To study the effects of uremic plasma on the ability of TRβ1 to bind to a specific TRE we analyzed the binding of hRXRα-hTRβ1 heterodimers to DR-4. In this assay, the protein-DNA complex was visualized by labeling the TRβ1 with ³⁵S. In the presence of T₃, the addition of increasing amounts (Figure 1; lanes 2–4) of plasma from normal individuals improved the binding of hRXRα-hTRβ1 to DNA. Conversely, TR incubation with uremic plasma (lanes 5–7) collected prior to hemodialysis significantly reduced the binding of heterodimers (RXR-TR) to DR-4. Band densitometry analysis of 5 independent experiments showed that uremic plasma reduced hRXRα-hTRβ1-DR-4 complex formation by 77 ± 15%, compared to plasma from normal subject (not shown). Similar results were observed for the thyroid response element F2 (inverted palindrome) in which uremic plasma also inhibited the RXR-TR binding to DNA (not shown). Pre-treatment of hTRβ1 with T₃ failed to improve the ability of the dimer hRXRα-hTRβ1 to bind to DNA.

We used uremic plasma from four different patients to determine whether the observed inhibition of RXRα-TRβ1 binding to DNA (DR-4) was patient specific. Uremic plasma samples from all four patients inhibited RXRα-TRβ1 binding to DR-4 to various degrees (not shown). However, we could not detect any correlation between TR binding impairment and abnormalities in plasma levels of urea, creatinine or thyroid hormone.

To exclude the possibility that the inhibition of hRXRα-hTRβ1 binding to DNA induced by uremic plasma could be due to proteolytic degradation of TRβ1, we incubated ³⁵S-labeled TRβ1, at the same conditions as in the gel-shift experiments, with normal or uremic plasma at 4°C, for thirty minutes. ³⁵S-labeled TRβ1 samples were then analyzed by SDS-PAGE. As expected, the major translation product of TRβ1 was 53 kD and incubation of theses products with normal or uremic plasma did not modify the translated [³⁵S]TRβ1 (Figure 2). These results indicated an absence of uremic proteolytic activity that might be involved in the decrease of TR-RXR complex formation on DNA.

It is yet unknown why uremic plasma diminished TRβ1-RXRα binding to DNA, when compared to non-uremic plasma. The observed effect could be ascribed either to a lack of some factor(s) typically present in normal plasma or to the presence of some inhibitory products present in uremic plasma. To test the second hypothesis, we analyzed the influence of uremic plasma, collected before and after hemodialysis, on TRβ1-RXRα-DR-4 complex formation using gel-shift assays. In these experiments, [³⁵S] TRβ1 was incubated with normal or uremic plasma, collected before (pre-HD) or 4 h after hemodialysis (post-HD). As shown in Figure 3, uremic plasma collected before HD (lanes 5–7) decreased the binding of hTRβ1-hRXRα to DNA (DR-4), relative to normal plasma (lanes 2–4). However, when these receptors were pre-incubated with uremic plasma collected from the same patient after hemodialysis (lane 8–10), an important improvement of hTRβ1-hRXRα binding to DR-4 was observed. Although hemodialysis improved complex formation, it did not completely recover the inhibition caused by uremic plasma. Densitometry analysis demonstrated that hemodialysis increased by 50% the hTRβ1-hRXRα heterodimer binding to DNA when compared to pre-HD uremic plasma (not shown).

In view that HD improved the binding of hTRβ1-hRXRα to DR-4 response element, and that previous studies showed that uremic solutions inhibited hVDR-RXRα binding to DR-3-response element (VDRE) 262729 we decided to determine whether uremic plasma collected before and after hemodialysis has the same effect on hVDR-RXRα binding to DR-3.

Accordingly, a similar experiment was carried out with hVDR treated with 1,25 (OH)₂ D₃ vitamin (VD₃). [³⁵S]hVDR was incubated with non-uremic plasma, uremic plasma collected before and after hemodialysis. As shown in Figure 4, in comparison to control (lane1), the incubation of VDR with plasma from normal individuals increased [³⁵S]hVDR-RXRα binding to DR-3 (lanes 2–4). In contrast, when [³⁵S]VDR was incubated with uremic plasma collected before HD, we observed a significant reduction of hVDR-hRXRα binding to DR-3 (lane 5–7). On the other hand, when compared to pre-HD uremic plasma pre-incubation with uremic plasma collected after hemodialysis (lanes 8–10) improved hVDR-hRXRα binding to DR-3. Taken together, these results suggest that dialyzable toxins were responsible for the reduced binding of hTRβ1-hRXRα and VDR-RXRα to DNA.

Next we decided to evaluate whether uremic plasma impairs DNA binding of other members of the nuclear receptor superfamily. We performed gel shift assays using the nuclear fatty acid receptor PPARγ and its response element, DR-1. Members of the PPAR family have been shown to play an important role in obesity and the plurimetabolic syndrome 30 and insulin resistance has also been described in uremic patients 31. In contrast to what was observed with TRβ1 and VDR, Figure 5 shows that the incubation of [S³⁵] hPPARγ with uremic plasma failed to reduce the binding of PPARγ-hRXRα to DR-1 (lanes 2–4 compared to lanes 5–7). Such finding suggests that the inhibitory effect of uremic plasma does not extend to all members of nuclear receptor superfamily.

Our findings demonstrated that hemodialysis partially corrected the inhibition of hTRβ1-hRXRα binding to DR-4 caused by uremic plasma. We therefore hypothesized that dialyzable toxins may cause the inhibitory effect of uremic plasma on protein-DNA complex formation. To gather more information on the nature of these toxins we heated normal and uremic plasma for 5 minutes at 100°C to test if the inhibitory factor was thermo-labile. We subsequently performed gel shift assays. To control for our experimental setup we used in these experiments [³²P] labeled DR-4 with unlabelled TRβ1-RXRα (Figure 6). Furthermore, in order to avoid any effect from plasmatic phosphatases on [³²P] DNA, we also incubated the normal and uremic plasmas with a phosphatase inhibitor cocktail (lane 6).

Heating the plasma from normal individual did not affect the binding of the hTRβ1-hRXRα heterodimer to [³²P] DR-4 (lane 1 compared to lane 2). As shown in prior experiments, uremic plasma significantly diminished the hTRβ1-hRXRα – [³²P] DR-4 complex formation (lane 4). However, heating the uremic plasma did improve the binding of the hTRβ1-hRXRα heterodimer to [³²P] DR-4 (lane 5). In addition, the effects of uremic plasma persisted in the presence of phosphatase inhibitors (lane 6).

Lastly we examined if the in vitro inhibitory effect of uremic plasma on TR binding to DNA could affect the functions of TR as a transcription regulator in a cell based assay (Figure 7). We transfected U937 cells with cDNA encoding the human TRβ1 and a reporter gene with a DR-4 response element upstream of the firefly luciferase coding sequence. We used RPMI-1640 medium treated with ultrafiltrate (UF) solution 10-fold concentrated from normal or uremic patients collected before or after hemodialysis. As shown in Figure 7, in the absence of any UF (Control), T₃ activated transcription by 5.7 ± 1.2 fold. When the cells were treated with UF from normal individuals, we did not observe significant changes in the transcription activation of the luciferase reporter (4.7 ± 0.74 fold; ns). Interestingly, when compared to the UF from normal individuals, UF collected Pre-HD from uremic patients reduced T₃-dependent transcriptional activation by 66% (1.6 ± 0.35, p < 0.05). On the other hand, U937 cells treated with UF collected Post-HD had no significant effect on T₃-dependent transcription activation (4.8 ± 1.47 fold; ns). These results suggest that uremic toxins impair T₃ induced transcriptional activation and that hemodialysis reduces their inhibitory effect.

Four patients from the chronic dialysis program of Soclimed Dialysis Clinic were enrolled in our study. All patients were men whose age ranged from 19 to 43 years with the mean age being 34 years. They appeared well nourished and clinically and laboratorial euthyroid; none had a history of thyroid disease, thyroid hormone therapy, treatment with amiodarone or clinically detectable goiter. Etiology of their chronic renal failure was as follows: chronic glomerulonephritis (2); hypertension (1); reflux nephropathy (1). Mean plasma urea level was 178 ± 44.8 mg/dL (120 to 233 mg/dL), while that of creatinine was 12.6 ± 2.7 mg/dL (9.9 to 16.3 mg/dL). Patients were on hemodialysis 3 times a week, during 4 hours using a 1.8 m² Fresenius^® Polysulfone filter. Normal control subjects consisted of three healthy men, age ranging from 23 to 41 years, with the mean age of 32 years. The experimental protocol was approved by the Human Rights in Research Committee of the University of Brasilia and all patients and normal individuals gave their informed consent.

For the in vitro DNA binding assay, uremic plasma was collected immediately before and after 4 h of hemodialysis, aliquoted into 20 μL samples and stocked at -20°C. Uremic ultrafiltrate (UF) was also collected pre and post 4 h hemodialysis. Lyophilisation was used to concentrate ultrafiltrate samples. The lyophilisates were re-suspended in bidistillated water to a 10-fold concentrated solution, as effects of the UF were not detectable at lower concentrations (1 fold, 2.5 fold, 5 fold concentrated; not shown). Samples were subsequently desalted by filtration using Centricon 3 filters. Following centrifugation, the pellet was re-suspended in RPMI-1640 medium, (10% newborn bovine serum; 2 mM glutamine; 50 units/mL penicillin; 50 μg/mL streptomycin) and pH corrected to 7. Normal UF was collected from control plasma of normal individuals. The treatment solution was prepared in the same manner as the uremic solution. All experiments were performed with the uremic sample from the same patient that showed the strongest inhibitory effect.

Gel shift assays were used to evaluate the binding of ³⁵S-labeled TR synthesized in reticulocyte lysate on 600 fmoles of unlabeled DR-4 (5'-AGCT TC

AGGTCA

AGGTCA

TGACCC

AGGTCA

In vitro receptor synthesis was performed using plasmids encoding hTRβ1 hRXRα, and hPPARγ 38 and hVDR 39 with the TNT-coupled Reticulocyte Lysate System (Promega, Madison, WI) containing a methionine-free aminoacid mixture, and either 20 μM cold methionine or ³⁵S-labeled methionine. DNA plasmid (0.2–2 μg) was added to TNT Quick Master Mix and incubated in 50 μL for 90 min at 30°C. To confirm efficiency of the translation reaction ³⁵S-labeled translated proteins were analyzed by sodium dodecyl sulfate gel electrophoresis (SDS-PAGE). The TR-DNA complex was visualized by labeling reticulocyte lysate-translated receptors with ³⁵S or by using labeled DNA with ³²P using polynucleotide kinase. Prior to incubation with DNA, the reticulocyte lysate-translated receptors were treated with TRβ1, VDR and PPARγ receptors ligands (T₃, 1,25-dihydroxy-vitamin D₃; and 9-cis-retinoic acid respectively) for 30 min at 4°C. Labeled nuclear receptor plus ligands were then incubated in different volumes (0.5, 1 and 2 μL) of uremic or normal plasma for 30 min at 4°C. Following plasma exposure, the nuclear receptors were incubated for another 20 min at room temperature (20–30°C) in a solution containing 2 μg non-specific DNA poly (dIdC) (Pharmacia LKB, Piscataway, NJ), cold specific response element (10 ng/reaction), nonradioactive RXRα and a binding buffer in a 20 μL reaction as previously described 23. When using non-labeled nuclear receptors the gel shift experiment was performed using ³²P-DR-4 (2000–5000 cpm). A phosphatase inhibitor cocktail 2 (Sigma, P 5726) was added when the DNA was labeled (³²P-DR-4). The binding buffer contained 0.2 mM Na₂HPO₄, 0.2 mM NaH₂PO₄, 1 mM MgCl₂, 0.5 mM EDTA, and 5% glycerol. Final samples were loaded on 5% nondenaturing polyacrylamide gel, previously run for 30 min at 200 V. To separate the protein-DNA complexes, the gel was run at 4°C for 90–180 min at 240 V, using a running buffer (pH 7.5 for 10X stock at room temperature) containing 6.7 mM Tris-base, 1 mM EDTA, and 3.3 mM Sodium Acetate. The polyacrylamide gel was dried at the end of electrophoresis and autoradiographed.

Human promonocyte U937 cells were maintained in culture as previously described 38. For transfection assays, cells were collected by centrifugation and resuspended in transfection solution (0.5 mL/ 1.5 × 10⁷ cells) containing PBS, 100 mM calcium and 0.1% dextrose and mixed with 2 μg of human TRβ1 expression vector, 4 μg of the luciferase (Luc) reporter and 500 ng control β-galactosidase vector. The reporter plasmid contained a synthetic TR response element containing two copies of DR-4 cloned immediately upstream of a minimal thymidine kinase (tk) promoter (-32/+45) linked to luciferase coding sequences 38. The cells were transferred to a cuvette and electroporated using a Bio-Rad gene pulser at 300 V and 960 μF.

Immediately after electroporation, the cells were transferred to fresh RPMI-1640 medium treated without or with normal or uremic ultrafiltrate solution (10 fold concentrated) collected before or after hemodialysis. Cells were then plated in 12-well dish and treated in triplicates with T₃ 10^-7M or ethanol (vehicle). After 24 h, cells were collected by centrifugation, lysed by the addition of 150 μL 1X lysis buffer (Promega) and assayed for luciferase (kit from Promega Corp.) and β-galactosidase (kit from Tropix, Inc., Bedford, MA) activities. All transfection experiments were performed at least three times.

Data were analyzed by Kruskal-Wallis test followed by Dunn's Multiple Comparison Test when applicable. P < 0.05 was considered statistically significant.