Fifty male 10- to 12-week-old C57BL/6J wild-type mice were randomly assigned, by the random pick of one out of two prepared treatment plans, to the following groups: sham, 5 ng/kg Stx, 10 ng/kg Stx, 15 ng/kg Stx, and 20 ng/kg Stx. Stx from the same stock used in previous experiments (17) purified from an O157:H7 enterohemorrhagic E. coli (EHEC) strain 86-24 patient isolate was used to induce experimental HUS. Five, 10, 15, or 20 ng/kg body weight Stx or 0.9% saline (sham) was injected intravenously on days 0, 3, and 6 of the experiment and mice were monitored three times per day according to the previously described HUS scoring system (17) along with 800 µL of subcutaneous administration of Ringer’s lactate solution for volume replacement. Mice of each group were sacrificed on either day 14 or day 21 after initial Stx injection (day 0). Thereby, the number of individuals implies five mice in each group according to Mayer and Muche’s recommendations (18). Mice were bred in-house under specific pathogen-free conditions in individual ventilated cages (IVCs). For acclimatization phase 1 week before the experiments and during the experimental procedures, mice were kept in open cages with bedding material and tubes as environmental enrichment and under standardized laboratory conditions (22 ± 1°C, 55% ± 10% humidity) with free access to standard rodent chow and water. All treatments and scorings were carried out blinded in consistent order and in a separate room to minimize stress. All animal experiments were approved by the regional animal welfare committee and the Thuringian State Office for Consumer Protection (registration number UKJ-20-018) and were performed in accordance with German legislation and the approved guidelines as well as in compliance with the ARRIVE guidelines. Outcome assessment comparing all groups were carried out blinded and allocations were corrected before statistical analysis.

Whole blood sampling and plasma preparation were carried out as described previously (17). Hemograms were measured immediately after collection using the hematology device scil Vet abc Plus+ (scil animal care company GmbH). Neutrophil gelatinase-associated lipocalin (NGAL), urea, erythropoietin (EPO), ferritin, hepcidin, and lactate dehydrogenase (LDH) activity were determined in plasma of mice using commercial kits according to the manufacturer’s protocol ( Supplementary Table S1 ).

Fresh kidney halves were fixed in 5% buffered formaldehyde solution (Fischar, Germany), dehydrated in descending alcohol solutions and embedded in paraffin (Thermo Fisher Scientific, USA) as previously described (17). FFPE sections of 2 µm thickness were prepared to receive renal cross-sections including areas of cortex and medulla. AZAN after Heidenhain (Morphisto, Germany) staining was performed according to the manufacturer’s instructions. For immunohistochemistry, sections were hydrated using descending alcohol solutions, blocked in 3% hydrogen peroxide (Carl Roth GmbH + Co. KG, Germany), and epitope retrieval was carried out at 110°C for 10 min in target retrieval solution (Dako, Denmark). 1× Tris buffer (10× Tris: 0.5 M Tris, 3 M NaCl, and HCl 37% dropwise until pH 7.6; 1% Tween20) was used as washing buffer. Unspecific antibody binding sites were blocked using bovine serum albumin (BSA; Sigma Aldrich, USA), serum (GE Healthcare, USA), and avidin/biotin solutions (Vector Laboratories, USA). Renal sections were incubated in primary antibody ( Supplementary Table S2 ) at 4°C overnight. The next day, sections were incubated in secondary antibody ( Supplementary Table S3 ); specificity was enhanced by VECTASTAIN Elite ABC HRP Kit (Vector Laboratories, USA) for 30 min each and ImmPACT DAB Peroxidase (HRP) Substrate (Vector Laboratories, USA) was added. Nuclei were counterstained with hematoxylin (Carl Roth GmbH + Co. KG, Germany). Quantification of Ly6G level was performed as described previously (19). Intersitital fibrosis was quantified counting AZAN positive caskets (blue staining) in a grid area (10 × 10 caskets; grid area: 0.097 mm²) for 20 adjacent cortical areas (magnification 400×). CD206 and CD86 abundance was quantified by counting positive grid intersections for 20 adjacent cortical areas (magnification 400×) using the grid mentioned above. Images were taken with the KEYENCE BZ-X800 microscope and BZ-X800 viewer after white balance and auto exposure at a magnification of 400×.

Cross-sections of frozen kidneys (20–30 mg), including tissue of renal cortex and medulla, were homogenized in lysis buffer [1% IGEPAL^® CA-630 (Sigma Aldrich, USA), 50 mM Tris-aminomethan (TRIS), pH 8.0 (Carl Roth GmbH + Co. KG, Germany), 150 mM NaCl (Carl Roth GmbH + Co. KG, Germany), 5 mM NaF (Sigma Aldrich, USA), 1 mM Na₃VO₄ (Sigma Aldrich, USA), and protease inhibitor cocktail 1:100 (Roche Holding, Switzerland)]. After centrifugation for 10 min at 14,000 × g, the total protein amount of the supernatant was determined using Pierce Detergent Compatible Bradford Assay Kit (Thermo Fisher Scientific, USA). Total protein (40 µg) was reduced in Laemmli buffer at 95°C for 5 min and loaded onto 10% TGX Stain-Free FastCast gels (Bio-Rad Laboratories, USA). Transfer, blocking, antibody incubation, and signal development were carried out as described previously (19) using anti-TGF-β (Cell Signaling Technologies, UK, #3711S) and anti-rabbit IgG, HRP-linked antibody (Cell Signaling Technologies, UK, #7074S). To quantify, the Image Lab software was used to normalize abundance of protein of interest to total lane protein visualized by stain-free technology (Bio-Rad Laboratories, USA) instead of a reference protein for loading control (20). When normalization factors exceeded 1.3 or fell below 0.7, bands were excluded from calculations.

Renal tissue lysates were prepared fresh as described in section immunoblotting. Renal erythropoietin (EPO) was determined using a commercial kit according to the manufacturer’s protocol ( Supplementary Table S1 ). Total protein (500 µg) was used to measure cytokines using the Bio-Plex Pro Mouse Chemokine 31-Plex panel assay (Bio-Rad, USA) according to the manufacturer’s instructions.

Data were statistically analyzed using GraphPad Prism 7.05 (GraphPad Software, USA) and are presented as mean standard deviation (SD) for n observations (animals). Gaussian distribution was tested using the Shapiro-Wilk normality test at a significance level of 0.05. When data were normally distributed, unpaired t-test was used to determine differences comparing two groups, one-way ANOVA followed by Sidak’s post-hoc test was used to determine differences comparing more than two groups, or two-way ANOVA followed by Dunnett’s post-hoc test was used to determine differences when all groups were compared, with time and concentration as factors. When data were not normally distributed, Mann–Whitney U test was used to determine differences comparing two groups. Furthermore, values that were three times the SD of the corresponding group were tested for outliers using the ROUT method with Q = 1% and were excluded.

To analyze disease progression phenotypically, weight loss was assessed in mice for a time period of 21 days after HUS induction. Mice with experimental HUS developed a concentration-dependent weight loss compared with sham mice starting from day 4 ( Figure 1A ), which is already known from our well-established 7-day HUS study (17). Average weight loss peaked on day 9 after HUS induction, from which mice gained weight again. Weight loss on day 14 was still significantly higher in mice treated with 10, 15, and 20 ng/kg Stx compared to sham mice. However, the weight of mice with experimental HUS returned to sham level 21 days after HUS induction. Urea and NGAL, both known to be elevated in plasma in the early phase of renal injury, were increased in plasma 14 days after HUS induction, but reduced compared to 7-day HUS mice (19), and returned to sham level 21 days after HUS induction ( Figures 1B, C ). Neutrophils, which represent first-line defender cells and play a pivotal role in the pathology of HUS, had migrated into renal tissue 14 days after HUS induction in mice that received 10 and 20 ng/kg Stx compared to sham mice as indicated by the murine surface marker lymphocyte antigen 6 family member G (Ly6G, Figure 1D ). Twenty-one days after HUS induction, renal neutrophil abundance was comparable to sham mice, independent on the applied Stx concentration. Taken together, weight assessment, renal tissue damage parameter, and acute-phase immune cells in renal tissue indicate that mice recover from the acute HUS disease 21 days after HUS induction.

Hemolytic anemia is a central symptom of HUS. Therefore, parameters reflecting anemia and hemolysis were analyzed in mice 14 and 21 days after HUS induction. Erythrocytes were reduced in mice with experimental HUS 14 and 21 days after HUS induction, but without statistical significance ( Figure 2A ). Hemoglobin and hematocrit levels were both significantly decreased in mice with experimental HUS, when exposed to 20 ng/kg Stx, 14 and 21 days after HUS induction ( Figures 2B, D ). The mean corpuscular volume of erythrocytes was significantly decreased in mice 21 days after HUS induction, irrespective of the applied Stx concentration ( Figure 2C ). EPO level and LDH activity were only measured in mice exposed to the highest Stx concentration based on significantly altered hematocrit and hemoglobin levels. LDH activity in plasma, indicating hemolysis, was increased in HUS mice only 14 days but not 21 days after HUS induction ( Figure 2E ), whereas EPO levels in plasma and renal tissue were unchanged, 14 and 21 days after HUS induction ( Figures 2F, G ). Plasma levels of hepcidin and ferritin, which are involved in resorption and storage of iron, were slightly increased 14 days after HUS induction (hepcidin: with 15 ng/kg and 20 ng/kg, ferritin: with 20 ng/kg) and returned to sham level 21 days after HUS induction ( Supplementary Figure S1 ). Overall, HUS mice exposed to 20 ng/kg Stx develop anemia that transitions from hemolytic to non-hemolytic 21 days after HUS induction, without impaired EPO production, unchanged hepcidin or ferritin levels, but lowered mean corpuscular volume of erythrocytes.

In order to analyze whether Stx can induce long-term consequences in mice other than non-hemolytic anemia, the exacerbated deposition of ECM as a measure of fibrosis development was visualized in renal tissue by AZAN staining. The quantification of the positive interstitial area revealed that Stx induced a concentration-dependent increase in fibrosis 21 days after HUS induction ( Figure 3A ). At day 14 after HUS induction, renal fibrosis was only significantly apparent in mice exposed to 10 ng/kg Stx. The abundance of TGF-β, one of the key driver cytokines in the induction and progression of fibrosis, was analyzed in renal tissue lysates. Since fibrosis was most advanced in mice exposed to 20 ng/kg Stx 21 days after HUS induction, TGF-β levels were only measured in these mice and sham mice at both time points. Because of the unpredictable ribotoxic effects of Stx, stain-free technology instead of loading controls was used for the relative quantification of TGF-β (see Material and Methods). Stx induced an increase of the latent TGF-β as well as the active monomer 14 and 21 days after HUS induction ( Figures 3B, C ). This indicates an ongoing pro-fibrotic signaling cascade even 21 days after HUS induction.

Dependent on the phenotype, macrophages secrete factors mitigating or driving renal fibrosis development. To characterize macrophages in renal tissue, markers for M1 pro-inflammatory phenotype (CD86) and M2 pro-fibrotic phenotype (CD206) were analyzed. At day 14 after HUS induction, macrophages of the M1 and M2 phenotype were present in renal tissue of mice with experimental HUS ( Figures 4A, B ). At day 21 after HUS induction, M1 macrophages were almost absent, whereas M2 macrophages remained highly abundant in a concentration-dependent manner ( Figures 4A, B ). To conclude, 14 days after HUS induction, M1 and M2 macrophages are present in the renal tissue of mice with experimental HUS, whereas 21 days after HUS induction, the predominant phenotypes are M2 macrophages, pointing towards a pro-fibrotic environment.

The acute phase of HUS is associated with and accelerated by excessive cytokine abundance in the kidney (21). A multiplex ELISA was performed on renal tissue lysates to detect the levels of 31 cytokines/chemokines. Because fibrosis and macrophage recruitment were concentration-dependent, only mice exposed to the highest Stx concentration were compared to sham mice. Besides the here mentioned cytokines/chemokines, all other cytokines/chemokines were unchanged compared to the sham group. HUS-associated pro-inflammatory cytokines and chemokines IL-6, IL-1β, GM-CSF, and TNF-α were unchanged in the renal tissue of mice 21 days after HUS induction compared to the sham group ( Figure 5A ). In contrast, concentrations of pro-fibrotic chemokines MCP-1, MCP-5, RANTES, and CCL22 were significantly higher in the renal tissue of mice 21 days after HUS induction compared to the sham group ( Figure 5B ). These chemokines especially attract macrophages, supporting the finding of increased macrophage infiltration into the kidneys of mice 14 and 21 days after HUS induction.

Tissue regeneration after Stx-induced injury possibly explains recovered weight, systemic urea, and NGAL 21 days after HUS induction. Recent studies revealed that NGAL, elevated as early as 2 h after injury (22), is a strong predictor of CKD progression; however, plasma levels in patients with CKD compared to those with AKI are controversial (23). Neutrophils are the first leukocytes accumulating in the renal tissue upon AKI, also through HUS, and their number and activation status correlate with severity (24–26). Although neutrophils promote AKI progression and outcome, their involvement in CKD is not fully understood. The presence of neutrophils has been identified as an indicator of inflammation in CKD; however, the neutrophil-to-leukocyte ratio (NLR) alone could not predict CKD progression (27). In the present study, acute inflammation is decreased 21 days after HUS induction, indicating that neutrophils might be involved in the pathogenesis but not progression of CKD.

The symptomatic trials of HUS includes hemolytic anemia, also detected in our established model of subacute HUS (17). Here, we proved that hemolysis terminated 21 days after HUS induction, since plasma LDH levels declined to sham niveau. However, decreased levels of anemia markers hemoglobin and hematocrit 14 and 21 days after HUS induction with 20 ng/kg Stx indicate a persistent anemia even beyond 21 days. Anemia is a characteristic of CKD that correlates with severity and worsens prognosis (28). Multiple factors can lead to CKD anemia, such as reduced EPO production, EPO resistance, a low erythrocyte life span, and absolute or functional iron deficiency (28). We recently discovered elevated EPO levels in plasma of mice 7 days after HUS induction (29), whereas, here, EPO levels were unchanged in plasma and renal tissue 14 and 21 days after HUS induction. Interestingly, MCV was reduced 21 days after HUS induction, which is accompanied by low erythrocyte counts and/or low hemoglobin diagnostic for microcytic anemia, and the main cause is iron deficiency (30). In this case, although EPO production is normal, responsiveness is reduced due to the iron dependency of erythroblast-to-reticulocyte transition (31). Indicators of iron deficiency anemia besides erythrocyte counts, MCV, and hemoglobin are proteins of the iron metabolism (31). These include hepcidin and ferritin; however, we did not detect lowered plasma abundance of neither hepcidin nor ferritin 21 days after HUS induction, rather comparable to sham level. The pitfall of these proteins is their upregulation induced by inflammatory mediators, which we previously detected in plasma of mice with subacute HUS, for ferritin (32), and here for hepcidin and ferritin after 14 days.

Repetitive damage to renal tissue can lead to CKD with inefficient blood perfusion, kidney function loss, and scarring, called fibrosis (33). Transitioned mesenchymal cells, pericytes, or activated tissue-resident fibroblasts excrete ECM molecules to an excessive and destructive extent (11). We detected interstitial fibrosis histologically by trichrome staining 21 days after HUS induction as the major proof of CKD development. The regulation of fibrogenesis is complex; however, TGF-β is known as the most important cytokine involved in the development and progression of CKD (33). In this study, we measured elevated levels of the latent as well as the active monomer of TGF-β in the kidneys of mice 14 and 21 days after HUS induction.

Macrophages are highly heterogeneous and subpopulation classification is based on their distinct function, origin, and tissue-specific location (34). Macrophages are referred to either classically activated M1 or alternatively activated M2 phenotype (34), identifiable by the distinct expression of cell surface markers such as CD86 primary on M1 and CD206 primary on M2 macrophages (35–38). The M1 phenotype is considered pro-inflammatory because of its capacity to express cytokines like IL-1, IL-6, and TNF-α, which contribute to leukocyte recruitment and activation. M2 macrophages express anti-inflammatory functions and are involved in tissue repair and fibrosis (34, 39). The profound presence of M2 macrophages in the renal tissue has been linked to defective healing and subsequent fibrosis in clinical and preclinical studies: In patients of diabetic nephropathy (DN), M2 macrophage abundance positively correlated with interstitial fibrosis, tubular atrophy and DN class (40). In biopsies of patients with new-onset immunoglobulin A (IgA) nephropathy, M2 macrophages were detected in sites of active fibrosis and cell numbers correlated with mesangial matrix expansion (41). Two further studies investigating renal allograft fibrosis discovered that high numbers of M2 macrophages correlated with alloimmune inflammation, injury, and fibrosis (42, 43). Along with clinical findings, several animal models highlight the role of M2 macrophages in fibrogenesis and progression. For instance, the depletion of M2 macrophages in a long-term model following IRI improved renal fibrosis, whereas adoptive transfer of M2c macrophages partially reversed these beneficial effects (44). Shen and colleagues investigated an M1-to-M2 phenotype switch in a model of UUO, and depletion of M2 macrophages attenuated EMT and subsequent fibrosis (14). In line with these (pre-)clinical findings, we discovered a transition in the intrarenal macrophage phenotype from HUS-induced AKI to CKD long-term state. Parallel to fibrosis development after 21 days, M2 macrophages revealed to be the predominant macrophage phenotype in the kidneys of mice with HUS.

In HUS pathology, leukocyte recruitment and activation contributes to disease progression and severity (21). Pro-inflammatory cytokine release by different cell types induced by Stx include IL-6, TNF-α, IL-1β, and GM-CSF, exacerbating renal injury and sensitivity to Stx (45, 46). In our 7-day model of HUS, we demonstrated the involvement of the classical and non-canonical NFκB pathway, including the renal abundance of the abovementioned cytokines and chemokines (47). Here, we prove that at least after 21 days, these cytokines were no longer released in the kidneys, supporting the hypothesis of an attenuated acute phase. Interestingly, the renal level of cytokines with monocyte/macrophage-chemotactic function was elevated 21 days after HUS induction, highlighting their involvement in CKD. The role of MCP-1 and RANTES in fibrogenesis was evaluated in a mouse model of UUO, in which both were increasingly expressed at sites of progressive fibrosis (48). In addition, Mansour and colleagues featured MCP-1 as a promising renal fibrosis biomarker for prediction and diagnosis (49). Besides its chemotactic capacity, the MCP-1/CCR2 axis has been associated with the transdifferentiation of macrophages to myofibroblasts in a model of Twist1-deficient mice with UUO in which the levels of MCP-1 and CCR2 and consequently renal fibrosis were reduced (50). MCP-5, possessing high similarity to MCP-1, revealed to have a pivotal role in pulmonary fibrogenesis, since cell-specific silencing of MCP-5 or neutralization attenuated pulmonary fibrosis (51, 52). CCL22 also seems to express a role in pulmonary fibrosis, with enhanced abundance in bronchoalveolar lavage fluid and serum of patients and in mice (53). Taken together, the chemokines that were highly abundant in the renal tissue of mice 21 days after HUS induction all possess the capacity to recruit especially macrophages contributing to fibrosis. As the corresponding receptors were proven to be of similar importance in the initiation of fibrotic signaling, their analysis would be of great interest.