Leader peptide or pro-segment mutants of renin are misrouted to mitochondria in autosomal dominant tubulointerstitial kidney disease

Renin (encoded by REN) is an aspartyl protease present in the blood and responsible for the cleavage of angiotensinogen to produce angiotensin I. It is thus involved in the first step of the renin-angiotensin system, which regulates blood pressure by controlling fluid and electrolyte balance and systemic vascular resistance (Sparks et al., 2014). Renin is synthesised as preprorenin, which is translocated into the endoplasmic reticulum (ER), where it is processed to prorenin, its inactive precursor, by cleavage of the leader peptide. Whereas prorenin is produced in many tissues, mature renin, produced by cleavage of the pro-segment, is exclusively produced in and secreted by renal juxtaglomerular cells (Paul et al., 2006).

Mutations in REN (1q32) have been associated with two different diseases. Homozygosity or compound heterozygosity for loss-of-function mutations, most commonly causing complete loss of renin synthesis, are associated with renal tubular dysgenesis (MIM#267430), a rare recessive condition characterised by perinatal mortality (Gribouval et al., 2005). In contrast, dominant mutations are associated with REN-related autosomal dominant tubulointerstitial kidney disease (ADTKD-REN) (MIM#613092) (Živná et al., 2009), leading to renal tubulointerstitial damage and progressive chronic kidney disease. Dominant REN mutations likely have a gain-of-function effect, as in heterozygous carriers of REN loss-of-function recessive mutations, the lack of one renin functional allele is not sufficient to induce ADTKD (Gubler and Antignac, 2010). Additional genes have been identified associated with ADTKD, such as uromodulin (UMOD), mucin-1 (MUC1), HNF1B and α1-subunit of translocon 61 (SEC61A1), with evidence supporting further genetic heterogeneity (Bleyer et al., 2017; Devuyst et al., 2019). ADTKD is one of the most common monogenic kidney diseases, with an estimated prevalence of ∼3% of patients with end-stage renal disease of genetic origin for ADTKD-UMOD alone (Groopman et al., 2019).

ADTKD-REN mutations reported to date are non-truncating, missense changes, in-frame deletion affecting residues within the protein leader peptide and missense changes located in the pro-segment or the mature part of the protein (Živná et al., 2020). Interestingly, some relevant differences were reported in the clinical picture of patients bearing mutations in the leader peptide and the pro-segment compared to that of patients with mutations in the mature part of renin. Mutations in the renin leader peptide or pro-segment lead to very early onset of clinical manifestations. Impaired kidney function and hypo-proliferative anaemia with low haemoglobin concentration can indeed be observed as early as 3 or 4 years of age. Anaemia resolves as the child enters adolescence, as long as renal function is not severely impaired. Hyperuricemia and gout usually develop in the second decade of life. Chronic kidney disease is slowly progressive and affected subjects reach end-stage kidney disease in the fifth or sixth decade of life. Mutations in the mature part of renin are associated with a milder renal phenotype, mainly characterised by late onset of kidney disease, hyperuricemia and gout. Indeed, the first signs of renal dysfunction appear in adulthood, at 50-60 years of age, and end-stage kidney disease is reached on average at the age of 70. Infantile hypo-proliferative anaemia was not reported in these patients (Živná et al., 2020).

Studies in cell models showed that all analysed ADTKD-REN mutations led to absent/decreased secretion, although their reported effect on intracellular maturation and trafficking was different. Extensive in vitro characterisation has been performed for the p.L381P isoform, a mutant of the mature protein, showing complete ER retention and induction of the unfolded protein response (Schaeffer et al., 2019). ER accumulation has also been confirmed for additional mutants in the mature part of renin (Živná et al., 2020). Mutant isoforms in the pro-segment showed partial accumulation in the ER-to-Golgi intermediate compartment (ERGIC) (Živná et al., 2020), whereas mutations in the leader peptide are associated with defective translocation in the ER and intracellular accumulation (Živná et al., 2009, 2020). Despite the apparently different cellular phenotype, mutations in the renin leader peptide and pro-segment are associated with similar disease features, raising the question as to whether they share a yet unravelled cellular phenotype.

In this study, we report a detailed in vitro characterisation of mutant renin isoforms in the leader peptide and the pro-segment. We show that these mutations lead to complete or partial mistargeting of renin to the mitochondria, which is accompanied by mitochondrial import defect and fragmentation of the mitochondrial network. This effect is due to the presence of the mutated leader peptide and/or of the pro-segment region; it does not depend on the mature renin protein or activity. These results represent an important additional step towards the current understanding of ADTKD-REN pathogenesis and suggest that different primary cellular responses could be the basis of the different ages of disease onset observed in patients.

Stemming on the hypothesis that mutations in the renin leader peptide could interfere with protein localisation, we performed in silico analysis using the software WoLF PSORT, which predicts the subcellular localisation of proteins based on their amino acid sequence. Although, as expected, wild-type (WT) renin is predicted to be secreted (extracellular), six out the eight described mutants of the leader peptide are predicted to be localised in the mitochondria (Fig. 1). The predicted localisation of mutants of the pro-segment is similar to that of the WT protein. Uromodulin and paraplegin (encoded by SPG7) are shown as examples of secreted and mitochondrial proteins, respectively, the localisation of which is correctly predicted by the software.

To verify the results of prediction analysis, we transiently expressed renin isoforms with mutations located in the leader peptide (p.W10R, p.L16del, p.L16R, p.W17R and p.C20R) and in the pro-segment (p.M39K and p.E348K) in Calu-6 cells, a human carcinoma-derived cell line expressing human renin endogenously at low levels (Lang et al., 1996). Western blot analysis showed that the mutants p.L16R, p.W17R and p.C20R were not or were barely secreted, whereas all other mutant isoforms were found in the extracellular medium (Fig. 2A). Moreover, in cell lysates of all mutants, we detected a low-molecular-mass additional band (Fig. 2A) that corresponds to a non-glycosylated form of renin, as shown by resistance to PNGase F (an enzyme removing N-linked oligosaccharides from glycoproteins) digestion (Fig. 2B). We hypothesised that this band could correspond to mutant renin that fails to interact with the ER translocon and is synthesised in the cytosol. Consistent with in silico prediction, analysis of renin subcellular localisation by immunofluorescence indicated mitochondrial localisation of all mutant isoforms, as shown by colocalisation with the mitochondrial marker TIM44 (encoded by TIMM44) (Fig. 2C). Whereas some mutants in the leader peptide were fully mistargeted to mitochondria (such as p.L16R, p.W17R and p.C20R), others were partially mislocalised (such as p.W10R and p.L16del). Mitochondrial localisation could also be detected for mutants in the pro-segment, even if reduced compared to that of mutants in the leader peptide. This is consistent with western blot results showing the non-glycosylated band only for mutants that were fully relocalised to mitochondria, whereas an additional signal, corresponding to the glycosylated protein, was seen for mutants that were partly rerouted to mitochondria and partly co-translationally inserted in the ER to enter the secretory pathway and acquire N-glycosylation (Fig. 2A). Similar results were obtained in HEK293 cells, a cell line of kidney origin that was previously used to study the effect of the REN mutation p.L381P mapping to the mature part of the protein (Schaeffer et al., 2019; Živná et al., 2009, 2020) (Fig. S1).

As immunofluorescence experiments showed co-staining between mitochondria and mutant renin isoforms, we wondered whether this reflected actual import of renin inside the organelle or its association on the outer membrane. To address this question, we performed a protease protection assay on mitochondria isolated from cells stably expressing representative mutants in the leader peptide (p.L16R and p.W17R). Digestion of the mitochondrial preparations with trypsin in different permeabilisation conditions showed that mutant renin isoforms were susceptible to protease digestion only in conditions permeabilizing the mitochondrial inner membrane. Mutant renin was in fact digested in the same conditions as TIM44, a subunit of the inner translocase facing the matrix; although it was still protected in conditions allowing digestion of CHCHD3, a protein of the intermembrane space (Fig. 2D). These results demonstrate that mutant renin isoforms are rerouted to mitochondria and imported in the mitochondrial matrix.

We then wondered which renin domains are necessary for mutant renin import in the mitochondria. We addressed this question by employing chimeric constructs consisting of green fluorescent protein (GFP) fused to the renin leader peptide alone or to the leader peptide and pro-segment with WT or mutated sequence. We focussed on two representative mutations in the leader peptide showing complete (p.L16R) or partial (p.L16del) mitochondrial relocalisation and on one mutation (p.E48K) in the pro-segment. Through immunofluorescence analysis of transiently transfected Calu-6 cells, we observed that the mutation p.L16R was sufficient to convert the renin leader peptide into a mitochondrial-targeting sequence (MTS). GFP fused to the p.L16R leader peptide was indeed fully targeted to mitochondria, as assessed by complete colocalisation with the mitochondrial marker TIM44. The L16del leader peptide alone did not have such an effect on GFP localisation (Fig. 3A), although it led to partial re-localisation of GFP to mitochondria when the WT pro-segment was added (Fig. 3B). Notably, the pro-segment by itself did not lead to mitochondrial localisation of a fusion construct with GFP (Fig. 3A). The pro-segment carrying the p.E48K mutation, when fused to the WT leader sequence and GFP, induced partial targeting of GFP to mitochondria (Fig. 3B). We conclude that some ADTKD-REN mutations in the renin leader peptide modify it to a functional MTS, whereas others require the presence of the WT pro-segment domain to create a signal for mitochondrial targeting. When the leader peptide is not affected, introduction of a mutation in the pro-segment leads to mitochondrial localisation. The mature part of the protein is not necessary for mitochondrial localisation. Similar results were obtained in HEK293 cells (Fig. S2) demonstrating that the cellular effect of renin mutations is not restricted to one cell type.

We then analysed whether mutant renin isoforms interfere with the efficiency of the machinery of mitochondrial protein import, owing to the presence of a non-canonical MTS. To do so, we co-expressed mutant renin isoforms with yellow fluorescent protein (YFP) fused to a canonical MTS to efficiently route it to mitochondria (Mito-YFP). We quantified the rate of mitochondrial import of Mito-YFP by western blot analysis. Indeed, the MTS was cleaved once YFP entered the mitochondria, leading to a shift of molecular mass that allowed us to distinguish the precursor from the mature, imported form. Cells treated with valinomycin were used as a positive control for defective mitochondrial protein import. In these cells, accumulation of the precursor protein still containing the MTS could be clearly appreciated (Fig. 4A). Co-expression of Mito-YFP with renin isoforms mutated in the leader peptide or pro-segment led to an increased amount of the precursor form of Mito-YFP compared to that seen in mock cells or cells expressing WT renin (Fig. 4A,B). This result strongly suggests that the presence of mutant renin interferes with the import of mitochondrially localised proteins. The import defect was not observed when co-expressing the p.L381P renin isoform mutated in the mature part and previously shown to be retained in the ER (Schaeffer et al., 2019). We also analysed mitochondrial import of YFP when co-expressed with renin targeted to the mitochondria via a canonical MTS (MTS-REN). Although MTS-REN was indeed localised in mitochondria (Fig. 4C), its co-expression did not lead to an import defect of Mito-YFP (Fig. 4B,D). These results demonstrate that the expression of the investigated renin mutants leads to a mitochondrial import defect and that such an effect is due to the pre-pro-sequence of renin.

We next analysed the effect of mutant renin mistargeting to mitochondria on mitochondrial network morphology, as mitochondrial fragmentation has been reported upon induction of various stress signals (Eisner et al., 2018). All isoforms mutated in the leader peptide or the pro-segment, fully or partially localizing to the mitochondria, led to mitochondrial fragmentation. As a control, the mutant isoform p.L381P did not show a similar effect. Importantly, this effect was not observed when expressing an exogenous protein, YFP, fused to an MTS to efficiently route it to mitochondria (Mito-YFP) (Fig. 5A).

We then assessed the effect of chimeric constructs with the renin leader or leader and pro-segment sequences fused to GFP (Fig. 3). Notably, all constructs previously shown to be targeted to mitochondria induced fragmentation, whereas the fusion of GFP with the L16del leader peptide alone, which is not targeted to mitochondria, had no effect (Fig. 5B). These results show that the presence of a mutated region of renin that is sufficient to drive mitochondrial localisation – i.e. the mutated leader peptide with, for some mutations, the addition of the pro-segment – induces fragmentation, whereas the mature part of renin is not required. Consistently, fragmentation was not induced by mature renin targeted to the mitochondria via a canonical MTS (MTS-REN), even when the pro-segment was added, although these isoforms were properly localised to mitochondria (Fig. 5C,D).

To assess whether defective ER translocation could be the basis of mitochondrial misrouting observed for mutants of the leader peptide or pro-segment, we assessed the fate of WT renin when inhibiting ER translocation. To do so, we treated cells expressing WT renin with ipomoeassin F (IpoF), a chemical compound shown to interact with SEC61α, the pore-forming subunit of the SEC61 translocon, thus inhibiting ER translocation (Zong et al., 2019). Immunofluorescence analysis showed that after IpoF treatment, WT renin was misrouted to mitochondria (Fig. 6A). The use of GFP fused to the WT leader peptide or leader peptide and pro-segment demonstrated that the presence of the pro-segment was necessary for mitochondrial localisation upon inhibition of ER translocation. Indeed, GFP fused to the renin leader peptide alone accumulated in the cytosol after IpoF treatment (Fig. 6A).

Interestingly, mitochondrial localisation of WT renin or of preprorenin GFP fusion constructs led to mitochondrial fragmentation. This is consistent with findings on mutant renin isoforms, confirming that the presence of the leader peptide and pro-segment are sufficient to induce such an effect (Fig. 6B). These results also demonstrate that the WT pre-pro-sequence of renin has the same effect of the mutated sequences in conditions forcing their interaction with mitochondria.

Finally, as ADTKD-REN patients are heterozygous for REN mutations and thus co-express both WT and mutant renin isoforms, we generated WT and mutant renin co-expressing cells to model the heterozygous condition. We used a dual-expression plasmid allowing co-expression of Flag-tagged WT renin and HA-tagged WT or mutant renin. Western blot analysis of lysates and conditioned medium showed the same signal pattern for the Flag-tagged WT protein when co-expressed with the HA-tagged WT isoform or different mutants (Fig. 7A). Consistently, immunofluorescence analysis showed that the localisation pattern of Flag-tagged WT renin is not different when co-expressed with HA-tagged WT or mutant isoforms (Fig. 7B). Hence, expression of mutant isoforms in the leader peptide or the pro-segment does not affect WT renin production and secretion, nor does it induce its accumulation as a non-glycosylated isoform.

In this study, we identified a novel cellular phenotype for renin mutations associated with ADTKD-REN that are localised in the leader peptide and in the pro-segment domains. Expression of renin isoforms with such mutations indeed leads to full or partial mistargeting of mutant renin to mitochondria, an event that is associated with mitochondrial import defect and fragmentation of the mitochondrial network. These effects are specific for these types of mutation, as they are not observed for an ADTKD-REN mutation that maps in the mature part of the protein and that is retained in the ER. The renin pre-pro-sequence, carrying mutations either in the leader peptide or in the pro-segment, is necessary and sufficient to induce mitochondrial localisation and a subsequent mitochondrial phenotype (fragmentation and defective import). Additionally, the mature part of renin is dispensable and has no effect when delivered to mitochondria through a canonical MTS. Rather, the effect of renin mutants is reproduced in cells expressing WT renin, or just its pre-pro-sequence fused to GFP, when reducing the efficiency of ER translocation by inhibiting SEC61A1 activity. Taken together, our findings (1) reveal a common cellular phenotype for mutations mapping to the leader peptide or pro-sequence of renin that share common disease features; (2) suggest that localisation of the renin pre-pro-peptide to mitochondria, whether it is induced by an ADTKD-related mutation or by interfering with ER entry, leads to a mitochondrial phenotype; and (3) provide new insight into the molecular pathogenesis of the disease.

To date, only few studies reported the characterisation of localisation of renin mutants in the leader peptide and the pro-segment. The authors showed that leader peptide mutants fail to enter the secretory pathway and accumulate intracellularly (Bleyer et al., 2010; Živná et al., 2009, 2020), whereas pro-segment mutants accumulate in the ERGIC (Živná et al., 2020). Our results clarify the fate of leader peptide mutants, showing that they are partially or fully directed to mitochondria, hence uncovering a new cellular phenotype. The full (mutant p.L16R, p.C20R and p.W17R) or partial (mutant p.L16del) mitochondrial localisation is consistent with previous studies showing that these mutants do not enter or partially enter the secretory pathway, respectively (Živná et al., 2009, 2020). Along the same line, the partial mitochondrial localisation of mutants in the pro-segment does not exclude a possible accumulation in the ERGIC, as previously reported. In fact, in our study, mutants in the pro-segment can also be detected in the ERGIC as they are partly entering the secretory pathway (Fig. S3). Our findings demonstrate a common cellular phenotype (i.e. mistargeting to mitochondria, mitochondrial fragmentation and import defect) for renin mutants in the leader peptide and pro-segment. This is opposed to the effect of mutations in the mature protein, leading to ER retention of mutant proteins. These different cellular phenotypes have a clinical correlation (Živná et al., 2020), as mutations in the leader peptide and pro-segment are associated with early-onset disease, whereas mutations in the mature part lead to a milder adult-onset form of the disease (Schaeffer et al., 2019; Živná et al., 2020).

The association of different clinical features induced by mutations localised in different domains of the protein is reminiscent of mutations in the INS gene encoding insulin. Processing of insulin is very similar to renin, as it is synthesised as pre-proinsulin and activated in secretory granules to mature insulin, followed by release upon stimulation (Vasiljevic et al., 2020). Both dominant and recessive mutations in INS can lead to diabetes. In analogy to what was observed for renin, recessive INS mutations leading to early-onset diabetes are loss-of-function mutations. Instead, dominant mutations leading to early-onset diabetes, called mutant INS-gene-induced diabetes of youth (MIDY), or late-onset diabetes act via a toxic gain-of-function mechanism. Although heterogeneity of the clinical presentation partly depends on genetic and environmental factors, it is also explained by different effects of INS dominant mutations. Whereas mutations causing MIDY phenotype are characterised by proinsulin misfolding in the ER and exert a dominant-negative effect on the trafficking of the WT protein, mutations associated with late-onset diabetes, including mutations in the leader peptide, act via a different mechanism, leading to cytosolic accumulation downstream of defective ER targeting and translocation (Liu et al., 2015).

Recently, a mutation in the leader peptide of another protein of the renin-angiotensin system, angiotensin-converting enzyme (ACE), has been reported (Fila et al., 2020). Interestingly, similarly to REN, recessive loss-of-function mutations in ACE are associated with renal tubular dysgenesis. This new mutation associates with slowly progressive kidney failure and anaemia, which is not explained by the level of kidney failure, highly reminiscent of the phenotype reported for REN mutations in the leader peptide. However, characterisation of the subcellular localisation of this mutant was not investigated. Several studies reported that mutations in the leader peptide of different proteins interfere with ER translocation, thus leading to decreased secretion, for example, mutations in collagen X in Schmid metaphyseal chondrodysplasia (Chan et al., 2001) or cationic trypsinogen or serine protease inhibitor Kazal type 1 in autosomal dominant hereditary pancreatitis (Hegyi and Sahin-Tóth, 2017). Mistargeting to mitochondria of mutated proteins has already been reported in kidney diseases, although these are due to other types of mutations. For example, a mutation adding an MTS to a native peroxisomal-targeting sequence has been reported in enoyl-CoA hydratase and 3-hydroxyacyl CoA dehydrogenase (EHHADH), leading to renal Fanconi syndrome. Mislocalisation of EHHADH interferes with mitochondrial function by disturbing β-oxidation of long-chain fatty acids (Assmann et al., 2016). A peculiar mechanism of mitochondrial mistargeting has also been reported in primary hyperoxaluria, in which mutations in alanine:glyoxylate aminotransferase (AGT) expose a cryptic MTS. These mutations indeed inhibit AGT dimerization, which, in normal conditions, masks the MTS (Fargue et al., 2013).

Our study demonstrates that mutant renin isoforms in the leader peptide or pro-segment are imported in the mitochondrial matrix. We show that fusion with the mutated leader peptide of renin targets GFP to mitochondria, suggesting that a single point mutation in the leader peptide is sufficient to convert the ER-targeting sequence into a mitochondrial one. As for some mutations in the leader peptide (i.e. p.L16del), addition of the pro-segment is needed, and considering that mutations in the pro-segment also drive mitochondria mistargeting, it is likely that structural features within the nascent chain, besides its primary sequence, are also involved. Recent studies on insulin indeed showed that ER translocation does not only require an efficient leader peptide, but it also depends on downstream sequences (Yang et al., 2021). Moreover, structural elements within the nascent chains have been demonstrated to regulate alternative targeting of secretory proteins to mitochondria for diverse proteins, for example, the prion-like protein shadoo (Pfeiffer et al., 2013). We thus hypothesise that mistargeting to mitochondria of mutant renin is driven by decreased efficiency in ER targeting, which is associated with the presence of a downstream domain compatible with mitochondrial import. Mutations in the pre-pro-sequence could lead to a decreased interaction with the signal recognition particle (SRP). This could be the basis of mitochondrial mistargeting as it has been previously reported that absence of SRP leads to aberrant co-translational targeting to mitochondria of proteins normally targeted to the ER (Costa et al., 2018). This is consistent with our data showing that interfering with ER translocation by blocking SEC61A1 with IpoF is sufficient to induce targeting of WT renin to mitochondria. IpoF does not have a general impact on translocation, but rather a selective effect on a subset of secretory proteins, possibly because they have reduced ability to compete with IpoF for SEC61 binding (Roboti et al., 2021). In line with these findings, renin was recently shown to be among the proteins for which ER translocation is severely affected when SEC61A1 activity is genetically reduced by ADTKD-associated mutations (Sicking et al., 2022). Collectively, these results suggest that defective ER translocation and mitochondrial mistargeting of renin could be a mechanism shared by ADTKD-SEC61A1 and ADTKD-REN. Moreover, these studies suggest that renin is among those secretory proteins that carry a weak leader peptide and hence are more likely to be affected in conditions affecting ER translocation.

Our study highlights the importance of the renin pre-pro-sequence to induce a mitochondrial phenotype (Fig. 8). Mitochondrial import defect and fragmentation were indeed observed upon expression of GFP fused to the mutated leader peptide with, in some cases, addition of the pro-segment. Mitochondria fragmentation was also observed when forcing the interaction with mitochondria of WT renin or GFP fused to the WT renin pre-pro-sequence by inhibiting ER translocation (IpoF treatment). On the contrary, mitochondrial phenotypes were not observed by targeting mature renin to the mitochondrial matrix using a canonical MTS. This clearly demonstrates that the toxic effect is exerted by renin pre-pro-sequence interaction with mitochondria, whereas the mature protein and its activity are not required. Further studies are needed to understand the molecular bases of this toxic effect that could, for instance, be mediated by the interaction of renin mutants with components of the mitochondrial import machinery in the outer and/or inner membrane and the consequences of defective import on mitochondrial function. A similar mechanism has been described in Parkinson's disease for α-synuclein, which interacts with the import machinery to lead to defective import (Di Maio et al., 2016). This, in turn, leads to mitochondrial complex-I dysfunction, which is rescued by overexpression of translocases of the outer and inner membrane (Franco-Iborra et al., 2018).

Given the functional studies performed so far on mutations in the different ADTKD genes (i.e. UMOD, MUC1, SEC61A1 and REN), ER stress and unfolded protein response have been proposed to play a central role in disease onset and progression (Devuyst et al., 2019; Mabillard et al., 2023; Schaeffer and Olinger, 2020). Our work, demonstrating mistargeting to mitochondria of renin mutants in the leader peptide and pro-segment, expands current knowledge of ADTKD pathogenesis. Of note, mitochondrial dysfunction, secondary to ER stress, has been reported in ADTKD-UMOD (Kemter et al., 2017). Additional studies are needed to characterise cell responses that are induced by mutants mistargeted to mitochondria and how this is eventually linked with the induction of inflammation and fibrosis in the kidney.

Human cDNA of a untagged renin clone (SC321832) was purchased from Origene (Rockville, MD, USA). Renin cDNA was subcloned in pcDNA3.1 (Thermo Fisher Scientific, Waltham, MA, USA) and HA was inserted at the C-terminal (Schaeffer et al., 2019). The indicated mutations were introduced in the pcDNA-hRenHA construct using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA, USA). Primers were designed using the software QuikChange Primer Design program.

Fusion constructs between GFP and the leader or leader and pro-segment domains were obtained by PCR as follows. Two PCRs were performed to produce the sequences of GFP with the leader peptide, pro-segment or leader peptide and pro-segment. The two PCRs were designed in order to obtain two products with complementary ends, which could be used as template in a third PCR to obtain the final construct with the 5′ and 3′ ends at BamHI and XbaI sites, respectively, for cloning in the pcDNA3.1 vector.

MTS-YFP was expressed from the pYFP-Mito vector (Clontech, Mountain View, CA, USA; 6115-1). The mature part of renin (renin) or the mature part of renin fused to the pro-segment (prorenin) was amplified by PCR from the pcDNA-hRenHA vector and cloned under the MTS from subunit VIII of human cytochrome c oxidase in the pYFP-Mito vector in place of the YFP using the BamHI and NotI cloning sites.

All constructs were verified by sequencing before use and are available upon request. The primers used in this study are provided in Tables S1-S3.

HEK293 cells [American Type Culture Collection (ATCC); CRL-1573] were grown in Dulbecco's modified Eagle's medium (DMEM; Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS; Euroclone, Pero, Italy), 200 U/ml penicillin, 200 μg/ml streptomycin and 2 mM glutamine at 37°C, 5% CO₂. Calu-6 cells (ATCC; HTB-56) were grown in Eagle's minimum essential medium (LGC Standards, London, UK; 30-2003) supplemented with 10% FBS, 200 U/ml penicillin and 200 μg/ml streptomycin at 37°C, 5% CO₂. Cells were transiently transfected using Lipofectamine 2000 (Thermo Fisher Scientific) following the manufacturer's protocol. All cell lines were routinely tested by PCR for the absence of mycoplasma contamination. When indicated cells were treated with 100 nM valinomycin (Cayman Chemical, Ann Arbor, MI, USA; 10009152) for 20 h.

Cells were incubated for 14 h in OptiMEM (Thermo Fisher Scientific) 36 h after transfection and lysed in Triton lysis buffer [50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% Triton X-100, 10 mM NaF, 0.5 mM sodium orthovanadate, 1 mM glycerophosphate and protease inhibitor cocktail (Sigma-Aldrich)] for 1 h at 4°C under rotation, followed by centrifugation for 10 min at 17,000 g. Soluble fractions were quantified by the Bio-Rad Protein Assay (Bio-Rad, Hercules, CA, USA). Conditioned medium was precipitated with four volumes of acetone and resuspended in PBS. When indicated, lysates were deglycosylated with PNGase F (New England Biolabs, Ipswich, MA, USA) according to the manufacturer's instructions. Then, 20 µg of each protein lysate and the equivalent of 200 µl of medium were loaded onto reducing SDS-polyacrylamide (10%) gels. Transblotted nitrocellulose membranes (Cytiva, Marlborough, MA, USA) were incubated with the indicated antibodies, followed by incubation with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibody. Protein bands were visualised with the Immobilon Western Chemiluminescent Horseradish Peroxidase Substrate kit (Millipore, Billerica, MA, USA).

The following primary antibodies were used: mouse purified anti-HA.11 epitope tag antibody (1:1000; BioLegend, San Diego, CA, USA; 901502), mouse monoclonal anti-GAPDH antibody (1:1000; Santa Cruz Biotechnology, Dallas, TX, USA; sc-32233), rabbit polyclonal anti-GFP antibody (1:5000; Thermo Fisher Scientific; A11122) and rabbit polyclonal anti-CHCHD3 antibody (1:1000; Novus Biologicals, Littleton, CO, USA; NBP1-83656). The following secondary antibodies were used: Amersham ECL rabbit IgG, HRP-linked whole antibody (from donkey) (1:7500; Cytiva; NA934-1ML) and Amersham ECL mouse IgG, HRP-linked whole antibody (from sheep) (1:7500; Cytiva; NA931-1ML).

Cells grown on coverslips were fixed in 4% paraformaldehyde for 15 min, permeabilised for 10 min with 0.5% Triton X-100 and blocked for 30 min with 10% donkey serum. Cells were labelled for 1 h 30 min at room temperature with the indicated antibodies, followed by a 1 h incubation with the appropriate Alexa Fluor-conjugated secondary antibodies. Cells were stained with 4,6-diamidino-2-phenylindole (DAPI) and mounted using fluorescence mounting medium (DAKO, Agilent). All pictures were taken with an UltraVIEW ERS spinning disk confocal microscope (UltraVIEW ERS-Imaging Suite Software, Zeiss 63×/1.4; PerkinElmer Life and Analytical Sciences, Boston, MA, USA) All images were imported in Photoshop CS (Adobe Systems, Mountain View, CA, USA) and adjusted for brightness and contrast. Pearson's correlation coefficient between the signal of renin and that of TIM44 was calculated using the Fiji plugin Coloc 2 (Schindelin et al., 2012).

The following primary antibodies were used: mouse purified anti-HA.11 Epitope Tag antibody (1:500; BioLegend; 901502), rabbit recombinant anti-TIM44 antibody (1:500; Abcam, Cambridge, UK; ab194829) and rabbit affinity isolated anti-ERGIC-53/p58 antibody (1:500; Sigma-Aldrich; E1031). The following secondary antibodies were used: donkey anti-mouse IgG (H+L) highly cross-adsorbed secondary antibody, Alexa Fluor™ 488 (1:500; Thermo Fisher Scientific; A21202), donkey anti-mouse IgG (H+L) highly cross-adsorbed secondary antibody, Alexa Fluor™ 594 (1:500; Thermo Fisher Scientific; A21203), donkey anti-rabbit IgG (H+L) highly cross-adsorbed secondary antibody, Alexa Fluor™ 488 (1:500; Thermo Fisher Scientific; A21206) and donkey anti-rabbit IgG (H+L) highly cross-adsorbed secondary antibody, Alexa Fluor™ 594 (1:500; Thermo Fisher Scientific; A21207)

Mitochondria were prepared as follows: ∼15×10⁶ cells were resuspended in 1 ml isolation buffer (225 mM mannitol, 75 mM sucrose, 1 mM EGTA and 5 mM HEPES) containing protease inhibitor cocktail (Sigma-Aldrich), diluted 1/1000 and lysed using a syringe. The lysate was centrifuged for 10 min at 600 g at 4°C. The supernatant was then centrifuged for 10 min at 7000 g at 4°C. The pellet was washed in 500 µl isolation buffer and centrifuged for 10 min at 7000 g at 4°C. The pellet containing the mitochondria was finally resuspended in 50 µl isolation buffer. Isolated mitochondria (50 µg) were permeabilised with the indicated amount of digitonin (BioVision, Waltham, MA, USA) to permeabilise only the outer or both the inner and outer mitochondrial membranes, and, where indicated, digested with 50 µg/ml trypsin for 30 min on ice. The reaction was stopped by addition of 10 mM phenylmethylsulfonyl fluoride.

Calu-6 cells were plated on coverslips in 35 mm wells (500,000 cells/well) and transfected with 300 ng pYFP-Mito to express MTS-YFP and 1.2 µg of the plasmid expressing the indicated construct using 7 µl Lipofectamine 2000 (Thermo Fisher Scientific). After 24 h of transfection, the YFP signal was analysed by live imaging. Pictures were acquired with z-stacks of 0.2 µm using a DeltaVision microscope with a 100× objective and deconvolved. For each transfection, 20 fields were analysed, in which the number of cells with fused, intermediate or fragmented mitochondrial networks were counted.

For experiments performed in conditions of ER translocation inhibition, cells were treated for 14 h with 100 nM IpoF. IpoF treatment was started 30 h after transfection. IpoF was a generous gift from Prof. Wei Shi (Ball State University, IN, USA).

We thank the San Raffaele Alembic facility for excellent technical assistance for microscopy experiments; Prof. Wei Shi (Ball State University, IN, USA) for kindly providing IpoF; Prof. Stephen High (University of Manchester, UK) for helpful suggestions; and Prof. Marisa Santostefano (Policlinico S. Orsola-Malpighi Hospital, Bologna, Italy), Prof. Bertrand Knebelmann (Necker Hospital, APHP, Paris, France) and Prof. Corinne Antignac (Necker Hospital, APHP, Paris, France) for sharing data on identification of new ADTKD-REN mutations.