Extracellular matrix derived fromTrypanosoma cruzi infected endothelial cells directs phenotypic expression

JOURNAL OF CELLULAR PHYSIOLOGY 145:340-346 (1990)

Extracellular Matrix Derived From Trypanosoma cruzi Infected Endothelial Cells Directs Phenotypic Expression STEPHEN A. MORRIS,* MURRAY WITTNER, LOUIS WEISS, VICTOR B. HATCHER, HERBERT B. TANOWITZ, JOHNP. BILEZIKIAN, AND PORTIA B. GORDON College of Physicians and Surgeons, New York, New York 10032 (S.A.M., j.P.5.); Albert Einstein College of Medicine, Bronx, New York 1046 I (M.W., 1.W., V.5.H., H.B.T., P.B.C.)

infection of confluent human umbilical vein endothelial cells by the parasite Trypanosoma cruzi results in the appearance of a n altered heparan sulfate proteoglycan (HSPG) isolated from the extracellular matrix of infected endothelial cells (ECMi). HSPG from ECMi differed from HSPG obtained from the extracellular matrix of uninfected endothelial cells (ECMu) by virtue of an 8-1 0-fold increase in sulfation and a different elution pattern using DEAE Sepharose chromatography. Analysis of the HSPG that binds to acidic fibroblast growth factor (aFGF) revealed that infection increased the proportion of HSPG that binds to aFGF by 35%. Heparitinase and alkaline borohydride treatment of aFGF-binding HSPG and chromatographic resolution on Sepharose CL4B column revealed an infectionassociated 1 0-fold increase in sulfation of the GAG side chain with no significant change in the migration of the core protein. in addition, the aFCF binding HSPG isolated from ECMi demonstrated a markedly attenuated synergistic mitogenic activity with aFGF in a cell proliferation assay. All of the infection associated changes in HSPG could be demonstrated in HSPG obtained from uninfected endothelial cell cultures grown on ECMi. Hence, the ECMi is associated with signals capable of modulating the ECM associated metabolism of uninfected endothelial cells. This facility of ECMi was also shown to extend to patterns of Gs protein synthesis as revealed by Western blot analysis. The observation that the ECM produced by infected endothelial cells can direct the synthetic patterns of uninfected endothelial cells in a manner uniquely observed in infected endothelial cells suggests a plausible pathway by which infection of only a few cells can ultimately result in the coordinate responses of neighboring uninfected cells. Chronic chagasic cardiomyopathy following Trypun o s o m cruzi infection affects 10-15 million people in the Western hemisphere. One of the major obstacles in understanding this often fatal chronic disease is the lack of correlation between the presence of parasites in the tissues and the extent of the disease (Rosenbaum, 1965). Experimentally, there is a similar lack of correlation between the degree of parasitic infection and altered cellular physiology. Thus, we observed identical alterations in CAMP generation, intracellular Ca2+ mobilization, patterns of G protein ADP ribosylation, and inositol phosphate metabolism inde endent of whether 5% or 50% of cultured endothelia cells harbored the intracellular amastigote form of the parasite (Morris et al., 1987, 1989). For these reasons, the possibility that infection may be associated with a toxin is an attractive, but as yet unsubstantiated, hypothesis. In this report, we demonstrate the T . cruzi infected human umbilical vein endothelial cells deposit into their extracellular matrix (ECM)heparan sulfate proteoglycans (HSPG, Gallagher et al., 1986) that are different and biologically less active than the HSPG de osited by uninfected endothelial cells. Furthermore, E8M from infected endothelial cells (ECMi)retains the

P

0 1990 WILEY-LISS, INC.

capacity to direct the synthesis not only of the infectionspecific HSPG in uninfected cells, but also altered patterns of Gs proteins, as detected by binding of anti Gas-antisera, suggesting that cells infected with the parasite produce signals that extend the conse uences of infection well beyond the relatively few cells arboring the parasite. MATERIALS AND METHODS Materials 'Na235S04(10-1,000 Ci/mmol)and NalZ5I( ~ 3 5mCi/ 0 ml) were obtained from New England Nuclear. [lz5I1Staph A protein was obtained from Amersham. Tissue culture supplies were obtained from Gibco, and all chemicals used were of reagent grade and obtained from sources as previously described (Gordon et al., 1989).

R

Received January 5, 1990; accepted August 7, 1990. *To whom reprint requestsicorrespondence should be addressed.

341

T.CRUZI. EXTRACELLULAR MATRIX, ENDOTHELIAL CELLS

Culture condition

matogra hy as previously described (Gordon et al., 1989). The DE E-Sepharose peak containing [35SlHSPG(Fig. 1) was dialyzed against 0.05 M Heges buffer, pH 7.5, containing 0.15 M NaC1, 4 X 10- M dextran sulfate (8,000 daltons), and 0.1% Triton X-100. [35Slheparan sulfate proteoglycans were then applied to an aFGFGrowth of parasites and infection of endothelial cells Affi-Gel15 column using purified aFGF (Gordon et al., The aFGF-Affi-Gel 15 column was equilibrated Forty-eight hours after plating, endothelial cells 1989). eluted with 0.05 M Hepes buffer, pH 7.5, containwere infected with the Tulahuen strain of T. cruzi and ing 0.15 M NaCl, 4 pM dextran sulfate (8,000 daltons), obtained directly from mice as previously described and 0.1% Triton X-100. [35SlHSPGwhich bound to the (Morris et al., 1984). column were eluted with 0.05 M Hepes buffer containing 2.0 M NaC1. aFGF binding [35SlHSPGwere precipMetabolic labeling of endothelial cells itated with 4 volumes of ethanol containing 1.5% Infected or uninfected cells were grown to confluence potassium for 18 hours at 4°C. This procedure (approximately7 days after plating at a concentration of precipitatedacetate 85-90% of the [35SlHSPG.The recovered 2.5 x lo3 cells/cm2), and metabolically labelled with HSPG were then resuspended in 4M guanidine hydromedium containing 10-50 FCi/ml NaZ3%O4; extra- chloride buffer, pH 6.3, and intact HSPG chromatocellular 35S-heparan sulfate proteoglycans were iso- graphed on a Sepharose CL-4B column equilibrated in lated as described (Gordon et al., 1985). the same buffer. Human umbilical vein endothelial cells were isolated according to the method of Jaffe (1980) and grown as previously described (Gordon et al., 1983).

Preparation of ECMi and ECMu for use as matrix for uninfected endothelial cells To examine the effect of ECM from infected cells on uninfected cell proteoglycan synthesis, infected endothelial cells were grown to confluence and ECMi isolated as described previously. Plates containing ECMi were incubated in 0.02 N NH40H at 37°C for 5 min, following which the plates were washed three times in Hepes buffer. In ex eriments not shown, we determined that this was procedure completely lysed and removed any viable parasitic forms without altering the characteristics of the ECM. In addition, similar treatment of try omastigotes rendered them noninfectious in endotheyial cells. Finally, uninfected endothelial cells subsequently rown on washed ECMi did not become infected. Unin ected cells were grown to confluence on these plates containing ECMi at a cell density equal to that used in all experiments. Confluent cultures were then incubated with [35SlNazS04as described above, and the ECM proteoglycans isolated. As controls, uninfected cells were grown on uninfected matrix (ECMu) infected cells grown on infected matrix (ECMi), and 38S-ECM proteoglycans isolated as described. DEAE sepharose chromatography

R

1

Separation of core protein and glycosaminoglycans of the heparan sulfate proteoglycans To isolate the core protein, fractions of aFGF-binding HSPG were first iodinated with Na1251on enzymobeads according to the method described in the BioRad bulletin for single reaction enzymobead radioiodination. Radiolabelled HSPG was then treated with 0.05 M Hepes buffer containing 10 microunits of heparitinase for 16 hours at 37"C, which was subsequently measured as CPM's lZ5I (Gordon et al., 1989). In other experi-

B

1800[ I600 1400 1200

1000 8 00

of S5S-proteoglycans

6 00

The labelled proteoglycans were applied to a column of DEAE-Sepharose pre-equilibrated with the 7.5 M urea buffer containing rotease inhibitors and 0.5% (w/v) Triton X-100 (Gor on et al., 1985). The column was washed and then eluted with a continuous NaCl gradient from 0.15 M to 1.15 M in the same buffer. The concentration of NaCl in the elution fractions is noted by the dotted line, and was determined as described previously (Gordon et al., 1989).

400

B

Isolation of aFGF-bindingECM heparan sulfate proteoglycans 35S-proteoglycans isolated after DEAE-Sepharose column chromatography were determined to be greater than 85% HSPG (Gordon et al., 1989). 35S-heparan sulfate proteoglycans were further purified using aFGF affinity chromatography and Sepharose CL-4B chro-

ff

200 0

10 20 30 40 50 60 70

Fraction# Fig. 1. The influence of T. cruzi infection on the extracellular matrix of endothelial cells. ECM-HSPG from confluent endothelial cells labelled with 10-50 $Xml Na236S0, was isolated by DEAE Sepharose chromatography as described. For each run shown here, total CPMs applied to the column were identical. 0 , uninfected endothelial cells grown on ECMU; 0 , infected cells grown on ECMi; 0 , uninfected cells grown on ECMi. It is to be noted that the figure shown here reflects data obtained from cultures where 5% of the cells were infected. Identical curves were obtained using cultures where 17,28, and 54% of the cells were infected. In all experiments, results are expressed as counts per minute (CPMs) per lo6 cells.

342

MORRIS ET AL.

ments, no labelling of core protein was carried out, and following heparatinase treatment, core protein was monitored by performing a BioRad protein assay on aliquots of each fraction following Sepharose CL-4B column chromatography, and fractions were monitored at OD of 595 nm. For isolation of GAG side chains, aFGF-binding HSPG was treated with 0.05 M NaOH, 1 M sodium borohydride for 24 hours, reprecipitated with 4 volumes of absolute ethanol containing 1.5% potassium acetate, as described previously. This method precipitated 85%of [35S-GAG]. The precipitates were then rechromatographed on Sepharose CL4B column to isolate GAG as described (Gordon et al., 1989). Quantitation of urionic acid in the GAG side chains was performed as previously reported (Gordon et al. 1989), and used to determine the specific activity of ['%I. Assay of aFGF's mitogenic stimulation of sparse endothelial cell cultures The isolation and purification of aFGF-binding HSPG or GAG side chains, subsequent Sepharose CL4B chromatography, and the assay used to determine the mitogenic potential of aFGF were performed as described (Gordon et al., 1989). Briefly, human endothelial cells were plated at a density of 2,500-5,000 cells/cm2 in Medium 199 containing 20% fetal calf serum with or without 14 pM aFGF and 50 nmol/ml of aFGF-Affi-Gel affinity fractionated [35SlHSPG (based on nmol of glucuronate) as described (Gordon et al., 1989).Amounts (nmols) of HSPG were determined by a modified carbazole assay (Bitter and Muir, 1962).Cells were incubated in the appropriate medium for 4 days at 37°C harvested with trypsin, and counted in a hemocytometer. Preparation of anti-Gas antisera and Western blot analysis of endothelial cell membrane proteins The antigen used to raise antisera contained a 16 amino acid sequence unique to the G,, subunit of Gs (Peninsula Laboratories, San Carlos, CAI. The polypeptide was conjugated to Keyhole limpet hemocyanin (KLH) via the amino terminal cysteine residue of the polypeptides using m-maeimido-benzyol-N-hydroxysuccinimide ester as described (Ransnas and Insel, 1988). The peptide-KLH conjugate was used to raise polyclonal, monospecific antisera in rabbits. A total of 100 pg peptide-KLH emulsified in complete Freund's adjuvant was injected intradermally at 15 to 20 sites on the rabbit and was followed by a booster injection of 50 pg peptide-KLH in incomplete Freund's adjuvant after 4 weeks. Sera from various rabbits was tested for binding to the antigen using an immunoblot technique, and those samples giving positive reaction at 1500 dilution were used. Western blot analysis were performed following overnight transfer of proteins run on SDS-PAGE (prepared as described above) to nitrocellulose paper in a buffer containing 20% methanol, glycine (150 mM), and Tris (20 mM) at 75 V and 250 mA. Nitrocellulose blots were then incubated with rabbit anti G,, antisera as previously described, and [lZ5I]StaphA protein was used to detect bound antibody. Specificity of the antisera to G,, was evaluated by two criteria. First, no antibody binding in the 35-65

kDa region was detectable in membranes prepared from cyc- cells, mutants of S49 lymphoma cells lacking Gs (Bourne et al., 1975). Second, in simultaneous assays performed under identical reaction conditions, cholera toxin dependent ADP ribosylation of dog heart membranes occurred in a molecular weight region (42-45 kDa) completely superimposable with antibody binding.

RESULTS

Influence of infection on the ECM We first compared the properties of the ECM obtained from uninfected and infected endothelial cells. ECM was isolated from confluent cultures of T. cruzi infected or uninfected human umbilical vein endothelial cells metabolically labeled with [35S1Na2S04 (Gordon et al., 1986). We have previously characterized the ECM of uninfected endothelial cells (Gordon et al., 1989), and include this data for the sake of comparison. As shown in Figure 1,ECM-derived [35Slproteoglycans synthesized by uninfected endothelial cells (ECMu) were eluted in 0.3 M NaC1. ECM derived [35Slproteoglycans synthesized by infected endothelial cells (ECMi) demonstrated an (4-fold) increase in 35S-incorporation based on counts per minute per lo6 cells, and were eluted as a single peak in a buffer of greater ionic strength, 0.5 M NaC1. These [35Slproteoglycansfrom both ECMu and ECMi were determined to be HSPG by comparing Se harose CL-4B chromatographic profiles before and a ter either nitrous acid, heparitinase, or chondroitinase ABC treatment, as previously described (Gordon et al., 1989; results not shown). Characterization of the infection associated changes in HSPG As previously demonstrated (Gordon et al., 19891, aFGF-binding HSPG derived from uninfected endothelial cells (approximately 50% of total HSPG isolated from the ECM) synergistically enhances the mitogenic capacity of acidic fibroblast growth factor on endothelial cells. In light of this potential to link biological activity with the biochemistry of a subfraction of HSPG, we examined more closely the influence of infection on aFGF binding HSPG. Virtually all (greater than 85%)of the HSPG derived from ECMi bound to an aFGF-Affi Gel affinity column (results not shown). Sepharose CL 4B elution patterns of pooled aFGFbinding L3%1HSPG from ECMu and ECMi differed in migration and configuration (Fig. 2). Material obtained from the ECM of uninfected endothelial cells (ECMu) migrated as a major peak (I)with a minor shoulder (11). The major peak eluted within the first fraction obtained following the void volume with the shoulder fraction. In contrast, the of the aFGF-binding HSPG obtained only one major peak with a different elution pattern than that of aFGF-binding HSPG from ECMu. In addition, endothelial cells infected with T. cruzi incorporated 8-10-fold more Naz 35S04into aFGF-binding ECM HSPG as compared to uninfected endothelial cells based on counts per minute per lo6 cells. We next sought to characterize further the HSPG with regard to its core protein and GAG side chain.

P

-

343

T. CRUZI, EXTRACELLULAR M[ATRIX,ENDOTHELIAL CELLS

300

r

Kav

0

A

0.2 0.4

0.6

0.0

derived HSPG, was significantly higher than the specific activity of GAG chains from ECMu derived HSPG (105k3 cpdnmol uronate HSPGi vs. 1 0 t 2 cpdnmol uronate HSPGu). These data are consistent with an infection associated change predominantly in the sulfation of the GAG side chain. We next performed a series of experiments (data not shown) to confirm that the altered ECMi proteoglycans followed as a consequence of host cell infection. First, trypomastigotes metabolically labelled with [35S]do not produce [35Slproteoglycans.Second, [35SlECM obtained from endothelial cells grown on ECMu that had been previously incubated with trypomastigotes for 2 days and then washed with 0.02 N NH40H to kill trypomastigotes prior to replating of fresh cells was identical to HSPG obtained from untreated ECMu. Third, ECMi obtained from cultures in which only 5% of the cells were infected had identical characteristics to ECMi derived from 50% infected cultures, rendering it unlikely that retained parasite products were responsible for the changes in ECM. Finally, we observed no difference in cell number between infected and uninfected endothelial cells; i.e., in this experimental design, the rate of proliferation was identical and uninfluenced by infection. Hence, infection associated changes in ECM are due specifically to the presence of the intracellular parasite.

1.0

i14000

12000

Vt

vo

11.4

"OF" 2400

T 2ooo-

- 0.0

m

- 0.6

- 0.4

cn

8

T

- 0.2 50

tI

0

t

0

4,

VO

Fraction [ml)

Fig. 2. Sepharose CL-4B column chromatography of aFGF-binding ECM . HSPG Influence of heparatinase and alkaline borohydride treatment. A: ECM repared from uninfected cells. B: ECMi prepared from infected endotRelia1 cells (identical results were obtained for ECM obtained from uninfected endothelial cells grown on ECMi). Closed squares (I), intact HSPG closed triangles (A),aFGF-binding HSPG treated with heparitinase measured as CPM's lz5I in panel A and aliquots of each fraction monitored at an OD of 595 nm for protein using the BioRad assay in panel B; closed circles (01, aFGF-binding HSPG treated with 0.05 M NaOH, 1 M sodium borohydride for 24 hours. Radiolabel results in both panels are expressed as counts per minute (cpm) per lo6 cells.

Heparatinase treatment of aFGF-binding HSPG resulted in the generation of core protein. Core protein of aFGF-bindin HSPG from uninfected or infected cells demonstrate the same elution profile (Fig. 2). Alkaline borohydride treatment of HSPG obtained from ECM of uninfected and infected endothelial cells permitted the subse uent isolation of GAG side chains. It was determined t at infected endothelial cells incorporated 5-fold more NaZ3%O4 into GAG chains than uninfected cells on a per cell basis. In addition the specific activity of the GAG chains, isolated from ECMI

%

1

Role of the ECM in the expression of infection associated changes in HSPG Consistent with our previous studies (Morris et al., 1989), infection associated changes in the ECM of endothelial cells are also independent of the magnitude of infection (legend to Fig. 1).We therefore pursued the possibility that the state of infection may be associated with a toxin like substance with an influence on surroundin uninfected endothelial cells. The possibility that suc a factor may be secreted into the medium by infected cells was ruled out by the observation that the medium of infected cells did not alter the synthetic patterns of the ECM of uninfected endothelial cells (results not shown). We determined whether the ECM itself may serve as the basis by which infection of only a few cells may influence a large number of uninfected neighboring cells. As a marker for the influence of infection, we examined HSPG obtained from uninfected endothelial cells grown on ECMi. Both nitrous acid and heparitinase treatment, as well as DEAE (Fi 1)or Sepharose CL 4B (Fig. 2) elution patterns of [3 IHSPG obtained from ECM of uninfected cells grown on ECMi gave results identical to [35SlHSPGfrom ECMi produced by infected endothelial cells grown on ECMi. In experiments not shown, it was determined that the ECM obtained from uninfected endothelial cells grown either on ECMu or on gelatin were identical with respect to 35S-HSPGcomposition. Collectively, these results indicated that the infection-associated changes in the HSPG of the ECM 1) do not require the continued presence of the parasite, and 2) the ECM synthesized by the infected host cell retains the ca acity to direct the synthesis of infection-specific HSP .

fl

4

4

344

MORRIS ET AL.

Determination of the influence of infection on the mitogenic activity of aFGF-binding HSPG In a test of the biological activity of HSPG, we next evaluated the synergistic influence of aFGF-binding HSPG and aFGF on endothelial cell proliferation (Gordon et al., 1989) and the influence of infection. Using cells plated at sparse cell densities in growth medium supplemented with aFGF, marked cell proliferation occurs (Fig. 3); growth medium without aFGF cannot support cell proliferation. The roliferative res onse to the combined presence o aFGF-binding ['SIHSPG from ECMu with aFGF was double that observed with aFGF alone. Addition of aFGF-binding [35SlHSPGfrom ECMi to growth medium containing aFGF did not synergistically enhance the proliferative response observed with aFGF alone. Thus, the HSPG obtained from ECMu and ECMi both bind with the same affinity to aFGF, but only the HSPG obtained from ECMu retains biological activity. Alone, HSPG from ECMu or ECMi did not influence endothelial cell proliferation.

Despite the infection associated chan es in the distribution of antisera binding, infection id not alter the ability of cholera toxin or GTPyS treatment to enhance antibody binding in either the 42 or 55 kDa region. Results identical to these were obtained using membranes prepared from infected endothelial cells grown on ECMi (not shown). DISCUSSION In this report, we document yet another example of the capacity of the extracellular matrix to direct biochemical characteristics of endothelial cells. Specifically, we have demonstrated that the ECM synthesized by endothelial cells infected with the arasite T. cruzi can direct uninfected endothelial ce s to synthesize aFGF-binding heparan sulfate proteoglycans and Gas guanine nucleotide-binding proteins biochemically similar or identical to those synthesized by infected endothelial cells. Whereas the role of the ECM in dictating phenotypic expression of cells has been extensively documented, t o our knowledge this has been restricted to cell differentiation. This report is one of the first to demonstrate an example of how the phenoRole of ECM in the expression of other t pic expression of endothelial cells can be altered by infection associated changes in endothelial t e ECM in the setting of infection. cell metabolism The actual biochemical basis by which the ECM In light of our documentation that other biochemical influences associated endothelial cells has yet to be characteristics of infection also appear to be independent of the magnitude of infection (Morris et al., 19891, we explored the possibility that these phenomena may also relate to the ECM. To this end, we examined the T infection associated changes in G protein biochemistry. 10 Infection of endothelial cells results in a decrease in the magnitude of cholera and pertussis toxin dependent 9ADP ribosylation (Morris et al., 1989), consistent with a decline in the magnitude of the Gs and Gi guanine anucleotide binding proteins. This influence of infection 7was observed whether 5 or 50% of the endothelial cells I d harbored the parasite. However, toxin catalyzed ADP 2 6 ribosylation reactions are frequently difficult to interX v, 5pret in light of the many variables that participate in J the reaction. To provide a more accurate assessment of i0i 4 the Gs protein moieties, we employed antisera raised to the G,, protein. In a Western blot analysis of mem3branes pre ared from uninfected endothelial cells 2grown on 8CMu (Fig. 4), we observed that antisera bound predominantly to the 42 kDa region, with minor Ibinding in a 55 and 68 kDa region. The intensity of antisera binding in all three regions increased when CONTROL oFGF aFGF HSPGi aFGF the membranes were previously incubated with cholera t t toxin or GTPyS. It is important to note that under HSPG, HSPG, HSPGi otherwise identical reaction conditions, antisera binding patterns obtained for membranes from uninfected Fig. 3. Potentiation of aFGFs mitogenic stimulation of sparse enendothelial cells grown on gelatin were identical to dothelial cell cultures by aFGF-binding HSPG Influence of infection. those shown here for uninfected endothelial cells grown Human endothelial cells were plated at a density of 2,500-5,000 on ECMu. These patterns of anti-Gas antisera binding cellsicm' in Medium 199 containing 20% fetal calf serum with or 14 pM aFGF and 50 nmoliml of aFGF-Affi-Gel affinity are to be contrasted with those obtained with mem- without [35S]HSPG(based on nmol of glucuronate) as described branes prepared from uninfected endothelial cells fractionated (Gordon et al., 1989). Amount (nmol) of HSPG was determined by a grown on ECMi (Fig. 4).Independent of whether the modified carbazole assay (Gordon et al., 1989).Cells were incubated in membranes were incubated in the presence or absence the appropriate medium for 4 days at 37"C, harvested with trypsin, and counted in a hemocytometer. Means of three separate determiof cholera toxin or GTPyS, we observed a marked nations are shown here; standard deviations are less than 15%of the decrease in the magnitude of antibody binding in the 42 mean, and for purposes of clarity are not presented. *Significantly kDa region. However, antisera binding to the 55 and 68 different from the combined mitogenic capacity of aFGF and HSPG kDa regions appeared to be markedly increased (Fig. 4). from uninfected cells at P