The importance of sialic acid, pH and ion concentration on the interaction of uromodulin and complement factor H

Abstract Uromodulin (UMOD) can bind complement factor H (cFH) and inhibit the activation of complement alternative pathway (AP) by enhancing the cofactor activity of cFH on degeneration of C3b. UMOD, an N‐glycans‐rich glycoprotein, is expressed in thick ascending limb of Henle's loop where the epithelia need to adapt to gradient change of pH and ion concentration. ELISA‐based cofactor activity of cFH and erythrocytes haemolytic assay was used to measure the impact of native and de‐glycosylated UMOD on the functions of cFH. The binding assay was performed under different pH and ion concentrations, using ELISA. The levels of sialic acid on UMOD, from healthy controls and patients with chronic kidney disease (CKD), were also detected by lectin‐ELISA. It was shown that removal of glycans decreased the binding between UMOD and cFH and abolished the ability of enhancing C3b degradation. In acidic condition, the binding became stronger, but it reduced as sodium concentration increased. A significant decrease of α‐2,3 sialic acids on UMOD was observed in CKD patients compared with that of healthy individuals. The sialic acids on UMOD, local pH and sodium concentration could impact the binding capacity between UMOD and cFH and thus regulate the activation of complement AP.

ion concentration as well as the state of hypoxia and ischaemia. [13][14][15] UMOD is also a highly glycosylated protein, about 28% of its molecular weight is carbohydrates. There are seven actual N-glycosylation sites, among which Asn275 carries one high-mannose chain, and the other six sites are composed of di-, tri-and tetra-antennary types of N-linked glycans. 16 Sialic acids are the outermost properties of the glycan chains. cFH has specific binding sites to sialic acids, 17,18 and sialic acid-bound cFH has 10-fold affinity for C3b compared with C3b to its activator. 19,20 Thus, we explored the impact of pH, ion concentration and sialic acids on the interaction of UMOD and cFH.

| CKD cohort
In this study, we recruited 17 healthy volunteers and 36 patients with chronic kidney disease (CKD). The average age of the 36 CKD patients was 38.0 ± 12.6 years. The renal diseases included IgA nephropathy, lupus nephritis, ischemic renal disease, ANCA-associated vasculitis, diabetic nephropathy, Alport syndrome, thrombotic microangiopathy, hypertensive kidney disease and glomerulonephritis.
We collected their urine, purified the UMOD to exclude the influence of proteinuria and same dosage of UMOD were used in the experiments to avoid effect of UMOD level from different patients.

| Uromodulin purification
UMOD was purified according to protocol from previous study. 21 Briefly, 1-L urine was mixed with 20 g diatomaceous earth (Celite 521, Acros Organics) for 20 minutes at 4℃ and the mixture was transferred into a funnel with filter paper (Whatman, 15 cm diameter). After filtration, the layer of diatomaceous earth was washed with 0.025 mol/L sodium-phosphate buffer and then mixed with deionized water for 30 minutes at 4℃, and centrifuged at 20 000 g for 30 minutes at 4℃.
The 0.1 mol/L phosphate buffer and NaCl were added into the supernatant to the final concentration of 0.0025 and 0.14 mol/L, respectively. After mixing for 5 minutes, 5-g diatomaceous earth was added, mixed for 20 minutes at 4℃. The mixture was transferred into a funnel with filter paper (Whatman, 7 cm diameter). After filtration, the layer of diatomaceous earth was washed with PBS and taken out to mix with deionized water for 30 minutes at 4℃ and centrifuged at 20 000 ×g for 30 minutes at 4℃. The supernatant was collected for dialysis in deionized water at 4℃ for overnight. After dialysis, the supernatant was concentrated by centrifugation at 3000 rpm using 30 000 NMWL filter (Amicon TM Ultra-15; Milipore), and then UMOD was lyophilized and kept at −80℃ before the experiment.
The purity of UMOD was evaluated with silver stain and western blot ( Figure S1). Purified UMOD was separated in 10% SDS-PAGE gel. After electrophoresis, the gel was silver stained following the procedure described in silver staining kit (Coolaber, Beijing). Purified UMOD was boiled with 5 x loading buffer for 10 minutes and separated in 10% polyacrylamide gel by SDS-PAGE. Proteins were then transferred to polyvinylidene difluoride (PVDF) membranes and blocked for 1 hours in 5% non-fat milk. Then, purified UMOD was detected by polyclonal antibody to human cFH (Cloud-Clone Corp.) and polyclonal antibody to human complement factor I (Cloud-Clone Corp.). After washing, the secondary horseradish peroxidase antibody was used. The membranes were exposed and analysed in GE Image Quant LAS 4000 chemiluminescence imaging analyser.

| Haemolytic assay
The assay was modified according to previous study. 22

| Removal of glycan chains
The PNGase F (P0704) and α2-   was prepared according to previous report. 23

| Statistical analyses
Statistical software SPSS 14.0 (SPSS) was used for statistical analysis. Quantitative data were expressed as mean ± SEM. For normally distributed data, one-way ANOVA was used for comparison of continuous data. For data which did not assume normal distribution, nonparametric tests were used to compare data significance. Statistical significance was considered as P < .05.

| Uromodulin strengthened the function of complement factor H
In this study, we further explored whether the existence of UMOD could accelerate C3b degradation mediated by cFH and factor I. The α-chain (108 kDa) of C3b was cleaved into two fragments at 68 and 43 kDa. The ratio of 43 kDa over 108 kDa indicates the level of C3b degradation. The control group with UMOD alone (without cFH) was set up. With the extension of time, the group without UMOD presented no obvious increase of C3b degradation, but in the group with UMOD, an increase of C3b degradation was observed. The control group with UMOD alone present no degradation of C3b ( Figure 1A, B).
The cFH protects sheep red blood cells (SRBCs) from haemolysis through inhibiting the activation of complement AP. 24 In our test, we set up a system which led to nearly 100% haemolysis of the SRBCs by adding 6μg anti-cFH antibody to the healthy serum. The haemolysis rate was reduced to 50% by adding exogenous 3 μg cFH in the system. When adding UMOD (5, 10 and 20 μg), the haemolysis rate was gradually reduced to 30%. We also set up control groups with various UMOD (5, 10 and 20 μg), but without cFH. The results showed that UMOD could not protect erythrocytes from haemolysis without cFH ( Figure 1C).   (Figure 2A). Under denatured condition, UMOD lost the ability of binding cFH either with or without N-glycans ( Figure 2B). Under native condition, the N-glycan chains were partially removed. When UMOD was treated with neuraminidase A under native condition, there was also a small band shifting, suggested partial removal of sialic acids (Figure 2A).
In order to exclude the influence of PNGase F and neuraminidase A on UMOD and cFH-binding analysis. We used exclusion chromatography to remove the PNGase F (36 kDa) in reaction mixture and silver stain showed that most of the PNGase F was removed ( Figure S2A).
No difference was identified between filtered and not filtered deglycosylated UMOD-cFH binding ( Figure S2B). Filtering with exclusion chromatography lost most of UMOD protein due to the viscosity of UMOD and similar molecular weight of de-glycosylated UMOD and neuraminidase. We tried another control experiment to demonstrate that no difference between the binding strength of UMOD with and without neuraminidase (not incubated at 37℃) to cFH ( Figure S2C).
Under native condition, the removal of N-glycan by PNGase F led to obviously decreased binding of UMOD to factor H ( Figure 2B). We then found that removal of sialic acid components by neuraminidase A also obviously reduced the binding of UMOD to cFH ( Figure 2C).
In cofactor activity assay of cFH, UMOD pretreated with PNGase F lost the ability to enhance C3b degradation compared with controls ( Figure 3A, B).
When UMOD was pretreated by neuraminidase A, the ability of enhancing C3b degradation of UMOD decreased, but in large dose (8 μg UMOD) group, the de-sia UMOD still enhanced the degradation compared with 0 μg UMOD group ( Figure 3C, D).

| Measurement of siaα (2,3) Gal/GalNAc in CKD patients
We recruited 17 healthy controls and 36 CKD patients with average serum creatinine at 212.0 ± 102. 6 μmol/L. We collected their urine, purified the UMOD to exclude the influence of proteinuria.
The same dosage of UMOD was used in the experiments. We meas-   showed that when the concentration of UMOD was constant, the binding ability of UMOD with cFH was increased as pH decreased. That was similar as the assay performed in PBS. Calcium and magnesium ions were removed from artificial urine, and pH of the urine was adjusted by hydrochloric acid. The binding capacity of UMOD and cFH was not influenced by calcium and magnesium ( Figure S3).

| D ISCUSS I ON
We previously found that UMOD could bind with cFH and enhance the cofactor activity of cFH in the cleavage of C3b by cFI. 7 And this work was also proved by Rhodes et al 25 It is well known that cFH is the key regulator of complement AP, and the regulatory domain of cFH locates in N-terminal short consensus repeat 1-4 (SCR1-4), whose function is to bind C3b, and to accelerate the degradation of C3 convertase. 7,26 C3b has α-chain and β-chain, and α-chain (108 kDa) can be cleaved into 68 and 43 kDa. 68 kDa product is stable, while 43 kDa product is increasing with degradation of αchain. 27 The With function tests of cFH, we further demonstrated that UMOD could enhance the cofactor function of cFH. The cFH is an important protective factor of cells through binding cell surface by c-terminals. 28 It can bind to renal tubular epithelial cells.
When ischemia-reperfusion injury occurs, cFH can inhibit the excessive activation of complement and reduce the damage of complement system to epithelial cells. 29 cFH is related to the severity and prognosis of IgA nephropathy. The level of cFH in serum and urine of CKD patients is higher than that of normal people. 30  On the other hand, UMOD is a highly N-glycosylated protein, with sialic acids on the outer part. 16 The carbohydrate structures of UMOD have different functions, such as binding microbial structure, 39 lectin structure and extracellular matrix. 40,41 The high-mannose side chain on UMOD mediates removal of type 1 fimbriated E. coli from urinary tract. 42 In renal allograft recipients, disturbed glycosylation of UMOD was detected and the abnormal glycosylation of UMOD reduced its binding capacity to cytokines, and associated with tubular injury. 5 Several evidences suggest that glycans of UMOD are essential for its immunosuppressive and cytokine-binding activities. [43][44][45] Under denatured condition, UMOD lost the ability of binding with cFH, so the native status was used to measure the influence of glycans. PNGase F can partly remove the glycans without selection. The loss of partial N-glycans of UMOD under native condition, led to significant decrease of its binding with cFH. Previous studies also have found that the C-terminus of cFH is responsible for its binding with sialic acid. 18 The binding can be enhanced by the existence of surface-deposited C3b. 19,20 Thus, sialic acid may be important for cFH to play the role in the context of immune surveillance. We further removed sialic acid on UMOD by neuraminidase A specifically. A decreased binding between UMOD and cFH was observed, and a lower ability of enhancing C3b degradation was also found. In this assay, the fact that no difference was identified between the binding strength of filtered and not filtered de-glycosylated UMOD-cFH suggested that the binding status was not influenced by PNGase F. Because of the viscosity of UMOD, filtering with exclusion chromatography lost a lot of UMOD protein.
The molecular weight of de-glycosylated UMOD is closer to that of neuraminidase; most of the UMOD was lost when separating them from mixed samples. We compared the binding of UMOD to cFH when they were mixed with or without neuraminidase at room temperature.
No difference was found between the two groups. Furthermore, the binding assay was performed in PBS, and PBS was not the working buffer of neuraminidase A and PNGase F, which greatly reduced the effect of them.
We also identified that the level of α-2,3 sialic acids of UMOD decreased in patients with CKD when compared with normal individuals. When CKD develops to four or five stage, patients usually suffer from tubular atrophy, renal fibrosis and decline of renal function, and the correlation between disease progression and primary disease is not as strong as early stage. 46 Therefore, we did not specifically distinguish the primary diseases when we included the patients. This finding indicated that not only level of UMOD but also the status of sialic acid on UMOD may be involved in the progression of CKD. We only identified decreased α-2,3 sialic acids, which is the type of sialic acids that can be recognized by cFH. 47,48 In summary, the binding of UMOD and cFH was regulated by local pH, local sodium concentration and sialic acid richness on UMOD.
Although we detected sialic acid levels in CKD patients, most of our experiments were in vitro, more data from human is needed.

ACK N OWLED G EM ENTS
We appreciate the volunteers participated in our study. The project is supported by the National Natural Science Foundation of China (81570664).

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

CO N S ENT S TATEM ENT
The study was approved by the local ethics committee of Peking University First Hospital (protocol No. 2017(protocol No. [1280). All participants in this study signed written informed consent forms.