Hypoxic human proximal tubular epithelial cells undergo ferroptosis and elicit an NLRP3 inflammasome response in CD1c+ dendritic cells

Inflammasomes are multiprotein platforms responsible for the release of pro-inflammatory cytokines interleukin (IL)-1β and IL-18. Mouse studies have identified inflammasome activation within dendritic cells (DC) as pivotal for driving tubulointerstitial fibrosis and inflammation, the hallmarks of chronic kidney disease (CKD). However, translation of this work to human CKD remains limited. Here, we examined the complex tubular cell death pathways mediating inflammasome activation in human kidney DC and, thus, CKD progression. Ex vivo patient-derived proximal tubular epithelial cells (PTEC) cultured under hypoxic (1% O2) conditions modelling the CKD microenvironment showed characteristics of ferroptotic cell death, including mitochondrial dysfunction, reductions in the lipid repair enzyme glutathione peroxidase 4 (GPX4) and increases in lipid peroxidation by-product 4-hydroxynonenal (4-HNE) compared with normoxic PTEC. The addition of ferroptosis inhibitor, ferrostatin-1, significantly reduced hypoxic PTEC death. Human CD1c+ DC activated in the presence of hypoxic PTEC displayed significantly increased production of inflammasome-dependent cytokines IL-1β and IL-18. Treatment of co-cultures with VX-765 (caspase-1/4 inhibitor) and MCC950 (NLRP3 inflammasome inhibitor) significantly attenuated IL-1β/IL-18 levels, supporting an NLRP3 inflammasome-dependent DC response. In line with these in vitro findings, in situ immunolabelling of human fibrotic kidney tissue revealed a significant accumulation of tubulointerstitial CD1c+ DC containing active inflammasome (ASC) specks adjacent to ferroptotic PTEC. These data establish ferroptosis as the primary pattern of PTEC necrosis under the hypoxic conditions of CKD. Moreover, this study identifies NLRP3 inflammasome signalling driven by complex tubulointerstitial PTEC-DC interactions as a key checkpoint for therapeutic targeting in human CKD.


INTRODUCTION
Chronic kidney disease (CKD) is a rapidly increasing healthcare burden, affecting~275 million people globally [1]. Irrespective of its aetiology or its associated co-morbidities, the key pathobiological features of CKD are inflammation and fibrosis within the tubulointerstitial compartment [2,3]. The structural and pathophysiological changes that arise from this tubulointerstitial damage generate a localised hypoxic environment [4,5]. Prolonged hypoxia triggers additional pathobiology that includes tubular cell death/necrosis, resulting in the release of damageassociated molecular patterns (DAMP). Tubulointerstitial immune cells, such as dendritic cells (DC), detect these DAMP via a suite of pattern recognition receptors, including nucleotide-binding oligomerization domain (NOD)-like receptors (NLR) [6], becoming activated and driving pro-inflammatory processes that culminate in ongoing tubulointerstitial fibrosis [7,8]. Despite the high prevalence of CKD, the specific tubular cell death-immune cell interactions that drive this hypoxic-mediated inflammatory loop remain poorly defined in humans.
Proximal tubular epithelial cells (PTEC) of the kidney are central players in CKD [9]. PTEC energy is primarily generated through mitochondrial fatty acid oxidation, an oxygen-intensive process [10,11]. Thus, PTEC are particularly susceptible to hypoxic-driven mitochondrial dysfunction and cell death in CKD [12]. Studies have previously shown that hypoxic PTEC, in concert with pro-inflammatory factors (e.g., interleukin (IL)-18), significantly enhance pathogenic immune cell activation and cytotoxic licensing [13]. Supporting this concept, elevated levels of IL-1β and IL-18 are reported in human fibrotic kidney tissue [13]. Both cytokines are members of the IL-1 family of proinflammatory cytokines [14,15]. However, the functional contribution of hypoxic PTEC death in driving IL-1β/IL-18 release within the diseased tubulointerstitium and the immunological source of these pro-inflammatory cytokines in human CKD remain elusive.
The secretion of active IL-1β and IL-18 is controlled by inflammasomes, cytosolic multiprotein response and signalling platforms [16][17][18]. The NLRP3 (nucleotide-binding domain and leucine-rich repeat-containing family, pyrin domain-containing 3) inflammasome is the best characterised of the inflammasomes K.T.K. Giuliani et al. [19]. Canonical NLRP3 inflammasome activation by DAMP signals involves both priming and activation steps that drive the formation of the inflammasome-caspase-1 holoenzyme. The formation of the inflammasome-caspase-1 complex is mediated by the adaptor protein ASC (apoptosis-associated speck-like protein containing a caspase activation and recruitment domain), which facilitates the recruitment and autolytic cleavage of caspase-1 on the inflammasome platform [19,20]. Once cleaved, the active caspase-1 drives the proteolytic maturation and subsequent release of active IL-1β and IL-18 [20].
In this study, we demonstrate that patient-derived primary PTEC cultured under hypoxic conditions modelling the human CKD microenvironment: (1) exhibit pathobiological features of ferroptotic cell death; and (2) trigger NLRP3 inflammasome-dependent IL-1β/IL-18 secretion in CD1c + DC. Moreover, we confirm these pathogenic cell-cell interactions in human fibrotic kidneys, identifying a significant accumulation of tubulointerstitial CD1c + DC containing active inflammasome (ASC) specks adjacent to ferroptotic PTEC. Our data establish NLRP3 inflammasome signalling driven by complex tubulointerstitial PTEC-CD1c + DC interactions as a novel clinical target for therapeutic translation in human CKD.

RESULTS
Human primary PTEC display mitochondrial dysfunction and DNA damage under hypoxic conditions We have previously established a pathogenic function for human primary PTEC under conditions of hypoxia, an established driver of CKD [13,31]. We extend on these previous reports to examine, for the first time, the pathobiological profile of human primary PTEC cultured under hypoxic (1% O 2 ) versus normoxic control (21% O 2 ) conditions. Hypoxic PTEC displayed significantly elevated mitochondrial superoxide (O 2 • − ) production (Fig. 1A), significantly reduced mitochondrial function (↓ mitochondrial membrane potential; ΔΨ mt ) (Fig. 1B), significantly increased expression of the DNA damage marker γ-H2AX (Fig. 1C) and significantly reduced cell viability (Fig. 1D) compared with normoxic conditions. Examination of hypoxic PTEC culture supernatants showed significantly elevated levels of high mobility group box protein-1 (HMGB1) (Fig. 1E), a pro-inflammatory DAMP released after necrotic cell damage [32].
Co-localisation of ferroptotic PTEC with CD1c + DC in fibrotic kidney tissue To translate these in vitro data to an in situ model, we performed immunohistochemical (IHC) staining of human fibrotic kidney tissue. PTEC in control/non-fibrotic and fibrotic kidney tissue were identified based on the expression of aquaporin-1 (AQP-1) (Fig. 3A), a marker restricted to the proximal regions of the tubular compartment [34]. Serial section analysis showed a selective reduction in GPX4 levels and increased 4-HNE in AQP-1 + PTEC within fibrotic kidney tissue (highlighted with black arrows) compared with control samples (Fig. 3B, C), supporting the concept of ferroptotic PTEC death in tubulointerstitial fibrosis. Quantitative analysis confirmed these findings, with significantly decreased GPX4 expression and significantly elevated 4-HNE in fibrotic tissue compared with control tissue (Fig. 3B, C).
Moreover, we showed a significantly increased tubulointerstitial accumulation of CD1c + cells (with dendritic morphology) in Fig. 1 Hypoxia induces pathways of mitochondrial oxidative damage/dysfunction and DNA damage in human primary PTEC. A Left panel: fold changes (relative to normoxia) in mitochondrial superoxide levels (% MitoSOX + cells) for PTEC cultured under normoxic and hypoxic conditions. Bar graphs represent mean ± standard error mean (SEM). Symbols represent individual donor PTEC; n = 5. *P < 0.05, paired t test. Right panel: representative MitoSOX staining (black unfilled) compared with unstained control (grey filled) for PTEC cultured under normoxic and hypoxic conditions. Mitochondrial superoxide levels (% MitoSOX + cells) are presented for each histogram, with fold change (FC) value relative to normoxic PTEC also shown. B Left panel: Fold changes (relative to normoxia) in mitochondrial membrane potential [measured as ratio of ΔMFI JC-1 red/ΔMFI JC-1 green; with delta median fluorescence intensity (ΔMFI) representing MFI test -MFI unstained control] for PTEC cultured under normoxic and hypoxic conditions. Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC; n = 4. *P < 0.05, paired t test. Right panel: representative JC-1 dot plots of PTEC cultured under normoxic and hypoxic conditions. Mitochondrial membrane potential (ΔΨ mt ) values are presented for each histogram, with fold change (FC) value relative to normoxic PTEC also shown. C Left panel: fold changes (relative to normoxia) in γ-H2AX protein levels (as a ratio of loading control β-tubulin) for PTEC cultured under normoxic and hypoxic conditions. Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC; n = 4. *P < 0.05, paired t test. Right panel: western blot for γ-H2AX for PTEC cultured under normoxic and hypoxic conditions (15 µg total protein per lane). Representative images from one of four donor PTEC are presented; full and uncropped western blot available as Supplementary Material. D, E Fold changes (relative to normoxia) in cell viability (MTT assay) (D) and HMGB1 (ELISA) (E) for PTEC cultured under normoxic and hypoxic conditions. Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC; n = 7 for viability data and n = 3 for HMGB1 data. **P < 0.01, paired t test.
fibrotic kidney tissue (Fig. 3D). Notably, these CD1c + DC were both localised to sites of tubulointerstitial injury and adjacent to ferroptotic PTEC (Fig. 3D). These collective findings provide evidence of spatial co-localisation of ferroptotic PTEC with CD1c + DC in tubulointerstitial fibrosis.
CD1c + DC activated in the presence of hypoxic PTEC produce significantly increased levels of inflammasome cytokines IL-1β/IL- 18 We established an in vitro co-culture model to examine the mechanistic interactions which could occur between adjacent K.T.K. Giuliani et al.
PTEC and CD1c + DC in fibrotic kidneys. We isolated human CD1c + DC by immuno-magnetic selection/flow cytometry sorting (>99% purity) and examined their pro-inflammatory cytokine profiles following co-culture with pre-conditioned normoxic (21% O 2 ) or hypoxic (1% O 2 ) PTEC in the absence or presence of poly I:C, a double-stranded RNA (dsRNA) analogue and TLR3 agonist. Poly I:C is a physiologically relevant activator, with TLR3 reported to be a critical sensor of endogenous RNA released during tissue necrosis [35,36].
In the absence of poly I:C stimulation, no to minimal changes in soluble cytokine levels were detected by LEGENDplex™ assay in the supernatants of either normoxic or hypoxic PTEC-DC cocultures (or PTEC/DC alone control cultures) ( To enable intracellular localisation/detection of IL-1β, hypoxic PTEC-DC co-cultures (in the presence of poly I:C) were pre-treated with either protein transport inhibitor brefeldin A to block endoplasmic reticulum/golgi complex transport of inflammasome components or caspase-1/4 inhibitor VX-765. Both modes of inhibition showed IL-1β expression in hypoxic co-cultures restricted to CD1c + DC (~15%), with no detectable levels in PTEC (Fig. 4D). These data point to CD1c + DC as a novel source of IL-1β within the hypoxic microenvironment of human CKD.
The addition of ferrostatin-1 to inhibit hypoxic PTEC ferroptosis elicited a partial attenuation of this CD1c + DC-derived IL-1β, with reduced cytokine levels in co-culture supernatants in 4/7 donor PTEC experiments (Fig. 4E). We conclude that although PTEC ferroptosis plays a role in CD1c + DC IL-1β production, there are other, dominant pathways (i.e., inflammasome signalling) in the complex and multifactorial PTEC-DC crosstalk of human CKD.
Hypoxic PTEC trigger CD1c + DC-derived IL-1β/IL-18 secretion via NLRP3 inflammasome activation Mechanistic investigations were undertaken to confirm the functional role of inflammasome-dependent signalling, in particular, NLRP3 inflammasome activation, in this human CKD model. The expression of CD1c + DC NLRP3 mRNA was evaluated in poly I:C-stimulated cultures. Poly I:C stimulation of normoxic or hypoxic DC alone failed to induce NLRP3 mRNA expression (Fig. 5A). This is in line with previous studies establishing extracellular poly I:C as a weak activator of the NLRP3 inflammasome [37,38]. In contrast, significantly increased NLRP3 mRNA was detected in CD1c + DC from hypoxic co-cultures (Fig. 5A). These data suggest that hypoxic PTEC-derived DAMP trigger CD1c + DC via activation of the NLRP3 inflammasome.
Significantly elevated tubulointerstitial CD1c + DC with ASC specks in fibrotic kidneys A hallmark of NLRP3 inflammasome activation is the ASC speck, a cytosolic supramolecular aggregate of the inflammasome adaptor protein ASC [41]. We demonstrated, for the first time, tubulointerstitial CD1c + DC with punctate ASC specks in fibrotic kidney tissue. These ASC-speck-positive CD1c + DC were positioned in close proximity to AQP-1 + PTEC (Fig. 6A, B). Quantitative analysis confirmed a significantly increased accumulation of ASC-speckpositive CD1c + DC in fibrotic kidney tissue compared with control/ non-fibrotic tissue (Fig. 6C), with 14.52 ± 8.18% of CD1c + DC containing ASC specks. Of note, this DC population was identified as a key cellular source of ASC specks in fibrotic kidneys, with 31.61 ± 16.23% of total ASC-speck-positive cells identified as CD1c + DC. Collectively, these findings support a specialised pathobiological role for PTEC-DC crosstalk in triggering inflammasome activation in human CKD.

DISCUSSION
Pro-inflammatory cytokines IL-1β and IL-18 are potent molecules in the propagation of tubulointerstitial hypoxia/fibrosis, the pathological hallmarks of CKD [13,42]. Here, we present data identifying CD1c + DC as a key source of IL-1β/IL-18 in human CKD, mediated via complex NLRP3 inflammasome-dependent interactions with damaged PTEC within the hypoxic microenvironment.
We established an in vitro model of tubulointerstitial hypoxia to dissect the mechanistic interplay between co-localised PTEC and CD1c + DC in human CKD. Under hypoxic conditions, our human primary PTEC displayed evidence of mitochondrial dysfunction (↑ mitochondrial O 2 • − , ↓ ΔΨ mt ) leading to DNA damage (↑ γ-H2AX). The pathobiological link between Fig. 2 Hypoxia induces ferroptosis in human primary PTEC. A Left panel: fold changes (relative to normoxia) in cellular necrosis (% Annexin-V + PI + cells) for PTEC cultured under normoxic and hypoxic conditions. Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC; n = 4. *P < 0.05, paired t test. Right panel: representative donor Annexin-V/PI dot plots of PTEC cultured under normoxic and hypoxic conditions. The percentage of Annexin-V + PI + necrotic cells for each dot plot are presented, with fold change (FC) value relative to normoxic PTEC also shown. B Left panel: fold changes (relative to normoxia) in GPX4 protein levels (as a ratio of loading control β-tubulin) for PTEC cultured under normoxic and hypoxic conditions. Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC; n = 4. **P < 0.01, paired t test. Right panel: GPX4 western blot for PTEC cultured under normoxic and hypoxic conditions (15 µg total protein per lane). Representative images from one of four donor PTEC are presented; full and uncropped western blot available as Supplementary Material. C Left panel: fold changes (relative to normoxia) in 4-HNE levels (measured as mean corrected total cellular fluorescence (CTCF) of >60 cells per condition) for PTEC cultured under normoxic and hypoxic conditions. Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC; n = 6. *P < 0.05, paired t test. Right panel: immunofluorescent labelling of representative PTEC cultured under normoxic and hypoxic conditions and stained for 4-HNE (green), β-tubulin (red) and DAPI (blue). 4-HNE positivity is highlighted with white arrows. Scale bars represent 20 µm. D Left panel: representative donor bar graph of cellular necrosis for PTEC cultured under normoxic (N) and hypoxic (H) conditions in the absence (-Fer-1; DMSO vehicle control) or presence of ferrostatin-1 (+Fer-1). Right panel: fold changes (relative to hypoxia/-Fer-1) in cellular necrosis (% Annexin-V + PI + cells) for hypoxic PTEC cultured in the absence or presence of ferrostatin-1. Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC; n = 7. *P < 0.05, paired t test.
mitochondrial oxidative stress and impaired mitochondrial function has been established in experimental CKD models and also kidney tissue from CKD patients [43][44][45][46][47]. Moreover, a recent study of kidney tissue from CKD patients reported extensive PTEC DNA damage, marked by phosphorylation of the histone H2A variant H2AX (γ-H2AX) [48]. The identification of these shared pathobiological pathways in our hypoxic human primary PTEC validates our in vitro model as a clinically relevant system for examining downstream tubulointerstitial disease processes (i.e., tubular cell death) observed in CKD patients.
If unresolved, sustained mitochondrial dysregulation triggers pathways of regulated cell death (necrosis). We demonstrated the selective induction of PTEC ferroptosis (↓ GPX4, ↑ 4-HNE) both in our in vitro hypoxic model of human CKD and in situ within fibrotic kidney tissue. Ferroptotic cell death is an ironand reactive oxygen species (ROS)-dependent necrosis [33]. Ferroptotic tubular cell death has been reported in human kidney biopsy specimens of diverse aetiologies, including acute tubular injury patients [49,50]. Our present study extends these findings to the hypoxic microenvironment of human CKD. Hypoxia can promote ferroptosis by damaging mitochondrial membranes, leading to elevated intracellular levels of (i) hydrogen peroxide (H 2 O 2 )-generated by dismutation of excess mitochondrial superoxide (O 2 • − ); and (ii) labile/free iron released from mitochondrial iron-containing proteins (e.g., cytochrome c) [51]. Subsequent iron-H 2 O 2 interactions yield highly toxic free radicals-hydroxyl radicals (OH • )-via the Fenton reaction [52]. These free radicals react with polyunsaturated fatty acids to generate lipid peroxides [53]. The lipid repair enzyme GPX4 can reduce these toxic lipid peroxides to their alcohol form [54]. A deficiency in GPX4 function/expression therefore results in an accumulation of lipid peroxides and their degradation products (i.e., oxidised phospholipids and 4-HNE) that, in turn, trigger ferroptotic cell death [55]. Notably, increased plasma levels of 4-HNE have been reported in CKD patients [56]. Here, we map this lipid peroxidation by-product to ferroptotic PTEC under in vitro hypoxic conditions and in human fibrotic kidney tissue.
Our findings substantiate the "wave of death" hypothesis in which intercellular propagation of ferroptotic cell death, and its release of pro-inflammatory DAMP, accounts for ongoing kidney tubular injury and loss of nephrons [57,58]. In particular, Ide et al. recently reported that 'damage associated' PTEC display both a dysregulation in the glutathione/GPX4 anti-ferroptotic defence pathway and promote a pro-inflammatory milieu via cell-cell signalling to myeloid cells [59]. Our study identifies tubulointerstitial CD1c + DC as the critical myeloid cell population in human CKD, co-localising with ferroptotic PTEC in fibrotic kidney tissue and sensing PTEC-derived DAMP via the NLRP3 inflammasome. ASC is a key adaptor protein with a pivotal role in NLRP3 inflammasome assembly and activation [19]. Following inflammasome activation, ASC assembles into a large protein complex or 'speck' [60]. ASC specks are released from inflammasomeactivated cells and accumulate in extracellular bodily fluids in human chronic inflammatory diseases [41,60,61]. Of particular relevance, ASC in the urine of CKD patients positively correlates with the concentration of proteinuria, a biomarker of kidney injury [62]. Our study confirms tubulointerstitial CD1c + DC as a key immunological source of ASC specks within human fibrotic kidney tissue.
Our in vitro interrogation of the danger signals mediating CD1c + DC inflammasome activation highlights the complexity of this cell-cell interaction. Firstly, triggering of CD1c + DC with TLR3 agonist poly I:C was required to elicit an NLRP3 inflammasome response in our CKD model. TLR3 has been recently implicated in kidney fibrosis and inflammation via activation of the transforming growth factor (TGF)-β/Smad/ TLR3 signalling pathway [63]. TLR3 recognition of self dsRNA released from necrotic cells also induces sterile inflammation [35,36], with poly I:C representative of these endogenous RNA DAMP in our in vitro model of tubulointerstitial hypoxia. Most notably, extracellular poly I:C is a weak to non-activator of the NLRP3 inflammasome [37,38], and was thus selected as a stimulus in our CKD model to trigger DC without masking the effect of inflammasome-activating DAMP from hypoxic PTEC. In line with these previous reports, the addition of poly I:C alone failed to induce CD1c + DC NLRP3 mRNA expression in our system, but importantly, sensitised the DC to subsequent NLRP3 inflammasome triggering by hypoxic PTEC.
Cellular products of lipid peroxidation (i.e., oxidized phospholipids) have been identified as inflammasome-activating DAMP [64][65][66]. However, pre-treatment of hypoxic PTEC with ferroptosis/ lipid peroxidation inhibitor ferrostatin-1 only partially attenuated downstream CD1c + DC IL-1β production (in 4/7 co-culture experiments), despite significantly blocking necrotic cell death in these PTEC. We conclude that there are alternate hypoxic PTECderived DAMP in addition to those directly associated with the ferroptotic cell death pathway. The functional role of other established inflammasome-activating DAMP (i.e., HMGB1 and ATP) was also examined. Accumulating evidence implicates both HMGB1 and ATP signalling pathways in the development of tubulointerstitial fibrosis [67][68][69][70]. However, inhibition of HMGB1 and extracellular ATP did not significantly modulate NLRP3 inflammasome activation in CD1c + DC. Collectively, these data are suggestive of redundancies in PTEC-derived DAMP signalling to CD1c + DC, mediated by complex and highly diverse inflammasome-activating stimuli. The corollary is that therapeutics targeting this component of the NLRP3 inflammasome-activation pathway in CD1c + DC are likely to be unsuccessful in CKD.
Targeting IL-1β/IL-18 signalling downstream of inflammasome activation has been assessed as an alternate strategy in preclinical and clinical studies of CKD. Of these two proinflammatory cytokines, IL-1β is currently the prime focus of immunotherapeutic strategies targeting the inflammasome. The IL-1 receptor (IL-1R) antagonist, anakinra, was shown to be effective in the treatment of gout flares in patients with advanced CKD [71]. However, the investigators reported no significant improvement in kidney function [71]. Similarly, in high-risk atherosclerosis patients with CKD, treatment with canakinumab, an IL-1β neutralising antibody, had no beneficial effect on kidney function, but significantly reduced major adverse cardiovascular events [72]. We propose that effective renoprotection requires blocking both IL-1β and IL-18 by targeting upstream of their activation (i.e., the NLRP3 inflammasome with MCC950). The efficacy of sulfonylurea-containing compound MCC950 was evaluated in a phase II clinical trial of rheumatoid arthritis-however, this clinical study was suspended due to off-target hepatotoxicity [73]. Attention has since focussed on the development of less toxic MCC950 analogues, with IZD-334 (NCT04086602) and inzomelid (NCT04015076) both in clinical trials for the treatment of cryopyrin-associated periodic syndrome (CAPS) [74]. Our study provides the proof of concept for repurposing of such MCC950-related compounds for the therapeutic targeting of CD1c + DC in CKD and other inflammatory kidney diseases.  ). B, C Secreted IL-1β (B) and IL-18 (C) protein levels (measured by LEGENDplex™ assay; pg/ml) following 24-h co-culture (DC & PTEC) of flow cytometry sorted CD1c + DC with preconditioned normoxic (white bars) or hypoxic (grey bars) PTEC in the absence (Nil) or presence of poly I:C. PTEC alone and DC alone cultures are included as controls; N.D. not detectable. Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC experiments; n = 4. ****P < 0.0001, one-way ANOVA with Tukey's multiple-comparison test. D Flow cytometric detection of intracellular IL-1β protein following 4-h co-culture of magnetic bead-enriched CD1c + DC with pre-conditioned hypoxic PTEC in the presence of poly I:C. Co-cultures were treated prior to the 4-h culture period with either brefeldin A or VX-765 to enable intracellular IL-1β accumulation/detection. Representative flow cytometric histograms of IL-1β staining (black unfilled) compared with unstained control (grey filled) for PTEC (gated on live, single, CD45 − cells) and CD1c + DC (gated on live, single, CD45 + lineage − HLA-DR + CD1c + cells) are presented. One representative of three individual donor PTEC experiments is shown. E Fold changes (relative to -Fer-1) in secreted IL-1β levels (measured by ELISA; pg/ml) following 24-h co-culture (DC & PTEC) of magnetic bead-enriched CD1c + DC with pre-conditioned hypoxic PTEC and poly I:C in the absence (-Fer-1) or presence of ferrostatin-1 (+Fer-1). Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC experiments; n = 7. ns not significant, paired t test.
Collectively, these results provide the first comprehensive molecular and functional characterisation of inflammasome signalling under the hypoxic conditions of human CKD. We provide evidence of ferroptotic PTEC death, release of putative DAMP and triggering of CD1c + DC via the NLRP3 inflammasome within the hypoxic/fibrotic tubulointerstitium. The broader application of these findings will enable the development of novel therapeutic approaches with greater specificity-i.e., clinical testing of next-generation inflammasome inhibitors, engineered to avoid off-target effects, but not yet trialled in kidney diseases.  ) or DAMP inhibitors (apyrase, isotype control or HMGB1 antibody); n = 6. The colour bar represents the z-score. Yellow indicates low expression; purple indicates high expression. C, D Fold changes (relative to NIL inhibitor) in secreted IL-1β (C) and IL-18 (D) protein levels (measured by LEGENDplex™ assay) following 24-h co-culture (DC & PTEC) of magnetic bead-enriched CD1c + DC with pre-conditioned hypoxic PTEC and poly I:C in the absence (NIL) or presence of inflammasome inhibitors (VX-765, MCC950) or DAMP inhibitors (apyrase, isotype control or HMGB1 antibody). Bar graphs represent mean ± SEM. Symbols represent individual donor PTEC experiments; n = 6. *P < 0.05; **P < 0.01; ****P < 0.0001, one-way ANOVA with Tukey's multiple-comparison test.

MATERIALS AND METHODS Kidney tissue specimens
on the Banff 97 working classification of kidney pathology [75]. Tissue specimens were grouped into 'control/non-fibrotic' (n = 17; 4 female/13 male; mean age of 56.3 ± 11.5 years) or 'fibrotic' cortical tissue (n = 5; 2 female/3 male; mean age of 54.4 ± 9.7 years) based on this histopathological assessment of the non-tumour kidney parenchyma.

Isolation and culture of human primary PTEC
Human primary PTEC were purified from kidney cortical tissue following the method of Glynne and Evans [76] and then cultured in Defined Medium (DM) as previously described [77]. Human primary PTEC isolated from 'control/non-fibrotic' cortical tissue were used for in vitro experiments. All PTEC were used in experiments at passage 4.
For further characterisation of normoxic versus hypoxic PTEC, cell culture supernatant was collected for analysis of soluble factors (e.g., HMGB1), with PTEC harvested by trypsin treatment for downstream experiments. PTEC mitochondrial superoxide production, mitochondrial function and necrosis were assessed by flow cytometry, with cell acquisition performed on an LSR Fortessa (BD Biosciences, San Jose, CA, USA) and data analysed with FlowJo software (TreeStar, Ashland, OR, USA). Harvested PTEC were also examined for protein expression by western blotting (detailed information for these PTEC assays in Supplementary Methods).

4-HNE immunofluorescence staining of human primary PTEC
CD1c + DC were resuspended at 1.33 × 10 6 cells/ml in Complete Medium (CM) (components detailed in Supplementary Methods). CD1c + DC (100,000 cells in CM; 75 µl total volume) were added to the preconditioned PTEC (without removal of PTEC DM culture medium) and co-cultured for 24 h (unless otherwise specified) under normoxic or hypoxic conditions (10% foetal bovine serum (FBS) final concentration; 275 µL final volume). PTEC and DC-only wells were included as controls. Where indicated, 50 µg/ml polyinosinic:polycytidylic acid (Poly I:C; Sigma-Aldrich), a synthetic double-stranded RNA analogue, was added to cultures. For blocking studies, hypoxic PTEC or CD1c + DC were preincubated with inhibitors prior to co-culture (detailed information provided in Supplementary Methods).
Culture supernatants were harvested, and levels of soluble proteins were determined using the LEGENDplex™ Human Inflammation Panel 1 multiplex bead array (BioLegend, San Diego, CA, USA). LEGENDplex™ data were analysed using LEGENDplex TM QOGNIT (v2020.12.15, San Carlos, CA, USA), with a standard curve calculated using a five-parameter logistic (5PL) regression. For heatmap generation, LEGENDplex TM data were collated, and z-scores were calculated across cytokines in the dataset [78]. In selected experiments, levels of IL-1β in culture supernatants were assayed using the Human IL-1β ELISA MAX TM Deluxe Set (BioLegend), following the manufacturer's recommendations. Non-adherent cells (i.e., CD1c + DC) from normoxic and hypoxic co-cultures were collected for quantitative real-time polymerase chain reaction (qRT-PCR) analysis (detailed information provided in Supplementary Methods). The cellular source of IL-1β in PTEC-CD1c + DC hypoxic co-cultures was examined by intracellular staining and flow cytometric analysis (detailed information provided in Supplementary Methods). Sections were developed by applying the host-species appropriate horseradish peroxidase (HRP) polymer (Biocare Medical, Pacheco, CA, USA), followed by ImmPACT DAB peroxidase substrate (Vector Laboratories, Burlingame, CA, USA). Images were obtained using an Aperio AT Turbo (Leica Biosystems, Mt Waverley, VIC, Australia) bright field microscope. Quantitative-IHC analysis (positive pixel intensity/µm 2 area) was undertaken from four randomly selected areas for each tissue sample using ImageScope (v12.2.2.5015, Leica Biosystems) (detailed information provided in Supplementary Methods).

Statistical analysis
Sample sizes were selected based on previous publications from our laboratory with similar experimental design [80,81]. Data were normalised as a fold change relative to control conditions (i.e., normoxia) for each donor. All statistical tests were performed using GraphPad Prism (v9.0.1; GraphPad Software, La Jolla, CA, USA). Statistical comparisons between unpaired groups were performed using a Welch's t test. Comparisons between paired groups were performed using a Student's t test and multiple paired comparisons were performed using a one-way analysis of variance (ANOVA) with Tukey's multiple-comparison post hoc test. P values ≤0.05 were considered statistically significant.

DATA AVAILABILITY
The datasets generated and/or analysed during this study are available from the corresponding author on reasonable request.