This study is the first to demonstrate the extensive involvement of proteoglycans in tissue remodeling in experimental CTD, demonstrating marked changes in proteoglycan expression in the intrarenal arteries, glomeruli and interstitium. The following key observations were made. First, whereas HS proteoglycans dominate in neointimal lesions in TV and in FGS, the CS/DS proteoglycan versican dominates in IF. Second, glomerular remodeling is associated with an impressive induction of perlecan expression in the glomerular BM. Third, the HS proteoglycan content becomes increased in cortical tubular BMs, especially due to increased collXVIII expression. Finally, allografts are characterized by marked tubulointerstitial lymphangiogenesis which correlates with IF development and impaired graft function.
Some earlier work demonstrated increased HS polysaccharides in fibrotic and sclerotic lesions in vessels, interstitium and mesangium of chronic renal transplant dysfunction 
, along with increased GAG-mediated chemokine binding 
. However, proteoglycan core-proteins were not identified in those studies. In non-transplant renal diseases, tubular upregulation of collXVIII/endostatin was reported in a number of experimental models 
. Mesangial expression of perlecan and agrin was reported in human diffuse mesangial sclerosis 
, in diabetic nephropathy 
, and some other human glomerulopathies with mesangial expansion 
. Concerning proteoglycan expression in the neointima, both perlecan and versican have been shown to be present in neointimal lesions formed after experimental or human stenting or denudation or related to atherosclerosis 
. The neointima in arteries of human cardiac allografts contain versican 
. The striking similarities in proteoglycan expression in transplantation-unrelated kidney diseases and chronic renal allograft dysfunction suggest comparable matrix remodeling programs. This might indicate that anti-fibrotic treatments in various kidney diseases might also reduce chronic transplant dysfunction.
We showed differential expression of HS proteoglycans in neointimal lesions and FGS on one hand, and CS/DS proteoglycans in IF on the other, suggesting the existence of spatial (i.e.
compartment-specific) proteoglycan responses during the development of CTD with potentially variable biological effects. Expression of the CS/DS proteoglycan versican in IF is likely involved in leukocyte recruitment and infiltration by its L-selectin-binding capacity 
. The abundant versican expression in IF supports our previous finding that L-selectin in the interstitium binds to CS/DS side chains and not HS side chains 
. Moreover, the high L-selectin-binding capacity of CS/DS proteoglycans in IF fits well with our observation that most leukocyte infiltration was observed in interstitial regions and to a far lesser extent in neointimal lesions and within the glomeruli. The marked expression of versican in IF is probably produced by interstitial myofibroblasts 
. Tubulointerstitial versican might contribute to the activation and proliferation of intra- and extrarenal myofibroblasts and may also mediate their recruitment. In line with this, we recently demonstrated that ~53% of interstitial myofibroblasts in IF are derived from extrarenal sources 
and may originate from a population of recirculating fibrocytes. Fibrocytes are mesenchymal progenitor cells exhibiting morphological characteristics of hematopoietic stem cells, monocytes and fibroblasts and have the capacity to differentiate into α-SMA-expressing myofibroblasts which is promoted by TGF-ß 
. Although HS proteoglycans have been shown to mediate hematopoietic progenitor cell homing 
this needs to be experimentally proven for CS/DS proteoglycans.
In contrast to interstitial myofibroblasts in IF, neointimal SMCs in experimental renal allografts are solely derived from an intrarenal source, probably the arterial media 
. In the current study, we observed a strong expression of perlecan in the neointima. Perlecan expression in arteries has been associated with inhibition of SMC proliferation and reduced intimal hyperplasia 
which favours for a role of perlecan in neointima stabilization. However, data reported by others indicate that arterial HS proteoglycans can actually activate SMC proliferation by modulating the function of basic fibroblast growth factor (bFGF/FGF2) 
. Although clear expression of collXVIII was observed in the neointima, its potential role in neointima formation is as yet unknown.
After transplantation, we observed a strong induction of perlecan in the glomerular and peritubular capillary BMs. Peritubular capillaries play an essential role in graft rejection 
. Upon capillary inflammation, endothelial cells become activated and changes occur in the BM, like splitting and multi-layering 
. The response in peritubular capillaries is similar to that observed in glomerular capillaries, and the thickened BM might be the resultant of processes associated with endothelial cell death and regeneration 
. Capillary BM changes are related to our previous data indicating endothelial chimerism (i.e.
presence of recipient-derived endothelial cells) in glomerular and peritubular capillaries in CTD 
. Both endothelial chimerism and perlecan expression in capillaries could be essential in capillary endothelial regeneration 
. Perlecan in capillary BM might thus play a role in maintaining the capillary endothelial integrity but also contribute to the inflammatory response 
We observed a major increase of collXVIII expression in the tubular BM after renal transplantation in both iso- and allografts. The integrity of the tubular BM and its changes are involved in inflammation, phenotypic changes of tubular epithelial cells, and the development of IF and tubular atrophy 
. Tubular epithelial cells can contribute to IF via epithelial-to-mesenchymal transition (EMT) in which epithelial cells transdifferentiate into interstitial myofibroblasts 
. CollXVIII and (weakly expressed) perlecan in the tubular BM could play a role in the EMT process by binding of chemokines and growth factors resulting in a concentration gradient in the tubular BM 
. This gradient might then direct migration of tubular epithelial cells into the interstitium during EMT. In line with this, preliminary data indeed suggest increased binding capacity of HS proteoglycans for FGF-2 in the tubular BM in allografts (not shown).
The more interrupted and less uniform expression of agrin in tubular BMs after transplantation supports the assumption that agrin normally plays a role in anchoring tubular epithelial cells, and focal loss of agrin could therefore be related to migration of transdifferentiated tubular cells in EMT or tubular atrophy 
. In addition to tubular atrophy and EMT, proteoglycan expression in the tubular BM could be involved in binding of L-selectin, thereby facilitating inflammatory responses in tubules 
. The potential causal role of BM HS proteoglycans in tubular atrophy or EMT are under current investigation in HS proteoglycan mutant mice.
We showed a marked induction of lymphangiogenesis in allografts, which was accompanied by the expression of perlecan at the abluminal side of lymphatic endothelium. Recovery of renal lymph drainage is shown to occur as early as 24 hours after renal transplantation 
, suggesting that lymph drainage and the process of lymphangiogenesis after renal transplantation is of potential functional relevance. However, it is still a matter of debate whether lymphangiogenesis and potential development of lymphoid structures in renal grafts is beneficial or detrimental to clinical outcome. Lymph vessels could be beneficial by mediating the drainage of extravasated fluid and the export of leukocytes 
. On the other hand, lymph vessels and additional development of lymphoid structures could also perpetuate the inflammatory response 
. We observed a clear correlation between the magnitude of lymphangiogenesis and severity of IF, suggesting that new lymph vessel formation may enhance the fibrotic process by stimulating the inflammatory process. This is supported by recent findings in diabetic nephropathy indicating that lymphangiogenesis is associated with inflammatory cell infiltration and progression of IF 
. In our study, increased lymphangiogenesis correlated with reduced graft function suggesting that therapies that target de novo
lymphatic formation might contribute to improved graft function. The existence of a causal relation between lymphangiogenesis and loss of graft function, however, needs to be established in future studies. The expression of perlecan in close proximity of lymphatic endothelial cells suggests a functional role for perlecan in lymphangiogenesis. This is supported by results from studies performed in a mouse model for regenerating skin which suggest that perlecan is involved in lymphatic endothelial cell migration, lymphatic organization and maturation 
. In addition, also versican, which was abundantly present in the interstitium, might play a role in lymphangiogenesis 
In conclusion, we identified increased spatial expression of HS and CS/DS proteoglycans in the intrarenal arteries, glomeruli and tubulointerstitium undergoing extensive tissue remodeling associated with CTD in renal allografts. Compartment-specific expression of proteoglycans in CTD might translate into compartment-specific responses to therapy. In line with this concept, we recently reported a differential response in renal allograft remodeling to aldosterone receptor blockade using spironolactone in which spironolactone ameliorated TV and FGS but not IF 
. The potential role of proteoglycans in the spironolactone-induced effects are currently under investigation.
Although our results are descriptive in nature, the observed differential expression of proteoglycans in renal allografts most likely also have functional consequences as the proteoglycan core proteins were shown to have GAGs that were able to bind L-selectin. Preliminary data furthermore suggest altered endogenous expression of natural proteoglycan ligands (such as FGF-2, HB-EGF, and L-selectin on leukocytes). As a resultant, the bioavailability of these ligands, which orchestrate tissue remodeling and inflammation, is most likely modulated due to altered proteoglycan expression as well as GAG side chain modifications.
Based on our results we propose that proteoglycans could be targets for intervention to ameliorate CTD. As an example, antibodies recognizing, and thereby blocking, specific HS-motifs/domains may inhibit leukocyte extravasation resulting in reduced inflammation. Also generated small inactive chemokine fragments might be used to block the HS proteoglycan-binding sites of their in vivo
active counterparts thereby making the HS proteoglycans less bioactive. Alternatively, we suggest the possibility to produce small HS-mimetics which may target more specifically a particular component of HS/heparin bioactivity 
. Therefore, focus should now be on the identification of the precise functional role of proteoglycans in chronic tissue remodeling after renal transplantation followed by exploration of the feasibility to use proteoglycans as targets for therapeutic intervention to ameliorate the development of CTD.