Tamm-Horsfall Protein Regulates Circulating and Renal Cytokines by Affecting Glomerular Filtration Rate and Acting as a Urinary Cytokine Trap*

Background: Kidneys play a key role in cytokine catabolism, but the mechanism remains unclear. Results: Deficiency of Tamm-Horsfall protein (THP), a kidney-specific protein, leads to marked increase of circulating cytokines, via decreased glomerular clearance and loss of urinary cytokine trapping. Conclusion: THP influences systemic cytokine clearance. Significance: Defects of THP may underlie disease pathogenesis of kidneys and other vital organs. Although few organ systems play a more important role than the kidneys in cytokine catabolism, the mechanism(s) regulating this pivotal physiological function and how its deficiency affects systemic cytokine homeostasis remain unclear. Here we show that elimination of Tamm-Horsfall protein (THP) expression from mouse kidneys caused a marked elevation of circulating IFN-γ, IL1α, TNF-α, IL6, CXCL1, and IL13. Accompanying this were enlarged spleens with prominent white-pulp macrophage infiltration. Lipopolysaccharide (LPS) exacerbated the increase of serum cytokines without a corresponding increase in their urinary excretion in THP knock-out (KO) mice. This, along with the rise of serum cystatin C and the reduced inulin and creatinine clearance from the circulation, suggested that diminished glomerular filtration may contribute to reduced cytokine clearance in THP KO mice both at the baseline and under stress. Unlike wild-type mice where renal and urinary cytokines formed specific in vivo complexes with THP, this “trapping” effect was absent in THP KO mice, thus explaining why cytokine signaling pathways were activated in renal epithelial cells in such mice. Our study provides new evidence implicating an important role of THP in influencing cytokine clearance and acting as a decoy receptor for urinary cytokines. Based on these and other data, we present a unifying model that underscores the role of THP as a major regulator of renal and systemic immunity.

capable of interacting with cognate membrane receptors present on multiple cell types and, in doing so, eliciting divergent biological responses (3). Although a physiological level of cytokines is an engine to drive the requisite biological activities, their overproduction can lead to detrimental consequences (4,5). In fact, emerging clinical and experimental evidence suggests that cytokines are the principal mediators of the complex inter-organ cross-talk and their imbalance underlies multiorgan failure (6). It is therefore no great surprise that under normal conditions cytokines in the circulation are kept at extremely low levels (2,7).
Of all organ systems, the kidneys play a leading role in the clearance of circulating cytokines (8 -11). With molecular masses of the mature peptides ranging from 8 to 25 kDa, most cytokines can pass freely through the molecular sieves of the ultrafiltration barrier of the glomeruli in the kidney (7). Indeed, radiolabeled cytokines administered intravenously in experimental animals are eliminated within minutes to hours from the bloodstream and are traced primarily to the urine (12,13). For some cytokines, a significant portion of the glomerulus-filtered load is endocytosed by the epithelial cells comprising the proximal tubules and destined for lysosomal degradation, with the rest excreted in the urine (14,15). Alternatively, urinary cytokines could be reabsorbed via the paracellular route along the renal tubules (16), but this possibility has yet to be experimentally validated. Despite the demonstrated importance of the kidneys in cytokine catabolism, many key questions remain. For instance, how is cytokine filtration through the glomeruli regulated? Would a reduced glomerular filtration affect renal cytokine clearance and hence systemic cytokine retention, and if so, would that have a systemic pathological effect? In addition, free urinary cytokines that have escaped the epithelial uptake are known to be harmful for renal tubular and urothelial cells (17)(18)(19). Does a decoy receptor exist in the urine that can serve as a scavenger to prevent the cytokines from overactivating the renal epithelial and urothelial cells and causing urinary tract pathology?
Tamm-Horsfall protein (THP) 2 is a 90-kDa, heavily glycosylated protein made by the functionally specialized epithelial cells of the kidney comprising the thick ascending limb of loop of Henle (TAL) (20 -25). Mature THP is anchored onto the luminal leaflet of the apical membrane of TAL until its controlled release into the urine by protease and/or lipase activities (26,27). Although reactive substances to polyclonal THP antibodies were found in non-kidney sources including serum, brain, liver, and amniotic fluid (23,28,29), recent RT-PCR and transgenic mice harboring a THP promoter/reporter transgene demonstrated that THP is expressed only in the kidneys (30,31). The relationship of THP with immunity was not appreciated until it was "rediscovered" as a potent immune suppressant in the urine of pregnant women and renamed uromodulin (32,33). However, the primary sequences of these two proteins are identical, although less is certain about their exact glycosylation contents (34 -36). In vitro, uromodulin binds avidly with IL1, IL2, and TNF, and inhibits antigen-mediated T cell proliferation or the IL1-mediated T-cell colony response by phytohemagglutinin (37)(38)(39)(40). Uromodulin also suppresses the cytotoxicity of monocytes. Interestingly, the urinary level of uromodulin increases during the second and third trimester of normal pregnancy, a physiological response attributed to an increased need for immune suppression during this period. In contrast, uromodulin is markedly reduced or even undetectable in hypertensive preeclampsia patients (41,42). Notably, most if not all of the in vitro immune suppressive properties of uromodulin were later found to be shared by THP isolated from healthy male urine, suggesting a functional commonality between THP and uromodulin (20,35). Despite these intriguing observations, controversies persisted regarding the role of THP in immunity. First and foremost, unlike the recombinant cytokines, native cytokines did not appear to interact with THP in vitro (43). Second, because normally THP is not significantly present in the interstitium or circulation where there are immune cells, its proposed immunosuppressive role would have to be local, e.g. to be confined to the urinary tract. Third, THP is an effective immune stimulator when directly exposed to neutrophils, monocytes, dendritic cells, via up-regulation of target cell cytokines and enhancement of their chemotaxis/phagocytosis (44 -46). The effects of THP on dendritic cell maturation and autoantibody production after intravenous THP injection are both dependent on TLR-4 signaling (46). These results seem to be in direct conflict with those suggesting a role for THP in immune suppression. Efforts in elucidating an in vivo role of THP in immune regulation and in reconciling the seemingly diametrically opposing functions of the THP have been hampered by the lack of experimental models, although more recent data support an immunosuppressive role for THP in the ischemia-reperfusion model of acute kidney injury (47,48).
In the present study, we explored the in vivo net effects of the loss of THP on circulating and renal cytokine homeostasis under both steady-state and stressed conditions. We studied the renal mechanisms whereby the absence of THP in knock-out mice alters cytokine clearance and hence systemic cytokine levels. Additionally, we examined whether, in the wild-type mice, renal and urinary THP forms natural complexes with cytokines and whether such a binding/trapping effect is absent in the THP KO mice using several independent but complementary approaches. We also determined the pathophysiological consequences of un-trapped urinary cytokines on renal epithelial activation. On the basis of our data, we present a unifying model that reconciles the current controversies and explains why THP plays an important role in renal and systemic cytokine homeostasis. Finally, because gene mutations, polymorphisms, and reduced expression of THP occur frequently in pathological conditions in humans, our findings suggest that local and systemic cytokine imbalances caused by THP deficiency may play a key role in the disease pathogenesis of the kidneys and other vital organs.

EXPERIMENTAL PROCEDURES
Genetically Engineered Mice Lacking Tamm-Horsfall Protein (Uromodulin)-Mice defective for the THP gene were generated using homologous recombination, resulting in the deletion of a region of the THP gene from 650 bp upstream of the transcriptional initiation site to the middle of intron 4 (49). Homozygous THP KO mice lacked THP mRNA or protein in the kidney, as evidenced by Northern blotting, RT-PCR, in situ hybridization, Western blotting, and immunohistochemistry. Both THP KO mice and their wild-type littermates used for this study were maintained in a specific pathogen-free facility and bred in parallel in a 129/SvEv background for 6 generations. The mice were tested vigorously to be free of any viral, bacterial, or parasitic infection. Mice, 2-3 months old, of both genders were used for this study. The genotype of the KO mice was verified using PCR of tail genomic DNA with specific primers (forward, 5Ј-gaagggactggctgctattg-3Ј; reverse, 5Ј-aatatcacgggtagccaacg-3Ј). The genotype of the wild-type mice was verified similarly with specific primers (forward, 5Ј-agggctttacaggggatggttg-3Ј; reverse, 5Ј-gattgcactcagggggctctgt-3Ј). All animal-related procedures were carried out in accordance with federal and local regulations and under an active protocol approved by the Institutional Animal Care and Use Committee (IACUC).
Quantitation of Serum, Urine, and Kidney Cytokines by ELISA-Whole blood was obtained from the retro-orbital sinus route of 2-3-month-old mice under anesthesia with ketamine and xylazine, and the serum was isolated routinely and stored at Ϫ80°C until use. The same cohorts (WT (n ϭ 23) and KO mice (n ϭ 22)) were used for the ELISAs to test serum cytokines, although, due to finite volume recovery, not all mice yielded sufficient amounts for all the assays, hence the slightly varied numbers for different cytokines. The 24-h urine samples were collected using autoclaved single-mouse metabolic cages (VWR; NALGENE*) into the collecting tubes containing 20 l of 10% thymol and 10 l of a mixture of protease inhibitors (80 M aprotinin, 5 mM bestatin, 1.5 mM E-64, 5 mM EDTA, 2 mM leupeptin, and 1 mM pepstatin A; Thermo Scientific). After centrifugation at 500 ϫ g, the urine supernatants were kept at Ϫ80°C until use. Total kidney proteins were extracted by homogenizing freshly dissected kidney tissue in lysis buffer containing a final concentration of 20 mM Hepes (pH 7.5), 150 mM NaCl, 1 mM EGTA, 1.5 mM MgCl 2 , 10% glycerol, 1% Triton X-100, and a mixture of protease inhibitors (same as above). After centrifugation at 17,000 ϫ g at 4°C for 20 min, the supernatant was collected and stored at Ϫ80°C until use. Protein concentrations were determined using a BCA reagent (Pierce) with bovine serum albumin as a standard.
Measurement of cytokine levels in serum, urine, and kidney was carried out using commercial ELISA kits consisting of antibodies and protein standards for mouse cytokines including IFN-␥, IL1␣, TNF-␣, and IL6 (Thermo Fisher Scientific), and CXCL1 (mouse equivalent of human IL8) and IL13 (R&D Systems). In the case of ELISA involving kidney proteins, cytokine standards were used as positive controls; and sample dilution buffer, bovine serum albumin, and purified mouse THP were used as negative controls.
Lipopolysaccharide Challenge-Lipopolysaccharide purified from Escherichia coli (0111:B4) using ion-exchange chromatography (500,000 endotoxin units/mg; Sigma) was reconstituted in autoclaved phosphate-buffered saline to a final concentration of 1 mg/ml. Four milligrams of LPS per kilogram of mouse body weight was injected intraperitoneally into randomly selected groups of THP KO and wild-type mice. At various time points post-injection (1.5, 3, 6, 12, and 24 h), the blood samples were obtained via the retro-orbital sinus route from anesthetic mice, whereas kidney specimens were procured after the mice were euthanized. The sera and total kidney protein extracts were then subject to ELISA as described above.
Determination of Serum Cystatin C and Renal Inulin and Creatinine Clearance-Serum cystatin C, a marker for renal glomerular filtration function (50,51), was measured by ELISA using a kit specific for the mouse protein (BioVendor), based on the manufacturer's instructions.
The glomerular filtration rate (GFR) was determined using intravenous delivery of FITC-conjugated inulin following a previously established protocol (52). Briefly, FITC-conjugated inulin (Sigma) was reconstituted in 0.9% NaCl to a final concentration of 5%. After dialysis in the dark against 0.9% NaCl (dialysis membrane cut-off: 1,000-Da; Spectrum Laboratories) to remove unbound FITC, the solution was filtered through a 0.22-m pore filter (Millipore), and the filtrate was then injected intravenously via the retro-orbital route under isoflurane-induced anesthesia. After injection, 20 l of blood samples were collected via the saphenous vein at 3, 7, 10, 15, 35, 55, and 75 min. The plasma was employed for determining the concentrations of FITC-inulin at 485-nm excitation and 538-nm emission wavelengths. The fluorescence readings were plotted into a 2-phase exponential decay curve using nonlinear regression with Internet-based SPSS software. GFR was determined with the formula: GFR ϭ I/(A/␣ ϩ B/␤) (I, the quantity of administered FITC-inulin; A and B, the y intercept values of the two decay rates; ␣ and ␤, decay constants for the distribution and elimination phases) (53). GFR was expressed as ml/min (with or without being referenced to the body weight).
Fractional creatinine clearance rate was determined by measuring the concentrations of the serum and 24-h urine samples using a commercial kit (Biovision, Mountain View, CA). After dilution in the assay buffer, serum and urine samples were incubated in a 96-well microplate with a solution containing creatinase, creatininase, enzyme mix, and color probe at 37°C for 1 h. The creatinine concentration was calculated by plotting the reading at 570 nm against a standard curve. Creatinine clearance was expressed as: urinary creatinine concentration ϫ 24-h urine volume/serum creatinine concentration.
Serum cystatin C level and inulin and creatinine clearance were determined at the baseline and after LPS challenge. In the latter situation, blood samples were obtained 12 h after the LPS challenge.
In Vivo Interaction of Renal Cytokines with Tamm-Horsfall Protein-THP KO mice and their wild-type controls (without LPS challenge) were sacrificed and their kidneys dissected out and homogenized in a lysis buffer designed for immunoprecipitation (50 mM Tris/HCl (pH 8.0), 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and a mixture of protease inhibitors (same as above)). After centrifugation at 17,000 ϫ g at 4°C for 20 min, the supernatant was collected and its total protein concentration was determined using the BCA reagent. The supernatant containing 1 mg of the total kidney protein extract was supplemented with protein A-conjugated agarose (Bio-Rad). After centrifugation, the supernatant was reconstituted with the addition of a rabbit anti-THP antibody (54) to a final dilution of 1:250, and the mixture was incubated at 4°C overnight. As negative controls, isotype, nonimmune, normal rabbit sera were similarly diluted and subjected to the same experimental procedures as with the rabbit anti-THP antibody. Fresh protein A-agarose was then added to the mixture and again incubated at 4°C overnight. After centrifugation, the agarose beads were washed 4 times in immunoprecipitation buffer and the bound protein complexes were eluted by boiling the agarose beads in 200 l of the aforementioned buffer for 10 min. Fifty microliters of the elutes were then added into the ELISA plates that were pre-coated with antibodies against mouse cytokines (Thermo Fisher Scientific), and ELISA was then carried out as described above. In parallel experiments, equal amounts (100 g) of the total kidney protein extracts were added directly into ELISA plates followed by ELISA to quantify the amount of renal cytokine input (THP-bound and unbound cytokines).
Native Polyacrylamide Gel Electrophoresis of THP-Cytokine Complex-One hundred micrograms of total kidney proteins extracted with the immunoprecipitation buffer described above were resolved on a 6% polyacrylamide gel in nondenaturing and nonreducing conditions. The proteins were then electrotransferred onto an Immobilon-PVDF membrane and immunoblotted with goat anti-mouse IL1␣ antibody. Following the incubation with a secondary donkey anti-goat antibody conjugated with horseradish peroxidase, the IL1␣ band was visualized with enhanced chemiluminescent (ECL) reagents (Pierce). The membrane was stripped and re-blotted with rabbit anti-THP antibody followed by a secondary, goat anti-rabbit antibody conjugated with peroxidase. To rule out nonspecific binding, the same membranes were stripped again and re-blotted with isotype antibodies (nonimmune normal goat serum in place of goat anti-mouse IL1␣ antibody; nonimmune normal rabbit serum in place of rabbit anti-THP antibody).
Co-sedimentation of Cytokine-THP Complex by Centrifugation-Freshly collected 24-h urine samples from wild-type and THP KO mice were centrifuged at 17,000 ϫ g at 4°C for 15 min. The pellets were re-suspended in distilled water; and the suspension and the corresponding supernatants were subjected to ELISA. The status of THP was determined by SDS-PAGE followed by Coomassie Blue staining as well as Western blotting using a rabbit anti-THP antibody (54).
Double Immunofluorescent Staining-Coronal sections (5 m thick) of paraffin-embedded kidneys from the wild-type mice (without LPS challenge) were de-paraffinized and underwent antigen unmasking in citrate buffer (pH 6.0) using micro-wave treatment (9/10 power output for 20 min). The sections were double-stained with (i) goat anti-IL1␣ and rabbit anti-THP or (ii) goat anti-IFN-␥ and rabbit anti-THP, followed by two secondary antibodies (Alexa Fluor 594-conjuated donkey anti-goat IgG (red fluorescence) and Alexa Fluor 488-conjuated donkey anti-rabbit IgG (green fluorescence)). Isotype, nonimmune antibodies (normal goat serum in place of anti-IL1␣ and goat anti-IFN-␥; normal rabbit serum in place of anti-THP) were used in parallel as negative controls with identical staining procedures.
Histopathology, Immunohistochemistry, and Western Blotting-Spleens from wild-type and THP KO mice were paraffinembedded, sectioned, and stained routinely with hemotoxylin and eosin. Alternatively, paraffin sections underwent antigen unmasking as described above and then immunohistochemical staining using a rat antibody against the mouse macrophage marker (F4/80(CI:A3)) (Novus). After incubation with a secondary goat anti-rat antibody conjugated with horseradish peroxidase, the reaction was developed in 50 mM Tris/HCl buffer (pH 7.4) containing 3,3Ј-diaminobenzidine tetrahydrochloride and H 2 O 2 . For quantification of the staining area and intensity, three antibody-stained sections per mouse of 3 WT and 3 THP KO mice were scanned and analyzed using NIH Image J software (rsb.info.nih.gov/ij/index.html). The average reading from WT mice was set at 1 and that from THP KO mice was expressed as fold-changes in reference to the WT mice.
For the assessment of renal neutrophil infiltration, paraffinembedded kidney sections from wild-type and THP KO mice underwent antigen retrieval as described, followed by incubation with a primary rat antibody against mLy-6G, a cell surface marker of mouse peripheral neutrophils (BioXCell; 1:10,000 dilution), and then with a secondary, donkey anti-rat antibody conjugated with Alexa Fluor 594 (Invitrogen; 1:500 dilution). The sections were double-stained with a rabbit anti-mouse collagen type IV polyclonal antibody (Abcam Inc.; 1:500 dilution) visualized by an Alexa Fluor 488-conjugated donkey anti-rabbit antibody (Invitrogen Corp.). The sections were subsequently counterstained with DAPI to reveal the nuclei. Positively stained neutrophils were counted at ϫ400 magnification from 20 nonoverlapping fields from each section, three sections each of kidney, and 8 mice per genotype. The neutrophil count was expressed as the number of neutrophils per 10 high-power fields. Cross-sections of a mouse bladder inoculated with a uropathogenic E. coli strain (UTI89) were used as positive controls for the neutrophil antibody.
Key signaling molecules in the cytokine activation pathway were determined by Western blotting using total kidney protein extracts. After SDS-PAGE, the proteins were transferred onto an Immobilon-PVDF membrane and blotted with primary antibodies against IB, NF-B (p65), phosphorylated Stat3, and MAPK (loading control), with subsequent steps carried out routinely.
Statistical Analysis-Mann-Whitney U test (two-sided) was used to assess the differences of spleen weights between wildtype and THP KO mice using SPSS software. Student's t tests were performed to evaluate the differences between the wildtype and THP KO mice for all other applicable assays using the same software. A p value of less than 0.05 was considered statistically significant.

Splenic Enlargement and Macrophage Infiltration in THP KO
Mice-An unexpected, yet frequent finding during gross anatomy of our THP KO mice was the markedly increased sizes of the spleens. Additionally, whereas spleens from the wild-type littermates always had smooth, shinning surfaces, those of the THP KO mice were often uneven, dull-colored and nodular (Fig. 1A). Two groups of randomly selected THP KO (n ϭ 32) and wild-type (n ϭ 32) mice, both aged 3 months and in the same (129/SvEv) background, were therefore sacrificed and their spleen and body weights measured. It became evident that the average spleen weight of THP KO mice was significantly greater than that of wild-type mice, either without (Fig. 1B, left  panel) or with (Fig. 1B, right panel) adjustment to the body weight. Severely enlarged spleens were also occasionally noted, with the spleen weights three times as heavy as those of the wild-type controls. H&E staining of the enlarged spleens showed disorganized white-pulp and prominent infiltration by the macrophages (Fig. 1C), the identity of which was confirmed by immunohistochemical staining using an antibody against mouse F4/80, a marker specific for the macrophages. Spleens from the THP KO mice contained a considerably greater number of macrophages than those from the wild-type mice (Fig.  1D). No other organ exhibited inflammatory changes in the THP KO mice (data not shown). These findings revealed a heretofore unsuspected, specific extra renal pathological consequence of the loss of THP expression in the kidney.
Marked Increase of Circulating Cytokines in THP KO Mice-Our finding that loss of THP, a protein made exclusively in the kidneys, would lead to splenomegaly prompted us to explore whether this was due to a rise in circulating cytokines. Using ELISA, we quantified the serum levels of inflammatory cytokines IFN-␥, IL1␣, TNF-␣, IL6, and CXCL1 (a mouse homologue of IL8) and a noninflammatory cytokine IL13. All but IL6 and IL13 were significantly higher in THP KO mice than in the wild-type controls (Fig. 2, A-F). IL6 and IL13 levels were also elevated, although they did not reach statistical significance (Fig. 2, D and F). The serum level of IFN-␥ (averaged 750 pg/ml in THP KO mice; Fig. 2A) was considerably higher than that of the other cytokines (averaged 10 -35 pg/ml; Fig. 2, B-F), suggesting IFN-␥, a key activator of macrophages (55), as a key reason for the splenomegaly observed in our THP KO mice. The relatively large standard deviation in Fig. 2 (also see below) might reflect the fact that, despite 6 generations of back-crossing, the mice we used had not reached an in-bred status, although other possibilities cannot be completely ruled out.
When challenged with lipopolysaccharide, the increase of the circulating cytokines over the steady-state level was also more pronounced in THP KO mice than in wild-type controls (Fig. 3, A-F). The time course of induction varied among the cytokines. TNF-␣ peaked around 1.5 h after LPS administration and leveled off quickly after 6 h (Fig. 3C), in accordance with its being an early responder. IL1␣ lagged in its peak time (e.g. at 3 h) and subsided also less quickly (after 12 h) (Fig. 3B). In FIGURE 3. Circulating cytokines in response to LPS treatment. Wild-type (WT, filled circle; n ϭ 8) and THP knock-out (KO, filled square) mice (n ϭ 8) were injected intraperitoneally with LPS, and their blood was obtained at 1.5, 3, 6, 12, and 24 h after injection and processed for ELISA. Note that although serum cytokine levels increased in both WT and KO mice in response to LPS, KO mice had significantly higher levels of cytokines, except IL13, at various time points than WT mice (asterisks denote the time points where statistical significance was present; see text for details). IL13 levels were also higher in KO mice than in WT mice at all time points, but did not reach statistical significance. Twenty-hour urine samples collected with metabolic cages were subject to ELISA for IFN-␥ and IL1␣. The concentrations were normalized against simultaneously measured creatinine. Total excretion in a 24-h period was also calculated based on the concentrations and urine volumes. Note that urinary excretion of IFN-␥ and IL1␣ was not higher in THP KO mice than in WT controls. Other cytokines shown in Figs. 2 and 3 were also measured with the same set of urine samples but were undetectable, thus not shown here.
comparison, IFN-␥ and IL6 did not reach the highest levels until 6 h and remained high until after 24 h (Fig. 3, A and D). CXCL1 and IL13, on the other hand, peaked early (e.g. at 1.5 h) and remained high throughout the 24-h observation period. Taken together, although LPS induced cytokines in both wildtype and THP KO mice, it had more profound effects on THP KO mice.
Effect of THP Deficiency on Urinary Cytokine Excretion-The kidney plays an important role in the clearance of cytokines (9,14,15), but the exact mechanisms involved are not well understood. For example, clearance depends on at least three principal factors: glomerular filtration rate, cytokine reabsorption, and tubular degradation. Because of the elevated levels of serum cytokines in THP KO mice, we examined their urinary excretion rate. The urinary concentrations of IFN-␥ and IL1␣ in THP KO mice were comparable with those in WT mice (with or without adjustment to creatinine), and the total amounts of their urinary excretion in 24 h were not significantly different between THP KO and wild-type mice (Fig. 4). This suggested that the clearance of IFN-␥ and IL1␣ was reduced in THP KO mice. Despite their elevated levels in the serum, urinary TNF-␣, IL6, CXCL1, and IL13 were undetectable in the THP KO mice as with the wild-type controls (data not shown). This may be because these latter cytokines undergo significant post-filtration handling (such as reabsorption) by the kidney without any residual urinary excretion. Our repeated attempts to determine the urinary levels of the various ILs in LPS-treated mice were met with limited success, because these mice became extremely ill and were not amenable to urine collection in metabolic cages.
Reduction in Glomerular Filtration as a Possible Contributor to Elevated Serum Cytokines-Because cytokine clearance depends in part on GFR (9, 14, 15), we used several methods to assess the differences in GFR between WT and THP KO mice. First, we measured the serum level of cystatin C, a low molecular mass (13.3 kDa) and positively charged protein removed from circulation almost exclusively by glomerular filtration (50,51,56,57). We found the steady-state level of serum cystatin C to be significantly higher in THP KO mice than in the wild-type mice (Fig. 5A, left two columns). We also determined the clearance of FITC-conjugated inulin from the circulation and found it to be significantly lower in THP KO mice without (Fig. 5B, upper panels) or with (Fig. 5B, lower panels) it being referenced to body weight. The extent of serum cystatin C elevation and inulin clearance reduction in THP KO mice worsened when they were challenged with LPS (Fig. 5, A and B, right two columns). Specifically, inulin clearance under the untreated condition in THP KO mice was 22.5% lower than that of WT mice (Fig. 5B). After LPS treatment it was 46.9% lower than in WT mice. Last, we determined the fractional excretion rate of creatinine both at steady-state and after LPS challenge (Fig. 5C), yielding data highly consistent with those of the inulin clearance. These data point to a glomerular filtration defect in THP KO mice, both at the baseline and upon LPS challenge, as a possible contributor of a reduced renal cytokine clearance. The reduction in inulin and creatinine clearance in THP KO mice was somewhat disproportional to the increase in serum cytokines, especially for IFN-␥ and IL1␣, suggesting that additional post-filtration tubular handling may be involved in regulating the level of circulating cytokines (see later).
Renal and Urinary Cytokines Bound to THP in Wild-type Mice, a "Trapping" Effect Lost in THP KO Mice-Although previous studies showed in vitro binding of several recombinant cytokines to THP, it remained highly controversial as to whether such a binding occurs in vivo with native cytokines (20,21,23,24). To clarify this issue, we tested whether cytokines could co-purify with THP from total kidney protein extracts of wild-type mice, but not from THP KO mice. Kidney protein extracts from both types of mice (without LPS challenge) were subject to immunoprecipitation using an anti-THP antibody, followed by ELISA of cytokines on the immunoprecipitated products. Compared with the THP KO mice from which little of the four tested cytokines (IFN-␥, IL1␣, TNF-␣, and IL13) could be co-precipitated by the THP antibody, these cytokines were readily precipitated in wild-type mice (Fig. 6). Approximately 1% IFN-␥, 25% IL1␣, and 10% IL13 of their respective total inputs were co-purified (Fig. 6, A, B, and D, compare the upper panels with the corresponding lower panels), whereas only 0.1% TNF-␣ was co-purified with THP (Fig. 6C, compare the upper panel with the lower panel). The relatively low proportion of IFN-␥ and TNF-␣ that were THP-bound could be a reflection of the relatively high total cytokine pool in the renal epithelial cells of which most was cytoplasmic and not in direct contact with the apical surface located or urinary THP (see Fig.  7C). It is also possible that different cytokines bound to THP with different affinities and/or capacities (see "Discussion"). Despite the overall higher cytokine concentrations in the starting materials of the THP KO mice (filled bars in the lower panels of Fig. 6, A-D), little if any cytokine was precipitated by THP antibody in THP KO mice, again supporting the specific cytokine-THP interaction occurring only in the WT mice. In additional negative control experiments done in parallel, normal nonimmune rabbit serum replacing the anti-THP failed to immunoprecipitate any appreciable level of cytokines (data not shown), further demonstrating the specificity of the THP-cytokine interaction in the WT mice. These experiments, performed three times with three independent cohorts, yielded highly consistent results.
To verify the co-immunoprecipitation data, we performed native gel electrophoresis by resolving total kidney proteins under nondenaturing and nonreducing conditions. Upon Western blotting, we found that IL1␣ co-migrates with THP in wild-type mice, but not in THP KO mice (Fig. 7A). Isotype control antibodies from the same animal species where the primary antibodies were generated did not produce any band, thus establishing the specificity of the antibody-antigen reaction (Fig. 7A, lower panels). Taking advantage of the property of urinary THP to form high-molecular mass polymers (58), we also tested whether urinary cytokines could be co-sedimented with THP by high-speed centrifugation. We found that, in contrast to the THP KO mice in which IL1␣ was almost exclusively present in the urine supernatant, an appreciable amount of this cytokine was found in urine pellets of wild-type mice where THP was present (Fig. 7B), thus further supporting the THPcytokine interaction. Finally, in wild-type mouse kidneys, IL1␣ and IFN-␥ were co-localized immunohistochemically with THP at the apical surface of the thick ascending limb of loop of Henle, where THP is normally located (Fig. 7C). Both IL1␣ and IFN-␥ had a wider tubular distribution (in proximal and distal tubular cells in addition to TAL), and was more cytoplasmic than THP, thus explaining why only some of the interleukins were THP-bound in the co-immunoprecipitation (IP) experiment (Fig. 6). . In vivo binding of renal cytokines to THP. Top panels, co-immunoprecipitation of THP-bound cytokines followed by cytokine ELISA. Total kidney protein extracts from wild-type (WT) and THP KO mice (2 representative mice per genotype shown) were subject to IP using a rabbit anti-THP antibody. The IP products were loaded into microtiter wells pre-coated with antibodies against IFN-␥ (A), IL1␣ (B), TNF-␣ (C), or IL13 (D), to capture THPbound cytokines. This was followed by conventional ELISA (see "Experimental Procedures" for details). Starting materials for IP contained the same amount of total kidney proteins. Lower panels, ELISA quantification of renal cytokine input (e.g. THP-bound and -unbound cytokines), expressed as per milligram of total kidney proteins. Note that, whereas THP KO mice had higher renal cytokines (lower panels, filled bars), little could be precipitated with anti-THP antibody (upper panels, filled bars). In contrast, in wild-type mice, each of the four tested cytokines could be precipitated by the anti-THP antibody (upper panels, open bars) strongly suggesting an in vivo THP-cytokine interaction (see text for details).

Effects of THP Deficiency on Renal Cytokine Content and
Production and Renal Inflammation-To test the potential effects of the unbound urinary cytokines in the absence of THP on renal epithelial cells, we compared the renal cytokine levels of THP KO mice versus those of wild-type mice (Fig. 8). The steady-state levels of IFN-␥, IL1␣, IL6, and IL13 were all significantly higher in THP KO mice than in wild-type mice (Fig. 9, A, B, D, and F). TNF-␣ and CXCL1 were also higher although the difference was not statistically significant (Fig. 8, C and E). Western blotting analyses showed that there was a down-regulation of IB and up-regulation of NFB and phosphorylated Stat3, supporting the cytokinemediated activation of signaling pathways in THP KO mouse kidneys (Fig. 9A). FIGURE 7. Native gel electrophoresis, centrifugal co-sedimentation, and immunohistochemical co-localization of the IL-THP complex. A, total kidney proteins were extracted with a nondenaturing/nonreducing buffer, resolved by a native polyacrylamide gel, electrotransferred onto an Immobilon PVDF membrane, and immunoblotted with anti-IL1-␣. The membrane was then stripped and re-immunoblotted with anti-THP antibody. To demonstrate strict specificity, the same membranes were stripped again and blotted with nonimmune, isotype antibodies from the same animal species as with anti-IL1␣ and anti-THP. Short lines on the left edge of the left panels denote the molecular mass standards (from top to bottom: 130, 95, 72,56,43,34,26,17,and 11 kDa). Note that IL1␣ co-migrated with THP in wild-type (WT) mice, but not in THP KO mice, indicating in vivo IL1␣-THP complex formation. Also note the lack of any nonspecific reaction with isotype antibodies. B, left panel, 24-h urine samples from wild-type and THP KO mice were centrifuged at high speed and their pellets were re-suspended and subject to ELISA of IL1␣. Note that urinary IL1␣ could be precipitated by centrifugation in the WT mice, which contained THP, but not in the THP KO mice, which lacked THP. B, right panel, Coomassie Blue-stained SDS-PAGE (left strip) and Western blotting (right strip; using rabbit anti-THP) of the pellets from the high-speed centrifugation (left panel), showing the presence of THP (arrow) in WT mice but not in THP KO mice. Short lines on the right edge of the right strip denote the molecular mass standards (from top to bottom: 130, 95, 72,56,43,34,26,and 17 kDa). C, paraffin-embedded kidney sections from wild-type mice were double-stained with (i) anti-IL1␣ and anti-THP antibodies (upper panels); (ii) anti-IFN-␥ and anti-THP (middle panels); and (iii) isotype, nonimmune antibodies as negative controls for anti-IL1␣/anti-IFN-␥ and anti-THP (lower panels). Note the significant co-localization of IL1␣ or IFN-␥ with THP (yellow color, arrows). IL1␣ and IFN-␥ were present in more renal tubules (arrowheads) than THP. All panels in C were ϫ200 magnification.
To discern whether the increased activation of cytokine signaling in THP KO mice was due to increased renal cytokine synthesis or increased cytokine reabsorption by or increased interactions with renal epithelial cells, we performed real-time PCRs of cytokine mRNA expression (Fig. 9B). The levels of cytokine production were not significantly different between WT and KO mice. In addition, a mouse neutrophil-specific antibody, which strongly labeled neutrophils in a uropathogenic E. coli-infected mouse bladder, did not detect apparent neutrophil infiltration in either WT or THP KO mice (Fig. 9C). These data are in favor of the idea that increased renal cytokine content (Figs. 6 and 8) results from increased reabsorption of urinary cytokines and/or increased interaction of urinary cytokines with renal epithelial cells in the absence of THP.

DISCUSSION
Functional THP Regulates Serum Cytokine Levels-One of major findings we made in this study was that the loss of THP, a urinary protein made exclusively by kidney epithelial cells, led to a marked elevation of circulating cytokines in THP KO mice (Figs. 2 and 3). This occurred reproducibly in multiple cohorts and both at steady-state (Fig. 2) and after LPS challenge (Fig. 3); it affected most cytokines examined (Figs. 2 and 3) and coincided with a prominent, nonrenal organ anomaly, e.g. splenomegaly (Fig. 1). The increase in serum cytokines was not accompanied by a corresponding increase in their urinary excretion (Fig. 4), suggesting decreased renal clearance, especially for IFN-␥ and IL1␣. Furthermore, our data indicate a significantly reduced glomerular filtration rate in THP KO mice, as evidenced by the increased serum concentration of cystatin C, a marker for the glomerular filtration function, as well as decreased inulin and creatinine clearance (Fig. 5). Together, our data suggest that THP deficiency can cause systemic cytokine imbalance at least partially by affecting glomerular filtration. It should be noted that reduced GFR was also observed in an independent THP KO model (59), although changes involving circulating cytokines were not investigated.
THP as Decoy Receptor for Urinary Cytokines-Our present study also provided the first experimental evidence that THP formed complexes with cytokines in vivo. Whether THP is capable of binding cytokines has been fiercely debated, mainly because virtually all previous binding studies were done in vitro using recombinant cytokines (37,60). Although recombinant human cytokines consistently bound to purified THP, this result could not be reproduced with native cytokines (43). In the present study, we employed four independent methods, e.g. co-IP, native gel electrophoresis, co-sedimentation, and double immunofluoresent staining, using in vivo isogenic materials, e.g. mouse native THP and mouse native cytokines (Figs. 6 and 7). We demonstrated that THP-cytokine complexes existed in situ in wild-type mice. In stark contrast, such complexes were lost during THP deficiency. In addition, THP appears to bind both inflammatory and noninflammatory cytokines, suggesting that it is the chemical properties of THP that determine its promiscuousness in cytokine interaction (see later). Furthermore, by comparing the renal cytokine levels of wild-type mice versus those of THP KO mice and determining the cytokine signaling pathways, we showed that the unbound urinary cytokines in THP KO mice were capable of binding to and activating renal epithelial cells (Figs. 8 and 9). Based on these results, we propose a new role for THP as a urinary trap that serves to interact with and inactivating glomerulus-filtered cytokines, thus preventing them from activating renal epithelial and urothelial cells. The trapping role played by urinary THP may not be limited to cytokines. Previous in vitro studies suggested that THP also binds to complement C1 and C1q (61)(62)(63). It is possible that THP is a promiscuous urinary trap not only for cytokines, but also for other immunological effectors. The previously noted IgG-THP interaction that has thus far been shown only in vitro (64) could have presented technical problems in ELISA, Western blotting, and immunohistochemisty involving THP-containing materials from wild-type mouse kidneys. However, it must be emphasized that we included multiple controls in our experiments, in particular the isotype nonimmune antibodies to replace THP-specific antibodies wherever applicable. We found no evidence of any nonimmune IgG-THP binding using in vivo materials.
It appears that cytokines bind with THP via distinct structural determinants. In vitro studies showed that the high-mannose moieties of THP are required for its interaction with IL1 and TNF, because de-glycosylation or inclusion of free mannoses effectively blocks the THP-cytokine interaction (65,66). Interestingly, high-mannose glycosylation of THP is extremely conserved, being present in THP isolated from all mammals studied to date (67,68). This suggests a strong structural/functional relationship in the THP-cytokine interaction and adds to the functional importance of THP-cytokine binding as an evolutionarily conserved mechanism in renal protection. It cannot FIGURE 8. Renal cytokine content. A-F, total protein extracts from groups of wild-type (WT) and THP KO mice were used to assess cytokine production using ELISA. The cytokine levels were expressed as a ratio to the total kidney protein input (e.g. pg of cytokine/mg of total proteins). Note that all the six cytokines assayed without LPS challenge were higher in KO mice than in WT controls, although the increases for TNF-␣ and CXCL1 were not statistically significant. be ruled out that binding affinities between THP and different cytokines may vary depending on different primary structures and secondary modifications of the cytokines.
Unifying Model of THP in Regulating Local and Systemic Immunity-On the basis of our current data and those published previously, we propose a model that depicts from a mechanistic standpoint how the kidney acts as a major catabolic organ for cytokines. Normally, small sized cytokines pass freely through the glomerular filtration barrier (7). This filtration process is regulated by a functional THP, which ensures high glomerular clearance, hence low circulating cytokines. Once filtered through the glomeruli, the bulk of urinary cytokines is retrieved by the proximal tubules and degraded via the lysosomal pathway (14,15). Those that have escaped the uptake and/or exceeded the absorption limit move downstream with the urine flow but are soon captured by urinary and/or membrane-bound THP at the TAL. Although free cytokines are capable of binding to and overactivating the renal epithelial and urothelial cells (18,19), THP binding traps the free cytokines, thus blocking their biological activities. The trapping effect may also prevent the uptake of cytokines by the distal tubular cells as recently suggested (69) or by para-cellular re-absorption (16), although the existence or lack thereof of the latter route needs experimental validation. Our model underscores the critical importance of THP, a urinary protein normally insulated from the immune system, in ensuring both renal protection and systemic immune homeostasis. In emphasizing the physiological roles of THP in immune suppression, we are by no means ruling out its immune stimulatory effect under certain pathological conditions (44,70,71). Such immune stimulatory responses are distinguishable from the suppressive functions of THP in context, biological end points, and clinical correlates.
THP Deficiency-mediated Cytokine Imbalance as Key Mediator of Chronic Kidney Diseases and Multiorgan Failure-Evidence is accumulating that THP deficiency, both qualitative and quantitative, exists in a wide spectrum of pathophysiological conditions in humans. Germline mutations in the THP gene, which result in THP misfolding and failure to exit from the endoplasmic reticulum, cause juvenile-onset hyperuricemic nephropathy, type II medullary kidney cystic disease, and glomerulocystic disease (25,72,73). Single nucleotide polymorphisms in the promoter and coding regions of the human THP gene, which may affect THP gene expression and function, are strongly associated with chronic kidney disease (51). Decreased urinary THP excretion has also been found in type I diabetes, acute tubular necrosis, and hyperprostaglandin E syndrome (74 -79). The exact cause and pathogenesis of many of these diseases remain unclear. Based on our data, it can be speculated that the local and systemic cytokine imbalance caused by THP deficiency could be a mediator in the genesis and progression of some of these conditions. Given the fact that cytokines play a key role in multiorgan failure caused by renal insufficiency (80), THP deficiency could also contribute to this complicated pathophysiological process for which little is currently known. Elucidating how renal cytokine homeostasis can be maintained during the various pathophysiological conditions will likely provide insights into improved therapeutic options.