compared with the apex during chronic pulmonary embolism in rats Transcriptional changes in right...

Gene expression changes associated with fibronectin-induced cardiacmyocyte hypertrophy

Hua Chen,1 Xueyin N. Huang,1 Alexandre F. R. Stewart,2 and Jorge L. Sepulveda1

1Departments of Pathology and 2Cardiovascular Institute, University of Pittsburgh, Pittsburgh, Pennsylvania 15213

Submitted 30 April 2004; accepted in final form 1 June 2004

Chen, Hua, Xueyin N. Huang, Alexandre F. R. Stewart, andJorge L. Sepulveda. Gene expression changes associated with fi-bronectin-induced cardiac myocyte hypertrophy. Physiol Genomics18: 273–283, 2004; 10.1152/physiolgenomics.00104.2004.—Fibro-nectin (FN) is an extracellular matrix protein that binds to integrinreceptors and couples cardiac myocytes to the basal lamina. CardiacFN expression is elevated in models of pressure overload, and FNcauses cultured cardiac myocytes to hypertrophy by a mechanism thathas not been characterized in detail. In this study, we analyzed thegene expression changes induced by FN in purified rat neonatalventricular myocytes using the Affymetrix RAE230A microarray, tounderstand how FN affects gene expression in cardiac myocytes andto separate the effects contributed by cardiac nonmyocytes in vivo.Pathway analysis using z-score statistics and comparison with amouse model of cardiac hypertrophy revealed several pathways stim-ulated by FN in cardiac myocytes. In addition to the known cardiacmyocyte hypertrophy markers, FN significantly induced metabolicpathways including virtually all of the enzymes of cholesterol bio-synthesis, fatty acid biosynthesis, and the mitochondrial electrontransport chain. FN also increased the expression of genes coding forribosomal proteins, translation factors, and the ubiquitin-proteasomepathway. Interestingly, cardiac myocytes plated on FN showed ele-vated expression of the fibrosis-promoting peptides connective tissuegrowth factor (CTGF), WNT1 inducible signaling pathway protein 2(WISP2), and secreted acidic cysteine-rich glycoprotein (SPARC).Our data complement in vivo studies and reveal several novel genesand pathways stimulated by FN, pointing to cardiac myocyte-specificmechanisms that lead to development of the hypertrophic phenotype.

extracellular matrix; integrin; ventricular remodeling; signal transduc-tion; gene expression profiling

OUTSIDE-IN SIGNALING CHARACTERIZES the cellular effects medi-ated by interaction of cellular receptors with the extracellularmatrix. In neonatal rat ventricular myocytes (NRVM), cluster-ing of integrin receptors induced by fibronectin (FN) (39, 40)or RGD peptides (3) elicits a hypertrophic response, charac-terized by a 1.5- to 3-fold increase in cellular area and corre-sponding increases in global mRNA and protein synthesis,together with enhanced myofibrillogenesis and sarcomeric as-sembly. FN also induces increased formation of focal adhe-sions and costameres, which are sites of contact of extracellularmatrix with integrins and associated cytoskeletal proteins as-sociated with the Z-disks of the sarcomere (39). As in othermodels of hypertrophy, the hypertrophic response is accompa-nied by changes in gene expression, with increased expressionof the natriuretic peptides atrial natriuretic peptide (ANP) andbrain natriuretic peptide (BNP) and fetal isoforms of sarco-

meric proteins such as �-myosin heavy chain (�MHC) and�-skeletal actin (�SkA).

The pathophysiological relevance of the FN-induced, inte-grin-mediated pathway is highlighted by the observation ofincreased deposition of FN in the extracellular matrix in animalmodels of cardiac hypertrophy and in patients with cardiacfailure (5, 20, 22, 29, 42, 45). The increased synthesis of FN isin part mediated by stimulation of cardiac fibroblasts by an-giotensin II, produced in response to mechanical overload (11,20, 26, 52). The extracellular matrix remodeling associatedwith interstitial fibrosis can lead to contractile and diastolicdysfunction and consequent heart failure. In contrast to thedevelopment of fibrosis in the hypertrophic heart, the geneexpression effects of increased FN deposition and consequentintegrin activation have been less appreciated. That integrin-mediated pathways are critical for the response to hypertrophicstimuli is suggested by the observations that inhibition ofoutside-in integrin signaling blocks induction of the hypertro-phic phenotype caused by phenylephrine (51, 56) or by me-chanical stretch (1, 33).

In this study, we examined the changes in gene expressioninduced by FN outside-in signaling in NRVM using high-density oligonucleotide microarrays. The results obtained pro-vide a framework to analyze in a comprehensive manner thetranscriptional response to integrin activation and suggestnovel pathways activated by FN in hypertrophic cardiac myo-cytes.

MATERIALS AND METHODS

Isolation and culture of NRVMs. Ventricular myocytes were iso-lated from 1- or 2-day-old newborn Sprague-Dawley rats according toan established protocol (14) and in compliance with the “GuidingPrinciples for the Care and Use of Laboratory Animals.” In brief,ventricles were dissected from 1 or 2 litters of about 10 pups each andincubated with Hanks’ balanced salt solution with 20 mM HEPES, pH7.4, containing 2.0 mg/ml of trypsin and 20 �g/ml DNase I untilcomplete cell dissociation. The cells were then washed, resuspendedin minimal Eagle’s medium (MEM) with 5% bovine calf serum, andpreplated in regular tissue culture dishes for 1 h to allow for attach-ment of nonmyocytes. After the preplating period, nonattached myo-cytes were removed by gentle pipetting and were plated in tissueculture dishes that were either untreated or coated with 0.5 �g/cm2 ofFN (Sigma catalog number F-1141; e.g., 1 ml of 10 �g/ml solution ona 6-cm diameter plate) or with 100 �g/cm2 of gelatin, for 3 to 4 h.Cells were plated at a density of 500 cells/mm2 and cultured for 24 hin MEM containing 5% bovine calf serum and 0.1 mM bromode-oxyuridine at 37°C in 1% CO2. Using this protocol, we routinelyachieve a purity of 90–95% myocytes, as assessed by sarcomeric�-actinin immunofluorescence. In some experiments, 1 �g/ml ofanti-�1-integrin (anti-CD29, clone Ha2/5, no azide/no endotoxin;Pharmingen, San Diego, CA) or rabbit IgG were incorporated at alltimes in the culture medium. After 24 h of culture, the medium wasreplaced with MEM containing 10 �g/ml of transferrin, 10 �g/ml of

Article published online before print. See web site for date of publication(http://physiolgenomics.physiology.org).

Address for reprint requests and other correspondence: J. L. Sepulveda,Dept. of Pathology, Univ. of Pittsburgh, CHP MT 5726, 200 Lothrop, Pitts-burgh, PA 15213 (E-mail: [email protected]).

Physiol Genomics 18: 273–283, 2004;10.1152/physiolgenomics.00104.2004.

1094-8341/04 $5.00 Copyright © 2004 the American Physiological Society 273

insulin, and 1 mg/ml of low-endotoxin bovine serum albumin (BSA).Insulin was omitted from the medium after 48 h of culture, andextracts were collected for RNA analysis after 72 h of culture.

Immunofluorescence and image analysis. Cardiac myocytes werecultured in 8-well glass chamber slides (Lab-Tek II; Nalge Nunc,Naperville, IL) coated with gelatin or FN, as described above.After 4 days of culture, the cells were washed with phosphate-buffered saline (PBS), fixed in PBS containing 2% paraformalde-hyde for 20 min, and permeabilized with pH 7.4 PBS containing0.1% Triton X-100 for 20 min. Nonspecific binding was blockedby incubation for 1 h at room temperature with PBS containing0.1% Tween 20 (PBS-T) and 5% normal donkey serum. Thefollowing primary antibodies were diluted in PBS-T with 5% goatserum and incubated for 1 h at room temperature: 1) rabbitanti-ANP (Peninsula Laboratories, Belmont, CA) and 2) mousemonoclonal anti-sarcomeric �-actinin (Sigma, St. Louis, MO). Theslides were then briefly rinsed in PBS-T and washed three times for5 min in PBS-T. Cy2-, Cy3-, or Cy5-conjugated AffiniPure donkeysecondary antibodies (anti-mouse or anti-rabbit), affinity purifiedto remove cross-reactivity with IgG from other species, were fromJackson ImmunoResearch Laboratories (West Grove, PA). Thesecondary antibodies were used at 1/300 dilution in PBS-T with5% donkey serum. In selected samples, rhodamine-phalloidin(Sigma, St. Louis, MO) was also added at 100 ng/ml for visual-ization of the actin cytoskeleton and sarcomere organization. Aftera 45-min to 1-h incubation at room temperature, the slides werewashed three times for 5 min with PBS-T and were mounted in50% glycerol in PBS containing 1 �g/ml bis-benzimide (Hoechst33258) nuclear stain. The cells were visualized in a Zeiss Axioplan2 confocal imaging system or in an Olympus BX40 epifluorescencemicroscope. Images were acquired with a SPOT camera, and imageanalysis was performed with Photoshop (Adobe Systems, San Jose,CA). To validate double or triple labeling experiments, appropriatenegative controls omitting one of the secondary or primary anti-bodies were performed to rule out unwanted cross-reactivity be-tween the various primary and secondary antibodies and to rule outoptical leaching between the different filter combinations. For cellsurface area measurements, several pictures were taken fromrepresentative fields using Nomarski interference optics on culturedishes or �-actinin immunofluorescence on fixed cells. At least 200myocytes for each data point were analyzed with the NIH IMAGEsoftware.

RNA purification and real-time RT-PCR. Cells were cultured asdescribed above, and RNA was purified using TRIzol (Invitrogen,Carlsbad, CA), according to manufacturer instructions. RNA wasprecipitated from the TRIzol reagent with isopropanol and resus-pended at 0.5–1.0 mg/ml using RNASecure solution (Ambion, Austin,TX). RNA was treated with DNase I (DNA-free, Ambion) beforereverse transcription. For reverse transcription, 60 ng of RNA weremixed with 5 �M random hexamers, 1 mM each dNTP, 7.5 mMMgCl2, 40 U RNasin (Promega, Madison, WI), 1� PCR buffer II(Applied Biosystems, Foster City, CA) and 250 U of SuperScript IIreverse transcriptase (Invitrogen). The reaction was incubated at 25°Cfor 10 min, 48°C for 45 min, and 95°C for 5 min, then cooled to 4°C.For SYBR Green quantitative real-time PCR, 2 �l of reverse tran-scription reaction was mixed with 400 nM each specific primer (seeTable 2) and 1� SYBR PCR Master Mix (Applied Biosystems). Thereaction was incubated in a model 7100 thermocycler (AppliedBiosystems) for 45 cycles consisting of denaturation at 95°C for 15 sand annealing/extension at 58–60°C for 1 min. The critical cycle (CT)was determined by the ABI Prism 7000 SDS software, and quantifi-cation of relative mRNA levels was performed by the software usinga calibration curve obtained by serial dilutions of a standard RNApreparation from rat heart.

Target RNA preparation and microarray hybridization. TargetRNA was extracted using TRIzol, as described above, and furtherpurified by the RNeasy mini kit (Qiagen, Valencia, CA). Double-

stranded cDNA was synthesized from total RNA (SuperScript IIsystem; Invitrogen). An in vitro transcription reaction was thenperformed to obtain biotin-labeled cRNA from the double-strandedcDNA (Enzo BioArray High Yield RNA Transcript Labeling kit;Enzo Diagnostics, Farmingdale, NY). The cRNA was fragmentedbefore hybridization, then mixed in a hybridization mixture con-taining probe array controls, BSA, and herring sperm DNA. Acleanup procedure was performed on the hybridization cocktailusing an RNeasy spin column (Qiagen), after which it was appliedto the Affymetrix RAE230A probe array. Hybridization was al-lowed to continue for 16 h at 45°C, after which the arrays werewashed and stained. Each probe array was scanned twice. Thecomputer workstation automatically overlaid the two scanned im-ages and averaged the intensities of each probe cell for greatestassay sensitivity.

Data analysis. GeneChip expression software (Affymetrix, SantaClara, CA) was used to determine the absolute analysis metrics usingthe probe arrays hybridization measure after scanning the arrays, asdirected by the Affymetrix GeneChip standard procedure. The aver-age difference for each probe set is an average of the differencesbetween the “perfect match” (PM) and control “mismatch” (MM)probe intensities and is directly related to the level of expression of thetranscript. The GeneChip Detection Algorithm was used to calculatea detection P value using the distance between the discriminationscore [R � (PM � MM)/(PM � MM)] and the threshold Tau �0.015, tested by a one-sided Wilcoxon signed rank test. Only thoseprobe sets with detection P values �0.05 were called “present” andconsidered for further analysis in this study. The level of expressionof each transcript (signal) was determined by GeneChip using aone-step Tukey biweight estimate. The data points for each gene ofthe FN or control arrays were normalized using a targeted normaliza-tion procedure in which the average signal from the 3 GAPDH probesets was defined as unchanged. The signals from all the experimentalarrays were then multiplied by a normalization factor to obtain similaraverage intensity between arrays for the 3 GAPDH probe sets. Thisnormalization method was chosen because the global increase in geneexpression associated with hypertrophy precludes global or evenselected “housekeeping” gene normalization procedures. We selectedGAPDH because it has been broadly used as an invariant control formRNA expression in studies of cardiac myocyte hypertrophy, and inour experience GAPDH mRNA expression by Northern blot orreal-time RT-PCR shows very little variation in cells plated in thepresence or absence of FN and other hypertrophic agents (unpublishedobservations). The corrected signals after the normalization procedurewere then used to determine the fold change of FN relative to controlby dividing the signal of each probe set in the FN array by respectivesignal in the control array. Each experiment was performed in dupli-cate, and the fold change ratios from each replicate were averaged inthe final analysis. The complete data set was deposited into the NCBIGene Expression Omnibus (GEO) database with the accession num-ber GSE1055.

Pathway analysis. Pathway analysis was performed using theGenMAPP software version 1.0 (Gladstone Institutes, UCSF, SanFrancisco, CA) (12), downloadable from http://www.genmapp.org.This software uses an identifier for each probe set to display theexpression level of each gene in a pathway. We used the Swiss-Protidentifiers for each gene based on the annotation data provided byAffymetrix, as well as a table relating rat GenBank identifiers toSwiss-Prot numbers obtained with the Dragon program (http://pevsnerlab.kennedykrieger.org/annotate.htm). When Swiss-Protidentifiers were not available, the GenBank identifier was used andreplaced in the respective gene slot in the pathway. The annotation foreach probe set of the RAE230A array was verified by downloadingthe latest annotation files from Affymetrix. In addition, we performedBLAST analysis using the probe target sequence described by Af-fymetrix for all the probe sets with significant levels of expression(present) where there were discrepancies between various probe sets

274 FIBRONECTIN-INDUCED CARDIOMYOCYTE HYPERTROPHY

Physiol Genomics • VOL 18 • www.physiolgenomics.org

corresponding to the same transcript or where the annotation was“transcribed sequences.” When genes in a pathway were not automat-ically found in the array with the GenMAPP software, we used NCBIBLASTN (http://www.ncbi.nlm.nih.gov/BLAST/) and BLAT align-ments to the UCSC rat genome browser (http://genome.ucsc.edu)(27) to identify sequences homologous to the probes in the array.Using these strategies, we assigned a specific gene (with the corre-sponding HUGO Gene Nomenclature Committee approved gene sym-bol) to 7,250 RAE230A probe sets corresponding to 5,664 differentgenes, compared with the original 5,193 probe sets corresponding to4,662 different genes from the Affymetrix annotation files. Moreover,in several instances, the original annotation was corrected based onthe results of BLAST analysis.

All the pathways in the GenMAPP database were analyzed todetermine whether a significant number of genes in each pathwaywere affected by FN vs. control (15). We examined the frequency ofgenes fulfilling a certain criterion (such as increase �1.5) in eachgroup or pathway and compared with null hypothesis expected fre-quency for that group or pathway based on the total number of genesexamined on the array. The z-score was derived by dividing thedifference between the observed number of genes meeting the crite-rion in a specific group or pathway and the expected number of genesbased on the total number of genes in the array meeting the criterionand standardizing by dividing by the standard deviation of the ob-served number of genes under the hypergeometric distribution. Theequation used was:

z �

� r � nR

N�

�n�R

N��1 �

R

N��1 �

n � 1

N � 1�

where N � the total number of genes measured in the array, R � thetotal number of genes meeting the criterion, n � the total number ofgenes in a specific group or pathway, and r � the number of genesmeeting the criterion in a specific group or pathway.

A positive z-score indicates that more genes than expected fulfilledthe criterion in a certain group or pathway; therefore, that group orpathway is likely to be affected by FN.

RESULTS

Induction of hypertrophy by FN. Cardiac myocytes plated onFN displayed several features of hypertrophy (Fig. 1). First,there was a significant increase in average cell area, both inlive, unstained myocytes (Fig. 1, A–C) as well as in fixed cellsimmunostained with anti-sarcomeric �-actinin antibody (Fig.1, compare E to G). Second, there was a marked increased insarcomeric organization and thickness (Fig. 1, D and E). Third,there was a significant increase in the expression of ANP andBNP, as measured by the percentage of actinin-positive cardiacmyocytes expressing ANP (Fig. 1, H–J). There was also a

Fig. 1. Induction of the hypertrophic pheno-type by fibronectin (FN). Cardiac myocyteswere plated in gelatin (A, D, F, and H) or inFN-coated plates (B, E, G, and I). A and B:Nomarski interference pictures of live neo-natal rat ventricular myocytes (NRVM) incultured dishes. C: changes in cell surfacearea. D and E: sarcomere assembly, afterfixation and staining of NRVM with phalloi-din. F and G show the same fields, respec-tively, as H and I. F and G: NRVM labeledwith anti-cardiac �-actinin, which highlightsthe Z-line of the sarcomeres. H and I:NRVM labeled with anti-ANP. Hypertro-phic NRVM show typical perinuclear fluo-rescence corresponding to ANP. J: quantifi-cation of the percentage of cells in severalfields with ANP-positive fluorescence. K:results of real-time quantitative RT-PCR inNRVM plated in uncoated, gelatin-coated,FN-coated, and FN-coated plates pretreatedwith either anti-�1-integrin antibody CD29or the same amount of control IgG. Resultsrepresent the average ratio of atrial natri-uretic peptide (ANP) or brain natriureticpeptide (BNP) mRNA relative to unchangedGAPDH. Error bars reflect the SD. The Pvalue was calculated using the 2-tailed un-paired Student’s t-test.

275FIBRONECTIN-INDUCED CARDIOMYOCYTE HYPERTROPHY


parallel increase in the mRNA levels of ANP and BNP relativeto GAPDH, as measured by quantitative real-time RT-PCR(Fig. 1K) and Northern blot (not shown). To demonstrate thatthis increase in ANP and BNP expression was due to FNoutside-in signaling through the integrin receptors, cells wereplated and cultured in the presence of anti-�1-integrin (CD29)antibody at all times before RNA extraction. Normal mouseIgG at the same concentrations was used as control. The results(Fig. 1K) show that �1-integrin binding to FN is required forinduction of the hypertrophic phenotype.

Microarray analysis of FN-induced gene expressionchanges. We analyzed global gene expression changes incardiac myocytes plated in FN-coated plates compared withuncoated plates using Affymetrix RAE230A microarrays.Each RAE230A array contains a total of 15,875 probe sets,excluding the control sets, which represent 14,046 uniqueUniGene identifiers (3,142 of UniGene identifiers are rep-resented by more than one probe set). A total of 7,250 probesets, representing 6,630 unique UniGene transcripts, werecalled present by the GeneChip detection algorithm in eitherthe replicate FN or control arrays. Only these were consid-ered for further analysis. Surprisingly, despite an obviouschange in phenotype with significant increases in cell area,sarcomeric assembly, and ANP/BNP expression, very fewgenes were markedly altered by FN. For example, only 296

unique transcripts were induced more than 2-fold, and only20 were induced more than 3-fold. Fourteen of the tran-scripts induced more than 3-fold were expressed at very lowlevels (signal �200) and most likely represent false-positivesignals. The only transcripts induced more than 3-fold withhigh levels of expression (signal 500) were ANP (4.4-fold), cytochrome P-450 subfamily 51 (4.0-fold), and com-ponent of oligomeric Golgi complex 1 (3.2-fold, Table 1).Similarly, of 242 genes downregulated more than 2-fold (to�0.5 of control level), only 2 unique transcripts wererepressed more than 3-fold and had initial expression levelsgreater than 500 (Table 1).

We performed real-time RT-PCR in independent samples toverify the accuracy and validity of our array results (Table 2).In general, the results of RT-PCR and microarray quantifica-tion were in agreement in the direction, if not the exactmagnitude, of change. All the genes tested were upregulated inboth assays. Moreover, we used anti-CD29 antibody to blockFN outside-in signaling through �1-integrins and measuredCyp51 induction (Fig. 2). The results confirm that most if notall of the induction of Cyp51 by FN is due to outside-insignaling involving �1-integrin.

Group and pathway analysis. Table 3 shows the z-scorescalculated for selected gene groups and pathways. The z-scorewas obtained using the formula described in MATERIALS AND

Table 1. Cardiac myocyte genes with normalized expression levels 200, induced or repressed more than 3-fold by FN

Probe Set GenBank Symbol Description Function Avg C Avg FN Avg FN/C

1367564 NM_012612 Nppa Natriuretic peptide precursor type A Peptide hormone 3,345 12,232 4.41367979 NM_012941 Cyp51a1 Cytochrome P-450, subfamily 51 Cholesterol synthesis 476 1,581 4.01373758 AI171855 Pp591 FAD-synthetase FAD synthesis 116 326 4.01371397 AI104068 Nosip Nitric oxide synthase interacting protein NO pathway 93 303 3.61372487 BI296006 Cog1 Component of oligomeric Golgi complex 1 Golgi function 191 543 3.21375048 AA799566 Mms191 MMS19 (MET18 S. cerevisiae)-like DNA repair 107 327 3.11383180 BF288097 Ppp1r2 Protein phosphatase inhibitor 2 (IPP-2) Signaling 456 144 0.31368272 D00252 Got1 Glutamate oxaloacetate transaminase 1 Aminoacid metabolism 401 164 0.31375850 BG371810 Transcribed sequence, no homology 356 116 0.31367581 AB001382 Spp1 Secreted phosphoprotein 1 Extracellular, adhesion 348 103 0.31387436 NM_022616 Cdc10 CDC10 (cell division cycle 10) Cell cycle 299 67 0.31369571 NM_023977 Gmx33 Trans-Golgi protein GMx33 Vesicle trafficking 296 83 0.31369085 NM_130738 Snurf SNRPN upstream reading frame Ribonucleoprotein 236 78 0.31372309 BI296558 Transcribed sequence, no homology 212 68 0.31377151 AI102833 Transcribed sequence, no homology 557 126 0.21387373 NM_022382 LOC64183 Myomegalin Transmembrane 512 73 0.21372325 BI303596 Emilin1 Elastin microfibril interfacer 1 Extracellular matrix 315 70 0.21390116 AA899937 Transcribed sequence, no homology 3,182 376 0.1

Cardiac myocytes were plated in fibronectin-coated (FN) or uncoated (C) plates, and mRNA expression levels were examined with Affymetrix RAE230Aoligonucleotide microarrays after 4 days of culture. The results represent the average of 2 independent microarray experiments for each condition.

Table 2. Primers and results of quantitative real-time RT-PCR in NRVM induced with fibronectin

Gene ID 5 Primer 3 Primer RT FN/C

GAPDH 1367557 ATGATTCTACCCACGGCAAG CTGGAAGATGGTGATGGGTT 1.0 1.0ANP 1367564 CCGATAGATCTGCCCTCTTG CTCTGGGCTCCAATCCTGT 7.9 4.4BNP 1367616 CTGGGAAGTCCTAGCCAGTC ACTTTTCTCTTATCAGCTCCAGCA 6.7 3.0CTGF 1367631 CGGGTTACCAATGACAATACC GTGCAGCCAGAAAGCTCAA 1.6 2.7Cyp51 1367979 TCGGGAGAGAAGTTTGCCTA AACGAAGCATAGTGGACCAAA 4.6 4.0Fdps 1367667 TCCAGCACTTCTCCCAGATT CAGCATCCTGTTTCCTTGGT 3.4 3.0Hmgcs1 1367932 TGGATGAAGGAGTGGGACTT GCAGGAAGTCTTGGCACTTT 2.7 2.6Seladin 1372012 ACCGAGAGGGCCTTGAATAC CATCCAGCCAAAGAGGTAGC 2.4 2.5

“ID” corresponds to the Affymetrix identification number for a representative RAE230A probe set complementary to the mRNA of interest; “RT” representsthe GAPDH-normalized average ratio of expression in neonatal rat ventricular myocytes (NRVM) plated in fibronectin vs. noncoated plates, whereas “FN/C”represents the same ratio measured by the RAE230A array.



METHODS, by comparing the proportion of genes increased by1.5 or more by FN, among all unique genes with a present call(1,170/6,630 � 18%), vs. the proportion of genes increased�1.5-fold among all present unique genes that belong to aspecific group or pathway. The z-score values greater than 2.0are generally considered significant. Note that the z-scoredepends on the total number of genes examined; therefore, it ishigher in larger groups. We validated this statistical approachby determining the z-score for irrelevant groups such as genesymbols starting with a specific letter and obtained z-scoresbelow 2.0.

Table 3 shows that general biosynthesis pathways relatedto hypertrophy were significantly upregulated by FN, butthere is specificity in which components of the generalassembly machinery were upregulated at the mRNA level.For example, not all genes involved in translation wereupregulated (z � 1.1), but ribosomal proteins and genes

involved in ribosomal metabolism were significantly up-regulated (z � 9.0–12.2). Similarly, mRNAs for most tran-scription factors and genes involved in transcription werenot induced (z � 0.7), but RNA polymerase-related geneswere (z � 3.8). In addition to assembly pathways, theproteolytic ubiquitin-proteasome machinery was mildly up-regulated (z � 2.6–3.4).

Among the cytoskeleton-associated genes, tubulins andtubulin-associated genes were significantly upregulated (z �3.9), whereas actin-related genes as a group were not (z �1.2). However, individual genes, such as skeletal-�-actin(Acta1), were significantly induced. Myosins and tropomyo-sins are a heterogeneous group with highly expressed sar-comeric-specific genes and a large variety of nonsarcomericmyosins. Among the highly expressed myosins and tropo-myosins (normalized levels 200), a significantly largerproportion of genes was upregulated (z � 2.8). Theseinclude tropomyosins 1, 2, and 3 (Tpm1–3), �MHC (Myh7),cardiac myosin-binding protein C (Mybpc3), cardiac myosinlight chains (Myl2, Myl3, Myl4), and smooth-muscle myo-sin light chain (Myl6).

Three metabolic pathways achieved high z-scores: choles-terol biosynthesis (z � 8.9), fatty acid biosynthesis (z � 4.2),and the mitochondrial respiratory chain (z � 6.4). Again, thez-scores are specific for these pathways, and they are muchreduced when the groups are widened. For example, consider-ing all genes involved in fatty acid metabolism reduces thez-score from 4.2 to 2.5, suggesting that our statistical analysiscorrectly identifies significantly coregulated genes specific tothe anabolic pathway.

Comparison with an aortic banding mouse model. Using theLocusLink gene symbol field as a link between the rat andmouse annotation databases and the Resourcerer programfor matching Affymetrix GeneChips (http://pga.tigr.org/tigr-scripts/magic/r1.pl), we compared our results to those ofaortic-banded mice, published by the Harvard CardiovascularGenomics group (8, 19) (Table 4 and Supplemental Table S1,

Fig. 2. Measurement of Cyp51 expression by quantitative RT-PCR in NRVM.NRVM were plated in gelatin (G) or FN plates. Some of the FN plates weretreated with CD29 anti-�1-integrin antibody (CD29) or control IgG (IgG).Values represent the ratios of Cyp51 levels normalized to GAPDH relative tobaseline (G), and error bars represent the SD. The P values were calculatedusing the 2-tailed unpaired Student’s t-test.

Table 3. Pathway analysis of genes affected by FN in NRVM

Universe Group or Pathway

zName Total FN/C � 1.5 Percent Name Total FN/C � 1.5 Percent

Unique genes, present 6,630 1,170 18% Ribosomal proteins 111 68 61% 12.2Unique genes, present 6,630 1,170 18% Description � *ribosom* 131 73 56% 11.5Unique genes, present 6,630 1,170 18% GO � *ribosom* 55 35 64% 9.0Unique genes, present 6,630 1,170 18% GO/description � *translation* 46 11 24% 1.1Unique genes, present 6,630 1,170 18% Cholesterol biosynthesis MAPP 17 17 100% 8.9Unique genes, present 6,630 1,170 18% Targets of SREBP1a 38 32 84% 10.8Unique genes, present 6,630 1,170 18% Fatty acid metabolism 48 15 31% 2.5Unique genes, present 6,630 1,170 18% Fatty acid biosynthesis 18 10 56% 4.2Unique genes, present 6,630 1,170 18% Respiratory chain 77 35 45% 6.4Unique genes, present 6,630 1,170 18% Proteasome 36 14 39% 3.4Unique genes, present 6,630 1,170 18% Ubiquitin 72 21 29% 2.6Unique genes, present 6,630 1,170 18% RNA polymerase 23 11 48% 3.8Unique genes, present 6,630 1,170 18% Transcription 141 28 20% 0.7Unique genes, present 6,630 1,170 18% Tubulin 14 8 57% 3.9Unique genes, present, 200 4,277 851 20% Myosin/tropomyosin 23 10 43% 2.8Unique genes, present, 200 4,277 851 20% Actin 57 15 26% 1.2

Pathway analysis and z-score calculation was performed by comparing the frequency of genes upregulated by FN in selected groups vs. the overall frequencyof upregulation in the entire universe of unique genes expressed as described in MATERIALS AND METHODS. Groups were defined by MAPP pathway software,or by searching the RAE230A array annotation database. “FN/C � 1.5” represents the number of genes in the group that were induced by FN by a factor of1.5 or greater. “Present” means that the genes were called “present” by GeneChip software detection algorithm in at least one array.



available at the Physiological Genomics web site).1 In thismouse model of cardiac hypertrophy, gene expression arrayswere done on RNA obtained from hearts at 1 h, 4 h, 24 h, 48 h,1 wk, and 8 wk postaortic banding and compared with sham-treated control animal (8, 19). We were able to match 4,304probe sets from the rat RAE230A with 4,440 probe sets in themouse MG-U74Av2 array, corresponding to 3,316 uniquegenes, of which 1,727 were expressed by NRVM and 452 wereupregulated �1.5-fold in our experiments. Table 4 showsgenes upregulated in vivo that were not significantly changedby FN in our purified NRVM. Interestingly, these includegenes coding for several extracellular matrix and transmem-brane proteins (biglycan, collagens, FN, epithelial membraneprotein 1, fibrillin 1, TIMP1) as well as genes possibly asso-ciated with noncardiac cell lineages such as B-cell transloca-tion gene 2, macrophage migration inhibitory factor, epithelialmembrane antigen, and MARCKS-like protein. These obser-vations highlight one advantage of purified cell culture studiesfor cell-autonomous pathway identification. Many of the genesidentified as upregulated in our studies did not show a changein vivo. One possibility for the discrepancy, in addition tohypertrophic stimulus- and species-specific pathways, is thatrelatively small changes (�3-fold) in mRNA abundance ex-clusive to cardiac myocytes may be undetectable when wholeheart RNA is examined, since cardiac myocytes may compriseless than 50% of the cells in the diseased heart. Our ability topurify NRVM and detect statistically significant changes as

small as 1.5-fold may uncover important pathways missed withwhole organ analyses.

On the other hand, we identified 124 genes significantlyupregulated both in the primary NRVM culture model and invivo (Supplemental Table S1). These include genes involved inapoptosis and cell cycle (8), calcium handling (9), metabolism(20), signal transduction (14), transcription (8), translation(19), and associated with the cytoskeleton or the sarcomeres(20). Many of the genes representing the pathways identifiedon Table 3 were also induced in vivo (Supplemental Table S1).

Protein synthesis. Ribosomal proteins represented the groupaffected the most by FN, in part because of their large number,resulting in the highest z-score (Table 3). Since many of theribosomal proteins were also induced in the mouse aorticbanding model (Supplemental Table S1), these data suggestthat ribosomal protein abundance increases in hypertrophiccardiac myocytes. In addition, we measured significant in-creases in the mRNA abundance of subunits of translationfactors eIF2, eIF3, and eEF1 and in aminoacyl-tRNA syntheta-ses Vars2, Cars, and Sars (Supplemental Table S1). To ourknowledge, this is the first report suggesting the involvementof these translation factors in cardiac hypertrophy. Interest-ingly, elongation factor eEF2 is regulated posttranscriptionallyby phosphorylation (7) and showed unchanged mRNA levelsin our experiments.

Protein degradation. We have identified the proteolyticubiquitin-proteasome pathway as a target for upregulation byFN (Table 3). The z-scores for ubiquitin and proteasome-related genes were 3.4 and 2.6, respectively. These relativelylow z-scores reflect the fact that several ubiquitin and protea-some-related genes were not upregulated, or only slightly

1The Supplementary Material for this article (Supplemental Table S1) isavailable online at http://physiolgenomics.physiology.org/cgi/content/full/00104.2004/DC1.

Table 4. Genes upregulated in in vivo hypertrophy but not in purified NRVM

Symbol Title Avg FN Avg FN/C

Aortic Banding

Base 1 h 4 h 24 h 48 h 1 wk 8 wk

Bgn Biglycan 1,938 0.9 2,713 1.0 0.9 1.3 3.1 2.8 2.7Btg2 B-cell translocation gene 2 362 0.8 1,282 2.7 2.5 2.3 2.1 1.7 1.1Colla1 Procollagen, type I, �1 8,958 1.0 603 1.6 0.9 1.8 4.8 3.6 2.9Colla2 Procollagen, type I, �2 5,779 1.0 879 1.1 0.9 1.1 2.9 3.7 2.2Col3a1 Procollagen, type III, �1 4,482 1.1 641 1.1 0.9 1.9 4.1 3.7 2.1Col5a1 Procollagen, type V, �1 985 1.3 1,383 0.7 0.8 1.2 2.4 2.0 1.3Col5a2 Procollagen, type V, �2 1,849 0.9 227 1.1 0.9 3.0 6.4 7.7 2.9Col6a1 Procollagen, type VI, �1 792 0.6 1,226 1.1 0.9 1.0 1.6 2.1 1.5Col6a2 Procollagen, type VI, �2 332 0.4 905 1.3 0.8 1.0 1.5 2.3 1.5Col8a1 Procollagen, type VII, �1 279 1.1 647 1.0 1.0 2.4 4.4 3.6 1.6Col12a1 Procollagen, type XII, �1 225 0.7 204 0.7 0.7 1.5 1.4 1.7 1.1Col15a1 Procollagen, type XV, �1 406 1.0 598 1.3 0.9 1.1 2.0 2.1 1.3Col18a1 Procollagen, type XVIII, �1 605 0.9 278 1.4 0.7 2.1 1.8 1.4 1.1Emp1 Epithelial membrane protein 1 614 0.9 1,461 1.5 1.9 2.0 2.7 2.3 1.9Fbn1 Fibrillin-1 797 0.7 466 0.7 1.3 2.0 3.6 3.4 1.8Fn1 Fibronectin 1 3,337 0.9 912 0.5 1.1 1.7 2.3 1.8 1.8Hspa1a Heat shock protein 70 a1 206 0.8 472 1.2 6.7 5.2 2.0 1.7 0.7Hspca Heat shock 90-kDa protein 1, alpha 2,996 0.8 1,624 1.7 2.4 2.5 1.4 1.4 1.5Mif Macrophage migration inhibitory factor 517 1.2 215 0.6 1.2 1.6 1.9 3.6 0.9Mlp MARCKS-like protein 1,049 0.9 320 1.1 1.1 2.4 1.9 1.2 1.1Mt2 Metallothionein-II (MT-II) 1,521 0.7 3,463 2.4 2.0 4.4 1.9 2.9 2.4Oazin Ornithine decarboxylase antizyme inhibitor 367 0.8 789 1.5 1.8 2.5 1.6 1.3 1.2Timp1 Tissue inhibitor of metalloproteinase 1 654 1.1 1,468 0.7 0.9 3.8 9.4 2.5 1.0

Selected genes that were upregulated in an aortic banding mouse model of cardiac hypertrophy (8, 19) but were not induced in NRVM stimulated with FNare shown. “Avg FN” indicates average level of expression after normalization in FN-treated NRVM. “Avg FN/C” indicates average ratio of expression inFN-treated vs. control animals. “Base” indicates average level of baseline expression in sham-operated animals. Numbers represent the ratio of gene expressionin aortic-banded vs. sham- operated animals. Values �1.5 are in bold font.



upregulated, by FN. Two of those genes (Uchl1, Usp18) werealso upregulated in vivo (Supplemental Table S1). Uchl1 codesfor a carboxy-terminal ubiquitin esterase involved in removingubiquitin from small adducts, whereas Usp18 is a member ofthe ubiquitin-specific proteases that disassemble polyubiquitinchains into free monomeric ubiquitin. Additional genes in-duced by FN and involved in the ubiquitin-proteasome path-way included ubiquitin proteases (Usp1, Usp2, Usp4, Usp10,Usp15, Usp39, Usp49), ubiquitin precursors (FAU, Uba52,Ubb, Ubc), members of the E2-ligase complex (Ube2d2,Ube2j1, Ube2v1, Ube2v2), E3-ligase and cofactors (Sugt1,Rbx1, Rnf7, Spop, Fbxl6, Fbxo7, Fbxo22, Fbxw5, Tceb2,Cop1), and components of proteasome 19S (Psmc2, Psmc3,Psmc5, Psmd1, Psmd9–11, Psmd13) and 20S (Psma7, Psmb3–6, Psmb8, Psmb10) subunits (results not shown).

Metabolic pathways. Virtually all of the expressed choles-terol biosynthesis genes present in the RAE230A array weresignificantly induced by FN (Fig. 3). A rate-limiting enzyme,the P-450 cytochrome CYP51, responsible for the sterol 14-�demethylase reaction, is coded by one of the genes with thehighest level of induction by FN (Table 1). Several genes in thecholesterol biosynthesis pathway (Cyp51a1, Dhcr7, Hmgcs1,and Sqle) were also induced by aortic banding (SupplementalTable S1). We confirmed the induction of several of thecholesterol biosynthesis genes by RT-PCR (Table 2).

Changes in fatty acid metabolism have been observed incardiac hypertrophy. These usually involve a decrease infatty acid �-oxidation and reversion to the fetal phenotypeof glucose utilization as a major source of energy. We didnot observe any major changes in glucose metabolism orfatty acid oxidation gene expression, although we did see aminor decrease in CPT1b (0.8-fold), the muscle isoform ofcarnitine phosphotransferase and rate-limiting enzyme fortransporting long-chain fatty acid into the mitochondria, aswell as a decrease in long-chain fatty acid-CoA ligasesFacl2 (0.8-fold) and Facl3 (0.6-fold) and fatty acid trans-porter CD36 (0.7-fold). In addition, we observed a mildincrease in the insulin-insensitive glucose transporterSlc2a1 (GLUT1, 1.4-fold), whereas the insulin-regulatedtransporter Slc2a4 (GLUT4) was undetectable. These resultsare consistent with a shift from exogenous fatty acid oxida-tion to glucose metabolism for energy production. In addi-tion to a decrease in fatty acid oxidation, our data suggest apotential increase in fatty acid synthesis. We detected sig-nificant amounts of fatty acid synthase (Fasn) mRNA (nor-malized levels � 1,028) in induced NRVM, and inductionby FN was 2.4-fold. Several other genes in the fatty acidbiosynthesis/elongation/desaturation pathway were alsopresent at high levels and induced by FN (z � 4.2, Table 3and Supplemental Table S1).

Fig. 3. Diagram of cholesterol biosynthesis pathway gene expression changes induced by FN in cardiac myocytes. The diagramwas adapted from files available at the GenMAPP website (http://www.genmapp.org), originally created by Michael Liebermanand Manny Ramirez. The diagram was modified to match the RAE230A probe sets and several new genes and interactions wereintroduced. Genes induced �1.5-fold are shown in orange, genes induced �2.0-fold are shown in red, and genes with expressionchanges between 0.7 and 1.5 are shown in yellow. Genes in gray were not called “present” by the GeneChip software, whereasgenes in white are not represented by any probe sets in the RAE230A array. The red arrows represent genes known to be inducedby the transcription factor SREBP1, whereas the broken red arrow shows the gene (mevalonate kinase) affected by SREBP2 butnot SREBP1.



Induction of cholesterol and fatty acid biosynthesis in othercell types has been linked to activation of the transcriptionfactors SREBP1 and SREBP2. Microarray analysis of mouseliver mRNA expression in SREBP1a and SREBP2 transgenicmice and SREBP cleavage-activating protein (SCAP) knock-out mice revealed 33 genes that are targets of both SREBP1aand SREBP2, 18 genes activated by SREBP1a but notSREBP2, and 10 genes that are targets of SREBP2 only (23).When we applied the pathway analysis to the 38 SREBP1atarget genes represented in the RAE230A array and expressedin NRVM, 32 were induced �1.5-fold by FN (Table 3). Thesegenes include all of the 17 cholesterol biosynthesis pathwaygenes in Fig. 3, plus the LDL receptor, involved in cholesteroluptake, INSIG1, a repressor of SREBP induced by negativefeedback when adequate sterol levels are present in the cell(23), SREBP1 itself, 9 fatty-acid synthesis genes (Me1, Acacb,Acac, Fasn, Acat2, Elovl6, Facl5, Fads1, Fads2), and 3 othergenes (Gstt3, Mac30, Tkt).

Respiratory chain. Induction by FN of several electrontransporters and other genes involved in the mitochondrialrespiratory chain was observed in our experiments (z � 6.4,Table 3). These include genes in the NADH-ubiquinone oxi-doreductase complex I (15 of 25 expressed genes were in-duced), ubiquinol cytochrome c reductase complex III (7 of 11induced), cytochrome c oxidase complex IV (10 of 16 in-duced), and ATP synthase complex V (6 of 16 induced).Complex IV includes Cox17, a protein essential for the assem-bly of functional cytochrome c oxidase (CCO) and for deliveryof copper ions to the mitochondrion, which was induced1.9-fold by FN. Interestingly, other genes involved in regula-tion in copper transport and metabolism were also induced byFN, including Slc31a1, a copper transporter, and Atox1, acopper chaperone (Supplemental Table S1).

Cytoskeletal and sarcomeric proteins. Our statistical analy-sis of pathways affected by FN in cardiac myocytes uncoveredsignificant changes in genes involved in microtubule assemblyand sarcomeric proteins (Table 3). FN induced �-tubulin genesTuba1 and Tuba4, �-tubulin genes Tubb, Tubb2, and Tubb5, aswell as microtubule-associated proteins Mtap1a and Mtap4(Supplemental Table S1) and tubulin- and microtubule-main-tenance chaperones, including T-complex subunits 1, 3, 4, 5,and 7 and tubulin cofactors A, B, D, and E (not shown).

In contrast to microtubule proteins, actins and actin poly-merization genes as a group were not significantly altered byFN (z � 1.2). However, several genes playing important rolesin actin polymerization were upregulated, including Actn1,Cfl1, Cnn2, Coro1c, Cyr61, Enah, Fbln2, and Tagln, drebrin 1,thymosin b4x, filamin C, and Arp2/3 complex members (Sup-plemental Table S1). Interestingly, melusin, a �1-integrinbinding protein linking mechanical stretch and cytoskeletaldynamics in cardiac myocytes (6), was induced 2.1-fold by FN.These results suggest that changes in actin polymerization andcytoskeleton dynamics may result from FN-induced cardiachypertrophy.

Genes coding for sarcomeric proteins were also significantlyinduced by FN (Table 3 and Supplemental Table S1). Theseinclude the fetal isoforms �SkA (Acta1) and cardiac �MHC(Myh7), well known to be upregulated in cardiac hypertrophy,as well as myosin light chains, cardiac myosin-binding proteinC, troponin C and T, and tropomyosin genes (SupplementalTable S1). The tropomyosin genes are regulated by alternative

splicing in a tissue-specific manner (57), and the RAE230Aarray probe sets can distinguish some of the alternativelyspliced isoforms. For example, probe set 1370287_a_at recog-nizes the smooth muscle-specific exon 9d and showed anexpression level of 1,157 and induction of 1.3-fold by FN,whereas probe set 1368724_a_at recognizes the striated mus-cle-specific exon 9b and showed an expression level of 16,414and induction of 2.1-fold by FN in cardiac myocytes. Ourresults suggest that transcriptional upregulation of these sarco-meric genes beyond the general increase in transcription ac-companying hypertrophy underlie not only increased sarco-meric assembly, but also actual shifts in the isoform composi-tion of the sarcomeres.

DISCUSSION

In this study we have confirmed that FN induces a hyper-trophic phenotype in NRVM characterized by increased cellu-lar area, enhanced myofibrillogenesis, and sarcomeric assem-bly, together with gene expression changes, including expres-sion of fetal genes such as ANP, BNP, �SkA, and �MHC.Moreover, we reported for the first time the results of globalgene expression analysis in hypertrophic primary cardiac myo-cytes using high-density oligonucleotide microarrays. Our datauncovered several pathways induced by FN and provide aframework for comparing whole heart gene expression changesassociated with in vivo cardiac hypertrophy with expressionchanges intrinsic to cardiac myocytes. The statistic pathwayanalysis we used allowed us to distinguish noise from trulyinduced genes, even when the induction was as small as1.5-fold. Our data, together with other studies in culture and invivo, confirm that hypertrophy involves global changes inbiosynthesis pathways, including increased transcription,translation, and synthesis of cholesterol and fatty acids, whichare consistent with the significant increases in cell size andcomplexity characteristic of hypertrophy. In addition, hyper-trophy is accompanied by induction of specific genes, such asthose involved in the sarcomeric assembly and the natriureticpeptides, which are upregulated to levels beyond the globalincrease in transcription and translation.

One important feature of hypertrophic cardiac myocytes isincreased protein synthesis, achieved in part by an increase intranslational capacity. The increase in translational capacity issupported by our observation of FN-induced expression ofribosomal protein genes, resulting in the highest z-score of thepathways examined (Table 3). It is known that hypertrophy andintegrin activation increases the translational efficiency ofribosomal protein mRNAs (3, 54). Therefore, it is likely thatthe abundance of ribosomal proteins increases during hyper-trophy due to increased transcription and increased translation.Together with increased transcription of the 45S rRNA precur-sor (2, 36), the resulting increased translational capacity allowsfor enhanced protein synthesis required for hypertrophicgrowth.

In addition to the increase in translational capacity, hyper-trophic cardiac myocytes are characterized by increased trans-lational efficiency. The rate-limiting step of regulation oftranslation efficiency involves the regulation of the activity oftranslation initiation factor eIF4E, which modulates the forma-tion of the eIF4F complex and affects the affinity of mRNA forthe ribosome. We did not detect a change in mRNA abundance



of eIF4E, but this protein is mostly regulated by phosphoryla-tion and association with 4E-binding proteins in cardiac myo-cytes (55). However, we did measure increases in mRNAlevels of eIF2 and eIF3, which are involved in recruitment ofcapped mRNAs to the ribosome, eEF1 subunits �1, �, �, and�, which bind aminoacyl-tRNA complexes (7), and valine,serine, and cysteine-tRNA synthetases, suggesting that synthe-sis of certain aminoacyl-tRNA and their recruitment to theribosome are limiting in hypertrophic cardiac myocytes.

The robust increase in protein synthesis during cardiacmyocyte hypertrophy is accompanied by moderate increases inprotein degradation, resulting in higher protein turnover andnet accumulation of proteins. The ubiquitin-proteasome path-way plays the major role in proteolytic degradation of mis-folded or damaged proteins, turnover of short-lived proteins,and also of sarcomeric proteins (17). We have identified theubiquitin-proteasome pathway as a target for upregulation byFN (Table 3), suggesting that the ubiquitin-proteasome path-way plays a role in protein turnover in hypertrophic cardiacmyocytes and that inhibition of this pathway may modify thehypertrophic phenotype.

Consistent with the growth in cell volume, the hypertrophiccardiac myocytes require increased membrane assembly,which is reflected by a coordinated induction in cholesterolbiosynthesis. Virtually all of the cholesterol biosynthesis genesdetected in the RAE230A array were significantly induced byFN (Fig. 3). This is a novel observation in cardiac myocytesand may be a mechanism by which cholesterol synthesisinhibitors prevent cardiac hypertrophy (41, 44), in addition totheir effect in the isoprenylation of small G proteins, such asRas and RhoA (30).

Although the downregulation of fatty acid oxidation incardiac hypertrophy has been extensively reported, the abilityof cardiac myocytes to synthesize and elongate fatty acids hasbeen poorly studied, since the heart is thought incapable ofsignificant fatty acid synthesis (16, 21). However, the ratio ofC18 to C16 fatty acids in the heart is twice that of plasma (10),purified cardiac myocytes were able to desaturate and elongatelabeled linoleic acid (34), and there is no satisfactory explana-tion for the fate of considerable amounts of malonyl-CoA inthe heart (21). In our NRVM cultures, we have detectedsignificant mRNA expression and induction by FN of fatty acidsynthetase (Fasn) and other genes in the fatty acid biosynthesispathway. Other studies have detected Fasn mRNA in the heartand reported regulation by starvation (28), glucocorticoids(48), and thyroid hormone (4). It is possible that in the perfusedadult heart (16, 21) fatty acid synthesis is undetectable, but incultured NRVM fatty acid synthesis plays a role in hypertro-phic growth.

Since all of the cholesterol biosynthesis genes and nine fattyacid biosynthesis genes induced by FN are also known targetsof transcription factor SREBP1 (23), our data suggest thatSREBP1 may be co-coordinating the induction of these twopathways to maintain the proper balance of cholesterol andphospholipid-associated fatty acids in hypertrophic cardiacmyocytes. We are currently testing the hypothesis that activa-tion of SREBP1 during cardiac hypertrophy is responsible forthe induction of transcription of cholesterol and fatty acidbiosynthesis genes.

Induction of respiratory chain components has not been wellstudied in hypertrophic myocytes. ATP synthase activity is

induced by increased contraction and inotropic agents in car-diac myocytes (13), and ATP synthase subunit c mRNA wasinduced by norepinephrine in the rat heart (32). In our exper-iments we observed induction of components I, III, IV, and Vof the mitochondrial respiratory chain. The highest induction inthe ATP synthase complex F1 was observed in subunit e(2.4-fold). Interestingly, regulation of subunit e by hypoxia atthe pretranslational level has been previously observed incardiac myocytes (31). It is possible that transcriptional induc-tion of several limiting components of the respiratory chain isrequired to meet the increased energy demand of the hypertro-phic myocytes.

Given the recently identified role of calcineurin in hypertro-phy (37), it is interesting that FN did not induce the expressionof any of the isoforms of calcineurin or the calcineurin/NFAT-target gene Dscr1, which can be induced by mechanical stretchor adrenergic stimulation (49, 53). On the other hand, FNinduced calmodulin 1 (Supplemental Table S1), which acti-vates calcineurin in response to elevated intracellular Ca2�

levels. FN also induced the calcineurin binding protein myoze-nin 2 (calsarcin-1), which tethers calcineurin to the Z-line ofsarcomeres (18), calcium/calmodulin-dependent protein kinaseI, which synergizes with calcineurin in promoting hypertrophy(43), and frequenin homolog, which can substitute for calmod-ulin and regulate cardiac Ito K� channels (38). It is possiblethat FN induces a permissive state for calcineurin activationbut is insufficient for the full induction of calcineurin, whichmay require increased intracellular Ca2� levels induced bymechanical stretch or adrenergic agonists.

We have identified tubulins and microtubule-associated pro-teins, such as Mtap4, as targets of FN-induced upregulation.Microtubule density is increased in cardiac hypertrophy andmay interfere with contractility and play a role in heart failureprogression (25, 46, 47, 50). Mtap4 is a microtubule-stabilizingprotein and is thought to be involved in microtubule densifi-cation during cardiac myocyte hypertrophy (46, 50). Interest-ingly, we also observed induction by FN of several chaperonesimplicated in tubulin and microtubule maintenance. Theseresults suggest that both pretranslational and posttranslationalmechanisms are involved in tubulin upregulation and microtu-bule stabilization in hypertrophic cardiac myocytes.

FN induced the connective tissue growth factor (CTGF)gene, coding for a cysteine-rich, glycosylated protein thatregulates proliferation of fibroblast and other mesenchymalcells and induces secretion of extracellular matrix proteins.CTGF is induced by TGF-� in cardiac fibroblasts and myo-cytes and is thought to play a role in heart fibrosis (9). FN alsoupregulated WNT1 inducible signaling pathway protein 2(WISP2), another member of the CTGF family, but its expres-sion in the heart has not been previously reported. Anotherextracellular protein induced by FN, the secreted acidic cys-teine-rich glycoprotein (osteonectin, SPARC), is abundant inremodeling tissues and in diseases associated with fibrosis. Itwas reported to be induced by �-adrenergic stimulation in themyocardium of adult rats (35) and in a transgenic mouse modelof hypertrophy (24). It is tempting to speculate that secretion ofthe cysteine-rich proteins SPARC, CTGF, and WISP2 bycardiac myocytes may be responsible in part for the fibroblastproliferation and fibrosis associated with pathological hyper-trophy and heart failure in vivo and may explain the increasedmRNA levels of various collagen genes, FN, biglycan, and



other extracellular matrix genes, which were not induced byFN in purified NRVM but were upregulated in the mouse aorticbanding model (Table 4).

In summary, our model of FN-induced hypertrophy togetherwith our statistical analysis of gene expression patterns andcomparison with the mouse aortic banding model revealedseveral pathway and gene expression changes not previouslyassociated with cardiac hypertrophy. Our studies also appear todistinguish cardiac myocyte-specific from non-myocyte-de-pendent pathways. Further work is required to determine howcritical these gene expression changes are for a completecardiac hypertrophy phenotype and to evaluate how modula-tion of these changes may affect progression to heart failure.

ACKNOWLEDGMENTS

We thank Dr. Jianhua Luo for help with microarray experiments and Dr.Antonia Sepulveda for helpful comments.

Present address of H. Chen: Department of Pathology, New York Univer-sity, New York, NY 10016.

GRANTS

This work was supported by the National Heart Foundation and by NationalHeart, Lung, and Blood Institute Grant HL-68714-03 (to J. L. Sepulveda).

REFERENCES

1. Aikawa R, Nagai T, Kudoh S, Zou Y, Tanaka M, Tamura M,Akazawa H, Takano H, Nagai R, and Komuro I. Integrins play a criticalrole in mechanical stress-induced p38 MAPK activation. Hypertension 39:233–238., 2002.

2. Allo SN, McDermott PJ, Carl LL, and Morgan HE. Phorbol esterstimulation of protein kinase C activity and ribosomal DNA transcription.Role in hypertrophic growth of cultured cardiomyocytes. J Biol Chem 266:22003–22009, 1991.

3. Balasubramanian S and Kuppuswamy D. RGD-containing peptidesactivate S6K1 through beta 3 integrin in adult cardiac muscle cells. J BiolChem 278: 42214–42224, 2003.

4. Blennemann B, Leahy P, Kim TS, and Freake HC. Tissue-specificregulation of lipogenic mRNAs by thyroid hormone. Mol Cell Endocrinol110: 1–8, 1995.

5. Boluyt MO, Long X, Eschenhagen T, Mende U, Schmitz W, CrowMT, and Lakatta EG. Isoproterenol infusion induces alterations inexpression of hypertrophy-associated genes in rat heart. Am J PhysiolHeart Circ Physiol 269: H638–H647, 1995.

6. Brancaccio M, Fratta L, Notte A, Hirsch E, Poulet R, Guazzone S, DeAcetis M, Vecchione C, Marino G, Altruda F, Silengo L, Tarone G,and Lembo G. Melusin, a muscle-specific integrin beta1-interactingprotein, is required to prevent cardiac failure in response to chronicpressure overload. Nat Med 9: 68–75, 2003.

7. Browne GJ and Proud CG. Regulation of peptide-chain elongation inmammalian cells. Eur J Biochem 269: 5360–5368, 2002.

8. Chen D, Izumo S, Kohane I, Kong SW, Nie Q, Riggi L, Schinke M,and Shioi T. Microarray-Based Analysis of Gene Expression DuringPressure-Overload Induced Cardiac Hypertrophy. 67th Cold Spring Har-bor Symposium on Quantitative Biology: The Cardiovascular System,Cold Spring Harbor, 2002.

9. Chen MM, Lam A, Abraham JA, Schreiner GF, and Joly AH. CTGFexpression is induced by TGF-beta in cardiac fibroblasts and cardiacmyocytes: a potential role in heart fibrosis. J Mol Cell Cardiol 32:1805–1819, 2000.

10. Cinti DL, Cook L, Nagi MN, and Suneja SK. The fatty acid chainelongation system of mammalian endoplasmic reticulum. Prog Lipid Res31: 1–51, 1992.

11. Craig R, Wagner M, McCardle T, Craig AG, and Glembotski CC. Thecytoprotective effects of the glycoprotein 130 receptor-coupled cytokine,cardiotrophin-1, require activation of NF-kappa B. J Biol Chem 276:37621–37629, 2001.

12. Dahlquist KD, Salomonis N, Vranizan K, Lawlor SC, and ConklinBR. GenMAPP, a new tool for viewing and analyzing microarray data onbiological pathways. Nat Genet 31: 19–20, 2002.

13. Das AM. Regulation of the mitochondrial ATP-synthase in health anddisease. Mol Genet Metab 79: 71–82, 2003.

14. Deng XF, Rokosh DG, and Simpson PC. Autonomous and growthfactor-induced hypertrophy in cultured neonatal mouse cardiac myocytes:comparison with rat. Circ Res 87: 781–788, 2000.

15. Doniger SW, Salomonis N, Dahlquist KD, Vranizan K, Lawlor SC,and Conklin BR. MAPPFinder: using Gene Ontology and GenMAPP tocreate a global gene-expression profile from microarray data. Genome Biol4: R7, 2003.

16. Dyck JR, Kudo N, Barr AJ, Davies SP, Hardie DG, and LopaschukGD. Phosphorylation control of cardiac acetyl-CoA carboxylase bycAMP-dependent protein kinase and 5-AMP activated protein kinase. EurJ Biochem 262: 184–190, 1999.

17. Eble DM, Spragia ML, Ferguson AG, and Samarel AM. Sarcomericmyosin heavy chain is degraded by the proteasome. Cell Tissue Res 296:541–548, 1999.

18. Frey N, Richardson JA, and Olson EN. Calsarcins, a novel family ofsarcomeric calcineurin-binding proteins. Proc Natl Acad Sci USA97:14632–14637, 2000.

19. Genomics of Cardiovascular Development Adaptation, and Remodel-ing. Pressure overload-induced cardiac hypertrophy. [Online]. NHLBIProgram for Genomic Applications, Harvard Medical School. http://cardiogenomics.med.harvard.edu/groups/proj1/pages/band_home.html#bandexp [accessed 2/2004].

20. Grimm D, Kromer EP, Bocker W, Bruckschlegel G, Holmer SR,Riegger GA, and Schunkert H. Regulation of extracellular matrixproteins in pressure-overload cardiac hypertrophy: effects of angiotensinconverting enzyme inhibition. J Hypertens 16: 1345–1355, 1998.

21. Hamilton C and Saggerson ED. Malonyl-CoA metabolism in cardiacmyocytes. Biochem J 350: 61–67, 2000.

22. Heling A, Zimmermann R, Kostin S, Maeno Y, Hein S, Devaux B,Bauer E, Klovekorn WP, Schlepper M, Schaper W, and Schaper J.Increased expression of cytoskeletal, linkage, and extracellular proteins infailing human myocardium. Circ Res 86: 846–853, 2000.

23. Horton JD, Shah NA, Warrington JA, Anderson NN, Park SW,Brown MS, and Goldstein JL. Combined analysis of oligonucleotidemicroarray data from transgenic and knockout mice identifies directSREBP target genes. Proc Natl Acad Sci USA 100: 12027–12032, 2003.

24. Ihara Y, Suzuki YJ, Kitta K, Jones LR, and Ikeda T. Modulation ofgene expression in transgenic mouse hearts overexpressing calsequestrin.Cell Calcium 32: 21–29, 2002.

25. Ishibashi Y, Takahashi M, Isomatsu Y, Qiao F, Iijima Y, Shiraishi H,Simsic JM, Baicu CF, Robbins J, Zile MR, and Cooper G IV. Role ofmicrotubules versus myosin heavy chain isoforms in contractile dysfunc-tion of hypertrophied murine cardiocytes. Am J Physiol Heart Circ Physiol285: H1270–H1285, 2003. First published May 15, 2003; 10.1152/ajpheart.00654.2002.

26. Kawano H, Do YS, Kawano Y, Starnes V, Barr M, Law RE, andHsueh WA. Angiotensin II has multiple profibrotic effects in humancardiac fibroblasts. Circulation 101: 1130–1137, 2000.

27. Kent WJ. BLAT: the BLAST-like alignment tool. Genome Res 12:656–664, 2002.

28. Kim TS and Freake HC. High carbohydrate diet and starvation regulatelipogenic mRNA in rats in a tissue-specific manner. J Nutr 126: 611–617,1996.

29. Laser M, Willey CD, Jiang W, Cooper G VI, Menick DR, Zile MR,and Kuppuswamy D. Integrin activation and focal complex formation incardiac hypertrophy. J Biol Chem 275: 35624–35630, 2000.

30. Laufs U, Marra D, Node K, and Liao JK. 3-Hydroxy-3-methylglutaryl-CoA reductase inhibitors attenuate vascular smooth muscle proliferationby preventing rho GTPase-induced down-regulation of p27(Kip1). J BiolChem 274: 21926–21931, 1999.

31. Levy FH and Kelly DP. Regulation of ATP synthase subunit e geneexpression by hypoxia: cell differentiation stage-specific control. Am JPhysiol Cell Physiol 272: C457–C465, 1997.

32. Li P, Li J, Feng X, Li Z, Hou R, Han C, and Zhang Y. Gene expressionprofile of cardiomyocytes in hypertrophic heart induced by continuousnorepinephrine infusion in the rats. Cell Mol Life Sci 60: 2200–2209,2003.

33. Liang F, Atakilit A, and Gardner DG. Integrin dependence of brainnatriuretic peptide gene promoter activation by mechanical strain. J BiolChem 275: 20355–20360, 2000.

34. Lopez Jimenez JA, Bordoni A, Lorenzini A, Rossi CA, Biagi PL, andHrelia S. Linoleic acid metabolism in primary cultures of adult rat



cardiomyocytes is impaired by aging. Biochem Biophys Res Commun 237:142–145, 1997.

35. Masson S, Arosio B, Luvara G, Gagliano N, Fiordaliso F, Santam-brogio D, Vergani C, Latini R, and Annoni G. Remodelling of cardiacextracellular matrix during beta-adrenergic stimulation: upregulation ofSPARC in the myocardium of adult rats. J Mol Cell Cardiol 30: 1505–1514, 1998.

36. McDermott PJ, Rothblum LI, Smith SD, and Morgan HE. Acceleratedrates of ribosomal RNA synthesis during growth of contracting heart cellsin culture. J Biol Chem 264: 18220–18227, 1989.

37. Molkentin JD, Lu JR, Antos CL, Markham B, Richardson J, RobbinsJ, Grant SR, and Olson EN. A calcineurin-dependent transcriptionalpathway for cardiac hypertrophy. Cell 93: 215–228, 1998.

38. Nakamura TY, Sturm E, Pountney DJ, Orenzoff B, Artman M, andCoetzee WA. Developmental expression of NCS-1 (frequenin), a regula-tor of Kv4 K� channels, in mouse heart. Pediatr Res 53: 554–557, 2003.

39. Ogawa E, Saito Y, Harada M, Kamitani S, Kuwahara K, Miyamoto Y,Ishikawa M, Hamanaka I, Kajiyama N, Takahashi N, Nakagawa O,Masuda I, Kishimoto I, and Nakao K. Outside-in signalling of fibronec-tin stimulates cardiomyocyte hypertrophy in cultured neonatal rat ventric-ular myocytes. J Mol Cell Cardiol 32: 765–776., 2000.

40. Ogawa E, Saito Y, Kuwahara K, Harada M, Miyamoto Y, HamanakaI, Kajiyama N, Takahashi N, Izumi T, Kawakami R, Kishimoto I,Naruse Y, Mori N, and Nakao K. Fibronectin signaling stimulates BNPgene transcription by inhibiting neuron-restrictive silencer element-depen-dent repression. Cardiovasc Res 53: 451–459, 2002.

41. Oi S, Haneda T, Osaki J, Kashiwagi Y, Nakamura Y, Kawabe J, andKikuchi K. Lovastatin prevents angiotensin II-induced cardiac hypertro-phy in cultured neonatal rat heart cells. Eur J Pharmacol 376: 139–148,1999.

42. Pardo Mindan FJ and Panizo A. Alterations in the extracellular matrixof the myocardium in essential hypertension. Eur Heart J 14, Suppl J:12–14, 1993.

43. Passier R, Zeng H, Frey N, Naya FJ, Nicol RL, McKinsey TA,Overbeek P, Richardson JA, Grant SR, and Olson EN. CaM kinasesignaling induces cardiac hypertrophy and activates the MEF2 transcrip-tion factor in vivo. J Clin Invest 105: 1395–1406, 2000.

44. Patel R, Nagueh SF, Tsybouleva N, Abdellatif M, Lutucuta S, KopelenHA, Quinones MA, Zoghbi WA, Entman ML, Roberts R, and MarianAJ. Simvastatin induces regression of cardiac hypertrophy and fibrosisand improves cardiac function in a transgenic rabbit model of humanhypertrophic cardiomyopathy. Circulation 104: 317–324, 2001.

45. Rothermund L, Kreutz R, Kossmehl P, Fredersdorf S, Shakibaei M,Schulze-Tanzil G, Paul M, and Grimm D. Early onset of chondroitinsulfate and osteopontin expression in angiotensin II-dependent left ven-tricular hypertrophy. Am J Hypertens 15: 644–652, 2002.

46. Sato H, Nagai T, Kuppuswamy D, Narishige T, Koide M, Menick DR,and Cooper G IV. Microtubule stabilization in pressure overload cardiachypertrophy. J Cell Biol 139: 963–973, 1997.

47. Scopacasa BS, Teixeira VP, and Franchini KG. Colchicine attenuatesleft ventricular hypertrophy but preserves cardiac function of aortic-constricted rats. J Appl Physiol 94: 1627–1633, 2003. First publishedDecember 13, 2002; 10.1152/japplphysiol.00744.2002.

48. Soncini M, Yet SF, Moon Y, Chun JY, and Sul HS. Hormonal andnutritional control of the fatty acid synthase promoter in transgenic mice.J Biol Chem 270: 30339–30343, 1995.

49. Taigen T, De Windt LJ, Lim HW, and Molkentin JD. Targetedinhibition of calcineurin prevents agonist-induced cardiomyocyte hyper-trophy. Proc Natl Acad Sci USA 97: 1196–1201, 2000.

50. Takahashi M, Shiraishi H, Ishibashi Y, Blade KL, McDermott PJ,Menick DR, Kuppuswamy D, and Cooper G IV. Phenotypic conse-quences of beta1-tubulin expression and MAP4 decoration of microtu-bules in adult cardiocytes. Am J Physiol Heart Circ Physiol 285: H2072–H2083, 2003. First published July 10, 2003; 10.1152/ajpheart.00396.2003.

51. Taylor JM, Rovin JD, and Parsons JT. A role for focal adhesion kinasein phenylephrine-induced hypertrophy of rat ventricular cardiomyocytes.J Biol Chem 275: 19250–19257, 2000.

52. Tsutsumi Y, Matsubara H, Ohkubo N, Mori Y, Nozawa Y, MurasawaS, Kijima K, Maruyama K, Masaki H, Moriguchi Y, Shibasaki Y,Kamihata H, Inada M, and Iwasaka T. Angiotensin II type 2 receptoris upregulated in human heart with interstitial fibrosis, and cardiac fibro-blasts are the major cell type for its expression. Circ Res 83: 1035–1046,1998.

53. Vega RB, Rothermel BA, Weinheimer CJ, Kovacs A, Naseem RH,Bassel-Duby R, Williams RS, and Olson EN. Dual roles of modulatorycalcineurin-interacting protein 1 in cardiac hypertrophy. Proc Natl AcadSci USA 100: 669–674, 2003.

54. Volarevic S and Thomas G. Role of S6 phosphorylation and S6 kinase incell growth. Prog Nucleic Acid Res Mol Biol 65: 101–127, 2001.

55. Wang L, Wang X, and Proud CG. Activation of mRNA translation in ratcardiac myocytes by insulin involves multiple rapamycin-sensitive steps.Am J Physiol Heart Circ Physiol 278: H1056–H1068, 2000.

56. Wei L, Wang L, Carson JA, Agan JE, Imanaka-Yoshida K, andSchwartz RJ. beta1 integrin and organized actin filaments facilitatecardiomyocyte- specific RhoA-dependent activation of the skeletal alpha-actin promoter. FASEB J 15: 785–796., 2001.

57. Wieczorek DF, Smith CW, and Nadal-Ginard B. The rat alpha-tropo-myosin gene generates a minimum of six different mRNAs coding forstriated, smooth, and nonmuscle isoforms by alternative splicing. Mol CellBiol 8: 679–694, 1988.



compared with the apex during chronic pulmonary embolism in rats Transcriptional changes in right...

Documents

Transcript of compared with the apex during chronic pulmonary embolism in rats Transcriptional changes in right...