Evolving adeno-associated viruses for gene transfer to the kidney via cross-species cycling of capsid libraries

Alan Rosales; Leo O Blondel; Joshua Hull; Qimeng Gao; Nihal Aykun; Jennifer L Peek; Alejandra Vargas; Sophia Fergione; Mingqing Song; Matthew H Wilson; Andrew S Barbas; Aravind Asokan

doi:10.1038/s41551-024-01341-0

. 2025 Feb 5;9(7):1086–1100. doi: 10.1038/s41551-024-01341-0

Evolving adeno-associated viruses for gene transfer to the kidney via cross-species cycling of capsid libraries

Alan Rosales ¹, Leo O Blondel ², Joshua Hull ², Qimeng Gao ², Nihal Aykun ², Jennifer L Peek ³, Alejandra Vargas ², Sophia Fergione ², Mingqing Song ², Matthew H Wilson ^3,^4,^5,⁶, Andrew S Barbas ², Aravind Asokan ^1,^2,^7,^✉

PMCID: PMC12261118 NIHMSID: NIHMS2091088 PMID: 39910375

Abstract

The difficulty of delivering genes to the kidney has limited the translation of genetic medicines, particularly for the more than 10% of the global population with chronic kidney disease. Here we show that new variants of adeno-associated viruses (AAVs) displaying robust and widespread transduction in the kidneys of mice, pigs and non-human-primates can be obtained by evolving capsid libraries via cross-species cycling in different kidney models. Specifically, the new variants, AAV.k13 and AAV.k20, were enriched from the libraries following sequential intravenous cycling through mouse and pig kidneys, ex vivo cycling in human organoid cultures, and ex vivo machine perfusion in isolated kidneys from rhesus macaques. The two variants transduced murine kidneys following intravenous administration, with selective tropism for proximal tubules, and led to markedly higher transgene expression than parental AAV9 vectors in proximal tubule epithelial cells within human organoid cultures and in autotransplanted pig kidneys. Following ureteral delivery, AAV.k20 efficiently transduced kidneys in pigs and macaques. The AAV.k13 and AAV.k20 variants are promising vectors for therapeutic gene-transfer applications in kidney diseases and transplantation.

Subject terms: Gene delivery, Gene therapy

Variants of adeno-associated viruses displaying robust and widespread transduction in murine, porcine and non-human-primate kidneys, and in human kidney organoids, can be obtained via the cross-species cycling of capsid libraries in the different kidney models.

Main

Chronic kidney disease (CKD) is estimated to affect 8–16% of the population worldwide and has increased by 31.7% over the past 10 years^1,2. Ultimately, CKD progresses to end-stage renal disease, where dialysis or kidney transplantation is the only viable option for renal replacement therapy^3,4. Maintenance dialysis therapy requires multiple prolonged sessions per week and is characterized by poor patient survival. Kidney transplantation has developed into a long-term therapy, but the field remains limited by donor organ scarcity and the need for lifelong immunosuppression^5,6. Many kidney diseases such as cystinuria, polycystic kidney disease and cystinosis among others have underlying genetic aetiologies that may be amenable to gene therapy or genome editing, underscoring the crucial unmet need for an effective and safe kidney-targeting gene delivery vehicle^7–10.

Within this framework, recombinant adeno-associated virus (AAV) vectors constitute a promising technology for human gene therapy^11,12. With the approval of five AAV-based products by the US Food and Drug Administration (FDA) for treating ocular, haematological and neuromuscular disorders, there are ~200 completed or active gene therapy clinical trials registered at ClinicalTrials.gov as of April 2024. However, little progress has been made so far in achieving effective therapeutic gene transfer to kidneys. Due to the intrinsic filtering function and complex physiology of kidneys, gene delivery has proven challenging^8,13.

Here we evolve new cross-species-compatible AAV-kidney (AAV.k) variants by cycling AAV capsid libraries across different kidney systems¹⁴. Our robust approach utilizes intravenous (i.v.) dosing in mice and pigs, ex vivo machine perfusion in isolated rhesus macaque kidneys using both ureteral and arterial delivery, as well as human organoid cultures to cycle AAV capsid libraries. By employing a multimodel and multispecies evolution strategy in vivo and ex vivo, we discover and characterize novel AAV.k variants that show promise as delivery vehicles for therapeutic kidney gene transfer.

Results

Cycling capsid libraries across different kidney systems yields new AAV variants

To engineer AAV.k variants, we generated AAV9-based capsid libraries, which were subjected to several evolution tracks involving evolution in mice and pigs via intravenous administration, infectious cycling ex vivo on human kidney organoids, and ex vivo machine perfusion through isolated non-human primate (NHP) kidneys via arterial and ureteral routes (Fig. 1a). Capsid libraries were constructed via saturation mutagenesis of AAV9 variable region (VR) IV, corresponding to amino acids 452–458 (VP1 subunit numbering). Studies from our lab and others have demonstrated the critical role played by this surface epitope located at the 3-fold symmetry axis in cellular uptake, transduction and neutralizing antibody recognition^14–19.

Fig. 1 — a, Schematic of AAV capsid library evolution in mice and pigs via intravenous dosing on human kidney organoids and ex vivo non-human primate kidneys following perfusion of the arterial and ureteral route. b, Next-generation sequencing of AAV capsid library evolved from human kidney organoids. In b–d and f, black dots represent individual 7-mer amino acid sequences c, NGS of AAV capsid library evolved from mouse kidney. d, NGS of AAV capsid library evolved from human kidney organoids following an initial evolution in mouse kidneys. e, Left: NGS of AAV capsid library evolved from pig kidney cortex (blue) and medulla (orange). Right: zoom of top right quadrant of NGS plot demonstrating enrichment of specific 7-mer amino acid sequences within the pig kidney cortex, with AAV9 encircled in brown. f, NGS of AAV capsid library evolved from human kidney organoids following an initial evolution in mouse kidneys and pig kidneys. g, Left: NGS of AAV capsid library evolved from ex vivo perfusion of non-human primate kidney via the arterial route (blue) and ureteral route (orange). Right: zoom of top right quadrant of NGS plot demonstrating enrichment of specific 7-mer amino acid sequences within ex vivo perfused non-human primate kidneys. In b–g, AAV9 is encircled in brown; the X axis represents the read depth of each sequence found within the library amplified from the specified tissue, and the Y axis represents the fold change enrichment vs the starting parental library. h, Consensus motif analysis of the top 100 enriched AAV for pig cortex and medulla evolution for residues 452 − 458 (VP1 numbering). i, Consensus motif analysis of the top 100 enriched AAV for ex vivo NHP arterial and ureteral evolution for residues 452 − 458 (VP1 numbering).

Wild-type AAV capsid libraries packaging genomes consisting of AAV2 Rep and AAV9 mutant Cap, flanked by AAV2 inverted terminal repeats (ITRs), were sequentially cycled and enriched variants amplified from mouse, pig and (isolated) NHP kidneys as well as human organoid cultures. Next-generation sequencing (NGS) analysis of each step in the library cycling process plotted as a function of fold enrichment (compared to parental library) and read depth enabled identification of clones enriched across species (Fig. 1b–g). Interestingly, in the case of track 1, evolving the parental library directly on human kidney organoids first provided little to no selective pressure as noted by the tight clustering of unique sequences in the top right quadrant of the NGS analysis graph (Fig. 1b). This is also noted in the case of track 3 after further evolving the library initially cycled through mouse kidneys (Fig. 1d). This separation of unique sequence clusters was maintained in human kidney organoids only after evolving in pig kidneys (Fig. 1e), as seen in track 5 (Fig. 1f). Lastly, track 6 yielded several unique sequences above the threshold for read depth and fold change (Fig. 1g).

Notably, in the case of the pig kidney isolated after intravenous library dosing, we determined specific variants enriched in the kidney cortex vs the medulla (Fig. 1e,h and Supplementary Fig. 1a). Of the ~12,300 and 10,000 unique sequences that were enriched in the cortex and medulla, respectively, only ~1,100 overlapping variants were determined. Moreover, further analysis in ex vivo machine-perfused NHP kidneys enabled comparison between clones enriched via ureteral vs arterial delivery routes (Fig. 1g,i and Supplementary Fig. 1b). In this scenario, 4,552 and 4,711 unique sequences were enriched from the arterial and ureteral routes, respectively, with 5,913 overlapping variants. To better understand the selective pressure exerted by the different tracks of AAV evolution in multiple kidney models, we compared the relative enrichment of the top two variants, AAV.k13 and AAV.k20 to the parental AAV9 in each track (Supplementary Table 2). Notably, parental AAV9 was only enriched between 1–3-fold across different tracks, while AAV.k13 exhibited a ~1,000-10,000-fold increase, with the highest fold change observed in track 5 (mouse–pig–human kidney organoid). Strikingly, the top capsid, AAV.k20, displayed enrichment up to ~20,000-fold in track 5 (mouse–pig–human kidney organoid) and ~25,000–28,000-fold increase with ex vivo perfused NHP kidneys. These results suggest that selective pressure from human kidney organoids and then NHP kidney afforded the most robust selection and yielded the best performing cross-species-compatible AAV.k variants.

Deeper analysis of NGS data from theoretical, parental (packaged into AAV) and evolved libraries at each step of the evolution process was performed by calculating amino acid frequency at each position in the variable region IV loop (Supplementary Fig. 2). The parental library shows a broad distribution of amino acids for each position, with some structural bias for Gly, Ala, Val, Ser and Arg residues as illustrated by red colour intensity. Cys residues, which are prone to disulfide formation and aromatic side-chain hydrophobic amino acids such as Trp, Tyr and Phe are generally not preferred from a structural compatibility standpoint. As the evolution progresses with increasing complexity of the model, we observed a strong deviation in amino acid preferences from mouse to higher organisms. Striking preferences for basic Arg residues, but not Lys in positions 456 and 458 were observed for AAV variants enriched in human organoid and NHP kidney tissues. The different routes of administration (arterial vs ureteral), however, did not display notably different trends in amino acid preferences as discussed earlier. In contrast, subtle yet possibly meaningful differences were observed for long-chain hydrophilic residues, that is, Asp, Glu, Asn and Gln, between the different regions (cortex vs medulla) in pig kidneys. Overall, AAV.k13 and AAV.k20 showed distinct amino acid sequences in the VR-IV region compared with AAV9 (sequence alignments shown in Supplementary Fig. 3a). The structure–function correlates of such preferences will require additional mutational analysis in future studies.

AAV.k13 and AAV.k20 transduce three-dimensional (3D) human kidney organoids with high efficiency

The top two AAV.k variants were first compared to AAV9 by determining vector genome titres following research-scale production. Regardless of transgene choice, no significant difference in total yields of AAV.k13 and AAV.k20 compared to AAV9 were observed (Supplementary Fig. 3b). Briefly, 3D human kidney organoids were generated in suspension and validated by quantitation of several stem cell markers and human kidney organoid differentiation markers (Extended Data Fig. 1a)²⁰. Transduction profiles of AAV9 and AAV.k variant vectors packaging self-complementary chicken–beta-actin hybrid (CBh)–mCherry incubated at 1 × 10¹¹ vector genomes (vg) or 1 × 10¹² vg total were compared in differentiated 3D human kidney organoids at 5 days post transduction (Fig. 2a). Native mCherry fluorescence from cryosections shows widespread and robust transduction at 1 × 10¹¹ vg for AAV.k13 and AAV.k20 but minimal expression to AAV9 (Fig. 2b). Quantification of relative mCherry fluorescence to DAPI nuclear staining revealed that AAV.k13 and AAV.k20 had ~8- and 12-fold higher fluorescence intensity, respectively, compared with AAV9 (Fig. 2d). At the higher dose of 1 × 10¹² vg per well, we observed a saturating effect, wherein AAV.k13 and AAV.k20 had ~1.5- and 3-fold higher fluorescence intensity, respectively, compared with AAV9 (Fig. 2e). While no significant difference in mCherry mRNA levels was noted between AAV9 and AAV.k13 at 1 × 10¹¹ vg, AAV.k20 had the highest mRNA levels across all capsids (Fig. 2f). Similarly, at 1 × 10¹² vg, there was no significant difference in mCherry mRNA levels between AAV9 and AAV.k13, while AAV.k20 had considerably higher levels of mCherry mRNA (Fig. 2f). We also compared AAV9, AAV.k13 and AAV.k20 to AAV5, which showed minimal to undetectable transduction in human kidney organoids at 1 × 10¹¹ vg and 1 × 10¹² vg (Extended Data Fig. 1b).

To determine cellular specificities for AAV9, AAV.k13 and AAV.k20, we analysed organoids transduced with 1 × 10¹² vg (Fig. 2c and Extended Data Fig. 1c). While some sporadic transduction was observed in all cases, it was noted that mCherry expression at proximal tubules labelled with lotus tetragonolubus lectin (LTL) at the brush border was markedly correlated for AAV.k13 and AAV.k20 compared with AAV9. This is highlighted by the robust mCherry expression encircling the LTL-labelled brush border of proximal tubules indicated by white arrowheads and white outlines (Fig. 2c and Extended Data Fig. 1c). In addition, co-immunostaining for PECAM-1 and nephrin revealed no appreciable overlap with mCherry expression for the three capsids (data not shown). While the exact nature of other cell types transduced in human kidney organoids remains to be examined, the results strongly justified evaluating the potential translatability of AAV.k vectors to kidney models of other species with specific focus on achieving robust and specific transduction of proximal tubule epithelial cell types using a ubiquitous promoter.

Further, 3D human kidney organoids were also transduced with AAV9, AAV.k13 and AAV.k20 packaging a single-stranded chicken–β-actin (CBA)–luciferase cassette at 1 × 10¹¹ vg and 1 × 10¹² vg total and analysed for uptake and luciferase activity at 5 days post transduction (Fig. 2a). While no significant uptake was observed between AAV9 and AAV.k13, AAV.k20 had significantly more uptake than AAV9 and AAV.k13 for 1 × 10¹¹ vg (Fig. 2h). Meanwhile, both AAV.k13 and AAV.k20 had notably higher uptake levels compared with AAV9 but were comparable to each other at 1 × 10¹² vg (Fig. 2h). Assessment of luciferase expression revealed higher but not significant luciferase expression with AAV.k13 compared with AAV9 for both 1 × 10¹¹ vg and 1 × 10¹² vg conditions (Fig. 2g). However, AAV.k20 revealed the highest luciferase expression compared with AAV9 and AAV.k13 for both 1 × 10¹¹ vg and 1 × 10¹² vg conditions (Fig. 2g). These results corroborate that both single-stranded (ss) and self-complementary (sc) vector genomes are compatible with AAV.k vectors in achieving robust kidney gene expression.

To further probe the tropism for proximal tubule cells, we evaluated AAV.k variants’ capability of transducing TH1 cells (Kerafast), an immortalized human renal proximal tubule epithelial cell line (RPTEC). TH1 cells seeded on a 96-well plate were transduced with AAV9, AAV.k13 or AAV.k20 packaging a single-stranded luciferase cassette driven by a CBA promoter at a multiplicity of infection (MOI) of 5,000, 10,000 and 50,000 (Extended Data Fig. 2a). At 5 days post transduction, a luciferase assay was performed to assess expression levels. At each of the different MOIs, AAV.k20 considerably outperformed AAV9 and AAV.k13, while AAV.k13 had slightly higher, but not significant luciferase expression compared with AAV9 (Extended Data Fig. 2b). Further, TH1 cells were seeded on a 96-well plate and transduced with AAV9, AAV.k13 or AAV.k20 packaging a self-complementary mCherry cassette driven by a CBh promoter at an MOI of 10,000 and 50,000 (Extended Data Fig. 2a). Assessment of fluorescence imaging of mCherry performed at 5 days post transduction revealed higher levels or mCherry transduction with AAV.k variants (AAV.k20 > AAV.k13 > AAV9) at an MOI of 50,000, but comparable levels of expression at an MOI of 10,000 (Extended Data Fig. 2c). These results highlight TH1 cells as an additional model for further studying the transduction mechanisms of AAV.k variants in vitro.

Extended Data Fig. 2 — A) Human renal proximal tubule epithelial cell (RPTEC) cells were seeded a 1E4 cells/well in a 96-well plate. RPTECs were then transduced at three MOIs: 5 K, 10 K, and 50 K with AAV9, AAV.k13 and AAV.k20 packaging a single-stranded cassette encoding for luciferase driven by a chicken β-actin (CBA) promoter or a self-complementary cassette encoding for mCherry driven by a chicken β-actin hybrid (CBh) promoter. A luciferase expression assay or fluorescence imaging was performed five days post transduction. B) Luciferase expression levels were determined for each MOI from each well and represented as relative luminescence units, where each dot represents a well. C) Representative fluorescent images of TH1 cells at different MOIs. For the box plots, the center line represents the median, the box limits are upper and lower quartiles, and whiskers represent 1.5x interquartile range. Statistical significance was determined by one-way ANOVA with Fishers LSD post-test for luciferase expression in TH1 cells. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns is not significant.

Source data

AAV.k13 and AAV.k20 display tropism for proximal tubules in mouse and NHP models

We compared the transduction profiles of AAV.k13, AAV.k20 and AAV9 vectors packaging a self-complementary CBh–mCherry cassette administered intravenously at a dose of 5 × 10¹³ vg kg⁻¹ at 4 weeks post administration (Fig. 3a). Native mCherry fluorescence from kidney cryosections shows widespread and robust transduction with AAV.k13 and AAV.k20 vectors compared with AAV9 (Fig. 3b). Specific kidney markers were then used to determine localization of mCherry expression using immunofluorescence microscopy. No localization of mCherry expression was observed in glomeruli, specifically, podocytes, on the basis of anti-nephrin immunofluorescence (Fig. 3f). Moreover, no localization of mCherry expression was observed in collecting ducts by Dolichos biflorus agglutinin (DBA) staining (Fig. 3f). Robust transduction of proximal tubules was observed on the basis of co-localization of mCherry to LTL staining (Fig. 3c). Semi-quantitative assessment of relative fluorescence (mCherry to LTL) revealed that AAV.k13 and AAV.k20 had ~3- and 9-fold higher fluorescence intensity, respectively, compared with AAV9 (Fig. 3d). To further characterize proximal tubule regions transduced by AAV.k13 and AAV.k20, anti-SGLT1 and anti-SGLT2 antibodies were used to stain the early S1 segment and distal S2 segment of proximal tubules, respectively (Fig. 3e). Relatively higher mCherry expression was determined in early proximal tubule segments on the basis of co-localization of mCherry with SGLT2 staining.

Fig. 3 — a, C57BL/6J mice (8-week-old) were injected intravenously (i.v.) with AAV9, AAV.k13 and AAV.k20 packaging a self-complementary cassette encoding for mCherry driven by a CBh promoter at a dose of 5 × 10¹³ vg kg⁻¹. Organs were collected at 30 days post injection. b, Representative images of native mCherry fluorescence in mouse kidney for mock, AAV9, AAV.k13 and AAV.k20. c, Representative images of native mCherry fluorescence and LTL staining of proximal tubules for AAV9, AAV.k13 and AAV.k20. White arrowheads depict mCherry expression in kidney cross-section. d, Quantification of native mCherry fluorescence intensity normalized to LTL signal in kidney for AAV9 compared to AAV.k13 and to AAV.k20, where fold change is listed above P values (total n = 3 mice, n = 13 images). e, Representative images of immunofluorescence labelling of S1 early segment of the proximal tubule (SGLT2) and S2 distal segment of the proximal tubule (SGLT1) and mCherry expression in the mouse kidney. f, Representative images for immunofluorescence labelling of mCherry and collecting ducts with DBA and podocytes with nephrin in mouse kidneys for AAV.k13 and AAV.k20. g, Representative images of native mCherry fluorescence and LTL staining of proximal tubules for AAV5. h,i, Representative images of native mCherry fluorescence in mouse livers (h) and mouse hearts (i) for AAV9, AAV.k13, AAV.k20 and AAV5. j,k, Vector genome copy numbers per μg DNA for liver (j) and kidney (k) were calculated by normalizing mCherry copy numbers to the total μg DNA input for qPCR quantification, and subsequently normalized to mock, where each dot represents an individual mouse (n = 3). Total viral genome load for liver or kidney was calculated on the basis of the average genomic DNA extraction from liver or kidney sample and average weight of mouse livers or kidneys. For boxplots, the centre line represents the median, the box limits are upper and lower quartiles, and whiskers represent 1.5× interquartile range. Statistical significance was determined using one-way ANOVA with Tukey’s post test for mCherry fluorescence in mouse kidneys, and using Brown–Forsythe and Welch ANOVA test for liver and kidney uptake compared to AAV.k20.

Source data

To further extend our analysis of human kidney organoid data, we compared AAV5 to AAV.k variants packaging a self-complementary CBh–mCherry cassette administered intravenously at a dose of 5 × 10¹³ vg kg⁻¹ at 4 weeks post administration (Fig. 3a). Vector genomes were normalized to mock tissue and compared across AAV5, AAV9, AAVk.13 and AAV.k20. No significant difference in vector genomes per μg of DNA or total viral genome load per kidney was noted between AAV5, AAV9, AAV.k13 and AAV.k20 (Fig. 3k). AAV5 showed higher accumulation in the liver overall, while there was no difference between AAV9, AAV.k13 and AAV.k20 (Fig. 3j). Notably, all capsids showed approximately one log order of magnitude higher vector genome copies in the liver compared with the kidney following intravenous dosing. mCherry expression in mice was correspondingly higher in the liver based on native fluorescence (Fig. 3h). mCherry expression in mice was comparable in the heart and minimal within the kidney in the case of AAV5 (Fig. 3i,g). Overall, these results corroborate that AAV.k variants are superior in transducing both human and mouse kidney tissue compared with AAV5 and AAV9.

We next evaluated the ability to deliver a therapeutically relevant transgene in C57BL/6J mice. Specifically, we chose programmed death-ligand 1 (PD-L1), which has been implicated in organ transplant tolerance, with expression on host target tissues potentially reducing the likelihood of transplant rejection²¹. Using AAV.k20, a self-complementary cassette encoding for C-terminus HA-tagged PD-L1 driven by a CBh promoter, mice (n = 4) were dosed at 5 × 10¹³ vg kg⁻¹ and kidneys were collected at 4 weeks post administration (Fig. 4a). Successful delivery of AAV.k20–scCBh–PD-L1–HA was confirmed by viral genomes detected in treated mouse kidneys (Fig. 4b). PD-L1–HA expression in mouse kidneys was confirmed by western blotting, with vinculin used as a loading control. All four treated mouse kidneys demonstrated robust expression of PD-L1–HA, while the mock kidney showed no expression (Fig. 4c). Therefore, in addition to scAAV mCherry data, the PD-L1 expression data further confirm the potential utility of AAV.k vectors for expression of clinically relevant transgenes.

Fig. 4 — a, C57BL/6J mice (8-week-old) were injected intravenously with AAV9, AAV.k13 and AAV.k20 packaging a self-complementary cassette encoding for C-terminus HA-tagged PD-L1 driven by a CBh promoter, or a single-stranded cassette encoding for luciferase driven by a CBA promoter at a dose of 5 × 10¹³ vg kg⁻¹. Organs were collected at 30 days post injection. b, Vector genome copy numbers per μg DNA for kidney were calculated by normalizing PD-L1 copy numbers to the total μg DNA input for qPCR quantification and plotted as log vg μg⁻¹ DNA, where each dot represents an individual mouse (n = 4). c, Western blot analysis of PD-L1–HA expression in mouse kidneys (#1–4) treated with AAV.k20–PD-L1–HA and untreated mock mouse kidney. Antibodies against HA-tag and vinculin (~145 kDa) were used to confirm expression of PD-L1–HA (~51 kDa) and equal loading, respectively. d–f, Vector genome copy numbers per μg DNA for heart (d), liver (e) and kidney (f) were calculated by normalizing luciferase copy numbers to the total μg DNA input for qPCR quantification and plotted as log vg μg⁻¹ DNA, where each dot represents an individual mouse (n = 4). Total viral genome load for heart/liver/kidney was calculated on the basis of average genomic DNA extraction from heart/liver/kidney samples and average weight of mouse hearts/livers/kidneys. g–i, Luciferase expression levels for mouse hearts (g), livers (h) and kidneys (i) were normalized for total protein concentration and represented as relative luminescence units per gram of tissue (RLU g⁻¹), where each dot represents an individual mouse (n = 4). j, AAV.k20 packaging a single-stranded cassette encoding for luciferase driven by a CBA promoter was delivered to both kidneys of the NHP in situ via the ureter following a midline laparotomy, and kidneys were recovered at 30 days post delivery. k, Vector genome copy numbers per μg DNA for kidney were calculated by normalizing luciferase copy numbers to the total μg DNA input for qPCR quantification and plotted as log vg μg⁻¹ DNA, where each dot represents a biopsy of the NHP AAV.k20-treated left and right kidneys (n = 10). The dashed black line represents baseline from a mock NHP kidney. l, Luciferase expression levels were normalized for total protein concentration and represented as RLU g⁻¹ tissue, where each dot represents a biopsy (n = 10). The dashed black line represents baseline from a mock NHP kidney. For boxplots, the centre line represents the median, the box limits are upper and lower quartiles, and whiskers represent 1.5× interquartile range. Statistical significance was determined using unpaired t-test for kidney PD-L1 viral genome uptake, and using Kruskal–Wallis test with post hoc Dunn’s test for luciferase biodistribution and expression in mouse kidneys. Individual P values for each comparison are included in each figure panel.

Source data

We also compared the transduction profiles of AAV9, AAV.k13 and AAV.k20 vectors packaging a single-stranded CBA–luciferase cassette administered intravenously at a dose of 5 × 10¹³ vg kg⁻¹ at 4 weeks post administration (Fig. 4a). While no significant difference in vector genomes was noted in mouse heart and liver, there was slightly less vector genomes in the mouse kidney for AAV.k20 compared with AAV9 (Fig. 4d–f). In addition, mice showed no significant difference in luciferase activity for heart and liver tissues between AAV9, AAV.k13 and AAV.k20 (Fig. 4g,h). Interestingly, assessment of luciferase expression in the mouse kidneys revealed that AAV.k13 and AAV.k20 had ~40- and 22-fold higher luciferase expression, respectively, compared with AAV9 despite little to no difference in kidney uptake of viral genomes (Fig. 4i). These results are consistent with human kidney organoid data and confirm that both ss- and scAAV vectors are compatible with AAV.k capsids in achieving robust kidney gene expression in mice. Although the liver accumulation of AAV.k13/AAV.k20 is comparable to AAV9 in mice following i.v. dosing, this prompted us to additionally evaluate direct administration routes to assess kidney transduction.

Given the success of AAV.k20 in human kidney organoids and in C57BL/6J mice, we further assessed its translatability by evaluating in situ delivery of AAV.k20 packaging a single-stranded luciferase cassette driven by a CBA promoter to a live rhesus macaque via the ureter (Fig. 4j). To interrogate the transduction profile in NHP kidneys, 10 biopsies were taken from the treated kidneys and assessed for viral genomes and luciferase expression levels (Supplementary Fig. 4a). A range of viral genome uptakes well above the mock kidney baseline was determined from the 10 biopsies for both kidneys (Fig. 4k and Supplementary Fig. 4b). Assessment of luciferase expression was determined for each biopsy and again revealed a range of luciferase activities compared with mock kidney for each biopsy (Fig. 4l and Supplementary Fig. 4c). Taken together, AAV.k20 demonstrates robust transduction in human kidney organoids, mouse and NHP kidneys, exemplifying the translational potential of this newly evolved capsid. Further, it is noteworthy to mention that AAV.k20 afforded robust kidney expression through intravenous or ureteral routes, with implications for future clinical dosing strategies.

AAV.k13 and AAV.k20 show robust and widespread transduction in pig kidneys ex vivo

To further assess translatability of AAV.k vectors in transplant applications and evaluate the effect of vascular and ureteral routes, AAV.k vectors were delivered to pig kidney grafts ex vivo. Kidney grafts treated either through direct administration during cold storage or during ex vivo machine perfusion were transplanted and assessed at 1–2 weeks post transplant to determine transgene expression. Briefly, in current clinical practice, kidney grafts are preserved by either static cold storage on ice or by ex vivo machine perfusion. Machine perfusion, although more complex, may have some advantages due to its ability to improve kidney viability, deliver targeted treatment and even reduce immunogenicity^22–25. Due to their similar structure and renal physiology to humans, pigs have been particularly well studied for preclinical evaluation of kidney-focused therapeutics^26–29. For AAV.k13, the efficacy of delivery during static cold storage was assessed. Following nephrectomy, the pig kidney was flushed with preservation solution and stored on ice. Half of the total AAV.k13 vector dose was administered via the arterial route and the other half through retrograde ureteral delivery. In this preliminary assessment, the pig kidney graft was stored on ice for 2 h before autotransplantation in the same animal (Fig. 5a). Following transplantation, immunohistochemistry (IHC) of mCherry protein performed on biopsies of pig kidney graft tissue revealed prominent and widespread expression in proximal tubules, with some expression in distal tubules but no signal in the glomerulus.

Fig. 5 — a, Nephrectomy was performed in a 50-kg pig, and the kidney graft was flushed with cold preservation solution and stored on ice. During the cold storage period, 5.9 × 10¹² vg of AAV.k13-scCBh-mCherry was administered for 2 h via the renal artery and 5.9 × 10¹² vg was administered via the ureteral route. Following static cold storage, kidneys were transplanted back into the same pig and kidneys were recovered at 14 days post transplant. Immunohistochemistry of mCherry was performed on biopsies of pig kidney tissue, where positive signal is stained brown. Representative IHC images have proximal tubules, distal tubules and glomeruli labelled. b, Nephrectomy was performed in a 48.9-kg pig and ex vivo machine perfusion was initiated for 2 h. AAV.k20–scCBh-mCherry (6.1 × 10¹² vg) was perfused for 2 h via the arterial route and 6.1 × 10¹² vg was administered retrograde via the ureteral route. Following machine perfusion, kidneys were transplanted back into the same pig, and kidneys were recovered at 9 days post transplant. IHC of mCherry was performed on biopsies of pig kidney tissue, where positive signal is stained brown. Representative IHC images have proximal tubules, distal tubules and glomeruli labelled. c, Nephrectomy was performed in a 42-kg pig and ex vivo machine perfusion was initiated. In this experiment, 7.5 × 10¹² vg of AAV.k20–scCBh-mCherry was administered for 2 h via the ureteral route. Following machine perfusion, kidneys were transplanted into the same pig, and kidneys were recovered at 9 days post transplant. IHC of mCherry was performed on biopsies of pig kidney tissue, where positive signal is stained brown. Representative IHC images have proximal tubules, distal tubules and glomeruli labelled. d, Nephrectomy was performed in pigs weighing between 40 and 50 kg and machine perfusion was initiated for 2 h. A total of 7 × 10¹² vg of AAV9, AAV.k13 and AAV.k20 packaging scCBh-mCherry was administered for 2 h via the ureteral route. Following machine perfusion, kidneys were transplanted into the same pig, and kidneys were recovered at 9 days post transplant. e, Representative IHC images are shown for AAV9, AAV.k13 and AAV.k20 pig kidneys, where positive signal is stained brown. f, Vector genome copy numbers per μg DNA for kidney were calculated by normalizing mCherry copy numbers to the total μg DNA input for qPCR quantification and plotted as log vg μg⁻¹ DNA, where each dot represents a biopsy of the respective capsid (n = 6). g, Assay analysing gene expression of each pig kidney biopsy. mRNA levels for mCherry were normalized to the house keeping gene, Actb, from RT–qPCR quantification and further normalized to AAV9 expression levels. Each dot is representative of a biopsy (n = 6). For boxplots, the centre line represents the median, the box limits are upper and lower quartiles, and whiskers represent 1.5× interquartile range. Statistical significance was determined using Kruskal–Wallis test with post hoc Dunn’s test for pig kidney uptake and mCherry mRNA expression. Individual P values for each comparison are included in each figure panel.

Source data

Next, the transduction profile of the AAV.k20 vector was assessed using the same transplant model but with AAV delivery during ex vivo machine perfusion (Fig. 5b). A similar pattern, albeit with a more robust expression profile, was observed with AAV.k20 vectors using a dual perfusion via both the arterial and ureteral route. Building on these results, the AAV.k20 vector was delivered via ureteral administration alone to assess potential clinical translatability³⁰. Following successful transplantation, IHC analysis reaffirmed robust expression throughout the proximal tubules of the graft as described earlier (Fig. 5c).

Given the success of retrograde ureteral delivery of the AAV.k20 vector during in situ perfusion in rhesus macaques and ex vivo machine perfusion in pig kidneys, a study comparing AAV9 and the AAV.k variants was performed using the pig model and optimized dosing parameters (Fig. 5d). All vectors were administered at the same total dose and route during ex vivo machine perfusion, and kidney grafts were assessed at 9 days following transplantation. IHC analysis for mCherry expression in kidneys perfused with AAV9 vector revealed little to no expression of mCherry in glomeruli and distal tubules. Most importantly, no expression was observed in proximal tubules with the AAV9 vector. In stark contrast with AAV9, AAV.k13 and AAV.k20 vectors demonstrated robust and widespread expression throughout proximal tubules, with some expression in distal tubules (Fig. 5e). To further interrogate the transduction profile of AAV9, AAV.k13 and AAV.k20, 6 biopsies were taken from each pig kidney for biodistribution and mCherry RNA expression analysis (Supplementary Fig. 5a). No significant mCherry immunostaining in untreated pig kidneys was observed (Supplementary Fig. 5b). Further, no significant differences in the levels of vector genomes were observed between AAV9, AAV.k13 and AAV.k20 biopsies (Fig. 5f and Supplementary Fig. 5c). In contrast, mCherry RNA levels normalized to AAV9 expression revealed significantly higher levels of mCherry RNA expression for AAV.k13 and AAV.k20 compared with AAV9 (Fig. 5g and Supplementary Fig. 5d). These results further support the translatability of AAV.k vectors across different preclinical kidney models, namely, murine, human organoid, porcine and NHP kidneys, but also highlight the potential translatability of retrograde ureteral (ex vivo or in situ) delivery of AAV.k vectors in the clinic for the treatment of kidney diseases, as well as organ transplant applications.

Discussion

We have described notably augmented AAV transduction efficiency in the kidney across multiple model systems using a cross-species evolution approach. In particular, two capsid variants, AAV.k13 and AAV.k20, show promise for preclinical development based on improved gene-transfer efficiency in mice following i.v. administration, and in human kidney organoids. Furthermore, we demonstrate that retrograde ureteral delivery of AAV.k13/k20 variants in NHP and pig kidney models is a promising approach for clinical translation with the ability to achieve robust kidney gene transfer at nominal AAV doses. Such an approach may also limit systemic exposure to high viral vector load, which may cause undesirable side effects³¹. These findings have noteworthy implications for therapeutic kidney gene transfer applications as well as the genetic manipulation of kidneys for organ transplant applications.

A key element of our approach beyond cross-species evolution¹⁴ is the multifaceted approach ranging from kidney organoids to different routes of administration (that is, intravenous vs arterial vs retrograde ureteral). In particular, we believe consideration of the structural, anatomical and physiological differences in kidneys across species played a key role in selective pressure^32,33. While mouse kidneys are unipapillary, pig kidneys are multipapillary and share similar cortical and medullary structures with human kidneys^34,35. Therefore, we performed high-throughput sequencing on variants enriched separately from the pig cortex and medulla. Only ~2% of the sequences (including our lead variants AAV.k13 and AAV.k20) were found to be enriched in both anatomical regions, with nearly half of the remaining unique sequences being recovered individually from the cortex and medulla. This observation may imply differential preferences in the regional biodistribution of different AAV variants within the kidney. Notably, when we utilized ex vivo machine perfusion techniques with NHP kidneys in our evolution approach, NGS data revealed that nearly a third of the enriched sequences (including AAV.k13 and AAV.k20) overlapped between arterial and ureteral routes, with the remaining unique sequences split approximately in half. These results may indicate that AAV capsid accessibility to proximal tubule epithelial cells was not particularly limited by the route of administration. Further analysis of region-specific and route-specific capsid variants in NHP and/or pig kidneys is likely to yield additional candidates enabling different preclinical development paths.

Comparison of AAV.k13, AAV.k20 and AAV9 following intravenous administration in mice did not reveal any significant differences in biodistribution. However, higher transgene expression was observed in kidneys with AAV.k variants, specifically within proximal tubules. This observation was translatable across human kidney organoids and transplanted pig kidneys. This preferential transduction profile can probably be explained by the fact that proximal tubules constitute a major portion of the kidney, with proximal tubule epithelial cells being a highly abundant cell population³⁶. From an anatomical perspective, the renal artery branches into the afferent and efferent arterioles, which form the vascular network that encapsulates proximal tubules and the rest of the nephron^32,33. Due to glomerular filtration, large macromolecules that do not enter the Bowman’s capsule are typically returned to the bloodstream through the efferent arterioles, specifically the peritubular capillaries. Albeit unclear at this writing, this phenomenon may offer a potential uptake pathway involving transcytotic uptake from capillaries into the basolateral membrane of proximal tubule epithelial cells. Such is corroborated in part by immunoco-localization staining of the S1 and S2 segments of the proximal tubules with mCherry in mouse kidney tissue.

Additional studies focused on dissecting the mechanistic underpinnings of AAV.k variant transduction in kidneys could involve isolation of different cell types within the kidney as well as tracking of viral capsids and genomes. It is noteworthy to mention that expanded cellular tropism using AAV.k vectors may be achievable by driving transgene expression using specific promoters for other kidney cell types and/or mining for additional enriched capsids as outlined earlier. Nevertheless, the evolved properties of AAV.k variants are distinct from those of parental AAV9, which does not appear to appreciably transduce proximal tubule epithelia. In addition, the propensity to transduce proximal tubule epithelia (or lack thereof) is contrasted by the transduction profile of AAV.k variants vs AAV9 in human organoids. While previous studies have reported no detectable luciferase expression in kidneys following systemic administration of AAV9, we observed a low level of expression³⁷. This discrepancy may be attributed to differences in experimental conditions, such as the mouse strains used (Balb/C vs C57BL/6)³⁸ and the vector dose administered (1 × 10¹¹ vg vs 1 × 10¹² vg). Nevertheless, our results demonstrate that AAV.k variants are more effective than AAV9 for kidney gene transfer, and future efforts to understand the kidney cell entry and post-entry mechanism(s) of AAV.k capsids could help further optimize transduction efficiency.

Outlook

A particularly exciting attribute of our study is the ability of AAV.k variants (notably, AAV.k20) to transduce the pig kidney when administered via the ureter. Therefore, AAV.k variants may enable transgene expression (secreted or cell-surface localized) in the kidney with noteworthy implications for renal transplantation. When combined with machine perfusion of donor organs, which can maintain the kidney in a functioning state, AAV.k vectors could enable expression of immunomodulatory agents before transplantation. This approach has great potential value with regard to evaluation of strategies to mitigate the risks of transplant rejection. Further, widespread expression was also observed in different species, specifically within the proximal tubules for AAV.k variants. These studies support a potential path for preclinical development in kidney disease models, with notable examples of renal diseases involving proximal tubules being polycystic kidney disease, cystinuria and cystinosis, among others⁷.

Methods

Study design

All experiments were approved by the Institutional Animal Care and Use Committee (IACUC) at Duke University. Our approach involved cycling an AAV9 VR-IV library intravenously in mice and pig, on human kidney organoids, and ex vivo machine perfusion in NHP kidney via arterial and ureteral routes. All mouse, pig and NHP protocols were approved by the IACUC at Duke University (mouse, Protocol A189-21-09; pigs, Protocol A140-21-06; NHP, Protocol A118-22-06). During the study, the care and use of animals was conducted in accordance with the guidelines set by Duke IACUC.

AAV capsid libraries and recombinant vector production

The AAV9 VR-IV plasmid library was designed, constructed and produced in-house as previously described^14,39. Briefly, using pAAV9-null (Addgene, 200182) as a template, the first PCR was performed with primer set #1 that amplifies ~600 bp directly downstream from the end of the library site consisting of the 21-nucleotide library NNS (7-mer library insert), where N is any nucleotide and S is a guanine or cytosine, and the XbaI restriction site. The second PCR was performed with primer set #2 that amplifies ~700 bp directly upstream from the start of the library site consisting of the BsiWI restriction site. Finally, both amplicons served as primers and template by amplifying from the homology region of the two amplicons. The final PCR amplicon contained the library insert. This region containing the library can then easily be cloned into pITR-AAV-Rep2Cap9-GFP (Addgene, 199744) using BsiWI and XbaI restriction sites. Recombinant AAV vectors packaging different genome cassettes were produced by triple plasmid transfection in HEK293 cells, purified and analysed as previously described^14,39.

Cycling of AAV capsid libraries in kidney model systems

AAV VR-IV libraries were produced as described in the section above. The evolution for track 1 was performed in human kidney organoids, where 1 × 10¹¹ vg of parental capsid library was added to a well. Genomic DNA was isolated at 24 h post transduction, and the AAV VR-IV library region was amplified from organoid genomic DNA using 94F-B and 98R-X primers targeting AAV9 Cap (Supplementary Table 1). PCR amplicons were digested and ligated into our AAV library plasmid backbone and used to generate the next library. Schematics were created in part using BioRender.com.

The evolution for track 2 was performed in mice following intravenous administration at a dose of 5 × 10¹³ vg kg⁻¹. The kidney was collected at 3 days post injection, where half of the sagittal dissection was used for genomic DNA isolation. AAV genomes were amplified from mouse genomic DNA. PCR amplicons were digested and ligated into our AAV library plasmid backbone and used to generate the subsequent AAV9 VR-IV library.

The evolution for track 3 was performed on human kidney organoids, where 1 × 10¹¹ vg of the capsid library evolved from track 2 in mice was added to a well. Genomic DNA was isolated at 24 h post transduction and the AAV VR-IV library region was amplified from organoid genomic DNA. PCR amplicons were digested and ligated into our AAV library plasmid backbone and used to generate the next library.

The evolution for track 4 occurred in pigs following intravenous administration. Briefly, Yorkshire pigs (n = 2) were used for AAV library cycling and evolution. Studies were performed in 3-week-old newly weaned piglets weighing ~7 kg. Male and female piglets were injected systemically via the heart vena cava with 1 × 10¹³ vg kg⁻¹ of the AAV VR-IV library. Kidneys were collected at 3 days post injection and prepared for library amplification. Pig kidneys from different animals were collected at 3 days post injection and dissected into multiple cortex and medulla regions before genomic DNA isolation. The AAV VR-IV library region was amplified and ligated into our AAV library plasmid backbone and used to produce the next round of AAV VR-IV library.

The evolution for track 5 occurred on differentiated human kidney organoids, where the pig cortex and medulla AAV capsid libraries were pooled at a 1:1 ratio and used to transduce organoids. Genomic DNA was isolated at 24 h post transduction, and the AAV VR-IV library region was amplified and used to generate the next round of AAV capsid library.

The evolution for track 6 was performed using an ex vivo NHP kidney machine perfusion system. Briefly, kidneys from rhesus macaques weighing 4–8 kg were perfused using the method described above. AAV capsid library generated from evolving on human kidney organoids was perfused for 4 h via the arterial route and ureteral route of two different NHP kidneys. NHP kidneys were flushed and dissected before extracting genomic DNA. All kidneys were perfused with PBS to remove AAV variants that may be surface bound but not internalized by kidney cell types. Additional details for ex vivo kidney perfusion of AAV vectors in pigs and NHPs are outlined below. The final evolved library was prepared as previously described. Next-generation sequencing was performed on libraries to track the progress of the evolution.

High-throughput sequencing analysis and identification of newly evolved AAV.k variants

AAV capsid library vectors were produced from AAV library plasmids that were generated after every evolution track. Evolved viral libraries were DNAse-I treated to extract viral genomes from capsids, and Illumina adapters were added via PCR. For all but the NHP, the first PCR added Illumina adapters using primers specific to the amplicon, while the second PCR added the Illumina indexing adapters. For NHP sequencing, the first PCR with Illumina adapters was done followed by amplicon EZ sequencing (2 × 250 bp configuration) (Genewiz). Libraries were then prepared for sequencing with the Illumina NovaSeq 6000 S-Prime Reagent kit (2 × 150 bp configuration) and sequenced on the Illumina NovaSeq system. PCR products were purified using the PureLink PCR Micro kit (Invitrogen) and concentrations were quantified using a Qubit spectrometer (Thermo Fisher). NGS reads were analysed with an in-house Python script that counts and ranks nucleotide sequences of the library regions. Briefly, for data analysis, the 5’AAGACTATT and 3’CAAACGCTA barcodes were used to identify sequences of interest in our AAV9 VR-IV NGS data. Nucleotides were converted into amino acids and counts for each amino acid sequence were recorded. Fold change was calculated by division of evolution round by the parental production run as a percent abundance of species identified in both. The amino acid sequences were then ranked. For species that were not observed in the parental but were observed in the evolution round at greater than 100 counts, a pseudo-threshold was set as if the sequence had been observed once in the parental round.

iPSC-derived 3D human kidney organoid differentiation and processing

A human EE0000002 DM1-iPSC cell line available through the National Institute of Neurological Disorders and Stroke (NINDS) Human Cell and Data Repository (NHCDR) was used in combination with an established protocol to generate differentiated human kidney organoids⁴⁰. At Day 14, kidney organoids were evenly distributed in a 6-well plate (~10 organoids per well) and prepared for transduction assays. Kidney organoids were transduced at Day 14 with either 1 × 10¹¹ vg or 1 × 10¹² vg of AAV vector packaging a self-complementary CBh–mCherry cassette or AAV vector packaging a single-stranded CBA–luciferase cassette by adding directly into the well. Media were replenished with fresh Stage II media at 24 h post transduction. Kidney organoids were collected at 5 days post transduction and placed in either 4% paraformaldehyde for 24 h, RNAlater (Invitrogen) or frozen for luciferase assays. Organoids placed in 4% paraformaldehyde were rinsed 3× with PBS and subsequently placed in 30% sucrose on a tube rotator at 4 °C for 24 h. Organoids were then embedded in Tissue-Tek O.C.T Compound (Sakura) moulds and carefully dipped in liquid nitrogen-chilled 2-methyl butane until frozen blocks were formed. Frozen organoid blocks were sectioned at 8 μm on a Leica CM1520 cryostat, mounted on Superfrost Plus microscope slides (Fisherbrand) and stored at −80 °C. DNA and RNA were extracted from organoids kept in RNAlater using the DNA/RNA/Protein Extraction kit (IBI) and kept at −20 °C and −80 °C, respectively.

iPSC-derived 3D human kidney organoid immunostaining

Cryosectioned organoids were allowed to come to room temperature for 10 min and rinsed once in PBS. Organoids were then incubated in 2% BSA + 0.5% Triton X-100 in PBS for 1 h at room temperature. Organoids were then quickly rinsed in PBS, incubated overnight at 4 °C with primary antibodies in PBS containing 2% BSA + 0.5% Triton X-100, washed 3× in PBS and incubated with secondary antibodies in PBS containing 2% BSA + 0.5% Triton X-100 at room temperature for 1 h. Subsequently, organoids were washed 3× in PBS and mounted using ProLong Gold antifade mountant with DNA stain DAPI (Invitrogen). Slides were then imaged using an Echo Revolve microscope or a Zeiss 880 Airyscan Fast Inverted Confocal microscope at the Duke University Light Microscopy Core Facility. Primary antibody used was RFP (1:200; rabbit, Rockland, 600-401-379). Secondary antibody used was anti-rabbit Alexa Fluor 594 (1:500; Invitrogen, A-11012) and LTL-FL (1:200; Vector Laboratories, FL-1321-2).

Renal proximal tubule epithelial cell (TH1) luciferase assay

TH1 cells, immortalized renal proximal tubule epithelial cells (RPTECs), were purchased from Kerafast and maintained in DMEM-high glucose, 10% FBS and 1% penicillin-streptomycin. For the luciferase assay, TH1 cells were seeded at 1 × 10⁴ cells per well in a tissue-treated 96-well white/clear-bottom plate and transduced with corresponding AAVs packaging a single-stranded luciferase cassette driven by a CBA promoter at MOIs of 5,000, 10,000 and 50,000. Ar 5 days post transduction, cells were lysed in the well with 50 μl 2× Passive Lysis Buffer (Promega) with 1× Halt Protease Inhibitor Cocktail (ThermoFisher). To measure luciferase activity, 50 μl of luciferin was added to each well. Luminescence was measured using a Varioskan LUX Multimode microplate reader (ThermoFisher) following manufacturer specifications. Data were graphed and statistical analysis was performed.

Determination of vector genome uptake in organoids by quantitative PCR

After extracting DNA, vector genomes were quantified via qPCR using a self-complementary CBh–mCherry plasmid standard and primers targeting an mCherry amplicon (Supplementary Table 1). The uptake of viral genomes was represented as the ratio of vector genomes per microgram of DNA extracted for both 1 × 10¹¹ vg per well and 1 × 10¹² vg per well conditions. Similarly, for the luciferase study, vector genomes were quantified via qPCR, using a single-stranded CBA–luciferase plasmid standard and primers targeting a luciferase amplicon (Supplementary Table 1). Quantitative PCR reactions were carried out using a Roche Light-Cycler 480 and SYBR Green I Master Mix (Roche Applied Sciences).

Intravenous administration in wild-type C57/B6 mice

For all mouse studies, 8–10-week-old male and female adult mice were systemically administered with AAV via the tail vein. The wild-type C57/B6 mouse colony was bred and maintained at Duke University. For the AAV capsid library evolution in C57/B6 mice (n = 2), AAV was injected systemically at 2 × 10¹³ vg kg⁻¹. Kidneys were collected at 3 days post injection and prepared for library amplification. For the mCherry (n = 3) and luciferase reporter studies (n = 4), C57/B6 mice were injected systemically via tail vein injection at a dose of 5 × 10¹³ vg kg⁻¹ AAV. Tissues were collected at 4 weeks post injection for all gene transfer studies in C57/B6 mice. For the PD-L1 studies (n = 4), C57/B6 mice were injected systemically via tail vein injection at a dose of 5 × 10¹³ vg kg⁻¹ AAV. Kidneys were collected at 4 weeks post injection for both studies.

Western blot analysis for PD-L1 expression in mice

Biopsies of mouse kidneys from AAV.k20–scCBh–PD-L1–HA treated mice and mock mouse were lysed by adding 200 μl of 1× RIPA buffer (ThermoFisher) before mechanical lysis using a Fastprep-24 tissue homogenizer (MP Biomedicals). Lysate was cleared by centrifugation (21,100 × g, 3 min, 4 °C) to remove remaining debris. Samples were incubated at 95 °C for 5 min in LDS loading buffer supplemented with 2.5% β-mercaptoethanol, separated by SDS–PAGE and transferred to PVDF membranes. Membranes were blocked with 5% milk powder in Tris buffered saline with −0.1% Tween 20 (TBST) and incubated for 1 h at room temperature. Membranes were subsequently incubated with primary antibodies for vinculin (E1E9V) XP (1:5,000; Cell Signaling, 13901) or anti-HA-Tag (1:1,000; Invitrogen, PA1-985) in 5% milk in TBST overnight at 4 °C. The following day, membranes were washed 3× for 5 min with TBST. Membranes were then incubated with goat anti-rabbit IgG and human anti-HRP (1:5,000; Southern Biotech, 4010) in 5% milk in TBST for 1 h at room temperature. Membranes were washed 3× for 5 min with TBST and incubated for 1 min with enhanced chemiluminescence substrate (ThermoFisher, 34049). Immunoblots were visualized on a Bio-Rad ChemiDoc imaging system.

Immunohistology of mouse tissues

Isolated C57/B6 mouse hearts and livers were post fixed in 10% formalin (VWR) overnight. Hearts and livers were washed 3× in 1× PBS, embedded in 3% agarose and sectioned to 50 μm thickness using a vibratome (Leica Biosystems). For mCherry evaluation, tissue was mounted on slides with ProLong Gold antifade mountant with DAPI (ThermoFisher) and imaged for native fluorescence. Livers were taken under high exposure and low exposure across all AAV capsids. Frozen sections were allowed to come to room temperature for a few seconds, then fixed in 4% paraformaldehyde for 15 min. After washing with PBS, sections were blocked for 1 h in Power Block (Biogenex Laboratories, HK0855K) with 10% donkey serum, then incubated overnight at 4 °C with primary antibodies in PBS containing 5% donkey serum + 2.5% BSA + 0.05% Tween 20. After washing, sections were then incubated with secondary antibodies and 0.25 μg ml⁻¹ DAPI for 1 h at room temperature in the same antibody buffer. After washing the sections with PBS, coverslips were mounted in ProLong Gold without DAPI (Invitrogen, P36934). After curing, slides were imaged using an inverted fluorescence microscope. Primary antibodies used were SGLT1 (1:1,000; rabbit, Izumi Kaji lab, 576-610), SGLT2 (1:100; mouse, Santa Cruz, sc-393350) and mCherry (1:500; rat, Invitrogen, M11217). Secondary antibodies used at 1:500 were donkey anti-rabbit 488 (Invitrogen, A21206), donkey anti-rat 594 (Invitrogen, A21209) and donkey anti-mouse 647 (Invitrogen, A31571).

All C57/B6 mouse kidneys were post fixed in 10% formalin overnight and washed 3× in 1× PBS before sucrose treatment. Collected kidney organoids and post-fixed mouse kidneys were incubated in 30% sucrose on a tube rotator for 48 h at 4 °C. Specimens were then embedded in Tissue-Tek O.C.T Compound (Sakura), snap frozen in liquid nitrogen-chilled 2-methylbutane, cryosectioned on a Leica CM1520 cryostat, mounted on Superfrost Plus microscope slides (Fisherbrand) and stored at −80 °C. Cryosectioned mouse kidneys were allowed to come to room temperature for 15 min and rinsed once in PBS. Mouse kidneys were then incubated in blocking buffer (5% normal goat serum, 0.1% Triton X-100 in 1× PBS) for 1 h at room temperature. Mouse kidneys were quickly rinsed 3× in PBS and incubated overnight at 4 °C with primary antibodies diluted in blocking buffer. Mouse kidney tissues were then rinsed 3× in PBS and incubated with secondary antibodies diluted in blocking buffer at room temperature for 1 h. Sections were washed 3× in PBS, followed by treatment with Vector TrueView Autofluorescence Quenching kit (Vector Laboratories) and quick washing in PBS, and mounted using ProLong Gold antifade mountant with DNA Stain DAPI (Invitrogen). Slides were then imaged using an Echo Revolve microscope or a Zeiss 880 Airyscan Fast Inverted Confocal microscope at the Duke University Light Microscopy Core Facility. Primary antibodies used were RFP (1:200; rabbit, Rockland, 600-401-379) and mouse nephrin (1:200; goat, R&D Systems, AF3159). Secondary antibodies used were anti-rabbit Alexa Fluor 594 (1:500; Invitrogen, A-11012), anti-goat Alex Fluor 488 (1:500; Invitrogen, A-11055), DBA-FL (1:200; Vector Laboratories, FL-1031) and LTL-FL (1:200; Vector Laboratories, FL-1321-2).

To quantify total fluorescence intensity for our mCherry studies in mice, the following equation was used: total fluorescence intensity = integrated density – (area of tissue region × mean fluorescence of background readings). Total fluorescence intensity of mCherry was then normalized to LTL intensity. All mouse staining quantification was performed using at least two sections and taking three images of each section for every mouse (n = 3). Similarly, total fluorescence intensity for our mCherry studies in human kidney organoids was performed using the same approach. For organoids, mCherry expression was normalized to DAPI.

Ex vivo kidney perfusion of AAV vectors in pigs and non-human primates

Machine perfusion of porcine and non-human primate kidneys was conducted using an automated and portable perfusion system that was specifically developed by BioMed Innovations, Organ Bank. The organ was perfused with a solution made of human albumin, bicarbonate-based dialysate (B. Braun Medical), calcium gluconate, heparin, multivitamins, dexamethasone and piperacillin/tazobactam at a temperature between 22 and 25 °C. An additional nutritional supplement called Clinimix (Baxter International) and regular insulin were continuously infused into the organ along with verapamil. The perfusate was oxygenated using a mixture of carbogen, which consists of 95% oxygen (O₂) and 5% carbon dioxide (CO₂), at a flow rate of 2–3 l min⁻¹. The renal artery was cannulated and linked to the device; however, the renal vein was intentionally left open to facilitate drainage. The ureter was cannulated similarly. A pressure-controlled pump was used to progressively elevate the mean arterial pressure to 70 mmHg after a 30-min warming-up time. Different AAV vectors were delivered either through the ureter route or the ureter and renal artery route; subsequently, the ureter was clamped to obstruct the outflow of urine.

For transplant studies, nephrectomy was performed in pigs and non-human primates, and ex vivo machine perfusion was initiated using the conditions described above. During machine perfusion, AAV vectors packaging a self-complementary CBh–mCherry cassette were administered via the arterial route or ureteral route. Following the machine perfusion period, kidney transplantation was performed by autotransplantation back into the same pig or non-human primate. Kidneys were collected at 9–14 days post transplantation for pig studies and 30 days post transplantation for the non-human primate study.

In situ delivery of AAV vectors in non-human primate kidneys

AAV was delivered to the kidneys of non-human primates in situ via the ureter following midline laparotomy. Both kidneys received AAV packaging a single-stranded cassette encoding for luciferase driven by a CBA promoter, employing a standard protocol. Initially, the distal part of the ureter was isolated and clamped, followed by cannulation with a 22G angiocatheter. Subsequently, the renal artery was isolated and heparin was administered systemically to ensure adequate anti-coagulation throughout the procedure. The renal artery was then temporarily clamped to arrest blood flow. Following AAV administration through the ureter, a 15-min waiting period ensued for optimal distribution and uptake of the viral vector within the kidney tissue before unclamping both renal artery and ureter to restore blood flow and urinary drainage.

Histology and immunohistochemistry staining of pig and non-human primate kidney tissues for mCherry

Biopsies from collected kidneys were fixed in 10% formalin and embedded in paraffin. Immunohistochemical staining was performed using chicken anti-mCherry (1:1,000; Novus Biologics, NBP2-25158). Immunohistochemistry was performed using a horseradish peroxidase conjugated anti-chicken secondary antibody, with diaminobenzidine as the chromogenic substrate. Whole-slide digital images were taken using the Aperio AT Turbo digital slide scanner system (Leica Biosystems) and viewed using the Imagescope (Leica Biosytems) digital pathology software.

Determination of AAV vector transcripts in organoids, pig kidney biopsies and non-human primate kidney biopsies by quantitative RT–PCR

Extracted RNA from organoids, pig kidney biopsies and non-human primate kidney biopsies were subjected to DNAse treatment using TURBO DNA-free kit (Invitrogen). Equal amounts of DNAse-treated RNA were used for complementary DNA synthesis using the High-Capacity RNA-to-cDNA kit (Applied Biosystems). Newly synthesized cDNA was used for qPCR using primers specific to mCherry and human GAPDH (pig Actb and NHP GAPDH, respectively) (Supplementary Table 1). Quantitative RT–PCR reactions were carried out using a Roche Light-Cycler 480 and SYBR Green I Master Mix (Roche Applied Sciences).

Determination of vector genome biodistribution in multiple species by qPCR

After extracting DNA from mouse tissues, pig kidney biopsies and non-human primate kidney biopsies, vector genomes were quantified via qPCR using a self-complementary CBh–mCherry plasmid standard and primers targeting an mCherry amplicon (Supplementary Table 1). The biodistribution of viral genomes is represented as the ratio of vector genomes per microgram of DNA extracted. Similarly, for the luciferase study, vector genomes were quantified via qPCR using a single-stranded CBA–luciferase plasmid standard and primers targeting a luciferase amplicon (Supplementary Table 1). Quantitative PCR reactions were carried out using a Roche Light-Cycler 480 and SYBR Green I Master Mix (Roche Applied Sciences).

Determination of luciferase activity in mouse tissues, organoids and non-human primate kidney biopsies

Organoids were homogenized with 100 μl of 2× Passive Lysis Buffer (Promega) with 1× Halt Protease Inhibitor Cocktail (ThermoFisher) by mechanical lysis via repetitive pipetting. Homogenized organoids were then centrifuged at 8,000 × g for 1 min to remove any debris, and supernatant was used for luciferase expression assay. To measure luciferase activity, 20 μl of supernatant from each organoid sample was loaded onto an assay plate along with 50 μl of luciferin. The relative light units obtained for each organoid sample was then normalized to the input tissue weight, measured in grams. Luminescence was measured using a Varioskan LUX Multimode microplate reader (ThermoFisher) following manufacturer specifications. Data were graphed and statistical analysis was performed.

To quantify luciferase expression, mouse tissue biopsies were thawed and ~20 mg of tissue were biopsied, with exact weights recorded. Tissues were lysed by adding 200 μl of 2× Passive Lysis Buffer (Promega) with 1× Halt Protease Inhibitor Cocktail (ThermoFisher) before mechanical lysis using an MP BIOMEDICAL benchtop homogenizer. Homogenized tissues were then centrifuged at 8,000 × g for 2 min to remove any debris, and supernatant was used for luciferase expression assay. To measure luciferase activity, 50 μl of supernatant from each tissue lysate was loaded onto an assay plate along with 50 μl of luciferin. The relative light units obtained for each sample were then normalized to the input tissue weight, measured in grams. Luminescence was measured using a Varioskan LUX Multimode microplate reader (ThermoFisher) following manufacturer specifications. Data were graphed and statistical analysis was performed. Similarly, for NHP kidney biopsies, exact weights of the biopsies were recorded. Tissue lysing and processing was performed exactly as described above.

Statistical analysis

GraphPad Prism software (v.10.0.2) was used for statistical analysis. All values are presented as mean ± s.d. For data sets with four groups, significance was determined using one-way analysis of variance (ANOVA) with Tukey’s post test, unless otherwise noted. Where indicated, *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001; NS is not significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information^{(2.4MB, pdf)}

Supplementary Figures and Tables.

Reporting Summary^{(86KB, pdf)}

Peer Review File^{(1.5MB, pdf)}

Source data

Source Data Figs. 2–5, Extended Data Figs. 1–2 and Supplementary Figs. 3–5^{(59.5KB, xlsx)}

Source data and statistics.

Source Data for Fig. 4c^{(258.5KB, pdf)}

Unprocessed western blots for vinculin and PDL1-HA

Acknowledgements

This study was funded by NIH grants awarded to A.A. (R01HL089221, UH3AR075336, evolution tracks 2 and 4 were funded in part by BridgeBio Inc.), A.A. and A.S.B. (U01AI170064), M.H.W. (R01EB033676, I01BX004258) and J.L.P. (F30DK134046). We thank members of the Asokan, Barbas and Wilson labs for feedback and support, and the Duke Light Microscopy Core Facility for technical assistance and support.

Extended data

Author contributions

A.R. and A.A. conceived the study. A.R. and A.A. designed the experiments with assistance from Q.G. and A.S.B. for ex vivo experiments, and J.L.P. and M.H.W. for mouse experiments. A.R. and L.O.B. carried out mouse and organoid experiments, and A.V. and S.F. assisted with pig and organoid experiments. A.R. and J.H. analysed next-generation sequencing data. Q.G. and N.A. performed ex vivo NHP evolution and ex vivo pig experiments. M.S. performed pig kidney IHC. A.R. and A.A. prepared the paper, with input from J.L.P., M.H.W., N.A. and A.S.B.

Peer review

Peer review information

Nature Biomedical Engineering thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available.

Data availability

The NGS datasets for the capsid libraries are available via the Sequence Read Archive accession code SUB14697692. The raw and analysed datasets generated during the study are available for research purposes from the corresponding author on reasonable request. Source data for the figures are provided with this paper.

Code availability

The scripts for AAV-capsid analysis have been deposited on Zenodo at 10.5281/zenodo.13620030 (ref. ⁴¹).

Competing interests

A.R., Q.G, A.S.B. and A.A. have filed a patent application (WO2024206226A1, Durham, NC, USA, 03/25/2024) on the subject matter of this manuscript. A.A. is a co-founder at TorqueBio and serves as an advisor to Atsena Therapeutics, Nvelop Therapeutics, Mammoth Bio, Ring Therapeutics and Ginkgo Bioworks. M.H.W. has served as a consultant for TorqueBio and serves as an advisor to SalioGen Therapeutics. A.S.B. serves on the medical advisory board for BioMedInnovations. The other authors declare no completing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

is available for this paper at 10.1038/s41551-024-01341-0.

Supplementary information

The online version contains supplementary material available at 10.1038/s41551-024-01341-0.

References

1.Hill, N. R. et al. Global prevalence of chronic kidney disease – a systematic review and meta-analysis. PLoS ONE11, e0158765 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Evans, M. et al. A narrative review of chronic kidney disease in clinical practice: current challenges and future perspectives. Adv. Ther.39, 33–43 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Schrezenmeier, E. et al. The underestimated burden of monogenic kidney disease in adults waitlisted for kidney transplantation. Genet. Med.23, 1219–1224 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Tonelli, M. et al. Systematic review: kidney transplantation compared with dialysis in clinically relevant outcomes. Am. J. Transplant.11, 2093–2109 (2011). [DOI] [PubMed] [Google Scholar]
5.Malek, S. K., Keys, B. J., Kumar, S., Milford, E. & Tullius, S. G. Racial and ethnic disparities in kidney transplantation. Transpl. Int.24, 419–424 (2011). [DOI] [PubMed] [Google Scholar]
6.Keith, D. S. & Vranic, G. M. Approach to the highly sensitized kidney transplant candidate. Clin. J. Am. Soc. Nephrol.11, 684–693 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Peek, J. L. & Wilson, M. H. Cell and gene therapy for kidney disease. Nat. Rev. Nephrol.19, 451–462 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Rubin, J. D. & Barry, M. A. Improving molecular therapy in the kidney. Mol. Diagn. Ther.24, 375–396 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Peek, J. L. & Wilson, M. H. Gene therapy for kidney disease: targeting cystinuria. Curr. Opin. Nephrol. Hypertens.31, 175–179 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Hildebrandt, F. Genetic kidney diseases. Lancet375, 1287–1295 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Pupo, A. et al. AAV vectors: the Rubik’s cube of human gene therapy. Mol. Ther.30, 3515–3541 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Wang, D., Tai, P. W. L. & Gao, G. Adeno-associated virus vector as a platform for gene therapy delivery. Nat. Rev. Drug Discov.18, 358–378 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Rubin, J. D., Nguyen, T. V., Allen, K. L., Ayasoufi, K. & Barry, M. A. Comparison of gene delivery to the kidney by adenovirus, adeno-associated virus, and lentiviral vectors after intravenous and direct kidney injections. Hum. Gene Ther.30, 1559–1571 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Gonzalez, T. J. et al. Cross-species evolution of a highly potent AAV variant for therapeutic gene transfer and genome editing. Nat. Commun.13, 5947 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Adachi, K., Enoki, T., Kawano, Y., Veraz, M. & Nakai, H. Drawing a high-resolution functional map of adeno-associated virus capsid by massively parallel sequencing. Nat. Commun.5, 3075 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Emmanuel, S. N. et al. Structurally mapping antigenic epitopes of adeno-associated virus 9: development of antibody escape variants. J. Virol.96, e0125121 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Tse, L. V. et al. Structure-guided evolution of antigenically distinct adeno-associated virus variants for immune evasion. Proc. Natl Acad. Sci. USA114, E4812–E4821 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Havlik, L. P. et al. Coevolution of adeno-associated virus capsid antigenicity and tropism through a structure-guided approach. J. Virol.94, 976–996 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Chan, K. Y. et al. Engineered AAVs for efficient noninvasive gene delivery to the central and peripheral nervous systems. Nat. Neurosci.20, 1172–1179 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Shi, G. & Jin, Y. Role of Oct4 in maintaining and regaining stem cell pluripotency. Stem Cell Res. Ther.1, 39 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Saha, A. et al. Programmed death ligand-1 expression on donor T cells drives graft-versus-host disease lethality. J. Clin. Invest.126, 2642–2660 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Zulpaite, R. et al. Ex-vivo kidney machine perfusion: therapeutic potential. Front. Med.8, 808719 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Yuzefovych, Y. et al. Genetic engineering of the kidney to permanently silence MHC transcripts during ex vivo organ perfusion. Front. Immunol.11, 499419 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Samoylova, M. L., Nash, A., Kuchibhatla, M., Barbas, A. S. & Brennan, T. V. Machine perfusion of donor kidneys may reduce graft rejection. Clin. Transplant.33, e13716 (2019). [DOI] [PubMed] [Google Scholar]
25.Moers, C. et al. Machine perfusion or cold storage in deceased-donor kidney transplantation. N. Engl. J. Med.360, 1460–1461 (2009). [PubMed] [Google Scholar]
26.Locke, J. E., Kumar, V., Anderson, D. & Porrett, P. M. Normal graft function after pig-to-human kidney xenotransplant. JAMA Surg.10.1001/jamasurg.2023.2774 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Gutierrez, K., Dicks, N., Glanzner, W. G., Agellon, L. B. & Bordignon, V. Efficacy of the porcine species in biomedical research. Front. Genet.6, 159179 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Koslowski, S., Latapy, C., Auvray, P., Blondel, M. & Meijer, L. An overview of in vivo and in vitro models for autosomal dominant polycystic kidney disease: a journey from 3D-cysts to mini-pigs. Int. J. Mol. Sci.21, 4537 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Lian, X. et al. The combination of metformin and 2-deoxyglucose significantly inhibits cyst formation in miniature pigs with polycystic kidney disease. Br. J. Pharmacol.176, 711–724 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Chung, D. C. et al. Adeno-associated virus-mediated gene transfer to renal tubule cells via a retrograde ureteral approach. Nephron Extra1, 217–223 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Au, H. K. E., Isalan, M. & Mielcarek, M. Gene therapy advances: a meta-analysis of AAV usage in clinical settings. Front. Med.8, 809118 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Dalal, R., Bruss, Z. S. & Sehdev, J. S. Physiology, Renal Blood Flow and Filtration (StatPearls, 2023). [PubMed]
33.Daehn, I. S. & Duffield, J. S. The glomerular filtration barrier: a structural target for novel kidney therapies. Nat. Rev. Drug Discov.20, 770–788 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Hansen-Estruch, C., Cooper, D. K. C. & Judd, E. Physiological aspects of pig kidney xenotransplantation and implications for management following transplant. Xenotransplantation29, e12743 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Soriano, R. M., Penfold, D. & Leslie, S. W. Anatomy, Abdomen and Pelvis: Kidneys (StatPearls, 2023). [PubMed]
36.Grynberg, K., Ma, F. Y. & Nikolic-Paterson, D. J. The JNK signaling pathway in renal fibrosis. Front. Physiol.8, 829 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Zincarelli, C., Soltys, S., Rengo, G. & Rabinowitz, J. E. Analysis of AAV serotypes 1–9 mediated gene expression and tropism in mice after systemic injection. Mol. Ther.16, 1073–1080 (2008). [DOI] [PubMed] [Google Scholar]
38.Hordeaux, J. et al. The neurotropic properties of AAV-PHP.B are limited to C57BL/6J mice. Mol. Ther.26, 664–668 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Gonzalez, T. J. et al. Structure-guided AAV capsid evolution strategies for enhanced CNS gene delivery. Nat. Protoc.10.1038/s41596-023-00875-y (2023). [DOI] [PubMed] [Google Scholar]
40.Przepiorski, A. et al. A simplified method for generating kidney organoids from human pluripotent stem cells. J. Vis. Exp. 10.3791/62452 (2021). [DOI] [PMC free article] [PubMed]
41.Hull, J. & Rosales, A. AAV capsid library analysis scripts for ‘Multiscale and cross-species evolution of adeno-associated viruses for kidney gene transfer’. Zenodo10.5281/zenodo.13620030 (2024).

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information^{(2.4MB, pdf)}

Supplementary Figures and Tables.

Reporting Summary^{(86KB, pdf)}

Peer Review File^{(1.5MB, pdf)}

Source Data Figs. 2–5, Extended Data Figs. 1–2 and Supplementary Figs. 3–5^{(59.5KB, xlsx)}

Source data and statistics.

Source Data for Fig. 4c^{(258.5KB, pdf)}

Unprocessed western blots for vinculin and PDL1-HA

Data Availability Statement

The scripts for AAV-capsid analysis have been deposited on Zenodo at 10.5281/zenodo.13620030 (ref. ⁴¹).

[CR1] 1.Hill, N. R. et al. Global prevalence of chronic kidney disease – a systematic review and meta-analysis. PLoS ONE11, e0158765 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR2] 2.Evans, M. et al. A narrative review of chronic kidney disease in clinical practice: current challenges and future perspectives. Adv. Ther.39, 33–43 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR3] 3.Schrezenmeier, E. et al. The underestimated burden of monogenic kidney disease in adults waitlisted for kidney transplantation. Genet. Med.23, 1219–1224 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR4] 4.Tonelli, M. et al. Systematic review: kidney transplantation compared with dialysis in clinically relevant outcomes. Am. J. Transplant.11, 2093–2109 (2011). [DOI] [PubMed] [Google Scholar]

[CR5] 5.Malek, S. K., Keys, B. J., Kumar, S., Milford, E. & Tullius, S. G. Racial and ethnic disparities in kidney transplantation. Transpl. Int.24, 419–424 (2011). [DOI] [PubMed] [Google Scholar]

[CR6] 6.Keith, D. S. & Vranic, G. M. Approach to the highly sensitized kidney transplant candidate. Clin. J. Am. Soc. Nephrol.11, 684–693 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR7] 7.Peek, J. L. & Wilson, M. H. Cell and gene therapy for kidney disease. Nat. Rev. Nephrol.19, 451–462 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Rubin, J. D. & Barry, M. A. Improving molecular therapy in the kidney. Mol. Diagn. Ther.24, 375–396 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR9] 9.Peek, J. L. & Wilson, M. H. Gene therapy for kidney disease: targeting cystinuria. Curr. Opin. Nephrol. Hypertens.31, 175–179 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR10] 10.Hildebrandt, F. Genetic kidney diseases. Lancet375, 1287–1295 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR11] 11.Pupo, A. et al. AAV vectors: the Rubik’s cube of human gene therapy. Mol. Ther.30, 3515–3541 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR12] 12.Wang, D., Tai, P. W. L. & Gao, G. Adeno-associated virus vector as a platform for gene therapy delivery. Nat. Rev. Drug Discov.18, 358–378 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR13] 13.Rubin, J. D., Nguyen, T. V., Allen, K. L., Ayasoufi, K. & Barry, M. A. Comparison of gene delivery to the kidney by adenovirus, adeno-associated virus, and lentiviral vectors after intravenous and direct kidney injections. Hum. Gene Ther.30, 1559–1571 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR14] 14.Gonzalez, T. J. et al. Cross-species evolution of a highly potent AAV variant for therapeutic gene transfer and genome editing. Nat. Commun.13, 5947 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR15] 15.Adachi, K., Enoki, T., Kawano, Y., Veraz, M. & Nakai, H. Drawing a high-resolution functional map of adeno-associated virus capsid by massively parallel sequencing. Nat. Commun.5, 3075 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR16] 16.Emmanuel, S. N. et al. Structurally mapping antigenic epitopes of adeno-associated virus 9: development of antibody escape variants. J. Virol.96, e0125121 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR17] 17.Tse, L. V. et al. Structure-guided evolution of antigenically distinct adeno-associated virus variants for immune evasion. Proc. Natl Acad. Sci. USA114, E4812–E4821 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR18] 18.Havlik, L. P. et al. Coevolution of adeno-associated virus capsid antigenicity and tropism through a structure-guided approach. J. Virol.94, 976–996 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR19] 19.Chan, K. Y. et al. Engineered AAVs for efficient noninvasive gene delivery to the central and peripheral nervous systems. Nat. Neurosci.20, 1172–1179 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR20] 20.Shi, G. & Jin, Y. Role of Oct4 in maintaining and regaining stem cell pluripotency. Stem Cell Res. Ther.1, 39 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR21] 21.Saha, A. et al. Programmed death ligand-1 expression on donor T cells drives graft-versus-host disease lethality. J. Clin. Invest.126, 2642–2660 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Zulpaite, R. et al. Ex-vivo kidney machine perfusion: therapeutic potential. Front. Med.8, 808719 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Yuzefovych, Y. et al. Genetic engineering of the kidney to permanently silence MHC transcripts during ex vivo organ perfusion. Front. Immunol.11, 499419 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Samoylova, M. L., Nash, A., Kuchibhatla, M., Barbas, A. S. & Brennan, T. V. Machine perfusion of donor kidneys may reduce graft rejection. Clin. Transplant.33, e13716 (2019). [DOI] [PubMed] [Google Scholar]

[CR25] 25.Moers, C. et al. Machine perfusion or cold storage in deceased-donor kidney transplantation. N. Engl. J. Med.360, 1460–1461 (2009). [PubMed] [Google Scholar]

[CR26] 26.Locke, J. E., Kumar, V., Anderson, D. & Porrett, P. M. Normal graft function after pig-to-human kidney xenotransplant. JAMA Surg.10.1001/jamasurg.2023.2774 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR27] 27.Gutierrez, K., Dicks, N., Glanzner, W. G., Agellon, L. B. & Bordignon, V. Efficacy of the porcine species in biomedical research. Front. Genet.6, 159179 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR28] 28.Koslowski, S., Latapy, C., Auvray, P., Blondel, M. & Meijer, L. An overview of in vivo and in vitro models for autosomal dominant polycystic kidney disease: a journey from 3D-cysts to mini-pigs. Int. J. Mol. Sci.21, 4537 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR29] 29.Lian, X. et al. The combination of metformin and 2-deoxyglucose significantly inhibits cyst formation in miniature pigs with polycystic kidney disease. Br. J. Pharmacol.176, 711–724 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Chung, D. C. et al. Adeno-associated virus-mediated gene transfer to renal tubule cells via a retrograde ureteral approach. Nephron Extra1, 217–223 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR31] 31.Au, H. K. E., Isalan, M. & Mielcarek, M. Gene therapy advances: a meta-analysis of AAV usage in clinical settings. Front. Med.8, 809118 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Dalal, R., Bruss, Z. S. & Sehdev, J. S. Physiology, Renal Blood Flow and Filtration (StatPearls, 2023). [PubMed]

[CR33] 33.Daehn, I. S. & Duffield, J. S. The glomerular filtration barrier: a structural target for novel kidney therapies. Nat. Rev. Drug Discov.20, 770–788 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Hansen-Estruch, C., Cooper, D. K. C. & Judd, E. Physiological aspects of pig kidney xenotransplantation and implications for management following transplant. Xenotransplantation29, e12743 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR35] 35.Soriano, R. M., Penfold, D. & Leslie, S. W. Anatomy, Abdomen and Pelvis: Kidneys (StatPearls, 2023). [PubMed]

[CR36] 36.Grynberg, K., Ma, F. Y. & Nikolic-Paterson, D. J. The JNK signaling pathway in renal fibrosis. Front. Physiol.8, 829 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR37] 37.Zincarelli, C., Soltys, S., Rengo, G. & Rabinowitz, J. E. Analysis of AAV serotypes 1–9 mediated gene expression and tropism in mice after systemic injection. Mol. Ther.16, 1073–1080 (2008). [DOI] [PubMed] [Google Scholar]

[CR38] 38.Hordeaux, J. et al. The neurotropic properties of AAV-PHP.B are limited to C57BL/6J mice. Mol. Ther.26, 664–668 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR39] 39.Gonzalez, T. J. et al. Structure-guided AAV capsid evolution strategies for enhanced CNS gene delivery. Nat. Protoc.10.1038/s41596-023-00875-y (2023). [DOI] [PubMed] [Google Scholar]

[CR40] 40.Przepiorski, A. et al. A simplified method for generating kidney organoids from human pluripotent stem cells. J. Vis. Exp. 10.3791/62452 (2021). [DOI] [PMC free article] [PubMed]

[CR41] 41.Hull, J. & Rosales, A. AAV capsid library analysis scripts for ‘Multiscale and cross-species evolution of adeno-associated viruses for kidney gene transfer’. Zenodo10.5281/zenodo.13620030 (2024).

PERMALINK

Evolving adeno-associated viruses for gene transfer to the kidney via cross-species cycling of capsid libraries

Alan Rosales

Leo O Blondel

Joshua Hull

Qimeng Gao

Nihal Aykun

Jennifer L Peek

Alejandra Vargas

Sophia Fergione

Mingqing Song

Matthew H Wilson

Andrew S Barbas

Aravind Asokan

Abstract

Main

Results

Cycling capsid libraries across different kidney systems yields new AAV variants

Fig. 1. A multidimensional evolution strategy across different kidney model systems in vivo and ex vivo yields AAV.k variants.

AAV.k13 and AAV.k20 transduce three-dimensional (3D) human kidney organoids with high efficiency

Extended Data Fig. 1. Characterization of human kidney organoid model and additional AAV9, AAV.k13 and AAV.k20 transduction profiles.

Fig. 2. AAV.k13 and AAV.k20 transduce 3D human kidney organoids more efficiently than AAV9.

Extended Data Fig. 2. Transduction assays performed in human renal proximal tubule epithelial cells (RPTECs) with AAV9, AAV.k13 and AAV.k20.

AAV.k13 and AAV.k20 display tropism for proximal tubules in mouse and NHP models

Fig. 3. AAV.k13 and AAV.k20 transduce mouse kidneys more efficiently than AAV9 following intravenous administration.

Fig. 4. AAV.k capsids packaging a single-stranded cassette transduce mouse kidneys more efficiently than AAV9 following i.v. administration and non-human primate kidneys after in situ ureteral delivery.

AAV.k13 and AAV.k20 show robust and widespread transduction in pig kidneys ex vivo

Fig. 5. AAV.k13 and AAV.k20 transduce transplanted pig kidneys more efficiently than AAV9.

Discussion

Outlook

Methods

Study design

AAV capsid libraries and recombinant vector production

Cycling of AAV capsid libraries in kidney model systems

High-throughput sequencing analysis and identification of newly evolved AAV.k variants

iPSC-derived 3D human kidney organoid differentiation and processing

iPSC-derived 3D human kidney organoid immunostaining

Renal proximal tubule epithelial cell (TH1) luciferase assay

Determination of vector genome uptake in organoids by quantitative PCR

Intravenous administration in wild-type C57/B6 mice

Western blot analysis for PD-L1 expression in mice

Immunohistology of mouse tissues

Ex vivo kidney perfusion of AAV vectors in pigs and non-human primates

In situ delivery of AAV vectors in non-human primate kidneys

Histology and immunohistochemistry staining of pig and non-human primate kidney tissues for mCherry

Determination of AAV vector transcripts in organoids, pig kidney biopsies and non-human primate kidney biopsies by quantitative RT–PCR

Determination of vector genome biodistribution in multiple species by qPCR

Determination of luciferase activity in mouse tissues, organoids and non-human primate kidney biopsies

Statistical analysis

Reporting summary

Supplementary information

Source data

Acknowledgements

Extended data

Author contributions

Peer review

Peer review information

Data availability

Code availability

Competing interests

Footnotes

Extended data

Supplementary information

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases