Differential Display Identification of 40 Genes with Altered Expression in Activated Human Smooth Muscle Cells

Detailed knowledge on the molecular and cellular mechanisms that control (re)-differentiation of vascular smooth muscle cells (SMCs) is critical to understanding the pathological processes underlying atherogenesis. We identified by differential display/reverse transcriptase-polymerase chain reaction 40 genes with altered expression in cultured SMCs upon stimulation with the conditioned medium of activated macrophages. This set of genes comprises 10 known genes and 30 novel genes, which we call “smags” (forsmooth muscle activation-specificgenes). To determine the in vivo significance of these (novel) genes in atherogenesis, we performed in situ hybridization experiments on vascular tissue. Specifically, FLICE (Fas-associated death domain-like interleukin-1β-converting enzyme)-like inhibitory protein (FLIP) is expressed in neointimal SMCs as well as in lesion macrophages and endothelial cells, whereas the expression of the novel genes smag-63,smag-64, and smag-84 is restricted to neointimal SMCs. Characterization of full-length smag-64 cDNA revealed that it encodes a novel protein of 66 amino acids.smag-82 cDNA comprises the complete, unknown, 3′-untranslated region of fibroblast growth factor-5. Collectively, our results illustrate the complex changes of SMC gene expression that occur in response to stimulation with cytokines and growth factors secreted by activated macrophages. Moreover, we identified interesting candidate genes that may play a role in the differentiation of SMCs during atherogenesis.

The arterial vessel wall is composed of a single, luminal layer of endothelial cells, multiple layers of smooth muscle cells (SMCs), 1 forming the media, and the adventitia. The principal function of medial SMCs is to provide the artery with elastic properties to maintain vascular tone in response to environmental stimuli. Therefore, SMCs are equipped with a contractile apparatus, involving cytoskeletal proteins, ion channels, and specific signaling molecules. Limited information is available on SMC-specific proteins except for those, which have been proposed to be associated with the cytoskeleton, notably SM ␣-actin, myosin heavy chain subtypes, desmin, vimentin, calponin, caldesmon, SM22␣ (reviewed in Ref. 1), and smoothelin, a recently described protein (2). A typical characteristic of vascular SMCs is that they easily undergo phenotypic changes, involving the transient transformation from resting, contractile, fully differentiated SMCs into proliferative, migratory, dedifferentiated SMCs, which synthesize vast amounts of extracellular matrix components (1,3). This plasticity of SMC phenotype is relevant in adult organisms for maintenance of an intact vascular system upon vascular injury as well as during angiogenesis. In addition, accelerated proliferation and migration of SMCs plays an important role in the formation of atherosclerotic lesions, post-angioplastic restenosis, and vesselwall extension in response to hypertension.
During atherosclerosis, activated endothelial cells allow massive infiltration of monocytes into the subendothelial space, where these cells differentiate into macrophages and are ultimately converted into lipid-laden foam cells (reviewed in Ref. 4). The continuous presence of macrophages in the vessel wall results in constitutive, local secretion of cytokines and growth factors. These, mostly unknown, factors affect the quiescent medial SMCs, which subsequently undergo a transition into migrating, proliferative SMCs. At present, limited information is available on the molecular mechanisms that control the (de)differentiation program of human SMCs and, more specifically, on the genes which are involved in these processes. It should be emphasized that the human genome has been estimated to comprise up to 100,000 genes and that a function has been assigned to only 5-10% of these genes. Therefore, we assume that most of the genes, which are involved in a multifactorial phenomenon like SMC phenotype transition, are unknown. An inventory of those (novel) genes will substantially improve our knowledge on the underlying mechanisms in SMC biology and pathology, and may allow the future identification of new targets for drug discovery. We cultured human SMCs in vitro and treated the cells with a stimulus that mimics the onset of atherogenesis when the initial lipid-laden macrophages have accumulated in the subendothelium. Differential gene expression of SMCs was analyzed by differential display/ RT-PCR (DD/RT-PCR) (5,6).
In this report, we describe the identification of 40 (partial) cDNAs corresponding to mRNAs with regulated expression in activated SMCs. Among these genes are four known genes, which have been described to be involved in atherosclerosis, a set of six known genes, which have been described in relation to different cellular processes and 30 novel genes of unknown function, the so-called "smooth muscle activation-specific genes (smags). The subsequent documentation of differential expression of these genes in vascular specimen of normal vascular tissue and at different stages of the disease and characterization of the corresponding full-length cDNAs allows us to propose a potential function of these (novel) genes in atherogenesis.

EXPERIMENTAL PROCEDURES
Human Tissue Samples-Arteries were dissected from human umbilical cords and immediately frozen or used for SMC explant cultures. Apparently normal vascular tissue and early stages of atherosclerosis were obtained during organ transplantation, whereas final stages of atherosclerotic vascular tissue was acquired during vascular surgery. In each case, informed consent was obtained of patients and/or relatives according to protocols of the Medical Ethical Committee of the Academic Medical Center of Amsterdam. Tissue samples for in situ hybridization and immunohistochemistry were fixed within 15 min after resection in 3.8% (v/v) formaldehyde in phosphate-buffered saline and were subsequently paraffin-embedded.
Tissue Culture-SMCs were obtained from human artery explant cultures (passage 5 to 7) and maintained in a 1:1 mixture of RPMI and M199 (Life Technologies, Inc., Gaithersburg, MD) with 20% (v/v) human serum, penicillin, streptomycin, and fungizone. The cultured cells were characterized by immunofluorescence with a monoclonal antibody directed against SM ␣-actin (1A4, Dako, Denmark), which was detected with a Cy3-conjugated goat anti-mouse antibody (Jackson Laboratories, Westgrove, PA). With this method, the cells show uniform fibrillar staining. Confluent cultures were made quiescent by incubation overnight in serum-free medium, supplemented with insulin (5 g/ml), transferrin (5 g/ml), selenite (5 g/ml), and vitamin C (0.2 mM).
Oxidation of Human Low Density Lipoprotein (LDL) and Production of Conditioned Medium of Macrophages-LDL (200 g of protein/ml) (Sigma) was dialyzed against phosphate-buffered saline and subsequently incubated for 7 h at 37°C with copper sulfate (10 M). The reaction was stopped by the addition of 100 M EDTA. The oxidized-LDL contained 40 -50 nmol of thiobarbituric acid-reactive substances per mg of protein (7). Human elutriated monocytes (kindly provided by Dr. E. Meul, CLB, Amsterdam, The Netherlands) were allowed to attach to 80-cm 2 tissue culture flasks in 10 ml of RPMI/M199 with 5% (v/v) human serum, penicillin, streptomycin, and Fungizone per 10 8 cells. Subsequently, the attached cells were incubated for 16 h in the presence of 25-30 g/ml oxidized LDL particles. The conditioned medium was harvested, stored at Ϫ20°C, and applied at a 10-fold dilution, which resulted reproducibly in the induction of interleukin-8 (IL-8).
RNA Isolation and Northern Blotting-Total RNA was isolated with the TRIzol TM method (Life Technologies, Inc., Gaithersburg, MD). Ten g of total RNA per lane was run on a formaldehyde-agarose gel and transferred to Hybond nylon membranes (Amersham Pharmacia Biotech). The blots were (pre)hybridized in formamide-containing buffers according to standard procedures (8). Probes were radiolabeled with the random primer labeling mixture (Life Technologies, Inc.). To detect SM22␣ mRNA and calponin mRNA, rat cDNA probes were kindly provided by Dr. C. Shanahan (Cambridge, United Kingdom) (9) and to detect SM ␣-actin mRNA the oligonucleotide (5Ј-AGTGCTGTCCTCT-TCTTCACACATA-3Ј) was phosphorylated with [␥-32 P]ATP (Amersham Pharmacia Biotech) and hybridized at 50°C under standard conditions without formamide or dextrane sulfate (8). Multiple tissue poly(A) ϩ RNA blots were purchased from CLONTECH (Palo Alto, CA) and hybridized according to the enclosed protocols. As a control for equal RNA loading on the multiple tissue blots we used the 2-kb human ␤-actin cDNA probe supplied with the filters. For the other Northerns equal loading was confirmed with a GAPDH probe.
Differential Display/RT-PCR (DD/RT-PCR) and Sequencing of DD/ RT-PCR Fragments-DD/RT-PCR was performed according to Liang and Pardee (5) with minor modifications (10) and 12 random decamers as described by Bauer et al. (6). Differential bands were recovered from the gel, subcloned with TA-cloning plasmids pCRII (InVitrogen, Carlsbad, CA) or pGEM-T (Promega, Madison, WI). The clones were sequenced using the AutoRead Sequencing-kit and Cy5-labeled T7 or SP6 oligonucleotides and analyzed on the ALF-automatic sequencer (materials and protocol from Amersham Pharmacia Biotech).
RNase Protection Assay-For the RNase protection assay, [ 32 P]UTPlabeled antisense riboprobes were obtained by in vitro transcription of cDNA fragments inserted in plasmids, containing T7 and SP6 RNA polymerase transcription initiation sites. A GAPDH riboprobe of 114 nt was used as a control for equal loading, yielding a protected fragment of 84 nucleotides (nt) (bp 517 to 601 of GenBank accession number M17851). The RNase protection assay was performed with 4 g of total RNA as described before (11). For quantification of the protected bands we applied the Molecular Dynamics (Sunnyvale, CA) PhosphorImager with Image Quant software.
Immunohistochemistry-Immunostaining was performed on 5-m paraffin sections with monoclonal antibody HAM56 (Dako, Denmark) to detect macrophages and monoclonal antibody 1A4 (Dako), directed against SM ␣-actin to identify SMCs. The secondary goat anti-mouse antibody was a biotin conjugate, which was subsequently detected with StreptABComplex horseradish peroxidase (Dako). Peroxidase activity was visualized with the substrate aminoethylcarbazole and hydrogen peroxide.
In Situ Hybridization-Riboprobes were synthesized as described for the RNase protection assay, but were radiolabeled with [ 35 S]UTP (Amersham Pharmacia Biotech). For FLICE-like inhibitory protein (FLIP) a probe corresponding with bp 124 to 968 of GenBank accession number U97074 was used, for FGF-5 bp 240 to 899 of GenBank accession number M37825, for smag-84 the complete 1228-bp insert of the EST clone with GenBank accession number W46259 was transcribed, for smag-64 bp 60 to 515 and for smag-63 the complete differential display fragment of 450 bp. In situ hybridization was performed as described previously (10). For each probe both antisense and sense riboprobes were assayed, the latter ones as a control for specificity of the signal.
Cytoplasmic RNA Isolation, cDNA Library Construction and Screening-Cytoplasmic RNA was isolated from quiescent SMCs (8) and mixed with equal amounts of RNA from SMCs that were stimulated for 1 to 24 h with conditioned medium of macrophages (see above). Polyadenylated RNA was isolated with the FastTrack kit (InVitrogen).
A cDNA library was constructed essentially according to the method described by Aruffo and Seed (12). First strand cDNA was synthesized with 5 g of mRNA by incubation for 1. After phenol/chloroform extraction, the double-stranded cDNA was precipitated with ethanol. Four g of BstXI adaptors (InVitrogen) were ligated to the cDNA by incubation for 16 h at 15°C in T4 DNA ligase buffer and 400 units of T4 DNA ligase (New England Biolabs, Beverly, MA). Size fractionation of the cDNA and removal of residual adaptors was performed by agarose gel electrophoresis. Two fractions, ranging from 500 bp to 2 kb and larger than 2 kb, were excised and purified using Qiaquick spin columns (Qiagen, Hilden, Germany). Purified cDNA fractions were ligated to BstXI-cut pUC-BstXI, which is a derivative of pUC18 containing an additional polylinker including BstXI sites (a generous gift of Dr. A. Caricasole, Hubrecht Laboratory, Utrecht, the Netherlands), for 16 h at 15°C using 400 units of T4 DNA ligase. Aliquots of the ligation mixture were used for transformation of ElectroMAX DH10B TN bacteria (Life Technologies, Inc.) by electroporation.
In Vitro Transcription/Translation-After cloning full-length cDNAs into the pSP64 Poly(A) vector (Promega), in vitro transcription and translation was performed using the TnT Coupled Reticulocyte Lysate System (Promega).

Characterization of Primary SMC Cultures
Although a substantial cellular part of the human atherosclerotic vessel wall consists of SMCs, the complexity and variability of these lesions, with respect to differences in cellular composition and differentiation status of those vascular cells, is too extensive to allow a direct comparison of gene expression patterns in normal and atherosclerotic tissues. Consequently, we applied in vitro cultured SMCs in our search for novel atherosclerotic, SMC-specific genes. At present, no stable human SMC line is available, which can either display a quiescent or a proliferative phenotype, like primary cells. SMCs were obtained from explant cultures of umbilical cord artery, adult iliac artery, or abdominal aorta and grown to confluency. Subsequently, we compared the relative expression levels of the SMC-specific differentiation markers SM22␣, SM ␣-actin, and calponin in these cultures by Northern blotting (Fig. 1). Each of these genes exhibits the highest expression level in normal vascular tissue (lane 1), which is predominantly composed of fully differentiated, medial SMCs. SM22␣ is expressed in all the SMC cultures, originating from different vascular sources (panel A), with a relatively high expression level in umbilical cord artery SMCs. However, SM ␣-actin (panel B) as well as calponin (panel C) mRNA expression is restricted to the in vitro cultured SMCs obtained from umbilical cord artery. From these data we conclude that SMCs derived from this human, neonatal vascular tissue most closely resemble medial SMCs and are likely to exhibit the largest differences in gene expression upon activation, which makes them most appropriate for our studies.

Differential Display/RT-PCR Analysis on Stimulated SMCs
To mimic the conditions to which vascular SMCs are exposed at the initiation of atherosclerosis, we stimulated cultured, umbilical cord artery SMCs with the conditioned medium of human macrophages, which were activated with oxidized LDL particles. Confluent SMC cultures were kept overnight in serum-free medium and were subsequently stimulated for different periods with medium, containing 5% (v/v) human serum and appropriately diluted conditioned medium of human macrophages. To discriminate between genes, which respond to serum and genes that are regulated by macrophage-secreted components, an additional stimulation was performed for a few selected periods with 5% (v/v) human serum only. To identify immediate early, early as well as late response genes, total RNA was isolated from SMCs after 0, 1, 2, 4, 8, and 24 h of stimulation. The incubations were performed in duplicate in separate experiments. For each RNA preparation, different cDNA fractions were generated with 12 two-base anchored oligo(dT) 11 oligonucleotides (5). These cDNA fractions were amplified with 12 different random decamers (6) and the reaction products were analyzed on denaturing gels. Differential cDNA fragments that are present at identical periods of activation in both sets of RNA, were isolated from the gel, reamplified, and subcloned. Subsequently, the sequence of the differential display fragments was determined. Finally, these partial cDNA sequences were analyzed by means of the nonredundant GenBank/EMBL data base to identify known genes with established functions. Homology searches against expressed sequence tag (EST) data bases and the High-Throughput Genome sequence data bases identified known sequences to which no function has been assigned yet and the novel sequences were deposited in GenBank (Table I)  indicated with an arrow in Fig. 2A was identified as part of the 3Ј-untranslated region of granulocyte-macrophage colony stimulating factor (GM-CSF) (13), which is induced at 2-8 h of stimulation. FLICE-like inhibitory protein (FLIP) mRNA is induced at 2-4 h and was isolated from a DD-RT/PCR reaction on a single set of mRNAs ( Fig. 2B) (14). Of the novel genes shown in this figure (Fig. 2, C-E), smag-84 is induced at 2-8 h and this sequence is represented in the EST data base (Gen-Bank accession number W46259). The expression of smag-64 is also enhanced at 2 to 4 h, whereas smag-63 exemplifies a gene, which is induced only after 2 h of activation of the SMCs.

Verification and Determination of the Extent of Induction/ Repression of mRNAs, Corresponding with the Partial cDNAs Isolated by DD/RT-PCR
To confirm the differential expression pattern observed by DD/RT-PCR, we performed Northern blotting analyses with the isolated differential display fragments (Fig. 3A). As a control for equal loading of the RNA gel, the filters were hybridized with a GAPDH probe as shown in the last panel. The kinetics of mRNA expression obtained in these experiments are all in accordance with the data obtained by DD/RT-PCR as is shown for GM-CSF (13), FLIP (14), smag-64 and smag-84, respectively (Figs. 2 and 3A). TR3 orphan receptor (15) and mitogen-induced orphan receptor (MINOR) (16) exhibit kinetics of mRNA expression corresponding to that of immediate-early response genes, which are already expressed after 1 h of stimulation and are shut-off after 4 h. The expression of GM-CSF is transient from 2 to 8 h (see also Fig. 2), whereas interleukin-8 (IL-8) (17), ICAM-1 (18), and nuclear factor-B (NF-B) (19) are expressed throughout the stimulation period with optimal expression at 4 to 8 h. FLIP, ICAM, smag-53, and smag-99 are induced by 5% (v/v) human serum both in the presence and absence (lane 6c) of the macrophage conditioned medium, suggesting that these genes are at least partially induced by serum. In Table II, top, we summarized the extent of induction of the genes shown in Fig. 3A as determined by PhosphorImager analysis, as well as the size of the corresponding mRNA, deduced from the relative migration in the gel. Only 12 of the 40 partial cDNAs, isolated from the DD/RT-PCR analysis, reveal a hybridization signal on the total RNA blots, whereas the expression of the other mRNAs corresponding with the available probes is too low to be detected with this technique.
To confirm the expression patterns of the genes with relatively low expression levels, we performed RNase protection assays. In the RNase protection experiments, hybridizations were performed simultaneously with both a radiolabeled GAPDH-riboprobe and the riboprobe of the gene of interest. The GAPDH-protected band demonstrates equal loading of the lanes (example given in the last panel of Fig. 3B). In the first lanes the results are shown of a control hybridization of the riboprobes with tRNA and subsequent RNase digestion, which shows that the probes are not protected against RNase upon hybridization with nonspecific RNA. In Fig. 3B, a summary of the data obtained in the RNase protection assays is given, showing only the protected bands for each of the probes tested. From our initial experiments, we learned that the expression patterns observed in the DD/RT-PCR gels are similar in the Northern blots, an observation that allowed us to assay in the RNase protection analyses only the most informative periods of activation. For example, the RNase protection data for smag-63 exactly correspond with the DD/RT-PCR data shown in Fig. 2E. Differential expression patterns were furthermore, confirmed for ELAM-ligand fucosyltransferase (20), also called fucosyltransferase IV (21), smag-82, which turned out to be fibroblast growth factor-5 (FGF-5, see next paragraph) (22), and 10 novel genes (smag-42, -50, -56, -60, -63, -66, -71, -75, -83, and -85). Remarkably, smag-50 is the only gene identified so far which shows specific down-regulation upon SMC stimulation. The relative induction of expression of these genes, as determined in the RNase protection assays, is summarized in Table II, lower portion.
Of the 40 differentially expressed partial cDNAs isolated in the DD/RT-PCR, we validated the expression pattern of 24 genes by Northern blotting analysis and RNase protection assays. For the remaining 16 partial cDNAs the exact expression levels need to be determined by a more sensitive technique. The majority of the differentially expressed genes are novel sequences (30 out of 40), exemplifying the power of a DD/RT-PCR analysis, a technique which is apparently not biased for known or abundant mRNAs.

Expression of FLIP, FGF-5, smag-63, smag-64, and smag-84 in Human Atherosclerotic Tissue
As a first step to determine if any of the differentially expressed genes might play a role in atherogenesis, we analyzed the expression of a selected number of these genes in human vascular tissue by in situ hybridization. We determined the mRNA expression of FLIP and FGF-5, as well as of smag-84, which is represented in the EST data base, and of the genuine novel genes smag-63 and smag-64 in vascular specimens at different stages of the disease. The vascular specimens, which were applied in these experiments, were obtained from organ donors who did not have a prior history of vascular disease or from patients during reconstructive vascular surgery. Immunohistochemical analyses were performed on consecutive sections of these vascular specimens to assess the cellular composition of the specimens, in terms of the presence of neointimal SMCs and macrophages. An example of vascular tissue with a neointima of only a limited number of SMC layers is shown in Fig. 4, A-E. In this neointima macrophages are absent (Fig.  4B). FLIP mRNA is highly expressed in the endothelial cell lining of the normal vessel wall (Fig. 4C), in the adventitial capillaries (data not shown), and in some neointimal SMCs, whereas smag-84 and smag-64 are not expressed in this specimen. In an early lesion, containing both neointimal SMCs (Fig.  4, F and G) and lipid-laden macrophages (Fig. 4H), we studied the expression of these genes and found that FLIP is expressed both in neointimal SMCs and macrophages (Fig. 4I). However, smag-84 and smag-64 are not expressed in the macrophage foam cells as is shown in Fig. 4, J and K, respectively. The expression of smag-84 is restricted to some neointimal SMCs in this lesion (arrows in Fig. 4J), whereas no expression of smag-64 was observed in this lesion. Fig. 4, L-O, illustrate that smag-64 is expressed in an advanced aortic lesion by a subset of neointimal SMCs, which are scattered throughout the lesion. The expression pattern of smag-63 shows high similarity to the expression of smag-64 (data not shown). Finally, the level of expression of smag-82 (which represents a FGF-5 splice variant, see the following paragraph) is too low in these specimens to be detected by in situ hybridization (data not shown).

Isolation of Full-length cDNAs
Identification of smag-82 as FGF-5-Fragments obtained from the DD/RT-PCR analysis comprise up to 500 bp and are usually derived from the 3Ј-untranslated region. Only 2 of the 30 novel genes display homology to EST sequences that are assembled in data bases either at NCBI or at the Institute for Genomic Research (TIGR) (Table I). Obviously, full-length cDNAs would be better suited to predict a potential function of the encoded gene product. Accordingly, we constructed a cDNA library using a mixture of mRNAs isolated from both quiescent SMCs and from SMCs that were stimulated for different periods. The differential display fragment of smag-82 was used to screen pools of this (activated) SMC cDNA library, which resulted in the isolation of two independent cDNA clones of 5025 (82H) and 3535 bp (82F), respectively. Sequence analysis of these clones revealed, first, that the cDNA of 3535 bp is identical to bp 492-4027 of smag-82H and is therefore a partial cDNA. Second, the differential display fragment of smag-82 is located between bp 3860 and 4015 of the full-length cDNA, indicating that the first strand cDNA synthesis of the DD/RT-PCR reaction was not initiated upon hybridization of the anchored primer at the very 3Ј end of the message but rather at an internal adenyl-rich stretch. Third, the 5Ј end of the cDNA is identical to the reported cDNA sequence of the short variant of FGF-5, which lacks the second exon (23). Until now only a partial cDNA of FGF-5 has been described (GenBank accession number M37825, 1123 bp), which encodes only the open reading frame (ORF) of human FGF-5. We now demonstrate that  smag-53, smag-64, smag-84, smag-99, and GAPDH as a control for equal loading. The relative intensity of the bands was determined by PhosphorImager, analyzed with ImageQuant software, and the data are summarized in Table II, top, in which also the length of the differentially expressed mRNAs is given. B, RNase protection assays to confirm differential expression of mRNAs identified in the DD/RT-PCR analysis. In this figure only the protected fragments of the RNase protection assays are given. In the first lane (tRNA control) it is shown that no aspecific hybridization is observed for each of the antisense riboprobes tested. The other lanes are labeled as in A. The blank lanes indicate that the corresponding time point was not assayed for that specific probe. For each differential display riboprobe, a simultaneous hybridization was performed with the GAPDH riboprobe as a control for the total amount of RNA applied in the experiment. In the last panel an example is given of the protected GAPDH band for a complete set of RNA samples. The relative intensity of the bands was determined by PhosphorImager, analyzed with ImageQuant software, and the data are summarized in Table II, bottom. the complete FGF-5 transcript extends 5174 nt, which is in accordance with the largest hybridizing band that we observed by Northern blotting analysis (data not shown). At present, the function of the relatively long 4654 bp 3Ј-untranslated sequence of FGF-5 is unknown.
Characterization of Full-length smag-64 -Screening of the SMC cDNA library with the differential display fragment of smag-64, revealed a full-length cDNA of 2828 bp (see scheme Fig. 5A). The differential display fragment is located between bp 1240 and 1680, indicative for initiation of first strand cDNA synthesis at an internal adenyl stretch. Moreover, sequence analysis of the differential display fragment revealed that this fragment was amplified at both ends with the GC-anchored oligo(dT) primer. The orientation of this fragment was established by RNase protection assays with sense and antisense riboprobes. Application of 5Ј rapid amplification of cDNA ends did not reveal any additional sequence information, confirming that 2828 bp represents the full-length cDNA for smag-64 (data not shown). The first ORF in smag-64 starts at bp 263 and extends only 11 amino acids, whereas a subsequent ORF, in another reading frame, initiates at bp 288 and theoretically encodes a peptide of 18 amino acids. However, the longest ORF of 66 amino acid residues initiates at bp 360. To determine whether this ORF is actually translated into a polypeptide, we performed an in vitro transcription and translation assay and analyzed the products by SDS-PAGE (Fig. 5C). Clearly, a pro-tein with an apparent molecular mass of approximately 7 kDa is synthesized (Fig. 5B). This finding indicates that the largest ORF is used as a template for protein synthesis and provides yet another cDNA sequence to the growing list of sequences that violate the first AUG rule (24). Detailed analysis of the cDNA sequence, preceeding the various potential initiator AUG codons, revealed that the first and second ORF contain only 1 nt of the consensus sequence for optimal ribosome binding (GCC(A/G)CC (24)), whereas the large ORF contains 4 nt which are identical to this consensus, an observation which may explain the expression of this 7-kDa protein. An analysis of this 66-amino acid sequence by using the available protein data bases at the ExPASy Molecular Biology Server did not detect a particular motif, indicative for a specific function or characteristic of this protein, except for two phosphokinase C phosphorylation sites (at amino acid sequences 32-35 (Thr-Leu-Arg) and 51-54 (Thr-Phe-Lys), respectively). Although, the amino acid sequence of smag-64 does not harbor an obvious DNA binding motif nor a nuclear localization signal, Reinhardt's method for cytoplasmic versus nuclear discrimination predicts that this protein is localized in the nucleus.
The expression of smag-64 in normal adult tissue was analyzed on Northern blots containing polyadenylated RNA of several tissues, including multiple non-vascular SMC-containing tissues (Fig. 5D). Relatively low levels of full-length mRNA of smag-64 (about 2.8 kb) is only expressed in normal lung, whereas smag-64 mRNA is absent in pancreas, kidney, liver, placenta, brain, heart, and the SMC-containing tissues prostate, stomach, bladder, small intestine, colon, and uterus. Significantly, in skeletal muscle we observed a smaller hybridizing mRNA of approximately 2.1 kb. This shorter 2.1-kb smag-64 mRNA, may be caused by alternative polyadenylation, because two additional polyadenylation signals are present in the 3Јuntranslated region of smag-64 at bp 732 and 1930, respectively. The latter one may result in a mRNA with a length of approximately 2.1 kb, which corresponds with the estimated length of the shorter mRNA observed. DISCUSSION We performed extensive DD/RT-PCR analysis on cultured human SMCs stimulated with conditioned media from activated macrophages as a means to identify potential candidate genes involved in mediation of alterations in SMC differentiation and growth associated with atherogenesis. The assay was performed on RNA samples isolated from SMCs, which were stimulated for different periods in two independent experiments, to achieve optimal efficacy and reproducibility of the DD/RT-PCR analysis. Only bands with a regulated expression in both experiments, displaying reproducible kinetics of mRNA expression were considered for further characterization. We applied 144 different primer combinations in our DD/RT-PCR analysis (see "Experimental Procedures"), which should represent, based on statistical calculations, 80% of the entire repertoire of mRNAs (6). Collectively, we identified 10 known and 30 novel genes with modulated expression in stimulated SMCs. Our analysis substantially extends the list of genes that are potentially involved in SMC de-differentiation during atherogenesis. Furthermore, our data indicate that multiple functions of SMCs are affected by the in vitro stimulus chosen (see Table  I). With regard to the differentially expressed known genes, the cellular adhesion molecule ICAM-1 has been reported to be induced in in vitro stimulated human SMCs (25), in addition to its induction in endothelial cells. Moreover, ICAM-1 is expressed in neointimal SMCs in the atherosclerotic lesion (26,27). Induction of GM-CSF expression in cultured SMCs has been well documented (4,28), whereas expression in the atherosclerotic vessel wall has only been reported in rabbits (29). a Primer pair represents the combination of anchored primer (letters) and decamers (numbers) (6) that identified the gene by DD/RT-PCR as described in Fig. 1.
b Kinetics of (optimal) expression of the gene (in hours). c The Northern blots revealed information on the length of the mRNA. For some genes multiple hybridizing bands were observed. Estimated length of the mRNA is given in kb.
d Quantification of the radioactive bands was performed using a PhosphorImager with Image Quant software. e These fragments were isolated from multiple reactions of different anchored-primers with the same decamer. f smag-64 was amplified with two anchored primers.
We and others detected a strong induction of IL-8 expression in activated SMCs (30), although so far, IL-8 expression had been only reported in endothelial cells and macrophages in the plaque (31,32). As IL-8 has been shown to be a mitogen and chemoattractant for cultured SMCs, it may be speculated that autocrine stimulation is involved in the pathologic behavior of SMCs in atherosclerosis (33). Fuc-TIV/ELAM-ligand fucosyltransferase is an intracellular enzyme mediating the biosynthesis of a complex carbohydrate group, the so-called Lewis X antigen, which forms an essential part of the ligands for selectins (21,22). Radioactive in situ hybridizations were performed on human vascular tissue with [ 35 S]UTP-labeled antisense riboprobes for FLIP, smag-64, and smag-84 to localize mRNA expression in atherosclerotic plaques. The expression in the apparently normal vessel wall was determined in a specimen, which was obtained from an iliac artery of a female 58-year-old organ donor (A-E). Expression in an early lesion is illustrated in a specimen obtained from a female organ donor of 41 years (F-K) and a specimen with an advanced lesion was obtained during vascular surgery on an abdominal aortic aneurysm of a 81-year-old woman (L-O). Serial sections of the specimens were analyzed by immunohistochemical staining with a SMC-specific antibody directed against SM ␣-actin (A, F, G, and N) and a macrophage-specific antibody (B, H, and M), revealing the presence of (neointimal) SMCs and macrophages, respectively. An overview of the advanced lesion is given by the chemical MassonTrichrome staining (L). In C-E and O, the positive signal was visualized by epipolarization microscopy, resulting in bright blue spots, whereas in I-K bright-field microscopy is applied, which results in small black dots as the positive signal. After immunohistochemistry and in situ hybridization, the sections were counterstained with hematoxylin to detect the nuclei (purple). The expression of FLIP is shown in C and I, smag-84 in D and J, and smag-64 in E, K, and O. For each of these genes also the sense 35  tion of this protein in the atherosclerotic lesion in neointimal SMCs, endothelial cells, and macrophages with specific antibodies that only recognize the activated form of NF-B (34). Recently, NF-B has been shown to regulate an anti-apoptotic mechanism in SMCs that involves direct induction of the expression of hIAP-1 (35). We demonstrate here that hIAP-1 as well as FLIP, which both have been characterized as inhibitors of the intracellular caspases (14,36), are induced upon stimulation of SMCs, exemplifying the defense strategy of activated SMCs against programmed cell death during atherogenesis. Finally, we show a transient induction of two members of a subfamily of the steroid/thyroid hormone receptor superfamily, TR3 orphan receptor, and MINOR (15,16). It has been reported that these transcription factors are activated in many different processes involving cellular activation, differentiation, as well as in apoptosis (37, 38).
We recently described a similar, extensive DD/RT-PCR anal-ysis with the same set of primers on resting and activated endothelial cells and found 106 differentially expressed genes (10). Remarkably, a very limited overlap of identified differentially expressed genes between endothelial cells and SMCs was observed. Among the known genes, only GM-CSF, IL-8, and hIAP-1 were differentially expressed in both endothelial cells and SMCs, whereas none of the novel genes or the genes represented in the EST data bases were identified in both screens. Among the novel genes isolated from endothelial cells, 33 out of 84 were represented in the EST data bases, whereas for SMCs only 2 out of 30 novel genes are ESTs. The mere fact that the human cDNA libraries, which were applied to generate the EST data bases, were made from only a very limited number of vascular SMC-containing tissues and not from tissues containing activated SMCs, may explain the low representation of our smags in these data bases. The physiological relevance of the genes, which were isolated FIG . 5. smag-64. A, schematic representation of smag-64 cDNA with the largest ORF between bp 360 and 560 and the differential display fragment located between bp 1240 and 1680. B, sequence of the 66-amino acid ORF of smag-64. C, in vitro transcription/translation of smag-64. The complete cDNA of smag-64 was cloned in the pSP64 poly(A)-vector and translated in a reticulocyte cell lysate in the presence of [ 35 S]methionine. The reaction products were analyzed by 15% (w/v) SDS-polyacrylamide gel electrophoresis, followed by autoradiography. As controls, the reaction was performed in the absence of plasmid (Ϫ lane) and in the presence of the original vector without insert ("vector" lane). The M r markers are indicated in kDa. D, polyadenylated RNA from different tissues on Northern blots was hybridized with smag-64. The arrow indicates the hybridizing band of approximately 2.8 kb in lung and the arrowhead indicates the hybridizing band in skeletal muscle at approximately 2.1 kb. Subsequent RNA samples; 1, pancreas; 2, kidney, 3, liver; 4, lung; 5, placenta; 6, brain; 7, prostate; 8, stomach; 9, heart; 10, bladder; 11, small intestine; 12, colon; 13, uterus; and 14, skeletal muscle. Hybridization with ␤-actin is shown as a control for the amount of mRNA loaded per lane. In heart and skeletal muscle there are two forms of ␤-actin mRNA, the 2-kb form and an additional form of 1.6 -1.8 kb.
from the cultured SMCs, in atherogenesis was substantiated by mRNA expression studies in human vascular tissue at different stages of the disease. We show that FLIP is expressed in endothelial cells of the normal vessel wall, which is in accordance with the observation that quiescent, in vitro cultured endothelial cells express FLIP (39). In the latter study, it was further shown that the oxidized LDL-mediated induction of apoptosis in endothelial cells is due to the fact that FLIP expression is repressed. So far, we did not observe a clear repression of FLIP mRNA expression in endothelial cells covering the atherosclerotic lesion. However, within the plaque FLIP is prominently expressed in subsets of neointimal SMCs as well as in lipid-laden macrophages, which may indicate that both these SMCs and macrophages are protected against apoptosis by overexpression of this caspase homologue. Clearly, neointimal SMCs that are in close vicinity within a lesion can be divided in distinct subpopulations, which are characterized by specific expression patterns of these (novel) genes, for example, smag-64 and smag-84 (Fig. 4).
Most importantly, this DD/RT-PCR analysis provided us with a set of partial cDNAs which have not been isolated before, and for which a potential role of the encoded proteins in the pathologic behavior of SMCs in atherosclerosis can be anticipated. The set of novel genes, now available for in situ hybridizations, will enable us to analyze heterogeneity of SMCs in normal vascular tissue and in atherosclerotic lesions in more detail. The in situ expression data presented in this study illustrate that also in advanced lesions, which have been formed during decades, differences in gene expression patterns exist between the neointimal SMCs. This may indicate that even at later stages of the disease dynamic processes are involved in further progression of the lesion. The identification of the corresponding full-length cDNAs of the differential display fragments will allow us to study the functional involvement of the corresponding gene products in atherogenesis and to better understand the underlying processes of this disease. Ultimately, these studies will encourage us to define new targets for therapy in atherosclerosis and restenosis in angioplasty.