The C-terminal tail of polycystin-1 suppresses cystic disease in a mitochondrial enzyme-dependent fashion

Autosomal dominant polycystic kidney disease (ADPKD) is the most prevalent potentially lethal monogenic disorder. Mutations in the PKD1 gene, which encodes polycystin-1 (PC1), account for approximately 78% of cases. PC1 is a large 462-kDa protein that undergoes cleavage in its N and C-terminal domains. C-terminal cleavage produces fragments that translocate to mitochondria. We show that transgenic expression of a protein corresponding to the final 200 amino acid (aa) residues of PC1 in two Pkd1-KO orthologous murine models of ADPKD suppresses cystic phenotype and preserves renal function. This suppression depends upon an interaction between the C-terminal tail of PC1 and the mitochondrial enzyme Nicotinamide Nucleotide Transhydrogenase (NNT). This interaction modulates tubular/cyst cell proliferation, the metabolic profile, mitochondrial function, and the redox state. Together, these results suggest that a short fragment of PC1 is sufficient to suppress cystic phenotype and open the door to the exploration of gene therapy strategies for ADPKD.

It is characterized by the progressive development of fluid-filled cysts whose expansion compromise renal function and can lead to end-stage renal disease.
Mutations in the PKD1 gene, which encodes polycystin-1 (PC1), are responsible for ~78% of cases (Cornec-Le Gall et al., 2019). Although PKD1 was identified over 25 years ago (Consortium, 1994), the downstream pathways leading to cystogenesis have not been fully elucidated, and current therapeutic options remain scarce and only modestly effective.
Over the past decade, metabolic abnormalities have emerged as a hallmark of ADPKD (Padovano et al., 2018). The first evidence for metabolic alterations in ADPKD was the observation of increased glycolysis and lactate production in cells from a Pkd1 knockout (KO) mouse model (Rowe et al., 2013). Limiting glucose availability decreased their proliferation, and treatment with 2-deoxyglucose (2DG) to inhibit glycolysis led to partial amelioration of the cystic phenotype in Pkd1-KO mice (Rowe et al., 2013, Chiaravalli et al., 2016. The dependence on glycolysis and the increased lactate levels are together suggestive of an aerobic glycolysis phenotype, similar to the Warburg effect observed in cancer cells (Priolo and Henske, 2013, Rowe et al., 2013, Chiaravalli et al., 2016. Subsequently, other related significant metabolic alterations have been observed in ADPKD cellular and animal models, such as defective fatty acid oxidation and decreased rates of oxidative phosphorylation (Padovano et al., 2017, Hajarnis et al., 2017, Menezes et al., 2016. Several promising therapeutic approaches that target these altered metabolic pathways have been reliably assess its subcellular localization through immunofluorescence microscopy. To assess whether the CTT distribution was consistent with its interaction with NNT, we transiently transfected WT HEK293 cells with the 2HA-PC1-CTT construct and found that it colocalized with endogenous NNT at the mitochondria ( Figure 2D). Of note, CTT expression was also observed in nuclei of transfected cells, consistent with previous findings (Chauvet et al., 2004).
Interestingly, the widely used C57BL/6J ("J") mouse strain carries a deletion of exons 7-11 in the Nnt gene, a mutation that completely abrogates its expression ( Figure 2E).
This variant was acquired prior to 1984, however was only identified in 2005, when it was associated with glucose intolerance in "J" mice (Toye et al., 2005). We moved the alleles required to produce the Pkd1-KO+CTT mice to this NNT-deficient background (J-Pkd1-KO+CTT). To determine whether NNT interacts with 2HA-PC1-CTT in vivo, we performed anti-HA pulldowns from N-Pkd1-KO+CTT, J-Pkd1-KO+CTT and N-Pkd1-KO total kidney lysates. We found that NNT is only detected in immunoprecipitates derived from animals that express the 2HA-tagged PC1-CTT and not in those derived from N-Pkd1-KO mice that do not express CTT ( Figure 2F). As expected, anti-HA immunoprecipitates from Pkd1-KO+CTT kidneys on the "J" background did not contain a 114-kDa NNT band that could be detected on immunoblotting with anti-NNT antibody. Taken together, these data demonstrate that the PC1-CTT interacts with the inner mitochondrial membrane protein NNT in mouse kidney epithelial cells in vivo.

The PC1-CTT/NNT interaction modulates disease progression
under the control of the Pax8 rtTA promoter, leading to excision of the floxed exon 2-4 region and consequent inactivation of Pkd1 in Pax8 rtTA -exppressing cells (Shibazaki et al., 2008, Ma et al., 2013. We determined the status of homozygosity or heterozygosity for both the Pax8 rtTA and TetO-Cre alleles by quantitative PCR (qPCR) for each animal included in the present cohort and found that Pax8 rtTA and TetO-Cre copy numbers were randomly distributed across the 4 groups (Pkd1-KO+CTT and Pkd1-KO in both "N" and "J" backgrounds), and did not correlate to any degree with disease severity, as determined by KW/BW ratio ( Figure S2). We also evaluated the efficiency of Pkd1 knock-out in all animals to determine whether there was any significant genotype-dependent effect on this parameter. We determined levels of nonrearranged WT Pkd1 by extracting genomic DNA from kidney tissue from each mouse contained in the cohort followed by quantitative genomic PCR using a reverse primer specific for Pkd1 exon 4 and a forward primer specific to its preceding intron. We calculated a relative ratio of non-rearranged Pkd1 by comparing the observed levels of non-rearranged Pkd1 product to the levels obtained from WT control kidneys and found that rearrangement levels are equivalent across all 4 mouse groups ( Figure S3).
Finally, the NNT mutation appears to be the major allelic difference between C57BL/6J and C57BL/6N strains (Simon et al., 2013) and is the only candidate genetic variation that has been directly associated with the metabolic (Toye et al., 2005, Ronchi et al., 2013, Fergusson et al., 2014, cardiologic (Murphy, 2015) and renal (Usami et al., 2018) differences observed between them. It is important to note, however, that the "N" and the "J" strains manifest at least one additional phenotypically significant genetic polymorphism. This is the case for the Crb1 gene, in which the rd8 retinal degeneration mutant is detected exclusively in the "N" mice and results in a recessive ocular phenotype. To our knowledge, this is the only other mutation that differs criteria of both P value <0.05 and fold change >2, of which 6 were upregulated and 38 were downregulated in N-Pkd1-KO+CTT compared to N-Pkd1-KO kidneys ( Figure 4B and Table S2). In contrast, analysis of both J-Pkd1-KO+CTT and J-Pkd1-KO only revealed a significant change (P value <0.05 and fold change >2) in 1 metabolite.
Interestingly, many of the metabolites whose levels are reduced in N-Pkd1-KO+CTT mice have been previously implicated in ADPKD pathogenesis and some of them are related to potential therapeutic targets ( Figure 4B). This is the case for the 4-fold decrease in methionine levels induced by CTT expression in the "N" mice. Methionine has been reported to promote cyst growth through increased Mettl3 expression and to constitute a potential therapeutic vulnerability in murine models (Ramalingam et al., 2021). Similarly, we report that CTT expression in N mice is associated with a 4-fold decrease in lactate levels, consistent with a reduction in the dependence on glycolysis that has been detected in ADPKD cell and mouse models (Rowe et al., 2013, Chiaravalli et al., 2016. Additionally, a recent study has found that Pkd1 -/mouse embryonic fibroblasts are dependent upon glutamine to fuel the TCA cycle and that these cells exhibit increased asparagine synthetase activity . This enzyme couples the conversion of aspartate-to-asparagine to the conversion of glutamine-to-glutamate. These findings are consistent with the results of plasma metabolomics in children and young adults with ADPKD relative to healthy controls, which revealed a significant increase in asparagine levels (Baliga et al., 2021). Interestingly, we detected a 5-fold decrease in asparagine levels and a 2.5-fold decrease in glutamate levels in the N-Pkd1-KO+CTT mice as compared to those detected in N-Pkd1-KO mice, suggesting a rescue of this feature of pathologic metabolic reprogramming. Finally, we observed significant reductions in the levels of known uremic toxins such as allantoin and 5-hydroxyindoleacetate in N-Pkd1-KO+CTT mice, as well as a significant reduction in the levels of the urea cycle metabolites ornithine, citrulline and arginine. Of note, the previously mentioned clinical metabolomic study (Baliga et al., 2021) suggests drastic changes in urea cycle activity in ADPKD and reveals a significant positive association between ornithine levels and height-adjusted total kidney volume (HtTKV) at baseline, as well as association between ornithine levels and change in HtTKV over a 3-year observation period. In conclusion, the metabolic signature associated with CTT expression in N-Pkd1-KO mice is marked by the reversal of dysregulated metabolites that are associated with ADPKD.

N-Pkd1-KO+CTT mice exhibit increased NNT expression and increased assembly of ATP synthase and mitochondrial complex IV at a "per mitochondrion" level
We next sought to determine whether CTT and CTT/NNT interactions alter the inventories of proteins that potentially affect mitochondrial function. To this end, we performed immunoblotting analyses on kidney lysates from N-Pkd1-KO+CTT and N-Pkd1-KO mice ( Figure 5A). We detected a 4-fold increase in NNT expression (normalized to actin) in CTT-expressing mice compared to N-Pkd1-KO littermates that did not inherit the transgene. Interestingly, this significant difference was shown to be a product of both increased mitochondrial mass in CTT expressing mice (as revealed by the significant 1.8-fold increase in TOMM20/actin ratio), as well as of increased NNT expression at a "per mitochondrion" level (as revealed by the significant 2.1-fold increase in NNT/TOMM20 ratio in these same mice) ( Figure 5B). Comparable findings were identified in tissue from cystic human kidneys, which exhibit a significant decrease in NNT protein expression as compared to healthy kidney tissue ( Figure 5D). Furthermore, western blotting employing a "mitococktail" antibody that interrogates the levels of stably assembled mitochondrial membrane complexes demonstrated increased levels of assembled ATP-synthase (complex V or CV) and cytochrome c oxidase (complex IV or CIV) at a "per mitochondrion" level, as revealed by a significant 2.4-fold increase in CV/TOMM20 and a 1.9-fold increase in CIV/TOMM20 in N-Pkd1-KO+CTT expressing mice ( Figures 5A and 5B). Interestingly, we did not observe any differences in the assembly of complexes I (CI), II (CII) or III (CIII) in these same mice.
A similar evaluation on the "J" background reveals no significant difference in mitochondrial mass (TOMM20/actin) or in mitochondrial complex assembly (Figures 5A and 5C), suggesting that the effects observed in the "N" background are specific to the CTT expression and the potential for its interaction with NNT.

Pkd1-KO mice
We next interrogated whether NNT alone, in the absence of CTT, serves as a significant gene modifier in the context of ADPKD. To that end, we compared data previously presented (Figures 1C-E and 3 A-C) derived from N-Pkd1-KO mice and J-Pkd1-KO to each other, in the absence of CTT expression. Surprisingly, NNTcompetent cystic mice exhibited non-significant trends towards increased KW/BW ratio ( Figure 6A) and BUN levels ( Figure 6B), as well as a significant 2.16-fold increase in serum creatinine levels ( Figure 6C) when compared to J-Pkd1-KO. Considering both mouse cohorts were analyzed at 16-weeks, a time window characterized by marked and advanced cystogenesis (Ma et al., 2013), it is not surprising that only the least sensitive parameter would show a statistically significant difference at this late stage of disease progression. It is well established that declining glomerular filtration rate (GFR) is a late consequence of cyst progression in ADPKD (Grantham et al., 2011) consistent with the observed difference in serum creatinine levels but not in other parameters that have already reached their maximal cystic disease-induced elevations. Furthermore, we performed immunohistochemistry (IHC) on mouse kidney tissue obtained from these Pkd1-KO mice on both backgrounds and from N-Pkd1-KO+CTT mice ( Figure 6D). As expected, the "J" cystic mice did not express NNT. The cystic "N" mice, both in the presence and absence of CTT, expressed NNT predominantly in distal segments of the nephron and to a lesser degree in proximal tubules. No NNT was detected in glomeruli or Bowman's capsule. A similar pattern of NNT renal distribution is reported in the Human Protein Atlas (proteinatlas.org) (Uhlen et al., 2015). It is important to note that CTT-expressing mice exhibit dramatic preservation of renal parenchyma architecture as compared to "J" and "N" mice lacking CTT expression. Finally, cystic mice on the "N" background were frequently found to exhibit mild to moderate hydronephrosis, characterized by an expanded pelvis ( Figure   6E) with no obvious downstream obstruction and negative staining for calcium oxalate deposition (data not shown). Thus, a cause for this specific morphologic feature in "N" mice remains to be determined.

PC1-CTT expression rescues NNT activity
We assessed whether and how the CTT might alter NNT activity. We initially adopted a targeted LC-MS method to directly quantify NAD(P)(H) levels in whole kidney homogenates from 16-week Pkd1-KO+CTT and Pkd1-KO mice on both the "N" and "J" backgrounds. N-Pkd1-KO+CTT exhibited an increase in NADPH/NADP and NADH/NAD + ratios when compared to N-Pkd1-KO only mice ( Figure 7B and 7C), suggesting CTT-induced effects on the modulation of these cofactors. As expected, expression of CTT in the J-Pkd1-KO mice did not affect either ratio. While NAD(P)(H) measurements provide indirect insights into the level of NNT function, these values alone cannot be employed to directly estimate NNT enzymatic activity since a large number of processes contribute to determining NAD(P)(H) levels. Previous studies in non-renal tissue have suggested that 50-70% of the mitochondrial NADPH pool is maintained by isocitrate dehydrogenase (IDH2) (Rydstrom, 2006, Nickel et al., 2015. In light of this potential complexity, we opted to directly evaluate the NNT activity in mitochondrial fractions from N-Pkd1-KO+CTT and N-Pkd1-KO kidneys. To ensure that the assessment of NNT enzymatic activity was not influenced by the cystic phenotype we conducted this experiment in pre-cystic, 10-week-old mice (Ma et al., 2013)( Figure 7D).
Immunoblotting of a mitochondrial fraction prepared from 70 mg of fresh kidney tissue revealed no significant differences in NNT expression among N-Pkd1-KO+CTT, N-Pkd1-KO and N-WT mice at 10 weeks of age (Figures 7G and 7H), in marked contrast to differences observed in 16-week-old animals ( Figures 5A and 5B). The absence of a significant cystic phenotype in all mice at 10 weeks was consistent with the lack of differences in BUN and serum creatinine ( Figures 7E and 7F). The fresh mitochondrial extract was employed in concert with a standard kinetic spectrophotometric assay of NNT enzymatic activity that detects NNT-mediated reduction of the NAD analog APAD (Shimomura et al, 2009). NNT is embedded in the inner mitochondrial membrane and its catalytic domains face the mitochondrial matrix. Thus, to make hydrophilic substrates accessible to the enzyme for the purposes of the assay, the mitochondrial membrane is permeabilized through the addition of detergents Brij35 and lysolecithin.
Moreover, a relatively high assay pH is used to minimize nonspecific contributions of NADH-linked dehydrogenases and reductases (Shimomura et al, 2009). We confirmed the specificity of the assay by comparing the activities detected in material prepared from WT "N" to "J" kidneys. While a consistent linear positive slope was observed in assay tracings of material from "N" mice, as expected assay tracings of material from WT "J" mice exhibited no upward slope ( Figure 7I). We find that N-Pkd1-KO mice exhibit a 20% decrease in NNT enzymatic activity as compared to that detected in healthy "N" WT controls (Figures 7I and 7J). Furthermore, CTT expression in N-Pkd1-KO mice rescues NNT enzymatic activity to the same level observed in the healthy "N" controls ( Figures 7I and 7J). Of note, this assay was performed on mitochondria extracted from whole-kidney tissue, which includes multiple cell types in addition to Cre-expressing tubular cells. Hence, the magnitude of the observed difference is likely an underestimation of the true effect that is manifest in Creexpressing cells.

DISCUSSION
We report the unexpected finding that expressing the C-terminal 200 aa (CTT) of PC1 in an orthologous murine model of ADPKD is sufficient to suppress the development of the cystic phenotype. CTT expression resulted in dramatic preservation of renal function and morphology, as evidenced by BUN and serum creatinine levels that are comparable to levels detected in healthy control mice. Mass spectrometric analysis performed on immunoprecipitates from a crude mitochondrial fraction revealed that NNT is the most significant PC1 binding partner in this setting. We confirmed that NNT interacts with CTT in vivo and showed that the CTT and NNT proteins colocalize in mitochondria when CTT is expressed in HEK293 cells. A previous study that employed a split GFP assay to assess PC1-CTT submitochondrial localization showed that PC1-CTT localizes specifically to the mitochondrial matrix or matrix-facing surface of the mitochondrial inner membrane (Lin et al., 2018), a finding consistent with the predicted topological requirements of the novel interaction that we have identified between CTT and the mitochondrial matrix-facing inner membrane enzyme NNT. We showed that the suppression of the cystic phenotype produced by CTT expression is dependent upon this interaction, since no significant rescue is observed when Pkd1-KO+CTT mice are compared to Pkd1-KO mice in the NNT-deficient "J" background. Similarly, tubular and cystic index and tubular proliferation parameters were only rescued in CTT-expressing Pkd1-KO mice generated in the "N" background. Bulk kidney tissue metabolite profiling analysis revealed differentially clustering metabolites only on the "N" background when comparing Pkd1-KO+CTT to Pkd1-KO mice. We showed that N-Pkd1-KO+CTT mice exhibit significant and concomitant downregulation of multiple ADPKD-associated metabolites that have been identified over the past decade and that have led to the suggestion of several metabolism-related potential therapeutic interventions for ADPKD. We assessed the effects of CTT expression on the stable assembly of electron transport chain (ETC) complexes. Previous studies have shown that increased quantities of stably assembled ETC complexes correlate with increased ETC activity and increased levels of oxidative phosphorylation (Sieber et al., 2016, Sing et al., 2014. We show that CTT expression in N-Pkd1-KO mice not only leads to increased mitochondrial mass, but also to increased stable assembly of CIV and ATPsynthase after normalization to mitochondrial content. In concert with the metabolomic analysis, which revealed that expression of CTT on the "N" background leads to an approximately 4-fold reduction in lactate levels, these data support the interpretation that the rescue model exhibits a profound shift back towards oxidative phosphorylation serving as the predominant source of ATP generation. Taken together, our data suggest that re-expression of the PC1 C-terminal tail exerts effects that act upstream of the previously identified ADPKD-relevant metabolic pathways. Interestingly, the presence or absence of NNT expression alone does not dramatically alter the outcome in Pkd1-KO mice. In the absence of CTT expression, both "N" and "J" Pkd1-KO mice develop severe cystic disease, and our data strongly suggest that the NNT-competent strain exhibits even more aggressive disease progression. While the morphological and physiological consequences of cystic disease are severe in both 16-week-old "J" and "N" Pkd1-KO mice, we detect a roughly 2-fold increase in serum creatinine levels in "N" vs "J" Pkd1-KO mice. This is especially interesting since declining GFR constitutes a fairly late consequence of cystic disease progression (Grantham et al., 2011). In line with our findings, a recent study characterized the disease progression in a mouse model homozygous for a Pkd1 hypomorphic variant on three different strain backgrounds: BalbC/cJ (BC), 129S6/SvEvTac (129) and C57BL/6J (Arroyo et al., 2021). While this study does not identify a specific modifier or mechanistic link, it presents robust evidence suggesting that disease progression in C57BL/6J was less severe than in the BC or 129 mice, which are expected to be NNT-expressing. Furthermore, this same study reveals the presence of mild hydronephrosis in both BC and 129 mice but not in C57BL/6J (Arroyo et al., 2021), which is consistent with the frequent presence of hydronephrosis that we observe in N-Pkd1-KO mice.
We hypothesize that the more aggressive phenotype in N-Pkd1-KO mice may be attributable to NNT's key role as an antioxidative enzyme due to its capacity to regenerate NADPH from NADP utilizing NADH as the electron donor, driven by the protonmotive force across the mitochondrial inner membrane. By facilitating the detoxification of the reactive oxygen species (ROS) that have been shown to accumulate in hyperproliferative ADPKD tissue (Lu et al., 2020), NNT may allow cystic tissue to overcome oxidative stress and further proliferate. This hypothesis is supported by the trend towards higher proliferation revealed by ki67 staining in "N" vs "J" Pkd1-KO mice. In fact, the possible contributions of NNT to the severity of conditions involving hyperproliferation have been recognized in the context of cancer.
NNT knockdown, which leads to decreased proliferation and increased apoptosis, is being explored as a therapeutic approach in models of various neoplasms, including gastric cancer (Li et al., 2018), adrenocortical carcinoma (Chortis et al., 2018) and hepatic adenocarcinoma (Ho et al., 2017).
The unexpected finding that NNT expression alone does not protect against cyst formation but that expression of CTT stimulates NNT activity and suppresses cyst formation in an NNT-dependent manner led us to further investigate the mechanism responsible for the NNT-dependent CTT rescue. Our initial efforts involved evaluating potential changes in redox environment secondary to CTT expression. Significant changes in bulk tissue NADH/NAD + and NADPH/NADP ratios were detected between cystic mice expressing or not CTT exclusively in the "N" but not in the "J" background, suggesting that CTT is capable of altering NNT-dependent oxidoreductase pathways.
The relative increase of NADPH in the N-Pkd1-KO+CTT rescue model suggests the possibility that CTT expression in N-Pkd1-KO mice reduces oxidative stress through enhanced NADPH-dependent reduction of oxidized glutathione (GSSG->GSH) and thioredoxin. The impact of oxidative stress on cyst progression in ADPKD models is complex. A previous study suggested that Pkd1 mutant cells exhibit higher concentrations of oxidized glutathione, which is consistent with increased levels of detrimental oxidative stress (Lin et al., 2018). A different study suggests that exogenous GSH administration may actually enhance cystogenesis by decreasing the vulnerability of mutant cells to ROS and oxidative stress (Flowers et al., 2018). Of note, a previous study has established a connection between NAD + metabolism and ADPKD by showing that administration of nicotinamide led to inhibition of NAD +dependent sirtuin-1 and partially suppressed cyst development in mice (Zhou et al., 2013). These effects were not confirmed in a subsequent patient trial (El Ters et al., 2020).
We further explored the NNT-dependent CTT rescue mechanism by evaluating NNT expression and enzymatic activity. We observed a significant increase in NNT expression in our CTT rescue model compared to cystic N-Pkd1-KO mice at the 16week time point. This finding correlates with the higher quantities of NNT detected in healthy human kidney tissue relative to kidney tissue from ADPKD patients.
Interestingly, differences in NNT levels between "N" mice that do or do not express CTT are not observed shortly after Pkd1 gene disruption, however, there is already significantly greater enzymatic activity in the N-Pkd1-KO+CTT rescue mice compared to N-Pkd1-KO mice alone at this pre-cystic stage, prior to the development of gross morphological alterations. It is important to note that, to make APAD and NADPH substrates accessible to NNT, it is necessary to disrupt membrane integrity, which eliminates the NNT-driving protonmotive force. Thus, while NNT enzymatic activity can be measured, the driving forces that determine the directionality of the enzymatic reaction remain to be defined. Consequently, we cannot state with certainty whether the effects of CTT expression on NNT ameliorates or exacerbates oxidative damage in cystic cells. It seems likely, however, that the suppressive effects of CTT on cyst development are associated with its capacity to effect NNT-dependent changes in the redox environment. It is, of course, also possible that the observed rescue is not a direct product of NNT enzymatic activity but rather a consequence of NNT-dependent alterations in the mitochondrial protonmotive force. Further experiments will be required to elucidate fully the mechanism of NNT-dependent CTT action.
In summary, we have identified a novel interaction between the mitochondrial enzyme NNT and the C-terminal tail of PC1. More importantly, we showed that expressing the Cystic mouse cohorts are composed of 53%-58% female and 42%-47% male mice.
Data are expressed as mean ± SEM. Pairwise comparisons were performed using Student's t-test. See also Figure S1. (E) Immunoblotting of total kidney lysate from WT "N" and "J" mice, confirming the presence and absence of NNT, respectively.
Scale bar:10 μm. See also table S1.     Data are expressed as mean ± SEM. Pairwise comparisons were performed using Mann-Whitney U test due to non-normally distributed data (B and C).
Multiple group comparisons were performed using one-way ANOVA followed by Tukey's multiple-comparisons test (E, F, H and J).

Lead Contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Michael J. Caplan (michael.caplan@yale.edu).

Materials Availability
The 2HA-PC1-CTT; Pkd1 fl/fl ; Pax8 rtTA ; TetO-Cre mice on both C57BL/6J and C57BL/6N backgrounds were generated in this study and are available through the Yale University School of Medicine.

Data and code availability
The mass spectrometry proteomics data have been deposited to the Linearized modified BAC DNA purified by CHEF electrophoresis was used for pronuclear injection to generate transgenic founder lines. The BAC transgenic lines were produced in (C57BL/6J X SJL/J) F2 zygotes. Founders were identified by PCR genotyping, verified by sequencing of PCR products and BAC copy number was determined by genomic quantitative PCR as described previously (Dong et al., 2021, Cai et al., 2014, Fedeles et al., 2011. Two BAC founders with BAC copy numbers 2 or 4 were used in this study. All strains were backcrossed at least four generations with C57BL6 and are therefore expected to be at least 90% C57BL6 congenic. These animals were then crossed with Pkd1 fl/fl ; Pax8 rtTA ; TetO-Cre mice to generate 2HA-PC1-CTT; Pkd1 fl/fl ; Pax8 rtTA ; TetO-Cre, on both C57BL6/J and C57BL6/N backgrounds.

Cell Lines
HEK293 cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin, and 1% l-glutamine at 37 o C. These cells were then subjected to transient transfection following the protocol described in the transient transfection section of Methods.

Human specimens
The Baltimore Polycystic Kidney Disease Research and Clinical Core Center (P30DK090868) provided human kidney tissue from both ADPKD patients and nonaffected controls. Samples were surgically harvested according to the guidelines established by the Institutional Review Board of the University of Maryland and were then de-identified. Sex information was not available. Immunoblotting with NNT and actin antibodies was performed using the protocol described in the Western blot section of Methods.

Mouse kidney tissue harvest
Mice were euthanized according to Yale IACUC protocols. Tail or toe tissue from previously genotyped mice was acquired for a second time when animals were under anesthesia for genotype confirmation. Retro-orbital blood was also collected from these anesthetized mice. The left kidney was excised, weighed, snap-frozen in liquified N2 and stored at -80 o C for biochemistry analysis. The right kidney was excised, weighed and fixed in 4% paraformaldehyde. Fixed kidneys were then sectioned in half along their sagittal axes, infiltrated with 30% sucrose overnight and embedded in OCT for further imaging. Fischer Scientific) were equilibrated in lysis buffer (50 μL beads per reaction in 500μl of lysis buffer) for 10 min at RT on a rocking shaker, and then incubated with a total of 1ml lysis buffer, the dynabeads were magnetically recovered and precipitated proteins were eluted in 60μl of 2x Laemmli sample buffer (catalog#1610747, Bio-Rad) with 300mM DTT. 1/3 of the eluted proteins (20μl) was loaded per sample per gel for immunoblotting.

Proteomic analysis
Proteomic analysis on material immunoprecipitated from the Pkd1 F/H -BAC kidneys was performed by the Mass Spectrometry and Proteomics Resource of the W.M. Keck Foundation Biotechnology Resource Laboratory at Yale according to their standard operating procedures. Immunoprecipitated proteins were subjected to chloroform:methanol:water protein extraction, after which they were reduced, alkylated and trypsin digested for follow up LC MS/MS bottom-up data collection.

Samples were analyzed with an Orbitrap Fusion mass spectrometer and Mascot
Search Engine software was utilized for protein identifications.

Transient transfection in cultured cells
We used Lipofectamine 2000 (catalog# 11668019, ThermoFischer Scientific) to transiently transfect HEK293 cells following the manufacturer's protocol. Cells were transfected with the previously described 2HA-PC1-CTT construct (Merrick et al., 2019). Briefly, the sequence encoding the final 200 aa of human  with an N-terminal 2xHA tag was cloned into the pcDNA3.1 zeo vector. The 2HA-PC1-CTT sequence is identical to that expressed in 2HA-PC1-CTT; Pkd1-KO mice.

Immunofluorescence staining in cells
HEK293 cells grown on poly-L coated coverslips were fixed with 4% PFA in PBS for 30 min at room temperature followed by a 15-min treatment with permeabilization buffer (PBS, 1mM MgCl2, 0.1mM CaCl2, 0.1% BSA, 0.3 % Triton X100). Cells were then blocked with goat serum dilution buffer (GSDB; 16% filtered goat serum, 0.3% Triton X-100, 20mM NaPi, pH 7.4, 150 mM NaCl) for 30 min, followed by a one-hour incubation with primary antibodies (1:100) diluted in GSDB. The primary antibodies utilized were anti-PC1-C-terminus (catalog #EJH002, Kerafast) and anti-NNT (catalog #459170, Invitrogen). Following 3 PBS washes, samples were incubated with secondary antibodies (1:200) diluted in GSDB for one hour and then washed again with PBS. Alexa Fluor conjugated antibodies (Alexa-594, 647; catalog #A11032 and #A31573 respectively, Life Technologies Invitrogen) were used as secondary reagents. Finally, coverslips were mounted on slides with VectaShield mounting medium (catalog # H-1000-10, Vector Laboratories) and imaged using a Zeiss LSM780 confocal microscope. Images are the product of 8-fold line averaging and contrast and brightness settings were chosen so that all pixels were in the linear range.
Anti-Na,K-ATPase α-subunit was used as a tubular marker. Three images were acquired in the upper, middle and lower third of the kidney by a blinded investigator who also quantified the percentage of ki67 positive nuclei relative to total tubular nuclei in these 9 independent images. A total of at least 2000 tubular nuclei were counted per animal.

Morphological analyses
Whole kidney images from hematoxylin and eosin-stained sagittal kidney sections were obtained at a 4x magnification using automated image acquisition by the scan slide module in MetaMorph (Molecular Devices). The whole kidney was defined as the region of interest and the ImageJ default auto threshold function was employed to measure cystic and tubular area relative to total kidney area.

Genomic DNA isolation and quantitative RT-PCR
The DNeasy Blood & Tissue kit was used to extract genomic DNA from all 2HA-PC1-CTT; Pkd1 fl/fl ; Pax8 rtTA ; TetO-Cre and Pkd1 fl/fl ; Pax8 rtTA ; TetO-Cre mice included in the 16-week cohort, starting with 20mg of kidney tissue from each animal and following the manufacturer's instructions. Of note, we performed the optional 2-min treatment with 4μl of RNAse A (100mg/ml) at room temperature to obtain RNA-free genomic DNA in transcriptionally active tissues. Quantitative RT-PCR (qRT-PCR) was performed using iTaq Universal SYBR Green Supermix (catalog# 172-5121, Bio-rad).
All samples were loaded in triplicates and reactions and data acquisition were performed using the Agilent real-time PCR system with its associated software. Mouse samples were normalized to DNA obtained from a control mouse that expressed a single copy of both Pax8 rtTA and TetO-Cre. All animals included in the present cohort presented a 1:1 or 2:1 ratio for both genes when compared to controls, confirming homozygosity or heterozygosity for both Pax8 rtTA and TetO-Cre alleles.

QUANTIFICATION AND STATISTICAL ANALYSES
Data quantification and plotting were performed using GraphPad Prism software (https://www.graphpad.com/scientific-software/prism/), with the exception of metabolomic data, which were analyzed and plotted with ClustVis and R as described in the "Metabolomics" section of methods. Data were expressed as means  SEM.
Student's t-test or Mann-Whitney U test was used for pairwise comparisons, as indicated in the figure legends. One-way analysis of variance (ANOVA) followed by Tukey's multiple-comparison test was used for multiple comparisons. P<0.05 was considered statistically significant. Sample sizes were determined based on our experience working with the Pkd1 fl/fl ; Pax8 rtTA ; TetO-Cre, and Pkd1 F/H -BAC mouse models; no prior power analysis was performed. Sex distribution was similar between the compared groups. The exclusion criteria were based on mouse well-being; no mice were excluded from this study.  Table S1: Comparative proteomic analysis of material immunoprecipitated from crude renal mitochondrial fractions with anti-HA antibodies from Pkd1 F/H -BAC and WT mice, generated on mixed backgrounds (related to Figure 2C). The table lists the X values (log2 Fold change BAC/control) and Y values (-log10 P value), as depicted in the volcano plot in Figure 2C, for all identified peptides and indicates the targets with P value <0.05.