Effects of Transforming Growth Factor-β1 on Renal Extracellular Matrix Components and Their Regulating Proteins

Abstract

Abstract. Transforming growth factor-β1 (TGF-β1) is widely regarded as a potent fibrogenic renal growth factor. In cell culture, TGF-β1 has been shown to increase various extracellular matrix (ECM) proteins and tissue inhibitors of metalloproteinases (TIMP), while decreasing matrix metalloproteinases (MMP), providing the optimum environment for progressive ECM accumulation. This study, which uses the isolated perfused rat kidney (IPRK), describes for the first time in a whole kidney preparation the action of TGF-β1 on factors associated with ECM processing. This model allows the study of the intact rat kidney with physiologic cell-cell interactions in the absence of confounding systemic influences. Left kidneys were removed from male Wistar rats by a nonischemic technique and perfused with a sterile, apyrogenic, endotoxin-free perfusate, based on the plasma volume expander Hemaccel (polygeline), at constant pressure in a recirculating IPRK system. Kidneys were perfused for 1 h either with (n = 3) or without (n = 3) recombinant human TGF-β1 (20 ng/ml). The effects of perfusion were controlled by comparison with the nonperfused contralateral kidney (n = 6). TGF-β1 was measured in the perfusate and urine, at the start and end of the experiment using an enzyme-linked immunosorbent assay to its biologically active form. After perfusion, sections of the kidneys were analyzed for changes in mRNA by Northern blotting. Significant increases in mRNA for fibronectin (7.5-fold, P < 0.01), heparan sulfate proteoglycan core protein (53-fold, P < 0.001), laminin β1 (12-fold, P < 0.001), collagen α1(IV) (17-fold, P < 0.001), collagen α1(III) (fourfold, P < 0.001), and MMP9 (twofold, P < 0.05) were observed after perfusion with TGF-β1. Measurement of TIMP1, TIMP2, TIMP3, MMP1, and MMP2 mRNA demonstrated no detectable change, whereas determination of mRNA for tissue transglutaminase, an enzyme capable of cross-linking many ECM components, showed an eightfold increase (P < 0.01). This study suggests that in the IPRK and in the absence of other exogenous growth factors, TGF-β1 selectively increases the synthesis of ECM and tissue transglutaminase without changes that would result in the reduction of ECM degradation.

The progression of chronic renal insufficiency is associated with progressive scarring and fibrosis of the kidneys. More specifically, renal fibrosis is associated with glomerulosclerosis and tubulointerstitial fibrosis. The mechanisms and mediators of such fibrotic processes have been the subject of ongoing investigations in recent years. A role has been attributed to a wide range of systemic hormones, autacoids, and growth-promoting peptides (growth factors) (¹). Of the latter, transforming growth factor-β1 (TGF-β1) is thought to be one of the most potent renal fibrogenic growth factors (²). It is expressed within the scarred kidneys of experimental animals (³) and humans (⁴). When the TGF-β1 gene is transfected into rat glomeruli, it leads to progressive glomerulosclerosis (⁵). Furthermore, mice transgenic to TGF-β1 develop spontaneous glomerulosclerosis (⁶). The fibrogenic effect of TGF-β is thought to be due to its stimulation of the synthesis of extracellular matrix (ECM) and the inhibition of its breakdown (⁷). In tissue culture, TGF-β stimulates the synthesis by glomerular cells of collagens, laminin, and fibronectin (²). The fibrogenic effect of TGF-β is thought to be enhanced by its capacity to inhibit the synthesis of matrix-breaking metalloproteinases (MMP) and stimulate their inhibitors (tissue inhibitors of metalloproteinases [TIMP]) (⁷).

TGF-β is also capable of modulating the activity of tissue transglutaminase (tTgase), an enzyme known to stabilize ECM and possibly prevent its breakdown (⁸). This enzyme crosslinks proteins through ε-(γ-glutamyl) lysine bonds, which are resistant to enzymatic degradation (⁹). A role has been suggested for this enzyme in the pathogenesis of fibrotic diseases (¹⁰). We have recently observed an increase in tissue tTgase in scarred kidneys and postulated a link between this enzyme, TGF-β, and renal fibrosis (¹¹).

Recent attention has focused on the context of growth factor action, including that of TGF-β1 (¹²). Transforming growth factor-β1 has variable and opposing effects depending on the local microenvironment (¹²). The effect of TGF-β1 on the synthesis of ECM by glomerular cells in culture depends to a large extent on the cell type; it stimulates synthesis of collagen IV by mesangial cells, and it stimulates synthesis of fibronectin by epithelial cells (¹³). In other studies based on human mesangial cells in culture, synergism has been observed between the effects of TGF-β1 and insulin-like growth factor-I (IGF-I) on collagen, fibronectin, and laminin synthesis (¹⁴).

We decided to investigate the direct effect of TGF-β1 on the rat kidney's synthesis of components of ECM and their regulating enzymes. For this we have used the isolated perfused rat kidney (IPRK), thus allowing us to test the direct and isolated effect of TGF-β1 on the rat kidney in the absence of confounding systemic factors. The IPRK also allows for interactions between various renal cells and therefore avoids the artificiality of testing a single cell line in isolation. It also tests the kidney with its own ECM support, thus avoiding the phenotypic changes induced by cell culture on plastic or artificial matrix substitutes. We have previously used this approach successfully to test the renal hemodynamic effects of amino acids (¹⁵), radiocontrast material (¹⁶), and growth factors (¹⁷). Others have applied it successfully to test and localize the effects of IGF-I on renal proteoglycan synthesis (¹⁸).

Materials and Methods

Isolated Rat Kidney Perfusion

Male Wistar rats (Sheffield University), 400 to 475 g, previously allowed free access to food and water were anesthetized with thiopentone (100 mg/kg body wt), and the right kidney was removed by a nonischemic technique. After an intravenous injection of 500 U/kg heparin, the right ureter was cannulated for the collection of urine and the right adrenal artery was tied. The right renal artery was cannulated via the superior mesenteric artery and perfusion was started in situ. The right kidney was then transferred to a recirculating perfusion system (Figure 1) and perfused ex vivo (¹⁵). All procedures were carried out under license according to regulations prepared Her Majesty's Government, United Kingdom (Animals Scientific Procedures Act 1986).

Figure 1: . Schematic representation of the isolated perfused rat kidney apparatus.

To obtain a perfusion system that tested limulus-negative for endotoxin at the end of the 1-h perfusion period, the circuit was dismantled after each kidney perfusion. Glassware was soaked in 3N nitric acid, and all plastics including the circuit tubing were sonicated in a weak detergent solution (5% decon). Both glassware and plastics were thoroughly rinsed in pyrogen-free water, autoclaved, and dried.

Perfusate Preparation

A perfusate with minimal endotoxin contamination was prepared before each experiment by the addition of 1.5 ml of 0.2 M MgSO₄, 3.6 ml of 1 M NaHCO₃, and 6.0 ml of 5% glucose to 150 ml of Hemaccel, a commercial plasma volume expander that is both sterile and apyrogenic before use. At the start of each experiment, the perfusate composition was (in mM/L): 155 Na²⁺, 135 Cl^-, 1.8 Mg²⁺, 1.8 SO₄^2-, 22 HCO₃^-, 2.37 K⁺, 5.8 Ca²⁺, and 10.3 glucose, together with polygeline (MW 35,000) 3.26%. Ultrafiltration dialysis using Centricon 3000 partition cups (Amicon) showed calcium to be extensively bound to polygeline (75.4 ± 5.4%). The pH of the perfusion solution at 37°C was 7.4 when gassed with 95% O₂/5% CO₂.

Experimental Protocol

Groups of three kidneys were used in the perfusions, and a group of six kidneys was used as nonperfused controls. Kidneys were perfused at constant pressure (100 mmHg) for a 60-min period either with or without the addition of 20 ng/ml TGF-β1 (Life Technologies BRL, Paisley, United Kingdom). This dose was selected based on tissue culture studies (^19,20,21,22) and subcutaneous in vivo injection of TGF-β1 that results in fibrosis (²³). Perfusion pressure was measured from a point inside the renal artery and maintained by a servo-controlled peristaltic pump (Watson-Marlow 503U; Watson-Marlow, Cornwall, United Kingdom). Renal perfusate flow (RPF) was continually monitored. GFR was assessed from the clearance of ¹⁴C inulin after the addition of 1 μCi ¹⁴C inulin to the perfusate. Urine and perfusate samples were collected at 5-min intervals, and ¹⁴C levels were determined by liquid scintillation counting. The renal clearance of ¹⁴C inulin was calculated as the ¹⁴C urinary excretion rate divided by the ¹⁴C-perfusate concentration. Urine flow was measured gravimetrically. Values of inulin clearance, RPF, and urine flow were expressed in ml/min per gram wet weight of the nonperfused kidney. The filtration fraction was calculated by dividing GFR by RPF. At the end of the perfusion period, the kidney was removed from the perfusion circuit and decapsulated, and three samples of 100- to 200-mg transverse sections were taken from areas adjacent to, but not including, the papilla such that they contained equal amounts of medullary and cortical tissue. Sections were snap-frozen and stored in liquid nitrogen for RNA analysis. Samples of perfusate were analyzed at the start and end of the perfusion period for endotoxin contamination and TGF-β1 levels. Urine was collected through a ureteral catheter for measurement of urinary TGF-β1. The effect of perfusion was assessed by comparison with the nonperfused contralateral kidneys.

RNA Analysis

Total RNA was extracted using Trizol™ (Life Technologies BRL) and quantified by spectrophotometry. Thirty milligrams of total RNA was electrophoresed on a 1.2% (wt/vol) agarose/4-morpholinepropanesulfonic acid/formaldehyde gel and viewed under ultraviolet (UV) light to verify loading and RNA integrity by the presence of intact ribosomal bands. RNA was then transferred to Hybond N nylon membranes (Amersham, Buckinghamshire, United Kingdom) by capillary blotting with 20 × SSC and fixed by UV irradiation (70 mJ/cm² using the UV cross-linker (Amersham). Filters were prehybridized in a mixture containing 50% (vol/vol) deionized formamide, 5× saline-sodium phosphate-ethylenediaminetetra-acetic acid, 5× Denhardt's reagent, 1% (wt/vol) SDS, and 200 mg/ml sonicated and denatured salmon sperm DNA at a probe-specific temperature for 1 h. Hybridization was performed under the same conditions with the addition of labeled probe to 1 × 10⁶ cpm/ml for 18 h. Filters were then washed to a stringency of 0.2%/0.2× saline-sodium phosphate-ethylenediaminetetra-acetic acid at temperatures up to 65°C (depending on the probe) for 1 h and then exposed to Kodak XOMAT AR film for up to 7 d with intensifying screens. Developed films were subsequently analyzed by scanning volume density analysis using the Bio-Rad GS-690 scanning densitometer (Hertfordshire, United Kingdom) and Molecular Analyst version 4. Loading was corrected by reference to the optical density of ethidium bromide-stained gels of 18S rRNA and the housekeeping genes cyclophilin and β-actin. Multiple use of housekeeping genes and ribosomal RNA was used to guard against TGF-β1 regulation. Determination of transcript size was by reference to RNA molecular weight markers (Promega, Madison, WI) and calculated using Molecular Analyst software. This was confirmed by visual comparison to the ribosomal RNA subunits.

cDNA Probes

Specific random-primed DNA probes were labeled from the following DNA sequences: rat fibronectin (²⁴), human laminin β1 (²⁵), rat collagen (α1) III (²⁶), human collagen (α1) IV (²⁷), mouse heparin sulfate proteoglycan core protein (²⁸), rat TIMP1 (²⁹), rat TIMP2 (³⁰), human TIMP3 (³¹), rat gelatinase A (MMP2) (³²), rat gelatinase B (MMP9), rat interstitial collagenase (MMP1) (³³), and mouse tTgase (³⁴). With the exception of MMP9, each cDNA was kindly supplied in a suitable vector by the relevant author. These plasmids were then used to transform bacteria (Escherichia coli strain DH5α), which were then used to amplify the plasmid. To produce an MMP9 cDNA, total RNA was reverse-transcribed from rat kidney and amplified by PCR using the primers CTCACCAGCGCCAGCCGACTTAT and GGCACTGAGGAATGATCTAA designed from published rat MMP9 sequences (³⁵). The 1.16-kb product was transferred into plasmid pCR 2.1 using the Invitrogen TA cloning kit (RD Systems, London, United Kingdom). Confirmation of the product was undertaken by restriction mapping and partial sequencing.

cDNA fragments were excised from the host vector by restriction enzyme digests, separated by electrophoresis on a 1% (wt/vol) agarose TAE gel, and purified using Bandprep™ (Pharmacia-Biotech, Uppsala, Sweden). Twenty-five nanograms of purified cDNA was random-primed with ³²P-labeled dCTP (Redivue, Amersham) using the Prime-a-Gene system (Promega). Unincorporated nucleotides were removed using Sephadex G50 Nick Columns™ (Pharmacia), and the random-primed cDNA probe was denatured before addition to the hybridization solution.

TGF-β1 Enzyme-Linked Immunosorbent Assay

Total and biologically active TGF-β1 antigen levels were measured in perfusate and urine using a TGF-β1 sandwich enzyme-linked immunosorbent assay system (Promega). The monoclonal primary antibody used was specific for biologically active TGF-β1, and total TGF-β1 was determined by activation of the latent form by acid hydrolysis at pH 2.3 according to the manufacturer's instructions.

Endotoxin Measurements

Preliminary investigations (data not shown) demonstrated that endotoxin contamination above 0.3 endotoxin units/ml caused transcriptional induction of fibrogenic growth factors (TGF-β1, platelet-derived growth factor, and IGF-I), TIMP, MMP, and ECM proteins (collagens III and IV, laminin, and fibronectin) within 15 min of perfusion. We were therefore careful to avoid endotoxin contamination of our IPRK system. To verify that perfusions were performed in a system with minimal endotoxin contamination, endotoxin levels in the perfusate were measured using the E-toxate system (Sigma Chemicals, Poole, United Kingdom) according to the manufacturer's instructions.

Statistical Analyses

Results are given as mean ± SEM. Evaluation of the significance of the difference between groups was undertaken by t test. A P value <0.05 was considered significant.

Results

Renal Function

In this polygeline perfusate-based system, renal perfusate flow was 23.6 ± 0.9 ml/min per g, [¹⁴C] inulin clearance was 0.41 ± 0.08 ml/min per g, giving a filtration fraction of 1.6 ± 0.3%. The addition of TGF-β1 did not significantly alter any of these measurements.

TGF-β1 Levels in the Perfusate

Measurement of biologically active TGF-β1 in the perfusate showed no detectable levels in any of the control perfusions. In the TGF-β1 perfusions, the detectable level of biologically active TGF-β1 was 4.0 ± 0.9 ng/ml at the start of the perfusion. There was still 88 ± 16% of the original TGF-β1 levels at the end of the 1-h perfusion period (3.2 ± 0.3 ng/ml). Determination of total TGF-β1 (data not shown) revealed no significant difference from the biologically active form.

Urine measurements of biologically active TGF-β1 showed no detectable levels in the control perfusions. In the TGF-β1 perfusions, detectable levels of the growth factor were not generally present in the urine initially (first 5-min perfusion), but by the end of the perfusion period (60 min), TGF-β1 was detectable (0.18 ± 0.01 ng/ml urine).

Effect of TGF-β1 on ECM mRNA

Perfusing kidneys for 1 h at a constant pressure in the absence of added growth factor did not cause significant changes in mRNA for collagen (α1) III, collagen (α1) IV, fibronectin, or heparan sulfate proteoglycan (HSPG) core protein when compared with the contralateral nonperfused kidney (Figures 2 and 3). Laminin β1 mRNA, however, was significantly decreased by approximately 2.5-fold (P < 0.05). A time course investigation revealed that this decrease was not progressive and simply an adjustment to the basal perfusion conditions, as laminin β1 mRNA reduction occurred within 15 min and remained stable for at least 3 h (data not shown).

Figure 2: . Autoradiographs from Northern blot analysis for mRNA expression of collagen (α1) III, collagen (α1), fibronectin, laminin β1, heparan sulfate proteoglycan (HSPG) core protein, and tissue transglutaminase in nonperfused kidneys (n = 6) and kidneys perfused with (n = 3) and without (n = 3) transforming growth factor-β1 (TGF-β1).

Figure 3: . Volume density analysis of the autoradiographs shown in Figure 2. Histogram data represent mean volume density (± SEM) corrected for loading using cyclophilin mRNA density from three perfusions and six nonperfused kidneys.

The addition of recombinant human TGF-β1 to the perfusate had a considerable effect on the mRNA levels of all ECM protein mRNA that were measured when compared against the control perfusion (Figure 2). The largest increase as a result of TGF-β1 perfusion was on HSPG core protein mRNA, which demonstrated a 53-fold increase (P < 0.001) over control, although caution should be used when interpreting this increase due to the threshold and saturation levels of the autoradiographic film. The HSPG probe used recognizes all HSPG; however, due to the predominance of perlecan in renal tissue and the single band recognition of the HSPG probe, it is likely that we are specifically measuring perlecan mRNA. Collagen α(IV) and laminin β1 chain showed similar increases in mRNA levels of 17-fold (P < 0.001) and 12-fold (P < 0.001), respectively. The smallest increases were seen in fibronectin (7.5-fold, P < 0.01) and collagen α1(III) (fourfold, P < 0.001).

Effect of TGF-β1 on ECM-Regulating Enzyme mRNA

Messenger RNA for interstitial collagenase (MMP1), gelatinase A (MMP2), gelatinase B (MMP9), TIMP1, TIMP2, and TIMP3 remained unchanged on perfusion when compared with the contralateral nonperfused kidney (Figures 4 and 5). Interestingly, tTgase mRNA was significantly decreased by twofold (P < 0.001) on perfusion (Figures 2 and 3), but again this was a nonprogressive basal adjustment to the IPRK as determined in the same manner as for laminin β1 above.

Figure 4: . Autoradiographs from Northern blot analysis for mRNA expression of interstitial collagenase (MMP1), gelatinase A (MMP2), gelatinase B (MMP9), TIMP1, TIMP2, and TIMP3 in nonperfused kidneys (n = 6) and kidneys perfused with (n = 3) and without (n = 3) TGF-β1.

Figure 5: . Volume density analysis of the autoradiographs shown in Figure 4. Histogram data represent mean volume density (± SEM) corrected for loading using cyclophilin mRNA density from three perfusions and six nonperfused kidneys.

In complete contrast to the action of TGF-β1 on ECM mRNA in the IPRK, TGF-β1 had no significant effect on the mRNA levels of the matrix metalloproteinases MMP1, MMP2, or TIMP 1, 2, and 3 when compared with the control perfusion (Figures 4 and 5). However, TGF-β1 did cause a slight 2.4-fold increase in MMP9 (P < 0.05). On the other hand, TGF-β1 showed a marked effect on tTgase mRNA with more than an eightfold increase over the control (P < 0.01) (Figures 2 and 3).

Control of mRNA Loading

It was not possible to use β-actin mRNA as a loading control because it showed regulation under TGF-β1 when compared against the 18S rRNA subunit as determined under UV light. Comparison of cyclophilin mRNA with the 18S rRNA subunit showed no TGF-β1 regulation of cyclophilin mRNA. There was no statistical difference (paired t test) between correction values calculated from 18S rRNA and cyclophilin mRNA (Figures 2 and 4). Cyclophilin mRNA was used as the loading control in this study because it was believed to be more representative: It was measured after blotting, hybridization, and using the same autoradiograph endpoint/measurement as the mRNA under investigation.

Discussion

In this investigation, we have been able to study the overall action of TGF-β1 in an intact organ, thus eliminating phenotypic and biochemical perturbations associated with isolated cell culture and without the potentially secondary or synergistic confounding effects of in vivo administration. The IPRK has been used extensively to study kidney physiology (³⁶). The model has been criticized because studies on an isolated organ deprived of blood may have their limitations. We have previously established the viability and functional stability of the model in our hands (^16,17). For instance, the IPRK displays a physiologic response to both vasoconstrictors and vasodilators (^16,17). The IPRK has also been used as a tool to investigate changes in ECM in response to potentially fibrogenic stimuli. Our group has previously shown that the infusion of IGF-I into an IPRK stimulates the transcription of ECM proteins (³⁷). Similarly, IGF-I was shown to induce the deposition of glomerular HSPG in the IPRK (¹⁸).

In this model, special care was taken to avoid contamination of the IPRK by endotoxin because pilot work (T. S. Johnson and J. L. Haylor, unpublished observations) showed that such a contamination considerably increases the baseline levels of mRNA for ECM proteins, growth factors, and cytokines. Others have also shown the effects of endotoxin on cytokines and chemoattractants in the IPRK (³⁸). We have therefore used a perfusate based on the volume expander Hemaccel (containing polygeline and extremely low levels of endotoxin contamination) and used extensive cleaning and sterilization of the perfusion apparatus to overcome this problem. The use of Hemaccel, although beneficial in terms of not introducing toxins, does tend to reduce the filtration fraction, as the polygeline is partially filtered. However, the IPRK in this format had similar amounts of ECM mRNA compared with the contralateral nonperfused kidney, thus allowing us to evaluate its response to stimuli on the synthesis of ECM, MMP, and TIMP.

In the IPRK, we have demonstrated findings on the effects of TGF-β1 on ECM synthesis similar to those reported in in vitro and in vivo studies. All ECM protein mRNA that we tested were induced by TGF-β1. Of particular interest was the fact that the ECM proteins involved in basement membrane formation (HSPG core protein, laminin β1, collagen (α1) IV, fibronectin) increased more markedly than the predominantly interstitial ECM proteins (collagen (α1) III), perhaps indicating a selective action of TGF-β1 toward basement membrane components. It is important to note that one of the major interstitial matrix components, collagen α1 (III), is induced slightly in this perfusion model, and therefore the action of TGF-β1 is from an elevated basal state that may affect the level of increase seen.

Of great interest was the finding that tTgase mRNA expression was induced more than eightfold by TGF-β1. We have recently described perturbations in tTgase expression in the subtotal nephrectomy model of renal scarring in rats (¹¹). This is particularly relevant because this enzyme has been ascribed a role in ECM processing in fibrotic disease (¹⁰) possibly by the stabilization of the fibril structure by incorporation of the ε-(γ-glutamyl) lysine cross-links (³⁹) that are resistant to attack by any known mammalian enzyme (⁹). Because tTgase has been demonstrated to cause collagen fibril cross-linking, even in the absence of lysyl oxidase, then induction of the enzyme by TGF-β1 could play an important role in the pathogenesis of tissue fibrosis. This would attribute an additional fibrogenic potential to TGF-β1 through the stabilization and resistance to breakdown of deposited ECM proteins. Moreover, because tTgase is an important element in activation of the latent TGF-β1-LAP complex (⁴⁰), this could provide a positive feedback loop for TGF-β1 activation. Qualitative changes to the ECM as potentially caused by tTgase have received little attention thus far. This is surprising in view of the important consequences of changes in matrix structure if such changes interfered with ECM breakdown by MMP in tissue and renal fibrogenesis.

In contrast to the majority of the published literature was the finding that TGF-β1 failed to upregulate TIMP1, TIMP2, and TIMP3, while also failing to alter MMP1 and MMP2 expression. However, the increase in MMP9, although small, was significant and is in keeping with recently published in vitro data suggesting that TGF-β1 can induce MMP9 in glomerular epithelial cells (⁴¹). This specific MMP9 induction, rather than inhibition (⁴²), would seem to indicate that TGF-β1 may play a role in selective remodeling of ECM rather than wholesale inhibition of breakdown pathways. Although there is in vitro evidence to support the fact that MMP2 is either induced (²²) or not affected (⁴³) by TGF-β1 alone, the remainder of these findings conflict with recognized TGF-β1 action. However, there are in vitro examples of TGF-β1 alone being unable to alter MMP and TIMP levels (¹⁹). One possible explanation of the failure of MMP and TIMP to respond to TGF-β1 may be due to differential dose—response to TGF-β1. Although we added recombinant human TGF-β1 to a concentration of 20 ng/ml biologically active compound (manufacturer's measurement), the measurable level (enzyme-linked immunosorbent assay) averaged 4.0 ng/ml. It is probable that this loss of free TGF-β1 is due to the hydrophobic nature of this protein, which encourages it to bind to glass and plastic. Because the glass perfusion columns are siliconized, such binding is most likely to have occurred to the disposable plastic cannulas and tubing used in the apparatus, although binding to a perfusate component cannot be ruled out. Therefore, it is possible that TGF-β1 may show more efficacy toward ECM induction at a lower dose compared to MMP and TIMP regulation. This, however, is unlikely because in vitro experimentation has shown induction of ECM and TIMP and suppression of MMP at the same doses of TGF-β1, and these respond in a dose-dependent manner to as little as 1 ng/ml of TGF-β1 (²¹). The discrepancy may simply be that the whole kidney differs in its response to that of isolated cells in culture.

Tissue culture conditions may not provide the best means of studying the effects of TGF-β1 on renal cells. It has been clearly demonstrated that there is growth factor signaling between different cells of the kidney (⁴⁴). This is never truly replicated in culture. In addition, there is now clear evidence that proteins involved in the structure of the ECM (collagen I and fibronectin), its turnover (TIMP1, MMP1), and regulation (TGF-β1) are affected by the disorganization of the cytoskeleton (⁴⁵) such as that occurring during cell culture passage and trypsinization. It is therefore likely that renal cells growing in culture also exhibit changes in the cytoskeleton, which may account for the matrix accumulating changes in TIMP, MMP, and ECM that are exhibited with increasing passage number by mesangial cells (⁴⁶). The behavior and synthetic capacity of cells grown in culture depends on the substrate on which these cells are grown (⁴⁷). Furthermore, one must also consider that most in vitro studies of TGF-β1 action on ECM regulatory components are performed in serum-free conditions, with tightly controlled levels of other biologically active molecules, a situation significantly different from that occurring in vivo. In the IPRK, due to the maintenance of organ structure, there is inevitably growth factor signaling between cells. All of these factors point toward cells in an intact organ having a higher biologic activity than in single cell culture. It may be that this results in the accelerated mRNA kinetics in response to TGF-β1 seen here compared with single cell culture, with detectable (but not maximal) responses to TGF-β1 within 60 min.

Interactions between TGF-β and other growth factors may modulate its effects on ECM components and their regulatory enzymes. In human fibroblasts, the expression of TIMP1 or MMP1 and MMP3 was unaffected by TGF-β1 alone (¹⁹). In the same cell line, endothelial-derived factor and fibroblast growth factor induction of MMP1, but not MMP3, was blocked by TGF-β1, whereas TIMP1 expression was synergistically increased (¹⁹). It was also demonstrated that TGF-β1 and interleukin-1β in combination could stimulate TIMP secretion in a number of fibroblast and synovial cell lines, whereas TGF-β1 alone had less effect (²⁰). Furthermore, data suggest that interleukin-1β-induced pro-collagenase secretion could be blocked by TGF-β1 (²⁰). There is also evidence to show that TGF-β1 will interact with retinoic acid synergistically to increase TIMP production in human skin and synovial fibroblasts (⁴⁸). These observations and others highlight the concept that the biologic response produced to TGF-β1 is a function of cell type and cell state together with the presence or absence of other growth factors (¹²).

This study in the IPRK therefore questions some of the in vitro findings on the action of TGF-β1 on MMP and TIMP. Although it is possible that these rapid changes in ECM protein mRNA and transglutaminase mRNA may be a consequence of hemodynamic changes in the kidney due to TGF-β1's ability to cause vasodilation by inhibiting nitric oxide synthase (⁴⁹), the lack of any changes in RPF, GFR, and the filtration fraction suggest that this must be a direct action of TGF-β1 of matrix gene transcription as reported in vitro. Thus, the question remains as to why the in vitro action of TGF-β1 on MMP1/2 and TIMP is not repeated here, and whether the true in vivo effects of TGF-β1 are due to TGF-β1 alone, or as a result of synergism with other growth factors. Synergistic regulation of TGF-β1 would provide auxiliary means of regulation for the potentially powerful effects of TGF-β1 on matrix remodeling, ensuring tight finite control on both ECM deposition and breakdown pathways. This study suggests that in the intact kidney, the fibrogenic effects of TGF-β1 alone by in vivo transfection (⁵), injection (²³), or in transgenic animals (⁶) may be the consequence of changes in the anabolic side of the ECM metabolism rather than the catabolic side. Determination of the actions and synergistic interplay of growth factors in intact functioning organs is paramount if effective therapies to prevent renal scarring are to be developed. Our data also suggest that the IPRK provides a useful ex vivo tool to investigate the effects of growth factors on kidney physiology and pathology.

Acknowledgments

The authors are grateful for the financial support of The Sheffield Kidney Research Foundation, The Sheffield Kidney Patients Association, and The Northern General Hospital Trust Research Committee.

The authors would also like to thank the following for supplying cDNA sequences: Dr Ray Boot Handford, School of Biological Sciences, Manchester University; Dr MJ Arthur, School of Medicine, University of Southampton; Professor Martin Griffin, The Nottingham Trent University; Dr Richard Hynes, Center for Cancer Research, Massachusetts Institute of Technology; Dr Markku Kurkinen, Dept of Medicine, UMDNJ Robert Wood Johnson Medical School; Dr DJ Prockop, Jefferson Institute of Molecular Medicine, Jefferson Medical Collage; Professor Karl Trygvasson, Biocenter and Department of Biochemistry, University of Oulu; Dr D Bergmsa, SKB Pharmaceuticals, USA; Professor CO Quinn, Pediatric Research Institute, St Louis University School of Medicine; Professor NC Partridge, Department pharmacologic and Physiological Sciences St Louis University Medical Center; Dr Carlos Lopez - Otin, Faculty of Medicine Oviedo University, Spain