Global gene expression analysis reveals a role for the alpha 1 integrin in renal pathogenesis.

Kidney fibrosis is the hallmark of most types of progressive kidney disease, including the genetic disorder Alport's syndrome. We undertook gene expression analysis in Alport's syndrome mouse kidneys using microchip arrays to characterize the development of fibrosis. In addition to matrix and matrix-remodeling genes, consistent with interstitial fibrosis, macrophage-related genes show elevated expression levels in Alport's syndrome kidneys. Immunohistochemical analysis of kidney sections illustrated that macrophages as well as myofibroblasts accumulate in the tubular interstitium. Deletion of alpha(1) integrin results in decreased accumulation of both myofibroblasts and macrophages in the tubular interstitium in Alport's syndrome mice and delays disease progression. Transforming growth factor beta antagonism, although reducing interstitial fibrosis, does not limit macrophage accumulation in the tubular interstitium and disease progression. In this study, we identified previously overlooked inflammatory events that occur in the tubulointerstitial region. We propose that in addition to the previously suggested role for the alpha(1)beta(1) integrin in mesangial expansion and abnormal laminin deposition, this integrin may be critical for monocyte accumulation that, in turn, may lead directly to renal failure. Our gene expression and immunohistochemical data indicate that macrophage accumulation is dependent on alpha(1) integrin expression on the macrophage cell surface and that anti-alpha(1) integrin strategies may be employed as therapeutics in the treatment of chronic inflammatory and fibrotic diseases.

Most types of progressive renal disease are characterized by kidney failure associated with glomerular and interstitial fibrosis. Within the glomerulus, mesangial matrix expands, and there is increased deposition of extracellular matrix components, including collagen. Glomerular filtration and renal blood flow is increasingly impaired, tubules atrophy, and a general inflammatory response to the tissue damage ensues. The progressive loss of glomerular function terminates in tubulointerstitial fibrosis. The etiology of kidney fibrosis is many-fold, ranging from disease-associated secondary pathology, e.g. diabetic nephropathy, to autosomal mutations of basement membrane proteins.
One such hereditary basement membrane disorder is Alport's syndrome, a result of mutations in the collagen4A5, collagen4A3, or collagen4A4 genes (1)(2)(3). Loss of any one of these three genes results in the absence of all three proteins due to their mandatory association to form the quaternary structure of collagen. This disease affects about 1 in 5000 humans (4). There are several animal models for Alport's syndrome, one of which is a collagen 4A3 gene knockout in mouse (5,6). This model is very useful for studying both the tissue pathology of the disease as well as the underlying gene expression changes that may drive the disease. It has been demonstrated that at least two biochemical pathways are involved in pathogenesis. The first pathway is TGF-␤ 1 -stimulated and has been well established in many other disease models as well (7)(8)(9)(10)(11). The second pathway is ␣ 1 ␤ 1 integrin-dependent (12). The ␣ 1 ␤ 1 integrin is a major integrin receptor expressed on mesangial cells, and Cosgrove et al. (12) hypothesized that loss of ␣ 1 integrin expression on the mesangial cells resulted in the abnormal expression of specific laminin isoforms. Focusing on glomerular pathogenesis, these authors demonstrated that mesangial expansion and podocyte foot process effacement are attenuated in Alport's animals lacking the ␣ 1 ␤ 1 integrin. They suggested that the abnormal expression of laminin isoforms displaced proper podocyte foot process adhesion to the glomerular basement membrane (12).
To further characterize the mechanisms of disease progression and to include the tubulointerstitial regions in our analysis, we undertook an investigation of gene expression using microchip arrays (13)(14)(15). This approach permitted the quantitative analysis of the expression of ϳ11,000 genes and expressed sequence tags. Changes in expression were monitored at two different time points of the disease in Alport's syndrome (col4A3Ϫ/Ϫ) mice, in Alport's syndrome mice that were also null for the ␣ 1 integrin gene, and in Alport's syndrome mice treated with a TGF-␤ antagonist. Organization of the genes whose expression profile changes in Alport's syndrome mice into families of similar biological function has allowed us to analyze the effects of ␣ 1 ␤ 1 integrin on the progression of the disease and the relationship of the ␣ 1 ␤ 1 integrin pathway to that of TGF-␤. In this study, we identify previously overlooked inflammatory events that occur in the tubulointerstitial region and propose that in addition to the suggested role for the ␣ 1 ␤ 1 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 The abbreviations used are: TGF-␤, transforming growth factor ␤; TGF-␤R, TGF-␤ receptor; MMP-1, matrix metalloprotease-1; TIMP-1, tissue inhibitor of metalloprotease-1; PAI-1, plasminogen activator inhibitor-1; MCP-1, macrophage chemoattractant protein-1; PCR, polymerase chain reaction; MES, 4-morpholineethanesulfonic acid. integrin in mesangial expansion and abnormal laminin deposition, this integrin may be critical for monocyte accumulation that, in turn, may lead directly to renal failure.

EXPERIMENTAL PROCEDURES
Mice-The Alport's syndrome (col4A3 Ϫ/Ϫ) mice have been described previously (5). These mice are on a 129 Sv/J background. Heterozygotes for the col4A3 mutation were bred to obtain both homozygous knockouts and wild-type littermates. The genotype of the mice was determined by PCR. In addition, at the time of sacrifice, urine samples were collected and analyzed for protein that is indicative of kidney failure to confirm the knockout genotype. At the desired age (4 or 7 weeks) mice were sacrificed, and kidneys were dissected and frozen immediately in liquid nitrogen or as described below for tissue sections. The col4A3Ϫ/Ϫ; ␣ 1 Ϫ/Ϫ mice have also been described previously (12) and are on a 129 Sv background. Kidneys were collected as described above at 7 weeks of age.
TGF-␤R-Fc Treatment-Murine TGF-␤R-Fc (12) dissolved in sterile phosphate-buffered saline (1 mg/kg, 100 g/ml) was injected intraperitoneal into Alport's syndrome (col4A3Ϫ/Ϫ) mice beginning at 3 weeks of age. Injections were continued twice weekly until they reached 7 weeks of age, at which time they were sacrificed. Kidneys from each animal were dissected and frozen immediately in liquid nitrogen for RNA isolation or as described below for tissue sections. A separate group of control animals that were wild-type littermates (col4A3ϩ/ϩ) were treated in an identical fashion. Six animals were originally included in each experimental group. After confirmation of genotype and phenotype, kidneys from five animals in each group were used for further experiments.
Isolation of RNA from Kidneys-Frozen kidney tissue was homogenized in Trizol (Life Technologies, Inc.) at 4°C, and total RNA was extracted. The RNA was further purified with a Qiagen RNeasy kit according to the manufacturer's instructions (Valencia, CA). RNA from each individual kidney was purified separately.
Preparation of Labeled cRNA and Hybridization to Microarrays-Total RNA (5 g) was pooled from five of each mouse type, i.e. 25 g of total RNA was used for cDNA synthesis. Double-stranded cDNA was synthesized with an oligo(dT) 24 primer with a T7 RNA polymerase promoter site added to its 3Ј end (Genset, La Jolla, CA) and Superscrip-tII reverse transcriptase (Life Technologies). All of the cDNA obtained was subjected to in vitro transcription with biotinylated UTP and CTP (Enzo Diagnostics, Farmingdale, NY). A total of 40 g of the cRNA product was fragmented and combined with herring sperm DNA (0.1 mg/ml), control oligonucleotide B2 (50 pM), and four control bacterial and phage cRNA (1.5 pM BioB, 5 pM BioC, 25 pM BioD, and 100 pM Cre) samples in hybridization buffer (100 mM MES, 1 M NaCl, 20 mM EDTA, and 0.01% Tween 20). These samples served as internal controls for target grid alignment and hybridization efficiency, as recommended by the target manufacturer (Affymetrix, Santa Clara, CA). The cRNA mixtures were hybridized (200 l) at 45°C to a Mu11K GeneChip (Affymetrix) for 16 h. The same cRNA mixture was hybridized to the subA GeneChip and subB GeneChip on successive days. Each target array was washed and scanned (Hewlett-Packard, GeneArray scanner G2500A) according to the procedures outlined by the manufacturer.
Analysis of GeneChip Data-Expression data were initially analyzed with the GeneChip software (Affymetrix). Arrays were globally scaled to an average intensity of 2500. Each of the Alport's syndrome mouse chips (TGF-␤R-Fc-treated col4A3Ϫ/Ϫ, untreated col4A3Ϫ/Ϫ, or col4A3Ϫ/Ϫ; ␣ 1 Ϫ/Ϫ) and the wild-type TGF-␤R-Fc-treated mouse chip was analyzed using the age-matched wild-type mouse chip as the biological base line. A value of 100 was assigned to all intensity measurements below 100 before differences between knockout and base-line intensities were calculated. Three parameters were used for analysis, average difference intensity change, difference call, and fold change. Gene expression levels that varied less than 4-fold relative to the biological base line or had a difference call of "no change," as determined by the GeneChip algorithms, were considered unchanged. Further analysis was based on those genes whose expression levels changed between the Alport's syndrome (col4A3Ϫ/Ϫ) and the wild-type (col4A3ϩ/ϩ) mice. We chose all 420 genes that had an average intensity greater than 100 arbitrary units in either the knockout or the wild-type mice and whose intensity was at least 4-fold different between knockout and wild type. In this set of 420 genes, 23 genes were represented two or three times and gave similar expression results. We only further analyzed one representative of each gene, i.e. there were 393 unique genes whose expression levels changed. For example, collagen1A2 is represented by probe arrays Msa.2220.0 and X58251 on the Mu11K subB chip. At 7 weeks in the Alport's syndrome (col4A3Ϫ/Ϫ) mice versus the wild-type (col4A3ϩ/ϩ) mice, these probe arrays measured 31-and 20-fold increases in collagen 1A2 expression, respectively. Genes were further classified according to biological function using the established classification scheme of Adams et al. (16). They were clustered within categories using the GENE CLUSTER and TREEVIEW programs (17). These clusters were used to group subsets of genes whose expression were affected by the ␣ 1 deletion or TGF-␤R-Fc treatment.
Preparation of Tissue Samples and Immunohistochemical Labeling-Fresh kidneys were embedded in OCT (Tissue Tek) and frozen in liquid nitrogen. Cryostat sections (5 m) were cut and fixed in ice-cold acetone. After blocking in 3% bovine serum albumin, phosphate-buffered saline, sections were stained with the antibodies anti-␣-smooth muscle actin-Cy3, anti-CD11b-PE, rat anti-laminin ␤1 (Sigma; Pharmingen, San Diego, CA; Chemicon, Temecula, CA, respectively), and hamster anti-␣ 1 integrin Ha31/8, which was conjugated with Alexa-488 (18). The anti-laminin antibody was detected with fluorescein isothiocyanate-conjugated goat anti-rat Ig (Jackson Immunoresearch, West Grove, PA). Sections were mounted with MOWIOL (Calbiochem). Images were collected using the Openlab (Improvision, Lexington, MA) image analysis system interfaced with a Leica DMR fluorescence microscope equipped with a Hamamatsu Orca CCD camera.
Reverse Transcription-PCR-Total RNA isolated as described above was reverse-transcribed (Superscript II, Life Technologies). PCR was performed on cDNA from individual mice on a TC-200 thermocycler (MJ Research, Waltham, MA). PCR reactions included primers for the amplification of a 541-base fragment of glyceraldehyde-3-phosphate dehydrogenase and the genes of interest (TIMP-1, collagen 1A2, Rantes, and MCP-1). Samples were subjected to 27 rounds of PCR at an annealing temperature of 60°C. PCR products were separated in a 2% agarose gel by electrophoresis and imaged with SYBR Green I (Molecular Probes) staining on a Storm fluorescence imager (Molecular Dynamics, Sunnyvale, CA). The relative amounts of glyceraldehyde-3-phosphate dehydrogenase and the gene of interest were determined by integration of the PCR bands using ImageQuant software (Molecular Dynamics).

RESULTS
Alport's Syndrome Mice-To identify genes whose expression changed as a result of kidney disease, we compared the transcript levels in the Alport's syndrome (col4A3Ϫ/Ϫ) mice to their wild-type littermates at 4 and 7 weeks. RNA from five animals was pooled in order to minimize biological variation. We used 4-fold changes in gene expression level as a conservative cut-off for significant changes. We then categorized the 393 differentially expressed genes into established classes according to biological function (16), following the strategy of Stanton et al. (19) for studying gene expression in response to myocardial infarction. Among these differentially expressed genes, 80 are expressed sequence tags or of unknown function. A summary representation of the known 313 classified genes is shown in Fig. 1 (a complete report of expression data is available on-line in supplemental Table I).
At 4 weeks of age, we found that of the 393 differentially expressed genes, 8 genes were up-regulated in Alport's syndrome mice, and 3 were down-regulated (Fig. 1, first column). At 7 weeks of age, we found that of the 393, 234 genes were up-regulated in the Alport's syndrome mouse, and 151 were down-regulated (Fig. 1, second column). The significant difference in the number of genes whose expression changes between 4 and 7 weeks is consistent with the progression of kidney pathogenesis in the Alport's syndrome mice. Staining of kidney tissue for a marker of interstitial myofibroblasts, ␣-smooth muscle actin, demonstrates that between 4 and 7 weeks, myofibroblasts are accumulating in the tubulointerstitium (Figs. 2,  A and B). Thus, by 7 weeks, kidney pathology is advanced, and by 8.5 weeks, Alport's syndrome mice die of end-stage renal failure (12).
Consistent with interstitial fibrosis and mesangial expansion, many matrix proteins show elevated expression levels at 7 weeks (Fig. 1, classified as Cell Structure Motility). These include collagens 1, 2, 3, 6, 8, and 15, laminins ␣5 and ␤3, and  Table I on-line for quantitative fold-change data. EGF, epidermal growth factor; IL, interleukin; M-CSF, macrophage colony-stimulating factor; PDGF, platelet-derived growth factor; MAP, mitogen-activated protein; TRAF, TNF receptor-associated fibronectin. Remodeling genes typically associated with tissue fibrosis are also affected in the Alport's syndrome mouse at 7 weeks. For example, the protease and protease inhibitors MMP-1, TIMP-1 and PAI-1 and the growth factors endothelin-1 and insulin-like growth factors show large changes in gene expression ranging from 5 to 50-fold (Fig. 1, classified under Protein Expression and Cell Signaling and Communication, respectively), and the expression levels of oxidative enzymes are perturbed, e.g. lysyl oxidase and amiloride-sensitive amine oxidase (Fig. 1, Metabolism). Thus, the large changes in matrix gene expression observed with the Affymetrix microarray technology are consistent with the pathological changes in protein composition observed in kidney fibrosis.
In addition to matrix and matrix-remodeling genes, functional classification revealed two additional categories of genes that were heavily affected in the Alport's mouse, Metabolism (73 genes) and Cell/Organism Defense (70 genes) genes (Fig.  1). Both carbohydrate and lipid metabolism are affected, e.g. the gene expression levels of enolase, phosphoenolpyruvate carboxykinase, hydroxymethylglutaryl-CoA reductase, and galactocerebrosidase are decreased. The large number of genes in both categories suggests that the cellular composition of the tissue has changed significantly during the course of the disease.
The enormous up-regulation of Cell/Organism Defense genes at 7 weeks, the majority of which (83%, Fig. 1, column 2) is immune-response-related, suggests that immune system cells, e.g. macrophages or lymphocytes, appear in the Alport's syndrome kidney. We stained kidney tissue with cell lineage markers for monocytes/macrophages, granulocytes, and T-lymphocytes. At 4 weeks, we observed that a small number of macrophages had begun to appear in the kidney (Fig. 2E). At 7 weeks, an extensive population of macrophages (CD11b/Mac-1 ϩ ) was present in the tubular interstitium (Fig. 2F). Staining of wild-type controls indicated that few macrophages were detectable in healthy tissue (Fig. 3D). Infiltrating T-cells (CD3 ϩ ) or granulocytes (Gr-1 ϩ ) were not observed in the Alport's syndrome kidneys (data not shown). ␣ 1 Integrin in Alport's Syndrome-We next investigated the effect of ␣ 1 integrin deletion on the expression of those genes whose levels were changed in the 7-week-old Alport's syndrome mice. We analyzed gene expression levels in the col4A3Ϫ/Ϫ;-␣ 1 Ϫ/Ϫ mice using age-matched wild-type (col4A3ϩ/ϩ) mice as the biological base line. Within each functional category, the fold changes observed (Fig. 1, column 3) were compared with the fold changes determined in the 4-week-and 7-week-old Alport's syndrome mice (Fig. 1, columns 1 and 2). 75% of the 393 Alport's-related genes were maintained fully or partially at wild-type levels upon deletion of the ␣ 1 integrin from the Alport's syndrome mice. As can be seen in Fig. 1, the ␣ 1 -sensitive genes include those for matrix proteins, metalloproteases, protease inhibitors, and growth factors as well as metabolic and cell defense genes.
Matrix proteins including collagens 1, 2, 3, and 6, laminin ␣5 and ␤3, and fibronectin undergo large changes in transcript levels. For example, collagen 1A2 dropped from 20-fold higher than wild type in the 7-week-old Alport's syndrome mouse to 6-fold higher in the 7-week old col4A3Ϫ/Ϫ;-␣ 1 Ϫ/Ϫ mouse. Matrix remodeling was also affected significantly. Thus, ␣ 1 integrin deletion has a very global effect on matrix deposition. In addition to laminin ␣2 deposition, ␣ 1 integrin deletion inhibited the accumulation of collagens, fibronectin, and other laminin chains. TIMP-1 gene expression, which promotes extracellular matrix accumulation by inhibiting metalloprotease degradation of matrix in tissue, was reduced dramatically by ␣ 1 integrin deletion. Gene expression dropped from 54-fold in the 7-week-old Alport's syndrome mouse to 4-fold in the 7-week-old col4A3Ϫ/Ϫ;-␣ 1 Ϫ/Ϫ mouse. A similar change was observed for PAI-1, which inhibited tissue plasminogen activator and urokinase plasminogen activator, proteases necessary for activation of plasmin, although the amount of PAI-1 transcript in Alport's syndrome mice was not as high as that of TIMP-1. Plasmin is responsible for degradation of matrix proteins fibronectin and laminin as well as activation of collagenases and gelatinases. Thus, both the cascade for activation of metalloproteases as well as metalloprotease activities in tissue were inhibited in Alport's syndrome. Deletion of ␣ 1 integrin in Alport's syndrome reduced the levels of metalloprotease inhibitor mRNA transcripts to near wild-type levels, i.e. levels observed in a healthy kidney. This reduction in matrix and matrix remodeling genes is consistent with the decrease in myofibroblast accumulation in the interstitium observed in 7-week-old col4A3Ϫ/Ϫ;-␣ 1 Ϫ/Ϫ mice (Fig. 2C).
The monocyte/macrophage-derived genes also showed a decrease in transcript level upon ␣ 1 integrin deletion ( Fig. 1 and on-line in Table I (MCP-1), macrophage-inducible protein (IP-10), macrophage colony-stimulating factor, macrophage mannose receptor, and F4/80 were elevated 24-, 6-, 13-, 20.9-, and 11.9-fold respectively, in the 7-week-old Alport's syndrome mice. In the 7-weekold col4A3Ϫ/Ϫ; ␣ 1 Ϫ/Ϫ mice, their expression levels were maintained at wild-type levels. Furthermore, the transcript levels of Fc receptors, receptors present on the macrophage cell surface, are reduced from high (5-50-fold) levels to wild-type levels upon deletion of ␣ 1 integrin in Alport's syndrome mice. In addition, macrophage metalloelastase, and carboxypeptidase E proteases, which are secreted to clear tissue debris and facilitate migration of leukocytes into the tissue, have gene expression levels 19 -27-fold above the wild-type base line in Alport's syndrome mice. Again, they are reduced to wild-type levels by ␣ 1 integrin deletion. Consistent with the reduced expression levels observed, infiltration by macrophages is reduced to levels seen in 4-week-old Alport's syndrome mice (Fig. 2G). Thus, deletion of ␣ 1 integrin reduces the number or delays the appearance of both myofibrobasts and macrophages in the tubular interstitium in Alport's syndrome mice.
Antagonism of TGF-␤ in Alport's Syndrome-The expression levels of 22 genes known to be TGF-␤-inducible in other systems are restored to near wild-type levels by deletion of ␣ 1 integrin ( Fig. 1 and on-line in Table I). These genes encompass all functional categories, and include, in addition to the previously discussed matrix and matrix remodeling proteins, transcription factors, oxidative stress genes, and growth factors. Previously, it had been shown that, at 7 weeks of age, TGF-␤1 expression is induced 4-fold in Alport's syndrome mice relative to wild-type mice (20) and deletion of ␣ 1 integrin in Alport's syndrome mice prevented induction of TGF-␤ at 7 weeks of age. Because TGF-␤ promotes the progression of renal disease and fibrosis, we addressed how many of the gene expression changes associated with the loss of the ␣ 1 integrin were due to a loss of TGF-␤ activity. In addition, we wanted to elucidate the individual roles of TGF-␤ and ␣ 1 in this disease. Therefore, we investigated the effect of treatment with a murine soluble TGF-␤ type II receptor (TGF-␤R-Fc) on gene expression (12). This receptor is an antagonist for TGF-␤1 and TGF-␤3 but not TGF-␤2 (12). Alport's syndrome mice were treated with TGF-␤R-Fc twice weekly from 3 weeks of age until they were sacrificed at 7 weeks of age. Expression of 62% of the 393 Alport'srelated genes was restored or partially restored to wild-type levels upon treatment of Alport's syndrome mice with the TGF-␤R-Fc (Fig. 1, column 4). This subset included 18 of the 22 genes identified as TGF-␤-inducible ( Fig. 1 and on-line in Table  I). The expression levels of the other four TGF-␤-inducible genes, COL1A1, tenascin, fibronectin, and c-Jun, are slightly reduced or unchanged upon treatment with TGF-␤R-Fc. Furthermore, of the antagonist-affected genes, 95% are also affected by the ␣ 1 deletion. That is, only 11 genes that do not respond to removal of ␣ 1 integrin in the Alport's syndrome mice have altered expression levels upon treatment with TGF-␤R-Fc. The transcript levels of matrix and matrix-remodeling proteins, transcription factors, and most growth factors are reduced upon antagonist treatment. However, there are many macrophage-related genes whose expression levels respond to ␣ 1 deletion but not TGF-␤ antagonism. For example, the expression levels of IP-10, macrophage colony-stimulating factor, macrophage mannose receptor, metalloelastase, and carboxypeptidase E are all relatively unchanged upon treatment with the TGF-␤ antagonist (Fig. 1, column 4, and on-line in Table I).
When we looked for the presence of ␣-smooth muscle actin and CD11b-positive cells in the TGF-␤ antagonist-treated kidneys, we observed cell populations consistent with the transcript profile. In 7-week-old TGF-␤R-Fc-treated Alport's Syndrome mice, accumulation of myofibroblasts in the interstitium was significantly decreased (Fig. 2D). However, macrophages are still present (Fig. 2H). Thus, treatment with a TGF-␤ antagonist reduces accumulation in the tubular interstitium in Alport's syndrome mice of myofibroblasts but not macrophages.
Localization of ␣ 1 Integrin-Because deletion of ␣ 1 integrin was associated with a delay in macrophage appearance in Alport's syndrome mice, we investigated the localization of ␣ 1 integrin and macrophages over the time course of the disease. Kidney tissue from Alport's syndrome (col4A3Ϫ/Ϫ) mice was double-stained with anti-CD11b, a macrophage marker, and  E, and H), and 8-week (C, F, and I) Alport's syndrome mice are stained with anti-␣ 1 integrin (green; A (the arrow identifies ␣ 1 integrin expression on mesangial cells) B, and C), anti-CD11b (red; D, E (the arrow identifies accumulating macrophages), F); and both anti-␣ 1 integrin (green) and anti-CD11b (red); yellow reflects cells expressing both ␣ 1 integrin and CD11b (G, H, and I (the arrow identifies an example of macrophages expressing both ␣ 1 integrin and CD11b)). Images were acquired as in Fig. 2. anti-␣ 1 integrin antibodies. We examined sections at 1-week intervals from 4 to 8 weeks as well as a wild-type control. As previously reported, ␣ 1 integrin is present in healthy kidney on the mesangial cell surface (Fig. 3A). However, there are very few macrophages in the healthy kidney (Fig. 3D). In the Alport's syndrome mice, ␣ 1 integrin is also present on the mesangial cell surface (Fig. 3B). Beginning at ϳ4 weeks, CD11bpositive macrophages begin to accumulate in the tubulointerstitium (Fig. 3E) of Alport's syndrome mice. As the disease progresses, the number of macrophages accumulating progressively increases. By week 8, macrophages appear throughout the tubulointerstitium (Fig. 3F). Double-staining of the tissue with anti-␣ 1 integrin antibody demonstrates that a substantial subset of the accumulated macrophages are ␣ 1 integrin-positive (Figs. 3, H and I). Hence, the delay of macrophage accumulation in col4A3Ϫ/Ϫ; ␣ 1 Ϫ/Ϫ mice is associated with the absence of ␣ 1 integrin on the macrophage surface in these knockout mice. DISCUSSION Although the initial trigger or triggers for fibrosis are not known, the hallmarks of tissue fibrosis are well characterized. Cytokines promote growth of mesangial cells in the glomerulus and enhance release of extracellular matrix proteins such as laminin and collagen. The cytokines also stimulate accumulation of myofibroblasts in the tubulointerstitium. The myofibroblasts contain ␣-smooth muscle actin and are associated with accumulation of additional matrix proteins, e.g. collagen and fibronectin. In addition to a change in the quantity and composition of matrix proteins, remodeling of the matrix is affected in tissue fibrosis. In normal tissue, remodeling and turnover of the matrix proteins are balanced by metalloproteases and their inhibitors. In fibrotic tissue, increased levels of matrix metalloprotease inhibitors are observed, and the activities of matrix metalloproteases can be elevated or decreased (21). The net result is increased matrix accumulation in the tubulointerstitium.
Our gene expression profiling experiments demonstrate that these changes are associated with kidney fibrosis in an Alport's syndrome mouse model. In addition to genes for matrix and matrix-remodeling proteins, gene expression levels of immuneresponse genes are up-regulated. Tissue damage is normally associated with an inflammatory response that results in macrophage infiltration. Our immunohistochemistry results demonstrate, however, that the macrophage infiltration precedes accumulation of matrix. Moreover, both the gene profiling experiments and immunohistochemistry experiments clearly demonstrate that despite the reduction of interstitial fibrosis by TGF-␤ antagonism, macrophages accumulate in the renal tubular interstitium of Alport's syndrome mice. In contrast, ␣ 1 integrin deletion reduces both fibrosis and macrophage accumulation. In a recent publication (12), it was reported that loss of the ␣ 1 integrin significantly extended the lifetime of Alport's syndrome mice. On the other hand, TGF-␤ antagonism, although reducing glomerular and interstitial fibrosis, did not extend life. Because the ␣ 1 ␤ 1 integrin is one of the major integrins expressed on the mesangial cell surface and may be involved in both mesangial expansion and signaling, Cosgrove et al. (12) focused their analysis on the glomerulus. These authors hypothesized that loss of ␣ 1 integrin expression on mesangial cells may result in aberrant expression of laminin isoforms that, in turn, result in podocyte effacement from the glomerular basement membrane and a loss of renal filtration. The combined transcript profiles and immunohistochemical analyses presented here suggest an alternative hypothesis that accumulation of monocytes in the tubular interstitium may lead to direct renal failure. One reason that TGF-␤ receptortreated Alport's syndrome mice do not have an extended life-time despite delayed interstitial fibrosis is that macrophage accumulation in the kidney is minimally altered. In contrast, macrophage accumulation in the kidney of ␣ 1 integrin null Alport's syndrome mice is reduced. Tissue damage resulting from the pro-inflammatory cytokines may lead to death. Macrophages produce TGF-␤, and the marked decrease in the expression of TGF-␤-inducible genes in the ␣ 1 integrin null Alport's syndrome mice at 7 weeks is most likely due to the absence of macrophages in these mice.
Although our experiments are specific to the Alport's syndrome model, the conclusions are applicable to fibrotic pathology resulting from many forms of progressive renal disease. Indeed, similar transcript profiles were identified in a murine model of lung fibrosis (20). Although ACE inhibitors are known to delay the progression of renal disease to end stage renal failure, no current therapy directly targets fibrosis. Thus, development of a therapy that targets the fibrotic process of the disease itself would be of immense clinical significance. It has been shown previously that blockade of the ␣ 1 integrin mitigates the inflammatory response in murine models of contact sensitivity, delayed-type hypersensitivity, and arthritis (18). Our data in the Alport's model of renal pathogenesis demonstrate that the ␣ 1 integrin is required for accumulation of monocytes in damaged tissue. Although present at low levels on resting monocytes (22), expression of ␣ 1 integrin is upregulated in macrophages upon cytokine activation (23). The ␣ 1 integrin is involved in cell proliferation (24) and leukocyte activation (25). Thus, the absence of ␣ 1 integrin on the cell surface may either directly inhibit migration and/or proliferation of macrophages or disrupt integrin-mediated macrophage activation. The presence of an immune response in a variety of fibrotic diseases suggests that ␣ 1 integrin may play a role in other fibrotic diseases as well. Thus, monocyte accumulation may be regulated by ␣ 1 integrin, and therapeutics aimed at blocking the function of the ␣ 1 integrin may be useful for treating diseases associated with chronic inflammation and fibrosis.