Three-dimensional matrix primes mesangial cells todown-regulation of a-smooth muscle actin via deactivation ofCArG box elements

MASANORI KITAMURA and YOSHIHISA ISHIKAWA

Glomerular Bioengineering Unit, Department of Medicine, University College London Medical School, London, England, UnitedKingdom

Three-dimensional matrix primes mesangial cells to down-regulation ofa-smooth muscle actin via deactivation of CArG box elements. Prolongedculture of mesangial cells forms multifocal nodular structures, termed“hillocks,” composed of cells and extracellular matrix (ECM), which maymimic the situation in the glomerular mesangium. Mesangial cells incor-porated in hillocks show repressed expression of a-smooth muscle actin, amarker of mesangial cell activation/dedifferentiation. The aim of thisstudy is to elucidate molecular mechanisms involved in this phenomenon,focusing on the activity of CArG box elements located in 59-flankingregion of the a-smooth muscle actin gene. Reporter mesangial cells werecreated to monitor the activity of CArG elements. These clones expressedb-galactosidase gene (lacZ) under the control of CArG boxes. Within thehillocks, reporter cells showed repressed expression of lacZ as well asa-smooth muscle actin compared to the cells in two-dimensional cultures.Consistent with this result, the reporter cells embedded in collagen gelexhibited down-regulation of lacZ and a-smooth muscle actin transcripts.Deactivation of CArG box elements by transfection with either a domi-nant negative mutant of serum response factor or a dominant negativeform of ternary complex factor Elk-1 led to depressed expression ofa-smooth muscle actin gene. These data suggested that three-dimensionalECM primes mesangial cells to down-regulation of a-smooth muscle actinvia deactivation of CArG box elements.

In the renal glomerulus, mesangial cells are surrounded by athree-dimensional extracellular matrix (ECM), the mesangialmatrix, consisting of basement membrane-type collagens, glyco-proteins and proteoglycans. Since ECM has profound biologicaleffects on a variety of cell types, mesangial matrix possiblyregulates the behavior and function of mesangial cells in thenormal glomerular microenvironment. Using gel matrices recon-stituted in vitro, we previously reported that a three-dimensionalbasement membrane gel inhibited migration and proliferation ofcultured mesangial cells [1]. Gene transfection studies showedthat mesangial cells with aberrant matrix-degrading activity exhib-ited accelerated mitogenesis and migration in vitro [2]. This line of

evidence implied that the mesangial matrix controls the behaviorof mesangial cells to maintain a normal cell phenotype in theglomerulus.

This hypothesis has been further supported by recent findings[3]. Prolonged culture of mesangial cells forms nodular structures,termed “hillocks,” composed of cells and surrounding ECM,which may mimic the situation in the glomerular mesangium [4].Incorporation of mesangial cells into this “natural” matrix struc-ture allowed for transition of the cellular phenotype towarddeactivation and differentiation [3]. Compared to hillock-unasso-ciated cells, the cells within hillocks showed repressed mitogenicactivity, elevated type IV/I collagen mRNA ratio and suppressedexpression of a-smooth muscle actin, which is similar to thebehavior of normal mesangial cells in vivo [5].

a-Smooth muscle actin is known to be a crucial marker foractivation and dedifferentiation of mesangial cells [5]. In thenormal glomerulus, expression of a-smooth muscle actin is absentor only faintly detectable, whereas it is markedly up-regulated inthe mesangium during a wide range of experimental and humanglomerular diseases [5–9]. The expression of a-smooth muscleactin is reversible and correlated with injury of the mesangiumand/or activation of mesangial cells [6, 8–11]. However, currently,information is limited regarding regulation of the a-smoothmuscle actin expression in pathophysiologic circumstances.

The 59-flanking region of the a-smooth muscle actin genecontains CC(A/T)6GG sequences (CArG box elements) [12–14],the binding sites of serum response factor (SRF) and the ternarycomplex factor (TCF) subfamily of Ets transcription factors [15].The CArG box motif, the core consensus sequence of serumresponse element (SRE), is crucial for serum-mediated inductionof a-smooth muscle actin in several cell types [12–14]. In ratmesangial cells, two CArG box motifs in the core promoterfragment are necessary and sufficient for the serum induction ofa-smooth muscle actin [16]. Based on these data together with ourprevious findings, ECMs might affect a-smooth muscle actinexpression via the down-regulation of CArG box elements. Usingthe nodular cultures and cultures in gel matrices, the present studyaims at examining this possibility. This report demonstrates thatthree-dimensional ECMs transduce a signal that deactivatesCArG box elements and, thereby, attenuate the expression ofa-smooth muscle actin in mesangial cells.

Key words: mesangial cell; a-smooth muscle actin; CArG box element;serum response element; extracellular matrix.

Received for publication April 30, 1997and in revised form October 2, 1997Accepted for publication October 7, 1997

© 1998 by the International Society of Nephrology

Kidney International, Vol. 53 (1998), pp. 690–697

690

METHODS

Mesangial cells and transfectants

Mesangial cells were cultured from isolated glomeruli of a maleSprague-Dawley rat [17]. Cell clones SM41, SM42 and SM43 wereestablished by a limiting dilution method and identified as beingof mesangial cell phenotype as described [1]. Cells (passages 10 to30) were maintained in Dulbecco’s modified Eagle’s medium/Ham’s F-12 (DME-F12; GIBCO BRL, Gaithersburg, MD, USA)supplemented with 100 U/ml penicillin G, 100 mg/ml streptomy-cin, 0.25 mg/ml of amphotericin B and 10 to 20% fetal calf serum(FCS; GIBCO BRL).

Using a modified calcium phosphate co-precipitation method[18], SM43 mesangial cells were stably transfected with an expres-sion plasmid pSRE-lacZ [19]. This construct introduces (i) abacterial b-galactosidase gene (lacZ) under the control of threecopies of human c-fos SRE adjacent to a minimal promoterderived from Rous sarcoma virus, and (ii) a neomycin phospho-transferase gene (neo) driven by a thymidine kinase promoter.Stable transfectants were selected in the presence of neomycinanalog G418 (750 mg/ml; Sigma Immunochemicals, St. Louis,MO, USA), and reporter cells MCAGZ1 and MCAGZ5 wereestablished. Modifiable expression of the transgene in response tofetal calf serum (FCS) was confirmed using 5-bromo-4-chloro-3-indolyl b-D-galactopyranoside (X-gal) assay [20, 21]. MLTRZ mesan-gial cells that express lacZ under the control of a Moloney murineleukemia virus long terminal repeat (LTR) [17] were utilized ascontrol cells. Other control cells of MACTZ that express lacZunder the control of a b-actin promoter were established bytransfection of SM43 cells with pHbAPr-1-b-gal [22] and pRc/CMV (Invitrogen, San Diego, CA, USA) that introduces neo.

To investigate the role of CArG box elements in the regulationof a-smooth muscle actin gene, SM43 cells were stably transfectedwith pRc/CMV and pCGNSRFpm1 (a gift from Dr. Ron Prywes[23]) coding for a mutant serum response factor (SRF), SRFpm1.Expression of transgenes in the established transfectants,MmuSRF1 and MmuSRF2, was confirmed by Northern blotanalysis. Mock transfectants were created by transfection withpRc/CMV alone. SRFpm1 has three point mutations in the basicregion that abolish DNA binding [23]. Since SRF binds to SRE asa dimer, overexpression of SRFpm1 may act in a transdominantnegative fashion. Indeed, SRFpm1 can heterodimerize with en-dogenous wild type SRF, form dimers with low DNA bindingactivity and thereby attenuate SRE activity in muscle cells (Dr.Ron Prywes, personal communication).

Transient transfection

Activity of CArG box elements in MmuSRFs was studied by atransient transfection assay as described before [24]. In brief,using the calcium phosphate coprecipitation method, cells in24-well plates (2 3 105/well) were transiently transfected withpSRE-LacZ (0.5 mg/well). After 48 hours, cells were subjected toX-gal assay to detect b-galactosidase activity. The activity ofCArG box elements was assessed by counting blue cells stained byX-gal. To exclude possible variability in transfection efficiencybetween different cell clones, the number of X-gal-positive cells ineach was normalized by the number of positive cells transfectedwith a control plasmid, pCXL [25] or pCI-bgal (a gift fromPromega, Madison, WI, USA). These expression plasmids intro-duce lacZ under the control of constitutively active viral regula-

tory elements, that is, the spleen necrosis virus LTR or theimmediate-early enhancer/promoter of human cytomegalovirus,respectively.

The ternary complex factor Elk-1 is required for activation ofCArG box elements via mitogen-activated protein kinase cascades[26]. To confirm the crucial role of CArG box elements in theregulation of a-smooth muscle actin gene, transient transfectionstudies were performed using a dominant negative mutant ofElk-1 [27]. Elk-1 forms complexes with SRF, and binding ofElk-1-SRF complexes to CArG motifs leads to induction of targetgenes. If a dominant-interfering form of Elk-1 is overexpressed,transacting ability of Elk-1-SRF complexes is abrogated [26]. Toassess activity of the a-smooth muscle actin promoter, an expres-sion plasmid paAP126 [28] was utilized. This reporter plasmidintroduces lacZ under the control of 2894 to 112 base pairs ofthe human a-smooth muscle actin gene. SM43 cells in 24-wellplates were transiently transfected with pSRE-LacZ, paAP126 orpCI-bgal (0.5 mg/well) together with an empty plasmid or pDN-Elk (2.5 mg/well) encoding a mutated Elk-1 with deletion of thecarboxy-terminal activation domain [26]. After 48 hours, cellswere subjected to X-gal assay to evaluate the activity of thea-smooth muscle actin promoter or CArG box elements. Thenumber of X-gal-positive cells transfected with pSRE-LacZ orpaAP126 was counted and normalized by the number of X-gal-positive cells transfected with pCI-bgal. All assays were per-formed in quadruplicate in the presence of 10% FCS.

Activity of Elk-1 was assessed in hillock-associated cells andunassociated cells using transfection with pGAL4-Elk andpUASg-LacZ. pGAL4-Elk encodes the Elk-1 activation domain(residues 307 to 428) fused to the GAL4 DNA-binding domain[26]. pUASg-LacZ contains lacZ downstream of a promoter(E1b) linked to five copies of the GAL4-binding site [29].Superconfluent SM43 cells (4 3 105/well) in 24-well plates weretransiently transfected with pUASg-LacZ (1 mg/well) or a controlplasmid pCIb-gal (1 mg/well), together with pGAL4-Elk (1 mg/well), under a low serum concentration (0.5% FCS). Aftercompleting the transfection, hillock formation was triggered byexposure of the cells to 20% FCS. After four days, cells weresubjected to X-gal assay to evaluate the activity of Elk-1. Thenumber of X-gal-positive cells transfected with pGAL4-Elk andpUASg-LacZ was counted and normalized by the number ofX-gal-positive cells transfected with pGAL4-Elk and pCI-bgal.These values were compared between hillock-associated cells andunassociated cells. Assays were performed in quadruplicate.

X-gal assay

X-gal assay was performed as described before [17]. In brief,transfectants were fixed in 0.5% glutaraldehyde, 2 mM MgCl2, and1.25 mM EGTA in phosphate buffered saline (PBS) for 10 minutesat room temperature, and incubated at 37°C for two to six hoursin X-gal solution containing 1 mg/ml X-gal (Sigma), 20 mM

K3Fe(CN)6, 20 mM K4Fe(CN)6 z 3H2O, 2 mM MgCl2, 0.01%sodium desoxycholate and 0.02% Nonidet P-40 in phosphatebuffered saline (PBS; pH 7.4).

Development of hillocks

Mesangial cells were cultured in 100 mm plastic plates withoutpassage for one to two weeks for the development of hillocks [3].Hillocks were isolated mechanically by a jet of cold PBS until the

Kitamura and Ishikawa: Deactivation of CArG elements by ECM 691

majority of the hillocks were removed from the plates. Isolatedhillocks were washed three times by cold PBS. After the removalof hillocks, the remaining cells were harvested using a rubberpoliceman. Isolated hillocks and hillock-unassociated cells fromidentical plates were used for Northern blot analyses. To testreversibility of altered gene expression, purified hillocks weretrypsinized and re-seeded into six-well culture plates. After sev-eral days, confluent cells were subjected to Northern analyses.

Stability of mRNAs in hillock-associated cells and unassociatedcells was examined as follows. After the development of hillocks,cultures were treated with the RNA synthesis inhibitor, actinomy-cin D (500 ng/ml; Serva, Heidelberg, Germany). Before or afterthe treatment for 12 and 24 hours, hillocks and hillock-unassoci-ated cells were harvested separately and used for Northernanalyses of lacZ and a-smooth muscle actin expression.

Collagen gel culture

MCAGZ5 cells were incorporated in collagen gels as describedpreviously [1, 30]. In brief, type I collagen solution (1.2 ml)prepared from tails of rats [30] were mixed on ice with (i) 53concentrated DME [NaHCO3 (2); GIBCO, 0.4 ml], (ii) recon-struction buffer (0.14 N NaOH 2 0.26 N NaHCO3; 0.2 ml), and iii)cell suspension (MCAGZ5; 1 3 106 cells/0.2 ml). The mixture waspoured into wells of six-well culture plates, left for 30 minutes at37°C to allow for gel formation and fed with 2 ml of DME-F12containing 20% FCS. After 24 hours of incubation, the collagengel was released from the bottom of each well. In this situation,substantial number of cells populated both on the basementplastic and within the floating gel. After incubating for anadditional 24 hours in fresh 10% FCS/DME-F12, cells on plasticand within the gels were harvested separately and subjected toNorthern blot analyses. To examine whether the three-dimen-sional structure is required for the action of ECM on CArG boxelements, MCAGZ5 cells were seeded on plastic pre-coated with orwithout type I collagen, cultured for 48 hours and used forNorthern analyses.

Northern blot analysis

Northern blot analysis was performed as described previously[31]. Total RNA was extracted by a single-step method, electro-phoresed on 1.2% agarose gels containing 10% formaldehyde andtransferred onto nitrocellulose membranes. As probes, rat a-vas-cular smooth muscle actin cDNA [32], b-galactosidase cDNA[33], SRF cDNA [23] and rat glyceraldehyde-3-phosphate dehy-drogenase (GAPDH) cDNA were labeled with 32P-dCTP using arandom priming method. The membranes were hybridized withprobes at 65°C overnight in a solution containing 4 3 SSC (600mM sodium chloride, 60 mM sodium citrate), 53 Denhardt’ssolution, 10% dextran sulfate, 50 mg/ml herring sperm DNA and50 mg/ml poly(A). The hybridized membranes were washed fourtimes in 4 3 SSC/0.1% SDS at 50°C and exposed to Kodak XARfilms with intensifying screens at 280°C. Densitometric analyseswere performed using a computerized program, NIH Image.

Statistical analysis

Data were expressed as means 6 SE. Statistical analysis wasperformed using the non-parametric Mann-Whitney test to com-pare data in different groups. P value of ,0.05 was used toindicate a statistically significant difference.

RESULTS

Expression of a-smooth muscle actin in hillock-associated cellsand unassociated cells

Mesangial cell clones, SM41, SM42 and SM43 were establishedfrom isolated glomeruli of a Sprague-Dawley rat and used fordevelopment of hillocks. Hillocks and hillock-unassociated cellswere isolated from each culture and subjected to Northern blotanalysis of a-smooth muscle actin expression. In all clones,expression of the transcript (1.5 kb) was significantly repressed inhillock-associated cells compared to unassociated cells (46.5 64.0% versus unassociated cells; 100%, mean 6 SE; Fig. 1). This isconsistent with our early report using uncloned mesangial cellsderived from a different rat strain [3].

Correlation between CArG box activity and a-smooth muscleactin expression

The CArG box motif is known to be the important regulatoryelement located in the 59-flanking region of the a-smooth muscleactin gene [16]. Using the established reporter clones MCAGZ1and MCAGZ5, correlation between CArG box activity anda-smooth muscle actin expression was investigated in nodularcultures (Fig. 2A). Northern blot analysis revealed that, in thepresence of 10% FCS, CArG box elements were constitutivelyactive in hillock-unassociated cells. This was correlated with theirsubstantial expression of the a-smooth muscle actin gene. In bothreporter clones, expression of lacZ was repressed in hillock-associated cells compared to unassociated cells. This was closelycorrelated with downregulation of the a-smooth muscle actinmRNA in hillock-associated cells. In control clones MLTRZ andMACTZ that express lacZ under the control of viral LTR or ab-actin promoter, expression of a-smooth muscle actin was simi-larly repressed in hillock-associated cells, but the expression oflacZ was unaltered.

To investigate reversibility of the suppressed activity of CArGelements in hillock-associated cells, hillocks were trypsinized,dispersed and re-seeded on a plastic substratum. Within severaldays, the dissociated hillocks regained higher levels of CArG box

Fig. 1. Expression of a-smooth muscle actin mRNA in hillock-associatedmesangial cells and unassociated cells. Hillock-unassociated cells (U) andassociated cells (A) were harvested from an identical culture of ratmesangial cell clones SM41, SM42 and SM43. Northern blot analysis wasperformed on a-smooth muscle actin (a-SMA) expression. Expression ofglyceraldehyde-3-phosphate dehydrogenase (GAPDH) and staining of28S and 18S ribosomal RNAs with ethidium bromide are shown as loadingcontrols.

Kitamura and Ishikawa: Deactivation of CArG elements by ECM692

activity and a-smooth muscle actin expression, both of which aretypical of hillock-unassociated cells (Fig. 2B).

Stability of mRNAs in hillock-associated cells and unassociatedcells was examined using an inhibitor of RNA synthesis. After thedevelopment of hillocks, cultures were treated with actinomycinD. Before and after the treatment for 12 and 24 hours, hillocksand hillock-unassociated cells were harvested separately fromidentical plates, and degradation rates of mRNAs were examinedby Northern blot analyses. As demonstrated in Figure 2C, stabilityof lacZ and a-smooth muscle actin transcripts in hillock-associ-ated cells was identical to that in hillock-unassociated cells.

Activity of Elk-1 in hillock-associated cells and unassociatedcells

The activity of CArG box motif is regulated by ternary complexfactors that form complexes with SRF [15]. To further investigatemechanism involved in the suppression of CArG box elements inhillocks, activity of Elk-1, a member of the TCF family, wascompared between hillock-associated cells and unassociated cells.Confluent mesangial cells were transiently transfected withpUASg-LacZ or a control plasmid pCIb-gal, together withpGAL4-Elk. After completing the transfection, hillocks wereinduced by exposure of the cells to 20% FCS. X-gal assay revealed

that hillock-unassociated cells exhibited substantial activity ofElk-1. This constitutive activity was significantly suppressed in thecells incorporated in hillocks (Fig. 3).

Depression of a-smooth muscle actin in mesangial cells bydeactivation of CArG box elements

To examine whether inactivation of CArG elements leads tosuppressed expression of a-smooth muscle actin gene, mesangialcells were stably transfected with a gene encoding a transdomi-nant negative mutant of SRF, the essential component for CArGbox activation [34]. Northern blot analysis revealed that mesangialcells constitutively expressed the endogenous SRF transcript.Established transfectants MmuSRF1 and MmuSRF2 exhibited anadditional transgene transcript (Fig. 4A). Transient transfectionassays revealed that, compared to mock transfected cells,MmuSRF cells showed attenuated activity of CArG box elements[60.0 6 2.9% versus control (100%); Fig. 4B]. Using these clones,expression of a-smooth muscle actin gene was investigated intwo-dimensional cultures. As shown in Figure 4C, the expressionof a-smooth muscle actin was downregulated in both MmuSRF1cells (31%) and MmuSRF2 cells (49%) compared to controls(100%).

To further confirm that the inactivation of CArG box elements

Fig. 2. Activity of CArG box elements in hillock-associated cells and unassociated cells. SM43mesangial cells were stably transfected withpSRE-LacZ that introduces a bacterial b-galac-tosidase gene (lacZ) under the control of threecopies of c-fos CArG box elements, and reportercells MCAGZ1 and MCAGZ5 were established. Ascontrol clones, MLTRZ and MACTZ mesangialcells were used. MLTRZ cells express lacZ underthe control of a Moloney murine leukemia viruslong terminal repeat, and MACTZ cells expresslacZ under the control of a b-actin promoter. (A)Correlation between CArG box activity anda-smooth muscle actin expression. Hillock-unas-sociated cells (U) and associated cells (A) wereharvested from an identical culture of the re-porter clone MCAGZ1, MCAGZ5, MLTRZ orMACTZ. Northern blot analyses were performedon lacZ and a-smooth muscle actin expression.(B) Reversibility of CArG box activity. Purifiedhillocks composed of MCAGZ5 cells weretrypsinized and re-seeded on plastic (A-.U).After several days, expression of lacZ as well asa-smooth muscle actin was compared to that inhillock-associated cells before trypsinization (A).(C) Stability of mRNAs. After the developmentof hillocks, cultures were treated with the inhib-itor of RNA synthesis, actinomycin D (500 ng/ml). Before or after the treatment for 12 and 24hours, hillock-associated cells (F) and unassoci-ated cells (E) were harvested separately fromidentical culture plates and used for Northernanalyses of lacZ and a-smooth muscle actin ex-pression.

Kitamura and Ishikawa: Deactivation of CArG elements by ECM 693

in hillocks was causative of the suppressed a-smooth muscle actinexpression, transfection studies were conducted. Mesangial cellswere transiently transfected with pSRE-LacZ, paAP126 or pCI-bgal, together with an empty plasmid or pDN-Elk, and activitiesof the a-smooth muscle actin promoter and CArG box elementswere assessed by X-gal assay. As previously reported [26, 27],introduction of the dominant negative form of Elk-1 markedlyrepressed the basal activity of CArG box elements [5.9 6 5.9%versus control (100%); Fig. 4D, left]. Consistent with the resultsfrom MmuSRF cells, inactivation of CArG box elements by theElk-1 mutant also substantially repressed activity of the a-smoothmuscle actin promoter [11.7 6 2.3% versus control (100%); Fig.4D, right].

Activity of CArG box elements and expression of a-smoothmuscle actin in mesangial cells embedded in collagen gels

To investigate the potential of three-dimensional ECMs formodulating the activity of CArG elements, MCAGZ5 cells wereembedded in collagen gels, and expression of lacZ and a-smoothmuscle actin was compared to that of the cells on plastic (Fig. 5A).Consistent with the results from hillocks, the three-dimensionalgel matrix dramatically inhibited the expression of lacZ as well asa-smooth muscle actin gene. Thin coating of plastic with type Icollagen failed to induce down-regulation of lacZ and a-smoothmuscle actin in the reporter mesangial cells (Fig. 5B).

DISCUSSION

Expression of a-smooth muscle actin alters in certain cell typesin response to local pathophysiologic conditions. For example,normally, fibroblasts in various organs do not express a-smoothmuscle actin, whereas after insults, activated fibroblasts (myofi-broblasts) abundantly synthesize this actin isoform [35]. Similarly,in glomerular mesangial cells, a-smooth muscle actin is expressedexclusively under pathologic situations and has been regarded asa crucial marker for cytoactivation and dedifferentiation [5].

Various peptide factors may function as potential regulators ofa-smooth muscle actin expression. Those include; angiotensin II,endothelin-1, arginin vasopressin, transforming growth factor-b,platelet-derived growth factor (PDGF), granulocyte macrophage-colony stimulating factor and interferon-g [36–41]. However,currently, information is limited regarding other environmentalfactors that regulate the expression of a-smooth muscle actin.Using nodular culture and collagen gel culture methods, this studyaddresses a novel potential of three-dimensional ECMs for down-regulation of a-smooth muscle actin expression in mesangial cells.The three-dimensional structure was found to be essential for thisaction, since a collagen-coated plastic failed to exert this biologicactivity.

In rat mesangial cells, a core fragment of the a-smooth muscleactin promoter from positions 2122 to 11 is crucial for seruminduction of the a-smooth muscle gene [16]. This sequencecontains two highly conserved CArG box motifs that alone confermaximum levels of serum inducibility on a heterologous minimalpromoter. Deletion of either sequence reduces the target genetranscription, suggesting that both CArG box elements are re-quired for efficient transcription [16]. This has led us to examinethe activity of CArG box elements in mesangial cells incorporatedin hillocks and collagen gels. In this study, reporter mesangial cellsin ECMs exhibited repressed activity of CArG box elementscompared to the cells in two-dimensional cultures. In bothexperimental settings, matrix-associated and unassociated cellswere located in the identical culture media containing 10 to 20%FCS that stimulates CArG box elements in MCAGZ cells [21].These data imply that three-dimensional ECMs may blunt thesensitivity of the cells to serum factor(s).

Extracellular matrix is known to generate intracellular signal-ling via the cell surface receptors, integrins, and modulate tran-scription of various genes [42]. Marx and coworkers reported thatthree-dimensional cultures using a collagen gel resulted in aremarkable down-regulation of PDGF-b-receptors on the surfaceof mesangial cells [43]. Since PDGF is a stimulator of CArGelements in mesangial cells [44] and is present in the basal growthmedia containing serum, the unresponsiveness to PDGF viadown-regulation of its receptor may explain the attenuated CArGbox activity.

Another possible mechanism is via focal adhesion kinase(FAK). FAK is a nonreceptor protein-tyrosine kinase implicatedin controlling cellular responses to the engagement of cell surfacereceptor integrins [45]. Following ligation with ECM components,integrins initiate phosphorylation of FAK and consequently mod-ulate cell behavior including spreading, migration, proliferationand gene expression [46]. Based on this action, the surroundingECM may continuously activate FAK and affect the activity ofCArG elements in hillock-associated cells. Interestingly, we re-cently found that v-Src, an well-known activator of FAK, re-pressed both activity of CArG box elements and expression ofa-smooth muscle actin in mesangial cells [47]. Three-dimensionalECMs might suppress CArG box activity via integrin-mediatedactivation of FAK.

To examine whether functional inactivation of CArG elementsleads to repressed a-smooth muscle actin expression, mesangialcells were transfected with a cDNA encoding a transdominantnegative mutant of SRF, an essential component for CArG boxactivation. The established transfectants showed reduced activityof CArG box elements and repressed expression of a-smooth

Fig. 3. Activity of Elk-1 in hillock-associated cells and unassociated cells.Superconfluent SM43 cells were transiently transfected with pUASg-LacZor a control plasmid pCIb-gal, together with pGAL4-Elk. After complet-ing the transfection, hillocks were induced by exposure of the cells to 20%FCS. Activity of Elk-1 was evaluated by 5-bromo-4-chloro-3-indolyl b-D-galactopyranoside (X-gal) assay. The number of X-gal-positive cellstransfected with pGAL4-Elk and pUASg-LacZ was counted and normal-ized by the number of X-gal-positive cells transfected with pGAL4-Elkand pCI-bgal. The value (mean 6 SE) of hillock-associated cells (A) wasexpressed as a relative percentage against the value of hillock-unassoci-ated cells (U: 100%). Assays were performed in quadruplicate. An asteriskindicates statistically significant difference (P , 0.05).

Kitamura and Ishikawa: Deactivation of CArG elements by ECM694

muscle actin gene. Furthermore, transient transfection with thedominant negative mutant of Elk-1, another crucial componentfor CArG box activation, markedly suppressed the activity ofa-smooth muscle actin promoter, confirming that inactivation ofCArG elements was causative of the depressed a-smooth muscleactin in mesangial cells surrounded by ECMs.

Currently, correlation between the activity of CArG box ele-ments and other phenotypic markers such as mitogenic activity isunknown. However, it is worthwhile to note that, in mesangialcells (i) PDGF is a prominent mitogenic factor in vivo and in vitro[48], (ii) PDGF induces egr-1 expression via activation of CArGbox elements [44], and (iii) inhibition of egr-1 by antisenseoligonucleotides suppresses PDGF-induced mitogenesis [49].Inactivation of CArG elements possibly affects other phenotypicproperties of mesangial cells towards deactivation and differenti-ation.

In this study, the role of three-dimensional ECMs was broughtinto focus to explain the phenotypical shift of mesangial cells inhillocks. However, the expression of a-smooth muscle actin and

the activity of CArG box elements might be controlled via othermechanisms. Following the three-dimensional organization ofcells, several environmental differences may be created, such asalterations in cell-cell contact, cell-matrix interaction and diffu-sion rates of oxygen, nutrients, growth factors and metabolicproducts. These differences in microenvironment may affect cellfunctions and could be, in part, responsible for the phenotypicchange of mesangial cells in hillocks.

In summary, the present data suggested the potential impor-tance of mesangial matrix as a modifier of mesangial cell pheno-type. The CArG box element was found to be a molecular targetinvolved in this regulatory process. To our knowledge, this studyprovides the first evidence that the CArG box activity is control-lable by signals initiated by three-dimensional ECMs.

ACKNOWLEDGMENTS

This work was supported by grants from Wellcome Trust, BaxterHealthcare Corporation (Extramural Grant Program) and National Kid-ney Research Fund to M. Kitamura.

Fig. 4. Effect of inactivation of CArG box elements on transcrip-tional activity of the a-smooth muscle actin gene. (A) Creation ofstable transfectants that express a transdominant negative mutantof serum response factor (SRF). SM43 cells were transfected withpCGNSRFpm1 encoding a dominant negative form of SRF, andstable transfectants MmuSRF1 and MmuSRF2 were established.Expression of endogenous SRF and exogenous mutant SRF(muSRF) in untransfected cells, mock transfected cells andMmuSRF cells was examined by Northern analysis. (B) Activity ofCArG box elements in MmuSRF cells. Mock transfected cells andMmuSRF1 cells were transiently transfected with pSRE-LacZ.After 48 hours, cells were subjected X-gal assay to evaluateb-galactosidase activity. The activity of CArG box elements wasassessed by counting blue cells stained by X-gal. To excludepossible variability in transfection efficiency between different cellclones, the number of X-gal-positive cells was normalized by thenumber of positive cells transfected with a control plasmid pCXL.Data were expressed as means 6 SE. Assays were performed inquadruplicate. An asterisk indicates statistically significant differ-

ence (P , 0.05). (C) Expression of a-smooth muscle actin mRNA in MmuSRF cells. Untransfected cells, mock transfected cells and MmuSRF cellsin monolayer culture were subjected to Northern blot analysis. (D) Inactivation of CArG box elements by a transdominant negative mutant of Elk-1and altered activity of the a-smooth muscle actin promoter. SM43 cells were transiently transfected with pSRE-LacZ, paAP126 or a control plasmidpCI-bgal, together with an empty plasmid or pDN-Elk. paAP126 introduces lacZ under the control of 2894 to 112 base pairs of the human a-smoothmuscle actin gene. pDN-Elk encodes a mutant Elk-1 that has deletion of the carboxy-terminal activation domain. After 48 hours, cells were subjectedto X-gal assay to evaluate the activities of CArG box elements and the a-smooth muscle actin promoter. The number of X-gal-positive cells transfectedwith pSRE-LacZ or paAP126 was counted and normalized by the number of X-gal-positive cells transfected with pCI-bgal. Data are presented asrelative percentages against controls [pSRE-LacZ or paAP126 1 empty vector (100%)]. Assays were performed in quadruplicate. Asterisks indicatestatistically significant differences (P , 0.05).

Kitamura and Ishikawa: Deactivation of CArG elements by ECM 695

Reprint requests to Masanori Kitamura, M.D., Glomerular BioengineeringUnit, Department of Medicine, University College London Medical School,The Rayne Institute, 5 University Street, London WC1E 6JJ, England, UnitedKingdom.E-mail: m.kitamura@medicine.ucl.ac.uk

APPENDIX

Abbreviations used in this article are: CArG box, CC(A/T)6GG se-quences; DME-F-12, Dulbecco’s modified Eagle’s medium/Ham’s F-12;ECM, extracellular matrix; FAK, focal adhesion kinase; FCS, fetal calfserum; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; lacZ, b-ga-lactosidase gene; LTR, long terminal repeat; neo, neomycin phosphotrans-ferase gene; PBS, phosphate buffered saline; PDGF, platelet-derivedgrowth factor; SRE, serum response element; SRF, serum response factor;TCF, ternary complex factor; X-gal, 5-bromo-4-chloro-3-indoyl b-D-galac-topyranoside.

REFERENCES1. KITAMURA M, MITARAI T, MARUYAMA N, NAGASAWA R, YOSHIDA H,

SAKAI O: Mesangial cell behavior in a three-dimensional extracellularmatrix. Kidney Int 40:653–661, 1991

2. KITAMURA M, SHIRASAWA T, MARUYAMA N: Gene transfer of metal-loproteinase transin induces aberrant behavior of cultured mesangialcells. Kidney Int 45:1580–1586, 1994

3. KITAMURA M, MITARAI T, NAGASAWA R, MARUYAMA N: Differenti-ated phenotype of glomerular mesangial cells in nodular culture. Am JPhysiol 270:F614–F622, 1996

4. STERZEL RB, LOVETT DH, FOELLMER HG, PERFETTO M, BIEMESDER-FER D, KASHGARIAN M: Mesangial cell hillocks. Nodular foci ofexaggerated growth of cells and matrix in prolonged culture. Am JPathol 125:130–140, 1986

5. JOHNSON RJ, FLOEGE J, YOSHIMURA A, IIDA H, COUSER WG, ALPERSCE: The activated mesangial cell: A glomerular “myofibroblast”?J Am Soc Nephrol 2(Suppl):S190–S197, 1992

6. ALPERS CE, HUDKINS KL, GOWN AM, JOHNSON RJ: Enhancedexpression of “muscle-specific” actin in glomerulonephritis. Kidney Int41:1134–1142, 1992

7. ELGER M, DRENCKHAHN D, NOBILING R, MUNDEL P, KRIZ W:Cultured rat mesangial cells contain smooth muscle a-actin not foundin vivo. Am J Pathol 142:497–509, 1993

8. JOHNSON RJ, IIDA H, ALPERS CE, MAJESKY MW, SCHWARTZ SM,PRITZL P, GORDON K, GOWN AM: Expression of smooth muscle cell

phenotype by rat mesangial cells in immune complex nephritis:a-smooth muscle actin is a marker of mesangial cell proliferation.J Clin Invest 87:847–858, 1991

9. MACPHERSON BR, LESLIE KO, LIZASO KV, SCHWARZ JE: Contractilecells of the kidney in primary glomerular disorders: An immunohis-tochemical study using an anti-alpha smooth muscle actin monoclonalantibody. Hum Pathol 24:710–716, 1993

10. FLOEGE J, JOHNSON RJ, GORDON K, IIDA H, PRITZL P, YOSHIMURA A,CAMPBELL C, ALPERS CE, COUSER WG: Increased synthesis ofextracellular matrix in mesangial proliferative nephritis. Kidney Int40:477–488, 1991

11. KIMURA K, HIROI J, TOJO A, SUZUKI N, OHBA S, MISE N, OMATA M,NAGAI R: Multiplicity of phenotypic modulation in glomerular cellsduring development of different glomerular diseases. (abstract) J AmSoc Nephrol 6:871, 1995

12. BLANK RS, MCQUINN TC, YIN KC, THOMPSON MM, TAKEYASU K,SCHWARTZ RJ, OWENS GK: Elements of the smooth muscle a-actinpromoter required in cis for transcriptional activation in smoothmuscle. J Biol Chem 267:984–989, 1992

13. KIM J-H, BUSHEL PR, KUMAR CC: Smooth muscle a-actin promoteractivity is induced by serum stimulation of fibroblast cells. BiochemBiophys Res Commun 190:1115–1121, 1993

14. MIN B, FOSTER DN, STRAUCH AR: The 59-flanking region of themouse vascular smooth muscle a-actin gene contains evolutionarilyconserved sequence motifs within a functional promoter. J Biol Chem265:16667–16675, 1990

15. TREISMAN R: The serum response element. Trends Biochem Sci17:423–426, 1992

16. SIMONSON MS, WALSH K, KUMAR CC, BUSHEL P, HERMAN WH: Twoproximal CArG elements regulate SM a-actin promoter, a geneticmarker of activated phenotype of mesangial cells. Am J Physiol268:F760–F769, 1995

17. KITAMURA M, TAYLOR S, UNWIN R, BURTON S, SHIMIZU F, FINE LG:Gene transfer into the rat renal glomerulus via a mesangial cell vector:Site-specific delivery, in situ amplification and sustained expression ofan exogenous gene in vivo. J Clin Invest 94:497–505, 1994

18. KITAMURA M, BURTON S, ENGLISH J, KAWACHI H, FINE LG: Transferof a mutated gene encoding active transforming growth factor-b1suppresses mitogenesis and IL-1 response in the glomerulus. KidneyInt 48:1747–1757, 1995

19. MEINKOTH J, ALBERTS A, FERAMISCO J: Construction of mammaliancell lines with indicator genes driven by regulated promoters. CIBAFoundation Symp 150:47–56, 1990

20. KITAMURA M: Creation of a reversible on/off system for site-specific invivo control of exogenous gene activity in the renal glomerulus. ProcNatl Acad Sci USA 93:7387–7391, 1996

21. KITAMURA M, KAWACHI H: Creation of an in vivo cytosensor usingengineered mesangial cells: Automatic sensing of glomerular inflam-mation controls transgene activity. J Clin Invest 100:1394–1399, 1997

22. KLUNDER I, HULSER DF: b-Galactosidase activity in transfected Ltk-cells is differentially regulated in monolayer and in spheroid cultures.Exp Cell Res 207:155–162, 1993

23. KIM J-H, JOHANSEN F-E, ROBERTSON N, CATINO JJ, ORYWES R,KUMAR CC: Suppression of Ras transformation by serum responsefactor. J Biol Chem 269:13740–13743, 1994

24. YOKOO T, KITAMURA M: Dual regulation of IL-1b-mediated matrixmetalloproteinase-9 expression in mesangial cells by NF-kB and AP-1.Am J Physiol 270:F123–F130, 1996

25. MIKAWA T, FISCHMAN DA, DOUGHERTY JP, BROWN AMC: In vivoanalysis of a new lacZ retrovirus vector suitable for cell lineagemarking in avian and other species. Exp Cell Res 195:516–523, 1991

26. WHITMARSH AJ, SHORE P, SHARROCKS AD, DAVIS RJ: Integration ofMAP kinase signal transduction pathways at the serum responseelement. Science 269:403–407, 1995

27. KITAMURA M, ISHIKAWA Y: Functional inactivation of signallingmolecules via transdominant negative mutants. Exp Nephrol 5:435–438, 1997

28. KUMAR CC, BUSHEL P, MOHAN-PETERSON S, RAMIREZ F: Regulationof smooth muscle a-actin promoter in ras-transformed cells: Useful-ness for setting up reporter gene-based assay system for drug screen-ing. Cancer Res 52:6877–6884, 1992

29. ARGENTON F, ARAVA Y, ARONHEIM A, WALKER MD: An activationdomain of the helix-loop-helix transcription factor E2A shows cell

Fig. 5. Activity of CArG box elements and expression of a-smooth muscleactin in mesangial cells embedded in collagen gels or grown on collagen-coated plastic. (A) In collagen gels. MCAGZ5 cells were incorporated intype I collagen gels as described in the Methods section. After 48 hours,cells grown on the basement plastic and within floating gels wereseparately harvested and subjected to Northern analyses of lacZ anda-smooth muscle actin expression. (B) On collagen-coated plastic.MCAGZ5 cells were seeded on plain plastic or plastic coated with collagen,cultured for 48 hours and subjected to Northern analyses.

Kitamura and Ishikawa: Deactivation of CArG elements by ECM696

type preference in vivo in microinjected Zebra Fish embryo. Mol CellBiol 16:1714–1721, 1996

30. KITAMURA M, MARUYAMA N, YOSHIDA H, NAGASAWA R, MITARAI T,SAKAI O: Extracellular matrix contraction by cultured mesangial cells:An assay system for mesangial cell-matrix interaction. Exp Mol Pathol54:181–200, 1991

31. KITAMURA M, SUTO T, YOKOO T, SHIMIZU F, FINE LG: Transforminggrowth factor-b1 is the predominant paracrine inhibitor of macro-phage cytokine synthesis produced by glomerular mesangial cells.J Immunol 156:2964–2971, 1996

32. MCHUGH KM, LESSARD JL: The nucleotide sequence of a rat vascularsmooth muscle a-actin cDNA. Nucleic Acids Res 16:4167, 1988

33. PRICE J, TURNER D, CEPKO C: Lineage analysis in the vertebratenervous system by retrovirus-mediated gene transfer. Proc Natl AcadSci USA 84:156–160, 1987

34. TREISMAN R: The SRE: A growth factor responsive transcriptionalregulator. Semin Cancer Biol 1:47–58, 1990

35. GABBIANI G: The biology of the myofibroblast. Kidney Int 41:530–532,1992

36. ANDRAWIS NS, RULEY EH, ABERNETHY DR: Angiotensin II regulateshuman vascular smooth muscle a-actin gene expression. BiochemBiophys Res Commun 196:962–968, 1993

37. HAHN AW, RESINK TJ, KERN F, BUHLER FR: Effects of endothelin-1on vascular smooth muscle cell phenotypic differentiation. J Cardio-vasc Pharmacol 20:S33–S36, 1992

38. VAN PUTTEN V, LI X, MASELLI J, NEMENOFF RA: Regulation ofsmooth muscle actin promoter by vasopressin and platelet-derivedgrowth factor in rat aortic vascular smooth muscle cells. Circ Res75:1126–1130, 1994

39. ARCINIEGAS E, SUTTON AB, ALLEN TD, SCHOR AM: Transforminggrowth factor beta-1 promotes the differentiation of endothelial cellsinto smooth muscle-like cells in vitro. J Cell Sci 103:521–529, 1992

40. RUBBIA-BRANDT L, SAPPINO A-P, GABBIANI G: Locally applied GM-CSF induces the accumulation of a-smooth muscle actin containingmyofibroblasts. Virchows Archiv B Cell Pathol 60:73–82, 1991

41. HANSSON GK, HELLSTRAND M, RYMO L, RUBBIA L, GABBIANI G:Interferon g inhibits both proliferation and expression of differentia-tion-specific a-smooth muscle actin in arterial smooth muscle cells. JExp Med 170:1595–1608, 1989

42. JULIANO RL, HASKILL S: Signal transduction from the extracellularmatrix. J Cell Biol 120:577–585, 1993

43. MARX M, DANIEL TO, KASHGARIAN M, MADRI JA: Spatial organiza-tion of the extracellular matrix modulates the expression of PDGF-receptor subunits in mesangial cells. Kidney Int 43:1027–1041, 1993

44. RUPPRECHT HD, SUKHATME VP, LACY J, STERZEL RB, COLEMAN DL:PDGF-induced Egr-1 expression in rat mesangial cells is mediatedthrough upstream serum response elements. Am J Physiol 265:F351–F360, 1993

45. SCHALLER MD, BORGMAN CA, COBB BS, VINES RR, REYNOLDS AB,PARSONS JT: pp125FAK, a structurally distinctive protein-tyrosinekinase associated with focal adhesion. Proc Natl Acad Sci USA89:5192–5196, 1992

46. HANKS SK, POLTE TR: Signaling through focal adhesion kinase.BioEssay 19:137–145, 1997

47. ISHIKAWA Y, KITAMURA M: Unpredicted suppression of the ERK-Elk1-SRE pathway by v-Src tyrosine kinase in cultured mesangialcells. (abstract) J Am Soc Nephrol 8:440A, 1997

48. JOHNSON RJ, FLOEGE J, COUSER WG, ALPERS CE: Role of platelet-derived growth factor in glomerular disease. J Am Soc Nephrol4:119–128, 1993

49. HOFER G, GRIMMER C, SUKHATME VP, STERZEL RB, RUPPRECHT HD:Transcription factor Egr-1 regulates glomerular mesangial cell prolif-eration. J Biol Chem 271:28306–28310, 1996

Kitamura and Ishikawa: Deactivation of CArG elements by ECM 697