Tubular iron deposition and iron handling proteins in human healthy kidney and chronic kidney disease

Iron is suggested to play a detrimental role in the progression of chronic kidney disease (CKD). The kidney recycles iron back into the circulation. However, the localization of proteins relevant for physiological tubular iron handling and their potential role in CKD remain unclear. We examined associations between iron deposition, expression of iron handling proteins and tubular injury in kidney biopsies from CKD patients and healthy controls using immunohistochemistry. Iron was deposited in proximal (PT) and distal tubules (DT) in 33% of CKD biopsies, predominantly in pathologies with glomerular dysfunction, but absent in controls. In healthy kidney, PT contained proteins required for iron recycling including putative iron importers ZIP8, ZIP14, DMT1, iron storage proteins L- and H-ferritin and iron exporter ferroportin, while DT only contained ZIP8, ZIP14, and DMT1. In CKD, iron deposition associated with increased intensity of iron importers (ZIP14, ZIP8), storage proteins (L-, H-ferritin), and/or decreased ferroportin abundance. This demonstrates that tubular iron accumulation may result from increased iron uptake and/or inadequate iron export. Iron deposition associated with oxidative injury as indicated by heme oxygenase-1 abundance. In conclusion, iron deposition is relatively common in CKD, and may result from altered molecular iron handling and may contribute to renal injury.

Proteinuria as a result of glomerular damage in CKD is linked to tubulointerstitial injury 29 , which is associated with increased filtration of TBI 5,9,30 . Increased exposure of TBI can lead to tubular accumulation of reactive iron as a result of inadequate or disturbed iron handling. In diabetic nephropathy, the kidney can also be exposed to non-transferrin-bound iron (NTBI) derived from filtered TBI as a result of acidification of the filtrate as it passes along the nephron 5,31 , or directly filtered from the circulation 32 . NTBI uptake from the tubular lumen is thought to be mediated by ZIP8, ZIP14 and/or DMT1 20 .
For the first time in human kidney, we characterized associations between the presence and localization of iron deposition, proteins involved in cellular iron handling and tubular injury in kidney biopsies of patients with CKD and healthy controls.

Results
Iron deposition in CKD. We found iron deposition in 33% of biopsies from various forms of CKD (n = 126), but not in controls (n = 8; Fig. 1, Table 1). Iron was deposited in a granular pattern in tubular epithelial cells and in the majority of CKD in both PT and DT. In minimal change disease, iron deposition was found in PT only. We detected iron deposition in kidney disorders characterized by nephrotic glomerulopathy (membranous glomerulopathy, focal segmental glomerular sclerosis (FSGS), minimal change disease), glomerulonephritis (Wegener's disease, anti-glomerular basement membrane (GBM) disease), mesangial glomerular expansion (diabetic nephropathy), and potentially mixed nephrotic and nephritic glomerular injury (lupus nephritis (LN), IgA nephropathy (IgAN), hypertensive glomerulopathy) ( Fig. 1b-k). These findings suggest that renal tubular iron deposition is a relatively common phenomenon in kidney disorders with glomerular dysfunction of different natures.
Tubular iron handling proteins in healthy kidney. In controls, ZIP8, ZIP14 and DMT1 were detected in both PT and DT ( Fig. 2a-c, Table 2). ZIP8 was detected at the apical side of the tubules, whereas ZIP14 was localized intracellularly. DMT1 was observed both apically and intracellularly. Interestingly, L-ferritin, H-ferritin and ferroportin were observed only in PT ( Fig. 2d-f, Table 2). Both ferritins showed intracellular orientation while ferroportin was expressed intracellularly and at the basolateral membrane. These findings indicate that PT express proteins for iron import, storage and export, while DT only express proteins involved in iron import in physiological conditions. Tubular iron handling proteins in CKD. To characterize tubular iron handling during proteinuric CKD, a subset of kidney diseases was selected, including diabetic nephropathy, classified as early and advanced (DNE and DNA, respectively), FSGS, LN and IgAN (Table 3). Tubular localization of iron handling proteins in CKD was similar to controls (Figs 3-6). Staining intensity, however, differed between CKD and controls. ZIP14 showed comparable intensity in PT and DT, and, therefore, both segments were analyzed together. ZIP14 intensity was significantly increased in FSGS and IgAN compared to control (both p < 0.001; Fig. 3g-l,s). In FSGS, ZIP14 intensity was increased in two patients only, while all IgAN biopsies showed enhanced ZIP14 staining. ZIP8 and DMT1 showed differential staining between PT and DT, and were, therefore, analyzed separately. ZIP8 intensity was increased in DT in DNE and FSGS compared to control (both p < 0.01; Fig. 3a-f,s), while ZIP8 intensity in PT ( Fig. 3a-f,s) was unchanged. DMT1 intensity in either PT or DT was not different from control ( Fig. 3m-r,s). L-ferritin staining was increased in FSGS (p < 0.001; Fig. 4a-f,m), mainly caused by 2 biopsies, while H-ferritin was overall increased in DNE (p < 0.05) and DNA (p < 0.001; Fig. 4g-l,m). Moreover, ferroportin showed decreased intensity in DNE (p < 0.01) and DNA (p < 0.001) in PT (Fig. 5). Interestingly, for some CKD biopsies, the observed changes in iron handling protein intensity coincided with iron deposition. In FSGS and IgAN biopsies with PT iron deposition, ZIP14 intensity was increased, which was accompanied by increased L-ferritin intensity in FSGS. In DNE and DNA with PT iron deposition, H-ferritin abundance was increased along with  decreased ferroportin intensity. In DT, ZIP8 and/or ZIP14 were increased concurrent with iron deposition in DNE, FSGS and IgAN, but neither were observed in DNA. In contrast, no changes in protein abundance of both tubules were seen in any of the LN biopsies despite iron deposition in 2 of the biopsies. These findings are summarized in Fig. 6.   In conclusion, our findings show associations between tubular iron deposition and abundance of iron handling proteins in most types of CKD, which differ between specific pathologies. Overall, tubular iron deposition was related to increased iron import (ZIP8, ZIP14) in PT and DT, in some CKD biopsies accompanied by increased storage (ferritin) or decreased export (ferroportin) in PT.
Tubular injury in CKD. Tubular injury was assessed with PAS and heme oxygenase-1 (HO-1) staining. We used PAS staining to assess gross histology and scored renal injury based on tubular atrophy, blebbed tubular structures and irregular tubular cytoplasm (Fig. 7a-d). Renal damage was increased in DNE and DNA compared to control and moderately elevated in FSGS (Fig. 7e). HO-1, a marker for oxidative cellular stress 33 , showed comparable intensity in both PT and DT, which was increased in CKD (control < DNE = DNA < IgAN < FSGS = LN; Fig. 8). Increased HO-1 staining coincided with iron deposition in DNE, FSGS, LN and IgAN, but was also induced in DT of DNA without iron deposition. In conclusion, we found oxidative cellular stress, but not histological damage, to associate with iron deposition.

Discussion
Increased iron accumulation may be a detrimental factor in progression of CKD, but the mechanisms of iron handling in the human kidney are not fully elucidated. In this study we found iron deposition in one third of biopsies from various CKD disorders, predominantly in pathologies with glomerular dysfunction. We observed that in the healthy kidney, PT contain proteins involved in iron import, storage and export, while DT only showed proteins involved in iron import. Associations between iron deposition, intensity of iron handling proteins and tubular injury in CKD were seen, which differed between the pathologies. Nevertheless, in the majority of CKD, tubular iron deposition was accompanied by an increase in iron import proteins ZIP8 and/or ZIP14 in both PT and DT. This coincided with an increase in iron storage proteins or decrease in iron exporter in PT and increased oxidative cellular injury in both PT and DT.
Our data show that iron deposition is a relatively common phenomenon in CKD with glomerular dysfunction. In physiological conditions, TBI filtered by the glomerulus is believed to be almost completely reabsorbed by the tubular epithelium. In this process, PT reabsorb the bulk of filtered proteins and DT play only a minor role 20 . We hypothesize that in nephropathic glomerulopathy or pathologies with mesangial glomerular damage, PT iron reabsorption is increased by large amounts of TBI leaking from the injured glomerulus, resulting in local iron deposition. Moreover, when the reabsorption capacity of PT is overwhelmed, DT might also be exposed to increased iron levels, leading to DT iron deposition. In addition, we found iron deposition in pathologies with nephritic glomerular damage, which present with both hematuria and proteinuria 34 . Our results complement previous studies of renal iron deposition in CKD biopsies 11,12 , which was localized to PT lysosomes 18,19 .
In our study, we demonstrated iron deposition also in DT in CKD. In addition, we show that iron deposition relates to oxidative cellular injury, as assessed by HO-1 staining. This strengthens the supposition that iron accumulation facilitates highly reactive radical formation that damage membranes, proteins and DNA, and, subsequently, causes tissue injury, which has been reported in renal tubular cells and animal models of CKD 6-9,35-37 . Interestingly, iron reduction via a low-iron diet or treatment with an iron chelator has been reported to reduce renal iron accumulation and tubulointerstitial injury in various CKD animal models [35][36][37] . Together with findings of increased urinary iron levels and renal iron deposition in patients with CKD [11][12][13][14][15][16][17][18][19] , this suggests that renal iron loading could contribute to disease progression in patients with CKD. Biopsies used in this study were taken for diagnostic purposes and are usually at a relatively early stage in disease progression, which is reflected by their tubular injury score. However, renal injury was already more established at the time of biopsy for diabetic nephropathy, especially for DNA. Nevertheless, our findings of iron deposition and associated oxidative cellular injury in FSGS, LN and IgAN biopsies suggest that iron may already be involved in the onset of renal disease and contribute to CKD progression.
We are the first to show ZIP14 and DMT1 in human kidney, whereas ZIP8 was previously described in human kidney 38 . Until now, DMT1 and ZIP14 were only demonstrated in murine PT, and DMT1 and ZIP8 in only mouse and rat DT [39][40][41][42][43][44][45] . Luminal TBI reabsorption is suggested to involve TfR1, but TfR1 localization, on the apical or basolateral membrane of tubular epithelial cells, is debated 31,41,[46][47][48] . Since we did not obtain reliable TfR1 stainings, our study is not able to add to the understanding of TfR1 localization. Following entry in the cytosol, iron is known to be oxidized by H-ferritin and stored in L-ferritin. Our results of intracellular expression of both ferritins in PT agree with other studies 48,49 . Cytosolic iron is suggested to be transported back to the systemic circulation via the exporter ferroportin 20 . Although both apical and basolateral localization of ferroportin are shown in murine PT 31,41,[46][47][48] , our human kidney results clearly demonstrate basolateral localization only. This discrepancy might be related to species difference 50 . In our studies, ferroportin was absent in DT, as is shown in rats 46 . The absence of ferritin and ferroportin in DT support that DT only play a minor role in physiological TBI reabsorption. Lack of effective iron storage and export could make these cells vulnerable to iron accumulation and related injury in case of high iron exposure.
For the majority of CKD disorders, we could distinguish an overall increased intensity of ZIP8 or ZIP14 that might have contributed to iron deposition. TfR1 and DMT1 are known to protect cells by limiting iron uptake via iron responsive element-iron responsive protein (IRE-IRP) regulation 51 , but this is not described for ZIP8 and ZIP14 52 . Interestingly, ZIP8 and ZIP14 increase with iron loading in hepatocytes 28,53 . Therefore, ZIP8 and ZIP14 are plausible candidates for unrestricted iron import and subsequent iron loading in kidney tubular epithelium from either endocytosis TBI or direct NTBI transport over the apical membrane 5,31,32 . Future assessment of the intracellular localization and function of ZIP8 and ZIP14 would be valuable to dissect potential iron transport routes in CKD.
We observed decreased ferroportin and concomitant increased H-ferritin intensity in diabetic nephropathy, which may explain PT iron deposition in these patients. Based on IRE-IRP regulation, however, we would expect ferroportin to increase with cellular iron loading 51 . At the systemic level, ferroportin is regulated by hepcidin, which causes ferroportin degradation 54 . Also renal ferroportin protein abundance is shown to decrease with high circulating hepcidin levels 55 . Interestingly, elevated hepcidin levels have been reported in patients with type 2 diabetes in the presence of chronic renal disease, obesity or inflammation 56,57 , which could explain the observed reduction in ferroportin intensity. Unfortunately, serum hepcidin levels were not available in our study. Moreover, renal biopsies in our study were obtained from patients with either type 1 or type 2 diabetes and the former patients do not have increased hepcidin levels 58 . Our findings of an association between decreased ferroportin and increased H-ferritin are corroborated by studies in human macrophages in which ferroportin silencing led to increased H-ferritin protein levels 59 , suggesting that H-ferritin upregulation could result from decreased ferroportin. Conversely, others showed a reciprocal relationship between H-ferritin and ferroportin by reporting reduced ferroportin mRNA and protein levels in conditional PT H-ferritin knockout mice 48 . Ferroportin is regulated by several multilayered signals, where hepcidin is reported to have a dominant effect 54 . This makes it difficult to determine whether renal iron deposition is due to ferroportin decreases or merely results from other mechanisms. Moreover, because of its multilayered regulation, the contribution of ferroportin to tubular iron loading may be different for various kidney diseases as observed in our study.
Besides ferroportin, also other cellular proteins involved in renal iron handling are regulated by processes beyond iron metabolism, including inflammation and oxidative stress 54,60,61 . These processes are also believed to contribute to the pathophysiology of CKD 62 . Therefore, changes in abundance of proteins we studied cannot solely be attributed to changes in iron handling. Vice versa, iron loading was observed in LN without changes in protein abundance. In these patients, iron accumulation could be mediated by other potential iron transporters that we were not able to include in this study, such as TfR1, the megalin-cubilin receptor complex or NGALR [22][23][24][25] . This underlines the complex mechanisms involved in the various CKD pathologies, which may (to various extents) all contribute to changes in renal iron handling proteins. In summary, our findings in human renal biopsies form the basis for further elucidating renal iron handling in health and disease. Future studies should focus on unraveling the molecular mechanisms of renal iron loading in individual CKD disorders. This will determine whether and how reduction of renal iron accumulation in CKD is a feasible target to halt disease progression.

Methods
Design. Biopsies from patients with CKD and controls were used to examine the presence and localization of iron deposition. Subsequently, a selection of prevalent kidney diseases 63 and controls was made to examine proteins involved in cellular iron handling and tubular injury.
Patients. Renal biopsy material was collected from patients with CKD and potential kidney donors (healthy controls). Biopsies were classified based on clinical presentation of the patient and histological assessment by a trained renal pathologist. Biopsies were collected in University Medical Centre Groningen (the Netherlands (NL)) and Royal Free Hospital (RFH; London, United Kingdom (UK)), during the period of 1996-2014 and 2011-2016, respectively. Biopsies from both hospitals were used to assess iron deposition. For assessment of iron handling proteins and tubular injury, CKD biopsies and healthy kidneys from RFH only were selected, based on tissue availability aiming at age-and gender-matching. Biopsies with excessive tissue damage ( >70% of tissue) were excluded.
Procedures and use of anonymized material left over from diagnostic care were performed according to Dutch ethical guidelines. Accordingly, this study did not require approval from an ethical committee and material from patients from UMCG could be used without informed consent. All patients from RFH signed informed consent. All material was collected according to the Declaration of Helsinki.
Histological stainings. Biopsy tissue was embedded in paraffin and cut into 5 µm sections. Renal histology was assessed by Periodic acid-Schiff (PAS) staining and iron deposition by Perls' staining, according to routine staining protocols.
Statistical analysis. Data were analyzed by one-way ANOVA with Dunnett's post test using GraphPad Prism, to evaluate staining intensity and tubular injury in CKD compared to control. Differences were considered statistically significant when p < 0.05.