEL assays and activated Capase-3 immuno-staining were performed on mesenteries of rats necropsied at the 1-, 2-, 4-, 12-, 24-, and 48-hour postdose time points. TUNEL and activated Capase-3 immunoreactivity in VSMC of rats given Fenoldopam were generally absent before detection of morphologic vascular damage (i.e., 1-, 2-, and 4-hour time points) but increased in incidence coincident with medial necrosis and hemorrhage first observed at the 12-hour time point. Severity of vascular medial necrosis and hemorrhage was increased beginning at the 24-hour time point. Perivascular inflammatory cell infiltrates were first observed in mesenteries of Fenoldopam-treated rats at the 24-hour time point and were increased in severity in EC at the 48-hour time point. Although apoptotic EC were observed, their presence was rare and occurred at a similar incidence in both control and Fenoldopam-treated rats. Based on this information, VSMC apoptosis appears to be a secondary event in the pathogenesis of Fenoldopam-induced vascular damage. In contrast, a previous study suggested apoptosis played an important role in the pathogenesis of vascular lesions secondary to PDE III inhibitors, as significant EC and VSMC apoptosis were noted early in development of the lesions (Zhang et al., 2002). However, the authors reported that inflammatory cell infiltrates were present early in lesion formation. In another report using a PDE IV inhibitor, perivascular inflammation occurred 16 hours postdose, but VSMC apoptosis only occurred ≥ 48 hours postdose (Slim et al., 2003). The observations of Slim et al. agree with findings in this study, in which VSMC apoptosis appears to be a secondary event after Fenoldopam or PDE IV inhibitor induced vascular damage but not PDE III inhibition.
A complex pattern of genes linked to regulation of vascular tone, inflammation, and coagulation were regulated in EC and VSMC. Notable was the pronounced up-regulation of interleukin-6 (IL-6)–dependent transcription factors, CCAAT/enhancer binding protein, beta and delta (Ramji and Foka, 2002). This correlated with a dramatic 100-fold increase in IL-6 expression only evident at the 4-hour time point in Fenoldopam-treated samples. Multiple genes linked to profibrinolytic functions were up-regulated in EC including elastase I (Oudijk et al., 2000), fibrinogen (Mosesson, 2005), gamma peptide (Moadell et al., 2000) and tissue plasminogen activator (Camera et al., 1998), which may explain, in part, the absence of thrombosis with Fenoldopam-mediated vascular injury. Directional changes in genes involved in regulation of vascular tone were suggestive of both vasconstriction and vasodilation. These conflicting signals may ultimately translate or reflect contractile dysfunction or vasospasm as suggested by up-regulation of activating transcription factor 3 in VSMC, a stress-induced gene associated with cardiac contractile dysfunction (Okamoto et al., 2001).
Numerous genes linked to ECM structure, remodeling, and cell-ECM signaling were modulated by treatment. Common to both elements was up-regulation of genes linked to proteoglycans, ADAMTS-1 and syndecan-4. ADAMTS-1, a disintegrin and metalloproteinase with thrombospondin motifs 1, catalyzes the proteolysis of the proteoglycan versican (Bellon et al., 2004), possibly contributing release of matricrytins (Davis et al., 2000). Syndecan-4, a transmembrane proteogylcan that binds ECM ligands with a signaling cytoplasmic domain and an immediate early gene responsive to oscillatory stress (Li and Chaikof, 2002), was transcriptionally up-regulated in response to arterial injury.
A surprising finding was the increase in genes primarily linked, in literature, to pancreas including protease and lipase super-family related transcripts, in response to treatment. This was particularly evident in EC (Tables 7, 9). Numerous genes encoding a range of proteolytic/peptidolytic and lipolytic enzymes were typically up-regulated and included elastases, trypsin, carboxypeptidase A1 and B1, pancreatic carboxyester lipase, and phospholipase A2, as well as genes such as colipase, syncolin, and ZG-16 (Tables 7, 8). The expression of genes known to be linked to pancreatic tissue observed in LCM-enriched EC and VSMC samples is not considered to be caused by contamination with pancreas. Serial sections from each rat were prepared alongside cryosections and stained with H&E. These serial sections were referenced during microdissection and then archived. Following GeneChip and TaqMan analysis, these sections were reexamined for the presence of pancreas. In only a few control samples, pancreatic tissue was identified and was observed to a greater extent in control compared with drug-treated sections. Perivascular adventitia was also microdissected and profiled (data not presented), and this constellation of enzymes was not present in these samples. In addition, loose-tissue debris was removed prior to microdissection, and beam size or film contact with tissue sections was restricted to 7.5 microns to ensure EC and SMC enrichment. GeneChip signal intensity from treated samples for these pancreatic genes were high (~100-fold up-regulation versus respective controls) and only detected in samples obtained from Fenoldopam-treated rats. The Affymetrix GeneChip data for these genes were also confirmed by TaqMan analysis and were observed in each independent EC-and VSMC-enriched sample. Also, expression of colipase, an enzyme involved in triglyceride hydrolysis reportedly limited to pancreas (Sims and Lowe, 1992), was confirmed to be present in EC by ISH. These data support that gross contamination of mesenteric tissue with pancreatic RNA was highly unlikely. In addition, some enzymes primarily considered to be pancreatic-specific have been identified in EC or other tissues and include pancreatic-type carboxyl ester lipase (Li and Hui, 1998), chymotrypsin (Wang et al., 1998), trypsinogen 1 (Koshikawa et al., 1994), chymotrypsin (Wang et al., 1998), and pancreatic carboxyl ester lipase (Shamir et al., 1996). In addition, Koshikawa et al. (1997) have reported expression of trypsino-gen-2, a proteinase involved in blood coagulation, angiogenesis, and control of blood pressure, in human vascular EC around gastric tumors and in patients with disseminated intravascular coagulation. Koshikawa et al. (1997) have also shown by ISH that trypsin is expressed by normal epithelial cells of various tissues such as esophagus, stomach, and intestines and suggest that trypsinogen-2 may play important roles in expression of normal cellular functions in various tissues and cells. Therefore, changes in these genes in EC, which have been previously linked to pancreas, may reflect a unique response of rat mesenteric EC and possibly contribute to its sensitivity to vascular injury induced following Fenoldopam treatment.
A common transcriptional theme in EC and VSMC response to Fenoldopam treatment was down-regulation of a large number of genes linked to immune function. These included components of CD3 and the T-cell receptor, genes not typically associated with these tissue elements. Such consistent signals are likely not a result of contamination of LCM samples by lymphocytes, particularly because samples were selectively enriched for each particular element and microdissection was performed at time points before any morphological evidence of injury, including inflammatory cell infiltration (Figure 1C, 1D). In addition, changes in immune-function genes within selected vascular cells types have also been reported in cocultured EC and VSMC in response to laminar shear stress for 4 or 24 hours (Heydarkhan-Hagvall et al., 2006). Taken together, these data may suggest a protective immunomodulatory response of the vasculature to minimize potential immune response to cryptic antigens released as a consequence of cellular damage.
Recently, Kerns et al. (2005) nominated various genes as potential biomarkers of drug-induced vascular injury. The data presented in this study are consistent with a number of genes identified by Kerns et al. In EC-enriched samples, von Willebrand Factor (vWF), thrombospondin, and tissue plasminogen activator (tPA) were up-regulated at either the 1-hour and/or 4-hour time point(s), and smooth muscle alpha-actin and caveolin-1 were up-regulated in VSMC-enriched samples. Expression of these genes further indicates the validity of the methods used in this study.
Several regulated genes were identified that have not been previously linked to vascular structure or function. Reelin, a secreted ECM protein with serine protease activity that is crucial for neuronal positioning during brain development (Fatemi, 2005), was regulated in VSMC. Late-gestation lung protein 1, an extracellular matrix protein and trypsin inhibitor that mediates steroid-induced mesenchymal-epithelial interactions in fetal lung development (Kaplan et al., 1999; Oyewumi et al., 2003), was up-regulated in EC-enriched samples. In addition, galectin-related interfiber protein, related to the soluble lectin family of galectins, which has been reported to only be expressed in lens (Leffler et al., 2004; Ogden et al., 1998), was down-regulated in VSMC-enriched samples.
Overall, this work demonstrates the feasibility of expression profiling, transcript amplification, and Affymetrix GeneChip microarray analysis of selected microdissected elements of rat mesenteric vasculature to characterize transcriptome changes associated with drug-induced vascular injury and to identify specific genes and potential biomarkers associated with this process. Verification of the feasibility of this approach has paved the way for exploring drug-induced modulation of the transcriptome in vascular beds and has allowed for the isolation and evaluation of effects at the level of individual vascular tissue components.
Footnotes
Acknowledgments
The authors thank Tom Covatta for photomicrograph and ISH imaging, Dawn Zimmerman and David Mc Farland for overseeing the ISH experiments and in-life portion of the study, respectively, and Lauren Tierney, Padma Narayanan, and Steve Clark for helpful discussions.